* RE: [PING*4] [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
@ 2021-11-11 5:32 Kumar N, Bhuvanendra
2021-11-27 7:56 ` Joel Brobecker
0 siblings, 1 reply; 2+ messages in thread
From: Kumar N, Bhuvanendra @ 2021-11-11 5:32 UTC (permalink / raw)
To: gdb-patches
Cc: George, Jini Susan, Achra, Nitika, Sharma, Alok Kumar, E,
Nagajyothi, Joseph, Ancel
[AMD Official Use Only]
Hi all,
Gentle PING*4 for the revised patch. Requesting some of you from the community to review the updated code patch please, i had addressed all the previous review comments.
Regards,
bhuvan
-----Original Message-----
From: Kumar N, Bhuvanendra
Sent: Thursday, November 4, 2021 11:46 AM
To: Andrew Burgess <andrew.burgess@embecosm.com>
Cc: gdb-patches@sourceware.org; George, Jini Susan <JiniSusan.George@amd.com>; Achra, Nitika <Nitika.Achra@amd.com>; Sharma, Alok Kumar <AlokKumar.Sharma@amd.com>; E, Nagajyothi <Nagajyothi.E@amd.com>
Subject: RE: [PING*3] [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
[AMD Official Use Only]
Hi Andrew and all,
Gentle PING*3 for the revised patch
Regards,
bhuvan
-----Original Message-----
From: Kumar N, Bhuvanendra
Sent: Wednesday, October 27, 2021 6:46 PM
To: Andrew Burgess <andrew.burgess@embecosm.com>
Cc: gdb-patches@sourceware.org; George, Jini Susan <JiniSusan.George@amd.com>; Achra, Nitika <Nitika.Achra@amd.com>; Sharma, Alok Kumar <AlokKumar.Sharma@amd.com>; E, Nagajyothi <Nagajyothi.E@amd.com>
Subject: RE: {PING*2] [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
[AMD Official Use Only]
Hi Andrew and all,
Gentle PING*2 for the revised patch
regards,
bhuvan
-----Original Message-----
From: Kumar N, Bhuvanendra
Sent: Monday, October 18, 2021 10:32 PM
To: 'Andrew Burgess' <andrew.burgess@embecosm.com>
Cc: 'gdb-patches@sourceware.org' <gdb-patches@sourceware.org>; George, Jini Susan <JiniSusan.George@amd.com>; Achra, Nitika <Nitika.Achra@amd.com>; Sharma, Alok Kumar <AlokKumar.Sharma@amd.com>; E, Nagajyothi <Nagajyothi.E@amd.com>
Subject: RE: [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
[AMD Official Use Only]
Hi Andrew and all,
Gentle ping for the revised patch
regards,
Bhuvan
-----Original Message-----
From: Kumar N, Bhuvanendra
Sent: Friday, October 8, 2021 2:35 AM
To: Andrew Burgess <andrew.burgess@embecosm.com>
Cc: gdb-patches@sourceware.org; George, Jini Susan <JiniSusan.George@amd.com>; Achra, Nitika <Nitika.Achra@amd.com>; Sharma, Alok Kumar <AlokKumar.Sharma@amd.com>; E, Nagajyothi <Nagajyothi.E@amd.com>
Subject: RE: [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
[AMD Official Use Only]
Hi Andrew,
Thanks again for your review comments. I have now introduced a new type(TYPE_CODE_NAMELIST) for namelist type as you suggested, could you please review the revised changes. Updated patch is attached and also inlined with this email. There are no regressions found during testing.
thanks,
bhuvan
PATCH inlined:
From 89b7f847042ccea44633f714db07400fa6dd3617 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E2=80=9Cbhkumarn=E2=80=9D?= <Bhuvanendra.KumarN@amd.com>
Date: Fri, 8 Oct 2021 02:01:18 +0530
Subject: [PATCH] gdb/fortran: Fix ptype and print commands for namelist variables.
GCC/gfortran support namelist(a fortran feature), it emits DW_TAG_namelist and DW_TAG_namelist_item dies. But gdb does not process these dies and support namelist variables during print and ptype commands. When tried to print, it bails out with the error message as shown below.
(gdb) print nml
No symbol "nml" in current context.
This commit is to make the print and ptype commands work for namelist variables and its items. Sample output of these commands is shared below, with fixed gdb.
(gdb) ptype nml
type = Type nml
integer(kind=4) :: a
integer(kind=4) :: b
End Type nml
(gdb) print nml
$1 = ( a = 10, b = 20 )
---
gdb/dwarf2/read.c | 43 ++++++++++++++++++----
gdb/f-typeprint.c | 6 +++-
gdb/f-valprint.c | 11 ++++++
gdb/gdbtypes.h | 2 ++
gdb/testsuite/gdb.fortran/namelist.exp | 49 ++++++++++++++++++++++++++
gdb/testsuite/gdb.fortran/namelist.f90 | 27 ++++++++++++++
include/dwarf2.def | 2 +-
7 files changed, 132 insertions(+), 8 deletions(-) create mode 100644 gdb/testsuite/gdb.fortran/namelist.exp
create mode 100644 gdb/testsuite/gdb.fortran/namelist.f90
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c index cbd9a3012eb..e6cd8ed48f0 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -9705,6 +9705,7 @@ process_die (struct die_info *die, struct dwarf2_cu *cu)
case DW_TAG_interface_type:
case DW_TAG_structure_type:
case DW_TAG_union_type:
+ case DW_TAG_namelist:
process_structure_scope (die, cu);
break;
case DW_TAG_enumeration_type:
@@ -14562,8 +14563,20 @@ dwarf2_add_field (struct field_info *fip, struct die_info *die,
fp = &new_field->field;
- if (die->tag == DW_TAG_member && ! die_is_declaration (die, cu))
- {
+ if ((die->tag == DW_TAG_member || die->tag == DW_TAG_namelist_item)
+ && ! die_is_declaration (die, cu))
+ {
+ /* For the DW_TAG_namelist_item die, use the referenced die. */
+ if (die->tag == DW_TAG_namelist_item)
+ {
+ struct attribute *attr1 = dwarf2_attr (die, DW_AT_namelist_item, cu);
+ struct die_info *item_die = nullptr;
+ struct dwarf2_cu *item_cu = cu;
+ if (attr1->form_is_ref ())
+ item_die = follow_die_ref (die, attr1, &item_cu);
+ if (item_die != nullptr)
+ die = item_die;
+ }
/* Data member other than a C++ static data member. */
/* Get type of field. */
@@ -15621,6 +15634,10 @@ read_structure_type (struct die_info *die, struct dwarf2_cu *cu)
{
type->set_code (TYPE_CODE_UNION);
}
+ else if (die->tag == DW_TAG_namelist)
+ {
+ type->set_code (TYPE_CODE_NAMELIST);
+ }
else
{
type->set_code (TYPE_CODE_STRUCT); @@ -15823,7 +15840,8 @@ handle_struct_member_die (struct die_info *child_die, struct type *type,
struct dwarf2_cu *cu)
{
if (child_die->tag == DW_TAG_member
- || child_die->tag == DW_TAG_variable)
+ || child_die->tag == DW_TAG_variable
+ || child_die->tag == DW_TAG_namelist_item)
{
/* NOTE: carlton/2002-11-05: A C++ static data member
should be a DW_TAG_member that is a declaration, but @@ -15867,7 +15885,9 @@ handle_struct_member_die (struct die_info *child_die, struct type *type, }
/* Finish creating a structure or union type, including filling in
- its members and creating a symbol for it. */
+ its members and creating a symbol for it. This function also
+ handles Fortran namelist variable, its items or members and
+ creating a symbol for it. */
static void
process_structure_scope (struct die_info *die, struct dwarf2_cu *cu) @@ -21971,8 +21991,17 @@ new_symbol (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
case DW_TAG_union_type:
case DW_TAG_set_type:
case DW_TAG_enumeration_type:
- SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
- SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
+ case DW_TAG_namelist:
+ if (die->tag == DW_TAG_namelist)
+ {
+ SYMBOL_ACLASS_INDEX (sym) = LOC_STATIC;
+ SYMBOL_DOMAIN (sym) = VAR_DOMAIN;
+ }
+ else
+ {
+ SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
+ SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
+ }
{
/* NOTE: carlton/2003-11-10: C++ class symbols shouldn't @@ -22909,6 +22938,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
