ARMAsmParser.cpp revision 6aa49435994c33257b7588cac24671785d17fa6e 
1//===-- 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 "ARMAddressingModes.h"
12#include "ARMSubtarget.h"
13#include "llvm/MC/MCParser/MCAsmLexer.h"
14#include "llvm/MC/MCParser/MCAsmParser.h"
15#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
16#include "llvm/MC/MCContext.h"
17#include "llvm/MC/MCStreamer.h"
18#include "llvm/MC/MCExpr.h"
19#include "llvm/MC/MCInst.h"
20#include "llvm/Target/TargetRegistry.h"
21#include "llvm/Target/TargetAsmParser.h"
22#include "llvm/Support/SourceMgr.h"
23#include "llvm/Support/raw_ostream.h"
24#include "llvm/ADT/SmallVector.h"
25#include "llvm/ADT/StringSwitch.h"
26#include "llvm/ADT/Twine.h"
27using namespace llvm;
28
29// The shift types for register controlled shifts in arm memory addressing
30enum ShiftType {
31  Lsl,
32  Lsr,
33  Asr,
34  Ror,
35  Rrx
36};
37
38namespace {
39
40class ARMOperand;
41
42class ARMAsmParser : public TargetAsmParser {
43  MCAsmParser &Parser;
44  TargetMachine &TM;
45
46  MCAsmParser &getParser() const { return Parser; }
47  MCAsmLexer &getLexer() const { return Parser.getLexer(); }
48
49  void Warning(SMLoc L, const Twine &Msg) { Parser.Warning(L, Msg); }
50  bool Error(SMLoc L, const Twine &Msg) { return Parser.Error(L, Msg); }
51
52  int TryParseRegister();
53  bool TryParseRegisterWithWriteBack(SmallVectorImpl<MCParsedAsmOperand*> &);
54  bool ParseRegisterList(SmallVectorImpl<MCParsedAsmOperand*> &);
55  bool ParseMemory(SmallVectorImpl<MCParsedAsmOperand*> &);
56  bool ParseOperand(SmallVectorImpl<MCParsedAsmOperand*> &);
57
58  bool ParseMemoryOffsetReg(bool &Negative,
59                            bool &OffsetRegShifted,
60                            enum ShiftType &ShiftType,
61                            const MCExpr *&ShiftAmount,
62                            const MCExpr *&Offset,
63                            bool &OffsetIsReg,
64                            int &OffsetRegNum,
65                            SMLoc &E);
66  bool ParseShift(enum ShiftType &St, const MCExpr *&ShiftAmount, SMLoc &E);
67  bool ParseDirectiveWord(unsigned Size, SMLoc L);
68  bool ParseDirectiveThumb(SMLoc L);
69  bool ParseDirectiveThumbFunc(SMLoc L);
70  bool ParseDirectiveCode(SMLoc L);
71  bool ParseDirectiveSyntax(SMLoc L);
72
73  bool MatchAndEmitInstruction(SMLoc IDLoc,
74                               SmallVectorImpl<MCParsedAsmOperand*> &Operands,
75                               MCStreamer &Out);
76
77  /// @name Auto-generated Match Functions
78  /// {
79
80#define GET_ASSEMBLER_HEADER
81#include "ARMGenAsmMatcher.inc"
82
83  /// }
84
85public:
86  ARMAsmParser(const Target &T, MCAsmParser &_Parser, TargetMachine &_TM)
87    : TargetAsmParser(T), Parser(_Parser), TM(_TM) {
88      // Initialize the set of available features.
89      setAvailableFeatures(ComputeAvailableFeatures(
90          &TM.getSubtarget<ARMSubtarget>()));
91    }
92
93  virtual bool ParseInstruction(StringRef Name, SMLoc NameLoc,
94                                SmallVectorImpl<MCParsedAsmOperand*> &Operands);
95  virtual bool ParseDirective(AsmToken DirectiveID);
96};
97} // end anonymous namespace
98
99namespace {
100
101/// ARMOperand - Instances of this class represent a parsed ARM machine
102/// instruction.
103class ARMOperand : public MCParsedAsmOperand {
104  enum KindTy {
105    CondCode,
106    CCOut,
107    Immediate,
108    Memory,
109    Register,
110    RegisterList,
111    DPRRegisterList,
112    SPRRegisterList,
113    Token
114  } Kind;
115
116  SMLoc StartLoc, EndLoc;
117  SmallVector<unsigned, 8> Registers;
118
119  union {
120    struct {
121      ARMCC::CondCodes Val;
122    } CC;
123
124    struct {
125      const char *Data;
126      unsigned Length;
127    } Tok;
128
129    struct {
130      unsigned RegNum;
131    } Reg;
132
133    struct {
134      const MCExpr *Val;
135    } Imm;
136
137    // This is for all forms of ARM address expressions
138    struct {
139      unsigned BaseRegNum;
140      unsigned OffsetRegNum;         // used when OffsetIsReg is true
141      const MCExpr *Offset;          // used when OffsetIsReg is false
142      const MCExpr *ShiftAmount;     // used when OffsetRegShifted is true
143      enum ShiftType ShiftType;      // used when OffsetRegShifted is true
144      unsigned OffsetRegShifted : 1; // only used when OffsetIsReg is true
145      unsigned Preindexed       : 1;
146      unsigned Postindexed      : 1;
147      unsigned OffsetIsReg      : 1;
148      unsigned Negative         : 1; // only used when OffsetIsReg is true
149      unsigned Writeback        : 1;
150    } Mem;
151  };
152
153  ARMOperand(KindTy K) : MCParsedAsmOperand(), Kind(K) {}
154public:
155  ARMOperand(const ARMOperand &o) : MCParsedAsmOperand() {
156    Kind = o.Kind;
157    StartLoc = o.StartLoc;
158    EndLoc = o.EndLoc;
159    switch (Kind) {
160    case CondCode:
161      CC = o.CC;
162      break;
163    case Token:
164      Tok = o.Tok;
165      break;
166    case CCOut:
167    case Register:
168      Reg = o.Reg;
169      break;
170    case RegisterList:
171    case DPRRegisterList:
172    case SPRRegisterList:
173      Registers = o.Registers;
174      break;
175    case Immediate:
176      Imm = o.Imm;
177      break;
178    case Memory:
179      Mem = o.Mem;
180      break;
181    }
182  }
183
184  /// getStartLoc - Get the location of the first token of this operand.
185  SMLoc getStartLoc() const { return StartLoc; }
186  /// getEndLoc - Get the location of the last token of this operand.
187  SMLoc getEndLoc() const { return EndLoc; }
188
189  ARMCC::CondCodes getCondCode() const {
190    assert(Kind == CondCode && "Invalid access!");
191    return CC.Val;
192  }
193
194  StringRef getToken() const {
195    assert(Kind == Token && "Invalid access!");
196    return StringRef(Tok.Data, Tok.Length);
197  }
198
199  unsigned getReg() const {
200    assert((Kind == Register || Kind == CCOut) && "Invalid access!");
201    return Reg.RegNum;
202  }
203
204  const SmallVectorImpl<unsigned> &getRegList() const {
205    assert((Kind == RegisterList || Kind == DPRRegisterList ||
206            Kind == SPRRegisterList) && "Invalid access!");
207    return Registers;
208  }
209
210  const MCExpr *getImm() const {
211    assert(Kind == Immediate && "Invalid access!");
212    return Imm.Val;
213  }
214
215  bool isCondCode() const { return Kind == CondCode; }
216  bool isCCOut() const { return Kind == CCOut; }
217  bool isImm() const { return Kind == Immediate; }
218  bool isReg() const { return Kind == Register; }
219  bool isRegList() const { return Kind == RegisterList; }
220  bool isDPRRegList() const { return Kind == DPRRegisterList; }
221  bool isSPRRegList() const { return Kind == SPRRegisterList; }
222  bool isToken() const { return Kind == Token; }
223  bool isMemory() const { return Kind == Memory; }
224  bool isMemMode5() const {
225    if (!isMemory() || Mem.OffsetIsReg || Mem.OffsetRegShifted ||
226        Mem.Writeback || Mem.Negative)
227      return false;
228
229    // If there is an offset expression, make sure it's valid.
230    if (!Mem.Offset) return true;
231
232    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
233    if (!CE) return false;
234
235    // The offset must be a multiple of 4 in the range 0-1020.
236    int64_t Value = CE->getValue();
237    return ((Value & 0x3) == 0 && Value <= 1020 && Value >= -1020);
238  }
239  bool isMemModeThumb() const {
240    if (!isMemory() || (!Mem.OffsetIsReg && !Mem.Offset) || Mem.Writeback)
241      return false;
242
243    if (!Mem.Offset) return true;
244
245    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
246    if (!CE) return false;
247
248    // The offset must be a multiple of 4 in the range 0-124.
249    uint64_t Value = CE->getValue();
250    return ((Value & 0x3) == 0 && Value <= 124);
251  }
252
253  void addExpr(MCInst &Inst, const MCExpr *Expr) const {
254    // Add as immediates when possible.  Null MCExpr = 0.
255    if (Expr == 0)
256      Inst.addOperand(MCOperand::CreateImm(0));
257    else if (const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Expr))
258      Inst.addOperand(MCOperand::CreateImm(CE->getValue()));
259    else
260      Inst.addOperand(MCOperand::CreateExpr(Expr));
261  }
262
263  void addCondCodeOperands(MCInst &Inst, unsigned N) const {
264    assert(N == 2 && "Invalid number of operands!");
265    Inst.addOperand(MCOperand::CreateImm(unsigned(getCondCode())));
266    unsigned RegNum = getCondCode() == ARMCC::AL ? 0: ARM::CPSR;
267    Inst.addOperand(MCOperand::CreateReg(RegNum));
268  }
269
270  void addCCOutOperands(MCInst &Inst, unsigned N) const {
271    assert(N == 1 && "Invalid number of operands!");
272    Inst.addOperand(MCOperand::CreateReg(getReg()));
273  }
274
275  void addRegOperands(MCInst &Inst, unsigned N) const {
276    assert(N == 1 && "Invalid number of operands!");
277    Inst.addOperand(MCOperand::CreateReg(getReg()));
278  }
279
280  void addRegListOperands(MCInst &Inst, unsigned N) const {
281    assert(N == 1 && "Invalid number of operands!");
282    const SmallVectorImpl<unsigned> &RegList = getRegList();
283    for (SmallVectorImpl<unsigned>::const_iterator
284           I = RegList.begin(), E = RegList.end(); I != E; ++I)
285      Inst.addOperand(MCOperand::CreateReg(*I));
286  }
287
288  void addDPRRegListOperands(MCInst &Inst, unsigned N) const {
289    addRegListOperands(Inst, N);
290  }
291
292  void addSPRRegListOperands(MCInst &Inst, unsigned N) const {
293    addRegListOperands(Inst, N);
294  }
295
296  void addImmOperands(MCInst &Inst, unsigned N) const {
297    assert(N == 1 && "Invalid number of operands!");
298    addExpr(Inst, getImm());
299  }
300
301  void addMemMode5Operands(MCInst &Inst, unsigned N) const {
302    assert(N == 2 && isMemMode5() && "Invalid number of operands!");
303
304    Inst.addOperand(MCOperand::CreateReg(Mem.BaseRegNum));
305    assert(!Mem.OffsetIsReg && "Invalid mode 5 operand");
306
307    // FIXME: #-0 is encoded differently than #0. Does the parser preserve
308    // the difference?
