[ 3.5.y.z extended stable ] Patch "hrtimers: Move SMP function call to thread context" has been added to staging queue

[Luis Henriques]

This is a note to let you know that I have just added a patch titled

    hrtimers: Move SMP function call to thread context

to the linux-3.5.y-queue branch of the 3.5.y.z extended stable tree 
which can be found at:

 http://kernel.ubuntu.com/git?p=ubuntu/linux.git;a=shortlog;h=refs/heads/linux-3.5.y-queue

If you, or anyone else, feels it should not be added to this tree, please 
reply to this email.

For more information about the 3.5.y.z tree, see
https://wiki.ubuntu.com/Kernel/Dev/ExtendedStable

Thanks.
-Luis

------

>From c13659bfc7281cbe868c20db9b34dbc67c84bfa4 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx at linutronix.de>
Date: Fri, 5 Jul 2013 12:09:18 +0200
Subject: [PATCH] hrtimers: Move SMP function call to thread context

commit 5ec2481b7b47a4005bb446d176e5d0257400c77d upstream.

smp_call_function_* must not be called from softirq context.

But clock_was_set() which calls on_each_cpu() is called from softirq
context to implement a delayed clock_was_set() for the timer interrupt
handler. Though that almost never gets invoked. A recent change in the
resume code uses the softirq based delayed clock_was_set to support
Xens resume mechanism.

linux-next contains a new warning which warns if smp_call_function_*
is called from softirq context which gets triggered by that Xen
change.

Fix this by moving the delayed clock_was_set() call to a work context.

Reported-and-tested-by: Artem Savkov <artem.savkov at gmail.com>
Reported-by: Sasha Levin <sasha.levin at oracle.com>
Cc: David Vrabel <david.vrabel at citrix.com>
Cc: Ingo Molnar <mingo at kernel.org>
Cc: H. Peter Anvin <hpa at zytor.com>,
Cc: Konrad Wilk <konrad.wilk at oracle.com>
Cc: John Stultz <john.stultz at linaro.org>
Cc: xen-devel at lists.xen.org
Signed-off-by: Thomas Gleixner <tglx at linutronix.de>
[ luis: backported to 3.5: adjusted context ]
Signed-off-by: Luis Henriques <luis.henriques at canonical.com>
---
 kernel/hrtimer.c | 28 +++++++++++++---------------
 1 file changed, 13 insertions(+), 15 deletions(-)

diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
index 60f7e32..a57ef25 100644
--- a/kernel/hrtimer.c
+++ b/kernel/hrtimer.c
@@ -707,17 +707,20 @@ static int hrtimer_switch_to_hres(void)
 	return 1;
 }

+static void clock_was_set_work(struct work_struct *work)
+{
+	clock_was_set();
+}
+
+static DECLARE_WORK(hrtimer_work, clock_was_set_work);
+
 /*
- * Called from timekeeping code to reprogramm the hrtimer interrupt
- * device. If called from the timer interrupt context we defer it to
- * softirq context.
+ * Called from timekeeping and resume code to reprogramm the hrtimer
+ * interrupt device on all cpus.
  */
 void clock_was_set_delayed(void)
 {
-	struct hrtimer_cpu_base *cpu_base = &__get_cpu_var(hrtimer_bases);
-
-	cpu_base->clock_was_set = 1;
-	__raise_softirq_irqoff(HRTIMER_SOFTIRQ);
+	schedule_work(&hrtimer_work);
 }

 #else
@@ -766,8 +769,10 @@ void hrtimers_resume(void)
 	WARN_ONCE(!irqs_disabled(),
 		  KERN_INFO "hrtimers_resume() called with IRQs enabled!");

+	/* Retrigger on the local CPU */
 	retrigger_next_event(NULL);
-	timerfd_clock_was_set();
+	/* And schedule a retrigger for all others */
+	clock_was_set_delayed();
 }

 static inline void timer_stats_hrtimer_set_start_info(struct hrtimer *timer)
@@ -1416,13 +1421,6 @@ void hrtimer_peek_ahead_timers(void)

 static void run_hrtimer_softirq(struct softirq_action *h)
 {
-	struct hrtimer_cpu_base *cpu_base = &__get_cpu_var(hrtimer_bases);
-
-	if (cpu_base->clock_was_set) {
-		cpu_base->clock_was_set = 0;
-		clock_was_set();
-	}
-
 	hrtimer_peek_ahead_timers();
 }

--
1.8.1.2