Bug 155323

Summary:	radeon driver confuses outputs on Fujitsu Siemens Amilo 1667G
Product:	[openSUSE] SUSE LINUX 10.0	Reporter:	Jiri Dluhos <jdluhos>
Component:	X.Org	Assignee:	Stefan Dirsch <sndirsch>
Status:	RESOLVED FIXED	QA Contact:	Stefan Dirsch <sndirsch>
Severity:	Major
Priority:	P5 - None	CC:	eich, suse-beta
Version:	RC 4
Target Milestone:	---
Hardware:	All
OS:	Other
Whiteboard:
Found By:	Other	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	The modified xorg.conf. The Xorg.0.log with the original configuration (where nothing is seen on the screen). The Xorg.0.log with the modified configuration (with the added MonitorLayout line). Output of 'hwinfo --gfxcard'

Description Jiri Dluhos 2006-03-05 17:34:24 UTC

The radeon driver seems to confuse the primary/secondary output on Fujitsu Siemens Amilo 1667G notebook (Radeon X700). By default, it thinks the external output is primary, and sends all graphics data to it. If no external monitor is connected, the LCD monitor switches off right after starting X, tricking the user into thinking that the whole machine has crashed.

Workaround: add this line to the xorg.conf file:

Option "MonitorLayout" "LVDS"

After this, the radeon driver works fine (without 3d acceleration, of course).

I believe this same problem may happen also with other notebooks with the Radeon graphics device.

Comment 1 Stefan Dirsch 2006-03-05 17:46:42 UTC

Thanks. What about testing this on 10.1? We don't plan to investigate such issues for SUSE 10.0.

Comment 2 Jiri Dluhos 2006-03-05 18:00:16 UTC

Yup, I will try it on 10.1 as soon as I get some time to install it.

Leaving NEEDINFO until I get my hands on 10.1.

Comment 3 Jiri Dluhos 2006-03-06 12:05:57 UTC

Positive, it does exactly the same at SL10.1 Beta6, and the same workaround works.

When I installed it, I noticed that it offers "Dual Head" mode by default. I kept that default unchanged, trying to mimic what a common user would do. Maybe this is the culprit - but when I looked at the /etc/X11/xorg.conf file, I haven't found any traces of dualhead configuration at all, it looked exactly the same as the xorg.conf file from 10.0 (we could say not only it does not work but it also lies ;-) ).

Comment 4 Stefan Dirsch 2006-03-06 12:47:13 UTC

Could you attach xorg.conf and /var/log/Xorg.0.log with *and* without your additional MonitorLayout line? Thanks.

Comment 5 Jiri Dluhos 2006-03-06 13:37:23 UTC

Created attachment 71350 [details]
The modified xorg.conf.

This xorg.conf contains only a single change against the original, the single line with the "MonitorLayout" option.

Comment 6 Jiri Dluhos 2006-03-06 13:47:58 UTC

Created attachment 71356 [details]
The Xorg.0.log with the original configuration (where nothing is seen on the screen).

Comment 7 Jiri Dluhos 2006-03-06 13:48:52 UTC

Created attachment 71357 [details]
The Xorg.0.log with the modified configuration (with the added MonitorLayout line).

Comment 8 Jiri Dluhos 2006-03-06 13:57:34 UTC

By the way, when I looked at the difference between the logs, I have seen an interesting thing:

565a567,570
> (WW) RADEON(0): Invalid Monitor type specified for 2nd port
> (**) RADEON(0): MonitorLayout Option:
>       Monitor1--Type LVDS, Monitor2--Type `:v
                                           ^^^^^

That second monitor type string looks like it was read from a random portion of memory...

Comment 9 Stefan Dirsch 2006-03-06 14:07:57 UTC

I think you need to specify the Monitor type of the 2nd port as well. Try
"LVDS,NONE" instead.

Unfortunately the wrong Monitor type as Primary is deteteced when no option
is given. :-(

[...]
(II) RADEON(0): Primary:
 Monitor   -- CRT
[...]

So this is the problem.

Comment 10 Stefan Dirsch 2006-03-07 17:40:48 UTC

I wonder whether MonitorLayout "LVDS,CRT" works for you as well. Using such an option would be a valid workaround.

Comment 11 Jiri Dluhos 2006-03-10 21:31:32 UTC

Yes, I have tried it, it works. :-)

Comment 12 Stefan Dirsch 2006-03-10 21:35:50 UTC

Thanks. Could you also add the output of "hwinfo --gfxcard"? Thanks.

Comment 13 Jiri Dluhos 2006-03-13 08:14:48 UTC

Created attachment 72493 [details]
Output of 'hwinfo --gfxcard'

Please find the attached info.

Comment 14 Stefan Dirsch 2006-03-13 17:15:06 UTC

fixed for Beta > 8. Using the workaround mentioned above.