- 08 Oct, 2016 40 commits
-
-
Michal Hocko authored
Since the lumpy reclaim is gone there is no source of higher order pages if CONFIG_COMPACTION=n except for the order-0 pages reclaim which is unreliable for that purpose to say the least. Hitting an OOM for !costly higher order requests is therefore all not that hard to imagine. We are trying hard to not invoke OOM killer as much as possible but there is simply no reliable way to detect whether more reclaim retries make sense. Disabling COMPACTION is not widespread but it seems that some users might have disable the feature without realizing full consequences (mostly along with disabling THP because compaction used to be THP mainly thing). This patch just adds a note if the OOM killer was triggered by higher order request with compaction disabled. This will help us identifying possible misconfiguration right from the oom report which is easier than to always keep in mind that somebody might have disabled COMPACTION without a good reason. Link: http://lkml.kernel.org/r/20160830111632.GD23963@dhcp22.suse.czSigned-off-by:
Michal Hocko <mhocko@suse.com> Cc: David Rientjes <rientjes@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Huang Ying authored
File pages use a set of radix tree tags (DIRTY, TOWRITE, WRITEBACK, etc.) to accelerate finding the pages with a specific tag in the radix tree during inode writeback. But for anonymous pages in the swap cache, there is no inode writeback. So there is no need to find the pages with some writeback tags in the radix tree. It is not necessary to touch radix tree writeback tags for pages in the swap cache. Per Rik van Riel's suggestion, a new flag AS_NO_WRITEBACK_TAGS is introduced for address spaces which don't need to update the writeback tags. The flag is set for swap caches. It may be used for DAX file systems, etc. With this patch, the swap out bandwidth improved 22.3% (from ~1.2GB/s to ~1.48GBps) in the vm-scalability swap-w-seq test case with 8 processes. The test is done on a Xeon E5 v3 system. The swap device used is a RAM simulated PMEM (persistent memory) device. The improvement comes from the reduced contention on the swap cache radix tree lock. To test sequential swapping out, the test case uses 8 processes, which sequentially allocate and write to the anonymous pages until RAM and part of the swap device is used up. Details of comparison is as follow, base base+patch ---------------- -------------------------- %stddev %change %stddev \ | \ 2506952 ± 2% +28.1% 3212076 ± 7% vm-scalability.throughput 1207402 ± 7% +22.3% 1476578 ± 6% vmstat.swap.so 10.86 ± 12% -23.4% 8.31 ± 16% perf-profile.cycles-pp._raw_spin_lock_irq.__add_to_swap_cache.add_to_swap_cache.add_to_swap.shrink_page_list 10.82 ± 13% -33.1% 7.24 ± 14% perf-profile.cycles-pp._raw_spin_lock_irqsave.__remove_mapping.shrink_page_list.shrink_inactive_list.shrink_zone_memcg 10.36 ± 11% -100.0% 0.00 ± -1% perf-profile.cycles-pp._raw_spin_lock_irqsave.__test_set_page_writeback.bdev_write_page.__swap_writepage.swap_writepage 10.52 ± 12% -100.0% 0.00 ± -1% perf-profile.cycles-pp._raw_spin_lock_irqsave.test_clear_page_writeback.end_page_writeback.page_endio.pmem_rw_page Link: http://lkml.kernel.org/r/1472578089-5560-1-git-send-email-ying.huang@intel.comSigned-off-by:"Huang, Ying" <ying.huang@intel.com> Acked-by:
Rik van Riel <riel@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: Shaohua Li <shli@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Tejun Heo <tj@kernel.org> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Dave Hansen <dave.hansen@intel.com> Signed-off-by:
-
zijun_hu authored
In ___alloc_bootmem_node_nopanic(), replace kzalloc() by kzalloc_node() in order to allocate memory within given node preferentially when slab is available Link: http://lkml.kernel.org/r/1f487f12-6af4-5e4f-a28c-1de2361cdcd8@zoho.comSigned-off-by:
zijun_hu <zijun_hu@htc.com> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by:
-
zijun_hu authored
Fix the following bugs: - the same ARCH_LOW_ADDRESS_LIMIT statements are duplicated between header and relevant source - don't ensure ARCH_LOW_ADDRESS_LIMIT perhaps defined by ARCH in asm/processor.h is preferred over default in linux/bootmem.h completely since the former header isn't included by the latter Link: http://lkml.kernel.org/r/e046aeaa-e160-6d9e-dc1b-e084c2fd999f@zoho.comSigned-off-by:
-
Srikar Dronamraju authored
Currently significant amount of memory is reserved only in kernel booted to capture kernel dump using the fa_dump method. Kernels compiled with CONFIG_DEFERRED_STRUCT_PAGE_INIT will initialize only certain size memory per node. The certain size takes into account the dentry and inode cache sizes. Currently the cache sizes are calculated based on the total system memory including the reserved memory. However such a kernel when booting the same kernel as fadump kernel will not be able to allocate the required amount of memory to suffice for the dentry and inode caches. This results in crashes like Hence only implement arch_reserved_kernel_pages() for CONFIG_FA_DUMP configurations. The amount reserved will be reduced while calculating the large caches and will avoid crashes like the below on large systems such as 32 TB systems. Dentry cache hash table entries: 536870912 (order: 16, 4294967296 bytes) vmalloc: allocation failure, allocated 4097114112 of 17179934720 bytes swapper/0: page allocation failure: order:0, mode:0x2080020(GFP_ATOMIC) CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.6-master+ #3 Call Trace: dump_stack+0xb0/0xf0 (unreliable) warn_alloc_failed+0x114/0x160 __vmalloc_node_range+0x304/0x340 __vmalloc+0x6c/0x90 alloc_large_system_hash+0x1b8/0x2c0 inode_init+0x94/0xe4 vfs_caches_init+0x8c/0x13c start_kernel+0x50c/0x578 start_here_common+0x20/0xa8 Link: http://lkml.kernel.org/r/1472476010-4709-4-git-send-email-srikar@linux.vnet.ibm.comSigned-off-by:Srikar Dronamraju <srikar@linux.vnet.ibm.com> Suggested-by:
