Merge pull request #27 from katsyoshi/implement-linker

implement a linker

Author: katsyoshi

.gitignore

@@ -4,6 +4,7 @@
 /coverage/
 /doc/
 /pkg/
+/AGENTS.md
 /spec/reports/
 /tmp/
 /Gemfile.lock
@@ -13,3 +14,5 @@
 /.gem_rbs_collection
 /node_modules
 /package*.json
+/*.md
+!/README.md

Gemfile

@@ -7,5 +7,6 @@ gemspec
 
 gem "rake", "~> 13.0"
 gem "fiddle"
+gem "irb"
 gem "steep"
 gem "test-unit"

lib/caotral/linker.rb

@@ -1,39 +1,51 @@
 # frozen_string_literal: true
-class Caotral::Linker
-  def self.link!(input:, output: "a.out", linker: "mold", debug: false, shared: false) = new(input:, output:, linker:, debug:, shared:).link
+require_relative "linker/reader"
+require_relative "linker/writer"
 
-  def initialize(input:, output: "a.out", linker: "mold", linker_options: [], shared: false, debug: false)
-    @input, @output, @linker = input, output, linker
-    @options = linker_options
-    @debug, @shared = debug, shared
-  end
+module Caotral
+  class Linker
+    def self.link!(input:, output: "a.out", linker: "mold", debug: false, shared: false) = new(input:, output:, linker:, debug:, shared:).link
 
-  def link(input: @input, output: @output, debug: @debug, shared: @shared) = IO.popen(link_command).close
-
-  def link_command(input: @input, output: @output, debug: @debug, shared: @shared)
-    ld_path = []
-    if @shared
-      ld_path << "--shared"
-      ld_path << "#{libpath}/crti.o"
-      ld_path << "#{gcc_libpath}/crtbeginS.o"
-      ld_path << "#{gcc_libpath}/crtendS.o"
-    else
-      ld_path << "-dynamic-linker"
-      ld_path << "/lib64/ld-linux-x86-64.so.2"
-      ld_path << "#{libpath}/crt1.o"
-      ld_path << "#{libpath}/crti.o"
-      ld_path << "#{gcc_libpath}/crtbegin.o"
-      # for not static compile
-      ld_path << "#{gcc_libpath}/crtend.o"
+    def initialize(input:, output: "a.out", linker: "mold", linker_options: [], shared: false, debug: false)
+      @input, @output, @linker = input, output, linker
+      @options = linker_options
+      @debug, @shared = debug, shared
     end
 
-    ld_path << "#{libpath}/libc.so"
-    ld_path << "#{libpath}/crtn.o"
-    cmd = [@linker, "-o", @output, "-m", "elf_x86_64", *@options, *ld_path, @input].join(' ')
-    puts cmd if @debug
-    cmd
-  end
+    def link(input: @input, output: @output, debug: @debug, shared: @shared) = IO.popen(link_command).close
+
+    def link_command(input: @input, output: @output, debug: @debug, shared: @shared)
+      ld_path = []
+      return to_elf(input:, output:, debug:) if @linker == "self"
 
-  def libpath = @libpath ||= File.dirname(Dir.glob("/usr/lib*/**/crti.o").last)
-  def gcc_libpath = @gcc_libpath ||= File.dirname(Dir.glob("/usr/lib/gcc/x86_64-*/*/crtbegin.o").last)
+      if @shared
+        ld_path << "--shared"
+        ld_path << "#{libpath}/crti.o"
+        ld_path << "#{gcc_libpath}/crtbeginS.o"
+        ld_path << "#{gcc_libpath}/crtendS.o"
+      else
+        ld_path << "-dynamic-linker"
+        ld_path << "/lib64/ld-linux-x86-64.so.2"
+        ld_path << "#{libpath}/crt1.o"
+        ld_path << "#{libpath}/crti.o"
+        ld_path << "#{gcc_libpath}/crtbegin.o"
+        # for not static compile
+        ld_path << "#{gcc_libpath}/crtend.o"
+      end
+
+      ld_path << "#{libpath}/libc.so"
+      ld_path << "#{libpath}/crtn.o"
+      cmd = [@linker, "-o", @output, "-m", "elf_x86_64", *@options, *ld_path, @input].join(' ')
+      puts cmd if @debug
+      cmd
+    end
+
+    def libpath = @libpath ||= File.dirname(Dir.glob("/usr/lib*/**/crti.o").last)
+    def gcc_libpath = @gcc_libpath ||= File.dirname(Dir.glob("/usr/lib/gcc/x86_64-*/*/crtbegin.o").last)
+
+    def to_elf(input: @input, output: @output, debug: @debug)
+      elf_obj = Caotral::Linker::Reader.new(input:, debug:).read
+      Caotral::Linker::Writer.new(elf_obj:, output:, debug:).write
+    end
+  end
 end

lib/caotral/linker/elf.rb

@@ -0,0 +1,14 @@
+require_relative "elf/header"
+require_relative "elf/sections"
+
+module Caotral
+  class Linker
+    class ELF
+      attr_reader :sections, :header
+      def initialize
+        @sections = Caotral::Linker::ELF::Sections.new
+        @header = Caotral::Linker::ELF::Header.new
+      end
+    end
+  end
+end

lib/caotral/linker/elf/header.rb

@@ -0,0 +1,97 @@
+module Caotral
+  class Linker
+    class ELF
+      class Header
+        include Caotral::Assembler::ELF::Utils
+        attr_reader :entry, :phoffset, :shoffset, :shnum, :shstrndx
+        IDENT = [0x7f, 0x45, 0x4c, 0x46, 0x02, 0x01, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00].freeze
+        IDENT_STR = IDENT.pack("C*").freeze
+        ELF_FILE_TYPE = { NONE: 0, REL: 1, EXEC: 2, DYN: 3, CORE: 4 }.freeze
+
+        def initialize(endian: :little, 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!(type: nil, entry: nil, phoffset: nil, shoffset: nil, shnum: nil, shstrndx: nil, phsize: nil, phnum: nil, ehsize: nil)
+          @type = num2bytes(type, 2) if check(type, 2)
+          @entry = num2bytes(entry, 8) if check(entry, 8)
+          @phoffset = num2bytes(phoffset, 8) if check(phoffset, 8)
+          @phsize = num2bytes(phsize, 2) if check(phsize, 2)
+          @phnum = num2bytes(phnum, 2) if check(phnum, 2)
+          @ehsize = num2bytes(ehsize, 2) if check(ehsize, 2)
+          @shoffset = num2bytes(shoffset, 8) if check(shoffset, 8)
+          @shnum = num2bytes(shnum, 2) if check(shnum, 2)
+          @shstrndx = num2bytes(shstrndx, 2) if check(shstrndx, 2)
+          self
+        end
+
+        def ehsize = get(:ehsize)
+        def phsize = get(:phsize)
+        def phnum = get(:phnum)
+        def shentsize = get(:shentsize)
+        def shnum = get(:shnum)
+        def shstrndx = get(:shstrndx)
+
+        LONG_TYPES = %w[entry phoffset shoffset].freeze
+        INT_TYPES = %w[type version].freeze
+        SHORT_TYPES = %w[ehsize phsize phnum shentsize shnum shstrndx].freeze
+        CHAR_TYPES = %w[arch flags].freeze
+        private_constant :LONG_TYPES, :INT_TYPES, :SHORT_TYPES, :CHAR_TYPES
+
+        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
+
+        def get(type)
+          val = instance_variable_get(:"@#{type.to_s}").pack("C*")
+          case type.to_s.downcase
+          when *LONG_TYPES; val.unpack("Q<")
+          when *INT_TYPES; val.unpack("L<")
+          when *SHORT_TYPES; val.unpack("S<")
+          when *CHAR_TYPES; val.unpack("C<")
+          else
+            raise "not specified: #{type}"
+          end.first
+        end
+      end
+    end
+  end
+end

lib/caotral/linker/elf/program_header.rb

@@ -0,0 +1,32 @@
+module Caotral
+  class Linker
+    class ELF
+      class ProgramHeader
+        include Caotral::Assembler::ELF::Utils
+        def initialize
+          @type = num2bytes(0, 4)
+          @flags = num2bytes(0, 4)
+          @offset = num2bytes(0, 8)
+          @vaddr = num2bytes(0, 8)
+          @paddr = num2bytes(0, 8)
+          @filesz = num2bytes(0, 8)
+          @memsz = num2bytes(0, 8)
+          @align = num2bytes(0, 8)
+        end
+        def build = bytes.flatten.pack("C*")
+        def set!(type: nil, flags: nil, offset: nil, vaddr: nil, paddr: nil, filesz: nil, memsz: nil, align: nil)
+          @type = num2bytes(type, 4) if check(type, 4)
+          @flags = num2bytes(flags, 4) if check(flags, 4)
+          @offset = num2bytes(offset, 8) if check(offset, 8)
+          @vaddr = num2bytes(vaddr, 8) if check(vaddr, 8)
+          @paddr = num2bytes(paddr, 8) if check(paddr, 8)
+          @filesz = num2bytes(filesz, 8) if check(filesz, 8)
+          @memsz = num2bytes(memsz, 8) if check(memsz, 8)
+          @align = num2bytes(align, 8) if check(align, 8)
+          self
+        end
+        private def bytes = [@type, @flags, @offset, @vaddr, @paddr, @filesz, @memsz, @align]
+      end
+    end
+  end
+end

