linux-stable/mm/compaction.c

3017 lines
82 KiB
C
Raw Normal View History

License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 14:07:57 +00:00
// SPDX-License-Identifier: GPL-2.0
/*
* linux/mm/compaction.c
*
* Memory compaction for the reduction of external fragmentation. Note that
* this heavily depends upon page migration to do all the real heavy
* lifting
*
* Copyright IBM Corp. 2007-2010 Mel Gorman <mel@csn.ul.ie>
*/
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
#include <linux/cpu.h>
#include <linux/swap.h>
#include <linux/migrate.h>
#include <linux/compaction.h>
#include <linux/mm_inline.h>
#include <linux/sched/signal.h>
#include <linux/backing-dev.h>
#include <linux/sysctl.h>
#include <linux/sysfs.h>
#include <linux/page-isolation.h>
#include <linux/kasan.h>
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
#include <linux/kthread.h>
#include <linux/freezer.h>
#include <linux/page_owner.h>
psi: pressure stall information for CPU, memory, and IO When systems are overcommitted and resources become contended, it's hard to tell exactly the impact this has on workload productivity, or how close the system is to lockups and OOM kills. In particular, when machines work multiple jobs concurrently, the impact of overcommit in terms of latency and throughput on the individual job can be enormous. In order to maximize hardware utilization without sacrificing individual job health or risk complete machine lockups, this patch implements a way to quantify resource pressure in the system. A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that expose the percentage of time the system is stalled on CPU, memory, or IO, respectively. Stall states are aggregate versions of the per-task delay accounting delays: cpu: some tasks are runnable but not executing on a CPU memory: tasks are reclaiming, or waiting for swapin or thrashing cache io: tasks are waiting for io completions These percentages of walltime can be thought of as pressure percentages, and they give a general sense of system health and productivity loss incurred by resource overcommit. They can also indicate when the system is approaching lockup scenarios and OOMs. To do this, psi keeps track of the task states associated with each CPU and samples the time they spend in stall states. Every 2 seconds, the samples are averaged across CPUs - weighted by the CPUs' non-idle time to eliminate artifacts from unused CPUs - and translated into percentages of walltime. A running average of those percentages is maintained over 10s, 1m, and 5m periods (similar to the loadaverage). [hannes@cmpxchg.org: doc fixlet, per Randy] Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org [hannes@cmpxchg.org: code optimization] Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org [hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter] Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org [hannes@cmpxchg.org: fix build] Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Tested-by: Daniel Drake <drake@endlessm.com> Tested-by: Suren Baghdasaryan <surenb@google.com> Cc: Christopher Lameter <cl@linux.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Johannes Weiner <jweiner@fb.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Enderborg <peter.enderborg@sony.com> Cc: Randy Dunlap <rdunlap@infradead.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Vinayak Menon <vinmenon@codeaurora.org> Cc: Randy Dunlap <rdunlap@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 22:06:27 +00:00
#include <linux/psi.h>
#include "internal.h"
#ifdef CONFIG_COMPACTION
static inline void count_compact_event(enum vm_event_item item)
{
count_vm_event(item);
}
static inline void count_compact_events(enum vm_event_item item, long delta)
{
count_vm_events(item, delta);
}
#else
#define count_compact_event(item) do { } while (0)
#define count_compact_events(item, delta) do { } while (0)
#endif
#if defined CONFIG_COMPACTION || defined CONFIG_CMA
#define CREATE_TRACE_POINTS
#include <trace/events/compaction.h>
#define block_start_pfn(pfn, order) round_down(pfn, 1UL << (order))
#define block_end_pfn(pfn, order) ALIGN((pfn) + 1, 1UL << (order))
#define pageblock_start_pfn(pfn) block_start_pfn(pfn, pageblock_order)
#define pageblock_end_pfn(pfn) block_end_pfn(pfn, pageblock_order)
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/*
* Fragmentation score check interval for proactive compaction purposes.
*/
static const unsigned int HPAGE_FRAG_CHECK_INTERVAL_MSEC = 500;
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/*
* Page order with-respect-to which proactive compaction
* calculates external fragmentation, which is used as
* the "fragmentation score" of a node/zone.
*/
#if defined CONFIG_TRANSPARENT_HUGEPAGE
#define COMPACTION_HPAGE_ORDER HPAGE_PMD_ORDER
#elif defined CONFIG_HUGETLBFS
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
#define COMPACTION_HPAGE_ORDER HUGETLB_PAGE_ORDER
#else
#define COMPACTION_HPAGE_ORDER (PMD_SHIFT - PAGE_SHIFT)
#endif
static unsigned long release_freepages(struct list_head *freelist)
{
struct page *page, *next;
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
unsigned long high_pfn = 0;
list_for_each_entry_safe(page, next, freelist, lru) {
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
unsigned long pfn = page_to_pfn(page);
list_del(&page->lru);
__free_page(page);
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
if (pfn > high_pfn)
high_pfn = pfn;
}
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
return high_pfn;
}
static void split_map_pages(struct list_head *list)
{
unsigned int i, order, nr_pages;
struct page *page, *next;
LIST_HEAD(tmp_list);
list_for_each_entry_safe(page, next, list, lru) {
list_del(&page->lru);
order = page_private(page);
nr_pages = 1 << order;
mm/page_alloc: introduce post allocation processing on page allocator This patch is motivated from Hugh and Vlastimil's concern [1]. There are two ways to get freepage from the allocator. One is using normal memory allocation API and the other is __isolate_free_page() which is internally used for compaction and pageblock isolation. Later usage is rather tricky since it doesn't do whole post allocation processing done by normal API. One problematic thing I already know is that poisoned page would not be checked if it is allocated by __isolate_free_page(). Perhaps, there would be more. We could add more debug logic for allocated page in the future and this separation would cause more problem. I'd like to fix this situation at this time. Solution is simple. This patch commonize some logic for newly allocated page and uses it on all sites. This will solve the problem. [1] http://marc.info/?i=alpine.LSU.2.11.1604270029350.7066%40eggly.anvils%3E [iamjoonsoo.kim@lge.com: mm-page_alloc-introduce-post-allocation-processing-on-page-allocator-v3] Link: http://lkml.kernel.org/r/1464230275-25791-7-git-send-email-iamjoonsoo.kim@lge.com Link: http://lkml.kernel.org/r/1466150259-27727-9-git-send-email-iamjoonsoo.kim@lge.com Link: http://lkml.kernel.org/r/1464230275-25791-7-git-send-email-iamjoonsoo.kim@lge.com Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Minchan Kim <minchan@kernel.org> Cc: Alexander Potapenko <glider@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Michal Hocko <mhocko@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-26 22:23:58 +00:00
post_alloc_hook(page, order, __GFP_MOVABLE);
if (order)
split_page(page, order);
for (i = 0; i < nr_pages; i++) {
list_add(&page->lru, &tmp_list);
page++;
}
}
list_splice(&tmp_list, list);
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
#ifdef CONFIG_COMPACTION
mm: migrate: support non-lru movable page migration We have allowed migration for only LRU pages until now and it was enough to make high-order pages. But recently, embedded system(e.g., webOS, android) uses lots of non-movable pages(e.g., zram, GPU memory) so we have seen several reports about troubles of small high-order allocation. For fixing the problem, there were several efforts (e,g,. enhance compaction algorithm, SLUB fallback to 0-order page, reserved memory, vmalloc and so on) but if there are lots of non-movable pages in system, their solutions are void in the long run. So, this patch is to support facility to change non-movable pages with movable. For the feature, this patch introduces functions related to migration to address_space_operations as well as some page flags. If a driver want to make own pages movable, it should define three functions which are function pointers of struct address_space_operations. 1. bool (*isolate_page) (struct page *page, isolate_mode_t mode); What VM expects on isolate_page function of driver is to return *true* if driver isolates page successfully. On returing true, VM marks the page as PG_isolated so concurrent isolation in several CPUs skip the page for isolation. If a driver cannot isolate the page, it should return *false*. Once page is successfully isolated, VM uses page.lru fields so driver shouldn't expect to preserve values in that fields. 2. int (*migratepage) (struct address_space *mapping, struct page *newpage, struct page *oldpage, enum migrate_mode); After isolation, VM calls migratepage of driver with isolated page. The function of migratepage is to move content of the old page to new page and set up fields of struct page newpage. Keep in mind that you should indicate to the VM the oldpage is no longer movable via __ClearPageMovable() under page_lock if you migrated the oldpage successfully and returns 0. If driver cannot migrate the page at the moment, driver can return -EAGAIN. On -EAGAIN, VM will retry page migration in a short time because VM interprets -EAGAIN as "temporal migration failure". On returning any error except -EAGAIN, VM will give up the page migration without retrying in this time. Driver shouldn't touch page.lru field VM using in the functions. 3. void (*putback_page)(struct page *); If migration fails on isolated page, VM should return the isolated page to the driver so VM calls driver's putback_page with migration failed page. In this function, driver should put the isolated page back to the own data structure. 4. non-lru movable page flags There are two page flags for supporting non-lru movable page. * PG_movable Driver should use the below function to make page movable under page_lock. void __SetPageMovable(struct page *page, struct address_space *mapping) It needs argument of address_space for registering migration family functions which will be called by VM. Exactly speaking, PG_movable is not a real flag of struct page. Rather than, VM reuses page->mapping's lower bits to represent it. #define PAGE_MAPPING_MOVABLE 0x2 page->mapping = page->mapping | PAGE_MAPPING_MOVABLE; so driver shouldn't access page->mapping directly. Instead, driver should use page_mapping which mask off the low two bits of page->mapping so it can get right struct address_space. For testing of non-lru movable page, VM supports __PageMovable function. However, it doesn't guarantee to identify non-lru movable page because page->mapping field is unified with other variables in struct page. As well, if driver releases the page after isolation by VM, page->mapping doesn't have stable value although it has PAGE_MAPPING_MOVABLE (Look at __ClearPageMovable). But __PageMovable is cheap to catch whether page is LRU or non-lru movable once the page has been isolated. Because LRU pages never can have PAGE_MAPPING_MOVABLE in page->mapping. It is also good for just peeking to test non-lru movable pages before more expensive checking with lock_page in pfn scanning to select victim. For guaranteeing non-lru movable page, VM provides PageMovable function. Unlike __PageMovable, PageMovable functions validates page->mapping and mapping->a_ops->isolate_page under lock_page. The lock_page prevents sudden destroying of page->mapping. Driver using __SetPageMovable should clear the flag via __ClearMovablePage under page_lock before the releasing the page. * PG_isolated To prevent concurrent isolation among several CPUs, VM marks isolated page as PG_isolated under lock_page. So if a CPU encounters PG_isolated non-lru movable page, it can skip it. Driver doesn't need to manipulate the flag because VM will set/clear it automatically. Keep in mind that if driver sees PG_isolated page, it means the page have been isolated by VM so it shouldn't touch page.lru field. PG_isolated is alias with PG_reclaim flag so driver shouldn't use the flag for own purpose. [opensource.ganesh@gmail.com: mm/compaction: remove local variable is_lru] Link: http://lkml.kernel.org/r/20160618014841.GA7422@leo-test Link: http://lkml.kernel.org/r/1464736881-24886-3-git-send-email-minchan@kernel.org Signed-off-by: Gioh Kim <gi-oh.kim@profitbricks.com> Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Ganesh Mahendran <opensource.ganesh@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Rafael Aquini <aquini@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: John Einar Reitan <john.reitan@foss.arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-26 22:23:05 +00:00
int PageMovable(struct page *page)
{
struct address_space *mapping;
VM_BUG_ON_PAGE(!PageLocked(page), page);
if (!__PageMovable(page))
return 0;
mapping = page_mapping(page);
if (mapping && mapping->a_ops && mapping->a_ops->isolate_page)
return 1;
return 0;
}
EXPORT_SYMBOL(PageMovable);
void __SetPageMovable(struct page *page, struct address_space *mapping)
{
VM_BUG_ON_PAGE(!PageLocked(page), page);
VM_BUG_ON_PAGE((unsigned long)mapping & PAGE_MAPPING_MOVABLE, page);
page->mapping = (void *)((unsigned long)mapping | PAGE_MAPPING_MOVABLE);
}
EXPORT_SYMBOL(__SetPageMovable);
void __ClearPageMovable(struct page *page)
{
VM_BUG_ON_PAGE(!PageMovable(page), page);
/*
* Clear registered address_space val with keeping PAGE_MAPPING_MOVABLE
* flag so that VM can catch up released page by driver after isolation.
* With it, VM migration doesn't try to put it back.
*/
page->mapping = (void *)((unsigned long)page->mapping &
PAGE_MAPPING_MOVABLE);
}
EXPORT_SYMBOL(__ClearPageMovable);
/* Do not skip compaction more than 64 times */
#define COMPACT_MAX_DEFER_SHIFT 6
/*
* Compaction is deferred when compaction fails to result in a page
* allocation success. 1 << compact_defer_shift, compactions are skipped up
* to a limit of 1 << COMPACT_MAX_DEFER_SHIFT
*/
static void defer_compaction(struct zone *zone, int order)
{
zone->compact_considered = 0;
zone->compact_defer_shift++;
if (order < zone->compact_order_failed)
zone->compact_order_failed = order;
if (zone->compact_defer_shift > COMPACT_MAX_DEFER_SHIFT)
zone->compact_defer_shift = COMPACT_MAX_DEFER_SHIFT;
trace_mm_compaction_defer_compaction(zone, order);
}
/* Returns true if compaction should be skipped this time */
static bool compaction_deferred(struct zone *zone, int order)
{
unsigned long defer_limit = 1UL << zone->compact_defer_shift;
if (order < zone->compact_order_failed)
return false;
/* Avoid possible overflow */
if (++zone->compact_considered >= defer_limit) {
zone->compact_considered = defer_limit;
return false;
}
trace_mm_compaction_deferred(zone, order);
return true;
}
/*
* Update defer tracking counters after successful compaction of given order,
* which means an allocation either succeeded (alloc_success == true) or is
* expected to succeed.
*/
void compaction_defer_reset(struct zone *zone, int order,
bool alloc_success)
{
if (alloc_success) {
zone->compact_considered = 0;
zone->compact_defer_shift = 0;
}
if (order >= zone->compact_order_failed)
zone->compact_order_failed = order + 1;
trace_mm_compaction_defer_reset(zone, order);
}
/* Returns true if restarting compaction after many failures */
static bool compaction_restarting(struct zone *zone, int order)
{
if (order < zone->compact_order_failed)
return false;
return zone->compact_defer_shift == COMPACT_MAX_DEFER_SHIFT &&
zone->compact_considered >= 1UL << zone->compact_defer_shift;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
/* Returns true if the pageblock should be scanned for pages to isolate. */
static inline bool isolation_suitable(struct compact_control *cc,
struct page *page)
{
if (cc->ignore_skip_hint)
return true;
return !get_pageblock_skip(page);
}
static void reset_cached_positions(struct zone *zone)
{
zone->compact_cached_migrate_pfn[0] = zone->zone_start_pfn;
zone->compact_cached_migrate_pfn[1] = zone->zone_start_pfn;
2016-03-15 21:57:45 +00:00
zone->compact_cached_free_pfn =
pageblock_start_pfn(zone_end_pfn(zone) - 1);
}
/*
* Compound pages of >= pageblock_order should consistently be skipped until
2017-11-17 23:26:34 +00:00
* released. It is always pointless to compact pages of such order (if they are
* migratable), and the pageblocks they occupy cannot contain any free pages.
*/
2017-11-17 23:26:34 +00:00
static bool pageblock_skip_persistent(struct page *page)
{
2017-11-17 23:26:34 +00:00
if (!PageCompound(page))
return false;
2017-11-17 23:26:34 +00:00
page = compound_head(page);
if (compound_order(page) >= pageblock_order)
return true;
return false;
}
2019-03-05 23:45:38 +00:00
static bool
__reset_isolation_pfn(struct zone *zone, unsigned long pfn, bool check_source,
bool check_target)
{
struct page *page = pfn_to_online_page(pfn);
mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints Mikhail Gavrilo reported the following bug being triggered in a Fedora kernel based on 5.1-rc1 but it is relevant to a vanilla kernel. kernel: page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) kernel: ------------[ cut here ]------------ kernel: kernel BUG at include/linux/mm.h:1021! kernel: invalid opcode: 0000 [#1] SMP NOPTI kernel: CPU: 6 PID: 116 Comm: kswapd0 Tainted: G C 5.1.0-0.rc1.git1.3.fc31.x86_64 #1 kernel: Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 1201 12/07/2018 kernel: RIP: 0010:__reset_isolation_pfn+0x244/0x2b0 kernel: Code: fe 06 e8 0f 8e fc ff 44 0f b6 4c 24 04 48 85 c0 0f 85 dc fe ff ff e9 68 fe ff ff 48 c7 c6 58 b7 2e 8c 4c 89 ff e8 0c 75 00 00 <0f> 0b 48 c7 c6 58 b7 2e 8c e8 fe 74 00 00 0f 0b 48 89 fa 41 b8 01 kernel: RSP: 0018:ffff9e2d03f0fde8 EFLAGS: 00010246 kernel: RAX: 0000000000000034 RBX: 000000000081f380 RCX: ffff8cffbddd6c20 kernel: RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8cffbddd6c20 kernel: RBP: 0000000000000001 R08: 0000009898b94613 R09: 0000000000000000 kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000100000 kernel: R13: 0000000000100000 R14: 0000000000000001 R15: ffffca7de07ce000 kernel: FS: 0000000000000000(0000) GS:ffff8cffbdc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fc1670e9000 CR3: 00000007f5276000 CR4: 00000000003406e0 kernel: Call Trace: kernel: __reset_isolation_suitable+0x62/0x120 kernel: reset_isolation_suitable+0x3b/0x40 kernel: kswapd+0x147/0x540 kernel: ? finish_wait+0x90/0x90 kernel: kthread+0x108/0x140 kernel: ? balance_pgdat+0x560/0x560 kernel: ? kthread_park+0x90/0x90 kernel: ret_from_fork+0x27/0x50 He bisected it down to e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints"). The problem is that the patch in question was sloppy with respect to the handling of zone boundaries. In some instances, it was possible for PFNs outside of a zone to be examined and if those were not properly initialised or poisoned then it would trigger the VM_BUG_ON. This patch corrects the zone boundary issues when resetting pageblock skip hints and Mikhail reported that the bug did not trigger after 30 hours of testing. Link: http://lkml.kernel.org/r/20190327085424.GL3189@techsingularity.net Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints") Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Qian Cai <cai@lca.pw> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
2019-04-04 10:54:09 +00:00
struct page *block_page;
2019-03-05 23:45:38 +00:00
struct page *end_page;
unsigned long block_pfn;
if (!page)
return false;
if (zone != page_zone(page))
return false;
if (pageblock_skip_persistent(page))
return false;
/*
* If skip is already cleared do no further checking once the
* restart points have been set.
*/
if (check_source && check_target && !get_pageblock_skip(page))
return true;
/*
* If clearing skip for the target scanner, do not select a
* non-movable pageblock as the starting point.
*/
if (!check_source && check_target &&
get_pageblock_migratetype(page) != MIGRATE_MOVABLE)
return false;
mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints Mikhail Gavrilo reported the following bug being triggered in a Fedora kernel based on 5.1-rc1 but it is relevant to a vanilla kernel. kernel: page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) kernel: ------------[ cut here ]------------ kernel: kernel BUG at include/linux/mm.h:1021! kernel: invalid opcode: 0000 [#1] SMP NOPTI kernel: CPU: 6 PID: 116 Comm: kswapd0 Tainted: G C 5.1.0-0.rc1.git1.3.fc31.x86_64 #1 kernel: Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 1201 12/07/2018 kernel: RIP: 0010:__reset_isolation_pfn+0x244/0x2b0 kernel: Code: fe 06 e8 0f 8e fc ff 44 0f b6 4c 24 04 48 85 c0 0f 85 dc fe ff ff e9 68 fe ff ff 48 c7 c6 58 b7 2e 8c 4c 89 ff e8 0c 75 00 00 <0f> 0b 48 c7 c6 58 b7 2e 8c e8 fe 74 00 00 0f 0b 48 89 fa 41 b8 01 kernel: RSP: 0018:ffff9e2d03f0fde8 EFLAGS: 00010246 kernel: RAX: 0000000000000034 RBX: 000000000081f380 RCX: ffff8cffbddd6c20 kernel: RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8cffbddd6c20 kernel: RBP: 0000000000000001 R08: 0000009898b94613 R09: 0000000000000000 kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000100000 kernel: R13: 0000000000100000 R14: 0000000000000001 R15: ffffca7de07ce000 kernel: FS: 0000000000000000(0000) GS:ffff8cffbdc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fc1670e9000 CR3: 00000007f5276000 CR4: 00000000003406e0 kernel: Call Trace: kernel: __reset_isolation_suitable+0x62/0x120 kernel: reset_isolation_suitable+0x3b/0x40 kernel: kswapd+0x147/0x540 kernel: ? finish_wait+0x90/0x90 kernel: kthread+0x108/0x140 kernel: ? balance_pgdat+0x560/0x560 kernel: ? kthread_park+0x90/0x90 kernel: ret_from_fork+0x27/0x50 He bisected it down to e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints"). The problem is that the patch in question was sloppy with respect to the handling of zone boundaries. In some instances, it was possible for PFNs outside of a zone to be examined and if those were not properly initialised or poisoned then it would trigger the VM_BUG_ON. This patch corrects the zone boundary issues when resetting pageblock skip hints and Mikhail reported that the bug did not trigger after 30 hours of testing. Link: http://lkml.kernel.org/r/20190327085424.GL3189@techsingularity.net Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints") Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Qian Cai <cai@lca.pw> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
2019-04-04 10:54:09 +00:00
/* Ensure the start of the pageblock or zone is online and valid */
block_pfn = pageblock_start_pfn(pfn);
block_pfn = max(block_pfn, zone->zone_start_pfn);
block_page = pfn_to_online_page(block_pfn);
mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints Mikhail Gavrilo reported the following bug being triggered in a Fedora kernel based on 5.1-rc1 but it is relevant to a vanilla kernel. kernel: page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) kernel: ------------[ cut here ]------------ kernel: kernel BUG at include/linux/mm.h:1021! kernel: invalid opcode: 0000 [#1] SMP NOPTI kernel: CPU: 6 PID: 116 Comm: kswapd0 Tainted: G C 5.1.0-0.rc1.git1.3.fc31.x86_64 #1 kernel: Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 1201 12/07/2018 kernel: RIP: 0010:__reset_isolation_pfn+0x244/0x2b0 kernel: Code: fe 06 e8 0f 8e fc ff 44 0f b6 4c 24 04 48 85 c0 0f 85 dc fe ff ff e9 68 fe ff ff 48 c7 c6 58 b7 2e 8c 4c 89 ff e8 0c 75 00 00 <0f> 0b 48 c7 c6 58 b7 2e 8c e8 fe 74 00 00 0f 0b 48 89 fa 41 b8 01 kernel: RSP: 0018:ffff9e2d03f0fde8 EFLAGS: 00010246 kernel: RAX: 0000000000000034 RBX: 000000000081f380 RCX: ffff8cffbddd6c20 kernel: RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8cffbddd6c20 kernel: RBP: 0000000000000001 R08: 0000009898b94613 R09: 0000000000000000 kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000100000 kernel: R13: 0000000000100000 R14: 0000000000000001 R15: ffffca7de07ce000 kernel: FS: 0000000000000000(0000) GS:ffff8cffbdc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fc1670e9000 CR3: 00000007f5276000 CR4: 00000000003406e0 kernel: Call Trace: kernel: __reset_isolation_suitable+0x62/0x120 kernel: reset_isolation_suitable+0x3b/0x40 kernel: kswapd+0x147/0x540 kernel: ? finish_wait+0x90/0x90 kernel: kthread+0x108/0x140 kernel: ? balance_pgdat+0x560/0x560 kernel: ? kthread_park+0x90/0x90 kernel: ret_from_fork+0x27/0x50 He bisected it down to e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints"). The problem is that the patch in question was sloppy with respect to the handling of zone boundaries. In some instances, it was possible for PFNs outside of a zone to be examined and if those were not properly initialised or poisoned then it would trigger the VM_BUG_ON. This patch corrects the zone boundary issues when resetting pageblock skip hints and Mikhail reported that the bug did not trigger after 30 hours of testing. Link: http://lkml.kernel.org/r/20190327085424.GL3189@techsingularity.net Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints") Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Qian Cai <cai@lca.pw> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
2019-04-04 10:54:09 +00:00
if (block_page) {
page = block_page;
pfn = block_pfn;
}
/* Ensure the end of the pageblock or zone is online and valid */
block_pfn = pageblock_end_pfn(pfn) - 1;
mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints Mikhail Gavrilo reported the following bug being triggered in a Fedora kernel based on 5.1-rc1 but it is relevant to a vanilla kernel. kernel: page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) kernel: ------------[ cut here ]------------ kernel: kernel BUG at include/linux/mm.h:1021! kernel: invalid opcode: 0000 [#1] SMP NOPTI kernel: CPU: 6 PID: 116 Comm: kswapd0 Tainted: G C 5.1.0-0.rc1.git1.3.fc31.x86_64 #1 kernel: Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 1201 12/07/2018 kernel: RIP: 0010:__reset_isolation_pfn+0x244/0x2b0 kernel: Code: fe 06 e8 0f 8e fc ff 44 0f b6 4c 24 04 48 85 c0 0f 85 dc fe ff ff e9 68 fe ff ff 48 c7 c6 58 b7 2e 8c 4c 89 ff e8 0c 75 00 00 <0f> 0b 48 c7 c6 58 b7 2e 8c e8 fe 74 00 00 0f 0b 48 89 fa 41 b8 01 kernel: RSP: 0018:ffff9e2d03f0fde8 EFLAGS: 00010246 kernel: RAX: 0000000000000034 RBX: 000000000081f380 RCX: ffff8cffbddd6c20 kernel: RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8cffbddd6c20 kernel: RBP: 0000000000000001 R08: 0000009898b94613 R09: 0000000000000000 kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000100000 kernel: R13: 0000000000100000 R14: 0000000000000001 R15: ffffca7de07ce000 kernel: FS: 0000000000000000(0000) GS:ffff8cffbdc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fc1670e9000 CR3: 00000007f5276000 CR4: 00000000003406e0 kernel: Call Trace: kernel: __reset_isolation_suitable+0x62/0x120 kernel: reset_isolation_suitable+0x3b/0x40 kernel: kswapd+0x147/0x540 kernel: ? finish_wait+0x90/0x90 kernel: kthread+0x108/0x140 kernel: ? balance_pgdat+0x560/0x560 kernel: ? kthread_park+0x90/0x90 kernel: ret_from_fork+0x27/0x50 He bisected it down to e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints"). The problem is that the patch in question was sloppy with respect to the handling of zone boundaries. In some instances, it was possible for PFNs outside of a zone to be examined and if those were not properly initialised or poisoned then it would trigger the VM_BUG_ON. This patch corrects the zone boundary issues when resetting pageblock skip hints and Mikhail reported that the bug did not trigger after 30 hours of testing. Link: http://lkml.kernel.org/r/20190327085424.GL3189@techsingularity.net Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints") Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Qian Cai <cai@lca.pw> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
2019-04-04 10:54:09 +00:00
block_pfn = min(block_pfn, zone_end_pfn(zone) - 1);
end_page = pfn_to_online_page(block_pfn);
if (!end_page)
return false;
2019-03-05 23:45:38 +00:00
/*
* Only clear the hint if a sample indicates there is either a
* free page or an LRU page in the block. One or other condition
* is necessary for the block to be a migration source/target.
*/
do {
if (pfn_valid_within(pfn)) {
if (check_source && PageLRU(page)) {
clear_pageblock_skip(page);
return true;
}
if (check_target && PageBuddy(page)) {
clear_pageblock_skip(page);
return true;
}
}
page += (1 << PAGE_ALLOC_COSTLY_ORDER);
pfn += (1 << PAGE_ALLOC_COSTLY_ORDER);
} while (page <= end_page);
2019-03-05 23:45:38 +00:00
return false;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
/*
* This function is called to clear all cached information on pageblocks that
* should be skipped for page isolation when the migrate and free page scanner
* meet.
*/
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
static void __reset_isolation_suitable(struct zone *zone)
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
{
2019-03-05 23:45:38 +00:00
unsigned long migrate_pfn = zone->zone_start_pfn;
mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints Mikhail Gavrilo reported the following bug being triggered in a Fedora kernel based on 5.1-rc1 but it is relevant to a vanilla kernel. kernel: page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) kernel: ------------[ cut here ]------------ kernel: kernel BUG at include/linux/mm.h:1021! kernel: invalid opcode: 0000 [#1] SMP NOPTI kernel: CPU: 6 PID: 116 Comm: kswapd0 Tainted: G C 5.1.0-0.rc1.git1.3.fc31.x86_64 #1 kernel: Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 1201 12/07/2018 kernel: RIP: 0010:__reset_isolation_pfn+0x244/0x2b0 kernel: Code: fe 06 e8 0f 8e fc ff 44 0f b6 4c 24 04 48 85 c0 0f 85 dc fe ff ff e9 68 fe ff ff 48 c7 c6 58 b7 2e 8c 4c 89 ff e8 0c 75 00 00 <0f> 0b 48 c7 c6 58 b7 2e 8c e8 fe 74 00 00 0f 0b 48 89 fa 41 b8 01 kernel: RSP: 0018:ffff9e2d03f0fde8 EFLAGS: 00010246 kernel: RAX: 0000000000000034 RBX: 000000000081f380 RCX: ffff8cffbddd6c20 kernel: RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8cffbddd6c20 kernel: RBP: 0000000000000001 R08: 0000009898b94613 R09: 0000000000000000 kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000100000 kernel: R13: 0000000000100000 R14: 0000000000000001 R15: ffffca7de07ce000 kernel: FS: 0000000000000000(0000) GS:ffff8cffbdc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fc1670e9000 CR3: 00000007f5276000 CR4: 00000000003406e0 kernel: Call Trace: kernel: __reset_isolation_suitable+0x62/0x120 kernel: reset_isolation_suitable+0x3b/0x40 kernel: kswapd+0x147/0x540 kernel: ? finish_wait+0x90/0x90 kernel: kthread+0x108/0x140 kernel: ? balance_pgdat+0x560/0x560 kernel: ? kthread_park+0x90/0x90 kernel: ret_from_fork+0x27/0x50 He bisected it down to e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints"). The problem is that the patch in question was sloppy with respect to the handling of zone boundaries. In some instances, it was possible for PFNs outside of a zone to be examined and if those were not properly initialised or poisoned then it would trigger the VM_BUG_ON. This patch corrects the zone boundary issues when resetting pageblock skip hints and Mikhail reported that the bug did not trigger after 30 hours of testing. Link: http://lkml.kernel.org/r/20190327085424.GL3189@techsingularity.net Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints") Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Qian Cai <cai@lca.pw> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
2019-04-04 10:54:09 +00:00
unsigned long free_pfn = zone_end_pfn(zone) - 1;
2019-03-05 23:45:38 +00:00
unsigned long reset_migrate = free_pfn;
unsigned long reset_free = migrate_pfn;
bool source_set = false;
bool free_set = false;
if (!zone->compact_blockskip_flush)
return;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
zone->compact_blockskip_flush = false;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
2019-03-05 23:45:38 +00:00
/*
* Walk the zone and update pageblock skip information. Source looks
* for PageLRU while target looks for PageBuddy. When the scanner
* is found, both PageBuddy and PageLRU are checked as the pageblock
* is suitable as both source and target.
*/
for (; migrate_pfn < free_pfn; migrate_pfn += pageblock_nr_pages,
free_pfn -= pageblock_nr_pages) {
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
cond_resched();
2019-03-05 23:45:38 +00:00
/* Update the migrate PFN */
if (__reset_isolation_pfn(zone, migrate_pfn, true, source_set) &&
migrate_pfn < reset_migrate) {
source_set = true;
reset_migrate = migrate_pfn;
zone->compact_init_migrate_pfn = reset_migrate;
zone->compact_cached_migrate_pfn[0] = reset_migrate;
zone->compact_cached_migrate_pfn[1] = reset_migrate;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
2019-03-05 23:45:38 +00:00
/* Update the free PFN */
if (__reset_isolation_pfn(zone, free_pfn, free_set, true) &&
free_pfn > reset_free) {
free_set = true;
reset_free = free_pfn;
zone->compact_init_free_pfn = reset_free;
zone->compact_cached_free_pfn = reset_free;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
}
2019-03-05 23:45:38 +00:00
/* Leave no distance if no suitable block was reset */
if (reset_migrate >= reset_free) {
zone->compact_cached_migrate_pfn[0] = migrate_pfn;
zone->compact_cached_migrate_pfn[1] = migrate_pfn;
zone->compact_cached_free_pfn = free_pfn;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
}
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
void reset_isolation_suitable(pg_data_t *pgdat)
{
int zoneid;
for (zoneid = 0; zoneid < MAX_NR_ZONES; zoneid++) {
struct zone *zone = &pgdat->node_zones[zoneid];
if (!populated_zone(zone))
continue;
/* Only flush if a full compaction finished recently */
if (zone->compact_blockskip_flush)
__reset_isolation_suitable(zone);
}
}
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
/*
* Sets the pageblock skip bit if it was clear. Note that this is a hint as
* locks are not required for read/writers. Returns true if it was already set.
*/
static bool test_and_set_skip(struct compact_control *cc, struct page *page,
unsigned long pfn)
{
bool skip;
/* Do no update if skip hint is being ignored */
if (cc->ignore_skip_hint)
return false;
if (!IS_ALIGNED(pfn, pageblock_nr_pages))
return false;
skip = get_pageblock_skip(page);
if (!skip && !cc->no_set_skip_hint)
set_pageblock_skip(page);
return skip;
}
static void update_cached_migrate(struct compact_control *cc, unsigned long pfn)
{
struct zone *zone = cc->zone;
pfn = pageblock_end_pfn(pfn);
/* Set for isolation rather than compaction */
if (cc->no_set_skip_hint)
return;
if (pfn > zone->compact_cached_migrate_pfn[0])
zone->compact_cached_migrate_pfn[0] = pfn;
if (cc->mode != MIGRATE_ASYNC &&
pfn > zone->compact_cached_migrate_pfn[1])
zone->compact_cached_migrate_pfn[1] = pfn;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
/*
* If no pages were isolated then mark this pageblock to be skipped in the
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
* future. The information is later cleared by __reset_isolation_suitable().
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
*/
mm: compaction: Restart compaction from near where it left off This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:45 +00:00
static void update_pageblock_skip(struct compact_control *cc,
mm, compaction: reduce premature advancement of the migration target scanner The fast isolation of free pages allows the cached PFN of the free scanner to advance faster than necessary depending on the contents of the free list. The key is that fast_isolate_freepages() can update zone->compact_cached_free_pfn via isolate_freepages_block(). When the fast search fails, the linear scan can start from a point that has skipped valid migration targets, particularly pageblocks with just low-order free pages. This can cause the migration source/target scanners to meet prematurely causing a reset. This patch starts by avoiding an update of the pageblock skip information and cached PFN from isolate_freepages_block() and puts the responsibility of updating that information in the callers. The fast scanner will update the cached PFN if and only if it finds a block that is higher than the existing cached PFN and sets the skip if the pageblock is full or nearly full. The linear scanner will update skipped information and the cached PFN only when a block is completely scanned. The total impact is that the free scanner advances more slowly as it is primarily driven by the linear scanner instead of the fast search. 5.0.0-rc1 5.0.0-rc1 noresched-v3r17 slowfree-v3r17 Amean fault-both-3 2965.68 ( 0.00%) 3036.75 ( -2.40%) Amean fault-both-5 3995.90 ( 0.00%) 4522.24 * -13.17%* Amean fault-both-7 5842.12 ( 0.00%) 6365.35 ( -8.96%) Amean fault-both-12 9550.87 ( 0.00%) 10340.93 ( -8.27%) Amean fault-both-18 13304.72 ( 0.00%) 14732.46 ( -10.73%) Amean fault-both-24 14618.59 ( 0.00%) 16288.96 ( -11.43%) Amean fault-both-30 16650.96 ( 0.00%) 16346.21 ( 1.83%) Amean fault-both-32 17145.15 ( 0.00%) 19317.49 ( -12.67%) The impact to latency is higher than the last version but it appears to be due to a slight increase in the free scan rates which is a potential side-effect of the patch. However, this is necessary for later patches that are more careful about how pageblocks are treated as earlier iterations of those patches hit corner cases where the restarts were punishing and very visible. Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:28 +00:00
struct page *page, unsigned long pfn)
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
{
mm: compaction: Restart compaction from near where it left off This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:45 +00:00
struct zone *zone = cc->zone;
if (cc->no_set_skip_hint)
return;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
if (!page)
return;
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
set_pageblock_skip(page);
mm: compaction: Restart compaction from near where it left off This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:45 +00:00
/* Update where async and sync compaction should restart */
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
if (pfn < zone->compact_cached_free_pfn)
zone->compact_cached_free_pfn = pfn;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
}
#else
static inline bool isolation_suitable(struct compact_control *cc,
struct page *page)
{
return true;
}
2017-11-17 23:26:34 +00:00
static inline bool pageblock_skip_persistent(struct page *page)
{
return false;
}
static inline void update_pageblock_skip(struct compact_control *cc,
mm, compaction: reduce premature advancement of the migration target scanner The fast isolation of free pages allows the cached PFN of the free scanner to advance faster than necessary depending on the contents of the free list. The key is that fast_isolate_freepages() can update zone->compact_cached_free_pfn via isolate_freepages_block(). When the fast search fails, the linear scan can start from a point that has skipped valid migration targets, particularly pageblocks with just low-order free pages. This can cause the migration source/target scanners to meet prematurely causing a reset. This patch starts by avoiding an update of the pageblock skip information and cached PFN from isolate_freepages_block() and puts the responsibility of updating that information in the callers. The fast scanner will update the cached PFN if and only if it finds a block that is higher than the existing cached PFN and sets the skip if the pageblock is full or nearly full. The linear scanner will update skipped information and the cached PFN only when a block is completely scanned. The total impact is that the free scanner advances more slowly as it is primarily driven by the linear scanner instead of the fast search. 5.0.0-rc1 5.0.0-rc1 noresched-v3r17 slowfree-v3r17 Amean fault-both-3 2965.68 ( 0.00%) 3036.75 ( -2.40%) Amean fault-both-5 3995.90 ( 0.00%) 4522.24 * -13.17%* Amean fault-both-7 5842.12 ( 0.00%) 6365.35 ( -8.96%) Amean fault-both-12 9550.87 ( 0.00%) 10340.93 ( -8.27%) Amean fault-both-18 13304.72 ( 0.00%) 14732.46 ( -10.73%) Amean fault-both-24 14618.59 ( 0.00%) 16288.96 ( -11.43%) Amean fault-both-30 16650.96 ( 0.00%) 16346.21 ( 1.83%) Amean fault-both-32 17145.15 ( 0.00%) 19317.49 ( -12.67%) The impact to latency is higher than the last version but it appears to be due to a slight increase in the free scan rates which is a potential side-effect of the patch. However, this is necessary for later patches that are more careful about how pageblocks are treated as earlier iterations of those patches hit corner cases where the restarts were punishing and very visible. Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:28 +00:00
struct page *page, unsigned long pfn)
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
{
}
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
static void update_cached_migrate(struct compact_control *cc, unsigned long pfn)
{
}
static bool test_and_set_skip(struct compact_control *cc, struct page *page,
unsigned long pfn)
{
return false;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
#endif /* CONFIG_COMPACTION */
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
/*
* Compaction requires the taking of some coarse locks that are potentially
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
* very heavily contended. For async compaction, trylock and record if the
* lock is contended. The lock will still be acquired but compaction will
* abort when the current block is finished regardless of success rate.
* Sync compaction acquires the lock.
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
*
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
* Always returns true which makes it easier to track lock state in callers.
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
*/
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
static bool compact_lock_irqsave(spinlock_t *lock, unsigned long *flags,
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
struct compact_control *cc)
__acquires(lock)
mm: compaction: acquire the zone->lru_lock as late as possible Richard Davies and Shaohua Li have both reported lock contention problems in compaction on the zone and LRU locks as well as significant amounts of time being spent in compaction. This series aims to reduce lock contention and scanning rates to reduce that CPU usage. Richard reported at https://lkml.org/lkml/2012/9/21/91 that this series made a big different to a problem he reported in August: http://marc.info/?l=kvm&m=134511507015614&w=2 Patch 1 defers acquiring the zone->lru_lock as long as possible. Patch 2 defers acquiring the zone->lock as lock as possible. Patch 3 reverts Rik's "skip-free" patches as the core concept gets reimplemented later and the remaining patches are easier to understand if this is reverted first. Patch 4 adds a pageblock-skip bit to the pageblock flags to cache what pageblocks should be skipped by the migrate and free scanners. This drastically reduces the amount of scanning compaction has to do. Patch 5 reimplements something similar to Rik's idea except it uses the pageblock-skip information to decide where the scanners should restart from and does not need to wrap around. I tested this on 3.6-rc6 + linux-next/akpm. Kernels tested were akpm-20120920 3.6-rc6 + linux-next/akpm as of Septeber 20th, 2012 lesslock Patches 1-6 revert Patches 1-7 cachefail Patches 1-8 skipuseless Patches 1-9 Stress high-order allocation tests looked ok. Success rates are more or less the same with the full series applied but there is an expectation that there is less opportunity to race with other allocation requests if there is less scanning. The time to complete the tests did not vary that much and are uninteresting as were the vmstat statistics so I will not present them here. Using ftrace I recorded how much scanning was done by compaction and got this 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 akpm-20120920 lockless revert-v2r2 cachefail skipuseless Total free scanned 360753976 515414028 565479007 17103281 18916589 Total free isolated 2852429 3597369 4048601 670493 727840 Total free efficiency 0.0079% 0.0070% 0.0072% 0.0392% 0.0385% Total migrate scanned 247728664 822729112 1004645830 17946827 14118903 Total migrate isolated 2555324 3245937 3437501 616359 658616 Total migrate efficiency 0.0103% 0.0039% 0.0034% 0.0343% 0.0466% The efficiency is worthless because of the nature of the test and the number of failures. The really interesting point as far as this patch series is concerned is the number of pages scanned. Note that reverting Rik's patches massively increases the number of pages scanned indicating that those patches really did make a difference to CPU usage. However, caching what pageblocks should be skipped has a much higher impact. With patches 1-8 applied, free page and migrate page scanning are both reduced by 95% in comparison to the akpm kernel. If the basic concept of Rik's patches are implemened on top then scanning then the free scanner barely changed but migrate scanning was further reduced. That said, tests on 3.6-rc5 indicated that the last patch had greater impact than what was measured here so it is a bit variable. One way or the other, this series has a large impact on the amount of scanning compaction does when there is a storm of THP allocations. This patch: Compaction's migrate scanner acquires the zone->lru_lock when scanning a range of pages looking for LRU pages to acquire. It does this even if there are no LRU pages in the range. If multiple processes are compacting then this can cause severe locking contention. To make matters worse commit b2eef8c0 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") releases the lru_lock every SWAP_CLUSTER_MAX pages that are scanned. This patch makes two changes to how the migrate scanner acquires the LRU lock. First, it only releases the LRU lock every SWAP_CLUSTER_MAX pages if the lock is contended. This reduces the number of times it unnecessarily disables and re-enables IRQs. The second is that it defers acquiring the LRU lock for as long as possible. If there are no LRU pages or the only LRU pages are transhuge then the LRU lock will not be acquired at all which reduces contention on zone->lru_lock. [minchan@kernel.org: augment comment] [akpm@linux-foundation.org: tweak comment text] Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:33 +00:00
{
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
/* Track if the lock is contended in async mode */
if (cc->mode == MIGRATE_ASYNC && !cc->contended) {
if (spin_trylock_irqsave(lock, *flags))
return true;
cc->contended = true;
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
}
mm, compaction: khugepaged should not give up due to need_resched() Async compaction aborts when it detects zone lock contention or need_resched() is true. David Rientjes has reported that in practice, most direct async compactions for THP allocation abort due to need_resched(). This means that a second direct compaction is never attempted, which might be OK for a page fault, but khugepaged is intended to attempt a sync compaction in such case and in these cases it won't. This patch replaces "bool contended" in compact_control with an int that distinguishes between aborting due to need_resched() and aborting due to lock contention. This allows propagating the abort through all compaction functions as before, but passing the abort reason up to __alloc_pages_slowpath() which decides when to continue with direct reclaim and another compaction attempt. Another problem is that try_to_compact_pages() did not act upon the reported contention (both need_resched() or lock contention) immediately and would proceed with another zone from the zonelist. When need_resched() is true, that means initializing another zone compaction, only to check again need_resched() in isolate_migratepages() and aborting. For zone lock contention, the unintended consequence is that the lock contended status reported back to the allocator is detrmined from the last zone where compaction was attempted, which is rather arbitrary. This patch fixes the problem in the following way: - async compaction of a zone aborting due to need_resched() or fatal signal pending means that further zones should not be tried. We report COMPACT_CONTENDED_SCHED to the allocator. - aborting zone compaction due to lock contention means we can still try another zone, since it has different set of locks. We report back COMPACT_CONTENDED_LOCK only if *all* zones where compaction was attempted, it was aborted due to lock contention. As a result of these fixes, khugepaged will proceed with second sync compaction as intended, when the preceding async compaction aborted due to need_resched(). Page fault compactions aborting due to need_resched() will spare some cycles previously wasted by initializing another zone compaction only to abort again. Lock contention will be reported only when compaction in all zones aborted due to lock contention, and therefore it's not a good idea to try again after reclaim. In stress-highalloc from mmtests configured to use __GFP_NO_KSWAPD, this has improved number of THP collapse allocations by 10%, which shows positive effect on khugepaged. The benchmark's success rates are unchanged as it is not recognized as khugepaged. Numbers of compact_stall and compact_fail events have however decreased by 20%, with compact_success still a bit improved, which is good. With benchmark configured not to use __GFP_NO_KSWAPD, there is 6% improvement in THP collapse allocations, and only slight improvement in stalls and failures. [akpm@linux-foundation.org: fix warnings] Reported-by: David Rientjes <rientjes@google.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:14 +00:00
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
spin_lock_irqsave(lock, *flags);
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
return true;
mm: compaction: acquire the zone->lru_lock as late as possible Richard Davies and Shaohua Li have both reported lock contention problems in compaction on the zone and LRU locks as well as significant amounts of time being spent in compaction. This series aims to reduce lock contention and scanning rates to reduce that CPU usage. Richard reported at https://lkml.org/lkml/2012/9/21/91 that this series made a big different to a problem he reported in August: http://marc.info/?l=kvm&m=134511507015614&w=2 Patch 1 defers acquiring the zone->lru_lock as long as possible. Patch 2 defers acquiring the zone->lock as lock as possible. Patch 3 reverts Rik's "skip-free" patches as the core concept gets reimplemented later and the remaining patches are easier to understand if this is reverted first. Patch 4 adds a pageblock-skip bit to the pageblock flags to cache what pageblocks should be skipped by the migrate and free scanners. This drastically reduces the amount of scanning compaction has to do. Patch 5 reimplements something similar to Rik's idea except it uses the pageblock-skip information to decide where the scanners should restart from and does not need to wrap around. I tested this on 3.6-rc6 + linux-next/akpm. Kernels tested were akpm-20120920 3.6-rc6 + linux-next/akpm as of Septeber 20th, 2012 lesslock Patches 1-6 revert Patches 1-7 cachefail Patches 1-8 skipuseless Patches 1-9 Stress high-order allocation tests looked ok. Success rates are more or less the same with the full series applied but there is an expectation that there is less opportunity to race with other allocation requests if there is less scanning. The time to complete the tests did not vary that much and are uninteresting as were the vmstat statistics so I will not present them here. Using ftrace I recorded how much scanning was done by compaction and got this 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 akpm-20120920 lockless revert-v2r2 cachefail skipuseless Total free scanned 360753976 515414028 565479007 17103281 18916589 Total free isolated 2852429 3597369 4048601 670493 727840 Total free efficiency 0.0079% 0.0070% 0.0072% 0.0392% 0.0385% Total migrate scanned 247728664 822729112 1004645830 17946827 14118903 Total migrate isolated 2555324 3245937 3437501 616359 658616 Total migrate efficiency 0.0103% 0.0039% 0.0034% 0.0343% 0.0466% The efficiency is worthless because of the nature of the test and the number of failures. The really interesting point as far as this patch series is concerned is the number of pages scanned. Note that reverting Rik's patches massively increases the number of pages scanned indicating that those patches really did make a difference to CPU usage. However, caching what pageblocks should be skipped has a much higher impact. With patches 1-8 applied, free page and migrate page scanning are both reduced by 95% in comparison to the akpm kernel. If the basic concept of Rik's patches are implemened on top then scanning then the free scanner barely changed but migrate scanning was further reduced. That said, tests on 3.6-rc5 indicated that the last patch had greater impact than what was measured here so it is a bit variable. One way or the other, this series has a large impact on the amount of scanning compaction does when there is a storm of THP allocations. This patch: Compaction's migrate scanner acquires the zone->lru_lock when scanning a range of pages looking for LRU pages to acquire. It does this even if there are no LRU pages in the range. If multiple processes are compacting then this can cause severe locking contention. To make matters worse commit b2eef8c0 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") releases the lru_lock every SWAP_CLUSTER_MAX pages that are scanned. This patch makes two changes to how the migrate scanner acquires the LRU lock. First, it only releases the LRU lock every SWAP_CLUSTER_MAX pages if the lock is contended. This reduces the number of times it unnecessarily disables and re-enables IRQs. The second is that it defers acquiring the LRU lock for as long as possible. If there are no LRU pages or the only LRU pages are transhuge then the LRU lock will not be acquired at all which reduces contention on zone->lru_lock. [minchan@kernel.org: augment comment] [akpm@linux-foundation.org: tweak comment text] Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:33 +00:00
}
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
/*
* Compaction requires the taking of some coarse locks that are potentially
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
* very heavily contended. The lock should be periodically unlocked to avoid
* having disabled IRQs for a long time, even when there is nobody waiting on
* the lock. It might also be that allowing the IRQs will result in
* need_resched() becoming true. If scheduling is needed, async compaction
* aborts. Sync compaction schedules.
* Either compaction type will also abort if a fatal signal is pending.
* In either case if the lock was locked, it is dropped and not regained.
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
*
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
* Returns true if compaction should abort due to fatal signal pending, or
* async compaction due to need_resched()
* Returns false when compaction can continue (sync compaction might have
* scheduled)
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
*/
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
static bool compact_unlock_should_abort(spinlock_t *lock,
unsigned long flags, bool *locked, struct compact_control *cc)
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
{
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
if (*locked) {
spin_unlock_irqrestore(lock, flags);
*locked = false;
}
mm, compaction: khugepaged should not give up due to need_resched() Async compaction aborts when it detects zone lock contention or need_resched() is true. David Rientjes has reported that in practice, most direct async compactions for THP allocation abort due to need_resched(). This means that a second direct compaction is never attempted, which might be OK for a page fault, but khugepaged is intended to attempt a sync compaction in such case and in these cases it won't. This patch replaces "bool contended" in compact_control with an int that distinguishes between aborting due to need_resched() and aborting due to lock contention. This allows propagating the abort through all compaction functions as before, but passing the abort reason up to __alloc_pages_slowpath() which decides when to continue with direct reclaim and another compaction attempt. Another problem is that try_to_compact_pages() did not act upon the reported contention (both need_resched() or lock contention) immediately and would proceed with another zone from the zonelist. When need_resched() is true, that means initializing another zone compaction, only to check again need_resched() in isolate_migratepages() and aborting. For zone lock contention, the unintended consequence is that the lock contended status reported back to the allocator is detrmined from the last zone where compaction was attempted, which is rather arbitrary. This patch fixes the problem in the following way: - async compaction of a zone aborting due to need_resched() or fatal signal pending means that further zones should not be tried. We report COMPACT_CONTENDED_SCHED to the allocator. - aborting zone compaction due to lock contention means we can still try another zone, since it has different set of locks. We report back COMPACT_CONTENDED_LOCK only if *all* zones where compaction was attempted, it was aborted due to lock contention. As a result of these fixes, khugepaged will proceed with second sync compaction as intended, when the preceding async compaction aborted due to need_resched(). Page fault compactions aborting due to need_resched() will spare some cycles previously wasted by initializing another zone compaction only to abort again. Lock contention will be reported only when compaction in all zones aborted due to lock contention, and therefore it's not a good idea to try again after reclaim. In stress-highalloc from mmtests configured to use __GFP_NO_KSWAPD, this has improved number of THP collapse allocations by 10%, which shows positive effect on khugepaged. The benchmark's success rates are unchanged as it is not recognized as khugepaged. Numbers of compact_stall and compact_fail events have however decreased by 20%, with compact_success still a bit improved, which is good. With benchmark configured not to use __GFP_NO_KSWAPD, there is 6% improvement in THP collapse allocations, and only slight improvement in stalls and failures. [akpm@linux-foundation.org: fix warnings] Reported-by: David Rientjes <rientjes@google.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:14 +00:00
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
if (fatal_signal_pending(current)) {
mm, compaction: simplify contended compaction handling Async compaction detects contention either due to failing trylock on zone->lock or lru_lock, or by need_resched(). Since 1f9efdef4f3f ("mm, compaction: khugepaged should not give up due to need_resched()") the code got quite complicated to distinguish these two up to the __alloc_pages_slowpath() level, so different decisions could be taken for khugepaged allocations. After the recent changes, khugepaged allocations don't check for contended compaction anymore, so we again don't need to distinguish lock and sched contention, and simplify the current convoluted code a lot. However, I believe it's also possible to simplify even more and completely remove the check for contended compaction after the initial async compaction for costly orders, which was originally aimed at THP page fault allocations. There are several reasons why this can be done now: - with the new defaults, THP page faults no longer do reclaim/compaction at all, unless the system admin has overridden the default, or application has indicated via madvise that it can benefit from THP's. In both cases, it means that the potential extra latency is expected and worth the benefits. - even if reclaim/compaction proceeds after this patch where it previously wouldn't, the second compaction attempt is still async and will detect the contention and back off, if the contention persists - there are still heuristics like deferred compaction and pageblock skip bits in place that prevent excessive THP page fault latencies Link: http://lkml.kernel.org/r/20160721073614.24395-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:49:30 +00:00
cc->contended = true;
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
return true;
}
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
mm, compaction: do not consider a need to reschedule as contention Scanning on large machines can take a considerable length of time and eventually need to be rescheduled. This is treated as an abort event but that's not appropriate as the attempt is likely to be retried after making numerous checks and taking another cycle through the page allocator. This patch will check the need to reschedule if necessary but continue the scanning. The main benefit is reduced scanning when compaction is taking a long time or the machine is over-saturated. It also avoids an unnecessary exit of compaction that ends up being retried by the page allocator in the outer loop. 5.0.0-rc1 5.0.0-rc1 synccached-v3r16 noresched-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2958.27 ( 0.00%) 2965.68 ( -0.25%) Amean fault-both-5 4091.90 ( 0.00%) 3995.90 ( 2.35%) Amean fault-both-7 5803.05 ( 0.00%) 5842.12 ( -0.67%) Amean fault-both-12 9481.06 ( 0.00%) 9550.87 ( -0.74%) Amean fault-both-18 14141.51 ( 0.00%) 13304.72 ( 5.92%) Amean fault-both-24 16438.00 ( 0.00%) 14618.59 ( 11.07%) Amean fault-both-30 17531.72 ( 0.00%) 16650.96 ( 5.02%) Amean fault-both-32 17101.96 ( 0.00%) 17145.15 ( -0.25%) Link: http://lkml.kernel.org/r/20190118175136.31341-18-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:24 +00:00
cond_resched();
mm, compaction: properly signal and act upon lock and need_sched() contention Compaction uses compact_checklock_irqsave() function to periodically check for lock contention and need_resched() to either abort async compaction, or to free the lock, schedule and retake the lock. When aborting, cc->contended is set to signal the contended state to the caller. Two problems have been identified in this mechanism. First, compaction also calls directly cond_resched() in both scanners when no lock is yet taken. This call either does not abort async compaction, or set cc->contended appropriately. This patch introduces a new compact_should_abort() function to achieve both. In isolate_freepages(), the check frequency is reduced to once by SWAP_CLUSTER_MAX pageblocks to match what the migration scanner does in the preliminary page checks. In case a pageblock is found suitable for calling isolate_freepages_block(), the checks within there are done on higher frequency. Second, isolate_freepages() does not check if isolate_freepages_block() aborted due to contention, and advances to the next pageblock. This violates the principle of aborting on contention, and might result in pageblocks not being scanned completely, since the scanning cursor is advanced. This problem has been noticed in the code by Joonsoo Kim when reviewing related patches. This patch makes isolate_freepages_block() check the cc->contended flag and abort. In case isolate_freepages() has already isolated some pages before aborting due to contention, page migration will proceed, which is OK since we do not want to waste the work that has been done, and page migration has own checks for contention. However, we do not want another isolation attempt by either of the scanners, so cc->contended flag check is added also to compaction_alloc() and compact_finished() to make sure compaction is aborted right after the migration. The outcome of the patch should be reduced lock contention by async compaction and lower latencies for higher-order allocations where direct compaction is involved. [akpm@linux-foundation.org: fix typo in comment] Reported-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Tested-by: Shawn Guo <shawn.guo@linaro.org> Tested-by: Kevin Hilman <khilman@linaro.org> Tested-by: Stephen Warren <swarren@nvidia.com> Tested-by: Fabio Estevam <fabio.estevam@freescale.com> Cc: David Rientjes <rientjes@google.com> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:10:41 +00:00
return false;
}
/*
* Isolate free pages onto a private freelist. If @strict is true, will abort
* returning 0 on any invalid PFNs or non-free pages inside of the pageblock
* (even though it may still end up isolating some pages).
*/
static unsigned long isolate_freepages_block(struct compact_control *cc,
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
unsigned long *start_pfn,
unsigned long end_pfn,
struct list_head *freelist,
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
unsigned int stride,
bool strict)
{
int nr_scanned = 0, total_isolated = 0;
mm, compaction: reduce premature advancement of the migration target scanner The fast isolation of free pages allows the cached PFN of the free scanner to advance faster than necessary depending on the contents of the free list. The key is that fast_isolate_freepages() can update zone->compact_cached_free_pfn via isolate_freepages_block(). When the fast search fails, the linear scan can start from a point that has skipped valid migration targets, particularly pageblocks with just low-order free pages. This can cause the migration source/target scanners to meet prematurely causing a reset. This patch starts by avoiding an update of the pageblock skip information and cached PFN from isolate_freepages_block() and puts the responsibility of updating that information in the callers. The fast scanner will update the cached PFN if and only if it finds a block that is higher than the existing cached PFN and sets the skip if the pageblock is full or nearly full. The linear scanner will update skipped information and the cached PFN only when a block is completely scanned. The total impact is that the free scanner advances more slowly as it is primarily driven by the linear scanner instead of the fast search. 5.0.0-rc1 5.0.0-rc1 noresched-v3r17 slowfree-v3r17 Amean fault-both-3 2965.68 ( 0.00%) 3036.75 ( -2.40%) Amean fault-both-5 3995.90 ( 0.00%) 4522.24 * -13.17%* Amean fault-both-7 5842.12 ( 0.00%) 6365.35 ( -8.96%) Amean fault-both-12 9550.87 ( 0.00%) 10340.93 ( -8.27%) Amean fault-both-18 13304.72 ( 0.00%) 14732.46 ( -10.73%) Amean fault-both-24 14618.59 ( 0.00%) 16288.96 ( -11.43%) Amean fault-both-30 16650.96 ( 0.00%) 16346.21 ( 1.83%) Amean fault-both-32 17145.15 ( 0.00%) 19317.49 ( -12.67%) The impact to latency is higher than the last version but it appears to be due to a slight increase in the free scan rates which is a potential side-effect of the patch. However, this is necessary for later patches that are more careful about how pageblocks are treated as earlier iterations of those patches hit corner cases where the restarts were punishing and very visible. Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:28 +00:00
struct page *cursor;
unsigned long flags = 0;
bool locked = false;
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
unsigned long blockpfn = *start_pfn;
unsigned int order;
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
/* Strict mode is for isolation, speed is secondary */
if (strict)
stride = 1;
cursor = pfn_to_page(blockpfn);
/* Isolate free pages. */
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
for (; blockpfn < end_pfn; blockpfn += stride, cursor += stride) {
int isolated;
struct page *page = cursor;
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
/*
* Periodically drop the lock (if held) regardless of its
* contention, to give chance to IRQs. Abort if fatal signal
* pending or async compaction detects need_resched()
*/
if (!(blockpfn % SWAP_CLUSTER_MAX)
&& compact_unlock_should_abort(&cc->zone->lock, flags,
&locked, cc))
break;
nr_scanned++;
if (!pfn_valid_within(blockpfn))
mm/compaction: break out of loop on !PageBuddy in isolate_freepages_block We received several reports of bad page state when freeing CMA pages previously allocated with alloc_contig_range: BUG: Bad page state in process Binder_A pfn:63202 page:d21130b0 count:0 mapcount:1 mapping: (null) index:0x7dfbf page flags: 0x40080068(uptodate|lru|active|swapbacked) Based on the page state, it looks like the page was still in use. The page flags do not make sense for the use case though. Further debugging showed that despite alloc_contig_range returning success, at least one page in the range still remained in the buddy allocator. There is an issue with isolate_freepages_block. In strict mode (which CMA uses), if any pages in the range cannot be isolated, isolate_freepages_block should return failure 0. The current check keeps track of the total number of isolated pages and compares against the size of the range: if (strict && nr_strict_required > total_isolated) total_isolated = 0; After taking the zone lock, if one of the pages in the range is not in the buddy allocator, we continue through the loop and do not increment total_isolated. If in the last iteration of the loop we isolate more than one page (e.g. last page needed is a higher order page), the check for total_isolated may pass and we fail to detect that a page was skipped. The fix is to bail out if the loop immediately if we are in strict mode. There's no benfit to continuing anyway since we need all pages to be isolated. Additionally, drop the error checking based on nr_strict_required and just check the pfn ranges. This matches with what isolate_freepages_range does. Signed-off-by: Laura Abbott <lauraa@codeaurora.org> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-03-10 22:49:44 +00:00
goto isolate_fail;
/*
* For compound pages such as THP and hugetlbfs, we can save
* potentially a lot of iterations if we skip them at once.
* The check is racy, but we can consider only valid values
* and the only danger is skipping too much.
*/
if (PageCompound(page)) {
const unsigned int order = compound_order(page);
if (likely(order < MAX_ORDER)) {
blockpfn += (1UL << order) - 1;
cursor += (1UL << order) - 1;
}
goto isolate_fail;
}
if (!PageBuddy(page))
mm/compaction: break out of loop on !PageBuddy in isolate_freepages_block We received several reports of bad page state when freeing CMA pages previously allocated with alloc_contig_range: BUG: Bad page state in process Binder_A pfn:63202 page:d21130b0 count:0 mapcount:1 mapping: (null) index:0x7dfbf page flags: 0x40080068(uptodate|lru|active|swapbacked) Based on the page state, it looks like the page was still in use. The page flags do not make sense for the use case though. Further debugging showed that despite alloc_contig_range returning success, at least one page in the range still remained in the buddy allocator. There is an issue with isolate_freepages_block. In strict mode (which CMA uses), if any pages in the range cannot be isolated, isolate_freepages_block should return failure 0. The current check keeps track of the total number of isolated pages and compares against the size of the range: if (strict && nr_strict_required > total_isolated) total_isolated = 0; After taking the zone lock, if one of the pages in the range is not in the buddy allocator, we continue through the loop and do not increment total_isolated. If in the last iteration of the loop we isolate more than one page (e.g. last page needed is a higher order page), the check for total_isolated may pass and we fail to detect that a page was skipped. The fix is to bail out if the loop immediately if we are in strict mode. There's no benfit to continuing anyway since we need all pages to be isolated. Additionally, drop the error checking based on nr_strict_required and just check the pfn ranges. This matches with what isolate_freepages_range does. Signed-off-by: Laura Abbott <lauraa@codeaurora.org> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-03-10 22:49:44 +00:00
goto isolate_fail;
/*
* If we already hold the lock, we can skip some rechecking.
* Note that if we hold the lock now, checked_pageblock was
* already set in some previous iteration (or strict is true),
* so it is correct to skip the suitable migration target
* recheck as well.
