Move GetRootListIndex into Heap.

Move GetRootListIndex into Heap.

R=hpayer@chromium.org
BUG=

Committed: https://crrev.com/cb08656b6df964b9312a5f73f61f1bdfed371cf0
Cr-Commit-Position: refs/heads/master@{#27891}

[Hannes Payer (out of office), 2015-04-16 14:04:28]

lgtm

[Yang, 2015-04-16 14:13:11]

On 2015/04/16 14:04:28, Hannes Payer wrote:
> lgtm

The serializer uses RootIndexMap for lookup. We cache an instance of it on the
isolate. Only non-writable roots are included in the RootIndexMap, and
test-heap/WritableVsImmortalRoots makes sure that immortal roots are
non-writable.

Maybe it makes sense to move that to the heap to do table lookups instead of
this linear search.

[Yang, 2015-04-16 14:18:02]

On 2015/04/16 14:13:11, Yang wrote:
> On 2015/04/16 14:04:28, Hannes Payer wrote:
> > lgtm
> 
> The serializer uses RootIndexMap for lookup. We cache an instance of it on the
> isolate. Only non-writable roots are included in the RootIndexMap, and
> test-heap/WritableVsImmortalRoots makes sure that immortal roots are
> non-writable.
> 
> Maybe it makes sense to move that to the heap to do table lookups instead of
> this linear search.

Forgot to mention: major GCs delete the hash table instance. This works because
the serializer doesn't trigger a GC. Probably does not apply in this use case,
so you may want to remove that GC part.

[titzer, 2015-04-16 14:41:03]

On 2015/04/16 14:18:02, Yang wrote:
> On 2015/04/16 14:13:11, Yang wrote:
> > On 2015/04/16 14:04:28, Hannes Payer wrote:
> > > lgtm
> > 
> > The serializer uses RootIndexMap for lookup. We cache an instance of it on
the
> > isolate. Only non-writable roots are included in the RootIndexMap, and
> > test-heap/WritableVsImmortalRoots makes sure that immortal roots are
> > non-writable.
> > 
> > Maybe it makes sense to move that to the heap to do table lookups instead of
> > this linear search.
> 
> Forgot to mention: major GCs delete the hash table instance. This works
because
> the serializer doesn't trigger a GC. Probably does not apply in this use case,
> so you may want to remove that GC part.

Acknowledged. I think we could also speed up some of the checks using a range
check on the first page. As is, this CL just moves it over to make those
optimizations easier, since it's now encapsulated in heap.

[titzer, 2015-04-16 14:41:08]

The CQ bit was checked by titzer@chromium.org

[commit-bot: I haz the power, 2015-04-16 14:41:22]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1095513003/1

[commit-bot: I haz the power, 2015-04-16 15:08:23]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-04-16 15:08:31]

Patchset 1 (id:??) landed as
https://crrev.com/cb08656b6df964b9312a5f73f61f1bdfed371cf0
Cr-Commit-Position: refs/heads/master@{#27891}