From 03a72fc5834fca19e24235c6950e6e09acb6fb39 Mon Sep 17 00:00:00 2001 From: Eugen Wissner Date: Sun, 25 Feb 2024 15:16:19 +0100 Subject: [PATCH] Extend the tester to compile sources --- .gitignore | 1 + CMakeLists.txt | 15 +- README | 37 + Rakefile | 90 ++ TODO | 30 +- dub.json | 9 + include/elna/interactive.hpp | 19 +- include/elna/lexer.hpp | 11 +- include/elna/result.hpp | 14 +- include/elna/state.hpp | 32 +- shell/history.cpp | 9 +- shell/interactive.cpp | 84 +- shell/result.cpp | 18 +- shell/state.cpp | 68 +- source/elna/arguments.d | 154 ++++ source/elna/backend.d | 105 +++ source/elna/elf.d | 1060 ++++++++++++++++++++++++ source/elna/extended.d | 335 ++++++++ source/elna/ir.d | 268 ++++++ source/elna/lexer.d | 254 ++++++ source/elna/parser.d | 372 +++++++++ source/elna/result.d | 107 +++ source/elna/riscv.d | 363 ++++++++ {shell => source}/lexer.cpp | 11 +- source/main.d | 33 + tests/7_member_sum.eln | 1 + tests/const_list.eln | 3 + tests/expectations/7_member_sum.txt | 1 + tests/expectations/const_list.txt | 1 + tests/expectations/left_nested_sum.txt | 1 + tests/expectations/print_number.txt | 1 + tests/expectations/subtraction.txt | 1 + tests/expectations/sum.txt | 1 + tests/expectations/sums.txt | 1 + tests/left_nested_sum.eln | 2 + tests/print_number.eln | 2 + tests/runner.cpp | 57 +- tests/subtraction.eln | 2 + tests/sum.eln | 2 + tests/sums.eln | 2 + 40 files changed, 3473 insertions(+), 104 deletions(-) create mode 100644 README create mode 100644 Rakefile create mode 100644 dub.json create mode 100644 source/elna/arguments.d create mode 100644 source/elna/backend.d create mode 100644 source/elna/elf.d create mode 100644 source/elna/extended.d create mode 100644 source/elna/ir.d create mode 100644 source/elna/lexer.d create mode 100644 source/elna/parser.d create mode 100644 source/elna/result.d create mode 100644 source/elna/riscv.d rename {shell => source}/lexer.cpp (90%) create mode 100644 source/main.d create mode 100644 tests/7_member_sum.eln create mode 100644 tests/const_list.eln create mode 100644 tests/expectations/7_member_sum.txt create mode 100644 tests/expectations/const_list.txt create mode 100644 tests/expectations/left_nested_sum.txt create mode 100644 tests/expectations/print_number.txt create mode 100644 tests/expectations/subtraction.txt create mode 100644 tests/expectations/sum.txt create mode 100644 tests/expectations/sums.txt create mode 100644 tests/left_nested_sum.eln create mode 100644 tests/print_number.eln create mode 100644 tests/subtraction.eln create mode 100644 tests/sum.eln create mode 100644 tests/sums.eln diff --git a/.gitignore b/.gitignore index a65d6bb..021b743 100644 --- a/.gitignore +++ b/.gitignore @@ -2,3 +2,4 @@ .cache/ CMakeFiles/ CMakeCache.txt +dub.selections.json diff --git a/CMakeLists.txt b/CMakeLists.txt index d316f83..8a2223e 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -3,8 +3,7 @@ project(Elna) set(CMAKE_EXPORT_COMPILE_COMMANDS 1) set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin) - -find_package(Boost COMPONENTS filesystem REQUIRED) +set(CMAKE_CXX_STANDARD 17) add_executable(tester tests/runner.cpp include/elna/tester.hpp) target_include_directories(tester PRIVATE include) @@ -13,9 +12,11 @@ add_executable(elnsh shell/main.cpp shell/interactive.cpp include/elna/interactive.hpp shell/history.cpp include/elna/history.hpp shell/state.cpp include/elna/state.hpp - shell/lexer.cpp include/elna/lexer.hpp - shell/result.cpp include/elna/result.hpp) -target_include_directories(elnsh PRIVATE include) -target_link_libraries(elnsh PUBLIC - ${Boost_FILESYSTEM_LIBRARY} + shell/result.cpp include/elna/result.hpp ) +target_include_directories(elnsh PRIVATE include) + +add_library(elna + source/lexer.cpp include/elna/lexer.hpp +) +target_include_directories(elna PRIVATE include) diff --git a/README b/README new file mode 100644 index 0000000..b6e03eb --- /dev/null +++ b/README @@ -0,0 +1,37 @@ +# Elna programming language + +Elna compiles simple mathematical operations to machine code. +The compiled program returns the result of the operation. + +## File extension + +.elna + +## Grammar PL/0 + +program = block "." ; + +block = [ "const" ident "=" number {"," ident "=" number} ";"] + [ "var" ident {"," ident} ";"] + { "procedure" ident ";" block ";" } statement ; + +statement = [ ident ":=" expression | "call" ident + | "?" ident | "!" expression + | "begin" statement {";" statement } "end" + | "if" condition "then" statement + | "while" condition "do" statement ]; + +condition = "odd" expression | + expression ("="|"#"|"<"|"<="|">"|">=") expression ; + +expression = [ "+"|"-"] term { ("+"|"-") term}; + +term = factor {("*"|"/") factor}; + +factor = ident | number | "(" expression ")"; + +## Operations + +"!" - Write a line. +"?" - Read user input. +"odd" - The only function, returns whether a number is odd. diff --git a/Rakefile b/Rakefile new file mode 100644 index 0000000..c623b05 --- /dev/null +++ b/Rakefile @@ -0,0 +1,90 @@ +require 'pathname' +require 'rake/clean' +require 'open3' + +DFLAGS = ['--warn-no-deprecated', '-L/usr/lib64/gcc-12'] +BINARY = 'build/bin/elna' +TESTS = FileList['tests/*.eln'].flat_map do |test| + build = Pathname.new 'build' + test_basename = Pathname.new(test).basename('') + + [build + 'riscv' + test_basename].map { |path| path.sub_ext('').to_path } +end + +SOURCES = FileList['source/**/*.d'] + +directory 'build' + +CLEAN.include 'build' +CLEAN.include '.dub' + +rule(/build\/riscv\/[^\/\.]+$/ => ->(file) { test_for_out(file, '.o') }) do |t| + sh '/opt/riscv/bin/riscv32-unknown-elf-ld', + '-o', t.name, + '-L/opt/riscv/lib/gcc/riscv32-unknown-elf/13.2.0/', + '-L/opt/riscv/riscv32-unknown-elf/lib', + '/opt/riscv/riscv32-unknown-elf/lib/crt0.o', + '/opt/riscv/lib/gcc/riscv32-unknown-elf/13.2.0/crtbegin.o', + t.source, + '--start-group', '-lgcc', '-lc', '-lgloss', '--end-group', + '/opt/riscv/lib/gcc/riscv32-unknown-elf/13.2.0/crtend.o' +end + +rule(/build\/riscv\/.+\.o$/ => ->(file) { test_for_object(file, '.eln') }) do |t| + Pathname.new(t.name).dirname.mkpath + sh BINARY, '-o', t.name, t.source +end + +file BINARY => SOURCES do |t| + sh({ 'DFLAGS' => (DFLAGS * ' ') }, 'dub', 'build', '--compiler=gdc') +end + +task default: BINARY + +desc 'Run all tests and check the results' +task test: TESTS +task test: BINARY do + TESTS.each do |test| + expected = Pathname + .new(test) + .sub_ext('.txt') + .sub(/^build\/[[:alpha:]]+\//, 'tests/expectations/') + .to_path + + puts "Running #{test}" + if test.include? '/riscv/' + spike = [ + '/opt/riscv/bin/spike', + '--isa=RV32IMAC', + '/opt/riscv/riscv32-unknown-elf/bin/pk', + test + ] + diff = ['diff', '-Nur', '--color', expected, '-'] + tail = ['tail', '-n', '1'] + + last_stdout, wait_threads = Open3.pipeline_r spike, tail, diff + else + raise 'Unsupported test platform' + end + print last_stdout.read + last_stdout.close + + fail unless wait_threads.last.value.exitstatus.zero? + end +end + +def test_for_object(out_file, extension) + test_source = Pathname + .new(out_file) + .sub_ext(extension) + .sub(/^build\/[[:alpha:]]+\//, 'tests/') + .to_path + [test_source, BINARY] +end + +def test_for_out(out_file, extension) + Pathname + .new(out_file) + .sub_ext(extension) + .to_path +end diff --git a/TODO b/TODO index 5b2a1dc..6a08d34 100644 --- a/TODO +++ b/TODO @@ -1,15 +1,23 @@ -- After inserting autocompletion move the cursor to the end of the insertion. +# Completion + +- Configure fzf to show only the current directory files - Support multiple selections for insertion in fzf. -- Vi mode. - Don't hardcode fzf binary path. -- Persist the history. - Send the word under the cursor to fzf as initial input. -- Replace hard coded ANSI codes with constants or functions. -- Show the hostname in the prompt. -- Crash if the cd directory doesn't exist. - Home directory expansion. -- Paste pastes only the first character. -- When we read the key and an error occures how it should be handled? -- Kitty protocol works with UTF8 code points for escape sequences with modifiers. - They can be wider than an unsigned char. -- Add a bar with additional information under the prompt (edit_bar). +- Show files in the PATH if starting at the beginning of the prompt + +# Appearance + +- Add a bar with additional information under the prompt (edit_bar), like the hostname. +- Show current time in the prompt. + +# Input + +- DEL handling. +- Starting long running process and killing it with Ctrl-C kills the shell. + +# Other + +- Persist the history. +- Replace hard coded ANSI codes with constants or functions. diff --git a/dub.json b/dub.json new file mode 100644 index 0000000..743f619 --- /dev/null +++ b/dub.json @@ -0,0 +1,9 @@ +{ + "dependencies": { + "tanya": "~>0.19.0" + }, + "name": "elna", + "targetType": "executable", + "targetPath": "build/bin", + "mainSourceFile": "source/main.d" +} diff --git a/include/elna/interactive.hpp b/include/elna/interactive.hpp index f5c1fb3..8106060 100644 --- a/include/elna/interactive.hpp +++ b/include/elna/interactive.hpp @@ -5,10 +5,11 @@ #include #include -#include "elna/lexer.hpp" #include "elna/state.hpp" #include "elna/history.hpp" +#define BOOST_PROCESS_USE_STD_FS + namespace elna { /** @@ -42,15 +43,14 @@ namespace elna * \param line The command and arguments. * \return Whether the input shoud continued (no exit requested). */ - bool execute(const std::vector& line); + bool execute(const std::string& line); /** * Runs the binary specified in its argument. * * \param program Program name in PATH. - * \param arguments Command arguments. */ - void launch(const std::string& program, const std::vector& arguments); + void launch(const std::string& program); /** * Enables the raw mode. @@ -74,7 +74,16 @@ namespace elna /** * Calls autocompletion. * + * \param User input. + * \param Cursor position in the user input. * \return Selected item. */ - std::string complete(); + std::string complete(const std::string& input, const std::size_t position); + + /** + * Prints the message from the exception. + * + * \param exception The exception to print. + */ + void print_exception(const std::exception& exception); } diff --git a/include/elna/lexer.hpp b/include/elna/lexer.hpp index 159cb8f..b9435d4 100644 --- a/include/elna/lexer.hpp +++ b/include/elna/lexer.hpp @@ -14,9 +14,10 @@ namespace elna class const_iterator { std::string::const_iterator m_buffer; - Position m_position; + source_position m_position; - const_iterator(std::string::const_iterator buffer, const Position& position); + const_iterator(std::string::const_iterator buffer, + const source_position position = source_position()); public: using iterator_category = std::forward_iterator_tag; @@ -25,7 +26,7 @@ namespace elna using pointer = const value_type *; using reference = const value_type&; - const Position& position() const noexcept; + const source_position& position() const noexcept; reference operator*() const noexcept; pointer operator->() const noexcept; @@ -68,11 +69,11 @@ namespace elna type of() const noexcept; const value& identifier() const noexcept; - const Position& position() const noexcept; + const source_position& position() const noexcept; private: std::string m_value; - Position m_position; + source_position m_position; type m_type; }; diff --git a/include/elna/result.hpp b/include/elna/result.hpp index b14f5dd..078b500 100644 --- a/include/elna/result.hpp +++ b/include/elna/result.hpp @@ -8,7 +8,7 @@ namespace elna /** * Position in the source text. */ - struct Position + struct source_position { /// Line. std::size_t line = 1; @@ -20,21 +20,21 @@ namespace elna /** * A compilation error consists of an error message and position. */ - struct CompileError + struct compile_error { private: - char const *message_; - Position position_; + char const *message; + source_position position; public: /** * @param message Error text. * @param position Error position in the source text. */ - CompileError(char const *message, Position position) noexcept; + compile_error(char const *message, const source_position position) noexcept; /// Error text. - char const *message() const noexcept; + const char *what() const noexcept; /// Error line in the source text. std::size_t line() const noexcept; @@ -44,5 +44,5 @@ namespace elna }; template - using result = boost::outcome_v2::result; + using result = boost::outcome_v2::result; } diff --git a/include/elna/state.hpp b/include/elna/state.hpp index b0284cd..4c7e8f4 100644 --- a/include/elna/state.hpp +++ b/include/elna/state.hpp @@ -38,30 +38,28 @@ namespace elna num_lock = 0b10000000, ///< 128 }; - class key + struct key { - std::variant store; - std::uint8_t modifiers; + using char_t = std::uint32_t; - public: key(); - explicit key(const unsigned char c, const std::uint8_t modifiers = 0); - explicit key(const unsigned char c, const modifier modifiers); + explicit key(const char_t c, const std::uint8_t modifiers = 0); + explicit key(const char_t c, const modifier modifiers); explicit key(const special_key control, const std::uint8_t modifiers = 0); explicit key(const special_key control, const modifier modifiers); - bool operator==(const unsigned char c) const; - friend bool operator==(const unsigned char c, const key& that); + bool operator==(const char_t c) const; + friend bool operator==(const char_t c, const key& that); bool operator==(const special_key control) const; friend bool operator==(const special_key control, const key& that); bool operator==(const key& that) const; - bool operator!=(const unsigned char c) const; - friend bool operator!=(const unsigned char c, const key& that); + bool operator!=(const char_t c) const; + friend bool operator!=(const char_t c, const key& that); bool operator!=(const special_key control) const; friend bool operator!=(const special_key control, const key& that); bool operator!=(const key& that) const; - unsigned char character() const; + char_t character() const; special_key control() const; bool modified(const std::uint8_t modifiers) const noexcept; bool modified(const modifier modifiers) const noexcept; @@ -69,6 +67,10 @@ namespace elna bool is_character() const noexcept; bool is_control() const noexcept; bool empty() const noexcept; + + private: + std::variant store; + std::uint8_t modifiers; }; /** @@ -143,12 +145,20 @@ namespace elna */ std::size_t absolute_position() const noexcept; + /** + * Cursor position relative to the start of the user input. + * + * \returns Cursor position. + */ + std::size_t relative_position() const noexcept; + /** * Inserts text at cursor position. * * \param text Text to insert. */ void insert(const std::string& text); + void insert(const key::char_t text); /** * Replaces the user input with the given string. diff --git a/shell/history.cpp b/shell/history.cpp index bb58dd4..aad7375 100644 --- a/shell/history.cpp +++ b/shell/history.cpp @@ -9,8 +9,8 @@ namespace elna void editor_history::push(const std::string& entry) { - commands.push_front(entry); - current_pointer = commands.cbegin(); + commands.push_back(entry); + current_pointer = commands.cend(); } void editor_history::clear() @@ -46,4 +46,9 @@ namespace elna { return this->commands.cend(); } + + editor_history::const_iterator editor_history::current() const noexcept + { + return this->current_pointer; + } } diff --git a/shell/interactive.cpp b/shell/interactive.cpp index 22af464..e10446a 100644 --- a/shell/interactive.cpp +++ b/shell/interactive.cpp @@ -1,9 +1,10 @@ +#include "elna/interactive.hpp" + +#include #include #include #include -#include "elna/interactive.hpp" - namespace elna { static termios original_termios; @@ -39,9 +40,8 @@ namespace elna break; } history.push(line.value().command_line()); - auto tokens = lex(line.value().command_line()); - if (!execute(tokens.assume_value())) + if (!execute(line.value().command_line())) { break; } @@ -53,6 +53,7 @@ namespace elna { editor_state state; std::string buffer; + std::string saved; state.draw(std::back_inserter(buffer)); write(STDOUT_FILENO, buffer.data(), buffer.size()); @@ -81,7 +82,7 @@ namespace elna break; case action::complete: write(STDOUT_FILENO, "\x1b[1E\x1b[2K", 8); - state.insert(complete()); + state.insert(complete(state.command_line(), state.relative_position())); write(STDOUT_FILENO, "\x1b[1A", 4); state.draw(std::back_inserter(buffer)); write(STDOUT_FILENO, buffer.data(), buffer.size()); @@ -104,18 +105,26 @@ namespace elna if (pressed_key == key(special_key::arrow_up, modifier::ctrl)) { - history_iterator = history.next(); + if (history.current() == history.cend()) + { + saved = state.command_line(); + } + history_iterator = history.prev(); } else if (pressed_key == key(special_key::arrow_down, modifier::ctrl)) { - history_iterator = history.prev(); + history_iterator = history.next(); + if (history_iterator == history.cend()) + { + state.load(saved); + } } if (history_iterator != history.cend()) { state.load(*history_iterator); - state.draw(std::back_inserter(buffer)); - write(STDOUT_FILENO, buffer.data(), buffer.size()); } + state.draw(std::back_inserter(buffer)); + write(STDOUT_FILENO, buffer.data(), buffer.size()); break; } } @@ -137,7 +146,7 @@ namespace elna { return key(); } - auto [number, last_char] = read_number(); + auto [number, last_char] = read_number(); std::uint8_t modifiers{ 0 }; if (last_char == ';') @@ -185,35 +194,51 @@ namespace elna return key(); } - bool execute(const std::vector& tokens) + void print_exception(const std::exception& exception) { - if (tokens.empty()) + std::string message{ exception.what() }; + message += "\r\n"; + write(STDERR_FILENO, message.data(), message.size()); + } + + bool execute(const std::string& line) + { + if (line.empty()) { return true; } - std::string program = tokens.front().identifier(); - if (program == "exit") + if (line == "exit") { return false; } - else if (program == "cd") + else if (boost::starts_with(line, "cd ")) { - std::filesystem::current_path(tokens[1].identifier()); + try + { + std::filesystem::current_path(line.substr(strlen("cd "))); + } + catch (const std::filesystem::filesystem_error& exception) + { + print_exception(exception); + } return true; } - std::vector arguments; - for (auto argument = tokens.cbegin() + 1; argument != std::cend(tokens); ++argument) - { - arguments.push_back(argument->identifier()); - } - launch(program, arguments); + launch(line); return true; } - void launch(const std::string& program, const std::vector& arguments) + void launch(const std::string& program) { - boost::process::system(boost::process::search_path(program), arguments); + try + { + boost::process::system(program); + } + catch (const boost::process::process_error& exception) + { + print_exception(exception); + } + enable_raw_mode(); } bool enable_raw_mode() @@ -240,10 +265,17 @@ namespace elna tcsetattr(STDIN_FILENO, TCSAFLUSH, &original_termios); } - std::string complete() + std::string complete(const std::string& input, const std::size_t position) { + std::filesystem::path program = boost::process::search_path("fzf"); + + if (program.empty()) + { + return ""; + } boost::process::ipstream output; - boost::process::system("/usr/bin/fzf", "--height=10", "--layout=reverse", + boost::process::system(program, + "--height=10", "--layout=reverse", "-1", "--no-multi", boost::process::std_out > output); std::string selections; diff --git a/shell/result.cpp b/shell/result.cpp index 8531512..8054eb3 100644 --- a/shell/result.cpp +++ b/shell/result.cpp @@ -2,24 +2,24 @@ namespace elna { - CompileError::CompileError(char const *message, Position position) noexcept + compile_error::compile_error(const char *message, const source_position position) noexcept { - this->message_ = message; - this->position_ = position; + this->message = message; + this->position = position; } - char const *CompileError::message() const noexcept + char const *compile_error::what() const noexcept { - return this->message_; + return this->message; } - std::size_t CompileError::line() const noexcept + std::size_t compile_error::line() const noexcept { - return this->position_.line; + return this->position.line; } - std::size_t CompileError::column() const noexcept + std::size_t compile_error::column() const noexcept { - return this->position_.column; + return this->position.column; } } diff --git a/shell/state.cpp b/shell/state.cpp index 61b41c8..f148980 100644 --- a/shell/state.cpp +++ b/shell/state.cpp @@ -1,3 +1,6 @@ +#include +#include + #include "elna/state.hpp" namespace elna @@ -12,16 +15,16 @@ namespace elna } key::key() - : store(static_cast(0x1b)) + : store(0x1bu) { } - key::key(const unsigned char c, const std::uint8_t modifiers) + key::key(const char_t c, const std::uint8_t modifiers) : store(c), modifiers(modifiers == 0 ? 0 : modifiers - 1) { } - key::key(const unsigned char c, const modifier modifiers) + key::key(const char_t c, const modifier modifiers) : store(c), modifiers(static_cast(modifiers)) { } @@ -36,9 +39,9 @@ namespace elna { } - unsigned char key::character() const + key::char_t key::character() const { - return std::get(this->store); + return std::get(this->store); } special_key key::control() const @@ -56,10 +59,10 @@ namespace elna return this->modifiers == static_cast(modifiers); } - bool key::operator==(const unsigned char c) const + bool key::operator==(const char_t c) const { - return std::holds_alternative(this->store) - && std::get(this->store) == c; + return std::holds_alternative(this->store) + && std::get(this->store) == c; } bool key::operator==(const special_key control) const @@ -73,7 +76,7 @@ namespace elna return this->store == that.store; } - bool key::operator!=(const unsigned char c) const + bool key::operator!=(const char_t c) const { return !(*this == c); } @@ -90,7 +93,7 @@ namespace elna bool key::is_character() const noexcept { - return std::holds_alternative(this->store); + return std::holds_alternative(this->store); } bool key::is_control() const noexcept @@ -131,6 +134,7 @@ namespace elna void editor_state::load(const std::string& text) { this->input = text; + this->position = this->input.size() + 1; } const std::string& editor_state::command_line() const noexcept @@ -143,9 +147,36 @@ namespace elna return this->prompt.size() + this->position; } + std::size_t editor_state::relative_position() const noexcept + { + return this->position; + } + void editor_state::insert(const std::string& text) { - this->input.insert(std::end(this->input), std::cbegin(text), std::cend(text)); + this->input.insert(this->position - 1, text); + this->position += text.size(); + } + + void editor_state::insert(const key::char_t text) + { + if (text == 0u) + { + this->input.insert(this->position - 1, '\0', 1); + ++this->position; + return; + } + key::char_t buffer = boost::endian::native_to_big(text); + + const char *begin = reinterpret_cast(&buffer); + const char *end = begin + sizeof(key::char_t); + const char *significant = std::find_if(begin, end, + [](const char byte) { + return byte != 0; + }); + + this->input.insert(this->position - 1, significant, end - significant); + this->position += end - significant; } action editor_state::process_keypress(const key& press) @@ -202,12 +233,21 @@ namespace elna this->input.erase(this->position - 1, 1); } return action::redraw; + case 9: // Tab. + return action::complete; case '\r': return action::finalize; default: - ++this->position; - this->input.push_back(press.character()); - return action::write; + if (this->position - 1 < input.size()) + { + insert(press.character()); + return action::redraw; + } + else + { + insert(press.character()); + return action::write; + } } } } diff --git a/source/elna/arguments.d b/source/elna/arguments.d new file mode 100644 index 0000000..1ad928a --- /dev/null +++ b/source/elna/arguments.d @@ -0,0 +1,154 @@ +/** + * Argument parsing. + */ +module elna.arguments; + +import std.algorithm; +import std.range; +import std.sumtype; + +struct ArgumentError +{ + enum Type + { + expectedOutputFile, + noInput, + superfluousArguments, + } + + private Type type_; + private string argument_; + + @property Type type() const @nogc nothrow pure @safe + { + return this.type_; + } + + @property string argument() const @nogc nothrow pure @safe + { + return this.argument_; + } + + void toString(OR)(OR range) + if (isOutputRage!OR) + { + final switch (Type) + { + case Type.expectedOutputFile: + put(range, "Expected an output filename after -o"); + break; + case Type.noInput: + put(range, "No input files specified"); + break; + } + } +} + +/** + * Supported compiler arguments. + */ +struct Arguments +{ + private bool assembler_; + private string output_; + private string inFile_; + + @property string inFile() @nogc nothrow pure @safe + { + return this.inFile_; + } + + /** + * Returns: Whether to generate assembly instead of an object file. + */ + @property bool assembler() const @nogc nothrow pure @safe + { + return this.assembler_; + } + + /** + * Returns: Output file. + */ + @property string output() const @nogc nothrow pure @safe + { + return this.output_; + } + + /** + * Parse command line arguments. + * + * The first argument is expected to be the program name (and it is + * ignored). + * + * Params: + * arguments = Command line arguments. + * + * Returns: Parsed arguments or an error. + */ + static SumType!(ArgumentError, Arguments) parse(string[] arguments) + @nogc nothrow pure @safe + { + if (!arguments.empty) + { + arguments.popFront; + } + alias ReturnType = typeof(return); + + return parseArguments(arguments).match!( + (Arguments parsed) { + if (parsed.inFile is null) + { + return ReturnType(ArgumentError(ArgumentError.Type.noInput)); + } + else if (!arguments.empty) + { + return ReturnType(ArgumentError( + ArgumentError.Type.superfluousArguments, + arguments.front + )); + } + return ReturnType(parsed); + }, + (ArgumentError argumentError) => ReturnType(argumentError) + ); + } + + private static SumType!(ArgumentError, Arguments) parseArguments(ref string[] arguments) + @nogc nothrow pure @safe + { + Arguments parsed; + + while (!arguments.empty) + { + if (arguments.front == "-s") + { + parsed.assembler_ = true; + } + else if (arguments.front == "-o") + { + if (arguments.empty) + { + return typeof(return)(ArgumentError( + ArgumentError.Type.expectedOutputFile, + arguments.front + )); + } + arguments.popFront; + parsed.output_ = arguments.front; + } + else if (arguments.front == "--") + { + arguments.popFront; + parsed.inFile_ = arguments.front; + arguments.popFront; + break; + } + else if (!arguments.front.startsWith("-")) + { + parsed.inFile_ = arguments.front; + } + arguments.popFront; + } + return typeof(return)(parsed); + } +} diff --git a/source/elna/backend.d b/source/elna/backend.d new file mode 100644 index 0000000..ec4ed7a --- /dev/null +++ b/source/elna/backend.d @@ -0,0 +1,105 @@ +module elna.backend; + +import core.stdc.stdio; +import core.stdc.stdlib; +import elna.elf; +import elna.ir; +import elna.extended; +import elna.riscv; +import elna.lexer; +import elna.parser; +import elna.result; +import std.algorithm; +import std.sumtype; +import std.typecons; +import tanya.os.error; +import tanya.container.array; +import tanya.container.string; +import tanya.memory.allocator; + +private Nullable!String readSource(string source) @nogc +{ + enum size_t bufferSize = 255; + auto sourceFilename = String(source); + + return readFile(sourceFilename).match!( + (ErrorCode errorCode) { + perror(sourceFilename.toStringz); + return Nullable!String(); + }, + (Array!ubyte contents) => nullable(String(cast(char[]) contents.get)) + ); +} + +int generate(string inFile, ref String outputFilename) @nogc +{ + auto sourceText = readSource(inFile); + if (sourceText.isNull) + { + return 3; + } + auto tokens = lex(sourceText.get.get); + if (!tokens.valid) + { + auto compileError = tokens.error.get; + printf("%lu:%lu: %s\n", compileError.line, compileError.column, compileError.message.ptr); + return 1; + } + auto ast = parse(tokens.result); + if (!ast.valid) + { + auto compileError = ast.error.get; + printf("%lu:%lu: %s\n", compileError.line, compileError.column, compileError.message.ptr); + return 2; + } + auto transformVisitor = cast(TransformVisitor) malloc(__traits(classInstanceSize, TransformVisitor)); + (cast(void*) transformVisitor)[0 .. __traits(classInstanceSize, TransformVisitor)] = __traits(initSymbol, TransformVisitor)[]; + + auto ir = transformVisitor.visit(ast.result); + + transformVisitor.__xdtor(); + free(cast(void*) transformVisitor); + + auto handle = File.open(outputFilename.toStringz, BitFlags!(File.Mode)(File.Mode.truncate)); + if (!handle.valid) + { + return 1; + } + auto program = writeNext(ir); + auto elf = Elf!ELFCLASS32(move(handle)); + auto readOnlyData = Array!ubyte(cast(const(ubyte)[]) "%d\n".ptr[0 .. 4]); // With \0. + + elf.addReadOnlyData(String(".CL0"), readOnlyData); + elf.addCode(program.name, program.text); + + elf.addExternSymbol(String("printf")); + foreach (ref reference; program.symbols) + { + elf.Rela relocationEntry = { + r_offset: cast(elf.Addr) reference.offset + }; + elf.Rela relocationSub = { + r_offset: cast(elf.Addr) reference.offset, + r_info: R_RISCV_RELAX + }; + + final switch (reference.target) + { + case Reference.Target.text: + relocationEntry.r_info = R_RISCV_CALL; + break; + case Reference.Target.high20: + relocationEntry.r_info = R_RISCV_HI20; + break; + case Reference.Target.lower12i: + relocationEntry.r_info = R_RISCV_LO12_I; + break; + } + + elf.relocate(reference.name, relocationEntry, relocationSub); + } + + elf.finish(); + + return 0; +} diff --git a/source/elna/elf.d b/source/elna/elf.d new file mode 100644 index 0000000..81dd7da --- /dev/null +++ b/source/elna/elf.d @@ -0,0 +1,1060 @@ +module elna.elf; + +import elna.extended; +import elna.result; +import std.algorithm; +import tanya.container.array; +import tanya.container.hashtable; +import tanya.container.string; + +/// Unsigned program address. +alias Elf64_Addr = ulong; +/// Unsigned file offset. +alias Elf64_Off = ulong; +/// Unsigned medium integer. +alias Elf64_Half = ushort; +/// Unsigned integer. +alias Elf64_Word = uint; +/// Signed integer. +alias Elf64_Sword = int; +/// Unsigned long integer. +alias Elf64_Xword = ulong; +/// Signed long integer. +alias Elf64_Sxword = long; + +/// Unsigned program address. +alias Elf32_Addr = uint; +/// Unsigned file offset. +alias Elf32_Off = uint; +/// Unsigned medium integer. +alias Elf32_Half = ushort; +/// Unsigned integer. +alias Elf32_Word = uint; +/// Signed integer. +alias Elf32_Sword = int; + +enum : size_t +{ + /// File identification. + EI_MAG0 = 0, + /// File identification. + EI_MAG1 = 1, + /// File identification. + EI_MAG2 = 2, + /// File identification. + EI_MAG3 = 3, + /// File class. + EI_CLASS = 4, + /// Data encoding. + EI_DATA = 5, + /// File version. + EI_VERSION = 6, + /// Start of padding bytes. + EI_PAD = 7, + /// Size of e_ident[] + EI_NIDENT = 16 +} + +enum : ubyte +{ + /// e_ident[EI_MAG0]. + ELFMAG0 = 0x7f, + /// e_ident[EI_MAG1]. + ELFMAG1 = 'E', + /// e_ident[EI_MAG2]. + ELFMAG2 = 'L', + /// e_ident[EI_MAG3]. + ELFMAG3 = 'F' +} + +/** + * File header. + */ +struct Elf64_Ehdr +{ + /// ELF identification. + ubyte[EI_NIDENT] e_ident; + /// Object file type. + Elf64_Half e_type; + /// Machine type. + Elf64_Half e_machine; + /// Object file version + Elf64_Word e_version; + /// Entry point address. + Elf64_Addr e_entry; + /// Program header offset. + Elf64_Off e_phoff; + /// Section header offset. + Elf64_Off e_shoff; + /// Processor-specific flags. + Elf64_Word e_flags; + /// ELF header size. + Elf64_Half e_ehsize; + /// Size of program header entry. + Elf64_Half e_phentsize; + /// Number of program header entries. + Elf64_Half e_phnum; + /// Size of section header entry. + Elf64_Half e_shentsize; + /// Number of section header entries. + Elf64_Half e_shnum; + /// Section name string table index. + Elf64_Half e_shstrndx; +} + +/** + * File header. + */ +struct Elf32_Ehdr { + /// ELF identification. + ubyte[EI_NIDENT] e_ident; + /// Object file type. + Elf32_Half e_type; + /// Machine type. + Elf32_Half e_machine; + /// Object file version + Elf32_Word e_version; + /// Entry point address. + Elf32_Addr e_entry; + /// Program header offset. + Elf32_Off e_phoff; + /// Section header offset. + Elf32_Off e_shoff; + /// Processor-specific flags. + Elf32_Word e_flags; + /// ELF header size. + Elf32_Half e_ehsize; + /// Size of program header entry. + Elf32_Half e_phentsize; + /// Number of program header entries. + Elf32_Half e_phnum; + /// Size of section header entry. + Elf32_Half e_shentsize; + /// Number of section header entries. + Elf32_Half e_shnum; + /// Section name string table index. + Elf32_Half e_shstrndx; +} + +/** + * Section header. + */ +struct Elf64_Shdr +{ + /// Section name. + Elf64_Word sh_name; + /// Section type. + Elf64_Word sh_type; + /// Section attributes. + Elf64_Xword sh_flags; + /// Virtual address in memory. + Elf64_Addr sh_addr; + /// Offset in file. + Elf64_Off sh_offset; + /// Size of section. + Elf64_Xword sh_size; + /// Link to other section. + Elf64_Word sh_link; + /// Miscellaneous information. + Elf64_Word sh_info; + /// Address alignment boundary. + Elf64_Xword sh_addralign; + /// Size of entries, if section has table. + Elf64_Xword sh_entsize; +} + +/** + * Section header. + */ +struct Elf32_Shdr +{ + /// Section name. + Elf32_Word sh_name; + /// Section type. + Elf32_Word sh_type; + /// Section attributes. + Elf32_Word sh_flags; + /// Virtual address in memory. + Elf32_Addr sh_addr; + /// Offset in file. + Elf32_Off sh_offset; + /// Size of section. + Elf32_Word sh_size; + /// Link to other section. + Elf32_Word sh_link; + /// Miscellaneous information. + Elf32_Word sh_info; + /// Address alignment boundary. + Elf32_Word sh_addralign; + /// Size of entries, if section has table. + Elf32_Word sh_entsize; +} + +/** + * Symbol table entry. + */ +struct Elf64_Sym +{ + /// Symbol name. + Elf64_Word st_name; + /// Type and Binding attributes. + ubyte st_info; + /// Reserved. + ubyte st_other; + /// Section table index. + Elf64_Half st_shndx; + /// Symbol value. + Elf64_Addr st_value; + /// Size of object (e.g., common). + Elf64_Xword st_size; +} + +/** + * Relocation entry. + */ +struct Elf64_Rel +{ + /// Address of reference. + Elf64_Addr r_offset; + /// Symbol index and type of relocation. + Elf64_Xword r_info; +} + +/** + * Relocation entry with explicit addend. + */ +struct Elf64_Rela +{ + /// Address of reference. + Elf64_Addr r_offset; + /// Symbol index and type of relocation. + Elf64_Xword r_info; + /// Constant part of expression. + Elf64_Sxword r_addend; +} + +/** + * Symbol table entry. + */ +struct Elf32_Sym +{ + /// Symbol name. + Elf32_Word st_name; + /// Symbol value. + Elf32_Addr st_value; + /// Size of object (e.g., common). + Elf32_Word st_size; + /// Type and Binding attributes. + ubyte st_info; + /// Reserved. + ubyte st_other; + /// Section table index. + Elf32_Half st_shndx; +} + +/** + * Relocation entry. + */ +struct Elf32_Rel +{ + /// Address of reference. + Elf32_Addr r_offset; + /// Symbol index and type of relocation. + Elf32_Word r_info; +} + +/** + * Relocation entry with explicit addend. + */ +struct Elf32_Rela +{ + /// Address of reference. + Elf32_Addr r_offset; + /// Symbol index and type of relocation. + Elf32_Word r_info; + /// Constant part of expression. + Elf32_Sword r_addend; +} + +/// Section Types, sh_type. +enum : Elf64_Word +{ + /// Marks an unused section header. + SHT_NULL = 0, + /// Contains information defined by the program. + SHT_PROGBITS = 1, + /// Contains a linker symbol table. + SHT_SYMTAB = 2, + /// Contains a string table. + SHT_STRTAB = 3, + /// Contains “Rela” type relocation entries. + SHT_RELA = 4, + /// Contains a symbol hash table + SHT_HASH = 5, + /// Contains dynamic linking tables + SHT_DYNAMIC = 6, + /// Contains note information + SHT_NOTE = 7, + /// Contains uninitialized space; does not occupy any space in the file. + SHT_NOBITS = 8, + /// Contains "Rel" type relocation entries. + SHT_REL = 9, + /// Reserved. + SHT_SHLIB = 10, + /// Contains a dynamic loader symbol table. + SHT_DYNSYM = 11, + /// Environment-specific use. + SHT_LOOS = 0x60000000, + SHT_HIOS = 0x6FFFFFFF, + /// Processor-specific use. + SHT_LOPROC = 0x70000000, + SHT_HIPROC = 0x7FFFFFFF, +} + +/** + * Section Attributes, sh_flags. + */ +enum : Elf64_Xword +{ + /// Section contains writable data. + SHF_WRITE = 0x1, + /// Section is allocated in memory image of program. + SHF_ALLOC = 0x2, + /// Section contains executable instructions. + SHF_EXECINSTR = 0x4, + /// Environment-specific use. + SHF_MASKOS = 0x0F000000, + /// Processor-specific use. + SHF_MASKPROC = 0xF0000000, +} + +ubyte ELF64_R_SYM(Elf64_Xword i) @nogc nothrow pure @safe +{ + return cast(ubyte) (i >> 32); +} + +Elf64_Xword ELF64_R_TYPE(Elf64_Xword i) @nogc nothrow pure @safe +{ + return i & 0xffffffffL; +} + +Elf64_Xword ELF64_R_INFO(Elf64_Xword s, Elf64_Xword t) @nogc nothrow pure @safe +{ + return (s << 32) + (t & 0xffffffffL); +} + +ubyte ELF32_ST_BIND(ubyte i) @nogc nothrow pure @safe +{ + return i >> 4; +} + +ubyte ELF32_ST_TYPE(ubyte i) @nogc nothrow pure @safe +{ + return i & 0xf; +} + +ubyte ELF32_ST_INFO(Elf32_Word b, ubyte t) @nogc nothrow pure @safe +{ + return cast(ubyte) ((b << 4) + (t & 0xf)); +} + +Elf32_Word ELF32_R_SYM(Elf32_Word i) @nogc nothrow pure @safe +{ + return i >> 8; +} + +ubyte ELF32_R_TYPE(Elf32_Word i) @nogc nothrow pure @safe +{ + return cast(ubyte) i; +} + +Elf32_Word ELF32_R_INFO(Elf32_Word s, Elf32_Word t) @nogc nothrow pure @safe +{ + return (s << 8) + t; +} + +enum : uint +{ + /// Not visible outside the object file. + STB_LOCAL = 0, + /// Global symbol, visible to all object files. + STB_GLOBAL = 1, + /// Global scope, but with lower precedence than global symbols. + STB_WEAK = 2, + /// Environment-specific use. + STB_LOOS = 10, + STB_HIOS = 12, + /// Processor-specific use. + STB_LOPROC = 13, + STB_HIPROC = 15, +} + +enum : uint +{ + /// No type specified (e.g., an absolute symbol). + STT_NOTYPE = 0, + /// Data object. + STT_OBJECT = 1, + /// Function entry point. + STT_FUNC = 2, + /// Symbol is associated with a section. + STT_SECTION = 3, + /// Source file associated with the object file. + STT_FILE = 4, + /// Environment-specific use. + STT_LOOS = 10, + STT_HIOS = 12, + /// Processor-specific use. + STT_LOPROC = 13, + STT_HIPROC = 15, +} + +/// Special Section Indices. +enum : ushort +{ + /// Used to mark an undefined or meaningless section reference. + SHN_UNDEF = 0, + /// This value specifies the lower bound of the range of reserved indexes. + SHN_LORESERVE = 0xff00, + /// Processor-specific use. + SHN_LOPROC = 0xFF00, + SHN_HIPROC = 0xFF1F, + /// Environment-specific use. + SHN_LOOS = 0xFF20, + SHN_HIOS = 0xFF3F, + /// Indicates that the corresponding reference is an absolute value. + SHN_ABS = 0xFFF1, + /** + * Indicates a symbol that has been declared as a common block (Fortran + * COMMON or C tentative declaration). + */ + SHN_COMMON = 0xFFF2, +} + +/** + * Object File Classes, e_ident[EI_CLASS]. + */ +enum : ubyte +{ + /// Invalid class. + ELFCLASSNONE = 0, + /// 32-bit objects. + ELFCLASS32 = 1, + /// 64-bit objects. + ELFCLASS64 = 2 +} + +enum : ubyte { + /// Invalid version. + EV_NONE = 0, + /// Current version. + EV_CURRENT = 1 +} + +/** + * Data Encodings, e_ident[EI_DATA]. + */ +enum : ubyte +{ + /// Object file data structures are little-endian. + ELFDATA2LSB = 1, + /// Object file data structures are big-endian. + ELFDATA2MSB = 2, +} + +/** + * Operating System and ABI Identifiers, e_ident[EI_OSABI]. + */ +enum EI_OSABI : ubyte +{ + /// System V ABI. + ELFOSABI_SYSV = 0, + /// HP-UX operating system. + ELFOSABI_HPUX = 1, + /// Standalone (embedded) application. + ELFOSABI_STANDALONE = 255, +} + +enum : Elf64_Half +{ + ET_NONE = 0, /// No file type. + ET_REL = 1, /// Relocatable object file. + ET_EXEC = 2, /// Executable file. + ET_DYN = 3, /// Shared object file. + ET_CORE = 4, /// Core file. + ET_LOOS = 0xFE00, /// Environment-specific use. + ET_HIOS = 0xFEFF, + ET_LOPROC = 0xFF00, /// Processor-specific use. + ET_HIPROC = 0xFFFF, +} + +enum : ubyte +{ + R_RISCV_NONE = 0, + /// 32-bit relocation. + R_RISCV_32 = 1, + /// 64-bit relocation. + R_RISCV_64 = 2, + /// Relocation against a local symbol in a shared object. + R_RISCV_RELATIVE = 3, + /// Must be in executable; not allowed in shared library. + R_RISCV_COPY = 4, + /// Indicates the symbol associated with a PLT entry. + R_RISCV_JUMP_SLOT = 5, + R_RISCV_TLS_DTPMOD32 = 6, + R_RISCV_TLS_DTPMOD64 = 7, + R_RISCV_TLS_DTPREL32 = 8, + R_RISCV_TLS_DTPREL64 = 9, + R_RISCV_TLS_TPREL32 = 10, + R_RISCV_TLS_TPREL64 = 11, + /// 12-bit PC-relative branch offset. + R_RISCV_BRANCH = 16, + /// 20-bit PC-relative jump offset. + R_RISCV_JAL = 17, + /// 32-bit PC-relative function call, macros `call`, `tail`. + R_RISCV_CALL = 18, + /// 32-bit PC-relative function call, macros `call`, `tail` (PIC). + R_RISCV_CALL_PLT = 19, + /// High 20 bits of 32-bit PC-relative GOT access, `%got_pcrel_hi(symbol)`. + R_RISCV_GOT_HI20 = 20, + /// High 20 bits of 32-bit PC-relative TLS IE GOT access, macro `la.tls.ie`. + R_RISCV_TLS_GOT_HI20 = 21, + /// High 20 bits of 32-bit PC-relative TLS GD GOT reference, macro `la.tls.gd`. + R_RISCV_TLS_GD_HI20 = 22, + /// High 20 bits of 32-bit PC-relative reference, `%pcrel_hi(symbol)`. + R_RISCV_PCREL_HI20 = 23, + /// Low 12 bits of a 32-bit PC-relative, `%pcrel_lo(address of %pcrel_hi)`, the addend must be 0. + R_RISCV_PCREL_LO12_I = 24, + /// Low 12 bits of a 32-bit PC-relative, `%pcrel_lo(address of %pcrel_hi)`, the addend must be 0. + R_RISCV_PCREL_LO12_S = 25, + /// High 20 bits of 32-bit absolute address, `%hi(symbol)`. + R_RISCV_HI20 = 26, + /// Low 12 bits of 32-bit absolute address, `%lo(symbol)`. + R_RISCV_LO12_I = 27, + /// Low 12 bits of 32-bit absolute address, `%lo(symbol)`. + R_RISCV_LO12_S = 28, + /// High 20 bits of TLS LE thread pointer offset, `%tprel_hi(symbol)`. + R_RISCV_TPREL_HI20 = 29, + /// Low 12 bits of TLS LE thread pointer offset, `%tprel_lo(symbol)`. + R_RISCV_TPREL_LO12_I = 30, + /// Low 12 bits of TLS LE thread pointer offset, `%tprel_lo(symbol)`. + R_RISCV_TPREL_LO12_S = 31, + /// TLS LE thread pointer usage, `%tprel_add(symbol)`. + R_RISCV_TPREL_ADD = 32, + /// 8-bit label addition. + R_RISCV_ADD8 = 33, + /// 16-bit label addition. + R_RISCV_ADD16 = 34, + /// 32-bit label addition. + R_RISCV_ADD32 = 35, + /// 64-bit label addition. + R_RISCV_ADD64 = 36, + /// 8-bit label subtraction. + R_RISCV_SUB8 = 37, + /// 16-bit label subtraction. + R_RISCV_SUB16 = 38, + /// 32-bit label subtraction. + R_RISCV_SUB32 = 39, + /// 64-bit label subtraction. + R_RISCV_SUB64 = 40, + /// GNU {Cpp} vtable hierarchy. + R_RISCV_GNU_VTINHERIT = 41, + /// GNU {Cpp} vtable member usage. + R_RISCV_GNU_VTENTRY = 42, + /// Alignment statement. + R_RISCV_ALIGN = 43, + /// 8-bit PC-relative branch offset. + R_RISCV_RVC_BRANCH = 44, + /// 11-bit PC-relative jump offset. + R_RISCV_RVC_JUMP = 45, + /// High 6 bits of 18-bit absolute address. + R_RISCV_RVC_LUI = 46, + /// Instruction can be relaxed, paired with a normal relocation at the same address. + R_RISCV_RELAX = 51, + /// Local label subtraction. + R_RISCV_SUB6 = 52, + /// Local label assignment. + R_RISCV_SET6 = 53, + /// Local label assignment. + R_RISCV_SET8 = 54, + /// Local label assignment. + R_RISCV_SET16 = 55, + /// Local label assignment. + R_RISCV_SET32 = 56, + /// 32-bit PC relative. + R_RISCV_32_PCREL = 57, + /// Relocation against a local ifunc symbol in a shared object. + R_RISCV_IRELATIVE = 58 +} + +auto pad(ubyte elfClass)(size_t value) @nogc +{ + static if (elfClass == ELFCLASS32) + { + return cast(Elf32_Word) (value / 4 + 1) * 4; + } + else static if (elfClass == ELFCLASS64) + { + return cast(Elf64_Xword) (value / 8 + 1) * 8; + } + else + { + static assert(false, "Invalid ELF class"); + } +} + +private struct Relocation(Sym, Rel) +{ + Sym symbol; + Array!Rel relocations; +} + +struct Elf(ubyte elfClass) +{ + static if (elfClass == ELFCLASS32) + { + alias Addr = Elf32_Addr; + alias Off = Elf32_Off; + alias Half = Elf32_Half; + alias Word = Elf32_Word; + alias Sword = Elf32_Sword; + alias Xword = Elf32_Word; + alias Sxword = Elf32_Sword; + + alias Ehdr = Elf32_Ehdr; + alias Shdr = Elf32_Shdr; + alias Rel = Elf32_Rel; + alias Rela = Elf32_Rela; + alias Sym = Elf32_Sym; + + alias R_SYM = ELF32_R_SYM; + alias R_TYPE = ELF32_R_TYPE; + alias R_INFO = ELF32_R_INFO; + alias ST_BIND = ELF32_ST_BIND; + alias ST_TYPE = ELF32_ST_TYPE; + alias ST_INFO = ELF32_ST_INFO; + } + else static if (elfClass == ELFCLASS64) + { + alias Addr = Elf64_Addr; + alias Off = Elf64_Off; + alias Half = Elf64_Half; + alias Word = Elf64_Word; + alias Sword = Elf64_Sword; + alias Xword = Elf64_Xword; + alias Sxword = Elf64_Sxword; + + alias Ehdr = Elf64_Ehdr; + alias Shdr = Elf64_Shdr; + alias Rel = Elf64_Rel; + alias Rela = Elf64_Rela; + alias Sym = Elf64_Sym; + + alias R_SYM = ELF64_R_SYM; + alias R_TYPE = ELF64_R_TYPE; + alias R_INFO = ELF64_R_INFO; + alias ST_BIND = ELF64_ST_BIND; + alias ST_TYPE = ELF64_ST_TYPE; + alias ST_INFO = ELF64_ST_INFO; + } + else + { + static assert(false, "Invalid ELF class"); + } + + private alias Relocation = .Relocation!(Sym, Rela); + + private Array!Shdr sectionHeaders; + private Off currentOffset = Elf32_Ehdr.sizeof; + static immutable char[52] sections = + "\0.symtab\0.strtab\0.shstrtab\0.text\0.rodata\0.rela.text\0"; + private String strings; + private File output; + private Array!ubyte readOnly; + + private HashTable!(String, Relocation) symbolTable; + + private enum HeaderName + { + text = 0x1b, + roData = 0x21, + string_ = 0x09, + headerString = 0x11, + symbol = 0x01, + rela = 0x29 + } + + static Elf opCall(File output) @nogc + { + Elf elf = Elf.init; + + elf.initializeSectionHeaders(); + elf.output = move(output); + + elf.output.seek(Ehdr.sizeof, File.Whence.set); + + elf.makeTextHeader(); + elf.makeRoDataHeader(); + elf.makeSymbolHeader(); + elf.makeRelaHeader(); + elf.makeStringHeader!(HeaderName.string_)(); + elf.makeStringHeader!(HeaderName.headerString)(); + + return elf; + } + + @disable this(this); + + void finish() @nogc + { + writeRoDataTable(); + writeSymbolTable(); + writeStringTables(); + + // End writing data, start writing headers. + + output.write((cast(ubyte*) this.sectionHeaders.get)[0 .. Shdr.sizeof * this.sectionHeaders.length]); + + writeFileHeader(); + } + + private Sym initializeSymbols() @nogc + { + // Zero symbol + Sym symbol; + symbol.st_name = 0; // Word + symbol.st_value = 0; // Addr + symbol.st_size = 0; // Word + symbol.st_info = 0; // char + symbol.st_other = 0; // char + symbol.st_shndx = 0; // Half word + + return symbol; + } + + private void makeStringHeader(HeaderName position)() @nogc + { + Shdr table; + + table.sh_name = position; + table.sh_type = SHT_STRTAB; + table.sh_flags = 0; + table.sh_addr = 0; + table.sh_offset = 0; + table.sh_size = 0; + table.sh_link = SHN_UNDEF; + table.sh_info = 0; + table.sh_addralign = 1; + table.sh_entsize = 0; + + this.sectionHeaders.insertBack(table); + } + + private void writeStringTables() @nogc + { + auto stringIndex = findHeader!(HeaderName.string_); + assert(stringIndex != -1); + + this.sectionHeaders[stringIndex].sh_offset = this.currentOffset; + this.sectionHeaders[stringIndex].sh_size = cast(Word) strings.length; + + output.write(cast(ubyte[]) this.strings.toStringz[0 .. this.strings.length + 1]); + this.currentOffset += this.strings.length + 1; + + auto headerStringIndex = findHeader!(HeaderName.headerString); + assert(stringIndex != -1); + + this.sectionHeaders[headerStringIndex].sh_offset = this.currentOffset; + this.sectionHeaders[headerStringIndex].sh_size = cast(Word) sections.length; + + output.write(cast(const(ubyte)[]) this.sections); + this.currentOffset += this.sections.length; + auto alignment = pad!ELFCLASS32(this.strings.length + 1 + this.sections.length); + const(ubyte)[4] padding = 0; + output.write(padding[0 .. alignment - this.strings.length - 1 - this.sections.length]); + this.currentOffset += alignment - this.strings.length - 1 - this.sections.length; + } + + private void makeSymbolHeader() @nogc + { + Shdr symbolTableHeader; + + symbolTableHeader.sh_name = HeaderName.symbol; + symbolTableHeader.sh_type = SHT_SYMTAB; + symbolTableHeader.sh_flags = 0; + symbolTableHeader.sh_addr = 0; + symbolTableHeader.sh_offset = 0; + symbolTableHeader.sh_size = 0; + // String table used by entries in this section. + symbolTableHeader.sh_link = 0; + symbolTableHeader.sh_info = 0; + symbolTableHeader.sh_addralign = 4; + symbolTableHeader.sh_entsize = Sym.sizeof; + + this.sectionHeaders.insertBack(symbolTableHeader); + } + + private void writeSymbolTable() @nogc + { + const index = findHeader!(HeaderName.symbol)(); + const stringIndex = findHeader!(HeaderName.string_)(); + const relaIndex = findHeader!(HeaderName.rela); + const textIndex = findHeader!(HeaderName.text)(); + + assert(index != -1); + assert(stringIndex != -1); + assert(relaIndex != -1); + assert(textIndex != -1); + + this.sectionHeaders[index].sh_offset = this.currentOffset; + this.sectionHeaders[index].sh_link = cast(Word) stringIndex; + this.sectionHeaders[index].sh_size = cast(Word) ((1 + symbolTable.length) * Sym.sizeof); + + this.sectionHeaders[relaIndex].sh_link = cast(Word) index; + this.sectionHeaders[relaIndex].sh_info = cast(Word) textIndex; + this.sectionHeaders[relaIndex].sh_offset = this.sectionHeaders[index].sh_offset + + this.sectionHeaders[index].sh_size; + + auto initialSymbol = initializeSymbols(); + output.write((cast(ubyte*) &initialSymbol)[0 .. Sym.sizeof]); + this.currentOffset += Sym.sizeof; + + int i = 1; + Array!Relocation symbols = Array!Relocation(this.symbolTable.byValue()); + auto rightRange = symbols[].partition!(symbol => ST_BIND(symbol.symbol.st_info) != STB_GLOBAL); + + // Greater than last local symbol. + this.sectionHeaders[index].sh_info = cast(Word) (symbols.length - rightRange.length + 1); + + foreach (ref symbol; symbols[]) + { + this.output.seek(this.sectionHeaders[relaIndex].sh_offset + this.sectionHeaders[relaIndex].sh_size, + File.Whence.set); + + if (!symbol.relocations.empty) + { + foreach (ref relocation; symbol.relocations[]) + { + relocation.r_info = R_INFO(i, R_TYPE(relocation.r_info)); + } + this.sectionHeaders[relaIndex].sh_flags = SHF_ALLOC; + const size = cast(Word) (Rela.sizeof * symbol.relocations.length); + + this.output.write((cast(ubyte*) symbol.relocations.get)[0 .. size]); + this.sectionHeaders[relaIndex].sh_size += size; + this.currentOffset += size; + } + + this.output.seek(this.sectionHeaders[index].sh_offset + i * Sym.sizeof, File.Whence.set); + output.write((cast(ubyte*) &symbol)[0 .. Sym.sizeof]); + this.currentOffset += Sym.sizeof; + ++i; + } + this.output.seek(0, File.Whence.end); + } + + void addCode(String name, ref Array!ubyte text) + @nogc + { + this.output.write(text.get); + + auto textHeaderIndex = findHeader!(HeaderName.text)(); + assert(textHeaderIndex != -1); + + this.strings.insertBack("\0"); + + Sym symbol; + // Main function + symbol.st_name = cast(Word) this.strings.length; + symbol.st_value = 0; + symbol.st_size = cast(Word) text.length; + symbol.st_info = ST_INFO(STB_GLOBAL, STT_FUNC); + symbol.st_other = 0; // char + // .text header index, half word + symbol.st_shndx = cast(Half) textHeaderIndex; + this.symbolTable[name] = Relocation(symbol); + + this.strings.insertBack(name[]); + + this.sectionHeaders[textHeaderIndex].sh_size += text.length; + this.currentOffset += text.length; + } + + void addReadOnlyData(String name, ref Array!ubyte data) @nogc + { + auto roDataIndex = findHeader!(HeaderName.roData)(); + assert(roDataIndex != -1); + + this.strings.insertBack("\0"); + + Sym symbol; + // Main function + symbol.st_name = cast(Word) this.strings.length; + symbol.st_value = 0; + symbol.st_size = cast(Word) data.length; + symbol.st_info = ST_INFO(STB_LOCAL, STT_NOTYPE); + symbol.st_other = 0; // char + // .text header index, half word + symbol.st_shndx = cast(Half) roDataIndex; + this.symbolTable[name] = Relocation(symbol); + + this.strings.insertBack(name[]); + this.readOnly.insertBack(data[]); + } + + void addExternSymbol(String name) @nogc + { + Sym usedSymbolEntry; + + this.strings.insertBack("\0"); + usedSymbolEntry.st_name = cast(Word) this.strings.length; + usedSymbolEntry.st_value = 0; + usedSymbolEntry.st_size = 0; + usedSymbolEntry.st_info = ST_INFO(STB_GLOBAL, STT_NOTYPE); + usedSymbolEntry.st_other = 0; + usedSymbolEntry.st_shndx = SHN_UNDEF; + + this.strings.insertBack(name[]); + this.strings.insertBack("\0"); + this.symbolTable[name] = Relocation(usedSymbolEntry); + } + + void relocate(String name, Rela[] usedSymbols...) @nogc + { + foreach (usedSymbol; usedSymbols) + { + Rela relocationEntry = usedSymbol; + + relocationEntry.r_info = usedSymbol.r_info; + this.symbolTable[name].relocations.insertBack(relocationEntry); + } + } + + private ptrdiff_t findHeader(HeaderName position)() + { + return countUntil!(header => header.sh_name == position)(this.sectionHeaders[]); + } + + private void makeTextHeader() @nogc + { + Shdr textHeader; + + textHeader.sh_name = HeaderName.text; + textHeader.sh_type = SHT_PROGBITS; + textHeader.sh_flags = SHF_EXECINSTR | SHF_ALLOC; + textHeader.sh_addr = 0; + textHeader.sh_offset = this.currentOffset; + textHeader.sh_size = 0; + textHeader.sh_link = SHN_UNDEF; + textHeader.sh_info = 0; + textHeader.sh_addralign = 1; + textHeader.sh_entsize = 0; + + this.sectionHeaders.insertBack(textHeader); + } + + private void initializeSectionHeaders() @nogc + { + Shdr table; + + table.sh_name = 0; + table.sh_type = SHT_NULL; + table.sh_flags = 0; + table.sh_addr = 0; + table.sh_offset = 0; + table.sh_size = 0; + table.sh_link = SHN_UNDEF; + table.sh_info = 0; + table.sh_addralign = 0; + table.sh_entsize = 0; + + this.sectionHeaders.insertBack(table); + } + + private void writeFileHeader() @nogc + { + Ehdr fileHeader; + auto headerStringIndex = findHeader!(HeaderName.headerString)(); + + assert(headerStringIndex != -1); + + // Magic number. + fileHeader.e_ident[0] = '\x7f'; + fileHeader.e_ident[1] = 'E'; + fileHeader.e_ident[2] = 'L'; + fileHeader.e_ident[3] = 'F'; + + fileHeader.e_ident[4] = ELFCLASS32; + fileHeader.e_ident[5] = ELFDATA2LSB; + fileHeader.e_ident[6] = EV_CURRENT; + fileHeader.e_ident[7] = EI_OSABI.ELFOSABI_SYSV; + fileHeader.e_ident[8] = 0; + + fileHeader.e_type = ET_REL; + fileHeader.e_machine = 0xf3; // EM_RISCV + fileHeader.e_version = EV_CURRENT; + fileHeader.e_entry = 0; + fileHeader.e_phoff = 0; + fileHeader.e_shoff = this.currentOffset; + fileHeader.e_flags = 0; + fileHeader.e_ehsize = Elf32_Ehdr.sizeof; + fileHeader.e_phentsize = 0; + fileHeader.e_phnum = 0; + fileHeader.e_shentsize = Elf32_Shdr.sizeof; + fileHeader.e_shnum = cast(Elf32_Half) this.sectionHeaders.length; + + // String table is the last one + fileHeader.e_shstrndx = cast(Half) headerStringIndex; + + output.seek(0, File.Whence.set); + output.write((cast(ubyte*) &fileHeader)[0 .. fileHeader.sizeof]); + } + + private void makeRoDataHeader() @nogc + { + Shdr table; + + table.sh_name = HeaderName.roData; + table.sh_type = SHT_PROGBITS; + table.sh_flags = SHF_ALLOC; + table.sh_addr = 0; + table.sh_offset = 0; + table.sh_size = 0; + table.sh_link = SHN_UNDEF; + table.sh_info = 0; + table.sh_addralign = 4; + table.sh_entsize = 0; + + this.sectionHeaders.insertBack(table); + } + + private void writeRoDataTable() @nogc + { + auto index = findHeader!(HeaderName.roData)(); + assert(index != -1); + + this.sectionHeaders[index].sh_offset = this.currentOffset; + this.sectionHeaders[index].sh_size = cast(Xword) this.readOnly.length; + + output.write(this.readOnly.get); + this.currentOffset += this.readOnly.length; + } + + private void makeRelaHeader() @nogc + { + Shdr table; + + table.sh_name = HeaderName.rela; + table.sh_type = SHT_RELA; + table.sh_flags = 0; + table.sh_addr = 0; + table.sh_offset = 0; + table.sh_size = 0; + table.sh_link = SHN_UNDEF; + table.sh_info = 0; + table.sh_addralign = 4; + table.sh_entsize = Rela.sizeof; + + this.sectionHeaders.insertBack(table); + } +} diff --git a/source/elna/extended.d b/source/elna/extended.d new file mode 100644 index 0000000..ff06a0c --- /dev/null +++ b/source/elna/extended.d @@ -0,0 +1,335 @@ +/** + * File I/O that can be moved into more generic library when and if finished. + */ +module elna.extended; + +import core.stdc.errno; +import core.stdc.stdio; +import std.sumtype; +import std.typecons; +import tanya.os.error; +import tanya.container.array; +import tanya.container.string; + +/** + * File handle abstraction. + */ +struct File +{ + /// Plattform dependent file type. + alias Handle = FILE*; + + /// Uninitialized file handle value. + enum Handle invalid = null; + + /** + * Relative position. + */ + enum Whence + { + /// Relative to the start of the file. + set = SEEK_SET, + /// Relative to the current cursor position. + currentt = SEEK_CUR, + /// Relative from the end of the file. + end = SEEK_END, + } + + /** + * File open modes. + */ + enum Mode + { + /// Open the file for reading. + read = 1 << 0, + /// Open the file for writing. The stream is positioned at the beginning + /// of the file. + write = 1 << 1, + /// Open the file for writing and remove its contents. + truncate = 1 << 2, + /// Open the file for writing. The stream is positioned at the end of + /// the file. + append = 1 << 3, + } + + private enum Status + { + invalid, + owned, + borrowed, + } + + private union Storage + { + Handle handle; + ErrorCode errorCode; + } + private Storage storage; + private Status status = Status.invalid; + + @disable this(scope return ref File f); + @disable this(); + + /** + * Closes the file. + */ + ~this() @nogc nothrow + { + if (this.status == Status.owned) + { + fclose(this.storage.handle); + } + this.storage.handle = invalid; + this.status = Status.invalid; + } + + /** + * Construct the object with the given system handle. The won't be claused + * in the descructor if this constructor is used. + * + * Params: + * handle = File handle to be wrapped by this structure. + */ + this(Handle handle) @nogc nothrow pure @safe + { + this.storage.handle = handle; + this.status = Status.borrowed; + } + + /** + * Returns: Plattform dependent file handle. + */ + @property Handle handle() @nogc nothrow pure @trusted + { + return valid ? this.storage.handle : invalid; + } + + /** + * Returns: An error code if an error has occurred. + */ + @property ErrorCode errorCode() @nogc nothrow pure @safe + { + return valid ? ErrorCode() : this.storage.errorCode; + } + + /** + * Returns: Whether a valid, opened file is represented. + */ + @property bool valid() @nogc nothrow pure @safe + { + return this.status != Status.invalid; + } + + /** + * Transfers the file into invalid state. + * + * Returns: The old file handle. + */ + Handle reset() @nogc nothrow pure @safe + { + if (!valid) + { + return invalid; + } + auto oldHandle = handle; + + this.status = Status.invalid; + this.storage.errorCode = ErrorCode(); + + return oldHandle; + } + + /** + * Sets stream position in the file. + * + * Params: + * offset = File offset. + * whence = File position to add the offset to. + * + * Returns: Error code if any. + */ + ErrorCode seek(size_t offset, Whence whence) @nogc nothrow + { + if (!valid) + { + return ErrorCode(ErrorCode.ErrorNo.badDescriptor); + } + if (fseek(this.storage.handle, offset, whence)) + { + return ErrorCode(cast(ErrorCode.ErrorNo) errno); + } + return ErrorCode(); + } + + /** + * Returns: Current offset or an error. + */ + SumType!(ErrorCode, size_t) tell() @nogc nothrow + { + if (!valid) + { + return typeof(return)(ErrorCode(ErrorCode.ErrorNo.badDescriptor)); + } + auto result = ftell(this.storage.handle); + + if (result < 0) + { + return typeof(return)(ErrorCode(cast(ErrorCode.ErrorNo) errno)); + } + return typeof(return)(cast(size_t) result); + } + + /** + * Params: + * buffer = Destination buffer. + * + * Returns: Bytes read. $(D_PSYMBOL ErrorCode.ErrorNo.success) means that + * while reading the file an unknown error has occurred. + */ + SumType!(ErrorCode, size_t) read(ubyte[] buffer) @nogc nothrow + { + if (!valid) + { + return typeof(return)(ErrorCode(ErrorCode.ErrorNo.badDescriptor)); + } + const bytesRead = fread(buffer.ptr, 1, buffer.length, this.storage.handle); + if (bytesRead == buffer.length || eof()) + { + return typeof(return)(bytesRead); + } + return typeof(return)(ErrorCode()); + } + + /** + * Params: + * buffer = Source buffer. + * + * Returns: Bytes written. $(D_PSYMBOL ErrorCode.ErrorNo.success) means that + * while reading the file an unknown error has occurred. + */ + SumType!(ErrorCode, size_t) write(const(ubyte)[] buffer) @nogc nothrow + { + if (!valid) + { + return typeof(return)(ErrorCode(ErrorCode.ErrorNo.badDescriptor)); + } + const bytesWritten = fwrite(buffer.ptr, 1, buffer.length, this.storage.handle); + if (bytesWritten == buffer.length) + { + return typeof(return)(buffer.length); + } + return typeof(return)(ErrorCode()); + } + + /** + * Returns: EOF status of the file. + */ + bool eof() @nogc nothrow + { + return valid && feof(this.storage.handle) != 0; + } + + /** + * Constructs a file object that will be closed in the destructor. + * + * Params: + * filename = The file to open. + * + * Returns: Opened file or an error. + */ + static File open(const(char)* filename, BitFlags!Mode mode) @nogc nothrow + { + char[3] modeBuffer = "\0\0\0"; + + if (mode.truncate) + { + modeBuffer[0] = 'w'; + if (mode.read) + { + modeBuffer[1] = '+'; + } + } + else if (mode.append) + { + modeBuffer[0] = 'a'; + if (mode.read) + { + modeBuffer[1] = '+'; + } + } + else if (mode.read) + { + modeBuffer[0] = 'r'; + if (mode.write) + { + modeBuffer[1] = '+'; + } + } + + auto newHandle = fopen(filename, modeBuffer.ptr); + auto newFile = File(newHandle); + + if (newHandle is null) + { + newFile.status = Status.invalid; + newFile.storage.errorCode = ErrorCode(cast(ErrorCode.ErrorNo) errno); + } + else + { + if (mode == BitFlags!Mode(Mode.write)) + { + rewind(newHandle); + } + newFile.status = Status.owned; + } + + return newFile; + } +} + +/** + * Reads the whole file and returns its contents. + * + * Params: + * sourceFilename = Source filename. + * + * Returns: File contents or an error. + * + * See_Also: $(D_PSYMBOL File.read) + */ +SumType!