Richard Earnshaw <rearn...@arm.com> writes:
> This test generates different warnings on ilp32 targets because the size
> of an integer matches the size of a pointer. Avoid this by using
> signed char.
>
> gcc/testsuite:
>
> PR testsuite/113428
> * gcc.dg/gomp/bad-array-section-c-3.c: Use signed char instead
> of int.
> ---
>
> I think this fixes the issues seen on ilp32 machines, without substantially
> changing what the test does, but a second set of eyes wouldn't hurt.
>
> gcc/testsuite/gcc.dg/gomp/bad-array-section-c-3.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
It looks good to me FWIW, but I'd feel more comfortable with an OK
from Jakub.
OK midday Monday UK time if no-one objects by then.
Richard
>
> diff --git a/gcc/testsuite/gcc.dg/gomp/bad-array-section-c-3.c
> b/gcc/testsuite/gcc.dg/gomp/bad-array-section-c-3.c
> index 8be15ced8c0..431af71c422 100644
> --- a/gcc/testsuite/gcc.dg/gomp/bad-array-section-c-3.c
> +++ b/gcc/testsuite/gcc.dg/gomp/bad-array-section-c-3.c
> @@ -1,15 +1,15 @@
> /* { dg-do compile } */
>
> struct S {
> - int *ptr;
> + signed char *ptr;
> };
>
> int main()
> {
> - int arr[20];
> + signed char arr[20];
>
> /* Reject array section in compound initialiser. */
> -#pragma omp target map( (struct S) { .ptr = (int *) arr[5:5] } )
> +#pragma omp target map( (struct S) { .ptr = (signed char *) arr[5:5] } )
> /* { dg-error {expected '\]' before ':' token} "" { target *-*-* } .-1 } */
> /* { dg-warning {cast to pointer from integer of different size} "" { target
> *-*-* } .-2 } */
> /* { dg-message {sorry, unimplemented: unsupported map expression} "" {
> target *-*-* } .-3 } */
