This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH 5/7] gdb/fortran: Don't include module symbols when searching for types
- From: Tom de Vries <tdevries at suse dot de>
- To: Andrew Burgess <andrew dot burgess at embecosm dot com>, gdb-patches at sourceware dot org
- Cc: Richard Bunt <Richard dot Bunt at arm dot com>
- Date: Thu, 29 Aug 2019 11:09:00 +0200
- Subject: Re: [PATCH 5/7] gdb/fortran: Don't include module symbols when searching for types
- References: <cover.1564243858.git.andrew.burgess@embecosm.com> <5184ae288c08432d0e873fd48b8c56619e3d2699.1564243858.git.andrew.burgess@embecosm.com> <20190828123722.GW6076@embecosm.com>
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