linux-kernel - Re: [PATCH] sched/cputime: Make IRQ time accounting configurable at boot time

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] [day] [month] [year] [list]

Message-ID: <20230630150103.pp7lhxko7gya65rx@airbuntu>
Date:   Fri, 30 Jun 2023 16:01:03 +0100
From:   Qais Yousef <qyousef@...alina.io>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Bart Van Assche <bvanassche@....org>, linux-kernel@...r.kernel.org,
        Ingo Molnar <mingo@...hat.com>,
        Juri Lelli <juri.lelli@...hat.com>,
        Vincent Guittot <vincent.guittot@...aro.org>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Steven Rostedt <rostedt@...dmis.org>
Subject: Re: [PATCH] sched/cputime: Make IRQ time accounting configurable at
 boot time

On 06/16/23 09:45, Peter Zijlstra wrote:
> On Thu, Jun 15, 2023 at 01:37:26PM -0700, Bart Van Assche wrote:
> > Some producers of Android devices want IRQ time accounting enabled while
> > others want IRQ time accounting disabled. Hence, make IRQ time accounting
> > configurable at boot time.
> 
> Why would they want this disabled? IRQ time accounting avoids a number
> of issues under high irq/softirq pressure.
> 
> Disabling this makes no sense.

I think it is assumed that IRQ time accounting is only used for stat
collection (which is what I thought too), but based on this response I can see
it is used to ensure we account for stolen time in update_rq_clock_task() so it
helps to make it account more accurately for the time a task actually spent
running and doing useful work.

Bart, could you profile the cause of the high overhead? The config message says
a small perf impact, but 40% mentioned in your v2 is high. Could you try to
break down the problem further as we might be overlooking the true source of
overhead or miss an opportunity to improve the accounting logic instead?

Thanks

--
Qais Yousef