Mel Gorman <mgorman@techsingularity.net> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Michal Hocko <mhocko@kernel.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com> Cc: Hari Bathini <hbathini@linux.vnet.ibm.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Balbir Singh <bsingharora@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by:
-
Srikar Dronamraju authored
The total reserved memory in a system is accounted but not available for use use outside mm/memblock.c. By exposing the total reserved memory, systems can better calculate the size of large hashes. Link: http://lkml.kernel.org/r/1472476010-4709-3-git-send-email-srikar@linux.vnet.ibm.comSigned-off-by:
-
Srikar Dronamraju authored
Currently arch specific code can reserve memory blocks but alloc_large_system_hash() may not take it into consideration when sizing the hashes. This can lead to bigger hash than required and lead to no available memory for other purposes. This is specifically true for systems with CONFIG_DEFERRED_STRUCT_PAGE_INIT enabled. One approach to solve this problem would be to walk through the memblock regions and calculate the available memory and base the size of hash system on the available memory. The other approach would be to depend on the architecture to provide the number of pages that are reserved. This change provides hooks to allow the architecture to provide the required info. Link: http://lkml.kernel.org/r/1472476010-4709-2-git-send-email-srikar@linux.vnet.ibm.comSigned-off-by:
-
Aneesh Kumar K.V authored
Use the existing enums instead of hardcoded index when looking at the zonelist. This makes it more readable. No functionality change by this patch. Link: http://lkml.kernel.org/r/1472227078-24852-1-git-send-email-aneesh.kumar@linux.vnet.ibm.comSigned-off-by:
Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com> Reviewed-by:
Anshuman Khandual <khandual@linux.vnet.ibm.com> Signed-off-by:
-
Michal Hocko authored
oom reaper was skipped for an mm which is shared with the kernel thread (aka use_mm()). The primary concern was that such a kthread might want to read from the userspace memory and see zero page as a result of the oom reaper action. This is no longer a problem after "mm: make sure that kthreads will not refault oom reaped memory" because any attempt to fault in when the MMF_UNSTABLE is set will result in SIGBUS and so the target user should see an error. This means that we can finally allow oom reaper also to tasks which share their mm with kthreads. Link: http://lkml.kernel.org/r/1472119394-11342-10-git-send-email-mhocko@kernel.orgSigned-off-by:
-
Michal Hocko authored
There are only few use_mm() users in the kernel right now. Most of them write to the target memory but vhost driver relies on copy_from_user/get_user from a kernel thread context. This makes it impossible to reap the memory of an oom victim which shares the mm with the vhost kernel thread because it could see a zero page unexpectedly and theoretically make an incorrect decision visible outside of the killed task context. To quote Michael S. Tsirkin: : Getting an error from __get_user and friends is handled gracefully. : Getting zero instead of a real value will cause userspace : memory corruption. The vhost kernel thread is bound to an open fd of the vhost device which is not tight to the mm owner life cycle in general. The device fd can be inherited or passed over to another process which means that we really have to be careful about unexpected memory corruption because unlike for normal oom victims the result will be visible outside of the oom victim context. Make sure that no kthread context (users of use_mm) can ever see corrupted data because of the oom reaper and hook into the page fault path by checking MMF_UNSTABLE mm flag. __oom_reap_task_mm will set the flag before it starts unmapping the address space while the flag is checked after the page fault has been handled. If the flag is set then SIGBUS is triggered so any g-u-p user will get a error code. Regular tasks do not need this protection because all which share the mm are killed when the mm is reaped and so the corruption will not outlive them. This patch shouldn't have any visible effect at this moment because the OOM killer doesn't invoke oom reaper for tasks with mm shared with kthreads yet. Link: http://lkml.kernel.org/r/1472119394-11342-9-git-send-email-mhocko@kernel.orgSigned-off-by:
"Michael S. Tsirkin" <mst@redhat.com> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Oleg Nesterov <oleg@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Vladimir Davydov <vdavydov@parallels.com> Signed-off-by:
-
Tetsuo Handa authored
There are no users of exit_oom_victim on !current task anymore so enforce the API to always work on the current. Link: http://lkml.kernel.org/r/1472119394-11342-8-git-send-email-mhocko@kernel.orgSigned-off-by:
Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by:
-
Michal Hocko authored
