OpenE2K/linux
6
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
36e5b0e391
linux/mm
History
Johannes Weiner 4dd72b4a47 mm: fadvise: avoid expensive remote LRU cache draining after FADV_DONTNEED 
When FADV_DONTNEED cannot drop all pages in the range, it observes that
some pages might still be on per-cpu LRU caches after recent
instantiation and so initiates remote calls to all CPUs to flush their
local caches.  However, in most cases, the fadvise happens from the same
context that instantiated the pages, and any pre-LRU pages in the
specified range are most likely sitting on the local CPU's LRU cache,
and so in many cases this results in unnecessary remote calls, which, in
a loaded system, can hold up the fadvise() call significantly.

[ I didn't record it in the extreme case we observed at Facebook,
  unfortunately. We had a slow-to-respond system and noticed it
  lru_add_drain_all() leading the profile during fadvise calls. This
  patch came out of thinking about the code and how we commonly call
  FADV_DONTNEED.

  FWIW, I wrote a silly directory tree walker/searcher that recurses
  through /usr to read and FADV_DONTNEED each file it finds. On a 2
  socket 40 ht machine, over 1% is spent in lru_add_drain_all(). With
  the patch, that cost is gone; the local drain cost shows at 0.09%. ]

Try to avoid the remote call by flushing the local LRU cache before even
attempting to invalidate anything.  It's a cheap operation, and the
local LRU cache is the most likely to hold any pre-LRU pages in the
specified fadvise range.

