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 */
|
2005-04-16 22:20:36 +00:00
|
|
|
#ifndef _LINUX_MMZONE_H
|
|
|
|
#define _LINUX_MMZONE_H
|
|
|
|
|
|
|
|
#ifndef __ASSEMBLY__
|
2008-04-28 09:12:54 +00:00
|
|
|
#ifndef __GENERATING_BOUNDS_H
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
#include <linux/list.h>
|
|
|
|
#include <linux/wait.h>
|
2007-10-17 06:25:54 +00:00
|
|
|
#include <linux/bitops.h>
|
2005-04-16 22:20:36 +00:00
|
|
|
#include <linux/cache.h>
|
|
|
|
#include <linux/threads.h>
|
|
|
|
#include <linux/numa.h>
|
|
|
|
#include <linux/init.h>
|
2005-10-30 01:16:53 +00:00
|
|
|
#include <linux/seqlock.h>
|
2006-03-27 09:15:57 +00:00
|
|
|
#include <linux/nodemask.h>
|
Add a bitmap that is used to track flags affecting a block of pages
Here is the latest revision of the anti-fragmentation patches. Of particular
note in this version is special treatment of high-order atomic allocations.
Care is taken to group them together and avoid grouping pages of other types
near them. Artifical tests imply that it works. I'm trying to get the
hardware together that would allow setting up of a "real" test. If anyone
already has a setup and test that can trigger the atomic-allocation problem,
I'd appreciate a test of these patches and a report. The second major change
is that these patches will apply cleanly with patches that implement
anti-fragmentation through zones.
kernbench shows effectively no performance difference varying between -0.2%
and +2% on a variety of test machines. Success rates for huge page allocation
are dramatically increased. For example, on a ppc64 machine, the vanilla
kernel was only able to allocate 1% of memory as a hugepage and this was due
to a single hugepage reserved as min_free_kbytes. With these patches applied,
17% was allocatable as superpages. With reclaim-related fixes from Andy
Whitcroft, it was 40% and further reclaim-related improvements should increase
this further.
Changelog Since V28
o Group high-order atomic allocations together
o It is no longer required to set min_free_kbytes to 10% of memory. A value
of 16384 in most cases will be sufficient
o Now applied with zone-based anti-fragmentation
o Fix incorrect VM_BUG_ON within buffered_rmqueue()
o Reorder the stack so later patches do not back out work from earlier patches
o Fix bug were journal pages were being treated as movable
o Bias placement of non-movable pages to lower PFNs
o More agressive clustering of reclaimable pages in reactions to workloads
like updatedb that flood the size of inode caches
Changelog Since V27
o Renamed anti-fragmentation to Page Clustering. Anti-fragmentation was giving
the mistaken impression that it was the 100% solution for high order
allocations. Instead, it greatly increases the chances high-order
allocations will succeed and lays the foundation for defragmentation and
memory hot-remove to work properly
o Redefine page groupings based on ability to migrate or reclaim instead of
basing on reclaimability alone
o Get rid of spurious inits
o Per-cpu lists are no longer split up per-type. Instead the per-cpu list is
searched for a page of the appropriate type
o Added more explanation commentary
o Fix up bug in pageblock code where bitmap was used before being initalised
Changelog Since V26
o Fix double init of lists in setup_pageset
Changelog Since V25
o Fix loop order of for_each_rclmtype_order so that order of loop matches args
o gfpflags_to_rclmtype uses gfp_t instead of unsigned long
o Rename get_pageblock_type() to get_page_rclmtype()
o Fix alignment problem in move_freepages()
o Add mechanism for assigning flags to blocks of pages instead of page->flags
o On fallback, do not examine the preferred list of free pages a second time
The purpose of these patches is to reduce external fragmentation by grouping
pages of related types together. When pages are migrated (or reclaimed under
memory pressure), large contiguous pages will be freed.
This patch works by categorising allocations by their ability to migrate;
Movable - The pages may be moved with the page migration mechanism. These are
generally userspace pages.
Reclaimable - These are allocations for some kernel caches that are
reclaimable or allocations that are known to be very short-lived.
Unmovable - These are pages that are allocated by the kernel that
are not trivially reclaimed. For example, the memory allocated for a
loaded module would be in this category. By default, allocations are
considered to be of this type
HighAtomic - These are high-order allocations belonging to callers that
cannot sleep or perform any IO. In practice, this is restricted to
jumbo frame allocation for network receive. It is assumed that the
allocations are short-lived
Instead of having one MAX_ORDER-sized array of free lists in struct free_area,
there is one for each type of reclaimability. Once a 2^MAX_ORDER block of
pages is split for a type of allocation, it is added to the free-lists for
that type, in effect reserving it. Hence, over time, pages of the different
types can be clustered together.
When the preferred freelists are expired, the largest possible block is taken
from an alternative list. Buddies that are split from that large block are
placed on the preferred allocation-type freelists to mitigate fragmentation.
This implementation gives best-effort for low fragmentation in all zones.
Ideally, min_free_kbytes needs to be set to a value equal to 4 * (1 <<
(MAX_ORDER-1)) pages in most cases. This would be 16384 on x86 and x86_64 for
example.
Our tests show that about 60-70% of physical memory can be allocated on a
desktop after a few days uptime. In benchmarks and stress tests, we are
finding that 80% of memory is available as contiguous blocks at the end of the
test. To compare, a standard kernel was getting < 1% of memory as large pages
on a desktop and about 8-12% of memory as large pages at the end of stress
tests.
Following this email are 12 patches that implement thie page grouping feature.
The first patch introduces a mechanism for storing flags related to a whole
block of pages. Then allocations are split between movable and all other
allocations. Following that are patches to deal with per-cpu pages and make
the mechanism configurable. The next patch moves free pages between lists
when partially allocated blocks are used for pages of another migrate type.
The second last patch groups reclaimable kernel allocations such as inode
caches together. The final patch related to groupings keeps high-order atomic
allocations.
The last two patches are more concerned with control of fragmentation. The
second last patch biases placement of non-movable allocations towards the
start of memory. This is with a view of supporting memory hot-remove of DIMMs
with higher PFNs in the future. The biasing could be enforced a lot heavier
but it would cost. The last patch agressively clusters reclaimable pages like
inode caches together.
The fragmentation reduction strategy needs to track if pages within a block
can be moved or reclaimed so that pages are freed to the appropriate list.
This patch adds a bitmap for flags affecting a whole a MAX_ORDER block of
pages.
In non-SPARSEMEM configurations, the bitmap is stored in the struct zone and
allocated during initialisation. SPARSEMEM statically allocates the bitmap in
a struct mem_section so that bitmaps do not have to be resized during memory
hotadd. This wastes a small amount of memory per unused section (usually
sizeof(unsigned long)) but the complexity of dynamically allocating the memory
is quite high.
Additional credit to Andy Whitcroft who reviewed up an earlier implementation
of the mechanism an suggested how to make it a *lot* cleaner.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:47 +00:00
|
|
|
#include <linux/pageblock-flags.h>
|
2013-02-23 00:34:30 +00:00
|
|
|
#include <linux/page-flags-layout.h>
|
2011-07-26 23:09:06 +00:00
|
|
|
#include <linux/atomic.h>
|
2019-05-14 22:41:32 +00:00
|
|
|
#include <linux/mm_types.h>
|
|
|
|
#include <linux/page-flags.h>
|
2021-06-29 02:41:41 +00:00
|
|
|
#include <linux/local_lock.h>
|
[PATCH] Sparsemem build fix
From: Ralf Baechle <ralf@linux-mips.org>
<linux/mmzone.h> uses PAGE_SIZE, PAGE_SHIFT from <asm/page.h> without
including that header itself. For some sparsemem configurations this may
result in build errors like:
CC init/initramfs.o
In file included from include/linux/gfp.h:4,
from include/linux/slab.h:15,
from include/linux/percpu.h:4,
from include/linux/rcupdate.h:41,
from include/linux/dcache.h:10,
from include/linux/fs.h:226,
from init/initramfs.c:2:
include/linux/mmzone.h:498:22: warning: "PAGE_SHIFT" is not defined
In file included from include/linux/gfp.h:4,
from include/linux/slab.h:15,
from include/linux/percpu.h:4,
from include/linux/rcupdate.h:41,
from include/linux/dcache.h:10,
from include/linux/fs.h:226,
from init/initramfs.c:2:
include/linux/mmzone.h:526: error: `PAGE_SIZE' undeclared here (not in a function)
include/linux/mmzone.h: In function `__pfn_to_section':
include/linux/mmzone.h:573: error: `PAGE_SHIFT' undeclared (first use in this function)
include/linux/mmzone.h:573: error: (Each undeclared identifier is reported only once
include/linux/mmzone.h:573: error: for each function it appears in.)
include/linux/mmzone.h: In function `pfn_valid':
include/linux/mmzone.h:578: error: `PAGE_SHIFT' undeclared (first use in this function)
make[1]: *** [init/initramfs.o] Error 1
make: *** [init] Error 2
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Seems-reasonable-to: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-04 09:51:29 +00:00
|
|
|
#include <asm/page.h>
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
/* Free memory management - zoned buddy allocator. */
|
|
|
|
#ifndef CONFIG_FORCE_MAX_ZONEORDER
|
|
|
|
#define MAX_ORDER 11
|
|
|
|
#else
|
|
|
|
#define MAX_ORDER CONFIG_FORCE_MAX_ZONEORDER
|
|
|
|
#endif
|
2006-05-20 22:00:31 +00:00
|
|
|
#define MAX_ORDER_NR_PAGES (1 << (MAX_ORDER - 1))
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2007-07-17 11:03:16 +00:00
|
|
|
/*
|
|
|
|
* PAGE_ALLOC_COSTLY_ORDER is the order at which allocations are deemed
|
|
|
|
* costly to service. That is between allocation orders which should
|
2012-04-15 12:42:28 +00:00
|
|
|
* coalesce naturally under reasonable reclaim pressure and those which
|
2007-07-17 11:03:16 +00:00
|
|
|
* will not.
|
|
|
|
*/
|
|
|
|
#define PAGE_ALLOC_COSTLY_ORDER 3
|
|
|
|
|
2017-05-03 21:52:52 +00:00
|
|
|
enum migratetype {
|
2011-12-29 12:09:50 +00:00
|
|
|
MIGRATE_UNMOVABLE,
|
|
|
|
MIGRATE_MOVABLE,
|
2015-11-07 00:28:18 +00:00
|
|
|
MIGRATE_RECLAIMABLE,
|
2015-11-07 00:28:37 +00:00
|
|
|
MIGRATE_PCPTYPES, /* the number of types on the pcp lists */
|
|
|
|
MIGRATE_HIGHATOMIC = MIGRATE_PCPTYPES,
|
2011-12-29 12:09:50 +00:00
|
|
|
#ifdef CONFIG_CMA
|
|
|
|
/*
|
|
|
|
* MIGRATE_CMA migration type is designed to mimic the way
|
|
|
|
* ZONE_MOVABLE works. Only movable pages can be allocated
|
|
|
|
* from MIGRATE_CMA pageblocks and page allocator never
|
|
|
|
* implicitly change migration type of MIGRATE_CMA pageblock.
|
|
|
|
*
|
|
|
|
* The way to use it is to change migratetype of a range of
|
|
|
|
* pageblocks to MIGRATE_CMA which can be done by
|
|
|
|
* __free_pageblock_cma() function. What is important though
|
|
|
|
* is that a range of pageblocks must be aligned to
|
2021-05-07 01:05:51 +00:00
|
|
|
* MAX_ORDER_NR_PAGES should biggest page be bigger than
|
2011-12-29 12:09:50 +00:00
|
|
|
* a single pageblock.
|
|
|
|
*/
|
|
|
|
MIGRATE_CMA,
|
|
|
|
#endif
|
2013-02-23 00:33:58 +00:00
|
|
|
#ifdef CONFIG_MEMORY_ISOLATION
|
2011-12-29 12:09:50 +00:00
|
|
|
MIGRATE_ISOLATE, /* can't allocate from here */
|
2013-02-23 00:33:58 +00:00
|
|
|
#endif
|
2011-12-29 12:09:50 +00:00
|
|
|
MIGRATE_TYPES
|
|
|
|
};
|
|
|
|
|
mm, page_owner: print migratetype of page and pageblock, symbolic flags
The information in /sys/kernel/debug/page_owner includes the migratetype
of the pageblock the page belongs to. This is also checked against the
page's migratetype (as declared by gfp_flags during its allocation), and
the page is reported as Fallback if its migratetype differs from the
pageblock's one. t This is somewhat misleading because in fact fallback
allocation is not the only reason why these two can differ. It also
doesn't direcly provide the page's migratetype, although it's possible
to derive that from the gfp_flags.
It's arguably better to print both page and pageblock's migratetype and
leave the interpretation to the consumer than to suggest fallback
allocation as the only possible reason. While at it, we can print the
migratetypes as string the same way as /proc/pagetypeinfo does, as some
of the numeric values depend on kernel configuration. For that, this
patch moves the migratetype_names array from #ifdef CONFIG_PROC_FS part
of mm/vmstat.c to mm/page_alloc.c and exports it.
With the new format strings for flags, we can now also provide symbolic
page and gfp flags in the /sys/kernel/debug/page_owner file. This
replaces the positional printing of page flags as single letters, which
might have looked nicer, but was limited to a subset of flags, and
required the user to remember the letters.
Example page_owner entry after the patch:
Page allocated via order 0, mask 0x24213ca(GFP_HIGHUSER_MOVABLE|__GFP_COLD|__GFP_NOWARN|__GFP_NORETRY)
PFN 520 type Movable Block 1 type Movable Flags 0xfffff8001006c(referenced|uptodate|lru|active|mappedtodisk)
[<ffffffff811682c4>] __alloc_pages_nodemask+0x134/0x230
[<ffffffff811b4058>] alloc_pages_current+0x88/0x120
[<ffffffff8115e386>] __page_cache_alloc+0xe6/0x120
[<ffffffff8116ba6c>] __do_page_cache_readahead+0xdc/0x240
[<ffffffff8116bd05>] ondemand_readahead+0x135/0x260
[<ffffffff8116bfb1>] page_cache_sync_readahead+0x31/0x50
[<ffffffff81160523>] generic_file_read_iter+0x453/0x760
[<ffffffff811e0d57>] __vfs_read+0xa7/0xd0
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-15 21:56:08 +00:00
|
|
|
/* In mm/page_alloc.c; keep in sync also with show_migration_types() there */
|
2018-12-28 08:35:55 +00:00
|
|
|
extern const char * const migratetype_names[MIGRATE_TYPES];
|
mm, page_owner: print migratetype of page and pageblock, symbolic flags
The information in /sys/kernel/debug/page_owner includes the migratetype
of the pageblock the page belongs to. This is also checked against the
page's migratetype (as declared by gfp_flags during its allocation), and
the page is reported as Fallback if its migratetype differs from the
pageblock's one. t This is somewhat misleading because in fact fallback
allocation is not the only reason why these two can differ. It also
doesn't direcly provide the page's migratetype, although it's possible
to derive that from the gfp_flags.
It's arguably better to print both page and pageblock's migratetype and
leave the interpretation to the consumer than to suggest fallback
allocation as the only possible reason. While at it, we can print the
migratetypes as string the same way as /proc/pagetypeinfo does, as some
of the numeric values depend on kernel configuration. For that, this
patch moves the migratetype_names array from #ifdef CONFIG_PROC_FS part
of mm/vmstat.c to mm/page_alloc.c and exports it.
With the new format strings for flags, we can now also provide symbolic
page and gfp flags in the /sys/kernel/debug/page_owner file. This
replaces the positional printing of page flags as single letters, which
might have looked nicer, but was limited to a subset of flags, and
required the user to remember the letters.
Example page_owner entry after the patch:
Page allocated via order 0, mask 0x24213ca(GFP_HIGHUSER_MOVABLE|__GFP_COLD|__GFP_NOWARN|__GFP_NORETRY)
PFN 520 type Movable Block 1 type Movable Flags 0xfffff8001006c(referenced|uptodate|lru|active|mappedtodisk)
[<ffffffff811682c4>] __alloc_pages_nodemask+0x134/0x230
[<ffffffff811b4058>] alloc_pages_current+0x88/0x120
[<ffffffff8115e386>] __page_cache_alloc+0xe6/0x120
[<ffffffff8116ba6c>] __do_page_cache_readahead+0xdc/0x240
[<ffffffff8116bd05>] ondemand_readahead+0x135/0x260
[<ffffffff8116bfb1>] page_cache_sync_readahead+0x31/0x50
[<ffffffff81160523>] generic_file_read_iter+0x453/0x760
[<ffffffff811e0d57>] __vfs_read+0xa7/0xd0
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-15 21:56:08 +00:00
|
|
|
|
2011-12-29 12:09:50 +00:00
|
|
|
#ifdef CONFIG_CMA
|
|
|
|
# define is_migrate_cma(migratetype) unlikely((migratetype) == MIGRATE_CMA)
|
2016-07-19 22:00:04 +00:00
|
|
|
# define is_migrate_cma_page(_page) (get_pageblock_migratetype(_page) == MIGRATE_CMA)
|
2011-12-29 12:09:50 +00:00
|
|
|
#else
|
|
|
|
# define is_migrate_cma(migratetype) false
|
2016-07-19 22:00:04 +00:00
|
|
|
# define is_migrate_cma_page(_page) false
|
2011-12-29 12:09:50 +00:00
|
|
|
#endif
|
2007-10-16 08:25:48 +00:00
|
|
|
|
2017-05-08 22:54:43 +00:00
|
|
|
static inline bool is_migrate_movable(int mt)
|
|
|
|
{
|
|
|
|
return is_migrate_cma(mt) || mt == MIGRATE_MOVABLE;
|
|
|
|
}
|
|
|
|
|
2007-10-16 08:25:48 +00:00
|
|
|
#define for_each_migratetype_order(order, type) \
|
|
|
|
for (order = 0; order < MAX_ORDER; order++) \
|
|
|
|
for (type = 0; type < MIGRATE_TYPES; type++)
|
|
|
|
|
Print out statistics in relation to fragmentation avoidance to /proc/pagetypeinfo
This patch provides fragmentation avoidance statistics via /proc/pagetypeinfo.
The information is collected only on request so there is no runtime overhead.
The statistics are in three parts:
The first part prints information on the size of blocks that pages are
being grouped on and looks like
Page block order: 10
Pages per block: 1024
The second part is a more detailed version of /proc/buddyinfo and looks like
Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10
Node 0, zone DMA, type Unmovable 0 0 0 0 0 0 0 0 0 0 0
Node 0, zone DMA, type Reclaimable 1 0 0 0 0 0 0 0 0 0 0
Node 0, zone DMA, type Movable 0 0 0 0 0 0 0 0 0 0 0
Node 0, zone DMA, type Reserve 0 4 4 0 0 0 0 1 0 1 0
Node 0, zone Normal, type Unmovable 111 8 4 4 2 3 1 0 0 0 0
Node 0, zone Normal, type Reclaimable 293 89 8 0 0 0 0 0 0 0 0
Node 0, zone Normal, type Movable 1 6 13 9 7 6 3 0 0 0 0
Node 0, zone Normal, type Reserve 0 0 0 0 0 0 0 0 0 0 4
The third part looks like
Number of blocks type Unmovable Reclaimable Movable Reserve
Node 0, zone DMA 0 1 2 1
Node 0, zone Normal 3 17 94 4
To walk the zones within a node with interrupts disabled, walk_zones_in_node()
is introduced and shared between /proc/buddyinfo, /proc/zoneinfo and
/proc/pagetypeinfo to reduce code duplication. It seems specific to what
vmstat.c requires but could be broken out as a general utility function in
mmzone.c if there were other other potential users.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: Andy Whitcroft <apw@shadowen.org>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:26:02 +00:00
|
|
|
extern int page_group_by_mobility_disabled;
|
|
|
|
|
2020-08-07 06:25:41 +00:00
|
|
|
#define MIGRATETYPE_MASK ((1UL << PB_migratetype_bits) - 1)
|
mm: page_alloc: use word-based accesses for get/set pageblock bitmaps
The test_bit operations in get/set pageblock flags are expensive. This
patch reads the bitmap on a word basis and use shifts and masks to isolate
the bits of interest. Similarly masks are used to set a local copy of the
bitmap and then use cmpxchg to update the bitmap if there have been no
other changes made in parallel.
In a test running dd onto tmpfs the overhead of the pageblock-related
functions went from 1.27% in profiles to 0.5%.
In addition to the performance benefits, this patch closes races that are
possible between:
a) get_ and set_pageblock_migratetype(), where get_pageblock_migratetype()
reads part of the bits before and other part of the bits after
set_pageblock_migratetype() has updated them.
b) set_pageblock_migratetype() and set_pageblock_skip(), where the non-atomic
read-modify-update set bit operation in set_pageblock_skip() will cause
lost updates to some bits changed in the set_pageblock_migratetype().
Joonsoo Kim first reported the case a) via code inspection. Vlastimil
Babka's testing with a debug patch showed that either a) or b) occurs
roughly once per mmtests' stress-highalloc benchmark (although not
necessarily in the same pageblock). Furthermore during development of
unrelated compaction patches, it was observed that frequent calls to
{start,undo}_isolate_page_range() the race occurs several thousands of
times and has resulted in NULL pointer dereferences in move_freepages()
and free_one_page() in places where free_list[migratetype] is
manipulated by e.g. list_move(). Further debugging confirmed that
migratetype had invalid value of 6, causing out of bounds access to the
free_list array.
That confirmed that the race exist, although it may be extremely rare,
and currently only fatal where page isolation is performed due to
memory hot remove. Races on pageblocks being updated by
set_pageblock_migratetype(), where both old and new migratetype are
lower MIGRATE_RESERVE, currently cannot result in an invalid value
being observed, although theoretically they may still lead to
unexpected creation or destruction of MIGRATE_RESERVE pageblocks.
Furthermore, things could get suddenly worse when memory isolation is
used more, or when new migratetypes are added.
After this patch, the race has no longer been observed in testing.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Reported-and-tested-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Hugh Dickins <hughd@google.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:10:16 +00:00
|
|
|
|
2014-06-04 23:10:17 +00:00
|
|
|
#define get_pageblock_migratetype(page) \
|
2020-08-07 06:25:51 +00:00
|
|
|
get_pfnblock_flags_mask(page, page_to_pfn(page), MIGRATETYPE_MASK)
|
2014-06-04 23:10:17 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
struct free_area {
|
2007-10-16 08:25:48 +00:00
|
|
|
struct list_head free_list[MIGRATE_TYPES];
|
2005-04-16 22:20:36 +00:00
|
|
|
unsigned long nr_free;
|
|
|
|
};
|
|
|
|
|
2019-05-14 22:41:32 +00:00
|
|
|
static inline struct page *get_page_from_free_area(struct free_area *area,
|
|
|
|
int migratetype)
|
|
|
|
{
|
|
|
|
return list_first_entry_or_null(&area->free_list[migratetype],
|
|
|
|
struct page, lru);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool free_area_empty(struct free_area *area, int migratetype)
|
|
|
|
{
|
|
|
|
return list_empty(&area->free_list[migratetype]);
|
|
|
|
}
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
struct pglist_data;
|
|
|
|
|
|
|
|
/*
|
2021-07-01 01:53:17 +00:00
|
|
|
* Add a wild amount of padding here to ensure data fall into separate
|
2005-04-16 22:20:36 +00:00
|
|
|
* cachelines. There are very few zone structures in the machine, so space
|
|
|
|
* consumption is not a concern here.
|
|
|
|
*/
|
|
|
|
#if defined(CONFIG_SMP)
|
|
|
|
struct zone_padding {
|
|
|
|
char x[0];
|
2006-01-08 09:01:27 +00:00
|
|
|
} ____cacheline_internodealigned_in_smp;
|
2005-04-16 22:20:36 +00:00
|
|
|
#define ZONE_PADDING(name) struct zone_padding name;
|
|
|
|
#else
|
|
|
|
#define ZONE_PADDING(name)
|
|
|
|
#endif
|
|
|
|
|
mm: change the call sites of numa statistics items
Patch series "Separate NUMA statistics from zone statistics", v2.
Each page allocation updates a set of per-zone statistics with a call to
zone_statistics(). As discussed in 2017 MM summit, these are a
substantial source of overhead in the page allocator and are very rarely
consumed. This significant overhead in cache bouncing caused by zone
counters (NUMA associated counters) update in parallel in multi-threaded
page allocation (pointed out by Dave Hansen).
A link to the MM summit slides:
http://people.netfilter.org/hawk/presentations/MM-summit2017/MM-summit2017-JesperBrouer.pdf
To mitigate this overhead, this patchset separates NUMA statistics from
zone statistics framework, and update NUMA counter threshold to a fixed
size of MAX_U16 - 2, as a small threshold greatly increases the update
frequency of the global counter from local per cpu counter (suggested by
Ying Huang). The rationality is that these statistics counters don't
need to be read often, unlike other VM counters, so it's not a problem
to use a large threshold and make readers more expensive.
With this patchset, we see 31.3% drop of CPU cycles(537-->369, see
below) for per single page allocation and reclaim on Jesper's
page_bench03 benchmark. Meanwhile, this patchset keeps the same style
of virtual memory statistics with little end-user-visible effects (only
move the numa stats to show behind zone page stats, see the first patch
for details).
I did an experiment of single page allocation and reclaim concurrently
using Jesper's page_bench03 benchmark on a 2-Socket Broadwell-based
server (88 processors with 126G memory) with different size of threshold
of pcp counter.
Benchmark provided by Jesper D Brouer(increase loop times to 10000000):
https://github.com/netoptimizer/prototype-kernel/tree/master/kernel/mm/bench
Threshold CPU cycles Throughput(88 threads)
32 799 241760478
64 640 301628829
125 537 358906028 <==> system by default
256 468 412397590
512 428 450550704
4096 399 482520943
20000 394 489009617
30000 395 488017817
65533 369(-31.3%) 521661345(+45.3%) <==> with this patchset
N/A 342(-36.3%) 562900157(+56.8%) <==> disable zone_statistics
This patch (of 3):
In this patch, NUMA statistics is separated from zone statistics
framework, all the call sites of NUMA stats are changed to use
numa-stats-specific functions, it does not have any functionality change
except that the number of NUMA stats is shown behind zone page stats
when users *read* the zone info.
E.g. cat /proc/zoneinfo
***Base*** ***With this patch***
nr_free_pages 3976 nr_free_pages 3976
nr_zone_inactive_anon 0 nr_zone_inactive_anon 0
nr_zone_active_anon 0 nr_zone_active_anon 0
nr_zone_inactive_file 0 nr_zone_inactive_file 0
nr_zone_active_file 0 nr_zone_active_file 0
nr_zone_unevictable 0 nr_zone_unevictable 0
nr_zone_write_pending 0 nr_zone_write_pending 0
nr_mlock 0 nr_mlock 0
nr_page_table_pages 0 nr_page_table_pages 0
nr_kernel_stack 0 nr_kernel_stack 0
nr_bounce 0 nr_bounce 0
nr_zspages 0 nr_zspages 0
numa_hit 0 *nr_free_cma 0*
numa_miss 0 numa_hit 0
numa_foreign 0 numa_miss 0
numa_interleave 0 numa_foreign 0
numa_local 0 numa_interleave 0
numa_other 0 numa_local 0
*nr_free_cma 0* numa_other 0
... ...
vm stats threshold: 10 vm stats threshold: 10
... ...
The next patch updates the numa stats counter size and threshold.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/1503568801-21305-2-git-send-email-kemi.wang@intel.com
Signed-off-by: Kemi Wang <kemi.wang@intel.com>
Reported-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christopher Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Ying Huang <ying.huang@intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-08 23:12:48 +00:00
|
|
|
#ifdef CONFIG_NUMA
|
|
|
|
enum numa_stat_item {
|
|
|
|
NUMA_HIT, /* allocated in intended node */
|
|
|
|
NUMA_MISS, /* allocated in non intended node */
|
|
|
|
NUMA_FOREIGN, /* was intended here, hit elsewhere */
|
|
|
|
NUMA_INTERLEAVE_HIT, /* interleaver preferred this zone */
|
|
|
|
NUMA_LOCAL, /* allocation from local node */
|
|
|
|
NUMA_OTHER, /* allocation from other node */
|
2021-06-29 02:41:44 +00:00
|
|
|
NR_VM_NUMA_EVENT_ITEMS
|
mm: change the call sites of numa statistics items
Patch series "Separate NUMA statistics from zone statistics", v2.
Each page allocation updates a set of per-zone statistics with a call to
zone_statistics(). As discussed in 2017 MM summit, these are a
substantial source of overhead in the page allocator and are very rarely
consumed. This significant overhead in cache bouncing caused by zone
counters (NUMA associated counters) update in parallel in multi-threaded
page allocation (pointed out by Dave Hansen).
A link to the MM summit slides:
http://people.netfilter.org/hawk/presentations/MM-summit2017/MM-summit2017-JesperBrouer.pdf
To mitigate this overhead, this patchset separates NUMA statistics from
zone statistics framework, and update NUMA counter threshold to a fixed
size of MAX_U16 - 2, as a small threshold greatly increases the update
frequency of the global counter from local per cpu counter (suggested by
Ying Huang). The rationality is that these statistics counters don't
need to be read often, unlike other VM counters, so it's not a problem
to use a large threshold and make readers more expensive.
With this patchset, we see 31.3% drop of CPU cycles(537-->369, see
below) for per single page allocation and reclaim on Jesper's
page_bench03 benchmark. Meanwhile, this patchset keeps the same style
of virtual memory statistics with little end-user-visible effects (only
move the numa stats to show behind zone page stats, see the first patch
for details).
I did an experiment of single page allocation and reclaim concurrently
using Jesper's page_bench03 benchmark on a 2-Socket Broadwell-based
server (88 processors with 126G memory) with different size of threshold
of pcp counter.
Benchmark provided by Jesper D Brouer(increase loop times to 10000000):
https://github.com/netoptimizer/prototype-kernel/tree/master/kernel/mm/bench
Threshold CPU cycles Throughput(88 threads)
32 799 241760478
64 640 301628829
125 537 358906028 <==> system by default
256 468 412397590
512 428 450550704
4096 399 482520943
20000 394 489009617
30000 395 488017817
65533 369(-31.3%) 521661345(+45.3%) <==> with this patchset
N/A 342(-36.3%) 562900157(+56.8%) <==> disable zone_statistics
This patch (of 3):
In this patch, NUMA statistics is separated from zone statistics
framework, all the call sites of NUMA stats are changed to use
numa-stats-specific functions, it does not have any functionality change
except that the number of NUMA stats is shown behind zone page stats
when users *read* the zone info.
E.g. cat /proc/zoneinfo
***Base*** ***With this patch***
nr_free_pages 3976 nr_free_pages 3976
nr_zone_inactive_anon 0 nr_zone_inactive_anon 0
nr_zone_active_anon 0 nr_zone_active_anon 0
nr_zone_inactive_file 0 nr_zone_inactive_file 0
nr_zone_active_file 0 nr_zone_active_file 0
nr_zone_unevictable 0 nr_zone_unevictable 0
nr_zone_write_pending 0 nr_zone_write_pending 0
nr_mlock 0 nr_mlock 0
nr_page_table_pages 0 nr_page_table_pages 0
nr_kernel_stack 0 nr_kernel_stack 0
nr_bounce 0 nr_bounce 0
nr_zspages 0 nr_zspages 0
numa_hit 0 *nr_free_cma 0*
numa_miss 0 numa_hit 0
numa_foreign 0 numa_miss 0
numa_interleave 0 numa_foreign 0
numa_local 0 numa_interleave 0
numa_other 0 numa_local 0
*nr_free_cma 0* numa_other 0
... ...
vm stats threshold: 10 vm stats threshold: 10
... ...
The next patch updates the numa stats counter size and threshold.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/1503568801-21305-2-git-send-email-kemi.wang@intel.com
Signed-off-by: Kemi Wang <kemi.wang@intel.com>
Reported-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christopher Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Ying Huang <ying.huang@intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-08 23:12:48 +00:00
|
|
|
};
|
|
|
|
#else
|
2021-06-29 02:41:44 +00:00
|
|
|
#define NR_VM_NUMA_EVENT_ITEMS 0
|
mm: change the call sites of numa statistics items
Patch series "Separate NUMA statistics from zone statistics", v2.
Each page allocation updates a set of per-zone statistics with a call to
zone_statistics(). As discussed in 2017 MM summit, these are a
substantial source of overhead in the page allocator and are very rarely
consumed. This significant overhead in cache bouncing caused by zone
counters (NUMA associated counters) update in parallel in multi-threaded
page allocation (pointed out by Dave Hansen).
A link to the MM summit slides:
http://people.netfilter.org/hawk/presentations/MM-summit2017/MM-summit2017-JesperBrouer.pdf
To mitigate this overhead, this patchset separates NUMA statistics from
zone statistics framework, and update NUMA counter threshold to a fixed
size of MAX_U16 - 2, as a small threshold greatly increases the update
frequency of the global counter from local per cpu counter (suggested by
Ying Huang). The rationality is that these statistics counters don't
need to be read often, unlike other VM counters, so it's not a problem
to use a large threshold and make readers more expensive.
With this patchset, we see 31.3% drop of CPU cycles(537-->369, see
below) for per single page allocation and reclaim on Jesper's
page_bench03 benchmark. Meanwhile, this patchset keeps the same style
of virtual memory statistics with little end-user-visible effects (only
move the numa stats to show behind zone page stats, see the first patch
for details).
I did an experiment of single page allocation and reclaim concurrently
using Jesper's page_bench03 benchmark on a 2-Socket Broadwell-based
server (88 processors with 126G memory) with different size of threshold
of pcp counter.
Benchmark provided by Jesper D Brouer(increase loop times to 10000000):
https://github.com/netoptimizer/prototype-kernel/tree/master/kernel/mm/bench
Threshold CPU cycles Throughput(88 threads)
32 799 241760478
64 640 301628829
125 537 358906028 <==> system by default
256 468 412397590
512 428 450550704
4096 399 482520943
20000 394 489009617
30000 395 488017817
65533 369(-31.3%) 521661345(+45.3%) <==> with this patchset
N/A 342(-36.3%) 562900157(+56.8%) <==> disable zone_statistics
This patch (of 3):
In this patch, NUMA statistics is separated from zone statistics
framework, all the call sites of NUMA stats are changed to use
numa-stats-specific functions, it does not have any functionality change
except that the number of NUMA stats is shown behind zone page stats
when users *read* the zone info.
E.g. cat /proc/zoneinfo
***Base*** ***With this patch***
nr_free_pages 3976 nr_free_pages 3976
nr_zone_inactive_anon 0 nr_zone_inactive_anon 0
nr_zone_active_anon 0 nr_zone_active_anon 0
nr_zone_inactive_file 0 nr_zone_inactive_file 0
nr_zone_active_file 0 nr_zone_active_file 0
nr_zone_unevictable 0 nr_zone_unevictable 0
nr_zone_write_pending 0 nr_zone_write_pending 0
nr_mlock 0 nr_mlock 0
nr_page_table_pages 0 nr_page_table_pages 0
nr_kernel_stack 0 nr_kernel_stack 0
nr_bounce 0 nr_bounce 0
nr_zspages 0 nr_zspages 0
numa_hit 0 *nr_free_cma 0*
numa_miss 0 numa_hit 0
numa_foreign 0 numa_miss 0
numa_interleave 0 numa_foreign 0
numa_local 0 numa_interleave 0
numa_other 0 numa_local 0
*nr_free_cma 0* numa_other 0
... ...
vm stats threshold: 10 vm stats threshold: 10
... ...
The next patch updates the numa stats counter size and threshold.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/1503568801-21305-2-git-send-email-kemi.wang@intel.com
Signed-off-by: Kemi Wang <kemi.wang@intel.com>
Reported-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christopher Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Ying Huang <ying.huang@intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-08 23:12:48 +00:00
|
|
|
#endif
|
|
|
|
|
2006-06-30 08:55:33 +00:00
|
|
|
enum zone_stat_item {
|
2007-02-10 09:43:02 +00:00
|
|
|
/* First 128 byte cacheline (assuming 64 bit words) */
|
2007-02-10 09:43:02 +00:00
|
|
|
NR_FREE_PAGES,
|
mm: add per-zone lru list stat
When I did stress test with hackbench, I got OOM message frequently
which didn't ever happen in zone-lru.
gfp_mask=0x26004c0(GFP_KERNEL|__GFP_REPEAT|__GFP_NOTRACK), order=0
..
..
__alloc_pages_nodemask+0xe52/0xe60
? new_slab+0x39c/0x3b0
new_slab+0x39c/0x3b0
___slab_alloc.constprop.87+0x6da/0x840
? __alloc_skb+0x3c/0x260
? _raw_spin_unlock_irq+0x27/0x60
? trace_hardirqs_on_caller+0xec/0x1b0
? finish_task_switch+0xa6/0x220
? poll_select_copy_remaining+0x140/0x140
__slab_alloc.isra.81.constprop.86+0x40/0x6d
? __alloc_skb+0x3c/0x260
kmem_cache_alloc+0x22c/0x260
? __alloc_skb+0x3c/0x260
__alloc_skb+0x3c/0x260
alloc_skb_with_frags+0x4e/0x1a0
sock_alloc_send_pskb+0x16a/0x1b0
? wait_for_unix_gc+0x31/0x90
? alloc_set_pte+0x2ad/0x310
unix_stream_sendmsg+0x28d/0x340
sock_sendmsg+0x2d/0x40
sock_write_iter+0x6c/0xc0
__vfs_write+0xc0/0x120
vfs_write+0x9b/0x1a0
? __might_fault+0x49/0xa0
SyS_write+0x44/0x90
do_fast_syscall_32+0xa6/0x1e0
sysenter_past_esp+0x45/0x74
Mem-Info:
active_anon:104698 inactive_anon:105791 isolated_anon:192
active_file:433 inactive_file:283 isolated_file:22
unevictable:0 dirty:0 writeback:296 unstable:0
slab_reclaimable:6389 slab_unreclaimable:78927
mapped:474 shmem:0 pagetables:101426 bounce:0
free:10518 free_pcp:334 free_cma:0
Node 0 active_anon:418792kB inactive_anon:423164kB active_file:1732kB inactive_file:1132kB unevictable:0kB isolated(anon):768kB isolated(file):88kB mapped:1896kB dirty:0kB writeback:1184kB shmem:0kB writeback_tmp:0kB unstable:0kB pages_scanned:1478632 all_unreclaimable? yes
DMA free:3304kB min:68kB low:84kB high:100kB present:15992kB managed:15916kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:4088kB kernel_stack:0kB pagetables:2480kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB
lowmem_reserve[]: 0 809 1965 1965
Normal free:3436kB min:3604kB low:4504kB high:5404kB present:897016kB managed:858460kB mlocked:0kB slab_reclaimable:25556kB slab_unreclaimable:311712kB kernel_stack:164608kB pagetables:30844kB bounce:0kB free_pcp:620kB local_pcp:104kB free_cma:0kB
lowmem_reserve[]: 0 0 9247 9247
HighMem free:33808kB min:512kB low:1796kB high:3080kB present:1183736kB managed:1183736kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:372252kB bounce:0kB free_pcp:428kB local_pcp:72kB free_cma:0kB
lowmem_reserve[]: 0 0 0 0
DMA: 2*4kB (UM) 2*8kB (UM) 0*16kB 1*32kB (U) 1*64kB (U) 2*128kB (UM) 1*256kB (U) 1*512kB (M) 0*1024kB 1*2048kB (U) 0*4096kB = 3192kB
Normal: 33*4kB (MH) 79*8kB (ME) 11*16kB (M) 4*32kB (M) 2*64kB (ME) 2*128kB (EH) 7*256kB (EH) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 3244kB
HighMem: 2590*4kB (UM) 1568*8kB (UM) 491*16kB (UM) 60*32kB (UM) 6*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 33064kB
Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
25121 total pagecache pages
24160 pages in swap cache
Swap cache stats: add 86371, delete 62211, find 42865/60187
Free swap = 4015560kB
Total swap = 4192252kB
524186 pages RAM
295934 pages HighMem/MovableOnly
9658 pages reserved
0 pages cma reserved
The order-0 allocation for normal zone failed while there are a lot of
reclaimable memory(i.e., anonymous memory with free swap). I wanted to
analyze the problem but it was hard because we removed per-zone lru stat
so I couldn't know how many of anonymous memory there are in normal/dma
zone.
When we investigate OOM problem, reclaimable memory count is crucial
stat to find a problem. Without it, it's hard to parse the OOM message
so I believe we should keep it.
With per-zone lru stat,
gfp_mask=0x26004c0(GFP_KERNEL|__GFP_REPEAT|__GFP_NOTRACK), order=0
Mem-Info:
active_anon:101103 inactive_anon:102219 isolated_anon:0
active_file:503 inactive_file:544 isolated_file:0
unevictable:0 dirty:0 writeback:34 unstable:0
slab_reclaimable:6298 slab_unreclaimable:74669
mapped:863 shmem:0 pagetables:100998 bounce:0
free:23573 free_pcp:1861 free_cma:0
Node 0 active_anon:404412kB inactive_anon:409040kB active_file:2012kB inactive_file:2176kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:3452kB dirty:0kB writeback:136kB shmem:0kB writeback_tmp:0kB unstable:0kB pages_scanned:1320845 all_unreclaimable? yes
DMA free:3296kB min:68kB low:84kB high:100kB active_anon:5540kB inactive_anon:0kB active_file:0kB inactive_file:0kB present:15992kB managed:15916kB mlocked:0kB slab_reclaimable:248kB slab_unreclaimable:2628kB kernel_stack:792kB pagetables:2316kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB
lowmem_reserve[]: 0 809 1965 1965
Normal free:3600kB min:3604kB low:4504kB high:5404kB active_anon:86304kB inactive_anon:0kB active_file:160kB inactive_file:376kB present:897016kB managed:858524kB mlocked:0kB slab_reclaimable:24944kB slab_unreclaimable:296048kB kernel_stack:163832kB pagetables:35892kB bounce:0kB free_pcp:3076kB local_pcp:656kB free_cma:0kB
lowmem_reserve[]: 0 0 9247 9247
HighMem free:86156kB min:512kB low:1796kB high:3080kB active_anon:312852kB inactive_anon:410024kB active_file:1924kB inactive_file:2012kB present:1183736kB managed:1183736kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:365784kB bounce:0kB free_pcp:3868kB local_pcp:720kB free_cma:0kB
lowmem_reserve[]: 0 0 0 0
DMA: 8*4kB (UM) 8*8kB (UM) 4*16kB (M) 2*32kB (UM) 2*64kB (UM) 1*128kB (M) 3*256kB (UME) 2*512kB (UE) 1*1024kB (E) 0*2048kB 0*4096kB = 3296kB
Normal: 240*4kB (UME) 160*8kB (UME) 23*16kB (ME) 3*32kB (UE) 3*64kB (UME) 2*128kB (ME) 1*256kB (U) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 3408kB
HighMem: 10942*4kB (UM) 3102*8kB (UM) 866*16kB (UM) 76*32kB (UM) 11*64kB (UM) 4*128kB (UM) 1*256kB (M) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 86344kB
Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
54409 total pagecache pages
53215 pages in swap cache
Swap cache stats: add 300982, delete 247765, find 157978/226539
Free swap = 3803244kB
Total swap = 4192252kB
524186 pages RAM
295934 pages HighMem/MovableOnly
9642 pages reserved
0 pages cma reserved
With that, we can see normal zone has a 86M reclaimable memory so we can
know something goes wrong(I will fix the problem in next patch) in
reclaim.
[mgorman@techsingularity.net: rename zone LRU stats in /proc/vmstat]
Link: http://lkml.kernel.org/r/20160725072300.GK10438@techsingularity.net
Link: http://lkml.kernel.org/r/1469110261-7365-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:47:26 +00:00
|
|
|
NR_ZONE_LRU_BASE, /* Used only for compaction and reclaim retry */
|
|
|
|
NR_ZONE_INACTIVE_ANON = NR_ZONE_LRU_BASE,
|
|
|
|
NR_ZONE_ACTIVE_ANON,
|
|
|
|
NR_ZONE_INACTIVE_FILE,
|
|
|
|
NR_ZONE_ACTIVE_FILE,
|
|
|
|
NR_ZONE_UNEVICTABLE,
|
2016-07-28 22:47:31 +00:00
|
|
|
NR_ZONE_WRITE_PENDING, /* Count of dirty, writeback and unstable pages */
|
2008-10-19 03:26:51 +00:00
|
|
|
NR_MLOCK, /* mlock()ed pages found and moved off LRU */
|
2009-09-22 00:01:32 +00:00
|
|
|
/* Second 128 byte cacheline */
|
2006-06-30 08:55:41 +00:00
|
|
|
NR_BOUNCE,
|
2016-07-26 22:24:45 +00:00
|
|
|
#if IS_ENABLED(CONFIG_ZSMALLOC)
|
|
|
|
NR_ZSPAGES, /* allocated in zsmalloc */
|
2006-06-30 08:55:44 +00:00
|
|
|
#endif
|
2012-10-08 23:32:02 +00:00
|
|
|
NR_FREE_CMA_PAGES,
|
2006-06-30 08:55:33 +00:00
|
|
|
NR_VM_ZONE_STAT_ITEMS };
|
|
|
|
|
mm, vmstat: add infrastructure for per-node vmstats
Patchset: "Move LRU page reclaim from zones to nodes v9"
This series moves LRUs from the zones to the node. While this is a
current rebase, the test results were based on mmotm as of June 23rd.
Conceptually, this series is simple but there are a lot of details.
Some of the broad motivations for this are;
1. The residency of a page partially depends on what zone the page was
allocated from. This is partially combatted by the fair zone allocation
policy but that is a partial solution that introduces overhead in the
page allocator paths.
2. Currently, reclaim on node 0 behaves slightly different to node 1. For
example, direct reclaim scans in zonelist order and reclaims even if
the zone is over the high watermark regardless of the age of pages
in that LRU. Kswapd on the other hand starts reclaim on the highest
unbalanced zone. A difference in distribution of file/anon pages due
to when they were allocated results can result in a difference in
again. While the fair zone allocation policy mitigates some of the
problems here, the page reclaim results on a multi-zone node will
always be different to a single-zone node.
it was scheduled on as a result.
3. kswapd and the page allocator scan zones in the opposite order to
avoid interfering with each other but it's sensitive to timing. This
mitigates the page allocator using pages that were allocated very recently
in the ideal case but it's sensitive to timing. When kswapd is allocating
from lower zones then it's great but during the rebalancing of the highest
zone, the page allocator and kswapd interfere with each other. It's worse
if the highest zone is small and difficult to balance.
4. slab shrinkers are node-based which makes it harder to identify the exact
relationship between slab reclaim and LRU reclaim.
The reason we have zone-based reclaim is that we used to have
large highmem zones in common configurations and it was necessary
to quickly find ZONE_NORMAL pages for reclaim. Today, this is much
less of a concern as machines with lots of memory will (or should) use
64-bit kernels. Combinations of 32-bit hardware and 64-bit hardware are
rare. Machines that do use highmem should have relatively low highmem:lowmem
ratios than we worried about in the past.
Conceptually, moving to node LRUs should be easier to understand. The
page allocator plays fewer tricks to game reclaim and reclaim behaves
similarly on all nodes.
The series has been tested on a 16 core UMA machine and a 2-socket 48
core NUMA machine. The UMA results are presented in most cases as the NUMA
machine behaved similarly.
pagealloc
---------
This is a microbenchmark that shows the benefit of removing the fair zone
allocation policy. It was tested uip to order-4 but only orders 0 and 1 are
shown as the other orders were comparable.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min total-odr0-1 490.00 ( 0.00%) 457.00 ( 6.73%)
Min total-odr0-2 347.00 ( 0.00%) 329.00 ( 5.19%)
Min total-odr0-4 288.00 ( 0.00%) 273.00 ( 5.21%)
Min total-odr0-8 251.00 ( 0.00%) 239.00 ( 4.78%)
Min total-odr0-16 234.00 ( 0.00%) 222.00 ( 5.13%)
Min total-odr0-32 223.00 ( 0.00%) 211.00 ( 5.38%)
Min total-odr0-64 217.00 ( 0.00%) 208.00 ( 4.15%)
Min total-odr0-128 214.00 ( 0.00%) 204.00 ( 4.67%)
Min total-odr0-256 250.00 ( 0.00%) 230.00 ( 8.00%)
Min total-odr0-512 271.00 ( 0.00%) 269.00 ( 0.74%)
Min total-odr0-1024 291.00 ( 0.00%) 282.00 ( 3.09%)
Min total-odr0-2048 303.00 ( 0.00%) 296.00 ( 2.31%)
Min total-odr0-4096 311.00 ( 0.00%) 309.00 ( 0.64%)
Min total-odr0-8192 316.00 ( 0.00%) 314.00 ( 0.63%)
Min total-odr0-16384 317.00 ( 0.00%) 315.00 ( 0.63%)
Min total-odr1-1 742.00 ( 0.00%) 712.00 ( 4.04%)
Min total-odr1-2 562.00 ( 0.00%) 530.00 ( 5.69%)
Min total-odr1-4 457.00 ( 0.00%) 433.00 ( 5.25%)
Min total-odr1-8 411.00 ( 0.00%) 381.00 ( 7.30%)
Min total-odr1-16 381.00 ( 0.00%) 356.00 ( 6.56%)
Min total-odr1-32 372.00 ( 0.00%) 346.00 ( 6.99%)
Min total-odr1-64 372.00 ( 0.00%) 343.00 ( 7.80%)
Min total-odr1-128 375.00 ( 0.00%) 351.00 ( 6.40%)
Min total-odr1-256 379.00 ( 0.00%) 351.00 ( 7.39%)
Min total-odr1-512 385.00 ( 0.00%) 355.00 ( 7.79%)
Min total-odr1-1024 386.00 ( 0.00%) 358.00 ( 7.25%)
Min total-odr1-2048 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-4096 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-8192 388.00 ( 0.00%) 363.00 ( 6.44%)
This shows a steady improvement throughout. The primary benefit is from
reduced system CPU usage which is obvious from the overall times;
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
User 189.19 191.80
System 2604.45 2533.56
Elapsed 2855.30 2786.39
The vmstats also showed that the fair zone allocation policy was definitely
removed as can be seen here;
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v8
DMA32 allocs 28794729769 0
Normal allocs 48432501431 77227309877
Movable allocs 0 0
tiobench on ext4
----------------
tiobench is a benchmark that artifically benefits if old pages remain resident
while new pages get reclaimed. The fair zone allocation policy mitigates this
problem so pages age fairly. While the benchmark has problems, it is important
that tiobench performance remains constant as it implies that page aging
problems that the fair zone allocation policy fixes are not re-introduced.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min PotentialReadSpeed 89.65 ( 0.00%) 90.21 ( 0.62%)
Min SeqRead-MB/sec-1 82.68 ( 0.00%) 82.01 ( -0.81%)
Min SeqRead-MB/sec-2 72.76 ( 0.00%) 72.07 ( -0.95%)
Min SeqRead-MB/sec-4 75.13 ( 0.00%) 74.92 ( -0.28%)
Min SeqRead-MB/sec-8 64.91 ( 0.00%) 65.19 ( 0.43%)
Min SeqRead-MB/sec-16 62.24 ( 0.00%) 62.22 ( -0.03%)
Min RandRead-MB/sec-1 0.88 ( 0.00%) 0.88 ( 0.00%)
Min RandRead-MB/sec-2 0.95 ( 0.00%) 0.92 ( -3.16%)
Min RandRead-MB/sec-4 1.43 ( 0.00%) 1.34 ( -6.29%)
Min RandRead-MB/sec-8 1.61 ( 0.00%) 1.60 ( -0.62%)
Min RandRead-MB/sec-16 1.80 ( 0.00%) 1.90 ( 5.56%)
Min SeqWrite-MB/sec-1 76.41 ( 0.00%) 76.85 ( 0.58%)
Min SeqWrite-MB/sec-2 74.11 ( 0.00%) 73.54 ( -0.77%)
Min SeqWrite-MB/sec-4 80.05 ( 0.00%) 80.13 ( 0.10%)
Min SeqWrite-MB/sec-8 72.88 ( 0.00%) 73.20 ( 0.44%)
Min SeqWrite-MB/sec-16 75.91 ( 0.00%) 76.44 ( 0.70%)
Min RandWrite-MB/sec-1 1.18 ( 0.00%) 1.14 ( -3.39%)
Min RandWrite-MB/sec-2 1.02 ( 0.00%) 1.03 ( 0.98%)
Min RandWrite-MB/sec-4 1.05 ( 0.00%) 0.98 ( -6.67%)
Min RandWrite-MB/sec-8 0.89 ( 0.00%) 0.92 ( 3.37%)
Min RandWrite-MB/sec-16 0.92 ( 0.00%) 0.93 ( 1.09%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 approx-v9
User 645.72 525.90
System 403.85 331.75
Elapsed 6795.36 6783.67
This shows that the series has little or not impact on tiobench which is
desirable and a reduction in system CPU usage. It indicates that the fair
zone allocation policy was removed in a manner that didn't reintroduce
one class of page aging bug. There were only minor differences in overall
reclaim activity
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Minor Faults 645838 647465
Major Faults 573 640
Swap Ins 0 0
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 46041453 44190646
Normal allocs 78053072 79887245
Movable allocs 0 0
Allocation stalls 24 67
Stall zone DMA 0 0
Stall zone DMA32 0 0
Stall zone Normal 0 2
Stall zone HighMem 0 0
Stall zone Movable 0 65
Direct pages scanned 10969 30609
Kswapd pages scanned 93375144 93492094
Kswapd pages reclaimed 93372243 93489370
Direct pages reclaimed 10969 30609
Kswapd efficiency 99% 99%
Kswapd velocity 13741.015 13781.934
Direct efficiency 100% 100%
Direct velocity 1.614 4.512
Percentage direct scans 0% 0%
kswapd activity was roughly comparable. There were differences in direct
reclaim activity but negligible in the context of the overall workload
(velocity of 4 pages per second with the patches applied, 1.6 pages per
second in the baseline kernel).
pgbench read-only large configuration on ext4
---------------------------------------------
pgbench is a database benchmark that can be sensitive to page reclaim
decisions. This also checks if removing the fair zone allocation policy
is safe
pgbench Transactions
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Hmean 1 188.26 ( 0.00%) 189.78 ( 0.81%)
Hmean 5 330.66 ( 0.00%) 328.69 ( -0.59%)
Hmean 12 370.32 ( 0.00%) 380.72 ( 2.81%)
Hmean 21 368.89 ( 0.00%) 369.00 ( 0.03%)
Hmean 30 382.14 ( 0.00%) 360.89 ( -5.56%)
Hmean 32 428.87 ( 0.00%) 432.96 ( 0.95%)
Negligible differences again. As with tiobench, overall reclaim activity
was comparable.
bonnie++ on ext4
----------------
No interesting performance difference, negligible differences on reclaim
stats.
paralleldd on ext4
------------------
This workload uses varying numbers of dd instances to read large amounts of
data from disk.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Amean Elapsd-1 186.04 ( 0.00%) 189.41 ( -1.82%)
Amean Elapsd-3 192.27 ( 0.00%) 191.38 ( 0.46%)
Amean Elapsd-5 185.21 ( 0.00%) 182.75 ( 1.33%)
Amean Elapsd-7 183.71 ( 0.00%) 182.11 ( 0.87%)
Amean Elapsd-12 180.96 ( 0.00%) 181.58 ( -0.35%)
Amean Elapsd-16 181.36 ( 0.00%) 183.72 ( -1.30%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
User 1548.01 1552.44
System 8609.71 8515.08
Elapsed 3587.10 3594.54
There is little or no change in performance but some drop in system CPU usage.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Minor Faults 362662 367360
Major Faults 1204 1143
Swap Ins 22 0
Swap Outs 2855 1029
DMA allocs 0 0
DMA32 allocs 31409797 28837521
Normal allocs 46611853 49231282
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 40845270 40869088
Kswapd pages reclaimed 40830976 40855294
Direct pages reclaimed 0 0
Kswapd efficiency 99% 99%
Kswapd velocity 11386.711 11369.769
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Page writes by reclaim 2855 1029
Page writes file 0 0
Page writes anon 2855 1029
Page reclaim immediate 771 1628
Sector Reads 293312636 293536360
Sector Writes 18213568 18186480
Page rescued immediate 0 0
Slabs scanned 128257 132747
Direct inode steals 181 56
Kswapd inode steals 59 1131
It basically shows that kswapd was active at roughly the same rate in
both kernels. There was also comparable slab scanning activity and direct
reclaim was avoided in both cases. There appears to be a large difference
in numbers of inodes reclaimed but the workload has few active inodes and
is likely a timing artifact.
stutter
-------
stutter simulates a simple workload. One part uses a lot of anonymous
memory, a second measures mmap latency and a third copies a large file.
The primary metric is checking for mmap latency.
stutter
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Min mmap 16.6283 ( 0.00%) 13.4258 ( 19.26%)
1st-qrtle mmap 54.7570 ( 0.00%) 34.9121 ( 36.24%)
2nd-qrtle mmap 57.3163 ( 0.00%) 46.1147 ( 19.54%)
3rd-qrtle mmap 58.9976 ( 0.00%) 47.1882 ( 20.02%)
Max-90% mmap 59.7433 ( 0.00%) 47.4453 ( 20.58%)
Max-93% mmap 60.1298 ( 0.00%) 47.6037 ( 20.83%)
Max-95% mmap 73.4112 ( 0.00%) 82.8719 (-12.89%)
Max-99% mmap 92.8542 ( 0.00%) 88.8870 ( 4.27%)
Max mmap 1440.6569 ( 0.00%) 121.4201 ( 91.57%)
Mean mmap 59.3493 ( 0.00%) 42.2991 ( 28.73%)
Best99%Mean mmap 57.2121 ( 0.00%) 41.8207 ( 26.90%)
Best95%Mean mmap 55.9113 ( 0.00%) 39.9620 ( 28.53%)
Best90%Mean mmap 55.6199 ( 0.00%) 39.3124 ( 29.32%)
Best50%Mean mmap 53.2183 ( 0.00%) 33.1307 ( 37.75%)
Best10%Mean mmap 45.9842 ( 0.00%) 20.4040 ( 55.63%)
Best5%Mean mmap 43.2256 ( 0.00%) 17.9654 ( 58.44%)
Best1%Mean mmap 32.9388 ( 0.00%) 16.6875 ( 49.34%)
This shows a number of improvements with the worst-case outlier greatly
improved.
Some of the vmstats are interesting
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Swap Ins 163 502
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 618719206 1381662383
Normal allocs 891235743 564138421
Movable allocs 0 0
Allocation stalls 2603 1
Direct pages scanned 216787 2
Kswapd pages scanned 50719775 41778378
Kswapd pages reclaimed 41541765 41777639
Direct pages reclaimed 209159 0
Kswapd efficiency 81% 99%
Kswapd velocity 16859.554 14329.059
Direct efficiency 96% 0%
Direct velocity 72.061 0.001
Percentage direct scans 0% 0%
Page writes by reclaim 6215049 0
Page writes file 6215049 0
Page writes anon 0 0
Page reclaim immediate 70673 90
Sector Reads 81940800 81680456
Sector Writes 100158984 98816036
Page rescued immediate 0 0
Slabs scanned 1366954 22683
While this is not guaranteed in all cases, this particular test showed
a large reduction in direct reclaim activity. It's also worth noting
that no page writes were issued from reclaim context.
This series is not without its hazards. There are at least three areas
that I'm concerned with even though I could not reproduce any problems in
that area.
1. Reclaim/compaction is going to be affected because the amount of reclaim is
no longer targetted at a specific zone. Compaction works on a per-zone basis
so there is no guarantee that reclaiming a few THP's worth page pages will
have a positive impact on compaction success rates.
