#include "llvm/Target/TargetAsmParser.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/SourceMgr.h"
+#include "llvm/ADT/OwningPtr.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Twine.h"
using namespace llvm;
bool Error(SMLoc L, const Twine &Msg) { return Parser.Error(L, Msg); }
- bool MaybeParseRegister(ARMOperand &Op, bool ParseWriteBack);
+ bool MaybeParseRegister(OwningPtr<ARMOperand> &Op, bool ParseWriteBack);
- bool ParseRegisterList(ARMOperand &Op);
+ bool ParseRegisterList(OwningPtr<ARMOperand> &Op);
- bool ParseMemory(ARMOperand &Op);
+ bool ParseMemory(OwningPtr<ARMOperand> &Op);
bool ParseMemoryOffsetReg(bool &Negative,
bool &OffsetRegShifted,
const MCExpr *&ShiftAmount,
const MCExpr *&Offset,
bool &OffsetIsReg,
- int &OffsetRegNum);
+ int &OffsetRegNum,
+ SMLoc &E);
- bool ParseShift(enum ShiftType &St, const MCExpr *&ShiftAmount);
+ bool ParseShift(enum ShiftType &St, const MCExpr *&ShiftAmount, SMLoc &E);
- bool ParseOperand(ARMOperand &Op);
+ bool ParseOperand(OwningPtr<ARMOperand> &Op);
bool ParseDirectiveWord(unsigned Size, SMLoc L);
/// its register number, or -1 if there is no match. To allow return values
/// to be used directly in register lists, arm registers have values between
/// 0 and 15.
- int MatchRegisterName(const StringRef &Name);
+ int MatchRegisterName(StringRef Name);
/// }
ARMAsmParser(const Target &T, MCAsmParser &_Parser)
: TargetAsmParser(T), Parser(_Parser) {}
- virtual bool ParseInstruction(const StringRef &Name, SMLoc NameLoc,
+ virtual bool ParseInstruction(StringRef Name, SMLoc NameLoc,
SmallVectorImpl<MCParsedAsmOperand*> &Operands);
virtual bool ParseDirective(AsmToken DirectiveID);
/// ARMOperand - Instances of this class represent a parsed ARM machine
/// instruction.
struct ARMOperand : public MCParsedAsmOperand {
- enum {
+private:
+ ARMOperand() {}
+public:
+ enum KindTy {
Token,
Register,
Immediate,
Memory
} Kind;
+ SMLoc StartLoc, EndLoc;
union {
struct {
struct {
const MCExpr *Val;
} Imm;
-
+
// This is for all forms of ARM address expressions
struct {
unsigned BaseRegNum;
} Mem;
};
+
+ ARMOperand(KindTy K, SMLoc S, SMLoc E)
+ : Kind(K), StartLoc(S), EndLoc(E) {}
+
+ ARMOperand(const ARMOperand &o) : MCParsedAsmOperand() {
+ Kind = o.Kind;
+ StartLoc = o.StartLoc;
+ EndLoc = o.EndLoc;
+ switch (Kind) {
+ case Token:
+ Tok = o.Tok;
+ break;
+ case Register:
+ Reg = o.Reg;
+ break;
+ case Immediate:
+ Imm = o.Imm;
+ break;
+ case Memory:
+ Mem = o.Mem;
+ break;
+ }
+ }
+
+ /// getStartLoc - Get the location of the first token of this operand.
+ SMLoc getStartLoc() const { return StartLoc; }
+ /// getEndLoc - Get the location of the last token of this operand.
