diff options
Diffstat (limited to 'contrib/llvm/lib/CodeGen/MIRParser')
-rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MILexer.cpp | 225 | ||||
-rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MILexer.h | 111 | ||||
-rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MIParser.cpp | 606 | ||||
-rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MIParser.h | 51 | ||||
-rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MIRParser.cpp | 513 |
5 files changed, 1506 insertions, 0 deletions
diff --git a/contrib/llvm/lib/CodeGen/MIRParser/MILexer.cpp b/contrib/llvm/lib/CodeGen/MIRParser/MILexer.cpp new file mode 100644 index 0000000..482c33a --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MILexer.cpp @@ -0,0 +1,225 @@ +//===- MILexer.cpp - Machine instructions lexer implementation ----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the lexing of machine instructions. +// +//===----------------------------------------------------------------------===// + +#include "MILexer.h" +#include "llvm/ADT/StringSwitch.h" +#include "llvm/ADT/Twine.h" +#include <cctype> + +using namespace llvm; + +namespace { + +/// This class provides a way to iterate and get characters from the source +/// string. +class Cursor { + const char *Ptr; + const char *End; + +public: + Cursor(NoneType) : Ptr(nullptr), End(nullptr) {} + + explicit Cursor(StringRef Str) { + Ptr = Str.data(); + End = Ptr + Str.size(); + } + + bool isEOF() const { return Ptr == End; } + + char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; } + + void advance(unsigned I = 1) { Ptr += I; } + + StringRef remaining() const { return StringRef(Ptr, End - Ptr); } + + StringRef upto(Cursor C) const { + assert(C.Ptr >= Ptr && C.Ptr <= End); + return StringRef(Ptr, C.Ptr - Ptr); + } + + StringRef::iterator location() const { return Ptr; } + + operator bool() const { return Ptr != nullptr; } +}; + +} // end anonymous namespace + +/// Skip the leading whitespace characters and return the updated cursor. +static Cursor skipWhitespace(Cursor C) { + while (isspace(C.peek())) + C.advance(); + return C; +} + +static bool isIdentifierChar(char C) { + return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.'; +} + +static MIToken::TokenKind getIdentifierKind(StringRef Identifier) { + return StringSwitch<MIToken::TokenKind>(Identifier) + .Case("_", MIToken::underscore) + .Case("implicit", MIToken::kw_implicit) + .Case("implicit-def", MIToken::kw_implicit_define) + .Case("dead", MIToken::kw_dead) + .Case("killed", MIToken::kw_killed) + .Case("undef", MIToken::kw_undef) + .Default(MIToken::Identifier); +} + +static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) { + if (!isalpha(C.peek()) && C.peek() != '_') + return None; + auto Range = C; + while (isIdentifierChar(C.peek())) + C.advance(); + auto Identifier = Range.upto(C); + Token = MIToken(getIdentifierKind(Identifier), Identifier); + return C; +} + +static Cursor maybeLexMachineBasicBlock( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + if (!C.remaining().startswith("%bb.")) + return None; + auto Range = C; + C.advance(4); // Skip '%bb.' + if (!isdigit(C.peek())) { + Token = MIToken(MIToken::Error, C.remaining()); + ErrorCallback(C.location(), "expected a number after '%bb.'"); + return C; + } + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + StringRef Number = NumberRange.upto(C); + unsigned StringOffset = 4 + Number.size(); // Drop '%bb.<id>' + if (C.peek() == '.') { + C.advance(); // Skip '.' + ++StringOffset; + while (isIdentifierChar(C.peek())) + C.advance(); + } + Token = MIToken(MIToken::MachineBasicBlock, Range.upto(C), APSInt(Number), + StringOffset); + return C; +} + +static Cursor lexVirtualRegister(Cursor C, MIToken &Token) { + auto Range = C; + C.advance(); // Skip '%' + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + Token = MIToken(MIToken::VirtualRegister, Range.upto(C), + APSInt(NumberRange.upto(C))); + return C; +} + +static Cursor maybeLexRegister(Cursor C, MIToken &Token) { + if (C.peek() != '%') + return None; + if (isdigit(C.peek(1))) + return lexVirtualRegister(C, Token); + auto Range = C; + C.advance(); // Skip '%' + while (isIdentifierChar(C.peek())) + C.advance(); + Token = MIToken(MIToken::NamedRegister, Range.upto(C), + /*StringOffset=*/1); // Drop the '%' + return C; +} + +static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token) { + if (C.peek() != '@') + return None; + auto Range = C; + C.advance(); // Skip the '@' + // TODO: add support for quoted names. + if (!isdigit(C.peek())) { + while (isIdentifierChar(C.peek())) + C.advance(); + Token = MIToken(MIToken::NamedGlobalValue, Range.upto(C), + /*StringOffset=*/1); // Drop the '@' + return C; + } + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + Token = + MIToken(MIToken::GlobalValue, Range.upto(C), APSInt(NumberRange.upto(C))); + return C; +} + +static Cursor maybeLexIntegerLiteral(Cursor C, MIToken &Token) { + if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1)))) + return None; + auto Range = C; + C.advance(); + while (isdigit(C.peek())) + C.advance(); + StringRef StrVal = Range.upto(C); + Token = MIToken(MIToken::IntegerLiteral, StrVal, APSInt(StrVal)); + return C; +} + +static MIToken::TokenKind symbolToken(char C) { + switch (C) { + case ',': + return MIToken::comma; + case '=': + return MIToken::equal; + case ':': + return MIToken::colon; + default: + return MIToken::Error; + } +} + +static Cursor maybeLexSymbol(Cursor C, MIToken &Token) { + auto Kind = symbolToken(C.peek()); + if (Kind == MIToken::Error) + return None; + auto Range = C; + C.advance(); + Token = MIToken(Kind, Range.upto(C)); + return C; +} + +StringRef llvm::lexMIToken( + StringRef Source, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + auto C = skipWhitespace(Cursor(Source)); + if (C.isEOF()) { + Token = MIToken(MIToken::Eof, C.remaining()); + return C.remaining(); + } + + if (Cursor R = maybeLexIdentifier(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexRegister(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexGlobalValue(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexIntegerLiteral(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexSymbol(C, Token)) + return R.remaining(); + + Token = MIToken(MIToken::Error, C.remaining()); + ErrorCallback(C.location(), + Twine("unexpected character '") + Twine(C.peek()) + "'"); + return C.remaining(); +} diff --git a/contrib/llvm/lib/CodeGen/MIRParser/MILexer.h b/contrib/llvm/lib/CodeGen/MIRParser/MILexer.h new file mode 100644 index 0000000..55460b5 --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MILexer.h @@ -0,0 +1,111 @@ +//===- MILexer.h - Lexer for machine instructions -------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file declares the function that lexes the machine instruction source +// string. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_LIB_CODEGEN_MIRPARSER_MILEXER_H +#define LLVM_LIB_CODEGEN_MIRPARSER_MILEXER_H + +#include "llvm/ADT/APSInt.h" +#include "llvm/ADT/StringRef.h" +#include "llvm/ADT/STLExtras.h" +#include <functional> + +namespace llvm { + +class Twine; + +/// A token produced by the machine instruction lexer. +struct MIToken { + enum TokenKind { + // Markers + Eof, + Error, + + // Tokens with no info. + comma, + equal, + underscore, + colon, + + // Keywords + kw_implicit, + kw_implicit_define, + kw_dead, + kw_killed, + kw_undef, + + // Identifier tokens + Identifier, + NamedRegister, + MachineBasicBlock, + NamedGlobalValue, + GlobalValue, + + // Other tokens + IntegerLiteral, + VirtualRegister + }; + +private: + TokenKind Kind; + unsigned StringOffset; + StringRef Range; + APSInt IntVal; + +public: + MIToken(TokenKind Kind, StringRef Range, unsigned StringOffset = 0) + : Kind(Kind), StringOffset(StringOffset), Range(Range) {} + + MIToken(TokenKind Kind, StringRef Range, const APSInt &IntVal, + unsigned StringOffset = 0) + : Kind(Kind), StringOffset(StringOffset), Range(Range), IntVal(IntVal) {} + + TokenKind kind() const { return Kind; } + + bool isError() const { return Kind == Error; } + + bool isRegister() const { + return Kind == NamedRegister || Kind == underscore || + Kind == VirtualRegister; + } + + bool isRegisterFlag() const { + return Kind == kw_implicit || Kind == kw_implicit_define || + Kind == kw_dead || Kind == kw_killed || Kind == kw_undef; + } + + bool is(TokenKind K) const { return Kind == K; } + + bool isNot(TokenKind K) const { return Kind != K; } + + StringRef::iterator location() const { return Range.begin(); } + + StringRef stringValue() const { return Range.drop_front(StringOffset); } + + const APSInt &integerValue() const { return IntVal; } + + bool hasIntegerValue() const { + return Kind == IntegerLiteral || Kind == MachineBasicBlock || + Kind == GlobalValue || Kind == VirtualRegister; + } +}; + +/// Consume a single machine instruction token in the given source and return +/// the remaining source string. +StringRef lexMIToken( + StringRef Source, MIToken &Token, + function_ref<void(StringRef::iterator, const Twine &)> ErrorCallback); + +} // end namespace llvm + +#endif diff --git a/contrib/llvm/lib/CodeGen/MIRParser/MIParser.cpp b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.cpp new file mode 100644 index 0000000..c000112 --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.cpp @@ -0,0 +1,606 @@ +//===- MIParser.cpp - Machine instructions parser implementation ----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the parsing of machine instructions. +// +//===----------------------------------------------------------------------===// + +#include "MIParser.h" +#include "MILexer.h" +#include "llvm/ADT/StringMap.h" +#include "llvm/AsmParser/SlotMapping.h" +#include "llvm/CodeGen/MachineBasicBlock.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineInstr.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Support/SourceMgr.h" +#include "llvm/Target/TargetSubtargetInfo.h" +#include "llvm/Target/TargetInstrInfo.h" + +using namespace llvm; + +namespace { + +/// A wrapper struct around the 'MachineOperand' struct that includes a source +/// range. +struct MachineOperandWithLocation { + MachineOperand Operand; + StringRef::iterator Begin; + StringRef::iterator End; + + MachineOperandWithLocation(const MachineOperand &Operand, + StringRef::iterator Begin, StringRef::iterator End) + : Operand(Operand), Begin(Begin), End(End) {} +}; + +class MIParser { + SourceMgr &SM; + MachineFunction &MF; + SMDiagnostic &Error; + StringRef Source, CurrentSource; + MIToken Token; + const PerFunctionMIParsingState &PFS; + /// Maps from indices to unnamed global values and metadata nodes. + const SlotMapping &IRSlots; + /// Maps from instruction names to op codes. + StringMap<unsigned> Names2InstrOpCodes; + /// Maps from register names to registers. + StringMap<unsigned> Names2Regs; + /// Maps from register mask names to register masks. + StringMap<const uint32_t *> Names2RegMasks; + /// Maps from subregister names to subregister indices. + StringMap<unsigned> Names2SubRegIndices; + +public: + MIParser(SourceMgr &SM, MachineFunction &MF, SMDiagnostic &Error, + StringRef Source, const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots); + + void lex(); + + /// Report an error at the current location with the given message. + /// + /// This function always return true. + bool error(const Twine &Msg); + + /// Report an error at the given location with the given message. + /// + /// This function always return true. + bool error(StringRef::iterator Loc, const Twine &Msg); + + bool parse(MachineInstr *&MI); + bool parseMBB(MachineBasicBlock *&MBB); + bool parseNamedRegister(unsigned &Reg); + + bool parseRegister(unsigned &Reg); + bool parseRegisterFlag(unsigned &Flags); + bool parseSubRegisterIndex(unsigned &SubReg); + bool