Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210422120459.447350175@infradead.org> <20210422123308.980003687@infradead.org>
 <CAEXW_YTHZF69YHD-r=ST97sagjnxEDy6492nDKaaJtkKMoQN9Q@mail.gmail.com>
 <3dbce4ff-44ed-73ca-2ea1-97b126dd664e@oracle.com> <CAEXW_YSX79vR9hdkjKcf08uZk85xhC8nOnesf8s6Cvp2kqKoFA@mail.gmail.com>
 <2b4ae2b4-62e5-96be-ddae-b261139842c1@oracle.com> <YJuadFpNsfHu6n0Y@hirez.programming.kicks-ass.net>
In-Reply-To: <YJuadFpNsfHu6n0Y@hirez.programming.kicks-ass.net>
From:   Josh Don <joshdon@google.com>
Date:   Wed, 12 May 2021 13:20:02 -0700
Message-ID: <CABk29NsMtsMvvVLbq7fGR0EMLZ9soFKGnE-SeectjWLym6YGcg@mail.gmail.com>
Subject: Re: [PATCH 17/19] sched: Inherit task cookie on fork()
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     Chris Hyser <chris.hyser@oracle.com>,
        Joel Fernandes <joel@joelfernandes.org>,
        Ingo Molnar <mingo@kernel.org>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Valentin Schneider <valentin.schneider@arm.com>,
        Mel Gorman <mgorman@suse.de>,
        LKML <linux-kernel@vger.kernel.org>,
        Thomas Glexiner <tglx@linutronix.de>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, May 12, 2021 at 2:05 AM Peter Zijlstra <peterz@infradead.org> wrote:
> Right, I need a Champion that actually cares about cgroups and has
> use-cases to go argue with TJ on this. I've proposed code that I think
> has sane semantics, but I'm not in a position to argue for it, given I
> think a world without cgroups is a better world :-)))

Not sure if Tejun has any thoughts on
http://lkml.kernel.org/r/CABk29NtahuW6UERvRdK5v8My_MfPsoESDKXUjGdvaQcHOJEMvg@mail.gmail.com.

We're looking at using the prctl interface with one of our main
internal users of core scheduling. As an example, suppose we have a
management process that wants to make tasks A and B share a cookie:
- Spawn a new thread m, which then does the following, and exits.
- PR_SCHED_CORE_CREATE for just its own PID
- PR_SCHED_CORE_SHARE_TO A
- PR_SCHED_CORE_SHARE_TO B

That seems to work ok; I'll follow up if there are any pain points
that aren't easily addressed with the prctl interface.