From d62f022c92fcc3ebe2269273628943a0089683fd Mon Sep 17 00:00:00 2001 From: Richard-Qin-X Date: Thu, 13 Aug 2026 06:19:30 +0800 Subject: [CodeGen] Align MIR parser and printer ownership --- backend/lib/CodeGen/MIRYamlMapping.cpp | 2926 ++++++++++++++++++++++++++++++++ 1 file changed, 2926 insertions(+) create mode 100644 backend/lib/CodeGen/MIRYamlMapping.cpp (limited to 'backend/lib/CodeGen/MIRYamlMapping.cpp') 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 +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +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::max()) + return false; + Value = static_cast(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 +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(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 parseReferenceOnly(); + [[nodiscard]] Expected parseValue(); + [[nodiscard]] Expected 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 expect(MITokenKind Kind, std::string_view Description); + [[nodiscard]] Expected finish(); + [[nodiscard]] Expected parseReference(); + [[nodiscard]] Expected parseExpression(); + [[nodiscard]] Expected parseDILocation(std::vector &Frames, + std::optional &Terminal); + + MILexer Lexer; + Resolver Resolve; + std::vector 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 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 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::max()) { + return true; + } + const auto TextID = static_cast(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(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 &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 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::max()) { + return true; + } + const auto TextID = static_cast(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 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>({ + {"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>({ + {"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 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 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 parseDIExpressionOpcodeName(std::string_view Name) noexcept { + const auto NumberedOpcode = [Name](std::string_view Prefix, + std::uint64_t Base) -> std::optional { + 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 +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 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 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 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 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 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 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 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(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 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 +MetadataTextParser::parseDILocation(std::vector &Frames, + std::optional &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 NestedFrames; + std::optional NestedTerminal; + std::unordered_set 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::max()) { + return makeUnexpected(ErrorCode::ParseError, + "DILocation column must be an unsigned 16-bit integer"); + } + Frame.Column = static_cast(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(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 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 Frames; + std::optional 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 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 { + 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 { + 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 { + 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 ExpectedTypes, + std::string_view Description) -> Expected { + 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 { + const MachineMetadataReference *Reference = Value.reference(); + return Reference == nullptr ? Expected{} : ValidateMetadata(*Reference, {}, Description); + }; + const auto ValidateLocation = + [&ValidateMetadata](const MachineDebugLocation &Location, + std::string_view Description) -> Expected { + 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 { + 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 { + 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 { + 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(&Description->Value)) { + auto Verification = ValidateGenericConstant(ValidateGenericConstant, *Generic); + if (!Verification) + return Verification; + } else if (const auto *Address = std::get_if(&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(); Global != nullptr) { + auto Verification = ValidateGlobal(Global->Name, "global-symbol operand"); + if (!Verification) + return Verification; + } + if (const auto *Address = Operand.getIf(); Address != nullptr) { + auto Verification = ValidateBlockAddress(Address->FunctionName, Address->BlockName, + "block-address operand"); + if (!Verification) + return Verification; + } + if (const auto *Metadata = Operand.getIf(); 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 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 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::max()) { + return true; + } + const auto TextID = static_cast(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 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 &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::min() || + Wide > std::numeric_limits::max()) { + return fail(Value->range(), "offset adjustment is outside the 32-bit signed range"); + } + State.OffsetAdjustment = static_cast(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::max()) { + return fail(Value->range(), "maximum call-frame size is outside the 32-bit unsigned range"); + } + if (Size == std::numeric_limits::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::max()) { + return fail(Value->range(), "frame value is outside the 32-bit unsigned range"); + } + *Destination = static_cast(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 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() != 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 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 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::max() || + scalar(*ClassNode, ClassName, "register class or bank")) { + return true; + } + const auto TextID = static_cast(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(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 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 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 { + 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::numeric_limits::max())) { + return makeUnexpected(ErrorCode::Unsupported, + "MIR document contains too many synchronization scopes"); + } + const auto ID = + SyncScopeID::fromValue(static_cast(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 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 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::max() || + Offset > std::numeric_limits::max()) { + return true; + } + Instruction = instructionAt(Function, static_cast(Block), + static_cast(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::max() || + physicalRegister(*RegNode, Reg)) { + return true; + } + Number = static_cast(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::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(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 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::max()) { + return true; + } + Values[Index] = static_cast(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 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 +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 Elements; + }; + std::vector Definitions; + std::unordered_set 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 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 -- cgit v1.2.3