Merge pull request #48 from katsyoshi/support-dlopen-in-ruby

Support dlopen in ruby

Author: katsyoshi

.github/workflows/main.yml

@@ -31,3 +31,5 @@ jobs:
       run: bundle exec rake steep:check
     - name: Run the default task
       run: bundle exec rake
+    - name: Run test using Ruby::Box
+      run: RUBY_BOX=1 bundle exec rake test

Rakefile

@@ -7,8 +7,13 @@ require "steep/cli"
 
 task default: %i[test]
 
+def supported_ruby_box? = RUBY_VERSION >= "4.0.0" && ENV["RUBY_BOX"] == "1" && defined?(Ruby::Box)
+
 Rake::TestTask.new do |t|
   t.test_files = FileList['test/**/*_test.rb']
+  if supported_ruby_box?
+    t.test_files = FileList['test/caotral/linker/fiddle_test.rb']
+  end
 end
 
 namespace :steep do

lib/caotral/binary/elf/reader.rb

@@ -84,9 +84,9 @@ def read
             @bin.pos = section.header.offset
             body_bin = @bin.read(section.header.size)
             section.body = case type
-                           when :strtab
+                           when :strtab, :dynstr
                              Caotral::Binary::ELF::Section::Strtab.new(body_bin)
-                           when :symtab
+                           when :symtab, :dynsym
                              symtab_entsize = section.header.entsize
                              count = body_bin.bytesize / symtab_entsize
                              count.times.map do |i|
@@ -144,6 +144,7 @@ def validate_relocations
           pt_load = @context.program_headers.find { |ph| ph.type == :LOAD }
           dynamic = @context.sections.find { |section| section.section_name.to_s == ".dynamic" }
           rela_plt = @context.sections.find { |section| section.section_name.to_s == ".rela.plt" }
+          rela_plt_exists = !rela_plt.body.empty?
           got_plt = @context.sections.find { |s| s.section_name.to_s == ".got.plt" }
           failed_messages = []
           unless rela_dyn && pt_load && dynamic
@@ -172,7 +173,7 @@ def validate_relocations
             failed_messages << "Relocation entries in .rela.dyn exceed LOAD segment range"
           end
 
-          if rela_plt
+          if rela_plt && rela_plt_exists
             jump_rel = dynamic.body.find { |dt| dt.jmp_rel? }&.un == rela_plt.header.addr
             plt_rel_size = dynamic.body.find { |dt| dt.plt_rel_size? }&.un == rela_plt.header.size
             plt_rel = dynamic.body.find { |dt| dt.plt_rel? }&.un == 7

lib/caotral/binary/elf/section/hash.rb

@@ -5,6 +5,7 @@ class ELF
       class Section
         class Hash
           include Caotral::Binary::ELF::Utils
+          attr_reader :bucket, :chain
           def initialize(nchain:, nbucket: 1)
             @nbucket = num2bytes(nbucket, 4)
             @nchain = num2bytes(nchain, 4)

lib/caotral/linker/builder.rb

@@ -6,6 +6,7 @@ class Linker
     class Builder
       include Caotral::Binary::ELF::Utils
       REL_TYPES = Caotral::Binary::ELF::Section::Rel::TYPES
+      DYNAMIC_TAGS = Caotral::Binary::ELF::Section::Dynamic::TAG_TYPES
       SYMTAB_BIND = { locals: 0, globals: 1, weaks: 2, }.freeze
       BIND_BY_VALUE = SYMTAB_BIND.invert.freeze
       RELOCATION_SECTION_NAMES = [".rela.text", ".rel.text", ".rela.data", ".rel.data"].freeze
@@ -18,6 +19,11 @@ class Builder
         REL_TYPES[:AMD64_GOTPCRELX],
         REL_TYPES[:AMD64_REX_GOTPCRELX],
       ].freeze
+      REJECT_DYNAMIC_TAGS = [
+        DYNAMIC_TAGS[:PLTRELSZ],
+        DYNAMIC_TAGS[:PLTREL],
+        DYNAMIC_TAGS[:JMPREL],
+      ].freeze
 
