Reland "Fix memory leak caused by field type in descriptor array."

Reland "Fix memory leak caused by field type in descriptor array."

BUG=v8:3877
LOG=NO

Committed: https://crrev.com/c56fa73699bc7d6799c7ebfe55848005bf9b62e4
Cr-Commit-Position: refs/heads/master@{#27362}

[ulan, 2015-03-19 10:46:00]

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

[ulan, 2015-03-19 10:46:00]

PTAL. Fix is in delta between PS1 and PS2

[Erik Corry, 2015-03-19 11:45:18]

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

[Erik Corry, 2015-03-19 11:45:18]

lgtm

https://codereview.chromium.org/1018283002/diff/20001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/1018283002/diff/20001/src/objects.cc#newcode1703
src/objects.cc:1703: if (type->IsClass()) return
Map::WeakCellForMap(type->AsClass()->Map());
You could assert here that it's a Smi?

[Toon Verwaest, 2015-03-19 12:20:34]

lgtm

This is getting a bit out of hand with double-wrapped types... can you remove
HeapType from the mix? I'm ok if you do it in a different CL though.

[ulan, 2015-03-23 10:59:08]

On 2015/03/19 12:20:34, Toon Verwaest wrote:
> lgtm
> 
> This is getting a bit out of hand with double-wrapped types... can you remove
> HeapType from the mix? I'm ok if you do it in a different CL though.

Thanks, I'll remove HeapType once this CL sticks.

[ulan, 2015-03-23 10:59:15]

The CQ bit was checked by ulan@chromium.org

[ulan, 2015-03-23 10:59:16]

The patchset sent to the CQ was uploaded after l-g-t-m from
erik.corry@gmail.com, verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1018283002/#ps40001
(title: "Rebase")

[commit-bot: I haz the power, 2015-03-23 10:59:26]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1018283002/40001

[commit-bot: I haz the power, 2015-03-23 11:20:44]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-03-23 11:21:04]

Patchset 3 (id:??) landed as
https://crrev.com/c56fa73699bc7d6799c7ebfe55848005bf9b62e4
Cr-Commit-Position: refs/heads/master@{#27362}

[Jakob Kummerow, 2015-04-11 12:00:56]

jkummerow@chromium.org changed reviewers:
+ jkummerow@chromium.org

[Jakob Kummerow, 2015-04-11 12:00:58]

https://codereview.chromium.org/1018283002/diff/40001/src/objects-debug.cc
File src/objects-debug.cc (right):

https://codereview.chromium.org/1018283002/diff/40001/src/objects-debug.cc#ne...
src/objects-debug.cc:293: bool type_is_any = field_type->Is(HeapType::Any());
Shouldn't this be the other way round, i.e. HeapType::Any()->Is(field_type) ?
AKAIF "t->Is(Any)" will be true for every type t, whereas "Any->Is(t)" is only
true when t == Any.