+ SMLoc getEndLoc() const { return EndLoc; }
StringRef getToken() const {
assert(Kind == Token && "Invalid access!");
Inst.addOperand(MCOperand::CreateReg(getReg()));
}
- static ARMOperand CreateToken(StringRef Str) {
- ARMOperand Res;
- Res.Kind = Token;
- Res.Tok.Data = Str.data();
- Res.Tok.Length = Str.size();
- return Res;
+ static void CreateToken(OwningPtr<ARMOperand> &Op, StringRef Str,
+ SMLoc S) {
+ Op.reset(new ARMOperand);
+ Op->Kind = Token;
+ Op->Tok.Data = Str.data();
+ Op->Tok.Length = Str.size();
+ Op->StartLoc = S;
+ Op->EndLoc = S;
}
- static ARMOperand CreateReg(unsigned RegNum, bool Writeback) {
- ARMOperand Res;
- Res.Kind = Register;
- Res.Reg.RegNum = RegNum;
- Res.Reg.Writeback = Writeback;
- return Res;
+ static void CreateReg(OwningPtr<ARMOperand> &Op, unsigned RegNum,
+ bool Writeback, SMLoc S, SMLoc E) {
+ Op.reset(new ARMOperand);
+ Op->Kind = Register;
+ Op->Reg.RegNum = RegNum;
+ Op->Reg.Writeback = Writeback;
+
+ Op->StartLoc = S;
+ Op->EndLoc = E;
}
- static ARMOperand CreateImm(const MCExpr *Val) {
- ARMOperand Res;
- Res.Kind = Immediate;
- Res.Imm.Val = Val;
- return Res;
+ static void CreateImm(OwningPtr<ARMOperand> &Op, const MCExpr *Val,
+ SMLoc S, SMLoc E) {
+ Op.reset(new ARMOperand);
+ Op->Kind = Immediate;
+ Op->Imm.Val = Val;
+
+ Op->StartLoc = S;
+ Op->EndLoc = E;
}
- static ARMOperand CreateMem(unsigned BaseRegNum, bool OffsetIsReg,
- const MCExpr *Offset, unsigned OffsetRegNum,
- bool OffsetRegShifted, enum ShiftType ShiftType,
- const MCExpr *ShiftAmount, bool Preindexed,
- bool Postindexed, bool Negative, bool Writeback) {
- ARMOperand Res;
- Res.Kind = Memory;
- Res.Mem.BaseRegNum = BaseRegNum;
- Res.Mem.OffsetIsReg = OffsetIsReg;
- Res.Mem.Offset = Offset;
- Res.Mem.OffsetRegNum = OffsetRegNum;
- Res.Mem.OffsetRegShifted = OffsetRegShifted;
- Res.Mem.ShiftType = ShiftType;
- Res.Mem.ShiftAmount = ShiftAmount;
- Res.Mem.Preindexed = Preindexed;
- Res.Mem.Postindexed = Postindexed;
- Res.Mem.Negative = Negative;
- Res.Mem.Writeback = Writeback;
- return Res;
+ static void CreateMem(OwningPtr<ARMOperand> &Op,
+ unsigned BaseRegNum, bool OffsetIsReg,
+ const MCExpr *Offset, unsigned OffsetRegNum,
+ bool OffsetRegShifted, enum ShiftType ShiftType,
+ const MCExpr *ShiftAmount, bool Preindexed,
+ bool Postindexed, bool Negative, bool Writeback,
+ SMLoc S, SMLoc E) {
+ Op.reset(new ARMOperand);
+ Op->Kind = Memory;
+ Op->Mem.BaseRegNum = BaseRegNum;
+ Op->Mem.OffsetIsReg = OffsetIsReg;
+ Op->Mem.Offset = Offset;
+ Op->Mem.OffsetRegNum = OffsetRegNum;
+ Op->Mem.OffsetRegShifted = OffsetRegShifted;
+ Op->Mem.ShiftType = ShiftType;
+ Op->Mem.ShiftAmount = ShiftAmount;
+ Op->Mem.Preindexed = Preindexed;
+ Op->Mem.Postindexed = Postindexed;
+ Op->Mem.Negative = Negative;
+ Op->Mem.Writeback = Writeback;
+
+ Op->StartLoc = S;
+ Op->EndLoc = E;
}
};
/// and false is returned. Else true is returned and no token is eaten.
/// TODO this is likely to change to allow different register types and or to
/// parse for a specific register type.
-bool ARMAsmParser::MaybeParseRegister(ARMOperand &Op, bool ParseWriteBack) {
+bool ARMAsmParser::MaybeParseRegister
+ (OwningPtr<ARMOperand> &Op, bool ParseWriteBack) {
+ SMLoc S, E;
const AsmToken &Tok = Parser.getTok();
assert(Tok.is(AsmToken::Identifier) && "Token is not an Identifier");
RegNum = MatchRegisterName(Tok.getString());
if (RegNum == -1)
return true;
+
+ S = Tok.getLoc();
+
Parser.Lex(); // Eat identifier token.
