kernel/hung_task.c: defer showing held locks
authorTetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Mon, 8 May 2017 22:55:11 +0000 (15:55 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 9 May 2017 00:15:10 +0000 (17:15 -0700)
commit780cbcf28781511d2cb235c375127265209796a8
tree1ad6a5a7285951266fd955896dace775a62d7940
parent31b8cc80776c1b5a17abda6e0bbb5c615b9d90e4
kernel/hung_task.c: defer showing held locks

When I was running my testcase which may block hundreds of threads on fs
locks, I got lockup due to output from debug_show_all_locks() added by
commit b2d4c2edb2e4 ("locking/hung_task: Show all locks").

For example, if 1000 threads were blocked in TASK_UNINTERRUPTIBLE state
and 500 out of 1000 threads hold some lock, debug_show_all_locks() from
for_each_process_thread() loop will report locks held by 500 threads for
1000 times.  This is a too much noise.

In order to make sure rcu_lock_break() is called frequently, we should
avoid calling debug_show_all_locks() from for_each_process_thread() loop
because debug_show_all_locks() effectively calls for_each_process_thread()
loop.  Let's defer calling debug_show_all_locks() till before panic() or
leaving for_each_process_thread() loop.

Link: http://lkml.kernel.org/r/1489296834-60436-1-git-send-email-penguin-kernel@I-love.SAKURA.ne.jp
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reviewed-by: Vegard Nossum <vegard.nossum@oracle.com>
Cc: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/hung_task.c