[MachineOutliner] Don't save/restore LR for tail calls.

[efriedma]

Summary

The cost computation correctly assumes we do this, but the actual lowering was wrong.

[paquette]

Thanks for finding this. LGTM!

lib/Target/AArch64/AArch64InstrInfo.cpp

for (MachineInstr &MI : MBB) {
     int64_t NewImm = (Offset + 16) / Scale;
     StackOffsetOperand.setImm(NewImm);
   }
 }
 
 void AArch64InstrInfo::insertOutlinerEpilogue(
     MachineBasicBlock &MBB, MachineFunction &MF,
     const MachineOutlinerInfo &MInfo) const {
-
   // Is there a call in the outlined range?
-  if (std::any_of(MBB.instr_begin(), MBB.instr_end(),
-                  [](MachineInstr &MI) { return MI.isCall(); })) {
+  auto IsNonTailCall = [](MachineInstr &MI) {
+    return MI.isCall() && !MI.isReturn();
+  };
+  if (std::any_of(MBB.instr_begin(), MBB.instr_end(), IsNonTailCall)) {
     // Fix up the instructions in the range, since we're going to modify the
     // stack.
     fixupPostOutline(MBB);
 
     // LR has to be a live in so that we can save it.
     MBB.addLiveIn(AArch64::LR);
 
     MachineBasicBlock::iterator It = MBB.begin();

test/CodeGen/AArch64/machine-outliner-tail.ll

+; RUN: llc -verify-machineinstrs -enable-machine-outliner -mtriple=aarch64-linux-gnu < %s | FileCheck %s
+
+; CHECK: OUTLINED_FUNCTION_0:
+; CHECK:      orr     w0, wzr, #0x1
+; CHECK-NEXT: orr     w1, wzr, #0x2
+; CHECK-NEXT: orr     w2, wzr, #0x3
+; CHECK-NEXT: orr     w3, wzr, #0x4
+; CHECK-NEXT: b       z
+
+define void @a() {
+entry:
+  tail call void @z(i32 1, i32 2, i32 3, i32 4)
+  ret void
+}
+
+declare void @z(i32, i32, i32, i32)
+
+define dso_local void @b(i32* nocapture readnone %p) {
+entry:
+  tail call void @z(i32 1, i32 2, i32 3, i32 4)
+  ret void
+}