On Mon, 2002-12-16 at 17:24, Michel Dänzer wrote: > The DRM shouldn't make a difference for how X runs per se. You probably > also installed xserver-xfree86-dri-trunk?
No, I actually didn't. I'll have to double check the details below, but from memory: 1 radeonfb from 2.4.20-pre4-ben0 gave dark text console with many horizontal lines moving vertically (hard to describe, but video quality was poor, and fblevel was working backwards -- 1 was bright, 15 was dim). I was able to start X (using agpgart.o + radeon.o), but the horizontal lines were present in X, plus the right 1/5 or so of the screen had many more horizontal lines. 2 OF framebuffer from 2.4.20-pre4-ben0 gave good text console but X would not work do to lack of framebuffer device usable by X. 3 radeonfb from 2.4.20-ben1 (rsync benh kernel Sunday evening) gave good text console, but syslog reported unknown north bridge (instead of Apple Uni) when agpgart + radeon loaded. I _think_ startx led to blank screen at this point, but I was able to ssh in and reboot the iBook. (There was some problem after startx that led to doing 4 below, but I will have to double check exactly what is was.) 4 Replaced radeonfb from Dänzer's drm-trunk-module-src. syslog now reported Apple Uni North when agpgart + radeon loads, and X starts fine. Unfortunately, I realize I downgraded xserver-xfree86 from unstable to testing before I did this, so XFree86 was from testing at this point. Everything looked good (glxinfo output looked promising) but glxgears showed a red box and frame rates (at default window size) of 40 fps (it looked like the red box was perhaps flashing at high frequency). There were minor visible artifacts after closing glxgears, at which time I killed X and was back at an uncorrupted text console. Tonight I will try 4 with unstable XFree86 and then perhaps with xserver-xfree86-dri-trunk.
