ARMAsmParser.cpp revision 8ab1112bdc30b8675bb12431d8b5b270da42f1b5 
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/// Shift types used 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    /// Combined record 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 isMemModeRegThumb() const {
240    if (!isMemory() || (!Mem.OffsetIsReg && !Mem.Offset) || Mem.Writeback)
241      return false;
242    return !Mem.Offset || !isa<MCConstantExpr>(Mem.Offset);
243  }
244  bool isMemModeImmThumb() const {
245    if (!isMemory() || (!Mem.OffsetIsReg && !Mem.Offset) || Mem.Writeback)
246      return false;
247
248    if (!Mem.Offset) return false;
249
250    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
251    if (!CE) return false;
252
253    // The offset must be a multiple of 4 in the range 0-124.
254    uint64_t Value = CE->getValue();
255    return ((Value & 0x3) == 0 && Value <= 124);
256  }
257
258  void addExpr(MCInst &Inst, const MCExpr *Expr) const {
259    // Add as immediates when possible.  Null MCExpr = 0.
260    if (Expr == 0)
261      Inst.addOperand(MCOperand::CreateImm(0));
262    else if (const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Expr))
263      Inst.addOperand(MCOperand::CreateImm(CE->getValue()));
264    else
265      Inst.addOperand(MCOperand::CreateExpr(Expr));
266  }
267
268  void addCondCodeOperands(MCInst &Inst, unsigned N) const {
269    assert(N == 2 && "Invalid number of operands!");
270    Inst.addOperand(MCOperand::CreateImm(unsigned(getCondCode())));
271    unsigned RegNum = getCondCode() == ARMCC::AL ? 0: ARM::CPSR;
272    Inst.addOperand(MCOperand::CreateReg(RegNum));
273  }
274
275  void addCCOutOperands(MCInst &Inst, unsigned N) const {
276    assert(N == 1 && "Invalid number of operands!");
277    Inst.addOperand(MCOperand::CreateReg(getReg()));
278  }
279
280  void addRegOperands(MCInst &Inst, unsigned N) const {
281    assert(N == 1 && "Invalid number of operands!");
282    Inst.addOperand(MCOperand::CreateReg(getReg()));
283  }
284
285  void addRegListOperands(MCInst &Inst, unsigned N) const {
286    assert(N == 1 && "Invalid number of operands!");
287    const SmallVectorImpl<unsigned> &RegList = getRegList();
288    for (SmallVectorImpl<unsigned>::const_iterator
289           I = RegList.begin(), E = RegList.end(); I != E; ++I)
290      Inst.addOperand(MCOperand::CreateReg(*I));
291  }
292
293  void addDPRRegListOperands(MCInst &Inst, unsigned N) const {
294    addRegListOperands(Inst, N);
295  }
296
297  void addSPRRegListOperands(MCInst &Inst, unsigned N) const {
298    addRegListOperands(Inst, N);
299  }
300
301  void addImmOperands(MCInst &Inst, unsigned N) const {
302    assert(N == 1 && "Invalid number of operands!");
303    addExpr(Inst, getImm());
304  }
305
306  void addMemMode5Operands(MCInst &Inst, unsigned N) const {
307    assert(N == 2 && isMemMode5() && "Invalid number of operands!");
308
309    Inst.addOperand(MCOperand::CreateReg(Mem.BaseRegNum));
310    assert(!Mem.OffsetIsReg && "Invalid mode 5 operand");
311
312    // FIXME: #-0 is encoded differently than #0. Does the parser preserve
313    // the difference?
314    if (Mem.Offset) {
315      const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
316      assert(CE && "Non-constant mode 5 offset operand!");
317
318      // The MCInst offset operand doesn't include the low two bits (like
319      // the instruction encoding).
