Hi Julian,
please hold back the commit.
Actually, it does not seem to work if one modifies the test
case a bit. The following code compiles – but I think it
should not:
type one
integer i, j
end type
type two
type(one) A, B
end type
type(two) x
!$acc enter data copyin(x%A, x%A%i)
!$acc enter data copyin(x%A, x%A%i, x%A%i)
end
At least the last line has x%A%i twice but it is accepted.
I am not sure whether the line before is valid or not
(x%A%i is copied twice; might make a difference if it were
a pointer but it isn't a pointer here.) – In C/C++, with OpenMP
and, hence, OpenACC, the first one is accepted but the second
one is rejected. C example:
struct one {
int i, j;
};
struct two {
struct one A, B;
};
void foo() {
struct two x;
/* Accepted: x.A + x.A.i, even though both map 'x.A.i'. */
#pragma omp target enter data map(to:x.A, x.A.i)
#pragma acc enter data copyin(x.A, x.A.i)
/* Rejected: 'x.A.i' mapped twice */
#pragma omp target enter data map(to:x.A, x.A.i, x.A.i)
#pragma acc enter data copyin(x.A, x.A.i, x.A.i)
}
Tobias
