From: Brad K. <bra...@ki...> - 2005-05-10 13:17:19
|
Brian Paul wrote: > Brad King wrote: >> Program received signal SIGSEGV, Segmentation fault. >> 0x410265b6 in clear_pixmap () from >> /home/kingb/Nightly/Mesa-build/lib/libMesaGL.so.1 > That helps a lot. I think one of my assertions was invalid when using a > single-buffered visual. I've replaced that assertion with a few others > that should expose any potentially remaining problem. Great, I'm glad to have helped find the problem. >> This machine was the only one on which this test failed last night so >> it is probably a Mesa problem. >> >> The second and fourth failures are probably not Mesa-related. The >> third failure might be a Mesa bug but it is not a crash. This page >> reports the valid (expected) image, the image the test produced, and >> the difference image: >> >> http://www.vtk.org/Testing/Sites/hythloth.kitware/Linux-gcc-3.3/20050506-0300-Nightly/Results/__Graphics_Testing_Tcl_OSCone-image.html > > Do all the tests generally run with the same GLX Visual? Each test is a completely separate process execution, and only one runs at a time. >> The test was passing 2 days ago and now this machine is the only one >> where it fails: >> >> http://www.vtk.org/Testing/Dashboard/20050506-0300-Nightly/TestDetail/__Graphics_Testing_Tcl_OSCone-image.html >> >> >> so it is probably also a Mesa problem. I can give you more >> information if you need it. > > Let me know what happens with my latest check-in. Okay, the crashes are gone but these two test failures: http://www.vtk.org/Testing/Sites/hythloth.kitware/Linux-gcc-3.3/20050510-0300-Nightly/Results/__Graphics_Testing_Tcl_OSCone-image.html http://www.vtk.org/Testing/Sites/hythloth.kitware/Linux-gcc-3.3/20050510-0300-Nightly/Results/__Parallel_Testing_Cxx_DistributedData-image.html are definately a Mesa problem. They started right after your big checkin. I setup symlinks to trick the binaries into using the system GL instead of the Mesa GL and these two tests pass. If I remove the symlinks they fail again. Let me know what other information you need to find the problem. Thanks, -Brad |