diff options
Diffstat (limited to 'backend/lib')
25 files changed, 4633 insertions, 4921 deletions
diff --git a/backend/lib/MIR/MILexer.cpp b/backend/lib/CodeGen/MIRParser/MILexer.cpp index f568cef..830c548 100644 --- a/backend/lib/MIR/MILexer.cpp +++ b/backend/lib/CodeGen/MIRParser/MILexer.cpp @@ -1,6 +1,6 @@ // SPDX-License-Identifier: GPL-3.0-or-later -#include "rcc/MIR/MILexer.h" +#include "MILexer.h" #include <algorithm> #include <array> diff --git a/backend/lib/CodeGen/MIRParser/MILexer.h b/backend/lib/CodeGen/MIRParser/MILexer.h new file mode 100644 index 0000000..e3d50e7 --- /dev/null +++ b/backend/lib/CodeGen/MIRParser/MILexer.h @@ -0,0 +1,181 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#ifndef RCC_LIB_CODEGEN_MIRPARSER_MILEXER_H +#define RCC_LIB_CODEGEN_MIRPARSER_MILEXER_H + +#include "rcc/Support/TextDiagnostic.h" + +#include <cstddef> +#include <cstdint> +#include <memory> +#include <span> +#include <string> +#include <string_view> +#include <utility> +#include <variant> + +namespace rcc::mir { + +enum class MITokenKind : std::uint8_t { + EndOfFile, + Error, + Newline, + + Comma, + Equal, + Underscore, + Colon, + DoubleColon, + Dot, + Exclaim, + LeftParen, + RightParen, + LeftBrace, + RightBrace, + Plus, + Minus, + Less, + Greater, + + Keyword, + MetadataKeyword, + Identifier, + NamedRegister, + NamedVirtualRegister, + MachineBasicBlockLabel, + MachineBasicBlock, + StackObject, + FixedStackObject, + NamedGlobalValue, + GlobalValue, + ExternalSymbol, + MCSymbol, + IntegerLiteral, + FloatingPointLiteral, + HexLiteral, + VirtualRegister, + ConstantPoolItem, + JumpTableIndex, + NamedIRBlock, + IRBlock, + NamedIRValue, + IRValue, + QuotedIRValue, + SubRegisterIndex, + StringConstant, +}; + +[[nodiscard]] std::string_view getMITokenKindName(MITokenKind Kind) noexcept; +[[nodiscard]] bool isMIKeyword(std::string_view Spelling) noexcept; +[[nodiscard]] bool isMIMetadataKeyword(std::string_view Spelling) noexcept; + +class MIToken final { +public: + [[nodiscard]] MITokenKind kind() const noexcept { + return Kind; + } + + [[nodiscard]] const TextRange &range() const noexcept { + return Range; + } + + [[nodiscard]] std::string_view spelling() const noexcept { + return Spelling; + } + + [[nodiscard]] bool hasStringValue() const noexcept { + return !std::holds_alternative<std::monostate>(StringValue); + } + + [[nodiscard]] std::string_view stringValue() const noexcept; + + [[nodiscard]] bool hasIntegerValue() const noexcept { + return !IntegerValue.empty(); + } + + [[nodiscard]] std::string_view integerValue() const noexcept { + return IntegerValue; + } + + [[nodiscard]] bool is(MITokenKind Expected) const noexcept { + return Kind == Expected; + } + + [[nodiscard]] bool isKeyword(std::string_view Expected) const noexcept { + return Kind == MITokenKind::Keyword && stringValue() == Expected; + } + + [[nodiscard]] bool isMetadataKeyword(std::string_view Expected) const noexcept { + return Kind == MITokenKind::MetadataKeyword && stringValue() == Expected; + } + + [[nodiscard]] bool isNewlineOrEnd() const noexcept { + return Kind == MITokenKind::Newline || Kind == MITokenKind::EndOfFile; + } + + [[nodiscard]] bool isErrorOrEnd() const noexcept { + return Kind == MITokenKind::Error || Kind == MITokenKind::EndOfFile; + } + + [[nodiscard]] bool isRegister() const noexcept; + [[nodiscard]] bool isRegisterFlag() const noexcept; + [[nodiscard]] bool isMemoryOperandFlag() const noexcept; + +private: + using StringValueStorage = std::variant<std::monostate, std::string_view, std::string>; + + MIToken(MITokenKind Kind, TextRange Range, std::string_view Spelling, + StringValueStorage StringValue = {}, std::string_view IntegerValue = {}) + : Kind(Kind), Range(Range), Spelling(Spelling), StringValue(std::move(StringValue)), + IntegerValue(IntegerValue) {} + + MITokenKind Kind = MITokenKind::Error; + TextRange Range; + std::string_view Spelling; + StringValueStorage StringValue; + std::string_view IntegerValue; + + friend class MILexer; +}; + +struct MILexerOptions final { + std::size_t MaximumTokenLength = std::size_t{1} << 24U; + std::size_t MaximumTokens = std::size_t{1} << 24U; +}; + +class MILexer final { +public: + MILexer(std::string SourceName, std::string SourceText, MILexerOptions Options = {}); + MILexer(const MILexer &) = delete; + MILexer &operator=(const MILexer &) = delete; + MILexer(MILexer &&) noexcept; + MILexer &operator=(MILexer &&) noexcept; + ~MILexer(); + + [[nodiscard]] MIToken next(); + + [[nodiscard]] const TextSource &source() const noexcept; + [[nodiscard]] std::span<const TextDiagnostic> diagnostics() const noexcept; + [[nodiscard]] bool failed() const noexcept; + [[nodiscard]] std::size_t tokenCount() const noexcept; + +private: + [[nodiscard]] static MIToken makeToken(MITokenKind Kind, TextRange Range, + std::string_view Spelling, + std::string_view IntegerValue = {}); + [[nodiscard]] static MIToken makeStringToken(MITokenKind Kind, TextRange Range, + std::string_view Spelling, + std::string_view StringValue, + std::string_view IntegerValue = {}); + [[nodiscard]] static MIToken makeOwnedStringToken(MITokenKind Kind, TextRange Range, + std::string_view Spelling, + std::string StringValue, + std::string_view IntegerValue = {}); + + class Impl; + std::unique_ptr<Impl> Implementation; +}; + +} // namespace rcc::mir + +#endif // RCC_LIB_CODEGEN_MIRPARSER_MILEXER_H diff --git a/backend/lib/MIR/MIParser.cpp b/backend/lib/CodeGen/MIRParser/MIParser.cpp index 3748187..e38b9f1 100644 --- a/backend/lib/MIR/MIParser.cpp +++ b/backend/lib/CodeGen/MIRParser/MIParser.cpp @@ -1,10 +1,12 @@ // SPDX-License-Identifier: GPL-3.0-or-later -#include "rcc/MIR/MIParser.h" +#include "rcc/CodeGen/MIRParser/MIParser.h" +#include "MILexer.h" #include "rcc/CodeGen/GenericOpcodes.h" #include "rcc/CodeGen/InlineAsm.h" #include "rcc/CodeGen/InstructionDescriptor.h" +#include "rcc/CodeGen/MIRYamlMapping.h" #include "rcc/CodeGen/MachineBasicBlock.h" #include "rcc/CodeGen/MachineCFGUpdater.h" #include "rcc/CodeGen/MachineFunction.h" @@ -16,8 +18,6 @@ #include "rcc/CodeGen/TargetMIRTextInfo.h" #include "rcc/CodeGen/TargetRegisterInfo.h" #include "rcc/MIR/LLVMConstantSyntax.h" -#include "rcc/MIR/MIDebugSyntax.h" -#include "rcc/MIR/MILexer.h" #include <algorithm> #include <array> diff --git a/backend/lib/MIR/MIRDocumentParser.cpp b/backend/lib/CodeGen/MIRParser/MIRParser.cpp index 08627d5..e22f1bc 100644 --- a/backend/lib/MIR/MIRDocumentParser.cpp +++ b/backend/lib/CodeGen/MIRParser/MIRParser.cpp @@ -1,10 +1,8 @@ // SPDX-License-Identifier: GPL-3.0-or-later -#include "MIRDocumentParser.h" +#include "rcc/CodeGen/MIRParser/MIRParser.h" -#include "MIRDocumentReferenceValidation.h" -#include "MIRMetadataTextParser.h" -#include "MIRParserUtilities.h" +#include "rcc/CodeGen/MIRYamlMapping.h" #include "rcc/CodeGen/MachineBasicBlock.h" #include "rcc/CodeGen/MachineConstantPool.h" #include "rcc/CodeGen/MachineFrameInfo.h" @@ -40,13 +38,6 @@ namespace rcc::mir { -using parser_detail::parseBool; -using parser_detail::parsePrefixedID; -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; -using parser_detail::trim; - bool MIRDocumentParser::fail(const TextRange &Range, std::string Message, std::string Category) { if (Result.Diagnostics.empty()) { Result.Diagnostics.push_back(TextDiagnostic{ @@ -117,82 +108,6 @@ bool MIRDocumentParser::scalar(const yaml::Node &Node, std::string_view &Value, return false; } -bool MIRDocumentParser::boolean(const yaml::Node &Node, bool &Value) { - std::string_view Text; - if (scalar(Node, Text, "boolean") || !parseBool(Text, Value)) - return fail(Node.range(), "expected 'true' or 'false'"); - return false; -} - -bool MIRDocumentParser::unsignedValue(const yaml::Node &Node, std::uint64_t &Value, - std::string_view Description) { - std::string_view Text; - if (scalar(Node, Text, Description) || !parseUnsigned(Text, Value)) - return fail(Node.range(), "expected " + std::string(Description)); - return false; -} - -bool MIRDocumentParser::signedValue(const yaml::Node &Node, std::int64_t &Value, - std::string_view Description) { - std::string_view Text; - if (scalar(Node, Text, Description) || !parseSigned(Text, Value)) - return fail(Node.range(), "expected " + std::string(Description)); - return false; -} - -bool MIRDocumentParser::alignment(const yaml::Node &Node, Alignment &Value, bool AllowZero) { - std::uint64_t Bytes = 0; - if (unsignedValue(Node, Bytes, "an unsigned alignment")) - return true; - if (Bytes == 0 && AllowZero) { - Value = Alignment{}; - return false; - } - auto Parsed = Alignment::create(Bytes); - if (!Parsed) - return importError(Node.range(), Parsed.error()); - Value = *Parsed; - return false; -} - -bool MIRDocumentParser::physicalRegister(std::string_view Text, const TextRange &Range, - PhysicalRegister &Reg) { - Text = trim(Text); - if (!Text.starts_with('$') || Text.size() == 1) - return fail(Range, "physical register must use the '$name' MIR spelling"); - const PhysicalRegisterDesc *Description = Target.registers().physicalRegister(Text.substr(1)); - if (Description == nullptr) - return fail(Range, "unknown physical register '" + std::string(Text) + "'"); - Reg = Description->Reg; - return false; -} - -bool MIRDocumentParser::physicalRegister(const yaml::Node &Node, PhysicalRegister &Reg) { - std::string_view Text; - return scalar(Node, Text, "physical register") || physicalRegister(Text, Node.range(), Reg); -} - -bool MIRDocumentParser::virtualRegister(std::string_view Text, const TextRange &Range, - MachineFunction &Function, MIParserContext &Context, - Register &Reg) { - const auto ID = parsePrefixedID(Text, "%"); - if (!ID) - return fail(Range, "virtual register must use the '%N' MIR spelling"); - auto Found = - std::ranges::find_if(Context.VirtualRegisters, - [ID](const MIVirtualRegisterSlot &Slot) { return Slot.TextID == *ID; }); - if (Found == Context.VirtualRegisters.end()) { - auto Created = Function.registerInfo().createVirtualRegister(); - if (!Created) - return importError(Range, Created.error()); - Context.VirtualRegisters.push_back({*ID, *Created}); - Reg = *Created; - return false; - } - Reg = Found->Reg; - return false; -} - MIRDocumentParseResult MIRDocumentParser::run() { yaml::ParseResult YAMLResult = yaml::parse(std::string(Result.Source.name()), std::string(Result.Source.text()), Options.YAML); @@ -396,3 +311,24 @@ bool MIRDocumentParser::parseFunction(const yaml::Node &Root, MIRDocument &Docum } } // namespace rcc::mir + +namespace rcc::mir { + +MIRDocument::MIRDocument() = default; +MIRDocument::MIRDocument(CompilationUnit Unit) : Unit(std::move(Unit)) {} +MIRDocument::MIRDocument(MIRDocument &&) noexcept = default; +MIRDocument &MIRDocument::operator=(MIRDocument &&) noexcept = default; +MIRDocument::~MIRDocument() = default; + +MIRDocumentParseResult::MIRDocumentParseResult(MIRDocumentParseResult &&) noexcept = default; +MIRDocumentParseResult & +MIRDocumentParseResult::operator=(MIRDocumentParseResult &&) noexcept = default; +MIRDocumentParseResult::~MIRDocumentParseResult() = default; + +MIRDocumentParseResult parseMIRDocument(const TargetDescription &Target, std::string SourceName, + std::string SourceText, MIRDocumentParserOptions Options) { + return MIRDocumentParser(Target, std::move(SourceName), std::move(SourceText), std::move(Options)) + .run(); +} + +} // namespace rcc::mir diff --git a/backend/lib/MIR/MIPrinter.cpp b/backend/lib/CodeGen/MIRPrinter.cpp index 0265564..d510c83 100644 --- a/backend/lib/MIR/MIPrinter.cpp +++ b/backend/lib/CodeGen/MIRPrinter.cpp @@ -1,14 +1,23 @@ // SPDX-License-Identifier: GPL-3.0-or-later -#include "rcc/MIR/MIPrinter.h" +#include "rcc/CodeGen/MIRPrinter.h" #include "rcc/CodeGen/InlineAsm.h" #include "rcc/CodeGen/InstructionDescriptor.h" +#include "rcc/CodeGen/MIRYamlMapping.h" #include "rcc/CodeGen/MachineBasicBlock.h" #include "rcc/CodeGen/MachineCFIInfo.h" +#include "rcc/CodeGen/MachineCompilationStatus.h" +#include "rcc/CodeGen/MachineConstantPool.h" +#include "rcc/CodeGen/MachineFrameInfo.h" #include "rcc/CodeGen/MachineFunction.h" +#include "rcc/CodeGen/MachineFunctionAttributes.h" +#include "rcc/CodeGen/MachineFunctionProperties.h" +#include "rcc/CodeGen/MachineFunctionRecords.h" #include "rcc/CodeGen/MachineInstr.h" +#include "rcc/CodeGen/MachineJumpTableInfo.h" #include "rcc/CodeGen/MachineMemOperand.h" +#include "rcc/CodeGen/MachineMetadata.h" #include "rcc/CodeGen/MachineOperand.h" #include "rcc/CodeGen/MachineRegisterInfo.h" #include "rcc/CodeGen/MachineVerifier.h" @@ -16,11 +25,11 @@ #include "rcc/CodeGen/TargetMIRTextInfo.h" #include "rcc/CodeGen/TargetRegisterInfo.h" #include "rcc/MIR/LLVMConstantSyntax.h" -#include "rcc/MIR/MIDebugSyntax.h" #include <algorithm> #include <array> #include <bit> +#include <charconv> #include <cstddef> #include <cstdint> #include <limits> @@ -30,12 +39,1476 @@ #include <string> #include <string_view> #include <unordered_map> +#include <unordered_set> #include <utility> +#include <variant> #include <vector> namespace rcc::mir { namespace { +enum class YAMLQuoting : std::uint8_t { + None, + Single, + Double, +}; + +[[nodiscard]] bool isASCIIAlphanumeric(unsigned char Character) noexcept { + return (Character >= '0' && Character <= '9') || (Character >= 'A' && Character <= 'Z') || + (Character >= 'a' && Character <= 'z'); +} + +[[nodiscard]] bool isYAMLNumeric(std::string_view Value) noexcept { + if (Value.empty() || Value == "+" || Value == "-") + return false; + if (Value == ".nan" || Value == ".NaN" || Value == ".NAN") + return true; + + std::string_view Tail = Value; + if (Tail.front() == '+' || Tail.front() == '-') + Tail.remove_prefix(1); + if (Tail == ".inf" || Tail == ".Inf" || Tail == ".INF") + return true; + if (Value.starts_with("0o")) { + return Value.size() > 2U && + Value.substr(2).find_first_not_of("01234567") == std::string_view::npos; + } + if (Value.starts_with("0x")) { + return Value.size() > 2U && + Value.substr(2).find_first_not_of("0123456789abcdefABCDEF") == std::string_view::npos; + } + + if (Tail.empty() || Tail.front() == 'e' || Tail.front() == 'E') + return false; + std::size_t Position = 0; + if (Tail.front() == '.') { + ++Position; + if (Position == Tail.size() || Tail[Position] < '0' || Tail[Position] > '9') + return false; + while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') + ++Position; + } else { + while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') + ++Position; + if (Position == 0) + return false; + } + + if (Position < Tail.size() && Tail[Position] == '.') { + ++Position; + while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') + ++Position; + } + if (Position == Tail.size()) + return true; + if (Tail[Position] != 'e' && Tail[Position] != 'E') + return false; + ++Position; + if (Position < Tail.size() && (Tail[Position] == '+' || Tail[Position] == '-')) + ++Position; + const std::size_t ExponentBegin = Position; + while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') + ++Position; + return Position == Tail.size() && Position != ExponentBegin; +} + +[[nodiscard]] YAMLQuoting yamlQuoting(std::string_view Value) noexcept { + if (Value.empty()) + return YAMLQuoting::Single; + + YAMLQuoting Result = YAMLQuoting::None; + const auto IsSpace = [](unsigned char Character) { + return Character == ' ' || Character == '\t' || Character == '\r' || Character == '\n' || + Character == '\f' || Character == '\v'; + }; + if (IsSpace(static_cast<unsigned char>(Value.front())) || + IsSpace(static_cast<unsigned char>(Value.back())) || Value == "null" || Value == "Null" || + Value == "NULL" || Value == "~" || Value == "true" || Value == "True" || Value == "TRUE" || + Value == "false" || Value == "False" || Value == "FALSE" || isYAMLNumeric(Value)) { + Result = YAMLQuoting::Single; + } + + switch (Value.front()) { + case '-': + case '?': + case ':': + case '\\': + case ',': + case '[': + case ']': + case '{': + case '}': + case '#': + case '&': + case '*': + case '!': + case '|': + case '>': + case '\'': + case '"': + case '%': + case '@': + case '`': + Result = YAMLQuoting::Single; + break; + default: + break; + } + + for (const char RawCharacter : Value) { + const auto Character = static_cast<unsigned char>(RawCharacter); + if (isASCIIAlphanumeric(Character)) + continue; + switch (Character) { + case '_': + case '-': + case '^': + case '.': + case ',': + case ' ': + case '\t': + continue; + case '\n': + case '\r': + case 0x7f: + return YAMLQuoting::Double; + default: + if (Character <= 0x1fU || (Character & 0x80U) != 0) + return YAMLQuoting::Double; + Result = YAMLQuoting::Single; + break; + } + } + return Result; +} + +[[nodiscard]] std::string yamlScalar(std::string_view Value) { + const YAMLQuoting Quoting = yamlQuoting(Value); + if (Quoting == YAMLQuoting::None) + return std::string(Value); + + std::string Result; + Result.reserve(Value.size() + 2U); + const char Quote = Quoting == YAMLQuoting::Single ? '\'' : '"'; + Result.push_back(Quote); + for (const char Character : Value) { + if (Quoting == YAMLQuoting::Single) { + Result.push_back(Character); + if (Character == '\'') + Result.push_back('\''); + continue; + } + switch (Character) { + case '\\': + Result.append("\\\\"); + break; + case '"': + Result.append("\\\""); + break; + case '\0': + Result.append("\\0"); + break; + case '\a': + Result.append("\\a"); + break; + case '\b': + Result.append("\\b"); + break; + case '\t': + Result.append("\\t"); + break; + case '\n': + Result.append("\\n"); + break; + case '\v': + Result.append("\\v"); + break; + case '\f': + Result.append("\\f"); + break; + case '\r': + Result.append("\\r"); + break; + case '\x1b': + Result.append("\\e"); + break; + default: { + const auto Byte = static_cast<unsigned char>(Character); + if (Byte < 0x20U || Byte == 0x7fU) { + static constexpr std::string_view Hex = "0123456789ABCDEF"; + Result.append("\\x"); + Result.push_back(Hex[Byte >> 4U]); + Result.push_back(Hex[Byte & 0xfU]); + } else { + Result.push_back(Character); + } + break; + } + } + } + Result.push_back(Quote); + return Result; +} + +void appendBlockScalar(std::string &Output, std::string_view Text, std::size_t Indent) { + std::size_t Position = 0; + do { + Output.append(Indent, ' '); + const std::size_t End = Text.find('\n', Position); + if (End == std::string_view::npos) { + Output.append(Text.substr(Position)); + Output.push_back('\n'); + return; + } + Output.append(Text.substr(Position, End - Position)); + Output.push_back('\n'); + Position = End + 1U; + } while (Position < Text.size()); +} + +[[nodiscard]] std::optional<std::uint32_t> parseDecimal32(std::string_view Text) noexcept { + std::uint32_t Value = 0; + if (Text.empty()) + return std::nullopt; + const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, 10); + if (Error != std::errc{} || End != Text.data() + Text.size()) + return std::nullopt; + return Value; +} + +[[nodiscard]] bool isDefaultFrameState(const MachineFrameState &State) { + return State == MachineFrameState{}; +} + +class MIRDocumentPrinter final { +public: + MIRDocumentPrinter(const MIRDocument &Document, const TargetDescription &Target, + MIRDocumentPrinterOptions Options) + : Document(Document), Target(Target), Options(std::move(Options)) {} + + [[nodiscard]] Expected<std::string> run(); + +private: + struct InstructionLocation final { + std::uint32_t Block = 0; + std::uint32_t Offset = 0; + }; + + [[nodiscard]] Expected<void> printFunction(const MachineFunction &Function, + const MIRFunctionDocumentInfo *DocumentInfo); + [[nodiscard]] Expected<MIParserContext> createBodyContext(const MachineFunction &Function) const; + [[nodiscard]] Expected<void> printRegisters(const MachineFunction &Function, + const MIParserContext &Context); + [[nodiscard]] Expected<void> printFrameInfo(const MachineFunction &Function, + const MIParserContext &Context); + [[nodiscard]] Expected<void> printStackObjects(const MachineFunction &Function, + const MIParserContext &Context, bool Fixed); + [[nodiscard]] Expected<void> printConstants(const MachineFunction &Function, + const MIParserContext &Context); + [[nodiscard]] Expected<void> printJumpTable(const MachineFunction &Function, + const MIParserContext &Context); + [[nodiscard]] Expected<void> printMetadata(const MachineFunction &Function, + const MIParserContext &Context); + [[nodiscard]] Expected<void> printTargetState(const MachineFunction &Function); + [[nodiscard]] Expected<void> printFunctionRecords(const MachineFunction &Function, + const MIParserContext &Context); + [[nodiscard]] Expected<void> verifyOmittedEmbeddedIRLayout(const MachineFunction &Function) const; + + [[nodiscard]] Expected<std::string> physicalRegister(PhysicalRegister Register) const; + [[nodiscard]] Expected<std::string> frameReference(FrameIndex Index, + const MIParserContext &Context) const; + [[nodiscard]] Expected<std::string> metadataReference(MachineMetadataReference Reference, + const MIParserContext &Context) const; + [[nodiscard]] Expected<std::string> metadataValue(const MachineMetadataValue &Value, + const MIParserContext &Context) const; + [[nodiscard]] Expected<std::string> debugLocation(const MachineDebugLocation &Location, + const MIParserContext &Context) const; + [[nodiscard]] Expected<std::string> diLocation(const MachineDILocation &Location, + std::size_t FrameIndex, + const MIParserContext &Context) const; + [[nodiscard]] Expected<InstructionLocation> instructionLocation(const MachineFunction &Function, + MachineInstrStableID ID) const; + [[nodiscard]] Expected<std::string> blockReference(const MachineFunction &Function, + MachineBlockStableID ID) const; + [[nodiscard]] Expected<std::string> targetNode(const MIRTargetNode &Node) const; + [[nodiscard]] const MIRFunctionDocumentInfo * + documentInfo(const MachineFunction &Function) const noexcept; + + void printBoolean(std::string_view Key, bool Value, bool Always = false); + void printUnsigned(std::string_view Key, std::uint64_t Value, bool Always = false); + void printSigned(std::string_view Key, std::int64_t Value, bool Always = false); + + const MIRDocument &Document; + const TargetDescription &Target; + MIRDocumentPrinterOptions Options; + std::string Output; +}; + +void MIRDocumentPrinter::printBoolean(std::string_view Key, bool Value, bool Always) { + if (!Always && Options.Mode == MIPrintMode::Readable && !Value) + return; + Output.append(Key); + Output.append(Value ? ": true\n" : ": false\n"); +} + +void MIRDocumentPrinter::printUnsigned(std::string_view Key, std::uint64_t Value, bool Always) { + if (!Always && Options.Mode == MIPrintMode::Readable && Value == 0) + return; + Output.append(Key); + Output.append(": "); + Output.append(std::to_string(Value)); + Output.push_back('\n'); +} + +void MIRDocumentPrinter::printSigned(std::string_view Key, std::int64_t Value, bool Always) { + if (!Always && Options.Mode == MIPrintMode::Readable && Value == 0) + return; + Output.append(Key); + Output.append(": "); + Output.append(std::to_string(Value)); + Output.push_back('\n'); +} + +Expected<std::string> MIRDocumentPrinter::physicalRegister(PhysicalRegister Register) const { + const auto Name = Target.registers().physicalRegisterMIRName(Register); + if (!Name) { + return makeUnexpected(ErrorCode::VerificationError, + "physical register has no canonical MIR name"); + } + return "$" + std::string(*Name); +} + +const MIRFunctionDocumentInfo * +MIRDocumentPrinter::documentInfo(const MachineFunction &Function) const noexcept { + const auto Infos = Document.functionDocumentInfo(); + const auto Found = std::ranges::find_if(Infos, [&Function](const MIRFunctionDocumentInfo &Info) { + return Info.Function == Function.stableID(); + }); + return Found == Infos.end() ? nullptr : &*Found; +} + +Expected<void> +MIRDocumentPrinter::verifyOmittedEmbeddedIRLayout(const MachineFunction &Function) const { + if (Options.EmitEmbeddedIR) + return {}; + + const EmbeddedIRReferenceIndex *Embedded = Document.embeddedIR(); + if (Embedded == nullptr) + return {}; + + bool HasLayoutDependentContent = false; + for (const VirtualRegisterInfo &RegisterInfo : Function.registerInfo().virtualRegisters()) { + if (!RegisterInfo.IsTombstone && RegisterInfo.Type && + RegisterInfo.Type->isPointerOrPointerVector()) { + HasLayoutDependentContent = true; + break; + } + } + if (!HasLayoutDependentContent) { + for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { + const MachineConstantPoolEntryDesc *Description = Entry.description(); + if (Description != nullptr && + std::holds_alternative<GenericConstantData>(Description->Value)) { + HasLayoutDependentContent = true; + break; + } + } + } + if (!HasLayoutDependentContent) + return {}; + + if (Options.DataLayout.empty()) { + return makeUnexpected( + ErrorCode::InvalidArgument, + "omitting embedded LLVM IR with layout-dependent MIR requires an explicit printer " + "DataLayout"); + } + + const LLVMConstantSyntaxOptions LayoutOptions{ + .DataLayout = Options.DataLayout, + .NamedTypes = {}, + }; + for (const VirtualRegisterInfo &RegisterInfo : Function.registerInfo().virtualRegisters()) { + if (RegisterInfo.IsTombstone || !RegisterInfo.Type || + !RegisterInfo.Type->isPointerOrPointerVector()) { + continue; + } + const auto AddressSpace = RegisterInfo.Type->addressSpace(); + if (!AddressSpace) { + return makeUnexpected(ErrorCode::VerificationError, + "pointer virtual register has no address space"); + } + auto PointerSize = llvmPointerSizeInBits(*AddressSpace, LayoutOptions); + if (!PointerSize) { + return makeUnexpected(ErrorCode::InvalidArgument, + "printer DataLayout cannot determine pointer width: " + + std::string(PointerSize.error().message())); + } + if (*PointerSize != RegisterInfo.Type->scalarBitWidth()) { + return makeUnexpected(ErrorCode::VerificationError, + "printer DataLayout does not match pointer virtual-register width"); + } + } + + for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { + const MachineConstantPoolEntryDesc *Description = Entry.description(); + if (Description == nullptr) + continue; + const auto *Generic = std::get_if<GenericConstantData>(&Description->Value); + if (Generic == nullptr) + continue; + auto StorageSize = llvmConstantStorageSize(Generic->Type, LayoutOptions); + if (!StorageSize) { + return makeUnexpected(ErrorCode::InvalidArgument, + "printer DataLayout cannot lay out generic constant: " + + std::string(StorageSize.error().message())); + } + if (*StorageSize != Description->Size) { + return makeUnexpected(ErrorCode::VerificationError, + "printer DataLayout does not match generic constant storage size"); + } + } + return {}; +} + +Expected<MIParserContext> +MIRDocumentPrinter::createBodyContext(const MachineFunction &Function) const { + MIParserContext Context; + Context.SyncScopes.assign(Document.syncScopes().begin(), Document.syncScopes().end()); + const auto VirtualRegisters = Function.registerInfo().virtualRegisters(); + for (std::size_t Index = 0; Index < VirtualRegisters.size(); ++Index) { + if (VirtualRegisters[Index].IsTombstone) + continue; + auto Reg = Register::makeVirtual(Index); + if (!Reg) + return std::unexpected(Reg.error()); + if (Index > std::numeric_limits<std::uint32_t>::max()) { + return makeUnexpected(ErrorCode::Unsupported, + "MIR cannot represent more than 2^32 virtual registers"); + } + Context.VirtualRegisters.push_back({.TextID = static_cast<std::uint32_t>(Index), .Reg = *Reg}); + } + + std::uint32_t StackID = 0; + std::uint32_t FixedID = 0; + for (const StackObject &Object : Function.frameInfo().objects()) { + const StackObjectDesc *Description = Object.description(); + if (Description == nullptr) + continue; + MIFrameObjectSlot Slot{ + .TextID = Object.isFixed() ? FixedID++ : StackID++, + .Index = Object.index(), + .Name = Description->AllocaName, + }; + (Object.isFixed() ? Context.FixedStackObjects : Context.StackObjects) + .push_back(std::move(Slot)); + } + + std::uint32_t ConstantID = 0; + for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { + if (Entry.description() != nullptr) + Context.ConstantPoolItems.push_back({.TextID = ConstantID++, .Index = Entry.index()}); + } + std::uint32_t JumpTableID = 0; + for (const MachineJumpTableEntry &Entry : Function.jumpTableInfo().tables()) { + if (Entry.description() != nullptr) + Context.JumpTables.push_back({.TextID = JumpTableID++, .Index = Entry.index()}); + } + + std::unordered_set<std::uint32_t> EmbeddedMetadataSlots; + if (const EmbeddedIRReferenceIndex *Embedded = Document.embeddedIR()) { + for (const EmbeddedIRMetadataDefinition &Definition : Embedded->metadataDefinitions()) { + const auto Slot = parseDecimal32(Definition.Name); + if (Slot) { + if (!EmbeddedMetadataSlots.insert(*Slot).second) { + return makeUnexpected(ErrorCode::VerificationError, + "embedded LLVM IR metadata slots are not unique"); + } + Context.Metadata.push_back( + {.TextID = *Slot, .Reference = MachineMetadataReference::embeddedIR(*Slot)}); + } + } + } + + std::unordered_set<std::uint32_t> UsedMetadataSlots = EmbeddedMetadataSlots; + for (const MachineMetadataNode &Node : Function.metadata().nodes()) { + if (!EmbeddedMetadataSlots.contains(Node.Slot)) + UsedMetadataSlots.insert(Node.Slot); + } + std::uint64_t NextMetadataSlot = 0; + for (const MachineMetadataNode &Node : Function.metadata().nodes()) { + std::uint32_t TextID = Node.Slot; + if (EmbeddedMetadataSlots.contains(TextID)) { + while (NextMetadataSlot <= std::numeric_limits<std::uint32_t>::max() && + UsedMetadataSlots.contains(static_cast<std::uint32_t>(NextMetadataSlot))) { + ++NextMetadataSlot; + } + if (NextMetadataSlot > std::numeric_limits<std::uint32_t>::max()) { + return makeUnexpected(ErrorCode::Unsupported, "MIR metadata text slot space is exhausted"); + } + TextID = static_cast<std::uint32_t>(NextMetadataSlot); + UsedMetadataSlots.insert(TextID); + ++NextMetadataSlot; + } + Context.Metadata.push_back( + {.TextID = TextID, .Reference = MachineMetadataReference::machineFunction(Node.Slot)}); + } + return Context; +} + +Expected<std::string> MIRDocumentPrinter::frameReference(FrameIndex Index, + const MIParserContext &Context) const { + const auto Stack = std::ranges::find_if( + Context.StackObjects, [Index](const MIFrameObjectSlot &Slot) { return Slot.Index == Index; }); + if (Stack != Context.StackObjects.end()) + return "%stack." + std::to_string(Stack->TextID); + const auto Fixed = + std::ranges::find_if(Context.FixedStackObjects, + [Index](const MIFrameObjectSlot &Slot) { return Slot.Index == Index; }); + if (Fixed != Context.FixedStackObjects.end()) + return "%fixed-stack." + std::to_string(Fixed->TextID); + return makeUnexpected(ErrorCode::VerificationError, + "frame state references an unknown stack object"); +} + +Expected<std::string> MIRDocumentPrinter::metadataReference(MachineMetadataReference Reference, + const MIParserContext &Context) const { + if (!Reference.isValid()) + return makeUnexpected(ErrorCode::VerificationError, "invalid metadata reference"); + const auto Slot = + std::ranges::find_if(Context.Metadata, [Reference](const MIMetadataSlot &Candidate) { + return Candidate.Reference == Reference; + }); + if (Slot == Context.Metadata.end()) { + return makeUnexpected(ErrorCode::VerificationError, "metadata reference has no MIR text slot"); + } + return "!" + std::to_string(Slot->TextID); +} + +Expected<std::string> MIRDocumentPrinter::metadataValue(const MachineMetadataValue &Value, + const MIParserContext &Context) const { + if (const MachineMetadataReference *Reference = Value.reference()) + return metadataReference(*Reference, Context); + const MachineDIExpression *Expression = Value.expression(); + if (Expression == nullptr) + return makeUnexpected(ErrorCode::VerificationError, "metadata value is not printable"); + return formatDIExpression(Expression->elements()); +} + +Expected<std::string> MIRDocumentPrinter::diLocation(const MachineDILocation &Location, + std::size_t FrameIndex, + const MIParserContext &Context) const { + if (FrameIndex >= Location.inlineChain().size()) { + return makeUnexpected(ErrorCode::VerificationError, + "inline DILocation has an invalid frame chain"); + } + const MachineDILocationFrame &Frame = Location.inlineChain()[FrameIndex]; + auto Scope = metadataReference(Frame.Scope, Context); + if (!Scope) + return std::unexpected(Scope.error()); + std::string Text = "!DILocation(line: " + std::to_string(Frame.Line); + if (Options.Mode == MIPrintMode::Complete || Frame.Column != 0) + Text.append(", column: " + std::to_string(Frame.Column)); + Text.append(", scope: " + *Scope); + if (FrameIndex + 1U < Location.inlineChain().size()) { + auto Nested = diLocation(Location, FrameIndex + 1U, Context); + if (!Nested) + return std::unexpected(Nested.error()); + Text.append(", inlinedAt: " + *Nested); + } else if (Location.terminalInlinedAt()) { + auto Terminal = metadataReference(*Location.terminalInlinedAt(), Context); + if (!Terminal) + return std::unexpected(Terminal.error()); + Text.append(", inlinedAt: " + *Terminal); + } + if (Options.Mode == MIPrintMode::Complete || Frame.IsImplicitCode) + Text.append(Frame.IsImplicitCode ? ", isImplicitCode: true" : ", isImplicitCode: false"); + if (Options.Mode == MIPrintMode::Complete || Frame.AtomGroup != 0) + Text.append(", atomGroup: " + std::to_string(Frame.AtomGroup)); + if (Options.Mode == MIPrintMode::Complete || Frame.AtomRank != 0) + Text.append(", atomRank: " + std::to_string(Frame.AtomRank)); + Text.push_back(')'); + return Text; +} + +Expected<std::string> MIRDocumentPrinter::debugLocation(const MachineDebugLocation &Location, + const MIParserContext &Context) const { + if (const MachineMetadataReference *Reference = Location.reference()) + return metadataReference(*Reference, Context); + if (const MachineDILocation *Inline = Location.inlineLocation()) + return diLocation(*Inline, 0, Context); + return makeUnexpected(ErrorCode::VerificationError, "debug location is not printable"); +} + +Expected<MIRDocumentPrinter::InstructionLocation> +MIRDocumentPrinter::instructionLocation(const MachineFunction &Function, + MachineInstrStableID ID) const { + std::uint64_t BlockNumber = 0; + for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; + Block = Block->next(), ++BlockNumber) { + if (BlockNumber > std::numeric_limits<std::uint32_t>::max()) + break; + std::uint64_t Offset = 0; + for (const MachineInstr *Instruction = Block->firstInstruction(); Instruction != nullptr; + Instruction = Instruction->next(), ++Offset) { + if (Instruction->stableID() != ID) + continue; + if (Offset > std::numeric_limits<std::uint32_t>::max()) { + return makeUnexpected(ErrorCode::Unsupported, "MIR instruction offset exceeds 32 bits"); + } + return InstructionLocation{static_cast<std::uint32_t>(BlockNumber), + static_cast<std::uint32_t>(Offset)}; + } + } + return makeUnexpected(ErrorCode::VerificationError, + "function record references an unlinked instruction"); +} + +Expected<std::string> MIRDocumentPrinter::blockReference(const MachineFunction &Function, + MachineBlockStableID ID) const { + std::uint64_t Number = 0; + for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; + Block = Block->next(), ++Number) { + if (Block->stableID() == ID) { + if (Number > std::numeric_limits<std::uint32_t>::max()) + break; + return "%bb." + std::to_string(Number); + } + } + return makeUnexpected(ErrorCode::VerificationError, + "frame record references an unlinked basic block"); +} + +Expected<std::string> MIRDocumentPrinter::targetNode(const MIRTargetNode &Node) const { + switch (Node.Kind) { + case MIRTargetNodeKind::Null: + return std::string("null"); + case MIRTargetNodeKind::Scalar: + return yamlScalar(Node.Scalar); + case MIRTargetNodeKind::Sequence: { + std::string Text = "[ "; + for (std::size_t Index = 0; Index < Node.Sequence.size(); ++Index) { + if (Index != 0) + Text.append(", "); + auto Element = targetNode(Node.Sequence[Index]); + if (!Element) + return std::unexpected(Element.error()); + Text.append(*Element); + } + Text.append(" ]"); + return Text; + } + case MIRTargetNodeKind::Mapping: { + std::string Text = "{ "; + for (std::size_t Index = 0; Index < Node.Mapping.size(); ++Index) { + if (Index != 0) + Text.append(", "); + Text.append(yamlScalar(Node.Mapping[Index].Key)); + Text.append(": "); + auto Value = targetNode(Node.Mapping[Index].Value); + if (!Value) + return std::unexpected(Value.error()); + Text.append(*Value); + } + Text.append(" }"); + return Text; + } + case MIRTargetNodeKind::Count: + break; + } + return makeUnexpected(ErrorCode::VerificationError, "target MIR node has an invalid kind"); +} + +Expected<void> MIRDocumentPrinter::printRegisters(const MachineFunction &Function, + const MIParserContext &Context) { + if (Context.VirtualRegisters.empty()) { + if (Options.Mode == MIPrintMode::Complete) + Output.append("registers: []\n"); + } else { + Output.append("registers:\n"); + for (const MIVirtualRegisterSlot &Slot : Context.VirtualRegisters) { + auto InfoResult = Function.registerInfo().virtualRegisterInfo(Slot.Reg); + if (!InfoResult) + return std::unexpected(InfoResult.error()); + const VirtualRegisterInfo &Info = InfoResult->get(); + std::string Class = "_"; + if (Info.RegClass) { + const auto Name = Target.registers().registerClassMIRName(*Info.RegClass); + if (!Name) + return makeUnexpected(ErrorCode::VerificationError, + "virtual register class has no canonical MIR name"); + Class = *Name; + } else if (Info.Bank) { + const auto Name = Target.registers().registerBankMIRName(*Info.Bank); + if (!Name) + return makeUnexpected(ErrorCode::VerificationError, + "virtual register bank has no canonical MIR name"); + Class = *Name; + } + if (Info.AllocationHints.size() > 1) { + return makeUnexpected( + ErrorCode::Unsupported, + "fixed LLVM MIR can serialize only one simple preferred-register hint"); + } + + Output.append(" - { id: "); + Output.append(std::to_string(Slot.TextID)); + Output.append(", class: "); + Output.append(yamlScalar(Class)); + if (!Info.AllocationHints.empty()) { + auto Preferred = physicalRegister(Info.AllocationHints.front()); + if (!Preferred) + return std::unexpected(Preferred.error()); + Output.append(", preferred-register: "); + Output.append(yamlScalar(*Preferred)); + } + if (Info.TargetFlags != 0) { + auto Names = Target.mirText().virtualRegisterFlagNames(Info.TargetFlags); + if (!Names) + return std::unexpected(Names.error()); + Output.append(", flags: [ "); + for (std::size_t Index = 0; Index < Names->size(); ++Index) { + if (Index != 0) + Output.append(", "); + Output.append(yamlScalar((*Names)[Index])); + } + Output.append(" ]"); + } + Output.append(" }\n"); + } + } + + if (Function.registerInfo().liveIns().empty()) { + if (Options.Mode == MIPrintMode::Complete) + Output.append("liveins: []\n"); + } else { + Output.append("liveins:\n"); + for (const PhysicalRegisterLiveIn &LiveIn : Function.registerInfo().liveIns()) { + auto Name = physicalRegister(LiveIn.PhysReg); + if (!Name) + return std::unexpected(Name.error()); + Output.append(" - { reg: "); + Output.append(yamlScalar(*Name)); + if (LiveIn.InitialVirtualRegister) { + const auto Slot = std::ranges::find_if( + Context.VirtualRegisters, [&LiveIn](const MIVirtualRegisterSlot &Candidate) { + return Candidate.Reg == *LiveIn.InitialVirtualRegister; + }); + if (Slot == Context.VirtualRegisters.end()) { + return makeUnexpected(ErrorCode::VerificationError, + "live-in references an unknown virtual register"); + } + Output.append(", virtual-reg: "); + Output.append(yamlScalar("%" + std::to_string(Slot->TextID))); + } + Output.append(" }\n"); + } + } + + if (Function.records().hasCalleeSavedRegisters()) { + Output.append("calleeSavedRegisters: [ "); + const auto Registers = Function.records().calleeSavedRegisters(); + for (std::size_t Index = 0; Index < Registers.size(); ++Index) { + if (Index != 0) + Output.append(", "); + auto Name = physicalRegister(Registers[Index]); + if (!Name) + return std::unexpected(Name.error()); + Output.append(yamlScalar(*Name)); + } + Output.append(" ]\n"); + } + return {}; +} + +Expected<void> MIRDocumentPrinter::printFrameInfo(const MachineFunction &Function, + const MIParserContext &Context) { + const MachineFrameState &State = Function.frameInfo().state(); + if (Options.Mode == MIPrintMode::Readable && isDefaultFrameState(State)) + return {}; + if (State.MaxCallFrameSize && + *State.MaxCallFrameSize > std::numeric_limits<std::uint32_t>::max()) { + return makeUnexpected(ErrorCode::Unsupported, + "fixed LLVM MIR maxCallFrameSize is limited to 32 bits"); + } + Output.append("frameInfo:\n"); + const auto Bool = [this](std::string_view Key, bool Value) { + if (Options.Mode == MIPrintMode::Readable && !Value) + return; + Output.append(" "); + Output.append(Key); + Output.append(Value ? ": true\n" : ": false\n"); + }; + const auto Unsigned = [this](std::string_view Key, std::uint64_t Value) { + if (Options.Mode == MIPrintMode::Readable && Value == 0) + return; + Output.append(" "); + Output.append(Key); + Output.append(": "); + Output.append(std::to_string(Value)); + Output.push_back('\n'); + }; + + Bool("isFrameAddressTaken", State.IsFrameAddressTaken); + Bool("isReturnAddressTaken", State.IsReturnAddressTaken); + Bool("hasStackMap", State.HasStackMap); + Bool("hasPatchPoint", State.HasPatchPoint); + Unsigned("stackSize", State.StackSize); + if (Options.Mode == MIPrintMode::Complete || State.OffsetAdjustment != 0) { + Output.append(" offsetAdjustment: "); + Output.append(std::to_string(State.OffsetAdjustment)); + Output.push_back('\n'); + } + Unsigned("maxAlignment", State.MaxAlignment ? State.MaxAlignment->bytes() : 0); + Bool("adjustsStack", State.AdjustsStack); + Bool("hasCalls", State.HasCalls); + if (Options.Mode == MIPrintMode::Complete || + State.FramePointerPolicy != MachineFramePointerPolicy::None) { + Output.append(" framePointerPolicy: "); + Output.append(getMachineFramePointerPolicyName(State.FramePointerPolicy)); + Output.push_back('\n'); + } + if (Options.Mode == MIPrintMode::Complete || State.StackProtector) { + std::string Reference; + if (State.StackProtector) { + auto PrintedReference = frameReference(*State.StackProtector, Context); + if (!PrintedReference) + return std::unexpected(PrintedReference.error()); + Reference = std::move(*PrintedReference); + } + Output.append(" stackProtector: "); + Output.append(yamlScalar(Reference)); + Output.push_back('\n'); + } + if (Options.Mode == MIPrintMode::Complete || State.FunctionContext) { + std::string Reference; + if (State.FunctionContext) { + auto PrintedReference = frameReference(*State.FunctionContext, Context); + if (!PrintedReference) + return std::unexpected(PrintedReference.error()); + Reference = std::move(*PrintedReference); + } + Output.append(" functionContext: "); + Output.append(yamlScalar(Reference)); + Output.push_back('\n'); + } + if (Options.Mode == MIPrintMode::Complete || State.MaxCallFrameSize) { + Unsigned("maxCallFrameSize", + State.MaxCallFrameSize.value_or(std::numeric_limits<std::uint32_t>::max())); + } + Unsigned("cvBytesOfCalleeSavedRegisters", State.CVBytesOfCalleeSavedRegisters); + Bool("hasOpaqueSPAdjustment", State.HasOpaqueSPAdjustment); + Bool("hasVAStart", State.HasVAStart); + Bool("hasMustTailInVarArgFunc", State.HasMustTailInVarArgFunc); + Bool("hasTailCall", State.HasTailCall); + Bool("isCalleeSavedInfoValid", State.IsCalleeSavedInfoValid); + Unsigned("localFrameSize", State.LocalFrameSize); + + const auto PrintPoints = + [this, &Function](std::string_view Key, + std::span<const MachineFrameSaveRestorePoint> Points) -> Expected<void> { + if (Points.empty()) + return {}; + Output.append(" "); + Output.append(Key); + Output.append(":\n"); + for (const MachineFrameSaveRestorePoint &Point : Points) { + auto Block = blockReference(Function, Point.Block); + if (!Block) + return std::unexpected(Block.error()); + Output.append(" - { point: "); + Output.append(yamlScalar(*Block)); + if (!Point.Registers.empty()) { + Output.append(", registers: [ "); + for (std::size_t Index = 0; Index < Point.Registers.size(); ++Index) { + if (Index != 0) + Output.append(", "); + auto Name = physicalRegister(Point.Registers[Index]); + if (!Name) + return std::unexpected(Name.error()); + Output.append(yamlScalar(*Name)); + } + Output.append(" ]"); + } + Output.append(" }\n"); + } + return {}; + }; + if (auto Printed = PrintPoints("savePoint", State.SavePoints); !Printed) + return Printed; + return PrintPoints("restorePoint", State.RestorePoints); +} + +Expected<void> MIRDocumentPrinter::printStackObjects(const MachineFunction &Function, + const MIParserContext &Context, bool Fixed) { + const auto &Slots = Fixed ? Context.FixedStackObjects : Context.StackObjects; + if (Slots.empty()) { + if (Options.Mode == MIPrintMode::Complete) + Output.append(Fixed ? "fixedStack: []\n" : "stack: []\n"); + return {}; + } + Output.append(Fixed ? "fixedStack:\n" : "stack:\n"); + for (const MIFrameObjectSlot &Slot : Slots) { + const StackObject *Object = Function.frameInfo().object(Slot.Index); + const StackObjectDesc *Description = Object == nullptr ? nullptr : Object->description(); + if (Description == nullptr) + return makeUnexpected(ErrorCode::VerificationError, "MIR stack slot is a tombstone"); + if (Description->IsDead || Description->AddressTaken || !Description->Lifetime.empty() || + Description->TargetFlags != 0) { + return makeUnexpected(ErrorCode::Unsupported, + "stack object contains state not represented by fixed LLVM MIR"); + } + + std::string_view Type; + switch (Description->Kind) { + case StackObjectKind::Default: + case StackObjectKind::FixedABI: + Type = "default"; + break; + case StackObjectKind::FixedSpillSlot: + case StackObjectKind::SpillSlot: + Type = "spill-slot"; + break; + case StackObjectKind::VariableSized: + Type = "variable-sized"; + break; + case StackObjectKind::DynamicAlloca: + case StackObjectKind::CalleeSaved: + case StackObjectKind::EmergencySpill: + case StackObjectKind::Target: + case StackObjectKind::Count: + return makeUnexpected(ErrorCode::Unsupported, + "stack object kind has no fixed LLVM MIR representation"); + } + + Output.append(" - { id: "); + Output.append(std::to_string(Slot.TextID)); + if (!Fixed && Description->AllocaName) { + Output.append(", name: "); + Output.append(yamlScalar(*Description->AllocaName)); + } + if (Options.Mode == MIPrintMode::Complete || Type != "default") { + Output.append(", type: "); + Output.append(Type); + } + if (Options.Mode == MIPrintMode::Complete || Description->Offset.value_or(0) != 0) { + Output.append(", offset: "); + Output.append(std::to_string(Description->Offset.value_or(0))); + } + if (Description->Kind != StackObjectKind::VariableSized) { + Output.append(", size: "); + Output.append(std::to_string(Description->Size.value_or(0))); + } + if (Options.Mode == MIPrintMode::Complete || Description->Align.bytes() != 1) { + Output.append(", alignment: "); + Output.append(std::to_string(Description->Align.bytes())); + } + if (Options.Mode == MIPrintMode::Complete || Description->Stack != StackID::defaultStack()) { + Output.append(", stack-id: "); + Output.append(getStackIDName(Description->Stack)); + } + if (Fixed && Description->Kind != StackObjectKind::FixedSpillSlot) { + if (Options.Mode == MIPrintMode::Complete || Description->IsImmutable) + Output.append(Description->IsImmutable ? ", isImmutable: true" : ", isImmutable: false"); + if (Options.Mode == MIPrintMode::Complete || Description->IsAliased) + Output.append(Description->IsAliased ? ", isAliased: true" : ", isAliased: false"); + } else if (!Fixed && (Description->IsImmutable || Description->IsAliased)) { + return makeUnexpected(ErrorCode::Unsupported, + "ordinary stack object immutable/aliased state is not serialized"); + } + if (Description->CalleeSavedRegister) { + auto Name = physicalRegister(*Description->CalleeSavedRegister); + if (!Name) + return std::unexpected(Name.error()); + Output.append(", callee-saved-register: "); + Output.append(yamlScalar(*Name)); + } + if (Options.Mode == MIPrintMode::Complete || !Description->CalleeSavedRestored) { + Output.append(Description->CalleeSavedRestored ? ", callee-saved-restored: true" + : ", callee-saved-restored: false"); + } + if (!Fixed && Description->LocalOffset) { + Output.append(", local-offset: "); + Output.append(std::to_string(*Description->LocalOffset)); + } + if (Description->DebugInfo) { + const MachineVariableDebugInfo &Debug = *Description->DebugInfo; + if (Debug.Variable) { + auto Value = metadataReference(*Debug.Variable, Context); + if (!Value) + return std::unexpected(Value.error()); + Output.append(", debug-info-variable: "); + Output.append(yamlScalar(*Value)); + } + if (Debug.Expression) { + auto Value = metadataValue(*Debug.Expression, Context); + if (!Value) + return std::unexpected(Value.error()); + Output.append(", debug-info-expression: "); + Output.append(yamlScalar(*Value)); + } + if (Debug.Location) { + auto Value = debugLocation(*Debug.Location, Context); + if (!Value) + return std::unexpected(Value.error()); + Output.append(", debug-info-location: "); + Output.append(yamlScalar(*Value)); + } + } + Output.append(" }\n"); + } + return {}; +} + +Expected<void> MIRDocumentPrinter::printConstants(const MachineFunction &Function, + const MIParserContext &Context) { + if (Context.ConstantPoolItems.empty()) { + if (Options.Mode == MIPrintMode::Complete) + Output.append("constants: []\n"); + return {}; + } + Output.append("constants:\n"); + for (const MIConstantPoolSlot &Slot : Context.ConstantPoolItems) { + const MachineConstantPoolEntry *Entry = Function.constantPool().entry(Slot.Index); + const MachineConstantPoolEntryDesc *Description = + Entry == nullptr ? nullptr : Entry->description(); + if (Description == nullptr) + return makeUnexpected(ErrorCode::VerificationError, "constant-pool slot is a tombstone"); + if (Description->Merge != ConstantPoolMergeKind::Never || + Description->Relocation != ConstantPoolRelocationMode::None || + Description->Placement != ConstantPoolPlacement{}) { + return makeUnexpected(ErrorCode::Unsupported, + "constant-pool placement state is not represented by fixed LLVM MIR"); + } + + bool IsTargetSpecific = false; + Expected<std::string> Value = + makeUnexpected(ErrorCode::Unsupported, "constant-pool kind is not MIR serializable"); + if (const auto *Generic = std::get_if<GenericConstantData>(&Description->Value)) { + Value = printLLVMConstant(*Generic); + } else if (const auto *TargetConstantValue = std::get_if<TargetConstant>(&Description->Value)) { + IsTargetSpecific = true; + Value = Target.mirText().printTargetConstant(*TargetConstantValue); + } + if (!Value) + return std::unexpected(Value.error()); + + Output.append(" - id: "); + Output.append(std::to_string(Slot.TextID)); + Output.append("\n value: "); + Output.append(yamlScalar(*Value)); + Output.append("\n alignment: "); + Output.append(std::to_string(Description->Align.bytes())); + Output.push_back('\n'); + if (IsTargetSpecific) + Output.append(" isTargetSpecific: true\n"); + } + return {}; +} + +Expected<void> MIRDocumentPrinter::printJumpTable(const MachineFunction &Function, + const MIParserContext &Context) { + if (Context.JumpTables.empty()) + return {}; + const MachineJumpTableEntry *FirstEntry = + Function.jumpTableInfo().table(Context.JumpTables.front().Index); + const MachineJumpTableDesc *First = FirstEntry == nullptr ? nullptr : FirstEntry->description(); + if (First == nullptr) + return makeUnexpected(ErrorCode::VerificationError, "jump-table slot is not defined"); + + std::string_view Kind; + std::uint8_t ExpectedSize = 8; + switch (First->Encoding) { + case JumpTableEntryEncoding::BlockAddress: + Kind = "block-address"; + break; + case JumpTableEntryEncoding::GPRelative64BlockAddress: + Kind = "gp-rel64-block-address"; + break; + case JumpTableEntryEncoding::GPRelative32BlockAddress: + Kind = "gp-rel32-block-address"; + ExpectedSize = 4; + break; + case JumpTableEntryEncoding::LabelDifference32: + Kind = "label-difference32"; + ExpectedSize = 4; + break; + case JumpTableEntryEncoding::LabelDifference64: + Kind = "label-difference64"; + break; + case JumpTableEntryEncoding::Inline: + Kind = "inline"; + ExpectedSize = 4; + break; + case JumpTableEntryEncoding::Custom32: + Kind = "custom32"; + ExpectedSize = 4; + break; + case JumpTableEntryEncoding::Target: + case JumpTableEntryEncoding::Count: + return makeUnexpected(ErrorCode::Unsupported, + "target jump-table encoding has no fixed LLVM MIR spelling"); + } + + Output.append("jumpTable:\n kind: "); + Output.append(Kind); + Output.append("\n entries:\n"); + for (const MIJumpTableSlot &Slot : Context.JumpTables) { + const MachineJumpTableEntry *Entry = Function.jumpTableInfo().table(Slot.Index); + const MachineJumpTableDesc *Description = Entry == nullptr ? nullptr : Entry->description(); + if (Description == nullptr) + return makeUnexpected(ErrorCode::VerificationError, "jump-table slot is not defined"); + if (Description->Encoding != First->Encoding || Description->EntrySize != ExpectedSize || + Description->Align.bytes() != ExpectedSize || + Description->Placement != JumpTablePlacement{} || Description->TargetEncoding != 0) { + return makeUnexpected(ErrorCode::Unsupported, + "jump-table entries cannot share one fixed LLVM MIR kind"); + } + Output.append(" - { id: "); + Output.append(std::to_string(Slot.TextID)); + Output.append(", blocks: [ "); + for (std::size_t Index = 0; Index < Description->Targets.size(); ++Index) { + if (Index != 0) + Output.append(", "); + if (Description->Targets[Index] == nullptr) + return makeUnexpected(ErrorCode::VerificationError, + "jump-table entry contains a null block"); + auto Block = blockReference(Function, Description->Targets[Index]->stableID()); + if (!Block) + return std::unexpected(Block.error()); + Output.append(yamlScalar(*Block)); + } + Output.append(" ] }\n"); + } + return {}; +} + +Expected<void> MIRDocumentPrinter::printMetadata(const MachineFunction &Function, + const MIParserContext &Context) { + if (Function.metadata().nodes().empty()) + return {}; + Output.append("machineMetadataNodes:\n"); + for (const MachineMetadataNode &Node : Function.metadata().nodes()) { + auto NodeReference = + metadataReference(MachineMetadataReference::machineFunction(Node.Slot), Context); + if (!NodeReference) + return std::unexpected(NodeReference.error()); + std::string Text = *NodeReference + " = "; + if (Node.IsDistinct) + Text.append("distinct "); + Text.append("!{"); + for (std::size_t Index = 0; Index < Node.Elements.size(); ++Index) { + if (Index != 0) + Text.append(", "); + const MachineMetadataElement &Element = Node.Elements[Index]; + if (Element.isNull()) { + Text.append("null"); + } else if (const MachineMetadataReference *Reference = Element.reference()) { + auto Printed = metadataReference(*Reference, Context); + if (!Printed) + return std::unexpected(Printed.error()); + Text.append(*Printed); + } else if (const std::string *String = Element.string()) { + Text.append("!\""); + for (const char Character : *String) { + if (Character == '\\' || Character == '"') + Text.push_back('\\'); + Text.push_back(Character); + } + Text.push_back('"'); + } else { + return makeUnexpected(ErrorCode::VerificationError, + "machine metadata tuple contains an invalid element"); + } + } + Text.push_back('}'); + Output.append(" - "); + Output.append(yamlScalar(Text)); + Output.push_back('\n'); + } + return {}; +} + +Expected<void> MIRDocumentPrinter::printTargetState(const MachineFunction &Function) { + if (!Function.targetState().hasValue()) + return {}; + auto Node = Target.mirText().printFunctionState(Function.targetState()); + if (!Node) + return std::unexpected(Node.error()); + auto Text = targetNode(*Node); + if (!Text) + return std::unexpected(Text.error()); + Output.append("machineFunctionInfo: "); + Output.append(*Text); + Output.push_back('\n'); + return {}; +} + +Expected<void> MIRDocumentPrinter::printFunctionRecords(const MachineFunction &Function, + const MIParserContext &Context) { + const MachineFunctionRecords &Records = Function.records(); + if (!Records.entryValues().empty()) { + Output.append("entry_values:\n"); + for (const MachineEntryValueInfo &Entry : Records.entryValues()) { + auto RegisterName = physicalRegister(Entry.Reg); + if (!RegisterName) + return std::unexpected(RegisterName.error()); + if (!Entry.DebugInfo.Variable || !Entry.DebugInfo.Expression || !Entry.DebugInfo.Location) { + return makeUnexpected(ErrorCode::Unsupported, + "MIR entry value requires complete debug metadata"); + } + auto Variable = metadataReference(*Entry.DebugInfo.Variable, Context); + auto Expression = metadataValue(*Entry.DebugInfo.Expression, Context); + auto Location = debugLocation(*Entry.DebugInfo.Location, Context); + if (!Variable) + return std::unexpected(Variable.error()); + if (!Expression) + return std::unexpected(Expression.error()); + if (!Location) + return std::unexpected(Location.error()); + Output.append(" - { entry-value-register: "); + Output.append(yamlScalar(*RegisterName)); + Output.append(", debug-info-variable: "); + Output.append(yamlScalar(*Variable)); + Output.append(", debug-info-expression: "); + Output.append(yamlScalar(*Expression)); + Output.append(", debug-info-location: "); + Output.append(yamlScalar(*Location)); + Output.append(" }\n"); + } + } else if (Options.Mode == MIPrintMode::Complete) { + Output.append("entry_values: []\n"); + } + + if (!Records.callSites().empty()) { + Output.append("callSites:\n"); + for (const MachineCallSiteInfo &CallSite : Records.callSites()) { + auto Location = instructionLocation(Function, CallSite.CallInstruction); + if (!Location) + return std::unexpected(Location.error()); + Output.append(" - { bb: "); + Output.append(std::to_string(Location->Block)); + Output.append(", offset: "); + Output.append(std::to_string(Location->Offset)); + if (!CallSite.ForwardedArguments.empty()) { + Output.append(", fwdArgRegs: [ "); + for (std::size_t Index = 0; Index < CallSite.ForwardedArguments.size(); ++Index) { + if (Index != 0) + Output.append(", "); + const MachineCallSiteArgument &Argument = CallSite.ForwardedArguments[Index]; + auto RegisterName = physicalRegister(Argument.Reg); + if (!RegisterName) + return std::unexpected(RegisterName.error()); + Output.append("{ arg: "); + Output.append(std::to_string(Argument.ArgumentNumber)); + Output.append(", reg: "); + Output.append(yamlScalar(*RegisterName)); + Output.append(" }"); + } + Output.append(" ]"); + } + if (!CallSite.CalleeTypeIDs.empty()) { + Output.append(", calleeTypeIds: [ "); + for (std::size_t Index = 0; Index < CallSite.CalleeTypeIDs.size(); ++Index) { + if (Index != 0) + Output.append(", "); + Output.append(std::to_string(CallSite.CalleeTypeIDs[Index])); + } + Output.append(" ]"); + } + Output.append(" }\n"); + } + } else if (Options.Mode == MIPrintMode::Complete) { + Output.append("callSites: []\n"); + } + + if (!Records.debugValueSubstitutions().empty()) { + Output.append("debugValueSubstitutions:\n"); + for (const MachineDebugValueSubstitution &Substitution : Records.debugValueSubstitutions()) { + Output.append(" - { srcinst: "); + Output.append(std::to_string(Substitution.Source.InstructionNumber)); + Output.append(", srcop: "); + Output.append(std::to_string(Substitution.Source.OperandNumber)); + Output.append(", dstinst: "); + Output.append(std::to_string(Substitution.Destination.InstructionNumber)); + Output.append(", dstop: "); + Output.append(std::to_string(Substitution.Destination.OperandNumber)); + Output.append(", subreg: "); + Output.append(std::to_string(Substitution.Subregister.value())); + Output.append(" }\n"); + } + } else if (Options.Mode == MIPrintMode::Complete) { + Output.append("debugValueSubstitutions: []\n"); + } + + if (!Records.calledGlobals().empty()) { + Output.append("calledGlobals:\n"); + for (const MachineCalledGlobal &Called : Records.calledGlobals()) { + auto Location = instructionLocation(Function, Called.CallInstruction); + if (!Location) + return std::unexpected(Location.error()); + Output.append(" - { bb: "); + Output.append(std::to_string(Location->Block)); + Output.append(", offset: "); + Output.append(std::to_string(Location->Offset)); + Output.append(", callee: "); + Output.append(yamlScalar(Called.Callee)); + Output.append(", flags: "); + Output.append(std::to_string(Called.Flags)); + Output.append(" }\n"); + } + } + + if (!Records.prefetchTargets().empty()) { + Output.append("prefetch-targets: [ "); + const auto Targets = Records.prefetchTargets(); + for (std::size_t Index = 0; Index < Targets.size(); ++Index) { + if (Index != 0) + Output.append(", "); + const MachinePrefetchTarget &TargetValue = Targets[Index]; + std::string Text = "bb_id " + std::to_string(TargetValue.Block.BaseID) + ", " + + std::to_string(TargetValue.Block.CloneID) + ", " + + std::to_string(TargetValue.CallsiteIndex); + Output.append(yamlScalar(Text)); + } + Output.append(" ]\n"); + } + return {}; +} + +Expected<void> MIRDocumentPrinter::printFunction(const MachineFunction &Function, + const MIRFunctionDocumentInfo *) { + if (Function.compilationStatus().state() != MachineCompilationState::Success) { + return makeUnexpected(ErrorCode::Unsupported, + "fixed LLVM MIR cannot serialize this compilation failure state"); + } + auto Context = createBodyContext(Function); + if (!Context) + return std::unexpected(Context.error()); + + Output.append("---\nname: "); + Output.append(yamlScalar(Function.name())); + Output.push_back('\n'); + const auto Alignment = Function.attributes().alignment(); + if (Options.Mode == MIPrintMode::Complete || Alignment) { + Output.append("alignment: "); + Output.append(std::to_string(Alignment ? Alignment->bytes() : 1U)); + Output.push_back('\n'); + } + + const MachineFunctionProperties &Properties = Function.properties(); + const MachineFunctionAttributes &Attributes = Function.attributes(); + printBoolean("exposesReturnsTwice", + Attributes.has(MachineFunctionAttribute::ExposesReturnsTwice)); + printBoolean("legalized", Properties.has(MachineFunctionProperty::Legalized)); + printBoolean("regBankSelected", Properties.has(MachineFunctionProperty::RegBankSelected)); + printBoolean("selected", Properties.has(MachineFunctionProperty::Selected)); + printBoolean("failedISel", Properties.has(MachineFunctionProperty::FailedISel)); + printBoolean("tracksRegLiveness", Properties.has(MachineFunctionProperty::TracksLiveness)); + printBoolean("hasWinCFI", Attributes.has(MachineFunctionAttribute::HasWinCFI)); + printBoolean("noPhis", Properties.has(MachineFunctionProperty::NoPHIs), true); + printBoolean("isSSA", Properties.has(MachineFunctionProperty::IsSSA), true); + printBoolean("noVRegs", Properties.has(MachineFunctionProperty::NoVRegs), true); + printBoolean("hasFakeUses", Attributes.has(MachineFunctionAttribute::HasFakeUses), true); + printBoolean("callsEHReturn", Attributes.has(MachineFunctionAttribute::CallsEHReturn)); + printBoolean("callsUnwindInit", Attributes.has(MachineFunctionAttribute::CallsUnwindInit)); + printBoolean("hasEHContTarget", Attributes.has(MachineFunctionAttribute::HasEHContTarget)); + printBoolean("hasEHScopes", Attributes.has(MachineFunctionAttribute::HasEHScopes)); + printBoolean("hasEHFunclets", Attributes.has(MachineFunctionAttribute::HasEHFunclets)); + printBoolean("isOutlined", Attributes.has(MachineFunctionAttribute::IsOutlined)); + printBoolean("debugInstrRef", Attributes.has(MachineFunctionAttribute::DebugInstrRef)); + printBoolean("failsVerification", Properties.has(MachineFunctionProperty::FailsVerification)); + printBoolean("tracksDebugUserValues", + Attributes.has(MachineFunctionAttribute::TracksDebugUserValues)); + + if (auto Printed = printRegisters(Function, *Context); !Printed) + return Printed; + if (auto Printed = printFrameInfo(Function, *Context); !Printed) + return Printed; + if (auto Printed = printStackObjects(Function, *Context, true); !Printed) + return Printed; + if (auto Printed = printStackObjects(Function, *Context, false); !Printed) + return Printed; + if (auto Printed = printFunctionRecords(Function, *Context); !Printed) + return Printed; + if (auto Printed = printConstants(Function, *Context); !Printed) + return Printed; + if (auto Printed = printTargetState(Function); !Printed) + return Printed; + if (auto Printed = printJumpTable(Function, *Context); !Printed) + return Printed; + if (auto Printed = printMetadata(Function, *Context); !Printed) + return Printed; + + auto Body = + printMachineFunctionBody(Function, Target, *Context, MIPrinterOptions{.Mode = Options.Mode}); + if (!Body) + return std::unexpected(Body.error()); + Output.append("body: |\n"); + appendBlockScalar(Output, *Body, 2); + Output.append("...\n"); + return {}; +} + +Expected<std::string> MIRDocumentPrinter::run() { + auto UnitVerification = Document.unit().verify(); + if (!UnitVerification) + return std::unexpected(UnitVerification.error()); + if (Document.unit().functionCount() == 0) { + return makeUnexpected(ErrorCode::InvalidArgument, + "MIR document must contain at least one machine function"); + } + + for (const MIRFunctionDocumentInfo &Info : Document.functionDocumentInfo()) { + if (Document.unit().function(Info.Function) == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + "MIR document info references an unknown machine function"); + } + } + + for (std::size_t Index = 0; Index < Document.unit().functionCount(); ++Index) { + const MachineFunction *Function = Document.unit().functionAt(Index); + if (Function == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + "compilation unit contains a null machine function"); + } + auto ReferenceVerification = + verifyMIRDocumentReferences(Document, *Function, Options.EmitEmbeddedIR); + if (!ReferenceVerification) + return std::unexpected(ReferenceVerification.error()); + auto LayoutVerification = verifyOmittedEmbeddedIRLayout(*Function); + if (!LayoutVerification) + return std::unexpected(LayoutVerification.error()); + } + + if (Options.EmitEmbeddedIR && Document.embeddedIR() != nullptr) { + Output.append("--- |\n"); + appendBlockScalar(Output, Document.embeddedIR()->source().text(), 2); + Output.append("...\n"); + } + for (std::size_t Index = 0; Index < Document.unit().functionCount(); ++Index) { + const MachineFunction *Function = Document.unit().functionAt(Index); + if (Function == nullptr) + return makeUnexpected(ErrorCode::VerificationError, + "compilation unit contains a null machine function"); + if (auto Printed = printFunction(*Function, documentInfo(*Function)); !Printed) + return std::unexpected(Printed.error()); + } + return std::move(Output); +} + +} // namespace + +Expected<std::string> printMIRDocument(const MIRDocument &Document, const TargetDescription &Target, + MIRDocumentPrinterOptions Options) { + return MIRDocumentPrinter(Document, Target, std::move(Options)).run(); +} + +} // namespace rcc::mir + +namespace rcc::mir { +namespace { + [[nodiscard]] bool isBareNameCharacter(char Character) noexcept { return (Character >= 'a' && Character <= 'z') || (Character >= 'A' && Character <= 'Z') || (Character >= '0' && Character <= '9') || Character == '_' || Character == '-' || diff --git a/backend/lib/CodeGen/MIRYamlMapping.cpp b/backend/lib/CodeGen/MIRYamlMapping.cpp new file mode 100644 index 0000000..9ca8cad --- /dev/null +++ b/backend/lib/CodeGen/MIRYamlMapping.cpp @@ -0,0 +1,2926 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "rcc/CodeGen/MIRYamlMapping.h" + +#include "MIRParser/MILexer.h" +#include "rcc/CodeGen/MachineBasicBlock.h" +#include "rcc/CodeGen/MachineConstantPool.h" +#include "rcc/CodeGen/MachineFrameInfo.h" +#include "rcc/CodeGen/MachineFunction.h" +#include "rcc/CodeGen/MachineFunctionAttributes.h" +#include "rcc/CodeGen/MachineFunctionProperties.h" +#include "rcc/CodeGen/MachineFunctionRecords.h" +#include "rcc/CodeGen/MachineInstr.h" +#include "rcc/CodeGen/MachineJumpTableInfo.h" +#include "rcc/CodeGen/MachineMetadata.h" +#include "rcc/CodeGen/MachineRegisterInfo.h" +#include "rcc/CodeGen/MachineVerifier.h" +#include "rcc/CodeGen/TargetDescription.h" +#include "rcc/CodeGen/TargetMIRTextInfo.h" +#include "rcc/CodeGen/TargetRegisterInfo.h" +#include "rcc/MIR/LLVMConstantSyntax.h" + +#include <algorithm> +#include <array> +#include <charconv> +#include <cstddef> +#include <cstdint> +#include <functional> +#include <initializer_list> +#include <limits> +#include <optional> +#include <ranges> +#include <span> +#include <string> +#include <string_view> +#include <system_error> +#include <unordered_map> +#include <unordered_set> +#include <utility> +#include <variant> +#include <vector> + +namespace rcc::mir { + +namespace parser_detail { + +[[nodiscard]] inline std::string_view trim(std::string_view Text) noexcept { + while (!Text.empty() && (Text.front() == ' ' || Text.front() == '\t' || Text.front() == '\r' || + Text.front() == '\n')) { + Text.remove_prefix(1); + } + while (!Text.empty() && (Text.back() == ' ' || Text.back() == '\t' || Text.back() == '\r' || + Text.back() == '\n')) { + Text.remove_suffix(1); + } + return Text; +} + +[[nodiscard]] inline bool parseUnsigned(std::string_view Text, std::uint64_t &Value) noexcept { + Text = trim(Text); + int Base = 10; + if (Text.starts_with("0x") || Text.starts_with("0X")) { + Text.remove_prefix(2); + Base = 16; + } + if (Text.empty()) + return false; + const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, Base); + return Error == std::errc{} && End == Text.data() + Text.size(); +} + +[[nodiscard]] inline bool parseSigned(std::string_view Text, std::int64_t &Value) noexcept { + Text = trim(Text); + if (Text.empty()) + return false; + const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, 10); + return Error == std::errc{} && End == Text.data() + Text.size(); +} + +[[nodiscard]] inline bool parseUnsigned32(std::string_view Text, std::uint32_t &Value) noexcept { + std::uint64_t Wide = 0; + if (!parseUnsigned(Text, Wide) || Wide > std::numeric_limits<std::uint32_t>::max()) + return false; + Value = static_cast<std::uint32_t>(Wide); + return true; +} + +[[nodiscard]] inline bool parseBool(std::string_view Text, bool &Value) noexcept { + if (Text == "true") { + Value = true; + return true; + } + if (Text == "false") { + Value = false; + return true; + } + return false; +} + +[[nodiscard]] inline std::optional<std::uint32_t> +parsePrefixedID(std::string_view Text, std::string_view Prefix) noexcept { + Text = trim(Text); + if (!Text.starts_with(Prefix)) + return std::nullopt; + Text.remove_prefix(Prefix.size()); + const std::size_t Suffix = Text.find_first_of(". \t"); + if (Suffix != std::string_view::npos) + Text = Text.substr(0, Suffix); + std::uint32_t ID = 0; + if (!parseUnsigned32(Text, ID)) + return std::nullopt; + return ID; +} + +} // namespace parser_detail + +class MetadataTextParser final { +public: + using Resolver = std::function<Expected<MachineMetadataReference>(std::uint32_t Slot)>; + + MetadataTextParser(std::string Text, Resolver Resolve) + : Lexer("MIR metadata scalar", std::move(Text)), Resolve(std::move(Resolve)) { + do { + Tokens.push_back(Lexer.next()); + } while (!Tokens.back().isErrorOrEnd()); + } + + [[nodiscard]] Expected<MachineMetadataReference> parseReferenceOnly(); + [[nodiscard]] Expected<MachineMetadataValue> parseValue(); + [[nodiscard]] Expected<MachineDebugLocation> parseLocation(); + + [[nodiscard]] const TextRange &diagnosticRange() const noexcept { + return token().range(); + } + +private: + [[nodiscard]] const MIToken &token() const noexcept { + return Tokens[Position]; + } + [[nodiscard]] bool at(MITokenKind Kind) const noexcept { + return token().is(Kind); + } + void advance() noexcept { + if (Position + 1U < Tokens.size()) + ++Position; + } + [[nodiscard]] bool consume(MITokenKind Kind) noexcept { + if (!at(Kind)) + return false; + advance(); + return true; + } + [[nodiscard]] Expected<void> expect(MITokenKind Kind, std::string_view Description); + [[nodiscard]] Expected<void> finish(); + [[nodiscard]] Expected<MachineMetadataReference> parseReference(); + [[nodiscard]] Expected<MachineDIExpression> parseExpression(); + [[nodiscard]] Expected<void> parseDILocation(std::vector<MachineDILocationFrame> &Frames, + std::optional<MachineMetadataReference> &Terminal); + + MILexer Lexer; + Resolver Resolve; + std::vector<MIToken> Tokens; + std::size_t Position = 0; +}; + +using parser_detail::parseBool; +using parser_detail::parsePrefixedID; +using parser_detail::parseSigned; +using parser_detail::parseUnsigned; +using parser_detail::parseUnsigned32; +using parser_detail::trim; + +bool MIRDocumentParser::boolean(const yaml::Node &Node, bool &Value) { + std::string_view Text; + if (scalar(Node, Text, "boolean") || !parseBool(Text, Value)) + return fail(Node.range(), "expected 'true' or 'false'"); + return false; +} + +bool MIRDocumentParser::unsignedValue(const yaml::Node &Node, std::uint64_t &Value, + std::string_view Description) { + std::string_view Text; + if (scalar(Node, Text, Description) || !parseUnsigned(Text, Value)) + return fail(Node.range(), "expected " + std::string(Description)); + return false; +} + +bool MIRDocumentParser::signedValue(const yaml::Node &Node, std::int64_t &Value, + std::string_view Description) { + std::string_view Text; + if (scalar(Node, Text, Description) || !parseSigned(Text, Value)) + return fail(Node.range(), "expected " + std::string(Description)); + return false; +} + +bool MIRDocumentParser::alignment(const yaml::Node &Node, Alignment &Value, bool AllowZero) { + std::uint64_t Bytes = 0; + if (unsignedValue(Node, Bytes, "an unsigned alignment")) + return true; + if (Bytes == 0 && AllowZero) { + Value = Alignment{}; + return false; + } + auto Parsed = Alignment::create(Bytes); + if (!Parsed) + return importError(Node.range(), Parsed.error()); + Value = *Parsed; + return false; +} + +bool MIRDocumentParser::physicalRegister(std::string_view Text, const TextRange &Range, + PhysicalRegister &Reg) { + Text = trim(Text); + if (!Text.starts_with('$') || Text.size() == 1) + return fail(Range, "physical register must use the '$name' MIR spelling"); + const PhysicalRegisterDesc *Description = Target.registers().physicalRegister(Text.substr(1)); + if (Description == nullptr) + return fail(Range, "unknown physical register '" + std::string(Text) + "'"); + Reg = Description->Reg; + return false; +} + +bool MIRDocumentParser::physicalRegister(const yaml::Node &Node, PhysicalRegister &Reg) { + std::string_view Text; + return scalar(Node, Text, "physical register") || physicalRegister(Text, Node.range(), Reg); +} + +bool MIRDocumentParser::virtualRegister(std::string_view Text, const TextRange &Range, + MachineFunction &Function, MIParserContext &Context, + Register &Reg) { + const auto ID = parsePrefixedID(Text, "%"); + if (!ID) + return fail(Range, "virtual register must use the '%N' MIR spelling"); + auto Found = + std::ranges::find_if(Context.VirtualRegisters, + [ID](const MIVirtualRegisterSlot &Slot) { return Slot.TextID == *ID; }); + if (Found == Context.VirtualRegisters.end()) { + auto Created = Function.registerInfo().createVirtualRegister(); + if (!Created) + return importError(Range, Created.error()); + Context.VirtualRegisters.push_back({*ID, *Created}); + Reg = *Created; + return false; + } + Reg = Found->Reg; + return false; +} + +bool MIRDocumentParser::parseConstants(const yaml::Node &Root, MachineFunction &Function, + MIParserContext &Context) { + const yaml::Node *Constants = mappingValue(Root, "constants"); + if (Constants == nullptr) + return false; + if (requireKind(*Constants, yaml::NodeKind::Sequence, "constant pool")) + return true; + constexpr std::array Keys{ + std::string_view{"id"}, + std::string_view{"value"}, + std::string_view{"alignment"}, + std::string_view{"isTargetSpecific"}, + }; + std::vector<LLVMNamedTypeDefinition> NamedTypes; + const std::string_view DataLayout = effectiveDataLayout(); + if (ActiveDocument->EmbeddedIR) { + const auto Definitions = ActiveDocument->EmbeddedIR->typeDefinitions(); + NamedTypes.reserve(Definitions.size()); + for (const EmbeddedIRTypeDefinition &Definition : Definitions) { + NamedTypes.push_back({.Name = Definition.Name, + .Definition = Definition.Definition, + .NumericName = Definition.NumericName}); + } + } + std::unordered_set<std::uint32_t> IDs; + for (const yaml::NodeID ConstantID : Constants->sequence()) { + const yaml::Node *Constant = node(ConstantID); + if (Constant == nullptr || + requireKind(*Constant, yaml::NodeKind::Mapping, "constant pool entry") || + validateKeys(*Constant, Keys, "constant pool entry")) { + return true; + } + const yaml::Node *IDNode = mappingValue(*Constant, "id"); + const yaml::Node *ValueNode = mappingValue(*Constant, "value"); + if (IDNode == nullptr || ValueNode == nullptr) + return fail(Constant->range(), "constant pool entry requires 'id' and 'value'"); + std::uint64_t WideID = 0; + if (unsignedValue(*IDNode, WideID, "a 32-bit constant pool ID") || + WideID > std::numeric_limits<std::uint32_t>::max()) { + return true; + } + const auto TextID = static_cast<std::uint32_t>(WideID); + if (!IDs.insert(TextID).second) + return fail(IDNode->range(), "duplicate constant pool ID"); + + bool IsTargetSpecific = false; + if (const yaml::Node *TargetNode = mappingValue(*Constant, "isTargetSpecific")) { + if (boolean(*TargetNode, IsTargetSpecific)) + return true; + } + if (IsTargetSpecific) { + std::string_view ValueText; + if (scalar(*ValueNode, ValueText, "target constant")) + return true; + auto TargetConstantValue = Target.mirText().parseTargetConstant(ValueText); + if (!TargetConstantValue) + return importError(ValueNode->range(), TargetConstantValue.error(), "mir-target-constant"); + auto Emission = TargetConstantValue->emit(); + if (!Emission) + return importError(ValueNode->range(), Emission.error(), "mir-target-constant"); + auto RequiredAlignment = TargetConstantValue->requiredAlignment(); + if (!RequiredAlignment) + return importError(ValueNode->range(), RequiredAlignment.error(), "mir-target-constant"); + + MachineConstantPoolEntryDesc Description; + Description.Size = Emission->Bytes.size(); + Description.Align = *RequiredAlignment; + Description.Value = std::move(*TargetConstantValue); + if (const yaml::Node *AlignmentNode = mappingValue(*Constant, "alignment")) { + if (alignment(*AlignmentNode, Description.Align, true)) + return true; + } + auto Added = Function.constantPool().addEntry(std::move(Description)); + if (!Added) + return importError(Constant->range(), Added.error()); + Context.ConstantPoolItems.push_back({.TextID = TextID, .Index = *Added}); + continue; + } + + std::string_view ValueText; + if (scalar(*ValueNode, ValueText, "LLVM constant")) + return true; + const LLVMConstantSyntaxOptions ConstantOptions{ + .DataLayout = DataLayout, + .NamedTypes = NamedTypes, + }; + auto Data = parseLLVMConstant(trim(ValueText), ConstantOptions); + if (!Data) + return importError(ValueNode->range(), Data.error(), "mir-llvm-constant"); + auto Size = llvmConstantStorageSize(Data->Type, ConstantOptions); + if (!Size) + return importError(ValueNode->range(), Size.error(), "mir-llvm-constant"); + + MachineConstantPoolEntryDesc Description; + Description.Size = *Size; + Description.Value = std::move(*Data); + if (const yaml::Node *AlignmentNode = mappingValue(*Constant, "alignment")) { + if (alignment(*AlignmentNode, Description.Align, true)) + return true; + } else { + auto Preferred = llvmConstantPreferredAlignment( + std::get<GenericConstantData>(Description.Value).Type, ConstantOptions); + if (!Preferred) + return importError(ValueNode->range(), Preferred.error(), "mir-llvm-constant"); + Description.Align = *Preferred; + } + auto Added = Function.constantPool().addEntry(std::move(Description)); + if (!Added) + return importError(Constant->range(), Added.error()); + Context.ConstantPoolItems.push_back({.TextID = TextID, .Index = *Added}); + } + return false; +} + +bool MIRDocumentParser::reserveJumpTables(const yaml::Node &Root, MachineFunction &Function, + MIParserContext &Context, + std::vector<PendingJumpTable> &Pending) { + const yaml::Node *JumpTable = mappingValue(Root, "jumpTable"); + if (JumpTable == nullptr) + return false; + if (requireKind(*JumpTable, yaml::NodeKind::Mapping, "jump table")) + return true; + constexpr std::array JumpTableKeys{ + std::string_view{"kind"}, + std::string_view{"entries"}, + }; + if (validateKeys(*JumpTable, JumpTableKeys, "jump table")) + return true; + const yaml::Node *KindNode = mappingValue(*JumpTable, "kind"); + if (KindNode == nullptr) + return fail(JumpTable->range(), "jump table requires 'kind'"); + std::string_view KindName; + if (scalar(*KindNode, KindName, "jump table kind")) + return true; + + JumpTableEntryEncoding Encoding; + std::uint8_t EntrySize = 8; + if (KindName == "block-address") + Encoding = JumpTableEntryEncoding::BlockAddress; + else if (KindName == "gp-rel64-block-address") + Encoding = JumpTableEntryEncoding::GPRelative64BlockAddress; + else if (KindName == "gp-rel32-block-address") { + Encoding = JumpTableEntryEncoding::GPRelative32BlockAddress; + EntrySize = 4; + } else if (KindName == "label-difference32") { + Encoding = JumpTableEntryEncoding::LabelDifference32; + EntrySize = 4; + } else if (KindName == "label-difference64") + Encoding = JumpTableEntryEncoding::LabelDifference64; + else if (KindName == "inline") { + Encoding = JumpTableEntryEncoding::Inline; + EntrySize = 4; + } else if (KindName == "custom32") { + Encoding = JumpTableEntryEncoding::Custom32; + EntrySize = 4; + } else { + return fail(KindNode->range(), "unknown enumerated scalar"); + } + + const yaml::Node *Entries = mappingValue(*JumpTable, "entries"); + if (Entries == nullptr) + return false; + if (requireKind(*Entries, yaml::NodeKind::Sequence, "jump table entries")) + return true; + constexpr std::array EntryKeys{ + std::string_view{"id"}, + std::string_view{"blocks"}, + }; + std::unordered_set<std::uint32_t> IDs; + for (const yaml::NodeID EntryID : Entries->sequence()) { + const yaml::Node *Entry = node(EntryID); + if (Entry == nullptr || requireKind(*Entry, yaml::NodeKind::Mapping, "jump table entry") || + validateKeys(*Entry, EntryKeys, "jump table entry")) { + return true; + } + const yaml::Node *IDNode = mappingValue(*Entry, "id"); + if (IDNode == nullptr) + return fail(Entry->range(), "jump table entry requires 'id'"); + std::uint64_t WideID = 0; + if (unsignedValue(*IDNode, WideID, "a 32-bit jump table ID") || + WideID > std::numeric_limits<std::uint32_t>::max()) { + return true; + } + const auto TextID = static_cast<std::uint32_t>(WideID); + if (!IDs.insert(TextID).second) + return fail(IDNode->range(), "duplicate jump table ID"); + auto Reserved = Function.jumpTableInfo().reserveTable(); + if (!Reserved) + return importError(Entry->range(), Reserved.error()); + PendingJumpTable NewPending{ + .Index = *Reserved, + .TextID = TextID, + .Encoding = Encoding, + .EntrySize = EntrySize, + .Blocks = {}, + }; + if (const yaml::Node *Blocks = mappingValue(*Entry, "blocks")) { + if (requireKind(*Blocks, yaml::NodeKind::Sequence, "jump table blocks")) + return true; + for (const yaml::NodeID BlockID : Blocks->sequence()) { + const yaml::Node *BlockNode = node(BlockID); + std::string_view Text; + if (BlockNode == nullptr || scalar(*BlockNode, Text, "basic block reference")) + return true; + NewPending.Blocks.emplace_back(Text); + } + } + if (NewPending.Blocks.empty()) + return fail(Entry->range(), "jump table entry must contain at least one block"); + Context.JumpTables.push_back({.TextID = TextID, .Index = *Reserved}); + Pending.push_back(std::move(NewPending)); + } + return false; +} + +bool MIRDocumentParser::defineJumpTables(MachineFunction &Function, + std::span<const PendingJumpTable> Pending) { + for (const PendingJumpTable &Table : Pending) { + MachineJumpTableDesc Description; + Description.Encoding = Table.Encoding; + Description.EntrySize = Table.EntrySize; + Description.Align = *Alignment::create(Table.EntrySize); + for (const std::string &Text : Table.Blocks) { + MachineBasicBlock *Block = nullptr; + if (parseBlockReference(Text, Result.Source.range(0, 0), Function, Block)) + return true; + if (Block == nullptr) + return fail(Result.Source.range(0, 0), "reference to undefined machine basic block"); + Description.Targets.push_back(Block); + } + auto DefineResult = + Function.jumpTableInfo().defineReservedTable(Table.Index, std::move(Description)); + if (!DefineResult) + return importError(Result.Source.range(0, 0), DefineResult.error()); + } + return false; +} + +} // namespace rcc::mir + +namespace rcc::mir { +namespace { + +[[nodiscard]] bool parseIndex(std::string_view Text, std::uint64_t &Value) noexcept { + if (Text.empty()) + return false; + const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, 10); + return Error == std::errc{} && End == Text.data() + Text.size(); +} + +[[nodiscard]] const auto &namedOpcodes() { + static constexpr auto Opcodes = std::to_array<std::pair<std::string_view, std::uint64_t>>({ + {"DW_OP_addr", 0x03}, + {"DW_OP_deref", 0x06}, + {"DW_OP_const1u", 0x08}, + {"DW_OP_const1s", 0x09}, + {"DW_OP_const2u", 0x0A}, + {"DW_OP_const2s", 0x0B}, + {"DW_OP_const4u", 0x0C}, + {"DW_OP_const4s", 0x0D}, + {"DW_OP_const8u", 0x0E}, + {"DW_OP_const8s", 0x0F}, + {"DW_OP_constu", 0x10}, + {"DW_OP_consts", 0x11}, + {"DW_OP_dup", 0x12}, + {"DW_OP_drop", 0x13}, + {"DW_OP_over", 0x14}, + {"DW_OP_pick", 0x15}, + {"DW_OP_swap", 0x16}, + {"DW_OP_rot", 0x17}, + {"DW_OP_xderef", 0x18}, + {"DW_OP_abs", 0x19}, + {"DW_OP_and", 0x1A}, + {"DW_OP_div", 0x1B}, + {"DW_OP_minus", 0x1C}, + {"DW_OP_mod", 0x1D}, + {"DW_OP_mul", 0x1E}, + {"DW_OP_neg", 0x1F}, + {"DW_OP_not", 0x20}, + {"DW_OP_or", 0x21}, + {"DW_OP_plus", 0x22}, + {"DW_OP_plus_uconst", 0x23}, + {"DW_OP_shl", 0x24}, + {"DW_OP_shr", 0x25}, + {"DW_OP_shra", 0x26}, + {"DW_OP_xor", 0x27}, + {"DW_OP_bra", 0x28}, + {"DW_OP_eq", 0x29}, + {"DW_OP_ge", 0x2A}, + {"DW_OP_gt", 0x2B}, + {"DW_OP_le", 0x2C}, + {"DW_OP_lt", 0x2D}, + {"DW_OP_ne", 0x2E}, + {"DW_OP_skip", 0x2F}, + {"DW_OP_regx", 0x90}, + {"DW_OP_fbreg", 0x91}, + {"DW_OP_bregx", 0x92}, + {"DW_OP_piece", 0x93}, + {"DW_OP_deref_size", 0x94}, + {"DW_OP_xderef_size", 0x95}, + {"DW_OP_nop", 0x96}, + {"DW_OP_push_object_address", 0x97}, + {"DW_OP_call2", 0x98}, + {"DW_OP_call4", 0x99}, + {"DW_OP_call_ref", 0x9A}, + {"DW_OP_form_tls_address", 0x9B}, + {"DW_OP_call_frame_cfa", 0x9C}, + {"DW_OP_bit_piece", 0x9D}, + {"DW_OP_implicit_value", 0x9E}, + {"DW_OP_stack_value", 0x9F}, + {"DW_OP_implicit_pointer", 0xA0}, + {"DW_OP_addrx", 0xA1}, + {"DW_OP_constx", 0xA2}, + {"DW_OP_entry_value", 0xA3}, + {"DW_OP_const_type", 0xA4}, + {"DW_OP_regval_type", 0xA5}, + {"DW_OP_deref_type", 0xA6}, + {"DW_OP_xderef_type", 0xA7}, + {"DW_OP_convert", 0xA8}, + {"DW_OP_reinterpret", 0xA9}, + {"DW_OP_LLVM_fragment", 0x1000}, + {"DW_OP_LLVM_convert", 0x1001}, + {"DW_OP_LLVM_tag_offset", 0x1002}, + {"DW_OP_LLVM_entry_value", 0x1003}, + {"DW_OP_LLVM_implicit_pointer", 0x1004}, + {"DW_OP_LLVM_arg", 0x1005}, + {"DW_OP_LLVM_extract_bits_sext", 0x1006}, + {"DW_OP_LLVM_extract_bits_zext", 0x1007}, + }); + return Opcodes; +} + +[[nodiscard]] const auto &attributeEncodings() { + static constexpr auto Encodings = std::to_array<std::pair<std::string_view, std::uint64_t>>({ + {"DW_ATE_address", 0x01}, + {"DW_ATE_boolean", 0x02}, + {"DW_ATE_complex_float", 0x03}, + {"DW_ATE_float", 0x04}, + {"DW_ATE_signed", 0x05}, + {"DW_ATE_signed_char", 0x06}, + {"DW_ATE_unsigned", 0x07}, + {"DW_ATE_unsigned_char", 0x08}, + {"DW_ATE_imaginary_float", 0x09}, + {"DW_ATE_packed_decimal", 0x0A}, + {"DW_ATE_numeric_string", 0x0B}, + {"DW_ATE_edited", 0x0C}, + {"DW_ATE_signed_fixed", 0x0D}, + {"DW_ATE_unsigned_fixed", 0x0E}, + {"DW_ATE_decimal_float", 0x0F}, + {"DW_ATE_UTF", 0x10}, + {"DW_ATE_UCS", 0x11}, + {"DW_ATE_ASCII", 0x12}, + }); + return Encodings; +} + +[[nodiscard]] std::optional<std::string> opcodeName(std::uint64_t Opcode) { + if (Opcode >= 0x30U && Opcode <= 0x4FU) + return "DW_OP_lit" + std::to_string(Opcode - 0x30U); + if (Opcode >= 0x50U && Opcode <= 0x6FU) + return "DW_OP_reg" + std::to_string(Opcode - 0x50U); + if (Opcode >= 0x70U && Opcode <= 0x8FU) + return "DW_OP_breg" + std::to_string(Opcode - 0x70U); + const auto &Opcodes = namedOpcodes(); + const auto Found = + std::ranges::find_if(Opcodes, [Opcode](const auto &Entry) { return Entry.second == Opcode; }); + if (Found == Opcodes.end()) + return std::nullopt; + return std::string(Found->first); +} + +[[nodiscard]] std::size_t opcodeSize(std::uint64_t Opcode) noexcept { + if (Opcode >= 0x70U && Opcode <= 0x8FU) + return 2; + switch (Opcode) { + case 0x92: + case 0x1000: + case 0x1001: + case 0x1006: + case 0x1007: + return 3; + case 0x10: + case 0x11: + case 0x23: + case 0x90: + case 0x94: + case 0x1002: + case 0x1003: + case 0x1005: + return 2; + default: + return 1; + } +} + +[[nodiscard]] std::optional<std::string_view> attributeName(std::uint64_t Encoding) noexcept { + const auto &Encodings = attributeEncodings(); + const auto Found = std::ranges::find_if( + Encodings, [Encoding](const auto &Entry) { return Entry.second == Encoding; }); + if (Found == Encodings.end()) + return std::nullopt; + return Found->first; +} + +} // namespace + +std::optional<std::uint64_t> parseDIExpressionOpcodeName(std::string_view Name) noexcept { + const auto NumberedOpcode = [Name](std::string_view Prefix, + std::uint64_t Base) -> std::optional<std::uint64_t> { + if (!Name.starts_with(Prefix)) + return std::nullopt; + std::uint64_t Index = 0; + if (!parseIndex(Name.substr(Prefix.size()), Index) || Index > 31) + return std::nullopt; + return Base + Index; + }; + if (auto Opcode = NumberedOpcode("DW_OP_lit", 0x30U)) + return Opcode; + if (auto Opcode = NumberedOpcode("DW_OP_reg", 0x50U)) + return Opcode; + if (auto Opcode = NumberedOpcode("DW_OP_breg", 0x70U)) + return Opcode; + + const auto &NamedOpcodes = namedOpcodes(); + const auto *const Found = + std::ranges::find_if(NamedOpcodes, [Name](const auto &Entry) { return Entry.first == Name; }); + if (Found == NamedOpcodes.end()) + return std::nullopt; + return Found->second; +} + +std::optional<std::uint64_t> +parseDIExpressionAttributeEncodingName(std::string_view Name) noexcept { + const auto &Encodings = attributeEncodings(); + const auto Found = + std::ranges::find_if(Encodings, [Name](const auto &Entry) { return Entry.first == Name; }); + if (Found == Encodings.end()) + return std::nullopt; + return Found->second; +} + +std::string invalidDIExpressionElementMessage(std::string_view Name) { + if (Name.starts_with("DW_OP_")) + return "invalid DWARF op '" + std::string(Name) + "'"; + if (Name.starts_with("DW_ATE_")) + return "invalid DWARF attribute encoding '" + std::string(Name) + "'"; + return "expected unsigned integer"; +} + +std::string formatDIExpression(std::span<const std::uint64_t> Elements) { + std::string Result = "!DIExpression("; + std::size_t Position = 0; + bool First = true; + while (Position < Elements.size()) { + const std::uint64_t Opcode = Elements[Position]; + const auto Name = opcodeName(Opcode); + const std::size_t Size = opcodeSize(Opcode); + if (!Name || Position + Size > Elements.size()) { + Result = "!DIExpression("; + for (std::size_t Index = 0; Index < Elements.size(); ++Index) { + if (Index != 0) + Result.append(", "); + Result.append(std::to_string(Elements[Index])); + } + Result.push_back(')'); + return Result; + } + + if (!First) + Result.append(", "); + First = false; + Result.append(*Name); + for (std::size_t Argument = 1; Argument < Size; ++Argument) { + Result.append(", "); + if (Opcode == 0x1001U && Argument == 2U) { + if (const auto Encoding = attributeName(Elements[Position + Argument])) { + Result.append(*Encoding); + continue; + } + } + Result.append(std::to_string(Elements[Position + Argument])); + } + Position += Size; + } + Result.push_back(')'); + return Result; +} + +} // namespace rcc::mir + +namespace rcc::mir { + +using parser_detail::parseSigned; +using parser_detail::parseUnsigned; +using parser_detail::parseUnsigned32; + +Expected<void> MetadataTextParser::expect(MITokenKind Kind, std::string_view Description) { + if (consume(Kind)) + return {}; + return makeUnexpected(ErrorCode::ParseError, + "expected " + std::string(Description) + " in MIR metadata scalar"); +} + +Expected<void> MetadataTextParser::finish() { + while (consume(MITokenKind::Newline)) { + } + if (Lexer.failed()) { + if (!Lexer.diagnostics().empty()) { + return makeUnexpected(ErrorCode::ParseError, + std::string(Lexer.diagnostics().front().Message)); + } + return makeUnexpected(ErrorCode::ParseError, "invalid MIR metadata scalar"); + } + if (!at(MITokenKind::EndOfFile)) + return makeUnexpected(ErrorCode::ParseError, "unexpected trailing MIR metadata tokens"); + return {}; +} + +Expected<MachineMetadataReference> MetadataTextParser::parseReference() { + if (auto Exclaim = expect(MITokenKind::Exclaim, "'!' before a metadata slot"); !Exclaim) + return std::unexpected(Exclaim.error()); + if (!at(MITokenKind::IntegerLiteral)) + return makeUnexpected(ErrorCode::ParseError, "expected a 32-bit metadata slot"); + std::uint32_t Slot = 0; + if (!parseUnsigned32(token().integerValue(), Slot)) + return makeUnexpected(ErrorCode::ParseError, "metadata slot is outside the 32-bit range"); + advance(); + return Resolve(Slot); +} + +Expected<MachineMetadataReference> MetadataTextParser::parseReferenceOnly() { + auto Reference = parseReference(); + if (!Reference) + return std::unexpected(Reference.error()); + if (auto Finished = finish(); !Finished) + return std::unexpected(Finished.error()); + return Reference; +} + +Expected<MachineDIExpression> MetadataTextParser::parseExpression() { + if (!at(MITokenKind::MetadataKeyword) || !token().isMetadataKeyword("!DIExpression")) { + return makeUnexpected(ErrorCode::ParseError, "expected an inline !DIExpression"); + } + advance(); + if (auto Left = expect(MITokenKind::LeftParen, "'(' after !DIExpression"); !Left) + return std::unexpected(Left.error()); + std::vector<std::uint64_t> Elements; + while (!at(MITokenKind::RightParen)) { + if (at(MITokenKind::Identifier)) { + const auto Opcode = parseDIExpressionOpcodeName(token().stringValue()); + const auto Encoding = parseDIExpressionAttributeEncodingName(token().stringValue()); + if (!Opcode && !Encoding) + return makeUnexpected(ErrorCode::ParseError, + invalidDIExpressionElementMessage(token().stringValue())); + Elements.push_back(Opcode ? *Opcode : *Encoding); + advance(); + } else if (at(MITokenKind::IntegerLiteral)) { + std::int64_t Signed = 0; + std::uint64_t Unsigned = 0; + if (parseSigned(token().integerValue(), Signed)) + Elements.push_back(static_cast<std::uint64_t>(Signed)); + else if (parseUnsigned(token().integerValue(), Unsigned)) + Elements.push_back(Unsigned); + else + return makeUnexpected(ErrorCode::ParseError, "DIExpression integer exceeds 64 bits"); + advance(); + } else { + return makeUnexpected(ErrorCode::ParseError, "expected a DIExpression opcode or integer"); + } + if (!consume(MITokenKind::Comma)) + break; + } + if (auto Right = expect(MITokenKind::RightParen, "')' after !DIExpression"); !Right) + return std::unexpected(Right.error()); + return MachineDIExpression(std::move(Elements)); +} + +Expected<MachineMetadataValue> MetadataTextParser::parseValue() { + MachineMetadataValue Value; + if (at(MITokenKind::Exclaim)) { + auto Reference = parseReference(); + if (!Reference) + return std::unexpected(Reference.error()); + auto Created = MachineMetadataValue::reference(*Reference); + if (!Created) + return std::unexpected(Created.error()); + Value = std::move(*Created); + } else { + auto Expression = parseExpression(); + if (!Expression) + return std::unexpected(Expression.error()); + Value = MachineMetadataValue::expression(std::move(*Expression)); + } + if (auto Finished = finish(); !Finished) + return std::unexpected(Finished.error()); + return Value; +} + +Expected<void> +MetadataTextParser::parseDILocation(std::vector<MachineDILocationFrame> &Frames, + std::optional<MachineMetadataReference> &Terminal) { + if (!at(MITokenKind::MetadataKeyword) || !token().isMetadataKeyword("!DILocation")) { + return makeUnexpected(ErrorCode::ParseError, "expected an inline !DILocation"); + } + advance(); + if (auto Left = expect(MITokenKind::LeftParen, "'(' after !DILocation"); !Left) + return Left; + + MachineDILocationFrame Frame; + bool HasLine = false; + bool HasScope = false; + std::vector<MachineDILocationFrame> NestedFrames; + std::optional<MachineMetadataReference> NestedTerminal; + std::unordered_set<std::string> Seen; + while (!at(MITokenKind::RightParen)) { + if (!at(MITokenKind::Identifier)) + return makeUnexpected(ErrorCode::ParseError, "expected a DILocation field name"); + const std::string Field(token().stringValue()); + if (!Seen.insert(Field).second) { + return makeUnexpected(ErrorCode::ParseError, "duplicate DILocation field '" + Field + "'"); + } + advance(); + if (auto Colon = expect(MITokenKind::Colon, "':' after a DILocation field"); !Colon) + return Colon; + + if (Field == "line") { + std::uint32_t Value = 0; + if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned32(token().integerValue(), Value)) { + return makeUnexpected(ErrorCode::ParseError, + "DILocation line must be an unsigned 32-bit integer"); + } + Frame.Line = Value; + HasLine = true; + advance(); + } else if (Field == "column") { + std::uint64_t Value = 0; + if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned(token().integerValue(), Value) || + Value > std::numeric_limits<std::uint16_t>::max()) { + return makeUnexpected(ErrorCode::ParseError, + "DILocation column must be an unsigned 16-bit integer"); + } + Frame.Column = static_cast<std::uint16_t>(Value); + advance(); + } else if (Field == "scope") { + auto Scope = parseReference(); + if (!Scope) + return std::unexpected(Scope.error()); + if (Scope->scope() != MachineMetadataScope::EmbeddedIR) { + return makeUnexpected(ErrorCode::ParseError, + "DILocation scope must reference embedded LLVM IR metadata"); + } + Frame.Scope = *Scope; + HasScope = true; + } else if (Field == "inlinedAt") { + if (at(MITokenKind::MetadataKeyword) && token().isMetadataKeyword("!DILocation")) { + if (auto Nested = parseDILocation(NestedFrames, NestedTerminal); !Nested) + return Nested; + } else { + auto Reference = parseReference(); + if (!Reference) + return std::unexpected(Reference.error()); + if (Reference->scope() != MachineMetadataScope::EmbeddedIR) { + return makeUnexpected(ErrorCode::ParseError, + "DILocation inlinedAt must reference embedded LLVM IR metadata"); + } + NestedTerminal = *Reference; + } + } else if (Field == "isImplicitCode") { + if (!at(MITokenKind::Identifier) || + (token().stringValue() != "true" && token().stringValue() != "false")) { + return makeUnexpected(ErrorCode::ParseError, + "DILocation isImplicitCode must be true or false"); + } + Frame.IsImplicitCode = token().stringValue() == "true"; + advance(); + } else if (Field == "atomGroup") { + std::uint64_t Value = 0; + if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned(token().integerValue(), Value) || + Value > MachineDILocation::MaxAtomGroup) { + return makeUnexpected(ErrorCode::ParseError, "DILocation atomGroup exceeds 61 bits"); + } + Frame.AtomGroup = Value; + advance(); + } else if (Field == "atomRank") { + std::uint64_t Value = 0; + if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned(token().integerValue(), Value) || + Value > 7) { + return makeUnexpected(ErrorCode::ParseError, "DILocation atomRank exceeds three bits"); + } + Frame.AtomRank = static_cast<std::uint8_t>(Value); + advance(); + } else { + return makeUnexpected(ErrorCode::ParseError, "unknown DILocation field '" + Field + "'"); + } + if (!consume(MITokenKind::Comma)) + break; + } + if (auto Right = expect(MITokenKind::RightParen, "')' after !DILocation"); !Right) + return Right; + if (!HasLine || !HasScope) + return makeUnexpected(ErrorCode::ParseError, "DILocation requires line and scope fields"); + Frames.push_back(Frame); + Frames.insert(Frames.end(), NestedFrames.begin(), NestedFrames.end()); + Terminal = NestedTerminal; + return {}; +} + +Expected<MachineDebugLocation> MetadataTextParser::parseLocation() { + MachineDebugLocation Location; + if (at(MITokenKind::Exclaim)) { + auto Reference = parseReference(); + if (!Reference) + return std::unexpected(Reference.error()); + auto Created = MachineDebugLocation::reference(*Reference); + if (!Created) + return std::unexpected(Created.error()); + Location = std::move(*Created); + } else { + std::vector<MachineDILocationFrame> Frames; + std::optional<MachineMetadataReference> Terminal; + if (auto Parsed = parseDILocation(Frames, Terminal); !Parsed) + return std::unexpected(Parsed.error()); + auto Created = MachineDILocation::create(std::move(Frames), Terminal); + if (!Created) + return std::unexpected(Created.error()); + Location = MachineDebugLocation::inlineLocation(std::move(*Created)); + } + if (auto Finished = finish(); !Finished) + return std::unexpected(Finished.error()); + return Location; +} + +} // namespace rcc::mir + +namespace rcc::mir { + +Expected<void> verifyMIRDocumentReferences(const MIRDocument &Document, + const MachineFunction &Function, bool UseEmbeddedIR) { + const EmbeddedIRReferenceIndex *Embedded = UseEmbeddedIR ? Document.embeddedIR() : nullptr; + const EmbeddedIRFunction *IRFunction = nullptr; + if (Embedded != nullptr) { + if (Function.name().empty()) { + std::size_t AnonymousOrdinal = 0; + bool FoundFunction = false; + for (std::size_t Index = 0; Index < Document.unit().functionCount(); ++Index) { + const MachineFunction *Candidate = Document.unit().functionAt(Index); + if (Candidate == &Function) { + FoundFunction = true; + break; + } + if (Candidate != nullptr && Candidate->name().empty()) + ++AnonymousOrdinal; + } + if (FoundFunction) + IRFunction = Embedded->unnamedFunction(AnonymousOrdinal); + } else { + IRFunction = Embedded->function(Function.name()); + } + if (IRFunction == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + "machine function has no matching embedded LLVM IR function"); + } + } + + const auto ValidateGlobal = [Embedded](std::string_view Name, + std::string_view Description) -> Expected<void> { + if (Embedded == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " requires embedded LLVM IR to resolve global '" + + std::string(Name) + "'"); + } + if (Embedded->hasGlobalValue(Name)) + return {}; + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " references an undefined embedded LLVM IR global '" + + std::string(Name) + "'"); + }; + const auto ValidateBlockAddress = [Embedded](std::string_view FunctionName, + std::string_view BlockName, + std::string_view Description) -> Expected<void> { + if (Embedded == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " requires embedded LLVM IR to resolve a block address"); + } + const EmbeddedIRFunction *AddressFunction = Embedded->function(FunctionName); + if (AddressFunction == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " references an undefined embedded LLVM IR function '" + + std::string(FunctionName) + "'"); + } + if (AddressFunction->block(BlockName) == nullptr) { + return makeUnexpected( + ErrorCode::VerificationError, + std::string(Description) + " references an undefined embedded LLVM IR block '" + + std::string(BlockName) + "' in function '" + std::string(FunctionName) + "'"); + } + return {}; + }; + const auto ValidateIRValue = [IRFunction](std::string_view Name, + std::string_view Description) -> Expected<void> { + if (IRFunction == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " requires embedded LLVM IR to resolve IR value '%" + + std::string(Name) + "'"); + } + if (IRFunction->value(Name) != nullptr) + return {}; + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " references an undefined embedded LLVM IR value '%" + + std::string(Name) + "'"); + }; + const auto ValidateMetadata = [Embedded](MachineMetadataReference Reference, + std::initializer_list<std::string_view> ExpectedTypes, + std::string_view Description) -> Expected<void> { + if (!Reference.isValid()) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + " has an invalid metadata reference"); + } + if (Reference.scope() != MachineMetadataScope::EmbeddedIR) + return {}; + if (Embedded == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + " requires embedded LLVM IR metadata !" + + std::to_string(Reference.slot())); + } + const EmbeddedIRMetadataDefinition *Definition = + Embedded->metadata(std::to_string(Reference.slot())); + if (Definition == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + + " references undefined embedded LLVM IR metadata !" + + std::to_string(Reference.slot())); + } + if (ExpectedTypes.size() == 0 || + std::ranges::find(ExpectedTypes, Definition->DynamicType) != ExpectedTypes.end()) { + return {}; + } + std::string Expected; + for (const std::string_view Type : ExpectedTypes) { + if (!Expected.empty()) + Expected.append(" or "); + Expected.append(Type); + } + return makeUnexpected(ErrorCode::VerificationError, + std::string(Description) + " requires " + Expected + " metadata, but !" + + std::to_string(Reference.slot()) + " is " + + (Definition->DynamicType.empty() + ? std::string("an unclassified metadata node") + : Definition->DynamicType)); + }; + const auto ValidateMetadataValue = + [&ValidateMetadata](const MachineMetadataValue &Value, + std::string_view Description) -> Expected<void> { + const MachineMetadataReference *Reference = Value.reference(); + return Reference == nullptr ? Expected<void>{} : ValidateMetadata(*Reference, {}, Description); + }; + const auto ValidateLocation = + [&ValidateMetadata](const MachineDebugLocation &Location, + std::string_view Description) -> Expected<void> { + if (const MachineMetadataReference *Reference = Location.reference()) + return ValidateMetadata(*Reference, {"DILocation"}, Description); + const MachineDILocation *Inline = Location.inlineLocation(); + if (Inline == nullptr) + return {}; + for (const MachineDILocationFrame &Frame : Inline->inlineChain()) { + auto Verification = + ValidateMetadata(Frame.Scope, {"DISubprogram", "DILexicalBlock", "DILexicalBlockFile"}, + "inline DILocation scope"); + if (!Verification) + return Verification; + } + if (Inline->terminalInlinedAt()) { + return ValidateMetadata(*Inline->terminalInlinedAt(), {"DILocation"}, + "inline DILocation inlinedAt"); + } + return {}; + }; + const auto ValidateVariableDebugInfo = + [&ValidateMetadata, &ValidateMetadataValue, + &ValidateLocation](const MachineVariableDebugInfo &DebugInfo) -> Expected<void> { + if (DebugInfo.Variable) { + auto Verification = + ValidateMetadata(*DebugInfo.Variable, {"DILocalVariable"}, "debug variable"); + if (!Verification) + return Verification; + } + if (DebugInfo.Expression) { + if (const MachineMetadataReference *Reference = DebugInfo.Expression->reference(); + Reference != nullptr) { + auto Verification = ValidateMetadata(*Reference, {"DIExpression"}, "debug expression"); + if (!Verification) + return Verification; + } + auto Verification = ValidateMetadataValue(*DebugInfo.Expression, "debug expression"); + if (!Verification) + return Verification; + } + if (DebugInfo.Location) + return ValidateLocation(*DebugInfo.Location, "debug location"); + return {}; + }; + const auto ValidateGenericType = [Embedded](const auto &Self, + const GenericConstantType &Type) -> Expected<void> { + if (Type.Kind == GenericConstantTypeKind::Named) { + if (Embedded == nullptr) { + return makeUnexpected( + ErrorCode::VerificationError, + "generic constant type requires embedded LLVM IR to resolve named type '%" + Type.Name + + "'"); + } + if (Embedded->typeDefinition(Type.Name) == nullptr) { + return makeUnexpected( + ErrorCode::VerificationError, + "generic constant type references an undefined embedded LLVM IR named type '%" + + Type.Name + "'"); + } + } + for (const GenericConstantType &Element : Type.Elements) { + auto Verification = Self(Self, Element); + if (!Verification) + return Verification; + } + return {}; + }; + const auto ValidateGenericConstant = + [&ValidateBlockAddress, &ValidateGenericType, + &ValidateGlobal](const auto &Self, const GenericConstantData &Constant) -> Expected<void> { + auto TypeVerification = ValidateGenericType(ValidateGenericType, Constant.Type); + if (!TypeVerification) + return TypeVerification; + if (Constant.SourceElementType) { + TypeVerification = ValidateGenericType(ValidateGenericType, *Constant.SourceElementType); + if (!TypeVerification) + return TypeVerification; + } + switch (Constant.Kind) { + case GenericConstantKind::GlobalAddress: + case GenericConstantKind::DSOLocalEquivalent: + case GenericConstantKind::NoCFI: + if (auto Verification = ValidateGlobal(Constant.Symbol, "generic constant"); !Verification) + return Verification; + break; + case GenericConstantKind::BlockAddress: + if (auto Verification = ValidateBlockAddress(Constant.Symbol, Constant.Block, + "generic constant blockaddress"); + !Verification) { + return Verification; + } + break; + default: + break; + } + for (const GenericConstantData &Element : Constant.Elements) { + auto Verification = Self(Self, Element); + if (!Verification) + return Verification; + } + return {}; + }; + + for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; + Block = Block->next()) { + if (const auto Name = Block->attributes().irBlockName(); Name) { + if (IRFunction == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + "machine basic block requires embedded LLVM IR to resolve block '" + + std::string(*Name) + "'"); + } + if (IRFunction->block(*Name) == nullptr) { + return makeUnexpected( + ErrorCode::VerificationError, + "machine basic block references an undefined embedded LLVM IR block '" + + std::string(*Name) + "'"); + } + } + } + for (const StackObject &Object : Function.frameInfo().objects()) { + const StackObjectDesc *Description = Object.description(); + if (Description == nullptr || !Description->AllocaName) + continue; + if (IRFunction == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + "stack object requires embedded LLVM IR to resolve alloca '" + + *Description->AllocaName + "'"); + } + if (IRFunction->alloca(*Description->AllocaName) == nullptr) { + return makeUnexpected(ErrorCode::VerificationError, + "stack object references an undefined embedded LLVM IR alloca '" + + *Description->AllocaName + "'"); + } + } + + for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { + const MachineConstantPoolEntryDesc *Description = Entry.description(); + if (Description == nullptr) + continue; + if (const auto *Generic = std::get_if<GenericConstantData>(&Description->Value)) { + auto Verification = ValidateGenericConstant(ValidateGenericConstant, *Generic); + if (!Verification) + return Verification; + } else if (const auto *Address = std::get_if<BlockAddressConstant>(&Description->Value)) { + auto Verification = ValidateBlockAddress(Address->FunctionName, Address->BlockName, + "constant-pool block address"); + if (!Verification) + return Verification; + } + } + + for (const MachineMetadataNode &Node : Function.metadata().nodes()) { + for (const MachineMetadataElement &Element : Node.Elements) { + const MachineMetadataReference *Reference = Element.reference(); + if (Reference == nullptr) + continue; + auto Verification = ValidateMetadata(*Reference, {}, "machine metadata node"); + if (!Verification) + return Verification; + } + } + + for (const StackObject &Object : Function.frameInfo().objects()) { + const StackObjectDesc *Description = Object.description(); + if (Description == nullptr || !Description->DebugInfo) + continue; + auto Verification = ValidateVariableDebugInfo(*Description->DebugInfo); + if (!Verification) + return Verification; + } + for (const MachineEntryValueInfo &Entry : Function.records().entryValues()) { + auto Verification = ValidateVariableDebugInfo(Entry.DebugInfo); + if (!Verification) + return Verification; + } + for (const MachineCalledGlobal &Called : Function.records().calledGlobals()) { + auto Verification = ValidateGlobal(Called.Callee, "called-global record"); + if (!Verification) + return Verification; + } + + for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; + Block = Block->next()) { + for (const MachineInstr *Instruction = Block->firstInstruction(); Instruction != nullptr; + Instruction = Instruction->next()) { + if (Instruction->debugLocation()) { + auto Verification = + ValidateLocation(*Instruction->debugLocation(), "instruction debug location"); + if (!Verification) + return Verification; + } + const std::array InstructionMetadata{ + Instruction->heapAllocMarker(), + Instruction->pcSections(), + Instruction->mmra(), + }; + for (const auto &Reference : InstructionMetadata) { + if (!Reference) + continue; + auto Verification = ValidateMetadata(*Reference, {}, "instruction metadata attachment"); + if (!Verification) + return Verification; + } + + for (const MachineOperand &Operand : Instruction->operands()) { + if (const auto *Global = Operand.getIf<GlobalSymbolOperand>(); Global != nullptr) { + auto Verification = ValidateGlobal(Global->Name, "global-symbol operand"); + if (!Verification) + return Verification; + } + if (const auto *Address = Operand.getIf<BlockAddressOperand>(); Address != nullptr) { + auto Verification = ValidateBlockAddress(Address->FunctionName, Address->BlockName, + "block-address operand"); + if (!Verification) + return Verification; + } + if (const auto *Metadata = Operand.getIf<MetadataOperand>(); Metadata != nullptr) { + auto Verification = ValidateMetadataValue(Metadata->Value, "metadata operand"); + if (!Verification) + return Verification; + } + } + + for (const MachineMemOperand *Memory : Instruction->memOperands()) { + const MachineMemOperandDesc &Description = Memory->description(); + const std::optional<std::string_view> BaseName = Description.Base.symbolName(); + switch (Description.Base.kind()) { + case MemoryBase::Kind::GlobalSymbol: + case MemoryBase::Kind::CallEntryGlobal: + if (BaseName) { + auto Verification = ValidateGlobal(*BaseName, "machine memory base"); + if (!Verification) + return Verification; + } + break; + case MemoryBase::Kind::IRValue: + if (BaseName) { + auto Verification = ValidateIRValue(*BaseName, "machine memory base"); + if (!Verification) + return Verification; + } + break; + default: + break; + } + const std::array MetadataReferences{ + Description.TBAA, Description.AliasScope, + Description.NoAlias, Description.NoAliasAddressSpace, + Description.RangeMetadata, + }; + for (const auto &Reference : MetadataReferences) { + if (!Reference) + continue; + auto Verification = ValidateMetadata(*Reference, {}, "machine memory operand metadata"); + if (!Verification) + return Verification; + } + if (Description.SyncScope.value() == 0) + continue; + const auto Scope = + std::ranges::find_if(Document.syncScopes(), [&Description](const auto &Slot) { + return Slot.ID == Description.SyncScope; + }); + if (Scope == Document.syncScopes().end()) { + return makeUnexpected( + ErrorCode::VerificationError, + "machine memory operand synchronization scope has no document name"); + } + } + } + } + return {}; +} + +} // namespace rcc::mir + +namespace rcc::mir { + +using parser_detail::parseBool; +using parser_detail::parsePrefixedID; +using parser_detail::parseSigned; +using parser_detail::parseUnsigned; +using parser_detail::parseUnsigned32; +using parser_detail::trim; + +bool MIRDocumentParser::parseStackObjects(const yaml::Node &Root, MachineFunction &Function, + MIParserContext &Context) { + auto ParseList = [this, &Function, &Context](const yaml::Node &List, bool Fixed) -> bool { + if (requireKind(List, yaml::NodeKind::Sequence, + Fixed ? "fixed stack objects" : "stack objects")) { + return true; + } + constexpr std::array FixedKeys{ + std::string_view{"id"}, + std::string_view{"type"}, + std::string_view{"offset"}, + std::string_view{"size"}, + std::string_view{"alignment"}, + std::string_view{"stack-id"}, + std::string_view{"isImmutable"}, + std::string_view{"isAliased"}, + std::string_view{"callee-saved-register"}, + std::string_view{"callee-saved-restored"}, + std::string_view{"debug-info-variable"}, + std::string_view{"debug-info-expression"}, + std::string_view{"debug-info-location"}, + }; + constexpr std::array StackKeys{ + std::string_view{"id"}, + std::string_view{"name"}, + std::string_view{"type"}, + std::string_view{"offset"}, + std::string_view{"size"}, + std::string_view{"alignment"}, + std::string_view{"stack-id"}, + std::string_view{"callee-saved-register"}, + std::string_view{"callee-saved-restored"}, + std::string_view{"local-offset"}, + std::string_view{"debug-info-variable"}, + std::string_view{"debug-info-expression"}, + std::string_view{"debug-info-location"}, + }; + std::unordered_set<std::uint32_t> IDs; + for (const yaml::NodeID ElementID : List.sequence()) { + const yaml::Node *Element = node(ElementID); + if (Element == nullptr || requireKind(*Element, yaml::NodeKind::Mapping, "stack object") || + (Fixed ? validateKeys(*Element, FixedKeys, "fixed stack object") + : validateKeys(*Element, StackKeys, "stack object"))) { + return true; + } + const yaml::Node *IDNode = mappingValue(*Element, "id"); + if (IDNode == nullptr) + return fail(Element->range(), "stack object requires 'id'"); + std::uint64_t WideID = 0; + if (unsignedValue(*IDNode, WideID, "a 32-bit stack object ID") || + WideID > std::numeric_limits<std::uint32_t>::max()) { + return true; + } + const auto TextID = static_cast<std::uint32_t>(WideID); + if (!IDs.insert(TextID).second) + return fail(IDNode->range(), "duplicate stack object ID"); + + StackObjectDesc Description; + Description.Kind = Fixed ? StackObjectKind::FixedABI : StackObjectKind::Default; + Description.Offset = 0; + if (const yaml::Node *NameNode = mappingValue(*Element, "name")) { + std::string_view Name; + if (scalar(*NameNode, Name, "stack object name")) + return true; + if (!Name.empty()) + Description.AllocaName = std::string(Name); + } + if (const yaml::Node *TypeNode = mappingValue(*Element, "type")) { + std::string_view Type; + if (scalar(*TypeNode, Type, "stack object type")) + return true; + if (Type == "default") + Description.Kind = Fixed ? StackObjectKind::FixedABI : StackObjectKind::Default; + else if (Type == "spill-slot") + Description.Kind = Fixed ? StackObjectKind::FixedSpillSlot : StackObjectKind::SpillSlot; + else if (!Fixed && Type == "variable-sized") + Description.Kind = StackObjectKind::VariableSized; + else + return fail(TypeNode->range(), "unknown stack object type '" + std::string(Type) + "'"); + } + if (const yaml::Node *OffsetNode = mappingValue(*Element, "offset")) { + std::int64_t Offset = 0; + if (signedValue(*OffsetNode, Offset, "a signed stack object offset")) + return true; + Description.Offset = Offset; + } else if (Fixed) { + Description.Offset = 0; + } + if (Description.Kind != StackObjectKind::VariableSized) { + const yaml::Node *SizeNode = mappingValue(*Element, "size"); + if (SizeNode == nullptr) { + if (!Fixed) + return fail(Element->range(), "fixed-size stack object requires 'size'"); + Description.Size = 0; + } else { + std::uint64_t Size = 0; + if (unsignedValue(*SizeNode, Size, "an unsigned stack object size")) + return true; + Description.Size = Size; + } + } + if (const yaml::Node *AlignmentNode = mappingValue(*Element, "alignment")) { + if (alignment(*AlignmentNode, Description.Align, true)) + return true; + } + if (const yaml::Node *StackIDNode = mappingValue(*Element, "stack-id")) { + std::string_view Name; + if (scalar(*StackIDNode, Name, "stack ID")) + return true; + const auto Stack = parseStackIDName(Name); + if (!Stack) + return fail(StackIDNode->range(), "unknown stack ID '" + std::string(Name) + "'"); + Description.Stack = *Stack; + } + if (const yaml::Node *ImmutableNode = mappingValue(*Element, "isImmutable")) { + if (boolean(*ImmutableNode, Description.IsImmutable)) + return true; + } + if (const yaml::Node *AliasedNode = mappingValue(*Element, "isAliased")) { + if (boolean(*AliasedNode, Description.IsAliased)) + return true; + } + if (const yaml::Node *SavedNode = mappingValue(*Element, "callee-saved-register")) { + std::string_view SavedName; + if (scalar(*SavedNode, SavedName, "physical register")) + return true; + if (!trim(SavedName).empty()) { + PhysicalRegister Reg; + if (physicalRegister(SavedName, SavedNode->range(), Reg)) + return true; + Description.CalleeSavedRegister = Reg; + } + } + if (const yaml::Node *RestoredNode = mappingValue(*Element, "callee-saved-restored")) { + if (boolean(*RestoredNode, Description.CalleeSavedRestored)) + return true; + } + if (const yaml::Node *LocalNode = mappingValue(*Element, "local-offset")) { + std::int64_t Offset = 0; + if (signedValue(*LocalNode, Offset, "a signed local frame offset")) + return true; + Description.LocalOffset = Offset; + } + MachineVariableDebugInfo DebugInfo; + if (parseVariableDebugInfo(*Element, Function, DebugInfo)) + return true; + if (DebugInfo.Variable || DebugInfo.Expression || DebugInfo.Location) + Description.DebugInfo = std::move(DebugInfo); + + auto Created = Function.frameInfo().createObject(std::move(Description)); + if (!Created) + return importError(Element->range(), Created.error()); + MIFrameObjectSlot Slot{ + .TextID = TextID, + .Index = *Created, + .Name = std::nullopt, + }; + if (const StackObjectDesc *CreatedDescription = + Function.frameInfo().object(*Created)->description(); + CreatedDescription != nullptr && CreatedDescription->AllocaName) { + Slot.Name = *CreatedDescription->AllocaName; + } + (Fixed ? Context.FixedStackObjects : Context.StackObjects).push_back(std::move(Slot)); + } + return false; + }; + + if (const yaml::Node *Fixed = mappingValue(Root, "fixedStack")) { + if (ParseList(*Fixed, true)) + return true; + } + if (const yaml::Node *Stack = mappingValue(Root, "stack")) { + if (ParseList(*Stack, false)) + return true; + } + return false; +} + +bool MIRDocumentParser::parseFrameReference(const yaml::Node &Node, const MIParserContext &Context, + FrameIndex &Index) { + std::string_view Text; + if (scalar(Node, Text, "frame index")) + return true; + bool Fixed = false; + std::optional<std::uint32_t> ID = parsePrefixedID(Text, "%stack."); + if (!ID) { + ID = parsePrefixedID(Text, "%fixed-stack."); + Fixed = true; + } + if (!ID) + return fail(Node.range(), "frame index must use '%stack.N' or '%fixed-stack.N'"); + const auto &Slots = Fixed ? Context.FixedStackObjects : Context.StackObjects; + const auto Found = std::ranges::find_if( + Slots, [ID](const MIFrameObjectSlot &Slot) { return Slot.TextID == *ID; }); + if (Found == Slots.end()) + return fail(Node.range(), "reference to undefined frame object"); + Index = Found->Index; + return false; +} + +MachineBasicBlock *MIRDocumentParser::blockAt(MachineFunction &Function, + std::uint32_t Number) const noexcept { + for (MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; Block = Block->next()) { + if (Block->denseNumber() == Number) + return Block; + } + return nullptr; +} + +MachineInstr *MIRDocumentParser::instructionAt(MachineFunction &Function, std::uint32_t BlockNumber, + std::uint32_t Offset) const noexcept { + MachineBasicBlock *Block = blockAt(Function, BlockNumber); + if (Block == nullptr) + return nullptr; + MachineInstr *Instruction = Block->firstInstruction(); + for (std::uint32_t Index = 0; Instruction != nullptr && Index < Offset; ++Index) + Instruction = Instruction->next(); + return Instruction; +} + +bool MIRDocumentParser::parseBlockReference(std::string_view Text, const TextRange &Range, + MachineFunction &Function, MachineBasicBlock *&Block) { + const auto Number = parsePrefixedID(Text, "%bb."); + if (!Number) + return fail(Range, "basic block reference must use the '%bb.N' MIR spelling"); + Block = blockAt(Function, *Number); + if (Block == nullptr) + return fail(Range, "reference to undefined machine basic block"); + return false; +} + +bool MIRDocumentParser::parseSaveRestorePoints(const yaml::Node &Node, MachineFunction &Function, + std::vector<MachineFrameSaveRestorePoint> &Points) { + if (requireKind(Node, yaml::NodeKind::Sequence, "frame save/restore points")) + return true; + constexpr std::array Keys{ + std::string_view{"point"}, + std::string_view{"registers"}, + }; + for (const yaml::NodeID PointID : Node.sequence()) { + const yaml::Node *PointNode = node(PointID); + if (PointNode == nullptr || + requireKind(*PointNode, yaml::NodeKind::Mapping, "frame save/restore point") || + validateKeys(*PointNode, Keys, "frame save/restore point")) { + return true; + } + const yaml::Node *BlockNode = mappingValue(*PointNode, "point"); + if (BlockNode == nullptr) + return fail(PointNode->range(), "frame save/restore point requires 'point'"); + std::string_view BlockText; + MachineBasicBlock *Block = nullptr; + if (scalar(*BlockNode, BlockText, "basic block reference") || + parseBlockReference(BlockText, BlockNode->range(), Function, Block)) { + return true; + } + if (Block == nullptr) + return fail(BlockNode->range(), "reference to undefined machine basic block"); + MachineFrameSaveRestorePoint Point{.Block = Block->stableID(), .Registers = {}}; + if (const yaml::Node *RegistersNode = mappingValue(*PointNode, "registers")) { + if (requireKind(*RegistersNode, yaml::NodeKind::Sequence, "save/restore point registers")) { + return true; + } + for (const yaml::NodeID RegID : RegistersNode->sequence()) { + const yaml::Node *RegNode = node(RegID); + PhysicalRegister Reg; + if (RegNode == nullptr || physicalRegister(*RegNode, Reg)) + return true; + Point.Registers.push_back(Reg); + } + } + Points.push_back(std::move(Point)); + } + return false; +} + +bool MIRDocumentParser::parseFrameState(const yaml::Node &Root, MachineFunction &Function, + const MIParserContext &Context) { + const yaml::Node *Frame = mappingValue(Root, "frameInfo"); + if (Frame == nullptr) + return false; + if (requireKind(*Frame, yaml::NodeKind::Mapping, "frameInfo")) + return true; + constexpr std::array Keys{ + std::string_view{"isFrameAddressTaken"}, + std::string_view{"isReturnAddressTaken"}, + std::string_view{"hasStackMap"}, + std::string_view{"hasPatchPoint"}, + std::string_view{"stackSize"}, + std::string_view{"offsetAdjustment"}, + std::string_view{"maxAlignment"}, + std::string_view{"adjustsStack"}, + std::string_view{"hasCalls"}, + std::string_view{"framePointerPolicy"}, + std::string_view{"stackProtector"}, + std::string_view{"functionContext"}, + std::string_view{"maxCallFrameSize"}, + std::string_view{"cvBytesOfCalleeSavedRegisters"}, + std::string_view{"hasOpaqueSPAdjustment"}, + std::string_view{"hasVAStart"}, + std::string_view{"hasMustTailInVarArgFunc"}, + std::string_view{"hasTailCall"}, + std::string_view{"isCalleeSavedInfoValid"}, + std::string_view{"localFrameSize"}, + std::string_view{"savePoint"}, + std::string_view{"restorePoint"}, + }; + if (validateKeys(*Frame, Keys, "frameInfo")) + return true; + + MachineFrameState State = Function.frameInfo().state(); + const std::array BoolBindings{ + std::pair{"isFrameAddressTaken", &State.IsFrameAddressTaken}, + std::pair{"isReturnAddressTaken", &State.IsReturnAddressTaken}, + std::pair{"hasStackMap", &State.HasStackMap}, + std::pair{"hasPatchPoint", &State.HasPatchPoint}, + std::pair{"adjustsStack", &State.AdjustsStack}, + std::pair{"hasCalls", &State.HasCalls}, + std::pair{"hasOpaqueSPAdjustment", &State.HasOpaqueSPAdjustment}, + std::pair{"hasVAStart", &State.HasVAStart}, + std::pair{"hasMustTailInVarArgFunc", &State.HasMustTailInVarArgFunc}, + std::pair{"hasTailCall", &State.HasTailCall}, + std::pair{"isCalleeSavedInfoValid", &State.IsCalleeSavedInfoValid}, + }; + for (const auto &[Key, Destination] : BoolBindings) { + if (const yaml::Node *Value = mappingValue(*Frame, Key)) { + if (boolean(*Value, *Destination)) + return true; + } + } + + if (const yaml::Node *Value = mappingValue(*Frame, "stackSize")) { + if (unsignedValue(*Value, State.StackSize, "an unsigned stack size")) + return true; + } + if (const yaml::Node *Value = mappingValue(*Frame, "offsetAdjustment")) { + std::int64_t Wide = 0; + if (signedValue(*Value, Wide, "a 32-bit signed offset adjustment") || + Wide < std::numeric_limits<std::int32_t>::min() || + Wide > std::numeric_limits<std::int32_t>::max()) { + return fail(Value->range(), "offset adjustment is outside the 32-bit signed range"); + } + State.OffsetAdjustment = static_cast<std::int32_t>(Wide); + } + if (const yaml::Node *Value = mappingValue(*Frame, "maxAlignment")) { + std::uint64_t Bytes = 0; + if (unsignedValue(*Value, Bytes, "an unsigned maximum alignment")) + return true; + if (Bytes == 0) { + State.MaxAlignment.reset(); + } else { + auto Parsed = Alignment::create(Bytes); + if (!Parsed) + return importError(Value->range(), Parsed.error()); + State.MaxAlignment = *Parsed; + } + } + if (const yaml::Node *Value = mappingValue(*Frame, "framePointerPolicy")) { + std::string_view Name; + if (scalar(*Value, Name, "frame-pointer policy")) + return true; + const auto Policy = parseMachineFramePointerPolicy(Name); + if (!Policy) + return fail(Value->range(), "unknown frame-pointer policy '" + std::string(Name) + "'"); + State.FramePointerPolicy = *Policy; + } + for (const auto &[Key, Destination] : {std::pair{"stackProtector", &State.StackProtector}, + std::pair{"functionContext", &State.FunctionContext}}) { + if (const yaml::Node *Value = mappingValue(*Frame, Key)) { + std::string_view Reference; + if (scalar(*Value, Reference, "frame index")) + return true; + if (Reference.empty()) { + Destination->reset(); + continue; + } + FrameIndex Index; + if (parseFrameReference(*Value, Context, Index)) + return true; + *Destination = Index; + } + } + if (const yaml::Node *Value = mappingValue(*Frame, "maxCallFrameSize")) { + std::uint64_t Size = 0; + if (unsignedValue(*Value, Size, "a 32-bit unsigned maximum call-frame size") || + Size > std::numeric_limits<std::uint32_t>::max()) { + return fail(Value->range(), "maximum call-frame size is outside the 32-bit unsigned range"); + } + if (Size == std::numeric_limits<std::uint32_t>::max()) + State.MaxCallFrameSize.reset(); + else + State.MaxCallFrameSize = Size; + } + for (const auto &[Key, Destination] : + {std::pair{"cvBytesOfCalleeSavedRegisters", &State.CVBytesOfCalleeSavedRegisters}, + std::pair{"localFrameSize", &State.LocalFrameSize}}) { + if (const yaml::Node *Value = mappingValue(*Frame, Key)) { + std::uint64_t Wide = 0; + if (unsignedValue(*Value, Wide, "a 32-bit unsigned frame value") || + Wide > std::numeric_limits<std::uint32_t>::max()) { + return fail(Value->range(), "frame value is outside the 32-bit unsigned range"); + } + *Destination = static_cast<std::uint32_t>(Wide); + } + } + if (const yaml::Node *Value = mappingValue(*Frame, "savePoint")) { + if (parseSaveRestorePoints(*Value, Function, State.SavePoints)) + return true; + } + if (const yaml::Node *Value = mappingValue(*Frame, "restorePoint")) { + if (parseSaveRestorePoints(*Value, Function, State.RestorePoints)) + return true; + } + auto StateResult = Function.frameInfo().setState(std::move(State)); + if (!StateResult) + return importError(Frame->range(), StateResult.error()); + return false; +} + +} // namespace rcc::mir + +namespace rcc::mir { + +using parser_detail::parseBool; +using parser_detail::parsePrefixedID; +using parser_detail::parseSigned; +using parser_detail::parseUnsigned; +using parser_detail::parseUnsigned32; +using parser_detail::trim; + +bool MIRDocumentParser::parseFunctionScalars(const yaml::Node &Root, MachineFunction &Function, + MIRFunctionDocumentInfo &DocumentInfo) { + if (const yaml::Node *AlignNode = mappingValue(Root, "alignment")) { + Alignment Align; + if (alignment(*AlignNode, Align, true)) + return true; + if (Align.bytes() != 1 || AlignNode->scalar() != "0") + Function.attributes().setAlignment(Align); + } + + constexpr std::array PropertyBindings{ + std::pair{"legalized", MachineFunctionProperty::Legalized}, + std::pair{"regBankSelected", MachineFunctionProperty::RegBankSelected}, + std::pair{"selected", MachineFunctionProperty::Selected}, + std::pair{"failedISel", MachineFunctionProperty::FailedISel}, + std::pair{"tracksRegLiveness", MachineFunctionProperty::TracksLiveness}, + std::pair{"noPhis", MachineFunctionProperty::NoPHIs}, + std::pair{"isSSA", MachineFunctionProperty::IsSSA}, + std::pair{"noVRegs", MachineFunctionProperty::NoVRegs}, + std::pair{"failsVerification", MachineFunctionProperty::FailsVerification}, + }; + for (const auto &[Key, Property] : PropertyBindings) { + const yaml::Node *ValueNode = mappingValue(Root, Key); + if (ValueNode == nullptr) + continue; + bool Value = false; + if (boolean(*ValueNode, Value)) + return true; + if (Value) + Function.properties().set(Property); + else + Function.properties().clear(Property); + if (Key == std::string_view("noPhis")) + DocumentInfo.NoPHIs = Value; + else if (Key == std::string_view("isSSA")) + DocumentInfo.IsSSA = Value; + else if (Key == std::string_view("noVRegs")) + DocumentInfo.NoVRegs = Value; + } + + constexpr std::array AttributeBindings{ + std::pair{"exposesReturnsTwice", MachineFunctionAttribute::ExposesReturnsTwice}, + std::pair{"hasWinCFI", MachineFunctionAttribute::HasWinCFI}, + std::pair{"hasFakeUses", MachineFunctionAttribute::HasFakeUses}, + std::pair{"callsEHReturn", MachineFunctionAttribute::CallsEHReturn}, + std::pair{"callsUnwindInit", MachineFunctionAttribute::CallsUnwindInit}, + std::pair{"hasEHContTarget", MachineFunctionAttribute::HasEHContTarget}, + std::pair{"hasEHScopes", MachineFunctionAttribute::HasEHScopes}, + std::pair{"hasEHFunclets", MachineFunctionAttribute::HasEHFunclets}, + std::pair{"isOutlined", MachineFunctionAttribute::IsOutlined}, + std::pair{"debugInstrRef", MachineFunctionAttribute::DebugInstrRef}, + std::pair{"tracksDebugUserValues", MachineFunctionAttribute::TracksDebugUserValues}, + }; + for (const auto &[Key, Attribute] : AttributeBindings) { + const yaml::Node *ValueNode = mappingValue(Root, Key); + if (ValueNode == nullptr) + continue; + bool Value = false; + if (boolean(*ValueNode, Value)) + return true; + if (Value) + Function.attributes().set(Attribute); + else + Function.attributes().clear(Attribute); + if (Key == std::string_view("hasFakeUses")) + DocumentInfo.HasFakeUses = Value; + } + + return false; +} + +bool MIRDocumentParser::recomputeProperties(const yaml::Node &Root, MachineFunction &Function, + const MIRFunctionDocumentInfo &DocumentInfo) { + bool HasPHI = false; + bool HasFakeUse = false; + bool IsSSA = true; + bool HasTiedOperands = false; + bool AllTiedOperandsRewritten = true; + bool HasFrameIndex = false; + bool HasExecutablePseudo = false; + std::unordered_set<Register> Definitions; + for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; + Block = Block->next()) { + for (const MachineInstr *Instruction = Block->firstInstruction(); Instruction != nullptr; + Instruction = Instruction->next()) { + const InstructionDescriptor *Descriptor = Target.instructionDescriptor(Instruction->opcode()); + if (Descriptor == nullptr) + return fail(Root.range(), "instruction has no target descriptor"); + HasPHI |= Descriptor->Flags.has(InstructionFlag::Phi); + HasFakeUse |= Descriptor->Flags.has(InstructionFlag::FakeUse); + HasExecutablePseudo |= Descriptor->Flags.has(InstructionFlag::Pseudo) && + !Descriptor->Flags.has(InstructionFlag::Meta) && + !Descriptor->Flags.has(InstructionFlag::TerminalEmission); + const auto Operands = Instruction->operands(); + for (std::size_t Index = 0; Index < Operands.size(); ++Index) { + const MachineOperand &Operand = Operands[Index]; + HasFrameIndex |= Operand.getIf<FrameIndexOperand>() != nullptr; + const auto Reg = Operand.registerValue(); + if (Reg && Reg->isVirtual() && Operand.isDef() && !Definitions.insert(*Reg).second) { + IsSSA = false; + } + if (!Operand.tiedOperandIndex() || !Operand.isUse()) + continue; + HasTiedOperands = true; + const MachineOperand *Definition = Instruction->operandAt(*Operand.tiedOperandIndex()); + if (Definition == nullptr || Definition->registerValue() != Reg) + AllTiedOperandsRewritten = false; + } + for (const MachineMemOperand *Memory : Instruction->memOperands()) + HasFrameIndex |= Memory->description().Base.frameIndex().has_value(); + } + } + const bool NoPHIs = !HasPHI; + const bool NoVRegs = + std::ranges::none_of(Function.registerInfo().virtualRegisters(), + [](const VirtualRegisterInfo &Info) { return !Info.IsTombstone; }); + + const auto ApplyComputed = [this, &Root, &Function](std::optional<bool> Explicit, bool Computed, + MachineFunctionProperty Property, + std::string_view Conflict) -> bool { + if (Explicit && *Explicit && !Computed) { + return fail(Root.range(), std::string(Function.name()) + std::string(Conflict)); + } + if (Explicit.value_or(Computed)) + Function.properties().set(Property); + else + Function.properties().clear(Property); + return false; + }; + if (ApplyComputed(DocumentInfo.NoPHIs, NoPHIs, MachineFunctionProperty::NoPHIs, + " has explicit property NoPhi, but contains at least one PHI") || + ApplyComputed(DocumentInfo.IsSSA, IsSSA, MachineFunctionProperty::IsSSA, + " has explicit property IsSSA, but is not valid SSA") || + ApplyComputed(DocumentInfo.NoVRegs, NoVRegs, MachineFunctionProperty::NoVRegs, + " has explicit property NoVRegs, but contains virtual registers")) { + return true; + } + + if (DocumentInfo.HasFakeUses && !