Greetings! I have been using the most recent beta of 3.1. There seems to be a memory access problem in the vertex transformation code in 64-bits which causes a SEGV. It seems fine in version 3.0. On Irix systems, this was only happening in the transform_points3_3d_no_rot_raw function. (In other words, when I turned on perspective in the application, it would not crash.) In fact, I was able to prevent it from crashing if I padded the vectors to 4 elements in the output buffer. On a DEC Alpha (osf1), it would happen in several functions, and padding the output vectors to 4 components in every transformation function seemed to have no effect. It still crashed. The actual error seems to be that the memory pointed to by `from' or `to' is invalid, or misaligned, or something else. The pointers themselves seem fine. Unfortunately, "atom" (a memory checking tool on the DECs) reported no errors. (...except an uninitalized memory read: in vbfill.c, line 103, ReducedPrimitives was accessed, but it wasn't initialized by the malloc() in gl_immediate_alloc() in vb.c. This couldn't be the problem, though.) Even more unfortunate, Purify (a more complete memory checker which I use on Irix systems) is not available for 64-bit executables and won't be until at least January thanks to a lawsuit with SGI. As I said, I think everything is fine in 3.0, so I will probably have to use it for a while. ------------------------------------------ - - - - - - - - - - - - - - - Jeremy Meredith | email: [EMAIL PROTECTED] Lawrence Livermore National Laboratory | phone: (925) 422-1197 P.O. Box 808 L-98 | fax : (925) 422-3389 Livermore, CA 94550 | loc : bldg 132N room 1025 _______________________________________________ Mesa-bug maillist - [EMAIL PROTECTED] http://lists.mesa3d.org/mailman/listinfo/mesa-bug _______________________________________________ Mesa-dev maillist - [EMAIL PROTECTED] http://lists.mesa3d.org/mailman/listinfo/mesa-dev