Link: http://lkml.kernel.org/r/20161214210017.GA1465@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-20 09:48:46 -08:00
..
kasan Power management material for v4.10-rc1 2016-12-13 10:41:53 -08:00
backing-dev.c writeback: track if we're sleeping on progress in balance_dirty_pages() 2016-11-08 08:28:55 -07:00
balloon_compaction.c
bootmem.c mm: kmemleak: avoid using __va() on addresses that don't have a lowmem mapping 2016-10-11 15:06:33 -07:00
cleancache.c
cma_debug.c
cma.c mm/cma.c: check the max limit for cma allocation 2016-11-11 08:12:37 -08:00
cma.h
compaction.c mm, compaction: allow compaction for GFP_NOFS requests 2016-12-14 16:04:07 -08:00
debug_page_ref.c
debug.c mm, debug: print raw struct page data in __dump_page() 2016-12-12 18:55:08 -08:00
dmapool.c
early_ioremap.c
fadvise.c mm: fadvise: avoid expensive remote LRU cache draining after FADV_DONTNEED 2016-12-20 09:48:46 -08:00
failslab.c
filemap.c Merge branch 'akpm' (patches from Andrew) 2016-12-14 17:25:18 -08:00
frame_vector.c mm: replace get_vaddr_frames() write/force parameters with gup_flags 2016-10-19 08:11:24 -07:00
frontswap.c
gup.c mm: unexport __get_user_pages_unlocked() 2016-12-14 16:04:09 -08:00
highmem.c
huge_memory.c mm: join struct fault_env and vm_fault 2016-12-14 16:04:09 -08:00
hugetlb_cgroup.c
hugetlb.c mm: add tlb_remove_check_page_size_change to track page size change 2016-12-12 18:55:07 -08:00
hwpoison-inject.c
init-mm.c mm: Add a user_ns owner to mm_struct and fix ptrace permission checks 2016-11-22 11:49:48 -06:00
internal.h mm: add orig_pte field into vm_fault 2016-12-14 16:04:09 -08:00
interval_tree.c
Kconfig mm: THP page cache support for ppc64 2016-12-12 18:55:08 -08:00
Kconfig.debug PM / Hibernate: allow hibernation with PAGE_POISONING_ZERO 2016-09-13 02:35:27 +02:00
khugepaged.c radix-tree: improve multiorder iterators 2016-12-14 16:04:10 -08:00
kmemcheck.c
kmemleak-test.c
kmemleak.c kmemleak: fix reference to Documentation 2016-12-12 18:55:07 -08:00
ksm.c mm,ksm: add __GFP_HIGH to the allocation in alloc_stable_node() 2016-10-07 18:46:29 -07:00
list_lru.c mm/list_lru.c: avoid error-path NULL pointer deref 2016-10-27 18:43:42 -07:00
maccess.c
madvise.c mm: add tlb_remove_check_page_size_change to track page size change 2016-12-12 18:55:07 -08:00
Makefile Disable the __builtin_return_address() warning globally after all 2016-10-12 10:23:41 -07:00
memblock.c mm: kmemleak: avoid using __va() on addresses that don't have a lowmem mapping 2016-10-11 15:06:33 -07:00
memcontrol.c Merge branch 'akpm' (patches from Andrew) 2016-12-12 20:50:02 -08:00
memory_hotplug.c mm: remove x86-only restriction of movable_node 2016-12-12 18:55:07 -08:00
memory-failure.c mm: hwpoison: fix thp split handling in memory_failure() 2016-11-11 08:12:37 -08:00
memory.c Merge branch 'akpm' (patches from Andrew) 2016-12-14 17:25:18 -08:00
mempolicy.c mm/mempolicy.c: forbid static or relative flags for local NUMA mode 2016-12-12 18:55:07 -08:00
mempool.c
memtest.c
migrate.c lib: radix-tree: check accounting of existing slot replacement users 2016-12-12 18:55:08 -08:00
mincore.c mm, swap: use offset of swap entry as key of swap cache 2016-10-07 18:46:28 -07:00
mlock.c thp: fix corner case of munlock() of PTE-mapped THPs 2016-11-30 16:32:52 -08:00
mm_init.c
mmap.c mm: vma_merge: correct false positive from __vma_unlink->validate_mm_rb 2016-10-07 18:46:29 -07:00
mmu_context.c
mmu_notifier.c
mmzone.c
mprotect.c mm/pkeys: generate pkey system call code only if ARCH_HAS_PKEYS is selected 2016-12-12 18:55:07 -08:00
mremap.c mremap: move_ptes: check pte dirty after its removal 2016-11-29 08:20:24 -08:00
msync.c
nobootmem.c mm: kmemleak: avoid using __va() on addresses that don't have a lowmem mapping 2016-10-11 15:06:33 -07:00
nommu.c Merge branch 'akpm' (patches from Andrew) 2016-12-14 17:25:18 -08:00
oom_kill.c oom: print nodemask in the oom report 2016-10-07 18:46:29 -07:00
page_alloc.c mm: add support for releasing multiple instances of a page 2016-12-14 16:04:08 -08:00
page_counter.c
page_ext.c mm/page_ext: support extra space allocation by page_ext user 2016-10-07 18:46:27 -07:00
page_idle.c
page_io.c writeback: add wbc_to_write_flags() 2016-11-02 10:24:03 -06:00
page_isolation.c mm/page_isolation: fix typo: "paes" -> "pages" 2016-10-07 18:46:29 -07:00
page_owner.c mm/page_owner: don't define fields on struct page_ext by hard-coding 2016-10-07 18:46:27 -07:00
page_poison.c
page-writeback.c radix-tree: delete radix_tree_range_tag_if_tagged() 2016-12-14 16:04:10 -08:00
pagewalk.c
percpu-km.c
percpu-vm.c
percpu.c Merge branch 'for-4.10' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu 2016-12-13 12:34:47 -08:00
pgtable-generic.c
process_vm_access.c mm: unexport __get_user_pages_unlocked() 2016-12-14 16:04:09 -08:00
quicklist.c
readahead.c mm: don't cap request size based on read-ahead setting 2016-12-12 18:55:08 -08:00
rmap.c mm, rmap: handle anon_vma_prepare() common case inline 2016-12-12 18:55:08 -08:00
shmem.c Merge uncontroversial parts of branch 'readlink' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs 2016-12-17 19:16:12 -08:00
slab_common.c mm/slab_common.c: check kmem_create_cache flags are common 2016-12-12 18:55:06 -08:00
slab.c Merge branch 'for-4.10' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq 2016-12-13 12:59:57 -08:00
slab.h mm, slab: maintain total slab count instead of active count 2016-12-12 18:55:07 -08:00
slob.c slub: move synchronize_sched out of slab_mutex on shrink 2016-12-12 18:55:06 -08:00
slub.c slub: avoid false-postive warning 2016-12-12 18:55:06 -08:00
sparse-vmemmap.c treewide: replace obsolete _refok by __ref 2016-08-02 17:31:41 -04:00
sparse.c treewide: replace obsolete _refok by __ref 2016-08-02 17:31:41 -04:00
swap_cgroup.c
swap_state.c mm, swap: use offset of swap entry as key of swap cache 2016-10-07 18:46:28 -07:00
swap.c thp: reduce usage of huge zero page's atomic counter 2016-10-07 18:46:28 -07:00
swapfile.c mm: add three more cond_resched() in swapoff 2016-12-12 18:55:08 -08:00
truncate.c mm: workingset: move shadow entry tracking to radix tree exceptional tracking 2016-12-12 18:55:08 -08:00
usercopy.c mm: usercopy: Check for module addresses 2016-09-20 16:07:39 -07:00
userfaultfd.c
util.c Merge branch 'mm-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2016-10-22 09:39:10 -07:00
vmacache.c mm: unrig VMA cache hit ratio 2016-10-07 18:46:27 -07:00
vmalloc.c mm: add preempt points into __purge_vmap_area_lazy() 2016-12-12 18:55:08 -08:00
vmpressure.c
vmscan.c Merge branch 'akpm' (patches from Andrew) 2016-12-12 20:50:02 -08:00
vmstat.c mm/vmstat: Convert to hotplug state machine 2016-12-02 00:52:35 +01:00
workingset.c mm: workingset: update shadow limit to reflect bigger active list 2016-12-12 18:55:08 -08:00
z3fold.c
zbud.c
zpool.c
zsmalloc.c mm/zsmalloc: Convert to hotplug state machine 2016-12-02 00:52:36 +01:00
zswap.c mm/zswap: Convert pool to hotplug state machine 2016-12-02 00:52:36 +01:00