Closed Bug 561714 Opened 14 years ago Closed 14 years ago

Continuous memory leak while idle after upgrading from 3.6.3 to lorentz/3.6.4 beta 1

Categories

(Firefox :: General, defect)

Product:
Firefox
Component:
General
Platform:
x86
Windows 7
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: asmodai, Unassigned)

Details

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv:1.9.2.4) Gecko/20100413 Firefox/3.6.4 (.NET CLR 3.5.30729)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv:1.9.2.4) Gecko/20100413 Firefox/3.6.4 (.NET CLR 3.5.30729)

Using Firefox 3.6.4 (beta 1) I am experiencing a continuous memory leak.

I have a fair number of tabs open, about 50. This takes up about 400 MB when I start Firefox. From there on I see, if I don't do anything with Firefox, the memory consumption (private working set in Windows task manager) increase in increments of 20 KB - 600 KB. Typically it is something like 70 MB in 5 minutes. This keeps on growing until the memory allocation reaches something like 1.5-1.6 GB on my laptop at which time it crashes and the crash reporting tool pops up.

Reproducible: Always

Steps to Reproduce:
I am not sure yet how to reproduce it. It started after an upgrade of Firefox 3.6.3 to Lorentz/3.6.4 beta 1.
Actual Results:  
Memory keeps on increasing until finally it goes out of memory and crashes.

Expected Results:  
Stable memory consumption.

I tried with all plugins and addons disabled, but it still kept on leaking.
It seems that http://site.icu-project.org/ is causing quite some leaking for me. I am still experiencing some leakage, but at a *much* reduced rate.
To explain some more. I am still starting up with that slew of tabs open and was shutting down one at a time to see which one(s) would flatten out the memory behaviour. The above URL, after trying a few times to close and open it kept on increasing the memory allocated. I am guessing it might be a JavaScript leak, but I am not sure and definitely not an expert in chasing those.
Talking this over with Henri Sivonen we managed to track it to interaction between that page and html5.enable=true. Setting html5.enable back to false for 3.6.4 and reloading the page in question keeps the memory consumption stable.
See also bug 538722.

Still need to check if this happens on trunk.
The support forum (https://support.mozilla.com/en-US/forum/1/626951#threadId660969) has some comments about Windows 7 and XP SP3 leaks for Firefox 3.6.2 and 3.6.3.  The leak is the same - do some work, wait, and eventually the memory usage (and processor usage) starts to creep up until the whole thing grinds to a halt.  Kill Firefox and the problem clears.  Disabling all the Java Console Add-ons except the latest seems to help a lot.  Most reports indicate AVG 9.0 is running but I would not know if that is relevant.  This problem seems to have been introduced in a late 3.5 version or perhaps 3.6.0.  I also have Adblock Plus 1.1.3 Add-on active.
This issue is still occurring in 3.7pre-alpha release. The process starts running at about 50mb, then grows. It seems to be holding better for me (not continually growing), but it peaked for me at a few hundred mb with only 6 or 7 tabs open
Stop bothering developers with this minor problem, you are the only one, who has it.
Still need to check if this happens on trunk &&&&&
Stop bothering developers with this minor problem, you are the only one, who has it.
Stop bothering developers with this minor problem, you are the only one, who has it
Stop bothering developers with this minor problem, you are the only one, who has it
excuses for comments 7-11, some is spamming various bugs, see bug 567347
It seems that recent 3.6 versions have removed html5.enable=True to do anything. At least, I was able to display HTML5 and MathML with 3.6.4 and with 3.6.10 at least it doesn't do that anymore. So it might very well be that this memory leak cannot be triggered anymore. I, at least, haven't been able to replicate the issue anymore.

So for me this issue can be closed, but not sure what you guys want to do with it.
Lets be nice and close this since I cannot reproduce any of the problems with recent 3.6 versions.
Status: UNCONFIRMED → RESOLVED
Closed: 14 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.