[Offload] Allow setting null arguments in olLaunchKernel

Author: RossBrunton

offload/liboffload/API/Kernel.td

@@ -43,19 +43,21 @@ def : Function {
     let name = "olLaunchKernel";
     let desc = "Enqueue a kernel launch with the specified size and parameters.";
     let details = [
-        "If a queue is not specified, kernel execution happens synchronously"
+        "If a queue is not specified, kernel execution happens synchronously",
+        "ArgumentsData may be set to NULL (to indicate no parameters)"
     ];
     let params = [
         Param<"ol_queue_handle_t", "Queue", "handle of the queue", PARAM_IN_OPTIONAL>,
         Param<"ol_device_handle_t", "Device", "handle of the device to execute on", PARAM_IN>,
         Param<"ol_kernel_handle_t", "Kernel", "handle of the kernel", PARAM_IN>,
-        Param<"const void*", "ArgumentsData", "pointer to the kernel argument struct", PARAM_IN>,
+        Param<"const void*", "ArgumentsData", "pointer to the kernel argument struct", PARAM_IN_OPTIONAL>,
         Param<"size_t", "ArgumentsSize", "size of the kernel argument struct", PARAM_IN>,
         Param<"const ol_kernel_launch_size_args_t*", "LaunchSizeArgs", "pointer to the struct containing launch size parameters", PARAM_IN>,
         Param<"ol_event_handle_t*", "EventOut", "optional recorded event for the enqueued operation", PARAM_OUT_OPTIONAL>
     ];
     let returns = [
         Return<"OL_ERRC_INVALID_ARGUMENT", ["`Queue == NULL && EventOut != NULL`"]>,
+        Return<"OL_ERRC_INVALID_ARGUMENT", ["`ArgumentsSize > 0 && ArgumentsData == NULL`"]>,
         Return<"OL_ERRC_INVALID_DEVICE", ["If Queue is non-null but does not belong to Device"]>,
     ];
 }

offload/liboffload/include/generated/OffloadAPI.h

@@ -692,20 +692,22 @@ typedef struct ol_kernel_launch_size_args_t {
 ///
 /// @details
 ///    - If a queue is not specified, kernel execution happens synchronously
+///    - ArgumentsData may be set to NULL (to indicate no parameters)
 ///
 /// @returns
 ///     - ::OL_RESULT_SUCCESS
 ///     - ::OL_ERRC_UNINITIALIZED
 ///     - ::OL_ERRC_DEVICE_LOST
 ///     - ::OL_ERRC_INVALID_ARGUMENT
 ///         + `Queue == NULL && EventOut != NULL`
+///     - ::OL_ERRC_INVALID_ARGUMENT
+///         + `ArgumentsSize > 0 && ArgumentsData == NULL`
 ///     - ::OL_ERRC_INVALID_DEVICE
 ///         + If Queue is non-null but does not belong to Device
 ///     - ::OL_ERRC_INVALID_NULL_HANDLE
 ///         + `NULL == Device`
 ///         + `NULL == Kernel`
 ///     - ::OL_ERRC_INVALID_NULL_POINTER
-///         + `NULL == ArgumentsData`
 ///         + `NULL == LaunchSizeArgs`
 OL_APIEXPORT ol_result_t OL_APICALL olLaunchKernel(
     // [in][optional] handle of the queue
@@ -714,7 +716,7 @@ OL_APIEXPORT ol_result_t OL_APICALL olLaunchKernel(
     ol_device_handle_t Device,
     // [in] handle of the kernel
     ol_kernel_handle_t Kernel,
-    // [in] pointer to the kernel argument struct
+    // [in][optional] pointer to the kernel argument struct
     const void *ArgumentsData,
     // [in] size of the kernel argument struct
     size_t ArgumentsSize,

offload/liboffload/include/generated/OffloadEntryPoints.inc

@@ -838,6 +838,12 @@ olLaunchKernel_val(ol_queue_handle_t Queue, ol_device_handle_t Device,
           "validation failure: Queue == NULL && EventOut != NULL");
     }
 
+    if (ArgumentsSize > 0 && ArgumentsData == NULL) {
+      return createOffloadError(
+          error::ErrorCode::INVALID_ARGUMENT,
+          "validation failure: ArgumentsSize > 0 && ArgumentsData == NULL");
+    }
+
     if (NULL == Device) {
       return createOffloadError(error::ErrorCode::INVALID_NULL_HANDLE,
                                 "validation failure: NULL == Device");
@@ -848,11 +854,6 @@ olLaunchKernel_val(ol_queue_handle_t Queue, ol_device_handle_t Device,
                                 "validation failure: NULL == Kernel");
     }
 
