From: "Jan Beulich" <JBeulich@novell.com>
To: "Ingo Molnar" <mingo@elte.hu>, "Jeremy Fitzhardinge" <jeremy@goop.org>
Cc: "Jun Nakajima" <jun.nakajima@intel.com>,
"Xiaohui Xin" <xiaohui.xin@intel.com>,
"Xin Li" <xin.li@intel.com>,
"Xen-devel" <xen-devel@lists.xensource.com>,
"Nick Piggin" <npiggin@suse.de>,
"Linux Kernel Mailing List" <linux-kernel@vger.kernel.org>,
"H. Peter Anvin" <hpa@zytor.com>
Subject: Re: [Xen-devel] Performance overhead of paravirt_ops on nativeidentified
Date: Thu, 14 May 2009 09:05:56 +0100 [thread overview]
Message-ID: <4A0BED040200007800000DB0@vpn.id2.novell.com> (raw)
In-Reply-To: <4A0B62F7.5030802@goop.org>
>>> Jeremy Fitzhardinge <jeremy@goop.org> 14.05.09 02:16 >>>
>One possibility is to inline _spin_lock, etc, when building an
>optimised kernel (ie, when there's no spinlock/preempt
>instrumentation/debugging enabled). That will remove the outer
>call/return pair, returning the instruction stream to a single
>call/return, which will presumably execute the same as the non-pvops
>case. The downsides arel 1) it will replicate the
>preempt_disable/enable code at eack lock/unlock callsite; this code is
>fairly small, but not nothing; and 2) the spinlock definitions are
>already a very heavily tangled mass of #ifdefs and other preprocessor
>magic, and making any changes will be non-trivial.
>
>The other obvious answer is to disable pv-spinlocks. Making them a
>separate config option is fairly easy, and it would be trivial to
>enable them only when Xen is enabled (as the only non-default user).
>But it doesn't really address the common case of a distro build which
>is going to have Xen support enabled, and leaves the open question of
>whether the native performance cost of pv-spinlocks is worth the
>performance improvement on a loaded Xen system (10% saving of overall
>system CPU when guests block rather than spin). Still it is a
>reasonable short-term workaround.
Wouldn't a third solution be to use ticket spinlocks everywhere, i.e. eliminate
the current indirection, and replace it by an indirection for just the contention
case? As I view it, the problem for Xen aren't really the ticket locks by
themselves, but rather the extra spinning involved, which is of concern only
if a lock is contended. We're using ticket locks quite happily in our kernels,
with directed instead of global wakeup from the unlock path. The only open
issue we currently have is that while for native keeping interrupts disabled
while spinning may be acceptable (though I'm not sure how -rt folks are
viewing this), in a pv environment one should really re-enable interrupts
here due to the potentially much higher latency.
Jan
next prev parent reply other threads:[~2009-05-14 8:06 UTC|newest]
Thread overview: 79+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-05-14 0:16 Performance overhead of paravirt_ops on native identified Jeremy Fitzhardinge
2009-05-14 1:10 ` H. Peter Anvin
2009-05-14 8:25 ` Peter Zijlstra
2009-05-14 14:05 ` H. Peter Anvin
2009-05-14 17:36 ` Jeremy Fitzhardinge
2009-05-14 17:50 ` H. Peter Anvin
2009-05-14 8:05 ` Jan Beulich [this message]
