On Wed, Feb 08, 2017 at 07:35:09PM -0500, John David Anglin wrote:
> The libgomp.oacc-c-c++-common/loop-dim-default.c test uses asms and needs to
> be skipped on hppa.
This broke the test everywhere, because the dg-skip-if is not in a comment.
I've fixed it (below), but am not sure why you had to change anything,
the NVPTX assembly is there on all targets.
The point is that acc_on_device (acc_device_nvidia) should fold to 0
except when it is compiled with offloading compiler, and thus
even on hpux that whole if body should be optimized away (that is why
we require -O2).
So, what is so different on hppa*-*-hpux* that this doesn't work there?
2017-02-09 Jakub Jelinek <ja...@redhat.com>
* testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c: Move
dg-skip-if directive into a comment.
--- libgomp/testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c.jj
2017-02-09 12:59:36.000000000 +0100
+++ libgomp/testsuite/libgomp.oacc-c-c++-common/loop-dim-default.c
2017-02-09 14:48:34.911816584 +0100
@@ -1,7 +1,7 @@
/* This code uses nvptx inline assembly guarded with acc_on_device, which is
- not optimized away at -O0, and then confuses the target assembler.
- { dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
- { dg-skip-if "" { hppa*-*-* } { "*" } { "" } } */
+ not optimized away at -O0, and then confuses the target assembler. */
+/* { dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
+/* { dg-skip-if "" { hppa*-*-* } { "*" } { "" } } */
/* { dg-additional-options "-fopenacc-dim=16:16" } */
#include <openacc.h>
Jakub
