Transition shape when object's capacity changes

This commit adds a `capacity` field to shapes, and adds shape
transitions whenever an object's capacity changes. Objects which are
allocated out of a bigger size pool will also make a transition from the
root shape to the shape with the correct capacity for their size pool
when they are allocated.

This commit will allow us to remove numiv from objects completely, and
will also mean we can guarantee that if two objects share shapes, their
IVs are in the same positions (an embedded and extended object cannot
share shapes). This will enable us to implement ivar sets in YJIT using
object shapes.

Co-Authored-By: Aaron Patterson <[email protected]>

4 files changed, 34 insertions, 30 deletions

diff --git a/test/-ext-/string/test_cstr.rb b/test/-ext-/string/test_cstr.rb
index d909781700..efc64119dc 100644
--- a/test/-ext-/string/test_cstr.rb
+++ b/test/-ext-/string/test_cstr.rb
@@ -43,7 +43,11 @@ class Test_StringCStr < Test::Unit::TestCase
   end
 
   def test_rb_str_new_frozen_embed
-    str = Bug::String.cstr_noembed("rbconfig.rb")
+    # "rbconfi" is the smallest "maximum embeddable string".  VWA adds
+    # a capacity field, which removes one pointer capacity for embedded objects,
+    # so if VWA is enabled, but there is only one size pool, then the
+    # maximum embeddable capacity on 32 bit machines is 8 bytes.
+    str = Bug::String.cstr_noembed("rbconfi")
     str = Bug::String.rb_str_new_frozen(str)
     assert_equal true, Bug::String.cstr_embedded?(str)
   end
diff --git a/test/objspace/test_objspace.rb b/test/objspace/test_objspace.rb
index 5994fadeff..2366ec3b61 100644
--- a/test/objspace/test_objspace.rb
+++ b/test/objspace/test_objspace.rb
@@ -277,7 +277,7 @@ class TestObjSpace < Test::Unit::TestCase
     info = nil
     ObjectSpace.trace_object_allocations do
       line = __LINE__ + 1
-      str = "hello world"
+      str = "hello w"
       info = ObjectSpace.dump(str)
     end
     assert_dump_object(info, line)
@@ -289,7 +289,7 @@ class TestObjSpace < Test::Unit::TestCase
       th = Thread.start {r.read}
       ObjectSpace.trace_object_allocations do
         line = __LINE__ + 1
-        str = "hello world"
+        str = "hello w"
         ObjectSpace.dump(str, output: w)
       end
       w.close
@@ -301,7 +301,7 @@ class TestObjSpace < Test::Unit::TestCase
   def assert_dump_object(info, line)
     loc = caller_locations(1, 1)[0]
     assert_match(/"type":"STRING"/, info)
-    assert_match(/"embedded":true, "bytesize":11, "value":"hello world", "encoding":"UTF-8"/, info)
+    assert_match(/"embedded":true, "bytesize":7, "value":"hello w", "encoding":"UTF-8"/, info)
     assert_match(/"file":"#{Regexp.escape __FILE__}", "line":#{line}/, info)
     assert_match(/"method":"#{loc.base_label}"/, info)
     JSON.parse(info) if defined?(JSON)
@@ -549,17 +549,17 @@ class TestObjSpace < Test::Unit::TestCase
     #
     # This test makes assertions on the assignment to `str`, so we look for
     # the second appearance of /TEST STRING/ in the output
-    test_string_in_dump_all = output.grep(/TEST STRING/)
-    assert_equal(test_string_in_dump_all.size, 2)
+    test_string_in_dump_all = output.grep(/TEST2/)
+    assert_equal(2, test_string_in_dump_all.size, "number of strings")
 
     entry_hash = JSON.parse(test_string_in_dump_all[1])
 
-    assert_equal(entry_hash["bytesize"], 11)
-    assert_equal(entry_hash["value"], "TEST STRING")
-    assert_equal(entry_hash["encoding"], "UTF-8")
-    assert_equal(entry_hash["file"], "-")
-    assert_equal(entry_hash["line"], 4)
-    assert_equal(entry_hash["method"], "dump_my_heap_please")
+    assert_equal(5, entry_hash["bytesize"], "bytesize is wrong")
+    assert_equal("TEST2", entry_hash["value"], "value is wrong")
+    assert_equal("UTF-8", entry_hash["encoding"], "encoding is wrong")
+    assert_equal("-", entry_hash["file"], "file is wrong")
+    assert_equal(4, entry_hash["line"], "line is wrong")
+    assert_equal("dump_my_heap_please", entry_hash["method"], "method is wrong")
     assert_not_nil(entry_hash["generation"])
   end
 
@@ -571,7 +571,7 @@ class TestObjSpace < Test::Unit::TestCase
         def dump_my_heap_please
           ObjectSpace.trace_object_allocations_start
           GC.start
-          str = "TEST STRING".force_encoding("UTF-8")
+          str = "TEST2".force_encoding("UTF-8")
           ObjectSpace.dump_all(output: :stdout)
         end
 
