Add "own" symbols support.

Add "own" symbols support.

"Own" symbols are symbols that can only denote own properties of
objects.

R=hpayer@chromium.org, verwaest@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=23056

Committed: https://code.google.com/p/v8/source/detail?r=23083

[Dmitry Lomov (no reviews), 2014-08-11 15:14:01]

PTAL. I _hope_ I cover all of "old-style" lookup, let me know if I missed
something or there are more tests to write here.

[Toon Verwaest, 2014-08-11 15:56:30]

lgtm

[Dmitry Lomov (no reviews), 2014-08-11 16:06:41]

https://codereview.chromium.org/464473002/diff/20001/src/lookup.h
File src/lookup.h (right):

https://codereview.chromium.org/464473002/diff/20001/src/lookup.h#newcode193
src/lookup.h:193: return static_cast<Configuration>(configuration && CHECK_OWN);
Oh noes this '&&' is wrong of course :)

[arv (Not doing code reviews), 2014-08-11 16:20:18]

I don't think there is a single case (today) where we want a private symbol that
is not own.

Suggestion: Once this lands. Change all the internal usages to private own. Then
remove the private non own symbol.

[Dmitry Lomov (no reviews), 2014-08-11 16:27:30]

On 2014/08/11 16:20:18, arv wrote:
> I don't think there is a single case (today) where we want a private symbol
that
> is not own.
> 
> Suggestion: Once this lands. Change all the internal usages to private own.
Then
> remove the private non own symbol.

Step by step is our motto :)

[Dmitry Lomov (no reviews), 2014-08-11 18:59:55]

Committed patchset #3 manually as 23056 (presubmit successful).

[Dmitry Lomov (no reviews), 2014-08-12 14:38:22]

Adding Hannes for heap.cc change. 
We now pretenure runtime functions list so that GC tests are happy

[Hannes Payer (out of office), 2014-08-12 15:24:00]

lgtm

[Dmitry Lomov (no reviews), 2014-08-12 15:28:40]

Committed patchset #4 manually as 23083 (presubmit successful).

[Dmitry Lomov (no reviews), 2014-08-18 13:38:09]

On 2014/08/11 16:20:18, arv wrote:
> I don't think there is a single case (today) where we want a private symbol
that
> is not own.
> 
> Suggestion: Once this lands. Change all the internal usages to private own.
Then
> remove the private non own symbol.

Turns out we actually do use private non-own symbol in at least one case:
https://code.google.com/p/chromium/codesearch#chromium/src/v8/src/messages.js...

[Toon Verwaest, 2014-08-18 14:20:58]

I'm pretty sure we don't care whether that particular case is "own" or not.
Adding Yang to be sure.

[Yang, 2014-08-18 14:23:35]

On 2014/08/18 14:20:58, Toon Verwaest wrote:
> I'm pretty sure we don't care whether that particular case is "own" or not.
> Adding Yang to be sure.

Yeah. I already talked to Dmitry about this. What we only want is to get the
private property off the holder of the getter, which is the Error object. We
iterate the prototype chain to find the holder anyways, so we could just get own
private property as well here.