&& die->tag != DW_TAG_class_type
&& die->tag != DW_TAG_interface_type
&& die->tag != DW_TAG_structure_type
+ && die->tag != DW_TAG_namelist
&& die->tag != DW_TAG_union_type)
return NULL;
@@ -22933,6 +22963,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
case DW_TAG_interface_type:
case DW_TAG_structure_type:
case DW_TAG_union_type:
+ case DW_TAG_namelist:
/* Some GCC versions emit spurious DW_AT_name attributes for unnamed
structures or unions. These were of the form "._%d" in GCC 4.1,
or simply "<anonymous struct>" or "<anonymous union>" in GCC 4.3 diff --git a/gdb/f-typeprint.c b/gdb/f-typeprint.c index 1791cb29451..5b34622dacb 100644
--- a/gdb/f-typeprint.c
+++ b/gdb/f-typeprint.c
@@ -121,6 +121,7 @@ f_language::f_type_print_varspec_prefix (struct type *type,
case TYPE_CODE_UNDEF:
case TYPE_CODE_STRUCT:
case TYPE_CODE_UNION:
+ case TYPE_CODE_NAMELIST:
case TYPE_CODE_ENUM:
case TYPE_CODE_INT:
case TYPE_CODE_FLT:
@@ -261,6 +262,7 @@ f_language::f_type_print_varspec_suffix (struct type *type,
case TYPE_CODE_UNDEF:
case TYPE_CODE_STRUCT:
case TYPE_CODE_UNION:
+ case TYPE_CODE_NAMELIST:
case TYPE_CODE_ENUM:
case TYPE_CODE_INT:
case TYPE_CODE_FLT:
@@ -305,7 +307,8 @@ f_language::f_type_print_base (struct type *type, struct ui_file *stream,
const char *prefix = "";
if (type->code () == TYPE_CODE_UNION)
prefix = "Type, C_Union :: ";
- else if (type->code () == TYPE_CODE_STRUCT)
+ else if (type->code () == TYPE_CODE_STRUCT
+ || type->code () == TYPE_CODE_NAMELIST)
prefix = "Type ";
fprintf_filtered (stream, "%*s%s%s", level, "", prefix, type->name ());
return;
@@ -391,6 +394,7 @@ f_language::f_type_print_base (struct type *type, struct ui_file *stream,
case TYPE_CODE_STRUCT:
case TYPE_CODE_UNION:
+ case TYPE_CODE_NAMELIST:
if (type->code () == TYPE_CODE_UNION)
fprintf_filtered (stream, "%*sType, C_Union :: ", level, "");
else
diff --git a/gdb/f-valprint.c b/gdb/f-valprint.c index 27d9a730978..6f80b5a12ff 100644
--- a/gdb/f-valprint.c
+++ b/gdb/f-valprint.c
@@ -295,6 +295,7 @@ f_language::value_print_inner (struct value *val, struct ui_file *stream,
case TYPE_CODE_STRUCT:
case TYPE_CODE_UNION:
+ case TYPE_CODE_NAMELIST:
/* Starting from the Fortran 90 standard, Fortran supports derived
types. */
fprintf_filtered (stream, "( ");
@@ -320,6 +321,16 @@ f_language::value_print_inner (struct value *val, struct ui_file *stream,
fputs_filtered (" = ", stream);
}
+ /* While printing namelist items, fetch the appropriate value
+ field before printing its value. */
+ if (type->code () == TYPE_CODE_NAMELIST)
+ {
+ struct block_symbol symni = lookup_symbol (field_name,
+ get_selected_block (0), VAR_DOMAIN, nullptr);
+ if (symni.symbol != nullptr)
+ field = value_of_variable (symni.symbol, symni.block);
+ }
+
common_val_print (field, stream, recurse + 1,
options, current_language);
diff --git a/gdb/gdbtypes.h b/gdb/gdbtypes.h index dc575c42996..ba8a61987db 100644
--- a/gdb/gdbtypes.h
+++ b/gdb/gdbtypes.h
@@ -195,6 +195,8 @@ enum type_code
/* * Fixed Point type. */
TYPE_CODE_FIXED_POINT,
+
+ TYPE_CODE_NAMELIST, /**< Fortran namelist. */
};
/* * Some bits for the type's instance_flags word. See the macros diff --git a/gdb/testsuite/gdb.fortran/namelist.exp b/gdb/testsuite/gdb.fortran/namelist.exp
new file mode 100644
index 00000000000..90762928455
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/namelist.exp
@@ -0,0 +1,49 @@
+# Copyright (C) 2021 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or #
+(at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, # but
+WITHOUT ANY WARRANTY; without even the implied warranty of #
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU
+General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License #
+along with this program. If not, see <http://www.gnu.org/licenses/>.
+
+# This file is part of the gdb testsuite. It contains tests for
+fortran # namelist.
+
+if { [skip_fortran_tests] } { return -1 }
+
+standard_testfile .f90
+load_lib "fortran.exp"
+
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile {debug f90}]} {
+ return -1
+}
+
+if ![fortran_runto_main] then {
+ perror "couldn't run to main"
+ continue
+}
+
+# Depending on the compiler being used, the type names can be printed differently.
+set int [fortran_int4]
+
+gdb_breakpoint [gdb_get_line_number "Display namelist"]
+gdb_continue_to_breakpoint "Display namelist"
+
+if {[test_compiler_info {gcc-*}]} {
+ gdb_test "ptype nml" \
+ "type = Type nml\r\n *$int :: a\r\n *$int :: b\r\n *End Type nml"
+ gdb_test "print nml" \
+ "\\$\[0-9\]+ = \\( a = 10, b = 20 \\)"
+} else {
+ gdb_test "ptype nml" \
+ "No symbol \"nml\" in current context\\."
+ gdb_test "print nml" \
+ "No symbol \"nml\" in current context\\."
+}
diff --git a/gdb/testsuite/gdb.fortran/namelist.f90 b/gdb/testsuite/gdb.fortran/namelist.f90
new file mode 100644
index 00000000000..fb36690d765
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/namelist.f90
@@ -0,0 +1,27 @@
+! Copyright (C) 2021 Free Software Foundation, Inc.
+!
+! This program is free software; you can redistribute it and/or modify
+! it under the terms of the GNU General Public License as published by
+! the Free Software Foundation; either version 3 of the License, or !
+(at your option) any later version.
+!
+! This program is distributed in the hope that it will be useful, ! but
+WITHOUT ANY WARRANTY; without even the implied warranty of !
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ! GNU
+General Public License for more details.
+!
+! You should have received a copy of the GNU General Public License !
+along with this program. If not, see <http://www.gnu.org/licenses/>.
+!
+! This file is the Fortran source file for namelist.exp.
+
+program main
+
+ integer :: a, b
+ namelist /nml/ a, b
+
+ a = 10
+ b = 20
+ Write(*,nml) ! Display namelist
+
+end program main
diff --git a/include/dwarf2.def b/include/dwarf2.def index 1ae6e1df298..6b8be1f6a16 100644
--- a/include/dwarf2.def
+++ b/include/dwarf2.def
@@ -289,7 +289,7 @@ DW_AT (DW_AT_frame_base, 0x40) DW_AT (DW_AT_friend, 0x41) DW_AT (DW_AT_identifier_case, 0x42) DW_AT (DW_AT_macro_info, 0x43) -DW_AT (DW_AT_namelist_items, 0x44)
+DW_AT (DW_AT_namelist_item, 0x44)
DW_AT (DW_AT_priority, 0x45)
DW_AT (DW_AT_segment, 0x46)
DW_AT (DW_AT_specification, 0x47)
--
2.17.1
-----Original Message-----
From: Andrew Burgess <andrew.burgess@embecosm.com>
Sent: Wednesday, October 6, 2021 9:22 PM
To: Kumar N, Bhuvanendra <Bhuvanendra.KumarN@amd.com>
Cc: gdb-patches@sourceware.org; George, Jini Susan <JiniSusan.George@amd.com>; Achra, Nitika <Nitika.Achra@amd.com>; Sharma, Alok Kumar <AlokKumar.Sharma@amd.com>; E, Nagajyothi <Nagajyothi.E@amd.com>
Subject: Re: [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
[CAUTION: External Email]
* Kumar N, Bhuvanendra <Bhuvanendra.KumarN@amd.com> [2021-09-22 12:38:20 +0000]:
> [AMD Official Use Only]
>
> Hi Andrew,
>
> Thanks a lot for your review comments, I have addressed all the review comments and tested. Updated patch is attached and also inlined with this email.
>
> >I'm a little uncomfortable with this if condition. This feels like a bit of a hack. Is there precedent anywhere else in GDB for using the type specific field like this?
