vm_getivar: store the common ancestor shape in the inline cache

This is a continuation of the `rb_shape_get_iv_index_with_hint` patch.

Now we return the common ancestor id to `vm_get_ivar`, and it uses it
as the new cache key. This helps the IC stay stable instead of flip flopping.

benchmark/vm_ivar_memoize.yml

@@ -0,0 +1,85 @@
+prelude: |
+  IVARS = 60
+  class Record
+    def initialize(offset = false)
+      @offset = 1 if offset
+      @first = 0
+      IVARS.times do |i|
+        instance_variable_set("@ivar_#{i}", i)
+      end
+    end
+
+    def first
+      @first
+    end
+
+    def lazy_set
+      @lazy_set ||= 123
+    end
+
+    def undef
+      @undef
+    end
+  end
+
+  Record.new # Need one alloc to right size
+
+  BASE = Record.new
+  LAZY = Record.new
+  LAZY.lazy_set
+
+  class Miss < Record
+    @first = 0
+    IVARS.times do |i|
+      instance_variable_set("@i_#{i}", i)
+    end
+  end
+
+  Miss.new # Need one alloc to right size
+  MISS = Miss.new
+
+  DIVERGENT = Record.new(true)
+
+benchmark:
+  vm_ivar_stable_shape: |
+    BASE.first
+    BASE.first
+    BASE.first
+    BASE.first
+    BASE.first
+    BASE.first
+  vm_ivar_memoize_unstable_shape: |
+    BASE.first
+    LAZY.first
+    BASE.first
+    LAZY.first
+    BASE.first
+    LAZY.first
+  vm_ivar_memoize_unstable_shape_miss: |
+    BASE.first
+    MISS.first
+    BASE.first
+    MISS.first
+    BASE.first
+    MISS.first
+  vm_ivar_unstable_undef: |
+    BASE.undef
+    LAZY.undef
+    BASE.undef
+    LAZY.undef
+    BASE.undef
+    LAZY.undef
+  vm_ivar_divergent_shape: |
+    BASE.first
+    DIVERGENT.first
+    BASE.first
+    DIVERGENT.first
+    BASE.first
+    DIVERGENT.first
+  vm_ivar_divergent_shape_imbalanced: |
+    BASE.first
+    DIVERGENT.first
+    DIVERGENT.first
+    DIVERGENT.first
+    DIVERGENT.first
+    DIVERGENT.first

shape.c

@@ -436,12 +436,16 @@ rb_shape_transition_shape_capa(rb_shape_t* shape)
 // Same as rb_shape_get_iv_index, but uses a provided valid shape id and index
 // to return a result faster if branches of the shape tree are closely related.
 bool
-rb_shape_get_iv_index_with_hint(rb_shape_t * shape, ID id, attr_index_t *value, shape_id_t shape_id_hint, attr_index_t index_hint)
+rb_shape_get_iv_index_with_hint(shape_id_t shape_id, ID id, attr_index_t *value, shape_id_t *shape_id_hint)
 {
-    if (shape_id_hint == INVALID_SHAPE_ID)
+    attr_index_t index_hint = *value;
+    rb_shape_t *shape = rb_shape_get_shape_by_id(shape_id);
+    if (*shape_id_hint == INVALID_SHAPE_ID) {
+        *shape_id_hint = shape_id;
         return rb_shape_get_iv_index(shape, id, value);
+    }
 
-    rb_shape_t * shape_hint = rb_shape_get_shape_by_id(shape_id_hint);
+    rb_shape_t * shape_hint = rb_shape_get_shape_by_id(*shape_id_hint);
 
     while (shape->next_iv_index > index_hint) {
         while (shape_hint->next_iv_index > shape->next_iv_index) {
@@ -450,12 +454,9 @@ rb_shape_get_iv_index_with_hint(rb_shape_t * shape, ID id, attr_index_t *value,
 
         if (shape_hint == shape) {
             // We've found a common ancestor so use the index hint
-            if (index_hint == (attr_index_t)-1) {
-                return false;
-            } else {
-                *value = index_hint;
-                return true;
-            }
+            *value = index_hint;
+            *shape_id_hint = rb_shape_id(shape);
+            return true;
         }
         if (shape->edge_name == id) {
             // We found the matching id before a common ancestor
@@ -467,6 +468,7 @@ rb_shape_get_iv_index_with_hint(rb_shape_t * shape, ID id, attr_index_t *value,
         shape = rb_shape_get_parent(shape);
     }
 
+    *shape_id_hint = shape_id;
     return rb_shape_get_iv_index(shape, id, value);
 }
 

shape.h

@@ -144,7 +144,7 @@ rb_shape_t* rb_shape_get_shape_by_id(shape_id_t shape_id);
 shape_id_t rb_shape_get_shape_id(VALUE obj);
 rb_shape_t * rb_shape_get_next_iv_shape(rb_shape_t * shape, ID id);
 bool rb_shape_get_iv_index(rb_shape_t * shape, ID id, attr_index_t * value);
-bool rb_shape_get_iv_index_with_hint(rb_shape_t * shape, ID id, attr_index_t * value, shape_id_t shape_id_hint, attr_index_t index_hint);
+bool rb_shape_get_iv_index_with_hint(shape_id_t shape_id, ID id, attr_index_t * value, shape_id_t *shape_id_hint);
 bool rb_shape_obj_too_complex(VALUE obj);
 
 void rb_shape_set_shape(VALUE obj, rb_shape_t* shape);

vm_insnhelper.c

@@ -1286,22 +1286,28 @@ vm_getivar(VALUE obj, ID id, const rb_iseq_t *iseq, IVC ic, const struct rb_call
         }
 #endif
 
-        rb_shape_t *shape = rb_shape_get_shape_by_id(shape_id);
-
         if (shape_id == OBJ_TOO_COMPLEX_SHAPE_ID) {
             if (!st_lookup(ROBJECT_IV_HASH(obj), id, &val)) {
                 val = default_value;
             }
         }
         else {
-            if (rb_shape_get_iv_index_with_hint(shape, id, &index, cached_id, index)) {
+            shape_id_t previous_cached_id = cached_id;
+            if (rb_shape_get_iv_index_with_hint(shape_id, id, &index, &cached_id)) {
                 // This fills in the cache with the shared cache object.
                 // "ent" is the shared cache object
-                fill_ivar_cache(iseq, ic, cc, is_attr, index, shape_id);
+                if (cached_id != previous_cached_id) {
+                    fill_ivar_cache(iseq, ic, cc, is_attr, index, cached_id);
+                }
 
-                // We fetched the ivar list above
-                val = ivar_list[index];
-                RUBY_ASSERT(!UNDEF_P(val));
+                if (index == ATTR_INDEX_NOT_SET) {
+                    val = default_value;
+                }
+                else {
+                    // We fetched the ivar list above
+                    val = ivar_list[index];
+                    RUBY_ASSERT(!UNDEF_P(val));
+                }
             }
             else {
                 if (is_attr) {