linux-kernel - Re: [RFC v2 3/7] Improve the tracking of active utilisation

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20160405190511.2b3029a9@utopia>
Date:	Tue, 5 Apr 2016 19:05:11 +0200
From:	luca abeni <luca.abeni@...tn.it>
To:	Peter Zijlstra <peterz@...radead.org>
Cc:	linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...hat.com>,
	Juri Lelli <juri.lelli@....com>
Subject: Re: [RFC v2 3/7] Improve the tracking of active utilisation

Hi Peter,

On Tue, 5 Apr 2016 14:42:42 +0200
Peter Zijlstra <peterz@...radead.org> wrote:

> On Fri, Apr 01, 2016 at 05:12:29PM +0200, Luca Abeni wrote:
> > @@ -526,7 +575,18 @@ static void update_dl_entity(struct sched_dl_entity *dl_se,
> >  	struct dl_rq *dl_rq = dl_rq_of_se(dl_se);
> >  	struct rq *rq = rq_of_dl_rq(dl_rq);
> >  
> > -	add_running_bw(dl_se, dl_rq);
> > +	/*
> > +	 * If the "inactive timer" is still active, stop it and leave
> > +	 * the active utilisation unchanged.
> > +	 * Otherwise, increase the active utilisation.
> > +	 * If the timer cannot be cancelled, inactive_task_timer() will
> > +	 * find the task state as TASK_RUNNING, and will do nothing, so
> > +	 * we are still safe.
> > +	 */
> > +	if (hrtimer_active(&dl_se->inactive_timer))
> > +		hrtimer_try_to_cancel(&dl_se->inactive_timer);
> 
> _try_, what happens if that fails?

I think inactive_task_timer() will run, but will see p->state == TASK_RUNNING
and will return immediately.

I think I have actually seen this happening during my tests, because adding
the "if (p->state == TASK_RUNNING)" in inactive_task_timer() fixed some issues
that I was seeing.



				Thanks,
					Luca