x86: FARJMP/FARCALL support Nasm syntax

llvm/lib/Target/X86/AsmParser/X86AsmParser.cpp

@@ -2424,6 +2424,22 @@ bool X86AsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
 {
   MCAsmParser &Parser = getParser();
   InstInfo = &Info;
+
+  // Nasm accepts JMP/CALL, but not LJMP/LCALL
+  if (KsSyntax == KS_OPT_SYNTAX_NASM) {
+      if (Name == "jmp" || Name == "call") {
+          AsmToken Tokens[3];
+          MutableArrayRef<AsmToken> Buf(Tokens);
+          size_t count = getLexer().peekTokens(Buf);
+          if (count == 3 && Tokens[0].getString() == ":") {
+              if (Name == "jmp")
+                  Name = "ljmp";
+              else if (Name == "call")
+                  Name = "lcall";
+          }
+      }
+  }
+
   StringRef PatchedName = Name;
 
   // FIXME: Hack to recognize setneb as setne.
@@ -2584,18 +2600,18 @@ bool X86AsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
          Parser.eatToEndOfStatement();
          return true;
       }
-      // for LJMP/LCALL, check for ':'. otherwise, check for comma and eat it
-      if (Name.startswith("ljmp") || Name.startswith("lcall")) {
-          if (getLexer().is(AsmToken::Colon)) {
-              //Parser.Lex();
+
+      if (getLexer().is(AsmToken::Colon)) {
+          // for LJMP/LCALL, check for ':'. otherwise, check for comma and eat it
+          if (Name.startswith("ljmp") || Name.startswith("lcall"))
               Operands.push_back(X86Operand::CreateToken(":", consumeToken()));
-          } else
+          else
               break;
       } else if (getLexer().is(AsmToken::Comma))
-        Parser.Lex();
+          Parser.Lex();
       else
-        break;
-     }
+          break;
+    }
 
     if (getLexer().isNot(AsmToken::EndOfStatement) && getLexer().isNot(AsmToken::Eof)) {
       //return ErrorAndEatStatement(getLexer().getLoc(),
@@ -2775,7 +2791,8 @@ bool X86AsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
   return false;
 }
 
-bool X86AsmParser::processInstruction(MCInst &Inst, const OperandVector &Ops) {
+bool X86AsmParser::processInstruction(MCInst &Inst, const OperandVector &Ops)
+{
   switch (Inst.getOpcode()) {
   default: return false;
   case X86::VMOVZPQILo2PQIrr:

suite/regress/x86_jmp_0x33_08.py

@@ -15,8 +15,13 @@ class TestX86Intel(regress.RegressTest):
         ks = Ks(KS_ARCH_X86, KS_MODE_32)
         # Assemble with zero addr
         try:
+            # Keystone (LLVM) syntax
             encoding, count = ks.asm("ljmp 0x33:08", 0)
             self.assertEqual(encoding, [ 0xea, 0x08, 0x00, 0x30, 0x00 ])
+            # NASM syntax
+            ks.syntax = KS_OPT_SYNTAX_NASM
+            encoding2, count = ks.asm("jmp 0x33:08", 0)
+            self.assertEqual(encoding, encoding2)
         except KsError as e:
             pass