From 379ff50dcc7ccf977379aee68fa9d10f69065d23 Mon Sep 17 00:00:00 2001 From: Richard-Qin-X Date: Sun, 9 Aug 2026 18:31:49 +0800 Subject: [MC] Implement RV64GC assembly emission --- backend/lib/MC/MCStreamer.cpp | 488 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 488 insertions(+) create mode 100644 backend/lib/MC/MCStreamer.cpp (limited to 'backend/lib/MC/MCStreamer.cpp') diff --git a/backend/lib/MC/MCStreamer.cpp b/backend/lib/MC/MCStreamer.cpp new file mode 100644 index 0000000..001fa28 --- /dev/null +++ b/backend/lib/MC/MCStreamer.cpp @@ -0,0 +1,488 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "rcc/MC/MCStreamer.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +namespace rcc::mc { +namespace { + +[[nodiscard]] bool containsInvalidLineCharacter(std::string_view Text) noexcept { + return Text.find('\0') != std::string_view::npos || Text.find('\n') != std::string_view::npos || + Text.find('\r') != std::string_view::npos; +} + +[[nodiscard]] bool isPowerOfTwo(std::uint64_t Value) noexcept { + return Value != 0 && (Value & (Value - 1U)) == 0; +} + +[[nodiscard]] unsigned log2Exact(std::uint64_t Value) noexcept { + unsigned Result = 0; + while (Value > 1) { + Value >>= 1U; + ++Result; + } + return Result; +} + +[[nodiscard]] std::string byteSpelling(std::uint8_t Byte) { + constexpr std::string_view Hex = "0123456789abcdef"; + std::string Result = "0x00"; + Result[2] = Hex[(Byte >> 4U) & 0xfU]; + Result[3] = Hex[Byte & 0xfU]; + return Result; +} + +[[nodiscard]] std::string sectionFlagSpelling(MCSectionFlags Flags) { + std::string Result; + const auto Add = [&](MCSectionFlag Flag, char Spelling) { + if (Flags.has(Flag)) + Result.push_back(Spelling); + }; + Add(MCSectionFlag::Alloc, 'a'); + Add(MCSectionFlag::Writable, 'w'); + Add(MCSectionFlag::Executable, 'x'); + Add(MCSectionFlag::Merge, 'M'); + Add(MCSectionFlag::Strings, 'S'); + Add(MCSectionFlag::TLS, 'T'); + Add(MCSectionFlag::Group, 'G'); + Add(MCSectionFlag::Retain, 'R'); + return Result; +} + +[[nodiscard]] std::string_view sectionTypeSpelling(MCSectionKind Kind) noexcept { + switch (Kind) { + case MCSectionKind::BSS: + case MCSectionKind::TLSBSS: + return "nobits"; + case MCSectionKind::Text: + case MCSectionKind::ReadOnly: + case MCSectionKind::Data: + case MCSectionKind::TLSData: + case MCSectionKind::EHFrame: + case MCSectionKind::Custom: + return "progbits"; + } + return {}; +} + +[[nodiscard]] bool isDirectiveName(std::string_view Name) noexcept { + if (Name.empty() || Name.front() == '.') + return false; + const auto Initial = [](unsigned char Character) { + return std::isalpha(Character) != 0 || Character == '_'; + }; + const auto Following = [](unsigned char Character) { + return std::isalnum(Character) != 0 || Character == '_' || Character == '.'; + }; + if (!Initial(static_cast(Name.front()))) + return false; + return std::ranges::all_of(Name.substr(1), [Following](char Character) { + return Following(static_cast(Character)); + }); +} + +[[nodiscard]] std::string joinIntegerOperands(std::span Operands) { + std::string Result; + for (const std::int64_t Operand : Operands) { + if (!Result.empty()) + Result += ", "; + Result += std::to_string(Operand); + } + return Result; +} + +[[nodiscard]] std::vector encodeULEB128(std::uint64_t Value) { + std::vector Bytes; + do { + std::uint8_t Byte = static_cast(Value & 0x7fU); + Value >>= 7U; + if (Value != 0) + Byte |= 0x80U; + Bytes.push_back(Byte); + } while (Value != 0); + return Bytes; +} + +} // namespace + +Expected AsmTextStreamer::appendLine(std::string_view Prefix, std::string_view Text) { + if (IsFinished) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "cannot emit after the assembly stream is finalized"); + if (Text.empty() || containsInvalidLineCharacter(Text)) { + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "assembly stream line must be nonempty and contain no line break"); + } + Output.append(Prefix); + Output.append(Text); + Output.push_back('\n'); + return {}; +} + +Expected AsmTextStreamer::requireActiveSection() const { + if (!CurrentSection) { + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "assembly emission requires an active section"); + } + if (IsFinished) { + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "cannot emit after the assembly stream is finalized"); + } + return {}; +} + +Expected AsmTextStreamer::switchSection(MCSectionRef Section) { + if (IsFinished) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "cannot switch section after assembly finalization"); + if (!Section) + return makeUnexpected(ErrorCode::InvalidArgument, "cannot switch to a null MC section"); + auto Verification = Section->description().verify(); + if (!Verification) + return std::unexpected(Verification.error()); + if (CurrentSection == Section) + return {}; + + auto Name = printMCSymbolName(Section->name()); + if (!Name) + return std::unexpected(Name.error()); + const MCSectionDesc &Description = Section->description(); + std::string Directive = ".section " + *Name + ",\"" + sectionFlagSpelling(Description.Flags) + + "\",@" + std::string(sectionTypeSpelling(Description.Kind)); + if (Description.EntrySize != 0) + Directive += "," + std::to_string(Description.EntrySize); + if (Description.Group) { + auto GroupName = printMCSymbolName(Description.Group->name()); + if (!GroupName) + return std::unexpected(GroupName.error()); + Directive += "," + *GroupName + ",comdat"; + } + if (Description.UniqueID) + Directive += ",unique," + std::to_string(*Description.UniqueID); + + auto Emitted = appendLine({}, Directive); + if (!Emitted) + return std::unexpected(Emitted.error()); + CurrentSection = std::move(Section); + return {}; +} + +Expected AsmTextStreamer::emitLabel(MCSymbolRef Symbol) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (!Symbol) + return makeUnexpected(ErrorCode::InvalidArgument, "cannot emit a null MC symbol label"); + const std::string Name(Symbol->name()); + if (!DefinedSymbols.insert(Name).second) { + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "MC symbol label is defined more than once"); + } + auto Printed = printMCSymbolName(Name); + if (!Printed) + return std::unexpected(Printed.error()); + return appendLine({}, *Printed + ":"); +} + +Expected AsmTextStreamer::emitInstruction(const MCInst &Instruction) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (!Instruction.opcode().isValid()) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "MC instruction has an invalid opcode"); + for (const MCOperand &Operand : Instruction.operands()) { + if (const MCExprRef *Expression = Operand.expressionValue(); Expression != nullptr) { + if (!*Expression) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "MC instruction contains a null expression operand"); + auto Verification = (*Expression)->verify(); + if (!Verification) + return std::unexpected(Verification.error()); + } + } + auto Text = Printer->printInstruction(Instruction); + if (!Text) + return std::unexpected(Text.error()); + return appendLine(" ", *Text); +} + +Expected AsmTextStreamer::emitValue(MCExprRef Value, std::uint8_t Size) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (!Value) + return makeUnexpected(ErrorCode::InvalidArgument, "cannot emit a null MC value expression"); + std::string_view Directive; + switch (Size) { + case 1: + Directive = ".byte "; + break; + case 2: + Directive = ".2byte "; + break; + case 4: + Directive = ".4byte "; + break; + case 8: + Directive = ".8byte "; + break; + default: + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "assembly value size must be 1, 2, 4, or 8 bytes"); + } + auto Printed = printMCExpression(*Value, *Printer); + if (!Printed) + return std::unexpected(Printed.error()); + return appendLine(" ", std::string(Directive) + *Printed); +} + +Expected AsmTextStreamer::emitBytes(std::span Bytes) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + for (std::size_t Begin = 0; Begin < Bytes.size(); Begin += 16) { + const std::size_t End = std::min(Bytes.size(), Begin + 16); + std::string Directive = ".byte "; + for (std::size_t Index = Begin; Index < End; ++Index) { + if (Index != Begin) + Directive += ", "; + Directive += byteSpelling(Bytes[Index]); + } + auto Emitted = appendLine(" ", Directive); + if (!Emitted) + return std::unexpected(Emitted.error()); + } + return {}; +} + +Expected AsmTextStreamer::emitZeroFill(std::uint64_t Size) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (Size == 0) + return {}; + return appendLine(" ", ".zero " + std::to_string(Size)); +} + +Expected AsmTextStreamer::emitAlignment(std::uint64_t Alignment, std::uint8_t Fill, + std::uint64_t MaximumBytes) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (!isPowerOfTwo(Alignment)) { + return makeUnexpected(ErrorCode::InvalidArgument, + "assembly alignment must be a nonzero power of two"); + } + std::string Directive = ".p2align " + std::to_string(log2Exact(Alignment)); + if (Fill != 0 || MaximumBytes != 0) + Directive += ", " + std::to_string(Fill); + if (MaximumBytes != 0) + Directive += ", " + std::to_string(MaximumBytes); + return appendLine(" ", Directive); +} + +Expected AsmTextStreamer::emitSymbolAttribute(MCSymbolRef Symbol, + MCSymbolAttribute Attribute) { + if (IsFinished) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "cannot emit a symbol attribute after assembly finalization"); + if (!Symbol) + return makeUnexpected(ErrorCode::InvalidArgument, "symbol attribute requires an MC symbol"); + auto Name = printMCSymbolName(Symbol->name()); + if (!Name) + return std::unexpected(Name.error()); + + switch (Attribute) { + case MCSymbolAttribute::Global: + return appendLine({}, ".globl " + *Name); + case MCSymbolAttribute::Weak: + return appendLine({}, ".weak " + *Name); + case MCSymbolAttribute::Local: + return appendLine({}, ".local " + *Name); + case MCSymbolAttribute::Hidden: + return appendLine({}, ".hidden " + *Name); + case MCSymbolAttribute::Protected: + return appendLine({}, ".protected " + *Name); + case MCSymbolAttribute::Function: + return appendLine({}, ".type " + *Name + ", @function"); + case MCSymbolAttribute::Object: + return appendLine({}, ".type " + *Name + ", @object"); + case MCSymbolAttribute::TLSObject: + return appendLine({}, ".type " + *Name + ", @tls_object"); + } + return makeUnexpected(ErrorCode::InternalError, "MC symbol attribute is invalid"); +} + +Expected AsmTextStreamer::emitSymbolSize(MCSymbolRef Symbol, MCExprRef Size) { + if (IsFinished) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "cannot emit a symbol size after assembly finalization"); + if (!Symbol || !Size) + return makeUnexpected(ErrorCode::InvalidArgument, + "symbol size event requires a symbol and an expression"); + auto Name = printMCSymbolName(Symbol->name()); + if (!Name) + return std::unexpected(Name.error()); + auto PrintedSize = printMCExpression(*Size, *Printer); + if (!PrintedSize) + return std::unexpected(PrintedSize.error()); + return appendLine({}, ".size " + *Name + ", " + *PrintedSize); +} + +Expected AsmTextStreamer::emitDirective(std::string_view Name, std::string_view Arguments) { + if (!isDirectiveName(Name) || containsInvalidLineCharacter(Arguments)) { + return makeUnexpected(ErrorCode::InvalidArgument, + "assembly directive name or arguments are invalid"); + } + std::string Text = "." + std::string(Name); + if (!Arguments.empty()) + Text += " " + std::string(Arguments); + return appendLine({}, Text); +} + +Expected AsmTextStreamer::emitCFI(const MCCFIInstruction &Instruction) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + auto Verification = Instruction.verify(); + if (!Verification) + return std::unexpected(Verification.error()); + + const std::string Operands = joinIntegerOperands(Instruction.Operands); + switch (Instruction.Operation) { + case MCCFIOperation::StartProc: + return emitDirective("cfi_startproc"); + case MCCFIOperation::EndProc: + return emitDirective("cfi_endproc"); + case MCCFIOperation::Sections: + return emitDirective("cfi_sections", ".eh_frame"); + case MCCFIOperation::Personality: + case MCCFIOperation::LSDA: { + auto Symbol = printMCSymbolName(Instruction.Symbol->name()); + if (!Symbol) + return std::unexpected(Symbol.error()); + return emitDirective(Instruction.Operation == MCCFIOperation::Personality ? "cfi_personality" + : "cfi_lsda", + Operands + ", " + *Symbol); + } + case MCCFIOperation::SignalFrame: + return emitDirective("cfi_signal_frame"); + case MCCFIOperation::ReturnColumn: + return emitDirective("cfi_return_column", Operands); + case MCCFIOperation::SameValue: + return emitDirective("cfi_same_value", Operands); + case MCCFIOperation::RememberState: + return emitDirective("cfi_remember_state"); + case MCCFIOperation::RestoreState: + return emitDirective("cfi_restore_state"); + case MCCFIOperation::Offset: + return emitDirective("cfi_offset", Operands); + case MCCFIOperation::DefCFARegister: + return emitDirective("cfi_def_cfa_register", Operands); + case MCCFIOperation::DefCFAOffset: + return emitDirective("cfi_def_cfa_offset", Operands); + case MCCFIOperation::DefCFA: + return emitDirective("cfi_def_cfa", Operands); + case MCCFIOperation::RelativeOffset: + return emitDirective("cfi_rel_offset", Operands); + case MCCFIOperation::AdjustCFAOffset: + return emitDirective("cfi_adjust_cfa_offset", Operands); + case MCCFIOperation::Escape: { + std::string Bytes; + for (const std::uint8_t Byte : Instruction.Bytes) { + if (!Bytes.empty()) + Bytes += ", "; + Bytes += byteSpelling(Byte); + } + return emitDirective("cfi_escape", Bytes); + } + case MCCFIOperation::Restore: + return emitDirective("cfi_restore", Operands); + case MCCFIOperation::Undefined: + return emitDirective("cfi_undefined", Operands); + case MCCFIOperation::Register: + return emitDirective("cfi_register", Operands); + case MCCFIOperation::WindowSave: + return emitDirective("cfi_window_save"); + case MCCFIOperation::NegateReturnAddressState: + return emitDirective("cfi_negate_ra_state"); + case MCCFIOperation::NegateReturnAddressStateWithPC: + return emitDirective("cfi_negate_ra_state_with_pc"); + case MCCFIOperation::GNUArgsSize: { + if (Instruction.Operands.front() < 0) { + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "GNU CFI argument size cannot be negative"); + } + std::vector Bytes{0x2eU}; + const auto Encoded = encodeULEB128(static_cast(Instruction.Operands.front())); + Bytes.insert(Bytes.end(), Encoded.begin(), Encoded.end()); + return emitCFI(MCCFIInstruction{.Operation = MCCFIOperation::Escape, + .Operands = {}, + .Bytes = std::move(Bytes), + .Symbol = {}}); + } + case MCCFIOperation::Label: + return emitLabel(Instruction.Symbol); + case MCCFIOperation::ValueOffset: + return emitDirective("cfi_val_offset", Operands); + } + return makeUnexpected(ErrorCode::InternalError, "MC CFI operation is invalid"); +} + +Expected AsmTextStreamer::emitInlineAssembly(std::string_view Text) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (Text.find('\0') != std::string_view::npos || Text.find('\r') != std::string_view::npos) { + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "inline assembly contains an invalid character"); + } + if (Text.empty()) + return {}; + Output.append(Text); + if (Text.back() != '\n') + Output.push_back('\n'); + return {}; +} + +Expected AsmTextStreamer::emitComment(std::string_view Text) { + auto Active = requireActiveSection(); + if (!Active) + return std::unexpected(Active.error()); + if (Text.empty() || containsInvalidLineCharacter(Text)) + return makeUnexpected(ErrorCode::InvalidArgument, "assembly comment is invalid"); + return appendLine(" # ", Text); +} + +Expected AsmTextStreamer::finish() { + if (IsFinished) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "assembly stream was already finalized"); + IsFinished = true; + return {}; +} + +Expected AsmTextStreamer::takeOutput() { + if (!IsFinished) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, + "assembly output cannot be taken before finalization"); + if (OutputWasTaken) + return makeUnexpected(ErrorCode::AssemblyLoweringFailed, "assembly output was already taken"); + OutputWasTaken = true; + return std::move(Output); +} + +} // namespace rcc::mc -- cgit v1.2.3