linux-kernel - Re: [PATCH 03/11] sched: Extend scheduler's asym packing

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:   Thu, 25 Aug 2016 14:18:37 +0100
From:   Morten Rasmussen <morten.rasmussen@....com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Srinivas Pandruvada <srinivas.pandruvada@...ux.intel.com>,
        mingo@...hat.com, tglx@...utronix.de, hpa@...or.com,
        rjw@...ysocki.net, x86@...nel.org, bp@...e.de,
        sudeep.holla@....com, ak@...ux.intel.com,
        linux-acpi@...r.kernel.org, linux-pm@...r.kernel.org,
        alexey.klimov@....com, viresh.kumar@...aro.org,
        akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
        lenb@...nel.org, tim.c.chen@...ux.intel.com,
        paul.gortmaker@...driver.com, jpoimboe@...hat.com,
        mcgrof@...nel.org, jgross@...e.com, robert.moore@...el.com,
        dvyukov@...gle.com, jeyu@...hat.com
Subject: Re: [PATCH 03/11] sched: Extend scheduler's asym packing

On Thu, Aug 25, 2016 at 01:45:22PM +0200, Peter Zijlstra wrote:
> On Thu, Aug 25, 2016 at 12:22:52PM +0100, Morten Rasmussen wrote:
> > I haven't reviewed the entire patch set in detail, but why can't the cpu
> > priority list be handed to the scheduler instead of moving scheduling
> > decisions out of the scheduler?
> 
> It basically does that. All that we allow here is the architecture to
> override the default order of what is considered priority.
> 
> The default (as per Power7) is naked cpu number, with lower cpu numbers
> having higher priority to higher numbers.
> 
> This patch set allows the architecture to provide a less_than operator
> (and through that a custom order).

But why not just pass the customized list into the scheduler? Seems
simpler?

A custom less_than operator opens up for potential exploitation that is
much more complicated than what is shown in this patch set. Paired up
with the utilization signals, which are now exposed outside the
scheduler through cpufreq, you can start making complex load-balancing
decisions outside the scheduler using the this new interface.