blob: a58ab3e8804a40c47677d9148fcf3096e7db81df [file] [log] [blame]
Kevin Enderbyca9c42c2009-09-15 00:27:25 +00001//===-- ARMAsmParser.cpp - Parse ARM assembly to MCInst instructions ------===//
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
10#include "ARM.h"
11#include "llvm/ADT/SmallVector.h"
12#include "llvm/ADT/Twine.h"
13#include "llvm/MC/MCAsmLexer.h"
14#include "llvm/MC/MCAsmParser.h"
15#include "llvm/MC/MCStreamer.h"
16#include "llvm/MC/MCExpr.h"
17#include "llvm/MC/MCInst.h"
18#include "llvm/Support/SourceMgr.h"
19#include "llvm/Target/TargetRegistry.h"
20#include "llvm/Target/TargetAsmParser.h"
21using namespace llvm;
22
23namespace {
24struct ARMOperand;
25
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000026// The shift types for register controlled shifts in arm memory addressing
27enum ShiftType {
28 Lsl,
29 Lsr,
30 Asr,
31 Ror,
32 Rrx
33};
34
Kevin Enderbyca9c42c2009-09-15 00:27:25 +000035class ARMAsmParser : public TargetAsmParser {
36 MCAsmParser &Parser;
37
38private:
39 MCAsmParser &getParser() const { return Parser; }
40
41 MCAsmLexer &getLexer() const { return Parser.getLexer(); }
42
43 void Warning(SMLoc L, const Twine &Msg) { Parser.Warning(L, Msg); }
44
45 bool Error(SMLoc L, const Twine &Msg) { return Parser.Error(L, Msg); }
46
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000047 bool ParseRegister(ARMOperand &Op);
48
Kevin Enderbyd7894f12009-10-09 21:12:28 +000049 bool ParseRegisterList(ARMOperand &Op);
50
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000051 bool ParseMemory(ARMOperand &Op);
52
Kevin Enderby9d369622009-10-12 22:39:54 +000053 bool ParseShift(enum ShiftType *St, const MCExpr *&ShiftAmount);
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000054
55 bool ParseOperand(ARMOperand &Op);
56
Kevin Enderbyca9c42c2009-09-15 00:27:25 +000057 bool ParseDirectiveWord(unsigned Size, SMLoc L);
58
Kevin Enderby515d5092009-10-15 20:48:48 +000059 bool ParseDirectiveThumb(SMLoc L);
60
61 bool ParseDirectiveThumbFunc(SMLoc L);
62
63 bool ParseDirectiveCode(SMLoc L);
64
65 bool ParseDirectiveSyntax(SMLoc L);
66
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000067 // TODO - For now hacked versions of the next two are in here in this file to
68 // allow some parser testing until the table gen versions are implemented.
69
70 /// @name Auto-generated Match Functions
71 /// {
72 bool MatchInstruction(SmallVectorImpl<ARMOperand> &Operands,
73 MCInst &Inst);
74
Kevin Enderbyd7894f12009-10-09 21:12:28 +000075 /// MatchRegisterName - Match the given string to a register name and return
76 /// its register number, or -1 if there is no match. To allow return values
77 /// to be used directly in register lists, arm registers have values between
78 /// 0 and 15.
79 int MatchRegisterName(const StringRef &Name);
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000080
81 /// }
82
83
Kevin Enderbyca9c42c2009-09-15 00:27:25 +000084public:
85 ARMAsmParser(const Target &T, MCAsmParser &_Parser)
86 : TargetAsmParser(T), Parser(_Parser) {}
87
88 virtual bool ParseInstruction(const StringRef &Name, MCInst &Inst);
89
90 virtual bool ParseDirective(AsmToken DirectiveID);
91};
92
93} // end anonymous namespace
94
Kevin Enderbya7ba3a82009-10-06 22:26:42 +000095namespace {
96
97/// ARMOperand - Instances of this class represent a parsed ARM machine
98/// instruction.