320      int64_t Offset = CE->getValue() / 4;
321      if (Offset >= 0)
322        Inst.addOperand(MCOperand::CreateImm(ARM_AM::getAM5Opc(ARM_AM::add,
323                                                               Offset)));
324      else
325        Inst.addOperand(MCOperand::CreateImm(ARM_AM::getAM5Opc(ARM_AM::sub,
326                                                               -Offset)));
327    } else {
328      Inst.addOperand(MCOperand::CreateImm(0));
329    }
330  }
331
332  void addMemModeRegThumbOperands(MCInst &Inst, unsigned N) const {
333    assert(N == 2 && isMemModeRegThumb() && "Invalid number of operands!");
334    Inst.addOperand(MCOperand::CreateReg(Mem.BaseRegNum));
335    Inst.addOperand(MCOperand::CreateReg(Mem.OffsetRegNum));
336  }
337
338  void addMemModeImmThumbOperands(MCInst &Inst, unsigned N) const {
339    assert(N == 2 && isMemModeImmThumb() && "Invalid number of operands!");
340    Inst.addOperand(MCOperand::CreateReg(Mem.BaseRegNum));
341    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Mem.Offset);
342    assert(CE && "Non-constant mode offset operand!");
343    Inst.addOperand(MCOperand::CreateImm(CE->getValue()));
344  }
345
346  virtual void dump(raw_ostream &OS) const;
347
348  static ARMOperand *CreateCondCode(ARMCC::CondCodes CC, SMLoc S) {
349    ARMOperand *Op = new ARMOperand(CondCode);
350    Op->CC.Val = CC;
351    Op->StartLoc = S;
352    Op->EndLoc = S;
353    return Op;
354  }
355
356  static ARMOperand *CreateCCOut(unsigned RegNum, SMLoc S) {
357    ARMOperand *Op = new ARMOperand(CCOut);
358    Op->Reg.RegNum = RegNum;
359    Op->StartLoc = S;
360    Op->EndLoc = S;
361    return Op;
362  }
363
364  static ARMOperand *CreateToken(StringRef Str, SMLoc S) {
365    ARMOperand *Op = new ARMOperand(Token);
366    Op->Tok.Data = Str.data();
367    Op->Tok.Length = Str.size();
368    Op->StartLoc = S;
369    Op->EndLoc = S;
370    return Op;
371  }
372
373  static ARMOperand *CreateReg(unsigned RegNum, SMLoc S, SMLoc E) {
374    ARMOperand *Op = new ARMOperand(Register);
375    Op->Reg.RegNum = RegNum;
376    Op->StartLoc = S;
377    Op->EndLoc = E;
378    return Op;
379  }
380
381  static ARMOperand *
382  CreateRegList(const SmallVectorImpl<std::pair<unsigned, SMLoc> > &Regs,
383                SMLoc StartLoc, SMLoc EndLoc) {
384    KindTy Kind = RegisterList;
385
386    if (ARM::DPRRegClass.contains(Regs.front().first))
387      Kind = DPRRegisterList;
388    else if (ARM::SPRRegClass.contains(Regs.front().first))
389      Kind = SPRRegisterList;
390
391    ARMOperand *Op = new ARMOperand(Kind);
392    for (SmallVectorImpl<std::pair<unsigned, SMLoc> >::const_iterator
393           I = Regs.begin(), E = Regs.end(); I != E; ++I)
394      Op->Registers.push_back(I->first);
395    array_pod_sort(Op->Registers.begin(), Op->Registers.end());
396    Op->StartLoc = StartLoc;
397    Op->EndLoc = EndLoc;
398    return Op;
399  }
400
401  static ARMOperand *CreateImm(const MCExpr *Val, SMLoc S, SMLoc E) {
402    ARMOperand *Op = new ARMOperand(Immediate);
403    Op->Imm.Val = Val;
404    Op->StartLoc = S;
405    Op->EndLoc = E;
406    return Op;
407  }
408
409  static ARMOperand *CreateMem(unsigned BaseRegNum, bool OffsetIsReg,
410                               const MCExpr *Offset, unsigned OffsetRegNum,
411                               bool OffsetRegShifted, enum ShiftType ShiftType,
412                               const MCExpr *ShiftAmount, bool Preindexed,
413                               bool Postindexed, bool Negative, bool Writeback,
414                               SMLoc S, SMLoc E) {
415    ARMOperand *Op = new ARMOperand(Memory);
416    Op->Mem.BaseRegNum = BaseRegNum;
417    Op->Mem.OffsetIsReg = OffsetIsReg;
418    Op->Mem.Offset = Offset;
419    Op->Mem.OffsetRegNum = OffsetRegNum;
420    Op->Mem.OffsetRegShifted = OffsetRegShifted;
421    Op->Mem.ShiftType = ShiftType;
422    Op->Mem.ShiftAmount = ShiftAmount;
423    Op->Mem.Preindexed = Preindexed;
424    Op->Mem.Postindexed = Postindexed;
425    Op->Mem.Negative = Negative;
426    Op->Mem.Writeback = Writeback;
427
428    Op->StartLoc = S;
429    Op->EndLoc = E;
430    return Op;
431  }
432};
433
434} // end anonymous namespace.
435
436void ARMOperand::dump(raw_ostream &OS) const {
437  switch (Kind) {
438  case CondCode:
439    OS << "<ARMCC::" << ARMCondCodeToString(getCondCode()) << ">";
440    break;
441  case CCOut:
442    OS << "<ccout " << getReg() << ">";
443    break;
444  case Immediate:
445    getImm()->print(OS);
446    break;
447  case Memory:
448    OS << "<memory>";
449    break;
450  case Register:
451    OS << "<register " << getReg() << ">";
452    break;
453  case RegisterList:
454  case DPRRegisterList:
455  case SPRRegisterList: {
456    OS << "<register_list ";
457
458    const SmallVectorImpl<unsigned> &RegList = getRegList();
459    for (SmallVectorImpl<unsigned>::const_iterator
460           I = RegList.begin(), E = RegList.end(); I != E; ) {
461      OS << *I;
462      if (++I < E) OS << ", ";
463    }
464
465    OS << ">";
466    break;
467  }
468  case Token:
469    OS << "'" << getToken() << "'";
470    break;
471  }
472}
473
474/// @name Auto-generated Match Functions
475/// {
476
477static unsigned MatchRegisterName(StringRef Name);
478
479/// }
480
481/// Try to parse a register name.  The token must be an Identifier when called,
482/// and if it is a register name the token is eaten and the register number is
483/// returned.  Otherwise return -1.