       attr_reader :symbols
 
@@ -176,22 +182,22 @@ def build
                 first_insertion = got_plt_offsets[sym].nil?
                 got_plt_offsets[sym] ||= got_plt_offset.tap { got_plt_offset += 8 }
                 if dynamic? && undefined && first_insertion
-                    got_plt_section.body << [0].pack("Q<")
-                    rps = Caotral::Binary::ELF::Section::Rel.new.set!(
-                      offset: got_plt_offsets[sym],
-                      info: ((sym) << 32) | REL_TYPES[:AMD64_JUMP_SLOT]
+                  got_plt_section.body << [0].pack("Q<")
+                  rps = Caotral::Binary::ELF::Section::Rel.new.set!(
+                    offset: got_plt_offsets[sym],
+                    info: ((sym) << 32) | REL_TYPES[:AMD64_JUMP_SLOT]
+                  )
+                  name = symtab_section.body[sym].name_string
+                  dynstr_index = dynstr.body.offset_of(name)
+                  if dynstr_index.nil?
+                    dynstr.body.names += name + "\0"
+                    dynsym.body << Caotral::Binary::ELF::Section::Symtab.new.set!(
+                      name: dynstr.body.offset_of(name),
+                      info: (1 << 4) | 2,
                     )
-                    name = symtab_section.body[sym].name_string
-                    dynstr_index = dynstr.body.offset_of(name)
-                    if dynstr_index.nil?
-                      dynstr.body.names += name + "\0"
-                      dynsym.body << Caotral::Binary::ELF::Section::Symtab.new.set!(
-                        name: dynstr.body.offset_of(name),
-                        info: (1 << 4) | 2,
-                      )
-                    end
-                    rela_plt_section.body << rps
-                    next
+                  end
+                  rela_plt_section.body << rps
+                  next
                 end
               elsif UNSUPPORTED_REL_TYPES.include?(rel.type)
                 raise Caotral::Binary::ELF::Error, "unsupported relocation type: #{rel.type_name}"
@@ -260,13 +266,40 @@ def build
         if dynamic?
           sections << dynstr
           sections << dynsym
-          sections << build_hash_section if @pie
+          hash_section = build_hash_section
+          sections << hash_section
           sections << rela_dyn_section
           sections << rela_plt_section
           sym = sections.index(dynsym)
           rela_dyn_section.header.set!(link: sym, type: rel_type(rela_dyn_section), entsize: rel_entsize(rela_dyn_section))
           rela_plt_section.header.set!(link: sym, type: rel_type(rela_plt_section), info: ref_index(sections, got_plt_section.section_name))
-          sections << build_dynamic_section
+          symtab_section.body.each do |sym|
+            next unless [SYMTAB_BIND[:globals], SYMTAB_BIND[:weaks]].include?(sym.bind)
+            next if sym.shndx == 0
+            copy_sym = sym.dup
+            shndx = copy_sym.shndx
+            name = dynstr.body.offset_of(sym.name_string)
+            if name.nil?
+              dynstr.body.names += copy_sym.name_string + "\0"
+              name = dynstr.body.offset_of(copy_sym.name_string)
+            end
+            copy_sym.name_string = sym.name_string
+            dynsym.body << copy_sym.set!(name:, shndx:, value: sym.value)
+          end
+          hash = Caotral::Binary::ELF::Section::Hash.new(nchain: dynsym.body.size)
+          hash.bucket[0] = num2bytes(1, 4) if dynsym.body.size > 1
+          dynsym.body.each_with_index do |sym, i|
+            next if i == 0
+            hash.chain[i] = num2bytes(0, 4)
+          end
+          hash_section.body = hash
+          dynamic_section = build_dynamic_section
+          if rela_plt_section.body.size == 0 && dynamic?
+            bodies = dynamic_section.body.reject { |ent| REJECT_DYNAMIC_TAGS.include?(ent.tag) }
+            dynamic_section.body = bodies
+          end
+
+          sections << dynamic_section
         end
         sections << symtab_section
 

lib/caotral/linker/writer.rb

@@ -82,7 +82,7 @@ def write
           rel.header.set!(offset: rel_offset, size: rel_size, entsize:)
         end
 
-        patch_dynamic_sections(file: f)
+        patch_dynamic_sections(file: f) if dynamic?
         patch_program_headers(file: f)
         write_program_headers(file: f)
 
@@ -108,6 +108,21 @@ def patch_dynamic_sections(file:)
           dyn.header.set!(addr:)
         end
 
+        cur = file.pos
+        file.seek(dynsym_section.header.offset)
+        dynsym_section.body.each do |dynsym_body|
+          if dynsym_body.shndx != 0
+            value = dynsym_body.value
+            secndx = @write_sections[dynsym_body.shndx]&.header&.addr
+            unless secndx.nil?
+              value += secndx
+              dynsym_body.set!(value:)
+            end
+          end
+          file.write(dynsym_body.build)
+        end
+        file.seek(cur)
+
         if dynamic? && dynamic_section && rela_dyn_section
           rdsh = rela_dyn_section&.header
           bodies = dynamic_section.body
@@ -238,23 +253,27 @@ def write_shared_dynamic_sections(file:)
           interp_section.header.set!(offset: interp_offset, size:, addr: text_addr + (interp_offset - tsh.offset))
         end
 
+        pad_to_align(file:, align: dynstr_section.header.addralign)
         dynstr_offset = file.pos
         file.write(dynstr_section.body.build)
         size = file.pos - dynstr_offset
         dynstr_section.header.set!(offset: dynstr_offset, size:, addr: text_addr + (dynstr_offset - tsh.offset))
 
+        pad_to_align(file:, align: dynsym_section.header.addralign)
         dynsym_offset = file.pos
         dynsym_section.body.each { |dynsym| file.write(dynsym.build) }
         size = file.pos - dynsym_offset
         dynsym_section.header.set!(offset: dynsym_offset, size:, addr: text_addr + (dynsym_offset - tsh.offset))
 
