task_work: add a scheduling point in task_work_run()
authorEric Dumazet <edumazet@google.com>
Tue, 21 Aug 2012 13:05:14 +0000 (15:05 +0200)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 21 Aug 2012 16:11:44 +0000 (09:11 -0700)
commitf341861fb0b701139849f8a85c2d3cdff466e8e8
treea6e6fedd03fdb923828fa7143a0363738cd7dfcb
parentec5da7f8dc023fc4dcbdd42e87dad231d2c2812d
task_work: add a scheduling point in task_work_run()

It seems commit 4a9d4b024a31 ("switch fput to task_work_add") re-
introduced the problem addressed in 944be0b22472 ("close_files(): add
scheduling point")

If a server process with a lot of files (say 2 million tcp sockets) is
killed, we can spend a lot of time in task_work_run() and trigger a soft
lockup.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/task_work.c