https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119117
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
Assignee|unassigned at gcc dot gnu.org |pinskia at gcc dot
gnu.org
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
>This got broken by the C2y if declarations implementation (PR c/117019, commit
>440be01).
No it was broken before that; just now exposed by it because it thinks `unknown
* b` will be a declaration but in this case unknown is the same as
`self->unknown`.
The way unknown gets resolved later on to be self->unknown is not happening.
Note it is only an issue with `*` here and that makes me think the patch for PR
20385 caused it.
I will take a look on how to fix this later this week.
But the issue is here:
```
/* Try to detect an unknown type name when we have "A B" or "A *B". */
if (c_parser_peek_token (parser)->type == CPP_NAME
&& c_parser_peek_token (parser)->id_kind == C_ID_ID
&& (c_parser_peek_2nd_token (parser)->type == CPP_NAME
|| c_parser_peek_2nd_token (parser)->type == CPP_MULT)
&& (!nested || !lookup_name (c_parser_peek_token (parser)->value)))
```
Specifically for the CPP_MULT case. I suspect there needs to a check to see if
we are inside a Objective-C method or not; similarly how nested is handled.
