linux-kernel - Re: [PATCH] sched: wakeup buddy

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]

Message-ID: <513FED85.8030603@linux.vnet.ibm.com>
Date:	Wed, 13 Mar 2013 11:07:49 +0800
From:	Michael Wang <wangyun@...ux.vnet.ibm.com>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
CC:	LKML <linux-kernel@...r.kernel.org>,
	Ingo Molnar <mingo@...nel.org>, Mike Galbraith <efault@....de>,
	Namhyung Kim <namhyung@...nel.org>,
	Alex Shi <alex.shi@...el.com>, Paul Turner <pjt@...gle.com>,
	Andrew Morton <akpm@...ux-foundation.org>,
	"Nikunj A. Dadhania" <nikunj@...ux.vnet.ibm.com>,
	Ram Pai <linuxram@...ibm.com>
Subject: Re: [PATCH] sched: wakeup buddy

On 03/12/2013 06:08 PM, Peter Zijlstra wrote:
> Does something simple like a per-task throttle of wake_affine() gain
> similar benefits? Say something like only do wake_affine() once every
> 10 ms or so (counting on the wakee, not waker).
> 
> The rationale being that wake_affine() relies on load-balance
> statistics that aren't updated that much faster, so calling it more
> often than that seems pointless.

That could works, if we do not have any logical to rely on and just want
to limit the rate of pull, this could be a good solution.

However, we already figure out the logical that wakeup related task
could benefit from closely running, this could promise us somewhat
reliable benefit.

It's just like, tell me how many times that two task continuously wakeup
each other means they related, I will try to pull them together since
they have a big chance to benefit from cache.

IMHO, that sounds a little easier for users than to make the decision on
tell me how long to pull tasks together, they may be confused...

In summary, I think we have two point here need to be considered:

1. what about the missed optimize timing, that may benefit
   some workload (although we haven't found such workload yet).

2. how many benefit could wake_affine() stuff bring to us,
   if limit rate benefit us, why don't we remove it?

Point 1 could be solved by disable wakeup buddy with 0 limitation, and
when users complain about their database performance, we just say "Calm
down and take a try on this knob ;-)".

Point 2 is about the wake_affine() stuff itself.

Later we could try to make the stuff better, but I have the feeling that
some key info is not there (may be we need Ingo's work atom idea here,
just wondering...), whatever, I think we still need a knob finally,
since it doesn't sounds like a general optimization which benefit all
the cases.

And I don't agree to remove the stuff since we have many theories that
this could benefit us, but before it really show the benefit in all the
cases, provide a way to keep it quiet sounds necessary...

Regards,
Michael Wang

> 
> Something like that should take a lot less lines to implement.
> 
> Just wondering..
> 
> --
> 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/
> 

--
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/