Ensure the permissions DOMTokenList is initialized before access

Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This caused the
code not to work correctly in cases where the attribute is initially accessed via JS.

BUG=614608
Committed: https://crrev.com/c94140a3b3e85c7fb622a75338ec4c60e3425dd9
Cr-Commit-Position: refs/heads/master@{#408305}

[raymes, 2016-07-27 05:59:24]

raymes@chromium.org changed reviewers:
+ esprehn@chromium.org

[esprehn, 2016-07-27 06:05:57]

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/html/HTMLIFrameElement.cpp (right):

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/html/HTMLIFrameElement.cpp:66: if
(!initializePermissionsAttribute())
How can this fail? Does the spec say it can be null?

I'm confused if null below can cause a crash why is returning null ever okay
here?

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/html/HTMLIFrameElement.h (right):

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/html/HTMLIFrameElement.h:43: DOMTokenList*
permissions();
This is logically const, I'd const cast in the method. That it's lazy allocated
is a detail that's not relevant to users.

[raymes, 2016-07-27 06:08:09]

Description was changed from

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused a crash in some cases.

BUG=614608
==========

to

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the code not to work correctly in cases where the attribute is initially
accessed via JS.

BUG=614608
==========

[raymes, 2016-07-27 06:08:19]

Description was changed from

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the code not to work correctly in cases where the attribute is initially
accessed via JS.

BUG=614608
==========

to

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the 
code not to work correctly in cases where the attribute is initially accessed
via JS.

BUG=614608
==========

[raymes, 2016-07-27 06:13:45]

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/html/HTMLIFrameElement.cpp (right):

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/html/HTMLIFrameElement.cpp:66: if
(!initializePermissionsAttribute())
On 2016/07/27 06:05:56, esprehn wrote:
> How can this fail? Does the spec say it can be null?
> 
> I'm confused if null below can cause a crash why is returning null ever okay
> here?

Sorry my description was inaccurate. It doesn't actually crash - it just doesn't
work properly if the attribute is initially accessed via JS. I've updated the
description. It did crash in the new test I added (without this change) because
it assumes this will be non-null.

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/html/HTMLIFrameElement.h (right):

https://codereview.chromium.org/2191443002/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/html/HTMLIFrameElement.h:43: DOMTokenList*
permissions();
On 2016/07/27 06:05:56, esprehn wrote:
> This is logically const, I'd const cast in the method. That it's lazy
allocated
> is a detail that's not relevant to users.

Done.

[esprehn, 2016-07-27 06:32:48]

lgtm

[raymes, 2016-07-27 23:44:51]

The CQ bit was checked by raymes@chromium.org

[commit-bot: I haz the power, 2016-07-27 23:45:01]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-28 01:21:56]

Description was changed from

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the 
code not to work correctly in cases where the attribute is initially accessed
via JS.

BUG=614608
==========

to

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the 
code not to work correctly in cases where the attribute is initially accessed
via JS.

BUG=614608
==========

[commit-bot: I haz the power, 2016-07-28 01:21:57]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-07-28 01:23:59]

Description was changed from

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the 
code not to work correctly in cases where the attribute is initially accessed
via JS.

BUG=614608
==========

to

==========
Ensure the permissions DOMTokenList is initialized before access

The permissions DOMTokenList in the HTMLIFrameElement is lazily initialized but
we weren't initializing it before it was accessed through its accessor. This
caused the
code not to work correctly in cases where the attribute is initially accessed
via JS.

BUG=614608

Committed: https://crrev.com/c94140a3b3e85c7fb622a75338ec4c60e3425dd9
Cr-Commit-Position: refs/heads/master@{#408305}
==========

[commit-bot: I haz the power, 2016-07-28 01:24:00]

Patchset 2 (id:??) landed as
https://crrev.com/c94140a3b3e85c7fb622a75338ec4c60e3425dd9
Cr-Commit-Position: refs/heads/master@{#408305}