309    if (Mem.Offset) {
310      const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
311      assert(CE && "Non-constant mode 5 offset operand!");
312
313      // The MCInst offset operand doesn't include the low two bits (like
314      // the instruction encoding).
315      int64_t Offset = CE->getValue() / 4;
316      if (Offset >= 0)
317        Inst.addOperand(MCOperand::CreateImm(ARM_AM::getAM5Opc(ARM_AM::add,
318                                                               Offset)));
319      else
320        Inst.addOperand(MCOperand::CreateImm(ARM_AM::getAM5Opc(ARM_AM::sub,
321                                                               -Offset)));
322    } else {
323      Inst.addOperand(MCOperand::CreateImm(0));
324    }
325  }
326
327  void addMemModeThumbOperands(MCInst &Inst, unsigned N) const {
328    assert(N == 3 && isMemModeThumb() && "Invalid number of operands!");
329    Inst.addOperand(MCOperand::CreateReg(Mem.BaseRegNum));
330
331    if (Mem.Offset) {
332      const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
333      assert(CE && "Non-constant mode offset operand!");
334      Inst.addOperand(MCOperand::CreateImm(CE->getValue()));
335      Inst.addOperand(MCOperand::CreateReg(0));
336    } else {
337      Inst.addOperand(MCOperand::CreateImm(0));
338      Inst.addOperand(MCOperand::CreateReg(Mem.OffsetRegNum));
339    }
340  }
341
342  virtual void dump(raw_ostream &OS) const;
343
344  static ARMOperand *CreateCondCode(ARMCC::CondCodes CC, SMLoc S) {
345    ARMOperand *Op = new ARMOperand(CondCode);
346    Op->CC.Val = CC;
347    Op->StartLoc = S;
348    Op->EndLoc = S;
349    return Op;
350  }
351
352  static ARMOperand *CreateCCOut(unsigned RegNum, SMLoc S) {
353    ARMOperand *Op = new ARMOperand(CCOut);
354    Op->Reg.RegNum = RegNum;
355    Op->StartLoc = S;
356    Op->EndLoc = S;
357    return Op;
358  }
359
360  static ARMOperand *CreateToken(StringRef Str, SMLoc S) {
361    ARMOperand *Op = new ARMOperand(Token);
362    Op->Tok.Data = Str.data();
363    Op->Tok.Length = Str.size();
364    Op->StartLoc = S;
365    Op->EndLoc = S;
366    return Op;
367  }
368
369  static ARMOperand *CreateReg(unsigned RegNum, SMLoc S, SMLoc E) {
370    ARMOperand *Op = new ARMOperand(Register);
371    Op->Reg.RegNum = RegNum;
372    Op->StartLoc = S;
373    Op->EndLoc = E;
374    return Op;
375  }
376
377  static ARMOperand *
378  CreateRegList(const SmallVectorImpl<std::pair<unsigned, SMLoc> > &Regs,
379                SMLoc StartLoc, SMLoc EndLoc) {
380    KindTy Kind = RegisterList;
381
382    if (ARM::DPRRegClass.contains(Regs.front().first))
383      Kind = DPRRegisterList;
384    else if (ARM::SPRRegClass.contains(Regs.front().first))
385      Kind = SPRRegisterList;
386
387    ARMOperand *Op = new ARMOperand(Kind);
388    for (SmallVectorImpl<std::pair<unsigned, SMLoc> >::const_iterator
389           I = Regs.begin(), E = Regs.end(); I != E; ++I)
390      Op->Registers.push_back(I->first);
391    array_pod_sort(Op->Registers.begin(), Op->Registers.end());
392    Op->StartLoc = StartLoc;
393    Op->EndLoc = EndLoc;
394    return Op;
395  }
396
397  static ARMOperand *CreateImm(const MCExpr *Val, SMLoc S, SMLoc E) {
398    ARMOperand *Op = new ARMOperand(Immediate);
399    Op->Imm.Val = Val;
400    Op->StartLoc = S;
401    Op->EndLoc = E;
402    return Op;
403  }
404
405  static ARMOperand *CreateMem(unsigned BaseRegNum, bool OffsetIsReg,
406                               const MCExpr *Offset, unsigned OffsetRegNum,
407                               bool OffsetRegShifted, enum ShiftType ShiftType,
408                               const MCExpr *ShiftAmount, bool Preindexed,
409                               bool Postindexed, bool Negative, bool Writeback,
410                               SMLoc S, SMLoc E) {
411    ARMOperand *Op = new ARMOperand(Memory);
412    Op->Mem.BaseRegNum = BaseRegNum;
413    Op->Mem.OffsetIsReg = OffsetIsReg;
414    Op->Mem.Offset = Offset;
415    Op->Mem.OffsetRegNum = OffsetRegNum;
416    Op->Mem.OffsetRegShifted = OffsetRegShifted;
417    Op->Mem.ShiftType = ShiftType;
418    Op->Mem.ShiftAmount = ShiftAmount;
419    Op->Mem.Preindexed = Preindexed;
420    Op->Mem.Postindexed = Postindexed;
421    Op->Mem.Negative = Negative;
422    Op->Mem.Writeback = Writeback;
423
424    Op->StartLoc = S;
425    Op->EndLoc = E;
426    return Op;
427  }
428};
429
430} // end anonymous namespace.