2. The Slab/LRU reclaim ratio is affected because the frequency the shrinkers
are called is now different. This may or may not be a problem but if it
is, it'll be because shrinkers are not called enough and some balancing
is required.
3. The anon/file reclaim ratio may be affected. Pages about to be dirtied are
distributed between zones and the fair zone allocation policy used to do
something very similar for anon. The distribution is now different but not
necessarily in any way that matters but it's still worth bearing in mind.
VM statistic counters for reclaim decisions are zone-based. If the kernel
is to reclaim on a per-node basis then we need to track per-node
statistics but there is no infrastructure for that. The most notable
change is that the old node_page_state is renamed to
sum_zone_node_page_state. The new node_page_state takes a pglist_data and
uses per-node stats but none exist yet. There is some renaming such as
vm_stat to vm_zone_stat and the addition of vm_node_stat and the renaming
of mod_state to mod_zone_state. Otherwise, this is mostly a mechanical
patch with no functional change. There is a lot of similarity between the
node and zone helpers which is unfortunate but there was no obvious way of
reusing the code and maintaining type safety.
Link: http://lkml.kernel.org/r/1467970510-21195-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:45:24 +00:00
|
|
|
enum node_stat_item {
|
2016-07-28 22:45:31 +00:00
|
|
|
NR_LRU_BASE,
|
|
|
|
NR_INACTIVE_ANON = NR_LRU_BASE, /* must match order of LRU_[IN]ACTIVE */
|
|
|
|
NR_ACTIVE_ANON, /* " " " " " */
|
|
|
|
NR_INACTIVE_FILE, /* " " " " " */
|
|
|
|
NR_ACTIVE_FILE, /* " " " " " */
|
|
|
|
NR_UNEVICTABLE, /* " " " " " */
|
2020-08-07 06:20:39 +00:00
|
|
|
NR_SLAB_RECLAIMABLE_B,
|
|
|
|
NR_SLAB_UNRECLAIMABLE_B,
|
2016-07-28 22:45:31 +00:00
|
|
|
NR_ISOLATED_ANON, /* Temporary isolated pages from anon lru */
|
|
|
|
NR_ISOLATED_FILE, /* Temporary isolated pages from file lru */
|
2018-10-26 22:06:39 +00:00
|
|
|
WORKINGSET_NODES,
|
2020-08-12 01:30:43 +00:00
|
|
|
WORKINGSET_REFAULT_BASE,
|
|
|
|
WORKINGSET_REFAULT_ANON = WORKINGSET_REFAULT_BASE,
|
|
|
|
WORKINGSET_REFAULT_FILE,
|
|
|
|
WORKINGSET_ACTIVATE_BASE,
|
|
|
|
WORKINGSET_ACTIVATE_ANON = WORKINGSET_ACTIVATE_BASE,
|
|
|
|
WORKINGSET_ACTIVATE_FILE,
|
|
|
|
WORKINGSET_RESTORE_BASE,
|
|
|
|
WORKINGSET_RESTORE_ANON = WORKINGSET_RESTORE_BASE,
|
|
|
|
WORKINGSET_RESTORE_FILE,
|
2016-07-28 22:46:08 +00:00
|
|
|
WORKINGSET_NODERECLAIM,
|
2016-07-28 22:46:17 +00:00
|
|
|
NR_ANON_MAPPED, /* Mapped anonymous pages */
|
2016-07-28 22:46:14 +00:00
|
|
|
NR_FILE_MAPPED, /* pagecache pages mapped into pagetables.
|
|
|
|
only modified from process context */
|
2016-07-28 22:46:20 +00:00
|
|
|
NR_FILE_PAGES,
|
|
|
|
NR_FILE_DIRTY,
|
|
|
|
NR_WRITEBACK,
|
|
|
|
NR_WRITEBACK_TEMP, /* Writeback using temporary buffers */
|
|
|
|
NR_SHMEM, /* shmem pages (included tmpfs/GEM pages) */
|
|
|
|
NR_SHMEM_THPS,
|
|
|
|
NR_SHMEM_PMDMAPPED,
|
2019-09-23 22:37:54 +00:00
|
|
|
NR_FILE_THPS,
|
|
|
|
NR_FILE_PMDMAPPED,
|
2016-07-28 22:46:20 +00:00
|
|
|
NR_ANON_THPS,
|
2016-07-28 22:46:23 +00:00
|
|
|
NR_VMSCAN_WRITE,
|
|
|
|
NR_VMSCAN_IMMEDIATE, /* Prioritise for reclaim when writeback ends */
|
|
|
|
NR_DIRTIED, /* page dirtyings since bootup */
|
|
|
|
NR_WRITTEN, /* page writings since bootup */
|
mm/vmscan: throttle reclaim until some writeback completes if congested
Patch series "Remove dependency on congestion_wait in mm/", v5.
This series that removes all calls to congestion_wait in mm/ and deletes
wait_iff_congested. It's not a clever implementation but
congestion_wait has been broken for a long time [1].
Even if congestion throttling worked, it was never a great idea. While
excessive dirty/writeback pages at the tail of the LRU is one
possibility that reclaim may be slow, there is also the problem of too
many pages being isolated and reclaim failing for other reasons
(elevated references, too many pages isolated, excessive LRU contention
etc).
This series replaces the "congestion" throttling with 3 different types.
- If there are too many dirty/writeback pages, sleep until a timeout or
enough pages get cleaned
- If too many pages are isolated, sleep until enough isolated pages are
either reclaimed or put back on the LRU
- If no progress is being made, direct reclaim tasks sleep until
another task makes progress with acceptable efficiency.
This was initially tested with a mix of workloads that used to trigger
corner cases that no longer work. A new test case was created called
"stutterp" (pagereclaim-stutterp-noreaders in mmtests) using a freshly
created XFS filesystem. Note that it may be necessary to increase the
timeout of ssh if executing remotely as ssh itself can get throttled and
the connection may timeout.
stutterp varies the number of "worker" processes from 4 up to NR_CPUS*4
to check the impact as the number of direct reclaimers increase. It has
four types of worker.
- One "anon latency" worker creates small mappings with mmap() and
times how long it takes to fault the mapping reading it 4K at a time
- X file writers which is fio randomly writing X files where the total
size of the files add up to the allowed dirty_ratio. fio is allowed
to run for a warmup period to allow some file-backed pages to
accumulate. The duration of the warmup is based on the best-case
linear write speed of the storage.
- Y file readers which is fio randomly reading small files
- Z anon memory hogs which continually map (100-dirty_ratio)% of memory
- Total estimated WSS = (100+dirty_ration) percentage of memory
X+Y+Z+1 == NR_WORKERS varying from 4 up to NR_CPUS*4
The intent is to maximise the total WSS with a mix of file and anon
memory where some anonymous memory must be swapped and there is a high
likelihood of dirty/writeback pages reaching the end of the LRU.
The test can be configured to have no background readers to stress
dirty/writeback pages. The results below are based on having zero
readers.
The short summary of the results is that the series works and stalls
until some event occurs but the timeouts may need adjustment.
The test results are not broken down by patch as the series should be
treated as one block that replaces a broken throttling mechanism with a
working one.
Finally, three machines were tested but I'm reporting the worst set of
results. The other two machines had much better latencies for example.
First the results of the "anon latency" latency
stutterp
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r4
Amean mmap-4 31.4003 ( 0.00%) 2661.0198 (-8374.52%)
Amean mmap-7 38.1641 ( 0.00%) 149.2891 (-291.18%)
Amean mmap-12 60.0981 ( 0.00%) 187.8105 (-212.51%)
Amean mmap-21 161.2699 ( 0.00%) 213.9107 ( -32.64%)
Amean mmap-30 174.5589 ( 0.00%) 377.7548 (-116.41%)
Amean mmap-48 8106.8160 ( 0.00%) 1070.5616 ( 86.79%)
Stddev mmap-4 41.3455 ( 0.00%) 27573.9676 (-66591.66%)
Stddev mmap-7 53.5556 ( 0.00%) 4608.5860 (-8505.23%)
Stddev mmap-12 171.3897 ( 0.00%) 5559.4542 (-3143.75%)
Stddev mmap-21 1506.6752 ( 0.00%) 5746.2507 (-281.39%)
Stddev mmap-30 557.5806 ( 0.00%) 7678.1624 (-1277.05%)
Stddev mmap-48 61681.5718 ( 0.00%) 14507.2830 ( 76.48%)
Max-90 mmap-4 31.4243 ( 0.00%) 83.1457 (-164.59%)
Max-90 mmap-7 41.0410 ( 0.00%) 41.0720 ( -0.08%)
Max-90 mmap-12 66.5255 ( 0.00%) 53.9073 ( 18.97%)
Max-90 mmap-21 146.7479 ( 0.00%) 105.9540 ( 27.80%)
Max-90 mmap-30 193.9513 ( 0.00%) 64.3067 ( 66.84%)
Max-90 mmap-48 277.9137 ( 0.00%) 591.0594 (-112.68%)
Max mmap-4 1913.8009 ( 0.00%) 299623.9695 (-15555.96%)
Max mmap-7 2423.9665 ( 0.00%) 204453.1708 (-8334.65%)
Max mmap-12 6845.6573 ( 0.00%) 221090.3366 (-3129.64%)
Max mmap-21 56278.6508 ( 0.00%) 213877.3496 (-280.03%)
Max mmap-30 19716.2990 ( 0.00%) 216287.6229 (-997.00%)
Max mmap-48 477923.9400 ( 0.00%) 245414.8238 ( 48.65%)
For most thread counts, the time to mmap() is unfortunately increased.
In earlier versions of the series, this was lower but a large number of
throttling events were reaching their timeout increasing the amount of
inefficient scanning of the LRU. There is no prioritisation of reclaim
tasks making progress based on each tasks rate of page allocation versus
progress of reclaim. The variance is also impacted for high worker
counts but in all cases, the differences in latency are not
statistically significant due to very large maximum outliers. Max-90
shows that 90% of the stalls are comparable but the Max results show the
massive outliers which are increased to to stalling.
It is expected that this will be very machine dependant. Due to the
test design, reclaim is difficult so allocations stall and there are
variances depending on whether THPs can be allocated or not. The amount
of memory will affect exactly how bad the corner cases are and how often
they trigger. The warmup period calculation is not ideal as it's based
on linear writes where as fio is randomly writing multiple files from
multiple tasks so the start state of the test is variable. For example,
these are the latencies on a single-socket machine that had more memory
Amean mmap-4 42.2287 ( 0.00%) 49.6838 * -17.65%*
Amean mmap-7 216.4326 ( 0.00%) 47.4451 * 78.08%*
Amean mmap-12 2412.0588 ( 0.00%) 51.7497 ( 97.85%)
Amean mmap-21 5546.2548 ( 0.00%) 51.8862 ( 99.06%)
Amean mmap-30 1085.3121 ( 0.00%) 72.1004 ( 93.36%)
The overall system CPU usage and elapsed time is as follows
5.15.0-rc3 5.15.0-rc3
vanilla mm-reclaimcongest-v5r4
Duration User 6989.03 983.42
Duration System 7308.12 799.68
Duration Elapsed 2277.67 2092.98
The patches reduce system CPU usage by 89% as the vanilla kernel is rarely
stalling.
The high-level /proc/vmstats show
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r2
Ops Direct pages scanned 1056608451.00 503594991.00
Ops Kswapd pages scanned 109795048.00 147289810.00
Ops Kswapd pages reclaimed 63269243.00 31036005.00
Ops Direct pages reclaimed 10803973.00 6328887.00
Ops Kswapd efficiency % 57.62 21.07
Ops Kswapd velocity 48204.98 57572.86
Ops Direct efficiency % 1.02 1.26
Ops Direct velocity 463898.83 196845.97
Kswapd scanned less pages but the detailed pattern is different. The
vanilla kernel scans slowly over time where as the patches exhibits
burst patterns of scan activity. Direct reclaim scanning is reduced by
52% due to stalling.
The pattern for stealing pages is also slightly different. Both kernels
exhibit spikes but the vanilla kernel when reclaiming shows pages being
reclaimed over a period of time where as the patches tend to reclaim in
spikes. The difference is that vanilla is not throttling and instead
scanning constantly finding some pages over time where as the patched
kernel throttles and reclaims in spikes.
Ops Percentage direct scans 90.59 77.37
For direct reclaim, vanilla scanned 90.59% of pages where as with the
patches, 77.37% were direct reclaim due to throttling
Ops Page writes by reclaim 2613590.00 1687131.00
Page writes from reclaim context are reduced.
Ops Page writes anon 2932752.00 1917048.00
And there is less swapping.
Ops Page reclaim immediate 996248528.00 107664764.00
The number of pages encountered at the tail of the LRU tagged for
immediate reclaim but still dirty/writeback is reduced by 89%.
Ops Slabs scanned 164284.00 153608.00
Slab scan activity is similar.
ftrace was used to gather stall activity
Vanilla
-------
1 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=16000
2 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=12000
8 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=8000
29 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=4000
82394 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=0
The fast majority of wait_iff_congested calls do not stall at all. What
is likely happening is that cond_resched() reschedules the task for a
short period when the BDI is not registering congestion (which it never
will in this test setup).
1 writeback_congestion_wait: usec_timeout=100000 usec_delayed=120000
2 writeback_congestion_wait: usec_timeout=100000 usec_delayed=132000
4 writeback_congestion_wait: usec_timeout=100000 usec_delayed=112000
380 writeback_congestion_wait: usec_timeout=100000 usec_delayed=108000
778 writeback_congestion_wait: usec_timeout=100000 usec_delayed=104000
congestion_wait if called always exceeds the timeout as there is no
trigger to wake it up.
Bottom line: Vanilla will throttle but it's not effective.
Patch series
------------
Kswapd throttle activity was always due to scanning pages tagged for
immediate reclaim at the tail of the LRU
1 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
4 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
94 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
112 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority of events did not stall or stalled for a short period.
Roughly 16% of stalls reached the timeout before expiry. For direct
reclaim, the number of times stalled for each reason were
6624 reason=VMSCAN_THROTTLE_ISOLATED
93246 reason=VMSCAN_THROTTLE_NOPROGRESS
96934 reason=VMSCAN_THROTTLE_WRITEBACK
The most common reason to stall was due to excessive pages tagged for
immediate reclaim at the tail of the LRU followed by a failure to make
forward. A relatively small number were due to too many pages isolated
from the LRU by parallel threads
For VMSCAN_THROTTLE_ISOLATED, the breakdown of delays was
9 usec_timeout=20000 usect_delayed=4000 reason=VMSCAN_THROTTLE_ISOLATED
12 usec_timeout=20000 usect_delayed=16000 reason=VMSCAN_THROTTLE_ISOLATED
83 usec_timeout=20000 usect_delayed=20000 reason=VMSCAN_THROTTLE_ISOLATED
6520 usec_timeout=20000 usect_delayed=0 reason=VMSCAN_THROTTLE_ISOLATED
Most did not stall at all. A small number reached the timeout.
For VMSCAN_THROTTLE_NOPROGRESS, the breakdown of stalls were all over
the map
1 usec_timeout=500000 usect_delayed=324000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=332000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=348000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=360000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=228000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=260000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=340000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=364000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=372000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=428000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=460000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=464000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=244000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=252000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=272000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=188000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=268000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=328000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=380000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=392000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=432000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=204000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=220000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=412000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=436000 reason=VMSCAN_THROTTLE_NOPROGRESS
6 usec_timeout=500000 usect_delayed=488000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=212000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=300000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=316000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=472000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=248000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=356000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=456000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=124000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=376000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=484000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=172000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=420000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=452000 reason=VMSCAN_THROTTLE_NOPROGRESS
11 usec_timeout=500000 usect_delayed=256000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=112000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=116000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=144000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=152000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=264000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=384000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=424000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=492000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=184000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=444000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=308000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=440000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=476000 reason=VMSCAN_THROTTLE_NOPROGRESS
16 usec_timeout=500000 usect_delayed=140000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=232000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=240000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=280000 reason=VMSCAN_THROTTLE_NOPROGRESS
18 usec_timeout=500000 usect_delayed=404000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=148000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=216000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=468000 reason=VMSCAN_THROTTLE_NOPROGRESS
21 usec_timeout=500000 usect_delayed=448000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=168000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=296000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=132000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=352000 reason=VMSCAN_THROTTLE_NOPROGRESS
26 usec_timeout=500000 usect_delayed=180000 reason=VMSCAN_THROTTLE_NOPROGRESS
27 usec_timeout=500000 usect_delayed=284000 reason=VMSCAN_THROTTLE_NOPROGRESS
28 usec_timeout=500000 usect_delayed=164000 reason=VMSCAN_THROTTLE_NOPROGRESS
29 usec_timeout=500000 usect_delayed=136000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=200000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=400000 reason=VMSCAN_THROTTLE_NOPROGRESS
31 usec_timeout=500000 usect_delayed=196000 reason=VMSCAN_THROTTLE_NOPROGRESS
32 usec_timeout=500000 usect_delayed=156000 reason=VMSCAN_THROTTLE_NOPROGRESS
33 usec_timeout=500000 usect_delayed=224000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=128000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=176000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=368000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=496000 reason=VMSCAN_THROTTLE_NOPROGRESS
37 usec_timeout=500000 usect_delayed=312000 reason=VMSCAN_THROTTLE_NOPROGRESS
38 usec_timeout=500000 usect_delayed=304000 reason=VMSCAN_THROTTLE_NOPROGRESS
40 usec_timeout=500000 usect_delayed=288000 reason=VMSCAN_THROTTLE_NOPROGRESS
43 usec_timeout=500000 usect_delayed=408000 reason=VMSCAN_THROTTLE_NOPROGRESS
55 usec_timeout=500000 usect_delayed=416000 reason=VMSCAN_THROTTLE_NOPROGRESS
56 usec_timeout=500000 usect_delayed=76000 reason=VMSCAN_THROTTLE_NOPROGRESS
58 usec_timeout=500000 usect_delayed=120000 reason=VMSCAN_THROTTLE_NOPROGRESS
59 usec_timeout=500000 usect_delayed=208000 reason=VMSCAN_THROTTLE_NOPROGRESS
61 usec_timeout=500000 usect_delayed=68000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=192000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=480000 reason=VMSCAN_THROTTLE_NOPROGRESS
79 usec_timeout=500000 usect_delayed=60000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=320000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=92000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=64000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=80000 reason=VMSCAN_THROTTLE_NOPROGRESS
88 usec_timeout=500000 usect_delayed=84000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=160000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=292000 reason=VMSCAN_THROTTLE_NOPROGRESS
94 usec_timeout=500000 usect_delayed=56000 reason=VMSCAN_THROTTLE_NOPROGRESS
118 usec_timeout=500000 usect_delayed=88000 reason=VMSCAN_THROTTLE_NOPROGRESS
119 usec_timeout=500000 usect_delayed=72000 reason=VMSCAN_THROTTLE_NOPROGRESS
126 usec_timeout=500000 usect_delayed=108000 reason=VMSCAN_THROTTLE_NOPROGRESS
146 usec_timeout=500000 usect_delayed=52000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=36000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=48000 reason=VMSCAN_THROTTLE_NOPROGRESS
159 usec_timeout=500000 usect_delayed=28000 reason=VMSCAN_THROTTLE_NOPROGRESS
178 usec_timeout=500000 usect_delayed=44000 reason=VMSCAN_THROTTLE_NOPROGRESS
183 usec_timeout=500000 usect_delayed=40000 reason=VMSCAN_THROTTLE_NOPROGRESS
237 usec_timeout=500000 usect_delayed=100000 reason=VMSCAN_THROTTLE_NOPROGRESS
266 usec_timeout=500000 usect_delayed=32000 reason=VMSCAN_THROTTLE_NOPROGRESS
313 usec_timeout=500000 usect_delayed=24000 reason=VMSCAN_THROTTLE_NOPROGRESS
347 usec_timeout=500000 usect_delayed=96000 reason=VMSCAN_THROTTLE_NOPROGRESS
470 usec_timeout=500000 usect_delayed=20000 reason=VMSCAN_THROTTLE_NOPROGRESS
559 usec_timeout=500000 usect_delayed=16000 reason=VMSCAN_THROTTLE_NOPROGRESS
964 usec_timeout=500000 usect_delayed=12000 reason=VMSCAN_THROTTLE_NOPROGRESS
2001 usec_timeout=500000 usect_delayed=104000 reason=VMSCAN_THROTTLE_NOPROGRESS
2447 usec_timeout=500000 usect_delayed=8000 reason=VMSCAN_THROTTLE_NOPROGRESS
7888 usec_timeout=500000 usect_delayed=4000 reason=VMSCAN_THROTTLE_NOPROGRESS
22727 usec_timeout=500000 usect_delayed=0 reason=VMSCAN_THROTTLE_NOPROGRESS
51305 usec_timeout=500000 usect_delayed=500000 reason=VMSCAN_THROTTLE_NOPROGRESS
The full timeout is often hit but a large number also do not stall at
all. The remainder slept a little allowing other reclaim tasks to make
progress.
While this timeout could be further increased, it could also negatively
impact worst-case behaviour when there is no prioritisation of what task
should make progress.
For VMSCAN_THROTTLE_WRITEBACK, the breakdown was
1 usec_timeout=100000 usect_delayed=44000 reason=VMSCAN_THROTTLE_WRITEBACK
2 usec_timeout=100000 usect_delayed=76000 reason=VMSCAN_THROTTLE_WRITEBACK
3 usec_timeout=100000 usect_delayed=80000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=48000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=84000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
7 usec_timeout=100000 usect_delayed=88000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=56000 reason=VMSCAN_THROTTLE_WRITEBACK
12 usec_timeout=100000 usect_delayed=64000 reason=VMSCAN_THROTTLE_WRITEBACK
16 usec_timeout=100000 usect_delayed=92000 reason=VMSCAN_THROTTLE_WRITEBACK
24 usec_timeout=100000 usect_delayed=68000 reason=VMSCAN_THROTTLE_WRITEBACK
28 usec_timeout=100000 usect_delayed=32000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=60000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=96000 reason=VMSCAN_THROTTLE_WRITEBACK
32 usec_timeout=100000 usect_delayed=52000 reason=VMSCAN_THROTTLE_WRITEBACK
42 usec_timeout=100000 usect_delayed=40000 reason=VMSCAN_THROTTLE_WRITEBACK
77 usec_timeout=100000 usect_delayed=28000 reason=VMSCAN_THROTTLE_WRITEBACK
99 usec_timeout=100000 usect_delayed=36000 reason=VMSCAN_THROTTLE_WRITEBACK
137 usec_timeout=100000 usect_delayed=24000 reason=VMSCAN_THROTTLE_WRITEBACK
190 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
339 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
518 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
852 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
3359 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
7147 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
83962 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority hit the timeout in direct reclaim context although a
sizable number did not stall at all. This is very different to kswapd
where only a tiny percentage of stalls due to writeback reached the
timeout.
Bottom line, the throttling appears to work and the wakeup events may
limit worst case stalls. There might be some grounds for adjusting
timeouts but it's likely futile as the worst-case scenarios depend on
the workload, memory size and the speed of the storage. A better
approach to improve the series further would be to prioritise tasks
based on their rate of allocation with the caveat that it may be very
expensive to track.
This patch (of 5):
Page reclaim throttles on wait_iff_congested under the following
conditions:
- kswapd is encountering pages under writeback and marked for immediate
reclaim implying that pages are cycling through the LRU faster than
pages can be cleaned.
- Direct reclaim will stall if all dirty pages are backed by congested
inodes.
wait_iff_congested is almost completely broken with few exceptions.
This patch adds a new node-based workqueue and tracks the number of
throttled tasks and pages written back since throttling started. If
enough pages belonging to the node are written back then the throttled
tasks will wake early. If not, the throttled tasks sleeps until the
timeout expires.
[neilb@suse.de: Uninterruptible sleep and simpler wakeups]
[hdanton@sina.com: Avoid race when reclaim starts]
[vbabka@suse.cz: vmstat irq-safe api, clarifications]
Link: https://lore.kernel.org/linux-mm/45d8b7a6-8548-65f5-cccf-9f451d4ae3d4@kernel.dk/ [1]
Link: https://lkml.kernel.org/r/20211022144651.19914-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20211022144651.19914-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: NeilBrown <neilb@suse.de>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: "Darrick J . Wong" <djwong@kernel.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 20:42:25 +00:00
|
|
|
NR_THROTTLED_WRITTEN, /* NR_WRITTEN while reclaim throttled */
|
2018-10-26 22:05:46 +00:00
|
|
|
NR_KERNEL_MISC_RECLAIMABLE, /* reclaimable non-slab kernel pages */
|
2020-04-02 04:05:37 +00:00
|
|
|
NR_FOLL_PIN_ACQUIRED, /* via: pin_user_page(), gup flag: FOLL_PIN */
|
|
|
|
NR_FOLL_PIN_RELEASED, /* pages returned via unpin_user_page() */
|
2020-08-07 06:21:37 +00:00
|
|
|
NR_KERNEL_STACK_KB, /* measured in KiB */
|
|
|
|
#if IS_ENABLED(CONFIG_SHADOW_CALL_STACK)
|
|
|
|
NR_KERNEL_SCS_KB, /* measured in KiB */
|
|
|
|
#endif
|
2020-12-15 03:07:17 +00:00
|
|
|
NR_PAGETABLE, /* used for pagetables */
|
2021-02-24 20:03:55 +00:00
|
|
|
#ifdef CONFIG_SWAP
|
|
|
|
NR_SWAPCACHE,
|
|
|
|
#endif
|
mm, vmstat: add infrastructure for per-node vmstats
Patchset: "Move LRU page reclaim from zones to nodes v9"
This series moves LRUs from the zones to the node. While this is a
current rebase, the test results were based on mmotm as of June 23rd.
Conceptually, this series is simple but there are a lot of details.
Some of the broad motivations for this are;
1. The residency of a page partially depends on what zone the page was
allocated from. This is partially combatted by the fair zone allocation
policy but that is a partial solution that introduces overhead in the
page allocator paths.
2. Currently, reclaim on node 0 behaves slightly different to node 1. For
example, direct reclaim scans in zonelist order and reclaims even if
the zone is over the high watermark regardless of the age of pages
in that LRU. Kswapd on the other hand starts reclaim on the highest
unbalanced zone. A difference in distribution of file/anon pages due
to when they were allocated results can result in a difference in
again. While the fair zone allocation policy mitigates some of the
problems here, the page reclaim results on a multi-zone node will
always be different to a single-zone node.
it was scheduled on as a result.
3. kswapd and the page allocator scan zones in the opposite order to
avoid interfering with each other but it's sensitive to timing. This
mitigates the page allocator using pages that were allocated very recently
in the ideal case but it's sensitive to timing. When kswapd is allocating
from lower zones then it's great but during the rebalancing of the highest
zone, the page allocator and kswapd interfere with each other. It's worse
if the highest zone is small and difficult to balance.
4. slab shrinkers are node-based which makes it harder to identify the exact
relationship between slab reclaim and LRU reclaim.
The reason we have zone-based reclaim is that we used to have
large highmem zones in common configurations and it was necessary
to quickly find ZONE_NORMAL pages for reclaim. Today, this is much
less of a concern as machines with lots of memory will (or should) use
64-bit kernels. Combinations of 32-bit hardware and 64-bit hardware are
rare. Machines that do use highmem should have relatively low highmem:lowmem
ratios than we worried about in the past.
Conceptually, moving to node LRUs should be easier to understand. The
page allocator plays fewer tricks to game reclaim and reclaim behaves
similarly on all nodes.
The series has been tested on a 16 core UMA machine and a 2-socket 48
core NUMA machine. The UMA results are presented in most cases as the NUMA
machine behaved similarly.
pagealloc
---------
This is a microbenchmark that shows the benefit of removing the fair zone
allocation policy. It was tested uip to order-4 but only orders 0 and 1 are
shown as the other orders were comparable.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min total-odr0-1 490.00 ( 0.00%) 457.00 ( 6.73%)
Min total-odr0-2 347.00 ( 0.00%) 329.00 ( 5.19%)
Min total-odr0-4 288.00 ( 0.00%) 273.00 ( 5.21%)
Min total-odr0-8 251.00 ( 0.00%) 239.00 ( 4.78%)
Min total-odr0-16 234.00 ( 0.00%) 222.00 ( 5.13%)
Min total-odr0-32 223.00 ( 0.00%) 211.00 ( 5.38%)
Min total-odr0-64 217.00 ( 0.00%) 208.00 ( 4.15%)
Min total-odr0-128 214.00 ( 0.00%) 204.00 ( 4.67%)
Min total-odr0-256 250.00 ( 0.00%) 230.00 ( 8.00%)
Min total-odr0-512 271.00 ( 0.00%) 269.00 ( 0.74%)
Min total-odr0-1024 291.00 ( 0.00%) 282.00 ( 3.09%)
Min total-odr0-2048 303.00 ( 0.00%) 296.00 ( 2.31%)
Min total-odr0-4096 311.00 ( 0.00%) 309.00 ( 0.64%)
Min total-odr0-8192 316.00 ( 0.00%) 314.00 ( 0.63%)
Min total-odr0-16384 317.00 ( 0.00%) 315.00 ( 0.63%)
Min total-odr1-1 742.00 ( 0.00%) 712.00 ( 4.04%)
Min total-odr1-2 562.00 ( 0.00%) 530.00 ( 5.69%)
Min total-odr1-4 457.00 ( 0.00%) 433.00 ( 5.25%)
Min total-odr1-8 411.00 ( 0.00%) 381.00 ( 7.30%)
Min total-odr1-16 381.00 ( 0.00%) 356.00 ( 6.56%)
Min total-odr1-32 372.00 ( 0.00%) 346.00 ( 6.99%)
Min total-odr1-64 372.00 ( 0.00%) 343.00 ( 7.80%)
Min total-odr1-128 375.00 ( 0.00%) 351.00 ( 6.40%)
Min total-odr1-256 379.00 ( 0.00%) 351.00 ( 7.39%)
Min total-odr1-512 385.00 ( 0.00%) 355.00 ( 7.79%)
Min total-odr1-1024 386.00 ( 0.00%) 358.00 ( 7.25%)
Min total-odr1-2048 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-4096 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-8192 388.00 ( 0.00%) 363.00 ( 6.44%)
This shows a steady improvement throughout. The primary benefit is from
reduced system CPU usage which is obvious from the overall times;
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
User 189.19 191.80
System 2604.45 2533.56
Elapsed 2855.30 2786.39
The vmstats also showed that the fair zone allocation policy was definitely
removed as can be seen here;
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v8
DMA32 allocs 28794729769 0
Normal allocs 48432501431 77227309877
Movable allocs 0 0
tiobench on ext4
----------------
tiobench is a benchmark that artifically benefits if old pages remain resident
while new pages get reclaimed. The fair zone allocation policy mitigates this
problem so pages age fairly. While the benchmark has problems, it is important
that tiobench performance remains constant as it implies that page aging
problems that the fair zone allocation policy fixes are not re-introduced.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min PotentialReadSpeed 89.65 ( 0.00%) 90.21 ( 0.62%)
Min SeqRead-MB/sec-1 82.68 ( 0.00%) 82.01 ( -0.81%)
Min SeqRead-MB/sec-2 72.76 ( 0.00%) 72.07 ( -0.95%)
Min SeqRead-MB/sec-4 75.13 ( 0.00%) 74.92 ( -0.28%)
Min SeqRead-MB/sec-8 64.91 ( 0.00%) 65.19 ( 0.43%)
Min SeqRead-MB/sec-16 62.24 ( 0.00%) 62.22 ( -0.03%)
Min RandRead-MB/sec-1 0.88 ( 0.00%) 0.88 ( 0.00%)
Min RandRead-MB/sec-2 0.95 ( 0.00%) 0.92 ( -3.16%)
Min RandRead-MB/sec-4 1.43 ( 0.00%) 1.34 ( -6.29%)
Min RandRead-MB/sec-8 1.61 ( 0.00%) 1.60 ( -0.62%)
Min RandRead-MB/sec-16 1.80 ( 0.00%) 1.90 ( 5.56%)
Min SeqWrite-MB/sec-1 76.41 ( 0.00%) 76.85 ( 0.58%)
Min SeqWrite-MB/sec-2 74.11 ( 0.00%) 73.54 ( -0.77%)
Min SeqWrite-MB/sec-4 80.05 ( 0.00%) 80.13 ( 0.10%)
Min SeqWrite-MB/sec-8 72.88 ( 0.00%) 73.20 ( 0.44%)
Min SeqWrite-MB/sec-16 75.91 ( 0.00%) 76.44 ( 0.70%)
Min RandWrite-MB/sec-1 1.18 ( 0.00%) 1.14 ( -3.39%)
Min RandWrite-MB/sec-2 1.02 ( 0.00%) 1.03 ( 0.98%)
Min RandWrite-MB/sec-4 1.05 ( 0.00%) 0.98 ( -6.67%)
Min RandWrite-MB/sec-8 0.89 ( 0.00%) 0.92 ( 3.37%)
Min RandWrite-MB/sec-16 0.92 ( 0.00%) 0.93 ( 1.09%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 approx-v9
User 645.72 525.90
System 403.85 331.75
Elapsed 6795.36 6783.67
This shows that the series has little or not impact on tiobench which is
desirable and a reduction in system CPU usage. It indicates that the fair
zone allocation policy was removed in a manner that didn't reintroduce
one class of page aging bug. There were only minor differences in overall
reclaim activity
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Minor Faults 645838 647465
Major Faults 573 640
Swap Ins 0 0
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 46041453 44190646
Normal allocs 78053072 79887245
Movable allocs 0 0
Allocation stalls 24 67
Stall zone DMA 0 0
Stall zone DMA32 0 0
Stall zone Normal 0 2
Stall zone HighMem 0 0
Stall zone Movable 0 65
Direct pages scanned 10969 30609
Kswapd pages scanned 93375144 93492094
Kswapd pages reclaimed 93372243 93489370
Direct pages reclaimed 10969 30609
Kswapd efficiency 99% 99%
Kswapd velocity 13741.015 13781.934
Direct efficiency 100% 100%
Direct velocity 1.614 4.512
Percentage direct scans 0% 0%
kswapd activity was roughly comparable. There were differences in direct
reclaim activity but negligible in the context of the overall workload
(velocity of 4 pages per second with the patches applied, 1.6 pages per
second in the baseline kernel).
pgbench read-only large configuration on ext4
---------------------------------------------
pgbench is a database benchmark that can be sensitive to page reclaim
decisions. This also checks if removing the fair zone allocation policy
is safe
pgbench Transactions
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Hmean 1 188.26 ( 0.00%) 189.78 ( 0.81%)
Hmean 5 330.66 ( 0.00%) 328.69 ( -0.59%)
Hmean 12 370.32 ( 0.00%) 380.72 ( 2.81%)
Hmean 21 368.89 ( 0.00%) 369.00 ( 0.03%)
Hmean 30 382.14 ( 0.00%) 360.89 ( -5.56%)
Hmean 32 428.87 ( 0.00%) 432.96 ( 0.95%)
Negligible differences again. As with tiobench, overall reclaim activity
was comparable.
bonnie++ on ext4
----------------
No interesting performance difference, negligible differences on reclaim
stats.
paralleldd on ext4
------------------
This workload uses varying numbers of dd instances to read large amounts of
data from disk.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Amean Elapsd-1 186.04 ( 0.00%) 189.41 ( -1.82%)
Amean Elapsd-3 192.27 ( 0.00%) 191.38 ( 0.46%)
Amean Elapsd-5 185.21 ( 0.00%) 182.75 ( 1.33%)
Amean Elapsd-7 183.71 ( 0.00%) 182.11 ( 0.87%)
Amean Elapsd-12 180.96 ( 0.00%) 181.58 ( -0.35%)
Amean Elapsd-16 181.36 ( 0.00%) 183.72 ( -1.30%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
User 1548.01 1552.44
System 8609.71 8515.08
Elapsed 3587.10 3594.54
There is little or no change in performance but some drop in system CPU usage.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Minor Faults 362662 367360
Major Faults 1204 1143
Swap Ins 22 0
Swap Outs 2855 1029
DMA allocs 0 0
DMA32 allocs 31409797 28837521
Normal allocs 46611853 49231282
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 40845270 40869088
Kswapd pages reclaimed 40830976 40855294
Direct pages reclaimed 0 0
Kswapd efficiency 99% 99%
Kswapd velocity 11386.711 11369.769
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Page writes by reclaim 2855 1029
Page writes file 0 0
Page writes anon 2855 1029
Page reclaim immediate 771 1628
Sector Reads 293312636 293536360
Sector Writes 18213568 18186480
Page rescued immediate 0 0
Slabs scanned 128257 132747
Direct inode steals 181 56
Kswapd inode steals 59 1131
It basically shows that kswapd was active at roughly the same rate in
both kernels. There was also comparable slab scanning activity and direct
reclaim was avoided in both cases. There appears to be a large difference
in numbers of inodes reclaimed but the workload has few active inodes and
is likely a timing artifact.
stutter
-------
stutter simulates a simple workload. One part uses a lot of anonymous
memory, a second measures mmap latency and a third copies a large file.
The primary metric is checking for mmap latency.
stutter
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Min mmap 16.6283 ( 0.00%) 13.4258 ( 19.26%)
1st-qrtle mmap 54.7570 ( 0.00%) 34.9121 ( 36.24%)
2nd-qrtle mmap 57.3163 ( 0.00%) 46.1147 ( 19.54%)
3rd-qrtle mmap 58.9976 ( 0.00%) 47.1882 ( 20.02%)
Max-90% mmap 59.7433 ( 0.00%) 47.4453 ( 20.58%)
Max-93% mmap 60.1298 ( 0.00%) 47.6037 ( 20.83%)
Max-95% mmap 73.4112 ( 0.00%) 82.8719 (-12.89%)
Max-99% mmap 92.8542 ( 0.00%) 88.8870 ( 4.27%)
Max mmap 1440.6569 ( 0.00%) 121.4201 ( 91.57%)
Mean mmap 59.3493 ( 0.00%) 42.2991 ( 28.73%)
Best99%Mean mmap 57.2121 ( 0.00%) 41.8207 ( 26.90%)
Best95%Mean mmap 55.9113 ( 0.00%) 39.9620 ( 28.53%)
Best90%Mean mmap 55.6199 ( 0.00%) 39.3124 ( 29.32%)
Best50%Mean mmap 53.2183 ( 0.00%) 33.1307 ( 37.75%)
Best10%Mean mmap 45.9842 ( 0.00%) 20.4040 ( 55.63%)
Best5%Mean mmap 43.2256 ( 0.00%) 17.9654 ( 58.44%)
Best1%Mean mmap 32.9388 ( 0.00%) 16.6875 ( 49.34%)
This shows a number of improvements with the worst-case outlier greatly
improved.
Some of the vmstats are interesting
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Swap Ins 163 502
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 618719206 1381662383
Normal allocs 891235743 564138421
Movable allocs 0 0
Allocation stalls 2603 1
Direct pages scanned 216787 2
Kswapd pages scanned 50719775 41778378
Kswapd pages reclaimed 41541765 41777639
Direct pages reclaimed 209159 0
Kswapd efficiency 81% 99%
Kswapd velocity 16859.554 14329.059
Direct efficiency 96% 0%
Direct velocity 72.061 0.001
Percentage direct scans 0% 0%
Page writes by reclaim 6215049 0
Page writes file 6215049 0
Page writes anon 0 0
Page reclaim immediate 70673 90
Sector Reads 81940800 81680456
Sector Writes 100158984 98816036
Page rescued immediate 0 0
Slabs scanned 1366954 22683
While this is not guaranteed in all cases, this particular test showed
a large reduction in direct reclaim activity. It's also worth noting
that no page writes were issued from reclaim context.
This series is not without its hazards. There are at least three areas
that I'm concerned with even though I could not reproduce any problems in
that area.
1. Reclaim/compaction is going to be affected because the amount of reclaim is
no longer targetted at a specific zone. Compaction works on a per-zone basis
so there is no guarantee that reclaiming a few THP's worth page pages will
have a positive impact on compaction success rates.
2. The Slab/LRU reclaim ratio is affected because the frequency the shrinkers
are called is now different. This may or may not be a problem but if it
is, it'll be because shrinkers are not called enough and some balancing
is required.
3. The anon/file reclaim ratio may be affected. Pages about to be dirtied are
distributed between zones and the fair zone allocation policy used to do
something very similar for anon. The distribution is now different but not
necessarily in any way that matters but it's still worth bearing in mind.
VM statistic counters for reclaim decisions are zone-based. If the kernel
is to reclaim on a per-node basis then we need to track per-node
statistics but there is no infrastructure for that. The most notable
change is that the old node_page_state is renamed to
sum_zone_node_page_state. The new node_page_state takes a pglist_data and
uses per-node stats but none exist yet. There is some renaming such as
vm_stat to vm_zone_stat and the addition of vm_node_stat and the renaming
of mod_state to mod_zone_state. Otherwise, this is mostly a mechanical
patch with no functional change. There is a lot of similarity between the
node and zone helpers which is unfortunate but there was no obvious way of
reusing the code and maintaining type safety.
Link: http://lkml.kernel.org/r/1467970510-21195-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:45:24 +00:00
|
|
|
NR_VM_NODE_STAT_ITEMS
|
|
|
|
};
|
|
|
|
|
2021-02-24 20:03:23 +00:00
|
|
|
/*
|
|
|
|
* Returns true if the item should be printed in THPs (/proc/vmstat
|
|
|
|
* currently prints number of anon, file and shmem THPs. But the item
|
|
|
|
* is charged in pages).
|
|
|
|
*/
|
|
|
|
static __always_inline bool vmstat_item_print_in_thp(enum node_stat_item item)
|
|
|
|
{
|
|
|
|
if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE))
|
|
|
|
return false;
|
|
|
|
|
2021-02-24 20:03:27 +00:00
|
|
|
return item == NR_ANON_THPS ||
|
2021-02-24 20:03:31 +00:00
|
|
|
item == NR_FILE_THPS ||
|
2021-02-24 20:03:35 +00:00
|
|
|
item == NR_SHMEM_THPS ||
|
2021-02-24 20:03:39 +00:00
|
|
|
item == NR_SHMEM_PMDMAPPED ||
|
|
|
|
item == NR_FILE_PMDMAPPED;
|
2021-02-24 20:03:23 +00:00
|
|
|
}
|
|
|
|
|
2020-08-07 06:20:35 +00:00
|
|
|
/*
|
|
|
|
* Returns true if the value is measured in bytes (most vmstat values are
|
|
|
|
* measured in pages). This defines the API part, the internal representation
|
|
|
|
* might be different.
|
|
|
|
*/
|
|
|
|
static __always_inline bool vmstat_item_in_bytes(int idx)
|
|
|
|
{
|
2020-08-07 06:20:39 +00:00
|
|
|
/*
|
|
|
|
* Global and per-node slab counters track slab pages.
|
|
|
|
* It's expected that changes are multiples of PAGE_SIZE.
|
|
|
|
* Internally values are stored in pages.
|
|
|
|
*
|
|
|
|
* Per-memcg and per-lruvec counters track memory, consumed
|
|
|
|
* by individual slab objects. These counters are actually
|
|
|
|
* byte-precise.
|
|
|
|
*/
|
|
|
|
return (idx == NR_SLAB_RECLAIMABLE_B ||
|
|
|
|
idx == NR_SLAB_UNRECLAIMABLE_B);
|
2020-08-07 06:20:35 +00:00
|
|
|
}
|
|
|
|
|
2008-10-19 03:26:32 +00:00
|
|
|
/*
|
|
|
|
* We do arithmetic on the LRU lists in various places in the code,
|
|
|
|
* so it is important to keep the active lists LRU_ACTIVE higher in
|
|
|
|
* the array than the corresponding inactive lists, and to keep
|
|
|
|
* the *_FILE lists LRU_FILE higher than the corresponding _ANON lists.
|
|
|
|
*
|
|
|
|
* This has to be kept in sync with the statistics in zone_stat_item
|
|
|
|
* above and the descriptions in vmstat_text in mm/vmstat.c
|
|
|
|
*/
|
|
|
|
#define LRU_BASE 0
|
|
|
|
#define LRU_ACTIVE 1
|
|
|
|
#define LRU_FILE 2
|
|
|
|
|
2008-10-19 03:26:14 +00:00
|
|
|
enum lru_list {
|
2008-10-19 03:26:32 +00:00
|
|
|
LRU_INACTIVE_ANON = LRU_BASE,
|
|
|
|
LRU_ACTIVE_ANON = LRU_BASE + LRU_ACTIVE,
|
|
|
|
LRU_INACTIVE_FILE = LRU_BASE + LRU_FILE,
|
|
|
|
LRU_ACTIVE_FILE = LRU_BASE + LRU_FILE + LRU_ACTIVE,
|
Unevictable LRU Infrastructure
When the system contains lots of mlocked or otherwise unevictable pages,
the pageout code (kswapd) can spend lots of time scanning over these
pages. Worse still, the presence of lots of unevictable pages can confuse
kswapd into thinking that more aggressive pageout modes are required,
resulting in all kinds of bad behaviour.
Infrastructure to manage pages excluded from reclaim--i.e., hidden from
vmscan. Based on a patch by Larry Woodman of Red Hat. Reworked to
maintain "unevictable" pages on a separate per-zone LRU list, to "hide"
them from vmscan.
Kosaki Motohiro added the support for the memory controller unevictable
lru list.
Pages on the unevictable list have both PG_unevictable and PG_lru set.
Thus, PG_unevictable is analogous to and mutually exclusive with
PG_active--it specifies which LRU list the page is on.
The unevictable infrastructure is enabled by a new mm Kconfig option
[CONFIG_]UNEVICTABLE_LRU.
A new function 'page_evictable(page, vma)' in vmscan.c tests whether or
not a page may be evictable. Subsequent patches will add the various
!evictable tests. We'll want to keep these tests light-weight for use in
shrink_active_list() and, possibly, the fault path.
To avoid races between tasks putting pages [back] onto an LRU list and
tasks that might be moving the page from non-evictable to evictable state,
the new function 'putback_lru_page()' -- inverse to 'isolate_lru_page()'
-- tests the "evictability" of a page after placing it on the LRU, before
dropping the reference. If the page has become unevictable,
putback_lru_page() will redo the 'putback', thus moving the page to the
unevictable list. This way, we avoid "stranding" evictable pages on the
unevictable list.
[akpm@linux-foundation.org: fix fallout from out-of-order merge]
[riel@redhat.com: fix UNEVICTABLE_LRU and !PROC_PAGE_MONITOR build]
[nishimura@mxp.nes.nec.co.jp: remove redundant mapping check]
[kosaki.motohiro@jp.fujitsu.com: unevictable-lru-infrastructure: putback_lru_page()/unevictable page handling rework]
[kosaki.motohiro@jp.fujitsu.com: kill unnecessary lock_page() in vmscan.c]
[kosaki.motohiro@jp.fujitsu.com: revert migration change of unevictable lru infrastructure]
[kosaki.motohiro@jp.fujitsu.com: revert to unevictable-lru-infrastructure-kconfig-fix.patch]
[kosaki.motohiro@jp.fujitsu.com: restore patch failure of vmstat-unevictable-and-mlocked-pages-vm-events.patch]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Debugged-by: Benjamin Kidwell <benjkidwell@yahoo.com>
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-10-19 03:26:39 +00:00
|
|
|
LRU_UNEVICTABLE,
|
|
|
|
NR_LRU_LISTS
|
|
|
|
};
|
2008-10-19 03:26:14 +00:00
|
|
|
|
mm/vmscan: throttle reclaim until some writeback completes if congested
Patch series "Remove dependency on congestion_wait in mm/", v5.
This series that removes all calls to congestion_wait in mm/ and deletes
wait_iff_congested. It's not a clever implementation but
congestion_wait has been broken for a long time [1].
Even if congestion throttling worked, it was never a great idea. While
excessive dirty/writeback pages at the tail of the LRU is one
possibility that reclaim may be slow, there is also the problem of too
many pages being isolated and reclaim failing for other reasons
(elevated references, too many pages isolated, excessive LRU contention
etc).
This series replaces the "congestion" throttling with 3 different types.
- If there are too many dirty/writeback pages, sleep until a timeout or
enough pages get cleaned
- If too many pages are isolated, sleep until enough isolated pages are
either reclaimed or put back on the LRU
- If no progress is being made, direct reclaim tasks sleep until
another task makes progress with acceptable efficiency.
This was initially tested with a mix of workloads that used to trigger
corner cases that no longer work. A new test case was created called
"stutterp" (pagereclaim-stutterp-noreaders in mmtests) using a freshly
created XFS filesystem. Note that it may be necessary to increase the
timeout of ssh if executing remotely as ssh itself can get throttled and
the connection may timeout.
stutterp varies the number of "worker" processes from 4 up to NR_CPUS*4
to check the impact as the number of direct reclaimers increase. It has
four types of worker.
- One "anon latency" worker creates small mappings with mmap() and
times how long it takes to fault the mapping reading it 4K at a time
- X file writers which is fio randomly writing X files where the total
size of the files add up to the allowed dirty_ratio. fio is allowed
to run for a warmup period to allow some file-backed pages to
accumulate. The duration of the warmup is based on the best-case
linear write speed of the storage.
- Y file readers which is fio randomly reading small files
- Z anon memory hogs which continually map (100-dirty_ratio)% of memory
- Total estimated WSS = (100+dirty_ration) percentage of memory
X+Y+Z+1 == NR_WORKERS varying from 4 up to NR_CPUS*4
The intent is to maximise the total WSS with a mix of file and anon
memory where some anonymous memory must be swapped and there is a high
likelihood of dirty/writeback pages reaching the end of the LRU.
The test can be configured to have no background readers to stress
dirty/writeback pages. The results below are based on having zero
readers.
The short summary of the results is that the series works and stalls
until some event occurs but the timeouts may need adjustment.
The test results are not broken down by patch as the series should be
treated as one block that replaces a broken throttling mechanism with a
working one.
Finally, three machines were tested but I'm reporting the worst set of
results. The other two machines had much better latencies for example.
First the results of the "anon latency" latency
stutterp
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r4
Amean mmap-4 31.4003 ( 0.00%) 2661.0198 (-8374.52%)
Amean mmap-7 38.1641 ( 0.00%) 149.2891 (-291.18%)
Amean mmap-12 60.0981 ( 0.00%) 187.8105 (-212.51%)
Amean mmap-21 161.2699 ( 0.00%) 213.9107 ( -32.64%)
Amean mmap-30 174.5589 ( 0.00%) 377.7548 (-116.41%)
Amean mmap-48 8106.8160 ( 0.00%) 1070.5616 ( 86.79%)
Stddev mmap-4 41.3455 ( 0.00%) 27573.9676 (-66591.66%)
Stddev mmap-7 53.5556 ( 0.00%) 4608.5860 (-8505.23%)
Stddev mmap-12 171.3897 ( 0.00%) 5559.4542 (-3143.75%)
Stddev mmap-21 1506.6752 ( 0.00%) 5746.2507 (-281.39%)
Stddev mmap-30 557.5806 ( 0.00%) 7678.1624 (-1277.05%)
Stddev mmap-48 61681.5718 ( 0.00%) 14507.2830 ( 76.48%)
Max-90 mmap-4 31.4243 ( 0.00%) 83.1457 (-164.59%)
Max-90 mmap-7 41.0410 ( 0.00%) 41.0720 ( -0.08%)
Max-90 mmap-12 66.5255 ( 0.00%) 53.9073 ( 18.97%)
Max-90 mmap-21 146.7479 ( 0.00%) 105.9540 ( 27.80%)
Max-90 mmap-30 193.9513 ( 0.00%) 64.3067 ( 66.84%)
Max-90 mmap-48 277.9137 ( 0.00%) 591.0594 (-112.68%)
Max mmap-4 1913.8009 ( 0.00%) 299623.9695 (-15555.96%)
Max mmap-7 2423.9665 ( 0.00%) 204453.1708 (-8334.65%)
Max mmap-12 6845.6573 ( 0.00%) 221090.3366 (-3129.64%)
Max mmap-21 56278.6508 ( 0.00%) 213877.3496 (-280.03%)
Max mmap-30 19716.2990 ( 0.00%) 216287.6229 (-997.00%)
Max mmap-48 477923.9400 ( 0.00%) 245414.8238 ( 48.65%)
For most thread counts, the time to mmap() is unfortunately increased.
In earlier versions of the series, this was lower but a large number of
throttling events were reaching their timeout increasing the amount of
inefficient scanning of the LRU. There is no prioritisation of reclaim
tasks making progress based on each tasks rate of page allocation versus
progress of reclaim. The variance is also impacted for high worker
counts but in all cases, the differences in latency are not
statistically significant due to very large maximum outliers. Max-90
shows that 90% of the stalls are comparable but the Max results show the
massive outliers which are increased to to stalling.
It is expected that this will be very machine dependant. Due to the
test design, reclaim is difficult so allocations stall and there are
variances depending on whether THPs can be allocated or not. The amount
of memory will affect exactly how bad the corner cases are and how often
they trigger. The warmup period calculation is not ideal as it's based
on linear writes where as fio is randomly writing multiple files from
multiple tasks so the start state of the test is variable. For example,
these are the latencies on a single-socket machine that had more memory
Amean mmap-4 42.2287 ( 0.00%) 49.6838 * -17.65%*
Amean mmap-7 216.4326 ( 0.00%) 47.4451 * 78.08%*
Amean mmap-12 2412.0588 ( 0.00%) 51.7497 ( 97.85%)
Amean mmap-21 5546.2548 ( 0.00%) 51.8862 ( 99.06%)
Amean mmap-30 1085.3121 ( 0.00%) 72.1004 ( 93.36%)
The overall system CPU usage and elapsed time is as follows
5.15.0-rc3 5.15.0-rc3
vanilla mm-reclaimcongest-v5r4
Duration User 6989.03 983.42
Duration System 7308.12 799.68
Duration Elapsed 2277.67 2092.98
The patches reduce system CPU usage by 89% as the vanilla kernel is rarely
stalling.
The high-level /proc/vmstats show
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r2
Ops Direct pages scanned 1056608451.00 503594991.00
Ops Kswapd pages scanned 109795048.00 147289810.00
Ops Kswapd pages reclaimed 63269243.00 31036005.00
Ops Direct pages reclaimed 10803973.00 6328887.00
Ops Kswapd efficiency % 57.62 21.07
Ops Kswapd velocity 48204.98 57572.86
Ops Direct efficiency % 1.02 1.26
Ops Direct velocity 463898.83 196845.97
Kswapd scanned less pages but the detailed pattern is different. The
vanilla kernel scans slowly over time where as the patches exhibits
burst patterns of scan activity. Direct reclaim scanning is reduced by
52% due to stalling.
The pattern for stealing pages is also slightly different. Both kernels
exhibit spikes but the vanilla kernel when reclaiming shows pages being
reclaimed over a period of time where as the patches tend to reclaim in
spikes. The difference is that vanilla is not throttling and instead
scanning constantly finding some pages over time where as the patched
kernel throttles and reclaims in spikes.
Ops Percentage direct scans 90.59 77.37
For direct reclaim, vanilla scanned 90.59% of pages where as with the
patches, 77.37% were direct reclaim due to throttling
Ops Page writes by reclaim 2613590.00 1687131.00
Page writes from reclaim context are reduced.
Ops Page writes anon 2932752.00 1917048.00
And there is less swapping.
Ops Page reclaim immediate 996248528.00 107664764.00
The number of pages encountered at the tail of the LRU tagged for
immediate reclaim but still dirty/writeback is reduced by 89%.
Ops Slabs scanned 164284.00 153608.00
Slab scan activity is similar.
ftrace was used to gather stall activity
Vanilla
-------
1 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=16000
2 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=12000
8 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=8000
29 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=4000
82394 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=0
The fast majority of wait_iff_congested calls do not stall at all. What
is likely happening is that cond_resched() reschedules the task for a
short period when the BDI is not registering congestion (which it never
will in this test setup).
1 writeback_congestion_wait: usec_timeout=100000 usec_delayed=120000
2 writeback_congestion_wait: usec_timeout=100000 usec_delayed=132000
4 writeback_congestion_wait: usec_timeout=100000 usec_delayed=112000
380 writeback_congestion_wait: usec_timeout=100000 usec_delayed=108000
778 writeback_congestion_wait: usec_timeout=100000 usec_delayed=104000
congestion_wait if called always exceeds the timeout as there is no
trigger to wake it up.
Bottom line: Vanilla will throttle but it's not effective.
Patch series
------------
Kswapd throttle activity was always due to scanning pages tagged for
immediate reclaim at the tail of the LRU
1 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
4 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
94 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
112 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority of events did not stall or stalled for a short period.
Roughly 16% of stalls reached the timeout before expiry. For direct
reclaim, the number of times stalled for each reason were
6624 reason=VMSCAN_THROTTLE_ISOLATED
93246 reason=VMSCAN_THROTTLE_NOPROGRESS
96934 reason=VMSCAN_THROTTLE_WRITEBACK
The most common reason to stall was due to excessive pages tagged for
immediate reclaim at the tail of the LRU followed by a failure to make
forward. A relatively small number were due to too many pages isolated
from the LRU by parallel threads
For VMSCAN_THROTTLE_ISOLATED, the breakdown of delays was
9 usec_timeout=20000 usect_delayed=4000 reason=VMSCAN_THROTTLE_ISOLATED
12 usec_timeout=20000 usect_delayed=16000 reason=VMSCAN_THROTTLE_ISOLATED
83 usec_timeout=20000 usect_delayed=20000 reason=VMSCAN_THROTTLE_ISOLATED
6520 usec_timeout=20000 usect_delayed=0 reason=VMSCAN_THROTTLE_ISOLATED
Most did not stall at all. A small number reached the timeout.