99struct ARMOperand {
100 enum {
101 Token,
102 Register,
Kevin Enderbycfe07242009-10-13 22:19:02 +0000103 Immediate,
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000104 Memory
105 } Kind;
106
107
108 union {
109 struct {
110 const char *Data;
111 unsigned Length;
112 } Tok;
113
114 struct {
115 unsigned RegNum;
Kevin Enderby99e6d4e2009-10-07 18:01:35 +0000116 bool Writeback;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000117 } Reg;
118
Kevin Enderbycfe07242009-10-13 22:19:02 +0000119 struct {
120 const MCExpr *Val;
121 } Imm;
122
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000123 // This is for all forms of ARM address expressions
124 struct {
125 unsigned BaseRegNum;
126 bool OffsetIsReg;
127 const MCExpr *Offset; // used when OffsetIsReg is false
128 unsigned OffsetRegNum; // used when OffsetIsReg is true
129 bool OffsetRegShifted; // only used when OffsetIsReg is true
130 enum ShiftType ShiftType; // used when OffsetRegShifted is true
131 const MCExpr *ShiftAmount; // used when OffsetRegShifted is true
132 bool Preindexed;
133 bool Postindexed;
134 bool Negative; // only used when OffsetIsReg is true
135 bool Writeback;
136 } Mem;
137
138 };
139
140 StringRef getToken() const {
141 assert(Kind == Token && "Invalid access!");
142 return StringRef(Tok.Data, Tok.Length);
143 }
144
145 unsigned getReg() const {
146 assert(Kind == Register && "Invalid access!");
147 return Reg.RegNum;
148 }
149
Kevin Enderbycfe07242009-10-13 22:19:02 +0000150 const MCExpr *getImm() const {
151 assert(Kind == Immediate && "Invalid access!");
152 return Imm.Val;
153 }
154
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000155 bool isToken() const {return Kind == Token; }
156
157 bool isReg() const { return Kind == Register; }
158
159 void addRegOperands(MCInst &Inst, unsigned N) const {
160 assert(N == 1 && "Invalid number of operands!");
161 Inst.addOperand(MCOperand::CreateReg(getReg()));
162 }
163
164 static ARMOperand CreateToken(StringRef Str) {
165 ARMOperand Res;
166 Res.Kind = Token;
167 Res.Tok.Data = Str.data();
168 Res.Tok.Length = Str.size();
169 return Res;
170 }
171
Kevin Enderby99e6d4e2009-10-07 18:01:35 +0000172 static ARMOperand CreateReg(unsigned RegNum, bool Writeback) {
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000173 ARMOperand Res;
174 Res.Kind = Register;
175 Res.Reg.RegNum = RegNum;
Kevin Enderby99e6d4e2009-10-07 18:01:35 +0000176 Res.Reg.Writeback = Writeback;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000177 return Res;
178 }
179
Kevin Enderbycfe07242009-10-13 22:19:02 +0000180 static ARMOperand CreateImm(const MCExpr *Val) {
181 ARMOperand Res;
182 Res.Kind = Immediate;
183 Res.Imm.Val = Val;
184 return Res;
185 }
186
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000187 static ARMOperand CreateMem(unsigned BaseRegNum, bool OffsetIsReg,
188 const MCExpr *Offset, unsigned OffsetRegNum,
189 bool OffsetRegShifted, enum ShiftType ShiftType,
190 const MCExpr *ShiftAmount, bool Preindexed,
191 bool Postindexed, bool Negative, bool Writeback) {
192 ARMOperand Res;
193 Res.Kind = Memory;
194 Res.Mem.BaseRegNum = BaseRegNum;
195 Res.Mem.OffsetIsReg = OffsetIsReg;
196 Res.Mem.Offset = Offset;
197 Res.Mem.OffsetRegNum = OffsetRegNum;
198 Res.Mem.OffsetRegShifted = OffsetRegShifted;
199 Res.Mem.ShiftType = ShiftType;
200 Res.Mem.ShiftAmount = ShiftAmount;
201 Res.Mem.Preindexed = Preindexed;
202 Res.Mem.Postindexed = Postindexed;
203 Res.Mem.Negative = Negative;
204 Res.Mem.Writeback = Writeback;
205 return Res;
206 }
207};
208
209} // end anonymous namespace.
