Merge remote-tracking branch 'upstream/sycl' into sycl

Author: bso-intel

llvm/test/Other/cfg-printer-branch-weights.ll

@@ -1,6 +1,9 @@
 ;RUN: opt < %s -analyze -dot-cfg -cfg-dot-filename-prefix=%t 2>/dev/null
 ;RUN: FileCheck %s -input-file=%t.f.dot
 
+;TODO: Investigate why this test doesn't work with dynamically linked libraries
+;REQUIRES: static-libs
+
 define void @f(i32) {
 entry:
   %check = icmp sgt i32 %0, 0

llvm/test/Other/cfg_deopt_unreach.ll

@@ -9,6 +9,9 @@
 ; RUN: opt < %s -analyze -dot-cfg -cfg-hide-unreachable-paths -cfg-hide-deoptimize-paths -cfg-dot-filename-prefix=%t/both-flags 2>/dev/null
 ; RUN: FileCheck %s -input-file=%t/both-flags.callee.dot -check-prefix=BOTH-FLAGS
 
+; TODO: Investigate why this test doesn't work with dynamically linked libraries
+; REQUIRES: static-libs
+
 declare i8 @llvm.experimental.deoptimize.i8(...)
 
 define i8 @callee(i1* %c) alwaysinline {

sycl/doc/GetStartedGuide.md

@@ -24,6 +24,7 @@ and a wide range of compute accelerators such as GPU and FPGA.
 * `git` - https://git-scm.com/downloads
 * `cmake` version 3.2 or later - http://www.cmake.org/download/
 * `python` - https://www.python.org/downloads/release/python-2716/
+* `ninja` - https://github.com/ninja-build/ninja/wiki/Pre-built-Ninja-packages
 * C++ compiler
   * Linux: `GCC` version 5.1.0 or later (including libstdc++) -
     https://gcc.gnu.org/install/
@@ -259,13 +260,8 @@ c:\oclcpu_rt_<cpu_version>\install.bat c:\tbb_<tbb_version>\tbb\bin\intel64\vc14
 
 To verify that built DPC++ toolchain is working correctly, run:
 
-**Linux**
+**Linux/Windows (64-bit)**
 ```bash
-make -j`nproc` check-all
-```
-
-**Windows (64-bit)**
-```bat
 ninja check-all
 ```
 

sycl/include/CL/sycl/detail/builtins.hpp

@@ -24,10 +24,6 @@
 #define __SYCL_EXTERN_IT2(Ret, prefix, call, Arg1, Arg2)
 #define __SYCL_EXTERN_IT2_SAME(Ret, prefix, call, Arg)
 #define __SYCL_EXTERN_IT3(Ret, prefix, call, Arg1, Arg2, Arg3)
-#define __SYCL_TPARAMS0(Type)
-#define __SYCL_TPARAMS1(Ret, A1)
-#define __SYCL_TPARAMS2(Ret, A1, A2)
-#define __SYCL_TPARAMS3(Ret, A1, A2, A3)
 #else
 #define __FUNC_PREFIX_OCL
 #define __FUNC_PREFIX_CORE
@@ -39,10 +35,6 @@
   extern Ret __SYCL_PPCAT(prefix, call)(Arg1, Arg2)
 #define __SYCL_EXTERN_IT3(Ret, prefix, call, Arg1, Arg2, Arg3)                 \
   extern Ret __SYCL_PPCAT(prefix, call)(Arg1, Arg2, Arg3)
-#define __SYCL_TPARAMS0(Type)
-#define __SYCL_TPARAMS1(Ret, A1)
-#define __SYCL_TPARAMS2(Ret, A1, A2)
-#define __SYCL_TPARAMS3(Ret, A1, A2, A3)
 #endif
 
 #define __SYCL_PPCAT_NX(A, B) A##B
@@ -55,7 +47,7 @@
     using Arg1 = cl::sycl::detail::ConvertToOpenCLType_t<T1>;                  \
     __SYCL_EXTERN_IT1(Ret, prefix, call, Arg1);                                \
     Arg1 arg1 = cl::sycl::detail::convertDataToType<T1, Arg1>(t1);             \
-    Ret ret = __SYCL_PPCAT(prefix, call) __SYCL_TPARAMS1(Ret, Arg1)(arg1);     \
+    Ret ret = __SYCL_PPCAT(prefix, call)(arg1);                                \
     return cl::sycl::detail::convertDataToType<Ret, R>(ret);                   \
   }
 