For VMSCAN_THROTTLE_NOPROGRESS, the breakdown of stalls were all over
the map
1 usec_timeout=500000 usect_delayed=324000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=332000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=348000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=360000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=228000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=260000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=340000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=364000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=372000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=428000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=460000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=464000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=244000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=252000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=272000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=188000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=268000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=328000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=380000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=392000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=432000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=204000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=220000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=412000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=436000 reason=VMSCAN_THROTTLE_NOPROGRESS
6 usec_timeout=500000 usect_delayed=488000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=212000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=300000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=316000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=472000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=248000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=356000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=456000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=124000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=376000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=484000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=172000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=420000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=452000 reason=VMSCAN_THROTTLE_NOPROGRESS
11 usec_timeout=500000 usect_delayed=256000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=112000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=116000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=144000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=152000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=264000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=384000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=424000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=492000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=184000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=444000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=308000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=440000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=476000 reason=VMSCAN_THROTTLE_NOPROGRESS
16 usec_timeout=500000 usect_delayed=140000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=232000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=240000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=280000 reason=VMSCAN_THROTTLE_NOPROGRESS
18 usec_timeout=500000 usect_delayed=404000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=148000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=216000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=468000 reason=VMSCAN_THROTTLE_NOPROGRESS
21 usec_timeout=500000 usect_delayed=448000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=168000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=296000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=132000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=352000 reason=VMSCAN_THROTTLE_NOPROGRESS
26 usec_timeout=500000 usect_delayed=180000 reason=VMSCAN_THROTTLE_NOPROGRESS
27 usec_timeout=500000 usect_delayed=284000 reason=VMSCAN_THROTTLE_NOPROGRESS
28 usec_timeout=500000 usect_delayed=164000 reason=VMSCAN_THROTTLE_NOPROGRESS
29 usec_timeout=500000 usect_delayed=136000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=200000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=400000 reason=VMSCAN_THROTTLE_NOPROGRESS
31 usec_timeout=500000 usect_delayed=196000 reason=VMSCAN_THROTTLE_NOPROGRESS
32 usec_timeout=500000 usect_delayed=156000 reason=VMSCAN_THROTTLE_NOPROGRESS
33 usec_timeout=500000 usect_delayed=224000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=128000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=176000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=368000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=496000 reason=VMSCAN_THROTTLE_NOPROGRESS
37 usec_timeout=500000 usect_delayed=312000 reason=VMSCAN_THROTTLE_NOPROGRESS
38 usec_timeout=500000 usect_delayed=304000 reason=VMSCAN_THROTTLE_NOPROGRESS
40 usec_timeout=500000 usect_delayed=288000 reason=VMSCAN_THROTTLE_NOPROGRESS
43 usec_timeout=500000 usect_delayed=408000 reason=VMSCAN_THROTTLE_NOPROGRESS
55 usec_timeout=500000 usect_delayed=416000 reason=VMSCAN_THROTTLE_NOPROGRESS
56 usec_timeout=500000 usect_delayed=76000 reason=VMSCAN_THROTTLE_NOPROGRESS
58 usec_timeout=500000 usect_delayed=120000 reason=VMSCAN_THROTTLE_NOPROGRESS
59 usec_timeout=500000 usect_delayed=208000 reason=VMSCAN_THROTTLE_NOPROGRESS
61 usec_timeout=500000 usect_delayed=68000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=192000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=480000 reason=VMSCAN_THROTTLE_NOPROGRESS
79 usec_timeout=500000 usect_delayed=60000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=320000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=92000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=64000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=80000 reason=VMSCAN_THROTTLE_NOPROGRESS
88 usec_timeout=500000 usect_delayed=84000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=160000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=292000 reason=VMSCAN_THROTTLE_NOPROGRESS
94 usec_timeout=500000 usect_delayed=56000 reason=VMSCAN_THROTTLE_NOPROGRESS
118 usec_timeout=500000 usect_delayed=88000 reason=VMSCAN_THROTTLE_NOPROGRESS
119 usec_timeout=500000 usect_delayed=72000 reason=VMSCAN_THROTTLE_NOPROGRESS
126 usec_timeout=500000 usect_delayed=108000 reason=VMSCAN_THROTTLE_NOPROGRESS
146 usec_timeout=500000 usect_delayed=52000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=36000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=48000 reason=VMSCAN_THROTTLE_NOPROGRESS
159 usec_timeout=500000 usect_delayed=28000 reason=VMSCAN_THROTTLE_NOPROGRESS
178 usec_timeout=500000 usect_delayed=44000 reason=VMSCAN_THROTTLE_NOPROGRESS
183 usec_timeout=500000 usect_delayed=40000 reason=VMSCAN_THROTTLE_NOPROGRESS
237 usec_timeout=500000 usect_delayed=100000 reason=VMSCAN_THROTTLE_NOPROGRESS
266 usec_timeout=500000 usect_delayed=32000 reason=VMSCAN_THROTTLE_NOPROGRESS
313 usec_timeout=500000 usect_delayed=24000 reason=VMSCAN_THROTTLE_NOPROGRESS
347 usec_timeout=500000 usect_delayed=96000 reason=VMSCAN_THROTTLE_NOPROGRESS
470 usec_timeout=500000 usect_delayed=20000 reason=VMSCAN_THROTTLE_NOPROGRESS
559 usec_timeout=500000 usect_delayed=16000 reason=VMSCAN_THROTTLE_NOPROGRESS
964 usec_timeout=500000 usect_delayed=12000 reason=VMSCAN_THROTTLE_NOPROGRESS
2001 usec_timeout=500000 usect_delayed=104000 reason=VMSCAN_THROTTLE_NOPROGRESS
2447 usec_timeout=500000 usect_delayed=8000 reason=VMSCAN_THROTTLE_NOPROGRESS
7888 usec_timeout=500000 usect_delayed=4000 reason=VMSCAN_THROTTLE_NOPROGRESS
22727 usec_timeout=500000 usect_delayed=0 reason=VMSCAN_THROTTLE_NOPROGRESS
51305 usec_timeout=500000 usect_delayed=500000 reason=VMSCAN_THROTTLE_NOPROGRESS
The full timeout is often hit but a large number also do not stall at
all. The remainder slept a little allowing other reclaim tasks to make
progress.
While this timeout could be further increased, it could also negatively
impact worst-case behaviour when there is no prioritisation of what task
should make progress.
For VMSCAN_THROTTLE_WRITEBACK, the breakdown was
1 usec_timeout=100000 usect_delayed=44000 reason=VMSCAN_THROTTLE_WRITEBACK
2 usec_timeout=100000 usect_delayed=76000 reason=VMSCAN_THROTTLE_WRITEBACK
3 usec_timeout=100000 usect_delayed=80000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=48000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=84000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
7 usec_timeout=100000 usect_delayed=88000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=56000 reason=VMSCAN_THROTTLE_WRITEBACK
12 usec_timeout=100000 usect_delayed=64000 reason=VMSCAN_THROTTLE_WRITEBACK
16 usec_timeout=100000 usect_delayed=92000 reason=VMSCAN_THROTTLE_WRITEBACK
24 usec_timeout=100000 usect_delayed=68000 reason=VMSCAN_THROTTLE_WRITEBACK
28 usec_timeout=100000 usect_delayed=32000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=60000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=96000 reason=VMSCAN_THROTTLE_WRITEBACK
32 usec_timeout=100000 usect_delayed=52000 reason=VMSCAN_THROTTLE_WRITEBACK
42 usec_timeout=100000 usect_delayed=40000 reason=VMSCAN_THROTTLE_WRITEBACK
77 usec_timeout=100000 usect_delayed=28000 reason=VMSCAN_THROTTLE_WRITEBACK
99 usec_timeout=100000 usect_delayed=36000 reason=VMSCAN_THROTTLE_WRITEBACK
137 usec_timeout=100000 usect_delayed=24000 reason=VMSCAN_THROTTLE_WRITEBACK
190 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
339 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
518 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
852 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
3359 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
7147 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
83962 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority hit the timeout in direct reclaim context although a
sizable number did not stall at all. This is very different to kswapd
where only a tiny percentage of stalls due to writeback reached the
timeout.
Bottom line, the throttling appears to work and the wakeup events may
limit worst case stalls. There might be some grounds for adjusting
timeouts but it's likely futile as the worst-case scenarios depend on
the workload, memory size and the speed of the storage. A better
approach to improve the series further would be to prioritise tasks
based on their rate of allocation with the caveat that it may be very
expensive to track.
This patch (of 5):
Page reclaim throttles on wait_iff_congested under the following
conditions:
- kswapd is encountering pages under writeback and marked for immediate
reclaim implying that pages are cycling through the LRU faster than
pages can be cleaned.
- Direct reclaim will stall if all dirty pages are backed by congested
inodes.
wait_iff_congested is almost completely broken with few exceptions.
This patch adds a new node-based workqueue and tracks the number of
throttled tasks and pages written back since throttling started. If
enough pages belonging to the node are written back then the throttled
tasks will wake early. If not, the throttled tasks sleeps until the
timeout expires.
[neilb@suse.de: Uninterruptible sleep and simpler wakeups]
[hdanton@sina.com: Avoid race when reclaim starts]
[vbabka@suse.cz: vmstat irq-safe api, clarifications]
Link: https://lore.kernel.org/linux-mm/45d8b7a6-8548-65f5-cccf-9f451d4ae3d4@kernel.dk/ [1]
Link: https://lkml.kernel.org/r/20211022144651.19914-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20211022144651.19914-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: NeilBrown <neilb@suse.de>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: "Darrick J . Wong" <djwong@kernel.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 20:42:25 +00:00
|
|
|
enum vmscan_throttle_state {
|
|
|
|
VMSCAN_THROTTLE_WRITEBACK,
|
2021-11-05 20:42:29 +00:00
|
|
|
VMSCAN_THROTTLE_ISOLATED,
|
2021-11-05 20:42:32 +00:00
|
|
|
VMSCAN_THROTTLE_NOPROGRESS,
|
mm: vmscan: Reduce throttling due to a failure to make progress
Mike Galbraith, Alexey Avramov and Darrick Wong all reported similar
problems due to reclaim throttling for excessive lengths of time. In
Alexey's case, a memory hog that should go OOM quickly stalls for
several minutes before stalling. In Mike and Darrick's cases, a small
memcg environment stalled excessively even though the system had enough
memory overall.
Commit 69392a403f49 ("mm/vmscan: throttle reclaim when no progress is
being made") introduced the problem although commit a19594ca4a8b
("mm/vmscan: increase the timeout if page reclaim is not making
progress") made it worse. Systems at or near an OOM state that cannot
be recovered must reach OOM quickly and memcg should kill tasks if a
memcg is near OOM.
To address this, only stall for the first zone in the zonelist, reduce
the timeout to 1 tick for VMSCAN_THROTTLE_NOPROGRESS and only stall if
the scan control nr_reclaimed is 0, kswapd is still active and there
were excessive pages pending for writeback. If kswapd has stopped
reclaiming due to excessive failures, do not stall at all so that OOM
triggers relatively quickly. Similarly, if an LRU is simply congested,
only lightly throttle similar to NOPROGRESS.
Alexey's original case was the most straight forward
for i in {1..3}; do tail /dev/zero; done
On vanilla 5.16-rc1, this test stalled heavily, after the patch the test
completes in a few seconds similar to 5.15.
Alexey's second test case added watching a youtube video while tail runs
10 times. On 5.15, playback only jitters slightly, 5.16-rc1 stalls a
lot with lots of frames missing and numerous audio glitches. With this
patch applies, the video plays similarly to 5.15.
[lkp@intel.com: Fix W=1 build warning]
Link: https://lore.kernel.org/r/99e779783d6c7fce96448a3402061b9dc1b3b602.camel@gmx.de
Link: https://lore.kernel.org/r/20211124011954.7cab9bb4@mail.inbox.lv
Link: https://lore.kernel.org/r/20211022144651.19914-1-mgorman@techsingularity.net
Link: https://lore.kernel.org/r/20211202150614.22440-1-mgorman@techsingularity.net
Link: https://linux-regtracking.leemhuis.info/regzbot/regression/20211124011954.7cab9bb4@mail.inbox.lv/
Reported-and-tested-by: Alexey Avramov <hakavlad@inbox.lv>
Reported-and-tested-by: Mike Galbraith <efault@gmx.de>
Reported-and-tested-by: Darrick J. Wong <djwong@kernel.org>
Reported-by: kernel test robot <lkp@intel.com>
Acked-by: Hugh Dickins <hughd@google.com>
Tracked-by: Thorsten Leemhuis <regressions@leemhuis.info>
Fixes: 69392a403f49 ("mm/vmscan: throttle reclaim when no progress is being made")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-12-02 15:06:14 +00:00
|
|
|
VMSCAN_THROTTLE_CONGESTED,
|
mm/vmscan: throttle reclaim until some writeback completes if congested
Patch series "Remove dependency on congestion_wait in mm/", v5.
This series that removes all calls to congestion_wait in mm/ and deletes
wait_iff_congested. It's not a clever implementation but
congestion_wait has been broken for a long time [1].
Even if congestion throttling worked, it was never a great idea. While
excessive dirty/writeback pages at the tail of the LRU is one
possibility that reclaim may be slow, there is also the problem of too
many pages being isolated and reclaim failing for other reasons
(elevated references, too many pages isolated, excessive LRU contention
etc).
This series replaces the "congestion" throttling with 3 different types.
- If there are too many dirty/writeback pages, sleep until a timeout or
enough pages get cleaned
- If too many pages are isolated, sleep until enough isolated pages are
either reclaimed or put back on the LRU
- If no progress is being made, direct reclaim tasks sleep until
another task makes progress with acceptable efficiency.
This was initially tested with a mix of workloads that used to trigger
corner cases that no longer work. A new test case was created called
"stutterp" (pagereclaim-stutterp-noreaders in mmtests) using a freshly
created XFS filesystem. Note that it may be necessary to increase the
timeout of ssh if executing remotely as ssh itself can get throttled and
the connection may timeout.
stutterp varies the number of "worker" processes from 4 up to NR_CPUS*4
to check the impact as the number of direct reclaimers increase. It has
four types of worker.
- One "anon latency" worker creates small mappings with mmap() and
times how long it takes to fault the mapping reading it 4K at a time
- X file writers which is fio randomly writing X files where the total
size of the files add up to the allowed dirty_ratio. fio is allowed
to run for a warmup period to allow some file-backed pages to
accumulate. The duration of the warmup is based on the best-case
linear write speed of the storage.
- Y file readers which is fio randomly reading small files
- Z anon memory hogs which continually map (100-dirty_ratio)% of memory
- Total estimated WSS = (100+dirty_ration) percentage of memory
X+Y+Z+1 == NR_WORKERS varying from 4 up to NR_CPUS*4
The intent is to maximise the total WSS with a mix of file and anon
memory where some anonymous memory must be swapped and there is a high
likelihood of dirty/writeback pages reaching the end of the LRU.
The test can be configured to have no background readers to stress
dirty/writeback pages. The results below are based on having zero
readers.
The short summary of the results is that the series works and stalls
until some event occurs but the timeouts may need adjustment.
The test results are not broken down by patch as the series should be
treated as one block that replaces a broken throttling mechanism with a
working one.
Finally, three machines were tested but I'm reporting the worst set of
results. The other two machines had much better latencies for example.
First the results of the "anon latency" latency
stutterp
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r4
Amean mmap-4 31.4003 ( 0.00%) 2661.0198 (-8374.52%)
Amean mmap-7 38.1641 ( 0.00%) 149.2891 (-291.18%)
Amean mmap-12 60.0981 ( 0.00%) 187.8105 (-212.51%)
Amean mmap-21 161.2699 ( 0.00%) 213.9107 ( -32.64%)
Amean mmap-30 174.5589 ( 0.00%) 377.7548 (-116.41%)
Amean mmap-48 8106.8160 ( 0.00%) 1070.5616 ( 86.79%)
Stddev mmap-4 41.3455 ( 0.00%) 27573.9676 (-66591.66%)
Stddev mmap-7 53.5556 ( 0.00%) 4608.5860 (-8505.23%)
Stddev mmap-12 171.3897 ( 0.00%) 5559.4542 (-3143.75%)
Stddev mmap-21 1506.6752 ( 0.00%) 5746.2507 (-281.39%)
Stddev mmap-30 557.5806 ( 0.00%) 7678.1624 (-1277.05%)
Stddev mmap-48 61681.5718 ( 0.00%) 14507.2830 ( 76.48%)
Max-90 mmap-4 31.4243 ( 0.00%) 83.1457 (-164.59%)
Max-90 mmap-7 41.0410 ( 0.00%) 41.0720 ( -0.08%)
Max-90 mmap-12 66.5255 ( 0.00%) 53.9073 ( 18.97%)
Max-90 mmap-21 146.7479 ( 0.00%) 105.9540 ( 27.80%)
Max-90 mmap-30 193.9513 ( 0.00%) 64.3067 ( 66.84%)
Max-90 mmap-48 277.9137 ( 0.00%) 591.0594 (-112.68%)
Max mmap-4 1913.8009 ( 0.00%) 299623.9695 (-15555.96%)
Max mmap-7 2423.9665 ( 0.00%) 204453.1708 (-8334.65%)
Max mmap-12 6845.6573 ( 0.00%) 221090.3366 (-3129.64%)
Max mmap-21 56278.6508 ( 0.00%) 213877.3496 (-280.03%)
Max mmap-30 19716.2990 ( 0.00%) 216287.6229 (-997.00%)
Max mmap-48 477923.9400 ( 0.00%) 245414.8238 ( 48.65%)
For most thread counts, the time to mmap() is unfortunately increased.
In earlier versions of the series, this was lower but a large number of
throttling events were reaching their timeout increasing the amount of
inefficient scanning of the LRU. There is no prioritisation of reclaim
tasks making progress based on each tasks rate of page allocation versus
progress of reclaim. The variance is also impacted for high worker
counts but in all cases, the differences in latency are not
statistically significant due to very large maximum outliers. Max-90
shows that 90% of the stalls are comparable but the Max results show the
massive outliers which are increased to to stalling.
It is expected that this will be very machine dependant. Due to the
test design, reclaim is difficult so allocations stall and there are
variances depending on whether THPs can be allocated or not. The amount
of memory will affect exactly how bad the corner cases are and how often
they trigger. The warmup period calculation is not ideal as it's based
on linear writes where as fio is randomly writing multiple files from
multiple tasks so the start state of the test is variable. For example,
these are the latencies on a single-socket machine that had more memory
Amean mmap-4 42.2287 ( 0.00%) 49.6838 * -17.65%*
Amean mmap-7 216.4326 ( 0.00%) 47.4451 * 78.08%*
Amean mmap-12 2412.0588 ( 0.00%) 51.7497 ( 97.85%)
Amean mmap-21 5546.2548 ( 0.00%) 51.8862 ( 99.06%)
Amean mmap-30 1085.3121 ( 0.00%) 72.1004 ( 93.36%)
The overall system CPU usage and elapsed time is as follows
5.15.0-rc3 5.15.0-rc3
vanilla mm-reclaimcongest-v5r4
Duration User 6989.03 983.42
Duration System 7308.12 799.68
Duration Elapsed 2277.67 2092.98
The patches reduce system CPU usage by 89% as the vanilla kernel is rarely
stalling.
The high-level /proc/vmstats show
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r2
Ops Direct pages scanned 1056608451.00 503594991.00
Ops Kswapd pages scanned 109795048.00 147289810.00
Ops Kswapd pages reclaimed 63269243.00 31036005.00
Ops Direct pages reclaimed 10803973.00 6328887.00
Ops Kswapd efficiency % 57.62 21.07
Ops Kswapd velocity 48204.98 57572.86
Ops Direct efficiency % 1.02 1.26
Ops Direct velocity 463898.83 196845.97
Kswapd scanned less pages but the detailed pattern is different. The
vanilla kernel scans slowly over time where as the patches exhibits
burst patterns of scan activity. Direct reclaim scanning is reduced by
52% due to stalling.
The pattern for stealing pages is also slightly different. Both kernels
exhibit spikes but the vanilla kernel when reclaiming shows pages being
reclaimed over a period of time where as the patches tend to reclaim in
spikes. The difference is that vanilla is not throttling and instead
scanning constantly finding some pages over time where as the patched
kernel throttles and reclaims in spikes.
Ops Percentage direct scans 90.59 77.37
For direct reclaim, vanilla scanned 90.59% of pages where as with the
patches, 77.37% were direct reclaim due to throttling
Ops Page writes by reclaim 2613590.00 1687131.00
Page writes from reclaim context are reduced.
Ops Page writes anon 2932752.00 1917048.00
And there is less swapping.
Ops Page reclaim immediate 996248528.00 107664764.00
The number of pages encountered at the tail of the LRU tagged for
immediate reclaim but still dirty/writeback is reduced by 89%.
Ops Slabs scanned 164284.00 153608.00
Slab scan activity is similar.
ftrace was used to gather stall activity
Vanilla
-------
1 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=16000
2 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=12000
8 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=8000
29 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=4000
82394 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=0
The fast majority of wait_iff_congested calls do not stall at all. What
is likely happening is that cond_resched() reschedules the task for a
short period when the BDI is not registering congestion (which it never
will in this test setup).
1 writeback_congestion_wait: usec_timeout=100000 usec_delayed=120000
2 writeback_congestion_wait: usec_timeout=100000 usec_delayed=132000
4 writeback_congestion_wait: usec_timeout=100000 usec_delayed=112000
380 writeback_congestion_wait: usec_timeout=100000 usec_delayed=108000
778 writeback_congestion_wait: usec_timeout=100000 usec_delayed=104000
congestion_wait if called always exceeds the timeout as there is no
trigger to wake it up.
Bottom line: Vanilla will throttle but it's not effective.
Patch series
------------
Kswapd throttle activity was always due to scanning pages tagged for
immediate reclaim at the tail of the LRU
1 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
4 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
94 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
112 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority of events did not stall or stalled for a short period.
Roughly 16% of stalls reached the timeout before expiry. For direct
reclaim, the number of times stalled for each reason were
6624 reason=VMSCAN_THROTTLE_ISOLATED
93246 reason=VMSCAN_THROTTLE_NOPROGRESS
96934 reason=VMSCAN_THROTTLE_WRITEBACK
The most common reason to stall was due to excessive pages tagged for
immediate reclaim at the tail of the LRU followed by a failure to make
forward. A relatively small number were due to too many pages isolated
from the LRU by parallel threads
For VMSCAN_THROTTLE_ISOLATED, the breakdown of delays was
9 usec_timeout=20000 usect_delayed=4000 reason=VMSCAN_THROTTLE_ISOLATED
12 usec_timeout=20000 usect_delayed=16000 reason=VMSCAN_THROTTLE_ISOLATED
83 usec_timeout=20000 usect_delayed=20000 reason=VMSCAN_THROTTLE_ISOLATED
6520 usec_timeout=20000 usect_delayed=0 reason=VMSCAN_THROTTLE_ISOLATED
Most did not stall at all. A small number reached the timeout.
For VMSCAN_THROTTLE_NOPROGRESS, the breakdown of stalls were all over
the map
1 usec_timeout=500000 usect_delayed=324000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=332000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=348000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=360000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=228000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=260000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=340000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=364000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=372000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=428000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=460000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=464000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=244000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=252000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=272000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=188000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=268000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=328000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=380000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=392000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=432000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=204000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=220000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=412000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=436000 reason=VMSCAN_THROTTLE_NOPROGRESS
6 usec_timeout=500000 usect_delayed=488000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=212000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=300000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=316000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=472000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=248000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=356000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=456000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=124000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=376000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=484000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=172000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=420000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=452000 reason=VMSCAN_THROTTLE_NOPROGRESS
11 usec_timeout=500000 usect_delayed=256000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=112000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=116000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=144000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=152000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=264000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=384000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=424000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=492000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=184000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=444000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=308000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=440000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=476000 reason=VMSCAN_THROTTLE_NOPROGRESS
16 usec_timeout=500000 usect_delayed=140000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=232000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=240000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=280000 reason=VMSCAN_THROTTLE_NOPROGRESS
18 usec_timeout=500000 usect_delayed=404000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=148000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=216000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=468000 reason=VMSCAN_THROTTLE_NOPROGRESS
21 usec_timeout=500000 usect_delayed=448000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=168000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=296000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=132000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=352000 reason=VMSCAN_THROTTLE_NOPROGRESS
26 usec_timeout=500000 usect_delayed=180000 reason=VMSCAN_THROTTLE_NOPROGRESS
27 usec_timeout=500000 usect_delayed=284000 reason=VMSCAN_THROTTLE_NOPROGRESS
28 usec_timeout=500000 usect_delayed=164000 reason=VMSCAN_THROTTLE_NOPROGRESS
29 usec_timeout=500000 usect_delayed=136000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=200000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=400000 reason=VMSCAN_THROTTLE_NOPROGRESS
31 usec_timeout=500000 usect_delayed=196000 reason=VMSCAN_THROTTLE_NOPROGRESS
32 usec_timeout=500000 usect_delayed=156000 reason=VMSCAN_THROTTLE_NOPROGRESS
33 usec_timeout=500000 usect_delayed=224000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=128000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=176000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=368000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=496000 reason=VMSCAN_THROTTLE_NOPROGRESS
37 usec_timeout=500000 usect_delayed=312000 reason=VMSCAN_THROTTLE_NOPROGRESS
38 usec_timeout=500000 usect_delayed=304000 reason=VMSCAN_THROTTLE_NOPROGRESS
40 usec_timeout=500000 usect_delayed=288000 reason=VMSCAN_THROTTLE_NOPROGRESS
43 usec_timeout=500000 usect_delayed=408000 reason=VMSCAN_THROTTLE_NOPROGRESS
55 usec_timeout=500000 usect_delayed=416000 reason=VMSCAN_THROTTLE_NOPROGRESS
56 usec_timeout=500000 usect_delayed=76000 reason=VMSCAN_THROTTLE_NOPROGRESS
58 usec_timeout=500000 usect_delayed=120000 reason=VMSCAN_THROTTLE_NOPROGRESS
59 usec_timeout=500000 usect_delayed=208000 reason=VMSCAN_THROTTLE_NOPROGRESS
61 usec_timeout=500000 usect_delayed=68000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=192000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=480000 reason=VMSCAN_THROTTLE_NOPROGRESS
79 usec_timeout=500000 usect_delayed=60000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=320000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=92000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=64000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=80000 reason=VMSCAN_THROTTLE_NOPROGRESS
88 usec_timeout=500000 usect_delayed=84000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=160000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=292000 reason=VMSCAN_THROTTLE_NOPROGRESS
94 usec_timeout=500000 usect_delayed=56000 reason=VMSCAN_THROTTLE_NOPROGRESS
118 usec_timeout=500000 usect_delayed=88000 reason=VMSCAN_THROTTLE_NOPROGRESS
119 usec_timeout=500000 usect_delayed=72000 reason=VMSCAN_THROTTLE_NOPROGRESS
126 usec_timeout=500000 usect_delayed=108000 reason=VMSCAN_THROTTLE_NOPROGRESS
146 usec_timeout=500000 usect_delayed=52000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=36000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=48000 reason=VMSCAN_THROTTLE_NOPROGRESS
159 usec_timeout=500000 usect_delayed=28000 reason=VMSCAN_THROTTLE_NOPROGRESS
178 usec_timeout=500000 usect_delayed=44000 reason=VMSCAN_THROTTLE_NOPROGRESS
183 usec_timeout=500000 usect_delayed=40000 reason=VMSCAN_THROTTLE_NOPROGRESS
237 usec_timeout=500000 usect_delayed=100000 reason=VMSCAN_THROTTLE_NOPROGRESS
266 usec_timeout=500000 usect_delayed=32000 reason=VMSCAN_THROTTLE_NOPROGRESS
313 usec_timeout=500000 usect_delayed=24000 reason=VMSCAN_THROTTLE_NOPROGRESS
347 usec_timeout=500000 usect_delayed=96000 reason=VMSCAN_THROTTLE_NOPROGRESS
470 usec_timeout=500000 usect_delayed=20000 reason=VMSCAN_THROTTLE_NOPROGRESS
559 usec_timeout=500000 usect_delayed=16000 reason=VMSCAN_THROTTLE_NOPROGRESS
964 usec_timeout=500000 usect_delayed=12000 reason=VMSCAN_THROTTLE_NOPROGRESS
2001 usec_timeout=500000 usect_delayed=104000 reason=VMSCAN_THROTTLE_NOPROGRESS
2447 usec_timeout=500000 usect_delayed=8000 reason=VMSCAN_THROTTLE_NOPROGRESS
7888 usec_timeout=500000 usect_delayed=4000 reason=VMSCAN_THROTTLE_NOPROGRESS
22727 usec_timeout=500000 usect_delayed=0 reason=VMSCAN_THROTTLE_NOPROGRESS
51305 usec_timeout=500000 usect_delayed=500000 reason=VMSCAN_THROTTLE_NOPROGRESS
The full timeout is often hit but a large number also do not stall at
all. The remainder slept a little allowing other reclaim tasks to make
progress.
While this timeout could be further increased, it could also negatively
impact worst-case behaviour when there is no prioritisation of what task
should make progress.
For VMSCAN_THROTTLE_WRITEBACK, the breakdown was
1 usec_timeout=100000 usect_delayed=44000 reason=VMSCAN_THROTTLE_WRITEBACK
2 usec_timeout=100000 usect_delayed=76000 reason=VMSCAN_THROTTLE_WRITEBACK
3 usec_timeout=100000 usect_delayed=80000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=48000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=84000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
7 usec_timeout=100000 usect_delayed=88000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=56000 reason=VMSCAN_THROTTLE_WRITEBACK
12 usec_timeout=100000 usect_delayed=64000 reason=VMSCAN_THROTTLE_WRITEBACK
16 usec_timeout=100000 usect_delayed=92000 reason=VMSCAN_THROTTLE_WRITEBACK
24 usec_timeout=100000 usect_delayed=68000 reason=VMSCAN_THROTTLE_WRITEBACK
28 usec_timeout=100000 usect_delayed=32000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=60000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=96000 reason=VMSCAN_THROTTLE_WRITEBACK
32 usec_timeout=100000 usect_delayed=52000 reason=VMSCAN_THROTTLE_WRITEBACK
42 usec_timeout=100000 usect_delayed=40000 reason=VMSCAN_THROTTLE_WRITEBACK
77 usec_timeout=100000 usect_delayed=28000 reason=VMSCAN_THROTTLE_WRITEBACK
99 usec_timeout=100000 usect_delayed=36000 reason=VMSCAN_THROTTLE_WRITEBACK
137 usec_timeout=100000 usect_delayed=24000 reason=VMSCAN_THROTTLE_WRITEBACK
190 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
339 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
518 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
852 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
3359 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
7147 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
83962 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority hit the timeout in direct reclaim context although a
sizable number did not stall at all. This is very different to kswapd
where only a tiny percentage of stalls due to writeback reached the
timeout.
Bottom line, the throttling appears to work and the wakeup events may
limit worst case stalls. There might be some grounds for adjusting
timeouts but it's likely futile as the worst-case scenarios depend on
the workload, memory size and the speed of the storage. A better
approach to improve the series further would be to prioritise tasks
based on their rate of allocation with the caveat that it may be very
expensive to track.
This patch (of 5):
Page reclaim throttles on wait_iff_congested under the following
conditions:
- kswapd is encountering pages under writeback and marked for immediate
reclaim implying that pages are cycling through the LRU faster than
pages can be cleaned.
- Direct reclaim will stall if all dirty pages are backed by congested
inodes.
wait_iff_congested is almost completely broken with few exceptions.
This patch adds a new node-based workqueue and tracks the number of
throttled tasks and pages written back since throttling started. If
enough pages belonging to the node are written back then the throttled
tasks will wake early. If not, the throttled tasks sleeps until the
timeout expires.
[neilb@suse.de: Uninterruptible sleep and simpler wakeups]
[hdanton@sina.com: Avoid race when reclaim starts]
[vbabka@suse.cz: vmstat irq-safe api, clarifications]
Link: https://lore.kernel.org/linux-mm/45d8b7a6-8548-65f5-cccf-9f451d4ae3d4@kernel.dk/ [1]
Link: https://lkml.kernel.org/r/20211022144651.19914-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20211022144651.19914-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: NeilBrown <neilb@suse.de>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: "Darrick J . Wong" <djwong@kernel.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 20:42:25 +00:00
|
|
|
NR_VMSCAN_THROTTLE,
|
|
|
|
};
|
|
|
|
|
2012-01-13 01:20:01 +00:00
|
|
|
#define for_each_lru(lru) for (lru = 0; lru < NR_LRU_LISTS; lru++)
|
2008-10-19 03:26:14 +00:00
|
|
|
|
2012-01-13 01:20:01 +00:00
|
|
|
#define for_each_evictable_lru(lru) for (lru = 0; lru <= LRU_ACTIVE_FILE; lru++)
|
Unevictable LRU Infrastructure
When the system contains lots of mlocked or otherwise unevictable pages,
the pageout code (kswapd) can spend lots of time scanning over these
pages. Worse still, the presence of lots of unevictable pages can confuse
kswapd into thinking that more aggressive pageout modes are required,
resulting in all kinds of bad behaviour.
Infrastructure to manage pages excluded from reclaim--i.e., hidden from
vmscan. Based on a patch by Larry Woodman of Red Hat. Reworked to
maintain "unevictable" pages on a separate per-zone LRU list, to "hide"
them from vmscan.
Kosaki Motohiro added the support for the memory controller unevictable
lru list.
Pages on the unevictable list have both PG_unevictable and PG_lru set.
Thus, PG_unevictable is analogous to and mutually exclusive with
PG_active--it specifies which LRU list the page is on.
The unevictable infrastructure is enabled by a new mm Kconfig option
[CONFIG_]UNEVICTABLE_LRU.
A new function 'page_evictable(page, vma)' in vmscan.c tests whether or
not a page may be evictable. Subsequent patches will add the various
!evictable tests. We'll want to keep these tests light-weight for use in
shrink_active_list() and, possibly, the fault path.
To avoid races between tasks putting pages [back] onto an LRU list and
tasks that might be moving the page from non-evictable to evictable state,
the new function 'putback_lru_page()' -- inverse to 'isolate_lru_page()'
-- tests the "evictability" of a page after placing it on the LRU, before
dropping the reference. If the page has become unevictable,
putback_lru_page() will redo the 'putback', thus moving the page to the
unevictable list. This way, we avoid "stranding" evictable pages on the
unevictable list.
[akpm@linux-foundation.org: fix fallout from out-of-order merge]
[riel@redhat.com: fix UNEVICTABLE_LRU and !PROC_PAGE_MONITOR build]
[nishimura@mxp.nes.nec.co.jp: remove redundant mapping check]
[kosaki.motohiro@jp.fujitsu.com: unevictable-lru-infrastructure: putback_lru_page()/unevictable page handling rework]
[kosaki.motohiro@jp.fujitsu.com: kill unnecessary lock_page() in vmscan.c]
[kosaki.motohiro@jp.fujitsu.com: revert migration change of unevictable lru infrastructure]
[kosaki.motohiro@jp.fujitsu.com: revert to unevictable-lru-infrastructure-kconfig-fix.patch]
[kosaki.motohiro@jp.fujitsu.com: restore patch failure of vmstat-unevictable-and-mlocked-pages-vm-events.patch]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Debugged-by: Benjamin Kidwell <benjkidwell@yahoo.com>
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-10-19 03:26:39 +00:00
|
|
|
|
mm: vmscan: enforce inactive:active ratio at the reclaim root
We split the LRU lists into inactive and an active parts to maximize
workingset protection while allowing just enough inactive cache space to
faciltate readahead and writeback for one-off file accesses (e.g. a
linear scan through a file, or logging); or just enough inactive anon to
maintain recent reference information when reclaim needs to swap.
With cgroups and their nested LRU lists, we currently don't do this
correctly. While recursive cgroup reclaim establishes a relative LRU
order among the pages of all involved cgroups, inactive:active size
decisions are done on a per-cgroup level. As a result, we'll reclaim a
cgroup's workingset when it doesn't have cold pages, even when one of its
siblings has plenty of it that should be reclaimed first.
For example: workload A has 50M worth of hot cache but doesn't do any
one-off file accesses; meanwhile, parallel workload B scans files and
rarely accesses the same page twice.
If these workloads were to run in an uncgrouped system, A would be
protected from the high rate of cache faults from B. But if they were put
in parallel cgroups for memory accounting purposes, B's fast cache fault
rate would push out the hot cache pages of A. This is unexpected and
undesirable - the "scan resistance" of the page cache is broken.
This patch moves inactive:active size balancing decisions to the root of
reclaim - the same level where the LRU order is established.
It does this by looking at the recursive size of the inactive and the
active file sets of the cgroup subtree at the beginning of the reclaim
cycle, and then making a decision - scan or skip active pages - that
applies throughout the entire run and to every cgroup involved.
With that in place, in the test above, the VM will recognize that there
are plenty of inactive pages in the combined cache set of workloads A and
B and prefer the one-off cache in B over the hot pages in A. The scan
resistance of the cache is restored.
[cai@lca.pw: fix some -Wenum-conversion warnings]
Link: http://lkml.kernel.org/r/1573848697-29262-1-git-send-email-cai@lca.pw
Link: http://lkml.kernel.org/r/20191107205334.158354-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-12-01 01:56:02 +00:00
|
|
|
static inline bool is_file_lru(enum lru_list lru)
|
2008-10-19 03:26:32 +00:00
|
|
|
{
|
2012-01-13 01:20:01 +00:00
|
|
|
return (lru == LRU_INACTIVE_FILE || lru == LRU_ACTIVE_FILE);
|
2008-10-19 03:26:32 +00:00
|
|
|
}
|
|
|
|
|
mm: vmscan: enforce inactive:active ratio at the reclaim root
We split the LRU lists into inactive and an active parts to maximize
workingset protection while allowing just enough inactive cache space to
faciltate readahead and writeback for one-off file accesses (e.g. a
linear scan through a file, or logging); or just enough inactive anon to
maintain recent reference information when reclaim needs to swap.
With cgroups and their nested LRU lists, we currently don't do this
correctly. While recursive cgroup reclaim establishes a relative LRU
order among the pages of all involved cgroups, inactive:active size
decisions are done on a per-cgroup level. As a result, we'll reclaim a
cgroup's workingset when it doesn't have cold pages, even when one of its
siblings has plenty of it that should be reclaimed first.
For example: workload A has 50M worth of hot cache but doesn't do any
one-off file accesses; meanwhile, parallel workload B scans files and
rarely accesses the same page twice.
If these workloads were to run in an uncgrouped system, A would be
protected from the high rate of cache faults from B. But if they were put
in parallel cgroups for memory accounting purposes, B's fast cache fault
rate would push out the hot cache pages of A. This is unexpected and
undesirable - the "scan resistance" of the page cache is broken.
This patch moves inactive:active size balancing decisions to the root of
reclaim - the same level where the LRU order is established.
It does this by looking at the recursive size of the inactive and the
active file sets of the cgroup subtree at the beginning of the reclaim
cycle, and then making a decision - scan or skip active pages - that
applies throughout the entire run and to every cgroup involved.
With that in place, in the test above, the VM will recognize that there
are plenty of inactive pages in the combined cache set of workloads A and
B and prefer the one-off cache in B over the hot pages in A. The scan
resistance of the cache is restored.
[cai@lca.pw: fix some -Wenum-conversion warnings]
Link: http://lkml.kernel.org/r/1573848697-29262-1-git-send-email-cai@lca.pw
Link: http://lkml.kernel.org/r/20191107205334.158354-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-12-01 01:56:02 +00:00
|
|
|
static inline bool is_active_lru(enum lru_list lru)
|
2008-10-19 03:26:14 +00:00
|
|
|
{
|
2012-01-13 01:20:01 +00:00
|
|
|
return (lru == LRU_ACTIVE_ANON || lru == LRU_ACTIVE_FILE);
|
2008-10-19 03:26:14 +00:00
|
|
|
}
|
|
|
|
|
2020-10-16 03:09:55 +00:00
|
|
|
#define ANON_AND_FILE 2
|
|
|
|
|
2019-12-01 01:55:52 +00:00
|
|
|
enum lruvec_flags {
|
|
|
|
LRUVEC_CONGESTED, /* lruvec has many dirty pages
|
|
|
|
* backed by a congested BDI
|
|
|
|
*/
|
|
|
|
};
|
|
|
|
|
2012-01-13 01:18:10 +00:00
|
|
|
struct lruvec {
|
2016-03-15 21:57:16 +00:00
|
|
|
struct list_head lists[NR_LRU_LISTS];
|
2020-12-15 20:34:29 +00:00
|
|
|
/* per lruvec lru_lock for memcg */
|
|
|
|
spinlock_t lru_lock;
|
mm: base LRU balancing on an explicit cost model
Currently, scan pressure between the anon and file LRU lists is balanced
based on a mixture of reclaim efficiency and a somewhat vague notion of
"value" of having certain pages in memory over others. That concept of
value is problematic, because it has caused us to count any event that
remotely makes one LRU list more or less preferrable for reclaim, even
when these events are not directly comparable and impose very different
costs on the system. One example is referenced file pages that we still
deactivate and referenced anonymous pages that we actually rotate back to
the head of the list.
There is also conceptual overlap with the LRU algorithm itself. By
rotating recently used pages instead of reclaiming them, the algorithm
already biases the applied scan pressure based on page value. Thus, when
rebalancing scan pressure due to rotations, we should think of reclaim
cost, and leave assessing the page value to the LRU algorithm.
Lastly, considering both value-increasing as well as value-decreasing
events can sometimes cause the same type of event to be counted twice,
i.e. how rotating a page increases the LRU value, while reclaiming it
succesfully decreases the value. In itself this will balance out fine,
but it quietly skews the impact of events that are only recorded once.
The abstract metric of "value", the murky relationship with the LRU
algorithm, and accounting both negative and positive events make the
current pressure balancing model hard to reason about and modify.
This patch switches to a balancing model of accounting the concrete,
actually observed cost of reclaiming one LRU over another. For now, that
cost includes pages that are scanned but rotated back to the list head.
Subsequent patches will add consideration for IO caused by refaulting of
recently evicted pages.
Replace struct zone_reclaim_stat with two cost counters in the lruvec, and
make everything that affects cost go through a new lru_note_cost()
function.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-9-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-06-03 23:02:53 +00:00
|
|
|
/*
|
|
|
|
* These track the cost of reclaiming one LRU - file or anon -
|
|
|
|
* over the other. As the observed cost of reclaiming one LRU
|
|
|
|
* increases, the reclaim scan balance tips toward the other.
|
|
|
|
*/
|
|
|
|
unsigned long anon_cost;
|
|
|
|
unsigned long file_cost;
|
2020-06-26 03:30:31 +00:00
|
|
|
/* Non-resident age, driven by LRU movement */
|
|
|
|
atomic_long_t nonresident_age;
|
2020-10-16 03:09:55 +00:00
|
|
|
/* Refaults at the time of last reclaim cycle */
|
|
|
|
unsigned long refaults[ANON_AND_FILE];
|
2019-12-01 01:55:52 +00:00
|
|
|
/* Various lruvec state flags (enum lruvec_flags) */
|
|
|
|
unsigned long flags;
|
2012-07-31 23:43:02 +00:00
|
|
|
#ifdef CONFIG_MEMCG
|
2016-07-28 22:45:31 +00:00
|
|
|
struct pglist_data *pgdat;
|
2012-05-29 22:06:58 +00:00
|
|
|
#endif
|
2012-01-13 01:18:10 +00:00
|
|
|
};
|
|
|
|
|
2019-12-01 01:55:18 +00:00
|
|
|
/* Isolate unmapped pages */
|
2012-05-29 22:06:54 +00:00
|
|
|
#define ISOLATE_UNMAPPED ((__force isolate_mode_t)0x2)
|
2012-01-13 01:19:38 +00:00
|
|
|
/* Isolate for asynchronous migration */
|
2012-05-29 22:06:54 +00:00
|
|
|
#define ISOLATE_ASYNC_MIGRATE ((__force isolate_mode_t)0x4)
|
2012-10-08 23:33:48 +00:00
|
|
|
/* Isolate unevictable pages */
|
|
|
|
#define ISOLATE_UNEVICTABLE ((__force isolate_mode_t)0x8)
|
2011-11-01 00:06:47 +00:00
|
|
|
|
|
|
|
/* LRU Isolation modes. */
|
2016-12-11 04:34:53 +00:00
|
|
|
typedef unsigned __bitwise isolate_mode_t;
|
2011-11-01 00:06:47 +00:00
|
|
|
|
2009-06-16 22:32:12 +00:00
|
|
|
enum zone_watermarks {
|
|
|
|
WMARK_MIN,
|
|
|
|
WMARK_LOW,
|
|
|
|
WMARK_HIGH,
|
|
|
|
NR_WMARK
|
|
|
|
};
|
|
|
|
|
2021-06-29 02:43:08 +00:00
|
|
|
/*
|
|
|
|
* One per migratetype for each PAGE_ALLOC_COSTLY_ORDER plus one additional
|
|
|
|
* for pageblock size for THP if configured.
|
|
|
|
*/
|
|
|
|
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
|
|
|
|
#define NR_PCP_THP 1
|
|
|
|
#else
|
|
|
|
#define NR_PCP_THP 0
|
|
|
|
#endif
|
|
|
|
#define NR_PCP_LISTS (MIGRATE_PCPTYPES * (PAGE_ALLOC_COSTLY_ORDER + 1 + NR_PCP_THP))
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Shift to encode migratetype and order in the same integer, with order
|
|
|
|
* in the least significant bits.
|
|
|
|
*/
|
|
|
|
#define NR_PCP_ORDER_WIDTH 8
|
|
|
|
#define NR_PCP_ORDER_MASK ((1<<NR_PCP_ORDER_WIDTH) - 1)
|
|
|
|
|
mm: reclaim small amounts of memory when an external fragmentation event occurs
An external fragmentation event was previously described as
When the page allocator fragments memory, it records the event using
the mm_page_alloc_extfrag event. If the fallback_order is smaller
than a pageblock order (order-9 on 64-bit x86) then it's considered
an event that will cause external fragmentation issues in the future.
The kernel reduces the probability of such events by increasing the
watermark sizes by calling set_recommended_min_free_kbytes early in the
lifetime of the system. This works reasonably well in general but if
there are enough sparsely populated pageblocks then the problem can still
occur as enough memory is free overall and kswapd stays asleep.
This patch introduces a watermark_boost_factor sysctl that allows a zone
watermark to be temporarily boosted when an external fragmentation causing
events occurs. The boosting will stall allocations that would decrease
free memory below the boosted low watermark and kswapd is woken if the
calling context allows to reclaim an amount of memory relative to the size
of the high watermark and the watermark_boost_factor until the boost is
cleared. When kswapd finishes, it wakes kcompactd at the pageblock order
to clean some of the pageblocks that may have been affected by the
fragmentation event. kswapd avoids any writeback, slab shrinkage and swap
from reclaim context during this operation to avoid excessive system
disruption in the name of fragmentation avoidance. Care is taken so that
kswapd will do normal reclaim work if the system is really low on memory.
This was evaluated using the same workloads as "mm, page_alloc: Spread
allocations across zones before introducing fragmentation".
1-socket Skylake machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 1 THP allocating thread
--------------------------------------
4.20-rc3 extfrag events < order 9: 804694
4.20-rc3+patch: 408912 (49% reduction)
4.20-rc3+patch1-4: 18421 (98% reduction)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Amean fault-base-1 653.58 ( 0.00%) 652.71 ( 0.13%)
Amean fault-huge-1 0.00 ( 0.00%) 178.93 * -99.00%*
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-1 0.00 ( 0.00%) 5.12 ( 100.00%)
Note that external fragmentation causing events are massively reduced by
this path whether in comparison to the previous kernel or the vanilla
kernel. The fault latency for huge pages appears to be increased but that
is only because THP allocations were successful with the patch applied.
1-socket Skylake machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------
4.20-rc3 extfrag events < order 9: 291392
4.20-rc3+patch: 191187 (34% reduction)
4.20-rc3+patch1-4: 13464 (95% reduction)
thpfioscale Fault Latencies
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Min fault-base-1 912.00 ( 0.00%) 905.00 ( 0.77%)
Min fault-huge-1 127.00 ( 0.00%) 135.00 ( -6.30%)
Amean fault-base-1 1467.55 ( 0.00%) 1481.67 ( -0.96%)
Amean fault-huge-1 1127.11 ( 0.00%) 1063.88 * 5.61%*
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-1 77.64 ( 0.00%) 83.46 ( 7.49%)
As before, massive reduction in external fragmentation events, some jitter
on latencies and an increase in THP allocation success rates.
2-socket Haswell machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 5 THP allocating threads
----------------------------------------------------------------
4.20-rc3 extfrag events < order 9: 215698
4.20-rc3+patch: 200210 (7% reduction)
4.20-rc3+patch1-4: 14263 (93% reduction)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Amean fault-base-5 1346.45 ( 0.00%) 1306.87 ( 2.94%)
Amean fault-huge-5 3418.60 ( 0.00%) 1348.94 ( 60.54%)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-5 0.78 ( 0.00%) 7.91 ( 910.64%)
There is a 93% reduction in fragmentation causing events, there is a big
reduction in the huge page fault latency and allocation success rate is
higher.
2-socket Haswell machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------
4.20-rc3 extfrag events < order 9: 166352
4.20-rc3+patch: 147463 (11% reduction)
4.20-rc3+patch1-4: 11095 (93% reduction)
thpfioscale Fault Latencies
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Amean fault-base-5 6217.43 ( 0.00%) 7419.67 * -19.34%*
Amean fault-huge-5 3163.33 ( 0.00%) 3263.80 ( -3.18%)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-5 95.14 ( 0.00%) 87.98 ( -7.53%)
There is a large reduction in fragmentation events with some jitter around
the latencies and success rates. As before, the high THP allocation
success rate does mean the system is under a lot of pressure. However, as
the fragmentation events are reduced, it would be expected that the
long-term allocation success rate would be higher.
Link: http://lkml.kernel.org/r/20181123114528.28802-5-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: David Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 08:35:52 +00:00
|
|
|
#define min_wmark_pages(z) (z->_watermark[WMARK_MIN] + z->watermark_boost)
|
|
|
|
#define low_wmark_pages(z) (z->_watermark[WMARK_LOW] + z->watermark_boost)
|
|
|
|
#define high_wmark_pages(z) (z->_watermark[WMARK_HIGH] + z->watermark_boost)
|
|
|
|
#define wmark_pages(z, i) (z->_watermark[i] + z->watermark_boost)
|
2009-06-16 22:32:12 +00:00
|
|
|
|
2021-06-29 02:41:41 +00:00
|
|
|
/* Fields and list protected by pagesets local_lock in page_alloc.c */
|
2005-04-16 22:20:36 +00:00
|
|
|
struct per_cpu_pages {
|
|
|
|
int count; /* number of pages in the list */
|
|
|
|
int high; /* high watermark, emptying needed */
|
|
|
|
int batch; /* chunk size for buddy add/remove */
|
2021-06-29 02:42:18 +00:00
|
|
|
short free_factor; /* batch scaling factor during free */
|
mm/page_alloc: split per cpu page lists and zone stats
The PCP (per-cpu page allocator in page_alloc.c) shares locking
requirements with vmstat and the zone lock which is inconvenient and
causes some issues. For example, the PCP list and vmstat share the same
per-cpu space meaning that it's possible that vmstat updates dirty cache
lines holding per-cpu lists across CPUs unless padding is used. Second,
PREEMPT_RT does not want to disable IRQs for too long in the page
allocator.
This series splits the locking requirements and uses locks types more
suitable for PREEMPT_RT, reduces the time when special locking is required
for stats and reduces the time when IRQs need to be disabled on
!PREEMPT_RT kernels.
Why local_lock? PREEMPT_RT considers the following sequence to be unsafe
as documented in Documentation/locking/locktypes.rst
local_irq_disable();
spin_lock(&lock);
The pcp allocator has this sequence for rmqueue_pcplist (local_irq_save)
-> __rmqueue_pcplist -> rmqueue_bulk (spin_lock). While it's possible to
separate this out, it generally means there are points where we enable
IRQs and reenable them again immediately. To prevent a migration and the
per-cpu pointer going stale, migrate_disable is also needed. That is a
custom lock that is similar, but worse, than local_lock. Furthermore, on
PREEMPT_RT, it's undesirable to leave IRQs disabled for too long. By
converting to local_lock which disables migration on PREEMPT_RT, the
locking requirements can be separated and start moving the protections for
PCP, stats and the zone lock to PREEMPT_RT-safe equivalent locking. As a
bonus, local_lock also means that PROVE_LOCKING does something useful.
After that, it's obvious that zone_statistics incurs too much overhead and
leaves IRQs disabled for longer than necessary on !PREEMPT_RT kernels.
zone_statistics uses perfectly accurate counters requiring IRQs be
disabled for parallel RMW sequences when inaccurate ones like vm_events
would do. The series makes the NUMA statistics (NUMA_HIT and friends)
inaccurate counters that then require no special protection on
!PREEMPT_RT.
The bulk page allocator can then do stat updates in bulk with IRQs enabled
which should improve the efficiency. Technically, this could have been
done without the local_lock and vmstat conversion work and the order
simply reflects the timing of when different series were implemented.
Finally, there are places where we conflate IRQs being disabled for the
PCP with the IRQ-safe zone spinlock. The remainder of the series reduces
the scope of what is protected by disabled IRQs on !PREEMPT_RT kernels.
By the end of the series, page_alloc.c does not call local_irq_save so the
locking scope is a bit clearer. The one exception is that modifying
NR_FREE_PAGES still happens in places where it's known the IRQs are
disabled as it's harmless for PREEMPT_RT and would be expensive to split
the locking there.
No performance data is included because despite the overhead of the stats,
it's within the noise for most workloads on !PREEMPT_RT. However, Jesper
Dangaard Brouer ran a page allocation microbenchmark on a E5-1650 v4 @
3.60GHz CPU on the first version of this series. Focusing on the array
variant of the bulk page allocator reveals the following.
(CPU: Intel(R) Xeon(R) CPU E5-1650 v4 @ 3.60GHz)
ARRAY variant: time_bulk_page_alloc_free_array: step=bulk size
Baseline Patched
1 56.383 54.225 (+3.83%)
2 40.047 35.492 (+11.38%)
3 37.339 32.643 (+12.58%)
4 35.578 30.992 (+12.89%)
8 33.592 29.606 (+11.87%)
16 32.362 28.532 (+11.85%)
32 31.476 27.728 (+11.91%)
64 30.633 27.252 (+11.04%)
128 30.596 27.090 (+11.46%)
While this is a positive outcome, the series is more likely to be
interesting to the RT people in terms of getting parts of the PREEMPT_RT
tree into mainline.
This patch (of 9):
The per-cpu page allocator lists and the per-cpu vmstat deltas are stored
in the same struct per_cpu_pages even though vmstats have no direct impact
on the per-cpu page lists. This is inconsistent because the vmstats for a
node are stored on a dedicated structure. The bigger issue is that the
per_cpu_pages structure is not cache-aligned and stat updates either cache
conflict with adjacent per-cpu lists incurring a runtime cost or padding
is required incurring a memory cost.
This patch splits the per-cpu pagelists and the vmstat deltas into
separate structures. It's mostly a mechanical conversion but some
variable renaming is done to clearly distinguish the per-cpu pages
structure (pcp) from the vmstats (pzstats).
Superficially, this appears to increase the size of the per_cpu_pages
structure but the movement of expire fills a structure hole so there is no
impact overall.
[mgorman@techsingularity.net: make it W=1 cleaner]
Link: https://lkml.kernel.org/r/20210514144622.GA3735@techsingularity.net
[mgorman@techsingularity.net: make it W=1 even cleaner]
Link: https://lkml.kernel.org/r/20210516140705.GB3735@techsingularity.net
[lkp@intel.com: check struct per_cpu_zonestat has a non-zero size]
[vbabka@suse.cz: Init zone->per_cpu_zonestats properly]
Link: https://lkml.kernel.org/r/20210512095458.30632-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20210512095458.30632-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-06-29 02:41:38 +00:00
|
|
|
#ifdef CONFIG_NUMA
|
2021-06-29 02:42:18 +00:00
|
|
|
short expire; /* When 0, remote pagesets are drained */
|
mm/page_alloc: split per cpu page lists and zone stats
The PCP (per-cpu page allocator in page_alloc.c) shares locking
requirements with vmstat and the zone lock which is inconvenient and
causes some issues. For example, the PCP list and vmstat share the same
per-cpu space meaning that it's possible that vmstat updates dirty cache
lines holding per-cpu lists across CPUs unless padding is used. Second,
PREEMPT_RT does not want to disable IRQs for too long in the page
allocator.
This series splits the locking requirements and uses locks types more
suitable for PREEMPT_RT, reduces the time when special locking is required
for stats and reduces the time when IRQs need to be disabled on
!PREEMPT_RT kernels.
Why local_lock? PREEMPT_RT considers the following sequence to be unsafe
as documented in Documentation/locking/locktypes.rst
local_irq_disable();
spin_lock(&lock);
The pcp allocator has this sequence for rmqueue_pcplist (local_irq_save)
-> __rmqueue_pcplist -> rmqueue_bulk (spin_lock). While it's possible to
separate this out, it generally means there are points where we enable
IRQs and reenable them again immediately. To prevent a migration and the
per-cpu pointer going stale, migrate_disable is also needed. That is a
custom lock that is similar, but worse, than local_lock. Furthermore, on
PREEMPT_RT, it's undesirable to leave IRQs disabled for too long. By
converting to local_lock which disables migration on PREEMPT_RT, the
locking requirements can be separated and start moving the protections for
PCP, stats and the zone lock to PREEMPT_RT-safe equivalent locking. As a
bonus, local_lock also means that PROVE_LOCKING does something useful.
After that, it's obvious that zone_statistics incurs too much overhead and
leaves IRQs disabled for longer than necessary on !PREEMPT_RT kernels.
zone_statistics uses perfectly accurate counters requiring IRQs be
disabled for parallel RMW sequences when inaccurate ones like vm_events
would do. The series makes the NUMA statistics (NUMA_HIT and friends)
inaccurate counters that then require no special protection on
!PREEMPT_RT.
The bulk page allocator can then do stat updates in bulk with IRQs enabled
which should improve the efficiency. Technically, this could have been
done without the local_lock and vmstat conversion work and the order
simply reflects the timing of when different series were implemented.
Finally, there are places where we conflate IRQs being disabled for the
PCP with the IRQ-safe zone spinlock. The remainder of the series reduces
the scope of what is protected by disabled IRQs on !PREEMPT_RT kernels.
By the end of the series, page_alloc.c does not call local_irq_save so the
locking scope is a bit clearer. The one exception is that modifying
NR_FREE_PAGES still happens in places where it's known the IRQs are
disabled as it's harmless for PREEMPT_RT and would be expensive to split
the locking there.
No performance data is included because despite the overhead of the stats,
it's within the noise for most workloads on !PREEMPT_RT. However, Jesper
Dangaard Brouer ran a page allocation microbenchmark on a E5-1650 v4 @
3.60GHz CPU on the first version of this series. Focusing on the array
variant of the bulk page allocator reveals the following.
(CPU: Intel(R) Xeon(R) CPU E5-1650 v4 @ 3.60GHz)
ARRAY variant: time_bulk_page_alloc_free_array: step=bulk size
Baseline Patched
1 56.383 54.225 (+3.83%)
2 40.047 35.492 (+11.38%)
3 37.339 32.643 (+12.58%)
4 35.578 30.992 (+12.89%)
8 33.592 29.606 (+11.87%)
16 32.362 28.532 (+11.85%)
32 31.476 27.728 (+11.91%)
64 30.633 27.252 (+11.04%)
128 30.596 27.090 (+11.46%)
While this is a positive outcome, the series is more likely to be
interesting to the RT people in terms of getting parts of the PREEMPT_RT
tree into mainline.
This patch (of 9):
The per-cpu page allocator lists and the per-cpu vmstat deltas are stored
in the same struct per_cpu_pages even though vmstats have no direct impact
on the per-cpu page lists. This is inconsistent because the vmstats for a
node are stored on a dedicated structure. The bigger issue is that the
per_cpu_pages structure is not cache-aligned and stat updates either cache
conflict with adjacent per-cpu lists incurring a runtime cost or padding
is required incurring a memory cost.
This patch splits the per-cpu pagelists and the vmstat deltas into
separate structures. It's mostly a mechanical conversion but some
variable renaming is done to clearly distinguish the per-cpu pages
structure (pcp) from the vmstats (pzstats).
Superficially, this appears to increase the size of the per_cpu_pages
structure but the movement of expire fills a structure hole so there is no
impact overall.
[mgorman@techsingularity.net: make it W=1 cleaner]
Link: https://lkml.kernel.org/r/20210514144622.GA3735@techsingularity.net
[mgorman@techsingularity.net: make it W=1 even cleaner]
Link: https://lkml.kernel.org/r/20210516140705.GB3735@techsingularity.net
[lkp@intel.com: check struct per_cpu_zonestat has a non-zero size]
[vbabka@suse.cz: Init zone->per_cpu_zonestats properly]
Link: https://lkml.kernel.org/r/20210512095458.30632-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20210512095458.30632-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-06-29 02:41:38 +00:00
|
|
|
#endif
|
page-allocator: split per-cpu list into one-list-per-migrate-type
The following two patches remove searching in the page allocator fast-path
by maintaining multiple free-lists in the per-cpu structure. At the time
the search was introduced, increasing the per-cpu structures would waste a
lot of memory as per-cpu structures were statically allocated at
compile-time. This is no longer the case.
The patches are as follows. They are based on mmotm-2009-08-27.
Patch 1 adds multiple lists to struct per_cpu_pages, one per
migratetype that can be stored on the PCP lists.
