Closed Bug 973776 Opened 10 years ago Closed 10 years ago

Memory leak in Firefox 27.0.1 on Mac when minimized to dock

Categories

(Firefox :: General, defect, P2)

27 Branch
x86_64
Windows 7
defect

Tracking

()

RESOLVED DUPLICATE of bug 937878

People

(Reporter: raghu.kuchi, Unassigned)

Details

(Whiteboard: [MemShrink])

Attachments

(14 files)

66.14 KB, application/pdf
Details
29.54 KB, image/png
Details
1.45 KB, text/rtf
Details
327.64 KB, application/x-gzip
Details
23.12 KB, image/jpeg
Details
19.72 KB, text/plain
Details
19.33 KB, text/plain
Details
41.28 KB, image/jpeg
Details
1.14 MB, video/quicktime
Details
10 years ago
(deleted), image/jpeg
Details
27.78 KB, image/jpeg
Details
27.78 KB, image/jpeg
Details
66.41 KB, image/jpeg
Details
204.75 KB, application/x-gzip
Details
Attached file memory details
We are developing an ajax based application which shows a tree and table control and  both the controls are reloaded after certain time. 

This application was working fine till Version 20. From next versions memory leak is started. The memory usage is starting from 120 MB and reaching to 1.5 GB in 12 hrs.


Trouble shooting information
------------------------------------
------------------------------------

Application Basics
------------------

Name: Firefox
Version: 27.0.1
User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:27.0) Gecko/20100101 Firefox/27.0

Crash Reports for the Last 3 Days
---------------------------------

All Crash Reports

Extensions
----------

Name: Firebug
Version: 1.12.6
Enabled: true
ID: firebug@software.joehewitt.com

Name: Html Validator
Version: 0.9.5.8
Enabled: true
ID: {3b56bcc7-54e5-44a2-9b44-66c3ef58c13e}

Name: Java Quick Starter
Version: 1.0
Enabled: true
ID: jqs@sun.com

Name: Proxy Selector
Version: 1.30
Enabled: true
ID: proxyselector@mozilla.org

Name: TabRenamizer
Version: 0.9.7
Enabled: true
ID: {792BDDFE-2E7C-42ed-B18D-18154D2761BD}

Important Modified Preferences
------------------------------

accessibility.typeaheadfind: true
accessibility.typeaheadfind.casesensitive: 1
accessibility.typeaheadfind.flashBar: 0
browser.cache.disk.capacity: 0
browser.cache.disk.smart_size.enabled: false
browser.cache.disk.smart_size.first_run: false
browser.places.importBookmarksHTML: false
browser.places.smartBookmarksVersion: 4
browser.sessionstore.upgradeBackup.latestBuildID: 20140212131424
browser.startup.homepage_override.buildID: 20140212131424
browser.startup.homepage_override.mstone: 27.0.1
dom.mozApps.used: true
extensions.lastAppVersion: 27.0.1
font.internaluseonly.changed: true
gfx.direct3d.last_used_feature_level_idx: 0
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1392627350
places.history.expiration.transient_current_max_pages: 104858
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
plugin.state.java: 2
print.printer_PDF-XChange_4.0.print_bgcolor: false
print.printer_PDF-XChange_4.0.print_bgimages: false
print.printer_PDF-XChange_4.0.print_colorspace:
print.printer_PDF-XChange_4.0.print_command:
print.printer_PDF-XChange_4.0.print_downloadfonts: false
print.printer_PDF-XChange_4.0.print_duplex: 0
print.printer_PDF-XChange_4.0.print_edge_bottom: 0
print.printer_PDF-XChange_4.0.print_edge_left: 0
print.printer_PDF-XChange_4.0.print_edge_right: 0
print.printer_PDF-XChange_4.0.print_edge_top: 0
print.printer_PDF-XChange_4.0.print_evenpages: true
print.printer_PDF-XChange_4.0.print_footercenter:
print.printer_PDF-XChange_4.0.print_footerleft: &PT
print.printer_PDF-XChange_4.0.print_footerright: &D
print.printer_PDF-XChange_4.0.print_headercenter:
print.printer_PDF-XChange_4.0.print_headerleft: &T
print.printer_PDF-XChange_4.0.print_headerright: &U
print.printer_PDF-XChange_4.0.print_in_color: true
print.printer_PDF-XChange_4.0.print_margin_bottom: 0.5
print.printer_PDF-XChange_4.0.print_margin_left: 0.5
print.printer_PDF-XChange_4.0.print_margin_right: 0.5
print.printer_PDF-XChange_4.0.print_margin_top: 0.5
print.printer_PDF-XChange_4.0.print_oddpages: true
print.printer_PDF-XChange_4.0.print_orientation: 0
print.printer_PDF-XChange_4.0.print_page_delay: 50
print.printer_PDF-XChange_4.0.print_paper_data: 9
print.printer_PDF-XChange_4.0.print_paper_height: 11.00
print.printer_PDF-XChange_4.0.print_paper_name:
print.printer_PDF-XChange_4.0.print_paper_size_type: 0
print.printer_PDF-XChange_4.0.print_paper_size_unit: 1
print.printer_PDF-XChange_4.0.print_paper_width: 8.50
print.printer_PDF-XChange_4.0.print_plex_name:
print.printer_PDF-XChange_4.0.print_resolution: 0
print.printer_PDF-XChange_4.0.print_resolution_name:
print.printer_PDF-XChange_4.0.print_reversed: false
print.printer_PDF-XChange_4.0.print_scaling: 1.00
print.printer_PDF-XChange_4.0.print_shrink_to_fit: true
print.printer_PDF-XChange_4.0.print_to_file: false
print.printer_PDF-XChange_4.0.print_unwriteable_margin_bottom: 0
print.printer_PDF-XChange_4.0.print_unwriteable_margin_left: 0
print.printer_PDF-XChange_4.0.print_unwriteable_margin_right: 0
print.printer_PDF-XChange_4.0.print_unwriteable_margin_top: 0
privacy.cpd.offlineApps: true
privacy.cpd.siteSettings: true
privacy.sanitize.migrateFx3Prefs: true
privacy.sanitize.timeSpan: 0
security.warn_viewing_mixed: false
storage.vacuum.last.index: 1
storage.vacuum.last.places.sqlite: 1391160095

Graphics
--------

Adapter Description: AMD RADEON HD 6350
Adapter Drivers: aticfx64 aticfx64 aticfx64 aticfx32 aticfx32 aticfx32 atiumd64 atidxx64 atidxx64 atiumdag atidxx32 atidxx32 atiumdva atiumd6a atitmm64
Adapter RAM: 512
Device ID: 0x68f9
Direct2D Enabled: true
DirectWrite Enabled: true (6.1.7601.18245)
Driver Date: 11-16-2010
Driver Version: 8.783.2.2000
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 10
Vendor ID: 0x1002
WebGL Renderer: Google Inc. -- ANGLE (AMD RADEON HD 6350 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: false
AzureCanvasBackend: direct2d
AzureContentBackend: direct2d
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

JavaScript
----------

Incremental GC: true

Accessibility
-------------

Activated: false
Prevent Accessibility: 0

Library Versions
----------------

NSPR
Expected minimum version: 4.10.2
Version in use: 4.10.2

NSS
Expected minimum version: 3.15.4 Basic ECC
Version in use: 3.15.4 Basic ECC

NSSSMIME
Expected minimum version: 3.15.4 Basic ECC
Version in use: 3.15.4 Basic ECC

NSSSSL
Expected minimum version: 3.15.4 Basic ECC
Version in use: 3.15.4 Basic ECC

NSSUTIL
Expected minimum version: 3.15.4
Version in use: 3.15.4
Can you link to the test application? Without more information, it isn't possible for people to investigate the leak further.
Flags: needinfo?(raghu.kuchi)
I'm a user of firefox 27.0.1 
Operating system: OsX 10.9.1 
Hardware: Retina Macbook with 16 Gb of ram
Plugins: Block site plus 1.1, Down them all 2.0.6, FoxyProxy basic 3.2.4, Karma blocker 0.4.8, lightbeam 1.08, self destructing cookies 0.4.3

Open tabs: Grooveshark and my personal wiki 

I see that firefox periodically asks for >8 gb of memory! This is a huge dealbreaker for me.
I POSTED THE REPORT BELOW AT https://support.mozilla.org/en-US/questions/987015?
IT WAS SUGGESTED TO ME THAT I POST IT HERE:


For my work I use FF under OSX 10.6. I have another, older machine with OSX 10.7 Lion installed running around the clock in my living quarters. I use FF on it for trivial operations such as checking the weather report, quick searches etc. I also always test new versions on it before installing it on my work machines, so as not to risk any workflow problems.

Soon after installing FF 27.0 I was very surprised to find that the machine running Lion was almost frozen, with 500Mb of free space left instead of the usual 45-50Gb. After some research, I figured that in the course of a few hours the swap files had multiplied to death — something I never, ever had experienced earlier.

At first I thought that this was the result of installing several plugin updates (Java, Flash and Silverlight) at the same time as the FF update. But a couple of days later, the same thing began to happen again. Having left the window of the folder containing the swap files open in the Finder, I could not only see that they had multiplied, but I could also watch them disappear one by one after quitting Firefox.

Installing FF 27.0.1 did not resolve this issue. But since going back to FF 26, the swap files have been staying put. I'll try again with 28, but until then I'm not going to run 27 on my work machines.

I don't quite know how to get in touch with the developpers, so if anyone is there watching, please forward this to them.

ADDITION: I have ≈ 24 Pin Tabs open continuously. But as sais, this never affected FF before this last version.
For me it is the Grooveshark website! I didn't see the surge for almost half a day when it is not open. 

Repeat, this is just with that version, never happened before.
Never had Grooveshark open on that machine. But I see that Grooveshark uses Flash. Could the last update of the Flash plugin, or rather its interaction with version 27.x of FF, be the culprit? I'm not a programmer. so it's really an open question...
I POSTED THE REPORT BELOW AT https://support.mozilla.org/en-US/questions/987143
IT WAS SUGGESTED TO ME THAT I POST IT HERE:


Here is what I know about this problem:

- Two different Macs (10.8.x, 10.9.x) with 32 GB RAM ran out of system swap due to the size of the Firefox process growing to over 20 GB, and both machines had virtually idle CPUs with no Firefox usage whatsoever - they were unattended
- It occurred immediately after upgrading from Firefox 26 to 27.0.1
- The set of pages, tabs, windows was the same before and after upgrading
- There does not seem to be a single identifiable URL that causes the problem, but opening a window with this URL and then iconifying it causes memory to be leaked at a substantial rate (at least on one OS X 10.8.5 machine)
 - http://msdn.microsoft.com/en-us/magazine/cc163744.aspx
- I normally have ~25 tabs distributed across ~8 windows, all of which but one are iconified
- Memory is consumed at the staggering rate of ~1 GB/hour while the offensive window is iconified
- Memory is not leaked per-se - about:memory confirms that heap-unclassified memory is growing rapidly, but as soon as the offending window is opened and no longer iconified, the memory usage immediately drops back to the normal ~1 GB or so
  - this is also verifiable in Activity Monitor.app, etc
- The problem occurs in safe mode, but regardless the only plugins installed are Shockwave Flash, QuickTime, and Java Applet
- I have killed Firefox repeatedly and can recreate the same memory usage pattern
- I have at least one Mac, however, on which trying to reproduce this does nothing
- I've already spent hours trying to determine even the initial cause of this
@jpabel: much more detailed, matches pretty accurately my own "symptoms". Except that mine were under 10.7.5.
I noticed the same symptoms as @jpabel starting with 27.0.1.  After I minimize my firefox window, memory usage starts going up tremendously.  Once I restore the window (if there is enough swap space left to do so -- it usually takes a while for it to swap back in), after it finally restores, memory usage drops to a normal amount.  I am on OS X 10.6.8.
Had something similar, reported here:
https://bugzilla.mozilla.org/show_bug.cgi?id=975945

For me, when opening a new tab, it was something with the 'tab page' thumbs that shot the RAM to the max, totally freezing my Mac, (something I thought was not even possible).

