Re: Linux 3.19-rc5

[Oleg Nesterov <oleg@redhat.com>]

On 01/31, Peter Zijlstra wrote:
>
> On Sat, Jan 31, 2015 at 10:32:23AM -0800, Linus Torvalds wrote:
> > On Fri, Jan 30, 2015 at 7:47 AM, Oleg Nesterov <oleg@redhat.com> wrote:
> > >
> > > Perhaps sched_annotate_sleep() shouldn't depend on CONFIG_DEBUG_ATOMIC_SLEEP
> > > too...
> >
> > Ugh. That thing is horrible. The naming doesn't make it obvious at all
> > that it's actually making sure that we have state set to TASK_RUNNING,
> > and I could easily imagine that it would cause similar "busy-loops
> > while scheduling" issues if anybody ever uses it in the wrong context.
>
> The alternative was putting unconditional __set_task_state(TASK_RUNNING)
> things in a few code paths. I didn't want to cause the extra code in
> case we didn't need them. Particularly the include/net/sock.h one, as I
> know the network people are cycle counters.

And personally I agree. sched_annotate_sleep() looks self-documented, it
is clear that it is used to suppress the warning.

Still. Can't we avoid this subtle change in behaviour DEBUG_ATOMIC_SLEEP
adds?

Oleg.


--- x/include/linux/kernel.h
+++ x/include/linux/kernel.h
@@ -176,7 +176,7 @@ extern int _cond_resched(void);
  */
 # define might_sleep() \
 	do { __might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
-# define sched_annotate_sleep()	__set_current_state(TASK_RUNNING)
+# define sched_annotate_sleep()	(current->task_state_change = 0)
 #else
   static inline void ___might_sleep(const char *file, int line,
 				   int preempt_offset) { }
--- x/kernel/sched/core.c
+++ x/kernel/sched/core.c
@@ -7292,7 +7292,7 @@ void __might_sleep(const char *file, int
 	 * since we will exit with TASK_RUNNING make sure we enter with it,
 	 * otherwise we will destroy state.
 	 */
-	if (WARN_ONCE(current->state != TASK_RUNNING,
+	if (WARN_ONCE(current->state != TASK_RUNNING && current->task_state_change,
 			"do not call blocking ops when !TASK_RUNNING; "
 			"state=%lx set at [<%p>] %pS\n",
 			current->state,