lib/caotral/linker/elf/section.rb

@@ -0,0 +1,15 @@
+class Caotral::Linker
+  class ELF
+    class Section
+      attr_accessor :header, :body, :section_name
+      def initialize(type:, section_name: nil, options: {})
+        type_string = type.to_s.capitalize
+        type_string = type_string.upcase if type_string == "Bss"
+        @section_name = (section_name.nil? ? type_string : section_name).to_s.downcase
+        @header, @body = nil, nil
+      end
+
+      def name = @section_name == "null" ? "" : "\0#{@section_name}"
+    end
+  end
+end

lib/caotral/linker/elf/section/rel.rb

@@ -0,0 +1,36 @@
+module Caotral
+  class Linker
+    class ELF
+      class Section
+        class Rel
+          include Caotral::Assembler::ELF::Utils
+          def initialize(addend: true)
+            @offset = num2bytes(0, 8)
+            @info = num2bytes(0, 8)
+            @addend = addend ? num2bytes(0, 8) : false
+          end
+
+          def set!(offset: nil, info: nil, addend: nil)
+            @offset = num2bytes(offset, 8) if check(offset, 8)
+            @info = num2bytes(info, 8) if check(info, 8)
+            @addend = num2bytes(addend, 8) if check(addend, 8)
+            self
+          end
+
+          def build = bytes.flatten.pack("C*")
+          def offset = @offset.pack("C*").unpack1("Q<")
+          def info = @info.pack("C*").unpack1("Q<")
+          def addend
+            raise "No addend field in this REL entry" unless addend?
+            @addend.pack("C*").unpack1("Q<")
+          end
+          def sym = @info.pack("C*").unpack1("Q<") >> 32
+          def type = @info.pack("C*").unpack1("Q<") & 0xffffffff
+          def addend? = !!@addend
+
+          private def bytes = addend? ? [@offset, @info, @addend] : [@offset, @info]
+        end
+      end
+    end
+  end
+end

lib/caotral/linker/elf/section/strtab.rb

@@ -0,0 +1,27 @@
+module Caotral
+  class Linker
+    class ELF
+      class Section
+        class Strtab
+          include Caotral::Assembler::ELF::Utils
+          attr_reader :names
+          def initialize(names = "\0main\0", **opts) = @names = names
+          def build = @names.bytes.pack("C*")
+          def offset_of(name)
+            offset = 0
+            @names.split("\0").each do |n|
+              return offset if n == name
+              offset += n.bytesize + 1
+            end
+            nil
+          end
+
+          def lookup(offset)
+            return "" if offset == 0
+            @names.byteslice(offset..).split("\0", 2).first
+          end
+        end
+      end
+    end
+  end
+end

lib/caotral/linker/elf/section/symtab.rb

@@ -0,0 +1,35 @@
+module Caotral
+  class Linker
+    class ELF
+      class Section
+        class Symtab
+          include Caotral::Assembler::ELF::Utils
+          attr_accessor :name_string
+          def initialize(**opts)
+            @entsize = []
+            @name = num2bytes(0, 4)
+            @info = num2bytes(0, 1)
+            @other = num2bytes(0, 1)
+            @shndx = num2bytes(0, 2)
+            @value = num2bytes(0, 8)
+            @size = num2bytes(0, 8)
+            @name_string = ""
+          end
+
+          def set!(name: nil, info: nil, other: nil, shndx: nil, value: nil, size: nil)
+            @name = num2bytes(name, 4) if check(name, 4)
+            @info = num2bytes(info, 1) if check(info, 1)
+            @other = num2bytes(other, 1) if check(other, 1)
+            @shndx = num2bytes(shndx, 2) if check(shndx, 2)
+            @value = num2bytes(value, 8) if check(value, 8)
+            @size = num2bytes(size, 8) if check(size, 8)
+            self
+          end
+
+          def name_offset = @name.pack("C*").unpack1("L<")
+          def value = @value.pack("C*").unpack1("Q<")
+        end
+      end
+    end
+  end
+end