(ErrorCode, Array!ubyte) readFile(String sourceFilename) @nogc +{ + enum size_t bufferSize = 255; + auto sourceFile = File.open(sourceFilename.toStringz, BitFlags!(File.Mode)(File.Mode.read)); + + if (!sourceFile.valid) + { + return typeof(return)(sourceFile.errorCode); + } + Array!ubyte sourceText; + size_t totalRead; + size_t bytesRead; + do + { + sourceText.length = sourceText.length + bufferSize; + const readStatus = sourceFile + .read(sourceText[totalRead .. $].get) + .match!( + (ErrorCode errorCode) => nullable(errorCode), + (size_t bytesRead_) { + bytesRead = bytesRead_; + return Nullable!ErrorCode(); + } + ); + if (!readStatus.isNull) + { + return typeof(return)(readStatus.get); + } + totalRead += bytesRead; + } + while (bytesRead == bufferSize); + + sourceText.length = totalRead; + + return typeof(return)(sourceText); +} diff --git a/source/elna/ir.d b/source/elna/ir.d new file mode 100644 index 0000000..41d834c --- /dev/null +++ b/source/elna/ir.d @@ -0,0 +1,268 @@ +module elna.ir; + +import parser = elna.parser; +import tanya.container.array; +import tanya.container.hashtable; +import tanya.container.string; +import tanya.memory.allocator; +public import elna.parser : BinaryOperator; + +/** + * Mapping between the parser and IR AST. + */ +struct ASTMapping +{ + alias Node = .Node; + alias Definition = .Definition; + alias VariableDeclaration = .VariableDeclaration; + alias Statement = .Operand; + alias BangStatement = .Operand; + alias Block = .Definition; + alias Expression = .Operand; + alias Number = .Number; + alias Variable = .Number; + alias BinaryExpression = .Variable; + +} + +/** + * IR visitor. +*/ +interface IRVisitor +{ + abstract void visit(Node) @nogc; + abstract void visit(Definition) @nogc; + abstract void visit(Operand) @nogc; + abstract void visit(BinaryExpression) @nogc; + abstract void visit(Variable) @nogc; + abstract void visit(VariableDeclaration) @nogc; + abstract void visit(Number) @nogc; +} + +/** + * AST node. + */ +abstract class Node +{ + abstract void accept(IRVisitor) @nogc; +} + +/** + * Definition. + */ +class Definition : Node +{ + char[] identifier; + Array!BinaryExpression statements; + Array!VariableDeclaration variableDeclarations; + Operand result; + + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +abstract class Statement : Node +{ +} + +abstract class Operand : Node +{ + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +class Number : Operand +{ + int value; + + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +class Variable : Operand +{ + size_t counter; + + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +class VariableDeclaration : Node +{ + String identifier; + + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +class BinaryExpression : Statement +{ + Operand lhs, rhs; + BinaryOperator operator; + + this(Operand lhs, Operand rhs, BinaryOperator operator) + @nogc + { + this.lhs = lhs; + this.rhs = rhs; + this.operator = operator; + } + + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +class BangExpression : Statement +{ + Operand operand; + + this(Operand operand) + { + this.operand = operand; + } + + override void accept(IRVisitor visitor) @nogc + { + visitor.visit(this); + } +} + +final class TransformVisitor : parser.ParserVisitor!ASTMapping +{ + private HashTable!(String, int) constants; + private Array!BinaryExpression statements; + + ASTMapping.Node visit(parser.Node node) @nogc + { + assert(false, "Not implemented"); + } + + ASTMapping.Definition visit(parser.Definition definition) @nogc + { + assert(false, "Not implemented"); + } + + ASTMapping.VariableDeclaration visit(parser.VariableDeclaration declaration) @nogc + { + assert(false, "Not implemented"); + } + + ASTMapping.BangStatement visit(parser.BangStatement statement) @nogc + { + return statement.expression.accept(this); + } + + ASTMapping.Block visit(parser.Block block) @nogc + { + auto target = defaultAllocator.make!Definition; + this.constants = transformConstants(block.definitions); + + target.result = block.statement.accept(this); + target.statements = this.statements; + + target.variableDeclarations = transformVariableDeclarations(block.variableDeclarations); + + return target; + } + + ASTMapping.Expression visit(parser.Expression expression) @nogc + { + if ((cast(parser.Number) expression) !is null) + { + return (cast(parser.Number) expression).accept(this); + } + if ((cast(parser.Variable) expression) !is null) + { + return (cast(parser.Variable) expression).accept(this); + } + else if ((cast(parser.BinaryExpression) expression) !is null) + { + return (cast(parser.BinaryExpression) expression).accept(this); + } + assert(false, "Invalid expression type"); + } + + ASTMapping.Number visit(parser.Number number) @nogc + { + auto numberExpression = defaultAllocator.make!Number; + numberExpression.value = number.value; + + return numberExpression; + } + + ASTMapping.Variable visit(parser.Variable variable) @nogc + { + auto numberExpression = defaultAllocator.make!Number; + numberExpression.value = this.constants[variable.identifier]; + + return numberExpression; + } + + ASTMapping.BinaryExpression visit(parser.BinaryExpression binaryExpression) @nogc + { + auto target = defaultAllocator.make!BinaryExpression( + binaryExpression.lhs.accept(this), + binaryExpression.rhs.accept(this), + binaryExpression.operator + ); + statements.insertBack(target); + + auto newVariable = defaultAllocator.make!Variable; + newVariable.counter = statements.length; + + return newVariable; + } + + private Number transformNumber(parser.Number number) @nogc + { + return defaultAllocator.make!Number(number.value); + } + + override Operand visit(parser.Statement statement) @nogc + { + if ((cast(parser.BangStatement) statement) !is null) + { + return (cast(parser.BangStatement) statement).accept(this); + } + assert(false, "Invalid statement type"); + } + + private HashTable!(String, int) transformConstants(ref Array!(parser.Definition) definitions) @nogc + { + typeof(return) constants; + + foreach (definition; definitions[]) + { + constants[definition.identifier] = definition.number.value; + } + + return constants; + } + + Array!VariableDeclaration transformVariableDeclarations(ref Array!(parser.VariableDeclaration) variableDeclarations) + @nogc + { + typeof(return) variables; + + foreach (ref variableDeclaration; variableDeclarations) + { + auto newDeclaration = defaultAllocator.make!VariableDeclaration; + newDeclaration.identifier = variableDeclaration.identifier; + variables.insertBack(newDeclaration); + } + + return variables; + } +} diff --git a/source/elna/lexer.d b/source/elna/lexer.d new file mode 100644 index 0000000..83fb873 --- /dev/null +++ b/source/elna/lexer.d @@ -0,0 +1,254 @@ +module elna.lexer; + +import core.stdc.stdlib; +import core.stdc.ctype; +import core.stdc.string; +import elna.result; +import std.range; +import tanya.container.array; +import tanya.container.string; + +struct Token +{ + enum Type + { + number, + operator, + let, + identifier, + equals, + var, + semicolon, + leftParen, + rightParen, + bang, + dot, + comma, + } + + union Value + { + int number; + String identifier; + } + + private Type type; + private Value value_; + private Position position_; + + @disable this(); + + this(Type type, Position position) @nogc nothrow pure @safe + { + this.type = type; + this.position_ = position; + } + + this(Type type, int value, Position position) @nogc nothrow pure @trusted + in (type == Type.number) + { + this(type, position); + this.value_.number = value; + } + + this()(Type type, auto ref String value, Position position) + @nogc nothrow pure @trusted + in (type == Type.identifier || type == Type.operator) + { + this(type, position); + this.value_.identifier = value; + } + + /** + * Params: + * type = Expected type. + * + * Returns: Whether this token is of the expected type. + */ + bool ofType(Type type) const @nogc nothrow pure @safe + { + return this.type == type; + } + + @property auto value(Type type)() @nogc nothrow pure @trusted + in (ofType(type)) + { + static if (type == Type.number) + { + return this.value_.number; + } + else static if (type == Type.identifier || type == Type.operator) + { + return this.value_.identifier; + } + else + { + static assert(false, "This type doesn't have a value"); + } + } + + /** + * Returns: The token position in the source text. + */ + @property const(Position) position() const @nogc nothrow pure @safe + { + return this.position_; + } +} + +/** + * Range over the source text that keeps track of the current position. + */ +struct Source +{ + char[] buffer; + Position position; + + this(char[] buffer) @nogc nothrow pure @safe + { + this.buffer = buffer; + } + + @disable this(); + + bool empty() @nogc nothrow pure @safe + { + return this.length == 0; + } + + char front() @nogc nothrow pure @safe + in (!empty) + { + return this.buffer[0]; + } + + void popFront() @nogc nothrow pure @safe + in (!empty) + { + this.buffer = buffer[1 .. $]; + ++this.position.column; + } + + void breakLine() @nogc nothrow pure @safe + in (!empty) + { + this.buffer = buffer[1 .. $]; + ++this.position.line; + this.position.column = 1; + } + + @property size_t length() const @nogc nothrow pure @safe + { + return this.buffer.length; + } + + char opIndex(size_t index) @nogc nothrow pure @safe + in (index < length) + { + return this.buffer[index]; + } + + char[] opSlice(size_t i, size_t j) @nogc nothrow pure @safe + in + { + assert(i <= j); + assert(j <= length); + } + do + { + return this.buffer[i .. j]; + } +} + +Result!(Array!Token) lex(char[] buffer) @nogc +{ + Array!Token tokens; + auto source = Source(buffer); + + while (!source.empty) + { + if (source.front == ' ') + { + source.popFront; + } + else if (source.front >= '0' && source.front <= '9') // Multi-digit. + { + tokens.insertBack(Token(Token.Type.number, source.front - '0', source.position)); + source.popFront; + } + else if (source.front == '=') + { + tokens.insertBack(Token(Token.Type.equals, source.position)); + source.popFront; + } + else if (source.front == '(') + { + tokens.insertBack(Token(Token.Type.leftParen, source.position)); + source.popFront; + } + else if (source.front == ')') + { + tokens.insertBack(Token(Token.Type.rightParen, source.position)); + source.popFront; + } + else if (source.front == ';') + { + tokens.insertBack(Token(Token.Type.semicolon, source.position)); + source.popFront; + } + else if (source.front == ',') + { + tokens.insertBack(Token(Token.Type.comma, source.position)); + source.popFront; + } + else if (source.front == '!') + { + tokens.insertBack(Token(Token.Type.bang, source.position)); + source.popFront; + } + else if (source.front == '.') + { + tokens.insertBack(Token(Token.Type.dot, source.position)); + source.popFront; + } + else if (isalpha(source.front)) + { + size_t i = 1; + while (i < source.length && isalpha(source[i])) + { + ++i; + } + if (source[0 .. i] == "const") + { + tokens.insertBack(Token(Token.Type.let, source.position)); + } + else if (source[0 .. i] == "var") + { + tokens.insertBack(Token(Token.Type.var, source.position)); + } + else + { + auto identifier = String(source[0 .. i]); + tokens.insertBack(Token(Token.Type.identifier, identifier, source.position)); + } + source.popFrontN(i); + } + else if (source.front == '+' || source.front == '-') + { + String operator; + + operator.insertBack(source.front); + tokens.insertBack(Token(Token.Type.operator, operator, source.position)); + source.popFront; + } + else if (source.front == '\n') + { + source.breakLine; + } + else + { + return typeof(return)("Unexptected next character", source.position); + } + } + return typeof(return)(tokens); +} diff --git a/source/elna/parser.d b/source/elna/parser.d new file mode 100644 index 0000000..c0bf048 --- /dev/null +++ b/source/elna/parser.d @@ -0,0 +1,372 @@ +module elna.parser; + +import elna.lexer; +import elna.result; +import