[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Failure to reach login screen - nouveau problem



On 02/04/16 08:50, Sven Joachim wrote:
> On 2016-04-02 06:51 +0300, Adam Wilson wrote:
> 
>> On Fri, 01 Apr 2016 19:58:17 +0200
>> Sven Joachim <svenjoac@gmx.de> wrote:
>>
>>> On 2016-04-01 13:40 -0400, Felix Miata wrote:
>>>
>>>> Peter Hillier-Brook composed on 2016-04-01 15:14 (UTC+0100):
>>>>
>>>>> Following an overnight shut down I couldn't reach the desktop manager
>>>>> the following day. Running a live system I was able to extract the
>>>>> following via 'dmesg'. Does anyone recognise this, or can offer a
>>>>> pointer to the problem?
>>>>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>>> [    0.705479] nouveau E[  DEVICE][0000:01:00.0] unknown chipset, 0x124020a1
>>>>> [    0.705515] nouveau E[     DRM] failed to create 0x80000080, -22
>>>>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>
>>> Did you upgrade your kernel prior to the reboot?

No. 'only' a few libs

>>>> I saw something like this on Mageia in just the past few days. The DRM
>>>> and Nouveau modules were failing to load automatically during boot. I
>>>> got X to work by modprobing nouveau and/or drm.
>>>>
>>>> What does 'lspci | grep VGA' show your gfxcard model is?
>>>
>>> If we can trust Google with the error message, it's a shiny GTX 970.

Spot on.

>>>> Possibly you have a firmware package installation failure.
>>>
>>> There is no firmware for this card in Debian yet
> 
> To clarify, the card should work without the firmware, but it only
> provides basic operation.  There's no acceleration without the firmware
> and Linux >= 4.6-rc1.
> 
>> This makes no sense. Unless OP has changed his hardware overnight,
>> then this shouldn't be happening, since it worked fine before.

No hardware changes and, as mentioned above, only a few libs via
Synaptic. I'm still trying to isolate the suspect lib.

> The hardware did presumably not change overnight, but the kernel might
> as well have.  You know, new kernels sometimes introduce bugs.

Vanilla Debian 8.0 with all updates to the date of my first post, so now
at 8.4. No new kernels. The reason I switched back to vanilla Debian was
to avoid the fairly frequent kernel changes in Kubuntu.

>> It seems we can't trust Google with anything.
> 
> We can't trust it with everything, but this time it gave some good
> results.  Several people seemed to have the OP's problem on various
> distributions.

I'm now back in business via the on-board Intel graphics, once I
persuaded kdm to start via "sudo /etc/init.d/kdm force-reload". After
the first boot this process became automatic again.

Thanks to all who offered help and advice.

Peter HB


Attachment: signature.asc
Description: OpenPGP digital signature


Reply to: