Re: [PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-12 Thread Dietmar Eggemann
On 12/10/2020 15:18, Peter Zijlstra wrote: > On Mon, Oct 12, 2020 at 02:52:00PM +0200, Peter Zijlstra wrote: >> On Fri, Oct 09, 2020 at 10:41:11PM +0200, Dietmar Eggemann wrote: >>> On 05/10/2020 16:57, Peter Zijlstra wrote: Since we now migrate tasks away before DYING, we should also move >>>

Re: [PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-12 Thread Peter Zijlstra
On Mon, Oct 12, 2020 at 02:52:00PM +0200, Peter Zijlstra wrote: > On Fri, Oct 09, 2020 at 10:41:11PM +0200, Dietmar Eggemann wrote: > > On 05/10/2020 16:57, Peter Zijlstra wrote: > > > Since we now migrate tasks away before DYING, we should also move > > > bandwidth unthrottle, otherwise we can gai

Re: [PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-12 Thread Peter Zijlstra
On Fri, Oct 09, 2020 at 10:41:11PM +0200, Dietmar Eggemann wrote: > On 05/10/2020 16:57, Peter Zijlstra wrote: > > Since we now migrate tasks away before DYING, we should also move > > bandwidth unthrottle, otherwise we can gain tasks from unthrottle > > after we expect all tasks to be gone already

Re: [PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-09 Thread Dietmar Eggemann
On 05/10/2020 16:57, Peter Zijlstra wrote: > Since we now migrate tasks away before DYING, we should also move > bandwidth unthrottle, otherwise we can gain tasks from unthrottle > after we expect all tasks to be gone already. > > Also; it looks like the RT balancers don't respect cpu_active() and

Re: [PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-06 Thread Peter Zijlstra
On Tue, Oct 06, 2020 at 02:46:28PM +0200, Vincent Guittot wrote: > > @@ -6993,6 +6995,14 @@ int sched_cpu_deactivate(unsigned int cp > > > > balance_push_set(cpu, true); > > > > + rq_lock_irqsave(rq, &rf); > > + if (rq->rd) { > > + update_rq_clock(); > > Tried to

Re: [PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-06 Thread Vincent Guittot
On Mon, 5 Oct 2020 at 17:09, Peter Zijlstra wrote: > > Since we now migrate tasks away before DYING, we should also move > bandwidth unthrottle, otherwise we can gain tasks from unthrottle > after we expect all tasks to be gone already. > > Also; it looks like the RT balancers don't respect cpu_ac

[PATCH -v2 07/17] sched: Fix hotplug vs CPU bandwidth control

2020-10-05 Thread Peter Zijlstra
Since we now migrate tasks away before DYING, we should also move bandwidth unthrottle, otherwise we can gain tasks from unthrottle after we expect all tasks to be gone already. Also; it looks like the RT balancers don't respect cpu_active() and instead rely on rq->online in part, complete this. T