- 17 Aug, 2018 40 commits
-
-
Kirill Tkhai authored
Imagine a big node with many cpus, memory cgroups and containers. Let we have 200 containers, every container has 10 mounts, and 10 cgroups. All container tasks don't touch foreign containers mounts. If there is intensive pages write, and global reclaim happens, a writing task has to iterate over all memcgs to shrink slab, before it's able to go to shrink_page_list(). Iteration over all the memcg slabs is very expensive: the task has to visit 200 * 10 = 2000 shrinkers for every memcg, and since there are 2000 memcgs, the total calls are 2000 * 2000 = 4000000. So, the shrinker makes 4 million do_shrink_slab() calls just to try to isolate SWAP_CLUSTER_MAX pages in one of the actively writing memcg via shrink_page_list(). I've observed a node spending almost 100% in kernel, making useless iteration over already shrinked slab. This patch adds bitmap of memcg-aware shrinkers to memcg. The size of the bitmap depends on bitmap_nr_ids, and during memcg life it's maintained to be enough to fit bitmap_nr_ids shrinkers. Every bit in the map is related to corresponding shrinker id. Next patches will maintain set bit only for really charged memcg. This will allow shrink_slab() to increase its performance in significant way. See the last patch for the numbers. [ktkhai@virtuozzo.com: v9] Link: http://lkml.kernel.org/r/153112549031.4097.3576147070498769979.stgit@localhost.localdomain [ktkhai@virtuozzo.com: add comment to mem_cgroup_css_online()] Link: http://lkml.kernel.org/r/521f9e5f-c436-b388-fe83-4dc870bfb489@virtuozzo.com Link: http://lkml.kernel.org/r/153063056619.1818.12550500883688681076.stgit@localhost.localdomainSigned-off-by:
Kirill Tkhai <ktkhai@virtuozzo.com> Acked-by:
Vladimir Davydov <vdavydov.dev@gmail.com> Tested-by:
Shakeel Butt <shakeelb@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Guenter Roeck <linux@roeck-us.net> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Josef Bacik <jbacik@fb.com> Cc: Li RongQing <lirongqing@baidu.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matthias Kaehlcke <mka@chromium.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Michal Hocko <mhocko@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Roman Gushchin <guro@fb.com> Cc: Sahitya Tummala <stummala@codeaurora.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Waiman Long <longman@redhat.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Kirill Tkhai authored
Next patch requires these defines are above their current position, so here they are moved to declarations. Link: http://lkml.kernel.org/r/153063055665.1818.5200425793649695598.stgit@localhost.localdomainSigned-off-by:
-
Kirill Tkhai authored
Introduce shrinker::id number, which is used to enumerate memcg-aware shrinkers. The number start from 0, and the code tries to maintain it as small as possible. This will be used to represent a memcg-aware shrinkers in memcg shrinkers map. Since all memcg-aware shrinkers are based on list_lru, which is per-memcg in case of !CONFIG_MEMCG_KMEM only, the new functionality will be under this config option. [ktkhai@virtuozzo.com: v9] Link: http://lkml.kernel.org/r/153112546435.4097.10607140323811756557.stgit@localhost.localdomain Link: http://lkml.kernel.org/r/153063054586.1818.6041047871606697364.stgit@localhost.localdomainSigned-off-by:
-
Kirill Tkhai authored
Introduce new config option, which is used to replace repeating CONFIG_MEMCG && !CONFIG_SLOB pattern. Next patches add a little more memcg+kmem related code, so let's keep the defines more clearly. Link: http://lkml.kernel.org/r/153063053670.1818.15013136946600481138.stgit@localhost.localdomainSigned-off-by:
-
Kirill Tkhai authored
Patch series "Improve shrink_slab() scalability (old complexity was O(n^2), new is O(n))", v8. This patcheset solves the problem with slow shrink_slab() occuring on the machines having many shrinkers and memory cgroups (i.e., with many containers). The problem is complexity of shrink_slab() is O(n^2) and it grows too fast with the growth of containers numbers. Let us have 200 containers, and every container has 10 mounts and 10 cgroups. All container tasks are isolated, and they don't touch foreign containers mounts. In case of global reclaim, a task has to iterate all over the memcgs and to call all the memcg-aware shrinkers for all of them. This means, the task has to visit 200 * 10 = 2000 shrinkers for every memcg, and since there are 2000 memcgs, the total calls of do_shrink_slab() are 2000 * 2000 = 4000000. 4 million calls are not a number operations, which can takes 1 cpu cycle. E.g., super_cache_count() accesses at least two lists, and makes arifmetical calculations. Even, if there are no charged objects, we do these calculations, and replaces cpu caches by read memory. I observed nodes spending almost 100% time in kernel, in case of intensive writing and global reclaim. The writer consumes pages fast, but it's need to shrink_slab() before the reclaimer reached shrink pages function (and frees SWAP_CLUSTER_MAX pages). Even if there is no writing, the iterations just waste the time, and slows reclaim down. Let's see the small test below: $echo 1 > /sys/fs/cgroup/memory/memory.use_hierarchy $mkdir /sys/fs/cgroup/memory/ct $echo 4000M > /sys/fs/cgroup/memory/ct/memory.kmem.limit_in_bytes $for i in `seq 0 4000`; do mkdir /sys/fs/cgroup/memory/ct/$i; echo $$ > /sys/fs/cgroup/memory/ct/$i/cgroup.procs; mkdir -p s/$i; mount -t tmpfs $i s/$i; touch s/$i/file; done Then, let's see drop caches time (5 sequential calls): $time echo 3 > /proc/sys/vm/drop_caches 0.00user 13.78system 0:13.78elapsed 99%CPU 0.00user 5.59system 0:05.60elapsed 99%CPU 0.00user 5.48system 0:05.48elapsed 99%CPU 0.00user 8.35system 0:08.35elapsed 99%CPU 0.00user 8.34system 0:08.35elapsed 99%CPU The last four calls don't actually shrink anything. So, the iterations over slab shrinkers take 5.48 seconds. Not so good for scalability. The patchset solves the problem by making shrink_slab() of O(n) complexity. There are following functional actions: 1) Assign id to every registered memcg-aware shrinker. 