PSI defaults to a FIFO-99 thread, reduce this to FIFO-1.
FIFO-99 is the very highest priority available to SCHED_FIFO and
it not a suitable default; it would indicate the psi work is the
most important work on the machine.
Since Real-Time tasks will have pre-allocated memory and locked it in
place, Real-Time tasks do not care about PSI. All it needs is to be
above OTHER.
Cc: Suren Baghdasaryan <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
---
kernel/sched/psi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/kernel/sched/psi.c
+++ b/kernel/sched/psi.c
@@ -1051,7 +1051,7 @@ struct psi_trigger *psi_trigger_create(s
if (!rcu_access_pointer(group->poll_kworker)) {
struct sched_param param = {
- .sched_priority = MAX_RT_PRIO - 1,
+ .sched_priority = 1,
};
struct kthread_worker *kworker;
On Thu, Aug 01, 2019 at 01:13:49PM +0200, Peter Zijlstra wrote:
> PSI defaults to a FIFO-99 thread, reduce this to FIFO-1.
>
> FIFO-99 is the very highest priority available to SCHED_FIFO and
> it not a suitable default; it would indicate the psi work is the
> most important work on the machine.
>
> Since Real-Time tasks will have pre-allocated memory and locked it in
> place, Real-Time tasks do not care about PSI. All it needs is to be
> above OTHER.
>
> Cc: Suren Baghdasaryan <[email protected]>
> Cc: Johannes Weiner <[email protected]>
> Cc: Thomas Gleixner <[email protected]>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Acked-by: Johannes Weiner <[email protected]>
Subject should be s/pci/psi/
On Thu, Aug 1, 2019 at 10:49 AM Johannes Weiner <[email protected]> wrote:
>
> On Thu, Aug 01, 2019 at 01:13:49PM +0200, Peter Zijlstra wrote:
> > PSI defaults to a FIFO-99 thread, reduce this to FIFO-1.
> >
> > FIFO-99 is the very highest priority available to SCHED_FIFO and
> > it not a suitable default; it would indicate the psi work is the
> > most important work on the machine.
> >
> > Since Real-Time tasks will have pre-allocated memory and locked it in
> > place, Real-Time tasks do not care about PSI. All it needs is to be
> > above OTHER.
> >
> > Cc: Suren Baghdasaryan <[email protected]>
> > Cc: Johannes Weiner <[email protected]>
> > Cc: Thomas Gleixner <[email protected]>
> > Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
>
> Acked-by: Johannes Weiner <[email protected]>
>
> Subject should be s/pci/psi/
Thanks for the patch. Please give me a couple hours for testing to
ensure no regressions before merging it.
On Thu, Aug 01, 2019 at 01:49:07PM -0400, Johannes Weiner wrote:
> On Thu, Aug 01, 2019 at 01:13:49PM +0200, Peter Zijlstra wrote:
> > PSI defaults to a FIFO-99 thread, reduce this to FIFO-1.
> >
> > FIFO-99 is the very highest priority available to SCHED_FIFO and
> > it not a suitable default; it would indicate the psi work is the
> > most important work on the machine.
> >
> > Since Real-Time tasks will have pre-allocated memory and locked it in
> > place, Real-Time tasks do not care about PSI. All it needs is to be
> > above OTHER.
> >
> > Cc: Suren Baghdasaryan <[email protected]>
> > Cc: Johannes Weiner <[email protected]>
> > Cc: Thomas Gleixner <[email protected]>
> > Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
>
> Acked-by: Johannes Weiner <[email protected]>
Thanks!
> Subject should be s/pci/psi/
Yeah, already fixed that.. obviously one sees such typoes only after
sending ;-)
On Thu, Aug 1, 2019 at 11:31 AM Suren Baghdasaryan <[email protected]> wrote:
>
> On Thu, Aug 1, 2019 at 10:49 AM Johannes Weiner <[email protected]> wrote:
> >
> > On Thu, Aug 01, 2019 at 01:13:49PM +0200, Peter Zijlstra wrote:
> > > PSI defaults to a FIFO-99 thread, reduce this to FIFO-1.
nit: above "PSI defaults" is more accurately "PSI polling defaults".
The core PSI thread that handles regular updates is a regular kthread.
> > >
> > > FIFO-99 is the very highest priority available to SCHED_FIFO and
> > > it not a suitable default; it would indicate the psi work is the
> > > most important work on the machine.
> > >
> > > Since Real-Time tasks will have pre-allocated memory and locked it in
> > > place, Real-Time tasks do not care about PSI. All it needs is to be
> > > above OTHER.
> > >
> > > Cc: Suren Baghdasaryan <[email protected]>
> > > Cc: Johannes Weiner <[email protected]>
> > > Cc: Thomas Gleixner <[email protected]>
> > > Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
> >
> > Acked-by: Johannes Weiner <[email protected]>
> >
> > Subject should be s/pci/psi/
>
> Thanks for the patch. Please give me a couple hours for testing to
> ensure no regressions before merging it.
Tested-by: Suren Baghdasaryan <[email protected]>