hung_task: fix false positive during vfork
Mandeep Singh Baines [Tue, 3 Jan 2012 22:41:13 +0000 (14:41 -0800)]
commit f9fab10bbd768b0e5254e53a4a8477a94bfc4b96 upstream.

vfork parent uninterruptibly and unkillably waits for its child to
exec/exit. This wait is of unbounded length. Ignore such waits
in the hung_task detector.

Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reported-by: Sasha Levin <levinsasha928@gmail.com>
LKML-Reference: <1325344394.28904.43.camel@lappy>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: John Kacur <jkacur@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Change-Id: I12ed227da1fb7099978c73b65a756caf51a95553
Reviewed-on: http://git-master/r/74230
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>
Tested-by: Varun Wadekar <vwadekar@nvidia.com>

kernel/hung_task.c

index ea64012..e972276 100644 (file)
@@ -74,11 +74,17 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout)
 
        /*
         * Ensure the task is not frozen.
-        * Also, when a freshly created task is scheduled once, changes
-        * its state to TASK_UNINTERRUPTIBLE without having ever been
-        * switched out once, it musn't be checked.
+        * Also, skip vfork and any other user process that freezer should skip.
         */
-       if (unlikely(t->flags & PF_FROZEN || !switch_count))
+       if (unlikely(t->flags & (PF_FROZEN | PF_FREEZER_SKIP)))
+           return;
+
+       /*
+        * When a freshly created task is scheduled once, changes its state to
+        * TASK_UNINTERRUPTIBLE without having ever been switched out once, it
+        * musn't be checked.
+        */
+       if (unlikely(!switch_count))
                return;
 
        if (switch_count != t->last_switch_count) {