Had Activity Monitor open, and the minute I open the tab page, memory jumps from 200MB or so to 1.6GB in 5 seconds. After opening a new tab, I have but a few seconds before my Mac becomes completely unresponsive, to the point of mouse freeze(!), to quit FF. The HD also start working endlessly during that time. 

Cleaning history, closing the Tab Page Preview (the thumbs) or disabling tap page altogether, prevents this from happening. So it's something to do with the 'Tab Page'.

(Running OS X 10.8.5, FF 27.0.1)
@bauermusic, that's not the case for me.  The new tab page doesn't cause my memory usage to spike.  It is specifically iconifying the application (like @jpabel mentioned) that does it for me.  If I leave the application open on the desktop, it doesn't happen either.
Update to my earlier report: I've had FF 26 running for 9 days straight after downgrading from 27.0.1, and there are still only three swap files in the /vm folder. The last of these has a time stamp of three days ago. So the leak does not appear to be related to the plugins, as I have not downgraded these.

To the last two posters: I don't use the Tab Page, nor do I iconify FF. The application was open "on the desktop" when the leaks happened.

I hope that someone is reading this and that the issue will be fixed by the time 28 comes out.
I have been experiencing similar memory symptoms since upgrading to 27.0.1.  Just today, I opened task manager and saw for 5 tabs, I had firefox taking more than 1GB, related plugins for a Netflix tab around 500MB.  On previous versions, I would barely top out at 3-400MB total for running more tabs and Netflix or other streaming content.
Windows XPsp3 - P4 2G of ram, Firefox 27.01. Woah! Upgraded from v. 20, When Firebug 1.13.0a9 (or earlier version) submenu has "Enable all panels" activated, even with only Console panel open, whenever a DATA URI is dynamically updated, my application takes forever to respond to GUI navigation, and slows down to a grinding halt! Never had this issue in Firefox 20 with Firebug installed. I have to close Firefox and restart it to get my app to function normally again. Not sure if Firebug issue or Firefox prob.
Im using Lubuntu 13.10 , Firefox 27.0.1, 6 tabs opened (4 about the problem, 2 For my use- YouTube and Facebook).

Virtual Memory:
920 MB

RSS (What is RSS? ):
440 MB

Data (Trubleshoting) Sorry for Hebrew-עברית, Guess most םof you cant read (: . Can I make an English Version? :
מידע יישום בסיסי
----------------

שם: Firefox
גרסה: 27.0.1
סוכן משתמש: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:27.0) Gecko/20100101 Firefox/27.0

Crash Reports for the Last 3 Days
---------------------------------

All Crash Reports

הרחבות
------

שם: Adblock Plus
גרסה: 2.5.1
מאופשר: true
מזהה: {d10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d}

שם: Flash-Aid
גרסה: 2.2.3
מאופשר: true
מזהה: flashaid@lovinglinux.megabyet.net

שם: Ubuntu Firefox Modifications
גרסה: 2.8
מאופשר: true
מזהה: ubufox@ubuntu.com

העדפות חשובות ששונו
-------------------

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size_cached_value: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.places.smartBookmarksVersion: 4
browser.sessionstore.upgradeBackup.latestBuildID: 20140218140052
browser.startup.homepage_override.buildID: 20140218140052
browser.startup.homepage_override.mstone: 27.0.1
dom.mozApps.used: true
extensions.lastAppVersion: 27.0.1
gfx.blacklist.direct2d: 4
gfx.blacklist.layers.direct3d10: 4
gfx.blacklist.layers.direct3d10-1: 4
gfx.blacklist.layers.direct3d9: 4
gfx.blacklist.layers.opengl: 4
gfx.blacklist.stagefright: 4
gfx.blacklist.webgl.angle: 4
gfx.blacklist.webgl.msaa: 4
gfx.blacklist.webgl.opengl: 4
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1393449403
places.history.expiration.transient_current_max_pages: 26271
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1393449403

גרפיקה
------

windowLayerManagerRemote: false
גרסת מנהל ההתקן: 1.3 Mesa 9.2.1
חלונות עם האצה גרפית: 0/1 Basic חסום עבור הכרטיס הגרפי שלך עקב בעיות לא פתורות במנהל ההתקן.
מזהה התקן: Mesa DRI R200 (RV280 5960) x86/MMX/SSE2 TCL DRI2
מזהה יצרן: Tungsten Graphics, Inc.
מנוע תצוגת WebGL: חסום עבור הכרטיס הגרפי שלך עקב בעיות לא פתורות במנהל ההתקן.
תיאור מתאם: Tungsten Graphics, Inc. -- Mesa DRI R200 (RV280 5960) x86/MMX/SSE2 TCL DRI2
AzureCanvasBackend: cairo
AzureContentBackend: cairo
AzureFallbackCanvasBackend: none
AzureSkiaAccelerated: 0

