Linux上的gdb fork（）exec

Question

I found the similar question on this site, but the solution doesn't work for me. In my main(), I have

if(fork() == 0) execl(program b, args);

I set couple break points at every beginning. Then I use "set follow-fork-mode" and run.

After fork, gdb attached to the child process, showing "Attaching after fork to child process 29730." But I won't have any chance to set break points in program b anymore, although I have 15 seconds sleep in program b. It executed until end or break.

How to set the break point in program b?

Thanks!

Answer 1

But I won't have any chance to set break points in program b anymore

The trick is to set a breakpoint in the child as a delayed breakpoint. When the child is execl() d, GDB will set the breakpoint in it. Example:

// a.c
#include <unistd.h>

int main()
{
  if (0 == fork()) execl("./b.out", "b.out", (char*)0);
  return 0;
}


// b.c
int foo() { return 0; }

int main() { return foo(); }


gcc -g a.c; gcc -g b.c -o b.out


gdb -nx -q ./a.out
Reading symbols from /tmp/a.out...done.
(gdb) b foo
Function "foo" not defined.
Make breakpoint pending on future shared library load? (y or [n]) y
Breakpoint 1 (foo) pending.

The breakpoint is pending because there is no foo() in a.out (but there is foo() in b.out ).

(gdb) set follow-fork-mode child
(gdb) run
Starting program: /tmp/a.out 
[New process 18759]
process 18759 is executing new program: /tmp/b.out
[Switching to process 18759]

Breakpoint 1, foo () at b.c:1
1   int foo() { return 0; }
(gdb) bt
#0  foo () at b.c:1
#1  0x00000000004004dd in main () at b.c:3
(gdb) quit