From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030225 Description of problem: I have an odd problem with Mesa 4.0.4, RedHat 8.0 worked fine, games rendered correctly. With RedHat 9.0, mesa 4.0.4, Ati Mobility 128 using the r128 driver, I see shadows and decals, such as blood, through walls and around corners. The decal rendered image is overlayed over all other images, thus if I kill someone in a room in Quake3, when I leave the room, their shadow and blood splatter is visible outside drawn on the outer wall. It seems that either the Z-buffer is not correctly working on depth or the clipping is not happening correctly, but this problem was not evident in RedHat 8.0 Any suggestions?? Version-Release number of selected component (if applicable): 4.0.4 How reproducible: Always Steps to Reproduce: 1.Using an Ati Mobility M3 128, 8 Meg card, start Quake3 or RTCW 2.Shoot and kill someone or stand in a room, surrounded by walls 3.Watch as the shadows of your opponents, gives away their position long before they come into view, as you can see the shadow through the walls. Actual Results: See shadows through walls, bad clips. Expected Results: Shadows and decals should have been clipped away so that the I could not see the opponent coming around the corner. Additional info: System: Dell Insprion 5000e, 384 Meg Ram, P3 750, Ati Rage Mobility M3 128, 8 meg card.
Please report this problem directly to the Mesa3D project upstream or the dri-devel mailing list. You can also file it in XFree86 bugzilla if you wish, and I will track the upstream progress on this problem and consider including any fixes they make. If you'd like me to track the upstream bug report, include the URL here after filing the bug: http://bugs.xfree86.org Thanks.
I passed the information on to Mesa mailing list, but they said it was a driver problem. I have a work around however, if I install Red Hat 8.0, keep the rage128 driver and modules, using them in Red Hat 9.0 is fine. And the clipping problem is no longer evident. So that would suggest that it is indeed a driver problem.
There's no URL to link to for me to track upstream progress on this, so I'm unable to track it. Presumeably the DRI project should be notified via sending a bug report to the URL provided above in XFree86 bugzilla. Closing report as WONTFIX, but feel free to add a URL to your XFree86.org bug report and reopen if you'd like it tracked upstream still.