On 07/02/16 21:34, Peter Saisanas wrote:
Do you happen to have an xorg.conf? if so, please remove it.
Xorg should probe for your screens and not need an xorg.conf.
It shouldn't need any kernel parameters passed.... But stranger
things have happened!
Thanks for your help Peter
No. There's no xorg.conf
Please post your xorg.0.log, just curious to see if the edid is
read correctly.
Attached as a text file
FYI, I tried it with a Quadro FX4500 and a flashed GeForce 7800
and D.H. seemed to work. I used a Dell 24" LCD @ 1900x1200
together with an old Philips 20" @ 1680x1050 and it worked... For
me anyway.... I should have an old GeForce 6600 laying around. I
believe you mentioned this is GPU you have? If you like, I can try
it out when i get a spare moment as well to see if it works on my
end.
It may end up being something specific with the GeForce 6600.
That would be great but it seems strange it wouldn't just work,
particularly as both xfce and xrandr see both monitors. What is odd is
that in the xfce display settings there are options to activate and
make both monitors master and to discover monitors but the "apply"
button is greyed out.
Regards
Clive
Hi Clive,
Thought I should create a specific thread with your issue.
I just tested a spare GeForce 6600 I had laying around and the dual head
display behaviour is definitely not working.
Whereas with the Quadro FX4500 it works fine (xrandr and D.H.).
So it is an issue specifically affecting the GeForce 6600 (maybe others
as well).
As you described with the 6600, it seems to detect the second display
head, (i.e. it reads the EDID correctly) but there is no digital output
when connecting a second display via the DVI-I interface (digital).
I don't have a spare DVI-A to VGA adapter, but just curious if it may
work better with a VGA display on the second head?
I have been able to force a digital output on the second head by
appending the following to the kernel command line:
append="video=DVI-I-2:D"
However it then seems to think the second head is a DFP (LVDS) LCD panel
(i.e. a laptop).
Unfortunately, also using the mentioned appended kernel command, the
second head display is quite bad, (wrong colours, incorrect geometry and
graphical corruption). But the output does get enabled.
So in other words, I don't think it is an xrandr issue at all.
Perhaps the issue is is either the nouveau driver doesn't configure the
second head correctly for this particular GPU family (GeForce 6600 or
NV43) or the Apple OEM FCODE rom used in the GeForce 6600 is has
incorrect parameters configured in the DCB?
Link below to understand what the DCB is:
ftp://download.nvidia.com/open-gpu-doc/DCB/1/DCB-4.0-Specification.html
I know that others may say "but it works fine in OSX"...
However Apple has written their own drivers so they may have their own
workaround implemented directly in their own driver.
This is a bit of a shame as this particular GPU is the most common one
supplied with the Powermac G5 11.2 (DC) machines...
If the issue is the DCB in the FCODE rom, there may be a magical
appended kernel command to workaround the problem.
Otherwise, perhaps we can try to patch a GeForce 6600 GPU VGA BIOS and
load it from a file so nouveau uses this patched VGA BIOS instead of
extracting it from the original FCODE rom from the OF device tree.
Of course, this is on the assumption that the D.H. issue is indeed
because of the incorrect DCB parameters in the GeForce 6600 Apple OEM
FCODE rom and not nouveau specifically with NV43.
Any takers?
Regards,
Peter
