Bug 113324

Summary:	radeon regression: external display no longer working in dual head configuration
Product:	[openSUSE] SUSE LINUX 10.0	Reporter:	Forgotten User OS1JNCFbCX <forgotten_OS1JNCFbCX>
Component:	X.Org	Assignee:	Matthias Hopf <mhopf>
Status:	RESOLVED WONTFIX	QA Contact:	Stefan Dirsch <sndirsch>
Severity:	Normal
Priority:	P3 - Medium	CC:	sndirsch
Version:	Beta 3
Target Milestone:	---
Hardware:	i686
OS:	SUSE Other
Whiteboard:
Found By:	Beta-Customer	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	used xorg.conf startup log

Description Forgotten User OS1JNCFbCX 2005-08-26 14:17:39 UTC

In 9.3 I did the following X configuration from within YaST on my Laptop: 
 
1600x1200 on the internal LCD for normal work 
 
1024x768 for an external video beamer for presentations 
 
Both did run in independent devices to allow me running other applications on 
the LCD the audience cannot see during the presentation. 
 
This does no longer work with new SaX2 because this produces a xorg.conf file 
using one device with MergedFB and a resolution of 2624x1200. Because of that 
the external viewport is 1024x1200 instead of 1024x768 and thus full screen 
modes of presentation tools show presentations in a way that the audience can 
no longer see the full slide because it is scrolling up and down all the time. 
 
I cannot see any way to tell new SaX2 to use two devices instead of MergedFB. 
 
I this a bug in SaX2, is there a way I did not see yet, or do I have to 
configure this feature manually in xorg.conf from now on?

Comment 1 Marcus Schaefer 2005-08-26 14:31:26 UTC

This is an intended behaviour. I think Matthias Hopf can tell you more  
about the possibilities of MergedFB. I assume you are running on a  
radeon based laptop because you are talking about MergedFB.  
  
To be able to use the old style you can call:  
  
    sax2 -b /usr/share/sax/profile/Radeon_DualHead 
 
but this is no longer supported actually

Comment 2 Matthias Hopf 2005-08-26 14:39:57 UTC

Marcus is right, with MergedFB the resolution of the common edge has to be the
same for both screens. If they are different, the driver automatically changes
it to a virtual desktop.

Please try to configure old style like in Comment 1, and reopen if this doesn't
work.

Comment 3 Forgotten User OS1JNCFbCX 2005-08-26 15:26:22 UTC

Thanks! 
 
With the above command SaX2 again creates a config file as expected. 
 
Unfortunately I have to make an X bug out of this again: The generated config 
does start up but an attached external monitor does not receive a signal.

Comment 4 Forgotten User OS1JNCFbCX 2005-08-26 15:27:31 UTC

Created attachment 47795 [details]
used xorg.conf

Comment 5 Forgotten User OS1JNCFbCX 2005-08-26 15:28:22 UTC

Created attachment 47796 [details]
startup log

Comment 6 Forgotten User OS1JNCFbCX 2005-08-26 15:30:32 UTC

This is a ThinkPad T42p with ATI Technologies Inc M10 NT [FireGL Mobility T2] 
(radeon driver). 
 
With 9.3 this was working. 
 
Do you need more information?

Comment 7 Stefan Dirsch 2005-08-26 15:38:12 UTC

Something for Matthias to investigate.

Comment 8 Matthias Hopf 2005-08-26 16:08:31 UTC

Will check...

Comment 9 Lenz Grimmer 2005-08-27 16:59:45 UTC

Duplicate of BUG #104640?

Comment 10 Forgotten User OS1JNCFbCX 2005-08-27 18:26:28 UTC

I don't think this is a real duplicate.  #104640 is mainly about switching
outputs.  This is not the problem described here.  My X server does everything
right but the output on the external connector is missing although according to
the logs it is set up.

But maybe the source of the problems is the same.

Comment 11 Matthias Hopf 2005-08-29 14:43:55 UTC

I'm not able to reproduce.

My T42p here has a RV350 NT, and whatever combination of device sections,
IgnoreEDID, NoDDC, MonitorLayout I try I always get an

  (WW) RADEON(1): Only one monitor detected, Second screen will NOT be created

With 9.3 I only get a broken cloned signal on the external monitor, with 10.0b3
I get a black screen on the external monitor. I tried this both with our current
beta and with SL9.3, no difference. Again, these seem to be different T42p :-(((

The only thing that works with this T42p is the new mergedFB mode.

Last hope: please try to use the radeon_drv.o from 9.3 with the current beta,
the ABI has remained stable, and I know of people who have successfully tried
this. Other than that I have to close this as WONTFIX, as I cannot change
actively anything on it, and I cannot reproduce this.

When more developer information is available for radeon chips, the output plug
detection/management routines have to be rewritten, they are pretty broken, and
it doesn't seem to get better with newer versions.