diff options
Diffstat (limited to 'contrib/llvm/lib/CodeGen/MIRParser')
| -rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MILexer.cpp | 604 | ||||
| -rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MILexer.h | 193 | ||||
| -rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MIParser.cpp | 2011 | ||||
| -rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MIParser.h | 99 | ||||
| -rw-r--r-- | contrib/llvm/lib/CodeGen/MIRParser/MIRParser.cpp | 739 |
5 files changed, 3646 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..28f9d4e --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MILexer.cpp @@ -0,0 +1,604 @@ +//===- 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/StringExtras.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 + +MIToken &MIToken::reset(TokenKind Kind, StringRef Range) { + this->Kind = Kind; + this->Range = Range; + return *this; +} + +MIToken &MIToken::setStringValue(StringRef StrVal) { + StringValue = StrVal; + return *this; +} + +MIToken &MIToken::setOwnedStringValue(std::string StrVal) { + StringValueStorage = std::move(StrVal); + StringValue = StringValueStorage; + return *this; +} + +MIToken &MIToken::setIntegerValue(APSInt IntVal) { + this->IntVal = std::move(IntVal); + return *this; +} + +/// Skip the leading whitespace characters and return the updated cursor. +static Cursor skipWhitespace(Cursor C) { + while (isblank(C.peek())) + C.advance(); + return C; +} + +static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; } + +/// Skip a line comment and return the updated cursor. +static Cursor skipComment(Cursor C) { + if (C.peek() != ';') + return C; + while (!isNewlineChar(C.peek()) && !C.isEOF()) + C.advance(); + return C; +} + +/// Return true if the given character satisfies the following regular +/// expression: [-a-zA-Z$._0-9] +static bool isIdentifierChar(char C) { + return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' || + C == '$'; +} + +/// Unescapes the given string value. +/// +/// Expects the string value to be quoted. +static std::string unescapeQuotedString(StringRef Value) { + assert(Value.front() == '"' && Value.back() == '"'); + Cursor C = Cursor(Value.substr(1, Value.size() - 2)); + + std::string Str; + Str.reserve(C.remaining().size()); + while (!C.isEOF()) { + char Char = C.peek(); + if (Char == '\\') { + if (C.peek(1) == '\\') { + // Two '\' become one + Str += '\\'; + C.advance(2); + continue; + } + if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) { + Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2)); + C.advance(3); + continue; + } + } + Str += Char; + C.advance(); + } + return Str; +} + +/// Lex a string constant using the following regular expression: \"[^\"]*\" +static Cursor lexStringConstant( + Cursor C, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + assert(C.peek() == '"'); + for (C.advance(); C.peek() != '"'; C.advance()) { + if (C.isEOF() || isNewlineChar(C.peek())) { + ErrorCallback( + C.location(), + "end of machine instruction reached before the closing '\"'"); + return None; + } + } + C.advance(); + return C; +} + +static Cursor lexName( + Cursor C, MIToken &Token, MIToken::TokenKind Type, unsigned PrefixLength, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + auto Range = C; + C.advance(PrefixLength); + if (C.peek() == '"') { + if (Cursor R = lexStringConstant(C, ErrorCallback)) { + StringRef String = Range.upto(R); + Token.reset(Type, String) + .setOwnedStringValue( + unescapeQuotedString(String.drop_front(PrefixLength))); + return R; + } + Token.reset(MIToken::Error, Range.remaining()); + return Range; + } + while (isIdentifierChar(C.peek())) + C.advance(); + Token.reset(Type, Range.upto(C)) + .setStringValue(Range.upto(C).drop_front(PrefixLength)); + return C; +} + +static Cursor maybeLexIntegerType(Cursor C, MIToken &Token) { + if (C.peek() != 'i' || !isdigit(C.peek(1))) + return None; + auto Range = C; + C.advance(); // Skip 'i' + while (isdigit(C.peek())) + C.advance(); + Token.reset(MIToken::IntegerType, Range.upto(C)); + return 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("def", MIToken::kw_def) + .Case("dead", MIToken::kw_dead) + .Case("killed", MIToken::kw_killed) + .Case("undef", MIToken::kw_undef) + .Case("internal", MIToken::kw_internal) + .Case("early-clobber", MIToken::kw_early_clobber) + .Case("debug-use", MIToken::kw_debug_use) + .Case("tied-def", MIToken::kw_tied_def) + .Case("frame-setup", MIToken::kw_frame_setup) + .Case("debug-location", MIToken::kw_debug_location) + .Case(".cfi_same_value", MIToken::kw_cfi_same_value) + .Case(".cfi_offset", MIToken::kw_cfi_offset) + .Case(".cfi_def_cfa_register", MIToken::kw_cfi_def_cfa_register) + .Case(".cfi_def_cfa_offset", MIToken::kw_cfi_def_cfa_offset) + .Case(".cfi_def_cfa", MIToken::kw_cfi_def_cfa) + .Case("blockaddress", MIToken::kw_blockaddress) + .Case("target-index", MIToken::kw_target_index) + .Case("half", MIToken::kw_half) + .Case("float", MIToken::kw_float) + .Case("double", MIToken::kw_double) + .Case("x86_fp80", MIToken::kw_x86_fp80) + .Case("fp128", MIToken::kw_fp128) + .Case("ppc_fp128", MIToken::kw_ppc_fp128) + .Case("target-flags", MIToken::kw_target_flags) + .Case("volatile", MIToken::kw_volatile) + .Case("non-temporal", MIToken::kw_non_temporal) + .Case("invariant", MIToken::kw_invariant) + .Case("align", MIToken::kw_align) + .Case("stack", MIToken::kw_stack) + .Case("got", MIToken::kw_got) + .Case("jump-table", MIToken::kw_jump_table) + .Case("constant-pool", MIToken::kw_constant_pool) + .Case("call-entry", MIToken::kw_call_entry) + .Case("liveout", MIToken::kw_liveout) + .Case("address-taken", MIToken::kw_address_taken) + .Case("landing-pad", MIToken::kw_landing_pad) + .Case("liveins", MIToken::kw_liveins) + .Case("successors", MIToken::kw_successors) + .Default(MIToken::Identifier); +} + +static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) { + if (!isalpha(C.peek()) && C.peek() != '_' && C.peek() != '.') + return None; + auto Range = C; + while (isIdentifierChar(C.peek())) + C.advance(); + auto Identifier = Range.upto(C); + Token.reset(getIdentifierKind(Identifier), Identifier) + .setStringValue(Identifier); + return C; +} + +static Cursor maybeLexMachineBasicBlock( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + bool IsReference = C.remaining().startswith("%bb."); + if (!IsReference && !C.remaining().startswith("bb.")) + return None; + auto Range = C; + unsigned PrefixLength = IsReference ? 4 : 3; + C.advance(PrefixLength); // Skip '%bb.' or 'bb.' + if (!isdigit(C.peek())) { + Token.reset(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 = PrefixLength + Number.size(); // Drop '%bb.<id>' + if (C.peek() == '.') { + C.advance(); // Skip '.' + ++StringOffset; + while (isIdentifierChar(C.peek())) + C.advance(); + } + Token.reset(IsReference ? MIToken::MachineBasicBlock + : MIToken::MachineBasicBlockLabel, + Range.upto(C)) + .setIntegerValue(APSInt(Number)) + .setStringValue(Range.upto(C).drop_front(StringOffset)); + return C; +} + +static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule, + MIToken::TokenKind Kind) { + if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size()))) + return None; + auto Range = C; + C.advance(Rule.size()); + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C))); + return C; +} + +static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule, + MIToken::TokenKind Kind) { + if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size()))) + return None; + auto Range = C; + C.advance(Rule.size()); + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + StringRef Number = NumberRange.upto(C); + unsigned StringOffset = Rule.size() + Number.size(); + if (C.peek() == '.') { + C.advance(); + ++StringOffset; + while (isIdentifierChar(C.peek())) + C.advance(); + } + Token.reset(Kind, Range.upto(C)) + .setIntegerValue(APSInt(Number)) + .setStringValue(Range.upto(C).drop_front(StringOffset)); + return C; +} + +static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) { + return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex); +} + +static Cursor maybeLexStackObject(Cursor C, MIToken &Token) { + return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject); +} + +static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) { + return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject); +} + +static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) { + return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem); +} + +static Cursor maybeLexIRBlock( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + const StringRef Rule = "%ir-block."; + if (!C.remaining().startswith(Rule)) + return None; + if (isdigit(C.peek(Rule.size()))) + return maybeLexIndex(C, Token, Rule, MIToken::IRBlock); + return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback); +} + +static Cursor maybeLexIRValue( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + const StringRef Rule = "%ir."; + if (!C.remaining().startswith(Rule)) + return None; + if (isdigit(C.peek(Rule.size()))) + return maybeLexIndex(C, Token, Rule, MIToken::IRValue); + return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback); +} + +static Cursor lexVirtualRegister(Cursor C, MIToken &Token) { + auto Range = C; + C.advance(); // Skip '%' + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + Token.reset(MIToken::VirtualRegister, Range.upto(C)) + .setIntegerValue(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.reset(MIToken::NamedRegister, Range.upto(C)) + .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%' + return C; +} + +static Cursor maybeLexGlobalValue( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + if (C.peek() != '@') + return None; + if (!isdigit(C.peek(1))) + return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1, + ErrorCallback); + auto Range = C; + C.advance(1); // Skip the '@' + auto NumberRange = C; + while (isdigit(C.peek())) + C.advance(); + Token.reset(MIToken::GlobalValue, Range.upto(C)) + .setIntegerValue(APSInt(NumberRange.upto(C))); + return C; +} + +static Cursor maybeLexExternalSymbol( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + if (C.peek() != '$') + return None; + return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1, + ErrorCallback); +} + +static bool isValidHexFloatingPointPrefix(char C) { + return C == 'H' || C == 'K' || C == 'L' || C == 'M'; +} + +static Cursor maybeLexHexFloatingPointLiteral(Cursor C, MIToken &Token) { + if (C.peek() != '0' || C.peek(1) != 'x') + return None; + Cursor Range = C; + C.advance(2); // Skip '0x' + if (isValidHexFloatingPointPrefix(C.peek())) + C.advance(); + while (isxdigit(C.peek())) + C.advance(); + Token.reset(MIToken::FloatingPointLiteral, Range.upto(C)); + return C; +} + +static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) { + C.advance(); + // Skip over [0-9]*([eE][-+]?