Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Peter Zijlstra
On Tue, Jul 04, 2017 at 11:57:12AM +0200, Vincent Guittot wrote: > On 4 July 2017 at 11:44, Peter Zijlstra wrote: > > But but but, how can that happen? Should it not all be under the same > > rq->lock and thus have only a single update_rq_clock() and thus be at > > the same 'instant' ? > > idle_

Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Vincent Guittot
On 4 July 2017 at 11:44, Peter Zijlstra wrote: > On Tue, Jul 04, 2017 at 11:12:34AM +0200, Vincent Guittot wrote: >> On 4 July 2017 at 10:34, Peter Zijlstra wrote: >> > On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote: >> >> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot

Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Peter Zijlstra
On Tue, Jul 04, 2017 at 11:12:34AM +0200, Vincent Guittot wrote: > On 4 July 2017 at 10:34, Peter Zijlstra wrote: > > On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote: > >> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote: > >> > The running state is a subset of runnab

Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Vincent Guittot
On 4 July 2017 at 10:34, Peter Zijlstra wrote: > On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote: >> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote: >> > The running state is a subset of runnable state which means that running >> > can't be set if runnable (weight)

Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Peter Zijlstra
On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote: > On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote: > > The running state is a subset of runnable state which means that running > > can't be set if runnable (weight) is cleared. There are corner cases > > where the curr

Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Vincent Guittot
On 4 July 2017 at 09:27, Peter Zijlstra wrote: > On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote: >> The running state is a subset of runnable state which means that running >> can't be set if runnable (weight) is cleared. There are corner cases >> where the current sched_entity ha

Re: [PATCH v2] sched/pelt: fix false running accounting

2017-07-04 Thread Peter Zijlstra
On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote: > The running state is a subset of runnable state which means that running > can't be set if runnable (weight) is cleared. There are corner cases > where the current sched_entity has been already dequeued but cfs_rq->curr > has not be