tanya.container.array; +import tanya.container.string; +import tanya.memory.allocator; + + +/** + * Parser visitor. + */ +interface ParserVisitor(Mapping) +{ + Mapping.Node visit(Node) @nogc; + Mapping.Definition visit(Definition) @nogc; + Mapping.VariableDeclaration visit(VariableDeclaration) @nogc; + Mapping.Statement visit(Statement) @nogc; + Mapping.BangStatement visit(BangStatement) @nogc; + Mapping.Block visit(Block) @nogc; + Mapping.Expression visit(Expression) @nogc; + Mapping.Number visit(Number) @nogc; + Mapping.Variable visit(Variable) @nogc; + Mapping.BinaryExpression visit(BinaryExpression) @nogc; +} + +/** + * AST node. + */ +abstract class Node +{ + Mapping.Node accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +/** + * Constant definition. + */ +class Definition : Node +{ + Number number; + String identifier; + + Mapping.Definition accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +/** + * Variable declaration. + */ +class VariableDeclaration : Node +{ + String identifier; + + Mapping.VariableDeclaration accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +abstract class Statement : Node +{ + Mapping.Statement accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +class BangStatement : Statement +{ + Expression expression; + + Mapping.BangStatement accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +class Block : Node +{ + Array!Definition definitions; + Array!VariableDeclaration variableDeclarations; + Statement statement; + + Mapping.Block accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +abstract class Expression : Node +{ + Mapping.Expression accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +class Number : Expression +{ + int value; + + Mapping.Number accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +class Variable : Expression +{ + String identifier; + + Mapping.Variable accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +enum BinaryOperator +{ + sum, + subtraction +} + +class BinaryExpression : Expression +{ + Expression lhs, rhs; + BinaryOperator operator; + + this(Expression lhs, Expression rhs, String operator) @nogc + { + this.lhs = lhs; + this.rhs = rhs; + if (operator == "+") + { + this.operator = BinaryOperator.sum; + } + else if (operator == "-") + { + this.operator = BinaryOperator.subtraction; + } + else + { + assert(false, "Invalid binary operator"); + } + } + + Mapping.BinaryExpression accept(Mapping)(ParserVisitor!Mapping visitor) @nogc + { + return visitor.visit(this); + } +} + +private Result!Expression parseFactor(ref Array!Token.Range tokens) @nogc +in (!tokens.empty, "Expected factor, got end of stream") +{ + if (tokens.front.ofType(Token.Type.identifier)) + { + auto variable = defaultAllocator.make!Variable; + variable.identifier = tokens.front.value!(Token.Type.identifier); + tokens.popFront; + return Result!Expression(variable); + } + else if (tokens.front.ofType(Token.Type.number)) + { + auto number = defaultAllocator.make!Number; + number.value = tokens.front.value!(Token.Type.number); + tokens.popFront; + return Result!Expression(number); + } + else if (tokens.front.ofType(Token.Type.leftParen)) + { + tokens.popFront; + + auto expression = parseExpression(tokens); + + tokens.popFront; + return expression; + } + return Result!Expression("Expected a factor", tokens.front.position); +} + +private Result!Expression parseTerm(ref Array!(Token).Range tokens) @nogc +{ + return parseFactor(tokens); +} + +private Result!Expression parseExpression(ref Array!(Token).Range tokens) @nogc +in (!tokens.empty, "Expected expression, got end of stream") +{ + auto term = parseTerm(tokens); + if (!term.valid || tokens.empty || !tokens.front.ofType(Token.Type.operator)) + { + return term; + } + auto operator = tokens.front.value!(Token.Type.operator); + tokens.popFront; + + auto expression = parseExpression(tokens); + + if (expression.valid) + { + auto binaryExpression = defaultAllocator + .make!BinaryExpression(term.result, expression.result, operator); + + return Result!Expression(binaryExpression); + } + else + { + return Result!Expression("Expected right-hand side to be an expression", tokens.front.position); + } +} + +private Result!Definition parseDefinition(ref Array!Token.Range tokens) @nogc +in (!tokens.empty, "Expected definition, got end of stream") +{ + auto definition = defaultAllocator.make!Definition; + definition.identifier = tokens.front.value!(Token.Type.identifier); // Copy. + + tokens.popFront(); + tokens.popFront(); // Skip the equals sign. + + if (tokens.front.ofType(Token.Type.number)) + { + auto number = defaultAllocator.make!Number; + number.value = tokens.front.value!(Token.Type.number); + definition.number = number; + tokens.popFront; + return Result!Definition(definition); + } + return Result!Definition("Expected a number", tokens.front.position); +} + +private Result!Statement parseStatement(ref Array!Token.Range tokens) @nogc +in (!tokens.empty, "Expected block, got end of stream") +{ + if (tokens.front.ofType(Token.Type.bang)) + { + tokens.popFront; + auto statement = defaultAllocator.make!BangStatement; + auto expression = parseExpression(tokens); + if (expression.valid) + { + statement.expression = expression.result; + } + else + { + return Result!Statement(expression.error.get); + } + return Result!Statement(statement); + } + return Result!Statement("Expected ! statement", tokens.front.position); +} + +private Result!(Array!Definition) parseDefinitions(ref Array!Token.Range tokens) @nogc +in (!tokens.empty, "Expected definition, got end of stream") +{ + tokens.popFront; // Skip const. + + Array!Definition definitions; + + while (!tokens.empty) + { + auto definition = parseDefinition(tokens); + if (!definition.valid) + { + return typeof(return)(definition.error.get); + } + definitions.insertBack(definition.result); + if (tokens.front.ofType(Token.Type.semicolon)) + { + break; + } + if (tokens.front.ofType(Token.Type.comma)) + { + tokens.popFront; + } + } + + return typeof(return)(definitions); +} + +private Result!(Array!VariableDeclaration) parseVariableDeclarations(ref Array!Token.Range tokens) @nogc +in (!tokens.empty, "Expected variable declarations, got end of stream") +{ + tokens.popFront; // Skip var. + + Array!VariableDeclaration variableDeclarations; + + while (!tokens.empty) + { + auto currentToken = tokens.front; + if (currentToken.ofType(Token.Type.identifier)) + { + auto variableDeclaration = defaultAllocator.make!VariableDeclaration; + variableDeclaration.identifier = currentToken.value!(Token.Type.identifier); + variableDeclarations.insertBack(variableDeclaration); + tokens.popFront; + } + else + { + return typeof(return)("Expected variable name", tokens.front.position); + } + if (tokens.empty) + { + return typeof(return)("Expected \";\" or \",\" name", currentToken.position); + } + if (tokens.front.ofType(Token.Type.semicolon)) + { + break; + } + if (tokens.front.ofType(Token.Type.comma)) + { + tokens.popFront; + } + } + + return typeof(return)(variableDeclarations); +} + +private Result!Block parseBlock(ref Array!Token.Range tokens) @nogc +in (!tokens.empty, "Expected block, got end of stream") +{ + auto block = defaultAllocator.make!Block; + if (tokens.front.ofType(Token.Type.let)) + { + auto constDefinitions = parseDefinitions(tokens); + if (constDefinitions.valid) + { + block.definitions = constDefinitions.result; + } + else + { + return Result!Block(constDefinitions.error.get); + } + tokens.popFront; + } + if (tokens.front.ofType(Token.Type.var)) + { + auto variableDeclarations = parseVariableDeclarations(tokens); + if (variableDeclarations.valid) + { + block.variableDeclarations = variableDeclarations.result; + } + else + { + return Result!Block(variableDeclarations.error.get); + } + tokens.popFront; + } + auto statement = parseStatement(tokens); + if (statement.valid) + { + block.statement = statement.result; + } + else + { + return Result!Block(statement.error.get); + } + + return Result!Block(block); +} + +Result!Block parse(ref Array!Token tokenStream) @nogc +{ + auto tokens = tokenStream[]; + return parseBlock(tokens); +} diff --git a/source/elna/result.d b/source/elna/result.d new file mode 100644 index 0000000..9427147 --- /dev/null +++ b/source/elna/result.d @@ -0,0 +1,107 @@ +module elna.result; + +import std.typecons; +import tanya.container.array; +import tanya.container.string; + +/** + * Position in the source text. + */ +struct Position +{ + /// Line. + size_t line = 1; + + /// Column. + size_t column = 1; +} + +struct CompileError +{ + private string message_; + + private Position position_; + + @disable this(); + + /** + * Params: + * message = Error text. + * position = Error position in the source text. + */ + this(string message, Position position) @nogc nothrow pure @safe + { + this.message_ = message; + this.position_ = position; + } + + /// Error text. + @property string message() const @nogc nothrow pure @safe + { + return this.message_; + } + + /// Error line in the source text. + @property size_t line() const @nogc nothrow pure @safe + { + return this.position_.line; + } + + /// Error column in the source text. + @property size_t column() const @nogc nothrow pure @safe + { + return this.position_.column; + } +} + +struct Result(T) +{ + Nullable!CompileError error; + T result; + + this(T result) + { + this.result = result; + this.error = typeof(this.error).init; + } + + this(string message, Position position) + { + this.result = T.init; + this.error = CompileError(message, position); + } + + this(CompileError compileError) + { + this.result = null; + this.error = compileError; + } + + @disable this(); + + @property bool valid() const + { + return error.isNull; + } +} + +struct Reference +{ + enum Target + { + text, + high20, + lower12i + } + + String name; + size_t offset; + Target target; +} + +struct Symbol +{ + String name; + Array!ubyte text; + Array!Reference symbols; +} diff --git a/source/elna/riscv.d b/source/elna/riscv.d new file mode 100644 index 0000000..bf3f2a5 --- /dev/null +++ b/source/elna/riscv.d @@ -0,0 +1,363 @@ +module elna.riscv; + +import core.stdc.stdlib; +import elna.extended; +import elna.ir; +import elna.result; +import std.algorithm; +import std.typecons; +import tanya.container.array; +import tanya.container.string; + +enum XRegister : ubyte +{ + zero = 0, + ra = 1, + sp = 2, + gp = 3, + tp = 4, + t0 = 5, + t1 = 6, + t2 = 7, + s0 = 8, + s1 = 9, + a0 = 10, + a1 = 11, + a2 = 12, + a3 = 13, + a4 = 14, + a5 = 15, + a6 = 16, + a7 = 17, + s2 = 18, + s3 = 19, + s4 = 20, + s5 = 21, + s6 = 22, + s7 = 23, + s8 = 24, + s9 = 25, + s10 = 26, + s11 = 27, + t3 = 28, + t4 = 29, + t5 = 30, + t6 = 31, +} + +enum Funct3 : ubyte +{ + addi = 0b000, + slti = 0b001, + sltiu = 0b011, + andi = 0b111, + ori = 0b110, + xori = 0b100, + slli = 0b000, + srli = 0b101, + srai = 0b101, + add = 0b000, + slt = 0b010, + sltu = 0b011, + and = 0b111, + or = 0b110, + xor = 0b100, + sll = 0b001, + srl = 0b101, + sub = 0b000, + sra = 0b101, + beq = 0b000, + bne = 0b001, + blt = 0b100, + bltu = 0b110, + bge = 0b101, + bgeu = 0b111, + fence = 0b000, + fenceI = 0b001, + csrrw = 0b001, + csrrs = 0b010, + csrrc = 0b011, + csrrwi = 0b101, + csrrsi = 0b110, + csrrci = 0b111, + priv = 0b000, + sb = 0b000, + sh = 0b001, + sw = 0b010, + lb = 0b000, + lh = 0b001, + lw = 0b010, + lbu = 0b100, + lhu = 0b101, + jalr = 0b000, +} + +enum Funct12 : ubyte +{ + ecall = 0b000000000000, + ebreak = 0b000000000001, +} + +enum Funct7 : ubyte +{ + none = 0, + sub = 0b0100000 +} + +enum BaseOpcode : ubyte +{ + opImm = 0b0010011, + lui = 0b0110111, + auipc = 0b0010111, + op = 0b0110011, + jal = 0b1101111, + jalr = 0b1100111, + branch = 0b1100011, + load = 0b0000011, + store = 0b0100011, + miscMem = 0b0001111, + system = 0b1110011, +} + +struct Instruction +{ + private uint instruction; + + this(BaseOpcode opcode) @nogc + { + this.instruction = opcode; + } + @disable this(); + + ref Instruction i(XRegister rd, Funct3 funct3, XRegister rs1, uint immediate) + return scope @nogc + { + this.instruction |= (rd << 7) + | (funct3 << 12) + | (rs1 << 15) + | (immediate << 20); + + return this; + } + + ref Instruction s(uint imm1, Funct3 funct3, XRegister rs1, XRegister rs2) + return scope @nogc + { + this.instruction |= ((imm1 & 0b11111) << 7) + | (funct3 << 12) + | (rs1 << 15) + | (rs2 << 20) + | ((imm1 & 0b111111100000) << 20); + + return this; + } + + ref Instruction r(XRegister rd, Funct3 funct3, XRegister rs1, XRegister rs2, Funct7 funct7 = Funct7.none) + return scope @nogc + { + this.instruction |= (rd << 7) + | (funct3 << 12) + | (rs1 << 15) + | (rs2 << 20) + | (funct7 << 25); + + return this; + } + + ref Instruction u(XRegister rd, uint imm) + return scope @nogc + { + this.instruction |= (rd << 7) | (imm << 12); + + return this; + } + + ubyte* encode() return scope @nogc + { + return cast(ubyte*) (&this.instruction); + } +} + +class RiscVVisitor : IRVisitor +{ + Array!Instruction instructions; + bool registerInUse; + uint variableCounter = 1; + Array!Reference references; + + override void visit(Node) @nogc + { + } + + override void visit(Definition definition) @nogc + { + const uint stackSize = cast(uint) (definition.statements.length * 4 + 12); + + // Prologue. + this.instructions.insertBack( + Instruction(BaseOpcode.opImm) + .i(XRegister.sp, Funct3.addi, XRegister.sp, -stackSize) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.store) + .s(stackSize - 4, Funct3.sw, XRegister.sp, XRegister.s0) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.store) + .s(stackSize - 8, Funct3.sw, XRegister.sp, XRegister.ra) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.opImm) + .i(XRegister.s0, Funct3.addi, XRegister.sp, stackSize) + ); + + foreach (statement; definition.statements[]) + { + statement.accept(this); + } + foreach (variableDeclaration; definition.variableDeclarations[]) + { + variableDeclaration.accept(this); + } + this.registerInUse = true; + definition.result.accept(this); + this.registerInUse = false; + + // Print the result. + this.instructions.insertBack( + Instruction(BaseOpcode.opImm) + .i(XRegister.a1, Funct3.addi, XRegister.a0, 0) + ); + this.references.insertBack(Reference(String(".CL0"), instructions.length * 4, Reference.Target.high20)); + this.instructions.insertBack( + Instruction(BaseOpcode.lui).u(XRegister.a5, 0) + ); + this.references.insertBack(Reference(String(".CL0"), instructions.length * 4, Reference.Target.lower12i)); + this.instructions.insertBack( + Instruction(BaseOpcode.opImm).i(XRegister.a0, Funct3.addi, XRegister.a5, 0) + ); + this.references.insertBack(Reference(String("printf"), instructions.length * 4, Reference.Target.text)); + this.instructions.insertBack( + Instruction(BaseOpcode.auipc).u(XRegister.ra, 0) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.jalr) + .i(XRegister.ra, Funct3.jalr, XRegister.ra, 0) + ); + // Set the return value (0). + this.instructions.insertBack( + Instruction(BaseOpcode.op) + .r(XRegister.a0, Funct3.and, XRegister.zero, XRegister.zero) + ); + + // Epilogue. + this.instructions.insertBack( + Instruction(BaseOpcode.load) + .i(XRegister.s0, Funct3.lw, XRegister.sp, stackSize - 4) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.load) + .i(XRegister.ra, Funct3.lw, XRegister.sp, stackSize - 8) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.opImm) + .i(XRegister.sp, Funct3.addi, XRegister.sp, stackSize) + ); + this.instructions.insertBack( + Instruction(BaseOpcode.jalr) + .i(XRegister.zero, Funct3.jalr, XRegister.ra, 0) + ); + } + + override void visit(Operand operand) @nogc + { + if ((cast(Variable) operand) !is null) + { + return (cast(Variable) operand).accept(this); + } + if ((cast(Number) operand) !is null) + { + return (cast(Number) operand).accept(this); + } + } + + override void visit(Variable variable) @nogc + { + const freeRegister = this.registerInUse ? XRegister.a0 : XRegister.t0; + + // movl -x(%rbp), %eax; where x is a number. + this.instructions.insertBack( + Instruction(BaseOpcode.load) + .i(freeRegister, Funct3.lw, XRegister.sp, + cast(byte) (variable.counter * 4)) + ); + } + + override void visit(VariableDeclaration) @nogc + { + } + + override void visit(Number number) @nogc + { + const freeRegister = this.registerInUse ? XRegister.a0 : XRegister.t0; + + this.instructions.insertBack( + Instruction(BaseOpcode.opImm) // movl $x, %eax; where $x is a number. + .i(freeRegister, Funct3.addi, XRegister.zero, number.value) + ); + } + + override void visit(BinaryExpression expression) @nogc + { + this.registerInUse = true; + expression.lhs.accept(this); + this.registerInUse = false; + expression.rhs.accept(this); + + // Calculate the result and assign it to a variable on the stack. + final switch (expression.operator) + { + case BinaryOperator.sum: + this.instructions.insertBack( + Instruction(BaseOpcode.op) + .r(XRegister.a0, Funct3.add, XRegister.a0, XRegister.t0) + ); + break; + case BinaryOperator.subtraction: + this.instructions.insertBack( + Instruction(BaseOpcode.op) + .r(XRegister.a0, Funct3.sub, XRegister.a0, XRegister.t0, Funct7.sub) + ); + break; + } + this.instructions.insertBack( // movl %eax, -x(%rbp); where x is a number. + Instruction(BaseOpcode.store) + .s(cast(uint) (this.variableCounter * 4), Funct3.sw, XRegister.sp, XRegister.a0) + ); + + ++this.variableCounter; + } +} + +Symbol writeNext(Definition ast) @nogc +{ + Array!Instruction instructions; + Array!Reference references; + auto visitor = cast(RiscVVisitor) malloc(__traits(classInstanceSize, RiscVVisitor)); + (cast(void*) visitor)[0 .. __traits(classInstanceSize, RiscVVisitor)] = __traits(initSymbol, RiscVVisitor)[]; + scope (exit) + { + visitor.__xdtor(); + free(cast(void*) visitor); + } + visitor.visit(ast); + + auto program = Symbol(String("main")); + + program.symbols = move(visitor.references); + foreach (ref instruction; visitor.instructions) + { + program.text.insertBack(instruction.encode[0 .. uint.sizeof]); + } + return program; +} diff --git a/shell/lexer.cpp b/source/lexer.cpp similarity index 90% rename from shell/lexer.cpp rename to source/lexer.cpp index e7ff7fc..b246ab6 100644 --- a/shell/lexer.cpp +++ b/source/lexer.cpp @@ -9,22 +9,23 @@ namespace elna source::const_iterator source::begin() const { - return source::const_iterator(std::cbegin(m_buffer), Position()); + return source::const_iterator(std::cbegin(m_buffer)); } source::const_iterator source::end() const { - Position end_position{ 0, 0 }; + source_position end_position{ 0, 0 }; return source::const_iterator(std::cend(m_buffer), end_position); } - source::const_iterator::const_iterator(std::string::const_iterator buffer, const Position& start_position) + source::const_iterator::const_iterator(std::string::const_iterator buffer, + const source_position start_position) : m_buffer(buffer), m_position(start_position) { } - const Position& source::const_iterator::position() const noexcept + const source_position& source::const_iterator::position() const noexcept { return this->m_position; } @@ -87,7 +88,7 @@ namespace elna return m_value; } - const Position& token::position() const noexcept + const source_position& token::position() const noexcept { return m_position; } diff --git a/source/main.d b/source/main.d new file mode 100644 index 0000000..dcb9fa0 --- /dev/null +++ b/source/main.d @@ -0,0 +1,33 @@ +import elna.backend; +import elna.ir; +import elna.arguments; +import std.path; +import std.sumtype; +import tanya.container.string; +import tanya.memory.allocator; +import tanya.memory.mmappool; + +int main(string[] args) +{ + defaultAllocator = MmapPool.instance; + + return Arguments.parse(args).match!( + (ArgumentError argumentError) => 4, + (Arguments arguments) { + String outputFilename; + if (arguments.output is null) + { + outputFilename = arguments + .inFile + .baseName + .withExtension("o"); + } + else + { + outputFilename = String(arguments.output); + } + + return generate(arguments.inFile, outputFilename); + } + ); +} diff --git a/tests/7_member_sum.eln b/tests/7_member_sum.eln new file mode 100644 index 0000000..a54a60a --- /dev/null +++ b/tests/7_member_sum.eln @@ -0,0 +1 @@ +! 3 + 4 + 5 + 1 + 2 + 4 + 3 diff --git a/tests/const_list.eln b/tests/const_list.eln new file mode 100644 index 0000000..375627f --- /dev/null +++ b/tests/const_list.eln @@ -0,0 +1,3 @@ +const a = 1, b = 2; +! a + b +. diff --git a/tests/expectations/7_member_sum.txt b/tests/expectations/7_member_sum.txt new file mode 100644 index 0000000..2bd5a0a --- /dev/null +++ b/tests/expectations/7_member_sum.txt @@ -0,0 +1 @@ +22 diff --git a/tests/expectations/const_list.txt b/tests/expectations/const_list.txt new file mode 100644 index 0000000..00750ed --- /dev/null +++ b/tests/expectations/const_list.txt @@ -0,0 +1 @@ +3 diff --git a/tests/expectations/left_nested_sum.txt b/tests/expectations/left_nested_sum.txt new file mode 100644 index 0000000..45a4fb7 --- /dev/null +++ b/tests/expectations/left_nested_sum.txt @@ -0,0 +1 @@ +8 diff --git a/tests/expectations/print_number.txt b/tests/expectations/print_number.txt new file mode 100644 index 0000000..00750ed --- /dev/null +++ b/tests/expectations/print_number.txt @@ -0,0 +1 @@ +3 diff --git a/tests/expectations/subtraction.txt b/tests/expectations/subtraction.txt new file mode 100644 index 0000000..d00491f --- /dev/null +++ b/tests/expectations/subtraction.txt @@ -0,0 +1 @@ +1 diff --git a/tests/expectations/sum.txt b/tests/expectations/sum.txt new file mode 100644 index 0000000..45a4fb7 --- /dev/null +++ b/tests/expectations/sum.txt @@ -0,0 +1 @@ +8 diff --git a/tests/expectations/sums.txt b/tests/expectations/sums.txt new file mode 100644 index 0000000..45a4fb7 --- /dev/null +++ b/tests/expectations/sums.txt @@ -0,0 +1 @@ +8 diff --git a/tests/left_nested_sum.eln b/tests/left_nested_sum.eln new file mode 100644 index 0000000..aae2a77 --- /dev/null +++ b/tests/left_nested_sum.eln @@ -0,0 +1,2 @@ +! (3 + 4) + 1 +. diff --git a/tests/print_number.eln b/tests/print_number.eln new file mode 100644 index 0000000..51b5ca0 --- /dev/null +++ b/tests/print_number.eln @@ -0,0 +1,2 @@ +! 3 +. diff --git a/tests/runner.cpp b/tests/runner.cpp index fcb06ab..27c6052 100755 --- a/tests/runner.cpp +++ b/tests/runner.cpp @@ -35,11 +35,48 @@ namespace elna } } + static std::string in_build_directory(const std::filesystem::path& path) + { + return "build/riscv" / path; + } + + static int build_test(const std::filesystem::directory_entry& test_entry) + { + const std::filesystem::path test_filename = test_entry.path().filename(); + + std::filesystem::path test_binary = in_build_directory(test_filename); + std::filesystem::path test_object = test_binary; + test_binary.replace_extension(); + test_object.replace_extension(".o"); + + std::filesystem::remove(test_binary); + std::filesystem::remove(test_object); + + auto status = boost::process::system("./build/bin/elna", + "-o", test_object.string(), test_entry.path().string()); + if (status != 0) + { + return status; + } + status = boost::process::system( + "/opt/riscv/bin/riscv32-unknown-elf-ld", + "-o", test_binary.string(), + "-L/opt/riscv/lib/gcc/riscv32-unknown-elf/13.2.0/", + "-L/opt/riscv/riscv32-unknown-elf/lib", + "/opt/riscv/riscv32-unknown-elf/lib/crt0.o", + "/opt/riscv/lib/gcc/riscv32-unknown-elf/13.2.0/crtbegin.o", + test_object.string(), + "--start-group", "-lgcc", "-lc", "-lgloss", "--end-group", + "/opt/riscv/lib/gcc/riscv32-unknown-elf/13.2.0/crtend.o" + ); + return status; + } + static int run_test(const std::filesystem::directory_entry& test_entry) { const std::filesystem::path test_filename = test_entry.path().filename(); - std::filesystem::path test_binary = "build/riscv" / test_filename; + std::filesystem::path test_binary = in_build_directory(test_filename); test_binary.replace_extension(); std::filesystem::path expectation_path = test_entry.path().parent_path() / "expectations" / test_filename; @@ -75,7 +112,23 @@ namespace elna { continue; } - results.add_exit_code(run_test(test_entry)); + int status{ 0 }; + + try + { + status = build_test(test_entry); + } + catch (const boost::process::process_error& exception) + { + std::cout << exception.what() << std::endl; + status = 3; + continue; + } + if (status == 0) + { + status = run_test(test_entry); + } + results.add_exit_code(status); } return results; } diff --git a/tests/subtraction.eln b/tests/subtraction.eln new file mode 100644 index 0000000..00917c6 --- /dev/null +++ b/tests/subtraction.eln @@ -0,0 +1,2 @@ +! 5 - 4 +. diff --git a/tests/sum.eln b/tests/sum.eln new file mode 100644 index 0000000..6b4e720 --- /dev/null +++ b/tests/sum.eln @@ -0,0 +1,2 @@ +! 1 + 7 +. diff --git a/tests/sums.eln b/tests/sums.eln new file mode 100644 index 0000000..59586e9 --- /dev/null +++ b/tests/sums.eln @@ -0,0 +1,2 @@ +! 1 + (3 + 4) +.