From: Don Zickus <dzickus@redhat.com>
To: Stephane Eranian <eranian@google.com>
Cc: Robert Richter <robert.richter@amd.com>,
Cyrill Gorcunov <gorcunov@gmail.com>,
"mingo@redhat.com" <mingo@redhat.com>,
"hpa@zytor.com" <hpa@zytor.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"yinghai@kernel.org" <yinghai@kernel.org>,
"andi@firstfloor.org" <andi@firstfloor.org>,
"peterz@infradead.org" <peterz@infradead.org>,
"ying.huang@intel.com" <ying.huang@intel.com>,
"fweisbec@gmail.com" <fweisbec@gmail.com>,
"ming.m.lin@intel.com" <ming.m.lin@intel.com>,
"tglx@linutronix.de" <tglx@linutronix.de>,
"mingo@elte.hu" <mingo@elte.hu>
Subject: Re: [tip:perf/urgent] perf, x86: Catch spurious interrupts after disabling counters
Date: Wed, 29 Sep 2010 16:03:23 -0400 [thread overview]
Message-ID: <20100929200323.GC26290@redhat.com> (raw)
In-Reply-To: <AANLkTimp=+1GzBOYaUZWtDF6teGt6FZe+RTpb9fAyOyd@mail.gmail.com>
On Wed, Sep 29, 2010 at 09:42:26PM +0200, Stephane Eranian wrote:
> On Wed, Sep 29, 2010 at 8:12 PM, Don Zickus <dzickus@redhat.com> wrote:
> > Robert,
> >
> > I think you missed Stephane's point. Say for example, kgdb is being used
> > while we are doing stuff with the perf counter (and say kgdb's handler is
> > a lower priority than perf; which isn't true I know, but let's say):
> >
> Yes, exactly my point. The reality is you cannot afford to have false positive
> because you may starve another subsystem from an important notification.
>
> I think it boils down to whether or not we need an error message (Dazed) in
> case no subsystem claimed the NMI. If you were to just silently consume the
> NMI when no subsystem claims it, then you would not have these issues.
>
> What Don has done is use a heuristic which gets activated when a PMU
> interrupt handler signals that more than one counter have overflowed. His
> claim is that this situation is likely to trigger back-to-back.
Actually its Robert's heuristic. :-)
>
> The reason this heuristic works is because it waits until ALL the subsystems
> have seen the notification before it declares that the NMI was PMU spurious.
> To do that is uses the DIE_NMI_UNKNOWN callchain. Handler on this chain
> get call last, after all subsystems have seen the notification once. I believe
> that is the only way to safely "consume" a "spurious" NMI and avoid
> the 'Dazed' message. Anything else runs the risks of starving the other
> subsystems.
I agree.
Cheers,
Don
next prev parent reply other threads:[~2010-09-29 20:03 UTC|newest]
Thread overview: 117+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-09-02 19:07 [PATCH 0/3 v2] nmi perf fixes Don Zickus
2010-09-02 19:07 ` [PATCH 1/3] perf, x86: Fix accidentally ack'ing a second event on intel perf counter Don Zickus
2010-09-02 19:26 ` Cyrill Gorcunov
2010-09-02 20:00 ` Don Zickus
2010-09-02 20:36 ` Cyrill Gorcunov
2010-09-03 7:10 ` [tip:perf/urgent] " tip-bot for Don Zickus
2010-09-03 7:39 ` Yinghai Lu
2010-09-03 15:00 ` Don Zickus
2010-09-03 17:15 ` Yinghai Lu
2010-09-03 18:35 ` Don Zickus
2010-09-03 19:24 ` Yinghai Lu
2010-09-03 20:10 ` Don Zickus
2010-10-04 23:24 ` Yinghai Lu
2010-10-11 20:25 ` Don Zickus
2010-09-02 19:07 ` [PATCH 2/3] perf, x86: Try to handle unknown nmis with an enabled PMU Don Zickus
2010-09-03 7:11 ` [tip:perf/urgent] " tip-bot for Robert Richter
2010-09-02 19:07 ` [PATCH 3/3] perf, x86: Fix handle_irq return values Don Zickus
2010-09-03 7:10 ` [tip:perf/urgent] " tip-bot for Peter Zijlstra
2010-09-10 11:41 ` [PATCH 0/3 v2] nmi perf fixes Peter Zijlstra
2010-09-10 12:10 ` Stephane Eranian
2010-09-10 12:13 ` Stephane Eranian
2010-09-10 13:27 ` Don Zickus
2010-09-10 14:46 ` Ingo Molnar
2010-09-10 15:17 ` Robert Richter
2010-09-10 15:58 ` Peter Zijlstra
2010-09-10 16:41 ` Ingo Molnar
2010-09-10 16:42 ` Ingo Molnar
2010-09-10 16:37 ` Ingo Molnar
2010-09-10 16:51 ` Ingo Molnar
