Revision 72853e2991a2702ae93aaf889ac7db743a415dd3 authored by Mel Gorman on 09 September 2010, 23:38:16 UTC, committed by Linus Torvalds on 10 September 2010, 01:57:25 UTC
When allocating a page, the system uses NR_FREE_PAGES counters to
determine if watermarks would remain intact after the allocation was made.
This check is made without interrupts disabled or the zone lock held and
so is race-prone by nature.  Unfortunately, when pages are being freed in
batch, the counters are updated before the pages are added on the list.
During this window, the counters are misleading as the pages do not exist
yet.  When under significant pressure on systems with large numbers of
CPUs, it's possible for processes to make progress even though they should
have been stalled.  This is particularly problematic if a number of the
processes are using GFP_ATOMIC as the min watermark can be accidentally
breached and in extreme cases, the system can livelock.

This patch updates the counters after the pages have been added to the
list.  This makes the allocator more cautious with respect to preserving
the watermarks and mitigates livelock possibilities.

[akpm@linux-foundation.org: avoid modifying incoming args]
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1 parent 5ee28a4
History
File Mode Size
lzo
raid6
reed_solomon
zlib_deflate
zlib_inflate
.gitignore -rw-r--r-- 51 bytes
Kconfig -rw-r--r-- 4.1 KB
Kconfig.debug -rw-r--r-- 39.7 KB
Kconfig.kgdb -rw-r--r-- 2.3 KB
Kconfig.kmemcheck -rw-r--r-- 2.9 KB
Makefile -rw-r--r-- 3.6 KB
argv_split.c -rw-r--r-- 1.8 KB
atomic64.c -rw-r--r-- 4.1 KB
atomic64_test.c -rw-r--r-- 3.5 KB
audit.c -rw-r--r-- 1.2 KB
bcd.c -rw-r--r-- 257 bytes
bitmap.c -rw-r--r-- 32.6 KB
bitrev.c -rw-r--r-- 2.1 KB
btree.c -rw-r--r-- 19.2 KB
bug.c -rw-r--r-- 4.7 KB
bust_spinlocks.c -rw-r--r-- 636 bytes
check_signature.c -rw-r--r-- 599 bytes
checksum.c -rw-r--r-- 4.9 KB
cmdline.c -rw-r--r-- 3.6 KB
cpu-notifier-error-inject.c -rw-r--r-- 1.5 KB
cpumask.c -rw-r--r-- 4.6 KB
crc-ccitt.c -rw-r--r-- 3.0 KB
crc-itu-t.c -rw-r--r-- 2.8 KB
crc-t10dif.c -rw-r--r-- 2.9 KB
crc16.c -rw-r--r-- 2.8 KB
crc32.c -rw-r--r-- 14.5 KB
crc32defs.h -rw-r--r-- 1.0 KB
crc7.c -rw-r--r-- 2.3 KB
ctype.c -rw-r--r-- 1.3 KB
debug_locks.c -rw-r--r-- 1.1 KB
debugobjects.c -rw-r--r-- 25.2 KB
dec_and_lock.c -rw-r--r-- 782 bytes
decompress.c -rw-r--r-- 1.2 KB
decompress_bunzip2.c -rw-r--r-- 23.5 KB
decompress_inflate.c -rw-r--r-- 3.6 KB
decompress_unlzma.c -rw-r--r-- 15.5 KB
decompress_unlzo.c -rw-r--r-- 4.9 KB
devres.c -rw-r--r-- 7.7 KB
div64.c -rw-r--r-- 2.3 KB
dma-debug.c -rw-r--r-- 31.5 KB
dump_stack.c -rw-r--r-- 290 bytes