210
211// Try to parse a register name. The token must be an Identifier when called,
212// and if it is a register name a Reg operand is created, the token is eaten
213// and false is returned. Else true is returned and no token is eaten.
214// TODO this is likely to change to allow different register types and or to
215// parse for a specific register type.
216bool ARMAsmParser::ParseRegister(ARMOperand &Op) {
217 const AsmToken &Tok = getLexer().getTok();
218 assert(Tok.is(AsmToken::Identifier) && "Token is not an Identifier");
219
220 // FIXME: Validate register for the current architecture; we have to do
221 // validation later, so maybe there is no need for this here.
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000222 int RegNum;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000223
224 RegNum = MatchRegisterName(Tok.getString());
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000225 if (RegNum == -1)
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000226 return true;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000227 getLexer().Lex(); // Eat identifier token.
228
Kevin Enderby99e6d4e2009-10-07 18:01:35 +0000229 bool Writeback = false;
230 const AsmToken &ExclaimTok = getLexer().getTok();
231 if (ExclaimTok.is(AsmToken::Exclaim)) {
232 Writeback = true;
233 getLexer().Lex(); // Eat exclaim token
234 }
235
236 Op = ARMOperand::CreateReg(RegNum, Writeback);
237
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000238 return false;
239}
240
Kevin Enderby515d5092009-10-15 20:48:48 +0000241// Parse a register list, return false if successful else return true or an
242// error. The first token must be a '{' when called.
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000243bool ARMAsmParser::ParseRegisterList(ARMOperand &Op) {
Kevin Enderby6bd266e2009-10-12 22:51:49 +0000244 assert(getLexer().getTok().is(AsmToken::LCurly) &&
Kevin Enderbycfe07242009-10-13 22:19:02 +0000245 "Token is not an Left Curly Brace");
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000246 getLexer().Lex(); // Eat left curly brace token.
247
248 const AsmToken &RegTok = getLexer().getTok();
249 SMLoc RegLoc = RegTok.getLoc();
250 if (RegTok.isNot(AsmToken::Identifier))
251 return Error(RegLoc, "register expected");
252 int RegNum = MatchRegisterName(RegTok.getString());
253 if (RegNum == -1)
254 return Error(RegLoc, "register expected");
255 getLexer().Lex(); // Eat identifier token.
256 unsigned RegList = 1 << RegNum;
257
258 int HighRegNum = RegNum;
259 // TODO ranges like "{Rn-Rm}"
260 while (getLexer().getTok().is(AsmToken::Comma)) {
261 getLexer().Lex(); // Eat comma token.
262
263 const AsmToken &RegTok = getLexer().getTok();
264 SMLoc RegLoc = RegTok.getLoc();
265 if (RegTok.isNot(AsmToken::Identifier))
266 return Error(RegLoc, "register expected");
267 int RegNum = MatchRegisterName(RegTok.getString());
268 if (RegNum == -1)
269 return Error(RegLoc, "register expected");
270
271 if (RegList & (1 << RegNum))
272 Warning(RegLoc, "register duplicated in register list");
273 else if (RegNum <= HighRegNum)
274 Warning(RegLoc, "register not in ascending order in register list");
275 RegList |= 1 << RegNum;
276 HighRegNum = RegNum;
277
278 getLexer().Lex(); // Eat identifier token.
279 }
280 const AsmToken &RCurlyTok = getLexer().getTok();
281 if (RCurlyTok.isNot(AsmToken::RCurly))
282 return Error(RCurlyTok.getLoc(), "'}' expected");
283 getLexer().Lex(); // Eat left curly brace token.
284
285 return false;
286}
287
Kevin Enderby515d5092009-10-15 20:48:48 +0000288// Parse an arm memory expression, return false if successful else return true
289// or an error. The first token must be a '[' when called.
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000290// TODO Only preindexing and postindexing addressing are started, unindexed
291// with option, etc are still to do.