@@ -68,8 +60,7 @@
     __SYCL_EXTERN_IT2(Ret, prefix, call, Arg1, Arg2);                          \
     Arg1 arg1 = cl::sycl::detail::convertDataToType<T1, Arg1>(t1);             \
     Arg2 arg2 = cl::sycl::detail::convertDataToType<T2, Arg2>(t2);             \
-    Ret ret = __SYCL_PPCAT(prefix, call)                                       \
-        __SYCL_TPARAMS2(Ret, Arg1, Arg2)(arg1, arg2);                          \
+    Ret ret = __SYCL_PPCAT(prefix, call)(arg1, arg2);                          \
     return cl::sycl::detail::convertDataToType<Ret, R>(ret);                   \
   }
 
@@ -81,8 +72,7 @@
     __SYCL_EXTERN_IT2_SAME(Ret, prefix, call, Arg);                            \
     Arg arg1 = cl::sycl::detail::convertDataToType<T, Arg>(t1);                \
     Arg arg2 = cl::sycl::detail::convertDataToType<T, Arg>(t2);                \
-    Ret ret =                                                                  \
-        __SYCL_PPCAT(prefix, call) __SYCL_TPARAMS1(Ret, Arg)(arg1, arg2);      \
+    Ret ret = __SYCL_PPCAT(prefix, call)(arg1, arg2);                          \
     return cl::sycl::detail::convertDataToType<Ret, R>(ret);                   \
   }
 
@@ -93,7 +83,7 @@
     __SYCL_EXTERN_IT2_SAME(Type, prefix, call, Type);                          \
     Type arg1 = cl::sycl::detail::convertDataToType<T, Type>(v1);              \
     Type arg2 = cl::sycl::detail::convertDataToType<T, Type>(v2);              \
-    Type ret = __SYCL_PPCAT(prefix, call) __SYCL_TPARAMS0(Type)(arg1, arg2);   \
+    Type ret = __SYCL_PPCAT(prefix, call)(arg1, arg2);                         \
     return cl::sycl::detail::convertDataToType<Type, T>(ret);                  \
   }
 
@@ -108,8 +98,7 @@
     Arg1 arg1 = cl::sycl::detail::convertDataToType<T1, Arg1>(t1);             \
     Arg2 arg2 = cl::sycl::detail::convertDataToType<T2, Arg2>(t2);             \
     Arg3 arg3 = cl::sycl::detail::convertDataToType<T3, Arg3>(t3);             \
-    Ret ret = __SYCL_PPCAT(prefix, call)                                       \
-        __SYCL_TPARAMS3(Ret, Arg1, Arg2, Arg3)(arg1, arg2, arg3);              \
+    Ret ret = __SYCL_PPCAT(prefix, call)(arg1, arg2, arg3);                    \
     return cl::sycl::detail::convertDataToType<Ret, R>(ret);                   \
   }
 
@@ -302,10 +291,6 @@ __SYCL_MAKE_CALL_ARG3(select, __FUNC_PREFIX_OCL) // select
 #undef __SYCL_PPCAT
 #undef __FUNC_PREFIX_OCL
 #undef __FUNC_PREFIX_CORE
-#undef __SYCL_TPARAMS0
-#undef __SYCL_TPARAMS1
-#undef __SYCL_TPARAMS2
-#undef __SYCL_TPARAMS3
 #undef __SYCL_EXTERN_IT1
 #undef __SYCL_EXTERN_IT2
 #undef __SYCL_EXTERN_IT2_SAME

sycl/source/CMakeLists.txt

@@ -105,8 +105,6 @@ set(SYCL_SOURCES
     "detail/scheduler/graph_builder.cpp"
     "detail/spec_constant_impl.cpp"
     "detail/sycl_mem_obj_t.cpp"
-    "detail/usm/clusm.cpp"
-    "detail/usm/usm_dispatch.cpp"
     "detail/usm/usm_impl.cpp"
     "detail/util.cpp"
     "accessor.cpp"

sycl/source/detail/context_impl.hpp

@@ -17,7 +17,6 @@
 #include <detail/kernel_program_cache.hpp>
 #include <detail/platform_impl.hpp>
 #include <detail/program_manager/program_manager.hpp>
-#include <detail/usm/usm_dispatch.hpp>
 