JavaScript
----------

איסוף זבל מצטבר: true

נגישות
------

מופעל: false
מניעת נגישות: 0

גרסאות ספריה
------------

NSPR
גרסת מינימום מצופה: 4.10.2
גרסה שבשימוש: 4.10.2

NSS
גרסת מינימום מצופה: 3.15.4 Basic ECC
גרסה שבשימוש: 3.15.4 Basic ECC

NSSSMIME
גרסת מינימום מצופה: 3.15.4 Basic ECC
גרסה שבשימוש: 3.15.4 Basic ECC

NSSSSL
גרסת מינימום מצופה: 3.15.4 Basic ECC
גרסה שבשימוש: 3.15.4 Basic ECC

NSSUTIL
גרסת מינימום מצופה: 3.15.4
גרסה שבשימוש: 3.15.4
Same issu on Mac OS 10.8.5, Firefox 27.0.1

In the "about:memory" heap-unclassified is growing rapidly 

Main Process
Explicit Allocations
1,507,774,639 B (100.0%) -- explicit
├──1,176,324,525 B (78.02%) ── heap-unclassified

Debug info (in french sory):

Paramètres de base de l'application
-----------------------------------

Nom: Firefox
Version: 27.0.1
Agent utilisateur: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:27.0) Gecko/20100101 Firefox/27.0

Rapports de plantage des 3 derniers jours
-----------------------------------------

Tous les rapports de plantage

Extensions
----------

Nom: Adblock Plus
Version: 2.5.1
Activée: true
ID: {d10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d}

Nom: Delicious Bookmarks
Version: 2.3.4
Activée: true
ID: {2fa4ed95-0317-4c6a-a74c-5f3e3912c1f9}

Nom: Détection de dispositifs Logitech
Version: 1.23.0.5
Activée: true
ID: DeviceDetection@logitech.com

Nom: Dictionnaire français «Classique & Réforme 1990»
Version: 4.3
Activée: true
ID: fr-classique-reforme1990@dictionaries.addons.mozilla.org

Nom: Dictionnaire français «Classique»
Version: 4.3
Activée: true
ID: fr-classique@dictionaries.addons.mozilla.org

Nom: Dictionnaire français «Moderne»
Version: 4.3
Activée: true
ID: fr-moderne@dictionaries.addons.mozilla.org

Nom: Dictionnaire français «Réforme 1990»
Version: 4.3
Activée: true
ID: fr-reforme1990@dictionaries.addons.mozilla.org

Nom: DownloadHelper
Version: 4.9.21
Activée: true
ID: {b9db16a4-6edc-47ec-a1f4-b86292ed211d}

Nom: Module d'Antidote
Version: 8.16.31
Activée: true
ID: antidote7_mac_firefox_103@druide.com

Nom: Redirector
Version: 2.8.1
Activée: true
ID: redirector@einaregilsson.com

Nom: Save Text Area
Version: 0.4.7
Activée: true
ID: {8522e648-adce-469a-8c3a-18659a6ab6e3}

Nom: UAControl
Version: 0.1.3.1
Activée: true
ID: uacontrol@qz.tsugumi.org

Préférences modifiées importantes
---------------------------------

accessibility.typeaheadfind: true
accessibility.typeaheadfind.flashBar: 0
browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size_cached_value: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.history_expire_days.mirror: 180
browser.places.importBookmarksHTML: false
browser.places.importDefaults: false
browser.places.leftPaneFolderId: -1
browser.places.migratePostDataAnnotations: false
browser.places.smartBookmarksVersion: 4
browser.places.updateRecentTagsUri: false
browser.sessionstore.upgradeBackup.latestBuildID: 20140212131424
browser.startup.homepage: http://localhost/~ldy/NetApp/
browser.startup.homepage_override.buildID: 20140212131424
browser.startup.homepage_override.mstone: 27.0.1
dom.mozApps.used: true
dom.w3c_touch_events.expose: false
extensions.lastAppVersion: 27.0.1
gfx.blacklist.webgl.msaa: 4
network.cookie.prefsMigrated: true
network.protocol-handler.warn-external.itmss: false
network.protocol-handler.warn-external.itpc: false
places.database.lastMaintenance: 1393915499
places.history.expiration.transient_current_max_pages: 104858
places.history.expiration.transient_optimal_database_size: 167772160
places.last_vacuum: 1300308867
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
print.macosx.pagesetup-2: PD94bWwgdmVyc2lvbj0iMS4wIiBlbmNvZGluZz0iVVRGLTgiPz4KPCFET0NUWVBFIHBsaXN0IFBVQkxJQyAiLS8vQXBwbGUvL0RURCBQTElTVCAxLjAvL0VO
print.print_bgcolor: false
print.print_bgimages: false
print.print_colorspace:
print.print_command:
print.print_downloadfonts: true
print.print_duplex: 0
print.print_evenpages: true
print.print_in_color: true
print.print_margin_bottom: 0.5
print.print_margin_left: 0.5
print.print_margin_right: 0.5
print.print_margin_top: 0.5
print.print_oddpages: true
print.print_orientation: 0
print.print_page_delay: 50
print.print_pagedelay: 500
print.print_paper_data: 0
print.print_paper_height: 11.00
print.print_paper_name:
print.print_paper_size: 0
print.print_paper_size_type: 1
print.print_paper_size_unit: 0
print.print_paper_width: 8.50
print.print_plex_name:
print.print_printer:
print.print_resolution: 0
print.print_resolution_name:
print.print_reversed: false
print.print_scaling: 1.00
print.print_shrink_to_fit: true
print.print_to_file: false
print.print_unwriteable_margin_bottom: 57
print.print_unwriteable_margin_left: 25
print.print_unwriteable_margin_right: 25
print.print_unwriteable_margin_top: 25
privacy.sanitize.migrateFx3Prefs: true
security.disable_button.openCertManager: false
security.OCSP.signingCA: Builtin Object Token:IPS CLASE1 root
security.OCSP.URL:
security.warn_viewing_mixed: false
storage.vacuum.last.index: 1
storage.vacuum.last.places.sqlite: 1392281384