Patch 2 notes that the pcpu drain path check empty lists multiple times. The
patch reduces the number of checks by maintaining a count of free
lists encountered. Lists containing pages will then free multiple
pages in batch
The patches were tested with kernbench, netperf udp/tcp, hackbench and
sysbench. The netperf tests were not bound to any CPU in particular and
were run such that the results should be 99% confidence that the reported
results are within 1% of the estimated mean. sysbench was run with a
postgres background and read-only tests. Similar to netperf, it was run
multiple times so that it's 99% confidence results are within 1%. The
patches were tested on x86, x86-64 and ppc64 as
x86: Intel Pentium D 3GHz with 8G RAM (no-brand machine)
kernbench - No significant difference, variance well within noise
netperf-udp - 1.34% to 2.28% gain
netperf-tcp - 0.45% to 1.22% gain
hackbench - Small variances, very close to noise
sysbench - Very small gains
x86-64: AMD Phenom 9950 1.3GHz with 8G RAM (no-brand machine)
kernbench - No significant difference, variance well within noise
netperf-udp - 1.83% to 10.42% gains
netperf-tcp - No conclusive until buffer >= PAGE_SIZE
4096 +15.83%
8192 + 0.34% (not significant)
16384 + 1%
hackbench - Small gains, very close to noise
sysbench - 0.79% to 1.6% gain
ppc64: PPC970MP 2.5GHz with 10GB RAM (it's a terrasoft powerstation)
kernbench - No significant difference, variance well within noise
netperf-udp - 2-3% gain for almost all buffer sizes tested
netperf-tcp - losses on small buffers, gains on larger buffers
possibly indicates some bad caching effect.
hackbench - No significant difference
sysbench - 2-4% gain
This patch:
Currently the per-cpu page allocator searches the PCP list for pages of
the correct migrate-type to reduce the possibility of pages being
inappropriate placed from a fragmentation perspective. This search is
potentially expensive in a fast-path and undesirable. Splitting the
per-cpu list into multiple lists increases the size of a per-cpu structure
and this was potentially a major problem at the time the search was
introduced. These problem has been mitigated as now only the necessary
number of structures is allocated for the running system.
This patch replaces a list search in the per-cpu allocator with one list
per migrate type. The potential snag with this approach is when bulk
freeing pages. We round-robin free pages based on migrate type which has
little bearing on the cache hotness of the page and potentially checks
empty lists repeatedly in the event the majority of PCP pages are of one
type.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: Nick Piggin <npiggin@suse.de>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-09-22 00:03:19 +00:00
|
|
|
|
|
|
|
/* Lists of pages, one per migrate type stored on the pcp-lists */
|
2021-06-29 02:43:08 +00:00
|
|
|
struct list_head lists[NR_PCP_LISTS];
|
2005-04-16 22:20:36 +00:00
|
|
|
};
|
|
|
|
|
mm/page_alloc: split per cpu page lists and zone stats
The PCP (per-cpu page allocator in page_alloc.c) shares locking
requirements with vmstat and the zone lock which is inconvenient and
causes some issues. For example, the PCP list and vmstat share the same
per-cpu space meaning that it's possible that vmstat updates dirty cache
lines holding per-cpu lists across CPUs unless padding is used. Second,
PREEMPT_RT does not want to disable IRQs for too long in the page
allocator.
This series splits the locking requirements and uses locks types more
suitable for PREEMPT_RT, reduces the time when special locking is required
for stats and reduces the time when IRQs need to be disabled on
!PREEMPT_RT kernels.
Why local_lock? PREEMPT_RT considers the following sequence to be unsafe
as documented in Documentation/locking/locktypes.rst
local_irq_disable();
spin_lock(&lock);
The pcp allocator has this sequence for rmqueue_pcplist (local_irq_save)
-> __rmqueue_pcplist -> rmqueue_bulk (spin_lock). While it's possible to
separate this out, it generally means there are points where we enable
IRQs and reenable them again immediately. To prevent a migration and the
per-cpu pointer going stale, migrate_disable is also needed. That is a
custom lock that is similar, but worse, than local_lock. Furthermore, on
PREEMPT_RT, it's undesirable to leave IRQs disabled for too long. By
converting to local_lock which disables migration on PREEMPT_RT, the
locking requirements can be separated and start moving the protections for
PCP, stats and the zone lock to PREEMPT_RT-safe equivalent locking. As a
bonus, local_lock also means that PROVE_LOCKING does something useful.
After that, it's obvious that zone_statistics incurs too much overhead and
leaves IRQs disabled for longer than necessary on !PREEMPT_RT kernels.
zone_statistics uses perfectly accurate counters requiring IRQs be
disabled for parallel RMW sequences when inaccurate ones like vm_events
would do. The series makes the NUMA statistics (NUMA_HIT and friends)
inaccurate counters that then require no special protection on
!PREEMPT_RT.
The bulk page allocator can then do stat updates in bulk with IRQs enabled
which should improve the efficiency. Technically, this could have been
done without the local_lock and vmstat conversion work and the order
simply reflects the timing of when different series were implemented.
Finally, there are places where we conflate IRQs being disabled for the
PCP with the IRQ-safe zone spinlock. The remainder of the series reduces
the scope of what is protected by disabled IRQs on !PREEMPT_RT kernels.
By the end of the series, page_alloc.c does not call local_irq_save so the
locking scope is a bit clearer. The one exception is that modifying
NR_FREE_PAGES still happens in places where it's known the IRQs are
disabled as it's harmless for PREEMPT_RT and would be expensive to split
the locking there.
No performance data is included because despite the overhead of the stats,
it's within the noise for most workloads on !PREEMPT_RT. However, Jesper
Dangaard Brouer ran a page allocation microbenchmark on a E5-1650 v4 @
3.60GHz CPU on the first version of this series. Focusing on the array
variant of the bulk page allocator reveals the following.
(CPU: Intel(R) Xeon(R) CPU E5-1650 v4 @ 3.60GHz)
ARRAY variant: time_bulk_page_alloc_free_array: step=bulk size
Baseline Patched
1 56.383 54.225 (+3.83%)
2 40.047 35.492 (+11.38%)
3 37.339 32.643 (+12.58%)
4 35.578 30.992 (+12.89%)
8 33.592 29.606 (+11.87%)
16 32.362 28.532 (+11.85%)
32 31.476 27.728 (+11.91%)
64 30.633 27.252 (+11.04%)
128 30.596 27.090 (+11.46%)
While this is a positive outcome, the series is more likely to be
interesting to the RT people in terms of getting parts of the PREEMPT_RT
tree into mainline.
This patch (of 9):
The per-cpu page allocator lists and the per-cpu vmstat deltas are stored
in the same struct per_cpu_pages even though vmstats have no direct impact
on the per-cpu page lists. This is inconsistent because the vmstats for a
node are stored on a dedicated structure. The bigger issue is that the
per_cpu_pages structure is not cache-aligned and stat updates either cache
conflict with adjacent per-cpu lists incurring a runtime cost or padding
is required incurring a memory cost.
This patch splits the per-cpu pagelists and the vmstat deltas into
separate structures. It's mostly a mechanical conversion but some
variable renaming is done to clearly distinguish the per-cpu pages
structure (pcp) from the vmstats (pzstats).
Superficially, this appears to increase the size of the per_cpu_pages
structure but the movement of expire fills a structure hole so there is no
impact overall.
[mgorman@techsingularity.net: make it W=1 cleaner]
Link: https://lkml.kernel.org/r/20210514144622.GA3735@techsingularity.net
[mgorman@techsingularity.net: make it W=1 even cleaner]
Link: https://lkml.kernel.org/r/20210516140705.GB3735@techsingularity.net
[lkp@intel.com: check struct per_cpu_zonestat has a non-zero size]
[vbabka@suse.cz: Init zone->per_cpu_zonestats properly]
Link: https://lkml.kernel.org/r/20210512095458.30632-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20210512095458.30632-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-06-29 02:41:38 +00:00
|
|
|
struct per_cpu_zonestat {
|
2006-06-30 08:55:33 +00:00
|
|
|
#ifdef CONFIG_SMP
|
|
|
|
s8 vm_stat_diff[NR_VM_ZONE_STAT_ITEMS];
|
mm/page_alloc: split per cpu page lists and zone stats
The PCP (per-cpu page allocator in page_alloc.c) shares locking
requirements with vmstat and the zone lock which is inconvenient and
causes some issues. For example, the PCP list and vmstat share the same
per-cpu space meaning that it's possible that vmstat updates dirty cache
lines holding per-cpu lists across CPUs unless padding is used. Second,
PREEMPT_RT does not want to disable IRQs for too long in the page
allocator.
This series splits the locking requirements and uses locks types more
suitable for PREEMPT_RT, reduces the time when special locking is required
for stats and reduces the time when IRQs need to be disabled on
!PREEMPT_RT kernels.
Why local_lock? PREEMPT_RT considers the following sequence to be unsafe
as documented in Documentation/locking/locktypes.rst
local_irq_disable();
spin_lock(&lock);
The pcp allocator has this sequence for rmqueue_pcplist (local_irq_save)
-> __rmqueue_pcplist -> rmqueue_bulk (spin_lock). While it's possible to
separate this out, it generally means there are points where we enable
IRQs and reenable them again immediately. To prevent a migration and the
per-cpu pointer going stale, migrate_disable is also needed. That is a
custom lock that is similar, but worse, than local_lock. Furthermore, on
PREEMPT_RT, it's undesirable to leave IRQs disabled for too long. By
converting to local_lock which disables migration on PREEMPT_RT, the
locking requirements can be separated and start moving the protections for
PCP, stats and the zone lock to PREEMPT_RT-safe equivalent locking. As a
bonus, local_lock also means that PROVE_LOCKING does something useful.
After that, it's obvious that zone_statistics incurs too much overhead and
leaves IRQs disabled for longer than necessary on !PREEMPT_RT kernels.
zone_statistics uses perfectly accurate counters requiring IRQs be
disabled for parallel RMW sequences when inaccurate ones like vm_events
would do. The series makes the NUMA statistics (NUMA_HIT and friends)
inaccurate counters that then require no special protection on
!PREEMPT_RT.
The bulk page allocator can then do stat updates in bulk with IRQs enabled
which should improve the efficiency. Technically, this could have been
done without the local_lock and vmstat conversion work and the order
simply reflects the timing of when different series were implemented.
Finally, there are places where we conflate IRQs being disabled for the
PCP with the IRQ-safe zone spinlock. The remainder of the series reduces
the scope of what is protected by disabled IRQs on !PREEMPT_RT kernels.
By the end of the series, page_alloc.c does not call local_irq_save so the
locking scope is a bit clearer. The one exception is that modifying
NR_FREE_PAGES still happens in places where it's known the IRQs are
disabled as it's harmless for PREEMPT_RT and would be expensive to split
the locking there.
No performance data is included because despite the overhead of the stats,
it's within the noise for most workloads on !PREEMPT_RT. However, Jesper
Dangaard Brouer ran a page allocation microbenchmark on a E5-1650 v4 @
3.60GHz CPU on the first version of this series. Focusing on the array
variant of the bulk page allocator reveals the following.
(CPU: Intel(R) Xeon(R) CPU E5-1650 v4 @ 3.60GHz)
ARRAY variant: time_bulk_page_alloc_free_array: step=bulk size
Baseline Patched
1 56.383 54.225 (+3.83%)
2 40.047 35.492 (+11.38%)
3 37.339 32.643 (+12.58%)
4 35.578 30.992 (+12.89%)
8 33.592 29.606 (+11.87%)
16 32.362 28.532 (+11.85%)
32 31.476 27.728 (+11.91%)
64 30.633 27.252 (+11.04%)
128 30.596 27.090 (+11.46%)
While this is a positive outcome, the series is more likely to be
interesting to the RT people in terms of getting parts of the PREEMPT_RT
tree into mainline.
This patch (of 9):
The per-cpu page allocator lists and the per-cpu vmstat deltas are stored
in the same struct per_cpu_pages even though vmstats have no direct impact
on the per-cpu page lists. This is inconsistent because the vmstats for a
node are stored on a dedicated structure. The bigger issue is that the
per_cpu_pages structure is not cache-aligned and stat updates either cache
conflict with adjacent per-cpu lists incurring a runtime cost or padding
is required incurring a memory cost.
This patch splits the per-cpu pagelists and the vmstat deltas into
separate structures. It's mostly a mechanical conversion but some
variable renaming is done to clearly distinguish the per-cpu pages
structure (pcp) from the vmstats (pzstats).
Superficially, this appears to increase the size of the per_cpu_pages
structure but the movement of expire fills a structure hole so there is no
impact overall.
[mgorman@techsingularity.net: make it W=1 cleaner]
Link: https://lkml.kernel.org/r/20210514144622.GA3735@techsingularity.net
[mgorman@techsingularity.net: make it W=1 even cleaner]
Link: https://lkml.kernel.org/r/20210516140705.GB3735@techsingularity.net
[lkp@intel.com: check struct per_cpu_zonestat has a non-zero size]
[vbabka@suse.cz: Init zone->per_cpu_zonestats properly]
Link: https://lkml.kernel.org/r/20210512095458.30632-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20210512095458.30632-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-06-29 02:41:38 +00:00
|
|
|
s8 stat_threshold;
|
|
|
|
#endif
|
|
|
|
#ifdef CONFIG_NUMA
|
2021-06-29 02:41:44 +00:00
|
|
|
/*
|
|
|
|
* Low priority inaccurate counters that are only folded
|
|
|
|
* on demand. Use a large type to avoid the overhead of
|
|
|
|
* folding during refresh_cpu_vm_stats.
|
|
|
|
*/
|
|
|
|
unsigned long vm_numa_event[NR_VM_NUMA_EVENT_ITEMS];
|
2006-06-30 08:55:33 +00:00
|
|
|
#endif
|
2010-01-05 06:34:51 +00:00
|
|
|
};
|
2005-06-22 00:14:47 +00:00
|
|
|
|
mm, vmstat: add infrastructure for per-node vmstats
Patchset: "Move LRU page reclaim from zones to nodes v9"
This series moves LRUs from the zones to the node. While this is a
current rebase, the test results were based on mmotm as of June 23rd.
Conceptually, this series is simple but there are a lot of details.
Some of the broad motivations for this are;
1. The residency of a page partially depends on what zone the page was
allocated from. This is partially combatted by the fair zone allocation
policy but that is a partial solution that introduces overhead in the
page allocator paths.
2. Currently, reclaim on node 0 behaves slightly different to node 1. For
example, direct reclaim scans in zonelist order and reclaims even if
the zone is over the high watermark regardless of the age of pages
in that LRU. Kswapd on the other hand starts reclaim on the highest
unbalanced zone. A difference in distribution of file/anon pages due
to when they were allocated results can result in a difference in
again. While the fair zone allocation policy mitigates some of the
problems here, the page reclaim results on a multi-zone node will
always be different to a single-zone node.
it was scheduled on as a result.
3. kswapd and the page allocator scan zones in the opposite order to
avoid interfering with each other but it's sensitive to timing. This
mitigates the page allocator using pages that were allocated very recently
in the ideal case but it's sensitive to timing. When kswapd is allocating
from lower zones then it's great but during the rebalancing of the highest
zone, the page allocator and kswapd interfere with each other. It's worse
if the highest zone is small and difficult to balance.
4. slab shrinkers are node-based which makes it harder to identify the exact
relationship between slab reclaim and LRU reclaim.
The reason we have zone-based reclaim is that we used to have
large highmem zones in common configurations and it was necessary
to quickly find ZONE_NORMAL pages for reclaim. Today, this is much
less of a concern as machines with lots of memory will (or should) use
64-bit kernels. Combinations of 32-bit hardware and 64-bit hardware are
rare. Machines that do use highmem should have relatively low highmem:lowmem
ratios than we worried about in the past.
Conceptually, moving to node LRUs should be easier to understand. The
page allocator plays fewer tricks to game reclaim and reclaim behaves
similarly on all nodes.
The series has been tested on a 16 core UMA machine and a 2-socket 48
core NUMA machine. The UMA results are presented in most cases as the NUMA
machine behaved similarly.
pagealloc
---------
This is a microbenchmark that shows the benefit of removing the fair zone
allocation policy. It was tested uip to order-4 but only orders 0 and 1 are
shown as the other orders were comparable.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min total-odr0-1 490.00 ( 0.00%) 457.00 ( 6.73%)
Min total-odr0-2 347.00 ( 0.00%) 329.00 ( 5.19%)
Min total-odr0-4 288.00 ( 0.00%) 273.00 ( 5.21%)
Min total-odr0-8 251.00 ( 0.00%) 239.00 ( 4.78%)
Min total-odr0-16 234.00 ( 0.00%) 222.00 ( 5.13%)
Min total-odr0-32 223.00 ( 0.00%) 211.00 ( 5.38%)
Min total-odr0-64 217.00 ( 0.00%) 208.00 ( 4.15%)
Min total-odr0-128 214.00 ( 0.00%) 204.00 ( 4.67%)
Min total-odr0-256 250.00 ( 0.00%) 230.00 ( 8.00%)
Min total-odr0-512 271.00 ( 0.00%) 269.00 ( 0.74%)
Min total-odr0-1024 291.00 ( 0.00%) 282.00 ( 3.09%)
Min total-odr0-2048 303.00 ( 0.00%) 296.00 ( 2.31%)
Min total-odr0-4096 311.00 ( 0.00%) 309.00 ( 0.64%)
Min total-odr0-8192 316.00 ( 0.00%) 314.00 ( 0.63%)
Min total-odr0-16384 317.00 ( 0.00%) 315.00 ( 0.63%)
Min total-odr1-1 742.00 ( 0.00%) 712.00 ( 4.04%)
Min total-odr1-2 562.00 ( 0.00%) 530.00 ( 5.69%)
Min total-odr1-4 457.00 ( 0.00%) 433.00 ( 5.25%)
Min total-odr1-8 411.00 ( 0.00%) 381.00 ( 7.30%)
Min total-odr1-16 381.00 ( 0.00%) 356.00 ( 6.56%)
Min total-odr1-32 372.00 ( 0.00%) 346.00 ( 6.99%)
Min total-odr1-64 372.00 ( 0.00%) 343.00 ( 7.80%)
Min total-odr1-128 375.00 ( 0.00%) 351.00 ( 6.40%)
Min total-odr1-256 379.00 ( 0.00%) 351.00 ( 7.39%)
Min total-odr1-512 385.00 ( 0.00%) 355.00 ( 7.79%)
Min total-odr1-1024 386.00 ( 0.00%) 358.00 ( 7.25%)
Min total-odr1-2048 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-4096 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-8192 388.00 ( 0.00%) 363.00 ( 6.44%)
This shows a steady improvement throughout. The primary benefit is from
reduced system CPU usage which is obvious from the overall times;
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
User 189.19 191.80
System 2604.45 2533.56
Elapsed 2855.30 2786.39
The vmstats also showed that the fair zone allocation policy was definitely
removed as can be seen here;
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v8
DMA32 allocs 28794729769 0
Normal allocs 48432501431 77227309877
Movable allocs 0 0
tiobench on ext4
----------------
tiobench is a benchmark that artifically benefits if old pages remain resident
while new pages get reclaimed. The fair zone allocation policy mitigates this
problem so pages age fairly. While the benchmark has problems, it is important
that tiobench performance remains constant as it implies that page aging
problems that the fair zone allocation policy fixes are not re-introduced.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min PotentialReadSpeed 89.65 ( 0.00%) 90.21 ( 0.62%)
Min SeqRead-MB/sec-1 82.68 ( 0.00%) 82.01 ( -0.81%)
Min SeqRead-MB/sec-2 72.76 ( 0.00%) 72.07 ( -0.95%)
Min SeqRead-MB/sec-4 75.13 ( 0.00%) 74.92 ( -0.28%)
Min SeqRead-MB/sec-8 64.91 ( 0.00%) 65.19 ( 0.43%)
Min SeqRead-MB/sec-16 62.24 ( 0.00%) 62.22 ( -0.03%)
Min RandRead-MB/sec-1 0.88 ( 0.00%) 0.88 ( 0.00%)
Min RandRead-MB/sec-2 0.95 ( 0.00%) 0.92 ( -3.16%)
Min RandRead-MB/sec-4 1.43 ( 0.00%) 1.34 ( -6.29%)
Min RandRead-MB/sec-8 1.61 ( 0.00%) 1.60 ( -0.62%)
Min RandRead-MB/sec-16 1.80 ( 0.00%) 1.90 ( 5.56%)
Min SeqWrite-MB/sec-1 76.41 ( 0.00%) 76.85 ( 0.58%)
Min SeqWrite-MB/sec-2 74.11 ( 0.00%) 73.54 ( -0.77%)
Min SeqWrite-MB/sec-4 80.05 ( 0.00%) 80.13 ( 0.10%)
Min SeqWrite-MB/sec-8 72.88 ( 0.00%) 73.20 ( 0.44%)
Min SeqWrite-MB/sec-16 75.91 ( 0.00%) 76.44 ( 0.70%)
Min RandWrite-MB/sec-1 1.18 ( 0.00%) 1.14 ( -3.39%)
Min RandWrite-MB/sec-2 1.02 ( 0.00%) 1.03 ( 0.98%)
Min RandWrite-MB/sec-4 1.05 ( 0.00%) 0.98 ( -6.67%)
Min RandWrite-MB/sec-8 0.89 ( 0.00%) 0.92 ( 3.37%)
Min RandWrite-MB/sec-16 0.92 ( 0.00%) 0.93 ( 1.09%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 approx-v9
User 645.72 525.90
System 403.85 331.75
Elapsed 6795.36 6783.67
This shows that the series has little or not impact on tiobench which is
desirable and a reduction in system CPU usage. It indicates that the fair
zone allocation policy was removed in a manner that didn't reintroduce
one class of page aging bug. There were only minor differences in overall
reclaim activity
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Minor Faults 645838 647465
Major Faults 573 640
Swap Ins 0 0
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 46041453 44190646
Normal allocs 78053072 79887245
Movable allocs 0 0
Allocation stalls 24 67
Stall zone DMA 0 0
Stall zone DMA32 0 0
Stall zone Normal 0 2
Stall zone HighMem 0 0
Stall zone Movable 0 65
Direct pages scanned 10969 30609
Kswapd pages scanned 93375144 93492094
Kswapd pages reclaimed 93372243 93489370
Direct pages reclaimed 10969 30609
Kswapd efficiency 99% 99%
Kswapd velocity 13741.015 13781.934
Direct efficiency 100% 100%
Direct velocity 1.614 4.512
Percentage direct scans 0% 0%
kswapd activity was roughly comparable. There were differences in direct
reclaim activity but negligible in the context of the overall workload
(velocity of 4 pages per second with the patches applied, 1.6 pages per
second in the baseline kernel).
pgbench read-only large configuration on ext4
---------------------------------------------
pgbench is a database benchmark that can be sensitive to page reclaim
decisions. This also checks if removing the fair zone allocation policy
is safe
pgbench Transactions
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Hmean 1 188.26 ( 0.00%) 189.78 ( 0.81%)
Hmean 5 330.66 ( 0.00%) 328.69 ( -0.59%)
Hmean 12 370.32 ( 0.00%) 380.72 ( 2.81%)
Hmean 21 368.89 ( 0.00%) 369.00 ( 0.03%)
Hmean 30 382.14 ( 0.00%) 360.89 ( -5.56%)
Hmean 32 428.87 ( 0.00%) 432.96 ( 0.95%)
Negligible differences again. As with tiobench, overall reclaim activity
was comparable.
bonnie++ on ext4
----------------
No interesting performance difference, negligible differences on reclaim
stats.
paralleldd on ext4
------------------
This workload uses varying numbers of dd instances to read large amounts of
data from disk.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Amean Elapsd-1 186.04 ( 0.00%) 189.41 ( -1.82%)
Amean Elapsd-3 192.27 ( 0.00%) 191.38 ( 0.46%)
Amean Elapsd-5 185.21 ( 0.00%) 182.75 ( 1.33%)
Amean Elapsd-7 183.71 ( 0.00%) 182.11 ( 0.87%)
Amean Elapsd-12 180.96 ( 0.00%) 181.58 ( -0.35%)
Amean Elapsd-16 181.36 ( 0.00%) 183.72 ( -1.30%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
User 1548.01 1552.44
System 8609.71 8515.08
Elapsed 3587.10 3594.54
There is little or no change in performance but some drop in system CPU usage.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Minor Faults 362662 367360
Major Faults 1204 1143
Swap Ins 22 0
Swap Outs 2855 1029
DMA allocs 0 0
DMA32 allocs 31409797 28837521
Normal allocs 46611853 49231282
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 40845270 40869088
Kswapd pages reclaimed 40830976 40855294
Direct pages reclaimed 0 0
Kswapd efficiency 99% 99%
Kswapd velocity 11386.711 11369.769
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Page writes by reclaim 2855 1029
Page writes file 0 0
Page writes anon 2855 1029
Page reclaim immediate 771 1628
Sector Reads 293312636 293536360
Sector Writes 18213568 18186480
Page rescued immediate 0 0
Slabs scanned 128257 132747
Direct inode steals 181 56
Kswapd inode steals 59 1131
It basically shows that kswapd was active at roughly the same rate in
both kernels. There was also comparable slab scanning activity and direct
reclaim was avoided in both cases. There appears to be a large difference
in numbers of inodes reclaimed but the workload has few active inodes and
is likely a timing artifact.
stutter
-------
stutter simulates a simple workload. One part uses a lot of anonymous
memory, a second measures mmap latency and a third copies a large file.
The primary metric is checking for mmap latency.
stutter
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Min mmap 16.6283 ( 0.00%) 13.4258 ( 19.26%)
1st-qrtle mmap 54.7570 ( 0.00%) 34.9121 ( 36.24%)
2nd-qrtle mmap 57.3163 ( 0.00%) 46.1147 ( 19.54%)
3rd-qrtle mmap 58.9976 ( 0.00%) 47.1882 ( 20.02%)
Max-90% mmap 59.7433 ( 0.00%) 47.4453 ( 20.58%)
Max-93% mmap 60.1298 ( 0.00%) 47.6037 ( 20.83%)
Max-95% mmap 73.4112 ( 0.00%) 82.8719 (-12.89%)
Max-99% mmap 92.8542 ( 0.00%) 88.8870 ( 4.27%)
Max mmap 1440.6569 ( 0.00%) 121.4201 ( 91.57%)
Mean mmap 59.3493 ( 0.00%) 42.2991 ( 28.73%)
Best99%Mean mmap 57.2121 ( 0.00%) 41.8207 ( 26.90%)
Best95%Mean mmap 55.9113 ( 0.00%) 39.9620 ( 28.53%)
Best90%Mean mmap 55.6199 ( 0.00%) 39.3124 ( 29.32%)
Best50%Mean mmap 53.2183 ( 0.00%) 33.1307 ( 37.75%)
Best10%Mean mmap 45.9842 ( 0.00%) 20.4040 ( 55.63%)
Best5%Mean mmap 43.2256 ( 0.00%) 17.9654 ( 58.44%)
Best1%Mean mmap 32.9388 ( 0.00%) 16.6875 ( 49.34%)
This shows a number of improvements with the worst-case outlier greatly
improved.
Some of the vmstats are interesting
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Swap Ins 163 502
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 618719206 1381662383
Normal allocs 891235743 564138421
Movable allocs 0 0
Allocation stalls 2603 1
Direct pages scanned 216787 2
Kswapd pages scanned 50719775 41778378
Kswapd pages reclaimed 41541765 41777639
Direct pages reclaimed 209159 0
Kswapd efficiency 81% 99%
Kswapd velocity 16859.554 14329.059
Direct efficiency 96% 0%
Direct velocity 72.061 0.001
Percentage direct scans 0% 0%
Page writes by reclaim 6215049 0
Page writes file 6215049 0
Page writes anon 0 0
Page reclaim immediate 70673 90
Sector Reads 81940800 81680456
Sector Writes 100158984 98816036
Page rescued immediate 0 0
Slabs scanned 1366954 22683
While this is not guaranteed in all cases, this particular test showed
a large reduction in direct reclaim activity. It's also worth noting
that no page writes were issued from reclaim context.
This series is not without its hazards. There are at least three areas
that I'm concerned with even though I could not reproduce any problems in
that area.
1. Reclaim/compaction is going to be affected because the amount of reclaim is
no longer targetted at a specific zone. Compaction works on a per-zone basis
so there is no guarantee that reclaiming a few THP's worth page pages will
have a positive impact on compaction success rates.
2. The Slab/LRU reclaim ratio is affected because the frequency the shrinkers
are called is now different. This may or may not be a problem but if it
is, it'll be because shrinkers are not called enough and some balancing
is required.
3. The anon/file reclaim ratio may be affected. Pages about to be dirtied are
distributed between zones and the fair zone allocation policy used to do
something very similar for anon. The distribution is now different but not
necessarily in any way that matters but it's still worth bearing in mind.
VM statistic counters for reclaim decisions are zone-based. If the kernel
is to reclaim on a per-node basis then we need to track per-node
statistics but there is no infrastructure for that. The most notable
change is that the old node_page_state is renamed to
sum_zone_node_page_state. The new node_page_state takes a pglist_data and
uses per-node stats but none exist yet. There is some renaming such as
vm_stat to vm_zone_stat and the addition of vm_node_stat and the renaming
of mod_state to mod_zone_state. Otherwise, this is mostly a mechanical
patch with no functional change. There is a lot of similarity between the
node and zone helpers which is unfortunate but there was no obvious way of
reusing the code and maintaining type safety.
Link: http://lkml.kernel.org/r/1467970510-21195-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:45:24 +00:00
|
|
|
struct per_cpu_nodestat {
|
|
|
|
s8 stat_threshold;
|
|
|
|
s8 vm_node_stat_diff[NR_VM_NODE_STAT_ITEMS];
|
|
|
|
};
|
|
|
|
|
2008-04-28 09:12:54 +00:00
|
|
|
#endif /* !__GENERATING_BOUNDS.H */
|
|
|
|
|
2006-09-26 06:31:13 +00:00
|
|
|
enum zone_type {
|
|
|
|
/*
|
2019-09-11 18:25:46 +00:00
|
|
|
* ZONE_DMA and ZONE_DMA32 are used when there are peripherals not able
|
|
|
|
* to DMA to all of the addressable memory (ZONE_NORMAL).
|
|
|
|
* On architectures where this area covers the whole 32 bit address
|
|
|
|
* space ZONE_DMA32 is used. ZONE_DMA is left for the ones with smaller
|
|
|
|
* DMA addressing constraints. This distinction is important as a 32bit
|
|
|
|
* DMA mask is assumed when ZONE_DMA32 is defined. Some 64-bit
|
|
|
|
* platforms may need both zones as they support peripherals with
|
|
|
|
* different DMA addressing limitations.
|
2006-09-26 06:31:13 +00:00
|
|
|
*/
|
2019-09-11 18:25:46 +00:00
|
|
|
#ifdef CONFIG_ZONE_DMA
|
2006-09-26 06:31:13 +00:00
|
|
|
ZONE_DMA,
|
2007-02-10 09:43:10 +00:00
|
|
|
#endif
|
2006-09-26 06:31:13 +00:00
|
|
|
#ifdef CONFIG_ZONE_DMA32
|
2006-09-26 06:31:13 +00:00
|
|
|
ZONE_DMA32,
|
2006-09-26 06:31:13 +00:00
|
|
|
#endif
|
2006-09-26 06:31:13 +00:00
|
|
|
/*
|
|
|
|
* Normal addressable memory is in ZONE_NORMAL. DMA operations can be
|
|
|
|
* performed on pages in ZONE_NORMAL if the DMA devices support
|
|
|
|
* transfers to all addressable memory.
|
|
|
|
*/
|
|
|
|
ZONE_NORMAL,
|
2006-09-26 06:31:14 +00:00
|
|
|
#ifdef CONFIG_HIGHMEM
|
2006-09-26 06:31:13 +00:00
|
|
|
/*
|
|
|
|
* A memory area that is only addressable by the kernel through
|
|
|
|
* mapping portions into its own address space. This is for example
|
|
|
|
* used by i386 to allow the kernel to address the memory beyond
|
|
|
|
* 900MB. The kernel will set up special mappings (page
|
|
|
|
* table entries on i386) for each page that the kernel needs to
|
|
|
|
* access.
|
|
|
|
*/
|
|
|
|
ZONE_HIGHMEM,
|
2006-09-26 06:31:14 +00:00
|
|
|
#endif
|
2020-10-13 23:55:35 +00:00
|
|
|
/*
|
|
|
|
* ZONE_MOVABLE is similar to ZONE_NORMAL, except that it contains
|
|
|
|
* movable pages with few exceptional cases described below. Main use
|
|
|
|
* cases for ZONE_MOVABLE are to make memory offlining/unplug more
|
|
|
|
* likely to succeed, and to locally limit unmovable allocations - e.g.,
|
|
|
|
* to increase the number of THP/huge pages. Notable special cases are:
|
|
|
|
*
|
|
|
|
* 1. Pinned pages: (long-term) pinning of movable pages might
|
2021-05-05 01:39:08 +00:00
|
|
|
* essentially turn such pages unmovable. Therefore, we do not allow
|
|
|
|
* pinning long-term pages in ZONE_MOVABLE. When pages are pinned and
|
|
|
|
* faulted, they come from the right zone right away. However, it is
|
|
|
|
* still possible that address space already has pages in
|
|
|
|
* ZONE_MOVABLE at the time when pages are pinned (i.e. user has
|
|
|
|
* touches that memory before pinning). In such case we migrate them
|
|
|
|
* to a different zone. When migration fails - pinning fails.
|
2020-10-13 23:55:35 +00:00
|
|
|
* 2. memblock allocations: kernelcore/movablecore setups might create
|
|
|
|
* situations where ZONE_MOVABLE contains unmovable allocations
|
|
|
|
* after boot. Memory offlining and allocations fail early.
|
|
|
|
* 3. Memory holes: kernelcore/movablecore setups might create very rare
|
|
|
|
* situations where ZONE_MOVABLE contains memory holes after boot,
|
|
|
|
* for example, if we have sections that are only partially
|
|
|
|
* populated. Memory offlining and allocations fail early.
|
|
|
|
* 4. PG_hwpoison pages: while poisoned pages can be skipped during
|
|
|
|
* memory offlining, such pages cannot be allocated.
|
|
|
|
* 5. Unmovable PG_offline pages: in paravirtualized environments,
|
|
|
|
* hotplugged memory blocks might only partially be managed by the
|
|
|
|
* buddy (e.g., via XEN-balloon, Hyper-V balloon, virtio-mem). The
|
|
|
|
* parts not manged by the buddy are unmovable PG_offline pages. In
|
|
|
|
* some cases (virtio-mem), such pages can be skipped during
|
|
|
|
* memory offlining, however, cannot be moved/allocated. These
|
|
|
|
* techniques might use alloc_contig_range() to hide previously
|
|
|
|
* exposed pages from the buddy again (e.g., to implement some sort
|
|
|
|
* of memory unplug in virtio-mem).
|
2021-05-05 01:39:04 +00:00
|
|
|
* 6. ZERO_PAGE(0), kernelcore/movablecore setups might create
|
|
|
|
* situations where ZERO_PAGE(0) which is allocated differently
|
|
|
|
* on different platforms may end up in a movable zone. ZERO_PAGE(0)
|
|
|
|
* cannot be migrated.
|
mm,memory_hotplug: allocate memmap from the added memory range
Physical memory hotadd has to allocate a memmap (struct page array) for
the newly added memory section. Currently, alloc_pages_node() is used
for those allocations.
This has some disadvantages:
a) an existing memory is consumed for that purpose
(eg: ~2MB per 128MB memory section on x86_64)
This can even lead to extreme cases where system goes OOM because
the physically hotplugged memory depletes the available memory before
it is onlined.
b) if the whole node is movable then we have off-node struct pages
which has performance drawbacks.
c) It might be there are no PMD_ALIGNED chunks so memmap array gets
populated with base pages.
This can be improved when CONFIG_SPARSEMEM_VMEMMAP is enabled.
Vmemap page tables can map arbitrary memory. That means that we can
reserve a part of the physically hotadded memory to back vmemmap page
tables. This implementation uses the beginning of the hotplugged memory
for that purpose.
There are some non-obviously things to consider though.
Vmemmap pages are allocated/freed during the memory hotplug events
(add_memory_resource(), try_remove_memory()) when the memory is
added/removed. This means that the reserved physical range is not
online although it is used. The most obvious side effect is that
pfn_to_online_page() returns NULL for those pfns. The current design
expects that this should be OK as the hotplugged memory is considered a
garbage until it is onlined. For example hibernation wouldn't save the
content of those vmmemmaps into the image so it wouldn't be restored on
resume but this should be OK as there no real content to recover anyway
while metadata is reachable from other data structures (e.g. vmemmap
page tables).
The reserved space is therefore (de)initialized during the {on,off}line
events (mhp_{de}init_memmap_on_memory). That is done by extracting page
allocator independent initialization from the regular onlining path.
The primary reason to handle the reserved space outside of
{on,off}line_pages is to make each initialization specific to the
purpose rather than special case them in a single function.
As per above, the functions that are introduced are:
- mhp_init_memmap_on_memory:
Initializes vmemmap pages by calling move_pfn_range_to_zone(), calls
kasan_add_zero_shadow(), and onlines as many sections as vmemmap pages
fully span.
- mhp_deinit_memmap_on_memory:
Offlines as many sections as vmemmap pages fully span, removes the
range from zhe zone by remove_pfn_range_from_zone(), and calls
kasan_remove_zero_shadow() for the range.
The new function memory_block_online() calls mhp_init_memmap_on_memory()
before doing the actual online_pages(). Should online_pages() fail, we
clean up by calling mhp_deinit_memmap_on_memory(). Adjusting of
present_pages is done at the end once we know that online_pages()
succedeed.
On offline, memory_block_offline() needs to unaccount vmemmap pages from
present_pages() before calling offline_pages(). This is necessary because
offline_pages() tears down some structures based on the fact whether the
node or the zone become empty. If offline_pages() fails, we account back
vmemmap pages. If it succeeds, we call mhp_deinit_memmap_on_memory().
Hot-remove:
We need to be careful when removing memory, as adding and
removing memory needs to be done with the same granularity.
To check that this assumption is not violated, we check the
memory range we want to remove and if a) any memory block has
vmemmap pages and b) the range spans more than a single memory
block, we scream out loud and refuse to proceed.
If all is good and the range was using memmap on memory (aka vmemmap pages),
we construct an altmap structure so free_hugepage_table does the right
thing and calls vmem_altmap_free instead of free_pagetable.
Link: https://lkml.kernel.org/r/20210421102701.25051-5-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.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 01:39:42 +00:00
|
|
|
* 7. Memory-hotplug: when using memmap_on_memory and onlining the
|
|
|
|
* memory to the MOVABLE zone, the vmemmap pages are also placed in
|
|
|
|
* such zone. Such pages cannot be really moved around as they are
|
|
|
|
* self-stored in the range, but they are treated as movable when
|
|
|
|
* the range they describe is about to be offlined.
|
2020-10-13 23:55:35 +00:00
|
|
|
*
|
|
|
|
* In general, no unmovable allocations that degrade memory offlining
|
|
|
|
* should end up in ZONE_MOVABLE. Allocators (like alloc_contig_range())
|
|
|
|
* have to expect that migrating pages in ZONE_MOVABLE can fail (even
|
|
|
|
* if has_unmovable_pages() states that there are no unmovable pages,
|
|
|
|
* there can be false negatives).
|
|
|
|
*/
|
2007-07-17 11:03:12 +00:00
|
|
|
ZONE_MOVABLE,
|
2015-08-09 19:29:06 +00:00
|
|
|
#ifdef CONFIG_ZONE_DEVICE
|
|
|
|
ZONE_DEVICE,
|
|
|
|
#endif
|
2008-04-28 09:12:54 +00:00
|
|
|
__MAX_NR_ZONES
|
2015-08-09 19:29:06 +00:00
|
|
|
|
2006-09-26 06:31:13 +00:00
|
|
|
};
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2008-04-28 09:12:54 +00:00
|
|
|
#ifndef __GENERATING_BOUNDS_H
|
|
|
|
|
2020-10-16 03:09:55 +00:00
|
|
|
#define ASYNC_AND_SYNC 2
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
struct zone {
|
2014-08-06 23:07:14 +00:00
|
|
|
/* Read-mostly fields */
|
2009-06-16 22:32:12 +00:00
|
|
|
|
|
|
|
/* zone watermarks, access with *_wmark_pages(zone) macros */
|
2018-12-28 08:35:44 +00:00
|
|
|
unsigned long _watermark[NR_WMARK];
|
mm: reclaim small amounts of memory when an external fragmentation event occurs
An external fragmentation event was previously described as
When the page allocator fragments memory, it records the event using
the mm_page_alloc_extfrag event. If the fallback_order is smaller
than a pageblock order (order-9 on 64-bit x86) then it's considered
an event that will cause external fragmentation issues in the future.
The kernel reduces the probability of such events by increasing the
watermark sizes by calling set_recommended_min_free_kbytes early in the
lifetime of the system. This works reasonably well in general but if
there are enough sparsely populated pageblocks then the problem can still
occur as enough memory is free overall and kswapd stays asleep.
This patch introduces a watermark_boost_factor sysctl that allows a zone
watermark to be temporarily boosted when an external fragmentation causing
events occurs. The boosting will stall allocations that would decrease
free memory below the boosted low watermark and kswapd is woken if the
calling context allows to reclaim an amount of memory relative to the size
of the high watermark and the watermark_boost_factor until the boost is
cleared. When kswapd finishes, it wakes kcompactd at the pageblock order
to clean some of the pageblocks that may have been affected by the
fragmentation event. kswapd avoids any writeback, slab shrinkage and swap
from reclaim context during this operation to avoid excessive system
disruption in the name of fragmentation avoidance. Care is taken so that
kswapd will do normal reclaim work if the system is really low on memory.
This was evaluated using the same workloads as "mm, page_alloc: Spread
allocations across zones before introducing fragmentation".
1-socket Skylake machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 1 THP allocating thread
--------------------------------------
4.20-rc3 extfrag events < order 9: 804694
4.20-rc3+patch: 408912 (49% reduction)
4.20-rc3+patch1-4: 18421 (98% reduction)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Amean fault-base-1 653.58 ( 0.00%) 652.71 ( 0.13%)
Amean fault-huge-1 0.00 ( 0.00%) 178.93 * -99.00%*
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-1 0.00 ( 0.00%) 5.12 ( 100.00%)
Note that external fragmentation causing events are massively reduced by
this path whether in comparison to the previous kernel or the vanilla
kernel. The fault latency for huge pages appears to be increased but that
is only because THP allocations were successful with the patch applied.
1-socket Skylake machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------
4.20-rc3 extfrag events < order 9: 291392
4.20-rc3+patch: 191187 (34% reduction)
4.20-rc3+patch1-4: 13464 (95% reduction)
thpfioscale Fault Latencies
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Min fault-base-1 912.00 ( 0.00%) 905.00 ( 0.77%)
Min fault-huge-1 127.00 ( 0.00%) 135.00 ( -6.30%)
Amean fault-base-1 1467.55 ( 0.00%) 1481.67 ( -0.96%)
Amean fault-huge-1 1127.11 ( 0.00%) 1063.88 * 5.61%*
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-1 77.64 ( 0.00%) 83.46 ( 7.49%)
As before, massive reduction in external fragmentation events, some jitter
on latencies and an increase in THP allocation success rates.
2-socket Haswell machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 5 THP allocating threads
----------------------------------------------------------------
4.20-rc3 extfrag events < order 9: 215698
4.20-rc3+patch: 200210 (7% reduction)
4.20-rc3+patch1-4: 14263 (93% reduction)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Amean fault-base-5 1346.45 ( 0.00%) 1306.87 ( 2.94%)
Amean fault-huge-5 3418.60 ( 0.00%) 1348.94 ( 60.54%)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-5 0.78 ( 0.00%) 7.91 ( 910.64%)
There is a 93% reduction in fragmentation causing events, there is a big
reduction in the huge page fault latency and allocation success rate is
higher.
2-socket Haswell machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------
4.20-rc3 extfrag events < order 9: 166352
4.20-rc3+patch: 147463 (11% reduction)
4.20-rc3+patch1-4: 11095 (93% reduction)
thpfioscale Fault Latencies
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Amean fault-base-5 6217.43 ( 0.00%) 7419.67 * -19.34%*
Amean fault-huge-5 3163.33 ( 0.00%) 3263.80 ( -3.18%)
4.20.0-rc3 4.20.0-rc3
lowzone-v5r8 boost-v5r8
Percentage huge-5 95.14 ( 0.00%) 87.98 ( -7.53%)
There is a large reduction in fragmentation events with some jitter around
the latencies and success rates. As before, the high THP allocation
success rate does mean the system is under a lot of pressure. However, as
the fragmentation events are reduced, it would be expected that the
long-term allocation success rate would be higher.
Link: http://lkml.kernel.org/r/20181123114528.28802-5-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: David Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 08:35:52 +00:00
|
|
|
unsigned long watermark_boost;
|
2009-06-16 22:32:12 +00:00
|
|
|
|
2015-11-07 00:28:37 +00:00
|
|
|
unsigned long nr_reserved_highatomic;
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
2015-11-07 00:28:46 +00:00
|
|
|
* We don't know if the memory that we're going to allocate will be
|
|
|
|
* freeable or/and it will be released eventually, so to avoid totally
|
|
|
|
* wasting several GB of ram we must reserve some of the lower zone
|
|
|
|
* memory (otherwise we risk to run OOM on the lower zones despite
|
|
|
|
* there being tons of freeable ram on the higher zones). This array is
|
|
|
|
* recalculated at runtime if the sysctl_lowmem_reserve_ratio sysctl
|
|
|
|
* changes.
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
2014-08-06 23:07:14 +00:00
|
|
|
long lowmem_reserve[MAX_NR_ZONES];
|
2012-01-10 23:07:42 +00:00
|
|
|
|
2005-06-22 00:14:47 +00:00
|
|
|
#ifdef CONFIG_NUMA
|
2006-09-27 08:50:08 +00:00
|
|
|
int node;
|
2014-08-06 23:07:14 +00:00
|
|
|
#endif
|
|
|
|
struct pglist_data *zone_pgdat;
|
mm/page_alloc: split per cpu page lists and zone stats
The PCP (per-cpu page allocator in page_alloc.c) shares locking
requirements with vmstat and the zone lock which is inconvenient and
causes some issues. For example, the PCP list and vmstat share the same
per-cpu space meaning that it's possible that vmstat updates dirty cache
lines holding per-cpu lists across CPUs unless padding is used. Second,
PREEMPT_RT does not want to disable IRQs for too long in the page
allocator.
This series splits the locking requirements and uses locks types more
suitable for PREEMPT_RT, reduces the time when special locking is required
for stats and reduces the time when IRQs need to be disabled on
!PREEMPT_RT kernels.
Why local_lock? PREEMPT_RT considers the following sequence to be unsafe
as documented in Documentation/locking/locktypes.rst
local_irq_disable();
spin_lock(&lock);
The pcp allocator has this sequence for rmqueue_pcplist (local_irq_save)
-> __rmqueue_pcplist -> rmqueue_bulk (spin_lock). While it's possible to
separate this out, it generally means there are points where we enable
IRQs and reenable them again immediately. To prevent a migration and the
per-cpu pointer going stale, migrate_disable is also needed. That is a
custom lock that is similar, but worse, than local_lock. Furthermore, on
PREEMPT_RT, it's undesirable to leave IRQs disabled for too long. By
converting to local_lock which disables migration on PREEMPT_RT, the
locking requirements can be separated and start moving the protections for
PCP, stats and the zone lock to PREEMPT_RT-safe equivalent locking. As a
bonus, local_lock also means that PROVE_LOCKING does something useful.
After that, it's obvious that zone_statistics incurs too much overhead and
leaves IRQs disabled for longer than necessary on !PREEMPT_RT kernels.
zone_statistics uses perfectly accurate counters requiring IRQs be
disabled for parallel RMW sequences when inaccurate ones like vm_events
would do. The series makes the NUMA statistics (NUMA_HIT and friends)
inaccurate counters that then require no special protection on
!PREEMPT_RT.
The bulk page allocator can then do stat updates in bulk with IRQs enabled
which should improve the efficiency. Technically, this could have been
done without the local_lock and vmstat conversion work and the order
simply reflects the timing of when different series were implemented.
Finally, there are places where we conflate IRQs being disabled for the
PCP with the IRQ-safe zone spinlock. The remainder of the series reduces
the scope of what is protected by disabled IRQs on !PREEMPT_RT kernels.
By the end of the series, page_alloc.c does not call local_irq_save so the
locking scope is a bit clearer. The one exception is that modifying
NR_FREE_PAGES still happens in places where it's known the IRQs are
disabled as it's harmless for PREEMPT_RT and would be expensive to split
the locking there.
No performance data is included because despite the overhead of the stats,
it's within the noise for most workloads on !PREEMPT_RT. However, Jesper
Dangaard Brouer ran a page allocation microbenchmark on a E5-1650 v4 @
3.60GHz CPU on the first version of this series. Focusing on the array
variant of the bulk page allocator reveals the following.
(CPU: Intel(R) Xeon(R) CPU E5-1650 v4 @ 3.60GHz)
ARRAY variant: time_bulk_page_alloc_free_array: step=bulk size
Baseline Patched
1 56.383 54.225 (+3.83%)
2 40.047 35.492 (+11.38%)
3 37.339 32.643 (+12.58%)
4 35.578 30.992 (+12.89%)
8 33.592 29.606 (+11.87%)
16 32.362 28.532 (+11.85%)
32 31.476 27.728 (+11.91%)
64 30.633 27.252 (+11.04%)
128 30.596 27.090 (+11.46%)
While this is a positive outcome, the series is more likely to be
interesting to the RT people in terms of getting parts of the PREEMPT_RT
tree into mainline.
This patch (of 9):
The per-cpu page allocator lists and the per-cpu vmstat deltas are stored
in the same struct per_cpu_pages even though vmstats have no direct impact
on the per-cpu page lists. This is inconsistent because the vmstats for a
node are stored on a dedicated structure. The bigger issue is that the
per_cpu_pages structure is not cache-aligned and stat updates either cache
conflict with adjacent per-cpu lists incurring a runtime cost or padding
is required incurring a memory cost.
This patch splits the per-cpu pagelists and the vmstat deltas into
separate structures. It's mostly a mechanical conversion but some
variable renaming is done to clearly distinguish the per-cpu pages
structure (pcp) from the vmstats (pzstats).
Superficially, this appears to increase the size of the per_cpu_pages
structure but the movement of expire fills a structure hole so there is no
impact overall.
[mgorman@techsingularity.net: make it W=1 cleaner]
Link: https://lkml.kernel.org/r/20210514144622.GA3735@techsingularity.net
[mgorman@techsingularity.net: make it W=1 even cleaner]
Link: https://lkml.kernel.org/r/20210516140705.GB3735@techsingularity.net
[lkp@intel.com: check struct per_cpu_zonestat has a non-zero size]
[vbabka@suse.cz: Init zone->per_cpu_zonestats properly]
Link: https://lkml.kernel.org/r/20210512095458.30632-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20210512095458.30632-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-06-29 02:41:38 +00:00
|
|
|
struct per_cpu_pages __percpu *per_cpu_pageset;
|
|
|
|
struct per_cpu_zonestat __percpu *per_cpu_zonestats;
|
2020-12-15 03:10:53 +00:00
|
|
|
/*
|
|
|
|
* the high and batch values are copied to individual pagesets for
|
|
|
|
* faster access
|
|
|
|
*/
|
|
|
|
int pageset_high;
|
|
|
|
int pageset_batch;
|
2014-08-06 23:07:14 +00:00
|
|
|
|
Add a bitmap that is used to track flags affecting a block of pages
Here is the latest revision of the anti-fragmentation patches. Of particular
note in this version is special treatment of high-order atomic allocations.
Care is taken to group them together and avoid grouping pages of other types
near them. Artifical tests imply that it works. I'm trying to get the
hardware together that would allow setting up of a "real" test. If anyone
already has a setup and test that can trigger the atomic-allocation problem,
I'd appreciate a test of these patches and a report. The second major change
is that these patches will apply cleanly with patches that implement
anti-fragmentation through zones.
kernbench shows effectively no performance difference varying between -0.2%
and +2% on a variety of test machines. Success rates for huge page allocation
are dramatically increased. For example, on a ppc64 machine, the vanilla
kernel was only able to allocate 1% of memory as a hugepage and this was due
to a single hugepage reserved as min_free_kbytes. With these patches applied,
17% was allocatable as superpages. With reclaim-related fixes from Andy
Whitcroft, it was 40% and further reclaim-related improvements should increase
this further.
Changelog Since V28
o Group high-order atomic allocations together
o It is no longer required to set min_free_kbytes to 10% of memory. A value
of 16384 in most cases will be sufficient
o Now applied with zone-based anti-fragmentation
o Fix incorrect VM_BUG_ON within buffered_rmqueue()
o Reorder the stack so later patches do not back out work from earlier patches
o Fix bug were journal pages were being treated as movable
o Bias placement of non-movable pages to lower PFNs
o More agressive clustering of reclaimable pages in reactions to workloads
like updatedb that flood the size of inode caches
Changelog Since V27
o Renamed anti-fragmentation to Page Clustering. Anti-fragmentation was giving
the mistaken impression that it was the 100% solution for high order
allocations. Instead, it greatly increases the chances high-order
allocations will succeed and lays the foundation for defragmentation and
memory hot-remove to work properly
o Redefine page groupings based on ability to migrate or reclaim instead of
basing on reclaimability alone
o Get rid of spurious inits
o Per-cpu lists are no longer split up per-type. Instead the per-cpu list is
searched for a page of the appropriate type
o Added more explanation commentary
o Fix up bug in pageblock code where bitmap was used before being initalised
Changelog Since V26
o Fix double init of lists in setup_pageset
Changelog Since V25
o Fix loop order of for_each_rclmtype_order so that order of loop matches args
o gfpflags_to_rclmtype uses gfp_t instead of unsigned long
o Rename get_pageblock_type() to get_page_rclmtype()
o Fix alignment problem in move_freepages()
o Add mechanism for assigning flags to blocks of pages instead of page->flags
o On fallback, do not examine the preferred list of free pages a second time
The purpose of these patches is to reduce external fragmentation by grouping
pages of related types together. When pages are migrated (or reclaimed under
memory pressure), large contiguous pages will be freed.
This patch works by categorising allocations by their ability to migrate;
Movable - The pages may be moved with the page migration mechanism. These are
generally userspace pages.
Reclaimable - These are allocations for some kernel caches that are
reclaimable or allocations that are known to be very short-lived.
Unmovable - These are pages that are allocated by the kernel that
are not trivially reclaimed. For example, the memory allocated for a
loaded module would be in this category. By default, allocations are
considered to be of this type
HighAtomic - These are high-order allocations belonging to callers that
cannot sleep or perform any IO. In practice, this is restricted to
jumbo frame allocation for network receive. It is assumed that the
allocations are short-lived
Instead of having one MAX_ORDER-sized array of free lists in struct free_area,
there is one for each type of reclaimability. Once a 2^MAX_ORDER block of
pages is split for a type of allocation, it is added to the free-lists for
that type, in effect reserving it. Hence, over time, pages of the different
types can be clustered together.
When the preferred freelists are expired, the largest possible block is taken
from an alternative list. Buddies that are split from that large block are
placed on the preferred allocation-type freelists to mitigate fragmentation.
This implementation gives best-effort for low fragmentation in all zones.
Ideally, min_free_kbytes needs to be set to a value equal to 4 * (1 <<
(MAX_ORDER-1)) pages in most cases. This would be 16384 on x86 and x86_64 for
example.
Our tests show that about 60-70% of physical memory can be allocated on a
desktop after a few days uptime. In benchmarks and stress tests, we are
finding that 80% of memory is available as contiguous blocks at the end of the
test. To compare, a standard kernel was getting < 1% of memory as large pages
on a desktop and about 8-12% of memory as large pages at the end of stress
tests.
Following this email are 12 patches that implement thie page grouping feature.
The first patch introduces a mechanism for storing flags related to a whole
block of pages. Then allocations are split between movable and all other
allocations. Following that are patches to deal with per-cpu pages and make
the mechanism configurable. The next patch moves free pages between lists
when partially allocated blocks are used for pages of another migrate type.
The second last patch groups reclaimable kernel allocations such as inode
caches together. The final patch related to groupings keeps high-order atomic
allocations.
The last two patches are more concerned with control of fragmentation. The
second last patch biases placement of non-movable allocations towards the
start of memory. This is with a view of supporting memory hot-remove of DIMMs
with higher PFNs in the future. The biasing could be enforced a lot heavier
but it would cost. The last patch agressively clusters reclaimable pages like
inode caches together.
The fragmentation reduction strategy needs to track if pages within a block
can be moved or reclaimed so that pages are freed to the appropriate list.
This patch adds a bitmap for flags affecting a whole a MAX_ORDER block of
pages.
In non-SPARSEMEM configurations, the bitmap is stored in the struct zone and
allocated during initialisation. SPARSEMEM statically allocates the bitmap in
a struct mem_section so that bitmaps do not have to be resized during memory
hotadd. This wastes a small amount of memory per unused section (usually
sizeof(unsigned long)) but the complexity of dynamically allocating the memory
is quite high.
Additional credit to Andy Whitcroft who reviewed up an earlier implementation
of the mechanism an suggested how to make it a *lot* cleaner.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:47 +00:00
|
|
|
#ifndef CONFIG_SPARSEMEM
|
|
|
|
/*
|
2007-10-16 08:26:01 +00:00
|
|
|
* Flags for a pageblock_nr_pages block. See pageblock-flags.h.
|
Add a bitmap that is used to track flags affecting a block of pages
Here is the latest revision of the anti-fragmentation patches. Of particular
note in this version is special treatment of high-order atomic allocations.
Care is taken to group them together and avoid grouping pages of other types
near them. Artifical tests imply that it works. I'm trying to get the
hardware together that would allow setting up of a "real" test. If anyone
already has a setup and test that can trigger the atomic-allocation problem,
I'd appreciate a test of these patches and a report. The second major change
is that these patches will apply cleanly with patches that implement
anti-fragmentation through zones.
kernbench shows effectively no performance difference varying between -0.2%
and +2% on a variety of test machines. Success rates for huge page allocation
are dramatically increased. For example, on a ppc64 machine, the vanilla
kernel was only able to allocate 1% of memory as a hugepage and this was due
to a single hugepage reserved as min_free_kbytes. With these patches applied,
17% was allocatable as superpages. With reclaim-related fixes from Andy
Whitcroft, it was 40% and further reclaim-related improvements should increase
this further.
Changelog Since V28
o Group high-order atomic allocations together
o It is no longer required to set min_free_kbytes to 10% of memory. A value
of 16384 in most cases will be sufficient
o Now applied with zone-based anti-fragmentation
o Fix incorrect VM_BUG_ON within buffered_rmqueue()
o Reorder the stack so later patches do not back out work from earlier patches
o Fix bug were journal pages were being treated as movable
o Bias placement of non-movable pages to lower PFNs
o More agressive clustering of reclaimable pages in reactions to workloads
like updatedb that flood the size of inode caches
Changelog Since V27
o Renamed anti-fragmentation to Page Clustering. Anti-fragmentation was giving
the mistaken impression that it was the 100% solution for high order
allocations. Instead, it greatly increases the chances high-order
allocations will succeed and lays the foundation for defragmentation and
memory hot-remove to work properly
o Redefine page groupings based on ability to migrate or reclaim instead of
basing on reclaimability alone
o Get rid of spurious inits
o Per-cpu lists are no longer split up per-type. Instead the per-cpu list is
searched for a page of the appropriate type
o Added more explanation commentary
o Fix up bug in pageblock code where bitmap was used before being initalised
Changelog Since V26
o Fix double init of lists in setup_pageset
Changelog Since V25
o Fix loop order of for_each_rclmtype_order so that order of loop matches args
o gfpflags_to_rclmtype uses gfp_t instead of unsigned long
o Rename get_pageblock_type() to get_page_rclmtype()
o Fix alignment problem in move_freepages()
o Add mechanism for assigning flags to blocks of pages instead of page->flags
o On fallback, do not examine the preferred list of free pages a second time
The purpose of these patches is to reduce external fragmentation by grouping
pages of related types together. When pages are migrated (or reclaimed under
memory pressure), large contiguous pages will be freed.
