Added the intrinsic to clang

Author: adam-yang

clang/include/clang/Basic/Builtins.td

@@ -4924,6 +4924,12 @@ def HLSLClip: LangBuiltin<"HLSL_LANG"> {
   let Prototype = "void(...)";
 }
 
+def HLSLGroupMemoryBarrierWithGroupSync: LangBuiltin<"HLSL_LANG"> {
+  let Spellings = ["__builtin_group_memory_barrier_with_group_sync"];
+  let Attributes = [NoThrow, Const];
+  let Prototype = "void(...)";
+}
+
 // Builtins for XRay.
 def XRayCustomEvent : Builtin {
   let Spellings = ["__xray_customevent"];

clang/lib/CodeGen/CGBuiltin.cpp

@@ -19446,6 +19446,10 @@ case Builtin::BI__builtin_hlsl_elementwise_isinf: {
     assert(E->getArg(0)->getType()->hasFloatingRepresentation() &&
            "clip operands types mismatch");
     return handleHlslClip(E, this);
+  case Builtin::BI__builtin_group_memory_barrier_with_group_sync: {
+    Intrinsic::ID ID = CGM.getHLSLRuntime().getGroupMemoryBarrierWithGroupSyncIntrinsic();
+    return EmitRuntimeCall(Intrinsic::getDeclaration(&CGM.getModule(), ID));
+  }
   }
   return nullptr;
 }

clang/lib/CodeGen/CGHLSLRuntime.h

@@ -103,6 +103,7 @@ class CGHLSLRuntime {
 
   GENERATE_HLSL_INTRINSIC_FUNCTION(CreateHandleFromBinding, handle_fromBinding)
   GENERATE_HLSL_INTRINSIC_FUNCTION(BufferUpdateCounter, bufferUpdateCounter)
+  GENERATE_HLSL_INTRINSIC_FUNCTION(GroupMemoryBarrierWithGroupSync, groupMemoryBarrierWithGroupSync)
 
   //===----------------------------------------------------------------------===//
   // End of reserved area for HLSL intrinsic getters.

clang/lib/Headers/hlsl/hlsl_intrinsics.h

@@ -2481,5 +2481,17 @@ float3 radians(float3);
 _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_radians)
 float4 radians(float4);
 
+//===----------------------------------------------------------------------===//
+// GroupMemoryBarrierWithGroupSync builtins
+//===----------------------------------------------------------------------===//
+
+/// \fn void GroupMemoryBarrierWithGroupSync(void)
+/// \brief Blocks execution of all threads in a group until all group shared
+/// accesses have been completed and all threads in the group have reached this
+/// call.
+
+_HLSL_BUILTIN_ALIAS(__builtin_group_memory_barrier_with_group_sync)
+void GroupMemoryBarrierWithGroupSync(void);
+
 } // namespace hlsl
 #endif //_HLSL_HLSL_INTRINSICS_H_

clang/lib/Sema/SemaHLSL.cpp

@@ -2239,6 +2239,10 @@ bool SemaHLSL::CheckBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
           << 1;
       return true;
     }
+  }
+  case Builtin::BI__builtin_group_memory_barrier_with_group_sync: {
+    if (SemaRef.checkArgCountAtMost(TheCall, 0))
+      return true;
     break;
   }
   }

clang/test/CodeGenHLSL/builtins/GroupMemoryBarrierWithGroupSync-errors.hlsl

@@ -0,0 +1,6 @@
+// RUN: %clang_cc1 -finclude-default-header -triple dxil-pc-shadermodel6.6-library %s -fnative-half-type -emit-llvm-only -disable-llvm-passes -verify -verify-ignore-unexpected
+
+void test_too_many_arg() {
+  __builtin_group_memory_barrier_with_group_sync(0);
+  // expected-error@-1 {{too many arguments to function call, expected at most 0, have 1}}
+}

clang/test/CodeGenHLSL/builtins/GroupMemoryBarrierWithGroupSync.hlsl

@@ -0,0 +1,22 @@
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   dxil-pc-shadermodel6.3-compute %s \
+// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s \
+// RUN:   -DTARGET=dx -DFNATTRS=noundef
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   spirv-unknown-vulkan-compute %s \
+// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s \
+// RUN:   -DTARGET=spv -DFNATTRS="spir_func noundef"
+
+// CHECK: define [[FNATTRS]] i32 @
+[numthreads(1, 1, 1)]
+void main() {
+  while (true) {
+// CHECK: call void @llvm.[[TARGET]].groupMemoryBarrierWithGroupSync()
+  GroupMemoryBarrierWithGroupSync();
+  break;
+  }
+}
+
+// CHECK: declare void @llvm.[[TARGET]].groupMemoryBarrierWithGroupSync() #[[ATTRS:[0-9]+]]
+// CHECK-NOT: attributes #[[ATTRS]] = {{.+}}memory(none){{.+}}
+// CHECK: attributes #[[ATTRS]] = {