Accélération graphique
----------------------

Fenêtres avec accélération graphique: 3/3 OpenGL (OMTC)
ID du périphérique: 0x a29
ID du vendeur: 0x10de
Rendu WebGL: NVIDIA Corporation -- NVIDIA GeForce GT 330M OpenGL Engine
windowLayerManagerRemote: true
AzureCanvasBackend: quartz
AzureContentBackend: quartz
AzureFallbackCanvasBackend: none
AzureSkiaAccelerated: 0

JavaScript
----------

Ramasse-miettes incrémentiel: true

Accessibilité
-------------

Activée: false
Empêcher l'accessibilité: 0

Versions des bibliothèques
--------------------------

NSPR
Version minimale attendue: 4.10.2
Version utilisée: 4.10.2

NSS
Version minimale attendue: 3.15.4 Basic ECC
Version utilisée: 3.15.4 Basic ECC

NSSSMIME
Version minimale attendue: 3.15.4 Basic ECC
Version utilisée: 3.15.4 Basic ECC

NSSSSL
Version minimale attendue: 3.15.4 Basic ECC
Version utilisée: 3.15.4 Basic ECC

NSSUTIL
Version minimale attendue: 3.15.4
Version utilisée: 3.15.4
Hi,
I acheive to reproduce the issue:
I open a window with 5 tab with the following URLS:
http://www.ldlc.com/fiche/PB00125160.html
http://www.ldlc.com/fiche/PB00138303.html
http://www.ldlc.com/navigation/disquette/
http://www.akasa.com.tw/search.php?seed=AK-ICR-16
http://www.grosbill.com/4-selection_grosbill_1_44_mo_usb2_0-610672-informatique-_lecteur_disquette

I miniaturized the windows and wait...
I keeped a open window with about:memory open 
At the beginning:
660.15 MB (100.0%) -- explicit
├──226.45 MB (34.30%) ++ window-objects
├──153.09 MB (23.19%) ── heap-unclassified
├──122.12 MB (18.50%) ++ js-non-window

After 3 minutes:
806.94 MB (100.0%) -- explicit
├──281.80 MB (34.92%) ── heap-unclassified
├──228.98 MB (28.38%) ++ window-objects
├──122.20 MB (15.14%) ++ js-non-window

After 3 more minutes
946.89 MB (100.0%) -- explicit
├──419.83 MB (44.34%) ── heap-unclassified
├──237.91 MB (25.13%) ++ window-objects
├──121.38 MB (12.82%) ++ js-non-window

Once the window is un-miniaturized:
690.14 MB (100.0%) -- explicit
├──238.50 MB (34.56%) ++ window-objects
├──160.48 MB (23.25%) ── heap-unclassified
├──123.89 MB (17.95%) ++ js-non-window
Look at my last comments here:

https://bugzilla.mozilla.org/show_bug.cgi?id=975945

1. Open (new tab) tab bar.
2. Open a second window and go to https://parse.com/ site.
3. Drag to link the page to the 'tab bar' page.
4. Get ready to quit FF while you still can. (I recommend have Activity Monitor open showing 'real memory', so you can catch it in time. I quit FF when it pass 1GB. It starts from normal ~200MB).

- RAM goes through the roof in less then 15 sec (not minutes, SECONDS!). Total system freeze, not just FF.

Tried it on a different Mac with same result.
When I look in the Resource Monitor within Windows 7, after I start to notice the major drag within Mozilla FF 27, The Disk operations max out in the I/O very frequently.  While watching it, when it maxes out, a long list of C:\System Volume Information\.... populate trying to write to disk.
Ditto on both redhat fedora 20 (FF 27.0.1) and ubuntu 13.10 (FF 27.0.1).  I just saw FF using 2G ram (RES) on the fedora box (x86_64, 6GB ram) and memory usage of 12GB (RES) on the ubuntu box (x86_64, 48GB ram).  Both machines have swap turned off.  Note that this is RES, not VIRT.
@Idemay a.o., update referring to my previous posts:

OSX 10.7.5 on 2007 MBP/4Gb RAM, only iTunes + a couple of minor apps running:
FF26, plenty of pin tabs, use hovering between 900Mb-1.45Gb real mem.
(same machine: swap files as stable as previously)

OSX 10.6.8 on 2011 MBP/8Gb RAM, 10-15 apps running:
FF26, plenty of pin tabs and other tabs open, use hovering between 990Mb and 1.05Gb real mem.

In both cases: no slowdown, no clogging, no increase in memory use.

Yesterday, FF27 installed by mistake on a third machine (2010/(Gb)): FF and computer stalled within short period of time.

Just confirming that it is a FF*27* issue, unrelated to machines' mounted RAM.
Unfortunately, this bug has devolved into various people trying to outline completely different memory usage problems they have, under different circumstances, with different OSes, at different times (immediately or later), and even with different outcomes in some cases. The worst problem is that there are no clear steps to reproduce from a clean profile. Just saying "a couple of minor apps" or "heap-unclassified is growing rapidly" without steps to reproduce it isn't helpful. That makes this bugreport basically unactionable: we can't use it to fix anything (because even where comments do contain useful information, all the comments as a whole are contradictory and confusing, and it isn't at all clear that everyone who commented is experiencing the same issue).

