1 //===-- X86AsmLexer.cpp - Tokenize X86 assembly to AsmTokens --------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 #include "MCTargetDesc/X86BaseInfo.h"
11 #include "llvm/MC/MCAsmInfo.h"
12 #include "llvm/MC/MCParser/MCAsmLexer.h"
13 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
14 #include "llvm/MC/MCTargetAsmLexer.h"
15 #include "llvm/Support/TargetRegistry.h"
16 #include "llvm/ADT/SmallVector.h"
22 class X86AsmLexer : public MCTargetAsmLexer {
23 const MCAsmInfo &AsmInfo;
25 bool tentativeIsValid;
26 AsmToken tentativeToken;
28 const AsmToken &lexTentative() {
29 tentativeToken = getLexer()->Lex();
30 tentativeIsValid = true;
31 return tentativeToken;
34 const AsmToken &lexDefinite() {
35 if (tentativeIsValid) {
36 tentativeIsValid = false;
37 return tentativeToken;
39 return getLexer()->Lex();
42 AsmToken LexTokenATT();
43 AsmToken LexTokenIntel();
47 SetError(SMLoc(), "No MCAsmLexer installed");
48 return AsmToken(AsmToken::Error, "", 0);
51 switch (AsmInfo.getAssemblerDialect()) {
53 SetError(SMLoc(), "Unhandled dialect");
54 return AsmToken(AsmToken::Error, "", 0);
58 return LexTokenIntel();
62 X86AsmLexer(const Target &T, const MCRegisterInfo &MRI, const MCAsmInfo &MAI)
63 : MCTargetAsmLexer(T), AsmInfo(MAI), tentativeIsValid(false) {
67 } // end anonymous namespace
69 #define GET_REGISTER_MATCHER
70 #include "X86GenAsmMatcher.inc"
72 AsmToken X86AsmLexer::LexTokenATT() {
73 AsmToken lexedToken = lexDefinite();
75 switch (lexedToken.getKind()) {
79 SetError(Lexer->getErrLoc(), Lexer->getErr());
82 case AsmToken::Percent: {
83 const AsmToken &nextToken = lexTentative();
84 if (nextToken.getKind() != AsmToken::Identifier)
88 if (unsigned regID = MatchRegisterName(nextToken.getString())) {
91 // FIXME: This is completely wrong when there is a space or other
92 // punctuation between the % and the register name.
93 StringRef regStr(lexedToken.getString().data(),
94 lexedToken.getString().size() +
95 nextToken.getString().size());
97 return AsmToken(AsmToken::Register, regStr,
98 static_cast<int64_t>(regID));
101 // Match register name failed. If this is "db[0-7]", match it as an alias
103 if (nextToken.getString().size() == 3 &&
104 nextToken.getString().startswith("db")) {
106 switch (nextToken.getString()[2]) {
107 case '0': RegNo = X86::DR0; break;
108 case '1': RegNo = X86::DR1; break;
109 case '2': RegNo = X86::DR2; break;
110 case '3': RegNo = X86::DR3; break;
111 case '4': RegNo = X86::DR4; break;
112 case '5': RegNo = X86::DR5; break;
113 case '6': RegNo = X86::DR6; break;
114 case '7': RegNo = X86::DR7; break;
120 // FIXME: This is completely wrong when there is a space or other
121 // punctuation between the % and the register name.
122 StringRef regStr(lexedToken.getString().data(),
123 lexedToken.getString().size() +
124 nextToken.getString().size());
125 return AsmToken(AsmToken::Register, regStr,
126 static_cast<int64_t>(RegNo));
136 AsmToken X86AsmLexer::LexTokenIntel() {
137 const AsmToken &lexedToken = lexDefinite();
139 switch(lexedToken.getKind()) {
142 case AsmToken::Error:
143 SetError(Lexer->getErrLoc(), Lexer->getErr());
145 case AsmToken::Identifier: {
146 unsigned regID = MatchRegisterName(lexedToken.getString().lower());
149 return AsmToken(AsmToken::Register,
150 lexedToken.getString(),
151 static_cast<int64_t>(regID));
157 extern "C" void LLVMInitializeX86AsmLexer() {
158 RegisterMCAsmLexer<X86AsmLexer> X(TheX86_32Target);
159 RegisterMCAsmLexer<X86AsmLexer> Y(TheX86_64Target);