[llvm] [InstCombine] Fold xor of bittests into bittest of xor'd value (PR #125676)

[Yingwei Zheng via llvm-commits]

================
@@ -4200,6 +4197,21 @@ Value *InstCombinerImpl::foldXorOfICmps(ICmpInst *LHS, ICmpInst *RHS,
           }
         }
     }
+
+    // Fold (icmp eq/ne (X & Pow2), 0) ^ (icmp eq/ne (Y & Pow2), 0) into
+    // (icmp eq/ne ((X ^ Y) & Pow2), 0)
+    Value *X, *Y;
+    const APInt *Mask;
+    if (ICmpInst::isEquality(PredL) && ICmpInst::isEquality(PredR) &&
+        LC->isZero() && RC->isZero() && LHS->hasOneUse() && RHS->hasOneUse() &&
+        match(LHS0, m_And(m_Value(X), m_Power2(Mask))) &&
+        match(RHS0, m_And(m_Value(Y), m_SpecificInt(*Mask)))) {
----------------
dtcxzyw wrote:

I will try it after landing this patch.

https://github.com/llvm/llvm-project/pull/125676