If you're having persistent memory problems, please do the following:

0) try to figure out steps to reproduce: under what *specific* conditions / with what website are you experiencing this?
1) use the profile manager to create a new test profile ( https://support.mozilla.org/en-US/kb/profile-manager-create-and-remove-firefox-profiles ). Don't install any add-ons.
2) reproduce in the new profile. If you can't, it's possible it's an issue with one of the add-ons you have installed in the old profile. Install them one by one in the new profile until you can reproduce, and then systematically disable them until you find the culprit / combination of culprits.
3) file a *new bug* with detailed steps to reproduce based on the clean profile.


I'm marking this bugreport as incomplete per the above explanation.
Status: UNCONFIRMED → RESOLVED
Closed: 10 years ago
Flags: needinfo?(raghu.kuchi)
Resolution: --- → INCOMPLETE
@:Gijs Kruitbosch

Listen, you attitude is anything but encouraging, or even constructive. I'm an end-user of FF with a company and 3 Macs using FF intensely. I'm not a programmer nor a FF beta-tester, and I have no intention of becoming either one. Which does doesn't mean that I don't feel, call it a "social responsibility" of sorts, towards the Mozilla project. This is precisely why I reported on the Mozilla forums an anomaly that resulted in FF27 being non-functional as opposed to FF26. Someone there that sounded sensible and knowledgeable suggested I report on Bugzilla, which is exactly what I did after going through the process of registering etc.

Now, if you're not interested in end-users filing bug reports and doing the amount of testing their time schedules allows for, and if as it seems you enjoy decisional power at the Mozilla Foundation just say so with a big sign:

"ONLY PROGRAMMERS, FULLY COMMITTED BETA-TESTERS, NERDS ALLOWED."

and we won't bother you anymore.

Dag,
Hey Dag,

Text can be hard to divine tone from, but I didn't read anything discouraging or non-constructive in Gijs' comment.

On the contrary, he actually outlined the next steps for people who are still experiencing problems along these lines:

> If you're having persistent memory problems, please do the following:
> 
> 0) try to figure out steps to reproduce: under what *specific* conditions /
> with what website are you experiencing this?
> 1) use the profile manager to create a new test profile (
> https://support.mozilla.org/en-US/kb/profile-manager-create-and-remove-
> firefox-profiles ). Don't install any add-ons.
> 2) reproduce in the new profile. If you can't, it's possible it's an issue
> with one of the add-ons you have installed in the old profile. Install them
> one by one in the new profile until you can reproduce, and then
> systematically disable them until you find the culprit / combination of
> culprits.
> 3) file a *new bug* with detailed steps to reproduce based on the clean
> profile.
> 
> 

I agree with Gijs that the information in this bug wasn't specific enough to derive anything actionable.

Bugs with detailed STR on clean profiles are the ones that are easiest to reproduce, which makes them easier to test, which makes them easier to debug, which makes them easier to fix. That's what Gijs is asking for.

So Dag - I'm sorry if it came off as sounding harsh, or that your contribution was unwelcome. Your contribution is _most_ welcome - but we're just asking for more specificity so we can take action on it.

I hope that clears things up.
I also can verify this issue which reared its' head originally after updating from version 25 (or maybe 26) to version 27.0 (and now still there in version 27.0.l).

OS    10.6.8; 16GB of RAM
Mac Pro Dual Quad
Same exact windows and tabs open; using same exact profile in each test.

Reverting to version 25-- Firefox runs normally with no indication of any memory leak in "Real" memory (uses average of 2GB after hours of use).

Any install of version 27.x.x will exhibit a memory leak over time (varies it appears) and FF will eventually grab all available free RAM; in excess of 12GB, until there is no free RAM remaining.  FF will eventually freeze.

Originally seem in version 27.0, and after finding that ver. 25 was still good i repeated this test multiple times and found it completely reproducible.  Tried version 27.0.1 when it was released and found the same issue-- again totally reproducible.


Summary:
Same computer, OS and configuration
Same Profile
Same windows and tabs
version 25 works fine
version 27.x.x exhibits Real Memory leak until all is exhausted
Reproducible

HAS to be memory management changes in version 27.x.x

Hope this helps....
Tom:

Are you able to reliably reproduce this with a clean profile? Does about:memory give some indication of where the memory is going?

-Mike
Flags: needinfo?(techsupport)
Mike,

Will reinstall ver 27 and try that when I get a chance-- probably Sunday...  But-- the same exact profile folder is being used with both 25.x and 27.x so; technically, this should not be profile related.

Will advise...

Tom
Flags: needinfo?(techsupport)
Mike,

Relaunched ver. 27 with a clean new profile (did ultimately recover bookmarks from a backup so I could open many of the same windows).  With only 7 or 8 windows open, no tabs, real memory runs about 700MB.

If I stick Firefox in the background the memory usage starts tyo grow almost immediately- my first short test of less than an hour had the Real Mem jump up to 3.55GB!   I tried to gram another Allocation report from the about:memory window but by the time I got yo it Real mem went back down to 850 MB or so...  I did however, get a screen grab from Activity Monitor.

I'm about to head out for dinner and expect all available RAM will be gone when I return based on this first hour.  Hopefully I can grab the 'about' report, as it usually freezes hard when all RAM is used up.  I don't know of any way to run a dynamic reprt from within Firefox so that it constantly updates.

Tom
Mike--

Reproducible in virgin version 27.0.1 relaunch and virgin profile....

Okay.  Back home and we now have a freeze with FF taking all available free memory.  Showing 6.78 GB in Activity Monitor.  I grabbed a sample Process file, but the app is non-responsive.  Finally able (after 3-4 minutes) to get a "measure and Save" in the about:memory window but I watched Real Memory usage go down to 1.40 GB after this.  It may be when it is in the background, so I will leave it forward overnight and see if the same occurs tomorrow when I check.  

I do not know where to send the files I have grabbed (Activity Monitor "Sample Process" and about;memory measure file ....

