Dealing with segmentation faults

[heartless]

Hi everyone,
I'd like to know how to prompt gdb to show me where segmentation fault in an application occurs and why. Any ideas?

 

[chroot]

Segfaults happen when your program tries to write to memory that doesn't belong to it. Common causes are:

1) Using uninitialized pointers.
2) Using uninitialized variables for pointer arithmentic (array indexing).
3) Letting a counter go beyond the boundaries of an array.

You should be able to use gdb to step through your code until you find the offending line.

- Warren

 

[heartless]

Yepp, that's what wikipedia says. However do you have any ideas how to find specific line or lines that cause segmentation fault in an application. I tried gdb for "normal" aps, and everything works perfectly, but I can't debug any applications with segmentation faults.

For example:
adrian@ubuntu:~/files/asm$ gdb env_show --quiet
(no debugging symbols found)
Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".
(gdb) r
Starting program: /home/adrian/files/asm/env_show

Program terminated with signal SIGSEGV, Segmentation fault.
The program no longer exists.
You can't do that without a process to debug.
(gdb) info registers
The program has no registers now.
(gdb) bt
No stack.
(gdb)

and so on, I'm not very experienced with gdb,
--Thanks,

 

[chroot]

Rather than just running your program, put a breakpoint in and step through it line by line.

- Warren

 

[chroot]

Also, I presume that you compiled your program with gcc -g to include debugging information in the executable.

- Warren

 

[nmtim]

Or run gdb on the core dump. Never done it myself, but the core file should have the program state at the time of the segfault.