linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [patch] high-res timers: core, do itimer rearming in process context, fix
@ 2006-12-19 21:16 Ingo Molnar
  2006-12-21  7:15 ` [patch] high-res timers: core, do itimer rearming in process context, fix2 Thomas Gleixner
  0 siblings, 1 reply; 2+ messages in thread
From: Ingo Molnar @ 2006-12-19 21:16 UTC (permalink / raw)
  To: Andrew Morton; +Cc: Thomas Gleixner, linux-kernel

Subject: [patch] high-res timers: core, do itimer rearming in process context, fix
From: Ingo Molnar <mingo@elte.hu>

restart itimers when they are not queued. (a non-queued hrtimer might
be callback-pending - but in that case we already missed the rearming
so the SIGALRM branch has to rearm.)

this fixes a threaded itimers stress-test hang.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 include/linux/hrtimer.h |    9 +++++++++
 kernel/hrtimer.c        |    9 ---------
 kernel/signal.c         |    2 +-
 3 files changed, 10 insertions(+), 10 deletions(-)

Index: linux/include/linux/hrtimer.h
===================================================================
--- linux.orig/include/linux/hrtimer.h
+++ linux/include/linux/hrtimer.h
@@ -309,6 +309,15 @@ static inline int hrtimer_active(const s
 	return timer->state != HRTIMER_STATE_INACTIVE;
 }
 
+/*
+ * Helper function to check, whether the timer is on one of the queues
+ */
+static inline int hrtimer_is_queued(struct hrtimer *timer)
+{
+	return timer->state &
+		(HRTIMER_STATE_ENQUEUED | HRTIMER_STATE_PENDING);
+}
+
 /* Forward a hrtimer so it expires after now: */
 extern unsigned long
 hrtimer_forward(struct hrtimer *timer, ktime_t now, ktime_t interval);
Index: linux/kernel/hrtimer.c
===================================================================
--- linux.orig/kernel/hrtimer.c
+++ linux/kernel/hrtimer.c
@@ -152,15 +152,6 @@ static void hrtimer_get_softirq_time(str
 }
 
 /*
- * Helper function to check, whether the timer is on one of the queues
- */
-static inline int hrtimer_is_queued(struct hrtimer *timer)
-{
-	return timer->state &
-		(HRTIMER_STATE_ENQUEUED | HRTIMER_STATE_PENDING);
-}
-
-/*
  * Helper function to check, whether the timer is running the callback
  * function
  */
Index: linux/kernel/signal.c
===================================================================
--- linux.orig/kernel/signal.c
+++ linux/kernel/signal.c
@@ -475,7 +475,7 @@ int dequeue_signal(struct task_struct *t
 		if (unlikely(signr == SIGALRM)) {
 			struct hrtimer *tmr = &tsk->signal->real_timer;
 
-			if (!hrtimer_active(tmr) &&
+			if (!hrtimer_is_queued(tmr) &&
 			    tsk->signal->it_real_incr.tv64 != 0) {
 				hrtimer_forward(tmr, hrtimer_cb_get_time(tmr),
 						tsk->signal->it_real_incr);

^ permalink raw reply	[flat|nested] 2+ messages in thread

* [patch] high-res timers: core, do itimer rearming in process context, fix2
  2006-12-19 21:16 [patch] high-res timers: core, do itimer rearming in process context, fix Ingo Molnar
@ 2006-12-21  7:15 ` Thomas Gleixner
  0 siblings, 0 replies; 2+ messages in thread
From: Thomas Gleixner @ 2006-12-21  7:15 UTC (permalink / raw)
  To: Ingo Molnar; +Cc: Andrew Morton, linux-kernel

Subject: [patch] high-res timers: core, do itimer rearming in process context, fix2
From: Thomas Gleixner <tglx@linutronix.de>

The rearming code in signal.c has to read the time and can not rely on
the timer->base->softirq time anymore, as it is not longer running in
softirq context.

Ensure, that the it_real_incr variable in the shared signal struct is
set to zero, when setitimer disables the itimer. Otherwise it could
happen that an inactive itimer gets rearmed by a SIGALRM.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 kernel/signal.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: 2.6.19-git19/kernel/signal.c
===================================================================
--- 2.6.19-git19.orig/kernel/signal.c
+++ 2.6.19-git19/kernel/signal.c
@@ -477,7 +477,7 @@ int dequeue_signal(struct task_struct *t
 
 			if (!hrtimer_is_queued(tmr) &&
 			    tsk->signal->it_real_incr.tv64 != 0) {
-				hrtimer_forward(tmr, hrtimer_cb_get_time(tmr),
+				hrtimer_forward(tmr, tmr->base->get_time(),
 						tsk->signal->it_real_incr);
 				hrtimer_restart(tmr);
 			}
Index: 2.6.19-git19/kernel/itimer.c
===================================================================
--- 2.6.19-git19.orig/kernel/itimer.c
+++ 2.6.19-git19/kernel/itimer.c
@@ -226,11 +226,14 @@ again:
 			spin_unlock_irq(&tsk->sighand->siglock);
 			goto again;
 		}
-		tsk->signal->it_real_incr =
-			timeval_to_ktime(value->it_interval);
 		expires = timeval_to_ktime(value->it_value);
-		if (expires.tv64 != 0)
+		if (expires.tv64 != 0) {
+			tsk->signal->it_real_incr =
+				timeval_to_ktime(value->it_interval);
 			hrtimer_start(timer, expires, HRTIMER_MODE_REL);
+		} else
+			tsk->signal->it_real_incr.tv64 = 0;
+
 		spin_unlock_irq(&tsk->sighand->siglock);
 		break;
 	case ITIMER_VIRTUAL:



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2006-12-21  7:11 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2006-12-19 21:16 [patch] high-res timers: core, do itimer rearming in process context, fix Ingo Molnar
2006-12-21  7:15 ` [patch] high-res timers: core, do itimer rearming in process context, fix2 Thomas Gleixner

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).