Tom
Just finished running a 4 hour test with version 27.0.1 and a virgin test profile. (OS X 10.6.8 MC PRO 16 GB RAM)
As soon as Firefox app is hidden in the background (I use MultiXFinder; replaced old ASM, to control my open apps and what is current and showing) it starts gaining memory-- over this time period the rate was 540 MB / hr.
Started at 1.49 GB; four hours later Real Memory was 4.28GB and gaining.  When FF was returned to active app it was totally unresponsive for some 3 minutes or so.  I could not grab a report from about:memory as when control returned the Real Memory returned to 1.43 GB.

Fully reproducible with new profile, 8 windows open.


Have log of tests, Activity Monitor screen grabs, and some earlier about:memory reports.
-More-

Also I was able to Sample Process in Activity Monitor while the memory was still at 4.28 GB....

Just do not know where or to whom to send any of these files (if any)
(In reply to Tom Mulhall from comment #31)
> -More-
> 
> Also I was able to Sample Process in Activity Monitor while the memory was
> still at 4.28 GB....
> 
> Just do not know where or to whom to send any of these files (if any)

Can you please attach the files to this bug (keep in mind that it'll be public, so make sure to scrub any sensitive or private information from them), and we'll see if we can find something actionable in there.
Mike--

Is there any way to attach multiple files as i can only see a way to do one at a time?

Thanks,

Tom
Sadly, one at a time is all that our instance of Bugzilla is able to do right now. :/
Mike,

Sorry that I was not able to get a Memory report from within Firefox, but it becomes totally unresponsive when RAM grows large.  I -may- relaunch ver 27 and try to grab one at a lower leak level if that would help.

Tom
Just ran a short (2 min or so) movie of the memory growing in real time after relaunching FF 27.0.1 with test configuration and virgin profile.
(In reply to Tom Mulhall from comment #42)
> Created attachment 8388321 [details]
> Memory Leak Real Time movie; a few minutes after relaunch and sending to
> background
> 
> Just ran a short (2 min or so) movie of the memory growing in real time
> after relaunching FF 27.0.1 with test configuration and virgin profile.

Hey Tom - it's possible you're hitting bug 975945...
Mike,

I only have seven or so main web page windows open with standard sites that we go to all the time (indeed, version 25 running with 23 windows open. many with multiple tabs with no issues (the test set is a small subset of those open every day).

Also, as I noted, this does not occur in version 25.0.1.  Indeed, FF is open 24/7 on this particular machine (my Mac Pro desktop machine in the office); often tossed in the background or hidden and the issue only appeared at ver. 27.0 on on.  There are no other open empty tabs or otherwise.

Additionally, as I sit right now there is 10.98 GB of Free RAM (will upload this).  If ver 27 runs it will eat every bit of that additional 11 GB's over time.  I have seen the RAM for FF as high as 12+ GB.

Tom
Attachment uploading working again....
Mike,

I sincerely doubt that6 this has anything to do with Bug 975945; especially since bauermusic has settled that down to parse.com (when he has page open, issues persists, closed it clears).

BTW-- did a relaunch two days ago and did not notice that the auto-update had run, thus version 27.0.1 overwrote my version 25.x.  I was shocked to see FF frozen the next morning with 12.8 GB of Real RAM grabbed until I noticed that ver. 27 had once again taken over.  Of course, jumping back to ver. 25.x cleared the issue.

I note that Status shows "RESOLVED INCOMPLETE".  How is this resolved?

Thanks all...

Tom
(In reply to Tom Mulhall from comment #47)
> Mike,
> I note that Status shows "RESOLVED INCOMPLETE".  How is this resolved?
> 
> Thanks all...
> 
> Tom

It's marked RESOLVED INCOMPLETE because at this point, we don't have enough actionable data to determine what the problem is. :(
Is there anything else that I could capture (other than the about:memory page which does not respond when the issue is active)?

I hate the thought that every version going forward -may- have this memory leak issue....

Tom
Tom: yes, it would be awful if future versions didn't bring a solution to this problem, but I don't know what anyone else can do. You're not a programmer, and neither am I. You, myself and plenty of others have reported as much as they could, but that doesn't seem to be satisfying for the "bright ones" (no insult intended). Personally, I don't feel like reinstalling 27 only to find the computer frozen a few hours later. When 28 comes out, I'll try it once. Until then, I'm sticking to 26.
Hey Nathan - any idea what Tom / tingo could do to give us more actionable data?
Flags: needinfo?(nfroyd)
The "huge memory usage when minimized" thing sounds like a problem we've previously had with some graphics stuff, but that was supposed to be fixed.  Maybe this is something similar.
Yeah, this sounds like bug 937878, which is fixed in Firefox 28.

Tom, do you only see the runaway memory usage when you minimize the Firefox window to the dock, or do you see it when Firefox is merely running as a non-foreground application (but not minimized to the dock)?
Flags: needinfo?(nfroyd) → needinfo?(techsupport)
I am not on a mac and have experienced significant memory issues with FF since going to 27.  I am an IT Director but need some guidance on exactly what you need me to do to give you actionable data.  Most of the time I have seen a lock has been when a flash plugin has been activated from a page or using streaming content like Netflix.  The browser's memory spirals out of control rendering my entire pc unusable. I never saw this with previous FF versions.  The only reason I found this bug was when it first started happening, I googled to see if there was a known issue with FF 27.
Nathan -

I just read bug 937878, and that definitely fits the pattern that I was seeing.  Obviously, I cannot speak for others.
(In reply to sandy from comment #54)
> I am not on a mac and have experienced significant memory issues with FF
> since going to 27.  I am an IT Director but need some guidance on exactly
> what you need me to do to give you actionable data.  Most of the time I have
> seen a lock has been when a flash plugin has been activated from a page or
> using streaming content like Netflix.  The browser's memory spirals out of
> control rendering my entire pc unusable. I never saw this with previous FF
> versions.  The only reason I found this bug was when it first started
> happening, I googled to see if there was a known issue with FF 27.

