linux-kernel - Re: [Question] report a race condition between CPU hotplug state machine and hrtimer 'sched_cfs_period

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, 28 Jun 2023 23:09:56 +0200
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Vincent Guittot <vincent.guittot@...aro.org>,
        Xiongfeng Wang <wangxiongfeng2@...wei.com>
Cc:     vschneid@...hat.com, Phil Auld <pauld@...hat.com>,
        vdonnefort@...gle.com,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Wei Li <liwei391@...wei.com>,
        "liaoyu (E)" <liaoyu15@...wei.com>, zhangqiao22@...wei.com,
        Peter Zijlstra <peterz@...radead.org>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Ingo Molnar <mingo@...nel.org>
Subject: Re: [Question] report a race condition between CPU hotplug state
 machine and hrtimer 'sched_cfs_period_timer' for cfs bandwidth throttling

On Wed, Jun 28 2023 at 15:30, Vincent Guittot wrote:
> On Tue, 27 Jun 2023 at 18:46, Vincent Guittot
>> > > +               struct hrtimer_clock_base *clock_base = cfs_b->period_timer.base;
>> > > +               int cpu = clock_base->cpu_base->cpu;
>> > > +               if (!cpu_active(cpu) && cpu != smp_processor_id())
>> > > +                       hrtimer_start_expires(&cfs_b->period_timer,
>> > > HRTIMER_MODE_ABS_PINNED);
>> > >                 return;
>> > > +       }

Can you please trim your replies?

>> I have been able to reproduce your problem and run your fix on top. I
>> still wonder if there is a
>
> Looks like I have been preempted and never finished the sentence. The
> full sentence is:
> I still wonder if there is a race condition where the hang can still
> happen but i haven't been able to find one so far

As I explained before. Assume the timer fires on the outgoing CPU and
the other CPU tries to rearm it concurrently. It will stay on the
outgoing CPU and not move over.

Thanks,

        tglx