linux-kernel - Re: [PATCH 00/11] perf tool: Add PERF_SAMPLE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 26 Oct 2012 12:23:36 +0200
From:	Jiri Olsa <jolsa@...hat.com>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
Cc:	Namhyung Kim <namhyung@...nel.org>, linux-kernel@...r.kernel.org,
	Arnaldo Carvalho de Melo <acme@...stprotocols.net>,
	Ingo Molnar <mingo@...e.hu>, Paul Mackerras <paulus@...ba.org>,
	Corey Ashford <cjashfor@...ux.vnet.ibm.com>,
	Frederic Weisbecker <fweisbec@...il.com>
Subject: Re: [PATCH 00/11] perf tool: Add PERF_SAMPLE_READ sample read support

On Fri, Oct 26, 2012 at 11:14:45AM +0200, Peter Zijlstra wrote:
> On Fri, 2012-10-26 at 10:29 +0900, Namhyung Kim wrote:
> > Hi Jiri,
> > 
> > On Sat, 20 Oct 2012 16:33:08 +0200, Jiri Olsa wrote:
> > > hi,
> > > adding support to read sample values through the PERF_SAMPLE_READ
> > > sample type. It's now possible to specify 'S' modifier for an event
> > > and get its sample value by PERF_SAMPLE_READ.
> > 
> > I have a question.  What's an actual impact of specifying 'S' modifiere
> > to a non-group event or even only a (non-leader) member of a group?  For
> > instance, 'cycles:S' or '{branches,branch-misses:S}'.
> 
> I would hope a syntax error from the parser ;-)

yeaa.. no ;)

$ ./perf record -e '{cycles:S,cache-misses}' ls
...
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.011 MB perf.data (~476 samples) ]

$ ./perf report
non matching sample_type

> 
> I haven't actually looked at the implementation, but I understood it to
> be a group modifier, not an event modifier.

we might want to be able to use PERF_SAMPLE_READ for single event
same as for groups

the difference between single event and group 'S' modifier:

  $ ./perf record -e 'cycles:S' ls
  - records 'cycles' samples and read period value via PERF_SAMPLE_READ

  $ ./perf record -e '{cycles,cache-misses}:S'
  - samples just on 'cycles' samples and read both period values
   (cycles and cache-misses) via PERF_SAMPLE_READ group format

  $ ./perf record -e '{cycles,cache-misses}:S,instructions' ls
  $ ./perf record -e '{cycles:S,cache-misses},instructions' ls
  $ ./perf record -e 'cycles:S,instructions' ls
  - non matching sample_type

hm, thats the unique sample_type issue again ;) Once we set
PERF_SAMPLE_READ for event or group, we need to set it for
all other events in session, otherwise the report fails

sooo, it looks like:
  - global record '-S' option instead, setting PERF_SAMPLE_READ sample type
    globaly for all events
  - and ':S' group modifier to enable sampling only on leader
    with group format reading for the rest of the group
  - ':S' group modifier alone would imply -S

How about that?

thanks,
jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/