https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125314
Steve Kargl <kargl at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |kargl at gcc dot gnu.org,
| |pault at gcc dot gnu.org
Summary|ice in |[PDT] ice in
|copy_reference_ops_from_ref |copy_reference_ops_from_ref
|, at tree-ssa-sccvn.cc:1147 |, at tree-ssa-sccvn.cc:1147
|with -O1. |with -O1.
--- Comment #4 from Steve Kargl <kargl at gcc dot gnu.org> ---
(In reply to David Binderman from comment #0)
> This Fortran 90 source code, from the Fujitsu test suite
> on github:
>
> module mod
> type ty10(ii)
> integer,kind::ii=20
> character(len=3) :: chk( ii )
> end type
This code mixes the new and exciting parameterized
derived type (PDT) feature with the very old and
musty 'entry' statement. Anyone mixing the two
deserves the pain of Dante's 6th circle.
Likely a corner case that Paul has yet to run into.
For the record, if you add -Wall to command line
with the original test case, one sees that the
'dope vector' for the returned entity is not
correctly set up.
% gfcx -c -O1 -Wall n8.f90
n8.f90:16:27:
16 | afun%str%chk(1:3)='abc'
| ^
Warning: '__result_master.0.afun.str.chk.data' is used uninitialized
[-Wuninitialized]
n8.f90:18:3:
18 | end
| ^
note: '__result_master.0.afun' declared here
n8.f90:16:27:
16 | afun%str%chk(1:3)='abc'
| ^
Warning: '__result_master.0.afun.str.chk.offset' is used uninitialized
[-Wuninitialized]
n8.f90:18:3:
18 | end
| ^
note: '__result_master.0.afun' declared here
n8.f90:16:27:
16 | afun%str%chk(1:3)='abc'
| ^
Warning: '__result_master.0.afun.str.chk.dim[0].stride' is used uninitialized
[-Wuninitialized]
n8.f90:18:3:
18 | end
| ^
note: '__result_master.0.afun' declared here
