linux-kernel - Re: Oops in microcode sysfs registration,

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: <20080729192214.2d3a4ca5@daedalus.pq.iki.fi>
Date:	Tue, 29 Jul 2008 19:22:14 +0300
From:	Pekka Paalanen <pq@....fi>
To:	Alistair John Strachan <alistair@...zero.co.uk>
Cc:	Linus Torvalds <torvalds@...ux-foundation.org>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	shaohua.li@...el.com, tigran@...azian.fsnet.co.uk,
	Ingo Molnar <mingo@...e.hu>,
	Thomas Gleixner <tglx@...utronix.de>,
	Steven Rostedt <rostedt@...dmis.org>
Subject: Re: Oops in microcode sysfs registration,

On Tue, 29 Jul 2008 14:57:58 +0100
Alistair John Strachan <alistair@...zero.co.uk> wrote:

> Noticing pq's mmiotrace was merged I tried to get a trace of the proprietary
> NVIDIA blob. Normally I wouldn't waste your time posting a tainted oops,
> however in this case it doesn't look related to the proprietary garbage and I
> think there's a real bug somewhere.
> 
> As I understand it, the mmiotrace tracing framework requires only one logical
> CPU to be active, automatically offlining the other CPUs. When mmiotrace is
> disabled, it automatically re-enables the CPUs it offlined. If I offline the
> spare CPUs myself, prior to enabling mmiotrace, I do not see the issue I'm
> about to describe. That's why tracing people have been CCed, even though that
> could be a red herring.

I have a wild hunch...
Could you try the following:
1. with all cpus enabled, load the nvidia proprietary driver
2. start and quit X
3. disable a cpu by hand
4. unload the proprietary driver
5. enable the cpu by hand

I have a vague recollection of the nvidia blob doing something bad
with notifiers, so if that crashes, and it does not crash when you
leave step 4 out, it's an nvidia problem. I assume you unloaded the
blob before disabling mmiotrace, right?

You may need to alter the sequence of things, but my guess is that
the blob may leave a notifier registered even when it is unloaded,
so it crashes when the notifier chain is traversed. I'm not sure the
backtrace really supports this scenario, but worth to try.

> The full dmesg and kernel config are available from
> http://devzero.co.uk/~alistair/2.6.27-rc1-mc-oops/
...
> Also, I'm sure this is reproducible without the NVIDIA garbage, but I was too lazy
> to test it. If you want me to repeat the experiment without the driver I would be
> more than happy to do so.

I'm not sure people are willing to look into this without a clean report,
so this would be cool. There's even a test module for mmiotrace in the
kernel, but I doubt it would make difference to use it or not, when trying
to reproduce the crash without the blob.

Thanks.

-- 
Pekka Paalanen
http://www.iki.fi/pq/
--
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/