[PATCH v3 6/6] Fortran: Nested functions, add scope parameter.

[Wiederhake, Tim]

Hi Yao,

Thanks for the feedback!

> -----Original Message-----
> From: Yao Qi [mailto:qiyaoltc@gmail.com]
> Sent: Monday, September 4, 2017 11:58 AM
> To: Wiederhake, Tim <tim.wiederhake@intel.com>
> Cc: gdb-patches@sourceware.org; Bernhard Heckel
> <bernhard.heckel@intel.com>
> Subject: Re: [PATCH v3 6/6] Fortran: Nested functions, add scope
> parameter.
> 
> On 17-08-11 13:06:51, Tim Wiederhake wrote:
> > From: Bernhard Heckel <bernhard.heckel@intel.com>
> >
> > In order to avoid name clashing in GDB, we add a scope
> > to nested subroutines. Enveloping function gives the
> > scope.
> 
> Hi Tim,
> Is there any reason that you split it from patch #5?  IMO, both #5 and
> #6 should be a single patch, and we need a NEWS entry.

I agree. Merged and NEWS entry written.

> According to your patch #5, I think both Ada and Fortran have the similar
> feature here.  I think we need to keep semantics same in both languages,
> like, "break func::nest_func"?

Sorry, I don't quite follow you. What is it that you'd like to see changed?

> >
> > ---
> >  gdb/doc/gdb.texinfo                        |  3 ++
> >  gdb/dwarf2read.c                           | 26 +++++++++++-
> >  gdb/testsuite/gdb.fortran/nested-funcs.exp | 28 +++++++++++--
> >  gdb/testsuite/gdb.fortran/nested-funcs.f90 | 66
> ++++++++++++++++++++++++++++--
> >  4 files changed, 114 insertions(+), 9 deletions(-)
> >  mode change 100755 => 100644 gdb/testsuite/gdb.fortran/nested-funcs.f90
> >
> > diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
> > index d098eba..983a570 100644
> > --- a/gdb/doc/gdb.texinfo
> > +++ b/gdb/doc/gdb.texinfo
> > @@ -15259,6 +15259,9 @@ The access component operator.  Normally used to
> access elements in derived
> >  types.  Also suitable for unions.  As unions aren't part of regular
> Fortran,
> >  this can only happen when accessing a register that uses a gdbarch-
> defined
> >  union type.
> > +@item ::
> > +The scope operator.  Normally used to access variables in modules or to
> set
> > +breakpoints on subroutines nested in modules or in other (internal)
> subroutines.
> >  @end table
> >
> >  @node Fortran Defaults
> > diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
> > index 894e6a0..92a9c49 100644
> > --- a/gdb/dwarf2read.c
> > +++ b/gdb/dwarf2read.c
> > @@ -7042,6 +7042,7 @@ partial_die_parent_scope (struct partial_die_info
> *pdi,
> >        return NULL;
> >      }
> >
> > +  /* Internal (nested) subroutines in Fortran get a prefix.  */
> >    if (pdi->tag == DW_TAG_enumerator)
> >      /* Enumerators should not get the name of the enumeration as a
> prefix.  */
> >      parent->scope = grandparent_scope;
> > @@ -7051,7 +7052,10 @@ partial_die_parent_scope (struct partial_die_info
> *pdi,
> >        || parent->tag == DW_TAG_class_type
> >        || parent->tag == DW_TAG_interface_type
> >        || parent->tag == DW_TAG_union_type
> > -      || parent->tag == DW_TAG_enumeration_type)
> > +      || parent->tag == DW_TAG_enumeration_type
> > +      || (cu->language == language_fortran
> > +	  && parent->tag == DW_TAG_subprogram
> > +	  && pdi->tag == DW_TAG_subprogram))
> >      {
> >        if (grandparent_scope == NULL)
> >  	parent->scope = parent->name;
> > @@ -8560,8 +8564,13 @@ process_die (struct die_info *die, struct
> dwarf2_cu *cu)
> >      case DW_TAG_type_unit:
> >        read_type_unit_scope (die, cu);
> >        break;
> > -    case DW_TAG_entry_point:
> >      case DW_TAG_subprogram:
> > +      /* Internal subprograms in Fortran get a prefix.  */
> > +      if (cu->language == language_fortran
> > +	  && die->parent != NULL
> > +	  && die->parent->tag == DW_TAG_subprogram)
> > +      cu->processing_has_namespace_info = 1;
> 
> We need some fall-through annotation to suppress warning in GCC 7.

Thanks, fixed.

> > +    case DW_TAG_entry_point:
> >      case DW_TAG_inlined_subroutine:
> >        read_func_scope (die, cu);
> >        break;
> > @@ -20169,6 +20178,19 @@ determine_prefix (struct die_info *die, struct
> dwarf2_cu *cu)
> >  	      return TYPE_TAG_NAME (parent_type);
> >  	    return "";
> >  	  }
> > +      case DW_TAG_subprogram:
> > +	/* Only internal subroutines in Fortran get a prefix with the name
> > +	   of the parent's subroutine.  */
> > +	if (cu->language == language_fortran)
> > +	  {
> > +	    if ((die->tag ==  DW_TAG_subprogram)
> > +		&& (dwarf2_name (parent, cu) != NULL))
> > +	      return dwarf2_name (parent, cu);
> > +	    else
> > +	      return "";
> > +	  }
> > +	else
> > +	  return determine_prefix (parent, cu);
> >  	/* Fall through.  */
> >        default:
> >  	return determine_prefix (parent, cu);
> > diff --git a/gdb/testsuite/gdb.fortran/nested-funcs.exp
> b/gdb/testsuite/gdb.fortran/nested-funcs.exp
> > index 4c2ee2a..0c8a416 100644
> > --- a/gdb/testsuite/gdb.fortran/nested-funcs.exp
> > +++ b/gdb/testsuite/gdb.fortran/nested-funcs.exp
> > @@ -31,8 +31,8 @@ if ![runto MAIN__] then {
> >  }
> >
> >  # Test if we can set a breakpoint in a nested function
> > -gdb_breakpoint "sub_nested_outer"
> > -gdb_continue_to_breakpoint "sub_nested_outer" ".*local_int = 19"
> > +gdb_breakpoint "testnestedfuncs::sub_nested_outer"
> > +gdb_continue_to_breakpoint "testnestedfuncs::sub_nested_outer"
> ".*local_int = 19"
> >
> >  # Test if we can access local and
> >  # non-local variables defined one level up.
> > @@ -43,13 +43,16 @@ gdb_test "set index = 42"
> >  gdb_test "print index" "= 42" "print index at BP_outer, manipulated"
> >  gdb_test "print local_int" "= 19" "print local_int in outer function"
> >
> > +
> 
> Unnecessary change, multiple instances.

Fixed as well.

> 
> >  # Non-local variable should be affected in one frame up as well.
> >  gdb_test "up"
> 
> 
> --
> Yao (齐尧)

Regards,
Tim
Intel Deutschland GmbH
Registered Address: Am Campeon 10-12, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de
Managing Directors: Christin Eisenschmid, Christian Lamprechter
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928