Commit 74070542 ("oom, suspend: fix oom_reaper vs. oom_killer_disable race") has workaround an existing race between oom_killer_disable and oom_reaper by adding another round of try_to_freeze_tasks after the oom killer was disabled. This was the easiest thing to do for a late 4.7 fix. Let's fix it properly now. After "oom: keep mm of the killed task available" we no longer have to call exit_oom_victim from the oom reaper because we have stable mm available and hide the oom_reaped mm by MMF_OOM_SKIP flag. So let's remove exit_oom_victim and the race described in the above commit doesn't exist anymore if. Unfortunately this alone is not sufficient for the oom_killer_disable usecase because now we do not have any reliable way to reach exit_oom_victim (the victim might get stuck on a way to exit for an unbounded amount of time). OOM killer can cope with that by checking mm flags and move on to another victim but we cannot do the same for oom_killer_disable as we would lose the guarantee of no further interference of the victim with the rest of the system. What we can do instead is to cap the maximum time the oom_killer_disable waits for victims. The only current user of this function (pm suspend) already has a concept of timeout for back off so we can reuse the same value there. Let's drop set_freezable for the oom_reaper kthread because it is no longer needed as the reaper doesn't wake or thaw any processes. Link: http://lkml.kernel.org/r/1472119394-11342-7-git-send-email-mhocko@kernel.orgSigned-off-by:
-
Michal Hocko authored
After "oom: keep mm of the killed task available" we can safely detect an oom victim by checking task->signal->oom_mm so we do not need the signal_struct counter anymore so let's get rid of it. This alone wouldn't be sufficient for nommu archs because exit_oom_victim doesn't hide the process from the oom killer anymore. We can, however, mark the mm with a MMF flag in __mmput. We can reuse MMF_OOM_REAPED and rename it to a more generic MMF_OOM_SKIP. Link: http://lkml.kernel.org/r/1472119394-11342-6-git-send-email-mhocko@kernel.orgSigned-off-by:
-
Michal Hocko authored
Lockdep complains that __mmdrop is not safe from the softirq context: ================================= [ INFO: inconsistent lock state ] 4.6.0-oomfortification2-00011-geeb3eadeab96-dirty #949 Tainted: G W --------------------------------- inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. swapper/1/0 [HC0[0]:SC1[1]:HE1:SE0] takes: (pgd_lock){+.?...}, at: pgd_free+0x19/0x6b {SOFTIRQ-ON-W} state was registered at: __lock_acquire+0xa06/0x196e lock_acquire+0x139/0x1e1 _raw_spin_lock+0x32/0x41 __change_page_attr_set_clr+0x2a5/0xacd change_page_attr_set_clr+0x16f/0x32c set_memory_nx+0x37/0x3a free_init_pages+0x9e/0xc7 alternative_instructions+0xa2/0xb3 check_bugs+0xe/0x2d start_kernel+0x3ce/0x3ea x86_64_start_reservations+0x2a/0x2c x86_64_start_kernel+0x17a/0x18d irq event stamp: 105916 hardirqs last enabled at (105916): free_hot_cold_page+0x37e/0x390 hardirqs last disabled at (105915): free_hot_cold_page+0x2c1/0x390 softirqs last enabled at (105878): _local_bh_enable+0x42/0x44 softirqs last disabled at (105879): irq_exit+0x6f/0xd1 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pgd_lock); <Interrupt> lock(pgd_lock); *** DEADLOCK *** 1 lock held by swapper/1/0: #0: (rcu_callback){......}, at: rcu_process_callbacks+0x390/0x800 stack backtrace: CPU: 1 PID: 0 Comm: swapper/1 Tainted: G W 4.6.0-oomfortification2-00011-geeb3eadeab96-dirty #949 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Debian-1.8.2-1 04/01/2014 Call Trace: <IRQ> print_usage_bug.part.25+0x259/0x268 mark_lock+0x381/0x567 __lock_acquire+0x993/0x196e lock_acquire+0x139/0x1e1 _raw_spin_lock+0x32/0x41 pgd_free+0x19/0x6b __mmdrop+0x25/0xb9 __put_task_struct+0x103/0x11e delayed_put_task_struct+0x157/0x15e rcu_process_callbacks+0x660/0x800 __do_softirq+0x1ec/0x4d5 irq_exit+0x6f/0xd1 smp_apic_timer_interrupt+0x42/0x4d apic_timer_interrupt+0x8e/0xa0 <EOI> arch_cpu_idle+0xf/0x11 default_idle_call+0x32/0x34 cpu_startup_entry+0x20c/0x399 start_secondary+0xfe/0x101 More over commit a79e53d8 ("x86/mm: Fix pgd_lock deadlock") was explicit about pgd_lock not to be called from the irq context. This means that __mmdrop called from free_signal_struct has to be postponed to a user context. We already have a similar mechanism for mmput_async so we can use it here as well. This is safe because mm_count is pinned by mm_users. This fixes bug introduced by "oom: keep mm of the killed task available" Link: http://lkml.kernel.org/r/1472119394-11342-5-git-send-email-mhocko@kernel.orgSigned-off-by: -
Michal Hocko authored
oom_reap_task has to call exit_oom_victim in order to make sure that the oom vicim will not block the oom killer for ever. This is, however, opening new problems (e.g oom_killer_disable exclusion - see commit 74070542 ("oom, suspend: fix oom_reaper vs. oom_killer_disable race")). exit_oom_victim should be only called from the victim's context ideally. One way to achieve this would be to rely on per mm_struct flags. We already have MMF_OOM_REAPED to hide a task from the oom killer since "mm, oom: hide mm which is shared with kthread or global init". The problem is that the exit path: do_exit exit_mm tsk->mm = NULL; mmput __mmput exit_oom_victim doesn't guarantee that exit_oom_victim will get called in a bounded amount of time. At least exit_aio depends on IO which might get blocked due to lack of memory and who knows what else is lurking there. This patch takes a different approach. We remember tsk->mm into the signal_struct and bind it to the signal struct life time for all oom victims. __oom_reap_task_mm as well as oom_scan_process_thread do not have to rely on find_lock_task_mm anymore and they will have a reliable reference to the mm struct. As a result all the oom specific communication inside the OOM killer can be done via tsk->signal->oom_mm. Increasing the signal_struct for something as unlikely as the oom killer is far from ideal but this approach will make the code much more reasonable and long term we even might want to move task->mm into the signal_struct anyway. In the next step we might want to make the oom killer exclusion and access to memory reserves completely independent which would be also nice. Link: http://lkml.kernel.org/r/1472119394-11342-4-git-send-email-mhocko@kernel.orgSigned-off-by:
-
Tetsuo Handa authored
"mm, oom_reaper: do not attempt to reap a task twice" tried to give the OOM reaper one more chance to retry using MMF_OOM_NOT_REAPABLE flag. But the usefulness of the flag is rather limited and actually never shown in practice. If the flag is set, it means that the holder of mm->mmap_sem cannot call up_write() due to presumably being blocked at unkillable wait waiting for other thread's memory allocation. But since one of threads sharing that mm will queue that mm immediately via task_will_free_mem() shortcut (otherwise, oom_badness() will select the same mm again due to oom_score_adj value unchanged), retrying MMF_OOM_NOT_REAPABLE mm is unlikely helpful. Let's always set MMF_OOM_REAPED. Link: http://lkml.kernel.org/r/1472119394-11342-3-git-send-email-mhocko@kernel.orgSigned-off-by:
-
Tetsuo Handa authored
Patch series "fortify oom killer even more", v2. This patch (of 9): __oom_reap_task() can be simplified a bit if it receives a valid mm from oom_reap_task() which also uses that mm when __oom_reap_task() failed. We can drop one find_lock_task_mm() call and also make the __oom_reap_task() code flow easier to follow. Moreover, this will make later patch in the series easier to review. Pinning mm's mm_count for longer time is not really harmful because this will not pin much memory. This patch doesn't introduce any functional change. Link: http://lkml.kernel.org/r/1472119394-11342-2-git-send-email-mhocko@kernel.orgSigned-off-by:
-
Huang Ying authored
This is a code clean up patch without functionality changes. The swap_cluster_list data structure and its operations are introduced to provide some better encapsulation for the free cluster and discard cluster list operations. This avoid some code duplication, improved the code readability, and reduced the total line number. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/1472067356-16004-1-git-send-email-ying.huang@intel.comSigned-off-by:
Minchan Kim <minchan@kernel.org> Acked-by:
-
Alexey Dobriyan authored
Current code doesn't count first FIND operation after VMA cache flush (which happen surprisingly often) artificially increasing cache hit ratio. On my regular setup the difference is: Before After ========================================================== * boot, login into KDE vmacache_find_calls 446216 vmacache_find_calls 492741 vmacache_find_hits 277596 vmacache_find_hits 276096 ~62.2% ~56.0% * rebuild kernel (no changes to code, usual config) vmacache_find_calls 1943007 vmacache_find_calls 2083718 vmacache_find_hits 1246123 vmacache_find_hits 1244146 ~64.1% ~59.7% * rebuild kernel (full rebuild, usual config) vmacache_find_calls 32163155 vmacache_find_calls 33677183 vmacache_find_hits 27889956 vmacache_find_hits 27877591 ~88.2% ~84.3% Total: ~4% cache hit ratio. If someone is counting _relative_ cache _miss_ ratio, misreporting is much higher. Link: http://lkml.kernel.org/r/20160822225009.GA3934@p183.telecom.bySigned-off-by:
Alexey Dobriyan <adobriyan@gmail.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Signed-off-by:
-
James Morse authored
Modify the comment describing struct mm_walk->test_walk()s behaviour to match the comment on walk_page_test() and the behaviour of walk_page_vma(). Fixes: fafaa426 ("pagewalk: improve vma handling") Link: http://lkml.kernel.org/r/1471622518-21980-1-git-send-email-james.morse@arm.comSigned-off-by:
James Morse <james.morse@arm.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by:
-
Bart Van Assche authored
If a fatal signal has been received, fail immediately instead of trying to read more data. If wait_on_page_locked_killable() was interrupted then this page is most likely is not PageUptodate() and in this case do_generic_file_read() will fail after lock_page_killable(). See also commit ebded027 ("mm: filemap: avoid unnecessary calls to lock_page when waiting for IO to complete during a read") [oleg@redhat.com: changelog addition] Link: http://lkml.kernel.org/r/63068e8e-8bee-b208-8441-a3c39a9d9eb6@sandisk.comSigned-off-by:
Bart Van Assche <bart.vanassche@sandisk.com> Reviewed-by:
Jan Kara <jack@suse.cz> Acked-by:
Oleg Nesterov <oleg@redhat.com> Acked-by:
-
Joonsoo Kim authored
There is a memory waste problem if we define field on struct page_ext by hard-coding. Entry size of struct page_ext includes the size of those fields even if it is disabled at runtime. Now, extra memory request at runtime is possible so page_owner don't need to define it's own fields by hard-coding. This patch removes hard-coded define and uses extra memory for storing page_owner information in page_owner. Most of code are just mechanical changes. Link: http://lkml.kernel.org/r/1471315879-32294-7-git-send-email-iamjoonsoo.kim@lge.comSigned-off-by:
Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by:
Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Michal Hocko <mhocko@kernel.org> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by:
-
Joonsoo Kim authored
Until now, if some page_ext users want to use it's own field on page_ext, it should be defined in struct page_ext by hard-coding. It has a problem that wastes memory in following situation. struct page_ext { #ifdef CONFIG_A int a; #endif #ifdef CONFIG_B int b; #endif }; Assume that kernel is built with both CONFIG_A and CONFIG_B. Even if we enable feature A and doesn't enable feature B at runtime, each entry of struct page_ext takes two int rather than one int. It's undesirable result so this patch tries to fix it. To solve above problem, this patch implements to support extra space allocation at runtime. When need() callback returns true, it's extra memory requirement is summed to entry size of page_ext. Also, offset for each user's extra memory space is returned. With this offset, user can use this extra space and there is no need to define needed field on page_ext by hard-coding. This patch only implements an infrastructure. Following patch will use it for page_owner which is only user having it's own fields on page_ext. Link: http://lkml.kernel.org/r/1471315879-32294-6-git-send-email-iamjoonsoo.kim@lge.comSigned-off-by: -
Joonsoo Kim authored
Here, 'offset' means entry index in page_ext array. Following patch will use 'offset' for field offset in each entry so rename current 'offset' to prevent confusion. Link: http://lkml.kernel.org/r/1471315879-32294-5-git-send-email-iamjoonsoo.kim@lge.comSigned-off-by:
-
Joonsoo Kim authored
There is no reason that page_owner specific function resides on vmstat.c. Link: http://lkml.kernel.org/r/1471315879-32294-4-git-send-email-iamjoonsoo.kim@lge.comSigned-off-by:
Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Acked-by:
-
Joonsoo Kim authored
What debug_pagealloc does is just mapping/unmapping page table. Basically, it doesn't need additional memory space to memorize something. But, with guard page feature, it requires additional memory to distinguish if the page is for guard or not. Guard page is only used when debug_guardpage_minorder is non-zero so this patch removes additional memory allocation (page_ext) if debug_guardpage_minorder is zero. It saves memory if we just use debug_pagealloc and not guard page. Link: http://lkml.kernel.org/r/1471315879-32294-3-git-send-email-iamjoonsoo.kim@lge.comSigned-off-by:
-
Joonsoo Kim authored
Patch series "Reduce memory waste by page extension user". This patchset tries to reduce memory waste by page extension user. First case is architecture supported debug_pagealloc. It doesn't requires additional memory if guard page isn't used. 8 bytes per page will be saved in this case. Second case is related to page owner feature. Until now, if page_ext users want to use it's own fields on page_ext, fields should be defined in struct page_ext by hard-coding. It has a following problem. struct page_ext { #ifdef CONFIG_A int a; #endif #ifdef CONFIG_B int b; #endif }; Assume that kernel is built with both CONFIG_A and CONFIG_B. Even if we enable feature A and doesn't enable feature B at runtime, each entry of struct page_ext takes two int rather than one int. It's undesirable waste so this patch tries to reduce it. By this patchset, we can save 20 bytes per page dedicated for page owner feature in some configurations. This patch (of 6): We can make code clean by moving decision condition for set_page_guard() into set_page_guard() itself. It will help code readability. There is no functional change. Link: http://lkml.kernel.org/r/1471315879-32294-2-git-send-email-iamjoonsoo.kim@lge.comSigned-off-by: -
Michal Hocko authored
throttle_vm_writeout() was introduced back in 2005 to fix OOMs caused by excessive pageout activity during the reclaim. Too many pages could be put under writeback therefore LRUs would be full of unreclaimable pages until the IO completes and in turn the OOM killer could be invoked. There have been some important changes introduced since then in the reclaim path though. Writers are throttled by balance_dirty_pages when initiating the buffered IO and later during the memory pressure, the direct reclaim is throttled by wait_iff_congested if the node is considered congested by dirty pages on LRUs and the underlying bdi is congested by the queued IO. The kswapd is throttled as well if it encounters pages marked for immediate reclaim or under writeback which signals that that there are too many pages under writeback already. Finally should_reclaim_retry does congestion_wait if the reclaim cannot make any progress and there are too many dirty/writeback pages. Another important aspect is that we do not issue any IO from the direct reclaim context anymore. In a heavy parallel load this could queue a lot of IO which would be very scattered and thus unefficient which would just make the problem worse. This three mechanisms should throttle and keep the amount of IO in a steady state even under heavy IO and memory pressure so yet another throttling point doesn't really seem helpful. Quite contrary, Mikulas Patocka has reported that swap backed by dm-crypt doesn't work properly because the swapout IO cannot make sufficient progress as the writeout path depends on dm_crypt worker which has to allocate memory to perform the encryption. In order to guarantee a forward progress it relies on the mempool allocator. mempool_alloc(), however, prefers to use the underlying (usually page) allocator before it grabs objects from the pool. Such an allocation can dive into the memory reclaim and consequently to throttle_vm_writeout. If there are too many dirty or pages under writeback it will get throttled even though it is in fact a flusher to clear pending pages. kworker/u4:0 D ffff88003df7f438 10488 6 2 0x00000000 Workqueue: kcryptd kcryptd_crypt [dm_crypt] Call Trace: schedule+0x3c/0x90 schedule_timeout+0x1d8/0x360 io_schedule_timeout+0xa4/0x110 congestion_wait+0x86/0x1f0 throttle_vm_writeout+0x44/0xd0 shrink_zone_memcg+0x613/0x720 shrink_zone+0xe0/0x300 do_try_to_free_pages+0x1ad/0x450 try_to_free_pages+0xef/0x300 __alloc_pages_nodemask+0x879/0x1210 alloc_pages_current+0xa1/0x1f0 new_slab+0x2d7/0x6a0 ___slab_alloc+0x3fb/0x5c0 __slab_alloc+0x51/0x90 kmem_cache_alloc+0x27b/0x310 mempool_alloc_slab+0x1d/0x30 mempool_alloc+0x91/0x230 bio_alloc_bioset+0xbd/0x260 kcryptd_crypt+0x114/0x3b0 [dm_crypt] Let's just drop throttle_vm_writeout altogether. It is not very much helpful anymore. I have tried to test a potential writeback IO runaway similar to the one described in the original patch which has introduced that [1]. Small virtual machine (512MB RAM, 4 CPUs, 2G of swap space and disk image on a rather slow NFS in a sync mode on the host) with 8 parallel writers each writing 1G worth of data. As soon as the pagecache fills up and the direct reclaim hits then I start anon memory consumer in a loop (allocating 300M and exiting after populating it) in the background to make the memory pressure even stronger as well as to disrupt the steady state for the IO. The direct reclaim is throttled because of the congestion as well as kswapd hitting congestion_wait due to nr_immediate but throttle_vm_writeout doesn't ever trigger the sleep throughout the test. Dirty+writeback are close to nr_dirty_threshold with some fluctuations caused by the anon consumer. [1] https://www2.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.9-rc1/2.6.9-rc1-mm3/broken-out/vm-pageout-throttling.patch Link: http://lkml.kernel.org/r/1471171473-21418-1-git-send-email-mhocko@kernel.orgSigned-off-by:Mikulas Patocka <mpatocka@redhat.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: NeilBrown <neilb@suse.com> Cc: Ondrej Kozina <okozina@redhat.com> Signed-off-by:
-
Xishi Qiu authored
On x86_64 MAX_ORDER_NR_PAGES is usually 4M, and a pageblock is usually 2M, so we only set one pageblock's migratetype in deferred_free_range() if pfn is aligned to MAX_ORDER_NR_PAGES. That means it causes uninitialized migratetype blocks, you can see from "cat /proc/pagetypeinfo", almost half blocks are Unmovable. Also we missed freeing the last block in deferred_init_memmap(), it causes memory leak. Fixes: ac5d2539 ("mm: meminit: reduce number of times pageblocks are set during struct page init") Link: http://lkml.kernel.org/r/57A3260F.4050709@huawei.comSigned-off-by:
Xishi Qiu <qiuxishi@huawei.com> Cc: Taku Izumi <izumi.taku@jp.fujitsu.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@suse.com> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com> Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Signed-off-by:
-
Xishi Qiu authored
Commit 342332e6 ("mm/page_alloc.c: introduce kernelcore=mirror option") rewrote the calculation of node spanned pages. But when we have a movable node, the size of node spanned pages is double added. That's because we have an empty normal zone, the present pages is zero, but its spanned pages is not zero. e.g. Zone ranges: DMA [mem 0x0000000000001000-0x0000000000ffffff] DMA32 [mem 0x0000000001000000-0x00000000ffffffff] Normal [mem 0x0000000100000000-0x0000007c7fffffff] Movable zone start for each node Node 1: 0x0000001080000000 Node 2: 0x0000002080000000 Node 3: 0x0000003080000000 Node 4: 0x0000003c80000000 Node 5: 0x0000004c80000000 Node 6: 0x0000005c80000000 Early memory node ranges node 0: [mem 0x0000000000001000-0x000000000009ffff] node 0: [mem 0x0000000000100000-0x000000007552afff] node 0: [mem 0x000000007bd46000-0x000000007bd46fff] node 0: [mem 0x000000007bdcd000-0x000000007bffffff] node 0: [mem 0x0000000100000000-0x000000107fffffff] node 1: [mem 0x0000001080000000-0x000000207fffffff] node 2: [mem 0x0000002080000000-0x000000307fffffff] node 3: [mem 0x0000003080000000-0x0000003c7fffffff] node 4: [mem 0x0000003c80000000-0x0000004c7fffffff] node 5: [mem 0x0000004c80000000-0x0000005c7fffffff] node 6: [mem 0x0000005c80000000-0x0000006c7fffffff] node 7: [mem 0x0000006c80000000-0x0000007c7fffffff] node1: Normal, start=0x1080000, present=0x0, spanned=0x1000000 Movable, start=0x1080000, present=0x1000000, spanned=0x1000000 pgdat, start=0x1080000, present=0x1000000, spanned=0x2000000 After this patch, the problem is fixed. node1: Normal, start=0x0, present=0x0, spanned=0x0 Movable, start=0x1080000, present=0x1000000, spanned=0x1000000 pgdat, start=0x1080000, present=0x1000000, spanned=0x1000000 Link: http://lkml.kernel.org/r/57A325E8.6070100@huawei.comSigned-off-by:
-
Vlastimil Babka authored
