| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 1 | //===-- X86AsmLexer.cpp - Tokenize X86 assembly to AsmTokens --------------===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | |
| Evan Cheng | 1142444 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 10 | #include "MCTargetDesc/X86BaseInfo.h" |
| Sean Callanan | 6654931 | 2010-01-25 21:59:20 +0000 | [diff] [blame] | 11 | #include "llvm/MC/MCAsmInfo.h" |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 12 | #include "llvm/MC/MCParser/MCAsmLexer.h" |
| 13 | #include "llvm/MC/MCParser/MCParsedAsmOperand.h" |
| Evan Cheng | 1142444 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 14 | #include "llvm/MC/MCTargetAsmLexer.h" |
| Evan Cheng | 2bb4035 | 2011-08-24 18:08:43 +0000 | [diff] [blame] | 15 | #include "llvm/Support/TargetRegistry.h" |
| Evan Cheng | f2596bc | 2011-07-23 00:45:41 +0000 | [diff] [blame] | 16 | #include "llvm/ADT/SmallVector.h" |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 17 | |
| 18 | using namespace llvm; |
| 19 | |
| 20 | namespace { |
| 21 | |
| Evan Cheng | 1142444 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 22 | class X86AsmLexer : public MCTargetAsmLexer { |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 23 | const MCAsmInfo &AsmInfo; |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 24 | |
| 25 | bool tentativeIsValid; |
| 26 | AsmToken tentativeToken; |
| 27 | |
| 28 | const AsmToken &lexTentative() { |
| Sean Callanan | 9aeccad | 2010-01-31 02:28:18 +0000 | [diff] [blame] | 29 | tentativeToken = getLexer()->Lex(); |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 30 | tentativeIsValid = true; |
| 31 | return tentativeToken; |
| 32 | } |
| 33 | |
| 34 | const AsmToken &lexDefinite() { |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 35 | if (tentativeIsValid) { |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 36 | tentativeIsValid = false; |
| 37 | return tentativeToken; |
| 38 | } |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 39 | return getLexer()->Lex(); |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 40 | } |
| Sean Callanan | 6654931 | 2010-01-25 21:59:20 +0000 | [diff] [blame] | 41 | |
| 42 | AsmToken LexTokenATT(); |
| 43 | AsmToken LexTokenIntel(); |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 44 | protected: |
| Sean Callanan | 6654931 | 2010-01-25 21:59:20 +0000 | [diff] [blame] | 45 | AsmToken LexToken() { |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 46 | if (!Lexer) { |
| 47 | SetError(SMLoc(), "No MCAsmLexer installed"); |
| 48 | return AsmToken(AsmToken::Error, "", 0); |
| 49 | } |
| 50 | |
| Sean Callanan | 6654931 | 2010-01-25 21:59:20 +0000 | [diff] [blame] | 51 | switch (AsmInfo.getAssemblerDialect()) { |
| 52 | default: |
| 53 | SetError(SMLoc(), "Unhandled dialect"); |
| 54 | return AsmToken(AsmToken::Error, "", 0); |
| 55 | case 0: |
| 56 | return LexTokenATT(); |
| 57 | case 1: |
| 58 | return LexTokenIntel(); |
| 59 | } |
| 60 | } |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 61 | public: |
| Evan Cheng | f60768a | 2011-07-25 20:53:02 +0000 | [diff] [blame] | 62 | X86AsmLexer(const Target &T, const MCRegisterInfo &MRI, const MCAsmInfo &MAI) |
| Evan Cheng | 1142444 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 63 | : MCTargetAsmLexer(T), AsmInfo(MAI), tentativeIsValid(false) { |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 64 | } |
| 65 | }; |
| 66 | |
| Chris Lattner | 3e4582a | 2010-09-06 19:11:01 +0000 | [diff] [blame] | 67 | } // end anonymous namespace |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 68 | |
| Chris Lattner | 3e4582a | 2010-09-06 19:11:01 +0000 | [diff] [blame] | 69 | #define GET_REGISTER_MATCHER |
| 70 | #include "X86GenAsmMatcher.inc" |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 71 | |
| Sean Callanan | 6654931 | 2010-01-25 21:59:20 +0000 | [diff] [blame] | 72 | AsmToken X86AsmLexer::LexTokenATT() { |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 73 | AsmToken lexedToken = lexDefinite(); |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 74 | |
| 75 | switch (lexedToken.getKind()) { |
| 76 | default: |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 77 | return lexedToken; |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 78 | case AsmToken::Error: |
| 79 | SetError(Lexer->getErrLoc(), Lexer->getErr()); |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 80 | return lexedToken; |
| 81 | |
| 82 | case AsmToken::Percent: { |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 83 | const AsmToken &nextToken = lexTentative(); |
| Chris Lattner | 056fd06 | 2010-06-24 07:16:25 +0000 | [diff] [blame] | 84 | if (nextToken.getKind() != AsmToken::Identifier) |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 85 | return lexedToken; |
| Chris Lattner | 056fd06 | 2010-06-24 07:16:25 +0000 | [diff] [blame] | 86 | |
| 87 | |
| 88 | if (unsigned regID = MatchRegisterName(nextToken.getString())) { |
| 89 | lexDefinite(); |
| 90 | |
| Chris Lattner | 8048662 | 2010-06-24 07:29:18 +0000 | [diff] [blame] | 91 | // FIXME: This is completely wrong when there is a space or other |
| 92 | // punctuation between the % and the register name. |
| Chris Lattner | 056fd06 | 2010-06-24 07:16:25 +0000 | [diff] [blame] | 93 | StringRef regStr(lexedToken.getString().data(), |
| 94 | lexedToken.getString().size() + |
| 95 | nextToken.getString().size()); |
| 96 | |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 97 | return AsmToken(AsmToken::Register, regStr, |
| Chris Lattner | 056fd06 | 2010-06-24 07:16:25 +0000 | [diff] [blame] | 98 | static_cast<int64_t>(regID)); |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 99 | } |
| Chris Lattner | 056fd06 | 2010-06-24 07:16:25 +0000 | [diff] [blame] | 100 | |
| Chris Lattner | 8048662 | 2010-06-24 07:29:18 +0000 | [diff] [blame] | 101 | // Match register name failed. If this is "db[0-7]", match it as an alias |
| 102 | // for dr[0-7]. |
| 103 | if (nextToken.getString().size() == 3 && |
| 104 | nextToken.getString().startswith("db")) { |
| 105 | int RegNo = -1; |
| 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; |
| 115 | } |
| 116 | |
| 117 | if (RegNo != -1) { |
| 118 | lexDefinite(); |
| 119 | |
| 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)); |
| 127 | } |
| 128 | } |
| 129 | |
| 130 | |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 131 | return lexedToken; |
| Sean Callanan | ad857fc | 2010-01-26 00:08:25 +0000 | [diff] [blame] | 132 | } |
| 133 | } |
| Sean Callanan | 6654931 | 2010-01-25 21:59:20 +0000 | [diff] [blame] | 134 | } |
| 135 | |
| 136 | AsmToken X86AsmLexer::LexTokenIntel() { |
| Sean Callanan | eb182e9 | 2010-01-26 01:00:10 +0000 | [diff] [blame] | 137 | const AsmToken &lexedToken = lexDefinite(); |
| 138 | |
| 139 | switch(lexedToken.getKind()) { |
| 140 | default: |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 141 | return lexedToken; |
| Sean Callanan | eb182e9 | 2010-01-26 01:00:10 +0000 | [diff] [blame] | 142 | case AsmToken::Error: |
| 143 | SetError(Lexer->getErrLoc(), Lexer->getErr()); |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 144 | return lexedToken; |
| 145 | case AsmToken::Identifier: { |
| Benjamin Kramer | 20baffb | 2011-11-06 20:37:06 +0000 | [diff] [blame] | 146 | unsigned regID = MatchRegisterName(lexedToken.getString().lower()); |
| Sean Callanan | eb182e9 | 2010-01-26 01:00:10 +0000 | [diff] [blame] | 147 | |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 148 | if (regID) |
| Sean Callanan | eb182e9 | 2010-01-26 01:00:10 +0000 | [diff] [blame] | 149 | return AsmToken(AsmToken::Register, |
| 150 | lexedToken.getString(), |
| 151 | static_cast<int64_t>(regID)); |
| Chris Lattner | c4e8430 | 2010-06-24 07:18:14 +0000 | [diff] [blame] | 152 | return lexedToken; |
| Sean Callanan | eb182e9 | 2010-01-26 01:00:10 +0000 | [diff] [blame] | 153 | } |
| 154 | } |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 155 | } |
| 156 | |
| 157 | extern "C" void LLVMInitializeX86AsmLexer() { |
| Evan Cheng | 1142444 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 158 | RegisterMCAsmLexer<X86AsmLexer> X(TheX86_32Target); |
| 159 | RegisterMCAsmLexer<X86AsmLexer> Y(TheX86_64Target); |
| Sean Callanan | 5051cb8 | 2010-01-23 02:43:15 +0000 | [diff] [blame] | 160 | } |