Re: Fix nested functions

[Jan Hubicka <jh at suse dot cz>]

> Jan Hubicka wrote:
> 
> > I see, adding "&& !LABEL_REF_NONLOCAL_P (trial)" would probably help
> > here, I will test the change tomorrow.
> 
> That crashes, because it then falls through to validate_change, 
> which succeeds in re-recognizing the insn as a direct jump;
> but now we have a direct jump with NULL JUMP_LABEL ...

Hmm, this reminds me the i386 case - fixing one problem leads to other
one.
I guess adding the mark_jump_label after validate_change will lead to
later crash in shorten_branches.
We really need to bail out earlier and give up idea of propagating
nonlocal LABEL_REF into the jump I would say :( 

Thanks for looking into it!
Honza
> 
> > the nonlocal gotos are usually indirect, so one mechanizm is marking the
> > indirect goto while the other is marking the actual label when it is
> > load to register, I think.
> 
> Makes sense, thanks!
> 
> Bye,
> Ulrich
> 
> -- 
>   Dr. Ulrich Weigand
>   weigand@informatik.uni-erlangen.de