Hello!
I'm running the latest wireless-testing.git kernel, which is 2.6.30-rc1
plus some wireless bits. I use intelfb with kernel mode switching and
X11 using fbdev. This runs on Fedora 10, fbdev driver is
xorg-x11-drv-fbdev-0.3.1-7.fc9.x86_64.
This worked fine with Linux 2.6.29, but after upgrading to 2.6.30-rc1
X11 hangs hard.
Bisecting found commit 66c1ca019078220dc1bf968f2bb18421100ef147 to be
responsible. It can be reverted on top of 2.6.30-rc1, and the resulting
kernel is working fine. The commit is called:
fbmem: fix fb_info->lock and mm->mmap_sem circular locking dependency
I'm especially concerned to see [email protected] in the cc in the log.
This patch fails on two systems for me, so it's hardly ready for stable
kernels.
The kernel config (very minimal, I used it to speed up bisecting),
xorg.conf and lspci output are attached.
--
Regards,
Pavel Roskin
On Thu, Apr 09, 2009 at 06:02:04PM -0400, Pavel Roskin wrote:
> Hello!
>
> I'm running the latest wireless-testing.git kernel, which is 2.6.30-rc1
> plus some wireless bits. I use intelfb with kernel mode switching and
> X11 using fbdev. This runs on Fedora 10, fbdev driver is
> xorg-x11-drv-fbdev-0.3.1-7.fc9.x86_64.
>
> This worked fine with Linux 2.6.29, but after upgrading to 2.6.30-rc1
> X11 hangs hard.
>
> Bisecting found commit 66c1ca019078220dc1bf968f2bb18421100ef147 to be
> responsible. It can be reverted on top of 2.6.30-rc1, and the resulting
> kernel is working fine. The commit is called:
>
> fbmem: fix fb_info->lock and mm->mmap_sem circular locking dependency
>
> I'm especially concerned to see [email protected] in the cc in the log.
> This patch fails on two systems for me, so it's hardly ready for stable
> kernels.
>
> The kernel config (very minimal, I used it to speed up bisecting),
> xorg.conf and lspci output are attached.
Thanks for reporting Pavel. I agree to revert 66c1ca ASAP.
See also http://lkml.org/lkml/2009/4/9/177.
-Andrea
* Andrea Righi ([email protected]) wrote:
> Thanks for reporting Pavel. I agree to revert 66c1ca ASAP.
Rather than dropping from -stable queue, looks like we should be adding:
513adb58685615b0b1d47a3f0d40f5352beff189 "fbdev: fix info->lock deadlock
in fbcon_event_notify()"?
thanks,
-chris
On Tue, Apr 21, 2009 at 06:34:24PM +0000, Chris Wright wrote:
> * Andrea Righi ([email protected]) wrote:
> > Thanks for reporting Pavel. I agree to revert 66c1ca ASAP.
>
> Rather than dropping from -stable queue, looks like we should be adding:
> 513adb58685615b0b1d47a3f0d40f5352beff189 "fbdev: fix info->lock deadlock
> in fbcon_event_notify()"?
>
> thanks,
> -chris
Correct.
Both Andreas and Krzysztof (in cc) tested this and reported that it
fixes the issue (see also http://patchwork.kernel.org/patch/17753/).
Thanks,
-Andrea
* Andrea Righi ([email protected]) wrote:
> On Tue, Apr 21, 2009 at 06:34:24PM +0000, Chris Wright wrote:
> > * Andrea Righi ([email protected]) wrote:
> > > Thanks for reporting Pavel. I agree to revert 66c1ca ASAP.
> >
> > Rather than dropping from -stable queue, looks like we should be adding:
> > 513adb58685615b0b1d47a3f0d40f5352beff189 "fbdev: fix info->lock deadlock
> > in fbcon_event_notify()"?
>
> Correct.
>
> Both Andreas and Krzysztof (in cc) tested this and reported that it
> fixes the issue (see also http://patchwork.kernel.org/patch/17753/).
queued. thanks,
-chris