[clang] [clang] VectorExprEvaluator::VisitCallExpr - add constant folding for X86 pslldqi/psrldqi intrinsics (PR #157403)

Simon Pilgrim via cfe-commits cfe-commits at lists.llvm.org
Wed Sep 24 05:12:16 PDT 2025


================
@@ -276,9 +277,8 @@ let Features = "sse2", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] i
   def psllw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
   def pslld128 : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>)">;
   def psllq128 : X86Builtin<"_Vector<2, long long int>(_Vector<2, long long int>, _Vector<2, long long int>)">;
-  def pmaddwd128 : X86Builtin<"_Vector<4, int>(_Vector<8, short>, _Vector<8, short>)">;
-  def pslldqi128_byteshift : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Constant int)">;
-  def psrldqi128_byteshift : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Constant int)">;
+  def pmaddwd128
+      : X86Builtin<"_Vector<4, int>(_Vector<8, short>, _Vector<8, short>)">;
----------------
RKSimon wrote:

(style) revert changes from code unrelated to the patch (this includes formatting)

https://github.com/llvm/llvm-project/pull/157403


More information about the cfe-commits mailing list