2) Maintain per-memcgroup bitmap of memcg-aware shrinkers, and set a shrinker-related bit after the first element is added to lru list (also, when removed child memcg elements are reparanted). 3) Split memcg-aware shrinkers and !memcg-aware shrinkers, and call a shrinker if its bit is set in memcg's shrinker bitmap. (Also, there is a functionality to clear the bit, after last element is shrinked). This gives significant performance increase. The result after patchset is applied: $time echo 3 > /proc/sys/vm/drop_caches 0.00user 1.10system 0:01.10elapsed 99%CPU 0.00user 0.00system 0:00.01elapsed 64%CPU 0.00user 0.01system 0:00.01elapsed 82%CPU 0.00user 0.00system 0:00.01elapsed 64%CPU 0.00user 0.01system 0:00.01elapsed 82%CPU The results show the performance increases at least in 548 times. So, the patchset makes shrink_slab() of less complexity and improves the performance in such types of load I pointed. This will give a profit in case of !global reclaim case, since there also will be less do_shrink_slab() calls. This patch (of 17): These two pairs of blocks of code are under the same #ifdef #else #endif. Link: http://lkml.kernel.org/r/153063052519.1818.9393587113056959488.stgit@localhost.localdomainSigned-off-by: -
Mike Rapoport authored
Most functions in memblock already use phys_addr_t to represent a physical address with __memblock_free_late() being an exception. This patch replaces u64 with phys_addr_t in __memblock_free_late() and switches several format strings from %llx to %pa to avoid casting from phys_addr_t to u64. Link: http://lkml.kernel.org/r/1530637506-1256-1-git-send-email-rppt@linux.vnet.ibm.comSigned-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Reviewed-by:
Pavel Tatashin <pasha.tatashin@oracle.com> Acked-by:
Michal Hocko <mhocko@suse.com> Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com> Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by:
-
Oscar Salvador authored
sparse_init_one_section() is being called from two sites: sparse_init() and sparse_add_one_section(). The former calls it from a for_each_present_section_nr() loop, and the latter marks the section as present before calling it. This means that when sparse_init_one_section() gets called, we already know that the section is present. So there is no point to double check that in the function. This removes the check and makes the function void. [ross.zwisler@linux.intel.com: fix error path in sparse_add_one_section] Link: http://lkml.kernel.org/r/20180706190658.6873-1-ross.zwisler@linux.intel.com [ross.zwisler@linux.intel.com: simplification suggested by Oscar] Link: http://lkml.kernel.org/r/20180706223358.742-1-ross.zwisler@linux.intel.com Link: http://lkml.kernel.org/r/20180702154325.12196-1-osalvador@techadventures.netSigned-off-by:
Oscar Salvador <osalvador@suse.de> Acked-by:
-
Michal Hocko authored
Commit 3812c8c8 ("mm: memcg: do not trap chargers with full callstack on OOM") has changed the ENOMEM semantic of memcg charges. Rather than invoking the oom killer from the charging context it delays the oom killer to the page fault path (pagefault_out_of_memory). This in turn means that many users (e.g. slab or g-u-p) will get ENOMEM when the corresponding memcg hits the hard limit and the memcg is is OOM. This is behavior is inconsistent with !memcg case where the oom killer is invoked from the allocation context and the allocator keeps retrying until it succeeds. The difference in the behavior is user visible. mmap(MAP_POPULATE) might result in not fully populated ranges while the mmap return code doesn't tell that to the userspace. Random syscalls might fail with ENOMEM etc. The primary motivation of the different memcg oom semantic was the deadlock avoidance. Things have changed since then, though. We have an async oom teardown by the oom reaper now and so we do not have to rely on the victim to tear down its memory anymore. Therefore we can return to the original semantic as long as the memcg oom killer is not handed over to the users space. There is still one thing to be careful about here though. If the oom killer is not able to make any forward progress - e.g. because there is no eligible task to kill - then we have to bail out of the charge path to prevent from same class of deadlocks. We have basically two options here. Either we fail the charge with ENOMEM or force the charge and allow overcharge. The first option has been considered more harmful than useful because rare inconsistencies in the ENOMEM behavior is hard to test for and error prone. Basically the same reason why the page allocator doesn't fail allocations under such conditions. The later might allow runaways but those should be really unlikely unless somebody misconfigures the system. E.g. allowing to migrate tasks away from the memcg to a different unlimited memcg with move_charge_at_immigrate disabled. Link: http://lkml.kernel.org/r/20180628151101.25307-1-mhocko@kernel.orgSigned-off-by:
Greg Thelen <gthelen@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Signed-off-by:
-
Mike Rapoport authored
The deferred memory initialization relies on section definitions, e.g PAGES_PER_SECTION, that are only available when CONFIG_SPARSEMEM=y on most architectures. Initially DEFERRED_STRUCT_PAGE_INIT depended on explicit ARCH_SUPPORTS_DEFERRED_STRUCT_PAGE_INIT configuration option, but since the commit 2e3ca40f ("mm: relax deferred struct page requirements") this requirement was relaxed and now it is possible to enable DEFERRED_STRUCT_PAGE_INIT on architectures that support DISCONTINGMEM and NO_BOOTMEM which causes build failures. For instance, setting SMP=y and DEFERRED_STRUCT_PAGE_INIT=y on arc causes the following build failure: CC mm/page_alloc.o mm/page_alloc.c: In function 'update_defer_init': mm/page_alloc.c:321:14: error: 'PAGES_PER_SECTION' undeclared (first use in this function); did you mean 'USEC_PER_SEC'? (pfn & (PAGES_PER_SECTION - 1)) == 0) { ^~~~~~~~~~~~~~~~~ USEC_PER_SEC mm/page_alloc.c:321:14: note: each undeclared identifier is reported only once for each function it appears in In file included from include/linux/cache.h:5:0, from include/linux/printk.h:9, from include/linux/kernel.h:14, from include/asm-generic/bug.h:18, from arch/arc/include/asm/bug.h:32, from include/linux/bug.h:5, from include/linux/mmdebug.h:5, from include/linux/mm.h:9, from mm/page_alloc.c:18: mm/page_alloc.c: In function 'deferred_grow_zone': mm/page_alloc.c:1624:52: error: 'PAGES_PER_SECTION' undeclared (first use in this function); did you mean 'USEC_PER_SEC'? unsigned long nr_pages_needed = ALIGN(1 << order, PAGES_PER_SECTION); ^ include/uapi/linux/kernel.h:11:47: note: in definition of macro '__ALIGN_KERNEL_MASK' #define __ALIGN_KERNEL_MASK(x, mask) (((x) + (mask)) & ~(mask)) ^~~~ include/linux/kernel.h:58:22: note: in expansion of macro '__ALIGN_KERNEL' #define ALIGN(x, a) __ALIGN_KERNEL((x), (a)) ^~~~~~~~~~~~~~ mm/page_alloc.c:1624:34: note: in expansion of macro 'ALIGN' unsigned long nr_pages_needed = ALIGN(1 << order, PAGES_PER_SECTION); ^~~~~ In file included from include/asm-generic/bug.h:18:0, from arch/arc/include/asm/bug.h:32, from include/linux/bug.h:5, from include/linux/mmdebug.h:5, from include/linux/mm.h:9, from mm/page_alloc.c:18: mm/page_alloc.c: In function 'free_area_init_node': mm/page_alloc.c:6379:50: error: 'PAGES_PER_SECTION' undeclared (first use in this function); did you mean 'USEC_PER_SEC'? pgdat->static_init_pgcnt = min_t(unsigned long, PAGES_PER_SECTION, ^ include/linux/kernel.h:812:22: note: in definition of macro '__typecheck' (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1))) ^ include/linux/kernel.h:836:24: note: in expansion of macro '__safe_cmp' __builtin_choose_expr(__safe_cmp(x, y), \ ^~~~~~~~~~ include/linux/kernel.h:904:27: note: in expansion of macro '__careful_cmp' #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <) ^~~~~~~~~~~~~ mm/page_alloc.c:6379:29: note: in expansion of macro 'min_t' pgdat->static_init_pgcnt = min_t(unsigned long, PAGES_PER_SECTION, ^~~~~ include/linux/kernel.h:836:2: error: first argument to '__builtin_choose_expr' not a constant __builtin_choose_expr(__safe_cmp(x, y), \ ^ include/linux/kernel.h:904:27: note: in expansion of macro '__careful_cmp' #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <) ^~~~~~~~~~~~~ mm/page_alloc.c:6379:29: note: in expansion of macro 'min_t' pgdat->static_init_pgcnt = min_t(unsigned long, PAGES_PER_SECTION, ^~~~~ scripts/Makefile.build:317: recipe for target 'mm/page_alloc.o' failed Let's make the DEFERRED_STRUCT_PAGE_INIT explicitly depend on SPARSEMEM as the systems that support DISCONTIGMEM do not seem to have that huge amounts of memory that would make DEFERRED_STRUCT_PAGE_INIT relevant. Link: http://lkml.kernel.org/r/1530279308-24988-1-git-send-email-rppt@linux.vnet.ibm.comSigned-off-by:
Randy Dunlap <rdunlap@infradead.org> Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com> Signed-off-by:
-
Andrey Ryabinin authored
KASAN learns about hotadded memory via the memory hotplug notifier. devm_memremap_pages() intentionally skips calling memory hotplug notifiers. So KASAN doesn't know anything about new memory added by devm_memremap_pages(). This causes a crash when KASAN tries to access non-existent shadow memory: BUG: unable to handle kernel paging request at ffffed0078000000 RIP: 0010:check_memory_region+0x82/0x1e0 Call Trace: memcpy+0x1f/0x50 pmem_do_bvec+0x163/0x720 pmem_make_request+0x305/0xac0 generic_make_request+0x54f/0xcf0 submit_bio+0x9c/0x370 submit_bh_wbc+0x4c7/0x700 block_read_full_page+0x5ef/0x870 do_read_cache_page+0x2b8/0xb30 read_dev_sector+0xbd/0x3f0 read_lba.isra.0+0x277/0x670 efi_partition+0x41a/0x18f0 check_partition+0x30d/0x5e9 rescan_partitions+0x18c/0x840 __blkdev_get+0x859/0x1060 blkdev_get+0x23f/0x810 __device_add_disk+0x9c8/0xde0 pmem_attach_disk+0x9a8/0xf50 nvdimm_bus_probe+0xf3/0x3c0 driver_probe_device+0x493/0xbd0 bus_for_each_drv+0x118/0x1b0 __device_attach+0x1cd/0x2b0 bus_probe_device+0x1ac/0x260 device_add+0x90d/0x1380 nd_async_device_register+0xe/0x50 async_run_entry_fn+0xc3/0x5d0 process_one_work+0xa0a/0x1810 worker_thread+0x87/0xe80 kthread+0x2d7/0x390 ret_from_fork+0x3a/0x50 Add kasan_add_zero_shadow()/kasan_remove_zero_shadow() - post mm_init() interface to map/unmap kasan_zero_page at requested virtual addresses. And use it to add/remove the shadow memory for hotplugged/unplugged device memory. Link: http://lkml.kernel.org/r/20180629164932.740-1-aryabinin@virtuozzo.com Fixes: 41e94a85 ("add devm_memremap_pages") Signed-off-by:
Andrey Ryabinin <aryabinin@virtuozzo.com> Reported-by:
Dave Chinner <david@fromorbit.com> Reviewed-by:
Dan Williams <dan.j.williams@intel.com> Tested-by:
-
Song Liu authored
khugepaged_enter_vma_merge() passes a stale vma->vm_flags to hugepage_vma_check(). The argument vm_flags contains the latest value. Therefore, it is necessary to pass this vm_flags into hugepage_vma_check(). With this bug, madvise(MADV_HUGEPAGE) for mmap files in shmem fails to put memory in huge pages. Here is an example of failed madvise(): /* mount /dev/shm with huge=advise: * mount -o remount,huge=advise /dev/shm */ /* create file /dev/shm/huge */ #define HUGE_FILE "/dev/shm/huge" fd = open(HUGE_FILE, O_RDONLY); ptr = mmap(NULL, FILE_SIZE, PROT_READ, MAP_PRIVATE, fd, 0); ret = madvise(ptr, FILE_SIZE, MADV_HUGEPAGE); madvise() will return 0, but this memory region is never put in huge page (check from /proc/meminfo: ShmemHugePages). Link: http://lkml.kernel.org/r/20180629181752.792831-1-songliubraving@fb.com Fixes: 02b75dc8160d ("mm: thp: register mm for khugepaged when merging vma for shmem") Signed-off-by:Song Liu <songliubraving@fb.com> Reviewed-by:
Rik van Riel <riel@surriel.com> Reviewed-by:
Yang Shi <yang.shi@linux.alibaba.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by:
-
Andrey Ryabinin authored
Signed integer overflow is undefined according to the C standard. The overflow in ksys_fadvise64_64() is deliberate, but since it is signed overflow, UBSAN complains: UBSAN: Undefined behaviour in mm/fadvise.c:76:10 signed integer overflow: 4 + 9223372036854775805 cannot be represented in type 'long long int' Use unsigned types to do math. Unsigned overflow is defined so UBSAN will not complain about it. This patch doesn't change generated code. [akpm@linux-foundation.org: add comment explaining the casts] Link: http://lkml.kernel.org/r/20180629184453.7614-1-aryabinin@virtuozzo.comSigned-off-by:
-
Colin Ian King authored
The mutexes swap_slots_cache_mutex and swap_slots_cache_enable_mutex are local to the source and do not need to be in global scope, so make them static. Cleans up sparse warnings: symbol 'swap_slots_cache_mutex' was not declared. Should it be static? symbol 'swap_slots_cache_enable_mutex' was not declared. Should it be static? Link: http://lkml.kernel.org/r/20180624182536.4937-1-colin.king@canonical.comSigned-off-by:
Colin Ian King <colin.king@canonical.com> Reviewed-by:
-
Colin Ian King authored
The functions zs_page_isolate, zs_page_migrate, zs_page_putback, lock_zspage, trylock_zspage and structure zsmalloc_aops are local to source and do not need to be in global scope, so make them static. Cleans up sparse warnings: symbol 'zs_page_isolate' was not declared. Should it be static? symbol 'zs_page_migrate' was not declared. Should it be static? symbol 'zs_page_putback' was not declared. Should it be static? symbol 'zsmalloc_aops' was not declared. Should it be static? symbol 'lock_zspage' was not declared. Should it be static? symbol 'trylock_zspage' was not declared. Should it be static? [arnd@arndb.de: hide unused lock_zspage] Link: http://lkml.kernel.org/r/20180706130924.3891230-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20180624213322.13776-1-colin.king@canonical.comSigned-off-by:
Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Reviewed-by:
-
Greg Thelen authored
Commit 93f78d88 ("writeback: move backing_dev_info->bdi_stat[] into bdi_writeback") replaced BDI_DIRTIED with WB_DIRTIED in account_page_redirty(). Update comment to track that change. BDI_DIRTIED => WB_DIRTIED BDI_WRITTEN => WB_WRITTEN Link: http://lkml.kernel.org/r/20180625171526.173483-1-gthelen@google.comSigned-off-by:
Jan Kara <jack@suse.cz> Acked-by:
Tejun Heo <tj@kernel.org> Signed-off-by:
-
Shakeel Butt authored
The buffer_head can consume a significant amount of system memory and is directly related to the amount of page cache. In our production environment we have observed that a lot of machines are spending a significant amount of memory as buffer_head and can not be left as system memory overhead. Charging buffer_head is not as simple as adding __GFP_ACCOUNT to the allocation. The buffer_heads can be allocated in a memcg different from the memcg of the page for which buffer_heads are being allocated. One concrete example is memory reclaim. The reclaim can trigger I/O of pages of any memcg on the system. So, the right way to charge buffer_head is to extract the memcg from the page for which buffer_heads are being allocated and then use targeted memcg charging API. [shakeelb@google.com: use __GFP_ACCOUNT for directed memcg charging] Link: http://lkml.kernel.org/r/20180702220208.213380-1-shakeelb@google.com Link: http://lkml.kernel.org/r/20180627191250.209150-3-shakeelb@google.comSigned-off-by:
Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@kernel.org> Cc: Jan Kara <jack@suse.cz> Cc: Amir Goldstein <amir73il@gmail.com> Cc: Greg Thelen <gthelen@google.com> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Roman Gushchin <guro@fb.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Signed-off-by:
-
Shakeel Butt authored