> >Wouldn't adding a TYPE_CODE_NAMELIST be more in keeping with how GDB currently does things, though I understand this is likely to make the patch slightly bigger. Did you consider this approach?
>
> Regarding your major comment, I did considered introducing a new type
> specific field for namelist earlier, instead of using any of those
> existing. Reason I did not included in my earlier patch was,
> type_specific_kind is declared with 3 bits in struct main_type and
> already there were 8 values, hence deferred expanding
> type_specific_kind etc.... But now I have done that, thanks for the
> suggestion.
Except that wasn't quite what I suggested; I actually asked whether namelist's should become their own type altogether, notice I suggested TYPE_CODE_NAMELIST.
Though what you have here is nice in its simplicity, but I can't shake the feeling that its not the right solution.
I'll need to think about this some more - maybe others have an opinion?
Thanks,
Andrew
>
> Regards,
> bhuvan
>
> PATH inlined:
>
> From 59053daf1018c3e4bcae8cd342d46b00f4f03648 Mon Sep 17 00:00:00 2001
> From: =?UTF-8?q?=E2=80=9Cbhkumarn=E2=80=9D?=
> <Bhuvanendra.KumarN@amd.com>
> Date: Wed, 22 Sep 2021 17:52:15 +0530
> Subject: [PATCH] Fix ptype and print commands for namelist variables(a
> fortran feature).
>
> GCC/gfortran support namelist(a fortran feature), it emits
> DW_TAG_namelist and DW_TAG_namelist_item dies. But gdb does not
> process these dies and support namelist variables during print and
> ptype commands. When tried to print, it bails out with the error message as shown below.
> (gdb) print nml
> No symbol "nml" in current context.
> This commit is to make the print and ptype commands work for namelist
> variables and its items. Sample output of these commands is shared
> below, with fixed gdb.
> (gdb) ptype nml
> type = Type nml
> integer(kind=4) :: a
> integer(kind=4) :: b
> End Type nml
> (gdb) print nml
> $1 = ( a = 10, b = 20 )
> ---
> gdb/ChangeLog | 11 ++++++
> gdb/dwarf2/read.c | 44 +++++++++++++++++++----
> gdb/f-valprint.c | 10 ++++++
> gdb/gdbtypes.h | 3 +-
> gdb/testsuite/ChangeLog | 5 +++
> gdb/testsuite/gdb.fortran/namelist.exp | 49
> ++++++++++++++++++++++++++
> gdb/testsuite/gdb.fortran/namelist.f90 | 27 ++++++++++++++
> include/dwarf2.def | 2 +-
> 8 files changed, 142 insertions(+), 9 deletions(-) create mode
> 100644 gdb/testsuite/gdb.fortran/namelist.exp
> create mode 100644 gdb/testsuite/gdb.fortran/namelist.f90
>
> diff --git a/gdb/ChangeLog b/gdb/ChangeLog index
> 36cb4c9e7e9..ec01c2957e9 100644
> --- a/gdb/ChangeLog
> +++ b/gdb/ChangeLog
> @@ -1,3 +1,14 @@
> +2021-08-23 Bhuvanendra Kumar N <Bhuvanendra.KumarN@amd.com>
> +
> + * dwarf2/read.c (process_die): Add new case for namelist.
> + (dwarf2_add_field): Process DW_TAG_namelist_item die.
> + (read_structure_type, handle_struct_member_die, new_symbol)
> + (dwarf2_name): Update.
> + * f-valprint.c (f_language::value_print_inner): Add support for
> + printing namelist items.
> + * include/dwarf2.def: (DW_AT_namelist_items): Renamed to ...
> + (DW_AT_namelist_item): ... this. As per dwarf standard.
> +
> 2021-06-08 Lancelot Six <lsix@lancelotsix.com>
>
> * python/lib/gdb/FrameDecorator.py (FrameDecorator): Use 'is None'
> diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c index
> 96009f1418f..88db06ec9e7 100644
> --- a/gdb/dwarf2/read.c
> +++ b/gdb/dwarf2/read.c
> @@ -9570,6 +9570,7 @@ process_die (struct die_info *die, struct dwarf2_cu *cu)
> case DW_TAG_interface_type:
> case DW_TAG_structure_type:
> case DW_TAG_union_type:
> + case DW_TAG_namelist:
> process_structure_scope (die, cu);
> break;
> case DW_TAG_enumeration_type:
> @@ -14417,8 +14418,20 @@ dwarf2_add_field (struct field_info *fip,
> struct die_info *die,
>
> fp = &new_field->field;
>
> - if (die->tag == DW_TAG_member && ! die_is_declaration (die, cu))
> - {
> + if ((die->tag == DW_TAG_member || die->tag == DW_TAG_namelist_item)
> + && ! die_is_declaration (die, cu))
> + {
> + /* For the DW_TAG_namelist_item die, use the referenced die. */
> + if (die->tag == DW_TAG_namelist_item)
> + {
> + struct attribute *attr1 = dwarf2_attr (die, DW_AT_namelist_item, cu);
> + struct die_info *item_die = nullptr;
> + struct dwarf2_cu *item_cu = cu;
> + if (attr1->form_is_ref ())
> + item_die = follow_die_ref (die, attr1, &item_cu);
> + if (item_die != nullptr)
> + die = item_die;
> + }
> /* Data member other than a C++ static data member. */
>
> /* Get type of field. */
> @@ -15448,7 +15461,10 @@ read_structure_type (struct die_info *die, struct dwarf2_cu *cu)
> }
>
> type = alloc_type (objfile);
> - INIT_CPLUS_SPECIFIC (type);
> + if (die->tag == DW_TAG_namelist)
> + TYPE_SPECIFIC_FIELD (type) = TYPE_SPECIFIC_NAMELIST; else
> + INIT_CPLUS_SPECIFIC (type);
>
> name = dwarf2_name (die, cu);
> if (name != NULL)
> @@ -15684,7 +15700,8 @@ handle_struct_member_die (struct die_info *child_die, struct type *type,
> struct dwarf2_cu *cu) {
> if (child_die->tag == DW_TAG_member
> - || child_die->tag == DW_TAG_variable)
> + || child_die->tag == DW_TAG_variable
> + || child_die->tag == DW_TAG_namelist_item)
> {
> /* NOTE: carlton/2002-11-05: A C++ static data member
> should be a DW_TAG_member that is a declaration, but @@
> -15728,7 +15745,9 @@ handle_struct_member_die (struct die_info
> *child_die, struct type *type, }
>
> /* Finish creating a structure or union type, including filling in
> - its members and creating a symbol for it. */
> + its members and creating a symbol for it. This function also
> + handles Fortran namelist variable, its items or members and
> + creating a symbol for it. */
>
> static void
> process_structure_scope (struct die_info *die, struct dwarf2_cu *cu)
> @@ -21807,8 +21826,17 @@ new_symbol (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
> case DW_TAG_union_type:
> case DW_TAG_set_type:
> case DW_TAG_enumeration_type:
> - SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> - SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
> + case DW_TAG_namelist:
> + if (die->tag == DW_TAG_namelist)
> + {
> + SYMBOL_ACLASS_INDEX (sym) = LOC_STATIC;
> + SYMBOL_DOMAIN (sym) = VAR_DOMAIN;
> + }
> + else
> + {
> + SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> + SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
> + }
>
> {
> /* NOTE: carlton/2003-11-10: C++ class symbols shouldn't @@
> -22744,6 +22772,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
> && die->tag != DW_TAG_class_type
> && die->tag != DW_TAG_interface_type
> && die->tag != DW_TAG_structure_type
> + && die->tag != DW_TAG_namelist
> && die->tag != DW_TAG_union_type)
> return NULL;
>
> @@ -22768,6 +22797,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
> case DW_TAG_interface_type:
> case DW_TAG_structure_type:
> case DW_TAG_union_type:
> + case DW_TAG_namelist:
> /* Some GCC versions emit spurious DW_AT_name attributes for unnamed
> structures or unions. These were of the form "._%d" in GCC 4.1,
> or simply "<anonymous struct>" or "<anonymous union>" in GCC
> 4.3 diff --git a/gdb/f-valprint.c b/gdb/f-valprint.c index
> 240daaf34f9..aa86fbc901e 100644
> --- a/gdb/f-valprint.c
> +++ b/gdb/f-valprint.c
> @@ -320,6 +320,16 @@ f_language::value_print_inner (struct value *val, struct ui_file *stream,
> fputs_filtered (" = ", stream);
> }
>
> + /* While printing namelist items, fetch the appropriate value
> + field before printing its value. */
> + if (TYPE_SPECIFIC_FIELD (type) == TYPE_SPECIFIC_NAMELIST)
> + {
> + struct block_symbol symni = lookup_symbol (field_name,
> + get_selected_block (0), VAR_DOMAIN, nullptr);
> + if (symni.symbol != nullptr)
> + field = value_of_variable (symni.symbol, symni.block);
> + }
> +
> common_val_print (field, stream, recurse + 1,
> options, current_language);
>
> diff --git a/gdb/gdbtypes.h b/gdb/gdbtypes.h index
> 0cc00e74a20..124282c3d53 100644
> --- a/gdb/gdbtypes.h
> +++ b/gdb/gdbtypes.h
> @@ -607,6 +607,7 @@ enum type_specific_kind
> TYPE_SPECIFIC_SELF_TYPE,
> TYPE_SPECIFIC_INT,
> TYPE_SPECIFIC_FIXED_POINT,
> + TYPE_SPECIFIC_NAMELIST,
> };
>
> union type_owner
> @@ -833,7 +834,7 @@ struct main_type
> /* * A discriminant telling us which field of the type_specific
> union is being used for this type, if any. */
>
> - ENUM_BITFIELD(type_specific_kind) type_specific_field : 3;
> + ENUM_BITFIELD(type_specific_kind) type_specific_field : 4;
>
> /* * Number of fields described for this type. This field appears
> at this location because it packs nicely here. */ diff --git
> a/gdb/testsuite/ChangeLog b/gdb/testsuite/ChangeLog index
> 87cf3e2a061..33f60c29b3c 100644
> --- a/gdb/testsuite/ChangeLog
> +++ b/gdb/testsuite/ChangeLog
> @@ -1,3 +1,8 @@
> +2021-07-26 Bhuvanendra Kumar N <Bhuvanendra.KumarN@amd.com>
> +
> + * gdb.fortran/namelist.exp: New file.
