Merge pull request #11 from katsyoshi/compile-to-elf

Compile to elf

Author: katsyoshi

.github/workflows/main.yml

@@ -14,8 +14,6 @@ jobs:
     strategy:
       matrix:
         ruby:
-          - '3.2'
-          - '3.3'
           - '3.4'
 
     steps:
@@ -25,6 +23,7 @@ jobs:
       with:
         ruby-version: ${{ matrix.ruby }}
         bundler-cache: true
+    - uses: rui314/setup-mold@v1
     - name: Install rbs
       run: bundle exec rbs collection install
     - name: Run type check

Rakefile

@@ -8,7 +8,7 @@ require "steep/cli"
 task default: %i[test]
 
 Rake::TestTask.new do |t|
-  t.test_files = FileList['test/test*.rb']
+  t.test_files = FileList['test/test*.rb', 'test/**/test*.rb']
 end
 
 namespace :steep do

exe/vaporware

@@ -4,11 +4,13 @@ require "vaporware"
 require "optparse"
 opt = OptionParser.new
 options = {}
-opt.on("-c", "--compiler [VAL]", "this option is selecting compiler precompiled file, default: gcc") { |v| options[:compiler] = v }
+opt.on("-c", "--compiler[=VAL]", "this option is selecting compiler precompiled file, default: \"self\"") { |v| options[:compiler] = v }
+opt.on("-a", "--assembler[=VAL]", "this option is selecting assembler assembler file, default: \"as\"") { |v| options[:assembler] = v }
 opt.on("-D", "--debug") { |v| options[:debug] = v }
-opt.on("-o", "--objects [VAL]") { |v| options[:dest] = v }
+opt.on("-o", "--objects[=VAL]") { |v| options[:dest] = v }
 opt.on("--compiler-options[=VAL]", "compiler options") { |v| options[:compiler_options] = v.split(",") }
-opt.on("-s", "--shared") { |v| options[:shared] = v }
+opt.on("-s", "--shared-library") { |v| options[:shared] = v }
+opt.on("-l", "--linker[=VAL]", "selecting linker: gold, lld, and mold, default: \"gold\".") { |v| options[:linker] = v }
 
 begin
   opt.parse!(ARGV)

lib/vaporware/compiler.rb

@@ -1,46 +1,26 @@
 # frozen_string_literal: true
 
 require_relative "compiler/generator"
+require_relative "compiler/assembler"
+require_relative "compiler/linker"
 
-module Vaporware
-  # Your code goes here...
-  class Compiler
-    def self.compile(source, compiler: "gcc", dest: "tmp", debug: false, compiler_options: ["-O0"], shared: false)
-      _precompile = "#{dest}.s"
-      s = new(source, _precompile: _precompile, debug:, shared:)
-      s.compile(compiler:, compiler_options:)
-    end
+class Vaporware::Compiler
+  attr_reader *%i(generator assembler linker)
 
-    def initialize(source, _precompile: "tmp.s", debug: false, shared: false)
-      @generator = Vaporware::Compiler::Generator.new(source, precompile: _precompile, debug:, shared:)
-    end
-
-    def compile(compiler: "gcc", compiler_options: ["-O0"])
-      @generator.register_var_and_method(@generator.ast)
+  def self.compile(source, assembler: "as", linker: "ld", dest: "tmp", debug: false, compiler_options: ["-O0"], shared: false)
+    compiler = new(input: source, output: dest, debug:, shared:, linker:, assembler:)
+    compiler.compile(compiler_options:)
+    compiler.assemble(input: dest.to_s + ".s", assembler:, debug:)
+    compiler.link
+  end
 
-      output = File.open(@generator.precompile, "w")
-      # prologue
-      output.puts ".intel_syntax noprefix"
-      if @generator.defined_methods.empty?
-        @generator.main = true
-        output.puts ".globl main"
-        output.puts "main:"
-        output.puts "  push rbp"
-        output.puts "  mov rbp, rsp"
-        output.puts "  sub rsp, #{@generator.defined_variables.size * 8}"
-        @generator.to_asm(@generator.ast, output)
-        # epilogue
-        @generator.epilogue(output)
-      else
-        @generator.prologue_methods(output)
-        output.puts ".globl main" unless @generator.shared
-        @generator.to_asm(@generator.ast, output)
-        # epilogue
-        @generator.epilogue(output)
-      end
-      output.close
-      compiler_options += @generator.compile_shared_option if @generator.shared
-      @generator.call_compiler(compiler:, compiler_options:)
-    end
+  def initialize(input:, output: File.basename(input, ".*"), linker: "ld", assembler: "as", debug: false, shared: false)
+    @generator = Vaporware::Compiler::Generator.new(input:, output: output + ".s", debug:, shared:)
+    @assembler = Vaporware::Compiler::Assembler.new(input: @generator.precompile, output: output + ".o", assembler:, debug:)
+    @linker = Vaporware::Compiler::Linker.new(input: @assembler.obj_file, output:, linker:, debug:, shared:)
   end
+
+  def assemble(input:, output: File.basename(input, ".*") + ".o", assembler: "as", assembler_options: [], debug: false) = @assembler.assemble(input:, output:, assembler:, assembler_options:, debug:)
+  def link = @linker.link
+  def compile(compiler_options: ["-O0"]) = @generator.compile
 end

lib/vaporware/compiler/assembler.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+require_relative "assembler/elf"
+require_relative "assembler/elf/utils"
+require_relative "assembler/elf/header"
+require_relative "assembler/elf/sections"
+require_relative "assembler/elf/section_header"
+
+class Vaporware::Compiler::Assembler
+  GCC_ASSEMBLERS = ["gcc", "as"]
+  CLANG_ASSEMBLERS = ["clang", "llvm"]
+  ASSEMBLERS = GCC_ASSEMBLERS + CLANG_ASSEMBLERS
+  class Error < StandardError; end
+
+  def self.assemble!(input, output = File.basename(input, ".*") + ".o", options = {}) = new(input:, output:, **options).assemble
+
+  def initialize(input:, output: File.basename(input, ".*") + ".o", assembler: "as", type: :relocatable, debug: false)
+    @input, @output = input, output
+    @elf = ELF.new(type:, input:, output:, debug:)
+    @assembler = assembler
+    @debug = debug
+  end
+
+  def assemble(assembler: @assembler, assembler_options: [], input: @input, output: @output, debug: false)
+    if ASSEMBLERS.include?(assembler)
+      IO.popen([command(assembler), *assembler_options, "-o", output, input].join(" ")).close
+    else
+      to_elf(input:, output:, debug:)
+    end
+
+    output
+  end
+  def obj_file = @output
+  def to_elf(input: @input, output: @output, debug: false) = @elf.build(input:, output:, debug:)
+
+  def command(asm)
+    case asm
+    when "as", "gcc"
+      gcc_assembler(asm)
+    when "clang", "llvm"
+      clang_assembler(asm)
+    else
+      raise Error, "Invalid assembler command: #{asm}"
+    end
+  end
+
+  private
+  def gcc_assembler(assembler)
+    case assembler
+    when "as", "gcc"
+      "as"
+    else
+      raise Error, "Invalid assembler command: #{assembler}"
+    end
+  end
+  
+  def clang_assembler(assembler)
+    case assembler
+    when "clang"
+      "clang"
+    when "llvm"
+      "clang -cc1as -triple x86_64-pc-linux-gnu -filetype obj -target-cpu x86-64 -mrelocation-model pic"
+    else
+      raise Error, "Invalid assembler command: #{assembler}"
+    end
+  end
+end

lib/vaporware/compiler/assembler/elf.rb

@@ -0,0 +1,78 @@
+class Vaporware::Compiler::Assembler
+  class ELF
+    class Error < StandardError; end
+    class Section; end
+    class SectionHeader; end
+    module Utils; end
+
+    def initialize(type:, input:, output:, debug:)
+      @input, @output = input, output
+      @header = Header.new(type:)
+      @sections = Sections.new
+    end
+
+    def build(input: @input, output: @output, debug: false)
+      program_size = 0
+      read!(input:)
+      init_assemble!
+
+      offset = 0x40
+      section_headers = []
+      names = []
+      bodies = {
+        null: nil,
+        text: nil,
+        data: nil,
+        bss: nil,
+        note: nil,
+        symtab: nil,
+        strtab: nil,
+        shstrtab: nil,
+      }
+      name_idx = 0
+      padding = nil
+      @sections.each do |section|
+        name = section.name
+        names << name
+        section.body.set!(name: names.join) if name == "\0.shstrtab"
+        bin = section.body.build
+        size = bin.bytesize
+        bin << "\0" until (bin.bytesize % 8) == 0 if ["\0.text", "\0.shstrtab"].include?(name)
+        bin << "\0" until ((bin.bytesize + offset) % 8) == 0 if ["\0.shstrtab"].include?(name)
+        bodies[section.section_name.to_sym] = bin
+        header = section.header
+        if offset > 0x40 && size > 0 && padding&.>(0)
+          offset += padding
+          padding = nil
+        end
+        padding = bin.size - size if size > 0
+        header.set!(name: name_idx, offset:, size:) unless name == ""
+        offset += size
+        section_headers << header.build
+        name_idx += name == "" ? 1 : name.size
+      end
+      @header.set!(shoffset: offset + padding)
+      w = File.open(output, "wb")
+      w.write([@header.build, *bodies.values, *section_headers].join)
+      w.close
+      [@header.build, *bodies.values, *section_headers]
+    end
+
+    private
+    def init_assemble! = (note!; symtab!)
+
+    def read!(input: @input, text: @sections.text.body)
+      read = { main: false }
+      File.open(input, "r") do |r|
+      r.each_line do |line|
+        read[:main] = line.match(/main:/) unless read[:main]
+        next unless read[:main] && !/main:/.match(line)
+        next if /\.L.+:/.match(line)
+        text.assemble!(line)
+      end
+      end
+    end
+    def note! = @sections.note.body.null!
+    def symtab! = @sections.symtab.body.set!(entsize: 0x18, name: 1, info: 0x10, other: 0, shndx: 1)
+  end
+end

