Turn old space cons strings into regular external strings (not short).

Turn old space cons strings into regular external strings (not short).

R=hpayer@chromium.org
BUG=v8:3530
LOG=N

Committed: https://code.google.com/p/v8/source/detail?r=23794

[Hannes Payer (out of office), 2014-07-03 11:37:19]

LGTM

https://codereview.chromium.org/368223002/diff/1/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/368223002/diff/1/src/objects.cc#newcode1071
src/objects.cc:1071: // string occupies is too small for a regular  external
string.
minus one blank before "external string"

https://codereview.chromium.org/368223002/diff/1/src/objects.cc#newcode1072
src/objects.cc:1072: // Instead, we resort to a short external string instead,
omitting
drop the second instead

https://codereview.chromium.org/368223002/diff/1/src/objects.cc#newcode1138
src/objects.cc:1138: // string occupies is too small for a regular  external
string.
same as above

[Hannes Payer (out of office), 2014-07-03 11:40:09]

https://codereview.chromium.org/368223002/diff/1/test/mjsunit/regress/regress...
File test/mjsunit/regress/regress-sliced-external-cons-regexp.js (right):

https://codereview.chromium.org/368223002/diff/1/test/mjsunit/regress/regress...
test/mjsunit/regress/regress-sliced-external-cons-regexp.js:10: gc();
you could call gc(true) to just invoke the scavenger

[Yang, 2014-07-03 11:45:44]

On 2014/07/03 11:40:09, Hannes Payer wrote:
>
https://codereview.chromium.org/368223002/diff/1/test/mjsunit/regress/regress...
> File test/mjsunit/regress/regress-sliced-external-cons-regexp.js (right):
> 
>
https://codereview.chromium.org/368223002/diff/1/test/mjsunit/regress/regress...
> test/mjsunit/regress/regress-sliced-external-cons-regexp.js:10: gc();
> you could call gc(true) to just invoke the scavenger

Just tested. I really do have to call gc() twice to trigger the bug.

[Yang, 2014-07-03 11:46:39]

Committed patchset #1 manually as r22192 (presubmit successful).

[Yang, 2014-08-19 15:32:13]

On 2014/07/03 11:46:39, Yang wrote:
> Committed patchset #1 manually as r22192 (presubmit successful).

Reusing this CL. Last time landing this failed due to memory corruption. Shall
we try again?

[Yang, 2014-08-19 15:32:52]

On 2014/08/19 15:32:13, Yang wrote:
> On 2014/07/03 11:46:39, Yang wrote:
> > Committed patchset #1 manually as r22192 (presubmit successful).
> 
> Reusing this CL. Last time landing this failed due to memory corruption. Shall
> we try again?

This fixes crbug.com/403521, if I reland
https://codereview.chromium.org/460643002/

[Hannes Payer (out of office), 2014-08-19 17:33:18]

lgtm

[Hannes Payer (out of office), 2014-08-19 17:33:18]

lgtm

[Yang, 2014-08-21 15:09:34]

On 2014/08/19 17:33:18, Hannes Payer wrote:
> lgtm

This still fails test-api/MakingExternalUnalignedAsciiString. In other words,
unaligned addresses in an external string that's in old pointer space still
causes GC to crash. Do we have a fix for that on the horizon?

[Yang, 2014-09-09 11:42:14]

Committed patchset #2 (id:20001) manually as 23794 (presubmit successful).