mirrors/linux-stable
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git synced 2024-10-06 08:46:46 +00:00
Code Issues Releases Wiki Activity
linux-stable/arch/parisc/kernel/cache.c

768 lines
21 KiB
C
Raw Normal View History

parisc: remove CVS keywords Signed-off-by: Alexander Beregalov <a.beregalov@gmail.com> Acked-by: Matthew Wilcox <willy@linux.intel.com> Acked-by: Grant Grundler <grundler@parisc-linux.org> Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-04-03 01:49:22 +00:00
/*
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file "COPYING" in the main directory of this archive
* for more details.
*
[PARISC] Misc. janitorial work Fix a spelling mistake, add a KERN_INFO flag, and fix some whitespace uglies. Signed-off-by: Helge Deller <deller@parisc-linux.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-03-27 19:52:14 +00:00
* Copyright (C) 1999-2006 Helge Deller <deller@gmx.de> (07-13-1999)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
* Copyright (C) 1999 SuSE GmbH Nuernberg
* Copyright (C) 2000 Philipp Rumpf (prumpf@tux.org)
*
* Cache and TLB management
*
*/
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/seq_file.h>
#include <linux/pagemap.h>
Detach sched.h from mm.h First thing mm.h does is including sched.h solely for can_do_mlock() inline function which has "current" dereference inside. By dealing with can_do_mlock() mm.h can be detached from sched.h which is good. See below, why. This patch a) removes unconditional inclusion of sched.h from mm.h b) makes can_do_mlock() normal function in mm/mlock.c c) exports can_do_mlock() to not break compilation d) adds sched.h inclusions back to files that were getting it indirectly. e) adds less bloated headers to some files (asm/signal.h, jiffies.h) that were getting them indirectly Net result is: a) mm.h users would get less code to open, read, preprocess, parse, ... if they don't need sched.h b) sched.h stops being dependency for significant number of files: on x86_64 allmodconfig touching sched.h results in recompile of 4083 files, after patch it's only 3744 (-8.3%). Cross-compile tested on all arm defconfigs, all mips defconfigs, all powerpc defconfigs, alpha alpha-up arm i386 i386-up i386-defconfig i386-allnoconfig ia64 ia64-up m68k mips parisc parisc-up powerpc powerpc-up s390 s390-up sparc sparc-up sparc64 sparc64-up um-x86_64 x86_64 x86_64-up x86_64-defconfig x86_64-allnoconfig as well as my two usual configs. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-20 21:22:52 +00:00
#include <linux/sched.h>
sched/headers: Prepare for new header dependencies before moving more code to <linux/sched/mm.h> We are going to split more MM APIs out of <linux/sched.h>, which will have to be picked up from a couple of .c files. The APIs that we are going to move are: arch_pick_mmap_layout() arch_get_unmapped_area() arch_get_unmapped_area_topdown() mm_update_next_owner() Include the header in the files that are going to need it. Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-02-08 17:51:31 +00:00
#include <linux/sched/mm.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#include <asm/pdc.h>
#include <asm/cache.h>
#include <asm/cacheflush.h>
#include <asm/tlbflush.h>
#include <asm/page.h>
#include <asm/processor.h>
[PARISC] Fix parisc_setup_cache_timing to choose a better flush threshold update comment about CAFL_STRIDE Signed-off-by: Kyle McMartin <kyle@parisc-linux.org> Fixed a bug in parisc_setup_cache_timing() which caused it to calculate a poor value for parisc_cache_flush_threshold. Thanks to Joel Soete for spotting the bug. Thanks to James Bottomley for pointing out the clean way to fix this. Signed-off-by: Stuart Brady <sdb@parisc-linux.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-10-22 02:44:14 +00:00
#include <asm/sections.h>
parisc: flush pages through tmpalias space The kernel has an 8M tmpailas space (originally designed for copying and clearing pages but now only used for clearing). The idea is to place zeros into the cache above a physical page rather than into the physical page and flush the cache, because often the zeros end up being replaced quickly anyway. We can also use the tmpalias space for flushing a page. The difference here is that we have to do tmpalias processing in the non access data and instruction traps. The principle is the same: as long as we know the physical address and have a virtual address congruent to the real one, the flush will be effective. In order to use the tmpalias space, the icache miss path has to be enhanced to check for the alias region to make the fic instruction effective. Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-12-22 16:22:11 +00:00
#include <asm/shmparam.h>
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
#include <asm/mmu_context.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Use __ro_after_init in cache.c Signed-off-by: Helge Deller <deller@gmx.de>
2019-05-10 18:47:11 +00:00
int split_tlb __ro_after_init;
int dcache_stride __ro_after_init;
int icache_stride __ro_after_init;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
EXPORT_SYMBOL(dcache_stride);
parisc: flush pages through tmpalias space The kernel has an 8M tmpailas space (originally designed for copying and clearing pages but now only used for clearing). The idea is to place zeros into the cache above a physical page rather than into the physical page and flush the cache, because often the zeros end up being replaced quickly anyway. We can also use the tmpalias space for flushing a page. The difference here is that we have to do tmpalias processing in the non access data and instruction traps. The principle is the same: as long as we know the physical address and have a virtual address congruent to the real one, the flush will be effective. In order to use the tmpalias space, the icache miss path has to be enhanced to check for the alias region to make the fic instruction effective. Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-12-22 16:22:11 +00:00
void flush_dcache_page_asm(unsigned long phys_addr, unsigned long vaddr);
EXPORT_SYMBOL(flush_dcache_page_asm);
parisc: Optimze cache flush algorithms The attached patch implements three optimizations: 1) Loops in flush_user_dcache_range_asm, flush_kernel_dcache_range_asm, purge_kernel_dcache_range_asm, flush_user_icache_range_asm, and flush_kernel_icache_range_asm are unrolled to reduce branch overhead. 2) The static branch prediction for cmpb instructions in pacache.S have been reviewed and the operand order adjusted where necessary. 3) For flush routines in cache.c, we purge rather flush when we have no context. The pdc instruction at level 0 is not required to write back dirty lines to memory. This provides a performance improvement over the fdc instruction if the feature is implemented. Version 2 adds alternative patching. The patch provides an average improvement of about 2%. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-20 00:48:12 +00:00
void purge_dcache_page_asm(unsigned long phys_addr, unsigned long vaddr);
parisc: flush pages through tmpalias space The kernel has an 8M tmpailas space (originally designed for copying and clearing pages but now only used for clearing). The idea is to place zeros into the cache above a physical page rather than into the physical page and flush the cache, because often the zeros end up being replaced quickly anyway. We can also use the tmpalias space for flushing a page. The difference here is that we have to do tmpalias processing in the non access data and instruction traps. The principle is the same: as long as we know the physical address and have a virtual address congruent to the real one, the flush will be effective. In order to use the tmpalias space, the icache miss path has to be enhanced to check for the alias region to make the fic instruction effective. Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-12-22 16:22:11 +00:00
void flush_icache_page_asm(unsigned long phys_addr, unsigned long vaddr);
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
/* Internal implementation in arch/parisc/kernel/pacache.S */
void flush_data_cache_local(void *); /* flushes local data-cache only */
void flush_instruction_cache_local(void); /* flushes local code-cache only */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Use per-pagetable spinlock PA-RISC uses a global spinlock to protect pagetable updates in the TLB fault handlers. When multiple cores are taking TLB faults simultaneously, the cache line containing the spinlock becomes a bottleneck. This patch embeds the spinlock in the top level page directory, so that every process has its own lock. It improves performance by 30% when doing parallel compilations. At least on the N class systems, only one PxTLB inter processor broadcast can be active at any one time on the Merced bus. If a Merced bus is found, this patch serializes the TLB flushes with the pa_tlb_flush_lock spinlock. v1: Initial patch by Mikulas v2: Added Merced detection by Helge v3: Revised TLB serialization by Dave & Helge Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2019-04-27 22:09:53 +00:00
/* On some machines (i.e., ones with the Merced bus), there can be
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
* only a single PxTLB broadcast at a time; this must be guaranteed
parisc: Use per-pagetable spinlock PA-RISC uses a global spinlock to protect pagetable updates in the TLB fault handlers. When multiple cores are taking TLB faults simultaneously, the cache line containing the spinlock becomes a bottleneck. This patch embeds the spinlock in the top level page directory, so that every process has its own lock. It improves performance by 30% when doing parallel compilations. At least on the N class systems, only one PxTLB inter processor broadcast can be active at any one time on the Merced bus. If a Merced bus is found, this patch serializes the TLB flushes with the pa_tlb_flush_lock spinlock. v1: Initial patch by Mikulas v2: Added Merced detection by Helge v3: Revised TLB serialization by Dave & Helge Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2019-04-27 22:09:53 +00:00
* by software. We need a spinlock around all TLB flushes to ensure
* this.
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*/
parisc: Use per-pagetable spinlock PA-RISC uses a global spinlock to protect pagetable updates in the TLB fault handlers. When multiple cores are taking TLB faults simultaneously, the cache line containing the spinlock becomes a bottleneck. This patch embeds the spinlock in the top level page directory, so that every process has its own lock. It improves performance by 30% when doing parallel compilations. At least on the N class systems, only one PxTLB inter processor broadcast can be active at any one time on the Merced bus. If a Merced bus is found, this patch serializes the TLB flushes with the pa_tlb_flush_lock spinlock. v1: Initial patch by Mikulas v2: Added Merced detection by Helge v3: Revised TLB serialization by Dave & Helge Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2019-04-27 22:09:53 +00:00
DEFINE_SPINLOCK(pa_tlb_flush_lock);
/* Swapper page setup lock. */
DEFINE_SPINLOCK(pa_swapper_pg_lock);
#if defined(CONFIG_64BIT) && defined(CONFIG_SMP)
parisc: Use __ro_after_init in cache.c Signed-off-by: Helge Deller <deller@gmx.de>
2019-05-10 18:47:11 +00:00
int pa_serialize_tlb_flushes __ro_after_init;
parisc: Use per-pagetable spinlock PA-RISC uses a global spinlock to protect pagetable updates in the TLB fault handlers. When multiple cores are taking TLB faults simultaneously, the cache line containing the spinlock becomes a bottleneck. This patch embeds the spinlock in the top level page directory, so that every process has its own lock. It improves performance by 30% when doing parallel compilations. At least on the N class systems, only one PxTLB inter processor broadcast can be active at any one time on the Merced bus. If a Merced bus is found, this patch serializes the TLB flushes with the pa_tlb_flush_lock spinlock. v1: Initial patch by Mikulas v2: Added Merced detection by Helge v3: Revised TLB serialization by Dave & Helge Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2019-04-27 22:09:53 +00:00
#endif
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Use __ro_after_init in cache.c Signed-off-by: Helge Deller <deller@gmx.de>
2019-05-10 18:47:11 +00:00
struct pdc_cache_info cache_info __ro_after_init;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#ifndef CONFIG_PA20
parisc: Use __ro_after_init in cache.c Signed-off-by: Helge Deller <deller@gmx.de>
2019-05-10 18:47:11 +00:00
static struct pdc_btlb_info btlb_info __ro_after_init;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
DEFINE_STATIC_KEY_TRUE(parisc_has_cache);
DEFINE_STATIC_KEY_TRUE(parisc_has_dcache);
DEFINE_STATIC_KEY_TRUE(parisc_has_icache);
static void cache_flush_local_cpu(void *dummy)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
if (static_branch_likely(&parisc_has_icache))
flush_instruction_cache_local();
if (static_branch_likely(&parisc_has_dcache))
flush_data_cache_local(NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
void flush_cache_all_local(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
cache_flush_local_cpu(NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
void flush_cache_all(void)
{
if (static_branch_likely(&parisc_has_cache))
on_each_cpu(cache_flush_local_cpu, NULL, 1);
}
static inline void flush_data_cache(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
if (static_branch_likely(&parisc_has_dcache))
on_each_cpu(flush_data_cache_local, NULL, 1);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
parisc: Avoid calling SMP cache flush functions on cache-less machines At least the qemu virtual machine does not provide D- and I-caches, so skip triggering SMP irqs to flush caches on such machines. Further optimize the caching code by using static branches and making some functions static. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-12 20:03:33 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
/* Kernel virtual address of pfn. */
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
#define pfn_va(pfn) __va(PFN_PHYS(pfn))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
void
parisc: Flush kernel data mapping in set_pte_at() when installing pte for user page For years, there have been random segmentation faults in userspace on SMP PA-RISC machines. It occurred to me that this might be a problem in set_pte_at(). MIPS and some other architectures do cache flushes when installing PTEs with the present bit set. Here I have adapted the code in update_mmu_cache() to flush the kernel mapping when the kernel flush is deferred, or when the kernel mapping may alias with the user mapping. This simplifies calls to update_mmu_cache(). I also changed the barrier in set_pte() from a compiler barrier to a full memory barrier. I know this change is not sufficient to fix the problem. It might not be needed. I have had a few days of operation with 5.14.16 to 5.15.1 and haven't seen any random segmentation faults on rp3440 or c8000 so far. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de> Cc: stable@kernel.org # 5.12+
2021-11-08 21:48:16 +00:00
__update_cache(pte_t pte)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
parisc: Flush kernel data mapping in set_pte_at() when installing pte for user page For years, there have been random segmentation faults in userspace on SMP PA-RISC machines. It occurred to me that this might be a problem in set_pte_at(). MIPS and some other architectures do cache flushes when installing PTEs with the present bit set. Here I have adapted the code in update_mmu_cache() to flush the kernel mapping when the kernel flush is deferred, or when the kernel mapping may alias with the user mapping. This simplifies calls to update_mmu_cache(). I also changed the barrier in set_pte() from a compiler barrier to a full memory barrier. I know this change is not sufficient to fix the problem. It might not be needed. I have had a few days of operation with 5.14.16 to 5.15.1 and haven't seen any random segmentation faults on rp3440 or c8000 so far. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de> Cc: stable@kernel.org # 5.12+
2021-11-08 21:48:16 +00:00
unsigned long pfn = pte_pfn(pte);
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
struct page *page;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
/* We don't have pte special. As a result, we can be called with
an invalid pfn and we don't need to flush the kernel dcache page.
