Keep track of array buffers in new space separately

Keep track of array buffers in new space separately

BUG=v8:3996
R=hpayer@chromium.org
LOG=n

Committed: https://crrev.com/506397d0a4241c19f5fab890e49e22d1d9b28bdc
Cr-Commit-Position: refs/heads/master@{#28978}

[Hannes Payer (out of office), 2015-05-08 13:39:30]

quick look over the cl, see comments. you got some buildbot failures. please
address first.

https://codereview.chromium.org/1133773002/diff/20001/src/heap/heap.cc
File src/heap/heap.cc (right):

https://codereview.chromium.org/1133773002/diff/20001/src/heap/heap.cc#newcod...
src/heap/heap.cc:1789: }  // namespace
Why did you add a namespace?

https://codereview.chromium.org/1133773002/diff/20001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1133773002/diff/20001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3110:
heap()->PromoteArrayBuffer(JSArrayBuffer::cast(target));
Please don't add any JS type specific functionality to this function.

You could iterate over the map after evacuation and see what got promoted.

[jochen (gone - plz use gerrit), 2015-05-08 14:42:56]

bots are happy now..

as discussed, i don't have pointers to the objects.

what about adding something like HeapObject::NeedsPromotionEpilog() and for
those objects call Heap::ObjectPromotionEpilog()? I could also move the js
function stuff from the scavenger promotion there...

https://codereview.chromium.org/1133773002/diff/20001/src/heap/heap.cc
File src/heap/heap.cc (right):

https://codereview.chromium.org/1133773002/diff/20001/src/heap/heap.cc#newcod...
src/heap/heap.cc:1789: }  // namespace
On 2015/05/08 13:39:29, Hannes Payer wrote:
> Why did you add a namespace?

so the helpers above don't pollute the global namespace

[Hannes Payer (out of office), 2015-05-12 06:33:29]

https://codereview.chromium.org/1133773002/diff/20001/src/heap/heap.cc
File src/heap/heap.cc (right):

https://codereview.chromium.org/1133773002/diff/20001/src/heap/heap.cc#newcod...
src/heap/heap.cc:1789: }  // namespace
On 2015/05/08 14:42:56, jochen (traveling) wrote:
> On 2015/05/08 13:39:29, Hannes Payer wrote:
> > Why did you add a namespace?
> 
> so the helpers above don't pollute the global namespace

I looked it up in the style guide. That's fine. However, we never do that in our
code base.

https://codereview.chromium.org/1133773002/diff/40001/src/heap/heap.cc
File src/heap/heap.cc (right):

https://codereview.chromium.org/1133773002/diff/40001/src/heap/heap.cc#newcod...
src/heap/heap.cc:2251: } else {
This logic could also be moved to the promotion queue. We should have there a
clean dispatch.

https://codereview.chromium.org/1133773002/diff/40001/src/heap/heap.cc#newcod...
src/heap/heap.cc:2256: heap->PromoteArrayBuffer(JSArrayBuffer::cast(target));
What about moving this logic into the promotion queue? That may be cleaner.

https://codereview.chromium.org/1133773002/diff/40001/src/heap/heap.h
File src/heap/heap.h (right):

https://codereview.chromium.org/1133773002/diff/40001/src/heap/heap.h#newcode...
src/heap/heap.h:1513: 
Please add a bunch of comments about what these functions are doing.

https://codereview.chromium.org/1133773002/diff/40001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1133773002/diff/40001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3110:
heap()->PromoteArrayBuffer(JSArrayBuffer::cast(target));
I think we should clean that up together with the specialized code in migrate
object. I like the idea of having a MigrationEpilog. Let's try that and see how
it looks.

[jochen (gone - plz use gerrit), 2015-06-11 21:38:10]

ptal

[Hannes Payer (out of office), 2015-06-12 05:15:03]

lgtm

[jochen (gone - plz use gerrit), 2015-06-12 05:23:30]

The CQ bit was checked by jochen@chromium.org

[commit-bot: I haz the power, 2015-06-12 05:23:40]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1133773002/60001

[commit-bot: I haz the power, 2015-06-12 05:35:29]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-06-12 05:35:46]

Patchset 4 (id:??) landed as
https://crrev.com/506397d0a4241c19f5fab890e49e22d1d9b28bdc
Cr-Commit-Position: refs/heads/master@{#28978}

[Benedikt Meurer, 2015-06-12 06:28:54]

A revert of this CL (patchset #4 id:60001) has been created in
https://codereview.chromium.org/1178403003/ by bmeurer@chromium.org.

The reason for reverting is: GC stress unhappy.