This patch works by categorising allocations by their ability to migrate;
Movable - The pages may be moved with the page migration mechanism. These are
generally userspace pages.
Reclaimable - These are allocations for some kernel caches that are
reclaimable or allocations that are known to be very short-lived.
Unmovable - These are pages that are allocated by the kernel that
are not trivially reclaimed. For example, the memory allocated for a
loaded module would be in this category. By default, allocations are
considered to be of this type
HighAtomic - These are high-order allocations belonging to callers that
cannot sleep or perform any IO. In practice, this is restricted to
jumbo frame allocation for network receive. It is assumed that the
allocations are short-lived
Instead of having one MAX_ORDER-sized array of free lists in struct free_area,
there is one for each type of reclaimability. Once a 2^MAX_ORDER block of
pages is split for a type of allocation, it is added to the free-lists for
that type, in effect reserving it. Hence, over time, pages of the different
types can be clustered together.
When the preferred freelists are expired, the largest possible block is taken
from an alternative list. Buddies that are split from that large block are
placed on the preferred allocation-type freelists to mitigate fragmentation.
This implementation gives best-effort for low fragmentation in all zones.
Ideally, min_free_kbytes needs to be set to a value equal to 4 * (1 <<
(MAX_ORDER-1)) pages in most cases. This would be 16384 on x86 and x86_64 for
example.
Our tests show that about 60-70% of physical memory can be allocated on a
desktop after a few days uptime. In benchmarks and stress tests, we are
finding that 80% of memory is available as contiguous blocks at the end of the
test. To compare, a standard kernel was getting < 1% of memory as large pages
on a desktop and about 8-12% of memory as large pages at the end of stress
tests.
Following this email are 12 patches that implement thie page grouping feature.
The first patch introduces a mechanism for storing flags related to a whole
block of pages. Then allocations are split between movable and all other
allocations. Following that are patches to deal with per-cpu pages and make
the mechanism configurable. The next patch moves free pages between lists
when partially allocated blocks are used for pages of another migrate type.
The second last patch groups reclaimable kernel allocations such as inode
caches together. The final patch related to groupings keeps high-order atomic
allocations.
The last two patches are more concerned with control of fragmentation. The
second last patch biases placement of non-movable allocations towards the
start of memory. This is with a view of supporting memory hot-remove of DIMMs
with higher PFNs in the future. The biasing could be enforced a lot heavier
but it would cost. The last patch agressively clusters reclaimable pages like
inode caches together.
The fragmentation reduction strategy needs to track if pages within a block
can be moved or reclaimed so that pages are freed to the appropriate list.
This patch adds a bitmap for flags affecting a whole a MAX_ORDER block of
pages.
In non-SPARSEMEM configurations, the bitmap is stored in the struct zone and
allocated during initialisation. SPARSEMEM statically allocates the bitmap in
a struct mem_section so that bitmaps do not have to be resized during memory
hotadd. This wastes a small amount of memory per unused section (usually
sizeof(unsigned long)) but the complexity of dynamically allocating the memory
is quite high.
Additional credit to Andy Whitcroft who reviewed up an earlier implementation
of the mechanism an suggested how to make it a *lot* cleaner.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:47 +00:00
|
|
|
* In SPARSEMEM, this map is stored in struct mem_section
|
|
|
|
*/
|
|
|
|
unsigned long *pageblock_flags;
|
|
|
|
#endif /* CONFIG_SPARSEMEM */
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/* zone_start_pfn == zone_start_paddr >> PAGE_SHIFT */
|
|
|
|
unsigned long zone_start_pfn;
|
|
|
|
|
2005-10-30 01:16:53 +00:00
|
|
|
/*
|
2012-12-12 21:52:12 +00:00
|
|
|
* spanned_pages is the total pages spanned by the zone, including
|
|
|
|
* holes, which is calculated as:
|
|
|
|
* spanned_pages = zone_end_pfn - zone_start_pfn;
|
2005-10-30 01:16:53 +00:00
|
|
|
*
|
2012-12-12 21:52:12 +00:00
|
|
|
* present_pages is physical pages existing within the zone, which
|
|
|
|
* is calculated as:
|
2013-03-26 17:30:44 +00:00
|
|
|
* present_pages = spanned_pages - absent_pages(pages in holes);
|
2012-12-12 21:52:12 +00:00
|
|
|
*
|
mm: track present early pages per zone
Patch series "mm/memory_hotplug: "auto-movable" online policy and memory groups", v3.
I. Goal
The goal of this series is improving in-kernel auto-online support. It
tackles the fundamental problems that:
1) We can create zone imbalances when onlining all memory blindly to
ZONE_MOVABLE, in the worst case crashing the system. We have to know
upfront how much memory we are going to hotplug such that we can
safely enable auto-onlining of all hotplugged memory to ZONE_MOVABLE
via "online_movable". This is far from practical and only applicable in
limited setups -- like inside VMs under the RHV/oVirt hypervisor which
will never hotplug more than 3 times the boot memory (and the
limitation is only in place due to the Linux limitation).
2) We see more setups that implement dynamic VM resizing, hot(un)plugging
memory to resize VM memory. In these setups, we might hotplug a lot of
memory, but it might happen in various small steps in both directions
(e.g., 2 GiB -> 8 GiB -> 4 GiB -> 16 GiB ...). virtio-mem is the
primary driver of this upstream right now, performing such dynamic
resizing NUMA-aware via multiple virtio-mem devices.
Onlining all hotplugged memory to ZONE_NORMAL means we basically have
no hotunplug guarantees. Onlining all to ZONE_MOVABLE means we can
easily run into zone imbalances when growing a VM. We want a mixture,
and we want as much memory as reasonable/configured in ZONE_MOVABLE.
Details regarding zone imbalances can be found at [1].
3) Memory devices consist of 1..X memory block devices, however, the
kernel doesn't really track the relationship. Consequently, also user
space has no idea. We want to make per-device decisions.
As one example, for memory hotunplug it doesn't make sense to use a
mixture of zones within a single DIMM: we want all MOVABLE if
possible, otherwise all !MOVABLE, because any !MOVABLE part will easily
block the whole DIMM from getting hotunplugged.
As another example, virtio-mem operates on individual units that span
1..X memory blocks. Similar to a DIMM, we want a unit to either be all
MOVABLE or !MOVABLE. A "unit" can be thought of like a DIMM, however,
all units of a virtio-mem device logically belong together and are
managed (added/removed) by a single driver. We want as much memory of
a virtio-mem device to be MOVABLE as possible.
4) We want memory onlining to be done right from the kernel while adding
memory, not triggered by user space via udev rules; for example, this
is reqired for fast memory hotplug for drivers that add individual
memory blocks, like virito-mem. We want a way to configure a policy in
the kernel and avoid implementing advanced policies in user space.
The auto-onlining support we have in the kernel is not sufficient. All we
have is a) online everything MOVABLE (online_movable) b) online everything
!MOVABLE (online_kernel) c) keep zones contiguous (online). This series
allows configuring c) to mean instead "online movable if possible
according to the coniguration, driven by a maximum MOVABLE:KERNEL ratio"
-- a new onlining policy.
II. Approach
This series does 3 things:
1) Introduces the "auto-movable" online policy that initially operates on
individual memory blocks only. It uses a maximum MOVABLE:KERNEL ratio
to make a decision whether a memory block will be onlined to
ZONE_MOVABLE or not. However, in the basic form, hotplugged KERNEL
memory does not allow for more MOVABLE memory (details in the
patches). CMA memory is treated like MOVABLE memory.
2) Introduces static (e.g., DIMM) and dynamic (e.g., virtio-mem) memory
groups and uses group information to make decisions in the
"auto-movable" online policy across memory blocks of a single memory
device (modeled as memory group). More details can be found in patch
#3 or in the DIMM example below.
3) Maximizes ZONE_MOVABLE memory within dynamic memory groups, by
allowing ZONE_NORMAL memory within a dynamic memory group to allow for
more ZONE_MOVABLE memory within the same memory group. The target use
case is dynamic VM resizing using virtio-mem. See the virtio-mem
example below.
I remember that the basic idea of using a ratio to implement a policy in
the kernel was once mentioned by Vitaly Kuznetsov, but I might be wrong (I
lost the pointer to that discussion).
For me, the main use case is using it along with virtio-mem (and DIMMs /
ppc64 dlpar where necessary) for dynamic resizing of VMs, increasing the
amount of memory we can hotunplug reliably again if we might eventually
hotplug a lot of memory to a VM.
III. Target Usage
The target usage will be:
1) Linux boots with "mhp_default_online_type=offline"
2) User space (e.g., systemd unit) configures memory onlining (according
to a config file and system properties), for example:
* Setting memory_hotplug.online_policy=auto-movable
* Setting memory_hotplug.auto_movable_ratio=301
* Setting memory_hotplug.auto_movable_numa_aware=true
3) User space enabled auto onlining via "echo online >
/sys/devices/system/memory/auto_online_blocks"
4) User space triggers manual onlining of all already-offline memory
blocks (go over offline memory blocks and set them to "online")
IV. Example
For DIMMs, hotplugging 4 GiB DIMMs to a 4 GiB VM with a configured ratio of
301% results in the following layout:
Memory block 0-15: DMA32 (early)
Memory block 32-47: Normal (early)
Memory block 48-79: Movable (DIMM 0)
Memory block 80-111: Movable (DIMM 1)
Memory block 112-143: Movable (DIMM 2)
Memory block 144-275: Normal (DIMM 3)
Memory block 176-207: Normal (DIMM 4)
... all Normal
(-> hotplugged Normal memory does not allow for more Movable memory)
For virtio-mem, using a simple, single virtio-mem device with a 4 GiB VM
will result in the following layout:
Memory block 0-15: DMA32 (early)
Memory block 32-47: Normal (early)
Memory block 48-143: Movable (virtio-mem, first 12 GiB)
Memory block 144: Normal (virtio-mem, next 128 MiB)
Memory block 145-147: Movable (virtio-mem, next 384 MiB)
Memory block 148: Normal (virtio-mem, next 128 MiB)
Memory block 149-151: Movable (virtio-mem, next 384 MiB)
... Normal/Movable mixture as above
(-> hotplugged Normal memory allows for more Movable memory within
the same device)
Which gives us maximum flexibility when dynamically growing/shrinking a
VM in smaller steps.
V. Doc Update
I'll update the memory-hotplug.rst documentation, once the overhaul [1] is
usptream. Until then, details can be found in patch #2.
VI. Future Work
1) Use memory groups for ppc64 dlpar
2) Being able to specify a portion of (early) kernel memory that will be
excluded from the ratio. Like "128 MiB globally/per node" are excluded.
This might be helpful when starting VMs with extremely small memory
footprint (e.g., 128 MiB) and hotplugging memory later -- not wanting
the first hotplugged units getting onlined to ZONE_MOVABLE. One
alternative would be a trigger to not consider ZONE_DMA memory
in the ratio. We'll have to see if this is really rrequired.
3) Indicate to user space that MOVABLE might be a bad idea -- especially
relevant when memory ballooning without support for balloon compaction
is active.
This patch (of 9):
For implementing a new memory onlining policy, which determines when to
online memory blocks to ZONE_MOVABLE semi-automatically, we need the
number of present early (boot) pages -- present pages excluding hotplugged
pages. Let's track these pages per zone.
Pass a page instead of the zone to adjust_present_page_count(), similar as
adjust_managed_page_count() and derive the zone from the page.
It's worth noting that a memory block to be offlined/onlined is either
completely "early" or "not early". add_memory() and friends can only add
complete memory blocks and we only online/offline complete (individual)
memory blocks.
Link: https://lkml.kernel.org/r/20210806124715.17090-1-david@redhat.com
Link: https://lkml.kernel.org/r/20210806124715.17090-2-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Cc: Hui Zhu <teawater@gmail.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Len Brown <lenb@kernel.org>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-09-08 02:55:19 +00:00
|
|
|
* present_early_pages is present pages existing within the zone
|
|
|
|
* located on memory available since early boot, excluding hotplugged
|
|
|
|
* memory.
|
|
|
|
*
|
2012-12-12 21:52:12 +00:00
|
|
|
* managed_pages is present pages managed by the buddy system, which
|
|
|
|
* is calculated as (reserved_pages includes pages allocated by the
|
|
|
|
* bootmem allocator):
|
|
|
|
* managed_pages = present_pages - reserved_pages;
|
|
|
|
*
|
2021-02-26 01:16:40 +00:00
|
|
|
* cma pages is present pages that are assigned for CMA use
|
|
|
|
* (MIGRATE_CMA).
|
|
|
|
*
|
2012-12-12 21:52:12 +00:00
|
|
|
* So present_pages may be used by memory hotplug or memory power
|
|
|
|
* management logic to figure out unmanaged pages by checking
|
|
|
|
* (present_pages - managed_pages). And managed_pages should be used
|
|
|
|
* by page allocator and vm scanner to calculate all kinds of watermarks
|
|
|
|
* and thresholds.
|
|
|
|
*
|
|
|
|
* Locking rules:
|
|
|
|
*
|
|
|
|
* zone_start_pfn and spanned_pages are protected by span_seqlock.
|
|
|
|
* It is a seqlock because it has to be read outside of zone->lock,
|
|
|
|
* and it is done in the main allocator path. But, it is written
|
|
|
|
* quite infrequently.
|
|
|
|
*
|
|
|
|
* The span_seq lock is declared along with zone->lock because it is
|
2005-10-30 01:16:53 +00:00
|
|
|
* frequently read in proximity to zone->lock. It's good to
|
|
|
|
* give them a chance of being in the same cacheline.
|
2012-12-12 21:52:12 +00:00
|
|
|
*
|
2013-07-03 22:03:14 +00:00
|
|
|
* Write access to present_pages at runtime should be protected by
|
mem-hotplug: implement get/put_online_mems
kmem_cache_{create,destroy,shrink} need to get a stable value of
cpu/node online mask, because they init/destroy/access per-cpu/node
kmem_cache parts, which can be allocated or destroyed on cpu/mem
hotplug. To protect against cpu hotplug, these functions use
{get,put}_online_cpus. However, they do nothing to synchronize with
memory hotplug - taking the slab_mutex does not eliminate the
possibility of race as described in patch 2.
What we need there is something like get_online_cpus, but for memory.
We already have lock_memory_hotplug, which serves for the purpose, but
it's a bit of a hammer right now, because it's backed by a mutex. As a
result, it imposes some limitations to locking order, which are not
desirable, and can't be used just like get_online_cpus. That's why in
patch 1 I substitute it with get/put_online_mems, which work exactly
like get/put_online_cpus except they block not cpu, but memory hotplug.
[ v1 can be found at https://lkml.org/lkml/2014/4/6/68. I NAK'ed it by
myself, because it used an rw semaphore for get/put_online_mems,
making them dead lock prune. ]
This patch (of 2):
{un}lock_memory_hotplug, which is used to synchronize against memory
hotplug, is currently backed by a mutex, which makes it a bit of a
hammer - threads that only want to get a stable value of online nodes
mask won't be able to proceed concurrently. Also, it imposes some
strong locking ordering rules on it, which narrows down the set of its
usage scenarios.
This patch introduces get/put_online_mems, which are the same as
get/put_online_cpus, but for memory hotplug, i.e. executing a code
inside a get/put_online_mems section will guarantee a stable value of
online nodes, present pages, etc.
lock_memory_hotplug()/unlock_memory_hotplug() are removed altogether.
Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Tang Chen <tangchen@cn.fujitsu.com>
Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: Jiang Liu <liuj97@gmail.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:07:18 +00:00
|
|
|
* mem_hotplug_begin/end(). Any reader who can't tolerant drift of
|
|
|
|
* present_pages should get_online_mems() to get a stable value.
|
2005-10-30 01:16:53 +00:00
|
|
|
*/
|
2018-12-28 08:34:24 +00:00
|
|
|
atomic_long_t managed_pages;
|
2012-12-12 21:52:12 +00:00
|
|
|
unsigned long spanned_pages;
|
|
|
|
unsigned long present_pages;
|
mm: track present early pages per zone
Patch series "mm/memory_hotplug: "auto-movable" online policy and memory groups", v3.
I. Goal
The goal of this series is improving in-kernel auto-online support. It
tackles the fundamental problems that:
1) We can create zone imbalances when onlining all memory blindly to
ZONE_MOVABLE, in the worst case crashing the system. We have to know
upfront how much memory we are going to hotplug such that we can
safely enable auto-onlining of all hotplugged memory to ZONE_MOVABLE
via "online_movable". This is far from practical and only applicable in
limited setups -- like inside VMs under the RHV/oVirt hypervisor which
will never hotplug more than 3 times the boot memory (and the
limitation is only in place due to the Linux limitation).
2) We see more setups that implement dynamic VM resizing, hot(un)plugging
memory to resize VM memory. In these setups, we might hotplug a lot of
memory, but it might happen in various small steps in both directions
(e.g., 2 GiB -> 8 GiB -> 4 GiB -> 16 GiB ...). virtio-mem is the
primary driver of this upstream right now, performing such dynamic
resizing NUMA-aware via multiple virtio-mem devices.
Onlining all hotplugged memory to ZONE_NORMAL means we basically have
no hotunplug guarantees. Onlining all to ZONE_MOVABLE means we can
easily run into zone imbalances when growing a VM. We want a mixture,
and we want as much memory as reasonable/configured in ZONE_MOVABLE.
Details regarding zone imbalances can be found at [1].
3) Memory devices consist of 1..X memory block devices, however, the
kernel doesn't really track the relationship. Consequently, also user
space has no idea. We want to make per-device decisions.
As one example, for memory hotunplug it doesn't make sense to use a
mixture of zones within a single DIMM: we want all MOVABLE if
possible, otherwise all !MOVABLE, because any !MOVABLE part will easily
block the whole DIMM from getting hotunplugged.
As another example, virtio-mem operates on individual units that span
1..X memory blocks. Similar to a DIMM, we want a unit to either be all
MOVABLE or !MOVABLE. A "unit" can be thought of like a DIMM, however,
all units of a virtio-mem device logically belong together and are
managed (added/removed) by a single driver. We want as much memory of
a virtio-mem device to be MOVABLE as possible.
4) We want memory onlining to be done right from the kernel while adding
memory, not triggered by user space via udev rules; for example, this
is reqired for fast memory hotplug for drivers that add individual
memory blocks, like virito-mem. We want a way to configure a policy in
the kernel and avoid implementing advanced policies in user space.
The auto-onlining support we have in the kernel is not sufficient. All we
have is a) online everything MOVABLE (online_movable) b) online everything
!MOVABLE (online_kernel) c) keep zones contiguous (online). This series
allows configuring c) to mean instead "online movable if possible
according to the coniguration, driven by a maximum MOVABLE:KERNEL ratio"
-- a new onlining policy.
II. Approach
This series does 3 things:
1) Introduces the "auto-movable" online policy that initially operates on
individual memory blocks only. It uses a maximum MOVABLE:KERNEL ratio
to make a decision whether a memory block will be onlined to
ZONE_MOVABLE or not. However, in the basic form, hotplugged KERNEL
memory does not allow for more MOVABLE memory (details in the
patches). CMA memory is treated like MOVABLE memory.
2) Introduces static (e.g., DIMM) and dynamic (e.g., virtio-mem) memory
groups and uses group information to make decisions in the
"auto-movable" online policy across memory blocks of a single memory
device (modeled as memory group). More details can be found in patch
#3 or in the DIMM example below.
3) Maximizes ZONE_MOVABLE memory within dynamic memory groups, by
allowing ZONE_NORMAL memory within a dynamic memory group to allow for
more ZONE_MOVABLE memory within the same memory group. The target use
case is dynamic VM resizing using virtio-mem. See the virtio-mem
example below.
I remember that the basic idea of using a ratio to implement a policy in
the kernel was once mentioned by Vitaly Kuznetsov, but I might be wrong (I
lost the pointer to that discussion).
For me, the main use case is using it along with virtio-mem (and DIMMs /
ppc64 dlpar where necessary) for dynamic resizing of VMs, increasing the
amount of memory we can hotunplug reliably again if we might eventually
hotplug a lot of memory to a VM.
III. Target Usage
The target usage will be:
1) Linux boots with "mhp_default_online_type=offline"
2) User space (e.g., systemd unit) configures memory onlining (according
to a config file and system properties), for example:
* Setting memory_hotplug.online_policy=auto-movable
* Setting memory_hotplug.auto_movable_ratio=301
* Setting memory_hotplug.auto_movable_numa_aware=true
3) User space enabled auto onlining via "echo online >
/sys/devices/system/memory/auto_online_blocks"
4) User space triggers manual onlining of all already-offline memory
blocks (go over offline memory blocks and set them to "online")
IV. Example
For DIMMs, hotplugging 4 GiB DIMMs to a 4 GiB VM with a configured ratio of
301% results in the following layout:
Memory block 0-15: DMA32 (early)
Memory block 32-47: Normal (early)
Memory block 48-79: Movable (DIMM 0)
Memory block 80-111: Movable (DIMM 1)
Memory block 112-143: Movable (DIMM 2)
Memory block 144-275: Normal (DIMM 3)
Memory block 176-207: Normal (DIMM 4)
... all Normal
(-> hotplugged Normal memory does not allow for more Movable memory)
For virtio-mem, using a simple, single virtio-mem device with a 4 GiB VM
will result in the following layout:
Memory block 0-15: DMA32 (early)
Memory block 32-47: Normal (early)
Memory block 48-143: Movable (virtio-mem, first 12 GiB)
Memory block 144: Normal (virtio-mem, next 128 MiB)
Memory block 145-147: Movable (virtio-mem, next 384 MiB)
Memory block 148: Normal (virtio-mem, next 128 MiB)
Memory block 149-151: Movable (virtio-mem, next 384 MiB)
... Normal/Movable mixture as above
(-> hotplugged Normal memory allows for more Movable memory within
the same device)
Which gives us maximum flexibility when dynamically growing/shrinking a
VM in smaller steps.
V. Doc Update
I'll update the memory-hotplug.rst documentation, once the overhaul [1] is
usptream. Until then, details can be found in patch #2.
VI. Future Work
1) Use memory groups for ppc64 dlpar
2) Being able to specify a portion of (early) kernel memory that will be
excluded from the ratio. Like "128 MiB globally/per node" are excluded.
This might be helpful when starting VMs with extremely small memory
footprint (e.g., 128 MiB) and hotplugging memory later -- not wanting
the first hotplugged units getting onlined to ZONE_MOVABLE. One
alternative would be a trigger to not consider ZONE_DMA memory
in the ratio. We'll have to see if this is really rrequired.
3) Indicate to user space that MOVABLE might be a bad idea -- especially
relevant when memory ballooning without support for balloon compaction
is active.
This patch (of 9):
For implementing a new memory onlining policy, which determines when to
online memory blocks to ZONE_MOVABLE semi-automatically, we need the
number of present early (boot) pages -- present pages excluding hotplugged
pages. Let's track these pages per zone.
Pass a page instead of the zone to adjust_present_page_count(), similar as
adjust_managed_page_count() and derive the zone from the page.
It's worth noting that a memory block to be offlined/onlined is either
completely "early" or "not early". add_memory() and friends can only add
complete memory blocks and we only online/offline complete (individual)
memory blocks.
Link: https://lkml.kernel.org/r/20210806124715.17090-1-david@redhat.com
Link: https://lkml.kernel.org/r/20210806124715.17090-2-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Cc: Hui Zhu <teawater@gmail.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Len Brown <lenb@kernel.org>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-09-08 02:55:19 +00:00
|
|
|
#if defined(CONFIG_MEMORY_HOTPLUG)
|
|
|
|
unsigned long present_early_pages;
|
|
|
|
#endif
|
2021-02-26 01:16:40 +00:00
|
|
|
#ifdef CONFIG_CMA
|
|
|
|
unsigned long cma_pages;
|
|
|
|
#endif
|
2014-08-06 23:07:14 +00:00
|
|
|
|
|
|
|
const char *name;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
mm/page_alloc: fix incorrect isolation behavior by rechecking migratetype
Before describing bugs itself, I first explain definition of freepage.
1. pages on buddy list are counted as freepage.
2. pages on isolate migratetype buddy list are *not* counted as freepage.
3. pages on cma buddy list are counted as CMA freepage, too.
Now, I describe problems and related patch.
Patch 1: There is race conditions on getting pageblock migratetype that
it results in misplacement of freepages on buddy list, incorrect
freepage count and un-availability of freepage.
Patch 2: Freepages on pcp list could have stale cached information to
determine migratetype of buddy list to go. This causes misplacement of
freepages on buddy list and incorrect freepage count.
Patch 4: Merging between freepages on different migratetype of
pageblocks will cause freepages accouting problem. This patch fixes it.
Without patchset [3], above problem doesn't happens on my CMA allocation
test, because CMA reserved pages aren't used at all. So there is no
chance for above race.
With patchset [3], I did simple CMA allocation test and get below
result:
- Virtual machine, 4 cpus, 1024 MB memory, 256 MB CMA reservation
- run kernel build (make -j16) on background
- 30 times CMA allocation(8MB * 30 = 240MB) attempts in 5 sec interval
- Result: more than 5000 freepage count are missed
With patchset [3] and this patchset, I found that no freepage count are
missed so that I conclude that problems are solved.
On my simple memory offlining test, these problems also occur on that
environment, too.
This patch (of 4):
There are two paths to reach core free function of buddy allocator,
__free_one_page(), one is free_one_page()->__free_one_page() and the
other is free_hot_cold_page()->free_pcppages_bulk()->__free_one_page().
Each paths has race condition causing serious problems. At first, this
patch is focused on first type of freepath. And then, following patch
will solve the problem in second type of freepath.
In the first type of freepath, we got migratetype of freeing page
without holding the zone lock, so it could be racy. There are two cases
of this race.
1. pages are added to isolate buddy list after restoring orignal
migratetype
CPU1 CPU2
get migratetype => return MIGRATE_ISOLATE
call free_one_page() with MIGRATE_ISOLATE
grab the zone lock
unisolate pageblock
release the zone lock
grab the zone lock
call __free_one_page() with MIGRATE_ISOLATE
freepage go into isolate buddy list,
although pageblock is already unisolated
This may cause two problems. One is that we can't use this page anymore
until next isolation attempt of this pageblock, because freepage is on
isolate buddy list. The other is that freepage accouting could be wrong
due to merging between different buddy list. Freepages on isolate buddy
list aren't counted as freepage, but ones on normal buddy list are
counted as freepage. If merge happens, buddy freepage on normal buddy
list is inevitably moved to isolate buddy list without any consideration
of freepage accouting so it could be incorrect.
2. pages are added to normal buddy list while pageblock is isolated.
It is similar with above case.
This also may cause two problems. One is that we can't keep these
freepages from being allocated. Although this pageblock is isolated,
freepage would be added to normal buddy list so that it could be
allocated without any restriction. And the other problem is same as
case 1, that it, incorrect freepage accouting.
This race condition would be prevented by checking migratetype again
with holding the zone lock. Because it is somewhat heavy operation and
it isn't needed in common case, we want to avoid rechecking as much as
possible. So this patch introduce new variable, nr_isolate_pageblock in
struct zone to check if there is isolated pageblock. With this, we can
avoid to re-check migratetype in common case and do it only if there is
isolated pageblock or migratetype is MIGRATE_ISOLATE. This solve above
mentioned problems.
Changes from v3:
Add one more check in free_one_page() that checks whether migratetype is
MIGRATE_ISOLATE or not. Without this, abovementioned case 1 could happens.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Tang Chen <tangchen@cn.fujitsu.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Laura Abbott <lauraa@codeaurora.org>
Cc: Heesub Shin <heesub.shin@samsung.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Ritesh Harjani <ritesh.list@gmail.com>
Cc: Gioh Kim <gioh.kim@lge.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-11-13 23:19:11 +00:00
|
|
|
#ifdef CONFIG_MEMORY_ISOLATION
|
|
|
|
/*
|
|
|
|
* Number of isolated pageblock. It is used to solve incorrect
|
|
|
|
* freepage counting problem due to racy retrieving migratetype
|
|
|
|
* of pageblock. Protected by zone->lock.
|
|
|
|
*/
|
|
|
|
unsigned long nr_isolate_pageblock;
|
|
|
|
#endif
|
|
|
|
|
2014-08-06 23:07:14 +00:00
|
|
|
#ifdef CONFIG_MEMORY_HOTPLUG
|
|
|
|
/* see spanned/present_pages for more description */
|
|
|
|
seqlock_t span_seqlock;
|
|
|
|
#endif
|
|
|
|
|
mm: remove per-zone hashtable of bitlock waitqueues
The per-zone waitqueues exist because of a scalability issue with the
page waitqueues on some NUMA machines, but it turns out that they hurt
normal loads, and now with the vmalloced stacks they also end up
breaking gfs2 that uses a bit_wait on a stack object:
wait_on_bit(&gh->gh_iflags, HIF_WAIT, TASK_UNINTERRUPTIBLE)
where 'gh' can be a reference to the local variable 'mount_gh' on the
stack of fill_super().
The reason the per-zone hash table breaks for this case is that there is
no "zone" for virtual allocations, and trying to look up the physical
page to get at it will fail (with a BUG_ON()).
It turns out that I actually complained to the mm people about the
per-zone hash table for another reason just a month ago: the zone lookup
also hurts the regular use of "unlock_page()" a lot, because the zone
lookup ends up forcing several unnecessary cache misses and generates
horrible code.
As part of that earlier discussion, we had a much better solution for
the NUMA scalability issue - by just making the page lock have a
separate contention bit, the waitqueue doesn't even have to be looked at
for the normal case.
Peter Zijlstra already has a patch for that, but let's see if anybody
even notices. In the meantime, let's fix the actual gfs2 breakage by
simplifying the bitlock waitqueues and removing the per-zone issue.
Reported-by: Andreas Gruenbacher <agruenba@redhat.com>
Tested-by: Bob Peterson <rpeterso@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-26 17:15:30 +00:00
|
|
|
int initialized;
|
2014-08-06 23:07:14 +00:00
|
|
|
|
2016-07-28 22:45:34 +00:00
|
|
|
/* Write-intensive fields used from the page allocator */
|
2014-08-06 23:07:14 +00:00
|
|
|
ZONE_PADDING(_pad1_)
|
2016-07-28 22:45:34 +00:00
|
|
|
|
2014-08-06 23:07:14 +00:00
|
|
|
/* free areas of different sizes */
|
|
|
|
struct free_area free_area[MAX_ORDER];
|
|
|
|
|
|
|
|
/* zone flags, see below */
|
|
|
|
unsigned long flags;
|
|
|
|
|
2016-07-28 22:45:34 +00:00
|
|
|
/* Primarily protects free_area */
|
2015-04-07 21:26:41 +00:00
|
|
|
spinlock_t lock;
|
|
|
|
|
2016-07-28 22:45:34 +00:00
|
|
|
/* Write-intensive fields used by compaction and vmstats. */
|
2014-08-06 23:07:14 +00:00
|
|
|
ZONE_PADDING(_pad2_)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* When free pages are below this point, additional steps are taken
|
|
|
|
* when reading the number of free pages to avoid per-cpu counter
|
|
|
|
* drift allowing watermarks to be breached
|
|
|
|
*/
|
|
|
|
unsigned long percpu_drift_mark;
|
|
|
|
|
|
|
|
#if defined CONFIG_COMPACTION || defined CONFIG_CMA
|
|
|
|
/* pfn where compaction free scanner should start */
|
|
|
|
unsigned long compact_cached_free_pfn;
|
2020-10-16 03:09:55 +00:00
|
|
|
/* pfn where compaction migration scanner should start */
|
|
|
|
unsigned long compact_cached_migrate_pfn[ASYNC_AND_SYNC];
|
2019-03-05 23:45:38 +00:00
|
|
|
unsigned long compact_init_migrate_pfn;
|
|
|
|
unsigned long compact_init_free_pfn;
|
2014-08-06 23:07:14 +00:00
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef CONFIG_COMPACTION
|
|
|
|
/*
|
|
|
|
* On compaction failure, 1<<compact_defer_shift compactions
|
|
|
|
* are skipped before trying again. The number attempted since
|
|
|
|
* last failure is tracked with compact_considered.
|
2020-08-12 01:31:10 +00:00
|
|
|
* compact_order_failed is the minimum compaction failed order.
|
2014-08-06 23:07:14 +00:00
|
|
|
*/
|
|
|
|
unsigned int compact_considered;
|
|
|
|
unsigned int compact_defer_shift;
|
|
|
|
int compact_order_failed;
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#if defined CONFIG_COMPACTION || defined CONFIG_CMA
|
|
|
|
/* Set to true when the PG_migrate_skip bits should be cleared */
|
|
|
|
bool compact_blockskip_flush;
|
|
|
|
#endif
|
|
|
|
|
2016-03-15 21:57:51 +00:00
|
|
|
bool contiguous;
|
|
|
|
|
2014-08-06 23:07:14 +00:00
|
|
|
ZONE_PADDING(_pad3_)
|
|
|
|
/* Zone statistics */
|
|
|
|
atomic_long_t vm_stat[NR_VM_ZONE_STAT_ITEMS];
|
2021-06-29 02:41:44 +00:00
|
|
|
atomic_long_t vm_numa_event[NR_VM_NUMA_EVENT_ITEMS];
|
2006-01-08 09:01:27 +00:00
|
|
|
} ____cacheline_internodealigned_in_smp;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2016-07-28 22:45:31 +00:00
|
|
|
enum pgdat_flags {
|
|
|
|
PGDAT_DIRTY, /* reclaim scanning has recently found
|
2013-07-03 22:01:50 +00:00
|
|
|
* many dirty file pages at the tail
|
|
|
|
* of the LRU.
|
|
|
|
*/
|
2016-07-28 22:45:31 +00:00
|
|
|
PGDAT_WRITEBACK, /* reclaim scanning has recently found
|
2013-07-03 22:01:51 +00:00
|
|
|
* many pages under writeback
|
|
|
|
*/
|
2016-07-28 22:46:32 +00:00
|
|
|
PGDAT_RECLAIM_LOCKED, /* prevents concurrent reclaim */
|
2014-10-09 22:28:17 +00:00
|
|
|
};
|
2007-10-17 06:25:54 +00:00
|
|
|
|
mm, page_alloc: do not wake kswapd with zone lock held
syzbot reported the following regression in the latest merge window and
it was confirmed by Qian Cai that a similar bug was visible from a
different context.
======================================================
WARNING: possible circular locking dependency detected
4.20.0+ #297 Not tainted
------------------------------------------------------
syz-executor0/8529 is trying to acquire lock:
000000005e7fb829 (&pgdat->kswapd_wait){....}, at:
__wake_up_common_lock+0x19e/0x330 kernel/sched/wait.c:120
but task is already holding lock:
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: spin_lock
include/linux/spinlock.h:329 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_bulk
mm/page_alloc.c:2548 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: __rmqueue_pcplist
mm/page_alloc.c:3021 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_pcplist
mm/page_alloc.c:3050 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue
mm/page_alloc.c:3072 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at:
get_page_from_freelist+0x1bae/0x52a0 mm/page_alloc.c:3491
It appears to be a false positive in that the only way the lock ordering
should be inverted is if kswapd is waking itself and the wakeup
allocates debugging objects which should already be allocated if it's
kswapd doing the waking. Nevertheless, the possibility exists and so
it's best to avoid the problem.
This patch flags a zone as needing a kswapd using the, surprisingly,
unused zone flag field. The flag is read without the lock held to do
the wakeup. It's possible that the flag setting context is not the same
as the flag clearing context or for small races to occur. However, each
race possibility is harmless and there is no visible degredation in
fragmentation treatment.
While zone->flag could have continued to be unused, there is potential
for moving some existing fields into the flags field instead.
Particularly read-mostly ones like zone->initialized and
zone->contiguous.
Link: http://lkml.kernel.org/r/20190103225712.GJ31517@techsingularity.net
Fixes: 1c30844d2dfe ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Reported-by: syzbot+93d94a001cfbce9e60e1@syzkaller.appspotmail.com
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Tested-by: Qian Cai <cai@lca.pw>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-01-08 23:23:39 +00:00
|
|
|
enum zone_flags {
|
|
|
|
ZONE_BOOSTED_WATERMARK, /* zone recently boosted watermarks.
|
|
|
|
* Cleared when kswapd is woken.
|
|
|
|
*/
|
2021-06-29 02:42:21 +00:00
|
|
|
ZONE_RECLAIM_ACTIVE, /* kswapd may be scanning the zone. */
|
mm, page_alloc: do not wake kswapd with zone lock held
syzbot reported the following regression in the latest merge window and
it was confirmed by Qian Cai that a similar bug was visible from a
different context.
======================================================
WARNING: possible circular locking dependency detected
4.20.0+ #297 Not tainted
------------------------------------------------------
syz-executor0/8529 is trying to acquire lock:
000000005e7fb829 (&pgdat->kswapd_wait){....}, at:
__wake_up_common_lock+0x19e/0x330 kernel/sched/wait.c:120
but task is already holding lock:
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: spin_lock
include/linux/spinlock.h:329 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_bulk
mm/page_alloc.c:2548 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: __rmqueue_pcplist
mm/page_alloc.c:3021 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_pcplist
mm/page_alloc.c:3050 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue
mm/page_alloc.c:3072 [inline]
000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at:
get_page_from_freelist+0x1bae/0x52a0 mm/page_alloc.c:3491
It appears to be a false positive in that the only way the lock ordering
should be inverted is if kswapd is waking itself and the wakeup
allocates debugging objects which should already be allocated if it's
kswapd doing the waking. Nevertheless, the possibility exists and so
it's best to avoid the problem.
This patch flags a zone as needing a kswapd using the, surprisingly,
unused zone flag field. The flag is read without the lock held to do
the wakeup. It's possible that the flag setting context is not the same
as the flag clearing context or for small races to occur. However, each
race possibility is harmless and there is no visible degredation in
fragmentation treatment.
While zone->flag could have continued to be unused, there is potential
for moving some existing fields into the flags field instead.
Particularly read-mostly ones like zone->initialized and
zone->contiguous.
Link: http://lkml.kernel.org/r/20190103225712.GJ31517@techsingularity.net
Fixes: 1c30844d2dfe ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Reported-by: syzbot+93d94a001cfbce9e60e1@syzkaller.appspotmail.com
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Tested-by: Qian Cai <cai@lca.pw>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-01-08 23:23:39 +00:00
|
|
|
};
|
|
|
|
|
2018-12-28 08:34:24 +00:00
|
|
|
static inline unsigned long zone_managed_pages(struct zone *zone)
|
|
|
|
{
|
|
|
|
return (unsigned long)atomic_long_read(&zone->managed_pages);
|
|
|
|
}
|
|
|
|
|
2021-02-26 01:16:40 +00:00
|
|
|
static inline unsigned long zone_cma_pages(struct zone *zone)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_CMA
|
|
|
|
return zone->cma_pages;
|
|
|
|
#else
|
|
|
|
return 0;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2013-03-22 22:04:43 +00:00
|
|
|
static inline unsigned long zone_end_pfn(const struct zone *zone)
|
2013-02-23 00:35:23 +00:00
|
|
|
{
|
|
|
|
return zone->zone_start_pfn + zone->spanned_pages;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool zone_spans_pfn(const struct zone *zone, unsigned long pfn)
|
|
|
|
{
|
|
|
|
return zone->zone_start_pfn <= pfn && pfn < zone_end_pfn(zone);
|
|
|
|
}
|
|
|
|
|
2013-02-23 00:35:24 +00:00
|
|
|
static inline bool zone_is_initialized(struct zone *zone)
|
|
|
|
{
|
mm: remove per-zone hashtable of bitlock waitqueues
The per-zone waitqueues exist because of a scalability issue with the
page waitqueues on some NUMA machines, but it turns out that they hurt
normal loads, and now with the vmalloced stacks they also end up
breaking gfs2 that uses a bit_wait on a stack object:
wait_on_bit(&gh->gh_iflags, HIF_WAIT, TASK_UNINTERRUPTIBLE)
where 'gh' can be a reference to the local variable 'mount_gh' on the
stack of fill_super().
The reason the per-zone hash table breaks for this case is that there is
no "zone" for virtual allocations, and trying to look up the physical
page to get at it will fail (with a BUG_ON()).
It turns out that I actually complained to the mm people about the
per-zone hash table for another reason just a month ago: the zone lookup
also hurts the regular use of "unlock_page()" a lot, because the zone
lookup ends up forcing several unnecessary cache misses and generates
horrible code.
As part of that earlier discussion, we had a much better solution for
the NUMA scalability issue - by just making the page lock have a
separate contention bit, the waitqueue doesn't even have to be looked at
for the normal case.
Peter Zijlstra already has a patch for that, but let's see if anybody
even notices. In the meantime, let's fix the actual gfs2 breakage by
simplifying the bitlock waitqueues and removing the per-zone issue.
Reported-by: Andreas Gruenbacher <agruenba@redhat.com>
Tested-by: Bob Peterson <rpeterso@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-26 17:15:30 +00:00
|
|
|
return zone->initialized;
|
2013-02-23 00:35:24 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool zone_is_empty(struct zone *zone)
|
|
|
|
{
|
|
|
|
return zone->spanned_pages == 0;
|
|
|
|
}
|
|
|
|
|
2017-07-06 22:38:11 +00:00
|
|
|
/*
|
|
|
|
* Return true if [start_pfn, start_pfn + nr_pages) range has a non-empty
|
|
|
|
* intersection with the given zone
|
|
|
|
*/
|
|
|
|
static inline bool zone_intersects(struct zone *zone,
|
|
|
|
unsigned long start_pfn, unsigned long nr_pages)
|
|
|
|
{
|
|
|
|
if (zone_is_empty(zone))
|
|
|
|
return false;
|
|
|
|
if (start_pfn >= zone_end_pfn(zone) ||
|
|
|
|
start_pfn + nr_pages <= zone->zone_start_pfn)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* The "priority" of VM scanning is how much of the queues we will scan in one
|
|
|
|
* go. A value of 12 for DEF_PRIORITY implies that we will scan 1/4096th of the
|
|
|
|
* queues ("queue_length >> 12") during an aging round.
|
|
|
|
*/
|
|
|
|
#define DEF_PRIORITY 12
|
|
|
|
|
[PATCH] memory page_alloc zonelist caching speedup
Optimize the critical zonelist scanning for free pages in the kernel memory
allocator by caching the zones that were found to be full recently, and
skipping them.
Remembers the zones in a zonelist that were short of free memory in the
last second. And it stashes a zone-to-node table in the zonelist struct,
to optimize that conversion (minimize its cache footprint.)
Recent changes:
This differs in a significant way from a similar patch that I
posted a week ago. Now, instead of having a nodemask_t of
recently full nodes, I have a bitmask of recently full zones.
This solves a problem that last weeks patch had, which on
systems with multiple zones per node (such as DMA zone) would
take seeing any of these zones full as meaning that all zones
on that node were full.
Also I changed names - from "zonelist faster" to "zonelist cache",
as that seemed to better convey what we're doing here - caching
some of the key zonelist state (for faster access.)
See below for some performance benchmark results. After all that
discussion with David on why I didn't need them, I went and got
some ;). I wanted to verify that I had not hurt the normal case
of memory allocation noticeably. At least for my one little
microbenchmark, I found (1) the normal case wasn't affected, and
(2) workloads that forced scanning across multiple nodes for
memory improved up to 10% fewer System CPU cycles and lower
elapsed clock time ('sys' and 'real'). Good. See details, below.
I didn't have the logic in get_page_from_freelist() for various
full nodes and zone reclaim failures correct. That should be
fixed up now - notice the new goto labels zonelist_scan,
this_zone_full, and try_next_zone, in get_page_from_freelist().
There are two reasons I persued this alternative, over some earlier
proposals that would have focused on optimizing the fake numa
emulation case by caching the last useful zone:
1) Contrary to what I said before, we (SGI, on large ia64 sn2 systems)
have seen real customer loads where the cost to scan the zonelist
was a problem, due to many nodes being full of memory before
we got to a node we could use. Or at least, I think we have.
This was related to me by another engineer, based on experiences
from some time past. So this is not guaranteed. Most likely, though.
The following approach should help such real numa systems just as
much as it helps fake numa systems, or any combination thereof.
2) The effort to distinguish fake from real numa, using node_distance,
so that we could cache a fake numa node and optimize choosing
it over equivalent distance fake nodes, while continuing to
properly scan all real nodes in distance order, was going to
require a nasty blob of zonelist and node distance munging.
The following approach has no new dependency on node distances or
zone sorting.
See comment in the patch below for a description of what it actually does.
Technical details of note (or controversy):
- See the use of "zlc_active" and "did_zlc_setup" below, to delay
adding any work for this new mechanism until we've looked at the
first zone in zonelist. I figured the odds of the first zone
having the memory we needed were high enough that we should just
look there, first, then get fancy only if we need to keep looking.
- Some odd hackery was needed to add items to struct zonelist, while
not tripping up the custom zonelists built by the mm/mempolicy.c
code for MPOL_BIND. My usual wordy comments below explain this.
Search for "MPOL_BIND".
- Some per-node data in the struct zonelist is now modified frequently,
with no locking. Multiple CPU cores on a node could hit and mangle
this data. The theory is that this is just performance hint data,
and the memory allocator will work just fine despite any such mangling.
The fields at risk are the struct 'zonelist_cache' fields 'fullzones'
(a bitmask) and 'last_full_zap' (unsigned long jiffies). It should
all be self correcting after at most a one second delay.
- This still does a linear scan of the same lengths as before. All
I've optimized is making the scan faster, not algorithmically
shorter. It is now able to scan a compact array of 'unsigned
short' in the case of many full nodes, so one cache line should
cover quite a few nodes, rather than each node hitting another
one or two new and distinct cache lines.
- If both Andi and Nick don't find this too complicated, I will be
(pleasantly) flabbergasted.
- I removed the comment claiming we only use one cachline's worth of
zonelist. We seem, at least in the fake numa case, to have put the
lie to that claim.
- I pay no attention to the various watermarks and such in this performance
hint. A node could be marked full for one watermark, and then skipped
over when searching for a page using a different watermark. I think
that's actually quite ok, as it will tend to slightly increase the
spreading of memory over other nodes, away from a memory stressed node.
===============
Performance - some benchmark results and analysis:
This benchmark runs a memory hog program that uses multiple
threads to touch alot of memory as quickly as it can.
Multiple runs were made, touching 12, 38, 64 or 90 GBytes out of
the total 96 GBytes on the system, and using 1, 19, 37, or 55
threads (on a 56 CPU system.) System, user and real (elapsed)
timings were recorded for each run, shown in units of seconds,
in the table below.
Two kernels were tested - 2.6.18-mm3 and the same kernel with
this zonelist caching patch added. The table also shows the
percentage improvement the zonelist caching sys time is over
(lower than) the stock *-mm kernel.
number 2.6.18-mm3 zonelist-cache delta (< 0 good) percent
GBs N ------------ -------------- ---------------- systime
mem threads sys user real sys user real sys user real better
12 1 153 24 177 151 24 176 -2 0 -1 1%
12 19 99 22 8 99 22 8 0 0 0 0%
12 37 111 25 6 112 25 6 1 0 0 -0%
12 55 115 25 5 110 23 5 -5 -2 0 4%
38 1 502 74 576 497 73 570 -5 -1 -6 0%
38 19 426 78 48 373 76 39 -53 -2 -9 12%
38 37 544 83 36 547 82 36 3 -1 0 -0%
38 55 501 77 23 511 80 24 10 3 1 -1%
64 1 917 125 1042 890 124 1014 -27 -1 -28 2%
64 19 1118 138 119 965 141 103 -153 3 -16 13%
64 37 1202 151 94 1136 150 81 -66 -1 -13 5%
64 55 1118 141 61 1072 140 58 -46 -1 -3 4%
90 1 1342 177 1519 1275 174 1450 -67 -3 -69 4%
90 19 2392 199 192 2116 189 176 -276 -10 -16 11%
90 37 3313 238 175 2972 225 145 -341 -13 -30 10%
90 55 1948 210 104 1843 213 100 -105 3 -4 5%
Notes:
1) This test ran a memory hog program that started a specified number N of
threads, and had each thread allocate and touch 1/N'th of
the total memory to be used in the test run in a single loop,
writing a constant word to memory, one store every 4096 bytes.
Watching this test during some earlier trial runs, I would see
each of these threads sit down on one CPU and stay there, for
the remainder of the pass, a different CPU for each thread.
2) The 'real' column is not comparable to the 'sys' or 'user' columns.
The 'real' column is seconds wall clock time elapsed, from beginning
to end of that test pass. The 'sys' and 'user' columns are total
CPU seconds spent on that test pass. For a 19 thread test run,
for example, the sum of 'sys' and 'user' could be up to 19 times the
number of 'real' elapsed wall clock seconds.
3) Tests were run on a fresh, single-user boot, to minimize the amount
of memory already in use at the start of the test, and to minimize
the amount of background activity that might interfere.
4) Tests were done on a 56 CPU, 28 Node system with 96 GBytes of RAM.
5) Notice that the 'real' time gets large for the single thread runs, even
though the measured 'sys' and 'user' times are modest. I'm not sure what
that means - probably something to do with it being slow for one thread to
be accessing memory along ways away. Perhaps the fake numa system, running
ostensibly the same workload, would not show this substantial degradation
of 'real' time for one thread on many nodes -- lets hope not.
6) The high thread count passes (one thread per CPU - on 55 of 56 CPUs)
ran quite efficiently, as one might expect. Each pair of threads needed
to allocate and touch the memory on the node the two threads shared, a
pleasantly parallizable workload.
7) The intermediate thread count passes, when asking for alot of memory forcing
them to go to a few neighboring nodes, improved the most with this zonelist
caching patch.
Conclusions:
* This zonelist cache patch probably makes little difference one way or the
other for most workloads on real numa hardware, if those workloads avoid
heavy off node allocations.
* For memory intensive workloads requiring substantial off-node allocations
on real numa hardware, this patch improves both kernel and elapsed timings
up to ten per-cent.
* For fake numa systems, I'm optimistic, but will have to leave that up to
Rohit Seth to actually test (once I get him a 2.6.18 backport.)
Signed-off-by: Paul Jackson <pj@sgi.com>
Cc: Rohit Seth <rohitseth@google.com>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Cc: David Rientjes <rientjes@cs.washington.edu>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 04:31:48 +00:00
|
|
|
/* Maximum number of zones on a zonelist */
|
|
|
|
#define MAX_ZONES_PER_ZONELIST (MAX_NUMNODES * MAX_NR_ZONES)
|
|
|
|
|
2016-01-14 23:19:00 +00:00
|
|
|
enum {
|
|
|
|
ZONELIST_FALLBACK, /* zonelist with fallback */
|
[PATCH] memory page_alloc zonelist caching speedup
Optimize the critical zonelist scanning for free pages in the kernel memory
allocator by caching the zones that were found to be full recently, and
skipping them.
Remembers the zones in a zonelist that were short of free memory in the
last second. And it stashes a zone-to-node table in the zonelist struct,
to optimize that conversion (minimize its cache footprint.)
Recent changes:
This differs in a significant way from a similar patch that I
posted a week ago. Now, instead of having a nodemask_t of
recently full nodes, I have a bitmask of recently full zones.
This solves a problem that last weeks patch had, which on
systems with multiple zones per node (such as DMA zone) would
take seeing any of these zones full as meaning that all zones
on that node were full.
Also I changed names - from "zonelist faster" to "zonelist cache",
as that seemed to better convey what we're doing here - caching
some of the key zonelist state (for faster access.)
See below for some performance benchmark results. After all that
discussion with David on why I didn't need them, I went and got
some ;). I wanted to verify that I had not hurt the normal case
of memory allocation noticeably. At least for my one little
microbenchmark, I found (1) the normal case wasn't affected, and
(2) workloads that forced scanning across multiple nodes for
memory improved up to 10% fewer System CPU cycles and lower
elapsed clock time ('sys' and 'real'). Good. See details, below.
I didn't have the logic in get_page_from_freelist() for various
full nodes and zone reclaim failures correct. That should be
fixed up now - notice the new goto labels zonelist_scan,
this_zone_full, and try_next_zone, in get_page_from_freelist().
There are two reasons I persued this alternative, over some earlier
proposals that would have focused on optimizing the fake numa
emulation case by caching the last useful zone:
1) Contrary to what I said before, we (SGI, on large ia64 sn2 systems)
have seen real customer loads where the cost to scan the zonelist
was a problem, due to many nodes being full of memory before
we got to a node we could use. Or at least, I think we have.
This was related to me by another engineer, based on experiences
from some time past. So this is not guaranteed. Most likely, though.
The following approach should help such real numa systems just as
much as it helps fake numa systems, or any combination thereof.
2) The effort to distinguish fake from real numa, using node_distance,
so that we could cache a fake numa node and optimize choosing
it over equivalent distance fake nodes, while continuing to
properly scan all real nodes in distance order, was going to
require a nasty blob of zonelist and node distance munging.
The following approach has no new dependency on node distances or
zone sorting.
See comment in the patch below for a description of what it actually does.
Technical details of note (or controversy):
- See the use of "zlc_active" and "did_zlc_setup" below, to delay
adding any work for this new mechanism until we've looked at the
first zone in zonelist. I figured the odds of the first zone
having the memory we needed were high enough that we should just
look there, first, then get fancy only if we need to keep looking.
- Some odd hackery was needed to add items to struct zonelist, while
not tripping up the custom zonelists built by the mm/mempolicy.c
code for MPOL_BIND. My usual wordy comments below explain this.
Search for "MPOL_BIND".
- Some per-node data in the struct zonelist is now modified frequently,
with no locking. Multiple CPU cores on a node could hit and mangle
this data. The theory is that this is just performance hint data,
and the memory allocator will work just fine despite any such mangling.
The fields at risk are the struct 'zonelist_cache' fields 'fullzones'
(a bitmask) and 'last_full_zap' (unsigned long jiffies). It should
all be self correcting after at most a one second delay.
- This still does a linear scan of the same lengths as before. All
I've optimized is making the scan faster, not algorithmically
shorter. It is now able to scan a compact array of 'unsigned
short' in the case of many full nodes, so one cache line should
cover quite a few nodes, rather than each node hitting another
one or two new and distinct cache lines.
- If both Andi and Nick don't find this too complicated, I will be
(pleasantly) flabbergasted.
- I removed the comment claiming we only use one cachline's worth of
zonelist. We seem, at least in the fake numa case, to have put the
lie to that claim.
- I pay no attention to the various watermarks and such in this performance
hint. A node could be marked full for one watermark, and then skipped
over when searching for a page using a different watermark. I think
that's actually quite ok, as it will tend to slightly increase the
spreading of memory over other nodes, away from a memory stressed node.
===============
Performance - some benchmark results and analysis:
This benchmark runs a memory hog program that uses multiple
threads to touch alot of memory as quickly as it can.
Multiple runs were made, touching 12, 38, 64 or 90 GBytes out of
the total 96 GBytes on the system, and using 1, 19, 37, or 55
threads (on a 56 CPU system.) System, user and real (elapsed)
timings were recorded for each run, shown in units of seconds,
in the table below.
Two kernels were tested - 2.6.18-mm3 and the same kernel with
this zonelist caching patch added. The table also shows the
percentage improvement the zonelist caching sys time is over
(lower than) the stock *-mm kernel.
number 2.6.18-mm3 zonelist-cache delta (< 0 good) percent
GBs N ------------ -------------- ---------------- systime
mem threads sys user real sys user real sys user real better
12 1 153 24 177 151 24 176 -2 0 -1 1%
12 19 99 22 8 99 22 8 0 0 0 0%
12 37 111 25 6 112 25 6 1 0 0 -0%
12 55 115 25 5 110 23 5 -5 -2 0 4%
38 1 502 74 576 497 73 570 -5 -1 -6 0%
38 19 426 78 48 373 76 39 -53 -2 -9 12%
38 37 544 83 36 547 82 36 3 -1 0 -0%
38 55 501 77 23 511 80 24 10 3 1 -1%
64 1 917 125 1042 890 124 1014 -27 -1 -28 2%
64 19 1118 138 119 965 141 103 -153 3 -16 13%
64 37 1202 151 94 1136 150 81 -66 -1 -13 5%
64 55 1118 141 61 1072 140 58 -46 -1 -3 4%
90 1 1342 177 1519 1275 174 1450 -67 -3 -69 4%
90 19 2392 199 192 2116 189 176 -276 -10 -16 11%
90 37 3313 238 175 2972 225 145 -341 -13 -30 10%
90 55 1948 210 104 1843 213 100 -105 3 -4 5%
Notes:
1) This test ran a memory hog program that started a specified number N of
threads, and had each thread allocate and touch 1/N'th of
the total memory to be used in the test run in a single loop,
writing a constant word to memory, one store every 4096 bytes.
Watching this test during some earlier trial runs, I would see
each of these threads sit down on one CPU and stay there, for
the remainder of the pass, a different CPU for each thread.
2) The 'real' column is not comparable to the 'sys' or 'user' columns.
The 'real' column is seconds wall clock time elapsed, from beginning
to end of that test pass. The 'sys' and 'user' columns are total
CPU seconds spent on that test pass. For a 19 thread test run,
for example, the sum of 'sys' and 'user' could be up to 19 times the
number of 'real' elapsed wall clock seconds.
3) Tests were run on a fresh, single-user boot, to minimize the amount
of memory already in use at the start of the test, and to minimize
the amount of background activity that might interfere.
4) Tests were done on a 56 CPU, 28 Node system with 96 GBytes of RAM.
5) Notice that the 'real' time gets large for the single thread runs, even
though the measured 'sys' and 'user' times are modest. I'm not sure what
that means - probably something to do with it being slow for one thread to
be accessing memory along ways away. Perhaps the fake numa system, running
ostensibly the same workload, would not show this substantial degradation
of 'real' time for one thread on many nodes -- lets hope not.
6) The high thread count passes (one thread per CPU - on 55 of 56 CPUs)
ran quite efficiently, as one might expect. Each pair of threads needed
to allocate and touch the memory on the node the two threads shared, a
pleasantly parallizable workload.
7) The intermediate thread count passes, when asking for alot of memory forcing
them to go to a few neighboring nodes, improved the most with this zonelist
caching patch.
Conclusions:
* This zonelist cache patch probably makes little difference one way or the
other for most workloads on real numa hardware, if those workloads avoid
heavy off node allocations.
* For memory intensive workloads requiring substantial off-node allocations
on real numa hardware, this patch improves both kernel and elapsed timings
up to ten per-cent.
* For fake numa systems, I'm optimistic, but will have to leave that up to
Rohit Seth to actually test (once I get him a 2.6.18 backport.)
Signed-off-by: Paul Jackson <pj@sgi.com>
Cc: Rohit Seth <rohitseth@google.com>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Cc: David Rientjes <rientjes@cs.washington.edu>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 04:31:48 +00:00
|
|
|
#ifdef CONFIG_NUMA
|
2016-01-14 23:19:00 +00:00
|
|
|
/*
|
|
|
|
* The NUMA zonelists are doubled because we need zonelists that
|
|
|
|
* restrict the allocations to a single node for __GFP_THISNODE.
|
|
|
|
*/
|
|
|
|
ZONELIST_NOFALLBACK, /* zonelist without fallback (__GFP_THISNODE) */
|
[PATCH] memory page_alloc zonelist caching speedup
Optimize the critical zonelist scanning for free pages in the kernel memory
allocator by caching the zones that were found to be full recently, and
skipping them.
Remembers the zones in a zonelist that were short of free memory in the
last second. And it stashes a zone-to-node table in the zonelist struct,
to optimize that conversion (minimize its cache footprint.)
Recent changes:
This differs in a significant way from a similar patch that I
posted a week ago. Now, instead of having a nodemask_t of
recently full nodes, I have a bitmask of recently full zones.
This solves a problem that last weeks patch had, which on
systems with multiple zones per node (such as DMA zone) would
take seeing any of these zones full as meaning that all zones
on that node were full.
Also I changed names - from "zonelist faster" to "zonelist cache",
as that seemed to better convey what we're doing here - caching
some of the key zonelist state (for faster access.)
See below for some performance benchmark results. After all that
discussion with David on why I didn't need them, I went and got
some ;). I wanted to verify that I had not hurt the normal case
of memory allocation noticeably. At least for my one little
microbenchmark, I found (1) the normal case wasn't affected, and
(2) workloads that forced scanning across multiple nodes for
memory improved up to 10% fewer System CPU cycles and lower
elapsed clock time ('sys' and 'real'). Good. See details, below.
I didn't have the logic in get_page_from_freelist() for various
full nodes and zone reclaim failures correct. That should be
fixed up now - notice the new goto labels zonelist_scan,
this_zone_full, and try_next_zone, in get_page_from_freelist().
There are two reasons I persued this alternative, over some earlier
proposals that would have focused on optimizing the fake numa
emulation case by caching the last useful zone:
1) Contrary to what I said before, we (SGI, on large ia64 sn2 systems)
have seen real customer loads where the cost to scan the zonelist
was a problem, due to many nodes being full of memory before
we got to a node we could use. Or at least, I think we have.
This was related to me by another engineer, based on experiences
from some time past. So this is not guaranteed. Most likely, though.
The following approach should help such real numa systems just as
much as it helps fake numa systems, or any combination thereof.
2) The effort to distinguish fake from real numa, using node_distance,
so that we could cache a fake numa node and optimize choosing
it over equivalent distance fake nodes, while continuing to
properly scan all real nodes in distance order, was going to
require a nasty blob of zonelist and node distance munging.
The following approach has no new dependency on node distances or
zone sorting.
See comment in the patch below for a description of what it actually does.
Technical details of note (or controversy):
- See the use of "zlc_active" and "did_zlc_setup" below, to delay
adding any work for this new mechanism until we've looked at the
first zone in zonelist. I figured the odds of the first zone
having the memory we needed were high enough that we should just
look there, first, then get fancy only if we need to keep looking.
- Some odd hackery was needed to add items to struct zonelist, while
not tripping up the custom zonelists built by the mm/mempolicy.c
code for MPOL_BIND. My usual wordy comments below explain this.
Search for "MPOL_BIND".
- Some per-node data in the struct zonelist is now modified frequently,
with no locking. Multiple CPU cores on a node could hit and mangle
this data. The theory is that this is just performance hint data,
and the memory allocator will work just fine despite any such mangling.
The fields at risk are the struct 'zonelist_cache' fields 'fullzones'
(a bitmask) and 'last_full_zap' (unsigned long jiffies). It should
all be self correcting after at most a one second delay.
- This still does a linear scan of the same lengths as before. All
I've optimized is making the scan faster, not algorithmically
shorter. It is now able to scan a compact array of 'unsigned
short' in the case of many full nodes, so one cache line should
cover quite a few nodes, rather than each node hitting another
one or two new and distinct cache lines.
- If both Andi and Nick don't find this too complicated, I will be
(pleasantly) flabbergasted.
- I removed the comment claiming we only use one cachline's worth of
zonelist. We seem, at least in the fake numa case, to have put the
lie to that claim.
- I pay no attention to the various watermarks and such in this performance
hint. A node could be marked full for one watermark, and then skipped
over when searching for a page using a different watermark. I think
that's actually quite ok, as it will tend to slightly increase the
spreading of memory over other nodes, away from a memory stressed node.
===============
Performance - some benchmark results and analysis:
This benchmark runs a memory hog program that uses multiple
threads to touch alot of memory as quickly as it can.
Multiple runs were made, touching 12, 38, 64 or 90 GBytes out of
the total 96 GBytes on the system, and using 1, 19, 37, or 55
threads (on a 56 CPU system.) System, user and real (elapsed)
timings were recorded for each run, shown in units of seconds,
in the table below.
Two kernels were tested - 2.6.18-mm3 and the same kernel with
this zonelist caching patch added. The table also shows the
percentage improvement the zonelist caching sys time is over
(lower than) the stock *-mm kernel.
number 2.6.18-mm3 zonelist-cache delta (< 0 good) percent
GBs N ------------ -------------- ---------------- systime
mem threads sys user real sys user real sys user real better
12 1 153 24 177 151 24 176 -2 0 -1 1%
12 19 99 22 8 99 22 8 0 0 0 0%
12 37 111 25 6 112 25 6 1 0 0 -0%
12 55 115 25 5 110 23 5 -5 -2 0 4%
38 1 502 74 576 497 73 570 -5 -1 -6 0%
38 19 426 78 48 373 76 39 -53 -2 -9 12%
38 37 544 83 36 547 82 36 3 -1 0 -0%
38 55 501 77 23 511 80 24 10 3 1 -1%
64 1 917 125 1042 890 124 1014 -27 -1 -28 2%
64 19 1118 138 119 965 141 103 -153 3 -16 13%
64 37 1202 151 94 1136 150 81 -66 -1 -13 5%
64 55 1118 141 61 1072 140 58 -46 -1 -3 4%
90 1 1342 177 1519 1275 174 1450 -67 -3 -69 4%
90 19 2392 199 192 2116 189 176 -276 -10 -16 11%
90 37 3313 238 175 2972 225 145 -341 -13 -30 10%
90 55 1948 210 104 1843 213 100 -105 3 -4 5%
Notes:
1) This test ran a memory hog program that started a specified number N of
threads, and had each thread allocate and touch 1/N'th of
the total memory to be used in the test run in a single loop,
writing a constant word to memory, one store every 4096 bytes.
Watching this test during some earlier trial runs, I would see
each of these threads sit down on one CPU and stay there, for
the remainder of the pass, a different CPU for each thread.
2) The 'real' column is not comparable to the 'sys' or 'user' columns.
The 'real' column is seconds wall clock time elapsed, from beginning
to end of that test pass. The 'sys' and 'user' columns are total
CPU seconds spent on that test pass. For a 19 thread test run,
for example, the sum of 'sys' and 'user' could be up to 19 times the
number of 'real' elapsed wall clock seconds.
3) Tests were run on a fresh, single-user boot, to minimize the amount
of memory already in use at the start of the test, and to minimize
the amount of background activity that might interfere.
4) Tests were done on a 56 CPU, 28 Node system with 96 GBytes of RAM.
5) Notice that the 'real' time gets large for the single thread runs, even
though the measured 'sys' and 'user' times are modest. I'm not sure what
that means - probably something to do with it being slow for one thread to
be accessing memory along ways away. Perhaps the fake numa system, running
ostensibly the same workload, would not show this substantial degradation
of 'real' time for one thread on many nodes -- lets hope not.
6) The high thread count passes (one thread per CPU - on 55 of 56 CPUs)
ran quite efficiently, as one might expect. Each pair of threads needed
to allocate and touch the memory on the node the two threads shared, a
pleasantly parallizable workload.
7) The intermediate thread count passes, when asking for alot of memory forcing
them to go to a few neighboring nodes, improved the most with this zonelist
caching patch.
Conclusions:
* This zonelist cache patch probably makes little difference one way or the
other for most workloads on real numa hardware, if those workloads avoid
heavy off node allocations.
* For memory intensive workloads requiring substantial off-node allocations
on real numa hardware, this patch improves both kernel and elapsed timings
up to ten per-cent.
* For fake numa systems, I'm optimistic, but will have to leave that up to
Rohit Seth to actually test (once I get him a 2.6.18 backport.)
Signed-off-by: Paul Jackson <pj@sgi.com>
Cc: Rohit Seth <rohitseth@google.com>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Cc: David Rientjes <rientjes@cs.washington.edu>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 04:31:48 +00:00
|
|
|
#endif
|
2016-01-14 23:19:00 +00:00
|
|
|
MAX_ZONELISTS
|
|
|
|
};
|
[PATCH] memory page_alloc zonelist caching speedup
Optimize the critical zonelist scanning for free pages in the kernel memory
allocator by caching the zones that were found to be full recently, and
skipping them.
Remembers the zones in a zonelist that were short of free memory in the
last second. And it stashes a zone-to-node table in the zonelist struct,
to optimize that conversion (minimize its cache footprint.)
Recent changes:
This differs in a significant way from a similar patch that I
posted a week ago. Now, instead of having a nodemask_t of
recently full nodes, I have a bitmask of recently full zones.
This solves a problem that last weeks patch had, which on
systems with multiple zones per node (such as DMA zone) would
take seeing any of these zones full as meaning that all zones
on that node were full.
Also I changed names - from "zonelist faster" to "zonelist cache",
as that seemed to better convey what we're doing here - caching
some of the key zonelist state (for faster access.)
See below for some performance benchmark results. After all that
discussion with David on why I didn't need them, I went and got
some ;). I wanted to verify that I had not hurt the normal case
of memory allocation noticeably. At least for my one little
microbenchmark, I found (1) the normal case wasn't affected, and
(2) workloads that forced scanning across multiple nodes for
memory improved up to 10% fewer System CPU cycles and lower
elapsed clock time ('sys' and 'real'). Good. See details, below.
I didn't have the logic in get_page_from_freelist() for various
full nodes and zone reclaim failures correct. That should be
fixed up now - notice the new goto labels zonelist_scan,
this_zone_full, and try_next_zone, in get_page_from_freelist().
There are two reasons I persued this alternative, over some earlier
proposals that would have focused on optimizing the fake numa
emulation case by caching the last useful zone:
1) Contrary to what I said before, we (SGI, on large ia64 sn2 systems)
have seen real customer loads where the cost to scan the zonelist
was a problem, due to many nodes being full of memory before
we got to a node we could use. Or at least, I think we have.
This was related to me by another engineer, based on experiences
from some time past. So this is not guaranteed. Most likely, though.
The following approach should help such real numa systems just as
much as it helps fake numa systems, or any combination thereof.
2) The effort to distinguish fake from real numa, using node_distance,
so that we could cache a fake numa node and optimize choosing
it over equivalent distance fake nodes, while continuing to
properly scan all real nodes in distance order, was going to
require a nasty blob of zonelist and node distance munging.
The following approach has no new dependency on node distances or
zone sorting.
See comment in the patch below for a description of what it actually does.
Technical details of note (or controversy):
- See the use of "zlc_active" and "did_zlc_setup" below, to delay
adding any work for this new mechanism until we've looked at the
first zone in zonelist. I figured the odds of the first zone
having the memory we needed were high enough that we should just
look there, first, then get fancy only if we need to keep looking.
- Some odd hackery was needed to add items to struct zonelist, while
not tripping up the custom zonelists built by the mm/mempolicy.c
code for MPOL_BIND. My usual wordy comments below explain this.
Search for "MPOL_BIND".
- Some per-node data in the struct zonelist is now modified frequently,
with no locking. Multiple CPU cores on a node could hit and mangle
this data. The theory is that this is just performance hint data,
and the memory allocator will work just fine despite any such mangling.
The fields at risk are the struct 'zonelist_cache' fields 'fullzones'
(a bitmask) and 'last_full_zap' (unsigned long jiffies). It should
all be self correcting after at most a one second delay.
- This still does a linear scan of the same lengths as before. All
I've optimized is making the scan faster, not algorithmically
shorter. It is now able to scan a compact array of 'unsigned
short' in the case of many full nodes, so one cache line should
cover quite a few nodes, rather than each node hitting another
one or two new and distinct cache lines.
- If both Andi and Nick don't find this too complicated, I will be
(pleasantly) flabbergasted.
- I removed the comment claiming we only use one cachline's worth of
zonelist. We seem, at least in the fake numa case, to have put the
lie to that claim.
- I pay no attention to the various watermarks and such in this performance
hint. A node could be marked full for one watermark, and then skipped
over when searching for a page using a different watermark. I think
that's actually quite ok, as it will tend to slightly increase the
spreading of memory over other nodes, away from a memory stressed node.
===============
Performance - some benchmark results and analysis:
This benchmark runs a memory hog program that uses multiple
threads to touch alot of memory as quickly as it can.
Multiple runs were made, touching 12, 38, 64 or 90 GBytes out of
the total 96 GBytes on the system, and using 1, 19, 37, or 55
threads (on a 56 CPU system.) System, user and real (elapsed)
timings were recorded for each run, shown in units of seconds,
in the table below.
Two kernels were tested - 2.6.18-mm3 and the same kernel with
this zonelist caching patch added. The table also shows the
percentage improvement the zonelist caching sys time is over
(lower than) the stock *-mm kernel.
number 2.6.18-mm3 zonelist-cache delta (< 0 good) percent
GBs N ------------ -------------- ---------------- systime
mem threads sys user real sys user real sys user real better
12 1 153 24 177 151 24 176 -2 0 -1 1%
12 19 99 22 8 99 22 8 0 0 0 0%
12 37 111 25 6 112 25 6 1 0 0 -0%
12 55 115 25 5 110 23 5 -5 -2 0 4%
38 1 502 74 576 497 73 570 -5 -1 -6 0%
38 19 426 78 48 373 76 39 -53 -2 -9 12%
38 37 544 83 36 547 82 36 3 -1 0 -0%
38 55 501 77 23 511 80 24 10 3 1 -1%
64 1 917 125 1042 890 124 1014 -27 -1 -28 2%
64 19 1118 138 119 965 141 103 -153 3 -16 13%
64 37 1202 151 94 1136 150 81 -66 -1 -13 5%
64 55 1118 141 61 1072 140 58 -46 -1 -3 4%
90 1 1342 177 1519 1275 174 1450 -67 -3 -69 4%
90 19 2392 199 192 2116 189 176 -276 -10 -16 11%
90 37 3313 238 175 2972 225 145 -341 -13 -30 10%
90 55 1948 210 104 1843 213 100 -105 3 -4 5%
Notes:
1) This test ran a memory hog program that started a specified number N of
threads, and had each thread allocate and touch 1/N'th of
the total memory to be used in the test run in a single loop,
writing a constant word to memory, one store every 4096 bytes.
Watching this test during some earlier trial runs, I would see
each of these threads sit down on one CPU and stay there, for
the remainder of the pass, a different CPU for each thread.
2) The 'real' column is not comparable to the 'sys' or 'user' columns.
The 'real' column is seconds wall clock time elapsed, from beginning
to end of that test pass. The 'sys' and 'user' columns are total
CPU seconds spent on that test pass. For a 19 thread test run,
for example, the sum of 'sys' and 'user' could be up to 19 times the
number of 'real' elapsed wall clock seconds.
3) Tests were run on a fresh, single-user boot, to minimize the amount
of memory already in use at the start of the test, and to minimize
the amount of background activity that might interfere.
4) Tests were done on a 56 CPU, 28 Node system with 96 GBytes of RAM.
5) Notice that the 'real' time gets large for the single thread runs, even
though the measured 'sys' and 'user' times are modest. I'm not sure what
that means - probably something to do with it being slow for one thread to
be accessing memory along ways away. Perhaps the fake numa system, running
ostensibly the same workload, would not show this substantial degradation
of 'real' time for one thread on many nodes -- lets hope not.
6) The high thread count passes (one thread per CPU - on 55 of 56 CPUs)
ran quite efficiently, as one might expect. Each pair of threads needed
to allocate and touch the memory on the node the two threads shared, a
pleasantly parallizable workload.
7) The intermediate thread count passes, when asking for alot of memory forcing
them to go to a few neighboring nodes, improved the most with this zonelist
caching patch.
Conclusions:
* This zonelist cache patch probably makes little difference one way or the
other for most workloads on real numa hardware, if those workloads avoid
heavy off node allocations.
* For memory intensive workloads requiring substantial off-node allocations
on real numa hardware, this patch improves both kernel and elapsed timings
up to ten per-cent.
* For fake numa systems, I'm optimistic, but will have to leave that up to
Rohit Seth to actually test (once I get him a 2.6.18 backport.)
Signed-off-by: Paul Jackson <pj@sgi.com>
Cc: Rohit Seth <rohitseth@google.com>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Cc: David Rientjes <rientjes@cs.washington.edu>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 04:31:48 +00:00
|
|
|
|
2008-04-28 09:12:17 +00:00
|
|
|
/*
|
|
|
|
* This struct contains information about a zone in a zonelist. It is stored
|
|
|
|
* here to avoid dereferences into large structures and lookups of tables
|
|
|
|
*/
|
|
|
|
struct zoneref {
|
|
|
|
struct zone *zone; /* Pointer to actual zone */
|
|
|
|
int zone_idx; /* zone_idx(zoneref->zone) */
|
|
|
|
};
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* One allocation request operates on a zonelist. A zonelist
|
|
|
|
* is a list of zones, the first one is the 'goal' of the
|
|
|
|
* allocation, the other zones are fallback zones, in decreasing
|
|
|
|
* priority.
|
|
|
|
*
|
2008-04-28 09:12:17 +00:00
|
|
|
* To speed the reading of the zonelist, the zonerefs contain the zone index
|
|
|
|
* of the entry being read. Helper functions to access information given
|
|
|
|
* a struct zoneref are
|
|
|
|
*
|
|
|
|
* zonelist_zone() - Return the struct zone * for an entry in _zonerefs
|
|
|
|
* zonelist_zone_idx() - Return the index of the zone for an entry
|
|
|
|
* zonelist_node_idx() - Return the index of the node for an entry
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
struct zonelist {
|
2008-04-28 09:12:17 +00:00
|
|
|
struct zoneref _zonerefs[MAX_ZONES_PER_ZONELIST + 1];
|
2005-04-16 22:20:36 +00:00
|
|
|
};
|
|
|
|
|
2021-06-29 02:42:52 +00:00
|
|
|
/*
|
|
|
|
* The array of struct pages for flatmem.
|
|
|
|
* It must be declared for SPARSEMEM as well because there are configurations
|
|
|
|
* that rely on that.
|
|
|
|
*/
|
2006-09-27 08:50:01 +00:00
|
|
|
extern struct page *mem_map;
|
|
|
|
|
2019-09-23 22:38:06 +00:00
|
|
|
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
|
|
|
|
struct deferred_split {
|
|
|
|
spinlock_t split_queue_lock;
|
|
|
|
struct list_head split_queue;
|
|
|
|
unsigned long split_queue_len;
|
|
|
|
};
|
|
|
|
#endif
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* On NUMA machines, each NUMA node would have a pg_data_t to describe
|
2017-07-10 22:49:32 +00:00
|
|
|
* it's memory layout. On UMA machines there is a single pglist_data which
|
|
|
|
* describes the whole memory.
|
2005-04-16 22:20:36 +00:00
|
|
|
*
|
|
|
|
* Memory statistics and page replacement data structures are maintained on a
|
|
|
|
* per-zone basis.
|
|
|
|
*/
|
|
|
|
typedef struct pglist_data {
|
2020-06-11 01:41:50 +00:00
|
|
|
/*
|
|
|
|
* node_zones contains just the zones for THIS node. Not all of the
|
|
|
|
* zones may be populated, but it is the full list. It is referenced by
|
|
|
|
* this node's node_zonelists as well as other node's node_zonelists.
|
|
|
|
*/
|
2005-04-16 22:20:36 +00:00
|
|
|
struct zone node_zones[MAX_NR_ZONES];
|
2020-06-11 01:41:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* node_zonelists contains references to all zones in all nodes.
|
|
|
|
* Generally the first zones will be references to this node's
|
|
|
|
* node_zones.
|
|
|
|
*/
|
2007-10-16 08:25:37 +00:00
|
|
|
struct zonelist node_zonelists[MAX_ZONELISTS];
|
2020-06-11 01:41:50 +00:00
|
|
|
|
|
|
|
int nr_zones; /* number of populated zones in this node */
|
2021-06-29 02:43:05 +00:00
|
|
|
#ifdef CONFIG_FLATMEM /* means !SPARSEMEM */
|
2005-04-16 22:20:36 +00:00
|
|
|
struct page *node_mem_map;
|
mm/page_ext: resurrect struct page extending code for debugging
When we debug something, we'd like to insert some information to every
page. For this purpose, we sometimes modify struct page itself. But,
this has drawbacks. First, it requires re-compile. This makes us
hesitate to use the powerful debug feature so development process is
slowed down. And, second, sometimes it is impossible to rebuild the
kernel due to third party module dependency. At third, system behaviour
would be largely different after re-compile, because it changes size of
struct page greatly and this structure is accessed by every part of
kernel. Keeping this as it is would be better to reproduce errornous
situation.
This feature is intended to overcome above mentioned problems. This
feature allocates memory for extended data per page in certain place
rather than the struct page itself. This memory can be accessed by the
accessor functions provided by this code. During the boot process, it
checks whether allocation of huge chunk of memory is needed or not. If
not, it avoids allocating memory at all. With this advantage, we can
include this feature into the kernel in default and can avoid rebuild and
solve related problems.
Until now, memcg uses this technique. But, now, memcg decides to embed
their variable to struct page itself and it's code to extend struct page
has been removed. I'd like to use this code to develop debug feature, so
this patch resurrect it.
To help these things to work well, this patch introduces two callbacks for
clients. One is the need callback which is mandatory if user wants to
avoid useless memory allocation at boot-time. The other is optional, init
callback, which is used to do proper initialization after memory is
allocated. Detailed explanation about purpose of these functions is in
code comment. Please refer it.
Others are completely same with previous extension code in memcg.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: Jungsoo Son <jungsoo.son@lge.com>
Cc: Ingo Molnar <mingo@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>
2014-12-13 00:55:46 +00:00
|
|
|
#ifdef CONFIG_PAGE_EXTENSION
|
|
|
|
struct page_ext *node_page_ext;
|
|
|
|
#endif
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#endif
|
2018-04-05 23:22:27 +00:00
|
|
|
#if defined(CONFIG_MEMORY_HOTPLUG) || defined(CONFIG_DEFERRED_STRUCT_PAGE_INIT)
|
2005-10-30 01:16:52 +00:00
|
|
|
/*
|
2018-12-28 08:37:10 +00:00
|
|
|
* Must be held any time you expect node_start_pfn,
|
|
|
|
* node_present_pages, node_spanned_pages or nr_zones to stay constant.
|
2020-06-03 22:59:24 +00:00
|
|
|
* Also synchronizes pgdat->first_deferred_pfn during deferred page
|
|
|
|
* init.
|
2005-10-30 01:16:52 +00:00
|
|
|
*
|
2013-07-03 22:02:09 +00:00
|
|
|
* pgdat_resize_lock() and pgdat_resize_unlock() are provided to
|
2018-04-05 23:22:27 +00:00
|
|
|
* manipulate node_size_lock without checking for CONFIG_MEMORY_HOTPLUG
|
|
|
|
* or CONFIG_DEFERRED_STRUCT_PAGE_INIT.
|
2013-07-03 22:02:09 +00:00
|
|
|
*
|
2013-07-03 22:02:08 +00:00
|
|
|
* Nests above zone->lock and zone->span_seqlock
|
2005-10-30 01:16:52 +00:00
|
|
|
*/
|
|
|
|
spinlock_t node_size_lock;
|
|
|
|
#endif
|
2005-04-16 22:20:36 +00:00
|
|
|
unsigned long node_start_pfn;
|
|
|
|
unsigned long node_present_pages; /* total number of physical pages */
|
|
|
|
unsigned long node_spanned_pages; /* total size of physical page
|
|
|
|
range, including holes */
|
|
|
|
int node_id;
|
|
|
|
wait_queue_head_t kswapd_wait;
|
2012-07-31 23:44:35 +00:00
|
|
|
wait_queue_head_t pfmemalloc_wait;
|
mm/vmscan: throttle reclaim until some writeback completes if congested
Patch series "Remove dependency on congestion_wait in mm/", v5.
This series that removes all calls to congestion_wait in mm/ and deletes
wait_iff_congested. It's not a clever implementation but
congestion_wait has been broken for a long time [1].
Even if congestion throttling worked, it was never a great idea. While
excessive dirty/writeback pages at the tail of the LRU is one
possibility that reclaim may be slow, there is also the problem of too
many pages being isolated and reclaim failing for other reasons
(elevated references, too many pages isolated, excessive LRU contention
etc).
This series replaces the "congestion" throttling with 3 different types.
- If there are too many dirty/writeback pages, sleep until a timeout or
enough pages get cleaned
- If too many pages are isolated, sleep until enough isolated pages are
either reclaimed or put back on the LRU
- If no progress is being made, direct reclaim tasks sleep until
another task makes progress with acceptable efficiency.
This was initially tested with a mix of workloads that used to trigger
corner cases that no longer work. A new test case was created called
"stutterp" (pagereclaim-stutterp-noreaders in mmtests) using a freshly
created XFS filesystem. Note that it may be necessary to increase the
timeout of ssh if executing remotely as ssh itself can get throttled and
the connection may timeout.
stutterp varies the number of "worker" processes from 4 up to NR_CPUS*4
to check the impact as the number of direct reclaimers increase. It has
four types of worker.
- One "anon latency" worker creates small mappings with mmap() and
times how long it takes to fault the mapping reading it 4K at a time
- X file writers which is fio randomly writing X files where the total
size of the files add up to the allowed dirty_ratio. fio is allowed
to run for a warmup period to allow some file-backed pages to
accumulate. The duration of the warmup is based on the best-case
linear write speed of the storage.
- Y file readers which is fio randomly reading small files
- Z anon memory hogs which continually map (100-dirty_ratio)% of memory
- Total estimated WSS = (100+dirty_ration) percentage of memory
X+Y+Z+1 == NR_WORKERS varying from 4 up to NR_CPUS*4
The intent is to maximise the total WSS with a mix of file and anon
memory where some anonymous memory must be swapped and there is a high
likelihood of dirty/writeback pages reaching the end of the LRU.
The test can be configured to have no background readers to stress
dirty/writeback pages. The results below are based on having zero
readers.
The short summary of the results is that the series works and stalls
until some event occurs but the timeouts may need adjustment.
The test results are not broken down by patch as the series should be
treated as one block that replaces a broken throttling mechanism with a
working one.
Finally, three machines were tested but I'm reporting the worst set of
results. The other two machines had much better latencies for example.
First the results of the "anon latency" latency
stutterp
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r4
Amean mmap-4 31.4003 ( 0.00%) 2661.0198 (-8374.52%)
Amean mmap-7 38.1641 ( 0.00%) 149.2891 (-291.18%)
Amean mmap-12 60.0981 ( 0.00%) 187.8105 (-212.51%)
Amean mmap-21 161.2699 ( 0.00%) 213.9107 ( -32.64%)
Amean mmap-30 174.5589 ( 0.00%) 377.7548 (-116.41%)
Amean mmap-48 8106.8160 ( 0.00%) 1070.5616 ( 86.79%)
Stddev mmap-4 41.3455 ( 0.00%) 27573.9676 (-66591.66%)
Stddev mmap-7 53.5556 ( 0.00%) 4608.5860 (-8505.23%)
Stddev mmap-12 171.3897 ( 0.00%) 5559.4542 (-3143.75%)
Stddev mmap-21 1506.6752 ( 0.00%) 5746.2507 (-281.39%)
Stddev mmap-30 557.5806 ( 0.00%) 7678.1624 (-1277.05%)
Stddev mmap-48 61681.5718 ( 0.00%) 14507.2830 ( 76.48%)
Max-90 mmap-4 31.4243 ( 0.00%) 83.1457 (-164.59%)
Max-90 mmap-7 41.0410 ( 0.00%) 41.0720 ( -0.08%)
Max-90 mmap-12 66.5255 ( 0.00%) 53.9073 ( 18.97%)
Max-90 mmap-21 146.7479 ( 0.00%) 105.9540 ( 27.80%)
Max-90 mmap-30 193.9513 ( 0.00%) 64.3067 ( 66.84%)
Max-90 mmap-48 277.9137 ( 0.00%) 591.0594 (-112.68%)
Max mmap-4 1913.8009 ( 0.00%) 299623.9695 (-15555.96%)
Max mmap-7 2423.9665 ( 0.00%) 204453.1708 (-8334.65%)
Max mmap-12 6845.6573 ( 0.00%) 221090.3366 (-3129.64%)
Max mmap-21 56278.6508 ( 0.00%) 213877.3496 (-280.03%)
Max mmap-30 19716.2990 ( 0.00%) 216287.6229 (-997.00%)
Max mmap-48 477923.9400 ( 0.00%) 245414.8238 ( 48.65%)
For most thread counts, the time to mmap() is unfortunately increased.
In earlier versions of the series, this was lower but a large number of
throttling events were reaching their timeout increasing the amount of
inefficient scanning of the LRU. There is no prioritisation of reclaim
tasks making progress based on each tasks rate of page allocation versus
progress of reclaim. The variance is also impacted for high worker
counts but in all cases, the differences in latency are not
statistically significant due to very large maximum outliers. Max-90
shows that 90% of the stalls are comparable but the Max results show the
massive outliers which are increased to to stalling.
It is expected that this will be very machine dependant. Due to the
test design, reclaim is difficult so allocations stall and there are
variances depending on whether THPs can be allocated or not. The amount
of memory will affect exactly how bad the corner cases are and how often
they trigger. The warmup period calculation is not ideal as it's based
on linear writes where as fio is randomly writing multiple files from
multiple tasks so the start state of the test is variable. For example,
these are the latencies on a single-socket machine that had more memory
Amean mmap-4 42.2287 ( 0.00%) 49.6838 * -17.65%*
Amean mmap-7 216.4326 ( 0.00%) 47.4451 * 78.08%*
Amean mmap-12 2412.0588 ( 0.00%) 51.7497 ( 97.85%)
Amean mmap-21 5546.2548 ( 0.00%) 51.8862 ( 99.06%)
Amean mmap-30 1085.3121 ( 0.00%) 72.1004 ( 93.36%)
The overall system CPU usage and elapsed time is as follows
5.15.0-rc3 5.15.0-rc3
vanilla mm-reclaimcongest-v5r4
Duration User 6989.03 983.42
Duration System 7308.12 799.68
Duration Elapsed 2277.67 2092.98
The patches reduce system CPU usage by 89% as the vanilla kernel is rarely
stalling.
The high-level /proc/vmstats show
5.15.0-rc1 5.15.0-rc1
vanilla mm-reclaimcongest-v5r2
Ops Direct pages scanned 1056608451.00 503594991.00
Ops Kswapd pages scanned 109795048.00 147289810.00
Ops Kswapd pages reclaimed 63269243.00 31036005.00
Ops Direct pages reclaimed 10803973.00 6328887.00
Ops Kswapd efficiency % 57.62 21.07
Ops Kswapd velocity 48204.98 57572.86
Ops Direct efficiency % 1.02 1.26
Ops Direct velocity 463898.83 196845.97
Kswapd scanned less pages but the detailed pattern is different. The
vanilla kernel scans slowly over time where as the patches exhibits
burst patterns of scan activity. Direct reclaim scanning is reduced by
52% due to stalling.
The pattern for stealing pages is also slightly different. Both kernels
exhibit spikes but the vanilla kernel when reclaiming shows pages being
reclaimed over a period of time where as the patches tend to reclaim in
spikes. The difference is that vanilla is not throttling and instead
scanning constantly finding some pages over time where as the patched
kernel throttles and reclaims in spikes.
Ops Percentage direct scans 90.59 77.37
For direct reclaim, vanilla scanned 90.59% of pages where as with the
patches, 77.37% were direct reclaim due to throttling
Ops Page writes by reclaim 2613590.00 1687131.00
Page writes from reclaim context are reduced.
Ops Page writes anon 2932752.00 1917048.00
And there is less swapping.
Ops Page reclaim immediate 996248528.00 107664764.00
The number of pages encountered at the tail of the LRU tagged for
immediate reclaim but still dirty/writeback is reduced by 89%.
Ops Slabs scanned 164284.00 153608.00
Slab scan activity is similar.
ftrace was used to gather stall activity
Vanilla
-------
1 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=16000
2 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=12000
8 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=8000
29 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=4000
82394 writeback_wait_iff_congested: usec_timeout=100000 usec_delayed=0
The fast majority of wait_iff_congested calls do not stall at all. What
is likely happening is that cond_resched() reschedules the task for a
short period when the BDI is not registering congestion (which it never
will in this test setup).
1 writeback_congestion_wait: usec_timeout=100000 usec_delayed=120000
2 writeback_congestion_wait: usec_timeout=100000 usec_delayed=132000
4 writeback_congestion_wait: usec_timeout=100000 usec_delayed=112000
380 writeback_congestion_wait: usec_timeout=100000 usec_delayed=108000
778 writeback_congestion_wait: usec_timeout=100000 usec_delayed=104000
congestion_wait if called always exceeds the timeout as there is no
trigger to wake it up.
Bottom line: Vanilla will throttle but it's not effective.
Patch series
------------
Kswapd throttle activity was always due to scanning pages tagged for
immediate reclaim at the tail of the LRU
1 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
4 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
94 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
112 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority of events did not stall or stalled for a short period.
Roughly 16% of stalls reached the timeout before expiry. For direct
reclaim, the number of times stalled for each reason were
6624 reason=VMSCAN_THROTTLE_ISOLATED
93246 reason=VMSCAN_THROTTLE_NOPROGRESS
96934 reason=VMSCAN_THROTTLE_WRITEBACK
The most common reason to stall was due to excessive pages tagged for
immediate reclaim at the tail of the LRU followed by a failure to make
forward. A relatively small number were due to too many pages isolated
from the LRU by parallel threads
For VMSCAN_THROTTLE_ISOLATED, the breakdown of delays was
9 usec_timeout=20000 usect_delayed=4000 reason=VMSCAN_THROTTLE_ISOLATED
12 usec_timeout=20000 usect_delayed=16000 reason=VMSCAN_THROTTLE_ISOLATED
83 usec_timeout=20000 usect_delayed=20000 reason=VMSCAN_THROTTLE_ISOLATED
6520 usec_timeout=20000 usect_delayed=0 reason=VMSCAN_THROTTLE_ISOLATED
Most did not stall at all. A small number reached the timeout.
For VMSCAN_THROTTLE_NOPROGRESS, the breakdown of stalls were all over
the map
1 usec_timeout=500000 usect_delayed=324000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=332000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=348000 reason=VMSCAN_THROTTLE_NOPROGRESS
1 usec_timeout=500000 usect_delayed=360000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=228000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=260000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=340000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=364000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=372000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=428000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=460000 reason=VMSCAN_THROTTLE_NOPROGRESS
2 usec_timeout=500000 usect_delayed=464000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=244000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=252000 reason=VMSCAN_THROTTLE_NOPROGRESS
3 usec_timeout=500000 usect_delayed=272000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=188000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=268000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=328000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=380000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=392000 reason=VMSCAN_THROTTLE_NOPROGRESS
4 usec_timeout=500000 usect_delayed=432000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=204000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=220000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=412000 reason=VMSCAN_THROTTLE_NOPROGRESS
5 usec_timeout=500000 usect_delayed=436000 reason=VMSCAN_THROTTLE_NOPROGRESS
6 usec_timeout=500000 usect_delayed=488000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=212000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=300000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=316000 reason=VMSCAN_THROTTLE_NOPROGRESS
7 usec_timeout=500000 usect_delayed=472000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=248000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=356000 reason=VMSCAN_THROTTLE_NOPROGRESS
8 usec_timeout=500000 usect_delayed=456000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=124000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=376000 reason=VMSCAN_THROTTLE_NOPROGRESS
9 usec_timeout=500000 usect_delayed=484000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=172000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=420000 reason=VMSCAN_THROTTLE_NOPROGRESS
10 usec_timeout=500000 usect_delayed=452000 reason=VMSCAN_THROTTLE_NOPROGRESS
11 usec_timeout=500000 usect_delayed=256000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=112000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=116000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=144000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=152000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=264000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=384000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=424000 reason=VMSCAN_THROTTLE_NOPROGRESS
12 usec_timeout=500000 usect_delayed=492000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=184000 reason=VMSCAN_THROTTLE_NOPROGRESS
13 usec_timeout=500000 usect_delayed=444000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=308000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=440000 reason=VMSCAN_THROTTLE_NOPROGRESS
14 usec_timeout=500000 usect_delayed=476000 reason=VMSCAN_THROTTLE_NOPROGRESS
16 usec_timeout=500000 usect_delayed=140000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=232000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=240000 reason=VMSCAN_THROTTLE_NOPROGRESS
17 usec_timeout=500000 usect_delayed=280000 reason=VMSCAN_THROTTLE_NOPROGRESS
18 usec_timeout=500000 usect_delayed=404000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=148000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=216000 reason=VMSCAN_THROTTLE_NOPROGRESS
20 usec_timeout=500000 usect_delayed=468000 reason=VMSCAN_THROTTLE_NOPROGRESS
21 usec_timeout=500000 usect_delayed=448000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=168000 reason=VMSCAN_THROTTLE_NOPROGRESS
23 usec_timeout=500000 usect_delayed=296000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=132000 reason=VMSCAN_THROTTLE_NOPROGRESS
25 usec_timeout=500000 usect_delayed=352000 reason=VMSCAN_THROTTLE_NOPROGRESS
26 usec_timeout=500000 usect_delayed=180000 reason=VMSCAN_THROTTLE_NOPROGRESS
27 usec_timeout=500000 usect_delayed=284000 reason=VMSCAN_THROTTLE_NOPROGRESS
28 usec_timeout=500000 usect_delayed=164000 reason=VMSCAN_THROTTLE_NOPROGRESS
29 usec_timeout=500000 usect_delayed=136000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=200000 reason=VMSCAN_THROTTLE_NOPROGRESS
30 usec_timeout=500000 usect_delayed=400000 reason=VMSCAN_THROTTLE_NOPROGRESS
31 usec_timeout=500000 usect_delayed=196000 reason=VMSCAN_THROTTLE_NOPROGRESS
32 usec_timeout=500000 usect_delayed=156000 reason=VMSCAN_THROTTLE_NOPROGRESS
33 usec_timeout=500000 usect_delayed=224000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=128000 reason=VMSCAN_THROTTLE_NOPROGRESS
35 usec_timeout=500000 usect_delayed=176000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=368000 reason=VMSCAN_THROTTLE_NOPROGRESS
36 usec_timeout=500000 usect_delayed=496000 reason=VMSCAN_THROTTLE_NOPROGRESS
37 usec_timeout=500000 usect_delayed=312000 reason=VMSCAN_THROTTLE_NOPROGRESS
38 usec_timeout=500000 usect_delayed=304000 reason=VMSCAN_THROTTLE_NOPROGRESS
40 usec_timeout=500000 usect_delayed=288000 reason=VMSCAN_THROTTLE_NOPROGRESS
43 usec_timeout=500000 usect_delayed=408000 reason=VMSCAN_THROTTLE_NOPROGRESS
55 usec_timeout=500000 usect_delayed=416000 reason=VMSCAN_THROTTLE_NOPROGRESS
56 usec_timeout=500000 usect_delayed=76000 reason=VMSCAN_THROTTLE_NOPROGRESS
58 usec_timeout=500000 usect_delayed=120000 reason=VMSCAN_THROTTLE_NOPROGRESS
59 usec_timeout=500000 usect_delayed=208000 reason=VMSCAN_THROTTLE_NOPROGRESS
61 usec_timeout=500000 usect_delayed=68000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=192000 reason=VMSCAN_THROTTLE_NOPROGRESS
71 usec_timeout=500000 usect_delayed=480000 reason=VMSCAN_THROTTLE_NOPROGRESS
79 usec_timeout=500000 usect_delayed=60000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=320000 reason=VMSCAN_THROTTLE_NOPROGRESS
82 usec_timeout=500000 usect_delayed=92000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=64000 reason=VMSCAN_THROTTLE_NOPROGRESS
85 usec_timeout=500000 usect_delayed=80000 reason=VMSCAN_THROTTLE_NOPROGRESS
88 usec_timeout=500000 usect_delayed=84000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=160000 reason=VMSCAN_THROTTLE_NOPROGRESS
90 usec_timeout=500000 usect_delayed=292000 reason=VMSCAN_THROTTLE_NOPROGRESS
94 usec_timeout=500000 usect_delayed=56000 reason=VMSCAN_THROTTLE_NOPROGRESS
118 usec_timeout=500000 usect_delayed=88000 reason=VMSCAN_THROTTLE_NOPROGRESS
119 usec_timeout=500000 usect_delayed=72000 reason=VMSCAN_THROTTLE_NOPROGRESS
126 usec_timeout=500000 usect_delayed=108000 reason=VMSCAN_THROTTLE_NOPROGRESS
146 usec_timeout=500000 usect_delayed=52000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=36000 reason=VMSCAN_THROTTLE_NOPROGRESS
148 usec_timeout=500000 usect_delayed=48000 reason=VMSCAN_THROTTLE_NOPROGRESS
159 usec_timeout=500000 usect_delayed=28000 reason=VMSCAN_THROTTLE_NOPROGRESS
178 usec_timeout=500000 usect_delayed=44000 reason=VMSCAN_THROTTLE_NOPROGRESS
183 usec_timeout=500000 usect_delayed=40000 reason=VMSCAN_THROTTLE_NOPROGRESS
237 usec_timeout=500000 usect_delayed=100000 reason=VMSCAN_THROTTLE_NOPROGRESS
266 usec_timeout=500000 usect_delayed=32000 reason=VMSCAN_THROTTLE_NOPROGRESS
313 usec_timeout=500000 usect_delayed=24000 reason=VMSCAN_THROTTLE_NOPROGRESS
347 usec_timeout=500000 usect_delayed=96000 reason=VMSCAN_THROTTLE_NOPROGRESS
470 usec_timeout=500000 usect_delayed=20000 reason=VMSCAN_THROTTLE_NOPROGRESS
559 usec_timeout=500000 usect_delayed=16000 reason=VMSCAN_THROTTLE_NOPROGRESS
964 usec_timeout=500000 usect_delayed=12000 reason=VMSCAN_THROTTLE_NOPROGRESS
2001 usec_timeout=500000 usect_delayed=104000 reason=VMSCAN_THROTTLE_NOPROGRESS
2447 usec_timeout=500000 usect_delayed=8000 reason=VMSCAN_THROTTLE_NOPROGRESS
7888 usec_timeout=500000 usect_delayed=4000 reason=VMSCAN_THROTTLE_NOPROGRESS
22727 usec_timeout=500000 usect_delayed=0 reason=VMSCAN_THROTTLE_NOPROGRESS
51305 usec_timeout=500000 usect_delayed=500000 reason=VMSCAN_THROTTLE_NOPROGRESS
The full timeout is often hit but a large number also do not stall at
all. The remainder slept a little allowing other reclaim tasks to make
progress.
While this timeout could be further increased, it could also negatively
impact worst-case behaviour when there is no prioritisation of what task
should make progress.
For VMSCAN_THROTTLE_WRITEBACK, the breakdown was
1 usec_timeout=100000 usect_delayed=44000 reason=VMSCAN_THROTTLE_WRITEBACK
2 usec_timeout=100000 usect_delayed=76000 reason=VMSCAN_THROTTLE_WRITEBACK
3 usec_timeout=100000 usect_delayed=80000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=48000 reason=VMSCAN_THROTTLE_WRITEBACK
5 usec_timeout=100000 usect_delayed=84000 reason=VMSCAN_THROTTLE_WRITEBACK
6 usec_timeout=100000 usect_delayed=72000 reason=VMSCAN_THROTTLE_WRITEBACK
7 usec_timeout=100000 usect_delayed=88000 reason=VMSCAN_THROTTLE_WRITEBACK
11 usec_timeout=100000 usect_delayed=56000 reason=VMSCAN_THROTTLE_WRITEBACK
12 usec_timeout=100000 usect_delayed=64000 reason=VMSCAN_THROTTLE_WRITEBACK
16 usec_timeout=100000 usect_delayed=92000 reason=VMSCAN_THROTTLE_WRITEBACK
24 usec_timeout=100000 usect_delayed=68000 reason=VMSCAN_THROTTLE_WRITEBACK
28 usec_timeout=100000 usect_delayed=32000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=60000 reason=VMSCAN_THROTTLE_WRITEBACK
30 usec_timeout=100000 usect_delayed=96000 reason=VMSCAN_THROTTLE_WRITEBACK
32 usec_timeout=100000 usect_delayed=52000 reason=VMSCAN_THROTTLE_WRITEBACK
42 usec_timeout=100000 usect_delayed=40000 reason=VMSCAN_THROTTLE_WRITEBACK
77 usec_timeout=100000 usect_delayed=28000 reason=VMSCAN_THROTTLE_WRITEBACK
99 usec_timeout=100000 usect_delayed=36000 reason=VMSCAN_THROTTLE_WRITEBACK
137 usec_timeout=100000 usect_delayed=24000 reason=VMSCAN_THROTTLE_WRITEBACK
190 usec_timeout=100000 usect_delayed=20000 reason=VMSCAN_THROTTLE_WRITEBACK
339 usec_timeout=100000 usect_delayed=16000 reason=VMSCAN_THROTTLE_WRITEBACK
518 usec_timeout=100000 usect_delayed=12000 reason=VMSCAN_THROTTLE_WRITEBACK
852 usec_timeout=100000 usect_delayed=8000 reason=VMSCAN_THROTTLE_WRITEBACK
3359 usec_timeout=100000 usect_delayed=4000 reason=VMSCAN_THROTTLE_WRITEBACK
7147 usec_timeout=100000 usect_delayed=0 reason=VMSCAN_THROTTLE_WRITEBACK
83962 usec_timeout=100000 usect_delayed=100000 reason=VMSCAN_THROTTLE_WRITEBACK
The majority hit the timeout in direct reclaim context although a
sizable number did not stall at all. This is very different to kswapd
where only a tiny percentage of stalls due to writeback reached the
timeout.
Bottom line, the throttling appears to work and the wakeup events may
limit worst case stalls. There might be some grounds for adjusting
timeouts but it's likely futile as the worst-case scenarios depend on
the workload, memory size and the speed of the storage. A better
approach to improve the series further would be to prioritise tasks
based on their rate of allocation with the caveat that it may be very
expensive to track.
This patch (of 5):
Page reclaim throttles on wait_iff_congested under the following
conditions:
- kswapd is encountering pages under writeback and marked for immediate
reclaim implying that pages are cycling through the LRU faster than
pages can be cleaned.
- Direct reclaim will stall if all dirty pages are backed by congested
inodes.
wait_iff_congested is almost completely broken with few exceptions.
This patch adds a new node-based workqueue and tracks the number of
throttled tasks and pages written back since throttling started. If
enough pages belonging to the node are written back then the throttled
tasks will wake early. If not, the throttled tasks sleeps until the
timeout expires.
[neilb@suse.de: Uninterruptible sleep and simpler wakeups]
[hdanton@sina.com: Avoid race when reclaim starts]
[vbabka@suse.cz: vmstat irq-safe api, clarifications]
Link: https://lore.kernel.org/linux-mm/45d8b7a6-8548-65f5-cccf-9f451d4ae3d4@kernel.dk/ [1]
Link: https://lkml.kernel.org/r/20211022144651.19914-1-mgorman@techsingularity.net
Link: https://lkml.kernel.org/r/20211022144651.19914-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: NeilBrown <neilb@suse.de>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: "Darrick J . Wong" <djwong@kernel.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 20:42:25 +00:00
|
|
|
|
|
|
|
/* workqueues for throttling reclaim for different reasons. */
|
|
|
|
wait_queue_head_t reclaim_wait[NR_VMSCAN_THROTTLE];
|
|
|
|
|
|
|
|
atomic_t nr_writeback_throttled;/* nr of writeback-throttled tasks */
|
|
|
|
unsigned long nr_reclaim_start; /* nr pages written while throttled
|
|
|
|
* when throttling started. */
|
mem-hotplug: implement get/put_online_mems
kmem_cache_{create,destroy,shrink} need to get a stable value of
cpu/node online mask, because they init/destroy/access per-cpu/node
kmem_cache parts, which can be allocated or destroyed on cpu/mem
hotplug. To protect against cpu hotplug, these functions use
{get,put}_online_cpus. However, they do nothing to synchronize with
memory hotplug - taking the slab_mutex does not eliminate the
possibility of race as described in patch 2.
What we need there is something like get_online_cpus, but for memory.
We already have lock_memory_hotplug, which serves for the purpose, but
it's a bit of a hammer right now, because it's backed by a mutex. As a
result, it imposes some limitations to locking order, which are not
desirable, and can't be used just like get_online_cpus. That's why in
patch 1 I substitute it with get/put_online_mems, which work exactly
like get/put_online_cpus except they block not cpu, but memory hotplug.
[ v1 can be found at https://lkml.org/lkml/2014/4/6/68. I NAK'ed it by
myself, because it used an rw semaphore for get/put_online_mems,
making them dead lock prune. ]
This patch (of 2):
{un}lock_memory_hotplug, which is used to synchronize against memory
hotplug, is currently backed by a mutex, which makes it a bit of a
hammer - threads that only want to get a stable value of online nodes
mask won't be able to proceed concurrently. Also, it imposes some
strong locking ordering rules on it, which narrows down the set of its
usage scenarios.
This patch introduces get/put_online_mems, which are the same as
get/put_online_cpus, but for memory hotplug, i.e. executing a code
inside a get/put_online_mems section will guarantee a stable value of
online nodes, present pages, etc.
lock_memory_hotplug()/unlock_memory_hotplug() are removed altogether.
Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Tang Chen <tangchen@cn.fujitsu.com>
Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: Jiang Liu <liuj97@gmail.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:07:18 +00:00
|
|
|
struct task_struct *kswapd; /* Protected by
|
|
|
|
mem_hotplug_begin/end() */
|
2016-07-28 22:45:49 +00:00
|
|
|
int kswapd_order;
|
2020-06-03 22:59:01 +00:00
|
|
|
enum zone_type kswapd_highest_zoneidx;
|
2016-07-28 22:45:49 +00:00
|
|
|
|
mm: fix 100% CPU kswapd busyloop on unreclaimable nodes
Patch series "mm: kswapd spinning on unreclaimable nodes - fixes and
cleanups".
Jia reported a scenario in which the kswapd of a node indefinitely spins
at 100% CPU usage. We have seen similar cases at Facebook.
The kernel's current method of judging its ability to reclaim a node (or
whether to back off and sleep) is based on the amount of scanned pages
in proportion to the amount of reclaimable pages. In Jia's and our
scenarios, there are no reclaimable pages in the node, however, and the
condition for backing off is never met. Kswapd busyloops in an attempt
to restore the watermarks while having nothing to work with.
This series reworks the definition of an unreclaimable node based not on
scanning but on whether kswapd is able to actually reclaim pages in
MAX_RECLAIM_RETRIES (16) consecutive runs. This is the same criteria
the page allocator uses for giving up on direct reclaim and invoking the
OOM killer. If it cannot free any pages, kswapd will go to sleep and
leave further attempts to direct reclaim invocations, which will either
make progress and re-enable kswapd, or invoke the OOM killer.
Patch #1 fixes the immediate problem Jia reported, the remainder are
smaller fixlets, cleanups, and overall phasing out of the old method.
Patch #6 is the odd one out. It's a nice cleanup to get_scan_count(),
and directly related to #5, but in itself not relevant to the series.
If the whole series is too ambitious for 4.11, I would consider the
first three patches fixes, the rest cleanups.
This patch (of 9):
Jia He reports a problem with kswapd spinning at 100% CPU when
requesting more hugepages than memory available in the system:
$ echo 4000 >/proc/sys/vm/nr_hugepages
top - 13:42:59 up 3:37, 1 user, load average: 1.09, 1.03, 1.01
Tasks: 1 total, 1 running, 0 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.0 us, 12.5 sy, 0.0 ni, 85.5 id, 2.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem: 31371520 total, 30915136 used, 456384 free, 320 buffers
KiB Swap: 6284224 total, 115712 used, 6168512 free. 48192 cached Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
76 root 20 0 0 0 0 R 100.0 0.000 217:17.29 kswapd3
At that time, there are no reclaimable pages left in the node, but as
kswapd fails to restore the high watermarks it refuses to go to sleep.
Kswapd needs to back away from nodes that fail to balance. Up until
commit 1d82de618ddd ("mm, vmscan: make kswapd reclaim in terms of
nodes") kswapd had such a mechanism. It considered zones whose
theoretically reclaimable pages it had reclaimed six times over as
unreclaimable and backed away from them. This guard was erroneously
removed as the patch changed the definition of a balanced node.
However, simply restoring this code wouldn't help in the case reported
here: there *are* no reclaimable pages that could be scanned until the
threshold is met. Kswapd would stay awake anyway.
Introduce a new and much simpler way of backing off. If kswapd runs
through MAX_RECLAIM_RETRIES (16) cycles without reclaiming a single
page, make it back off from the node. This is the same number of shots
direct reclaim takes before declaring OOM. Kswapd will go to sleep on
that node until a direct reclaimer manages to reclaim some pages, thus
proving the node reclaimable again.
[hannes@cmpxchg.org: check kswapd failure against the cumulative nr_reclaimed count]
Link: http://lkml.kernel.org/r/20170306162410.GB2090@cmpxchg.org
[shakeelb@google.com: fix condition for throttle_direct_reclaim]
Link: http://lkml.kernel.org/r/20170314183228.20152-1-shakeelb@google.com
Link: http://lkml.kernel.org/r/20170228214007.5621-2-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Shakeel Butt <shakeelb@google.com>
Reported-by: Jia He <hejianet@gmail.com>
Tested-by: Jia He <hejianet@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 21:51:51 +00:00
|
|
|
int kswapd_failures; /* Number of 'reclaimed == 0' runs */
|
|
|
|
|
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
|
|
|
#ifdef CONFIG_COMPACTION
|
|
|
|
int kcompactd_max_order;
|
2020-06-03 22:59:01 +00:00
|
|
|
enum zone_type 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
|
|
|
wait_queue_head_t kcompactd_wait;
|
|
|
|
struct task_struct *kcompactd;
|
2021-09-02 21:59:59 +00:00
|
|
|
bool proactive_compact_trigger;
|
2012-03-23 19:56:34 +00:00
|
|
|
#endif
|
2016-07-28 22:46:11 +00:00
|
|
|
/*
|
|
|
|
* This is a per-node reserve of pages that are not available
|
|
|
|
* to userspace allocations.
|
|
|
|
*/
|
|
|
|
unsigned long totalreserve_pages;
|
|
|
|
|
2016-07-28 22:46:32 +00:00
|
|
|
#ifdef CONFIG_NUMA
|
|
|
|
/*
|
2020-01-31 06:15:19 +00:00
|
|
|
* node reclaim becomes active if more unmapped pages exist.
|
2016-07-28 22:46:32 +00:00
|
|
|
*/
|
|
|
|
unsigned long min_unmapped_pages;
|
|
|
|
unsigned long min_slab_pages;
|
|
|
|
#endif /* CONFIG_NUMA */
|
|
|
|
|
2016-07-28 22:45:28 +00:00
|
|
|
/* Write-intensive fields used by page reclaim */
|
|
|
|
ZONE_PADDING(_pad1_)
|
2015-06-30 21:57:02 +00:00
|
|
|
|
|
|
|
#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
|
|
|
|
/*
|
|
|
|
* If memory initialisation on large machines is deferred then this
|
|
|
|
* is the first PFN that needs to be initialised.
|
|
|
|
*/
|
|
|
|
unsigned long first_deferred_pfn;
|
|
|
|
#endif /* CONFIG_DEFERRED_STRUCT_PAGE_INIT */
|
2016-02-03 00:57:08 +00:00
|
|
|
|
|
|
|
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
|
2019-09-23 22:38:06 +00:00
|
|
|
struct deferred_split deferred_split_queue;
|
2016-02-03 00:57:08 +00:00
|
|
|
#endif
|
mm, vmstat: add infrastructure for per-node vmstats
Patchset: "Move LRU page reclaim from zones to nodes v9"
This series moves LRUs from the zones to the node. While this is a
current rebase, the test results were based on mmotm as of June 23rd.
Conceptually, this series is simple but there are a lot of details.
Some of the broad motivations for this are;
1. The residency of a page partially depends on what zone the page was
allocated from. This is partially combatted by the fair zone allocation
policy but that is a partial solution that introduces overhead in the
page allocator paths.
2. Currently, reclaim on node 0 behaves slightly different to node 1. For
example, direct reclaim scans in zonelist order and reclaims even if
the zone is over the high watermark regardless of the age of pages
in that LRU. Kswapd on the other hand starts reclaim on the highest
unbalanced zone. A difference in distribution of file/anon pages due
to when they were allocated results can result in a difference in
again. While the fair zone allocation policy mitigates some of the
problems here, the page reclaim results on a multi-zone node will
always be different to a single-zone node.
it was scheduled on as a result.
3. kswapd and the page allocator scan zones in the opposite order to
avoid interfering with each other but it's sensitive to timing. This
mitigates the page allocator using pages that were allocated very recently
in the ideal case but it's sensitive to timing. When kswapd is allocating
from lower zones then it's great but during the rebalancing of the highest
zone, the page allocator and kswapd interfere with each other. It's worse
if the highest zone is small and difficult to balance.
4. slab shrinkers are node-based which makes it harder to identify the exact
relationship between slab reclaim and LRU reclaim.
The reason we have zone-based reclaim is that we used to have
large highmem zones in common configurations and it was necessary
to quickly find ZONE_NORMAL pages for reclaim. Today, this is much
less of a concern as machines with lots of memory will (or should) use
64-bit kernels. Combinations of 32-bit hardware and 64-bit hardware are
rare. Machines that do use highmem should have relatively low highmem:lowmem
ratios than we worried about in the past.
Conceptually, moving to node LRUs should be easier to understand. The
page allocator plays fewer tricks to game reclaim and reclaim behaves
similarly on all nodes.
