* [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types.
2013-12-11 12:35 [PATCH v2 0/2] enable ptype/whatis for fortran types/modules Keven Boell
2013-12-11 12:35 ` [PATCH v2 2/2] fortran: enable ptype/whatis for modules Keven Boell
@ 2013-12-11 12:35 ` Keven Boell
2013-12-11 16:52 ` Doug Evans
1 sibling, 1 reply; 9+ messages in thread
From: Keven Boell @ 2013-12-11 12:35 UTC (permalink / raw)
To: gdb-patches; +Cc: keven.boell, sanimir.agovic
(gdb) ptype type
old> No symbol "type" in current context.
new> type = Type type
integer(kind=4) :: t_i
End Type type
2013-11-19 Sanimir Agovic <sanimir.agovic@intel.com>
Keven Boell <keven.boell@intel.com>
* f-exp.y (yylex): Add domain array to enable lookup
in multiple domains. Loop over lookup domains and try
to find requested symbol. Add STRUCT_DOMAIN to lookup
domains to be able to query for user defined types.
testsuite/
* gdb.fortran/type.f90: New file.
* gdb.fortran/whatis_type.f90: New file.
Signed-off-by: Keven Boell <keven.boell@intel.com>
---
gdb/f-exp.y | 33 ++++++++++++--------
gdb/testsuite/gdb.fortran/type.f90 | 28 +++++++++++++++++
gdb/testsuite/gdb.fortran/whatis_type.exp | 48 +++++++++++++++++++++++++++++
3 files changed, 97 insertions(+), 12 deletions(-)
create mode 100644 gdb/testsuite/gdb.fortran/type.f90
create mode 100644 gdb/testsuite/gdb.fortran/whatis_type.exp
diff --git a/gdb/f-exp.y b/gdb/f-exp.y
index 567cd00..a7e59df 100644
--- a/gdb/f-exp.y
+++ b/gdb/f-exp.y
@@ -1175,21 +1175,30 @@ yylex (void)
char *tmp = copy_name (yylval.sval);
struct symbol *sym;
struct field_of_this_result is_a_field_of_this;
+ enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
+ int i;
int hextype;
-
- /* Initialize this in case we *don't* use it in this call; that
- way we can refer to it unconditionally below. */
- memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
-
- sym = lookup_symbol (tmp, expression_context_block,
- VAR_DOMAIN,
- parse_language->la_language == language_cplus
- ? &is_a_field_of_this : NULL);
- if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
+
+ for (i = 0; i < ARRAY_SIZE (lookup_domains); ++i)
{
- yylval.tsym.type = SYMBOL_TYPE (sym);
- return TYPENAME;
+ /* Initialize this in case we *don't* use it in this call; that
+ way we can refer to it unconditionally below. */
+ memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
+
+ sym = lookup_symbol (tmp, expression_context_block,
+ lookup_domains[i],
+ parse_language->la_language == language_cplus
+ ? &is_a_field_of_this : NULL);
+ if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
+ {
+ yylval.tsym.type = SYMBOL_TYPE (sym);
+ return TYPENAME;
+ }
+
+ if (sym)
+ break;
}
+
yylval.tsym.type
= language_lookup_primitive_type_by_name (parse_language,
parse_gdbarch, tmp);
diff --git a/gdb/testsuite/gdb.fortran/type.f90 b/gdb/testsuite/gdb.fortran/type.f90
new file mode 100644
index 0000000..e0c699e
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/type.f90
@@ -0,0 +1,28 @@
+! Copyright 2013 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/>.
+
+program type
+ implicit none
+
+ type :: t1
+ integer :: t1_i
+ real :: t1_r
+ end type t1
+
+ type (t1) :: t1v
+
+ t1v%t1_i = 42
+ t1v%t1_r = 42.24 ! bp1
+end program type
diff --git a/gdb/testsuite/gdb.fortran/whatis_type.exp b/gdb/testsuite/gdb.fortran/whatis_type.exp
new file mode 100644
index 0000000..5a90b40
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/whatis_type.exp
@@ -0,0 +1,48 @@
+# Copyright 2013 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/>.
+
+if { [skip_fortran_tests] } { continue }
+
+standard_testfile type.f90
+
+if { [prepare_for_testing ${testfile}.exp ${testfile} \
+ ${srcfile} {debug f90}] } {
+ return -1
+}
+
+if ![runto MAIN__] {
+ perror "Couldn't run to MAIN__"
+ continue
+}
+
+gdb_breakpoint [gdb_get_line_number "bp1"]
+gdb_continue_to_breakpoint "bp1"
+
+set t1_i "integer\\\(kind=4\\\) :: t1_i"
+set t1_r "real\\\(kind=4\\\) :: t1_r"
+
+gdb_test "whatis t1" \
+ "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
+ "whatis t1"
+gdb_test "whatis t1v" \
+ "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
+ "whatis t1v"
+
+gdb_test "ptype t1" \
+ "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
+ "ptype t1"
+gdb_test "ptype t1v" \
+ "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
+ "ptype t1"
--
1.7.9.5
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH v2 0/2] enable ptype/whatis for fortran types/modules
@ 2013-12-11 12:35 Keven Boell
2013-12-11 12:35 ` [PATCH v2 2/2] fortran: enable ptype/whatis for modules Keven Boell
2013-12-11 12:35 ` [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types Keven Boell
0 siblings, 2 replies; 9+ messages in thread
From: Keven Boell @ 2013-12-11 12:35 UTC (permalink / raw)
To: gdb-patches; +Cc: keven.boell, sanimir.agovic
This patch series enables ptype/whatis for fortran
user defined types and modules.
Recognize fortran user defined types:
(gdb) ptype type
old> No symbol "type" in current context.
new> type = Type type
integer(kind=4) :: t_i
End Type type
Recognize fortran modules:
(gdb) ptype modname
old> No symbol "modname" in current context.
new> type = module modname
Keven Boell (1):
fortran: enable ptype/whatis for modules.
Sanimir Agovic (1):
fortran: enable ptype/whatis for user defined types.
gdb/cp-namespace.c | 1 +
gdb/dwarf2read.c | 20 ++++++++++++
gdb/f-exp.y | 34 ++++++++++++--------
gdb/symtab.h | 4 +++
gdb/testsuite/gdb.fortran/module.exp | 13 ++++++--
gdb/testsuite/gdb.fortran/module.f90 | 12 ++++++++
gdb/testsuite/gdb.fortran/type.f90 | 28 +++++++++++++++++
gdb/testsuite/gdb.fortran/whatis_type.exp | 48 +++++++++++++++++++++++++++++
8 files changed, 145 insertions(+), 15 deletions(-)
create mode 100644 gdb/testsuite/gdb.fortran/type.f90
create mode 100644 gdb/testsuite/gdb.fortran/whatis_type.exp
--
1.7.9.5
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH v2 2/2] fortran: enable ptype/whatis for modules.
2013-12-11 12:35 [PATCH v2 0/2] enable ptype/whatis for fortran types/modules Keven Boell
@ 2013-12-11 12:35 ` Keven Boell
2013-12-11 17:07 ` Doug Evans
2013-12-11 12:35 ` [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types Keven Boell
1 sibling, 1 reply; 9+ messages in thread
From: Keven Boell @ 2013-12-11 12:35 UTC (permalink / raw)
To: gdb-patches; +Cc: keven.boell, sanimir.agovic
Added new domain MODULE_DOMAIN for fortran modules to avoid
issues with sharing namespaces (e.g. when a variable currently
in scope has the same name as a module).
(gdb) ptype modname
old> No symbol "modname" in current context.
new> type = module modname
This fixes PR 15209 and also addresses the issue
with sharing namespaces:
https://sourceware.org/ml/gdb-patches/2013-02/msg00643.html
2013-11-19 Keven Boell <keven.boell@intel.com>
Sanimir Agovic <sanimir.agovic@intel.com>
* cp-namespace.c (cp_lookup_nested_symbol): Enable
nested lookups for fortran modules.
* dwarf2read.c (read_module): Add fortran module to
the symbol table.
(add_partial_symbol, add_partial_module): Add fortran
module to the partial symbol table.
(new_symbol_full): Create full symbol for fortran module.
* f-exp.y (yylex): Add new module domain to be parsed.
* symtab.h: New domain for fortran modules.
testsuite/
* gdb.fortran/module.exp: Completion matches fortran module
names as well. ptype/whatis on modules return a proper type.
Add new check for having the correct scope.
Signed-off-by: Keven Boell <keven.boell@intel.com>
---
gdb/cp-namespace.c | 1 +
gdb/dwarf2read.c | 20 ++++++++++++++++++++
gdb/f-exp.y | 3 ++-
gdb/symtab.h | 4 ++++
gdb/testsuite/gdb.fortran/module.exp | 13 ++++++++++---
gdb/testsuite/gdb.fortran/module.f90 | 12 ++++++++++++
6 files changed, 49 insertions(+), 4 deletions(-)
diff --git a/gdb/cp-namespace.c b/gdb/cp-namespace.c
index 36134c0..06d8c76 100644
--- a/gdb/cp-namespace.c
+++ b/gdb/cp-namespace.c
@@ -784,6 +784,7 @@ cp_lookup_nested_symbol (struct type *parent_type,
case TYPE_CODE_STRUCT:
case TYPE_CODE_NAMESPACE:
case TYPE_CODE_UNION:
+ case TYPE_CODE_MODULE:
{
/* NOTE: carlton/2003-11-10: We don't treat C++ class members
of classes like, say, data or function members. Instead,
diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
index 1c7dfc5..8d07f86 100644
--- a/gdb/dwarf2read.c
+++ b/gdb/dwarf2read.c
@@ -6801,6 +6801,13 @@ add_partial_symbol (struct partial_die_info *pdi, struct dwarf2_cu *cu)
&objfile->global_psymbols,
0, (CORE_ADDR) 0, cu->language, objfile);
break;
+ case DW_TAG_module:
+ add_psymbol_to_list (actual_name, strlen (actual_name),
+ built_actual_name != NULL,
+ MODULE_DOMAIN, LOC_TYPEDEF,
+ &objfile->global_psymbols,
+ 0, (CORE_ADDR) 0, cu->language, objfile);
+ break;
case DW_TAG_class_type:
case DW_TAG_interface_type:
case DW_TAG_structure_type:
@@ -6871,6 +6878,10 @@ static void
add_partial_module (struct partial_die_info *pdi, CORE_ADDR *lowpc,
CORE_ADDR *highpc, int need_pc, struct dwarf2_cu *cu)
{
+ /* Add a symbol for the namespace. */
+
+ add_partial_symbol (pdi, cu);
+
/* Now scan partial symbols in that module. */
if (pdi->has_children)
@@ -13660,6 +13671,10 @@ static void
read_module (struct die_info *die, struct dwarf2_cu *cu)
{
struct die_info *child_die = die->child;
+ struct type *type;
+
+ type = read_type_die (die, cu);
+ new_symbol (die, type, cu);
while (child_die && child_die->tag)
{
@@ -17690,6 +17705,11 @@ new_symbol_full (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
list_to_add = &global_symbols;
break;
+ case DW_TAG_module:
+ SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
+ SYMBOL_DOMAIN (sym) = MODULE_DOMAIN;
+ list_to_add = &global_symbols;
+ break;
case DW_TAG_common_block:
SYMBOL_ACLASS_INDEX (sym) = LOC_COMMON_BLOCK;
SYMBOL_DOMAIN (sym) = COMMON_BLOCK_DOMAIN;
diff --git a/gdb/f-exp.y b/gdb/f-exp.y
index a7e59df..efddf1f 100644
--- a/gdb/f-exp.y
+++ b/gdb/f-exp.y
@@ -1175,7 +1175,8 @@ yylex (void)
char *tmp = copy_name (yylval.sval);
struct symbol *sym;
struct field_of_this_result is_a_field_of_this;
- enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
+ enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN,
+ MODULE_DOMAIN};
int i;
int hextype;
diff --git a/gdb/symtab.h b/gdb/symtab.h
index 7cc6667..cca6a76 100644
--- a/gdb/symtab.h
+++ b/gdb/symtab.h
@@ -418,6 +418,10 @@ typedef enum domain_enum_tag
STRUCT_DOMAIN,
+ /* MODULE_DOMAIN is used in Fortran to hold module type names. */
+
+ MODULE_DOMAIN,
+
/* LABEL_DOMAIN may be used for names of labels (for gotos). */
LABEL_DOMAIN,
diff --git a/gdb/testsuite/gdb.fortran/module.exp b/gdb/testsuite/gdb.fortran/module.exp
index 6a2b87d..8f5f1b7 100644
--- a/gdb/testsuite/gdb.fortran/module.exp
+++ b/gdb/testsuite/gdb.fortran/module.exp
@@ -43,6 +43,13 @@ gdb_breakpoint [gdb_get_line_number "i-is-2"]
gdb_continue_to_breakpoint "i-is-2" ".*i-is-2.*"
gdb_test "print var_i" " = 2" "print var_i value 2"
+gdb_breakpoint [gdb_get_line_number "i-is-3"]
+gdb_continue_to_breakpoint "i-is-3" ".*i-is-3.*"
+# Ensure that the scope is correctly resolved.
+gdb_test "p mod3" "Attempt to use a type name as an expression" "print mod3"
+gdb_test "p mod2" " = 3" "print mod2"
+gdb_test "p mod1" " = 3" "print mod1"
+
gdb_breakpoint [gdb_get_line_number "a-b-c-d"]
gdb_continue_to_breakpoint "a-b-c-d" ".*a-b-c-d.*"
gdb_test "print var_a" "No symbol \"var_a\" in current context\\."
@@ -54,7 +61,7 @@ gdb_test "print var_x" " = 30" "print var_x value 30"
gdb_test "print var_y" "No symbol \"var_y\" in current context\\."
gdb_test "print var_z" " = 31" "print var_x value 31"
-gdb_test "ptype modmany" {No symbol "modmany" in current context.}
+gdb_test "ptype modmany" "type = module modmany"
proc complete {expr list} {
set cmd "complete p $expr"
@@ -62,8 +69,8 @@ proc complete {expr list} {
gdb_test $cmd $expect "complete $expr"
}
set modmany_list {modmany::var_a modmany::var_b modmany::var_c modmany::var_i}
-complete "modm" $modmany_list
-complete "modmany" $modmany_list
+complete "modm" "modmany $modmany_list"
+complete "modmany" "modmany $modmany_list"
complete "modmany::" $modmany_list
complete "modmany::var" $modmany_list
diff --git a/gdb/testsuite/gdb.fortran/module.f90 b/gdb/testsuite/gdb.fortran/module.f90
index ada7262..d10b1fb 100644
--- a/gdb/testsuite/gdb.fortran/module.f90
+++ b/gdb/testsuite/gdb.fortran/module.f90
@@ -23,6 +23,12 @@ module mod2
integer :: var_i = 2
end module mod2
+module mod3
+ integer :: mod2 = 3
+ integer :: mod1 = 3
+ integer :: var_i = 3
+end module mod3
+
module modmany
integer :: var_a = 10, var_b = 11, var_c = 12, var_i = 14
end module modmany
@@ -43,6 +49,11 @@ end module moduse
var_i = var_i ! i-is-2
end
+ subroutine sub3
+ USE mod3
+ var_i = var_i ! i-is-3
+ END
+
program module
use modmany, only: var_b, var_d => var_c, var_i
@@ -50,6 +61,7 @@ end module moduse
call sub1
call sub2
+ call sub3
if (var_b .ne. 11) call abort
if (var_d .ne. 12) call abort
--
1.7.9.5
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types.
2013-12-11 12:35 ` [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types Keven Boell
@ 2013-12-11 16:52 ` Doug Evans
2013-12-12 13:04 ` Keven Boell
0 siblings, 1 reply; 9+ messages in thread
From: Doug Evans @ 2013-12-11 16:52 UTC (permalink / raw)
To: Keven Boell; +Cc: gdb-patches, Sanimir Agovic
On Wed, Dec 11, 2013 at 4:35 AM, Keven Boell <keven.boell@intel.com> wrote:
> (gdb) ptype type
> old> No symbol "type" in current context.
> new> type = Type type
> integer(kind=4) :: t_i
> End Type type
>
> 2013-11-19 Sanimir Agovic <sanimir.agovic@intel.com>
> Keven Boell <keven.boell@intel.com>
>
> * f-exp.y (yylex): Add domain array to enable lookup
> in multiple domains. Loop over lookup domains and try
> to find requested symbol. Add STRUCT_DOMAIN to lookup
> domains to be able to query for user defined types.
>
> testsuite/
> * gdb.fortran/type.f90: New file.
> * gdb.fortran/whatis_type.f90: New file.
>
> Signed-off-by: Keven Boell <keven.boell@intel.com>
> ---
> gdb/f-exp.y | 33 ++++++++++++--------
> gdb/testsuite/gdb.fortran/type.f90 | 28 +++++++++++++++++
> gdb/testsuite/gdb.fortran/whatis_type.exp | 48 +++++++++++++++++++++++++++++
> 3 files changed, 97 insertions(+), 12 deletions(-)
> create mode 100644 gdb/testsuite/gdb.fortran/type.f90
> create mode 100644 gdb/testsuite/gdb.fortran/whatis_type.exp
>
> diff --git a/gdb/f-exp.y b/gdb/f-exp.y
> index 567cd00..a7e59df 100644
> --- a/gdb/f-exp.y
> +++ b/gdb/f-exp.y
> @@ -1175,21 +1175,30 @@ yylex (void)
> char *tmp = copy_name (yylval.sval);
> struct symbol *sym;
> struct field_of_this_result is_a_field_of_this;
> + enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
> + int i;
> int hextype;
> -
> - /* Initialize this in case we *don't* use it in this call; that
> - way we can refer to it unconditionally below. */
> - memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
> -
> - sym = lookup_symbol (tmp, expression_context_block,
> - VAR_DOMAIN,
> - parse_language->la_language == language_cplus
> - ? &is_a_field_of_this : NULL);
> - if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
> +
> + for (i = 0; i < ARRAY_SIZE (lookup_domains); ++i)
> {
> - yylval.tsym.type = SYMBOL_TYPE (sym);
> - return TYPENAME;
> + /* Initialize this in case we *don't* use it in this call; that
> + way we can refer to it unconditionally below. */
> + memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
> +
> + sym = lookup_symbol (tmp, expression_context_block,
> + lookup_domains[i],
> + parse_language->la_language == language_cplus
> + ? &is_a_field_of_this : NULL);
> + if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
> + {
> + yylval.tsym.type = SYMBOL_TYPE (sym);
> + return TYPENAME;
> + }
> +
> + if (sym)
> + break;
> }
> +
> yylval.tsym.type
> = language_lookup_primitive_type_by_name (parse_language,
> parse_gdbarch, tmp);
I don't have any comments on this part of the patch,
Looks ok but I if there's a fortran-specific issue here I wouldn't know it. :-)
> diff --git a/gdb/testsuite/gdb.fortran/type.f90 b/gdb/testsuite/gdb.fortran/type.f90
> new file mode 100644
> index 0000000..e0c699e
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/type.f90
> @@ -0,0 +1,28 @@
> +! Copyright 2013 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/>.