+
+ E = Parser.getTok().getLoc();
bool Writeback = false;
if (ParseWriteBack) {
const AsmToken &ExclaimTok = Parser.getTok();
if (ExclaimTok.is(AsmToken::Exclaim)) {
+ E = ExclaimTok.getLoc();
Writeback = true;
Parser.Lex(); // Eat exclaim token
}
}
- Op = ARMOperand::CreateReg(RegNum, Writeback);
+ ARMOperand::CreateReg(Op, RegNum, Writeback, S, E);
return false;
}
/// Parse a register list, return false if successful else return true or an
/// error. The first token must be a '{' when called.
-bool ARMAsmParser::ParseRegisterList(ARMOperand &Op) {
+bool ARMAsmParser::ParseRegisterList(OwningPtr<ARMOperand> &Op) {
+ SMLoc S, E;
assert(Parser.getTok().is(AsmToken::LCurly) &&
"Token is not an Left Curly Brace");
+ S = Parser.getTok().getLoc();
Parser.Lex(); // Eat left curly brace token.
const AsmToken &RegTok = Parser.getTok();
const AsmToken &RCurlyTok = Parser.getTok();
if (RCurlyTok.isNot(AsmToken::RCurly))
return Error(RCurlyTok.getLoc(), "'}' expected");
+ E = RCurlyTok.getLoc();
Parser.Lex(); // Eat left curly brace token.
return false;
/// or an error. The first token must be a '[' when called.
/// TODO Only preindexing and postindexing addressing are started, unindexed
/// with option, etc are still to do.
-bool ARMAsmParser::ParseMemory(ARMOperand &Op) {
+bool ARMAsmParser::ParseMemory(OwningPtr<ARMOperand> &Op) {
+ SMLoc S, E;
assert(Parser.getTok().is(AsmToken::LBrac) &&
"Token is not an Left Bracket");
+ S = Parser.getTok().getLoc();
Parser.Lex(); // Eat left bracket token.
const AsmToken &BaseRegTok = Parser.getTok();
return Error(BaseRegTok.getLoc(), "register expected");
if (MaybeParseRegister(Op, false))
return Error(BaseRegTok.getLoc(), "register expected");
- int BaseRegNum = Op.getReg();
+ int BaseRegNum = Op->getReg();
bool Preindexed = false;
bool Postindexed = false;
const MCExpr *ShiftAmount;
const MCExpr *Offset;
if(ParseMemoryOffsetReg(Negative, OffsetRegShifted, ShiftType, ShiftAmount,
- Offset, OffsetIsReg, OffsetRegNum))
+ Offset, OffsetIsReg, OffsetRegNum, E))
return true;
const AsmToken &RBracTok = Parser.getTok();
if (RBracTok.isNot(AsmToken::RBrac))
return Error(RBracTok.getLoc(), "']' expected");
+ E = RBracTok.getLoc();
Parser.Lex(); // Eat right bracket token.
const AsmToken &ExclaimTok = Parser.getTok();
if (ExclaimTok.is(AsmToken::Exclaim)) {
+ E = ExclaimTok.getLoc();
Writeback = true;
Parser.Lex(); // Eat exclaim token
}
- Op = ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset, OffsetRegNum,
- OffsetRegShifted, ShiftType, ShiftAmount,
- Preindexed, Postindexed, Negative, Writeback);
+ ARMOperand::CreateMem(Op, BaseRegNum, OffsetIsReg, Offset, OffsetRegNum,
+ OffsetRegShifted, ShiftType, ShiftAmount,
+ Preindexed, Postindexed, Negative, Writeback, S, E);
return false;
}
// The "[Rn" we have so far was not followed by a comma.
// the "[Rn".
