Bug#683167: Attempting to start X with intel driver hangs system
 

found 683167 linux/3.5-1~exerimental.1
quit

Hi Maik,

Maik Zumstrull wrote:

> [Subject: Also in 3.5]

Please keep in mind that these appear as emails in a crowded inbox, so the
subject line can be a good place to put valuable context.

> Now running linux-image-3.5-trunk-amd64 (3.5-1~experimental.1). Still happens.
[...]
> This is on an Ivy Bridge Xeon E3 with GPU. I thought it was the X
> driver, because i915 KMS works fine for the console, it only crashes
> when trying to bring up X. But I pulled in the 2.20.2 driver today and
> it still happens. So I now think the problem is in the kernel.
[...]
> 3.2 crashes as well, but not as hard as 3.4. With 3.2, X doesn't come
> up, but you can switch to a console and reboot. 3.4 needs the reset
> button.

Does the squeeze 2.6.32.y kernel reproduce the same trouble? (It should run
fine on a wheezy/sid system.) Could you retrieve full "dmesg" output
from the 3.2.y failure?

Thanks and hope that helps,
Jonathan


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/20120803190134.GA21793@copier

Bug#683167: Attempting to start X with intel driver hangs system
 

On Fri, Aug 3, 2012 at 9:01 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:

> Does the squeeze 2.6.32.y kernel reproduce the same trouble? (It should run
> fine on a wheezy/sid system.) Could you retrieve full "dmesg" output
> from the 3.2.y failure?

2.6.32 has never heard of this GPU, so the i915 module doesn't
initialize. The X server exits immediately, saying no device was
found, but doesn't crash. The kernel panics on trying to reboot, but
the stack trace shows a bunch of EFI-related functions, so it's
probably unrelated.

But let's make a note that yes, I installed with EFI boot, which is
probably a lot less tested than legacy BIOS boot. I haven't tried if
it is happier with BIOS boot, I could do that tomorrow if you think
it's important.

3.2 behavior is now similar: i915.ko doesn't know the GPU, and X just
exits because i915 isn't there. My last tests were with a 2.19 X
driver, presumably the update to 2.20 made the 3.2 crash go away.


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAO=zWDKYaZa85b5pztsKdL+TffhN4OKSJpVCdWVtnsJN=q91Y Q@mail.gmail.com

Bug#683167: Attempting to start X with intel driver hangs system
 

Maik Zumstrull wrote:

> 2.6.32 has never heard of this GPU, so the i915 module doesn't
> initialize. The X server exits immediately, saying no device was
> found, but doesn't crash. The kernel panics on trying to reboot, but
> the stack trace shows a bunch of EFI-related functions, so it's
> probably unrelated.

I see, thanks.

Support for this GPU seems to have been added by

cc22a938fc1d drm/i915: add Ivy Bridge GT2 Server entries

which was merged in 3.4-rc2.

[...]
> This is on an Ivy Bridge Xeon E3 with GPU. I thought it was the X
> driver, because i915 KMS works fine for the console, it only crashes
> when trying to bring up X. But I pulled in the 2.20.2 driver today and
> it still happens. So I now think the problem is in the kernel.

Can you get a trace of the crash using a serial console[1] or
netconsole[2]?

If so, please report this to bugs.freedesktop.org, following
instructions from "http://intellinuxgraphics.org/how_to_report_bug.html",
and let us know the bug number so we can track it.

Hope that helps,
Jonathan

[1] http://www.kernel.org/doc/Documentation/serial-console.txt
[2] http://www.kernel.org/doc/Documentation/networking/netconsole.txt


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/20120804080404.GA52729@Jonathans-MacBook-Air.local

Bug#683167: Attempting to start X with intel driver hangs system
 

On Sat, Aug 4, 2012 at 10:04 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

> Can you get a trace of the crash using a serial console[1] or
> netconsole[2]?

I got netconsole working, and some messages came through (I replugged
a USB device to trigger some), but nothing when it crashes. I don't
have the cabling here for serial (it's only a pinheader on this
board), but I could get it next week if that's what it takes.


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAO=zWDJntYQ3YBsXUyv5Ya+waaB_pGJym_BoBavXbL19noThY Q@mail.gmail.com

Bug#683167: Attempting to start X with intel driver hangs system
 

Maik Zumstrull wrote:
> On Sat, Aug 4, 2012 at 10:04 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

>> Can you get a trace of the crash using a serial console[1] or
>> netconsole[2]?
>
> I got netconsole working, and some messages came through (I replugged
> a USB device to trigger some), but nothing when it crashes.

