Description of problem: When I run SPM2 (Statistical Parametric Mapping, a graphical program run in matlab) in Matlab 6.5.1 the letters of the graphical windows of SPM2 are almost unreadable and matlab crashes after some procedures (hard to give reproducable procedure). BUT, this does not happen if I ssh into this computer from another computer (with a different, "older" os), then matlab and spm runs perfectly. I have tried with matlab 6.0 but the same problem appears. This used to work fine until I changed the os from Mandrake to Fedora. It seems to be some kind of "graphical" problem since matlab also produces crash-dumps some times when I click on a window of the spm program. Could it be an XFree86 bug? Version-Release number of selected component (if applicable): XFree86: 4.3.0-42 glibc: 2.3.2 How reproducible: Always Steps to Reproduce: 1.start matlab 2.spm 3. Actual results: Expected results: Additional info:
Created attachment 97256 [details] matlab_crash_dump
The crash dump you've attached makes it rather clear IMHO that this is a bug in Matlab itself. If you have not already done so, please read the entire crash dump that you've attached, and follow the detailed instructions given inside the crash dump to report the problem to Matlab developers. This Matlab bug is not an XFree86 bug/problem. Closing 'NOTABUG'
I sent the matlab crash dump to the matlab developers a week ago. They will perhaps have a bugfix in the next version of matlab. Is there any way to downgrade Fedora so that I can still use Fedora and run Matlab (since matlab has worked on other Linuxes)? I would really appreciate your help! I am an inexperienced Linux user, what more than X is running on the server if you are using a program on a client via ssh? Thanks for your help! /Eva-Lena
Eva-Lena, what type of video card does your system have? (I can reproduce this on some systems that I have access to, but not others, and I'm starting to wonder if this is specific to certain video cards or anything like that.)
[I know this bug is closed as NOTABUG, but this is a good place to put this information in case anyone else stumbles upon the same problem.] Never mind my previous comment. The virtually-100% reproducible test case my user found causes this crash on XFree86 4.1.0 and up, but not XFree86 4.0.x and earlier. It doesn't seem to be the X server -- running a 3.3.6 server with the other packages at 4.1.0 also reproduces the crash. *However*, the remote end (where Matlab is actually running) also seems to play a part. One remote computer I'm testing with, which has a weird mix of RH 7.1, 7.2 and recompiled RHEL 2.1 packages, running a recompiled RHEL 2.1 glibc and a mainline 2.4.26 kernel, does not exhibit the crash. Two other remote computers, running Red Hat 9, do have Matlab crashing; at least one of the two is running the exact same Matlab binaries as the computer that's not crashing.