[0-9]+)? + while (isdigit(C.peek())) + C.advance(); + if ((C.peek() == 'e' || C.peek() == 'E') && + (isdigit(C.peek(1)) || + ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) { + C.advance(2); + while (isdigit(C.peek())) + C.advance(); + } + Token.reset(MIToken::FloatingPointLiteral, Range.upto(C)); + return C; +} + +static Cursor maybeLexNumericalLiteral(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(); + if (C.peek() == '.') + return lexFloatingPointLiteral(Range, C, Token); + StringRef StrVal = Range.upto(C); + Token.reset(MIToken::IntegerLiteral, StrVal).setIntegerValue(APSInt(StrVal)); + return C; +} + +static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) { + return StringSwitch<MIToken::TokenKind>(Identifier) + .Case("!tbaa", MIToken::md_tbaa) + .Case("!alias.scope", MIToken::md_alias_scope) + .Case("!noalias", MIToken::md_noalias) + .Case("!range", MIToken::md_range) + .Default(MIToken::Error); +} + +static Cursor maybeLexExlaim( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + if (C.peek() != '!') + return None; + auto Range = C; + C.advance(1); + if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) { + Token.reset(MIToken::exclaim, Range.upto(C)); + return C; + } + while (isIdentifierChar(C.peek())) + C.advance(); + StringRef StrVal = Range.upto(C); + Token.reset(getMetadataKeywordKind(StrVal), StrVal); + if (Token.isError()) + ErrorCallback(Token.location(), + "use of unknown metadata keyword '" + StrVal + "'"); + return C; +} + +static MIToken::TokenKind symbolToken(char C) { + switch (C) { + case ',': + return MIToken::comma; + case '=': + return MIToken::equal; + case ':': + return MIToken::colon; + case '(': + return MIToken::lparen; + case ')': + return MIToken::rparen; + case '{': + return MIToken::lbrace; + case '}': + return MIToken::rbrace; + case '+': + return MIToken::plus; + case '-': + return MIToken::minus; + default: + return MIToken::Error; + } +} + +static Cursor maybeLexSymbol(Cursor C, MIToken &Token) { + MIToken::TokenKind Kind; + unsigned Length = 1; + if (C.peek() == ':' && C.peek(1) == ':') { + Kind = MIToken::coloncolon; + Length = 2; + } else + Kind = symbolToken(C.peek()); + if (Kind == MIToken::Error) + return None; + auto Range = C; + C.advance(Length); + Token.reset(Kind, Range.upto(C)); + return C; +} + +static Cursor maybeLexNewline(Cursor C, MIToken &Token) { + if (!isNewlineChar(C.peek())) + return None; + auto Range = C; + C.advance(); + Token.reset(MIToken::Newline, Range.upto(C)); + return C; +} + +static Cursor maybeLexEscapedIRValue( + Cursor C, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + if (C.peek() != '`') + return None; + auto Range = C; + C.advance(); + auto StrRange = C; + while (C.peek() != '`') { + if (C.isEOF() || isNewlineChar(C.peek())) { + ErrorCallback( + C.location(), + "end of machine instruction reached before the closing '`'"); + Token.reset(MIToken::Error, Range.remaining()); + return C; + } + C.advance(); + } + StringRef Value = StrRange.upto(C); + C.advance(); + Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value); + return C; +} + +StringRef llvm::lexMIToken( + StringRef Source, MIToken &Token, + function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) { + auto C = skipComment(skipWhitespace(Cursor(Source))); + if (C.isEOF()) { + Token.reset(MIToken::Eof, C.remaining()); + return C.remaining(); + } + + if (Cursor R = maybeLexIntegerType(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexIdentifier(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexJumpTableIndex(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexStackObject(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexFixedStackObject(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexConstantPoolItem(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexRegister(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexHexFloatingPointLiteral(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexNumericalLiteral(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexExlaim(C, Token, ErrorCallback)) + return R.remaining(); + if (Cursor R = maybeLexSymbol(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexNewline(C, Token)) + return R.remaining(); + if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback)) + return R.remaining(); + + Token.reset(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..ff54aa3 --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MILexer.h @@ -0,0 +1,193 @@ +//===- 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, + Newline, + + // Tokens with no info. + comma, + equal, + underscore, + colon, + coloncolon, + exclaim, + lparen, + rparen, + lbrace, + rbrace, + plus, + minus, + + // Keywords + kw_implicit, + kw_implicit_define, + kw_def, + kw_dead, + kw_killed, + kw_undef, + kw_internal, + kw_early_clobber, + kw_debug_use, + kw_tied_def, + kw_frame_setup, + kw_debug_location, + kw_cfi_same_value, + kw_cfi_offset, + kw_cfi_def_cfa_register, + kw_cfi_def_cfa_offset, + kw_cfi_def_cfa, + kw_blockaddress, + kw_target_index, + kw_half, + kw_float, + kw_double, + kw_x86_fp80, + kw_fp128, + kw_ppc_fp128, + kw_target_flags, + kw_volatile, + kw_non_temporal, + kw_invariant, + kw_align, + kw_stack, + kw_got, + kw_jump_table, + kw_constant_pool, + kw_call_entry, + kw_liveout, + kw_address_taken, + kw_landing_pad, + kw_liveins, + kw_successors, + + // Named metadata keywords + md_tbaa, + md_alias_scope, + md_noalias, + md_range, + + // Identifier tokens + Identifier, + IntegerType, + NamedRegister, + MachineBasicBlockLabel, + MachineBasicBlock, + StackObject, + FixedStackObject, + NamedGlobalValue, + GlobalValue, + ExternalSymbol, + + // Other tokens + IntegerLiteral, + FloatingPointLiteral, + VirtualRegister, + ConstantPoolItem, + JumpTableIndex, + NamedIRBlock, + IRBlock, + NamedIRValue, + IRValue, + QuotedIRValue // `<constant value>` + }; + +private: + TokenKind Kind; + StringRef Range; + StringRef StringValue; + std::string StringValueStorage; + APSInt IntVal; + +public: + MIToken() : Kind(Error) {} + + MIToken &reset(TokenKind Kind, StringRef Range); + + MIToken &setStringValue(StringRef StrVal); + MIToken &setOwnedStringValue(std::string StrVal); + MIToken &setIntegerValue(APSInt IntVal); + + TokenKind kind() const { return Kind; } + + bool isError() const { return Kind == Error; } + + bool isNewlineOrEOF() const { return Kind == Newline || Kind == Eof; } + + bool isErrorOrEOF() const { return Kind == Error || Kind == Eof; } + + bool isRegister() const { + return Kind == NamedRegister || Kind == underscore || + Kind == VirtualRegister; + } + + bool isRegisterFlag() const { + return Kind == kw_implicit || Kind == kw_implicit_define || + Kind == kw_def || Kind == kw_dead || Kind == kw_killed || + Kind == kw_undef || Kind == kw_internal || + Kind == kw_early_clobber || Kind == kw_debug_use; + } + + bool isMemoryOperandFlag() const { + return Kind == kw_volatile || Kind == kw_non_temporal || + Kind == kw_invariant; + } + + bool is(TokenKind K) const { return Kind == K; } + + bool isNot(TokenKind K) const { return Kind != K; } + + StringRef::iterator location() const { return Range.begin(); } + + StringRef range() const { return Range; } + + /// Return the token's string value. + StringRef stringValue() const { return StringValue; } + + const APSInt &integerValue() const { return IntVal; } + + bool hasIntegerValue() const { + return Kind == IntegerLiteral || Kind == MachineBasicBlock || + Kind == MachineBasicBlockLabel || Kind == StackObject || + Kind == FixedStackObject || Kind == GlobalValue || + Kind == VirtualRegister || Kind == ConstantPoolItem || + Kind == JumpTableIndex || Kind == IRBlock || Kind == IRValue; + } +}; + +/// 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..f2f6584 --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.cpp @@ -0,0 +1,2011 @@ +//===- 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/Parser.h" +#include "llvm/AsmParser/SlotMapping.h" +#include "llvm/CodeGen/MachineBasicBlock.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineInstr.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineMemOperand.h" +#include "llvm/CodeGen/MachineModuleInfo.h" +#include "llvm/IR/Instructions.h" +#include "llvm/IR/Constants.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/ModuleSlotTracker.h" +#include "llvm/IR/ValueSymbolTable.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 and other attributes. +struct ParsedMachineOperand { + MachineOperand Operand; + StringRef::iterator Begin; + StringRef::iterator End; + Optional<unsigned> TiedDefIdx; + + ParsedMachineOperand(const MachineOperand &Operand, StringRef::iterator Begin, + StringRef::iterator End, Optional<unsigned> &TiedDefIdx) + : Operand(Operand), Begin(Begin), End(End), TiedDefIdx(TiedDefIdx) { + if (TiedDefIdx) + assert(Operand.isReg() && Operand.isUse() && + "Only used register operands can be tied"); + } +}; + +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; + /// Maps from slot numbers to function's unnamed basic blocks. + DenseMap<unsigned, const BasicBlock *> Slots2BasicBlocks; + /// Maps from slot numbers to function's unnamed values. + DenseMap<unsigned, const Value *> Slots2Values; + /// Maps from target index names to target indices. + StringMap<int> Names2TargetIndices; + /// Maps from direct target flag names to the direct target flag values. + StringMap<unsigned> Names2DirectTargetFlags; + /// Maps from direct target flag names to the bitmask target flag values. + StringMap<unsigned> Names2BitmaskTargetFlags; + +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 + parseBasicBlockDefinitions(DenseMap<unsigned, MachineBasicBlock *> &MBBSlots); + bool parseBasicBlocks(); + bool parse(MachineInstr *&MI); + bool parseStandaloneMBB(MachineBasicBlock *&MBB); + bool parseStandaloneNamedRegister(unsigned &Reg); + bool parseStandaloneVirtualRegister(unsigned &Reg); + bool parseStandaloneStackObject(int &FI); + bool parseStandaloneMDNode(MDNode *&Node); + + bool + parseBasicBlockDefinition(DenseMap<unsigned, MachineBasicBlock *> &MBBSlots); + bool parseBasicBlock(MachineBasicBlock &MBB); + bool parseBasicBlockLiveins(MachineBasicBlock &MBB); + bool parseBasicBlockSuccessors(MachineBasicBlock &MBB); + + bool parseRegister(unsigned &Reg); + bool parseRegisterFlag(unsigned &Flags); + bool parseSubRegisterIndex(unsigned &SubReg); + bool parseRegisterTiedDefIndex(unsigned &TiedDefIdx); + bool parseRegisterOperand(MachineOperand &Dest, + Optional<unsigned> &TiedDefIdx, bool IsDef = false); + bool parseImmediateOperand(MachineOperand &Dest); + bool parseIRConstant(StringRef::iterator Loc, StringRef Source, + const Constant *&C); + bool parseIRConstant(StringRef::iterator Loc, const Constant *&C); + bool parseTypedImmediateOperand(MachineOperand &Dest); + bool