 #include <map>
 #include <memory>
@@ -152,7 +151,6 @@ class context_impl {
   bool MPluginInterop;
   bool MHostContext;
   bool MUseCUDAPrimaryContext;
-  std::shared_ptr<usm::USMDispatcher> MUSMDispatch;
   std::map<DeviceLibExt, RT::PiProgram> MCachedLibPrograms;
   mutable KernelProgramCache MKernelProgramCache;
 };

sycl/source/detail/memory_manager.cpp

@@ -10,7 +10,6 @@
 #include <detail/context_impl.hpp>
 #include <detail/event_impl.hpp>
 #include <detail/queue_impl.hpp>
-#include <detail/usm/usm_dispatch.hpp>
 
 #include <algorithm>
 #include <cassert>

sycl/source/detail/platform_impl.cpp

@@ -65,7 +65,7 @@ struct DevDescT {
   int platformVerSize = 0;
 };
 
-static std::vector<DevDescT> getWhiteListDesc() {
+static std::vector<DevDescT> getAllowListDesc() {
   const char *str = SYCLConfig<SYCL_DEVICE_ALLOWLIST>::get();
   if (!str)
     return {};
@@ -101,15 +101,13 @@ static std::vector<DevDescT> getWhiteListDesc() {
     }
 
     if (':' != *str)
-      throw sycl::runtime_error("Malformed device white list",
-                                PI_INVALID_VALUE);
+      throw sycl::runtime_error("Malformed device allowlist", PI_INVALID_VALUE);
 
     // Skip ':'
     str += 1;
 
     if ('{' != *str || '{' != *(str + 1))
-      throw sycl::runtime_error("Malformed device white list",
-                                PI_INVALID_VALUE);
+      throw sycl::runtime_error("Malformed device allowlist", PI_INVALID_VALUE);
 
     // Skip opening sequence "{{"
     str += 2;
@@ -121,8 +119,7 @@ static std::vector<DevDescT> getWhiteListDesc() {
       ++str;
 
     if ('\0' == *str)
-      throw sycl::runtime_error("Malformed device white list",
-                                PI_INVALID_VALUE);
+      throw sycl::runtime_error("Malformed device allowlist", PI_INVALID_VALUE);
 
     *size = str - *valuePtr;
 
@@ -136,20 +133,18 @@ static std::vector<DevDescT> getWhiteListDesc() {
     if ('|' == *str)
       decDescs.emplace_back();
     else if (',' != *str)
-      throw sycl::runtime_error("Malformed device white list",
-                                PI_INVALID_VALUE);
+      throw sycl::runtime_error("Malformed device allowlist", PI_INVALID_VALUE);
 
     ++str;
   }
 
   return decDescs;
 }
 
-static void filterWhiteList(vector_class<RT::PiDevice> &PiDevices,
-                            RT::PiPlatform PiPlatform,
-                            const plugin &Plugin) {
-  const std::vector<DevDescT> WhiteList(getWhiteListDesc());
-  if (WhiteList.empty())
+static void filterAllowList(vector_class<RT::PiDevice> &PiDevices,
+                            RT::PiPlatform PiPlatform, const plugin &Plugin) {
+  const std::vector<DevDescT> AllowList(getAllowListDesc());
+  if (AllowList.empty())
     return;
 
   const string_class PlatformName =
@@ -170,7 +165,7 @@ static void filterWhiteList(vector_class<RT::PiDevice> &PiDevices,
     const string_class DeviceDriverVer = sycl::detail::get_device_info<
         string_class, info::device::driver_version>::get(Device, Plugin);
 
-    for (const DevDescT &Desc : WhiteList) {
+    for (const DevDescT &Desc : AllowList) {
       if (nullptr != Desc.platformName &&
           !std::regex_match(PlatformName,
                             std::regex(std::string(Desc.platformName,
@@ -229,9 +224,9 @@ platform_impl::get_devices(info::device_type DeviceType) const {
                                        pi::cast<RT::PiDeviceType>(DeviceType),
                                        NumDevices, PiDevices.data(), nullptr);
 
-  // Filter out devices that are not present in the white list
+  // Filter out devices that are not present in the allowlist
   if (SYCLConfig<SYCL_DEVICE_ALLOWLIST>::get())
-    filterWhiteList(PiDevices, MPlatform, this->getPlugin());
+    filterAllowList(PiDevices, MPlatform, this->getPlugin());
 
   std::transform(PiDevices.begin(), PiDevices.end(), std::back_inserter(Res),
                  [this](const RT::PiDevice &PiDevice) -> device {

sycl/source/detail/queue_impl.cpp

@@ -12,7 +12,6 @@
 #include <CL/sycl/detail/pi.hpp>
 #include <CL/sycl/device.hpp>
 #include <detail/queue_impl.hpp>
-#include <detail/usm/usm_dispatch.hpp>
 
 #include <cstring>