fs/dcache.c: avoid soft-lockup in dput()
authorWei Fang <fangwei1@huawei.com>
Wed, 6 Jul 2016 03:32:20 +0000 (11:32 +0800)
committerAl Viro <viro@zeniv.linux.org.uk>
Sun, 24 Jul 2016 20:37:16 +0000 (16:37 -0400)
commit47be61845c775643f1aa4d2a54343549f943c94c
tree7bc3d07cecb30116a51bdf11dfa0540bfacc4e43
parent285b102d3b745f3c2c110c9c327741d87e64aacc
fs/dcache.c: avoid soft-lockup in dput()

We triggered soft-lockup under stress test which
open/access/write/close one file concurrently on more than
five different CPUs:

WARN: soft lockup - CPU#0 stuck for 11s! [who:30631]
...
[<ffffffc0003986f8>] dput+0x100/0x298
[<ffffffc00038c2dc>] terminate_walk+0x4c/0x60
[<ffffffc00038f56c>] path_lookupat+0x5cc/0x7a8
[<ffffffc00038f780>] filename_lookup+0x38/0xf0
[<ffffffc000391180>] user_path_at_empty+0x78/0xd0
[<ffffffc0003911f4>] user_path_at+0x1c/0x28
[<ffffffc00037d4fc>] SyS_faccessat+0xb4/0x230

->d_lock trylock may failed many times because of concurrently
operations, and dput() may execute a long time.

Fix this by replacing cpu_relax() with cond_resched().
dput() used to be sleepable, so make it sleepable again
should be safe.

Cc: <stable@vger.kernel.org>
Signed-off-by: Wei Fang <fangwei1@huawei.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
fs/dcache.c