[SYCL] Use perfect forwarding around host tasks properly (#20732)

Fixes #20722.

Author: aelovikov-intel

sycl/include/sycl/handler.hpp

@@ -1457,8 +1457,8 @@ class __SYCL_EXPORT handler {
   template <typename SharedPtrT>
   void setHandlerKernelBundle(SharedPtrT &&NewKernelBundleImpPtr);
 
-  void SetHostTask(std::function<void()> &&Func);
-  void SetHostTask(std::function<void(interop_handle)> &&Func);
+  void SetHostTask(std::function<void()> Func);
+  void SetHostTask(std::function<void(interop_handle)> Func);
 
   template <typename FuncT>
   std::enable_if_t<detail::check_fn_signature<std::remove_reference_t<FuncT>,
@@ -1472,7 +1472,7 @@ class __SYCL_EXPORT handler {
     // accessors during finalize
     setArgsToAssociatedAccessors();
 
-    SetHostTask(std::move(Func));
+    SetHostTask(std::forward<FuncT>(Func));
   }
 
   template <typename FuncT>
@@ -1485,7 +1485,7 @@ class __SYCL_EXPORT handler {
     // accessors during finalize
     setArgsToAssociatedAccessors();
 
-    SetHostTask(std::move(Func));
+    SetHostTask(std::forward<FuncT>(Func));
     setType(detail::CGType::EnqueueNativeCommand);
   }
 
@@ -1707,7 +1707,7 @@ class __SYCL_EXPORT handler {
                    detail::check_fn_signature<std::remove_reference_t<FuncT>,
                                               void(interop_handle)>::value>
   host_task(FuncT &&Func) {
-    host_task_impl(Func);
+    host_task_impl(std::forward<FuncT>(Func));
   }
 
   /// Enqueues a command to the SYCL runtime to invoke \p Func immediately.
@@ -1716,7 +1716,7 @@ class __SYCL_EXPORT handler {
                                               void(interop_handle)>::value>
   ext_codeplay_enqueue_native_command([[maybe_unused]] FuncT &&Func) {
 #ifndef __SYCL_DEVICE_ONLY__
-    ext_codeplay_enqueue_native_command_impl(Func);
+    ext_codeplay_enqueue_native_command_impl(std::forward<FuncT>(Func));
 #endif
   }
 

sycl/source/handler.cpp

@@ -2045,13 +2045,13 @@ void *handler::storeRawArg(const void *Ptr, size_t Size) {
   return Storage;
 }
 
-void handler::SetHostTask(std::function<void()> &&Func) {
+void handler::SetHostTask(std::function<void()> Func) {
   setNDRangeDescriptor(range<1>(1));
   impl->MHostTask.reset(new detail::HostTask(std::move(Func)));
   setType(detail::CGType::CodeplayHostTask);
 }
 
-void handler::SetHostTask(std::function<void(interop_handle)> &&Func) {
+void handler::SetHostTask(std::function<void(interop_handle)> Func) {
   setNDRangeDescriptor(range<1>(1));
   impl->MHostTask.reset(new detail::HostTask(std::move(Func)));
   setType(detail::CGType::CodeplayHostTask);

sycl/test/abi/sycl_symbols_linux.dump

@@ -3545,8 +3545,8 @@ _ZN4sycl3_V17handler10depends_onERKSt6vectorINS0_5eventESaIS3_EE
 _ZN4sycl3_V17handler10depends_onERKSt6vectorISt10shared_ptrINS0_6detail10event_implEESaIS6_EE
 _ZN4sycl3_V17handler10mem_adviseEPKvmi
 _ZN4sycl3_V17handler10processArgEPvRKNS0_6detail19kernel_param_kind_tEimRmbb
-_ZN4sycl3_V17handler11SetHostTaskEOSt8functionIFvNS0_14interop_handleEEE
-_ZN4sycl3_V17handler11SetHostTaskEOSt8functionIFvvEE
+_ZN4sycl3_V17handler11SetHostTaskESt8functionIFvNS0_14interop_handleEEE
+_ZN4sycl3_V17handler11SetHostTaskESt8functionIFvvEE
 _ZN4sycl3_V17handler11copyCodeLocERKS1_
 _ZN4sycl3_V17handler11saveCodeLocENS0_6detail13code_locationE
 _ZN4sycl3_V17handler11saveCodeLocENS0_6detail13code_locationEb

sycl/test/abi/sycl_symbols_windows.dump

@@ -712,8 +712,8 @@
 ?SecondaryQueue@SubmissionInfo@detail@_V1@sycl@@QEBAAEBV?$shared_ptr@Vqueue_impl@detail@_V1@sycl@@@std@@XZ
 ?SecondaryQueue@SubmissionInfo@v1@detail@_V1@sycl@@QEAAAEAV?$shared_ptr@Vqueue_impl@detail@_V1@sycl@@@std@@XZ
 ?SecondaryQueue@SubmissionInfo@v1@detail@_V1@sycl@@QEBAAEBV?$shared_ptr@Vqueue_impl@detail@_V1@sycl@@@std@@XZ
-?SetHostTask@handler@_V1@sycl@@AEAAX$$QEAV?$function@$$A6AXVinterop_handle@_V1@sycl@@@Z@std@@@Z
-?SetHostTask@handler@_V1@sycl@@AEAAX$$QEAV?$function@$$A6AXXZ@std@@@Z
+?SetHostTask@handler@_V1@sycl@@AEAAXV?$function@$$A6AXVinterop_handle@_V1@sycl@@@Z@std@@@Z
+?SetHostTask@handler@_V1@sycl@@AEAAXV?$function@$$A6AXXZ@std@@@Z
 ?SetKernelLaunchpropertiesIfNotEmpty@handler@_V1@sycl@@AEAAXAEBU?$PropsHolder@Uwork_group_scratch_size@experimental@oneapi@ext@_V1@sycl@@Ucache_config@2intel@456@Uuse_root_sync_key@23456@Uwork_group_progress_key@23456@Usub_group_progress_key@23456@Uwork_item_progress_key@23456@U?$cluster_size@$00@cuda@23456@U?$cluster_size@$01@cuda@23456@U?$cluster_size@$02@cuda@23456@@kernel_launch_properties_v1@detail@23@@Z
 ?__abs_diff_impl@_V1@sycl@@YA?AV?$vec@C$00@12@V312@0@Z
 ?__abs_diff_impl@_V1@sycl@@YA?AV?$vec@C$01@12@V312@0@Z