Re: [RFC][PATCH 12/13] sched/deadline: Introduce deadline servers

[Juri Lelli <juri.lelli@redhat.com>]

On 08/08/19 11:27, Juri Lelli wrote:
> On 08/08/19 10:57, Dietmar Eggemann wrote:
> > On 8/8/19 10:46 AM, Juri Lelli wrote:
> > > On 08/08/19 10:11, Dietmar Eggemann wrote:
> > >> On 8/8/19 9:56 AM, Peter Zijlstra wrote:
> > >>> On Wed, Aug 07, 2019 at 06:31:59PM +0200, Dietmar Eggemann wrote:
> > >>>> On 7/26/19 4:54 PM, Peter Zijlstra wrote:
> > >>>>>
> > >>>>>
> > >>>>> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> > >>>>
> > >>>> [...]
> > >>>>
> > >>>>> @@ -889,6 +891,8 @@ static void update_curr(struct cfs_rq *c
> > >>>>>  		trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime);
> > >>>>>  		cgroup_account_cputime(curtask, delta_exec);
> > >>>>>  		account_group_exec_runtime(curtask, delta_exec);
> > >>>>> +		if (curtask->server)
> > >>>>> +			dl_server_update(curtask->server, delta_exec);
> > >>>>>  	}
> > >>>>
> > >>>> I get a lockdep_assert_held(&rq->lock) related warning in start_dl_timer()
> > >>>> when running the full stack.
> > >>>
> > >>> That would seem to imply a stale curtask->server value; the hunk below:
> > >>>
> > >>> --- a/kernel/sched/core.c
> > >>> +++ b/kernel/sched/core.c
> > >>> @@ -3756,8 +3756,11 @@ pick_next_task(struct rq *rq, struct tas
> > >>>
> > >>>         for_each_class(class) {
> > >>>                 p = class->pick_next_task(rq, NULL, NULL);
> > >>> -               if (p)
> > >>> +               if (p) {
> > >>> +                       if (p->sched_class == class && p->server)
> > >>> +                               p->server = NULL;
> > >>>                         return p;
> > >>> +               }
> > >>>         }
> > >>>
> > >>>
> > >>> Was supposed to clear p->server, but clearly something is going 'funny'.
> > >>
> > >> What about the fast path in pick_next_task()?
> > >>
> > >> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > >> index bffe849b5a42..f1ea6ae16052 100644
> > >> --- a/kernel/sched/core.c
> > >> +++ b/kernel/sched/core.c
> > >> @@ -3742,6 +3742,9 @@ pick_next_task(struct rq *rq, struct task_struct *prev, struct rq_flags *rf)
> > >>                 if (unlikely(!p))
> > >>                         p = idle_sched_class.pick_next_task(rq, prev, rf);
> > >>  
> > >> +               if (p->sched_class == &fair_sched_class && p->server)
> > >> +                       p->server = NULL;
> > >> +
> > > 
> > > Hummm, but then who sets it back to the correct server. AFAIU
> > > update_curr() needs a ->server to do the correct DL accounting?
> > 
> > Ah, OK, this would kill the whole functionality ;-)
> > 
> 
> I'm thinking we could use &rq->fair_server. It seems to pass the point
> we are discussing about, but then virt box becomes unresponsive (busy
> loops).

I'd like to take this last sentence back, I was able to run a few boot +
hackbench + shutdown cycles with the following applied (guess too much
debug printks around before).

--->8---
 kernel/sched/deadline.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 3fe82b8f7825..d4a20072d4c0 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1312,6 +1312,10 @@ void dl_server_update(struct sched_dl_entity *dl_se, s64 delta_exec)
 
 void dl_server_start(struct sched_dl_entity *dl_se)
 {
+	if (!dl_server(dl_se)) {
+		dl_se->dl_server = 1;
+		setup_new_dl_entity(dl_se);
+	}
 	enqueue_dl_entity(dl_se, dl_se, ENQUEUE_WAKEUP);
 }
 
@@ -1324,12 +1328,9 @@ void dl_server_init(struct sched_dl_entity *dl_se, struct rq *rq,
 		    dl_server_has_tasks_f has_tasks,
 		    dl_server_pick_f pick)
 {
-	dl_se->dl_server = 1;
 	dl_se->rq = rq;
 	dl_se->server_has_tasks = has_tasks;
 	dl_se->server_pick = pick;
-
-	setup_new_dl_entity(dl_se);
 }
 
 /*
@@ -2829,6 +2830,7 @@ static void __dl_clear_params(struct sched_dl_entity *dl_se)
 	dl_se->dl_yielded		= 0;
 	dl_se->dl_non_contending	= 0;
 	dl_se->dl_overrun		= 0;
+	dl_se->dl_server		= 0;
 }
 
 void init_dl_entity(struct sched_dl_entity *dl_se)

--->8---
 kernel/sched/core.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 7a161472decd..355cb1382aef 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3783,6 +3783,9 @@ pick_next_task(struct rq *rq, struct task_struct *prev, struct rq_flags *rf)
 		if (unlikely(!p))
 			p = idle_sched_class.pick_next_task(rq, prev, rf);
 
+		if (p->sched_class == &fair_sched_class)
+			p->server = &rq->fair_server;
+
 		return p;
 	}