422330 - Slow scrolling performance with dotted or dashed borders

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[Oleg Romashin (:romaxa)]

Scrolling performance very slow on URL with latest trunk.

Don't know exact regression bug with FF2.0 it works fast.

Steps:
1) open URL
2) scroll UP/down with mouse wheel.

CPU 1.8GHz, 1Gb Ram, ATI fglrx, Linux.

Comment 1

[Boris Zbarsky [:bzbarsky]]

Over half the time is under nsTextFrame::PaintText.

Can you hunt down a regression range for this?  That would make it easier to tell what's going on...  It would also help to have either a minimal testcase or at least a testcase attached to this bug.

Are we invalidating/repainting too much or something?

Nominating, but I don't think this should block necessarily.  I do think it should be wanted (next or 1.9).

Comment 2

[Oleg Romashin (:romaxa)]

First range:
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2007/04/2007-04-30-04-trunk/
and
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2007/05/2007-05-01-04-trunk/

Comment 3

[Boris Zbarsky [:bzbarsky]]

Ah, border rendering.  For what it's worth, a minimal testcase would likely help quite a bit here...  Specifically, what border styles are causing the slowness?

Comment 4

[Gábor Stefanik]

Looks like dashed borders are no better than dotted ones. This page has a lot of long dashed borders, while a major border rewrite bug is slow scrolling with long dotted borders.

Comment 5

[Boris Zbarsky [:bzbarsky]]

That's a likely possibility, yes.  I can't find the bug on slow dotted borders linked from bug 368247, though... What's the number?

Comment 6

[Mike Schroepfer]

Setting to wanted based on comment 1.  Unless vlad wants to crank a fix out :-P

Comment 7

[dolphinling]

There are a bunch of bugs on this currently, none of which are perfectly clean, and this was first, so updating summary, fixing fields, etc.

(see also bug 379834, where a much worse case of this problem with a different cause was fixed)

Comment 10

[dolphinling]

Ah, shoot, didn't see bug 424423 until just now because it wasn't marked a regression from 368247. Marking this depending on that, though it's probably a dupe.

Comment 11

[Oleg Romashin (:romaxa)]

Why did you change URL on https://bugzilla.mozilla.org/attachment.cgi?id=291413 ?

Current URL works on trunk faster than on FF2...

for me better testcase http://erased.info/post15530 or 
https://bugzilla.mozilla.org/attachment.cgi?id=310196

Comment 12

[Boris Zbarsky [:bzbarsky]]

romaxa, please just attach whatever page is showing issues to the bug. Linking to sites which may not be around (or might change) tomorrow, is not a good way to get bugs fixed....

Comment 13

[dolphinling]

romaxa: Sorry, for me attachment 291413 [details] is slower on trunk than 2.0, but only a little bit so you're right it probably doesn't show the same problem. Switching url field to attachment 310196 [details], which won't disappear. 

Comment 14

[Oleg Romashin (:romaxa)]

Attachment: Divs with dashed borders

Comment 16

[Schot]

Today I installed Firefox 3 from the previous Firefox 2 stable version.  I'm
very sad to say that I'm having this same problem affecting my site:
http://darkmatters.org

There is heavy use of png which I am guessing is the catalyst to this problem. 
Like Roman this is likely a transparency issue though I can't even guess as to
why FF2 works fine but FF3 does not.

Scroll lag on my site is very severe in FF3.

Comment 17

[Boris Zbarsky [:bzbarsky]]

Schot, I'm not seeing any dotted or dashed borders on that site.  Am I just missing them?

If there aren't any, then you're NOT seeing this bug, but another bug.  You should file it (and cc me on it).

Comment 18

[Schot]

My mistake Borris.  The problem I was having was resolved by changing my background from fixed to repeat.  It seems the combination of a fixed Background + the use of transparency or png is causing the slow down/lag I had experienced.  I did end up finding the proper reported bug:
https://bugzilla.mozilla.org/show_bug.cgi?id=418948

Comment 19

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

This should be significantly better now after bug 424423 was fixed; can someone who's seeing the problem check with a nightly?

Comment 20

[Bernhard Redl]

> can someone who's seeing the problem check with a nightly?

nothing better with firefox 3.5 on ubuntu. Tested Google Chrome & Opera (scrolling = instant)

Comment 21

[Gábor Stefanik]

(In reply to comment #20)
> nothing better with firefox 3.5 on ubuntu. Tested Google Chrome & Opera
> (scrolling = instant)

Firefox 3.5 != latest nightly; the latest nightly contains Firefox 3.next (Namoroka)-specific code.

Comment 22

[Boris Zbarsky [:bzbarsky]]

"latest nightly" as of when comment 19 was made is over a year ago.  3.5 contains the patch referenced in comment 19.

Comment 23

[Bernhard Redl]

Firefox 3.5 != latest nightly;

ok sry seems that i got something wrong. Thought that this patch must have been included in 3.5. 

Do you think current nightly will show better scrolling performance?

Comment 24

[Boris Zbarsky [:bzbarsky]]

That patch was in fact included in 3.5; see comment 22.

Comment 25

[RNicoletto]

Isn't this a DUPE of bug 266702?

Comment 26

[Mike Taylor [:miketaylr]]

Attachment: slow-borders.html

Here's an example where scrolling is unusably janky on OSX, with a retina screen:

http://www.2ality.com/2015/03/es6-generators.html

When I plugged in my external non-retina monitor to reduce, the jank is almost gone entirely--but it's still not as smooth as Chrome.

There's a single CSS class in this reduced version of that blog page:

.post-outer {
border: dotted 1px #ccbb99;
box-shadow: 0 0 0 rgba(0, 0, 0, .1);
}

Comment 27

[BugBot [:suhaib / :marco/ :calixte]]

The severity field for this bug is relatively low, S3. However, the bug has 3 duplicates.
:dholbert, could you consider increasing the bug severity?

For more information, please visit auto_nag documentation.

Comment 28

[BugBot (nomail) [:suhaib / :marco/ :calixte]]

The last needinfo from me was triggered in error by recent activity on the bug. I'm clearing the needinfo since this is a very old bug and I don't know if it's still relevant.