*/
if (!locked) {
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
locked = compact_lock_irqsave(&cc->zone->lock,
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
&flags, cc);
/* Recheck this is a buddy page under lock */
if (!PageBuddy(page))
goto isolate_fail;
}
/* Found a free page, will break it into order-0 pages */
order = buddy_order(page);
isolated = __isolate_free_page(page, order);
mm, compaction: abort free scanner if split fails If the memory compaction free scanner cannot successfully split a free page (only possible due to per-zone low watermark), terminate the free scanner rather than continuing to scan memory needlessly. If the watermark is insufficient for a free page of order <= cc->order, then terminate the scanner since all future splits will also likely fail. This prevents the compaction freeing scanner from scanning all memory on very large zones (very noticeable for zones > 128GB, for instance) when all splits will likely fail while holding zone->lock. compaction_alloc() iterating a 128GB zone has been benchmarked to take over 400ms on some systems whereas any free page isolated and ready to be split ends up failing in split_free_page() because of the low watermark check and thus the iteration continues. The next time compaction occurs, the freeing scanner will likely start at the end of the zone again since no success was made previously and we get the same lengthy iteration until the zone is brought above the low watermark. All thp page faults can take >400ms in such a state without this fix. Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1606211820350.97086@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Hugh Dickins <hughd@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-06-24 21:50:10 +00:00
if (!isolated)
break;
set_page_private(page, order);
mm, compaction: abort free scanner if split fails If the memory compaction free scanner cannot successfully split a free page (only possible due to per-zone low watermark), terminate the free scanner rather than continuing to scan memory needlessly. If the watermark is insufficient for a free page of order <= cc->order, then terminate the scanner since all future splits will also likely fail. This prevents the compaction freeing scanner from scanning all memory on very large zones (very noticeable for zones > 128GB, for instance) when all splits will likely fail while holding zone->lock. compaction_alloc() iterating a 128GB zone has been benchmarked to take over 400ms on some systems whereas any free page isolated and ready to be split ends up failing in split_free_page() because of the low watermark check and thus the iteration continues. The next time compaction occurs, the freeing scanner will likely start at the end of the zone again since no success was made previously and we get the same lengthy iteration until the zone is brought above the low watermark. All thp page faults can take >400ms in such a state without this fix. Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1606211820350.97086@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Hugh Dickins <hughd@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-06-24 21:50:10 +00:00
total_isolated += isolated;
mm, compaction: abort free scanner if split fails If the memory compaction free scanner cannot successfully split a free page (only possible due to per-zone low watermark), terminate the free scanner rather than continuing to scan memory needlessly. If the watermark is insufficient for a free page of order <= cc->order, then terminate the scanner since all future splits will also likely fail. This prevents the compaction freeing scanner from scanning all memory on very large zones (very noticeable for zones > 128GB, for instance) when all splits will likely fail while holding zone->lock. compaction_alloc() iterating a 128GB zone has been benchmarked to take over 400ms on some systems whereas any free page isolated and ready to be split ends up failing in split_free_page() because of the low watermark check and thus the iteration continues. The next time compaction occurs, the freeing scanner will likely start at the end of the zone again since no success was made previously and we get the same lengthy iteration until the zone is brought above the low watermark. All thp page faults can take >400ms in such a state without this fix. Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1606211820350.97086@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Hugh Dickins <hughd@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-06-24 21:50:10 +00:00
cc->nr_freepages += isolated;
list_add_tail(&page->lru, freelist);
mm, compaction: abort free scanner if split fails If the memory compaction free scanner cannot successfully split a free page (only possible due to per-zone low watermark), terminate the free scanner rather than continuing to scan memory needlessly. If the watermark is insufficient for a free page of order <= cc->order, then terminate the scanner since all future splits will also likely fail. This prevents the compaction freeing scanner from scanning all memory on very large zones (very noticeable for zones > 128GB, for instance) when all splits will likely fail while holding zone->lock. compaction_alloc() iterating a 128GB zone has been benchmarked to take over 400ms on some systems whereas any free page isolated and ready to be split ends up failing in split_free_page() because of the low watermark check and thus the iteration continues. The next time compaction occurs, the freeing scanner will likely start at the end of the zone again since no success was made previously and we get the same lengthy iteration until the zone is brought above the low watermark. All thp page faults can take >400ms in such a state without this fix. Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1606211820350.97086@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Hugh Dickins <hughd@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-06-24 21:50:10 +00:00
if (!strict && cc->nr_migratepages <= cc->nr_freepages) {
blockpfn += isolated;
break;
}
mm, compaction: abort free scanner if split fails If the memory compaction free scanner cannot successfully split a free page (only possible due to per-zone low watermark), terminate the free scanner rather than continuing to scan memory needlessly. If the watermark is insufficient for a free page of order <= cc->order, then terminate the scanner since all future splits will also likely fail. This prevents the compaction freeing scanner from scanning all memory on very large zones (very noticeable for zones > 128GB, for instance) when all splits will likely fail while holding zone->lock. compaction_alloc() iterating a 128GB zone has been benchmarked to take over 400ms on some systems whereas any free page isolated and ready to be split ends up failing in split_free_page() because of the low watermark check and thus the iteration continues. The next time compaction occurs, the freeing scanner will likely start at the end of the zone again since no success was made previously and we get the same lengthy iteration until the zone is brought above the low watermark. All thp page faults can take >400ms in such a state without this fix. Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1606211820350.97086@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Hugh Dickins <hughd@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-06-24 21:50:10 +00:00
/* Advance to the end of split page */
blockpfn += isolated - 1;
cursor += isolated - 1;
continue;
mm/compaction: break out of loop on !PageBuddy in isolate_freepages_block We received several reports of bad page state when freeing CMA pages previously allocated with alloc_contig_range: BUG: Bad page state in process Binder_A pfn:63202 page:d21130b0 count:0 mapcount:1 mapping: (null) index:0x7dfbf page flags: 0x40080068(uptodate|lru|active|swapbacked) Based on the page state, it looks like the page was still in use. The page flags do not make sense for the use case though. Further debugging showed that despite alloc_contig_range returning success, at least one page in the range still remained in the buddy allocator. There is an issue with isolate_freepages_block. In strict mode (which CMA uses), if any pages in the range cannot be isolated, isolate_freepages_block should return failure 0. The current check keeps track of the total number of isolated pages and compares against the size of the range: if (strict && nr_strict_required > total_isolated) total_isolated = 0; After taking the zone lock, if one of the pages in the range is not in the buddy allocator, we continue through the loop and do not increment total_isolated. If in the last iteration of the loop we isolate more than one page (e.g. last page needed is a higher order page), the check for total_isolated may pass and we fail to detect that a page was skipped. The fix is to bail out if the loop immediately if we are in strict mode. There's no benfit to continuing anyway since we need all pages to be isolated. Additionally, drop the error checking based on nr_strict_required and just check the pfn ranges. This matches with what isolate_freepages_range does. Signed-off-by: Laura Abbott <lauraa@codeaurora.org> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-03-10 22:49:44 +00:00
isolate_fail:
if (strict)
break;
else
continue;
}
mm, compaction: abort free scanner if split fails If the memory compaction free scanner cannot successfully split a free page (only possible due to per-zone low watermark), terminate the free scanner rather than continuing to scan memory needlessly. If the watermark is insufficient for a free page of order <= cc->order, then terminate the scanner since all future splits will also likely fail. This prevents the compaction freeing scanner from scanning all memory on very large zones (very noticeable for zones > 128GB, for instance) when all splits will likely fail while holding zone->lock. compaction_alloc() iterating a 128GB zone has been benchmarked to take over 400ms on some systems whereas any free page isolated and ready to be split ends up failing in split_free_page() because of the low watermark check and thus the iteration continues. The next time compaction occurs, the freeing scanner will likely start at the end of the zone again since no success was made previously and we get the same lengthy iteration until the zone is brought above the low watermark. All thp page faults can take >400ms in such a state without this fix. Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1606211820350.97086@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Hugh Dickins <hughd@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-06-24 21:50:10 +00:00
if (locked)
spin_unlock_irqrestore(&cc->zone->lock, flags);
/*
* There is a tiny chance that we have read bogus compound_order(),
* so be careful to not go outside of the pageblock.
*/
if (unlikely(blockpfn > end_pfn))
blockpfn = end_pfn;
trace_mm_compaction_isolate_freepages(*start_pfn, blockpfn,
nr_scanned, total_isolated);
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
/* Record how far we have got within the block */
*start_pfn = blockpfn;
/*
* If strict isolation is requested by CMA then check that all the
* pages requested were isolated. If there were any failures, 0 is
* returned and CMA will fail.
*/
mm/compaction: break out of loop on !PageBuddy in isolate_freepages_block We received several reports of bad page state when freeing CMA pages previously allocated with alloc_contig_range: BUG: Bad page state in process Binder_A pfn:63202 page:d21130b0 count:0 mapcount:1 mapping: (null) index:0x7dfbf page flags: 0x40080068(uptodate|lru|active|swapbacked) Based on the page state, it looks like the page was still in use. The page flags do not make sense for the use case though. Further debugging showed that despite alloc_contig_range returning success, at least one page in the range still remained in the buddy allocator. There is an issue with isolate_freepages_block. In strict mode (which CMA uses), if any pages in the range cannot be isolated, isolate_freepages_block should return failure 0. The current check keeps track of the total number of isolated pages and compares against the size of the range: if (strict && nr_strict_required > total_isolated) total_isolated = 0; After taking the zone lock, if one of the pages in the range is not in the buddy allocator, we continue through the loop and do not increment total_isolated. If in the last iteration of the loop we isolate more than one page (e.g. last page needed is a higher order page), the check for total_isolated may pass and we fail to detect that a page was skipped. The fix is to bail out if the loop immediately if we are in strict mode. There's no benfit to continuing anyway since we need all pages to be isolated. Additionally, drop the error checking based on nr_strict_required and just check the pfn ranges. This matches with what isolate_freepages_range does. Signed-off-by: Laura Abbott <lauraa@codeaurora.org> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-03-10 22:49:44 +00:00
if (strict && blockpfn < end_pfn)
total_isolated = 0;
cc->total_free_scanned += nr_scanned;
mm: compaction: Add scanned and isolated counters for compaction Compaction already has tracepoints to count scanned and isolated pages but it requires that ftrace be enabled and if that information has to be written to disk then it can be disruptive. This patch adds vmstat counters for compaction called compact_migrate_scanned, compact_free_scanned and compact_isolated. With these counters, it is possible to define a basic cost model for compaction. This approximates of how much work compaction is doing and can be compared that with an oprofile showing TLB misses and see if the cost of compaction is being offset by THP for example. Minimally a compaction patch can be evaluated in terms of whether it increases or decreases cost. The basic cost model looks like this Fundamental unit u: a word sizeof(void *) Ca = cost of struct page access = sizeof(struct page) / u Cmc = Cost migrate page copy = (Ca + PAGE_SIZE/u) * 2 Cmf = Cost migrate failure = Ca * 2 Ci = Cost page isolation = (Ca + Wi) where Wi is a constant that should reflect the approximate cost of the locking operation. Csm = Cost migrate scanning = Ca Csf = Cost free scanning = Ca Overall cost = (Csm * compact_migrate_scanned) + (Csf * compact_free_scanned) + (Ci * compact_isolated) + (Cmc * pgmigrate_success) + (Cmf * pgmigrate_failed) Where the values are read from /proc/vmstat. This is very basic and ignores certain costs such as the allocation cost to do a migrate page copy but any improvement to the model would still use the same vmstat counters. Signed-off-by: Mel Gorman <mgorman@suse.de> Reviewed-by: Rik van Riel <riel@redhat.com>
2012-10-19 11:00:10 +00:00
if (total_isolated)
count_compact_events(COMPACTISOLATED, total_isolated);
return total_isolated;
}
/**
* isolate_freepages_range() - isolate free pages.
* @cc: Compaction control structure.
* @start_pfn: The first PFN to start isolating.
* @end_pfn: The one-past-last PFN.
*
* Non-free pages, invalid PFNs, or zone boundaries within the
* [start_pfn, end_pfn) range are considered errors, cause function to
* undo its actions and return zero.
*
* Otherwise, function returns one-past-the-last PFN of isolated page
* (which may be greater then end_pfn if end fell in a middle of
* a free page).
*/
unsigned long
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
isolate_freepages_range(struct compact_control *cc,
unsigned long start_pfn, unsigned long end_pfn)
{
unsigned long isolated, pfn, block_start_pfn, block_end_pfn;
LIST_HEAD(freelist);
mm, compaction: reduce zone checking frequency in the migration scanner The unification of the migrate and free scanner families of function has highlighted a difference in how the scanners ensure they only isolate pages of the intended zone. This is important for taking zone lock or lru lock of the correct zone. Due to nodes overlapping, it is however possible to encounter a different zone within the range of the zone being compacted. The free scanner, since its inception by commit 748446bb6b5a ("mm: compaction: memory compaction core"), has been checking the zone of the first valid page in a pageblock, and skipping the whole pageblock if the zone does not match. This checking was completely missing from the migration scanner at first, and later added by commit dc9086004b3d ("mm: compaction: check for overlapping nodes during isolation for migration") in a reaction to a bug report. But the zone comparison in migration scanner is done once per a single scanned page, which is more defensive and thus more costly than a check per pageblock. This patch unifies the checking done in both scanners to once per pageblock, through a new pageblock_pfn_to_page() function, which also includes pfn_valid() checks. It is more defensive than the current free scanner checks, as it checks both the first and last page of the pageblock, but less defensive by the migration scanner per-page checks. It assumes that node overlapping may result (on some architecture) in a boundary between two nodes falling into the middle of a pageblock, but that there cannot be a node0 node1 node0 interleaving within a single pageblock. The result is more code being shared and a bit less per-page CPU cost in the migration scanner. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:11 +00:00
pfn = start_pfn;
block_start_pfn = pageblock_start_pfn(pfn);
if (block_start_pfn < cc->zone->zone_start_pfn)
block_start_pfn = cc->zone->zone_start_pfn;
block_end_pfn = pageblock_end_pfn(pfn);
mm, compaction: reduce zone checking frequency in the migration scanner The unification of the migrate and free scanner families of function has highlighted a difference in how the scanners ensure they only isolate pages of the intended zone. This is important for taking zone lock or lru lock of the correct zone. Due to nodes overlapping, it is however possible to encounter a different zone within the range of the zone being compacted. The free scanner, since its inception by commit 748446bb6b5a ("mm: compaction: memory compaction core"), has been checking the zone of the first valid page in a pageblock, and skipping the whole pageblock if the zone does not match. This checking was completely missing from the migration scanner at first, and later added by commit dc9086004b3d ("mm: compaction: check for overlapping nodes during isolation for migration") in a reaction to a bug report. But the zone comparison in migration scanner is done once per a single scanned page, which is more defensive and thus more costly than a check per pageblock. This patch unifies the checking done in both scanners to once per pageblock, through a new pageblock_pfn_to_page() function, which also includes pfn_valid() checks. It is more defensive than the current free scanner checks, as it checks both the first and last page of the pageblock, but less defensive by the migration scanner per-page checks. It assumes that node overlapping may result (on some architecture) in a boundary between two nodes falling into the middle of a pageblock, but that there cannot be a node0 node1 node0 interleaving within a single pageblock. The result is more code being shared and a bit less per-page CPU cost in the migration scanner. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:11 +00:00
for (; pfn < end_pfn; pfn += isolated,
block_start_pfn = block_end_pfn,
mm, compaction: reduce zone checking frequency in the migration scanner The unification of the migrate and free scanner families of function has highlighted a difference in how the scanners ensure they only isolate pages of the intended zone. This is important for taking zone lock or lru lock of the correct zone. Due to nodes overlapping, it is however possible to encounter a different zone within the range of the zone being compacted. The free scanner, since its inception by commit 748446bb6b5a ("mm: compaction: memory compaction core"), has been checking the zone of the first valid page in a pageblock, and skipping the whole pageblock if the zone does not match. This checking was completely missing from the migration scanner at first, and later added by commit dc9086004b3d ("mm: compaction: check for overlapping nodes during isolation for migration") in a reaction to a bug report. But the zone comparison in migration scanner is done once per a single scanned page, which is more defensive and thus more costly than a check per pageblock. This patch unifies the checking done in both scanners to once per pageblock, through a new pageblock_pfn_to_page() function, which also includes pfn_valid() checks. It is more defensive than the current free scanner checks, as it checks both the first and last page of the pageblock, but less defensive by the migration scanner per-page checks. It assumes that node overlapping may result (on some architecture) in a boundary between two nodes falling into the middle of a pageblock, but that there cannot be a node0 node1 node0 interleaving within a single pageblock. The result is more code being shared and a bit less per-page CPU cost in the migration scanner. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:11 +00:00
block_end_pfn += pageblock_nr_pages) {
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
/* Protect pfn from changing by isolate_freepages_block */
unsigned long isolate_start_pfn = pfn;
block_end_pfn = min(block_end_pfn, end_pfn);
/*
* pfn could pass the block_end_pfn if isolated freepage
* is more than pageblock order. In this case, we adjust
* scanning range to right one.
*/
if (pfn >= block_end_pfn) {
block_start_pfn = pageblock_start_pfn(pfn);
block_end_pfn = pageblock_end_pfn(pfn);
block_end_pfn = min(block_end_pfn, end_pfn);
}
if (!pageblock_pfn_to_page(block_start_pfn,
block_end_pfn, cc->zone))
mm, compaction: reduce zone checking frequency in the migration scanner The unification of the migrate and free scanner families of function has highlighted a difference in how the scanners ensure they only isolate pages of the intended zone. This is important for taking zone lock or lru lock of the correct zone. Due to nodes overlapping, it is however possible to encounter a different zone within the range of the zone being compacted. The free scanner, since its inception by commit 748446bb6b5a ("mm: compaction: memory compaction core"), has been checking the zone of the first valid page in a pageblock, and skipping the whole pageblock if the zone does not match. This checking was completely missing from the migration scanner at first, and later added by commit dc9086004b3d ("mm: compaction: check for overlapping nodes during isolation for migration") in a reaction to a bug report. But the zone comparison in migration scanner is done once per a single scanned page, which is more defensive and thus more costly than a check per pageblock. This patch unifies the checking done in both scanners to once per pageblock, through a new pageblock_pfn_to_page() function, which also includes pfn_valid() checks. It is more defensive than the current free scanner checks, as it checks both the first and last page of the pageblock, but less defensive by the migration scanner per-page checks. It assumes that node overlapping may result (on some architecture) in a boundary between two nodes falling into the middle of a pageblock, but that there cannot be a node0 node1 node0 interleaving within a single pageblock. The result is more code being shared and a bit less per-page CPU cost in the migration scanner. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:11 +00:00
break;
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
isolated = isolate_freepages_block(cc, &isolate_start_pfn,
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
block_end_pfn, &freelist, 0, true);
/*
* In strict mode, isolate_freepages_block() returns 0 if
* there are any holes in the block (ie. invalid PFNs or
* non-free pages).
*/
if (!isolated)
break;
/*
* If we managed to isolate pages, it is always (1 << n) *
* pageblock_nr_pages for some non-negative n. (Max order
* page may span two pageblocks).
*/
}
/* __isolate_free_page() does not map the pages */
split_map_pages(&freelist);
if (pfn < end_pfn) {
/* Loop terminated early, cleanup. */
release_freepages(&freelist);
return 0;
}
/* We don't use freelists for anything. */
return pfn;
}
/* Similar to reclaim, but different enough that they don't share logic */
static bool too_many_isolated(pg_data_t *pgdat)
{
unsigned long active, inactive, isolated;
inactive = node_page_state(pgdat, NR_INACTIVE_FILE) +
node_page_state(pgdat, NR_INACTIVE_ANON);
active = node_page_state(pgdat, NR_ACTIVE_FILE) +
node_page_state(pgdat, NR_ACTIVE_ANON);
isolated = node_page_state(pgdat, NR_ISOLATED_FILE) +
node_page_state(pgdat, NR_ISOLATED_ANON);
return isolated > (inactive + active) / 2;
}
/**
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
* isolate_migratepages_block() - isolate all migrate-able pages within
* a single pageblock
* @cc: Compaction control structure.
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
* @low_pfn: The first PFN to isolate
* @end_pfn: The one-past-the-last PFN to isolate, within same pageblock
* @isolate_mode: Isolation mode to be used.
*
* Isolate all pages that can be migrated from the range specified by
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
* [low_pfn, end_pfn). The range is expected to be within same pageblock.
* Returns errno, like -EAGAIN or -EINTR in case e.g signal pending or congestion,
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
* -ENOMEM in case we could not allocate a page, or 0.
* cc->migrate_pfn will contain the next pfn to scan.
*
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
* The pages are isolated on cc->migratepages list (not required to be empty),
* and cc->nr_migratepages is updated accordingly.
*/
static int
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
unsigned long end_pfn, isolate_mode_t isolate_mode)
{
pg_data_t *pgdat = cc->zone->zone_pgdat;
unsigned long nr_scanned = 0, nr_isolated = 0;
struct lruvec *lruvec;
unsigned long flags = 0;
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
struct lruvec *locked = NULL;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
struct page *page = NULL, *valid_page = NULL;
unsigned long start_pfn = low_pfn;
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
bool skip_on_failure = false;
unsigned long next_skip_pfn = 0;
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
bool skip_updated = false;
int ret = 0;
cc->migrate_pfn = low_pfn;
/*
* Ensure that there are not too many pages isolated from the LRU
* list by either parallel reclaimers or compaction. If there are,
* delay for some time until fewer pages are isolated
*/
while (unlikely(too_many_isolated(pgdat))) {
/* stop isolation if there are still pages not migrated */
if (cc->nr_migratepages)
return -EAGAIN;
/* async migration should just abort */
if (cc->mode == MIGRATE_ASYNC)
return -EAGAIN;
congestion_wait(BLK_RW_ASYNC, HZ/10);
if (fatal_signal_pending(current))
return -EINTR;
}
mm, compaction: do not consider a need to reschedule as contention Scanning on large machines can take a considerable length of time and eventually need to be rescheduled. This is treated as an abort event but that's not appropriate as the attempt is likely to be retried after making numerous checks and taking another cycle through the page allocator. This patch will check the need to reschedule if necessary but continue the scanning. The main benefit is reduced scanning when compaction is taking a long time or the machine is over-saturated. It also avoids an unnecessary exit of compaction that ends up being retried by the page allocator in the outer loop. 5.0.0-rc1 5.0.0-rc1 synccached-v3r16 noresched-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2958.27 ( 0.00%) 2965.68 ( -0.25%) Amean fault-both-5 4091.90 ( 0.00%) 3995.90 ( 2.35%) Amean fault-both-7 5803.05 ( 0.00%) 5842.12 ( -0.67%) Amean fault-both-12 9481.06 ( 0.00%) 9550.87 ( -0.74%) Amean fault-both-18 14141.51 ( 0.00%) 13304.72 ( 5.92%) Amean fault-both-24 16438.00 ( 0.00%) 14618.59 ( 11.07%) Amean fault-both-30 17531.72 ( 0.00%) 16650.96 ( 5.02%) Amean fault-both-32 17101.96 ( 0.00%) 17145.15 ( -0.25%) Link: http://lkml.kernel.org/r/20190118175136.31341-18-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:24 +00:00
cond_resched();
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
if (cc->direct_compaction && (cc->mode == MIGRATE_ASYNC)) {
skip_on_failure = true;
next_skip_pfn = block_end_pfn(low_pfn, cc->order);
}
/* Time to isolate some pages for migration */
for (; low_pfn < end_pfn; low_pfn++) {
mm, compaction: always skip all compound pages by order in migrate scanner The compaction migrate scanner tries to skip THP pages by their order, to reduce number of iterations for pages it cannot isolate. The check is only done if PageLRU() is true, which means it applies to THP pages, but not e.g. hugetlbfs pages or any other non-LRU compound pages, which we have to iterate by base pages. This limitation comes from the assumption that it's only safe to read compound_order() when we have the zone's lru_lock and THP cannot be split under us. But the only danger (after filtering out order values that are not below MAX_ORDER, to prevent overflows) is that we skip too much or too little after reading a bogus compound_order() due to a rare race. This is the same reasoning as patch 99c0fd5e51c4 ("mm, compaction: skip buddy pages by their order in the migrate scanner") introduced for unsafely reading PageBuddy() order. After this patch, all pages are tested for PageCompound() and we skip them by compound_order(). The test is done after the test for balloon_page_movable() as we don't want to assume if balloon pages (or other pages with own isolation and migration implementation if a generic API gets implemented) are compound or not. When tested with stress-highalloc from mmtests on 4GB system with 1GB hugetlbfs pages, the vmstat compact_migrate_scanned count decreased by 15%. [kirill.shutemov@linux.intel.com: change PageTransHuge checks to PageCompound for different series was squashed here] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:46 +00:00
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
if (skip_on_failure && low_pfn >= next_skip_pfn) {
/*
* We have isolated all migration candidates in the
* previous order-aligned block, and did not skip it due
* to failure. We should migrate the pages now and
* hopefully succeed compaction.
*/
if (nr_isolated)
break;
/*
* We failed to isolate in the previous order-aligned
* block. Set the new boundary to the end of the
* current block. Note we can't simply increase
* next_skip_pfn by 1 << order, as low_pfn might have
* been incremented by a higher number due to skipping
* a compound or a high-order buddy page in the
* previous loop iteration.
*/
next_skip_pfn = block_end_pfn(low_pfn, cc->order);
}
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
/*
* Periodically drop the lock (if held) regardless of its
mm: compaction: avoid 100% CPU usage during compaction when a task is killed "howaboutsynergy" reported via kernel buzilla number 204165 that compact_zone_order was consuming 100% CPU during a stress test for prolonged periods of time. Specifically the following command, which should exit in 10 seconds, was taking an excessive time to finish while the CPU was pegged at 100%. stress -m 220 --vm-bytes 1000000000 --timeout 10 Tracing indicated a pattern as follows stress-3923 [007] 519.106208: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106212: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106216: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106219: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106223: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106227: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106231: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106235: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106238: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106242: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 Note that compaction is entered in rapid succession while scanning and isolating nothing. The problem is that when a task that is compacting receives a fatal signal, it retries indefinitely instead of exiting while making no progress as a fatal signal is pending. It's not easy to trigger this condition although enabling zswap helps on the basis that the timing is altered. A very small window has to be hit for the problem to occur (signal delivered while compacting and isolating a PFN for migration that is not aligned to SWAP_CLUSTER_MAX). This was reproduced locally -- 16G single socket system, 8G swap, 30% zswap configured, vm-bytes 22000000000 using Colin Kings stress-ng implementation from github running in a loop until the problem hits). Tracing recorded the problem occurring almost 200K times in a short window. With this patch, the problem hit 4 times but the task existed normally instead of consuming CPU. This problem has existed for some time but it was made worse by commit cf66f0700c8f ("mm, compaction: do not consider a need to reschedule as contention"). Before that commit, if the same condition was hit then locks would be quickly contended and compaction would exit that way. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=204165 Link: http://lkml.kernel.org/r/20190718085708.GE24383@techsingularity.net Fixes: cf66f0700c8f ("mm, compaction: do not consider a need to reschedule as contention") Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> [5.1+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-08-03 04:48:51 +00:00
* contention, to give chance to IRQs. Abort completely if
* a fatal signal is pending.
mm, compaction: periodically drop lock and restore IRQs in scanners Compaction scanners regularly check for lock contention and need_resched() through the compact_checklock_irqsave() function. However, if there is no contention, the lock can be held and IRQ disabled for potentially long time. This has been addressed by commit b2eef8c0d091 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") for the migration scanner. However, the refactoring done by commit 2a1402aa044b ("mm: compaction: acquire the zone->lru_lock as late as possible") has changed the conditions so that the lock is dropped only when there's contention on the lock or need_resched() is true. Also, need_resched() is checked only when the lock is already held. The comment "give a chance to irqs before checking need_resched" is therefore misleading, as IRQs remain disabled when the check is done. This patch restores the behavior intended by commit b2eef8c0d091 and also tries to better balance and make more deterministic the time spent by checking for contention vs the time the scanners might run between the checks. It also avoids situations where checking has not been done often enough before. The result should be avoiding both too frequent and too infrequent contention checking, and especially the potentially long-running scans with IRQs disabled and no checking of need_resched() or for fatal signal pending, which can happen when many consecutive pages or pageblocks fail the preliminary tests and do not reach the later call site to compact_checklock_irqsave(), as explained below. Before the patch: In the migration scanner, compact_checklock_irqsave() was called each loop, if reached. If not reached, some lower-frequency checking could still be done if the lock was already held, but this would not result in aborting contended async compaction until reaching compact_checklock_irqsave() or end of pageblock. In the free scanner, it was similar but completely without the periodical checking, so lock can be potentially held until reaching the end of pageblock. After the patch, in both scanners: The periodical check is done as the first thing in the loop on each SWAP_CLUSTER_MAX aligned pfn, using the new compact_unlock_should_abort() function, which always unlocks the lock (if locked) and aborts async compaction if scheduling is needed. It also aborts any type of compaction when a fatal signal is pending. The compact_checklock_irqsave() function is replaced with a slightly different compact_trylock_irqsave(). The biggest difference is that the function is not called at all if the lock is already held. The periodical need_resched() checking is left solely to compact_unlock_should_abort(). The lock contention avoidance for async compaction is achieved by the periodical unlock by compact_unlock_should_abort() and by using trylock in compact_trylock_irqsave() and aborting when trylock fails. Sync compaction does not use trylock. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:16 +00:00
*/
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
if (!(low_pfn % SWAP_CLUSTER_MAX)) {
if (locked) {
unlock_page_lruvec_irqrestore(locked, flags);
locked = NULL;
}
if (fatal_signal_pending(current)) {
cc->contended = true;
ret = -EINTR;
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
goto fatal_pending;
}
cond_resched();
mm: compaction: avoid 100% CPU usage during compaction when a task is killed "howaboutsynergy" reported via kernel buzilla number 204165 that compact_zone_order was consuming 100% CPU during a stress test for prolonged periods of time. Specifically the following command, which should exit in 10 seconds, was taking an excessive time to finish while the CPU was pegged at 100%. stress -m 220 --vm-bytes 1000000000 --timeout 10 Tracing indicated a pattern as follows stress-3923 [007] 519.106208: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106212: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106216: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106219: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106223: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106227: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106231: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106235: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106238: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106242: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 Note that compaction is entered in rapid succession while scanning and isolating nothing. The problem is that when a task that is compacting receives a fatal signal, it retries indefinitely instead of exiting while making no progress as a fatal signal is pending. It's not easy to trigger this condition although enabling zswap helps on the basis that the timing is altered. A very small window has to be hit for the problem to occur (signal delivered while compacting and isolating a PFN for migration that is not aligned to SWAP_CLUSTER_MAX). This was reproduced locally -- 16G single socket system, 8G swap, 30% zswap configured, vm-bytes 22000000000 using Colin Kings stress-ng implementation from github running in a loop until the problem hits). Tracing recorded the problem occurring almost 200K times in a short window. With this patch, the problem hit 4 times but the task existed normally instead of consuming CPU. This problem has existed for some time but it was made worse by commit cf66f0700c8f ("mm, compaction: do not consider a need to reschedule as contention"). Before that commit, if the same condition was hit then locks would be quickly contended and compaction would exit that way. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=204165 Link: http://lkml.kernel.org/r/20190718085708.GE24383@techsingularity.net Fixes: cf66f0700c8f ("mm, compaction: do not consider a need to reschedule as contention") Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> [5.1+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-08-03 04:48:51 +00:00
}
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
if (!pfn_valid_within(low_pfn))
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
goto isolate_fail;
nr_scanned++;
page = pfn_to_page(low_pfn);
mm: compaction: check for overlapping nodes during isolation for migration When isolating pages for migration, migration starts at the start of a zone while the free scanner starts at the end of the zone. Migration avoids entering a new zone by never going beyond the free scanned. Unfortunately, in very rare cases nodes can overlap. When this happens, migration isolates pages without the LRU lock held, corrupting lists which will trigger errors in reclaim or during page free such as in the following oops BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 IP: [<ffffffff810f795c>] free_pcppages_bulk+0xcc/0x450 PGD 1dda554067 PUD 1e1cb58067 PMD 0 Oops: 0000 [#1] SMP CPU 37 Pid: 17088, comm: memcg_process_s Tainted: G X RIP: free_pcppages_bulk+0xcc/0x450 Process memcg_process_s (pid: 17088, threadinfo ffff881c2926e000, task ffff881c2926c0c0) Call Trace: free_hot_cold_page+0x17e/0x1f0 __pagevec_free+0x90/0xb0 release_pages+0x22a/0x260 pagevec_lru_move_fn+0xf3/0x110 putback_lru_page+0x66/0xe0 unmap_and_move+0x156/0x180 migrate_pages+0x9e/0x1b0 compact_zone+0x1f3/0x2f0 compact_zone_order+0xa2/0xe0 try_to_compact_pages+0xdf/0x110 __alloc_pages_direct_compact+0xee/0x1c0 __alloc_pages_slowpath+0x370/0x830 __alloc_pages_nodemask+0x1b1/0x1c0 alloc_pages_vma+0x9b/0x160 do_huge_pmd_anonymous_page+0x160/0x270 do_page_fault+0x207/0x4c0 page_fault+0x25/0x30 The "X" in the taint flag means that external modules were loaded but but is unrelated to the bug triggering. The real problem was because the PFN layout looks like this Zone PFN ranges: DMA 0x00000010 -> 0x00001000 DMA32 0x00001000 -> 0x00100000 Normal 0x00100000 -> 0x01e80000 Movable zone start PFN for each node early_node_map[14] active PFN ranges 0: 0x00000010 -> 0x0000009b 0: 0x00000100 -> 0x0007a1ec 0: 0x0007a354 -> 0x0007a379 0: 0x0007f7ff -> 0x0007f800 0: 0x00100000 -> 0x00680000 1: 0x00680000 -> 0x00e80000 0: 0x00e80000 -> 0x01080000 1: 0x01080000 -> 0x01280000 0: 0x01280000 -> 0x01480000 1: 0x01480000 -> 0x01680000 0: 0x01680000 -> 0x01880000 1: 0x01880000 -> 0x01a80000 0: 0x01a80000 -> 0x01c80000 1: 0x01c80000 -> 0x01e80000 The fix is straight-forward. isolate_migratepages() has to make a similar check to isolate_freepage to ensure that it never isolates pages from a zone it does not hold the LRU lock for. This was discovered in a 3.0-based kernel but it affects 3.1.x, 3.2.x and current mainline. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-02-09 01:13:38 +00:00
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
/*
* Check if the pageblock has already been marked skipped.
* Only the aligned PFN is checked as the caller isolates
* COMPACT_CLUSTER_MAX at a time so the second call must
* not falsely conclude that the block should be skipped.
*/
if (!valid_page && IS_ALIGNED(low_pfn, pageblock_nr_pages)) {
if (!cc->ignore_skip_hint && get_pageblock_skip(page)) {
low_pfn = end_pfn;
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
page = NULL;
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
goto isolate_abort;
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
valid_page = page;
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
if (PageHuge(page) && cc->alloc_contig) {
ret = isolate_or_dissolve_huge_page(page, &cc->migratepages);
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
/*
* Fail isolation in case isolate_or_dissolve_huge_page()
* reports an error. In case of -ENOMEM, abort right away.
*/
if (ret < 0) {
/* Do not report -EBUSY down the chain */
if (ret == -EBUSY)
ret = 0;
low_pfn += (1UL << compound_order(page)) - 1;
goto isolate_fail;
}
if (PageHuge(page)) {
/*
* Hugepage was successfully isolated and placed
* on the cc->migratepages list.
*/
low_pfn += compound_nr(page) - 1;
goto isolate_success_no_list;
}
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
/*
* Ok, the hugepage was dissolved. Now these pages are
* Buddy and cannot be re-allocated because they are
* isolated. Fall-through as the check below handles
* Buddy pages.
*/
}
/*
mm, compaction: skip buddy pages by their order in the migrate scanner The migration scanner skips PageBuddy pages, but does not consider their order as checking page_order() is generally unsafe without holding the zone->lock, and acquiring the lock just for the check wouldn't be a good tradeoff. Still, this could avoid some iterations over the rest of the buddy page, and if we are careful, the race window between PageBuddy() check and page_order() is small, and the worst thing that can happen is that we skip too much and miss some isolation candidates. This is not that bad, as compaction can already fail for many other reasons like parallel allocations, and those have much larger race window. This patch therefore makes the migration scanner obtain the buddy page order and use it to skip the whole buddy page, if the order appears to be in the valid range. It's important that the page_order() is read only once, so that the value used in the checks and in the pfn calculation is the same. But in theory the compiler can replace the local variable by multiple inlines of page_order(). Therefore, the patch introduces page_order_unsafe() that uses ACCESS_ONCE to prevent this. Testing with stress-highalloc from mmtests shows a 15% reduction in number of pages scanned by migration scanner. The reduction is >60% with __GFP_NO_KSWAPD allocations, along with success rates better by few percent. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:23 +00:00
* Skip if free. We read page order here without zone lock
* which is generally unsafe, but the race window is small and
* the worst thing that can happen is that we skip some
* potential isolation targets.
*/
mm, compaction: skip buddy pages by their order in the migrate scanner The migration scanner skips PageBuddy pages, but does not consider their order as checking page_order() is generally unsafe without holding the zone->lock, and acquiring the lock just for the check wouldn't be a good tradeoff. Still, this could avoid some iterations over the rest of the buddy page, and if we are careful, the race window between PageBuddy() check and page_order() is small, and the worst thing that can happen is that we skip too much and miss some isolation candidates. This is not that bad, as compaction can already fail for many other reasons like parallel allocations, and those have much larger race window. This patch therefore makes the migration scanner obtain the buddy page order and use it to skip the whole buddy page, if the order appears to be in the valid range. It's important that the page_order() is read only once, so that the value used in the checks and in the pfn calculation is the same. But in theory the compiler can replace the local variable by multiple inlines of page_order(). Therefore, the patch introduces page_order_unsafe() that uses ACCESS_ONCE to prevent this. Testing with stress-highalloc from mmtests shows a 15% reduction in number of pages scanned by migration scanner. The reduction is >60% with __GFP_NO_KSWAPD allocations, along with success rates better by few percent. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:23 +00:00
if (PageBuddy(page)) {
unsigned long freepage_order = buddy_order_unsafe(page);
mm, compaction: skip buddy pages by their order in the migrate scanner The migration scanner skips PageBuddy pages, but does not consider their order as checking page_order() is generally unsafe without holding the zone->lock, and acquiring the lock just for the check wouldn't be a good tradeoff. Still, this could avoid some iterations over the rest of the buddy page, and if we are careful, the race window between PageBuddy() check and page_order() is small, and the worst thing that can happen is that we skip too much and miss some isolation candidates. This is not that bad, as compaction can already fail for many other reasons like parallel allocations, and those have much larger race window. This patch therefore makes the migration scanner obtain the buddy page order and use it to skip the whole buddy page, if the order appears to be in the valid range. It's important that the page_order() is read only once, so that the value used in the checks and in the pfn calculation is the same. But in theory the compiler can replace the local variable by multiple inlines of page_order(). Therefore, the patch introduces page_order_unsafe() that uses ACCESS_ONCE to prevent this. Testing with stress-highalloc from mmtests shows a 15% reduction in number of pages scanned by migration scanner. The reduction is >60% with __GFP_NO_KSWAPD allocations, along with success rates better by few percent. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:23 +00:00
/*
* Without lock, we cannot be sure that what we got is
* a valid page order. Consider only values in the
* valid order range to prevent low_pfn overflow.
*/
if (freepage_order > 0 && freepage_order < MAX_ORDER)
low_pfn += (1UL << freepage_order) - 1;
continue;
mm, compaction: skip buddy pages by their order in the migrate scanner The migration scanner skips PageBuddy pages, but does not consider their order as checking page_order() is generally unsafe without holding the zone->lock, and acquiring the lock just for the check wouldn't be a good tradeoff. Still, this could avoid some iterations over the rest of the buddy page, and if we are careful, the race window between PageBuddy() check and page_order() is small, and the worst thing that can happen is that we skip too much and miss some isolation candidates. This is not that bad, as compaction can already fail for many other reasons like parallel allocations, and those have much larger race window. This patch therefore makes the migration scanner obtain the buddy page order and use it to skip the whole buddy page, if the order appears to be in the valid range. It's important that the page_order() is read only once, so that the value used in the checks and in the pfn calculation is the same. But in theory the compiler can replace the local variable by multiple inlines of page_order(). Therefore, the patch introduces page_order_unsafe() that uses ACCESS_ONCE to prevent this. Testing with stress-highalloc from mmtests shows a 15% reduction in number of pages scanned by migration scanner. The reduction is >60% with __GFP_NO_KSWAPD allocations, along with success rates better by few percent. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:23 +00:00
}
/*
mm, compaction: always skip all compound pages by order in migrate scanner The compaction migrate scanner tries to skip THP pages by their order, to reduce number of iterations for pages it cannot isolate. The check is only done if PageLRU() is true, which means it applies to THP pages, but not e.g. hugetlbfs pages or any other non-LRU compound pages, which we have to iterate by base pages. This limitation comes from the assumption that it's only safe to read compound_order() when we have the zone's lru_lock and THP cannot be split under us. But the only danger (after filtering out order values that are not below MAX_ORDER, to prevent overflows) is that we skip too much or too little after reading a bogus compound_order() due to a rare race. This is the same reasoning as patch 99c0fd5e51c4 ("mm, compaction: skip buddy pages by their order in the migrate scanner") introduced for unsafely reading PageBuddy() order. After this patch, all pages are tested for PageCompound() and we skip them by compound_order(). The test is done after the test for balloon_page_movable() as we don't want to assume if balloon pages (or other pages with own isolation and migration implementation if a generic API gets implemented) are compound or not. When tested with stress-highalloc from mmtests on 4GB system with 1GB hugetlbfs pages, the vmstat compact_migrate_scanned count decreased by 15%. [kirill.shutemov@linux.intel.com: change PageTransHuge checks to PageCompound for different series was squashed here] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:46 +00:00
* Regardless of being on LRU, compound pages such as THP and
* hugetlbfs are not to be compacted unless we are attempting
* an allocation much larger than the huge page size (eg CMA).
* We can potentially save a lot of iterations if we skip them
* at once. The check is racy, but we can consider only valid
* values and the only danger is skipping too much.
*/
if (PageCompound(page) && !cc->alloc_contig) {
const unsigned int order = compound_order(page);
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
if (likely(order < MAX_ORDER))
low_pfn += (1UL << order) - 1;
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
goto isolate_fail;
mm: compaction: acquire the zone->lru_lock as late as possible Richard Davies and Shaohua Li have both reported lock contention problems in compaction on the zone and LRU locks as well as significant amounts of time being spent in compaction. This series aims to reduce lock contention and scanning rates to reduce that CPU usage. Richard reported at https://lkml.org/lkml/2012/9/21/91 that this series made a big different to a problem he reported in August: http://marc.info/?l=kvm&m=134511507015614&w=2 Patch 1 defers acquiring the zone->lru_lock as long as possible. Patch 2 defers acquiring the zone->lock as lock as possible. Patch 3 reverts Rik's "skip-free" patches as the core concept gets reimplemented later and the remaining patches are easier to understand if this is reverted first. Patch 4 adds a pageblock-skip bit to the pageblock flags to cache what pageblocks should be skipped by the migrate and free scanners. This drastically reduces the amount of scanning compaction has to do. Patch 5 reimplements something similar to Rik's idea except it uses the pageblock-skip information to decide where the scanners should restart from and does not need to wrap around. I tested this on 3.6-rc6 + linux-next/akpm. Kernels tested were akpm-20120920 3.6-rc6 + linux-next/akpm as of Septeber 20th, 2012 lesslock Patches 1-6 revert Patches 1-7 cachefail Patches 1-8 skipuseless Patches 1-9 Stress high-order allocation tests looked ok. Success rates are more or less the same with the full series applied but there is an expectation that there is less opportunity to race with other allocation requests if there is less scanning. The time to complete the tests did not vary that much and are uninteresting as were the vmstat statistics so I will not present them here. Using ftrace I recorded how much scanning was done by compaction and got this 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 akpm-20120920 lockless revert-v2r2 cachefail skipuseless Total free scanned 360753976 515414028 565479007 17103281 18916589 Total free isolated 2852429 3597369 4048601 670493 727840 Total free efficiency 0.0079% 0.0070% 0.0072% 0.0392% 0.0385% Total migrate scanned 247728664 822729112 1004645830 17946827 14118903 Total migrate isolated 2555324 3245937 3437501 616359 658616 Total migrate efficiency 0.0103% 0.0039% 0.0034% 0.0343% 0.0466% The efficiency is worthless because of the nature of the test and the number of failures. The really interesting point as far as this patch series is concerned is the number of pages scanned. Note that reverting Rik's patches massively increases the number of pages scanned indicating that those patches really did make a difference to CPU usage. However, caching what pageblocks should be skipped has a much higher impact. With patches 1-8 applied, free page and migrate page scanning are both reduced by 95% in comparison to the akpm kernel. If the basic concept of Rik's patches are implemened on top then scanning then the free scanner barely changed but migrate scanning was further reduced. That said, tests on 3.6-rc5 indicated that the last patch had greater impact than what was measured here so it is a bit variable. One way or the other, this series has a large impact on the amount of scanning compaction does when there is a storm of THP allocations. This patch: Compaction's migrate scanner acquires the zone->lru_lock when scanning a range of pages looking for LRU pages to acquire. It does this even if there are no LRU pages in the range. If multiple processes are compacting then this can cause severe locking contention. To make matters worse commit b2eef8c0 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") releases the lru_lock every SWAP_CLUSTER_MAX pages that are scanned. This patch makes two changes to how the migrate scanner acquires the LRU lock. First, it only releases the LRU lock every SWAP_CLUSTER_MAX pages if the lock is contended. This reduces the number of times it unnecessarily disables and re-enables IRQs. The second is that it defers acquiring the LRU lock for as long as possible. If there are no LRU pages or the only LRU pages are transhuge then the LRU lock will not be acquired at all which reduces contention on zone->lru_lock. [minchan@kernel.org: augment comment] [akpm@linux-foundation.org: tweak comment text] Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:33 +00:00
}
mm: migrate: support non-lru movable page migration We have allowed migration for only LRU pages until now and it was enough to make high-order pages. But recently, embedded system(e.g., webOS, android) uses lots of non-movable pages(e.g., zram, GPU memory) so we have seen several reports about troubles of small high-order allocation. For fixing the problem, there were several efforts (e,g,. enhance compaction algorithm, SLUB fallback to 0-order page, reserved memory, vmalloc and so on) but if there are lots of non-movable pages in system, their solutions are void in the long run. So, this patch is to support facility to change non-movable pages with movable. For the feature, this patch introduces functions related to migration to address_space_operations as well as some page flags. If a driver want to make own pages movable, it should define three functions which are function pointers of struct address_space_operations. 1. bool (*isolate_page) (struct page *page, isolate_mode_t mode); What VM expects on isolate_page function of driver is to return *true* if driver isolates page successfully. On returing true, VM marks the page as PG_isolated so concurrent isolation in several CPUs skip the page for isolation. If a driver cannot isolate the page, it should return *false*. Once page is successfully isolated, VM uses page.lru fields so driver shouldn't expect to preserve values in that fields. 2. int (*migratepage) (struct address_space *mapping, struct page *newpage, struct page *oldpage, enum migrate_mode); After isolation, VM calls migratepage of driver with isolated page. The function of migratepage is to move content of the old page to new page and set up fields of struct page newpage. Keep in mind that you should indicate to the VM the oldpage is no longer movable via __ClearPageMovable() under page_lock if you migrated the oldpage successfully and returns 0. If driver cannot migrate the page at the moment, driver can return -EAGAIN. On -EAGAIN, VM will retry page migration in a short time because VM interprets -EAGAIN as "temporal migration failure". On returning any error except -EAGAIN, VM will give up the page migration without retrying in this time. Driver shouldn't touch page.lru field VM using in the functions. 3. void (*putback_page)(struct page *); If migration fails on isolated page, VM should return the isolated page to the driver so VM calls driver's putback_page with migration failed page. In this function, driver should put the isolated page back to the own data structure. 4. non-lru movable page flags There are two page flags for supporting non-lru movable page. * PG_movable Driver should use the below function to make page movable under page_lock. void __SetPageMovable(struct page *page, struct address_space *mapping) It needs argument of address_space for registering migration family functions which will be called by VM. Exactly speaking, PG_movable is not a real flag of struct page. Rather than, VM reuses page->mapping's lower bits to represent it. #define PAGE_MAPPING_MOVABLE 0x2 page->mapping = page->mapping | PAGE_MAPPING_MOVABLE; so driver shouldn't access page->mapping directly. Instead, driver should use page_mapping which mask off the low two bits of page->mapping so it can get right struct address_space. For testing of non-lru movable page, VM supports __PageMovable function. However, it doesn't guarantee to identify non-lru movable page because page->mapping field is unified with other variables in struct page. As well, if driver releases the page after isolation by VM, page->mapping doesn't have stable value although it has PAGE_MAPPING_MOVABLE (Look at __ClearPageMovable). But __PageMovable is cheap to catch whether page is LRU or non-lru movable once the page has been isolated. Because LRU pages never can have PAGE_MAPPING_MOVABLE in page->mapping. It is also good for just peeking to test non-lru movable pages before more expensive checking with lock_page in pfn scanning to select victim. For guaranteeing non-lru movable page, VM provides PageMovable function. Unlike __PageMovable, PageMovable functions validates page->mapping and mapping->a_ops->isolate_page under lock_page. The lock_page prevents sudden destroying of page->mapping. Driver using __SetPageMovable should clear the flag via __ClearMovablePage under page_lock before the releasing the page. * PG_isolated To prevent concurrent isolation among several CPUs, VM marks isolated page as PG_isolated under lock_page. So if a CPU encounters PG_isolated non-lru movable page, it can skip it. Driver doesn't need to manipulate the flag because VM will set/clear it automatically. Keep in mind that if driver sees PG_isolated page, it means the page have been isolated by VM so it shouldn't touch page.lru field. PG_isolated is alias with PG_reclaim flag so driver shouldn't use the flag for own purpose. [opensource.ganesh@gmail.com: mm/compaction: remove local variable is_lru] Link: http://lkml.kernel.org/r/20160618014841.GA7422@leo-test Link: http://lkml.kernel.org/r/1464736881-24886-3-git-send-email-minchan@kernel.org Signed-off-by: Gioh Kim <gi-oh.kim@profitbricks.com> Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Ganesh Mahendran <opensource.ganesh@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Rafael Aquini <aquini@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: John Einar Reitan <john.reitan@foss.arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-26 22:23:05 +00:00
/*
* Check may be lockless but that's ok as we recheck later.
* It's possible to migrate LRU and non-lru movable pages.
* Skip any other type of page
*/
if (!PageLRU(page)) {
/*
* __PageMovable can return false positive so we need
* to verify it under page_lock.
*/
if (unlikely(__PageMovable(page)) &&
!PageIsolated(page)) {
if (locked) {
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
unlock_page_lruvec_irqrestore(locked, flags);
locked = NULL;
mm: migrate: support non-lru movable page migration We have allowed migration for only LRU pages until now and it was enough to make high-order pages. But recently, embedded system(e.g., webOS, android) uses lots of non-movable pages(e.g., zram, GPU memory) so we have seen several reports about troubles of small high-order allocation. For fixing the problem, there were several efforts (e,g,. enhance compaction algorithm, SLUB fallback to 0-order page, reserved memory, vmalloc and so on) but if there are lots of non-movable pages in system, their solutions are void in the long run. So, this patch is to support facility to change non-movable pages with movable. For the feature, this patch introduces functions related to migration to address_space_operations as well as some page flags. If a driver want to make own pages movable, it should define three functions which are function pointers of struct address_space_operations. 1. bool (*isolate_page) (struct page *page, isolate_mode_t mode); What VM expects on isolate_page function of driver is to return *true* if driver isolates page successfully. On returing true, VM marks the page as PG_isolated so concurrent isolation in several CPUs skip the page for isolation. If a driver cannot isolate the page, it should return *false*. Once page is successfully isolated, VM uses page.lru fields so driver shouldn't expect to preserve values in that fields. 2. int (*migratepage) (struct address_space *mapping, struct page *newpage, struct page *oldpage, enum migrate_mode); After isolation, VM calls migratepage of driver with isolated page. The function of migratepage is to move content of the old page to new page and set up fields of struct page newpage. Keep in mind that you should indicate to the VM the oldpage is no longer movable via __ClearPageMovable() under page_lock if you migrated the oldpage successfully and returns 0. If driver cannot migrate the page at the moment, driver can return -EAGAIN. On -EAGAIN, VM will retry page migration in a short time because VM interprets -EAGAIN as "temporal migration failure". On returning any error except -EAGAIN, VM will give up the page migration without retrying in this time. Driver shouldn't touch page.lru field VM using in the functions. 3. void (*putback_page)(struct page *); If migration fails on isolated page, VM should return the isolated page to the driver so VM calls driver's putback_page with migration failed page. In this function, driver should put the isolated page back to the own data structure. 4. non-lru movable page flags There are two page flags for supporting non-lru movable page. * PG_movable Driver should use the below function to make page movable under page_lock. void __SetPageMovable(struct page *page, struct address_space *mapping) It needs argument of address_space for registering migration family functions which will be called by VM. Exactly speaking, PG_movable is not a real flag of struct page. Rather than, VM reuses page->mapping's lower bits to represent it. #define PAGE_MAPPING_MOVABLE 0x2 page->mapping = page->mapping | PAGE_MAPPING_MOVABLE; so driver shouldn't access page->mapping directly. Instead, driver should use page_mapping which mask off the low two bits of page->mapping so it can get right struct address_space. For testing of non-lru movable page, VM supports __PageMovable function. However, it doesn't guarantee to identify non-lru movable page because page->mapping field is unified with other variables in struct page. As well, if driver releases the page after isolation by VM, page->mapping doesn't have stable value although it has PAGE_MAPPING_MOVABLE (Look at __ClearPageMovable). But __PageMovable is cheap to catch whether page is LRU or non-lru movable once the page has been isolated. Because LRU pages never can have PAGE_MAPPING_MOVABLE in page->mapping. It is also good for just peeking to test non-lru movable pages before more expensive checking with lock_page in pfn scanning to select victim. For guaranteeing non-lru movable page, VM provides PageMovable function. Unlike __PageMovable, PageMovable functions validates page->mapping and mapping->a_ops->isolate_page under lock_page. The lock_page prevents sudden destroying of page->mapping. Driver using __SetPageMovable should clear the flag via __ClearMovablePage under page_lock before the releasing the page. * PG_isolated To prevent concurrent isolation among several CPUs, VM marks isolated page as PG_isolated under lock_page. So if a CPU encounters PG_isolated non-lru movable page, it can skip it. Driver doesn't need to manipulate the flag because VM will set/clear it automatically. Keep in mind that if driver sees PG_isolated page, it means the page have been isolated by VM so it shouldn't touch page.lru field. PG_isolated is alias with PG_reclaim flag so driver shouldn't use the flag for own purpose. [opensource.ganesh@gmail.com: mm/compaction: remove local variable is_lru] Link: http://lkml.kernel.org/r/20160618014841.GA7422@leo-test Link: http://lkml.kernel.org/r/1464736881-24886-3-git-send-email-minchan@kernel.org Signed-off-by: Gioh Kim <gi-oh.kim@profitbricks.com> Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Ganesh Mahendran <opensource.ganesh@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Rafael Aquini <aquini@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: John Einar Reitan <john.reitan@foss.arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-26 22:23:05 +00:00
}
mm/migration: make isolate_movable_page() return int type Patch series "HWPOISON: soft offlining for non-lru movable page", v6. After Minchan's commit bda807d44454 ("mm: migrate: support non-lru movable page migration"), some type of non-lru page like zsmalloc and virtio-balloon page also support migration. Therefore, we can: 1) soft offlining no-lru movable pages, which means when memory corrected errors occur on a non-lru movable page, we can stop to use it by migrating data onto another page and disable the original (maybe half-broken) one. 2) enable memory hotplug for non-lru movable pages, i.e. we may offline blocks, which include such pages, by using non-lru page migration. This patchset is heavily dependent on non-lru movable page migration. This patch (of 4): Change the return type of isolate_movable_page() from bool to int. It will return 0 when isolate movable page successfully, and return -EBUSY when it isolates failed. There is no functional change within this patch but prepare for later patch. [xieyisheng1@huawei.com: v6] Link: http://lkml.kernel.org/r/1486108770-630-2-git-send-email-xieyisheng1@huawei.com Link: http://lkml.kernel.org/r/1485867981-16037-2-git-send-email-ysxie@foxmail.com Signed-off-by: Yisheng Xie <xieyisheng1@huawei.com> Suggested-by: Michal Hocko <mhocko@kernel.org> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Andi Kleen <ak@linux.intel.com> Cc: Hanjun Guo <guohanjun@huawei.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Reza Arbab <arbab@linux.vnet.ibm.com> Cc: Taku Izumi <izumi.taku@jp.fujitsu.com> Cc: Vitaly Kuznetsov <vkuznets@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xishi Qiu <qiuxishi@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-24 22:57:29 +00:00
if (!isolate_movable_page(page, isolate_mode))
mm: migrate: support non-lru movable page migration We have allowed migration for only LRU pages until now and it was enough to make high-order pages. But recently, embedded system(e.g., webOS, android) uses lots of non-movable pages(e.g., zram, GPU memory) so we have seen several reports about troubles of small high-order allocation. For fixing the problem, there were several efforts (e,g,. enhance compaction algorithm, SLUB fallback to 0-order page, reserved memory, vmalloc and so on) but if there are lots of non-movable pages in system, their solutions are void in the long run. So, this patch is to support facility to change non-movable pages with movable. For the feature, this patch introduces functions related to migration to address_space_operations as well as some page flags. If a driver want to make own pages movable, it should define three functions which are function pointers of struct address_space_operations. 1. bool (*isolate_page) (struct page *page, isolate_mode_t mode); What VM expects on isolate_page function of driver is to return *true* if driver isolates page successfully. On returing true, VM marks the page as PG_isolated so concurrent isolation in several CPUs skip the page for isolation. If a driver cannot isolate the page, it should return *false*. Once page is successfully isolated, VM uses page.lru fields so driver shouldn't expect to preserve values in that fields. 2. int (*migratepage) (struct address_space *mapping, struct page *newpage, struct page *oldpage, enum migrate_mode); After isolation, VM calls migratepage of driver with isolated page. The function of migratepage is to move content of the old page to new page and set up fields of struct page newpage. Keep in mind that you should indicate to the VM the oldpage is no longer movable via __ClearPageMovable() under page_lock if you migrated the oldpage successfully and returns 0. If driver cannot migrate the page at the moment, driver can return -EAGAIN. On -EAGAIN, VM will retry page migration in a short time because VM interprets -EAGAIN as "temporal migration failure". On returning any error except -EAGAIN, VM will give up the page migration without retrying in this time. Driver shouldn't touch page.lru field VM using in the functions. 3. void (*putback_page)(struct page *); If migration fails on isolated page, VM should return the isolated page to the driver so VM calls driver's putback_page with migration failed page. In this function, driver should put the isolated page back to the own data structure. 4. non-lru movable page flags There are two page flags for supporting non-lru movable page. * PG_movable Driver should use the below function to make page movable under page_lock. void __SetPageMovable(struct page *page, struct address_space *mapping) It needs argument of address_space for registering migration family functions which will be called by VM. Exactly speaking, PG_movable is not a real flag of struct page. Rather than, VM reuses page->mapping's lower bits to represent it. #define PAGE_MAPPING_MOVABLE 0x2 page->mapping = page->mapping | PAGE_MAPPING_MOVABLE; so driver shouldn't access page->mapping directly. Instead, driver should use page_mapping which mask off the low two bits of page->mapping so it can get right struct address_space. For testing of non-lru movable page, VM supports __PageMovable function. However, it doesn't guarantee to identify non-lru movable page because page->mapping field is unified with other variables in struct page. As well, if driver releases the page after isolation by VM, page->mapping doesn't have stable value although it has PAGE_MAPPING_MOVABLE (Look at __ClearPageMovable). But __PageMovable is cheap to catch whether page is LRU or non-lru movable once the page has been isolated. Because LRU pages never can have PAGE_MAPPING_MOVABLE in page->mapping. It is also good for just peeking to test non-lru movable pages before more expensive checking with lock_page in pfn scanning to select victim. For guaranteeing non-lru movable page, VM provides PageMovable function. Unlike __PageMovable, PageMovable functions validates page->mapping and mapping->a_ops->isolate_page under lock_page. The lock_page prevents sudden destroying of page->mapping. Driver using __SetPageMovable should clear the flag via __ClearMovablePage under page_lock before the releasing the page. * PG_isolated To prevent concurrent isolation among several CPUs, VM marks isolated page as PG_isolated under lock_page. So if a CPU encounters PG_isolated non-lru movable page, it can skip it. Driver doesn't need to manipulate the flag because VM will set/clear it automatically. Keep in mind that if driver sees PG_isolated page, it means the page have been isolated by VM so it shouldn't touch page.lru field. PG_isolated is alias with PG_reclaim flag so driver shouldn't use the flag for own purpose. [opensource.ganesh@gmail.com: mm/compaction: remove local variable is_lru] Link: http://lkml.kernel.org/r/20160618014841.GA7422@leo-test Link: http://lkml.kernel.org/r/1464736881-24886-3-git-send-email-minchan@kernel.org Signed-off-by: Gioh Kim <gi-oh.kim@profitbricks.com> Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Ganesh Mahendran <opensource.ganesh@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Rafael Aquini <aquini@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: John Einar Reitan <john.reitan@foss.arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-26 22:23:05 +00:00
goto isolate_success;
}
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
goto isolate_fail;
mm: migrate: support non-lru movable page migration We have allowed migration for only LRU pages until now and it was enough to make high-order pages. But recently, embedded system(e.g., webOS, android) uses lots of non-movable pages(e.g., zram, GPU memory) so we have seen several reports about troubles of small high-order allocation. For fixing the problem, there were several efforts (e,g,. enhance compaction algorithm, SLUB fallback to 0-order page, reserved memory, vmalloc and so on) but if there are lots of non-movable pages in system, their solutions are void in the long run. So, this patch is to support facility to change non-movable pages with movable. For the feature, this patch introduces functions related to migration to address_space_operations as well as some page flags. If a driver want to make own pages movable, it should define three functions which are function pointers of struct address_space_operations. 1. bool (*isolate_page) (struct page *page, isolate_mode_t mode); What VM expects on isolate_page function of driver is to return *true* if driver isolates page successfully. On returing true, VM marks the page as PG_isolated so concurrent isolation in several CPUs skip the page for isolation. If a driver cannot isolate the page, it should return *false*. Once page is successfully isolated, VM uses page.lru fields so driver shouldn't expect to preserve values in that fields. 2. int (*migratepage) (struct address_space *mapping, struct page *newpage, struct page *oldpage, enum migrate_mode); After isolation, VM calls migratepage of driver with isolated page. The function of migratepage is to move content of the old page to new page and set up fields of struct page newpage. Keep in mind that you should indicate to the VM the oldpage is no longer movable via __ClearPageMovable() under page_lock if you migrated the oldpage successfully and returns 0. If driver cannot migrate the page at the moment, driver can return -EAGAIN. On -EAGAIN, VM will retry page migration in a short time because VM interprets -EAGAIN as "temporal migration failure". On returning any error except -EAGAIN, VM will give up the page migration without retrying in this time. Driver shouldn't touch page.lru field VM using in the functions. 3. void (*putback_page)(struct page *); If migration fails on isolated page, VM should return the isolated page to the driver so VM calls driver's putback_page with migration failed page. In this function, driver should put the isolated page back to the own data structure. 4. non-lru movable page flags There are two page flags for supporting non-lru movable page. * PG_movable Driver should use the below function to make page movable under page_lock. void __SetPageMovable(struct page *page, struct address_space *mapping) It needs argument of address_space for registering migration family functions which will be called by VM. Exactly speaking, PG_movable is not a real flag of struct page. Rather than, VM reuses page->mapping's lower bits to represent it. #define PAGE_MAPPING_MOVABLE 0x2 page->mapping = page->mapping | PAGE_MAPPING_MOVABLE; so driver shouldn't access page->mapping directly. Instead, driver should use page_mapping which mask off the low two bits of page->mapping so it can get right struct address_space. For testing of non-lru movable page, VM supports __PageMovable function. However, it doesn't guarantee to identify non-lru movable page because page->mapping field is unified with other variables in struct page. As well, if driver releases the page after isolation by VM, page->mapping doesn't have stable value although it has PAGE_MAPPING_MOVABLE (Look at __ClearPageMovable). But __PageMovable is cheap to catch whether page is LRU or non-lru movable once the page has been isolated. Because LRU pages never can have PAGE_MAPPING_MOVABLE in page->mapping. It is also good for just peeking to test non-lru movable pages before more expensive checking with lock_page in pfn scanning to select victim. For guaranteeing non-lru movable page, VM provides PageMovable function. Unlike __PageMovable, PageMovable functions validates page->mapping and mapping->a_ops->isolate_page under lock_page. The lock_page prevents sudden destroying of page->mapping. Driver using __SetPageMovable should clear the flag via __ClearMovablePage under page_lock before the releasing the page. * PG_isolated To prevent concurrent isolation among several CPUs, VM marks isolated page as PG_isolated under lock_page. So if a CPU encounters PG_isolated non-lru movable page, it can skip it. Driver doesn't need to manipulate the flag because VM will set/clear it automatically. Keep in mind that if driver sees PG_isolated page, it means the page have been isolated by VM so it shouldn't touch page.lru field. PG_isolated is alias with PG_reclaim flag so driver shouldn't use the flag for own purpose. [opensource.ganesh@gmail.com: mm/compaction: remove local variable is_lru] Link: http://lkml.kernel.org/r/20160618014841.GA7422@leo-test Link: http://lkml.kernel.org/r/1464736881-24886-3-git-send-email-minchan@kernel.org Signed-off-by: Gioh Kim <gi-oh.kim@profitbricks.com> Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Ganesh Mahendran <opensource.ganesh@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Rafael Aquini <aquini@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: John Einar Reitan <john.reitan@foss.arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-26 22:23:05 +00:00
}
mm, compaction: always skip all compound pages by order in migrate scanner The compaction migrate scanner tries to skip THP pages by their order, to reduce number of iterations for pages it cannot isolate. The check is only done if PageLRU() is true, which means it applies to THP pages, but not e.g. hugetlbfs pages or any other non-LRU compound pages, which we have to iterate by base pages. This limitation comes from the assumption that it's only safe to read compound_order() when we have the zone's lru_lock and THP cannot be split under us. But the only danger (after filtering out order values that are not below MAX_ORDER, to prevent overflows) is that we skip too much or too little after reading a bogus compound_order() due to a rare race. This is the same reasoning as patch 99c0fd5e51c4 ("mm, compaction: skip buddy pages by their order in the migrate scanner") introduced for unsafely reading PageBuddy() order. After this patch, all pages are tested for PageCompound() and we skip them by compound_order(). The test is done after the test for balloon_page_movable() as we don't want to assume if balloon pages (or other pages with own isolation and migration implementation if a generic API gets implemented) are compound or not. When tested with stress-highalloc from mmtests on 4GB system with 1GB hugetlbfs pages, the vmstat compact_migrate_scanned count decreased by 15%. [kirill.shutemov@linux.intel.com: change PageTransHuge checks to PageCompound for different series was squashed here] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:46 +00:00
/*
* Migration will fail if an anonymous page is pinned in memory,
* so avoid taking lru_lock and isolating it unnecessarily in an
* admittedly racy check.
*/
if (!page_mapping(page) &&
page_count(page) > page_mapcount(page))
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
goto isolate_fail;
mm, compaction: allow compaction for GFP_NOFS requests compaction has been disabled for GFP_NOFS and GFP_NOIO requests since the direct compaction was introduced by commit 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). The main reason is that the migration of page cache pages might recurse back to fs/io layer and we could potentially deadlock. This is overly conservative because all the anonymous memory is migrateable in the GFP_NOFS context just fine. This might be a large portion of the memory in many/most workkloads. Remove the GFP_NOFS restriction and make sure that we skip all fs pages (those with a mapping) while isolating pages to be migrated. We cannot consider clean fs pages because they might need a metadata update so only isolate pages without any mapping for nofs requests. The effect of this patch will be probably very limited in many/most workloads because higher order GFP_NOFS requests are quite rare, although different configurations might lead to very different results. David Chinner has mentioned a heavy metadata workload with 64kB block which to quote him: : Unfortunately, there was an era of cargo cult configuration tweaks in the : Ceph community that has resulted in a large number of production machines : with XFS filesystems configured this way. And a lot of them store large : numbers of small files and run under significant sustained memory : pressure. : : I slowly working towards getting rid of these high order allocations and : replacing them with the equivalent number of single page allocations, but : I haven't got that (complex) change working yet. We can do the following to simulate that workload: $ mkfs.xfs -f -n size=64k <dev> $ mount <dev> /mnt/scratch $ time ./fs_mark -D 10000 -S0 -n 100000 -s 0 -L 32 \ -d /mnt/scratch/0 -d /mnt/scratch/1 \ -d /mnt/scratch/2 -d /mnt/scratch/3 \ -d /mnt/scratch/4 -d /mnt/scratch/5 \ -d /mnt/scratch/6 -d /mnt/scratch/7 \ -d /mnt/scratch/8 -d /mnt/scratch/9 \ -d /mnt/scratch/10 -d /mnt/scratch/11 \ -d /mnt/scratch/12 -d /mnt/scratch/13 \ -d /mnt/scratch/14 -d /mnt/scratch/15 and indeed is hammers the system with many high order GFP_NOFS requests as per a simle tracepoint during the load: $ echo '!(gfp_flags & 0x80) && (gfp_flags &0x400000)' > $TRACE_MNT/events/kmem/mm_page_alloc/filter I am getting 5287609 order=0 37 order=1 1594905 order=2 3048439 order=3 6699207 order=4 66645 order=5 My testing was done in a kvm guest so performance numbers should be taken with a grain of salt but there seems to be a difference when the patch is applied: * Original kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4300.1 20745838 3 3200000 0 4239.9 23849857 5 4800000 0 4243.4 25939543 6 6400000 0 4248.4 19514050 8 8000000 0 4262.1 20796169 9 9600000 0 4257.6 21288675 11 11200000 0 4259.7 19375120 13 12800000 0 4220.7 22734141 14 14400000 0 4238.5 31936458 16 16000000 0 4231.5 23409901 18 17600000 0 4045.3 23577700 19 19200000 0 2783.4 58299526 21 20800000 0 2678.2 40616302 23 22400000 0 2693.5 83973996 and xfs complaining about memory allocation not making progress [ 2304.372647] XFS: fs_mark(3289) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) [ 2304.443323] XFS: fs_mark(3285) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) [ 4796.772477] XFS: fs_mark(3424) possible memory allocation deadlock size 46936 in kmem_alloc (mode:0x2408240) [ 4796.775329] XFS: fs_mark(3423) possible memory allocation deadlock size 51416 in kmem_alloc (mode:0x2408240) [ 4797.388808] XFS: fs_mark(3424) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) * Patched kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4289.1 19243934 3 3200000 0 4241.6 32828865 5 4800000 0 4248.7 32884693 6 6400000 0 4314.4 19608921 8 8000000 0 4269.9 24953292 9 9600000 0 4270.7 33235572 11 11200000 0 4346.4 40817101 13 12800000 0 4285.3 29972397 14 14400000 0 4297.2 20539765 16 16000000 0 4219.6 18596767 18 17600000 0 4273.8 49611187 19 19200000 0 4300.4 27944451 21 20800000 0 4270.6 22324585 22 22400000 0 4317.6 22650382 24 24000000 0 4065.2 22297964 So the dropdown at Count 19200000 didn't happen and there was only a single warning about allocation not making progress [ 3063.815003] XFS: fs_mark(3272) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) This suggests that the patch has helped even though there is not all that much of anonymous memory as the workload mostly generates fs metadata. I assume the success rate would be higher with more anonymous memory which should be the case in many workloads. [akpm@linux-foundation.org: fix comment] Link: http://lkml.kernel.org/r/20161012114721.31853-1-mhocko@kernel.org Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 23:04:07 +00:00
/*
* Only allow to migrate anonymous pages in GFP_NOFS context
* because those do not depend on fs locks.
