In the function tick_unfreeze(), it would resume timekeeping for the
first cpu, and for other cpus it does local tick resume only, not
to touch watchdog which is per-cpu as well.
This probably is not the problem for suspend to ram, but for suspend
to idle is. Since watchdog would be setup for each unplugged cores when
plugging during resume and would be touched in its setup process. But for
suspend to idle the system wouldn't unplug/plug cores, so we should
touch watchdog once resume from s2idle to avoid 'soft lockup' warnings
due to timeout detected by watchdog.
Signed-off-by: Chunyan Zhang <[email protected]>
---
kernel/time/tick-common.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/kernel/time/tick-common.c b/kernel/time/tick-common.c
index 59225b484e4e..7e5d3524e924 100644
--- a/kernel/time/tick-common.c
+++ b/kernel/time/tick-common.c
@@ -11,6 +11,7 @@
#include <linux/err.h>
#include <linux/hrtimer.h>
#include <linux/interrupt.h>
+#include <linux/nmi.h>
#include <linux/percpu.h>
#include <linux/profile.h>
#include <linux/sched.h>
@@ -558,6 +559,7 @@ void tick_unfreeze(void)
trace_suspend_resume(TPS("timekeeping_freeze"),
smp_processor_id(), false);
} else {
+ touch_softlockup_watchdog();
tick_resume_local();
}
--
2.20.1
The following commit has been merged into the timers/core branch of tip:
Commit-ID: 5167c506d62dd9ffab73eba23c79b0a8845c9fe1
Gitweb: https://git.kernel.org/tip/5167c506d62dd9ffab73eba23c79b0a8845c9fe1
Author: Chunyan Zhang <[email protected]>
AuthorDate: Fri, 10 Jan 2020 16:39:02 +08:00
Committer: Thomas Gleixner <[email protected]>
CommitterDate: Wed, 15 Jan 2020 21:29:45 +01:00
tick/common: Touch watchdog in tick_unfreeze() on all CPUs
Suspend to IDLE invokes tick_unfreeze() on resume. tick_unfreeze() on the
first resuming CPU resumes timekeeping, which also has the side effect of
resetting the softlockup watchdog on this CPU.
But on the secondary CPUs the watchdog is not reset in the resume /
unfreeze() path, which can result in false softlockup warnings on those
CPUs depending on the time spent in suspend.
Prevent this by clearing the softlock watchdog in the unfreeze path also
on the secondary resuming CPUs.
[ tglx: Massaged changelog ]
Signed-off-by: Chunyan Zhang <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
kernel/time/tick-common.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/kernel/time/tick-common.c b/kernel/time/tick-common.c
index 59225b4..7e5d352 100644
--- a/kernel/time/tick-common.c
+++ b/kernel/time/tick-common.c
@@ -11,6 +11,7 @@
#include <linux/err.h>
#include <linux/hrtimer.h>
#include <linux/interrupt.h>
+#include <linux/nmi.h>
#include <linux/percpu.h>
#include <linux/profile.h>
#include <linux/sched.h>
@@ -558,6 +559,7 @@ void tick_unfreeze(void)
trace_suspend_resume(TPS("timekeeping_freeze"),
smp_processor_id(), false);
} else {
+ touch_softlockup_watchdog();
tick_resume_local();
}