> +
> +program type
> + implicit none
> +
> + type :: t1
> + integer :: t1_i
> + real :: t1_r
> + end type t1
> +
> + type (t1) :: t1v
> +
> + t1v%t1_i = 42
> + t1v%t1_r = 42.24 ! bp1
> +end program type
> diff --git a/gdb/testsuite/gdb.fortran/whatis_type.exp b/gdb/testsuite/gdb.fortran/whatis_type.exp
> new file mode 100644
> index 0000000..5a90b40
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/whatis_type.exp
> @@ -0,0 +1,48 @@
> +# Copyright 2013 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/>.
> +
> +if { [skip_fortran_tests] } { continue }
> +
> +standard_testfile type.f90
> +
> +if { [prepare_for_testing ${testfile}.exp ${testfile} \
> + ${srcfile} {debug f90}] } {
> + return -1
> +}
> +
> +if ![runto MAIN__] {
> + perror "Couldn't run to MAIN__"
> + continue
> +}
I never liked calling perror for this particular failure.
grepping, I see it's still used a lot though I'll bet that's all
cut-n-paste (and presumably you're just going with the flow here as
well).
For new patches I'd like to just use fail here (that's what runto will
do internally if asked).
> +
> +gdb_breakpoint [gdb_get_line_number "bp1"]
> +gdb_continue_to_breakpoint "bp1"
> +
> +set t1_i "integer\\\(kind=4\\\) :: t1_i"
> +set t1_r "real\\\(kind=4\\\) :: t1_r"
> +
> +gdb_test "whatis t1" \
> + "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
> + "whatis t1"
> +gdb_test "whatis t1v" \
> + "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
> + "whatis t1v"
> +
> +gdb_test "ptype t1" \
> + "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
> + "ptype t1"
> +gdb_test "ptype t1v" \
> + "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
> + "ptype t1"
Typo: "ptype t1v"
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 2/2] fortran: enable ptype/whatis for modules.
2013-12-11 12:35 ` [PATCH v2 2/2] fortran: enable ptype/whatis for modules Keven Boell
@ 2013-12-11 17:07 ` Doug Evans
2013-12-12 13:04 ` Keven Boell
0 siblings, 1 reply; 9+ messages in thread
From: Doug Evans @ 2013-12-11 17:07 UTC (permalink / raw)
To: Keven Boell; +Cc: gdb-patches, Sanimir Agovic
On Wed, Dec 11, 2013 at 4:35 AM, Keven Boell <keven.boell@intel.com> wrote:
> Added new domain MODULE_DOMAIN for fortran modules to avoid
> issues with sharing namespaces (e.g. when a variable currently
> in scope has the same name as a module).
>
> (gdb) ptype modname
> old> No symbol "modname" in current context.
> new> type = module modname
>
> This fixes PR 15209 and also addresses the issue
> with sharing namespaces:
> https://sourceware.org/ml/gdb-patches/2013-02/msg00643.html
>
> 2013-11-19 Keven Boell <keven.boell@intel.com>
> Sanimir Agovic <sanimir.agovic@intel.com>
>
> * cp-namespace.c (cp_lookup_nested_symbol): Enable
> nested lookups for fortran modules.
> * dwarf2read.c (read_module): Add fortran module to
> the symbol table.
> (add_partial_symbol, add_partial_module): Add fortran
> module to the partial symbol table.
> (new_symbol_full): Create full symbol for fortran module.
> * f-exp.y (yylex): Add new module domain to be parsed.
> * symtab.h: New domain for fortran modules.
>
> testsuite/
>
> * gdb.fortran/module.exp: Completion matches fortran module
> names as well. ptype/whatis on modules return a proper type.
> Add new check for having the correct scope.
>
> Signed-off-by: Keven Boell <keven.boell@intel.com>
> ---
> gdb/cp-namespace.c | 1 +
> gdb/dwarf2read.c | 20 ++++++++++++++++++++
> gdb/f-exp.y | 3 ++-
> gdb/symtab.h | 4 ++++
> gdb/testsuite/gdb.fortran/module.exp | 13 ++++++++++---
> gdb/testsuite/gdb.fortran/module.f90 | 12 ++++++++++++
> 6 files changed, 49 insertions(+), 4 deletions(-)
>
> diff --git a/gdb/cp-namespace.c b/gdb/cp-namespace.c
> index 36134c0..06d8c76 100644
> --- a/gdb/cp-namespace.c
> +++ b/gdb/cp-namespace.c
> @@ -784,6 +784,7 @@ cp_lookup_nested_symbol (struct type *parent_type,
> case TYPE_CODE_STRUCT:
> case TYPE_CODE_NAMESPACE:
> case TYPE_CODE_UNION:
> + case TYPE_CODE_MODULE:
My gut reaction is to shudder and hope we can assert fail for this
case instead. :-)
[having to handle non-c++ stuff in cp-anything.c seems odd at first glance]
The comment for this function says it's only called when parent_type
is a c++ class or namespace.
How do we get here for a module?
[And if we can legitimately get here, a comment explaining "why things
are the way they are" is required (i.e. explain why we have to handle
TYPE_CODE_MODULE here).
I say that a lot, I know. But it really helps me when I'm in the code later.]
> {
> /* NOTE: carlton/2003-11-10: We don't treat C++ class members
> of classes like, say, data or function members. Instead,
> diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
> index 1c7dfc5..8d07f86 100644
> --- a/gdb/dwarf2read.c
> +++ b/gdb/dwarf2read.c
> @@ -6801,6 +6801,13 @@ add_partial_symbol (struct partial_die_info *pdi, struct dwarf2_cu *cu)
> &objfile->global_psymbols,
> 0, (CORE_ADDR) 0, cu->language, objfile);
> break;
> + case DW_TAG_module:
> + add_psymbol_to_list (actual_name, strlen (actual_name),
> + built_actual_name != NULL,
> + MODULE_DOMAIN, LOC_TYPEDEF,
> + &objfile->global_psymbols,
> + 0, (CORE_ADDR) 0, cu->language, objfile);
> + break;
> case DW_TAG_class_type:
> case DW_TAG_interface_type:
> case DW_TAG_structure_type:
> @@ -6871,6 +6878,10 @@ static void
> add_partial_module (struct partial_die_info *pdi, CORE_ADDR *lowpc,
> CORE_ADDR *highpc, int need_pc, struct dwarf2_cu *cu)
> {
> + /* Add a symbol for the namespace. */
> +
> + add_partial_symbol (pdi, cu);
> +
> /* Now scan partial symbols in that module. */
>
> if (pdi->has_children)
> @@ -13660,6 +13671,10 @@ static void
> read_module (struct die_info *die, struct dwarf2_cu *cu)
> {
> struct die_info *child_die = die->child;
> + struct type *type;
> +
> + type = read_type_die (die, cu);
> + new_symbol (die, type, cu);
>
> while (child_die && child_die->tag)
> {
> @@ -17690,6 +17705,11 @@ new_symbol_full (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
> SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> list_to_add = &global_symbols;
> break;
> + case DW_TAG_module:
> + SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> + SYMBOL_DOMAIN (sym) = MODULE_DOMAIN;
> + list_to_add = &global_symbols;
> + break;
> case DW_TAG_common_block:
> SYMBOL_ACLASS_INDEX (sym) = LOC_COMMON_BLOCK;
> SYMBOL_DOMAIN (sym) = COMMON_BLOCK_DOMAIN;
> diff --git a/gdb/f-exp.y b/gdb/f-exp.y
> index a7e59df..efddf1f 100644
> --- a/gdb/f-exp.y
> +++ b/gdb/f-exp.y
> @@ -1175,7 +1175,8 @@ yylex (void)
> char *tmp = copy_name (yylval.sval);
> struct symbol *sym;
> struct field_of_this_result is_a_field_of_this;
> - enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
> + enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN,
> + MODULE_DOMAIN};
I'm sure there's a rule that says this needs to be formatted
differently, though I'm not sure of the details.
