Merge branch 'intel' into math_marray_tests

Author: JackAKirk

.github/CODEOWNERS

@@ -1,9 +1,8 @@
-* @vladimirlaz @romanovvlad @bader
+* @pvchupin
 
 # Use runtime team as the umbrella for most of the tests
 /SYCL/ @intel/llvm-reviewers-runtime
 
-
 # SYCL sub-directory matchers are grouped by code owner first, followed by
 # alphabetical order within the group. Please, keep this ordering.
 

.github/workflows/clang-format.yml

@@ -20,6 +20,7 @@ jobs:
     - name: Run clang-format for the patch
       shell: bash {0}
       run: |
+        git config --global --add safe.directory /__w/llvm-test-suite/llvm-test-suite
         git clang-format ${GITHUB_SHA}^1
         git diff > ./clang-format.patch
 

SYCL/Assert/assert_in_kernels_ndebug.cpp

@@ -1,5 +1,5 @@
-// FIXME unsupported on CUDA and HIP until fallback libdevice becomes available
-// UNSUPPORTED: cuda || hip
+// FIXME unsupported on HIP until fallback libdevice becomes available
+// UNSUPPORTED: hip
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -DNDEBUG %S/assert_in_kernels.cpp -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out %CPU_CHECK_PLACEHOLDER
 // RUN: %GPU_RUN_PLACEHOLDER %t.out %GPU_CHECK_PLACEHOLDER

SYCL/Assert/assert_in_simultaneous_kernels.cpp

@@ -16,11 +16,11 @@
 // RUN: %ACC_RUN_PLACEHOLDER %t.out &> %t.txt
 // RUN: %ACC_RUN_PLACEHOLDER FileCheck %s --check-prefix=CHECK-ACC --input-file %t.txt
 //
-// CHECK:      {{.*}}assert_in_simultaneous_kernels.hpp:12: void assertFunc(): global id: [9,7,0], local id: [0,0,0]
+// CHECK:      {{.*}}assert_in_simultaneous_kernels.hpp:13: void assertFunc(): global id: [9,7,0], local id: [0,0,0]
 // CHECK-SAME: Assertion `false && "from assert statement"` failed.
 // CHECK-NOT:  The test ended.
 //
-// CHECK-ACC-NOT: {{.*}}assert_in_simultaneous_kernels.hpp:12: void assertFunc(): global id: [9,7,0], local id: [0,0,0]
+// CHECK-ACC-NOT: {{.*}}assert_in_simultaneous_kernels.hpp:13: void assertFunc(): global id: [9,7,0], local id: [0,0,0]
 // CHECK-ACC: The test ended.
 
 #include "assert_in_simultaneous_kernels.hpp"

SYCL/Assert/assert_in_simultaneous_kernels.hpp

@@ -1,5 +1,6 @@
 #include <CL/sycl.hpp>
 #include <cassert>
+#include <cstdio>
 #include <iostream>
 #include <thread>
 
@@ -44,6 +45,15 @@ void runTestForTid(queue *Q, size_t Tid) {
 }
 
 int main(int Argc, const char *Argv[]) {
+  // On windows stderr output becomes messed up if several thread
+  // output simultaneously. Hence, setting explicit line buffering here.
+#ifndef __SYCL_DEVICE_ONLY__
+  if (setvbuf(stderr, nullptr, _IOLBF, BUFSIZ)) {
+    std::cerr << "Can't set line-buffering mode fo stderr\n";
+    return 1;
+  }
+#endif
+
   std::vector<std::thread> threadPool;
   threadPool.reserve(NUM_THREADS);
 

SYCL/Assert/assert_in_simultaneous_kernels_win.cpp

@@ -17,11 +17,11 @@
 //
 // FIXME Windows version prints '(null)' instead of '<unknown func>' once in a
 // while for some insane reason.
-// CHECK:      {{.*}}assert_in_simultaneous_kernels.hpp:12: {{<unknown func>|(null)}}: global id: [9,7,0], local id: [0,0,0]
+// CHECK:      {{.*}}assert_in_simultaneous_kernels.hpp:13: {{<unknown func>|(null)}}: global id: [9,7,0], local id: [0,0,0]
 // CHECK-SAME: Assertion `false && "from assert statement"` failed.
 // CHECK-NOT:  The test ended.
 //
-// CHECK-ACC-NOT: {{.*}}assert_in_simultaneous_kernels.hpp:12: {{<unknown func>|(null)}}: global id: [9,7,0], local id: [0,0,0]
+// CHECK-ACC-NOT: {{.*}}assert_in_simultaneous_kernels.hpp:13: {{<unknown func>|(null)}}: global id: [9,7,0], local id: [0,0,0]
 // CHECK-ACC:  The test ended.
 
