I've got a 9250 running at a screen resolution of 1280x1024. Driver 6.7.196 has regressed with respect to 6.7.195 in that mplayer 1.0rc2 using the Xv video out (and probably other Xv-using programs, not tested) now displays videos in an upper-left-aligned slice of the screen I estimate to be 640x480, with the rest filled with whatever was there before (blue, black, junk, whatever). In 6.7.195, the image is rescaled correctly and fills the whole screen. I haven't dug deep enough to see what's going wrong yet, just deep enough to see that this is a behaviour change which correlates with ATI driver upgrade and thus plausibly its bug. (There is no useful difference in xvinfo output or in what mplayer dumps to stdout between the ATI driver versions.) I'm sorry this bug report is so short on specifics: if you know something I can do to provide more, please say.
Not sure what could have caused this by a quick look at the git log, but you could always use git-bisect to identify which commit caused it.
please attach your xorg log and config.
Created attachment 12673 [details] Log of working Xv boot, 6.7.195
Created attachment 12674 [details] Log of broken Xv boot, 6.7.196 Possibly significant, from the diff: drmOpenDevice: node name is /dev/dri/card0 -drmOpenDevice: open result is 7, (OK) +drmOpenDevice: open result is -1, (No such device or address) +drmOpenDevice: open result is -1, (No such device or address) +drmOpenDevice: Open failed This is all with the same kernel, 2.6.23.1.
Bisected. Bad commit: commit 5db3afaa1fdb69d382ac769ef40191a4b964d28e Author: Alex Deucher <alex@t41p.hsd1.va.comcast.net> Date: Thu Oct 11 20:25:20 2007 -0400 RADEON: return status unknown for flaky chip/connector combinations This should at least get something on the screen. XPRESS chips, I'm looking in your general direction... It is... unclear to me why on earth this should break Xv this way, but it does. (We already know from the logs that the card being detected is unchanged in my case.)
does: xrandr --output DVI-0 --off fix it?
That does indeed fix it. xrandr shows differences between the working and failing drivers: -DVI-0 disconnected (normal left inverted right x axis y axis) +DVI-0 unknown connection 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm + 1024x768 60.0* + 800x600 60.3 + 640x480 59.9 That second connection is a) the wrong size and b) is definitely physically disconnected and c) probably shouldn't be being used as the source of sizing information in any case, should it? If you have one connected device, and one 'unknown', surely the video overlay should be sized for the device that's known to be connected? (I might have spotted this myself, only I didn't even have xrandr installed until just now; it's not as though this CRT can be reflected or rotated as far as I know, nor do I have a driving desire to do such things with it...)
Can you try again with my latest commits in ati git master? I've limited the unknown status to XPRESS chips.
Yep, that's fixed it: it also fixed an apparently-related bug that cropped up at some point whereby other full-screen apps not using video overlays (e.g. wesnoth, xscreensaver) got less of the screen than they thought they had. Two bugs with one stone :) (Marking FIXED: I hope that's not a problem)
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.