Harold Gutch <logix@foobar.franken.de> wrote:

> So my stack seems to get lost during that ap_fini_vhost_config()
> call.  Does anybody have an idea what might be wrong here?

It might be a bug in gdb or in the way kernel handles breakpoint
traps.  Sometimes (and I don't know the exact conditions) when
stepping, or hitting a breakpoint, a debuggee would get a fatal
signal.  By the looks of it (registers, code), it seems that one
instruction in the debuggee has *not* been executed.  E.g. something
like

	mov.l	.L999, r1
	mov.l	@r1, r2

and it would crash on the second instruction and r1 has "old" value,
as if the first mov has not been executed.

I've been bitten by this regularly, but had not time to debug it.

SY, Uwe
-- 
uwe@ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen