During RCU-boost testing with the TREE03 rcutorture config, I found that
after a few hours, the machine locks up.
On tracing, I found that there is a live lock happening between 2 CPUs.
One CPU has an RT task running, while another CPU is being offlined
which also has an RT task running. During this offlining, all threads
are migrated. The migration thread is repeatedly scheduled to migrate
actively running tasks on the CPU being offlined. This results in a live
lock because select_fallback_rq() keeps picking the CPU that an RT task
is already running on only to get pushed back to the CPU being offlined.
It is anyway pointless to pick CPUs for pushing tasks to if they are
being offlined only to get migrated away to somewhere else. This could
also add unwanted latency to this task.
Fix these issues by not selecting CPUs in RT if they are not 'active'
for scheduling, using the cpu_active_mask. Other parts in core.c already
use cpu_active_mask to prevent tasks from being put on CPUs going
offline.
Tested-by: Paul E. McKenney <[email protected]>
Cc: [email protected]
Signed-off-by: Joel Fernandes (Google) <[email protected]>
---
kernel/sched/cpupri.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/kernel/sched/cpupri.c b/kernel/sched/cpupri.c
index a286e726eb4b..42c40cfdf836 100644
--- a/kernel/sched/cpupri.c
+++ b/kernel/sched/cpupri.c
@@ -101,6 +101,7 @@ static inline int __cpupri_find(struct cpupri *cp, struct task_struct *p,
if (lowest_mask) {
cpumask_and(lowest_mask, &p->cpus_mask, vec->mask);
+ cpumask_and(lowest_mask, lowest_mask, cpu_active_mask);
/*
* We have to ensure that we have at least one bit
--
2.42.0.515.g380fc7ccd1-goog
The following commit has been merged into the sched/urgent branch of tip:
Commit-ID: fc09027786c900368de98d03d40af058bcb01ad9
Gitweb: https://git.kernel.org/tip/fc09027786c900368de98d03d40af058bcb01ad9
Author: Joel Fernandes (Google) <[email protected]>
AuthorDate: Sat, 23 Sep 2023 01:14:08
Committer: Ingo Molnar <[email protected]>
CommitterDate: Thu, 28 Sep 2023 22:58:13 +02:00
sched/rt: Fix live lock between select_fallback_rq() and RT push
During RCU-boost testing with the TREE03 rcutorture config, I found that
after a few hours, the machine locks up.
On tracing, I found that there is a live lock happening between 2 CPUs.
One CPU has an RT task running, while another CPU is being offlined
which also has an RT task running. During this offlining, all threads
are migrated. The migration thread is repeatedly scheduled to migrate
actively running tasks on the CPU being offlined. This results in a live
lock because select_fallback_rq() keeps picking the CPU that an RT task
is already running on only to get pushed back to the CPU being offlined.
It is anyway pointless to pick CPUs for pushing tasks to if they are
being offlined only to get migrated away to somewhere else. This could
also add unwanted latency to this task.
Fix these issues by not selecting CPUs in RT if they are not 'active'
for scheduling, using the cpu_active_mask. Other parts in core.c already
use cpu_active_mask to prevent tasks from being put on CPUs going
offline.
With this fix I ran the tests for days and could not reproduce the
hang. Without the patch, I hit it in a few hours.
Signed-off-by: Joel Fernandes (Google) <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Tested-by: Paul E. McKenney <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
---
kernel/sched/cpupri.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/kernel/sched/cpupri.c b/kernel/sched/cpupri.c
index a286e72..42c40cf 100644
--- a/kernel/sched/cpupri.c
+++ b/kernel/sched/cpupri.c
@@ -101,6 +101,7 @@ static inline int __cpupri_find(struct cpupri *cp, struct task_struct *p,
if (lowest_mask) {
cpumask_and(lowest_mask, &p->cpus_mask, vec->mask);
+ cpumask_and(lowest_mask, lowest_mask, cpu_active_mask);
/*
* We have to ensure that we have at least one bit
Hey Joel
On 09/23/23 01:14, Joel Fernandes (Google) wrote:
> During RCU-boost testing with the TREE03 rcutorture config, I found that
> after a few hours, the machine locks up.
>
> On tracing, I found that there is a live lock happening between 2 CPUs.
> One CPU has an RT task running, while another CPU is being offlined
> which also has an RT task running. During this offlining, all threads
> are migrated. The migration thread is repeatedly scheduled to migrate
> actively running tasks on the CPU being offlined. This results in a live
> lock because select_fallback_rq() keeps picking the CPU that an RT task
> is already running on only to get pushed back to the CPU being offlined.
>
> It is anyway pointless to pick CPUs for pushing tasks to if they are
> being offlined only to get migrated away to somewhere else. This could
> also add unwanted latency to this task.
>
> Fix these issues by not selecting CPUs in RT if they are not 'active'
> for scheduling, using the cpu_active_mask. Other parts in core.c already
> use cpu_active_mask to prevent tasks from being put on CPUs going
> offline.
I think this is the same report as this one from Xuewen
https://lore.kernel.org/lkml/[email protected]/
You also have only 2 CPUs system, so when one CPU is offline, the other becomes
overloaded.
I think it is worth trying to fix rto_next_cpu() to consider the
cpu_active_mask too (see my reply to Xuewen then). Keep your change as well.
I think Xuewen and I considered disabling the overloaded logic when we devolve
to a UP system too as the whole logic doesn't make sense anymore then. This is
more of an optimization than correctness though. Don't feel strongly about it
now though.
Thanks!
--
Qais Yousef
>
> Tested-by: Paul E. McKenney <[email protected]>
> Cc: [email protected]
> Signed-off-by: Joel Fernandes (Google) <[email protected]>
> ---
> kernel/sched/cpupri.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/kernel/sched/cpupri.c b/kernel/sched/cpupri.c
> index a286e726eb4b..42c40cfdf836 100644
> --- a/kernel/sched/cpupri.c
> +++ b/kernel/sched/cpupri.c
> @@ -101,6 +101,7 @@ static inline int __cpupri_find(struct cpupri *cp, struct task_struct *p,
>
> if (lowest_mask) {
> cpumask_and(lowest_mask, &p->cpus_mask, vec->mask);
> + cpumask_and(lowest_mask, lowest_mask, cpu_active_mask);
>
> /*
> * We have to ensure that we have at least one bit
> --
> 2.42.0.515.g380fc7ccd1-goog
>