[Bug fortran/96080] OpenACC/Fortran runtime library routines vs. Fortran 'pointer'
albert at tugraz dot at via Gcc-bugs <[email protected]>
| Newsgroups | gmane.comp.gcc.bugs |
|---|---|
| Message-ID | <[email protected]/bugzilla/> |
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96080
Christopher Albert <albert at tugraz dot at> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jvdelisle at gcc dot gnu.org
--- Comment #5 from Christopher Albert <albert at tugraz dot at> ---
Open for review, comments welcome:
https://github.com/lazy-fortran/gcc/pull/32
The PR123280 change (remove CONTIGUOUS from the array_h dummy
arguments in libgomp/openacc.f90) fixes this bug too, as suspected
over on PR123280. Confirmed with an independent runtime reproducer
using a pointer alias of a copyin'd target array: on current master it
stops with array-temporary warnings because acc_is_present and
acc_update_self receive a temporary; with the change it exits cleanly
on a real NVIDIA device, matching nvfortran.