[llvm] Re apply 130577 narrow math for and operand (PR #133896)

[Matt Arsenault via llvm-commits]

================
@@ -1561,6 +1561,87 @@ void AMDGPUCodeGenPrepareImpl::expandDivRem64(BinaryOperator &I) const {
   llvm_unreachable("not a division");
 }
 
+Type *findSmallestLegalBits(Instruction *I, int OrigBit, int MaxBitsNeeded,
+                            const TargetLowering *TLI, const DataLayout &DL) {
+  if (MaxBitsNeeded >= OrigBit)
+    return nullptr;
+
+  Type *NewType = I->getType()->getWithNewBitWidth(MaxBitsNeeded);
+  while (OrigBit > MaxBitsNeeded) {
+    if (TLI->isOperationLegalOrCustom(
+            TLI->InstructionOpcodeToISD(I->getOpcode()),
+            TLI->getValueType(DL, NewType, true)))
+      return NewType;
+
+    MaxBitsNeeded *= 2;
+    NewType = I->getType()->getWithNewBitWidth(MaxBitsNeeded);
+  }
+  return nullptr;
+}
+
+static bool tryNarrowMathIfNoOverflow(Instruction *I, const TargetLowering *TLI,
+                                      const TargetTransformInfo &TTI,
----------------
arsenm wrote:

Use target specific subclass, or move this back to CGP 

https://github.com/llvm/llvm-project/pull/133896