mm, oom: fix race when specifying a thread as the oom origin
authorDavid Rientjes <rientjes@google.com>
Wed, 12 Dec 2012 00:02:56 +0000 (16:02 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 12 Dec 2012 01:22:27 +0000 (17:22 -0800)
commite1e12d2f3104be886073ac6c5c4678f30b1b9e51
treeb08cba1dba28e18cf7c2ffd8d076ce744e368b5f
parenta9c58b907dbc6821533dfc295b63caf111ff1f16
mm, oom: fix race when specifying a thread as the oom origin

test_set_oom_score_adj() and compare_swap_oom_score_adj() are used to
specify that current should be killed first if an oom condition occurs in
between the two calls.

The usage is

short oom_score_adj = test_set_oom_score_adj(OOM_SCORE_ADJ_MAX);
...
compare_swap_oom_score_adj(OOM_SCORE_ADJ_MAX, oom_score_adj);

to store the thread's oom_score_adj, temporarily change it to the maximum
score possible, and then restore the old value if it is still the same.

This happens to still be racy, however, if the user writes
OOM_SCORE_ADJ_MAX to /proc/pid/oom_score_adj in between the two calls.
The compare_swap_oom_score_adj() will then incorrectly reset the old value
prior to the write of OOM_SCORE_ADJ_MAX.

To fix this, introduce a new oom_flags_t member in struct signal_struct
that will be used for per-thread oom killer flags.  KSM and swapoff can
now use a bit in this member to specify that threads should be killed
first in oom conditions without playing around with oom_score_adj.

This also allows the correct oom_score_adj to always be shown when reading
/proc/pid/oom_score.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: Anton Vorontsov <anton.vorontsov@linaro.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/oom.h
include/linux/sched.h
include/linux/types.h
mm/ksm.c
mm/oom_kill.c
mm/swapfile.c