> + * gdb.fortran/namelist.f90: New file.
> +
> 2021-06-10 Bhuvanendra Kumar N <Bhuvanendra.KumarN@amd.com>
>
> * gdb.fortran/ptype-on-functions.exp: Add type info of formal
> diff --git a/gdb/testsuite/gdb.fortran/namelist.exp
> b/gdb/testsuite/gdb.fortran/namelist.exp
> new file mode 100644
> index 00000000000..90762928455
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/namelist.exp
> @@ -0,0 +1,49 @@
> +# Copyright (C) 2021 Free Software Foundation, Inc.
> +
> +# This program is free software; you can redistribute it and/or
> +modify # it under the terms of the GNU General Public License as
> +published by # the Free Software Foundation; either version 3 of the
> +License, or # (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful, #
> +but WITHOUT ANY WARRANTY; without even the implied warranty of #
> +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU
> +General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License #
> +along with this program. If not, see <https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.gnu.org%2Flicenses%2F&data=04%7C01%7CBhuvanendra.KumarN%40amd.com%7C1b602ae2197248e5a9a008d988e12ef7%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637691323019155507%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=MOunn7GZCinobEL44IujJYFEaB83wh9TWkuufnnjbQ8%3D&reserved=0>.
> +
> +# This file is part of the gdb testsuite. It contains tests for
> +fortran # namelist.
> +
> +if { [skip_fortran_tests] } { return -1 }
> +
> +standard_testfile .f90
> +load_lib "fortran.exp"
> +
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile {debug f90}]} {
> + return -1
> +}
> +
> +if ![fortran_runto_main] then {
> + perror "couldn't run to main"
> + continue
> +}
> +
> +# Depending on the compiler being used, the type names can be printed differently.
> +set int [fortran_int4]
> +
> +gdb_breakpoint [gdb_get_line_number "Display namelist"]
> +gdb_continue_to_breakpoint "Display namelist"
> +
> +if {[test_compiler_info {gcc-*}]} {
> + gdb_test "ptype nml" \
> + "type = Type nml\r\n *$int :: a\r\n *$int :: b\r\n *End Type nml"
> + gdb_test "print nml" \
> + "\\$\[0-9\]+ = \\( a = 10, b = 20 \\)"
> +} else {
> + gdb_test "ptype nml" \
> + "No symbol \"nml\" in current context\\."
> + gdb_test "print nml" \
> + "No symbol \"nml\" in current context\\."
> +}
> diff --git a/gdb/testsuite/gdb.fortran/namelist.f90
> b/gdb/testsuite/gdb.fortran/namelist.f90
> new file mode 100644
> index 00000000000..fb36690d765
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/namelist.f90
> @@ -0,0 +1,27 @@
> +! Copyright (C) 2021 Free Software Foundation, Inc.
> +!
> +! This program is free software; you can redistribute it and/or
> +modify ! it under the terms of the GNU General Public License as
> +published by ! the Free Software Foundation; either version 3 of the
> +License, or ! (at your option) any later version.
> +!
> +! This program is distributed in the hope that it will be useful, !
> +but WITHOUT ANY WARRANTY; without even the implied warranty of !
> +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ! GNU
> +General Public License for more details.
> +!
> +! You should have received a copy of the GNU General Public License !
> +along with this program. If not, see <https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.gnu.org%2Flicenses%2F&data=04%7C01%7CBhuvanendra.KumarN%40amd.com%7C1b602ae2197248e5a9a008d988e12ef7%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637691323019155507%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=MOunn7GZCinobEL44IujJYFEaB83wh9TWkuufnnjbQ8%3D&reserved=0>.
> +!
> +! This file is the Fortran source file for namelist.exp.
> +
> +program main
> +
> + integer :: a, b
> + namelist /nml/ a, b
> +
> + a = 10
> + b = 20
> + Write(*,nml) ! Display namelist
> +
> +end program main
> diff --git a/include/dwarf2.def b/include/dwarf2.def index
> 1ae6e1df298..6b8be1f6a16 100644
> --- a/include/dwarf2.def
> +++ b/include/dwarf2.def
> @@ -289,7 +289,7 @@ DW_AT (DW_AT_frame_base, 0x40) DW_AT
> (DW_AT_friend, 0x41) DW_AT (DW_AT_identifier_case, 0x42) DW_AT
> (DW_AT_macro_info, 0x43) -DW_AT (DW_AT_namelist_items, 0x44)
> +DW_AT (DW_AT_namelist_item, 0x44)
> DW_AT (DW_AT_priority, 0x45)
> DW_AT (DW_AT_segment, 0x46)
> DW_AT (DW_AT_specification, 0x47)
> --
> 2.17.1
>
>
> -----Original Message-----
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Sent: Monday, September 20, 2021 3:11 PM
> To: Kumar N, Bhuvanendra <Bhuvanendra.KumarN@amd.com>
> Cc: gdb-patches@sourceware.org; George, Jini Susan
> <JiniSusan.George@amd.com>; Achra, Nitika <Nitika.Achra@amd.com>;
> Sharma, Alok Kumar <AlokKumar.Sharma@amd.com>; E, Nagajyothi
> <Nagajyothi.E@amd.com>
> Subject: Re: [PATCH] Fix ptype and print commands for namelist
> variables(a fortran feature)
>
> [CAUTION: External Email]
>
> * Kumar N, Bhuvanendra via Gdb-patches <gdb-patches@sourceware.org> [2021-08-24 09:16:19 +0000]:
>
> > [AMD Official Use Only]
> >
> > Hi all,
> >
> > Requesting code review for this GDB patch. Required patch is attached and also inlined below with this email.
> >
> > Problem description/summary:
> >
> > GCC/gfortran support namelist(a fortran feature), it emits DW_TAG_namelist and DW_TAG_namelist_item dies. But gdb does not process these dies and support namelist variables during print and ptype commands. When tried to print, it bails out with the error message as shown below.
> > (gdb) print nml
> > No symbol "nml" in current context.
> >
> > Fix details:
> >
> > This fix is to make the print and ptype commands work for namelist variables and its items. Sample output of these commands is shared below, with fixed gdb.
> >
> > (gdb) ptype nml
> > type = Type nml
> > integer(kind=4) :: a
> > integer(kind=4) :: b
> > End Type nml
> > (gdb) print nml
> > $1 = ( a = 10, b = 20 )
>
> bhuvan,
>
> Thanks for working on this. I have some small style issues, but I have a bigger question which you'll find inline.
>
> Thanks,
> Andrew
>
>
> >
> > gdb/ChangeLog:
> >
> > * dwarf2/read.c (process_die): Add new case for namelist.
> > (dwarf2_add_field): Process DW_TAG_namelist_item die.
> > (read_structure_type, handle_struct_member_die, new_symbol)
> > (dwarf2_name): Update.
> > * f-valprint.c (f_language::value_print_inner): Add support for
> > printing namelist items.
> > * include/dwarf2.def: (DW_AT_namelist_items): Renamed to ...
> > (DW_AT_namelist_item): ... this. As per dwarf standard.
> >
> > gdb/testsuite/ChangeLog:
> >
> > * gdb.fortran/namelist.exp: New file.
> > * gdb.fortran/namelist.f90: New file.
> >
> > NOTE: Similarly renaming DW_AT_namelist_items to DW_AT_namelist_item as per DWARF standard naming convention in GCC/gfortran repo (https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fgcc-mirror%2Fgcc&data=04%7C01%7CBhuvanendra.KumarN%40amd.com%7C1b602ae2197248e5a9a008d988e12ef7%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637691323019155507%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=K8mncjSuCt9BEaGJTnlJ9qp0%2FpFSVvfkVxbe%2FNFrej0%3D&reserved=0) will be handled in separate PATCH. I will raise separate patch for this.
> >
> > regards,
> > bhuvan
> >
> > Patch inlined:
> >
> > From 0775cbf3716bae9480c3f1f1d9d8860ac561929e Mon Sep 17 00:00:00
> > 2001
> > From: =?UTF-8?q?=E2=80=9Cbhkumarn=E2=80=9D?=
> > Bhuvanendra.KumarN@amd.com<mailto:Bhuvanendra.KumarN@amd.com>
> > Date: Mon, 24 Aug 2021 11:49:14 +0530
> > Subject: [PATCH] Fix ptype and print commands for namelist
> > variables(a fortran feature).
> >
> > GCC/gfortran support namelist(a fortran feature), it emits
> > DW_TAG_namelist and DW_TAG_namelist_item dies. But gdb does not
> > process these dies and support namelist variables during print and
> > ptype commands. When tried to print, it bails out with the error message as shown below.
> > (gdb) print nml
> > No symbol "nml" in current context.
> > This commit is to make the print and ptype commands work for
> > namelist variables and its items. Sample output of these commands is
> > shared below, with fixed gdb.
> > (gdb) ptype nml
> > type = Type nml
> > integer(kind=4) :: a
> > integer(kind=4) :: b
> > End Type nml
> > (gdb) print nml
> > $1 = ( a = 10, b = 20 )
> > ---
> > gdb/ChangeLog | 11 +++++++
> > gdb/dwarf2/read.c | 41 +++++++++++++++++++----
> > gdb/f-valprint.c | 10 ++++++
> > gdb/testsuite/ChangeLog | 5 +++
> > gdb/testsuite/gdb.fortran/namelist.exp | 45
> > ++++++++++++++++++++++++++
> > gdb/testsuite/gdb.fortran/namelist.f90 | 27 ++++++++++++++++
> > include/dwarf2.def | 2 +-
> > 7 files changed, 134 insertions(+), 7 deletions(-) create mode
> > 100644 gdb/testsuite/gdb.fortran/namelist.exp
> > create mode 100644 gdb/testsuite/gdb.fortran/namelist.f90
> >
> > diff --git a/gdb/ChangeLog b/gdb/ChangeLog index
> > 36cb4c9e7e9..ec01c2957e9 100644
> > --- a/gdb/ChangeLog
> > +++ b/gdb/ChangeLog
> > @@ -1,3 +1,14 @@
> > +2021-08-23 Bhuvanendra Kumar N
> > +Bhuvanendra.KumarN@amd.com<mailto:Bhuvanendra.KumarN@amd.com>
> > +
> > + * dwarf2/read.c (process_die): Add new case for namelist.
> > + (dwarf2_add_field): Process DW_TAG_namelist_item die.
> > + (read_structure_type, handle_struct_member_die, new_symbol)
> > + (dwarf2_name): Update.
> > + * f-valprint.c (f_language::value_print_inner): Add support for
> > + printing namelist items.
> > + * include/dwarf2.def: (DW_AT_namelist_items): Renamed to ...
> > + (DW_AT_namelist_item): ... this. As per dwarf standard.
> > +
> > 2021-06-08 Lancelot Six lsix@lancelotsix.com<mailto:lsix@lancelotsix.com>
> > * python/lib/gdb/FrameDecorator.py (FrameDecorator): Use 'is None'
> > diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c index
> > 96009f1418f..54528d67498 100644
> > --- a/gdb/dwarf2/read.c
> > +++ b/gdb/dwarf2/read.c
> > @@ -9570,6 +9570,7 @@ process_die (struct die_info *die, struct dwarf2_cu *cu)
> > case DW_TAG_interface_type:
> > case DW_TAG_structure_type:
> > case DW_TAG_union_type:
> > + case DW_TAG_namelist:
> > process_structure_scope (die, cu);
> > break;
> > case DW_TAG_enumeration_type:
> > @@ -14417,8 +14418,20 @@ dwarf2_add_field (struct field_info *fip, struct die_info *die,
> > fp = &new_field->field;
> > - if (die->tag == DW_TAG_member && ! die_is_declaration (die, cu))
> > - {
> > + if ((die->tag == DW_TAG_member || die->tag == DW_TAG_namelist_item) &&
> > + ! die_is_declaration (die, cu))
>
> The '&&' operator should start the line.
>
> > + {
> > + /* For the DW_TAG_namelist_item die, use the referenced die. */
> > + if (die->tag == DW_TAG_namelist_item)
> > + {
> > + struct attribute *attr1 = dwarf2_attr (die, DW_AT_namelist_item, cu);
> > + struct die_info *item_die = NULL;
>
> Throughout this patch, please use nullptr instead of NULL.
>
> > + struct dwarf2_cu *item_cu = cu;
> > + if (attr1->form_is_ref ())
> > + item_die = follow_die_ref (die, attr1, &item_cu);
> > + if (item_die != NULL)
> > + die = item_die;
> > + }
> > /* Data member other than a C++ static data member. */
> > /* Get type of field. */
> > @@ -15449,6 +15462,8 @@ read_structure_type (struct die_info *die, struct dwarf2_cu *cu)
> > type = alloc_type (objfile);
> > INIT_CPLUS_SPECIFIC (type);
> > + if (die->tag == DW_TAG_namelist)
> > + INIT_NONE_SPECIFIC (type);
>
> I think you should use if/then/else and place INIT_CPLUS_SPECIFIC in the else block.
>
> > name = dwarf2_name (die, cu);
> > if (name != NULL)
> > @@ -15684,7 +15699,8 @@ handle_struct_member_die (struct die_info *child_die, struct type *type,
> > struct dwarf2_cu *cu) {
> > if (child_die->tag == DW_TAG_member
> > - || child_die->tag == DW_TAG_variable)
> > + || child_die->tag == DW_TAG_variable
> > + || child_die->tag == DW_TAG_namelist_item)
> > {
> > /* NOTE: carlton/2002-11-05: A C++ static data member
> > should be a DW_TAG_member that is a declaration, but @@
> > -15728,7
> > +15744,9 @@ handle_struct_member_die (struct die_info *child_die,
> > struct type *type, }
> > /* Finish creating a structure or union type, including filling in
> > - its members and creating a symbol for it. */
> > + its members and creating a symbol for it. This function also
> > + handles Fortran namelist variable, its items or members and
> > + creating a symbol for it. */
> > static void
> > process_structure_scope (struct die_info *die, struct dwarf2_cu *cu)
> > @@ -21807,8 +21825,17 @@ new_symbol (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
> > case DW_TAG_union_type:
> > case DW_TAG_set_type:
> > case DW_TAG_enumeration_type:
> > - SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> > - SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
> > + case DW_TAG_namelist:
> > + if (die->tag == DW_TAG_namelist)
> > + {
> > + SYMBOL_ACLASS_INDEX (sym) = LOC_STATIC;
> > + SYMBOL_DOMAIN (sym) = VAR_DOMAIN;
> > + }
> > + else
> > + {
> > + SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> > + SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
> > + }
> > {
> > /* NOTE: carlton/2003-11-10: C++ class symbols shouldn't @@
> > -22744,6 +22771,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
> > && die->tag != DW_TAG_class_type
> > && die->tag != DW_TAG_interface_type
> > && die->tag != DW_TAG_structure_type
> > + && die->tag != DW_TAG_namelist
> > && die->tag != DW_TAG_union_type)
> > return NULL;
> > @@ -22768,6 +22796,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
> > case DW_TAG_interface_type:
> > case DW_TAG_structure_type:
> > case DW_TAG_union_type:
> > + case DW_TAG_namelist:
> > /* Some GCC versions emit spurious DW_AT_name attributes for unnamed
> > structures or unions. These were of the form "._%d" in GCC 4.1,
> > or simply "<anonymous struct>" or "<anonymous union>" in GCC
> > 4.3 diff --git a/gdb/f-valprint.c b/gdb/f-valprint.c index
> > 240daaf34f9..8ed35e2fb1f 100644
> > --- a/gdb/f-valprint.c
> > +++ b/gdb/f-valprint.c
> > @@ -320,6 +320,16 @@ f_language::value_print_inner (struct value *val, struct ui_file *stream,
> > fputs_filtered (" = ", stream);
> > }
> > + /* While printing namelist items, fetch the appropriate value
> > + field before printing its value. */
> > + if (TYPE_SPECIFIC_FIELD (type) == TYPE_SPECIFIC_NONE)
>
> I'm a little uncomfortable with this if condition. This feels like a bit of a hack. Is there precedent anywhere else in GDB for using the type specific field like this?
