Please test!
xf86-video-ati 6.11.0 is coming soon, please test ati git master and file bugs if you run into any problems.
xf86-video-ati 6.11.0 is coming soon, please test ati git master and file bugs if you run into any problems.
You must be logged in to post a comment.
February 12th, 2009 at 6:20 am
It’s working fine on my RV530 on a MacBook Pro.
There is however a regression over 6.10 (”OpenOffice toolbar button corruption [EXA, r500]”, caused by fa8e5a4fc236f8f15f462cb0d6164b194a65a118):
https://bugs.freedesktop.org/show_bug.cgi?id=19923
and a regression over 6.9 (”screen corruption while loading X”, caused by 346228b20c69a965c9e7d67157f42c1d36a5b6c3):
http://bugs.freedesktop.org/show_bug.cgi?id=18108
February 13th, 2009 at 1:19 pm
Hmm. Display hangs with no output on my HD4670 & all git checkouts as of today if I turn DRI on. With DRI off, display appears to work fine, but no XV. Is this the currently expected behaviour?
February 15th, 2009 at 3:45 pm
If you’re running master then DRI support and acceleration for 6xx/7xx are not there. It probably shouldn’t crash but acceleration is not expected to work on your card either.
The 6.11 release will come out of master so that’s definitely where agd5f is looking for test feedback.
If you are running the 6xx-7xx-support branches of radeon and drm then both EXA and Xv acceleration should work.
February 15th, 2009 at 5:49 pm
This was the 6xx-7xx-support branches of radeon and drm. I’ll have another prod when I have time.
February 22nd, 2009 at 8:09 am
master works fine, or at least the git master checkouts that have been dropped into Debian experimental do.
However, the r6xx-r7xx-support branch, with r6xx-r7xx builds of both git drm & xf86-video-ati leads to either:
without DRI: Working X, but no acceleration whatsoever: solid window moves are slow.
with DRI: display turns off after stating X. There’s nothing in the Xorg.log file after “(II) RADEON(0): Direct rendering enabled” appears. The Xserver quits, leaving the display non-operative.
The kernel says:
[drm] Initialized drm 1.1.0 20060810
No module found in object
radeon 0000:01:00.0: setting latency timer to 64
[drm] Initialized radeon 1.29.0 20080613 on minor 0
mtrr: no MTRR for d0000000,10000000 found
[drm] Setting GART location based on new memory map
[drm] Loading RV730 PFP Microcode
[drm] Loading RV730 CP Microcode
[drm] Resetting GPU
[drm] writeback test succeeded in 1 usecs
[drm] Resetting GPU
[drm] Setting GART location based on new memory map
[drm] Loading RV730 PFP Microcode
[drm] Loading RV730 CP Microcode
[drm] Resetting GPU
[drm] writeback test succeeded in 1 usecs
[drm] Resetting GPU
[drm] Setting GART location based on new memory map
[drm] Loading RV730 PFP Microcode
[drm] Loading RV730 CP Microcode
[drm] Resetting GPU
[drm] writeback test succeeded in 1 usecs
[drm] Resetting GPU
The machine is still alive: I can log into it, tell X to use the old driver by changing the module path & everything works fine.
Ah ha: I’ve got hold of the Xserver output, which finishes with:
/usr/bin/X: symbol lookup error: /usr/local/lib/xorg/modules/drivers//radeon_drv.so: undefined symbol: R600DrawInit
That’s weird: the function is defined in r600_exa.c
Will iivestigate…
February 22nd, 2009 at 8:21 am
Further investigation reveals that I’d failed to regenerate the makefiles after switching branches. Argh.
It would be nice if configure *noticed* that the Makefile.in files had all changes & it would make sense to regenerate the Makefiles.
Will try again and report back!
February 22nd, 2009 at 8:38 am
It works!
So PEBKAC. Apart from the aforementioned whinge about automake/autoconf & it’s a bit unfortunate that Xorg leaves the card in a non-working state on exit, because it doesn’t check for symbol availability when it loads radeon_drv.so
February 4th, 2010 at 8:02 pm
Hey…
Hi there, I don’t want to change this subject, but how exactly does one program a website like this one? Is it complicated? Where exactly do you begin?…
February 4th, 2010 at 9:14 pm
Joe…
Check out my domain sometime….
March 8th, 2010 at 8:26 pm
ati video card…
…