linux-kernel - perf/nmi : NMI Dazed and confused messages on core2

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-next>] [day] [month] [year] [list]

Message-ID: <alpine.DEB.2.10.1311141626010.8746@vincent-weaver-1.um.maine.edu>
Date:	Thu, 14 Nov 2013 16:32:28 -0500 (EST)
From:	Vince Weaver <vincent.weaver@...ne.edu>
To:	LKML <linux-kernel@...r.kernel.org>
cc:	Ingo Molnar <mingo@...nel.org>,
	Peter Zijlstra <peterz@...radead.org>,
	Paul Mackerras <paulus@...ba.org>,
	Arnaldo Carvalho de Melo <acme@...stprotocols.net>,
	Don Zickus <dzickus@...hat.com>
Subject: perf/nmi : NMI Dazed and confused messages on core2

Hello

when running the perf_fuzzer on a core2 machine I often get the syslog 
spammed with messages like this:

[ 7911.810186] Uhhuh. NMI received for unknown reason 3d on CPU 0.
[ 7911.812756] Do you have a strange power saving mode enabled?
[ 7911.812756] Dazed and confused, but trying to continue
[ 7912.554225] Uhhuh. NMI received for unknown reason 2d on CPU 1.
[ 7912.556758] Do you have a strange power saving mode enabled?
[ 7912.556758] Dazed and confused, but trying to continue
[ 7912.838169] Uhhuh. NMI received for unknown reason 2d on CPU 0.
[ 7912.840726] Do you have a strange power saving mode enabled?
[ 7912.840726] Dazed and confused, but trying to continue

I've managed to narrow this down to a small program that will reproduce 
the issue (attached) although I can only get it to reproduce on a core2 
system (it doesn't work on recent atom or ivy bridge machines I have 
access to).

So maybe it's just an obscure core2 problem and not worth fixing, but I 
thought I'd report it anyway.

It looks like it might be related to setting up a cache event to trigger 
with high frequency while some other events are also triggering.

Vince Weaver
vincent.weaver@...ne.edu
http://www.eece.maine.edu/~vweaver/
View attachment "core2_nmi_flood.c" of type "TEXT/x-csrc" (12487 bytes)