Oilpan: Remove page navigation GCs from the prologues of V8's GCs

Oilpan: Remove page navigation GCs from the prologues of V8's GCs

I've removed synchronous Oilpan's GCs from the epilogues of V8's major GCs
but it didn't fix the regression of v8_gc_total metrics caused by Oilpan's shipping.

This CL tries to remove page navigation GCs from the prologues of V8's GCs.
I confirmed that this doesn't cause regression in memory.top_7_stress.

BUG=582831
Committed: https://crrev.com/00a90c6253df77840fcf2a638cfa4346a71479be
Cr-Commit-Position: refs/heads/master@{#374395}

[haraken, 2016-02-08 04:29:51]

haraken@chromium.org changed reviewers:
+ sigbjornf@opera.com

[haraken, 2016-02-08 04:29:51]

PTAL

[haraken, 2016-02-08 04:40:07]

Results for memory.top_7_stress: http://haraken.info/a/results.html

No regression in the peak memory usaged.

[sof, 2016-02-09 14:17:22]

lgtm to try (and I sure hope this doesn't lead to more iterations of time vs
space whack-a-mole.)

[haraken, 2016-02-09 14:27:59]

The CQ bit was checked by haraken@chromium.org

[commit-bot: I haz the power, 2016-02-09 14:29:16]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1679683003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1679683003/1

[commit-bot: I haz the power, 2016-02-09 17:00:31]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-02-09 17:02:15]

Description was changed from

==========
Oilpan: Remove page navigation GCs from the prologues of V8's GCs

I've removed synchronous Oilpan's GCs from the epilogues of V8's major GCs
but it didn't fix the regression of v8_gc_total metrics caused by Oilpan's
shipping.

This CL tries to remove page navigation GCs from the prologues of V8's GCs.
I confirmed that this doesn't cause regression in memory.top_7_stress.

BUG=582831
==========

to

==========
Oilpan: Remove page navigation GCs from the prologues of V8's GCs

I've removed synchronous Oilpan's GCs from the epilogues of V8's major GCs
but it didn't fix the regression of v8_gc_total metrics caused by Oilpan's
shipping.

This CL tries to remove page navigation GCs from the prologues of V8's GCs.
I confirmed that this doesn't cause regression in memory.top_7_stress.

BUG=582831

Committed: https://crrev.com/00a90c6253df77840fcf2a638cfa4346a71479be
Cr-Commit-Position: refs/heads/master@{#374395}
==========

[commit-bot: I haz the power, 2016-02-09 17:02:16]

Patchset 1 (id:??) landed as
https://crrev.com/00a90c6253df77840fcf2a638cfa4346a71479be
Cr-Commit-Position: refs/heads/master@{#374395}

[waffles, 2016-02-09 20:49:19]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1688483002/ by waffles@chromium.org.

The reason for reverting is: Speculative revert to fix a Mac browser test that
relies on 'loadstop' being transmitted in certain contexts..

[sof, 2016-02-09 20:55:01]

On 2016/02/09 20:49:19, waffles wrote:
> A revert of this CL (patchset #1 id:1) has been created in
> https://codereview.chromium.org/1688483002/ by mailto:waffles@chromium.org.
> 
> The reason for reverting is: Speculative revert to fix a Mac browser test that
> relies on 'loadstop' being transmitted in certain contexts..

Do you have some more details?

[sof, 2016-02-09 21:13:56]

On 2016/02/09 20:55:01, sof wrote:
> On 2016/02/09 20:49:19, waffles wrote:
> > A revert of this CL (patchset #1 id:1) has been created in
> > https://codereview.chromium.org/1688483002/ by mailto:waffles@chromium.org.
> > 
> > The reason for reverting is: Speculative revert to fix a Mac browser test
that
> > relies on 'loadstop' being transmitted in certain contexts..
> 
> Do you have some more details?

Ping. There's nothing actionable based on this revert.

[sof, 2016-02-09 21:17:29]

+waffles@: could you please supply some details?

[waffles, 2016-02-09 23:26:02]

I will add details to the bug ASAP. The issue cleared with this revert, but the
same test is now failing on another bot.

[haraken, 2016-02-09 23:43:22]

On 2016/02/09 23:26:02, waffles wrote:
> I will add details to the bug ASAP. The issue cleared with this revert, but
the
> same test is now failing on another bot.

It is possible that this CL changed a GC timing and exposed the bug. However,
the GC timing has been undeterministic (in nature), so I'm not sure how helpful
reverting this CL is. Either way, this CL just exposed the existing bug. You
need to fix the bug instead of reverting this CL.