[5/7] gdb/fortran: Don't include module symbols when searching for types

  Currently the 'info types' command will return symbols that correspond
to Fortran modules.  This because the symbols are created with domain
MODULE_DOMAIN, but address_class LOC_TYPEDEF, which is the same
address_class as type symbols.

This commit explicitly prevents MODULE_DOMAIN symbols from appearing
when we search for symbols in the TYPES_DOMAIN, this prevents the
module symbols from appearing in the output of 'info types'.

gdb/ChangeLog:

	* symtab.c (search_symbols): Don't include MODULE_DOMAIN symbols
	when searching for types.

gdb/testsuite/ChangeLog:

	* gdb.fortran/info-types.exp: Add module.
	* gdb.fortran/info-types.f90: Update expected results.
---
 gdb/ChangeLog                            |  5 +++++
 gdb/symtab.c                             |  3 ++-
 gdb/testsuite/ChangeLog                  |  5 +++++
 gdb/testsuite/gdb.fortran/info-types.exp |  4 +++-
 gdb/testsuite/gdb.fortran/info-types.f90 | 11 +++++++++++
 5 files changed, 26 insertions(+), 2 deletions(-)
  

* Andrew Burgess <andrew.burgess@embecosm.com> [2019-07-27 17:22:33 +0100]:

> Currently the 'info types' command will return symbols that correspond
> to Fortran modules.  This because the symbols are created with domain
> MODULE_DOMAIN, but address_class LOC_TYPEDEF, which is the same
> address_class as type symbols.
> 
> This commit explicitly prevents MODULE_DOMAIN symbols from appearing
> when we search for symbols in the TYPES_DOMAIN, this prevents the
> module symbols from appearing in the output of 'info types'.
> 
> gdb/ChangeLog:
> 
> 	* symtab.c (search_symbols): Don't include MODULE_DOMAIN symbols
> 	when searching for types.
> 
> gdb/testsuite/ChangeLog:
> 
> 	* gdb.fortran/info-types.exp: Add module.
> 	* gdb.fortran/info-types.f90: Update expected results.

I've now pushed this patch as it only depended on the previous patch
in this series.

Thanks,
Andrew


> ---
>  gdb/ChangeLog                            |  5 +++++
>  gdb/symtab.c                             |  3 ++-
>  gdb/testsuite/ChangeLog                  |  5 +++++
>  gdb/testsuite/gdb.fortran/info-types.exp |  4 +++-
>  gdb/testsuite/gdb.fortran/info-types.f90 | 11 +++++++++++
>  5 files changed, 26 insertions(+), 2 deletions(-)
> 
> diff --git a/gdb/symtab.c b/gdb/symtab.c
> index 84038d15dff..d3a209608b7 100644
> --- a/gdb/symtab.c
> +++ b/gdb/symtab.c
> @@ -4539,7 +4539,8 @@ search_symbols (const char *regexp, enum search_domain kind,
>  				      || treg_matches_sym_type_name (*treg,
>  								     sym)))
>  			      || (kind == TYPES_DOMAIN
> -				  && SYMBOL_CLASS (sym) == LOC_TYPEDEF))))
> +				  && SYMBOL_CLASS (sym) == LOC_TYPEDEF
> +				  && SYMBOL_DOMAIN (sym) != MODULE_DOMAIN))))
>  		    {
>  		      /* match */
>  		      result.emplace_back (i, sym);
> diff --git a/gdb/testsuite/gdb.fortran/info-types.exp b/gdb/testsuite/gdb.fortran/info-types.exp
> index 9571dc45593..81e67395e8a 100644
> --- a/gdb/testsuite/gdb.fortran/info-types.exp
> +++ b/gdb/testsuite/gdb.fortran/info-types.exp
> @@ -42,4 +42,6 @@ gdb_test "info types" \
>  	 "\[\t \]+${character1}" \
>  	 "\[\t \]+${integer4}" \
>  	 "\[\t \]+${logical4}" \
> -	 "16:\[\t \]+Type s1;" ]
> +	 "20:\[\t \]+Type __vtype_mod1_M1t1;" \
> +	 "17:\[\t \]+Type m1t1;" \
> +	 "22:\[\t \]+Type s1;" ]
> diff --git a/gdb/testsuite/gdb.fortran/info-types.f90 b/gdb/testsuite/gdb.fortran/info-types.f90
> index 21c9d9df63c..0e27e1ddf08 100644
> --- a/gdb/testsuite/gdb.fortran/info-types.f90
> +++ b/gdb/testsuite/gdb.fortran/info-types.f90
> @@ -13,13 +13,24 @@
>  ! You should have received a copy of the GNU General Public License
>  ! along with this program.  If not, see <http://www.gnu.org/licenses/>.
>  
> +module mod1
> +  type :: m1t1
> +     integer :: b
> +  end type m1t1
> +end module mod1
> +
>  program info_types_test
> +  use mod1
> +
>    type :: s1
>       integer :: a
>    end type s1
>  
>    logical :: l
>    type (s1) :: var_a
> +  type (m1t1) :: var_b
> +
>    var_a%a = 1
> +  var_b%b = 2
>    l = .FALSE.
>  end program info_types_test
> -- 
> 2.14.5
>
  

