[HLSL][SPIRV] Add convergence tokens to entry point wrapper (#112757)

Inlining currently assumes that either all function use controled
convergence or none of them do. This is why we need to have the entry
point wrapper use controled convergence.


https://github.com/llvm/llvm-project/blob/c85611e8583e6392d56075ebdfa60893b6284813/llvm/lib/Transforms/Utils/InlineFunction.cpp#L2431-L2439

Author: s-perron

clang/lib/CodeGen/CGHLSLRuntime.cpp

@@ -404,6 +404,16 @@ void CGHLSLRuntime::emitEntryFunction(const FunctionDecl *FD,
   BasicBlock *BB = BasicBlock::Create(Ctx, "entry", EntryFn);
   IRBuilder<> B(BB);
   llvm::SmallVector<Value *> Args;
+
+  SmallVector<OperandBundleDef, 1> OB;
+  if (CGM.shouldEmitConvergenceTokens()) {
+    assert(EntryFn->isConvergent());
+    llvm::Value *I = B.CreateIntrinsic(
+        llvm::Intrinsic::experimental_convergence_entry, {}, {});
+    llvm::Value *bundleArgs[] = {I};
+    OB.emplace_back("convergencectrl", bundleArgs);
+  }
+
   // FIXME: support struct parameters where semantics are on members.
   // See: https://github.com/llvm/llvm-project/issues/57874
   unsigned SRetOffset = 0;
@@ -419,7 +429,7 @@ void CGHLSLRuntime::emitEntryFunction(const FunctionDecl *FD,
     Args.push_back(emitInputSemantic(B, *PD, Param.getType()));
   }
 
-  CallInst *CI = B.CreateCall(FunctionCallee(Fn), Args);
+  CallInst *CI = B.CreateCall(FunctionCallee(Fn), Args, OB);
   CI->setCallingConv(Fn->getCallingConv());
   // FIXME: Handle codegen for return type semantics.
   // See: https://github.com/llvm/llvm-project/issues/57875
@@ -474,14 +484,22 @@ void CGHLSLRuntime::generateGlobalCtorDtorCalls() {
   for (auto &F : M.functions()) {
     if (!F.hasFnAttribute("hlsl.shader"))
       continue;
-    IRBuilder<> B(&F.getEntryBlock(), F.getEntryBlock().begin());
+    auto *Token = getConvergenceToken(F.getEntryBlock());
+    Instruction *IP = &*F.getEntryBlock().begin();
+    SmallVector<OperandBundleDef, 1> OB;
+    if (Token) {
+      llvm::Value *bundleArgs[] = {Token};
+      OB.emplace_back("convergencectrl", bundleArgs);
+      IP = Token->getNextNode();
+    }
+    IRBuilder<> B(IP);
     for (auto *Fn : CtorFns)
-      B.CreateCall(FunctionCallee(Fn));
+      B.CreateCall(FunctionCallee(Fn), {}, OB);
 
     // Insert global dtors before the terminator of the last instruction
     B.SetInsertPoint(F.back().getTerminator());
     for (auto *Fn : DtorFns)
-      B.CreateCall(FunctionCallee(Fn));
+      B.CreateCall(FunctionCallee(Fn), {}, OB);
   }
 
   // No need to keep global ctors/dtors for non-lib profile after call to
@@ -579,3 +597,18 @@ llvm::Function *CGHLSLRuntime::createResourceBindingInitFn() {
   Builder.CreateRetVoid();
   return InitResBindingsFunc;
 }
+
+llvm::Instruction *CGHLSLRuntime::getConvergenceToken(BasicBlock &BB) {
+  if (!CGM.shouldEmitConvergenceTokens())
+    return nullptr;
+
+  auto E = BB.end();
+  for (auto I = BB.begin(); I != E; ++I) {
+    auto *II = dyn_cast<llvm::IntrinsicInst>(&*I);
+    if (II && llvm::isConvergenceControlIntrinsic(II->getIntrinsicID())) {
+      return II;
+    }
+  }
+  llvm_unreachable("Convergence token should have been emitted.");
+  return nullptr;
+}

clang/lib/CodeGen/CGHLSLRuntime.h

@@ -143,6 +143,7 @@ class CGHLSLRuntime {
 
   bool needsResourceBindingInitFn();
   llvm::Function *createResourceBindingInitFn();
+  llvm::Instruction *getConvergenceToken(llvm::BasicBlock &BB);
 
 private:
   void addBufferResourceAnnotation(llvm::GlobalVariable *GV,

clang/test/CodeGenHLSL/convergence/entry.point.hlsl

@@ -0,0 +1,11 @@
+// RUN: %clang_cc1 -triple spirv-pc-vulkan-compute -finclude-default-header -disable-llvm-passes -emit-llvm -o - %s | FileCheck %s
+
+// CHECK-LABEL: define void @main()
+// CHECK-NEXT: entry:
+// CHECK-NEXT: [[token:%[0-9]+]] = call token @llvm.experimental.convergence.entry()
+// CHECK-NEXT: call spir_func void @_Z4mainv() [ "convergencectrl"(token [[token]]) ]
+
+[numthreads(1,1,1)]
+void main() {
+}
+