-        if @pie
+        if dynamic?
+          pad_to_align(file:, align: hash_section.header.addralign)
           hash_offset = file.pos
           file.write(hash_section.body.build)
           size = file.pos - hash_offset
           hash_section.header.set!(offset: hash_offset, size:, addr: text_addr + (hash_offset - tsh.offset))
         end
 
+        pad_to_align(file:, align: dynamic_section.header.addralign)
         dynamic_offset = file.pos
         dynamic_section.body.each { |dynamic| file.write(dynamic.build) }
         size = file.pos - dynamic_offset
@@ -316,6 +335,12 @@ def write_section_headers(file:, shoffset:)
         file.write(@elf_obj.header.build)
       end
 
+      def pad_to_align(file:, align:)
+        pos = file.pos
+        padding = (align - (pos % align)) % align
+        file.write("\0" * padding)
+      end
+
       def program_header_flags(flag) = Caotral::Binary::ELF::ProgramHeader::PF[flag.to_sym]
       def elf_type = Caotral::Binary::ELF::Header::TYPE[dynamic? ? :DYN : :EXEC]
 
@@ -343,7 +368,12 @@ def program_headers
           pph = Caotral::Binary::ELF::ProgramHeader.new
           pph.set!(type: 6)
         end
-        @program_headers = [pph, lph, iph, dph].compact
+        # ruby's dlopen support
+        if dynamic?
+          gsph = Caotral::Binary::ELF::ProgramHeader.new
+          gsph.set!(type: 0x6474e551, flags: program_header_flags(:RW))
+        end
+        @program_headers = [pph, lph, iph, dph, gsph].compact
       end
       def pie_program_header = @pie_program_header ||= program_headers.find { |ph| ph.type == :PHDR }
       def load_program_header = @load_program_header ||= program_headers.find { |ph| ph.type == :LOAD }
@@ -366,7 +396,7 @@ def plt_section = @plt_section ||= @write_sections.find { |s| ".plt" === s.secti
       def got_plt_section = @got_plt_section ||= @write_sections.find { |s| ".got.plt" === s.section_name.to_s }
       def rela_plt_section = @rela_plt_section ||= @write_sections.find { |s| ".rela.plt" === s.section_name.to_s }
 
-      def dynamic_sections = @dynamic_sections ||= [interp_section, dynstr_section, dynsym_section, dynamic_section, rela_dyn_section, rela_plt_section].compact
+      def dynamic_sections = @dynamic_sections ||= [interp_section, dynstr_section, dynsym_section, hash_section, dynamic_section, rela_dyn_section, rela_plt_section].compact
     end
   end
 end

sample/C/add.c

@@ -0,0 +1,3 @@
+int add(int x, int y) {
+  return x + y;
+}

sample/fiddle_add.rb

@@ -0,0 +1,7 @@
+require "fiddle/import"
+
+module X
+  extend Fiddle::Importer
+  dlload "./libtmp.so"
+  extern "int add(int, int)"
+end

test/caotral/linker/fiddle_test.rb

@@ -0,0 +1,40 @@
+require_relative "../../test_suite"
+
+class Caotral::Linker::FiddleMethodTest < Test::Unit::TestCase
+  include TestProcessHelper
+  def setup
+    @generated = []
+    omit("Ruby::Box is not supported in this environment") unless supported_ruby_box?
+  end
+
+  def teardown
+    @generated.each do |file|
+      File.delete(file) if File.exist?(file)
+    end
+  end
+
+  def test_sample_call_add_method
+    @generated = ["libtmp.so", "libtmp.so.o"]
+    @file = "sample/C/add.c"
+    IO.popen(["gcc", "-fPIC", "-c", "-o", "libtmp.so.o", "%s" % @file]).close
+    Caotral::Linker.link!(inputs: ["libtmp.so.o"], output: "libtmp.so", linker: "self", shared: true, executable: false)
+    elf = Caotral::Binary::ELF::Reader.read!(input: "./libtmp.so")
+    box = Ruby::Box.new
+    box.require("./sample/fiddle_add.rb")
+    assert_equal(10, box::X.add(3, 7))
+    dynsym = elf.find_by_name(".dynsym")
+    rela_plt = elf.find_by_name(".rela.plt")
+    dynamic = elf.find_by_name(".dynamic")
+    dynstr = elf.find_by_name(".dynstr")
+    dynstrs = dynstr.body.names.split("\x00")
+    assert(dynstrs.include?("add"))
+    assert_equal(2, dynsym.body.size)
+    assert_equal("add", dynstr.body.lookup(dynsym.body[1].name_offset))
+    assert_equal(0, rela_plt.body.size)
+    assert_equal(nil, dynamic.body.find { |dt| dt.plt_rel? })
+    assert_equal(nil, dynamic.body.find { |dt| dt.plt_rel_size? })
+    assert_equal(nil, dynamic.body.find { |dt| dt.jmp_rel? })
+  end
+
+  private def supported_ruby_box? = RUBY_VERSION >= "4.0.0" && ENV["RUBY_BOX"] == "1" && defined?(Ruby::Box)
+end