On Thu, 2013-02-28 at 11:31 +0100, Bjørn Mork wrote:
> Dale Schroeder <dale@BriannasSaladDressing.com> writes:
>
> > I've been able to narrow this down a bit more. One of my other
> > systems still had 3.2.0-3-686-pae (3.2.23-1) in its apt archives. This
> > kernel also boots successfully and does not hang at mdm startup. So,
> > the problem did not exist in any of the 3.2.0-3 series released to
> > Wheezy, but was introduced sometime between this image and 3.2.32-1,
> > the 1st of the 3.2.0-4 series released to Wheezy.
>
>
> I note that the changelog for
>
> linux (3.2.29-1) unstable; urgency=low
>
> includes
>
> * [x86] drm/i915: Fix i8xx interrupt handling (Closes: #655152)
>
> which is extremely suspiscious in this context. I wonder if anyone
> experiencing this bug has tried reverting this patch?:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=59;filename=drm-i915-i8xx-interrupt-handler.patch;att=1;bug=655152
>
> Note that it is another shot in the dark - I have absolutely no idea
> what's going on here. But I have a feeling that patch is replacing an
> annoying bug on one platform with a critical bug on another.
>
> Not sure that is a good tradeoff...
In fact the same chip is involved in both initial bug reports:
#655152:
00:02.0 VGA compatible controller [0300]: Intel Corporation 82865G Integrated Graphics Controller [8086:2572] (rev 02) (prog-if 00 [VGA controller])
#692607:
00:02.0 VGA compatible controller [0300]: Intel Corporation 82865G Integrated Graphics Controller [8086:2572] (rev 02) (prog-if 00 [VGA controller])
[...]
> Anyone able to spot the missing INIT_WORK()'s? Based on the
> I915_HAS_HOTPLUG(dev) test, I assume that leaving the first one out was
> intentional. But the second one cannot be left out, as demonstrated by
> these bug reports.
Agreed. And this is a faithful backport of the upstream change, i.e.
the bug was not introduced in backporting. However it seems to have
been almost immediately fixed upstream by:
commit 8b2e326dc7c5aa6952c88656d04d0d81fd85a6f8
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Apr 24 22:59:41 2012 +0100
drm/i915: Unconditionally initialise the interrupt workers
> I am attaching a proposed fix on top of the 655152 patch, which I have
> not tested at all on actual Debian kernel sources. Might need context
> adjustments. I'd appreciate it if anyone with crashing hardware could
> test it.
It applies and builds.
Instructions for building a patched package are at:
http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official
Ben.
--
Ben Hutchings
Time is nature's way of making sure that everything doesn't happen at once.
Attachment:
signature.asc
Description: This is a digitally signed message part