This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
sneaky egcs 1.1/newmat09 problem
- To: egcs-bugs at cygnus dot com
- Subject: sneaky egcs 1.1/newmat09 problem
- From: Conrad Sanderson <conrad at me dot gu dot edu dot au>
- Date: Mon, 21 Sep 1998 16:56:02 +1000 (EST)
I found a problem with the egcs 1.1 C++ compiler when using the newmat09
matrix library (available from http://webnz.com/robert/index.html ).
[ The new header file for the library has been used ]
Under egcs 1.0.3 the newmat09 library works as expected.
The bug occurs when using the eigenvalue decomposition function of the
form:
EigenValues(A,D,V); // A symmetric; V is a matrix, D diagonal
it gives a segmentation fault - i know this sounds rather flakey, but
unfortunately I don't have the time (and the knowledge) to start analyzing
the newmat library.
The bug occurs whether using optimization or not. I've recompiled the
library using these switches:
-O3 -mpentium -archpentium -D_REENTRANT -fschedule-insns2 -fschedule-insns
and without any switches at all, with the bug recurring.
Please contact me if you require more info.
My system is Red Hat 5.1 + patches, on Pentium II 266, kernel 2.0.34.
egcs 1.1 and 1.0.3 were obtained as RPMs from the RedHat contrib site.
PS. Many thanks for all your work on egcs