Radeon HD 6530D crashing Xorg with a CentOS 6.6 kernel

[Darby Vicker]

On Sat, Jan 3, 2015 at 11:10 PM, Darby Vicker <darby.vicker at gmail.com> wrote:
> On Wed, Dec 24, 2014 at 7:58 PM, Michel Dänzer <michel at daenzer.net> wrote:
>> On 24.12.2014 22:26, Darby Vicker wrote:
>>> On Wed, Dec 24, 2014 at 12:50 AM, Michel Dänzer <michel at daenzer.net> wrote:
>>>> On 24.12.2014 14:18, Darby Vicker wrote:
>>>>> On Tue, Dec 23, 2014 at 9:02 PM, Michel Dänzer <michel at daenzer.net> wrote:
>>>>>> On 23.12.2014 14:11, Darby Vicker wrote:
>>>>>>> Hello,
>>>>>>>
>>>>>>> Sorry for the cross post but I figure this is a better place to get
>>>>>>> help.  With the Xorg ABI update in CentOS/RHEL 6.6 [1] I can't get X
>>>>>>> to come up properly.  More info in this thread on the CentOS list:
>>>>>>>
>>>>>>> http://lists.centos.org/pipermail/centos/2014-December/148341.html
>>>>>>>
>>>>>>> To summarize:
>>>>>>>
>>>>>>> * Stock ATI drivers shipped with CentOS crash X with a 6.6 kernel
>>>>>>> * fglrx drivers also crash X with a 6.6 kernel[2]
>>>>>>> * a centos 6.6 livecd still exibits the problem (trying to rule out an
>>>>>>> issue with my specific installation)
>>>>>>>
>>>>>>> I have Xorg.0.log files for boots from the 4 combinations of boots
>>>>>>> from a CentOS 6.5 and 6.6 kernel and the stock and fglrx drivers.
>>>>>>> Nothing obvious stands out to me but I could definitely be missing
>>>>>>> something.  What's the best way to send the logs - just inline?
>>>>>>
>>>>>> Please attach the *.stock log files.
>>>>>
>>>>> Sure - attached.
>>>>
>>>> I can't see any graphics driver related problem in
>>>> Xorg.0.log.centos6.6.stock.gz; the X server seems to start up and shut
>>>> down cleanly.
>>>>
>>>> Why do you think the graphics drivers crash? Maybe the problem is on the
>>>> client side, e.g. related to gdm or the login manager you're using.
>>>
>>> I should be more specific about the problem.  The problem occurs at
>>> the point where the boot splash screen transitions to when (I assume)
>>> X starts and the login screen comes up.  When I boot the CentOS 6.6
>>> kernel, the screen just goes black and shortly after that the monitor
>>> goes to sleep with a "No signal" message.  The virtual terminals (e.g.
>>> cntl-alt-F2) don't work at that point either.  I was assuming that
>>> since the Xorg ABI has changed between the CentOS 6.5 and 6.6 release,
>>> it was very likely that the problem was related to Xorg.
>>
>> If there was a Xorg ABI issue, you could see in the Xorg log file that
>> the driver fails to load.
>>
>>
>>> Do you have any suggestions for isolating the problem?  I'm not sure what else to
>>> try.  I did also try a boot to single user mode with the 6.6 kernel
>>> and then did a "startx".  My thought was to try to start X with
>>> minimal other services started to see if something else was
>>> interfering with X.  But the same thing happens - the screen goes
>>> blank and the monitor goes to sleep.  Any other suggestions are
>>> welcomed.
>>
>> If you boot the 6.5 kernel, does it still work with the same userspace
>> that shows the problem with the 6.6 kernel? If so, it sounds like
>> there's just some regression in the 6.6 kernel. Maybe there's something
>> in dmesg, or maybe you can try isolating the change between the 6.5 and
>> 6.6 kernels which caused the regression.
>>
>>
>> --
>> Earthling Michel Dänzer               |               http://www.amd.com
>> Libre software enthusiast             |             Mesa and X developer
>
> dmesg output attached for both.  These are the differences that stand out to me:
>
>
> 1) Its allocating a different amount of memory.
>
> 1005c974
> < radeon 0000:00:01.0: GTT: 512M 0x0000000020000000 - 0x000000003FFFFFFF
> ---
>> radeon 0000:00:01.0: GTT: 1024M 0x0000000020000000 - 0x000000005FFFFFFF
>
>
> 2) different driver version
>
> 1060,1062c1040,1042
> < [drm] Initialized radeon 2.30.0 20080528 for 0000:00:01.0 on minor 0
> ---
>> [drm] Initialized radeon 2.37.0 20080528 for 0000:00:01.0 on minor 0
>
>
>
> I'd be interested in your take on this though.

I was able to verify that this messages appears in :0-greeter.log for
a failed GUI start up (i.e. CentOS 6.6 boot) and doesn't appear in a
6.5 boot:

gnome-settings-daemon: Fatal IO error 11 (Resource temporarily
unavailable) on X server :0.

Do you know if there is a way to bypass gdm to see if that is the
problem?  For example, is there an alternative to GDM or some way to
try to start the X-session directly?