parseFPImmediateOperand(MachineOperand &Dest); + bool parseMBBReference(MachineBasicBlock *&MBB); + bool parseMBBOperand(MachineOperand &Dest); + bool parseStackFrameIndex(int &FI); + bool parseStackObjectOperand(MachineOperand &Dest); + bool parseFixedStackFrameIndex(int &FI); + bool parseFixedStackObjectOperand(MachineOperand &Dest); + bool parseGlobalValue(GlobalValue *&GV); + bool parseGlobalAddressOperand(MachineOperand &Dest); + bool parseConstantPoolIndexOperand(MachineOperand &Dest); + bool parseJumpTableIndexOperand(MachineOperand &Dest); + bool parseExternalSymbolOperand(MachineOperand &Dest); + bool parseMDNode(MDNode *&Node); + bool parseMetadataOperand(MachineOperand &Dest); + bool parseCFIOffset(int &Offset); + bool parseCFIRegister(unsigned &Reg); + bool parseCFIOperand(MachineOperand &Dest); + bool parseIRBlock(BasicBlock *&BB, const Function &F); + bool parseBlockAddressOperand(MachineOperand &Dest); + bool parseTargetIndexOperand(MachineOperand &Dest); + bool parseLiveoutRegisterMaskOperand(MachineOperand &Dest); + bool parseMachineOperand(MachineOperand &Dest, + Optional<unsigned> &TiedDefIdx); + bool parseMachineOperandAndTargetFlags(MachineOperand &Dest, + Optional<unsigned> &TiedDefIdx); + bool parseOffset(int64_t &Offset); + bool parseAlignment(unsigned &Alignment); + bool parseOperandsOffset(MachineOperand &Op); + bool parseIRValue(const Value *&V); + bool parseMemoryOperandFlag(unsigned &Flags); + bool parseMemoryPseudoSourceValue(const PseudoSourceValue *&PSV); + bool parseMachinePointerInfo(MachinePointerInfo &Dest); + bool parseMachineMemoryOperand(MachineMemOperand *&Dest); + +private: + /// Convert the integer literal in the current token into an unsigned integer. + /// + /// Return true if an error occurred. + bool getUnsigned(unsigned &Result); + + /// Convert the integer literal in the current token into an uint64. + /// + /// Return true if an error occurred. + bool getUint64(uint64_t &Result); + + /// If the current token is of the given kind, consume it and return false. + /// Otherwise report an error and return true. + bool expectAndConsume(MIToken::TokenKind TokenKind); + + /// If the current token is of the given kind, consume it and return true. + /// Otherwise return false. + bool consumeIfPresent(MIToken::TokenKind TokenKind); + + 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, unsigned &Flags); + + bool assignRegisterTies(MachineInstr &MI, + ArrayRef<ParsedMachineOperand> Operands); + + bool verifyImplicitOperands(ArrayRef<ParsedMachineOperand> 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); + + const BasicBlock *getIRBlock(unsigned Slot); + const BasicBlock *getIRBlock(unsigned Slot, const Function &F); + + const Value *getIRValue(unsigned Slot); + + void initNames2TargetIndices(); + + /// Try to convert a name of target index to the corresponding target index. + /// + /// Return true if the name isn't a name of a target index. + bool getTargetIndex(StringRef Name, int &Index); + + void initNames2DirectTargetFlags(); + + /// Try to convert a name of a direct target flag to the corresponding + /// target flag. + /// + /// Return true if the name isn't a name of a direct flag. + bool getDirectTargetFlag(StringRef Name, unsigned &Flag); + + void initNames2BitmaskTargetFlags(); + + /// Try to convert a name of a bitmask target flag to the corresponding + /// target flag. + /// + /// Return true if the name isn't a name of a bitmask target flag. + bool getBitmaskTargetFlag(StringRef Name, unsigned &Flag); +}; + +} // 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), + 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())); + const MemoryBuffer &Buffer = *SM.getMemoryBuffer(SM.getMainFileID()); + if (Loc >= Buffer.getBufferStart() && Loc <= Buffer.getBufferEnd()) { + // Create an ordinary diagnostic when the source manager's buffer is the + // source string. + Error = SM.GetMessage(SMLoc::getFromPointer(Loc), SourceMgr::DK_Error, Msg); + return true; + } + // Create a diagnostic for a YAML string literal. + Error = SMDiagnostic(SM, SMLoc(), Buffer.getBufferIdentifier(), 1, + Loc - Source.data(), SourceMgr::DK_Error, Msg.str(), + Source, None, None); + return true; +} + +static const char *toString(MIToken::TokenKind TokenKind) { + switch (TokenKind) { + case MIToken::comma: + return "','"; + case MIToken::equal: + return "'='"; + case MIToken::colon: + return "':'"; + case MIToken::lparen: + return "'('"; + case MIToken::rparen: + return "')'"; + default: + return "<unknown token>"; + } +} + +bool MIParser::expectAndConsume(MIToken::TokenKind TokenKind) { + if (Token.isNot(TokenKind)) + return error(Twine("expected ") + toString(TokenKind)); + lex(); + return false; +} + +bool MIParser::consumeIfPresent(MIToken::TokenKind TokenKind) { + if (Token.isNot(TokenKind)) + return false; + lex(); + return true; +} + +bool MIParser::parseBasicBlockDefinition( + DenseMap<unsigned, MachineBasicBlock *> &MBBSlots) { + assert(Token.is(MIToken::MachineBasicBlockLabel)); + unsigned ID = 0; + if (getUnsigned(ID)) + return true; + auto Loc = Token.location(); + auto Name = Token.stringValue(); + lex(); + bool HasAddressTaken = false; + bool IsLandingPad = false; + unsigned Alignment = 0; + BasicBlock *BB = nullptr; + if (consumeIfPresent(MIToken::lparen)) { + do { + // TODO: Report an error when multiple same attributes are specified. + switch (Token.kind()) { + case MIToken::kw_address_taken: + HasAddressTaken = true; + lex(); + break; + case MIToken::kw_landing_pad: + IsLandingPad = true; + lex(); + break; + case MIToken::kw_align: + if (parseAlignment(Alignment)) + return true; + break; + case MIToken::IRBlock: + // TODO: Report an error when both name and ir block are specified. + if (parseIRBlock(BB, *MF.getFunction())) + return true; + lex(); + break; + default: + break; + } + } while (consumeIfPresent(MIToken::comma)); + if (expectAndConsume(MIToken::rparen)) + return true; + } + if (expectAndConsume(MIToken::colon)) + return true; + + if (!Name.empty()) { + BB = dyn_cast_or_null<BasicBlock>( + MF.getFunction()->getValueSymbolTable().lookup(Name)); + if (!BB) + return error(Loc, Twine("basic block '") + Name + + "' is not defined in the function '" + + MF.getName() + "'"); + } + auto *MBB = MF.CreateMachineBasicBlock(BB); + MF.insert(MF.end(), MBB); + bool WasInserted = MBBSlots.insert(std::make_pair(ID, MBB)).second; + if (!WasInserted) + return error(Loc, Twine("redefinition of machine basic block with id #") + + Twine(ID)); + if (Alignment) + MBB->setAlignment(Alignment); + if (HasAddressTaken) + MBB->setHasAddressTaken(); + MBB->setIsEHPad(IsLandingPad); + return false; +} + +bool MIParser::parseBasicBlockDefinitions( + DenseMap<unsigned, MachineBasicBlock *> &MBBSlots) { + lex(); + // Skip until the first machine basic block. + while (Token.is(MIToken::Newline)) + lex(); + if (Token.isErrorOrEOF()) + return Token.isError(); + if (Token.isNot(MIToken::MachineBasicBlockLabel)) + return error("expected a basic block definition before instructions"); + unsigned BraceDepth = 0; + do { + if (parseBasicBlockDefinition(MBBSlots)) + return true; + bool IsAfterNewline = false; + // Skip until the next machine basic block. + while (true) { + if ((Token.is(MIToken::MachineBasicBlockLabel) && IsAfterNewline) || + Token.isErrorOrEOF()) + break; + else if (Token.is(MIToken::MachineBasicBlockLabel)) + return error("basic block definition should be located at the start of " + "the line"); + else if (consumeIfPresent(MIToken::Newline)) { + IsAfterNewline = true; + continue; + } + IsAfterNewline = false; + if (Token.is(MIToken::lbrace)) + ++BraceDepth; + if (Token.is(MIToken::rbrace)) { + if (!BraceDepth) + return error("extraneous closing brace ('}')"); + --BraceDepth; + } + lex(); + } + // Verify that we closed all of the '{' at the end of a file or a block. + if (!Token.isError() && BraceDepth) + return error("expected '}'"); // FIXME: Report a note that shows '{'. + } while (!Token.isErrorOrEOF()); + return Token.isError(); +} + +bool MIParser::parseBasicBlockLiveins(MachineBasicBlock &MBB) { + assert(Token.is(MIToken::kw_liveins)); + lex(); + if (expectAndConsume(MIToken::colon)) + return true; + if (Token.isNewlineOrEOF()) // Allow an empty list of liveins. + return false; + do { + if (Token.isNot(MIToken::NamedRegister)) + return error("expected a named register"); + unsigned Reg = 0; + if (parseRegister(Reg)) + return true; + MBB.addLiveIn(Reg); + lex(); + } while (consumeIfPresent(MIToken::comma)); + return false; +} + +bool MIParser::parseBasicBlockSuccessors(MachineBasicBlock &MBB) { + assert(Token.is(MIToken::kw_successors)); + lex(); + if (expectAndConsume(MIToken::colon)) + return true; + if (Token.isNewlineOrEOF()) // Allow an empty list of successors. + return false; + do { + if (Token.isNot(MIToken::MachineBasicBlock)) + return error("expected a machine basic block reference"); + MachineBasicBlock *SuccMBB = nullptr; + if (parseMBBReference(SuccMBB)) + return true; + lex(); + unsigned Weight = 0; + if (consumeIfPresent(MIToken::lparen)) { + if (Token.isNot(MIToken::IntegerLiteral)) + return error("expected an integer literal after '('"); + if (getUnsigned(Weight)) + return true; + lex(); + if (expectAndConsume(MIToken::rparen)) + return true; + } + MBB.addSuccessor(SuccMBB, BranchProbability::getRaw(Weight)); + } while (consumeIfPresent(MIToken::comma)); + MBB.normalizeSuccProbs(); + return false; +} + +bool MIParser::parseBasicBlock(MachineBasicBlock &MBB) { + // Skip the definition. + assert(Token.is(MIToken::MachineBasicBlockLabel)); + lex(); + if (consumeIfPresent(MIToken::lparen)) { + while (Token.isNot(MIToken::rparen) && !Token.isErrorOrEOF()) + lex(); + consumeIfPresent(MIToken::rparen); + } + consumeIfPresent(MIToken::colon); + + // Parse the liveins and successors. + // N.B: Multiple lists of successors and liveins are allowed and they're + // merged into one. + // Example: + // liveins: %edi + // liveins: %esi + // + // is equivalent to + // liveins: %edi, %esi + while (true) { + if (Token.is(MIToken::kw_successors)) { + if (parseBasicBlockSuccessors(MBB)) + return true; + } else if (Token.is(MIToken::kw_liveins)) { + if (parseBasicBlockLiveins(MBB)) + return true; + } else if (consumeIfPresent(MIToken::Newline)) { + continue; + } else + break; + if (!Token.isNewlineOrEOF()) + return error("expected line break at the end of a list"); + lex(); + } + + // Parse the instructions. + bool IsInBundle = false; + MachineInstr *PrevMI = nullptr; + while (true) { + if (Token.is(MIToken::MachineBasicBlockLabel) || Token.is(MIToken::Eof)) + return false; + else if (consumeIfPresent(MIToken::Newline)) + continue; + if (consumeIfPresent(MIToken::rbrace)) { + // The first parsing pass should verify that all closing '}' have an + // opening '{'. + assert(IsInBundle); + IsInBundle = false; + continue; + } + MachineInstr *MI = nullptr; + if (parse(MI)) + return true; + MBB.insert(MBB.end(), MI); + if (IsInBundle) { + PrevMI->setFlag(MachineInstr::BundledSucc); + MI->setFlag(MachineInstr::BundledPred); + } + PrevMI = MI; + if (Token.is(MIToken::lbrace)) { + if (IsInBundle) + return error("nested instruction bundles are not allowed"); + lex(); + // This instruction is the start of the bundle. + MI->setFlag(MachineInstr::BundledSucc); + IsInBundle = true; + if (!Token.is(MIToken::Newline)) + // The next instruction can be on the same line. + continue; + } + assert(Token.isNewlineOrEOF() && "MI is not fully parsed"); + lex(); + } + return false; +} + +bool MIParser::parseBasicBlocks() { + lex(); + // Skip until the first machine basic block. + while (Token.is(MIToken::Newline)) + lex(); + if (Token.isErrorOrEOF()) + return Token.isError(); + // The first parsing pass should have verified that this token is a MBB label + // in the 'parseBasicBlockDefinitions' method. + assert(Token.is(MIToken::MachineBasicBlockLabel)); + do { + MachineBasicBlock *MBB = nullptr; + if (parseMBBReference(MBB)) + return true; + if (parseBasicBlock(*MBB)) + return true; + // The method 'parseBasicBlock' should parse the whole block until the next + // block or the end of file. + assert(Token.is(MIToken::MachineBasicBlockLabel) || Token.is(MIToken::Eof)); + } while (Token.isNot(MIToken::Eof)); + return false; +} + +bool MIParser::parse(MachineInstr *&MI) { + // Parse any register operands before '=' + MachineOperand MO = MachineOperand::CreateImm(0); + SmallVector<ParsedMachineOperand, 8> Operands; + while (Token.isRegister() || Token.isRegisterFlag()) { + auto Loc = Token.location(); + Optional<unsigned> TiedDefIdx; + if (parseRegisterOperand(MO, TiedDefIdx, /*IsDef=*/true)) + return true; + Operands.push_back( + ParsedMachineOperand(MO, Loc, Token.location(), TiedDefIdx)); + if (Token.isNot(MIToken::comma)) + break; + lex(); + } + if (!Operands.empty() && expectAndConsume(MIToken::equal)) + return true; + + unsigned OpCode, Flags = 0; + if (Token.isError() || parseInstruction(OpCode, Flags)) + return true; + + // Parse the remaining machine operands. + while (!Token.isNewlineOrEOF() && Token.isNot(MIToken::kw_debug_location) && + Token.isNot(MIToken::coloncolon) && Token.isNot(MIToken::lbrace)) { + auto Loc = Token.location(); + Optional<unsigned> TiedDefIdx; + if (parseMachineOperandAndTargetFlags(MO, TiedDefIdx)) + return true; + Operands.push_back( + ParsedMachineOperand(MO, Loc, Token.location(), TiedDefIdx)); + if (Token.isNewlineOrEOF() || Token.is(MIToken::coloncolon) || + Token.is(MIToken::lbrace)) + break; + if (Token.isNot(MIToken::comma)) + return error("expected ',' before the next machine operand"); + lex(); + } + + DebugLoc DebugLocation; + if (Token.is(MIToken::kw_debug_location)) { + lex(); + if (Token.isNot(MIToken::exclaim)) + return error("expected a metadata node after 'debug-location'"); + MDNode *Node = nullptr; + if (parseMDNode(Node)) + return true; + DebugLocation = DebugLoc(Node); + } + + // Parse the machine memory operands. + SmallVector<MachineMemOperand *, 2> MemOperands; + if (Token.is(MIToken::coloncolon)) { + lex(); + while (!Token.isNewlineOrEOF()) { + MachineMemOperand *MemOp = nullptr; + if (parseMachineMemoryOperand(MemOp)) + return true; + MemOperands.push_back(MemOp); + if (Token.isNewlineOrEOF()) + break; + if (Token.isNot(MIToken::comma)) + return error("expected ',' before the next machine memory 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, DebugLocation, /*NoImplicit=*/true); + MI->setFlags(Flags); + for (const auto &Operand : Operands) + MI->addOperand(MF, Operand.Operand); + if (assignRegisterTies(*MI, Operands)) + return true; + if (MemOperands.empty()) + return false; + MachineInstr::mmo_iterator MemRefs = + MF.allocateMemRefsArray(MemOperands.size()); + std::copy(MemOperands.begin(), MemOperands.end(), MemRefs); + MI->setMemRefs(MemRefs, MemRefs + MemOperands.size()); + return false; +} + +bool MIParser::parseStandaloneMBB(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::parseStandaloneNamedRegister(unsigned &Reg) { + lex(); + if (Token.isNot(MIToken::NamedRegister)) + return error("expected a named register"); + if (parseRegister(Reg)) + return true; + lex(); + if (Token.isNot(MIToken::Eof)) + return error("expected end of string after the register reference"); + return false; +} + +bool MIParser::parseStandaloneVirtualRegister(unsigned &Reg) { + lex(); + if (Token.isNot(MIToken::VirtualRegister)) + return error("expected a virtual register"); + if (parseRegister(Reg)) + return true; + lex(); + if (Token.isNot(MIToken::Eof)) + return error("expected end of string after the register reference"); + return false; +} + +bool MIParser::parseStandaloneStackObject(int &FI) { + lex(); + if (Token.isNot(MIToken::StackObject)) + return error("expected a stack object"); + if (parseStackFrameIndex(FI)) + return true; + if (Token.isNot(MIToken::Eof)) + return error("expected end of string after the stack object reference"); + return false; +} + +bool MIParser::parseStandaloneMDNode(MDNode *&Node) { + lex(); + if (Token.isNot(MIToken::exclaim)) + return error("expected a metadata node"); + if (parseMDNode(Node)) + return true; + if (Token.isNot(MIToken::Eof)) + return error("expected end of string after the metadata node"); + 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(); +} + +/// Return true if the parsed machine operands contain a given machine operand. +static bool isImplicitOperandIn(const MachineOperand &ImplicitOperand, + ArrayRef<ParsedMachineOperand> Operands) { + for (const auto &I : Operands) { + if (ImplicitOperand.isIdenticalTo(I.Operand)) + return true; + } + return false; +} + +bool MIParser::verifyImplicitOperands(ArrayRef<ParsedMachineOperand> 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 MCPhysReg *ImpDefs = MCID.getImplicitDefs(); *ImpDefs; ++ImpDefs) + ImplicitOperands.push_back( + MachineOperand::CreateReg(*ImpDefs, true, true)); + if (MCID.ImplicitUses) + for (const MCPhysReg *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"); + for (const auto &I : ImplicitOperands) { + if (isImplicitOperandIn(I, Operands)) + continue; + return error(Operands.empty() ? Token.location() : Operands.back().End, + Twine("missing implicit register operand '") + + printImplicitRegisterFlag(I) + " %" + + getRegisterName(TRI, I.getReg()) + "'"); + } + return false; +} + +bool MIParser::parseInstruction(unsigned &OpCode, unsigned &Flags) { + if (Token.is(MIToken::kw_frame_setup)) { + Flags |= MachineInstr::FrameSetup; + lex(); + } + 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) { + const unsigned OldFlags = Flags; + switch (Token.kind()) { + case MIToken::kw_implicit: + Flags |= RegState::Implicit; + break; + case MIToken::kw_implicit_define: + Flags |= RegState::ImplicitDefine; + break; + case MIToken::kw_def: + Flags |= RegState::Define; + 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; + case MIToken::kw_internal: + Flags |= RegState::InternalRead; + break; + case MIToken::kw_early_clobber: + Flags |= RegState::EarlyClobber; + break; + case MIToken::kw_debug_use: + Flags |= RegState::Debug; + break; + default: + llvm_unreachable("The current token should be a register flag"); + } + if (OldFlags == Flags) + // We know that the same flag is specified more than once when the flags + // weren't modified. + return error("duplicate '" + Token.stringValue() + "' 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::parseRegisterTiedDefIndex(unsigned &TiedDefIdx) { + if (!consumeIfPresent(MIToken::kw_tied_def)) + return error("expected 'tied-def' after '('"); + if (Token.isNot(MIToken::IntegerLiteral)) + return error("expected an integer literal after 'tied-def'"); + if (getUnsigned(TiedDefIdx)) + return true; + lex(); + if (expectAndConsume(MIToken::rparen)) + return true; + return false; +} + +bool MIParser::assignRegisterTies(MachineInstr &MI, + ArrayRef<ParsedMachineOperand> Operands) { + SmallVector<std::pair<unsigned, unsigned>, 4> TiedRegisterPairs; + for (unsigned I = 0, E = Operands.size(); I != E; ++I) { + if (!Operands[I].TiedDefIdx) + continue; + // The parser ensures that this operand is a register use, so we just have + // to check the tied-def operand. + unsigned DefIdx = Operands[I].TiedDefIdx.getValue(); + if (DefIdx >= E) + return error(Operands[I].Begin, + Twine("use of invalid tied-def operand index '" + + Twine(DefIdx) + "'; instruction has only ") + + Twine(E) + " operands"); + const auto &DefOperand = Operands[DefIdx].Operand; + if (!DefOperand.isReg() || !DefOperand.isDef()) + // FIXME: add note with the def operand. + return error(Operands[I].Begin, + Twine("use of invalid tied-def operand index '") + + Twine(DefIdx) + "'; the operand #" + Twine(DefIdx) + + " isn't a defined register"); + // Check that the tied-def operand wasn't tied elsewhere. + for (const auto &TiedPair : TiedRegisterPairs) { + if (TiedPair.first == DefIdx) + return error(Operands[I].Begin, + Twine("the tied-def operand #") + Twine(DefIdx) + + " is already tied with another register operand"); + } + TiedRegisterPairs.push_back(std::make_pair(DefIdx, I)); + } + // FIXME: Verify that for non INLINEASM instructions, the def and use tied + // indices must be less than tied max. + for (const auto &TiedPair : TiedRegisterPairs) + MI.tieOperands(TiedPair.first, TiedPair.second); + return false; +} + +bool MIParser::parseRegisterOperand(MachineOperand &Dest, + Optional<unsigned> &TiedDefIdx, + 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; + } + if ((Flags & RegState::Define) == 0 && consumeIfPresent(MIToken::lparen)) { + unsigned Idx; + if (parseRegisterTiedDefIndex(Idx)) + return true; + TiedDefIdx = Idx; + } + Dest = MachineOperand::CreateReg( + Reg, Flags & RegState::Define, Flags & RegState::Implicit, + Flags & RegState::Kill, Flags & RegState::Dead, Flags & RegState::Undef, + Flags & RegState::EarlyClobber, SubReg, Flags & RegState::Debug, + Flags & RegState::InternalRead); + return false; +} + +bool MIParser::parseImmediateOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::IntegerLiteral)); + const APSInt &Int = Token.integerValue(); + if (Int.getMinSignedBits() > 64) + return error("integer literal is too large to be an immediate operand"); + Dest = MachineOperand::CreateImm(Int.getExtValue()); + lex(); + return false; +} + +bool MIParser::parseIRConstant(StringRef::iterator Loc, StringRef StringValue, + const Constant *&C) { + auto Source = StringValue.str(); // The source has to be null terminated. + SMDiagnostic Err; + C = parseConstantValue(Source.c_str(), Err, *MF.getFunction()->getParent(), + &IRSlots); + if (!C) + return error(Loc + Err.getColumnNo(), Err.getMessage()); + return false; +} + +bool