On Tue, 2009-04-07 at 18:58 +0200, Thomas Gleixner wrote:
quoted text
> Fernando,
> 
> On Tue, 7 Apr 2009, Fernando Lopez-Lezcano wrote:
> 
> > On Mon, 2009-04-06 at 21:54 +0200, Thomas Gleixner wrote:
> > > Fernando,
> > > 
> > > On Sun, 5 Apr 2009, Fernando Lopez-Lezcano wrote:
> > > > > Hmm. Can you upload your full patch queue and the binary rpm so I can
> > > > > give it a test ride ?
> > > > 
> > > > Sure, no problem. I just did a release to planetcore-testing repo (with
> > > > a warning to the planetccrma list), you can find it here:
> > > > 
> > > > http://ccrma.stanford.edu/planetccrma/mirror/fedora/linux/planetcore/testing/10/i386/r...
> > > > http://ccrma.stanford.edu/planetccrma/mirror/fedora/linux/planetcore/testing/10/x86_64...
> > > > 
> > > > (the one I have tested and hangs is the i386 PAE version)
> > > > Source here:
> > > > 
> > > > http://ccrma.stanford.edu/planetccrma/mirror/fedora/linux/planetcore/testing/10/SRPMS/
> > > > 
> > > > (if you don't want to figure out the .src.rpm - I would not blame you -
> > > > I can give you a list of the patches that were actually applied - of the
> > > > ones that are included in the package)
> > > 
> > > No problem. I know how to distangle those beasts.
> > > 
> > > Still I have no reproducer yet. One of my 32bit test machines survived
> > > a 200 cycle test w/o showing it.
> > 
> > Arghh, I don't know what it could be then. You mean 200 power cycles??
> > Wow... it only takes one here. Thanks for testing. 
> 
> Does it trigger when you just boot up and the immediately shut down
> w/o doing any work on it ?

Yes. Just tried again. Clarification: triggers on a "halt -p", _not_ on
a reboot. 

quoted text
> Could you try whether the following triggers the bug as well: Boot up,
> do work or whatever. Make sure your data is sycned to disk. :)
> 
> # echo 0 >/sys/devices/system/cpu/cpu1/online
> 
> If that does not trigger, then check whether it survives the
> shutdown/reboot. If it does we look further.

It triggers it immediately (in my laptop), I tried from a text console
and it spews things and at the bottom a poke_text thing like before
(after a while I got a couple of messages from iwl3945 about a timeout
and somehow it started to reboot, either when I pressed the power button
or when I did <ctrl><alt><del>)

quoted text
> > > Is there anything special what you run, load ... ?
> > 
> > Not that I can think of right now. This happens on 32 bit installs on
> > both Fedora 9 and Fedora 10. Nothing special installed, of course I have
> > the Planet CCRMA packages but they don't involve additional kernel
> > modules or anything that (I think) deals with startup or shutdown. 
> 
> Can you provide the list of modules loaded ?

Three lists attached, first all modules after I login, then modules
after a single user boot and then two more after I removed bluetooth and
80211* modules on a hunch, fails to power off in all cases (I first saw
a ref to blue something at the very top of the dump, so removed that,
then saw a reference to 80211 at the top, removed that, then I gave
up :-) 

-- Fernando