Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 18 Jul 2022 12:10:29 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Peng Zhang <zhangpeng.00@bytedance.com>
Cc:     mingo@redhat.com, vincent.guittot@linaro.org,
        dietmar.eggemann@arm.com, bsegall@google.com,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] sched/fair: optimize entity_tick()
Message-ID: <YtUxlVPGftwLWQNV@worktop.programming.kicks-ass.net>
References: <20220718094937.37360-1-zhangpeng.00@bytedance.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220718094937.37360-1-zhangpeng.00@bytedance.com>
Precedence: bulk

On Mon, Jul 18, 2022 at 05:49:37PM +0800, Peng Zhang wrote:
> From: "zhangpeng.00" <zhangpeng.00@bytedance.com>
> 
> entity_tick() will be called to check preemption by calculating each
> sched_entity's time_slice in a task group tree from bottom to top.
> 
> If a need_resched_flag is set by a checking of a sched_entity,
> there is no necessary to check preemption of the sched_entity's
> parent and ancestor entity.
> 
> So we can test if the need_resched_flag has been set before calling
> check_preempt_tick() to avoid unnecessary calculations.
> 
> Signed-off-by: zhangpeng.00 <zhangpeng.00@bytedance.com>

Optimization patches should come with numbers showing they're worth it.

Also; did you consider side-effects from check_preempt_tick(), like
clearing buddy state etc.. ?