292bool ARMAsmParser::ParseMemory(ARMOperand &Op) {
Kevin Enderby6bd266e2009-10-12 22:51:49 +0000293 assert(getLexer().getTok().is(AsmToken::LBrac) &&
294 "Token is not an Left Bracket");
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000295 getLexer().Lex(); // Eat left bracket token.
296
297 const AsmToken &BaseRegTok = getLexer().getTok();
298 if (BaseRegTok.isNot(AsmToken::Identifier))
299 return Error(BaseRegTok.getLoc(), "register expected");
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000300 int BaseRegNum = MatchRegisterName(BaseRegTok.getString());
301 if (BaseRegNum == -1)
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000302 return Error(BaseRegTok.getLoc(), "register expected");
303 getLexer().Lex(); // Eat identifier token.
304
305 bool Preindexed = false;
306 bool Postindexed = false;
307 bool OffsetIsReg = false;
308 bool Negative = false;
309 bool Writeback = false;
310
311 // First look for preindexed address forms:
312 // [Rn, +/-Rm]
313 // [Rn, #offset]
314 // [Rn, +/-Rm, shift]
315 // that is after the "[Rn" we now have see if the next token is a comma.
316 const AsmToken &Tok = getLexer().getTok();
317 if (Tok.is(AsmToken::Comma)) {
318 Preindexed = true;
319 getLexer().Lex(); // Eat comma token.
320
321 const AsmToken &NextTok = getLexer().getTok();
322 if (NextTok.is(AsmToken::Plus))
323 getLexer().Lex(); // Eat plus token.
324 else if (NextTok.is(AsmToken::Minus)) {
325 Negative = true;
326 getLexer().Lex(); // Eat minus token
327 }
328
329 // See if there is a register following the "[Rn," we have so far.
330 const AsmToken &OffsetRegTok = getLexer().getTok();
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000331 int OffsetRegNum = MatchRegisterName(OffsetRegTok.getString());
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000332 bool OffsetRegShifted = false;
333 enum ShiftType ShiftType;
334 const MCExpr *ShiftAmount;
335 const MCExpr *Offset;
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000336 if (OffsetRegNum != -1) {
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000337 OffsetIsReg = true;
338 getLexer().Lex(); // Eat identifier token for the offset register.
339 // Look for a comma then a shift
340 const AsmToken &Tok = getLexer().getTok();
341 if (Tok.is(AsmToken::Comma)) {
342 getLexer().Lex(); // Eat comma token.
343
344 const AsmToken &Tok = getLexer().getTok();
345 if (ParseShift(&ShiftType, ShiftAmount))
346 return Error(Tok.getLoc(), "shift expected");
347 OffsetRegShifted = true;
348 }
349 }
350 else { // "[Rn," we have so far was not followed by "Rm"
351 // Look for #offset following the "[Rn,"
352 const AsmToken &HashTok = getLexer().getTok();
353 if (HashTok.isNot(AsmToken::Hash))
354 return Error(HashTok.getLoc(), "'#' expected");
355 getLexer().Lex(); // Eat hash token.
356
357 if (getParser().ParseExpression(Offset))
358 return true;
359 }
360 const AsmToken &RBracTok = getLexer().getTok();
361 if (RBracTok.isNot(AsmToken::RBrac))
362 return Error(RBracTok.getLoc(), "']' expected");
363 getLexer().Lex(); // Eat right bracket token.
364
365 const AsmToken &ExclaimTok = getLexer().getTok();
366 if (ExclaimTok.is(AsmToken::Exclaim)) {
367 Writeback = true;
368 getLexer().Lex(); // Eat exclaim token
369 }
370 Op = ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset, OffsetRegNum,
371 OffsetRegShifted, ShiftType, ShiftAmount,
372 Preindexed, Postindexed, Negative, Writeback);
373 return false;
374 }
375 // The "[Rn" we have so far was not followed by a comma.
376 else if (Tok.is(AsmToken::RBrac)) {
377 // This is a post indexing addressing forms:
378 // [Rn], #offset
379 // [Rn], +/-Rm
380 // [Rn], +/-Rm, shift
381 // that is a ']' follows after the "[Rn".
382 Postindexed = true;
383 Writeback = true;
384 getLexer().Lex(); // Eat right bracket token.
385
386 const AsmToken &CommaTok = getLexer().getTok();
387 if (CommaTok.isNot(AsmToken::Comma))
388 return Error(CommaTok.getLoc(), "',' expected");
389 getLexer().Lex(); // Eat comma token.
390
391 const AsmToken &NextTok = getLexer().getTok();
392 if (NextTok.is(AsmToken::Plus))
393 getLexer().Lex(); // Eat plus token.
394 else if (NextTok.is(AsmToken::Minus)) {
395 Negative = true;
396 getLexer().Lex(); // Eat minus token
397 }
398
399 // See if there is a register following the "[Rn]," we have so far.
400 const AsmToken &OffsetRegTok = getLexer().getTok();
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000401 int OffsetRegNum = MatchRegisterName(OffsetRegTok.getString());
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000402 bool OffsetRegShifted = false;
403 enum ShiftType ShiftType;
404 const MCExpr *ShiftAmount;
405 const MCExpr *Offset;
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000406 if (OffsetRegNum != -1) {
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000407 OffsetIsReg = true;
408 getLexer().Lex(); // Eat identifier token for the offset register.
409 // Look for a comma then a shift
410 const AsmToken &Tok = getLexer().getTok();
411 if (Tok.is(AsmToken::Comma)) {
412 getLexer().Lex(); // Eat comma token.
413
414 const AsmToken &Tok = getLexer().getTok();
415 if (ParseShift(&ShiftType, ShiftAmount))
416 return Error(Tok.getLoc(), "shift expected");
417 OffsetRegShifted = true;
418 }
419 }
420 else { // "[Rn]," we have so far was not followed by "Rm"
421 // Look for #offset following the "[Rn],"
422 const AsmToken &HashTok = getLexer().getTok();
423 if (HashTok.isNot(AsmToken::Hash))
424 return Error(HashTok.getLoc(), "'#' expected");
425 getLexer().Lex(); // Eat hash token.
426
427 if (getParser().ParseExpression(Offset))
428 return true;
429 }
430 Op = ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset, OffsetRegNum,
431 OffsetRegShifted, ShiftType, ShiftAmount,
432 Preindexed, Postindexed, Negative, Writeback);
433 return false;
434 }
435
436 return true;
437}
438
439/// ParseShift as one of these two:
440/// ( lsl | lsr | asr | ror ) , # shift_amount
441/// rrx
442/// and returns true if it parses a shift otherwise it returns false.
Kevin Enderby9d369622009-10-12 22:39:54 +0000443bool ARMAsmParser::ParseShift(ShiftType *St, const MCExpr *&ShiftAmount) {
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000444 const AsmToken &Tok = getLexer().getTok();
445 if (Tok.isNot(AsmToken::Identifier))
446 return true;
447 const StringRef &ShiftName = Tok.getString();
448 if (ShiftName == "lsl" || ShiftName == "LSL")
449 *St = Lsl;
450 else if (ShiftName == "lsr" || ShiftName == "LSR")
451 *St = Lsr;
452 else if (ShiftName == "asr" || ShiftName == "ASR")
453 *St = Asr;
454 else if (ShiftName == "ror" || ShiftName == "ROR")
455 *St = Ror;
456 else if (ShiftName == "rrx" || ShiftName == "RRX")
457 *St = Rrx;
458 else
459 return true;
460 getLexer().Lex(); // Eat shift type token.
461
462 // For all but a Rotate right there must be a '#' and a shift amount
463 if (*St != Rrx) {
464 // Look for # following the shift type
465 const AsmToken &HashTok = getLexer().getTok();
466 if (HashTok.isNot(AsmToken::Hash))
467 return Error(HashTok.getLoc(), "'#' expected");
468 getLexer().Lex(); // Eat hash token.
469
470 if (getParser().ParseExpression(ShiftAmount))
471 return true;
472 }
473
474 return false;
475}
476
Kevin Enderby515d5092009-10-15 20:48:48 +0000477// A hack to allow some testing, to be replaced by a real table gen version.
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000478int ARMAsmParser::MatchRegisterName(const StringRef &Name) {
479 if (Name == "r0" || Name == "R0")
480 return 0;
481 else if (Name == "r1" || Name == "R1")
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000482 return 1;
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000483 else if (Name == "r2" || Name == "R2")
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000484 return 2;
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000485 else if (Name == "r3" || Name == "R3")
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000486 return 3;
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000487 else if (Name == "r3" || Name == "R3")
488 return 3;
489 else if (Name == "r4" || Name == "R4")
490 return 4;
491 else if (Name == "r5" || Name == "R5")
492 return 5;
493 else if (Name == "r6" || Name == "R6")
494 return 6;
495 else if (Name == "r7" || Name == "R7")
496 return 7;
497 else if (Name == "r8" || Name == "R8")
498 return 8;
499 else if (Name == "r9" || Name == "R9")
500 return 9;
501 else if (Name == "r10" || Name == "R10")
502 return 10;
503 else if (Name == "r11" || Name == "R11" || Name == "fp")
504 return 11;
505 else if (Name == "r12" || Name == "R12" || Name == "ip")
506 return 12;
507 else if (Name == "r13" || Name == "R13" || Name == "sp")
Kevin Enderby99e6d4e2009-10-07 18:01:35 +0000508 return 13;
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000509 else if (Name == "r14" || Name == "R14" || Name == "lr")
510 return 14;
511 else if (Name == "r15" || Name == "R15" || Name == "pc")
512 return 15;
513 return -1;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000514}
515
Kevin Enderby515d5092009-10-15 20:48:48 +0000516// A hack to allow some testing, to be replaced by a real table gen version.
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000517bool ARMAsmParser::MatchInstruction(SmallVectorImpl<ARMOperand> &Operands,
518 MCInst &Inst) {
519 struct ARMOperand Op0 = Operands[0];
520 assert(Op0.Kind == ARMOperand::Token && "First operand not a Token");
521 const StringRef &Mnemonic = Op0.getToken();
522 if (Mnemonic == "add" ||
Kevin Enderby99e6d4e2009-10-07 18:01:35 +0000523 Mnemonic == "stmfd" ||
524 Mnemonic == "str" ||
525 Mnemonic == "ldmfd" ||
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000526 Mnemonic == "ldr" ||
Kevin Enderbycfe07242009-10-13 22:19:02 +0000527 Mnemonic == "mov" ||
Kevin Enderby515d5092009-10-15 20:48:48 +0000528 Mnemonic == "sub" ||
529 Mnemonic == "bl" ||
530 Mnemonic == "push" ||
531 Mnemonic == "blx" ||
532 Mnemonic == "pop")
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000533 return false;
534
535 return true;
536}
537
Kevin Enderby515d5092009-10-15 20:48:48 +0000538// Parse a arm instruction operand. For now this parses the operand regardless
539// of the mnemonic.
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000540bool ARMAsmParser::ParseOperand(ARMOperand &Op) {
541 switch (getLexer().getKind()) {
542 case AsmToken::Identifier:
543 if (!ParseRegister(Op))
544 return false;
Kevin Enderby515d5092009-10-15 20:48:48 +0000545 // This was not a register so parse other operands that start with an
546 // identifier (like labels) as expressions and create them as immediates.
547 const MCExpr *IdVal;
548 if (getParser().ParseExpression(IdVal))
549 return true;
550 Op = ARMOperand::CreateImm(IdVal);
551 return false;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000552 case AsmToken::LBrac:
Kevin Enderby515d5092009-10-15 20:48:48 +0000553 return ParseMemory(Op);
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000554 case AsmToken::LCurly:
Kevin Enderby515d5092009-10-15 20:48:48 +0000555 return ParseRegisterList(Op);
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000556 case AsmToken::Hash:
Kevin Enderby079469f2009-10-13 23:33:38 +0000557 // #42 -> immediate.
558 // TODO: ":lower16:" and ":upper16:" modifiers after # before immediate
Kevin Enderbycfe07242009-10-13 22:19:02 +0000559 getLexer().Lex();
Kevin Enderby515d5092009-10-15 20:48:48 +0000560 const MCExpr *ImmVal;
561 if (getParser().ParseExpression(ImmVal))
Kevin Enderbycfe07242009-10-13 22:19:02 +0000562 return true;
Kevin Enderby515d5092009-10-15 20:48:48 +0000563 Op = ARMOperand::CreateImm(ImmVal);
Kevin Enderbycfe07242009-10-13 22:19:02 +0000564 return false;
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000565 default:
Kevin Enderbyd7894f12009-10-09 21:12:28 +0000566 return Error(getLexer().getTok().getLoc(), "unexpected token in operand");
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000567 }
568}
569
Kevin Enderby515d5092009-10-15 20:48:48 +0000570// Parse an arm instruction mnemonic followed by its operands.
Kevin Enderbyca9c42c2009-09-15 00:27:25 +0000571bool ARMAsmParser::ParseInstruction(const StringRef &Name, MCInst &Inst) {
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000572 SmallVector<ARMOperand, 7> Operands;
573
574 Operands.push_back(ARMOperand::CreateToken(Name));
575
Kevin Enderbyca9c42c2009-09-15 00:27:25 +0000576 SMLoc Loc = getLexer().getTok().getLoc();
Kevin Enderbya7ba3a82009-10-06 22:26:42 +0000577 if (getLexer().isNot(AsmToken::EndOfStatement)) {
578
579 // Read the first operand.
580 Operands.push_back(ARMOperand());
581 if (ParseOperand(Operands.back()))
582 return true;
583
584 while (getLexer().is(AsmToken::Comma)) {
585 getLexer().Lex(); // Eat the comma.
586
587 // Parse and remember the operand.
588 Operands.push_back(ARMOperand());
589 if (ParseOperand(Operands.back()))
590 return true;
591 }
592 }
593 if (!MatchInstruction(Operands, Inst))
594 return false;
595
596 Error(Loc, "ARMAsmParser::ParseInstruction only partly implemented");
Kevin Enderbyca9c42c2009-09-15 00:27:25 +0000597 return true;
598}
599
Kevin Enderby515d5092009-10-15 20:48:48 +0000600/// ParseDirective parses the arm specific directives
Kevin Enderbyca9c42c2009-09-15 00:27:25 +0000601bool ARMAsmParser::ParseDirective(AsmToken DirectiveID) {
602 StringRef IDVal = DirectiveID.getIdentifier();
603 if (IDVal == ".word")
604 return ParseDirectiveWord(4, DirectiveID.getLoc());
Kevin Enderby515d5092009-10-15 20:48:48 +0000605 else if (IDVal == ".thumb")
606 return ParseDirectiveThumb(DirectiveID.getLoc());
607 else if (IDVal == ".thumb_func")
608 return ParseDirectiveThumbFunc(DirectiveID.getLoc());
609 else if (IDVal == ".code")
610 return ParseDirectiveCode(DirectiveID.getLoc());
611 else if (IDVal == ".syntax")
612 return ParseDirectiveSyntax(DirectiveID.getLoc());
Kevin Enderbyca9c42c2009-09-15 00:27:25 +0000613 return true;
614}
615
616/// ParseDirectiveWord
617/// ::= .word [ expression (, expression)* ]
618bool ARMAsmParser::ParseDirectiveWord(unsigned Size, SMLoc L) {
619 if (getLexer().isNot(AsmToken::EndOfStatement)) {
620 for (;;) {
621 const MCExpr *Value;
622 if (getParser().ParseExpression(Value))
623 return true;
624
625 getParser().getStreamer().EmitValue(Value, Size);
626
627 if (getLexer().is(AsmToken::EndOfStatement))
628 break;
629
630 // FIXME: Improve diagnostic.
