linux-kernel - Re: [patch 6/8] pull RT tasks

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:	Fri, 19 Oct 2007 15:43:50 -0400 (EDT)
From:	Steven Rostedt <rostedt@...dmis.org>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
cc:	LKML <linux-kernel@...r.kernel.org>,
	RT <linux-rt-users@...r.kernel.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Ingo Molnar <mingo@...e.hu>,
	Thomas Gleixner <tglx@...utronix.de>,
	Gregory Haskins <ghaskins@...ell.com>,
	Paul Jackson <pj@....com>
Subject: Re: [patch 6/8] pull RT tasks


--
On Fri, 19 Oct 2007, Peter Zijlstra wrote:

> > > +
> > > +	if (likely(!atomic_read(&rt_overload)))
> > > +		return 0;
> >
> > This seems to be the only usage of rt_overload. I'm not sure its worth
> > keeping it around for this.
>
> Ingo just brought up a good point. With large smp (where large is >64)
> this will all suck chunks.
>
> rt_overload will bounce around the system, and the rto_cpumask updates
> might already hurt.
>
> The idea would be to do this per cpuset, these naturally limit the
> migraiton posibilities of tasks and would thus be the natural locality
> to break this data structure.
>

That sounds like a good idea. RT balancing on >64 CPUs should be limited.
Having a bounding cpuset would help.

I'll try to come up with something.

Thanks!

-- Steve

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/