Maybe begin a new line at the {, and put } on a line by itself?
> int i;
> int hextype;
>
> diff --git a/gdb/symtab.h b/gdb/symtab.h
> index 7cc6667..cca6a76 100644
> --- a/gdb/symtab.h
> +++ b/gdb/symtab.h
> @@ -418,6 +418,10 @@ typedef enum domain_enum_tag
>
> STRUCT_DOMAIN,
>
> + /* MODULE_DOMAIN is used in Fortran to hold module type names. */
> +
> + MODULE_DOMAIN,
> +
> /* LABEL_DOMAIN may be used for names of labels (for gotos). */
>
> LABEL_DOMAIN,
> diff --git a/gdb/testsuite/gdb.fortran/module.exp b/gdb/testsuite/gdb.fortran/module.exp
> index 6a2b87d..8f5f1b7 100644
> --- a/gdb/testsuite/gdb.fortran/module.exp
> +++ b/gdb/testsuite/gdb.fortran/module.exp
> @@ -43,6 +43,13 @@ gdb_breakpoint [gdb_get_line_number "i-is-2"]
> gdb_continue_to_breakpoint "i-is-2" ".*i-is-2.*"
> gdb_test "print var_i" " = 2" "print var_i value 2"
>
> +gdb_breakpoint [gdb_get_line_number "i-is-3"]
> +gdb_continue_to_breakpoint "i-is-3" ".*i-is-3.*"
> +# Ensure that the scope is correctly resolved.
> +gdb_test "p mod3" "Attempt to use a type name as an expression" "print mod3"
> +gdb_test "p mod2" " = 3" "print mod2"
> +gdb_test "p mod1" " = 3" "print mod1"
> +
> gdb_breakpoint [gdb_get_line_number "a-b-c-d"]
> gdb_continue_to_breakpoint "a-b-c-d" ".*a-b-c-d.*"
> gdb_test "print var_a" "No symbol \"var_a\" in current context\\."
> @@ -54,7 +61,7 @@ gdb_test "print var_x" " = 30" "print var_x value 30"
> gdb_test "print var_y" "No symbol \"var_y\" in current context\\."
> gdb_test "print var_z" " = 31" "print var_x value 31"
>
> -gdb_test "ptype modmany" {No symbol "modmany" in current context.}
> +gdb_test "ptype modmany" "type = module modmany"
>
> proc complete {expr list} {
> set cmd "complete p $expr"
> @@ -62,8 +69,8 @@ proc complete {expr list} {
> gdb_test $cmd $expect "complete $expr"
> }
> set modmany_list {modmany::var_a modmany::var_b modmany::var_c modmany::var_i}
> -complete "modm" $modmany_list
> -complete "modmany" $modmany_list
> +complete "modm" "modmany $modmany_list"
> +complete "modmany" "modmany $modmany_list"
> complete "modmany::" $modmany_list
> complete "modmany::var" $modmany_list
>
> diff --git a/gdb/testsuite/gdb.fortran/module.f90 b/gdb/testsuite/gdb.fortran/module.f90
> index ada7262..d10b1fb 100644
> --- a/gdb/testsuite/gdb.fortran/module.f90
> +++ b/gdb/testsuite/gdb.fortran/module.f90
> @@ -23,6 +23,12 @@ module mod2
> integer :: var_i = 2
> end module mod2
>
> +module mod3
> + integer :: mod2 = 3
> + integer :: mod1 = 3
> + integer :: var_i = 3
> +end module mod3
> +
> module modmany
> integer :: var_a = 10, var_b = 11, var_c = 12, var_i = 14
> end module modmany
> @@ -43,6 +49,11 @@ end module moduse
> var_i = var_i ! i-is-2
> end
>
> + subroutine sub3
> + USE mod3
> + var_i = var_i ! i-is-3
> + END
> +
> program module
>
> use modmany, only: var_b, var_d => var_c, var_i
> @@ -50,6 +61,7 @@ end module moduse
>
> call sub1
> call sub2
> + call sub3
>
> if (var_b .ne. 11) call abort
> if (var_d .ne. 12) call abort
> --
> 1.7.9.5
>
The rest seems ok to me.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types.
2013-12-11 16:52 ` Doug Evans
@ 2013-12-12 13:04 ` Keven Boell
2013-12-16 17:10 ` Doug Evans
0 siblings, 1 reply; 9+ messages in thread
From: Keven Boell @ 2013-12-12 13:04 UTC (permalink / raw)
To: Doug Evans, Keven Boell; +Cc: gdb-patches, Sanimir Agovic
Thank you for the review. Please see my comments below.
On 11.12.2013 17:52, Doug Evans wrote:
> On Wed, Dec 11, 2013 at 4:35 AM, Keven Boell <keven.boell@intel.com> wrote:
>> (gdb) ptype type
>> old> No symbol "type" in current context.
>> new> type = Type type
>> integer(kind=4) :: t_i
>> End Type type
>>
>> 2013-11-19 Sanimir Agovic <sanimir.agovic@intel.com>
>> Keven Boell <keven.boell@intel.com>
>>
>> * f-exp.y (yylex): Add domain array to enable lookup
>> in multiple domains. Loop over lookup domains and try
>> to find requested symbol. Add STRUCT_DOMAIN to lookup
>> domains to be able to query for user defined types.
>>
>> testsuite/
>> * gdb.fortran/type.f90: New file.
>> * gdb.fortran/whatis_type.f90: New file.
>>
>> Signed-off-by: Keven Boell <keven.boell@intel.com>
>> ---
>> gdb/f-exp.y | 33 ++++++++++++--------
>> gdb/testsuite/gdb.fortran/type.f90 | 28 +++++++++++++++++
>> gdb/testsuite/gdb.fortran/whatis_type.exp | 48 +++++++++++++++++++++++++++++
>> 3 files changed, 97 insertions(+), 12 deletions(-)
>> create mode 100644 gdb/testsuite/gdb.fortran/type.f90
>> create mode 100644 gdb/testsuite/gdb.fortran/whatis_type.exp
>>
>> diff --git a/gdb/f-exp.y b/gdb/f-exp.y
>> index 567cd00..a7e59df 100644
>> --- a/gdb/f-exp.y
>> +++ b/gdb/f-exp.y
>> @@ -1175,21 +1175,30 @@ yylex (void)
>> char *tmp = copy_name (yylval.sval);
>> struct symbol *sym;
>> struct field_of_this_result is_a_field_of_this;
>> + enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
>> + int i;
>> int hextype;
>> -
>> - /* Initialize this in case we *don't* use it in this call; that
>> - way we can refer to it unconditionally below. */
>> - memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
>> -
>> - sym = lookup_symbol (tmp, expression_context_block,
>> - VAR_DOMAIN,
>> - parse_language->la_language == language_cplus
>> - ? &is_a_field_of_this : NULL);
>> - if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
>> +
>> + for (i = 0; i < ARRAY_SIZE (lookup_domains); ++i)
>> {
>> - yylval.tsym.type = SYMBOL_TYPE (sym);
>> - return TYPENAME;
>> + /* Initialize this in case we *don't* use it in this call; that
>> + way we can refer to it unconditionally below. */
>> + memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
>> +
>> + sym = lookup_symbol (tmp, expression_context_block,
>> + lookup_domains[i],
>> + parse_language->la_language == language_cplus
>> + ? &is_a_field_of_this : NULL);
>> + if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
>> + {
>> + yylval.tsym.type = SYMBOL_TYPE (sym);
>> + return TYPENAME;
>> + }
>> +
>> + if (sym)
>> + break;
>> }
>> +
>> yylval.tsym.type
>> = language_lookup_primitive_type_by_name (parse_language,
>> parse_gdbarch, tmp);
>
> I don't have any comments on this part of the patch,
> Looks ok but I if there's a fortran-specific issue here I wouldn't know it. :-)
>
>> diff --git a/gdb/testsuite/gdb.fortran/type.f90 b/gdb/testsuite/gdb.fortran/type.f90
>> new file mode 100644
>> index 0000000..e0c699e
>> --- /dev/null
>> +++ b/gdb/testsuite/gdb.fortran/type.f90
>> @@ -0,0 +1,28 @@
>> +! Copyright 2013 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/>.
>> +
>> +program type
>> + implicit none
>> +
>> + type :: t1
>> + integer :: t1_i
>> + real :: t1_r
>> + end type t1
>> +
>> + type (t1) :: t1v
>> +
>> + t1v%t1_i = 42
>> + t1v%t1_r = 42.24 ! bp1
>> +end program type
>> diff --git a/gdb/testsuite/gdb.fortran/whatis_type.exp b/gdb/testsuite/gdb.fortran/whatis_type.exp
>> new file mode 100644
>> index 0000000..5a90b40
>> --- /dev/null
>> +++ b/gdb/testsuite/gdb.fortran/whatis_type.exp
>> @@ -0,0 +1,48 @@
>> +# Copyright 2013 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/>.
>> +
>> +if { [skip_fortran_tests] } { continue }
>> +
>> +standard_testfile type.f90
>> +
>> +if { [prepare_for_testing ${testfile}.exp ${testfile} \
>> + ${srcfile} {debug f90}] } {
>> + return -1
>> +}
>> +
>> +if ![runto MAIN__] {
>> + perror "Couldn't run to MAIN__"
>> + continue
>> +}
>
> I never liked calling perror for this particular failure.
> grepping, I see it's still used a lot though I'll bet that's all
> cut-n-paste (and presumably you're just going with the flow here as
> well).
> For new patches I'd like to just use fail here (that's what runto will
> do internally if asked).
>
Indeed, I used another test as template :-)
Changed it to call fail with an appropriate message. I've attached the updated patch below.
>> +
>> +gdb_breakpoint [gdb_get_line_number "bp1"]
>> +gdb_continue_to_breakpoint "bp1"
>> +
>> +set t1_i "integer\\\(kind=4\\\) :: t1_i"
>> +set t1_r "real\\\(kind=4\\\) :: t1_r"
>> +
>> +gdb_test "whatis t1" \
>> + "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
>> + "whatis t1"
>> +gdb_test "whatis t1v" \
>> + "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
>> + "whatis t1v"
>> +
>> +gdb_test "ptype t1" \
>> + "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
>> + "ptype t1"
>> +gdb_test "ptype t1v" \
>> + "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
>> + "ptype t1"
>
> Typo: "ptype t1v"
>
Fixed.
Updated patch:
diff --git a/gdb/f-exp.y b/gdb/f-exp.y
index 567cd00..a7e59df 100644
--- a/gdb/f-exp.y
+++ b/gdb/f-exp.y
@@ -1175,21 +1175,30 @@ yylex (void)
char *tmp = copy_name (yylval.sval);
struct symbol *sym;
struct field_of_this_result is_a_field_of_this;
+ enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
+ int i;
int hextype;
-
- /* Initialize this in case we *don't* use it in this call; that
- way we can refer to it unconditionally below. */
- memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
-
- sym = lookup_symbol (tmp, expression_context_block,
- VAR_DOMAIN,
- parse_language->la_language == language_cplus
- ? &is_a_field_of_this : NULL);
- if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
+
+ for (i = 0; i < ARRAY_SIZE (lookup_domains); ++i)
{
- yylval.tsym.type = SYMBOL_TYPE (sym);
- return TYPENAME;
+ /* Initialize this in case we *don't* use it in this call; that
+ way we can refer to it unconditionally below. */
+ memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
+
+ sym = lookup_symbol (tmp, expression_context_block,
+ lookup_domains[i],
+ parse_language->la_language == language_cplus
+ ? &is_a_field_of_this : NULL);
+ if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
+ {
+ yylval.tsym.type = SYMBOL_TYPE (sym);
+ return TYPENAME;
+ }
+
+ if (sym)
+ break;
}
+
yylval.tsym.type
= language_lookup_primitive_type_by_name (parse_language,
parse_gdbarch, tmp);
diff --git a/gdb/testsuite/gdb.fortran/type.f90 b/gdb/testsuite/gdb.fortran/type.f90
new file mode 100644
index 0000000..e0c699e
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/type.f90
@@ -0,0 +1,28 @@
+! Copyright 2013 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/>.
+
+program type
+ implicit none
+
+ type :: t1
+ integer :: t1_i
+ real :: t1_r
+ end type t1
+
+ type (t1) :: t1v
+
+ t1v%t1_i = 42
+ t1v%t1_r = 42.24 ! bp1
+end program type
diff --git a/gdb/testsuite/gdb.fortran/whatis_type.exp b/gdb/testsuite/gdb.fortran/whatis_type.exp
new file mode 100644
index 0000000..9d6151f
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/whatis_type.exp
@@ -0,0 +1,48 @@
+# Copyright 2013 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/>.
+
+if { [skip_fortran_tests] } { continue }
+
+standard_testfile type.f90
+
+if { [prepare_for_testing ${testfile}.exp ${testfile} \
+ ${srcfile} {debug f90}] } {
+ return -1
+}
+
+if ![runto MAIN__] {
+ fail "run to MAIN__"
+ return
+}
+
+gdb_breakpoint [gdb_get_line_number "bp1"]
+gdb_continue_to_breakpoint "bp1"
+
+set t1_i "integer\\\(kind=4\\\) :: t1_i"
+set t1_r "real\\\(kind=4\\\) :: t1_r"
+
+gdb_test "whatis t1" \
+ "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
+ "whatis t1"
+gdb_test "whatis t1v" \
+ "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
+ "whatis t1v"
+
+gdb_test "ptype t1" \
+ "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
+ "ptype t1"
+gdb_test "ptype t1v" \
+ "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
+ "ptype t1v"
--
1.7.9.5
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 2/2] fortran: enable ptype/whatis for modules.
2013-12-11 17:07 ` Doug Evans
@ 2013-12-12 13:04 ` Keven Boell
2013-12-16 17:14 ` Doug Evans
0 siblings, 1 reply; 9+ messages in thread
From: Keven Boell @ 2013-12-12 13:04 UTC (permalink / raw)
To: Doug Evans, Keven Boell; +Cc: gdb-patches, Sanimir Agovic
On 11.12.2013 18:07, Doug Evans wrote:
> On Wed, Dec 11, 2013 at 4:35 AM, Keven Boell <keven.boell@intel.com> wrote:
>> Added new domain MODULE_DOMAIN for fortran modules to avoid
>> issues with sharing namespaces (e.g. when a variable currently
>> in scope has the same name as a module).
>>
>> (gdb) ptype modname
>> old> No symbol "modname" in current context.
>> new> type = module modname
>>
>> This fixes PR 15209 and also addresses the issue
>> with sharing namespaces:
>> https://sourceware.org/ml/gdb-patches/2013-02/msg00643.html
>>
>> 2013-11-19 Keven Boell <keven.boell@intel.com>
>> Sanimir Agovic <sanimir.agovic@intel.com>
>>
>> * cp-namespace.c (cp_lookup_nested_symbol): Enable
>> nested lookups for fortran modules.
>> * dwarf2read.c (read_module): Add fortran module to
>> the symbol table.
>> (add_partial_symbol, add_partial_module): Add fortran
>> module to the partial symbol table.
>> (new_symbol_full): Create full symbol for fortran module.
>> * f-exp.y (yylex): Add new module domain to be parsed.
>> * symtab.h: New domain for fortran modules.
>>
>> testsuite/
>>
>> * gdb.fortran/module.exp: Completion matches fortran module
>> names as well. ptype/whatis on modules return a proper type.
>> Add new check for having the correct scope.
>>
>> Signed-off-by: Keven Boell <keven.boell@intel.com>
>> ---
>> gdb/cp-namespace.c | 1 +
>> gdb/dwarf2read.c | 20 ++++++++++++++++++++
>> gdb/f-exp.y | 3 ++-
>> gdb/symtab.h | 4 ++++
>> gdb/testsuite/gdb.fortran/module.exp | 13 ++++++++++---
>> gdb/testsuite/gdb.fortran/module.f90 | 12 ++++++++++++
>> 6 files changed, 49 insertions(+), 4 deletions(-)
>>
>> diff --git a/gdb/cp-namespace.c b/gdb/cp-namespace.c
>> index 36134c0..06d8c76 100644
>> --- a/gdb/cp-namespace.c
>> +++ b/gdb/cp-namespace.c
>> @@ -784,6 +784,7 @@ cp_lookup_nested_symbol (struct type *parent_type,
>> case TYPE_CODE_STRUCT:
>> case TYPE_CODE_NAMESPACE:
>> case TYPE_CODE_UNION:
>> + case TYPE_CODE_MODULE:
>
> My gut reaction is to shudder and hope we can assert fail for this
> case instead. :-)
> [having to handle non-c++ stuff in cp-anything.c seems odd at first glance]
> The comment for this function says it's only called when parent_type
> is a c++ class or namespace.
> How do we get here for a module?
> [And if we can legitimately get here, a comment explaining "why things
> are the way they are" is required (i.e. explain why we have to handle
> TYPE_CODE_MODULE here).
> I say that a lot, I know. But it really helps me when I'm in the code later.]
>
GDB gets there via f-exp.y with a call to lookup_symbol. Actually GDB takes
the C++ code path for looking up non-local symbols in this case by calling
the function pointer la_lookup_symbol_nonlocal, which points to the C++ specific
code (cp_lookup_symbol_nonlocal) and ends up at this location.
I see that this might be confusing and added a comment describing this behavior.
>> {
>> /* NOTE: carlton/2003-11-10: We don't treat C++ class members
>> of classes like, say, data or function members. Instead,
>> diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
>> index 1c7dfc5..8d07f86 100644
>> --- a/gdb/dwarf2read.c
>> +++ b/gdb/dwarf2read.c
>> @@ -6801,6 +6801,13 @@ add_partial_symbol (struct partial_die_info *pdi, struct dwarf2_cu *cu)
>> &objfile->global_psymbols,
>> 0, (CORE_ADDR) 0, cu->language, objfile);
>> break;
>> + case DW_TAG_module:
>> + add_psymbol_to_list (actual_name, strlen (actual_name),
>> + built_actual_name != NULL,
>> + MODULE_DOMAIN, LOC_TYPEDEF,
>> + &objfile->global_psymbols,
>> + 0, (CORE_ADDR) 0, cu->language, objfile);
>> + break;
>> case DW_TAG_class_type:
>> case DW_TAG_interface_type:
>> case DW_TAG_structure_type:
>> @@ -6871,6 +6878,10 @@ static void
>> add_partial_module (struct partial_die_info *pdi, CORE_ADDR *lowpc,
>> CORE_ADDR *highpc, int need_pc, struct dwarf2_cu *cu)
>> {
>> + /* Add a symbol for the namespace. */
>> +
>> + add_partial_symbol (pdi, cu);
>> +
>> /* Now scan partial symbols in that module. */
>>
>> if (pdi->has_children)
>> @@ -13660,6 +13671,10 @@ static void
>> read_module (struct die_info *die, struct dwarf2_cu *cu)
>> {
>> struct die_info *child_die = die->child;
>> + struct type *type;
>> +
>> + type = read_type_die (die, cu);
>> + new_symbol (die, type, cu);
>>
>> while (child_die && child_die->tag)
>> {
>> @@ -17690,6 +17705,11 @@ new_symbol_full (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
>> SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
>> list_to_add = &global_symbols;
>> break;
>> + case DW_TAG_module:
>> + SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
>> + SYMBOL_DOMAIN (sym) = MODULE_DOMAIN;
>> + list_to_add = &global_symbols;
>> + break;
>> case DW_TAG_common_block:
>> SYMBOL_ACLASS_INDEX (sym) = LOC_COMMON_BLOCK;
>> SYMBOL_DOMAIN (sym) = COMMON_BLOCK_DOMAIN;
>> diff --git a/gdb/f-exp.y b/gdb/f-exp.y
>> index a7e59df..efddf1f 100644
>> --- a/gdb/f-exp.y
>> +++ b/gdb/f-exp.y
>> @@ -1175,7 +1175,8 @@ yylex (void)
>> char *tmp = copy_name (yylval.sval);
>> struct symbol *sym;
>> struct field_of_this_result is_a_field_of_this;
>> - enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
>> + enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN,
>> + MODULE_DOMAIN};
>
> I'm sure there's a rule that says this needs to be formatted
> differently, though I'm not sure of the details.
> Maybe begin a new line at the {, and put } on a line by itself?
>
I had a look how this is done elsewhere in the code and noticed that there are several styles. I changed it to the most common one.
>> int i;
>> int hextype;
>>
>> diff --git a/gdb/symtab.h b/gdb/symtab.h
>> index 7cc6667..cca6a76 100644
>> --- a/gdb/symtab.h
>> +++ b/gdb/symtab.h
>> @@ -418,6 +418,10 @@ typedef enum domain_enum_tag
>>
>> STRUCT_DOMAIN,
>>
>> + /* MODULE_DOMAIN is used in Fortran to hold module type names. */
>> +
>> + MODULE_DOMAIN,
>> +
>> /* LABEL_DOMAIN may be used for names of labels (for gotos). */
>>
>> LABEL_DOMAIN,
>> diff --git a/gdb/testsuite/gdb.fortran/module.exp b/gdb/testsuite/gdb.fortran/module.exp
>> index 6a2b87d..8f5f1b7 100644
>> --- a/gdb/testsuite/gdb.fortran/module.exp
>> +++ b/gdb/testsuite/gdb.fortran/module.exp
>> @@ -43,6 +43,13 @@ gdb_breakpoint [gdb_get_line_number "i-is-2"]
>> gdb_continue_to_breakpoint "i-is-2" ".*i-is-2.*"
>> gdb_test "print var_i" " = 2" "print var_i value 2"
>>
>> +gdb_breakpoint [gdb_get_line_number "i-is-3"]
>> +gdb_continue_to_breakpoint "i-is-3" ".*i-is-3.*"
>> +# Ensure that the scope is correctly resolved.
>> +gdb_test "p mod3" "Attempt to use a type name as an expression" "print mod3"
>> +gdb_test "p mod2" " = 3" "print mod2"
>> +gdb_test "p mod1" " = 3" "print mod1"
>> +
>> gdb_breakpoint [gdb_get_line_number "a-b-c-d"]
>> gdb_continue_to_breakpoint "a-b-c-d" ".*a-b-c-d.*"
>> gdb_test "print var_a" "No symbol \"var_a\" in current context\\."
>> @@ -54,7 +61,7 @@ gdb_test "print var_x" " = 30" "print var_x value 30"
>> gdb_test "print var_y" "No symbol \"var_y\" in current context\\."
>> gdb_test "print var_z" " = 31" "print var_x value 31"
>>
>> -gdb_test "ptype modmany" {No symbol "modmany" in current context.}
>> +gdb_test "ptype modmany" "type = module modmany"
>>
>> proc complete {expr list} {
>> set cmd "complete p $expr"
>> @@ -62,8 +69,8 @@ proc complete {expr list} {
>> gdb_test $cmd $expect "complete $expr"
>> }
>> set modmany_list {modmany::var_a modmany::var_b modmany::var_c modmany::var_i}
>> -complete "modm" $modmany_list
>> -complete "modmany" $modmany_list
>> +complete "modm" "modmany $modmany_list"
>> +complete "modmany" "modmany $modmany_list"
>> complete "modmany::" $modmany_list
>> complete "modmany::var" $modmany_list
>>
>> diff --git a/gdb/testsuite/gdb.fortran/module.f90 b/gdb/testsuite/gdb.fortran/module.f90
>> index ada7262..d10b1fb 100644
>> --- a/gdb/testsuite/gdb.fortran/module.f90
>> +++ b/gdb/testsuite/gdb.fortran/module.f90
>> @@ -23,6 +23,12 @@ module mod2
>> integer :: var_i = 2
>> end module mod2
>>
>> +module mod3
>> + integer :: mod2 = 3
>> + integer :: mod1 = 3
>> + integer :: var_i = 3
>> +end module mod3
>> +
>> module modmany
>> integer :: var_a = 10, var_b = 11, var_c = 12, var_i = 14
>> end module modmany
>> @@ -43,6 +49,11 @@ end module moduse
>> var_i = var_i ! i-is-2
>> end
>>
>> + subroutine sub3
>> + USE mod3
>> + var_i = var_i ! i-is-3
>> + END
>> +
>> program module
>>
>> use modmany, only: var_b, var_d => var_c, var_i
>> @@ -50,6 +61,7 @@ end module moduse
>>
>> call sub1
>> call sub2
>> + call sub3
>>
>> if (var_b .ne. 11) call abort
>> if (var_d .ne. 12) call abort
>> --
>> 1.7.9.5
>>
>
>
> The rest seems ok to me.
