All of lore.kernel.org
 help / color / mirror / Atom feed
* Re: [PATCH -v3] perf, x86: try to handle unknown nmis with running perfctrs
@ 2010-08-20 15:05 Don Zickus
  2010-08-20 15:25 ` Ingo Molnar
  2010-08-20 23:31 ` Don Zickus
  0 siblings, 2 replies; 70+ messages in thread
From: Don Zickus @ 2010-08-20 15:05 UTC (permalink / raw)
  To: Ingo Molnar
  Cc: Peter Zijlstra, Robert Richter, Cyrill Gorcunov, Lin Ming,
	fweisbec, linux-kernel, Huang, Ying, Yinghai Lu, Andi Kleen

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=utf-8, Size: 537 bytes --]

I'll test tip later today to see if I can reproduce it.

Cheers,
Don

Ingo Molnar <mingo@elte.hu> wrote:

>
>it's not working so well, i'm getting:
>
> Uhhuh. NMI received for unknown reason 00 on CPU 9.
> Do you have a strange power saving mode enabled?
> Dazed and confused, but trying to continue
>
>on a nehalem box, after a perf top and perf stat run.
>
>Thanks,
>
>	Ingo
ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥

^ permalink raw reply	[flat|nested] 70+ messages in thread
* Re: A question of perf NMI handler
@ 2010-08-04 15:18 Cyrill Gorcunov
  2010-08-04 15:50 ` Don Zickus
  0 siblings, 1 reply; 70+ messages in thread
From: Cyrill Gorcunov @ 2010-08-04 15:18 UTC (permalink / raw)
  To: Peter Zijlstra
  Cc: Don Zickus, Robert Richter, Lin Ming, Ingo Molnar, fweisbec,
	linux-kernel, Huang, Ying, Yinghai Lu

On Wed, Aug 04, 2010 at 05:02:41PM +0200, Peter Zijlstra wrote:
> On Wed, 2010-08-04 at 10:52 -0400, Don Zickus wrote:
> > > Right so I looked up your thing and while that limits the damage in that
> > > at some point it will let NMIs pass, it will still consume too many.
> > > Meaning that Yinghai will have to potentially press his NMI button
> > > several times before it registers.
> > 
> > Ok.  Thanks for reviewing.  How does it consume to many?  I probably don't
> > understand how perf is being used in the non-simple scenarios.
> 
> Suppose you have 4 counters (AMD, intel-nhm+), when more than 2 overflow
> the first will raise the PMI, if the other 2+ overflow before we disable
> the PMU it will try to raise 2+ more PMIs, but because hardware only has
> a single interrupt pending bit it will at most cause a single extra
> interrupt after we finish servicing the first one.
> 
> So then the first interrupt will see 3+ overflows, return 3+, and will
> thus eat 2+ NMIs, only one of which will be the pending interrupt,
> leaving 1+ NMIs from other sources to consume unhandled.
> 
> In which case Yinghai will have to press his NMI button 2+ times before
> it registers.
> 
> That said, that might be a better situation than always consuming
> unknown NMIs.. 
> 

Well, first I guess having Yinghai CC'ed is a bonus ;)
The second thing is that I don't get why perf handler can't be _last_
call in default_do_nmi, if there were any nmi with reason (serr or parity)
I think they should be calling first which of course don't eliminate
the former issue but somewhat make it weaken.

	-- Cyrill

^ permalink raw reply	[flat|nested] 70+ messages in thread

end of thread, other threads:[~2010-08-27 19:35 UTC | newest]

Thread overview: 70+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-08-20 15:05 [PATCH -v3] perf, x86: try to handle unknown nmis with running perfctrs Don Zickus
2010-08-20 15:25 ` Ingo Molnar
2010-08-23  8:53   ` Ingo Molnar
2010-08-24 16:22     ` Cyrill Gorcunov
2010-08-24 17:09       ` Robert Richter
2010-08-24 17:20         ` Cyrill Gorcunov
2010-08-24 17:21           ` Cyrill Gorcunov
2010-08-24 17:15     ` Robert Richter
2010-08-24 17:28       ` Cyrill Gorcunov
2010-08-24 18:46         ` Don Zickus
2010-08-24 18:54           ` Cyrill Gorcunov
2010-08-24 19:52             ` Cyrill Gorcunov
2010-08-24 20:27               ` Don Zickus
2010-08-24 20:40                 ` Cyrill Gorcunov
2010-08-25 23:52                   ` Frederic Weisbecker
2010-08-26  9:11                     ` Cyrill Gorcunov
2010-08-25 10:20                 ` Robert Richter
2010-08-26 21:14     ` Don Zickus
2010-08-27  7:51       ` Robert Richter
2010-08-27 13:39         ` Don Zickus
2010-08-27  8:10       ` Robert Richter
2010-08-27 13:44         ` Don Zickus
2010-08-27 14:05           ` Robert Richter
2010-08-27 15:05             ` Don Zickus
2010-08-27 15:48               ` Robert Richter
2010-08-27 18:57         ` Don Zickus
2010-08-27 19:00           ` Yinghai Lu
2010-08-27 19:33           ` Robert Richter
2010-08-25  9:48   ` Robert Richter
2010-08-25 10:41     ` Ingo Molnar
2010-08-25 11:00       ` Ingo Molnar
2010-08-25 20:11         ` Don Zickus
2010-08-25 20:24           ` Cyrill Gorcunov
2010-08-25 21:20             ` Don Zickus
2010-08-25 21:36               ` Cyrill Gorcunov
2010-08-26  9:00           ` Robert Richter
2010-08-26  9:18             ` Cyrill Gorcunov
2010-08-26 14:31               ` Don Zickus
2010-08-26 15:22               ` Don Zickus
2010-08-26 15:34                 ` Cyrill Gorcunov
2010-08-26 16:40                   ` Don Zickus
2010-08-26 18:02                     ` Cyrill Gorcunov
2010-08-27  7:57                       ` Robert Richter
2010-08-27  8:11                         ` Peter Zijlstra
2010-08-27  8:31                           ` Robert Richter
2010-08-25 11:02       ` Robert Richter
2010-08-25 11:19         ` Ingo Molnar
2010-08-20 23:31 ` Don Zickus
  -- strict thread matches above, loose matches on Subject: below --
2010-08-04 15:18 A question of perf NMI handler Cyrill Gorcunov
2010-08-04 15:50 ` Don Zickus
2010-08-04 16:10   ` Cyrill Gorcunov
2010-08-04 16:20     ` Don Zickus
2010-08-04 16:39       ` Cyrill Gorcunov
2010-08-04 18:48         ` Robert Richter
2010-08-04 19:26           ` Cyrill Gorcunov
2010-08-06  6:52             ` Robert Richter
2010-08-06 14:21               ` Don Zickus
2010-08-09 19:48                 ` [PATCH] perf, x86: try to handle unknown nmis with running perfctrs Robert Richter
2010-08-17 15:22                   ` [PATCH -v3] " Robert Richter
2010-08-17 16:17                     ` Cyrill Gorcunov
2010-08-19 10:45                     ` Peter Zijlstra
2010-08-19 12:39                       ` Robert Richter
2010-08-19 14:12                       ` Don Zickus
2010-08-19 14:27                         ` Peter Zijlstra
2010-08-19 15:20                           ` Don Zickus
2010-08-19 17:43                           ` Cyrill Gorcunov
2010-08-19 17:53                             ` Peter Zijlstra
2010-08-19 21:58                           ` Don Zickus
2010-08-20  8:50                             ` Peter Zijlstra
2010-08-20  1:50                           ` Don Zickus
2010-08-20  8:16                             ` Ingo Molnar
2010-08-20 10:04                               ` Peter Zijlstra
2010-08-20 10:30                                 ` Cyrill Gorcunov
2010-08-20 12:39                                 ` Don Zickus
2010-08-20 13:27                                   ` Ingo Molnar
2010-08-20 13:51                                     ` Don Zickus
2010-08-20 14:17                                       ` Ingo Molnar
2010-08-20 20:45                                         ` Cyrill Gorcunov
2010-08-24 21:48                                         ` Don Zickus
2010-08-20  8:36                             ` Robert Richter

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.