When workqueue cpumask changes are committed the associated rescuer (if
one exists) affinity is not touched and this might be a problem down the
line for isolated setups.
Make sure rescuers affinity is updated every time a workqueue cpumask
changes, so that rescuers can't break isolation.
Signed-off-by: Juri Lelli <[email protected]>
---
kernel/workqueue.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 2ef6573909070..df7f2f2bfd0c8 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -4416,6 +4416,12 @@ static void apply_wqattrs_commit(struct apply_wqattrs_ctx *ctx)
link_pwq(ctx->dfl_pwq);
swap(ctx->wq->dfl_pwq, ctx->dfl_pwq);
+ /* rescuer needs to respect wq cpumask changes */
+ if (ctx->wq->rescuer) {
+ set_cpus_allowed_ptr(ctx->wq->rescuer->task, ctx->attrs->cpumask);
+ wake_up_process(ctx->wq->rescuer->task);
+ }
+
mutex_unlock(&ctx->wq->mutex);
}
--
2.43.0
Hello, Juri
On Wed, Jan 17, 2024 at 12:20 AM Juri Lelli <[email protected]> wrote:
> + /* rescuer needs to respect wq cpumask changes */
> + if (ctx->wq->rescuer) {
> + set_cpus_allowed_ptr(ctx->wq->rescuer->task, ctx->attrs->cpumask);
> + wake_up_process(ctx->wq->rescuer->task);
> + }
> +
What's the reason to wake up the rescuer?
I support this patch except for the wakeup:
Reviewed-by: Lai Jiangshan <[email protected]>
Thanks
Lai
Hello Lai,
On 17/01/24 11:56, Lai Jiangshan wrote:
> Hello, Juri
>
> On Wed, Jan 17, 2024 at 12:20 AM Juri Lelli <[email protected]> wrote:
>
> > + /* rescuer needs to respect wq cpumask changes */
> > + if (ctx->wq->rescuer) {
> > + set_cpus_allowed_ptr(ctx->wq->rescuer->task, ctx->attrs->cpumask);
> > + wake_up_process(ctx->wq->rescuer->task);
> > + }
> > +
>
> What's the reason to wake up the rescuer?
I believe we want to wake it up so that it can possibly be moved
"instantly" to a cpu inside its new cpumask affinity. If we don't wake
it up it might be sleeping on a cpu outside its affinity which might
have become isolated and this cpu could be affected by that wakeup if
that only happens later on when possibly the rescuer needs to perform
some work.
Does is make more sense to you?
> I support this patch except for the wakeup:
> Reviewed-by: Lai Jiangshan <[email protected]>
Thanks for looking at this!
Juri