450350 - CSS background image painting chooses tiled vs non-tiled path based on dirty area

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

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

Attachment: fix

See bug 421438. Tiled and non-tiled code paths draw the same tile in different ways. However, the code in PaintBackgroundSC decides which path to take based on the dirty area. I don't have a testcase, but this could definitely lead to repainting artifacts in certain cases. It's easy enough to fix, especially now that bug 449324 is fixed.

Comment 1

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

It seems to me like it would be a bug if these codepaths produced different results in different cases.  (I suspect it's probably true for at least some zoomed cases, though.)

Comment 2

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

See bug 421438 comment #1, item 3. I don't think making DrawTile and DrawImage fully consisistent is an option.

Comment 3

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

Comment on attachment 333496 [details] [diff] [review]
fix

ok, r+sr=dbaron.

I think you want borderAreaOriginSnapped rather than aBorderArea.TopLeft(), though.  (I'm not completely sure... I could think about it more if you want.)

Comment 4

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

Yeah, I do want borderAreaOriginSnapped because that's what sourceRect, absTileRect and hence the current tiled-path test depend on.

Comment 5

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

Pushed 1d6f5b3d7ec3.

Comment 6

[Szabolcs Horvát]

Websites such as 

http://www.cssmagazine.it/

and

http://www.98rock.com/cc-common/new/6/indexrock.html

scroll much more slowly in the 2008-08-26 nightly build than in the 2008-08-25 one or in Firefox 3.0.1.  In addition to slow scrolling, some page elements (like the Google ads on cssmagazine.it) lag behind while scrolling.

Is it possible that this regression was caused by this check-in?

(Note: I tried this on WinXP.)

Comment 7

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

Yes. But I'd appreciate it if you filed a new bug with a reduced testcase.