[v1] cgroup/cpuset: Extend kthread_is_per_cpu() check to all PF_NO_SETAFFINITY tasks

[PATCH] cgroup/cpuset: Extend kthread_is_per_cpu() check to all PF_NO_SETAFFINITY tasks

Posted by Waiman Long 7 months, 1 week ago

Commit ec5fbdfb99d1 ("cgroup/cpuset: Enable update_tasks_cpumask()
on top_cpuset") enabled us to pull CPUs dedicated to child partitions
from tasks in top_cpuset by ignoring per cpu kthreads. However, there
can be other kthreads that are not per cpu but have PF_NO_SETAFFINITY
flag set to indicate that we shouldn't mess with their CPU affinity.

As all the per cpu kthreads have PF_NO_SETAFFINITY set, fix this
issue by dropping the kthread_is_per_cpu() check and checking the
PF_NO_SETAFFINITY flag instead.

Fixes: ec5fbdfb99d1 ("cgroup/cpuset: Enable update_tasks_cpumask() on top_cpuset")
Signed-off-by: Waiman Long <longman@redhat.com>
---
 kernel/cgroup/cpuset.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index d0143b3dce47..581032d73e4b 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -1130,9 +1130,9 @@ void cpuset_update_tasks_cpumask(struct cpuset *cs, struct cpumask *new_cpus)
 
 		if (top_cs) {
 			/*
-			 * Percpu kthreads in top_cpuset are ignored
+			 * PF_NO_SETAFFINITY tasks are ignored
 			 */
-			if (kthread_is_per_cpu(task))
+			if (task->flags & PF_NO_SETAFFINITY)
 				continue;
 			cpumask_andnot(new_cpus, possible_mask, subpartitions_cpus);
 		} else {
-- 
2.49.0

Re: [PATCH] cgroup/cpuset: Extend kthread_is_per_cpu() check to all PF_NO_SETAFFINITY tasks

Posted by Michal Koutný 7 months, 1 week ago

Hello.

On Wed, May 07, 2025 at 02:30:11PM -0400, Waiman Long <longman@redhat.com> wrote:
> Commit ec5fbdfb99d1 ("cgroup/cpuset: Enable update_tasks_cpumask()
> on top_cpuset") enabled us to pull CPUs dedicated to child partitions
> from tasks in top_cpuset by ignoring per cpu kthreads. However, there
> can be other kthreads that are not per cpu but have PF_NO_SETAFFINITY
> flag set to indicate that we shouldn't mess with their CPU affinity.
> 
> As all the per cpu kthreads have PF_NO_SETAFFINITY set, fix this
> issue by dropping the kthread_is_per_cpu() check and checking the
> PF_NO_SETAFFINITY flag instead.

I think there's a reason for existence of kthread_is_per_cpu(), so I'd
stick with more robust

-			if (kthread_is_per_cpu(task))
+			if (kthread_is_per_cpu(task) || task->flags & PF_NO_SETAFFINITY)

Perhaps with a comment that illuminates the kthread_is_per_cpu vs
PF_NO_SETAFFINITY difference.

0.02€,
Michal