linux-kernel - Re: perf: multiplexing broken on amd fam16h?

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]

Message-ID: <alpine.DEB.2.20.1507211600590.29103@vincent-weaver-1.umelst.maine.edu>
Date:	Tue, 21 Jul 2015 16:08:28 -0400 (EDT)
From:	Vince Weaver <vincent.weaver@...ne.edu>
To:	Peter Zijlstra <peterz@...radead.org>
cc:	Vince Weaver <vincent.weaver@...ne.edu>,
	linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...hat.com>,
	Arnaldo Carvalho de Melo <acme@...nel.org>
Subject: Re: perf: multiplexing broken on amd fam16h?

On Tue, 21 Jul 2015, Peter Zijlstra wrote:

> On Tue, Jul 21, 2015 at 02:14:54PM -0400, Vince Weaver wrote:
> > Hello
> > 
> > I have a new AMD A8 system and perf multiplexing support is acting weird.
> > 
> > If I run the perf_event_tests check_multiplexing test it fails like this:
> > 
> > Event	TotalCount	RawCount	Scale					ScaledCount	Error
> > 0	500282143	150403975	-2.00 (c8050b5 fffffffff9c224ef)	-300807950	160.13%
> > 
> > So the problem is that the TOTAL_TIME_RUNNING value is negative, which as 
> > far as I know shouldn't be possible.  
> 
> Correct, that should not happen. I don't have a Fam16 machine, but I can
> try and stare at the code a wee bit. Although there's another few perf
> issues I am still looking into as well :/

well I've traced it back to see that in
	update_event_times()
run_end is less than event->tstamp_running, which is why the result
is negative.

[ 1279.857205] VMW: 1 run_end=1279846999345 tstamp_running=1279929016607
[ 1279.857222] VMW: 0 run_end=1279855184453 tstamp_running=1279931111760

It seems to happen both in the PERF_EVENT_STATE_INACTIVE and 
!PERF_EVENT_STATE_INACTIVE cases.

tracking back why those values are wrong is proving to be a bit trickier.

Vince
--
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/