@@ -586,7 +586,7 @@ class TestObjSpace < Test::Unit::TestCase
         def dump_my_heap_please
           ObjectSpace.trace_object_allocations_start
           GC.start
-          (str = "TEST STRING").force_encoding("UTF-8")
+          (str = "TEST2").force_encoding("UTF-8")
           ObjectSpace.dump_all().path
         end
 
diff --git a/test/ruby/test_gc_compact.rb b/test/ruby/test_gc_compact.rb
index 92a2be1174..bae29a3162 100644
--- a/test/ruby/test_gc_compact.rb
+++ b/test/ruby/test_gc_compact.rb
@@ -210,7 +210,7 @@ class TestGCCompact < Test::Unit::TestCase
   end
 
   def test_moving_arrays_down_size_pools
-    omit if !GC.using_rvargc?
+    omit if GC::INTERNAL_CONSTANTS[:SIZE_POOL_COUNT] == 1
     assert_separately([], "#{<<~"begin;"}\n#{<<~"end;"}", timeout: 10, signal: :SEGV)
     begin;
       ARY_COUNT = 500
@@ -229,7 +229,8 @@ class TestGCCompact < Test::Unit::TestCase
   end
 
   def test_moving_arrays_up_size_pools
-    omit if !GC.using_rvargc?
+    omit if GC::INTERNAL_CONSTANTS[:SIZE_POOL_COUNT] == 1
+
     assert_separately([], "#{<<~"begin;"}\n#{<<~"end;"}", timeout: 10, signal: :SEGV)
     begin;
       ARY_COUNT = 500
@@ -250,6 +251,8 @@ class TestGCCompact < Test::Unit::TestCase
   end
 
   def test_moving_objects_between_size_pools
+    omit if GC::INTERNAL_CONSTANTS[:SIZE_POOL_COUNT] == 1
+
     assert_separately([], "#{<<~"begin;"}\n#{<<~"end;"}", timeout: 10, signal: :SEGV)
     begin;
       class Foo
@@ -274,7 +277,8 @@ class TestGCCompact < Test::Unit::TestCase
   end
 
   def test_moving_strings_up_size_pools
-    omit if !GC.using_rvargc?
+    omit if GC::INTERNAL_CONSTANTS[:SIZE_POOL_COUNT] == 1
+
     assert_separately([], "#{<<~"begin;"}\n#{<<~"end;"}", timeout: 10, signal: :SEGV)
     begin;
       STR_COUNT = 500
@@ -292,7 +296,8 @@ class TestGCCompact < Test::Unit::TestCase
   end
 
   def test_moving_strings_down_size_pools
-    omit if !GC.using_rvargc?
+    omit if GC::INTERNAL_CONSTANTS[:SIZE_POOL_COUNT] == 1
+
     assert_separately([], "#{<<~"begin;"}\n#{<<~"end;"}", timeout: 10, signal: :SEGV)
     begin;
       STR_COUNT = 500
diff --git a/test/ruby/test_shapes.rb b/test/ruby/test_shapes.rb
index 23696acc70..326ff3a453 100644
--- a/test/ruby/test_shapes.rb
+++ b/test/ruby/test_shapes.rb
@@ -86,15 +86,10 @@ class TestShapes < Test::Unit::TestCase
     assert_equal(2, bar_shape.next_iv_index)
   end
 
-  def test_new_obj_has_root_shape
-    assert_shape_equal(RubyVM::Shape.root_shape, RubyVM::Shape.of(Object.new))
-  end
+  class TestObject; end
 
-  def test_frozen_new_obj_has_frozen_root_shape
-    assert_shape_equal(
-      RubyVM::Shape.frozen_root_shape,
-      RubyVM::Shape.of(Object.new.freeze)
-    )
+  def test_new_obj_has_root_shape
+    assert_shape_equal(RubyVM::Shape.root_shape, RubyVM::Shape.of(TestObject.new))
   end
 
   def test_str_has_root_shape
@@ -109,12 +104,12 @@ class TestShapes < Test::Unit::TestCase
     assert_shape_equal(RubyVM::Shape.root_shape, RubyVM::Shape.of({}))
   end
 
-  def test_true_has_frozen_root_shape
-    assert_shape_equal(RubyVM::Shape.frozen_root_shape, RubyVM::Shape.of(true))
+  def test_true_has_special_const_shape_id
+    assert_equal(RubyVM::Shape::SPECIAL_CONST_SHAPE_ID, RubyVM::Shape.of(true).id)
   end
 
-  def test_nil_has_frozen_root_shape
-    assert_shape_equal(RubyVM::Shape.frozen_root_shape, RubyVM::Shape.of(nil))
+  def test_nil_has_special_const_shape_id
+    assert_equal(RubyVM::Shape::SPECIAL_CONST_SHAPE_ID, RubyVM::Shape.of(nil).id)
   end
 
   def test_basic_shape_transition