linux/mm
Mel Gorman c67fe3752a mm: compaction: Abort async compaction if locks are contended or taking too long
Jim Schutt reported a problem that pointed at compaction contending
heavily on locks.  The workload is straight-forward and in his own words;

	The systems in question have 24 SAS drives spread across 3 HBAs,
	running 24 Ceph OSD instances, one per drive.  FWIW these servers
	are dual-socket Intel 5675 Xeons w/48 GB memory.  I've got ~160
	Ceph Linux clients doing dd simultaneously to a Ceph file system
	backed by 12 of these servers.

Early in the test everything looks fine

  procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu-------
   r  b       swpd       free       buff      cache   si   so    bi    bo   in   cs  us sy  id wa st
  31 15          0     287216        576   38606628    0    0     2  1158    2   14   1  3  95  0  0
  27 15          0     225288        576   38583384    0    0    18 2222016 203357 134876  11 56  17 15  0
  28 17          0     219256        576   38544736    0    0    11 2305932 203141 146296  11 49  23 17  0
   6 18          0     215596        576   38552872    0    0     7 2363207 215264 166502  12 45  22 20  0
  22 18          0     226984        576   38596404    0    0     3 2445741 223114 179527  12 43  23 22  0

and then it goes to pot

  procs -------------------memory------------------ ---swap-- -----io---- --system-- -----cpu-------
   r  b       swpd       free       buff      cache   si   so    bi    bo   in   cs  us sy  id wa st
  163  8          0     464308        576   36791368    0    0    11 22210  866  536   3 13  79  4  0
  207 14          0     917752        576   36181928    0    0   712 1345376 134598 47367   7 90   1  2  0
  123 12          0     685516        576   36296148    0    0   429 1386615 158494 60077   8 84   5  3  0
  123 12          0     598572        576   36333728    0    0  1107 1233281 147542 62351   7 84   5  4  0
  622  7          0     660768        576   36118264    0    0   557 1345548 151394 59353   7 85   4  3  0
  223 11          0     283960        576   36463868    0    0    46 1107160 121846 33006   6 93   1  1  0

Note that system CPU usage is very high blocks being written out has
dropped by 42%. He analysed this with perf and found

  perf record -g -a sleep 10
  perf report --sort symbol --call-graph fractal,5
    34.63%  [k] _raw_spin_lock_irqsave
            |
            |--97.30%-- isolate_freepages
            |          compaction_alloc
            |          unmap_and_move
            |          migrate_pages
            |          compact_zone
            |          compact_zone_order
            |          try_to_compact_pages
            |          __alloc_pages_direct_compact
            |          __alloc_pages_slowpath
            |          __alloc_pages_nodemask
            |          alloc_pages_vma
            |          do_huge_pmd_anonymous_page
            |          handle_mm_fault
            |          do_page_fault
            |          page_fault
            |          |
            |          |--87.39%-- skb_copy_datagram_iovec
            |          |          tcp_recvmsg
            |          |          inet_recvmsg
            |          |          sock_recvmsg
            |          |          sys_recvfrom
            |          |          system_call
            |          |          __recv
            |          |          |
            |          |           --100.00%-- (nil)
            |          |
            |           --12.61%-- memcpy
             --2.70%-- [...]

There was other data but primarily it is all showing that compaction is
contended heavily on the zone->lock and zone->lru_lock.

commit [b2eef8c0: mm: compaction: minimise the time IRQs are disabled
while isolating pages for migration] noted that it was possible for
migration to hold the lru_lock for an excessive amount of time. Very
broadly speaking this patch expands the concept.

This patch introduces compact_checklock_irqsave() to check if a lock
is contended or the process needs to be scheduled. If either condition
is true then async compaction is aborted and the caller is informed.
The page allocator will fail a THP allocation if compaction failed due
to contention. This patch also introduces compact_trylock_irqsave()
which will acquire the lock only if it is not contended and the process
does not need to schedule.