431
432void ARMOperand::dump(raw_ostream &OS) const {
433  switch (Kind) {
434  case CondCode:
435    OS << ARMCondCodeToString(getCondCode());
436    break;
437  case CCOut:
438    OS << "<ccout " << getReg() << ">";
439    break;
440  case Immediate:
441    getImm()->print(OS);
442    break;
443  case Memory:
444    OS << "<memory>";
445    break;
446  case Register:
447    OS << "<register " << getReg() << ">";
448    break;
449  case RegisterList:
450  case DPRRegisterList:
451  case SPRRegisterList: {
452    OS << "<register_list ";
453
454    const SmallVectorImpl<unsigned> &RegList = getRegList();
455    for (SmallVectorImpl<unsigned>::const_iterator
456           I = RegList.begin(), E = RegList.end(); I != E; ) {
457      OS << *I;
458      if (++I < E) OS << ", ";
459    }
460
461    OS << ">";
462    break;
463  }
464  case Token:
465    OS << "'" << getToken() << "'";
466    break;
467  }
468}
469
470/// @name Auto-generated Match Functions
471/// {
472
473static unsigned MatchRegisterName(StringRef Name);
474
475/// }
476
477/// Try to parse a register name.  The token must be an Identifier when called,
478/// and if it is a register name the token is eaten and the register number is
479/// returned.  Otherwise return -1.
480///
481int ARMAsmParser::TryParseRegister() {
482  const AsmToken &Tok = Parser.getTok();
483  assert(Tok.is(AsmToken::Identifier) && "Token is not an Identifier");
484
485  // FIXME: Validate register for the current architecture; we have to do
486  // validation later, so maybe there is no need for this here.
487  unsigned RegNum = MatchRegisterName(Tok.getString());
488  if (RegNum == 0)
489    return -1;
490  Parser.Lex(); // Eat identifier token.
491  return RegNum;
492}
493
494
495/// Try to parse a register name.  The token must be an Identifier when called.
496/// If it's a register, an AsmOperand is created. Another AsmOperand is created
497/// if there is a "writeback". 'true' if it's not a register.
498///
499/// TODO this is likely to change to allow different register types and or to
500/// parse for a specific register type.
501bool ARMAsmParser::
502TryParseRegisterWithWriteBack(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
503  SMLoc S = Parser.getTok().getLoc();
504  int RegNo = TryParseRegister();
505  if (RegNo == -1)
506    return true;
507
508  Operands.push_back(ARMOperand::CreateReg(RegNo, S, Parser.getTok().getLoc()));
509
510  const AsmToken &ExclaimTok = Parser.getTok();
511  if (ExclaimTok.is(AsmToken::Exclaim)) {
512    Operands.push_back(ARMOperand::CreateToken(ExclaimTok.getString(),
513                                               ExclaimTok.getLoc()));
514    Parser.Lex(); // Eat exclaim token
515  }
516
517  return false;
518}
519
520/// Parse a register list, return it if successful else return null.  The first
521/// token must be a '{' when called.
522bool ARMAsmParser::
523ParseRegisterList(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
524  assert(Parser.getTok().is(AsmToken::LCurly) &&
525         "Token is not a Left Curly Brace");
526  SMLoc S = Parser.getTok().getLoc();
527
528  // Read the rest of the registers in the list.
529  unsigned PrevRegNum = 0;
530  SmallVector<std::pair<unsigned, SMLoc>, 32> Registers;
531
532  do {
533    bool IsRange = Parser.getTok().is(AsmToken::Minus);
534    Parser.Lex(); // Eat non-identifier token.
535
536    const AsmToken &RegTok = Parser.getTok();
537    SMLoc RegLoc = RegTok.getLoc();
538    if (RegTok.isNot(AsmToken::Identifier)) {
539      Error(RegLoc, "register expected");
540      return true;
541    }
542
543    int RegNum = TryParseRegister();
544    if (RegNum == -1) {
545      Error(RegLoc, "register expected");
546      return true;
547    }
548
549    if (IsRange) {
550      int Reg = PrevRegNum;
551      do {
552        ++Reg;
553        Registers.push_back(std::make_pair(Reg, RegLoc));
554      } while (Reg != RegNum);
555    } else {
556      Registers.push_back(std::make_pair(RegNum, RegLoc));
557    }
558
559    PrevRegNum = RegNum;
560  } while (Parser.getTok().is(AsmToken::Comma) ||
561           Parser.getTok().is(AsmToken::Minus));
562
563  // Process the right curly brace of the list.
564  const AsmToken &RCurlyTok = Parser.getTok();
565  if (RCurlyTok.isNot(AsmToken::RCurly)) {
566    Error(RCurlyTok.getLoc(), "'}' expected");
567    return true;
568  }
569
570  SMLoc E = RCurlyTok.getLoc();
571  Parser.Lex(); // Eat right curly brace token.
572
573  // Verify the register list.
