Cerca nel supporto

Attenzione alle mail truffa. Mozilla non chiederà mai di chiamare o mandare messaggi a un numero di telefono o di inviare dati personali. Segnalare qualsiasi attività sospetta utilizzando l'opzione “Segnala abuso”.

Learn More

Questa discussione è archiviata. Inserire una nuova richiesta se occorre aiuto.

Memory & Jank issues (/questions/1006397 &see998289)

  • 9 risposte
  • 9 hanno questo problema
  • 2 visualizzazioni
  • Ultima risposta di BillM

more options

Firefox doesn't play well with others: it allocates excessive memory and refuses to release it.

To eliminate the obvious... I do not have ABP installed. I use about:memory and Windows Resource Monitor (or taskmgr) to observe memory use (on Windows 7).

Firefox allocates a lot of memory (often upwards of 1.5GB on a 2GB machine) and is VERY loathe to release it. Even exiting the browser (by closing the window) still shows Firefox in memory (although "greyed out" in either resmon or taskmgr) ... and it refuses to "let go" for upwards of 60 seconds, even with nothing else running on the machine. (I don't have time to watch it past 60 sec. ... and starting a second Firefox process, while in this state, was tested, but does not "speed up" release of memory by the first process.)

> "I do not doubt what you say or that you have issues, just that it is not yet the sort of report that allows us to file a good bug with good Steps To Reproduce that we may pass on to an engineer to investigate."

And frankly, that is part of the problem. Routine, real-world use of the Web has outgrown the Mozilla/Firefox business model. I've submitted crash-reports in the past two weeks, where a specific web page would reliably exhaust well-over 1GB of RAM for Firefox ... but they disappear into a black hole. Meanwhile I spent hours of my time trying to find the cause myself, to produce a "good bug" for you... including close to an hour documenting this one.

The browser became "a utility" years ago, and the time has come where it must be treated (and supported) like one. Just possibly the day of a volunteer-built-and-supported browser is... over. (And I say this as a loyal Firefox user since at least 0.8)

Firefox doesn't play well with others: it allocates excessive memory and refuses to release it. To eliminate the obvious... I do not have ABP installed. I use about:memory and Windows Resource Monitor (or taskmgr) to observe memory use (on Windows 7). Firefox allocates a lot of memory (often upwards of 1.5GB on a 2GB machine) and is VERY loathe to release it. Even exiting the browser (by closing the window) still shows Firefox in memory (although "greyed out" in either resmon or taskmgr) ... and it refuses to "let go" for upwards of 60 seconds, even with nothing else running on the machine. (I don't have time to watch it past 60 sec. ... and starting a second Firefox process, while in this state, was tested, but does not "speed up" release of memory by the first process.) > "I do not doubt what you say or that you have issues, just that it is not yet the sort of report that allows us to file a good bug with good Steps To Reproduce that we may pass on to an engineer to investigate." And frankly, that is part of the problem. Routine, real-world use of the Web has outgrown the Mozilla/Firefox business model. I've submitted crash-reports in the past two weeks, where a specific web page would reliably exhaust well-over 1GB of RAM for Firefox ... but they disappear into a black hole. Meanwhile I spent hours of my time trying to find the cause myself, to produce a "good bug" for you... including close to an hour documenting this one. The browser became "a utility" years ago, and the time has come where it must be treated (and supported) like one. Just possibly the day of a volunteer-built-and-supported browser is... over. (And I say this as a loyal Firefox user since at least 0.8)

Tutte le risposte (9)

more options

Hello,

Looking at your system details, you appear to have multiple versions of the same plugin installed which can cause problems. While you have disabled "Shockwave for Director", it might be a good idea to remove the older versions. Same with Flash - you have both version 13.0.0.214 and 15.0.0.152.

