[Intel-gfx] i945GME support

[4ernov]

Hello, Chris and list,

I continue working under my problem with two LVDS monitors connected
to SDVO-B and SDVO-C. I finally managed to make one monitor work again
using generic kernel version 2.6.36 and xf86-video-intel drivers
version 2.12.0. Unfortunately, either with the latest unstable
versions or 2.14.0 stable version nothing seems to work and the
problems are identical to what was discussed before (thousands of
"(EE) intel(0): Failed to submit batch buffer, expect rendering
corruption or even a frozen display: Invalid argument." messages).

The hardware retailer also reported that they got to work both
monitors with IEGD drivers under Debian 5.0 (which contains Xserver
1.6.x). I've checked that it's true and the xorg.conf is attached.
Obviously, it was generated by IEGD GUI application. Is it possible to
make some finetuning for SDVO part using xf86-video-intel drivers?
Maybe the problem is with settings?

And the main question remaining: how both monitors can be enabled in
KMS-enabled console? I think they should show the same console but
only one monitor (SDVO-B) is on and another is off.

I hope that someone could help me as that's so hard problem for me and
I see no ways but to rollback to Xserver 1.6.0 and IEGD drivers which
is highly undesired option. Any information or help would be greatly
appreciated.

Thanks in advance,
Alexey

2011/1/21 4ernov <4ernov at gmail.com>:
> Well, I'm still experimenting with my setup. There're some interesting things:
> 1. It works under Windows XP with IEGD drivers, both monitors are working.
> 2. I found that I can't make one monitor working in X anymore, with any kernel.
>
> But under KMS in console the first monitor seems to work properly. So,
> is there also any way to get two monitors working in console? As far
> as I know, they should show the same picture in console. Maybe
> there're some settings to change?
>
> 2011/1/12 4ernov <4ernov at gmail.com>:
>> Hmm.. yes, dmesg is definitely printed not complete. I've copied the
>> longer log of the last boot from /var/log/syslog, perhaps it helps.
>> Yes, you're right, TV seems to be spurious, there're no connectors and
>> also it wasn't claimed.
>> /sys/kernel/debug/dri/0/i915_error_state seems to detect no errors, it is:
>> no error state collected
>>
>> Is it possible to know the problem why the GPU is wedged somehow?
>> Maybe it's some hardware fault?
>>
>> 2011/1/12 Chris Wilson <chris at chris-wilson.co.uk>:
>>> On Tue, 11 Jan 2011 19:39:03 +0300, 4ernov <4ernov at gmail.com> wrote:
>>>> Hello, Chris,
>>>>
>>>> I've finally managed to install the upstream drm-intel kernel module
>>>> and xf86-video-intel Xorg driver and test the system with parameters
>>>> you mentioned. So, dmesg and Xorg logs are attached.
>>>> I also got additional information concerning connections: according to
>>>> it, two LVDS outputs are connected through SDVOb and SDVOc channels.
>>>> And the symphoms on the fresh drivers are the same: one monitor
>>>> (SDVOb) works successfully in console mode using KMS, Xorg attempts to
>>>> start at the begginning but fails, the second monitor doesn't work.
>>>>
>>>> Any help or information would be greatly appreciated.
>>>
>>> That's not the complete dmesg from boot. The Xorg.log indicates that the
>>> source of the problem (if we ignore the spurious? TV detection) to be the
>>> GPU is wedged. However, the dmesg makes no mention of this.
>>>
>>> Is there an error recorded in /sys/kernel/debug/dri/0/i915_error_state?
>>> -Chris
>>>
>>> --
>>> Chris Wilson, Intel Open Source Technology Centre
>>>
>>
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: xorg.conf
Type: application/octet-stream
Size: 8796 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20110207/b4fd37ef/attachment.obj>