[WIP][SYCL][CUDA] Fix LIT testing with CUDA devices after open sourcing

sycl/include/CL/sycl/detail/pi.hpp

@@ -107,7 +107,7 @@ std::string platformInfoToString(pi_platform_info info);
 template <class To, class From> To cast(From value);
 
 // Holds the PluginInformation for the plugin that is bound.
-// Currently a global varaible is used to store OpenCL plugin information to be
+// Currently a global variable is used to store OpenCL plugin information to be
 // used with SYCL Interoperability Constructors.
 extern std::shared_ptr<plugin> GlobalPlugin;
 

sycl/plugins/cuda/CMakeLists.txt

@@ -6,7 +6,8 @@ message(STATUS "Including the PI API CUDA backend.")
 
 find_package(CUDA 10.0 REQUIRED)
 
-add_library(cudadrv SHARED IMPORTED)
+# Make imported library global to use it within the project.
+add_library(cudadrv SHARED IMPORTED GLOBAL)
 
 set_target_properties(
   cudadrv PROPERTIES 

sycl/plugins/cuda/pi_cuda.cpp

@@ -831,7 +831,8 @@ pi_result cuda_piDeviceGetInfo(pi_device device, pi_device_info param_name,
                    pi_uint64{max_alloc});
   }
   case PI_DEVICE_INFO_IMAGE_SUPPORT: {
-    return getInfo(param_value_size, param_value, param_value_size_ret, false);
+    return getInfo(param_value_size, param_value, param_value_size_ret,
+                   PI_FALSE);
   }
   case PI_DEVICE_INFO_MAX_READ_IMAGE_ARGS: {
     return getInfo(param_value_size, param_value, param_value_size_ret, 0);
@@ -2783,6 +2784,11 @@ pi_result piPluginInit(pi_plugin *PluginInit) {
   // PI interface supports higher version or the same version.
   strncpy(PluginInit->PluginVersion, SupportedVersion, 4);
 
+  // Set whole function table to zero to make it easier to detect if
+  // functions are not set up below.
+  std::memset(&(PluginInit->PiFunctionTable), 0,
+              sizeof(PluginInit->PiFunctionTable));
+
 // Forward calls to OpenCL RT.
 #define _PI_CL(pi_api, cuda_api)                                         \
   (PluginInit->PiFunctionTable).pi_api = (decltype(&::pi_api))(&cuda_api);
@@ -2837,6 +2843,7 @@ pi_result piPluginInit(pi_plugin *PluginInit) {
   _PI_CL(piKernelRetain, cuda_piKernelRetain)
   _PI_CL(piKernelRelease, cuda_piKernelRelease)
   _PI_CL(piKernelSetExecInfo, cuda_piKernelSetExecInfo)
+
   // Event
   _PI_CL(piEventCreate, cuda_piEventCreate)
   _PI_CL(piEventGetInfo, cuda_piEventGetInfo)
@@ -2868,6 +2875,7 @@ pi_result piPluginInit(pi_plugin *PluginInit) {
   _PI_CL(piEnqueueMemImageFill, cuda_piEnqueueMemImageFill)
   _PI_CL(piEnqueueMemBufferMap, cuda_piEnqueueMemBufferMap)
   _PI_CL(piEnqueueMemUnmap, cuda_piEnqueueMemUnmap)
+
   _PI_CL(piextKernelSetArgMemObj, cuda_piextKernelSetArgMemObj)
 
 #undef _PI_CL

sycl/test/aot/accelerator.cpp

@@ -36,6 +36,7 @@ void simple_vadd(const std::array<T, N>& VA, const std::array<T, N>& VB,
           std::cerr << "Unknown async exception was caught." << std::endl;
         }
       }
+      throw "ERROR: Asynchronous exception(s)";
     });
 
   cl::sycl::range<1> numOfItems{N};
@@ -67,12 +68,12 @@ int main() {
   simple_vadd(D, E, F);
   for (unsigned int i = 0; i < array_size; i++) {
     if (C[i] != A[i] + B[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << C[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << C[i]
                 << "!\n";
       return 1;
     }
     if (F[i] != D[i] + E[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << F[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << F[i]
                 << "!\n";
       return 1;
     }

sycl/test/aot/cpu.cpp

@@ -36,6 +36,7 @@ void simple_vadd(const std::array<T, N>& VA, const std::array<T, N>& VB,
           std::cerr << "Unknown async exception was caught." << std::endl;
         }
       }
+      throw "ERROR: Asynchronous exception(s)";
     });
 
   cl::sycl::range<1> numOfItems{N};
@@ -67,12 +68,12 @@ int main() {
   simple_vadd(D, E, F);
   for (unsigned int i = 0; i < array_size; i++) {
     if (C[i] != A[i] + B[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << C[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << C[i]
                 << "!\n";
       return 1;
     }
     if (F[i] != D[i] + E[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << F[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << F[i]
                 << "!\n";
       return 1;
     }

sycl/test/aot/gpu.cpp

@@ -1,9 +1,11 @@
 // REQUIRES: ocloc, gpu
+// UNSUPPORTED: cuda
+// CUDA is not compatible with SPIR.
 
 // RUN: %clangxx -fsycl -fsycl-targets=spir64_gen-unknown-unknown-sycldevice -Xsycl-target-backend=spir64_gen-unknown-unknown-sycldevice "-device skl" %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
-// XFAIL: cuda
+
 //==----- gpu.cpp - AOT compilation for gen devices using GEN compiler  ------==//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
@@ -36,6 +38,7 @@ void simple_vadd(const std::array<T, N>& VA, const std::array<T, N>& VB,
           std::cerr << "Unknown async exception was caught." << std::endl;
         }
       }
+      throw "ERROR: Asynchronous exception(s)";
     });
 
   cl::sycl::range<1> numOfItems{N};
@@ -67,12 +70,12 @@ int main() {
   simple_vadd(D, E, F);
   for (unsigned int i = 0; i < array_size; i++) {
     if (C[i] != A[i] + B[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << C[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << C[i]
                 << "!\n";
       return 1;
     }
     if (F[i] != D[i] + E[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << F[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << F[i]
                 << "!\n";
       return 1;
     }

sycl/test/aot/multiple-devices.cpp

@@ -7,6 +7,8 @@
 //===------------------------------------------------------------------------===//
 
 // REQUIRES: opencl-aot, ocloc, aoc, cpu, gpu, accelerator
+// UNSUPPORTED: cuda
+// CUDA is not compatible with SPIR.
 
 // 1-command compilation case
 // Targeting CPU, GPU, FPGA
@@ -88,6 +90,7 @@ void simple_vadd(const std::array<T, N>& VA, const std::array<T, N>& VB,
           std::cerr << "Unknown async exception was caught." << std::endl;
         }
       }
+      throw "ERROR: Asynchronous exception(s)";
     });
 
   cl::sycl::range<1> numOfItems{N};
@@ -119,12 +122,12 @@ int main() {
   simple_vadd(D, E, F);
   for (unsigned int i = 0; i < array_size; i++) {
     if (C[i] != A[i] + B[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << C[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << C[i]
                 << "!\n";
       return 1;
     }
     if (F[i] != D[i] + E[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << F[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << F[i]
                 << "!\n";
       return 1;
     }

sycl/test/aot/with-llvm-bc.cpp

@@ -40,6 +40,7 @@ void simple_vadd(const std::array<T, N>& VA, const std::array<T, N>& VB,
           std::cerr << "Unknown async exception was caught." << std::endl;
         }
       }
+      throw "ERROR: Asynchronous exception(s)";
     });
 
   cl::sycl::range<1> numOfItems{N};
@@ -71,12 +72,12 @@ int main() {
   simple_vadd(D, E, F);
   for (unsigned int i = 0; i < array_size; i++) {
     if (C[i] != A[i] + B[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << C[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << C[i]
                 << "!\n";
       return 1;
     }
     if (F[i] != D[i] + E[i]) {
-      std::cout << "The results are incorrect (element " << i << " is " << F[i]
+      std::cerr << "The results are incorrect (element " << i << " is " << F[i]
                 << "!\n";
       return 1;
     }

sycl/test/basic_tests/accessor/accessor.cpp

@@ -11,8 +11,8 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+#include "../../helpers.hpp"
 #include <CL/sycl.hpp>
-#include <cassert>
 
 namespace sycl {
 using namespace cl::sycl;
@@ -77,19 +77,19 @@ int main() {
     auto acc_src = buf_src.get_access<sycl::access::mode::read>();
     auto acc_dst = buf_dst.get_access<sycl::access::mode::read_write>();
 
-    assert(!acc_src.is_placeholder());
-    assert(acc_src.get_size() == sizeof(src));
-    assert(acc_src.get_count() == 2);
-    assert(acc_src.get_range() == sycl::range<1>(2));
+    CHECK(!acc_src.is_placeholder());
+    CHECK(acc_src.get_size() == sizeof(src));
+    CHECK(acc_src.get_count() == 2);
+    CHECK(acc_src.get_range() == sycl::range<1>(2));
 
     // Make sure that operator[] is defined for both size_t and id<1>.
     // Implicit conversion from IdxSzT to size_t guarantees that no
     // implicit conversion from size_t to id<1> will happen.
-    assert(acc_src[IdxSzT(0)] + acc_src[IdxID1(1)] == 10);
+    CHECK(acc_src[IdxSzT(0)] + acc_src[IdxID1(1)] == 10);
 
     acc_dst[0] = acc_src[0] + acc_src[IdxID1(0)];
     acc_dst[id1] = acc_src[1] + acc_src[IdxSzT(1)];
-    assert(dst[0] == 6 && dst[1] == 14);
+    CHECK(dst[0] == 6 && dst[1] == 14);
   }
 
   // Three-dimensional host accessor.
@@ -101,18 +101,18 @@ int main() {
       sycl::buffer<int, 3> buf(data, sycl::range<3>(2, 3, 4));
       auto acc = buf.get_access<sycl::access::mode::read_write>();
 
-      assert(!acc.is_placeholder());
-      assert(acc.get_size() == sizeof(data));
-      assert(acc.get_count() == 24);
-      assert(acc.get_range() == sycl::range<3>(2, 3, 4));
+      CHECK(!acc.is_placeholder());
+      CHECK(acc.get_size() == sizeof(data));
+      CHECK(acc.get_count() == 24);
+      CHECK(acc.get_range() == sycl::range<3>(2, 3, 4));
 
       for (int i = 0; i < 2; ++i)
         for (int j = 0; j < 3; ++j)
           for (int k = 0; k < 4; ++k)
             acc[IdxID3(i, j, k)] += acc[sycl::id<3>(i, j, k)];
     }
     for (int i = 0; i < 24; ++i) {
-      assert(data[i] == 2 * i);
+      CHECK(data[i] == 2 * i);
     }
   }
   int data = 5;
@@ -125,16 +125,16 @@ int main() {
 
     Queue.submit([&](sycl::handler &cgh) {
       auto acc = buf.get_access<sycl::access::mode::read_write>(cgh);
-      assert(!acc.is_placeholder());
-      assert(acc.get_size() == sizeof(int));
-      assert(acc.get_count() == 1);
-      assert(acc.get_range() == sycl::range<1>(1));
+      CHECK(!acc.is_placeholder());
+      CHECK(acc.get_size() == sizeof(int));
+      CHECK(acc.get_count() == 1);
+      CHECK(acc.get_range() == sycl::range<1>(1));
       cgh.single_task<class kernel>(
           [=]() { acc[IdxSzT(0)] += acc[IdxID1(0)]; });
     });
     Queue.wait();
   }
-  assert(data == 10);
+  CHECK(data == 10);
 
