From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Tejun Heo <tj@kernel.org>, John Stultz <john.stultz@linaro.org>,
Ingo Molnar <mingo@redhat.com>,
lkml <linux-kernel@vger.kernel.org>,
Dmitry Shmidt <dimitrysh@google.com>,
Rom Lemarchand <romlem@google.com>,
Colin Cross <ccross@google.com>, Todd Kjos <tkjos@google.com>,
Oleg Nesterov <oleg@redhat.com>
Subject: Re: Severe performance regression w/ 4.4+ on Android due to cgroup locking changes
Date: Wed, 13 Jul 2016 14:03:15 -0700 [thread overview]
Message-ID: <20160713210315.GO7094@linux.vnet.ibm.com> (raw)
In-Reply-To: <20160713205102.GZ30909@twins.programming.kicks-ass.net>
On Wed, Jul 13, 2016 at 10:51:02PM +0200, Peter Zijlstra wrote:
> On Wed, Jul 13, 2016 at 04:39:44PM -0400, Tejun Heo wrote:
>
> > > There is a synchronize_sched() in there, so sorta. That thing is heavily
> > > geared towards readers, as is the only 'sane' choice for global locks.
> >
> > It used to use the expedited variant until 001dac627ff3
> > ("locking/percpu-rwsem: Make use of the rcu_sync infrastructure"), so
> > it might have been okay before then.
>
> Right, but expedited stuff sprays IPIs around the entire system. That's
> stuff other people complain about.
Do anyone other than the non-NO_HZ_FULL low-latency guys and the -rt
guys care?
> > The options that I can see are
> >
> > 1. Somehow make percpu_rwsem's write behavior more responsive in a way
> > which is acceptable all use cases. This would be great but
> > probably impossible.
> >
> > 2. Add a fast-writer option to percpu_rwsem so that users which care
> > about write latency can opt in for higher processing overhead for
> > lower latency.
>
> So, IIRC, the trade-off is a full memory barrier in read_lock and
> read_unlock() vs sync_sched() in write.
>
> Full memory barriers are expensive and while the combined cost might
> well exceed the cost of the sync_sched() it doesn't suffer the latency
> issues.
>
> Not sure if we can frob the two in a single codebase, but I can have a
> poke if Oleg or Paul doesn't beat me to it.
Take the patch that I just sent out and make the choice of normal
vs. expedited depend on CONFIG_PREEMPT_RT or whatever the -rt guys are
calling it these days. Is there a low-latency Kconfig option other
than CONFIG_NO_HZ_FULL?
The memory-barrier approach can definitely be made to work, but is
going to be more complex due to the need to wait for readers.
Thanx, Paul
next prev parent reply other threads:[~2016-07-13 21:03 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-07-13 0:00 Severe performance regression w/ 4.4+ on Android due to cgroup locking changes John Stultz
2016-07-13 8:21 ` Peter Zijlstra
2016-07-13 14:42 ` Paul E. McKenney
2016-07-13 18:13 ` Dmitry Shmidt
2016-07-13 18:32 ` Paul E. McKenney
2016-07-13 18:21 ` Tejun Heo
2016-07-13 18:33 ` Tejun Heo
2016-07-13 20:13 ` John Stultz
2016-07-13 20:18 ` Tejun Heo
2016-07-13 20:26 ` Peter Zijlstra
2016-07-13 20:39 ` Tejun Heo
2016-07-13 20:51 ` Peter Zijlstra
2016-07-13 21:01 ` Tejun Heo
2016-07-13 21:03 ` Paul E. McKenney [this message]
2016-07-13 21:05 ` Tejun Heo
2016-07-13 21:18 ` Paul E. McKenney
2016-07-13 21:42 ` Paul E. McKenney
2016-07-13 21:46 ` John Stultz
2016-07-13 22:17 ` Paul E. McKenney
2016-07-13 22:39 ` John Stultz
2016-07-13 23:02 ` Paul E. McKenney
2016-07-13 23:04 ` Paul E. McKenney
2016-07-14 11:35 ` Tejun Heo
2016-07-14 12:04 ` Peter Zijlstra
2016-07-14 12:08 ` Tejun Heo
2016-07-14 12:20 ` Peter Zijlstra
2016-07-14 15:07 ` Tejun Heo
2016-07-14 15:24 ` Tejun Heo
2016-07-14 16:32 ` Peter Zijlstra
2016-07-14 17:34 ` Oleg Nesterov
2016-07-14 16:54 ` John Stultz
2016-07-13 22:25 ` John Stultz
2016-07-13 22:01 ` Tejun Heo
2016-07-13 22:33 ` Paul E. McKenney
2016-07-14 6:49 ` Peter Zijlstra
2016-07-14 11:20 ` Tejun Heo
2016-07-14 12:11 ` Peter Zijlstra
2016-07-14 15:14 ` Tejun Heo
2016-07-14 13:18 ` Peter Zijlstra
2016-07-14 14:14 ` Peter Zijlstra
2016-07-14 14:58 ` Oleg Nesterov
2016-07-14 16:14 ` Peter Zijlstra
2016-07-14 16:37 ` Peter Zijlstra
2016-07-14 17:05 ` Oleg Nesterov
2016-07-14 16:23 ` Paul E. McKenney
2016-07-14 16:45 ` Peter Zijlstra
2016-07-14 17:15 ` Paul E. McKenney
2016-07-14 16:43 ` John Stultz
2016-07-14 16:49 ` Peter Zijlstra
2016-07-14 17:02 ` John Stultz
2016-07-14 17:13 ` Oleg Nesterov
2016-07-14 17:30 ` John Stultz
2016-07-14 17:41 ` Oleg Nesterov
2016-07-14 17:51 ` John Stultz
2016-07-14 18:09 ` Oleg Nesterov
2016-07-14 18:36 ` Peter Zijlstra
2016-07-14 19:35 ` Peter Zijlstra
2016-07-13 20:57 ` John Stultz
2016-07-13 20:52 ` Paul E. McKenney
2016-07-13 20:57 ` Peter Zijlstra
2016-07-13 21:08 ` Paul E. McKenney
2016-07-13 21:01 ` Dmitry Shmidt
2016-07-13 21:03 ` John Stultz
2016-07-13 21:05 ` Paul E. McKenney
2016-07-13 20:31 ` Dmitry Shmidt
2016-07-13 20:44 ` Colin Cross
2016-07-13 20:54 ` Tejun Heo
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=20160713210315.GO7094@linux.vnet.ibm.com \
--to=paulmck@linux.vnet.ibm.com \
--cc=ccross@google.com \
--cc=dimitrysh@google.com \
--cc=john.stultz@linaro.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=oleg@redhat.com \
--cc=peterz@infradead.org \
--cc=romlem@google.com \
--cc=tj@kernel.org \
--cc=tkjos@google.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).