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;
Message-ID: <7b2586c0-c380-c856-7d63-b7bff6f3a640@huawei.com>
Date:   Mon, 8 Aug 2022 20:23:46 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101
 Thunderbird/91.3.2
Subject: Re: [Question] Reading /proc/stat has a time backward issue
From:   "Lihua (lihua, ran)" <hucool.lihua@huawei.com>
To:     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>,
        "open list:SCHEDULER" <linux-kernel@vger.kernel.org>,
        <frederic@kernel.org>
References: <85d5087b-450c-351f-270d-c61303cf3187@huawei.com>
 <1f1f625a-148d-0398-f840-1f9b4e964189@huawei.com>
In-Reply-To: <1f1f625a-148d-0398-f840-1f9b4e964189@huawei.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

ping...

Your suggestions are valuable, I don't have a good way to fix this.

thanks all.

在 2022/8/4 15:44, Lihua (lihua, ran) 写道:
> ping...
> 
> Any good suggestions?
> 
> thanks all.
> 
> 在 2022/7/27 12:02, Lihua (lihua, ran) 写道:
>> Hi all,
>>
>> I found a problem that the statistical time goes backward, the value read first is 319, and the value read again is 318. As follows：
>> first：
>> cat /proc/stat |  grep cpu1
>> cpu1    319    0    496    41665    0    0    0    0    0    0
>> then：
>> cat /proc/stat |  grep cpu1
>> cpu1    318    0    497    41674    0    0    0    0    0    0
>>
>> Time goes back, which is counterintuitive.
>>
>> After debug this, I found that the problem is caused by the implementation of kcpustat_cpu_fetch_vtime. As follows：
>>
>>                                CPU0                                                                          CPU1
>> First:
>> show_stat():
>>      ->kcpustat_cpu_fetch()
>>          ->kcpustat_cpu_fetch_vtime()
>>              ->cpustat[CPUTIME_USER] = kcpustat_cpu(cpu) + vtime->utime + delta;              rq->curr is in user mod
>>               ---> When CPU1 rq->curr running on userspace, need add utime and delta
>>                                                                                               --->  rq->curr->vtime->utime is less than 1 tick
>> Then:
>> show_stat():
>>      ->kcpustat_cpu_fetch()
>>          ->kcpustat_cpu_fetch_vtime()
>>              ->cpustat[CPUTIME_USER] = kcpustat_cpu(cpu);                                     rq->curr is in kernel mod
>>              ---> When CPU1 rq->curr running on kernel space, just got kcpustat
>>
>> Because the values ​​of utime、 stime and delta are temporarily written to cpustat. Therefore, there are two problems  read from /proc/stat:
>> 1. There may be a regression phenomenon;
>> 2. When there are many tasks, the statistics are not accurate enough when utime and stime do not exceed one TICK.
>> The time goes back is counterintuitive, and I want to discuss whether there is a good solution without compromising performance.
>>
>> Thanks a lot.