This occurs with FireGL card in C8000. */
if (!pfn_valid(pfn))
return;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
page = pfn_to_page(pfn);
mm: fix races between swapoff and flush dcache Thanks to commit 4b3ef9daa4fc ("mm/swap: split swap cache into 64MB trunks"), after swapoff the address_space associated with the swap device will be freed. So page_mapping() users which may touch the address_space need some kind of mechanism to prevent the address_space from being freed during accessing. The dcache flushing functions (flush_dcache_page(), etc) in architecture specific code may access the address_space of swap device for anonymous pages in swap cache via page_mapping() function. But in some cases there are no mechanisms to prevent the swap device from being swapoff, for example, CPU1 CPU2 __get_user_pages() swapoff() flush_dcache_page() mapping = page_mapping() ... exit_swap_address_space() ... kvfree(spaces) mapping_mapped(mapping) The address space may be accessed after being freed. But from cachetlb.txt and Russell King, flush_dcache_page() only care about file cache pages, for anonymous pages, flush_anon_page() should be used. The implementation of flush_dcache_page() in all architectures follows this too. They will check whether page_mapping() is NULL and whether mapping_mapped() is true to determine whether to flush the dcache immediately. And they will use interval tree (mapping->i_mmap) to find all user space mappings. While mapping_mapped() and mapping->i_mmap isn't used by anonymous pages in swap cache at all. So, to fix the race between swapoff and flush dcache, __page_mapping() is add to return the address_space for file cache pages and NULL otherwise. All page_mapping() invoking in flush dcache functions are replaced with page_mapping_file(). [akpm@linux-foundation.org: simplify page_mapping_file(), per Mike] Link: http://lkml.kernel.org/r/20180305083634.15174-1-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Andrew Morton <akpm@linux-foundation.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Chen Liqin <liqin.linux@gmail.com> Cc: Russell King <linux@armlinux.org.uk> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: Guan Xuetao <gxt@mprc.pku.edu.cn> Cc: "David S. Miller" <davem@davemloft.net> Cc: Chris Zankel <chris@zankel.net> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Ley Foon Tan <lftan@altera.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Andi Kleen <ak@linux.intel.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-04-05 23:24:39 +00:00
if (page_mapping_file(page) &&
test_bit(PG_dcache_dirty, &page->flags)) {
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
flush_kernel_dcache_page_addr(pfn_va(pfn));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
clear_bit(PG_dcache_dirty, &page->flags);
[PARISC] parisc specific kmap API implementation for pa8800 This patch fixes the pa8800 at a gross level (there are still other subtle incoherency issues which can still cause crashes and HPMCs). What it does is try to force eject inequivalent aliases before they become visible to the L2 cache (which is where we get the incoherence problems). A new function (parisc_requires_coherency) is introduced in asm/processor.h to identify the pa8x00 processors (8800 and 8900) which have the issue. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-08-23 16:00:04 +00:00
} else if (parisc_requires_coherency())
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
flush_kernel_dcache_page_addr(pfn_va(pfn));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
void
show_cache_info(struct seq_file *m)
{
[PARISC] Match show_cache_info with reality show_cache_info and struct pdc_cache_cf were out of sync with published documentation. Fix the reporting of cache associativity and update the pdc_cache_cf bitfields to match documentation. Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-14 20:26:25 +00:00
char buf[32];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
seq_printf(m, "I-cache\t\t: %ld KB\n",
cache_info.ic_size/1024 );
[PARISC] Fixes /proc/cpuinfo cache output on B160L Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-23 20:24:20 +00:00
if (cache_info.dc_loop != 1)
[PARISC] Match show_cache_info with reality show_cache_info and struct pdc_cache_cf were out of sync with published documentation. Fix the reporting of cache associativity and update the pdc_cache_cf bitfields to match documentation. Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-14 20:26:25 +00:00
snprintf(buf, 32, "%lu-way associative", cache_info.dc_loop);
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
seq_printf(m, "D-cache\t\t: %ld KB (%s%s, %s, alias=%d)\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
cache_info.dc_size/1024,
(cache_info.dc_conf.cc_wt ? "WT":"WB"),
(cache_info.dc_conf.cc_sh ? ", shared I/D":""),
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
((cache_info.dc_loop == 1) ? "direct mapped" : buf),
cache_info.dc_conf.cc_alias
);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
seq_printf(m, "ITLB entries\t: %ld\n" "DTLB entries\t: %ld%s\n",
cache_info.it_size,
cache_info.dt_size,
cache_info.dt_conf.tc_sh ? " - shared with ITLB":""
);
#ifndef CONFIG_PA20
/* BTLB - Block TLB */
if (btlb_info.max_size==0) {
seq_printf(m, "BTLB\t\t: not supported\n" );
} else {
seq_printf(m,
"BTLB fixed\t: max. %d pages, pagesize=%d (%dMB)\n"
"BTLB fix-entr.\t: %d instruction, %d data (%d combined)\n"
"BTLB var-entr.\t: %d instruction, %d data (%d combined)\n",
btlb_info.max_size, (int)4096,
btlb_info.max_size>>8,
btlb_info.fixed_range_info.num_i,
btlb_info.fixed_range_info.num_d,
btlb_info.fixed_range_info.num_comb,
btlb_info.variable_range_info.num_i,
btlb_info.variable_range_info.num_d,
btlb_info.variable_range_info.num_comb
);
}
#endif
}
void __init
parisc_cache_init(void)
{
if (pdc_cache_info(&cache_info) < 0)
panic("parisc_cache_init: pdc_cache_info failed");
#if 0
printk("ic_size %lx dc_size %lx it_size %lx\n",
cache_info.ic_size,
cache_info.dc_size,
cache_info.it_size);
printk("DC base 0x%lx stride 0x%lx count 0x%lx loop 0x%lx\n",
cache_info.dc_base,
cache_info.dc_stride,
cache_info.dc_count,
cache_info.dc_loop);
printk("dc_conf = 0x%lx alias %d blk %d line %d shift %d\n",
*(unsigned long *) (&cache_info.dc_conf),
cache_info.dc_conf.cc_alias,
cache_info.dc_conf.cc_block,
cache_info.dc_conf.cc_line,
cache_info.dc_conf.cc_shift);
[PARISC] Match show_cache_info with reality show_cache_info and struct pdc_cache_cf were out of sync with published documentation. Fix the reporting of cache associativity and update the pdc_cache_cf bitfields to match documentation. Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-14 20:26:25 +00:00
printk(" wt %d sh %d cst %d hv %d\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
cache_info.dc_conf.cc_wt,
cache_info.dc_conf.cc_sh,
cache_info.dc_conf.cc_cst,
[PARISC] Match show_cache_info with reality show_cache_info and struct pdc_cache_cf were out of sync with published documentation. Fix the reporting of cache associativity and update the pdc_cache_cf bitfields to match documentation. Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-14 20:26:25 +00:00
cache_info.dc_conf.cc_hv);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk("IC base 0x%lx stride 0x%lx count 0x%lx loop 0x%lx\n",
cache_info.ic_base,
cache_info.ic_stride,
cache_info.ic_count,
cache_info.ic_loop);
parisc: Imporove debug info about space registers and TLB configuration Signed-off-by: Helge Deller <deller@gmx.de>
2015-11-25 21:43:45 +00:00
printk("IT base 0x%lx stride 0x%lx count 0x%lx loop 0x%lx off_base 0x%lx off_stride 0x%lx off_count 0x%lx\n",
cache_info.it_sp_base,
cache_info.it_sp_stride,
cache_info.it_sp_count,
cache_info.it_loop,
cache_info.it_off_base,
cache_info.it_off_stride,
cache_info.it_off_count);
printk("DT base 0x%lx stride 0x%lx count 0x%lx loop 0x%lx off_base 0x%lx off_stride 0x%lx off_count 0x%lx\n",
cache_info.dt_sp_base,
cache_info.dt_sp_stride,
cache_info.dt_sp_count,
cache_info.dt_loop,
cache_info.dt_off_base,
cache_info.dt_off_stride,
cache_info.dt_off_count);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk("ic_conf = 0x%lx alias %d blk %d line %d shift %d\n",
*(unsigned long *) (&cache_info.ic_conf),
cache_info.ic_conf.cc_alias,
cache_info.ic_conf.cc_block,
cache_info.ic_conf.cc_line,
cache_info.ic_conf.cc_shift);
[PARISC] Match show_cache_info with reality show_cache_info and struct pdc_cache_cf were out of sync with published documentation. Fix the reporting of cache associativity and update the pdc_cache_cf bitfields to match documentation. Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-14 20:26:25 +00:00
printk(" wt %d sh %d cst %d hv %d\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
cache_info.ic_conf.cc_wt,
cache_info.ic_conf.cc_sh,
cache_info.ic_conf.cc_cst,
[PARISC] Match show_cache_info with reality show_cache_info and struct pdc_cache_cf were out of sync with published documentation. Fix the reporting of cache associativity and update the pdc_cache_cf bitfields to match documentation. Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-14 20:26:25 +00:00
cache_info.ic_conf.cc_hv);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Imporove debug info about space registers and TLB configuration Signed-off-by: Helge Deller <deller@gmx.de>
2015-11-25 21:43:45 +00:00
printk("D-TLB conf: sh %d page %d cst %d aid %d sr %d\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
cache_info.dt_conf.tc_sh,
cache_info.dt_conf.tc_page,
cache_info.dt_conf.tc_cst,
cache_info.dt_conf.tc_aid,
parisc: Imporove debug info about space registers and TLB configuration Signed-off-by: Helge Deller <deller@gmx.de>
2015-11-25 21:43:45 +00:00
cache_info.dt_conf.tc_sr);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Imporove debug info about space registers and TLB configuration Signed-off-by: Helge Deller <deller@gmx.de>
2015-11-25 21:43:45 +00:00
printk("I-TLB conf: sh %d page %d cst %d aid %d sr %d\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
cache_info.it_conf.tc_sh,
cache_info.it_conf.tc_page,
cache_info.it_conf.tc_cst,
cache_info.it_conf.tc_aid,
parisc: Imporove debug info about space registers and TLB configuration Signed-off-by: Helge Deller <deller@gmx.de>
2015-11-25 21:43:45 +00:00
cache_info.it_conf.tc_sr);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
split_tlb = 0;
if (cache_info.dt_conf.tc_sh == 0 || cache_info.dt_conf.tc_sh == 2) {
if (cache_info.dt_conf.tc_sh == 2)
printk(KERN_WARNING "Unexpected TLB configuration. "
"Will flush I/D separately (could be optimized).\n");
split_tlb = 1;
}
/* "New and Improved" version from Jim Hull
* (1 << (cc_block-1)) * (cc_line << (4 + cnf.cc_shift))
[PARISC] Fix parisc_setup_cache_timing to choose a better flush threshold update comment about CAFL_STRIDE Signed-off-by: Kyle McMartin <kyle@parisc-linux.org> Fixed a bug in parisc_setup_cache_timing() which caused it to calculate a poor value for parisc_cache_flush_threshold. Thanks to Joel Soete for spotting the bug. Thanks to James Bottomley for pointing out the clean way to fix this. Signed-off-by: Stuart Brady <sdb@parisc-linux.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-10-22 02:44:14 +00:00
* The following CAFL_STRIDE is an optimized version, see
* http://lists.parisc-linux.org/pipermail/parisc-linux/2004-June/023625.html
* http://lists.parisc-linux.org/pipermail/parisc-linux/2004-June/023671.html
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*/
#define CAFL_STRIDE(cnf) (cnf.cc_line << (3 + cnf.cc_block + cnf.cc_shift))
dcache_stride = CAFL_STRIDE(cache_info.dc_conf);
icache_stride = CAFL_STRIDE(cache_info.ic_conf);
#undef CAFL_STRIDE
#ifndef CONFIG_PA20
if (pdc_btlb_info(&btlb_info) < 0) {
memset(&btlb_info, 0, sizeof btlb_info);
}
#endif
if ((boot_cpu_data.pdc.capabilities & PDC_MODEL_NVA_MASK) ==
PDC_MODEL_NVA_UNSUPPORTED) {
printk(KERN_WARNING "parisc_cache_init: Only equivalent aliasing supported!\n");
#if 0
panic("SMP kernel required to avoid non-equivalent aliasing");
#endif
}
}
parisc: Move CPU startup-related functions into .text section If CONFIG_HOTPLUG_CPU is enabled, those functions will be run again after bootup. So they need to reside in the .text section. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-26 14:10:54 +00:00
void disable_sr_hashing(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
[PARISC] Ensure Space ID hashing is turned off Check PDC_CACHE to see if spaceid hashing is turned on, and fail to boot if that is the case. However, some old machines do not implement the PDC_CACHE_RET_SPID firmware call, so continue to boot if the call fails because of PDC_BAD_OPTION (but fail in all other error returns). Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-16 22:20:00 +00:00
int srhash_type, retval;
unsigned long space_bits;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
switch (boot_cpu_data.cpu_type) {
case pcx: /* We shouldn't get this far. setup.c should prevent it. */
BUG();
return;
case pcxs:
case pcxt:
case pcxt_:
srhash_type = SRHASH_PCXST;
break;
case pcxl:
srhash_type = SRHASH_PCXL;
break;
case pcxl2: /* pcxl2 doesn't support space register hashing */
return;
default: /* Currently all PA2.0 machines use the same ins. sequence */
srhash_type = SRHASH_PA20;
break;
}
disable_sr_hashing_asm(srhash_type);
[PARISC] Ensure Space ID hashing is turned off Check PDC_CACHE to see if spaceid hashing is turned on, and fail to boot if that is the case. However, some old machines do not implement the PDC_CACHE_RET_SPID firmware call, so continue to boot if the call fails because of PDC_BAD_OPTION (but fail in all other error returns). Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-06-16 22:20:00 +00:00
retval = pdc_spaceid_bits(&space_bits);
/* If this procedure isn't implemented, don't panic. */
if (retval < 0 && retval != PDC_BAD_OPTION)
panic("pdc_spaceid_bits call failed.\n");
if (space_bits != 0)
panic("SpaceID hashing is still on!\n");
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
static inline void
parisc: flush pages through tmpalias space The kernel has an 8M tmpailas space (originally designed for copying and clearing pages but now only used for clearing). The idea is to place zeros into the cache above a physical page rather than into the physical page and flush the cache, because often the zeros end up being replaced quickly anyway. We can also use the tmpalias space for flushing a page. The difference here is that we have to do tmpalias processing in the non access data and instruction traps. The principle is the same: as long as we know the physical address and have a virtual address congruent to the real one, the flush will be effective. In order to use the tmpalias space, the icache miss path has to be enhanced to check for the alias region to make the fic instruction effective. Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-12-22 16:22:11 +00:00
__flush_cache_page(struct vm_area_struct *vma, unsigned long vmaddr,
unsigned long physaddr)
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
{
parisc: Avoid flushing cache on cache-less machines Avoid flushing caches in __flush_cache_page() and __purge_cache_page() if the machine hasn't data or instruction caches - as e.g. in qemu. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-18 21:59:59 +00:00
if (!static_branch_likely(&parisc_has_cache))
return;
parisc: disable preemption while flushing D- or I-caches through TMPALIAS region It is necessary to disable preemption during cache flushes done through the TMPALIAS region to ensure that the TLB setup is not clobbered by another flush. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2013-02-02 23:41:24 +00:00
preempt_disable();
parisc: flush pages through tmpalias space The kernel has an 8M tmpailas space (originally designed for copying and clearing pages but now only used for clearing). The idea is to place zeros into the cache above a physical page rather than into the physical page and flush the cache, because often the zeros end up being replaced quickly anyway. We can also use the tmpalias space for flushing a page. The difference here is that we have to do tmpalias processing in the non access data and instruction traps. The principle is the same: as long as we know the physical address and have a virtual address congruent to the real one, the flush will be effective. In order to use the tmpalias space, the icache miss path has to be enhanced to check for the alias region to make the fic instruction effective. Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-12-22 16:22:11 +00:00
flush_dcache_page_asm(physaddr, vmaddr);
if (vma->vm_flags & VM_EXEC)
flush_icache_page_asm(physaddr, vmaddr);
parisc: disable preemption while flushing D- or I-caches through TMPALIAS region It is necessary to disable preemption during cache flushes done through the TMPALIAS region to ensure that the TLB setup is not clobbered by another flush. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2013-02-02 23:41:24 +00:00
preempt_enable();
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
static void flush_user_cache_page(struct vm_area_struct *vma, unsigned long vmaddr)
parisc: Optimze cache flush algorithms The attached patch implements three optimizations: 1) Loops in flush_user_dcache_range_asm, flush_kernel_dcache_range_asm, purge_kernel_dcache_range_asm, flush_user_icache_range_asm, and flush_kernel_icache_range_asm are unrolled to reduce branch overhead. 2) The static branch prediction for cmpb instructions in pacache.S have been reviewed and the operand order adjusted where necessary. 3) For flush routines in cache.c, we purge rather flush when we have no context. The pdc instruction at level 0 is not required to write back dirty lines to memory. This provides a performance improvement over the fdc instruction if the feature is implemented. Version 2 adds alternative patching. The patch provides an average improvement of about 2%. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-20 00:48:12 +00:00
{
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
unsigned long flags, space, pgd, prot;
#ifdef CONFIG_TLB_PTLOCK
unsigned long pgd_lock;
#endif
vmaddr &= PAGE_MASK;
parisc: Optimze cache flush algorithms The attached patch implements three optimizations: 1) Loops in flush_user_dcache_range_asm, flush_kernel_dcache_range_asm, purge_kernel_dcache_range_asm, flush_user_icache_range_asm, and flush_kernel_icache_range_asm are unrolled to reduce branch overhead. 2) The static branch prediction for cmpb instructions in pacache.S have been reviewed and the operand order adjusted where necessary. 3) For flush routines in cache.c, we purge rather flush when we have no context. The pdc instruction at level 0 is not required to write back dirty lines to memory. This provides a performance improvement over the fdc instruction if the feature is implemented. Version 2 adds alternative patching. The patch provides an average improvement of about 2%. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-20 00:48:12 +00:00
preempt_disable();
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
/* Set context for flush */
local_irq_save(flags);
prot = mfctl(8);
space = mfsp(SR_USER);
pgd = mfctl(25);
#ifdef CONFIG_TLB_PTLOCK
pgd_lock = mfctl(28);
#endif
switch_mm_irqs_off(NULL, vma->vm_mm, NULL);
local_irq_restore(flags);
flush_user_dcache_range_asm(vmaddr, vmaddr + PAGE_SIZE);
parisc: Optimze cache flush algorithms The attached patch implements three optimizations: 1) Loops in flush_user_dcache_range_asm, flush_kernel_dcache_range_asm, purge_kernel_dcache_range_asm, flush_user_icache_range_asm, and flush_kernel_icache_range_asm are unrolled to reduce branch overhead. 2) The static branch prediction for cmpb instructions in pacache.S have been reviewed and the operand order adjusted where necessary. 3) For flush routines in cache.c, we purge rather flush when we have no context. The pdc instruction at level 0 is not required to write back dirty lines to memory. This provides a performance improvement over the fdc instruction if the feature is implemented. Version 2 adds alternative patching. The patch provides an average improvement of about 2%. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-20 00:48:12 +00:00
if (vma->vm_flags & VM_EXEC)
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
flush_user_icache_range_asm(vmaddr, vmaddr + PAGE_SIZE);
flush_tlb_page(vma, vmaddr);
/* Restore previous context */
local_irq_save(flags);
#ifdef CONFIG_TLB_PTLOCK
mtctl(pgd_lock, 28);
#endif
mtctl(pgd, 25);
mtsp(space, SR_USER);
mtctl(prot, 8);
local_irq_restore(flags);
parisc: Optimze cache flush algorithms The attached patch implements three optimizations: 1) Loops in flush_user_dcache_range_asm, flush_kernel_dcache_range_asm, purge_kernel_dcache_range_asm, flush_user_icache_range_asm, and flush_kernel_icache_range_asm are unrolled to reduce branch overhead. 2) The static branch prediction for cmpb instructions in pacache.S have been reviewed and the operand order adjusted where necessary. 3) For flush routines in cache.c, we purge rather flush when we have no context. The pdc instruction at level 0 is not required to write back dirty lines to memory. This provides a performance improvement over the fdc instruction if the feature is implemented. Version 2 adds alternative patching. The patch provides an average improvement of about 2%. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-20 00:48:12 +00:00
preempt_enable();
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
static inline pte_t *get_ptep(struct mm_struct *mm, unsigned long addr)
{
pte_t *ptep = NULL;
pgd_t *pgd = mm->pgd;
p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
if (!pgd_none(*pgd)) {
p4d = p4d_offset(pgd, addr);
if (!p4d_none(*p4d)) {
pud = pud_offset(p4d, addr);
if (!pud_none(*pud)) {
pmd = pmd_offset(pud, addr);
if (!pmd_none(*pmd))
ptep = pte_offset_map(pmd, addr);
}
}
}
return ptep;
}
static inline bool pte_needs_flush(pte_t pte)
{
return (pte_val(pte) & (_PAGE_PRESENT | _PAGE_ACCESSED | _PAGE_NO_CACHE))
== (_PAGE_PRESENT | _PAGE_ACCESSED);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
void flush_dcache_page(struct page *page)
{
