mm: correct page->pfmemalloc to fix deactivate_slab regression
authorAlex Shi <alex.shi@intel.com>
Tue, 21 Aug 2012 23:16:08 +0000 (16:16 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 21 Aug 2012 23:45:03 +0000 (16:45 -0700)
commitb121186ab1b12e2a96a945d88eae0735b4542158
tree9d142b77b090af2c508302457435f33143630a0b
parent5ed12f12825c6c0451d703bfe918a7fc190e2738
mm: correct page->pfmemalloc to fix deactivate_slab regression

Commit cfd19c5a9ecf ("mm: only set page->pfmemalloc when
ALLOC_NO_WATERMARKS was used") tried to narrow down page->pfmemalloc
setting, but it missed some places the pfmemalloc should be set.

So, in __slab_alloc, the unalignment pfmemalloc and ALLOC_NO_WATERMARKS
cause incorrect deactivate_slab() on our core2 server:

    64.73%           fio  [kernel.kallsyms]     [k] _raw_spin_lock
                     |
                     --- _raw_spin_lock
                        |
                        |---0.34%-- deactivate_slab
                        |          __slab_alloc
                        |          kmem_cache_alloc
                        |          |

That causes our fio sync write performance to have a 40% regression.

Move the checking in get_page_from_freelist() which resolves this issue.

Signed-off-by: Alex Shi <alex.shi@intel.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: David Miller <davem@davemloft.net
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Tested-by: Eric Dumazet <eric.dumazet@gmail.com>
Tested-by: Sage Weil <sage@inktank.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/page_alloc.c