From: Peter Zijlstra <peterz@infradead.org>
To: Rik van Riel <riel@redhat.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>,
Glauber Costa <glommer@redhat.com>,
kvm@vger.kernel.org, avi@redhat.com, zamsden@redhat.com,
mtosatti@redhat.com, mingo@elte.hu
Subject: Re: [RFC v2 4/7] change kernel accounting to include steal time
Date: Tue, 31 Aug 2010 10:11:49 +0200 [thread overview]
Message-ID: <1283242309.1820.1471.camel@laptop> (raw)
In-Reply-To: <4C7C38BC.1090907@redhat.com>
On Mon, 2010-08-30 at 19:03 -0400, Rik van Riel wrote:
>
> > I think it basically comes down to adding "sched_clock_unstolen()" which
> > the scheduler can use to measure time a process spends running, and
> > sched_clock() for measuring sleep times. In the normal case,
> > sched_clock_unstolen() would be the same as sched_clock().
>
> That requires the host to export (any time the guest is scheduled
> in), the amount of CPU time the VCPU thread has used, and the time
> the VCPU was scheduled in.
>
> Since the VCPU must be running when it is examining these variables,
> it can calculate the additional time (since it was last scheduled)
> to account to the task, and remember the currently calculated time
> in its own per-vcpu variable, so next time it can get a delta again.
I think its easier (and sufficient) for the host to tell the guest how
long it was _not_ running. That can simply be passed in when you start
the vcpu again and doesn't need a fancy communication channel.
The guests sched_clock() will measure wall time, the guests
sched_clock_stolen() will report the accumulation of these stolen times.
Then you can make sched_clock_unstolen() be sched_clock() -
sched_clock_stolen(). And like Jeremy said, if you make the sched_fair
stuff use sched_clock_unstolen() things should more or less work.
The problem with all that is that you'll start to schedule on unstolen
time instead of wall-time, which might not give the best results for
things like latencies etc.. but I guess that's one of the prices you pay
for using virt.
Also, like said yesterday, you need some factor in update_cpu_power(), a
quick hack might be to add all stolen time to sched_rt_avg_update().
next prev parent reply other threads:[~2010-08-31 8:12 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-08-30 16:06 [RFC v2 0/7] kvm stael time implementation Glauber Costa
2010-08-30 16:06 ` [RFC v2 1/7] change headers preparing for steal time Glauber Costa
2010-08-30 16:06 ` [RFC 1/8] Implement getnsboottime kernel API Glauber Costa
2010-08-30 16:06 ` [RFC v2 2/7] always call kvm_write_guest Glauber Costa
2010-08-30 16:06 ` [RFC 2/8] change headers preparing for steal time Glauber Costa
2010-08-30 16:06 ` [RFC 3/8] always call kvm_write_guest Glauber Costa
2010-08-30 16:06 ` [RFC v2 3/7] measure time out of guest Glauber Costa
2010-08-30 16:06 ` [RFC v2 4/7] change kernel accounting to include steal time Glauber Costa
2010-08-30 16:06 ` [RFC 4/8] measure time out of guest Glauber Costa
2010-08-30 16:06 ` [RFC 5/8] change kernel accounting to include steal time Glauber Costa
2010-08-30 16:06 ` [RFC v2 5/7] kvm steal time implementation Glauber Costa
2010-08-30 16:06 ` [RFC 6/8] " Glauber Costa
2010-08-30 16:06 ` [RFC v2 6/7] touch softlockup watchdog Glauber Costa
2010-08-30 16:06 ` [RFC v2 7/7] tell guest about steal time feature Glauber Costa
2010-08-30 16:06 ` [RFC 7/8] touch softlockup watchdog Glauber Costa
2010-08-30 16:06 ` [RFC 8/8] tell guest about steal time feature Glauber Costa
2010-08-30 17:33 ` [RFC v2 6/7] touch softlockup watchdog Jeremy Fitzhardinge
2010-08-30 18:07 ` Glauber Costa
2010-08-30 16:46 ` [RFC 5/8] change kernel accounting to include steal time Peter Zijlstra
2010-08-30 17:26 ` Glauber Costa
2010-08-30 17:30 ` [RFC v2 4/7] " Jeremy Fitzhardinge
2010-08-30 18:39 ` Rik van Riel
2010-08-30 19:07 ` Jeremy Fitzhardinge
2010-08-30 19:14 ` Peter Zijlstra
2010-08-30 19:17 ` Rik van Riel
2010-08-30 19:20 ` Peter Zijlstra
2010-08-30 19:45 ` Rik van Riel
2010-08-30 22:56 ` Jeremy Fitzhardinge
2010-08-30 23:03 ` Rik van Riel
2010-08-31 8:11 ` Peter Zijlstra [this message]
2010-09-02 18:19 ` Glauber Costa
2010-09-03 3:24 ` Jeremy Fitzhardinge
2010-09-03 7:18 ` Peter Zijlstra
2010-09-01 23:56 ` [RFC 1/8] Implement getnsboottime kernel API Zachary Amsden
2010-08-30 16:37 ` [RFC v2 0/7] kvm stael time implementation Peter Zijlstra
2010-08-30 16:45 ` Jeremy Fitzhardinge
2010-08-30 17:21 ` Glauber Costa
2010-08-30 17:20 ` Jeremy Fitzhardinge
2010-08-30 17:06 [RFC v2 0/7] kvm steal time implementation proposal Glauber Costa
2010-08-30 17:06 ` [RFC v2 4/7] change kernel accounting to include steal time 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=1283242309.1820.1471.camel@laptop \
--to=peterz@infradead.org \
--cc=avi@redhat.com \
--cc=glommer@redhat.com \
--cc=jeremy@goop.org \
--cc=kvm@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=mtosatti@redhat.com \
--cc=riel@redhat.com \
--cc=zamsden@redhat.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;
as well as URLs for NNTP newsgroup(s).