[llvm] [InstCombine] Do not simplify lshr/shl arg if it is part of fshl rotate pattern. (PR #66115)

[Simon Pilgrim via llvm-commits]

================
@@ -611,6 +611,23 @@ Value *InstCombinerImpl::SimplifyDemandedUseBits(Value *V, APInt DemandedMask,
                                                     DemandedMask, Known))
             return R;
 
+      // Do not simplify if lshr is part of fshl rotate pattern
+      if (I->hasOneUser()) {
+        auto *Op = I->user_back();
+        if (Op->getOpcode() == BinaryOperator::Or) {
+          const APInt *LShrAmt;
+          Value *LShrVal;
+          auto *Operand =
+              Op->getOperand(0) == I ? Op->getOperand(1) : Op->getOperand(0);
+          if (match(Operand,
+                    m_OneUse(m_LShr(m_Value(LShrVal), m_APInt(LShrAmt)))))
----------------
RKSimon wrote:

Use m_Specific(I->getOperand(0))?

https://github.com/llvm/llvm-project/pull/66115