libgomp: Add openacc_{cuda,cublas,cudart} effective targets and use them in openacc testsuite

When gcc is configured for nvptx offloading with --without-cuda-driver
and full CUDA isn't installed, many libgomp.oacc-*/* tests fail,
some of them because cuda.h header can't be found, others because
the tests can't be linked against -lcuda, -lcudart or -lcublas.
I usually only have akmod-nvidia and xorg-x11-drv-nvidia-cuda rpms
installed, so libcuda.so.1 can be dlopened and the offloading works,
but linking against those libraries isn't possible nor are the
headers around (for the plugin itself there is the fallback
libgomp/plugin/cuda/cuda.h).

The following patch adds 3 new effective targets and uses them in tests that
needs those.

2021-05-27  Jakub Jelinek  <jakub@redhat.com>

	* testsuite/lib/libgomp.exp (check_effective_target_openacc_cuda,
	check_effective_target_openacc_cublas,
	check_effective_target_openacc_cudart): New.
	* testsuite/libgomp.oacc-fortran/host_data-4.f90: Require effective
	target openacc_cublas.
	* testsuite/libgomp.oacc-fortran/host_data-2.f90: Likewise.
	* testsuite/libgomp.oacc-fortran/host_data-3.f: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-91.c: Require effective
	target openacc_cuda.
	* testsuite/libgomp.oacc-c-c++-common/lib-70.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-90.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-75.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-69.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-74.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-81.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-72.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-85.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/pr87835.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-82.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-73.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-83.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-78.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-76.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-84.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/lib-79.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/host_data-1.c: Require effective
	targets openacc_cublas and openacc_cudart.
	* testsuite/libgomp.oacc-c-c++-common/context-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/context-2.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/context-3.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/context-4.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/acc_get_property-nvptx.c:
	Require effective target openacc_cudart.
	* testsuite/libgomp.oacc-c-c++-common/asyncwait-1.c: Add -DUSE_CUDA_H
	for effective target openacc_cuda and add && defined USE_CUDA_H to
	preprocessor conditionals.  Guard -lcuda also on openacc_cuda
	effective target.

libgomp/testsuite/lib/libgomp.exp

@@ -479,3 +479,56 @@ proc check_effective_target_openacc_radeon_accel_selected { } {
     return 0;
 }
 
+# Return 1 if cuda.h and -lcuda are available.
+
+proc check_effective_target_openacc_cuda { } {
+    return [check_no_compiler_messages openacc_cuda executable {
+#include <cuda.h>
+int main() {
+    CUdevice dev;
+    CUresult r = cuDeviceGet (&dev, 0);
+    if (r != CUDA_SUCCESS)
+	return 1;
+    return 0;
+} } "-lcuda" ]
+}
+
+# Return 1 if cublas_v2.h and -lcublas are available.
+
+proc check_effective_target_openacc_cublas { } {
+    return [check_no_compiler_messages openacc_cublas executable {
+#include <cuda.h>
+#include <cublas_v2.h>
+int main() {
+    cublasStatus_t s;
+    cublasHandle_t h;
+    CUdevice dev;
+    CUresult r = cuDeviceGet (&dev, 0);
+    if (r != CUDA_SUCCESS)
+	return 1;
+    s = cublasCreate (&h);
+    if (s != CUBLAS_STATUS_SUCCESS)
+	return 1;
+    return 0;
+} } "-lcuda -lcublas" ]
+}
+
+# Return 1 if cuda_runtime_api.h and -lcudart are available.
+
+proc check_effective_target_openacc_cudart { } {
+    return [check_no_compiler_messages openacc_cudart executable {
+#include <cuda.h>
+#include <cuda_runtime_api.h>
+int main() {
+    cudaError_t e;
+    int devn;
+    CUdevice dev;
+    CUresult r = cuDeviceGet (&dev, 0);
+    if (r != CUDA_SUCCESS)
+	return 1;
+    e = cudaGetDevice (&devn);
+    if (e != cudaSuccess)
+	return 1;
+    return 0;
+} } "-lcuda -lcudart" ]
+}

libgomp/testsuite/libgomp.oacc-c-c++-common/acc_get_property-nvptx.c

@@ -4,6 +4,7 @@
 /* { dg-additional-sources acc_get_property-aux.c } */
 /* { dg-additional-options "-lcuda -lcudart" } */
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
+/* { dg-require-effective-target openacc_cudart } */
 
 #include <openacc.h>
 #include <cuda.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/asyncwait-1.c

@@ -1,9 +1,10 @@
 /* { dg-do run } */
-/* { dg-additional-options "-lcuda" { target openacc_nvidia_accel_selected } } */
+/* { dg-additional-options "-DUSE_CUDA_H" { target openacc_cuda } } */
+/* { dg-additional-options "-lcuda" { target { openacc_nvidia_accel_selected && openacc_cuda } } } */
 