2010-09-10 15:56 ` [PATCH] x86: fix duplicate calls of the nmi handler Robert Richter
2010-09-10 16:15 ` Peter Zijlstra
2010-09-11 9:41 ` Ingo Molnar
2010-09-11 11:44 ` Robert Richter
2010-09-11 12:45 ` Ingo Molnar
2010-09-12 9:52 ` Robert Richter
2010-09-13 14:37 ` Robert Richter
2010-09-14 17:41 ` Robert Richter
2010-09-15 16:20 ` [PATCH] perf, x86: catch spurious interrupts after disabling counters Robert Richter
2010-09-15 16:36 ` Stephane Eranian
2010-09-15 17:00 ` Robert Richter
2010-09-15 17:32 ` Stephane Eranian
2010-09-15 18:44 ` Robert Richter
2010-09-15 19:34 ` Cyrill Gorcunov
2010-09-15 20:21 ` Stephane Eranian
2010-09-15 20:39 ` Cyrill Gorcunov
2010-09-15 22:27 ` Robert Richter
2010-09-16 14:51 ` Frederic Weisbecker
2010-09-15 16:46 ` Cyrill Gorcunov
2010-09-15 16:47 ` Stephane Eranian
2010-09-15 17:02 ` Cyrill Gorcunov
2010-09-15 17:28 ` Robert Richter
2010-09-15 17:40 ` Cyrill Gorcunov
2010-09-15 22:10 ` Robert Richter
2010-09-16 6:53 ` Cyrill Gorcunov
2010-09-16 17:34 ` Peter Zijlstra
2010-09-17 8:51 ` Robert Richter
2010-09-17 9:14 ` Peter Zijlstra
2010-09-17 13:06 ` Stephane Eranian
2010-09-20 8:41 ` Robert Richter
2010-09-24 0:02 ` Don Zickus
2010-09-24 3:18 ` Don Zickus
2010-09-24 10:03 ` Robert Richter
2010-09-24 13:38 ` Stephane Eranian
2010-09-30 12:33 ` Peter Zijlstra
2010-09-24 18:11 ` Don Zickus
2010-09-24 10:41 ` [tip:perf/urgent] perf, x86: Catch " tip-bot for Robert Richter
2010-09-29 12:26 ` Stephane Eranian
2010-09-29 12:53 ` Robert Richter
2010-09-29 12:54 ` Robert Richter
2010-09-29 13:13 ` Stephane Eranian
2010-09-29 13:28 ` Stephane Eranian
2010-09-29 15:01 ` Robert Richter
2010-09-29 15:12 ` Robert Richter
2010-09-29 15:27 ` Cyrill Gorcunov
2010-09-29 15:33 ` Stephane Eranian
2010-09-29 15:45 ` Cyrill Gorcunov
2010-09-29 15:51 ` Cyrill Gorcunov
2010-09-29 16:32 ` Robert Richter
2010-09-29 16:48 ` Cyrill Gorcunov
2010-09-29 16:00 ` Stephane Eranian
2010-09-29 17:09 ` Robert Richter
2010-09-29 17:41 ` Cyrill Gorcunov
2010-09-29 18:12 ` Don Zickus
2010-09-29 19:42 ` Stephane Eranian
2010-09-29 20:03 ` Don Zickus [this message]
2010-09-30 9:12 ` Robert Richter
2010-09-30 19:44 ` Don Zickus
2010-10-01 7:17 ` Robert Richter
[not found] ` <AANLkTimUyLaVaBigjm0-CwRsdh4UXWDiss2ffX53S+k_@mail.gmail.com>
2010-10-01 11:53 ` Stephane Eranian
2010-10-02 9:35 ` Robert Richter
2010-10-04 8:53 ` Stephane Eranian
2010-10-04 9:07 ` Andi Kleen
2010-10-04 17:28 ` Stephane Eranian
2010-09-29 16:31 ` Robert Richter
2010-09-29 16:22 ` Robert Richter
2010-09-29 19:01 ` Don Zickus
2010-09-29 13:39 ` Robert Richter
2010-09-29 13:56 ` Stephane Eranian
2010-09-29 14:00 ` Stephane Eranian
2010-10-02 9:50 ` Robert Richter
2010-10-02 17:40 ` Stephane Eranian
2010-09-29 15:02 ` Cyrill Gorcunov
2010-09-16 17:42 ` [PATCH] x86: fix duplicate calls of the nmi handler Peter Zijlstra
2010-09-16 20:18 ` Stephane Eranian
2010-09-17 7:09 ` Peter Zijlstra
2010-09-17 0:13 ` Huang Ying
2010-09-17 7:52 ` Peter Zijlstra
2010-09-17 8:13 ` Robert Richter
2010-09-17 8:37 ` Cyrill Gorcunov
2010-09-17 8:47 ` Huang Ying
2010-09-10 13:34 ` [PATCH 0/3 v2] nmi perf fixes Peter Zijlstra
2010-09-10 13:52 ` Peter Zijlstra
2010-09-13 8:55 ` Cyrill Gorcunov
2010-09-13 9:54 ` Stephane Eranian
2010-09-13 10:07 ` Cyrill Gorcunov
2010-09-13 10:10 ` Stephane Eranian
2010-09-13 10:12 ` Cyrill Gorcunov
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=20100929200323.GC26290@redhat.com \
--to=dzickus@redhat.com \
--cc=andi@firstfloor.org \
--cc=eranian@google.com \
--cc=fweisbec@gmail.com \
--cc=gorcunov@gmail.com \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=ming.m.lin@intel.com \
--cc=mingo@elte.hu \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=robert.richter@amd.com \
--cc=tglx@linutronix.de \
--cc=ying.huang@intel.com \
--cc=yinghai@kernel.org \
/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).