2009-05-14 8:33 ` [Xen-devel] Performance overhead of paravirt_ops on nativeidentified Peter Zijlstra
2009-05-14 17:45 ` Jeremy Fitzhardinge
2009-05-15 8:10 ` Jan Beulich
2009-05-15 18:50 ` Jeremy Fitzhardinge
2009-05-18 7:19 ` Jan Beulich
2009-05-20 22:42 ` Jeremy Fitzhardinge
2009-05-15 18:18 ` [tip:x86/urgent] x86: Fix performance regression caused by paravirt_ops on native kernels tip-bot for Jeremy Fitzhardinge
2009-05-21 22:42 ` Performance overhead of paravirt_ops on native identified Chuck Ebbert
2009-05-21 22:48 ` Jeremy Fitzhardinge
2009-05-21 23:10 ` H. Peter Anvin
2009-05-22 1:26 ` Xin, Xiaohui
2009-05-22 3:39 ` H. Peter Anvin
2009-05-22 4:27 ` Jeremy Fitzhardinge
2009-05-22 5:59 ` Xin, Xiaohui
2009-05-22 16:33 ` H. Peter Anvin
2009-05-22 22:44 ` Jeremy Fitzhardinge
2009-05-22 22:47 ` H. Peter Anvin
2009-05-25 9:15 ` [benchmark] 1% performance overhead of paravirt_ops on native kernels Ingo Molnar
2009-05-26 18:42 ` Jeremy Fitzhardinge
2009-05-28 6:17 ` Nick Piggin
2009-05-28 20:57 ` Jeremy Fitzhardinge
2009-05-30 10:23 ` Ingo Molnar
2009-06-02 14:18 ` Chris Mason
2009-06-02 14:49 ` Ulrich Drepper
2009-06-02 15:03 ` Chris Mason
2009-06-02 15:22 ` Ulrich Drepper
2009-06-02 16:20 ` Chris Mason
2009-06-02 18:13 ` Pekka Enberg
2009-06-02 18:06 ` Pekka Enberg
2009-06-02 18:27 ` Chris Mason
2009-06-03 6:33 ` Jeremy Fitzhardinge
2009-06-02 19:14 ` Thomas Gleixner
2009-06-02 19:51 ` Chris Mason
2009-06-03 12:38 ` Rusty Russell
2009-06-03 16:09 ` Linus Torvalds
[not found] ` <200906041554.37102.rusty@rustcorp.com.au>
2009-06-04 15:02 ` Linus Torvalds
2009-06-04 21:52 ` Dave McCracken
2009-06-05 7:31 ` Gerd Hoffmann
2009-06-05 14:31 ` Rusty Russell
2009-06-06 18:54 ` Anders K. Pedersen
2009-06-05 4:46 ` Rusty Russell
2009-06-05 14:54 ` Linus Torvalds
2009-06-07 0:53 ` Rusty Russell
2009-06-08 14:53 ` Linus Torvalds
2009-06-09 9:39 ` Nick Piggin
2009-06-09 11:17 ` Ingo Molnar
2009-06-09 12:10 ` Nick Piggin
2009-06-09 12:25 ` Ingo Molnar
2009-06-09 12:42 ` Nick Piggin
2009-06-09 12:56 ` Avi Kivity
2009-06-09 15:18 ` Linus Torvalds
2009-06-09 23:33 ` Paul Mackerras
2009-06-10 1:26 ` Ingo Molnar
2009-06-09 15:07 ` Linus Torvalds
2009-06-09 15:09 ` H. Peter Anvin
2009-06-09 18:06 ` Linus Torvalds
2009-06-09 18:07 ` Linus Torvalds
2009-06-09 22:48 ` Matthew Garrett
2009-06-09 22:54 ` H. Peter Anvin
2009-06-09 14:54 ` Linus Torvalds
2009-06-09 14:57 ` Ingo Molnar
2009-06-09 15:55 ` Avi Kivity
2009-06-09 15:38 ` Nick Piggin
2009-06-09 16:00 ` Linus Torvalds
2009-06-09 16:21 ` Nick Piggin
2009-06-09 16:26 ` Linus Torvalds
2009-06-09 16:45 ` Nick Piggin
2009-06-09 17:08 ` Linus Torvalds
2009-06-10 5:53 ` Nick Piggin
2009-06-17 9:40 ` Pavel Machek
2009-06-17 9:56 ` Nick Piggin
2009-06-10 6:29 ` Peter Zijlstra
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=4A0BED040200007800000DB0@vpn.id2.novell.com \
--to=jbeulich@novell.com \
--cc=hpa@zytor.com \
--cc=jeremy@goop.org \
--cc=jun.nakajima@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=npiggin@suse.de \
--cc=xen-devel@lists.xensource.com \
--cc=xiaohui.xin@intel.com \
--cc=xin.li@intel.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
all inboxes | Powered by JetHome®