484///
485int ARMAsmParser::TryParseRegister() {
486  const AsmToken &Tok = Parser.getTok();
487  assert(Tok.is(AsmToken::Identifier) && "Token is not an Identifier");
488
489  // FIXME: Validate register for the current architecture; we have to do
490  // validation later, so maybe there is no need for this here.
491  unsigned RegNum = MatchRegisterName(Tok.getString());
492  if (RegNum == 0)
493    return -1;
494  Parser.Lex(); // Eat identifier token.
495  return RegNum;
496}
497
498
499/// Try to parse a register name.  The token must be an Identifier when called.
500/// If it's a register, an AsmOperand is created. Another AsmOperand is created
501/// if there is a "writeback". 'true' if it's not a register.
502///
503/// TODO this is likely to change to allow different register types and or to
504/// parse for a specific register type.
505bool ARMAsmParser::
506TryParseRegisterWithWriteBack(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
507  SMLoc S = Parser.getTok().getLoc();
508  int RegNo = TryParseRegister();
509  if (RegNo == -1)
510    return true;
511
512  Operands.push_back(ARMOperand::CreateReg(RegNo, S, Parser.getTok().getLoc()));
513
514  const AsmToken &ExclaimTok = Parser.getTok();
515  if (ExclaimTok.is(AsmToken::Exclaim)) {
516    Operands.push_back(ARMOperand::CreateToken(ExclaimTok.getString(),
517                                               ExclaimTok.getLoc()));
518    Parser.Lex(); // Eat exclaim token
519  }
520
521  return false;
522}
523
524/// Parse a register list, return it if successful else return null.  The first
525/// token must be a '{' when called.
526bool ARMAsmParser::
527ParseRegisterList(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
528  assert(Parser.getTok().is(AsmToken::LCurly) &&
529         "Token is not a Left Curly Brace");
530  SMLoc S = Parser.getTok().getLoc();
531
532  // Read the rest of the registers in the list.
533  unsigned PrevRegNum = 0;
534  SmallVector<std::pair<unsigned, SMLoc>, 32> Registers;
535
536  do {
537    bool IsRange = Parser.getTok().is(AsmToken::Minus);
538    Parser.Lex(); // Eat non-identifier token.
539
540    const AsmToken &RegTok = Parser.getTok();
541    SMLoc RegLoc = RegTok.getLoc();
542    if (RegTok.isNot(AsmToken::Identifier)) {
543      Error(RegLoc, "register expected");
544      return true;
545    }
546
547    int RegNum = TryParseRegister();
548    if (RegNum == -1) {
549      Error(RegLoc, "register expected");
550      return true;
551    }
552
553    if (IsRange) {
554      int Reg = PrevRegNum;
555      do {
556        ++Reg;
557        Registers.push_back(std::make_pair(Reg, RegLoc));
558      } while (Reg != RegNum);
559    } else {
560      Registers.push_back(std::make_pair(RegNum, RegLoc));
561    }
562
563    PrevRegNum = RegNum;
564  } while (Parser.getTok().is(AsmToken::Comma) ||
565           Parser.getTok().is(AsmToken::Minus));
566
567  // Process the right curly brace of the list.
568  const AsmToken &RCurlyTok = Parser.getTok();
569  if (RCurlyTok.isNot(AsmToken::RCurly)) {
570    Error(RCurlyTok.getLoc(), "'}' expected");
571    return true;
572  }
573
574  SMLoc E = RCurlyTok.getLoc();
575  Parser.Lex(); // Eat right curly brace token.
576
577  // Verify the register list.
578  SmallVectorImpl<std::pair<unsigned, SMLoc> >::const_iterator
579    RI = Registers.begin(), RE = Registers.end();
580
581  DenseMap<unsigned, bool> RegMap;
582  RegMap[RI->first] = true;
583
584  unsigned HighRegNum = RI->first;
585  bool EmittedWarning = false;
586
587  for (++RI; RI != RE; ++RI) {
588    const std::pair<unsigned, SMLoc> &RegInfo = *RI;
589    unsigned Reg = RegInfo.first;
590
591    if (RegMap[Reg]) {
592      Error(RegInfo.second, "register duplicated in register list");
593      return true;
594    }
595
596    if (!EmittedWarning && Reg < HighRegNum)
597      Warning(RegInfo.second,
598              "register not in ascending order in register list");
599
600    RegMap[Reg] = true;
601    HighRegNum = std::max(Reg, HighRegNum);
602  }
603
604  Operands.push_back(ARMOperand::CreateRegList(Registers, S, E));
605  return false;
606}
607
608/// Parse an ARM memory expression, return false if successful else return true
609/// or an error.  The first token must be a '[' when called.