Ok, thanks. Please attach a full log from netconsole from booting and
reproducing the problem. (It's ok if it doesn't say anything
interesting at the crucial moment.)

If you can get the corresponding Xorg log, that would be interesting,
too.

Hope that helps,
Jonathan


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/20120804202717.GD53341@Jonathans-MacBook-Air.local

Bug#683167: Attempting to start X with intel driver hangs system
 

On Fri, 2012-08-03 at 21:29 +0200, Maik Zumstrull wrote:
> On Fri, Aug 3, 2012 at 9:01 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:
>
> > Does the squeeze 2.6.32.y kernel reproduce the same trouble? (It should run
> > fine on a wheezy/sid system.) Could you retrieve full "dmesg" output
> > from the 3.2.y failure?
>
> 2.6.32 has never heard of this GPU, so the i915 module doesn't
> initialize. The X server exits immediately, saying no device was
> found, but doesn't crash. The kernel panics on trying to reboot, but
> the stack trace shows a bunch of EFI-related functions, so it's
> probably unrelated.
[...]

That's possibly the same bug as #626022. I have seen the same problem
on one of my own systems and investigated a little way, but didn't reach
a conclusion yet.

Ben.

--
Ben Hutchings
Experience is directly proportional to the value of equipment destroyed.
- Carolyn Scheppner

Bug#683167: Attempting to start X with intel driver hangs system
 

On Sat, Aug 4, 2012 at 10:27 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:

> Please attach a full log from netconsole from booting and
> reproducing the problem. (It's ok if it doesn't say anything
> interesting at the crucial moment.)

Done.

> If you can get the corresponding Xorg log, that would be interesting,
> too.

It doesn't seem to get around to writing one. The existing file was
for a session with the vesa driver.

Bug#683167: Attempting to start X with intel driver hangs system
 

Maik Zumstrull wrote:
> On Sat, Aug 4, 2012 at 10:27 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:

>> Please attach a full log from netconsole from booting and
>> reproducing the problem. (It's ok if it doesn't say anything
>> interesting at the crucial moment.)
>
> Done.

Thanks for a quick reply.

One more test and then we should take this upstream: does a boot
without attempting to start X work ok? Do you get a working
framebuffer console? You can try this by either choosing the
"recovery mode" option from the grub menu or by adding 'text' as an
extra word (no quotes) at the end of the kernel command line.

Either way, please report this upstream following instructions from

http://intellinuxgraphics.org/how_to_report_bug.html

and let us know the bug number so we can track it. The upstream
maintainers will want a kernel log with the boot option
'drm.debug=0x0e' appended as an extra word (no quotes) at the end of
the kernel command line.

Thanks again and hope that helps,
Jonathan


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/20120806003951.GB1376@Jonathans-MacBook-Air.local

Bug#683167: Attempting to start X with intel driver hangs system
 

On Mon, Aug 6, 2012 at 2:39 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

> One more test and then we should take this upstream: does a boot
> without attempting to start X work ok? Do you get a working
> framebuffer console?

Yes, that works fine.

> Either way, please report this upstream following instructions from
>
> http://intellinuxgraphics.org/how_to_report_bug.html
>
> and let us know the bug number so we can track it.

https://bugs.freedesktop.org/show_bug.cgi?id=53222

> The upstream maintainers will want a kernel log with the boot option
> 'drm.debug=0x0e' appended

Their doc says 0x06? Either way, will do that tomorrow or so.


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAO=zWDKUJ9WdJ1H4NcRi+ELFaz+hx_bJaeT2Zf+-kq_nahZb7A@mail.gmail.com

Bug#683167: Attempting to start X with intel driver hangs system
 

Maik Zumstrull wrote:
> On Mon, Aug 6, 2012 at 2:39 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

>> The upstream maintainers will want a kernel log with the boot option
>> 'drm.debug=0x0e' appended
>
> Their doc says 0x06? Either way, will do that tomorrow or so.

Thanks! 0xe == 0x6 | 0x8 prints a little more debug info which I
don't think is too important but it's what I've seen them ask for on
bugzilla lately.

Ciao,
Jonathan


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/20120807221333.GA180@mannheim-rule.local