resolve merge conflict

Author: cperkinsintel

.github/CODEOWNERS

@@ -139,18 +139,18 @@ devops/dependencies-igc-dev.json @intel/sycl-matrix-reviewers @intel/dpcpp-esimd
 # Benchmarking scripts
 devops/scripts/benchmarks/ @intel/llvm-reviewers-benchmarking
 
-# Kernel fusion JIT compiler
-sycl-jit/ @intel/dpcpp-kernel-fusion-reviewers
-sycl/doc/design/KernelFusionJIT.md @intel/dpcpp-kernel-fusion-reviewers
-sycl/doc/extensions/experimental/sycl_ext_codeplay_kernel_fusion.asciidoc @intel/dpcpp-kernel-fusion-reviewers
-sycl/include/sycl/ext/codeplay/experimental/fusion_properties.hpp  @intel/dpcpp-kernel-fusion-reviewers
-sycl/include/sycl/ext/codeplay/experimental/fusion_wrapper.hpp @intel/dpcpp-kernel-fusion-reviewers
-sycl/source/detail/fusion/ @intel/dpcpp-kernel-fusion-reviewers
-sycl/source/detail/jit_compiler.hpp @intel/dpcpp-kernel-fusion-reviewers
-sycl/source/detail/jit_compiler.cpp @intel/dpcpp-kernel-fusion-reviewers
-sycl/source/detail/jit_device_binaries.hpp @intel/dpcpp-kernel-fusion-reviewers
-sycl/source/detail/jit_device_binaries.cpp @intel/dpcpp-kernel-fusion-reviewers
-sycl/test-e2e/KernelFusion @intel/dpcpp-kernel-fusion-reviewers
+# JIT compiler
+sycl-jit/ @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/doc/design/KernelFusionJIT.md @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/doc/extensions/experimental/sycl_ext_codeplay_kernel_fusion.asciidoc @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/include/sycl/ext/codeplay/experimental/fusion_properties.hpp  @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/include/sycl/ext/codeplay/experimental/fusion_wrapper.hpp @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/source/detail/fusion/ @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/source/detail/jit_compiler.hpp @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/source/detail/jit_compiler.cpp @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/source/detail/jit_device_binaries.hpp @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/source/detail/jit_device_binaries.cpp @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
+sycl/test-e2e/KernelFusion @intel/llvm-reviewers-runtime @cperkinsintel @aelovikov-intel
 
 # Matrix
 sycl/include/sycl/ext/oneapi/matrix/ @intel/sycl-matrix-reviewers

clang/lib/CodeGen/BackendUtil.cpp

@@ -1093,8 +1093,9 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
              ThinOrFullLTOPhase) {
             MPM.addPass(createModuleToFunctionPassAdaptor(
                 InferAddressSpacesPass(clang::targets::SPIR_GENERIC_AS)));
-            MPM.addPass(
-                createModuleToFunctionPassAdaptor(SYCLOptimizeBarriersPass()));
+            if (Level != OptimizationLevel::O0)
+              MPM.addPass(createModuleToFunctionPassAdaptor(
+                  SYCLOptimizeBarriersPass()));
           });
     }
 

clang/lib/Sema/SemaSYCL.cpp

@@ -6822,13 +6822,13 @@ class FreeFunctionPrinter {
         continue;
       }
 
-      TemplateName TN = TST->getTemplateName();
+      TemplateName CTN = CTST->getTemplateName();
+      CTN.getAsTemplateDecl()->printQualifiedName(ParmListOstream);
+      ParmListOstream << "<";
+
       auto SpecArgs = TST->template_arguments();
       auto DeclArgs = CTST->template_arguments();
 
-      TN.getAsTemplateDecl()->printQualifiedName(ParmListOstream);
-      ParmListOstream << "<";
-
       for (size_t I = 0, E = std::max(DeclArgs.size(), SpecArgs.size()),
                   SE = SpecArgs.size();
            I < E; ++I) {

clang/test/CodeGenSYCL/free-function-kernel-type-alias-arg.cpp

@@ -0,0 +1,63 @@
+// RUN: %clang_cc1 -fsycl-is-device -internal-isystem %S/Inputs -triple spir64-unknown-unknown -sycl-std=2020 -fsycl-int-header=%t.h %s
+// RUN: FileCheck -input-file=%t.h %s
+//
+// The purpose of this test is to ensure that forward declarations of free
+// function kernels are emitted properly.
+// However, this test checks a specific scenario:
+// - free function arguments are type aliases (through using or typedef)
+
+namespace ns {
+
+using IntUsing = int;
+typedef int IntTypedef;
+
+template <typename T>
+struct Foo {};
+
+using FooIntUsing = Foo<int>;
+typedef Foo<int> FooIntTypedef;
+
+template <typename T1, typename T2>
+struct Bar {};
+
+template<typename T1>
+using BarUsing = Bar<T1, float>;
+
+class Baz {
+public:
+  using type = BarUsing<double>;
+};
+
+} // namespace ns
+
+[[__sycl_detail__::add_ir_attributes_function("sycl-nd-range-kernel", 2)]]
+void int_using(ns::IntUsing Arg) {}
+
+// CHECK: void int_using(int Arg);
+
+[[__sycl_detail__::add_ir_attributes_function("sycl-nd-range-kernel", 2)]]
+void int_typedef(ns::IntTypedef Arg) {}
+
+// CHECK: void int_typedef(int Arg);
+
+[[__sycl_detail__::add_ir_attributes_function("sycl-nd-range-kernel", 2)]]
+void foo_using(ns::FooIntUsing Arg) {}
+
+// CHECK: void foo_using(ns::Foo<int> Arg);
+
+[[__sycl_detail__::add_ir_attributes_function("sycl-nd-range-kernel", 2)]]
+void foo_typedef(ns::FooIntTypedef Arg) {}
+
+// CHECK: void foo_typedef(ns::Foo<int> Arg);
+
+template<typename T>
+[[__sycl_detail__::add_ir_attributes_function("sycl-nd-range-kernel", 2)]]
+void bar_using(ns::BarUsing<T> Arg) {}
+template void bar_using(ns::BarUsing<int>);
+
+// CHECK: template <typename T> void bar_using(ns::Bar<T, float>);
+
+[[__sycl_detail__::add_ir_attributes_function("sycl-nd-range-kernel", 2)]]
+void baz_type(ns::Baz::type Arg) {}
+
+// CHECK: void baz_type(ns::Bar<double, float> Arg);