610///
611/// TODO Only preindexing and postindexing addressing are started, unindexed
612/// with option, etc are still to do.
613bool ARMAsmParser::
614ParseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
615  SMLoc S, E;
616  assert(Parser.getTok().is(AsmToken::LBrac) &&
617         "Token is not a Left Bracket");
618  S = Parser.getTok().getLoc();
619  Parser.Lex(); // Eat left bracket token.
620
621  const AsmToken &BaseRegTok = Parser.getTok();
622  if (BaseRegTok.isNot(AsmToken::Identifier)) {
623    Error(BaseRegTok.getLoc(), "register expected");
624    return true;
625  }
626  int BaseRegNum = TryParseRegister();
627  if (BaseRegNum == -1) {
628    Error(BaseRegTok.getLoc(), "register expected");
629    return true;
630  }
631
632  bool Preindexed = false;
633  bool Postindexed = false;
634  bool OffsetIsReg = false;
635  bool Negative = false;
636  bool Writeback = false;
637
638  // First look for preindexed address forms, that is after the "[Rn" we now
639  // have to see if the next token is a comma.
640  const AsmToken &Tok = Parser.getTok();
641  if (Tok.is(AsmToken::Comma)) {
642    Preindexed = true;
643    Parser.Lex(); // Eat comma token.
644    int OffsetRegNum;
645    bool OffsetRegShifted;
646    enum ShiftType ShiftType;
647    const MCExpr *ShiftAmount = 0;
648    const MCExpr *Offset = 0;
649    if (ParseMemoryOffsetReg(Negative, OffsetRegShifted, ShiftType, ShiftAmount,
650                             Offset, OffsetIsReg, OffsetRegNum, E))
651      return true;
652    const AsmToken &RBracTok = Parser.getTok();
653    if (RBracTok.isNot(AsmToken::RBrac)) {
654      Error(RBracTok.getLoc(), "']' expected");
655      return true;
656    }
657    E = RBracTok.getLoc();
658    Parser.Lex(); // Eat right bracket token.
659
660
661    const AsmToken &ExclaimTok = Parser.getTok();
662    ARMOperand *WBOp = 0;
663    if (ExclaimTok.is(AsmToken::Exclaim)) {
664      WBOp = ARMOperand::CreateToken(ExclaimTok.getString(),
665                                     ExclaimTok.getLoc());
666      Writeback = true;
667      Parser.Lex(); // Eat exclaim token
668    }
669
670    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset,
671                                             OffsetRegNum, OffsetRegShifted,
672                                             ShiftType, ShiftAmount, Preindexed,
673                                             Postindexed, Negative, Writeback,
674                                             S, E));
675    if (WBOp)
676      Operands.push_back(WBOp);
677
678    return false;
679  }
680  // The "[Rn" we have so far was not followed by a comma.
681  else if (Tok.is(AsmToken::RBrac)) {
682    // If there's anything other than the right brace, this is a post indexing
683    // addressing form.
684    E = Tok.getLoc();
685    Parser.Lex(); // Eat right bracket token.
686
687    int OffsetRegNum = 0;
688    bool OffsetRegShifted = false;
689    enum ShiftType ShiftType = Lsl;
690    const MCExpr *ShiftAmount = 0;
691    const MCExpr *Offset = 0;
692
693    const AsmToken &NextTok = Parser.getTok();
694
695    if (NextTok.isNot(AsmToken::EndOfStatement)) {
696      Postindexed = true;
697      Writeback = true;
698
699      if (NextTok.isNot(AsmToken::Comma)) {
700        Error(NextTok.getLoc(), "',' expected");
701        return true;
702      }
703
704      Parser.Lex(); // Eat comma token.
705
706      if (ParseMemoryOffsetReg(Negative, OffsetRegShifted, ShiftType,
707                               ShiftAmount, Offset, OffsetIsReg, OffsetRegNum,
708                               E))
709        return true;
710    }
711
712    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset,
713                                             OffsetRegNum, OffsetRegShifted,
714                                             ShiftType, ShiftAmount, Preindexed,
715                                             Postindexed, Negative, Writeback,
716                                             S, E));
717    return false;
718  }
719
720  return true;
721}
722
723/// Parse the offset of a memory operand after we have seen "[Rn," or "[Rn],"
724/// we will parse the following (were +/- means that a plus or minus is
725/// optional):
726///   +/-Rm
727///   +/-Rm, shift
728///   #offset
729/// we return false on success or an error otherwise.