-    if (NULL == ArgumentsData) {
-      return createOffloadError(error::ErrorCode::INVALID_NULL_POINTER,
-                                "validation failure: NULL == ArgumentsData");
-    }
-
     if (NULL == LaunchSizeArgs) {
       return createOffloadError(error::ErrorCode::INVALID_NULL_POINTER,
                                 "validation failure: NULL == LaunchSizeArgs");

offload/unittests/OffloadAPI/device_code/CMakeLists.txt

@@ -61,6 +61,7 @@ endif()
 
 add_offload_test_device_code(foo.c foo)
 add_offload_test_device_code(bar.c bar)
+add_offload_test_device_code(noargs.c noargs)
 
 add_custom_target(OffloadUnitTestsDeviceBins DEPENDS ${BIN_PATHS})
 

offload/unittests/OffloadAPI/device_code/noargs.c

@@ -0,0 +1,3 @@
+#include <gpuintrin.h>
+
+__gpu_kernel void noargs() { (void)0; }

offload/unittests/OffloadAPI/kernel/olLaunchKernel.cpp

@@ -10,14 +10,14 @@
 #include <OffloadAPI.h>
 #include <gtest/gtest.h>
 
-struct olLaunchKernelTest : OffloadQueueTest {
-  void SetUp() override {
+struct LaunchKernelTestBase : OffloadQueueTest {
+  void SetUpKernel(const char *kernel) {
     RETURN_ON_FATAL_FAILURE(OffloadQueueTest::SetUp());
-    ASSERT_TRUE(TestEnvironment::loadDeviceBinary("foo", Device, DeviceBin));
+    ASSERT_TRUE(TestEnvironment::loadDeviceBinary(kernel, Device, DeviceBin));
     ASSERT_GE(DeviceBin->getBufferSize(), 0lu);
     ASSERT_SUCCESS(olCreateProgram(Device, DeviceBin->getBufferStart(),
                                    DeviceBin->getBufferSize(), &Program));
-    ASSERT_SUCCESS(olGetKernel(Program, "foo", &Kernel));
+    ASSERT_SUCCESS(olGetKernel(Program, kernel, &Kernel));
     LaunchArgs.Dimensions = 1;
     LaunchArgs.GroupSizeX = 64;
     LaunchArgs.GroupSizeY = 1;
@@ -43,8 +43,20 @@ struct olLaunchKernelTest : OffloadQueueTest {
   ol_kernel_launch_size_args_t LaunchArgs{};
 };
 
+struct olLaunchKernelTest : LaunchKernelTestBase {
+  void SetUp() override {
+    RETURN_ON_FATAL_FAILURE(LaunchKernelTestBase::SetUpKernel("foo"));
+  }
+};
 OFFLOAD_TESTS_INSTANTIATE_DEVICE_FIXTURE(olLaunchKernelTest);
 
+struct olLaunchKernelNoArgsTest : LaunchKernelTestBase {
+  void SetUp() override {
+    RETURN_ON_FATAL_FAILURE(LaunchKernelTestBase::SetUpKernel("noargs"));
+  }
+};
+OFFLOAD_TESTS_INSTANTIATE_DEVICE_FIXTURE(olLaunchKernelNoArgsTest);
+
 TEST_P(olLaunchKernelTest, Success) {
   void *Mem;
   ASSERT_SUCCESS(olMemAlloc(Device, OL_ALLOC_TYPE_MANAGED, 64, &Mem));
@@ -65,6 +77,13 @@ TEST_P(olLaunchKernelTest, Success) {
   ASSERT_SUCCESS(olMemFree(Mem));
 }
 
+TEST_P(olLaunchKernelNoArgsTest, Success) {
+  ASSERT_SUCCESS(
+      olLaunchKernel(Queue, Device, Kernel, nullptr, 0, &LaunchArgs, nullptr));
+
+  ASSERT_SUCCESS(olWaitQueue(Queue));
+}
+
 TEST_P(olLaunchKernelTest, SuccessSynchronous) {
   void *Mem;
   ASSERT_SUCCESS(olMemAlloc(Device, OL_ALLOC_TYPE_MANAGED, 64, &Mem));