Patch series "Directed kmem charging", v8. The Linux kernel's memory cgroup allows limiting the memory usage of the jobs running on the system to provide isolation between the jobs. All the kernel memory allocated in the context of the job and marked with __GFP_ACCOUNT will also be included in the memory usage and be limited by the job's limit. The kernel memory can only be charged to the memcg of the process in whose context kernel memory was allocated. However there are cases where the allocated kernel memory should be charged to the memcg different from the current processes's memcg. This patch series contains two such concrete use-cases i.e. fsnotify and buffer_head. The fsnotify event objects can consume a lot of system memory for large or unlimited queues if there is either no or slow listener. The events are allocated in the context of the event producer. However they should be charged to the event consumer. Similarly the buffer_head objects can be allocated in a memcg different from the memcg of the page for which buffer_head objects are being allocated. To solve this issue, this patch series introduces mechanism to charge kernel memory to a given memcg. In case of fsnotify events, the memcg of the consumer can be used for charging and for buffer_head, the memcg of the page can be charged. For directed charging, the caller can use the scope API memalloc_[un]use_memcg() to specify the memcg to charge for all the __GFP_ACCOUNT allocations within the scope. This patch (of 2): A lot of memory can be consumed by the events generated for the huge or unlimited queues if there is either no or slow listener. This can cause system level memory pressure or OOMs. So, it's better to account the fsnotify kmem caches to the memcg of the listener. However the listener can be in a different memcg than the memcg of the producer and these allocations happen in the context of the event producer. This patch introduces remote memcg charging API which the producer can use to charge the allocations to the memcg of the listener. There are seven fsnotify kmem caches and among them allocations from dnotify_struct_cache, dnotify_mark_cache, fanotify_mark_cache and inotify_inode_mark_cachep happens in the context of syscall from the listener. So, SLAB_ACCOUNT is enough for these caches. The objects from fsnotify_mark_connector_cachep are not accounted as they are small compared to the notification mark or events and it is unclear whom to account connector to since it is shared by all events attached to the inode. The allocations from the event caches happen in the context of the event producer. For such caches we will need to remote charge the allocations to the listener's memcg. Thus we save the memcg reference in the fsnotify_group structure of the listener. This patch has also moved the members of fsnotify_group to keep the size same, at least for 64 bit build, even with additional member by filling the holes. [shakeelb@google.com: use GFP_KERNEL_ACCOUNT rather than open-coding it] Link: http://lkml.kernel.org/r/20180702215439.211597-1-shakeelb@google.com Link: http://lkml.kernel.org/r/20180627191250.209150-2-shakeelb@google.comSigned-off-by:
-
Roman Gushchin authored
Introduce the mem_cgroup_put() helper, which helps to eliminate guarding memcg css release with "#ifdef CONFIG_MEMCG" in multiple places. Link: http://lkml.kernel.org/r/20180623000600.5818-2-guro@fb.comSigned-off-by:
Roman Gushchin <guro@fb.com> Reviewed-by:
Michal Hocko <mhocko@kernel.org> Acked-by:
David Rientjes <rientjes@google.com> Signed-off-by:
-
Luis R. Rodriguez authored
Some architectures just don't have PAGE_KERNEL_EXEC. The mm/nommu.c and mm/vmalloc.c code have been using PAGE_KERNEL as a fallback for years. Move this fallback to asm-generic. Link: http://lkml.kernel.org/r/20180510185507.2439-3-mcgrof@kernel.orgSigned-off-by:
Luis R. Rodriguez <mcgrof@kernel.org> Suggested-by:
Matthew Wilcox <willy@infradead.org> Reviewed-by:
-
Luis R. Rodriguez authored
Some architectures do not define certain PAGE_KERNEL_* flags, this is either because: a) The way to implement some of these flags is *not yet ported*, or b) The architecture *has no way* to describe them Over time we have accumulated a few PAGE_KERNEL_* fallback workarounds for architectures in the kernel which do not define them using *relatively safe* equivalents. Move these scattered fallback hacks into asm-generic. We start off with PAGE_KERNEL_RO using PAGE_KERNEL as a fallback. This has been in place on the firmware loader for years. Move the fallback into the respective asm-generic header. Link: http://lkml.kernel.org/r/20180510185507.2439-2-mcgrof@kernel.orgSigned-off-by:
-
Oscar Salvador authored
Callers of register_mem_sect_under_node() are always passing a valid memory_block (not NULL), so we can safely drop the check for NULL. In the same way, register_mem_sect_under_node() is only called in case the node is online, so we can safely remove that check as well. Link: http://lkml.kernel.org/r/20180622111839.10071-5-osalvador@techadventures.netSigned-off-by:
Reza Arbab <arbab@linux.vnet.ibm.com> Tested-by:
Jonathan Cameron <Jonathan.Cameron@huawei.com> Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by:
-
Oscar Salvador authored
link_mem_sections() and walk_memory_range() share most of the code, so we can use convert link_mem_sections() into a dummy function that calls walk_memory_range() with a callback to register_mem_sect_under_node(). This patch converts register_mem_sect_under_node() in order to match a walk_memory_range's callback, getting rid of the check_nid argument and checking instead if the system is still boothing, since we only have to check for the nid if the system is in such state. Link: http://lkml.kernel.org/r/20180622111839.10071-4-osalvador@techadventures.netSigned-off-by:
Pavel Tatashin <pavel.tatashin@microsoft.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by:
-
Oscar Salvador authored
When hotplugging memory, it is possible that two calls are being made to register_mem_sect_under_node(). One comes from __add_section()->hotplug_memory_register() and the other from add_memory_resource()->link_mem_sections() if we had to register a new node. In case we had to register a new node, hotplug_memory_register() will only handle/allocate the memory_block's since register_mem_sect_under_node() will return right away because the node it is not online yet. I think it is better if we leave hotplug_memory_register() to handle/allocate only memory_block's and make link_mem_sections() to call register_mem_sect_under_node(). So this patch removes the call to register_mem_sect_under_node() from hotplug_memory_register(), and moves the call to link_mem_sections() out of the condition, so it will always be called. In this way we only have one place where the memory sections are registered. Link: http://lkml.kernel.org/r/20180622111839.10071-3-osalvador@techadventures.netSigned-off-by:
-
Oscar Salvador authored
This is a small cleanup for the memhotplug code. A lot more could be done, but it is better to start somewhere. I tried to unify/remove duplicated code. The following is what this patchset does: 1) add_memory_resource() has code to allocate a node in case it was offline. Since try_online_node has some code for that as well, I just made add_memory_resource() to use that so we can remove duplicated code.. This is better explained in patch 1/4. 2) register_mem_sect_under_node() will be called only from link_mem_sections() 3) Make register_mem_sect_under_node() a callback of walk_memory_range() 4) Drop unnecessary checks from register_mem_sect_under_node() I have done some tests and I could not see anything broken because of this patchset. add_memory_resource() contains code to allocate a new node in case it is necessary. Since try_online_node() also has some code for this purpose, let us make use of that and remove duplicate code. This introduces __try_online_node(), which is called by add_memory_resource() and try_online_node(). __try_online_node() has two new parameters, start_addr of the node, and if the node should be onlined and registered right away. This is always wanted if we are calling from do_cpu_up(), but not when we are calling from memhotplug code. Nothing changes from the point of view of the users of try_online_node(), since try_online_node passes start_addr=0 and online_node=true to __try_online_node(). Link: http://lkml.kernel.org/r/20180622111839.10071-2-osalvador@techadventures.netSigned-off-by:
-
Andrew Morton authored
__list_lru_count_one() has a single callsite. Acked-by:
-
Sebastian Andrzej Siewior authored
shadow_lru_isolate() disables interrupts and acquires a lock. It could use spin_lock_irq() instead. It also uses local_irq_enable() while it could use spin_unlock_irq()/xa_unlock_irq(). Use proper suffix for lock/unlock in order to enable/disable interrupts during release/acquire of a lock. Link: http://lkml.kernel.org/r/20180622151221.28167-3-bigeasy@linutronix.deSigned-off-by:
Sebastian Andrzej Siewior <bigeasy@linutronix.de> Reviewed-by:
-
Sebastian Andrzej Siewior authored
Patch series "mm: use irq locking suffix instead local_irq_disable()". A small series which avoids using local_irq_disable()/local_irq_enable() but instead does spin_lock_irq()/spin_unlock_irq() so it is within the context of the lock which it belongs to. Patch #1 is a cleanup where local_irq_.*() remained after the lock was removed. This patch (of 2): In 0c7c1bed ("mm: make counting of list_lru_one::nr_items lockless") the spin_lock(&nlru->lock); statement was replaced with rcu_read_lock(); in __list_lru_count_one(). The comment in count_shadow_nodes() says that the local_irq_disable() is required because the lock must be acquired with disabled interrupts and (spin_lock()) does not do so. Since the lock is replaced with rcu_read_lock() the local_irq_disable() is no longer needed. The code path is list_lru_shrink_count() -> list_lru_count_one() -> __list_lru_count_one() -> rcu_read_lock() -> list_lru_from_memcg_idx() -> rcu_read_unlock() Remove the local_irq_disable() statement. Link: http://lkml.kernel.org/r/20180622151221.28167-2-bigeasy@linutronix.deSigned-off-by:
-
Michal Hocko authored
There is no real reason to blow up just because the caller doesn't know that __get_free_pages cannot return highmem pages. Simply fix that up silently. Even if we have some confused users such a fixup will not be harmful. [akpm@linux-foundation.org: mask off __GFP_HIGHMEM] Link: http://lkml.kernel.org/r/20180622162841.25114-1-mhocko@kernel.orgSigned-off-by:
-
Huang Ying authored
This is to take better advantage of the general huge page copying optimization. Where, the target subpage will be copied last to avoid the cache lines of target subpage to be evicted when copying other subpages. This works better if the address of the target subpage is available when copying huge page. So hugetlbfs page fault handlers are changed to pass that information to hugetlb_cow(). This will benefit workloads which don't access the begin of the hugetlbfs huge page after the page fault under heavy cache contention. Link: http://lkml.kernel.org/r/20180524005851.4079-5-ying.huang@intel.comSigned-off-by:
"Huang, Ying" <ying.huang@intel.com> Reviewed-by:
Mike Kravetz <mike.kravetz@oracle.com> Cc: Michal Hocko <mhocko@suse.com> Cc: David Rientjes <rientjes@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Andi Kleen <andi.kleen@intel.com> Cc: Jan Kara <jack@suse.cz> Cc: Matthew Wilcox <willy@infradead.org> Cc: Hugh Dickins <hughd@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Shaohua Li <shli@fb.com> Cc: Christopher Lameter <cl@linux.com> Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> Cc: Punit Agrawal <punit.agrawal@arm.com> Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com> Signed-off-by:
-
Huang Ying authored
To take better advantage of general huge page copying optimization, the target subpage address will be passed to hugetlb_cow(), then copy_user_huge_page(). So we will use both target subpage address and huge page size aligned address in hugetlb_cow(). To distinguish between them, "haddr" is used for huge page size aligned address to be consistent with Transparent Huge Page naming convention. Now, only huge page size aligned address is used in hugetlb_cow(), so the "address" is renamed to "haddr" in hugetlb_cow() in this patch. Next patch will use target subpage address in hugetlb_cow() too. The patch is just code cleanup without any functionality changes. Link: http://lkml.kernel.org/r/20180524005851.4079-4-ying.huang@intel.comSigned-off-by:
-
Huang Ying authored
Huge page helps to reduce TLB miss rate, but it has higher cache footprint, sometimes this may cause some issue. For example, when copying huge page on x86_64 platform, the cache footprint is 4M. But on a Xeon E5 v3 2699 CPU, there are 18 cores, 36 threads, and only 45M LLC (last level cache). That is, in average, there are 2.5M LLC for each core and 1.25M LLC for each thread. If the cache contention is heavy when copying the huge page, and we copy the huge page from the begin to the end, it is possible that the begin of huge page is evicted from the cache after we finishing copying the end of the huge page. And it is possible for the application to access the begin of the huge page after copying the huge page. In c79b57e4 ("mm: hugetlb: clear target sub-page last when clearing huge page"), to keep the cache lines of the target subpage hot, the order to clear the subpages in the huge page in clear_huge_page() is changed to clearing the subpage which is furthest from the target subpage firstly, and the target subpage last. The similar order changing helps huge page copying too. That is implemented in this patch. Because we have put the order algorithm into a separate function, the implementation is quite simple. The patch is a generic optimization which should benefit quite some workloads, not for a specific use case. To demonstrate the performance benefit of the patch, we tested it with vm-scalability run on transparent huge page. With this patch, the throughput increases ~16.6% in vm-scalability anon-cow-seq test case with 36 processes on a 2 socket Xeon E5 v3 2699 system (36 cores, 72 threads). The test case set /sys/kernel/mm/transparent_hugepage/enabled to be always, mmap() a big anonymous memory area and populate it, then forked 36 child processes, each writes to the anonymous memory area from the begin to the end, so cause copy on write. For each child process, other child processes could be seen as other workloads which generate heavy cache pressure. At the same time, the IPC (instruction per cycle) increased from 0.63 to 0.78, and the time spent in user space is reduced ~7.2%. Link: http://lkml.kernel.org/r/20180524005851.4079-3-ying.huang@intel.comSigned-off-by:
-
Huang Ying authored
Patch series "mm, huge page: Copy target sub-page last when copy huge page", v2. Huge page helps to reduce TLB miss rate, but it has higher cache footprint, sometimes this may cause some issue. For example, when copying huge page on x86_64 platform, the cache footprint is 4M. But on a Xeon E5 v3 2699 CPU, there are 18 cores, 36 threads, and only 45M LLC (last level cache). That is, in average, there are 2.5M LLC for each core and 1.25M LLC for each thread. If the cache contention is heavy when copying the huge page, and we copy the huge page from the begin to the end, it is possible that the begin of huge page is evicted from the cache after we finishing copying the end of the huge page. And it is possible for the application to access the begin of the huge page after copying the huge page. In c79b57e4 ("mm: hugetlb: clear target sub-page last when clearing huge page"), to keep the cache lines of the target subpage hot, the order to clear the subpages in the huge page in clear_huge_page() is changed to clearing the subpage which is furthest from the target subpage firstly, and the target subpage last. The similar order changing helps huge page copying too. That is implemented in this patchset. The patchset is a generic optimization which should benefit quite some workloads, not for a specific use case. To demonstrate the performance benefit of the patchset, we have tested it with vm-scalability run on transparent huge page. With this patchset, the throughput increases ~16.6% in vm-scalability anon-cow-seq test case with 36 processes on a 2 socket Xeon E5 v3 2699 system (36 cores, 72 threads). The test case set /sys/kernel/mm/transparent_hugepage/enabled to be always, mmap() a big anonymous memory area and populate it, then forked 36 child processes, each writes to the anonymous memory area from the begin to the end, so cause copy on write. For each child process, other child processes could be seen as other workloads which generate heavy cache pressure. At the same time, the