>
The updated patch:
diff --git a/gdb/cp-namespace.c b/gdb/cp-namespace.c
index 36134c0..e666c3e 100644
--- a/gdb/cp-namespace.c
+++ b/gdb/cp-namespace.c
@@ -784,6 +784,10 @@ cp_lookup_nested_symbol (struct type *parent_type,
case TYPE_CODE_STRUCT:
case TYPE_CODE_NAMESPACE:
case TYPE_CODE_UNION:
+ /* NOTE: Handle modules here as well, because Fortran is re-using the C++
+ specific code to lookup nested symbols in modules, by calling the
+ function pointer la_lookup_symbol_nonlocal, which ends up here. */
+ case TYPE_CODE_MODULE:
{
/* NOTE: carlton/2003-11-10: We don't treat C++ class members
of classes like, say, data or function members. Instead,
diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
index 1c7dfc5..8d07f86 100644
--- a/gdb/dwarf2read.c
+++ b/gdb/dwarf2read.c
@@ -6801,6 +6801,13 @@ add_partial_symbol (struct partial_die_info *pdi, struct dwarf2_cu *cu)
&objfile->global_psymbols,
0, (CORE_ADDR) 0, cu->language, objfile);
break;
+ case DW_TAG_module:
+ add_psymbol_to_list (actual_name, strlen (actual_name),
+ built_actual_name != NULL,
+ MODULE_DOMAIN, LOC_TYPEDEF,
+ &objfile->global_psymbols,
+ 0, (CORE_ADDR) 0, cu->language, objfile);
+ break;
case DW_TAG_class_type:
case DW_TAG_interface_type:
case DW_TAG_structure_type:
@@ -6871,6 +6878,10 @@ static void
add_partial_module (struct partial_die_info *pdi, CORE_ADDR *lowpc,
CORE_ADDR *highpc, int need_pc, struct dwarf2_cu *cu)
{
+ /* Add a symbol for the namespace. */
+
+ add_partial_symbol (pdi, cu);
+
/* Now scan partial symbols in that module. */
if (pdi->has_children)
@@ -13660,6 +13671,10 @@ static void
read_module (struct die_info *die, struct dwarf2_cu *cu)
{
struct die_info *child_die = die->child;
+ struct type *type;
+
+ type = read_type_die (die, cu);
+ new_symbol (die, type, cu);
while (child_die && child_die->tag)
{
@@ -17690,6 +17705,11 @@ new_symbol_full (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
list_to_add = &global_symbols;
break;
+ case DW_TAG_module:
+ SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
+ SYMBOL_DOMAIN (sym) = MODULE_DOMAIN;
+ list_to_add = &global_symbols;
+ break;
case DW_TAG_common_block:
SYMBOL_ACLASS_INDEX (sym) = LOC_COMMON_BLOCK;
SYMBOL_DOMAIN (sym) = COMMON_BLOCK_DOMAIN;
diff --git a/gdb/f-exp.y b/gdb/f-exp.y
index a7e59df..e97a6dd 100644
--- a/gdb/f-exp.y
+++ b/gdb/f-exp.y
@@ -1175,7 +1175,12 @@ yylex (void)
char *tmp = copy_name (yylval.sval);
struct symbol *sym;
struct field_of_this_result is_a_field_of_this;
- enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
+ enum domain_enum_tag lookup_domains[] =
+ {
+ STRUCT_DOMAIN,
+ VAR_DOMAIN,
+ MODULE_DOMAIN
+ };
int i;
int hextype;
diff --git a/gdb/symtab.h b/gdb/symtab.h
index 7cc6667..cca6a76 100644
--- a/gdb/symtab.h
+++ b/gdb/symtab.h
@@ -418,6 +418,10 @@ typedef enum domain_enum_tag
STRUCT_DOMAIN,
+ /* MODULE_DOMAIN is used in Fortran to hold module type names. */
+
+ MODULE_DOMAIN,
+
/* LABEL_DOMAIN may be used for names of labels (for gotos). */
LABEL_DOMAIN,
diff --git a/gdb/testsuite/gdb.fortran/module.exp b/gdb/testsuite/gdb.fortran/module.exp
index 6a2b87d..8f5f1b7 100644
--- a/gdb/testsuite/gdb.fortran/module.exp
+++ b/gdb/testsuite/gdb.fortran/module.exp
@@ -43,6 +43,13 @@ gdb_breakpoint [gdb_get_line_number "i-is-2"]
gdb_continue_to_breakpoint "i-is-2" ".*i-is-2.*"
gdb_test "print var_i" " = 2" "print var_i value 2"
+gdb_breakpoint [gdb_get_line_number "i-is-3"]
+gdb_continue_to_breakpoint "i-is-3" ".*i-is-3.*"
+# Ensure that the scope is correctly resolved.
+gdb_test "p mod3" "Attempt to use a type name as an expression" "print mod3"
+gdb_test "p mod2" " = 3" "print mod2"
+gdb_test "p mod1" " = 3" "print mod1"
+
gdb_breakpoint [gdb_get_line_number "a-b-c-d"]
gdb_continue_to_breakpoint "a-b-c-d" ".*a-b-c-d.*"
gdb_test "print var_a" "No symbol \"var_a\" in current context\\."
@@ -54,7 +61,7 @@ gdb_test "print var_x" " = 30" "print var_x value 30"
gdb_test "print var_y" "No symbol \"var_y\" in current context\\."
gdb_test "print var_z" " = 31" "print var_x value 31"
-gdb_test "ptype modmany" {No symbol "modmany" in current context.}
+gdb_test "ptype modmany" "type = module modmany"
proc complete {expr list} {
set cmd "complete p $expr"
@@ -62,8 +69,8 @@ proc complete {expr list} {
gdb_test $cmd $expect "complete $expr"
}
set modmany_list {modmany::var_a modmany::var_b modmany::var_c modmany::var_i}
-complete "modm" $modmany_list
-complete "modmany" $modmany_list
+complete "modm" "modmany $modmany_list"
+complete "modmany" "modmany $modmany_list"
complete "modmany::" $modmany_list
complete "modmany::var" $modmany_list
diff --git a/gdb/testsuite/gdb.fortran/module.f90 b/gdb/testsuite/gdb.fortran/module.f90
index ada7262..d10b1fb 100644
--- a/gdb/testsuite/gdb.fortran/module.f90
+++ b/gdb/testsuite/gdb.fortran/module.f90
@@ -23,6 +23,12 @@ module mod2
integer :: var_i = 2
end module mod2
+module mod3
+ integer :: mod2 = 3
+ integer :: mod1 = 3
+ integer :: var_i = 3
+end module mod3
+
module modmany
integer :: var_a = 10, var_b = 11, var_c = 12, var_i = 14
end module modmany
@@ -43,6 +49,11 @@ end module moduse
var_i = var_i ! i-is-2
end
+ subroutine sub3
+ USE mod3
+ var_i = var_i ! i-is-3
+ END
+
program module
use modmany, only: var_b, var_d => var_c, var_i
@@ -50,6 +61,7 @@ end module moduse
call sub1
call sub2
+ call sub3
if (var_b .ne. 11) call abort
if (var_d .ne. 12) call abort
--
1.7.9.5
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types.
2013-12-12 13:04 ` Keven Boell
@ 2013-12-16 17:10 ` Doug Evans
0 siblings, 0 replies; 9+ messages in thread
From: Doug Evans @ 2013-12-16 17:10 UTC (permalink / raw)
To: Keven Boell; +Cc: Keven Boell, gdb-patches, Sanimir Agovic
On Thu, Dec 12, 2013 at 5:04 AM, Keven Boell
<keven.boell@linux.intel.com> wrote:
> Updated patch:
>
> diff --git a/gdb/f-exp.y b/gdb/f-exp.y
> index 567cd00..a7e59df 100644
> --- a/gdb/f-exp.y
> +++ b/gdb/f-exp.y
> @@ -1175,21 +1175,30 @@ yylex (void)
> char *tmp = copy_name (yylval.sval);
> struct symbol *sym;
> struct field_of_this_result is_a_field_of_this;
> + enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
> + int i;
> int hextype;
> -
> - /* Initialize this in case we *don't* use it in this call; that
> - way we can refer to it unconditionally below. */
> - memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
> -
> - sym = lookup_symbol (tmp, expression_context_block,
> - VAR_DOMAIN,
> - parse_language->la_language == language_cplus
> - ? &is_a_field_of_this : NULL);
> - if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
> +
> + for (i = 0; i < ARRAY_SIZE (lookup_domains); ++i)
> {
> - yylval.tsym.type = SYMBOL_TYPE (sym);
> - return TYPENAME;
> + /* Initialize this in case we *don't* use it in this call; that
> + way we can refer to it unconditionally below. */
> + memset (&is_a_field_of_this, 0, sizeof (is_a_field_of_this));
> +
> + sym = lookup_symbol (tmp, expression_context_block,
> + lookup_domains[i],
> + parse_language->la_language == language_cplus
> + ? &is_a_field_of_this : NULL);
> + if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
> + {
> + yylval.tsym.type = SYMBOL_TYPE (sym);
> + return TYPENAME;
> + }
> +
> + if (sym)
> + break;
> }
> +
> yylval.tsym.type
> = language_lookup_primitive_type_by_name (parse_language,
> parse_gdbarch, tmp);
> diff --git a/gdb/testsuite/gdb.fortran/type.f90 b/gdb/testsuite/gdb.fortran/type.f90
> new file mode 100644
> index 0000000..e0c699e
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/type.f90
> @@ -0,0 +1,28 @@
> +! Copyright 2013 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/>.
> +
> +program type
> + implicit none
> +
> + type :: t1
> + integer :: t1_i
> + real :: t1_r
> + end type t1
> +
> + type (t1) :: t1v
> +
> + t1v%t1_i = 42
> + t1v%t1_r = 42.24 ! bp1
> +end program type
> diff --git a/gdb/testsuite/gdb.fortran/whatis_type.exp b/gdb/testsuite/gdb.fortran/whatis_type.exp
> new file mode 100644
> index 0000000..9d6151f
> --- /dev/null
> +++ b/gdb/testsuite/gdb.fortran/whatis_type.exp
> @@ -0,0 +1,48 @@
> +# Copyright 2013 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/>.
> +
> +if { [skip_fortran_tests] } { continue }
> +
> +standard_testfile type.f90
> +
> +if { [prepare_for_testing ${testfile}.exp ${testfile} \
> + ${srcfile} {debug f90}] } {
> + return -1
> +}
> +
> +if ![runto MAIN__] {
> + fail "run to MAIN__"
> + return
> +}
> +
> +gdb_breakpoint [gdb_get_line_number "bp1"]
> +gdb_continue_to_breakpoint "bp1"
> +
> +set t1_i "integer\\\(kind=4\\\) :: t1_i"
> +set t1_r "real\\\(kind=4\\\) :: t1_r"
> +
> +gdb_test "whatis t1" \
> + "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
> + "whatis t1"
> +gdb_test "whatis t1v" \
> + "type = Type t1\r\n${t1_i}\r\n${t1_r}\r\nEnd Type t1" \
> + "whatis t1v"
> +
> +gdb_test "ptype t1" \
> + "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
> + "ptype t1"
> +gdb_test "ptype t1v" \
> + "type = Type t1\r\n ${t1_i}\r\n ${t1_r}\r\nEnd Type t1" \
> + "ptype t1v"
> --
Looks ok to me.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 2/2] fortran: enable ptype/whatis for modules.
2013-12-12 13:04 ` Keven Boell
@ 2013-12-16 17:14 ` Doug Evans
0 siblings, 0 replies; 9+ messages in thread
From: Doug Evans @ 2013-12-16 17:14 UTC (permalink / raw)
To: Keven Boell; +Cc: Keven Boell, gdb-patches, Sanimir Agovic
On Thu, Dec 12, 2013 at 5:04 AM, Keven Boell
<keven.boell@linux.intel.com> wrote:
> The updated patch:
>
> diff --git a/gdb/cp-namespace.c b/gdb/cp-namespace.c
> index 36134c0..e666c3e 100644
> --- a/gdb/cp-namespace.c
> +++ b/gdb/cp-namespace.c
> @@ -784,6 +784,10 @@ cp_lookup_nested_symbol (struct type *parent_type,
> case TYPE_CODE_STRUCT:
> case TYPE_CODE_NAMESPACE:
> case TYPE_CODE_UNION:
> + /* NOTE: Handle modules here as well, because Fortran is re-using the C++
> + specific code to lookup nested symbols in modules, by calling the
> + function pointer la_lookup_symbol_nonlocal, which ends up here. */
> + case TYPE_CODE_MODULE:
> {
> /* NOTE: carlton/2003-11-10: We don't treat C++ class members
> of classes like, say, data or function members. Instead,
> diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
> index 1c7dfc5..8d07f86 100644
> --- a/gdb/dwarf2read.c
> +++ b/gdb/dwarf2read.c
> @@ -6801,6 +6801,13 @@ add_partial_symbol (struct partial_die_info *pdi, struct dwarf2_cu *cu)
> &objfile->global_psymbols,
> 0, (CORE_ADDR) 0, cu->language, objfile);
> break;
> + case DW_TAG_module:
> + add_psymbol_to_list (actual_name, strlen (actual_name),
> + built_actual_name != NULL,
> + MODULE_DOMAIN, LOC_TYPEDEF,
> + &objfile->global_psymbols,
> + 0, (CORE_ADDR) 0, cu->language, objfile);
> + break;
> case DW_TAG_class_type:
> case DW_TAG_interface_type:
> case DW_TAG_structure_type:
> @@ -6871,6 +6878,10 @@ static void
> add_partial_module (struct partial_die_info *pdi, CORE_ADDR *lowpc,
> CORE_ADDR *highpc, int need_pc, struct dwarf2_cu *cu)
> {
> + /* Add a symbol for the namespace. */
> +
> + add_partial_symbol (pdi, cu);
> +
> /* Now scan partial symbols in that module. */
>
> if (pdi->has_children)
> @@ -13660,6 +13671,10 @@ static void
> read_module (struct die_info *die, struct dwarf2_cu *cu)
> {
> struct die_info *child_die = die->child;
> + struct type *type;
> +
> + type = read_type_die (die, cu);
> + new_symbol (die, type, cu);
>
> while (child_die && child_die->tag)
> {
> @@ -17690,6 +17705,11 @@ new_symbol_full (struct die_info *die, struct type *type, struct dwarf2_cu *cu,
> SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> list_to_add = &global_symbols;
> break;
> + case DW_TAG_module:
> + SYMBOL_ACLASS_INDEX (sym) = LOC_TYPEDEF;
> + SYMBOL_DOMAIN (sym) = MODULE_DOMAIN;
> + list_to_add = &global_symbols;
> + break;
> case DW_TAG_common_block:
> SYMBOL_ACLASS_INDEX (sym) = LOC_COMMON_BLOCK;
> SYMBOL_DOMAIN (sym) = COMMON_BLOCK_DOMAIN;
> diff --git a/gdb/f-exp.y b/gdb/f-exp.y
> index a7e59df..e97a6dd 100644
> --- a/gdb/f-exp.y
> +++ b/gdb/f-exp.y
> @@ -1175,7 +1175,12 @@ yylex (void)
> char *tmp = copy_name (yylval.sval);
> struct symbol *sym;
> struct field_of_this_result is_a_field_of_this;
> - enum domain_enum_tag lookup_domains[] = {STRUCT_DOMAIN, VAR_DOMAIN};
> + enum domain_enum_tag lookup_domains[] =
> + {
> + STRUCT_DOMAIN,
> + VAR_DOMAIN,
> + MODULE_DOMAIN
> + };
> int i;
> int hextype;
>
> diff --git a/gdb/symtab.h b/gdb/symtab.h
> index 7cc6667..cca6a76 100644
> --- a/gdb/symtab.h
> +++ b/gdb/symtab.h
> @@ -418,6 +418,10 @@ typedef enum domain_enum_tag
>
> STRUCT_DOMAIN,
>
> + /* MODULE_DOMAIN is used in Fortran to hold module type names. */
> +
> + MODULE_DOMAIN,
> +
> /* LABEL_DOMAIN may be used for names of labels (for gotos). */
>
> LABEL_DOMAIN,
> diff --git a/gdb/testsuite/gdb.fortran/module.exp b/gdb/testsuite/gdb.fortran/module.exp
> index 6a2b87d..8f5f1b7 100644
> --- a/gdb/testsuite/gdb.fortran/module.exp
> +++ b/gdb/testsuite/gdb.fortran/module.exp
> @@ -43,6 +43,13 @@ gdb_breakpoint [gdb_get_line_number "i-is-2"]
> gdb_continue_to_breakpoint "i-is-2" ".*i-is-2.*"
> gdb_test "print var_i" " = 2" "print var_i value 2"
>
> +gdb_breakpoint [gdb_get_line_number "i-is-3"]
> +gdb_continue_to_breakpoint "i-is-3" ".*i-is-3.*"
> +# Ensure that the scope is correctly resolved.
> +gdb_test "p mod3" "Attempt to use a type name as an expression" "print mod3"
> +gdb_test "p mod2" " = 3" "print mod2"
> +gdb_test "p mod1" " = 3" "print mod1"
> +
> gdb_breakpoint [gdb_get_line_number "a-b-c-d"]
> gdb_continue_to_breakpoint "a-b-c-d" ".*a-b-c-d.*"
> gdb_test "print var_a" "No symbol \"var_a\" in current context\\."
> @@ -54,7 +61,7 @@ gdb_test "print var_x" " = 30" "print var_x value 30"
> gdb_test "print var_y" "No symbol \"var_y\" in current context\\."
> gdb_test "print var_z" " = 31" "print var_x value 31"
>
> -gdb_test "ptype modmany" {No symbol "modmany" in current context.}
> +gdb_test "ptype modmany" "type = module modmany"
>
> proc complete {expr list} {
> set cmd "complete p $expr"
> @@ -62,8 +69,8 @@ proc complete {expr list} {
> gdb_test $cmd $expect "complete $expr"
> }
> set modmany_list {modmany::var_a modmany::var_b modmany::var_c modmany::var_i}
> -complete "modm" $modmany_list
> -complete "modmany" $modmany_list
> +complete "modm" "modmany $modmany_list"
> +complete "modmany" "modmany $modmany_list"
> complete "modmany::" $modmany_list
> complete "modmany::var" $modmany_list
>
> diff --git a/gdb/testsuite/gdb.fortran/module.f90 b/gdb/testsuite/gdb.fortran/module.f90
> index ada7262..d10b1fb 100644
> --- a/gdb/testsuite/gdb.fortran/module.f90
> +++ b/gdb/testsuite/gdb.fortran/module.f90
> @@ -23,6 +23,12 @@ module mod2
> integer :: var_i = 2
> end module mod2
>
> +module mod3
> + integer :: mod2 = 3
> + integer :: mod1 = 3
> + integer :: var_i = 3
> +end module mod3
> +
> module modmany
> integer :: var_a = 10, var_b = 11, var_c = 12, var_i = 14
> end module modmany
> @@ -43,6 +49,11 @@ end module moduse
> var_i = var_i ! i-is-2
> end
>
> + subroutine sub3
> + USE mod3
> + var_i = var_i ! i-is-3
> + END
> +
> program module
>
> use modmany, only: var_b, var_d => var_c, var_i
> @@ -50,6 +61,7 @@ end module moduse
>
> call sub1
> call sub2
> + call sub3
>
> if (var_b .ne. 11) call abort
> if (var_d .ne. 12) call abort
> --
Looks ok to me.
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2013-12-16 17:14 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-12-11 12:35 [PATCH v2 0/2] enable ptype/whatis for fortran types/modules Keven Boell
2013-12-11 12:35 ` [PATCH v2 2/2] fortran: enable ptype/whatis for modules Keven Boell
2013-12-11 17:07 ` Doug Evans
2013-12-12 13:04 ` Keven Boell
2013-12-16 17:14 ` Doug Evans
2013-12-11 12:35 ` [PATCH v2 1/2] fortran: enable ptype/whatis for user defined types Keven Boell
2013-12-11 16:52 ` Doug Evans
2013-12-12 13:04 ` Keven Boell
2013-12-16 17:10 ` Doug Evans
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).