On Fri, Jun 16, 2006 at 11:18:37PM +0200, Michel Dänzer wrote:
> I suggest tracking this down from the other direction. r128_cce_depth
> corresponds to the DRM_R128_DEPTH ioctl, which is used by the
> r128{Read,Write}Depth{Pixels,Span}Locked functions in Mesa. Apparently,
> one of those functions can be called when the hardware lock isn't held
> under some circumstances.
If you can see the message I just posted, you'll that that appears to
be the problem. It's not just some circumstances, it's in fact *all*
circumstances that the functions are called without the lock.
Unfortunatly my OpenGL book is a few thousand Km away. I'm going to try
writing a short OpenGL program proving that any manipulation at all of
the depth buffer causes this error. I'm kind of amazed no-one has
noticed this before, although direct depth buffer manipulation is
probably not very common.
Have a nice day,
--
Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/
> From each according to his ability. To each according to his ability to litigate.
Attachment:
signature.asc
Description: Digital signature