[llvm] [RISCV] Implement Statepoint and Patchpoint lowering to call instructions (PR #77337)

[Pengcheng Wang via llvm-commits]

================
@@ -152,8 +153,40 @@ void RISCVAsmPrinter::LowerPATCHPOINT(MCStreamer &OutStreamer, StackMaps &SM,
 
   PatchPointOpers Opers(&MI);
 
+  const MachineOperand &CalleeMO = Opers.getCallTarget();
   unsigned EncodedBytes = 0;
 
+  if (CalleeMO.isImm()) {
+    uint64_t CallTarget = CalleeMO.getImm();
+    if (CallTarget) {
+      assert((CallTarget & 0xFFFF'FFFF'FFFF) == CallTarget &&
+             "High 16 bits of call target should be zero.");
+      // Materialize the jump address:
+      SmallVector<MCInst, 8> Seq;
+      RISCVMatInt::generateMCInstSeq(CallTarget, *STI, RISCV::X1, Seq);
+      for (MCInst &Inst : Seq) {
+        MCInst CInst;
+        bool compressed = RISCVRVC::compress(CInst, Inst, *STI);
+        EmitToStreamer(OutStreamer, Inst);
----------------
wangpc-pp wrote:

I have created a PR (#88120) so that we don't have to call `RISCVRVC::compress` twice.
And after that, you can write:
```
bool Compressed = EmitToStreamer(OutStreamer, Inst);
EncodedBytes += Compressed ? 2 : 4;
```

https://github.com/llvm/llvm-project/pull/77337