I have an MPI application with which combines both C and Fortran sources. Occasionally it crashes due to a memory related bug, but I am having trouble finding the bug (it is somewhere in someone else's code, which at the moment I'm not very familiar with). I haven't yet been able to catch it with gdb, but sometimes a glibc backtrace is output as shown below.
The bug is probably close to "(main_main_+0x3bca)[0x804d5ce]", (but with a memory error, I know this may not be the case). My question is, does anyone know how to convert +0x3bca or 0x804d5ce into a particular line of the code?
Any other suggestions on tracking down the bug would also be appreciated. I'm quite familiar with the basics of gdb.
*** glibc detected *** /home/.../src/finite_element: munmap_chunk(): invalid pointer: 0x09d83018 ***
======= Backtrace: =========
/lib/i386-linux-gnu/libc.so.6(+0x73e42)[0xb7409e42]
/lib/i386-linux-gnu/libc.so.6(+0x74525)[0xb740a525]
/home/.../src/finite_element(main_main_+0x3bca)[0x804d5ce]
/home/.../src/finite_element[0x804e195]
/home/.../src/finite_element(main+0x34)[0x804e1e8]
/lib/i386-linux-gnu/libc.so.6(__libc_start_main+0xf3)[0xb73af4d3]
/home/davepc/finite-element/src/finite_element[0x8049971]
======= Memory map: ========
08048000-08056000 r-xp 00000000 08:05 1346306 /home/.../src/finite_element
08056000-08057000 r--p 0000d000 08:05 1346306 /home/.../src/finite_element
08057000-08058000 rw-p 0000e000 08:05 1346306 /home/.../src/finite_element
09d1b000-09d8f000 rw-p 00000000 00:00 0 [heap]
b2999000-b699b000 rw-s 00000000 08:03 15855 /tmp/openmpi-sessions-_0/37612/1/shared_mem_pool.babel
b699b000-b6b1d000 rw-p 00000000 00:00 0
b6b31000-b6b3d000 r-xp 00000000 08:03 407798 /usr/lib/openmpi/lib/openmpi/mca_osc_rdma.so
b6b3d000-b6b3e000 r--p 0000b000 08:03 407798 /usr/lib/openmpi/lib/openmpi/mca_osc_rdma.so
b6b3e000-b6b3f000 rw-p 0000c000 08:03 407798 /usr/lib/openmpi/lib/openmpi/mca_osc_rdma.so
<snip>
Thank you...