Date: Thu, 15 Oct 2009 07:41:41 -0400 From: John Baldwin <jhb@freebsd.org> To: freebsd-current@freebsd.org Cc: Taku YAMAMOTO <taku@tackymt.homeip.net>, Larry Rosenman <ler@lerctr.org> Subject: Re: Scheduler weirdness Message-ID: <200910150741.41569.jhb@freebsd.org> In-Reply-To: <20091012134159.8f6e4d66.taku@tackymt.homeip.net> References: <6729ad0409e449f8dbda69ecd8feb618.squirrel@webmail.lerctr.org> <20091012134159.8f6e4d66.taku@tackymt.homeip.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday 12 October 2009 12:41:59 am Taku YAMAMOTO wrote:
> On Sun, 11 Oct 2009 17:59:52 -0500
> "Larry Rosenman" <ler@lerctr.org> wrote:
>
> >
> > Ok, running RELENG_8 from Friday (10/9/2009).
> >
> > If I have 4 Folding-at-home processes running (they nice themselves) the
> > system is slow as a dog.
>
> Ah, I reminded a local patch regarding SCHED_ULE against niced threads.
> Something like this:
>
> --- sys/kern/sched_ule.c.orig 2009-04-29 12:26:30.000000000 +0900
> +++ sys/kern/sched_ule.c 2009-04-30 08:13:30.951440396 +0900
> @@ -1406,7 +1406,7 @@ sched_priority(struct thread *td)
> * score. Negative nice values make it easier for a thread to be
> * considered interactive.
> */
> - score = imax(0, sched_interact_score(td) - td->td_proc->p_nice);
> + score = imax(0, sched_interact_score(td) + td->td_proc->p_nice);
> if (score < sched_interact) {
> pri = PRI_MIN_REALTIME;
> pri += ((PRI_MAX_REALTIME - PRI_MIN_REALTIME) / sched_interact)
Thanks, I've committed this after Jeff reviewed it and will MFC it soon.
--
John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200910150741.41569.jhb>