574  SmallVectorImpl<std::pair<unsigned, SMLoc> >::const_iterator
575    RI = Registers.begin(), RE = Registers.end();
576
577  DenseMap<unsigned, bool> RegMap;
578  RegMap[RI->first] = true;
579
580  unsigned HighRegNum = RI->first;
581  bool EmittedWarning = false;
582
583  for (++RI; RI != RE; ++RI) {
584    const std::pair<unsigned, SMLoc> &RegInfo = *RI;
585    unsigned Reg = RegInfo.first;
586
587    if (RegMap[Reg]) {
588      Error(RegInfo.second, "register duplicated in register list");
589      return true;
590    }
591
592    if (!EmittedWarning && Reg < HighRegNum)
593      Warning(RegInfo.second,
594              "register not in ascending order in register list");
595
596    RegMap[Reg] = true;
597    HighRegNum = std::max(Reg, HighRegNum);
598  }
599
600  Operands.push_back(ARMOperand::CreateRegList(Registers, S, E));
601  return false;
602}
603
604/// Parse an ARM memory expression, return false if successful else return true
605/// or an error.  The first token must be a '[' when called.
606///
607/// TODO Only preindexing and postindexing addressing are started, unindexed
608/// with option, etc are still to do.
609bool ARMAsmParser::
610ParseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
611  SMLoc S, E;
612  assert(Parser.getTok().is(AsmToken::LBrac) &&
613         "Token is not a Left Bracket");
614  S = Parser.getTok().getLoc();
615  Parser.Lex(); // Eat left bracket token.
616
617  const AsmToken &BaseRegTok = Parser.getTok();
618  if (BaseRegTok.isNot(AsmToken::Identifier)) {
619    Error(BaseRegTok.getLoc(), "register expected");
620    return true;
621  }
622  int BaseRegNum = TryParseRegister();
623  if (BaseRegNum == -1) {
624    Error(BaseRegTok.getLoc(), "register expected");
625    return true;
626  }
627
628  bool Preindexed = false;
629  bool Postindexed = false;
630  bool OffsetIsReg = false;
631  bool Negative = false;
632  bool Writeback = false;
633
634  // First look for preindexed address forms, that is after the "[Rn" we now
635  // have to see if the next token is a comma.
636  const AsmToken &Tok = Parser.getTok();
637  if (Tok.is(AsmToken::Comma)) {
638    Preindexed = true;
639    Parser.Lex(); // Eat comma token.
640    int OffsetRegNum;
641    bool OffsetRegShifted;
642    enum ShiftType ShiftType;
643    const MCExpr *ShiftAmount = 0;
644    const MCExpr *Offset = 0;
645    if (ParseMemoryOffsetReg(Negative, OffsetRegShifted, ShiftType, ShiftAmount,
646                             Offset, OffsetIsReg, OffsetRegNum, E))
647      return true;
648    const AsmToken &RBracTok = Parser.getTok();
649    if (RBracTok.isNot(AsmToken::RBrac)) {
650      Error(RBracTok.getLoc(), "']' expected");
651      return true;
652    }
653    E = RBracTok.getLoc();
654    Parser.Lex(); // Eat right bracket token.
655
656
657    const AsmToken &ExclaimTok = Parser.getTok();
658    ARMOperand *WBOp = 0;
659    if (ExclaimTok.is(AsmToken::Exclaim)) {
660      WBOp = ARMOperand::CreateToken(ExclaimTok.getString(),
661                                     ExclaimTok.getLoc());
662      Writeback = true;
663      Parser.Lex(); // Eat exclaim token
664    }
665
666    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset,
667                                             OffsetRegNum, OffsetRegShifted,
668                                             ShiftType, ShiftAmount, Preindexed,
669                                             Postindexed, Negative, Writeback,
670                                             S, E));
671    if (WBOp)
672      Operands.push_back(WBOp);
673
674    return false;
675  }
676  // The "[Rn" we have so far was not followed by a comma.
677  else if (Tok.is(AsmToken::RBrac)) {
678    // If there's anything other than the right brace, this is a post indexing
679    // addressing form.
680    E = Tok.getLoc();
681    Parser.Lex(); // Eat right bracket token.
682
683    int OffsetRegNum = 0;
684    bool OffsetRegShifted = false;
685    enum ShiftType ShiftType = Lsl;
686    const MCExpr *ShiftAmount = 0;
687    const MCExpr *Offset = 0;
688
689    const AsmToken &NextTok = Parser.getTok();
690
691    if (NextTok.isNot(AsmToken::EndOfStatement)) {
692      Postindexed = true;
693      Writeback = true;
694
695      if (NextTok.isNot(AsmToken::Comma)) {
696        Error(NextTok.getLoc(), "',' expected");
697        return true;
698      }
699
700      Parser.Lex(); // Eat comma token.
701
702      if (ParseMemoryOffsetReg(Negative, OffsetRegShifted, ShiftType,
703                               ShiftAmount, Offset, OffsetIsReg, OffsetRegNum,
704                               E))
705        return true;
706    }
707
708    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset,
709                                             OffsetRegNum, OffsetRegShifted,
710                                             ShiftType, ShiftAmount, Preindexed,
711                                             Postindexed, Negative, Writeback,
712                                             S, E));
713    return false;
714  }
715
716  return true;
717}
718
719/// Parse the offset of a memory operand after we have seen "[Rn," or "[Rn],"
720/// we will parse the following (were +/- means that a plus or minus is
721/// optional):
722///   +/-Rm
723///   +/-Rm, shift
724///   #offset
725/// we return false on success or an error otherwise.