   // Device accessor with 2-dimensional subscript operators.
   {
@@ -158,7 +158,7 @@ int main() {
         for (int j = 0; j < 3; j++) {
           std::cout << "array[" << i << "][" << j << "]=" << array[i][j]
                     << std::endl;
-          assert(array[i][j] == i * 3 + j);
+          CHECK(array[i][j] == i * 3 + j);
         }
       }
     }
@@ -188,7 +188,7 @@ int main() {
           for (int k = 0; k < 4; k++) {
             std::cout << "array[" << i << "][" << j << "][" << k
                       << "]=" << array[i][j][k] << std::endl;
-            assert(array[i][j][k] == k + 4 * (j + 3 * i));
+            CHECK(array[i][j][k] == k + 4 * (j + 3 * i));
           }
         }
       }
@@ -211,11 +211,11 @@ int main() {
 
       auto host_acc = buf.get_access<sycl::access::mode::read>();
       for (int i = 0; i != 3; ++i)
-        assert(host_acc[i] == 42);
+        CHECK(host_acc[i] == 42);
 
     } catch (cl::sycl::exception e) {
-      std::cout << "SYCL exception caught: " << e.what();
-      return 1;
+      std::cerr << "SYCL exception caught: " << e.what();
+      throw;
     }
   }
 
@@ -236,8 +236,8 @@ int main() {
       auto host_acc =
         buf.get_access<sycl::access::mode::discard_read_write>();
     } catch (cl::sycl::exception e) {
-      std::cout << "SYCL exception caught: " << e.what();
-      return 1;
+      std::cerr << "SYCL exception caught: " << e.what();
+      throw;
     }
   }
 
@@ -262,7 +262,7 @@ int main() {
       }
       for (int i = 0; i < 10; i++) {
         std::cout << "array[" << i << "]=" << array[i] << std::endl;
-        assert(array[i] == 333);
+        CHECK(array[i] == 333);
       }
     }
   }
@@ -296,8 +296,8 @@ int main() {
       for (int i = 0; i < 10; i++) {
         std::cout << "array1[" << i << "]=" << array1[i] << std::endl;
         std::cout << "array2[" << i << "]=" << array2[i] << std::endl;
-        assert(array1[i] == 333);
-        assert(array2[i] == 666);
+        CHECK(array1[i] == 333);
+        CHECK(array2[i] == 666);
       }
     }
   }
@@ -326,7 +326,7 @@ int main() {
       }
       for (int i = 0; i < 10; i++) {
         std::cout << "array[" << i << "]=" << array[i] << std::endl;
-        assert(array[i] == 333);
+        CHECK(array[i] == 333);
       }
     }
   }
@@ -349,11 +349,11 @@ int main() {
 
       auto host_acc = buf.get_access<sycl::access::mode::read>();
       for (int i = 0; i != 3; ++i)
-        assert(host_acc[i] == 42);
+        CHECK(host_acc[i] == 42);
 
     } catch (cl::sycl::exception e) {
-      std::cout << "SYCL exception caught: " << e.what();
-      return 1;
+      std::cerr << "SYCL exception caught: " << e.what();
+      throw;
     }
   }
 
@@ -374,10 +374,10 @@ int main() {
           });
         });
       }
-      assert(data == 399);
+      CHECK(data == 399);
     } catch (sycl::exception e) {
-      std::cout << "SYCL exception caught: " << e.what();
-      return 1;
+      std::cerr << "SYCL exception caught: " << e.what();
+      throw;
     }
   }
 
@@ -424,8 +424,8 @@ int main() {
                    sycl::access::target::host_buffer>
         acc6(buf3, sycl::range<1>(1));
 
-    assert(acc4 == 2);
-    assert(acc5[0] == 4);
-    assert(acc6[0] == 6);
+    CHECK(acc4 == 2);
+    CHECK(acc5[0] == 4);
+    CHECK(acc6[0] == 6);
   }
 }