730bool ARMAsmParser::ParseMemoryOffsetReg(bool &Negative,
731                                        bool &OffsetRegShifted,
732                                        enum ShiftType &ShiftType,
733                                        const MCExpr *&ShiftAmount,
734                                        const MCExpr *&Offset,
735                                        bool &OffsetIsReg,
736                                        int &OffsetRegNum,
737                                        SMLoc &E) {
738  Negative = false;
739  OffsetRegShifted = false;
740  OffsetIsReg = false;
741  OffsetRegNum = -1;
742  const AsmToken &NextTok = Parser.getTok();
743  E = NextTok.getLoc();
744  if (NextTok.is(AsmToken::Plus))
745    Parser.Lex(); // Eat plus token.
746  else if (NextTok.is(AsmToken::Minus)) {
747    Negative = true;
748    Parser.Lex(); // Eat minus token
749  }
750  // See if there is a register following the "[Rn," or "[Rn]," we have so far.
751  const AsmToken &OffsetRegTok = Parser.getTok();
752  if (OffsetRegTok.is(AsmToken::Identifier)) {
753    SMLoc CurLoc = OffsetRegTok.getLoc();
754    OffsetRegNum = TryParseRegister();
755    if (OffsetRegNum != -1) {
756      OffsetIsReg = true;
757      E = CurLoc;
758    }
759  }
760
761  // If we parsed a register as the offset then there can be a shift after that.
762  if (OffsetRegNum != -1) {
763    // Look for a comma then a shift
764    const AsmToken &Tok = Parser.getTok();
765    if (Tok.is(AsmToken::Comma)) {
766      Parser.Lex(); // Eat comma token.
767
768      const AsmToken &Tok = Parser.getTok();
769      if (ParseShift(ShiftType, ShiftAmount, E))
770        return Error(Tok.getLoc(), "shift expected");
771      OffsetRegShifted = true;
772    }
773  }
774  else { // the "[Rn," or "[Rn,]" we have so far was not followed by "Rm"
775    // Look for #offset following the "[Rn," or "[Rn],"
776    const AsmToken &HashTok = Parser.getTok();
777    if (HashTok.isNot(AsmToken::Hash))
778      return Error(HashTok.getLoc(), "'#' expected");
779
780    Parser.Lex(); // Eat hash token.
781
782    if (getParser().ParseExpression(Offset))
783     return true;
784    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
785  }
786  return false;
787}
788
789/// ParseShift as one of these two:
790///   ( lsl | lsr | asr | ror ) , # shift_amount
791///   rrx
792/// and returns true if it parses a shift otherwise it returns false.
793bool ARMAsmParser::ParseShift(ShiftType &St, const MCExpr *&ShiftAmount,
794                              SMLoc &E) {
795  const AsmToken &Tok = Parser.getTok();
796  if (Tok.isNot(AsmToken::Identifier))
797    return true;
798  StringRef ShiftName = Tok.getString();
799  if (ShiftName == "lsl" || ShiftName == "LSL")
800    St = Lsl;
801  else if (ShiftName == "lsr" || ShiftName == "LSR")
802    St = Lsr;
803  else if (ShiftName == "asr" || ShiftName == "ASR")
804    St = Asr;
805  else if (ShiftName == "ror" || ShiftName == "ROR")
806    St = Ror;
807  else if (ShiftName == "rrx" || ShiftName == "RRX")
808    St = Rrx;
809  else
810    return true;
811  Parser.Lex(); // Eat shift type token.
812
813  // Rrx stands alone.
814  if (St == Rrx)
815    return false;
816
817  // Otherwise, there must be a '#' and a shift amount.
818  const AsmToken &HashTok = Parser.getTok();
819  if (HashTok.isNot(AsmToken::Hash))
820    return Error(HashTok.getLoc(), "'#' expected");
821  Parser.Lex(); // Eat hash token.
822
823  if (getParser().ParseExpression(ShiftAmount))
824    return true;
825
826  return false;
827}
828
829/// Parse a arm instruction operand.  For now this parses the operand regardless
830/// of the mnemonic.
831bool ARMAsmParser::ParseOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands){
832  SMLoc S, E;
833  switch (getLexer().getKind()) {
834  default:
835    Error(Parser.getTok().getLoc(), "unexpected token in operand");
836    return true;
837  case AsmToken::Identifier: {
838    if (!TryParseRegisterWithWriteBack(Operands))
839      return false;
840
841    // This was not a register so parse other operands that start with an
842    // identifier (like labels) as expressions and create them as immediates.
843    const MCExpr *IdVal;
844    S = Parser.getTok().getLoc();
845    if (getParser().ParseExpression(IdVal))
846      return true;
847    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
848    Operands.push_back(ARMOperand::CreateImm(IdVal, S, E));
849    return false;
850  }
851  case AsmToken::LBrac:
852    return ParseMemory(Operands);
853  case AsmToken::LCurly:
854    return ParseRegisterList(Operands);
855  case AsmToken::Hash:
856    // #42 -> immediate.