726bool ARMAsmParser::ParseMemoryOffsetReg(bool &Negative,
727                                        bool &OffsetRegShifted,
728                                        enum ShiftType &ShiftType,
729                                        const MCExpr *&ShiftAmount,
730                                        const MCExpr *&Offset,
731                                        bool &OffsetIsReg,
732                                        int &OffsetRegNum,
733                                        SMLoc &E) {
734  Negative = false;
735  OffsetRegShifted = false;
736  OffsetIsReg = false;
737  OffsetRegNum = -1;
738  const AsmToken &NextTok = Parser.getTok();
739  E = NextTok.getLoc();
740  if (NextTok.is(AsmToken::Plus))
741    Parser.Lex(); // Eat plus token.
742  else if (NextTok.is(AsmToken::Minus)) {
743    Negative = true;
744    Parser.Lex(); // Eat minus token
745  }
746  // See if there is a register following the "[Rn," or "[Rn]," we have so far.
747  const AsmToken &OffsetRegTok = Parser.getTok();
748  if (OffsetRegTok.is(AsmToken::Identifier)) {
749    SMLoc CurLoc = OffsetRegTok.getLoc();
750    OffsetRegNum = TryParseRegister();
751    if (OffsetRegNum != -1) {
752      OffsetIsReg = true;
753      E = CurLoc;
754    }
755  }
756
757  // If we parsed a register as the offset then there can be a shift after that.
758  if (OffsetRegNum != -1) {
759    // Look for a comma then a shift
760    const AsmToken &Tok = Parser.getTok();
761    if (Tok.is(AsmToken::Comma)) {
762      Parser.Lex(); // Eat comma token.
763
764      const AsmToken &Tok = Parser.getTok();
765      if (ParseShift(ShiftType, ShiftAmount, E))
766        return Error(Tok.getLoc(), "shift expected");
767      OffsetRegShifted = true;
768    }
769  }
770  else { // the "[Rn," or "[Rn,]" we have so far was not followed by "Rm"
771    // Look for #offset following the "[Rn," or "[Rn],"
772    const AsmToken &HashTok = Parser.getTok();
773    if (HashTok.isNot(AsmToken::Hash))
774      return Error(HashTok.getLoc(), "'#' expected");
775
776    Parser.Lex(); // Eat hash token.
777
778    if (getParser().ParseExpression(Offset))
779     return true;
780    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
781  }
782  return false;
783}
784
785/// ParseShift as one of these two:
786///   ( lsl | lsr | asr | ror ) , # shift_amount
787///   rrx
788/// and returns true if it parses a shift otherwise it returns false.
789bool ARMAsmParser::ParseShift(ShiftType &St, const MCExpr *&ShiftAmount,
790                              SMLoc &E) {
791  const AsmToken &Tok = Parser.getTok();
792  if (Tok.isNot(AsmToken::Identifier))
793    return true;
794  StringRef ShiftName = Tok.getString();
795  if (ShiftName == "lsl" || ShiftName == "LSL")
796    St = Lsl;
797  else if (ShiftName == "lsr" || ShiftName == "LSR")
798    St = Lsr;
799  else if (ShiftName == "asr" || ShiftName == "ASR")
800    St = Asr;
801  else if (ShiftName == "ror" || ShiftName == "ROR")
802    St = Ror;
803  else if (ShiftName == "rrx" || ShiftName == "RRX")
804    St = Rrx;
805  else
806    return true;
807  Parser.Lex(); // Eat shift type token.
808
809  // Rrx stands alone.
810  if (St == Rrx)
811    return false;
812
813  // Otherwise, there must be a '#' and a shift amount.
814  const AsmToken &HashTok = Parser.getTok();
815  if (HashTok.isNot(AsmToken::Hash))
816    return Error(HashTok.getLoc(), "'#' expected");
817  Parser.Lex(); // Eat hash token.
818
819  if (getParser().ParseExpression(ShiftAmount))
820    return true;
821
822  return false;
823}
824
825/// Parse a arm instruction operand.  For now this parses the operand regardless
826/// of the mnemonic.
827bool ARMAsmParser::ParseOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands){
828  SMLoc S, E;
829  switch (getLexer().getKind()) {
830  default:
831    Error(Parser.getTok().getLoc(), "unexpected token in operand");
832    return true;
833  case AsmToken::Identifier: {
834    if (!TryParseRegisterWithWriteBack(Operands))
835      return false;
836
837    // This was not a register so parse other operands that start with an
838    // identifier (like labels) as expressions and create them as immediates.
839    const MCExpr *IdVal;
840    S = Parser.getTok().getLoc();
841    if (getParser().ParseExpression(IdVal))
842      return true;
843    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
844    Operands.push_back(ARMOperand::CreateImm(IdVal, S, E));
845    return false;
846  }
847  case AsmToken::LBrac:
848    return ParseMemory(Operands);
849  case AsmToken::LCurly:
850    return ParseRegisterList(Operands);
851  case AsmToken::Hash:
852    // #42 -> immediate.
853    // TODO: ":lower16:" and ":upper16:" modifiers after # before immediate
854    S = Parser.getTok().getLoc();
855    Parser.Lex();
856    const MCExpr *ImmVal;
857    if (getParser().ParseExpression(ImmVal))
858      return true;
859    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
860    Operands.push_back(ARMOperand::CreateImm(ImmVal, S, E));
861    return false;
862  }
863}
864
865/// Parse an arm instruction mnemonic followed by its operands.
866bool ARMAsmParser::ParseInstruction(StringRef Name, SMLoc NameLoc,
867                               SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
868  // Create the leading tokens for the mnemonic, split by '.' characters.
869  size_t Start = 0, Next = Name.find('.');
870  StringRef Head = Name.slice(Start, Next);
871
872  // Determine the predicate, if any.
873  //
874  // FIXME: We need a way to check whether a prefix supports predication,
875  // otherwise we will end up with an ambiguity for instructions that happen to
876  // end with a predicate name.
877  // FIXME: Likewise, some arithmetic instructions have an 's' prefix which
878  // indicates to update the condition codes. Those instructions have an
879  // additional immediate operand which encodes the prefix as reg0 or CPSR.
880  // Just checking for a suffix of 's' definitely creates ambiguities; e.g,
881  // the SMMLS instruction.
882  unsigned CC = StringSwitch<unsigned>(Head.substr(Head.size()-2))
883    .Case("eq", ARMCC::EQ)
884    .Case("ne", ARMCC::NE)
885    .Case("hs", ARMCC::HS)
886    .Case("lo", ARMCC::LO)
887    .Case("mi", ARMCC::MI)
888    .Case("pl", ARMCC::PL)
889    .Case("vs", ARMCC::VS)
890    .Case("vc", ARMCC::VC)
891    .Case("hi", ARMCC::HI)
892    .Case("ls", ARMCC::LS)
893    .Case("ge", ARMCC::GE)
894    .Case("lt", ARMCC::LT)
895    .Case("gt", ARMCC::GT)
896    .Case("le", ARMCC::LE)
897    .Case("al", ARMCC::AL)
898    .Default(~0U);
899
900  if (CC == ~0U ||
901      (CC == ARMCC::LS && (Head == "vmls" || Head == "vnmls"))) {
902    CC = ARMCC::AL;
903  } else {
904    Head = Head.slice(0, Head.size() - 2);
905  }
906
907  Operands.push_back(ARMOperand::CreateToken(Head, NameLoc));
908
909  if (Head != "trap")
910    // FIXME: Should only add this operand for predicated instructions
911    Operands.push_back(ARMOperand::CreateCondCode(ARMCC::CondCodes(CC),
912                                                  NameLoc));
913
914  // Add the remaining tokens in the mnemonic.
915  while (Next != StringRef::npos) {
916    Start = Next;
917    Next = Name.find('.', Start + 1);
918    Head = Name.slice(Start, Next);
919
920    Operands.push_back(ARMOperand::CreateToken(Head, NameLoc));
921  }
922
923  // Read the remaining operands.
924  if (getLexer().isNot(AsmToken::EndOfStatement)) {
925    // Read the first operand.
926    if (ParseOperand(Operands)) {
927      Parser.EatToEndOfStatement();
928      return true;
929    }
930
931    while (getLexer().is(AsmToken::Comma)) {
932      Parser.Lex();  // Eat the comma.
933
934      // Parse and remember the operand.
935      if (ParseOperand(Operands)) {
936        Parser.EatToEndOfStatement();
937        return true;
938      }
939    }
940  }
941
942  if (getLexer().isNot(AsmToken::EndOfStatement)) {
943    Parser.EatToEndOfStatement();
944    return TokError("unexpected token in argument list");
945  }
946
947  Parser.Lex(); // Consume the EndOfStatement
948  return false;
949}
950
951bool ARMAsmParser::
952MatchAndEmitInstruction(SMLoc IDLoc,
953                        SmallVectorImpl<MCParsedAsmOperand*> &Operands,
954                        MCStreamer &Out) {
955  MCInst Inst;
956  unsigned ErrorInfo;
957  switch (MatchInstructionImpl(Operands, Inst, ErrorInfo)) {
958  case Match_Success:
959    Out.EmitInstruction(Inst);
960    return false;
961  case Match_MissingFeature:
962    Error(IDLoc, "instruction requires a CPU feature not currently enabled");
963    return true;
964  case Match_InvalidOperand: {
965    SMLoc ErrorLoc = IDLoc;
966    if (ErrorInfo != ~0U) {
967      if (ErrorInfo >= Operands.size())
968        return Error(IDLoc, "too few operands for instruction");
969
970      ErrorLoc = ((ARMOperand*)Operands[ErrorInfo])->getStartLoc();
971      if (ErrorLoc == SMLoc()) ErrorLoc = IDLoc;
972    }
973
974    return Error(ErrorLoc, "invalid operand for instruction");
975  }
976  case Match_MnemonicFail:
977    return Error(IDLoc, "unrecognized instruction mnemonic");
978  }
979
980  llvm_unreachable("Implement any new match types added!");
981  return true;
982}
983
984/// ParseDirective parses the arm specific directives
985bool ARMAsmParser::ParseDirective(AsmToken DirectiveID) {
986  StringRef IDVal = DirectiveID.getIdentifier();
987  if (IDVal == ".word")
988    return ParseDirectiveWord(4, DirectiveID.getLoc());
989  else if (IDVal == ".thumb")
990    return ParseDirectiveThumb(DirectiveID.getLoc());
991  else if (IDVal == ".thumb_func")
992    return ParseDirectiveThumbFunc(DirectiveID.getLoc());
993  else if (IDVal == ".code")
994    return ParseDirectiveCode(DirectiveID.getLoc());
995  else if (IDVal == ".syntax")
996    return ParseDirectiveSyntax(DirectiveID.getLoc());
997  return true;
998}
999
1000/// ParseDirectiveWord
1001///  ::= .word [ expression (, expression)* ]
1002bool ARMAsmParser::ParseDirectiveWord(unsigned Size, SMLoc L) {
1003  if (getLexer().isNot(AsmToken::EndOfStatement)) {
1004    for (;;) {
1005      const MCExpr *Value;
1006      if (getParser().ParseExpression(Value))
1007        return true;
1008
1009      getParser().getStreamer().EmitValue(Value, Size, 0/*addrspace*/);
1010
1011      if (getLexer().is(AsmToken::EndOfStatement))
1012        break;
1013
1014      // FIXME: Improve diagnostic.