For Flash, you can download the uninstaller here: http://helpx.adobe.com/flash-player/kb/uninstall-flash-player-windows.html

It will likely remove both versions (which may be a good idea since you don't know what the uninstaller will remove from version 13 that 15 uses) - so you can download a fresh copy of Adobe Flash here: https://get.adobe.com/flashplayer/

more options

Removed old versions of Flash; no material difference in behavior. (But cleaner/simpler is theoretically better; it used-to-be that new versions of Flash would replace previous ones. The problems with Flash upgrades in the past couple years were well-documented... and I re-experienced all of them, as I removed all old versions and installed Flash 15.0.0)

I don't believe this is clearly related to 998289. Not using AdBlock Plus. My extensions are NoScript, Logitech SetPoint 6.5, HTTPS-Everywhere (but disabling it doesn't change the behavior), and a few cosmetic extensions for pre-Australis look and feel (Addon Bar (restored) 2.0, Back/forward dropmarker 1.0, Downloads Window 0.5.3)

The core issue at this time is sloppy memory management; allocating a lot, and FAILING to release it and exit in a timely manner, when closing the Firefox window.

Modificato da BillM il

more options

Note that this problem is still outstanding. The consumption of excessive memory occurs across multiple websites, usually with nothing even-remotely "fancy" - e.g., no Flash. (I have taken the advice to purge old Flash versions, with little apparent effect.)

Just minutes ago, I had a single browser window open to a single site (no tabs) ... clicked the site's Exit button, and Firefox continued to INCREASE its memory consumption, until I eventually killed the process on my laptop.

How many ways can we spell, "UNACCEPTABLE?" We need better diagnostic tools, folks.

more options

Significant improvement with 33.0.1! Memory use (working set) is now in the 180MB vicinity and noticeably more stable than earlier 3x.* releases. I suspect you've at least pulled the low-hanging fruit off the tree. Will monitor behavior over the next few days to see if this improvement "sticks" (or to correlate any unusual memory use with specific activity or pages). I am noticing slow increase in working set size over time in a stable environment - e.g. has risen from 180 to 192MB in the time spent writing this message, then dropped-back to 189MB, rose again above 199MB... with no other activity than this page.

more options

Start Firefox in Safe Mode {web link} While you are in safe mode;

Type about:preferences<Enter> in the address bar

Select Advanced > General. Look for and turn off Use Hardware Acceleration.

Poke around safe web sites. Are there any problems?

Then restart.

more options

Now suspecting a "slow memory leak" in Firefox 33. My general habit is to start Firefox and run it for 8+ hours without restarting the process.

Config: safe mode; hardware acceleration already disabled. Baseline memory consumption (start in safe mode, navigate to about:preferences, then to about:memory) reveals 154MB in use by Firefox (as shown by win7 Resource Monitor).

about:memory shows: 91MB explicit allocation, 20 decommitted, 1600+ event counts, 39MB js-main-runtime, 19MB js-main-runtime-gc-heap-committed, 149MB private, 409MB vsize.

After running GC from about:memory, those data are: 90MB explicit allocation, 26 decommitted, 1300 event counts, 32MB js-main-runtime, 15MB js-main-runtime-gc-heap-committed, 154MB private, 414MB vsize. (notice increases in both private and vsize, from doing nothing but garbage-collection. Memory leak?

In the 15 minutes used to gather/report these data, Resource Monitor shows Firefox memory has risen to 159.6MB... and about:memory data have risen to 92.14MB explicit allocation, 25 decommitted, 1300 event counts, 33MB js-main-runtime, 15MB js-main-runtime-gc-heap-committed, 155MB private, 414MB vsize... while the Firefox window sat completely idle (except for the single-click on "Measure" to refresh the about:memory data).

Will leave this Firefox instance alone, untouched, for some hours and see what changes.

more options

(Of best-laid plans and insufficient machines) Have not been able to leave my machine completely idle long enough to prove/disprove slow leaks. It would be nice if someone in the moz labs would try this.

more options

I've called the big guys to help you. Good luck.

more options

(Noted: No follow-up from 'big guys'.)

New variation on the same problem: Firefox 33.0.1 is a memory HOG.

I opened two tabs, to "watch" an NFL game... one tab for patriots.com (a text-only description of game activity), the second for radio.com (to hear the audio from a broadcast radio station).

Imagine my surprise: Windows Performance Monitor shows the firefox.exe process with these two tabs (and nothing else) consumes nearly 1.5 GB of RAM.

Repeat the same experiment with IE 11 ... it uses a total of just over 0.5GB (in two processes - one per tab, vs. Firefox's single process - but processes are relatively cheap compared to RAM.)

(Note there is NO video involved here.... and I am not upgrading to 34.x until I'm convinced it's reasonably stable. The ongoing problem reports suggest that it's not.)

Modificato da BillM il