 #include "assert_in_simultaneous_kernels.hpp"

SYCL/Assert/assert_in_simultaneously_multiple_tus.cpp

@@ -1,9 +1,6 @@
 // FIXME unsupported on CUDA and HIP until fallback libdevice becomes available
-// UNSUPPORTED: cuda || hip
-// clang-format off
-// Failed on Linux on unrelated change (FileCheck error: '.../assert_in_simultaneously_multiple_tus.cpp.tmp.txt' is empty)
-// clang-format on
-// REQUIRES: TEMPORARILY_DISABLED
+// FIXME flaky output on Level Zero
+// UNSUPPORTED: cuda || hip || level_zero
 // RUN: %clangxx -DSYCL_FALLBACK_ASSERT=1 -fsycl -fsycl-targets=%sycl_triple -I %S/Inputs %s %S/Inputs/kernels_in_file2.cpp -o %t.out %threads_lib
 // RUN: %CPU_RUN_PLACEHOLDER %t.out &> %t.txt || true
 // RUN: %CPU_RUN_PLACEHOLDER FileCheck %s --input-file %t.txt
@@ -27,6 +24,7 @@
 
 #include "Inputs/kernels_in_file2.hpp"
 #include <CL/sycl.hpp>
+#include <cstdio>
 #include <iostream>
 #include <thread>
 
@@ -85,6 +83,15 @@ void runTestForTid(queue *Q, size_t Tid) {
 }
 
 int main(int Argc, const char *Argv[]) {
+#ifndef __SYCL_DEVICE_ONLY__
+  // On windows stderr output becomes messed up if several thread
+  // output simultaneously. Hence, setting explicit line buffering here.
+  if (setvbuf(stderr, nullptr, _IOLBF, BUFSIZ)) {
+    std::cerr << "Can't set line-buffering mode fo stderr\n";
+    return 1;
+  }
+#endif
+
   std::vector<std::thread> threadPool;
   threadPool.reserve(NUM_THREADS);
 

SYCL/AtomicRef/add.cpp

@@ -1,7 +1,7 @@
 // See https://github.com/intel/llvm-test-suite/issues/867 for detailed status
 // UNSUPPORTED: hip
 
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-device-code-split=per_kernel -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-device-code-split=per_kernel -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/add.h

@@ -287,10 +287,6 @@ template <access::address_space space, typename T, typename Difference = T,
 void add_test_scopes(queue q, size_t N) {
   std::vector<memory_scope> scopes =
       q.get_device().get_info<info::device::atomic_memory_scope_capabilities>();
-  if (std::find(scopes.begin(), scopes.end(), memory_scope::system) !=
-      scopes.end()) {
-    add_test<space, T, Difference, order, memory_scope::system>(q, N);
-  }
   if (std::find(scopes.begin(), scopes.end(), memory_scope::work_group) !=
       scopes.end()) {
     add_test<space, T, Difference, order, memory_scope::work_group>(q, N);

SYCL/AtomicRef/add_generic.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/add_generic_local.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/add_generic_local_native_fp.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/add_generic_native_fp.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/add_local.cpp

@@ -1,12 +1,12 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %ACC_RUN_PLACEHOLDER %t.out
 
-// Barrier is not supported on host. HIP and ACC  do not support floating
-// point atomics.
-// XFAIL: host, hip, acc
+// Barrier is not supported on host. HIP does not support floating point
+// atomics.
+// XFAIL: host, hip
 
 #include "add.h"
 

SYCL/AtomicRef/add_local_native_fp.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/add_native_fp.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/and.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/and.h

@@ -136,10 +136,6 @@ template <access::address_space space, typename T,
 void and_test_scopes(queue q) {
   std::vector<memory_scope> scopes =
       q.get_device().get_info<info::device::atomic_memory_scope_capabilities>();
-  if (std::find(scopes.begin(), scopes.end(), memory_scope::system) !=
-      scopes.end()) {
-    and_test<space, T, order, memory_scope::system>(q);
-  }
   if (std::find(scopes.begin(), scopes.end(), memory_scope::work_group) !=
       scopes.end()) {
     and_test<space, T, order, memory_scope::work_group>(q);

SYCL/AtomicRef/and_generic.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/and_generic_local.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/AtomicRef/and_local.cpp

@@ -1,4 +1,4 @@
-// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out