1015      if (getLexer().isNot(AsmToken::Comma))
1016        return Error(L, "unexpected token in directive");
1017      Parser.Lex();
1018    }
1019  }
1020
1021  Parser.Lex();
1022  return false;
1023}
1024
1025/// ParseDirectiveThumb
1026///  ::= .thumb
1027bool ARMAsmParser::ParseDirectiveThumb(SMLoc L) {
1028  if (getLexer().isNot(AsmToken::EndOfStatement))
1029    return Error(L, "unexpected token in directive");
1030  Parser.Lex();
1031
1032  // TODO: set thumb mode
1033  // TODO: tell the MC streamer the mode
1034  // getParser().getStreamer().Emit???();
1035  return false;
1036}
1037
1038/// ParseDirectiveThumbFunc
1039///  ::= .thumbfunc symbol_name
1040bool ARMAsmParser::ParseDirectiveThumbFunc(SMLoc L) {
1041  const AsmToken &Tok = Parser.getTok();
1042  if (Tok.isNot(AsmToken::Identifier) && Tok.isNot(AsmToken::String))
1043    return Error(L, "unexpected token in .thumb_func directive");
1044  StringRef Name = Tok.getString();
1045  Parser.Lex(); // Consume the identifier token.
1046  if (getLexer().isNot(AsmToken::EndOfStatement))
1047    return Error(L, "unexpected token in directive");
1048  Parser.Lex();
1049
1050  // Mark symbol as a thumb symbol.
1051  MCSymbol *Func = getParser().getContext().GetOrCreateSymbol(Name);
1052  getParser().getStreamer().EmitThumbFunc(Func);
1053  return false;
1054}
1055
1056/// ParseDirectiveSyntax
1057///  ::= .syntax unified | divided
1058bool ARMAsmParser::ParseDirectiveSyntax(SMLoc L) {
1059  const AsmToken &Tok = Parser.getTok();
1060  if (Tok.isNot(AsmToken::Identifier))
1061    return Error(L, "unexpected token in .syntax directive");
1062  StringRef Mode = Tok.getString();
1063  if (Mode == "unified" || Mode == "UNIFIED")
1064    Parser.Lex();
1065  else if (Mode == "divided" || Mode == "DIVIDED")
1066    Parser.Lex();
1067  else
1068    return Error(L, "unrecognized syntax mode in .syntax directive");
1069
1070  if (getLexer().isNot(AsmToken::EndOfStatement))
1071    return Error(Parser.getTok().getLoc(), "unexpected token in directive");
1072  Parser.Lex();
1073
1074  // TODO tell the MC streamer the mode
1075  // getParser().getStreamer().Emit???();
1076  return false;
1077}
1078
1079/// ParseDirectiveCode
1080///  ::= .code 16 | 32
1081bool ARMAsmParser::ParseDirectiveCode(SMLoc L) {
1082  const AsmToken &Tok = Parser.getTok();
1083  if (Tok.isNot(AsmToken::Integer))
1084    return Error(L, "unexpected token in .code directive");
1085  int64_t Val = Parser.getTok().getIntVal();
1086  if (Val == 16)
1087    Parser.Lex();
1088  else if (Val == 32)
1089    Parser.Lex();
1090  else
1091    return Error(L, "invalid operand to .code directive");
1092
1093  if (getLexer().isNot(AsmToken::EndOfStatement))
1094    return Error(Parser.getTok().getLoc(), "unexpected token in directive");
1095  Parser.Lex();
1096
1097  if (Val == 16)
1098    getParser().getStreamer().EmitAssemblerFlag(MCAF_Code16);
1099  else
1100    getParser().getStreamer().EmitAssemblerFlag(MCAF_Code32);
1101
1102  return false;
1103}
1104
1105extern "C" void LLVMInitializeARMAsmLexer();
1106
1107/// Force static initialization.
1108extern "C" void LLVMInitializeARMAsmParser() {
1109  RegisterAsmParser<ARMAsmParser> X(TheARMTarget);
1110  RegisterAsmParser<ARMAsmParser> Y(TheThumbTarget);
1111  LLVMInitializeARMAsmLexer();
1112}
1113
1114#define GET_REGISTER_MATCHER
1115#define GET_MATCHER_IMPLEMENTATION
1116#include "ARMGenAsmMatcher.inc"
1117