mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Steven Rostedt <rostedt@goodmis.org>
To: Robert Richter <robert.richter@amd.com>
Cc: "Andi Kleen" <andi@firstfloor.org>, "Ingo Molnar" <mingo@elte.hu>,
	"Peter Zijlstra" <a.p.zijlstra@chello.nl>,
	"Frédéric Weisbecker" <fweisbec@gmail.com>,
	"Mike Galbraith" <efault@gmx.de>,
	LKML <linux-kernel@vger.kernel.org>,
	"Arnaldo Carvalho de Melo" <acme@redhat.com>,
	"Pekka Enberg" <penberg@cs.helsinki.fi>,
	"Paul Mackerras" <paulus@samba.org>,
	oprofile-list <oprofile-list@lists.sourceforge.net>,
	"H. Peter Anvin" <hpa@zytor.com>,
	"Thomas Gleixner" <tglx@linutronix.de>,
	"Linus Torvalds" <torvalds@linux-foundation.org>,
	"Arjan van de Ven" <arjan@infradead.org>
Subject: Re: [PATCH 0/26] oprofile: Performance counter multiplexing
Date: Fri, 05 Mar 2010 12:01:13 -0500	[thread overview]
Message-ID: <1267808473.10871.1775.camel@gandalf.stny.rr.com> (raw)
In-Reply-To: <20100305164615.GY13205@erda.amd.com>

On Fri, 2010-03-05 at 17:46 +0100, Robert Richter wrote:
> On 26.02.10 15:51:04, Andi Kleen wrote:
> > That said the biggest problem with oprofile right now that the
> > new buffer it's using is quite a lot less reliable and drops
> > events left and right on any non trivial load. That makes
> > oprofile very unreliable, especially in call graph mode.
> 
> (cc'ing Steve)
> 

Thanks.

> Andi,
> 
> the tests I run with oprofile do not indicate unreliable ring_buffer
> behavior. Maybe my use cases and loads are different. Can you describe
> a setup where this may happen for sure? What is the impact, do you
> have lost samples or inconsistent buffer data? Is the data loss in
> kernel or user space? Also, I am not aware that the ring_buffer is
> unreliable for ftrace or tracepoints, where it is heavily used. I
> really want to find the root cause here.

Yes, the ftrace ring buffer (which I believe oprofile now uses) is
lockless and re-entrant. That means if a interrupt goes off while one
event is being recorded, and that interrupt writes to the same ring
buffer, the ring buffer will be able to record it without dropping the
event.

As long as the readers keep up, no event will ever be dropped. Perhaps
you need to increase the size of the ring buffers?

-- Steve



  reply	other threads:[~2010-03-05 17:01 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-07-28 17:07 Robert Richter
2009-07-28 17:07 ` [PATCH 01/26] x86/oprofile: Rework and simplify nmi_cpu_setup() Robert Richter
2009-07-28 17:07 ` [PATCH 02/26] x86/oprofile: Whitespaces changes only Robert Richter
2009-07-28 17:07 ` [PATCH 03/26] oprofile: Implement performance counter multiplexing Robert Richter
2009-07-28 17:07 ` [PATCH 04/26] x86/oprofile: Fix usage of NUM_CONTROLS/NUM_COUNTERS macros Robert Richter
2009-07-28 17:07 ` [PATCH 05/26] x86/oprofile: Use per_cpu() instead of __get_cpu_var() Robert Richter
2009-07-28 17:07 ` [PATCH 06/26] x86/oprofile: Fix initialization of switch_index Robert Richter
2009-07-28 17:07 ` [PATCH 07/26] oprofile: oprofile_set_timeout(), return with error for invalid args Robert Richter
2009-07-28 17:07 ` [PATCH 08/26] oprofile: Rename variable timeout_jiffies and move to oprofile_files.c Robert Richter
2009-07-28 17:07 ` [PATCH 09/26] oprofile: Remove oprofile_multiplexing_init() Robert Richter
2009-07-28 17:07 ` [PATCH 10/26] oprofile: Grouping multiplexing code in oprof.c Robert Richter
2009-07-28 17:07 ` [PATCH 11/26] oprofile: Introduce op_x86_phys_to_virt() Robert Richter
2009-07-28 17:07 ` [PATCH 12/26] oprofile: Grouping multiplexing code in op_model_amd.c Robert Richter
2009-07-28 17:07 ` [PATCH 13/26] x86/oprofile: Implement multiplexing setup/shutdown functions Robert Richter
2009-07-28 17:07 ` [PATCH 14/26] x86/oprofile: Moving nmi_setup_cpu_mux() in nmi_int.c Robert Richter
2009-07-28 17:07 ` [PATCH 15/26] x86/oprofile: Moving nmi_cpu_save/restore_mpx_registers() " Robert Richter
2009-07-28 17:07 ` [PATCH 16/26] x86/oprofile: Moving nmi_cpu_switch() " Robert Richter
2009-07-28 17:07 ` [PATCH 17/26] x86/oprofile: Remove const qualifier from struct op_x86_model_spec Robert Richter
2009-07-28 17:07 ` [PATCH 18/26] x86/oprofile: Remove unused num_virt_controls " Robert Richter
2009-07-28 17:07 ` [PATCH 19/26] x86/oprofile: Modify initialization of num_virt_counters Robert Richter
2009-07-28 17:07 ` [PATCH 20/26] x86/oprofile: Add function has_mux() to check multiplexing support Robert Richter
2009-07-28 17:07 ` [PATCH 21/26] x86/oprofile: Enable multiplexing only if the model supports it Robert Richter
2009-07-28 17:07 ` [PATCH 22/26] x86/oprofile: Implement mux_clone() Robert Richter
2009-07-28 17:07 ` [PATCH 23/26] oprofile: Adding switch counter to oprofile statistic variables Robert Richter
2009-07-28 17:07 ` [PATCH 24/26] x86/oprofile: Implement op_x86_virt_to_phys() Robert Richter
2009-07-28 17:07 ` [PATCH 25/26] x86/oprofile: Add counter reservation check for virtual counters Robert Richter
2009-07-28 17:07 ` [PATCH 26/26] x86/oprofile: Small coding style fixes Robert Richter
2009-07-28 18:45 ` [PATCH 0/26] oprofile: Performance counter multiplexing Andi Kleen
2009-07-28 22:18   ` Suthikulpanit, Suravee
2009-08-03 11:22 ` Ingo Molnar
2009-08-03 14:25   ` Arjan van de Ven
2009-08-04 20:11     ` Ingo Molnar
2009-08-05 12:35       ` Robert Richter
2009-08-06  7:31         ` Pekka Enberg
2009-08-06 10:51           ` Ingo Molnar
2009-08-13 16:13             ` Robert Richter
     [not found]           ` <20090806105134.GD11236__37984.4768475325$1249556453$gmane$org@elte.hu>
2010-02-26 14:51             ` Andi Kleen
2010-03-05 16:46               ` Robert Richter
2010-03-05 17:01                 ` Steven Rostedt [this message]
2010-03-08  3:51                 ` Andi Kleen
2009-08-03 16:30   ` Robert Richter
2009-08-04 17:05     ` Ingo Molnar

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=1267808473.10871.1775.camel@gandalf.stny.rr.com \
    --to=rostedt@goodmis.org \
    --cc=a.p.zijlstra@chello.nl \
    --cc=acme@redhat.com \
    --cc=andi@firstfloor.org \
    --cc=arjan@infradead.org \
    --cc=efault@gmx.de \
    --cc=fweisbec@gmail.com \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=oprofile-list@lists.sourceforge.net \
    --cc=paulus@samba.org \
    --cc=penberg@cs.helsinki.fi \
    --cc=robert.richter@amd.com \
    --cc=tglx@linutronix.de \
    --cc=torvalds@linux-foundation.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

all inboxes | Powered by JetHome®