Re: Some annoying problems with ATi Radeon 7200!
Michel Dänzer wrote:
>>I configed it more than one time and I can't remember that debconf asked
>>me about FBDev...
> Maybe it's a low priority question, otherwise it's probably autodetected
> from /proc/fb .
Could be, but my /proc/fb is empty.
>>However, I have serious problems with the dri packags:
>>I can't login anymore, there is a problem with my .xsession; from
>>Xlib: connection to ":0.0" refused by server
>>Xlib: Protocol not supported by server
>>xrdb: Can't open display ':0'
>>Should I set some permissions somewhere?
>>Also, my xdm login screen looks entirely different, but I guess that is
>>normal... (It looks a lot more ugly, by the way... ;-)
> In the next version,
> /usr/share/doc/xserver-xfree86-dri-trunk/README.Debian will contain:
> * xdm: in /etc/X11/xdm/xdm-config, set
> DisplayManager*authorize: false
> Does this help?
It seems to!
At least, I could log in now. At first I thought I did something wrong,
but the xdm login screen looks normal now, not white anymore!
Anyway, there is still a big problem though, my 3D seems broken:
Radeon DRI driver:
Compatibility mode for DRM driver version 1.1.1
TCL will be disabled, expect reduced performance
(prefer DRM radeon.o 1.3.x or newer)
disabling TCL support
glxgears: radeon_ioctl.h:168: radeonAllocCmdBuf: Assertion
`rmesa->dri.drmMinor >= 3' failed.
The XFree86.0.log seems just fine. (Yep, it starts with
XFree86 Version 4.2.0 (DRI trunk) / X Window System
There are no errors there.
By the way, I saw this:
(II) RADEON(0): AGP Fast Write disabled by default
What is AGP Fast Write? I saw in my BIOS an option to turn this on. Is
One very good thing: it seems that the ghosting is a bit less now! I
couldn't test with openGL programs (which caused the worst ghosting
effects), but it seems that it is a bit less now... (dragging windows
around is still causing some of the ghosting though...))
>>How can I tell what depth?
> xdpyinfo|grep 'depth of root'
It's 24 planes.
> BTW, I just recall experiencing color problems on the 7200 in the gf's
> Cube. Seems to be a radeonfb glitch which happens intermittently and
Well, I don't need fb anyway.
> I don't see anything unusual in the log. Does Option "HWcursor" make the
> problem go away?
Do I have to put that in the driver section?
I did so, but:
(II) RADEON(0): Using hardware cursor (scanline 4808)
(WW) RADEON(0): Option "HWcursor" is not used
According to my XFree86 log.
Note that the first line was also already in my original log, with the
4.2.1. drivers from Debian.
PS: MSX FOR EVER! (Questions? http://faq.msxnet.org/ )
PPS: Visit my homepage at http://manuel.msxnet.org/