dynamic_debug.c -rw-r--r-- 18.0 KB
extable.c -rw-r--r-- 2.4 KB
fault-inject.c -rw-r--r-- 7.9 KB
find_last_bit.c -rw-r--r-- 1.1 KB
find_next_bit.c -rw-r--r-- 6.4 KB
flex_array.c -rw-r--r-- 9.8 KB
gcd.c -rw-r--r-- 291 bytes
gen_crc32table.c -rw-r--r-- 2.2 KB
genalloc.c -rw-r--r-- 5.1 KB
halfmd4.c -rw-r--r-- 2.0 KB
hexdump.c -rw-r--r-- 6.5 KB
hweight.c -rw-r--r-- 1.9 KB
idr.c -rw-r--r-- 21.5 KB
inflate.c -rw-r--r-- 38.6 KB
int_sqrt.c -rw-r--r-- 533 bytes
iomap.c -rw-r--r-- 7.4 KB
iomap_copy.c -rw-r--r-- 2.1 KB
iommu-helper.c -rw-r--r-- 1006 bytes
ioremap.c -rw-r--r-- 2.1 KB
irq_regs.c -rw-r--r-- 578 bytes
is_single_threaded.c -rw-r--r-- 1.3 KB
kasprintf.c -rw-r--r-- 704 bytes
kernel_lock.c -rw-r--r-- 3.2 KB
klist.c -rw-r--r-- 9.3 KB
kobject.c -rw-r--r-- 23.3 KB
kobject_uevent.c -rw-r--r-- 10.3 KB
kref.c -rw-r--r-- 1.6 KB
lcm.c -rw-r--r-- 265 bytes
libcrc32c.c -rw-r--r-- 2.1 KB
list_debug.c -rw-r--r-- 1.7 KB
list_sort.c -rw-r--r-- 5.3 KB
locking-selftest-hardirq.h -rw-r--r-- 207 bytes
locking-selftest-mutex.h -rw-r--r-- 120 bytes
locking-selftest-rlock-hardirq.h -rw-r--r-- 74 bytes
locking-selftest-rlock-softirq.h -rw-r--r-- 74 bytes
locking-selftest-rlock.h -rw-r--r-- 158 bytes
locking-selftest-rsem.h -rw-r--r-- 163 bytes
locking-selftest-softirq.h -rw-r--r-- 207 bytes
locking-selftest-spin-hardirq.h -rw-r--r-- 73 bytes
locking-selftest-spin-softirq.h -rw-r--r-- 73 bytes
locking-selftest-spin.h -rw-r--r-- 118 bytes
locking-selftest-wlock-hardirq.h -rw-r--r-- 74 bytes
locking-selftest-wlock-softirq.h -rw-r--r-- 74 bytes
locking-selftest-wlock.h -rw-r--r-- 158 bytes
locking-selftest-wsem.h -rw-r--r-- 163 bytes
locking-selftest.c -rw-r--r-- 28.6 KB
lru_cache.c -rw-r--r-- 14.6 KB
nlattr.c -rw-r--r-- 12.2 KB
parser.c -rw-r--r-- 6.1 KB
percpu_counter.c -rw-r--r-- 3.7 KB
plist.c -rw-r--r-- 2.9 KB
prio_heap.c -rw-r--r-- 1.4 KB
prio_tree.c -rw-r--r-- 12.2 KB
proportions.c -rw-r--r-- 9.3 KB
radix-tree.c -rw-r--r-- 35.9 KB
random32.c -rw-r--r-- 3.9 KB
ratelimit.c -rw-r--r-- 1.5 KB
rational.c -rw-r--r-- 1.5 KB
rbtree.c -rw-r--r-- 9.9 KB
reciprocal_div.c -rw-r--r-- 159 bytes
rwsem-spinlock.c -rw-r--r-- 6.9 KB
rwsem.c -rw-r--r-- 8.1 KB
scatterlist.c -rw-r--r-- 12.5 KB
sha1.c -rw-r--r-- 2.4 KB
show_mem.c -rw-r--r-- 1.3 KB
smp_processor_id.c -rw-r--r-- 1.1 KB
sort.c -rw-r--r-- 2.5 KB
spinlock_debug.c -rw-r--r-- 6.9 KB
string.c -rw-r--r-- 14.3 KB
string_helpers.c -rw-r--r-- 1.7 KB
swiotlb.c -rw-r--r-- 25.4 KB
syscall.c -rw-r--r-- 2.4 KB
textsearch.c -rw-r--r-- 9.6 KB
ts_bm.c -rw-r--r-- 5.3 KB
ts_fsm.c -rw-r--r-- 10.6 KB
ts_kmp.c -rw-r--r-- 4.3 KB
uuid.c -rw-r--r-- 1.4 KB
vsprintf.c -rw-r--r-- 50.2 KB

back to top