[llvm] [RISCV] Add MachineCombiner to fold (sh3add Z, (add X, (slli Y, 6))) -> (sh3add (sh3add Y, Z), X). (PR #87884)

[Pengcheng Wang via llvm-commits]

================
@@ -1918,6 +1999,78 @@ static void combineFPFusedMultiply(MachineInstr &Root, MachineInstr &Prev,
   DelInstrs.push_back(&Root);
 }
 
+// Combine (sh3add Z, (add X, (slli Y, 5))) to (sh3add (sh2add Y, Z), X).
+static void
+genShXAddAddShift(MachineInstr &Root, unsigned AddOpIdx,
+                  SmallVectorImpl<MachineInstr *> &InsInstrs,
+                  SmallVectorImpl<MachineInstr *> &DelInstrs,
+                  DenseMap<unsigned, unsigned> &InstrIdxForVirtReg) {
+  MachineFunction *MF = Root.getMF();
+  MachineRegisterInfo &MRI = MF->getRegInfo();
+  const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo();
+
+  unsigned OuterShiftAmt;
+  switch (Root.getOpcode()) {
----------------
wangpc-pp wrote:

Maybe we can add a helper to get shift amount.

https://github.com/llvm/llvm-project/pull/87884