On 28-08-19 14:37, Andrew Burgess wrote:
> * Andrew Burgess <andrew.burgess@embecosm.com> [2019-07-27 17:22:33 +0100]:
> 
>> Currently the 'info types' command will return symbols that correspond
>> to Fortran modules.  This because the symbols are created with domain
>> MODULE_DOMAIN, but address_class LOC_TYPEDEF, which is the same
>> address_class as type symbols.
>>
>> This commit explicitly prevents MODULE_DOMAIN symbols from appearing
>> when we search for symbols in the TYPES_DOMAIN, this prevents the
>> module symbols from appearing in the output of 'info types'.
>>
>> gdb/ChangeLog:
>>
>> 	* symtab.c (search_symbols): Don't include MODULE_DOMAIN symbols
>> 	when searching for types.
>>
>> gdb/testsuite/ChangeLog:
>>
>> 	* gdb.fortran/info-types.exp: Add module.
>> 	* gdb.fortran/info-types.f90: Update expected results.
> 
> I've now pushed this patch as it only depended on the previous patch
> in this series.

Also, this test-case times out with check-read1, due to the many
"Non-debugging symbols" that are parsed by a single .*.

Thanks,
- Tom
  

On 29-08-19 11:09, Tom de Vries wrote:
> On 28-08-19 14:37, Andrew Burgess wrote:
>> * Andrew Burgess <andrew.burgess@embecosm.com> [2019-07-27 17:22:33 +0100]:
>>
>>> Currently the 'info types' command will return symbols that correspond
>>> to Fortran modules.  This because the symbols are created with domain
>>> MODULE_DOMAIN, but address_class LOC_TYPEDEF, which is the same
>>> address_class as type symbols.
>>>
>>> This commit explicitly prevents MODULE_DOMAIN symbols from appearing
>>> when we search for symbols in the TYPES_DOMAIN, this prevents the
>>> module symbols from appearing in the output of 'info types'.
>>>
>>> gdb/ChangeLog:
>>>
>>> 	* symtab.c (search_symbols): Don't include MODULE_DOMAIN symbols
>>> 	when searching for types.
>>>
>>> gdb/testsuite/ChangeLog:
>>>
>>> 	* gdb.fortran/info-types.exp: Add module.
>>> 	* gdb.fortran/info-types.f90: Update expected results.
>>
>> I've now pushed this patch as it only depended on the previous patch
>> in this series.
> 
> Also, this test-case times out with check-read1, due to the many
> "Non-debugging symbols" that are parsed by a single .*.

Sorry, wrong thread, this is about gdb.base/info-var.exp:
...
FAIL: gdb.base/info-var.exp: info variables (timeout)
FAIL: gdb.base/info-var.exp: info variables -n (timeout)
FAIL: gdb.base/info-var.exp: info variables -n global (timeout)
...

Thanks,
- Tom