1261315 - Even 64-bit version of FireFox (in e10s mode) is limited with maximum 8 GB per process?

Status: RESOLVED INCOMPLETE

(Core :: General, defect, P2)

Description

[User Dderss]

User Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:46.0) Gecko/20100101 Firefox/46.0
Build ID: 20160324011246

Steps to reproduce:

I was just browsing regularly in a big session for a prolonged time.


Actual results:

During this browsing performance has gradually degraded (but this is a separate issue), PluginContainer.exe has grown to have 8 GB footprint, and then it was abruptly yanked, and restarted as a new process. 

Visually it shows this way: most of tabs' icons got disappear with grey Earth icon instead, current tab's address bar also becomes blanks (with no history). However, the browser becomes fast again and tabs' icons get recovered if you active those tabs.

This bug is absolutely REPRODUCIBLE, happened to me thrice with the same pattern. However, crash report is not generated, so there is nothing to link to this bug report in this regard.


Expected results:

Normal continuous speedy operation with PluginContainer growing to be as big as needed -- whether it is 8 GB, 16 GB or even bigger value.

But before that, maybe Crash Reporter should be redone to catch up this type of crashes.

Comment 1

[Brad Lassey [:blassey] (use needinfo?)]

Andrew, do you have any suggestions for tracking this down?

Comment 2

[Andrew McCreight [:mccr8]]

(In reply to Brad Lassey [:blassey] (use needinfo?) from comment #1)
> Andrew, do you have any suggestions for tracking this down?

Not really. It must be some Windows-specific issue, because I've seen people complaining about a leaky Firefox using more than 10GB of memory.

Also, it would be good to know if this can be reproduced in non-e10s mode.

Comment 3

[User Dderss]

Important addition:

After Plugin-Container.exe had died and was automatically restarted without crashing the browser -- as described earlier -- I continued to use FireFox for another prolonger period of time. But this suddenly ended with the browser instantly dying without any prior signs or effects that could lead me to anticipate that. 

Even more strangely, after I started FireFox, there was no entry on that crash in "about:crashes". As I wrote in the previous post above, I can understand that crash reporter does not catch PluginContainer crashes, but I expected that it should catch main  FireFox.exe process crashes. 

The reason for that why it did not happen might be connected to the fact that Plugin-Container.exe has died earlier, what has somehow lead to crash reporter not being able to catch up the final dying of the browser itself.

Comment 4

[User Dderss]

(In reply to Andrew McCreight [:mccr8] from comment #2)
> (In reply to Brad Lassey [:blassey] (use needinfo?) from comment #1)
> > Andrew, do you have any suggestions for tracking this down?
> 
> Not really. It must be some Windows-specific issue, because I've seen people
> complaining about a leaky Firefox using more than 10GB of memory.
> 
> Also, it would be good to know if this can be reproduced in non-e10s mode.

I switched to using e10s mode fairly recently. My session was the same big. However, in non-e10s mode the browser is de-facto unusable once to click-to-load too many tabs in the session. It becomes so slow that there was no case where FF's footprint ever reached 9.5 GB (as in e10s mode: 1.5 GB of main FireFox.exe process and 8 GB Plugin-Container.exe). FireFox would crash long before that.

Besides, in this case the issue seems to be with Plugin-Container.exe, not FireFox.exe, so in non-e10s mode there would not be a way to test the issue as plugin-container.exe remains small there (unless something like Flash is running, but I do not have this evil any more), it will not grow to 8 GB.

Comment 5

[Nathan Froyd [:froydnj]]

Can you paste the contents of about:support as a comment or as an attachment to the bug?

Comment 6

[User Dderss]

"about:support" is empty for me for months already; nothing to paste here, there are only names of parameters, but nothing is filled out, values are empty.

Comment 7

[Nathan Froyd [:froydnj]]

(In reply to User Dderss from comment #6)
> "about:support" is empty for me for months already; nothing to paste here,
> there are only names of parameters, but nothing is filled out, values are
> empty.

That sounds extremely peculiar.  Can you provide a screenshot of the browser visiting about:support?  Are you using any addons?

