linux-kernel - Re: [PATCH] sched: Tweak default dynamic preempt mode selection

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:   Mon, 08 Nov 2021 15:21:56 +0000
From:   Valentin Schneider <valentin.schneider@....com>
To:     Mike Galbraith <efault@....de>, linux-kernel@...r.kernel.org
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...nel.org>,
        Frederic Weisbecker <frederic@...nel.org>
Subject: Re: [PATCH] sched: Tweak default dynamic preempt mode selection

On 08/11/21 13:27, Mike Galbraith wrote:
> On Mon, 2021-11-08 at 11:17 +0000, Valentin Schneider wrote:
>> On 06/11/21 05:40, Mike Galbraith wrote:
>> >
>> > Starting with a 5.15 config, to select RT you currently must first
>> > select a model you don't want, then reject PREEMPT_DYNAMIC and you'll
>> > be offered the full menu of models immediately. With your patch added,
>> > that became worse.  After rejecting PREEMPT_DYNAMIC, I had to go
>> > through new 5.15+ options before finally being offered the full menu.
>> >
>>
>> Do you mean at the syncconfig step?
>
> Um, not sure what that is, but it sounds about right.
>
>> I've only really played with upstream
>> arm64 / x86 defconfigs and didn't have to fight with any prompts, though
>> yes for x86 the default-y PREEMPT_DYNAMIC makes it a bit annoying to select
>> PREEMPT_RT.
>
> As long as RT depends on EXPERT it'll be a bit annoying regardless.  I
> just thought it worth mention that what you want now and what RT will
> presumably want upon merge completion appear to be mutually exclusive.
>

Hmm actually I think your approach should work, i.e. have

  config PREEMPT_DYNAMIC
    depends on [...] && !PREEMPT_RT

rather than

  config PREEMPT_RT
    depends on [...] && !PREEMPT_DYNAMIC

This essentially gives priority to the preemption model type over the
preemption model dynamicness, which I think makes sense. I can fold that in
v2.

>       -Mike