clang/test/CodeGenSYCL/kernel-early-optimization-pipeline.cpp

@@ -12,6 +12,7 @@
 // CHECK: InferFunctionAttrsPass
 // CHECK: AlwaysInlinerPass
 // CHECK: ModuleInlinerWrapperPass
+// CHECK: SYCLOptimizeBarriersPass
 // CHECK: ConstantMergePass
 // CHECK: SYCLMutatePrintfAddrspacePass
 // CHECK: SYCLPropagateAspectsUsagePass
@@ -22,4 +23,13 @@
 //
 // RUN: %clang_cc1 -O2 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -fno-sycl-early-optimizations -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-NOEARLYOPT
 // CHECK-NOEARLYOPT-NOT: ConstantMergePass1
+// CHECK-NOEARLYOPT-NOT: SYCLOptimizeBarriersPass
 // CHECK-NOEARLYOPT: SYCLMutatePrintfAddrspacePass
+
+// RUN: %clang_cc1 -O0 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-O0
+// CHECK-O0-NOT: SYCLOptimizeBarriersPass
+
+// Passes registered via registerOptimizerLastEPCallback don't run on empty
+// code
+__attribute__((sycl_device)) void bar() {
+}

devops/dependencies-igc-dev.json

@@ -1,10 +1,10 @@
 {
   "linux": {
     "igc_dev": {
-      "github_tag": "igc-dev-4e160b3",
-      "version": "4e160b3",
-      "updated_at": "2025-09-11T02:09:40Z",
-      "url": "https://api.github.com/repos/intel/intel-graphics-compiler/actions/artifacts/3981554499/zip",
+      "github_tag": "igc-dev-6d3f69e",
+      "version": "6d3f69e",
+      "updated_at": "2025-09-20T11:44:56Z",
+      "url": "https://api.github.com/repos/intel/intel-graphics-compiler/actions/artifacts/4062749149/zip",
       "root": "{DEPS_ROOT}/opencl/runtime/linux/oclgpu"
     }
   }

devops/scripts/benchmarks/benches/compute.py

@@ -218,19 +218,6 @@ def benchmarks(self) -> list[Benchmark]:
 
             # Add GraphApiSubmitGraph benchmarks
             for in_order_queue in [0, 1]:
-                for profiler_type in profiler_types:
-                    benches.append(
-                        GraphApiSubmitGraph(
-                            self,
-                            runtime,
-                            in_order_queue,
-                            self.submit_graph_num_kernels[-1],
-                            0,
-                            profiler_type,
-                            useEvents=0,
-                            useHostTasks=1,
-                        )
-                    )
                 for num_kernels in self.submit_graph_num_kernels:
                     for measure_completion_time in [0, 1]:
                         for use_events in [0, 1]:
@@ -1009,6 +996,9 @@ def __init__(
             profiler_type,
         )
 
+    def supported_runtimes(self) -> list[RUNTIMES]:
+        return super().supported_runtimes() + [RUNTIMES.SYCL_PREVIEW]
+
     def explicit_group(self):
         return f"SubmitGraph {self.ioq_str}{self.measure_str}{self.use_events_str}{self.host_tasks_str}, {self.numKernels} kernels"
 

libclc/utils/libclc-remangler/LibclcRemangler.cpp

@@ -883,6 +883,9 @@ class LibCLCRemangler : public ASTConsumer {
   }
 
   bool remangleFunction(Function &Func, llvm::Module *M) {
+    if (Func.hasLocalLinkage())
+      return true;
+
     if (!Func.getName().starts_with("_Z"))
       return true;
 

libdevice/include/sanitizer_defs.hpp

@@ -61,9 +61,6 @@ extern SYCL_EXTERNAL __attribute__((convergent)) void
 __spirv_ControlBarrier(int32_t Execution, int32_t Memory,
                        int32_t Semantics) noexcept;
 
-template <typename T>
-extern SYCL_EXTERNAL T __spirv_SpecConstant(int ID, T default_value) noexcept;
-
 extern "C" SYCL_EXTERNAL void __devicelib_exit();
 
 #endif // __SPIR__ || __SPIRV__

libdevice/include/sanitizer_utils.hpp

@@ -7,8 +7,8 @@
 //===----------------------------------------------------------------------===//
 #pragma once
 
-#include "sanitizer_common/sanitizer_libdevice.hpp"
 #include "sanitizer_defs.hpp"
+#include "spirv_vars.h"
 
 #if defined(__SPIR__) || defined(__SPIRV__)
 
@@ -49,11 +49,6 @@ inline __SYCL_PRIVATE__ void *ToPrivate(void *ptr) {
   return __spirv_GenericCastToPtrExplicit_ToPrivate(ptr, 7);
 }
 
-inline DeviceType GetDeviceTy() {
-  return static_cast<DeviceType>(
-      __spirv_SpecConstant(SPEC_CONSTANT_DEVICE_TYPE_ID, 0));
-}
-
 template <typename T> T Memset(T ptr, int value, size_t size) {
   for (size_t i = 0; i < size; i++) {
     ptr[i] = value;