538611 - setTimeout() very slow when using onmousemove event - Linux optimisation needed?

Status: RESOLVED WORKSFORME

(Core :: DOM: Events, defect, P3)

Description

[Evengard]

User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; ru; rv:1.9.1.6) Gecko/20091216 Iceweasel/3.5.6 (like Firefox/3.5.6; Debian-3.5.6-1)
Build Identifier: Mozilla/5.0 (X11; U; Linux x86_64; ru; rv:1.9.1.6) Gecko/20091216 Iceweasel/3.5.6 (like Firefox/3.5.6; Debian-3.5.6-1)

On the site http://forum.noxworld.ru there is a script which creates some effects to the mouse cursor. It is using the document.onmousemove event and a self-callable function (for updating the effect with given speed) which calls itself via setTimeout. This function updates the DOM (some CSS settings of some elements).
The problem is that then I move the mouse - the effects slows down like if setTimeout was given a bigger timeout!
It is less noticeable in safe mode, but it is twice slower than the same on WebKit (epiphany-browser), or Opera, or Chrome, or even Konqueror with it's KHTML!
So if possible please speed such effects up... Tracemonkey should be the fastest engine of the world, no?
NOTE: when not moving the cursor after the efects appeared they render smoothly... So it is a probably event-related misfunction...

Reproducible: Always

Steps to Reproduce:
1. Go to http://forum.noxworld.ru
2. Move the mouse - notice the speed of the effects
3. Do the same on an another nonGecko browser - notice the difference
Actual Results:  
In Gecko the effects are slow when moving the mouse, in other browsers - it is smooth

Expected Results:  
It should be smooth on Gecko too

I heard that Windows users don't get such problem, so it is probably a Linux-only slowdown? Optimisation for Linux needed!

Comment 1

[Evengard]

Very strange... When the page is loading - all is VERY smooth... When the page stops loading - it became buggy! Why?...

Comment 2

[Evengard]

NOTE: all my tests are done in --safe-mode
Later I will try to make a kind of videoshowing how smooth is it when page is loading and what happens next

Comment 3

[Evengard]

Erhm... The kernel update solved this problem... It is now just fine...

Comment 4

[Evengard]

Apparently it was the IO scheduler problem as well as a lot of other problems...

Comment 5

[Evengard]

Reopening the bug 'cause in Konqueror and other engines it is still ALOT smoother...

Comment 6

[Evengard]

Another example of slow running JavaScript (not related with events) is Google Waves...
Would opt of renaming this bug to simply "Linux Optimisation Needed"

Comment 7

[Wayne Mery (:wsmwk)]

Evengard is this in an X environment? can you create a reduced testcase and attach to the bug?

Comment 8

[Wayne Mery (:wsmwk)]

Evengard thinks this no longer exists.