WebKit Bugzilla
New
Browse
Search+
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED INVALID
95929
renderer()->view() is null
https://bugs.webkit.org/show_bug.cgi?id=95929
Summary
renderer()->view() is null
snyh
Reported
2012-09-05 21:06:48 PDT
in the begin of function "RenderLayer::convertToLayerCoord" ----------------------------------------------------- if (position == FixedPosition && !renderer()->inRenderFlowThread() && (!ancestorLayer || ancestorLayer == renderer()->view()->layer())) { // If the fixed layer's container is the root, just add in the offset of the view. We can obtain this by calling // localToAbsolute() on the RenderView. FloatPoint absPos = renderer()->localToAbsolute(FloatPoint(), true); location += LayoutSize(absPos.x(), absPos.y()); return; } ------------------------------------------------------ renderer()->view() may be null when this layer hasn't destroy.
Attachments
Add attachment
proposed patch, testcase, etc.
Alexey Proskuryakov
Comment 1
2012-09-06 16:23:34 PDT
Do you happen to have a test case where this happens in practice?
snyh
Comment 2
2012-09-06 18:26:48 PDT
(In reply to
comment #1
)
> Do you happen to have a test case where this happens in practice?
Sorry, I can't get an simply test case. I have modify the RenderLayer, FrameView etc. If you need an test case with modify some lite WebCore's code, I can provide one at the later. But the point is use renderLayer->absoluteBoundingBox() when the time page reload with two or more div has style "z-index: 0+; position:fixed". the webcore will crash. I can sure the RenderLayer and its m_renderer hasn't destroy when doing this. The problem is at this time the m_renderer has detached from RenderView? Now I have test if frame()->document()->renderView() valid when doing other things in my code.
snyh
Comment 3
2012-09-06 18:32:36 PDT
Or in RenderLayer::convertToLayerCoord simply test renderer()->view() is valid. But I see many place invoke renderer()->view()->otherthing without test. So Is this is my fault?
Julien Chaffraix
Comment 4
2012-09-27 18:05:12 PDT
(In reply to
comment #3
)
> Or in RenderLayer::convertToLayerCoord simply test renderer()->view() is valid. > > But I see many place invoke renderer()->view()->otherthing without test. > So Is this is my fault?
Very likely. renderer()->view() is NULL during Document's tear down FWIW. If you have modified WebKit (especially some core rendering objects like RenderLayer), you are on your own. Unless this reproduces against an unmodified WebKit codebase and you provide a way for us to reproduce that, we will consider this to be a bug in your code, not WebKit.
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug