mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Glauber Costa <glommer@parallels.com>
To: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: <linux-kernel@vger.kernel.org>, <xemul@parallels.com>,
	<paul@paulmenage.org>, <lizf@cn.fujitsu.com>,
	<daniel.lezcano@free.fr>, <jbottomley@parallels.com>
Subject: Re: [PATCH 6/9] Include idle and iowait fields in cpuacct
Date: Tue, 20 Sep 2011 10:29:01 -0300	[thread overview]
Message-ID: <4E78951D.1020901@parallels.com> (raw)
In-Reply-To: <1316523928.13664.12.camel@twins>

On 09/20/2011 10:05 AM, Peter Zijlstra wrote:
> On Tue, 2011-09-20 at 09:58 -0300, Glauber Costa wrote:
>> On 09/20/2011 09:58 AM, Peter Zijlstra wrote:
>>> On Tue, 2011-09-20 at 09:36 -0300, Glauber Costa wrote:
>>>> On 09/20/2011 06:21 AM, Peter Zijlstra wrote:
>>>>> On Wed, 2011-09-14 at 17:04 -0300, Glauber Costa wrote:
>>>>>> These are slightly different from the others though:
>>>>>> (note to reviewers: might be better to put those in a separate
>>>>>> array?)
>>>>>>
>>>>>> Since idle/iowait are a property of the system - by definition,
>>>>>> no process from any cgroup is running when the system is idle,
>>>>>> they are system wide. So what these fields really mean, are baselines
>>>>>> for when the cgroup was created. It allows the cgroup to start
>>>>>> counting idle/iowait from 0.
>>>>>
>>>>> Alternatively you can make iowait based on nr_uninterruptible per cgroup
>>>>> and count all ticks _this_ cgroup was idle.
>>>> You think?
>>>>
>>>> Humm,humm... maybe...
>>>> iowait can indeed be seen as a process group characteristic. I was
>>>> mainly concerned about overhead here, specially for the idle case:
>>>
>>> The overhead of accounting per cgroup nr_uninterruptible is the worst I
>>> think, that's in the sleep/wakeup paths.
>>>
>>>> If we are idle, there is no task context we can draw from, since the
>>>> task in the cpu is the idle task. So we end up having to touch all
>>>> cgroups... Or am I missing something?
>>>>
>>>> Sounds expensive.
>>>
>>> Count the total number of ticks on the cpu (I think we already have
>>> that) and subtract the number of ticks in this cgroup (I think we also
>>> already have that), which should yield: number of ticks not in this
>>> cgroup, aka number of ticks this cgroup was idle.
>> No , no... remember steal time.
>
> Of course I don't.. that's virt stuff, I repress that with all my might.
> But add or subtract steal ticks someplace and it doesn't come out right?

That's what I am here for...

But back to your answer:

 >>> Count the total number of ticks on the cpu (I think we already have
 >>> that) and subtract the number of ticks in this cgroup (I think we also
 >>> already have that),

Not sure if we have ticks in this cgroup... anyway, it can be done. We 
need a baseline for what was the tick situation when the cgroup started 
anyway.


  reply	other threads:[~2011-09-20 13:29 UTC|newest]

Thread overview: 58+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-09-14 20:04 [PATCH 0/9] Per-cgroup /proc/stat Glauber Costa
2011-09-14 20:04 ` [PATCH 1/9] Remove parent field in cpuacct cgroup Glauber Costa
2011-09-19 16:03   ` Peter Zijlstra
2011-09-19 16:09     ` Glauber Costa
2011-09-19 16:19       ` Peter Zijlstra
2011-09-19 16:30         ` Glauber Costa
2011-09-19 16:39           ` Peter Zijlstra
2011-09-19 16:41             ` Glauber Costa
2011-09-19 18:40               ` Peter Zijlstra
2011-09-20 17:29                 ` Srivatsa Vaddagiri
2011-09-22 15:11                   ` Balbir Singh
2011-09-22 15:17                     ` Peter Zijlstra
2011-09-23  8:09                       ` Balbir Singh
2011-09-23 14:35                         ` Peter Zijlstra
2011-09-23 15:45                         ` Glauber Costa
2011-09-19 18:35           ` Peter Zijlstra
2011-09-19 18:38             ` Glauber Costa
2011-09-19 18:44               ` Peter Zijlstra
2011-09-19 19:14                 ` Glauber Costa
2011-09-19 19:18                   ` Peter Zijlstra
2011-09-19 19:19                     ` Glauber Costa
2011-09-14 20:04 ` [PATCH 2/9] Make cpuacct fields per cpu variables Glauber Costa
2011-09-19 16:10   ` Peter Zijlstra
2011-09-14 20:04 ` [PATCH 3/9] Include nice values in cpuacct Glauber Costa
2011-09-19 16:19   ` Peter Zijlstra
2011-09-19 16:26     ` Glauber Costa
2011-09-19 18:36       ` Peter Zijlstra
2011-09-19 18:37         ` Glauber Costa
2011-09-14 20:04 ` [PATCH 4/9] Include irq and softirq fields " Glauber Costa
2011-09-19 18:38   ` Peter Zijlstra
2011-09-19 18:40     ` Glauber Costa
2011-09-14 20:04 ` [PATCH 5/9] Include guest " Glauber Costa
2011-09-14 20:04 ` [PATCH 6/9] Include idle and iowait " Glauber Costa
2011-09-20  9:21   ` Peter Zijlstra
2011-09-20 12:36     ` Glauber Costa
2011-09-20 12:58       ` Peter Zijlstra
2011-09-20 12:58         ` Glauber Costa
2011-09-20 13:05           ` Peter Zijlstra
2011-09-20 13:29             ` Glauber Costa [this message]
2011-09-14 20:04 ` [PATCH 7/9] Create cpuacct.proc.stat file Glauber Costa
2011-09-20  9:22   ` Peter Zijlstra
2011-09-20 12:22     ` Glauber Costa
2011-09-14 20:04 ` [PATCH 8/9] per-cgroup boot time Glauber Costa
2011-09-20  9:25   ` Peter Zijlstra
2011-09-20 12:37     ` Glauber Costa
2011-09-20 13:04       ` Peter Zijlstra
2011-09-20 13:06         ` Glauber Costa
2011-09-20 13:31           ` Peter Zijlstra
2011-09-14 20:04 ` [PATCH 9/9] Report steal time for cgroup Glauber Costa
2011-09-20  9:29   ` Peter Zijlstra
2011-09-14 20:13 ` [PATCH 0/9] Per-cgroup /proc/stat Peter Zijlstra
2011-09-14 20:20   ` Glauber Costa
2011-09-15  8:53     ` Peter Zijlstra
2011-09-14 20:23   ` Andi Kleen
2011-09-15  8:56     ` Peter Zijlstra
2011-09-19 23:07       ` Paul Turner
2011-09-20  8:33         ` Peter Zijlstra
2011-09-20 21:37         ` Glauber Costa

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4E78951D.1020901@parallels.com \
    --to=glommer@parallels.com \
    --cc=a.p.zijlstra@chello.nl \
    --cc=daniel.lezcano@free.fr \
    --cc=jbottomley@parallels.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lizf@cn.fujitsu.com \
    --cc=paul@paulmenage.org \
    --cc=xemul@parallels.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

Powered by JetHome