[clang] [llvm] Add support for flag output operand "=@cc" for SystemZ. (PR #125970)

[Ulrich Weigand via llvm-commits]

================
@@ -8132,7 +8722,9 @@ SDValue SystemZTargetLowering::combineBR_CCMASK(
   SDValue Chain = N->getOperand(0);
   SDValue CCReg = N->getOperand(4);
 
-  if (combineCCMask(CCReg, CCValidVal, CCMaskVal))
+  // combineCCIPMMask tries to combine srl/ipm sequence for flag output operand.
+  if (combineCCIPMMask(CCReg, CCValidVal, CCMaskVal) ||
+      combineCCMask(CCReg, CCValidVal, CCMaskVal))
----------------
uweigand wrote:

Why do we need the separate routine here?  `combineCCMask` already attempts to handle cases involving IPM - those that result from intrinsics that set CC.  Note that in general, we should apply the exact same set of optimizations whether the CC value was generated by an intrinsic or by an inline asm.

https://github.com/llvm/llvm-project/pull/125970