ابحث في الدعم

Avoid support scams. We will never ask you to call or text a phone number or share personal information. Please report suspicious activity using the “Report Abuse” option.

Learn More

Firefox memory leak crashes every session

  • 37 ردًا
  • 8 have this problem
  • 8 views
  • آخر ردّ كتبه alphaa10000

more options

Every Firefox session, I use plenty of tabs, stacked atop each other in a vertical bar along the left side of the screen, using an extension named "Tree Style Tab" and its affiliated extension, "Open Link in New Tab" These never caused a problem for me during the first two years of use, but in the last three years or so, Firefox began to eat huge chunks of my memory.

Since I use Tree Style Tab routinely to keep perhaps 30 or so tabs open as I do research, my first thought was something had changed with one or both of the two extensions, and they had developed a problem with using memory efficiently. To test my hypothesis, I simply disabled both by starting Firefox in SafeMode.

Of course, that also disabled every other extension and plug-in, as well, but proved that my problem does not come from any extension or plug-in. Because while in SafeMode, Firefox continued to eat memory at a phenomenal rate until operations became extremely slow and the system finally froze in-place. That is, I began with some 3.1 Gb of physical memory (my 32-bit XP Pro SP3 system retains 1Gb from a total of 4GB), but soon found Firefox was using a whopping 1.7 Gb for only 20 tabs which should have taken no more than perhaps 300Mb, at most.

Yes, I have varied the mix of tabs, so if there were any memory-intensive connection / website, that connection would have stood out plainly. But the same effect is observed no matter how the tab connections are varied.

All my plug-ins and extensions are up-to-date, and Firefox is at version 38. All caches have been cleared.

Every Firefox session, I use plenty of tabs, stacked atop each other in a vertical bar along the left side of the screen, using an extension named "Tree Style Tab" and its affiliated extension, "Open Link in New Tab" These never caused a problem for me during the first two years of use, but in the last three years or so, Firefox began to eat huge chunks of my memory. Since I use Tree Style Tab routinely to keep perhaps 30 or so tabs open as I do research, my first thought was something had changed with one or both of the two extensions, and they had developed a problem with using memory efficiently. To test my hypothesis, I simply disabled both by starting Firefox in SafeMode. Of course, that also disabled every other extension and plug-in, as well, but proved that my problem does not come from any extension or plug-in. Because while in SafeMode, Firefox continued to eat memory at a phenomenal rate until operations became extremely slow and the system finally froze in-place. That is, I began with some 3.1 Gb of physical memory (my 32-bit XP Pro SP3 system retains 1Gb from a total of 4GB), but soon found Firefox was using a whopping 1.7 Gb for only 20 tabs which should have taken no more than perhaps 300Mb, at most. Yes, I have varied the mix of tabs, so if there were any memory-intensive connection / website, that connection would have stood out plainly. But the same effect is observed no matter how the tab connections are varied. All my plug-ins and extensions are up-to-date, and Firefox is at version 38. All caches have been cleared.

All Replies (17)

more options

John99 said


Managing Bookmarks

... Trying to work from bookmarks instead of many open tabs should also ease problems. It is probably worth remembering that even without any additional addon Firefox's open tabs-- 1, Can be searched see Awesome Bar - Search your Firefox bookmarks, history and tabs from the address bar_changing-results-on-the-fly 2. Have a scrollable dropdown list from the Navigation toolbar

Actually, working with bookmarks was my first approach to organizing tabs, but it quickly grew out of hand because my daily intake from a variety of sources required me to move quickly from one bookmarked collection of tabs to another, and back again.

Currently, I use the FF extension Session Manager, which permits me not only to capture all tabs related to a subject area under one title ("medical research") but to add/remove individual site tabs and to join/merge two session tab collections. This means (more often, at least) I can work from the same session title/collection of tabs, and modify it, tab by tab, as the work progresses.


About:Memory Diagnostic

Firefox's about:memory as mentioned previously should catch and by diffing the results highlight the sort of leak you are seeing.

Strongly of interest, since I am eager to determine what causes Flash to crash with some sites, but not others. I am reading the Mozilla tutorial on such diagnostics, now.

Potentially may be worth filing a bug

Are any of the sites concerned well known and popular ones where you see a continual memory leak. It may be worth investigating further and considering filing a bug...

I may do that, John, because cbsnews.com is almost the epicenter of the Flash problem. Whatever CBS web coders are doing with their site, it probably invites the memory leak problem with other visitors, as well. CBS Interactive web coders can be reached with only a determined drilling effort through the CBSNews.com website, but they usually faithfully respond, so I may try that, as well.

Modified by alphaa10000

more options

alphaa10000 said

12345

How do I delete a duplicated post?


Reply by edit: You are not able to delete posts only edit the content of your own posts. I have now deleted #answer-755981 for you.

Modified by John99

more options

John99 said


Managing Bookmarks

... Trying to work from bookmarks instead of many open tabs should also ease problems. It is probably worth remembering that even without any additional addon Firefox's open tabs-- 1, Can be searched see Awesome Bar - Search your Firefox bookmarks, history and tabs from the address bar_changing-results-on-the-fly 2. Have a scrollable dropdown list from the Navigation toolbar

Actually, working with bookmarks was my first approach to organizing tabs, but it quickly grew out of hand because my daily intake from a variety of sources required me to move quickly from one bookmarked collection of tabs to another, and back again.

Currently, I use the FF extension Session Manager, which permits me not only to capture all tabs related to a subject area under one title ("medical research") but to add/remove individual site tabs and to join/merge two session tab collections. This means (more often, at least) I can work from the same session title/collection of tabs, and modify it, tab by tab, as the work progresses.


About:Memory Diagnostic

Firefox's about:memory as mentioned previously should catch and by diffing the results highlight the sort of leak you are seeing.

Strongly of interest, since I am eager to determine what causes Flash to crash with some sites, but not others. I am reading the Mozilla tutorial on such diagnostics, now.

Potentially may be worth filing a bug

Are any of the sites concerned well known and popular ones where you see a continual memory leak. It may be worth investigating further and considering filing a bug...

I may do that, John, because cbsnews.com is almost the epicenter of the Flash problem. Whatever CBS web coders are doing with their site, it probably invites the memory leak problem with other visitors, as well. CBS Interactive web coders can be reached with only a determined drilling effort through the CBSNews.com website, but they usually faithfully respond, so I may try that, as well.

more options

Try to boot the computer in Windows Safe Mode with network support (press F8 on the boot screen).

Are all plugins set the "Ask to Activate" or "Never Activate"?

more options

Tell me more-- how will Windows Safe Mode help with Firefox memory leak? Or is that suggestion directed at something else?

In answer to your second question, using Firefox Safe Mode, all Add-Ons were disabled, and when rebooted into normal mode, all plug-ins were set to "Ask to Activate".

Modified by alphaa10000

more options

We have already tried disabling plugins and disabling FlashPlayer. That is how you discovered the current issue is independent of FlashPlayer use.

cor-el may not have read all of this thread, or may have noted it has not been tried yet.

Windows Safe mode will take additional factors out of the equation and is used as a troubleshooting step where the problem is thought to be more of a System wide issue or at least external to Firefox

more options

Bad news about Firefox 64 bit. Apparently the Windows 64bit Firefox is only compatible with Windows 7 or later, so if you wish to upgrade to 64 bit Firefox you will need to change to Windows 7 or above.

Or consider some other avenue such as using a multi boot so you may use Linux on the machine that is usually Windows XP.


James said Note you will need 64-bit Windows 7 or newer Windows to use Win64 as they will not work on 64-bit WinXP, Vista and servers as you can still use 32-bit. Bug 1093741

The 64-bit Firefox.exe itself will no longer run on older operating systems than Windows 7 since bug 1094013. BTW Firefox.exe itself has no checks. The operating system is the one that checks the subsystem version of the executable.
more options

John, that confirms my decision to move production to a Windows 7 machine.

I'll keep the XP volume for legacy applications, but all this troubleshooting made it clear 32bit limits my ability to run as many tabs as I prefer.

Thanks for the update.

Modified by alphaa10000

more options

cor-el said

Try to boot the computer in Windows Safe Mode with network support Are all plugins set the "Ask to Activate" or "Never Activate"?

Tried that, but made no difference. The YouTube samples which had crashes in normal mode refused to run at all in SafeMode, but this was expected. Memory leakage as strong as ever.

I was under the impression YouTube had gone totally HTML5 with its video, but perhaps I chose the selections which had not been converted. All selections were from 2013 and 2014.

Firefox plugins were set to "Ask to Activate"

more options

John, since I would like to learn more about FF memory-handling functions, I am reading https://support.mozilla.org/en-US/kb/firefox-uses-too-much-memory-ram

From this vantage point, I have a lot of reading and experimentation ahead. But so far, having reviewed the main page for the article, perhaps the next step is to look into a heap profiler like DMD, and check its cumulative mode. As you pointed out, this may hold clues about the specific source of the problem.

As mentioned earlier, I plan to move on to Windows 7, a good, solid Windows release, and a 64bit platform which could eliminate some of the memory limitations under 32bit. That step, in itself, could transform the situation dramatically (Thanks also to Ed, who pointed out an OOM or two in my crash reports)

* While using SysInternals Process Explorer 12.04 to monitor some memory processes (since restoring the default Adobe Flash version), I noticed that, after having terminated a FF session in good order, its memory segment was not released, at all. That is, all 572k was kept in memory while I loaded another FF session. What do these symptoms suggest about the current version 39.0.3 of FF?

more options

alphaa10000 said

* While using SysInternals Process Explorer 12.04 to monitor some memory processes (since restoring the default Adobe Flash version), I noticed that, after having terminated a FF session in good order, its memory segment was not released, at all. That is, all 572k was kept in memory while I loaded another FF session. What do these symptoms suggest about the current version 39.0.3 of FF?

Not sure but I suspect it may not be too important.

If Firefox leaked 572k per termination then after a 100 terminations that is something in the order of 50 or 60 MB but Firefox would ordinarily be using 100s or many 100s of MB so unless you repeatedly terminate Firefox; but almost never turn off the computer; you may not see problems.

more options

John, my late-night hours are catching up with me-- I meant to say FF left a 572MB segment in memory after what seemed a normal termination. This happens often, perhaps every five sessions.

My concern was that failure to clear the "rogue" memory segment on FF termination could be related in some way to the memory leak. Both events represent a memory issue of major proportions.

Of course, I still subscribe to the idea Flash misuses memory resources, and hogs them especially at media-heavy websites. But the rogue segment is above and beyond what Flash normally does.

Modified by alphaa10000

more options

That is more serious, but maybe you just need to wait a while longer, up to 15 minutes !!

I am not even sure if it would be a Firefox issue or an OS issue. I don't often even use Windows, and it is a long while since I have used Sysinternals Process Explorer.

As a rule of thumb a process releases its memory when it terminates. So even if Firefox had memory leaks I would expect Windows to release all associated memory when Firefox terminates. It is almost a definition of a memory leak to say: It is memory allocated to a process that the process has lost track of, and is unable to deallocate that memory itself, until the process closes and the OS frees the memory. Also with the Firefox process terminated, I would not expect it to be found by SysInternals Process Explorer. Maybe SysInternals is just indicating the last process that segment was allocated to.

Are you allowing Firefox enough time to shutdown ? I think now after about 10 minutes Firefox will force a crash if it fails to shutdown. However that crash report will still take time to process and submit, and there may also be crash reports involved from FlashPlayer hangs.

more options
The 64-bit Firefox.exe itself will no longer run on older operating systems than Windows 7 since bug 1094013. BTW Firefox.exe itself has no checks. The operating system is the one that checks the subsystem version of the executable.

Does this mean I can run FF 64 on XP SP3, with no major problems beyond a failure to check the executable?

Are you allowing Firefox enough time to shutdown ? I think now after about 10 minutes Firefox will force a crash if it fails to shutdown. However that crash report will still take time to process and submit, and there may also be crash reports involved from FlashPlayer hangs.

John, I usually wait up to two or three minutes, which seems enough time for FF to sort itself out. Based on past behavior under similar but problem-free conditions, shutdown required fewer than 20 seconds.

