linux-kernel - Re: [RFC][PATCH 10/10] sched/eevdf: Use sched_attr::sched

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, 08 May 2024 15:52:32 +0200
From: Mike Galbraith <efault@....de>
To: Chen Yu <yu.c.chen@...el.com>, Peter Zijlstra <peterz@...radead.org>
Cc: mingo@...hat.com, juri.lelli@...hat.com, vincent.guittot@...aro.org, 
 dietmar.eggemann@....com, rostedt@...dmis.org, bsegall@...gle.com,
 mgorman@...e.de,  bristot@...hat.com, vschneid@...hat.com,
 linux-kernel@...r.kernel.org,  kprateek.nayak@....com,
 wuyun.abel@...edance.com, tglx@...utronix.de,  tim.c.chen@...el.com,
 yu.c.chen.y@...il.com
Subject: Re: [RFC][PATCH 10/10] sched/eevdf: Use sched_attr::sched_runtime
 to set request/slice suggestion

On Tue, 2024-05-07 at 23:15 +0800, Chen Yu wrote:
> On 2024-04-05 at 12:28:04 +0200, Peter Zijlstra wrote:
> > Allow applications to directly set a suggested request/slice length using
> > sched_attr::sched_runtime.
> >
> > The implementation clamps the value to: 0.1[ms] <= slice <= 100[ms]
> > which is 1/10 the size of HZ=1000 and 10 times the size of HZ=100.
> >
> > Applications should strive to use their periodic runtime at a high
> > confidence interval (95%+) as the target slice. Using a smaller slice
> > will introduce undue preemptions, while using a larger value will
> > increase latency.
> >
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
> >
>
> Is it possible to leverage this task slice to do better task wakeup placement?

Slice being unrelated to placement makes its use in a placement related
knob look wrong.  Even the smallest possible slice is orders of
magnitude larger than the cycle time of TCP_RR, making slice nearly
irrelevant to the issue being demonstrating via TCP_RR.  Even for that
huge socket box, it won't take long as cycle time increases toward that
smallest possible slice for the cost of needless wait to bury placement
decision costs.

> The idea is that, the smaller the slice the wakee has, the less idle CPU it
> should scan. This can reduce wake latency and inhibit costly task migration,
> especially on large systems.

Sure, this is an age old issue that's scaled up to size extra ugly in
that huge socket box.  Any solution needs to scale as well methinks, a
simple fixed yardstick won't work, as the costs being mitigated vary
wildly with platform size/shape.

	-Mike