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:   Thomas Gleixner <tglx@linutronix.de>
To:     "Eric W. Biederman" <ebiederm@xmission.com>
Cc:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        linux-kernel@vger.kernel.org, Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        Oleg Nesterov <oleg@redhat.com>,
        Matt Fleming <matt@codeblueprint.co.uk>
Subject: Re: [PATCH] signal: Allow RT tasks to cache one sigqueue struct
In-Reply-To: <87a6r95067.fsf@nanos.tec.linutronix.de>
References: <20210303142025.wbbt2nnr6dtgwjfi@linutronix.de> <m1zgzj7uv2.fsf@fess.ebiederm.org> <20210304081142.digtkddajkadwwq5@linutronix.de> <87tupr55ea.fsf@nanos.tec.linutronix.de> <m11rcu7nbr.fsf@fess.ebiederm.org> <87lfb263h2.fsf@nanos.tec.linutronix.de> <87sg524z6t.fsf@nanos.tec.linutronix.de> <m1pn06oeno.fsf@fess.ebiederm.org> <87mtva4l6o.fsf@nanos.tec.linutronix.de> <87a6r95067.fsf@nanos.tec.linutronix.de>
Date:   Thu, 11 Mar 2021 15:20:00 +0100
Message-ID: <877dmd4vsf.fsf@nanos.tec.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On Thu, Mar 11 2021 at 13:45, Thomas Gleixner wrote:

> On Thu, Mar 11 2021 at 00:56, Thomas Gleixner wrote:
>> Rant aside, there is no massive benefit of doing that caching in
>> general, but there is not much of a downside either and for particular
>> use cases it's useful even outside of PREEMPT_RT.
>>
>> IMO, having it there unconditionally is better than yet another special
>> cased hackery.
>
> Just did some micro instrumentation to measure the time spent in
> __sigqueue_alloc/free() with and without the caching.
>
> Unsurprisingly that results in a time reduction of ~67% saving about 3us
> per alloc/free pair. Not hugely relevant for a kernel build but for
> anything which is signal heavy it'll make an difference.

That's all fastpath allocations and nothing which hit the slow path,
which would be way worse.

Thanks,

        tglx