Linux 3.12.22-rt35-rc1

Dear RT Folks,

This is the RT stable review cycle of patch 3.12.22-rt35-rc1.

Please scream at me if I messed something up. Please test the patches too.

The -rc release will be uploaded to kernel.org and will be deleted when the final release is out. This is just a review release (or release candidate).

The pre-releases will not be pushed to the git repository, only the final release is.

If all goes well, this patch will be converted to the next main release on 7/7/2014.

Enjoy,

-- Steve

To build 3.12.22-rt35-rc1 directly, the following patches should be applied:

http://www.kernel.org/pub/linux/kernel/v3.x/linux-3.12.tar.xz

http://www.kernel.org/pub/linux/kernel/v3.x/patch-3.12.22.xz

http://www.kernel.org/pub/linux/kernel/projects/rt/3.12/patch-3.12.22-rt35-rc1.patch.xz

You can also build from 3.12.22-rt34 by applying the incremental patch:

http://www.kernel.org/pub/linux/kernel/projects/rt/3.12/incr/patch-3.12.22-rt34-rt35-rc1.patch.xz

Changes from 3.12.22-rt34:

---

Ben Hutchings (1):
Fix latency histogram after "hrtimer: Set expiry time before switch_hrtimer_base()"

Sebastian Andrzej Siewior (2):
Revert "migrate_disable pushd down in atomic_dec_and_spin_lock"
timer: do not spin_trylock() on UP

Steven Rostedt (Red Hat) (1):

      Linux 3.12.22-rt35-rc1

----

 kernel/hrtimer.c | 10 +++++-----
 kernel/rtmutex.c |  6 +++---
 kernel/timer.c   | 13 +++++++++++++
 localversion-rt  |  2 +-
 4 files changed, 22 insertions(+), 9 deletions(-)

3.12.22-rt35-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

This reverts commit ff9c870c3e27d58c9512fad122e91436681fee5a.

---
 kernel/rtmutex.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

index 4057bc607923..d1b2ca08b160 100644
@@ -952,12 +952,12 @@ int atomic_dec_and_spin_lock(atomic_t *atomic, spinlock_t *lock)
 	/* Subtract 1 from counter unless that drops it to 0 (ie. it was 1) */
 	if (atomic_add_unless(atomic, -1, 1))
 		return 0;
+	migrate_disable();
 	rt_spin_lock(lock);
-	if (atomic_dec_and_test(atomic)){
-		migrate_disable();
+	if (atomic_dec_and_test(atomic))
 		return 1;
-	}
 	rt_spin_unlock(lock);
+	migrate_enable();
 	return 0;
 }
 EXPORT_SYMBOL(atomic_dec_and_spin_lock);

2.0.0

3.12.22-rt35-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

This will void a warning comming from the spin-lock debugging code. The lock avoiding idea is from Steven Rostedt.

---
 kernel/timer.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

index 611a5973d9e1..3b79da2f9c67 100644
@@ -1461,6 +1461,19 @@ void run_local_timers(void)
 	 * the timer softirq.
 	 */
 #ifdef CONFIG_PREEMPT_RT_FULL
+
+#ifndef CONFIG_SMP
+	/*
+	 * The spin_do_trylock() later may fail as the lock may be hold before
+	 * the interrupt arrived. The spin-lock debugging code will raise a
+	 * warning if the try_lock fails on UP. Since this is only an
+	 * optimization for the FULL_NO_HZ case (not to run the timer softirq on
+	 * an nohz_full CPU) we don't really care and shedule the softirq.
+	 */
+	raise_softirq(TIMER_SOFTIRQ);
+	return;
+#endif
+
 	/* On RT, irq work runs from softirq */
 	if (irq_work_needs_cpu()) {
 		raise_softirq(TIMER_SOFTIRQ);

2.0.0

3.12.22-rt35-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Ben Hutchings <[email protected]>

In an rt-kernel with CONFIG_MISSED_TIMER_OFFSETS_HIST enabled, __hrtimer_start_range_ns() now crashes, as new_base is not assigned before it is used.

---
 kernel/hrtimer.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

index bc0e47dc0a05..c967b718ab37 100644
@@ -1105,6 +1105,11 @@ int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim,
 #endif
 	}
 
+	hrtimer_set_expires_range_ns(timer, tim, delta_ns);
+
+	/* Switch the timer base, if necessary: */
+	new_base = switch_hrtimer_base(timer, base, mode & HRTIMER_MODE_PINNED);
+
 #ifdef CONFIG_MISSED_TIMER_OFFSETS_HIST
 	{
 		ktime_t now = new_base->get_time();

@@ -1116,11 +1121,6 @@ int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim,
 	}
 #endif
 
-	hrtimer_set_expires_range_ns(timer, tim, delta_ns);
-
-	/* Switch the timer base, if necessary: */
-	new_base = switch_hrtimer_base(timer, base, mode & HRTIMER_MODE_PINNED);
-
 	timer_stats_hrtimer_set_start_info(timer);
 
 	leftmost = enqueue_hrtimer(timer, new_base);

2.0.0

3.12.22-rt35-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: "Steven Rostedt (Red Hat)" <[email protected]>

---
 localversion-rt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 21988f9ad53f..87569d22628e 100644
@@ -1 +1 @@
--rt34
+-rt35-rc1

2.0.0

[PATCH RT 5/4] sched: Do not clear PF_NO_SETAFFINITY flag in select_fallback_rq() ?

-Mike

Is that in 3.14-rt yet? I didn't see it. As with stable mainline, nothing goes into stable-rt until it is in -rt upstream.

-- Steve

Oh right, no, not yet.

-Mike