I reported this bug a long time ago. I had been hoping that by now some
update to Debian would contain a fixed Nouveau that would work as
expected with this old hardware. I'm still hoping ...
I have also been playing around with this myself -- I don't know enough
about video hardware (or anything about the Nouveau driver) to try to
meddle with the source, but I've changed configuration options in a
number of ways to see what would happen.
The first thing I should say is that I think the error lines in
Xorg.0.log are misleading. I had tried to disable acceleration in
Nouveau because it had caused problems in Stretch and earlier. The error
message lines:
[ 45.875] (EE) NOUVEAU(0): Error creating GPU channel: -19
[ 45.875] (EE) NOUVEAU(0): Error initialising acceleration. Falling
back to NoAccel
[ 45.875] (**) NOUVEAU(0): [COPY] acceleration disabled
seemed to me to be showing that the system was trying to use
acceleration. I have tried disabling acceleration by providing a kernel
parameter (either "noaccel" or "nouveau-noaccel=1") which I think worked
in Stretch, but this does NOT seem to disable acceleration in Buster,
which why I was getting the errors above. Those errors appear to have
masked the actual error, making it harder to diagnose the problem.
If I create a fairly minimal xorg.cong file containing just:
Section "Device"
Identifier "GeForce_6150"
Driver "nouveau"
VendorName "NVIDIA Corporation"
Option "NoAccel"
Option "AccelMethod" "None"
EndSection
Section "Monitor"
Identifier "Default monitor"
VendorName "DELL"
ModelName "U2410"
EndSection
Section "Screen"
Identifier "Screen0"
Device "GeForce_6150"
Monitor "Default monitor"
EndSection
then acceleration DOES seem to be disabled and I no longer get the
errors in Xorg.0.log.
In this state Buster boots to the login/greeting screen in the monitor's
default resolution of 1920x1200 and I can enter a username and password.
These are validated correctly and the screen blanks for a few moments
and returns to the login/greeter.
If I open a commandline console with ALT+F1 I can log in. If I then try
to start an X session with startx I get the following.
(Sorry, this is retyped by hand on another machine and may not be 100%
correct (and lines may have wrapped). Many of the original lines were
terminated with LF but not CR at the end and so the following lines
appeared indented.
X.Org X Server 1.20.4
X Protocol Version 11, Revision 0
Build Operating System: Linux 4.19.0-12-amd64 x86_64 Debian
Current Operating System: Linux welly-buster 4.19.0-16-amd64 #1 SMP
Debian 4.19.191-1 (2021-03-19) x86_64
Kernel command line: BOOT_IMAGE=/vmlinuz-4.19.0-16-amd64
root=UUID=4abd25ab-9425-42ee-8612-75c0164d13d5 ro noapic
Build Date: 01 December 2020 05:59:57PM
xorg-server 2:1.20.4-1+deb10u2 (https://www.debian.org/support)
Current version of pixman: 0.36.0
Before reporting problems, check http://wiki.x.org
to make sure you have the latest version.
Markers: (--) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warnoing, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/home/daniel/.local/share/xorg/Xorg.1.log", Time: Sat
Mar 27 23:32:08 2021
(==) Using config file: "/etc/X11/xorg.conf"
(==) Uing system config directory "/usr/share/X11/xorg.conf.d"
resize called 1920 1200
usr/lib/xorg/Xorg: symbol lookup error:
/usr/lib/xorg/modules/drivers/nouveau_drv.so: undefined symbol:
exaGetPixmapDriverPrivate
xinit: connection to X server lost
It now seems to me that the main problem is this undefined symbol error,
which is causing the xorg to fail to load.
I hope this is useful. Is there any more information I can provide to
help get this matter resolved?
--
Regards,
Daniel James
