Fix Marshal and dumping String in a multibyte encoding

Author: andrykonchin

spec/ruby/core/marshal/dump_spec.rb

@@ -193,6 +193,11 @@ def _dump(level)
       Marshal.dump(MarshalSpec::ClassWithOverriddenName).should == "\x04\bc)MarshalSpec::ClassWithOverriddenName"
     end
 
+    it "dumps a class with multibyte characters in name" do
+      source_object = eval("MarshalSpec::MultibyteぁあぃいClass".force_encoding(Encoding::UTF_8))
+      Marshal.dump(source_object).should == "\x04\bc,MarshalSpec::Multibyte\xE3\x81\x81\xE3\x81\x82\xE3\x81\x83\xE3\x81\x84Class"
+    end
+
     it "raises TypeError with an anonymous Class" do
       -> { Marshal.dump(Class.new) }.should raise_error(TypeError, /can't dump anonymous class/)
     end
@@ -211,6 +216,11 @@ def _dump(level)
       Marshal.dump(MarshalSpec::ModuleWithOverriddenName).should == "\x04\bc*MarshalSpec::ModuleWithOverriddenName"
     end
 
+    it "dumps a module with multibyte characters in name" do
+      source_object = eval("MarshalSpec::MultibyteけげこごModule".force_encoding(Encoding::UTF_8))
+      Marshal.dump(source_object).should == "\x04\bm-MarshalSpec::Multibyte\xE3\x81\x91\xE3\x81\x92\xE3\x81\x93\xE3\x81\x94Module"
+    end
+
     it "raises TypeError with an anonymous Module" do
       -> { Marshal.dump(Module.new) }.should raise_error(TypeError)
     end
@@ -664,6 +674,11 @@ class << obj
       Marshal.dump(obj).should include("MarshalSpec::TimeWithOverriddenName")
     end
 
+    it "dumps a Time subclass with multibyte characters in name" do
+      source_object = eval("MarshalSpec::MultibyteぁあぃいTime".force_encoding(Encoding::UTF_8))
+      Marshal.dump(source_object).should == "\x04\bc+MarshalSpec::Multibyte\xE3\x81\x81\xE3\x81\x82\xE3\x81\x83\xE3\x81\x84Time"
+    end
+
     it "raises TypeError with an anonymous Time subclass" do
       -> { Marshal.dump(Class.new(Time).now) }.should raise_error(TypeError)
     end

spec/ruby/core/marshal/fixtures/marshal_data.rb

@@ -262,6 +262,17 @@ def self.name
     end
   end
 
+  module_eval(<<~ruby.force_encoding(Encoding::UTF_8))
+    class MultibyteぁあぃいClass
+    end
+
+    module MultibyteけげこごModule
+    end
+
+    class MultibyteぁあぃいTime < Time
+    end
+  ruby
+
   DATA = {
     "nil" => [nil, "\004\b0"],
     "1..2" => [(1..2),

src/main/ruby/truffleruby/core/marshal.rb

@@ -59,15 +59,15 @@ class Class
     end
 
     name = Primitive.module_name self
-    "c#{ms.serialize_integer(name.length)}#{name}"
+    "c#{ms.serialize_integer(name.bytesize)}#{name.b}"
   end
 end
 
 class Module
   private def __marshal__(ms)
     raise TypeError, "can't dump anonymous module #{self}" if Primitive.module_anonymous?(self)
     name = Primitive.module_name self
-    "m#{ms.serialize_integer(name.length)}#{name}"
+    "m#{ms.serialize_integer(name.bytesize)}#{name.b}"
   end
 end
 
@@ -95,7 +95,7 @@ class Float
       end
     end
 
-    sl = str.length
+    sl = str.bytesize
     if sign == 1
       ss = '-'
       sl += 1
@@ -168,7 +168,7 @@ def __custom_marshal__(ms)
     out << Truffle::Type.binary_string("u#{ms.serialize(name.to_sym)}")
 
     str = _dump
-    out << ms.serialize_integer(str.length) + str
+    out << ms.serialize_integer(str.bytesize) + str.b
 
     count = ivars.size + extra_values.size
     out << ms.serialize_integer(count)