MIParser::parseIRConstant(StringRef::iterator Loc, const Constant *&C) { + if (parseIRConstant(Loc, StringRef(Loc, Token.range().end() - Loc), C)) + return true; + lex(); + return false; +} + +bool MIParser::parseTypedImmediateOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::IntegerType)); + auto Loc = Token.location(); + lex(); + if (Token.isNot(MIToken::IntegerLiteral)) + return error("expected an integer literal"); + const Constant *C = nullptr; + if (parseIRConstant(Loc, C)) + return true; + Dest = MachineOperand::CreateCImm(cast<ConstantInt>(C)); + return false; +} + +bool MIParser::parseFPImmediateOperand(MachineOperand &Dest) { + auto Loc = Token.location(); + lex(); + if (Token.isNot(MIToken::FloatingPointLiteral)) + return error("expected a floating point literal"); + const Constant *C = nullptr; + if (parseIRConstant(Loc, C)) + return true; + Dest = MachineOperand::CreateFPImm(cast<ConstantFP>(C)); + 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) || + Token.is(MIToken::MachineBasicBlockLabel)); + 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::parseStackFrameIndex(int &FI) { + assert(Token.is(MIToken::StackObject)); + unsigned ID; + if (getUnsigned(ID)) + return true; + auto ObjectInfo = PFS.StackObjectSlots.find(ID); + if (ObjectInfo == PFS.StackObjectSlots.end()) + return error(Twine("use of undefined stack object '%stack.") + Twine(ID) + + "'"); + StringRef Name; + if (const auto *Alloca = + MF.getFrameInfo()->getObjectAllocation(ObjectInfo->second)) + Name = Alloca->getName(); + if (!Token.stringValue().empty() && Token.stringValue() != Name) + return error(Twine("the name of the stack object '%stack.") + Twine(ID) + + "' isn't '" + Token.stringValue() + "'"); + lex(); + FI = ObjectInfo->second; + return false; +} + +bool MIParser::parseStackObjectOperand(MachineOperand &Dest) { + int FI; + if (parseStackFrameIndex(FI)) + return true; + Dest = MachineOperand::CreateFI(FI); + return false; +} + +bool MIParser::parseFixedStackFrameIndex(int &FI) { + assert(Token.is(MIToken::FixedStackObject)); + unsigned ID; + if (getUnsigned(ID)) + return true; + auto ObjectInfo = PFS.FixedStackObjectSlots.find(ID); + if (ObjectInfo == PFS.FixedStackObjectSlots.end()) + return error(Twine("use of undefined fixed stack object '%fixed-stack.") + + Twine(ID) + "'"); + lex(); + FI = ObjectInfo->second; + return false; +} + +bool MIParser::parseFixedStackObjectOperand(MachineOperand &Dest) { + int FI; + if (parseFixedStackFrameIndex(FI)) + return true; + Dest = MachineOperand::CreateFI(FI); + return false; +} + +bool MIParser::parseGlobalValue(GlobalValue *&GV) { + switch (Token.kind()) { + case MIToken::NamedGlobalValue: { + const Module *M = MF.getFunction()->getParent(); + GV = M->getNamedValue(Token.stringValue()); + if (!GV) + return error(Twine("use of undefined global value '") + Token.range() + + "'"); + break; + } + 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) + + "'"); + GV = IRSlots.GlobalValues[GVIdx]; + break; + } + default: + llvm_unreachable("The current token should be a global value"); + } + return false; +} + +bool MIParser::parseGlobalAddressOperand(MachineOperand &Dest) { + GlobalValue *GV = nullptr; + if (parseGlobalValue(GV)) + return true; + lex(); + Dest = MachineOperand::CreateGA(GV, /*Offset=*/0); + if (parseOperandsOffset(Dest)) + return true; + return false; +} + +bool MIParser::parseConstantPoolIndexOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::ConstantPoolItem)); + unsigned ID; + if (getUnsigned(ID)) + return true; + auto ConstantInfo = PFS.ConstantPoolSlots.find(ID); + if (ConstantInfo == PFS.ConstantPoolSlots.end()) + return error("use of undefined constant '%const." + Twine(ID) + "'"); + lex(); + Dest = MachineOperand::CreateCPI(ID, /*Offset=*/0); + if (parseOperandsOffset(Dest)) + return true; + return false; +} + +bool MIParser::parseJumpTableIndexOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::JumpTableIndex)); + unsigned ID; + if (getUnsigned(ID)) + return true; + auto JumpTableEntryInfo = PFS.JumpTableSlots.find(ID); + if (JumpTableEntryInfo == PFS.JumpTableSlots.end()) + return error("use of undefined jump table '%jump-table." + Twine(ID) + "'"); + lex(); + Dest = MachineOperand::CreateJTI(JumpTableEntryInfo->second); + return false; +} + +bool MIParser::parseExternalSymbolOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::ExternalSymbol)); + const char *Symbol = MF.createExternalSymbolName(Token.stringValue()); + lex(); + Dest = MachineOperand::CreateES(Symbol); + if (parseOperandsOffset(Dest)) + return true; + return false; +} + +bool MIParser::parseMDNode(MDNode *&Node) { + assert(Token.is(MIToken::exclaim)); + auto Loc = Token.location(); + lex(); + if (Token.isNot(MIToken::IntegerLiteral) || Token.integerValue().isSigned()) + return error("expected metadata id after '!'"); + unsigned ID; + if (getUnsigned(ID)) + return true; + auto NodeInfo = IRSlots.MetadataNodes.find(ID); + if (NodeInfo == IRSlots.MetadataNodes.end()) + return error(Loc, "use of undefined metadata '!" + Twine(ID) + "'"); + lex(); + Node = NodeInfo->second.get(); + return false; +} + +bool MIParser::parseMetadataOperand(MachineOperand &Dest) { + MDNode *Node = nullptr; + if (parseMDNode(Node)) + return true; + Dest = MachineOperand::CreateMetadata(Node); + return false; +} + +bool MIParser::parseCFIOffset(int &Offset) { + if (Token.isNot(MIToken::IntegerLiteral)) + return error("expected a cfi offset"); + if (Token.integerValue().getMinSignedBits() > 32) + return error("expected a 32 bit integer (the cfi offset is too large)"); + Offset = (int)Token.integerValue().getExtValue(); + lex(); + return false; +} + +bool MIParser::parseCFIRegister(unsigned &Reg) { + if (Token.isNot(MIToken::NamedRegister)) + return error("expected a cfi register"); + unsigned LLVMReg; + if (parseRegister(LLVMReg)) + return true; + const auto *TRI = MF.getSubtarget().getRegisterInfo(); + assert(TRI && "Expected target register info"); + int DwarfReg = TRI->getDwarfRegNum(LLVMReg, true); + if (DwarfReg < 0) + return error("invalid DWARF register"); + Reg = (unsigned)DwarfReg; + lex(); + return false; +} + +bool MIParser::parseCFIOperand(MachineOperand &Dest) { + auto Kind = Token.kind(); + lex(); + auto &MMI = MF.getMMI(); + int Offset; + unsigned Reg; + unsigned CFIIndex; + switch (Kind) { + case MIToken::kw_cfi_same_value: + if (parseCFIRegister(Reg)) + return true; + CFIIndex = + MMI.addFrameInst(MCCFIInstruction::createSameValue(nullptr, Reg)); + break; + case MIToken::kw_cfi_offset: + if (parseCFIRegister(Reg) || expectAndConsume(MIToken::comma) || + parseCFIOffset(Offset)) + return true; + CFIIndex = + MMI.addFrameInst(MCCFIInstruction::createOffset(nullptr, Reg, Offset)); + break; + case MIToken::kw_cfi_def_cfa_register: + if (parseCFIRegister(Reg)) + return true; + CFIIndex = + MMI.addFrameInst(MCCFIInstruction::createDefCfaRegister(nullptr, Reg)); + break; + case MIToken::kw_cfi_def_cfa_offset: + if (parseCFIOffset(Offset)) + return true; + // NB: MCCFIInstruction::createDefCfaOffset negates the offset. + CFIIndex = MMI.addFrameInst( + MCCFIInstruction::createDefCfaOffset(nullptr, -Offset)); + break; + case MIToken::kw_cfi_def_cfa: + if (parseCFIRegister(Reg) || expectAndConsume(MIToken::comma) || + parseCFIOffset(Offset)) + return true; + // NB: MCCFIInstruction::createDefCfa negates the offset. + CFIIndex = + MMI.addFrameInst(MCCFIInstruction::createDefCfa(nullptr, Reg, -Offset)); + break; + default: + // TODO: Parse the other CFI operands. + llvm_unreachable("The current token should be a cfi operand"); + } + Dest = MachineOperand::CreateCFIIndex(CFIIndex); + return false; +} + +bool MIParser::parseIRBlock(BasicBlock *&BB, const Function &F) { + switch (Token.kind()) { + case MIToken::NamedIRBlock: { + BB = dyn_cast_or_null<BasicBlock>( + F.getValueSymbolTable().lookup(Token.stringValue())); + if (!BB) + return error(Twine("use of undefined IR block '") + Token.range() + "'"); + break; + } + case MIToken::IRBlock: { + unsigned SlotNumber = 0; + if (getUnsigned(SlotNumber)) + return true; + BB = const_cast<BasicBlock *>(getIRBlock(SlotNumber, F)); + if (!BB) + return error(Twine("use of undefined IR block '%ir-block.") + + Twine(SlotNumber) + "'"); + break; + } + default: + llvm_unreachable("The current token should be an IR block reference"); + } + return false; +} + +bool MIParser::parseBlockAddressOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::kw_blockaddress)); + lex(); + if (expectAndConsume(MIToken::lparen)) + return true; + if (Token.isNot(MIToken::GlobalValue) && + Token.isNot(MIToken::NamedGlobalValue)) + return error("expected a global value"); + GlobalValue *GV = nullptr; + if (parseGlobalValue(GV)) + return true; + auto *F = dyn_cast<Function>(GV); + if (!F) + return error("expected an IR function reference"); + lex(); + if (expectAndConsume(MIToken::comma)) + return true; + BasicBlock *BB = nullptr; + if (Token.isNot(MIToken::IRBlock) && Token.isNot(MIToken::NamedIRBlock)) + return error("expected an IR block reference"); + if (parseIRBlock(BB, *F)) + return true; + lex(); + if (expectAndConsume(MIToken::rparen)) + return true; + Dest = MachineOperand::CreateBA(BlockAddress::get(F, BB), /*Offset=*/0); + if (parseOperandsOffset(Dest)) + return true; + return false; +} + +bool MIParser::parseTargetIndexOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::kw_target_index)); + lex(); + if (expectAndConsume(MIToken::lparen)) + return true; + if (Token.isNot(MIToken::Identifier)) + return error("expected the name of the target index"); + int Index = 0; + if (getTargetIndex(Token.stringValue(), Index)) + return error("use of undefined target index '" + Token.stringValue() + "'"); + lex(); + if (expectAndConsume(MIToken::rparen)) + return true; + Dest = MachineOperand::CreateTargetIndex(unsigned(Index), /*Offset=*/0); + if (parseOperandsOffset(Dest)) + return true; + return false; +} + +bool MIParser::parseLiveoutRegisterMaskOperand(MachineOperand &Dest) { + assert(Token.is(MIToken::kw_liveout)); + const auto *TRI = MF.getSubtarget().getRegisterInfo(); + assert(TRI && "Expected target register info"); + uint32_t *Mask = MF.allocateRegisterMask(TRI->getNumRegs()); + lex(); + if (expectAndConsume(MIToken::lparen)) + return true; + while (true) { + if (Token.isNot(MIToken::NamedRegister)) + return error("expected a named register"); + unsigned Reg = 0; + if (parseRegister(Reg)) + return true; + lex(); + Mask[Reg / 32] |= 1U << (Reg % 32); + // TODO: Report an error if the same register is used more than once. + if (Token.isNot(MIToken::comma)) + break; + lex(); + } + if (expectAndConsume(MIToken::rparen)) + return true; + Dest = MachineOperand::CreateRegLiveOut(Mask); + return false; +} + +bool MIParser::parseMachineOperand(MachineOperand &Dest, + Optional<unsigned> &TiedDefIdx) { + switch (Token.kind()) { + case MIToken::kw_implicit: + case MIToken::kw_implicit_define: + case MIToken::kw_def: + case MIToken::kw_dead: + case MIToken::kw_killed: + case MIToken::kw_undef: + case MIToken::kw_internal: + case MIToken::kw_early_clobber: + case MIToken::kw_debug_use: + case MIToken::underscore: + case MIToken::NamedRegister: + case MIToken::VirtualRegister: + return parseRegisterOperand(Dest, TiedDefIdx); + case MIToken::IntegerLiteral: + return parseImmediateOperand(Dest); + case MIToken::IntegerType: + return parseTypedImmediateOperand(Dest); + case MIToken::kw_half: + case MIToken::kw_float: + case MIToken::kw_double: + case MIToken::kw_x86_fp80: + case MIToken::kw_fp128: + case MIToken::kw_ppc_fp128: + return parseFPImmediateOperand(Dest); + case MIToken::MachineBasicBlock: + return parseMBBOperand(Dest); + case MIToken::StackObject: + return parseStackObjectOperand(Dest); + case MIToken::FixedStackObject: + return parseFixedStackObjectOperand(Dest); + case MIToken::GlobalValue: + case MIToken::NamedGlobalValue: + return parseGlobalAddressOperand(Dest); + case MIToken::ConstantPoolItem: + return parseConstantPoolIndexOperand(Dest); + case MIToken::JumpTableIndex: + return parseJumpTableIndexOperand(Dest); + case MIToken::ExternalSymbol: + return parseExternalSymbolOperand(Dest); + case MIToken::exclaim: + return parseMetadataOperand(Dest); + case MIToken::kw_cfi_same_value: + case MIToken::kw_cfi_offset: + case MIToken::kw_cfi_def_cfa_register: + case MIToken::kw_cfi_def_cfa_offset: + case MIToken::kw_cfi_def_cfa: + return parseCFIOperand(Dest); + case MIToken::kw_blockaddress: + return parseBlockAddressOperand(Dest); + case MIToken::kw_target_index: + return parseTargetIndexOperand(Dest); + case MIToken::kw_liveout: + return parseLiveoutRegisterMaskOperand(Dest); + case MIToken::Error: + return true; + case MIToken::Identifier: + if (const auto *RegMask = getRegMask(Token.stringValue())) { + Dest = MachineOperand::CreateRegMask(RegMask); + lex(); + break; + } + // fallthrough + default: + // FIXME: Parse the MCSymbol machine operand. + return error("expected a machine operand"); + } + return false; +} + +bool MIParser::parseMachineOperandAndTargetFlags( + MachineOperand &Dest, Optional<unsigned> &TiedDefIdx) { + unsigned TF = 0; + bool HasTargetFlags = false; + if (Token.is(MIToken::kw_target_flags)) { + HasTargetFlags = true; + lex(); + if (expectAndConsume(MIToken::lparen)) + return true; + if (Token.isNot(MIToken::Identifier)) + return error("expected the name of the target flag"); + if (getDirectTargetFlag(Token.stringValue(), TF)) { + if (getBitmaskTargetFlag(Token.stringValue(), TF)) + return error("use of undefined target flag '" + Token.stringValue() + + "'"); + } + lex(); + while (Token.is(MIToken::comma)) { + lex(); + if (Token.isNot(MIToken::Identifier)) + return error("expected the name of the target flag"); + unsigned BitFlag = 0; + if (getBitmaskTargetFlag(Token.stringValue(), BitFlag)) + return error("use of undefined target flag '" + Token.stringValue() + + "'"); + // TODO: Report an error when using a duplicate bit target flag. + TF |= BitFlag; + lex(); + } + if (expectAndConsume(MIToken::rparen)) + return true; + } + auto Loc = Token.location(); + if (parseMachineOperand(Dest, TiedDefIdx)) + return true; + if (!HasTargetFlags) + return false; + if (Dest.isReg()) + return error(Loc, "register operands can't have target flags"); + Dest.setTargetFlags(TF); + return false; +} + +bool MIParser::parseOffset(int64_t &Offset) { + if (Token.isNot(MIToken::plus) && Token.isNot(MIToken::minus)) + return false; + StringRef Sign = Token.range(); + bool IsNegative = Token.is(MIToken::minus); + lex(); + if (Token.isNot(MIToken::IntegerLiteral)) + return error("expected an integer literal after '" + Sign + "'"); + if (Token.integerValue().getMinSignedBits() > 64) + return error("expected 64-bit integer (too large)"); + Offset = Token.integerValue().getExtValue(); + if (IsNegative) + Offset = -Offset; + lex(); + return false; +} + +bool MIParser::parseAlignment(unsigned &Alignment) { + assert(Token.is(MIToken::kw_align)); + lex(); + if (Token.isNot(MIToken::IntegerLiteral) || Token.integerValue().isSigned()) + return error("expected an integer literal after 'align'"); + if (getUnsigned(Alignment)) + return true; + lex(); + return false; +} + +bool MIParser::parseOperandsOffset(MachineOperand &Op) { + int64_t Offset = 0; + if (parseOffset(Offset)) + return true; + Op.setOffset(Offset); + return false; +} + +bool MIParser::parseIRValue(const Value *&V) { + switch (Token.kind()) { + case MIToken::NamedIRValue: { + V = MF.getFunction()->getValueSymbolTable().lookup(Token.stringValue()); + break; + } + case MIToken::IRValue: { + unsigned SlotNumber = 0; + if (getUnsigned(SlotNumber)) + return true; + V = getIRValue(SlotNumber); + break; + } + case MIToken::NamedGlobalValue: + case MIToken::GlobalValue: { + GlobalValue *GV = nullptr; + if (parseGlobalValue(GV)) + return true; + V = GV; + break; + } + case MIToken::QuotedIRValue: { + const Constant *C = nullptr; + if (parseIRConstant(Token.location(), Token.stringValue(), C)) + return true; + V = C; + break; + } + default: + llvm_unreachable("The current token should be an IR block reference"); + } + if (!V) + return error(Twine("use of undefined IR value '") + Token.range() + "'"); + return false; +} + +bool MIParser::getUint64(uint64_t &Result) { + assert(Token.hasIntegerValue()); + if (Token.integerValue().getActiveBits() > 64) + return error("expected 64-bit integer (too large)"); + Result = Token.integerValue().getZExtValue(); + return false; +} + +bool MIParser::parseMemoryOperandFlag(unsigned &Flags) { + const unsigned OldFlags = Flags; + switch (Token.kind()) { + case MIToken::kw_volatile: + Flags |= MachineMemOperand::MOVolatile; + break; + case MIToken::kw_non_temporal: + Flags |= MachineMemOperand::MONonTemporal; + break; + case MIToken::kw_invariant: + Flags |= MachineMemOperand::MOInvariant; + break; + // TODO: parse the target specific memory operand flags. + default: + llvm_unreachable("The current token should be a memory operand flag"); + } + if (OldFlags == Flags) + // We know that the same flag is specified more than once when the flags + // weren't modified. + return error("duplicate '" + Token.stringValue() + "' memory operand flag"); + lex(); + return false; +} + +bool MIParser::parseMemoryPseudoSourceValue(const PseudoSourceValue *&PSV) { + switch (Token.kind()) { + case MIToken::kw_stack: + PSV = MF.getPSVManager().getStack(); + break; + case MIToken::kw_got: + PSV = MF.getPSVManager().getGOT(); + break; + case MIToken::kw_jump_table: + PSV = MF.getPSVManager().getJumpTable(); + break; + case MIToken::kw_constant_pool: + PSV = MF.getPSVManager().getConstantPool(); + break; + case MIToken::FixedStackObject: { + int FI; + if (parseFixedStackFrameIndex(FI)) + return true; + PSV = MF.getPSVManager().getFixedStack(FI); + // The token was already consumed, so use return here instead of break. + return false; + } + case MIToken::kw_call_entry: { + lex(); + switch (Token.kind()) { + case MIToken::GlobalValue: + case MIToken::NamedGlobalValue: { + GlobalValue *GV = nullptr; + if (parseGlobalValue(GV)) + return true; + PSV = MF.getPSVManager().getGlobalValueCallEntry(GV); + break; + } + case MIToken::ExternalSymbol: + PSV = MF.getPSVManager().getExternalSymbolCallEntry( + MF.createExternalSymbolName(Token.stringValue())); + break; + default: + return error( + "expected a global value or an external symbol after 'call-entry'"); + } + break; + } + default: + llvm_unreachable("The current token should be pseudo source value"); + } + lex(); + return false; +} + +bool MIParser::parseMachinePointerInfo(MachinePointerInfo &Dest) { + if (Token.is(MIToken::kw_constant_pool) || Token.is(MIToken::kw_stack) || + Token.is(MIToken::kw_got) || Token.is(MIToken::kw_jump_table) || + Token.is(MIToken::FixedStackObject) || Token.is(MIToken::kw_call_entry)) { + const PseudoSourceValue *PSV = nullptr; + if (parseMemoryPseudoSourceValue(PSV)) + return true; + int64_t Offset = 0; + if (parseOffset(Offset)) + return true; + Dest = MachinePointerInfo(PSV, Offset); + return false; + } + if (Token.isNot(MIToken::NamedIRValue) && Token.isNot(MIToken::IRValue) && + Token.isNot(MIToken::GlobalValue) && + Token.isNot(MIToken::NamedGlobalValue) && + Token.isNot(MIToken::QuotedIRValue)) + return error("expected an IR value reference"); + const Value *V = nullptr; + if (parseIRValue(V)) + return true; + if (!V->getType()->isPointerTy()) + return error("expected a pointer IR value"); + lex(); + int64_t Offset = 0; + if (parseOffset(Offset)) + return true; + Dest = MachinePointerInfo(V, Offset); + return false; +} + +bool MIParser::parseMachineMemoryOperand(MachineMemOperand *&Dest) { + if (expectAndConsume(MIToken::lparen)) + return true; + unsigned Flags = 0; + while (Token.isMemoryOperandFlag()) { + if (parseMemoryOperandFlag(Flags)) + return true; + } + if (Token.isNot(MIToken::Identifier) || + (Token.stringValue() != "load" && Token.stringValue() != "store")) + return error("expected 'load' or 'store' memory operation"); + if (Token.stringValue() == "load") + Flags |= MachineMemOperand::MOLoad; + else + Flags |= MachineMemOperand::MOStore; + lex(); + + if (Token.isNot(MIToken::IntegerLiteral)) + return error("expected the size integer literal after memory operation"); + uint64_t Size; + if (getUint64(Size)) + return true; + lex(); + + const char *Word = Flags & MachineMemOperand::MOLoad ? "from" : "into"; + if (Token.isNot(MIToken::Identifier) || Token.stringValue() != Word) + return error(Twine("expected '") + Word + "'"); + lex(); + + MachinePointerInfo Ptr = MachinePointerInfo(); + if (parseMachinePointerInfo(Ptr)) + return true; + unsigned BaseAlignment = Size; + AAMDNodes AAInfo; + MDNode *Range = nullptr; + while (consumeIfPresent(MIToken::comma)) { + switch (Token.kind()) { + case MIToken::kw_align: + if (parseAlignment(BaseAlignment)) + return true; + break; + case MIToken::md_tbaa: + lex(); + if (parseMDNode(AAInfo.TBAA)) + return true; + break; + case MIToken::md_alias_scope: + lex(); + if (parseMDNode(AAInfo.Scope)) + return true; + break; + case MIToken::md_noalias: + lex(); + if (parseMDNode(AAInfo.NoAlias)) + return true; + break; + case MIToken::md_range: + lex(); + if (parseMDNode(Range)) + return true; + break; + // TODO: Report an error on duplicate metadata nodes. + default: + return error("expected 'align' or '!tbaa' or '!alias.scope' or " + "'!noalias' or '!range'"); + } + } + if (expectAndConsume(MIToken::rparen)) + return true; + Dest = + MF.getMachineMemOperand(Ptr, Flags, Size, BaseAlignment, AAInfo, Range); + 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(); +} + +static void initSlots2BasicBlocks( + const Function &F, + DenseMap<unsigned, const BasicBlock *> &Slots2BasicBlocks) { + ModuleSlotTracker MST(F.getParent(), /*ShouldInitializeAllMetadata=*/false); + MST.incorporateFunction(F); + for (auto &BB : F) { + if (BB.hasName()) + continue; + int Slot = MST.getLocalSlot(&BB); + if (Slot == -1) + continue; + Slots2BasicBlocks.insert(std::make_pair(unsigned(Slot), &BB)); + } +} + +static const BasicBlock *getIRBlockFromSlot( + unsigned Slot, + const DenseMap<unsigned, const BasicBlock *> &Slots2BasicBlocks) { + auto BlockInfo = Slots2BasicBlocks.find(Slot); + if (BlockInfo == Slots2BasicBlocks.end()) + return nullptr; + return BlockInfo->second; +} + +const BasicBlock *MIParser::getIRBlock(unsigned Slot) { + if (Slots2BasicBlocks.empty()) + initSlots2BasicBlocks(*MF.getFunction(), Slots2BasicBlocks); + return getIRBlockFromSlot(Slot, Slots2BasicBlocks); +} + +const BasicBlock *MIParser::getIRBlock(unsigned Slot, const Function &F) { + if (&F == MF.getFunction()) + return getIRBlock(Slot); + DenseMap<unsigned, const BasicBlock *> CustomSlots2BasicBlocks; + initSlots2BasicBlocks(F, CustomSlots2BasicBlocks); + return getIRBlockFromSlot(Slot, CustomSlots2BasicBlocks); +} + +static void mapValueToSlot(const Value *V, ModuleSlotTracker &MST, + DenseMap<unsigned, const Value *> &Slots2Values) { + int Slot = MST.getLocalSlot(V); + if (Slot == -1) + return; + Slots2Values.insert(std::make_pair(unsigned(Slot), V)); +} + +/// Creates the mapping from slot numbers to function's unnamed IR values. +static void initSlots2Values(const Function &F, + DenseMap<unsigned, const Value *> &Slots2Values) { + ModuleSlotTracker MST(F.getParent(), /*ShouldInitializeAllMetadata=*/false); + MST.incorporateFunction(F); + for (const auto &Arg : F.args()) + mapValueToSlot(&Arg, MST, Slots2Values); + for (const auto &BB : F) { + mapValueToSlot(&BB, MST, Slots2Values); + for (const auto &I : BB) + mapValueToSlot(&I, MST, Slots2Values); + } +} + +const Value *MIParser::getIRValue(unsigned Slot) { + if (Slots2Values.empty()) + initSlots2Values(*MF.getFunction(), Slots2Values); + auto ValueInfo = Slots2Values.find(Slot); + if (ValueInfo == Slots2Values.end()) + return nullptr; + return ValueInfo->second; +} + +void MIParser::initNames2TargetIndices() { + if (!Names2TargetIndices.empty()) + return; + const auto *TII = MF.getSubtarget().getInstrInfo(); + assert(TII && "Expected target instruction info"); + auto Indices = TII->getSerializableTargetIndices(); + for (const auto &I : Indices) + Names2TargetIndices.insert(std::make_pair(StringRef(I.second), I.first)); +} + +bool MIParser::getTargetIndex(StringRef Name, int &Index) { + initNames2TargetIndices(); + auto IndexInfo = Names2TargetIndices.find(Name); + if (IndexInfo == Names2TargetIndices.end()) + return true; + Index = IndexInfo->second; + return false; +} + +void MIParser::initNames2DirectTargetFlags() { + if (!Names2DirectTargetFlags.empty()) + return; + const auto *TII = MF.getSubtarget().getInstrInfo(); + assert(TII && "Expected target instruction info"); + auto Flags = TII->getSerializableDirectMachineOperandTargetFlags(); + for (const auto &I : Flags) + Names2DirectTargetFlags.insert( + std::make_pair(StringRef(I.second), I.first)); +} + +bool MIParser::getDirectTargetFlag(StringRef Name, unsigned &Flag) { + initNames2DirectTargetFlags(); + auto FlagInfo = Names2DirectTargetFlags.find(Name); + if (FlagInfo == Names2DirectTargetFlags.end()) + return true; + Flag = FlagInfo->second; + return false; +} + +void MIParser::initNames2BitmaskTargetFlags() { + if (!Names2BitmaskTargetFlags.empty()) + return; + const auto *TII = MF.getSubtarget().getInstrInfo(); + assert(TII && "Expected target instruction info"); + auto Flags = TII->getSerializableBitmaskMachineOperandTargetFlags(); + for (const auto &I : Flags) + Names2BitmaskTargetFlags.insert( + std::make_pair(StringRef(I.second), I.first)); +} + +bool MIParser::getBitmaskTargetFlag(StringRef Name, unsigned &Flag) { + initNames2BitmaskTargetFlags(); + auto FlagInfo = Names2BitmaskTargetFlags.find(Name); + if (FlagInfo == Names2BitmaskTargetFlags.end()) + return true; + Flag = FlagInfo->second; + return false; +} + +bool llvm::parseMachineBasicBlockDefinitions(MachineFunction &MF, StringRef Src, + PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error) { + SourceMgr SM; + SM.AddNewSourceBuffer( + MemoryBuffer::getMemBuffer(Src, "", /*RequiresNullTerminator=*/false), + SMLoc()); + return MIParser(SM, MF, Error, Src, PFS, IRSlots) + .parseBasicBlockDefinitions(PFS.MBBSlots); +} + +bool llvm::parseMachineInstructions(MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error) { + SourceMgr SM; + SM.AddNewSourceBuffer( + MemoryBuffer::getMemBuffer(Src, "", /*RequiresNullTerminator=*/false), + SMLoc()); + return MIParser(SM, MF, Error, Src, PFS, IRSlots).parseBasicBlocks(); +} + +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).parseStandaloneMBB(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) + .parseStandaloneNamedRegister(Reg); +} + +bool llvm::parseVirtualRegisterReference(unsigned &Reg, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error) { + return MIParser(SM, MF, Error, Src, PFS, IRSlots) + .parseStandaloneVirtualRegister(Reg); +} + +bool llvm::parseStackObjectReference(int &FI, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error) { + return MIParser(SM, MF, Error, Src, PFS, IRSlots) + .parseStandaloneStackObject(FI); +} + +bool llvm::parseMDNode(MDNode *&Node, SourceMgr &SM, MachineFunction &MF, + StringRef Src, const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, SMDiagnostic &Error) { + return MIParser(SM, MF, Error, Src, PFS, IRSlots).parseStandaloneMDNode(Node); +} diff --git a/contrib/llvm/lib/CodeGen/MIRParser/MIParser.h b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.h new file mode 100644 index 0000000..8aef704 --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MIParser.h @@ -0,0 +1,99 @@ +//===- 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 BasicBlock; +class MachineBasicBlock; +class MachineInstr; +class MachineFunction; +class MDNode; +struct SlotMapping; +class SMDiagnostic; +class SourceMgr; + +struct PerFunctionMIParsingState { + DenseMap<unsigned, MachineBasicBlock *> MBBSlots; + DenseMap<unsigned, unsigned> VirtualRegisterSlots; + DenseMap<unsigned, int> FixedStackObjectSlots; + DenseMap<unsigned, int> StackObjectSlots; + DenseMap<unsigned, unsigned> ConstantPoolSlots; + DenseMap<unsigned, unsigned> JumpTableSlots; +}; + +/// Parse the machine basic block definitions, and skip the machine +/// instructions. +/// +/// This function runs the first parsing pass on the machine function's body. +/// It parses only the machine basic block definitions and creates the machine +/// basic blocks in the given machine function. +/// +/// The machine instructions aren't parsed during the first pass because all +/// the machine basic blocks aren't defined yet - this makes it impossible to +/// resolve the machine basic block references. +/// +/// Return true if an error occurred. +bool parseMachineBasicBlockDefinitions(MachineFunction &MF, StringRef Src, + PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error); + +/// Parse the machine instructions. +/// +/// This function runs the second parsing pass on the machine function's body. +/// It skips the machine basic block definitions and parses only the machine +/// instructions and basic block attributes like liveins and successors. +/// +/// The second parsing pass assumes that the first parsing pass already ran +/// on the given source string. +/// +/// Return true if an error occurred. +bool parseMachineInstructions(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); + +bool parseVirtualRegisterReference(unsigned &Reg, SourceMgr &SM, + MachineFunction &MF, StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, + SMDiagnostic &Error); + +bool parseStackObjectReference(int &FI, SourceMgr &SM, MachineFunction &MF, + StringRef Src, + const PerFunctionMIParsingState &PFS, + const SlotMapping &IRSlots, SMDiagnostic &Error); + +bool parseMDNode(MDNode *&Node, 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..422efbc --- /dev/null +++ b/contrib/llvm/lib/CodeGen/MIRParser/MIRParser.cpp @@ -0,0 +1,739 @@ +//===- 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/MachineConstantPool.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineModuleInfo.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); + + bool initializeRegisterInfo(MachineFunction &MF, + const yaml::MachineFunction &YamlMF, + PerFunctionMIParsingState &PFS); + + void inferRegisterInfo(MachineFunction &MF, + const yaml::MachineFunction &YamlMF); + + bool initializeFrameInfo(MachineFunction &MF, + const yaml::MachineFunction &YamlMF, + PerFunctionMIParsingState &PFS); + + bool parseCalleeSavedRegister(MachineFunction &MF, + PerFunctionMIParsingState &PFS, + std::vector<CalleeSavedInfo> &CSIInfo, + const yaml::StringValue &RegisterSource, + int FrameIdx); + + bool parseStackObjectsDebugInfo(MachineFunction &MF, + PerFunctionMIParsingState &PFS, + const yaml::MachineStackObject &Object, + int FrameIdx); + + bool initializeConstantPool(MachineConstantPool &ConstantPool, + const yaml::MachineFunction &YamlMF, + const MachineFunction &MF, + DenseMap<unsigned, unsigned> &ConstantPoolSlots); + + bool initializeJumpTableInfo(MachineFunction &MF, + const yaml::MachineJumpTable &YamlJTI, + PerFunctionMIParsingState &PFS); + +private: + bool parseMDNode(MDNode *&Node, const yaml::StringValue &Source, + MachineFunction &MF, const PerFunctionMIParsingState &PFS); + + bool parseMBBReference(MachineBasicBlock *&MBB, + const yaml::StringValue &Source, MachineFunction &MF, + const PerFunctionMIParsingState &PFS); + + /// Return a MIR diagnostic converted from an MI string diagnostic. + SMDiagnostic diagFromMIStringDiag(const SMDiagnostic &Error, + SMRange SourceRange); + + /// Return a MIR diagnostic converted from a diagnostic located in a YAML + /// block scalar string. + SMDiagnostic diagFromBlockStringDiag(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(diagFromBlockStringDiag(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, YamlMF, PFS)) + return true; + if (!YamlMF.Constants.empty()) { + auto *ConstantPool = MF.getConstantPool(); + assert(ConstantPool && "Constant pool must be created"); + if (initializeConstantPool(*ConstantPool, YamlMF, MF, + PFS.ConstantPoolSlots)) + return true; + } + + SMDiagnostic Error; + if (parseMachineBasicBlockDefinitions(MF, YamlMF.Body.Value.Value, PFS, + IRSlots, Error)) { + reportDiagnostic( + diagFromBlockStringDiag(Error, YamlMF.Body.Value.SourceRange)); + return true; + } + + if (MF.empty()) + return error(Twine("machine function '") + Twine(MF.getName()) + + "' requires at least one machine basic block in its body"); + // Initialize the frame