]> git.kernelconcepts.de Git - karo-tx-linux.git/commitdiff
memcg: do not call reclaim if !__GFP_WAIT
authorVladimir Davydov <vdavydov@parallels.com>
Wed, 10 Jun 2015 18:14:46 +0000 (11:14 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 10 Jun 2015 23:43:43 +0000 (16:43 -0700)
When trimming memcg consumption excess (see memory.high), we call
try_to_free_mem_cgroup_pages without checking if we are allowed to sleep
in the current context, which can result in a deadlock.  Fix this.

Fixes: 241994ed8649 ("mm: memcontrol: default hierarchy interface for memory")
Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/memcontrol.c

index 14c2f2017e37cc405e52cb12bc30b128997f1f8e..9da23a7ec4c00b04ced3e349e8dc8a5c6ca016b3 100644 (file)
@@ -2323,6 +2323,8 @@ done_restock:
        css_get_many(&memcg->css, batch);
        if (batch > nr_pages)
                refill_stock(memcg, batch - nr_pages);
+       if (!(gfp_mask & __GFP_WAIT))
+               goto done;
        /*
         * If the hierarchy is above the normal consumption range,
         * make the charging task trim their excess contribution.