*DocumentInfo.HasFakeUses && HasFakeUse) { + return fail(Root.range(), + std::string(Function.name()) + + " has explicit property hasFakeUses=false, but contains fake uses"); + } + if (DocumentInfo.HasFakeUses.value_or(HasFakeUse)) + Function.attributes().set(MachineFunctionAttribute::HasFakeUses); + else + Function.attributes().clear(MachineFunctionAttribute::HasFakeUses); + + const bool Selected = Function.properties().has(MachineFunctionProperty::Selected); + const bool StructurallyPostSSA = + Selected && NoPHIs && !Function.properties().has(MachineFunctionProperty::IsSSA); + if (AllTiedOperandsRewritten && (HasTiedOperands || StructurallyPostSSA)) + Function.properties().set(MachineFunctionProperty::TiedOpsRewritten); + else + Function.properties().clear(MachineFunctionProperty::TiedOpsRewritten); + + if (HasFrameIndex) + Function.properties().clear(MachineFunctionProperty::NoFrameIndices); + else + Function.properties().set(MachineFunctionProperty::NoFrameIndices); + + const MachineFrameState &FrameState = Function.frameInfo().state(); + const bool StructurallyFinalizedFrame = Selected && NoVRegs && !HasFrameIndex && + FrameState.IsCalleeSavedInfoValid && + FrameState.MaxAlignment.has_value(); + if (StructurallyFinalizedFrame) { + if (!Function.frameInfo().isFrozen()) { + auto Frozen = Function.frameInfo().freeze(FrameState.StackSize, *FrameState.MaxAlignment); + if (!Frozen) + return importError(Root.range(), Frozen.error(), "mir-frame-finalization"); + } + Function.properties().set(MachineFunctionProperty::FrameFinalized); + } else { + Function.properties().clear(MachineFunctionProperty::FrameFinalized); + } + + if (StructurallyFinalizedFrame && !HasExecutablePseudo) + Function.properties().set(MachineFunctionProperty::NoExecutablePseudos); + else + Function.properties().clear(MachineFunctionProperty::NoExecutablePseudos); + return false; +} + +bool MIRDocumentParser::parseRegisters(const yaml::Node &Root, MachineFunction &Function, + MIParserContext &Context) { + const yaml::Node *Registers = mappingValue(Root, "registers"); + if (Registers != nullptr) { + if (requireKind(*Registers, yaml::NodeKind::Sequence, "registers")) + return true; + constexpr std::array RegisterKeys{ + std::string_view{"id"}, + std::string_view{"class"}, + std::string_view{"preferred-register"}, + std::string_view{"flags"}, + }; + std::unordered_set<std::uint32_t> IDs; + for (const yaml::NodeID ElementID : Registers->sequence()) { + const yaml::Node *Element = node(ElementID); + if (Element == nullptr || requireKind(*Element, yaml::NodeKind::Mapping, "register entry") || + validateKeys(*Element, RegisterKeys, "register")) { + return true; + } + const yaml::Node *IDNode = mappingValue(*Element, "id"); + const yaml::Node *ClassNode = mappingValue(*Element, "class"); + if (IDNode == nullptr || ClassNode == nullptr) + return fail(Element->range(), "register entry requires 'id' and 'class'"); + std::uint64_t WideID = 0; + std::string_view ClassName; + if (unsignedValue(*IDNode, WideID, "a 32-bit virtual-register ID") || + WideID > std::numeric_limits<std::uint32_t>::max() || + scalar(*ClassNode, ClassName, "register class or bank")) { + return true; + } + const auto TextID = static_cast<std::uint32_t>(WideID); + if (!IDs.insert(TextID).second) + return fail(IDNode->range(), "duplicate virtual register ID"); + auto Created = Function.registerInfo().createVirtualRegister(); + if (!Created) + return importError(IDNode->range(), Created.error()); + Context.VirtualRegisters.push_back({TextID, *Created}); + + if (ClassName != "_") { + if (const RegisterClassDesc *Class = Target.registers().registerClass(ClassName)) { + auto SetResult = Function.registerInfo().setRegisterClass(*Created, Class->ID); + if (!SetResult) + return importError(ClassNode->range(), SetResult.error()); + SetResult = Function.registerInfo().setRegisterBank(*Created, Class->DefaultBank); + if (!SetResult) + return importError(ClassNode->range(), SetResult.error()); + } else if (const RegisterBankDesc *Bank = Target.registers().registerBank(ClassName)) { + auto SetResult = Function.registerInfo().setRegisterBank(*Created, Bank->ID); + if (!SetResult) + return importError(ClassNode->range(), SetResult.error()); + } else { + return fail(ClassNode->range(), + "unknown register class or bank '" + std::string(ClassName) + "'"); + } + } + + if (const yaml::Node *Preferred = mappingValue(*Element, "preferred-register")) { + std::string_view PreferredName; + PhysicalRegister Reg; + if (scalar(*Preferred, PreferredName, "physical register")) + return true; + if (!trim(PreferredName).empty()) { + if (physicalRegister(PreferredName, Preferred->range(), Reg)) + return true; + auto AddResult = Function.registerInfo().addAllocationHint(*Created, Reg); + if (!AddResult) + return importError(Preferred->range(), AddResult.error()); + } + } + if (const yaml::Node *Flags = mappingValue(*Element, "flags")) { + if (requireKind(*Flags, yaml::NodeKind::Sequence, "virtual-register flags")) + return true; + std::uint8_t Mask = 0; + for (const yaml::NodeID FlagID : Flags->sequence()) { + const yaml::Node *FlagNode = node(FlagID); + std::string_view Name; + if (FlagNode == nullptr || scalar(*FlagNode, Name, "virtual-register flag")) + return true; + const auto Flag = Target.mirText().virtualRegisterFlag(Name); + if (!Flag) + return fail(FlagNode->range(), + "use of undefined register flag '" + std::string(Name) + "'"); + if ((Mask & *Flag) != 0) + return fail(FlagNode->range(), "duplicate virtual-register flag"); + Mask = static_cast<std::uint8_t>(Mask | *Flag); + } + auto SetResult = Function.registerInfo().setTargetRegisterFlags(*Created, Mask); + if (!SetResult) + return importError(Flags->range(), SetResult.error()); + } + } + } + + const yaml::Node *LiveIns = mappingValue(Root, "liveins"); + if (LiveIns != nullptr) { + if (requireKind(*LiveIns, yaml::NodeKind::Sequence, "liveins")) + return true; + constexpr std::array LiveInKeys{ + std::string_view{"reg"}, + std::string_view{"virtual-reg"}, + }; + for (const yaml::NodeID ElementID : LiveIns->sequence()) { + const yaml::Node *Element = node(ElementID); + if (Element == nullptr || requireKind(*Element, yaml::NodeKind::Mapping, "live-in entry") || + validateKeys(*Element, LiveInKeys, "live-in")) { + return true; + } + const yaml::Node *RegNode = mappingValue(*Element, "reg"); + if (RegNode == nullptr) + return fail(Element->range(), "live-in entry requires 'reg'"); + PhysicalRegister Reg; + if (physicalRegister(*RegNode, Reg)) + return true; + std::optional<Register> Initial; + if (const yaml::Node *VirtualNode = mappingValue(*Element, "virtual-reg")) { + std::string_view Text; + if (scalar(*VirtualNode, Text, "virtual register")) + return true; + if (!trim(Text).empty()) { + Register Virtual; + if (virtualRegister(Text, VirtualNode->range(), Function, Context, Virtual)) + return true; + Initial = Virtual; + } + } + auto AddResult = Function.registerInfo().addLiveIn(Reg, Initial); + if (!AddResult) + return importError(Element->range(), AddResult.error()); + } + } + + if (const yaml::Node *CalleeSaved = mappingValue(Root, "calleeSavedRegisters")) { + if (requireKind(*CalleeSaved, yaml::NodeKind::Sequence, "callee-saved registers")) + return true; + std::vector<PhysicalRegister> RegistersValue; + for (const yaml::NodeID RegID : CalleeSaved->sequence()) { + const yaml::Node *RegNode = node(RegID); + PhysicalRegister Reg; + if (RegNode == nullptr || physicalRegister(*RegNode, Reg)) + return true; + RegistersValue.push_back(Reg); + } + auto SetResult = Function.records().setCalleeSavedRegisters(std::move(RegistersValue)); + if (!SetResult) + return importError(CalleeSaved->range(), SetResult.error()); + } + return false; +} + +TextRange MIRDocumentParser::scalarDiagnosticRange(const yaml::Node &Scalar, + const TextRange &ScalarRange) const noexcept { + if (Scalar.scalarStyle() != yaml::ScalarStyle::Literal && + Scalar.scalarStyle() != yaml::ScalarStyle::Folded) { + std::size_t ContentOffset = Scalar.range().Begin.Offset; + if (Scalar.scalarStyle() == yaml::ScalarStyle::SingleQuoted || + Scalar.scalarStyle() == yaml::ScalarStyle::DoubleQuoted) { + ++ContentOffset; + } + return Result.Source.range(ContentOffset + ScalarRange.Begin.Offset, + ContentOffset + ScalarRange.End.Offset); + } + + const std::size_t FirstBodyLine = Scalar.range().Begin.Line + 1U; + const std::size_t LastBodyLine = Scalar.range().End.Line; + if (FirstBodyLine > LastBodyLine || FirstBodyLine == 0) + return Scalar.range(); + + std::size_t Indent = 0; + bool FoundContent = false; + for (std::size_t Line = FirstBodyLine; Line <= LastBodyLine; ++Line) { + const std::string_view Text = Result.Source.line(Line); + std::size_t CurrentIndent = 0; + while (CurrentIndent < Text.size() && Text[CurrentIndent] == ' ') + ++CurrentIndent; + if (CurrentIndent == Text.size()) + continue; + Indent = CurrentIndent; + FoundContent = true; + break; + } + if (!FoundContent) + return Scalar.range(); + + const TextPosition DefaultPosition = Scalar.range().Begin; + const auto MapPosition = [this, FirstBodyLine, LastBodyLine, Indent, + DefaultPosition](TextPosition Position) { + const std::size_t SourceLine = FirstBodyLine + Position.Line - 1U; + if (Position.Line == 0 || SourceLine > LastBodyLine || SourceLine > Result.Source.lineCount()) { + return DefaultPosition; + } + const std::string_view SourceLineText = Result.Source.line(SourceLine); + const std::size_t BodyColumn = Position.Column == 0 ? 0 : Position.Column - 1U; + const std::size_t SourceColumn = std::min(Indent + BodyColumn, SourceLineText.size()); + const std::size_t Offset = Result.Source.lineOffsets()[SourceLine - 1U] + SourceColumn; + return Result.Source.positionAt(Offset); + }; + return {.Begin = MapPosition(ScalarRange.Begin), .End = MapPosition(ScalarRange.End)}; +} + +bool MIRDocumentParser::parseBody(const yaml::Node &Root, MachineFunction &Function, + MIParserContext &Context) { + const yaml::Node *Body = mappingValue(Root, "body"); + std::string BodyText; + TextRange BodyRange = Root.range(); + if (Body != nullptr) { + std::string_view Text; + if (scalar(*Body, Text, "machine function body")) + return true; + BodyText = Text; + BodyRange = Body->range(); + } + + if (!effectiveDataLayout().empty()) { + const LLVMConstantSyntaxOptions PointerOptions = pointerLayoutOptions(); + Context.PointerSizeInBits = [PointerOptions](std::uint32_t AddressSpace) { + return llvmPointerSizeInBits(AddressSpace, PointerOptions); + }; + } + Context.SyncScopes = ActiveDocument->SyncScopes; + Context.InternSyncScope = [this](std::string_view Name) -> Expected<SyncScopeID> { + if (Name.empty()) + return makeUnexpected(ErrorCode::ParseError, "synchronization scope name cannot be empty"); + const auto Existing = + std::ranges::find_if(ActiveDocument->SyncScopes, + [Name](const MISyncScopeSlot &Slot) { return Slot.Name == Name; }); + if (Existing != ActiveDocument->SyncScopes.end()) + return Existing->ID; + if (ActiveDocument->SyncScopes.size() >= + static_cast<std::size_t>(std::numeric_limits<std::uint32_t>::max())) { + return makeUnexpected(ErrorCode::Unsupported, + "MIR document contains too many synchronization scopes"); + } + const auto ID = + SyncScopeID::fromValue(static_cast<std::uint32_t>(ActiveDocument->SyncScopes.size() + 1U)); + ActiveDocument->SyncScopes.push_back({.Name = std::string(Name), .ID = ID}); + return ID; + }; + MIParserOptions BodyOptions = Options.MachineInstructions; + BodyOptions.VerifyAfterParse = false; + MIParser Parser(Function, Target, + std::string(Result.Source.name()) + ":" + std::string(Function.name()), + std::move(BodyText), Context, BodyOptions); + auto Parsed = Parser.parse(); + if (!Parsed) { + if (!Parser.diagnostics().empty()) { + const TextRange Range = + Body == nullptr ? BodyRange + : scalarDiagnosticRange(*Body, Parser.diagnostics().front().Range); + return fail(Range, Parser.diagnostics().front().Message, "mir-body"); + } + return importError(BodyRange, Parsed.error(), "mir-body"); + } + return false; +} + +bool MIRDocumentParser::convertTargetNode(const yaml::Node &Source, MIRTargetNode &Destination) { + switch (Source.kind()) { + case yaml::NodeKind::Null: + Destination = MIRTargetNode::null(); + return false; + case yaml::NodeKind::Scalar: + Destination = MIRTargetNode::scalar(std::string(Source.scalar())); + return false; + case yaml::NodeKind::Sequence: { + std::vector<MIRTargetNode> Values; + Values.reserve(Source.sequence().size()); + for (const yaml::NodeID ElementID : Source.sequence()) { + const yaml::Node *Element = node(ElementID); + if (Element == nullptr) + return fail(Source.range(), "target MIR sequence contains an invalid node"); + MIRTargetNode Value; + if (convertTargetNode(*Element, Value)) + return true; + Values.push_back(std::move(Value)); + } + Destination = MIRTargetNode::sequence(std::move(Values)); + return false; + } + case yaml::NodeKind::Mapping: { + std::vector<MIRTargetMappingEntry> Entries; + Entries.reserve(Source.mapping().size()); + for (const yaml::MappingEntry &SourceEntry : Source.mapping()) { + const yaml::Node *ValueNode = node(SourceEntry.Value); + if (ValueNode == nullptr) + return fail(SourceEntry.KeyRange, "target MIR mapping contains an invalid node"); + MIRTargetNode Value; + if (convertTargetNode(*ValueNode, Value)) + return true; + Entries.push_back({SourceEntry.Key, std::move(Value)}); + } + Destination = MIRTargetNode::mapping(std::move(Entries)); + return false; + } + } + return fail(Source.range(), "target MIR node has an unknown kind"); +} + +bool MIRDocumentParser::parseTargetState(const yaml::Node &Root, MachineFunction &Function) { + const yaml::Node *StateNode = mappingValue(Root, "machineFunctionInfo"); + if (StateNode == nullptr) + return false; + MIRTargetNode TargetNode; + if (convertTargetNode(*StateNode, TargetNode)) + return true; + auto State = Target.mirText().parseFunctionState(TargetNode); + if (!State) + return importError(StateNode->range(), State.error(), "mir-target-state"); + auto SetResult = Function.setTargetState(std::move(*State)); + if (!SetResult) + return importError(StateNode->range(), SetResult.error(), "mir-target-state"); + return false; +} + +bool MIRDocumentParser::validateDocumentReferences(const MachineFunction &Function) { + auto Verification = verifyMIRDocumentReferences(*ActiveDocument, Function, true); + if (Verification) + return false; + return importError(Result.Source.range(0, 0), Verification.error()); +} + +} // namespace rcc::mir + +namespace rcc::mir { + +using parser_detail::parseBool; +using parser_detail::parsePrefixedID; +using parser_detail::parseSigned; +using parser_detail::parseUnsigned; +using parser_detail::parseUnsigned32; +using parser_detail::trim; + +bool MIRDocumentParser::parseFunctionRecords(const yaml::Node &Root, MachineFunction &Function) { + auto ParseLocation = [this, &Function](const yaml::Node &Entry, + MachineInstr *&Instruction) -> bool { + const yaml::Node *BlockNode = mappingValue(Entry, "bb"); + const yaml::Node *OffsetNode = mappingValue(Entry, "offset"); + if (BlockNode == nullptr || OffsetNode == nullptr) + return fail(Entry.range(), "instruction location requires 'bb' and 'offset'"); + std::uint64_t Block = 0; + std::uint64_t Offset = 0; + if (unsignedValue(*BlockNode, Block, "a 32-bit basic block number") || + unsignedValue(*OffsetNode, Offset, "a 32-bit instruction offset") || + Block > std::numeric_limits<std::uint32_t>::max() || + Offset > std::numeric_limits<std::uint32_t>::max()) { + return true; + } + Instruction = instructionAt(Function, static_cast<std::uint32_t>(Block), + static_cast<std::uint32_t>(Offset)); + if (Instruction == nullptr) + return fail(Entry.range(), "instruction location is outside the machine function body"); + return false; + }; + + if (const yaml::Node *CallSites = mappingValue(Root, "callSites")) { + if (requireKind(*CallSites, yaml::NodeKind::Sequence, "call-site records")) + return true; + constexpr std::array CallSiteKeys{ + std::string_view{"bb"}, + std::string_view{"offset"}, + std::string_view{"fwdArgRegs"}, + std::string_view{"calleeTypeIds"}, + }; + constexpr std::array ArgumentKeys{ + std::string_view{"arg"}, + std::string_view{"reg"}, + }; + for (const yaml::NodeID CallSiteID : CallSites->sequence()) { + const yaml::Node *CallSiteNode = node(CallSiteID); + if (CallSiteNode == nullptr || + requireKind(*CallSiteNode, yaml::NodeKind::Mapping, "call-site record") || + validateKeys(*CallSiteNode, CallSiteKeys, "call-site record")) { + return true; + } + MachineInstr *Call = nullptr; + if (ParseLocation(*CallSiteNode, Call)) + return true; + if (Call == nullptr) + return fail(CallSiteNode->range(), + "instruction location is outside the machine function body"); + MachineCallSiteInfo Info{ + .CallInstruction = Call->stableID(), + .ForwardedArguments = {}, + .CalleeTypeIDs = {}, + }; + if (const yaml::Node *Arguments = mappingValue(*CallSiteNode, "fwdArgRegs")) { + if (requireKind(*Arguments, yaml::NodeKind::Sequence, "forwarded call arguments")) { + return true; + } + for (const yaml::NodeID ArgumentID : Arguments->sequence()) { + const yaml::Node *Argument = node(ArgumentID); + if (Argument == nullptr || + requireKind(*Argument, yaml::NodeKind::Mapping, "forwarded call argument") || + validateKeys(*Argument, ArgumentKeys, "forwarded call argument")) { + return true; + } + const yaml::Node *NumberNode = mappingValue(*Argument, "arg"); + const yaml::Node *RegNode = mappingValue(*Argument, "reg"); + if (NumberNode == nullptr || RegNode == nullptr) + return fail(Argument->range(), "forwarded call argument requires 'arg' and 'reg'"); + std::uint16_t Number = 0; + std::uint64_t WideNumber = 0; + PhysicalRegister Reg; + if (unsignedValue(*NumberNode, WideNumber, "a 16-bit argument number") || + WideNumber > std::numeric_limits<std::uint16_t>::max() || + physicalRegister(*RegNode, Reg)) { + return true; + } + Number = static_cast<std::uint16_t>(WideNumber); + Info.ForwardedArguments.push_back({Reg, Number}); + } + } + if (const yaml::Node *TypeIDs = mappingValue(*CallSiteNode, "calleeTypeIds")) { + if (requireKind(*TypeIDs, yaml::NodeKind::Sequence, "callee type IDs")) + return true; + for (const yaml::NodeID TypeID : TypeIDs->sequence()) { + const yaml::Node *TypeNode = node(TypeID); + std::uint64_t Value = 0; + if (TypeNode == nullptr || + unsignedValue(*TypeNode, Value, "an unsigned callee type ID")) { + return true; + } + Info.CalleeTypeIDs.push_back(Value); + } + } + auto AddResult = Function.records().addCallSite(std::move(Info)); + if (!AddResult) + return importError(CallSiteNode->range(), AddResult.error()); + } + } + + if (const yaml::Node *CalledGlobals = mappingValue(Root, "calledGlobals")) { + if (requireKind(*CalledGlobals, yaml::NodeKind::Sequence, "called-global records")) + return true; + constexpr std::array Keys{ + std::string_view{"bb"}, + std::string_view{"offset"}, + std::string_view{"callee"}, + std::string_view{"flags"}, + }; + for (const yaml::NodeID CalledID : CalledGlobals->sequence()) { + const yaml::Node *CalledNode = node(CalledID); + if (CalledNode == nullptr || + requireKind(*CalledNode, yaml::NodeKind::Mapping, "called-global record") || + validateKeys(*CalledNode, Keys, "called-global record")) { + return true; + } + MachineInstr *Call = nullptr; + if (ParseLocation(*CalledNode, Call)) + return true; + if (Call == nullptr) + return fail(CalledNode->range(), + "instruction location is outside the machine function body"); + const yaml::Node *CalleeNode = mappingValue(*CalledNode, "callee"); + const yaml::Node *FlagsNode = mappingValue(*CalledNode, "flags"); + if (CalleeNode == nullptr || FlagsNode == nullptr) + return fail(CalledNode->range(), "called-global record requires 'callee' and 'flags'"); + std::string_view Callee; + std::uint64_t WideFlags = 0; + if (scalar(*CalleeNode, Callee, "callee name") || + unsignedValue(*FlagsNode, WideFlags, "32-bit called-global flags") || + WideFlags > std::numeric_limits<std::uint32_t>::max()) { + return true; + } + if (ActiveDocument->EmbeddedIR && !ActiveDocument->EmbeddedIR->hasGlobalValue(Callee)) { + return fail(CalleeNode->range(), "called global references an undefined embedded global"); + } + auto AddResult = + Function.records().addCalledGlobal({.CallInstruction = Call->stableID(), + .Callee = std::string(Callee), + .Flags = static_cast<std::uint32_t>(WideFlags)}); + if (!AddResult) + return importError(CalledNode->range(), AddResult.error()); + } + } + + if (const yaml::Node *Substitutions = mappingValue(Root, "debugValueSubstitutions")) { + if (requireKind(*Substitutions, yaml::NodeKind::Sequence, "debug value substitutions")) { + return true; + } + constexpr std::array Keys{ + std::string_view{"srcinst"}, std::string_view{"srcop"}, std::string_view{"dstinst"}, + std::string_view{"dstop"}, std::string_view{"subreg"}, + }; + for (const yaml::NodeID SubstitutionID : Substitutions->sequence()) { + const yaml::Node *Substitution = node(SubstitutionID); + if (Substitution == nullptr || + requireKind(*Substitution, yaml::NodeKind::Mapping, "debug value substitution") || + validateKeys(*Substitution, Keys, "debug value substitution")) { + return true; + } + std::array<std::uint32_t, 5> Values{}; + for (std::size_t Index = 0; Index < Keys.size(); ++Index) { + const yaml::Node *ValueNode = mappingValue(*Substitution, Keys[Index]); + if (ValueNode == nullptr) + return fail(Substitution->range(), "debug value substitution requires every fixed field"); + std::uint64_t Wide = 0; + if (unsignedValue(*ValueNode, Wide, "a 32-bit debug substitution value") || + Wide > std::numeric_limits<std::uint32_t>::max()) { + return true; + } + Values[Index] = static_cast<std::uint32_t>(Wide); + } + auto AddResult = Function.records().addDebugValueSubstitution( + {.Source = {Values[0], Values[1]}, + .Destination = {Values[2], Values[3]}, + .Subregister = SubRegIndex::fromValue(Values[4])}); + if (!AddResult) + return importError(Substitution->range(), AddResult.error()); + } + } + + if (const yaml::Node *PrefetchTargets = mappingValue(Root, "prefetch-targets")) { + if (requireKind(*PrefetchTargets, yaml::NodeKind::Sequence, "prefetch targets")) + return true; + for (const yaml::NodeID TargetID : PrefetchTargets->sequence()) { + const yaml::Node *TargetNode = node(TargetID); + std::string_view Text; + if (TargetNode == nullptr || scalar(*TargetNode, Text, "prefetch target")) + return true; + + std::size_t ContentOffset = TargetNode->range().Begin.Offset; + if (TargetNode->scalarStyle() == yaml::ScalarStyle::SingleQuoted || + TargetNode->scalarStyle() == yaml::ScalarStyle::DoubleQuoted) { + ++ContentOffset; + } + const auto ErrorRange = [this, ContentOffset](std::size_t Begin, std::size_t End = 1U) { + return Result.Source.range(ContentOffset + Begin, ContentOffset + Begin + End); + }; + const auto SkipSpace = [&Text](std::size_t &Cursor) { + while (Cursor < Text.size() && (Text[Cursor] == ' ' || Text[Cursor] == '\t')) + ++Cursor; + }; + const auto ParseNumber = [&Text, &SkipSpace](std::size_t &Cursor, std::uint32_t &Value) { + SkipSpace(Cursor); + const std::size_t Begin = Cursor; + while (Cursor < Text.size() && Text[Cursor] >= '0' && Text[Cursor] <= '9') + ++Cursor; + return Begin != Cursor && parseUnsigned32(Text.substr(Begin, Cursor - Begin), Value); + }; + + std::size_t Cursor = 0; + SkipSpace(Cursor); + if (!Text.substr(Cursor).starts_with("bb_id") || + (Cursor + std::string_view("bb_id").size() < Text.size() && + Text[Cursor + std::string_view("bb_id").size()] != ' ' && + Text[Cursor + std::string_view("bb_id").size()] != '\t')) { + return fail(ErrorRange(Cursor), "expected 'bb_id'"); + } + Cursor += std::string_view("bb_id").size(); + + std::array<std::uint32_t, 3> Values{}; + if (!ParseNumber(Cursor, Values[0])) + return fail(ErrorRange(0), "expected unsigned integer"); + + SkipSpace(Cursor); + if (Cursor < Text.size() && (Text[Cursor] == ',' || Text[Cursor] == '.')) { + ++Cursor; + if (!ParseNumber(Cursor, Values[1])) + return fail(ErrorRange(0), "expected unsigned integer"); + } else if (Cursor < Text.size() && Text[Cursor] >= '0' && Text[Cursor] <= '9') { + if (!ParseNumber(Cursor, Values[1])) + return fail(ErrorRange(0), "expected unsigned integer"); + } + + SkipSpace(Cursor); + if (Cursor >= Text.size() || Text[Cursor] != ',') + return fail(ErrorRange(Cursor), "expected ','"); + ++Cursor; + if (!ParseNumber(Cursor, Values[2])) + return fail(ErrorRange(0), "expected unsigned integer"); + SkipSpace(Cursor); + if (Cursor != Text.size()) + return fail(ErrorRange(0), "unexpected content after prefetch target"); + + auto AddResult = Function.records().addPrefetchTarget( + {.Block = {Values[0], Values[1]}, .CallsiteIndex = Values[2]}); + if (!AddResult) + return importError(TargetNode->range(), AddResult.error()); + } + } + + if (const yaml::Node *EntryValues = mappingValue(Root, "entry_values")) { + if (requireKind(*EntryValues, yaml::NodeKind::Sequence, "entry values")) + return true; + constexpr std::array Keys{ + std::string_view{"entry-value-register"}, + std::string_view{"debug-info-variable"}, + std::string_view{"debug-info-expression"}, + std::string_view{"debug-info-location"}, + }; + for (const yaml::NodeID EntryID : EntryValues->sequence()) { + const yaml::Node *Entry = node(EntryID); + if (Entry == nullptr || requireKind(*Entry, yaml::NodeKind::Mapping, "entry value") || + validateKeys(*Entry, Keys, "entry value")) { + return true; + } + const yaml::Node *RegNode = mappingValue(*Entry, "entry-value-register"); + const yaml::Node *VariableNode = mappingValue(*Entry, "debug-info-variable"); + const yaml::Node *ExpressionNode = mappingValue(*Entry, "debug-info-expression"); + const yaml::Node *LocationNode = mappingValue(*Entry, "debug-info-location"); + if (RegNode == nullptr || VariableNode == nullptr || ExpressionNode == nullptr || + LocationNode == nullptr) { + return fail(Entry->range(), "entry value requires all four fixed fields"); + } + PhysicalRegister Reg; + if (physicalRegister(*RegNode, Reg)) + return true; + MachineVariableDebugInfo DebugInfo; + if (parseVariableDebugInfo(*Entry, Function, DebugInfo)) + return true; + if (!DebugInfo.Variable || !DebugInfo.Expression || !DebugInfo.Location) { + return fail(Entry->range(), "entry value debug metadata cannot be empty", + "mir-debug-metadata"); + } + auto AddResult = + Function.records().addEntryValue({.Reg = Reg, .DebugInfo = std::move(DebugInfo)}); + if (!AddResult) + return importError(Entry->range(), AddResult.error()); + } + } + return false; +} + +} // namespace rcc::mir + +namespace rcc::mir { + +using parser_detail::parseBool; +using parser_detail::parsePrefixedID; +using parser_detail::parseSigned; +using parser_detail::parseUnsigned; +using parser_detail::parseUnsigned32; +using parser_detail::trim; + +Expected<MachineMetadataReference> +MIRDocumentParser::resolveMetadataReference(const MachineFunction &Function, + std::uint32_t Slot) const { + if (ActiveDocument != nullptr && ActiveDocument->EmbeddedIR && + ActiveDocument->EmbeddedIR->metadata(std::to_string(Slot)) != nullptr) { + return MachineMetadataReference::embeddedIR(Slot); + } + if (Function.metadata().node(Slot) != nullptr) + return MachineMetadataReference::machineFunction(Slot); + return makeUnexpected(ErrorCode::ParseError, + "metadata reference uses undefined slot '!" + std::to_string(Slot) + "'"); +} + +bool MIRDocumentParser::parseVariableDebugInfo(const yaml::Node &Entry, MachineFunction &Function, + MachineVariableDebugInfo &DebugInfo) { + const auto Resolve = [this, &Function](std::uint32_t Slot) { + return resolveMetadataReference(Function, Slot); + }; + if (const yaml::Node *VariableNode = mappingValue(Entry, "debug-info-variable")) { + std::string_view Text; + if (scalar(*VariableNode, Text, "debug variable")) + return true; + if (!Text.empty()) { + MetadataTextParser Parser(std::string(Text), Resolve); + auto Reference = Parser.parseReferenceOnly(); + if (!Reference) + return importError(VariableNode->range(), Reference.error(), "mir-debug-metadata"); + if (Reference->scope() != MachineMetadataScope::EmbeddedIR) { + return fail(VariableNode->range(), + "debug variable must reference embedded LLVM IR metadata", + "mir-debug-metadata"); + } + DebugInfo.Variable = *Reference; + } + } + if (const yaml::Node *ExpressionNode = mappingValue(Entry, "debug-info-expression")) { + std::string_view Text; + if (scalar(*ExpressionNode, Text, "debug expression")) + return true; + if (!Text.empty()) { + MetadataTextParser Parser(std::string(Text), Resolve); + auto Value = Parser.parseValue(); + if (!Value) + return importError(scalarDiagnosticRange(*ExpressionNode, Parser.diagnosticRange()), + Value.error(), "mir-debug-metadata"); + if (Value->reference() != nullptr && + Value->reference()->scope() != MachineMetadataScope::EmbeddedIR) { + return fail(ExpressionNode->range(), + "debug expression reference must use embedded LLVM IR metadata", + "mir-debug-metadata"); + } + DebugInfo.Expression = std::move(*Value); + } + } + if (const yaml::Node *LocationNode = mappingValue(Entry, "debug-info-location")) { + std::string_view Text; + if (scalar(*LocationNode, Text, "debug location")) + return true; + if (!Text.empty()) { + MetadataTextParser Parser(std::string(Text), Resolve); + auto Location = Parser.parseLocation(); + if (!Location) + return importError(LocationNode->range(), Location.error(), "mir-debug-metadata"); + if (Location->reference() != nullptr && + Location->reference()->scope() != MachineMetadataScope::EmbeddedIR) { + return fail(LocationNode->range(), + "variable debug location must reference embedded LLVM IR metadata", + "mir-debug-metadata"); + } + DebugInfo.Location = std::move(*Location); + } + } + if (!DebugInfo.Variable && !DebugInfo.Expression && !DebugInfo.Location) + return false; + auto Verification = DebugInfo.verifyStructure(); + if (!Verification) + return importError(Entry.range(), Verification.error(), "mir-debug-metadata"); + return false; +} + +bool MIRDocumentParser::parseMetadata(const yaml::Node &Root, MachineFunction &Function, + MIParserContext &Context) { + if (ActiveDocument->EmbeddedIR) { + for (const EmbeddedIRMetadataDefinition &Reference : + ActiveDocument->EmbeddedIR->metadataDefinitions()) { + std::uint32_t ID = 0; + if (!parseUnsigned32(Reference.Name, ID)) + continue; + if (std::ranges::any_of(Context.Metadata, + [ID](const MIMetadataSlot &Slot) { return Slot.TextID == ID; })) { + return fail(Root.range(), "embedded LLVM IR metadata slots are not unique"); + } + Context.Metadata.push_back( + {.TextID = ID, .Reference = MachineMetadataReference::embeddedIR(ID)}); + } + } + + const yaml::Node *Metadata = mappingValue(Root, "machineMetadataNodes"); + if (Metadata == nullptr) + return false; + if (requireKind(*Metadata, yaml::NodeKind::Sequence, "machine metadata nodes")) + return true; + + enum class ParsedElementKind : std::uint8_t { + Null, + Reference, + String, + }; + struct ParsedElement final { + ParsedElementKind Kind = ParsedElementKind::Reference; + std::uint32_t Reference = 0; + std::string String; + TextRange Range; + }; + struct Definition final { + const yaml::Node *Node = nullptr; + std::uint32_t Slot = 0; + bool IsDistinct = false; + std::vector<ParsedElement> Elements; + }; + std::vector<Definition> Definitions; + std::unordered_set<std::uint32_t> Slots; + + const auto TranslateRange = [this](const yaml::Node &Node, const TextRange &Range) { + std::size_t ContentBegin = Node.range().Begin.Offset; + if (Node.scalarStyle() == yaml::ScalarStyle::SingleQuoted || + Node.scalarStyle() == yaml::ScalarStyle::DoubleQuoted) { + ++ContentBegin; + } + return Result.Source.range(ContentBegin + Range.Begin.Offset, ContentBegin + Range.End.Offset); + }; + + for (const yaml::NodeID DefinitionID : Metadata->sequence()) { + const yaml::Node *DefinitionNode = node(DefinitionID); + std::string_view Text; + if (DefinitionNode == nullptr || scalar(*DefinitionNode, Text, "machine metadata definition")) + return true; + + MILexer Lexer("machine-metadata", std::string(Text)); + MIToken Token = Lexer.next(); + const auto LexerFailed = [this, &Lexer, DefinitionNode, &TranslateRange, &Token]() -> bool { + if (!Token.is(MITokenKind::Error)) + return false; + const auto Diagnostics = Lexer.diagnostics(); + const TextRange Range = Diagnostics.empty() + ? TranslateRange(*DefinitionNode, Token.range()) + : TranslateRange(*DefinitionNode, Diagnostics.front().Range); + return fail(Range, Diagnostics.empty() ? "invalid machine metadata node" + : Diagnostics.front().Message); + }; + const auto Advance = [&Lexer, &Token, &LexerFailed]() -> bool { + Token = Lexer.next(); + return LexerFailed(); + }; + const auto FailAtToken = [this, DefinitionNode, &TranslateRange](const MIToken &At, + std::string Message) { + return fail(TranslateRange(*DefinitionNode, At.range()), std::move(Message)); + }; + + if (LexerFailed()) + return true; + if (!Token.is(MITokenKind::Exclaim)) + return FailAtToken(Token, "expected a metadata node"); + if (Advance()) + return true; + std::uint32_t Slot = 0; + if (!Token.is(MITokenKind::IntegerLiteral) || !parseUnsigned32(Token.integerValue(), Slot)) { + return FailAtToken(Token, "expected metadata id after '!'"); + } + if (Advance()) + return true; + if (!Token.is(MITokenKind::Equal)) + return FailAtToken(Token, "expected '='"); + if (Advance()) + return true; + + const bool IsDistinct = Token.isKeyword("distinct"); + if (IsDistinct && Advance()) + return true; + if (!Token.is(MITokenKind::Exclaim)) + return FailAtToken(Token, "expected a metadata node"); + if (Advance()) + return true; + if (!Token.is(MITokenKind::LeftBrace)) + return FailAtToken(Token, "expected '{' here"); + if (Advance()) + return true; + + std::vector<ParsedElement> Elements; + if (!Token.is(MITokenKind::RightBrace)) { + while (true) { + if (Token.hasStringValue() && Token.stringValue() == "null") { + Elements.push_back({.Kind = ParsedElementKind::Null, + .Reference = 0, + .String = {}, + .Range = TranslateRange(*DefinitionNode, Token.range())}); + } else { + if (!Token.is(MITokenKind::Exclaim)) + return FailAtToken(Token, "expected '!' here"); + const TextRange ReferenceRange = TranslateRange(*DefinitionNode, Token.range()); + if (Advance()) + return true; + + if (Token.is(MITokenKind::StringConstant)) { + Elements.push_back({.Kind = ParsedElementKind::String, + .Reference = 0, + .String = std::string(Token.stringValue()), + .Range = TranslateRange(*DefinitionNode, Token.range())}); + } else { + std::uint32_t Reference = 0; + if (!Token.is(MITokenKind::IntegerLiteral) || + !parseUnsigned32(Token.integerValue(), Reference)) { + return FailAtToken(Token, "expected metadata id after '!'"); + } + Elements.push_back({.Kind = ParsedElementKind::Reference, + .Reference = Reference, + .String = {}, + .Range = ReferenceRange}); + } + } + if (Advance()) + return true; + if (!Token.is(MITokenKind::Comma)) + break; + if (Advance()) + return true; + } + } + if (!Token.is(MITokenKind::RightBrace)) + return FailAtToken(Token, "expected end of metadata node"); + if (Advance()) + return true; + if (!Token.is(MITokenKind::EndOfFile)) + return FailAtToken(Token, "expected end of metadata node"); + if (!Slots.insert(Slot).second) + return FailAtToken(Token, "Metadata id is already used"); + + Definitions.push_back({.Node = DefinitionNode, + .Slot = Slot, + .IsDistinct = IsDistinct, + .Elements = std::move(Elements)}); + Context.Metadata.push_back( + {.TextID = Slot, .Reference = MachineMetadataReference::machineFunction(Slot)}); + } + + for (const Definition &Definition : Definitions) { + MachineMetadataNode NewNode{ + .Slot = Definition.Slot, + .IsDistinct = Definition.IsDistinct, + .Elements = {}, + }; + for (const ParsedElement &Element : Definition.Elements) { + if (Element.Kind == ParsedElementKind::Null) { + NewNode.Elements.push_back(MachineMetadataElement::null()); + continue; + } + if (Element.Kind == ParsedElementKind::String) { + NewNode.Elements.push_back(MachineMetadataElement::string(Element.String)); + continue; + } + + MachineMetadataReference Reference; + if (ActiveDocument->EmbeddedIR != std::nullopt && + ActiveDocument->EmbeddedIR->metadata(std::to_string(Element.Reference)) != nullptr) { + Reference = MachineMetadataReference::embeddedIR(Element.Reference); + } else if (Slots.contains(Element.Reference)) { + Reference = MachineMetadataReference::machineFunction(Element.Reference); + } else { + return fail(Element.Range, + "use of undefined metadata '!" + std::to_string(Element.Reference) + "'"); + } + auto Resolved = MachineMetadataElement::reference(Reference); + if (!Resolved) + return importError(Element.Range, Resolved.error()); + NewNode.Elements.push_back(std::move(*Resolved)); + } + auto AddResult = Function.metadata().addNode(std::move(NewNode)); + if (!AddResult) + return importError(Definition.Node->range(), AddResult.error()); + } + return false; +} + +} // namespace rcc::mir diff --git a/backend/lib/MIR/CMakeLists.txt b/backend/lib/MIR/CMakeLists.txt index f227390..cf5dbcd 100644 --- a/backend/lib/MIR/CMakeLists.txt +++ b/backend/lib/MIR/CMakeLists.txt @@ -4,30 +4,20 @@ set( RCCBackendMIRSources EmbeddedIR.cpp LLVMConstantSyntax.cpp - MIDebugSyntax.cpp - MILexer.cpp - MIRDataParser.cpp - MIRDocument.cpp - MIRDocumentParser.cpp - MIRDocumentReferenceValidation.cpp - MIRDocumentPrinter.cpp - MIRFrameParser.cpp - MIRFunctionParser.cpp - MIRFunctionRecordsParser.cpp - MIRMetadataParser.cpp - MIRMetadataTextParser.cpp - MIParser.cpp - MIPrinter.cpp + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MILexer.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MIParser.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MIRParser.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRPrinter.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRYamlMapping.cpp" ) set( RCCBackendMIRHeaders + "${RCC_BACKEND_SOURCE_DIR}/include/rcc/CodeGen/MIRParser/MIParser.h" + "${RCC_BACKEND_SOURCE_DIR}/include/rcc/CodeGen/MIRParser/MIRParser.h" + "${RCC_BACKEND_SOURCE_DIR}/include/rcc/CodeGen/MIRPrinter.h" + "${RCC_BACKEND_SOURCE_DIR}/include/rcc/CodeGen/MIRYamlMapping.h" "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/EmbeddedIR.h" "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/LLVMConstantSyntax.h" - "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/MIDebugSyntax.h" - "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/MILexer.h" - "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/MIRDocument.h" - "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/MIParser.h" - "${RCC_BACKEND_SOURCE_DIR}/include/rcc/MIR/MIPrinter.h" ) add_library(RCCBackendMIR ${RCCBackendMIRSources}) @@ -61,41 +51,20 @@ install( rcc_backend_register_format_files( "${CMAKE_CURRENT_SOURCE_DIR}/EmbeddedIR.cpp" "${CMAKE_CURRENT_SOURCE_DIR}/LLVMConstantSyntax.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIDebugSyntax.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MILexer.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDataParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocument.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentParser.h" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentReferenceValidation.h" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentReferenceValidation.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentPrinter.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRFrameParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRFunctionParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRFunctionRecordsParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRMetadataParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRMetadataTextParser.h" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRMetadataTextParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRParserUtilities.h" - "${CMAKE_CURRENT_SOURCE_DIR}/MIParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIPrinter.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MILexer.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MILexer.h" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MIParser.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MIRParser.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRPrinter.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRYamlMapping.cpp" ${RCCBackendMIRHeaders} ) rcc_backend_register_tidy_files( "${CMAKE_CURRENT_SOURCE_DIR}/EmbeddedIR.cpp" "${CMAKE_CURRENT_SOURCE_DIR}/LLVMConstantSyntax.