This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Strange results and debugging problems with Pentium-egcs 1.0
- To: egcs at cygnus dot com
- Subject: Strange results and debugging problems with Pentium-egcs 1.0
- From: Tobias Polzin <polzin00 at marvin dot informatik dot uni-dortmund dot de>
- Date: Tue, 23 Dec 1997 17:20:16 +0100
I have strange problems with a C++-package when it's compiled with egcs on
a Pentium (RedHat-Linux): The program calculates wrong results. (like 2+2=5,
just a bit more complicated; it worked with Pentium+gcc2.7, Sun+gcc2.7,
Alpha+egcs).
I wanted to debug the program to find the problem, but the debugging did
not work correctly: gdb reported object pointers as <error type> and
char* as <incomplete type>, debugging was not possible.
With simpler programs I did not have these problem. The installation of
egcs was done by the department's specialist for such a thing, and I did
it again, but with the same results.
Does anyone have similar, perhaps simpler problems of the same kind?
Or perhaps a solution?
Happy easter,
Tobias Polzin
(gdb) info stack
#0 warn_if (EXP=0x80e117b "1", line=913, file=0x80e10c2 "reduceGraph.C")
at datatypes.C:9
#1 0x8071977 in ReduceGraph::updateContract (this=<error type>)
at reduceGraph.C:913
#2 0x8065bce in nv (G=<error type>) at newReduction.C:373
#3 0x8061ec5 in run_r (order=<incomplete type>, R=<incomplete type>,
breakV_R=0) at reduction.C:819
#4 0x8063276 in reduction () at reduction.C:891
#5 0x804ae0b in multiplex (co=<incomplete type>) at multiplx.C:59
#6 0x804aaae in run_GSteiner () at GSteiner.C:30
#7 0x804a71c in main (argc=2, argv=0xbffff90c, env=0xbffff918) at steiner.C:73
(