Oilpan: Move ChromeClient classes into Oilpan heap.

Oilpan: Move ChromeClient classes into Oilpan heap.

This touches a lot of classes including core ones, but each change should be
mostly straightforward.

As a side effect, Page::PageClients is marked as STACK_ALLOCATED(). A couple
of use sites are changed so PageClients is allocated on stack. This mutation
is safe because the members of PageClients are copied on the creation of
a Page, thus a PageClients object doesn't have to outlive the corresponding
Page object.

This big fuss fixes just one raw reference instance in InspectorOverlayImpl.

BUG=509911
R=haraken@chromium.org, keishi@chromium.org, oilpan-reviews@chromium.org

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=200640

[haraken, 2015-08-13 10:52:06]

Is there any reason we can't ship oilpan for ChromeClient by default?

https://codereview.chromium.org/1291903003/diff/1/Source/core/loader/EmptyCli...
File Source/core/loader/EmptyClients.cpp (right):

https://codereview.chromium.org/1291903003/diff/1/Source/core/loader/EmptyCli...
Source/core/loader/EmptyClients.cpp:48: static ChromeClient* dummyChromeClient =
EmptyChromeClient::create().leakPtr();

Not related to this CL, it would be better to use DEFINE_STATIC_LOCAL.

[Yuta Kitamura, 2015-08-14 04:40:21]

On 2015/08/13 10:52:06, haraken wrote:
> Is there any reason we can't ship oilpan for ChromeClient by default?

BlinkGCPlugin will complain for a raw pointer field of ChromeClient*
in Page::PageClients.

With that being said, this patch fixes a raw pointer in one place
but creates another... Should Page::PageClients be converted
together?

[Yuta Kitamura, 2015-08-14 04:40:38]

https://codereview.chromium.org/1291903003/diff/1/Source/core/loader/EmptyCli...
File Source/core/loader/EmptyClients.cpp (right):

https://codereview.chromium.org/1291903003/diff/1/Source/core/loader/EmptyCli...
Source/core/loader/EmptyClients.cpp:48: static ChromeClient* dummyChromeClient =
EmptyChromeClient::create().leakPtr();
On 2015/08/13 10:52:06, haraken wrote:
> 
> Not related to this CL, it would be better to use DEFINE_STATIC_LOCAL.

Done.

[haraken, 2015-08-14 04:41:07]

On 2015/08/14 04:40:21, Yuta Kitamura wrote:
> On 2015/08/13 10:52:06, haraken wrote:
> > Is there any reason we can't ship oilpan for ChromeClient by default?
> 
> BlinkGCPlugin will complain for a raw pointer field of ChromeClient*
> in Page::PageClients.
> 
> With that being said, this patch fixes a raw pointer in one place
> but creates another... Should Page::PageClients be converted
> together?

Sounds nicer.

[Yuta Kitamura, 2015-08-14 12:16:39]

On 2015/08/14 04:41:07, haraken wrote:
> On 2015/08/14 04:40:21, Yuta Kitamura wrote:
> > On 2015/08/13 10:52:06, haraken wrote:
> > > Is there any reason we can't ship oilpan for ChromeClient by default?
> > 
> > BlinkGCPlugin will complain for a raw pointer field of ChromeClient*
> > in Page::PageClients.
> > 
> > With that being said, this patch fixes a raw pointer in one place
> > but creates another... Should Page::PageClients be converted
> > together?
> 
> Sounds nicer.

Done (PageClients is now STACK_ALLOCATED()).

It's turned out that it's impossible to ship Oilpan for ChromeClient
unconditionally, because Page contains a pointer to it.

So the patch set 5 is ready for review, PTAL.

[haraken, 2015-08-14 14:38:41]

LGTM

https://codereview.chromium.org/1291903003/diff/80001/Source/web/InspectorOve...
File Source/web/InspectorOverlayImpl.cpp (right):

https://codereview.chromium.org/1291903003/diff/80001/Source/web/InspectorOve...
Source/web/InspectorOverlayImpl.cpp:155: ChromeClient& m_client;

This needs to be a RawPtrWillBeMember. It is dangerous to leave a raw pointer
that is not explicitly cleared when the pointing object dies.

[Yuta Kitamura, 2015-08-17 08:03:22]

https://codereview.chromium.org/1291903003/diff/80001/Source/web/InspectorOve...
File Source/web/InspectorOverlayImpl.cpp (right):

https://codereview.chromium.org/1291903003/diff/80001/Source/web/InspectorOve...
Source/web/InspectorOverlayImpl.cpp:155: ChromeClient& m_client;
On 2015/08/14 14:38:41, haraken wrote:
> 
> This needs to be a RawPtrWillBeMember. It is dangerous to leave a raw pointer
> that is not explicitly cleared when the pointing object dies.

Ouch, this was the one I intended to fix, but has slipped off
during several refactorings...

Done.

[Yuta Kitamura, 2015-08-17 09:43:06]

The CQ bit was checked by yutak@chromium.org

[Yuta Kitamura, 2015-08-17 09:43:06]

The patchset sent to the CQ was uploaded after l-g-t-m from haraken@chromium.org
Link to the patchset: https://codereview.chromium.org/1291903003/#ps100001
(title: "Fix a raw reference in InspectorOverlayImpl.cpp.")

[commit-bot: I haz the power, 2015-08-17 09:43:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1291903003/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1291903003/100001

[commit-bot: I haz the power, 2015-08-17 10:51:18]

Committed patchset #6 (id:100001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=200640