mm: fix races between swapoff and flush dcache Thanks to commit 4b3ef9daa4fc ("mm/swap: split swap cache into 64MB trunks"), after swapoff the address_space associated with the swap device will be freed. So page_mapping() users which may touch the address_space need some kind of mechanism to prevent the address_space from being freed during accessing. The dcache flushing functions (flush_dcache_page(), etc) in architecture specific code may access the address_space of swap device for anonymous pages in swap cache via page_mapping() function. But in some cases there are no mechanisms to prevent the swap device from being swapoff, for example, CPU1 CPU2 __get_user_pages() swapoff() flush_dcache_page() mapping = page_mapping() ... exit_swap_address_space() ... kvfree(spaces) mapping_mapped(mapping) The address space may be accessed after being freed. But from cachetlb.txt and Russell King, flush_dcache_page() only care about file cache pages, for anonymous pages, flush_anon_page() should be used. The implementation of flush_dcache_page() in all architectures follows this too. They will check whether page_mapping() is NULL and whether mapping_mapped() is true to determine whether to flush the dcache immediately. And they will use interval tree (mapping->i_mmap) to find all user space mappings. While mapping_mapped() and mapping->i_mmap isn't used by anonymous pages in swap cache at all. So, to fix the race between swapoff and flush dcache, __page_mapping() is add to return the address_space for file cache pages and NULL otherwise. All page_mapping() invoking in flush dcache functions are replaced with page_mapping_file(). [akpm@linux-foundation.org: simplify page_mapping_file(), per Mike] Link: http://lkml.kernel.org/r/20180305083634.15174-1-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Andrew Morton <akpm@linux-foundation.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Chen Liqin <liqin.linux@gmail.com> Cc: Russell King <linux@armlinux.org.uk> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: Guan Xuetao <gxt@mprc.pku.edu.cn> Cc: "David S. Miller" <davem@davemloft.net> Cc: Chris Zankel <chris@zankel.net> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Ley Foon Tan <lftan@altera.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Andi Kleen <ak@linux.intel.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-04-05 23:24:39 +00:00
struct address_space *mapping = page_mapping_file(page);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct vm_area_struct *mpnt;
unsigned long offset;
parisc: flush pages through tmpalias space The kernel has an 8M tmpailas space (originally designed for copying and clearing pages but now only used for clearing). The idea is to place zeros into the cache above a physical page rather than into the physical page and flush the cache, because often the zeros end up being replaced quickly anyway. We can also use the tmpalias space for flushing a page. The difference here is that we have to do tmpalias processing in the non access data and instruction traps. The principle is the same: as long as we know the physical address and have a virtual address congruent to the real one, the flush will be effective. In order to use the tmpalias space, the icache miss path has to be enhanced to check for the alias region to make the fic instruction effective. Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-12-22 16:22:11 +00:00
unsigned long addr, old_addr = 0;
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
unsigned long count = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
pgoff_t pgoff;
if (mapping && !mapping_mapped(mapping)) {
set_bit(PG_dcache_dirty, &page->flags);
return;
}
mm: remove flush_kernel_dcache_page flush_kernel_dcache_page is a rather confusing interface that implements a subset of flush_dcache_page by not being able to properly handle page cache mapped pages. The only callers left are in the exec code as all other previous callers were incorrect as they could have dealt with page cache pages. Replace the calls to flush_kernel_dcache_page with calls to flush_dcache_page, which for all architectures does either exactly the same thing, can contains one or more of the following: 1) an optimization to defer the cache flush for page cache pages not mapped into userspace 2) additional flushing for mapped page cache pages if cache aliases are possible Link: https://lkml.kernel.org/r/20210712060928.4161649-7-hch@lst.de Signed-off-by: Christoph Hellwig <hch@lst.de> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Cc: Alex Shi <alexs@kernel.org> Cc: Geoff Levand <geoff@infradead.org> Cc: Greentime Hu <green.hu@gmail.com> Cc: Guo Ren <guoren@kernel.org> Cc: Helge Deller <deller@gmx.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Nick Hu <nickhu@andestech.com> Cc: Paul Cercueil <paul@crapouillou.net> Cc: Rich Felker <dalias@libc.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: Vincent Chen <deanbo422@gmail.com> Cc: Yoshinori Sato <ysato@users.osdn.me> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-09-02 21:56:36 +00:00
flush_kernel_dcache_page_addr(page_address(page));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!mapping)
return;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
pgoff = page->index;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
/*
* We have carefully arranged in arch_get_unmapped_area() that
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
* *any* mappings of a file are always congruently mapped (whether
* declared as MAP_PRIVATE or MAP_SHARED), so we only need
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
* to flush one address here for them all to become coherent
* on machines that support equivalent aliasing
*/
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
flush_dcache_mmap_lock(mapping);
mm: replace vma prio_tree with an interval tree Implement an interval tree as a replacement for the VMA prio_tree. The algorithms are similar to lib/interval_tree.c; however that code can't be directly reused as the interval endpoints are not explicitly stored in the VMA. So instead, the common algorithm is moved into a template and the details (node type, how to get interval endpoints from the node, etc) are filled in using the C preprocessor. Once the interval tree functions are available, using them as a replacement to the VMA prio tree is a relatively simple, mechanical job. Signed-off-by: Michel Lespinasse <walken@google.com> Cc: Rik van Riel <riel@redhat.com> Cc: Hillf Danton <dhillf@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 23:31:25 +00:00
vma_interval_tree_foreach(mpnt, &mapping->i_mmap, pgoff, pgoff) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
offset = (pgoff - mpnt->vm_pgoff) << PAGE_SHIFT;
addr = mpnt->vm_start + offset;
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
if (parisc_requires_coherency()) {
pte_t *ptep;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
ptep = get_ptep(mpnt->vm_mm, addr);
if (ptep && pte_needs_flush(*ptep))
flush_user_cache_page(mpnt, addr);
} else {
/*
* The TLB is the engine of coherence on parisc:
* The CPU is entitled to speculate any page
* with a TLB mapping, so here we kill the
* mapping then flush the page along a special
* flush only alias mapping. This guarantees that
* the page is no-longer in the cache for any
* process and nor may it be speculatively read
* in (until the user or kernel specifically
* accesses it, of course)
*/
flush_tlb_page(mpnt, addr);
if (old_addr == 0 || (old_addr & (SHM_COLOUR - 1))
!= (addr & (SHM_COLOUR - 1))) {
__flush_cache_page(mpnt, addr, page_to_phys(page));
/*
* Software is allowed to have any number
* of private mappings to a page.
*/
if (!(mpnt->vm_flags & VM_SHARED))
continue;
if (old_addr)
pr_err("INEQUIVALENT ALIASES 0x%lx and 0x%lx in file %pD\n",
old_addr, addr, mpnt->vm_file);
old_addr = addr;
}
[PATCH] mm: parisc pte atomicity There's a worrying function translation_exists in parisc cacheflush.h, unaffected by split ptlock since flush_dcache_page is using it on some other mm, without any relevant lock. Oh well, make it a slightly more robust by factoring the pfn check within it. And it looked liable to confuse a camouflaged swap or file entry with a good pte: fix that too. Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 01:16:36 +00:00
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
WARN_ON(++count == 4096);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
flush_dcache_mmap_unlock(mapping);
}
EXPORT_SYMBOL(flush_dcache_page);
/* Defined in arch/parisc/kernel/pacache.S */
EXPORT_SYMBOL(flush_kernel_dcache_range_asm);
EXPORT_SYMBOL(flush_kernel_icache_range_asm);
#define FLUSH_THRESHOLD 0x80000 /* 0.5MB */
parisc: Use __ro_after_init in cache.c Signed-off-by: Helge Deller <deller@gmx.de>
2019-05-10 18:47:11 +00:00
static unsigned long parisc_cache_flush_threshold __ro_after_init = FLUSH_THRESHOLD;
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
parisc: Reorder TLB flush timing calculation On boot (mostly reboot), my c8000 sometimes crashes after it prints the TLB flush threshold. The lockup is hard. The front LED flashes red and the box must be unplugged to reset the error. I noticed that when the crash occurs the TLB flush threshold is about one quarter what it is on a successful boot. If I disabled the calculation, the crash didn't occur. There also seemed to be a timing dependency affecting the crash. I finally realized that the flush_tlb_all() timing test runs just after the secondary CPUs are started. There seems to be a problem with running flush_tlb_all() too soon after the CPUs are started. The timing for the range test always seemed okay. So, I reversed the order of the two timing tests and I haven't had a crash at this point so far. I added a couple of information messages which I have left to help with diagnosis if the problem should appear on another machine. This version reduces the minimum TLB flush threshold to 16 KiB. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-17 00:49:56 +00:00
#define FLUSH_TLB_THRESHOLD (16*1024) /* 16 KiB minimum TLB threshold */
parisc: Rewrite tlb flush threshold calculation Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2020-10-02 19:28:31 +00:00
static unsigned long parisc_tlb_flush_threshold __ro_after_init = ~0UL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
void __init parisc_setup_cache_timing(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned long rangetime, alltime;
parisc: Rewrite tlb flush threshold calculation Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2020-10-02 19:28:31 +00:00
unsigned long size;
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
unsigned long threshold, threshold2;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
alltime = mfctl(16);
flush_data_cache();
alltime = mfctl(16) - alltime;
[PARISC] Fix parisc_setup_cache_timing to choose a better flush threshold update comment about CAFL_STRIDE Signed-off-by: Kyle McMartin <kyle@parisc-linux.org> Fixed a bug in parisc_setup_cache_timing() which caused it to calculate a poor value for parisc_cache_flush_threshold. Thanks to Joel Soete for spotting the bug. Thanks to James Bottomley for pointing out the clean way to fix this. Signed-off-by: Stuart Brady <sdb@parisc-linux.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-10-22 02:44:14 +00:00
size = (unsigned long)(_end - _text);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
rangetime = mfctl(16);
[PARISC] Fix parisc_setup_cache_timing to choose a better flush threshold update comment about CAFL_STRIDE Signed-off-by: Kyle McMartin <kyle@parisc-linux.org> Fixed a bug in parisc_setup_cache_timing() which caused it to calculate a poor value for parisc_cache_flush_threshold. Thanks to Joel Soete for spotting the bug. Thanks to James Bottomley for pointing out the clean way to fix this. Signed-off-by: Stuart Brady <sdb@parisc-linux.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-10-22 02:44:14 +00:00
flush_kernel_dcache_range((unsigned long)_text, size);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
rangetime = mfctl(16) - rangetime;
printk(KERN_DEBUG "Whole cache flush %lu cycles, flushing %lu bytes %lu cycles\n",
alltime, size, rangetime);
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
threshold = L1_CACHE_ALIGN((unsigned long)((uint64_t)size * alltime / rangetime));
pr_info("Calculated flush threshold is %lu KiB\n",
threshold/1024);
/*
* The threshold computed above isn't very reliable. The following
* heuristic works reasonably well on c8000/rp3440.
*/
threshold2 = cache_info.dc_size * num_online_cpus();
parisc_cache_flush_threshold = threshold2;
parisc: Fix races in parisc_setup_cache_timing() Helge reported to me the following startup crash: [ 0.000000] Linux version 4.8.0-1-parisc64-smp (debian-kernel@lists.debian.org) (gcc version 5.4.1 20161019 (GCC) ) #1 SMP Debian 4.8.7-1 (2016-11-13) [ 0.000000] The 64-bit Kernel has started... [ 0.000000] Kernel default page size is 4 KB. Huge pages enabled with 1 MB physical and 2 MB virtual size. [ 0.000000] Determining PDC firmware type: System Map. [ 0.000000] model 9000/785/J5000 [ 0.000000] Total Memory: 2048 MB [ 0.000000] Memory: 2018528K/2097152K available (9272K kernel code, 3053K rwdata, 1319K rodata, 1024K init, 840K bss, 78624K reserved, 0K cma-reserved) [ 0.000000] virtual kernel memory layout: [ 0.000000] vmalloc : 0x0000000000008000 - 0x000000003f000000 (1007 MB) [ 0.000000] memory : 0x0000000040000000 - 0x00000000c0000000 (2048 MB) [ 0.000000] .init : 0x0000000040100000 - 0x0000000040200000 (1024 kB) [ 0.000000] .data : 0x0000000040b0e000 - 0x0000000040f533e0 (4372 kB) [ 0.000000] .text : 0x0000000040200000 - 0x0000000040b0e000 (9272 kB) [ 0.768910] Brought up 1 CPUs [ 0.992465] NET: Registered protocol family 16 [ 2.429981] Releasing cpu 1 now, hpa=fffffffffffa2000 [ 2.635751] CPU(s): 2 out of 2 PA8500 (PCX-W) at 440.000000 MHz online [ 2.726692] Setting cache flush threshold to 1024 kB [ 2.729932] Not-handled unaligned insn 0x43ffff80 [ 2.798114] Setting TLB flush threshold to 140 kB [ 2.928039] Unaligned handler failed, ret = -1 [ 3.000419] _______________________________ [ 3.000419] < Your System ate a SPARC! Gah! > [ 3.000419] ------------------------------- [ 3.000419] \ ^__^ [ 3.000419] (__)\ )\/\ [ 3.000419] U ||----w | [ 3.000419] || || [ 9.340055] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.8.0-1-parisc64-smp #1 Debian 4.8.7-1 [ 9.448082] task: 00000000bfd48060 task.stack: 00000000bfd50000 [ 9.528040] [ 10.760029] IASQ: 0000000000000000 0000000000000000 IAOQ: 000000004025d154 000000004025d158 [ 10.868052] IIR: 43ffff80 ISR: 0000000000340000 IOR: 000001ff54150960 [ 10.960029] CPU: 1 CR30: 00000000bfd50000 CR31: 0000000011111111 [ 11.052057] ORIG_R28: 000000004021e3b4 [ 11.100045] IAOQ[0]: irq_exit+0x94/0x120 [ 11.152062] IAOQ[1]: irq_exit+0x98/0x120 [ 11.208031] RP(r2): irq_exit+0xb8/0x120 [ 11.256074] Backtrace: [ 11.288067] [<00000000402cd944>] cpu_startup_entry+0x1e4/0x598 [ 11.368058] [<0000000040109528>] smp_callin+0x2c0/0x2f0 [ 11.436308] [<00000000402b53fc>] update_curr+0x18c/0x2d0 [ 11.508055] [<00000000402b73b8>] dequeue_entity+0x2c0/0x1030 [ 11.584040] [<00000000402b3cc0>] set_next_entity+0x80/0xd30 [ 11.660069] [<00000000402c1594>] pick_next_task_fair+0x614/0x720 [ 11.740085] [<000000004020dd34>] __schedule+0x394/0xa60 [ 11.808054] [<000000004020e488>] schedule+0x88/0x118 [ 11.876039] [<0000000040283d3c>] rescuer_thread+0x4d4/0x5b0 [ 11.948090] [<000000004028fc4c>] kthread+0x1ec/0x248 [ 12.016053] [<0000000040205020>] end_fault_vector+0x20/0xc0 [ 12.092239] [<00000000402050c0>] _switch_to_ret+0x0/0xf40 [ 12.164044] [ 12.184036] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.8.0-1-parisc64-smp #1 Debian 4.8.7-1 [ 12.244040] Backtrace: [ 12.244040] [<000000004021c480>] show_stack+0x68/0x80 [ 12.244040] [<00000000406f332c>] dump_stack+0xec/0x168 [ 12.244040] [<000000004021c74c>] die_if_kernel+0x25c/0x430 [ 12.244040] [<000000004022d320>] handle_unaligned+0xb48/0xb50 [ 12.244040] [ 12.632066] ---[ end trace 9ca05a7215c7bbb2 ]--- [ 12.692036] Kernel panic - not syncing: Attempted to kill the idle task! We have the insn 0x43ffff80 in IIR but from IAOQ we should have: 4025d150: 0f f3 20 df ldd,s r19(r31),r31 4025d154: 0f 9f 00 9c ldw r31(ret0),ret0 4025d158: bf 80 20 58 cmpb,*<> r0,ret0,4025d18c <irq_exit+0xcc> Cpu0 has just completed running parisc_setup_cache_timing: [ 2.429981] Releasing cpu 1 now, hpa=fffffffffffa2000 [ 2.635751] CPU(s): 2 out of 2 PA8500 (PCX-W) at 440.000000 MHz online [ 2.726692] Setting cache flush threshold to 1024 kB [ 2.729932] Not-handled unaligned insn 0x43ffff80 [ 2.798114] Setting TLB flush threshold to 140 kB [ 2.928039] Unaligned handler failed, ret = -1 From the backtrace, cpu1 is in smp_callin: void __init smp_callin(void) { int slave_id = cpu_now_booting; smp_cpu_init(slave_id); preempt_disable(); flush_cache_all_local(); /* start with known state */ flush_tlb_all_local(NULL); local_irq_enable(); /* Interrupts have been off until now */ cpu_startup_entry(CPUHP_AP_ONLINE_IDLE); So, it has just flushed its caches and the TLB. It would seem either the flushes in parisc_setup_cache_timing or smp_callin have corrupted kernel memory. The attached patch reworks parisc_setup_cache_timing to remove the races in setting the cache and TLB flush thresholds. It also corrects the number of bytes flushed in the TLB calculation. The patch flushes the cache and TLB on cpu0 before starting the secondary processors so that they are started from a known state. Tested with a few reboots on c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2016-11-21 02:12:36 +00:00
printk(KERN_INFO "Cache flush threshold set to %lu KiB\n",
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
parisc_cache_flush_threshold/1024);
/* calculate TLB flush threshold */
parisc: Fix TLB related boot crash on SMP machines At bootup we run measurements to calculate the best threshold for when we should be using full TLB flushes instead of just flushing a specific amount of TLB entries. This performance test is run over the kernel text segment. But running this TLB performance test on the kernel text segment turned out to crash some SMP machines when the kernel text pages were mapped as huge pages. To avoid those crashes this patch simply skips this test on some SMP machines and calculates an optimal threshold based on the maximum number of available TLB entries and number of online CPUs. On a technical side, this seems to happen: The TLB measurement code uses flush_tlb_kernel_range() to flush specific TLB entries with a page size of 4k (pdtlb 0(sr1,addr)). On UP systems this purge instruction seems to work without problems even if the pages were mapped as huge pages. But on SMP systems the TLB purge instruction is broadcasted to other CPUs. Those CPUs then crash the machine because the page size is not as expected. C8000 machines with PA8800/PA8900 CPUs were not affected by this problem, because the required cache coherency prohibits to use huge pages at all. Sadly I didn't found any documentation about this behaviour, so this finding is purely based on testing with phyiscal SMP machines (A500-44 and J5000, both were 2-way boxes). Cc: <stable@vger.kernel.org> # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2016-12-08 20:00:46 +00:00
/* On SMP machines, skip the TLB measure of kernel text which
* has been mapped as huge pages. */
if (num_online_cpus() > 1 && !parisc_requires_coherency()) {
threshold = max(cache_info.it_size, cache_info.dt_size);
threshold *= PAGE_SIZE;
threshold /= num_online_cpus();
goto set_tlb_threshold;
}
parisc: Rewrite tlb flush threshold calculation Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2020-10-02 19:28:31 +00:00
size = (unsigned long)_end - (unsigned long)_text;
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
rangetime = mfctl(16);
parisc: Rewrite tlb flush threshold calculation Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2020-10-02 19:28:31 +00:00
flush_tlb_kernel_range((unsigned long)_text, (unsigned long)_end);
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
rangetime = mfctl(16) - rangetime;
parisc: Reorder TLB flush timing calculation On boot (mostly reboot), my c8000 sometimes crashes after it prints the TLB flush threshold. The lockup is hard. The front LED flashes red and the box must be unplugged to reset the error. I noticed that when the crash occurs the TLB flush threshold is about one quarter what it is on a successful boot. If I disabled the calculation, the crash didn't occur. There also seemed to be a timing dependency affecting the crash. I finally realized that the flush_tlb_all() timing test runs just after the secondary CPUs are started. There seems to be a problem with running flush_tlb_all() too soon after the CPUs are started. The timing for the range test always seemed okay. So, I reversed the order of the two timing tests and I haven't had a crash at this point so far. I added a couple of information messages which I have left to help with diagnosis if the problem should appear on another machine. This version reduces the minimum TLB flush threshold to 16 KiB. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-17 00:49:56 +00:00
alltime = mfctl(16);
flush_tlb_all();
alltime = mfctl(16) - alltime;
printk(KERN_INFO "Whole TLB flush %lu cycles, Range flush %lu bytes %lu cycles\n",
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
alltime, size, rangetime);
parisc: Reorder TLB flush timing calculation On boot (mostly reboot), my c8000 sometimes crashes after it prints the TLB flush threshold. The lockup is hard. The front LED flashes red and the box must be unplugged to reset the error. I noticed that when the crash occurs the TLB flush threshold is about one quarter what it is on a successful boot. If I disabled the calculation, the crash didn't occur. There also seemed to be a timing dependency affecting the crash. I finally realized that the flush_tlb_all() timing test runs just after the secondary CPUs are started. There seems to be a problem with running flush_tlb_all() too soon after the CPUs are started. The timing for the range test always seemed okay. So, I reversed the order of the two timing tests and I haven't had a crash at this point so far. I added a couple of information messages which I have left to help with diagnosis if the problem should appear on another machine. This version reduces the minimum TLB flush threshold to 16 KiB. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2018-10-17 00:49:56 +00:00
threshold = PAGE_ALIGN((num_online_cpus() * size * alltime) / rangetime);
printk(KERN_INFO "Calculated TLB flush threshold %lu KiB\n",
threshold/1024);
parisc: Fix TLB related boot crash on SMP machines At bootup we run measurements to calculate the best threshold for when we should be using full TLB flushes instead of just flushing a specific amount of TLB entries. This performance test is run over the kernel text segment. But running this TLB performance test on the kernel text segment turned out to crash some SMP machines when the kernel text pages were mapped as huge pages. To avoid those crashes this patch simply skips this test on some SMP machines and calculates an optimal threshold based on the maximum number of available TLB entries and number of online CPUs. On a technical side, this seems to happen: The TLB measurement code uses flush_tlb_kernel_range() to flush specific TLB entries with a page size of 4k (pdtlb 0(sr1,addr)). On UP systems this purge instruction seems to work without problems even if the pages were mapped as huge pages. But on SMP systems the TLB purge instruction is broadcasted to other CPUs. Those CPUs then crash the machine because the page size is not as expected. C8000 machines with PA8800/PA8900 CPUs were not affected by this problem, because the required cache coherency prohibits to use huge pages at all. Sadly I didn't found any documentation about this behaviour, so this finding is purely based on testing with phyiscal SMP machines (A500-44 and J5000, both were 2-way boxes). Cc: <stable@vger.kernel.org> # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2016-12-08 20:00:46 +00:00
set_tlb_threshold:
parisc: Rewrite tlb flush threshold calculation Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2020-10-02 19:28:31 +00:00
if (threshold > FLUSH_TLB_THRESHOLD)
