Add v8::HeapGraphNode::GetHeapValue method.

Add v8::HeapGraphNode::GetHeapValue method.

This allows getting a heap object by a heap snapshot node for further
inspection. Very useful for objects from bindings that mostly have
getter-provided fields.

R=vegorov@google.com
BUG=webkit/61179
TEST=test-heap-profiler/GetHeapValue,GetHeapValueForDeletedObject

Committed: http://code.google.com/p/v8/source/detail?r=9430

[Vyacheslav Egorov (Chromium), 2011-09-26 14:22:09]

http://codereview.chromium.org/8046006/diff/1/src/profile-generator.cc
File src/profile-generator.cc (right):

http://codereview.chromium.org/8046006/diff/1/src/profile-generator.cc#newcod...
src/profile-generator.cc:1540: if (ids_.FindObject(obj->address()) == id) {
this is not going to work if obj was in new space or was on the page that was
compacted.

we either will fail to find the object or will return the wrong one.

http://codereview.chromium.org/8046006/diff/1/src/profile-generator.cc#newcod...
src/profile-generator.cc:1543: // Can't break -- kFilterUnreachable requires
full heap traversal.
We can fix this by switching kFilterUnreachable from intrusive marking to
mark-bits based marking (just saying).

[mnaganov (inactive), 2011-09-26 14:28:18]

http://codereview.chromium.org/8046006/diff/1/src/profile-generator.cc
File src/profile-generator.cc (right):

http://codereview.chromium.org/8046006/diff/1/src/profile-generator.cc#newcod...
src/profile-generator.cc:1540: if (ids_.FindObject(obj->address()) == id) {
On 2011/09/26 14:22:10, Vyacheslav Egorov wrote:
> this is not going to work if obj was in new space or was on the page that was
> compacted.
> 
> we either will fail to find the object or will return the wrong one.

What do you mean? We start keeping IDs for heap objects after the first heap
snapshot has been done. We should have 'HeapProfiler::ObjectMoveEvent' in proper
places to track existing object moves, and new objects are added when we are
taking the next snapshot.

http://codereview.chromium.org/8046006/diff/1/src/profile-generator.cc#newcod...
src/profile-generator.cc:1543: // Can't break -- kFilterUnreachable requires
full heap traversal.
On 2011/09/26 14:22:10, Vyacheslav Egorov wrote:
> We can fix this by switching kFilterUnreachable from intrusive marking to
> mark-bits based marking (just saying).

Yes, that's a good idea. I will create a bug to track this.

[Vyacheslav Egorov (Chromium), 2011-09-26 14:32:44]

Sorry, I've completely forgotten about ObjectMove events, shame on me!

LGTM.