https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81841
janus at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2017-08-27
Summary|THREADPRIVATE (OpenMP) |[5/6/7/8 Regression]
|wrongly rejected in BLOCK |THREADPRIVATE (OpenMP)
|DATA |wrongly rejected in BLOCK
| |DATA
Ever confirmed|0 |1
--- Comment #2 from janus at gcc dot gnu.org ---
(In reply to dbroemmel from comment #0)
> A 'patch' against the git/trunk version would be:
>
> index 9ac50f0..1a815a9 100644
> --- a/gcc/fortran/parse.c
> +++ b/gcc/fortran/parse.c
> @@ -3694,6 +3694,7 @@ loop:
> case ST_EQUIVALENCE:
> case ST_IMPLICIT:
> case ST_IMPLICIT_NONE:
> + case ST_OMP_THREADPRIVATE:
> case ST_PARAMETER:
> case ST_STRUCTURE_DECL:
> case ST_TYPE:
>
> I've not extensively tested the OpenMP behaviour though, only a few simple
> tests.
If you have successfully built GCC/gfortran with this modification, you should
check whether it shows any failures in the testsuite (by running "make
check-gfortran" in the 'gcc' subdirectory of your build dir).
If OpenMP actually allows this, then it is obviously a regression.