parseRegisterOperand(MachineOperand &Dest, bool IsDef = false); + bool parseImmediateOperand(MachineOperand &Dest); + bool parseMBBReference(MachineBasicBlock *&MBB); + bool parseMBBOperand(MachineOperand &Dest); + bool parseGlobalAddressOperand(MachineOperand &Dest); + bool parseMachineOperand(MachineOperand &Dest); + +private: + /// Convert the integer literal in the current token into an unsigned integer. + /// + /// Return true if an error occurred. + bool getUnsigned(unsigned &Result); + + void initNames2InstrOpCodes(); + + /// Try to convert an instruction name to an opcode. Return true if the + /// instruction name is invalid. + bool parseInstrName(StringRef InstrName, unsigned &OpCode); + + bool parseInstruction(unsigned &OpCode); + + bool verifyImplicitOperands(ArrayRef<MachineOperandWithLocation> Operands, + const MCInstrDesc &MCID); + + void initNames2Regs(); + + /// Try to convert a register name to a register number. Return true if the + /// register name is invalid. + bool getRegisterByName(StringRef RegName, unsigned &Reg); + + void initNames2RegMasks(); + + /// Check if the given identifier is a name of a register mask. + /// + /// Return null if the identifier isn't a register mask. + const uint32_t *getRegMask(StringRef Identifier); + + void initNames2SubRegIndices(); + + /// Check if the given identifier is a name of a subregister index. + /// + /// Return 0 if the name isn't a subregister index class. + unsigned getSubRegIndex(StringRef Name); +}; + +} // end anonymous namespace + +MIParser::MIParser(SourceMgr &SM, MachineFunction &MF, SMDiagnostic &Error, + StringRef Source, const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots) + : SM(SM), MF(MF), Error(Error), Source(Source), CurrentSource(Source), + Token(MIToken::Error, StringRef()), PFS(PFS), IRSlots(IRSlots) {} + +void MIParser::lex() { + CurrentSource = lexMIToken( + CurrentSource, Token, + [this](StringRef::iterator Loc, const Twine &Msg) { error(Loc, Msg); }); +} + +bool MIParser::error(const Twine &Msg) { return error(Token.location(), Msg); } + +bool MIParser::error(StringRef::iterator Loc, const Twine &Msg) { + assert(Loc >= Source.data() && Loc <= (Source.data() + Source.size())); + Error = SMDiagnostic( + SM, SMLoc(), + SM.getMemoryBuffer(SM.getMainFileID())->getBufferIdentifier(), 1, + Loc - Source.data(), SourceMgr::DK_Error, Msg.str(), Source, None, None); + return true; +} + +bool MIParser::parse(MachineInstr *&MI) { + lex(); + + // Parse any register operands before '=' + // TODO: Allow parsing of multiple operands before '=' + MachineOperand MO = MachineOperand::CreateImm(0); + SmallVector<MachineOperandWithLocation, 8> Operands; + if (Token.isRegister() || Token.isRegisterFlag()) { + auto Loc = Token.location(); + if (parseRegisterOperand(MO, /*IsDef=*/true)) + return true; + Operands.push_back(MachineOperandWithLocation(MO, Loc, Token.location())); + if (Token.isNot(MIToken::equal)) + return error("expected '='"); + lex(); + } + + unsigned OpCode; + if (Token.isError() || parseInstruction(OpCode)) + return true; + + // TODO: Parse the instruction flags and memory operands. + + // Parse the remaining machine operands. + while (Token.isNot(MIToken::Eof)) { + auto Loc = Token.location(); + if (parseMachineOperand(MO)) + return true; + Operands.push_back(MachineOperandWithLocation(MO, Loc, Token.location())); + if (Token.is(MIToken::Eof)) + break; + if (Token.isNot(MIToken::comma)) + return error("expected ',' before the next machine operand"); + lex(); + } + + const auto &MCID = MF.getSubtarget().getInstrInfo()->get(OpCode); + if (!MCID.isVariadic()) { + // FIXME: Move the implicit operand verification to the machine verifier. + if (verifyImplicitOperands(Operands, MCID)) + return true; + } + + // TODO: Check for extraneous machine operands. + MI = MF.CreateMachineInstr(MCID, DebugLoc(), /*NoImplicit=*/true); + for (const auto &Operand : Operands) + MI->addOperand(MF, Operand.Operand); + return false; +} + +bool MIParser::parseMBB(MachineBasicBlock *&MBB) { + lex(); + if (Token.isNot(MIToken::MachineBasicBlock)) + return error("expected a machine basic block reference"); + if (parseMBBReference(MBB)) + return true; + lex(); + if (Token.isNot(MIToken::Eof)) + return error( + "expected end of string after the machine basic block reference"); + return false; +} + +bool MIParser::parseNamedRegister(unsigned &Reg) { + lex(); + if (Token.isNot(MIToken::NamedRegister)) + return error("expected a named register"); + if (parseRegister(Reg)) + return 0; + lex(); + if (Token.isNot(MIToken::Eof)) + return error("expected end of string after the register reference"); + return false; +} + +static const char *printImplicitRegisterFlag(const MachineOperand &MO) { + assert(MO.isImplicit()); + return MO.isDef() ? "implicit-def" : "implicit"; +} + +static std::string getRegisterName(const TargetRegisterInfo *TRI, + unsigned Reg) { + assert(TargetRegisterInfo::isPhysicalRegister(Reg) && "expected phys reg"); + return StringRef(TRI->getName(Reg)).lower(); +} + +bool MIParser::verifyImplicitOperands( + ArrayRef<MachineOperandWithLocation> Operands, const MCInstrDesc &MCID) { + if (MCID.isCall()) + // We can't verify call instructions as they can contain arbitrary implicit + // register and register mask operands. + return false; + + // Gather all the expected implicit operands. + SmallVector<MachineOperand, 4> ImplicitOperands; + if (MCID.ImplicitDefs) + for (const uint16_t *ImpDefs = MCID.getImplicitDefs(); *ImpDefs; ++ImpDefs) + ImplicitOperands.push_back( + MachineOperand::CreateReg(*ImpDefs, true, true)); + if (MCID.ImplicitUses) + for (const uint16_t *ImpUses = MCID.getImplicitUses(); *ImpUses; ++ImpUses) + ImplicitOperands.push_back( + MachineOperand::CreateReg(*ImpUses, false, true)); + + const auto *TRI = MF.getSubtarget().getRegisterInfo(); + assert(TRI && "Expected target register info"); + size_t I = ImplicitOperands.size(), J = Operands.size(); + while (I) { + --I; + if (J) { + --J; + const auto &ImplicitOperand = ImplicitOperands[I]; + const auto &Operand = Operands[J].Operand; + if (ImplicitOperand.isIdenticalTo(Operand)) + continue; + if (Operand.isReg() && Operand.isImplicit()) { + return error(Operands[J].Begin, + Twine("expected an implicit register operand '") + + printImplicitRegisterFlag(ImplicitOperand) + " %" + + getRegisterName(TRI, ImplicitOperand.getReg()) + "'"); + } + } + // TODO: Fix source location when Operands[J].end is right before '=', i.e: + // insead of reporting an error at this location: + // %eax = MOV32r0 + // ^ + // report the error at the following location: + // %eax = MOV32r0 + // ^ + return error(J < Operands.size() ? Operands[J].End : Token.location(), + Twine("missing implicit register operand '") + + printImplicitRegisterFlag(ImplicitOperands[I]) + " %" + + getRegisterName(TRI, ImplicitOperands[I].getReg()) + "'"); + } + return false; +} + +bool MIParser::parseInstruction(unsigned &OpCode) { + if (Token.isNot(MIToken::Identifier)) + return error("expected a machine instruction"); + StringRef InstrName = Token.stringValue(); + if (parseInstrName(InstrName, OpCode)) + return error(Twine("unknown machine instruction name '") + InstrName + "'"); + lex(); + return false; +} + +bool MIParser::parseRegister(unsigned &Reg) { + switch (Token.kind()) { + case MIToken::underscore: + Reg = 0; + break; + case MIToken::NamedRegister: { + StringRef Name = Token.stringValue(); + if (getRegisterByName(Name, Reg)) + return error(Twine("unknown register name '") + Name + "'"); + break; + } + case MIToken::VirtualRegister: { + unsigned ID; + if (getUnsigned(ID)) + return true; + const auto RegInfo = PFS.VirtualRegisterSlots.find(ID); + if (RegInfo == PFS.VirtualRegisterSlots.end()) + return error(Twine("use of undefined virtual register '%") + Twine(ID) + + "'"); + Reg = RegInfo->second; + break; + } + // TODO: Parse other register kinds. + default: + llvm_unreachable("The current token should be a register"); + } + return false; +} + +bool MIParser::parseRegisterFlag(unsigned &Flags) { + switch (Token.kind()) { + case MIToken::kw_implicit: + Flags |= RegState::Implicit; + break; + case MIToken::kw_implicit_define: + Flags |= RegState::ImplicitDefine; + break; + case MIToken::kw_dead: + Flags |= RegState::Dead; + break; + case MIToken::kw_killed: + Flags |= RegState::Kill; + break; + case MIToken::kw_undef: + Flags |= RegState::Undef; + break; + // TODO: report an error when we specify the same flag more than once. + // TODO: parse the other register flags. + default: + llvm_unreachable("The current token should be a register flag"); + } + lex(); + return false; +} + +bool MIParser::parseSubRegisterIndex(unsigned &SubReg) { + assert(Token.is(MIToken::colon)); + lex(); + if (Token.isNot(MIToken::Identifier)) + return error("expected a subregister index after ':'"); + auto Name = Token.stringValue(); + SubReg = getSubRegIndex(Name); + if (!SubReg) + return error(Twine("use of unknown subregister index '") + Name + "'"); + lex(); + return false; +} + +bool MIParser::parseRegisterOperand(MachineOperand &Dest, bool IsDef) { + unsigned Reg; + unsigned Flags = IsDef ? RegState::Define : 0; + while (Token.isRegisterFlag()) { + if (parseRegisterFlag(Flags)) + return true; + } + if (!Token.isRegister()) + return error("expected a register after register flags"); + if (parseRegister(Reg)) + return true; + lex(); + unsigned SubReg = 0; + if (Token.is(MIToken::colon)) { + if (parseSubRegisterIndex(SubReg)) + return true; + } + Dest = MachineOperand::CreateReg( + Reg, Flags & RegState::Define, Flags & RegState::Implicit, + Flags & RegState::Kill, Flags & RegState::Dead, Flags & RegState::Undef, + /*isEarlyClobber=*/false, SubReg); + return false; +} + +bool MIParser::parseImmediateOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::IntegerLiteral)); + const APSInt &Int = Token.integerValue(); + if (Int.getMinSignedBits() > 64) + // TODO: Replace this with an error when we can parse CIMM Machine Operands. + llvm_unreachable("Can't parse large integer literals yet!"); + Dest = MachineOperand::CreateImm(Int.getExtValue()); + lex(); + return false; +} + +bool MIParser::getUnsigned(unsigned &Result) { + assert(Token.hasIntegerValue() && "Expected a token with an integer value"); + const uint64_t Limit = uint64_t(std::numeric_limits<unsigned>::max()) + 1; + uint64_t Val64 = Token.integerValue().getLimitedValue(Limit); + if (Val64 == Limit) + return error("expected 32-bit integer (too large)"); + Result = Val64; + return false; +} + +bool MIParser::parseMBBReference(MachineBasicBlock *&MBB) { + assert(Token.is(MIToken::MachineBasicBlock)); + unsigned Number; + if (getUnsigned(Number)) + return true; + auto MBBInfo = PFS.MBBSlots.find(Number); + if (MBBInfo == PFS.MBBSlots.end()) + return error(Twine("use of undefined machine basic block #") + + Twine(Number)); + MBB = MBBInfo->second; + if (!Token.stringValue().empty() && Token.stringValue() != MBB->getName()) + return error(Twine("the name of machine basic block #") + Twine(Number) + + " isn't '" + Token.stringValue() + "'"); + return false; +} + +bool MIParser::parseMBBOperand(MachineOperand &Dest) { + MachineBasicBlock *MBB; + if (parseMBBReference(MBB)) + return true; + Dest = MachineOperand::CreateMBB(MBB); + lex(); + return false; +} + +bool MIParser::parseGlobalAddressOperand(MachineOperand &Dest) { + switch (Token.kind()) { + case MIToken::NamedGlobalValue: { + auto Name = Token.stringValue(); + const Module *M = MF.getFunction()->getParent(); + if (const auto *GV = M->getNamedValue(Name)) { + Dest = MachineOperand::CreateGA(GV, /*Offset=*/0); + break; + } + return error(Twine("use of undefined global value '@") + Name + "'"); + } + case MIToken::GlobalValue: { + unsigned GVIdx; + if (getUnsigned(GVIdx)) + return true; + if (GVIdx >= IRSlots.GlobalValues.size()) + return error(Twine("use of undefined global value '@") + Twine(GVIdx) + + "'"); + Dest = MachineOperand::CreateGA(IRSlots.GlobalValues[GVIdx], + /*Offset=*/0); + break; + } + default: + llvm_unreachable("The current token should be a global value"); + } + // TODO: Parse offset and target flags. + lex(); + return false; +} + +bool MIParser::parseMachineOperand(MachineOperand &Dest) { + switch (Token.kind()) { + case MIToken::kw_implicit: + case MIToken::kw_implicit_define: + case MIToken::kw_dead: + case MIToken::kw_killed: + case MIToken::kw_undef: + case MIToken::underscore: + case MIToken::NamedRegister: + case MIToken::VirtualRegister: + return parseRegisterOperand(Dest); + case MIToken::IntegerLiteral: + return parseImmediateOperand(Dest); + case MIToken::MachineBasicBlock: + return parseMBBOperand(Dest); + case MIToken::GlobalValue: + case MIToken::NamedGlobalValue: + return parseGlobalAddressOperand(Dest); + case MIToken::Error: + return true; + case MIToken::Identifier: + if (const auto *RegMask = getRegMask(Token.stringValue())) { + Dest = MachineOperand::CreateRegMask(RegMask); + lex(); + break; + } + // fallthrough + default: + // TODO: parse the other machine operands. + return error("expected a machine operand"); + } + return false; +} + +void MIParser::initNames2InstrOpCodes() { + if (!Names2InstrOpCodes.empty()) + return; + const auto *TII = MF.getSubtarget().getInstrInfo(); + assert(TII && "Expected target instruction info"); + for (unsigned I = 0, E = TII->getNumOpcodes(); I < E; ++I) + Names2InstrOpCodes.insert(std::make_pair(StringRef(TII->getName(I)), I)); +} + +bool MIParser::parseInstrName(StringRef InstrName, unsigned &OpCode) { + initNames2InstrOpCodes(); + auto InstrInfo = Names2InstrOpCodes.find(InstrName); + if (InstrInfo == Names2InstrOpCodes.end()) + return true; + OpCode = InstrInfo->getValue(); + return false; +} + +void MIParser::initNames2Regs() { + if (!Names2Regs.empty()) + return; + // The '%noreg' register is the register 0. + Names2Regs.insert(std::make_pair("noreg", 0)); + const auto *TRI = MF.getSubtarget().getRegisterInfo(); + assert(TRI && "Expected target register info"); + for (unsigned I = 0, E = TRI->getNumRegs(); I < E; ++I) { + bool WasInserted = + Names2Regs.insert(std::make_pair(StringRef(TRI->getName(I)).lower(), I)) + .second; + (void)WasInserted; + assert(WasInserted && "Expected registers to be unique case-insensitively"); + } +} + +bool MIParser::getRegisterByName(StringRef RegName, unsigned &Reg) { + initNames2Regs(); + auto RegInfo = Names2Regs.find(RegName); + if (RegInfo == Names2Regs.end()) + return true; + Reg = RegInfo->getValue(); + return false; +} + +void MIParser::initNames2RegMasks() { + if (!Names2RegMasks.empty()) + return; + const auto *TRI = MF.getSubtarget().getRegisterInfo(); + assert(TRI && "Expected target register info"); + ArrayRef<const uint32_t *> RegMasks = TRI->getRegMasks(); + ArrayRef<const char *> RegMaskNames = TRI->getRegMaskNames(); + assert(RegMasks.size() == RegMaskNames.size()); + for (size_t I = 0, E = RegMasks.size(); I < E; ++I) + Names2RegMasks.insert( + std::make_pair(StringRef(RegMaskNames[I]).lower(), RegMasks[I])); +} + +const uint32_t *MIParser::getRegMask(StringRef Identifier) { + initNames2RegMasks(); + auto RegMaskInfo = Names2RegMasks.find(Identifier); + if (RegMaskInfo == Names2RegMasks.end()) + return nullptr; + return RegMaskInfo->getValue(); +} + +void MIParser::initNames2SubRegIndices() { + if (!Names2SubRegIndices.empty()) + return; + const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo(); + for (unsigned I = 1, E = TRI->getNumSubRegIndices(); I < E; ++I) + Names2SubRegIndices.insert( + std::make_pair(StringRef(TRI->getSubRegIndexName(I)).lower(), I)); +} + +unsigned MIParser::getSubRegIndex(StringRef Name) { + initNames2SubRegIndices(); + auto SubRegInfo = Names2SubRegIndices.find(Name); + if (SubRegInfo == Names2SubRegIndices.end()) + return 0; + return SubRegInfo->getValue(); +} + +bool llvm::parseMachineInstr(MachineInstr *&MI, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, SMDiagnostic &Error) { + return MIParser(SM, MF, Error, Src, PFS, IRSlots).parse(MI); +} + +bool llvm::parseMBBReference(MachineBasicBlock *&MBB, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, SMDiagnostic &Error) { + return MIParser(SM, MF, Error, Src, PFS, IRSlots).parseMBB(MBB); +} + +bool llvm::parseNamedRegisterReference(unsigned &Reg, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error) { + return MIParser(SM, MF, Error, Src, PFS, IRSlots).parseNamedRegister(Reg); +} diff --git a/contrib/llvm/lib/CodeGen/MIRParser/MIParser.h b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.h new file mode 100644 index 0000000..fca4c4e --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.h @@ -0,0 +1,51 @@ +//===- MIParser.h - Machine Instructions Parser ---------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file declares the function that parses the machine instructions. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_LIB_CODEGEN_MIRPARSER_MIPARSER_H +#define LLVM_LIB_CODEGEN_MIRPARSER_MIPARSER_H + +#include "llvm/ADT/DenseMap.h" +#include "llvm/ADT/StringRef.h" + +namespace llvm { + +class MachineBasicBlock; +class MachineInstr; +class MachineFunction; +struct SlotMapping; +class SMDiagnostic; +class SourceMgr; + +struct PerFunctionMIParsingState { + DenseMap<unsigned, MachineBasicBlock *> MBBSlots; + DenseMap<unsigned, unsigned> VirtualRegisterSlots; +}; + +bool parseMachineInstr(MachineInstr *&MI, SourceMgr &SM, MachineFunction &MF, + StringRef Src, const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, SMDiagnostic &Error); + +bool parseMBBReference(MachineBasicBlock *&MBB, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, SMDiagnostic &Error); + +bool parseNamedRegisterReference(unsigned &Reg, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error); + +} // end