631 if (getLexer().isNot(AsmToken::Comma))
632 return Error(L, "unexpected token in directive");
633 getLexer().Lex();
634 }
635 }
636
637 getLexer().Lex();
638 return false;
639}
640
Kevin Enderby515d5092009-10-15 20:48:48 +0000641/// ParseDirectiveThumb
642/// ::= .thumb
643bool ARMAsmParser::ParseDirectiveThumb(SMLoc L) {
644 if (getLexer().isNot(AsmToken::EndOfStatement))
645 return Error(L, "unexpected token in directive");
646 getLexer().Lex();
647
648 // TODO: set thumb mode
649 // TODO: tell the MC streamer the mode
650 // getParser().getStreamer().Emit???();
651 return false;
652}
653
654/// ParseDirectiveThumbFunc
655/// ::= .thumbfunc symbol_name
656bool ARMAsmParser::ParseDirectiveThumbFunc(SMLoc L) {
657 const AsmToken &Tok = getLexer().getTok();
658 if (Tok.isNot(AsmToken::Identifier) && Tok.isNot(AsmToken::String))
659 return Error(L, "unexpected token in .syntax directive");
660 StringRef SymbolName = getLexer().getTok().getIdentifier();
661 getLexer().Lex(); // Consume the identifier token.
662
663 if (getLexer().isNot(AsmToken::EndOfStatement))
664 return Error(L, "unexpected token in directive");
665 getLexer().Lex();
666
667 // TODO: mark symbol as a thumb symbol
668 // getParser().getStreamer().Emit???();
669 return false;
670}
671
672/// ParseDirectiveSyntax
673/// ::= .syntax unified | divided
674bool ARMAsmParser::ParseDirectiveSyntax(SMLoc L) {
675 const AsmToken &Tok = getLexer().getTok();
676 if (Tok.isNot(AsmToken::Identifier))
677 return Error(L, "unexpected token in .syntax directive");
678 const StringRef &Mode = Tok.getString();
679 bool unified_syntax;
680 if (Mode == "unified" || Mode == "UNIFIED") {
681 getLexer().Lex();
682 unified_syntax = true;
683 }
684 else if (Mode == "divided" || Mode == "DIVIDED") {
685 getLexer().Lex();
686 unified_syntax = false;
687 }
688 else
689 return Error(L, "unrecognized syntax mode in .syntax directive");
690
691 if (getLexer().isNot(AsmToken::EndOfStatement))
692 return Error(getLexer().getTok().getLoc(), "unexpected token in directive");
693 getLexer().Lex();
694
695 // TODO tell the MC streamer the mode
696 // getParser().getStreamer().Emit???();
697 return false;
698}
699
700/// ParseDirectiveCode
701/// ::= .code 16 | 32
702bool ARMAsmParser::ParseDirectiveCode(SMLoc L) {
703 const AsmToken &Tok = getLexer().getTok();
704 if (Tok.isNot(AsmToken::Integer))
705 return Error(L, "unexpected token in .code directive");
706 int64_t Val = getLexer().getTok().getIntVal();
707 bool thumb_mode;
708 if (Val == 16) {
709 getLexer().Lex();
710 thumb_mode = true;
711 }
712 else if (Val == 32) {
713 getLexer().Lex();
714 thumb_mode = false;
715 }
716 else
717 return Error(L, "invalid operand to .code directive");
718
719 if (getLexer().isNot(AsmToken::EndOfStatement))
720 return Error(getLexer().getTok().getLoc(), "unexpected token in directive");
721 getLexer().Lex();
722
723 // TODO tell the MC streamer the mode
724 // getParser().getStreamer().Emit???();
725 return false;
726}
727
Kevin Enderbyca9c42c2009-09-15 00:27:25 +0000728// Force static initialization.
729extern "C" void LLVMInitializeARMAsmParser() {
730 RegisterAsmParser<ARMAsmParser> X(TheARMTarget);
731 RegisterAsmParser<ARMAsmParser> Y(TheThumbTarget);
732}