Reported-by: Jim Schutt <jaschut@sandia.gov>
Tested-by: Jim Schutt <jaschut@sandia.gov>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-08-21 16:45:03 -07:00
..
backing-dev.c vfs: kill write_super and sync_supers 2012-08-04 01:24:44 +04:00
bootmem.c bootmem: make ___alloc_bootmem_node_nopanic() really nopanic 2012-07-17 16:21:29 -07:00
bounce.c bounce: allow use of bounce pool via config option 2012-07-18 16:40:35 -04:00
cleancache.c ->encode_fh() API change 2012-05-29 23:28:33 -04:00
compaction.c mm: compaction: Abort async compaction if locks are contended or taking too long 2012-08-21 16:45:03 -07:00
debug-pagealloc.c
dmapool.c
fadvise.c mm, fadvise: don't return -EINVAL when filesystem cannot implement fadvise() 2012-07-31 18:42:42 -07:00
failslab.c
filemap_xip.c fs: Protect write paths by sb_start_write - sb_end_write 2012-07-31 09:45:47 +04:00
filemap.c fs: Protect write paths by sb_start_write - sb_end_write 2012-07-31 09:45:47 +04:00
fremap.c
frontswap.c mm/frontswap: cleanup doc and comment error 2012-07-23 11:16:20 -04:00
highmem.c mm: add support for direct_IO to highmem pages 2012-07-31 18:42:47 -07:00
huge_memory.c mm/memcg: apply add/del_page to lruvec 2012-05-29 16:22:28 -07:00
hugetlb_cgroup.c hugetlb/cgroup: remove exclude and wakeup rmdir calls from migrate 2012-07-31 18:42:41 -07:00
hugetlb.c mm: hugetlbfs: close race during teardown of hugetlbfs shared page tables 2012-07-31 18:42:50 -07:00
hwpoison-inject.c memcg: rename config variables 2012-07-31 18:42:43 -07:00
init-mm.c
internal.h mm: compaction: Abort async compaction if locks are contended or taking too long 2012-08-21 16:45:03 -07:00
Kconfig mm: factor out memory isolate functions 2012-07-31 18:42:45 -07:00
Kconfig.debug
kmemcheck.c
kmemleak-test.c
kmemleak.c
ksm.c
maccess.c
madvise.c mm: Hold a file reference in madvise_remove 2012-07-06 10:34:38 -07:00
Makefile mm: factor out memory isolate functions 2012-07-31 18:42:45 -07:00
memblock.c mm/memblock.c:memblock_double_array(): cosmetic cleanups 2012-07-31 18:42:41 -07:00
memcontrol.c memcg: add mem_cgroup_from_css() helper 2012-07-31 18:42:49 -07:00
memory_hotplug.c mm/hotplug: free zone->pageset when a zone becomes empty 2012-07-31 18:42:44 -07:00
memory-failure.c memcg: rename config variables 2012-07-31 18:42:43 -07:00
memory.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2012-08-01 10:26:23 -07:00
mempolicy.c Merge branch 'slab/next' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/linux 2012-07-30 11:32:24 -07:00
mempool.c mempool: add @gfp_mask to mempool_create_node() 2012-06-25 11:53:47 +02:00
migrate.c mm: memcg: fix compaction/migration failing due to memcg limits 2012-07-31 18:42:48 -07:00
mincore.c
mlock.c
mm_init.c
mmap.c mm: change nr_ptes BUG_ON to WARN_ON 2012-08-21 16:45:02 -07:00
mmu_context.c
mmu_notifier.c mm: mmu_notifier: fix freed page still mapped in secondary MMU 2012-07-31 18:42:49 -07:00
mmzone.c memcg: rename config variables 2012-07-31 18:42:43 -07:00
mprotect.c Merge branch 'akpm' (Andrew's patch-bomb) 2012-03-22 09:04:48 -07:00
mremap.c mm: account the total_vm in the vm_stat_account() 2012-07-31 18:42:39 -07:00
msync.c
nobootmem.c memblock: free allocated memblock_reserved_regions later 2012-07-11 16:04:50 -07:00
nommu.c nommu: fix compilation of nommu.c 2012-06-04 17:17:31 -04:00
oom_kill.c mm, memcg: move all oom handling to memcontrol.c 2012-07-31 18:42:45 -07:00
page_alloc.c mm: compaction: Abort async compaction if locks are contended or taking too long 2012-08-21 16:45:03 -07:00
page_cgroup.c memcg: rename config variables 2012-07-31 18:42:43 -07:00
page_io.c mm: add support for direct_IO to highmem pages 2012-07-31 18:42:47 -07:00
page_isolation.c memory-hotplug: fix kswapd looping forever problem 2012-07-31 18:42:45 -07:00
page-writeback.c vfs: kill write_super and sync_supers 2012-08-04 01:24:44 +04:00
pagewalk.c mm: fix kernel-doc warnings 2012-06-20 14:39:36 -07:00
percpu-km.c
percpu-vm.c mm: fix kernel-doc warnings 2012-06-20 14:39:36 -07:00
percpu.c kmemleak: Fix the kmemleak tracking of the percpu areas with !SMP 2012-05-09 10:13:29 -07:00
pgtable-generic.c arch/tile: allow building Linux with transparent huge pages enabled 2012-05-25 12:48:21 -04:00
prio_tree.c
process_vm_access.c aio/vfs: cleanup of rw_copy_check_uvector() and compat_rw_copy_check_uvector() 2012-05-31 17:49:32 -07:00
quicklist.c
readahead.c mm: move readahead syscall to mm/readahead.c 2012-05-29 16:22:23 -07:00
rmap.c mm: remove swap token code 2012-05-29 16:22:19 -07:00
shmem.c tmpfs: distribute interleave better across nodes 2012-07-31 18:42:50 -07:00
slab_common.c mm: Fix build warning in kmem_cache_create() 2012-07-30 13:15:40 +03:00
slab.c mm: micro-optimise slab to avoid a function call 2012-07-31 18:42:46 -07:00
slab.h mm, sl[aou]b: Use a common mutex definition 2012-07-09 12:13:41 +03:00
slob.c slob: Fix early boot kernel crash 2012-07-12 10:13:22 +03:00
slub.c mm: slub: optimise the SLUB fast path to avoid pfmemalloc checks 2012-07-31 18:42:45 -07:00
sparse-vmemmap.c
sparse.c mm/sparse: remove index_init_lock 2012-07-31 18:42:49 -07:00
swap_state.c mm: add support for a filesystem to activate swap files and use direct_IO for writing swap pages 2012-07-31 18:42:47 -07:00
swap.c mm: add support for direct_IO to highmem pages 2012-07-31 18:42:47 -07:00
swapfile.c mm: swapfile: clean up unuse_pte race handling 2012-07-31 18:42:48 -07:00
truncate.c mm/fs: remove truncate_range 2012-05-29 16:22:23 -07:00
util.c new helper: vm_mmap_pgoff() 2012-06-01 10:37:18 -04:00
vmalloc.c mm: make vb_alloc() more foolproof 2012-07-31 18:42:39 -07:00
vmscan.c memcg: gix memory accounting scalability in shrink_page_list 2012-07-31 18:42:49 -07:00
vmstat.c mm: account for the number of times direct reclaimers get throttled 2012-07-31 18:42:46 -07:00