Reload length of retained_maps array after GC.

Reload length of retained_maps array after GC.

This fixes flaky GC stress failure:

> Fatal error in ../src/heap/mark-compact.cc, line 2127
> Check failed: retained_maps->Get(i)->IsWeakCell().

BUG=
TEST=test-heap/RegressArrayListGC

Committed: https://crrev.com/1efcca7f04704c23a6dac29002570c10a7280d67
Cr-Commit-Position: refs/heads/master@{#27412}

[ulan, 2015-03-24 10:09:03]

ulan@chromium.org changed reviewers:
+ hpayer@chromium.org

[ulan, 2015-03-24 10:09:04]

PTAL

[ulan, 2015-03-24 13:31:47]

ulan@chromium.org changed reviewers:
+ jarin@chromium.org

[Jarin, 2015-03-24 13:50:04]

lgtm with nits. Hannes should still have a look.

https://codereview.chromium.org/1026113004/diff/20001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/1026113004/diff/20001/src/objects.cc#newcode8285
src/objects.cc:8285: length = array->Length();
Would it make sense to here (and in the other Add method, too) that the new
length is smaller than the original one?

https://codereview.chromium.org/1026113004/diff/20001/src/objects.h
File src/objects.h (right):

https://codereview.chromium.org/1026113004/diff/20001/src/objects.h#newcode2636
src/objects.h:2636: // Use this if GC can delete elements from array.
nit: ... from the array.

[ulan, 2015-03-24 14:01:36]

Thanks!

As discussed, let's land it and then I'll address Hannes' comment (if any) in
another CL.

https://codereview.chromium.org/1026113004/diff/20001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/1026113004/diff/20001/src/objects.cc#newcode8285
src/objects.cc:8285: length = array->Length();
On 2015/03/24 13:50:04, jarin wrote:
> Would it make sense to here (and in the other Add method, too) that the new
> length is smaller than the original one?

Done.

https://codereview.chromium.org/1026113004/diff/20001/src/objects.h
File src/objects.h (right):

https://codereview.chromium.org/1026113004/diff/20001/src/objects.h#newcode2636
src/objects.h:2636: // Use this if GC can delete elements from array.
On 2015/03/24 13:50:04, jarin wrote:
> nit: ... from the array.

Done.

[ulan, 2015-03-24 14:01:42]

The CQ bit was checked by ulan@chromium.org

[ulan, 2015-03-24 14:01:42]

The patchset sent to the CQ was uploaded after l-g-t-m from jarin@chromium.org
Link to the patchset: https://codereview.chromium.org/1026113004/#ps40001
(title: "Address comments")

[commit-bot: I haz the power, 2015-03-24 14:01:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1026113004/40001

[commit-bot: I haz the power, 2015-03-24 14:13:09]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-03-24 14:13:10]

Try jobs failed on following builders:
  v8_linux_nodcheck_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_nodcheck_rel/build...)

[ulan, 2015-03-24 14:15:57]

The CQ bit was checked by ulan@chromium.org

[ulan, 2015-03-24 14:15:57]

The patchset sent to the CQ was uploaded after l-g-t-m from jarin@chromium.org
Link to the patchset: https://codereview.chromium.org/1026113004/#ps60001
(title: "Remove unused var")

[commit-bot: I haz the power, 2015-03-24 14:16:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1026113004/60001

[commit-bot: I haz the power, 2015-03-24 14:36:00]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-03-24 14:36:21]

Patchset 4 (id:??) landed as
https://crrev.com/1efcca7f04704c23a6dac29002570c10a7280d67
Cr-Commit-Position: refs/heads/master@{#27412}

[Erik Corry, 2015-03-25 09:53:06]

erik.corry@gmail.com changed reviewers:
+ erik.corry@gmail.com

[Erik Corry, 2015-03-25 09:53:07]

https://codereview.chromium.org/1026113004/diff/60001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/1026113004/diff/60001/src/objects.cc#newcode8299
src/objects.cc:8299: length = array->Length();
Missing DCHECK from the single-argument version.

Did you try just doing this unconditionally, and was it too slow?  It looks a
bit like a trap for the unwary now.

https://codereview.chromium.org/1026113004/diff/60001/src/objects.cc#newcode8309
src/objects.cc:8309: int capacity = array->length();
// This is the length of the FixedArray

https://codereview.chromium.org/1026113004/diff/60001/src/objects.cc#newcode8313
src/objects.cc:8313: capacity = capacity + Max(capacity / 2, 2);
Division by 2 of a signed value can be an expensive operation.  Consider either
making it unsigned or using shift.

https://codereview.chromium.org/1026113004/diff/60001/src/objects.cc#newcode8315
src/objects.cc:8315: if (empty) array->SetLength(0);
// This sets the length of the ArrayList, using the first entry in the fixed
array