information after creating all the MBBs so that the + // MBB references in the frame information can be resolved. + if (initializeFrameInfo(MF, YamlMF, PFS)) + return true; + // Initialize the jump table after creating all the MBBs so that the MBB + // references can be resolved. + if (!YamlMF.JumpTableInfo.Entries.empty() && + initializeJumpTableInfo(MF, YamlMF.JumpTableInfo, PFS)) + return true; + // Parse the machine instructions after creating all of the MBBs so that the + // parser can resolve the MBB references. + if (parseMachineInstructions(MF, YamlMF.Body.Value.Value, PFS, IRSlots, + Error)) { + reportDiagnostic( + diagFromBlockStringDiag(Error, YamlMF.Body.Value.SourceRange)); + return true; + } + inferRegisterInfo(MF, YamlMF); + // FIXME: This is a temporary workaround until the reserved registers can be + // serialized. + MF.getRegInfo().freezeReservedRegs(MF); + MF.verify(); + return false; +} + +bool MIRParserImpl::initializeRegisterInfo(MachineFunction &MF, + const yaml::MachineFunction &YamlMF, + PerFunctionMIParsingState &PFS) { + MachineRegisterInfo &RegInfo = MF.getRegInfo(); + assert(RegInfo.isSSA()); + if (!YamlMF.IsSSA) + RegInfo.leaveSSA(); + assert(RegInfo.tracksLiveness()); + if (!YamlMF.TracksRegLiveness) + RegInfo.invalidateLiveness(); + RegInfo.enableSubRegLiveness(YamlMF.TracksSubRegLiveness); + + SMDiagnostic Error; + // 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); + if (!PFS.VirtualRegisterSlots.insert(std::make_pair(VReg.ID.Value, Reg)) + .second) + return error(VReg.ID.SourceRange.Start, + Twine("redefinition of virtual register '%") + + Twine(VReg.ID.Value) + "'"); + if (!VReg.PreferredRegister.Value.empty()) { + unsigned PreferredReg = 0; + if (parseNamedRegisterReference(PreferredReg, SM, MF, + VReg.PreferredRegister.Value, PFS, + IRSlots, Error)) + return error(Error, VReg.PreferredRegister.SourceRange); + RegInfo.setSimpleHint(Reg, PreferredReg); + } + } + + // Parse the liveins. + for (const auto &LiveIn : YamlMF.LiveIns) { + unsigned Reg = 0; + if (parseNamedRegisterReference(Reg, SM, MF, LiveIn.Register.Value, PFS, + IRSlots, Error)) + return error(Error, LiveIn.Register.SourceRange); + unsigned VReg = 0; + if (!LiveIn.VirtualRegister.Value.empty()) { + if (parseVirtualRegisterReference( + VReg, SM, MF, LiveIn.VirtualRegister.Value, PFS, IRSlots, Error)) + return error(Error, LiveIn.VirtualRegister.SourceRange); + } + RegInfo.addLiveIn(Reg, VReg); + } + + // Parse the callee saved register mask. + BitVector CalleeSavedRegisterMask(RegInfo.getUsedPhysRegsMask().size()); + if (!YamlMF.CalleeSavedRegisters) + return false; + for (const auto &RegSource : YamlMF.CalleeSavedRegisters.getValue()) { + unsigned Reg = 0; + if (parseNamedRegisterReference(Reg, SM, MF, RegSource.Value, PFS, IRSlots, + Error)) + return error(Error, RegSource.SourceRange); + CalleeSavedRegisterMask[Reg] = true; + } + RegInfo.setUsedPhysRegMask(CalleeSavedRegisterMask.flip()); + return false; +} + +void MIRParserImpl::inferRegisterInfo(MachineFunction &MF, + const yaml::MachineFunction &YamlMF) { + if (YamlMF.CalleeSavedRegisters) + return; + for (const MachineBasicBlock &MBB : MF) { + for (const MachineInstr &MI : MBB) { + for (const MachineOperand &MO : MI.operands()) { + if (!MO.isRegMask()) + continue; + MF.getRegInfo().addPhysRegsUsedFromRegMask(MO.getRegMask()); + } + } + } +} + +bool MIRParserImpl::initializeFrameInfo(MachineFunction &MF, + const yaml::MachineFunction &YamlMF, + PerFunctionMIParsingState &PFS) { + MachineFrameInfo &MFI = *MF.getFrameInfo(); + const Function &F = *MF.getFunction(); + 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); + if (!YamlMFI.SavePoint.Value.empty()) { + MachineBasicBlock *MBB = nullptr; + if (parseMBBReference(MBB, YamlMFI.SavePoint, MF, PFS)) + return true; + MFI.setSavePoint(MBB); + } + if (!YamlMFI.RestorePoint.Value.empty()) { + MachineBasicBlock *MBB = nullptr; + if (parseMBBReference(MBB, YamlMFI.RestorePoint, MF, PFS)) + return true; + MFI.setRestorePoint(MBB); + } + + std::vector<CalleeSavedInfo> CSIInfo; + // 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); + if (!PFS.FixedStackObjectSlots.insert(std::make_pair(Object.ID.Value, + ObjectIdx)) + .second) + return error(Object.ID.SourceRange.Start, + Twine("redefinition of fixed stack object '%fixed-stack.") + + Twine(Object.ID.Value) + "'"); + if (parseCalleeSavedRegister(MF, PFS, CSIInfo, Object.CalleeSavedRegister, + ObjectIdx)) + return true; + } + + // Initialize the ordinary frame objects. + for (const auto &Object : YamlMF.StackObjects) { + int ObjectIdx; + const AllocaInst *Alloca = nullptr; + const yaml::StringValue &Name = Object.Name; + if (!Name.Value.empty()) { + Alloca = dyn_cast_or_null<AllocaInst>( + F.getValueSymbolTable().lookup(Name.Value)); + if (!Alloca) + return error(Name.SourceRange.Start, + "alloca instruction named '" + Name.Value + + "' isn't defined in the function '" + F.getName() + + "'"); + } + if (Object.Type == yaml::MachineStackObject::VariableSized) + ObjectIdx = MFI.CreateVariableSizedObject(Object.Alignment, Alloca); + else + ObjectIdx = MFI.CreateStackObject( + Object.Size, Object.Alignment, + Object.Type == yaml::MachineStackObject::SpillSlot, Alloca); + MFI.setObjectOffset(ObjectIdx, Object.Offset); + if (!PFS.StackObjectSlots.insert(std::make_pair(Object.ID.Value, ObjectIdx)) + .second) + return error(Object.ID.SourceRange.Start, + Twine("redefinition of stack object '%stack.") + + Twine(Object.ID.Value) + "'"); + if (parseCalleeSavedRegister(MF, PFS, CSIInfo, Object.CalleeSavedRegister, + ObjectIdx)) + return true; + if (Object.LocalOffset) + MFI.mapLocalFrameObject(ObjectIdx, Object.LocalOffset.getValue()); + if (parseStackObjectsDebugInfo(MF, PFS, Object, ObjectIdx)) + return true; + } + MFI.setCalleeSavedInfo(CSIInfo); + if (!CSIInfo.empty()) + MFI.setCalleeSavedInfoValid(true); + + // Initialize the various stack object references after initializing the + // stack objects. + if (!YamlMFI.StackProtector.Value.empty()) { + SMDiagnostic Error; + int FI; + if (parseStackObjectReference(FI, SM, MF, YamlMFI.StackProtector.Value, PFS, + IRSlots, Error)) + return error(Error, YamlMFI.StackProtector.SourceRange); + MFI.setStackProtectorIndex(FI); + } + return false; +} + +bool MIRParserImpl::parseCalleeSavedRegister( + MachineFunction &MF, PerFunctionMIParsingState &PFS, + std::vector<CalleeSavedInfo> &CSIInfo, + const yaml::StringValue &RegisterSource, int FrameIdx) { + if (RegisterSource.Value.empty()) + return false; + unsigned Reg = 0; + SMDiagnostic Error; + if (parseNamedRegisterReference(Reg, SM, MF, RegisterSource.Value, PFS, + IRSlots, Error)) + return error(Error, RegisterSource.SourceRange); + CSIInfo.push_back(CalleeSavedInfo(Reg, FrameIdx)); + return false; +} + +/// Verify that given node is of a certain type. Return true on error. +template <typename T> +static bool typecheckMDNode(T *&Result, MDNode *Node, + const yaml::StringValue &Source, + StringRef TypeString, MIRParserImpl &Parser) { + if (!Node) + return false; + Result = dyn_cast<T>(Node); + if (!Result) + return Parser.error(Source.SourceRange.Start, + "expected a reference to a '" + TypeString + + "' metadata node"); + return false; +} + +bool MIRParserImpl::parseStackObjectsDebugInfo( + MachineFunction &MF, PerFunctionMIParsingState &PFS, + const yaml::MachineStackObject &Object, int FrameIdx) { + // Debug information can only be attached to stack objects; Fixed stack + // objects aren't supported. + assert(FrameIdx >= 0 && "Expected a stack object frame index"); + MDNode *Var = nullptr, *Expr = nullptr, *Loc = nullptr; + if (parseMDNode(Var, Object.DebugVar, MF, PFS) || + parseMDNode(Expr, Object.DebugExpr, MF, PFS) || + parseMDNode(Loc, Object.DebugLoc, MF, PFS)) + return true; + if (!Var && !Expr && !Loc) + return false; + DILocalVariable *DIVar = nullptr; + DIExpression *DIExpr = nullptr; + DILocation *DILoc = nullptr; + if (typecheckMDNode(DIVar, Var, Object.DebugVar, "DILocalVariable", *this) || + typecheckMDNode(DIExpr, Expr, Object.DebugExpr, "DIExpression", *this) || + typecheckMDNode(DILoc, Loc, Object.DebugLoc, "DILocation", *this)) + return true; + MF.getMMI().setVariableDbgInfo(DIVar, DIExpr, unsigned(FrameIdx), DILoc); + return false; +} + +bool MIRParserImpl::parseMDNode(MDNode *&Node, const yaml::StringValue &Source, + MachineFunction &MF, + const PerFunctionMIParsingState &PFS) { + if (Source.Value.empty()) + return false; + SMDiagnostic Error; + if (llvm::parseMDNode(Node, SM, MF, Source.Value, PFS, IRSlots, Error)) + return error(Error, Source.SourceRange); + return false; +} + +bool MIRParserImpl::initializeConstantPool( + MachineConstantPool &ConstantPool, const yaml::MachineFunction &YamlMF, + const MachineFunction &MF, + DenseMap<unsigned, unsigned> &ConstantPoolSlots) { + const auto &M = *MF.getFunction()->getParent(); + SMDiagnostic Error; + for (const auto &YamlConstant : YamlMF.Constants) { + const Constant *Value = dyn_cast_or_null<Constant>( + parseConstantValue(YamlConstant.Value.Value, Error, M)); + if (!Value) + return error(Error, YamlConstant.Value.SourceRange); + unsigned Alignment = + YamlConstant.Alignment + ? YamlConstant.Alignment + : M.getDataLayout().getPrefTypeAlignment(Value->getType()); + unsigned Index = ConstantPool.getConstantPoolIndex(Value, Alignment); + if (!ConstantPoolSlots.insert(std::make_pair(YamlConstant.ID.Value, Index)) + .second) + return error(YamlConstant.ID.SourceRange.Start, + Twine("redefinition of constant pool item '%const.") + + Twine(YamlConstant.ID.Value) + "'"); + } + return false; +} + +bool MIRParserImpl::initializeJumpTableInfo( + MachineFunction &MF, const yaml::MachineJumpTable &YamlJTI, + PerFunctionMIParsingState &PFS) { + MachineJumpTableInfo *JTI = MF.getOrCreateJumpTableInfo(YamlJTI.Kind); + for (const auto &Entry : YamlJTI.Entries) { + std::vector<MachineBasicBlock *> Blocks; + for (const auto &MBBSource : Entry.Blocks) { + MachineBasicBlock *MBB = nullptr; + if (parseMBBReference(MBB, MBBSource.Value, MF, PFS)) + return true; + Blocks.push_back(MBB); + } + unsigned Index = JTI->createJumpTableIndex(Blocks); + if (!PFS.JumpTableSlots.insert(std::make_pair(Entry.ID.Value, Index)) + .second) + return error(Entry.ID.SourceRange.Start, + Twine("redefinition of jump table entry '%jump-table.") + + Twine(Entry.ID.Value) + "'"); + } + return false; +} + +bool MIRParserImpl::parseMBBReference(MachineBasicBlock *&MBB, + const yaml::StringValue &Source, + MachineFunction &MF, + const PerFunctionMIParsingState &PFS) { + SMDiagnostic Error; + if (llvm::parseMBBReference(MBB, SM, MF, Source.Value, PFS, IRSlots, Error)) + return error(Error, Source.SourceRange); + 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::diagFromBlockStringDiag(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)); +} |
