Created attachment 30989 [details]
Xorg.0.log for radeonhd driver without UseAtomBIOS, or AtomBIOS pll=on.

I originally wrote to the list and this is the text and replies from that.

 > I have a gigabyte GA-MA690G-S3H motherboard with the AMD 690G chipset.
 > Until recently I was using the VGA output connected to a 1280x1024 CRT and
 > the DVI_D output connected to a 1920x1200 LCD, this worked fine.  I now
 > have two 1920x1200 LCDs one connected to the VGA and one to the DVI_D,
 > however, I have been unable to get the VGA to display at 1920x1200.
 > Instead everything in software on the box thinks the VGA is at 1920x1200
 > (xrandr, and X start up logs), however, the screen is actually at
 > 1600x1200.  The radeon (NOT HD) driver seemed to work a little better in
 > that both screens were put in the same mode, however, the bottom screen in
 > the setup would always get corruption.  I am now using the ati driver which
 > seems to work fine for this setup.
 >
 > I am not sure how much testing I can do since this PC is used daily by
 > myself and my wife and taking down X and restarting it would be a real
 > pain.  I might be able to do some testing with a second X server running on
 > a different VT if people think that won't destroy my current setup.
 >
 > I thought that maybe I am doing something wrong and if not then maybe
 > someone on the list can test this setup and confirm it and then raise a bug
 > if they can.

Egbert Wrote:

 > Please provide a a verbose log file. To obtain this start X from the
 > console with: 'X -logverbose 7' and terminate it with ctrl-alt-backspace.

I Wrote:

 > Sorry it's been so long, I have read some of the posts on this list
 > and tried the "UseAtomBios" setting and this then fixes the problem.
 > I can if you want still produce the "X -logverbose 7" log file, but
 > thought I should let you know that I have a solution.

Matthias Hopf Wrote:

 > Can you try to use the undocumented
 > 
 >   Option "AtomBios" "pll=on"
 > 
 > *instead* of the "UseAtomBios" option? This would help to narrow down
 > the issue.

It's been a long time since then and I am sorry for not getting back sooner, however, it looks like you guys have been doing some great work, so the time wasn't wasted :).

Using xf86-video-radeonhd version 1.3.0 I still have the same problem without the UseAtomBios setting.  The files radeonhd.log and radeonhd.xorg.conf attached are the Xorg.0.log and xorg.conf used for that test run, logverbose was set to 7.

Setting AtomBios pll=on as suggested by Matthias does NOT solve the problem, in fact the results are the same. The files radeonhd_pll.log and radeonhd_pll.xorg.conf are the Xorg.0.log and xorg.conf used for that test run.

OK, so now the confession, what motivation my return to looking at this problem ?  Well, UseAtomBIOS is no longer working well, the screens have the correct resolution, however, the LCD connected via DVI blanks for a second far to regularly, sometimes many times a minute, it a very annoying.  When browsing the web some pages are unviewable as the screen is constantly blank with the occasional flash of the page.  I also sometimes get image corruption with this, the right eighth of the screen overlaps onto the left eighth of the screen, with the image being ghosted so that it is full strength at the top and not at all at the bottom. (kind of hard to describe), I guess I could take a photo of it if you want.

I have experimented with this a little and found that I still get blanking, but a lot less, at 1680x1050, at 1280x1024 everything seems fine.  Also if I setup xorg.conf so that the same image is displayed on startup on both monitors, then everything is fine.  If I then use Xrandr to make them different, the problem comes back, using xrandr to make them the same again, doesn't fix the problem.  I ran this test (although maybe not the last bit or using xrandr to make them the same) with logverbose 7, the files radeonhd_atombios.log and radeonhd_atombios.xorg.conf are the Xorg.0.log and xorg.conf used for that test run.

I am now in a position to do some a little testing in the evenings, so please, if you want me to try anything let me know.  Also I understand that you might not be so interested in the "UseAtomBIOS" problem, that's fine as long as we get the driver working correctly in the other case.