Remove the weak list of views from array buffers

Remove the weak list of views from array buffers

Instead, views have to check their array buffer for whether
it's neutered or not.

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

Committed: https://crrev.com/5ae083a05a6743d6cb91585f449539f7846a5d8c
Cr-Commit-Position: refs/heads/master@{#27995}

[jochen (gone - plz use gerrit), 2015-04-17 12:49:13]

depends on https://codereview.chromium.org/1092923002 so tests pass

[Toon Verwaest, 2015-04-17 13:25:55]

lgtm

https://codereview.chromium.org/1094863002/diff/1/src/accessors.cc
File src/accessors.cc (left):

https://codereview.chromium.org/1094863002/diff/1/src/accessors.cc#oldcode89
src/accessors.cc:89: }
Just removing this without adding new support for loading the length from typed
arrays is going to tank gb performance by ~5%. I don't feel strongly about
implementing it in *this* CL, but we should repair the loss.

https://codereview.chromium.org/1094863002/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1094863002/diff/1/src/hydrogen.cc#newcode9928
src/hydrogen.cc:9928: HValue* was_neutered_test =
For now this is fine by me, but we should at least get data on the alternative
approach that only has a single branch by folding it into the boundscheck.

https://codereview.chromium.org/1094863002/diff/1/src/objects-inl.h
File src/objects-inl.h (right):

https://codereview.chromium.org/1094863002/diff/1/src/objects-inl.h#newcode4383
src/objects-inl.h:4383: !JSArrayBuffer::cast(view->buffer())->was_neutered()) {
What about a nicer method to avoid duplicating this all over the place?

[jochen (gone - plz use gerrit), 2015-04-17 13:32:05]

yeah, the plan is to start improving performance again once we got rid of the
weak list between array buffers as well as the polymorphism

[jochen (gone - plz use gerrit), 2015-04-20 14:15:42]

The CQ bit was checked by jochen@chromium.org

[jochen (gone - plz use gerrit), 2015-04-20 14:15:42]

The patchset sent to the CQ was uploaded after l-g-t-m from
verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1094863002/#ps20001
(title: "updates")

[commit-bot: I haz the power, 2015-04-20 14:16:01]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1094863002/20001

[commit-bot: I haz the power, 2015-04-20 14:47:04]

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

[commit-bot: I haz the power, 2015-04-20 14:47:05]

Try jobs failed on following builders:
  v8_linux_dbg on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_dbg/builds/4078)

[jochen (gone - plz use gerrit), 2015-04-22 07:22:28]

The CQ bit was checked by jochen@chromium.org

[jochen (gone - plz use gerrit), 2015-04-22 07:22:28]

The patchset sent to the CQ was uploaded after l-g-t-m from
verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1094863002/#ps40001
(title: "updates")

[commit-bot: I haz the power, 2015-04-22 07:22:42]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1094863002/40001

[jochen (gone - plz use gerrit), 2015-04-22 07:37:05]

The CQ bit was unchecked by jochen@chromium.org

[Hannes Payer (out of office), 2015-04-22 07:45:53]

LGTM, few nits

https://codereview.chromium.org/1094863002/diff/40001/src/api.cc
File src/api.cc (right):

https://codereview.chromium.org/1094863002/diff/40001/src/api.cc#newcode6343
src/api.cc:6343: if (!source) return 0;
Isn't that more an assert? source is not null when not neutered?

https://codereview.chromium.org/1094863002/diff/40001/src/heap/heap.h
File src/heap/heap.h (left):

https://codereview.chromium.org/1094863002/diff/40001/src/heap/heap.h#oldcode...
src/heap/heap.h:2180: bool migration_failure_;
Since we are not using the list of new views anymore, migration_failure_ and
previous_migration_failure_ can be deleted.

https://codereview.chromium.org/1094863002/diff/40001/test/cctest/test-weakty...
File test/cctest/test-weaktypedarrays.cc (right):

https://codereview.chromium.org/1094863002/diff/40001/test/cctest/test-weakty...
test/cctest/test-weaktypedarrays.cc:65: #if 0
Do the tests still make sense? If not, just remove them and get rid of #if 0. To
find all the views you would need to iterate the heap...

[jochen (gone - plz use gerrit), 2015-04-22 08:21:42]

https://codereview.chromium.org/1094863002/diff/40001/src/api.cc
File src/api.cc (right):

https://codereview.chromium.org/1094863002/diff/40001/src/api.cc#newcode6343
src/api.cc:6343: if (!source) return 0;
On 2015/04/22 at 07:45:53, Hannes Payer wrote:
> Isn't that more an assert? source is not null when not neutered?

true. if it's neutered, byte_length() above will be zero, and so we shouldn't
get here at all

turning this into an assert() doesn't really help, because the memcpy will just
crash as well

https://codereview.chromium.org/1094863002/diff/40001/test/cctest/test-weakty...
File test/cctest/test-weaktypedarrays.cc (right):

https://codereview.chromium.org/1094863002/diff/40001/test/cctest/test-weakty...
test/cctest/test-weaktypedarrays.cc:65: #if 0
On 2015/04/22 at 07:45:53, Hannes Payer wrote:
> Do the tests still make sense? If not, just remove them and get rid of #if 0.
To find all the views you would need to iterate the heap...

we'd need a way to iterate the entire heap (or at least new space and old space)
to find the views. with that, the tests would make sense again.

[jochen (gone - plz use gerrit), 2015-04-22 08:36:35]

The CQ bit was checked by jochen@chromium.org

[jochen (gone - plz use gerrit), 2015-04-22 08:36:36]

The patchset sent to the CQ was uploaded after l-g-t-m from hpayer@chromium.org,
verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1094863002/#ps100001
(title: "updates")

[commit-bot: I haz the power, 2015-04-22 08:36:45]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1094863002/100001

[commit-bot: I haz the power, 2015-04-22 09:19:43]

Committed patchset #6 (id:100001)

[commit-bot: I haz the power, 2015-04-22 13:19:13]

Patchset 6 (id:??) landed as
https://crrev.com/5ae083a05a6743d6cb91585f449539f7846a5d8c
Cr-Commit-Position: refs/heads/master@{#27995}

[jochen (gone - plz use gerrit), 2015-04-22 15:03:02]

A revert of this CL (patchset #6 id:100001) has been created in
https://codereview.chromium.org/1061753008/ by jochen@chromium.org.

The reason for reverting is: I'm reverting this while working on the regression
fix.