[llvm] [RISCV] Implement RISCVInstrInfo::getMemOperandsWithOffsetWidth (PR #73681)

[Craig Topper via llvm-commits]

================
@@ -2193,11 +2193,51 @@ MachineInstr *RISCVInstrInfo::emitLdStWithAddr(MachineInstr &MemI,
       .setMIFlags(MemI.getFlags());
 }
 
+bool RISCVInstrInfo::getMemOperandsWithOffsetWidth(
+    const MachineInstr &LdSt, SmallVectorImpl<const MachineOperand *> &BaseOps,
+    int64_t &Offset, bool &OffsetIsScalable, unsigned &Width,
+    const TargetRegisterInfo *TRI) const {
+  if (!LdSt.mayLoadOrStore())
+    return false;
+
+  // Conservatively, only handle scalar loads/stores for now.
+  switch (LdSt.getOpcode()) {
+  case RISCV::LB:
+  case RISCV::LBU:
+  case RISCV::SB:
+  case RISCV::LH:
+  case RISCV::LHU:
+  case RISCV::FLH:
+  case RISCV::SH:
+  case RISCV::FSH:
+  case RISCV::LW:
+  case RISCV::LWU:
+  case RISCV::FLW:
+  case RISCV::SW:
+  case RISCV::FSW:
+  case RISCV::LD:
+  case RISCV::FLD:
+  case RISCV::SD:
+  case RISCV::FSD:
+    break;
+  default:
+    return false;
+  }
+  const MachineOperand *BaseOp;
+  OffsetIsScalable = false;
+  if (!getMemOperandWithOffsetWidth(LdSt, BaseOp, Offset, Width, TRI))
+    return false;
+  BaseOps.push_back(BaseOp);
+  return true;
+}
+
 // Set BaseReg (the base register operand), Offset (the byte offset being
 // accessed) and the access Width of the passed instruction that reads/writes
 // memory. Returns false if the instruction does not read/write memory or the
 // BaseReg/Offset/Width can't be determined. Is not guaranteed to always
 // recognise base operands and offsets in all cases.
+// TODO: Add an IsScalable bool ref argument (like the equivalent AArch64
+// function) and set it as appropriate.
 bool RISCVInstrInfo::getMemOperandWithOffsetWidth(
----------------
topperc wrote:

Should we fold this into the new function and get rid of this signature? The default implementation of this signature calls the SmallVector version.

https://github.com/llvm/llvm-project/pull/73681