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIDebugSyntax.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MILexer.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDataParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocument.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentReferenceValidation.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRDocumentPrinter.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRFrameParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRFunctionParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRFunctionRecordsParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRMetadataParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIRMetadataTextParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIParser.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/MIPrinter.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MILexer.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MIParser.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRParser/MIRParser.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRPrinter.cpp" + "${RCC_BACKEND_SOURCE_DIR}/lib/CodeGen/MIRYamlMapping.cpp" ) diff --git a/backend/lib/MIR/MIDebugSyntax.cpp b/backend/lib/MIR/MIDebugSyntax.cpp deleted file mode 100644 index e614a21..0000000 --- a/backend/lib/MIR/MIDebugSyntax.cpp +++ /dev/null @@ -1,264 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "rcc/MIR/MIDebugSyntax.h" - -#include <array> -#include <charconv> -#include <cstdint> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <system_error> -#include <utility> - -namespace rcc::mir { -namespace { - -[[nodiscard]] bool parseIndex(std::string_view Text, std::uint64_t &Value) noexcept { - if (Text.empty()) - return false; - const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, 10); - return Error == std::errc{} && End == Text.data() + Text.size(); -} - -[[nodiscard]] const auto &namedOpcodes() { - static constexpr auto Opcodes = std::to_array<std::pair<std::string_view, std::uint64_t>>({ - {"DW_OP_addr", 0x03}, - {"DW_OP_deref", 0x06}, - {"DW_OP_const1u", 0x08}, - {"DW_OP_const1s", 0x09}, - {"DW_OP_const2u", 0x0A}, - {"DW_OP_const2s", 0x0B}, - {"DW_OP_const4u", 0x0C}, - {"DW_OP_const4s", 0x0D}, - {"DW_OP_const8u", 0x0E}, - {"DW_OP_const8s", 0x0F}, - {"DW_OP_constu", 0x10}, - {"DW_OP_consts", 0x11}, - {"DW_OP_dup", 0x12}, - {"DW_OP_drop", 0x13}, - {"DW_OP_over", 0x14}, - {"DW_OP_pick", 0x15}, - {"DW_OP_swap", 0x16}, - {"DW_OP_rot", 0x17}, - {"DW_OP_xderef", 0x18}, - {"DW_OP_abs", 0x19}, - {"DW_OP_and", 0x1A}, - {"DW_OP_div", 0x1B}, - {"DW_OP_minus", 0x1C}, - {"DW_OP_mod", 0x1D}, - {"DW_OP_mul", 0x1E}, - {"DW_OP_neg", 0x1F}, - {"DW_OP_not", 0x20}, - {"DW_OP_or", 0x21}, - {"DW_OP_plus", 0x22}, - {"DW_OP_plus_uconst", 0x23}, - {"DW_OP_shl", 0x24}, - {"DW_OP_shr", 0x25}, - {"DW_OP_shra", 0x26}, - {"DW_OP_xor", 0x27}, - {"DW_OP_bra", 0x28}, - {"DW_OP_eq", 0x29}, - {"DW_OP_ge", 0x2A}, - {"DW_OP_gt", 0x2B}, - {"DW_OP_le", 0x2C}, - {"DW_OP_lt", 0x2D}, - {"DW_OP_ne", 0x2E}, - {"DW_OP_skip", 0x2F}, - {"DW_OP_regx", 0x90}, - {"DW_OP_fbreg", 0x91}, - {"DW_OP_bregx", 0x92}, - {"DW_OP_piece", 0x93}, - {"DW_OP_deref_size", 0x94}, - {"DW_OP_xderef_size", 0x95}, - {"DW_OP_nop", 0x96}, - {"DW_OP_push_object_address", 0x97}, - {"DW_OP_call2", 0x98}, - {"DW_OP_call4", 0x99}, - {"DW_OP_call_ref", 0x9A}, - {"DW_OP_form_tls_address", 0x9B}, - {"DW_OP_call_frame_cfa", 0x9C}, - {"DW_OP_bit_piece", 0x9D}, - {"DW_OP_implicit_value", 0x9E}, - {"DW_OP_stack_value", 0x9F}, - {"DW_OP_implicit_pointer", 0xA0}, - {"DW_OP_addrx", 0xA1}, - {"DW_OP_constx", 0xA2}, - {"DW_OP_entry_value", 0xA3}, - {"DW_OP_const_type", 0xA4}, - {"DW_OP_regval_type", 0xA5}, - {"DW_OP_deref_type", 0xA6}, - {"DW_OP_xderef_type", 0xA7}, - {"DW_OP_convert", 0xA8}, - {"DW_OP_reinterpret", 0xA9}, - {"DW_OP_LLVM_fragment", 0x1000}, - {"DW_OP_LLVM_convert", 0x1001}, - {"DW_OP_LLVM_tag_offset", 0x1002}, - {"DW_OP_LLVM_entry_value", 0x1003}, - {"DW_OP_LLVM_implicit_pointer", 0x1004}, - {"DW_OP_LLVM_arg", 0x1005}, - {"DW_OP_LLVM_extract_bits_sext", 0x1006}, - {"DW_OP_LLVM_extract_bits_zext", 0x1007}, - }); - return Opcodes; -} - -[[nodiscard]] const auto &attributeEncodings() { - static constexpr auto Encodings = std::to_array<std::pair<std::string_view, std::uint64_t>>({ - {"DW_ATE_address", 0x01}, - {"DW_ATE_boolean", 0x02}, - {"DW_ATE_complex_float", 0x03}, - {"DW_ATE_float", 0x04}, - {"DW_ATE_signed", 0x05}, - {"DW_ATE_signed_char", 0x06}, - {"DW_ATE_unsigned", 0x07}, - {"DW_ATE_unsigned_char", 0x08}, - {"DW_ATE_imaginary_float", 0x09}, - {"DW_ATE_packed_decimal", 0x0A}, - {"DW_ATE_numeric_string", 0x0B}, - {"DW_ATE_edited", 0x0C}, - {"DW_ATE_signed_fixed", 0x0D}, - {"DW_ATE_unsigned_fixed", 0x0E}, - {"DW_ATE_decimal_float", 0x0F}, - {"DW_ATE_UTF", 0x10}, - {"DW_ATE_UCS", 0x11}, - {"DW_ATE_ASCII", 0x12}, - }); - return Encodings; -} - -[[nodiscard]] std::optional<std::string> opcodeName(std::uint64_t Opcode) { - if (Opcode >= 0x30U && Opcode <= 0x4FU) - return "DW_OP_lit" + std::to_string(Opcode - 0x30U); - if (Opcode >= 0x50U && Opcode <= 0x6FU) - return "DW_OP_reg" + std::to_string(Opcode - 0x50U); - if (Opcode >= 0x70U && Opcode <= 0x8FU) - return "DW_OP_breg" + std::to_string(Opcode - 0x70U); - const auto &Opcodes = namedOpcodes(); - const auto Found = - std::ranges::find_if(Opcodes, [Opcode](const auto &Entry) { return Entry.second == Opcode; }); - if (Found == Opcodes.end()) - return std::nullopt; - return std::string(Found->first); -} - -[[nodiscard]] std::size_t opcodeSize(std::uint64_t Opcode) noexcept { - if (Opcode >= 0x70U && Opcode <= 0x8FU) - return 2; - switch (Opcode) { - case 0x92: - case 0x1000: - case 0x1001: - case 0x1006: - case 0x1007: - return 3; - case 0x10: - case 0x11: - case 0x23: - case 0x90: - case 0x94: - case 0x1002: - case 0x1003: - case 0x1005: - return 2; - default: - return 1; - } -} - -[[nodiscard]] std::optional<std::string_view> attributeName(std::uint64_t Encoding) noexcept { - const auto &Encodings = attributeEncodings(); - const auto Found = std::ranges::find_if( - Encodings, [Encoding](const auto &Entry) { return Entry.second == Encoding; }); - if (Found == Encodings.end()) - return std::nullopt; - return Found->first; -} - -} // namespace - -std::optional<std::uint64_t> parseDIExpressionOpcodeName(std::string_view Name) noexcept { - const auto NumberedOpcode = [Name](std::string_view Prefix, - std::uint64_t Base) -> std::optional<std::uint64_t> { - if (!Name.starts_with(Prefix)) - return std::nullopt; - std::uint64_t Index = 0; - if (!parseIndex(Name.substr(Prefix.size()), Index) || Index > 31) - return std::nullopt; - return Base + Index; - }; - if (auto Opcode = NumberedOpcode("DW_OP_lit", 0x30U)) - return Opcode; - if (auto Opcode = NumberedOpcode("DW_OP_reg", 0x50U)) - return Opcode; - if (auto Opcode = NumberedOpcode("DW_OP_breg", 0x70U)) - return Opcode; - - const auto &NamedOpcodes = namedOpcodes(); - const auto *const Found = - std::ranges::find_if(NamedOpcodes, [Name](const auto &Entry) { return Entry.first == Name; }); - if (Found == NamedOpcodes.end()) - return std::nullopt; - return Found->second; -} - -std::optional<std::uint64_t> -parseDIExpressionAttributeEncodingName(std::string_view Name) noexcept { - const auto &Encodings = attributeEncodings(); - const auto Found = - std::ranges::find_if(Encodings, [Name](const auto &Entry) { return Entry.first == Name; }); - if (Found == Encodings.end()) - return std::nullopt; - return Found->second; -} - -std::string invalidDIExpressionElementMessage(std::string_view Name) { - if (Name.starts_with("DW_OP_")) - return "invalid DWARF op '" + std::string(Name) + "'"; - if (Name.starts_with("DW_ATE_")) - return "invalid DWARF attribute encoding '" + std::string(Name) + "'"; - return "expected unsigned integer"; -} - -std::string formatDIExpression(std::span<const std::uint64_t> Elements) { - std::string Result = "!DIExpression("; - std::size_t Position = 0; - bool First = true; - while (Position < Elements.size()) { - const std::uint64_t Opcode = Elements[Position]; - const auto Name = opcodeName(Opcode); - const std::size_t Size = opcodeSize(Opcode); - if (!Name || Position + Size > Elements.size()) { - Result = "!DIExpression("; - for (std::size_t Index = 0; Index < Elements.size(); ++Index) { - if (Index != 0) - Result.append(", "); - Result.append(std::to_string(Elements[Index])); - } - Result.push_back(')'); - return Result; - } - - if (!First) - Result.append(", "); - First = false; - Result.append(*Name); - for (std::size_t Argument = 1; Argument < Size; ++Argument) { - Result.append(", "); - if (Opcode == 0x1001U && Argument == 2U) { - if (const auto Encoding = attributeName(Elements[Position + Argument])) { - Result.append(*Encoding); - continue; - } - } - Result.append(std::to_string(Elements[Position + Argument])); - } - Position += Size; - } - Result.push_back(')'); - return Result; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRDataParser.cpp b/backend/lib/MIR/MIRDataParser.cpp deleted file mode 100644 index 2ac1f2d..0000000 --- a/backend/lib/MIR/MIRDataParser.cpp +++ /dev/null @@ -1,287 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentParser.h" -#include "MIRDocumentReferenceValidation.h" -#include "MIRMetadataTextParser.h" -#include "MIRParserUtilities.h" -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionAttributes.h" -#include "rcc/CodeGen/MachineFunctionProperties.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/MachineRegisterInfo.h" -#include "rcc/CodeGen/MachineVerifier.h" -#include "rcc/CodeGen/TargetDescription.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/CodeGen/TargetRegisterInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" - -#include <algorithm> -#include <array> -#include <cstddef> -#include <cstdint> -#include <initializer_list> -#include <limits> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <unordered_map> -#include <unordered_set> -#include <utility> -#include <vector> - -namespace rcc::mir { - -using parser_detail::parseBool; -using parser_detail::parsePrefixedID; -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; -using parser_detail::trim; - -bool MIRDocumentParser::parseConstants(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context) { - const yaml::Node *Constants = mappingValue(Root, "constants"); - if (Constants == nullptr) - return false; - if (requireKind(*Constants, yaml::NodeKind::Sequence, "constant pool")) - return true; - constexpr std::array Keys{ - std::string_view{"id"}, - std::string_view{"value"}, - std::string_view{"alignment"}, - std::string_view{"isTargetSpecific"}, - }; - std::vector<LLVMNamedTypeDefinition> NamedTypes; - const std::string_view DataLayout = effectiveDataLayout(); - if (ActiveDocument->EmbeddedIR) { - const auto Definitions = ActiveDocument->EmbeddedIR->typeDefinitions(); - NamedTypes.reserve(Definitions.size()); - for (const EmbeddedIRTypeDefinition &Definition : Definitions) { - NamedTypes.push_back({.Name = Definition.Name, - .Definition = Definition.Definition, - .NumericName = Definition.NumericName}); - } - } - std::unordered_set<std::uint32_t> IDs; - for (const yaml::NodeID ConstantID : Constants->sequence()) { - const yaml::Node *Constant = node(ConstantID); - if (Constant == nullptr || - requireKind(*Constant, yaml::NodeKind::Mapping, "constant pool entry") || - validateKeys(*Constant, Keys, "constant pool entry")) { - return true; - } - const yaml::Node *IDNode = mappingValue(*Constant, "id"); - const yaml::Node *ValueNode = mappingValue(*Constant, "value"); - if (IDNode == nullptr || ValueNode == nullptr) - return fail(Constant->range(), "constant pool entry requires 'id' and 'value'"); - std::uint64_t WideID = 0; - if (unsignedValue(*IDNode, WideID, "a 32-bit constant pool ID") || - WideID > std::numeric_limits<std::uint32_t>::max()) { - return true; - } - const auto TextID = static_cast<std::uint32_t>(WideID); - if (!IDs.insert(TextID).second) - return fail(IDNode->range(), "duplicate constant pool ID"); - - bool IsTargetSpecific = false; - if (const yaml::Node *TargetNode = mappingValue(*Constant, "isTargetSpecific")) { - if (boolean(*TargetNode, IsTargetSpecific)) - return true; - } - if (IsTargetSpecific) { - std::string_view ValueText; - if (scalar(*ValueNode, ValueText, "target constant")) - return true; - auto TargetConstantValue = Target.mirText().parseTargetConstant(ValueText); - if (!TargetConstantValue) - return importError(ValueNode->range(), TargetConstantValue.error(), "mir-target-constant"); - auto Emission = TargetConstantValue->emit(); - if (!Emission) - return importError(ValueNode->range(), Emission.error(), "mir-target-constant"); - auto RequiredAlignment = TargetConstantValue->requiredAlignment(); - if (!RequiredAlignment) - return importError(ValueNode->range(), RequiredAlignment.error(), "mir-target-constant"); - - MachineConstantPoolEntryDesc Description; - Description.Size = Emission->Bytes.size(); - Description.Align = *RequiredAlignment; - Description.Value = std::move(*TargetConstantValue); - if (const yaml::Node *AlignmentNode = mappingValue(*Constant, "alignment")) { - if (alignment(*AlignmentNode, Description.Align, true)) - return true; - } - auto Added = Function.constantPool().addEntry(std::move(Description)); - if (!Added) - return importError(Constant->range(), Added.error()); - Context.ConstantPoolItems.push_back({.TextID = TextID, .Index = *Added}); - continue; - } - - std::string_view ValueText; - if (scalar(*ValueNode, ValueText, "LLVM constant")) - return true; - const LLVMConstantSyntaxOptions ConstantOptions{ - .DataLayout = DataLayout, - .NamedTypes = NamedTypes, - }; - auto Data = parseLLVMConstant(trim(ValueText), ConstantOptions); - if (!Data) - return importError(ValueNode->range(), Data.error(), "mir-llvm-constant"); - auto Size = llvmConstantStorageSize(Data->Type, ConstantOptions); - if (!Size) - return importError(ValueNode->range(), Size.error(), "mir-llvm-constant"); - - MachineConstantPoolEntryDesc Description; - Description.Size = *Size; - Description.Value = std::move(*Data); - if (const yaml::Node *AlignmentNode = mappingValue(*Constant, "alignment")) { - if (alignment(*AlignmentNode, Description.Align, true)) - return true; - } else { - auto Preferred = llvmConstantPreferredAlignment( - std::get<GenericConstantData>(Description.Value).Type, ConstantOptions); - if (!Preferred) - return importError(ValueNode->range(), Preferred.error(), "mir-llvm-constant"); - Description.Align = *Preferred; - } - auto Added = Function.constantPool().addEntry(std::move(Description)); - if (!Added) - return importError(Constant->range(), Added.error()); - Context.ConstantPoolItems.push_back({.TextID = TextID, .Index = *Added}); - } - return false; -} - -bool MIRDocumentParser::reserveJumpTables(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context, - std::vector<PendingJumpTable> &Pending) { - const yaml::Node *JumpTable = mappingValue(Root, "jumpTable"); - if (JumpTable == nullptr) - return false; - if (requireKind(*JumpTable, yaml::NodeKind::Mapping, "jump table")) - return true; - constexpr std::array JumpTableKeys{ - std::string_view{"kind"}, - std::string_view{"entries"}, - }; - if (validateKeys(*JumpTable, JumpTableKeys, "jump table")) - return true; - const yaml::Node *KindNode = mappingValue(*JumpTable, "kind"); - if (KindNode == nullptr) - return fail(JumpTable->range(), "jump table requires 'kind'"); - std::string_view KindName; - if (scalar(*KindNode, KindName, "jump table kind")) - return true; - - JumpTableEntryEncoding Encoding; - std::uint8_t EntrySize = 8; - if (KindName == "block-address") - Encoding = JumpTableEntryEncoding::BlockAddress; - else if (KindName == "gp-rel64-block-address") - Encoding = JumpTableEntryEncoding::GPRelative64BlockAddress; - else if (KindName == "gp-rel32-block-address") { - Encoding = JumpTableEntryEncoding::GPRelative32BlockAddress; - EntrySize = 4; - } else if (KindName == "label-difference32") { - Encoding = JumpTableEntryEncoding::LabelDifference32; - EntrySize = 4; - } else if (KindName == "label-difference64") - Encoding = JumpTableEntryEncoding::LabelDifference64; - else if (KindName == "inline") { - Encoding = JumpTableEntryEncoding::Inline; - EntrySize = 4; - } else if (KindName == "custom32") { - Encoding = JumpTableEntryEncoding::Custom32; - EntrySize = 4; - } else { - return fail(KindNode->range(), "unknown enumerated scalar"); - } - - const yaml::Node *Entries = mappingValue(*JumpTable, "entries"); - if (Entries == nullptr) - return false; - if (requireKind(*Entries, yaml::NodeKind::Sequence, "jump table entries")) - return true; - constexpr std::array EntryKeys{ - std::string_view{"id"}, - std::string_view{"blocks"}, - }; - std::unordered_set<std::uint32_t> IDs; - for (const yaml::NodeID EntryID : Entries->sequence()) { - const yaml::Node *Entry = node(EntryID); - if (Entry == nullptr || requireKind(*Entry, yaml::NodeKind::Mapping, "jump table entry") || - validateKeys(*Entry, EntryKeys, "jump table entry")) { - return true; - } - const yaml::Node *IDNode = mappingValue(*Entry, "id"); - if (IDNode == nullptr) - return fail(Entry->range(), "jump table entry requires 'id'"); - std::uint64_t WideID = 0; - if (unsignedValue(*IDNode, WideID, "a 32-bit jump table ID") || - WideID > std::numeric_limits<std::uint32_t>::max()) { - return true; - } - const auto TextID = static_cast<std::uint32_t>(WideID); - if (!IDs.insert(TextID).second) - return fail(IDNode->range(), "duplicate jump table ID"); - auto Reserved = Function.jumpTableInfo().reserveTable(); - if (!Reserved) - return importError(Entry->range(), Reserved.error()); - PendingJumpTable NewPending{ - .Index = *Reserved, - .TextID = TextID, - .Encoding = Encoding, - .EntrySize = EntrySize, - .Blocks = {}, - }; - if (const yaml::Node *Blocks = mappingValue(*Entry, "blocks")) { - if (requireKind(*Blocks, yaml::NodeKind::Sequence, "jump table blocks")) - return true; - for (const yaml::NodeID BlockID : Blocks->sequence()) { - const yaml::Node *BlockNode = node(BlockID); - std::string_view Text; - if (BlockNode == nullptr || scalar(*BlockNode, Text, "basic block reference")) - return true; - NewPending.Blocks.emplace_back(Text); - } - } - if (NewPending.Blocks.empty()) - return fail(Entry->range(), "jump table entry must contain at least one block"); - Context.JumpTables.push_back({.TextID = TextID, .Index = *Reserved}); - Pending.push_back(std::move(NewPending)); - } - return false; -} - -bool MIRDocumentParser::defineJumpTables(MachineFunction &Function, - std::span<const PendingJumpTable> Pending) { - for (const PendingJumpTable &Table : Pending) { - MachineJumpTableDesc Description; - Description.Encoding = Table.Encoding; - Description.EntrySize = Table.EntrySize; - Description.Align = *Alignment::create(Table.EntrySize); - for (const std::string &Text : Table.Blocks) { - MachineBasicBlock *Block = nullptr; - if (parseBlockReference(Text, Result.Source.range(0, 0), Function, Block)) - return true; - if (Block == nullptr) - return fail(Result.Source.range(0, 0), "reference to undefined machine basic block"); - Description.Targets.push_back(Block); - } - auto DefineResult = - Function.jumpTableInfo().defineReservedTable(Table.Index, std::move(Description)); - if (!DefineResult) - return importError(Result.Source.range(0, 0), DefineResult.error()); - } - return false; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRDocument.cpp b/backend/lib/MIR/MIRDocument.cpp deleted file mode 100644 index ad9ec40..0000000 --- a/backend/lib/MIR/MIRDocument.cpp +++ /dev/null @@ -1,29 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "rcc/MIR/MIRDocument.h" - -#include "MIRDocumentParser.h" - -#include <string> -#include <utility> - -namespace rcc::mir { - -MIRDocument::MIRDocument() = default; -MIRDocument::MIRDocument(CompilationUnit Unit) : Unit(std::move(Unit)) {} -MIRDocument::MIRDocument(MIRDocument &&) noexcept = default; -MIRDocument &MIRDocument::operator=(MIRDocument &&) noexcept = default; -MIRDocument::~MIRDocument() = default; - -MIRDocumentParseResult::MIRDocumentParseResult(MIRDocumentParseResult &&) noexcept = default; -MIRDocumentParseResult & -MIRDocumentParseResult::operator=(MIRDocumentParseResult &&) noexcept = default; -MIRDocumentParseResult::~MIRDocumentParseResult() = default; - -MIRDocumentParseResult parseMIRDocument(const TargetDescription &Target, std::string SourceName, - std::string SourceText, MIRDocumentParserOptions Options) { - return MIRDocumentParser(Target, std::move(SourceName), std::move(SourceText), std::move(Options)) - .run(); -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRDocumentParser.h b/backend/lib/MIR/MIRDocumentParser.h deleted file mode 100644 index 4d1d1ea..0000000 --- a/backend/lib/MIR/MIRDocumentParser.h +++ /dev/null @@ -1,132 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -// NOLINTNEXTLINE(llvm-header-guard) -#ifndef RCC_LIB_MIR_MIRDOCUMENTPARSER_H -#define RCC_LIB_MIR_MIRDOCUMENTPARSER_H - -#include "rcc/CodeGen/MachineDebugInfo.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" -#include "rcc/MIR/MIRDocument.h" - -#include <cstdint> -#include <span> -#include <string> -#include <string_view> -#include <utility> -#include <vector> - -namespace rcc { - -class MachineBasicBlock; -class MachineFunction; -class MachineInstr; - -namespace mir { - -struct PendingJumpTable final { - JumpTableIndex Index; - std::uint32_t TextID = 0; - JumpTableEntryEncoding Encoding = JumpTableEntryEncoding::BlockAddress; - std::uint8_t EntrySize = 8; - std::vector<std::string> Blocks; -}; - -class MIRDocumentParser final { -public: - MIRDocumentParser(const TargetDescription &Target, std::string SourceName, std::string SourceText, - MIRDocumentParserOptions Options) - : Target(Target), Options(std::move(Options)), - Result(std::move(SourceName), std::move(SourceText)) {} - - [[nodiscard]] MIRDocumentParseResult run(); - -private: - [[nodiscard]] bool fail(const TextRange &Range, std::string Message, - std::string Category = "mir-yaml"); - [[nodiscard]] bool importError(const TextRange &Range, const Error &Failure, - std::string Category = "mir-yaml"); - [[nodiscard]] const yaml::Node *node(yaml::NodeID ID) const noexcept { - return Stream->node(ID); - } - [[nodiscard]] const yaml::Node *mappingValue(const yaml::Node &Mapping, - std::string_view Key) const noexcept; - [[nodiscard]] bool requireKind(const yaml::Node &Node, yaml::NodeKind Kind, - std::string_view Description); - [[nodiscard]] bool validateKeys(const yaml::Node &Mapping, std::span<const std::string_view> Keys, - std::string_view Description); - [[nodiscard]] bool scalar(const yaml::Node &Node, std::string_view &Value, - std::string_view Description); - [[nodiscard]] bool boolean(const yaml::Node &Node, bool &Value); - [[nodiscard]] bool unsignedValue(const yaml::Node &Node, std::uint64_t &Value, - std::string_view Description); - [[nodiscard]] bool signedValue(const yaml::Node &Node, std::int64_t &Value, - std::string_view Description); - [[nodiscard]] bool alignment(const yaml::Node &Node, Alignment &Value, bool AllowZero); - [[nodiscard]] bool physicalRegister(const yaml::Node &Node, PhysicalRegister &Reg); - [[nodiscard]] bool physicalRegister(std::string_view Text, const TextRange &Range, - PhysicalRegister &Reg); - [[nodiscard]] bool virtualRegister(std::string_view Text, const TextRange &Range, - MachineFunction &Function, MIParserContext &Context, - Register &Reg); - [[nodiscard]] Expected<MachineMetadataReference> - resolveMetadataReference(const MachineFunction &Function, std::uint32_t Slot) const; - [[nodiscard]] bool parseVariableDebugInfo(const yaml::Node &Entry, MachineFunction &Function, - MachineVariableDebugInfo &DebugInfo); - [[nodiscard]] bool parseFunction(const yaml::Node &Root, MIRDocument &Document); - [[nodiscard]] bool parseFunctionScalars(const yaml::Node &Root, MachineFunction &Function, - MIRFunctionDocumentInfo &DocumentInfo); - [[nodiscard]] bool recomputeProperties(const yaml::Node &Root, MachineFunction &Function, - const MIRFunctionDocumentInfo &DocumentInfo); - [[nodiscard]] bool parseRegisters(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context); - [[nodiscard]] bool parseMetadata(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context); - [[nodiscard]] bool parseStackObjects(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context); - [[nodiscard]] bool parseConstants(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context); - [[nodiscard]] bool parseFrameState(const yaml::Node &Root, MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] bool reserveJumpTables(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context, - std::vector<PendingJumpTable> &Pending); - [[nodiscard]] bool defineJumpTables(MachineFunction &Function, - std::span<const PendingJumpTable> Pending); - [[nodiscard]] bool parseBody(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context); - [[nodiscard]] TextRange scalarDiagnosticRange(const yaml::Node &Scalar, - const TextRange &ScalarRange) const noexcept; - [[nodiscard]] bool parseFunctionRecords(const yaml::Node &Root, MachineFunction &Function); - [[nodiscard]] bool parseTargetState(const yaml::Node &Root, MachineFunction &Function); - [[nodiscard]] bool convertTargetNode(const yaml::Node &Source, MIRTargetNode &Destination); - [[nodiscard]] bool validateDocumentReferences(const MachineFunction &Function); - [[nodiscard]] std::string_view effectiveDataLayout() const noexcept; - [[nodiscard]] LLVMConstantSyntaxOptions pointerLayoutOptions() const noexcept; - - [[nodiscard]] MachineBasicBlock *blockAt(MachineFunction &Function, - std::uint32_t Number) const noexcept; - [[nodiscard]] MachineInstr *instructionAt(MachineFunction &Function, std::uint32_t Block, - std::uint32_t Offset) const noexcept; - [[nodiscard]] bool parseBlockReference(std::string_view Text, const TextRange &Range, - MachineFunction &Function, MachineBasicBlock *&Block); - [[nodiscard]] bool parseFrameReference(const yaml::Node &Node, const MIParserContext &Context, - FrameIndex &Index); - [[nodiscard]] bool parseSaveRestorePoints(const yaml::Node &Node, MachineFunction &Function, - std::vector<MachineFrameSaveRestorePoint> &Points); - - const TargetDescription &Target; - MIRDocumentParserOptions Options; - MIRDocumentParseResult Result; - const yaml::Stream *Stream = nullptr; - MIRDocument *ActiveDocument = nullptr; - MachineFunction *ActiveFunction = nullptr; -}; - -} // namespace mir -} // namespace rcc - -#endif // RCC_LIB_MIR_MIRDOCUMENTPARSER_H diff --git a/backend/lib/MIR/MIRDocumentPrinter.cpp b/backend/lib/MIR/MIRDocumentPrinter.cpp deleted file mode 100644 index 06998bb..0000000 --- a/backend/lib/MIR/MIRDocumentPrinter.cpp +++ /dev/null @@ -1,1499 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentReferenceValidation.h" -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineCompilationStatus.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionAttributes.h" -#include "rcc/CodeGen/MachineFunctionProperties.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/MachineRegisterInfo.h" -#include "rcc/CodeGen/TargetDescription.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/CodeGen/TargetRegisterInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" -#include "rcc/MIR/MIDebugSyntax.h" -#include "rcc/MIR/MIRDocument.h" - -#include <algorithm> -#include <array> -#include <charconv> -#include <cstddef> -#include <cstdint> -#include <limits> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <unordered_set> -#include <utility> -#include <variant> -#include <vector> - -namespace rcc::mir { -namespace { - -enum class YAMLQuoting : std::uint8_t { - None, - Single, - Double, -}; - -[[nodiscard]] bool isASCIIAlphanumeric(unsigned char Character) noexcept { - return (Character >= '0' && Character <= '9') || (Character >= 'A' && Character <= 'Z') || - (Character >= 'a' && Character <= 'z'); -} - -[[nodiscard]] bool isYAMLNumeric(std::string_view Value) noexcept { - if (Value.empty() || Value == "+" || Value == "-") - return false; - if (Value == ".nan" || Value == ".NaN" || Value == ".NAN") - return true; - - std::string_view Tail = Value; - if (Tail.front() == '+' || Tail.front() == '-') - Tail.remove_prefix(1); - if (Tail == ".inf" || Tail == ".Inf" || Tail == ".INF") - return true; - if (Value.starts_with("0o")) { - return Value.size() > 2U && - Value.substr(2).find_first_not_of("01234567") == std::string_view::npos; - } - if (Value.starts_with("0x")) { - return Value.size() > 2U && - Value.substr(2).find_first_not_of("0123456789abcdefABCDEF") == std::string_view::npos; - } - - if (Tail.empty() || Tail.front() == 'e' || Tail.front() == 'E') - return false; - std::size_t Position = 0; - if (Tail.front() == '.') { - ++Position; - if (Position == Tail.size() || Tail[Position] < '0' || Tail[Position] > '9') - return false; - while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') - ++Position; - } else { - while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') - ++Position; - if (Position == 0) - return false; - } - - if (Position < Tail.size() && Tail[Position] == '.') { - ++Position; - while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') - ++Position; - } - if (Position == Tail.size()) - return true; - if (Tail[Position] != 'e' && Tail[Position] != 'E') - return false; - ++Position; - if (Position < Tail.size() && (Tail[Position] == '+' || Tail[Position] == '-')) - ++Position; - const std::size_t ExponentBegin = Position; - while (Position < Tail.size() && Tail[Position] >= '0' && Tail[Position] <= '9') - ++Position; - return Position == Tail.size() && Position != ExponentBegin; -} - -[[nodiscard]] YAMLQuoting yamlQuoting(std::string_view Value) noexcept { - if (Value.empty()) - return YAMLQuoting::Single; - - YAMLQuoting Result = YAMLQuoting::None; - const auto IsSpace = [](unsigned char Character) { - return Character == ' ' || Character == '\t' || Character == '\r' || Character == '\n' || - Character == '\f' || Character == '\v'; - }; - if (IsSpace(static_cast<unsigned char>(Value.front())) || - IsSpace(static_cast<unsigned char>(Value.back())) || Value == "null" || Value == "Null" || - Value == "NULL" || Value == "~" || Value == "true" || Value == "True" || Value == "TRUE" || - Value == "false" || Value == "False" || Value == "FALSE" || isYAMLNumeric(Value)) { - Result = YAMLQuoting::Single; - } - - switch (Value.front()) { - case '-': - case '?': - case ':': - case '\\': - case ',': - case '[': - case ']': - case '{': - case '}': - case '#': - case '&': - case '*': - case '!': - case '|': - case '>': - case '\'': - case '"': - case '%': - case '@': - case '`': - Result = YAMLQuoting::Single; - break; - default: - break; - } - - for (const char RawCharacter : Value) { - const auto Character = static_cast<unsigned char>(RawCharacter); - if (isASCIIAlphanumeric(Character)) - continue; - switch (Character) { - case '_': - case '-': - case '^': - case '.': - case ',': - case ' ': - case '\t': - continue; - case '\n': - case '\r': - case 0x7f: - return YAMLQuoting::Double; - default: - if (Character <= 0x1fU || (Character & 0x80U) != 0) - return YAMLQuoting::Double; - Result = YAMLQuoting::Single; - break; - } - } - return Result; -} - -[[nodiscard]] std::string yamlScalar(std::string_view Value) { - const YAMLQuoting Quoting = yamlQuoting(Value); - if (Quoting == YAMLQuoting::None) - return std::string(Value); - - std::string Result; - Result.reserve(Value.size() + 2U); - const char Quote = Quoting == YAMLQuoting::Single ? '\'' : '"'; - Result.push_back(Quote); - for (const char Character : Value) { - if (Quoting == YAMLQuoting::Single) { - Result.push_back(Character); - if (Character == '\'') - Result.push_back('\''); - continue; - } - switch (Character) { - case '\\': - Result.append("\\\\"); - break; - case '"': - Result.append("\\\""); - break; - case '\0': - Result.append("\\0"); - break; - case '\a': - Result.append("\\a"); - break; - case '\b': - Result.append("\\b"); - break; - case '\t': - Result.append("\\t"); - break; - case '\n': - Result.append("\\n"); - break; - case '\v': - Result.append("\\v"); - break; - case '\f': - Result.append("\\f"); - break; - case '\r': - Result.append("\\r"); - break; - case '\x1b': - Result.append("\\e"); - break; - default: { - const auto Byte = static_cast<unsigned char>(Character); - if (Byte < 0x20U || Byte == 0x7fU) { - static constexpr std::string_view Hex = "0123456789ABCDEF"; - Result.append("\\x"); - Result.push_back(Hex[Byte >> 4U]); - Result.push_back(Hex[Byte & 0xfU]); - } else { - Result.push_back(Character); - } - break; - } - } - } - Result.push_back(Quote); - return Result; -} - -void appendBlockScalar(std::string &Output, std::string_view Text, std::size_t Indent) { - std::size_t Position = 0; - do { - Output.append(Indent, ' '); - const std::size_t End = Text.find('\n', Position); - if (End == std::string_view::npos) { - Output.append(Text.substr(Position)); - Output.push_back('\n'); - return; - } - Output.append(Text.substr(Position, End - Position)); - Output.push_back('\n'); - Position = End + 1U; - } while (Position < Text.size()); -} - -[[nodiscard]] std::optional<std::uint32_t> parseDecimal32(std::string_view Text) noexcept { - std::uint32_t Value = 0; - if (Text.empty()) - return std::nullopt; - const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, 10); - if (Error != std::errc{} || End != Text.data() + Text.size()) - return std::nullopt; - return Value; -} - -[[nodiscard]] bool isDefaultFrameState(const MachineFrameState &State) { - return State == MachineFrameState{}; -} - -class MIRDocumentPrinter final { -public: - MIRDocumentPrinter(const MIRDocument &Document, const TargetDescription &Target, - MIRDocumentPrinterOptions Options) - : Document(Document), Target(Target), Options(std::move(Options)) {} - - [[nodiscard]] Expected<std::string> run(); - -private: - struct InstructionLocation final { - std::uint32_t Block = 0; - std::uint32_t Offset = 0; - }; - - [[nodiscard]] Expected<void> printFunction(const MachineFunction &Function, - const MIRFunctionDocumentInfo *DocumentInfo); - [[nodiscard]] Expected<MIParserContext> createBodyContext(const MachineFunction &Function) const; - [[nodiscard]] Expected<void> printRegisters(const MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] Expected<void> printFrameInfo(const MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] Expected<void> printStackObjects(const MachineFunction &Function, - const MIParserContext &Context, bool Fixed); - [[nodiscard]] Expected<void> printConstants(const MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] Expected<void> printJumpTable(const MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] Expected<void> printMetadata(const MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] Expected<void> printTargetState(const MachineFunction &Function); - [[nodiscard]] Expected<void> printFunctionRecords(const MachineFunction &Function, - const MIParserContext &Context); - [[nodiscard]] Expected<void> verifyOmittedEmbeddedIRLayout(const MachineFunction &Function) const; - - [[nodiscard]] Expected<std::string> physicalRegister(PhysicalRegister Register) const; - [[nodiscard]] Expected<std::string> frameReference(FrameIndex Index, - const MIParserContext &Context) const; - [[nodiscard]] Expected<std::string> metadataReference(MachineMetadataReference Reference, - const MIParserContext &Context) const; - [[nodiscard]] Expected<std::string> metadataValue(const MachineMetadataValue &Value, - const MIParserContext &Context) const; - [[nodiscard]] Expected<std::string> debugLocation(const MachineDebugLocation &Location, - const MIParserContext &Context) const; - [[nodiscard]] Expected<std::string> diLocation(const MachineDILocation &Location, - std::size_t FrameIndex, - const MIParserContext &Context) const; - [[nodiscard]] Expected<InstructionLocation> instructionLocation(const MachineFunction &Function, - MachineInstrStableID ID) const; - [[nodiscard]] Expected<std::string> blockReference(const MachineFunction &Function, - MachineBlockStableID ID) const; - [[nodiscard]] Expected<std::string> targetNode(const MIRTargetNode &Node) const; - [[nodiscard]] const MIRFunctionDocumentInfo * - documentInfo(const MachineFunction &Function) const noexcept; - - void printBoolean(std::string_view Key, bool Value, bool Always = false); - void printUnsigned(std::string_view Key, std::uint64_t Value, bool Always = false); - void printSigned(std::string_view Key, std::int64_t Value, bool Always = false); - - const MIRDocument &Document; - const TargetDescription &Target; - MIRDocumentPrinterOptions Options; - std::string Output; -}; - -void MIRDocumentPrinter::printBoolean(std::string_view Key, bool Value, bool Always) { - if (!Always && Options.Mode == MIPrintMode::Readable && !Value) - return; - Output.append(Key); - Output.append(Value ? ": true\n" : ": false\n"); -} - -void MIRDocumentPrinter::printUnsigned(std::string_view Key, std::uint64_t Value, bool Always) { - if (!Always && Options.Mode == MIPrintMode::Readable && Value == 0) - return; - Output.append(Key); - Output.append(": "); - Output.append(std::to_string(Value)); - Output.push_back('\n'); -} - -void MIRDocumentPrinter::printSigned(std::string_view Key, std::int64_t Value, bool Always) { - if (!Always && Options.Mode == MIPrintMode::Readable && Value == 0) - return; - Output.append(Key); - Output.append(": "); - Output.append(std::to_string(Value)); - Output.push_back('\n'); -} - -Expected<std::string> MIRDocumentPrinter::physicalRegister(PhysicalRegister Register) const { - const auto Name = Target.registers().physicalRegisterMIRName(Register); - if (!Name) { - return makeUnexpected(ErrorCode::VerificationError, - "physical register has no canonical MIR name"); - } - return "$" + std::string(*Name); -} - -const MIRFunctionDocumentInfo * -MIRDocumentPrinter::documentInfo(const MachineFunction &Function) const noexcept { - const auto Infos = Document.functionDocumentInfo(); - const auto Found = std::ranges::find_if(Infos, [&Function](const MIRFunctionDocumentInfo &Info) { - return Info.Function == Function.stableID(); - }); - return Found == Infos.end() ? nullptr : &*Found; -} - -Expected<void> -MIRDocumentPrinter::verifyOmittedEmbeddedIRLayout(const MachineFunction &Function) const { - if (Options.EmitEmbeddedIR) - return {}; - - const EmbeddedIRReferenceIndex *Embedded = Document.embeddedIR(); - if (Embedded == nullptr) - return {}; - - bool HasLayoutDependentContent = false; - for (const VirtualRegisterInfo &RegisterInfo : Function.registerInfo().virtualRegisters()) { - if (!RegisterInfo.IsTombstone && RegisterInfo.Type && - RegisterInfo.Type->isPointerOrPointerVector()) { - HasLayoutDependentContent = true; - break; - } - } - if (!HasLayoutDependentContent) { - for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { - const MachineConstantPoolEntryDesc *Description = Entry.description(); - if (Description != nullptr && - std::holds_alternative<GenericConstantData>(Description->Value)) { - HasLayoutDependentContent = true; - break; - } - } - } - if (!HasLayoutDependentContent) - return {}; - - if (Options.DataLayout.empty()) { - return makeUnexpected( - ErrorCode::InvalidArgument, - "omitting embedded LLVM IR with layout-dependent MIR requires an explicit printer " - "DataLayout"); - } - - const LLVMConstantSyntaxOptions LayoutOptions{ - .DataLayout = Options.DataLayout, - .NamedTypes = {}, - }; - for (const VirtualRegisterInfo &RegisterInfo : Function.registerInfo().virtualRegisters()) { - if (RegisterInfo.IsTombstone || !RegisterInfo.Type || - !RegisterInfo.Type->isPointerOrPointerVector()) { - continue; - } - const auto AddressSpace = RegisterInfo.Type->addressSpace(); - if (!AddressSpace) { - return makeUnexpected(ErrorCode::VerificationError, - "pointer virtual register has no address space"); - } - auto PointerSize = llvmPointerSizeInBits(*AddressSpace, LayoutOptions); - if (!PointerSize) { - return makeUnexpected(ErrorCode::InvalidArgument, - "printer DataLayout cannot determine pointer width: " + - std::string(PointerSize.error().message())); - } - if (*PointerSize != RegisterInfo.Type->scalarBitWidth()) { - return makeUnexpected(ErrorCode::VerificationError, - "printer DataLayout does not match pointer virtual-register width"); - } - } - - for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { - const MachineConstantPoolEntryDesc *Description = Entry.description(); - if (Description == nullptr) - continue; - const auto *Generic = std::get_if<GenericConstantData>(&Description->Value); - if (Generic == nullptr) - continue; - auto StorageSize = llvmConstantStorageSize(Generic->Type, LayoutOptions); - if (!StorageSize) { - return makeUnexpected(ErrorCode::InvalidArgument, - "printer DataLayout cannot lay out generic constant: " + - std::string(StorageSize.error().message())); - } - if (*StorageSize != Description->Size) { - return makeUnexpected(ErrorCode::VerificationError, - "printer DataLayout does not match generic constant storage size"); - } - } - return {}; -} - -Expected<MIParserContext> -MIRDocumentPrinter::createBodyContext(const MachineFunction &Function) const { - MIParserContext Context; - Context.SyncScopes.assign(Document.syncScopes().begin(), Document.syncScopes().end()); - const auto VirtualRegisters = Function.registerInfo().virtualRegisters(); - for (std::size_t Index = 0; Index < VirtualRegisters.size(); ++Index) { - if (VirtualRegisters[Index].IsTombstone) - continue; - auto Reg = Register::makeVirtual(Index); - if (!Reg) - return std::unexpected(Reg.error()); - if (Index > std::numeric_limits<std::uint32_t>::max()) { - return makeUnexpected(ErrorCode::Unsupported, - "MIR cannot represent more than 2^32 virtual registers"); - } - Context.VirtualRegisters.push_back({.TextID = static_cast<std::uint32_t>(Index), .Reg = *Reg}); - } - - std::uint32_t StackID = 0; - std::uint32_t FixedID = 0; - for (const StackObject &Object : Function.frameInfo().objects()) { - const StackObjectDesc *Description = Object.description(); - if (Description == nullptr) - continue; - MIFrameObjectSlot Slot{ - .TextID = Object.isFixed() ? FixedID++ : StackID++, - .Index = Object.index(), - .Name = Description->AllocaName, - }; - (Object.isFixed() ? Context.FixedStackObjects : Context.StackObjects) - .push_back(std::move(Slot)); - } - - std::uint32_t ConstantID = 0; - for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { - if (Entry.description() != nullptr) - Context.ConstantPoolItems.push_back({.TextID = ConstantID++, .Index = Entry.index()}); - } - std::uint32_t JumpTableID = 0; - for (const MachineJumpTableEntry &Entry : Function.jumpTableInfo().tables()) { - if (Entry.description() != nullptr) - Context.JumpTables.push_back({.TextID = JumpTableID++, .Index = Entry.index()}); - } - - std::unordered_set<std::uint32_t> EmbeddedMetadataSlots; - if (const EmbeddedIRReferenceIndex *Embedded = Document.embeddedIR()) { - for (const EmbeddedIRMetadataDefinition &Definition : Embedded->metadataDefinitions()) { - const auto Slot = parseDecimal32(Definition.Name); - if (Slot) { - if (!EmbeddedMetadataSlots.insert(*Slot).second) { - return makeUnexpected(ErrorCode::VerificationError, - "embedded LLVM IR metadata slots are not unique"); - } - Context.Metadata.push_back( - {.TextID = *Slot, .Reference = MachineMetadataReference::embeddedIR(*Slot)}); - } - } - } - - std::unordered_set<std::uint32_t> UsedMetadataSlots = EmbeddedMetadataSlots; - for (const MachineMetadataNode &Node : Function.metadata().nodes()) { - if (!EmbeddedMetadataSlots.contains(Node.Slot)) - UsedMetadataSlots.insert(Node.Slot); - } - std::uint64_t NextMetadataSlot = 0; - for (const MachineMetadataNode &Node : Function.metadata().nodes()) { - std::uint32_t TextID = Node.Slot; - if (EmbeddedMetadataSlots.contains(TextID)) { - while (NextMetadataSlot <= std::numeric_limits<std::uint32_t>::max() && - UsedMetadataSlots.contains(static_cast<std::uint32_t>(NextMetadataSlot))) { - ++NextMetadataSlot; - } - if (NextMetadataSlot > std::numeric_limits<std::uint32_t>::max()) { - return makeUnexpected(ErrorCode::Unsupported, "MIR metadata text slot space is exhausted"); - } - TextID = static_cast<std::uint32_t>(NextMetadataSlot); - UsedMetadataSlots.insert(TextID); - ++NextMetadataSlot; - } - Context.Metadata.push_back( - {.TextID = TextID, .Reference = MachineMetadataReference::machineFunction(Node.Slot)}); - } - return Context; -} - -Expected<std::string> MIRDocumentPrinter::frameReference(FrameIndex Index, - const MIParserContext &Context) const { - const auto Stack = std::ranges::find_if( - Context.StackObjects, [Index](const MIFrameObjectSlot &Slot) { return Slot.Index == Index; }); - if (Stack != Context.StackObjects.end()) - return "%stack." + std::to_string(Stack->TextID); - const auto Fixed = - std::ranges::find_if(Context.FixedStackObjects, - [Index](const MIFrameObjectSlot &Slot) { return Slot.Index == Index; }); - if (Fixed != Context.FixedStackObjects.end()) - return "%fixed-stack." + std::to_string(Fixed->TextID); - return makeUnexpected(ErrorCode::VerificationError, - "frame state references an unknown stack object"); -} - -Expected<std::string> MIRDocumentPrinter::metadataReference(MachineMetadataReference Reference, - const MIParserContext &Context) const { - if (!Reference.isValid()) - return makeUnexpected(ErrorCode::VerificationError, "invalid metadata reference"); - const auto Slot = - std::ranges::find_if(Context.Metadata, [Reference](const MIMetadataSlot &Candidate) { - return Candidate.Reference == Reference; - }); - if (Slot == Context.Metadata.end()) { - return makeUnexpected(ErrorCode::VerificationError, "metadata reference has no MIR text slot"); - } - return "!" + std::to_string(Slot->TextID); -} - -Expected<std::string> MIRDocumentPrinter::metadataValue(const MachineMetadataValue &Value, - const MIParserContext &Context) const { - if (const MachineMetadataReference *Reference = Value.reference()) - return metadataReference(*Reference, Context); - const MachineDIExpression *Expression = Value.expression(); - if (Expression == nullptr) - return makeUnexpected(ErrorCode::VerificationError, "metadata value is not printable"); - return formatDIExpression(Expression->elements()); -} - -Expected<std::string> MIRDocumentPrinter::diLocation(const MachineDILocation &Location, - std::size_t FrameIndex, - const MIParserContext &Context) const { - if (FrameIndex >= Location.inlineChain().size()) { - return makeUnexpected(ErrorCode::VerificationError, - "inline DILocation has an invalid frame chain"); - } - const MachineDILocationFrame &Frame = Location.inlineChain()[FrameIndex]; - auto Scope = metadataReference(Frame.Scope, Context); - if (!Scope) - return std::unexpected(Scope.error()); - std::string Text = "!DILocation(line: " + std::to_string(Frame.Line); - if (Options.Mode == MIPrintMode::Complete || Frame.Column != 0) - Text.append(", column: " + std::to_string(Frame.Column)); - Text.append(", scope: " + *Scope); - if (FrameIndex + 1U < Location.inlineChain().size()) { - auto Nested = diLocation(Location, FrameIndex + 1U, Context); - if (!Nested) - return std::unexpected(Nested.error()); - Text.append(", inlinedAt: " + *Nested); - } else if (Location.terminalInlinedAt()) { - auto Terminal = metadataReference(*Location.terminalInlinedAt(), Context); - if (!Terminal) - return std::unexpected(Terminal.error()); - Text.append(", inlinedAt: " + *Terminal); - } - if (Options.Mode == MIPrintMode::Complete || Frame.IsImplicitCode) - Text.append(Frame.IsImplicitCode ? ", isImplicitCode: true" : ", isImplicitCode: false"); - if (Options.Mode == MIPrintMode::Complete || Frame.AtomGroup != 0) - Text.append(", atomGroup: " + std::to_string(Frame.AtomGroup)); - if (Options.Mode == MIPrintMode::Complete || Frame.AtomRank != 0) - Text.append(", atomRank: " + std::to_string(Frame.AtomRank)); - Text.push_back(')'); - return Text; -} - -Expected<std::string> MIRDocumentPrinter::debugLocation(const MachineDebugLocation &Location, - const MIParserContext &Context) const { - if (const MachineMetadataReference *Reference = Location.reference()) - return metadataReference(*Reference, Context); - if (const MachineDILocation *Inline = Location.inlineLocation()) - return diLocation(*Inline, 0, Context); - return makeUnexpected(ErrorCode::VerificationError, "debug location is not printable"); -} - -Expected<MIRDocumentPrinter::InstructionLocation> -MIRDocumentPrinter::instructionLocation(const MachineFunction &Function, - MachineInstrStableID ID) const { - std::uint64_t BlockNumber = 0; - for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; - Block = Block->next(), ++BlockNumber) { - if (BlockNumber > std::numeric_limits<std::uint32_t>::max()) - break; - std::uint64_t Offset = 0; - for (const MachineInstr *Instruction = Block->firstInstruction(); Instruction != nullptr; - Instruction = Instruction->next(), ++Offset) { - if (Instruction->stableID() != ID) - continue; - if (Offset > std::numeric_limits<std::uint32_t>::max()) { - return makeUnexpected(ErrorCode::Unsupported, "MIR instruction offset exceeds 32 bits"); - } - return InstructionLocation{static_cast<std::uint32_t>(BlockNumber), - static_cast<std::uint32_t>(Offset)}; - } - } - return makeUnexpected(ErrorCode::VerificationError, - "function record references an unlinked instruction"); -} - -Expected<std::string> MIRDocumentPrinter::blockReference(const MachineFunction &Function, - MachineBlockStableID ID) const { - std::uint64_t Number = 0; - for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; - Block = Block->next(), ++Number) { - if (Block->stableID() == ID) { - if (Number > std::numeric_limits<std::uint32_t>::max()) - break; - return "%bb." + std::to_string(Number); - } - } - return makeUnexpected(ErrorCode::VerificationError, - "frame record references an unlinked basic block"); -} - -Expected<std::string> MIRDocumentPrinter::targetNode(const MIRTargetNode &Node) const { - switch (Node.Kind) { - case MIRTargetNodeKind::Null: - return std::string("null"); - case MIRTargetNodeKind::Scalar: - return yamlScalar(Node.Scalar); - case MIRTargetNodeKind::Sequence: { - std::string Text = "[ "; - for (std::size_t Index = 0; Index < Node.Sequence.size(); ++Index) { - if (Index != 0) - Text.append(", "); - auto Element = targetNode(Node.Sequence[Index]); - if (!Element) - return std::unexpected(Element.error()); - Text.append(*Element); - } - Text.append(" ]"); - return Text; - } - case MIRTargetNodeKind::Mapping: { - std::string Text = "{ "; - for (std::size_t Index = 0; Index < Node.Mapping.size(); ++Index) { - if (Index != 0) - Text.append(", "); - Text.append(yamlScalar(Node.Mapping[Index].Key)); - Text.append(": "); - auto Value = targetNode(Node.Mapping[Index].Value); - if (!Value) - return std::unexpected(Value.error()); - Text.append(*Value); - } - Text.append(" }"); - return Text; - } - case MIRTargetNodeKind::Count: - break; - } - return makeUnexpected(ErrorCode::VerificationError, "target MIR node has an invalid kind"); -} - -Expected<void> MIRDocumentPrinter::printRegisters(const MachineFunction &Function, - const MIParserContext &Context) { - if (Context.VirtualRegisters.empty()) { - if (Options.Mode == MIPrintMode::Complete) - Output.append("registers: []\n"); - } else { - Output.append("registers:\n"); - for (const MIVirtualRegisterSlot &Slot : Context.VirtualRegisters) { - auto InfoResult = Function.registerInfo().virtualRegisterInfo(Slot.Reg); - if (!InfoResult) - return std::unexpected(InfoResult.error()); - const VirtualRegisterInfo &Info = InfoResult->get(); - std::string Class = "_"; - if (Info.RegClass) { - const auto Name = Target.registers().registerClassMIRName(*Info.RegClass); - if (!Name) - return makeUnexpected(ErrorCode::VerificationError, - "virtual register class has no canonical MIR name"); - Class = *Name; - } else if (Info.Bank) { - const auto Name = Target.registers().registerBankMIRName(*Info.Bank); - if (!Name) - return makeUnexpected(ErrorCode::VerificationError, - "virtual register bank has no canonical MIR name"); - Class = *Name; - } - if (Info.AllocationHints.size() > 1) { - return makeUnexpected( - ErrorCode::Unsupported, - "fixed LLVM MIR can serialize only one simple preferred-register hint"); - } - - Output.append(" - { id: "); - Output.append(std::to_string(Slot.TextID)); - Output.append(", class: "); - Output.append(yamlScalar(Class)); - if (!Info.AllocationHints.empty()) { - auto Preferred = physicalRegister(Info.AllocationHints.front()); - if (!Preferred) - return std::unexpected(Preferred.error()); - Output.append(", preferred-register: "); - Output.append(yamlScalar(*Preferred)); - } - if (Info.TargetFlags != 0) { - auto Names = Target.mirText().virtualRegisterFlagNames(Info.TargetFlags); - if (!Names) - return std::unexpected(Names.error()); - Output.append(", flags: [ "); - for (std::size_t Index = 0; Index < Names->size(); ++Index) { - if (Index != 0) - Output.append(", "); - Output.append(yamlScalar((*Names)[Index])); - } - Output.append(" ]"); - } - Output.append(" }\n"); - } - } - - if (Function.registerInfo().liveIns().empty()) { - if (Options.Mode == MIPrintMode::Complete) - Output.append("liveins: []\n"); - } else { - Output.append("liveins:\n"); - for (const PhysicalRegisterLiveIn &LiveIn : Function.registerInfo().liveIns()) { - auto Name = physicalRegister(LiveIn.PhysReg); - if (!Name) - return std::unexpected(Name.error()); - Output.append(" - { reg: "); - Output.append(yamlScalar(*Name)); - if (LiveIn.InitialVirtualRegister) { - const auto Slot = std::ranges::find_if( - Context.VirtualRegisters, [&LiveIn](const MIVirtualRegisterSlot &Candidate) { - return Candidate.Reg == *LiveIn.InitialVirtualRegister; - }); - if (Slot == Context.VirtualRegisters.end()) { - return makeUnexpected(ErrorCode::VerificationError, - "live-in references an unknown virtual register"); - } - Output.append(", virtual-reg: "); - Output.append(yamlScalar("%" + std::to_string(Slot->TextID))); - } - Output.append(" }\n"); - } - } - - if (Function.records().hasCalleeSavedRegisters()) { - Output.append("calleeSavedRegisters: [ "); - const auto Registers = Function.records().calleeSavedRegisters(); - for (std::size_t Index = 0; Index < Registers.size(); ++Index) { - if (Index != 0) - Output.append(", "); - auto Name = physicalRegister(Registers[Index]); - if (!Name) - return std::unexpected(Name.error()); - Output.append(yamlScalar(*Name)); - } - Output.append(" ]\n"); - } - return {}; -} - -Expected<void> MIRDocumentPrinter::printFrameInfo(const MachineFunction &Function, - const MIParserContext &Context) { - const MachineFrameState &State = Function.frameInfo().state(); - if (Options.Mode == MIPrintMode::Readable && isDefaultFrameState(State)) - return {}; - if (State.MaxCallFrameSize && - *State.MaxCallFrameSize > std::numeric_limits<std::uint32_t>::max()) { - return makeUnexpected(ErrorCode::Unsupported, - "fixed LLVM MIR maxCallFrameSize is limited to 32 bits"); - } - Output.append("frameInfo:\n"); - const auto Bool = [this](std::string_view Key, bool Value) { - if (Options.Mode == MIPrintMode::Readable && !Value) - return; - Output.append(" "); - Output.append(Key); - Output.append(Value ? ": true\n" : ": false\n"); - }; - const auto Unsigned = [this](std::string_view Key, std::uint64_t Value) { - if (Options.Mode == MIPrintMode::Readable && Value == 0) - return; - Output.append(" "); - Output.append(Key); - Output.append(": "); - Output.append(std::to_string(Value)); - Output.push_back('\n'); - }; - - Bool("isFrameAddressTaken", State.IsFrameAddressTaken); - Bool("isReturnAddressTaken", State.IsReturnAddressTaken); - Bool("hasStackMap", State.HasStackMap); - Bool("hasPatchPoint", State.HasPatchPoint); - Unsigned("stackSize", State.StackSize); - if (Options.Mode == MIPrintMode::Complete || State.OffsetAdjustment != 0) { - Output.append(" offsetAdjustment: "); - Output.append(std::to_string(State.OffsetAdjustment)); - Output.push_back('\n'); - } - Unsigned("maxAlignment", State.MaxAlignment ? State.MaxAlignment->bytes() : 0); - Bool("adjustsStack", State.AdjustsStack); - Bool("hasCalls", State.HasCalls); - if (Options.Mode == MIPrintMode::Complete || - State.FramePointerPolicy != MachineFramePointerPolicy::None) { - Output.append(" framePointerPolicy: "); - Output.append(getMachineFramePointerPolicyName(State.FramePointerPolicy)); - Output.push_back('\n'); - } - if (Options.Mode == MIPrintMode::Complete || State.StackProtector) { - std::string Reference; - if (State.StackProtector) { - auto PrintedReference = frameReference(*State.StackProtector, Context); - if (!PrintedReference) - return std::unexpected(PrintedReference.error()); - Reference = std::move(*PrintedReference); - } - Output.append(" stackProtector: "); - Output.append(yamlScalar(Reference)); - Output.push_back('\n'); - } - if (Options.Mode == MIPrintMode::Complete || State.FunctionContext) { - std::string Reference; - if (State.FunctionContext) { - auto PrintedReference = frameReference(*State.FunctionContext, Context); - if (!PrintedReference) - return std::unexpected(PrintedReference.error()); - Reference = std::move(*PrintedReference); - } - Output.append(" functionContext: "); - Output.append(yamlScalar(Reference)); - Output.push_back('\n'); - } - if (Options.Mode == MIPrintMode::Complete || State.MaxCallFrameSize) { - Unsigned("maxCallFrameSize", - State.MaxCallFrameSize.value_or(std::numeric_limits<std::uint32_t>::max())); - } - Unsigned("cvBytesOfCalleeSavedRegisters", State.CVBytesOfCalleeSavedRegisters); - Bool("hasOpaqueSPAdjustment", State.HasOpaqueSPAdjustment); - Bool("hasVAStart", State.HasVAStart); - Bool("hasMustTailInVarArgFunc", State.HasMustTailInVarArgFunc); - Bool("hasTailCall", State.HasTailCall); - Bool("isCalleeSavedInfoValid", State.IsCalleeSavedInfoValid); - Unsigned("localFrameSize", State.LocalFrameSize); - - const auto PrintPoints = - [this, &Function](std::string_view Key, - std::span<const MachineFrameSaveRestorePoint> Points) -> Expected<void> { - if (Points.empty()) - return {}; - Output.append(" "); - Output.append(Key); - Output.append(":\n"); - for (const MachineFrameSaveRestorePoint &Point : Points) { - auto Block = blockReference(Function, Point.Block); - if (!Block) - return std::unexpected(Block.error()); - Output.append(" - { point: "); - Output.append(yamlScalar(*Block)); - if (!Point.Registers.empty()) { - Output.append(", registers: [ "); - for (std::size_t Index = 0; Index < Point.Registers.size(); ++Index) { - if (Index != 0) - Output.append(", "); - auto Name = physicalRegister(Point.Registers[Index]); - if (!Name) - return std::unexpected(Name.error()); - Output.append(yamlScalar(*Name)); - } - Output.append(" ]"); - } - Output.append(" }\n"); - } - return {}; - }; - if (auto Printed = PrintPoints("savePoint", State.SavePoints); !Printed) - return Printed; - return PrintPoints("restorePoint", State.RestorePoints); -} - -Expected<void> MIRDocumentPrinter::printStackObjects(const MachineFunction &Function, - const MIParserContext &Context, bool Fixed) { - const auto &Slots = Fixed ? Context.FixedStackObjects : Context.StackObjects; - if (Slots.empty()) { - if (Options.Mode == MIPrintMode::Complete) - Output.append(Fixed ? "fixedStack: []\n" : "stack: []\n"); - return {}; - } - Output.append(Fixed ? "fixedStack:\n" : "stack:\n"); - for (const MIFrameObjectSlot &Slot : Slots) { - const StackObject *Object = Function.frameInfo().object(Slot.Index); - const StackObjectDesc *Description = Object == nullptr ? nullptr : Object->description(); - if (Description == nullptr) - return makeUnexpected(ErrorCode::VerificationError, "MIR stack slot is a tombstone"); - if (Description->IsDead || Description->AddressTaken || !Description->Lifetime.empty() || - Description->TargetFlags != 0) { - return makeUnexpected(ErrorCode::Unsupported, - "stack object contains state not represented by fixed LLVM MIR"); - } - - std::string_view Type; - switch (Description->Kind) { - case StackObjectKind::Default: - case StackObjectKind::FixedABI: - Type = "default"; - break; - case StackObjectKind::FixedSpillSlot: - case StackObjectKind::SpillSlot: - Type = "spill-slot"; - break; - case StackObjectKind::VariableSized: - Type = "variable-sized"; - break; - case StackObjectKind::DynamicAlloca: - case StackObjectKind::CalleeSaved: - case StackObjectKind::EmergencySpill: - case StackObjectKind::Target: - case StackObjectKind::Count: - return makeUnexpected(ErrorCode::Unsupported, - "stack object kind has no fixed LLVM MIR representation"); - } - - Output.append(" - { id: "); - Output.append(std::to_string(Slot.TextID)); - if (!Fixed && Description->AllocaName) { - Output.append(", name: "); - Output.append(yamlScalar(*Description->AllocaName)); - } - if (Options.Mode == MIPrintMode::Complete || Type != "default") { - Output.append(", type: "); - Output.append(Type); - } - if (Options.Mode == MIPrintMode::Complete || Description->Offset.value_or(0) != 0) { - Output.append(", offset: "); - Output.append(std::to_string(Description->Offset.value_or(0))); - } - if (Description->Kind != StackObjectKind::VariableSized) { - Output.append(", size: "); - Output.append(std::to_string(Description->Size.value_or(0))); - } - if (Options.Mode == MIPrintMode::Complete || Description->Align.bytes() != 1) { - Output.append(", alignment: "); - Output.append(std::to_string(Description->Align.bytes())); - } - if (Options.Mode == MIPrintMode::Complete || Description->Stack != StackID::defaultStack()) { - Output.append(", stack-id: "); - Output.append(getStackIDName(Description->Stack)); - } - if (Fixed && Description->Kind != StackObjectKind::FixedSpillSlot) { - if (Options.Mode == MIPrintMode::Complete || Description->IsImmutable) - Output.append(Description->IsImmutable ? ", isImmutable: true" : ", isImmutable: false"); - if (Options.Mode == MIPrintMode::Complete || Description->IsAliased) - Output.append(Description->IsAliased ? ", isAliased: true" : ", isAliased: false"); - } else if (!Fixed && (Description->IsImmutable || Description->IsAliased)) { - return makeUnexpected(ErrorCode::Unsupported, - "ordinary stack object immutable/aliased state is not serialized"); - } - if (Description->CalleeSavedRegister) { - auto Name = physicalRegister(*Description->CalleeSavedRegister); - if (!Name) - return std::unexpected(Name.error()); - Output.append(", callee-saved-register: "); - Output.append(yamlScalar(*Name)); - } - if (Options.Mode == MIPrintMode::Complete || !Description->CalleeSavedRestored) { - Output.append(Description->CalleeSavedRestored ? ", callee-saved-restored: true" - : ", callee-saved-restored: false"); - } - if (!Fixed && Description->LocalOffset) { - Output.append(", local-offset: "); - Output.append(std::to_string(*Description->LocalOffset)); - } - if (Description->DebugInfo) { - const MachineVariableDebugInfo &Debug = *Description->DebugInfo; - if (Debug.Variable) { - auto Value = metadataReference(*Debug.Variable, Context); - if (!Value) - return std::unexpected(Value.error()); - Output.append(", debug-info-variable: "); - Output.append(yamlScalar(*Value)); - } - if (Debug.Expression) { - auto Value = metadataValue(*Debug.Expression, Context); - if (!Value) - return std::unexpected(Value.error()); - Output.append(", debug-info-expression: "); - Output.append(yamlScalar(*Value)); - } - if (Debug.Location) { - auto Value = debugLocation(*Debug.Location, Context); - if (!Value) - return std::unexpected(Value.error()); - Output.append(", debug-info-location: "); - Output.append(yamlScalar(*Value)); - } - } - Output.append(" }\n"); - } - return {}; -} - -Expected<void> MIRDocumentPrinter::printConstants(const MachineFunction &Function, - const MIParserContext &Context) { - if (Context.ConstantPoolItems.empty()) { - if (Options.Mode == MIPrintMode::Complete) - Output.append("constants: []\n"); - return {}; - } - Output.append("constants:\n"); - for (const MIConstantPoolSlot &Slot : Context.ConstantPoolItems) { - const MachineConstantPoolEntry *Entry = Function.constantPool().entry(Slot.Index); - const MachineConstantPoolEntryDesc *Description = - Entry == nullptr ? nullptr : Entry->description(); - if (Description == nullptr) - return makeUnexpected(ErrorCode::VerificationError, "constant-pool slot is a tombstone"); - if (Description->Merge != ConstantPoolMergeKind::Never || - Description->Relocation != ConstantPoolRelocationMode::None || - Description->Placement != ConstantPoolPlacement{}) { - return makeUnexpected(ErrorCode::Unsupported, - "constant-pool placement state is not represented by fixed LLVM MIR"); - } - - bool IsTargetSpecific = false; - Expected<std::string> Value = - makeUnexpected(ErrorCode::Unsupported, "constant-pool kind is not MIR serializable"); - if (const auto *Generic = std::get_if<GenericConstantData>(&Description->Value)) { - Value = printLLVMConstant(*Generic); - } else if (const auto *TargetConstantValue = std::get_if<TargetConstant>(&Description->Value)) { - IsTargetSpecific = true; - Value = Target.mirText().printTargetConstant(*TargetConstantValue); - } - if (!Value) - return std::unexpected(Value.error()); - - Output.append(" - id: "); - Output.append(std::to_string(Slot.TextID)); - Output.append("\n value: "); - Output.append(yamlScalar(*Value)); - Output.append("\n alignment: "); - Output.append(std::to_string(Description->Align.bytes())); - Output.push_back('\n'); - if (IsTargetSpecific) - Output.append(" isTargetSpecific: true\n"); - } - return {}; -} - -Expected<void> MIRDocumentPrinter::printJumpTable(const MachineFunction &Function, - const MIParserContext &Context) { - if (Context.JumpTables.empty()) - return {}; - const MachineJumpTableEntry *FirstEntry = - Function.jumpTableInfo().table(Context.JumpTables.front().Index); - const MachineJumpTableDesc *First = FirstEntry == nullptr ? nullptr : FirstEntry->description(); - if (First == nullptr) - return makeUnexpected(ErrorCode::VerificationError, "jump-table slot is not defined"); - - std::string_view Kind; - std::uint8_t ExpectedSize = 8; - switch (First->Encoding) { - case JumpTableEntryEncoding::BlockAddress: - Kind = "block-address"; - break; - case JumpTableEntryEncoding::GPRelative64BlockAddress: - Kind = "gp-rel64-block-address"; - break; - case JumpTableEntryEncoding::GPRelative32BlockAddress: - Kind = "gp-rel32-block-address"; - ExpectedSize = 4; - break; - case JumpTableEntryEncoding::LabelDifference32: - Kind = "label-difference32"; - ExpectedSize = 4; - break; - case JumpTableEntryEncoding::LabelDifference64: - Kind = "label-difference64"; - break; - case JumpTableEntryEncoding::Inline: - Kind = "inline"; - ExpectedSize = 4; - break; - case JumpTableEntryEncoding::Custom32: - Kind = "custom32"; - ExpectedSize = 4; - break; - case JumpTableEntryEncoding::Target: - case JumpTableEntryEncoding::Count: - return makeUnexpected(ErrorCode::Unsupported, - "target jump-table encoding has no fixed LLVM MIR spelling"); - } - - Output.append("jumpTable:\n kind: "); - Output.append(Kind); - Output.append("\n entries:\n"); - for (const MIJumpTableSlot &Slot : Context.JumpTables) { - const MachineJumpTableEntry *Entry = Function.jumpTableInfo().table(Slot.Index); - const MachineJumpTableDesc *Description = Entry == nullptr ? nullptr : Entry->description(); - if (Description == nullptr) - return makeUnexpected(ErrorCode::VerificationError, "jump-table slot is not defined"); - if (Description->Encoding != First->Encoding || Description->EntrySize != ExpectedSize || - Description->Align.bytes() != ExpectedSize || - Description->Placement != JumpTablePlacement{} || Description->TargetEncoding != 0) { - return makeUnexpected(ErrorCode::Unsupported, - "jump-table entries cannot share one fixed LLVM MIR kind"); - } - Output.append(" - { id: "); - Output.append(std::to_string(Slot.TextID)); - Output.append(", blocks: [ "); - for (std::size_t Index = 0; Index < Description->Targets.size(); ++Index) { - if (Index != 0) - Output.append(", "); - if (Description->Targets[Index] == nullptr) - return makeUnexpected(ErrorCode::VerificationError, - "jump-table entry contains a null block"); - auto Block = blockReference(Function, Description->Targets[Index]->stableID()); - if (!Block) - return std::unexpected(Block.error()); - Output.append(yamlScalar(*Block)); - } - Output.append(" ] }\n"); - } - return {}; -} - -Expected<void> MIRDocumentPrinter::printMetadata(const MachineFunction &Function, - const MIParserContext &Context) { - if (Function.metadata().nodes().empty()) - return {}; - Output.append("machineMetadataNodes:\n"); - for (const MachineMetadataNode &Node : Function.metadata().nodes()) { - auto NodeReference = - metadataReference(MachineMetadataReference::machineFunction(Node.Slot), Context); - if (!NodeReference) - return std::unexpected(NodeReference.error()); - std::string Text = *NodeReference + " = "; - if (Node.IsDistinct) - Text.append("distinct "); - Text.append("!{"); - for (std::size_t Index = 0; Index < Node.Elements.size(); ++Index) { - if (Index != 0) - Text.append(", "); - const MachineMetadataElement &Element = Node.Elements[Index]; - if (Element.isNull()) { - Text.append("null"); - } else if (const MachineMetadataReference *Reference = Element.reference()) { - auto Printed = metadataReference(*Reference, Context); - if (!Printed) - return std::unexpected(Printed.error()); - Text.append(*Printed); - } else if (const std::string *String = Element.string()) { - Text.append("!\""); - for (const char Character : *String) { - if (Character == '\\' || Character == '"') - Text.push_back('\\'); - Text.push_back(Character); - } - Text.push_back('"'); - } else { - return makeUnexpected(ErrorCode::VerificationError, - "machine metadata tuple contains an invalid element"); - } - } - Text.push_back('}'); - Output.append(" - "); - Output.append(yamlScalar(Text)); - Output.push_back('\n'); - } - return {}; -} - -Expected<void> MIRDocumentPrinter::printTargetState(const MachineFunction &Function) { - if (!Function.targetState().hasValue()) - return {}; - auto Node = Target.mirText().printFunctionState(Function.targetState()); - if (!Node) - return std::unexpected(Node.error()); - auto Text = targetNode(*Node); - if (!Text) - return std::unexpected(Text.error()); - Output.append("machineFunctionInfo: "); - Output.append(*Text); - Output.push_back('\n'); - return {}; -} - -Expected<void> MIRDocumentPrinter::printFunctionRecords(const MachineFunction &Function, - const MIParserContext &Context) { - const MachineFunctionRecords &Records = Function.records(); - if (!Records.entryValues().empty()) { - Output.append("entry_values:\n"); - for (const MachineEntryValueInfo &Entry : Records.entryValues()) { - auto RegisterName = physicalRegister(Entry.Reg); - if (!RegisterName) - return std::unexpected(RegisterName.error()); - if (!Entry.DebugInfo.Variable || !Entry.DebugInfo.Expression || !Entry.DebugInfo.Location) { - return makeUnexpected(ErrorCode::Unsupported, - "MIR entry value requires complete debug metadata"); - } - auto Variable = metadataReference(*Entry.DebugInfo.Variable, Context); - auto Expression = metadataValue(*Entry.DebugInfo.Expression, Context); - auto Location = debugLocation(*Entry.DebugInfo.Location, Context); - if (!Variable) - return std::unexpected(Variable.error()); - if (!Expression) - return std::unexpected(Expression.error()); - if (!Location) - return std::unexpected(Location.error()); - Output.append(" - { entry-value-register: "); - Output.append(yamlScalar(*RegisterName)); - Output.append(", debug-info-variable: "); - Output.append(yamlScalar(*Variable)); - Output.append(", debug-info-expression: "); - Output.append(yamlScalar(*Expression)); - Output.append(", debug-info-location: "); - Output.append(yamlScalar(*Location)); - Output.append(" }\n"); - } - } else if (Options.Mode == MIPrintMode::Complete) { - Output.append("entry_values: []\n"); - } - - if (!Records.callSites().empty()) { - Output.append("callSites:\n"); - for (const MachineCallSiteInfo &CallSite : Records.callSites()) { - auto Location = instructionLocation(Function, CallSite.CallInstruction); - if (!Location) - return std::unexpected(Location.error()); - Output.append(" - { bb: "); - Output.append(std::to_string(Location->Block)); - Output.append(", offset: "); - Output.append(std::to_string(Location->Offset)); - if (!CallSite.ForwardedArguments.empty()) { - Output.append(", fwdArgRegs: [ "); - for (std::size_t Index = 0; Index < CallSite.ForwardedArguments.size(); ++Index) { - if (Index != 0) - Output.append(", "); - const MachineCallSiteArgument &Argument = CallSite.ForwardedArguments[Index]; - auto RegisterName = physicalRegister(Argument.Reg); - if (!RegisterName) - return std::unexpected(RegisterName.error()); - Output.append("{ arg: "); - Output.append(std::to_string(Argument.ArgumentNumber)); - Output.append(", reg: "); - Output.append(yamlScalar(*RegisterName)); - Output.append(" }"); - } - Output.append(" ]"); - } - if (!CallSite.CalleeTypeIDs.empty()) { - Output.append(", calleeTypeIds: [ "); - for (std::size_t Index = 0; Index < CallSite.CalleeTypeIDs.size(); ++Index) { - if (Index != 0) - Output.append(", "); - Output.append(std::to_string(CallSite.CalleeTypeIDs[Index])); - } - Output.append(" ]"); - } - Output.append(" }\n"); - } - } else if (Options.Mode == MIPrintMode::Complete) { - Output.append("callSites: []\n"); - } - - if (!Records.debugValueSubstitutions().empty()) { - Output.append("debugValueSubstitutions:\n"); - for (const MachineDebugValueSubstitution &Substitution : Records.debugValueSubstitutions()) { - Output.append(" - { srcinst: "); - Output.append(std::to_string(Substitution.Source.InstructionNumber)); - Output.append(", srcop: "); - Output.append(std::to_string(Substitution.Source.OperandNumber)); - Output.append(", dstinst: "); - Output.append(std::to_string(Substitution.Destination.InstructionNumber)); - Output.append(", dstop: "); - Output.append(std::to_string(Substitution.Destination.OperandNumber)); - Output.append(", subreg: "); - Output.append(std::to_string(Substitution.Subregister.value())); - Output.append(" }\n"); - } - } else if (Options.Mode == MIPrintMode::Complete) { - Output.append("debugValueSubstitutions: []\n"); - } - - if (!Records.calledGlobals().empty()) { - Output.append("calledGlobals:\n"); - for (const MachineCalledGlobal &Called : Records.calledGlobals()) { - auto Location = instructionLocation(Function, Called.CallInstruction); - if (!Location) - return std::unexpected(Location.error()); - Output.append(" - { bb: "); - Output.append(std::to_string(Location->Block)); - Output.append(", offset: "); - Output.append(std::to_string(Location->Offset)); - Output.append(", callee: "); - Output.append(yamlScalar(Called.Callee)); - Output.append(", flags: "); - Output.append(std::to_string(Called.Flags)); - Output.append(" }\n"); - } - } - - if (!Records.prefetchTargets().empty()) { - Output.append("prefetch-targets: [ "); - const auto Targets = Records.prefetchTargets(); - for (std::size_t Index = 0; Index < Targets.size(); ++Index) { - if (Index != 0) - Output.append(", "); - const MachinePrefetchTarget &TargetValue = Targets[Index]; - std::string Text = "bb_id " + std::to_string(TargetValue.Block.BaseID) + ", " + - std::to_string(TargetValue.Block.CloneID) + ", " + - std::to_string(TargetValue.CallsiteIndex); - Output.append(yamlScalar(Text)); - } - Output.append(" ]\n"); - } - return {}; -} - -Expected<void> MIRDocumentPrinter::printFunction(const MachineFunction &Function, - const MIRFunctionDocumentInfo *) { - if (Function.compilationStatus().state() != MachineCompilationState::Success) { - return makeUnexpected(ErrorCode::Unsupported, - "fixed LLVM MIR cannot serialize this compilation failure state"); - } - auto Context = createBodyContext(Function); - if (!Context) - return std::unexpected(Context.error()); - - Output.append("---\nname: "); - Output.append(yamlScalar(Function.name())); - Output.push_back('\n'); - const auto Alignment = Function.attributes().alignment(); - if (Options.Mode == MIPrintMode::Complete || Alignment) { - Output.append("alignment: "); - Output.append(std::to_string(Alignment ? Alignment->bytes() : 1U)); - Output.push_back('\n'); - } - - const MachineFunctionProperties &Properties = Function.properties(); - const MachineFunctionAttributes &Attributes = Function.attributes(); - printBoolean("exposesReturnsTwice", - Attributes.has(MachineFunctionAttribute::ExposesReturnsTwice)); - printBoolean("legalized", Properties.has(MachineFunctionProperty::Legalized)); - printBoolean("regBankSelected", Properties.has(MachineFunctionProperty::RegBankSelected)); - printBoolean("selected", Properties.has(MachineFunctionProperty::Selected)); - printBoolean("failedISel", Properties.has(MachineFunctionProperty::FailedISel)); - printBoolean("tracksRegLiveness", Properties.has(MachineFunctionProperty::TracksLiveness)); - printBoolean("hasWinCFI", Attributes.has(MachineFunctionAttribute::HasWinCFI)); - printBoolean("noPhis", Properties.has(MachineFunctionProperty::NoPHIs), true); - printBoolean("isSSA", Properties.has(MachineFunctionProperty::IsSSA), true); - printBoolean("noVRegs", Properties.has(MachineFunctionProperty::NoVRegs), true); - printBoolean("hasFakeUses", Attributes.has(MachineFunctionAttribute::HasFakeUses), true); - printBoolean("callsEHReturn", Attributes.has(MachineFunctionAttribute::CallsEHReturn)); - printBoolean("callsUnwindInit", Attributes.has(MachineFunctionAttribute::CallsUnwindInit)); - printBoolean("hasEHContTarget", Attributes.has(MachineFunctionAttribute::HasEHContTarget)); - printBoolean("hasEHScopes", Attributes.has(MachineFunctionAttribute::HasEHScopes)); - printBoolean("hasEHFunclets", Attributes.has(MachineFunctionAttribute::HasEHFunclets)); - printBoolean("isOutlined", Attributes.has(MachineFunctionAttribute::IsOutlined)); - printBoolean("debugInstrRef", Attributes.has(MachineFunctionAttribute::DebugInstrRef)); - printBoolean("failsVerification", Properties.has(MachineFunctionProperty::FailsVerification)); - printBoolean("tracksDebugUserValues", - Attributes.has(MachineFunctionAttribute::TracksDebugUserValues)); - - if (auto Printed = printRegisters(Function, *Context); !Printed) - return Printed; - if (auto Printed = printFrameInfo(Function, *Context); !Printed) - return Printed; - if (auto Printed = printStackObjects(Function, *Context, true); !Printed) - return Printed; - if (auto Printed = printStackObjects(Function, *Context, false); !Printed) - return Printed; - if (auto Printed = printFunctionRecords(Function, *Context); !Printed) - return Printed; - if (auto Printed = printConstants(Function, *Context); !Printed) - return Printed; - if (auto Printed = printTargetState(Function); !Printed) - return Printed; - if (auto Printed = printJumpTable(Function, *Context); !Printed) - return Printed; - if (auto Printed = printMetadata(Function, *Context); !Printed) - return Printed; - - auto Body = - printMachineFunctionBody(Function, Target, *Context, MIPrinterOptions{.Mode = Options.Mode}); - if (!Body) - return std::unexpected(Body.error()); - Output.append("body: |\n"); - appendBlockScalar(Output, *Body, 2); - Output.append("...\n"); - return {}; -} - -Expected<std::string> MIRDocumentPrinter::run() { - auto UnitVerification = Document.unit().verify(); - if (!UnitVerification) - return std::unexpected(UnitVerification.error()); - if (Document.unit().functionCount() == 0) { - return makeUnexpected(ErrorCode::InvalidArgument, - "MIR document must contain at least one machine function"); - } - - for (const MIRFunctionDocumentInfo &Info : Document.functionDocumentInfo()) { - if (Document.unit().function(Info.Function) == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - "MIR document info references an unknown machine function"); - } - } - - for (std::size_t Index = 0; Index < Document.unit().functionCount(); ++Index) { - const MachineFunction *Function = Document.unit().functionAt(Index); - if (Function == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - "compilation unit contains a null machine function"); - } - auto ReferenceVerification = - verifyMIRDocumentReferences(Document, *Function, Options.EmitEmbeddedIR); - if (!ReferenceVerification) - return std::unexpected(ReferenceVerification.error()); - auto LayoutVerification = verifyOmittedEmbeddedIRLayout(*Function); - if (!LayoutVerification) - return std::unexpected(LayoutVerification.error()); - } - - if (Options.EmitEmbeddedIR && Document.embeddedIR() != nullptr) { - Output.append("--- |\n"); - appendBlockScalar(Output, Document.embeddedIR()->source().text(), 2); - Output.append("...\n"); - } - for (std::size_t Index = 0; Index < Document.unit().functionCount(); ++Index) { - const MachineFunction *Function = Document.unit().functionAt(Index); - if (Function == nullptr) - return makeUnexpected(ErrorCode::VerificationError, - "compilation unit contains a null machine function"); - if (auto Printed = printFunction(*Function, documentInfo(*Function)); !Printed) - return std::unexpected(Printed.error()); - } - return std::move(Output); -} - -} // namespace - -Expected<std::string> printMIRDocument(const MIRDocument &Document, const TargetDescription &Target, - MIRDocumentPrinterOptions Options) { - return MIRDocumentPrinter(Document, Target, std::move(Options)).run(); -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRDocumentReferenceValidation.cpp b/backend/lib/MIR/MIRDocumentReferenceValidation.cpp deleted file mode 100644 index 69b0dd1..0000000 --- a/backend/lib/MIR/MIRDocumentReferenceValidation.cpp +++ /dev/null @@ -1,426 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentReferenceValidation.h" - -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/MIR/MIRDocument.h" - -#include <algorithm> -#include <array> -#include <initializer_list> -#include <optional> -#include <ranges> -#include <string> -#include <string_view> -#include <variant> - -namespace rcc::mir { - -Expected<void> verifyMIRDocumentReferences(const MIRDocument &Document, - const MachineFunction &Function, bool UseEmbeddedIR) { - const EmbeddedIRReferenceIndex *Embedded = UseEmbeddedIR ? Document.embeddedIR() : nullptr; - const EmbeddedIRFunction *IRFunction = nullptr; - if (Embedded != nullptr) { - if (Function.name().empty()) { - std::size_t AnonymousOrdinal = 0; - bool FoundFunction = false; - for (std::size_t Index = 0; Index < Document.unit().functionCount(); ++Index) { - const MachineFunction *Candidate = Document.unit().functionAt(Index); - if (Candidate == &Function) { - FoundFunction = true; - break; - } - if (Candidate != nullptr && Candidate->name().empty()) - ++AnonymousOrdinal; - } - if (FoundFunction) - IRFunction = Embedded->unnamedFunction(AnonymousOrdinal); - } else { - IRFunction = Embedded->function(Function.name()); - } - if (IRFunction == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - "machine function has no matching embedded LLVM IR function"); - } - } - - const auto ValidateGlobal = [Embedded](std::string_view Name, - std::string_view Description) -> Expected<void> { - if (Embedded == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " requires embedded LLVM IR to resolve global '" + - std::string(Name) + "'"); - } - if (Embedded->hasGlobalValue(Name)) - return {}; - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " references an undefined embedded LLVM IR global '" + - std::string(Name) + "'"); - }; - const auto ValidateBlockAddress = [Embedded](std::string_view FunctionName, - std::string_view BlockName, - std::string_view Description) -> Expected<void> { - if (Embedded == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " requires embedded LLVM IR to resolve a block address"); - } - const EmbeddedIRFunction *AddressFunction = Embedded->function(FunctionName); - if (AddressFunction == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " references an undefined embedded LLVM IR function '" + - std::string(FunctionName) + "'"); - } - if (AddressFunction->block(BlockName) == nullptr) { - return makeUnexpected( - ErrorCode::VerificationError, - std::string(Description) + " references an undefined embedded LLVM IR block '" + - std::string(BlockName) + "' in function '" + std::string(FunctionName) + "'"); - } - return {}; - }; - const auto ValidateIRValue = [IRFunction](std::string_view Name, - std::string_view Description) -> Expected<void> { - if (IRFunction == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " requires embedded LLVM IR to resolve IR value '%" + - std::string(Name) + "'"); - } - if (IRFunction->value(Name) != nullptr) - return {}; - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " references an undefined embedded LLVM IR value '%" + - std::string(Name) + "'"); - }; - const auto ValidateMetadata = [Embedded](MachineMetadataReference Reference, - std::initializer_list<std::string_view> ExpectedTypes, - std::string_view Description) -> Expected<void> { - if (!Reference.isValid()) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + " has an invalid metadata reference"); - } - if (Reference.scope() != MachineMetadataScope::EmbeddedIR) - return {}; - if (Embedded == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + " requires embedded LLVM IR metadata !" + - std::to_string(Reference.slot())); - } - const EmbeddedIRMetadataDefinition *Definition = - Embedded->metadata(std::to_string(Reference.slot())); - if (Definition == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + - " references undefined embedded LLVM IR metadata !" + - std::to_string(Reference.slot())); - } - if (ExpectedTypes.size() == 0 || - std::ranges::find(ExpectedTypes, Definition->DynamicType) != ExpectedTypes.end()) { - return {}; - } - std::string Expected; - for (const std::string_view Type : ExpectedTypes) { - if (!Expected.empty()) - Expected.append(" or "); - Expected.append(Type); - } - return makeUnexpected(ErrorCode::VerificationError, - std::string(Description) + " requires " + Expected + " metadata, but !" + - std::to_string(Reference.slot()) + " is " + - (Definition->DynamicType.empty() - ? std::string("an unclassified metadata node") - : Definition->DynamicType)); - }; - const auto ValidateMetadataValue = - [&ValidateMetadata](const MachineMetadataValue &Value, - std::string_view Description) -> Expected<void> { - const MachineMetadataReference *Reference = Value.reference(); - return Reference == nullptr ? Expected<void>{} : ValidateMetadata(*Reference, {}, Description); - }; - const auto ValidateLocation = - [&ValidateMetadata](const MachineDebugLocation &Location, - std::string_view Description) -> Expected<void> { - if (const MachineMetadataReference *Reference = Location.reference()) - return ValidateMetadata(*Reference, {"DILocation"}, Description); - const MachineDILocation *Inline = Location.inlineLocation(); - if (Inline == nullptr) - return {}; - for (const MachineDILocationFrame &Frame : Inline->inlineChain()) { - auto Verification = - ValidateMetadata(Frame.Scope, {"DISubprogram", "DILexicalBlock", "DILexicalBlockFile"}, - "inline DILocation scope"); - if (!Verification) - return Verification; - } - if (Inline->terminalInlinedAt()) { - return ValidateMetadata(*Inline->terminalInlinedAt(), {"DILocation"}, - "inline DILocation inlinedAt"); - } - return {}; - }; - const auto ValidateVariableDebugInfo = - [&ValidateMetadata, &ValidateMetadataValue, - &ValidateLocation](const MachineVariableDebugInfo &DebugInfo) -> Expected<void> { - if (DebugInfo.Variable) { - auto Verification = - ValidateMetadata(*DebugInfo.Variable, {"DILocalVariable"}, "debug variable"); - if (!Verification) - return Verification; - } - if (DebugInfo.Expression) { - if (const MachineMetadataReference *Reference = DebugInfo.Expression->reference(); - Reference != nullptr) { - auto Verification = ValidateMetadata(*Reference, {"DIExpression"}, "debug expression"); - if (!Verification) - return Verification; - } - auto Verification = ValidateMetadataValue(*DebugInfo.Expression, "debug expression"); - if (!Verification) - return Verification; - } - if (DebugInfo.Location) - return ValidateLocation(*DebugInfo.Location, "debug location"); - return {}; - }; - const auto ValidateGenericType = [Embedded](const auto &Self, - const GenericConstantType &Type) -> Expected<void> { - if (Type.Kind == GenericConstantTypeKind::Named) { - if (Embedded == nullptr) { - return makeUnexpected( - ErrorCode::VerificationError, - "generic constant type requires embedded LLVM IR to resolve named type '%" + Type.Name + - "'"); - } - if (Embedded->typeDefinition(Type.Name) == nullptr) { - return makeUnexpected( - ErrorCode::VerificationError, - "generic constant type references an undefined embedded LLVM IR named type '%" + - Type.Name + "'"); - } - } - for (const GenericConstantType &Element : Type.Elements) { - auto Verification = Self(Self, Element); - if (!Verification) - return Verification; - } - return {}; - }; - const auto ValidateGenericConstant = - [&ValidateBlockAddress, &ValidateGenericType, - &ValidateGlobal](const auto &Self, const GenericConstantData &Constant) -> Expected<void> { - auto TypeVerification = ValidateGenericType(ValidateGenericType, Constant.Type); - if (!TypeVerification) - return TypeVerification; - if (Constant.SourceElementType) { - TypeVerification = ValidateGenericType(ValidateGenericType, *Constant.SourceElementType); - if (!TypeVerification) - return TypeVerification; - } - switch (Constant.Kind) { - case GenericConstantKind::GlobalAddress: - case GenericConstantKind::DSOLocalEquivalent: - case GenericConstantKind::NoCFI: - if (auto Verification = ValidateGlobal(Constant.Symbol, "generic constant"); !Verification) - return Verification; - break; - case GenericConstantKind::BlockAddress: - if (auto Verification = ValidateBlockAddress(Constant.Symbol, Constant.Block, - "generic constant blockaddress"); - !Verification) { - return Verification; - } - break; - default: - break; - } - for (const GenericConstantData &Element : Constant.Elements) { - auto Verification = Self(Self, Element); - if (!Verification) - return Verification; - } - return {}; - }; - - for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; - Block = Block->next()) { - if (const auto Name = Block->attributes().irBlockName(); Name) { - if (IRFunction == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - "machine basic block requires embedded LLVM IR to resolve block '" + - std::string(*Name) + "'"); - } - if (IRFunction->block(*Name) == nullptr) { - return makeUnexpected( - ErrorCode::VerificationError, - "machine basic block references an undefined embedded LLVM IR block '" + - std::string(*Name) + "'"); - } - } - } - for (const StackObject &Object : Function.frameInfo().objects()) { - const StackObjectDesc *Description = Object.description(); - if (Description == nullptr || !Description->AllocaName) - continue; - if (IRFunction == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - "stack object requires embedded LLVM IR to resolve alloca '" + - *Description->AllocaName + "'"); - } - if (IRFunction->alloca(*Description->AllocaName) == nullptr) { - return makeUnexpected(ErrorCode::VerificationError, - "stack object references an undefined embedded LLVM IR alloca '" + - *Description->AllocaName + "'"); - } - } - - for (const MachineConstantPoolEntry &Entry : Function.constantPool().entries()) { - const MachineConstantPoolEntryDesc *Description = Entry.description(); - if (Description == nullptr) - continue; - if (const auto *Generic = std::get_if<GenericConstantData>(&Description->Value)) { - auto Verification = ValidateGenericConstant(ValidateGenericConstant, *Generic); - if (!Verification) - return Verification; - } else if (const auto *Address = std::get_if<BlockAddressConstant>(&Description->Value)) { - auto Verification = ValidateBlockAddress(Address->FunctionName, Address->BlockName, - "constant-pool block address"); - if (!Verification) - return Verification; - } - } - - for (const MachineMetadataNode &Node : Function.metadata().nodes()) { - for (const MachineMetadataElement &Element : Node.Elements) { - const MachineMetadataReference *Reference = Element.reference(); - if (Reference == nullptr) - continue; - auto Verification = ValidateMetadata(*Reference, {}, "machine metadata node"); - if (!Verification) - return Verification; - } - } - - for (const StackObject &Object : Function.frameInfo().objects()) { - const StackObjectDesc *Description = Object.description(); - if (Description == nullptr || !Description->DebugInfo) - continue; - auto Verification = ValidateVariableDebugInfo(*Description->DebugInfo); - if (!Verification) - return Verification; - } - for (const MachineEntryValueInfo &Entry : Function.records().entryValues()) { - auto Verification = ValidateVariableDebugInfo(Entry.DebugInfo); - if (!Verification) - return Verification; - } - for (const MachineCalledGlobal &Called : Function.records().calledGlobals()) { - auto Verification = ValidateGlobal(Called.Callee, "called-global record"); - if (!Verification) - return Verification; - } - - for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; - Block = Block->next()) { - for (const MachineInstr *Instruction = Block->firstInstruction(); Instruction != nullptr; - Instruction = Instruction->next()) { - if (Instruction->debugLocation()) { - auto Verification = - ValidateLocation(*Instruction->debugLocation(), "instruction debug location"); - if (!Verification) - return Verification; - } - const std::array InstructionMetadata{ - Instruction->heapAllocMarker(), - Instruction->pcSections(), - Instruction->mmra(), - }; - for (const auto &Reference : InstructionMetadata) { - if (!Reference) - continue; - auto Verification = ValidateMetadata(*Reference, {}, "instruction metadata attachment"); - if (!Verification) - return Verification; - } - - for (const MachineOperand &Operand : Instruction->operands()) { - if (const auto *Global = Operand.getIf<GlobalSymbolOperand>(); Global != nullptr) { - auto Verification = ValidateGlobal(Global->Name, "global-symbol operand"); - if (!Verification) - return Verification; - } - if (const auto *Address = Operand.getIf<BlockAddressOperand>(); Address != nullptr) { - auto Verification = ValidateBlockAddress(Address->FunctionName, Address->BlockName, - "block-address operand"); - if (!Verification) - return Verification; - } - if (const auto *Metadata = Operand.getIf<MetadataOperand>(); Metadata != nullptr) { - auto Verification = ValidateMetadataValue(Metadata->Value, "metadata operand"); - if (!Verification) - return Verification; - } - } - - for (const MachineMemOperand *Memory : Instruction->memOperands()) { - const MachineMemOperandDesc &Description = Memory->description(); - const std::optional<std::string_view> BaseName = Description.Base.symbolName(); - switch (Description.Base.kind()) { - case MemoryBase::Kind::GlobalSymbol: - case MemoryBase::Kind::CallEntryGlobal: - if (BaseName) { - auto Verification = ValidateGlobal(*BaseName, "machine memory base"); - if (!Verification) - return Verification; - } - break; - case MemoryBase::Kind::IRValue: - if (BaseName) { - auto Verification = ValidateIRValue(*BaseName, "machine memory base"); - if (!Verification) - return Verification; - } - break; - default: - break; - } - const std::array MetadataReferences{ - Description.TBAA, Description.AliasScope, - Description.NoAlias, Description.NoAliasAddressSpace, - Description.RangeMetadata, - }; - for (const auto &Reference : MetadataReferences) { - if (!Reference) - continue; - auto Verification = ValidateMetadata(*Reference, {}, "machine memory operand metadata"); - if (!Verification) - return Verification; - } - if (Description.SyncScope.value() == 0) - continue; - const auto Scope = - std::ranges::find_if(Document.syncScopes(), [&Description](const auto &Slot) { - return Slot.ID == Description.SyncScope; - }); - if (Scope == Document.syncScopes().end()) { - return makeUnexpected( - ErrorCode::VerificationError, - "machine memory operand synchronization scope has no document name"); - } - } - } - } - return {}; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRDocumentReferenceValidation.h b/backend/lib/MIR/MIRDocumentReferenceValidation.h deleted file mode 100644 index 2ab6291..0000000 --- a/backend/lib/MIR/MIRDocumentReferenceValidation.h +++ /dev/null @@ -1,25 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -// The private header has no stable include-root-relative guard name. -// NOLINTNEXTLINE(llvm-header-guard) -#ifndef RCC_LIB_MIR_MIRDOCUMENTREFERENCEVALIDATION_H -#define RCC_LIB_MIR_MIRDOCUMENTREFERENCEVALIDATION_H - -#include "rcc/Support/Error.h" - -namespace rcc { - -class MachineFunction; - -namespace mir { - -class MIRDocument; - -[[nodiscard]] Expected<void> verifyMIRDocumentReferences(const MIRDocument &Document, - const MachineFunction &Function, - bool UseEmbeddedIR); - -} // namespace mir -} // namespace rcc - -#endif // RCC_LIB_MIR_MIRDOCUMENTREFERENCEVALIDATION_H diff --git a/backend/lib/MIR/MIRFrameParser.cpp b/backend/lib/MIR/MIRFrameParser.cpp deleted file mode 100644 index 4367fbd..0000000 --- a/backend/lib/MIR/MIRFrameParser.cpp +++ /dev/null @@ -1,467 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentParser.h" -#include "MIRDocumentReferenceValidation.h" -#include "MIRMetadataTextParser.h" -#include "MIRParserUtilities.h" -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionAttributes.h" -#include "rcc/CodeGen/MachineFunctionProperties.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/MachineRegisterInfo.h" -#include "rcc/CodeGen/MachineVerifier.h" -#include "rcc/CodeGen/TargetDescription.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/CodeGen/TargetRegisterInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" - -#include <algorithm> -#include <array> -#include <cstddef> -#include <cstdint> -#include <initializer_list> -#include <limits> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <unordered_map> -#include <unordered_set> -#include <utility> -#include <vector> - -namespace rcc::mir { - -using parser_detail::parseBool; -using parser_detail::parsePrefixedID; -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; -using parser_detail::trim; - -bool MIRDocumentParser::parseStackObjects(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context) { - auto ParseList = [this, &Function, &Context](const yaml::Node &List, bool Fixed) -> bool { - if (requireKind(List, yaml::NodeKind::Sequence, - Fixed ? "fixed stack objects" : "stack objects")) { - return true; - } - constexpr std::array FixedKeys{ - std::string_view{"id"}, - std::string_view{"type"}, - std::string_view{"offset"}, - std::string_view{"size"}, - std::string_view{"alignment"}, - std::string_view{"stack-id"}, - std::string_view{"isImmutable"}, - std::string_view{"isAliased"}, - std::string_view{"callee-saved-register"}, - std::string_view{"callee-saved-restored"}, - std::string_view{"debug-info-variable"}, - std::string_view{"debug-info-expression"}, - std::string_view{"debug-info-location"}, - }; - constexpr std::array StackKeys{ - std::string_view{"id"}, - std::string_view{"name"}, - std::string_view{"type"}, - std::string_view{"offset"}, - std::string_view{"size"}, - std::string_view{"alignment"}, - std::string_view{"stack-id"}, - std::string_view{"callee-saved-register"}, - std::string_view{"callee-saved-restored"}, - std::string_view{"local-offset"}, - std::string_view{"debug-info-variable"}, - std::string_view{"debug-info-expression"}, - std::string_view{"debug-info-location"}, - }; - std::unordered_set<std::uint32_t> IDs; - for (const yaml::NodeID ElementID : List.sequence()) { - const yaml::Node *Element = node(ElementID); - if (Element == nullptr || requireKind(*Element, yaml::NodeKind::Mapping, "stack object") || - (Fixed ? validateKeys(*Element, FixedKeys, "fixed stack object") - : validateKeys(*Element, StackKeys, "stack object"))) { - return true; - } - const yaml::Node *IDNode = mappingValue(*Element, "id"); - if (IDNode == nullptr) - return fail(Element->range(), "stack object requires 'id'"); - std::uint64_t WideID = 0; - if (unsignedValue(*IDNode, WideID, "a 32-bit stack object ID") || - WideID > std::numeric_limits<std::uint32_t>::max()) { - return true; - } - const auto TextID = static_cast<std::uint32_t>(WideID); - if (!IDs.insert(TextID).second) - return fail(IDNode->range(), "duplicate stack object ID"); - - StackObjectDesc Description; - Description.Kind = Fixed ? StackObjectKind::FixedABI : StackObjectKind::Default; - Description.Offset = 0; - if (const yaml::Node *NameNode = mappingValue(*Element, "name")) { - std::string_view Name; - if (scalar(*NameNode, Name, "stack object name")) - return true; - if (!Name.empty()) - Description.AllocaName = std::string(Name); - } - if (const yaml::Node *TypeNode = mappingValue(*Element, "type")) { - std::string_view Type; - if (scalar(*TypeNode, Type, "stack object type")) - return true; - if (Type == "default") - Description.Kind = Fixed ? StackObjectKind::FixedABI : StackObjectKind::Default; - else if (Type == "spill-slot") - Description.Kind = Fixed ? StackObjectKind::FixedSpillSlot : StackObjectKind::SpillSlot; - else if (!Fixed && Type == "variable-sized") - Description.Kind = StackObjectKind::VariableSized; - else - return fail(TypeNode->range(), "unknown stack object type '" + std::string(Type) + "'"); - } - if (const yaml::Node *OffsetNode = mappingValue(*Element, "offset")) { - std::int64_t Offset = 0; - if (signedValue(*OffsetNode, Offset, "a signed stack object offset")) - return true; - Description.Offset = Offset; - } else if (Fixed) { - Description.Offset = 0; - } - if (Description.Kind != StackObjectKind::VariableSized) { - const yaml::Node *SizeNode = mappingValue(*Element, "size"); - if (SizeNode == nullptr) { - if (!Fixed) - return fail(Element->range(), "fixed-size stack object requires 'size'"); - Description.Size = 0; - } else { - std::uint64_t Size = 0; - if (unsignedValue(*SizeNode, Size, "an unsigned stack object size")) - return true; - Description.Size = Size; - } - } - if (const yaml::Node *AlignmentNode = mappingValue(*Element, "alignment")) { - if (alignment(*AlignmentNode, Description.Align, true)) - return true; - } - if (const yaml::Node *StackIDNode = mappingValue(*Element, "stack-id")) { - std::string_view Name; - if (scalar(*StackIDNode, Name, "stack ID")) - return true; - const auto Stack = parseStackIDName(Name); - if (!Stack) - return fail(StackIDNode->range(), "unknown stack ID '" + std::string(Name) + "'"); - Description.Stack = *Stack; - } - if (const yaml::Node *ImmutableNode = mappingValue(*Element, "isImmutable")) { - if (boolean(*ImmutableNode, Description.IsImmutable)) - return true; - } - if (const yaml::Node *AliasedNode = mappingValue(*Element, "isAliased")) { - if (boolean(*AliasedNode, Description.IsAliased)) - return true; - } - if (const yaml::Node *SavedNode = mappingValue(*Element, "callee-saved-register")) { - std::string_view SavedName; - if (scalar(*SavedNode, SavedName, "physical register")) - return true; - if (!trim(SavedName).empty()) { - PhysicalRegister Reg; - if (physicalRegister(SavedName, SavedNode->range(), Reg)) - return true; - Description.CalleeSavedRegister = Reg; - } - } - if (const yaml::Node *RestoredNode = mappingValue(*Element, "callee-saved-restored")) { - if (boolean(*RestoredNode, Description.CalleeSavedRestored)) - return true; - } - if (const yaml::Node *LocalNode = mappingValue(*Element, "local-offset")) { - std::int64_t Offset = 0; - if (signedValue(*LocalNode, Offset, "a signed local frame offset")) - return true; - Description.LocalOffset = Offset; - } - MachineVariableDebugInfo DebugInfo; - if (parseVariableDebugInfo(*Element, Function, DebugInfo)) - return true; - if (DebugInfo.Variable || DebugInfo.Expression || DebugInfo.Location) - Description.DebugInfo = std::move(DebugInfo); - - auto Created = Function.frameInfo().createObject(std::move(Description)); - if (!Created) - return importError(Element->range(), Created.error()); - MIFrameObjectSlot Slot{ - .TextID = TextID, - .Index = *Created, - .Name = std::nullopt, - }; - if (const StackObjectDesc *CreatedDescription = - Function.frameInfo().object(*Created)->description(); - CreatedDescription != nullptr && CreatedDescription->AllocaName) { - Slot.Name = *CreatedDescription->AllocaName; - } - (Fixed ? Context.FixedStackObjects : Context.StackObjects).push_back(std::move(Slot)); - } - return false; - }; - - if (const yaml::Node *Fixed = mappingValue(Root, "fixedStack")) { - if (ParseList(*Fixed, true)) - return true; - } - if (const yaml::Node *Stack = mappingValue(Root, "stack")) { - if (ParseList(*Stack, false)) - return true; - } - return false; -} - -bool MIRDocumentParser::parseFrameReference(const yaml::Node &Node, const MIParserContext &Context, - FrameIndex &Index) { - std::string_view Text; - if (scalar(Node, Text, "frame index")) - return true; - bool Fixed = false; - std::optional<std::uint32_t> ID = parsePrefixedID(Text, "%stack."); - if (!ID) { - ID = parsePrefixedID(Text, "%fixed-stack."); - Fixed = true; - } - if (!ID) - return fail(Node.range(), "frame index must use '%stack.N' or '%fixed-stack.N'"); - const auto &Slots = Fixed ? Context.FixedStackObjects : Context.StackObjects; - const auto Found = std::ranges::find_if( - Slots, [ID](const MIFrameObjectSlot &Slot) { return Slot.TextID == *ID; }); - if (Found == Slots.end()) - return fail(Node.range(), "reference to undefined frame object"); - Index = Found->Index; - return false; -} - -MachineBasicBlock *MIRDocumentParser::blockAt(MachineFunction &Function, - std::uint32_t Number) const noexcept { - for (MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; Block = Block->next()) { - if (Block->denseNumber() == Number) - return Block; - } - return nullptr; -} - -MachineInstr *MIRDocumentParser::instructionAt(MachineFunction &Function, std::uint32_t BlockNumber, - std::uint32_t Offset) const noexcept { - MachineBasicBlock *Block = blockAt(Function, BlockNumber); - if (Block == nullptr) - return nullptr; - MachineInstr *Instruction = Block->firstInstruction(); - for (std::uint32_t Index = 0; Instruction != nullptr && Index < Offset; ++Index) - Instruction = Instruction->next(); - return Instruction; -} - -bool MIRDocumentParser::parseBlockReference(std::string_view Text, const TextRange &Range, - MachineFunction &Function, MachineBasicBlock *&Block) { - const auto Number = parsePrefixedID(Text, "%bb."); - if (!Number) - return fail(Range, "basic block reference must use the '%bb.N' MIR spelling"); - Block = blockAt(Function, *Number); - if (Block == nullptr) - return fail(Range, "reference to undefined machine basic block"); - return false; -} - -bool MIRDocumentParser::parseSaveRestorePoints(const yaml::Node &Node, MachineFunction &Function, - std::vector<MachineFrameSaveRestorePoint> &Points) { - if (requireKind(Node, yaml::NodeKind::Sequence, "frame save/restore points")) - return true; - constexpr std::array Keys{ - std::string_view{"point"}, - std::string_view{"registers"}, - }; - for (const yaml::NodeID PointID : Node.sequence()) { - const yaml::Node *PointNode = node(PointID); - if (PointNode == nullptr || - requireKind(*PointNode, yaml::NodeKind::Mapping, "frame save/restore point") || - validateKeys(*PointNode, Keys, "frame save/restore point")) { - return true; - } - const yaml::Node *BlockNode = mappingValue(*PointNode, "point"); - if (BlockNode == nullptr) - return fail(PointNode->range(), "frame save/restore point requires 'point'"); - std::string_view BlockText; - MachineBasicBlock *Block = nullptr; - if (scalar(*BlockNode, BlockText, "basic block reference") || - parseBlockReference(BlockText, BlockNode->range(), Function, Block)) { - return true; - } - if (Block == nullptr) - return fail(BlockNode->range(), "reference to undefined machine basic block"); - MachineFrameSaveRestorePoint Point{.Block = Block->stableID(), .Registers = {}}; - if (const yaml::Node *RegistersNode = mappingValue(*PointNode, "registers")) { - if (requireKind(*RegistersNode, yaml::NodeKind::Sequence, "save/restore point registers")) { - return true; - } - for (const yaml::NodeID RegID : RegistersNode->sequence()) { - const yaml::Node *RegNode = node(RegID); - PhysicalRegister Reg; - if (RegNode == nullptr || physicalRegister(*RegNode, Reg)) - return true; - Point.Registers.push_back(Reg); - } - } - Points.push_back(std::move(Point)); - } - return false; -} - -bool MIRDocumentParser::parseFrameState(const yaml::Node &Root, MachineFunction &Function, - const MIParserContext &Context) { - const yaml::Node *Frame = mappingValue(Root, "frameInfo"); - if (Frame == nullptr) - return false; - if (requireKind(*Frame, yaml::NodeKind::Mapping, "frameInfo")) - return true; - constexpr std::array Keys{ - std::string_view{"isFrameAddressTaken"}, - std::string_view{"isReturnAddressTaken"}, - std::string_view{"hasStackMap"}, - std::string_view{"hasPatchPoint"}, - std::string_view{"stackSize"}, - std::string_view{"offsetAdjustment"}, - std::string_view{"maxAlignment"}, - std::string_view{"adjustsStack"}, - std::string_view{"hasCalls"}, - std::string_view{"framePointerPolicy"}, - std::string_view{"stackProtector"}, - std::string_view{"functionContext"}, - std::string_view{"maxCallFrameSize"}, - std::string_view{"cvBytesOfCalleeSavedRegisters"}, - std::string_view{"hasOpaqueSPAdjustment"}, - std::string_view{"hasVAStart"}, - std::string_view{"hasMustTailInVarArgFunc"}, - std::string_view{"hasTailCall"}, - std::string_view{"isCalleeSavedInfoValid"}, - std::string_view{"localFrameSize"}, - std::string_view{"savePoint"}, - std::string_view{"restorePoint"}, - }; - if (validateKeys(*Frame, Keys, "frameInfo")) - return true; - - MachineFrameState State = Function.frameInfo().state(); - const std::array BoolBindings{ - std::pair{"isFrameAddressTaken", &State.IsFrameAddressTaken}, - std::pair{"isReturnAddressTaken", &State.IsReturnAddressTaken}, - std::pair{"hasStackMap", &State.HasStackMap}, - std::pair{"hasPatchPoint", &State.HasPatchPoint}, - std::pair{"adjustsStack", &State.AdjustsStack}, - std::pair{"hasCalls", &State.HasCalls}, - std::pair{"hasOpaqueSPAdjustment", &State.HasOpaqueSPAdjustment}, - std::pair{"hasVAStart", &State.HasVAStart}, - std::pair{"hasMustTailInVarArgFunc", &State.HasMustTailInVarArgFunc}, - std::pair{"hasTailCall", &State.HasTailCall}, - std::pair{"isCalleeSavedInfoValid", &State.IsCalleeSavedInfoValid}, - }; - for (const auto &[Key, Destination] : BoolBindings) { - if (const yaml::Node *Value = mappingValue(*Frame, Key)) { - if (boolean(*Value, *Destination)) - return true; - } - } - - if (const yaml::Node *Value = mappingValue(*Frame, "stackSize")) { - if (unsignedValue(*Value, State.StackSize, "an unsigned stack size")) - return true; - } - if (const yaml::Node *Value = mappingValue(*Frame, "offsetAdjustment")) { - std::int64_t Wide = 0; - if (signedValue(*Value, Wide, "a 32-bit signed offset adjustment") || - Wide < std::numeric_limits<std::int32_t>::min() || - Wide > std::numeric_limits<std::int32_t>::max()) { - return fail(Value->range(), "offset adjustment is outside the 32-bit signed range"); - } - State.OffsetAdjustment = static_cast<std::int32_t>(Wide); - } - if (const yaml::Node *Value = mappingValue(*Frame, "maxAlignment")) { - std::uint64_t Bytes = 0; - if (unsignedValue(*Value, Bytes, "an unsigned maximum alignment")) - return true; - if (Bytes == 0) { - State.MaxAlignment.reset(); - } else { - auto Parsed = Alignment::create(Bytes); - if (!Parsed) - return importError(Value->range(), Parsed.error()); - State.MaxAlignment = *Parsed; - } - } - if (const yaml::Node *Value = mappingValue(*Frame, "framePointerPolicy")) { - std::string_view Name; - if (scalar(*Value, Name, "frame-pointer policy")) - return true; - const auto Policy = parseMachineFramePointerPolicy(Name); - if (!Policy) - return fail(Value->range(), "unknown frame-pointer policy '" + std::string(Name) + "'"); - State.FramePointerPolicy = *Policy; - } - for (const auto &[Key, Destination] : {std::pair{"stackProtector", &State.StackProtector}, - std::pair{"functionContext", &State.FunctionContext}}) { - if (const yaml::Node *Value = mappingValue(*Frame, Key)) { - std::string_view Reference; - if (scalar(*Value, Reference, "frame index")) - return true; - if (Reference.empty()) { - Destination->reset(); - continue; - } - FrameIndex Index; - if (parseFrameReference(*Value, Context, Index)) - return true; - *Destination = Index; - } - } - if (const yaml::Node *Value = mappingValue(*Frame, "maxCallFrameSize")) { - std::uint64_t Size = 0; - if (unsignedValue(*Value, Size, "a 32-bit unsigned maximum call-frame size") || - Size > std::numeric_limits<std::uint32_t>::max()) { - return fail(Value->range(), "maximum call-frame size is outside the 32-bit unsigned range"); - } - if (Size == std::numeric_limits<std::uint32_t>::max()) - State.MaxCallFrameSize.reset(); - else - State.MaxCallFrameSize = Size; - } - for (const auto &[Key, Destination] : - {std::pair{"cvBytesOfCalleeSavedRegisters", &State.CVBytesOfCalleeSavedRegisters}, - std::pair{"localFrameSize", &State.LocalFrameSize}}) { - if (const yaml::Node *Value = mappingValue(*Frame, Key)) { - std::uint64_t Wide = 0; - if (unsignedValue(*Value, Wide, "a 32-bit unsigned frame value") || - Wide > std::numeric_limits<std::uint32_t>::max()) { - return fail(Value->range(), "frame value is outside the 32-bit unsigned range"); - } - *Destination = static_cast<std::uint32_t>(Wide); - } - } - if (const yaml::Node *Value = mappingValue(*Frame, "savePoint")) { - if (parseSaveRestorePoints(*Value, Function, State.SavePoints)) - return true; - } - if (const yaml::Node *Value = mappingValue(*Frame, "restorePoint")) { - if (parseSaveRestorePoints(*Value, Function, State.RestorePoints)) - return true; - } - auto StateResult = Function.frameInfo().setState(std::move(State)); - if (!StateResult) - return importError(Frame->range(), StateResult.error()); - return false; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRFunctionParser.cpp b/backend/lib/MIR/MIRFunctionParser.cpp deleted file mode 100644 index 6acf774..0000000 --- a/backend/lib/MIR/MIRFunctionParser.cpp +++ /dev/null @@ -1,552 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentParser.h" -#include "MIRDocumentReferenceValidation.h" -#include "MIRMetadataTextParser.h" -#include "MIRParserUtilities.h" -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionAttributes.h" -#include "rcc/CodeGen/MachineFunctionProperties.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/MachineRegisterInfo.h" -#include "rcc/CodeGen/MachineVerifier.h" -#include "rcc/CodeGen/TargetDescription.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/CodeGen/TargetRegisterInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" - -#include <algorithm> -#include <array> -#include <cstddef> -#include <cstdint> -#include <initializer_list> -#include <limits> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <unordered_map> -#include <unordered_set> -#include <utility> -#include <vector> - -namespace rcc::mir { - -using parser_detail::parseBool; -using parser_detail::parsePrefixedID; -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; -using parser_detail::trim; - -bool MIRDocumentParser::parseFunctionScalars(const yaml::Node &Root, MachineFunction &Function, - MIRFunctionDocumentInfo &DocumentInfo) { - if (const yaml::Node *AlignNode = mappingValue(Root, "alignment")) { - Alignment Align; - if (alignment(*AlignNode, Align, true)) - return true; - if (Align.bytes() != 1 || AlignNode->scalar() != "0") - Function.attributes().setAlignment(Align); - } - - constexpr std::array PropertyBindings{ - std::pair{"legalized", MachineFunctionProperty::Legalized}, - std::pair{"regBankSelected", MachineFunctionProperty::RegBankSelected}, - std::pair{"selected", MachineFunctionProperty::Selected}, - std::pair{"failedISel", MachineFunctionProperty::FailedISel}, - std::pair{"tracksRegLiveness", MachineFunctionProperty::TracksLiveness}, - std::pair{"noPhis", MachineFunctionProperty::NoPHIs}, - std::pair{"isSSA", MachineFunctionProperty::IsSSA}, - std::pair{"noVRegs", MachineFunctionProperty::NoVRegs}, - std::pair{"failsVerification", MachineFunctionProperty::FailsVerification}, - }; - for (const auto &[Key, Property] : PropertyBindings) { - const yaml::Node *ValueNode = mappingValue(Root, Key); - if (ValueNode == nullptr) - continue; - bool Value = false; - if (boolean(*ValueNode, Value)) - return true; - if (Value) - Function.properties().set(Property); - else - Function.properties().clear(Property); - if (Key == std::string_view("noPhis")) - DocumentInfo.NoPHIs = Value; - else if (Key == std::string_view("isSSA")) - DocumentInfo.IsSSA = Value; - else if (Key == std::string_view("noVRegs")) - DocumentInfo.NoVRegs = Value; - } - - constexpr std::array AttributeBindings{ - std::pair{"exposesReturnsTwice", MachineFunctionAttribute::ExposesReturnsTwice}, - std::pair{"hasWinCFI", MachineFunctionAttribute::HasWinCFI}, - std::pair{"hasFakeUses", MachineFunctionAttribute::HasFakeUses}, - std::pair{"callsEHReturn", MachineFunctionAttribute::CallsEHReturn}, - std::pair{"callsUnwindInit", MachineFunctionAttribute::CallsUnwindInit}, - std::pair{"hasEHContTarget", MachineFunctionAttribute::HasEHContTarget}, - std::pair{"hasEHScopes", MachineFunctionAttribute::HasEHScopes}, - std::pair{"hasEHFunclets", MachineFunctionAttribute::HasEHFunclets}, - std::pair{"isOutlined", MachineFunctionAttribute::IsOutlined}, - std::pair{"debugInstrRef", MachineFunctionAttribute::DebugInstrRef}, - std::pair{"tracksDebugUserValues", MachineFunctionAttribute::TracksDebugUserValues}, - }; - for (const auto &[Key, Attribute] : AttributeBindings) { - const yaml::Node *ValueNode = mappingValue(Root, Key); - if (ValueNode == nullptr) - continue; - bool Value = false; - if (boolean(*ValueNode, Value)) - return true; - if (Value) - Function.attributes().set(Attribute); - else - Function.attributes().clear(Attribute); - if (Key == std::string_view("hasFakeUses")) - DocumentInfo.HasFakeUses = Value; - } - - return false; -} - -bool MIRDocumentParser::recomputeProperties(const yaml::Node &Root, MachineFunction &Function, - const MIRFunctionDocumentInfo &DocumentInfo) { - bool HasPHI = false; - bool HasFakeUse = false; - bool IsSSA = true; - bool HasTiedOperands = false; - bool AllTiedOperandsRewritten = true; - bool HasFrameIndex = false; - bool HasExecutablePseudo = false; - std::unordered_set<Register> Definitions; - for (const MachineBasicBlock *Block = Function.firstBlock(); Block != nullptr; - Block = Block->next()) { - for (const MachineInstr *Instruction = Block->firstInstruction(); Instruction != nullptr; - Instruction = Instruction->next()) { - const InstructionDescriptor *Descriptor = Target.instructionDescriptor(Instruction->opcode()); - if (Descriptor == nullptr) - return fail(Root.range(), "instruction has no target descriptor"); - HasPHI |= Descriptor->Flags.has(InstructionFlag::Phi); - HasFakeUse |= Descriptor->Flags.has(InstructionFlag::FakeUse); - HasExecutablePseudo |= Descriptor->Flags.has(InstructionFlag::Pseudo) && - !Descriptor->Flags.has(InstructionFlag::Meta) && - !Descriptor->Flags.has(InstructionFlag::TerminalEmission); - const auto Operands = Instruction->operands(); - for (std::size_t Index = 0; Index < Operands.size(); ++Index) { - const MachineOperand &Operand = Operands[Index]; - HasFrameIndex |= Operand.getIf<FrameIndexOperand>() != nullptr; - const auto Reg = Operand.registerValue(); - if (Reg && Reg->isVirtual() && Operand.isDef() && !Definitions.insert(*Reg).second) { - IsSSA = false; - } - if (!Operand.tiedOperandIndex() || !Operand.isUse()) - continue; - HasTiedOperands = true; - const MachineOperand *Definition = Instruction->operandAt(*Operand.tiedOperandIndex()); - if (Definition == nullptr || Definition->registerValue() != Reg) - AllTiedOperandsRewritten = false; - } - for (const MachineMemOperand *Memory : Instruction->memOperands()) - HasFrameIndex |= Memory->description().Base.frameIndex().has_value(); - } - } - const bool NoPHIs = !HasPHI; - const bool NoVRegs = - std::ranges::none_of(Function.registerInfo().virtualRegisters(), - [](const VirtualRegisterInfo &Info) { return !Info.IsTombstone; }); - - const auto ApplyComputed = [this, &Root, &Function](std::optional<bool> Explicit, bool Computed, - MachineFunctionProperty Property, - std::string_view Conflict) -> bool { - if (Explicit && *Explicit && !Computed) { - return fail(Root.range(), std::string(Function.name()) + std::string(Conflict)); - } - if (Explicit.value_or(Computed)) - Function.properties().set(Property); - else - Function.properties().clear(Property); - return false; - }; - if (ApplyComputed(DocumentInfo.NoPHIs, NoPHIs, MachineFunctionProperty::NoPHIs, - " has explicit property NoPhi, but contains at least one PHI") || - ApplyComputed(DocumentInfo.IsSSA, IsSSA, MachineFunctionProperty::IsSSA, - " has explicit property IsSSA, but is not valid SSA") || - ApplyComputed(DocumentInfo.NoVRegs, NoVRegs, MachineFunctionProperty::NoVRegs, - " has explicit property NoVRegs, but contains virtual registers")) { - return true; - } - - if (DocumentInfo.HasFakeUses && !*DocumentInfo.HasFakeUses && HasFakeUse) { - return fail(Root.range(), - std::string(Function.name()) + - " has explicit property hasFakeUses=false, but contains fake uses"); - } - if (DocumentInfo.HasFakeUses.value_or(HasFakeUse)) - Function.attributes().set(MachineFunctionAttribute::HasFakeUses); - else - Function.attributes().clear(MachineFunctionAttribute::HasFakeUses); - - const bool Selected = Function.properties().has(MachineFunctionProperty::Selected); - const bool StructurallyPostSSA = - Selected && NoPHIs && !Function.properties().has(MachineFunctionProperty::IsSSA); - if (AllTiedOperandsRewritten && (HasTiedOperands || StructurallyPostSSA)) - Function.properties().set(MachineFunctionProperty::TiedOpsRewritten); - else - Function.properties().clear(MachineFunctionProperty::TiedOpsRewritten); - - if (HasFrameIndex) - Function.properties().clear(MachineFunctionProperty::NoFrameIndices); - else - Function.properties().set(MachineFunctionProperty::NoFrameIndices); - - const MachineFrameState &FrameState = Function.frameInfo().state(); - const bool StructurallyFinalizedFrame = Selected && NoVRegs && !HasFrameIndex && - FrameState.IsCalleeSavedInfoValid && - FrameState.MaxAlignment.has_value(); - if (StructurallyFinalizedFrame) { - if (!Function.frameInfo().isFrozen()) { - auto Frozen = Function.frameInfo().freeze(FrameState.StackSize, *FrameState.MaxAlignment); - if (!Frozen) - return importError(Root.range(), Frozen.error(), "mir-frame-finalization"); - } - Function.properties().set(MachineFunctionProperty::FrameFinalized); - } else { - Function.properties().clear(MachineFunctionProperty::FrameFinalized); - } - - if (StructurallyFinalizedFrame && !HasExecutablePseudo) - Function.properties().set(MachineFunctionProperty::NoExecutablePseudos); - else - Function.properties().clear(MachineFunctionProperty::NoExecutablePseudos); - return false; -} - -bool MIRDocumentParser::parseRegisters(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context) { - const yaml::Node *Registers = mappingValue(Root, "registers"); - if (Registers != nullptr) { - if (requireKind(*Registers, yaml::NodeKind::Sequence, "registers")) - return true; - constexpr std::array RegisterKeys{ - std::string_view{"id"}, - std::string_view{"class"}, - std::string_view{"preferred-register"}, - std::string_view{"flags"}, - }; - std::unordered_set<std::uint32_t> IDs; - for (const yaml::NodeID ElementID : Registers->sequence()) { - const yaml::Node *Element = node(ElementID); - if (Element == nullptr || requireKind(*Element, yaml::NodeKind::Mapping, "register entry") || - validateKeys(*Element, RegisterKeys, "register")) { - return true; - } - const yaml::Node *IDNode = mappingValue(*Element, "id"); - const yaml::Node *ClassNode = mappingValue(*Element, "class"); - if (IDNode == nullptr || ClassNode == nullptr) - return fail(Element->range(), "register entry requires 'id' and 'class'"); - std::uint64_t WideID = 0; - std::string_view ClassName; - if (unsignedValue(*IDNode, WideID, "a 32-bit virtual-register ID") || - WideID > std::numeric_limits<std::uint32_t>::max() || - scalar(*ClassNode, ClassName, "register class or bank")) { - return true; - } - const auto TextID = static_cast<std::uint32_t>(WideID); - if (!IDs.insert(TextID).second) - return fail(IDNode->range(), "duplicate virtual register ID"); - auto Created = Function.registerInfo().createVirtualRegister(); - if (!Created) - return importError(IDNode->range(), Created.error()); - Context.VirtualRegisters.push_back({TextID, *Created}); - - if (ClassName != "_") { - if (const RegisterClassDesc *Class = Target.registers().registerClass(ClassName)) { - auto SetResult = Function.registerInfo().setRegisterClass(*Created, Class->ID); - if (!SetResult) - return importError(ClassNode->range(), SetResult.error()); - SetResult = Function.registerInfo().setRegisterBank(*Created, Class->DefaultBank); - if (!SetResult) - return importError(ClassNode->range(), SetResult.error()); - } else if (const RegisterBankDesc *Bank = Target.registers().registerBank(ClassName)) { - auto SetResult = Function.registerInfo().setRegisterBank(*Created, Bank->ID); - if (!SetResult) - return importError(ClassNode->range(), SetResult.error()); - } else { - return fail(ClassNode->range(), - "unknown register class or bank '" + std::string(ClassName) + "'"); - } - } - - if (const yaml::Node *Preferred = mappingValue(*Element, "preferred-register")) { - std::string_view PreferredName; - PhysicalRegister Reg; - if (scalar(*Preferred, PreferredName, "physical register")) - return true; - if (!trim(PreferredName).empty()) { - if (physicalRegister(PreferredName, Preferred->range(), Reg)) - return true; - auto AddResult = Function.registerInfo().addAllocationHint(*Created, Reg); - if (!AddResult) - return importError(Preferred->range(), AddResult.error()); - } - } - if (const yaml::Node *Flags = mappingValue(*Element, "flags")) { - if (requireKind(*Flags, yaml::NodeKind::Sequence, "virtual-register flags")) - return true; - std::uint8_t Mask = 0; - for (const yaml::NodeID FlagID : Flags->sequence()) { - const yaml::Node *FlagNode = node(FlagID); - std::string_view Name; - if (FlagNode == nullptr || scalar(*FlagNode, Name, "virtual-register flag")) - return true; - const auto Flag = Target.mirText().virtualRegisterFlag(Name); - if (!Flag) - return fail(FlagNode->range(), - "use of undefined register flag '" + std::string(Name) + "'"); - if ((Mask & *Flag) != 0) - return fail(FlagNode->range(), "duplicate virtual-register flag"); - Mask = static_cast<std::uint8_t>(Mask | *Flag); - } - auto SetResult = Function.registerInfo().setTargetRegisterFlags(*Created, Mask); - if (!SetResult) - return importError(Flags->range(), SetResult.error()); - } - } - } - - const yaml::Node *LiveIns = mappingValue(Root, "liveins"); - if (LiveIns != nullptr) { - if (requireKind(*LiveIns, yaml::NodeKind::Sequence, "liveins")) - return true; - constexpr std::array LiveInKeys{ - std::string_view{"reg"}, - std::string_view{"virtual-reg"}, - }; - for (const yaml::NodeID ElementID : LiveIns->sequence()) { - const yaml::Node *Element = node(ElementID); - if (Element == nullptr || requireKind(*Element, yaml::NodeKind::Mapping, "live-in entry") || - validateKeys(*Element, LiveInKeys, "live-in")) { - return true; - } - const yaml::Node *RegNode = mappingValue(*Element, "reg"); - if (RegNode == nullptr) - return fail(Element->range(), "live-in entry requires 'reg'"); - PhysicalRegister Reg; - if (physicalRegister(*RegNode, Reg)) - return true; - std::optional<Register> Initial; - if (const yaml::Node *VirtualNode = mappingValue(*Element, "virtual-reg")) { - std::string_view Text; - if (scalar(*VirtualNode, Text, "virtual register")) - return true; - if (!trim(Text).empty()) { - Register Virtual; - if (virtualRegister(Text, VirtualNode->range(), Function, Context, Virtual)) - return true; - Initial = Virtual; - } - } - auto AddResult = Function.registerInfo().addLiveIn(Reg, Initial); - if (!AddResult) - return importError(Element->range(), AddResult.error()); - } - } - - if (const yaml::Node *CalleeSaved = mappingValue(Root, "calleeSavedRegisters")) { - if (requireKind(*CalleeSaved, yaml::NodeKind::Sequence, "callee-saved registers")) - return true; - std::vector<PhysicalRegister> RegistersValue; - for (const yaml::NodeID RegID : CalleeSaved->sequence()) { - const yaml::Node *RegNode = node(RegID); - PhysicalRegister Reg; - if (RegNode == nullptr || physicalRegister(*RegNode, Reg)) - return true; - RegistersValue.push_back(Reg); - } - auto SetResult = Function.records().setCalleeSavedRegisters(std::move(RegistersValue)); - if (!SetResult) - return importError(CalleeSaved->range(), SetResult.error()); - } - return false; -} - -TextRange MIRDocumentParser::scalarDiagnosticRange(const yaml::Node &Scalar, - const TextRange &ScalarRange) const noexcept { - if (Scalar.scalarStyle() != yaml::ScalarStyle::Literal && - Scalar.scalarStyle() != yaml::ScalarStyle::Folded) { - std::size_t ContentOffset = Scalar.range().Begin.Offset; - if (Scalar.scalarStyle() == yaml::ScalarStyle::SingleQuoted || - Scalar.scalarStyle() == yaml::ScalarStyle::DoubleQuoted) { - ++ContentOffset; - } - return Result.Source.range(ContentOffset + ScalarRange.Begin.Offset, - ContentOffset + ScalarRange.End.Offset); - } - - const std::size_t FirstBodyLine = Scalar.range().Begin.Line + 1U; - const std::size_t LastBodyLine = Scalar.range().End.Line; - if (FirstBodyLine > LastBodyLine || FirstBodyLine == 0) - return Scalar.range(); - - std::size_t Indent = 0; - bool FoundContent = false; - for (std::size_t Line = FirstBodyLine; Line <= LastBodyLine; ++Line) { - const std::string_view Text = Result.Source.line(Line); - std::size_t CurrentIndent = 0; - while (CurrentIndent < Text.size() && Text[CurrentIndent] == ' ') - ++CurrentIndent; - if (CurrentIndent == Text.size()) - continue; - Indent = CurrentIndent; - FoundContent = true; - break; - } - if (!FoundContent) - return Scalar.range(); - - const TextPosition DefaultPosition = Scalar.range().Begin; - const auto MapPosition = [this, FirstBodyLine, LastBodyLine, Indent, - DefaultPosition](TextPosition Position) { - const std::size_t SourceLine = FirstBodyLine + Position.Line - 1U; - if (Position.Line == 0 || SourceLine > LastBodyLine || SourceLine > Result.Source.lineCount()) { - return DefaultPosition; - } - const std::string_view SourceLineText = Result.Source.line(SourceLine); - const std::size_t BodyColumn = Position.Column == 0 ? 