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;
From:   Valentin Schneider <valentin.schneider@arm.com>
To:     Frederic Weisbecker <frederic@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@kernel.org>
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Frederic Weisbecker <frederic@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        "Paul E . McKenney" <paulmck@kernel.org>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        Yejune Deng <yejune.deng@gmail.com>
Subject: Re: [PATCH] sched: Fix PF_NO_SETAFFINITY blind inheritance
In-Reply-To: <20210525235849.441842-1-frederic@kernel.org>
References: <20210525235849.441842-1-frederic@kernel.org>
Date:   Wed, 26 May 2021 11:10:07 +0100
Message-ID: <87a6ohhklc.mognet@arm.com>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk


Hi Frederic,

Sorry about that one; thanks for having dug into it.

On 26/05/21 01:58, Frederic Weisbecker wrote:
> One way to solve the PF_NO_SETAFFINITY issue is to not inherit this flag
> on copy_process() at all. The cases where it matters are:
>
> * fork_idle(): explicitly set the flag already.
> * fork() syscalls: userspace tasks that shouldn't be concerned by that.
> * create_io_thread(): the callers explicitly attribute the flag to the
>                       newly created tasks.
> * kernel_thread():
>       _ Fix the issues on init/1 and kthreadd
>       _ Fix the issues on kthreadd children.
>       _ Usermode helper created by an unbound workqueue. This shouldn't
>         matter. In the worst case it gives more control to userspace
>         on setting affinity to these short living tasks although this can
>         be tuned with inherited unbound workqueues affinity already.
>

(I just saw it got shoved into tip already, but in any case:)

That makes sense to me. Regarding the UMH point, I don't believe there are
others like it creeping around; otherwise we might've had to go with e.g.

  p->flags &= ~(... | (PF_NO_SETAFFINITY * !!(p->flags & PF_IDLE)))

but per the above that doesn't seem necessary.

Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>