linux-kernel - Re: [PATCH] cpufreq: schedutil: Don't use the limits

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: <20210215063008.hsdkrs4bw7wt3wye@vireshk-i7>
Date:   Mon, 15 Feb 2021 12:00:08 +0530
From:   Viresh Kumar <viresh.kumar@...aro.org>
To:     Yue Hu <zbestahu@....com>
Cc:     "Rafael J. Wysocki" <rafael@...nel.org>,
        Yue Hu <zbestahu@...il.com>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Rafael Wysocki <rafael.j.wysocki@...el.com>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Vincent Guittot <vincent.guittot@...aro.org>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Benjamin Segall <bsegall@...gle.com>,
        Mel Gorman <mgorman@...e.de>,
        Daniel Bristot de Oliveira <bristot@...hat.com>,
        Linux PM <linux-pm@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Yue Hu <huyue2@...ong.com>, zhangwen@...ong.com
Subject: Re: [PATCH] cpufreq: schedutil: Don't use the limits_changed flag
 any more

On 14-02-21, 11:44, Yue Hu wrote:
> On Fri, 12 Feb 2021 17:14:03 +0100
> "Rafael J. Wysocki" <rafael@...nel.org> wrote:
> > This may be running in parallel with sugov_update_next_freq() on a
> > different CPU, so the latter may clear need_freq_update right after it
> > has been set here unless I'm overlooking something.
> 
> Whether this logic is also happening for limits_changed in
> sugo_should_update_freq() or not?

It is but it shouldn't have any side effects as we calculate the next
frequency after cleaning the limits_changed flag. Your patch would
have been fine, but it is not anymore because of commit 23a881852f3e
("cpufreq: schedutil: Don't skip freq update if need_freq_update is
set").

It made a considerable change after which your patch adds a bug. With
23a881852f3e, need_freq_update is updated/cleared after the next
frequency is calculated, while earlier it was cleared before it. And
so even with the race condition taking place, there were no issues.

-- 
viresh