From: Frederic Weisbecker <frederic@xxxxxxxxxx>
The hrtimers migration on CPU-down hotplug process has been moved
earlier, before the CPU actually goes to die. This leaves a small window
of opportunity to queue an hrtimer in a blind spot, leaving it ignored.
For example a practical case has been reported with RCU waking up a
SCHED_FIFO task right before the CPUHP_AP_IDLE_DEAD stage, queuing that
way a sched/rt timer to the local offline CPU.
Make sure such situations never go unnoticed and warn when that happens.
Reported-by: Paul E. McKenney <paulmck@xxxxxxxxxx>
Fixes: 5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier")
Signed-off-by: Frederic Weisbecker <frederic@xxxxxxxxxx>
Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>
---
include/linux/hrtimer.h | 3 ++-
kernel/time/hrtimer.c | 3 +++
2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index 87e3bedf8eb0..4f2cf7309486 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -179,7 +179,8 @@ struct hrtimer_cpu_base {
unsigned int hres_active : 1,
in_hrtirq : 1,
hang_detected : 1,
- softirq_activated : 1;
+ softirq_activated : 1,
+ online : 1;
#ifdef CONFIG_HIGH_RES_TIMERS
unsigned int nr_events;
unsigned short nr_retries;
diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 760793998cdd..edb0f821dcea 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -1085,6 +1085,7 @@ static int enqueue_hrtimer(struct hrtimer *timer,
enum hrtimer_mode mode)
{
debug_activate(timer, mode);
+ WARN_ON_ONCE(!base->cpu_base->online);
base->cpu_base->active_bases |= 1 << base->index;
@@ -2183,6 +2184,7 @@ int hrtimers_prepare_cpu(unsigned int cpu)
cpu_base->softirq_next_timer = NULL;
cpu_base->expires_next = KTIME_MAX;
cpu_base->softirq_expires_next = KTIME_MAX;
+ cpu_base->online = 1;
hrtimer_cpu_base_init_expiry_lock(cpu_base);
return 0;
}
@@ -2250,6 +2252,7 @@ int hrtimers_cpu_dying(unsigned int dying_cpu)
smp_call_function_single(ncpu, retrigger_next_event, NULL, 0);
raw_spin_unlock(&new_base->lock);
+ old_base->online = 0;
raw_spin_unlock(&old_base->lock);
return 0;
--
2.43.0