lib/vaporware/compiler/assembler/elf/header.rb

@@ -0,0 +1,62 @@
+require_relative "../elf"
+
+class Vaporware::Compiler::Assembler::ELF::Header
+  include Vaporware::Compiler::Assembler::ELF::Utils
+  IDENT = [0x7f, 0x45, 0x4c, 0x46, 0x02, 0x01, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00].freeze
+  ELF_FILE_TYPE = { NONE: 0, REL: 1, EXEC: 2, DYN: 3, CORE: 4 }.freeze
+
+  def initialize(endian: :littel, type: :rel, arc: :amd64)
+    @ident = IDENT
+    @type = num2bytes(ELF_FILE_TYPE[elf(type)], 2)
+    @arch = arch(arc)
+    @version = num2bytes(1, 4)
+    @entry = num2bytes(0x00, 8)
+    @phoffset = num2bytes(0x00, 8)
+    @shoffset = num2bytes(0x00, 8)
+    @flags = num2bytes(0x00, 4)
+    @ehsize = num2bytes(0x40, 2)
+    @phsize = num2bytes(0x00, 2)
+    @phnum = num2bytes(0x00, 2)
+    @shentsize = num2bytes(0x40, 2)
+    @shnum = num2bytes(0x08, 2)
+    @shstrndx = num2bytes(0x07, 2)
+  end
+
+  def build = bytes.flatten.pack("C*")
+
+  def set!(entry: nil, phoffset: nil, shoffset: nil, shnum: nil, shstrndx: nil)
+    @entry = num2bytes(entry, 8) if check(entry, 8)
+    @phoffset = num2bytes(phoffset, 8) if check(phoffset, 8)
+    @shoffset = num2bytes(shoffset, 8) if check(shoffset, 8)
+    @shnum = num2bytes(shnum, 4) if check(shnum, 4)
+    @shstrndx = num2bytes(shstrndx, 4) if check(shstrndx, 4)
+  end
+
+  private
+
+  def bytes = [
+    @ident, @type, @arch, @version, @entry, @phoffset,
+    @shoffset, @flags, @ehsize, @phsize, @phnum, @shentsize,
+    @shnum, @shstrndx
+  ]
+
+  def arch(machine)
+    case machine.to_s
+    in "amd64" | "x86_64" | "x64"
+      [0x3e, 0x00]
+    end
+  end
+
+  def elf(type)
+    case type.to_s
+    in "relocatable" | "rel"
+      :REL
+    in "exe" | "ex" | "exec"
+      :EXEC
+    in "shared" | "share" | "dynamic" | "dyn"
+      :DYN
+    else
+      :NONE
+    end
+  end
+end

lib/vaporware/compiler/assembler/elf/section.rb

@@ -0,0 +1,25 @@
+require_relative "section/text"
+require_relative "section/bss"
+require_relative "section/data"
+require_relative "section/note"
+require_relative "section/null"
+require_relative "section/symtab"
+require_relative "section/strtab"
+require_relative "section/shstrtab"
+require_relative "section_header"
+
+class Vaporware::Compiler::Assembler::ELF::Section
+  attr_reader :header, :body, :name, :section_name
+  def initialize(type:, options: {})
+    type_string = type.to_s.capitalize
+    type_string = type_string.upcase if type_string == "Bss"
+    @section_name = type_string.downcase
+    @name = section_name == "null" ? "" : "\0.#{section_name}" 
+    @header = Vaporware::Compiler::Assembler::ELF::SectionHeader.new.send("#{@section_name}!")
+    @body = Module.const_get("Vaporware::Compiler::Assembler::ELF::Section::#{type_string}").new(**options)
+  end
+
+  def name=(name)
+    @name = name
+  end
+end

lib/vaporware/compiler/assembler/elf/section/bss.rb

@@ -0,0 +1,10 @@
+class Vaporware::Compiler::Assembler::ELF::Section::BSS
+  include Vaporware::Compiler::Assembler::ELF::Utils
+  def initialize(**opts) = nil
+  def build = bytes.flatten.pack("C*")
+  def set! = self
+
+  private
+  def bytes = []
+  def check(val, bytes) = false
+end

lib/vaporware/compiler/assembler/elf/section/data.rb

@@ -0,0 +1,7 @@
+class Vaporware::Compiler::Assembler::ELF::Section::Data
+  include Vaporware::Compiler::Assembler::ELF::Utils
+  def initialize(**opts) = nil
+  def build = bytes.flatten.pack("C*")
+  def set! = self
+  private def bytes = []
+end