From 10861d8f252c45b7c4b579e530e744911d3170d2 Mon Sep 17 00:00:00 2001
From: Alexey Bader <alexey.bader@intel.com>
Date: Wed, 9 Sep 2020 14:46:18 +0300
Subject: [PATCH 1/3] [SYCL][NFC] Remove workaround for SPIR-V translator
 issues

---
 sycl/test/basic_tests/stream/stream.cpp | 6 +-----
 sycl/test/hier_par/hier_par_wgscope.cpp | 5 +----
 sycl/test/sub_group/generic-shuffle.cpp | 5 +----
 sycl/test/sub_group/generic_reduce.cpp  | 7 ++-----
 sycl/test/sub_group/load_store.cpp      | 5 +----
 sycl/test/sub_group/scan_fp16.cpp       | 5 +----
 sycl/test/sub_group/shuffle.cpp         | 5 +----
 sycl/test/sub_group/shuffle_fp16.cpp    | 5 +----
 sycl/test/sub_group/shuffle_fp64.cpp    | 5 +----
 9 files changed, 10 insertions(+), 38 deletions(-)

diff --git a/sycl/test/basic_tests/stream/stream.cpp b/sycl/test/basic_tests/stream/stream.cpp
index 34cef884152a..6b5e6925298d 100644
--- a/sycl/test/basic_tests/stream/stream.cpp
+++ b/sycl/test/basic_tests/stream/stream.cpp
@@ -1,8 +1,4 @@
-// XFAIL: cpu
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out | FileCheck %s
 // RUN: %CPU_RUN_PLACEHOLDER %t.out %CPU_CHECK_PLACEHOLDER
 // RUN: %GPU_RUN_ON_LINUX_PLACEHOLDER %t.out %GPU_CHECK_ON_LINUX_PLACEHOLDER
diff --git a/sycl/test/hier_par/hier_par_wgscope.cpp b/sycl/test/hier_par/hier_par_wgscope.cpp
index b0e58d305e1f..42022748dfe5 100644
--- a/sycl/test/hier_par/hier_par_wgscope.cpp
+++ b/sycl/test/hier_par/hier_par_wgscope.cpp
@@ -1,7 +1,4 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
diff --git a/sycl/test/sub_group/generic-shuffle.cpp b/sycl/test/sub_group/generic-shuffle.cpp
index 786a9b57a0ec..b8bb972d646a 100644
--- a/sycl/test/sub_group/generic-shuffle.cpp
+++ b/sycl/test/sub_group/generic-shuffle.cpp
@@ -1,10 +1,7 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
diff --git a/sycl/test/sub_group/generic_reduce.cpp b/sycl/test/sub_group/generic_reduce.cpp
index 98401e091e73..fdad3289ec10 100644
--- a/sycl/test/sub_group/generic_reduce.cpp
+++ b/sycl/test/sub_group/generic_reduce.cpp
@@ -1,11 +1,8 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-unnamed-lambda -std=c++14 %s -o %t.out
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple -std=c++14 -D SG_GPU %s -o %t_gpu.out
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -std=c++14 %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-unnamed-lambda -fsycl-targets=%sycl_triple -std=c++14 -D SG_GPU %s -o %t_gpu.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t_gpu.out
diff --git a/sycl/test/sub_group/load_store.cpp b/sycl/test/sub_group/load_store.cpp
index c51a01fbdaef..dae9db67e979 100644
--- a/sycl/test/sub_group/load_store.cpp
+++ b/sycl/test/sub_group/load_store.cpp
@@ -1,12 +1,9 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda || cpu
 // CUDA compilation and runtime do not yet support sub-groups.
 // #2252 Disable until all variants of built-ins are available in OpenCL CPU
 // runtime for every supported ISA
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
diff --git a/sycl/test/sub_group/scan_fp16.cpp b/sycl/test/sub_group/scan_fp16.cpp
index 13e3f45b2794..bfdb09f8f7f3 100644
--- a/sycl/test/sub_group/scan_fp16.cpp
+++ b/sycl/test/sub_group/scan_fp16.cpp
@@ -1,10 +1,7 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
 //==--------------- scan_fp16.cpp - SYCL sub_group scan test --------*- C++ -*---==//
diff --git a/sycl/test/sub_group/shuffle.cpp b/sycl/test/sub_group/shuffle.cpp
index b86b00dbf89b..5207716148ef 100644
--- a/sycl/test/sub_group/shuffle.cpp
+++ b/sycl/test/sub_group/shuffle.cpp
@@ -1,10 +1,7 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
diff --git a/sycl/test/sub_group/shuffle_fp16.cpp b/sycl/test/sub_group/shuffle_fp16.cpp
index 930035d37a67..62f07fc612de 100644
--- a/sycl/test/sub_group/shuffle_fp16.cpp
+++ b/sycl/test/sub_group/shuffle_fp16.cpp
@@ -1,10 +1,7 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 //
 //==------------ shuffle_fp16.cpp - SYCL sub_group shuffle test -----*- C++ -*---==//
diff --git a/sycl/test/sub_group/shuffle_fp64.cpp b/sycl/test/sub_group/shuffle_fp64.cpp
index 3b5c3994cec8..3b1ed5690760 100644
--- a/sycl/test/sub_group/shuffle_fp64.cpp
+++ b/sycl/test/sub_group/shuffle_fp64.cpp
@@ -1,10 +1,7 @@
-// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
-// See https://github.com/intel/llvm/issues/2264 for more details.
-
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

From 72f7a2540ed38e745013f59ded690d95188bc527 Mon Sep 17 00:00:00 2001
From: Alexey Bader <alexey.bader@intel.com>
Date: Wed, 9 Sep 2020 15:52:01 +0300
Subject: [PATCH 2/3] Use XFAIL to work-around issues caused by enabling
 optimizations

This way fixes will be detected automatically.
---
 sycl/test/basic_tests/boolean.cpp       | 3 ++-
 sycl/test/basic_tests/stream/stream.cpp | 4 ++++
 sycl/test/hier_par/hier_par_wgscope.cpp | 4 ++++
 sycl/test/sub_group/generic_reduce.cpp  | 4 ++++
 sycl/test/sub_group/load_store.cpp      | 4 ++++
 5 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/sycl/test/basic_tests/boolean.cpp b/sycl/test/basic_tests/boolean.cpp
index dec4aec312ed..031a0afafda0 100644
--- a/sycl/test/basic_tests/boolean.cpp
+++ b/sycl/test/basic_tests/boolean.cpp
@@ -1,7 +1,8 @@
 // TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
 // See https://github.com/intel/llvm/issues/2264 for more details.
+// XFAIL: gpu && (level_zero || opencl)
 
-// RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-targets=%sycl_triple %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
diff --git a/sycl/test/basic_tests/stream/stream.cpp b/sycl/test/basic_tests/stream/stream.cpp
index 6b5e6925298d..d7e42e6f5c8a 100644
--- a/sycl/test/basic_tests/stream/stream.cpp
+++ b/sycl/test/basic_tests/stream/stream.cpp
@@ -1,3 +1,7 @@
+// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
+// See https://github.com/intel/llvm/issues/2264 for more details.
+// XFAIL: gpu && (level_zero || opencl)
+
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out | FileCheck %s
 // RUN: %CPU_RUN_PLACEHOLDER %t.out %CPU_CHECK_PLACEHOLDER
diff --git a/sycl/test/hier_par/hier_par_wgscope.cpp b/sycl/test/hier_par/hier_par_wgscope.cpp
index 42022748dfe5..2167f3bfba8b 100644
--- a/sycl/test/hier_par/hier_par_wgscope.cpp
+++ b/sycl/test/hier_par/hier_par_wgscope.cpp
@@ -1,3 +1,7 @@
+// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
+// See https://github.com/intel/llvm/issues/2264 for more details.
+// XFAIL: gpu && (level_zero || opencl) && linux
+
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
diff --git a/sycl/test/sub_group/generic_reduce.cpp b/sycl/test/sub_group/generic_reduce.cpp
index fdad3289ec10..3ef504343af4 100644
--- a/sycl/test/sub_group/generic_reduce.cpp
+++ b/sycl/test/sub_group/generic_reduce.cpp
@@ -1,3 +1,7 @@
+// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
+// See https://github.com/intel/llvm/issues/2264 for more details.
+// XFAIL: gpu && linux
+
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.
 //
diff --git a/sycl/test/sub_group/load_store.cpp b/sycl/test/sub_group/load_store.cpp
index dae9db67e979..f6c96a02324d 100644
--- a/sycl/test/sub_group/load_store.cpp
+++ b/sycl/test/sub_group/load_store.cpp
@@ -1,3 +1,7 @@
+// TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
+// See https://github.com/intel/llvm/issues/2264 for more details.
+// XFAIL: gpu
+
 // UNSUPPORTED: cuda || cpu
 // CUDA compilation and runtime do not yet support sub-groups.
 // #2252 Disable until all variants of built-ins are available in OpenCL CPU

From 81f33074dad153abcc060c7f35bc196d1da451c8 Mon Sep 17 00:00:00 2001
From: Alexey Bader <alexey.bader@intel.com>
Date: Wed, 9 Sep 2020 16:55:59 +0300
Subject: [PATCH 3/3] Adjust checks for Windows

---
 sycl/test/basic_tests/stream/stream.cpp | 2 +-
 sycl/test/hier_par/hier_par_wgscope.cpp | 2 +-
 sycl/test/sub_group/generic_reduce.cpp  | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/sycl/test/basic_tests/stream/stream.cpp b/sycl/test/basic_tests/stream/stream.cpp
index d7e42e6f5c8a..441320da38f5 100644
--- a/sycl/test/basic_tests/stream/stream.cpp
+++ b/sycl/test/basic_tests/stream/stream.cpp
@@ -1,6 +1,6 @@
 // TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
 // See https://github.com/intel/llvm/issues/2264 for more details.
-// XFAIL: gpu && (level_zero || opencl)
+// XFAIL: gpu && (level_zero || opencl) && linux
 
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out | FileCheck %s
diff --git a/sycl/test/hier_par/hier_par_wgscope.cpp b/sycl/test/hier_par/hier_par_wgscope.cpp
index 2167f3bfba8b..698ac60bf848 100644
--- a/sycl/test/hier_par/hier_par_wgscope.cpp
+++ b/sycl/test/hier_par/hier_par_wgscope.cpp
@@ -1,6 +1,6 @@
 // TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
 // See https://github.com/intel/llvm/issues/2264 for more details.
-// XFAIL: gpu && (level_zero || opencl) && linux
+// XFAIL: gpu && (level_zero || opencl)
 
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env SYCL_DEVICE_TYPE=HOST %t.out
diff --git a/sycl/test/sub_group/generic_reduce.cpp b/sycl/test/sub_group/generic_reduce.cpp
index 3ef504343af4..676c6b284a39 100644
--- a/sycl/test/sub_group/generic_reduce.cpp
+++ b/sycl/test/sub_group/generic_reduce.cpp
@@ -1,6 +1,6 @@
 // TODO: Enable compilation w/o -fno-sycl-early-optimizations option.
 // See https://github.com/intel/llvm/issues/2264 for more details.
-// XFAIL: gpu && linux
+// XFAIL: gpu
 
 // UNSUPPORTED: cuda
 // CUDA compilation and runtime do not yet support sub-groups.