Judging from your time of response, and your statement you do not use Windows, are you a UK Apple user? Just curious, as usual.

Modified by alphaa10000

more options

I noticed that comment myself. I am not sure if it is actually possible to use the 64 bit Firefox on XP, but clearly reading other comments it is not intended to support XP. (I has worked with XP in the past IIRC) I suppose the easiest way to find out is to try it if you have 64bit XP, but even if it were to work now it probably will not remain fully functional and secure.

I am in the uk. I do have Windows 7 and XP. I think I may have a machine bootable in Windows 3 still, I certainly have a hdd from that era that works and was bootable when I tried it last year. I have not used Apple's Macs for decades, I can see their appeal but tend to go for the cheapest options and often get old machines free and put Linux on them.

As for the shutdown, Firefox has various checks built in. It is important it closes correctly, so that it is not wasting resources, and so that it can restart normally. Partly due to a series of problems with hangs at closedown, that were not causing crashes a last resort type check was introduced some time about Fx34. If Firefox is trying to closedown but after about ten minutes has failed it forces a crash. That should not normally happen, but does ensure the problem should be noticed. If a regression of some sort causes that type of problem, even some third party software, there should be a clear and noticeable spike in the Crash reports of that particular signature, and the reports should help in discovering what the issue is.

In your case if something odd is happening at closedown it is worth giving Firefox an extended period of time in order to see if it initiates such a crash.

more options

NEWS FLASH:::

to remedy or partially u can use a different player such as VLC set it in applications or contents and set it to play as default for video playback. u can find vlc player at videolan.org its open source i've used for years and its a great option and alternative.

http;//videolan.org

more options

Thanks, SB. For now, I have decided to stay with HTML5 and see how the video play runs with that. Yet, I have used VLC player and was happy enough with the results. The version I used does require fine tuning, however, and that can get a bit tricky unless user has sufficient video experience and enough time to experiment,

In any case, glad VLC worked for you-- I am an open source supporter, as well. We users face a major problem with Windows in the near future-- even if Windows 10 is to be free for one year to qualified systems. The whole relationship with Microsoft is now to be determined by what the corporation deems suitable hardware for its flagship product.

If that definition one day disqualifies certain hardware, users of such newly-disqualified systems cannot even hope for a Windows remedy for a Windows driver issue. Ironically, the forced connection to a perpetual Windows update machine means disqualification is much more likely, the longer a system remains useful.

And since Windows 10 is considered permanently installed in a given system, if that system loses its qualification, both hardware and its Windows become a dead box, married permanently to any problems which may develop. In fact, users may be notified of a "problem" before they experience it, as the semi-plausible justification for disqualification. Clearly, OEMs of today are more eager than ever to cut sharply the cost of user support. Not only does user support reduce net profit, but OEMs have shareholders to keep happy. But what happened to pleasing customers, to increase market share?

For users under the new Microsoft-dictated order, the only direct remedy is to buy a new, Windows-qualified system and a new license. Since the new hardware and Windows license cost more if bought separately, the new regime becomes a bonanza for new retail system makers, ensuring widespread OEM support.

Such increased vulnerability to technical change becomes expensive for ordinary computer users, especially when coupled with reluctance of third-party software producers to develop new drivers for old hardware. Under the new Windows 10 regime, even newer hardware is vulnerable to an accelerated pace of Windows internal changes, not to mention known (and unknown) downstream hardware impacts and third-party abandonments.

Better control of the situation is afforded with LInux, it goes almost without saying. So, I am inclined to migrate toward an environment which is more stable and often more hardware-friendly. Even if an undiscovered problem is buried within Linux and device driver code, there is no ticking clock in the background dooming the hardware to failure and decreed obsolescence. Linux coders may work around the problem and the user also has the option to remain with a given version of Linux which does work acceptably-- no "overnight upgrade", mandatory changes in Windows to force disqualification issues.

The open source community is the natural ally of every computer system user, but especially the Microsoft-dependent. In fact, that may be one way open source can expand its following exponentially, provided Microsoft and device OEM licensing is not hostile to efforts to extend life of user-owned hardware.

Modified by alphaa10000

  1. 1
  2. 2