RE: [PATCH v4 3/3] dwarf, fortran: add support for DW_TAG_entry_point

["Kempke, Nils-Christian via Gdb-patches" <gdb-patches@sourceware.org>]

> -----Original Message-----
> From: Tom Tromey <tom@tromey.com>
> Sent: Friday, July 15, 2022 9:57 PM
> To: Nils-Christian Kempke via Gdb-patches <gdb-patches@sourceware.org>
> Cc: Kempke, Nils-Christian <nils-christian.kempke@intel.com>;
> aburgess@redhat.com; Tim Wiederhake <tim.wiederhake@intel.com>;
> Bernhard Heckel <bernhard.heckel@intel.com>;
> jinisusan.george@amd.com; tom@tromey.com
> Subject: Re: [PATCH v4 3/3] dwarf, fortran: add support for
> DW_TAG_entry_point
> 
> >>>>> Nils-Christian Kempke via Gdb-patches <gdb-
> patches@sourceware.org> writes:
> 
> > Fortran provides additional entry points for subroutines and functions.
> > These entry points may use only a subset (or a different set) of the
> > parameters of the original subroutine.  The entry points may be described
> > via the DWARF tag DW_TAG_entry_point.
> 
> > @@ -21797,12 +21852,15 @@ determine_prefix (struct die_info *die,
> struct dwarf2_cu *cu)
> >  	return "";
> >        case DW_TAG_subprogram:
> >  	/* Nested subroutines in Fortran get a prefix with the name
> > -	   of the parent's subroutine.  */
> > +	   of the parent's subroutine.  Entry points are prefixed by the
> > +	   parent's namespace.  */
> >  	if (cu->lang () == language_fortran)
> >  	  {
> >  	    if ((die->tag ==  DW_TAG_subprogram)
> >  		&& (dwarf2_name (parent, cu) != NULL))
> >  	      return dwarf2_name (parent, cu);
> > +	    else if (die->tag == DW_TAG_entry_point)
> > +	      return determine_prefix (parent, cu);
> >  	  }
> 
> It seems to me that this will also need a special case in
> cooked_indexer::index_dies to ensure that the cooked_index_entry for the
> entry point has the correct parent.  Parent relationships are how name
> components are determined there, so IIUC the entry point's parent should
> be the same as the subroutine's parent -- not the subroutine itself.
> 
> One way to see whether this works is to write a .gdb_index and then dump
> it and examine the resulting name of the entry point.
> 
> thanks,
> Tom

Hi Tom,

Thanks for the review - ah, I see. Let me quickly recap just to see whether I
understood this: when writing the gdb_index we use the
full_name (struct obstack *storage) method of the cooked_index_entry.

In order for the name scope to be correct we'd have to set the function's
parent as the entry's parent.

I changed this now to correctly set the cooked_index_entry's parent but,
I could not test this with my Fortran examples/code.  The problem is that
full_name (struct obstack *storage) does not consider the scope when
the language is set to Fortran - while I could verify that the parent is set
correctly and the name would be printed correctly if I hack the language if
a bit - with Fortran code (the currently only user of DW_TAG_entry_point
I know), full_name will return 'canonical' which is still the name of the
entry point without scope.

I will send a v5 shortly with the above mentioned change.

Thanks,
Nils
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva  
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928