parisc: Fix races in parisc_setup_cache_timing() Helge reported to me the following startup crash: [ 0.000000] Linux version 4.8.0-1-parisc64-smp (debian-kernel@lists.debian.org) (gcc version 5.4.1 20161019 (GCC) ) #1 SMP Debian 4.8.7-1 (2016-11-13) [ 0.000000] The 64-bit Kernel has started... [ 0.000000] Kernel default page size is 4 KB. Huge pages enabled with 1 MB physical and 2 MB virtual size. [ 0.000000] Determining PDC firmware type: System Map. [ 0.000000] model 9000/785/J5000 [ 0.000000] Total Memory: 2048 MB [ 0.000000] Memory: 2018528K/2097152K available (9272K kernel code, 3053K rwdata, 1319K rodata, 1024K init, 840K bss, 78624K reserved, 0K cma-reserved) [ 0.000000] virtual kernel memory layout: [ 0.000000] vmalloc : 0x0000000000008000 - 0x000000003f000000 (1007 MB) [ 0.000000] memory : 0x0000000040000000 - 0x00000000c0000000 (2048 MB) [ 0.000000] .init : 0x0000000040100000 - 0x0000000040200000 (1024 kB) [ 0.000000] .data : 0x0000000040b0e000 - 0x0000000040f533e0 (4372 kB) [ 0.000000] .text : 0x0000000040200000 - 0x0000000040b0e000 (9272 kB) [ 0.768910] Brought up 1 CPUs [ 0.992465] NET: Registered protocol family 16 [ 2.429981] Releasing cpu 1 now, hpa=fffffffffffa2000 [ 2.635751] CPU(s): 2 out of 2 PA8500 (PCX-W) at 440.000000 MHz online [ 2.726692] Setting cache flush threshold to 1024 kB [ 2.729932] Not-handled unaligned insn 0x43ffff80 [ 2.798114] Setting TLB flush threshold to 140 kB [ 2.928039] Unaligned handler failed, ret = -1 [ 3.000419] _______________________________ [ 3.000419] < Your System ate a SPARC! Gah! > [ 3.000419] ------------------------------- [ 3.000419] \ ^__^ [ 3.000419] (__)\ )\/\ [ 3.000419] U ||----w | [ 3.000419] || || [ 9.340055] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.8.0-1-parisc64-smp #1 Debian 4.8.7-1 [ 9.448082] task: 00000000bfd48060 task.stack: 00000000bfd50000 [ 9.528040] [ 10.760029] IASQ: 0000000000000000 0000000000000000 IAOQ: 000000004025d154 000000004025d158 [ 10.868052] IIR: 43ffff80 ISR: 0000000000340000 IOR: 000001ff54150960 [ 10.960029] CPU: 1 CR30: 00000000bfd50000 CR31: 0000000011111111 [ 11.052057] ORIG_R28: 000000004021e3b4 [ 11.100045] IAOQ[0]: irq_exit+0x94/0x120 [ 11.152062] IAOQ[1]: irq_exit+0x98/0x120 [ 11.208031] RP(r2): irq_exit+0xb8/0x120 [ 11.256074] Backtrace: [ 11.288067] [<00000000402cd944>] cpu_startup_entry+0x1e4/0x598 [ 11.368058] [<0000000040109528>] smp_callin+0x2c0/0x2f0 [ 11.436308] [<00000000402b53fc>] update_curr+0x18c/0x2d0 [ 11.508055] [<00000000402b73b8>] dequeue_entity+0x2c0/0x1030 [ 11.584040] [<00000000402b3cc0>] set_next_entity+0x80/0xd30 [ 11.660069] [<00000000402c1594>] pick_next_task_fair+0x614/0x720 [ 11.740085] [<000000004020dd34>] __schedule+0x394/0xa60 [ 11.808054] [<000000004020e488>] schedule+0x88/0x118 [ 11.876039] [<0000000040283d3c>] rescuer_thread+0x4d4/0x5b0 [ 11.948090] [<000000004028fc4c>] kthread+0x1ec/0x248 [ 12.016053] [<0000000040205020>] end_fault_vector+0x20/0xc0 [ 12.092239] [<00000000402050c0>] _switch_to_ret+0x0/0xf40 [ 12.164044] [ 12.184036] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.8.0-1-parisc64-smp #1 Debian 4.8.7-1 [ 12.244040] Backtrace: [ 12.244040] [<000000004021c480>] show_stack+0x68/0x80 [ 12.244040] [<00000000406f332c>] dump_stack+0xec/0x168 [ 12.244040] [<000000004021c74c>] die_if_kernel+0x25c/0x430 [ 12.244040] [<000000004022d320>] handle_unaligned+0xb48/0xb50 [ 12.244040] [ 12.632066] ---[ end trace 9ca05a7215c7bbb2 ]--- [ 12.692036] Kernel panic - not syncing: Attempted to kill the idle task! We have the insn 0x43ffff80 in IIR but from IAOQ we should have: 4025d150: 0f f3 20 df ldd,s r19(r31),r31 4025d154: 0f 9f 00 9c ldw r31(ret0),ret0 4025d158: bf 80 20 58 cmpb,*<> r0,ret0,4025d18c <irq_exit+0xcc> Cpu0 has just completed running parisc_setup_cache_timing: [ 2.429981] Releasing cpu 1 now, hpa=fffffffffffa2000 [ 2.635751] CPU(s): 2 out of 2 PA8500 (PCX-W) at 440.000000 MHz online [ 2.726692] Setting cache flush threshold to 1024 kB [ 2.729932] Not-handled unaligned insn 0x43ffff80 [ 2.798114] Setting TLB flush threshold to 140 kB [ 2.928039] Unaligned handler failed, ret = -1 From the backtrace, cpu1 is in smp_callin: void __init smp_callin(void) { int slave_id = cpu_now_booting; smp_cpu_init(slave_id); preempt_disable(); flush_cache_all_local(); /* start with known state */ flush_tlb_all_local(NULL); local_irq_enable(); /* Interrupts have been off until now */ cpu_startup_entry(CPUHP_AP_ONLINE_IDLE); So, it has just flushed its caches and the TLB. It would seem either the flushes in parisc_setup_cache_timing or smp_callin have corrupted kernel memory. The attached patch reworks parisc_setup_cache_timing to remove the races in setting the cache and TLB flush thresholds. It also corrects the number of bytes flushed in the TLB calculation. The patch flushes the cache and TLB on cpu0 before starting the secondary processors so that they are started from a known state. Tested with a few reboots on c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2016-11-21 02:12:36 +00:00
parisc_tlb_flush_threshold = threshold;
parisc: Rewrite tlb flush threshold calculation Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2020-10-02 19:28:31 +00:00
else
parisc_tlb_flush_threshold = FLUSH_TLB_THRESHOLD;
parisc: Fix races in parisc_setup_cache_timing() Helge reported to me the following startup crash: [ 0.000000] Linux version 4.8.0-1-parisc64-smp (debian-kernel@lists.debian.org) (gcc version 5.4.1 20161019 (GCC) ) #1 SMP Debian 4.8.7-1 (2016-11-13) [ 0.000000] The 64-bit Kernel has started... [ 0.000000] Kernel default page size is 4 KB. Huge pages enabled with 1 MB physical and 2 MB virtual size. [ 0.000000] Determining PDC firmware type: System Map. [ 0.000000] model 9000/785/J5000 [ 0.000000] Total Memory: 2048 MB [ 0.000000] Memory: 2018528K/2097152K available (9272K kernel code, 3053K rwdata, 1319K rodata, 1024K init, 840K bss, 78624K reserved, 0K cma-reserved) [ 0.000000] virtual kernel memory layout: [ 0.000000] vmalloc : 0x0000000000008000 - 0x000000003f000000 (1007 MB) [ 0.000000] memory : 0x0000000040000000 - 0x00000000c0000000 (2048 MB) [ 0.000000] .init : 0x0000000040100000 - 0x0000000040200000 (1024 kB) [ 0.000000] .data : 0x0000000040b0e000 - 0x0000000040f533e0 (4372 kB) [ 0.000000] .text : 0x0000000040200000 - 0x0000000040b0e000 (9272 kB) [ 0.768910] Brought up 1 CPUs [ 0.992465] NET: Registered protocol family 16 [ 2.429981] Releasing cpu 1 now, hpa=fffffffffffa2000 [ 2.635751] CPU(s): 2 out of 2 PA8500 (PCX-W) at 440.000000 MHz online [ 2.726692] Setting cache flush threshold to 1024 kB [ 2.729932] Not-handled unaligned insn 0x43ffff80 [ 2.798114] Setting TLB flush threshold to 140 kB [ 2.928039] Unaligned handler failed, ret = -1 [ 3.000419] _______________________________ [ 3.000419] < Your System ate a SPARC! Gah! > [ 3.000419] ------------------------------- [ 3.000419] \ ^__^ [ 3.000419] (__)\ )\/\ [ 3.000419] U ||----w | [ 3.000419] || || [ 9.340055] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.8.0-1-parisc64-smp #1 Debian 4.8.7-1 [ 9.448082] task: 00000000bfd48060 task.stack: 00000000bfd50000 [ 9.528040] [ 10.760029] IASQ: 0000000000000000 0000000000000000 IAOQ: 000000004025d154 000000004025d158 [ 10.868052] IIR: 43ffff80 ISR: 0000000000340000 IOR: 000001ff54150960 [ 10.960029] CPU: 1 CR30: 00000000bfd50000 CR31: 0000000011111111 [ 11.052057] ORIG_R28: 000000004021e3b4 [ 11.100045] IAOQ[0]: irq_exit+0x94/0x120 [ 11.152062] IAOQ[1]: irq_exit+0x98/0x120 [ 11.208031] RP(r2): irq_exit+0xb8/0x120 [ 11.256074] Backtrace: [ 11.288067] [<00000000402cd944>] cpu_startup_entry+0x1e4/0x598 [ 11.368058] [<0000000040109528>] smp_callin+0x2c0/0x2f0 [ 11.436308] [<00000000402b53fc>] update_curr+0x18c/0x2d0 [ 11.508055] [<00000000402b73b8>] dequeue_entity+0x2c0/0x1030 [ 11.584040] [<00000000402b3cc0>] set_next_entity+0x80/0xd30 [ 11.660069] [<00000000402c1594>] pick_next_task_fair+0x614/0x720 [ 11.740085] [<000000004020dd34>] __schedule+0x394/0xa60 [ 11.808054] [<000000004020e488>] schedule+0x88/0x118 [ 11.876039] [<0000000040283d3c>] rescuer_thread+0x4d4/0x5b0 [ 11.948090] [<000000004028fc4c>] kthread+0x1ec/0x248 [ 12.016053] [<0000000040205020>] end_fault_vector+0x20/0xc0 [ 12.092239] [<00000000402050c0>] _switch_to_ret+0x0/0xf40 [ 12.164044] [ 12.184036] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.8.0-1-parisc64-smp #1 Debian 4.8.7-1 [ 12.244040] Backtrace: [ 12.244040] [<000000004021c480>] show_stack+0x68/0x80 [ 12.244040] [<00000000406f332c>] dump_stack+0xec/0x168 [ 12.244040] [<000000004021c74c>] die_if_kernel+0x25c/0x430 [ 12.244040] [<000000004022d320>] handle_unaligned+0xb48/0xb50 [ 12.244040] [ 12.632066] ---[ end trace 9ca05a7215c7bbb2 ]--- [ 12.692036] Kernel panic - not syncing: Attempted to kill the idle task! We have the insn 0x43ffff80 in IIR but from IAOQ we should have: 4025d150: 0f f3 20 df ldd,s r19(r31),r31 4025d154: 0f 9f 00 9c ldw r31(ret0),ret0 4025d158: bf 80 20 58 cmpb,*<> r0,ret0,4025d18c <irq_exit+0xcc> Cpu0 has just completed running parisc_setup_cache_timing: [ 2.429981] Releasing cpu 1 now, hpa=fffffffffffa2000 [ 2.635751] CPU(s): 2 out of 2 PA8500 (PCX-W) at 440.000000 MHz online [ 2.726692] Setting cache flush threshold to 1024 kB [ 2.729932] Not-handled unaligned insn 0x43ffff80 [ 2.798114] Setting TLB flush threshold to 140 kB [ 2.928039] Unaligned handler failed, ret = -1 From the backtrace, cpu1 is in smp_callin: void __init smp_callin(void) { int slave_id = cpu_now_booting; smp_cpu_init(slave_id); preempt_disable(); flush_cache_all_local(); /* start with known state */ flush_tlb_all_local(NULL); local_irq_enable(); /* Interrupts have been off until now */ cpu_startup_entry(CPUHP_AP_ONLINE_IDLE); So, it has just flushed its caches and the TLB. It would seem either the flushes in parisc_setup_cache_timing or smp_callin have corrupted kernel memory. The attached patch reworks parisc_setup_cache_timing to remove the races in setting the cache and TLB flush thresholds. It also corrects the number of bytes flushed in the TLB calculation. The patch flushes the cache and TLB on cpu0 before starting the secondary processors so that they are started from a known state. Tested with a few reboots on c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2016-11-21 02:12:36 +00:00
printk(KERN_INFO "TLB flush threshold set to %lu KiB\n",
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
parisc_tlb_flush_threshold/1024);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[PARISC] parisc specific kmap API implementation for pa8800 This patch fixes the pa8800 at a gross level (there are still other subtle incoherency issues which can still cause crashes and HPMCs). What it does is try to force eject inequivalent aliases before they become visible to the L2 cache (which is where we get the incoherence problems). A new function (parisc_requires_coherency) is introduced in asm/processor.h to identify the pa8x00 processors (8800 and 8900) which have the issue. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-08-23 16:00:04 +00:00
parisc: fixes and cleanups in page cache flushing (1/4) This is the first patch in a series of 4, with which the page cache flushing of parisc will gets fixed and enhanced. This even fixes the nasty "minifail" bug (http://wiki.parisc-linux.org/TestCases?highlight=%28minifail%29) which prevented parisc to stay an official debian port. Basically the flush in copy_user_page together with the TLB patch from commit 7139bc1579901b53db7e898789e916ee2fb52d78 is what fixes the minifail bug. This patch still uses the TMPALIAS approach. The new copy_user_page implementation calls flush_dcache_page_asm to flush the user dcache page (crucial for minifail fix) via a kernel TMPALIAS mapping. After that, it just copies the page using the kernel mapping. It does a final flush if needed. Generally it is hard to avoid doing some cache flushes using the kernel mapping (e.g., copy_to_user_page and copy_from_user_page). This patch depends on a subsequent change to pacache.S implementing clear_page_asm and copy_page_asm. These are optimized routines to clear and copy a page. The calls in clear_user_page and copy_user_page could be replaced by calls to memset and memcpy, respectively. I tested prefetch optimizations in clear_page_asm and copy_page_asm but didn't see any significant performance improvement on rp3440. I'm not sure if these are routines are significantly faster than memset and/or memcpy, but they are there for further performance evaluation. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2013-02-03 22:59:09 +00:00
extern void purge_kernel_dcache_page_asm(unsigned long);
extern void clear_user_page_asm(void *, unsigned long);
extern void copy_user_page_asm(void *, void *, unsigned long);
[PARISC] parisc specific kmap API implementation for pa8800 This patch fixes the pa8800 at a gross level (there are still other subtle incoherency issues which can still cause crashes and HPMCs). What it does is try to force eject inequivalent aliases before they become visible to the L2 cache (which is where we get the incoherence problems). A new function (parisc_requires_coherency) is introduced in asm/processor.h to identify the pa8x00 processors (8800 and 8900) which have the issue. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-08-23 16:00:04 +00:00
void flush_kernel_dcache_page_addr(void *addr)
{
parisc: ensure broadcast tlb purge runs single threaded The TLB flushing functions on hppa, which causes PxTLB broadcasts on the system bus, needs to be protected by irq-safe spinlocks to avoid irq handlers to deadlock the kernel. The deadlocks only happened during I/O intensive loads and triggered pretty seldom, which is why this bug went so long unnoticed. Signed-off-by: Helge Deller <deller@gmx.de> [edited to use spin_lock_irqsave on UP as well since we'd been locking there all this time anyway, --kyle] Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-06-16 20:51:48 +00:00
unsigned long flags;
[PARISC] parisc specific kmap API implementation for pa8800 This patch fixes the pa8800 at a gross level (there are still other subtle incoherency issues which can still cause crashes and HPMCs). What it does is try to force eject inequivalent aliases before they become visible to the L2 cache (which is where we get the incoherence problems). A new function (parisc_requires_coherency) is introduced in asm/processor.h to identify the pa8x00 processors (8800 and 8900) which have the issue. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-08-23 16:00:04 +00:00
flush_kernel_dcache_page_asm(addr);
parisc: ensure broadcast tlb purge runs single threaded The TLB flushing functions on hppa, which causes PxTLB broadcasts on the system bus, needs to be protected by irq-safe spinlocks to avoid irq handlers to deadlock the kernel. The deadlocks only happened during I/O intensive loads and triggered pretty seldom, which is why this bug went so long unnoticed. Signed-off-by: Helge Deller <deller@gmx.de> [edited to use spin_lock_irqsave on UP as well since we'd been locking there all this time anyway, --kyle] Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-06-16 20:51:48 +00:00
purge_tlb_start(flags);
parisc: Use constants to encode the space registers like SR_KERNEL Use the provided space register constants instead of hardcoded values. Signed-off-by: Helge Deller <deller@gmx.de>
2022-02-17 14:26:27 +00:00
pdtlb(SR_KERNEL, addr);
parisc: ensure broadcast tlb purge runs single threaded The TLB flushing functions on hppa, which causes PxTLB broadcasts on the system bus, needs to be protected by irq-safe spinlocks to avoid irq handlers to deadlock the kernel. The deadlocks only happened during I/O intensive loads and triggered pretty seldom, which is why this bug went so long unnoticed. Signed-off-by: Helge Deller <deller@gmx.de> [edited to use spin_lock_irqsave on UP as well since we'd been locking there all this time anyway, --kyle] Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-06-16 20:51:48 +00:00
purge_tlb_end(flags);
[PARISC] parisc specific kmap API implementation for pa8800 This patch fixes the pa8800 at a gross level (there are still other subtle incoherency issues which can still cause crashes and HPMCs). What it does is try to force eject inequivalent aliases before they become visible to the L2 cache (which is where we get the incoherence problems). A new function (parisc_requires_coherency) is introduced in asm/processor.h to identify the pa8x00 processors (8800 and 8900) which have the issue. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-08-23 16:00:04 +00:00
}
EXPORT_SYMBOL(flush_kernel_dcache_page_addr);
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
static void flush_cache_page_if_present(struct vm_area_struct *vma,
unsigned long vmaddr, unsigned long pfn)
parisc: fix cache-flushing This commit: f8dae00684d678afa13041ef170cecfd1297ed40: parisc: Ensure full cache coherency for kmap/kunmap caused negative caching side-effects, e.g. hanging processes with expect and too many inequivalent alias messages from flush_dcache_page() on Debian 5 systems. This patch now partly reverts it and has been in production use on our debian buildd makeservers since a week without any major problems. Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2014-01-31 20:33:17 +00:00
{
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
pte_t *ptep = get_ptep(vma->vm_mm, vmaddr);
/*
* The pte check is racy and sometimes the flush will trigger
* a non-access TLB miss. Hopefully, the page has already been
* flushed.
*/
if (ptep && pte_needs_flush(*ptep))
flush_cache_page(vma, vmaddr, pfn);
}
void copy_user_highpage(struct page *to, struct page *from,
unsigned long vaddr, struct vm_area_struct *vma)
{
void *kto, *kfrom;
kfrom = kmap_local_page(from);
kto = kmap_local_page(to);
flush_cache_page_if_present(vma, vaddr, page_to_pfn(from));
copy_page_asm(kto, kfrom);
kunmap_local(kto);
kunmap_local(kfrom);
}
void copy_to_user_page(struct vm_area_struct *vma, struct page *page,
unsigned long user_vaddr, void *dst, void *src, int len)
{
flush_cache_page_if_present(vma, user_vaddr, page_to_pfn(page));
memcpy(dst, src, len);
flush_kernel_dcache_range_asm((unsigned long)dst, (unsigned long)dst + len);
}
void copy_from_user_page(struct vm_area_struct *vma, struct page *page,
unsigned long user_vaddr, void *dst, void *src, int len)
{
flush_cache_page_if_present(vma, user_vaddr, page_to_pfn(page));
memcpy(dst, src, len);
parisc: fix cache-flushing This commit: f8dae00684d678afa13041ef170cecfd1297ed40: parisc: Ensure full cache coherency for kmap/kunmap caused negative caching side-effects, e.g. hanging processes with expect and too many inequivalent alias messages from flush_dcache_page() on Debian 5 systems. This patch now partly reverts it and has been in production use on our debian buildd makeservers since a week without any major problems. Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2014-01-31 20:33:17 +00:00
}
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
/* __flush_tlb_range()
*
* returns 1 if all TLBs were flushed.
*/
int __flush_tlb_range(unsigned long sid, unsigned long start,
unsigned long end)
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
{
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
unsigned long flags;
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
if ((!IS_ENABLED(CONFIG_SMP) || !arch_irqs_disabled()) &&
end - start >= parisc_tlb_flush_threshold) {
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
flush_tlb_all();
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
return 1;
}
parisc: ensure broadcast tlb purge runs single threaded The TLB flushing functions on hppa, which causes PxTLB broadcasts on the system bus, needs to be protected by irq-safe spinlocks to avoid irq handlers to deadlock the kernel. The deadlocks only happened during I/O intensive loads and triggered pretty seldom, which is why this bug went so long unnoticed. Signed-off-by: Helge Deller <deller@gmx.de> [edited to use spin_lock_irqsave on UP as well since we'd been locking there all this time anyway, --kyle] Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-06-16 20:51:48 +00:00
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
/* Purge TLB entries for small ranges using the pdtlb and
pitlb instructions. These instructions execute locally
but cause a purge request to be broadcast to other TLBs. */
while (start < end) {
parisc: ensure broadcast tlb purge runs single threaded The TLB flushing functions on hppa, which causes PxTLB broadcasts on the system bus, needs to be protected by irq-safe spinlocks to avoid irq handlers to deadlock the kernel. The deadlocks only happened during I/O intensive loads and triggered pretty seldom, which is why this bug went so long unnoticed. Signed-off-by: Helge Deller <deller@gmx.de> [edited to use spin_lock_irqsave on UP as well since we'd been locking there all this time anyway, --kyle] Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-06-16 20:51:48 +00:00
purge_tlb_start(flags);
parisc: Use constants to encode the space registers like SR_KERNEL Use the provided space register constants instead of hardcoded values. Signed-off-by: Helge Deller <deller@gmx.de>
2022-02-17 14:26:27 +00:00
mtsp(sid, SR_TEMP1);
pdtlb(SR_TEMP1, start);
pitlb(SR_TEMP1, start);
parisc: ensure broadcast tlb purge runs single threaded The TLB flushing functions on hppa, which causes PxTLB broadcasts on the system bus, needs to be protected by irq-safe spinlocks to avoid irq handlers to deadlock the kernel. The deadlocks only happened during I/O intensive loads and triggered pretty seldom, which is why this bug went so long unnoticed. Signed-off-by: Helge Deller <deller@gmx.de> [edited to use spin_lock_irqsave on UP as well since we'd been locking there all this time anyway, --kyle] Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2009-06-16 20:51:48 +00:00
purge_tlb_end(flags);
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
start += PAGE_SIZE;
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
}
parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results The increased use of pdtlb/pitlb instructions seemed to increase the frequency of random segmentation faults building packages. Further, we had a number of cases where TLB inserts would repeatedly fail and all forward progress would stop. The Haskell ghc package caused a lot of trouble in this area. The final indication of a race in pte handling was this syslog entry on sibaris (C8000): swap_free: Unused swap offset entry 00000004 BUG: Bad page map in process mysqld pte:00000100 pmd:019bbec5 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1 Backtrace: [<0000000040173eb0>] show_stack+0x20/0x38 [<0000000040444424>] dump_stack+0x9c/0x110 [<00000000402a0d38>] print_bad_pte+0x1a8/0x278 [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770 [<00000000402a4090>] zap_page_range+0xf0/0x198 [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0 Note that the pte value is 0 except for the accessed bit 0x100. This bit shouldn't be set without the present bit. It should be noted that the madvise system call is probably a trigger for many of the random segmentation faults. In looking at the kernel code, I found the following problems: 1) The pte_clear define didn't take TLB lock when clearing a pte. 2) We didn't test pte present bit inside lock in exception support. 3) The pte and tlb locks needed to merged in order to ensure consistency between page table and TLB. This also has the effect of serializing TLB broadcasts on SMP systems. The attached change implements the above and a few other tweaks to try to improve performance. Based on the timing code, TLB purges are very slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it beneficial to test the split_tlb variable to avoid duplicate purges. Probably, all PA 2.0 machines have combined TLBs. I dropped using __flush_tlb_range in flush_tlb_mm as I realized all applications and most threads have a stack size that is too large to make this useful. I added some comments to this effect. Since implementing 1 through 3, I haven't had any random segmentation faults on mx3210 (rp3440) in about one week of building code and running as a Debian buildd. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v3.18+ Signed-off-by: Helge Deller <deller@gmx.de>
2015-07-01 21:18:37 +00:00
return 0;
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
static void flush_cache_pages(struct vm_area_struct *vma, unsigned long start, unsigned long end)
parisc: fixes and cleanups in page cache flushing (3/4) flush_cache_mm, for the non current case also uses flush_dcache_page_asm and flush_icache_page_asm which are TMPALIAS flushes. For the non current case, the algorithm used by get_ptep is derived from the vmalloc_to_page implementation in vmalloc.c. It is essentially a generic page table lookup. The other alternative was to duplicate the lookup in entry.S. The break point for switching to a full cache flush is somewhat arbitrary. The same approach is used in flush_cache_range for non current case. In a GCC build and check, many small programs are executed and this change provided a significant performance enhancement, e.g. GCC build time was cut almost in half on a rp3440 at j4. Previously, we always flushed the entire cache. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2013-02-03 23:01:47 +00:00
{
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
unsigned long addr, pfn;
pte_t *ptep;
parisc: fixes and cleanups in page cache flushing (3/4) flush_cache_mm, for the non current case also uses flush_dcache_page_asm and flush_icache_page_asm which are TMPALIAS flushes. For the non current case, the algorithm used by get_ptep is derived from the vmalloc_to_page implementation in vmalloc.c. It is essentially a generic page table lookup. The other alternative was to duplicate the lookup in entry.S. The break point for switching to a full cache flush is somewhat arbitrary. The same approach is used in flush_cache_range for non current case. In a GCC build and check, many small programs are executed and this change provided a significant performance enhancement, e.g. GCC build time was cut almost in half on a rp3440 at j4. Previously, we always flushed the entire cache. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2013-02-03 23:01:47 +00:00
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
for (addr = start; addr < end; addr += PAGE_SIZE) {
/*
* The vma can contain pages that aren't present. Although
* the pte search is expensive, we need the pte to find the
* page pfn and to check whether the page should be flushed.