>
> Wouldn't adding a TYPE_CODE_NAMELIST be more in keeping with how GDB currently does things, though I understand this is likely to make the patch slightly bigger. Did you consider this approach?
>
>
>
>
> > + {
> > + struct block_symbol symni = lookup_symbol(field_name,
> > + get_selected_block (0), VAR_DOMAIN, nullptr);
>
> There's a missing space before '(' here, and two lines below.
>
> > + if (symni.symbol != NULL)
> > + field = value_of_variable(symni.symbol, symni.block);
> > + }
> > +
> > common_val_print (field, stream, recurse + 1,
> > options, current_language); diff --git
> > a/gdb/testsuite/ChangeLog b/gdb/testsuite/ChangeLog index
> > 87cf3e2a061..33f60c29b3c 100644
> > --- a/gdb/testsuite/ChangeLog
> > +++ b/gdb/testsuite/ChangeLog
> > @@ -1,3 +1,8 @@
> > +2021-07-26 Bhuvanendra Kumar N
> > +Bhuvanendra.KumarN@amd.com<mailto:Bhuvanendra.KumarN@amd.com>
> > +
> > + * gdb.fortran/namelist.exp: New file.
> > + * gdb.fortran/namelist.f90: New file.
> > +
> > 2021-06-10 Bhuvanendra Kumar N Bhuvanendra.KumarN@amd.com<mailto:Bhuvanendra.KumarN@amd.com>
> > * gdb.fortran/ptype-on-functions.exp: Add type info of formal
> > diff --git a/gdb/testsuite/gdb.fortran/namelist.exp
> > b/gdb/testsuite/gdb.fortran/namelist.exp
> > new file mode 100644
> > index 00000000000..e4df8c7debb
> > --- /dev/null
> > +++ b/gdb/testsuite/gdb.fortran/namelist.exp
> > @@ -0,0 +1,45 @@
> > +# Copyright 2020-2021 Free Software Foundation, Inc.
>
> The '2020' date is only needed if the patch was original posted in 2020, or if this file is copied/based on some other in-tree file that is copyright 2020+.
>
> > +
> > +# This program is free software; you can redistribute it and/or
> > +modify # it under the terms of the GNU General Public License as
> > +published by # the Free Software Foundation; either version 3 of
> > +the License, or # (at your option) any later version.
> > +#
> > +# This program is distributed in the hope that it will be useful, #
> > +but WITHOUT ANY WARRANTY; without even the implied warranty of #
> > +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU
> > +General Public License for more details.
> > +#
> > +# You should have received a copy of the GNU General Public License
> > +# along with this program. If not, see https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.gnu.org%2Flicenses%2F&data=04%7C01%7CBhuvanendra.KumarN%40amd.com%7C1b602ae2197248e5a9a008d988e12ef7%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637691323019160499%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=KoXo6%2BRQEYSwRQ6VV%2Batjx0FoVfnufps0RJnK88k2y0%3D&reserved=0.
> > +
> > +# This file is part of the gdb testsuite. It contains tests for
> > +fortran # namelist.
> > +
> > +if { [skip_fortran_tests] } { return -1 }
> > +
> > +standard_testfile .f90
> > +load_lib "fortran.exp"
> > +
> > +if {[prepare_for_testing "failed to prepare" $testfile $srcfile {debug f90}]} {
> > + return -1
> > +}
> > +
> > +if ![fortran_runto_main] then {
> > + perror "couldn't run to main"
> > + continue
> > +}
> > +
> > +# Depending on the compiler being used, the type names can be printed differently.
> > +set int [fortran_int4]
> > +
> > +gdb_breakpoint [gdb_get_line_number "Display namelist"]
> > +gdb_continue_to_breakpoint "Display namelist"
> > +
> > +if {[test_compiler_info {gcc-*}]} {
> > + gdb_test "ptype nml" "type = Type nml\r\n *$int :: a\r\n *$int :: b\r\n *End Type nml"
> > + gdb_test "print nml" \\$\[0-9\]+ = \\( a = 10, b = 20
> > +\\)<file://$/%5b0-9/%5d+%20=%20/(%20a%20=%2010,%20b%20=%2020%20/)>
>
> The pattern should be wrapped onto the newline.
>
> Thanks,
> Andrew
>
> > +} else {
> > + gdb_test "ptype nml" "No symbol \"nml\" in current context\\."
> > + gdb_test "print nml" "No symbol \"nml\" in current context\\."
> > +}
> > diff --git a/gdb/testsuite/gdb.fortran/namelist.f90
> > b/gdb/testsuite/gdb.fortran/namelist.f90
> > new file mode 100644
> > index 00000000000..00704eddf27
> > --- /dev/null
> > +++ b/gdb/testsuite/gdb.fortran/namelist.f90
> > @@ -0,0 +1,27 @@
> > +! Copyright 2020-2021 Free Software Foundation, Inc.
> > +!
> > +! This program is free software; you can redistribute it and/or
> > +modify ! it under the terms of the GNU General Public License as
> > +published by ! the Free Software Foundation; either version 3 of
> > +the License, or ! (at your option) any later version.
> > +!
> > +! This program is distributed in the hope that it will be useful, !
> > +but WITHOUT ANY WARRANTY; without even the implied warranty of !
> > +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ! GNU
> > +General Public License for more details.
> > +!
> > +! You should have received a copy of the GNU General Public License !
> > +along with this program. If not, see https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.gnu.org%2Flicenses%2F&data=04%7C01%7CBhuvanendra.KumarN%40amd.com%7C1b602ae2197248e5a9a008d988e12ef7%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637691323019160499%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=KoXo6%2BRQEYSwRQ6VV%2Batjx0FoVfnufps0RJnK88k2y0%3D&reserved=0.
> > +!
> > +! This file is the Fortran source file for namelist.exp.
> > +
> > +program main
> > +
> > + integer :: a, b
> > + namelist /nml/ a, b
> > +
> > + a = 10
> > + b = 20
> > + Write(*,nml) ! Display namelist
> > +
> > +end program main
> > diff --git a/include/dwarf2.def b/include/dwarf2.def index
> > 1ae6e1df298..6b8be1f6a16 100644
> > --- a/include/dwarf2.def
> > +++ b/include/dwarf2.def
> > @@ -289,7 +289,7 @@ DW_AT (DW_AT_frame_base, 0x40) DW_AT
> > (DW_AT_friend, 0x41) DW_AT (DW_AT_identifier_case, 0x42) DW_AT
> > (DW_AT_macro_info, 0x43) -DW_AT (DW_AT_namelist_items, 0x44)
> > +DW_AT (DW_AT_namelist_item, 0x44)
> > DW_AT (DW_AT_priority, 0x45)
> > DW_AT (DW_AT_segment, 0x46)
> > DW_AT (DW_AT_specification, 0x47)
> > --
> > 2.17.1
> >
> >
> >
>
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PING*4] [PATCH] Fix ptype and print commands for namelist variables(a fortran feature)
2021-11-11 5:32 [PING*4] [PATCH] Fix ptype and print commands for namelist variables(a fortran feature) Kumar N, Bhuvanendra
@ 2021-11-27 7:56 ` Joel Brobecker
0 siblings, 0 replies; 2+ messages in thread
From: Joel Brobecker @ 2021-11-27 7:56 UTC (permalink / raw)
To: Kumar N, Bhuvanendra via Gdb-patches
Cc: George, Jini Susan, Achra, Nitika, Sharma, Alok Kumar, Joseph,
Ancel, E, Nagajyothi, Joel Brobecker
Hi Bhuvan,
> Gentle PING*4 for the revised patch. Requesting some of you from the
> community to review the updated code patch please, i had addressed all
> the previous review comments.
