[llvm] [x64][win] Add compiler support for x64 import call optimization (equivalent to MSVC /d2guardretpoline) (PR #126631)

[Eli Friedman via llvm-commits]

================
@@ -2469,7 +2535,49 @@ void X86AsmPrinter::emitInstruction(const MachineInstr *MI) {
   case X86::CALL64pcrel32:
     if (IndCSPrefix && MI->hasRegisterImplicitUseOperand(X86::R11))
       EmitAndCountInstruction(MCInstBuilder(X86::CS_PREFIX));
+
+    if (EnableImportCallOptimization && isImportedFunction(MI->getOperand(0))) {
+      emitLabelAndRecordForImportCallOptimization(
+          IMAGE_RETPOLINE_AMD64_IMPORT_CALL);
+
+      MCInst TmpInst;
+      MCInstLowering.Lower(MI, TmpInst);
+
+      // For Import Call Optimization to work, we need a the call instruction
+      // with a rex prefix, and a 5-byte nop after the call instruction.
+      EmitAndCountInstruction(MCInstBuilder(X86::REX64_PREFIX));
+      emitCallInstruction(TmpInst);
+      emitNop(*OutStreamer, 5, Subtarget);
+      return;
+    }
+
     break;
+  case X86::CALL64r:
+    if (EnableImportCallOptimization) {
+      MCInst TmpInst;
+      MCInstLowering.Lower(MI, TmpInst);
+
+      // Import call optimization requires all indirect calls go via RAX.
+      ensureRaxUsedForOperand(TmpInst);
+
+      emitLabelAndRecordForImportCallOptimization(
+          IMAGE_RETPOLINE_AMD64_INDIR_CALL);
+      emitCallInstruction(TmpInst);
+
+      // For Import Call Optimization to work, we a 3-byte nop after the call
----------------
efriedma-quic wrote:

```suggestion
      // For Import Call Optimization to work, we need a 3-byte nop after the call
```

https://github.com/llvm/llvm-project/pull/126631