Re: x86_64 frame unwind info

[Jan Hubicka <jh at suse dot cz>]

> 
> You should realize that flag_non_call_exceptions currently implements
> flag_asynchronous_unwind_tables.  For now you should just make

I've realized that, but I tought it is just our current laziness, as the
non-call-exception came in as asynchronous-exception that are flawed by
some other problems, so that the fact we make unwind info exact isjust
leftout and that we can optimize the info same way as we do for
!flag_non_call_exceptions just considering any instruction that traps
as possible "call".
> flag_non_call_exceptions imply flag_asynchronous_unwind_tables and
> adjust dwarf2out appropriately.

OK, I will do.
> 
> Now, something that should arguably be checked here is for_eh.
> If !for_eh, then this is .debug_frame, which is for debugger
> consumption, which probably ought to keep the CFA up to date
> all the time.  But for_eh isn't available here, and it would
> take some restructuring to have for_eh discard information later.
> 
> > -       int any_eh_needed = 0;
> > +       int any_eh_needed = flag_asynchronous_unwind_tables;
> 
> Much better.

:) Nice to hear that I am improving.  My understanding of the dwarf2out
concept is still in the state of flux, but I am getting closer :)

Honza
> 
> 
> r~