Version 1.2.4.
Improved performance of floating point number allocation for ARM platforms.
Fixed crash when using the instanceof operator on functions with number values in their prototype chain (issue 341).
Optimized virtual frame operations in the code generator to speed up compilation time and allocated the frames in the zone.
Made the representation of virtual frames and jump targets in the code generator much more compact.
Avoided linear search for non-locals in scope code when resolving variables inside with and eval scopes.
Optimized lexical scanner by dealing with whitespace as part of the token scanning instead of as a separate step before it.
Changed the scavenging collector so that promoted objects do not reside in the old generation while their remembered set is being swept for pointers into the young generation.
Fixed numeric overflow handling when compiling count operations.
git-svn-id: http://v8.googlecode.com/svn/trunk@1983 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
diff --git a/src/heap-inl.h b/src/heap-inl.h
index 86165ee..3d734e9 100644
--- a/src/heap-inl.h
+++ b/src/heap-inl.h
@@ -191,6 +191,27 @@
}
+void Heap::ScavengeObject(HeapObject** p, HeapObject* object) {
+ ASSERT(InFromSpace(object));
+
+ // We use the first word (where the map pointer usually is) of a heap
+ // object to record the forwarding pointer. A forwarding pointer can
+ // point to an old space, the code space, or the to space of the new
+ // generation.
+ MapWord first_word = object->map_word();
+
+ // If the first word is a forwarding address, the object has already been
+ // copied.
+ if (first_word.IsForwardingAddress()) {
+ *p = first_word.ToForwardingAddress();
+ return;
+ }
+
+ // Call the slow part of scavenge object.
+ return ScavengeObjectSlow(p, object);
+}
+
+
Object* Heap::GetKeyedLookupCache() {
if (keyed_lookup_cache()->IsUndefined()) {
Object* obj = LookupCache::Allocate(4);