*/
if (!(cc->gfp_mask & __GFP_FS) && page_mapping(page))
goto isolate_fail;
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
/*
* Be careful not to clear PageLRU until after we're
* sure the page is not being freed elsewhere -- the
* page release code relies on it.
*/
if (unlikely(!get_page_unless_zero(page)))
goto isolate_fail;
if (!__isolate_lru_page_prepare(page, isolate_mode))
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
goto isolate_fail_put;
/* Try isolate the page */
if (!TestClearPageLRU(page))
goto isolate_fail_put;
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
lruvec = mem_cgroup_page_lruvec(page, pgdat);
/* If we already hold the lock, we can skip some rechecking */
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
if (lruvec != locked) {
if (locked)
unlock_page_lruvec_irqrestore(locked, flags);
compact_lock_irqsave(&lruvec->lru_lock, &flags, cc);
locked = lruvec;
lruvec_memcg_debug(lruvec, page);
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
/* Try get exclusive access under lock */
if (!skip_updated) {
skip_updated = true;
if (test_and_set_skip(cc, page, low_pfn))
goto isolate_abort;
}
mm: compaction: acquire the zone->lru_lock as late as possible Richard Davies and Shaohua Li have both reported lock contention problems in compaction on the zone and LRU locks as well as significant amounts of time being spent in compaction. This series aims to reduce lock contention and scanning rates to reduce that CPU usage. Richard reported at https://lkml.org/lkml/2012/9/21/91 that this series made a big different to a problem he reported in August: http://marc.info/?l=kvm&m=134511507015614&w=2 Patch 1 defers acquiring the zone->lru_lock as long as possible. Patch 2 defers acquiring the zone->lock as lock as possible. Patch 3 reverts Rik's "skip-free" patches as the core concept gets reimplemented later and the remaining patches are easier to understand if this is reverted first. Patch 4 adds a pageblock-skip bit to the pageblock flags to cache what pageblocks should be skipped by the migrate and free scanners. This drastically reduces the amount of scanning compaction has to do. Patch 5 reimplements something similar to Rik's idea except it uses the pageblock-skip information to decide where the scanners should restart from and does not need to wrap around. I tested this on 3.6-rc6 + linux-next/akpm. Kernels tested were akpm-20120920 3.6-rc6 + linux-next/akpm as of Septeber 20th, 2012 lesslock Patches 1-6 revert Patches 1-7 cachefail Patches 1-8 skipuseless Patches 1-9 Stress high-order allocation tests looked ok. Success rates are more or less the same with the full series applied but there is an expectation that there is less opportunity to race with other allocation requests if there is less scanning. The time to complete the tests did not vary that much and are uninteresting as were the vmstat statistics so I will not present them here. Using ftrace I recorded how much scanning was done by compaction and got this 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 akpm-20120920 lockless revert-v2r2 cachefail skipuseless Total free scanned 360753976 515414028 565479007 17103281 18916589 Total free isolated 2852429 3597369 4048601 670493 727840 Total free efficiency 0.0079% 0.0070% 0.0072% 0.0392% 0.0385% Total migrate scanned 247728664 822729112 1004645830 17946827 14118903 Total migrate isolated 2555324 3245937 3437501 616359 658616 Total migrate efficiency 0.0103% 0.0039% 0.0034% 0.0343% 0.0466% The efficiency is worthless because of the nature of the test and the number of failures. The really interesting point as far as this patch series is concerned is the number of pages scanned. Note that reverting Rik's patches massively increases the number of pages scanned indicating that those patches really did make a difference to CPU usage. However, caching what pageblocks should be skipped has a much higher impact. With patches 1-8 applied, free page and migrate page scanning are both reduced by 95% in comparison to the akpm kernel. If the basic concept of Rik's patches are implemened on top then scanning then the free scanner barely changed but migrate scanning was further reduced. That said, tests on 3.6-rc5 indicated that the last patch had greater impact than what was measured here so it is a bit variable. One way or the other, this series has a large impact on the amount of scanning compaction does when there is a storm of THP allocations. This patch: Compaction's migrate scanner acquires the zone->lru_lock when scanning a range of pages looking for LRU pages to acquire. It does this even if there are no LRU pages in the range. If multiple processes are compacting then this can cause severe locking contention. To make matters worse commit b2eef8c0 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") releases the lru_lock every SWAP_CLUSTER_MAX pages that are scanned. This patch makes two changes to how the migrate scanner acquires the LRU lock. First, it only releases the LRU lock every SWAP_CLUSTER_MAX pages if the lock is contended. This reduces the number of times it unnecessarily disables and re-enables IRQs. The second is that it defers acquiring the LRU lock for as long as possible. If there are no LRU pages or the only LRU pages are transhuge then the LRU lock will not be acquired at all which reduces contention on zone->lru_lock. [minchan@kernel.org: augment comment] [akpm@linux-foundation.org: tweak comment text] Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:33 +00:00
mm, compaction: always skip all compound pages by order in migrate scanner The compaction migrate scanner tries to skip THP pages by their order, to reduce number of iterations for pages it cannot isolate. The check is only done if PageLRU() is true, which means it applies to THP pages, but not e.g. hugetlbfs pages or any other non-LRU compound pages, which we have to iterate by base pages. This limitation comes from the assumption that it's only safe to read compound_order() when we have the zone's lru_lock and THP cannot be split under us. But the only danger (after filtering out order values that are not below MAX_ORDER, to prevent overflows) is that we skip too much or too little after reading a bogus compound_order() due to a rare race. This is the same reasoning as patch 99c0fd5e51c4 ("mm, compaction: skip buddy pages by their order in the migrate scanner") introduced for unsafely reading PageBuddy() order. After this patch, all pages are tested for PageCompound() and we skip them by compound_order(). The test is done after the test for balloon_page_movable() as we don't want to assume if balloon pages (or other pages with own isolation and migration implementation if a generic API gets implemented) are compound or not. When tested with stress-highalloc from mmtests on 4GB system with 1GB hugetlbfs pages, the vmstat compact_migrate_scanned count decreased by 15%. [kirill.shutemov@linux.intel.com: change PageTransHuge checks to PageCompound for different series was squashed here] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:46 +00:00
/*
* Page become compound since the non-locked check,
* and it's on LRU. It can only be a THP so the order
* is safe to read and it's 0 for tail pages.
*/
if (unlikely(PageCompound(page) && !cc->alloc_contig)) {
low_pfn += compound_nr(page) - 1;
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
SetPageLRU(page);
goto isolate_fail_put;
}
}
/* The whole page is taken off the LRU; skip the tail pages. */
if (PageCompound(page))
low_pfn += compound_nr(page) - 1;
/* Successfully isolated */
del_page_from_lru_list(page, lruvec);
mod_node_page_state(page_pgdat(page),
NR_ISOLATED_ANON + page_is_file_lru(page),
thp_nr_pages(page));
isolate_success:
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
list_add(&page->lru, &cc->migratepages);
isolate_success_no_list:
mm/compaction: count pages and stop correctly during page isolation In isolate_migratepages_block, when cc->alloc_contig is true, we are able to isolate compound pages. But nr_migratepages and nr_isolated did not count compound pages correctly, causing us to isolate more pages than we thought. So count compound pages as the number of base pages they contain. Otherwise, we might be trapped in too_many_isolated while loop, since the actual isolated pages can go up to COMPACT_CLUSTER_MAX*512=16384, where COMPACT_CLUSTER_MAX is 32, since we stop isolation after cc->nr_migratepages reaches to COMPACT_CLUSTER_MAX. In addition, after we fix the issue above, cc->nr_migratepages could never be equal to COMPACT_CLUSTER_MAX if compound pages are isolated, thus page isolation could not stop as we intended. Change the isolation stop condition to '>='. The issue can be triggered as follows: In a system with 16GB memory and an 8GB CMA region reserved by hugetlb_cma, if we first allocate 10GB THPs and mlock them (so some THPs are allocated in the CMA region and mlocked), reserving 6 1GB hugetlb pages via /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages will get stuck (looping in too_many_isolated function) until we kill either task. With the patch applied, oom will kill the application with 10GB THPs and let hugetlb page reservation finish. [ziy@nvidia.com: v3] Link: https://lkml.kernel.org/r/20201030183809.3616803-1-zi.yan@sent.com Fixes: 1da2f328fa64 ("cmm,thp,compaction,cma: allow THP migration for CMA allocations") Signed-off-by: Zi Yan <ziy@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Yang Shi <shy828301@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@surriel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: <stable@vger.kernel.org> Link: https://lkml.kernel.org/r/20201029200435.3386066-1-zi.yan@sent.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-11-14 06:51:40 +00:00
cc->nr_migratepages += compound_nr(page);
nr_isolated += compound_nr(page);
mm, compaction: avoid rescanning the same pageblock multiple times Pageblocks are marked for skip when no pages are isolated after a scan. However, it's possible to hit corner cases where the migration scanner gets stuck near the boundary between the source and target scanner. Due to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are migrated can be reallocated before the pageblock is complete. The pageblock is not necessarily skipped so it can be rescanned multiple times. Similarly, a pageblock with some dirty/writeback pages may fail to migrate and be rescanned until writeback completes which is wasteful. This patch tracks if a pageblock is being rescanned. If so, then the entire pageblock will be migrated as one operation. This narrows the race window during which pages can be reallocated during migration. Secondly, if there are pages that cannot be isolated then the pageblock will still be fully scanned and marked for skipping. On the second rescan, the pageblock skip is set and the migration scanner makes progress. 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3200.68 ( 0.00%) 3002.07 ( 6.21%) Amean fault-both-5 4847.75 ( 0.00%) 4684.47 ( 3.37%) Amean fault-both-7 6658.92 ( 0.00%) 6815.54 ( -2.35%) Amean fault-both-12 11077.62 ( 0.00%) 10864.02 ( 1.93%) Amean fault-both-18 12403.97 ( 0.00%) 12247.52 ( 1.26%) Amean fault-both-24 15607.10 ( 0.00%) 15683.99 ( -0.49%) Amean fault-both-30 18752.27 ( 0.00%) 18620.02 ( 0.71%) Amean fault-both-32 21207.54 ( 0.00%) 19250.28 * 9.23%* 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Percentage huge-3 96.86 ( 0.00%) 95.00 ( -1.91%) Percentage huge-5 93.72 ( 0.00%) 94.22 ( 0.53%) Percentage huge-7 94.31 ( 0.00%) 92.35 ( -2.08%) Percentage huge-12 92.66 ( 0.00%) 91.90 ( -0.82%) Percentage huge-18 91.51 ( 0.00%) 89.58 ( -2.11%) Percentage huge-24 90.50 ( 0.00%) 90.03 ( -0.52%) Percentage huge-30 91.57 ( 0.00%) 89.14 ( -2.65%) Percentage huge-32 91.00 ( 0.00%) 90.58 ( -0.46%) Negligible difference but this was likely a case when the specific corner case was not hit. A previous run of the same patch based on an earlier iteration of the series showed large differences where migration rates could be halved when the corner case was hit. The specific corner case where migration scan rates go through the roof was due to a dirty/writeback pageblock located at the boundary of the migration/free scanner did not happen in this case. When it does happen, the scan rates multipled by massive margins. Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:07 +00:00
/*
* Avoid isolating too much unless this block is being
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
* rescanned (e.g. dirty/writeback pages, parallel allocation)
* or a lock is contended. For contention, isolate quickly to
* potentially remove one source of contention.
mm, compaction: avoid rescanning the same pageblock multiple times Pageblocks are marked for skip when no pages are isolated after a scan. However, it's possible to hit corner cases where the migration scanner gets stuck near the boundary between the source and target scanner. Due to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are migrated can be reallocated before the pageblock is complete. The pageblock is not necessarily skipped so it can be rescanned multiple times. Similarly, a pageblock with some dirty/writeback pages may fail to migrate and be rescanned until writeback completes which is wasteful. This patch tracks if a pageblock is being rescanned. If so, then the entire pageblock will be migrated as one operation. This narrows the race window during which pages can be reallocated during migration. Secondly, if there are pages that cannot be isolated then the pageblock will still be fully scanned and marked for skipping. On the second rescan, the pageblock skip is set and the migration scanner makes progress. 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3200.68 ( 0.00%) 3002.07 ( 6.21%) Amean fault-both-5 4847.75 ( 0.00%) 4684.47 ( 3.37%) Amean fault-both-7 6658.92 ( 0.00%) 6815.54 ( -2.35%) Amean fault-both-12 11077.62 ( 0.00%) 10864.02 ( 1.93%) Amean fault-both-18 12403.97 ( 0.00%) 12247.52 ( 1.26%) Amean fault-both-24 15607.10 ( 0.00%) 15683.99 ( -0.49%) Amean fault-both-30 18752.27 ( 0.00%) 18620.02 ( 0.71%) Amean fault-both-32 21207.54 ( 0.00%) 19250.28 * 9.23%* 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Percentage huge-3 96.86 ( 0.00%) 95.00 ( -1.91%) Percentage huge-5 93.72 ( 0.00%) 94.22 ( 0.53%) Percentage huge-7 94.31 ( 0.00%) 92.35 ( -2.08%) Percentage huge-12 92.66 ( 0.00%) 91.90 ( -0.82%) Percentage huge-18 91.51 ( 0.00%) 89.58 ( -2.11%) Percentage huge-24 90.50 ( 0.00%) 90.03 ( -0.52%) Percentage huge-30 91.57 ( 0.00%) 89.14 ( -2.65%) Percentage huge-32 91.00 ( 0.00%) 90.58 ( -0.46%) Negligible difference but this was likely a case when the specific corner case was not hit. A previous run of the same patch based on an earlier iteration of the series showed large differences where migration rates could be halved when the corner case was hit. The specific corner case where migration scan rates go through the roof was due to a dirty/writeback pageblock located at the boundary of the migration/free scanner did not happen in this case. When it does happen, the scan rates multipled by massive margins. Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:07 +00:00
*/
mm/compaction: count pages and stop correctly during page isolation In isolate_migratepages_block, when cc->alloc_contig is true, we are able to isolate compound pages. But nr_migratepages and nr_isolated did not count compound pages correctly, causing us to isolate more pages than we thought. So count compound pages as the number of base pages they contain. Otherwise, we might be trapped in too_many_isolated while loop, since the actual isolated pages can go up to COMPACT_CLUSTER_MAX*512=16384, where COMPACT_CLUSTER_MAX is 32, since we stop isolation after cc->nr_migratepages reaches to COMPACT_CLUSTER_MAX. In addition, after we fix the issue above, cc->nr_migratepages could never be equal to COMPACT_CLUSTER_MAX if compound pages are isolated, thus page isolation could not stop as we intended. Change the isolation stop condition to '>='. The issue can be triggered as follows: In a system with 16GB memory and an 8GB CMA region reserved by hugetlb_cma, if we first allocate 10GB THPs and mlock them (so some THPs are allocated in the CMA region and mlocked), reserving 6 1GB hugetlb pages via /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages will get stuck (looping in too_many_isolated function) until we kill either task. With the patch applied, oom will kill the application with 10GB THPs and let hugetlb page reservation finish. [ziy@nvidia.com: v3] Link: https://lkml.kernel.org/r/20201030183809.3616803-1-zi.yan@sent.com Fixes: 1da2f328fa64 ("cmm,thp,compaction,cma: allow THP migration for CMA allocations") Signed-off-by: Zi Yan <ziy@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Yang Shi <shy828301@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@surriel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: <stable@vger.kernel.org> Link: https://lkml.kernel.org/r/20201029200435.3386066-1-zi.yan@sent.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-11-14 06:51:40 +00:00
if (cc->nr_migratepages >= COMPACT_CLUSTER_MAX &&
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
!cc->rescan && !cc->contended) {
++low_pfn;
break;
}
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
continue;
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
isolate_fail_put:
/* Avoid potential deadlock in freeing page under lru_lock */
if (locked) {
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
unlock_page_lruvec_irqrestore(locked, flags);
locked = NULL;
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
}
put_page(page);
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
isolate_fail:
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
if (!skip_on_failure && ret != -ENOMEM)
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
continue;
/*
* We have isolated some pages, but then failed. Release them
* instead of migrating, as we cannot form the cc->order buddy
* page anyway.
*/
if (nr_isolated) {
if (locked) {
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
unlock_page_lruvec_irqrestore(locked, flags);
locked = NULL;
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
}
putback_movable_pages(&cc->migratepages);
cc->nr_migratepages = 0;
nr_isolated = 0;
}
if (low_pfn < next_skip_pfn) {
low_pfn = next_skip_pfn - 1;
/*
* The check near the loop beginning would have updated
* next_skip_pfn too, but this is a bit simpler.
*/
next_skip_pfn += 1UL << cc->order;
}
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
if (ret == -ENOMEM)
break;
}
mm, compaction: skip buddy pages by their order in the migrate scanner The migration scanner skips PageBuddy pages, but does not consider their order as checking page_order() is generally unsafe without holding the zone->lock, and acquiring the lock just for the check wouldn't be a good tradeoff. Still, this could avoid some iterations over the rest of the buddy page, and if we are careful, the race window between PageBuddy() check and page_order() is small, and the worst thing that can happen is that we skip too much and miss some isolation candidates. This is not that bad, as compaction can already fail for many other reasons like parallel allocations, and those have much larger race window. This patch therefore makes the migration scanner obtain the buddy page order and use it to skip the whole buddy page, if the order appears to be in the valid range. It's important that the page_order() is read only once, so that the value used in the checks and in the pfn calculation is the same. But in theory the compiler can replace the local variable by multiple inlines of page_order(). Therefore, the patch introduces page_order_unsafe() that uses ACCESS_ONCE to prevent this. Testing with stress-highalloc from mmtests shows a 15% reduction in number of pages scanned by migration scanner. The reduction is >60% with __GFP_NO_KSWAPD allocations, along with success rates better by few percent. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:23 +00:00
/*
* The PageBuddy() check could have potentially brought us outside
* the range to be scanned.
*/
if (unlikely(low_pfn > end_pfn))
low_pfn = end_pfn;
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
page = NULL;
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
isolate_abort:
mm: compaction: Abort async compaction if locks are contended or taking too long Jim Schutt reported a problem that pointed at compaction contending heavily on locks. The workload is straight-forward and in his own words; The systems in question have 24 SAS drives spread across 3 HBAs, running 24 Ceph OSD instances, one per drive. FWIW these servers are dual-socket Intel 5675 Xeons w/48 GB memory. I've got ~160 Ceph Linux clients doing dd simultaneously to a Ceph file system backed by 12 of these servers. Early in the test everything looks fine procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 31 15 0 287216 576 38606628 0 0 2 1158 2 14 1 3 95 0 0 27 15 0 225288 576 38583384 0 0 18 2222016 203357 134876 11 56 17 15 0 28 17 0 219256 576 38544736 0 0 11 2305932 203141 146296 11 49 23 17 0 6 18 0 215596 576 38552872 0 0 7 2363207 215264 166502 12 45 22 20 0 22 18 0 226984 576 38596404 0 0 3 2445741 223114 179527 12 43 23 22 0 and then it goes to pot procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu------- r b swpd free buff cache si so bi bo in cs us sy id wa st 163 8 0 464308 576 36791368 0 0 11 22210 866 536 3 13 79 4 0 207 14 0 917752 576 36181928 0 0 712 1345376 134598 47367 7 90 1 2 0 123 12 0 685516 576 36296148 0 0 429 1386615 158494 60077 8 84 5 3 0 123 12 0 598572 576 36333728 0 0 1107 1233281 147542 62351 7 84 5 4 0 622 7 0 660768 576 36118264 0 0 557 1345548 151394 59353 7 85 4 3 0 223 11 0 283960 576 36463868 0 0 46 1107160 121846 33006 6 93 1 1 0 Note that system CPU usage is very high blocks being written out has dropped by 42%. He analysed this with perf and found perf record -g -a sleep 10 perf report --sort symbol --call-graph fractal,5 34.63% [k] _raw_spin_lock_irqsave | |--97.30%-- isolate_freepages | compaction_alloc | unmap_and_move | migrate_pages | compact_zone | compact_zone_order | try_to_compact_pages | __alloc_pages_direct_compact | __alloc_pages_slowpath | __alloc_pages_nodemask | alloc_pages_vma | do_huge_pmd_anonymous_page | handle_mm_fault | do_page_fault | page_fault | | | |--87.39%-- skb_copy_datagram_iovec | | tcp_recvmsg | | inet_recvmsg | | sock_recvmsg | | sys_recvfrom | | system_call | | __recv | | | | | --100.00%-- (nil) | | | --12.61%-- memcpy --2.70%-- [...] There was other data but primarily it is all showing that compaction is contended heavily on the zone->lock and zone->lru_lock. commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled while isolating pages for migration] noted that it was possible for migration to hold the lru_lock for an excessive amount of time. Very broadly speaking this patch expands the concept. This patch introduces compact_checklock_irqsave() to check if a lock is contended or the process needs to be scheduled. If either condition is true then async compaction is aborted and the caller is informed. The page allocator will fail a THP allocation if compaction failed due to contention. This patch also introduces compact_trylock_irqsave() which will acquire the lock only if it is not contended and the process does not need to schedule. Reported-by: Jim Schutt <jaschut@sandia.gov> Tested-by: Jim Schutt <jaschut@sandia.gov> Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 23:16:17 +00:00
if (locked)
mm/lru: replace pgdat lru_lock with lruvec lock This patch moves per node lru_lock into lruvec, thus bring a lru_lock for each of memcg per node. So on a large machine, each of memcg don't have to suffer from per node pgdat->lru_lock competition. They could go fast with their self lru_lock. After move memcg charge before lru inserting, page isolation could serialize page's memcg, then per memcg lruvec lock is stable and could replace per node lru lock. In isolate_migratepages_block(), compact_unlock_should_abort and lock_page_lruvec_irqsave are open coded to work with compact_control. Also add a debug func in locking which may give some clues if there are sth out of hands. Daniel Jordan's testing show 62% improvement on modified readtwice case on his 2P * 10 core * 2 HT broadwell box. https://lore.kernel.org/lkml/20200915165807.kpp7uhiw7l3loofu@ca-dmjordan1.us.oracle.com/ Hugh Dickins helped on the patch polish, thanks! [alex.shi@linux.alibaba.com: fix comment typo] Link: https://lkml.kernel.org/r/5b085715-292a-4b43-50b3-d73dc90d1de5@linux.alibaba.com [alex.shi@linux.alibaba.com: use page_memcg()] Link: https://lkml.kernel.org/r/5a4c2b72-7ee8-2478-fc0e-85eb83aafec4@linux.alibaba.com Link: https://lkml.kernel.org/r/1604566549-62481-18-git-send-email-alex.shi@linux.alibaba.com Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Rong Chen <rong.a.chen@intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Yang <richard.weiyang@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:29 +00:00
unlock_page_lruvec_irqrestore(locked, flags);
mm/compaction: do page isolation first in compaction Currently, compaction would get the lru_lock and then do page isolation which works fine with pgdat->lru_lock, since any page isoltion would compete for the lru_lock. If we want to change to memcg lru_lock, we have to isolate the page before getting lru_lock, thus isoltion would block page's memcg change which relay on page isoltion too. Then we could safely use per memcg lru_lock later. The new page isolation use previous introduced TestClearPageLRU() + pgdat lru locking which will be changed to memcg lru lock later. Hugh Dickins <hughd@google.com> fixed following bugs in this patch's early version: Fix lots of crashes under compaction load: isolate_migratepages_block() must clean up appropriately when rejecting a page, setting PageLRU again if it had been cleared; and a put_page() after get_page_unless_zero() cannot safely be done while holding locked_lruvec - it may turn out to be the final put_page(), which will take an lruvec lock when PageLRU. And move __isolate_lru_page_prepare back after get_page_unless_zero to make trylock_page() safe: trylock_page() is not safe to use at this time: its setting PG_locked can race with the page being freed or allocated ("Bad page"), and can also erase flags being set by one of those "sole owners" of a freshly allocated page who use non-atomic __SetPageFlag(). Link: https://lkml.kernel.org/r/1604566549-62481-16-git-send-email-alex.shi@linux.alibaba.com Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com> Acked-by: Hugh Dickins <hughd@google.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexander Duyck <alexander.duyck@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: "Chen, Rong A" <rong.a.chen@intel.com> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Jann Horn <jannh@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttilä <mika.penttila@nextfour.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Cc: Yang Shi <yang.shi@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-12-15 20:34:20 +00:00
if (page) {
SetPageLRU(page);
put_page(page);
}
/*
mm, compaction: avoid rescanning the same pageblock multiple times Pageblocks are marked for skip when no pages are isolated after a scan. However, it's possible to hit corner cases where the migration scanner gets stuck near the boundary between the source and target scanner. Due to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are migrated can be reallocated before the pageblock is complete. The pageblock is not necessarily skipped so it can be rescanned multiple times. Similarly, a pageblock with some dirty/writeback pages may fail to migrate and be rescanned until writeback completes which is wasteful. This patch tracks if a pageblock is being rescanned. If so, then the entire pageblock will be migrated as one operation. This narrows the race window during which pages can be reallocated during migration. Secondly, if there are pages that cannot be isolated then the pageblock will still be fully scanned and marked for skipping. On the second rescan, the pageblock skip is set and the migration scanner makes progress. 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3200.68 ( 0.00%) 3002.07 ( 6.21%) Amean fault-both-5 4847.75 ( 0.00%) 4684.47 ( 3.37%) Amean fault-both-7 6658.92 ( 0.00%) 6815.54 ( -2.35%) Amean fault-both-12 11077.62 ( 0.00%) 10864.02 ( 1.93%) Amean fault-both-18 12403.97 ( 0.00%) 12247.52 ( 1.26%) Amean fault-both-24 15607.10 ( 0.00%) 15683.99 ( -0.49%) Amean fault-both-30 18752.27 ( 0.00%) 18620.02 ( 0.71%) Amean fault-both-32 21207.54 ( 0.00%) 19250.28 * 9.23%* 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Percentage huge-3 96.86 ( 0.00%) 95.00 ( -1.91%) Percentage huge-5 93.72 ( 0.00%) 94.22 ( 0.53%) Percentage huge-7 94.31 ( 0.00%) 92.35 ( -2.08%) Percentage huge-12 92.66 ( 0.00%) 91.90 ( -0.82%) Percentage huge-18 91.51 ( 0.00%) 89.58 ( -2.11%) Percentage huge-24 90.50 ( 0.00%) 90.03 ( -0.52%) Percentage huge-30 91.57 ( 0.00%) 89.14 ( -2.65%) Percentage huge-32 91.00 ( 0.00%) 90.58 ( -0.46%) Negligible difference but this was likely a case when the specific corner case was not hit. A previous run of the same patch based on an earlier iteration of the series showed large differences where migration rates could be halved when the corner case was hit. The specific corner case where migration scan rates go through the roof was due to a dirty/writeback pageblock located at the boundary of the migration/free scanner did not happen in this case. When it does happen, the scan rates multipled by massive margins. Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:07 +00:00
* Updated the cached scanner pfn once the pageblock has been scanned
* Pages will either be migrated in which case there is no point
* scanning in the near future or migration failed in which case the
* failure reason may persist. The block is marked for skipping if
* there were no pages isolated in the block or if the block is
* rescanned twice in a row.
*/
mm, compaction: avoid rescanning the same pageblock multiple times Pageblocks are marked for skip when no pages are isolated after a scan. However, it's possible to hit corner cases where the migration scanner gets stuck near the boundary between the source and target scanner. Due to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are migrated can be reallocated before the pageblock is complete. The pageblock is not necessarily skipped so it can be rescanned multiple times. Similarly, a pageblock with some dirty/writeback pages may fail to migrate and be rescanned until writeback completes which is wasteful. This patch tracks if a pageblock is being rescanned. If so, then the entire pageblock will be migrated as one operation. This narrows the race window during which pages can be reallocated during migration. Secondly, if there are pages that cannot be isolated then the pageblock will still be fully scanned and marked for skipping. On the second rescan, the pageblock skip is set and the migration scanner makes progress. 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3200.68 ( 0.00%) 3002.07 ( 6.21%) Amean fault-both-5 4847.75 ( 0.00%) 4684.47 ( 3.37%) Amean fault-both-7 6658.92 ( 0.00%) 6815.54 ( -2.35%) Amean fault-both-12 11077.62 ( 0.00%) 10864.02 ( 1.93%) Amean fault-both-18 12403.97 ( 0.00%) 12247.52 ( 1.26%) Amean fault-both-24 15607.10 ( 0.00%) 15683.99 ( -0.49%) Amean fault-both-30 18752.27 ( 0.00%) 18620.02 ( 0.71%) Amean fault-both-32 21207.54 ( 0.00%) 19250.28 * 9.23%* 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Percentage huge-3 96.86 ( 0.00%) 95.00 ( -1.91%) Percentage huge-5 93.72 ( 0.00%) 94.22 ( 0.53%) Percentage huge-7 94.31 ( 0.00%) 92.35 ( -2.08%) Percentage huge-12 92.66 ( 0.00%) 91.90 ( -0.82%) Percentage huge-18 91.51 ( 0.00%) 89.58 ( -2.11%) Percentage huge-24 90.50 ( 0.00%) 90.03 ( -0.52%) Percentage huge-30 91.57 ( 0.00%) 89.14 ( -2.65%) Percentage huge-32 91.00 ( 0.00%) 90.58 ( -0.46%) Negligible difference but this was likely a case when the specific corner case was not hit. A previous run of the same patch based on an earlier iteration of the series showed large differences where migration rates could be halved when the corner case was hit. The specific corner case where migration scan rates go through the roof was due to a dirty/writeback pageblock located at the boundary of the migration/free scanner did not happen in this case. When it does happen, the scan rates multipled by massive margins. Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:07 +00:00
if (low_pfn == end_pfn && (!nr_isolated || cc->rescan)) {
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
if (valid_page && !skip_updated)
set_pageblock_skip(valid_page);
update_cached_migrate(cc, low_pfn);
}
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
trace_mm_compaction_isolate_migratepages(start_pfn, low_pfn,
nr_scanned, nr_isolated);
mm: compaction: avoid 100% CPU usage during compaction when a task is killed "howaboutsynergy" reported via kernel buzilla number 204165 that compact_zone_order was consuming 100% CPU during a stress test for prolonged periods of time. Specifically the following command, which should exit in 10 seconds, was taking an excessive time to finish while the CPU was pegged at 100%. stress -m 220 --vm-bytes 1000000000 --timeout 10 Tracing indicated a pattern as follows stress-3923 [007] 519.106208: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106212: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106216: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106219: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106223: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106227: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106231: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106235: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106238: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 stress-3923 [007] 519.106242: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0 Note that compaction is entered in rapid succession while scanning and isolating nothing. The problem is that when a task that is compacting receives a fatal signal, it retries indefinitely instead of exiting while making no progress as a fatal signal is pending. It's not easy to trigger this condition although enabling zswap helps on the basis that the timing is altered. A very small window has to be hit for the problem to occur (signal delivered while compacting and isolating a PFN for migration that is not aligned to SWAP_CLUSTER_MAX). This was reproduced locally -- 16G single socket system, 8G swap, 30% zswap configured, vm-bytes 22000000000 using Colin Kings stress-ng implementation from github running in a loop until the problem hits). Tracing recorded the problem occurring almost 200K times in a short window. With this patch, the problem hit 4 times but the task existed normally instead of consuming CPU. This problem has existed for some time but it was made worse by commit cf66f0700c8f ("mm, compaction: do not consider a need to reschedule as contention"). Before that commit, if the same condition was hit then locks would be quickly contended and compaction would exit that way. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=204165 Link: http://lkml.kernel.org/r/20190718085708.GE24383@techsingularity.net Fixes: cf66f0700c8f ("mm, compaction: do not consider a need to reschedule as contention") Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> [5.1+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-08-03 04:48:51 +00:00
fatal_pending:
cc->total_migrate_scanned += nr_scanned;
mm: compaction: Add scanned and isolated counters for compaction Compaction already has tracepoints to count scanned and isolated pages but it requires that ftrace be enabled and if that information has to be written to disk then it can be disruptive. This patch adds vmstat counters for compaction called compact_migrate_scanned, compact_free_scanned and compact_isolated. With these counters, it is possible to define a basic cost model for compaction. This approximates of how much work compaction is doing and can be compared that with an oprofile showing TLB misses and see if the cost of compaction is being offset by THP for example. Minimally a compaction patch can be evaluated in terms of whether it increases or decreases cost. The basic cost model looks like this Fundamental unit u: a word sizeof(void *) Ca = cost of struct page access = sizeof(struct page) / u Cmc = Cost migrate page copy = (Ca + PAGE_SIZE/u) * 2 Cmf = Cost migrate failure = Ca * 2 Ci = Cost page isolation = (Ca + Wi) where Wi is a constant that should reflect the approximate cost of the locking operation. Csm = Cost migrate scanning = Ca Csf = Cost free scanning = Ca Overall cost = (Csm * compact_migrate_scanned) + (Csf * compact_free_scanned) + (Ci * compact_isolated) + (Cmc * pgmigrate_success) + (Cmf * pgmigrate_failed) Where the values are read from /proc/vmstat. This is very basic and ignores certain costs such as the allocation cost to do a migrate page copy but any improvement to the model would still use the same vmstat counters. Signed-off-by: Mel Gorman <mgorman@suse.de> Reviewed-by: Rik van Riel <riel@redhat.com>
2012-10-19 11:00:10 +00:00
if (nr_isolated)
count_compact_events(COMPACTISOLATED, nr_isolated);
mm: compaction: Add scanned and isolated counters for compaction Compaction already has tracepoints to count scanned and isolated pages but it requires that ftrace be enabled and if that information has to be written to disk then it can be disruptive. This patch adds vmstat counters for compaction called compact_migrate_scanned, compact_free_scanned and compact_isolated. With these counters, it is possible to define a basic cost model for compaction. This approximates of how much work compaction is doing and can be compared that with an oprofile showing TLB misses and see if the cost of compaction is being offset by THP for example. Minimally a compaction patch can be evaluated in terms of whether it increases or decreases cost. The basic cost model looks like this Fundamental unit u: a word sizeof(void *) Ca = cost of struct page access = sizeof(struct page) / u Cmc = Cost migrate page copy = (Ca + PAGE_SIZE/u) * 2 Cmf = Cost migrate failure = Ca * 2 Ci = Cost page isolation = (Ca + Wi) where Wi is a constant that should reflect the approximate cost of the locking operation. Csm = Cost migrate scanning = Ca Csf = Cost free scanning = Ca Overall cost = (Csm * compact_migrate_scanned) + (Csf * compact_free_scanned) + (Ci * compact_isolated) + (Cmc * pgmigrate_success) + (Cmf * pgmigrate_failed) Where the values are read from /proc/vmstat. This is very basic and ignores certain costs such as the allocation cost to do a migrate page copy but any improvement to the model would still use the same vmstat counters. Signed-off-by: Mel Gorman <mgorman@suse.de> Reviewed-by: Rik van Riel <riel@redhat.com>
2012-10-19 11:00:10 +00:00
cc->migrate_pfn = low_pfn;
return ret;
}
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
/**
* isolate_migratepages_range() - isolate migrate-able pages in a PFN range
* @cc: Compaction control structure.
* @start_pfn: The first PFN to start isolating.
* @end_pfn: The one-past-last PFN.
*
mm: make alloc_contig_range handle free hugetlb pages alloc_contig_range will fail if it ever sees a HugeTLB page within the range we are trying to allocate, even when that page is free and can be easily reallocated. This has proved to be problematic for some users of alloc_contic_range, e.g: CMA and virtio-mem, where those would fail the call even when those pages lay in ZONE_MOVABLE and are free. We can do better by trying to replace such page. Free hugepages are tricky to handle so as to no userspace application notices disruption, we need to replace the current free hugepage with a new one. In order to do that, a new function called alloc_and_dissolve_huge_page is introduced. This function will first try to get a new fresh hugepage, and if it succeeds, it will replace the old one in the free hugepage pool. The free page replacement is done under hugetlb_lock, so no external users of hugetlb will notice the change. To allocate the new huge page, we use alloc_buddy_huge_page(), so we do not have to deal with any counters, and prep_new_huge_page() is not called. This is valulable because in case we need to free the new page, we only need to call __free_pages(). Once we know that the page to be replaced is a genuine 0-refcounted huge page, we remove the old page from the freelist by remove_hugetlb_page(). Then, we can call __prep_new_huge_page() and __prep_account_new_huge_page() for the new huge page to properly initialize it and increment the hstate->nr_huge_pages counter (previously decremented by remove_hugetlb_page()). Once done, the page is enqueued by enqueue_huge_page() and it is ready to be used. There is one tricky case when page's refcount is 0 because it is in the process of being released. A missing PageHugeFreed bit will tell us that freeing is in flight so we retry after dropping the hugetlb_lock. The race window should be small and the next retry should make a forward progress. E.g: CPU0 CPU1 free_huge_page() isolate_or_dissolve_huge_page PageHuge() == T alloc_and_dissolve_huge_page alloc_buddy_huge_page() spin_lock_irq(hugetlb_lock) // PageHuge() && !PageHugeFreed && // !PageCount() spin_unlock_irq(hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) update_and_free_page PageHuge() == F __free_pages() 2) enqueue_huge_page SetPageHugeFreed() spin_unlock_irq(&hugetlb_lock) spin_lock_irq(hugetlb_lock) 1) PageHuge() == F (freed by case#1 from CPU0) 2) PageHuge() == T PageHugeFreed() == T - proceed with replacing the page In the case above we retry as the window race is quite small and we have high chances to succeed next time. With regard to the allocation, we restrict it to the node the page belongs to with __GFP_THISNODE, meaning we do not fallback on other node's zones. Note that gigantic hugetlb pages are fenced off since there is a cyclic dependency between them and alloc_contig_range. Link: https://lkml.kernel.org/r/20210419075413.1064-6-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:35:26 +00:00
* Returns -EAGAIN when contented, -EINTR in case of a signal pending, -ENOMEM
* in case we could not allocate a page, or 0.
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
*/
int
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
isolate_migratepages_range(struct compact_control *cc, unsigned long start_pfn,
unsigned long end_pfn)
{
unsigned long pfn, block_start_pfn, block_end_pfn;
int ret = 0;
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
/* Scan block by block. First and last block may be incomplete */
pfn = start_pfn;
block_start_pfn = pageblock_start_pfn(pfn);
if (block_start_pfn < cc->zone->zone_start_pfn)
block_start_pfn = cc->zone->zone_start_pfn;
block_end_pfn = pageblock_end_pfn(pfn);
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
for (; pfn < end_pfn; pfn = block_end_pfn,
block_start_pfn = block_end_pfn,
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
block_end_pfn += pageblock_nr_pages) {
block_end_pfn = min(block_end_pfn, end_pfn);
if (!pageblock_pfn_to_page(block_start_pfn,
block_end_pfn, cc->zone))
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
continue;
ret = isolate_migratepages_block(cc, pfn, block_end_pfn,
ISOLATE_UNEVICTABLE);
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
if (ret)
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
break;
mm/compaction.c: avoid premature range skip in isolate_migratepages_range Commit edc2ca612496 ("mm, compaction: move pageblock checks up from isolate_migratepages_range()") commonizes isolate_migratepages variants and make them use isolate_migratepages_block(). isolate_migratepages_block() could stop the execution when enough pages are isolated, but, there is no code in isolate_migratepages_range() to handle this case. In the result, even if isolate_migratepages_block() returns prematurely without checking all pages in the range, isolate_migratepages_block() is called repeately on the following pageblock and some pages in the previous range are skipped to check. Then, CMA is failed frequently due to this fact. To fix this problem, this patch let isolate_migratepages_range() know the situation that enough pages are isolated and stop the isolation in that case. Note that isolate_migratepages() has no such problem, because, it always stops the isolation after just one call of isolate_migratepages_block(). Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-29 21:50:20 +00:00
mm/compaction: count pages and stop correctly during page isolation In isolate_migratepages_block, when cc->alloc_contig is true, we are able to isolate compound pages. But nr_migratepages and nr_isolated did not count compound pages correctly, causing us to isolate more pages than we thought. So count compound pages as the number of base pages they contain. Otherwise, we might be trapped in too_many_isolated while loop, since the actual isolated pages can go up to COMPACT_CLUSTER_MAX*512=16384, where COMPACT_CLUSTER_MAX is 32, since we stop isolation after cc->nr_migratepages reaches to COMPACT_CLUSTER_MAX. In addition, after we fix the issue above, cc->nr_migratepages could never be equal to COMPACT_CLUSTER_MAX if compound pages are isolated, thus page isolation could not stop as we intended. Change the isolation stop condition to '>='. The issue can be triggered as follows: In a system with 16GB memory and an 8GB CMA region reserved by hugetlb_cma, if we first allocate 10GB THPs and mlock them (so some THPs are allocated in the CMA region and mlocked), reserving 6 1GB hugetlb pages via /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages will get stuck (looping in too_many_isolated function) until we kill either task. With the patch applied, oom will kill the application with 10GB THPs and let hugetlb page reservation finish. [ziy@nvidia.com: v3] Link: https://lkml.kernel.org/r/20201030183809.3616803-1-zi.yan@sent.com Fixes: 1da2f328fa64 ("cmm,thp,compaction,cma: allow THP migration for CMA allocations") Signed-off-by: Zi Yan <ziy@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Yang Shi <shy828301@gmail.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@surriel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: <stable@vger.kernel.org> Link: https://lkml.kernel.org/r/20201029200435.3386066-1-zi.yan@sent.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-11-14 06:51:40 +00:00
if (cc->nr_migratepages >= COMPACT_CLUSTER_MAX)
mm/compaction.c: avoid premature range skip in isolate_migratepages_range Commit edc2ca612496 ("mm, compaction: move pageblock checks up from isolate_migratepages_range()") commonizes isolate_migratepages variants and make them use isolate_migratepages_block(). isolate_migratepages_block() could stop the execution when enough pages are isolated, but, there is no code in isolate_migratepages_range() to handle this case. In the result, even if isolate_migratepages_block() returns prematurely without checking all pages in the range, isolate_migratepages_block() is called repeately on the following pageblock and some pages in the previous range are skipped to check. Then, CMA is failed frequently due to this fact. To fix this problem, this patch let isolate_migratepages_range() know the situation that enough pages are isolated and stop the isolation in that case. Note that isolate_migratepages() has no such problem, because, it always stops the isolation after just one call of isolate_migratepages_block(). Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-29 21:50:20 +00:00
break;
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
}
return ret;
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
}
#endif /* CONFIG_COMPACTION || CONFIG_CMA */
#ifdef CONFIG_COMPACTION
static bool suitable_migration_source(struct compact_control *cc,
struct page *page)
{
mm, compaction: restrict async compaction to pageblocks of same migratetype The migrate scanner in async compaction is currently limited to MIGRATE_MOVABLE pageblocks. This is a heuristic intended to reduce latency, based on the assumption that non-MOVABLE pageblocks are unlikely to contain movable pages. However, with the exception of THP's, most high-order allocations are not movable. Should the async compaction succeed, this increases the chance that the non-MOVABLE allocations will fallback to a MOVABLE pageblock, making the long-term fragmentation worse. This patch attempts to help the situation by changing async direct compaction so that the migrate scanner only scans the pageblocks of the requested migratetype. If it's a non-MOVABLE type and there are such pageblocks that do contain movable pages, chances are that the allocation can succeed within one of such pageblocks, removing the need for a fallback. If that fails, the subsequent sync attempt will ignore this restriction. In testing based on 4.9 kernel with stress-highalloc from mmtests configured for order-4 GFP_KERNEL allocations, this patch has reduced the number of unmovable allocations falling back to movable pageblocks by 30%. The number of movable allocations falling back is reduced by 12%. Link: http://lkml.kernel.org/r/20170307131545.28577-8-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:54:49 +00:00
int block_mt;
if (pageblock_skip_persistent(page))
return false;
mm, compaction: restrict async compaction to pageblocks of same migratetype The migrate scanner in async compaction is currently limited to MIGRATE_MOVABLE pageblocks. This is a heuristic intended to reduce latency, based on the assumption that non-MOVABLE pageblocks are unlikely to contain movable pages. However, with the exception of THP's, most high-order allocations are not movable. Should the async compaction succeed, this increases the chance that the non-MOVABLE allocations will fallback to a MOVABLE pageblock, making the long-term fragmentation worse. This patch attempts to help the situation by changing async direct compaction so that the migrate scanner only scans the pageblocks of the requested migratetype. If it's a non-MOVABLE type and there are such pageblocks that do contain movable pages, chances are that the allocation can succeed within one of such pageblocks, removing the need for a fallback. If that fails, the subsequent sync attempt will ignore this restriction. In testing based on 4.9 kernel with stress-highalloc from mmtests configured for order-4 GFP_KERNEL allocations, this patch has reduced the number of unmovable allocations falling back to movable pageblocks by 30%. The number of movable allocations falling back is reduced by 12%. Link: http://lkml.kernel.org/r/20170307131545.28577-8-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:54:49 +00:00
if ((cc->mode != MIGRATE_ASYNC) || !cc->direct_compaction)
return true;
mm, compaction: restrict async compaction to pageblocks of same migratetype The migrate scanner in async compaction is currently limited to MIGRATE_MOVABLE pageblocks. This is a heuristic intended to reduce latency, based on the assumption that non-MOVABLE pageblocks are unlikely to contain movable pages. However, with the exception of THP's, most high-order allocations are not movable. Should the async compaction succeed, this increases the chance that the non-MOVABLE allocations will fallback to a MOVABLE pageblock, making the long-term fragmentation worse. This patch attempts to help the situation by changing async direct compaction so that the migrate scanner only scans the pageblocks of the requested migratetype. If it's a non-MOVABLE type and there are such pageblocks that do contain movable pages, chances are that the allocation can succeed within one of such pageblocks, removing the need for a fallback. If that fails, the subsequent sync attempt will ignore this restriction. In testing based on 4.9 kernel with stress-highalloc from mmtests configured for order-4 GFP_KERNEL allocations, this patch has reduced the number of unmovable allocations falling back to movable pageblocks by 30%. The number of movable allocations falling back is reduced by 12%. Link: http://lkml.kernel.org/r/20170307131545.28577-8-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:54:49 +00:00
block_mt = get_pageblock_migratetype(page);
if (cc->migratetype == MIGRATE_MOVABLE)
return is_migrate_movable(block_mt);
else
return block_mt == cc->migratetype;
}
/* Returns true if the page is within a block suitable for migration to */
2016-10-08 00:00:37 +00:00
static bool suitable_migration_target(struct compact_control *cc,
struct page *page)
{
/* If the page is a large free page, then disallow migration */
if (PageBuddy(page)) {
/*
* We are checking page_order without zone->lock taken. But
* the only small danger is that we skip a potentially suitable
* pageblock, so it's not worth to check order for valid range.
*/
if (buddy_order_unsafe(page) >= pageblock_order)
return false;
}
if (cc->ignore_block_suitable)
return true;
/* If the block is MIGRATE_MOVABLE or MIGRATE_CMA, allow migration */
if (is_migrate_movable(get_pageblock_migratetype(page)))
return true;
/* Otherwise skip the block */
return false;
}
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
static inline unsigned int
freelist_scan_limit(struct compact_control *cc)
{
unsigned short shift = BITS_PER_LONG - 1;
return (COMPACT_CLUSTER_MAX >> min(shift, cc->fast_search_fail)) + 1;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
}
mm, compaction: more robust check for scanners meeting Assorted compaction cleanups and optimizations. The interesting patches are 4 and 5. In 4, skipping of compound pages in single iteration is improved for migration scanner, so it works also for !PageLRU compound pages such as hugetlbfs, slab etc. Patch 5 introduces this kind of skipping in the free scanner. The trick is that we can read compound_order() without any protection, if we are careful to filter out values larger than MAX_ORDER. The only danger is that we skip too much. The same trick was already used for reading the freepage order in the migrate scanner. To demonstrate improvements of Patches 4 and 5 I've run stress-highalloc from mmtests, set to simulate THP allocations (including __GFP_COMP) on a 4GB system where 1GB was occupied by hugetlbfs pages. I'll include just the relevant stats: Patch 3 Patch 4 Patch 5 Compaction stalls 7523 7529 7515 Compaction success 323 304 322 Compaction failures 7200 7224 7192 Page migrate success 247778 264395 240737 Page migrate failure 15358 33184 21621 Compaction pages isolated 906928 980192 909983 Compaction migrate scanned 2005277 1692805 1498800 Compaction free scanned 13255284 11539986 9011276 Compaction cost 288 305 277 With 5 iterations per patch, the results are still noisy, but we can see that Patch 4 does reduce migrate_scanned by 15% thanks to skipping the hugetlbfs pages at once. Interestingly, free_scanned is also reduced and I have no idea why. Patch 5 further reduces free_scanned as expected, by 15%. Other stats are unaffected modulo noise. [1] https://lkml.org/lkml/2015/1/19/158 This patch (of 5): Compaction should finish when the migration and free scanner meet, i.e. they reach the same pageblock. Currently however, the test in compact_finished() simply just compares the exact pfns, which may yield a false negative when the free scanner position is in the middle of a pageblock and the migration scanner reaches the begining of the same pageblock. This hasn't been a problem until commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner") allowed the free scanner position to be in the middle of a pageblock between invocations. The hot-fix 1d5bfe1ffb5b ("mm, compaction: prevent infinite loop in compact_zone") prevented the issue by adding a special check in the migration scanner to satisfy the current detection of scanners meeting. However, the proper fix is to make the detection more robust. This patch introduces the compact_scanners_met() function that returns true when the free scanner position is in the same or lower pageblock than the migration scanner. The special case in isolate_migratepages() introduced by 1d5bfe1ffb5b is removed. Suggested-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:36 +00:00
/*
* Test whether the free scanner has reached the same or lower pageblock than
* the migration scanner, and compaction should thus terminate.
*/
static inline bool compact_scanners_met(struct compact_control *cc)
{
return (cc->free_pfn >> pageblock_order)
<= (cc->migrate_pfn >> pageblock_order);
}
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
/*
* Used when scanning for a suitable migration target which scans freelists
* in reverse. Reorders the list such as the unscanned pages are scanned
* first on the next iteration of the free scanner
*/
static void
move_freelist_head(struct list_head *freelist, struct page *freepage)
{
LIST_HEAD(sublist);
if (!list_is_last(freelist, &freepage->lru)) {
list_cut_before(&sublist, freelist, &freepage->lru);
if (!list_empty(&sublist))
list_splice_tail(&sublist, freelist);
}
}
/*
* Similar to move_freelist_head except used by the migration scanner
* when scanning forward. It's possible for these list operations to
* move against each other if they search the free list exactly in
* lockstep.
*/
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
static void
move_freelist_tail(struct list_head *freelist, struct page *freepage)
{
LIST_HEAD(sublist);
if (!list_is_first(freelist, &freepage->lru)) {
list_cut_position(&sublist, freelist, &freepage->lru);
if (!list_empty(&sublist))
list_splice_tail(&sublist, freelist);
}
}
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
static void
fast_isolate_around(struct compact_control *cc, unsigned long pfn, unsigned long nr_isolated)
{
unsigned long start_pfn, end_pfn;
mm, compaction: make fast_isolate_freepages() stay within zone Compaction always operates on pages from a single given zone when isolating both pages to migrate and freepages. Pageblock boundaries are intersected with zone boundaries to be safe in case zone starts or ends in the middle of pageblock. The use of pageblock_pfn_to_page() protects against non-contiguous pageblocks. The functions fast_isolate_freepages() and fast_isolate_around() don't currently protect the fast freepage isolation thoroughly enough against these corner cases, and can result in freepage isolation operate outside of zone boundaries: - in fast_isolate_freepages() if we get a pfn from the first pageblock of a zone that starts in the middle of that pageblock, 'highest' can be a pfn outside of the zone. If we fail to isolate anything in this function, we may then call fast_isolate_around() on a pfn outside of the zone and there effectively do a set_pageblock_skip(page_to_pfn(highest)) which may currently hit a VM_BUG_ON() in some configurations - fast_isolate_around() checks only the zone end boundary and not beginning, nor that the pageblock is contiguous (with pageblock_pfn_to_page()) so it's possible that we end up calling isolate_freepages_block() on a range of pfn's from two different zones and end up e.g. isolating freepages under the wrong zone's lock. This patch should fix the above issues. Link: https://lkml.kernel.org/r/20210217173300.6394-1-vbabka@suse.cz Fixes: 5a811889de10 ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Hildenbrand <david@redhat.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Rapoport <rppt@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:39 +00:00
struct page *page;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
/* Do not search around if there are enough pages already */
if (cc->nr_freepages >= cc->nr_migratepages)
return;
/* Minimise scanning during async compaction */
if (cc->direct_compaction && cc->mode == MIGRATE_ASYNC)
return;
/* Pageblock boundaries */
mm, compaction: make fast_isolate_freepages() stay within zone Compaction always operates on pages from a single given zone when isolating both pages to migrate and freepages. Pageblock boundaries are intersected with zone boundaries to be safe in case zone starts or ends in the middle of pageblock. The use of pageblock_pfn_to_page() protects against non-contiguous pageblocks. The functions fast_isolate_freepages() and fast_isolate_around() don't currently protect the fast freepage isolation thoroughly enough against these corner cases, and can result in freepage isolation operate outside of zone boundaries: - in fast_isolate_freepages() if we get a pfn from the first pageblock of a zone that starts in the middle of that pageblock, 'highest' can be a pfn outside of the zone. If we fail to isolate anything in this function, we may then call fast_isolate_around() on a pfn outside of the zone and there effectively do a set_pageblock_skip(page_to_pfn(highest)) which may currently hit a VM_BUG_ON() in some configurations - fast_isolate_around() checks only the zone end boundary and not beginning, nor that the pageblock is contiguous (with pageblock_pfn_to_page()) so it's possible that we end up calling isolate_freepages_block() on a range of pfn's from two different zones and end up e.g. isolating freepages under the wrong zone's lock. This patch should fix the above issues. Link: https://lkml.kernel.org/r/20210217173300.6394-1-vbabka@suse.cz Fixes: 5a811889de10 ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Hildenbrand <david@redhat.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Rapoport <rppt@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:39 +00:00
start_pfn = max(pageblock_start_pfn(pfn), cc->zone->zone_start_pfn);
end_pfn = min(pageblock_end_pfn(pfn), zone_end_pfn(cc->zone));
page = pageblock_pfn_to_page(start_pfn, end_pfn, cc->zone);
if (!page)
return;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
/* Scan before */
if (start_pfn != pfn) {
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
isolate_freepages_block(cc, &start_pfn, pfn, &cc->freepages, 1, false);
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
if (cc->nr_freepages >= cc->nr_migratepages)
return;
}
/* Scan after */
start_pfn = pfn + nr_isolated;
mm/compaction.c: correct zone boundary handling when isolating pages from a pageblock syzbot reported the following error from a tree with a head commit of baf76f0c58ae ("slip: make slhc_free() silently accept an error pointer") BUG: unable to handle kernel paging request at ffffea0003348000 #PF error: [normal kernel read fault] PGD 12c3f9067 P4D 12c3f9067 PUD 12c3f8067 PMD 0 Oops: 0000 [#1] PREEMPT SMP KASAN CPU: 1 PID: 28916 Comm: syz-executor.2 Not tainted 5.1.0-rc6+ #89 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:constant_test_bit arch/x86/include/asm/bitops.h:314 [inline] RIP: 0010:PageCompound include/linux/page-flags.h:186 [inline] RIP: 0010:isolate_freepages_block+0x1c0/0xd40 mm/compaction.c:579 Code: 01 d8 ff 4d 85 ed 0f 84 ef 07 00 00 e8 29 00 d8 ff 4c 89 e0 83 85 38 ff ff ff 01 48 c1 e8 03 42 80 3c 38 00 0f 85 31 0a 00 00 <4d> 8b 2c 24 31 ff 49 c1 ed 10 41 83 e5 01 44 89 ee e8 3a 01 d8 ff RSP: 0018:ffff88802b31eab8 EFLAGS: 00010246 RAX: 1ffffd4000669000 RBX: 00000000000cd200 RCX: ffffc9000a235000 RDX: 000000000001ca5e RSI: ffffffff81988cc7 RDI: 0000000000000001 RBP: ffff88802b31ebd8 R08: ffff88805af700c0 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffffea0003348000 R13: 0000000000000000 R14: ffff88802b31f030 R15: dffffc0000000000 FS: 00007f61648dc700(0000) GS:ffff8880ae900000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffea0003348000 CR3: 0000000037c64000 CR4: 00000000001426e0 Call Trace: fast_isolate_around mm/compaction.c:1243 [inline] fast_isolate_freepages mm/compaction.c:1418 [inline] isolate_freepages mm/compaction.c:1438 [inline] compaction_alloc+0x1aee/0x22e0 mm/compaction.c:1550 There is no reproducer and it is difficult to hit -- 1 crash every few days. The issue is very similar to the fix in commit 6b0868c820ff ("mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints"). When isolating free pages around a target pageblock, the boundary handling is off by one and can stray into the next pageblock. Triggering the syzbot error requires that the end of pageblock is section or zone aligned, and that the next section is unpopulated. A more subtle consequence of the bug is that pageblocks were being improperly used as migration targets which potentially hurts fragmentation avoidance in the long-term one page at a time. A debugging patch revealed that it's definitely possible to stray outside of a pageblock which is not intended. While syzbot cannot be used to verify this patch, it was confirmed that the debugging warning no longer triggers with this patch applied. It has also been confirmed that the THP allocation stress tests are not degraded by this patch. Link: http://lkml.kernel.org/r/20190510182124.GI18914@techsingularity.net Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints") Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Reported-by: syzbot+d84c80f9fe26a0f7a734@syzkaller.appspotmail.com Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: Qian Cai <cai@lca.pw> Cc: Michal Hocko <mhocko@suse.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> # v5.1+ Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-17 21:31:41 +00:00
if (start_pfn < end_pfn)
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
isolate_freepages_block(cc, &start_pfn, end_pfn, &cc->freepages, 1, false);
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
/* Skip this pageblock in the future as it's full or nearly full */
if (cc->nr_freepages < cc->nr_migratepages)
set_pageblock_skip(page);
}
/* Search orders in round-robin fashion */
static int next_search_order(struct compact_control *cc, int order)
{
order--;
if (order < 0)
order = cc->order - 1;
/* Search wrapped around? */
if (order == cc->search_order) {
cc->search_order--;
if (cc->search_order < 0)
cc->search_order = cc->order - 1;
return -1;
}
return order;
}
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
static unsigned long
fast_isolate_freepages(struct compact_control *cc)
{
unsigned int limit = min(1U, freelist_scan_limit(cc) >> 1);
unsigned int nr_scanned = 0;
mm, compaction: move high_pfn to the for loop scope In fast_isolate_freepages, high_pfn will be used if a prefered one (ie PFN >= low_fn) not found. But the high_pfn is not reset before searching an free area, so when it was used as freepage, it may from another free area searched before. As a result move_freelist_head(freelist, freepage) will have unexpected behavior (eg corrupt the MOVABLE freelist) Unable to handle kernel paging request at virtual address dead000000000200 Mem abort info: ESR = 0x96000044 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000044 CM = 0, WnR = 1 [dead000000000200] address between user and kernel address ranges -000|list_cut_before(inline) -000|move_freelist_head(inline) -000|fast_isolate_freepages(inline) -000|isolate_freepages(inline) -000|compaction_alloc(?, ?) -001|unmap_and_move(inline) -001|migrate_pages([NSD:0xFFFFFF80088CBBD0] from = 0xFFFFFF80088CBD88, [NSD:0xFFFFFF80088CBBC8] get_new_p -002|__read_once_size(inline) -002|static_key_count(inline) -002|static_key_false(inline) -002|trace_mm_compaction_migratepages(inline) -002|compact_zone(?, [NSD:0xFFFFFF80088CBCB0] capc = 0x0) -003|kcompactd_do_work(inline) -003|kcompactd([X19] p = 0xFFFFFF93227FBC40) -004|kthread([X20] _create = 0xFFFFFFE1AFB26380) -005|ret_from_fork(asm) The issue was reported on an smart phone product with 6GB ram and 3GB zram as swap device. This patch fixes the issue by reset high_pfn before searching each free area, which ensure freepage and freelist match when call move_freelist_head in fast_isolate_freepages(). Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Link: https://lkml.kernel.org/r/20210112094720.1238444-1-wu-yan@tcl.com Fixes: 5a811889de10f1eb ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Rokudo Yan <wu-yan@tcl.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-05 02:32:20 +00:00
unsigned long low_pfn, min_pfn, highest = 0;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
unsigned long nr_isolated = 0;
unsigned long distance;
struct page *page = NULL;
bool scan_start = false;
int order;
/* Full compaction passes in a negative order */
if (cc->order <= 0)
return cc->free_pfn;
/*
* If starting the scan, use a deeper search and use the highest
* PFN found if a suitable one is not found.
*/
2019-03-05 23:45:38 +00:00
if (cc->free_pfn >= cc->zone->compact_init_free_pfn) {
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
limit = pageblock_nr_pages >> 1;
scan_start = true;
}
/*
* Preferred point is in the top quarter of the scan space but take
* a pfn from the top half if the search is problematic.
*/
distance = (cc->free_pfn - cc->migrate_pfn);
low_pfn = pageblock_start_pfn(cc->free_pfn - (distance >> 2));
min_pfn = pageblock_start_pfn(cc->free_pfn - (distance >> 1));
if (WARN_ON_ONCE(min_pfn > low_pfn))
low_pfn = min_pfn;
/*
* Search starts from the last successful isolation order or the next
* order to search after a previous failure
*/
cc->search_order = min_t(unsigned int, cc->order - 1, cc->search_order);
for (order = cc->search_order;
!page && order >= 0;
order = next_search_order(cc, order)) {
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
struct free_area *area = &cc->zone->free_area[order];
struct list_head *freelist;
struct page *freepage;
unsigned long flags;
unsigned int order_scanned = 0;
mm, compaction: move high_pfn to the for loop scope In fast_isolate_freepages, high_pfn will be used if a prefered one (ie PFN >= low_fn) not found. But the high_pfn is not reset before searching an free area, so when it was used as freepage, it may from another free area searched before. As a result move_freelist_head(freelist, freepage) will have unexpected behavior (eg corrupt the MOVABLE freelist) Unable to handle kernel paging request at virtual address dead000000000200 Mem abort info: ESR = 0x96000044 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000044 CM = 0, WnR = 1 [dead000000000200] address between user and kernel address ranges -000|list_cut_before(inline) -000|move_freelist_head(inline) -000|fast_isolate_freepages(inline) -000|isolate_freepages(inline) -000|compaction_alloc(?, ?) -001|unmap_and_move(inline) -001|migrate_pages([NSD:0xFFFFFF80088CBBD0] from = 0xFFFFFF80088CBD88, [NSD:0xFFFFFF80088CBBC8] get_new_p -002|__read_once_size(inline) -002|static_key_count(inline) -002|static_key_false(inline) -002|trace_mm_compaction_migratepages(inline) -002|compact_zone(?, [NSD:0xFFFFFF80088CBCB0] capc = 0x0) -003|kcompactd_do_work(inline) -003|kcompactd([X19] p = 0xFFFFFF93227FBC40) -004|kthread([X20] _create = 0xFFFFFFE1AFB26380) -005|ret_from_fork(asm) The issue was reported on an smart phone product with 6GB ram and 3GB zram as swap device. This patch fixes the issue by reset high_pfn before searching each free area, which ensure freepage and freelist match when call move_freelist_head in fast_isolate_freepages(). Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Link: https://lkml.kernel.org/r/20210112094720.1238444-1-wu-yan@tcl.com Fixes: 5a811889de10f1eb ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Rokudo Yan <wu-yan@tcl.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-05 02:32:20 +00:00
unsigned long high_pfn = 0;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
if (!area->nr_free)
continue;
spin_lock_irqsave(&cc->zone->lock, flags);
freelist = &area->free_list[MIGRATE_MOVABLE];
list_for_each_entry_reverse(freepage, freelist, lru) {
unsigned long pfn;
order_scanned++;
nr_scanned++;
pfn = page_to_pfn(freepage);
if (pfn >= highest)
mm, compaction: make fast_isolate_freepages() stay within zone Compaction always operates on pages from a single given zone when isolating both pages to migrate and freepages. Pageblock boundaries are intersected with zone boundaries to be safe in case zone starts or ends in the middle of pageblock. The use of pageblock_pfn_to_page() protects against non-contiguous pageblocks. The functions fast_isolate_freepages() and fast_isolate_around() don't currently protect the fast freepage isolation thoroughly enough against these corner cases, and can result in freepage isolation operate outside of zone boundaries: - in fast_isolate_freepages() if we get a pfn from the first pageblock of a zone that starts in the middle of that pageblock, 'highest' can be a pfn outside of the zone. If we fail to isolate anything in this function, we may then call fast_isolate_around() on a pfn outside of the zone and there effectively do a set_pageblock_skip(page_to_pfn(highest)) which may currently hit a VM_BUG_ON() in some configurations - fast_isolate_around() checks only the zone end boundary and not beginning, nor that the pageblock is contiguous (with pageblock_pfn_to_page()) so it's possible that we end up calling isolate_freepages_block() on a range of pfn's from two different zones and end up e.g. isolating freepages under the wrong zone's lock. This patch should fix the above issues. Link: https://lkml.kernel.org/r/20210217173300.6394-1-vbabka@suse.cz Fixes: 5a811889de10 ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Hildenbrand <david@redhat.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Rapoport <rppt@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:39 +00:00
highest = max(pageblock_start_pfn(pfn),
cc->zone->zone_start_pfn);
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
if (pfn >= low_pfn) {
cc->fast_search_fail = 0;
cc->search_order = order;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
page = freepage;
break;
}
if (pfn >= min_pfn && pfn > high_pfn) {
high_pfn = pfn;
/* Shorten the scan if a candidate is found */
limit >>= 1;
}
if (order_scanned >= limit)
break;
}
/* Use a minimum pfn if a preferred one was not found */
if (!page && high_pfn) {
page = pfn_to_page(high_pfn);
/* Update freepage for the list reorder below */
freepage = page;
}
/* Reorder to so a future search skips recent pages */
move_freelist_head(freelist, freepage);
/* Isolate the page if available */
if (page) {
if (__isolate_free_page(page, order)) {
set_page_private(page, order);
nr_isolated = 1 << order;
cc->nr_freepages += nr_isolated;
list_add_tail(&page->lru, &cc->freepages);
count_compact_events(COMPACTISOLATED, nr_isolated);
} else {
/* If isolation fails, abort the search */
mm/compaction.c: abort search if isolation fails Running LTP oom01 in a tight loop or memory stress testing put the system in a low-memory situation could triggers random memory corruption like page flag corruption below due to in fast_isolate_freepages(), if isolation fails, next_search_order() does not abort the search immediately could lead to improper accesses. UBSAN: Undefined behaviour in ./include/linux/mm.h:1195:50 index 7 is out of range for type 'zone [5]' Call Trace: dump_stack+0x62/0x9a ubsan_epilogue+0xd/0x7f __ubsan_handle_out_of_bounds+0x14d/0x192 __isolate_free_page+0x52c/0x600 compaction_alloc+0x886/0x25f0 unmap_and_move+0x37/0x1e70 migrate_pages+0x2ca/0xb20 compact_zone+0x19cb/0x3620 kcompactd_do_work+0x2df/0x680 kcompactd+0x1d8/0x6c0 kthread+0x32c/0x3f0 ret_from_fork+0x35/0x40 ------------[ cut here ]------------ kernel BUG at mm/page_alloc.c:3124! invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI RIP: 0010:__isolate_free_page+0x464/0x600 RSP: 0000:ffff888b9e1af848 EFLAGS: 00010007 RAX: 0000000030000000 RBX: ffff888c39fcf0f8 RCX: 0000000000000000 RDX: 1ffff111873f9e25 RSI: 0000000000000004 RDI: ffffed1173c35ef6 RBP: ffff888b9e1af898 R08: fffffbfff4fc2461 R09: fffffbfff4fc2460 R10: fffffbfff4fc2460 R11: ffffffffa7e12303 R12: 0000000000000008 R13: dffffc0000000000 R14: 0000000000000000 R15: 0000000000000007 FS: 0000000000000000(0000) GS:ffff888ba8e80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fc7abc00000 CR3: 0000000752416004 CR4: 00000000001606a0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: compaction_alloc+0x886/0x25f0 unmap_and_move+0x37/0x1e70 migrate_pages+0x2ca/0xb20 compact_zone+0x19cb/0x3620 kcompactd_do_work+0x2df/0x680 kcompactd+0x1d8/0x6c0 kthread+0x32c/0x3f0 ret_from_fork+0x35/0x40 Link: http://lkml.kernel.org/r/20190320192648.52499-1-cai@lca.pw Fixes: dbe2d4e4f12e ("mm, compaction: round-robin the order while searching the free lists for a target") Signed-off-by: Qian Cai <cai@lca.pw> Acked-by: Mel Gorman <mgorman@techsingularity.net> Cc: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
2019-04-04 10:54:41 +00:00
order = cc->search_order + 1;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
page = NULL;
}
}
spin_unlock_irqrestore(&cc->zone->lock, flags);
/*
* Smaller scan on next order so the total scan ig related
* to freelist_scan_limit.
*/
if (order_scanned >= limit)
limit = min(1U, limit >> 1);
}
if (!page) {
cc->fast_search_fail++;
if (scan_start) {
/*
* Use the highest PFN found above min. If one was
* not found, be pessimistic for direct compaction
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
* and use the min mark.
*/
if (highest) {
page = pfn_to_page(highest);
cc->free_pfn = highest;
} else {
mm, compaction: make sure we isolate a valid PFN When we have holes in a normal memory zone, we could endup having cached_migrate_pfns which may not necessarily be valid, under heavy memory pressure with swapping enabled ( via __reset_isolation_suitable(), triggered by kswapd). Later if we fail to find a page via fast_isolate_freepages(), we may end up using the migrate_pfn we started the search with, as valid page. This could lead to accessing NULL pointer derefernces like below, due to an invalid mem_section pointer. Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008 [47/1825] Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 0000000082f94ae9 [0000000000000008] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP ... CPU: 10 PID: 6080 Comm: qemu-system-aar Not tainted 510-rc1+ #6 Hardware name: AmpereComputing(R) OSPREY EV-883832-X3-0001/OSPREY, BIOS 4819 09/25/2018 pstate: 60000005 (nZCv daif -PAN -UAO) pc : set_pfnblock_flags_mask+0x58/0xe8 lr : compaction_alloc+0x300/0x950 [...] Process qemu-system-aar (pid: 6080, stack limit = 0x0000000095070da5) Call trace: set_pfnblock_flags_mask+0x58/0xe8 compaction_alloc+0x300/0x950 migrate_pages+0x1a4/0xbb0 compact_zone+0x750/0xde8 compact_zone_order+0xd8/0x118 try_to_compact_pages+0xb4/0x290 __alloc_pages_direct_compact+0x84/0x1e0 __alloc_pages_nodemask+0x5e0/0xe18 alloc_pages_vma+0x1cc/0x210 do_huge_pmd_anonymous_page+0x108/0x7c8 __handle_mm_fault+0xdd4/0x1190 handle_mm_fault+0x114/0x1c0 __get_user_pages+0x198/0x3c0 get_user_pages_unlocked+0xb4/0x1d8 __gfn_to_pfn_memslot+0x12c/0x3b8 gfn_to_pfn_prot+0x4c/0x60 kvm_handle_guest_abort+0x4b0/0xcd8 handle_exit+0x140/0x1b8 kvm_arch_vcpu_ioctl_run+0x260/0x768 kvm_vcpu_ioctl+0x490/0x898 do_vfs_ioctl+0xc4/0x898 ksys_ioctl+0x8c/0xa0 __arm64_sys_ioctl+0x28/0x38 el0_svc_common+0x74/0x118 el0_svc_handler+0x38/0x78 el0_svc+0x8/0xc Code: f8607840 f100001f 8b011401 9a801020 (f9400400) ---[ end trace af6a35219325a9b6 ]--- The issue was reported on an arm64 server with 128GB with holes in the zone (e.g, [32GB@4GB, 96GB@544GB]), with a swap device enabled, while running 100 KVM guest instances. This patch fixes the issue by ensuring that the page belongs to a valid PFN when we fallback to using the lower limit of the scan range upon failure in fast_isolate_freepages(). Link: http://lkml.kernel.org/r/1558711908-15688-1-git-send-email-suzuki.poulose@arm.com Fixes: 5a811889de10f1eb ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Reported-by: Marc Zyngier <marc.zyngier@arm.com> Reviewed-by: Mel Gorman <mgorman@techsingularity.net> Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Qian Cai <cai@lca.pw> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-06-01 05:30:59 +00:00
if (cc->direct_compaction && pfn_valid(min_pfn)) {
mm: memmap_init: iterate over memblock regions rather that check each PFN When called during boot the memmap_init_zone() function checks if each PFN is valid and actually belongs to the node being initialized using early_pfn_valid() and early_pfn_in_nid(). Each such check may cost up to O(log(n)) where n is the number of memory banks, so for large amount of memory overall time spent in early_pfn*() becomes substantial. Since the information is anyway present in memblock, we can iterate over memblock memory regions in memmap_init() and only call memmap_init_zone() for PFN ranges that are know to be valid and in the appropriate node. [cai@lca.pw: fix a compilation warning from Clang] Link: http://lkml.kernel.org/r/CF6E407F-17DC-427C-8203-21979FB882EF@lca.pw [bhe@redhat.com: fix the incorrect hole in fast_isolate_freepages()] Link: http://lkml.kernel.org/r/8C537EB7-85EE-4DCF-943E-3CC0ED0DF56D@lca.pw Link: http://lkml.kernel.org/r/20200521014407.29690-1-bhe@redhat.com Signed-off-by: Baoquan He <bhe@redhat.com> Signed-off-by: Mike Rapoport <rppt@linux.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Hoan Tran <hoan@os.amperecomputing.com> [arm64] Cc: Brian Cain <bcain@codeaurora.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Greentime Hu <green.hu@gmail.com> Cc: Greg Ungerer <gerg@linux-m68k.org> Cc: Guan Xuetao <gxt@pku.edu.cn> Cc: Guo Ren <guoren@kernel.org> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Helge Deller <deller@gmx.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Ley Foon Tan <ley.foon.tan@intel.com> Cc: Mark Salter <msalter@redhat.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Simek <monstr@monstr.eu> Cc: Nick Hu <nickhu@andestech.com> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Richard Weinberger <richard@nod.at> Cc: Rich Felker <dalias@libc.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Stafford Horne <shorne@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Tony Luck <tony.luck@intel.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Cc: Qian Cai <cai@lca.pw> Link: http://lkml.kernel.org/r/20200412194859.12663-16-rppt@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-06-03 22:57:55 +00:00
page = pageblock_pfn_to_page(min_pfn,
mm, compaction: make fast_isolate_freepages() stay within zone Compaction always operates on pages from a single given zone when isolating both pages to migrate and freepages. Pageblock boundaries are intersected with zone boundaries to be safe in case zone starts or ends in the middle of pageblock. The use of pageblock_pfn_to_page() protects against non-contiguous pageblocks. The functions fast_isolate_freepages() and fast_isolate_around() don't currently protect the fast freepage isolation thoroughly enough against these corner cases, and can result in freepage isolation operate outside of zone boundaries: - in fast_isolate_freepages() if we get a pfn from the first pageblock of a zone that starts in the middle of that pageblock, 'highest' can be a pfn outside of the zone. If we fail to isolate anything in this function, we may then call fast_isolate_around() on a pfn outside of the zone and there effectively do a set_pageblock_skip(page_to_pfn(highest)) which may currently hit a VM_BUG_ON() in some configurations - fast_isolate_around() checks only the zone end boundary and not beginning, nor that the pageblock is contiguous (with pageblock_pfn_to_page()) so it's possible that we end up calling isolate_freepages_block() on a range of pfn's from two different zones and end up e.g. isolating freepages under the wrong zone's lock. This patch should fix the above issues. Link: https://lkml.kernel.org/r/20210217173300.6394-1-vbabka@suse.cz Fixes: 5a811889de10 ("mm, compaction: use free lists to quickly locate a migration target") Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Hildenbrand <david@redhat.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Rapoport <rppt@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:39 +00:00
min(pageblock_end_pfn(min_pfn),
zone_end_pfn(cc->zone)),
mm: memmap_init: iterate over memblock regions rather that check each PFN When called during boot the memmap_init_zone() function checks if each PFN is valid and actually belongs to the node being initialized using early_pfn_valid() and early_pfn_in_nid(). Each such check may cost up to O(log(n)) where n is the number of memory banks, so for large amount of memory overall time spent in early_pfn*() becomes substantial. Since the information is anyway present in memblock, we can iterate over memblock memory regions in memmap_init() and only call memmap_init_zone() for PFN ranges that are know to be valid and in the appropriate node. [cai@lca.pw: fix a compilation warning from Clang] Link: http://lkml.kernel.org/r/CF6E407F-17DC-427C-8203-21979FB882EF@lca.pw [bhe@redhat.com: fix the incorrect hole in fast_isolate_freepages()] Link: http://lkml.kernel.org/r/8C537EB7-85EE-4DCF-943E-3CC0ED0DF56D@lca.pw Link: http://lkml.kernel.org/r/20200521014407.29690-1-bhe@redhat.com Signed-off-by: Baoquan He <bhe@redhat.com> Signed-off-by: Mike Rapoport <rppt@linux.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Hoan Tran <hoan@os.amperecomputing.com> [arm64] Cc: Brian Cain <bcain@codeaurora.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Greentime Hu <green.hu@gmail.com> Cc: Greg Ungerer <gerg@linux-m68k.org> Cc: Guan Xuetao <gxt@pku.edu.cn> Cc: Guo Ren <guoren@kernel.org> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Helge Deller <deller@gmx.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Ley Foon Tan <ley.foon.tan@intel.com> Cc: Mark Salter <msalter@redhat.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michal Simek <monstr@monstr.eu> Cc: Nick Hu <nickhu@andestech.com> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Richard Weinberger <richard@nod.at> Cc: Rich Felker <dalias@libc.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Stafford Horne <shorne@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Tony Luck <tony.luck@intel.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Cc: Qian Cai <cai@lca.pw> Link: http://lkml.kernel.org/r/20200412194859.12663-16-rppt@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-06-03 22:57:55 +00:00
cc->zone);
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
cc->free_pfn = min_pfn;
}
}
}
}
mm, compaction: reduce premature advancement of the migration target scanner The fast isolation of free pages allows the cached PFN of the free scanner to advance faster than necessary depending on the contents of the free list. The key is that fast_isolate_freepages() can update zone->compact_cached_free_pfn via isolate_freepages_block(). When the fast search fails, the linear scan can start from a point that has skipped valid migration targets, particularly pageblocks with just low-order free pages. This can cause the migration source/target scanners to meet prematurely causing a reset. This patch starts by avoiding an update of the pageblock skip information and cached PFN from isolate_freepages_block() and puts the responsibility of updating that information in the callers. The fast scanner will update the cached PFN if and only if it finds a block that is higher than the existing cached PFN and sets the skip if the pageblock is full or nearly full. The linear scanner will update skipped information and the cached PFN only when a block is completely scanned. The total impact is that the free scanner advances more slowly as it is primarily driven by the linear scanner instead of the fast search. 5.0.0-rc1 5.0.0-rc1 noresched-v3r17 slowfree-v3r17 Amean fault-both-3 2965.68 ( 0.00%) 3036.75 ( -2.40%) Amean fault-both-5 3995.90 ( 0.00%) 4522.24 * -13.17%* Amean fault-both-7 5842.12 ( 0.00%) 6365.35 ( -8.96%) Amean fault-both-12 9550.87 ( 0.00%) 10340.93 ( -8.27%) Amean fault-both-18 13304.72 ( 0.00%) 14732.46 ( -10.73%) Amean fault-both-24 14618.59 ( 0.00%) 16288.96 ( -11.43%) Amean fault-both-30 16650.96 ( 0.00%) 16346.21 ( 1.83%) Amean fault-both-32 17145.15 ( 0.00%) 19317.49 ( -12.67%) The impact to latency is higher than the last version but it appears to be due to a slight increase in the free scan rates which is a potential side-effect of the patch. However, this is necessary for later patches that are more careful about how pageblocks are treated as earlier iterations of those patches hit corner cases where the restarts were punishing and very visible. Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:28 +00:00
if (highest && highest >= cc->zone->compact_cached_free_pfn) {
highest -= pageblock_nr_pages;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
cc->zone->compact_cached_free_pfn = highest;
mm, compaction: reduce premature advancement of the migration target scanner The fast isolation of free pages allows the cached PFN of the free scanner to advance faster than necessary depending on the contents of the free list. The key is that fast_isolate_freepages() can update zone->compact_cached_free_pfn via isolate_freepages_block(). When the fast search fails, the linear scan can start from a point that has skipped valid migration targets, particularly pageblocks with just low-order free pages. This can cause the migration source/target scanners to meet prematurely causing a reset. This patch starts by avoiding an update of the pageblock skip information and cached PFN from isolate_freepages_block() and puts the responsibility of updating that information in the callers. The fast scanner will update the cached PFN if and only if it finds a block that is higher than the existing cached PFN and sets the skip if the pageblock is full or nearly full. The linear scanner will update skipped information and the cached PFN only when a block is completely scanned. The total impact is that the free scanner advances more slowly as it is primarily driven by the linear scanner instead of the fast search. 5.0.0-rc1 5.0.0-rc1 noresched-v3r17 slowfree-v3r17 Amean fault-both-3 2965.68 ( 0.00%) 3036.75 ( -2.40%) Amean fault-both-5 3995.90 ( 0.00%) 4522.24 * -13.17%* Amean fault-both-7 5842.12 ( 0.00%) 6365.35 ( -8.96%) Amean fault-both-12 9550.87 ( 0.00%) 10340.93 ( -8.27%) Amean fault-both-18 13304.72 ( 0.00%) 14732.46 ( -10.73%) Amean fault-both-24 14618.59 ( 0.00%) 16288.96 ( -11.43%) Amean fault-both-30 16650.96 ( 0.00%) 16346.21 ( 1.83%) Amean fault-both-32 17145.15 ( 0.00%) 19317.49 ( -12.67%) The impact to latency is higher than the last version but it appears to be due to a slight increase in the free scan rates which is a potential side-effect of the patch. However, this is necessary for later patches that are more careful about how pageblocks are treated as earlier iterations of those patches hit corner cases where the restarts were punishing and very visible. Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:28 +00:00
}
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
cc->total_free_scanned += nr_scanned;
if (!page)
return cc->free_pfn;
low_pfn = page_to_pfn(page);
fast_isolate_around(cc, low_pfn, nr_isolated);
return low_pfn;
}
/*
* Based on information in the current compact_control, find blocks
* suitable for isolating free pages from and then isolate them.
*/
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
static void isolate_freepages(struct compact_control *cc)
{
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
struct zone *zone = cc->zone;
struct page *page;
unsigned long block_start_pfn; /* start of current pageblock */
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
unsigned long isolate_start_pfn; /* exact pfn we start at */
unsigned long block_end_pfn; /* end of current pageblock */
unsigned long low_pfn; /* lowest pfn scanner is able to scan */
struct list_head *freelist = &cc->freepages;
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
unsigned int stride;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
/* Try a small search of the free lists for a candidate */
isolate_start_pfn = fast_isolate_freepages(cc);
if (cc->nr_freepages)
goto splitmap;
/*
* Initialise the free scanner. The starting point is where we last
mm/compaction: make isolate_freepages start at pageblock boundary The compaction freepage scanner implementation in isolate_freepages() starts by taking the current cc->free_pfn value as the first pfn. In a for loop, it scans from this first pfn to the end of the pageblock, and then subtracts pageblock_nr_pages from the first pfn to obtain the first pfn for the next for loop iteration. This means that when cc->free_pfn starts at offset X rather than being aligned on pageblock boundary, the scanner will start at offset X in all scanned pageblock, ignoring potentially many free pages. Currently this can happen when a) zone's end pfn is not pageblock aligned, or b) through zone->compact_cached_free_pfn with CONFIG_HOLES_IN_ZONE enabled and a hole spanning the beginning of a pageblock This patch fixes the problem by aligning the initial pfn in isolate_freepages() to pageblock boundary. This also permits replacing the end-of-pageblock alignment within the for loop with a simple pageblock_nr_pages increment. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Heesub Shin <heesub.shin@samsung.com> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Dongjun Shin <d.j.shin@samsung.com> Cc: Sunghwan Yun <sunghwan.yun@samsung.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-05-06 19:50:03 +00:00
* successfully isolated from, zone-cached value, or the end of the
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
* zone when isolating for the first time. For looping we also need
* this pfn aligned down to the pageblock boundary, because we do
* block_start_pfn -= pageblock_nr_pages in the for loop.
* For ending point, take care when isolating in last pageblock of a
* zone which ends in the middle of a pageblock.
mm/compaction: make isolate_freepages start at pageblock boundary The compaction freepage scanner implementation in isolate_freepages() starts by taking the current cc->free_pfn value as the first pfn. In a for loop, it scans from this first pfn to the end of the pageblock, and then subtracts pageblock_nr_pages from the first pfn to obtain the first pfn for the next for loop iteration. This means that when cc->free_pfn starts at offset X rather than being aligned on pageblock boundary, the scanner will start at offset X in all scanned pageblock, ignoring potentially many free pages. Currently this can happen when a) zone's end pfn is not pageblock aligned, or b) through zone->compact_cached_free_pfn with CONFIG_HOLES_IN_ZONE enabled and a hole spanning the beginning of a pageblock This patch fixes the problem by aligning the initial pfn in isolate_freepages() to pageblock boundary. This also permits replacing the end-of-pageblock alignment within the for loop with a simple pageblock_nr_pages increment. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Heesub Shin <heesub.shin@samsung.com> Acked-by: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Dongjun Shin <d.j.shin@samsung.com> Cc: Sunghwan Yun <sunghwan.yun@samsung.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-05-06 19:50:03 +00:00
* The low boundary is the end of the pageblock the migration scanner
* is using.
*/
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
isolate_start_pfn = cc->free_pfn;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
block_start_pfn = pageblock_start_pfn(isolate_start_pfn);
block_end_pfn = min(block_start_pfn + pageblock_nr_pages,
zone_end_pfn(zone));
low_pfn = pageblock_end_pfn(cc->migrate_pfn);
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
stride = cc->mode == MIGRATE_ASYNC ? COMPACT_CLUSTER_MAX : 1;
/*
* Isolate free pages until enough are available to migrate the
* pages on cc->migratepages. We stop searching if the migrate
* and free page scanners meet or enough free pages are isolated.
*/
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
for (; block_start_pfn >= low_pfn;
block_end_pfn = block_start_pfn,
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
block_start_pfn -= pageblock_nr_pages,
isolate_start_pfn = block_start_pfn) {
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
unsigned long nr_isolated;
/*
* This can iterate a massively long zone without finding any
* suitable migration targets, so periodically check resched.
*/
if (!(block_start_pfn % (SWAP_CLUSTER_MAX * pageblock_nr_pages)))
mm, compaction: do not consider a need to reschedule as contention Scanning on large machines can take a considerable length of time and eventually need to be rescheduled. This is treated as an abort event but that's not appropriate as the attempt is likely to be retried after making numerous checks and taking another cycle through the page allocator. This patch will check the need to reschedule if necessary but continue the scanning. The main benefit is reduced scanning when compaction is taking a long time or the machine is over-saturated. It also avoids an unnecessary exit of compaction that ends up being retried by the page allocator in the outer loop. 5.0.0-rc1 5.0.0-rc1 synccached-v3r16 noresched-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2958.27 ( 0.00%) 2965.68 ( -0.25%) Amean fault-both-5 4091.90 ( 0.00%) 3995.90 ( 2.35%) Amean fault-both-7 5803.05 ( 0.00%) 5842.12 ( -0.67%) Amean fault-both-12 9481.06 ( 0.00%) 9550.87 ( -0.74%) Amean fault-both-18 14141.51 ( 0.00%) 13304.72 ( 5.92%) Amean fault-both-24 16438.00 ( 0.00%) 14618.59 ( 11.07%) Amean fault-both-30 17531.72 ( 0.00%) 16650.96 ( 5.02%) Amean fault-both-32 17101.96 ( 0.00%) 17145.15 ( -0.25%) Link: http://lkml.kernel.org/r/20190118175136.31341-18-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:24 +00:00
cond_resched();
mm, compaction: reduce zone checking frequency in the migration scanner The unification of the migrate and free scanner families of function has highlighted a difference in how the scanners ensure they only isolate pages of the intended zone. This is important for taking zone lock or lru lock of the correct zone. Due to nodes overlapping, it is however possible to encounter a different zone within the range of the zone being compacted. The free scanner, since its inception by commit 748446bb6b5a ("mm: compaction: memory compaction core"), has been checking the zone of the first valid page in a pageblock, and skipping the whole pageblock if the zone does not match. This checking was completely missing from the migration scanner at first, and later added by commit dc9086004b3d ("mm: compaction: check for overlapping nodes during isolation for migration") in a reaction to a bug report. But the zone comparison in migration scanner is done once per a single scanned page, which is more defensive and thus more costly than a check per pageblock. This patch unifies the checking done in both scanners to once per pageblock, through a new pageblock_pfn_to_page() function, which also includes pfn_valid() checks. It is more defensive than the current free scanner checks, as it checks both the first and last page of the pageblock, but less defensive by the migration scanner per-page checks. It assumes that node overlapping may result (on some architecture) in a boundary between two nodes falling into the middle of a pageblock, but that there cannot be a node0 node1 node0 interleaving within a single pageblock. The result is more code being shared and a bit less per-page CPU cost in the migration scanner. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:11 +00:00
page = pageblock_pfn_to_page(block_start_pfn, block_end_pfn,
zone);
if (!page)
continue;
/* Check the block is suitable for migration */
2016-10-08 00:00:37 +00:00
if (!suitable_migration_target(cc, page))
continue;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
/* If isolation recently failed, do not retry */
if (!isolation_suitable(cc, page))
continue;
mm, compaction: remember position within pageblock in free pages scanner Unlike the migration scanner, the free scanner remembers the beginning of the last scanned pageblock in cc->free_pfn. It might be therefore rescanning pages uselessly when called several times during single compaction. This might have been useful when pages were returned to the buddy allocator after a failed migration, but this is no longer the case. This patch changes the meaning of cc->free_pfn so that if it points to a middle of a pageblock, that pageblock is scanned only from cc->free_pfn to the end. isolate_freepages_block() will record the pfn of the last page it looked at, which is then used to update cc->free_pfn. In the mmtests stress-highalloc benchmark, this has resulted in lowering the ratio between pages scanned by both scanners, from 2.5 free pages per migrate page, to 2.25 free pages per migrate page, without affecting success rates. With __GFP_NO_KSWAPD allocations, this appears to result in a worse ratio (2.1 instead of 1.8), but page migration successes increased by 10%, so this could mean that more useful work can be done until need_resched() aborts this kind of compaction. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:20 +00:00
/* Found a block suitable for isolating free pages from. */
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
nr_isolated = isolate_freepages_block(cc, &isolate_start_pfn,
block_end_pfn, freelist, stride, false);
mm, compaction: reduce premature advancement of the migration target scanner The fast isolation of free pages allows the cached PFN of the free scanner to advance faster than necessary depending on the contents of the free list. The key is that fast_isolate_freepages() can update zone->compact_cached_free_pfn via isolate_freepages_block(). When the fast search fails, the linear scan can start from a point that has skipped valid migration targets, particularly pageblocks with just low-order free pages. This can cause the migration source/target scanners to meet prematurely causing a reset. This patch starts by avoiding an update of the pageblock skip information and cached PFN from isolate_freepages_block() and puts the responsibility of updating that information in the callers. The fast scanner will update the cached PFN if and only if it finds a block that is higher than the existing cached PFN and sets the skip if the pageblock is full or nearly full. The linear scanner will update skipped information and the cached PFN only when a block is completely scanned. The total impact is that the free scanner advances more slowly as it is primarily driven by the linear scanner instead of the fast search. 5.0.0-rc1 5.0.0-rc1 noresched-v3r17 slowfree-v3r17 Amean fault-both-3 2965.68 ( 0.00%) 3036.75 ( -2.40%) Amean fault-both-5 3995.90 ( 0.00%) 4522.24 * -13.17%* Amean fault-both-7 5842.12 ( 0.00%) 6365.35 ( -8.96%) Amean fault-both-12 9550.87 ( 0.00%) 10340.93 ( -8.27%) Amean fault-both-18 13304.72 ( 0.00%) 14732.46 ( -10.73%) Amean fault-both-24 14618.59 ( 0.00%) 16288.96 ( -11.43%) Amean fault-both-30 16650.96 ( 0.00%) 16346.21 ( 1.83%) Amean fault-both-32 17145.15 ( 0.00%) 19317.49 ( -12.67%) The impact to latency is higher than the last version but it appears to be due to a slight increase in the free scan rates which is a potential side-effect of the patch. However, this is necessary for later patches that are more careful about how pageblocks are treated as earlier iterations of those patches hit corner cases where the restarts were punishing and very visible. Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:28 +00:00
/* Update the skip hint if the full pageblock was scanned */
if (isolate_start_pfn == block_end_pfn)
update_pageblock_skip(cc, page, block_start_pfn);
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
/* Are enough freepages isolated? */
if (cc->nr_freepages >= cc->nr_migratepages) {
if (isolate_start_pfn >= block_end_pfn) {
/*
* Restart at previous pageblock if more
* freepages can be isolated next time.
*/
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
isolate_start_pfn =
block_start_pfn - pageblock_nr_pages;
}
mm, compaction: properly signal and act upon lock and need_sched() contention Compaction uses compact_checklock_irqsave() function to periodically check for lock contention and need_resched() to either abort async compaction, or to free the lock, schedule and retake the lock. When aborting, cc->contended is set to signal the contended state to the caller. Two problems have been identified in this mechanism. First, compaction also calls directly cond_resched() in both scanners when no lock is yet taken. This call either does not abort async compaction, or set cc->contended appropriately. This patch introduces a new compact_should_abort() function to achieve both. In isolate_freepages(), the check frequency is reduced to once by SWAP_CLUSTER_MAX pageblocks to match what the migration scanner does in the preliminary page checks. In case a pageblock is found suitable for calling isolate_freepages_block(), the checks within there are done on higher frequency. Second, isolate_freepages() does not check if isolate_freepages_block() aborted due to contention, and advances to the next pageblock. This violates the principle of aborting on contention, and might result in pageblocks not being scanned completely, since the scanning cursor is advanced. This problem has been noticed in the code by Joonsoo Kim when reviewing related patches. This patch makes isolate_freepages_block() check the cc->contended flag and abort. In case isolate_freepages() has already isolated some pages before aborting due to contention, page migration will proceed, which is OK since we do not want to waste the work that has been done, and page migration has own checks for contention. However, we do not want another isolation attempt by either of the scanners, so cc->contended flag check is added also to compaction_alloc() and compact_finished() to make sure compaction is aborted right after the migration. The outcome of the patch should be reduced lock contention by async compaction and lower latencies for higher-order allocations where direct compaction is involved. [akpm@linux-foundation.org: fix typo in comment] Reported-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Tested-by: Shawn Guo <shawn.guo@linaro.org> Tested-by: Kevin Hilman <khilman@linaro.org> Tested-by: Stephen Warren <swarren@nvidia.com> Tested-by: Fabio Estevam <fabio.estevam@freescale.com> Cc: David Rientjes <rientjes@google.com> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:10:41 +00:00
break;
} else if (isolate_start_pfn < block_end_pfn) {
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
/*
* If isolation failed early, do not continue
* needlessly.
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
*/
break;
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
}
mm, compaction: sample pageblocks for free pages Once fast searching finishes, there is a possibility that the linear scanner is scanning full blocks found by the fast scanner earlier. This patch uses an adaptive stride to sample pageblocks for free pages. The more consecutive full pageblocks encountered, the larger the stride until a pageblock with free pages is found. The scanners might meet slightly sooner but it is an acceptable risk given that the search of the free lists may still encounter the pages and adjust the cached PFN of the free scanner accordingly. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2752.37 ( 0.00%) 2729.95 ( 0.81%) Amean fault-both-5 4341.69 ( 0.00%) 4397.80 ( -1.29%) Amean fault-both-7 6308.75 ( 0.00%) 6097.61 ( 3.35%) Amean fault-both-12 10241.81 ( 0.00%) 9407.15 ( 8.15%) Amean fault-both-18 13736.09 ( 0.00%) 10857.63 * 20.96%* Amean fault-both-24 16853.95 ( 0.00%) 13323.24 * 20.95%* Amean fault-both-30 15862.61 ( 0.00%) 17345.44 ( -9.35%) Amean fault-both-32 18450.85 ( 0.00%) 16892.00 ( 8.45%) The latency is mildly improved offseting some overhead from earlier patches that are prerequisites for the rest of the series. However, a major impact is on the free scan rate with an 82% reduction. 5.0.0-rc1 5.0.0-rc1 roundrobin-v3r17 samplefree-v3r17 Compaction migrate scanned 21607271 20116887 Compaction free scanned 95336406 16668703 It's also the first time in the series where the number of pages scanned by the migration scanner is greater than the free scanner due to the increased search efficiency. Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:34 +00:00
/* Adjust stride depending on isolation */
if (nr_isolated) {
stride = 1;
continue;
}
stride = min_t(unsigned int, COMPACT_CLUSTER_MAX, stride << 1);
}
mm: compaction: detect when scanners meet in isolate_freepages Compaction of a zone is finished when the migrate scanner (which begins at the zone's lowest pfn) meets the free page scanner (which begins at the zone's highest pfn). This is detected in compact_zone() and in the case of direct compaction, the compact_blockskip_flush flag is set so that kswapd later resets the cached scanner pfn's, and a new compaction may again start at the zone's borders. The meeting of the scanners can happen during either scanner's activity. However, it may currently fail to be detected when it occurs in the free page scanner, due to two problems. First, isolate_freepages() keeps free_pfn at the highest block where it isolated pages from, for the purposes of not missing the pages that are returned back to allocator when migration fails. Second, failing to isolate enough free pages due to scanners meeting results in -ENOMEM being returned by migrate_pages(), which makes compact_zone() bail out immediately without calling compact_finished() that would detect scanners meeting. This failure to detect scanners meeting might result in repeated attempts at compaction of a zone that keep starting from the cached pfn's close to the meeting point, and quickly failing through the -ENOMEM path, without the cached pfns being reset, over and over. This has been observed (through additional tracepoints) in the third phase of the mmtests stress-highalloc benchmark, where the allocator runs on an otherwise idle system. The problem was observed in the DMA32 zone, which was used as a fallback to the preferred Normal zone, but on the 4GB system it was actually the largest zone. The problem is even amplified for such fallback zone - the deferred compaction logic, which could (after being fixed by a previous patch) reset the cached scanner pfn's, is only applied to the preferred zone and not for the fallbacks. The problem in the third phase of the benchmark was further amplified by commit 81c0a2bb515f ("mm: page_alloc: fair zone allocator policy") which resulted in a non-deterministic regression of the allocation success rate from ~85% to ~65%. This occurs in about half of benchmark runs, making bisection problematic. It is unlikely that the commit itself is buggy, but it should put more pressure on the DMA32 zone during phases 1 and 2, which may leave it more fragmented in phase 3 and expose the bugs that this patch fixes. The fix is to make scanners meeting in isolate_freepage() stay that way, and to check in compact_zone() for scanners meeting when migrate_pages() returns -ENOMEM. The result is that compact_finished() also detects scanners meeting and sets the compact_blockskip_flush flag to make kswapd reset the scanner pfn's. The results in stress-highalloc benchmark show that the "regression" by commit 81c0a2bb515f in phase 3 no longer occurs, and phase 1 and 2 allocation success rates are also significantly improved. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:09 +00:00
/*
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
* Record where the free scanner will restart next time. Either we
* broke from the loop and set isolate_start_pfn based on the last
* call to isolate_freepages_block(), or we met the migration scanner
* and the loop terminated due to isolate_start_pfn < low_pfn
mm: compaction: detect when scanners meet in isolate_freepages Compaction of a zone is finished when the migrate scanner (which begins at the zone's lowest pfn) meets the free page scanner (which begins at the zone's highest pfn). This is detected in compact_zone() and in the case of direct compaction, the compact_blockskip_flush flag is set so that kswapd later resets the cached scanner pfn's, and a new compaction may again start at the zone's borders. The meeting of the scanners can happen during either scanner's activity. However, it may currently fail to be detected when it occurs in the free page scanner, due to two problems. First, isolate_freepages() keeps free_pfn at the highest block where it isolated pages from, for the purposes of not missing the pages that are returned back to allocator when migration fails. Second, failing to isolate enough free pages due to scanners meeting results in -ENOMEM being returned by migrate_pages(), which makes compact_zone() bail out immediately without calling compact_finished() that would detect scanners meeting. This failure to detect scanners meeting might result in repeated attempts at compaction of a zone that keep starting from the cached pfn's close to the meeting point, and quickly failing through the -ENOMEM path, without the cached pfns being reset, over and over. This has been observed (through additional tracepoints) in the third phase of the mmtests stress-highalloc benchmark, where the allocator runs on an otherwise idle system. The problem was observed in the DMA32 zone, which was used as a fallback to the preferred Normal zone, but on the 4GB system it was actually the largest zone. The problem is even amplified for such fallback zone - the deferred compaction logic, which could (after being fixed by a previous patch) reset the cached scanner pfn's, is only applied to the preferred zone and not for the fallbacks. The problem in the third phase of the benchmark was further amplified by commit 81c0a2bb515f ("mm: page_alloc: fair zone allocator policy") which resulted in a non-deterministic regression of the allocation success rate from ~85% to ~65%. This occurs in about half of benchmark runs, making bisection problematic. It is unlikely that the commit itself is buggy, but it should put more pressure on the DMA32 zone during phases 1 and 2, which may leave it more fragmented in phase 3 and expose the bugs that this patch fixes. The fix is to make scanners meeting in isolate_freepage() stay that way, and to check in compact_zone() for scanners meeting when migrate_pages() returns -ENOMEM. The result is that compact_finished() also detects scanners meeting and sets the compact_blockskip_flush flag to make kswapd reset the scanner pfn's. The results in stress-highalloc benchmark show that the "regression" by commit 81c0a2bb515f in phase 3 no longer occurs, and phase 1 and 2 allocation success rates are also significantly improved. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:09 +00:00
*/
mm, compaction: simplify handling restart position in free pages scanner Handling the position where compaction free scanner should restart (stored in cc->free_pfn) got more complex with commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner"). Currently the position is updated in each loop iteration of isolate_freepages(), although it should be enough to update it only when breaking from the loop. There's also an extra check outside the loop updates the position in case we have met the migration scanner. This can be simplified if we move the test for having isolated enough from the for-loop header next to the test for contention, and determining the restart position only in these cases. We can reuse the isolate_start_pfn variable for this instead of setting cc->free_pfn directly. Outside the loop, we can simply set cc->free_pfn to current value of isolate_start_pfn without any extra check. Also add a VM_BUG_ON to catch possible mistake in the future, in case we later add a new condition that terminates isolate_freepages_block() prematurely without also considering the condition in isolate_freepages(). Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:39 +00:00
cc->free_pfn = isolate_start_pfn;
mm, compaction: use free lists to quickly locate a migration target Similar to the migration scanner, this patch uses the free lists to quickly locate a migration target. The search is different in that lower orders will be searched for a suitable high PFN if necessary but the search is still bound. This is justified on the grounds that the free scanner typically scans linearly much more than the migration scanner. If a free page is found, it is isolated and compaction continues if enough pages were isolated. For SYNC* scanning, the full pageblock is scanned for any remaining free pages so that is can be marked for skipping in the near future. 1-socket thpfioscale 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Amean fault-both-3 3024.41 ( 0.00%) 3200.68 ( -5.83%) Amean fault-both-5 4749.30 ( 0.00%) 4847.75 ( -2.07%) Amean fault-both-7 6454.95 ( 0.00%) 6658.92 ( -3.16%) Amean fault-both-12 10324.83 ( 0.00%) 11077.62 ( -7.29%) Amean fault-both-18 12896.82 ( 0.00%) 12403.97 ( 3.82%) Amean fault-both-24 13470.60 ( 0.00%) 15607.10 * -15.86%* Amean fault-both-30 17143.99 ( 0.00%) 18752.27 ( -9.38%) Amean fault-both-32 17743.91 ( 0.00%) 21207.54 * -19.52%* The impact on latency is variable but the search is optimistic and sensitive to the exact system state. Success rates are similar but the major impact is to the rate of scanning 5.0.0-rc1 5.0.0-rc1 isolmig-v3r15 findfree-v3r16 Compaction migrate scanned 25646769 29507205 Compaction free scanned 201558184 100359571 The free scan rates are reduced by 50%. The 2-socket reductions for the free scanner are more dramatic which is a likely reflection that the machine has more memory. [dan.carpenter@oracle.com: fix static checker warning] [vbabka@suse.cz: correct number of pages scanned for lower orders] Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:01 +00:00
splitmap:
/* __isolate_free_page() does not map the pages */
split_map_pages(freelist);
}
/*
* This is a migrate-callback that "allocates" freepages by taking pages
* from the isolated freelists in the block we are migrating to.
*/
static struct page *compaction_alloc(struct page *migratepage,
unsigned long data)
{
struct compact_control *cc = (struct compact_control *)data;
struct page *freepage;
if (list_empty(&cc->freepages)) {
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
isolate_freepages(cc);
if (list_empty(&cc->freepages))
return NULL;
}
freepage = list_entry(cc->freepages.next, struct page, lru);
list_del(&freepage->lru);
cc->nr_freepages--;
return freepage;
}
/*
mm, compaction: return failed migration target pages back to freelist Greg reported that he found isolated free pages were returned back to the VM rather than the compaction freelist. This will cause holes behind the free scanner and cause it to reallocate additional memory if necessary later. He detected the problem at runtime seeing that ext4 metadata pages (esp the ones read by "sbi->s_group_desc[i] = sb_bread(sb, block)") were constantly visited by compaction calls of migrate_pages(). These pages had a non-zero b_count which caused fallback_migrate_page() -> try_to_release_page() -> try_to_free_buffers() to fail. Memory compaction works by having a "freeing scanner" scan from one end of a zone which isolates pages as migration targets while another "migrating scanner" scans from the other end of the same zone which isolates pages for migration. When page migration fails for an isolated page, the target page is returned to the system rather than the freelist built by the freeing scanner. This may require the freeing scanner to continue scanning memory after suitable migration targets have already been returned to the system needlessly. This patch returns destination pages to the freeing scanner freelist when page migration fails. This prevents unnecessary work done by the freeing scanner but also encourages memory to be as compacted as possible at the end of the zone. Signed-off-by: David Rientjes <rientjes@google.com> Reported-by: Greg Thelen <gthelen@google.com> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:08:26 +00:00
* This is a migrate-callback that "frees" freepages back to the isolated
* freelist. All pages on the freelist are from the same zone, so there is no
* special handling needed for NUMA.
*/
static void compaction_free(struct page *page, unsigned long data)
{
struct compact_control *cc = (struct compact_control *)data;
list_add(&page->lru, &cc->freepages);
cc->nr_freepages++;
}
/* possible outcome of isolate_migratepages */
typedef enum {
ISOLATE_ABORT, /* Abort compaction now */
ISOLATE_NONE, /* No pages isolated, continue scanning */
ISOLATE_SUCCESS, /* Pages isolated, migrate */
} isolate_migrate_t;
mm: allow compaction of unevictable pages Currently, pages which are marked as unevictable are protected from compaction, but not from other types of migration. The POSIX real time extension explicitly states that mlock() will prevent a major page fault, but the spirit of this is that mlock() should give a process the ability to control sources of latency, including minor page faults. However, the mlock manpage only explicitly says that a locked page will not be written to swap and this can cause some confusion. The compaction code today does not give a developer who wants to avoid swap but wants to have large contiguous areas available any method to achieve this state. This patch introduces a sysctl for controlling compaction behavior with respect to the unevictable lru. Users who demand no page faults after a page is present can set compact_unevictable_allowed to 0 and users who need the large contiguous areas can enable compaction on locked memory by leaving the default value of 1. To illustrate this problem I wrote a quick test program that mmaps a large number of 1MB files filled with random data. These maps are created locked and read only. Then every other mmap is unmapped and I attempt to allocate huge pages to the static huge page pool. When the compact_unevictable_allowed sysctl is 0, I cannot allocate hugepages after fragmenting memory. When the value is set to 1, allocations succeed. Signed-off-by: Eric B Munson <emunson@akamai.com> Acked-by: Michal Hocko <mhocko@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Christoph Lameter <cl@linux.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-15 23:13:20 +00:00
/*
* Allow userspace to control policy on scanning the unevictable LRU for
* compactable pages.
*/
#ifdef CONFIG_PREEMPT_RT
int sysctl_compact_unevictable_allowed __read_mostly = 0;
#else
mm: allow compaction of unevictable pages Currently, pages which are marked as unevictable are protected from compaction, but not from other types of migration. The POSIX real time extension explicitly states that mlock() will prevent a major page fault, but the spirit of this is that mlock() should give a process the ability to control sources of latency, including minor page faults. However, the mlock manpage only explicitly says that a locked page will not be written to swap and this can cause some confusion. The compaction code today does not give a developer who wants to avoid swap but wants to have large contiguous areas available any method to achieve this state. This patch introduces a sysctl for controlling compaction behavior with respect to the unevictable lru. Users who demand no page faults after a page is present can set compact_unevictable_allowed to 0 and users who need the large contiguous areas can enable compaction on locked memory by leaving the default value of 1. To illustrate this problem I wrote a quick test program that mmaps a large number of 1MB files filled with random data. These maps are created locked and read only. Then every other mmap is unmapped and I attempt to allocate huge pages to the static huge page pool. When the compact_unevictable_allowed sysctl is 0, I cannot allocate hugepages after fragmenting memory. When the value is set to 1, allocations succeed. Signed-off-by: Eric B Munson <emunson@akamai.com> Acked-by: Michal Hocko <mhocko@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Christoph Lameter <cl@linux.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-15 23:13:20 +00:00
int sysctl_compact_unevictable_allowed __read_mostly = 1;
#endif
mm: allow compaction of unevictable pages Currently, pages which are marked as unevictable are protected from compaction, but not from other types of migration. The POSIX real time extension explicitly states that mlock() will prevent a major page fault, but the spirit of this is that mlock() should give a process the ability to control sources of latency, including minor page faults. However, the mlock manpage only explicitly says that a locked page will not be written to swap and this can cause some confusion. The compaction code today does not give a developer who wants to avoid swap but wants to have large contiguous areas available any method to achieve this state. This patch introduces a sysctl for controlling compaction behavior with respect to the unevictable lru. Users who demand no page faults after a page is present can set compact_unevictable_allowed to 0 and users who need the large contiguous areas can enable compaction on locked memory by leaving the default value of 1. To illustrate this problem I wrote a quick test program that mmaps a large number of 1MB files filled with random data. These maps are created locked and read only. Then every other mmap is unmapped and I attempt to allocate huge pages to the static huge page pool. When the compact_unevictable_allowed sysctl is 0, I cannot allocate hugepages after fragmenting memory. When the value is set to 1, allocations succeed. Signed-off-by: Eric B Munson <emunson@akamai.com> Acked-by: Michal Hocko <mhocko@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Christoph Lameter <cl@linux.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-15 23:13:20 +00:00
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
static inline void
update_fast_start_pfn(struct compact_control *cc, unsigned long pfn)
{
if (cc->fast_start_pfn == ULONG_MAX)
return;
if (!cc->fast_start_pfn)
cc->fast_start_pfn = pfn;
cc->fast_start_pfn = min(cc->fast_start_pfn, pfn);
}
static inline unsigned long
reinit_migrate_pfn(struct compact_control *cc)
{
if (!cc->fast_start_pfn || cc->fast_start_pfn == ULONG_MAX)
return cc->migrate_pfn;
cc->migrate_pfn = cc->fast_start_pfn;
cc->fast_start_pfn = ULONG_MAX;
return cc->migrate_pfn;
}
/*
* Briefly search the free lists for a migration source that already has
* some free pages to reduce the number of pages that need migration
* before a pageblock is free.
*/
static unsigned long fast_find_migrateblock(struct compact_control *cc)
{
unsigned int limit = freelist_scan_limit(cc);
unsigned int nr_scanned = 0;
unsigned long distance;
unsigned long pfn = cc->migrate_pfn;
unsigned long high_pfn;
int order;
bool found_block = false;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
/* Skip hints are relied on to avoid repeats on the fast search */
if (cc->ignore_skip_hint)
return pfn;
/*
* If the migrate_pfn is not at the start of a zone or the start
* of a pageblock then assume this is a continuation of a previous
* scan restarted due to COMPACT_CLUSTER_MAX.
*/
if (pfn != cc->zone->zone_start_pfn && pfn != pageblock_start_pfn(pfn))
return pfn;
/*
* For smaller orders, just linearly scan as the number of pages
* to migrate should be relatively small and does not necessarily
* justify freeing up a large block for a small allocation.
*/
if (cc->order <= PAGE_ALLOC_COSTLY_ORDER)
return pfn;
/*
* Only allow kcompactd and direct requests for movable pages to
* quickly clear out a MOVABLE pageblock for allocation. This
* reduces the risk that a large movable pageblock is freed for
* an unmovable/reclaimable small allocation.
*/
if (cc->direct_compaction && cc->migratetype != MIGRATE_MOVABLE)
return pfn;
/*
* When starting the migration scanner, pick any pageblock within the
* first half of the search space. Otherwise try and pick a pageblock
* within the first eighth to reduce the chances that a migration
* target later becomes a source.
*/
distance = (cc->free_pfn - cc->migrate_pfn) >> 1;
if (cc->migrate_pfn != cc->zone->zone_start_pfn)
distance >>= 2;
high_pfn = pageblock_start_pfn(cc->migrate_pfn + distance);
for (order = cc->order - 1;
order >= PAGE_ALLOC_COSTLY_ORDER && !found_block && nr_scanned < limit;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
order--) {
struct free_area *area = &cc->zone->free_area[order];
struct list_head *freelist;
unsigned long flags;
struct page *freepage;
if (!area->nr_free)
continue;
spin_lock_irqsave(&cc->zone->lock, flags);
freelist = &area->free_list[MIGRATE_MOVABLE];
list_for_each_entry(freepage, freelist, lru) {
unsigned long free_pfn;
if (nr_scanned++ >= limit) {
move_freelist_tail(freelist, freepage);
break;
}
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
free_pfn = page_to_pfn(freepage);
if (free_pfn < high_pfn) {
/*
* Avoid if skipped recently. Ideally it would
* move to the tail but even safe iteration of
* the list assumes an entry is deleted, not
* reordered.
*/
if (get_pageblock_skip(freepage))
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
continue;
/* Reorder to so a future search skips recent pages */
move_freelist_tail(freelist, freepage);
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
update_fast_start_pfn(cc, free_pfn);
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
pfn = pageblock_start_pfn(free_pfn);
cc->fast_search_fail = 0;
found_block = true;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
set_pageblock_skip(freepage);
break;
}
}
spin_unlock_irqrestore(&cc->zone->lock, flags);
}
cc->total_migrate_scanned += nr_scanned;
/*
* If fast scanning failed then use a cached entry for a page block
* that had free pages as the basis for starting a linear scan.
*/
if (!found_block) {
cc->fast_search_fail++;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
pfn = reinit_migrate_pfn(cc);
}
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
return pfn;
}
/*
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
* Isolate all pages that can be migrated from the first suitable block,
* starting at the block pointed to by the migrate scanner pfn within
* compact_control.
*/
static isolate_migrate_t isolate_migratepages(struct compact_control *cc)
{
unsigned long block_start_pfn;
unsigned long block_end_pfn;
unsigned long low_pfn;
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
struct page *page;
const isolate_mode_t isolate_mode =
mm: allow compaction of unevictable pages Currently, pages which are marked as unevictable are protected from compaction, but not from other types of migration. The POSIX real time extension explicitly states that mlock() will prevent a major page fault, but the spirit of this is that mlock() should give a process the ability to control sources of latency, including minor page faults. However, the mlock manpage only explicitly says that a locked page will not be written to swap and this can cause some confusion. The compaction code today does not give a developer who wants to avoid swap but wants to have large contiguous areas available any method to achieve this state. This patch introduces a sysctl for controlling compaction behavior with respect to the unevictable lru. Users who demand no page faults after a page is present can set compact_unevictable_allowed to 0 and users who need the large contiguous areas can enable compaction on locked memory by leaving the default value of 1. To illustrate this problem I wrote a quick test program that mmaps a large number of 1MB files filled with random data. These maps are created locked and read only. Then every other mmap is unmapped and I attempt to allocate huge pages to the static huge page pool. When the compact_unevictable_allowed sysctl is 0, I cannot allocate hugepages after fragmenting memory. When the value is set to 1, allocations succeed. Signed-off-by: Eric B Munson <emunson@akamai.com> Acked-by: Michal Hocko <mhocko@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Christoph Lameter <cl@linux.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-15 23:13:20 +00:00
(sysctl_compact_unevictable_allowed ? ISOLATE_UNEVICTABLE : 0) |
(cc->mode != MIGRATE_SYNC ? ISOLATE_ASYNC_MIGRATE : 0);
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
bool fast_find_block;
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
/*
* Start at where we last stopped, or beginning of the zone as
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
* initialized by compact_zone(). The first failure will use
* the lowest PFN as the starting point for linear scanning.
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
*/
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
low_pfn = fast_find_migrateblock(cc);
block_start_pfn = pageblock_start_pfn(low_pfn);
if (block_start_pfn < cc->zone->zone_start_pfn)
block_start_pfn = cc->zone->zone_start_pfn;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
/*
* fast_find_migrateblock marks a pageblock skipped so to avoid
* the isolation_suitable check below, check whether the fast
* search was successful.
*/
fast_find_block = low_pfn != cc->migrate_pfn && !cc->fast_search_fail;
/* Only scan within a pageblock boundary */
block_end_pfn = pageblock_end_pfn(low_pfn);
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
/*
* Iterate over whole pageblocks until we find the first suitable.
* Do not cross the free scanner.
*/
for (; block_end_pfn <= cc->free_pfn;
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
fast_find_block = false,
cc->migrate_pfn = low_pfn = block_end_pfn,
block_start_pfn = block_end_pfn,
block_end_pfn += pageblock_nr_pages) {
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
/*
* This can potentially iterate a massively long zone with
* many pageblocks unsuitable, so periodically check if we
* need to schedule.
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
*/
if (!(low_pfn % (SWAP_CLUSTER_MAX * pageblock_nr_pages)))
mm, compaction: do not consider a need to reschedule as contention Scanning on large machines can take a considerable length of time and eventually need to be rescheduled. This is treated as an abort event but that's not appropriate as the attempt is likely to be retried after making numerous checks and taking another cycle through the page allocator. This patch will check the need to reschedule if necessary but continue the scanning. The main benefit is reduced scanning when compaction is taking a long time or the machine is over-saturated. It also avoids an unnecessary exit of compaction that ends up being retried by the page allocator in the outer loop. 5.0.0-rc1 5.0.0-rc1 synccached-v3r16 noresched-v3r17 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2958.27 ( 0.00%) 2965.68 ( -0.25%) Amean fault-both-5 4091.90 ( 0.00%) 3995.90 ( 2.35%) Amean fault-both-7 5803.05 ( 0.00%) 5842.12 ( -0.67%) Amean fault-both-12 9481.06 ( 0.00%) 9550.87 ( -0.74%) Amean fault-both-18 14141.51 ( 0.00%) 13304.72 ( 5.92%) Amean fault-both-24 16438.00 ( 0.00%) 14618.59 ( 11.07%) Amean fault-both-30 17531.72 ( 0.00%) 16650.96 ( 5.02%) Amean fault-both-32 17101.96 ( 0.00%) 17145.15 ( -0.25%) Link: http://lkml.kernel.org/r/20190118175136.31341-18-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:24 +00:00
cond_resched();
page = pageblock_pfn_to_page(block_start_pfn,
block_end_pfn, cc->zone);
mm, compaction: reduce zone checking frequency in the migration scanner The unification of the migrate and free scanner families of function has highlighted a difference in how the scanners ensure they only isolate pages of the intended zone. This is important for taking zone lock or lru lock of the correct zone. Due to nodes overlapping, it is however possible to encounter a different zone within the range of the zone being compacted. The free scanner, since its inception by commit 748446bb6b5a ("mm: compaction: memory compaction core"), has been checking the zone of the first valid page in a pageblock, and skipping the whole pageblock if the zone does not match. This checking was completely missing from the migration scanner at first, and later added by commit dc9086004b3d ("mm: compaction: check for overlapping nodes during isolation for migration") in a reaction to a bug report. But the zone comparison in migration scanner is done once per a single scanned page, which is more defensive and thus more costly than a check per pageblock. This patch unifies the checking done in both scanners to once per pageblock, through a new pageblock_pfn_to_page() function, which also includes pfn_valid() checks. It is more defensive than the current free scanner checks, as it checks both the first and last page of the pageblock, but less defensive by the migration scanner per-page checks. It assumes that node overlapping may result (on some architecture) in a boundary between two nodes falling into the middle of a pageblock, but that there cannot be a node0 node1 node0 interleaving within a single pageblock. The result is more code being shared and a bit less per-page CPU cost in the migration scanner. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:11 +00:00
if (!page)
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
continue;
mm, compaction: keep migration source private to a single compaction instance Due to either a fast search of the free list or a linear scan, it is possible for multiple compaction instances to pick the same pageblock for migration. This is lucky for one scanner and increased scanning for all the others. It also allows a race between requests on which first allocates the resulting free block. This patch tests and updates the pageblock skip for the migration scanner carefully. When isolating a block, it will check and skip if the block is already in use. Once the zone lock is acquired, it will be rechecked so that only one scanner can set the pageblock skip for exclusive use. Any scanner contending will continue with a linear scan. The skip bit is still set if no pages can be isolated in a range. While this may result in redundant scanning, it avoids unnecessarily acquiring the zone lock when there are no suitable migration sources. 1-socket thpscale Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3390.40 ( 0.00%) 3024.41 ( 10.80%) Amean fault-both-5 5082.28 ( 0.00%) 4749.30 ( 6.55%) Amean fault-both-7 7012.51 ( 0.00%) 6454.95 ( 7.95%) Amean fault-both-12 11346.63 ( 0.00%) 10324.83 ( 9.01%) Amean fault-both-18 15324.19 ( 0.00%) 12896.82 * 15.84%* Amean fault-both-24 16088.50 ( 0.00%) 13470.60 * 16.27%* Amean fault-both-30 18723.42 ( 0.00%) 17143.99 ( 8.44%) Amean fault-both-32 18612.01 ( 0.00%) 17743.91 ( 4.66%) 5.0.0-rc1 5.0.0-rc1 findmig-v3r15 isolmig-v3r15 Percentage huge-3 89.83 ( 0.00%) 92.96 ( 3.48%) Percentage huge-5 91.96 ( 0.00%) 93.26 ( 1.41%) Percentage huge-7 92.85 ( 0.00%) 93.63 ( 0.84%) Percentage huge-12 92.74 ( 0.00%) 92.80 ( 0.07%) Percentage huge-18 91.71 ( 0.00%) 91.62 ( -0.10%) Percentage huge-24 92.13 ( 0.00%) 91.50 ( -0.69%) Percentage huge-30 93.79 ( 0.00%) 92.73 ( -1.13%) Percentage huge-32 91.27 ( 0.00%) 91.94 ( 0.74%) This shows a reasonable reduction in latency as multiple compaction scanners do not operate on the same blocks with a similar allocation success rate. Compaction migrate scanned 41093126 25646769 Migration scan rates are reduced by 38%. Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:58 +00:00
/*
* If isolation recently failed, do not retry. Only check the
* pageblock once. COMPACT_CLUSTER_MAX causes a pageblock
* to be visited multiple times. Assume skip was checked
* before making it "skip" so other compaction instances do
* not scan the same block.
*/
if (IS_ALIGNED(low_pfn, pageblock_nr_pages) &&
!fast_find_block && !isolation_suitable(cc, page))
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
continue;
/*
* For async compaction, also only scan in MOVABLE blocks
* without huge pages. Async compaction is optimistic to see
* if the minimum amount of work satisfies the allocation.
* The cached PFN is updated as it's possible that all
* remaining blocks between source and target are unsuitable
* and the compaction scanners fail to meet.
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
*/
if (!suitable_migration_source(cc, page)) {
update_cached_migrate(cc, block_end_pfn);
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
continue;
}
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
/* Perform the isolation */
if (isolate_migratepages_block(cc, low_pfn, block_end_pfn,
isolate_mode))
mm, compaction: move pageblock checks up from isolate_migratepages_range() isolate_migratepages_range() is the main function of the compaction scanner, called either on a single pageblock by isolate_migratepages() during regular compaction, or on an arbitrary range by CMA's __alloc_contig_migrate_range(). It currently perfoms two pageblock-wide compaction suitability checks, and because of the CMA callpath, it tracks if it crossed a pageblock boundary in order to repeat those checks. However, closer inspection shows that those checks are always true for CMA: - isolation_suitable() is true because CMA sets cc->ignore_skip_hint to true - migrate_async_suitable() check is skipped because CMA uses sync compaction We can therefore move the compaction-specific checks to isolate_migratepages() and simplify isolate_migratepages_range(). Furthermore, we can mimic the freepage scanner family of functions, which has isolate_freepages_block() function called both by compaction from isolate_freepages() and by CMA from isolate_freepages_range(), where each use-case adds own specific glue code. This allows further code simplification. Thus, we rename isolate_migratepages_range() to isolate_migratepages_block() and limit its functionality to a single pageblock (or its subset). For CMA, a new different isolate_migratepages_range() is created as a CMA-specific wrapper for the _block() function. The checks specific to compaction are moved to isolate_migratepages(). As part of the unification of these two families of functions, we remove the redundant zone parameter where applicable, since zone pointer is already passed in cc->zone. Furthermore, going back to compact_zone() and compact_finished() when pageblock is found unsuitable (now by isolate_migratepages()) is wasteful - the checks are meant to skip pageblocks quickly. The patch therefore also introduces a simple loop into isolate_migratepages() so that it does not return immediately on failed pageblock checks, but keeps going until isolate_migratepages_range() gets called once. Similarily to isolate_freepages(), the function periodically checks if it needs to reschedule or abort async compaction. [iamjoonsoo.kim@lge.com: fix isolated page counting bug in compaction] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:09 +00:00
return ISOLATE_ABORT;
/*
* Either we isolated something and proceed with migration. Or
* we failed and compact_zone should decide if we should
* continue or not.
*/
break;
}
return cc->nr_migratepages ? ISOLATE_SUCCESS : ISOLATE_NONE;
}
/*
* order == -1 is expected when compacting via
* /proc/sys/vm/compact_memory
*/
static inline bool is_via_compact_memory(int order)
{
return order == -1;
}
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
static bool kswapd_is_running(pg_data_t *pgdat)
{
return pgdat->kswapd && (pgdat->kswapd->state == TASK_RUNNING);
}
/*
* A zone's fragmentation score is the external fragmentation wrt to the
mm/compaction: correct deferral logic for proactive compaction should_proactive_compact_node() returns true when sum of the weighted fragmentation score of all the zones in the node is greater than the wmark_high of compaction, which then triggers the proactive compaction that operates on the individual zones of the node. But proactive compaction runs on the zone only when its weighted fragmentation score is greater than wmark_low(=wmark_high - 10). This means that the sum of the weighted fragmentation scores of all the zones can exceed the wmark_high but individual weighted fragmentation zone scores can still be less than wmark_low which makes the unnecessary trigger of the proactive compaction only to return doing nothing. Issue with the return of proactive compaction with out even trying is its deferral. It is simply deferred for 1 << COMPACT_MAX_DEFER_SHIFT if the scores across the proactive compaction is same, thinking that compaction didn't make any progress but in reality it didn't even try. With the delay between successive retries for proactive compaction is 500msec, it can result into the deferral for ~30sec with out even trying the proactive compaction. Test scenario is that: compaction_proactiveness=50 thus the wmark_low = 50 and wmark_high = 60. System have 2 zones(Normal and Movable) with sizes 5GB and 6GB respectively. After opening some apps on the android, the weighted fragmentation scores of these zones are 47 and 49 respectively. Since the sum of these fragmentation scores are above the wmark_high which triggers the proactive compaction and there since the individual zones weighted fragmentation scores are below wmark_low, it returns without trying the proactive compaction. As a result the weighted fragmentation scores of the zones are still 47 and 49 which makes the existing logic to defer the compaction thinking that noprogress is made across the compaction. Fix this by checking just zone fragmentation score, not the weighted, in __compact_finished() and use the zones weighted fragmentation score in fragmentation_score_node(). In the test case above, If the weighted average of is above wmark_high, then individual score (not adjusted) of atleast one zone has to be above wmark_high. Thus it avoids the unnecessary trigger and deferrals of the proactive compaction. Link: https://lkml.kernel.org/r/1610989938-31374-1-git-send-email-charante@codeaurora.org Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Suggested-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Acked-by: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Vinayak Menon <vinmenon@codeaurora.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:32 +00:00
* COMPACTION_HPAGE_ORDER. It returns a value in the range [0, 100].
*/
static unsigned int fragmentation_score_zone(struct zone *zone)
{
return extfrag_for_order(zone, COMPACTION_HPAGE_ORDER);
}
/*
* A weighted zone's fragmentation score is the external fragmentation
* wrt to the COMPACTION_HPAGE_ORDER scaled by the zone's size. It
* returns a value in the range [0, 100].
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
*
* The scaling factor ensures that proactive compaction focuses on larger
* zones like ZONE_NORMAL, rather than smaller, specialized zones like
* ZONE_DMA32. For smaller zones, the score value remains close to zero,
* and thus never exceeds the high threshold for proactive compaction.
*/
mm/compaction: correct deferral logic for proactive compaction should_proactive_compact_node() returns true when sum of the weighted fragmentation score of all the zones in the node is greater than the wmark_high of compaction, which then triggers the proactive compaction that operates on the individual zones of the node. But proactive compaction runs on the zone only when its weighted fragmentation score is greater than wmark_low(=wmark_high - 10). This means that the sum of the weighted fragmentation scores of all the zones can exceed the wmark_high but individual weighted fragmentation zone scores can still be less than wmark_low which makes the unnecessary trigger of the proactive compaction only to return doing nothing. Issue with the return of proactive compaction with out even trying is its deferral. It is simply deferred for 1 << COMPACT_MAX_DEFER_SHIFT if the scores across the proactive compaction is same, thinking that compaction didn't make any progress but in reality it didn't even try. With the delay between successive retries for proactive compaction is 500msec, it can result into the deferral for ~30sec with out even trying the proactive compaction. Test scenario is that: compaction_proactiveness=50 thus the wmark_low = 50 and wmark_high = 60. System have 2 zones(Normal and Movable) with sizes 5GB and 6GB respectively. After opening some apps on the android, the weighted fragmentation scores of these zones are 47 and 49 respectively. Since the sum of these fragmentation scores are above the wmark_high which triggers the proactive compaction and there since the individual zones weighted fragmentation scores are below wmark_low, it returns without trying the proactive compaction. As a result the weighted fragmentation scores of the zones are still 47 and 49 which makes the existing logic to defer the compaction thinking that noprogress is made across the compaction. Fix this by checking just zone fragmentation score, not the weighted, in __compact_finished() and use the zones weighted fragmentation score in fragmentation_score_node(). In the test case above, If the weighted average of is above wmark_high, then individual score (not adjusted) of atleast one zone has to be above wmark_high. Thus it avoids the unnecessary trigger and deferrals of the proactive compaction. Link: https://lkml.kernel.org/r/1610989938-31374-1-git-send-email-charante@codeaurora.org Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Suggested-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Acked-by: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Vinayak Menon <vinmenon@codeaurora.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:32 +00:00
static unsigned int fragmentation_score_zone_weighted(struct zone *zone)
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
{
unsigned long score;
mm/compaction: correct deferral logic for proactive compaction should_proactive_compact_node() returns true when sum of the weighted fragmentation score of all the zones in the node is greater than the wmark_high of compaction, which then triggers the proactive compaction that operates on the individual zones of the node. But proactive compaction runs on the zone only when its weighted fragmentation score is greater than wmark_low(=wmark_high - 10). This means that the sum of the weighted fragmentation scores of all the zones can exceed the wmark_high but individual weighted fragmentation zone scores can still be less than wmark_low which makes the unnecessary trigger of the proactive compaction only to return doing nothing. Issue with the return of proactive compaction with out even trying is its deferral. It is simply deferred for 1 << COMPACT_MAX_DEFER_SHIFT if the scores across the proactive compaction is same, thinking that compaction didn't make any progress but in reality it didn't even try. With the delay between successive retries for proactive compaction is 500msec, it can result into the deferral for ~30sec with out even trying the proactive compaction. Test scenario is that: compaction_proactiveness=50 thus the wmark_low = 50 and wmark_high = 60. System have 2 zones(Normal and Movable) with sizes 5GB and 6GB respectively. After opening some apps on the android, the weighted fragmentation scores of these zones are 47 and 49 respectively. Since the sum of these fragmentation scores are above the wmark_high which triggers the proactive compaction and there since the individual zones weighted fragmentation scores are below wmark_low, it returns without trying the proactive compaction. As a result the weighted fragmentation scores of the zones are still 47 and 49 which makes the existing logic to defer the compaction thinking that noprogress is made across the compaction. Fix this by checking just zone fragmentation score, not the weighted, in __compact_finished() and use the zones weighted fragmentation score in fragmentation_score_node(). In the test case above, If the weighted average of is above wmark_high, then individual score (not adjusted) of atleast one zone has to be above wmark_high. Thus it avoids the unnecessary trigger and deferrals of the proactive compaction. Link: https://lkml.kernel.org/r/1610989938-31374-1-git-send-email-charante@codeaurora.org Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Suggested-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Acked-by: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Vinayak Menon <vinmenon@codeaurora.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:32 +00:00
score = zone->present_pages * fragmentation_score_zone(zone);
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
return div64_ul(score, zone->zone_pgdat->node_present_pages + 1);
}
/*
* The per-node proactive (background) compaction process is started by its
* corresponding kcompactd thread when the node's fragmentation score
* exceeds the high threshold. The compaction process remains active till
* the node's score falls below the low threshold, or one of the back-off
* conditions is met.
*/
static unsigned int fragmentation_score_node(pg_data_t *pgdat)
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
{
unsigned int score = 0;
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
int zoneid;
for (zoneid = 0; zoneid < MAX_NR_ZONES; zoneid++) {
struct zone *zone;
zone = &pgdat->node_zones[zoneid];
mm/compaction: correct deferral logic for proactive compaction should_proactive_compact_node() returns true when sum of the weighted fragmentation score of all the zones in the node is greater than the wmark_high of compaction, which then triggers the proactive compaction that operates on the individual zones of the node. But proactive compaction runs on the zone only when its weighted fragmentation score is greater than wmark_low(=wmark_high - 10). This means that the sum of the weighted fragmentation scores of all the zones can exceed the wmark_high but individual weighted fragmentation zone scores can still be less than wmark_low which makes the unnecessary trigger of the proactive compaction only to return doing nothing. Issue with the return of proactive compaction with out even trying is its deferral. It is simply deferred for 1 << COMPACT_MAX_DEFER_SHIFT if the scores across the proactive compaction is same, thinking that compaction didn't make any progress but in reality it didn't even try. With the delay between successive retries for proactive compaction is 500msec, it can result into the deferral for ~30sec with out even trying the proactive compaction. Test scenario is that: compaction_proactiveness=50 thus the wmark_low = 50 and wmark_high = 60. System have 2 zones(Normal and Movable) with sizes 5GB and 6GB respectively. After opening some apps on the android, the weighted fragmentation scores of these zones are 47 and 49 respectively. Since the sum of these fragmentation scores are above the wmark_high which triggers the proactive compaction and there since the individual zones weighted fragmentation scores are below wmark_low, it returns without trying the proactive compaction. As a result the weighted fragmentation scores of the zones are still 47 and 49 which makes the existing logic to defer the compaction thinking that noprogress is made across the compaction. Fix this by checking just zone fragmentation score, not the weighted, in __compact_finished() and use the zones weighted fragmentation score in fragmentation_score_node(). In the test case above, If the weighted average of is above wmark_high, then individual score (not adjusted) of atleast one zone has to be above wmark_high. Thus it avoids the unnecessary trigger and deferrals of the proactive compaction. Link: https://lkml.kernel.org/r/1610989938-31374-1-git-send-email-charante@codeaurora.org Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Suggested-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Acked-by: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Vinayak Menon <vinmenon@codeaurora.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-02-24 20:09:32 +00:00
score += fragmentation_score_zone_weighted(zone);
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
}
return score;
}
static unsigned int fragmentation_score_wmark(pg_data_t *pgdat, bool low)
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
{
unsigned int wmark_low;
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/*
* Cap the low watermark to avoid excessive compaction
* activity in case a user sets the proactiveness tunable
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
* close to 100 (maximum).
*/
wmark_low = max(100U - sysctl_compaction_proactiveness, 5U);
return low ? wmark_low : min(wmark_low + 10, 100U);
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
}
static bool should_proactive_compact_node(pg_data_t *pgdat)
{
int wmark_high;
if (!sysctl_compaction_proactiveness || kswapd_is_running(pgdat))
return false;
wmark_high = fragmentation_score_wmark(pgdat, false);
return fragmentation_score_node(pgdat) > wmark_high;
}
static enum compact_result __compact_finished(struct compact_control *cc)
{
mm: compaction: partially revert capture of suitable high-order page Eric Wong reported on 3.7 and 3.8-rc2 that ppoll() got stuck when waiting for POLLIN on a local TCP socket. It was easier to trigger if there was disk IO and dirty pages at the same time and he bisected it to commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). The intention of that patch was to improve high-order allocations under memory pressure after changes made to reclaim in 3.6 drastically hurt THP allocations but the approach was flawed. For Eric, the problem was that page->pfmemalloc was not being cleared for captured pages leading to a poor interaction with swap-over-NFS support causing the packets to be dropped. However, I identified a few more problems with the patch including the fact that it can increase contention on zone->lock in some cases which could result in async direct compaction being aborted early. In retrospect the capture patch took the wrong approach. What it should have done is mark the pageblock being migrated as MIGRATE_ISOLATE if it was allocating for THP and avoided races that way. While the patch was showing to improve allocation success rates at the time, the benefit is marginal given the relative complexity and it should be revisited from scratch in the context of the other reclaim-related changes that have taken place since the patch was first written and tested. This patch partially reverts commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). Reported-and-tested-by: Eric Wong <normalperson@yhbt.net> Tested-by: Eric Dumazet <eric.dumazet@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: David Miller <davem@davemloft.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-11 22:32:16 +00:00
unsigned int order;
const int migratetype = cc->migratetype;
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
int ret;
/* Compaction run completes if the migrate and free scanner meet */
mm, compaction: more robust check for scanners meeting Assorted compaction cleanups and optimizations. The interesting patches are 4 and 5. In 4, skipping of compound pages in single iteration is improved for migration scanner, so it works also for !PageLRU compound pages such as hugetlbfs, slab etc. Patch 5 introduces this kind of skipping in the free scanner. The trick is that we can read compound_order() without any protection, if we are careful to filter out values larger than MAX_ORDER. The only danger is that we skip too much. The same trick was already used for reading the freepage order in the migrate scanner. To demonstrate improvements of Patches 4 and 5 I've run stress-highalloc from mmtests, set to simulate THP allocations (including __GFP_COMP) on a 4GB system where 1GB was occupied by hugetlbfs pages. I'll include just the relevant stats: Patch 3 Patch 4 Patch 5 Compaction stalls 7523 7529 7515 Compaction success 323 304 322 Compaction failures 7200 7224 7192 Page migrate success 247778 264395 240737 Page migrate failure 15358 33184 21621 Compaction pages isolated 906928 980192 909983 Compaction migrate scanned 2005277 1692805 1498800 Compaction free scanned 13255284 11539986 9011276 Compaction cost 288 305 277 With 5 iterations per patch, the results are still noisy, but we can see that Patch 4 does reduce migrate_scanned by 15% thanks to skipping the hugetlbfs pages at once. Interestingly, free_scanned is also reduced and I have no idea why. Patch 5 further reduces free_scanned as expected, by 15%. Other stats are unaffected modulo noise. [1] https://lkml.org/lkml/2015/1/19/158 This patch (of 5): Compaction should finish when the migration and free scanner meet, i.e. they reach the same pageblock. Currently however, the test in compact_finished() simply just compares the exact pfns, which may yield a false negative when the free scanner position is in the middle of a pageblock and the migration scanner reaches the begining of the same pageblock. This hasn't been a problem until commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner") allowed the free scanner position to be in the middle of a pageblock between invocations. The hot-fix 1d5bfe1ffb5b ("mm, compaction: prevent infinite loop in compact_zone") prevented the issue by adding a special check in the migration scanner to satisfy the current detection of scanners meeting. However, the proper fix is to make the detection more robust. This patch introduces the compact_scanners_met() function that returns true when the free scanner position is in the same or lower pageblock than the migration scanner. The special case in isolate_migratepages() introduced by 1d5bfe1ffb5b is removed. Suggested-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:36 +00:00
if (compact_scanners_met(cc)) {
mm: compaction: reset scanner positions immediately when they meet Compaction used to start its migrate and free page scaners at the zone's lowest and highest pfn, respectively. Later, caching was introduced to remember the scanners' progress across compaction attempts so that pageblocks are not re-scanned uselessly. Additionally, pageblocks where isolation failed are marked to be quickly skipped when encountered again in future compactions. Currently, both the reset of cached pfn's and clearing of the pageblock skip information for a zone is done in __reset_isolation_suitable(). This function gets called when: - compaction is restarting after being deferred - compact_blockskip_flush flag is set in compact_finished() when the scanners meet (and not again cleared when direct compaction succeeds in allocation) and kswapd acts upon this flag before going to sleep This behavior is suboptimal for several reasons: - when direct sync compaction is called after async compaction fails (in the allocation slowpath), it will effectively do nothing, unless kswapd happens to process the compact_blockskip_flush flag meanwhile. This is racy and goes against the purpose of sync compaction to more thoroughly retry the compaction of a zone where async compaction has failed. The restart-after-deferring path cannot help here as deferring happens only after the sync compaction fails. It is also done only for the preferred zone, while the compaction might be done for a fallback zone. - the mechanism of marking pageblock to be skipped has little value since the cached pfn's are reset only together with the pageblock skip flags. This effectively limits pageblock skip usage to parallel compactions. This patch changes compact_finished() so that cached pfn's are reset immediately when the scanners meet. Clearing pageblock skip flags is unchanged, as well as the other situations where cached pfn's are reset. This allows the sync-after-async compaction to retry pageblocks not marked as skipped, such as blocks !MIGRATE_MOVABLE blocks that async compactions now skips without marking them. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@redhat.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:11 +00:00
/* Let the next compaction start anew. */
reset_cached_positions(cc->zone);
mm: compaction: reset scanner positions immediately when they meet Compaction used to start its migrate and free page scaners at the zone's lowest and highest pfn, respectively. Later, caching was introduced to remember the scanners' progress across compaction attempts so that pageblocks are not re-scanned uselessly. Additionally, pageblocks where isolation failed are marked to be quickly skipped when encountered again in future compactions. Currently, both the reset of cached pfn's and clearing of the pageblock skip information for a zone is done in __reset_isolation_suitable(). This function gets called when: - compaction is restarting after being deferred - compact_blockskip_flush flag is set in compact_finished() when the scanners meet (and not again cleared when direct compaction succeeds in allocation) and kswapd acts upon this flag before going to sleep This behavior is suboptimal for several reasons: - when direct sync compaction is called after async compaction fails (in the allocation slowpath), it will effectively do nothing, unless kswapd happens to process the compact_blockskip_flush flag meanwhile. This is racy and goes against the purpose of sync compaction to more thoroughly retry the compaction of a zone where async compaction has failed. The restart-after-deferring path cannot help here as deferring happens only after the sync compaction fails. It is also done only for the preferred zone, while the compaction might be done for a fallback zone. - the mechanism of marking pageblock to be skipped has little value since the cached pfn's are reset only together with the pageblock skip flags. This effectively limits pageblock skip usage to parallel compactions. This patch changes compact_finished() so that cached pfn's are reset immediately when the scanners meet. Clearing pageblock skip flags is unchanged, as well as the other situations where cached pfn's are reset. This allows the sync-after-async compaction to retry pageblocks not marked as skipped, such as blocks !MIGRATE_MOVABLE blocks that async compactions now skips without marking them. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@redhat.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:11 +00:00
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
/*
* Mark that the PG_migrate_skip information should be cleared
mm, kswapd: replace kswapd compaction with waking up kcompactd Similarly to direct reclaim/compaction, kswapd attempts to combine reclaim and compaction to attempt making memory allocation of given order available. The details differ from direct reclaim e.g. in having high watermark as a goal. The code involved in kswapd's reclaim/compaction decisions has evolved to be quite complex. Testing reveals that it doesn't actually work in at least one scenario, and closer inspection suggests that it could be greatly simplified without compromising on the goal (make high-order page available) or efficiency (don't reclaim too much). The simplification relieas of doing all compaction in kcompactd, which is simply woken up when high watermarks are reached by kswapd's reclaim. The scenario where kswapd compaction doesn't work was found with mmtests test stress-highalloc configured to attempt order-9 allocations without direct reclaim, just waking up kswapd. There was no compaction attempt from kswapd during the whole test. Some added instrumentation shows what happens: - balance_pgdat() sets end_zone to Normal, as it's not balanced - reclaim is attempted on DMA zone, which sets nr_attempted to 99, but it cannot reclaim anything, so sc.nr_reclaimed is 0 - for zones DMA32 and Normal, kswapd_shrink_zone uses testorder=0, so it merely checks if high watermarks were reached for base pages. This is true, so no reclaim is attempted. For DMA, testorder=0 wasn't used, as compaction_suitable() returned COMPACT_SKIPPED - even though the pgdat_needs_compaction flag wasn't set to false, no compaction happens due to the condition sc.nr_reclaimed > nr_attempted being false (as 0 < 99) - priority-- due to nr_reclaimed being 0, repeat until priority reaches 0 pgdat_balanced() is false as only the small zone DMA appears balanced (curiously in that check, watermark appears OK and compaction_suitable() returns COMPACT_PARTIAL, because a lower classzone_idx is used there) Now, even if it was decided that reclaim shouldn't be attempted on the DMA zone, the scenario would be the same, as (sc.nr_reclaimed=0 > nr_attempted=0) is also false. The condition really should use >= as the comment suggests. Then there is a mismatch in the check for setting pgdat_needs_compaction to false using low watermark, while the rest uses high watermark, and who knows what other subtlety. Hopefully this demonstrates that this is unsustainable. Luckily we can simplify this a lot. The reclaim/compaction decisions make sense for direct reclaim scenario, but in kswapd, our primary goal is to reach high watermark in order-0 pages. Afterwards we can attempt compaction just once. Unlike direct reclaim, we don't reclaim extra pages (over the high watermark), the current code already disallows it for good reasons. After this patch, we simply wake up kcompactd to process the pgdat, after we have either succeeded or failed to reach the high watermarks in kswapd, which goes to sleep. We pass kswapd's order and classzone_idx, so kcompactd can apply the same criteria to determine which zones are worth compacting. Note that we use the classzone_idx from wakeup_kswapd(), not balanced_classzone_idx which can include higher zones that kswapd tried to balance too, but didn't consider them in pgdat_balanced(). Since kswapd now cannot create high-order pages itself, we need to adjust how it determines the zones to be balanced. The key element here is adding a "highorder" parameter to zone_balanced, which, when set to false, makes it consider only order-0 watermark instead of the desired higher order (this was done previously by kswapd_shrink_zone(), but not elsewhere). This false is passed for example in pgdat_balanced(). Importantly, wakeup_kswapd() uses true to make sure kswapd and thus kcompactd are woken up for a high-order allocation failure. The last thing is to decide what to do with pageblock_skip bitmap handling. Compaction maintains a pageblock_skip bitmap to record pageblocks where isolation recently failed. This bitmap can be reset by three ways: 1) direct compaction is restarting after going through the full deferred cycle 2) kswapd goes to sleep, and some other direct compaction has previously finished scanning the whole zone and set zone->compact_blockskip_flush. Note that a successful direct compaction clears this flag. 3) compaction was invoked manually via trigger in /proc The case 2) is somewhat fuzzy to begin with, but after introducing kcompactd we should update it. The check for direct compaction in 1), and to set the flush flag in 2) use current_is_kswapd(), which doesn't work for kcompactd. Thus, this patch adds bool direct_compaction to compact_control to use in 2). For the case 1) we remove the check completely - unlike the former kswapd compaction, kcompactd does use the deferred compaction functionality, so flushing tied to restarting from deferred compaction makes sense here. Note that when kswapd goes to sleep, kcompactd is woken up, so it will see the flushed pageblock_skip bits. This is different from when the former kswapd compaction observed the bits and I believe it makes more sense. Kcompactd can afford to be more thorough than a direct compaction trying to limit allocation latency, or kswapd whose primary goal is to reclaim. For testing, I used stress-highalloc configured to do order-9 allocations with GFP_NOWAIT|__GFP_HIGH|__GFP_COMP, so they relied just on kswapd/kcompactd reclaim/compaction (the interfering kernel builds in phases 1 and 2 work as usual): stress-highalloc 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Success 1 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 1 Mean 1.40 ( 0.00%) 6.20 (-55.00%) Success 1 Max 2.00 ( 0.00%) 7.00 (-16.67%) Success 2 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 2 Mean 1.80 ( 0.00%) 6.40 (-52.38%) Success 2 Max 3.00 ( 0.00%) 7.00 (-16.67%) Success 3 Min 34.00 ( 0.00%) 62.00 ( 1.59%) Success 3 Mean 41.80 ( 0.00%) 63.80 ( 1.24%) Success 3 Max 53.00 ( 0.00%) 65.00 ( 2.99%) User 3166.67 3181.09 System 1153.37 1158.25 Elapsed 1768.53 1799.37 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Direct pages scanned 32938 32797 Kswapd pages scanned 2183166 2202613 Kswapd pages reclaimed 2152359 2143524 Direct pages reclaimed 32735 32545 Percentage direct scans 1% 1% THP fault alloc 579 612 THP collapse alloc 304 316 THP splits 0 0 THP fault fallback 793 778 THP collapse fail 11 16 Compaction stalls 1013 1007 Compaction success 92 67 Compaction failures 920 939 Page migrate success 238457 721374 Page migrate failure 23021 23469 Compaction pages isolated 504695 1479924 Compaction migrate scanned 661390 8812554 Compaction free scanned 13476658 84327916 Compaction cost 262 838 After this patch we see improvements in allocation success rate (especially for phase 3) along with increased compaction activity. The compaction stalls (direct compaction) in the interfering kernel builds (probably THP's) also decreased somewhat thanks to kcompactd activity, yet THP alloc successes improved a bit. Note that elapsed and user time isn't so useful for this benchmark, because of the background interference being unpredictable. It's just to quickly spot some major unexpected differences. System time is somewhat more useful and that didn't increase. Also (after adjusting mmtests' ftrace monitor): Time kswapd awake 2547781 2269241 Time kcompactd awake 0 119253 Time direct compacting 939937 557649 Time kswapd compacting 0 0 Time kcompactd compacting 0 119099 The decrease of overal time spent compacting appears to not match the increased compaction stats. I suspect the tasks get rescheduled and since the ftrace monitor doesn't see that, the reported time is wall time, not CPU time. But arguably direct compactors care about overall latency anyway, whether busy compacting or waiting for CPU doesn't matter. And that latency seems to almost halved. It's also interesting how much time kswapd spent awake just going through all the priorities and failing to even try compacting, over and over. We can also configure stress-highalloc to perform both direct reclaim/compaction and wakeup kswapd/kcompactd, by using GFP_KERNEL|__GFP_HIGH|__GFP_COMP: stress-highalloc 4.5-rc1+before 4.5-rc1+after -direct -direct Success 1 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 1 Mean 8.00 ( 0.00%) 10.00 (-19.05%) Success 1 Max 12.00 ( 0.00%) 11.00 ( 15.38%) Success 2 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 2 Mean 8.20 ( 0.00%) 10.00 (-16.28%) Success 2 Max 13.00 ( 0.00%) 11.00 ( 8.33%) Success 3 Min 75.00 ( 0.00%) 74.00 ( 1.33%) Success 3 Mean 75.60 ( 0.00%) 75.20 ( 0.53%) Success 3 Max 77.00 ( 0.00%) 76.00 ( 0.00%) User 3344.73 3246.04 System 1194.24 1172.29 Elapsed 1838.04 1836.76 4.5-rc1+before 4.5-rc1+after -direct -direct Direct pages scanned 125146 120966 Kswapd pages scanned 2119757 2135012 Kswapd pages reclaimed 2073183 2108388 Direct pages reclaimed 124909 120577 Percentage direct scans 5% 5% THP fault alloc 599 652 THP collapse alloc 323 354 THP splits 0 0 THP fault fallback 806 793 THP collapse fail 17 16 Compaction stalls 2457 2025 Compaction success 906 518 Compaction failures 1551 1507 Page migrate success 2031423 2360608 Page migrate failure 32845 40852 Compaction pages isolated 4129761 4802025 Compaction migrate scanned 11996712 21750613 Compaction free scanned 214970969 344372001 Compaction cost 2271 2694 In this scenario, this patch doesn't change the overall success rate as direct compaction already tries all it can. There's however significant reduction in direct compaction stalls (that is, the number of allocations that went into direct compaction). The number of successes (i.e. direct compaction stalls that ended up with successful allocation) is reduced by the same number. This means the offload to kcompactd is working as expected, and direct compaction is reduced either due to detecting contention, or compaction deferred by kcompactd. In the previous version of this patchset there was some apparent reduction of success rate, but the changes in this version (such as using sync compaction only), new baseline kernel, and/or averaging results from 5 executions (my bet), made this go away. Ftrace-based stats seem to roughly agree: Time kswapd awake 2532984 2326824 Time kcompactd awake 0 257916 Time direct compacting 864839 735130 Time kswapd compacting 0 0 Time kcompactd compacting 0 257585 Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:15 +00:00
* by kswapd when it goes to sleep. kcompactd does not set the
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
* flag itself as the decision to be clear should be directly
* based on an allocation request.
*/
mm, kswapd: replace kswapd compaction with waking up kcompactd Similarly to direct reclaim/compaction, kswapd attempts to combine reclaim and compaction to attempt making memory allocation of given order available. The details differ from direct reclaim e.g. in having high watermark as a goal. The code involved in kswapd's reclaim/compaction decisions has evolved to be quite complex. Testing reveals that it doesn't actually work in at least one scenario, and closer inspection suggests that it could be greatly simplified without compromising on the goal (make high-order page available) or efficiency (don't reclaim too much). The simplification relieas of doing all compaction in kcompactd, which is simply woken up when high watermarks are reached by kswapd's reclaim. The scenario where kswapd compaction doesn't work was found with mmtests test stress-highalloc configured to attempt order-9 allocations without direct reclaim, just waking up kswapd. There was no compaction attempt from kswapd during the whole test. Some added instrumentation shows what happens: - balance_pgdat() sets end_zone to Normal, as it's not balanced - reclaim is attempted on DMA zone, which sets nr_attempted to 99, but it cannot reclaim anything, so sc.nr_reclaimed is 0 - for zones DMA32 and Normal, kswapd_shrink_zone uses testorder=0, so it merely checks if high watermarks were reached for base pages. This is true, so no reclaim is attempted. For DMA, testorder=0 wasn't used, as compaction_suitable() returned COMPACT_SKIPPED - even though the pgdat_needs_compaction flag wasn't set to false, no compaction happens due to the condition sc.nr_reclaimed > nr_attempted being false (as 0 < 99) - priority-- due to nr_reclaimed being 0, repeat until priority reaches 0 pgdat_balanced() is false as only the small zone DMA appears balanced (curiously in that check, watermark appears OK and compaction_suitable() returns COMPACT_PARTIAL, because a lower classzone_idx is used there) Now, even if it was decided that reclaim shouldn't be attempted on the DMA zone, the scenario would be the same, as (sc.nr_reclaimed=0 > nr_attempted=0) is also false. The condition really should use >= as the comment suggests. Then there is a mismatch in the check for setting pgdat_needs_compaction to false using low watermark, while the rest uses high watermark, and who knows what other subtlety. Hopefully this demonstrates that this is unsustainable. Luckily we can simplify this a lot. The reclaim/compaction decisions make sense for direct reclaim scenario, but in kswapd, our primary goal is to reach high watermark in order-0 pages. Afterwards we can attempt compaction just once. Unlike direct reclaim, we don't reclaim extra pages (over the high watermark), the current code already disallows it for good reasons. After this patch, we simply wake up kcompactd to process the pgdat, after we have either succeeded or failed to reach the high watermarks in kswapd, which goes to sleep. We pass kswapd's order and classzone_idx, so kcompactd can apply the same criteria to determine which zones are worth compacting. Note that we use the classzone_idx from wakeup_kswapd(), not balanced_classzone_idx which can include higher zones that kswapd tried to balance too, but didn't consider them in pgdat_balanced(). Since kswapd now cannot create high-order pages itself, we need to adjust how it determines the zones to be balanced. The key element here is adding a "highorder" parameter to zone_balanced, which, when set to false, makes it consider only order-0 watermark instead of the desired higher order (this was done previously by kswapd_shrink_zone(), but not elsewhere). This false is passed for example in pgdat_balanced(). Importantly, wakeup_kswapd() uses true to make sure kswapd and thus kcompactd are woken up for a high-order allocation failure. The last thing is to decide what to do with pageblock_skip bitmap handling. Compaction maintains a pageblock_skip bitmap to record pageblocks where isolation recently failed. This bitmap can be reset by three ways: 1) direct compaction is restarting after going through the full deferred cycle 2) kswapd goes to sleep, and some other direct compaction has previously finished scanning the whole zone and set zone->compact_blockskip_flush. Note that a successful direct compaction clears this flag. 3) compaction was invoked manually via trigger in /proc The case 2) is somewhat fuzzy to begin with, but after introducing kcompactd we should update it. The check for direct compaction in 1), and to set the flush flag in 2) use current_is_kswapd(), which doesn't work for kcompactd. Thus, this patch adds bool direct_compaction to compact_control to use in 2). For the case 1) we remove the check completely - unlike the former kswapd compaction, kcompactd does use the deferred compaction functionality, so flushing tied to restarting from deferred compaction makes sense here. Note that when kswapd goes to sleep, kcompactd is woken up, so it will see the flushed pageblock_skip bits. This is different from when the former kswapd compaction observed the bits and I believe it makes more sense. Kcompactd can afford to be more thorough than a direct compaction trying to limit allocation latency, or kswapd whose primary goal is to reclaim. For testing, I used stress-highalloc configured to do order-9 allocations with GFP_NOWAIT|__GFP_HIGH|__GFP_COMP, so they relied just on kswapd/kcompactd reclaim/compaction (the interfering kernel builds in phases 1 and 2 work as usual): stress-highalloc 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Success 1 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 1 Mean 1.40 ( 0.00%) 6.20 (-55.00%) Success 1 Max 2.00 ( 0.00%) 7.00 (-16.67%) Success 2 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 2 Mean 1.80 ( 0.00%) 6.40 (-52.38%) Success 2 Max 3.00 ( 0.00%) 7.00 (-16.67%) Success 3 Min 34.00 ( 0.00%) 62.00 ( 1.59%) Success 3 Mean 41.80 ( 0.00%) 63.80 ( 1.24%) Success 3 Max 53.00 ( 0.00%) 65.00 ( 2.99%) User 3166.67 3181.09 System 1153.37 1158.25 Elapsed 1768.53 1799.37 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Direct pages scanned 32938 32797 Kswapd pages scanned 2183166 2202613 Kswapd pages reclaimed 2152359 2143524 Direct pages reclaimed 32735 32545 Percentage direct scans 1% 1% THP fault alloc 579 612 THP collapse alloc 304 316 THP splits 0 0 THP fault fallback 793 778 THP collapse fail 11 16 Compaction stalls 1013 1007 Compaction success 92 67 Compaction failures 920 939 Page migrate success 238457 721374 Page migrate failure 23021 23469 Compaction pages isolated 504695 1479924 Compaction migrate scanned 661390 8812554 Compaction free scanned 13476658 84327916 Compaction cost 262 838 After this patch we see improvements in allocation success rate (especially for phase 3) along with increased compaction activity. The compaction stalls (direct compaction) in the interfering kernel builds (probably THP's) also decreased somewhat thanks to kcompactd activity, yet THP alloc successes improved a bit. Note that elapsed and user time isn't so useful for this benchmark, because of the background interference being unpredictable. It's just to quickly spot some major unexpected differences. System time is somewhat more useful and that didn't increase. Also (after adjusting mmtests' ftrace monitor): Time kswapd awake 2547781 2269241 Time kcompactd awake 0 119253 Time direct compacting 939937 557649 Time kswapd compacting 0 0 Time kcompactd compacting 0 119099 The decrease of overal time spent compacting appears to not match the increased compaction stats. I suspect the tasks get rescheduled and since the ftrace monitor doesn't see that, the reported time is wall time, not CPU time. But arguably direct compactors care about overall latency anyway, whether busy compacting or waiting for CPU doesn't matter. And that latency seems to almost halved. It's also interesting how much time kswapd spent awake just going through all the priorities and failing to even try compacting, over and over. We can also configure stress-highalloc to perform both direct reclaim/compaction and wakeup kswapd/kcompactd, by using GFP_KERNEL|__GFP_HIGH|__GFP_COMP: stress-highalloc 4.5-rc1+before 4.5-rc1+after -direct -direct Success 1 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 1 Mean 8.00 ( 0.00%) 10.00 (-19.05%) Success 1 Max 12.00 ( 0.00%) 11.00 ( 15.38%) Success 2 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 2 Mean 8.20 ( 0.00%) 10.00 (-16.28%) Success 2 Max 13.00 ( 0.00%) 11.00 ( 8.33%) Success 3 Min 75.00 ( 0.00%) 74.00 ( 1.33%) Success 3 Mean 75.60 ( 0.00%) 75.20 ( 0.53%) Success 3 Max 77.00 ( 0.00%) 76.00 ( 0.00%) User 3344.73 3246.04 System 1194.24 1172.29 Elapsed 1838.04 1836.76 4.5-rc1+before 4.5-rc1+after -direct -direct Direct pages scanned 125146 120966 Kswapd pages scanned 2119757 2135012 Kswapd pages reclaimed 2073183 2108388 Direct pages reclaimed 124909 120577 Percentage direct scans 5% 5% THP fault alloc 599 652 THP collapse alloc 323 354 THP splits 0 0 THP fault fallback 806 793 THP collapse fail 17 16 Compaction stalls 2457 2025 Compaction success 906 518 Compaction failures 1551 1507 Page migrate success 2031423 2360608 Page migrate failure 32845 40852 Compaction pages isolated 4129761 4802025 Compaction migrate scanned 11996712 21750613 Compaction free scanned 214970969 344372001 Compaction cost 2271 2694 In this scenario, this patch doesn't change the overall success rate as direct compaction already tries all it can. There's however significant reduction in direct compaction stalls (that is, the number of allocations that went into direct compaction). The number of successes (i.e. direct compaction stalls that ended up with successful allocation) is reduced by the same number. This means the offload to kcompactd is working as expected, and direct compaction is reduced either due to detecting contention, or compaction deferred by kcompactd. In the previous version of this patchset there was some apparent reduction of success rate, but the changes in this version (such as using sync compaction only), new baseline kernel, and/or averaging results from 5 executions (my bet), made this go away. Ftrace-based stats seem to roughly agree: Time kswapd awake 2532984 2326824 Time kcompactd awake 0 257916 Time direct compacting 864839 735130 Time kswapd compacting 0 0 Time kcompactd compacting 0 257585 Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:15 +00:00
if (cc->direct_compaction)
cc->zone->compact_blockskip_flush = true;
mm: compaction: clear PG_migrate_skip based on compaction and reclaim activity Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:47 +00:00
if (cc->whole_zone)
return COMPACT_COMPLETE;
else
return COMPACT_PARTIAL_SKIPPED;
mm: compaction: cache if a pageblock was scanned and no pages were isolated When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:41 +00:00
}
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
if (cc->proactive_compaction) {
int score, wmark_low;
pg_data_t *pgdat;
pgdat = cc->zone->zone_pgdat;
if (kswapd_is_running(pgdat))
return COMPACT_PARTIAL_SKIPPED;
score = fragmentation_score_zone(cc->zone);
wmark_low = fragmentation_score_wmark(pgdat, true);
if (score > wmark_low)
ret = COMPACT_CONTINUE;
else
ret = COMPACT_SUCCESS;
goto out;
}
if (is_via_compact_memory(cc->order))
return COMPACT_CONTINUE;
/*
* Always finish scanning a pageblock to reduce the possibility of
* fallbacks in the future. This is particularly important when
* migration source is unmovable/reclaimable but it's not worth
* special casing.
*/
if (!IS_ALIGNED(cc->migrate_pfn, pageblock_nr_pages))
return COMPACT_CONTINUE;
mm, compaction: finish whole pageblock to reduce fragmentation The main goal of direct compaction is to form a high-order page for allocation, but it should also help against long-term fragmentation when possible. Most lower-than-pageblock-order compactions are for non-movable allocations, which means that if we compact in a movable pageblock and terminate as soon as we create the high-order page, it's unlikely that the fallback heuristics will claim the whole block. Instead there might be a single unmovable page in a pageblock full of movable pages, and the next unmovable allocation might pick another pageblock and increase long-term fragmentation. To help against such scenarios, this patch changes the termination criteria for compaction so that the current pageblock is finished even though the high-order page already exists. Note that it might be possible that the high-order page formed elsewhere in the zone due to parallel activity, but this patch doesn't try to detect that. This is only done with sync compaction, because async compaction is limited to pageblock of the same migratetype, where it cannot result in a migratetype fallback. (Async compaction also eagerly skips order-aligned blocks where isolation fails, which is against the goal of migrating away as much of the pageblock as possible.) As a result of this patch, long-term memory fragmentation should be reduced. In testing based on 4.9 kernel with stress-highalloc from mmtests configured for order-4 GFP_KERNEL allocations, this patch has reduced the number of unmovable allocations falling back to movable pageblocks by 20%. The number Link: http://lkml.kernel.org/r/20170307131545.28577-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:54:52 +00:00
/* Direct compactor: Is a suitable page free? */
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
ret = COMPACT_NO_SUITABLE_PAGE;
mm: compaction: partially revert capture of suitable high-order page Eric Wong reported on 3.7 and 3.8-rc2 that ppoll() got stuck when waiting for POLLIN on a local TCP socket. It was easier to trigger if there was disk IO and dirty pages at the same time and he bisected it to commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). The intention of that patch was to improve high-order allocations under memory pressure after changes made to reclaim in 3.6 drastically hurt THP allocations but the approach was flawed. For Eric, the problem was that page->pfmemalloc was not being cleared for captured pages leading to a poor interaction with swap-over-NFS support causing the packets to be dropped. However, I identified a few more problems with the patch including the fact that it can increase contention on zone->lock in some cases which could result in async direct compaction being aborted early. In retrospect the capture patch took the wrong approach. What it should have done is mark the pageblock being migrated as MIGRATE_ISOLATE if it was allocating for THP and avoided races that way. While the patch was showing to improve allocation success rates at the time, the benefit is marginal given the relative complexity and it should be revisited from scratch in the context of the other reclaim-related changes that have taken place since the patch was first written and tested. This patch partially reverts commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). Reported-and-tested-by: Eric Wong <normalperson@yhbt.net> Tested-by: Eric Dumazet <eric.dumazet@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: David Miller <davem@davemloft.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-11 22:32:16 +00:00
for (order = cc->order; order < MAX_ORDER; order++) {
struct free_area *area = &cc->zone->free_area[order];
mm/compaction: enhance compaction finish condition Compaction has anti fragmentation algorithm. It is that freepage should be more than pageblock order to finish the compaction if we don't find any freepage in requested migratetype buddy list. This is for mitigating fragmentation, but, there is a lack of migratetype consideration and it is too excessive compared to page allocator's anti fragmentation algorithm. Not considering migratetype would cause premature finish of compaction. For example, if allocation request is for unmovable migratetype, freepage with CMA migratetype doesn't help that allocation and compaction should not be stopped. But, current logic regards this situation as compaction is no longer needed, so finish the compaction. Secondly, condition is too excessive compared to page allocator's logic. We can steal freepage from other migratetype and change pageblock migratetype on more relaxed conditions in page allocator. This is designed to prevent fragmentation and we can use it here. Imposing hard constraint only to the compaction doesn't help much in this case since page allocator would cause fragmentation again. To solve these problems, this patch borrows anti fragmentation logic from page allocator. It will reduce premature compaction finish in some cases and reduce excessive compaction work. stress-highalloc test in mmtests with non movable order 7 allocation shows considerable increase of compaction success rate. Compaction success rate (Compaction success * 100 / Compaction stalls, %) 31.82 : 42.20 I tested it on non-reboot 5 runs stress-highalloc benchmark and found that there is no more degradation on allocation success rate than before. That roughly means that this patch doesn't result in more fragmentations. Vlastimil suggests additional idea that we only test for fallbacks when migration scanner has scanned a whole pageblock. It looked good for fragmentation because chance of stealing increase due to making more free pages in certain pageblock. So, I tested it, but, it results in decreased compaction success rate, roughly 38.00. I guess the reason that if system is low memory condition, watermark check could be failed due to not enough order 0 free page and so, sometimes, we can't reach a fallback check although migrate_pfn is aligned to pageblock_nr_pages. I can insert code to cope with this situation but it makes code more complicated so I don't include his idea at this patch. [akpm@linux-foundation.org: fix CONFIG_CMA=n build] Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-14 22:45:21 +00:00
bool can_steal;
mm: compaction: partially revert capture of suitable high-order page Eric Wong reported on 3.7 and 3.8-rc2 that ppoll() got stuck when waiting for POLLIN on a local TCP socket. It was easier to trigger if there was disk IO and dirty pages at the same time and he bisected it to commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). The intention of that patch was to improve high-order allocations under memory pressure after changes made to reclaim in 3.6 drastically hurt THP allocations but the approach was flawed. For Eric, the problem was that page->pfmemalloc was not being cleared for captured pages leading to a poor interaction with swap-over-NFS support causing the packets to be dropped. However, I identified a few more problems with the patch including the fact that it can increase contention on zone->lock in some cases which could result in async direct compaction being aborted early. In retrospect the capture patch took the wrong approach. What it should have done is mark the pageblock being migrated as MIGRATE_ISOLATE if it was allocating for THP and avoided races that way. While the patch was showing to improve allocation success rates at the time, the benefit is marginal given the relative complexity and it should be revisited from scratch in the context of the other reclaim-related changes that have taken place since the patch was first written and tested. This patch partially reverts commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). Reported-and-tested-by: Eric Wong <normalperson@yhbt.net> Tested-by: Eric Dumazet <eric.dumazet@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: David Miller <davem@davemloft.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-11 22:32:16 +00:00
/* Job done if page is free of the right migratetype */
if (!free_area_empty(area, migratetype))
return COMPACT_SUCCESS;
mm: compaction: partially revert capture of suitable high-order page Eric Wong reported on 3.7 and 3.8-rc2 that ppoll() got stuck when waiting for POLLIN on a local TCP socket. It was easier to trigger if there was disk IO and dirty pages at the same time and he bisected it to commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). The intention of that patch was to improve high-order allocations under memory pressure after changes made to reclaim in 3.6 drastically hurt THP allocations but the approach was flawed. For Eric, the problem was that page->pfmemalloc was not being cleared for captured pages leading to a poor interaction with swap-over-NFS support causing the packets to be dropped. However, I identified a few more problems with the patch including the fact that it can increase contention on zone->lock in some cases which could result in async direct compaction being aborted early. In retrospect the capture patch took the wrong approach. What it should have done is mark the pageblock being migrated as MIGRATE_ISOLATE if it was allocating for THP and avoided races that way. While the patch was showing to improve allocation success rates at the time, the benefit is marginal given the relative complexity and it should be revisited from scratch in the context of the other reclaim-related changes that have taken place since the patch was first written and tested. This patch partially reverts commit 1fb3f8ca0e92 ("mm: compaction: capture a suitable high-order page immediately when it is made available"). Reported-and-tested-by: Eric Wong <normalperson@yhbt.net> Tested-by: Eric Dumazet <eric.dumazet@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: David Miller <davem@davemloft.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-11 22:32:16 +00:00
mm/compaction: enhance compaction finish condition Compaction has anti fragmentation algorithm. It is that freepage should be more than pageblock order to finish the compaction if we don't find any freepage in requested migratetype buddy list. This is for mitigating fragmentation, but, there is a lack of migratetype consideration and it is too excessive compared to page allocator's anti fragmentation algorithm. Not considering migratetype would cause premature finish of compaction. For example, if allocation request is for unmovable migratetype, freepage with CMA migratetype doesn't help that allocation and compaction should not be stopped. But, current logic regards this situation as compaction is no longer needed, so finish the compaction. Secondly, condition is too excessive compared to page allocator's logic. We can steal freepage from other migratetype and change pageblock migratetype on more relaxed conditions in page allocator. This is designed to prevent fragmentation and we can use it here. Imposing hard constraint only to the compaction doesn't help much in this case since page allocator would cause fragmentation again. To solve these problems, this patch borrows anti fragmentation logic from page allocator. It will reduce premature compaction finish in some cases and reduce excessive compaction work. stress-highalloc test in mmtests with non movable order 7 allocation shows considerable increase of compaction success rate. Compaction success rate (Compaction success * 100 / Compaction stalls, %) 31.82 : 42.20 I tested it on non-reboot 5 runs stress-highalloc benchmark and found that there is no more degradation on allocation success rate than before. That roughly means that this patch doesn't result in more fragmentations. Vlastimil suggests additional idea that we only test for fallbacks when migration scanner has scanned a whole pageblock. It looked good for fragmentation because chance of stealing increase due to making more free pages in certain pageblock. So, I tested it, but, it results in decreased compaction success rate, roughly 38.00. I guess the reason that if system is low memory condition, watermark check could be failed due to not enough order 0 free page and so, sometimes, we can't reach a fallback check although migrate_pfn is aligned to pageblock_nr_pages. I can insert code to cope with this situation but it makes code more complicated so I don't include his idea at this patch. [akpm@linux-foundation.org: fix CONFIG_CMA=n build] Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-14 22:45:21 +00:00
#ifdef CONFIG_CMA
/* MIGRATE_MOVABLE can fallback on MIGRATE_CMA */
if (migratetype == MIGRATE_MOVABLE &&
!free_area_empty(area, MIGRATE_CMA))
return COMPACT_SUCCESS;
mm/compaction: enhance compaction finish condition Compaction has anti fragmentation algorithm. It is that freepage should be more than pageblock order to finish the compaction if we don't find any freepage in requested migratetype buddy list. This is for mitigating fragmentation, but, there is a lack of migratetype consideration and it is too excessive compared to page allocator's anti fragmentation algorithm. Not considering migratetype would cause premature finish of compaction. For example, if allocation request is for unmovable migratetype, freepage with CMA migratetype doesn't help that allocation and compaction should not be stopped. But, current logic regards this situation as compaction is no longer needed, so finish the compaction. Secondly, condition is too excessive compared to page allocator's logic. We can steal freepage from other migratetype and change pageblock migratetype on more relaxed conditions in page allocator. This is designed to prevent fragmentation and we can use it here. Imposing hard constraint only to the compaction doesn't help much in this case since page allocator would cause fragmentation again. To solve these problems, this patch borrows anti fragmentation logic from page allocator. It will reduce premature compaction finish in some cases and reduce excessive compaction work. stress-highalloc test in mmtests with non movable order 7 allocation shows considerable increase of compaction success rate. Compaction success rate (Compaction success * 100 / Compaction stalls, %) 31.82 : 42.20 I tested it on non-reboot 5 runs stress-highalloc benchmark and found that there is no more degradation on allocation success rate than before. That roughly means that this patch doesn't result in more fragmentations. Vlastimil suggests additional idea that we only test for fallbacks when migration scanner has scanned a whole pageblock. It looked good for fragmentation because chance of stealing increase due to making more free pages in certain pageblock. So, I tested it, but, it results in decreased compaction success rate, roughly 38.00. I guess the reason that if system is low memory condition, watermark check could be failed due to not enough order 0 free page and so, sometimes, we can't reach a fallback check although migrate_pfn is aligned to pageblock_nr_pages. I can insert code to cope with this situation but it makes code more complicated so I don't include his idea at this patch. [akpm@linux-foundation.org: fix CONFIG_CMA=n build] Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-04-14 22:45:21 +00:00
#endif
/*
* Job done if allocation would steal freepages from
* other migratetype buddy lists.
*/
if (find_suitable_fallback(area, order, migratetype,
mm, compaction: finish whole pageblock to reduce fragmentation The main goal of direct compaction is to form a high-order page for allocation, but it should also help against long-term fragmentation when possible. Most lower-than-pageblock-order compactions are for non-movable allocations, which means that if we compact in a movable pageblock and terminate as soon as we create the high-order page, it's unlikely that the fallback heuristics will claim the whole block. Instead there might be a single unmovable page in a pageblock full of movable pages, and the next unmovable allocation might pick another pageblock and increase long-term fragmentation. To help against such scenarios, this patch changes the termination criteria for compaction so that the current pageblock is finished even though the high-order page already exists. Note that it might be possible that the high-order page formed elsewhere in the zone due to parallel activity, but this patch doesn't try to detect that. This is only done with sync compaction, because async compaction is limited to pageblock of the same migratetype, where it cannot result in a migratetype fallback. (Async compaction also eagerly skips order-aligned blocks where isolation fails, which is against the goal of migrating away as much of the pageblock as possible.) As a result of this patch, long-term memory fragmentation should be reduced. In testing based on 4.9 kernel with stress-highalloc from mmtests configured for order-4 GFP_KERNEL allocations, this patch has reduced the number of unmovable allocations falling back to movable pageblocks by 20%. The number Link: http://lkml.kernel.org/r/20170307131545.28577-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:54:52 +00:00
true, &can_steal) != -1) {
/* movable pages are OK in any pageblock */
if (migratetype == MIGRATE_MOVABLE)
return COMPACT_SUCCESS;
/*
* We are stealing for a non-movable allocation. Make
* sure we finish compacting the current pageblock
* first so it is as free as possible and we won't
* have to steal another one soon. This only applies
* to sync compaction, as async compaction operates
* on pageblocks of the same migratetype.
*/
if (cc->mode == MIGRATE_ASYNC ||
IS_ALIGNED(cc->migrate_pfn,
pageblock_nr_pages)) {
return COMPACT_SUCCESS;
}
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
ret = COMPACT_CONTINUE;
break;
mm, compaction: finish whole pageblock to reduce fragmentation The main goal of direct compaction is to form a high-order page for allocation, but it should also help against long-term fragmentation when possible. Most lower-than-pageblock-order compactions are for non-movable allocations, which means that if we compact in a movable pageblock and terminate as soon as we create the high-order page, it's unlikely that the fallback heuristics will claim the whole block. Instead there might be a single unmovable page in a pageblock full of movable pages, and the next unmovable allocation might pick another pageblock and increase long-term fragmentation. To help against such scenarios, this patch changes the termination criteria for compaction so that the current pageblock is finished even though the high-order page already exists. Note that it might be possible that the high-order page formed elsewhere in the zone due to parallel activity, but this patch doesn't try to detect that. This is only done with sync compaction, because async compaction is limited to pageblock of the same migratetype, where it cannot result in a migratetype fallback. (Async compaction also eagerly skips order-aligned blocks where isolation fails, which is against the goal of migrating away as much of the pageblock as possible.) As a result of this patch, long-term memory fragmentation should be reduced. In testing based on 4.9 kernel with stress-highalloc from mmtests configured for order-4 GFP_KERNEL allocations, this patch has reduced the number of unmovable allocations falling back to movable pageblocks by 20%. The number Link: http://lkml.kernel.org/r/20170307131545.28577-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:54:52 +00:00
}
}
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
out:
mm, compaction: finish pageblock scanning on contention Async migration aborts on spinlock contention but contention can be high when there are multiple compaction attempts and kswapd is active. The consequence is that the migration scanners move forward uselessly while still contending on locks for longer while leaving suitable migration sources behind. This patch will acquire the lock but track when contention occurs. When it does, the current pageblock will finish as compaction may succeed for that block and then abort. This will have a variable impact on latency as in some cases useless scanning is avoided (reduces latency) but a lock will be contended (increase latency) or a single contended pageblock is scanned that would otherwise have been skipped (increase latency). 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3002.07 ( 0.00%) 3153.17 ( -5.03%) Amean fault-both-5 4684.47 ( 0.00%) 4280.52 ( 8.62%) Amean fault-both-7 6815.54 ( 0.00%) 5811.50 * 14.73%* Amean fault-both-12 10864.02 ( 0.00%) 9276.85 ( 14.61%) Amean fault-both-18 12247.52 ( 0.00%) 11032.67 ( 9.92%) Amean fault-both-24 15683.99 ( 0.00%) 14285.70 ( 8.92%) Amean fault-both-30 18620.02 ( 0.00%) 16293.76 * 12.49%* Amean fault-both-32 19250.28 ( 0.00%) 16721.02 * 13.14%* 5.0.0-rc1 5.0.0-rc1 norescan-v3r16 finishcontend-v3r16 Percentage huge-1 0.00 ( 0.00%) 0.00 ( 0.00%) Percentage huge-3 95.00 ( 0.00%) 96.82 ( 1.92%) Percentage huge-5 94.22 ( 0.00%) 95.40 ( 1.26%) Percentage huge-7 92.35 ( 0.00%) 95.92 ( 3.86%) Percentage huge-12 91.90 ( 0.00%) 96.73 ( 5.25%) Percentage huge-18 89.58 ( 0.00%) 96.77 ( 8.03%) Percentage huge-24 90.03 ( 0.00%) 96.05 ( 6.69%) Percentage huge-30 89.14 ( 0.00%) 96.81 ( 8.60%) Percentage huge-32 90.58 ( 0.00%) 97.41 ( 7.54%) There is a variable impact that is mostly good on latency while allocation success rates are slightly higher. System CPU usage is reduced by about 10% but scan rate impact is mixed Compaction migrate scanned 27997659.00 20148867 Compaction free scanned 120782791.00 118324914 Migration scan rates are reduced 28% which is expected as a pageblock is used by the async scanner instead of skipped. The impact on the free scanner is known to be variable. Overall the primary justification for this patch is that completing scanning of a pageblock is very important for later patches. [yuehaibing@huawei.com: fix unused variable warning] Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: YueHaibing <yuehaibing@huawei.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:11 +00:00
if (cc->contended || fatal_signal_pending(current))
ret = COMPACT_CONTENDED;
return ret;
}
static enum compact_result compact_finished(struct compact_control *cc)
{
int ret;
ret = __compact_finished(cc);
trace_mm_compaction_finished(cc->zone, cc->order, ret);
if (ret == COMPACT_NO_SUITABLE_PAGE)
ret = COMPACT_CONTINUE;
return ret;
}
static enum compact_result __compaction_suitable(struct zone *zone, int order,
unsigned int alloc_flags,
int highest_zoneidx,
mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders "mm: consider compaction feedback also for costly allocation" has removed the upper bound for the reclaim/compaction retries based on the number of reclaimed pages for costly orders. While this is desirable the patch did miss a mis interaction between reclaim, compaction and the retry logic. The direct reclaim tries to get zones over min watermark while compaction backs off and returns COMPACT_SKIPPED when all zones are below low watermark + 1<<order gap. If we are getting really close to OOM then __compaction_suitable can keep returning COMPACT_SKIPPED a high order request (e.g. hugetlb order-9) while the reclaim is not able to release enough pages to get us over low watermark. The reclaim is still able to make some progress (usually trashing over few remaining pages) so we are not able to break out from the loop. I have seen this happening with the same test described in "mm: consider compaction feedback also for costly allocation" on a swapless system. The original problem got resolved by "vmscan: consider classzone_idx in compaction_ready" but it shows how things might go wrong when we approach the oom event horizont. The reason why compaction requires being over low rather than min watermark is not clear to me. This check was there essentially since 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). It is clearly an implementation detail though and we shouldn't pull it into the generic retry logic while we should be able to cope with such eventuality. The only place in should_compact_retry where we retry without any upper bound is for compaction_withdrawn() case. Introduce compaction_zonelist_suitable function which checks the given zonelist and returns true only if there is at least one zone which would would unblock __compaction_suitable if more memory got reclaimed. In this implementation it checks __compaction_suitable with NR_FREE_PAGES plus part of the reclaimable memory as the target for the watermark check. The reclaimable memory is reduced linearly by the allocation order. The idea is that we do not want to reclaim all the remaining memory for a single allocation request just unblock __compaction_suitable which doesn't guarantee we will make a further progress. The new helper is then used if compaction_withdrawn() feedback was provided so we do not retry if there is no outlook for a further progress. !costly requests shouldn't be affected much - e.g. order-2 pages would require to have at least 64kB on the reclaimable LRUs while order-9 would need at least 32M which should be enough to not lock up. [vbabka@suse.cz: fix classzone_idx vs. high_zoneidx usage in compaction_zonelist_suitable] [akpm@linux-foundation.org: fix it for Mel's mm-page_alloc-remove-field-from-alloc_context.patch] Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Vladimir Davydov <vdavydov@virtuozzo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 23:57:12 +00:00
unsigned long wmark_target)
{
unsigned long watermark;
if (is_via_compact_memory(order))
return COMPACT_CONTINUE;
watermark = wmark_pages(zone, alloc_flags & ALLOC_WMARK_MASK);
mm, compaction: pass classzone_idx and alloc_flags to watermark checking Compaction relies on zone watermark checks for decisions such as if it's worth to start compacting in compaction_suitable() or whether compaction should stop in compact_finished(). The watermark checks take classzone_idx and alloc_flags parameters, which are related to the memory allocation request. But from the context of compaction they are currently passed as 0, including the direct compaction which is invoked to satisfy the allocation request, and could therefore know the proper values. The lack of proper values can lead to mismatch between decisions taken during compaction and decisions related to the allocation request. Lack of proper classzone_idx value means that lowmem_reserve is not taken into account. This has manifested (during recent changes to deferred compaction) when DMA zone was used as fallback for preferred Normal zone. compaction_suitable() without proper classzone_idx would think that the watermarks are already satisfied, but watermark check in get_page_from_freelist() would fail. Because of this problem, deferring compaction has extra complexity that can be removed in the following patch. The issue (not confirmed in practice) with missing alloc_flags is opposite in nature. For allocations that include ALLOC_HIGH, ALLOC_HIGHER or ALLOC_CMA in alloc_flags (the last includes all MOVABLE allocations on CMA-enabled systems) the watermark checking in compaction with 0 passed will be stricter than in get_page_from_freelist(). In these cases compaction might be running for a longer time than is really needed. Another issue compaction_suitable() is that the check for "does the zone need compaction at all?" comes only after the check "does the zone have enough free free pages to succeed compaction". The latter considers extra pages for migration and can therefore in some situations fail and return COMPACT_SKIPPED, although the high-order allocation would succeed and we should return COMPACT_PARTIAL. This patch fixes these problems by adding alloc_flags and classzone_idx to struct compact_control and related functions involved in direct compaction and watermark checking. Where possible, all other callers of compaction_suitable() pass proper values where those are known. This is currently limited to classzone_idx, which is sometimes known in kswapd context. However, the direct reclaim callers should_continue_reclaim() and compaction_ready() do not currently know the proper values, so the coordination between reclaim and compaction may still not be as accurate as it could. This can be fixed later, if it's shown to be an issue. Additionaly the checks in compact_suitable() are reordered to address the second issue described above. The effect of this patch should be slightly better high-order allocation success rates and/or less compaction overhead, depending on the type of allocations and presence of CMA. It allows simplifying deferred compaction code in a followup patch. When testing with stress-highalloc, there was some slight improvement (which might be just due to variance) in success rates of non-THP-like allocations. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:22 +00:00
/*
* If watermarks for high-order allocation are already met, there
* should be no need for compaction at all.
*/
if (zone_watermark_ok(zone, order, watermark, highest_zoneidx,
mm, compaction: pass classzone_idx and alloc_flags to watermark checking Compaction relies on zone watermark checks for decisions such as if it's worth to start compacting in compaction_suitable() or whether compaction should stop in compact_finished(). The watermark checks take classzone_idx and alloc_flags parameters, which are related to the memory allocation request. But from the context of compaction they are currently passed as 0, including the direct compaction which is invoked to satisfy the allocation request, and could therefore know the proper values. The lack of proper values can lead to mismatch between decisions taken during compaction and decisions related to the allocation request. Lack of proper classzone_idx value means that lowmem_reserve is not taken into account. This has manifested (during recent changes to deferred compaction) when DMA zone was used as fallback for preferred Normal zone. compaction_suitable() without proper classzone_idx would think that the watermarks are already satisfied, but watermark check in get_page_from_freelist() would fail. Because of this problem, deferring compaction has extra complexity that can be removed in the following patch. The issue (not confirmed in practice) with missing alloc_flags is opposite in nature. For allocations that include ALLOC_HIGH, ALLOC_HIGHER or ALLOC_CMA in alloc_flags (the last includes all MOVABLE allocations on CMA-enabled systems) the watermark checking in compaction with 0 passed will be stricter than in get_page_from_freelist(). In these cases compaction might be running for a longer time than is really needed. Another issue compaction_suitable() is that the check for "does the zone need compaction at all?" comes only after the check "does the zone have enough free free pages to succeed compaction". The latter considers extra pages for migration and can therefore in some situations fail and return COMPACT_SKIPPED, although the high-order allocation would succeed and we should return COMPACT_PARTIAL. This patch fixes these problems by adding alloc_flags and classzone_idx to struct compact_control and related functions involved in direct compaction and watermark checking. Where possible, all other callers of compaction_suitable() pass proper values where those are known. This is currently limited to classzone_idx, which is sometimes known in kswapd context. However, the direct reclaim callers should_continue_reclaim() and compaction_ready() do not currently know the proper values, so the coordination between reclaim and compaction may still not be as accurate as it could. This can be fixed later, if it's shown to be an issue. Additionaly the checks in compact_suitable() are reordered to address the second issue described above. The effect of this patch should be slightly better high-order allocation success rates and/or less compaction overhead, depending on the type of allocations and presence of CMA. It allows simplifying deferred compaction code in a followup patch. When testing with stress-highalloc, there was some slight improvement (which might be just due to variance) in success rates of non-THP-like allocations. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:22 +00:00
alloc_flags))
return COMPACT_SUCCESS;
mm, compaction: pass classzone_idx and alloc_flags to watermark checking Compaction relies on zone watermark checks for decisions such as if it's worth to start compacting in compaction_suitable() or whether compaction should stop in compact_finished(). The watermark checks take classzone_idx and alloc_flags parameters, which are related to the memory allocation request. But from the context of compaction they are currently passed as 0, including the direct compaction which is invoked to satisfy the allocation request, and could therefore know the proper values. The lack of proper values can lead to mismatch between decisions taken during compaction and decisions related to the allocation request. Lack of proper classzone_idx value means that lowmem_reserve is not taken into account. This has manifested (during recent changes to deferred compaction) when DMA zone was used as fallback for preferred Normal zone. compaction_suitable() without proper classzone_idx would think that the watermarks are already satisfied, but watermark check in get_page_from_freelist() would fail. Because of this problem, deferring compaction has extra complexity that can be removed in the following patch. The issue (not confirmed in practice) with missing alloc_flags is opposite in nature. For allocations that include ALLOC_HIGH, ALLOC_HIGHER or ALLOC_CMA in alloc_flags (the last includes all MOVABLE allocations on CMA-enabled systems) the watermark checking in compaction with 0 passed will be stricter than in get_page_from_freelist(). In these cases compaction might be running for a longer time than is really needed. Another issue compaction_suitable() is that the check for "does the zone need compaction at all?" comes only after the check "does the zone have enough free free pages to succeed compaction". The latter considers extra pages for migration and can therefore in some situations fail and return COMPACT_SKIPPED, although the high-order allocation would succeed and we should return COMPACT_PARTIAL. This patch fixes these problems by adding alloc_flags and classzone_idx to struct compact_control and related functions involved in direct compaction and watermark checking. Where possible, all other callers of compaction_suitable() pass proper values where those are known. This is currently limited to classzone_idx, which is sometimes known in kswapd context. However, the direct reclaim callers should_continue_reclaim() and compaction_ready() do not currently know the proper values, so the coordination between reclaim and compaction may still not be as accurate as it could. This can be fixed later, if it's shown to be an issue. Additionaly the checks in compact_suitable() are reordered to address the second issue described above. The effect of this patch should be slightly better high-order allocation success rates and/or less compaction overhead, depending on the type of allocations and presence of CMA. It allows simplifying deferred compaction code in a followup patch. When testing with stress-highalloc, there was some slight improvement (which might be just due to variance) in success rates of non-THP-like allocations. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:22 +00:00
/*
* Watermarks for order-0 must be met for compaction to be able to
mm, compaction: use proper alloc_flags in __compaction_suitable() The __compaction_suitable() function checks the low watermark plus a compact_gap() gap to decide if there's enough free memory to perform compaction. This check uses direct compactor's alloc_flags, but that's wrong, since these flags are not applicable for freepage isolation. For example, alloc_flags may indicate access to memory reserves, making compaction proceed, and then fail watermark check during the isolation. A similar problem exists for ALLOC_CMA, which may be part of alloc_flags, but not during freepage isolation. In this case however it makes sense to use ALLOC_CMA both in __compaction_suitable() and __isolate_free_page(), since there's actually nothing preventing the freepage scanner to isolate from CMA pageblocks, with the assumption that a page that could be migrated once by compaction can be migrated also later by CMA allocation. Thus we should count pages in CMA pageblocks when considering compaction suitability and when isolating freepages. To sum up, this patch should remove some false positives from __compaction_suitable(), and allow compaction to proceed when free pages required for compaction reside in the CMA pageblocks. Link: http://lkml.kernel.org/r/20160810091226.6709-10-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:57 +00:00
* isolate free pages for migration targets. This means that the
* watermark and alloc_flags have to match, or be more pessimistic than
* the check in __isolate_free_page(). We don't use the direct
* compactor's alloc_flags, as they are not relevant for freepage
* isolation. We however do use the direct compactor's highest_zoneidx
* to skip over zones where lowmem reserves would prevent allocation
* even if compaction succeeds.
mm, compaction: require only min watermarks for non-costly orders The __compaction_suitable() function checks the low watermark plus a compact_gap() gap to decide if there's enough free memory to perform compaction. Then __isolate_free_page uses low watermark check to decide if particular free page can be isolated. In the latter case, using low watermark is needlessly pessimistic, as the free page isolations are only temporary. For __compaction_suitable() the higher watermark makes sense for high-order allocations where more freepages increase the chance of success, and we can typically fail with some order-0 fallback when the system is struggling to reach that watermark. But for low-order allocation, forming the page should not be that hard. So using low watermark here might just prevent compaction from even trying, and eventually lead to OOM killer even if we are above min watermarks. So after this patch, we use min watermark for non-costly orders in __compaction_suitable(), and for all orders in __isolate_free_page(). [vbabka@suse.cz: clarify __isolate_free_page() comment] Link: http://lkml.kernel.org/r/7ae4baec-4eca-e70b-2a69-94bea4fb19fa@suse.cz Link: http://lkml.kernel.org/r/20160810091226.6709-11-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:58:00 +00:00
* For costly orders, we require low watermark instead of min for
* compaction to proceed to increase its chances.
Revert "mm/cma: manage the memory of the CMA area by using the ZONE_MOVABLE" This reverts the following commits that change CMA design in MM. 3d2054ad8c2d ("ARM: CMA: avoid double mapping to the CMA area if CONFIG_HIGHMEM=y") 1d47a3ec09b5 ("mm/cma: remove ALLOC_CMA") bad8c6c0b114 ("mm/cma: manage the memory of the CMA area by using the ZONE_MOVABLE") Ville reported a following error on i386. Inode-cache hash table entries: 65536 (order: 6, 262144 bytes) microcode: microcode updated early to revision 0x4, date = 2013-06-28 Initializing CPU#0 Initializing HighMem for node 0 (000377fe:00118000) Initializing Movable for node 0 (00000001:00118000) BUG: Bad page state in process swapper pfn:377fe page:f53effc0 count:0 mapcount:-127 mapping:00000000 index:0x0 flags: 0x80000000() raw: 80000000 00000000 00000000 ffffff80 00000000 00000100 00000200 00000001 page dumped because: nonzero mapcount Modules linked in: CPU: 0 PID: 0 Comm: swapper Not tainted 4.17.0-rc5-elk+ #145 Hardware name: Dell Inc. Latitude E5410/03VXMC, BIOS A15 07/11/2013 Call Trace: dump_stack+0x60/0x96 bad_page+0x9a/0x100 free_pages_check_bad+0x3f/0x60 free_pcppages_bulk+0x29d/0x5b0 free_unref_page_commit+0x84/0xb0 free_unref_page+0x3e/0x70 __free_pages+0x1d/0x20 free_highmem_page+0x19/0x40 add_highpages_with_active_regions+0xab/0xeb set_highmem_pages_init+0x66/0x73 mem_init+0x1b/0x1d7 start_kernel+0x17a/0x363 i386_start_kernel+0x95/0x99 startup_32_smp+0x164/0x168 The reason for this error is that the span of MOVABLE_ZONE is extended to whole node span for future CMA initialization, and, normal memory is wrongly freed here. I submitted the fix and it seems to work, but, another problem happened. It's so late time to fix the later problem so I decide to reverting the series. Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Acked-by: Laura Abbott <labbott@redhat.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-05-23 01:18:21 +00:00
* ALLOC_CMA is used, as pages in CMA pageblocks are considered
* suitable migration targets
*/
mm, compaction: require only min watermarks for non-costly orders The __compaction_suitable() function checks the low watermark plus a compact_gap() gap to decide if there's enough free memory to perform compaction. Then __isolate_free_page uses low watermark check to decide if particular free page can be isolated. In the latter case, using low watermark is needlessly pessimistic, as the free page isolations are only temporary. For __compaction_suitable() the higher watermark makes sense for high-order allocations where more freepages increase the chance of success, and we can typically fail with some order-0 fallback when the system is struggling to reach that watermark. But for low-order allocation, forming the page should not be that hard. So using low watermark here might just prevent compaction from even trying, and eventually lead to OOM killer even if we are above min watermarks. So after this patch, we use min watermark for non-costly orders in __compaction_suitable(), and for all orders in __isolate_free_page(). [vbabka@suse.cz: clarify __isolate_free_page() comment] Link: http://lkml.kernel.org/r/7ae4baec-4eca-e70b-2a69-94bea4fb19fa@suse.cz Link: http://lkml.kernel.org/r/20160810091226.6709-11-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:58:00 +00:00
watermark = (order > PAGE_ALLOC_COSTLY_ORDER) ?
low_wmark_pages(zone) : min_wmark_pages(zone);
watermark += compact_gap(order);
if (!__zone_watermark_ok(zone, 0, watermark, highest_zoneidx,
Revert "mm/cma: manage the memory of the CMA area by using the ZONE_MOVABLE" This reverts the following commits that change CMA design in MM. 3d2054ad8c2d ("ARM: CMA: avoid double mapping to the CMA area if CONFIG_HIGHMEM=y") 1d47a3ec09b5 ("mm/cma: remove ALLOC_CMA") bad8c6c0b114 ("mm/cma: manage the memory of the CMA area by using the ZONE_MOVABLE") Ville reported a following error on i386. Inode-cache hash table entries: 65536 (order: 6, 262144 bytes) microcode: microcode updated early to revision 0x4, date = 2013-06-28 Initializing CPU#0 Initializing HighMem for node 0 (000377fe:00118000) Initializing Movable for node 0 (00000001:00118000) BUG: Bad page state in process swapper pfn:377fe page:f53effc0 count:0 mapcount:-127 mapping:00000000 index:0x0 flags: 0x80000000() raw: 80000000 00000000 00000000 ffffff80 00000000 00000100 00000200 00000001 page dumped because: nonzero mapcount Modules linked in: CPU: 0 PID: 0 Comm: swapper Not tainted 4.17.0-rc5-elk+ #145 Hardware name: Dell Inc. Latitude E5410/03VXMC, BIOS A15 07/11/2013 Call Trace: dump_stack+0x60/0x96 bad_page+0x9a/0x100 free_pages_check_bad+0x3f/0x60 free_pcppages_bulk+0x29d/0x5b0 free_unref_page_commit+0x84/0xb0 free_unref_page+0x3e/0x70 __free_pages+0x1d/0x20 free_highmem_page+0x19/0x40 add_highpages_with_active_regions+0xab/0xeb set_highmem_pages_init+0x66/0x73 mem_init+0x1b/0x1d7 start_kernel+0x17a/0x363 i386_start_kernel+0x95/0x99 startup_32_smp+0x164/0x168 The reason for this error is that the span of MOVABLE_ZONE is extended to whole node span for future CMA initialization, and, normal memory is wrongly freed here. I submitted the fix and it seems to work, but, another problem happened. It's so late time to fix the later problem so I decide to reverting the series. Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Acked-by: Laura Abbott <labbott@redhat.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-05-23 01:18:21 +00:00
ALLOC_CMA, wmark_target))
return COMPACT_SKIPPED;
return COMPACT_CONTINUE;
}
/*
* compaction_suitable: Is this suitable to run compaction on this zone now?
* Returns
* COMPACT_SKIPPED - If there are too few free pages for compaction
* COMPACT_SUCCESS - If the allocation would succeed without compaction
* COMPACT_CONTINUE - If compaction should run now
*/
enum compact_result compaction_suitable(struct zone *zone, int order,
unsigned int alloc_flags,
int highest_zoneidx)
{
enum compact_result ret;
int fragindex;
ret = __compaction_suitable(zone, order, alloc_flags, highest_zoneidx,
zone_page_state(zone, NR_FREE_PAGES));
/*
* fragmentation index determines if allocation failures are due to
* low memory or external fragmentation
*
mm, compaction: pass classzone_idx and alloc_flags to watermark checking Compaction relies on zone watermark checks for decisions such as if it's worth to start compacting in compaction_suitable() or whether compaction should stop in compact_finished(). The watermark checks take classzone_idx and alloc_flags parameters, which are related to the memory allocation request. But from the context of compaction they are currently passed as 0, including the direct compaction which is invoked to satisfy the allocation request, and could therefore know the proper values. The lack of proper values can lead to mismatch between decisions taken during compaction and decisions related to the allocation request. Lack of proper classzone_idx value means that lowmem_reserve is not taken into account. This has manifested (during recent changes to deferred compaction) when DMA zone was used as fallback for preferred Normal zone. compaction_suitable() without proper classzone_idx would think that the watermarks are already satisfied, but watermark check in get_page_from_freelist() would fail. Because of this problem, deferring compaction has extra complexity that can be removed in the following patch. The issue (not confirmed in practice) with missing alloc_flags is opposite in nature. For allocations that include ALLOC_HIGH, ALLOC_HIGHER or ALLOC_CMA in alloc_flags (the last includes all MOVABLE allocations on CMA-enabled systems) the watermark checking in compaction with 0 passed will be stricter than in get_page_from_freelist(). In these cases compaction might be running for a longer time than is really needed. Another issue compaction_suitable() is that the check for "does the zone need compaction at all?" comes only after the check "does the zone have enough free free pages to succeed compaction". The latter considers extra pages for migration and can therefore in some situations fail and return COMPACT_SKIPPED, although the high-order allocation would succeed and we should return COMPACT_PARTIAL. This patch fixes these problems by adding alloc_flags and classzone_idx to struct compact_control and related functions involved in direct compaction and watermark checking. Where possible, all other callers of compaction_suitable() pass proper values where those are known. This is currently limited to classzone_idx, which is sometimes known in kswapd context. However, the direct reclaim callers should_continue_reclaim() and compaction_ready() do not currently know the proper values, so the coordination between reclaim and compaction may still not be as accurate as it could. This can be fixed later, if it's shown to be an issue. Additionaly the checks in compact_suitable() are reordered to address the second issue described above. The effect of this patch should be slightly better high-order allocation success rates and/or less compaction overhead, depending on the type of allocations and presence of CMA. It allows simplifying deferred compaction code in a followup patch. When testing with stress-highalloc, there was some slight improvement (which might be just due to variance) in success rates of non-THP-like allocations. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:22 +00:00
* index of -1000 would imply allocations might succeed depending on
* watermarks, but we already failed the high-order watermark check
* index towards 0 implies failure is due to lack of memory
* index towards 1000 implies failure is due to fragmentation
*
* Only compact if a failure would be due to fragmentation. Also
* ignore fragindex for non-costly orders where the alternative to
* a successful reclaim/compaction is OOM. Fragindex and the
* vm.extfrag_threshold sysctl is meant as a heuristic to prevent
* excessive compaction for costly orders, but it should not be at the
* expense of system stability.
*/
if (ret == COMPACT_CONTINUE && (order > PAGE_ALLOC_COSTLY_ORDER)) {
fragindex = fragmentation_index(zone, order);
if (fragindex >= 0 && fragindex <= sysctl_extfrag_threshold)
ret = COMPACT_NOT_SUITABLE_ZONE;
}
trace_mm_compaction_suitable(zone, order, ret);
if (ret == COMPACT_NOT_SUITABLE_ZONE)
ret = COMPACT_SKIPPED;
return ret;
}
mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders "mm: consider compaction feedback also for costly allocation" has removed the upper bound for the reclaim/compaction retries based on the number of reclaimed pages for costly orders. While this is desirable the patch did miss a mis interaction between reclaim, compaction and the retry logic. The direct reclaim tries to get zones over min watermark while compaction backs off and returns COMPACT_SKIPPED when all zones are below low watermark + 1<<order gap. If we are getting really close to OOM then __compaction_suitable can keep returning COMPACT_SKIPPED a high order request (e.g. hugetlb order-9) while the reclaim is not able to release enough pages to get us over low watermark. The reclaim is still able to make some progress (usually trashing over few remaining pages) so we are not able to break out from the loop. I have seen this happening with the same test described in "mm: consider compaction feedback also for costly allocation" on a swapless system. The original problem got resolved by "vmscan: consider classzone_idx in compaction_ready" but it shows how things might go wrong when we approach the oom event horizont. The reason why compaction requires being over low rather than min watermark is not clear to me. This check was there essentially since 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). It is clearly an implementation detail though and we shouldn't pull it into the generic retry logic while we should be able to cope with such eventuality. The only place in should_compact_retry where we retry without any upper bound is for compaction_withdrawn() case. Introduce compaction_zonelist_suitable function which checks the given zonelist and returns true only if there is at least one zone which would would unblock __compaction_suitable if more memory got reclaimed. In this implementation it checks __compaction_suitable with NR_FREE_PAGES plus part of the reclaimable memory as the target for the watermark check. The reclaimable memory is reduced linearly by the allocation order. The idea is that we do not want to reclaim all the remaining memory for a single allocation request just unblock __compaction_suitable which doesn't guarantee we will make a further progress. The new helper is then used if compaction_withdrawn() feedback was provided so we do not retry if there is no outlook for a further progress. !costly requests shouldn't be affected much - e.g. order-2 pages would require to have at least 64kB on the reclaimable LRUs while order-9 would need at least 32M which should be enough to not lock up. [vbabka@suse.cz: fix classzone_idx vs. high_zoneidx usage in compaction_zonelist_suitable] [akpm@linux-foundation.org: fix it for Mel's mm-page_alloc-remove-field-from-alloc_context.patch] Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Vladimir Davydov <vdavydov@virtuozzo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 23:57:12 +00:00
bool compaction_zonelist_suitable(struct alloc_context *ac, int order,
int alloc_flags)
{
struct zone *zone;
struct zoneref *z;
/*
* Make sure at least one zone would pass __compaction_suitable if we continue
* retrying the reclaim.
*/
for_each_zone_zonelist_nodemask(zone, z, ac->zonelist,
ac->highest_zoneidx, ac->nodemask) {
mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders "mm: consider compaction feedback also for costly allocation" has removed the upper bound for the reclaim/compaction retries based on the number of reclaimed pages for costly orders. While this is desirable the patch did miss a mis interaction between reclaim, compaction and the retry logic. The direct reclaim tries to get zones over min watermark while compaction backs off and returns COMPACT_SKIPPED when all zones are below low watermark + 1<<order gap. If we are getting really close to OOM then __compaction_suitable can keep returning COMPACT_SKIPPED a high order request (e.g. hugetlb order-9) while the reclaim is not able to release enough pages to get us over low watermark. The reclaim is still able to make some progress (usually trashing over few remaining pages) so we are not able to break out from the loop. I have seen this happening with the same test described in "mm: consider compaction feedback also for costly allocation" on a swapless system. The original problem got resolved by "vmscan: consider classzone_idx in compaction_ready" but it shows how things might go wrong when we approach the oom event horizont. The reason why compaction requires being over low rather than min watermark is not clear to me. This check was there essentially since 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). It is clearly an implementation detail though and we shouldn't pull it into the generic retry logic while we should be able to cope with such eventuality. The only place in should_compact_retry where we retry without any upper bound is for compaction_withdrawn() case. Introduce compaction_zonelist_suitable function which checks the given zonelist and returns true only if there is at least one zone which would would unblock __compaction_suitable if more memory got reclaimed. In this implementation it checks __compaction_suitable with NR_FREE_PAGES plus part of the reclaimable memory as the target for the watermark check. The reclaimable memory is reduced linearly by the allocation order. The idea is that we do not want to reclaim all the remaining memory for a single allocation request just unblock __compaction_suitable which doesn't guarantee we will make a further progress. The new helper is then used if compaction_withdrawn() feedback was provided so we do not retry if there is no outlook for a further progress. !costly requests shouldn't be affected much - e.g. order-2 pages would require to have at least 64kB on the reclaimable LRUs while order-9 would need at least 32M which should be enough to not lock up. [vbabka@suse.cz: fix classzone_idx vs. high_zoneidx usage in compaction_zonelist_suitable] [akpm@linux-foundation.org: fix it for Mel's mm-page_alloc-remove-field-from-alloc_context.patch] Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Vladimir Davydov <vdavydov@virtuozzo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 23:57:12 +00:00
unsigned long available;
enum compact_result compact_result;
/*
* Do not consider all the reclaimable memory because we do not
* want to trash just for a single high order allocation which
* is even not guaranteed to appear even if __compaction_suitable
* is happy about the watermark check.
*/
available = zone_reclaimable_pages(zone) / order;
mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders "mm: consider compaction feedback also for costly allocation" has removed the upper bound for the reclaim/compaction retries based on the number of reclaimed pages for costly orders. While this is desirable the patch did miss a mis interaction between reclaim, compaction and the retry logic. The direct reclaim tries to get zones over min watermark while compaction backs off and returns COMPACT_SKIPPED when all zones are below low watermark + 1<<order gap. If we are getting really close to OOM then __compaction_suitable can keep returning COMPACT_SKIPPED a high order request (e.g. hugetlb order-9) while the reclaim is not able to release enough pages to get us over low watermark. The reclaim is still able to make some progress (usually trashing over few remaining pages) so we are not able to break out from the loop. I have seen this happening with the same test described in "mm: consider compaction feedback also for costly allocation" on a swapless system. The original problem got resolved by "vmscan: consider classzone_idx in compaction_ready" but it shows how things might go wrong when we approach the oom event horizont. The reason why compaction requires being over low rather than min watermark is not clear to me. This check was there essentially since 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). It is clearly an implementation detail though and we shouldn't pull it into the generic retry logic while we should be able to cope with such eventuality. The only place in should_compact_retry where we retry without any upper bound is for compaction_withdrawn() case. Introduce compaction_zonelist_suitable function which checks the given zonelist and returns true only if there is at least one zone which would would unblock __compaction_suitable if more memory got reclaimed. In this implementation it checks __compaction_suitable with NR_FREE_PAGES plus part of the reclaimable memory as the target for the watermark check. The reclaimable memory is reduced linearly by the allocation order. The idea is that we do not want to reclaim all the remaining memory for a single allocation request just unblock __compaction_suitable which doesn't guarantee we will make a further progress. The new helper is then used if compaction_withdrawn() feedback was provided so we do not retry if there is no outlook for a further progress. !costly requests shouldn't be affected much - e.g. order-2 pages would require to have at least 64kB on the reclaimable LRUs while order-9 would need at least 32M which should be enough to not lock up. [vbabka@suse.cz: fix classzone_idx vs. high_zoneidx usage in compaction_zonelist_suitable] [akpm@linux-foundation.org: fix it for Mel's mm-page_alloc-remove-field-from-alloc_context.patch] Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Vladimir Davydov <vdavydov@virtuozzo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 23:57:12 +00:00
available += zone_page_state_snapshot(zone, NR_FREE_PAGES);
compact_result = __compaction_suitable(zone, order, alloc_flags,
ac->highest_zoneidx, available);
if (compact_result != COMPACT_SKIPPED)
mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders "mm: consider compaction feedback also for costly allocation" has removed the upper bound for the reclaim/compaction retries based on the number of reclaimed pages for costly orders. While this is desirable the patch did miss a mis interaction between reclaim, compaction and the retry logic. The direct reclaim tries to get zones over min watermark while compaction backs off and returns COMPACT_SKIPPED when all zones are below low watermark + 1<<order gap. If we are getting really close to OOM then __compaction_suitable can keep returning COMPACT_SKIPPED a high order request (e.g. hugetlb order-9) while the reclaim is not able to release enough pages to get us over low watermark. The reclaim is still able to make some progress (usually trashing over few remaining pages) so we are not able to break out from the loop. I have seen this happening with the same test described in "mm: consider compaction feedback also for costly allocation" on a swapless system. The original problem got resolved by "vmscan: consider classzone_idx in compaction_ready" but it shows how things might go wrong when we approach the oom event horizont. The reason why compaction requires being over low rather than min watermark is not clear to me. This check was there essentially since 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). It is clearly an implementation detail though and we shouldn't pull it into the generic retry logic while we should be able to cope with such eventuality. The only place in should_compact_retry where we retry without any upper bound is for compaction_withdrawn() case. Introduce compaction_zonelist_suitable function which checks the given zonelist and returns true only if there is at least one zone which would would unblock __compaction_suitable if more memory got reclaimed. In this implementation it checks __compaction_suitable with NR_FREE_PAGES plus part of the reclaimable memory as the target for the watermark check. The reclaimable memory is reduced linearly by the allocation order. The idea is that we do not want to reclaim all the remaining memory for a single allocation request just unblock __compaction_suitable which doesn't guarantee we will make a further progress. The new helper is then used if compaction_withdrawn() feedback was provided so we do not retry if there is no outlook for a further progress. !costly requests shouldn't be affected much - e.g. order-2 pages would require to have at least 64kB on the reclaimable LRUs while order-9 would need at least 32M which should be enough to not lock up. [vbabka@suse.cz: fix classzone_idx vs. high_zoneidx usage in compaction_zonelist_suitable] [akpm@linux-foundation.org: fix it for Mel's mm-page_alloc-remove-field-from-alloc_context.patch] Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Vladimir Davydov <vdavydov@virtuozzo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 23:57:12 +00:00
return true;
}
return false;
}
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
static enum compact_result
compact_zone(struct compact_control *cc, struct capture_control *capc)
{
enum compact_result ret;
unsigned long start_pfn = cc->zone->zone_start_pfn;
unsigned long end_pfn = zone_end_pfn(cc->zone);
unsigned long last_migrated_pfn;
const bool sync = cc->mode != MIGRATE_ASYNC;
bool update_cached;
/*
* These counters track activities during zone compaction. Initialize
* them before compacting a new zone.
*/
cc->total_migrate_scanned = 0;
cc->total_free_scanned = 0;
cc->nr_migratepages = 0;
cc->nr_freepages = 0;
INIT_LIST_HEAD(&cc->freepages);
INIT_LIST_HEAD(&cc->migratepages);
cc->migratetype = gfp_migratetype(cc->gfp_mask);
ret = compaction_suitable(cc->zone, cc->order, cc->alloc_flags,
cc->highest_zoneidx);
/* Compaction is likely to fail */
if (ret == COMPACT_SUCCESS || ret == COMPACT_SKIPPED)
return ret;
/* huh, compaction_suitable is returning something unexpected */
VM_BUG_ON(ret != COMPACT_CONTINUE);
mm: compaction: reset cached scanner pfn's before reading them Compaction caches pfn's for its migrate and free scanners to avoid scanning the whole zone each time. In compact_zone(), the cached values are read to set up initial values for the scanners. There are several situations when these cached pfn's are reset to the first and last pfn of the zone, respectively. One of these situations is when a compaction has been deferred for a zone and is now being restarted during a direct compaction, which is also done in compact_zone(). However, compact_zone() currently reads the cached pfn's *before* resetting them. This means the reset doesn't affect the compaction that performs it, and with good chance also subsequent compactions, as update_pageblock_skip() is likely to be called and update the cached pfn's to those being processed. Another chance for a successful reset is when a direct compaction detects that migration and free scanners meet (which has its own problems addressed by another patch) and sets update_pageblock_skip flag which kswapd uses to do the reset because it goes to sleep. This is clearly a bug that results in non-deterministic behavior, so this patch moves the cached pfn reset to be performed *before* the values are read. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:08 +00:00
/*
* Clear pageblock skip if there were failures recently and compaction
mm, kswapd: replace kswapd compaction with waking up kcompactd Similarly to direct reclaim/compaction, kswapd attempts to combine reclaim and compaction to attempt making memory allocation of given order available. The details differ from direct reclaim e.g. in having high watermark as a goal. The code involved in kswapd's reclaim/compaction decisions has evolved to be quite complex. Testing reveals that it doesn't actually work in at least one scenario, and closer inspection suggests that it could be greatly simplified without compromising on the goal (make high-order page available) or efficiency (don't reclaim too much). The simplification relieas of doing all compaction in kcompactd, which is simply woken up when high watermarks are reached by kswapd's reclaim. The scenario where kswapd compaction doesn't work was found with mmtests test stress-highalloc configured to attempt order-9 allocations without direct reclaim, just waking up kswapd. There was no compaction attempt from kswapd during the whole test. Some added instrumentation shows what happens: - balance_pgdat() sets end_zone to Normal, as it's not balanced - reclaim is attempted on DMA zone, which sets nr_attempted to 99, but it cannot reclaim anything, so sc.nr_reclaimed is 0 - for zones DMA32 and Normal, kswapd_shrink_zone uses testorder=0, so it merely checks if high watermarks were reached for base pages. This is true, so no reclaim is attempted. For DMA, testorder=0 wasn't used, as compaction_suitable() returned COMPACT_SKIPPED - even though the pgdat_needs_compaction flag wasn't set to false, no compaction happens due to the condition sc.nr_reclaimed > nr_attempted being false (as 0 < 99) - priority-- due to nr_reclaimed being 0, repeat until priority reaches 0 pgdat_balanced() is false as only the small zone DMA appears balanced (curiously in that check, watermark appears OK and compaction_suitable() returns COMPACT_PARTIAL, because a lower classzone_idx is used there) Now, even if it was decided that reclaim shouldn't be attempted on the DMA zone, the scenario would be the same, as (sc.nr_reclaimed=0 > nr_attempted=0) is also false. The condition really should use >= as the comment suggests. Then there is a mismatch in the check for setting pgdat_needs_compaction to false using low watermark, while the rest uses high watermark, and who knows what other subtlety. Hopefully this demonstrates that this is unsustainable. Luckily we can simplify this a lot. The reclaim/compaction decisions make sense for direct reclaim scenario, but in kswapd, our primary goal is to reach high watermark in order-0 pages. Afterwards we can attempt compaction just once. Unlike direct reclaim, we don't reclaim extra pages (over the high watermark), the current code already disallows it for good reasons. After this patch, we simply wake up kcompactd to process the pgdat, after we have either succeeded or failed to reach the high watermarks in kswapd, which goes to sleep. We pass kswapd's order and classzone_idx, so kcompactd can apply the same criteria to determine which zones are worth compacting. Note that we use the classzone_idx from wakeup_kswapd(), not balanced_classzone_idx which can include higher zones that kswapd tried to balance too, but didn't consider them in pgdat_balanced(). Since kswapd now cannot create high-order pages itself, we need to adjust how it determines the zones to be balanced. The key element here is adding a "highorder" parameter to zone_balanced, which, when set to false, makes it consider only order-0 watermark instead of the desired higher order (this was done previously by kswapd_shrink_zone(), but not elsewhere). This false is passed for example in pgdat_balanced(). Importantly, wakeup_kswapd() uses true to make sure kswapd and thus kcompactd are woken up for a high-order allocation failure. The last thing is to decide what to do with pageblock_skip bitmap handling. Compaction maintains a pageblock_skip bitmap to record pageblocks where isolation recently failed. This bitmap can be reset by three ways: 1) direct compaction is restarting after going through the full deferred cycle 2) kswapd goes to sleep, and some other direct compaction has previously finished scanning the whole zone and set zone->compact_blockskip_flush. Note that a successful direct compaction clears this flag. 3) compaction was invoked manually via trigger in /proc The case 2) is somewhat fuzzy to begin with, but after introducing kcompactd we should update it. The check for direct compaction in 1), and to set the flush flag in 2) use current_is_kswapd(), which doesn't work for kcompactd. Thus, this patch adds bool direct_compaction to compact_control to use in 2). For the case 1) we remove the check completely - unlike the former kswapd compaction, kcompactd does use the deferred compaction functionality, so flushing tied to restarting from deferred compaction makes sense here. Note that when kswapd goes to sleep, kcompactd is woken up, so it will see the flushed pageblock_skip bits. This is different from when the former kswapd compaction observed the bits and I believe it makes more sense. Kcompactd can afford to be more thorough than a direct compaction trying to limit allocation latency, or kswapd whose primary goal is to reclaim. For testing, I used stress-highalloc configured to do order-9 allocations with GFP_NOWAIT|__GFP_HIGH|__GFP_COMP, so they relied just on kswapd/kcompactd reclaim/compaction (the interfering kernel builds in phases 1 and 2 work as usual): stress-highalloc 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Success 1 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 1 Mean 1.40 ( 0.00%) 6.20 (-55.00%) Success 1 Max 2.00 ( 0.00%) 7.00 (-16.67%) Success 2 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 2 Mean 1.80 ( 0.00%) 6.40 (-52.38%) Success 2 Max 3.00 ( 0.00%) 7.00 (-16.67%) Success 3 Min 34.00 ( 0.00%) 62.00 ( 1.59%) Success 3 Mean 41.80 ( 0.00%) 63.80 ( 1.24%) Success 3 Max 53.00 ( 0.00%) 65.00 ( 2.99%) User 3166.67 3181.09 System 1153.37 1158.25 Elapsed 1768.53 1799.37 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Direct pages scanned 32938 32797 Kswapd pages scanned 2183166 2202613 Kswapd pages reclaimed 2152359 2143524 Direct pages reclaimed 32735 32545 Percentage direct scans 1% 1% THP fault alloc 579 612 THP collapse alloc 304 316 THP splits 0 0 THP fault fallback 793 778 THP collapse fail 11 16 Compaction stalls 1013 1007 Compaction success 92 67 Compaction failures 920 939 Page migrate success 238457 721374 Page migrate failure 23021 23469 Compaction pages isolated 504695 1479924 Compaction migrate scanned 661390 8812554 Compaction free scanned 13476658 84327916 Compaction cost 262 838 After this patch we see improvements in allocation success rate (especially for phase 3) along with increased compaction activity. The compaction stalls (direct compaction) in the interfering kernel builds (probably THP's) also decreased somewhat thanks to kcompactd activity, yet THP alloc successes improved a bit. Note that elapsed and user time isn't so useful for this benchmark, because of the background interference being unpredictable. It's just to quickly spot some major unexpected differences. System time is somewhat more useful and that didn't increase. Also (after adjusting mmtests' ftrace monitor): Time kswapd awake 2547781 2269241 Time kcompactd awake 0 119253 Time direct compacting 939937 557649 Time kswapd compacting 0 0 Time kcompactd compacting 0 119099 The decrease of overal time spent compacting appears to not match the increased compaction stats. I suspect the tasks get rescheduled and since the ftrace monitor doesn't see that, the reported time is wall time, not CPU time. But arguably direct compactors care about overall latency anyway, whether busy compacting or waiting for CPU doesn't matter. And that latency seems to almost halved. It's also interesting how much time kswapd spent awake just going through all the priorities and failing to even try compacting, over and over. We can also configure stress-highalloc to perform both direct reclaim/compaction and wakeup kswapd/kcompactd, by using GFP_KERNEL|__GFP_HIGH|__GFP_COMP: stress-highalloc 4.5-rc1+before 4.5-rc1+after -direct -direct Success 1 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 1 Mean 8.00 ( 0.00%) 10.00 (-19.05%) Success 1 Max 12.00 ( 0.00%) 11.00 ( 15.38%) Success 2 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 2 Mean 8.20 ( 0.00%) 10.00 (-16.28%) Success 2 Max 13.00 ( 0.00%) 11.00 ( 8.33%) Success 3 Min 75.00 ( 0.00%) 74.00 ( 1.33%) Success 3 Mean 75.60 ( 0.00%) 75.20 ( 0.53%) Success 3 Max 77.00 ( 0.00%) 76.00 ( 0.00%) User 3344.73 3246.04 System 1194.24 1172.29 Elapsed 1838.04 1836.76 4.5-rc1+before 4.5-rc1+after -direct -direct Direct pages scanned 125146 120966 Kswapd pages scanned 2119757 2135012 Kswapd pages reclaimed 2073183 2108388 Direct pages reclaimed 124909 120577 Percentage direct scans 5% 5% THP fault alloc 599 652 THP collapse alloc 323 354 THP splits 0 0 THP fault fallback 806 793 THP collapse fail 17 16 Compaction stalls 2457 2025 Compaction success 906 518 Compaction failures 1551 1507 Page migrate success 2031423 2360608 Page migrate failure 32845 40852 Compaction pages isolated 4129761 4802025 Compaction migrate scanned 11996712 21750613 Compaction free scanned 214970969 344372001 Compaction cost 2271 2694 In this scenario, this patch doesn't change the overall success rate as direct compaction already tries all it can. There's however significant reduction in direct compaction stalls (that is, the number of allocations that went into direct compaction). The number of successes (i.e. direct compaction stalls that ended up with successful allocation) is reduced by the same number. This means the offload to kcompactd is working as expected, and direct compaction is reduced either due to detecting contention, or compaction deferred by kcompactd. In the previous version of this patchset there was some apparent reduction of success rate, but the changes in this version (such as using sync compaction only), new baseline kernel, and/or averaging results from 5 executions (my bet), made this go away. Ftrace-based stats seem to roughly agree: Time kswapd awake 2532984 2326824 Time kcompactd awake 0 257916 Time direct compacting 864839 735130 Time kswapd compacting 0 0 Time kcompactd compacting 0 257585 Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:15 +00:00
* is about to be retried after being deferred.
mm: compaction: reset cached scanner pfn's before reading them Compaction caches pfn's for its migrate and free scanners to avoid scanning the whole zone each time. In compact_zone(), the cached values are read to set up initial values for the scanners. There are several situations when these cached pfn's are reset to the first and last pfn of the zone, respectively. One of these situations is when a compaction has been deferred for a zone and is now being restarted during a direct compaction, which is also done in compact_zone(). However, compact_zone() currently reads the cached pfn's *before* resetting them. This means the reset doesn't affect the compaction that performs it, and with good chance also subsequent compactions, as update_pageblock_skip() is likely to be called and update the cached pfn's to those being processed. Another chance for a successful reset is when a direct compaction detects that migration and free scanners meet (which has its own problems addressed by another patch) and sets update_pageblock_skip flag which kswapd uses to do the reset because it goes to sleep. This is clearly a bug that results in non-deterministic behavior, so this patch moves the cached pfn reset to be performed *before* the values are read. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:08 +00:00
*/
if (compaction_restarting(cc->zone, cc->order))
__reset_isolation_suitable(cc->zone);
mm: compaction: reset cached scanner pfn's before reading them Compaction caches pfn's for its migrate and free scanners to avoid scanning the whole zone each time. In compact_zone(), the cached values are read to set up initial values for the scanners. There are several situations when these cached pfn's are reset to the first and last pfn of the zone, respectively. One of these situations is when a compaction has been deferred for a zone and is now being restarted during a direct compaction, which is also done in compact_zone(). However, compact_zone() currently reads the cached pfn's *before* resetting them. This means the reset doesn't affect the compaction that performs it, and with good chance also subsequent compactions, as update_pageblock_skip() is likely to be called and update the cached pfn's to those being processed. Another chance for a successful reset is when a direct compaction detects that migration and free scanners meet (which has its own problems addressed by another patch) and sets update_pageblock_skip flag which kswapd uses to do the reset because it goes to sleep. This is clearly a bug that results in non-deterministic behavior, so this patch moves the cached pfn reset to be performed *before* the values are read. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:08 +00:00
mm: compaction: Restart compaction from near where it left off This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:45 +00:00
/*
* Setup to move all movable pages to the end of the zone. Used cached
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
* information on where the scanners should start (unless we explicitly
* want to compact the whole zone), but check that it is initialised
* by ensuring the values are within zone boundaries.
mm: compaction: Restart compaction from near where it left off This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:45 +00:00
*/
mm, compaction: use free lists to quickly locate a migration source The migration scanner is a linear scan of a zone with a potentiall large search space. Furthermore, many pageblocks are unusable such as those filled with reserved pages or partially filled with pages that cannot migrate. These still get scanned in the common case of allocating a THP and the cost accumulates. The patch uses a partial search of the free lists to locate a migration source candidate that is marked as MOVABLE when allocating a THP. It prefers picking a block with a larger number of free pages already on the basis that there are fewer pages to migrate to free the entire block. The lowest PFN found during searches is tracked as the basis of the start for the linear search after the first search of the free list fails. After the search, the free list is shuffled so that the next search will not encounter the same page. If the search fails then the subsequent searches will be shorter and the linear scanner is used. If this search fails, or if the request is for a small or unmovable/reclaimable allocation then the linear scanner is still used. It is somewhat pointless to use the list search in those cases. Small free pages must be used for the search and there is no guarantee that movable pages are located within that block that are contiguous. 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Amean fault-both-3 3771.41 ( 0.00%) 3390.40 ( 10.10%) Amean fault-both-5 5409.05 ( 0.00%) 5082.28 ( 6.04%) Amean fault-both-7 7040.74 ( 0.00%) 7012.51 ( 0.40%) Amean fault-both-12 11887.35 ( 0.00%) 11346.63 ( 4.55%) Amean fault-both-18 16718.19 ( 0.00%) 15324.19 ( 8.34%) Amean fault-both-24 21157.19 ( 0.00%) 16088.50 * 23.96%* Amean fault-both-30 21175.92 ( 0.00%) 18723.42 * 11.58%* Amean fault-both-32 21339.03 ( 0.00%) 18612.01 * 12.78%* 5.0.0-rc1 5.0.0-rc1 noboost-v3r10 findmig-v3r15 Percentage huge-3 86.50 ( 0.00%) 89.83 ( 3.85%) Percentage huge-5 92.52 ( 0.00%) 91.96 ( -0.61%) Percentage huge-7 92.44 ( 0.00%) 92.85 ( 0.44%) Percentage huge-12 92.98 ( 0.00%) 92.74 ( -0.25%) Percentage huge-18 91.70 ( 0.00%) 91.71 ( 0.02%) Percentage huge-24 91.59 ( 0.00%) 92.13 ( 0.60%) Percentage huge-30 90.14 ( 0.00%) 93.79 ( 4.04%) Percentage huge-32 90.03 ( 0.00%) 91.27 ( 1.37%) This shows an improvement in allocation latencies with similar allocation success rates. While not presented, there was a 31% reduction in migration scanning and a 8% reduction on system CPU usage. A 2-socket machine showed similar benefits. [mgorman@techsingularity.net: several fixes] Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net [vbabka@suse.cz: migrate block that was found-fast, some optimisations] Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <Vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:44:54 +00:00
cc->fast_start_pfn = 0;
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
if (cc->whole_zone) {
mm: compaction: Restart compaction from near where it left off This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by: Mel Gorman <mgorman@suse.de> Acked-by: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by: Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:32:45 +00:00
cc->migrate_pfn = start_pfn;
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
cc->free_pfn = pageblock_start_pfn(end_pfn - 1);
} else {
cc->migrate_pfn = cc->zone->compact_cached_migrate_pfn[sync];
cc->free_pfn = cc->zone->compact_cached_free_pfn;
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
if (cc->free_pfn < start_pfn || cc->free_pfn >= end_pfn) {
cc->free_pfn = pageblock_start_pfn(end_pfn - 1);
cc->zone->compact_cached_free_pfn = cc->free_pfn;
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
}
if (cc->migrate_pfn < start_pfn || cc->migrate_pfn >= end_pfn) {
cc->migrate_pfn = start_pfn;
cc->zone->compact_cached_migrate_pfn[0] = cc->migrate_pfn;
cc->zone->compact_cached_migrate_pfn[1] = cc->migrate_pfn;
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
}
2019-03-05 23:45:38 +00:00
if (cc->migrate_pfn <= cc->zone->compact_init_migrate_pfn)
mm, compaction: make whole_zone flag ignore cached scanner positions Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:35 +00:00
cc->whole_zone = true;
}
last_migrated_pfn = 0;
/*
* Migrate has separate cached PFNs for ASYNC and SYNC* migration on
* the basis that some migrations will fail in ASYNC mode. However,
* if the cached PFNs match and pageblocks are skipped due to having
* no isolation candidates, then the sync state does not matter.
* Until a pageblock with isolation candidates is found, keep the
* cached PFNs in sync to avoid revisiting the same blocks.
*/
update_cached = !sync &&
cc->zone->compact_cached_migrate_pfn[0] == cc->zone->compact_cached_migrate_pfn[1];
trace_mm_compaction_begin(start_pfn, cc->migrate_pfn,
cc->free_pfn, end_pfn, sync);
mm: compaction: trace compaction begin and end The broad goal of the series is to improve allocation success rates for huge pages through memory compaction, while trying not to increase the compaction overhead. The original objective was to reintroduce capturing of high-order pages freed by the compaction, before they are split by concurrent activity. However, several bugs and opportunities for simple improvements were found in the current implementation, mostly through extra tracepoints (which are however too ugly for now to be considered for sending). The patches mostly deal with two mechanisms that reduce compaction overhead, which is caching the progress of migrate and free scanners, and marking pageblocks where isolation failed to be skipped during further scans. Patch 1 (from mgorman) adds tracepoints that allow calculate time spent in compaction and potentially debug scanner pfn values. Patch 2 encapsulates the some functionality for handling deferred compactions for better maintainability, without a functional change type is not determined without being actually needed. Patch 3 fixes a bug where cached scanner pfn's are sometimes reset only after they have been read to initialize a compaction run. Patch 4 fixes a bug where scanners meeting is sometimes not properly detected and can lead to multiple compaction attempts quitting early without doing any work. Patch 5 improves the chances of sync compaction to process pageblocks that async compaction has skipped due to being !MIGRATE_MOVABLE. Patch 6 improves the chances of sync direct compaction to actually do anything when called after async compaction fails during allocation slowpath. The impact of patches were validated using mmtests's stress-highalloc benchmark with mmtests's stress-highalloc benchmark on a x86_64 machine with 4GB memory. Due to instability of the results (mostly related to the bugs fixed by patches 2 and 3), 10 iterations were performed, taking min,mean,max values for success rates and mean values for time and vmstat-based metrics. First, the default GFP_HIGHUSER_MOVABLE allocations were tested with the patches stacked on top of v3.13-rc2. Patch 2 is OK to serve as baseline due to no functional changes in 1 and 2. Comments below. stress-highalloc 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-nothp 3-nothp 4-nothp 5-nothp 6-nothp Success 1 Min 9.00 ( 0.00%) 10.00 (-11.11%) 43.00 (-377.78%) 43.00 (-377.78%) 33.00 (-266.67%) Success 1 Mean 27.50 ( 0.00%) 25.30 ( 8.00%) 45.50 (-65.45%) 45.90 (-66.91%) 46.30 (-68.36%) Success 1 Max 36.00 ( 0.00%) 36.00 ( 0.00%) 47.00 (-30.56%) 48.00 (-33.33%) 52.00 (-44.44%) Success 2 Min 10.00 ( 0.00%) 8.00 ( 20.00%) 46.00 (-360.00%) 45.00 (-350.00%) 35.00 (-250.00%) Success 2 Mean 26.40 ( 0.00%) 23.50 ( 10.98%) 47.30 (-79.17%) 47.60 (-80.30%) 48.10 (-82.20%) Success 2 Max 34.00 ( 0.00%) 33.00 ( 2.94%) 48.00 (-41.18%) 50.00 (-47.06%) 54.00 (-58.82%) Success 3 Min 65.00 ( 0.00%) 63.00 ( 3.08%) 85.00 (-30.77%) 84.00 (-29.23%) 85.00 (-30.77%) Success 3 Mean 76.70 ( 0.00%) 70.50 ( 8.08%) 86.20 (-12.39%) 85.50 (-11.47%) 86.00 (-12.13%) Success 3 Max 87.00 ( 0.00%) 86.00 ( 1.15%) 88.00 ( -1.15%) 87.00 ( 0.00%) 87.00 ( 0.00%) 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-nothp 3-nothp 4-nothp 5-nothp 6-nothp User 6437.72 6459.76 5960.32 5974.55 6019.67 System 1049.65 1049.09 1029.32 1031.47 1032.31 Elapsed 1856.77 1874.48 1949.97 1994.22 1983.15 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-nothp 3-nothp 4-nothp 5-nothp 6-nothp Minor Faults 253952267 254581900 250030122 250507333 250157829 Major Faults 420 407 506 530 530 Swap Ins 4 9 9 6 6 Swap Outs 398 375 345 346 333 Direct pages scanned 197538 189017 298574 287019 299063 Kswapd pages scanned 1809843 1801308 1846674 1873184 1861089 Kswapd pages reclaimed 1806972 1798684 1844219 1870509 1858622 Direct pages reclaimed 197227 188829 298380 286822 298835 Kswapd efficiency 99% 99% 99% 99% 99% Kswapd velocity 953.382 970.449 952.243 934.569 922.286 Direct efficiency 99% 99% 99% 99% 99% Direct velocity 104.058 101.832 153.961 143.200 148.205 Percentage direct scans 9% 9% 13% 13% 13% Zone normal velocity 347.289 359.676 348.063 339.933 332.983 Zone dma32 velocity 710.151 712.605 758.140 737.835 737.507 Zone dma velocity 0.000 0.000 0.000 0.000 0.000 Page writes by reclaim 557.600 429.000 353.600 426.400 381.800 Page writes file 159 53 7 79 48 Page writes anon 398 375 345 346 333 Page reclaim immediate 825 644 411 575 420 Sector Reads 2781750 2769780 2878547 2939128 2910483 Sector Writes 12080843 12083351 12012892 12002132 12010745 Page rescued immediate 0 0 0 0 0 Slabs scanned 1575654 1545344 1778406 1786700 1794073 Direct inode steals 9657 10037 15795 14104 14645 Kswapd inode steals 46857 46335 50543 50716 51796 Kswapd skipped wait 0 0 0 0 0 THP fault alloc 97 91 81 71 77 THP collapse alloc 456 506 546 544 565 THP splits 6 5 5 4 4 THP fault fallback 0 1 0 0 0 THP collapse fail 14 14 12 13 12 Compaction stalls 1006 980 1537 1536 1548 Compaction success 303 284 562 559 578 Compaction failures 702 696 974 976 969 Page migrate success 1177325 1070077 3927538 3781870 3877057 Page migrate failure 0 0 0 0 0 Compaction pages isolated 2547248 2306457 8301218 8008500 8200674 Compaction migrate scanned 42290478 38832618 153961130 154143900 159141197 Compaction free scanned 89199429 79189151 356529027 351943166 356326727 Compaction cost 1566 1426 5312 5156 5294 NUMA PTE updates 0 0 0 0 0 NUMA hint faults 0 0 0 0 0 NUMA hint local faults 0 0 0 0 0 NUMA hint local percent 100 100 100 100 100 NUMA pages migrated 0 0 0 0 0 AutoNUMA cost 0 0 0 0 0 Observations: - The "Success 3" line is allocation success rate with system idle (phases 1 and 2 are with background interference). I used to get stable values around 85% with vanilla 3.11. The lower min and mean values came with 3.12. This was bisected to commit 81c0a2bb ("mm: page_alloc: fair zone allocator policy") As explained in comment for patch 3, I don't think the commit is wrong, but that it makes the effect of compaction bugs worse. From patch 3 onwards, the results are OK and match the 3.11 results. - Patch 4 also clearly helps phases 1 and 2, and exceeds any results I've seen with 3.11 (I didn't measure it that thoroughly then, but it was never above 40%). - Compaction cost and number of scanned pages is higher, especially due to patch 4. However, keep in mind that patches 3 and 4 fix existing bugs in the current design of compaction overhead mitigation, they do not change it. If overhead is found unacceptable, then it should be decreased differently (and consistently, not due to random conditions) than the current implementation does. In contrast, patches 5 and 6 (which are not strictly bug fixes) do not increase the overhead (but also not success rates). This might be a limitation of the stress-highalloc benchmark as it's quite uniform. Another set of results is when configuring stress-highalloc t allocate with similar flags as THP uses: (GFP_HIGHUSER_MOVABLE|__GFP_NOMEMALLOC|__GFP_NORETRY|__GFP_NO_KSWAPD) stress-highalloc 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-thp 3-thp 4-thp 5-thp 6-thp Success 1 Min 2.00 ( 0.00%) 7.00 (-250.00%) 18.00 (-800.00%) 19.00 (-850.00%) 26.00 (-1200.00%) Success 1 Mean 19.20 ( 0.00%) 17.80 ( 7.29%) 29.20 (-52.08%) 29.90 (-55.73%) 32.80 (-70.83%) Success 1 Max 27.00 ( 0.00%) 29.00 ( -7.41%) 35.00 (-29.63%) 36.00 (-33.33%) 37.00 (-37.04%) Success 2 Min 3.00 ( 0.00%) 8.00 (-166.67%) 21.00 (-600.00%) 21.00 (-600.00%) 32.00 (-966.67%) Success 2 Mean 19.30 ( 0.00%) 17.90 ( 7.25%) 32.20 (-66.84%) 32.60 (-68.91%) 35.70 (-84.97%) Success 2 Max 27.00 ( 0.00%) 30.00 (-11.11%) 36.00 (-33.33%) 37.00 (-37.04%) 39.00 (-44.44%) Success 3 Min 62.00 ( 0.00%) 62.00 ( 0.00%) 85.00 (-37.10%) 75.00 (-20.97%) 64.00 ( -3.23%) Success 3 Mean 66.30 ( 0.00%) 65.50 ( 1.21%) 85.60 (-29.11%) 83.40 (-25.79%) 83.50 (-25.94%) Success 3 Max 70.00 ( 0.00%) 69.00 ( 1.43%) 87.00 (-24.29%) 86.00 (-22.86%) 87.00 (-24.29%) 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-thp 3-thp 4-thp 5-thp 6-thp User 6547.93 6475.85 6265.54 6289.46 6189.96 System 1053.42 1047.28 1043.23 1042.73 1038.73 Elapsed 1835.43 1821.96 1908.67 1912.74 1956.38 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-thp 3-thp 4-thp 5-thp 6-thp Minor Faults 256805673 253106328 253222299 249830289 251184418 Major Faults 395 375 423 434 448 Swap Ins 12 10 10 12 9 Swap Outs 530 537 487 455 415 Direct pages scanned 71859 86046 153244 152764 190713 Kswapd pages scanned 1900994 1870240 1898012 1892864 1880520 Kswapd pages reclaimed 1897814 1867428 1894939 1890125 1877924 Direct pages reclaimed 71766 85908 153167 152643 190600 Kswapd efficiency 99% 99% 99% 99% 99% Kswapd velocity 1029.000 1067.782 1000.091 991.049 951.218 Direct efficiency 99% 99% 99% 99% 99% Direct velocity 38.897 49.127 80.747 79.983 96.468 Percentage direct scans 3% 4% 7% 7% 9% Zone normal velocity 351.377 372.494 348.910 341.689 335.310 Zone dma32 velocity 716.520 744.414 731.928 729.343 712.377 Zone dma velocity 0.000 0.000 0.000 0.000 0.000 Page writes by reclaim 669.300 604.000 545.700 538.900 429.900 Page writes file 138 66 58 83 14 Page writes anon 530 537 487 455 415 Page reclaim immediate 806 655 772 548 517 Sector Reads 2711956 2703239 2811602 2818248 2839459 Sector Writes 12163238 12018662 12038248 11954736 11994892 Page rescued immediate 0 0 0 0 0 Slabs scanned 1385088 1388364 1507968 1513292 1558656 Direct inode steals 1739 2564 4622 5496 6007 Kswapd inode steals 47461 46406 47804 48013 48466 Kswapd skipped wait 0 0 0 0 0 THP fault alloc 110 82 84 69 70 THP collapse alloc 445 482 467 462 539 THP splits 6 5 4 5 3 THP fault fallback 3 0 0 0 0 THP collapse fail 15 14 14 14 13 Compaction stalls 659 685 1033 1073 1111 Compaction success 222 225 410 427 456 Compaction failures 436 460 622 646 655 Page migrate success 446594 439978 1085640 1095062 1131716 Page migrate failure 0 0 0 0 0 Compaction pages isolated 1029475 1013490 2453074 2482698 2565400 Compaction migrate scanned 9955461 11344259 24375202 27978356 30494204 Compaction free scanned 27715272 28544654 80150615 82898631 85756132 Compaction cost 552 555 1344 1379 1436 NUMA PTE updates 0 0 0 0 0 NUMA hint faults 0 0 0 0 0 NUMA hint local faults 0 0 0 0 0 NUMA hint local percent 100 100 100 100 100 NUMA pages migrated 0 0 0 0 0 AutoNUMA cost 0 0 0 0 0 There are some differences from the previous results for THP-like allocations: - Here, the bad result for unpatched kernel in phase 3 is much more consistent to be between 65-70% and not related to the "regression" in 3.12. Still there is the improvement from patch 4 onwards, which brings it on par with simple GFP_HIGHUSER_MOVABLE allocations. - Compaction costs have increased, but nowhere near as much as the non-THP case. Again, the patches should be worth the gained determininsm. - Patches 5 and 6 somewhat increase the number of migrate-scanned pages. This is most likely due to __GFP_NO_KSWAPD flag, which means the cached pfn's and pageblock skip bits are not reset by kswapd that often (at least in phase 3 where no concurrent activity would wake up kswapd) and the patches thus help the sync-after-async compaction. It doesn't however show that the sync compaction would help so much with success rates, which can be again seen as a limitation of the benchmark scenario. This patch (of 6): Add two tracepoints for compaction begin and end of a zone. Using this it is possible to calculate how much time a workload is spending within compaction and potentially debug problems related to cached pfns for scanning. In combination with the direct reclaim and slab trace points it should be possible to estimate most allocation-related overhead for a workload. Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:05 +00:00
/* lru_add_drain_all could be expensive with involving other CPUs */
lru_add_drain();
while ((ret = compact_finished(cc)) == COMPACT_CONTINUE) {
int err;
unsigned long iteration_start_pfn = cc->migrate_pfn;
mm, compaction: avoid rescanning the same pageblock multiple times Pageblocks are marked for skip when no pages are isolated after a scan. However, it's possible to hit corner cases where the migration scanner gets stuck near the boundary between the source and target scanner. Due to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are migrated can be reallocated before the pageblock is complete. The pageblock is not necessarily skipped so it can be rescanned multiple times. Similarly, a pageblock with some dirty/writeback pages may fail to migrate and be rescanned until writeback completes which is wasteful. This patch tracks if a pageblock is being rescanned. If so, then the entire pageblock will be migrated as one operation. This narrows the race window during which pages can be reallocated during migration. Secondly, if there are pages that cannot be isolated then the pageblock will still be fully scanned and marked for skipping. On the second rescan, the pageblock skip is set and the migration scanner makes progress. 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3200.68 ( 0.00%) 3002.07 ( 6.21%) Amean fault-both-5 4847.75 ( 0.00%) 4684.47 ( 3.37%) Amean fault-both-7 6658.92 ( 0.00%) 6815.54 ( -2.35%) Amean fault-both-12 11077.62 ( 0.00%) 10864.02 ( 1.93%) Amean fault-both-18 12403.97 ( 0.00%) 12247.52 ( 1.26%) Amean fault-both-24 15607.10 ( 0.00%) 15683.99 ( -0.49%) Amean fault-both-30 18752.27 ( 0.00%) 18620.02 ( 0.71%) Amean fault-both-32 21207.54 ( 0.00%) 19250.28 * 9.23%* 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Percentage huge-3 96.86 ( 0.00%) 95.00 ( -1.91%) Percentage huge-5 93.72 ( 0.00%) 94.22 ( 0.53%) Percentage huge-7 94.31 ( 0.00%) 92.35 ( -2.08%) Percentage huge-12 92.66 ( 0.00%) 91.90 ( -0.82%) Percentage huge-18 91.51 ( 0.00%) 89.58 ( -2.11%) Percentage huge-24 90.50 ( 0.00%) 90.03 ( -0.52%) Percentage huge-30 91.57 ( 0.00%) 89.14 ( -2.65%) Percentage huge-32 91.00 ( 0.00%) 90.58 ( -0.46%) Negligible difference but this was likely a case when the specific corner case was not hit. A previous run of the same patch based on an earlier iteration of the series showed large differences where migration rates could be halved when the corner case was hit. The specific corner case where migration scan rates go through the roof was due to a dirty/writeback pageblock located at the boundary of the migration/free scanner did not happen in this case. When it does happen, the scan rates multipled by massive margins. Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:07 +00:00
/*
* Avoid multiple rescans which can happen if a page cannot be
* isolated (dirty/writeback in async mode) or if the migrated
* pages are being allocated before the pageblock is cleared.
* The first rescan will capture the entire pageblock for
* migration. If it fails, it'll be marked skip and scanning
* will proceed as normal.
*/
cc->rescan = false;
if (pageblock_start_pfn(last_migrated_pfn) ==
pageblock_start_pfn(iteration_start_pfn)) {
mm, compaction: avoid rescanning the same pageblock multiple times Pageblocks are marked for skip when no pages are isolated after a scan. However, it's possible to hit corner cases where the migration scanner gets stuck near the boundary between the source and target scanner. Due to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are migrated can be reallocated before the pageblock is complete. The pageblock is not necessarily skipped so it can be rescanned multiple times. Similarly, a pageblock with some dirty/writeback pages may fail to migrate and be rescanned until writeback completes which is wasteful. This patch tracks if a pageblock is being rescanned. If so, then the entire pageblock will be migrated as one operation. This narrows the race window during which pages can be reallocated during migration. Secondly, if there are pages that cannot be isolated then the pageblock will still be fully scanned and marked for skipping. On the second rescan, the pageblock skip is set and the migration scanner makes progress. 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 3200.68 ( 0.00%) 3002.07 ( 6.21%) Amean fault-both-5 4847.75 ( 0.00%) 4684.47 ( 3.37%) Amean fault-both-7 6658.92 ( 0.00%) 6815.54 ( -2.35%) Amean fault-both-12 11077.62 ( 0.00%) 10864.02 ( 1.93%) Amean fault-both-18 12403.97 ( 0.00%) 12247.52 ( 1.26%) Amean fault-both-24 15607.10 ( 0.00%) 15683.99 ( -0.49%) Amean fault-both-30 18752.27 ( 0.00%) 18620.02 ( 0.71%) Amean fault-both-32 21207.54 ( 0.00%) 19250.28 * 9.23%* 5.0.0-rc1 5.0.0-rc1 findfree-v3r16 norescan-v3r16 Percentage huge-3 96.86 ( 0.00%) 95.00 ( -1.91%) Percentage huge-5 93.72 ( 0.00%) 94.22 ( 0.53%) Percentage huge-7 94.31 ( 0.00%) 92.35 ( -2.08%) Percentage huge-12 92.66 ( 0.00%) 91.90 ( -0.82%) Percentage huge-18 91.51 ( 0.00%) 89.58 ( -2.11%) Percentage huge-24 90.50 ( 0.00%) 90.03 ( -0.52%) Percentage huge-30 91.57 ( 0.00%) 89.14 ( -2.65%) Percentage huge-32 91.00 ( 0.00%) 90.58 ( -0.46%) Negligible difference but this was likely a case when the specific corner case was not hit. A previous run of the same patch based on an earlier iteration of the series showed large differences where migration rates could be halved when the corner case was hit. The specific corner case where migration scan rates go through the roof was due to a dirty/writeback pageblock located at the boundary of the migration/free scanner did not happen in this case. When it does happen, the scan rates multipled by massive margins. Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:07 +00:00
cc->rescan = true;
}
switch (isolate_migratepages(cc)) {
case ISOLATE_ABORT:
ret = COMPACT_CONTENDED;
putback_movable_pages(&cc->migratepages);
cc->nr_migratepages = 0;
goto out;
case ISOLATE_NONE:
if (update_cached) {
cc->zone->compact_cached_migrate_pfn[1] =
cc->zone->compact_cached_migrate_pfn[0];
}
mm, compaction: more focused lru and pcplists draining The goal of memory compaction is to create high-order freepages through page migration. Page migration however puts pages on the per-cpu lru_add cache, which is later flushed to per-cpu pcplists, and only after pcplists are drained the pages can actually merge. This can happen due to the per-cpu caches becoming full through further freeing, or explicitly. During direct compaction, it is useful to do the draining explicitly so that pages merge as soon as possible and compaction can detect success immediately and keep the latency impact at minimum. However the current implementation is far from ideal. Draining is done only in __alloc_pages_direct_compact(), after all zones were already compacted, and the decisions to continue or stop compaction in individual zones was done without the last batch of migrations being merged. It is also missing the draining of lru_add cache before the pcplists. This patch moves the draining for direct compaction into compact_zone(). It adds the missing lru_cache draining and uses the newly introduced single zone pcplists draining to reduce overhead and avoid impact on unrelated zones. Draining is only performed when it can actually lead to merging of a page of desired order (passed by cc->order). This means it is only done when migration occurred in the previously scanned cc->order aligned block(s) and the migration scanner is now pointing to the next cc->order aligned block. The patch has been tested with stress-highalloc benchmark from mmtests. Although overal allocation success rates of the benchmark were not affected, the number of detected compaction successes has doubled. This suggests that allocations were previously successful due to implicit merging caused by background activity, making a later allocation attempt succeed immediately, but not attributing the success to compaction. Since stress-highalloc always tries to allocate almost the whole memory, it cannot show the improvement in its reported success rate metric. However after this patch, compaction should detect success and terminate earlier, reducing the direct compaction latencies in a real scenario. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:34 +00:00
/*
* We haven't isolated and migrated anything, but
* there might still be unflushed migrations from
* previous cc->order aligned block.
*/
goto check_drain;
case ISOLATE_SUCCESS:
update_cached = false;
last_migrated_pfn = iteration_start_pfn;
}
mm, compaction: return failed migration target pages back to freelist Greg reported that he found isolated free pages were returned back to the VM rather than the compaction freelist. This will cause holes behind the free scanner and cause it to reallocate additional memory if necessary later. He detected the problem at runtime seeing that ext4 metadata pages (esp the ones read by "sbi->s_group_desc[i] = sb_bread(sb, block)") were constantly visited by compaction calls of migrate_pages(). These pages had a non-zero b_count which caused fallback_migrate_page() -> try_to_release_page() -> try_to_free_buffers() to fail. Memory compaction works by having a "freeing scanner" scan from one end of a zone which isolates pages as migration targets while another "migrating scanner" scans from the other end of the same zone which isolates pages for migration. When page migration fails for an isolated page, the target page is returned to the system rather than the freelist built by the freeing scanner. This may require the freeing scanner to continue scanning memory after suitable migration targets have already been returned to the system needlessly. This patch returns destination pages to the freeing scanner freelist when page migration fails. This prevents unnecessary work done by the freeing scanner but also encourages memory to be as compacted as possible at the end of the zone. Signed-off-by: David Rientjes <rientjes@google.com> Reported-by: Greg Thelen <gthelen@google.com> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:08:26 +00:00
err = migrate_pages(&cc->migratepages, compaction_alloc,
compaction_free, (unsigned long)cc, cc->mode,
MR_COMPACTION);
mm/compaction: do not count migratepages when unnecessary During compaction, update_nr_listpages() has been used to count remaining non-migrated and free pages after a call to migrage_pages(). The freepages counting has become unneccessary, and it turns out that migratepages counting is also unnecessary in most cases. The only situation when it's needed to count cc->migratepages is when migrate_pages() returns with a negative error code. Otherwise, the non-negative return value is the number of pages that were not migrated, which is exactly the count of remaining pages in the cc->migratepages list. Furthermore, any non-zero count is only interesting for the tracepoint of mm_compaction_migratepages events, because after that all remaining unmigrated pages are put back and their count is set to 0. This patch therefore removes update_nr_listpages() completely, and changes the tracepoint definition so that the manual counting is done only when the tracepoint is enabled, and only when migrate_pages() returns a negative error code. Furthermore, migrate_pages() and the tracepoints won't be called when there's nothing to migrate. This potentially avoids some wasted cycles and reduces the volume of uninteresting mm_compaction_migratepages events where "nr_migrated=0 nr_failed=0". In the stress-highalloc mmtest, this was about 75% of the events. The mm_compaction_isolate_migratepages event is better for determining that nothing was isolated for migration, and this one was just duplicating the info. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:08:32 +00:00
trace_mm_compaction_migratepages(cc->nr_migratepages, err,
&cc->migratepages);
mm/compaction: do not count migratepages when unnecessary During compaction, update_nr_listpages() has been used to count remaining non-migrated and free pages after a call to migrage_pages(). The freepages counting has become unneccessary, and it turns out that migratepages counting is also unnecessary in most cases. The only situation when it's needed to count cc->migratepages is when migrate_pages() returns with a negative error code. Otherwise, the non-negative return value is the number of pages that were not migrated, which is exactly the count of remaining pages in the cc->migratepages list. Furthermore, any non-zero count is only interesting for the tracepoint of mm_compaction_migratepages events, because after that all remaining unmigrated pages are put back and their count is set to 0. This patch therefore removes update_nr_listpages() completely, and changes the tracepoint definition so that the manual counting is done only when the tracepoint is enabled, and only when migrate_pages() returns a negative error code. Furthermore, migrate_pages() and the tracepoints won't be called when there's nothing to migrate. This potentially avoids some wasted cycles and reduces the volume of uninteresting mm_compaction_migratepages events where "nr_migrated=0 nr_failed=0". In the stress-highalloc mmtest, this was about 75% of the events. The mm_compaction_isolate_migratepages event is better for determining that nothing was isolated for migration, and this one was just duplicating the info. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:08:32 +00:00
/* All pages were either migrated or will be released */
cc->nr_migratepages = 0;
if (err) {
putback_movable_pages(&cc->migratepages);
mm: compaction: detect when scanners meet in isolate_freepages Compaction of a zone is finished when the migrate scanner (which begins at the zone's lowest pfn) meets the free page scanner (which begins at the zone's highest pfn). This is detected in compact_zone() and in the case of direct compaction, the compact_blockskip_flush flag is set so that kswapd later resets the cached scanner pfn's, and a new compaction may again start at the zone's borders. The meeting of the scanners can happen during either scanner's activity. However, it may currently fail to be detected when it occurs in the free page scanner, due to two problems. First, isolate_freepages() keeps free_pfn at the highest block where it isolated pages from, for the purposes of not missing the pages that are returned back to allocator when migration fails. Second, failing to isolate enough free pages due to scanners meeting results in -ENOMEM being returned by migrate_pages(), which makes compact_zone() bail out immediately without calling compact_finished() that would detect scanners meeting. This failure to detect scanners meeting might result in repeated attempts at compaction of a zone that keep starting from the cached pfn's close to the meeting point, and quickly failing through the -ENOMEM path, without the cached pfns being reset, over and over. This has been observed (through additional tracepoints) in the third phase of the mmtests stress-highalloc benchmark, where the allocator runs on an otherwise idle system. The problem was observed in the DMA32 zone, which was used as a fallback to the preferred Normal zone, but on the 4GB system it was actually the largest zone. The problem is even amplified for such fallback zone - the deferred compaction logic, which could (after being fixed by a previous patch) reset the cached scanner pfn's, is only applied to the preferred zone and not for the fallbacks. The problem in the third phase of the benchmark was further amplified by commit 81c0a2bb515f ("mm: page_alloc: fair zone allocator policy") which resulted in a non-deterministic regression of the allocation success rate from ~85% to ~65%. This occurs in about half of benchmark runs, making bisection problematic. It is unlikely that the commit itself is buggy, but it should put more pressure on the DMA32 zone during phases 1 and 2, which may leave it more fragmented in phase 3 and expose the bugs that this patch fixes. The fix is to make scanners meeting in isolate_freepage() stay that way, and to check in compact_zone() for scanners meeting when migrate_pages() returns -ENOMEM. The result is that compact_finished() also detects scanners meeting and sets the compact_blockskip_flush flag to make kswapd reset the scanner pfn's. The results in stress-highalloc benchmark show that the "regression" by commit 81c0a2bb515f in phase 3 no longer occurs, and phase 1 and 2 allocation success rates are also significantly improved. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:09 +00:00
/*
* migrate_pages() may return -ENOMEM when scanners meet
* and we want compact_finished() to detect it
*/
mm, compaction: more robust check for scanners meeting Assorted compaction cleanups and optimizations. The interesting patches are 4 and 5. In 4, skipping of compound pages in single iteration is improved for migration scanner, so it works also for !PageLRU compound pages such as hugetlbfs, slab etc. Patch 5 introduces this kind of skipping in the free scanner. The trick is that we can read compound_order() without any protection, if we are careful to filter out values larger than MAX_ORDER. The only danger is that we skip too much. The same trick was already used for reading the freepage order in the migrate scanner. To demonstrate improvements of Patches 4 and 5 I've run stress-highalloc from mmtests, set to simulate THP allocations (including __GFP_COMP) on a 4GB system where 1GB was occupied by hugetlbfs pages. I'll include just the relevant stats: Patch 3 Patch 4 Patch 5 Compaction stalls 7523 7529 7515 Compaction success 323 304 322 Compaction failures 7200 7224 7192 Page migrate success 247778 264395 240737 Page migrate failure 15358 33184 21621 Compaction pages isolated 906928 980192 909983 Compaction migrate scanned 2005277 1692805 1498800 Compaction free scanned 13255284 11539986 9011276 Compaction cost 288 305 277 With 5 iterations per patch, the results are still noisy, but we can see that Patch 4 does reduce migrate_scanned by 15% thanks to skipping the hugetlbfs pages at once. Interestingly, free_scanned is also reduced and I have no idea why. Patch 5 further reduces free_scanned as expected, by 15%. Other stats are unaffected modulo noise. [1] https://lkml.org/lkml/2015/1/19/158 This patch (of 5): Compaction should finish when the migration and free scanner meet, i.e. they reach the same pageblock. Currently however, the test in compact_finished() simply just compares the exact pfns, which may yield a false negative when the free scanner position is in the middle of a pageblock and the migration scanner reaches the begining of the same pageblock. This hasn't been a problem until commit e14c720efdd7 ("mm, compaction: remember position within pageblock in free pages scanner") allowed the free scanner position to be in the middle of a pageblock between invocations. The hot-fix 1d5bfe1ffb5b ("mm, compaction: prevent infinite loop in compact_zone") prevented the issue by adding a special check in the migration scanner to satisfy the current detection of scanners meeting. However, the proper fix is to make the detection more robust. This patch introduces the compact_scanners_met() function that returns true when the free scanner position is in the same or lower pageblock than the migration scanner. The special case in isolate_migratepages() introduced by 1d5bfe1ffb5b is removed. Suggested-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 22:02:36 +00:00
if (err == -ENOMEM && !compact_scanners_met(cc)) {
ret = COMPACT_CONTENDED;
goto out;
}
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
/*
* We failed to migrate at least one page in the current
* order-aligned block, so skip the rest of it.
*/
if (cc->direct_compaction &&
(cc->mode == MIGRATE_ASYNC)) {
cc->migrate_pfn = block_end_pfn(
cc->migrate_pfn - 1, cc->order);
/* Draining pcplists is useless in this case */
last_migrated_pfn = 0;
mm, compaction: skip blocks where isolation fails in async direct compaction The goal of direct compaction is to quickly make a high-order page available for the pending allocation. Within an aligned block of pages of desired order, a single allocated page that cannot be isolated for migration means that the block cannot fully merge to a buddy page that would satisfy the allocation request. Therefore we can reduce the allocation stall by skipping the rest of the block immediately on isolation failure. For async compaction, this also means a higher chance of succeeding until it detects contention. We however shouldn't completely sacrifice the second objective of compaction, which is to reduce overal long-term memory fragmentation. As a compromise, perform the eager skipping only in direct async compaction, while sync compaction (including kcompactd) remains thorough. Testing was done using stress-highalloc from mmtests, configured for order-4 GFP_KERNEL allocations: 4.6-rc1 4.6-rc1 before after Success 1 Min 24.00 ( 0.00%) 27.00 (-12.50%) Success 1 Mean 30.20 ( 0.00%) 31.60 ( -4.64%) Success 1 Max 37.00 ( 0.00%) 35.00 ( 5.41%) Success 2 Min 42.00 ( 0.00%) 32.00 ( 23.81%) Success 2 Mean 44.00 ( 0.00%) 44.80 ( -1.82%) Success 2 Max 48.00 ( 0.00%) 52.00 ( -8.33%) Success 3 Min 91.00 ( 0.00%) 92.00 ( -1.10%) Success 3 Mean 92.20 ( 0.00%) 92.80 ( -0.65%) Success 3 Max 94.00 ( 0.00%) 93.00 ( 1.06%) We can see that success rates are unaffected by the skipping. 4.6-rc1 4.6-rc1 before after User 2587.42 2566.53 System 482.89 471.20 Elapsed 1395.68 1382.00 Times are not so useful metric for this benchmark as main portion is the interfering kernel builds, but results do hint at reduced system times. 4.6-rc1 4.6-rc1 before after Direct pages scanned 163614 159608 Kswapd pages scanned 2070139 2078790 Kswapd pages reclaimed 2061707 2069757 Direct pages reclaimed 163354 159505 Reduced direct reclaim was unintended, but could be explained by more successful first attempt at (async) direct compaction, which is attempted before the first reclaim attempt in __alloc_pages_slowpath(). Compaction stalls 33052 39853 Compaction success 12121 19773 Compaction failures 20931 20079 Compaction is indeed more successful, and thus less likely to get deferred, so there are also more direct compaction stalls. Page migrate success 3781876 3326819 Page migrate failure 45817 41774 Compaction pages isolated 7868232 6941457 Compaction migrate scanned 168160492 127269354 Compaction migrate prescanned 0 0 Compaction free scanned 2522142582 2326342620 Compaction free direct alloc 0 0 Compaction free dir. all. miss 0 0 Compaction cost 5252 4476 The patch reduces migration scanned pages by 25% thanks to the eager skipping. [hughd@google.com: prevent nr_isolated_* from going negative] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-20 00:11:55 +00:00
}
}
mm, compaction: more focused lru and pcplists draining The goal of memory compaction is to create high-order freepages through page migration. Page migration however puts pages on the per-cpu lru_add cache, which is later flushed to per-cpu pcplists, and only after pcplists are drained the pages can actually merge. This can happen due to the per-cpu caches becoming full through further freeing, or explicitly. During direct compaction, it is useful to do the draining explicitly so that pages merge as soon as possible and compaction can detect success immediately and keep the latency impact at minimum. However the current implementation is far from ideal. Draining is done only in __alloc_pages_direct_compact(), after all zones were already compacted, and the decisions to continue or stop compaction in individual zones was done without the last batch of migrations being merged. It is also missing the draining of lru_add cache before the pcplists. This patch moves the draining for direct compaction into compact_zone(). It adds the missing lru_cache draining and uses the newly introduced single zone pcplists draining to reduce overhead and avoid impact on unrelated zones. Draining is only performed when it can actually lead to merging of a page of desired order (passed by cc->order). This means it is only done when migration occurred in the previously scanned cc->order aligned block(s) and the migration scanner is now pointing to the next cc->order aligned block. The patch has been tested with stress-highalloc benchmark from mmtests. Although overal allocation success rates of the benchmark were not affected, the number of detected compaction successes has doubled. This suggests that allocations were previously successful due to implicit merging caused by background activity, making a later allocation attempt succeed immediately, but not attributing the success to compaction. Since stress-highalloc always tries to allocate almost the whole memory, it cannot show the improvement in its reported success rate metric. However after this patch, compaction should detect success and terminate earlier, reducing the direct compaction latencies in a real scenario. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:34 +00:00
check_drain:
/*
* Has the migration scanner moved away from the previous
* cc->order aligned block where we migrated from? If yes,
* flush the pages that were freed, so that they can merge and
* compact_finished() can detect immediately if allocation
* would succeed.
*/
if (cc->order > 0 && last_migrated_pfn) {
mm, compaction: more focused lru and pcplists draining The goal of memory compaction is to create high-order freepages through page migration. Page migration however puts pages on the per-cpu lru_add cache, which is later flushed to per-cpu pcplists, and only after pcplists are drained the pages can actually merge. This can happen due to the per-cpu caches becoming full through further freeing, or explicitly. During direct compaction, it is useful to do the draining explicitly so that pages merge as soon as possible and compaction can detect success immediately and keep the latency impact at minimum. However the current implementation is far from ideal. Draining is done only in __alloc_pages_direct_compact(), after all zones were already compacted, and the decisions to continue or stop compaction in individual zones was done without the last batch of migrations being merged. It is also missing the draining of lru_add cache before the pcplists. This patch moves the draining for direct compaction into compact_zone(). It adds the missing lru_cache draining and uses the newly introduced single zone pcplists draining to reduce overhead and avoid impact on unrelated zones. Draining is only performed when it can actually lead to merging of a page of desired order (passed by cc->order). This means it is only done when migration occurred in the previously scanned cc->order aligned block(s) and the migration scanner is now pointing to the next cc->order aligned block. The patch has been tested with stress-highalloc benchmark from mmtests. Although overal allocation success rates of the benchmark were not affected, the number of detected compaction successes has doubled. This suggests that allocations were previously successful due to implicit merging caused by background activity, making a later allocation attempt succeed immediately, but not attributing the success to compaction. Since stress-highalloc always tries to allocate almost the whole memory, it cannot show the improvement in its reported success rate metric. However after this patch, compaction should detect success and terminate earlier, reducing the direct compaction latencies in a real scenario. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:34 +00:00
unsigned long current_block_start =
block_start_pfn(cc->migrate_pfn, cc->order);
mm, compaction: more focused lru and pcplists draining The goal of memory compaction is to create high-order freepages through page migration. Page migration however puts pages on the per-cpu lru_add cache, which is later flushed to per-cpu pcplists, and only after pcplists are drained the pages can actually merge. This can happen due to the per-cpu caches becoming full through further freeing, or explicitly. During direct compaction, it is useful to do the draining explicitly so that pages merge as soon as possible and compaction can detect success immediately and keep the latency impact at minimum. However the current implementation is far from ideal. Draining is done only in __alloc_pages_direct_compact(), after all zones were already compacted, and the decisions to continue or stop compaction in individual zones was done without the last batch of migrations being merged. It is also missing the draining of lru_add cache before the pcplists. This patch moves the draining for direct compaction into compact_zone(). It adds the missing lru_cache draining and uses the newly introduced single zone pcplists draining to reduce overhead and avoid impact on unrelated zones. Draining is only performed when it can actually lead to merging of a page of desired order (passed by cc->order). This means it is only done when migration occurred in the previously scanned cc->order aligned block(s) and the migration scanner is now pointing to the next cc->order aligned block. The patch has been tested with stress-highalloc benchmark from mmtests. Although overal allocation success rates of the benchmark were not affected, the number of detected compaction successes has doubled. This suggests that allocations were previously successful due to implicit merging caused by background activity, making a later allocation attempt succeed immediately, but not attributing the success to compaction. Since stress-highalloc always tries to allocate almost the whole memory, it cannot show the improvement in its reported success rate metric. However after this patch, compaction should detect success and terminate earlier, reducing the direct compaction latencies in a real scenario. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:34 +00:00
if (last_migrated_pfn < current_block_start) {
mm/swap: Use local_lock for protection The various struct pagevec per CPU variables are protected by disabling either preemption or interrupts across the critical sections. Inside these sections spinlocks have to be acquired. These spinlocks are regular spinlock_t types which are converted to "sleeping" spinlocks on PREEMPT_RT enabled kernels. Obviously sleeping locks cannot be acquired in preemption or interrupt disabled sections. local locks provide a trivial way to substitute preempt and interrupt disable instances. On a non PREEMPT_RT enabled kernel local_lock() maps to preempt_disable() and local_lock_irq() to local_irq_disable(). Create lru_rotate_pvecs containing the pagevec and the locallock. Create lru_pvecs containing the remaining pagevecs and the locallock. Add lru_add_drain_cpu_zone() which is used from compact_zone() to avoid exporting the pvec structure. Change the relevant call sites to acquire these locks instead of using preempt_disable() / get_cpu() / get_cpu_var() and local_irq_disable() / local_irq_save(). There is neither a functional change nor a change in the generated binary code for non PREEMPT_RT enabled non-debug kernels. When lockdep is enabled local locks have lockdep maps embedded. These allow lockdep to validate the protections, i.e. inappropriate usage of a preemption only protected sections would result in a lockdep warning while the same problem would not be noticed with a plain preempt_disable() based protection. local locks also improve readability as they provide a named scope for the protections while preempt/interrupt disable are opaque scopeless. Finally local locks allow PREEMPT_RT to substitute them with real locking primitives to ensure the correctness of operation in a fully preemptible kernel. [ bigeasy: Adopted to use local_lock ] Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org> Acked-by: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20200527201119.1692513-4-bigeasy@linutronix.de
2020-05-27 20:11:15 +00:00
lru_add_drain_cpu_zone(cc->zone);
mm, compaction: more focused lru and pcplists draining The goal of memory compaction is to create high-order freepages through page migration. Page migration however puts pages on the per-cpu lru_add cache, which is later flushed to per-cpu pcplists, and only after pcplists are drained the pages can actually merge. This can happen due to the per-cpu caches becoming full through further freeing, or explicitly. During direct compaction, it is useful to do the draining explicitly so that pages merge as soon as possible and compaction can detect success immediately and keep the latency impact at minimum. However the current implementation is far from ideal. Draining is done only in __alloc_pages_direct_compact(), after all zones were already compacted, and the decisions to continue or stop compaction in individual zones was done without the last batch of migrations being merged. It is also missing the draining of lru_add cache before the pcplists. This patch moves the draining for direct compaction into compact_zone(). It adds the missing lru_cache draining and uses the newly introduced single zone pcplists draining to reduce overhead and avoid impact on unrelated zones. Draining is only performed when it can actually lead to merging of a page of desired order (passed by cc->order). This means it is only done when migration occurred in the previously scanned cc->order aligned block(s) and the migration scanner is now pointing to the next cc->order aligned block. The patch has been tested with stress-highalloc benchmark from mmtests. Although overal allocation success rates of the benchmark were not affected, the number of detected compaction successes has doubled. This suggests that allocations were previously successful due to implicit merging caused by background activity, making a later allocation attempt succeed immediately, but not attributing the success to compaction. Since stress-highalloc always tries to allocate almost the whole memory, it cannot show the improvement in its reported success rate metric. However after this patch, compaction should detect success and terminate earlier, reducing the direct compaction latencies in a real scenario. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:34 +00:00
/* No more flushing until we migrate again */
last_migrated_pfn = 0;
mm, compaction: more focused lru and pcplists draining The goal of memory compaction is to create high-order freepages through page migration. Page migration however puts pages on the per-cpu lru_add cache, which is later flushed to per-cpu pcplists, and only after pcplists are drained the pages can actually merge. This can happen due to the per-cpu caches becoming full through further freeing, or explicitly. During direct compaction, it is useful to do the draining explicitly so that pages merge as soon as possible and compaction can detect success immediately and keep the latency impact at minimum. However the current implementation is far from ideal. Draining is done only in __alloc_pages_direct_compact(), after all zones were already compacted, and the decisions to continue or stop compaction in individual zones was done without the last batch of migrations being merged. It is also missing the draining of lru_add cache before the pcplists. This patch moves the draining for direct compaction into compact_zone(). It adds the missing lru_cache draining and uses the newly introduced single zone pcplists draining to reduce overhead and avoid impact on unrelated zones. Draining is only performed when it can actually lead to merging of a page of desired order (passed by cc->order). This means it is only done when migration occurred in the previously scanned cc->order aligned block(s) and the migration scanner is now pointing to the next cc->order aligned block. The patch has been tested with stress-highalloc benchmark from mmtests. Although overal allocation success rates of the benchmark were not affected, the number of detected compaction successes has doubled. This suggests that allocations were previously successful due to implicit merging caused by background activity, making a later allocation attempt succeed immediately, but not attributing the success to compaction. Since stress-highalloc always tries to allocate almost the whole memory, it cannot show the improvement in its reported success rate metric. However after this patch, compaction should detect success and terminate earlier, reducing the direct compaction latencies in a real scenario. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:34 +00:00
}
}
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
/* Stop if a page has been captured */
if (capc && capc->page) {
ret = COMPACT_SUCCESS;
break;
}
}
out:
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
/*
* Release free pages and update where the free scanner should restart,
* so we don't leave any returned pages behind in the next attempt.
*/
if (cc->nr_freepages > 0) {
unsigned long free_pfn = release_freepages(&cc->freepages);
cc->nr_freepages = 0;
VM_BUG_ON(free_pfn == 0);
/* The cached pfn is always the first in a pageblock */
free_pfn = pageblock_start_pfn(free_pfn);
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
/*
* Only go back, not forward. The cached pfn might have been
* already reset to zone end in compact_finished()
*/
if (free_pfn > cc->zone->compact_cached_free_pfn)
cc->zone->compact_cached_free_pfn = free_pfn;
mm, compaction: always update cached scanner positions Compaction caches the migration and free scanner positions between compaction invocations, so that the whole zone gets eventually scanned and there is no bias towards the initial scanner positions at the beginning/end of the zone. The cached positions are continuously updated as scanners progress and the updating stops as soon as a page is successfully isolated. The reasoning behind this is that a pageblock where isolation succeeded is likely to succeed again in near future and it should be worth revisiting it. However, the downside is that potentially many pages are rescanned without successful isolation. At worst, there might be a page where isolation from LRU succeeds but migration fails (potentially always). So upon encountering this page, cached position would always stop being updated for no good reason. It might have been useful to let such page be rescanned with sync compaction after async one failed, but this is now handled by caching scanner position for async and sync mode separately since commit 35979ef33931 ("mm, compaction: add per-zone migration pfn cache for async compaction"). After this patch, cached positions are updated unconditionally. In stress-highalloc benchmark, this has decreased the numbers of scanned pages by few percent, without affecting allocation success rates. To prevent free scanner from leaving free pages behind after they are returned due to page migration failure, the cached scanner pfn is changed to point to the pageblock of the returned free page with the highest pfn, before leaving compact_zone(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:31 +00:00
}
count_compact_events(COMPACTMIGRATE_SCANNED, cc->total_migrate_scanned);
count_compact_events(COMPACTFREE_SCANNED, cc->total_free_scanned);
trace_mm_compaction_end(start_pfn, cc->migrate_pfn,
cc->free_pfn, end_pfn, sync, ret);
mm: compaction: trace compaction begin and end The broad goal of the series is to improve allocation success rates for huge pages through memory compaction, while trying not to increase the compaction overhead. The original objective was to reintroduce capturing of high-order pages freed by the compaction, before they are split by concurrent activity. However, several bugs and opportunities for simple improvements were found in the current implementation, mostly through extra tracepoints (which are however too ugly for now to be considered for sending). The patches mostly deal with two mechanisms that reduce compaction overhead, which is caching the progress of migrate and free scanners, and marking pageblocks where isolation failed to be skipped during further scans. Patch 1 (from mgorman) adds tracepoints that allow calculate time spent in compaction and potentially debug scanner pfn values. Patch 2 encapsulates the some functionality for handling deferred compactions for better maintainability, without a functional change type is not determined without being actually needed. Patch 3 fixes a bug where cached scanner pfn's are sometimes reset only after they have been read to initialize a compaction run. Patch 4 fixes a bug where scanners meeting is sometimes not properly detected and can lead to multiple compaction attempts quitting early without doing any work. Patch 5 improves the chances of sync compaction to process pageblocks that async compaction has skipped due to being !MIGRATE_MOVABLE. Patch 6 improves the chances of sync direct compaction to actually do anything when called after async compaction fails during allocation slowpath. The impact of patches were validated using mmtests's stress-highalloc benchmark with mmtests's stress-highalloc benchmark on a x86_64 machine with 4GB memory. Due to instability of the results (mostly related to the bugs fixed by patches 2 and 3), 10 iterations were performed, taking min,mean,max values for success rates and mean values for time and vmstat-based metrics. First, the default GFP_HIGHUSER_MOVABLE allocations were tested with the patches stacked on top of v3.13-rc2. Patch 2 is OK to serve as baseline due to no functional changes in 1 and 2. Comments below. stress-highalloc 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-nothp 3-nothp 4-nothp 5-nothp 6-nothp Success 1 Min 9.00 ( 0.00%) 10.00 (-11.11%) 43.00 (-377.78%) 43.00 (-377.78%) 33.00 (-266.67%) Success 1 Mean 27.50 ( 0.00%) 25.30 ( 8.00%) 45.50 (-65.45%) 45.90 (-66.91%) 46.30 (-68.36%) Success 1 Max 36.00 ( 0.00%) 36.00 ( 0.00%) 47.00 (-30.56%) 48.00 (-33.33%) 52.00 (-44.44%) Success 2 Min 10.00 ( 0.00%) 8.00 ( 20.00%) 46.00 (-360.00%) 45.00 (-350.00%) 35.00 (-250.00%) Success 2 Mean 26.40 ( 0.00%) 23.50 ( 10.98%) 47.30 (-79.17%) 47.60 (-80.30%) 48.10 (-82.20%) Success 2 Max 34.00 ( 0.00%) 33.00 ( 2.94%) 48.00 (-41.18%) 50.00 (-47.06%) 54.00 (-58.82%) Success 3 Min 65.00 ( 0.00%) 63.00 ( 3.08%) 85.00 (-30.77%) 84.00 (-29.23%) 85.00 (-30.77%) Success 3 Mean 76.70 ( 0.00%) 70.50 ( 8.08%) 86.20 (-12.39%) 85.50 (-11.47%) 86.00 (-12.13%) Success 3 Max 87.00 ( 0.00%) 86.00 ( 1.15%) 88.00 ( -1.15%) 87.00 ( 0.00%) 87.00 ( 0.00%) 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-nothp 3-nothp 4-nothp 5-nothp 6-nothp User 6437.72 6459.76 5960.32 5974.55 6019.67 System 1049.65 1049.09 1029.32 1031.47 1032.31 Elapsed 1856.77 1874.48 1949.97 1994.22 1983.15 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-nothp 3-nothp 4-nothp 5-nothp 6-nothp Minor Faults 253952267 254581900 250030122 250507333 250157829 Major Faults 420 407 506 530 530 Swap Ins 4 9 9 6 6 Swap Outs 398 375 345 346 333 Direct pages scanned 197538 189017 298574 287019 299063 Kswapd pages scanned 1809843 1801308 1846674 1873184 1861089 Kswapd pages reclaimed 1806972 1798684 1844219 1870509 1858622 Direct pages reclaimed 197227 188829 298380 286822 298835 Kswapd efficiency 99% 99% 99% 99% 99% Kswapd velocity 953.382 970.449 952.243 934.569 922.286 Direct efficiency 99% 99% 99% 99% 99% Direct velocity 104.058 101.832 153.961 143.200 148.205 Percentage direct scans 9% 9% 13% 13% 13% Zone normal velocity 347.289 359.676 348.063 339.933 332.983 Zone dma32 velocity 710.151 712.605 758.140 737.835 737.507 Zone dma velocity 0.000 0.000 0.000 0.000 0.000 Page writes by reclaim 557.600 429.000 353.600 426.400 381.800 Page writes file 159 53 7 79 48 Page writes anon 398 375 345 346 333 Page reclaim immediate 825 644 411 575 420 Sector Reads 2781750 2769780 2878547 2939128 2910483 Sector Writes 12080843 12083351 12012892 12002132 12010745 Page rescued immediate 0 0 0 0 0 Slabs scanned 1575654 1545344 1778406 1786700 1794073 Direct inode steals 9657 10037 15795 14104 14645 Kswapd inode steals 46857 46335 50543 50716 51796 Kswapd skipped wait 0 0 0 0 0 THP fault alloc 97 91 81 71 77 THP collapse alloc 456 506 546 544 565 THP splits 6 5 5 4 4 THP fault fallback 0 1 0 0 0 THP collapse fail 14 14 12 13 12 Compaction stalls 1006 980 1537 1536 1548 Compaction success 303 284 562 559 578 Compaction failures 702 696 974 976 969 Page migrate success 1177325 1070077 3927538 3781870 3877057 Page migrate failure 0 0 0 0 0 Compaction pages isolated 2547248 2306457 8301218 8008500 8200674 Compaction migrate scanned 42290478 38832618 153961130 154143900 159141197 Compaction free scanned 89199429 79189151 356529027 351943166 356326727 Compaction cost 1566 1426 5312 5156 5294 NUMA PTE updates 0 0 0 0 0 NUMA hint faults 0 0 0 0 0 NUMA hint local faults 0 0 0 0 0 NUMA hint local percent 100 100 100 100 100 NUMA pages migrated 0 0 0 0 0 AutoNUMA cost 0 0 0 0 0 Observations: - The "Success 3" line is allocation success rate with system idle (phases 1 and 2 are with background interference). I used to get stable values around 85% with vanilla 3.11. The lower min and mean values came with 3.12. This was bisected to commit 81c0a2bb ("mm: page_alloc: fair zone allocator policy") As explained in comment for patch 3, I don't think the commit is wrong, but that it makes the effect of compaction bugs worse. From patch 3 onwards, the results are OK and match the 3.11 results. - Patch 4 also clearly helps phases 1 and 2, and exceeds any results I've seen with 3.11 (I didn't measure it that thoroughly then, but it was never above 40%). - Compaction cost and number of scanned pages is higher, especially due to patch 4. However, keep in mind that patches 3 and 4 fix existing bugs in the current design of compaction overhead mitigation, they do not change it. If overhead is found unacceptable, then it should be decreased differently (and consistently, not due to random conditions) than the current implementation does. In contrast, patches 5 and 6 (which are not strictly bug fixes) do not increase the overhead (but also not success rates). This might be a limitation of the stress-highalloc benchmark as it's quite uniform. Another set of results is when configuring stress-highalloc t allocate with similar flags as THP uses: (GFP_HIGHUSER_MOVABLE|__GFP_NOMEMALLOC|__GFP_NORETRY|__GFP_NO_KSWAPD) stress-highalloc 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-thp 3-thp 4-thp 5-thp 6-thp Success 1 Min 2.00 ( 0.00%) 7.00 (-250.00%) 18.00 (-800.00%) 19.00 (-850.00%) 26.00 (-1200.00%) Success 1 Mean 19.20 ( 0.00%) 17.80 ( 7.29%) 29.20 (-52.08%) 29.90 (-55.73%) 32.80 (-70.83%) Success 1 Max 27.00 ( 0.00%) 29.00 ( -7.41%) 35.00 (-29.63%) 36.00 (-33.33%) 37.00 (-37.04%) Success 2 Min 3.00 ( 0.00%) 8.00 (-166.67%) 21.00 (-600.00%) 21.00 (-600.00%) 32.00 (-966.67%) Success 2 Mean 19.30 ( 0.00%) 17.90 ( 7.25%) 32.20 (-66.84%) 32.60 (-68.91%) 35.70 (-84.97%) Success 2 Max 27.00 ( 0.00%) 30.00 (-11.11%) 36.00 (-33.33%) 37.00 (-37.04%) 39.00 (-44.44%) Success 3 Min 62.00 ( 0.00%) 62.00 ( 0.00%) 85.00 (-37.10%) 75.00 (-20.97%) 64.00 ( -3.23%) Success 3 Mean 66.30 ( 0.00%) 65.50 ( 1.21%) 85.60 (-29.11%) 83.40 (-25.79%) 83.50 (-25.94%) Success 3 Max 70.00 ( 0.00%) 69.00 ( 1.43%) 87.00 (-24.29%) 86.00 (-22.86%) 87.00 (-24.29%) 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-thp 3-thp 4-thp 5-thp 6-thp User 6547.93 6475.85 6265.54 6289.46 6189.96 System 1053.42 1047.28 1043.23 1042.73 1038.73 Elapsed 1835.43 1821.96 1908.67 1912.74 1956.38 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 3.13-rc2 2-thp 3-thp 4-thp 5-thp 6-thp Minor Faults 256805673 253106328 253222299 249830289 251184418 Major Faults 395 375 423 434 448 Swap Ins 12 10 10 12 9 Swap Outs 530 537 487 455 415 Direct pages scanned 71859 86046 153244 152764 190713 Kswapd pages scanned 1900994 1870240 1898012 1892864 1880520 Kswapd pages reclaimed 1897814 1867428 1894939 1890125 1877924 Direct pages reclaimed 71766 85908 153167 152643 190600 Kswapd efficiency 99% 99% 99% 99% 99% Kswapd velocity 1029.000 1067.782 1000.091 991.049 951.218 Direct efficiency 99% 99% 99% 99% 99% Direct velocity 38.897 49.127 80.747 79.983 96.468 Percentage direct scans 3% 4% 7% 7% 9% Zone normal velocity 351.377 372.494 348.910 341.689 335.310 Zone dma32 velocity 716.520 744.414 731.928 729.343 712.377 Zone dma velocity 0.000 0.000 0.000 0.000 0.000 Page writes by reclaim 669.300 604.000 545.700 538.900 429.900 Page writes file 138 66 58 83 14 Page writes anon 530 537 487 455 415 Page reclaim immediate 806 655 772 548 517 Sector Reads 2711956 2703239 2811602 2818248 2839459 Sector Writes 12163238 12018662 12038248 11954736 11994892 Page rescued immediate 0 0 0 0 0 Slabs scanned 1385088 1388364 1507968 1513292 1558656 Direct inode steals 1739 2564 4622 5496 6007 Kswapd inode steals 47461 46406 47804 48013 48466 Kswapd skipped wait 0 0 0 0 0 THP fault alloc 110 82 84 69 70 THP collapse alloc 445 482 467 462 539 THP splits 6 5 4 5 3 THP fault fallback 3 0 0 0 0 THP collapse fail 15 14 14 14 13 Compaction stalls 659 685 1033 1073 1111 Compaction success 222 225 410 427 456 Compaction failures 436 460 622 646 655 Page migrate success 446594 439978 1085640 1095062 1131716 Page migrate failure 0 0 0 0 0 Compaction pages isolated 1029475 1013490 2453074 2482698 2565400 Compaction migrate scanned 9955461 11344259 24375202 27978356 30494204 Compaction free scanned 27715272 28544654 80150615 82898631 85756132 Compaction cost 552 555 1344 1379 1436 NUMA PTE updates 0 0 0 0 0 NUMA hint faults 0 0 0 0 0 NUMA hint local faults 0 0 0 0 0 NUMA hint local percent 100 100 100 100 100 NUMA pages migrated 0 0 0 0 0 AutoNUMA cost 0 0 0 0 0 There are some differences from the previous results for THP-like allocations: - Here, the bad result for unpatched kernel in phase 3 is much more consistent to be between 65-70% and not related to the "regression" in 3.12. Still there is the improvement from patch 4 onwards, which brings it on par with simple GFP_HIGHUSER_MOVABLE allocations. - Compaction costs have increased, but nowhere near as much as the non-THP case. Again, the patches should be worth the gained determininsm. - Patches 5 and 6 somewhat increase the number of migrate-scanned pages. This is most likely due to __GFP_NO_KSWAPD flag, which means the cached pfn's and pageblock skip bits are not reset by kswapd that often (at least in phase 3 where no concurrent activity would wake up kswapd) and the patches thus help the sync-after-async compaction. It doesn't however show that the sync compaction would help so much with success rates, which can be again seen as a limitation of the benchmark scenario. This patch (of 6): Add two tracepoints for compaction begin and end of a zone. Using this it is possible to calculate how much time a workload is spending within compaction and potentially debug problems related to cached pfns for scanning. In combination with the direct reclaim and slab trace points it should be possible to estimate most allocation-related overhead for a workload. Signed-off-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-21 23:51:05 +00:00
return ret;
}
static enum compact_result compact_zone_order(struct zone *zone, int order,
mm, compaction: simplify contended compaction handling Async compaction detects contention either due to failing trylock on zone->lock or lru_lock, or by need_resched(). Since 1f9efdef4f3f ("mm, compaction: khugepaged should not give up due to need_resched()") the code got quite complicated to distinguish these two up to the __alloc_pages_slowpath() level, so different decisions could be taken for khugepaged allocations. After the recent changes, khugepaged allocations don't check for contended compaction anymore, so we again don't need to distinguish lock and sched contention, and simplify the current convoluted code a lot. However, I believe it's also possible to simplify even more and completely remove the check for contended compaction after the initial async compaction for costly orders, which was originally aimed at THP page fault allocations. There are several reasons why this can be done now: - with the new defaults, THP page faults no longer do reclaim/compaction at all, unless the system admin has overridden the default, or application has indicated via madvise that it can benefit from THP's. In both cases, it means that the potential extra latency is expected and worth the benefits. - even if reclaim/compaction proceeds after this patch where it previously wouldn't, the second compaction attempt is still async and will detect the contention and back off, if the contention persists - there are still heuristics like deferred compaction and pageblock skip bits in place that prevent excessive THP page fault latencies Link: http://lkml.kernel.org/r/20160721073614.24395-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:49:30 +00:00
gfp_t gfp_mask, enum compact_priority prio,
unsigned int alloc_flags, int highest_zoneidx,
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
struct page **capture)
{
enum compact_result ret;
struct compact_control cc = {
.order = order,
.search_order = order,
.gfp_mask = gfp_mask,
.zone = zone,
.mode = (prio == COMPACT_PRIO_ASYNC) ?
MIGRATE_ASYNC : MIGRATE_SYNC_LIGHT,
mm, compaction: pass classzone_idx and alloc_flags to watermark checking Compaction relies on zone watermark checks for decisions such as if it's worth to start compacting in compaction_suitable() or whether compaction should stop in compact_finished(). The watermark checks take classzone_idx and alloc_flags parameters, which are related to the memory allocation request. But from the context of compaction they are currently passed as 0, including the direct compaction which is invoked to satisfy the allocation request, and could therefore know the proper values. The lack of proper values can lead to mismatch between decisions taken during compaction and decisions related to the allocation request. Lack of proper classzone_idx value means that lowmem_reserve is not taken into account. This has manifested (during recent changes to deferred compaction) when DMA zone was used as fallback for preferred Normal zone. compaction_suitable() without proper classzone_idx would think that the watermarks are already satisfied, but watermark check in get_page_from_freelist() would fail. Because of this problem, deferring compaction has extra complexity that can be removed in the following patch. The issue (not confirmed in practice) with missing alloc_flags is opposite in nature. For allocations that include ALLOC_HIGH, ALLOC_HIGHER or ALLOC_CMA in alloc_flags (the last includes all MOVABLE allocations on CMA-enabled systems) the watermark checking in compaction with 0 passed will be stricter than in get_page_from_freelist(). In these cases compaction might be running for a longer time than is really needed. Another issue compaction_suitable() is that the check for "does the zone need compaction at all?" comes only after the check "does the zone have enough free free pages to succeed compaction". The latter considers extra pages for migration and can therefore in some situations fail and return COMPACT_SKIPPED, although the high-order allocation would succeed and we should return COMPACT_PARTIAL. This patch fixes these problems by adding alloc_flags and classzone_idx to struct compact_control and related functions involved in direct compaction and watermark checking. Where possible, all other callers of compaction_suitable() pass proper values where those are known. This is currently limited to classzone_idx, which is sometimes known in kswapd context. However, the direct reclaim callers should_continue_reclaim() and compaction_ready() do not currently know the proper values, so the coordination between reclaim and compaction may still not be as accurate as it could. This can be fixed later, if it's shown to be an issue. Additionaly the checks in compact_suitable() are reordered to address the second issue described above. The effect of this patch should be slightly better high-order allocation success rates and/or less compaction overhead, depending on the type of allocations and presence of CMA. It allows simplifying deferred compaction code in a followup patch. When testing with stress-highalloc, there was some slight improvement (which might be just due to variance) in success rates of non-THP-like allocations. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-12-10 23:43:22 +00:00
.alloc_flags = alloc_flags,
.highest_zoneidx = highest_zoneidx,
mm, kswapd: replace kswapd compaction with waking up kcompactd Similarly to direct reclaim/compaction, kswapd attempts to combine reclaim and compaction to attempt making memory allocation of given order available. The details differ from direct reclaim e.g. in having high watermark as a goal. The code involved in kswapd's reclaim/compaction decisions has evolved to be quite complex. Testing reveals that it doesn't actually work in at least one scenario, and closer inspection suggests that it could be greatly simplified without compromising on the goal (make high-order page available) or efficiency (don't reclaim too much). The simplification relieas of doing all compaction in kcompactd, which is simply woken up when high watermarks are reached by kswapd's reclaim. The scenario where kswapd compaction doesn't work was found with mmtests test stress-highalloc configured to attempt order-9 allocations without direct reclaim, just waking up kswapd. There was no compaction attempt from kswapd during the whole test. Some added instrumentation shows what happens: - balance_pgdat() sets end_zone to Normal, as it's not balanced - reclaim is attempted on DMA zone, which sets nr_attempted to 99, but it cannot reclaim anything, so sc.nr_reclaimed is 0 - for zones DMA32 and Normal, kswapd_shrink_zone uses testorder=0, so it merely checks if high watermarks were reached for base pages. This is true, so no reclaim is attempted. For DMA, testorder=0 wasn't used, as compaction_suitable() returned COMPACT_SKIPPED - even though the pgdat_needs_compaction flag wasn't set to false, no compaction happens due to the condition sc.nr_reclaimed > nr_attempted being false (as 0 < 99) - priority-- due to nr_reclaimed being 0, repeat until priority reaches 0 pgdat_balanced() is false as only the small zone DMA appears balanced (curiously in that check, watermark appears OK and compaction_suitable() returns COMPACT_PARTIAL, because a lower classzone_idx is used there) Now, even if it was decided that reclaim shouldn't be attempted on the DMA zone, the scenario would be the same, as (sc.nr_reclaimed=0 > nr_attempted=0) is also false. The condition really should use >= as the comment suggests. Then there is a mismatch in the check for setting pgdat_needs_compaction to false using low watermark, while the rest uses high watermark, and who knows what other subtlety. Hopefully this demonstrates that this is unsustainable. Luckily we can simplify this a lot. The reclaim/compaction decisions make sense for direct reclaim scenario, but in kswapd, our primary goal is to reach high watermark in order-0 pages. Afterwards we can attempt compaction just once. Unlike direct reclaim, we don't reclaim extra pages (over the high watermark), the current code already disallows it for good reasons. After this patch, we simply wake up kcompactd to process the pgdat, after we have either succeeded or failed to reach the high watermarks in kswapd, which goes to sleep. We pass kswapd's order and classzone_idx, so kcompactd can apply the same criteria to determine which zones are worth compacting. Note that we use the classzone_idx from wakeup_kswapd(), not balanced_classzone_idx which can include higher zones that kswapd tried to balance too, but didn't consider them in pgdat_balanced(). Since kswapd now cannot create high-order pages itself, we need to adjust how it determines the zones to be balanced. The key element here is adding a "highorder" parameter to zone_balanced, which, when set to false, makes it consider only order-0 watermark instead of the desired higher order (this was done previously by kswapd_shrink_zone(), but not elsewhere). This false is passed for example in pgdat_balanced(). Importantly, wakeup_kswapd() uses true to make sure kswapd and thus kcompactd are woken up for a high-order allocation failure. The last thing is to decide what to do with pageblock_skip bitmap handling. Compaction maintains a pageblock_skip bitmap to record pageblocks where isolation recently failed. This bitmap can be reset by three ways: 1) direct compaction is restarting after going through the full deferred cycle 2) kswapd goes to sleep, and some other direct compaction has previously finished scanning the whole zone and set zone->compact_blockskip_flush. Note that a successful direct compaction clears this flag. 3) compaction was invoked manually via trigger in /proc The case 2) is somewhat fuzzy to begin with, but after introducing kcompactd we should update it. The check for direct compaction in 1), and to set the flush flag in 2) use current_is_kswapd(), which doesn't work for kcompactd. Thus, this patch adds bool direct_compaction to compact_control to use in 2). For the case 1) we remove the check completely - unlike the former kswapd compaction, kcompactd does use the deferred compaction functionality, so flushing tied to restarting from deferred compaction makes sense here. Note that when kswapd goes to sleep, kcompactd is woken up, so it will see the flushed pageblock_skip bits. This is different from when the former kswapd compaction observed the bits and I believe it makes more sense. Kcompactd can afford to be more thorough than a direct compaction trying to limit allocation latency, or kswapd whose primary goal is to reclaim. For testing, I used stress-highalloc configured to do order-9 allocations with GFP_NOWAIT|__GFP_HIGH|__GFP_COMP, so they relied just on kswapd/kcompactd reclaim/compaction (the interfering kernel builds in phases 1 and 2 work as usual): stress-highalloc 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Success 1 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 1 Mean 1.40 ( 0.00%) 6.20 (-55.00%) Success 1 Max 2.00 ( 0.00%) 7.00 (-16.67%) Success 2 Min 1.00 ( 0.00%) 5.00 (-66.67%) Success 2 Mean 1.80 ( 0.00%) 6.40 (-52.38%) Success 2 Max 3.00 ( 0.00%) 7.00 (-16.67%) Success 3 Min 34.00 ( 0.00%) 62.00 ( 1.59%) Success 3 Mean 41.80 ( 0.00%) 63.80 ( 1.24%) Success 3 Max 53.00 ( 0.00%) 65.00 ( 2.99%) User 3166.67 3181.09 System 1153.37 1158.25 Elapsed 1768.53 1799.37 4.5-rc1+before 4.5-rc1+after -nodirect -nodirect Direct pages scanned 32938 32797 Kswapd pages scanned 2183166 2202613 Kswapd pages reclaimed 2152359 2143524 Direct pages reclaimed 32735 32545 Percentage direct scans 1% 1% THP fault alloc 579 612 THP collapse alloc 304 316 THP splits 0 0 THP fault fallback 793 778 THP collapse fail 11 16 Compaction stalls 1013 1007 Compaction success 92 67 Compaction failures 920 939 Page migrate success 238457 721374 Page migrate failure 23021 23469 Compaction pages isolated 504695 1479924 Compaction migrate scanned 661390 8812554 Compaction free scanned 13476658 84327916 Compaction cost 262 838 After this patch we see improvements in allocation success rate (especially for phase 3) along with increased compaction activity. The compaction stalls (direct compaction) in the interfering kernel builds (probably THP's) also decreased somewhat thanks to kcompactd activity, yet THP alloc successes improved a bit. Note that elapsed and user time isn't so useful for this benchmark, because of the background interference being unpredictable. It's just to quickly spot some major unexpected differences. System time is somewhat more useful and that didn't increase. Also (after adjusting mmtests' ftrace monitor): Time kswapd awake 2547781 2269241 Time kcompactd awake 0 119253 Time direct compacting 939937 557649 Time kswapd compacting 0 0 Time kcompactd compacting 0 119099 The decrease of overal time spent compacting appears to not match the increased compaction stats. I suspect the tasks get rescheduled and since the ftrace monitor doesn't see that, the reported time is wall time, not CPU time. But arguably direct compactors care about overall latency anyway, whether busy compacting or waiting for CPU doesn't matter. And that latency seems to almost halved. It's also interesting how much time kswapd spent awake just going through all the priorities and failing to even try compacting, over and over. We can also configure stress-highalloc to perform both direct reclaim/compaction and wakeup kswapd/kcompactd, by using GFP_KERNEL|__GFP_HIGH|__GFP_COMP: stress-highalloc 4.5-rc1+before 4.5-rc1+after -direct -direct Success 1 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 1 Mean 8.00 ( 0.00%) 10.00 (-19.05%) Success 1 Max 12.00 ( 0.00%) 11.00 ( 15.38%) Success 2 Min 4.00 ( 0.00%) 9.00 (-50.00%) Success 2 Mean 8.20 ( 0.00%) 10.00 (-16.28%) Success 2 Max 13.00 ( 0.00%) 11.00 ( 8.33%) Success 3 Min 75.00 ( 0.00%) 74.00 ( 1.33%) Success 3 Mean 75.60 ( 0.00%) 75.20 ( 0.53%) Success 3 Max 77.00 ( 0.00%) 76.00 ( 0.00%) User 3344.73 3246.04 System 1194.24 1172.29 Elapsed 1838.04 1836.76 4.5-rc1+before 4.5-rc1+after -direct -direct Direct pages scanned 125146 120966 Kswapd pages scanned 2119757 2135012 Kswapd pages reclaimed 2073183 2108388 Direct pages reclaimed 124909 120577 Percentage direct scans 5% 5% THP fault alloc 599 652 THP collapse alloc 323 354 THP splits 0 0 THP fault fallback 806 793 THP collapse fail 17 16 Compaction stalls 2457 2025 Compaction success 906 518 Compaction failures 1551 1507 Page migrate success 2031423 2360608 Page migrate failure 32845 40852 Compaction pages isolated 4129761 4802025 Compaction migrate scanned 11996712 21750613 Compaction free scanned 214970969 344372001 Compaction cost 2271 2694 In this scenario, this patch doesn't change the overall success rate as direct compaction already tries all it can. There's however significant reduction in direct compaction stalls (that is, the number of allocations that went into direct compaction). The number of successes (i.e. direct compaction stalls that ended up with successful allocation) is reduced by the same number. This means the offload to kcompactd is working as expected, and direct compaction is reduced either due to detecting contention, or compaction deferred by kcompactd. In the previous version of this patchset there was some apparent reduction of success rate, but the changes in this version (such as using sync compaction only), new baseline kernel, and/or averaging results from 5 executions (my bet), made this go away. Ftrace-based stats seem to roughly agree: Time kswapd awake 2532984 2326824 Time kcompactd awake 0 257916 Time direct compacting 864839 735130 Time kswapd compacting 0 0 Time kcompactd compacting 0 257585 Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:15 +00:00
.direct_compaction = true,
mm, compaction: add the ultimate direct compaction priority During reclaim/compaction loop, it's desirable to get a final answer from unsuccessful compaction so we can either fail the allocation or invoke the OOM killer. However, heuristics such as deferred compaction or pageblock skip bits can cause compaction to skip parts or whole zones and lead to premature OOM's, failures or excessive reclaim/compaction retries. To remedy this, we introduce a new direct compaction priority called COMPACT_PRIO_SYNC_FULL, which instructs direct compaction to: - ignore deferred compaction status for a zone - ignore pageblock skip hints - ignore cached scanner positions and scan the whole zone The new priority should get eventually picked up by should_compact_retry() and this should improve success rates for costly allocations using __GFP_REPEAT, such as hugetlbfs allocations, and reduce some corner-case OOM's for non-costly allocations. Link: http://lkml.kernel.org/r/20160810091226.6709-6-vbabka@suse.cz [vbabka@suse.cz: use the MIN_COMPACT_PRIORITY alias] Link: http://lkml.kernel.org/r/d443b884-87e7-1c93-8684-3a3a35759fb1@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:47 +00:00
.whole_zone = (prio == MIN_COMPACT_PRIORITY),
2016-10-08 00:00:37 +00:00
.ignore_skip_hint = (prio == MIN_COMPACT_PRIORITY),
.ignore_block_suitable = (prio == MIN_COMPACT_PRIORITY)
};
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
struct capture_control capc = {
.cc = &cc,
.page = NULL,
};
mm, compaction: make capture control handling safe wrt interrupts Hugh reports: "While stressing compaction, one run oopsed on NULL capc->cc in __free_one_page()'s task_capc(zone): compact_zone_order() had been interrupted, and a page was being freed in the return from interrupt. Though you would not expect it from the source, both gccs I was using (4.8.1 and 7.5.0) had chosen to compile compact_zone_order() with the ".cc = &cc" implemented by mov %rbx,-0xb0(%rbp) immediately before callq compact_zone - long after the "current->capture_control = &capc". An interrupt in between those finds capc->cc NULL (zeroed by an earlier rep stos). This could presumably be fixed by a barrier() before setting current->capture_control in compact_zone_order(); but would also need more care on return from compact_zone(), in order not to risk leaking a page captured by interrupt just before capture_control is reset. Maybe that is the preferable fix, but I felt safer for task_capc() to exclude the rather surprising possibility of capture at interrupt time" I have checked that gcc10 also behaves the same. The advantage of fix in compact_zone_order() is that we don't add another test in the page freeing hot path, and that it might prevent future problems if we stop exposing pointers to uninitialized structures in current task. So this patch implements the suggestion for compact_zone_order() with barrier() (and WRITE_ONCE() to prevent store tearing) for setting current->capture_control, and prevents page leaking with WRITE_ONCE/READ_ONCE in the proper order. Link: http://lkml.kernel.org/r/20200616082649.27173-1-vbabka@suse.cz Fixes: 5e1f0f098b46 ("mm, compaction: capture a page under direct compaction") Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Hugh Dickins <hughd@google.com> Suggested-by: Hugh Dickins <hughd@google.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Li Wang <liwang@redhat.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: <stable@vger.kernel.org> [5.1+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-06-26 03:29:24 +00:00
/*
* Make sure the structs are really initialized before we expose the
* capture control, in case we are interrupted and the interrupt handler
* frees a page.
*/
barrier();
WRITE_ONCE(current->capture_control, &capc);
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
ret = compact_zone(&cc, &capc);
VM_BUG_ON(!list_empty(&cc.freepages));
VM_BUG_ON(!list_empty(&cc.migratepages));
mm, compaction: make capture control handling safe wrt interrupts Hugh reports: "While stressing compaction, one run oopsed on NULL capc->cc in __free_one_page()'s task_capc(zone): compact_zone_order() had been interrupted, and a page was being freed in the return from interrupt. Though you would not expect it from the source, both gccs I was using (4.8.1 and 7.5.0) had chosen to compile compact_zone_order() with the ".cc = &cc" implemented by mov %rbx,-0xb0(%rbp) immediately before callq compact_zone - long after the "current->capture_control = &capc". An interrupt in between those finds capc->cc NULL (zeroed by an earlier rep stos). This could presumably be fixed by a barrier() before setting current->capture_control in compact_zone_order(); but would also need more care on return from compact_zone(), in order not to risk leaking a page captured by interrupt just before capture_control is reset. Maybe that is the preferable fix, but I felt safer for task_capc() to exclude the rather surprising possibility of capture at interrupt time" I have checked that gcc10 also behaves the same. The advantage of fix in compact_zone_order() is that we don't add another test in the page freeing hot path, and that it might prevent future problems if we stop exposing pointers to uninitialized structures in current task. So this patch implements the suggestion for compact_zone_order() with barrier() (and WRITE_ONCE() to prevent store tearing) for setting current->capture_control, and prevents page leaking with WRITE_ONCE/READ_ONCE in the proper order. Link: http://lkml.kernel.org/r/20200616082649.27173-1-vbabka@suse.cz Fixes: 5e1f0f098b46 ("mm, compaction: capture a page under direct compaction") Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Hugh Dickins <hughd@google.com> Suggested-by: Hugh Dickins <hughd@google.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Li Wang <liwang@redhat.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: <stable@vger.kernel.org> [5.1+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-06-26 03:29:24 +00:00
/*
* Make sure we hide capture control first before we read the captured
* page pointer, otherwise an interrupt could free and capture a page
* and we would leak it.
*/
WRITE_ONCE(current->capture_control, NULL);
*capture = READ_ONCE(capc.page);
/*
* Technically, it is also possible that compaction is skipped but
* the page is still captured out of luck(IRQ came and freed the page).
* Returning COMPACT_SUCCESS in such cases helps in properly accounting
* the COMPACT[STALL|FAIL] when compaction is skipped.
*/
if (*capture)
ret = COMPACT_SUCCESS;
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
return ret;
}
int sysctl_extfrag_threshold = 500;
/**
* try_to_compact_pages - Direct compact to satisfy a high-order allocation
* @gfp_mask: The GFP mask of the current allocation
* @order: The order of the current allocation
* @alloc_flags: The allocation flags of the current allocation
* @ac: The context of current allocation
* @prio: Determines how hard direct compaction should try to succeed
mm, compaction: fully assume capture is not NULL in compact_zone_order() Dan reports: The patch 5e1f0f098b46: "mm, compaction: capture a page under direct compaction" from Mar 5, 2019, leads to the following Smatch complaint: mm/compaction.c:2321 compact_zone_order() error: we previously assumed 'capture' could be null (see line 2313) mm/compaction.c 2288 static enum compact_result compact_zone_order(struct zone *zone, int order, 2289 gfp_t gfp_mask, enum compact_priority prio, 2290 unsigned int alloc_flags, int classzone_idx, 2291 struct page **capture) ^^^^^^^ 2313 if (capture) ^^^^^^^ Check for NULL 2314 current->capture_control = &capc; 2315 2316 ret = compact_zone(&cc, &capc); 2317 2318 VM_BUG_ON(!list_empty(&cc.freepages)); 2319 VM_BUG_ON(!list_empty(&cc.migratepages)); 2320 2321 *capture = capc.page; ^^^^^^^^ Unchecked dereference. 2322 current->capture_control = NULL; 2323 In practice this is not an issue, as the only caller path passes non-NULL capture: __alloc_pages_direct_compact() struct page *page = NULL; try_to_compact_pages(capture = &page); compact_zone_order(capture = capture); So let's remove the unnecessary check, which should also make Smatch happy. Fixes: 5e1f0f098b46 ("mm, compaction: capture a page under direct compaction") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Suggested-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Mel Gorman <mgorman@techsingularity.net> Link: http://lkml.kernel.org/r/18b0df3c-0589-d96c-23fa-040798fee187@suse.cz Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-04-02 04:10:35 +00:00
* @capture: Pointer to free page created by compaction will be stored here
*
* This is the main entry point for direct page compaction.
*/
enum compact_result try_to_compact_pages(gfp_t gfp_mask, unsigned int order,
unsigned int alloc_flags, const struct alloc_context *ac,
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
enum compact_priority prio, struct page **capture)
{
int may_perform_io = gfp_mask & __GFP_IO;
struct zoneref *z;
struct zone *zone;
enum compact_result rc = COMPACT_SKIPPED;
mm, compaction: allow compaction for GFP_NOFS requests compaction has been disabled for GFP_NOFS and GFP_NOIO requests since the direct compaction was introduced by commit 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). The main reason is that the migration of page cache pages might recurse back to fs/io layer and we could potentially deadlock. This is overly conservative because all the anonymous memory is migrateable in the GFP_NOFS context just fine. This might be a large portion of the memory in many/most workkloads. Remove the GFP_NOFS restriction and make sure that we skip all fs pages (those with a mapping) while isolating pages to be migrated. We cannot consider clean fs pages because they might need a metadata update so only isolate pages without any mapping for nofs requests. The effect of this patch will be probably very limited in many/most workloads because higher order GFP_NOFS requests are quite rare, although different configurations might lead to very different results. David Chinner has mentioned a heavy metadata workload with 64kB block which to quote him: : Unfortunately, there was an era of cargo cult configuration tweaks in the : Ceph community that has resulted in a large number of production machines : with XFS filesystems configured this way. And a lot of them store large : numbers of small files and run under significant sustained memory : pressure. : : I slowly working towards getting rid of these high order allocations and : replacing them with the equivalent number of single page allocations, but : I haven't got that (complex) change working yet. We can do the following to simulate that workload: $ mkfs.xfs -f -n size=64k <dev> $ mount <dev> /mnt/scratch $ time ./fs_mark -D 10000 -S0 -n 100000 -s 0 -L 32 \ -d /mnt/scratch/0 -d /mnt/scratch/1 \ -d /mnt/scratch/2 -d /mnt/scratch/3 \ -d /mnt/scratch/4 -d /mnt/scratch/5 \ -d /mnt/scratch/6 -d /mnt/scratch/7 \ -d /mnt/scratch/8 -d /mnt/scratch/9 \ -d /mnt/scratch/10 -d /mnt/scratch/11 \ -d /mnt/scratch/12 -d /mnt/scratch/13 \ -d /mnt/scratch/14 -d /mnt/scratch/15 and indeed is hammers the system with many high order GFP_NOFS requests as per a simle tracepoint during the load: $ echo '!(gfp_flags & 0x80) && (gfp_flags &0x400000)' > $TRACE_MNT/events/kmem/mm_page_alloc/filter I am getting 5287609 order=0 37 order=1 1594905 order=2 3048439 order=3 6699207 order=4 66645 order=5 My testing was done in a kvm guest so performance numbers should be taken with a grain of salt but there seems to be a difference when the patch is applied: * Original kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4300.1 20745838 3 3200000 0 4239.9 23849857 5 4800000 0 4243.4 25939543 6 6400000 0 4248.4 19514050 8 8000000 0 4262.1 20796169 9 9600000 0 4257.6 21288675 11 11200000 0 4259.7 19375120 13 12800000 0 4220.7 22734141 14 14400000 0 4238.5 31936458 16 16000000 0 4231.5 23409901 18 17600000 0 4045.3 23577700 19 19200000 0 2783.4 58299526 21 20800000 0 2678.2 40616302 23 22400000 0 2693.5 83973996 and xfs complaining about memory allocation not making progress [ 2304.372647] XFS: fs_mark(3289) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) [ 2304.443323] XFS: fs_mark(3285) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) [ 4796.772477] XFS: fs_mark(3424) possible memory allocation deadlock size 46936 in kmem_alloc (mode:0x2408240) [ 4796.775329] XFS: fs_mark(3423) possible memory allocation deadlock size 51416 in kmem_alloc (mode:0x2408240) [ 4797.388808] XFS: fs_mark(3424) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) * Patched kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4289.1 19243934 3 3200000 0 4241.6 32828865 5 4800000 0 4248.7 32884693 6 6400000 0 4314.4 19608921 8 8000000 0 4269.9 24953292 9 9600000 0 4270.7 33235572 11 11200000 0 4346.4 40817101 13 12800000 0 4285.3 29972397 14 14400000 0 4297.2 20539765 16 16000000 0 4219.6 18596767 18 17600000 0 4273.8 49611187 19 19200000 0 4300.4 27944451 21 20800000 0 4270.6 22324585 22 22400000 0 4317.6 22650382 24 24000000 0 4065.2 22297964 So the dropdown at Count 19200000 didn't happen and there was only a single warning about allocation not making progress [ 3063.815003] XFS: fs_mark(3272) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) This suggests that the patch has helped even though there is not all that much of anonymous memory as the workload mostly generates fs metadata. I assume the success rate would be higher with more anonymous memory which should be the case in many workloads. [akpm@linux-foundation.org: fix comment] Link: http://lkml.kernel.org/r/20161012114721.31853-1-mhocko@kernel.org Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 23:04:07 +00:00
/*
* Check if the GFP flags allow compaction - GFP_NOIO is really
* tricky context because the migration might require IO
*/
if (!may_perform_io)
mm, compaction: defer each zone individually instead of preferred zone When direct sync compaction is often unsuccessful, it may become deferred for some time to avoid further useless attempts, both sync and async. Successful high-order allocations un-defer compaction, while further unsuccessful compaction attempts prolong the compaction deferred period. Currently the checking and setting deferred status is performed only on the preferred zone of the allocation that invoked direct compaction. But compaction itself is attempted on all eligible zones in the zonelist, so the behavior is suboptimal and may lead both to scenarios where 1) compaction is attempted uselessly, or 2) where it's not attempted despite good chances of succeeding, as shown on the examples below: 1) A direct compaction with Normal preferred zone failed and set deferred compaction for the Normal zone. Another unrelated direct compaction with DMA32 as preferred zone will attempt to compact DMA32 zone even though the first compaction attempt also included DMA32 zone. In another scenario, compaction with Normal preferred zone failed to compact Normal zone, but succeeded in the DMA32 zone, so it will not defer compaction. In the next attempt, it will try Normal zone which will fail again, instead of skipping Normal zone and trying DMA32 directly. 2) Kswapd will balance DMA32 zone and reset defer status based on watermarks looking good. A direct compaction with preferred Normal zone will skip compaction of all zones including DMA32 because Normal was still deferred. The allocation might have succeeded in DMA32, but won't. This patch makes compaction deferring work on individual zone basis instead of preferred zone. For each zone, it checks compaction_deferred() to decide if the zone should be skipped. If watermarks fail after compacting the zone, defer_compaction() is called. The zone where watermarks passed can still be deferred when the allocation attempt is unsuccessful. When allocation is successful, compaction_defer_reset() is called for the zone containing the allocated page. This approach should approximate calling defer_compaction() only on zones where compaction was attempted and did not yield allocated page. There might be corner cases but that is inevitable as long as the decision to stop compacting dues not guarantee that a page will be allocated. Due to a new COMPACT_DEFERRED return value, some functions relying implicitly on COMPACT_SKIPPED = 0 had to be updated, with comments made more accurate. The did_some_progress output parameter of __alloc_pages_direct_compact() is removed completely, as the caller actually does not use it after compaction sets it - it is only considered when direct reclaim sets it. During testing on a two-node machine with a single very small Normal zone on node 1, this patch has improved success rates in stress-highalloc mmtests benchmark. The success here were previously made worse by commit 3a025760fc15 ("mm: page_alloc: spill to remote nodes before waking kswapd") as kswapd was no longer resetting often enough the deferred compaction for the Normal zone, and DMA32 zones on both nodes were thus not considered for compaction. On different machine, success rates were improved with __GFP_NO_KSWAPD allocations. [akpm@linux-foundation.org: fix CONFIG_COMPACTION=n build] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Minchan Kim <minchan@kernel.org> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:02 +00:00
return COMPACT_SKIPPED;
trace_mm_compaction_try_to_compact_pages(order, gfp_mask, prio);
/* Compact each zone in the list */
for_each_zone_zonelist_nodemask(zone, z, ac->zonelist,
ac->highest_zoneidx, ac->nodemask) {
enum compact_result status;
mm, compaction: add the ultimate direct compaction priority During reclaim/compaction loop, it's desirable to get a final answer from unsuccessful compaction so we can either fail the allocation or invoke the OOM killer. However, heuristics such as deferred compaction or pageblock skip bits can cause compaction to skip parts or whole zones and lead to premature OOM's, failures or excessive reclaim/compaction retries. To remedy this, we introduce a new direct compaction priority called COMPACT_PRIO_SYNC_FULL, which instructs direct compaction to: - ignore deferred compaction status for a zone - ignore pageblock skip hints - ignore cached scanner positions and scan the whole zone The new priority should get eventually picked up by should_compact_retry() and this should improve success rates for costly allocations using __GFP_REPEAT, such as hugetlbfs allocations, and reduce some corner-case OOM's for non-costly allocations. Link: http://lkml.kernel.org/r/20160810091226.6709-6-vbabka@suse.cz [vbabka@suse.cz: use the MIN_COMPACT_PRIORITY alias] Link: http://lkml.kernel.org/r/d443b884-87e7-1c93-8684-3a3a35759fb1@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:47 +00:00
if (prio > MIN_COMPACT_PRIORITY
&& compaction_deferred(zone, order)) {
rc = max_t(enum compact_result, COMPACT_DEFERRED, rc);
mm, compaction: defer each zone individually instead of preferred zone When direct sync compaction is often unsuccessful, it may become deferred for some time to avoid further useless attempts, both sync and async. Successful high-order allocations un-defer compaction, while further unsuccessful compaction attempts prolong the compaction deferred period. Currently the checking and setting deferred status is performed only on the preferred zone of the allocation that invoked direct compaction. But compaction itself is attempted on all eligible zones in the zonelist, so the behavior is suboptimal and may lead both to scenarios where 1) compaction is attempted uselessly, or 2) where it's not attempted despite good chances of succeeding, as shown on the examples below: 1) A direct compaction with Normal preferred zone failed and set deferred compaction for the Normal zone. Another unrelated direct compaction with DMA32 as preferred zone will attempt to compact DMA32 zone even though the first compaction attempt also included DMA32 zone. In another scenario, compaction with Normal preferred zone failed to compact Normal zone, but succeeded in the DMA32 zone, so it will not defer compaction. In the next attempt, it will try Normal zone which will fail again, instead of skipping Normal zone and trying DMA32 directly. 2) Kswapd will balance DMA32 zone and reset defer status based on watermarks looking good. A direct compaction with preferred Normal zone will skip compaction of all zones including DMA32 because Normal was still deferred. The allocation might have succeeded in DMA32, but won't. This patch makes compaction deferring work on individual zone basis instead of preferred zone. For each zone, it checks compaction_deferred() to decide if the zone should be skipped. If watermarks fail after compacting the zone, defer_compaction() is called. The zone where watermarks passed can still be deferred when the allocation attempt is unsuccessful. When allocation is successful, compaction_defer_reset() is called for the zone containing the allocated page. This approach should approximate calling defer_compaction() only on zones where compaction was attempted and did not yield allocated page. There might be corner cases but that is inevitable as long as the decision to stop compacting dues not guarantee that a page will be allocated. Due to a new COMPACT_DEFERRED return value, some functions relying implicitly on COMPACT_SKIPPED = 0 had to be updated, with comments made more accurate. The did_some_progress output parameter of __alloc_pages_direct_compact() is removed completely, as the caller actually does not use it after compaction sets it - it is only considered when direct reclaim sets it. During testing on a two-node machine with a single very small Normal zone on node 1, this patch has improved success rates in stress-highalloc mmtests benchmark. The success here were previously made worse by commit 3a025760fc15 ("mm: page_alloc: spill to remote nodes before waking kswapd") as kswapd was no longer resetting often enough the deferred compaction for the Normal zone, and DMA32 zones on both nodes were thus not considered for compaction. On different machine, success rates were improved with __GFP_NO_KSWAPD allocations. [akpm@linux-foundation.org: fix CONFIG_COMPACTION=n build] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Minchan Kim <minchan@kernel.org> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:02 +00:00
continue;
}
mm, compaction: defer each zone individually instead of preferred zone When direct sync compaction is often unsuccessful, it may become deferred for some time to avoid further useless attempts, both sync and async. Successful high-order allocations un-defer compaction, while further unsuccessful compaction attempts prolong the compaction deferred period. Currently the checking and setting deferred status is performed only on the preferred zone of the allocation that invoked direct compaction. But compaction itself is attempted on all eligible zones in the zonelist, so the behavior is suboptimal and may lead both to scenarios where 1) compaction is attempted uselessly, or 2) where it's not attempted despite good chances of succeeding, as shown on the examples below: 1) A direct compaction with Normal preferred zone failed and set deferred compaction for the Normal zone. Another unrelated direct compaction with DMA32 as preferred zone will attempt to compact DMA32 zone even though the first compaction attempt also included DMA32 zone. In another scenario, compaction with Normal preferred zone failed to compact Normal zone, but succeeded in the DMA32 zone, so it will not defer compaction. In the next attempt, it will try Normal zone which will fail again, instead of skipping Normal zone and trying DMA32 directly. 2) Kswapd will balance DMA32 zone and reset defer status based on watermarks looking good. A direct compaction with preferred Normal zone will skip compaction of all zones including DMA32 because Normal was still deferred. The allocation might have succeeded in DMA32, but won't. This patch makes compaction deferring work on individual zone basis instead of preferred zone. For each zone, it checks compaction_deferred() to decide if the zone should be skipped. If watermarks fail after compacting the zone, defer_compaction() is called. The zone where watermarks passed can still be deferred when the allocation attempt is unsuccessful. When allocation is successful, compaction_defer_reset() is called for the zone containing the allocated page. This approach should approximate calling defer_compaction() only on zones where compaction was attempted and did not yield allocated page. There might be corner cases but that is inevitable as long as the decision to stop compacting dues not guarantee that a page will be allocated. Due to a new COMPACT_DEFERRED return value, some functions relying implicitly on COMPACT_SKIPPED = 0 had to be updated, with comments made more accurate. The did_some_progress output parameter of __alloc_pages_direct_compact() is removed completely, as the caller actually does not use it after compaction sets it - it is only considered when direct reclaim sets it. During testing on a two-node machine with a single very small Normal zone on node 1, this patch has improved success rates in stress-highalloc mmtests benchmark. The success here were previously made worse by commit 3a025760fc15 ("mm: page_alloc: spill to remote nodes before waking kswapd") as kswapd was no longer resetting often enough the deferred compaction for the Normal zone, and DMA32 zones on both nodes were thus not considered for compaction. On different machine, success rates were improved with __GFP_NO_KSWAPD allocations. [akpm@linux-foundation.org: fix CONFIG_COMPACTION=n build] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Minchan Kim <minchan@kernel.org> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:02 +00:00
status = compact_zone_order(zone, order, gfp_mask, prio,
alloc_flags, ac->highest_zoneidx, capture);
rc = max(status, rc);
mm, compaction: don't recheck watermarks after COMPACT_SUCCESS Joonsoo has reminded me that in a later patch changing watermark checks throughout compaction I forgot to update checks in try_to_compact_pages() and compactd_do_work(). Closer inspection however shows that they are redundant now in the success case, because compact_zone() now reliably reports this with COMPACT_SUCCESS. So effectively the checks just repeat (a subset) of checks that have just passed. So instead of checking watermarks again, just test the return value. Note it's also possible that compaction would declare failure e.g. because its find_suitable_fallback() is more strict than simple watermark check, and then the watermark check we are removing would then still succeed. After this patch this is not possible and it's arguably better, because for long-term fragmentation avoidance we should rather try a different zone than allocate with the unsuitable fallback. If compaction of all zones fail and the allocation is important enough, it will retry and succeed anyway. Also remove the stray "bool success" variable from kcompactd_do_work(). Link: http://lkml.kernel.org/r/20160810091226.6709-5-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:44 +00:00
/* The allocation should succeed, stop compacting */
if (status == COMPACT_SUCCESS) {
mm, compaction: defer each zone individually instead of preferred zone When direct sync compaction is often unsuccessful, it may become deferred for some time to avoid further useless attempts, both sync and async. Successful high-order allocations un-defer compaction, while further unsuccessful compaction attempts prolong the compaction deferred period. Currently the checking and setting deferred status is performed only on the preferred zone of the allocation that invoked direct compaction. But compaction itself is attempted on all eligible zones in the zonelist, so the behavior is suboptimal and may lead both to scenarios where 1) compaction is attempted uselessly, or 2) where it's not attempted despite good chances of succeeding, as shown on the examples below: 1) A direct compaction with Normal preferred zone failed and set deferred compaction for the Normal zone. Another unrelated direct compaction with DMA32 as preferred zone will attempt to compact DMA32 zone even though the first compaction attempt also included DMA32 zone. In another scenario, compaction with Normal preferred zone failed to compact Normal zone, but succeeded in the DMA32 zone, so it will not defer compaction. In the next attempt, it will try Normal zone which will fail again, instead of skipping Normal zone and trying DMA32 directly. 2) Kswapd will balance DMA32 zone and reset defer status based on watermarks looking good. A direct compaction with preferred Normal zone will skip compaction of all zones including DMA32 because Normal was still deferred. The allocation might have succeeded in DMA32, but won't. This patch makes compaction deferring work on individual zone basis instead of preferred zone. For each zone, it checks compaction_deferred() to decide if the zone should be skipped. If watermarks fail after compacting the zone, defer_compaction() is called. The zone where watermarks passed can still be deferred when the allocation attempt is unsuccessful. When allocation is successful, compaction_defer_reset() is called for the zone containing the allocated page. This approach should approximate calling defer_compaction() only on zones where compaction was attempted and did not yield allocated page. There might be corner cases but that is inevitable as long as the decision to stop compacting dues not guarantee that a page will be allocated. Due to a new COMPACT_DEFERRED return value, some functions relying implicitly on COMPACT_SKIPPED = 0 had to be updated, with comments made more accurate. The did_some_progress output parameter of __alloc_pages_direct_compact() is removed completely, as the caller actually does not use it after compaction sets it - it is only considered when direct reclaim sets it. During testing on a two-node machine with a single very small Normal zone on node 1, this patch has improved success rates in stress-highalloc mmtests benchmark. The success here were previously made worse by commit 3a025760fc15 ("mm: page_alloc: spill to remote nodes before waking kswapd") as kswapd was no longer resetting often enough the deferred compaction for the Normal zone, and DMA32 zones on both nodes were thus not considered for compaction. On different machine, success rates were improved with __GFP_NO_KSWAPD allocations. [akpm@linux-foundation.org: fix CONFIG_COMPACTION=n build] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Minchan Kim <minchan@kernel.org> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:02 +00:00
/*
* We think the allocation will succeed in this zone,
* but it is not certain, hence the false. The caller
* will repeat this with true if allocation indeed
* succeeds in this zone.
*/
compaction_defer_reset(zone, order, false);
mm, compaction: khugepaged should not give up due to need_resched() Async compaction aborts when it detects zone lock contention or need_resched() is true. David Rientjes has reported that in practice, most direct async compactions for THP allocation abort due to need_resched(). This means that a second direct compaction is never attempted, which might be OK for a page fault, but khugepaged is intended to attempt a sync compaction in such case and in these cases it won't. This patch replaces "bool contended" in compact_control with an int that distinguishes between aborting due to need_resched() and aborting due to lock contention. This allows propagating the abort through all compaction functions as before, but passing the abort reason up to __alloc_pages_slowpath() which decides when to continue with direct reclaim and another compaction attempt. Another problem is that try_to_compact_pages() did not act upon the reported contention (both need_resched() or lock contention) immediately and would proceed with another zone from the zonelist. When need_resched() is true, that means initializing another zone compaction, only to check again need_resched() in isolate_migratepages() and aborting. For zone lock contention, the unintended consequence is that the lock contended status reported back to the allocator is detrmined from the last zone where compaction was attempted, which is rather arbitrary. This patch fixes the problem in the following way: - async compaction of a zone aborting due to need_resched() or fatal signal pending means that further zones should not be tried. We report COMPACT_CONTENDED_SCHED to the allocator. - aborting zone compaction due to lock contention means we can still try another zone, since it has different set of locks. We report back COMPACT_CONTENDED_LOCK only if *all* zones where compaction was attempted, it was aborted due to lock contention. As a result of these fixes, khugepaged will proceed with second sync compaction as intended, when the preceding async compaction aborted due to need_resched(). Page fault compactions aborting due to need_resched() will spare some cycles previously wasted by initializing another zone compaction only to abort again. Lock contention will be reported only when compaction in all zones aborted due to lock contention, and therefore it's not a good idea to try again after reclaim. In stress-highalloc from mmtests configured to use __GFP_NO_KSWAPD, this has improved number of THP collapse allocations by 10%, which shows positive effect on khugepaged. The benchmark's success rates are unchanged as it is not recognized as khugepaged. Numbers of compact_stall and compact_fail events have however decreased by 20%, with compact_success still a bit improved, which is good. With benchmark configured not to use __GFP_NO_KSWAPD, there is 6% improvement in THP collapse allocations, and only slight improvement in stalls and failures. [akpm@linux-foundation.org: fix warnings] Reported-by: David Rientjes <rientjes@google.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:14 +00:00
mm, compaction: simplify contended compaction handling Async compaction detects contention either due to failing trylock on zone->lock or lru_lock, or by need_resched(). Since 1f9efdef4f3f ("mm, compaction: khugepaged should not give up due to need_resched()") the code got quite complicated to distinguish these two up to the __alloc_pages_slowpath() level, so different decisions could be taken for khugepaged allocations. After the recent changes, khugepaged allocations don't check for contended compaction anymore, so we again don't need to distinguish lock and sched contention, and simplify the current convoluted code a lot. However, I believe it's also possible to simplify even more and completely remove the check for contended compaction after the initial async compaction for costly orders, which was originally aimed at THP page fault allocations. There are several reasons why this can be done now: - with the new defaults, THP page faults no longer do reclaim/compaction at all, unless the system admin has overridden the default, or application has indicated via madvise that it can benefit from THP's. In both cases, it means that the potential extra latency is expected and worth the benefits. - even if reclaim/compaction proceeds after this patch where it previously wouldn't, the second compaction attempt is still async and will detect the contention and back off, if the contention persists - there are still heuristics like deferred compaction and pageblock skip bits in place that prevent excessive THP page fault latencies Link: http://lkml.kernel.org/r/20160721073614.24395-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:49:30 +00:00
break;
mm, compaction: khugepaged should not give up due to need_resched() Async compaction aborts when it detects zone lock contention or need_resched() is true. David Rientjes has reported that in practice, most direct async compactions for THP allocation abort due to need_resched(). This means that a second direct compaction is never attempted, which might be OK for a page fault, but khugepaged is intended to attempt a sync compaction in such case and in these cases it won't. This patch replaces "bool contended" in compact_control with an int that distinguishes between aborting due to need_resched() and aborting due to lock contention. This allows propagating the abort through all compaction functions as before, but passing the abort reason up to __alloc_pages_slowpath() which decides when to continue with direct reclaim and another compaction attempt. Another problem is that try_to_compact_pages() did not act upon the reported contention (both need_resched() or lock contention) immediately and would proceed with another zone from the zonelist. When need_resched() is true, that means initializing another zone compaction, only to check again need_resched() in isolate_migratepages() and aborting. For zone lock contention, the unintended consequence is that the lock contended status reported back to the allocator is detrmined from the last zone where compaction was attempted, which is rather arbitrary. This patch fixes the problem in the following way: - async compaction of a zone aborting due to need_resched() or fatal signal pending means that further zones should not be tried. We report COMPACT_CONTENDED_SCHED to the allocator. - aborting zone compaction due to lock contention means we can still try another zone, since it has different set of locks. We report back COMPACT_CONTENDED_LOCK only if *all* zones where compaction was attempted, it was aborted due to lock contention. As a result of these fixes, khugepaged will proceed with second sync compaction as intended, when the preceding async compaction aborted due to need_resched(). Page fault compactions aborting due to need_resched() will spare some cycles previously wasted by initializing another zone compaction only to abort again. Lock contention will be reported only when compaction in all zones aborted due to lock contention, and therefore it's not a good idea to try again after reclaim. In stress-highalloc from mmtests configured to use __GFP_NO_KSWAPD, this has improved number of THP collapse allocations by 10%, which shows positive effect on khugepaged. The benchmark's success rates are unchanged as it is not recognized as khugepaged. Numbers of compact_stall and compact_fail events have however decreased by 20%, with compact_success still a bit improved, which is good. With benchmark configured not to use __GFP_NO_KSWAPD, there is 6% improvement in THP collapse allocations, and only slight improvement in stalls and failures. [akpm@linux-foundation.org: fix warnings] Reported-by: David Rientjes <rientjes@google.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:14 +00:00
}
if (prio != COMPACT_PRIO_ASYNC && (status == COMPACT_COMPLETE ||
mm, compaction: simplify contended compaction handling Async compaction detects contention either due to failing trylock on zone->lock or lru_lock, or by need_resched(). Since 1f9efdef4f3f ("mm, compaction: khugepaged should not give up due to need_resched()") the code got quite complicated to distinguish these two up to the __alloc_pages_slowpath() level, so different decisions could be taken for khugepaged allocations. After the recent changes, khugepaged allocations don't check for contended compaction anymore, so we again don't need to distinguish lock and sched contention, and simplify the current convoluted code a lot. However, I believe it's also possible to simplify even more and completely remove the check for contended compaction after the initial async compaction for costly orders, which was originally aimed at THP page fault allocations. There are several reasons why this can be done now: - with the new defaults, THP page faults no longer do reclaim/compaction at all, unless the system admin has overridden the default, or application has indicated via madvise that it can benefit from THP's. In both cases, it means that the potential extra latency is expected and worth the benefits. - even if reclaim/compaction proceeds after this patch where it previously wouldn't, the second compaction attempt is still async and will detect the contention and back off, if the contention persists - there are still heuristics like deferred compaction and pageblock skip bits in place that prevent excessive THP page fault latencies Link: http://lkml.kernel.org/r/20160721073614.24395-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:49:30 +00:00
status == COMPACT_PARTIAL_SKIPPED))
mm, compaction: defer each zone individually instead of preferred zone When direct sync compaction is often unsuccessful, it may become deferred for some time to avoid further useless attempts, both sync and async. Successful high-order allocations un-defer compaction, while further unsuccessful compaction attempts prolong the compaction deferred period. Currently the checking and setting deferred status is performed only on the preferred zone of the allocation that invoked direct compaction. But compaction itself is attempted on all eligible zones in the zonelist, so the behavior is suboptimal and may lead both to scenarios where 1) compaction is attempted uselessly, or 2) where it's not attempted despite good chances of succeeding, as shown on the examples below: 1) A direct compaction with Normal preferred zone failed and set deferred compaction for the Normal zone. Another unrelated direct compaction with DMA32 as preferred zone will attempt to compact DMA32 zone even though the first compaction attempt also included DMA32 zone. In another scenario, compaction with Normal preferred zone failed to compact Normal zone, but succeeded in the DMA32 zone, so it will not defer compaction. In the next attempt, it will try Normal zone which will fail again, instead of skipping Normal zone and trying DMA32 directly. 2) Kswapd will balance DMA32 zone and reset defer status based on watermarks looking good. A direct compaction with preferred Normal zone will skip compaction of all zones including DMA32 because Normal was still deferred. The allocation might have succeeded in DMA32, but won't. This patch makes compaction deferring work on individual zone basis instead of preferred zone. For each zone, it checks compaction_deferred() to decide if the zone should be skipped. If watermarks fail after compacting the zone, defer_compaction() is called. The zone where watermarks passed can still be deferred when the allocation attempt is unsuccessful. When allocation is successful, compaction_defer_reset() is called for the zone containing the allocated page. This approach should approximate calling defer_compaction() only on zones where compaction was attempted and did not yield allocated page. There might be corner cases but that is inevitable as long as the decision to stop compacting dues not guarantee that a page will be allocated. Due to a new COMPACT_DEFERRED return value, some functions relying implicitly on COMPACT_SKIPPED = 0 had to be updated, with comments made more accurate. The did_some_progress output parameter of __alloc_pages_direct_compact() is removed completely, as the caller actually does not use it after compaction sets it - it is only considered when direct reclaim sets it. During testing on a two-node machine with a single very small Normal zone on node 1, this patch has improved success rates in stress-highalloc mmtests benchmark. The success here were previously made worse by commit 3a025760fc15 ("mm: page_alloc: spill to remote nodes before waking kswapd") as kswapd was no longer resetting often enough the deferred compaction for the Normal zone, and DMA32 zones on both nodes were thus not considered for compaction. On different machine, success rates were improved with __GFP_NO_KSWAPD allocations. [akpm@linux-foundation.org: fix CONFIG_COMPACTION=n build] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Minchan Kim <minchan@kernel.org> Reviewed-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:02 +00:00
/*
* We think that allocation won't succeed in this zone
* so we defer compaction there. If it ends up
* succeeding after all, it will be reset.
*/
defer_compaction(zone, order);
mm, compaction: khugepaged should not give up due to need_resched() Async compaction aborts when it detects zone lock contention or need_resched() is true. David Rientjes has reported that in practice, most direct async compactions for THP allocation abort due to need_resched(). This means that a second direct compaction is never attempted, which might be OK for a page fault, but khugepaged is intended to attempt a sync compaction in such case and in these cases it won't. This patch replaces "bool contended" in compact_control with an int that distinguishes between aborting due to need_resched() and aborting due to lock contention. This allows propagating the abort through all compaction functions as before, but passing the abort reason up to __alloc_pages_slowpath() which decides when to continue with direct reclaim and another compaction attempt. Another problem is that try_to_compact_pages() did not act upon the reported contention (both need_resched() or lock contention) immediately and would proceed with another zone from the zonelist. When need_resched() is true, that means initializing another zone compaction, only to check again need_resched() in isolate_migratepages() and aborting. For zone lock contention, the unintended consequence is that the lock contended status reported back to the allocator is detrmined from the last zone where compaction was attempted, which is rather arbitrary. This patch fixes the problem in the following way: - async compaction of a zone aborting due to need_resched() or fatal signal pending means that further zones should not be tried. We report COMPACT_CONTENDED_SCHED to the allocator. - aborting zone compaction due to lock contention means we can still try another zone, since it has different set of locks. We report back COMPACT_CONTENDED_LOCK only if *all* zones where compaction was attempted, it was aborted due to lock contention. As a result of these fixes, khugepaged will proceed with second sync compaction as intended, when the preceding async compaction aborted due to need_resched(). Page fault compactions aborting due to need_resched() will spare some cycles previously wasted by initializing another zone compaction only to abort again. Lock contention will be reported only when compaction in all zones aborted due to lock contention, and therefore it's not a good idea to try again after reclaim. In stress-highalloc from mmtests configured to use __GFP_NO_KSWAPD, this has improved number of THP collapse allocations by 10%, which shows positive effect on khugepaged. The benchmark's success rates are unchanged as it is not recognized as khugepaged. Numbers of compact_stall and compact_fail events have however decreased by 20%, with compact_success still a bit improved, which is good. With benchmark configured not to use __GFP_NO_KSWAPD, there is 6% improvement in THP collapse allocations, and only slight improvement in stalls and failures. [akpm@linux-foundation.org: fix warnings] Reported-by: David Rientjes <rientjes@google.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:14 +00:00
/*
* We might have stopped compacting due to need_resched() in
* async compaction, or due to a fatal signal detected. In that
mm, compaction: simplify contended compaction handling Async compaction detects contention either due to failing trylock on zone->lock or lru_lock, or by need_resched(). Since 1f9efdef4f3f ("mm, compaction: khugepaged should not give up due to need_resched()") the code got quite complicated to distinguish these two up to the __alloc_pages_slowpath() level, so different decisions could be taken for khugepaged allocations. After the recent changes, khugepaged allocations don't check for contended compaction anymore, so we again don't need to distinguish lock and sched contention, and simplify the current convoluted code a lot. However, I believe it's also possible to simplify even more and completely remove the check for contended compaction after the initial async compaction for costly orders, which was originally aimed at THP page fault allocations. There are several reasons why this can be done now: - with the new defaults, THP page faults no longer do reclaim/compaction at all, unless the system admin has overridden the default, or application has indicated via madvise that it can benefit from THP's. In both cases, it means that the potential extra latency is expected and worth the benefits. - even if reclaim/compaction proceeds after this patch where it previously wouldn't, the second compaction attempt is still async and will detect the contention and back off, if the contention persists - there are still heuristics like deferred compaction and pageblock skip bits in place that prevent excessive THP page fault latencies Link: http://lkml.kernel.org/r/20160721073614.24395-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:49:30 +00:00
* case do not try further zones
mm, compaction: khugepaged should not give up due to need_resched() Async compaction aborts when it detects zone lock contention or need_resched() is true. David Rientjes has reported that in practice, most direct async compactions for THP allocation abort due to need_resched(). This means that a second direct compaction is never attempted, which might be OK for a page fault, but khugepaged is intended to attempt a sync compaction in such case and in these cases it won't. This patch replaces "bool contended" in compact_control with an int that distinguishes between aborting due to need_resched() and aborting due to lock contention. This allows propagating the abort through all compaction functions as before, but passing the abort reason up to __alloc_pages_slowpath() which decides when to continue with direct reclaim and another compaction attempt. Another problem is that try_to_compact_pages() did not act upon the reported contention (both need_resched() or lock contention) immediately and would proceed with another zone from the zonelist. When need_resched() is true, that means initializing another zone compaction, only to check again need_resched() in isolate_migratepages() and aborting. For zone lock contention, the unintended consequence is that the lock contended status reported back to the allocator is detrmined from the last zone where compaction was attempted, which is rather arbitrary. This patch fixes the problem in the following way: - async compaction of a zone aborting due to need_resched() or fatal signal pending means that further zones should not be tried. We report COMPACT_CONTENDED_SCHED to the allocator. - aborting zone compaction due to lock contention means we can still try another zone, since it has different set of locks. We report back COMPACT_CONTENDED_LOCK only if *all* zones where compaction was attempted, it was aborted due to lock contention. As a result of these fixes, khugepaged will proceed with second sync compaction as intended, when the preceding async compaction aborted due to need_resched(). Page fault compactions aborting due to need_resched() will spare some cycles previously wasted by initializing another zone compaction only to abort again. Lock contention will be reported only when compaction in all zones aborted due to lock contention, and therefore it's not a good idea to try again after reclaim. In stress-highalloc from mmtests configured to use __GFP_NO_KSWAPD, this has improved number of THP collapse allocations by 10%, which shows positive effect on khugepaged. The benchmark's success rates are unchanged as it is not recognized as khugepaged. Numbers of compact_stall and compact_fail events have however decreased by 20%, with compact_success still a bit improved, which is good. With benchmark configured not to use __GFP_NO_KSWAPD, there is 6% improvement in THP collapse allocations, and only slight improvement in stalls and failures. [akpm@linux-foundation.org: fix warnings] Reported-by: David Rientjes <rientjes@google.com> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Christoph Lameter <cl@linux.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-10-09 22:27:14 +00:00
*/
mm, compaction: simplify contended compaction handling Async compaction detects contention either due to failing trylock on zone->lock or lru_lock, or by need_resched(). Since 1f9efdef4f3f ("mm, compaction: khugepaged should not give up due to need_resched()") the code got quite complicated to distinguish these two up to the __alloc_pages_slowpath() level, so different decisions could be taken for khugepaged allocations. After the recent changes, khugepaged allocations don't check for contended compaction anymore, so we again don't need to distinguish lock and sched contention, and simplify the current convoluted code a lot. However, I believe it's also possible to simplify even more and completely remove the check for contended compaction after the initial async compaction for costly orders, which was originally aimed at THP page fault allocations. There are several reasons why this can be done now: - with the new defaults, THP page faults no longer do reclaim/compaction at all, unless the system admin has overridden the default, or application has indicated via madvise that it can benefit from THP's. In both cases, it means that the potential extra latency is expected and worth the benefits. - even if reclaim/compaction proceeds after this patch where it previously wouldn't, the second compaction attempt is still async and will detect the contention and back off, if the contention persists - there are still heuristics like deferred compaction and pageblock skip bits in place that prevent excessive THP page fault latencies Link: http://lkml.kernel.org/r/20160721073614.24395-9-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:49:30 +00:00
if ((prio == COMPACT_PRIO_ASYNC && need_resched())
|| fatal_signal_pending(current))
break;
}
return rc;
}
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/*
* Compact all zones within a node till each zone's fragmentation score
* reaches within proactive compaction thresholds (as determined by the
* proactiveness tunable).
*
* It is possible that the function returns before reaching score targets
* due to various back-off conditions, such as, contention on per-node or
* per-zone locks.
*/
static void proactive_compact_node(pg_data_t *pgdat)
{
int zoneid;
struct zone *zone;
struct compact_control cc = {
.order = -1,
.mode = MIGRATE_SYNC_LIGHT,
.ignore_skip_hint = true,
.whole_zone = true,
.gfp_mask = GFP_KERNEL,
.proactive_compaction = true,
};
for (zoneid = 0; zoneid < MAX_NR_ZONES; zoneid++) {
zone = &pgdat->node_zones[zoneid];
if (!populated_zone(zone))
continue;
cc.zone = zone;
compact_zone(&cc, NULL);
VM_BUG_ON(!list_empty(&cc.freepages));
VM_BUG_ON(!list_empty(&cc.migratepages));
}
}
/* Compact all zones within a node */
static void compact_node(int nid)
{
pg_data_t *pgdat = NODE_DATA(nid);
int zoneid;
struct zone *zone;
struct compact_control cc = {
.order = -1,
.mode = MIGRATE_SYNC,
.ignore_skip_hint = true,
.whole_zone = true,
mm, compaction: allow compaction for GFP_NOFS requests compaction has been disabled for GFP_NOFS and GFP_NOIO requests since the direct compaction was introduced by commit 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). The main reason is that the migration of page cache pages might recurse back to fs/io layer and we could potentially deadlock. This is overly conservative because all the anonymous memory is migrateable in the GFP_NOFS context just fine. This might be a large portion of the memory in many/most workkloads. Remove the GFP_NOFS restriction and make sure that we skip all fs pages (those with a mapping) while isolating pages to be migrated. We cannot consider clean fs pages because they might need a metadata update so only isolate pages without any mapping for nofs requests. The effect of this patch will be probably very limited in many/most workloads because higher order GFP_NOFS requests are quite rare, although different configurations might lead to very different results. David Chinner has mentioned a heavy metadata workload with 64kB block which to quote him: : Unfortunately, there was an era of cargo cult configuration tweaks in the : Ceph community that has resulted in a large number of production machines : with XFS filesystems configured this way. And a lot of them store large : numbers of small files and run under significant sustained memory : pressure. : : I slowly working towards getting rid of these high order allocations and : replacing them with the equivalent number of single page allocations, but : I haven't got that (complex) change working yet. We can do the following to simulate that workload: $ mkfs.xfs -f -n size=64k <dev> $ mount <dev> /mnt/scratch $ time ./fs_mark -D 10000 -S0 -n 100000 -s 0 -L 32 \ -d /mnt/scratch/0 -d /mnt/scratch/1 \ -d /mnt/scratch/2 -d /mnt/scratch/3 \ -d /mnt/scratch/4 -d /mnt/scratch/5 \ -d /mnt/scratch/6 -d /mnt/scratch/7 \ -d /mnt/scratch/8 -d /mnt/scratch/9 \ -d /mnt/scratch/10 -d /mnt/scratch/11 \ -d /mnt/scratch/12 -d /mnt/scratch/13 \ -d /mnt/scratch/14 -d /mnt/scratch/15 and indeed is hammers the system with many high order GFP_NOFS requests as per a simle tracepoint during the load: $ echo '!(gfp_flags & 0x80) && (gfp_flags &0x400000)' > $TRACE_MNT/events/kmem/mm_page_alloc/filter I am getting 5287609 order=0 37 order=1 1594905 order=2 3048439 order=3 6699207 order=4 66645 order=5 My testing was done in a kvm guest so performance numbers should be taken with a grain of salt but there seems to be a difference when the patch is applied: * Original kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4300.1 20745838 3 3200000 0 4239.9 23849857 5 4800000 0 4243.4 25939543 6 6400000 0 4248.4 19514050 8 8000000 0 4262.1 20796169 9 9600000 0 4257.6 21288675 11 11200000 0 4259.7 19375120 13 12800000 0 4220.7 22734141 14 14400000 0 4238.5 31936458 16 16000000 0 4231.5 23409901 18 17600000 0 4045.3 23577700 19 19200000 0 2783.4 58299526 21 20800000 0 2678.2 40616302 23 22400000 0 2693.5 83973996 and xfs complaining about memory allocation not making progress [ 2304.372647] XFS: fs_mark(3289) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) [ 2304.443323] XFS: fs_mark(3285) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) [ 4796.772477] XFS: fs_mark(3424) possible memory allocation deadlock size 46936 in kmem_alloc (mode:0x2408240) [ 4796.775329] XFS: fs_mark(3423) possible memory allocation deadlock size 51416 in kmem_alloc (mode:0x2408240) [ 4797.388808] XFS: fs_mark(3424) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) * Patched kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4289.1 19243934 3 3200000 0 4241.6 32828865 5 4800000 0 4248.7 32884693 6 6400000 0 4314.4 19608921 8 8000000 0 4269.9 24953292 9 9600000 0 4270.7 33235572 11 11200000 0 4346.4 40817101 13 12800000 0 4285.3 29972397 14 14400000 0 4297.2 20539765 16 16000000 0 4219.6 18596767 18 17600000 0 4273.8 49611187 19 19200000 0 4300.4 27944451 21 20800000 0 4270.6 22324585 22 22400000 0 4317.6 22650382 24 24000000 0 4065.2 22297964 So the dropdown at Count 19200000 didn't happen and there was only a single warning about allocation not making progress [ 3063.815003] XFS: fs_mark(3272) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) This suggests that the patch has helped even though there is not all that much of anonymous memory as the workload mostly generates fs metadata. I assume the success rate would be higher with more anonymous memory which should be the case in many workloads. [akpm@linux-foundation.org: fix comment] Link: http://lkml.kernel.org/r/20161012114721.31853-1-mhocko@kernel.org Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 23:04:07 +00:00
.gfp_mask = GFP_KERNEL,
};
for (zoneid = 0; zoneid < MAX_NR_ZONES; zoneid++) {
zone = &pgdat->node_zones[zoneid];
if (!populated_zone(zone))
continue;
cc.zone = zone;
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
compact_zone(&cc, NULL);
VM_BUG_ON(!list_empty(&cc.freepages));
VM_BUG_ON(!list_empty(&cc.migratepages));
}
}
/* Compact all nodes in the system */
static void compact_nodes(void)
{
int nid;
compact_pgdat: workaround lockdep warning in kswapd I get this lockdep warning from swapping load on linux-next, due to "vmscan: kswapd carefully call compaction". ================================= [ INFO: inconsistent lock state ] 3.3.0-rc2-next-20120201 #5 Not tainted --------------------------------- inconsistent {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} usage. kswapd0/28 [HC0[0]:SC0[0]:HE1:SE1] takes: (pcpu_alloc_mutex){+.+.?.}, at: [<ffffffff810d6684>] pcpu_alloc+0x67/0x325 {RECLAIM_FS-ON-W} state was registered at: [<ffffffff81099b75>] mark_held_locks+0xd7/0x103 [<ffffffff8109a13c>] lockdep_trace_alloc+0x85/0x9e [<ffffffff810f6bdc>] __kmalloc+0x6c/0x14b [<ffffffff810d57fd>] pcpu_mem_zalloc+0x59/0x62 [<ffffffff810d5d16>] pcpu_extend_area_map+0x26/0xb1 [<ffffffff810d679f>] pcpu_alloc+0x182/0x325 [<ffffffff810d694d>] __alloc_percpu+0xb/0xd [<ffffffff8142ebfd>] snmp_mib_init+0x1e/0x2e [<ffffffff8185cd8d>] ipv4_mib_init_net+0x7a/0x184 [<ffffffff813dc963>] ops_init.clone.0+0x6b/0x73 [<ffffffff813dc9cc>] register_pernet_operations+0x61/0xa0 [<ffffffff813dca8e>] register_pernet_subsys+0x29/0x42 [<ffffffff8185d044>] inet_init+0x1ad/0x252 [<ffffffff810002e3>] do_one_initcall+0x7a/0x12f [<ffffffff81832bc5>] kernel_init+0x9d/0x11e [<ffffffff814e51e4>] kernel_thread_helper+0x4/0x10 irq event stamp: 656613 hardirqs last enabled at (656613): [<ffffffff814e0ddc>] __mutex_unlock_slowpath+0x104/0x128 hardirqs last disabled at (656612): [<ffffffff814e0d34>] __mutex_unlock_slowpath+0x5c/0x128 softirqs last enabled at (655568): [<ffffffff8105b4a5>] __do_softirq+0x120/0x136 softirqs last disabled at (654757): [<ffffffff814e52dc>] call_softirq+0x1c/0x30 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pcpu_alloc_mutex); <Interrupt> lock(pcpu_alloc_mutex); *** DEADLOCK *** no locks held by kswapd0/28. stack backtrace: Pid: 28, comm: kswapd0 Not tainted 3.3.0-rc2-next-20120201 #5 Call Trace: [<ffffffff810981f4>] print_usage_bug+0x1bf/0x1d0 [<ffffffff81096c3e>] ? print_irq_inversion_bug+0x1d9/0x1d9 [<ffffffff810982c0>] mark_lock_irq+0xbb/0x22e [<ffffffff810c5399>] ? free_hot_cold_page+0x13d/0x14f [<ffffffff81098684>] mark_lock+0x251/0x331 [<ffffffff81098893>] mark_irqflags+0x12f/0x141 [<ffffffff81098e32>] __lock_acquire+0x58d/0x753 [<ffffffff810d6684>] ? pcpu_alloc+0x67/0x325 [<ffffffff81099433>] lock_acquire+0x54/0x6a [<ffffffff810d6684>] ? pcpu_alloc+0x67/0x325 [<ffffffff8107a5b8>] ? add_preempt_count+0xa9/0xae [<ffffffff814e0a21>] mutex_lock_nested+0x5e/0x315 [<ffffffff810d6684>] ? pcpu_alloc+0x67/0x325 [<ffffffff81098f81>] ? __lock_acquire+0x6dc/0x753 [<ffffffff810c9fb0>] ? __pagevec_release+0x2c/0x2c [<ffffffff810d6684>] pcpu_alloc+0x67/0x325 [<ffffffff810c9fb0>] ? __pagevec_release+0x2c/0x2c [<ffffffff810d694d>] __alloc_percpu+0xb/0xd [<ffffffff8106c35e>] schedule_on_each_cpu+0x23/0x110 [<ffffffff810c9fcb>] lru_add_drain_all+0x10/0x12 [<ffffffff810f126f>] __compact_pgdat+0x20/0x182 [<ffffffff810f15c2>] compact_pgdat+0x27/0x29 [<ffffffff810c306b>] ? zone_watermark_ok+0x1a/0x1c [<ffffffff810cdf6f>] balance_pgdat+0x732/0x751 [<ffffffff810ce0ed>] kswapd+0x15f/0x178 [<ffffffff810cdf8e>] ? balance_pgdat+0x751/0x751 [<ffffffff8106fd11>] kthread+0x84/0x8c [<ffffffff814e51e4>] kernel_thread_helper+0x4/0x10 [<ffffffff810787ed>] ? finish_task_switch+0x85/0xea [<ffffffff814e3861>] ? retint_restore_args+0xe/0xe [<ffffffff8106fc8d>] ? __init_kthread_worker+0x56/0x56 [<ffffffff814e51e0>] ? gs_change+0xb/0xb The RECLAIM_FS notations indicate that it's doing the GFP_FS checking that Nick hacked into lockdep a while back: I think we're intended to read that "<Interrupt>" in the DEADLOCK scenario as "<Direct reclaim>". I'm hazy, I have not reached any conclusion as to whether it's right to complain or not; but I believe it's uneasy about kswapd now doing the mutex_lock(&pcpu_alloc_mutex) which lru_add_drain_all() entails. Nor have I reached any conclusion as to whether it's important for kswapd to do that draining or not. But so as not to get blocked on this, with lockdep disabled from giving further reports, here's a patch which removes the lru_add_drain_all() from kswapd's callpath (and calls it only once from compact_nodes(), instead of once per node). Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: Mel Gorman <mel@csn.ul.ie> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 23:33:53 +00:00
/* Flush pending updates to the LRU lists */
lru_add_drain_all();
for_each_online_node(nid)
compact_node(nid);
}
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/*
* Tunable for proactive compaction. It determines how
* aggressively the kernel should compact memory in the
* background. It takes values in the range [0, 100].
*/
unsigned int __read_mostly sysctl_compaction_proactiveness = 20;
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/*
* This is the entry point for compacting all nodes via
* /proc/sys/vm/compact_memory
*/
int sysctl_compaction_handler(struct ctl_table *table, int write,
void *buffer, size_t *length, loff_t *ppos)
{
if (write)
compact_nodes();
return 0;
}
#if defined(CONFIG_SYSFS) && defined(CONFIG_NUMA)
static ssize_t sysfs_compact_node(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t count)
{
compact_pgdat: workaround lockdep warning in kswapd I get this lockdep warning from swapping load on linux-next, due to "vmscan: kswapd carefully call compaction". ================================= [ INFO: inconsistent lock state ] 3.3.0-rc2-next-20120201 #5 Not tainted --------------------------------- inconsistent {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} usage. kswapd0/28 [HC0[0]:SC0[0]:HE1:SE1] takes: (pcpu_alloc_mutex){+.+.?.}, at: [<ffffffff810d6684>] pcpu_alloc+0x67/0x325 {RECLAIM_FS-ON-W} state was registered at: [<ffffffff81099b75>] mark_held_locks+0xd7/0x103 [<ffffffff8109a13c>] lockdep_trace_alloc+0x85/0x9e [<ffffffff810f6bdc>] __kmalloc+0x6c/0x14b [<ffffffff810d57fd>] pcpu_mem_zalloc+0x59/0x62 [<ffffffff810d5d16>] pcpu_extend_area_map+0x26/0xb1 [<ffffffff810d679f>] pcpu_alloc+0x182/0x325 [<ffffffff810d694d>] __alloc_percpu+0xb/0xd [<ffffffff8142ebfd>] snmp_mib_init+0x1e/0x2e [<ffffffff8185cd8d>] ipv4_mib_init_net+0x7a/0x184 [<ffffffff813dc963>] ops_init.clone.0+0x6b/0x73 [<ffffffff813dc9cc>] register_pernet_operations+0x61/0xa0 [<ffffffff813dca8e>] register_pernet_subsys+0x29/0x42 [<ffffffff8185d044>] inet_init+0x1ad/0x252 [<ffffffff810002e3>] do_one_initcall+0x7a/0x12f [<ffffffff81832bc5>] kernel_init+0x9d/0x11e [<ffffffff814e51e4>] kernel_thread_helper+0x4/0x10 irq event stamp: 656613 hardirqs last enabled at (656613): [<ffffffff814e0ddc>] __mutex_unlock_slowpath+0x104/0x128 hardirqs last disabled at (656612): [<ffffffff814e0d34>] __mutex_unlock_slowpath+0x5c/0x128 softirqs last enabled at (655568): [<ffffffff8105b4a5>] __do_softirq+0x120/0x136 softirqs last disabled at (654757): [<ffffffff814e52dc>] call_softirq+0x1c/0x30 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pcpu_alloc_mutex); <Interrupt> lock(pcpu_alloc_mutex); *** DEADLOCK *** no locks held by kswapd0/28. stack backtrace: Pid: 28, comm: kswapd0 Not tainted 3.3.0-rc2-next-20120201 #5 Call Trace: [<ffffffff810981f4>] print_usage_bug+0x1bf/0x1d0 [<ffffffff81096c3e>] ? print_irq_inversion_bug+0x1d9/0x1d9 [<ffffffff810982c0>] mark_lock_irq+0xbb/0x22e [<ffffffff810c5399>] ? free_hot_cold_page+0x13d/0x14f [<ffffffff81098684>] mark_lock+0x251/0x331 [<ffffffff81098893>] mark_irqflags+0x12f/0x141 [<ffffffff81098e32>] __lock_acquire+0x58d/0x753 [<ffffffff810d6684>] ? pcpu_alloc+0x67/0x325 [<ffffffff81099433>] lock_acquire+0x54/0x6a [<ffffffff810d6684>] ? pcpu_alloc+0x67/0x325 [<ffffffff8107a5b8>] ? add_preempt_count+0xa9/0xae [<ffffffff814e0a21>] mutex_lock_nested+0x5e/0x315 [<ffffffff810d6684>] ? pcpu_alloc+0x67/0x325 [<ffffffff81098f81>] ? __lock_acquire+0x6dc/0x753 [<ffffffff810c9fb0>] ? __pagevec_release+0x2c/0x2c [<ffffffff810d6684>] pcpu_alloc+0x67/0x325 [<ffffffff810c9fb0>] ? __pagevec_release+0x2c/0x2c [<ffffffff810d694d>] __alloc_percpu+0xb/0xd [<ffffffff8106c35e>] schedule_on_each_cpu+0x23/0x110 [<ffffffff810c9fcb>] lru_add_drain_all+0x10/0x12 [<ffffffff810f126f>] __compact_pgdat+0x20/0x182 [<ffffffff810f15c2>] compact_pgdat+0x27/0x29 [<ffffffff810c306b>] ? zone_watermark_ok+0x1a/0x1c [<ffffffff810cdf6f>] balance_pgdat+0x732/0x751 [<ffffffff810ce0ed>] kswapd+0x15f/0x178 [<ffffffff810cdf8e>] ? balance_pgdat+0x751/0x751 [<ffffffff8106fd11>] kthread+0x84/0x8c [<ffffffff814e51e4>] kernel_thread_helper+0x4/0x10 [<ffffffff810787ed>] ? finish_task_switch+0x85/0xea [<ffffffff814e3861>] ? retint_restore_args+0xe/0xe [<ffffffff8106fc8d>] ? __init_kthread_worker+0x56/0x56 [<ffffffff814e51e0>] ? gs_change+0xb/0xb The RECLAIM_FS notations indicate that it's doing the GFP_FS checking that Nick hacked into lockdep a while back: I think we're intended to read that "<Interrupt>" in the DEADLOCK scenario as "<Direct reclaim>". I'm hazy, I have not reached any conclusion as to whether it's right to complain or not; but I believe it's uneasy about kswapd now doing the mutex_lock(&pcpu_alloc_mutex) which lru_add_drain_all() entails. Nor have I reached any conclusion as to whether it's important for kswapd to do that draining or not. But so as not to get blocked on this, with lockdep disabled from giving further reports, here's a patch which removes the lru_add_drain_all() from kswapd's callpath (and calls it only once from compact_nodes(), instead of once per node). Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@redhat.com> Acked-by: Mel Gorman <mel@csn.ul.ie> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 23:33:53 +00:00
int nid = dev->id;
if (nid >= 0 && nid < nr_node_ids && node_online(nid)) {
/* Flush pending updates to the LRU lists */
lru_add_drain_all();
compact_node(nid);
}
return count;
}
static DEVICE_ATTR(compact, 0200, NULL, sysfs_compact_node);
int compaction_register_node(struct node *node)
{
return device_create_file(&node->dev, &dev_attr_compact);
}
void compaction_unregister_node(struct node *node)
{
return device_remove_file(&node->dev, &dev_attr_compact);
}
#endif /* CONFIG_SYSFS && CONFIG_NUMA */
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
static inline bool kcompactd_work_requested(pg_data_t *pgdat)
{
mm: fix kcompactd hang during memory offlining Assume memory47 is the last online block left in node1. This will hang: # echo offline > /sys/devices/system/node/node1/memory47/state After a couple of minutes, the following pops up in dmesg: INFO: task bash:957 blocked for more than 120 seconds. Not tainted 4.6.0-rc6+ #6 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. bash D ffff8800b7adbaf8 0 957 951 0x00000000 Call Trace: schedule+0x35/0x80 schedule_timeout+0x1ac/0x270 wait_for_completion+0xe1/0x120 kthread_stop+0x4f/0x110 kcompactd_stop+0x26/0x40 __offline_pages.constprop.28+0x7e6/0x840 offline_pages+0x11/0x20 memory_block_action+0x73/0x1d0 memory_subsys_offline+0x47/0x60 device_offline+0x86/0xb0 store_mem_state+0xda/0xf0 dev_attr_store+0x18/0x30 sysfs_kf_write+0x37/0x40 kernfs_fop_write+0x11d/0x170 __vfs_write+0x37/0x120 vfs_write+0xa9/0x1a0 SyS_write+0x55/0xc0 entry_SYSCALL_64_fastpath+0x1a/0xa4 kcompactd is waiting for kcompactd_max_order > 0 when it's woken up to actually exit. Check kthread_should_stop() to break out of the wait. Fixes: 698b1b306 ("mm, compaction: introduce kcompactd"). Reported-by: Reza Arbab <arbab@linux.vnet.ibm.com> Tested-by: Reza Arbab <arbab@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-05 23:22:32 +00:00
return pgdat->kcompactd_max_order > 0 || kthread_should_stop();
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
}
static bool kcompactd_node_suitable(pg_data_t *pgdat)
{
int zoneid;
struct zone *zone;
enum zone_type highest_zoneidx = pgdat->kcompactd_highest_zoneidx;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
for (zoneid = 0; zoneid <= highest_zoneidx; zoneid++) {
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
zone = &pgdat->node_zones[zoneid];
if (!populated_zone(zone))
continue;
if (compaction_suitable(zone, pgdat->kcompactd_max_order, 0,
highest_zoneidx) == COMPACT_CONTINUE)
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
return true;
}
return false;
}
static void kcompactd_do_work(pg_data_t *pgdat)
{
/*
* With no special task, compact all zones so that a page of requested
* order is allocatable.
*/
int zoneid;
struct zone *zone;
struct compact_control cc = {
.order = pgdat->kcompactd_max_order,
.search_order = pgdat->kcompactd_max_order,
.highest_zoneidx = pgdat->kcompactd_highest_zoneidx,
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
.mode = MIGRATE_SYNC_LIGHT,
.ignore_skip_hint = false,
mm, compaction: allow compaction for GFP_NOFS requests compaction has been disabled for GFP_NOFS and GFP_NOIO requests since the direct compaction was introduced by commit 56de7263fcf3 ("mm: compaction: direct compact when a high-order allocation fails"). The main reason is that the migration of page cache pages might recurse back to fs/io layer and we could potentially deadlock. This is overly conservative because all the anonymous memory is migrateable in the GFP_NOFS context just fine. This might be a large portion of the memory in many/most workkloads. Remove the GFP_NOFS restriction and make sure that we skip all fs pages (those with a mapping) while isolating pages to be migrated. We cannot consider clean fs pages because they might need a metadata update so only isolate pages without any mapping for nofs requests. The effect of this patch will be probably very limited in many/most workloads because higher order GFP_NOFS requests are quite rare, although different configurations might lead to very different results. David Chinner has mentioned a heavy metadata workload with 64kB block which to quote him: : Unfortunately, there was an era of cargo cult configuration tweaks in the : Ceph community that has resulted in a large number of production machines : with XFS filesystems configured this way. And a lot of them store large : numbers of small files and run under significant sustained memory : pressure. : : I slowly working towards getting rid of these high order allocations and : replacing them with the equivalent number of single page allocations, but : I haven't got that (complex) change working yet. We can do the following to simulate that workload: $ mkfs.xfs -f -n size=64k <dev> $ mount <dev> /mnt/scratch $ time ./fs_mark -D 10000 -S0 -n 100000 -s 0 -L 32 \ -d /mnt/scratch/0 -d /mnt/scratch/1 \ -d /mnt/scratch/2 -d /mnt/scratch/3 \ -d /mnt/scratch/4 -d /mnt/scratch/5 \ -d /mnt/scratch/6 -d /mnt/scratch/7 \ -d /mnt/scratch/8 -d /mnt/scratch/9 \ -d /mnt/scratch/10 -d /mnt/scratch/11 \ -d /mnt/scratch/12 -d /mnt/scratch/13 \ -d /mnt/scratch/14 -d /mnt/scratch/15 and indeed is hammers the system with many high order GFP_NOFS requests as per a simle tracepoint during the load: $ echo '!(gfp_flags & 0x80) && (gfp_flags &0x400000)' > $TRACE_MNT/events/kmem/mm_page_alloc/filter I am getting 5287609 order=0 37 order=1 1594905 order=2 3048439 order=3 6699207 order=4 66645 order=5 My testing was done in a kvm guest so performance numbers should be taken with a grain of salt but there seems to be a difference when the patch is applied: * Original kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4300.1 20745838 3 3200000 0 4239.9 23849857 5 4800000 0 4243.4 25939543 6 6400000 0 4248.4 19514050 8 8000000 0 4262.1 20796169 9 9600000 0 4257.6 21288675 11 11200000 0 4259.7 19375120 13 12800000 0 4220.7 22734141 14 14400000 0 4238.5 31936458 16 16000000 0 4231.5 23409901 18 17600000 0 4045.3 23577700 19 19200000 0 2783.4 58299526 21 20800000 0 2678.2 40616302 23 22400000 0 2693.5 83973996 and xfs complaining about memory allocation not making progress [ 2304.372647] XFS: fs_mark(3289) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) [ 2304.443323] XFS: fs_mark(3285) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) [ 4796.772477] XFS: fs_mark(3424) possible memory allocation deadlock size 46936 in kmem_alloc (mode:0x2408240) [ 4796.775329] XFS: fs_mark(3423) possible memory allocation deadlock size 51416 in kmem_alloc (mode:0x2408240) [ 4797.388808] XFS: fs_mark(3424) possible memory allocation deadlock size 65728 in kmem_alloc (mode:0x2408240) * Patched kernel FSUse% Count Size Files/sec App Overhead 1 1600000 0 4289.1 19243934 3 3200000 0 4241.6 32828865 5 4800000 0 4248.7 32884693 6 6400000 0 4314.4 19608921 8 8000000 0 4269.9 24953292 9 9600000 0 4270.7 33235572 11 11200000 0 4346.4 40817101 13 12800000 0 4285.3 29972397 14 14400000 0 4297.2 20539765 16 16000000 0 4219.6 18596767 18 17600000 0 4273.8 49611187 19 19200000 0 4300.4 27944451 21 20800000 0 4270.6 22324585 22 22400000 0 4317.6 22650382 24 24000000 0 4065.2 22297964 So the dropdown at Count 19200000 didn't happen and there was only a single warning about allocation not making progress [ 3063.815003] XFS: fs_mark(3272) possible memory allocation deadlock size 65624 in kmem_alloc (mode:0x2408240) This suggests that the patch has helped even though there is not all that much of anonymous memory as the workload mostly generates fs metadata. I assume the success rate would be higher with more anonymous memory which should be the case in many workloads. [akpm@linux-foundation.org: fix comment] Link: http://lkml.kernel.org/r/20161012114721.31853-1-mhocko@kernel.org Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <js1304@gmail.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 23:04:07 +00:00
.gfp_mask = GFP_KERNEL,
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
};
trace_mm_compaction_kcompactd_wake(pgdat->node_id, cc.order,
cc.highest_zoneidx);
count_compact_event(KCOMPACTD_WAKE);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
for (zoneid = 0; zoneid <= cc.highest_zoneidx; zoneid++) {
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
int status;
zone = &pgdat->node_zones[zoneid];
if (!populated_zone(zone))
continue;
if (compaction_deferred(zone, cc.order))
continue;
if (compaction_suitable(zone, cc.order, 0, zoneid) !=
COMPACT_CONTINUE)
continue;
mm: fix kcompactd hang during memory offlining Assume memory47 is the last online block left in node1. This will hang: # echo offline > /sys/devices/system/node/node1/memory47/state After a couple of minutes, the following pops up in dmesg: INFO: task bash:957 blocked for more than 120 seconds. Not tainted 4.6.0-rc6+ #6 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. bash D ffff8800b7adbaf8 0 957 951 0x00000000 Call Trace: schedule+0x35/0x80 schedule_timeout+0x1ac/0x270 wait_for_completion+0xe1/0x120 kthread_stop+0x4f/0x110 kcompactd_stop+0x26/0x40 __offline_pages.constprop.28+0x7e6/0x840 offline_pages+0x11/0x20 memory_block_action+0x73/0x1d0 memory_subsys_offline+0x47/0x60 device_offline+0x86/0xb0 store_mem_state+0xda/0xf0 dev_attr_store+0x18/0x30 sysfs_kf_write+0x37/0x40 kernfs_fop_write+0x11d/0x170 __vfs_write+0x37/0x120 vfs_write+0xa9/0x1a0 SyS_write+0x55/0xc0 entry_SYSCALL_64_fastpath+0x1a/0xa4 kcompactd is waiting for kcompactd_max_order > 0 when it's woken up to actually exit. Check kthread_should_stop() to break out of the wait. Fixes: 698b1b306 ("mm, compaction: introduce kcompactd"). Reported-by: Reza Arbab <arbab@linux.vnet.ibm.com> Tested-by: Reza Arbab <arbab@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-05 23:22:32 +00:00
if (kthread_should_stop())
return;
cc.zone = zone;
mm, compaction: capture a page under direct compaction Compaction is inherently race-prone as a suitable page freed during compaction can be allocated by any parallel task. This patch uses a capture_control structure to isolate a page immediately when it is freed by a direct compactor in the slow path of the page allocator. The intent is to avoid redundant scanning. 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Amean fault-both-1 0.00 ( 0.00%) 0.00 * 0.00%* Amean fault-both-3 2582.11 ( 0.00%) 2563.68 ( 0.71%) Amean fault-both-5 4500.26 ( 0.00%) 4233.52 ( 5.93%) Amean fault-both-7 5819.53 ( 0.00%) 6333.65 ( -8.83%) Amean fault-both-12 9321.18 ( 0.00%) 9759.38 ( -4.70%) Amean fault-both-18 9782.76 ( 0.00%) 10338.76 ( -5.68%) Amean fault-both-24 15272.81 ( 0.00%) 13379.55 * 12.40%* Amean fault-both-30 15121.34 ( 0.00%) 16158.25 ( -6.86%) Amean fault-both-32 18466.67 ( 0.00%) 18971.21 ( -2.73%) Latency is only moderately affected but the devil is in the details. A closer examination indicates that base page fault latency is reduced but latency of huge pages is increased as it takes creater care to succeed. Part of the "problem" is that allocation success rates are close to 100% even when under pressure and compaction gets harder 5.0.0-rc1 5.0.0-rc1 selective-v3r17 capture-v3r19 Percentage huge-3 96.70 ( 0.00%) 98.23 ( 1.58%) Percentage huge-5 96.99 ( 0.00%) 95.30 ( -1.75%) Percentage huge-7 94.19 ( 0.00%) 97.24 ( 3.24%) Percentage huge-12 94.95 ( 0.00%) 97.35 ( 2.53%) Percentage huge-18 96.74 ( 0.00%) 97.30 ( 0.58%) Percentage huge-24 97.07 ( 0.00%) 97.55 ( 0.50%) Percentage huge-30 95.69 ( 0.00%) 98.50 ( 2.95%) Percentage huge-32 96.70 ( 0.00%) 99.27 ( 2.65%) And scan rates are reduced as expected by 6% for the migration scanner and 29% for the free scanner indicating that there is less redundant work. Compaction migrate scanned 20815362 19573286 Compaction free scanned 16352612 11510663 [mgorman@techsingularity.net: remove redundant check] Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-05 23:45:41 +00:00
status = compact_zone(&cc, NULL);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
mm, compaction: don't recheck watermarks after COMPACT_SUCCESS Joonsoo has reminded me that in a later patch changing watermark checks throughout compaction I forgot to update checks in try_to_compact_pages() and compactd_do_work(). Closer inspection however shows that they are redundant now in the success case, because compact_zone() now reliably reports this with COMPACT_SUCCESS. So effectively the checks just repeat (a subset) of checks that have just passed. So instead of checking watermarks again, just test the return value. Note it's also possible that compaction would declare failure e.g. because its find_suitable_fallback() is more strict than simple watermark check, and then the watermark check we are removing would then still succeed. After this patch this is not possible and it's arguably better, because for long-term fragmentation avoidance we should rather try a different zone than allocate with the unsuitable fallback. If compaction of all zones fail and the allocation is important enough, it will retry and succeed anyway. Also remove the stray "bool success" variable from kcompactd_do_work(). Link: http://lkml.kernel.org/r/20160810091226.6709-5-vbabka@suse.cz Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Joonsoo Kim <iamjoonsoo.kim@lge.com> Tested-by: Lorenzo Stoakes <lstoakes@gmail.com> Acked-by: Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 23:57:44 +00:00
if (status == COMPACT_SUCCESS) {
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
compaction_defer_reset(zone, cc.order, false);
} else if (status == COMPACT_PARTIAL_SKIPPED || status == COMPACT_COMPLETE) {
mm, compaction: drain pcps for zone when kcompactd fails It's possible for free pages to become stranded on per-cpu pagesets (pcps) that, if drained, could be merged with buddy pages on the zone's free area to form large order pages, including up to MAX_ORDER. Consider a verbose example using the tools/vm/page-types tool at the beginning of a ZONE_NORMAL ('B' indicates a buddy page and 'S' indicates a slab page). Pages on pcps do not have any page flags set. 109954 1 _______S________________________________________________________ 109955 2 __________B_____________________________________________________ 109957 1 ________________________________________________________________ 109958 1 __________B_____________________________________________________ 109959 7 ________________________________________________________________ 109960 1 __________B_____________________________________________________ 109961 9 ________________________________________________________________ 10996a 1 __________B_____________________________________________________ 10996b 3 ________________________________________________________________ 10996e 1 __________B_____________________________________________________ 10996f 1 ________________________________________________________________ ... 109f8c 1 __________B_____________________________________________________ 109f8d 2 ________________________________________________________________ 109f8f 2 __________B_____________________________________________________ 109f91 f ________________________________________________________________ 109fa0 1 __________B_____________________________________________________ 109fa1 7 ________________________________________________________________ 109fa8 1 __________B_____________________________________________________ 109fa9 1 ________________________________________________________________ 109faa 1 __________B_____________________________________________________ 109fab 1 _______S________________________________________________________ The compaction migration scanner is attempting to defragment this memory since it is at the beginning of the zone. It has done so quite well, all movable pages have been migrated. From pfn [0x109955, 0x109fab), there are only buddy pages and pages without flags set. These pages may be stranded on pcps that could otherwise allow this memory to be coalesced if freed back to the zone free area. It is possible that some of these pages may not be on pcps and that something has called alloc_pages() and used the memory directly, but we rely on the absence of __GFP_MOVABLE in these cases to allocate from MIGATE_UNMOVABLE pageblocks to try to keep these MIGRATE_MOVABLE pageblocks as free as possible. These buddy and pcp pages, spanning 1,621 pages, could be coalesced and allow for three transparent hugepages to be dynamically allocated. Running the numbers for all such spans on the system, it was found that there were over 400 such spans of only buddy pages and pages without flags set at the time this /proc/kpageflags sample was collected. Without this support, there were _no_ order-9 or order-10 pages free. When kcompactd fails to defragment memory such that a cc.order page can be allocated, drain all pcps for the zone back to the buddy allocator so this stranding cannot occur. Compaction for that order will subsequently be deferred, which acts as a ratelimit on this drain. Link: http://lkml.kernel.org/r/alpine.DEB.2.20.1803010340100.88270@chino.kir.corp.google.com Signed-off-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-04-05 23:24:02 +00:00
/*
* Buddy pages may become stranded on pcps that could
* otherwise coalesce on the zone's free area for
* order >= cc.order. This is ratelimited by the
* upcoming deferral.
*/
drain_all_pages(zone);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
/*
* We use sync migration mode here, so we defer like
* sync direct compaction does.
*/
defer_compaction(zone, cc.order);
}
count_compact_events(KCOMPACTD_MIGRATE_SCANNED,
cc.total_migrate_scanned);
count_compact_events(KCOMPACTD_FREE_SCANNED,
cc.total_free_scanned);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
VM_BUG_ON(!list_empty(&cc.freepages));
VM_BUG_ON(!list_empty(&cc.migratepages));
}
/*
* Regardless of success, we are done until woken up next. But remember
* the requested order/highest_zoneidx in case it was higher/tighter
* than our current ones
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
*/
if (pgdat->kcompactd_max_order <= cc.order)
pgdat->kcompactd_max_order = 0;
if (pgdat->kcompactd_highest_zoneidx >= cc.highest_zoneidx)
pgdat->kcompactd_highest_zoneidx = pgdat->nr_zones - 1;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
}
void wakeup_kcompactd(pg_data_t *pgdat, int order, int highest_zoneidx)
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
{
if (!order)
return;
if (pgdat->kcompactd_max_order < order)
pgdat->kcompactd_max_order = order;
if (pgdat->kcompactd_highest_zoneidx > highest_zoneidx)
pgdat->kcompactd_highest_zoneidx = highest_zoneidx;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
/*
* Pairs with implicit barrier in wait_event_freezable()
* such that wakeups are not missed.
*/
if (!wq_has_sleeper(&pgdat->kcompactd_wait))
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
return;
if (!kcompactd_node_suitable(pgdat))
return;
trace_mm_compaction_wakeup_kcompactd(pgdat->node_id, order,
highest_zoneidx);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
wake_up_interruptible(&pgdat->kcompactd_wait);
}
/*
* The background compaction daemon, started as a kernel thread
* from the init process.
*/
static int kcompactd(void *p)
{
mm/mempool: minor coding style tweaks Various coding style tweaks to various files under mm/ [daizhiyuan@phytium.com.cn: mm/swapfile: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614223624-16055-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/sparse: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614227288-19363-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/vmscan: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614227649-19853-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/compaction: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614228218-20770-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/oom_kill: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614228360-21168-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/shmem: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614228504-21491-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/page_alloc: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614228613-21754-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/filemap: minor coding style tweaks] Link: https://lkml.kernel.org/r/1614228936-22337-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/mlock: minor coding style tweaks] Link: https://lkml.kernel.org/r/1613956588-2453-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/frontswap: minor coding style tweaks] Link: https://lkml.kernel.org/r/1613962668-15045-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/vmalloc: minor coding style tweaks] Link: https://lkml.kernel.org/r/1613963379-15988-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/memory_hotplug: minor coding style tweaks] Link: https://lkml.kernel.org/r/1613971784-24878-1-git-send-email-daizhiyuan@phytium.com.cn [daizhiyuan@phytium.com.cn: mm/mempolicy: minor coding style tweaks] Link: https://lkml.kernel.org/r/1613972228-25501-1-git-send-email-daizhiyuan@phytium.com.cn Link: https://lkml.kernel.org/r/1614222374-13805-1-git-send-email-daizhiyuan@phytium.com.cn Signed-off-by: Zhiyuan Dai <daizhiyuan@phytium.com.cn> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-05-05 01:40:12 +00:00
pg_data_t *pgdat = (pg_data_t *)p;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
struct task_struct *tsk = current;
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
unsigned int proactive_defer = 0;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id);
if (!cpumask_empty(cpumask))
set_cpus_allowed_ptr(tsk, cpumask);
set_freezable();
pgdat->kcompactd_max_order = 0;
pgdat->kcompactd_highest_zoneidx = pgdat->nr_zones - 1;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
while (!kthread_should_stop()) {
psi: pressure stall information for CPU, memory, and IO When systems are overcommitted and resources become contended, it's hard to tell exactly the impact this has on workload productivity, or how close the system is to lockups and OOM kills. In particular, when machines work multiple jobs concurrently, the impact of overcommit in terms of latency and throughput on the individual job can be enormous. In order to maximize hardware utilization without sacrificing individual job health or risk complete machine lockups, this patch implements a way to quantify resource pressure in the system. A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that expose the percentage of time the system is stalled on CPU, memory, or IO, respectively. Stall states are aggregate versions of the per-task delay accounting delays: cpu: some tasks are runnable but not executing on a CPU memory: tasks are reclaiming, or waiting for swapin or thrashing cache io: tasks are waiting for io completions These percentages of walltime can be thought of as pressure percentages, and they give a general sense of system health and productivity loss incurred by resource overcommit. They can also indicate when the system is approaching lockup scenarios and OOMs. To do this, psi keeps track of the task states associated with each CPU and samples the time they spend in stall states. Every 2 seconds, the samples are averaged across CPUs - weighted by the CPUs' non-idle time to eliminate artifacts from unused CPUs - and translated into percentages of walltime. A running average of those percentages is maintained over 10s, 1m, and 5m periods (similar to the loadaverage). [hannes@cmpxchg.org: doc fixlet, per Randy] Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org [hannes@cmpxchg.org: code optimization] Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org [hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter] Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org [hannes@cmpxchg.org: fix build] Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Tested-by: Daniel Drake <drake@endlessm.com> Tested-by: Suren Baghdasaryan <surenb@google.com> Cc: Christopher Lameter <cl@linux.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Johannes Weiner <jweiner@fb.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Enderborg <peter.enderborg@sony.com> Cc: Randy Dunlap <rdunlap@infradead.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tejun Heo <tj@kernel.org> Cc: Vinayak Menon <vinmenon@codeaurora.org> Cc: Randy Dunlap <rdunlap@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 22:06:27 +00:00
unsigned long pflags;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
trace_mm_compaction_kcompactd_sleep(pgdat->node_id);
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
if (wait_event_freezable_timeout(pgdat->kcompactd_wait,
kcompactd_work_requested(pgdat),
msecs_to_jiffies(HPAGE_FRAG_CHECK_INTERVAL_MSEC))) {
psi_memstall_enter(&pflags);
kcompactd_do_work(pgdat);
psi_memstall_leave(&pflags);
continue;
}
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
mm: proactive compaction For some applications, we need to allocate almost all memory as hugepages. However, on a running system, higher-order allocations can fail if the memory is fragmented. Linux kernel currently does on-demand compaction as we request more hugepages, but this style of compaction incurs very high latency. Experiments with one-time full memory compaction (followed by hugepage allocations) show that kernel is able to restore a highly fragmented memory state to a fairly compacted memory state within <1 sec for a 32G system. Such data suggests that a more proactive compaction can help us allocate a large fraction of memory as hugepages keeping allocation latencies low. For a more proactive compaction, the approach taken here is to define a new sysctl called 'vm.compaction_proactiveness' which dictates bounds for external fragmentation which kcompactd tries to maintain. The tunable takes a value in range [0, 100], with a default of 20. Note that a previous version of this patch [1] was found to introduce too many tunables (per-order extfrag{low, high}), but this one reduces them to just one sysctl. Also, the new tunable is an opaque value instead of asking for specific bounds of "external fragmentation", which would have been difficult to estimate. The internal interpretation of this opaque value allows for future fine-tuning. Currently, we use a simple translation from this tunable to [low, high] "fragmentation score" thresholds (low=100-proactiveness, high=low+10%). The score for a node is defined as weighted mean of per-zone external fragmentation. A zone's present_pages determines its weight. To periodically check per-node score, we reuse per-node kcompactd threads, which are woken up every 500 milliseconds to check the same. If a node's score exceeds its high threshold (as derived from user-provided proactiveness value), proactive compaction is started until its score reaches its low threshold value. By default, proactiveness is set to 20, which implies threshold values of low=80 and high=90. This patch is largely based on ideas from Michal Hocko [2]. See also the LWN article [3]. Performance data ================ System: x64_64, 1T RAM, 80 CPU threads. Kernel: 5.6.0-rc3 + this patch echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/enabled echo madvise | sudo tee /sys/kernel/mm/transparent_hugepage/defrag Before starting the driver, the system was fragmented from a userspace program that allocates all memory and then for each 2M aligned section, frees 3/4 of base pages using munmap. The workload is mainly anonymous userspace pages, which are easy to move around. I intentionally avoided unmovable pages in this test to see how much latency we incur when hugepage allocations hit direct compaction. 1. Kernel hugepage allocation latencies With the system in such a fragmented state, a kernel driver then allocates as many hugepages as possible and measures allocation latency: (all latency values are in microseconds) - With vanilla 5.6.0-rc3 percentile latency –––––––––– ––––––– 5 7894 10 9496 25 12561 30 15295 40 18244 50 21229 60 27556 75 30147 80 31047 90 32859 95 33799 Total 2M hugepages allocated = 383859 (749G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) - With 5.6.0-rc3 + this patch, with proactiveness=20 sysctl -w vm.compaction_proactiveness=20 percentile latency –––––––––– ––––––– 5 2 10 2 25 3 30 3 40 3 50 4 60 4 75 4 80 4 90 5 95 429 Total 2M hugepages allocated = 384105 (750G worth of hugepages out of 762G total free => 98% of free memory could be allocated as hugepages) 2. JAVA heap allocation In this test, we first fragment memory using the same method as for (1). Then, we start a Java process with a heap size set to 700G and request the heap to be allocated with THP hugepages. We also set THP to madvise to allow hugepage backing of this heap. /usr/bin/time java -Xms700G -Xmx700G -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch The above command allocates 700G of Java heap using hugepages. - With vanilla 5.6.0-rc3 17.39user 1666.48system 27:37.89elapsed - With 5.6.0-rc3 + this patch, with proactiveness=20 8.35user 194.58system 3:19.62elapsed Elapsed time remains around 3:15, as proactiveness is further increased. Note that proactive compaction happens throughout the runtime of these workloads. The situation of one-time compaction, sufficient to supply hugepages for following allocation stream, can probably happen for more extreme proactiveness values, like 80 or 90. In the above Java workload, proactiveness is set to 20. The test starts with a node's score of 80 or higher, depending on the delay between the fragmentation step and starting the benchmark, which gives more-or-less time for the initial round of compaction. As t he benchmark consumes hugepages, node's score quickly rises above the high threshold (90) and proactive compaction starts again, which brings down the score to the low threshold level (80). Repeat. bpftrace also confirms proactive compaction running 20+ times during the runtime of this Java benchmark. kcompactd threads consume 100% of one of the CPUs while it tries to bring a node's score within thresholds. Backoff behavior ================ Above workloads produce a memory state which is easy to compact. However, if memory is filled with unmovable pages, proactive compaction should essentially back off. To test this aspect: - Created a kernel driver that allocates almost all memory as hugepages followed by freeing first 3/4 of each hugepage. - Set proactiveness=40 - Note that proactive_compact_node() is deferred maximum number of times with HPAGE_FRAG_CHECK_INTERVAL_MSEC of wait between each check (=> ~30 seconds between retries). [1] https://patchwork.kernel.org/patch/11098289/ [2] https://lore.kernel.org/linux-mm/20161230131412.GI13301@dhcp22.suse.cz/ [3] https://lwn.net/Articles/817905/ Signed-off-by: Nitin Gupta <nigupta@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Tested-by: Oleksandr Natalenko <oleksandr@redhat.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Nitin Gupta <ngupta@nitingupta.dev> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Link: http://lkml.kernel.org/r/20200616204527.19185-1-nigupta@nvidia.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 01:31:00 +00:00
/* kcompactd wait timeout */
if (should_proactive_compact_node(pgdat)) {
unsigned int prev_score, score;
if (proactive_defer) {
proactive_defer--;
continue;
}
prev_score = fragmentation_score_node(pgdat);
proactive_compact_node(pgdat);
score = fragmentation_score_node(pgdat);
/*
* Defer proactive compaction if the fragmentation
* score did not go down i.e. no progress made.
*/
proactive_defer = score < prev_score ?
0 : 1 << COMPACT_MAX_DEFER_SHIFT;
}
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
}
return 0;
}
/*
* This kcompactd start function will be called by init and node-hot-add.
* On node-hot-add, kcompactd will moved to proper cpus if cpus are hot-added.
*/
int kcompactd_run(int nid)
{
pg_data_t *pgdat = NODE_DATA(nid);
int ret = 0;
if (pgdat->kcompactd)
return 0;
pgdat->kcompactd = kthread_run(kcompactd, pgdat, "kcompactd%d", nid);
if (IS_ERR(pgdat->kcompactd)) {
pr_err("Failed to start kcompactd on node %d\n", nid);
ret = PTR_ERR(pgdat->kcompactd);
pgdat->kcompactd = NULL;
}
return ret;
}
/*
* Called by memory hotplug when all memory in a node is offlined. Caller must
* hold mem_hotplug_begin/end().
*/
void kcompactd_stop(int nid)
{
struct task_struct *kcompactd = NODE_DATA(nid)->kcompactd;
if (kcompactd) {
kthread_stop(kcompactd);
NODE_DATA(nid)->kcompactd = NULL;
}
}
/*
* It's optimal to keep kcompactd on the same CPUs as their memory, but
* not required for correctness. So if the last cpu in a node goes
* away, we get changed to run anywhere: as the first one comes back,
* restore their cpu bindings.
*/
static int kcompactd_cpu_online(unsigned int cpu)
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
{
int nid;
for_each_node_state(nid, N_MEMORY) {
pg_data_t *pgdat = NODE_DATA(nid);
const struct cpumask *mask;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
mask = cpumask_of_node(pgdat->node_id);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
/* One of our CPUs online: restore mask */
set_cpus_allowed_ptr(pgdat->kcompactd, mask);
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
}
return 0;
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
}
static int __init kcompactd_init(void)
{
int nid;
int ret;
ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
"mm/compaction:online",
kcompactd_cpu_online, NULL);
if (ret < 0) {
pr_err("kcompactd: failed to register hotplug callbacks.\n");
return ret;
}
mm, compaction: introduce kcompactd Memory compaction can be currently performed in several contexts: - kswapd balancing a zone after a high-order allocation failure - direct compaction to satisfy a high-order allocation, including THP page fault attemps - khugepaged trying to collapse a hugepage - manually from /proc The purpose of compaction is two-fold. The obvious purpose is to satisfy a (pending or future) high-order allocation, and is easy to evaluate. The other purpose is to keep overal memory fragmentation low and help the anti-fragmentation mechanism. The success wrt the latter purpose is more The current situation wrt the purposes has a few drawbacks: - compaction is invoked only when a high-order page or hugepage is not available (or manually). This might be too late for the purposes of keeping memory fragmentation low. - direct compaction increases latency of allocations. Again, it would be better if compaction was performed asynchronously to keep fragmentation low, before the allocation itself comes. - (a special case of the previous) the cost of compaction during THP page faults can easily offset the benefits of THP. - kswapd compaction appears to be complex, fragile and not working in some scenarios. It could also end up compacting for a high-order allocation request when it should be reclaiming memory for a later order-0 request. To improve the situation, we should be able to benefit from an equivalent of kswapd, but for compaction - i.e. a background thread which responds to fragmentation and the need for high-order allocations (including hugepages) somewhat proactively. One possibility is to extend the responsibilities of kswapd, which could however complicate its design too much. It should be better to let kswapd handle reclaim, as order-0 allocations are often more critical than high-order ones. Another possibility is to extend khugepaged, but this kthread is a single instance and tied to THP configs. This patch goes with the option of a new set of per-node kthreads called kcompactd, and lays the foundations, without introducing any new tunables. The lifecycle mimics kswapd kthreads, including the memory hotplug hooks. For compaction, kcompactd uses the standard compaction_suitable() and ompact_finished() criteria and the deferred compaction functionality. Unlike direct compaction, it uses only sync compaction, as there's no allocation latency to minimize. This patch doesn't yet add a call to wakeup_kcompactd. The kswapd compact/reclaim loop for high-order pages will be replaced by waking up kcompactd in the next patch with the description of what's wrong with the old approach. Waking up of the kcompactd threads is also tied to kswapd activity and follows these rules: - we don't want to affect any fastpaths, so wake up kcompactd only from the slowpath, as it's done for kswapd - if kswapd is doing reclaim, it's more important than compaction, so don't invoke kcompactd until kswapd goes to sleep - the target order used for kswapd is passed to kcompactd Future possible future uses for kcompactd include the ability to wake up kcompactd on demand in special situations, such as when hugepages are not available (currently not done due to __GFP_NO_KSWAPD) or when a fragmentation event (i.e. __rmqueue_fallback()) occurs. It's also possible to perform periodic compaction with kcompactd. [arnd@arndb.de: fix build errors with kcompactd] [paul.gortmaker@windriver.com: don't use modular references for non modular code] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 21:18:08 +00:00
for_each_node_state(nid, N_MEMORY)
kcompactd_run(nid);
return 0;
}
subsys_initcall(kcompactd_init)
#endif /* CONFIG_COMPACTION */