Re: Gdbserver Threading Issues

[Daniel Jacobowitz <drow@mvista.com>]

On Fri, Dec 06, 2002 at 05:11:54PM -0500, Paul Mundt wrote:
> On Fri, 2002-12-06 at 14:28, Daniel Jacobowitz wrote:
> > Not without more information.  Look at the remote session transcript,
> > see what is going on.
> 
> Hum, still debugging this .. I'm still at a loss as to why breakpoints
> would be having any effect on anything, though from the looks of it the
> newly created threads are tracked just fine by
> thread_db_find_new_threads() in gdbserver/thread-db.c
> 
> Also single stepping through maybe_attach_thread() everything looks
> alright, though still the threads don't show up on the GDB side.

Thread events aren't reported immediately for latency reasons; but they
should all be delivered the first time the child stops...

> Also, on the gdb side of things, it doesn't look like there are any
> events happening .. thus handle_inferior_event() is never stepped into
> at thread creation/running/exit time. Although td_thr_event_enable()
> still returns TD_OK.

Wait, what's going on?  Are you saying td_thr_event_enable is being
called on your host GDB?  That's not going to work right.

> Notably, if I connect gdb to gdbserver while gdb and gdbserver are both
> running under gdb control, SIG32 issues appear again.. 

gdbserver may get a SIG32 while debugging; set the process debugging
gdbserver to "nostop noprint pass" that signal.  Ditto SIG33, I think.


-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer