Can you help figure out source of Firefox 4.0.1 memory leak?

[VanillaMozilla]

All right guys, people at overclockers forums have used my private messaged links to test the Firefox 4.0.1 Memory Leak and one by one they are all confirming it, they have been able to crash Firefox 4.0.1!

Using all available virtual memory might be a bug if it happens and it's possible to avoid it, but it's still not a memory leak. How that event is handled depends on how your system is set up, but I forget the details. I've actually tested Fx for this in the past. My system warns you that resources are exhausted, and then I think the application just stops.

VanillaMozilla since you marked my report invalid, I sent you a private message about the details so you can see why I couldn't post the links.

Your message apparently included a memory monitor to install on my computer. Sorry, no, I'll use Task Manager. You can probably test just by viewing some local images instead of a Web site. I had a link to a Web site I used to use for this, but I don't think I have it any more.

[VanillaMozilla]

I just tested by viewing local .jpg files. I loaded 32 images totaling 480 MB uncompressed, into a computer with 384 MB RAM. It handled it beautifully. I watched it periodically deallocating RAM as it should. (You need to watch VM in the Windows Task Manager.) At one point it jumped from 342 to 163 MB. The graphical swap file display showed a history of deallocating memory as expected.

I'm watching it right now. A moment ago I had the 32 images loaded in separate tabs, and total memory usage was 76,688 K (about 76 MB). Virtual memory, which is essentially the total allocated, was only 150,536 K. I just closed all those tabs, and VM quickly dropped to 82,172 K.

I've seen many claims such as yours over the years. I've worked long and hard trying to duplicate such problems, but whenever I've tried, Firefox has always performed correctly. If you still think you have evidence of a problem, you have to describe it accurately and prove it.

[c627627]

VanillaMozilla you have a private message.

[VanillaMozilla]

C627627, if you report a bug about a memory leak, you need to first find out what a memory leak is. Wikipedia probably has an article on it. High memory use is not the same as a memory leak.

If you have evidence of some other kind of bug, by all means report it, but make sure the report is complete and accurate. But first you should consider discussing it on the MozillaZine Firefox Bugs forum. Do be sure to read the announcement at the top of the forum, which includes, among other matters, searching Bugzilla before reporting. In the case of a crash due to exhausted resources, it would probably help your case if you make sure you understand exactly what is supposed to happen and what your system settings should be regarding the swap file.

[c627627]

I will follow procedure and protocol to file/reclassify/resubmit a report that includes a description of Firefox 4.0.1 eventually crashing after not releasing back the amount of RAM it consumes.

Every single person who followed my private messaged links was able to replicate this problem, they either crashed Firefox 4.0.1 or watched it chew up over 2GB of RAM. Clearly the way forward is to establish if you can replicate this Firefox 4.0.1 flaw or not. You can do so by right-click opening the links I sent you in separate TABs then simply watch your memory usage sky rocket. The more links you open, the faster your Firefox 4.0.1 will crash.


If and when you do that - we can discuss alternate, more appropriate ways to have everyone else publicly replicate it. My bug report here
https://bugzilla.mozilla.org/show_bug.cgi?id=657759
will also be modified or resubmitted of course.

The links will not crash Firefox 3.6 or any other browser. They will release back the consumed RAM if any other browser opens identical links. Only Firefox 4.0.1 has this problem.

[VanillaMozilla]

Be sure to publish minimum steps to duplicate the problem. I would suggest using local images rather than those Web sites. Be careful to distinguish between an actual crash and simply consuming all available resources. Those images have to be stored somewhere, and if you have exhausted all available RAM or swap file, the operation will always fail (this is true of ANY computer program). If this is what is happening, be sure you state precisely what you think the bug is.

[c627627]

VanillaMozilla, were you able to either crash Firefox 4 or watch it continue to consume RAM by simply right click opening the links I sent you in new TABs?

After you crash Firefox 4, repeat the same using Firefox 3.6 or any other browser so you can conclude that *only* Firefox 4 is affected by this thereby conclusively proving that the problem is unique to Firefox 4 and could not be explained away as stated.


I invite anyone reading this to Private Message me for Firefox 4 crashing links as I unfortunately cannot post them publicly.

[VanillaMozilla]

I already told you, it deallocates memory perfectly using local files. It may handle remote images differently, but I'm not particularly inclined to pursue it further. As far as I can tell from the information you've provided, you're just exhausting all resources, which is not a bug (is this getting repetitive?). It's possible that some other browsers handle this by compressing images on the fly, but that's a high-level design decision.

Your task is not to get it past me in this forum, but to convince someone to fix it. Provide an accurate description of how to duplicate the problem and what happens, and you might have a chance. I've already given you some steps to take that are not mandatory, but that might help your case. I can't help you further.

[c627627]

Very well I understand.

I also believe that you understand that the complaint here is that a dozen pages opening a dozen 2MB images each, can crash the latest Firefox 4. I am surprised that you would classify a browser which does that as simply handling image compression differently. It certainly is not indicative of a browser which deallocates memory perfectly.

I understand you are not in a position at least at this time to confirm this problem and I will try to get other people elsewhere to confirm it. Hopefully I will find links which can be publicly posted to use as conclusive proof that there is a major problem with Firefox 4.0.1.

EDIT: All right never mind: Here's how to crash Firefox 4.0.1 viewtopic.php?f=38&t=2204639

[VanillaMozilla]

Those are not 2 MB images. Uncompressed, they are much larger. You are attempting to store between 6 GB and 12 GB of data in memory, including RAM and Windows swap file. Maybe more. I am relying on your own numbers, which are vague. If your computer can't do that, then neither can any program you can run on it.

Storing these images in compressed form was considered at one time, but the idea was not implemented, for whatever reason. Firefox performs splendidly for viewing images on my system, but if you just have to view 12 GB of pictures simultaneously, maybe you should use a browser that's optimized for this sort of thing, and accept whatever penalty or benefit that entails.

Do what you want with respect to a bug report, but I can just about guarantee you will get nowhere if you are simply terminating the program by deliberately exceeding the bounds of your computer.

[c627627]

Whether other browsers can open 50% more pages or 80% more pages than Firefox 4 can doesn't matter. It is a significantly higher percentage difference in favor of other browsers that should not exists and is indicative of a flaw in Firefox 4.


The number of pages opened vary by system but if you can open 10 pages on Firefox 3.6 and other browsers but cannot open those same 10 pages on Firefox 4, using that same computer system, then you can explain it away but the bottom line is that Firefox 4 should be able to open as many pages as every other browser can.

[squeaky]

I've just switched to a new laptop with win7 on board, downloaded FF4, and have copied my profile across. Unlike my old laptop (see earlier posts) this one is not showing a steady climb in mem usage while sitting "idle".

Hmmm... ?

[VanillaMozilla]

Squeaky, there are other differences besides your profile. Possible differences: OS, memory, processor type, plugins, memory available, different Web sites visited, and whether hardware acceleration is enabled on the computer. Plugins can allocate their own memory. Firefox uses more cache if there is more memory available. Ideally, eliminate as many variables as possible and narrow it down.

[VanillaMozilla]

C627627, both triagers trashed your bug report, and both gave you hints on what you can do to prove your point. If you can't provide that info, then you'll never get anything changed. Now deal with it.

And note that this is a support forum, not the Bugs forum, and not Bugzilla. Bye.

[squeaky]

Yes, sure, but what I'm saying is that with the same set of plug-ins and add-ons and the same set of open tabs - but a different operating system (I've switched from WinXP Home to Win7) and yes, different chipset etc. I'm seeing a different behaviour when it comes to memory usage.

In this case a much better behaviour

I'm not arguing for or against this being a "bug" - just saying, in the remote case that it might of some slight help, that to the best of my knowledge because I copied my profile straight across from winXP to win7 on a different machine I'm seeing a difference in behaviour. So it could be some function of a given chipset or system memory management that's at the root of this ??