namespace llvm + +#endif diff --git a/contrib/llvm/lib/CodeGen/MIRParser/MIRParser.cpp b/contrib/llvm/lib/CodeGen/MIRParser/MIRParser.cpp new file mode 100644 index 0000000..16b0e16 --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MIRParser.cpp @@ -0,0 +1,513 @@ +//===- MIRParser.cpp - MIR serialization format parser implementation -----===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the class that parses the optional LLVM IR and machine +// functions that are stored in MIR files. +// +//===----------------------------------------------------------------------===// + +#include "llvm/CodeGen/MIRParser/MIRParser.h" +#include "MIParser.h" +#include "llvm/ADT/DenseMap.h" +#include "llvm/ADT/StringRef.h" +#include "llvm/ADT/StringMap.h" +#include "llvm/ADT/STLExtras.h" +#include "llvm/AsmParser/Parser.h" +#include "llvm/AsmParser/SlotMapping.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineRegisterInfo.h" +#include "llvm/CodeGen/MIRYamlMapping.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/DiagnosticInfo.h" +#include "llvm/IR/Instructions.h" +#include "llvm/IR/LLVMContext.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/ValueSymbolTable.h" +#include "llvm/Support/LineIterator.h" +#include "llvm/Support/SMLoc.h" +#include "llvm/Support/SourceMgr.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/YAMLTraits.h" +#include <memory> + +using namespace llvm; + +namespace llvm { + +/// This class implements the parsing of LLVM IR that's embedded inside a MIR +/// file. +class MIRParserImpl { + SourceMgr SM; + StringRef Filename; + LLVMContext &Context; + StringMap<std::unique_ptr<yaml::MachineFunction>> Functions; + SlotMapping IRSlots; + /// Maps from register class names to register classes. + StringMap<const TargetRegisterClass *> Names2RegClasses; + +public: + MIRParserImpl(std::unique_ptr<MemoryBuffer> Contents, StringRef Filename, + LLVMContext &Context); + + void reportDiagnostic(const SMDiagnostic &Diag); + + /// Report an error with the given message at unknown location. + /// + /// Always returns true. + bool error(const Twine &Message); + + /// Report an error with the given message at the given location. + /// + /// Always returns true. + bool error(SMLoc Loc, const Twine &Message); + + /// Report a given error with the location translated from the location in an + /// embedded string literal to a location in the MIR file. + /// + /// Always returns true. + bool error(const SMDiagnostic &Error, SMRange SourceRange); + + /// Try to parse the optional LLVM module and the machine functions in the MIR + /// file. + /// + /// Return null if an error occurred. + std::unique_ptr<Module> parse(); + + /// Parse the machine function in the current YAML document. + /// + /// \param NoLLVMIR - set to true when the MIR file doesn't have LLVM IR. + /// A dummy IR function is created and inserted into the given module when + /// this parameter is true. + /// + /// Return true if an error occurred. + bool parseMachineFunction(yaml::Input &In, Module &M, bool NoLLVMIR); + + /// Initialize the machine function to the state that's described in the MIR + /// file. + /// + /// Return true if error occurred. + bool initializeMachineFunction(MachineFunction &MF); + + /// Initialize the machine basic block using it's YAML representation. + /// + /// Return true if an error occurred. + bool initializeMachineBasicBlock(MachineFunction &MF, MachineBasicBlock &MBB, + const yaml::MachineBasicBlock &YamlMBB, + const PerFunctionMIParsingState &PFS); + + bool + initializeRegisterInfo(const MachineFunction &MF, + MachineRegisterInfo &RegInfo, + const yaml::MachineFunction &YamlMF, + DenseMap<unsigned, unsigned> &VirtualRegisterSlots); + + bool initializeFrameInfo(MachineFrameInfo &MFI, + const yaml::MachineFunction &YamlMF); + +private: + /// Return a MIR diagnostic converted from an MI string diagnostic. + SMDiagnostic diagFromMIStringDiag(const SMDiagnostic &Error, + SMRange SourceRange); + + /// Return a MIR diagnostic converted from an LLVM assembly diagnostic. + SMDiagnostic diagFromLLVMAssemblyDiag(const SMDiagnostic &Error, + SMRange SourceRange); + + /// Create an empty function with the given name. + void createDummyFunction(StringRef Name, Module &M); + + void initNames2RegClasses(const MachineFunction &MF); + + /// Check if the given identifier is a name of a register class. + /// + /// Return null if the name isn't a register class. + const TargetRegisterClass *getRegClass(const MachineFunction &MF, + StringRef Name); +}; + +} // end namespace llvm + +MIRParserImpl::MIRParserImpl(std::unique_ptr<MemoryBuffer> Contents, + StringRef Filename, LLVMContext &Context) + : SM(), Filename(Filename), Context(Context) { + SM.AddNewSourceBuffer(std::move(Contents), SMLoc()); +} + +bool MIRParserImpl::error(const Twine &Message) { + Context.diagnose(DiagnosticInfoMIRParser( + DS_Error, SMDiagnostic(Filename, SourceMgr::DK_Error, Message.str()))); + return true; +} + +bool MIRParserImpl::error(SMLoc Loc, const Twine &Message) { + Context.diagnose(DiagnosticInfoMIRParser( + DS_Error, SM.GetMessage(Loc, SourceMgr::DK_Error, Message))); + return true; +} + +bool MIRParserImpl::error(const SMDiagnostic &Error, SMRange SourceRange) { + assert(Error.getKind() == SourceMgr::DK_Error && "Expected an error"); + reportDiagnostic(diagFromMIStringDiag(Error, SourceRange)); + return true; +} + +void MIRParserImpl::reportDiagnostic(const SMDiagnostic &Diag) { + DiagnosticSeverity Kind; + switch (Diag.getKind()) { + case SourceMgr::DK_Error: + Kind = DS_Error; + break; + case SourceMgr::DK_Warning: + Kind = DS_Warning; + break; + case SourceMgr::DK_Note: + Kind = DS_Note; + break; + } + Context.diagnose(DiagnosticInfoMIRParser(Kind, Diag)); +} + +static void handleYAMLDiag(const SMDiagnostic &Diag, void *Context) { + reinterpret_cast<MIRParserImpl *>(Context)->reportDiagnostic(Diag); +} + +std::unique_ptr<Module> MIRParserImpl::parse() { + yaml::Input In(SM.getMemoryBuffer(SM.getMainFileID())->getBuffer(), + /*Ctxt=*/nullptr, handleYAMLDiag, this); + In.setContext(&In); + + if (!In.setCurrentDocument()) { + if (In.error()) + return nullptr; + // Create an empty module when the MIR file is empty. + return llvm::make_unique<Module>(Filename, Context); + } + + std::unique_ptr<Module> M; + bool NoLLVMIR = false; + // Parse the block scalar manually so that we can return unique pointer + // without having to go trough YAML traits. + if (const auto *BSN = + dyn_cast_or_null<yaml::BlockScalarNode>(In.getCurrentNode())) { + SMDiagnostic Error; + M = parseAssembly(MemoryBufferRef(BSN->getValue(), Filename), Error, + Context, &IRSlots); + if (!M) { + reportDiagnostic(diagFromLLVMAssemblyDiag(Error, BSN->getSourceRange())); + return M; + } + In.nextDocument(); + if (!In.setCurrentDocument()) + return M; + } else { + // Create an new, empty module. + M = llvm::make_unique<Module>(Filename, Context); + NoLLVMIR = true; + } + + // Parse the machine functions. + do { + if (parseMachineFunction(In, *M, NoLLVMIR)) + return nullptr; + In.nextDocument(); + } while (In.setCurrentDocument()); + + return M; +} + +bool MIRParserImpl::parseMachineFunction(yaml::Input &In, Module &M, + bool NoLLVMIR) { + auto MF = llvm::make_unique<yaml::MachineFunction>(); + yaml::yamlize(In, *MF, false); + if (In.error()) + return true; + auto FunctionName = MF->Name; + if (Functions.find(FunctionName) != Functions.end()) + return error(Twine("redefinition of machine function '") + FunctionName + + "'"); + Functions.insert(std::make_pair(FunctionName, std::move(MF))); + if (NoLLVMIR) + createDummyFunction(FunctionName, M); + else if (!M.getFunction(FunctionName)) + return error(Twine("function '") + FunctionName + + "' isn't defined in the provided LLVM IR"); + return false; +} + +void MIRParserImpl::createDummyFunction(StringRef Name, Module &M) { + auto &Context = M.getContext(); + Function *F = cast<Function>(M.getOrInsertFunction( + Name, FunctionType::get(Type::getVoidTy(Context), false))); + BasicBlock *BB = BasicBlock::Create(Context, "entry", F); + new UnreachableInst(Context, BB); +} + +bool MIRParserImpl::initializeMachineFunction(MachineFunction &MF) { + auto It = Functions.find(MF.getName()); + if (It == Functions.end()) + return error(Twine("no machine function information for function '") + + MF.getName() + "' in the MIR file"); + // TODO: Recreate the machine function. + const yaml::MachineFunction &YamlMF = *It->getValue(); + if (YamlMF.Alignment) + MF.setAlignment(YamlMF.Alignment); + MF.setExposesReturnsTwice(YamlMF.ExposesReturnsTwice); + MF.setHasInlineAsm(YamlMF.HasInlineAsm); + PerFunctionMIParsingState PFS; + if (initializeRegisterInfo(MF, MF.getRegInfo(), YamlMF, + PFS.VirtualRegisterSlots)) + return true; + if (initializeFrameInfo(*MF.getFrameInfo(), YamlMF)) + return true; + + const auto &F = *MF.getFunction(); + for (const auto &YamlMBB : YamlMF.BasicBlocks) { + const BasicBlock *BB = nullptr; + const yaml::StringValue &Name = YamlMBB.Name; + if (!Name.Value.empty()) { + BB = dyn_cast_or_null<BasicBlock>( + F.getValueSymbolTable().lookup(Name.Value)); + if (!BB) + return error(Name.SourceRange.Start, + Twine("basic block '") + Name.Value + + "' is not defined in the function '" + MF.getName() + + "'"); + } + auto *MBB = MF.CreateMachineBasicBlock(BB); + MF.insert(MF.end(), MBB); + bool WasInserted = + PFS.MBBSlots.insert(std::make_pair(YamlMBB.ID, MBB)).second; + if (!WasInserted) + return error(Twine("redefinition of machine basic block with id #") + + Twine(YamlMBB.ID)); + } + + if (YamlMF.BasicBlocks.empty()) + return error(Twine("machine function '") + Twine(MF.getName()) + + "' requires at least one machine basic block in its body"); + // Initialize the machine basic blocks after creating them all so that the + // machine instructions parser can resolve the MBB references. + unsigned I = 0; + for (const auto &YamlMBB : YamlMF.BasicBlocks) { + if (initializeMachineBasicBlock(MF, *MF.getBlockNumbered(I++), YamlMBB, + PFS)) + return true; + } + return false; +} + +bool MIRParserImpl::initializeMachineBasicBlock( + MachineFunction &MF, MachineBasicBlock &MBB, + const yaml::MachineBasicBlock &YamlMBB, + const PerFunctionMIParsingState &PFS) { + MBB.setAlignment(YamlMBB.Alignment); + if (YamlMBB.AddressTaken) + MBB.setHasAddressTaken(); + MBB.setIsLandingPad(YamlMBB.IsLandingPad); + SMDiagnostic Error; + // Parse the successors. + for (const auto &MBBSource : YamlMBB.Successors) { + MachineBasicBlock *SuccMBB = nullptr; + if (parseMBBReference(SuccMBB, SM, MF, MBBSource.Value, PFS, IRSlots, + Error)) + return error(Error, MBBSource.SourceRange); + // TODO: Report an error when adding the same successor more than once. + MBB.addSuccessor(SuccMBB); + } + // Parse the liveins. + for (const auto &LiveInSource : YamlMBB.LiveIns) { + unsigned Reg = 0; + if (parseNamedRegisterReference(Reg, SM, MF, LiveInSource.Value, PFS, + IRSlots, Error)) + return error(Error, LiveInSource.SourceRange); + MBB.addLiveIn(Reg); + } + // Parse the instructions. + for (const auto &MISource : YamlMBB.Instructions) { + MachineInstr *MI = nullptr; + if (parseMachineInstr(MI, SM, MF, MISource.Value, PFS, IRSlots, Error)) + return error(Error, MISource.SourceRange); + MBB.insert(MBB.end(), MI); + } + return false; +} + +bool MIRParserImpl::initializeRegisterInfo( + const MachineFunction &MF, MachineRegisterInfo &RegInfo, + const yaml::MachineFunction &YamlMF, + DenseMap<unsigned, unsigned> &VirtualRegisterSlots) { + assert(RegInfo.isSSA()); + if (!YamlMF.IsSSA) + RegInfo.leaveSSA(); + assert(RegInfo.tracksLiveness()); + if (!YamlMF.TracksRegLiveness) + RegInfo.invalidateLiveness(); + RegInfo.enableSubRegLiveness(YamlMF.TracksSubRegLiveness); + + // Parse the virtual register information. + for (const auto &VReg : YamlMF.VirtualRegisters) { + const auto *RC = getRegClass(MF, VReg.Class.Value); + if (!RC) + return error(VReg.Class.SourceRange.Start, + Twine("use of undefined register class '") + + VReg.Class.Value + "'"); + unsigned Reg = RegInfo.createVirtualRegister(RC); + // TODO: Report an error when the same virtual register with the same ID is + // redefined. + VirtualRegisterSlots.insert(std::make_pair(VReg.ID, Reg)); + } + return false; +} + +bool MIRParserImpl::initializeFrameInfo(MachineFrameInfo &MFI, + const yaml::MachineFunction &YamlMF) { + const yaml::MachineFrameInfo &YamlMFI = YamlMF.FrameInfo; + MFI.setFrameAddressIsTaken(YamlMFI.IsFrameAddressTaken); + MFI.setReturnAddressIsTaken(YamlMFI.IsReturnAddressTaken); + MFI.setHasStackMap(YamlMFI.HasStackMap); + MFI.setHasPatchPoint(YamlMFI.HasPatchPoint); + MFI.setStackSize(YamlMFI.StackSize); + MFI.setOffsetAdjustment(YamlMFI.OffsetAdjustment); + if (YamlMFI.MaxAlignment) + MFI.ensureMaxAlignment(YamlMFI.MaxAlignment); + MFI.setAdjustsStack(YamlMFI.AdjustsStack); + MFI.setHasCalls(YamlMFI.HasCalls); + MFI.setMaxCallFrameSize(YamlMFI.MaxCallFrameSize); + MFI.setHasOpaqueSPAdjustment(YamlMFI.HasOpaqueSPAdjustment); + MFI.setHasVAStart(YamlMFI.HasVAStart); + MFI.setHasMustTailInVarArgFunc(YamlMFI.HasMustTailInVarArgFunc); + + // Initialize the fixed frame objects. + for (const auto &Object : YamlMF.FixedStackObjects) { + int ObjectIdx; + if (Object.Type != yaml::FixedMachineStackObject::SpillSlot) + ObjectIdx = MFI.CreateFixedObject(Object.Size, Object.Offset, + Object.IsImmutable, Object.IsAliased); + else + ObjectIdx = MFI.CreateFixedSpillStackObject(Object.Size, Object.Offset); + MFI.setObjectAlignment(ObjectIdx, Object.Alignment); + // TODO: Store the mapping between fixed object IDs and object indices to + // parse fixed stack object references correctly. + } + + // Initialize the ordinary frame objects. + for (const auto &Object : YamlMF.StackObjects) { + int ObjectIdx; + if (Object.Type == yaml::MachineStackObject::VariableSized) + ObjectIdx = + MFI.CreateVariableSizedObject(Object.Alignment, /*Alloca=*/nullptr); + else + ObjectIdx = MFI.CreateStackObject( + Object.Size, Object.Alignment, + Object.Type == yaml::MachineStackObject::SpillSlot); + MFI.setObjectOffset(ObjectIdx, Object.Offset); + // TODO: Store the mapping between object IDs and object indices to parse + // stack object references correctly. + } + return false; +} + +SMDiagnostic MIRParserImpl::diagFromMIStringDiag(const SMDiagnostic &Error, + SMRange SourceRange) { + assert(SourceRange.isValid() && "Invalid source range"); + SMLoc Loc = SourceRange.Start; + bool HasQuote = Loc.getPointer() < SourceRange.End.getPointer() && + *Loc.getPointer() == '\''; + // Translate the location of the error from the location in the MI string to + // the corresponding location in the MIR file. + Loc = Loc.getFromPointer(Loc.getPointer() + Error.getColumnNo() + + (HasQuote ? 1 : 0)); + + // TODO: Translate any source ranges as well. + return SM.GetMessage(Loc, Error.getKind(), Error.getMessage(), None, + Error.getFixIts()); +} + +SMDiagnostic MIRParserImpl::diagFromLLVMAssemblyDiag(const SMDiagnostic &Error, + SMRange SourceRange) { + assert(SourceRange.isValid()); + + // Translate the location of the error from the location in the llvm IR string + // to the corresponding location in the MIR file. + auto LineAndColumn = SM.getLineAndColumn(SourceRange.Start); + unsigned Line = LineAndColumn.first + Error.getLineNo() - 1; + unsigned Column = Error.getColumnNo(); + StringRef LineStr = Error.getLineContents(); + SMLoc Loc = Error.getLoc(); + + // Get the full line and adjust the column number by taking the indentation of + // LLVM IR into account. + for (line_iterator L(*SM.getMemoryBuffer(SM.getMainFileID()), false), E; + L != E; ++L) { + if (L.line_number() == Line) { + LineStr = *L; + Loc = SMLoc::getFromPointer(LineStr.data()); + auto Indent = LineStr.find(Error.getLineContents()); + if (Indent != StringRef::npos) + Column += Indent; + break; + } + } + + return SMDiagnostic(SM, Loc, Filename, Line, Column, Error.getKind(), + Error.getMessage(), LineStr, Error.getRanges(), + Error.getFixIts()); +} + +void MIRParserImpl::initNames2RegClasses(const MachineFunction &MF) { + if (!Names2RegClasses.empty()) + return; + const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo(); + for (unsigned I = 0, E = TRI->getNumRegClasses(); I < E; ++I) { + const auto *RC = TRI->getRegClass(I); + Names2RegClasses.insert( + std::make_pair(StringRef(TRI->getRegClassName(RC)).lower(), RC)); + } +} + +const TargetRegisterClass *MIRParserImpl::getRegClass(const MachineFunction &MF, + StringRef Name) { + initNames2RegClasses(MF); + auto RegClassInfo = Names2RegClasses.find(Name); + if (RegClassInfo == Names2RegClasses.end()) + return nullptr; + return RegClassInfo->getValue(); +} + +MIRParser::MIRParser(std::unique_ptr<MIRParserImpl> Impl) + : Impl(std::move(Impl)) {} + +MIRParser::~MIRParser() {} + +std::unique_ptr<Module> MIRParser::parseLLVMModule() { return Impl->parse(); } + +bool MIRParser::initializeMachineFunction(MachineFunction &MF) { + return Impl->initializeMachineFunction(MF); +} + +std::unique_ptr<MIRParser> llvm::createMIRParserFromFile(StringRef Filename, + SMDiagnostic &Error, + LLVMContext &Context) { + auto FileOrErr = MemoryBuffer::getFile(Filename); + if (std::error_code EC = FileOrErr.getError()) { + Error = SMDiagnostic(Filename, SourceMgr::DK_Error, + "Could not open input file: " + EC.message()); + return nullptr; + } + return createMIRParser(std::move(FileOrErr.get()), Context); +} + +std::unique_ptr<MIRParser> +llvm::createMIRParser(std::unique_ptr<MemoryBuffer> Contents, + LLVMContext &Context) { + auto Filename = Contents->getBufferIdentifier(); + return llvm::make_unique<MIRParser>( + llvm::make_unique<MIRParserImpl>(std::move(Contents), Filename, Context)); +} |