ubuntu-x-swat team mailing list archive
-
ubuntu-x-swat team
-
Mailing list archive
-
Message #113434
Re: [Bug 732089] [NEW] [gm45] GPU lockup dbe8d52d (EIR: 0x00000010 ESR: 0x00000010 PGTBL_ER: 0x00000001 IPEHR: 0x01000000)
On Wed, Mar 09, 2011 at 04:33:35PM -0000, Robbie Williamson wrote:
> Public bug reported:
>
> Binary package hint: xserver-xorg-video-intel
>
> Wasn't doing anything in particular, the message just popped up telling
> me I had a system error...it's been happening everytime I log in, I just
> finally decided to report it. I don't notice any odd behavior, just the
> message.
Hi Robbie,
Believe it or not I love hearing what you just said...
(And I think we may have a fix!)
This is an extremely commonly reported bug (there's gobs of dupes),
however it seems difficult to locate someone who can reproduce it
reliably, as it sounds like you can.
Basically, because of wanting fast/pretty boot with plymouth and all
that, the kernel loads up the generic vesafb framebuffer driver early
on, then switches to a better driver later (the intel kernel drm driver
in this case). I gather that the plumbing layer bits can try to bring
up this intel driver before vesafb has been unloaded. And that makes
the GPU unhappy and it locks up briefly; however it's able to reset the
GPU and keep on chugging (at least, most of the time).
Or something like that.
Anyway, the good news is the kernel team has posted a possible fix for
this, which adds checks throughout the plumbing layer to prevent the
kernel drivers from colliding during boot. This patches gdm, linux,
module-init-tools, plymouth, and udev (no X packages are involved in
this bug afaik).
When the PPA's kernel has finished building, I'll post a note to the
master bug (#702090). I would really appreciate it if you could load
that PPA on your machine and do whatever you normally do to see the bug,
and verify that it no longer occurs with this PPA.
References