sched: Fix preempt_schedule_common() triggering tracing recursion
authorFrederic Weisbecker <fweisbec@gmail.com>
Mon, 16 Feb 2015 18:20:07 +0000 (19:20 +0100)
committerIngo Molnar <mingo@kernel.org>
Wed, 18 Feb 2015 13:27:38 +0000 (14:27 +0100)
Since the function graph tracer needs to disable preemption, it might
call preempt_schedule() after reenabling  it if something triggered the
need for rescheduling in between.

Therefore we can't trace preempt_schedule() itself because we would
face a function tracing recursion otherwise as the tracer is always
called before PREEMPT_ACTIVE gets set to prevent that recursion. This is
why preempt_schedule() is tagged as "notrace".

But the same issue applies to every function called by preempt_schedule()
before PREEMPT_ACTIVE is actually set. And preempt_schedule_common() is
one such example. Unfortunately we forgot to tag it as notrace as well
and as a result we are encountering tracing recursion since it got
introduced by:

   a18b5d0181923 ("sched: Fix missing preemption opportunity")

Let's fix that by applying the appropriate function tag to
preempt_schedule_common().

Reported-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1424110807-15057-1-git-send-email-fweisbec@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/core.c

index f77acd98f50a3ed0742cd3423a980df900b69b55..c314000f5e5295aa980459f9ea67deb2e3dd04da 100644 (file)
@@ -2839,7 +2839,7 @@ void __sched schedule_preempt_disabled(void)
        preempt_disable();
 }
 
-static void preempt_schedule_common(void)
+static void __sched notrace preempt_schedule_common(void)
 {
        do {
                __preempt_count_add(PREEMPT_ACTIVE);