After being chastised for not submitting a valid bug report in comment 22,

https://bugzilla.mozilla.org/show_bug.cgi?id=973776#c22

the bug was marked as RESOLVED INCOMPLETE.  So I would suggest following the
instructions in comment 22, and creating a new bug.
Nathan — not meaning to interfere, but considering that Tom hasn't answered yet: personaly, I never, ever, minimize FF (or anything else) to the Dock. I cannot tell whether FF had been in the background or not when I first reported the bug, but somehow I doubt that it was.
Nathan (et al)--

I am starting the testing on your question.  Will first leave FF 27.x just in the background and watch the memory in Activity Monitor.  If that is negative than I will minimize to the dock and check...

Tom
Flags: needinfo?(techsupport)
(In reply to jpabel from comment #55)
> Nathan -
> 
> I just read bug 937878, and that definitely fits the pattern that I was
> seeing.  Obviously, I cannot speak for others.

Confirmed that bug 937878 fixed my problem by installing FF 28 which was posted this morning.  Perhaps others could try installing FF 28 to see if it fixes their problems.
I will update now.
I'ive already tried FireFox 28 as Beta, as I thought it may work better, but seems the problem is still there, as only a few tabs takes more than 1 GB (Chrome can use like 700mb while having 2 tabs, and I believe that for our FireFox it's a joke).

GO GO FIREFOX!
Nathan-

Testing is proceeding.  

Did one hour with FF in background (remained on desktop but not forward app) with six windows open including this bugzilla page.  That is completed and now I am running with all windows minimized to the dock.

Astonishing difference!!! About 25 minutes to go but I am seeing a massive difference with the windows in the dock-- far worse.

Full report coming shortly.  I will try to open another window and leave the rest in the dock and try to grab a about:memory report if it does not freeze.

Tom
Okay!!!

FF in background; 6 windows open.  
Real Ram
Start           After 1 hour        Delta
472 MB             746 MB           274 MB

Virtual Mem
Start           After 1 hour        Delta
440 MB            686 MB            246 MB

FF all windows Minimized to Dock; same 6 windows.  
Real Ram
Start           After 1 hour        Delta
476 MB            2.867 GB         2.39 GB

Virtual Mem
Start           After 1 hour        Delta
429 MB            4.85 GB          4.42 GB

Conclusion:
Obviously, pages minimized to dock are causing the issue.  I opened a sibgle new window and was able to garb an about:memory report.  Will upload...

Tom
Tested with 6 windows minimized to the dock
Status: RESOLVED → REOPENED
Ever confirmed: true
Resolution: INCOMPLETE → ---
Whiteboard: [MemShrink]
about:memory grabbed after 1 hour with minimized windows
Tom, did you try upgrading to 28?  I think that's released now, and it may fix this problem.  Thanks!
Andrew,

Just downloaded and testing now-- looks VERY promising after 15 minutes!  I want to give it another 15 min. before I conclude anything.
Light years better after upgrading to 28.  Barely breaking 400MB for the same type of tabs and content that was hitting 1-2GB in 27.
Voilá!  

Installed and tested version 28.0 for the memory leak (with test windows minimized to dock) and am pleased to report that this bug appears to be DEAD!

Both Real Memory and Virtual Memory remain steady, or slightly decreased, after 30 minutes of testing (formerly would have jumped multi-gigabytes in this time frame).
Thanks all-- you guys are pure genius!

Tom
I guess I'll close this again.  Thanks for your investigation!
Status: REOPENED → RESOLVED
Closed: 10 years ago10 years ago
Resolution: --- → INCOMPLETE
Using FireFox 28 Beta, with Facebook, Gmail And YouTube (And this window):
RSS (RAM?): 480mb
Virtual: 1.1GB

AM I the only one?
Also using ABP and Flash-Aid.
Since Tom Mulhall provided the most data, let's make this bug about his problem. Which turned out to be a dup of bug 937878.

For those others who have been having problems on Mac, please try FF28 (it was just released). If that doesn't fix your problem (or you aren't on Mac) please file a new bug, *one per person*. Gijs' comment 22 was true -- when a bug's symptoms are vague and multiple people pile on, it makes it really hard for Firefox developers to keep track of what's happening.

If each separate person files a separate bug, the chances of the problem being fixed are higher. If multiple bugs end up being about the same problem, they can be dup'd together later on. Clear steps to reproduce are also extremely valuable, but one bug per person is the starting point for a useful outcome. Thanks!
Resolution: INCOMPLETE → DUPLICATE
Summary: Memory leak in Firefox 27.0.1 → Memory leak in Firefox 27.0.1 on Mac when minimized to dock
Confirming (a bit late, it's CEST here) that the bug appears to be gone, after running FF28 under 10.9.2 for almost six hours straight, with a YouTube compilation going nonstop for all that time. Actually, FF's memory use has even gone DOWN, to ± 750Mb.

But Nicholas, as much as I also appreciate Tom Mulhall's contribution, I don't think it's correct to rename the bug as you did: as stated yesterday, in my case the memory leak had nothing to do with minimizing windows.

Else, can anyone confirm that the bug is also gone under 10.6.8?
As I saw onky now, this bug related to mac only?
(In reply to yinonzadok from comment #75)
> As I saw onky now, this bug related to mac only?

Yes.

If you are seeing memory problems with Firefox 27 and before on Mac, please upgrade to 28 and see if doing so fixes the issues.

If you are seeing memory problems with Firefox 28 on non-Mac platforms, or you are still seeing problems with Firefox 28 on Mac, please open a new bug with information.
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: