I've been trying to track down a bug in the r128 driver causing lots of the following messages in the kernel logs:
r128_cce_depth called without lock held The problem appears (to me anyway) to be in Mesa-6.4.2/src/mesa/drivers/dri/r128/r128_span.c. There are a few macro definitions there that disable locking, yet the code clearly calls r128WriteDepthSpanLocked() which requires locking. It would seem the fix is to undef HW_LOCK and HW_UNLOCK before the depth buffer stuff so the locking occurs as normal. Any thoughts? I've tried to excercise the code by doing a glReadPixels() on the depth buffer, but it doesn't appear to trigger it. So far, the only way to reliably trigger it is using google-earth. Any thoughts on how to create a simple test case? I know it's not a kernel issue, since the same bug manifests itself on both 2.4 and 2.6 kernels. All I know is that the old version of XFree86 didn't have this problem. For a full history, see the Debian bug report at: http://bugs.debian.org/373949 -- Martijn van Oosterhout <[EMAIL PROTECTED]> http://svana.org/kleptog/ _______________________________________________ Mesa3d-dev mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
