Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Fri, 13 Dec 2019 09:10:33 +0100
From:   Peter Zijlstra <peterz@infradead.org>
To:     "chengjian (D)" <cj.chengjian@huawei.com>
Cc:     mingo@kernel.org, linux-kernel@vger.kernel.org,
        chenwandun@huawei.com, xiexiuqi@huawei.com, liwei391@huawei.com,
        huawei.libin@huawei.com, bobo.shaobowang@huawei.com,
        juri.lelli@redhat.com, vincent.guittot@linaro.org
Subject: Re: [PATCH] sched/fair: Optimize select_idle_cpu
Message-ID: <20191213081033.GZ2844@hirez.programming.kicks-ass.net>
References: <20191212144102.181510-1-cj.chengjian@huawei.com>
 <20191212150429.GZ2827@hirez.programming.kicks-ass.net>
 <52112146-a4ee-d09f-b61e-9aa35e2e5298@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <52112146-a4ee-d09f-b61e-9aa35e2e5298@huawei.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Dec 13, 2019 at 09:51:30AM +0800, chengjian (D) wrote:
> Hi, Peter
> 
> ??? I saw the same work in select_idle_core, and I was wondering why the
> per_cpu variable was
> 
> needed for this yesterday. Now I think I probably understand : cpumask may
> be too large,
> 
> putting it on the stack may cause overflow. Is this correct ?

Yes, for instance when NR_CPUS=4096, struct cpumask ends up being 512
bytes, and that is _far_ too large for an on-stack variable, remember we
have relatively small fixed size stacks.