The series has been tested on a 16 core UMA machine and a 2-socket 48
core NUMA machine. The UMA results are presented in most cases as the NUMA
machine behaved similarly.
pagealloc
---------
This is a microbenchmark that shows the benefit of removing the fair zone
allocation policy. It was tested uip to order-4 but only orders 0 and 1 are
shown as the other orders were comparable.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min total-odr0-1 490.00 ( 0.00%) 457.00 ( 6.73%)
Min total-odr0-2 347.00 ( 0.00%) 329.00 ( 5.19%)
Min total-odr0-4 288.00 ( 0.00%) 273.00 ( 5.21%)
Min total-odr0-8 251.00 ( 0.00%) 239.00 ( 4.78%)
Min total-odr0-16 234.00 ( 0.00%) 222.00 ( 5.13%)
Min total-odr0-32 223.00 ( 0.00%) 211.00 ( 5.38%)
Min total-odr0-64 217.00 ( 0.00%) 208.00 ( 4.15%)
Min total-odr0-128 214.00 ( 0.00%) 204.00 ( 4.67%)
Min total-odr0-256 250.00 ( 0.00%) 230.00 ( 8.00%)
Min total-odr0-512 271.00 ( 0.00%) 269.00 ( 0.74%)
Min total-odr0-1024 291.00 ( 0.00%) 282.00 ( 3.09%)
Min total-odr0-2048 303.00 ( 0.00%) 296.00 ( 2.31%)
Min total-odr0-4096 311.00 ( 0.00%) 309.00 ( 0.64%)
Min total-odr0-8192 316.00 ( 0.00%) 314.00 ( 0.63%)
Min total-odr0-16384 317.00 ( 0.00%) 315.00 ( 0.63%)
Min total-odr1-1 742.00 ( 0.00%) 712.00 ( 4.04%)
Min total-odr1-2 562.00 ( 0.00%) 530.00 ( 5.69%)
Min total-odr1-4 457.00 ( 0.00%) 433.00 ( 5.25%)
Min total-odr1-8 411.00 ( 0.00%) 381.00 ( 7.30%)
Min total-odr1-16 381.00 ( 0.00%) 356.00 ( 6.56%)
Min total-odr1-32 372.00 ( 0.00%) 346.00 ( 6.99%)
Min total-odr1-64 372.00 ( 0.00%) 343.00 ( 7.80%)
Min total-odr1-128 375.00 ( 0.00%) 351.00 ( 6.40%)
Min total-odr1-256 379.00 ( 0.00%) 351.00 ( 7.39%)
Min total-odr1-512 385.00 ( 0.00%) 355.00 ( 7.79%)
Min total-odr1-1024 386.00 ( 0.00%) 358.00 ( 7.25%)
Min total-odr1-2048 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-4096 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-8192 388.00 ( 0.00%) 363.00 ( 6.44%)
This shows a steady improvement throughout. The primary benefit is from
reduced system CPU usage which is obvious from the overall times;
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
User 189.19 191.80
System 2604.45 2533.56
Elapsed 2855.30 2786.39
The vmstats also showed that the fair zone allocation policy was definitely
removed as can be seen here;
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v8
DMA32 allocs 28794729769 0
Normal allocs 48432501431 77227309877
Movable allocs 0 0
tiobench on ext4
----------------
tiobench is a benchmark that artifically benefits if old pages remain resident
while new pages get reclaimed. The fair zone allocation policy mitigates this
problem so pages age fairly. While the benchmark has problems, it is important
that tiobench performance remains constant as it implies that page aging
problems that the fair zone allocation policy fixes are not re-introduced.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min PotentialReadSpeed 89.65 ( 0.00%) 90.21 ( 0.62%)
Min SeqRead-MB/sec-1 82.68 ( 0.00%) 82.01 ( -0.81%)
Min SeqRead-MB/sec-2 72.76 ( 0.00%) 72.07 ( -0.95%)
Min SeqRead-MB/sec-4 75.13 ( 0.00%) 74.92 ( -0.28%)
Min SeqRead-MB/sec-8 64.91 ( 0.00%) 65.19 ( 0.43%)
Min SeqRead-MB/sec-16 62.24 ( 0.00%) 62.22 ( -0.03%)
Min RandRead-MB/sec-1 0.88 ( 0.00%) 0.88 ( 0.00%)
Min RandRead-MB/sec-2 0.95 ( 0.00%) 0.92 ( -3.16%)
Min RandRead-MB/sec-4 1.43 ( 0.00%) 1.34 ( -6.29%)
Min RandRead-MB/sec-8 1.61 ( 0.00%) 1.60 ( -0.62%)
Min RandRead-MB/sec-16 1.80 ( 0.00%) 1.90 ( 5.56%)
Min SeqWrite-MB/sec-1 76.41 ( 0.00%) 76.85 ( 0.58%)
Min SeqWrite-MB/sec-2 74.11 ( 0.00%) 73.54 ( -0.77%)
Min SeqWrite-MB/sec-4 80.05 ( 0.00%) 80.13 ( 0.10%)
Min SeqWrite-MB/sec-8 72.88 ( 0.00%) 73.20 ( 0.44%)
Min SeqWrite-MB/sec-16 75.91 ( 0.00%) 76.44 ( 0.70%)
Min RandWrite-MB/sec-1 1.18 ( 0.00%) 1.14 ( -3.39%)
Min RandWrite-MB/sec-2 1.02 ( 0.00%) 1.03 ( 0.98%)
Min RandWrite-MB/sec-4 1.05 ( 0.00%) 0.98 ( -6.67%)
Min RandWrite-MB/sec-8 0.89 ( 0.00%) 0.92 ( 3.37%)
Min RandWrite-MB/sec-16 0.92 ( 0.00%) 0.93 ( 1.09%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 approx-v9
User 645.72 525.90
System 403.85 331.75
Elapsed 6795.36 6783.67
This shows that the series has little or not impact on tiobench which is
desirable and a reduction in system CPU usage. It indicates that the fair
zone allocation policy was removed in a manner that didn't reintroduce
one class of page aging bug. There were only minor differences in overall
reclaim activity
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Minor Faults 645838 647465
Major Faults 573 640
Swap Ins 0 0
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 46041453 44190646
Normal allocs 78053072 79887245
Movable allocs 0 0
Allocation stalls 24 67
Stall zone DMA 0 0
Stall zone DMA32 0 0
Stall zone Normal 0 2
Stall zone HighMem 0 0
Stall zone Movable 0 65
Direct pages scanned 10969 30609
Kswapd pages scanned 93375144 93492094
Kswapd pages reclaimed 93372243 93489370
Direct pages reclaimed 10969 30609
Kswapd efficiency 99% 99%
Kswapd velocity 13741.015 13781.934
Direct efficiency 100% 100%
Direct velocity 1.614 4.512
Percentage direct scans 0% 0%
kswapd activity was roughly comparable. There were differences in direct
reclaim activity but negligible in the context of the overall workload
(velocity of 4 pages per second with the patches applied, 1.6 pages per
second in the baseline kernel).
pgbench read-only large configuration on ext4
---------------------------------------------
pgbench is a database benchmark that can be sensitive to page reclaim
decisions. This also checks if removing the fair zone allocation policy
is safe
pgbench Transactions
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Hmean 1 188.26 ( 0.00%) 189.78 ( 0.81%)
Hmean 5 330.66 ( 0.00%) 328.69 ( -0.59%)
Hmean 12 370.32 ( 0.00%) 380.72 ( 2.81%)
Hmean 21 368.89 ( 0.00%) 369.00 ( 0.03%)
Hmean 30 382.14 ( 0.00%) 360.89 ( -5.56%)
Hmean 32 428.87 ( 0.00%) 432.96 ( 0.95%)
Negligible differences again. As with tiobench, overall reclaim activity
was comparable.
bonnie++ on ext4
----------------
No interesting performance difference, negligible differences on reclaim
stats.
paralleldd on ext4
------------------
This workload uses varying numbers of dd instances to read large amounts of
data from disk.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Amean Elapsd-1 186.04 ( 0.00%) 189.41 ( -1.82%)
Amean Elapsd-3 192.27 ( 0.00%) 191.38 ( 0.46%)
Amean Elapsd-5 185.21 ( 0.00%) 182.75 ( 1.33%)
Amean Elapsd-7 183.71 ( 0.00%) 182.11 ( 0.87%)
Amean Elapsd-12 180.96 ( 0.00%) 181.58 ( -0.35%)
Amean Elapsd-16 181.36 ( 0.00%) 183.72 ( -1.30%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
User 1548.01 1552.44
System 8609.71 8515.08
Elapsed 3587.10 3594.54
There is little or no change in performance but some drop in system CPU usage.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Minor Faults 362662 367360
Major Faults 1204 1143
Swap Ins 22 0
Swap Outs 2855 1029
DMA allocs 0 0
DMA32 allocs 31409797 28837521
Normal allocs 46611853 49231282
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 40845270 40869088
Kswapd pages reclaimed 40830976 40855294
Direct pages reclaimed 0 0
Kswapd efficiency 99% 99%
Kswapd velocity 11386.711 11369.769
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Page writes by reclaim 2855 1029
Page writes file 0 0
Page writes anon 2855 1029
Page reclaim immediate 771 1628
Sector Reads 293312636 293536360
Sector Writes 18213568 18186480
Page rescued immediate 0 0
Slabs scanned 128257 132747
Direct inode steals 181 56
Kswapd inode steals 59 1131
It basically shows that kswapd was active at roughly the same rate in
both kernels. There was also comparable slab scanning activity and direct
reclaim was avoided in both cases. There appears to be a large difference
in numbers of inodes reclaimed but the workload has few active inodes and
is likely a timing artifact.
stutter
-------
stutter simulates a simple workload. One part uses a lot of anonymous
memory, a second measures mmap latency and a third copies a large file.
The primary metric is checking for mmap latency.
stutter
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Min mmap 16.6283 ( 0.00%) 13.4258 ( 19.26%)
1st-qrtle mmap 54.7570 ( 0.00%) 34.9121 ( 36.24%)
2nd-qrtle mmap 57.3163 ( 0.00%) 46.1147 ( 19.54%)
3rd-qrtle mmap 58.9976 ( 0.00%) 47.1882 ( 20.02%)
Max-90% mmap 59.7433 ( 0.00%) 47.4453 ( 20.58%)
Max-93% mmap 60.1298 ( 0.00%) 47.6037 ( 20.83%)
Max-95% mmap 73.4112 ( 0.00%) 82.8719 (-12.89%)
Max-99% mmap 92.8542 ( 0.00%) 88.8870 ( 4.27%)
Max mmap 1440.6569 ( 0.00%) 121.4201 ( 91.57%)
Mean mmap 59.3493 ( 0.00%) 42.2991 ( 28.73%)
Best99%Mean mmap 57.2121 ( 0.00%) 41.8207 ( 26.90%)
Best95%Mean mmap 55.9113 ( 0.00%) 39.9620 ( 28.53%)
Best90%Mean mmap 55.6199 ( 0.00%) 39.3124 ( 29.32%)
Best50%Mean mmap 53.2183 ( 0.00%) 33.1307 ( 37.75%)
Best10%Mean mmap 45.9842 ( 0.00%) 20.4040 ( 55.63%)
Best5%Mean mmap 43.2256 ( 0.00%) 17.9654 ( 58.44%)
Best1%Mean mmap 32.9388 ( 0.00%) 16.6875 ( 49.34%)
This shows a number of improvements with the worst-case outlier greatly
improved.
Some of the vmstats are interesting
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Swap Ins 163 502
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 618719206 1381662383
Normal allocs 891235743 564138421
Movable allocs 0 0
Allocation stalls 2603 1
Direct pages scanned 216787 2
Kswapd pages scanned 50719775 41778378
Kswapd pages reclaimed 41541765 41777639
Direct pages reclaimed 209159 0
Kswapd efficiency 81% 99%
Kswapd velocity 16859.554 14329.059
Direct efficiency 96% 0%
Direct velocity 72.061 0.001
Percentage direct scans 0% 0%
Page writes by reclaim 6215049 0
Page writes file 6215049 0
Page writes anon 0 0
Page reclaim immediate 70673 90
Sector Reads 81940800 81680456
Sector Writes 100158984 98816036
Page rescued immediate 0 0
Slabs scanned 1366954 22683
While this is not guaranteed in all cases, this particular test showed
a large reduction in direct reclaim activity. It's also worth noting
that no page writes were issued from reclaim context.
This series is not without its hazards. There are at least three areas
that I'm concerned with even though I could not reproduce any problems in
that area.
1. Reclaim/compaction is going to be affected because the amount of reclaim is
no longer targetted at a specific zone. Compaction works on a per-zone basis
so there is no guarantee that reclaiming a few THP's worth page pages will
have a positive impact on compaction success rates.
2. The Slab/LRU reclaim ratio is affected because the frequency the shrinkers
are called is now different. This may or may not be a problem but if it
is, it'll be because shrinkers are not called enough and some balancing
is required.
3. The anon/file reclaim ratio may be affected. Pages about to be dirtied are
distributed between zones and the fair zone allocation policy used to do
something very similar for anon. The distribution is now different but not
necessarily in any way that matters but it's still worth bearing in mind.
VM statistic counters for reclaim decisions are zone-based. If the kernel
is to reclaim on a per-node basis then we need to track per-node
statistics but there is no infrastructure for that. The most notable
change is that the old node_page_state is renamed to
sum_zone_node_page_state. The new node_page_state takes a pglist_data and
uses per-node stats but none exist yet. There is some renaming such as
vm_stat to vm_zone_stat and the addition of vm_node_stat and the renaming
of mod_state to mod_zone_state. Otherwise, this is mostly a mechanical
patch with no functional change. There is a lot of similarity between the
node and zone helpers which is unfortunate but there was no obvious way of
reusing the code and maintaining type safety.
Link: http://lkml.kernel.org/r/1467970510-21195-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:45:24 +00:00
|
|
|
|
2016-07-28 22:45:31 +00:00
|
|
|
/* Fields commonly accessed by the page reclaim scanner */
|
2019-12-01 01:55:34 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* NOTE: THIS IS UNUSED IF MEMCG IS ENABLED.
|
|
|
|
*
|
|
|
|
* Use mem_cgroup_lruvec() to look up lruvecs.
|
|
|
|
*/
|
|
|
|
struct lruvec __lruvec;
|
2016-07-28 22:45:31 +00:00
|
|
|
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
ZONE_PADDING(_pad2_)
|
|
|
|
|
mm, vmstat: add infrastructure for per-node vmstats
Patchset: "Move LRU page reclaim from zones to nodes v9"
This series moves LRUs from the zones to the node. While this is a
current rebase, the test results were based on mmotm as of June 23rd.
Conceptually, this series is simple but there are a lot of details.
Some of the broad motivations for this are;
1. The residency of a page partially depends on what zone the page was
allocated from. This is partially combatted by the fair zone allocation
policy but that is a partial solution that introduces overhead in the
page allocator paths.
2. Currently, reclaim on node 0 behaves slightly different to node 1. For
example, direct reclaim scans in zonelist order and reclaims even if
the zone is over the high watermark regardless of the age of pages
in that LRU. Kswapd on the other hand starts reclaim on the highest
unbalanced zone. A difference in distribution of file/anon pages due
to when they were allocated results can result in a difference in
again. While the fair zone allocation policy mitigates some of the
problems here, the page reclaim results on a multi-zone node will
always be different to a single-zone node.
it was scheduled on as a result.
3. kswapd and the page allocator scan zones in the opposite order to
avoid interfering with each other but it's sensitive to timing. This
mitigates the page allocator using pages that were allocated very recently
in the ideal case but it's sensitive to timing. When kswapd is allocating
from lower zones then it's great but during the rebalancing of the highest
zone, the page allocator and kswapd interfere with each other. It's worse
if the highest zone is small and difficult to balance.
4. slab shrinkers are node-based which makes it harder to identify the exact
relationship between slab reclaim and LRU reclaim.
The reason we have zone-based reclaim is that we used to have
large highmem zones in common configurations and it was necessary
to quickly find ZONE_NORMAL pages for reclaim. Today, this is much
less of a concern as machines with lots of memory will (or should) use
64-bit kernels. Combinations of 32-bit hardware and 64-bit hardware are
rare. Machines that do use highmem should have relatively low highmem:lowmem
ratios than we worried about in the past.
Conceptually, moving to node LRUs should be easier to understand. The
page allocator plays fewer tricks to game reclaim and reclaim behaves
similarly on all nodes.
The series has been tested on a 16 core UMA machine and a 2-socket 48
core NUMA machine. The UMA results are presented in most cases as the NUMA
machine behaved similarly.
pagealloc
---------
This is a microbenchmark that shows the benefit of removing the fair zone
allocation policy. It was tested uip to order-4 but only orders 0 and 1 are
shown as the other orders were comparable.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min total-odr0-1 490.00 ( 0.00%) 457.00 ( 6.73%)
Min total-odr0-2 347.00 ( 0.00%) 329.00 ( 5.19%)
Min total-odr0-4 288.00 ( 0.00%) 273.00 ( 5.21%)
Min total-odr0-8 251.00 ( 0.00%) 239.00 ( 4.78%)
Min total-odr0-16 234.00 ( 0.00%) 222.00 ( 5.13%)
Min total-odr0-32 223.00 ( 0.00%) 211.00 ( 5.38%)
Min total-odr0-64 217.00 ( 0.00%) 208.00 ( 4.15%)
Min total-odr0-128 214.00 ( 0.00%) 204.00 ( 4.67%)
Min total-odr0-256 250.00 ( 0.00%) 230.00 ( 8.00%)
Min total-odr0-512 271.00 ( 0.00%) 269.00 ( 0.74%)
Min total-odr0-1024 291.00 ( 0.00%) 282.00 ( 3.09%)
Min total-odr0-2048 303.00 ( 0.00%) 296.00 ( 2.31%)
Min total-odr0-4096 311.00 ( 0.00%) 309.00 ( 0.64%)
Min total-odr0-8192 316.00 ( 0.00%) 314.00 ( 0.63%)
Min total-odr0-16384 317.00 ( 0.00%) 315.00 ( 0.63%)
Min total-odr1-1 742.00 ( 0.00%) 712.00 ( 4.04%)
Min total-odr1-2 562.00 ( 0.00%) 530.00 ( 5.69%)
Min total-odr1-4 457.00 ( 0.00%) 433.00 ( 5.25%)
Min total-odr1-8 411.00 ( 0.00%) 381.00 ( 7.30%)
Min total-odr1-16 381.00 ( 0.00%) 356.00 ( 6.56%)
Min total-odr1-32 372.00 ( 0.00%) 346.00 ( 6.99%)
Min total-odr1-64 372.00 ( 0.00%) 343.00 ( 7.80%)
Min total-odr1-128 375.00 ( 0.00%) 351.00 ( 6.40%)
Min total-odr1-256 379.00 ( 0.00%) 351.00 ( 7.39%)
Min total-odr1-512 385.00 ( 0.00%) 355.00 ( 7.79%)
Min total-odr1-1024 386.00 ( 0.00%) 358.00 ( 7.25%)
Min total-odr1-2048 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-4096 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-8192 388.00 ( 0.00%) 363.00 ( 6.44%)
This shows a steady improvement throughout. The primary benefit is from
reduced system CPU usage which is obvious from the overall times;
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
User 189.19 191.80
System 2604.45 2533.56
Elapsed 2855.30 2786.39
The vmstats also showed that the fair zone allocation policy was definitely
removed as can be seen here;
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v8
DMA32 allocs 28794729769 0
Normal allocs 48432501431 77227309877
Movable allocs 0 0
tiobench on ext4
----------------
tiobench is a benchmark that artifically benefits if old pages remain resident
while new pages get reclaimed. The fair zone allocation policy mitigates this
problem so pages age fairly. While the benchmark has problems, it is important
that tiobench performance remains constant as it implies that page aging
problems that the fair zone allocation policy fixes are not re-introduced.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min PotentialReadSpeed 89.65 ( 0.00%) 90.21 ( 0.62%)
Min SeqRead-MB/sec-1 82.68 ( 0.00%) 82.01 ( -0.81%)
Min SeqRead-MB/sec-2 72.76 ( 0.00%) 72.07 ( -0.95%)
Min SeqRead-MB/sec-4 75.13 ( 0.00%) 74.92 ( -0.28%)
Min SeqRead-MB/sec-8 64.91 ( 0.00%) 65.19 ( 0.43%)
Min SeqRead-MB/sec-16 62.24 ( 0.00%) 62.22 ( -0.03%)
Min RandRead-MB/sec-1 0.88 ( 0.00%) 0.88 ( 0.00%)
Min RandRead-MB/sec-2 0.95 ( 0.00%) 0.92 ( -3.16%)
Min RandRead-MB/sec-4 1.43 ( 0.00%) 1.34 ( -6.29%)
Min RandRead-MB/sec-8 1.61 ( 0.00%) 1.60 ( -0.62%)
Min RandRead-MB/sec-16 1.80 ( 0.00%) 1.90 ( 5.56%)
Min SeqWrite-MB/sec-1 76.41 ( 0.00%) 76.85 ( 0.58%)
Min SeqWrite-MB/sec-2 74.11 ( 0.00%) 73.54 ( -0.77%)
Min SeqWrite-MB/sec-4 80.05 ( 0.00%) 80.13 ( 0.10%)
Min SeqWrite-MB/sec-8 72.88 ( 0.00%) 73.20 ( 0.44%)
Min SeqWrite-MB/sec-16 75.91 ( 0.00%) 76.44 ( 0.70%)
Min RandWrite-MB/sec-1 1.18 ( 0.00%) 1.14 ( -3.39%)
Min RandWrite-MB/sec-2 1.02 ( 0.00%) 1.03 ( 0.98%)
Min RandWrite-MB/sec-4 1.05 ( 0.00%) 0.98 ( -6.67%)
Min RandWrite-MB/sec-8 0.89 ( 0.00%) 0.92 ( 3.37%)
Min RandWrite-MB/sec-16 0.92 ( 0.00%) 0.93 ( 1.09%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 approx-v9
User 645.72 525.90
System 403.85 331.75
Elapsed 6795.36 6783.67
This shows that the series has little or not impact on tiobench which is
desirable and a reduction in system CPU usage. It indicates that the fair
zone allocation policy was removed in a manner that didn't reintroduce
one class of page aging bug. There were only minor differences in overall
reclaim activity
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Minor Faults 645838 647465
Major Faults 573 640
Swap Ins 0 0
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 46041453 44190646
Normal allocs 78053072 79887245
Movable allocs 0 0
Allocation stalls 24 67
Stall zone DMA 0 0
Stall zone DMA32 0 0
Stall zone Normal 0 2
Stall zone HighMem 0 0
Stall zone Movable 0 65
Direct pages scanned 10969 30609
Kswapd pages scanned 93375144 93492094
Kswapd pages reclaimed 93372243 93489370
Direct pages reclaimed 10969 30609
Kswapd efficiency 99% 99%
Kswapd velocity 13741.015 13781.934
Direct efficiency 100% 100%
Direct velocity 1.614 4.512
Percentage direct scans 0% 0%
kswapd activity was roughly comparable. There were differences in direct
reclaim activity but negligible in the context of the overall workload
(velocity of 4 pages per second with the patches applied, 1.6 pages per
second in the baseline kernel).
pgbench read-only large configuration on ext4
---------------------------------------------
pgbench is a database benchmark that can be sensitive to page reclaim
decisions. This also checks if removing the fair zone allocation policy
is safe
pgbench Transactions
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Hmean 1 188.26 ( 0.00%) 189.78 ( 0.81%)
Hmean 5 330.66 ( 0.00%) 328.69 ( -0.59%)
Hmean 12 370.32 ( 0.00%) 380.72 ( 2.81%)
Hmean 21 368.89 ( 0.00%) 369.00 ( 0.03%)
Hmean 30 382.14 ( 0.00%) 360.89 ( -5.56%)
Hmean 32 428.87 ( 0.00%) 432.96 ( 0.95%)
Negligible differences again. As with tiobench, overall reclaim activity
was comparable.
bonnie++ on ext4
----------------
No interesting performance difference, negligible differences on reclaim
stats.
paralleldd on ext4
------------------
This workload uses varying numbers of dd instances to read large amounts of
data from disk.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Amean Elapsd-1 186.04 ( 0.00%) 189.41 ( -1.82%)
Amean Elapsd-3 192.27 ( 0.00%) 191.38 ( 0.46%)
Amean Elapsd-5 185.21 ( 0.00%) 182.75 ( 1.33%)
Amean Elapsd-7 183.71 ( 0.00%) 182.11 ( 0.87%)
Amean Elapsd-12 180.96 ( 0.00%) 181.58 ( -0.35%)
Amean Elapsd-16 181.36 ( 0.00%) 183.72 ( -1.30%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
User 1548.01 1552.44
System 8609.71 8515.08
Elapsed 3587.10 3594.54
There is little or no change in performance but some drop in system CPU usage.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Minor Faults 362662 367360
Major Faults 1204 1143
Swap Ins 22 0
Swap Outs 2855 1029
DMA allocs 0 0
DMA32 allocs 31409797 28837521
Normal allocs 46611853 49231282
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 40845270 40869088
Kswapd pages reclaimed 40830976 40855294
Direct pages reclaimed 0 0
Kswapd efficiency 99% 99%
Kswapd velocity 11386.711 11369.769
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Page writes by reclaim 2855 1029
Page writes file 0 0
Page writes anon 2855 1029
Page reclaim immediate 771 1628
Sector Reads 293312636 293536360
Sector Writes 18213568 18186480
Page rescued immediate 0 0
Slabs scanned 128257 132747
Direct inode steals 181 56
Kswapd inode steals 59 1131
It basically shows that kswapd was active at roughly the same rate in
both kernels. There was also comparable slab scanning activity and direct
reclaim was avoided in both cases. There appears to be a large difference
in numbers of inodes reclaimed but the workload has few active inodes and
is likely a timing artifact.
stutter
-------
stutter simulates a simple workload. One part uses a lot of anonymous
memory, a second measures mmap latency and a third copies a large file.
The primary metric is checking for mmap latency.
stutter
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Min mmap 16.6283 ( 0.00%) 13.4258 ( 19.26%)
1st-qrtle mmap 54.7570 ( 0.00%) 34.9121 ( 36.24%)
2nd-qrtle mmap 57.3163 ( 0.00%) 46.1147 ( 19.54%)
3rd-qrtle mmap 58.9976 ( 0.00%) 47.1882 ( 20.02%)
Max-90% mmap 59.7433 ( 0.00%) 47.4453 ( 20.58%)
Max-93% mmap 60.1298 ( 0.00%) 47.6037 ( 20.83%)
Max-95% mmap 73.4112 ( 0.00%) 82.8719 (-12.89%)
Max-99% mmap 92.8542 ( 0.00%) 88.8870 ( 4.27%)
Max mmap 1440.6569 ( 0.00%) 121.4201 ( 91.57%)
Mean mmap 59.3493 ( 0.00%) 42.2991 ( 28.73%)
Best99%Mean mmap 57.2121 ( 0.00%) 41.8207 ( 26.90%)
Best95%Mean mmap 55.9113 ( 0.00%) 39.9620 ( 28.53%)
Best90%Mean mmap 55.6199 ( 0.00%) 39.3124 ( 29.32%)
Best50%Mean mmap 53.2183 ( 0.00%) 33.1307 ( 37.75%)
Best10%Mean mmap 45.9842 ( 0.00%) 20.4040 ( 55.63%)
Best5%Mean mmap 43.2256 ( 0.00%) 17.9654 ( 58.44%)
Best1%Mean mmap 32.9388 ( 0.00%) 16.6875 ( 49.34%)
This shows a number of improvements with the worst-case outlier greatly
improved.
Some of the vmstats are interesting
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Swap Ins 163 502
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 618719206 1381662383
Normal allocs 891235743 564138421
Movable allocs 0 0
Allocation stalls 2603 1
Direct pages scanned 216787 2
Kswapd pages scanned 50719775 41778378
Kswapd pages reclaimed 41541765 41777639
Direct pages reclaimed 209159 0
Kswapd efficiency 81% 99%
Kswapd velocity 16859.554 14329.059
Direct efficiency 96% 0%
Direct velocity 72.061 0.001
Percentage direct scans 0% 0%
Page writes by reclaim 6215049 0
Page writes file 6215049 0
Page writes anon 0 0
Page reclaim immediate 70673 90
Sector Reads 81940800 81680456
Sector Writes 100158984 98816036
Page rescued immediate 0 0
Slabs scanned 1366954 22683
While this is not guaranteed in all cases, this particular test showed
a large reduction in direct reclaim activity. It's also worth noting
that no page writes were issued from reclaim context.
This series is not without its hazards. There are at least three areas
that I'm concerned with even though I could not reproduce any problems in
that area.
1. Reclaim/compaction is going to be affected because the amount of reclaim is
no longer targetted at a specific zone. Compaction works on a per-zone basis
so there is no guarantee that reclaiming a few THP's worth page pages will
have a positive impact on compaction success rates.
2. The Slab/LRU reclaim ratio is affected because the frequency the shrinkers
are called is now different. This may or may not be a problem but if it
is, it'll be because shrinkers are not called enough and some balancing
is required.
3. The anon/file reclaim ratio may be affected. Pages about to be dirtied are
distributed between zones and the fair zone allocation policy used to do
something very similar for anon. The distribution is now different but not
necessarily in any way that matters but it's still worth bearing in mind.
VM statistic counters for reclaim decisions are zone-based. If the kernel
is to reclaim on a per-node basis then we need to track per-node
statistics but there is no infrastructure for that. The most notable
change is that the old node_page_state is renamed to
sum_zone_node_page_state. The new node_page_state takes a pglist_data and
uses per-node stats but none exist yet. There is some renaming such as
vm_stat to vm_zone_stat and the addition of vm_node_stat and the renaming
of mod_state to mod_zone_state. Otherwise, this is mostly a mechanical
patch with no functional change. There is a lot of similarity between the
node and zone helpers which is unfortunate but there was no obvious way of
reusing the code and maintaining type safety.
Link: http://lkml.kernel.org/r/1467970510-21195-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-07-28 22:45:24 +00:00
|
|
|
/* Per-node vmstats */
|
|
|
|
struct per_cpu_nodestat __percpu *per_cpu_nodestats;
|
|
|
|
atomic_long_t vm_stat[NR_VM_NODE_STAT_ITEMS];
|
2005-04-16 22:20:36 +00:00
|
|
|
} pg_data_t;
|
|
|
|
|
|
|
|
#define node_present_pages(nid) (NODE_DATA(nid)->node_present_pages)
|
|
|
|
#define node_spanned_pages(nid) (NODE_DATA(nid)->node_spanned_pages)
|
2021-06-29 02:43:05 +00:00
|
|
|
#ifdef CONFIG_FLATMEM
|
[PATCH] remove non-DISCONTIG use of pgdat->node_mem_map
This patch effectively eliminates direct use of pgdat->node_mem_map outside
of the DISCONTIG code. On a flat memory system, these fields aren't
currently used, neither are they on a sparsemem system.
There was also a node_mem_map(nid) macro on many architectures. Its use
along with the use of ->node_mem_map itself was not consistent. It has
been removed in favor of two new, more explicit, arch-independent macros:
pgdat_page_nr(pgdat, pagenr)
nid_page_nr(nid, pagenr)
I called them "pgdat" and "nid" because we overload the term "node" to mean
"NUMA node", "DISCONTIG node" or "pg_data_t" in very confusing ways. I
believe the newer names are much clearer.
These macros can be overridden in the sparsemem case with a theoretically
slower operation using node_start_pfn and pfn_to_page(), instead. We could
make this the only behavior if people want, but I don't want to change too
much at once. One thing at a time.
This patch removes more code than it adds.
Compile tested on alpha, alpha discontig, arm, arm-discontig, i386, i386
generic, NUMAQ, Summit, ppc64, ppc64 discontig, and x86_64. Full list
here: http://sr71.net/patches/2.6.12/2.6.12-rc1-mhp2/configs/
Boot tested on NUMAQ, x86 SMP and ppc64 power4/5 LPARs.
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin J. Bligh <mbligh@aracnet.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:37 +00:00
|
|
|
#define pgdat_page_nr(pgdat, pagenr) ((pgdat)->node_mem_map + (pagenr))
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#else
|
|
|
|
#define pgdat_page_nr(pgdat, pagenr) pfn_to_page((pgdat)->node_start_pfn + (pagenr))
|
|
|
|
#endif
|
[PATCH] remove non-DISCONTIG use of pgdat->node_mem_map
This patch effectively eliminates direct use of pgdat->node_mem_map outside
of the DISCONTIG code. On a flat memory system, these fields aren't
currently used, neither are they on a sparsemem system.
There was also a node_mem_map(nid) macro on many architectures. Its use
along with the use of ->node_mem_map itself was not consistent. It has
been removed in favor of two new, more explicit, arch-independent macros:
pgdat_page_nr(pgdat, pagenr)
nid_page_nr(nid, pagenr)
I called them "pgdat" and "nid" because we overload the term "node" to mean
"NUMA node", "DISCONTIG node" or "pg_data_t" in very confusing ways. I
believe the newer names are much clearer.
These macros can be overridden in the sparsemem case with a theoretically
slower operation using node_start_pfn and pfn_to_page(), instead. We could
make this the only behavior if people want, but I don't want to change too
much at once. One thing at a time.
This patch removes more code than it adds.
Compile tested on alpha, alpha discontig, arm, arm-discontig, i386, i386
generic, NUMAQ, Summit, ppc64, ppc64 discontig, and x86_64. Full list
here: http://sr71.net/patches/2.6.12/2.6.12-rc1-mhp2/configs/
Boot tested on NUMAQ, x86 SMP and ppc64 power4/5 LPARs.
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin J. Bligh <mbligh@aracnet.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:37 +00:00
|
|
|
#define nid_page_nr(nid, pagenr) pgdat_page_nr(NODE_DATA(nid),(pagenr))
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2011-06-16 08:28:07 +00:00
|
|
|
#define node_start_pfn(nid) (NODE_DATA(nid)->node_start_pfn)
|
2013-02-23 00:35:27 +00:00
|
|
|
#define node_end_pfn(nid) pgdat_end_pfn(NODE_DATA(nid))
|
2011-06-16 08:28:07 +00:00
|
|
|
|
2013-02-23 00:35:27 +00:00
|
|
|
static inline unsigned long pgdat_end_pfn(pg_data_t *pgdat)
|
|
|
|
{
|
|
|
|
return pgdat->node_start_pfn + pgdat->node_spanned_pages;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool pgdat_is_empty(pg_data_t *pgdat)
|
|
|
|
{
|
|
|
|
return !pgdat->node_start_pfn && !pgdat->node_spanned_pages;
|
|
|
|
}
|
2011-06-16 08:28:07 +00:00
|
|
|
|
2005-10-30 01:16:52 +00:00
|
|
|
#include <linux/memory_hotplug.h>
|
|
|
|
|
2017-09-06 23:20:24 +00:00
|
|
|
void build_all_zonelists(pg_data_t *pgdat);
|
2018-04-05 23:25:16 +00:00
|
|
|
void wakeup_kswapd(struct zone *zone, gfp_t gfp_mask, int order,
|
2020-06-03 22:59:01 +00:00
|
|
|
enum zone_type highest_zoneidx);
|
2016-05-20 23:57:12 +00:00
|
|
|
bool __zone_watermark_ok(struct zone *z, unsigned int order, unsigned long mark,
|
2020-06-03 22:59:01 +00:00
|
|
|
int highest_zoneidx, unsigned int alloc_flags,
|
2016-05-20 23:57:12 +00:00
|
|
|
long free_pages);
|
2014-06-04 23:10:21 +00:00
|
|
|
bool zone_watermark_ok(struct zone *z, unsigned int order,
|
2020-06-03 22:59:01 +00:00
|
|
|
unsigned long mark, int highest_zoneidx,
|
2016-05-20 00:13:38 +00:00
|
|
|
unsigned int alloc_flags);
|
2014-06-04 23:10:21 +00:00
|
|
|
bool zone_watermark_ok_safe(struct zone *z, unsigned int order,
|
2020-06-03 22:59:01 +00:00
|
|
|
unsigned long mark, int highest_zoneidx);
|
2020-09-26 04:19:28 +00:00
|
|
|
/*
|
|
|
|
* Memory initialization context, use to differentiate memory added by
|
|
|
|
* the platform statically or via memory hotplug interface.
|
|
|
|
*/
|
|
|
|
enum meminit_context {
|
|
|
|
MEMINIT_EARLY,
|
|
|
|
MEMINIT_HOTPLUG,
|
2007-01-11 07:15:30 +00:00
|
|
|
};
|
2020-09-26 04:19:28 +00:00
|
|
|
|
mm: remove return value from init_currently_empty_zone
Patch series "mm: make movable onlining suck less", v4.
Movable onlining is a real hack with many downsides - mainly
reintroduction of lowmem/highmem issues we used to have on 32b systems -
but it is the only way to make the memory hotremove more reliable which
is something that people are asking for.
The current semantic of memory movable onlinening is really cumbersome,
however. The main reason for this is that the udev driven approach is
basically unusable because udev races with the memory probing while only
the last memory block or the one adjacent to the existing zone_movable
are allowed to be onlined movable. In short the criterion for the
successful online_movable changes under udev's feet. A reliable udev
approach would require a 2 phase approach where the first successful
movable online would have to check all the previous blocks and online
them in descending order. This is hard to be considered sane.
This patchset aims at making the onlining semantic more usable. First
of all it allows to online memory movable as long as it doesn't clash
with the existing ZONE_NORMAL. That means that ZONE_NORMAL and
ZONE_MOVABLE cannot overlap. Currently I preserve the original ordering
semantic so the zone always precedes the movable zone but I have plans
to remove this restriction in future because it is not really necessary.
First 3 patches are cleanups which should be ready to be merged right
away (unless I have missed something subtle of course).
Patch 4 deals with ZONE_DEVICE dependencies down the __add_pages path.
Patch 5 deals with implicit assumptions of register_one_node on pgdat
initialization.
Patches 6-10 deal with offline holes in the zone for pfn walkers. I
hope I got all of them right but people familiar with compaction should
double check this.
Patch 11 is the core of the change. In order to make it easier to
review I have tried it to be as minimalistic as possible and the large
code removal is moved to patch 14.
Patch 12 is a trivial follow up cleanup. Patch 13 fixes sparse warnings
and finally patch 14 removes the unused code.
I have tested the patches in kvm:
# qemu-system-x86_64 -enable-kvm -monitor pty -m 2G,slots=4,maxmem=4G -numa node,mem=1G -numa node,mem=1G ...
and then probed the additional memory by
(qemu) object_add memory-backend-ram,id=mem1,size=1G
(qemu) device_add pc-dimm,id=dimm1,memdev=mem1
Then I have used this simple script to probe the memory block by hand
# cat probe_memblock.sh
#!/bin/sh
BLOCK_NR=$1
# echo $((0x100000000+$BLOCK_NR*(128<<20))) > /sys/devices/system/memory/probe
# for i in $(seq 10); do sh probe_memblock.sh $i; done
# grep . /sys/devices/system/memory/memory3?/valid_zones 2>/dev/null
/sys/devices/system/memory/memory33/valid_zones:Normal Movable
/sys/devices/system/memory/memory34/valid_zones:Normal Movable
/sys/devices/system/memory/memory35/valid_zones:Normal Movable
/sys/devices/system/memory/memory36/valid_zones:Normal Movable
/sys/devices/system/memory/memory37/valid_zones:Normal Movable
/sys/devices/system/memory/memory38/valid_zones:Normal Movable
/sys/devices/system/memory/memory39/valid_zones:Normal Movable
The main difference to the original implementation is that all new
memblocks can be both online_kernel and online_movable initially because
there is no clash obviously. For the comparison the original
implementation would have
/sys/devices/system/memory/memory33/valid_zones:Normal
/sys/devices/system/memory/memory34/valid_zones:Normal
/sys/devices/system/memory/memory35/valid_zones:Normal
/sys/devices/system/memory/memory36/valid_zones:Normal
/sys/devices/system/memory/memory37/valid_zones:Normal
/sys/devices/system/memory/memory38/valid_zones:Normal
/sys/devices/system/memory/memory39/valid_zones:Normal Movable
Now
# echo online_movable > /sys/devices/system/memory/memory34/state
# grep . /sys/devices/system/memory/memory3?/valid_zones 2>/dev/null
/sys/devices/system/memory/memory33/valid_zones:Normal Movable
/sys/devices/system/memory/memory34/valid_zones:Movable
/sys/devices/system/memory/memory35/valid_zones:Movable
/sys/devices/system/memory/memory36/valid_zones:Movable
/sys/devices/system/memory/memory37/valid_zones:Movable
/sys/devices/system/memory/memory38/valid_zones:Movable
/sys/devices/system/memory/memory39/valid_zones:Movable
Block 33 can still be online both kernel and movable while all
the remaining can be only movable.
/proc/zonelist says
Node 0, zone Normal
pages free 0
min 0
low 0
high 0
spanned 0
present 0
--
Node 0, zone Movable
pages free 32753
min 85
low 117
high 149
spanned 32768
present 32768
A new memblock at a lower address will result in a new memblock (32)
which will still allow both Normal and Movable.
# sh probe_memblock.sh 0
# grep . /sys/devices/system/memory/memory3[2-5]/valid_zones 2>/dev/null
/sys/devices/system/memory/memory32/valid_zones:Normal Movable
/sys/devices/system/memory/memory33/valid_zones:Normal Movable
/sys/devices/system/memory/memory34/valid_zones:Movable
/sys/devices/system/memory/memory35/valid_zones:Movable
and online_kernel will convert it to the zone normal properly
while 33 can be still onlined both ways.
# echo online_kernel > /sys/devices/system/memory/memory32/state
# grep . /sys/devices/system/memory/memory3[2-5]/valid_zones 2>/dev/null
/sys/devices/system/memory/memory32/valid_zones:Normal
/sys/devices/system/memory/memory33/valid_zones:Normal Movable
/sys/devices/system/memory/memory34/valid_zones:Movable
/sys/devices/system/memory/memory35/valid_zones:Movable
/proc/zoneinfo will now tell
Node 0, zone Normal
pages free 65441
min 165
low 230
high 295
spanned 65536
present 65536
--
Node 0, zone Movable
pages free 32740
min 82
low 114
high 146
spanned 32768
present 32768
so both zones have one memblock spanned and present.
Onlining 39 should associate this block to the movable zone
# echo online > /sys/devices/system/memory/memory39/state
/proc/zoneinfo will now tell
Node 0, zone Normal
pages free 32765
min 80
low 112
high 144
spanned 32768
present 32768
--
Node 0, zone Movable
pages free 65501
min 160
low 225
high 290
spanned 196608
present 65536
so we will have a movable zone which spans 6 memblocks, 2 present and 4
representing a hole.
Offlining both movable blocks will lead to the zone with no present
pages which is the expected behavior I believe.
# echo offline > /sys/devices/system/memory/memory39/state
# echo offline > /sys/devices/system/memory/memory34/state
# grep -A6 "Movable\|Normal" /proc/zoneinfo
Node 0, zone Normal
pages free 32735
min 90
low 122
high 154
spanned 32768
present 32768
--
Node 0, zone Movable
pages free 0
min 0
low 0
high 0
spanned 196608
present 0
As a bonus we will get a nice cleanup in the memory hotplug codebase.
This patch (of 16):
init_currently_empty_zone doesn't have any error to return yet it is
still an int and callers try to be defensive and try to handle potential
error. Remove this nonsense and simplify all callers.
This patch shouldn't have any visible effect
Link: http://lkml.kernel.org/r/20170515085827.16474-2-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Acked-by: Balbir Singh <bsingharora@gmail.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Daniel Kiper <daniel.kiper@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Reza Arbab <arbab@linux.vnet.ibm.com>
Cc: Tobias Regnery <tobias.regnery@gmail.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-06 22:37:35 +00:00
|
|
|
extern void init_currently_empty_zone(struct zone *zone, unsigned long start_pfn,
|
2015-11-06 02:47:06 +00:00
|
|
|
unsigned long size);
|
2006-06-23 09:03:10 +00:00
|
|
|
|
memcg: fix hotplugged memory zone oops
When MEMCG is configured on (even when it's disabled by boot option),
when adding or removing a page to/from its lru list, the zone pointer
used for stats updates is nowadays taken from the struct lruvec. (On
many configurations, calculating zone from page is slower.)
But we have no code to update all the lruvecs (per zone, per memcg) when
a memory node is hotadded. Here's an extract from the oops which
results when running numactl to bind a program to a newly onlined node:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000f60
IP: __mod_zone_page_state+0x9/0x60
Pid: 1219, comm: numactl Not tainted 3.6.0-rc5+ #180 Bochs Bochs
Process numactl (pid: 1219, threadinfo ffff880039abc000, task ffff8800383c4ce0)
Call Trace:
__pagevec_lru_add_fn+0xdf/0x140
pagevec_lru_move_fn+0xb1/0x100
__pagevec_lru_add+0x1c/0x30
lru_add_drain_cpu+0xa3/0x130
lru_add_drain+0x2f/0x40
...
The natural solution might be to use a memcg callback whenever memory is
hotadded; but that solution has not been scoped out, and it happens that
we do have an easy location at which to update lruvec->zone. The lruvec
pointer is discovered either by mem_cgroup_zone_lruvec() or by
mem_cgroup_page_lruvec(), and both of those do know the right zone.
So check and set lruvec->zone in those; and remove the inadequate
attempt to set lruvec->zone from lruvec_init(), which is called before
NODE_DATA(node) has been allocated in such cases.
Ah, there was one exceptionr. For no particularly good reason,
mem_cgroup_force_empty_list() has its own code for deciding lruvec.
Change it to use the standard mem_cgroup_zone_lruvec() and
mem_cgroup_get_lru_size() too. In fact it was already safe against such
an oops (the lru lists in danger could only be empty), but we're better
proofed against future changes this way.
I've marked this for stable (3.6) since we introduced the problem in 3.5
(now closed to stable); but I have no idea if this is the only fix
needed to get memory hotadd working with memcg in 3.6, and received no
answer when I enquired twice before.
Reported-by: Tang Chen <tangchen@cn.fujitsu.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-11-16 22:14:54 +00:00
|
|
|
extern void lruvec_init(struct lruvec *lruvec);
|
2012-05-29 22:06:58 +00:00
|
|
|
|
2016-07-28 22:45:31 +00:00
|
|
|
static inline struct pglist_data *lruvec_pgdat(struct lruvec *lruvec)
|
2012-05-29 22:06:58 +00:00
|
|
|
{
|
2012-07-31 23:43:02 +00:00
|
|
|
#ifdef CONFIG_MEMCG
|
2016-07-28 22:45:31 +00:00
|
|
|
return lruvec->pgdat;
|
2012-05-29 22:06:58 +00:00
|
|
|
#else
|
2019-12-01 01:55:34 +00:00
|
|
|
return container_of(lruvec, struct pglist_data, __lruvec);
|
2012-05-29 22:06:58 +00:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2010-05-26 21:45:00 +00:00
|
|
|
#ifdef CONFIG_HAVE_MEMORYLESS_NODES
|
|
|
|
int local_memory_node(int node_id);
|
|
|
|
#else
|
|
|
|
static inline int local_memory_node(int node_id) { return node_id; };
|
|
|
|
#endif
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* zone_idx() returns 0 for the ZONE_DMA zone, 1 for the ZONE_NORMAL zone, etc.
|
|
|
|
*/
|
|
|
|
#define zone_idx(zone) ((zone) - (zone)->zone_pgdat->node_zones)
|
|
|
|
|
2021-02-26 01:17:05 +00:00
|
|
|
#ifdef CONFIG_ZONE_DEVICE
|
|
|
|
static inline bool zone_is_zone_device(struct zone *zone)
|
|
|
|
{
|
|
|
|
return zone_idx(zone) == ZONE_DEVICE;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static inline bool zone_is_zone_device(struct zone *zone)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2016-09-01 23:14:55 +00:00
|
|
|
/*
|
|
|
|
* Returns true if a zone has pages managed by the buddy allocator.
|
|
|
|
* All the reclaim decisions have to use this function rather than
|
|
|
|
* populated_zone(). If the whole zone is reserved then we can easily
|
|
|
|
* end up with populated_zone() && !managed_zone().
|
|
|
|
*/
|
|
|
|
static inline bool managed_zone(struct zone *zone)
|
|
|
|
{
|
2018-12-28 08:34:24 +00:00
|
|
|
return zone_managed_pages(zone);
|
2016-09-01 23:14:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Returns true if a zone has memory */
|
|
|
|
static inline bool populated_zone(struct zone *zone)
|
2006-01-06 08:11:15 +00:00
|
|
|
{
|
2016-09-01 23:14:55 +00:00
|
|
|
return zone->present_pages;
|
2006-01-06 08:11:15 +00:00
|
|
|
}
|
|
|
|
|
2018-08-22 04:53:32 +00:00
|
|
|
#ifdef CONFIG_NUMA
|
|
|
|
static inline int zone_to_nid(struct zone *zone)
|
|
|
|
{
|
|
|
|
return zone->node;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void zone_set_nid(struct zone *zone, int nid)
|
|
|
|
{
|
|
|
|
zone->node = nid;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static inline int zone_to_nid(struct zone *zone)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void zone_set_nid(struct zone *zone, int nid) {}
|
|
|
|
#endif
|
|
|
|
|
2007-07-17 11:03:12 +00:00
|
|
|
extern int movable_zone;
|
|
|
|
|
2006-09-26 06:31:13 +00:00
|
|
|
static inline int is_highmem_idx(enum zone_type idx)
|
2005-04-16 22:20:36 +00:00
|
|
|
{
|
2006-09-26 06:31:14 +00:00
|
|
|
#ifdef CONFIG_HIGHMEM
|
2007-07-17 11:03:12 +00:00
|
|
|
return (idx == ZONE_HIGHMEM ||
|
2021-06-29 02:41:04 +00:00
|
|
|
(idx == ZONE_MOVABLE && movable_zone == ZONE_HIGHMEM));
|
2006-09-26 06:31:14 +00:00
|
|
|
#else
|
|
|
|
return 0;
|
|
|
|
#endif
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
|
mm_zone: add function to check if managed dma zone exists
Patch series "Handle warning of allocation failure on DMA zone w/o
managed pages", v4.
**Problem observed:
On x86_64, when crash is triggered and entering into kdump kernel, page
allocation failure can always be seen.
---------------------------------
DMA: preallocated 128 KiB GFP_KERNEL pool for atomic allocations
swapper/0: page allocation failure: order:5, mode:0xcc1(GFP_KERNEL|GFP_DMA), nodemask=(null),cpuset=/,mems_allowed=0
CPU: 0 PID: 1 Comm: swapper/0
Call Trace:
dump_stack+0x7f/0xa1
warn_alloc.cold+0x72/0xd6
......
__alloc_pages+0x24d/0x2c0
......
dma_atomic_pool_init+0xdb/0x176
do_one_initcall+0x67/0x320
? rcu_read_lock_sched_held+0x3f/0x80
kernel_init_freeable+0x290/0x2dc
? rest_init+0x24f/0x24f
kernel_init+0xa/0x111
ret_from_fork+0x22/0x30
Mem-Info:
------------------------------------
***Root cause:
In the current kernel, it assumes that DMA zone must have managed pages
and try to request pages if CONFIG_ZONE_DMA is enabled. While this is not
always true. E.g in kdump kernel of x86_64, only low 1M is presented and
locked down at very early stage of boot, so that this low 1M won't be
added into buddy allocator to become managed pages of DMA zone. This
exception will always cause page allocation failure if page is requested
from DMA zone.
***Investigation:
This failure happens since below commit merged into linus's tree.
1a6a9044b967 x86/setup: Remove CONFIG_X86_RESERVE_LOW and reservelow= options
23721c8e92f7 x86/crash: Remove crash_reserve_low_1M()
f1d4d47c5851 x86/setup: Always reserve the first 1M of RAM
7c321eb2b843 x86/kdump: Remove the backup region handling
6f599d84231f x86/kdump: Always reserve the low 1M when the crashkernel option is specified
Before them, on x86_64, the low 640K area will be reused by kdump kernel.
So in kdump kernel, the content of low 640K area is copied into a backup
region for dumping before jumping into kdump. Then except of those firmware
reserved region in [0, 640K], the left area will be added into buddy
allocator to become available managed pages of DMA zone.
However, after above commits applied, in kdump kernel of x86_64, the low
1M is reserved by memblock, but not released to buddy allocator. So any
later page allocation requested from DMA zone will fail.
At the beginning, if crashkernel is reserved, the low 1M need be locked
down because AMD SME encrypts memory making the old backup region
mechanims impossible when switching into kdump kernel.
Later, it was also observed that there are BIOSes corrupting memory
under 1M. To solve this, in commit f1d4d47c5851, the entire region of
low 1M is always reserved after the real mode trampoline is allocated.
Besides, recently, Intel engineer mentioned their TDX (Trusted domain
extensions) which is under development in kernel also needs to lock down
the low 1M. So we can't simply revert above commits to fix the page allocation
failure from DMA zone as someone suggested.
***Solution:
Currently, only DMA atomic pool and dma-kmalloc will initialize and
request page allocation with GFP_DMA during bootup.
So only initializ DMA atomic pool when DMA zone has available managed
pages, otherwise just skip the initialization.
For dma-kmalloc(), for the time being, let's mute the warning of
allocation failure if requesting pages from DMA zone while no manged
pages. Meanwhile, change code to use dma_alloc_xx/dma_map_xx API to
replace kmalloc(GFP_DMA), or do not use GFP_DMA when calling kmalloc() if
not necessary. Christoph is posting patches to fix those under
drivers/scsi/. Finally, we can remove the need of dma-kmalloc() as people
suggested.
This patch (of 3):
In some places of the current kernel, it assumes that dma zone must have
managed pages if CONFIG_ZONE_DMA is enabled. While this is not always
true. E.g in kdump kernel of x86_64, only low 1M is presented and locked
down at very early stage of boot, so that there's no managed pages at all
in DMA zone. This exception will always cause page allocation failure if
page is requested from DMA zone.
Here add function has_managed_dma() and the relevant helper functions to
check if there's DMA zone with managed pages. It will be used in later
patches.
Link: https://lkml.kernel.org/r/20211223094435.248523-1-bhe@redhat.com
Link: https://lkml.kernel.org/r/20211223094435.248523-2-bhe@redhat.com
Fixes: 6f599d84231f ("x86/kdump: Always reserve the low 1M when the crashkernel option is specified")
Signed-off-by: Baoquan He <bhe@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: John Donnelly <john.p.donnelly@oracle.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Christoph Lameter <cl@linux.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Laight <David.Laight@ACULAB.COM>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Robin Murphy <robin.murphy@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>
2022-01-14 22:07:37 +00:00
|
|
|
#ifdef CONFIG_ZONE_DMA
|
|
|
|
bool has_managed_dma(void);
|
|
|
|
#else
|
|
|
|
static inline bool has_managed_dma(void)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/**
|
2018-10-30 22:07:40 +00:00
|
|
|
* is_highmem - helper function to quickly check if a struct zone is a
|
2005-04-16 22:20:36 +00:00
|
|
|
* highmem zone or not. This is an attempt to keep references
|
|
|
|
* to ZONE_{DMA/NORMAL/HIGHMEM/etc} in general code to a minimum.
|
2021-04-30 06:02:16 +00:00
|
|
|
* @zone: pointer to struct zone variable
|
|
|
|
* Return: 1 for a highmem zone, 0 otherwise
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
static inline int is_highmem(struct zone *zone)
|
|
|
|
{
|
2006-09-26 06:31:14 +00:00
|
|
|
#ifdef CONFIG_HIGHMEM
|
2016-05-20 00:11:57 +00:00
|
|
|
return is_highmem_idx(zone_idx(zone));
|
2006-09-26 06:31:14 +00:00
|
|
|
#else
|
|
|
|
return 0;
|
|
|
|
#endif
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* These two functions are used to setup the per zone pages min values */
|
|
|
|
struct ctl_table;
|
2020-04-24 06:43:36 +00:00
|
|
|
|
2020-04-24 06:43:38 +00:00
|
|
|
int min_free_kbytes_sysctl_handler(struct ctl_table *, int, void *, size_t *,
|
|
|
|
loff_t *);
|
|
|
|
int watermark_scale_factor_sysctl_handler(struct ctl_table *, int, void *,
|
|
|
|
size_t *, loff_t *);
|
2018-04-10 23:30:11 +00:00
|
|
|
extern int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES];
|
2020-04-24 06:43:38 +00:00
|
|
|
int lowmem_reserve_ratio_sysctl_handler(struct ctl_table *, int, void *,
|
|
|
|
size_t *, loff_t *);
|
2021-06-29 02:42:24 +00:00
|
|
|
int percpu_pagelist_high_fraction_sysctl_handler(struct ctl_table *, int,
|
|
|
|
void *, size_t *, loff_t *);
|
2006-07-03 07:24:13 +00:00
|
|
|
int sysctl_min_unmapped_ratio_sysctl_handler(struct ctl_table *, int,
|
2020-04-24 06:43:38 +00:00
|
|
|
void *, size_t *, loff_t *);
|
2006-09-26 06:31:52 +00:00
|
|
|
int sysctl_min_slab_ratio_sysctl_handler(struct ctl_table *, int,
|
2020-04-24 06:43:38 +00:00
|
|
|
void *, size_t *, loff_t *);
|
|
|
|
int numa_zonelist_order_handler(struct ctl_table *, int,
|
|
|
|
void *, size_t *, loff_t *);
|
2021-06-29 02:42:24 +00:00
|
|
|
extern int percpu_pagelist_high_fraction;
|
2007-07-16 06:38:01 +00:00
|
|
|
extern char numa_zonelist_order[];
|
2017-09-06 23:20:13 +00:00
|
|
|
#define NUMA_ZONELIST_ORDER_LEN 16
|
2007-07-16 06:38:01 +00:00
|
|
|
|
2021-06-29 02:43:01 +00:00
|
|
|
#ifndef CONFIG_NUMA
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
extern struct pglist_data contig_page_data;
|
2021-07-01 01:53:56 +00:00
|
|
|
static inline struct pglist_data *NODE_DATA(int nid)
|
|
|
|
{
|
|
|
|
return &contig_page_data;
|
|
|
|
}
|
2005-04-16 22:20:36 +00:00
|
|
|
#define NODE_MEM_MAP(nid) mem_map
|
|
|
|
|
2021-06-29 02:43:01 +00:00
|
|
|
#else /* CONFIG_NUMA */
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
#include <asm/mmzone.h>
|
|
|
|
|
2021-06-29 02:43:01 +00:00
|
|
|
#endif /* !CONFIG_NUMA */
|
2005-06-23 07:07:40 +00:00
|
|
|
|
2006-03-27 09:16:02 +00:00
|
|
|
extern struct pglist_data *first_online_pgdat(void);
|
|
|
|
extern struct pglist_data *next_online_pgdat(struct pglist_data *pgdat);
|
|
|
|
extern struct zone *next_zone(struct zone *zone);
|
2006-03-27 09:15:57 +00:00
|
|
|
|
|
|
|
/**
|
2008-05-23 20:05:01 +00:00
|
|
|
* for_each_online_pgdat - helper macro to iterate over all online nodes
|
2021-04-30 06:02:16 +00:00
|
|
|
* @pgdat: pointer to a pg_data_t variable
|
2006-03-27 09:15:57 +00:00
|
|
|
*/
|
|
|
|
#define for_each_online_pgdat(pgdat) \
|
|
|
|
for (pgdat = first_online_pgdat(); \
|
|
|
|
pgdat; \
|
|
|
|
pgdat = next_online_pgdat(pgdat))
|
|
|
|
/**
|
|
|
|
* for_each_zone - helper macro to iterate over all memory zones
|
2021-04-30 06:02:16 +00:00
|
|
|
* @zone: pointer to struct zone variable
|
2006-03-27 09:15:57 +00:00
|
|
|
*
|
|
|
|
* The user only needs to declare the zone variable, for_each_zone
|
|
|
|
* fills it in.
|
|
|
|
*/
|
|
|
|
#define for_each_zone(zone) \
|
|
|
|
for (zone = (first_online_pgdat())->node_zones; \
|
|
|
|
zone; \
|
|
|
|
zone = next_zone(zone))
|
|
|
|
|
2009-03-31 22:19:31 +00:00
|
|
|
#define for_each_populated_zone(zone) \
|
|
|
|
for (zone = (first_online_pgdat())->node_zones; \
|
|
|
|
zone; \
|
|
|
|
zone = next_zone(zone)) \
|
|
|
|
if (!populated_zone(zone)) \
|
|
|
|
; /* do nothing */ \
|
|
|
|
else
|
|
|
|
|
2008-04-28 09:12:17 +00:00
|
|
|
static inline struct zone *zonelist_zone(struct zoneref *zoneref)
|
|
|
|
{
|
|
|
|
return zoneref->zone;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int zonelist_zone_idx(struct zoneref *zoneref)
|
|
|
|
{
|
|
|
|
return zoneref->zone_idx;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int zonelist_node_idx(struct zoneref *zoneref)
|
|
|
|
{
|
2018-08-22 04:53:32 +00:00
|
|
|
return zone_to_nid(zoneref->zone);
|
2008-04-28 09:12:17 +00:00
|
|
|
}
|
|
|
|
|
2016-05-20 00:13:30 +00:00
|
|
|
struct zoneref *__next_zones_zonelist(struct zoneref *z,
|
|
|
|
enum zone_type highest_zoneidx,
|
|
|
|
nodemask_t *nodes);
|
|
|
|
|
2008-04-28 09:12:18 +00:00
|
|
|
/**
|
|
|
|
* next_zones_zonelist - Returns the next zone at or below highest_zoneidx within the allowed nodemask using a cursor within a zonelist as a starting point
|
2021-04-30 06:02:16 +00:00
|
|
|
* @z: The cursor used as a starting point for the search
|
|
|
|
* @highest_zoneidx: The zone index of the highest zone to return
|
|
|
|
* @nodes: An optional nodemask to filter the zonelist with
|
2008-04-28 09:12:18 +00:00
|
|
|
*
|
|
|
|
* This function returns the next zone at or below a given zone index that is
|
|
|
|
* within the allowed nodemask using a cursor as the starting point for the
|
2008-09-13 09:33:19 +00:00
|
|
|
* search. The zoneref returned is a cursor that represents the current zone
|
|
|
|
* being examined. It should be advanced by one before calling
|
|
|
|
* next_zones_zonelist again.
|
2021-04-30 06:02:16 +00:00
|
|
|
*
|
|
|
|
* Return: the next zone at or below highest_zoneidx within the allowed
|
|
|
|
* nodemask using a cursor within a zonelist as a starting point
|
2008-04-28 09:12:18 +00:00
|
|
|
*/
|
2016-05-20 00:13:30 +00:00
|
|
|
static __always_inline struct zoneref *next_zones_zonelist(struct zoneref *z,
|
2008-04-28 09:12:18 +00:00
|
|
|
enum zone_type highest_zoneidx,
|
2016-05-20 00:13:30 +00:00
|
|
|
nodemask_t *nodes)
|
|
|
|
{
|
|
|
|
if (likely(!nodes && zonelist_zone_idx(z) <= highest_zoneidx))
|
|
|
|
return z;
|
|
|
|
return __next_zones_zonelist(z, highest_zoneidx, nodes);
|
|
|
|
}
|
2008-04-28 09:12:17 +00:00
|
|
|
|
2008-04-28 09:12:18 +00:00
|
|
|
/**
|
|
|
|
* first_zones_zonelist - Returns the first zone at or below highest_zoneidx within the allowed nodemask in a zonelist
|
2021-04-30 06:02:16 +00:00
|
|
|
* @zonelist: The zonelist to search for a suitable zone
|
|
|
|
* @highest_zoneidx: The zone index of the highest zone to return
|
|
|
|
* @nodes: An optional nodemask to filter the zonelist with
|
2008-04-28 09:12:18 +00:00
|
|
|
*
|
|
|
|
* This function returns the first zone at or below a given zone index that is
|
|
|
|
* within the allowed nodemask. The zoneref returned is a cursor that can be
|
2008-09-13 09:33:19 +00:00
|
|
|
* used to iterate the zonelist with next_zones_zonelist by advancing it by
|
|
|
|
* one before calling.
|
mm, page_alloc: fix check for NULL preferred_zone
Patch series "fix premature OOM regression in 4.7+ due to cpuset races".
This is v2 of my attempt to fix the recent report based on LTP cpuset
stress test [1]. The intention is to go to stable 4.9 LTSS with this,
as triggering repeated OOMs is not nice. That's why the patches try to
be not too intrusive.
Unfortunately why investigating I found that modifying the testcase to
use per-VMA policies instead of per-task policies will bring the OOM's
back, but that seems to be much older and harder to fix problem. I have
posted a RFC [2] but I believe that fixing the recent regressions has a
higher priority.
Longer-term we might try to think how to fix the cpuset mess in a better
and less error prone way. I was for example very surprised to learn,
that cpuset updates change not only task->mems_allowed, but also
nodemask of mempolicies. Until now I expected the parameter to
alloc_pages_nodemask() to be stable. I wonder why do we then treat
cpusets specially in get_page_from_freelist() and distinguish HARDWALL
etc, when there's unconditional intersection between mempolicy and
cpuset. I would expect the nodemask adjustment for saving overhead in
g_p_f(), but that clearly doesn't happen in the current form. So we
have both crazy complexity and overhead, AFAICS.
[1] https://lkml.kernel.org/r/CAFpQJXUq-JuEP=QPidy4p_=FN0rkH5Z-kfB4qBvsf6jMS87Edg@mail.gmail.com
[2] https://lkml.kernel.org/r/7c459f26-13a6-a817-e508-b65b903a8378@suse.cz
This patch (of 4):
Since commit c33d6c06f60f ("mm, page_alloc: avoid looking up the first
zone in a zonelist twice") we have a wrong check for NULL preferred_zone,
which can theoretically happen due to concurrent cpuset modification. We
check the zoneref pointer which is never NULL and we should check the zone
pointer. Also document this in first_zones_zonelist() comment per Michal
Hocko.
Fixes: c33d6c06f60f ("mm, page_alloc: avoid looking up the first zone in a zonelist twice")
Link: http://lkml.kernel.org/r/20170120103843.24587-2-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Ganapatrao Kulkarni <gpkulkarni@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-01-24 23:18:32 +00:00
|
|
|
*
|
|
|
|
* When no eligible zone is found, zoneref->zone is NULL (zoneref itself is
|
|
|
|
* never NULL). This may happen either genuinely, or due to concurrent nodemask
|
|
|
|
* update due to cpuset modification.
|
2021-04-30 06:02:16 +00:00
|
|
|
*
|
|
|
|
* Return: Zoneref pointer for the first suitable zone found
|
2008-04-28 09:12:18 +00:00
|
|
|
*/
|
2008-04-28 09:12:17 +00:00
|
|
|
static inline struct zoneref *first_zones_zonelist(struct zonelist *zonelist,
|
2008-04-28 09:12:18 +00:00
|
|
|
enum zone_type highest_zoneidx,
|
2016-05-20 00:14:10 +00:00
|
|
|
nodemask_t *nodes)
|
2008-04-28 09:12:16 +00:00
|
|
|
{
|
2016-05-20 00:14:10 +00:00
|
|
|
return next_zones_zonelist(zonelist->_zonerefs,
|
2015-02-11 23:25:47 +00:00
|
|
|
highest_zoneidx, nodes);
|
2008-04-28 09:12:16 +00:00
|
|
|
}
|
|
|
|
|
2008-04-28 09:12:18 +00:00
|
|
|
/**
|
|
|
|
* for_each_zone_zonelist_nodemask - helper macro to iterate over valid zones in a zonelist at or below a given zone index and within a nodemask
|
2021-04-30 06:02:16 +00:00
|
|
|
* @zone: The current zone in the iterator
|
|
|
|
* @z: The current pointer within zonelist->_zonerefs being iterated
|
|
|
|
* @zlist: The zonelist being iterated
|
|
|
|
* @highidx: The zone index of the highest zone to return
|
|
|
|
* @nodemask: Nodemask allowed by the allocator
|
2008-04-28 09:12:18 +00:00
|
|
|
*
|
|
|
|
* This iterator iterates though all zones at or below a given zone index and
|
|
|
|
* within a given nodemask
|
|
|
|
*/
|
|
|
|
#define for_each_zone_zonelist_nodemask(zone, z, zlist, highidx, nodemask) \
|
2016-05-20 00:14:10 +00:00
|
|
|
for (z = first_zones_zonelist(zlist, highidx, nodemask), zone = zonelist_zone(z); \
|
2008-04-28 09:12:18 +00:00
|
|
|
zone; \
|
2015-02-11 23:25:47 +00:00
|
|
|
z = next_zones_zonelist(++z, highidx, nodemask), \
|
2016-05-20 00:14:10 +00:00
|
|
|
zone = zonelist_zone(z))
|
|
|
|
|
2020-10-13 23:55:57 +00:00
|
|
|
#define for_next_zone_zonelist_nodemask(zone, z, highidx, nodemask) \
|
2016-05-20 00:14:10 +00:00
|
|
|
for (zone = z->zone; \
|
|
|
|
zone; \
|
|
|
|
z = next_zones_zonelist(++z, highidx, nodemask), \
|
|
|
|
zone = zonelist_zone(z))
|
|
|
|
|
2008-04-28 09:12:16 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* for_each_zone_zonelist - helper macro to iterate over valid zones in a zonelist at or below a given zone index
|
2021-04-30 06:02:16 +00:00
|
|
|
* @zone: The current zone in the iterator
|
|
|
|
* @z: The current pointer within zonelist->zones being iterated
|
|
|
|
* @zlist: The zonelist being iterated
|
|
|
|
* @highidx: The zone index of the highest zone to return
|
2008-04-28 09:12:16 +00:00
|
|
|
*
|
|
|
|
* This iterator iterates though all zones at or below a given zone index.
|
|
|
|
*/
|
|
|
|
#define for_each_zone_zonelist(zone, z, zlist, highidx) \
|
2008-04-28 09:12:18 +00:00
|
|
|
for_each_zone_zonelist_nodemask(zone, z, zlist, highidx, NULL)
|
2008-04-28 09:12:16 +00:00
|
|
|
|
mm/page_alloc: detect allocation forbidden by cpuset and bail out early
There was a report that starting an Ubuntu in docker while using cpuset
to bind it to movable nodes (a node only has movable zone, like a node
for hotplug or a Persistent Memory node in normal usage) will fail due
to memory allocation failure, and then OOM is involved and many other
innocent processes got killed.
It can be reproduced with command:
$ docker run -it --rm --cpuset-mems 4 ubuntu:latest bash -c "grep Mems_allowed /proc/self/status"
(where node 4 is a movable node)
runc:[2:INIT] invoked oom-killer: gfp_mask=0x500cc2(GFP_HIGHUSER|__GFP_ACCOUNT), order=0, oom_score_adj=0
CPU: 8 PID: 8291 Comm: runc:[2:INIT] Tainted: G W I E 5.8.2-0.g71b519a-default #1 openSUSE Tumbleweed (unreleased)
Hardware name: Dell Inc. PowerEdge R640/0PHYDR, BIOS 2.6.4 04/09/2020
Call Trace:
dump_stack+0x6b/0x88
dump_header+0x4a/0x1e2
oom_kill_process.cold+0xb/0x10
out_of_memory.part.0+0xaf/0x230
out_of_memory+0x3d/0x80
__alloc_pages_slowpath.constprop.0+0x954/0xa20
__alloc_pages_nodemask+0x2d3/0x300
pipe_write+0x322/0x590
new_sync_write+0x196/0x1b0
vfs_write+0x1c3/0x1f0
ksys_write+0xa7/0xe0
do_syscall_64+0x52/0xd0
entry_SYSCALL_64_after_hwframe+0x44/0xa9
Mem-Info:
active_anon:392832 inactive_anon:182 isolated_anon:0
active_file:68130 inactive_file:151527 isolated_file:0
unevictable:2701 dirty:0 writeback:7
slab_reclaimable:51418 slab_unreclaimable:116300
mapped:45825 shmem:735 pagetables:2540 bounce:0
free:159849484 free_pcp:73 free_cma:0
Node 4 active_anon:1448kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:0kB dirty:0kB writeback:0kB shmem:0kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 0kB writeback_tmp:0kB all_unreclaimable? no
Node 4 Movable free:130021408kB min:9140kB low:139160kB high:269180kB reserved_highatomic:0KB active_anon:1448kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:130023424kB managed:130023424kB mlocked:0kB kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:292kB local_pcp:84kB free_cma:0kB
lowmem_reserve[]: 0 0 0 0 0
Node 4 Movable: 1*4kB (M) 0*8kB 0*16kB 1*32kB (M) 0*64kB 0*128kB 1*256kB (M) 1*512kB (M) 1*1024kB (M) 0*2048kB 31743*4096kB (M) = 130021156kB
oom-kill:constraint=CONSTRAINT_CPUSET,nodemask=(null),cpuset=docker-9976a269caec812c134fa317f27487ee36e1129beba7278a463dd53e5fb9997b.scope,mems_allowed=4,global_oom,task_memcg=/system.slice/containerd.service,task=containerd,pid=4100,uid=0
Out of memory: Killed process 4100 (containerd) total-vm:4077036kB, anon-rss:51184kB, file-rss:26016kB, shmem-rss:0kB, UID:0 pgtables:676kB oom_score_adj:0
oom_reaper: reaped process 8248 (docker), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
oom_reaper: reaped process 2054 (node_exporter), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
oom_reaper: reaped process 1452 (systemd-journal), now anon-rss:0kB, file-rss:8564kB, shmem-rss:4kB
oom_reaper: reaped process 2146 (munin-node), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
oom_reaper: reaped process 8291 (runc:[2:INIT]), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
The reason is that in this case, the target cpuset nodes only have
movable zone, while the creation of an OS in docker sometimes needs to
allocate memory in non-movable zones (dma/dma32/normal) like
GFP_HIGHUSER, and the cpuset limit forbids the allocation, then
out-of-memory killing is involved even when normal nodes and movable
nodes both have many free memory.
The OOM killer cannot help to resolve the situation as there is no
usable memory for the request in the cpuset scope. The only reasonable
measure to take is to fail the allocation right away and have the caller
to deal with it.
So add a check for cases like this in the slowpath of allocation, and
bail out early returning NULL for the allocation.
As page allocation is one of the hottest path in kernel, this check will
hurt all users with sane cpuset configuration, add a static branch check
and detect the abnormal config in cpuset memory binding setup so that
the extra check cost in page allocation is not paid by everyone.
[thanks to Micho Hocko and David Rientjes for suggesting not handling
it inside OOM code, adding cpuset check, refining comments]
Link: https://lkml.kernel.org/r/1632481657-68112-1-git-send-email-feng.tang@intel.com
Signed-off-by: Feng Tang <feng.tang@intel.com>
Suggested-by: Michal Hocko <mhocko@suse.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Zefan Li <lizefan.x@bytedance.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 20:40:34 +00:00
|
|
|
/* Whether the 'nodes' are all movable nodes */
|
|
|
|
static inline bool movable_only_nodes(nodemask_t *nodes)
|
|
|
|
{
|
|
|
|
struct zonelist *zonelist;
|
|
|
|
struct zoneref *z;
|
|
|
|
int nid;
|
|
|
|
|
|
|
|
if (nodes_empty(*nodes))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We can chose arbitrary node from the nodemask to get a
|
|
|
|
* zonelist as they are interlinked. We just need to find
|
|
|
|
* at least one zone that can satisfy kernel allocations.
|
|
|
|
*/
|
|
|
|
nid = first_node(*nodes);
|
|
|
|
zonelist = &NODE_DATA(nid)->node_zonelists[ZONELIST_FALLBACK];
|
|
|
|
z = first_zones_zonelist(zonelist, ZONE_NORMAL, nodes);
|
|
|
|
return (!z->zone) ? true : false;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#ifdef CONFIG_SPARSEMEM
|
|
|
|
#include <asm/sparsemem.h>
|
|
|
|
#endif
|
|
|
|
|
2006-01-06 08:10:53 +00:00
|
|
|
#ifdef CONFIG_FLATMEM
|
|
|
|
#define pfn_to_nid(pfn) (0)
|
|
|
|
#endif
|
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#ifdef CONFIG_SPARSEMEM
|
|
|
|
|
|
|
|
/*
|
|
|
|
* PA_SECTION_SHIFT physical address to/from section number
|
|
|
|
* PFN_SECTION_SHIFT pfn to/from section number
|
|
|
|
*/
|
|
|
|
#define PA_SECTION_SHIFT (SECTION_SIZE_BITS)
|
|
|
|
#define PFN_SECTION_SHIFT (SECTION_SIZE_BITS - PAGE_SHIFT)
|
|
|
|
|
|
|
|
#define NR_MEM_SECTIONS (1UL << SECTIONS_SHIFT)
|
|
|
|
|
|
|
|
#define PAGES_PER_SECTION (1UL << PFN_SECTION_SHIFT)
|
|
|
|
#define PAGE_SECTION_MASK (~(PAGES_PER_SECTION-1))
|
|
|
|
|
Add a bitmap that is used to track flags affecting a block of pages
Here is the latest revision of the anti-fragmentation patches. Of particular
note in this version is special treatment of high-order atomic allocations.
Care is taken to group them together and avoid grouping pages of other types
near them. Artifical tests imply that it works. I'm trying to get the
hardware together that would allow setting up of a "real" test. If anyone
already has a setup and test that can trigger the atomic-allocation problem,
I'd appreciate a test of these patches and a report. The second major change
is that these patches will apply cleanly with patches that implement
anti-fragmentation through zones.
kernbench shows effectively no performance difference varying between -0.2%
and +2% on a variety of test machines. Success rates for huge page allocation
are dramatically increased. For example, on a ppc64 machine, the vanilla
kernel was only able to allocate 1% of memory as a hugepage and this was due
to a single hugepage reserved as min_free_kbytes. With these patches applied,
17% was allocatable as superpages. With reclaim-related fixes from Andy
Whitcroft, it was 40% and further reclaim-related improvements should increase
this further.
Changelog Since V28
o Group high-order atomic allocations together
o It is no longer required to set min_free_kbytes to 10% of memory. A value
of 16384 in most cases will be sufficient
o Now applied with zone-based anti-fragmentation
o Fix incorrect VM_BUG_ON within buffered_rmqueue()
o Reorder the stack so later patches do not back out work from earlier patches
o Fix bug were journal pages were being treated as movable
o Bias placement of non-movable pages to lower PFNs
o More agressive clustering of reclaimable pages in reactions to workloads
like updatedb that flood the size of inode caches
Changelog Since V27
o Renamed anti-fragmentation to Page Clustering. Anti-fragmentation was giving
the mistaken impression that it was the 100% solution for high order
allocations. Instead, it greatly increases the chances high-order
allocations will succeed and lays the foundation for defragmentation and
memory hot-remove to work properly
o Redefine page groupings based on ability to migrate or reclaim instead of
basing on reclaimability alone
o Get rid of spurious inits
o Per-cpu lists are no longer split up per-type. Instead the per-cpu list is
searched for a page of the appropriate type
o Added more explanation commentary
o Fix up bug in pageblock code where bitmap was used before being initalised
Changelog Since V26
o Fix double init of lists in setup_pageset
Changelog Since V25
o Fix loop order of for_each_rclmtype_order so that order of loop matches args
o gfpflags_to_rclmtype uses gfp_t instead of unsigned long
o Rename get_pageblock_type() to get_page_rclmtype()
o Fix alignment problem in move_freepages()
o Add mechanism for assigning flags to blocks of pages instead of page->flags
o On fallback, do not examine the preferred list of free pages a second time
The purpose of these patches is to reduce external fragmentation by grouping
pages of related types together. When pages are migrated (or reclaimed under
memory pressure), large contiguous pages will be freed.
This patch works by categorising allocations by their ability to migrate;
Movable - The pages may be moved with the page migration mechanism. These are
generally userspace pages.
Reclaimable - These are allocations for some kernel caches that are
reclaimable or allocations that are known to be very short-lived.
Unmovable - These are pages that are allocated by the kernel that
are not trivially reclaimed. For example, the memory allocated for a
loaded module would be in this category. By default, allocations are
considered to be of this type
HighAtomic - These are high-order allocations belonging to callers that
cannot sleep or perform any IO. In practice, this is restricted to
jumbo frame allocation for network receive. It is assumed that the
allocations are short-lived
Instead of having one MAX_ORDER-sized array of free lists in struct free_area,
there is one for each type of reclaimability. Once a 2^MAX_ORDER block of
pages is split for a type of allocation, it is added to the free-lists for
that type, in effect reserving it. Hence, over time, pages of the different
types can be clustered together.
When the preferred freelists are expired, the largest possible block is taken
from an alternative list. Buddies that are split from that large block are
placed on the preferred allocation-type freelists to mitigate fragmentation.
This implementation gives best-effort for low fragmentation in all zones.
Ideally, min_free_kbytes needs to be set to a value equal to 4 * (1 <<
(MAX_ORDER-1)) pages in most cases. This would be 16384 on x86 and x86_64 for
example.
Our tests show that about 60-70% of physical memory can be allocated on a
desktop after a few days uptime. In benchmarks and stress tests, we are
finding that 80% of memory is available as contiguous blocks at the end of the
test. To compare, a standard kernel was getting < 1% of memory as large pages
on a desktop and about 8-12% of memory as large pages at the end of stress
tests.
Following this email are 12 patches that implement thie page grouping feature.
The first patch introduces a mechanism for storing flags related to a whole
block of pages. Then allocations are split between movable and all other
allocations. Following that are patches to deal with per-cpu pages and make
the mechanism configurable. The next patch moves free pages between lists
when partially allocated blocks are used for pages of another migrate type.
The second last patch groups reclaimable kernel allocations such as inode
caches together. The final patch related to groupings keeps high-order atomic
allocations.
The last two patches are more concerned with control of fragmentation. The
second last patch biases placement of non-movable allocations towards the
start of memory. This is with a view of supporting memory hot-remove of DIMMs
with higher PFNs in the future. The biasing could be enforced a lot heavier
but it would cost. The last patch agressively clusters reclaimable pages like
inode caches together.
The fragmentation reduction strategy needs to track if pages within a block
can be moved or reclaimed so that pages are freed to the appropriate list.
This patch adds a bitmap for flags affecting a whole a MAX_ORDER block of
pages.
In non-SPARSEMEM configurations, the bitmap is stored in the struct zone and
allocated during initialisation. SPARSEMEM statically allocates the bitmap in
a struct mem_section so that bitmaps do not have to be resized during memory
hotadd. This wastes a small amount of memory per unused section (usually
sizeof(unsigned long)) but the complexity of dynamically allocating the memory
is quite high.
Additional credit to Andy Whitcroft who reviewed up an earlier implementation
of the mechanism an suggested how to make it a *lot* cleaner.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:47 +00:00
|
|
|
#define SECTION_BLOCKFLAGS_BITS \
|
2007-10-16 08:26:01 +00:00
|
|
|
((1UL << (PFN_SECTION_SHIFT - pageblock_order)) * NR_PAGEBLOCK_BITS)
|
Add a bitmap that is used to track flags affecting a block of pages
Here is the latest revision of the anti-fragmentation patches. Of particular
note in this version is special treatment of high-order atomic allocations.
Care is taken to group them together and avoid grouping pages of other types
near them. Artifical tests imply that it works. I'm trying to get the
hardware together that would allow setting up of a "real" test. If anyone
already has a setup and test that can trigger the atomic-allocation problem,
I'd appreciate a test of these patches and a report. The second major change
is that these patches will apply cleanly with patches that implement
anti-fragmentation through zones.
kernbench shows effectively no performance difference varying between -0.2%
and +2% on a variety of test machines. Success rates for huge page allocation
are dramatically increased. For example, on a ppc64 machine, the vanilla
kernel was only able to allocate 1% of memory as a hugepage and this was due
to a single hugepage reserved as min_free_kbytes. With these patches applied,
17% was allocatable as superpages. With reclaim-related fixes from Andy
Whitcroft, it was 40% and further reclaim-related improvements should increase
this further.
Changelog Since V28
o Group high-order atomic allocations together
o It is no longer required to set min_free_kbytes to 10% of memory. A value
of 16384 in most cases will be sufficient
o Now applied with zone-based anti-fragmentation
o Fix incorrect VM_BUG_ON within buffered_rmqueue()
o Reorder the stack so later patches do not back out work from earlier patches
o Fix bug were journal pages were being treated as movable
o Bias placement of non-movable pages to lower PFNs
o More agressive clustering of reclaimable pages in reactions to workloads
like updatedb that flood the size of inode caches
Changelog Since V27
o Renamed anti-fragmentation to Page Clustering. Anti-fragmentation was giving
the mistaken impression that it was the 100% solution for high order
allocations. Instead, it greatly increases the chances high-order
allocations will succeed and lays the foundation for defragmentation and
memory hot-remove to work properly
o Redefine page groupings based on ability to migrate or reclaim instead of
basing on reclaimability alone
o Get rid of spurious inits
o Per-cpu lists are no longer split up per-type. Instead the per-cpu list is
searched for a page of the appropriate type
o Added more explanation commentary
o Fix up bug in pageblock code where bitmap was used before being initalised
Changelog Since V26
o Fix double init of lists in setup_pageset
Changelog Since V25
o Fix loop order of for_each_rclmtype_order so that order of loop matches args
o gfpflags_to_rclmtype uses gfp_t instead of unsigned long
o Rename get_pageblock_type() to get_page_rclmtype()
o Fix alignment problem in move_freepages()
o Add mechanism for assigning flags to blocks of pages instead of page->flags
o On fallback, do not examine the preferred list of free pages a second time
The purpose of these patches is to reduce external fragmentation by grouping
pages of related types together. When pages are migrated (or reclaimed under
memory pressure), large contiguous pages will be freed.
This patch works by categorising allocations by their ability to migrate;
Movable - The pages may be moved with the page migration mechanism. These are
generally userspace pages.
Reclaimable - These are allocations for some kernel caches that are
reclaimable or allocations that are known to be very short-lived.
Unmovable - These are pages that are allocated by the kernel that
are not trivially reclaimed. For example, the memory allocated for a
loaded module would be in this category. By default, allocations are
considered to be of this type
HighAtomic - These are high-order allocations belonging to callers that
cannot sleep or perform any IO. In practice, this is restricted to
jumbo frame allocation for network receive. It is assumed that the
allocations are short-lived
Instead of having one MAX_ORDER-sized array of free lists in struct free_area,
there is one for each type of reclaimability. Once a 2^MAX_ORDER block of
pages is split for a type of allocation, it is added to the free-lists for
that type, in effect reserving it. Hence, over time, pages of the different
types can be clustered together.
When the preferred freelists are expired, the largest possible block is taken
from an alternative list. Buddies that are split from that large block are
placed on the preferred allocation-type freelists to mitigate fragmentation.
This implementation gives best-effort for low fragmentation in all zones.
Ideally, min_free_kbytes needs to be set to a value equal to 4 * (1 <<
(MAX_ORDER-1)) pages in most cases. This would be 16384 on x86 and x86_64 for
example.
Our tests show that about 60-70% of physical memory can be allocated on a
desktop after a few days uptime. In benchmarks and stress tests, we are
finding that 80% of memory is available as contiguous blocks at the end of the
test. To compare, a standard kernel was getting < 1% of memory as large pages
on a desktop and about 8-12% of memory as large pages at the end of stress
tests.
Following this email are 12 patches that implement thie page grouping feature.
The first patch introduces a mechanism for storing flags related to a whole
block of pages. Then allocations are split between movable and all other
allocations. Following that are patches to deal with per-cpu pages and make
the mechanism configurable. The next patch moves free pages between lists
when partially allocated blocks are used for pages of another migrate type.
The second last patch groups reclaimable kernel allocations such as inode
caches together. The final patch related to groupings keeps high-order atomic
allocations.
The last two patches are more concerned with control of fragmentation. The
second last patch biases placement of non-movable allocations towards the
start of memory. This is with a view of supporting memory hot-remove of DIMMs
with higher PFNs in the future. The biasing could be enforced a lot heavier
but it would cost. The last patch agressively clusters reclaimable pages like
inode caches together.
The fragmentation reduction strategy needs to track if pages within a block
can be moved or reclaimed so that pages are freed to the appropriate list.
This patch adds a bitmap for flags affecting a whole a MAX_ORDER block of
pages.
In non-SPARSEMEM configurations, the bitmap is stored in the struct zone and
allocated during initialisation. SPARSEMEM statically allocates the bitmap in
a struct mem_section so that bitmaps do not have to be resized during memory
hotadd. This wastes a small amount of memory per unused section (usually
sizeof(unsigned long)) but the complexity of dynamically allocating the memory
is quite high.
Additional credit to Andy Whitcroft who reviewed up an earlier implementation
of the mechanism an suggested how to make it a *lot* cleaner.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:47 +00:00
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#if (MAX_ORDER - 1 + PAGE_SHIFT) > SECTION_SIZE_BITS
|
|
|
|
#error Allocator MAX_ORDER exceeds SECTION_SIZE
|
|
|
|
#endif
|
|
|
|
|
2017-10-03 23:16:29 +00:00
|
|
|
static inline unsigned long pfn_to_section_nr(unsigned long pfn)
|
|
|
|
{
|
|
|
|
return pfn >> PFN_SECTION_SHIFT;
|
|
|
|
}
|
|
|
|
static inline unsigned long section_nr_to_pfn(unsigned long sec)
|
|
|
|
{
|
|
|
|
return sec << PFN_SECTION_SHIFT;
|
|
|
|
}
|
2011-05-25 00:12:33 +00:00
|
|
|
|
2011-05-25 00:12:51 +00:00
|
|
|
#define SECTION_ALIGN_UP(pfn) (((pfn) + PAGES_PER_SECTION - 1) & PAGE_SECTION_MASK)
|
|
|
|
#define SECTION_ALIGN_DOWN(pfn) ((pfn) & PAGE_SECTION_MASK)
|
|
|
|
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
#define SUBSECTION_SHIFT 21
|
2020-01-30 20:06:07 +00:00
|
|
|
#define SUBSECTION_SIZE (1UL << SUBSECTION_SHIFT)
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
|
|
|
|
#define PFN_SUBSECTION_SHIFT (SUBSECTION_SHIFT - PAGE_SHIFT)
|
|
|
|
#define PAGES_PER_SUBSECTION (1UL << PFN_SUBSECTION_SHIFT)
|
|
|
|
#define PAGE_SUBSECTION_MASK (~(PAGES_PER_SUBSECTION-1))
|
|
|
|
|
|
|
|
#if SUBSECTION_SHIFT > SECTION_SIZE_BITS
|
|
|
|
#error Subsection size exceeds section size
|
|
|
|
#else
|
|
|
|
#define SUBSECTIONS_PER_SECTION (1UL << (SECTION_SIZE_BITS - SUBSECTION_SHIFT))
|
|
|
|
#endif
|
|
|
|
|
2019-07-18 22:58:40 +00:00
|
|
|
#define SUBSECTION_ALIGN_UP(pfn) ALIGN((pfn), PAGES_PER_SUBSECTION)
|
|
|
|
#define SUBSECTION_ALIGN_DOWN(pfn) ((pfn) & PAGE_SUBSECTION_MASK)
|
|
|
|
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
struct mem_section_usage {
|
2020-04-07 03:07:06 +00:00
|
|
|
#ifdef CONFIG_SPARSEMEM_VMEMMAP
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
DECLARE_BITMAP(subsection_map, SUBSECTIONS_PER_SECTION);
|
2020-04-07 03:07:06 +00:00
|
|
|
#endif
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
/* See declaration of similar field in struct zone */
|
|
|
|
unsigned long pageblock_flags[0];
|
|
|
|
};
|
|
|
|
|
2019-07-18 22:58:04 +00:00
|
|
|
void subsection_map_init(unsigned long pfn, unsigned long nr_pages);
|
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
struct page;
|
mm/page_ext: resurrect struct page extending code for debugging
When we debug something, we'd like to insert some information to every
page. For this purpose, we sometimes modify struct page itself. But,
this has drawbacks. First, it requires re-compile. This makes us
hesitate to use the powerful debug feature so development process is
slowed down. And, second, sometimes it is impossible to rebuild the
kernel due to third party module dependency. At third, system behaviour
would be largely different after re-compile, because it changes size of
struct page greatly and this structure is accessed by every part of
kernel. Keeping this as it is would be better to reproduce errornous
situation.
This feature is intended to overcome above mentioned problems. This
feature allocates memory for extended data per page in certain place
rather than the struct page itself. This memory can be accessed by the
accessor functions provided by this code. During the boot process, it
checks whether allocation of huge chunk of memory is needed or not. If
not, it avoids allocating memory at all. With this advantage, we can
include this feature into the kernel in default and can avoid rebuild and
solve related problems.
Until now, memcg uses this technique. But, now, memcg decides to embed
their variable to struct page itself and it's code to extend struct page
has been removed. I'd like to use this code to develop debug feature, so
this patch resurrect it.
To help these things to work well, this patch introduces two callbacks for
clients. One is the need callback which is mandatory if user wants to
avoid useless memory allocation at boot-time. The other is optional, init
callback, which is used to do proper initialization after memory is
allocated. Detailed explanation about purpose of these functions is in
code comment. Please refer it.
Others are completely same with previous extension code in memcg.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: Jungsoo Son <jungsoo.son@lge.com>
Cc: Ingo Molnar <mingo@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>
2014-12-13 00:55:46 +00:00
|
|
|
struct page_ext;
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
struct mem_section {
|
2005-06-23 07:08:00 +00:00
|
|
|
/*
|
|
|
|
* This is, logically, a pointer to an array of struct
|
|
|
|
* pages. However, it is stored with some other magic.
|
|
|
|
* (see sparse.c::sparse_init_one_section())
|
|
|
|
*
|
2006-06-23 09:03:41 +00:00
|
|
|
* Additionally during early boot we encode node id of
|
|
|
|
* the location of the section here to guide allocation.
|
|
|
|
* (see sparse.c::memory_present())
|
|
|
|
*
|
2005-06-23 07:08:00 +00:00
|
|
|
* Making it a UL at least makes someone do a cast
|
|
|
|
* before using it wrong.
|
|
|
|
*/
|
|
|
|
unsigned long section_mem_map;
|
2007-10-16 08:25:56 +00:00
|
|
|
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
struct mem_section_usage *usage;
|
mm/page_ext: resurrect struct page extending code for debugging
When we debug something, we'd like to insert some information to every
page. For this purpose, we sometimes modify struct page itself. But,
this has drawbacks. First, it requires re-compile. This makes us
hesitate to use the powerful debug feature so development process is
slowed down. And, second, sometimes it is impossible to rebuild the
kernel due to third party module dependency. At third, system behaviour
would be largely different after re-compile, because it changes size of
struct page greatly and this structure is accessed by every part of
kernel. Keeping this as it is would be better to reproduce errornous
situation.
This feature is intended to overcome above mentioned problems. This
feature allocates memory for extended data per page in certain place
rather than the struct page itself. This memory can be accessed by the
accessor functions provided by this code. During the boot process, it
checks whether allocation of huge chunk of memory is needed or not. If
not, it avoids allocating memory at all. With this advantage, we can
include this feature into the kernel in default and can avoid rebuild and
solve related problems.
Until now, memcg uses this technique. But, now, memcg decides to embed
their variable to struct page itself and it's code to extend struct page
has been removed. I'd like to use this code to develop debug feature, so
this patch resurrect it.
To help these things to work well, this patch introduces two callbacks for
clients. One is the need callback which is mandatory if user wants to
avoid useless memory allocation at boot-time. The other is optional, init
callback, which is used to do proper initialization after memory is
allocated. Detailed explanation about purpose of these functions is in
code comment. Please refer it.
Others are completely same with previous extension code in memcg.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: Jungsoo Son <jungsoo.son@lge.com>
Cc: Ingo Molnar <mingo@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>
2014-12-13 00:55:46 +00:00
|
|
|
#ifdef CONFIG_PAGE_EXTENSION
|
|
|
|
/*
|
2016-05-20 23:58:04 +00:00
|
|
|
* If SPARSEMEM, pgdat doesn't have page_ext pointer. We use
|
mm/page_ext: resurrect struct page extending code for debugging
When we debug something, we'd like to insert some information to every
page. For this purpose, we sometimes modify struct page itself. But,
this has drawbacks. First, it requires re-compile. This makes us
hesitate to use the powerful debug feature so development process is
slowed down. And, second, sometimes it is impossible to rebuild the
kernel due to third party module dependency. At third, system behaviour
would be largely different after re-compile, because it changes size of
struct page greatly and this structure is accessed by every part of
kernel. Keeping this as it is would be better to reproduce errornous
situation.
This feature is intended to overcome above mentioned problems. This
feature allocates memory for extended data per page in certain place
rather than the struct page itself. This memory can be accessed by the
accessor functions provided by this code. During the boot process, it
checks whether allocation of huge chunk of memory is needed or not. If
not, it avoids allocating memory at all. With this advantage, we can
include this feature into the kernel in default and can avoid rebuild and
solve related problems.
Until now, memcg uses this technique. But, now, memcg decides to embed
their variable to struct page itself and it's code to extend struct page
has been removed. I'd like to use this code to develop debug feature, so
this patch resurrect it.
To help these things to work well, this patch introduces two callbacks for
clients. One is the need callback which is mandatory if user wants to
avoid useless memory allocation at boot-time. The other is optional, init
callback, which is used to do proper initialization after memory is
allocated. Detailed explanation about purpose of these functions is in
code comment. Please refer it.
Others are completely same with previous extension code in memcg.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: Jungsoo Son <jungsoo.son@lge.com>
Cc: Ingo Molnar <mingo@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>
2014-12-13 00:55:46 +00:00
|
|
|
* section. (see page_ext.h about this.)
|
|
|
|
*/
|
|
|
|
struct page_ext *page_ext;
|
|
|
|
unsigned long pad;
|
|
|
|
#endif
|
2013-07-03 22:04:44 +00:00
|
|
|
/*
|
|
|
|
* WARNING: mem_section must be a power-of-2 in size for the
|
|
|
|
* calculation and use of SECTION_ROOT_MASK to make sense.
|
|
|
|
*/
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
};
|
|
|
|
|
2005-09-03 22:54:28 +00:00
|
|
|
#ifdef CONFIG_SPARSEMEM_EXTREME
|
|
|
|
#define SECTIONS_PER_ROOT (PAGE_SIZE / sizeof (struct mem_section))
|
|
|
|
#else
|
|
|
|
#define SECTIONS_PER_ROOT 1
|
|
|
|
#endif
|
2005-09-03 22:54:26 +00:00
|
|
|
|
2005-09-03 22:54:28 +00:00
|
|
|
#define SECTION_NR_TO_ROOT(sec) ((sec) / SECTIONS_PER_ROOT)
|
2010-05-24 21:32:47 +00:00
|
|
|
#define NR_SECTION_ROOTS DIV_ROUND_UP(NR_MEM_SECTIONS, SECTIONS_PER_ROOT)
|
2005-09-03 22:54:28 +00:00
|
|
|
#define SECTION_ROOT_MASK (SECTIONS_PER_ROOT - 1)
|
2005-09-03 22:54:26 +00:00
|
|
|
|
2005-09-03 22:54:28 +00:00
|
|
|
#ifdef CONFIG_SPARSEMEM_EXTREME
|
2017-09-29 14:08:16 +00:00
|
|
|
extern struct mem_section **mem_section;
|
2005-09-03 22:54:26 +00:00
|
|
|
#else
|
2005-09-03 22:54:28 +00:00
|
|
|
extern struct mem_section mem_section[NR_SECTION_ROOTS][SECTIONS_PER_ROOT];
|
|
|
|
#endif
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
static inline unsigned long *section_to_usemap(struct mem_section *ms)
|
|
|
|
{
|
|
|
|
return ms->usage->pageblock_flags;
|
|
|
|
}
|
|
|
|
|
2005-06-23 07:08:00 +00:00
|
|
|
static inline struct mem_section *__nr_to_section(unsigned long nr)
|
|
|
|
{
|
2017-09-29 14:08:16 +00:00
|
|
|
#ifdef CONFIG_SPARSEMEM_EXTREME
|
|
|
|
if (!mem_section)
|
|
|
|
return NULL;
|
|
|
|
#endif
|
2005-09-03 22:54:28 +00:00
|
|
|
if (!mem_section[SECTION_NR_TO_ROOT(nr)])
|
|
|
|
return NULL;
|
|
|
|
return &mem_section[SECTION_NR_TO_ROOT(nr)][nr & SECTION_ROOT_MASK];
|
2005-06-23 07:08:00 +00:00
|
|
|
}
|
mm/sparsemem: introduce struct mem_section_usage
Patch series "mm: Sub-section memory hotplug support", v10.
The memory hotplug section is an arbitrary / convenient unit for memory
hotplug. 'Section-size' units have bled into the user interface
('memblock' sysfs) and can not be changed without breaking existing
userspace. The section-size constraint, while mostly benign for typical
memory hotplug, has and continues to wreak havoc with 'device-memory'
use cases, persistent memory (pmem) in particular. Recall that pmem
uses devm_memremap_pages(), and subsequently arch_add_memory(), to
allocate a 'struct page' memmap for pmem. However, it does not use the
'bottom half' of memory hotplug, i.e. never marks pmem pages online and
never exposes the userspace memblock interface for pmem. This leaves an
opening to redress the section-size constraint.
To date, the libnvdimm subsystem has attempted to inject padding to
satisfy the internal constraints of arch_add_memory(). Beyond
complicating the code, leading to bugs [2], wasting memory, and limiting
configuration flexibility, the padding hack is broken when the platform
changes this physical memory alignment of pmem from one boot to the
next. Device failure (intermittent or permanent) and physical
reconfiguration are events that can cause the platform firmware to
change the physical placement of pmem on a subsequent boot, and device
failure is an everyday event in a data-center.
It turns out that sections are only a hard requirement of the
user-facing interface for memory hotplug and with a bit more
infrastructure sub-section arch_add_memory() support can be added for
kernel internal usages like devm_memremap_pages(). Here is an analysis
of the current design assumptions in the current code and how they are
addressed in the new implementation:
Current design assumptions:
- Sections that describe boot memory (early sections) are never
unplugged / removed.
- pfn_valid(), in the CONFIG_SPARSEMEM_VMEMMAP=y, case devolves to a
valid_section() check
- __add_pages() and helper routines assume all operations occur in
PAGES_PER_SECTION units.
- The memblock sysfs interface only comprehends full sections
New design assumptions:
- Sections are instrumented with a sub-section bitmask to track (on
x86) individual 2MB sub-divisions of a 128MB section.
- Partially populated early sections can be extended with additional
sub-sections, and those sub-sections can be removed with
arch_remove_memory(). With this in place we no longer lose usable
memory capacity to padding.
- pfn_valid() is updated to look deeper than valid_section() to also
check the active-sub-section mask. This indication is in the same
cacheline as the valid_section() so the performance impact is
expected to be negligible. So far the lkp robot has not reported any
regressions.
- Outside of the core vmemmap population routines which are replaced,
other helper routines like shrink_{zone,pgdat}_span() are updated to
handle the smaller granularity. Core memory hotplug routines that
deal with online memory are not touched.
- The existing memblock sysfs user api guarantees / assumptions are not
touched since this capability is limited to !online
!memblock-sysfs-accessible sections.
Meanwhile the issue reports continue to roll in from users that do not
understand when and how the 128MB constraint will bite them. The current
implementation relied on being able to support at least one misaligned
namespace, but that immediately falls over on any moderately complex
namespace creation attempt. Beyond the initial problem of 'System RAM'
colliding with pmem, and the unsolvable problem of physical alignment
changes, Linux is now being exposed to platforms that collide pmem ranges
with other pmem ranges by default [3]. In short, devm_memremap_pages()
has pushed the venerable section-size constraint past the breaking point,
and the simplicity of section-aligned arch_add_memory() is no longer
tenable.
These patches are exposed to the kbuild robot on a subsection-v10 branch
[4], and a preview of the unit test for this functionality is available
on the 'subsection-pending' branch of ndctl [5].
[2]: https://lore.kernel.org/r/155000671719.348031.2347363160141119237.stgit@dwillia2-desk3.amr.corp.intel.com
[3]: https://github.com/pmem/ndctl/issues/76
[4]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/log/?h=subsection-v10
[5]: https://github.com/pmem/ndctl/commit/7c59b4867e1c
This patch (of 13):
Towards enabling memory hotplug to track partial population of a section,
introduce 'struct mem_section_usage'.
A pointer to a 'struct mem_section_usage' instance replaces the existing
pointer to a 'pageblock_flags' bitmap. Effectively it adds one more
'unsigned long' beyond the 'pageblock_flags' (usemap) allocation to house
a new 'subsection_map' bitmap. The new bitmap enables the memory
hot{plug,remove} implementation to act on incremental sub-divisions of a
section.
SUBSECTION_SHIFT is defined as global constant instead of per-architecture
value like SECTION_SIZE_BITS in order to allow cross-arch compatibility of
subsection users. Specifically a common subsection size allows for the
possibility that persistent memory namespace configurations be made
compatible across architectures.
The primary motivation for this functionality is to support platforms that
mix "System RAM" and "Persistent Memory" within a single section, or
multiple PMEM ranges with different mapping lifetimes within a single
section. The section restriction for hotplug has caused an ongoing saga
of hacks and bugs for devm_memremap_pages() users.
Beyond the fixups to teach existing paths how to retrieve the 'usemap'
from a section, and updates to usemap allocation path, there are no
expected behavior changes.
Link: http://lkml.kernel.org/r/156092349845.979959.73333291612799019.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> [ppc64]
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Qian Cai <cai@lca.pw>
Cc: Logan Gunthorpe <logang@deltatee.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-07-18 22:57:57 +00:00
|
|
|
extern size_t mem_section_usage_size(void);
|
2005-06-23 07:08:00 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We use the lower bits of the mem_map pointer to store
|
2018-02-01 00:20:26 +00:00
|
|
|
* a little bit of information. The pointer is calculated
|
|
|
|
* as mem_map - section_nr_to_pfn(pnum). The result is
|
|
|
|
* aligned to the minimum alignment of the two values:
|
|
|
|
* 1. All mem_map arrays are page-aligned.
|
|
|
|
* 2. section_nr_to_pfn() always clears PFN_SECTION_SHIFT
|
|
|
|
* lowest bits. PFN_SECTION_SHIFT is arch-specific
|
|
|
|
* (equal SECTION_SIZE_BITS - PAGE_SHIFT), and the
|
|
|
|
* worst combination is powerpc with 256k pages,
|
|
|
|
* which results in PFN_SECTION_SHIFT equal 6.
|
|
|
|
* To sum it up, at least 6 bits are available.
|
2005-06-23 07:08:00 +00:00
|
|
|
*/
|
2021-02-26 01:17:05 +00:00
|
|
|
#define SECTION_MARKED_PRESENT (1UL<<0)
|
|
|
|
#define SECTION_HAS_MEM_MAP (1UL<<1)
|
|
|
|
#define SECTION_IS_ONLINE (1UL<<2)
|
|
|
|
#define SECTION_IS_EARLY (1UL<<3)
|
|
|
|
#define SECTION_TAINT_ZONE_DEVICE (1UL<<4)
|
|
|
|
#define SECTION_MAP_LAST_BIT (1UL<<5)
|
|
|
|
#define SECTION_MAP_MASK (~(SECTION_MAP_LAST_BIT-1))
|
2021-09-02 21:57:07 +00:00
|
|
|
#define SECTION_NID_SHIFT 6
|
2005-06-23 07:08:00 +00:00
|
|
|
|
|
|
|
static inline struct page *__section_mem_map_addr(struct mem_section *section)
|
|
|
|
{
|
|
|
|
unsigned long map = section->section_mem_map;
|
|
|
|
map &= SECTION_MAP_MASK;
|
|
|
|
return (struct page *)map;
|
|
|
|
}
|
|
|
|
|
2007-10-16 08:24:11 +00:00
|
|
|
static inline int present_section(struct mem_section *section)
|
2005-06-23 07:08:00 +00:00
|
|
|
{
|
2005-09-03 22:54:26 +00:00
|
|
|
return (section && (section->section_mem_map & SECTION_MARKED_PRESENT));
|
2005-06-23 07:08:00 +00:00
|
|
|
}
|
|
|
|
|
2007-10-16 08:24:11 +00:00
|
|
|
static inline int present_section_nr(unsigned long nr)
|
|
|
|
{
|
|
|
|
return present_section(__nr_to_section(nr));
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int valid_section(struct mem_section *section)
|
2005-06-23 07:08:00 +00:00
|
|
|
{
|
2005-09-03 22:54:26 +00:00
|
|
|
return (section && (section->section_mem_map & SECTION_HAS_MEM_MAP));
|
2005-06-23 07:08:00 +00:00
|
|
|
}
|
|
|
|
|
2019-07-18 22:58:00 +00:00
|
|
|
static inline int early_section(struct mem_section *section)
|
|
|
|
{
|
|
|
|
return (section && (section->section_mem_map & SECTION_IS_EARLY));
|
|
|
|
}
|
|
|
|
|
2005-06-23 07:08:00 +00:00
|
|
|
static inline int valid_section_nr(unsigned long nr)
|
|
|
|
{
|
|
|
|
return valid_section(__nr_to_section(nr));
|
|
|
|
}
|
|
|
|
|
2017-07-06 22:37:56 +00:00
|
|
|
static inline int online_section(struct mem_section *section)
|
|
|
|
{
|
|
|
|
return (section && (section->section_mem_map & SECTION_IS_ONLINE));
|
|
|
|
}
|
|
|
|
|
2021-02-26 01:17:05 +00:00
|
|
|
static inline int online_device_section(struct mem_section *section)
|
|
|
|
{
|
|
|
|
unsigned long flags = SECTION_IS_ONLINE | SECTION_TAINT_ZONE_DEVICE;
|
|
|
|
|
|
|
|
return section && ((section->section_mem_map & flags) == flags);
|
|
|
|
}
|
|
|
|
|
2017-07-06 22:37:56 +00:00
|
|
|
static inline int online_section_nr(unsigned long nr)
|
|
|
|
{
|
|
|
|
return online_section(__nr_to_section(nr));
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_MEMORY_HOTPLUG
|
|
|
|
void online_mem_sections(unsigned long start_pfn, unsigned long end_pfn);
|
|
|
|
void offline_mem_sections(unsigned long start_pfn, unsigned long end_pfn);
|
|
|
|
#endif
|
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
static inline struct mem_section *__pfn_to_section(unsigned long pfn)
|
|
|
|
{
|
2005-06-23 07:08:00 +00:00
|
|
|
return __nr_to_section(pfn_to_section_nr(pfn));
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
}
|
|
|
|
|
2019-07-18 22:57:37 +00:00
|
|
|
extern unsigned long __highest_present_section_nr;
|
mm, sparsemem: break out of loops early
There are a number of times that we loop over NR_MEM_SECTIONS, looking
for section_present() on each section. But, when we have very large
physical address spaces (large MAX_PHYSMEM_BITS), NR_MEM_SECTIONS
becomes very large, making the loops quite long.
With MAX_PHYSMEM_BITS=46 and a section size of 128MB, the current loops
are 512k iterations, which we barely notice on modern hardware. But,
raising MAX_PHYSMEM_BITS higher (like we will see on systems that
support 5-level paging) makes this 64x longer and we start to notice,
especially on slower systems like simulators. A 10-second delay for
512k iterations is annoying. But, a 640- second delay is crippling.
This does not help if we have extremely sparse physical address spaces,
but those are quite rare. We expect that most of the "slow" systems
where this matters will also be quite small and non-sparse.
To fix this, we track the highest section we've ever encountered. This
lets us know when we will *never* see another section_present(), and
lets us break out of the loops earlier.
Doing the whole for_each_present_section_nr() macro is probably
overkill, but it will ensure that any future loop iterations that we
grow are more likely to be correct.
Kirrill said "It shaved almost 40 seconds from boot time in qemu with
5-level paging enabled for me".
Link: http://lkml.kernel.org/r/20170504174434.C45A4735@viggo.jf.intel.com
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Tested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-06 22:36:44 +00:00
|
|
|
|
2019-07-18 22:58:04 +00:00
|
|
|
static inline int subsection_map_index(unsigned long pfn)
|
|
|
|
{
|
|
|
|
return (pfn & ~(PAGE_SECTION_MASK)) / PAGES_PER_SUBSECTION;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_SPARSEMEM_VMEMMAP
|
|
|
|
static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
|
|
|
|
{
|
|
|
|
int idx = subsection_map_index(pfn);
|
|
|
|
|
|
|
|
return test_bit(idx, ms->usage->subsection_map);
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
|
|
|
|
{
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
ARM: 6913/1: sparsemem: allow pfn_valid to be overridden when using SPARSEMEM
In commit eb33575c ("[ARM] Double check memmap is actually valid with a
memmap has unexpected holes V2"), a new function, memmap_valid_within,
was introduced to mmzone.h so that holes in the memmap which pass
pfn_valid in SPARSEMEM configurations can be detected and avoided.
The fix to this problem checks that the pfn <-> page linkages are
correct by calculating the page for the pfn and then checking that
page_to_pfn on that page returns the original pfn. Unfortunately, in
SPARSEMEM configurations, this results in reading from the page flags to
determine the correct section. Since the memmap here has been freed,
junk is read from memory and the check is no longer robust.
In the best case, reading from /proc/pagetypeinfo will give you the
wrong answer. In the worst case, you get SEGVs, Kernel OOPses and hung
CPUs. Furthermore, ioremap implementations that use pfn_valid to
disallow the remapping of normal memory will break.
This patch allows architectures to provide their own pfn_valid function
instead of using the default implementation used by sparsemem. The
architecture-specific version is aware of the memmap state and will
return false when passed a pfn for a freed page within a valid section.
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2011-05-19 12:21:14 +00:00
|
|
|
#ifndef CONFIG_HAVE_ARCH_PFN_VALID
|
2021-07-01 01:51:13 +00:00
|
|
|
/**
|
|
|
|
* pfn_valid - check if there is a valid memory map entry for a PFN
|
|
|
|
* @pfn: the page frame number to check
|
|
|
|
*
|
|
|
|
* Check if there is a valid memory map entry aka struct page for the @pfn.
|
|
|
|
* Note, that availability of the memory map entry does not imply that
|
|
|
|
* there is actual usable memory at that @pfn. The struct page may
|
|
|
|
* represent a hole or an unusable page frame.
|
|
|
|
*
|
|
|
|
* Return: 1 for PFNs that have memory map entries and 0 otherwise
|
|
|
|
*/
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
static inline int pfn_valid(unsigned long pfn)
|
|
|
|
{
|
2019-07-18 22:58:04 +00:00
|
|
|
struct mem_section *ms;
|
|
|
|
|
2021-07-01 01:51:26 +00:00
|
|
|
/*
|
|
|
|
* Ensure the upper PAGE_SHIFT bits are clear in the
|
|
|
|
* pfn. Else it might lead to false positives when
|
|
|
|
* some of the upper bits are set, but the lower bits
|
|
|
|
* match a valid pfn.
|
|
|
|
*/
|
|
|
|
if (PHYS_PFN(PFN_PHYS(pfn)) != pfn)
|
|
|
|
return 0;
|
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
|
|
|
|
return 0;
|
2021-11-05 20:38:15 +00:00
|
|
|
ms = __pfn_to_section(pfn);
|
2019-07-18 22:58:04 +00:00
|
|
|
if (!valid_section(ms))
|
|
|
|
return 0;
|
|
|
|
/*
|
|
|
|
* Traditionally early sections always returned pfn_valid() for
|
|
|
|
* the entire section-sized span.
|
|
|
|
*/
|
|
|
|
return early_section(ms) || pfn_section_valid(ms, pfn);
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
}
|
ARM: 6913/1: sparsemem: allow pfn_valid to be overridden when using SPARSEMEM
In commit eb33575c ("[ARM] Double check memmap is actually valid with a
memmap has unexpected holes V2"), a new function, memmap_valid_within,
was introduced to mmzone.h so that holes in the memmap which pass
pfn_valid in SPARSEMEM configurations can be detected and avoided.
The fix to this problem checks that the pfn <-> page linkages are
correct by calculating the page for the pfn and then checking that
page_to_pfn on that page returns the original pfn. Unfortunately, in
SPARSEMEM configurations, this results in reading from the page flags to
determine the correct section. Since the memmap here has been freed,
junk is read from memory and the check is no longer robust.
In the best case, reading from /proc/pagetypeinfo will give you the
wrong answer. In the worst case, you get SEGVs, Kernel OOPses and hung
CPUs. Furthermore, ioremap implementations that use pfn_valid to
disallow the remapping of normal memory will break.
This patch allows architectures to provide their own pfn_valid function
instead of using the default implementation used by sparsemem. The
architecture-specific version is aware of the memmap state and will
return false when passed a pfn for a freed page within a valid section.
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2011-05-19 12:21:14 +00:00
|
|
|
#endif
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
|
2020-04-02 04:09:27 +00:00
|
|
|
static inline int pfn_in_present_section(unsigned long pfn)
|
2007-10-16 08:24:11 +00:00
|
|
|
{
|
|
|
|
if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
|
|
|
|
return 0;
|
2021-11-05 20:38:15 +00:00
|
|
|
return present_section(__pfn_to_section(pfn));
|
2007-10-16 08:24:11 +00:00
|
|
|
}
|
|
|
|
|
2020-02-04 01:34:02 +00:00
|
|
|
static inline unsigned long next_present_section_nr(unsigned long section_nr)
|
|
|
|
{
|
|
|
|
while (++section_nr <= __highest_present_section_nr) {
|
|
|
|
if (present_section_nr(section_nr))
|
|
|
|
return section_nr;
|
|
|
|
}
|
|
|
|
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
/*
|
|
|
|
* These are _only_ used during initialisation, therefore they
|
|
|
|
* can use __initdata ... They could have names to indicate
|
|
|
|
* this restriction.
|
|
|
|
*/
|
|
|
|
#ifdef CONFIG_NUMA
|
2006-01-06 08:10:54 +00:00
|
|
|
#define pfn_to_nid(pfn) \
|
|
|
|
({ \
|
|
|
|
unsigned long __pfn_to_nid_pfn = (pfn); \
|
|
|
|
page_to_nid(pfn_to_page(__pfn_to_nid_pfn)); \
|
|
|
|
})
|
2006-01-06 08:10:53 +00:00
|
|
|
#else
|
|
|
|
#define pfn_to_nid(pfn) (0)
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#endif
|
|
|
|
|
|
|
|
void sparse_init(void);
|
|
|
|
#else
|
|
|
|
#define sparse_init() do {} while (0)
|
2005-09-03 22:54:29 +00:00
|
|
|
#define sparse_index_init(_sec, _nid) do {} while (0)
|
2020-04-02 04:09:27 +00:00
|
|
|
#define pfn_in_present_section pfn_valid
|
2019-07-18 22:58:04 +00:00
|
|
|
#define subsection_map_init(_pfn, _nr_pages) do {} while (0)
|
[PATCH] sparsemem memory model
Sparsemem abstracts the use of discontiguous mem_maps[]. This kind of
mem_map[] is needed by discontiguous memory machines (like in the old
CONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem
replaces DISCONTIGMEM when enabled, and it is hoped that it can eventually
become a complete replacement.
A significant advantage over DISCONTIGMEM is that it's completely separated
from CONFIG_NUMA. When producing this patch, it became apparent in that NUMA
and DISCONTIG are often confused.
Another advantage is that sparse doesn't require each NUMA node's ranges to be
contiguous. It can handle overlapping ranges between nodes with no problems,
where DISCONTIGMEM currently throws away that memory.
Sparsemem uses an array to provide different pfn_to_page() translations for
each SECTION_SIZE area of physical memory. This is what allows the mem_map[]
to be chopped up.
In order to do quick pfn_to_page() operations, the section number of the page
is encoded in page->flags. Part of the sparsemem infrastructure enables
sharing of these bits more dynamically (at compile-time) between the
page_zone() and sparsemem operations. However, on 32-bit architectures, the
number of bits is quite limited, and may require growing the size of the
page->flags type in certain conditions. Several things might force this to
occur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of
memory), an increase in the physical address space, or an increase in the
number of used page->flags.
One thing to note is that, once sparsemem is present, the NUMA node
information no longer needs to be stored in the page->flags. It might provide
speed increases on certain platforms and will be stored there if there is
room. But, if out of room, an alternate (theoretically slower) mechanism is
used.
This patch introduces CONFIG_FLATMEM. It is used in almost all cases where
there used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM
often have to compile out the same areas of code.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Signed-off-by: Martin Bligh <mbligh@aracnet.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Yasunori Goto <y-goto@jp.fujitsu.com>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-23 07:07:54 +00:00
|
|
|
#endif /* CONFIG_SPARSEMEM */
|
|
|
|
|
2008-04-28 09:12:54 +00:00
|
|
|
#endif /* !__GENERATING_BOUNDS.H */
|
2005-04-16 22:20:36 +00:00
|
|
|
#endif /* !__ASSEMBLY__ */
|
|
|
|
#endif /* _LINUX_MMZONE_H */
|