Feedback

Author: HendrikHuebner

clang/lib/CIR/CodeGen/CIRGenBuiltin.cpp

@@ -540,12 +540,20 @@ RValue CIRGenFunction::emitBuiltinExpr(const GlobalDecl &gd, unsigned builtinID,
   }
 
   // Now see if we can emit a target-specific builtin.
-  if (mlir::Value v = emitTargetBuiltinExpr(builtinID, e, returnValue)) {
+  std::optional<mlir::Value> valueOpt =
+      emitTargetBuiltinExpr(builtinID, e, returnValue);
+  if (valueOpt) {
+    // A builtin was emitted but had no return value.
+    if (*valueOpt == nullptr) {
+      return RValue::get(nullptr);
+    }
+
     switch (evalKind) {
     case cir::TEK_Scalar:
-      if (mlir::isa<cir::VoidType>(v.getType()))
+      if (mlir::isa<cir::VoidType>(valueOpt->getType())) {
         return RValue::get(nullptr);
-      return RValue::get(v);
+      }
+      return RValue::get(*valueOpt);
     case cir::TEK_Aggregate:
       cgm.errorNYI(e->getSourceRange(), "aggregate return value from builtin");
       return getUndefRValue(e->getType());
@@ -561,7 +569,7 @@ RValue CIRGenFunction::emitBuiltinExpr(const GlobalDecl &gd, unsigned builtinID,
   return getUndefRValue(e->getType());
 }
 
-static mlir::Value emitTargetArchBuiltinExpr(CIRGenFunction *cgf,
+static std::optional<mlir::Value> emitTargetArchBuiltinExpr(CIRGenFunction *cgf,
                                              unsigned builtinID,
                                              const CallExpr *e,
                                              ReturnValueSlot &returnValue,
@@ -616,7 +624,7 @@ static mlir::Value emitTargetArchBuiltinExpr(CIRGenFunction *cgf,
   }
 }
 
-mlir::Value
+std::optional<mlir::Value>
 CIRGenFunction::emitTargetBuiltinExpr(unsigned builtinID, const CallExpr *e,
                                       ReturnValueSlot &returnValue) {
   if (getContext().BuiltinInfo.isAuxBuiltinID(builtinID)) {

clang/lib/CIR/CodeGen/CIRGenBuiltinX86.cpp

@@ -48,6 +48,10 @@ static mlir::Value emitPrefetch(CIRGenFunction &cgf, unsigned builtinID,
   bool isWrite{};
   int locality{};
 
+  assert(builtinID == X86::BI_mm_prefetch ||
+         builtinID == X86::BI_m_prefetchw ||
+         builtinID == X86::BI_m_prefetch && "Expected prefetch builtin");
+
   if (builtinID == X86::BI_mm_prefetch) {
     int hint = getIntValueFromConstOp(ops[1]);
     isWrite = (hint >> 2) & 0x1;
@@ -61,7 +65,7 @@ static mlir::Value emitPrefetch(CIRGenFunction &cgf, unsigned builtinID,
   return {};
 }
 
-mlir::Value CIRGenFunction::emitX86BuiltinExpr(unsigned builtinID,
+std::optional<mlir::Value> CIRGenFunction::emitX86BuiltinExpr(unsigned builtinID,
                                                const CallExpr *e) {
   if (builtinID == Builtin::BI__builtin_cpu_is) {
     cgm.errorNYI(e->getSourceRange(), "__builtin_cpu_is");

clang/lib/CIR/CodeGen/CIRGenFunction.h

@@ -1746,7 +1746,7 @@ class CIRGenFunction : public CIRGenTypeCache {
                                      bool buildingTopLevelCase);
   mlir::LogicalResult emitSwitchStmt(const clang::SwitchStmt &s);
 
-  mlir::Value emitTargetBuiltinExpr(unsigned builtinID,
+  std::optional<mlir::Value> emitTargetBuiltinExpr(unsigned builtinID,
                                     const clang::CallExpr *e,
                                     ReturnValueSlot &returnValue);
 
@@ -1788,7 +1788,7 @@ class CIRGenFunction : public CIRGenTypeCache {
 
   mlir::LogicalResult emitWhileStmt(const clang::WhileStmt &s);
 
-  mlir::Value emitX86BuiltinExpr(unsigned builtinID, const CallExpr *e);
+  std::optional<mlir::Value> emitX86BuiltinExpr(unsigned builtinID, const CallExpr *e);
 
   /// Given an assignment `*lhs = rhs`, emit a test that checks if \p rhs is
   /// nonnull, if 1\p LHS is marked _Nonnull.

clang/test/CIR/CodeGen/X86/prefetchw-builtin.c

@@ -20,17 +20,17 @@ void test_m_prefetch_w(void *p) {
   // LLVM-LABEL: test_m_prefetch_w
   // OGCG-LABEL: test_m_prefetch_w
   return _m_prefetchw(p);
-  // CIR: cir.prefetch write locality(0) %{{.*}} : !cir.ptr<!void>
-  // LLVM: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 0, i32 1)
-  // OGCG: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 0, i32 1)
+  // CIR: cir.prefetch write locality(3) %{{.*}} : !cir.ptr<!void>
+  // LLVM: call void @llvm.prefetch.p0(ptr {{.*}}, i32 1, i32 3, i32 1)
+  // OGCG: call void @llvm.prefetch.p0(ptr {{.*}}, i32 1, i32 3, i32 1)
 }
 
 void test_m_prefetch(void *p) {
   // CIR-LABEL: test_m_prefetch
   // LLVM-LABEL: test_m_prefetch
   // OGCG-LABEL: test_m_prefetch
   return _m_prefetch(p);
-  // CIR: cir.prefetch read locality(0) %{{.*}} : !cir.ptr<!void>
-  // LLVM: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 0, i32 1)
-  // OGCG: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 0, i32 1)
+  // CIR: cir.prefetch read locality(3) %{{.*}} : !cir.ptr<!void>
+  // LLVM: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 3, i32 1)
+  // OGCG: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 3, i32 1)
 }

clang/test/CIR/CodeGen/X86/sse-builtins.c

@@ -44,3 +44,12 @@ void test_mm_prefetch_local(char const* p) {
   // LLVM: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 3, i32 1)
   // OGCG: call void @llvm.prefetch.p0(ptr {{.*}}, i32 0, i32 3, i32 1)
 }
+
+void test_mm_prefetch_write(char const* p) {
+  // CIR-LABEL: test_mm_prefetch_write
+  // LLVM-LABEL: test_mm_prefetch_write
+  _mm_prefetch(p, 7);
+  // CIR: cir.prefetch write locality(3) %{{.*}} : !cir.ptr<!void>
+  // LLVM: call void @llvm.prefetch.p0(ptr {{.*}}, i32 1, i32 3, i32 1)
+  // OGCG: call void @llvm.prefetch.p0(ptr {{.*}}, i32 1, i32 3, i32 1)
+}