Re: x86/mce: machine check warning during poweroff

From: Linus Torvalds
Date: Fri Jan 13 2012 - 16:08:38 EST


On Fri, Jan 13, 2012 at 12:53 PM, Srivatsa S. Bhat
<srivatsa.bhat@xxxxxxxxxxxxxxxxxx> wrote:
>
> Wait a minute, did you mention "second attempt"? I think I have something
> interesting..

Yes, I think you're hitting the exact same thing.

I *think* that what is going on is that we free some data structure
too early, and we didn't use to free them before. I tried to see if I
could catch it with slab and list debugging, but I didn't see
anything, and the machine I used for suspend/resume had other issues
too (wireless network - which is the *only* network on that machine -
hung on resume), so I ended up punting and just disabling MCE to
concentrate on those issues.

On eof the differences between sysdev and 'struct device' is that
sysdev doesn't bother refcounting parents etc. So there could have
been some refcount problem that was never relevant with the old sysdev
code. I dunno.

The wireless issues got resolved for me, and I haven't gotten back to
MCE yet. I was *really* hoping that somebody else could figure it out,
since I'm not the only one seeing it..

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