410752 - [mlk] Memory leak with javascript : Image refresh

Status: RESOLVED WORKSFORME

(Firefox :: General, defect)

Description

[Olivier]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9b3pre) Gecko/2008010304 Minefield/3.0b3pre

I'm using a javascript to refresh an image (jpg) every second or every 500ms in order to have a video with a low framerate.

This javascript cause a memory leak on firefox. On the 3.0b3pre the leak is about 1,5MB per minute but on firefox 2.x it's about 1MB per second !

As you can see in attached file, the refreshed picture is genereated by a php file and the name of the picture is different each time to avoid the browser cache.

Reproducible: Always

Steps to Reproduce:
1. Call recursively (with setTimeout) a javascript function 
2. Refresh the src <img /> with the function
3.
Actual Results:  
Memory leak : 

Minute          Used memory(KB)
----------------------------
0                31080
1                32420
2                34156
3                36104

Expected Results:  
No memory leak :)

Comment 1

[Mardeg]

I thought proper cache http headers is the way to avoid the cache, doesn't different file names makes all the images *stay* in memory cache?

Comment 2

[Olivier]

Attachment: Javascript code. Function causing memory leak

The first function is the source of memory leak.
Other are just here for a better understanding

Comment 3

[Olivier]

(In reply to comment #1)
> I thought proper cache http headers is the way to avoid the cache, doesn't
> different file names makes all the images *stay* in memory cache?
> 

Not working with appropriated header (no-cache and expire)

Comment 4

[Jo Hermans]

dupe of bug 228233 ?

Comment 5

[Thomas Bertels]

Looks like it.

Comment 6

[Boris Zbarsky [:bzbarsky]]

Doesn't look like it to me at all.  Bug 228233 as described in bug 228233 comment 14 requires the following two to happen:

1) The data is cached
2) The new load is started before the previous load gets a response from the
   server.

Per description of this bug, neither is true.  Reopening.  This needs someone to actually debug then.

Same symptoms don't mean same causes; otherwise all crash bugs would be duplicates of each other.

What this bug needs are:

1)  An actual HTML page (not just a script fragment) showing the problem.
2)  Preferably a minimal testcase derived from said page.

Comment 7

[Olivier]

Hi,
You can find a test page here : http://www.lanforums.com/bugff/
Clic on the start button and watch your memory (on Windows XP).

I tested this on FF 2.0.0.11 and 2.0.0.7 and the problem is the same.
Nevertheless on the last FF 3 Beta the bug seems to be solved.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Comment 0 seems to say there is a problem with "Gecko/2008010304 Minefield/3.0b3pre" as well.  Is that problem gone with a newer trunk build?

To be honest, it's not likely that anything in this area will get changed in the 2.x builds; I'm primarily interested in the trunk issues, if any.

Comment 9

[Steve Chapel]

I can't reproduce with Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9b3pre) Gecko/2008011804 Minefield/3.0b3pre. Mem Usage and VM Size both go up and down a lot, but both stay under 70 MB, even if I run the http://www.lanforums.com/bugff/ test for over an hour.

Comment 10

[Olivier]

As you can see on this screenshot : 
http://www.enregistrersous.com/images2/180645505420080121100724.jpg

i have a huge amount of memory used (still growing up) and this , only after 4 minutes on http://www.lanforums.com/bugff/

Comment 11

[Steve Chapel]

As Boris said, the likelihood of this being fixed for Firefox 2 is very low. That version is getting only important stability and security fixes at this point. The problem seems to be fixed for Firefox 3.