[26531] in Athena Bugs
Re: linux 9.4.14: matlab
daemon@ATHENA.MIT.EDU (Alex T Prengel)
Thu Aug 25 16:42:28 2005
Message-Id: <200508252041.j7PKfFML024247@dit.mit.edu>
To: cfox@mit.edu, bugs@mit.edu
In-Reply-To: "[26528] in Athena Bugs"
Date: Thu, 25 Aug 2005 16:41:15 -0400
From: Alex T Prengel <alexp@mit.edu>
X-Spam-Score: 1.041
X-Spam-Level: * (1.041)
X-Spam-Flag: NO
cc: cfyi@mit.edu
cc: alexp@mit.edu
cc: testers@mit.edu
Errors-To: bugs-bounces@mit.edu
>>Here's another reproduceable crash, submitted by OLC user ckemp:
>> X = rand(2,1);
>> X*X'
>and another, reported to swmaint by user gbarter:
>
> >> phi=rand(20,10);
> >> coor=rand(10,2);
> >> phi*coor
>
>Matlab crashes and exits.
Well, I've fixed these and the plot3 problem for now. We were setting
LD_ASSUME_KERNEL to 2.4.1, to get around the problem below when using
Symbolic Toolbox functions:
>> syms a b
>> a=b^2;
??? Invalid MEX-file '/mit/matlab_v7.0.4/distrib/toolbox/symbolic/maplemex.mexglx': /mit/matlab_v7.0.4/distrib/bin/glnx86/libmaple.so: symbol errno, version GLIBC_2.0 not defined in file libc.so.6 with link time reference.
when I took this setting out, the first set of problems went away, but of
course the last one came back... since the latter is likely to affect
fewer users than the other problems, I'm leaving LD_ASSUME_KERNEL unset
for now.
The libmaple.so problem has been around for about 2 years, and the Mathworks
support site still says to set LD_ASSUME_KERNEL to 2.4.1 to work around
it... :-(
Alex