The compaction_ready() is used during direct reclaim for costly order allocations to skip reclaim for zones where compaction should be attempted instead. It's combining the standard compaction_suitable() check with its own watermark check based on high watermark with extra gap, and the result is confusing at best. This patch attempts to better structure and document the checks involved. First, compaction_suitable() can determine that the allocation should either succeed already, or that compaction doesn't have enough free pages to proceed. The third possibility is that compaction has enough free pages, but we still decide to reclaim first - unless we are already above the high watermark with gap. This does not mean that the reclaim will actually reach this watermark during single attempt, this is rather an over-reclaim protection. So document the code as such. The check for compaction_deferred() is removed completely, as it in fact had no proper role here. The result after this patch is mainly a less confusing code. We also skip some over-reclaim in cases where the allocation should already succed. Link: http://lkml.kernel.org/r/20160810091226.6709-12-vbabka@suse.czSigned-off-by:
Lorenzo Stoakes <lstoakes@gmail.com> Acked-by:
-
Vlastimil Babka authored
The __compaction_suitable() function checks the low watermark plus a compact_gap() gap to decide if there's enough free memory to perform compaction. Then __isolate_free_page uses low watermark check to decide if particular free page can be isolated. In the latter case, using low watermark is needlessly pessimistic, as the free page isolations are only temporary. For __compaction_suitable() the higher watermark makes sense for high-order allocations where more freepages increase the chance of success, and we can typically fail with some order-0 fallback when the system is struggling to reach that watermark. But for low-order allocation, forming the page should not be that hard. So using low watermark here might just prevent compaction from even trying, and eventually lead to OOM killer even if we are above min watermarks. So after this patch, we use min watermark for non-costly orders in __compaction_suitable(), and for all orders in __isolate_free_page(). [vbabka@suse.cz: clarify __isolate_free_page() comment] Link: http://lkml.kernel.org/r/7ae4baec-4eca-e70b-2a69-94bea4fb19fa@suse.cz Link: http://lkml.kernel.org/r/20160810091226.6709-11-vbabka@suse.czSigned-off-by:
-
Vlastimil Babka authored
The __compaction_suitable() function checks the low watermark plus a compact_gap() gap to decide if there's enough free memory to perform compaction. This check uses direct compactor's alloc_flags, but that's wrong, since these flags are not applicable for freepage isolation. For example, alloc_flags may indicate access to memory reserves, making compaction proceed, and then fail watermark check during the isolation. A similar problem exists for ALLOC_CMA, which may be part of alloc_flags, but not during freepage isolation. In this case however it makes sense to use ALLOC_CMA both in __compaction_suitable() and __isolate_free_page(), since there's actually nothing preventing the freepage scanner to isolate from CMA pageblocks, with the assumption that a page that could be migrated once by compaction can be migrated also later by CMA allocation. Thus we should count pages in CMA pageblocks when considering compaction suitability and when isolating freepages. To sum up, this patch should remove some false positives from __compaction_suitable(), and allow compaction to proceed when free pages required for compaction reside in the CMA pageblocks. Link: http://lkml.kernel.org/r/20160810091226.6709-10-vbabka@suse.czSigned-off-by:
-
Vlastimil Babka authored
Compaction uses a watermark gap of (2UL << order) pages at various places and it's not immediately obvious why. Abstract it through a compact_gap() wrapper to create a single place with a thorough explanation. [vbabka@suse.cz: clarify the comment of compact_gap()] Link: http://lkml.kernel.org/r/7b6aed1f-fdf8-2063-9ff4-bbe4de712d37@suse.cz Link: http://lkml.kernel.org/r/20160810091226.6709-9-vbabka@suse.czSigned-off-by:
-
Vlastimil Babka authored
The __compact_finished() function uses low watermark in a check that has to pass if the direct compaction is to finish and allocation should succeed. This is too pessimistic, as the allocation will typically use min watermark. It may happen that during compaction, we drop below the low watermark (due to parallel activity), but still form the target high-order page. By checking against low watermark, we might needlessly continue compaction. Similarly, __compaction_suitable() uses low watermark in a check whether allocation can succeed without compaction. Again, this is unnecessarily pessimistic. After this patch, these check will use direct compactor's alloc_flags to determine the watermark, which is effectively the min watermark. Link: http://lkml.kernel.org/r/20160810091226.6709-8-vbabka@suse.czSigned-off-by:
-
Vlastimil Babka authored
During reclaim/compaction loop, it's desirable to get a final answer from unsuccessful compaction so we can either fail the allocation or invoke the OOM killer. However, heuristics such as deferred compaction or pageblock skip bits can cause compaction to skip parts or whole zones and lead to premature OOM's, failures or excessive reclaim/compaction retries. To remedy this, we introduce a new direct compaction priority called COMPACT_PRIO_SYNC_FULL, which instructs direct compaction to: - ignore deferred compaction status for a zone - ignore pageblock skip hints - ignore cached scanner positions and scan the whole zone The new priority should get eventually picked up by should_compact_retry() and this should improve success rates for costly allocations using __GFP_REPEAT, such as hugetlbfs allocations, and reduce some corner-case OOM's for non-costly allocations. Link: http://lkml.kernel.org/r/20160810091226.6709-6-vbabka@suse.cz [vbabka@suse.cz: use the MIN_COMPACT_PRIORITY alias] Link: http://lkml.kernel.org/r/d443b884-87e7-1c93-8684-3a3a35759fb1@suse.czSigned-off-by:
-
Vlastimil Babka authored
Joonsoo has reminded me that in a later patch changing watermark checks throughout compaction I forgot to update checks in try_to_compact_pages() and compactd_do_work(). Closer inspection however shows that they are redundant now in the success case, because compact_zone() now reliably reports this with COMPACT_SUCCESS. So effectively the checks just repeat (a subset) of checks that have just passed. So instead of checking watermarks again, just test the return value. Note it's also possible that compaction would declare failure e.g. because its find_suitable_fallback() is more strict than simple watermark check, and then the watermark check we are removing would then still succeed. After this patch this is not possible and it's arguably better, because for long-term fragmentation avoidance we should rather try a different zone than allocate with the unsuitable fallback. If compaction of all zones fail and the allocation is important enough, it will retry and succeed anyway. Also remove the stray "bool success" variable from kcompactd_do_work(). Link: http://lkml.kernel.org/r/20160810091226.6709-5-vbabka@suse.czSigned-off-by:
Michal Hocko <mhocko@kernel.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by:
-
Vlastimil Babka authored
COMPACT_PARTIAL has historically meant that compaction returned after doing some work without fully compacting a zone. It however didn't distinguish if compaction terminated because it succeeded in creating the requested high-order page. This has changed recently and now we only return COMPACT_PARTIAL when compaction thinks it succeeded, or the high-order watermark check in compaction_suitable() passes and no compaction needs to be done. So at this point we can make the return value clearer by renaming it to COMPACT_SUCCESS. The next patch will remove some redundant tests for success where compaction just returned COMPACT_SUCCESS. Link: http://lkml.kernel.org/r/20160810091226.6709-4-vbabka@suse.czSigned-off-by:
-
Vlastimil Babka authored
Since kswapd compaction moved to kcompactd, compact_pgdat() is not called anymore, so we remove it. The only caller of __compact_pgdat() is compact_node(), so we merge them and remove code that was only reachable from kswapd. Link: http://lkml.kernel.org/r/20160810091226.6709-3-vbabka@suse.czSigned-off-by:
-
Vlastimil Babka authored
Patch series "make direct compaction more deterministic") This is mostly a followup to Michal's oom detection rework, which highlighted the need for direct compaction to provide better feedback in reclaim/compaction loop, so that it can reliably recognize when compaction cannot make further progress, and allocation should invoke OOM killer or fail. We've discussed this at LSF/MM [1] where I proposed expanding the async/sync migration mode used in compaction to more general "priorities". This patchset adds one new priority that just overrides all the heuristics and makes compaction fully scan all zones. I don't currently think that we need more fine-grained priorities, but we'll see. Other than that there's some smaller fixes and cleanups, mainly related to the THP-specific hacks. I've tested this with stress-highalloc in GFP_KERNEL order-4 and THP-like order-9 scenarios. There's some improvement for compaction stats for the order-4, which is likely due to the better watermarks handling. In the previous version I reported mostly noise wrt compaction stats, and decreased direct reclaim - now the reclaim is without difference. I believe this is due to the less aggressive compaction priority increase in patch 6. "before" is a mmotm tree prior to 4.7 release plus the first part of the series that was sent and merged separately before after order-4: Compaction stalls 27216 30759 Compaction success 19598 25475 Compaction failures 7617 5283 Page migrate success 370510 464919 Page migrate failure 25712 27987 Compaction pages isolated 849601 1041581 Compaction migrate scanned 143146541 101084990 Compaction free scanned 208355124 144863510 Compaction cost 1403 1210 order-9: Compaction stalls 7311 7401 Compaction success 1634 1683 Compaction failures 5677 5718 Page migrate success 194657 183988 Page migrate failure 4753 4170 Compaction pages isolated 498790 456130 Compaction migrate scanned 565371 524174 Compaction free scanned 4230296 4250744 Compaction cost 215 203 [1] https://lwn.net/Articles/684611/ This patch (of 11): A recent patch has added whole_zone flag that compaction sets when scanning starts from the zone boundary, in order to report that zone has been fully scanned in one attempt. For allocations that want to try really hard or cannot fail, we will want to introduce a mode where scanning whole zone is guaranteed regardless of the cached positions. This patch reuses the whole_zone flag in a way that if it's already passed true to compaction, the cached scanner positions are ignored. Employing this flag during reclaim/compaction loop will be done in the next patch. This patch however converts compaction invoked from userspace via procfs to use this flag. Before this patch, the cached positions were first reset to zone boundaries and then read back from struct zone, so there was a window where a parallel compaction could replace the reset values, making the manual compaction less effective. Using the flag instead of performing reset is more robust. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20160810091226.6709-2-vbabka@suse.czSigned-off-by:
-