857    // TODO: ":lower16:" and ":upper16:" modifiers after # before immediate
858    S = Parser.getTok().getLoc();
859    Parser.Lex();
860    const MCExpr *ImmVal;
861    if (getParser().ParseExpression(ImmVal))
862      return true;
863    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
864    Operands.push_back(ARMOperand::CreateImm(ImmVal, S, E));
865    return false;
866  }
867}
868
869// FIXME: Would be nice to autogen this.
870static unsigned SplitMnemonicAndCC(StringRef &Mnemonic) {
871  // Ignore some mnemonics we know aren't predicated forms.
872  if (Mnemonic == "teq" || Mnemonic == "vceq" ||
873      Mnemonic == "movs" ||
874      Mnemonic == "svc" ||
875      (Mnemonic == "mls" || Mnemonic == "smmls" || Mnemonic == "vcls" ||
876       Mnemonic == "vmls" || Mnemonic == "vnmls") ||
877      Mnemonic == "vacge" || Mnemonic == "vcge" ||
878      Mnemonic == "vclt" ||
879      Mnemonic == "vacgt" || Mnemonic == "vcgt" ||
880      Mnemonic == "vcle" ||
881      (Mnemonic == "smlal" || Mnemonic == "umaal" || Mnemonic == "umlal" ||
882       Mnemonic == "vabal" || Mnemonic == "vmlal" || Mnemonic == "vpadal" ||
883       Mnemonic == "vqdmlal"))
884    return ARMCC::AL;
885
886  // Otherwise, determine the predicate.
887  //
888  // FIXME: We need a way to check whether a prefix supports predication,
889  // otherwise we will end up with an ambiguity for instructions that happen to
890  // end with a predicate name.
891  unsigned CC = StringSwitch<unsigned>(Mnemonic.substr(Mnemonic.size()-2))
892    .Case("eq", ARMCC::EQ)
893    .Case("ne", ARMCC::NE)
894    .Case("hs", ARMCC::HS)
895    .Case("lo", ARMCC::LO)
896    .Case("mi", ARMCC::MI)
897    .Case("pl", ARMCC::PL)
898    .Case("vs", ARMCC::VS)
899    .Case("vc", ARMCC::VC)
900    .Case("hi", ARMCC::HI)
901    .Case("ls", ARMCC::LS)
902    .Case("ge", ARMCC::GE)
903    .Case("lt", ARMCC::LT)
904    .Case("gt", ARMCC::GT)
905    .Case("le", ARMCC::LE)
906    .Case("al", ARMCC::AL)
907    .Default(~0U);
908  if (CC != ~0U) {
909    Mnemonic = Mnemonic.slice(0, Mnemonic.size() - 2);
910    return CC;
911  }
912
913  return ARMCC::AL;
914}
915
916/// Parse an arm instruction mnemonic followed by its operands.
917bool ARMAsmParser::ParseInstruction(StringRef Name, SMLoc NameLoc,
918                               SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
919  // Create the leading tokens for the mnemonic, split by '.' characters.
920  size_t Start = 0, Next = Name.find('.');
921  StringRef Head = Name.slice(Start, Next);
922
923  // Determine the predicate, if any.
924  unsigned CC = SplitMnemonicAndCC(Head);
925
926  Operands.push_back(ARMOperand::CreateToken(Head, NameLoc));
927
928  // FIXME: Should only add this operand for predicated instructions
929  if (Head != "trap") {
930    Operands.push_back(ARMOperand::CreateCondCode(ARMCC::CondCodes(CC),
931                                                  NameLoc));
932  }
933
934  // Add the remaining tokens in the mnemonic.
935  while (Next != StringRef::npos) {
936    Start = Next;
937    Next = Name.find('.', Start + 1);
938    Head = Name.slice(Start, Next);
939
940    Operands.push_back(ARMOperand::CreateToken(Head, NameLoc));
941  }
942
943  // Read the remaining operands.
944  if (getLexer().isNot(AsmToken::EndOfStatement)) {
945    // Read the first operand.
946    if (ParseOperand(Operands)) {
947      Parser.EatToEndOfStatement();
948      return true;
949    }
950
951    while (getLexer().is(AsmToken::Comma)) {
952      Parser.Lex();  // Eat the comma.
953
954      // Parse and remember the operand.
955      if (ParseOperand(Operands)) {
956        Parser.EatToEndOfStatement();
957        return true;
958      }
959    }
960  }
961
962  if (getLexer().isNot(AsmToken::EndOfStatement)) {
963    Parser.EatToEndOfStatement();
964    return TokError("unexpected token in argument list");
965  }
966
967  Parser.Lex(); // Consume the EndOfStatement
968  return false;
969}
970
971bool ARMAsmParser::
972MatchAndEmitInstruction(SMLoc IDLoc,
973                        SmallVectorImpl<MCParsedAsmOperand*> &Operands,
974                        MCStreamer &Out) {
975  MCInst Inst;
976  unsigned ErrorInfo;
977  MatchResultTy MatchResult, MatchResult2;
978  MatchResult = MatchInstructionImpl(Operands, Inst, ErrorInfo);
979  if (MatchResult != Match_Success) {
980    // If we get a Match_InvalidOperand it might be some arithmetic instruction
981    // that does not update the condition codes.  So try adding a CCOut operand
982    // with a value of reg0.
983    if (MatchResult == Match_InvalidOperand) {
984      Operands.insert(Operands.begin() + 1,
985                      ARMOperand::CreateCCOut(0,
986                                  ((ARMOperand*)Operands[0])->getStartLoc()));
987      MatchResult2 = MatchInstructionImpl(Operands, Inst, ErrorInfo);
988      if (MatchResult2 == Match_Success)
989        MatchResult = Match_Success;
990      else {
991        ARMOperand *CCOut = ((ARMOperand*)Operands[1]);
992        Operands.erase(Operands.begin() + 1);
993        delete CCOut;
994      }
995    }
996    // If we get a Match_MnemonicFail it might be some arithmetic instruction
997    // that updates the condition codes if it ends in 's'.  So see if the
998    // mnemonic ends in 's' and if so try removing the 's' and adding a CCOut
999    // operand with a value of CPSR.
1000    else if(MatchResult == Match_MnemonicFail) {
1001      // Get the instruction mnemonic, which is the first token.
1002      StringRef Mnemonic = ((ARMOperand*)Operands[0])->getToken();
1003      if (Mnemonic.substr(Mnemonic.size()-1) == "s") {
1004        // removed the 's' from the mnemonic for matching.
1005        StringRef MnemonicNoS = Mnemonic.slice(0, Mnemonic.size() - 1);
1006        SMLoc NameLoc = ((ARMOperand*)Operands[0])->getStartLoc();
1007        ARMOperand *OldMnemonic = ((ARMOperand*)Operands[0]);
1008        Operands.erase(Operands.begin());
1009        delete OldMnemonic;
1010        Operands.insert(Operands.begin(),
1011                        ARMOperand::CreateToken(MnemonicNoS, NameLoc));
1012        Operands.insert(Operands.begin() + 1,
1013                        ARMOperand::CreateCCOut(ARM::CPSR, NameLoc));
1014        MatchResult2 = MatchInstructionImpl(Operands, Inst, ErrorInfo);
1015        if (MatchResult2 == Match_Success)
1016          MatchResult = Match_Success;
1017        else {
1018          ARMOperand *OldMnemonic = ((ARMOperand*)Operands[0]);
1019          Operands.erase(Operands.begin());
1020          delete OldMnemonic;
1021          Operands.insert(Operands.begin(),
1022                          ARMOperand::CreateToken(Mnemonic, NameLoc));
1023          ARMOperand *CCOut = ((ARMOperand*)Operands[1]);
1024          Operands.erase(Operands.begin() + 1);
1025          delete CCOut;
1026        }
1027      }
1028    }
1029  }
1030  switch (MatchResult) {
1031  case Match_Success:
1032    Out.EmitInstruction(Inst);
1033    return false;
1034  case Match_MissingFeature:
1035    Error(IDLoc, "instruction requires a CPU feature not currently enabled");
1036    return true;
1037  case Match_InvalidOperand: {
1038    SMLoc ErrorLoc = IDLoc;
1039    if (ErrorInfo != ~0U) {
1040      if (ErrorInfo >= Operands.size())
1041        return Error(IDLoc, "too few operands for instruction");
1042
1043      ErrorLoc = ((ARMOperand*)Operands[ErrorInfo])->getStartLoc();
1044      if (ErrorLoc == SMLoc()) ErrorLoc = IDLoc;
1045    }
1046
1047    return Error(ErrorLoc, "invalid operand for instruction");
1048  }
1049  case Match_MnemonicFail:
1050    return Error(IDLoc, "unrecognized instruction mnemonic");
1051  }
1052
1053  llvm_unreachable("Implement any new match types added!");
1054  return true;
1055}
1056
1057/// ParseDirective parses the arm specific directives
1058bool ARMAsmParser::ParseDirective(AsmToken DirectiveID) {
1059  StringRef IDVal = DirectiveID.getIdentifier();
1060  if (IDVal == ".word")
1061    return ParseDirectiveWord(4, DirectiveID.getLoc());
1062  else if (IDVal == ".thumb")
1063    return ParseDirectiveThumb(DirectiveID.getLoc());
1064  else if (IDVal == ".thumb_func")
1065    return ParseDirectiveThumbFunc(DirectiveID.getLoc());
1066  else if (IDVal == ".code")
1067    return ParseDirectiveCode(DirectiveID.getLoc());
1068  else if (IDVal == ".syntax")
1069    return ParseDirectiveSyntax(DirectiveID.getLoc());
1070  return true;
1071}
1072
1073/// ParseDirectiveWord
1074///  ::= .word [ expression (, expression)* ]
1075bool ARMAsmParser::ParseDirectiveWord(unsigned Size, SMLoc L) {
1076  if (getLexer().isNot(AsmToken::EndOfStatement)) {
1077    for (;;) {
1078      const MCExpr *Value;
1079      if (getParser().ParseExpression(Value))
1080        return true;
1081
1082      getParser().getStreamer().EmitValue(Value, Size, 0/*addrspace*/);
1083
1084      if (getLexer().is(AsmToken::EndOfStatement))
1085        break;
1086
1087      // FIXME: Improve diagnostic.
