Re: [PATCH] -gused

[Devang Patel <dpatel at apple dot com>]

On Friday, June 20, 2003, at 4:23 PM, Richard Henderson wrote:

> On Fri, Jun 20, 2003 at 03:37:50PM -0700, Devang Patel wrote:
> > --- gcc/dbxout.c	20 Jun 2003 22:27:25 -0000
> > + extern int flag_debug_only_used_symbols;
>
> Should be in flags.h.

Yes. I'll remove this redundant line.

> > -     dbxout_symbol (decl, 0);
> > +     {
> > +       int saved_tree_used = TREE_USED (decl);
> > +       TREE_USED (decl) = 1;
> > +       dbxout_symbol (decl, 0);
> > +       TREE_USED (decl) = saved_tree_used;
> > +     }
>
> Why?  Should definitely be a comment there, whatever the answer is.

It is to intercept call from outside. dbxout_symbol now only emits 
symbols
which are used so this is done to avoid overwriting TREE_USED bit
unnecessarily.

> > *************** dbxout_finish (filename)
> > *** 719,724 ****
> > --- 732,743 ----
> >   #ifdef DBX_OUTPUT_MAIN_SOURCE_FILE_END
> >     DBX_OUTPUT_MAIN_SOURCE_FILE_END (asmfile, filename);
> >   #endif /* DBX_OUTPUT_MAIN_SOURCE_FILE_END */
> > +   if (symbol_queue)
> > +     {
> > +       free (symbol_queue);
> > +       symbol_queue = NULL;
> > +       symbol_queue_size = 0;
> > +     }
>
> Didn't get moved?

Need to write free_symbol_queue() :-)

> > +       if ((TREE_CODE (type) == RECORD_TYPE
> > + 	   || TREE_CODE (type) == UNION_TYPE
> > + 	   || TREE_CODE (type) == QUAL_UNION_TYPE
> > + 	   || TREE_CODE (type) == ENUMERAL_TYPE)
> > + 	  && TYPE_STUB_DECL (type)
> > + 	  && TREE_CODE_CLASS (TREE_CODE (TYPE_STUB_DECL (type))) == 'd'
> > + 	  && ! DECL_IGNORED_P (TYPE_STUB_DECL (type)))
>
> You have this complex test replicated many times.  Seems like
> you'd just send in "type" and be done with it.
>
> > + extern tree *symbol_queue;
> > + extern int symbol_queue_index;
> > + extern int symbol_queue_size;
>
> VARRAY?

I have not thought much about this.

> Any reason this is part of the public interface?

Because  I need to write free_symbol_queue() :-)

> > + debug_queue_symbol (tree decl)
>
> I'd actually had it in my head to have final call the queueing
> functions directly, instead of the calling the debug functions
> directly.  Not that this just be a library for the debug functions
> to call.

Can you explain this little bit more?.

-Devang