IPC (instruction per cycle) increased from 0.63 to 0.78, and the time spent in user space is reduced ~7.2%. This patch (of 4): In c79b57e4 ("mm: hugetlb: clear target sub-page last when clearing huge page"), to keep the cache lines of the target subpage hot, the order to clear the subpages in the huge page in clear_huge_page() is changed to clearing the subpage which is furthest from the target subpage firstly, and the target subpage last. This optimization could be applied to copying huge page too with the same order algorithm. To avoid code duplication and reduce maintenance overhead, in this patch, the order algorithm is moved out of clear_huge_page() into a separate function: process_huge_page(). So that we can use it for copying huge page too. This will change the direct calls to clear_user_highpage() into the indirect calls. But with the proper inline support of the compilers, the indirect call will be optimized to be the direct call. Our tests show no performance change with the patch. This patch is a code cleanup without functionality change. Link: http://lkml.kernel.org/r/20180524005851.4079-2-ying.huang@intel.comSigned-off-by:
-
Jens Axboe authored
a_ops->readpages() is only ever used for read-ahead. Ensure that we pass this information down to the block layer. Link: http://lkml.kernel.org/r/20180621010725.17813-5-axboe@kernel.dkSigned-off-by:
Jens Axboe <axboe@kernel.dk> Reviewed-by:
-
Jens Axboe authored
a_ops->readpages() is only ever used for read-ahead. Ensure that we pass this information down to the block layer. Link: http://lkml.kernel.org/r/20180621010725.17813-4-axboe@kernel.dkSigned-off-by:
-
Jens Axboe authored
a_ops->readpages() is only ever used for read-ahead, yet we don't flag the IO being submitted as such. Fix that up. Any file system that uses mpage_readpages() as its ->readpages() implementation will now get this right. Since we're passing in whether the IO is read-ahead or not, we don't need to pass in the 'gfp' separately, as it is dependent on the IO being read-ahead. Kill off that member. Add some documentation notes on ->readpages() being purely for read-ahead. Link: http://lkml.kernel.org/r/20180621010725.17813-3-axboe@kernel.dkSigned-off-by:
-
Jens Axboe authored
Patch series "Submit ->readpages() IO as read-ahead", v4. The only caller of ->readpages() is from read-ahead, yet we don't submit IO flagged with REQ_RAHEAD. This means we don't see it in blktrace, for instance, which is a shame. Additionally, it's preventing further functional changes in the block layer for deadling with read-ahead more intelligently. We already make assumptions about ->readpages() just being for read-ahead in the mpage implementation, using readahead_gfp_mask(mapping) as out GFP mask of choice. This small series fixes up mpage_readpages() to submit with REQ_RAHEAD, which takes care of file systems using mpage_readpages(). The first patch is a prep patch, that makes do_mpage_readpage() take an argument structure. This patch (of 4): We're currently passing 8 arguments to this function, clean it up a bit by packing the arguments in an args structure we pass to it. No intentional functional changes in this patch. [akpm@linux-foundation.org: coding-style fixes] Link: http://lkml.kernel.org/r/20180621010725.17813-2-axboe@kernel.dkSigned-off-by:
-
Yang Shi authored
/sys/kernel/mm/transparent_hugepage/khugepaged/pages_collapsed is used to record the counter of collapsed THP, but it just gets inc'ed in anonymous THP collapse path, do this for shmem THP collapse too. Link: http://lkml.kernel.org/r/1529622949-75504-2-git-send-email-yang.shi@linux.alibaba.comSigned-off-by:
Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by:
-
Yang Shi authored
When merging anonymous page vma, if the size of the vma can fit in at least one hugepage, the mm will be registered for khugepaged for collapsing THP in the future. But it skips shmem vmas. Do so for shmem also, but not for file-private mappings when merging a vma in order to increase the odds of collapsing a hugepage via khugepaged. hugepage_vma_check() sounds like a good fit to do the check. And move the definition of it before khugepaged_enter_vma_merge() to avoid a build error. Link: http://lkml.kernel.org/r/1529697791-6950-1-git-send-email-yang.shi@linux.alibaba.comSigned-off-by:
-
Jia-Ju Bai authored
The argument "gfp_t flags" is not used in kasan_unpoison_element() and remove_element(), so remove it. Link: http://lkml.kernel.org/r/20180621070332.16633-1-baijiaju1990@gmail.comSigned-off-by:
Jia-Ju Bai <baijiaju1990@gmail.com> Reviewed-by:
-
Greg Thelen authored
Use smaller scan_control fields for order, priority, and reclaim_idx. Convert fields from int => s8. All easily fit within a byte: - allocation order range: 0..MAX_ORDER(64?) - priority range: 0..12(DEF_PRIORITY) - reclaim_idx range: 0..6(__MAX_NR_ZONES) Since 6538b8ea ("x86_64: expand kernel stack to 16K") x86_64 stack overflows are not an issue. But it's inefficient to use ints. Use s8 (signed byte) rather than u8 to allow for loops like: do { ... } while (--sc.priority >= 0); Add BUILD_BUG_ON to verify that s8 is capable of storing max values. This reduces sizeof(struct scan_control): - 96 => 80 bytes (x86_64) - 68 => 56 bytes (i386) scan_control structure field order is changed to utilize padding. After this patch there is 1 bit of scan_control padding. akpm: makes my vmscan.o's .text 572 bytes smaller as well. Link: http://lkml.kernel.org/r/20180530061212.84915-1-gthelen@google.comSigned-off-by:
-
