HeapProfiler: check that heap snapshot has no unretained entries except root.

HeapProfiler: check that heap snapshot has no unretained entries except root.

TEST=AllocationSitesAreVisible
BUG=
R=mvstanton@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=15591

[mvstanton, 2013-07-10 12:10:49]

This is great, thanks Ilya. lgtm.
--Michael

[loislo, 2013-07-10 12:40:49]

Committed patchset #2 manually as r15591 (presubmit successful).

[yurys, 2013-07-10 16:27:48]

I'm wondering if there was some problem that prompted adding this test?

[loislo, 2013-07-10 17:13:06]

On 2013/07/10 16:27:48, Yury Semikhatsky wrote:
> I'm wondering if there was some problem that prompted adding this test?

This check is a downstream version of our expectation about heap-snapshot
structure.

5 days ago danno landed a patch
https://code.google.com/p/v8/source/detail?r=15512
And in the combination with today's patch
https://chromiumcodereview.appspot.com/18751003/ we got ~20 arrays in the heap
snapshot with no retainers.
As a result our blink tests for heap profiler became red.
Actually they were timing out and it was very hard to understand the reason of
that.

[yurys, 2013-07-11 16:53:24]

On 2013/07/10 17:13:06, loislo wrote:
> On 2013/07/10 16:27:48, Yury Semikhatsky wrote:
> > I'm wondering if there was some problem that prompted adding this test?
> 
> This check is a downstream version of our expectation about heap-snapshot
> structure.
> 
> 5 days ago danno landed a patch
> https://code.google.com/p/v8/source/detail?r=15512
> And in the combination with today's patch
> https://chromiumcodereview.appspot.com/18751003/ we got ~20 arrays in the heap
> snapshot with no retainers.
> As a result our blink tests for heap profiler became red.
> Actually they were timing out and it was very hard to understand the reason of
> that.

Thanks for the clarification! Next time please include such information into the
CL description or into a bug.