Re: [RFA 3/4] Improved linker-debugger interface

[Gary Benson <gbenson@redhat.com>]

Jan Kratochvil wrote:
> On Wed, 18 Jul 2012 01:42:12 +0200, Sergio Durigan Junior wrote:
> > I was considering suggesting the removal of the `gdb_assert'
> > calls, and instead make it a simple check and return properly
> > if the OBJFILE does not support probes.
> > 
> >     unsigned
> >     get_probe_argument_count (struct objfile *objfile, struct probe *probe)
> >     {
> >       if (objfile->sf == NULL || objfile->sf->sym_probe_fns == NULL)
> >         return 0;
> > 
> >       return objfile->sf->sym_probe_fns->sym_get_probe_argument_count (objfile,
> >                                                                        probe);
> >     }
> > 
> > It would cover the case when OBJFILE does not support probe and
> > would not be less safe.
> 
> In fact why 'struct probe' does not contain its 'struct objfile *'?
> It does not make sense to pass probe with different objfile.  So the
> parameter 'objfile' should not be passed at all.

That would certainly simplify my code.  Sergio, is this possible to
do?

Cheers,
Gary

-- 
http://gbenson.net/