547132 - Clip the scrollport before we perform display list analysis for scrolling

Status: RESOLVED WONTFIX

(Core :: Layout, defect)

Description

[Robert O'Callahan (:roc) (email my personal email if necessary)]

When we unify chrome and content rendering (bug 130078), suppose we have a scrollable element that is partially offscreen, i.e. it's "under" the chrome. When we perform display list analysis for scrolling that element, the scrollport is intersected by a bunch of elements in the chrome, which causes the regions to get complicated so we simplify the regions and make some suboptimal decisions.

There is a simple way to fix this: before we call ComputeRepaintRegionForCopy, we can clip the scrollport to reflect any clipping performed by ancestor frames.

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 427692 [details] [diff] [review]
fix

I was going to say that you should start from mScrollPort instead of the frame's rect... but I guess that's taken care of by the intersecting that you do the very first time through the loop.

The bit about stopping at the display root only affects popups, right?  We'll still correctly handle (post-unification of view trees?) the case of not blitting if there are SVG effects in chrome that contains a content window we're scrolling, right?

r=dbaron

Comment 3

[Robert O'Callahan (:roc) (email my personal email if necessary)]

(In reply to comment #2)
> (From update of attachment 427692 [details] [diff] [review])
> I was going to say that you should start from mScrollPort instead of the
> frame's rect... but I guess that's taken care of by the intersecting that you
> do the very first time through the loop.

Yes, I'll add a comment about that.

> The bit about stopping at the display root only affects popups, right?

Correct.

> We'll still correctly handle (post-unification of view trees?) the case of not
> blitting if there are SVG effects in chrome that contains a content window
> we're scrolling, right?

Yes, because the displayroot is the root frame of the chrome window in that case.

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I might obsolete this with retained layer scrolling before we actually need it, so I'll hold off on landing it for now.

Comment 5

[Timothy Nikkel (:tnikkel)]

Obsolete now?

Comment 6

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Yes!