 #include <openacc.h>
 #include <stdlib.h>
-#if defined ACC_DEVICE_TYPE_nvidia
+#if defined ACC_DEVICE_TYPE_nvidia && defined USE_CUDA_H
 #include "cuda.h"
 #endif
 
@@ -13,7 +14,7 @@
 int
 main (int argc, char **argv)
 {
-#if defined ACC_DEVICE_TYPE_nvidia
+#if defined ACC_DEVICE_TYPE_nvidia && defined USE_CUDA_H
     CUresult r;
     CUstream stream1;
 #endif
@@ -22,7 +23,7 @@ main (int argc, char **argv)
     int i;
     int nbytes;
 
-#if defined ACC_DEVICE_TYPE_nvidia
+#if defined ACC_DEVICE_TYPE_nvidia && defined USE_CUDA_H
     acc_init (acc_device_nvidia);
 #endif
 
@@ -216,7 +217,7 @@ main (int argc, char **argv)
     }
 
 
-#if defined ACC_DEVICE_TYPE_nvidia
+#if defined ACC_DEVICE_TYPE_nvidia && defined USE_CUDA_H
     r = cuStreamCreate (&stream1, CU_STREAM_NON_BLOCKING);
     if (r != CUDA_SUCCESS)
     {
@@ -650,7 +651,7 @@ main (int argc, char **argv)
     }
 
 
-#if defined ACC_DEVICE_TYPE_nvidia
+#if defined ACC_DEVICE_TYPE_nvidia && defined USE_CUDA_H
     r = cuStreamCreate (&stream1, CU_STREAM_NON_BLOCKING);
     if (r != CUDA_SUCCESS)
     {
@@ -902,7 +903,7 @@ main (int argc, char **argv)
             abort ();
     }
 
-#if defined ACC_DEVICE_TYPE_nvidia
+#if defined ACC_DEVICE_TYPE_nvidia && defined USE_CUDA_H
     acc_shutdown (acc_device_nvidia);
 #endif
 

libgomp/testsuite/libgomp.oacc-c-c++-common/context-1.c

@@ -1,5 +1,7 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda -lcublas -lcudart" } */
+/* { dg-require-effective-target openacc_cublas } */
+/* { dg-require-effective-target openacc_cudart } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/context-2.c

@@ -1,5 +1,7 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda -lcublas -lcudart" } */
+/* { dg-require-effective-target openacc_cublas } */
+/* { dg-require-effective-target openacc_cudart } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/context-3.c

@@ -1,5 +1,7 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda -lcublas -lcudart" } */
+/* { dg-require-effective-target openacc_cublas } */
+/* { dg-require-effective-target openacc_cudart } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/context-4.c

@@ -1,5 +1,7 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda -lcublas -lcudart" } */
+/* { dg-require-effective-target openacc_cublas } */
+/* { dg-require-effective-target openacc_cudart } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/host_data-1.c

@@ -1,5 +1,7 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lm -lcuda -lcublas -lcudart -Wall -Wextra" } */
+/* { dg-require-effective-target openacc_cublas } */
+/* { dg-require-effective-target openacc_cudart } */
 
 #include <stdlib.h>
 #include <math.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-69.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-70.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-72.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-73.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-74.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-75.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-76.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-78.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-79.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-81.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-82.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-83.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdio.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-84.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdlib.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-85.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdlib.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-90.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <pthread.h>
 #include <stdio.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/lib-91.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <stdlib.h>
 #include <unistd.h>

libgomp/testsuite/libgomp.oacc-c-c++-common/pr87835.c

@@ -1,5 +1,6 @@
 /* { dg-do run { target openacc_nvidia_accel_selected } } */
 /* { dg-additional-options "-lcuda" } */
+/* { dg-require-effective-target openacc_cuda } */
 
 #include <openacc.h>
 #include <stdlib.h>

libgomp/testsuite/libgomp.oacc-fortran/host_data-2.f90

@@ -3,6 +3,7 @@
 
 ! { dg-do run { target openacc_nvidia_accel_selected } }
 ! { dg-additional-options "-lcublas -Wall -Wextra" }
+! { dg-require-effective-target openacc_cublas }
 
 program test
   implicit none

libgomp/testsuite/libgomp.oacc-fortran/host_data-3.f

@@ -2,6 +2,7 @@
 
 ! { dg-do run { target openacc_nvidia_accel_selected } }
 ! { dg-additional-options "-lcublas -Wall -Wextra" }
+! { dg-require-effective-target openacc_cublas }
 
       include "cublas-fixed.h"
 

libgomp/testsuite/libgomp.oacc-fortran/host_data-4.f90

@@ -2,6 +2,7 @@
 
 ! { dg-do run { target openacc_nvidia_accel_selected } }
 ! { dg-additional-options "-lcublas -Wall -Wextra" }
+! { dg-require-effective-target openacc_cublas }
 
 module cublas
   interface