Remove internalized cons string types.

Remove internalized cons string types.

Currently, internalizing a cons string could result in either an
in-place converted internalized cons string or a newly created
internalized sequential string, depending on allocation success.

The former could end up being embedded into an IC, which is not
supported.

R=mstarzinger@chromium.org
BUG=357103
LOG=N

Committed: https://code.google.com/p/v8/source/detail?r=20394

[Michael Starzinger, 2014-04-01 10:55:18]

LGTM with one suggestion.

https://codereview.chromium.org/218993011/diff/1/src/objects.cc
File src/objects.cc (left):

https://codereview.chromium.org/218993011/diff/1/src/objects.cc#oldcode13877
src/objects.cc:13877: string_ = string_->TryFlattenGetString();
I am not an expert on this, but doesn't it still make sense to try to flatten
any cons-string before attempting to internalize?

[Yang, 2014-04-01 11:15:23]

On 2014/04/01 10:55:18, Michael Starzinger wrote:
> LGTM with one suggestion.
> 
> https://codereview.chromium.org/218993011/diff/1/src/objects.cc
> File src/objects.cc (left):
> 
> https://codereview.chromium.org/218993011/diff/1/src/objects.cc#oldcode13877
> src/objects.cc:13877: string_ = string_->TryFlattenGetString();
> I am not an expert on this, but doesn't it still make sense to try to flatten
> any cons-string before attempting to internalize?

Flattening a string only affects cons strings. If flattening fails, we will end
up at AllocateInternalizedStringImpl. If flattening does not fail, TryFlatten
does essentially the same thing as AllocateInternalizedStringImpl. So there is
no point doing the flattening.

[Yang, 2014-04-01 11:30:42]

Committed patchset #1 manually as r20394 (presubmit successful).