SWDEV-279657 - Fix hipMallocManaged-N256M failure

Add concurrentManagedAccess detection in hipMallocManaged test.
Skip test when device doesn't support concurrentManagedAccess.

Change-Id: Ie54046feef3baba857a7068972ec1fc1a60c2dfd

hipamd/include/hip/nvidia_detail/hip_runtime_api.h

@@ -1561,6 +1561,9 @@ inline static hipError_t hipDeviceGetAttribute(int* pi, hipDeviceAttribute_t att
         case hipDeviceAttributeCooperativeMultiDeviceLaunch:
             cdattr = cudaDevAttrCooperativeMultiDeviceLaunch;
             break;
+        case hipDeviceAttributeConcurrentManagedAccess:
+            cdattr = cudaDevAttrConcurrentManagedAccess;
+            break;
         default:
             return hipCUDAErrorTohipError(cudaErrorInvalidValue);
     }

hipamd/tests/src/runtimeApi/memory/hipMallocManaged.cpp

@@ -20,6 +20,15 @@ void add(int n, float *x, float *y)
 int main(int argc, char *argv[])
 {
     HipTest::parseStandardArguments(argc, argv, true);
+    int concurrentManagedAccess = 0;
+    HIPCHECK(hipDeviceGetAttribute(&concurrentManagedAccess,
+                           hipDeviceAttributeConcurrentManagedAccess,
+                           p_gpuDevice));
+    if(!concurrentManagedAccess) {
+        printf("info: concurrent managed access not supported on device %d\n Skipped\n",
+               p_gpuDevice);
+        passed();
+    }
 
     printf("info: set device to %d\n", p_gpuDevice);
     HIPCHECK(hipSetDevice(p_gpuDevice));