0 : Position.Column - 1U; - const std::size_t SourceColumn = std::min(Indent + BodyColumn, SourceLineText.size()); - const std::size_t Offset = Result.Source.lineOffsets()[SourceLine - 1U] + SourceColumn; - return Result.Source.positionAt(Offset); - }; - return {.Begin = MapPosition(ScalarRange.Begin), .End = MapPosition(ScalarRange.End)}; -} - -bool MIRDocumentParser::parseBody(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context) { - const yaml::Node *Body = mappingValue(Root, "body"); - std::string BodyText; - TextRange BodyRange = Root.range(); - if (Body != nullptr) { - std::string_view Text; - if (scalar(*Body, Text, "machine function body")) - return true; - BodyText = Text; - BodyRange = Body->range(); - } - - if (!effectiveDataLayout().empty()) { - const LLVMConstantSyntaxOptions PointerOptions = pointerLayoutOptions(); - Context.PointerSizeInBits = [PointerOptions](std::uint32_t AddressSpace) { - return llvmPointerSizeInBits(AddressSpace, PointerOptions); - }; - } - Context.SyncScopes = ActiveDocument->SyncScopes; - Context.InternSyncScope = [this](std::string_view Name) -> Expected<SyncScopeID> { - if (Name.empty()) - return makeUnexpected(ErrorCode::ParseError, "synchronization scope name cannot be empty"); - const auto Existing = - std::ranges::find_if(ActiveDocument->SyncScopes, - [Name](const MISyncScopeSlot &Slot) { return Slot.Name == Name; }); - if (Existing != ActiveDocument->SyncScopes.end()) - return Existing->ID; - if (ActiveDocument->SyncScopes.size() >= - static_cast<std::size_t>(std::numeric_limits<std::uint32_t>::max())) { - return makeUnexpected(ErrorCode::Unsupported, - "MIR document contains too many synchronization scopes"); - } - const auto ID = - SyncScopeID::fromValue(static_cast<std::uint32_t>(ActiveDocument->SyncScopes.size() + 1U)); - ActiveDocument->SyncScopes.push_back({.Name = std::string(Name), .ID = ID}); - return ID; - }; - MIParserOptions BodyOptions = Options.MachineInstructions; - BodyOptions.VerifyAfterParse = false; - MIParser Parser(Function, Target, - std::string(Result.Source.name()) + ":" + std::string(Function.name()), - std::move(BodyText), Context, BodyOptions); - auto Parsed = Parser.parse(); - if (!Parsed) { - if (!Parser.diagnostics().empty()) { - const TextRange Range = - Body == nullptr ? BodyRange - : scalarDiagnosticRange(*Body, Parser.diagnostics().front().Range); - return fail(Range, Parser.diagnostics().front().Message, "mir-body"); - } - return importError(BodyRange, Parsed.error(), "mir-body"); - } - return false; -} - -bool MIRDocumentParser::convertTargetNode(const yaml::Node &Source, MIRTargetNode &Destination) { - switch (Source.kind()) { - case yaml::NodeKind::Null: - Destination = MIRTargetNode::null(); - return false; - case yaml::NodeKind::Scalar: - Destination = MIRTargetNode::scalar(std::string(Source.scalar())); - return false; - case yaml::NodeKind::Sequence: { - std::vector<MIRTargetNode> Values; - Values.reserve(Source.sequence().size()); - for (const yaml::NodeID ElementID : Source.sequence()) { - const yaml::Node *Element = node(ElementID); - if (Element == nullptr) - return fail(Source.range(), "target MIR sequence contains an invalid node"); - MIRTargetNode Value; - if (convertTargetNode(*Element, Value)) - return true; - Values.push_back(std::move(Value)); - } - Destination = MIRTargetNode::sequence(std::move(Values)); - return false; - } - case yaml::NodeKind::Mapping: { - std::vector<MIRTargetMappingEntry> Entries; - Entries.reserve(Source.mapping().size()); - for (const yaml::MappingEntry &SourceEntry : Source.mapping()) { - const yaml::Node *ValueNode = node(SourceEntry.Value); - if (ValueNode == nullptr) - return fail(SourceEntry.KeyRange, "target MIR mapping contains an invalid node"); - MIRTargetNode Value; - if (convertTargetNode(*ValueNode, Value)) - return true; - Entries.push_back({SourceEntry.Key, std::move(Value)}); - } - Destination = MIRTargetNode::mapping(std::move(Entries)); - return false; - } - } - return fail(Source.range(), "target MIR node has an unknown kind"); -} - -bool MIRDocumentParser::parseTargetState(const yaml::Node &Root, MachineFunction &Function) { - const yaml::Node *StateNode = mappingValue(Root, "machineFunctionInfo"); - if (StateNode == nullptr) - return false; - MIRTargetNode TargetNode; - if (convertTargetNode(*StateNode, TargetNode)) - return true; - auto State = Target.mirText().parseFunctionState(TargetNode); - if (!State) - return importError(StateNode->range(), State.error(), "mir-target-state"); - auto SetResult = Function.setTargetState(std::move(*State)); - if (!SetResult) - return importError(StateNode->range(), SetResult.error(), "mir-target-state"); - return false; -} - -bool MIRDocumentParser::validateDocumentReferences(const MachineFunction &Function) { - auto Verification = verifyMIRDocumentReferences(*ActiveDocument, Function, true); - if (Verification) - return false; - return importError(Result.Source.range(0, 0), Verification.error()); -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRFunctionRecordsParser.cpp b/backend/lib/MIR/MIRFunctionRecordsParser.cpp deleted file mode 100644 index 04857bb..0000000 --- a/backend/lib/MIR/MIRFunctionRecordsParser.cpp +++ /dev/null @@ -1,341 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentParser.h" -#include "MIRDocumentReferenceValidation.h" -#include "MIRMetadataTextParser.h" -#include "MIRParserUtilities.h" -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionAttributes.h" -#include "rcc/CodeGen/MachineFunctionProperties.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/MachineRegisterInfo.h" -#include "rcc/CodeGen/MachineVerifier.h" -#include "rcc/CodeGen/TargetDescription.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/CodeGen/TargetRegisterInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" - -#include <algorithm> -#include <array> -#include <cstddef> -#include <cstdint> -#include <initializer_list> -#include <limits> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <unordered_map> -#include <unordered_set> -#include <utility> -#include <vector> - -namespace rcc::mir { - -using parser_detail::parseBool; -using parser_detail::parsePrefixedID; -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; -using parser_detail::trim; - -bool MIRDocumentParser::parseFunctionRecords(const yaml::Node &Root, MachineFunction &Function) { - auto ParseLocation = [this, &Function](const yaml::Node &Entry, - MachineInstr *&Instruction) -> bool { - const yaml::Node *BlockNode = mappingValue(Entry, "bb"); - const yaml::Node *OffsetNode = mappingValue(Entry, "offset"); - if (BlockNode == nullptr || OffsetNode == nullptr) - return fail(Entry.range(), "instruction location requires 'bb' and 'offset'"); - std::uint64_t Block = 0; - std::uint64_t Offset = 0; - if (unsignedValue(*BlockNode, Block, "a 32-bit basic block number") || - unsignedValue(*OffsetNode, Offset, "a 32-bit instruction offset") || - Block > std::numeric_limits<std::uint32_t>::max() || - Offset > std::numeric_limits<std::uint32_t>::max()) { - return true; - } - Instruction = instructionAt(Function, static_cast<std::uint32_t>(Block), - static_cast<std::uint32_t>(Offset)); - if (Instruction == nullptr) - return fail(Entry.range(), "instruction location is outside the machine function body"); - return false; - }; - - if (const yaml::Node *CallSites = mappingValue(Root, "callSites")) { - if (requireKind(*CallSites, yaml::NodeKind::Sequence, "call-site records")) - return true; - constexpr std::array CallSiteKeys{ - std::string_view{"bb"}, - std::string_view{"offset"}, - std::string_view{"fwdArgRegs"}, - std::string_view{"calleeTypeIds"}, - }; - constexpr std::array ArgumentKeys{ - std::string_view{"arg"}, - std::string_view{"reg"}, - }; - for (const yaml::NodeID CallSiteID : CallSites->sequence()) { - const yaml::Node *CallSiteNode = node(CallSiteID); - if (CallSiteNode == nullptr || - requireKind(*CallSiteNode, yaml::NodeKind::Mapping, "call-site record") || - validateKeys(*CallSiteNode, CallSiteKeys, "call-site record")) { - return true; - } - MachineInstr *Call = nullptr; - if (ParseLocation(*CallSiteNode, Call)) - return true; - if (Call == nullptr) - return fail(CallSiteNode->range(), - "instruction location is outside the machine function body"); - MachineCallSiteInfo Info{ - .CallInstruction = Call->stableID(), - .ForwardedArguments = {}, - .CalleeTypeIDs = {}, - }; - if (const yaml::Node *Arguments = mappingValue(*CallSiteNode, "fwdArgRegs")) { - if (requireKind(*Arguments, yaml::NodeKind::Sequence, "forwarded call arguments")) { - return true; - } - for (const yaml::NodeID ArgumentID : Arguments->sequence()) { - const yaml::Node *Argument = node(ArgumentID); - if (Argument == nullptr || - requireKind(*Argument, yaml::NodeKind::Mapping, "forwarded call argument") || - validateKeys(*Argument, ArgumentKeys, "forwarded call argument")) { - return true; - } - const yaml::Node *NumberNode = mappingValue(*Argument, "arg"); - const yaml::Node *RegNode = mappingValue(*Argument, "reg"); - if (NumberNode == nullptr || RegNode == nullptr) - return fail(Argument->range(), "forwarded call argument requires 'arg' and 'reg'"); - std::uint16_t Number = 0; - std::uint64_t WideNumber = 0; - PhysicalRegister Reg; - if (unsignedValue(*NumberNode, WideNumber, "a 16-bit argument number") || - WideNumber > std::numeric_limits<std::uint16_t>::max() || - physicalRegister(*RegNode, Reg)) { - return true; - } - Number = static_cast<std::uint16_t>(WideNumber); - Info.ForwardedArguments.push_back({Reg, Number}); - } - } - if (const yaml::Node *TypeIDs = mappingValue(*CallSiteNode, "calleeTypeIds")) { - if (requireKind(*TypeIDs, yaml::NodeKind::Sequence, "callee type IDs")) - return true; - for (const yaml::NodeID TypeID : TypeIDs->sequence()) { - const yaml::Node *TypeNode = node(TypeID); - std::uint64_t Value = 0; - if (TypeNode == nullptr || - unsignedValue(*TypeNode, Value, "an unsigned callee type ID")) { - return true; - } - Info.CalleeTypeIDs.push_back(Value); - } - } - auto AddResult = Function.records().addCallSite(std::move(Info)); - if (!AddResult) - return importError(CallSiteNode->range(), AddResult.error()); - } - } - - if (const yaml::Node *CalledGlobals = mappingValue(Root, "calledGlobals")) { - if (requireKind(*CalledGlobals, yaml::NodeKind::Sequence, "called-global records")) - return true; - constexpr std::array Keys{ - std::string_view{"bb"}, - std::string_view{"offset"}, - std::string_view{"callee"}, - std::string_view{"flags"}, - }; - for (const yaml::NodeID CalledID : CalledGlobals->sequence()) { - const yaml::Node *CalledNode = node(CalledID); - if (CalledNode == nullptr || - requireKind(*CalledNode, yaml::NodeKind::Mapping, "called-global record") || - validateKeys(*CalledNode, Keys, "called-global record")) { - return true; - } - MachineInstr *Call = nullptr; - if (ParseLocation(*CalledNode, Call)) - return true; - if (Call == nullptr) - return fail(CalledNode->range(), - "instruction location is outside the machine function body"); - const yaml::Node *CalleeNode = mappingValue(*CalledNode, "callee"); - const yaml::Node *FlagsNode = mappingValue(*CalledNode, "flags"); - if (CalleeNode == nullptr || FlagsNode == nullptr) - return fail(CalledNode->range(), "called-global record requires 'callee' and 'flags'"); - std::string_view Callee; - std::uint64_t WideFlags = 0; - if (scalar(*CalleeNode, Callee, "callee name") || - unsignedValue(*FlagsNode, WideFlags, "32-bit called-global flags") || - WideFlags > std::numeric_limits<std::uint32_t>::max()) { - return true; - } - if (ActiveDocument->EmbeddedIR && !ActiveDocument->EmbeddedIR->hasGlobalValue(Callee)) { - return fail(CalleeNode->range(), "called global references an undefined embedded global"); - } - auto AddResult = - Function.records().addCalledGlobal({.CallInstruction = Call->stableID(), - .Callee = std::string(Callee), - .Flags = static_cast<std::uint32_t>(WideFlags)}); - if (!AddResult) - return importError(CalledNode->range(), AddResult.error()); - } - } - - if (const yaml::Node *Substitutions = mappingValue(Root, "debugValueSubstitutions")) { - if (requireKind(*Substitutions, yaml::NodeKind::Sequence, "debug value substitutions")) { - return true; - } - constexpr std::array Keys{ - std::string_view{"srcinst"}, std::string_view{"srcop"}, std::string_view{"dstinst"}, - std::string_view{"dstop"}, std::string_view{"subreg"}, - }; - for (const yaml::NodeID SubstitutionID : Substitutions->sequence()) { - const yaml::Node *Substitution = node(SubstitutionID); - if (Substitution == nullptr || - requireKind(*Substitution, yaml::NodeKind::Mapping, "debug value substitution") || - validateKeys(*Substitution, Keys, "debug value substitution")) { - return true; - } - std::array<std::uint32_t, 5> Values{}; - for (std::size_t Index = 0; Index < Keys.size(); ++Index) { - const yaml::Node *ValueNode = mappingValue(*Substitution, Keys[Index]); - if (ValueNode == nullptr) - return fail(Substitution->range(), "debug value substitution requires every fixed field"); - std::uint64_t Wide = 0; - if (unsignedValue(*ValueNode, Wide, "a 32-bit debug substitution value") || - Wide > std::numeric_limits<std::uint32_t>::max()) { - return true; - } - Values[Index] = static_cast<std::uint32_t>(Wide); - } - auto AddResult = Function.records().addDebugValueSubstitution( - {.Source = {Values[0], Values[1]}, - .Destination = {Values[2], Values[3]}, - .Subregister = SubRegIndex::fromValue(Values[4])}); - if (!AddResult) - return importError(Substitution->range(), AddResult.error()); - } - } - - if (const yaml::Node *PrefetchTargets = mappingValue(Root, "prefetch-targets")) { - if (requireKind(*PrefetchTargets, yaml::NodeKind::Sequence, "prefetch targets")) - return true; - for (const yaml::NodeID TargetID : PrefetchTargets->sequence()) { - const yaml::Node *TargetNode = node(TargetID); - std::string_view Text; - if (TargetNode == nullptr || scalar(*TargetNode, Text, "prefetch target")) - return true; - - std::size_t ContentOffset = TargetNode->range().Begin.Offset; - if (TargetNode->scalarStyle() == yaml::ScalarStyle::SingleQuoted || - TargetNode->scalarStyle() == yaml::ScalarStyle::DoubleQuoted) { - ++ContentOffset; - } - const auto ErrorRange = [this, ContentOffset](std::size_t Begin, std::size_t End = 1U) { - return Result.Source.range(ContentOffset + Begin, ContentOffset + Begin + End); - }; - const auto SkipSpace = [&Text](std::size_t &Cursor) { - while (Cursor < Text.size() && (Text[Cursor] == ' ' || Text[Cursor] == '\t')) - ++Cursor; - }; - const auto ParseNumber = [&Text, &SkipSpace](std::size_t &Cursor, std::uint32_t &Value) { - SkipSpace(Cursor); - const std::size_t Begin = Cursor; - while (Cursor < Text.size() && Text[Cursor] >= '0' && Text[Cursor] <= '9') - ++Cursor; - return Begin != Cursor && parseUnsigned32(Text.substr(Begin, Cursor - Begin), Value); - }; - - std::size_t Cursor = 0; - SkipSpace(Cursor); - if (!Text.substr(Cursor).starts_with("bb_id") || - (Cursor + std::string_view("bb_id").size() < Text.size() && - Text[Cursor + std::string_view("bb_id").size()] != ' ' && - Text[Cursor + std::string_view("bb_id").size()] != '\t')) { - return fail(ErrorRange(Cursor), "expected 'bb_id'"); - } - Cursor += std::string_view("bb_id").size(); - - std::array<std::uint32_t, 3> Values{}; - if (!ParseNumber(Cursor, Values[0])) - return fail(ErrorRange(0), "expected unsigned integer"); - - SkipSpace(Cursor); - if (Cursor < Text.size() && (Text[Cursor] == ',' || Text[Cursor] == '.')) { - ++Cursor; - if (!ParseNumber(Cursor, Values[1])) - return fail(ErrorRange(0), "expected unsigned integer"); - } else if (Cursor < Text.size() && Text[Cursor] >= '0' && Text[Cursor] <= '9') { - if (!ParseNumber(Cursor, Values[1])) - return fail(ErrorRange(0), "expected unsigned integer"); - } - - SkipSpace(Cursor); - if (Cursor >= Text.size() || Text[Cursor] != ',') - return fail(ErrorRange(Cursor), "expected ','"); - ++Cursor; - if (!ParseNumber(Cursor, Values[2])) - return fail(ErrorRange(0), "expected unsigned integer"); - SkipSpace(Cursor); - if (Cursor != Text.size()) - return fail(ErrorRange(0), "unexpected content after prefetch target"); - - auto AddResult = Function.records().addPrefetchTarget( - {.Block = {Values[0], Values[1]}, .CallsiteIndex = Values[2]}); - if (!AddResult) - return importError(TargetNode->range(), AddResult.error()); - } - } - - if (const yaml::Node *EntryValues = mappingValue(Root, "entry_values")) { - if (requireKind(*EntryValues, yaml::NodeKind::Sequence, "entry values")) - return true; - constexpr std::array Keys{ - std::string_view{"entry-value-register"}, - std::string_view{"debug-info-variable"}, - std::string_view{"debug-info-expression"}, - std::string_view{"debug-info-location"}, - }; - for (const yaml::NodeID EntryID : EntryValues->sequence()) { - const yaml::Node *Entry = node(EntryID); - if (Entry == nullptr || requireKind(*Entry, yaml::NodeKind::Mapping, "entry value") || - validateKeys(*Entry, Keys, "entry value")) { - return true; - } - const yaml::Node *RegNode = mappingValue(*Entry, "entry-value-register"); - const yaml::Node *VariableNode = mappingValue(*Entry, "debug-info-variable"); - const yaml::Node *ExpressionNode = mappingValue(*Entry, "debug-info-expression"); - const yaml::Node *LocationNode = mappingValue(*Entry, "debug-info-location"); - if (RegNode == nullptr || VariableNode == nullptr || ExpressionNode == nullptr || - LocationNode == nullptr) { - return fail(Entry->range(), "entry value requires all four fixed fields"); - } - PhysicalRegister Reg; - if (physicalRegister(*RegNode, Reg)) - return true; - MachineVariableDebugInfo DebugInfo; - if (parseVariableDebugInfo(*Entry, Function, DebugInfo)) - return true; - if (!DebugInfo.Variable || !DebugInfo.Expression || !DebugInfo.Location) { - return fail(Entry->range(), "entry value debug metadata cannot be empty", - "mir-debug-metadata"); - } - auto AddResult = - Function.records().addEntryValue({.Reg = Reg, .DebugInfo = std::move(DebugInfo)}); - if (!AddResult) - return importError(Entry->range(), AddResult.error()); - } - } - return false; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRMetadataParser.cpp b/backend/lib/MIR/MIRMetadataParser.cpp deleted file mode 100644 index 63727ee..0000000 --- a/backend/lib/MIR/MIRMetadataParser.cpp +++ /dev/null @@ -1,333 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRDocumentParser.h" -#include "MIRDocumentReferenceValidation.h" -#include "MIRMetadataTextParser.h" -#include "MIRParserUtilities.h" -#include "rcc/CodeGen/MachineBasicBlock.h" -#include "rcc/CodeGen/MachineConstantPool.h" -#include "rcc/CodeGen/MachineFrameInfo.h" -#include "rcc/CodeGen/MachineFunction.h" -#include "rcc/CodeGen/MachineFunctionAttributes.h" -#include "rcc/CodeGen/MachineFunctionProperties.h" -#include "rcc/CodeGen/MachineFunctionRecords.h" -#include "rcc/CodeGen/MachineInstr.h" -#include "rcc/CodeGen/MachineJumpTableInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/CodeGen/MachineRegisterInfo.h" -#include "rcc/CodeGen/MachineVerifier.h" -#include "rcc/CodeGen/TargetDescription.h" -#include "rcc/CodeGen/TargetMIRTextInfo.h" -#include "rcc/CodeGen/TargetRegisterInfo.h" -#include "rcc/MIR/LLVMConstantSyntax.h" -#include "rcc/MIR/MILexer.h" - -#include <algorithm> -#include <array> -#include <cstddef> -#include <cstdint> -#include <initializer_list> -#include <limits> -#include <optional> -#include <ranges> -#include <span> -#include <string> -#include <string_view> -#include <unordered_map> -#include <unordered_set> -#include <utility> -#include <vector> - -namespace rcc::mir { - -using parser_detail::parseBool; -using parser_detail::parsePrefixedID; -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; -using parser_detail::trim; - -Expected<MachineMetadataReference> -MIRDocumentParser::resolveMetadataReference(const MachineFunction &Function, - std::uint32_t Slot) const { - if (ActiveDocument != nullptr && ActiveDocument->EmbeddedIR && - ActiveDocument->EmbeddedIR->metadata(std::to_string(Slot)) != nullptr) { - return MachineMetadataReference::embeddedIR(Slot); - } - if (Function.metadata().node(Slot) != nullptr) - return MachineMetadataReference::machineFunction(Slot); - return makeUnexpected(ErrorCode::ParseError, - "metadata reference uses undefined slot '!" + std::to_string(Slot) + "'"); -} - -bool MIRDocumentParser::parseVariableDebugInfo(const yaml::Node &Entry, MachineFunction &Function, - MachineVariableDebugInfo &DebugInfo) { - const auto Resolve = [this, &Function](std::uint32_t Slot) { - return resolveMetadataReference(Function, Slot); - }; - if (const yaml::Node *VariableNode = mappingValue(Entry, "debug-info-variable")) { - std::string_view Text; - if (scalar(*VariableNode, Text, "debug variable")) - return true; - if (!Text.empty()) { - MetadataTextParser Parser(std::string(Text), Resolve); - auto Reference = Parser.parseReferenceOnly(); - if (!Reference) - return importError(VariableNode->range(), Reference.error(), "mir-debug-metadata"); - if (Reference->scope() != MachineMetadataScope::EmbeddedIR) { - return fail(VariableNode->range(), - "debug variable must reference embedded LLVM IR metadata", - "mir-debug-metadata"); - } - DebugInfo.Variable = *Reference; - } - } - if (const yaml::Node *ExpressionNode = mappingValue(Entry, "debug-info-expression")) { - std::string_view Text; - if (scalar(*ExpressionNode, Text, "debug expression")) - return true; - if (!Text.empty()) { - MetadataTextParser Parser(std::string(Text), Resolve); - auto Value = Parser.parseValue(); - if (!Value) - return importError(scalarDiagnosticRange(*ExpressionNode, Parser.diagnosticRange()), - Value.error(), "mir-debug-metadata"); - if (Value->reference() != nullptr && - Value->reference()->scope() != MachineMetadataScope::EmbeddedIR) { - return fail(ExpressionNode->range(), - "debug expression reference must use embedded LLVM IR metadata", - "mir-debug-metadata"); - } - DebugInfo.Expression = std::move(*Value); - } - } - if (const yaml::Node *LocationNode = mappingValue(Entry, "debug-info-location")) { - std::string_view Text; - if (scalar(*LocationNode, Text, "debug location")) - return true; - if (!Text.empty()) { - MetadataTextParser Parser(std::string(Text), Resolve); - auto Location = Parser.parseLocation(); - if (!Location) - return importError(LocationNode->range(), Location.error(), "mir-debug-metadata"); - if (Location->reference() != nullptr && - Location->reference()->scope() != MachineMetadataScope::EmbeddedIR) { - return fail(LocationNode->range(), - "variable debug location must reference embedded LLVM IR metadata", - "mir-debug-metadata"); - } - DebugInfo.Location = std::move(*Location); - } - } - if (!DebugInfo.Variable && !DebugInfo.Expression && !DebugInfo.Location) - return false; - auto Verification = DebugInfo.verifyStructure(); - if (!Verification) - return importError(Entry.range(), Verification.error(), "mir-debug-metadata"); - return false; -} - -bool MIRDocumentParser::parseMetadata(const yaml::Node &Root, MachineFunction &Function, - MIParserContext &Context) { - if (ActiveDocument->EmbeddedIR) { - for (const EmbeddedIRMetadataDefinition &Reference : - ActiveDocument->EmbeddedIR->metadataDefinitions()) { - std::uint32_t ID = 0; - if (!parseUnsigned32(Reference.Name, ID)) - continue; - if (std::ranges::any_of(Context.Metadata, - [ID](const MIMetadataSlot &Slot) { return Slot.TextID == ID; })) { - return fail(Root.range(), "embedded LLVM IR metadata slots are not unique"); - } - Context.Metadata.push_back( - {.TextID = ID, .Reference = MachineMetadataReference::embeddedIR(ID)}); - } - } - - const yaml::Node *Metadata = mappingValue(Root, "machineMetadataNodes"); - if (Metadata == nullptr) - return false; - if (requireKind(*Metadata, yaml::NodeKind::Sequence, "machine metadata nodes")) - return true; - - enum class ParsedElementKind : std::uint8_t { - Null, - Reference, - String, - }; - struct ParsedElement final { - ParsedElementKind Kind = ParsedElementKind::Reference; - std::uint32_t Reference = 0; - std::string String; - TextRange Range; - }; - struct Definition final { - const yaml::Node *Node = nullptr; - std::uint32_t Slot = 0; - bool IsDistinct = false; - std::vector<ParsedElement> Elements; - }; - std::vector<Definition> Definitions; - std::unordered_set<std::uint32_t> Slots; - - const auto TranslateRange = [this](const yaml::Node &Node, const TextRange &Range) { - std::size_t ContentBegin = Node.range().Begin.Offset; - if (Node.scalarStyle() == yaml::ScalarStyle::SingleQuoted || - Node.scalarStyle() == yaml::ScalarStyle::DoubleQuoted) { - ++ContentBegin; - } - return Result.Source.range(ContentBegin + Range.Begin.Offset, ContentBegin + Range.End.Offset); - }; - - for (const yaml::NodeID DefinitionID : Metadata->sequence()) { - const yaml::Node *DefinitionNode = node(DefinitionID); - std::string_view Text; - if (DefinitionNode == nullptr || scalar(*DefinitionNode, Text, "machine metadata definition")) - return true; - - MILexer Lexer("machine-metadata", std::string(Text)); - MIToken Token = Lexer.next(); - const auto LexerFailed = [this, &Lexer, DefinitionNode, &TranslateRange, &Token]() -> bool { - if (!Token.is(MITokenKind::Error)) - return false; - const auto Diagnostics = Lexer.diagnostics(); - const TextRange Range = Diagnostics.empty() - ? TranslateRange(*DefinitionNode, Token.range()) - : TranslateRange(*DefinitionNode, Diagnostics.front().Range); - return fail(Range, Diagnostics.empty() ? "invalid machine metadata node" - : Diagnostics.front().Message); - }; - const auto Advance = [&Lexer, &Token, &LexerFailed]() -> bool { - Token = Lexer.next(); - return LexerFailed(); - }; - const auto FailAtToken = [this, DefinitionNode, &TranslateRange](const MIToken &At, - std::string Message) { - return fail(TranslateRange(*DefinitionNode, At.range()), std::move(Message)); - }; - - if (LexerFailed()) - return true; - if (!Token.is(MITokenKind::Exclaim)) - return FailAtToken(Token, "expected a metadata node"); - if (Advance()) - return true; - std::uint32_t Slot = 0; - if (!Token.is(MITokenKind::IntegerLiteral) || !parseUnsigned32(Token.integerValue(), Slot)) { - return FailAtToken(Token, "expected metadata id after '!'"); - } - if (Advance()) - return true; - if (!Token.is(MITokenKind::Equal)) - return FailAtToken(Token, "expected '='"); - if (Advance()) - return true; - - const bool IsDistinct = Token.isKeyword("distinct"); - if (IsDistinct && Advance()) - return true; - if (!Token.is(MITokenKind::Exclaim)) - return FailAtToken(Token, "expected a metadata node"); - if (Advance()) - return true; - if (!Token.is(MITokenKind::LeftBrace)) - return FailAtToken(Token, "expected '{' here"); - if (Advance()) - return true; - - std::vector<ParsedElement> Elements; - if (!Token.is(MITokenKind::RightBrace)) { - while (true) { - if (Token.hasStringValue() && Token.stringValue() == "null") { - Elements.push_back({.Kind = ParsedElementKind::Null, - .Reference = 0, - .String = {}, - .Range = TranslateRange(*DefinitionNode, Token.range())}); - } else { - if (!Token.is(MITokenKind::Exclaim)) - return FailAtToken(Token, "expected '!' here"); - const TextRange ReferenceRange = TranslateRange(*DefinitionNode, Token.range()); - if (Advance()) - return true; - - if (Token.is(MITokenKind::StringConstant)) { - Elements.push_back({.Kind = ParsedElementKind::String, - .Reference = 0, - .String = std::string(Token.stringValue()), - .Range = TranslateRange(*DefinitionNode, Token.range())}); - } else { - std::uint32_t Reference = 0; - if (!Token.is(MITokenKind::IntegerLiteral) || - !parseUnsigned32(Token.integerValue(), Reference)) { - return FailAtToken(Token, "expected metadata id after '!'"); - } - Elements.push_back({.Kind = ParsedElementKind::Reference, - .Reference = Reference, - .String = {}, - .Range = ReferenceRange}); - } - } - if (Advance()) - return true; - if (!Token.is(MITokenKind::Comma)) - break; - if (Advance()) - return true; - } - } - if (!Token.is(MITokenKind::RightBrace)) - return FailAtToken(Token, "expected end of metadata node"); - if (Advance()) - return true; - if (!Token.is(MITokenKind::EndOfFile)) - return FailAtToken(Token, "expected end of metadata node"); - if (!Slots.insert(Slot).second) - return FailAtToken(Token, "Metadata id is already used"); - - Definitions.push_back({.Node = DefinitionNode, - .Slot = Slot, - .IsDistinct = IsDistinct, - .Elements = std::move(Elements)}); - Context.Metadata.push_back( - {.TextID = Slot, .Reference = MachineMetadataReference::machineFunction(Slot)}); - } - - for (const Definition &Definition : Definitions) { - MachineMetadataNode NewNode{ - .Slot = Definition.Slot, - .IsDistinct = Definition.IsDistinct, - .Elements = {}, - }; - for (const ParsedElement &Element : Definition.Elements) { - if (Element.Kind == ParsedElementKind::Null) { - NewNode.Elements.push_back(MachineMetadataElement::null()); - continue; - } - if (Element.Kind == ParsedElementKind::String) { - NewNode.Elements.push_back(MachineMetadataElement::string(Element.String)); - continue; - } - - MachineMetadataReference Reference; - if (ActiveDocument->EmbeddedIR != std::nullopt && - ActiveDocument->EmbeddedIR->metadata(std::to_string(Element.Reference)) != nullptr) { - Reference = MachineMetadataReference::embeddedIR(Element.Reference); - } else if (Slots.contains(Element.Reference)) { - Reference = MachineMetadataReference::machineFunction(Element.Reference); - } else { - return fail(Element.Range, - "use of undefined metadata '!" + std::to_string(Element.Reference) + "'"); - } - auto Resolved = MachineMetadataElement::reference(Reference); - if (!Resolved) - return importError(Element.Range, Resolved.error()); - NewNode.Elements.push_back(std::move(*Resolved)); - } - auto AddResult = Function.metadata().addNode(std::move(NewNode)); - if (!AddResult) - return importError(Definition.Node->range(), AddResult.error()); - } - return false; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRMetadataTextParser.cpp b/backend/lib/MIR/MIRMetadataTextParser.cpp deleted file mode 100644 index 7875b06..0000000 --- a/backend/lib/MIR/MIRMetadataTextParser.cpp +++ /dev/null @@ -1,259 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -#include "MIRMetadataTextParser.h" - -#include "MIRParserUtilities.h" -#include "rcc/MIR/MIDebugSyntax.h" - -#include <cstdint> -#include <limits> -#include <optional> -#include <string> -#include <string_view> -#include <unordered_set> -#include <utility> -#include <vector> - -namespace rcc::mir { - -using parser_detail::parseSigned; -using parser_detail::parseUnsigned; -using parser_detail::parseUnsigned32; - -Expected<void> MetadataTextParser::expect(MITokenKind Kind, std::string_view Description) { - if (consume(Kind)) - return {}; - return makeUnexpected(ErrorCode::ParseError, - "expected " + std::string(Description) + " in MIR metadata scalar"); -} - -Expected<void> MetadataTextParser::finish() { - while (consume(MITokenKind::Newline)) { - } - if (Lexer.failed()) { - if (!Lexer.diagnostics().empty()) { - return makeUnexpected(ErrorCode::ParseError, - std::string(Lexer.diagnostics().front().Message)); - } - return makeUnexpected(ErrorCode::ParseError, "invalid MIR metadata scalar"); - } - if (!at(MITokenKind::EndOfFile)) - return makeUnexpected(ErrorCode::ParseError, "unexpected trailing MIR metadata tokens"); - return {}; -} - -Expected<MachineMetadataReference> MetadataTextParser::parseReference() { - if (auto Exclaim = expect(MITokenKind::Exclaim, "'!' before a metadata slot"); !Exclaim) - return std::unexpected(Exclaim.error()); - if (!at(MITokenKind::IntegerLiteral)) - return makeUnexpected(ErrorCode::ParseError, "expected a 32-bit metadata slot"); - std::uint32_t Slot = 0; - if (!parseUnsigned32(token().integerValue(), Slot)) - return makeUnexpected(ErrorCode::ParseError, "metadata slot is outside the 32-bit range"); - advance(); - return Resolve(Slot); -} - -Expected<MachineMetadataReference> MetadataTextParser::parseReferenceOnly() { - auto Reference = parseReference(); - if (!Reference) - return std::unexpected(Reference.error()); - if (auto Finished = finish(); !Finished) - return std::unexpected(Finished.error()); - return Reference; -} - -Expected<MachineDIExpression> MetadataTextParser::parseExpression() { - if (!at(MITokenKind::MetadataKeyword) || !token().isMetadataKeyword("!DIExpression")) { - return makeUnexpected(ErrorCode::ParseError, "expected an inline !DIExpression"); - } - advance(); - if (auto Left = expect(MITokenKind::LeftParen, "'(' after !DIExpression"); !Left) - return std::unexpected(Left.error()); - std::vector<std::uint64_t> Elements; - while (!at(MITokenKind::RightParen)) { - if (at(MITokenKind::Identifier)) { - const auto Opcode = parseDIExpressionOpcodeName(token().stringValue()); - const auto Encoding = parseDIExpressionAttributeEncodingName(token().stringValue()); - if (!Opcode && !Encoding) - return makeUnexpected(ErrorCode::ParseError, - invalidDIExpressionElementMessage(token().stringValue())); - Elements.push_back(Opcode ? *Opcode : *Encoding); - advance(); - } else if (at(MITokenKind::IntegerLiteral)) { - std::int64_t Signed = 0; - std::uint64_t Unsigned = 0; - if (parseSigned(token().integerValue(), Signed)) - Elements.push_back(static_cast<std::uint64_t>(Signed)); - else if (parseUnsigned(token().integerValue(), Unsigned)) - Elements.push_back(Unsigned); - else - return makeUnexpected(ErrorCode::ParseError, "DIExpression integer exceeds 64 bits"); - advance(); - } else { - return makeUnexpected(ErrorCode::ParseError, "expected a DIExpression opcode or integer"); - } - if (!consume(MITokenKind::Comma)) - break; - } - if (auto Right = expect(MITokenKind::RightParen, "')' after !DIExpression"); !Right) - return std::unexpected(Right.error()); - return MachineDIExpression(std::move(Elements)); -} - -Expected<MachineMetadataValue> MetadataTextParser::parseValue() { - MachineMetadataValue Value; - if (at(MITokenKind::Exclaim)) { - auto Reference = parseReference(); - if (!Reference) - return std::unexpected(Reference.error()); - auto Created = MachineMetadataValue::reference(*Reference); - if (!Created) - return std::unexpected(Created.error()); - Value = std::move(*Created); - } else { - auto Expression = parseExpression(); - if (!Expression) - return std::unexpected(Expression.error()); - Value = MachineMetadataValue::expression(std::move(*Expression)); - } - if (auto Finished = finish(); !Finished) - return std::unexpected(Finished.error()); - return Value; -} - -Expected<void> -MetadataTextParser::parseDILocation(std::vector<MachineDILocationFrame> &Frames, - std::optional<MachineMetadataReference> &Terminal) { - if (!at(MITokenKind::MetadataKeyword) || !token().isMetadataKeyword("!DILocation")) { - return makeUnexpected(ErrorCode::ParseError, "expected an inline !DILocation"); - } - advance(); - if (auto Left = expect(MITokenKind::LeftParen, "'(' after !DILocation"); !Left) - return Left; - - MachineDILocationFrame Frame; - bool HasLine = false; - bool HasScope = false; - std::vector<MachineDILocationFrame> NestedFrames; - std::optional<MachineMetadataReference> NestedTerminal; - std::unordered_set<std::string> Seen; - while (!at(MITokenKind::RightParen)) { - if (!at(MITokenKind::Identifier)) - return makeUnexpected(ErrorCode::ParseError, "expected a DILocation field name"); - const std::string Field(token().stringValue()); - if (!Seen.insert(Field).second) { - return makeUnexpected(ErrorCode::ParseError, "duplicate DILocation field '" + Field + "'"); - } - advance(); - if (auto Colon = expect(MITokenKind::Colon, "':' after a DILocation field"); !Colon) - return Colon; - - if (Field == "line") { - std::uint32_t Value = 0; - if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned32(token().integerValue(), Value)) { - return makeUnexpected(ErrorCode::ParseError, - "DILocation line must be an unsigned 32-bit integer"); - } - Frame.Line = Value; - HasLine = true; - advance(); - } else if (Field == "column") { - std::uint64_t Value = 0; - if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned(token().integerValue(), Value) || - Value > std::numeric_limits<std::uint16_t>::max()) { - return makeUnexpected(ErrorCode::ParseError, - "DILocation column must be an unsigned 16-bit integer"); - } - Frame.Column = static_cast<std::uint16_t>(Value); - advance(); - } else if (Field == "scope") { - auto Scope = parseReference(); - if (!Scope) - return std::unexpected(Scope.error()); - if (Scope->scope() != MachineMetadataScope::EmbeddedIR) { - return makeUnexpected(ErrorCode::ParseError, - "DILocation scope must reference embedded LLVM IR metadata"); - } - Frame.Scope = *Scope; - HasScope = true; - } else if (Field == "inlinedAt") { - if (at(MITokenKind::MetadataKeyword) && token().isMetadataKeyword("!DILocation")) { - if (auto Nested = parseDILocation(NestedFrames, NestedTerminal); !Nested) - return Nested; - } else { - auto Reference = parseReference(); - if (!Reference) - return std::unexpected(Reference.error()); - if (Reference->scope() != MachineMetadataScope::EmbeddedIR) { - return makeUnexpected(ErrorCode::ParseError, - "DILocation inlinedAt must reference embedded LLVM IR metadata"); - } - NestedTerminal = *Reference; - } - } else if (Field == "isImplicitCode") { - if (!at(MITokenKind::Identifier) || - (token().stringValue() != "true" && token().stringValue() != "false")) { - return makeUnexpected(ErrorCode::ParseError, - "DILocation isImplicitCode must be true or false"); - } - Frame.IsImplicitCode = token().stringValue() == "true"; - advance(); - } else if (Field == "atomGroup") { - std::uint64_t Value = 0; - if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned(token().integerValue(), Value) || - Value > MachineDILocation::MaxAtomGroup) { - return makeUnexpected(ErrorCode::ParseError, "DILocation atomGroup exceeds 61 bits"); - } - Frame.AtomGroup = Value; - advance(); - } else if (Field == "atomRank") { - std::uint64_t Value = 0; - if (!at(MITokenKind::IntegerLiteral) || !parseUnsigned(token().integerValue(), Value) || - Value > 7) { - return makeUnexpected(ErrorCode::ParseError, "DILocation atomRank exceeds three bits"); - } - Frame.AtomRank = static_cast<std::uint8_t>(Value); - advance(); - } else { - return makeUnexpected(ErrorCode::ParseError, "unknown DILocation field '" + Field + "'"); - } - if (!consume(MITokenKind::Comma)) - break; - } - if (auto Right = expect(MITokenKind::RightParen, "')' after !DILocation"); !Right) - return Right; - if (!HasLine || !HasScope) - return makeUnexpected(ErrorCode::ParseError, "DILocation requires line and scope fields"); - Frames.push_back(Frame); - Frames.insert(Frames.end(), NestedFrames.begin(), NestedFrames.end()); - Terminal = NestedTerminal; - return {}; -} - -Expected<MachineDebugLocation> MetadataTextParser::parseLocation() { - MachineDebugLocation Location; - if (at(MITokenKind::Exclaim)) { - auto Reference = parseReference(); - if (!Reference) - return std::unexpected(Reference.error()); - auto Created = MachineDebugLocation::reference(*Reference); - if (!Created) - return std::unexpected(Created.error()); - Location = std::move(*Created); - } else { - std::vector<MachineDILocationFrame> Frames; - std::optional<MachineMetadataReference> Terminal; - if (auto Parsed = parseDILocation(Frames, Terminal); !Parsed) - return std::unexpected(Parsed.error()); - auto Created = MachineDILocation::create(std::move(Frames), Terminal); - if (!Created) - return std::unexpected(Created.error()); - Location = MachineDebugLocation::inlineLocation(std::move(*Created)); - } - if (auto Finished = finish(); !Finished) - return std::unexpected(Finished.error()); - return Location; -} - -} // namespace rcc::mir diff --git a/backend/lib/MIR/MIRMetadataTextParser.h b/backend/lib/MIR/MIRMetadataTextParser.h deleted file mode 100644 index 032523d..0000000 --- a/backend/lib/MIR/MIRMetadataTextParser.h +++ /dev/null @@ -1,74 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -// NOLINTNEXTLINE(llvm-header-guard) -#ifndef RCC_LIB_MIR_MIRMETADATATEXTPARSER_H -#define RCC_LIB_MIR_MIRMETADATATEXTPARSER_H - -#include "rcc/CodeGen/MachineDebugInfo.h" -#include "rcc/CodeGen/MachineMetadata.h" -#include "rcc/MIR/MILexer.h" -#include "rcc/Support/Error.h" - -#include <cstddef> -#include <cstdint> -#include <functional> -#include <optional> -#include <string> -#include <string_view> -#include <utility> -#include <vector> - -namespace rcc::mir { - -class MetadataTextParser final { -public: - using Resolver = std::function<Expected<MachineMetadataReference>(std::uint32_t Slot)>; - - MetadataTextParser(std::string Text, Resolver Resolve) - : Lexer("MIR metadata scalar", std::move(Text)), Resolve(std::move(Resolve)) { - do { - Tokens.push_back(Lexer.next()); - } while (!Tokens.back().isErrorOrEnd()); - } - - [[nodiscard]] Expected<MachineMetadataReference> parseReferenceOnly(); - [[nodiscard]] Expected<MachineMetadataValue> parseValue(); - [[nodiscard]] Expected<MachineDebugLocation> parseLocation(); - - [[nodiscard]] const TextRange &diagnosticRange() const noexcept { - return token().range(); - } - -private: - [[nodiscard]] const MIToken &token() const noexcept { - return Tokens[Position]; - } - [[nodiscard]] bool at(MITokenKind Kind) const noexcept { - return token().is(Kind); - } - void advance() noexcept { - if (Position + 1U < Tokens.size()) - ++Position; - } - [[nodiscard]] bool consume(MITokenKind Kind) noexcept { - if (!at(Kind)) - return false; - advance(); - return true; - } - [[nodiscard]] Expected<void> expect(MITokenKind Kind, std::string_view Description); - [[nodiscard]] Expected<void> finish(); - [[nodiscard]] Expected<MachineMetadataReference> parseReference(); - [[nodiscard]] Expected<MachineDIExpression> parseExpression(); - [[nodiscard]] Expected<void> parseDILocation(std::vector<MachineDILocationFrame> &Frames, - std::optional<MachineMetadataReference> &Terminal); - - MILexer Lexer; - Resolver Resolve; - std::vector<MIToken> Tokens; - std::size_t Position = 0; -}; - -} // namespace rcc::mir - -#endif // RCC_LIB_MIR_MIRMETADATATEXTPARSER_H diff --git a/backend/lib/MIR/MIRParserUtilities.h b/backend/lib/MIR/MIRParserUtilities.h deleted file mode 100644 index 2f69548..0000000 --- a/backend/lib/MIR/MIRParserUtilities.h +++ /dev/null @@ -1,85 +0,0 @@ -// SPDX-License-Identifier: GPL-3.0-or-later - -// NOLINTNEXTLINE(llvm-header-guard) -#ifndef RCC_LIB_MIR_MIRPARSERUTILITIES_H -#define RCC_LIB_MIR_MIRPARSERUTILITIES_H - -#include <charconv> -#include <cstdint> -#include <limits> -#include <optional> -#include <string_view> - -namespace rcc::mir::parser_detail { - -[[nodiscard]] inline std::string_view trim(std::string_view Text) noexcept { - while (!Text.empty() && (Text.front() == ' ' || Text.front() == '\t' || Text.front() == '\r' || - Text.front() == '\n')) { - Text.remove_prefix(1); - } - while (!Text.empty() && (Text.back() == ' ' || Text.back() == '\t' || Text.back() == '\r' || - Text.back() == '\n')) { - Text.remove_suffix(1); - } - return Text; -} - -[[nodiscard]] inline bool parseUnsigned(std::string_view Text, std::uint64_t &Value) noexcept { - Text = trim(Text); - int Base = 10; - if (Text.starts_with("0x") || Text.starts_with("0X")) { - Text.remove_prefix(2); - Base = 16; - } - if (Text.empty()) - return false; - const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, Base); - return Error == std::errc{} && End == Text.data() + Text.size(); -} - -[[nodiscard]] inline bool parseSigned(std::string_view Text, std::int64_t &Value) noexcept { - Text = trim(Text); - if (Text.empty()) - return false; - const auto [End, Error] = std::from_chars(Text.data(), Text.data() + Text.size(), Value, 10); - return Error == std::errc{} && End == Text.data() + Text.size(); -} - -[[nodiscard]] inline bool parseUnsigned32(std::string_view Text, std::uint32_t &Value) noexcept { - std::uint64_t Wide = 0; - if (!parseUnsigned(Text, Wide) || Wide > std::numeric_limits<std::uint32_t>::max()) - return false; - Value = static_cast<std::uint32_t>(Wide); - return true; -} - -[[nodiscard]] inline bool parseBool(std::string_view Text, bool &Value) noexcept { - if (Text == "true") { - Value = true; - return true; - } - if (Text == "false") { - Value = false; - return true; - } - return false; -} - -[[nodiscard]] inline std::optional<std::uint32_t> -parsePrefixedID(std::string_view Text, std::string_view Prefix) noexcept { - Text = trim(Text); - if (!Text.starts_with(Prefix)) - return std::nullopt; - Text.remove_prefix(Prefix.size()); - const std::size_t Suffix = Text.find_first_of(". \t"); - if (Suffix != std::string_view::npos) - Text = Text.substr(0, Suffix); - std::uint32_t ID = 0; - if (!parseUnsigned32(Text, ID)) - return std::nullopt; - return ID; -} - -} // namespace rcc::mir::parser_detail - -#endif // RCC_LIB_MIR_MIRPARSERUTILITIES_H diff --git a/backend/lib/Tooling/ArtifactStore.cpp b/backend/lib/Tooling/ArtifactStore.cpp index db330b1..115f58f 100644 --- a/backend/lib/Tooling/ArtifactStore.cpp +++ b/backend/lib/Tooling/ArtifactStore.cpp @@ -3,10 +3,10 @@ #include "rcc/Tooling/ArtifactStore.h" #include "rcc/CodeGen/CompilationUnit.h" +#include "rcc/CodeGen/MIRPrinter.h" #include "rcc/CodeGen/MachineFunction.h" #include "rcc/CodeGen/MachineVerifier.h" #include "rcc/CodeGen/TargetDescription.h" -#include "rcc/MIR/MIRDocument.h" #include "rcc/Support/SHA256.h" #include "rcc/Support/TextDiagnostic.h" #include "rcc/Tooling/BackendInvocation.h" diff --git a/backend/lib/Tooling/BackendDriver.cpp b/backend/lib/Tooling/BackendDriver.cpp index c726fd3..0402f5a 100644 --- a/backend/lib/Tooling/BackendDriver.cpp +++ b/backend/lib/Tooling/BackendDriver.cpp @@ -2,8 +2,8 @@ #include "rcc/Tooling/BackendDriver.h" +#include "rcc/CodeGen/MIRPrinter.h" #include "rcc/CodeGen/MachineFunction.h" -#include "rcc/MIR/MIRDocument.h" #include "rcc/Target/RISCV/RISCVAssemblyPrinter.h" #include "rcc/Target/RISCV/RISCVObjectFile.h" diff --git a/backend/lib/Tooling/MIRFuzzer.cpp b/backend/lib/Tooling/MIRFuzzer.cpp index e75e9f9..83762c4 100644 --- a/backend/lib/Tooling/MIRFuzzer.cpp +++ b/backend/lib/Tooling/MIRFuzzer.cpp @@ -5,6 +5,7 @@ #include "rcc/CodeGen/GenericOpcodes.h" #include "rcc/CodeGen/InstructionDescriptor.h" #include "rcc/CodeGen/LowLevelType.h" +#include "rcc/CodeGen/MIRPrinter.h" #include "rcc/CodeGen/MachineBasicBlock.h" #include "rcc/CodeGen/MachineFunction.h" #include "rcc/CodeGen/MachineIRBuilder.h" @@ -13,7 +14,6 @@ #include "rcc/CodeGen/MachineRegisterInfo.h" #include "rcc/CodeGen/MachineVerifier.h" #include "rcc/CodeGen/TargetDescription.h" -#include "rcc/MIR/MIRDocument.h" #include "rcc/Target/RISCV/RISCVPipeline.h" #include "rcc/Tooling/ArtifactStore.h" #include "rcc/Tooling/BackendDriver.h" diff --git a/backend/lib/Tooling/PipelineDriver.cpp b/backend/lib/Tooling/PipelineDriver.cpp index d7a7d16..68cbb57 100644 --- a/backend/lib/Tooling/PipelineDriver.cpp +++ b/backend/lib/Tooling/PipelineDriver.cpp @@ -3,10 +3,10 @@ #include "rcc/Tooling/PipelineDriver.h" #include "rcc/CodeGen/CompilationUnit.h" +#include "rcc/CodeGen/MIRPrinter.h" #include "rcc/CodeGen/MachineFunction.h" #include "rcc/CodeGen/MachineOptimizationPipeline.h" #include "rcc/CodeGen/MachineVerifier.h" -#include "rcc/MIR/MIRDocument.h" #include "rcc/Target/RISCV/RISCVPipeline.h" #include <algorithm> |