Postindexed = true;
Writeback = true;
+ E = Tok.getLoc();
Parser.Lex(); // Eat right bracket token.
int OffsetRegNum = 0;
const AsmToken &NextTok = Parser.getTok();
if (NextTok.isNot(AsmToken::EndOfStatement)) {
if (NextTok.isNot(AsmToken::Comma))
- return Error(NextTok.getLoc(), "',' expected");
+ return Error(NextTok.getLoc(), "',' expected");
Parser.Lex(); // Eat comma token.
if(ParseMemoryOffsetReg(Negative, OffsetRegShifted, ShiftType,
- ShiftAmount, Offset, OffsetIsReg, OffsetRegNum))
+ ShiftAmount, Offset, OffsetIsReg, OffsetRegNum,
+ E))
return true;
}
- Op = ARMOperand::CreateMem(BaseRegNum, OffsetIsReg, Offset, OffsetRegNum,
- OffsetRegShifted, ShiftType, ShiftAmount,
- Preindexed, Postindexed, Negative, Writeback);
+ ARMOperand::CreateMem(Op, BaseRegNum, OffsetIsReg, Offset, OffsetRegNum,
+ OffsetRegShifted, ShiftType, ShiftAmount,
+ Preindexed, Postindexed, Negative, Writeback, S, E);
return false;
}
/// #offset
/// we return false on success or an error otherwise.
bool ARMAsmParser::ParseMemoryOffsetReg(bool &Negative,
- bool &OffsetRegShifted,
+ bool &OffsetRegShifted,
enum ShiftType &ShiftType,
const MCExpr *&ShiftAmount,
const MCExpr *&Offset,
bool &OffsetIsReg,
- int &OffsetRegNum) {
- ARMOperand Op;
+ int &OffsetRegNum,
+ SMLoc &E) {
+ OwningPtr<ARMOperand> Op;
Negative = false;
OffsetRegShifted = false;
OffsetIsReg = false;
OffsetRegNum = -1;
const AsmToken &NextTok = Parser.getTok();
+ E = NextTok.getLoc();
if (NextTok.is(AsmToken::Plus))
Parser.Lex(); // Eat plus token.
else if (NextTok.is(AsmToken::Minus)) {
const AsmToken &OffsetRegTok = Parser.getTok();
if (OffsetRegTok.is(AsmToken::Identifier)) {
OffsetIsReg = !MaybeParseRegister(Op, false);
- if (OffsetIsReg)
- OffsetRegNum = Op.getReg();
+ if (OffsetIsReg) {
+ E = Op->getEndLoc();
+ OffsetRegNum = Op->getReg();
+ }
}
// If we parsed a register as the offset then their can be a shift after that
if (OffsetRegNum != -1) {
Parser.Lex(); // Eat comma token.
const AsmToken &Tok = Parser.getTok();
- if (ParseShift(ShiftType, ShiftAmount))
- return Error(Tok.getLoc(), "shift expected");
+ if (ParseShift(ShiftType, ShiftAmount, E))
+ return Error(Tok.getLoc(), "shift expected");
OffsetRegShifted = true;
}
}
const AsmToken &HashTok = Parser.getTok();
if (HashTok.isNot(AsmToken::Hash))
return Error(HashTok.getLoc(), "'#' expected");
+
Parser.Lex(); // Eat hash token.
if (getParser().ParseExpression(Offset))
return true;
+ E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
}
return false;
}
/// ( lsl | lsr | asr | ror ) , # shift_amount
/// rrx
/// and returns true if it parses a shift otherwise it returns false.
-bool ARMAsmParser::ParseShift(ShiftType &St, const MCExpr *&ShiftAmount) {
+bool ARMAsmParser::ParseShift(ShiftType &St,
+ const MCExpr *&ShiftAmount,
+ SMLoc &E) {
const AsmToken &Tok = Parser.getTok();
if (Tok.isNot(AsmToken::Identifier))
return true;
- const StringRef &ShiftName = Tok.getString();
+ StringRef ShiftName = Tok.getString();
if (ShiftName == "lsl" || ShiftName == "LSL")
St = Lsl;
else if (ShiftName == "lsr" || ShiftName == "LSR")
}
/// A hack to allow some testing, to be replaced by a real table gen version.
-int ARMAsmParser::MatchRegisterName(const StringRef &Name) {
+int ARMAsmParser::MatchRegisterName(StringRef Name) {
if (Name == "r0" || Name == "R0")
return 0;
else if (Name == "r1" || Name == "R1")
MCInst &Inst) {
ARMOperand &Op0 = *(ARMOperand*)Operands[0];
assert(Op0.Kind == ARMOperand::Token && "First operand not a Token");
- const StringRef &Mnemonic = Op0.getToken();
+ StringRef Mnemonic = Op0.getToken();
if (Mnemonic == "add" ||
Mnemonic == "stmfd" ||
Mnemonic == "str" ||
/// Parse a arm instruction operand. For now this parses the operand regardless
/// of the mnemonic.
-bool ARMAsmParser::ParseOperand(ARMOperand &Op) {
+bool ARMAsmParser::ParseOperand(OwningPtr<ARMOperand> &Op) {
+ SMLoc S, E;
+
switch (getLexer().getKind()) {
case AsmToken::Identifier:
if (!MaybeParseRegister(Op, true))
// This was not a register so parse other operands that start with an
// identifier (like labels) as expressions and create them as immediates.
const MCExpr *IdVal;
+ S = Parser.getTok().getLoc();
if (getParser().ParseExpression(IdVal))
return true;
- Op = ARMOperand::CreateImm(IdVal);
+ E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
+ ARMOperand::CreateImm(Op, IdVal, S, E);
return false;
case AsmToken::LBrac:
return ParseMemory(Op);
case AsmToken::Hash:
// #42 -> immediate.
// TODO: ":lower16:" and ":upper16:" modifiers after # before immediate
+ S = Parser.getTok().getLoc();
Parser.Lex();
const MCExpr *ImmVal;
if (getParser().ParseExpression(ImmVal))
return true;
- Op = ARMOperand::CreateImm(ImmVal);
+ E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
+ ARMOperand::CreateImm(Op, ImmVal, S, E);
return false;
default:
return Error(Parser.getTok().getLoc(), "unexpected token in operand");
}
/// Parse an arm instruction mnemonic followed by its operands.
-bool ARMAsmParser::ParseInstruction(const StringRef &Name, SMLoc NameLoc,
+bool ARMAsmParser::ParseInstruction(StringRef Name, SMLoc NameLoc,
SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
- Operands.push_back(new ARMOperand(ARMOperand::CreateToken(Name)));
+ OwningPtr<ARMOperand> Op;
+ ARMOperand::CreateToken(Op, Name, NameLoc);
+
+ Operands.push_back(Op.take());
- SMLoc Loc = Parser.getTok().getLoc();
if (getLexer().isNot(AsmToken::EndOfStatement)) {
// Read the first operand.
- ARMOperand Op;
+ OwningPtr<ARMOperand> Op;
if (ParseOperand(Op)) return true;
- Operands.push_back(new ARMOperand(Op));
+ Operands.push_back(Op.take());
while (getLexer().is(AsmToken::Comma)) {
Parser.Lex(); // Eat the comma.
// Parse and remember the operand.
if (ParseOperand(Op)) return true;
- Operands.push_back(new ARMOperand(Op));
+ Operands.push_back(Op.take());
}
}
return false;
const AsmToken &Tok = Parser.getTok();
if (Tok.isNot(AsmToken::Identifier))
return Error(L, "unexpected token in .syntax directive");
- const StringRef &Mode = Tok.getString();
- bool unified_syntax;
- if (Mode == "unified" || Mode == "UNIFIED") {
+ StringRef Mode = Tok.getString();
+ if (Mode == "unified" || Mode == "UNIFIED")
Parser.Lex();
- unified_syntax = true;
- }
- else if (Mode == "divided" || Mode == "DIVIDED") {
+ else if (Mode == "divided" || Mode == "DIVIDED")
Parser.Lex();
- unified_syntax = false;
- }
else
return Error(L, "unrecognized syntax mode in .syntax directive");
if (Tok.isNot(AsmToken::Integer))
return Error(L, "unexpected token in .code directive");
int64_t Val = Parser.getTok().getIntVal();
- bool thumb_mode;
- if (Val == 16) {
+ if (Val == 16)
Parser.Lex();
- thumb_mode = true;
- }
- else if (Val == 32) {
+ else if (Val == 32)
Parser.Lex();
- thumb_mode = false;
- }
else
return Error(L, "invalid operand to .code directive");
return false;
}
+extern "C" void LLVMInitializeARMAsmLexer();
+
/// Force static initialization.
extern "C" void LLVMInitializeARMAsmParser() {
RegisterAsmParser<ARMAsmParser> X(TheARMTarget);
RegisterAsmParser<ARMAsmParser> Y(TheThumbTarget);
+ LLVMInitializeARMAsmLexer();
}