I'm very sorry for the lack of feedback for this patch. I'll attempt
to help you push this patch through, knowing that I have no knowledge
of Fortran.
I only have time to do code reviews in the weekend, at the moment,
so hopefully others might be able to join in and help. But otherwise,
please bear with me.
> From 89b7f847042ccea44633f714db07400fa6dd3617 Mon Sep 17 00:00:00 2001
> From: =?UTF-8?q?=E2=80=9Cbhkumarn=E2=80=9D?= <Bhuvanendra.KumarN@amd.com>
> Date: Fri, 8 Oct 2021 02:01:18 +0530
> Subject: [PATCH] gdb/fortran: Fix ptype and print commands for namelist variables.
>
> GCC/gfortran support namelist(a fortran feature), it emits DW_TAG_namelist and DW_TAG_namelist_item dies. But gdb does not process these dies and support namelist variables during print and ptype commands. When tried to print, it bails out with the error message as shown below.
> (gdb) print nml
> No symbol "nml" in current context.
> This commit is to make the print and ptype commands work for namelist variables and its items. Sample output of these commands is shared below, with fixed gdb.
> (gdb) ptype nml
> type = Type nml
> integer(kind=4) :: a
> integer(kind=4) :: b
> End Type nml
> (gdb) print nml
> $1 = ( a = 10, b = 20 )
This is a small detail, but typically, for text which is not
a copy/paste of other things like GDB sessions, etc, we try
to keep the length of each line down to something that fits
within an 80-characters terminal. Adding the 4 characters that
commands like "git log" or "git show" indents the commit message
with, would you mind reformatting your explanations to fit within
those 76-characters per line, please?
Also, for those who are not familiar with Fortran, I think it would
be useful to have a small introduction to that feature that goes
beyond a bit of Fortran code in a testcase, and how GDB is proposed
to show it. I'd like to understand what they mean semantically,
how those are typically used, and how they are stored in memory
(to understand how they work). This will help me better review
this change.
Can you provide that kind of info?
> ---
> gdb/dwarf2/read.c | 43 ++++++++++++++++++----
> gdb/f-typeprint.c | 6 +++-
> gdb/f-valprint.c | 11 ++++++
> gdb/gdbtypes.h | 2 ++
> gdb/testsuite/gdb.fortran/namelist.exp | 49 ++++++++++++++++++++++++++
> gdb/testsuite/gdb.fortran/namelist.f90 | 27 ++++++++++++++
> include/dwarf2.def | 2 +-
> 7 files changed, 132 insertions(+), 8 deletions(-) create mode 100644 gdb/testsuite/gdb.fortran/namelist.exp
> create mode 100644 gdb/testsuite/gdb.fortran/namelist.f90
>
> diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c index cbd9a3012eb..e6cd8ed48f0 100644
> --- a/gdb/dwarf2/read.c
> +++ b/gdb/dwarf2/read.c
> @@ -9705,6 +9705,7 @@ process_die (struct die_info *die, struct dwarf2_cu *cu)
> case DW_TAG_interface_type:
> case DW_TAG_structure_type:
> case DW_TAG_union_type:
> + case DW_TAG_namelist:
> process_structure_scope (die, cu);
> break;
> case DW_TAG_enumeration_type:
> @@ -14562,8 +14563,20 @@ dwarf2_add_field (struct field_info *fip, struct die_info *die,
>
> fp = &new_field->field;
>
> - if (die->tag == DW_TAG_member && ! die_is_declaration (die, cu))
> - {
> + if ((die->tag == DW_TAG_member || die->tag == DW_TAG_namelist_item)
> + && ! die_is_declaration (die, cu))
The indentation is incorrect. The "&&" opeator should line up
with the indentation of the condition expression. Thus:
> + if ((die->tag == DW_TAG_member || die->tag == DW_TAG_namelist_item)
> + && ! die_is_declaration (die, cu))
> + {
> + /* For the DW_TAG_namelist_item die, use the referenced die. */
Can you explain why, and put that information in the comment?
Also, this is only a minor comment that you don't have to agree with,
but FWIW, when I read your comment, I intuitively thought that
namelist items were always references. On the other hand, reading
the code itself, it doesn't appear to be so.
> + if (die->tag == DW_TAG_namelist_item)
> + {
> + struct attribute *attr1 = dwarf2_attr (die, DW_AT_namelist_item, cu);
> + struct die_info *item_die = nullptr;
> + struct dwarf2_cu *item_cu = cu;
> + if (attr1->form_is_ref ())
> + item_die = follow_die_ref (die, attr1, &item_cu);
> + if (item_die != nullptr)
> + die = item_die;
> + }
> /* Data member other than a C++ static data member. */
I think it would be better if you kept that comment at the start
of the if block, so above your special handling of DW_TAG_namelist_item.
> /* Get type of field. */
> @@ -15621,6 +15634,10 @@ read_structure_type (struct die_info *die, struct dwarf2_cu *cu)
> {
> type->set_code (TYPE_CODE_UNION);
> }
> + else if (die->tag == DW_TAG_namelist)
> + {
> + type->set_code (TYPE_CODE_NAMELIST);
> + }
> else
> {
> type->set_code (TYPE_CODE_STRUCT); @@ -15823,7 +15840,8 @@ handle_struct_member_die (struct die_info *child_die, struct type *type,
> struct dwarf2_cu *cu)
> {
> if (child_die->tag == DW_TAG_member
> - || child_die->tag == DW_TAG_variable)
> + || child_die->tag == DW_TAG_variable
> + || child_die->tag == DW_TAG_namelist_item)
> {
> /* NOTE: carlton/2002-11-05: A C++ static data member
> should be a DW_TAG_member that is a declaration, but @@ -15867,7 +15885,9 @@ handle_struct_member_die (struct die_info *child_die, struct type *type, }
>
> /* Finish creating a structure or union type, including filling in
> - its members and creating a symbol for it. */
> + its members and creating a symbol for it. This function also
> + handles Fortran namelist variable, its items or members and
> + creating a symbol for it. */
>
> static void
> process_structure_scope (struct die_info *die, struct dwarf2_cu *cu) @@ -21971,8 +21991,17 @@ new_symbol (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
> case DW_TAG_union_type:
> case DW_TAG_set_type:
> case DW_TAG_enumeration_type:
> - SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> - SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
> + case DW_TAG_namelist:
> + if (die->tag == DW_TAG_namelist)
> + {
> + SYMBOL_ACLASS_INDEX (sym) = LOC_STATIC;
> + SYMBOL_DOMAIN (sym) = VAR_DOMAIN;
> + }
> + else
> + {
> + SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> + SYMBOL_DOMAIN (sym) = STRUCT_DOMAIN;
> + }
I have to say, I'm a little uncomfortable with this part of
the change -- perhaps simply because I don't understand it.
Does it make sense to handle DW_TAG_namelist here? This second
handles a group of *types*, wheres the change suggests that
namelist are more like *variables*. Should the two be handled
at the same place? What made you pick this part of the code
to insert support for DW_TAG_namelist? Or am I misunderstanding
the concept of namelists (which brings me back to the question
at the beginning)?
> {
> /* NOTE: carlton/2003-11-10: C++ class symbols shouldn't @@ -22909,6 +22938,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
> && die->tag != DW_TAG_class_type
> && die->tag != DW_TAG_interface_type
> && die->tag != DW_TAG_structure_type
> + && die->tag != DW_TAG_namelist
> && die->tag != DW_TAG_union_type)
> return NULL;
>
> @@ -22933,6 +22963,7 @@ dwarf2_name (struct die_info *die, struct dwarf2_cu *cu)
> case DW_TAG_interface_type:
> case DW_TAG_structure_type:
> case DW_TAG_union_type:
> + case DW_TAG_namelist:
> /* Some GCC versions emit spurious DW_AT_name attributes for unnamed
> structures or unions. These were of the form "._%d" in GCC 4.1,
> or simply "<anonymous struct>" or "<anonymous union>" in GCC 4.3 diff --git a/gdb/f-typeprint.c b/gdb/f-typeprint.c index 1791cb29451..5b34622dacb 100644
> --- a/gdb/f-typeprint.c
> +++ b/gdb/f-typeprint.c
> @@ -121,6 +121,7 @@ f_language::f_type_print_varspec_prefix (struct type *type,
> case TYPE_CODE_UNDEF:
> case TYPE_CODE_STRUCT:
> case TYPE_CODE_UNION:
> + case TYPE_CODE_NAMELIST:
> case TYPE_CODE_ENUM:
> case TYPE_CODE_INT:
> case TYPE_CODE_FLT:
> @@ -261,6 +262,7 @@ f_language::f_type_print_varspec_suffix (struct type *type,
> case TYPE_CODE_UNDEF:
> case TYPE_CODE_STRUCT:
> case TYPE_CODE_UNION:
> + case TYPE_CODE_NAMELIST:
> case TYPE_CODE_ENUM:
> case TYPE_CODE_INT:
> case TYPE_CODE_FLT:
> @@ -305,7 +307,8 @@ f_language::f_type_print_base (struct type *type, struct ui_file *stream,
> const char *prefix = "";
> if (type->code () == TYPE_CODE_UNION)
> prefix = "Type, C_Union :: ";
> - else if (type->code () == TYPE_CODE_STRUCT)
> + else if (type->code () == TYPE_CODE_STRUCT
> + || type->code () == TYPE_CODE_NAMELIST)
> prefix = "Type ";
> fprintf_filtered (stream, "%*s%s%s", level, "", prefix, type->name ());
> return;
> @@ -391,6 +394,7 @@ f_language::f_type_print_base (struct type *type, struct ui_file *stream,
>
> case TYPE_CODE_STRUCT:
> case TYPE_CODE_UNION:
> + case TYPE_CODE_NAMELIST:
> if (type->code () == TYPE_CODE_UNION)
> fprintf_filtered (stream, "%*sType, C_Union :: ", level, "");
> else
> diff --git a/gdb/f-valprint.c b/gdb/f-valprint.c index 27d9a730978..6f80b5a12ff 100644
> --- a/gdb/f-valprint.c
> +++ b/gdb/f-valprint.c
> @@ -295,6 +295,7 @@ f_language::value_print_inner (struct value *val, struct ui_file *stream,
>
> case TYPE_CODE_STRUCT:
> case TYPE_CODE_UNION:
> + case TYPE_CODE_NAMELIST:
> /* Starting from the Fortran 90 standard, Fortran supports derived
> types. */
> fprintf_filtered (stream, "( ");
> @@ -320,6 +321,16 @@ f_language::value_print_inner (struct value *val, struct ui_file *stream,
> fputs_filtered (" = ", stream);
> }
>
> + /* While printing namelist items, fetch the appropriate value
> + field before printing its value. */
> + if (type->code () == TYPE_CODE_NAMELIST)
> + {
> + struct block_symbol symni = lookup_symbol (field_name,
> + get_selected_block (0), VAR_DOMAIN, nullptr);
Can you fix the indentation here, please? On the second line,
the start of the parameters should be aligned with the start
of the parameters on the lines above. Also, remember that indentation
is 2 characters, not 4. You could have...
struct block_symbol symni = lookup_symbol (field_name,
get_selected_block (0),
VAR_DOMAIN, nullptr);
... or (one parameter per line) ...
struct block_symbol symni = lookup_symbol (field_name,
get_selected_block (0),
VAR_DOMAIN,
nullptr);
... or move all the parameters on the next line, with:
struct block_symbol symni = lookup_symbol
(field_name, get_selected_block (0), VAR_DOMAIN, nullptr);
> + if (symni.symbol != nullptr)
> + field = value_of_variable (symni.symbol, symni.block);
> + }
> +
> common_val_print (field, stream, recurse + 1,
> options, current_language);
>
> diff --git a/gdb/gdbtypes.h b/gdb/gdbtypes.h index dc575c42996..ba8a61987db 100644
> --- a/gdb/gdbtypes.h
> +++ b/gdb/gdbtypes.h
> @@ -195,6 +195,8 @@ enum type_code
>
> /* * Fixed Point type. */
> TYPE_CODE_FIXED_POINT,
> +
> + TYPE_CODE_NAMELIST, /**< Fortran namelist. */
In terms of documenting the enumerate, can you use the same style
as the other enums? This is actually important, because I believe
this is to allow us to use Doxygen to generate documentation (something
we may not have done for a while, but I'd rather we don't make it
worse).
>
> /* * Some bits for the type's instance_flags word. See the macros diff --git a/gdb/testsuite/gdb.fortran/namelist.exp b/gdb/testsuite/gdb.fortran/namelist.exp
> new file mode 100644
> index 00000000000..90762928455
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/namelist.exp
> @@ -0,0 +1,49 @@
> +# Copyright (C) 2021 Free Software Foundation, Inc.
> +
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 3 of the License, or #
> +(at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful, # but
> +WITHOUT ANY WARRANTY; without even the implied warranty of #
> +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU
> +General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License #
> +along with this program. If not, see <http://www.gnu.org/licenses/>.
> +
> +# This file is part of the gdb testsuite. It contains tests for
> +fortran # namelist.
I think there are trailing spaces at the end of several lines above.
While at it, would you mind please removing them all?
Or, actually, given the above, I'm wondering if this wasn't caused
by your mailer making some reformatting changes..
> +
> +if { [skip_fortran_tests] } { return -1 }
> +
> +standard_testfile .f90
> +load_lib "fortran.exp"
> +
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile {debug f90}]} {
> + return -1
> +}
> +
> +if ![fortran_runto_main] then {
> + perror "couldn't run to main"
> + continue
> +}
> +
> +# Depending on the compiler being used, the type names can be printed differently.
Same request as in the commit message: It would be nice if we could
keep the code and comments without 80-characters whenever reasonable.
> +set int [fortran_int4]
> +
> +gdb_breakpoint [gdb_get_line_number "Display namelist"]
> +gdb_continue_to_breakpoint "Display namelist"
> +
> +if {[test_compiler_info {gcc-*}]} {
> + gdb_test "ptype nml" \
> + "type = Type nml\r\n *$int :: a\r\n *$int :: b\r\n *End Type nml"
> + gdb_test "print nml" \
> + "\\$\[0-9\]+ = \\( a = 10, b = 20 \\)"
> +} else {
> + gdb_test "ptype nml" \
> + "No symbol \"nml\" in current context\\."
> + gdb_test "print nml" \
> + "No symbol \"nml\" in current context\\."
> +}
> diff --git a/gdb/testsuite/gdb.fortran/namelist.f90 b/gdb/testsuite/gdb.fortran/namelist.f90
> new file mode 100644
> index 00000000000..fb36690d765
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/namelist.f90
> @@ -0,0 +1,27 @@
> +! Copyright (C) 2021 Free Software Foundation, Inc.
> +!
> +! This program is free software; you can redistribute it and/or modify
> +! it under the terms of the GNU General Public License as published by
> +! the Free Software Foundation; either version 3 of the License, or !
> +(at your option) any later version.
> +!
> +! This program is distributed in the hope that it will be useful, ! but
> +WITHOUT ANY WARRANTY; without even the implied warranty of !
> +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ! GNU
> +General Public License for more details.
> +!
> +! You should have received a copy of the GNU General Public License !
> +along with this program. If not, see <http://www.gnu.org/licenses/>.
> +!
> +! This file is the Fortran source file for namelist.exp.
> +
> +program main
> +
> + integer :: a, b
> + namelist /nml/ a, b
> +
> + a = 10
> + b = 20
> + Write(*,nml) ! Display namelist
> +
> +end program main
> diff --git a/include/dwarf2.def b/include/dwarf2.def index 1ae6e1df298..6b8be1f6a16 100644
> --- a/include/dwarf2.def
> +++ b/include/dwarf2.def
> @@ -289,7 +289,7 @@ DW_AT (DW_AT_frame_base, 0x40) DW_AT (DW_AT_friend, 0x41) DW_AT (DW_AT_identifier_case, 0x42) DW_AT (DW_AT_macro_info, 0x43) -DW_AT (DW_AT_namelist_items, 0x44)
> +DW_AT (DW_AT_namelist_item, 0x44)
> DW_AT (DW_AT_priority, 0x45)
> DW_AT (DW_AT_segment, 0x46)
> DW_AT (DW_AT_specification, 0x47)
> --
> 2.17.1
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2021-11-27 7:56 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-11-11 5:32 [PING*4] [PATCH] Fix ptype and print commands for namelist variables(a fortran feature) Kumar N, Bhuvanendra
2021-11-27 7:56 ` Joel Brobecker
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).