Re: [patch 1/3] Clear stale specific locs, not whole bpts [rediff]

[Frederic Riss <frederic.riss@gmail.com>]

Hi !

On 7 June 2010 15:40, Jan Kratochvil <jan.kratochvil@redhat.com> wrote:
> Checked-in:
>        http://sourceware.org/ml/gdb-cvs/2010-06/msg00058.html

Seems that this commit is partially broken. In non-stop mode, after
deleteing a step breakpoint, GDB call print_it_typical () on a bpstat
that references that moribund breakpoint. Looking at the
print_it_typical () code, you'll quickly see that this means crash.
It's very easy to reproduce:

------ foo.c ---------------------------
int foo () {
	return 0;
}

int main (int argc, char *argv[])
{
	return foo ();
}
------ foo.c ---------------------------

gcc -g foo.c && gdb a.out -ex 'set non-stop' -ex start -ex n

Cheers,
Fred