libgomp: Split testcase in order to XFAIL __sync_val_compare_and_swap_16

On PowerPC, only, __sync_val_compare_and_swap_16 is generated for the
testcase – which nvptx currently does not support.

libgomp/ChangeLog:

	* testsuite/libgomp.c-c++-common/reduction-16.c: Move int128 test to ...
	* testsuite/libgomp.c-c++-common/reduction-17.c: ... this new test.
	* testsuite/libgomp.c-c++-common/reduction-17a.c: New test.

 .../testsuite/libgomp.c-c++-common/reduction-16.c  |  8 +----
 .../testsuite/libgomp.c-c++-common/reduction-17.c  | 41 ++++++++++++++++++++++
 .../testsuite/libgomp.c-c++-common/reduction-17a.c | 11 ++++++
 3 files changed, 53 insertions(+), 7 deletions(-)

diff --git a/libgomp/testsuite/libgomp.c-c++-common/reduction-16.c b/libgomp/testsuite/libgomp.c-c++-common/reduction-16.c
index d0e82b04790..02ec4cd36b8 100644
--- a/libgomp/testsuite/libgomp.c-c++-common/reduction-16.c
+++ b/libgomp/testsuite/libgomp.c-c++-common/reduction-16.c
@@ -1,4 +1,5 @@
 /* { dg-do run } */
+/* See also reduction-17.c for an int128 testcase.  */
 
 #include <stdlib.h>
 
@@ -32,9 +33,6 @@ GENERATE_TEST(char)
 GENERATE_TEST(short)
 GENERATE_TEST(int)
 GENERATE_TEST(long)
-#ifdef __SIZEOF_INT128__
-GENERATE_TEST(__int128)
-#endif
 
 int main(void)
 {
@@ -46,8 +44,4 @@ int main(void)
     abort ();
   if (test_long ())
     abort ();
-#ifdef __SIZEOF_INT128__
-  if (test___int128 ())
-    abort ();
-#endif
 }
diff --git a/libgomp/testsuite/libgomp.c-c++-common/reduction-17.c b/libgomp/testsuite/libgomp.c-c++-common/reduction-17.c
new file mode 100644
index 00000000000..44dabf6e139
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c-c++-common/reduction-17.c
@@ -0,0 +1,41 @@
+/* { dg-do run { target { ! powerpc*-*-linux* } || { ! offload_target_nvptx } } } */
+/* { dg-require-effective-target int128 } */
+
+/* See also reduction-17a.c for powerpc*-*-linux* + offload_target_nvptx.  */
+/* See also reduction-16.c for a char/short/int/long testcase.  */
+
+#include <stdlib.h>
+
+#define N 512
+
+#define GENERATE_TEST(T)	\
+int test_##T (void)		\
+{				\
+  T a[N], res = 0;		\
+				\
+  for (int i = 0; i < N; ++i)	\
+    a[i] = i & 1;		\
+				\
+_Pragma("omp target teams distribute reduction(||:res) defaultmap(tofrom:scalar)") \
+  for (int i = 0; i < N; ++i)	\
+    res = res || a[i];		\
+				\
+  /* res should be non-zero.  */\
+  if (!res)			\
+    return 1;			\
+				\
+_Pragma("omp target teams distribute reduction(&&:res) defaultmap(tofrom:scalar)") \
+  for (int i = 0; i < N; ++i)	\
+    res = res && a[i];		\
+				\
+  /* res should be zero.  */	\
+  return res;			\
+}
+
+GENERATE_TEST(__int128)
+
+int main(void)
+{
+  if (test___int128 ())
+    abort ();
+}
diff --git a/libgomp/testsuite/libgomp.c-c++-common/reduction-17a.c b/libgomp/testsuite/libgomp.c-c++-common/reduction-17a.c
new file mode 100644
index 00000000000..1e6a7bc990f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c-c++-common/reduction-17a.c
@@ -0,0 +1,11 @@
+/* Duplicates reduction-17.c as it fails on PowerPC with nvptx offload
+   due to 'unresolved symbol __sync_val_compare_and_swap_16' on the nvptx side.  */
+
+/* { dg-do link { target { powerpc*-*-linux* } && { offload_target_nvptx } } } */
+/* { dg-require-effective-target int128 } */
+/* { dg-xfail-if "__sync_val_compare_and_swap_16" { *-*-* } } */
+
+/* See also reduction-17a.c for either not powerpc*-*-linux* or not offload_target_nvptx.  */
+/* See also reduction-16.c for a char/short/int/long testcase.  */
+
+#include "reduction-17.c"