810275 - Scrolling on some sites isn't smooth

Status: VERIFIED FIXED

(Core :: Layout, defect)

Description

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

Go to http://myanimelist.net/animelist/Oosran&status=7&order=0 and start scrolling. Scrolling with mouse isn't smooth and feels very heavy. Scrolling with scrollbar use 100% CPU core power.

Comment 1

[Mayank Bansal]

The page is divided into boxes. When scrolling, each box constantly repaints.

Heres a profile :
http://people.mozilla.com/~bgirard/cleopatra/#report=97a2dbcd2062b59f6aa46e92f40b2a46489560d6&mergeUnbranched=true&invertCallback=true&selection=%255B%2522(total)%2522%252C%2522NtGdiBitBlt%2522%255D

Comment 2

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

Additional info:
It not happens with HW Acc disabled

Comment 3

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

This bug seems pretty bad.

Comment 4

[Matt Woodrow (:mattwoodrow)]

Attachment: Compensate for the offset when computing activeScrolledRootTopLeft

The primary problem here is that the page has a fixed background image, but it isn't on the canvas frame. We don't layerize that, so we have to repaint everything when we scroll.

The patch fixes a smaller bug, but doesn't help on this example since we the image invalidates the same pixels.

Comment 5

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

I'm confused about why this is a regression, then. We've always repainted the world for such pages.

Comment 6

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

We should profile painting performance here. The page isn't complicated, we should be able to paint it pretty fast with D2D, and that's probably the right way to fix this bug.

Comment 7

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/58ce0ddf7611

I suspect that the large amount of nsDisplayOpacity objects here will hurt d2d since we have to flush every time.

Comment 8

[Ed Morley [:emorley]]

Push backed out for reftest failures:
https://tbpl.mozilla.org/?tree=Mozilla-Inbound&rev=d30b99cfd933&jobname=%20reftest

https://hg.mozilla.org/integration/mozilla-inbound/rev/dd00a3bad021

Comment 9

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/4d2479573ecd

Comment 10

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/4d2479573ecd

Comment 11

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

FYI - This patch doesn't fix main problem like Matt Woodrow (:mattwoodrow) mention. Scrolling on this page seems better with patch, but unfortunately it's still feels heavy compared to disabled HW Acc.

Comment 12

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

Can someone with privileges close this as RESOLVED FIXED?

Comment 13

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

with pleasure!

Comment 14

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

Thank you! Marking as VERIFIED as I can't reproduce the main problem anymore.