[Bug fortran/45030] -fwhole-file: Bogus error message with ENTRY and different result types

[burnus at gcc dot gnu dot org]

------- Comment #2 from burnus at gcc dot gnu dot org  2010-07-23 11:24 -------
Created an attachment (id=21293)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=21293&action=view)
Draft patch

Draft patch. I was wondering whether one needs to take care of

(a) use names in case of use association - or can one not reach that function
in this case?

(b) C binding names, i.e.

subroutine foo() bind(C, name=bar)
...
entry foo2() bind(C, name=bar2)
...
end

interface
  subroutine Hello() bind(C,name=bar2)
  end
end interface

call Hello() ! << this should use the same backend decl as "ENTRY foo2".

(c) Whether this just fixes the error checking or whether the backend_decl is
also correct.

What do you think, Paul?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45030