On Tue, 15 Jul 2003, Mark Lane wrote:

>Date: Tue, 15 Jul 2003 15:20:11 -0600
>From: Mark Lane <[EMAIL PROTECTED]>
>To: [EMAIL PROTECTED]
>Reply-To: [EMAIL PROTECTED]
>Content-Type: text/plain; charset=us-ascii
>Subject: Opteron, PCI RADEON and PCIGART
>
>Anyone know why GinGin64 would have problems with Forcing PCIMode with a Radeon
>7000. The log shows it looping errors about RADEONCP.

Yes, because Radeon PCI DRI support is experimental only on
non-Alpha CPU architectures, so PCI is disabled by default.  ;o)

This requires you to use ForcePCIMode which is more or less a
nice way of saying "I understand that this is an experimental
feature, but please enable the experimental non-stable PCI DRI
code so I can try it anyway, and use it if it randomly happens to
work on my system."

The pcigart code is known to work for some people on x86, alpha, 
and ppc hardware currently.  It does not however work on AMD64 in 
my experience though, which implies that either there are issues 
needing debugging on AMD64 WRT pcigart - which is entirely a 
reasonable assumption, or else the fixes which allow pcigart to 
work on x86/ppc did not make it into the AMD64 kernel you are 
using.

I suspect that it is just a generic AMD64 issue with pcigart 
requiring someone to debug it who is interested in seeing it 
work.

Someone did this for x86 about 5 months back.  I'm interested in
seeing it work on AMD64 also, but it's unfortunately I don't 
have time to muck with it currently.  I planned on fixing AMD64 
MTRR support in XFree86 first before tackling the PCI radeon 
issue.  There's a bug report in Red Hat bugzilla for this problem 
if you'd like to add yourself to the CC list.  Also when I get a 
chance to poke at it, assuming it is still unsolved, please let 
me know if you're interested in doing some realtime 
troubleshooting and experimenting on #dri-devel on 
irc.freenode.net, as it is likely to be useful for narrowing 
things down.

My recommendation, is to try using the latest official DRM code 
from the DRI project's CVS.  Just the DRM, not all of DRI-CVS.  

If this works for you, please let me know.

Thanks for the report, hope this helps.
TTYL

-- 
Mike A. Harris


_______________________________________________
XFree86 mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xfree86

Reply via email to