1088      if (getLexer().isNot(AsmToken::Comma))
1089        return Error(L, "unexpected token in directive");
1090      Parser.Lex();
1091    }
1092  }
1093
1094  Parser.Lex();
1095  return false;
1096}
1097
1098/// ParseDirectiveThumb
1099///  ::= .thumb
1100bool ARMAsmParser::ParseDirectiveThumb(SMLoc L) {
1101  if (getLexer().isNot(AsmToken::EndOfStatement))
1102    return Error(L, "unexpected token in directive");
1103  Parser.Lex();
1104
1105  // TODO: set thumb mode
1106  // TODO: tell the MC streamer the mode
1107  // getParser().getStreamer().Emit???();
1108  return false;
1109}
1110
1111/// ParseDirectiveThumbFunc
1112///  ::= .thumbfunc symbol_name
1113bool ARMAsmParser::ParseDirectiveThumbFunc(SMLoc L) {
1114  const AsmToken &Tok = Parser.getTok();
1115  if (Tok.isNot(AsmToken::Identifier) && Tok.isNot(AsmToken::String))
1116    return Error(L, "unexpected token in .thumb_func directive");
1117  StringRef Name = Tok.getString();
1118  Parser.Lex(); // Consume the identifier token.
1119  if (getLexer().isNot(AsmToken::EndOfStatement))
1120    return Error(L, "unexpected token in directive");
1121  Parser.Lex();
1122
1123  // Mark symbol as a thumb symbol.
1124  MCSymbol *Func = getParser().getContext().GetOrCreateSymbol(Name);
1125  getParser().getStreamer().EmitThumbFunc(Func);
1126  return false;
1127}
1128
1129/// ParseDirectiveSyntax
1130///  ::= .syntax unified | divided
1131bool ARMAsmParser::ParseDirectiveSyntax(SMLoc L) {
1132  const AsmToken &Tok = Parser.getTok();
1133  if (Tok.isNot(AsmToken::Identifier))
1134    return Error(L, "unexpected token in .syntax directive");
1135  StringRef Mode = Tok.getString();
1136  if (Mode == "unified" || Mode == "UNIFIED")
1137    Parser.Lex();
1138  else if (Mode == "divided" || Mode == "DIVIDED")
1139    Parser.Lex();
1140  else
1141    return Error(L, "unrecognized syntax mode in .syntax directive");
1142
1143  if (getLexer().isNot(AsmToken::EndOfStatement))
1144    return Error(Parser.getTok().getLoc(), "unexpected token in directive");
1145  Parser.Lex();
1146
1147  // TODO tell the MC streamer the mode
1148  // getParser().getStreamer().Emit???();
1149  return false;
1150}
1151
1152/// ParseDirectiveCode
1153///  ::= .code 16 | 32
1154bool ARMAsmParser::ParseDirectiveCode(SMLoc L) {
1155  const AsmToken &Tok = Parser.getTok();
1156  if (Tok.isNot(AsmToken::Integer))
1157    return Error(L, "unexpected token in .code directive");
1158  int64_t Val = Parser.getTok().getIntVal();
1159  if (Val == 16)
1160    Parser.Lex();
1161  else if (Val == 32)
1162    Parser.Lex();
1163  else
1164    return Error(L, "invalid operand to .code directive");
1165
1166  if (getLexer().isNot(AsmToken::EndOfStatement))
1167    return Error(Parser.getTok().getLoc(), "unexpected token in directive");
1168  Parser.Lex();
1169
1170  if (Val == 16)
1171    getParser().getStreamer().EmitAssemblerFlag(MCAF_Code16);
1172  else
1173    getParser().getStreamer().EmitAssemblerFlag(MCAF_Code32);
1174
1175  return false;
1176}
1177
1178extern "C" void LLVMInitializeARMAsmLexer();
1179
1180/// Force static initialization.
1181extern "C" void LLVMInitializeARMAsmParser() {
1182  RegisterAsmParser<ARMAsmParser> X(TheARMTarget);
1183  RegisterAsmParser<ARMAsmParser> Y(TheThumbTarget);
1184  LLVMInitializeARMAsmLexer();
1185}
1186
1187#define GET_REGISTER_MATCHER
1188#define GET_MATCHER_IMPLEMENTATION
1189#include "ARMGenAsmMatcher.inc"
1190