From: Peter Zijlstra <peterz@infradead.org>
To: Giovanni Gherdovich <ggherdovich@suse.cz>
Cc: Ingo Molnar <mingo@redhat.com>,
Mike Galbraith <mgalbraith@suse.de>,
Stanislaw Gruszka <sgruszka@redhat.com>,
linux-kernel@vger.kernel.org, Mel Gorman <mgorman@suse.com>,
mgorman@techsingularity.net
Subject: Re: [PATCH] sched/cputime: Mitigate performance regression in times()/clock_gettime()
Date: Wed, 3 Aug 2016 12:02:54 +0200 [thread overview]
Message-ID: <20160803100254.GE6879@twins.programming.kicks-ass.net> (raw)
In-Reply-To: <1470175492.1849.3.camel@suse.cz>
On Wed, Aug 03, 2016 at 12:04:52AM +0200, Giovanni Gherdovich wrote:
> > > +#if defined(CONFIG_FAIR_GROUP_SCHED)
> >
> > This here wants a comment on why we're doing this. Because I'm sure
> > that if someone were to read this code in a few weeks they'd go
> > WTF!?
>
> I had that config variable set in the machine I was testing on, and
> thought that for some reason it was related to my observations. I will
> repeat the experiment without it, and if I obtain the same results I
> will drop the conditional. Otherwise I will motivate its necessity.
>
No, I meant we want a comment here explaining the reason for these
prefetches.
You'll need the #ifdef because se->cfs_rq doesn't exist otherwise.
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -2998,6 +2998,11 @@ unsigned long long task_sched_runtime(struct
> task_struct *p)
> * thread, breaking clock_gettime().
> */
> if (task_current(rq, p) && task_on_rq_queued(p)) {
> +#if defined(CONFIG_FAIR_GROUP_SCHED)
> + struct sched_entity *curr = (&p->se)->cfs_rq->curr;
> + prefetch(curr);
> + prefetch(&curr->exec_start);
> +#endif
> update_rq_clock(rq);
> p->sched_class->update_curr(rq);
> }
> -- -- >8 -- -- >8 -- -- >8 -- -- >8 -- -- >8 -- -- >8 -- -- >8
>
> I post below the snippets of generated code with and without CSE that
> I got running 'disassemble /m task_sched_runtime' in gdb; you'll see
> they're identical. If you prefer the explicit hint I'll include it in
> v2, but it's probably safe to say it isn't needed.
I much prefer the manual CSE, its much more readable.
Also, maybe pull the whole thing into a helper function with a
descriptive name, like:
/*
* XXX comment on why this is needed goes here...
*/
static inline void prefetch_curr_exec_start(struct task_struct *p)
{
#ifdef CONFIG_FAIR_GROUP_SCHED
struct sched_entity *curr = (&p->se)->cfs_rq->curr;
prefetch(curr);
prefetch(&curr->exec_start);
#endif
}
next prev parent reply other threads:[~2016-08-03 10:41 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-07-26 14:07 [PATCH] sched/cputime: Mitigate performance regression in times()/clock_gettime() Giovanni Gherdovich
2016-07-26 14:43 ` kbuild test robot
2016-07-27 11:41 ` Giovanni Gherdovich
2016-08-02 10:37 ` Peter Zijlstra
2016-08-02 13:26 ` Mike Galbraith
2016-08-02 22:04 ` Giovanni Gherdovich
2016-08-03 10:02 ` Peter Zijlstra [this message]
2016-08-03 10:34 ` Peter Zijlstra
2016-08-05 7:58 ` Giovanni Gherdovich
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=20160803100254.GE6879@twins.programming.kicks-ass.net \
--to=peterz@infradead.org \
--cc=ggherdovich@suse.cz \
--cc=linux-kernel@vger.kernel.org \
--cc=mgalbraith@suse.de \
--cc=mgorman@suse.com \
--cc=mgorman@techsingularity.net \
--cc=mingo@redhat.com \
--cc=sgruszka@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.