[SYCL] Test for in-order queue event release verification.

Signed-off-by: rbegam <[email protected]>

Author: rbegam

SYCL/InorderQueue/in_order_event_release.cpp

@@ -0,0 +1,62 @@
+// UNSUPPORTED: cuda
+// CUDA does not support unnamed lambdas.
+//
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -fsycl-unnamed-lambda %s -o %t.out
+// RUN: %HOST_RUN_PLACEHOLDER %t.out
+// RUN: %ACC_RUN_PLACEHOLDER %t.out
+// RUN: %CPU_RUN_PLACEHOLDER %t.out
+// RUN: %GPU_RUN_PLACEHOLDER %t.out
+
+//==------ oq_event_release.cpp - SYCL ordered queue event release shortcut test --------==//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include <CL/sycl.hpp>
+#include <iostream>
+using namespace cl::sycl;
+
+int main() {
+  queue q{property::queue::in_order()};
+  auto dev = q.get_device();
+  auto ctx = q.get_context();
+  const int N = 8;
+
+  if (dev.get_info<info::device::usm_shared_allocations>()) {
+    auto A = (int *)malloc_shared(N * sizeof(int), dev, ctx);
+
+    for (int i = 0; i < N; i++) {
+      A[i] = 1;
+    }
+
+    {
+      id<1> offset(0);
+      q.parallel_for<class Baz>(range<1>{N}, offset, [=](id<1> ID) {
+        auto i = ID[0];
+        A[i]++;
+      });
+
+      q.wait();
+    }
+
+    {
+      nd_range<1> NDR(range<1>{N}, range<1>{2});
+      q.parallel_for<class NDFoo>(NDR, [=](nd_item<1> Item) {
+        auto i = Item.get_global_id(0);
+        A[i]++;
+      });
+
+      q.wait();
+    }
+
+    for (int i = 0; i < N; i++) {
+      if (A[i] != 3)
+        return 1;
+    }
+  }
+
+  return 0;
+}