On Tuesday 07 July 2009, walt wrote: > On 07/06/2009 04:23 PM, Mick wrote: > > On Monday 06 July 2009, walt wrote: > >> On 07/06/2009 04:39 AM, Mick wrote: > >>> On Monday 06 July 2009, walt wrote: > >>>> On 07/06/2009 03:29 AM, Mick wrote: > >>>>> I cleared out some xorg related packages at some point because my > >>>>> machine will not shutdown without a kernel oops when it is trying to > >>>>> unload the radeon driver. > >>>> > >>>> Hm. There is no reason to unload any driver or module just to shut > >>>> down. Are you sure that's the reason for the oops? > >>> > >>> Well, no, but when I Ctrl+Alt+F1 to a console and shutdown from there > >>> there are no oops with the usual messages about > >>> "vblank_disable_fn+0x79/0xd0 [drm]" and "EIP: [<e0cc14fe>] > >>> radeon_get_vblank_counter+0x7e/0xd0 [radeon]". > >> > >> Wait, when exactly do you see the 'usual' oops messages? > > > > When I try to shutdown the machine they come up on the console - > > currently running 2.6.29-gentoo-r5. Typically, the oops occurs after the > > wireless driver and alsasound modules are unloaded and occasionally after > > syslog-ng is stopped... > > Okay, I know about the alsa modules stuff, but not the wireless. The alsa > modules should *not* be unloaded on shutdown because there is no reason to > do it and it causes problems if you try. (And I'm guessing that wireless is > the same.) > > Look at /etc/conf.d/alsasound. You should have two lines there as follows: > > # Deprecated options: > # Upstream feels, and we wholehartedly agree, that this was a silly idea > UNLOAD_ON_STOP="no" > KILLPROC_ON_STOP="no" > > When I'm more awake I'll take a look at wireless.
Thanks walt, but this problem is *not* related either to the usual alsasound culprit, or wireless. I have tried it with and without wireless and with and without unloading alsasound. All the errors are about radeon, nothing to do with anything else. As a side note, the new xorg server has caused 3d rendering (when I still had installed glxinfo and could compare speeds) to run like a dog. Until the 2.6.29 versions I never had kernel oops, or at least not without compiling something really inappropriate in the kernel. -- Regards, Mick
signature.asc
Description: This is a digitally signed message part.
