linux-kernel - Re: [BUG] cpu controller can't provide fair CPU time for each group

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:	Wed, 11 Nov 2009 21:51:31 +0100
From:	Peter Zijlstra <peterz@...radead.org>
To:	Chris Friesen <cfriesen@...tel.com>
Cc:	Yasunori Goto <y-goto@...fujitsu.com>,
	Miao Xie <miaox@...fujitsu.com>,
	Linux-Kernel <linux-kernel@...r.kernel.org>,
	containers <containers@...ts.linux-foundation.org>,
	Ingo Molnar <mingo@...e.hu>
Subject: Re: [BUG] cpu controller can't provide fair CPU time for each group

On Wed, 2009-11-11 at 14:39 -0600, Chris Friesen wrote:
> On 11/11/2009 01:20 AM, Peter Zijlstra wrote:
> > On Wed, 2009-11-11 at 15:21 +0900, Yasunori Goto wrote:
> > 
> >> When users use cpuset/cpu affinity, then they would like to controll cpu affinity.
> >> Not CPU time.
> > 
> > What are people using affinity for? The only use of affinity is to
> > restrict or disable the load-balancer. Don't complain the load-balancer
> > doesn't work when you're taking active steps to hinder its work.
> 
> I have one active user of scheduler groups (using CKRM though, but they
> want to switch to a new kernel using CFS and sched groups in the near
> future).
> 
> They want to run their app on one cpu by itself with as little
> interference as possible.  Pure cpu processing, not even any I/O except
> via shared memory buffers.  Everything else gets done on the other cpu,
> but they want to control how much of the other cpu is assigned to packet
> processing, how much to system maintenance, normal user shell commands, etc.
> 
> This would seem like a case where some sort of cpuset/affinity  and
> sched groups would be expected to play nice together.

Agreed, and I'd like the load-balance partition feature of cpusets and
the cpu task-groups to work well together, except that the current
interface doesn't allow that nicely.



--
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/