https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123597
Tobias Burnus <burnus at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|--- |14.4
CC| |sandra at gcc dot gnu.org
--- Comment #3 from Tobias Burnus <burnus at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #1)
> No, it is not shared by all threads. The int sum = 0; declaration is still
> inside of
> #pragma omp parallel shared(A)
OK, I have simplified it too much. The larger program had:
#pragma omp target teams distribute parallel for collapse(2)
and then the gimple dump has:
int sum;
#pragma omp distribute collapse(2) private(j.31) private(i.30)
for (i.30 = 0; i.30 < D.3342; i.30 = i.30 + 1)
for (j.31 = 0; j.31 < D.3344; j.31 = j.31 + 1)
{
#pragma omp parallel shared(D.3344) shared(D.3342) shared(sum) \
shared(A) shared(N)
which has the sharing issue.
* * *
> BTW, it changed with r14-3490-g53891f18f32588d86ba0ec1c5e6206df63be714b
which is:
commit 53891f18f32588d86ba0ec1c5e6206df63be714b
Author: Sandra Loosemore
Date: Thu Aug 24 17:35:00 2023 +0000
OpenMP: C++ support for imperfectly-nested loops
