Bug#720329: xserver-xorg-video-radeon: Screen keeping black at start but for the cursor

[Adam Smith]

Hi Michel

I sent most of this already to the Debian PowerPC mailing list so you
may have already seen it....

Michel Dänzer wrote:

>BTW, if you or anyone else having stability issues with radeon KMS in
>AGP mode could try if the attached kernel patch makes a difference, that
>would be interesting.

I tried the patch, but sadly it made no difference.  It still crashed
with agpmode=1 or 4.  This is on a G4 iBook mobility radeon 9200 (r200).

Interestingly, using the Mate desktop environment, KMS is noticeably
more stable.  It can take some time for a crash to happen.  Even when
running compiz 0.8.6 I can have 'the cube' (sad I know) working and
still no crash.  In contrast, an LXDE, XFCE or KDE environment often
causes an almost instant crash.

This rather mirrors a problem I have with UMS and EXA.  Unless I set my
FBTexPercent very low I often get corruption in application windows.
This can be just the window title bar, or toolbar or even the whole
window.  Even setting FBTexPerecent to zero I can still get this
corruption.  It's very annoying.  Again, LXDE (or XFCE, KDE) is very
prone to this problem (in particular PCManFM), but Mate doesn't seem so
badly effected and it is why I'm running it.  

I was wondering if the KMS crash and this UMS problem were linked in
some way since MATE seems to solve or lessen both problems?  

Attached is a screenshot of the UMS corruption.  It's a XFCE
environment, but the application is PCManFM.  The trash icon at the
bottom seems almost permanently corrupted.  The corruption can be
random pixels, or a repeated pattern.  Often it is resembles another
part or the screen stretched, shrunk, repeated etc.

I have access to another iBook (mobility radeon 9550 -r300) and this
has the same UMS corruption.  I've searched for existing bug reports
about it since surely it must be a common problem, but couldn't find
any.  I've not reported it before myself because I thought UMS wasn't
supported anymore by the radeon developers.  However, I've convinced
myself now that it is linked in some way to the KMS crashes...

I'm very keen to get KMS/UMS working properly on PowerPC, so if you
need any more patches testing then I will gladly compile them.

Btw, thanks for all the work you do with PowerPC!

Adam


-------------- next part --------------
A non-text attachment was scrubbed...
Name: corrupt2.png
Type: image/png
Size: 313811 bytes
Desc: not available
URL: <http://lists.x.org/archives/xorg-driver-ati/attachments/20140320/ee62dd3f/attachment-0001.png>