*/
ptep = get_ptep(vma->vm_mm, addr);
if (ptep && pte_needs_flush(*ptep)) {
if (parisc_requires_coherency()) {
flush_user_cache_page(vma, addr);
} else {
pfn = pte_pfn(*ptep);
if (WARN_ON(!pfn_valid(pfn)))
return;
__flush_cache_page(vma, addr, PFN_PHYS(pfn));
parisc: use pgtable-nopXd instead of 4level-fixup parisc has two or three levels of page tables and can use appropriate pgtable-nopXd and folding of the upper layers. Replace usage of include/asm-generic/4level-fixup.h and explicit definitions of __PAGETABLE_PxD_FOLDED in parisc with include/asm-generic/pgtable-nopmd.h for two-level configurations and with include/asm-generic/pgtable-nopud.h for three-lelve configurations and adjust page table manipulation macros and functions accordingly. Link: http://lkml.kernel.org/r/1572938135-31886-9-git-send-email-rppt@kernel.org Signed-off-by: Mike Rapoport <rppt@linux.ibm.com> Acked-by: Helge Deller <deller@gmx.de> Cc: Anatoly Pugachev <matorola@gmail.com> Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: "David S. Miller" <davem@davemloft.net> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Greentime Hu <green.hu@gmail.com> Cc: Greg Ungerer <gerg@linux-m68k.org> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Jeff Dike <jdike@addtoit.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Mark Salter <msalter@redhat.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Michal Simek <monstr@monstr.eu> Cc: Peter Rosin <peda@axentia.se> Cc: Richard Weinberger <richard@nod.at> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Russell King <linux@armlinux.org.uk> Cc: Russell King <rmk+kernel@armlinux.org.uk> Cc: Sam Creasey <sammy@sammy.net> Cc: Vincent Chen <deanbo422@gmail.com> Cc: Vineet Gupta <Vineet.Gupta1@synopsys.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-12-05 00:54:12 +00:00
}
parisc: fixes and cleanups in page cache flushing (3/4) flush_cache_mm, for the non current case also uses flush_dcache_page_asm and flush_icache_page_asm which are TMPALIAS flushes. For the non current case, the algorithm used by get_ptep is derived from the vmalloc_to_page implementation in vmalloc.c. It is essentially a generic page table lookup. The other alternative was to duplicate the lookup in entry.S. The break point for switching to a full cache flush is somewhat arbitrary. The same approach is used in flush_cache_range for non current case. In a GCC build and check, many small programs are executed and this change provided a significant performance enhancement, e.g. GCC build time was cut almost in half on a rp3440 at j4. Previously, we always flushed the entire cache. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2013-02-03 23:01:47 +00:00
}
}
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
static inline unsigned long mm_total_size(struct mm_struct *mm)
parisc: deduplicate code in flush_cache_mm() and flush_cache_range() Parts of both functions are the same, so deduplicate them. No functional change. Signed-off-by: Sven Schnelle <svens@stackframe.org> Signed-off-by: Helge Deller <deller@gmx.de>
2021-10-09 18:24:36 +00:00
{
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
struct vm_area_struct *vma;
unsigned long usize = 0;
parisc: deduplicate code in flush_cache_mm() and flush_cache_range() Parts of both functions are the same, so deduplicate them. No functional change. Signed-off-by: Sven Schnelle <svens@stackframe.org> Signed-off-by: Helge Deller <deller@gmx.de>
2021-10-09 18:24:36 +00:00
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
for (vma = mm->mmap; vma && usize < parisc_cache_flush_threshold; vma = vma->vm_next)
usize += vma->vm_end - vma->vm_start;
return usize;
parisc: deduplicate code in flush_cache_mm() and flush_cache_range() Parts of both functions are the same, so deduplicate them. No functional change. Signed-off-by: Sven Schnelle <svens@stackframe.org> Signed-off-by: Helge Deller <deller@gmx.de>
2021-10-09 18:24:36 +00:00
}
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
void flush_cache_mm(struct mm_struct *mm)
{
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
struct vm_area_struct *vma;
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
/*
* Flushing the whole cache on each cpu takes forever on
* rp3440, etc. So, avoid it if the mm isn't too big.
*
* Note that we must flush the entire cache on machines
* with aliasing caches to prevent random segmentation
* faults.
*/
if (!parisc_requires_coherency()
|| mm_total_size(mm) >= parisc_cache_flush_threshold) {
if (WARN_ON(IS_ENABLED(CONFIG_SMP) && arch_irqs_disabled()))
return;
flush_tlb_all();
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
flush_cache_all();
return;
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
/* Flush mm */
parisc: deduplicate code in flush_cache_mm() and flush_cache_range() Parts of both functions are the same, so deduplicate them. No functional change. Signed-off-by: Sven Schnelle <svens@stackframe.org> Signed-off-by: Helge Deller <deller@gmx.de>
2021-10-09 18:24:36 +00:00
for (vma = mm->mmap; vma; vma = vma->vm_next)
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
flush_cache_pages(vma, vma->vm_start, vma->vm_end);
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
void flush_cache_range(struct vm_area_struct *vma, unsigned long start, unsigned long end)
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
{
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
if (!parisc_requires_coherency()
|| end - start >= parisc_cache_flush_threshold) {
if (WARN_ON(IS_ENABLED(CONFIG_SMP) && arch_irqs_disabled()))
return;
flush_tlb_range(vma, start, end);
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
flush_cache_all();
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
return;
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
flush_cache_pages(vma, start, end);
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
void flush_cache_page(struct vm_area_struct *vma, unsigned long vmaddr, unsigned long pfn)
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
{
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
if (WARN_ON(!pfn_valid(pfn)))
return;
if (parisc_requires_coherency())
flush_user_cache_page(vma, vmaddr);
else
__flush_cache_page(vma, vmaddr, PFN_PHYS(pfn));
}
void flush_anon_page(struct vm_area_struct *vma, struct page *page, unsigned long vmaddr)
{
if (!PageAnon(page))
return;
if (parisc_requires_coherency()) {
flush_user_cache_page(vma, vmaddr);
return;
parisc: Fix cache routines to ignore vma's with an invalid pfn The parisc architecture does not have a pte special bit. As a result, special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags. VM_MIXEDMAP mappings may or may not have a "struct page" backing. When pfn_valid() is false, there is no "struct page" backing. Otherwise, they are treated as normal pages. The FireGL driver uses the VM_MIXEDMAP without a backing "struct page". This treatment caused a panic due to a TLB data miss in update_mmu_cache. This appeared to be in the code generated for page_address(). We were in fact using a very circular bit of code to determine the physical address of the PFN in various cache routines. This wasn't valid when there was no "struct page" backing. The needed address can in fact be determined simply from the PFN itself without using the "struct page". The attached patch updates update_mmu_cache(), flush_cache_mm(), flush_cache_range() and flush_cache_page() to check pfn_valid() and to directly compute the PFN physical and virtual addresses. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: <stable@vger.kernel.org> # 3.10 Signed-off-by: Helge Deller <deller@gmx.de>
2013-07-23 16:27:52 +00:00
}
parisc: Rewrite cache flush code for PA8800/PA8900 Originally, I was convinced that we needed to use tmpalias flushes everwhere, for both user and kernel flushes. However, when I modified flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000 would crash quite early when booting. The PDC returns alias values of 0 for the icache and dcache. This indicates that either the alias boundary is greater than 16MB or equivalent aliasing doesn't work. I modified the tmpalias code to make it easy to try alternate boundaries. I tried boundaries up to 128MB but still kernel tmpalias flushes didn't work on c8000. This led me to conclude that tmpalias flushes don't work on PA8800 and PA8900 machines, and that we needed to flush directly using the virtual address of user and kernel pages. This is likely the major cause of instability on the c8000 and rp34xx machines. Flushing user pages requires doing a temporary context switch as we have to flush pages that don't belong to the current context. Further, we have to deal with pages that aren't present. If a page isn't present, the flush instructions fault on every line. Other code has been rearranged and simplified based on testing. For example, I introduced a flush_cache_dup_mm routine. flush_cache_mm and flush_cache_dup_mm differ in that flush_cache_mm calls purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages. In some implementations, pdc is more efficient than fdc. Based on my testing, I don't believe there's any performance benefit on the c8000. Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de>
2022-05-16 15:14:47 +00:00
flush_tlb_page(vma, vmaddr);
preempt_disable();
flush_dcache_page_asm(page_to_phys(page), vmaddr);
preempt_enable();
[PARISC] Clean up the cache and tlb headers No changes in functionality. Signed-off-by: Randolph Chung <tausq@debian.org> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-12-12 13:51:54 +00:00
}
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
void flush_kernel_vmap_range(void *vaddr, int size)
{
unsigned long start = (unsigned long)vaddr;
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
unsigned long end = start + size;
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
if ((!IS_ENABLED(CONFIG_SMP) || !arch_irqs_disabled()) &&
(unsigned long)size >= parisc_cache_flush_threshold) {
flush_tlb_kernel_range(start, end);
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
flush_data_cache();
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
return;
}
Revert "parisc: Fix invalidate/flush vmap routines" This reverts commit 53d862fac4a09b9c56cca0433fa9de5732fd05a1. It turned out that flush_kernel_vmap_range() is being called with interrupts disabled. There's no way to flush entire cache with interrupts disabled. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-27 16:53:42 +00:00
flush_kernel_dcache_range_asm(start, end);
flush_tlb_kernel_range(start, end);
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
}
EXPORT_SYMBOL(flush_kernel_vmap_range);
void invalidate_kernel_vmap_range(void *vaddr, int size)
{
unsigned long start = (unsigned long)vaddr;
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
unsigned long end = start + size;
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
if ((!IS_ENABLED(CONFIG_SMP) || !arch_irqs_disabled()) &&
(unsigned long)size >= parisc_cache_flush_threshold) {
flush_tlb_kernel_range(start, end);
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
flush_data_cache();
parisc: Fix ordering of cache and TLB flushes The change to flush_kernel_vmap_range() wasn't sufficient to avoid the SMP stalls.  The problem is some drivers call these routines with interrupts disabled.  Interrupts need to be enabled for flush_tlb_all() and flush_cache_all() to work.  This version adds checks to ensure interrupts are not disabled before calling routines that need IPI interrupts.  When interrupts are disabled, we now drop into slower code. The attached change fixes the ordering of cache and TLB flushes in several cases.  When we flush the cache using the existing PTE/TLB entries, we need to flush the TLB after doing the cache flush.  We don't need to do this when we flush the entire instruction and data caches as these flushes don't use the existing TLB entries.  The same is true for tmpalias region flushes. The flush_kernel_vmap_range() and invalidate_kernel_vmap_range() routines have been updated. Secondly, we added a new purge_kernel_dcache_range_asm() routine to pacache.S and use it in invalidate_kernel_vmap_range().  Nominally, purges are faster than flushes as the cache lines don't have to be written back to memory. Hopefully, this is sufficient to resolve the remaining problems due to cache speculation.  So far, testing indicates that this is the case.  I did work up a patch using tmpalias flushes, but there is a performance hit because we need the physical address for each page, and we also need to sequence access to the tmpalias flush code.  This increases the probability of stalls. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # 4.9+ Signed-off-by: Helge Deller <deller@gmx.de>
2018-02-27 13:16:07 +00:00
return;
}
Revert "parisc: Fix invalidate/flush vmap routines" This reverts commit 53d862fac4a09b9c56cca0433fa9de5732fd05a1. It turned out that flush_kernel_vmap_range() is being called with interrupts disabled. There's no way to flush entire cache with interrupts disabled. Signed-off-by: Helge Deller <deller@gmx.de>
2022-03-27 16:53:42 +00:00
purge_kernel_dcache_range_asm(start, end);
flush_tlb_kernel_range(start, end);
parisc: Optimize flush_kernel_vmap_range and invalidate_kernel_vmap_range The previously submitted patch did not resolve the random segmentation faults observed on the phantom buildd system. There are still unresolved problems with the Debian 4.8 and 4.9 kernels on C8000. The attached patch removes the flush of the offset map pages and does a whole data cache flush for large ranges. No other arch flushes the offset map in these routines as far as I can tell. I have not observed any random segmentation faults on rp3440 in two weeks of testing with 4.10.0 and 4.10.1. Signed-off-by: John David Anglin <dave.anglin@bell.net> Cc: stable@vger.kernel.org # v4.8+ Signed-off-by: Helge Deller <deller@gmx.de>
2017-03-11 23:03:34 +00:00
}
EXPORT_SYMBOL(invalidate_kernel_vmap_range);
Reference in a new issue Copy permalink