From: Peter Zijlstra <peterz@infradead.org>
To: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: paulus <paulus@samba.org>,
stephane eranian <eranian@googlemail.com>,
Robert Richter <robert.richter@amd.com>,
Will Deacon <will.deacon@arm.com>,
Paul Mundt <lethal@linux-sh.org>,
Frederic Weisbecker <fweisbec@gmail.com>,
Cyrill Gorcunov <gorcunov@gmail.com>,
Lin Ming <ming.m.lin@intel.com>,
Yanmin <yanmin_zhang@linux.intel.com>,
Deng-Cheng Zhu <dengcheng.zhu@gmail.com>,
David Miller <davem@davemloft.net>,
linux-kernel@vger.kernel.org
Subject: Re: [RFC][PATCH 00/11] perf pmu interface -v2
Date: Mon, 28 Jun 2010 17:13:29 +0200 [thread overview]
Message-ID: <1277738009.3561.129.camel@laptop> (raw)
In-Reply-To: <4C262949.3030804@linux.vnet.ibm.com>
On Sat, 2010-06-26 at 09:22 -0700, Corey Ashford wrote:
> These patches look like they are really taking us in the right
> direction. Thanks for all your effort on this!
Yeah, although I seem to have managed to wreck all architectures tested
so far (I found some iffy on x86 too), I guess I need to carefully look
at things.
> As for the "hardware write batching", can you describe a bit more about
> what you have in mind there? I wonder if this might have something to
> do with accounting for PMU hardware which is slow to access, for
> example, via I2C via an internal bridge.
Right, so the write batching is basically delaying writing out the PMU
state to hardware until pmu::pmu_enable() time. It avoids having to
re-program the hardware when, due to a scheduling constraint, we have to
move counters around.
So say, we context switch a task, and remove the old events and add the
new ones under a single pmu::pmu_disable()/::pmu_enable() pair, we will
only hit the hardware twice (once to disable, once to enable), instead
of for each individual ::del()/::add().
For this to work we need to have an association between a context and a
pmu, otherwise its very hard to know what pmu to disable/enable; the
alternative is all of them which isn't very attractive.
Then again, it doesn't make sense to have task-counters on an I2C
attached PMU simply because writing to the PMU could cause context
switches.
next prev parent reply other threads:[~2010-06-28 15:13 UTC|newest]
Thread overview: 56+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-06-24 14:28 [RFC][PATCH 00/11] perf pmu interface -v2 Peter Zijlstra
2010-06-24 14:28 ` [PATCH 01/11] perf, x86: Fix Nehalem PMU quirk Peter Zijlstra
2010-06-24 14:28 ` [PATCH 02/11] perf: Fix argument of perf_arch_fetch_caller_regs Peter Zijlstra
2010-06-24 14:28 ` [PATCH 03/11] perf, sparc64: Fix maybe_change_configuration() PCR setting Peter Zijlstra
2010-06-24 14:28 ` [RFC][PATCH 04/11] perf: deconstify struct pmu Peter Zijlstra
2010-06-24 14:28 ` [RFC][PATCH 05/11] perf: register pmu implementations Peter Zijlstra
2010-06-28 13:21 ` Frederic Weisbecker
2010-06-28 15:16 ` Peter Zijlstra
2010-06-28 15:29 ` Frederic Weisbecker
2010-07-09 3:08 ` Paul Mackerras
2010-07-09 8:14 ` Peter Zijlstra
2010-06-24 14:28 ` [RFC][PATCH 06/11] perf: Unindent labels Peter Zijlstra
2010-06-24 14:28 ` [RFC][PATCH 07/11] perf: Reduce perf_disable() usage Peter Zijlstra
2010-06-24 14:28 ` [RFC][PATCH 08/11] perf: Per PMU disable Peter Zijlstra
2010-07-09 7:31 ` Paul Mackerras
2010-07-09 8:36 ` Peter Zijlstra
2010-06-24 14:28 ` [RFC][PATCH 09/11] perf: Default PMU ops Peter Zijlstra
2010-06-29 14:49 ` Frederic Weisbecker
2010-06-29 14:57 ` Peter Zijlstra
2010-06-29 15:00 ` Frederic Weisbecker
2010-06-29 14:58 ` Frederic Weisbecker
2010-06-29 14:59 ` Peter Zijlstra
2010-06-29 15:03 ` Frederic Weisbecker
2010-06-29 16:34 ` Peter Zijlstra
2010-06-29 18:07 ` Frederic Weisbecker
2010-06-29 18:09 ` Peter Zijlstra
2010-06-29 18:11 ` Frederic Weisbecker
2010-06-29 18:19 ` Peter Zijlstra
2010-06-29 18:21 ` Frederic Weisbecker
2010-06-24 14:28 ` [RFC][PATCH 10/11] perf: Shrink hw_perf_event Peter Zijlstra
2010-06-29 15:06 ` Frederic Weisbecker
2010-06-24 14:28 ` [RFC][PATCH 11/11] perf: Rework the PMU methods Peter Zijlstra
2010-06-29 15:37 ` Frederic Weisbecker
2010-06-29 16:40 ` Peter Zijlstra
2010-06-29 18:09 ` Frederic Weisbecker
2010-06-25 11:11 ` [RFC][PATCH 00/11] perf pmu interface -v2 Will Deacon
2010-06-25 11:16 ` Peter Zijlstra
2010-06-25 14:36 ` Will Deacon
2010-06-25 14:50 ` Peter Zijlstra
2010-07-01 14:36 ` Peter Zijlstra
2010-07-01 15:02 ` Peter Zijlstra
2010-07-01 15:31 ` MattFleming
2010-07-01 15:39 ` Peter Zijlstra
2010-07-01 16:04 ` Matt Fleming
2010-07-02 2:57 ` Paul Mundt
2010-07-02 9:52 ` Peter Zijlstra
2010-07-05 11:14 ` Paul Mundt
2010-07-08 11:13 ` Peter Zijlstra
2010-07-08 11:19 ` Ingo Molnar
2010-07-18 19:37 ` Matt Fleming
2010-07-02 12:55 ` Will Deacon
2010-06-26 11:22 ` Matt Fleming
2010-06-26 16:22 ` Corey Ashford
2010-06-28 15:13 ` Peter Zijlstra [this message]
2010-06-30 17:19 ` Corey Ashford
2010-06-30 18:11 ` 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=1277738009.3561.129.camel@laptop \
--to=peterz@infradead.org \
--cc=cjashfor@linux.vnet.ibm.com \
--cc=davem@davemloft.net \
--cc=dengcheng.zhu@gmail.com \
--cc=eranian@googlemail.com \
--cc=fweisbec@gmail.com \
--cc=gorcunov@gmail.com \
--cc=lethal@linux-sh.org \
--cc=linux-kernel@vger.kernel.org \
--cc=ming.m.lin@intel.com \
--cc=paulus@samba.org \
--cc=robert.richter@amd.com \
--cc=will.deacon@arm.com \
--cc=yanmin_zhang@linux.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;
as well as URLs for NNTP newsgroup(s).