Reland "[heap] Refactor evacuation for young and old gen into visitors."

Reland "[heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N

Committed: https://crrev.com/f0f707d8326cd8e0135c1e7a8bbb9893e6d04510
Cr-Commit-Position: refs/heads/master@{#32617}

[Michael Lippautz, 2015-12-04 10:55:44]

Description was changed from

==========
Reland "heap] Refactor evacuation for young and old gen into visitors."

BUG=
==========

to

==========
Reland "heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N
==========

[Michael Lippautz, 2015-12-04 11:24:16]

Patchset #2 (id:20001) has been deleted

[Michael Lippautz, 2015-12-04 11:29:35]

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

[Michael Lippautz, 2015-12-04 11:29:35]

PTAL. This is the CL that triggered lots of aborted pages.

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3391: AlwaysAllocateScope always_allocate(isolate());
I am open for suggestions here:
- I'd rather not add it to the visitor constructor, as we soon will have a
newspace vistor in the same scope (for parallel evacuation of all generations).
This could mess up the semantics again.
- Adding it to the actual visiting function of a single object would mean that
we have an atomic inc/dec for visiting each object. Maybe this is good enough?
- We could also add it to IterateLiveObjectsOnPage (depending on a parameter).
Also not ideal...

Adding it here adds the overhead on a page level.

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3401: p->ResetLiveBytes();
Slight inconsistency in the old refactoring. We always reset the live bytes
(although it should not matter).

[Michael Lippautz, 2015-12-04 11:34:37]

Description was changed from

==========
Reland "heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N
==========

to

==========
Reland "[heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N
==========

[Hannes Payer (out of office), 2015-12-04 12:00:41]

lgtm

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3391: AlwaysAllocateScope always_allocate(isolate());
On 2015/12/04 11:29:35, Michael Lippautz wrote:
> I am open for suggestions here:
> - I'd rather not add it to the visitor constructor, as we soon will have a
> newspace vistor in the same scope (for parallel evacuation of all
generations).
> This could mess up the semantics again.
> - Adding it to the actual visiting function of a single object would mean that
> we have an atomic inc/dec for visiting each object. Maybe this is good enough?
> - We could also add it to IterateLiveObjectsOnPage (depending on a parameter).
> Also not ideal...
> 
> Adding it here adds the overhead on a page level.

Acknowledged.

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3401: p->ResetLiveBytes();
On 2015/12/04 11:29:35, Michael Lippautz wrote:
> Slight inconsistency in the old refactoring. We always reset the live bytes
> (although it should not matter).

Careful here, he still want to account for live objects on the aborted page.

[Michael Lippautz, 2015-12-04 13:29:49]

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1499893002/diff/40001/src/heap/mark-compact.c...
src/heap/mark-compact.cc:3401: p->ResetLiveBytes();
On 2015/12/04 12:00:41, Hannes Payer wrote:
> On 2015/12/04 11:29:35, Michael Lippautz wrote:
> > Slight inconsistency in the old refactoring. We always reset the live bytes
> > (although it should not matter).
> 
> Careful here, he still want to account for live objects on the aborted page. 

Done.

[Michael Lippautz, 2015-12-04 13:33:42]

The CQ bit was checked by mlippautz@chromium.org

[Michael Lippautz, 2015-12-04 13:33:42]

The patchset sent to the CQ was uploaded after l-g-t-m from hpayer@chromium.org
Link to the patchset: https://codereview.chromium.org/1499893002/#ps60001
(title: "Moved resetting live bytes back")

[commit-bot: I haz the power, 2015-12-04 13:33:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1499893002/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1499893002/60001

[commit-bot: I haz the power, 2015-12-04 14:19:27]

Description was changed from

==========
Reland "[heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N
==========

to

==========
Reland "[heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N
==========

[commit-bot: I haz the power, 2015-12-04 14:19:28]

Committed patchset #3 (id:60001)

[commit-bot: I haz the power, 2015-12-04 14:19:37]

Description was changed from

==========
Reland "[heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N
==========

to

==========
Reland "[heap] Refactor evacuation for young and old gen into visitors."

Create a visitor for evacuating objects for young and old generation. This is
the first step of preparing a task to process, both,  newspace and oldspace
pages in parallel.

BUG=chromium:524425
LOG=N

Committed: https://crrev.com/f0f707d8326cd8e0135c1e7a8bbb9893e6d04510
Cr-Commit-Position: refs/heads/master@{#32617}
==========

[commit-bot: I haz the power, 2015-12-04 14:19:38]

Patchset 3 (id:??) landed as
https://crrev.com/f0f707d8326cd8e0135c1e7a8bbb9893e6d04510
Cr-Commit-Position: refs/heads/master@{#32617}