Comment 8

[User Dderss]

Attachment: Empty About Support.png

Yes, I was surprised about this. "About:support" stays empty even in fresh-restarted safe-mode -- as seen in the attached screenshot.

Comment 9

[Tracy Walker [:tracy]]

Even in safe mode, add-ons can have flipped prefs that might affect this.  Can you try in a new profile?

Comment 10

[User Dderss]

In a new profile "about:support" works, but what preferences can possible damage this function that much that it stops working even in safe mode?

Comment 11

[Jim Mathies [:jimm]]

User, could you do us a favor and try displaying about:support, and then copy paste anything you have in your browser console here? We're wondering if there are errors there that might indicate what's going on.

Comment 12

[User Dderss]

Good guess James; thanks. Here is error that might be related to why about:support does not work in my profile:

1460507104572	addons.manager	WARN	Exception calling callback: TypeError: a.name is null (resource://gre/modules/Troubleshoot.jsm:240:13) JS Stack trace: extensions/</<@Troubleshoot.jsm:240:13 < extensions/<@Troubleshoot.jsm:237:7 < [e-mail]:179:5 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2527:9 < [e-mail]:375:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < [e-mail]:46:7 < [e-mail]:254:12 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2517:9 < [e-mail]:381:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < this.Experiments.PreviousExperimentProvider.prototype<[e-mail]:2179:7 < [e-mail]:254:12 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2517:9 < [e-mail]:381:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < [e-mail]:1032:7 < [e-mail]:254:12 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2517:9 < [e-mail]:381:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < [e-mail]:141:7 < [e-mail]:254:12 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2517:9 < [e-mail]:381:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < [e-mail]:653:7 < [e-mail]:254:12 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2517:9 < [e-mail]:381:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < [e-mail]:449:7 < [e-mail]:254:12 < AddonManagerInternal.getAddonsByTypes/<[e-mail]:2517:9 < [e-mail]:381:7 < AddonManagerInternal.getAddonsByTypes/<.nextObject/<@AddonManager.jsm:2522:11 < this.XPIProvider.getAddonsByTypes/<@XPIProvider.jsm:3956:7 < makeSafe/<@XPIProviderUtils.js:161:17 < [e-mail]:196:7 < asyncMap/</<@XPIProviderUtils.js:203:9 < [e-mail]:150:5 < [e-mail]:580:7 < this.AddonRepository.getCachedAddonByID<@AddonRepository.jsm:592:5

Comment 13

[Jim Mathies [:jimm]]

Dave, can make any sense of this console spew?

Comment 14

[Dave Townsend [:mossop]]

(In reply to Jim Mathies [:jimm] from comment #13)
> Dave, can make any sense of this console spew?

Apparently there is an add-on with no name installed and so we fail here: https://dxr.mozilla.org/mozilla-central/source/toolkit/modules/Troubleshoot.jsm#246

Comment 15

[Benjamin Smedberg]

Dderss, at this point I'm primarily interested in whether you can reproduce this kind of memory leak with no addons. If you can, I'm *very* interested in getting more diagnostic data, both about:memory dumps and about:support and perhaps also some minidumps.

If this is only reproducible in the profile with addons, the next step is probably to isolate particular addons until we find the one(s) that cause this, and either fix or disable those, but this doesn't need to block initial rollout of e10s which is only for users with no addons.

Putting up a patch to fix about:support is simple, but doesn't address the important part here.

Comment 16

[Jim Mathies [:jimm]]

It looks like you might be able to get about:support working by removing 'safe call'.

Comment 17

[User Dderss]

Thanks.

How can I delete this faulty entry from the list of add-ons/extensions/plugins?

In the past I could use normal editors (for XML and other types) or SQL editor to manage similar situations, but lately FireFox started to compress files with a non-standard file format that has twisted header, not allowing normal ways to unpack-edit-repack.

Comment 18

[User Dderss]

(If that add-on is called "Safe call", then I can not delete it via usual way since it is not listed through UI as, apparently, it has no name.)

Comment 19

[Benjamin Smedberg]

I filed/fixed the about:support issue in bug 1264725.

Comment 20

[User Dderss]

Thanks. I did not notice that the broken extension actually does have a name representation in UI, it was "null", so I finally uninstalled it.

So here is full "about:support" information (lots of extensions are turned off):

Сведения о приложении
---------------------

Имя: Firefox
Версия: 46.0b10
ID сборки: 20160411042519
Канал обновления: beta
User Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:46.0) Gecko/20100101 Firefox/46.0
Многопроцессные окна: 1/1 (Включены пользователем)
Безопасный Режим: false

Сообщения о падениях за последние 3 дня
---------------------------------------

Все сообщения о падениях (включая 3 ожидающих отправки сообщения в заданном диапазоне времени)

Расширения
----------

Имя: 1C:Enterprise Extension
Версия: 3.7
Включено: true
ID: [e-mail]

Имя: Доступ к Рутрекеру
Версия: 1.0.8
Включено: true
ID: [e-mail]

Имя: About sessionstore
Версия: 0.32.1-signed
Включено: true
ID: aboutsessionstore@dt

Имя: Adblock Plus
Версия: 2.7.2
Включено: true
ID: {d10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d}

Имя: Add-on Compatibility Reporter
Версия: 2.0.6.1-signed
Включено: true
ID: [e-mail]

Имя: Contextual Google Image Search
Версия: 2.1.1.1-signed
Включено: true
ID: {D46504D3-4959-4351-AED6-C7EA276DBB93}

Имя: Element Hiding Helper для Adblock Plus
Версия: 1.3.7
Включено: true
ID: [e-mail]

Имя: Firefox Hello
Версия: 1.2.6
Включено: true
ID: [e-mail]

Имя: Flagfox
Версия: 5.1.9
Включено: true
ID: {1018e4d6-728f-4b20-ad56-37578a4de76b}

Имя: Flash and Video Download
Версия: 1.81
Включено: true
ID: {bee6eb20-01e0-ebd1-da83-080329fb9a3a}

Имя: Google search link fix
Версия: 1.5.3
Включено: true
ID: jid0-XWJxt5VvCXkKzQK99PhZqAn7Xbg@jetpack

Имя: Greasemonkey
Версия: 3.7
Включено: true
ID: {e4a8a97b-f2ed-450b-b12d-ee082ba24781}

Имя: HighlightAll
Версия: 1.8
Включено: true
ID: {26FD1F83-A45B-4c74-AF5A-F2EE0EE4D691}

Имя: Multi-process staged rollout
Версия: 1.0
Включено: true
ID: [e-mail]

Имя: Open Link In Pinned Tab
Версия: 0.1.2.1-signed
Включено: true
ID: jid1-btKbvVtcxyAJ6Q@jetpack

Имя: Page Title in URL Bar
Версия: 5.2.4.1-signed
Включено: true
ID: [e-mail]

Имя: Pocket
Версия: 1.0
Включено: true
ID: [e-mail]

Имя: Prevent Out Of Virtual Memory Crashes
Версия: 1.0.12.1-signed
Включено: true
ID: PreventOutOfVirtualMemoryCrashes@ZXSpectrum

Имя: Quick Context Search
Версия: 1.5.2.1-signed
Включено: true
ID: quickcontextsearch@pf

Имя: Save Image in Folder
Версия: 1.3.18
Включено: true
ID: {5e594888-3e8e-47da-b2c6-b0b545112f84}

Имя: SQLite Manager
Версия: 0.8.3.1-signed
Включено: true
ID: [e-mail]

Имя: stealthy
Версия: 3.0.1
Включено: true
ID: [e-mail]

Имя: Tab Mix Plus
Версия: 0.4.2.2
Включено: true
ID: {dc572301-7619-498c-a57d-39143191b318}

Имя: Tab Scope
Версия: 1.6.2.1-signed
Включено: true
ID: [e-mail]

Имя: TinEye Reverse Image Search
Версия: 1.2.1
Включено: true
ID: [e-mail]

Имя: YouTube High Definition
Версия: 45.1
Включено: true
ID: {7b1bf0b6-a1b9-42b0-b75d-252036438bdc}

Имя: Youtube's Autoplay No More
Версия: 0.3
Включено: true
ID: jid1-XQEcUtyD5PwB8w@jetpack

Имя: ADB Helper
Версия: 0.8.7
Включено: false
ID: [e-mail]

Имя: Add Page Title To URL Bar
Версия: 1.0.21
Включено: false
ID: AddPageTitleToURLBar@ZXSpectrum

Имя: Aptana Debugger
Версия: 1.7.2
Включено: false
ID: [e-mail]

Имя: BitComet Video Downloader
Версия: 1.37
Включено: false
ID: {B042753D-F57E-4e8e-A01B-7379A6D4CEFB}

Имя: CacheViewer Continued
Версия: 0.9.4.1
Включено: false
ID: {30E08C68-889E-11E0-95EF-DA7E4824019B}

Имя: Check4Change
Версия: 1.9.8.1
Включено: false
ID: [e-mail]

Имя: Click to Play per-element
Версия: 0.3.3
Включено: false
ID: [e-mail]

Имя: DOM Inspector
Версия: 2.0.16
Включено: false
ID: [e-mail]

Имя: Download Status Bar
Версия: 13.4.2.2
Включено: false
ID: {6c28e999-e900-4635-a39d-b1ec90ba0c0f}

Имя: Download Statusbar
Версия: 0.9.10.1-signed
Включено: false
ID: {D4DD63FA-01E4-46a7-B6B1-EDAB7D6AD389}

Имя: Firebug
Версия: 2.0.16
Включено: false
ID: [e-mail]

Имя: FireFTP
Версия: 2.0.26
Включено: false
ID: {a7c6cf7f-112c-4500-a7ea-39801a327e5f}

Имя: geckoprofiler
Версия: 1.16.6
Включено: false
ID: jid0-edalmuivkozlouyij0lpdx548bc@jetpack

Имя: GitHub Extension Installer
Версия: 1.7.0
Включено: false
ID: {86054B0A-BD85-42F9-8E58-8794EC6F6EA1}

Имя: HTTPS-Everywhere
Версия: 5.1.6
Включено: false
ID: [e-mail]

Имя: JavaScript Deobfuscator
Версия: 2.0.2
Включено: false
ID: [e-mail]

Имя: Log Helper
Версия: 0.6.1-signed
Включено: false
ID: [e-mail]

Имя: Media Hint
Версия: 0.1.13.1
Включено: false
ID: mediahint@jetpack

Имя: Menu Wizard
Версия: 4.05
Включено: false
ID: s3menu@wizard

Имя: Multiple Tab Handler (Обработка группы вкладок)
Версия: 0.8.2015113001
Включено: false
ID: [e-mail]

Имя: NoScript
Версия: 2.9.0.11
Включено: false
ID: {73a6fe31-595d-460b-a920-fcc0f8843232}

Имя: NoSquint
Версия: 2.1.9.1-signed
Включено: false
ID: [e-mail]

Имя: Suspend background tabs
Версия: 0.0.1a
Включено: false
ID: {1474867A-4655-11E3-BC1F-0DB06188709B}

Имя: Valence
Версия: 0.3.5
Включено: false
ID: [e-mail]

Имя: Wiktionary and Google Translate
Версия: 7.5
Включено: false
ID: [e-mail]

Графика
-------

Аппаратно ускоренные окна: 1/1 Direct3D 11 (OMTC)
Асинхронное панорамирование/зум: включён ввод колесиком
Версия драйвера: 15.301.1201.0
Визуализатор WebGL: Google Inc. -- ANGLE (AMD Radeon HD 6900 Series Direct3D11 vs_5_0 ps_5_0)
Дата разработки драйвера: 12-23-2015
Драйвера видеокарты: aticfx64 aticfx64 aticfx64 aticfx32 aticfx32 aticfx32 atiumd64 atidxx64 atidxx64 atiumdag atidxx32 atidxx32 atiumdva atiumd6a atitmm64
Код подсистемы: 00000000
Код производителя: 0x1002
Код устройства: 0x6718
Объём видеопамяти: 2048
Описание видеокарты: AMD Radeon HD 6900 Series
Параметры ClearType: Gamma: 2200 Pixel Structure: R ClearType Level: 100 Enhanced Contrast: 50
Поддержка аппаратного декодирования H264: Yes
Direct2D включён: true
DirectWrite включён: true (6.2.9200.17568)
GPU №2 активен: false
windowLayerManagerRemote: true
AzureCanvasBackend: direct2d 1.1
AzureContentBackend: direct2d 1.1
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

Важные изменённые настройки
---------------------------

accessibility.blockautorefresh: true
accessibility.force_disabled: 1
accessibility.typeaheadfind: true
accessibility.typeaheadfind.flashBar: 0
browser.cache.disk.capacity: 1048576
browser.cache.disk.filesystem_reported: 1
browser.cache.disk.hashstats_reported: 1
browser.cache.disk.smart_size_cached_value: 640000
browser.cache.disk.smart_size.enabled: false
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 3
browser.display.background_color: #CCCCCC
browser.download.folderList: 2
browser.download.importedFromSqlite: true
browser.download.manager.alertOnEXEOpen: true
browser.download.manager.showAlertOnComplete: false
browser.download.manager.showWhenStarting: false
browser.history_expire_days_min: 99999
browser.history_expire_days.mirror: 99999
browser.link.open_newwindow.restriction: 0
browser.places.createdSmartBookmarks: true
browser.places.importBookmarksHTML: false
browser.places.importDefaults: false
browser.places.leftPaneFolderId: -1
browser.places.migratePostDataAnnotations: false
browser.places.smartBookmarksVersion: 7
browser.places.updateRecentTagsUri: false
browser.search.openintab: true
browser.search.param.yahoo-fr: megaup
browser.search.param.yahoo-fr-cjkt: megaup
browser.search.useDBForOrder: true
browser.sessionstore.max_serialize_back: 30
browser.sessionstore.max_tabs_undo: 50
browser.sessionstore.max_windows_undo: 20
browser.sessionstore.upgradeBackup.latestBuildID: 20160411042519
browser.startup.homepage: about:home
browser.startup.homepage_override.buildID: 20160411042519
browser.startup.homepage_override.mstone: 46.0
browser.tabs.closeWindowWithLastTab: false
browser.tabs.insertRelatedAfterCurrent: false
browser.tabs.remote.autostart: true
browser.tabs.remote.force-enable: true
browser.tabs.selectOwnerOnClose: false
browser.tabs.warnOnClose: false
browser.urlbar.suggest.searches: true
browser.urlbar.userMadeSearchSuggestionsChoice: true
dom.apps.reset-permissions: true
dom.event.clipboardevents.enabled: false
dom.max_chrome_script_run_time: 40
dom.max_script_run_time: 0
dom.mozApps.used: true
dom.w3c_touch_events.expose: false
extensions.lastAppVersion: 46.0
font.default: sans-serif
font.internaluseonly.changed: false
font.language.group: x-cyrillic
font.name.sans-serif.ja: Arial Unicode MS
font.name.serif.x-cyrillic: Arial
font.name.serif.x-western: Tahoma
general.useragent.extra.firefox: Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/46.0.2490.80 Safari/537.36
general.useragent.extra.microsoftdotnet: (.NET CLR 3.5.30729)
gfx.crash-guard.d3d11layers.appVersion: 46.0
gfx.crash-guard.d3d11layers.deviceID: 0x6718
gfx.crash-guard.d3d11layers.driverVersion: 15.301.1201.0
gfx.crash-guard.d3d11layers.feature-d2d: true
gfx.crash-guard.d3d11layers.feature-d3d11: true
gfx.crash-guard.glcontext.gfx.driver-init.direct3d11-angle: true
gfx.crash-guard.glcontext.gfx.driver-init.webgl-angle: true
gfx.crash-guard.glcontext.gfx.driver-init.webgl-angle-force-d3d11: false
gfx.crash-guard.glcontext.gfx.driver-init.webgl-angle-force-warp: false
gfx.crash-guard.glcontext.gfx.driver-init.webgl-angle-try-d3d11: true
gfx.crash-guard.status.d3d11layers: 2
gfx.crash-guard.status.d3d9video: 2
gfx.crash-guard.status.glcontext: 2
gfx.direct3d.last_used_feature_level_idx: 0
gfx.direct3d.prefer_10_1: true
gfx.driver-init.appVersion: 42.0
gfx.driver-init.deviceID: 0x6718
gfx.driver-init.driverVersion: 15.201.1151.1005
gfx.driver-init.feature-d2d: true
gfx.driver-init.feature-d3d11: true
gfx.driver-init.status: 2
keyword.URL: http://search.yahoo.com/search?ei=utf-8&fr=megaup&p=
media.gmp-eme-adobe.abi: x86_64-msvc-x64
media.gmp-eme-adobe.lastUpdate: 1458390462
media.gmp-eme-adobe.version: 17
media.gmp-gmpopenh264.abi: x86_64-msvc-x64
media.gmp-gmpopenh264.lastUpdate: 1451415256
media.gmp-gmpopenh264.version: 1.5.3
media.gmp-manager.buildID: 20160411042519
media.gmp-manager.lastCheck: 1460658748
media.hardware-video-decoding.failed: false
media.video-queue.default-size: 15
network.cookie.prefsMigrated: true
network.http.keep-alive.timeout: 900
network.http.max-persistent-connections-per-server: 11
network.predictor.cleaned-up: true
places.database.lastMaintenance: 1460686872
places.history.expiration.max_pages: 9999999
places.history.expiration.transient_current_max_pages: 9999999
places.history.expiration.transient_optimal_database_size: 139549408
places.last_vacuum: 1300678081
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
plugin.state.flash: 0
plugin.state.npatgpc: 1
plugin.state.npctrl: 0
plugin.state.npqtplugin: 2
plugin.state.npvizzedrgr: 0
plugin.state.npvlc: 2
plugin.state.npwlpg: 1
print.print_bgcolor: false
print.print_bgimages: false
print.print_command:
print.print_downloadfonts: true
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_pagedelay: 500
print.print_paper_data: 0
print.print_paper_height: 11,00
print.print_paper_size_type: 1
print.print_paper_size_unit: 0
print.print_paper_width: 8,50
print.print_printer: Brother MFC-990CW Printer
print.print_reversed: false
print.print_scaling: 1,00
print.print_shrink_to_fit: true
print.print_to_file: false
print.printer_Brother_MFC-990CW_Printer.print_bgcolor: false
print.printer_Brother_MFC-990CW_Printer.print_bgimages: false
print.printer_Brother_MFC-990CW_Printer.print_command:
print.printer_Brother_MFC-990CW_Printer.print_downloadfonts: true
print.printer_Brother_MFC-990CW_Printer.print_edge_bottom: 0
print.printer_Brother_MFC-990CW_Printer.print_edge_left: 0
print.printer_Brother_MFC-990CW_Printer.print_edge_right: 0
print.printer_Brother_MFC-990CW_Printer.print_edge_top: 0
print.printer_Brother_MFC-990CW_Printer.print_evenpages: true
print.printer_Brother_MFC-990CW_Printer.print_footercenter:
print.printer_Brother_MFC-990CW_Printer.print_footerleft: &PT
print.printer_Brother_MFC-990CW_Printer.print_footerright: &D
print.printer_Brother_MFC-990CW_Printer.print_headercenter:
print.printer_Brother_MFC-990CW_Printer.print_headerleft: &T
print.printer_Brother_MFC-990CW_Printer.print_headerright: &U
print.printer_Brother_MFC-990CW_Printer.print_in_color: true
print.printer_Brother_MFC-990CW_Printer.print_margin_bottom: 0.5
print.printer_Brother_MFC-990CW_Printer.print_margin_left: 0.5
print.printer_Brother_MFC-990CW_Printer.print_margin_right: 0.5
print.printer_Brother_MFC-990CW_Printer.print_margin_top: 0.5
print.printer_Brother_MFC-990CW_Printer.print_oddpages: true
print.printer_Brother_MFC-990CW_Printer.print_orientation: 0
print.printer_Brother_MFC-990CW_Printer.print_pagedelay: 500
print.printer_Brother_MFC-990CW_Printer.print_paper_data: 0
print.printer_Brother_MFC-990CW_Printer.print_paper_height: 11,00
print.printer_Brother_MFC-990CW_Printer.print_paper_size_type: 0
print.printer_Brother_MFC-990CW_Printer.print_paper_size_unit: 1
print.printer_Brother_MFC-990CW_Printer.print_paper_width: 8,50
print.printer_Brother_MFC-990CW_Printer.print_reversed: false
print.printer_Brother_MFC-990CW_Printer.print_scaling: 1,00
print.printer_Brother_MFC-990CW_Printer.print_shrink_to_fit: true
print.printer_Brother_MFC-990CW_Printer.print_to_file: false
print.printer_Brother_MFC-990CW_Printer.print_unwriteable_margin_bottom: 0
print.printer_Brother_MFC-990CW_Printer.print_unwriteable_margin_left: 0
print.printer_Brother_MFC-990CW_Printer.print_unwriteable_margin_right: 0
print.printer_Brother_MFC-990CW_Printer.print_unwriteable_margin_top: 0
print.printer_Brother_MFC-990CW.print_bgcolor: false
print.printer_Brother_MFC-990CW.print_bgimages: false
print.printer_Brother_MFC-990CW.print_command:
print.printer_Brother_MFC-990CW.print_downloadfonts: true
print.printer_Brother_MFC-990CW.print_edge_bottom: 0
print.printer_Brother_MFC-990CW.print_edge_left: 0
print.printer_Brother_MFC-990CW.print_edge_right: 0
print.printer_Brother_MFC-990CW.print_edge_top: 0
print.printer_Brother_MFC-990CW.print_evenpages: true
print.printer_Brother_MFC-990CW.print_footercenter:
print.printer_Brother_MFC-990CW.print_footerleft: &PT
print.printer_Brother_MFC-990CW.print_footerright: &D
print.printer_Brother_MFC-990CW.print_headercenter:
print.printer_Brother_MFC-990CW.print_headerleft: &T
print.printer_Brother_MFC-990CW.print_headerright: &U
print.printer_Brother_MFC-990CW.print_in_color: true
print.printer_Brother_MFC-990CW.print_margin_bottom: 0.5
print.printer_Brother_MFC-990CW.print_margin_left: 0.5
print.printer_Brother_MFC-990CW.print_margin_right: 0.5
print.printer_Brother_MFC-990CW.print_margin_top: 0.5
print.printer_Brother_MFC-990CW.print_oddpages: true
print.printer_Brother_MFC-990CW.print_orientation: 0
print.printer_Brother_MFC-990CW.print_pagedelay: 500
print.printer_Brother_MFC-990CW.print_paper_data: 0
print.printer_Brother_MFC-990CW.print_paper_height: 11,00
print.printer_Brother_MFC-990CW.print_paper_size_type: 0
print.printer_Brother_MFC-990CW.print_paper_size_unit: 1
print.printer_Brother_MFC-990CW.print_paper_width: 8,50
print.printer_Brother_MFC-990CW.print_reversed: false
print.printer_Brother_MFC-990CW.print_scaling: 1,00
print.printer_Brother_MFC-990CW.print_shrink_to_fit: true
print.printer_Brother_MFC-990CW.print_to_file: false
print.printer_Brother_MFC-990CW.print_unwriteable_margin_bottom: 0
print.printer_Brother_MFC-990CW.print_unwriteable_margin_left: 0
print.printer_Brother_MFC-990CW.print_unwriteable_margin_right: 0
print.printer_Brother_MFC-990CW.print_unwriteable_margin_top: 0
print.printer_Microsoft_XPS_Document_Writer.print_bgcolor: false
print.printer_Microsoft_XPS_Document_Writer.print_bgimages: false
print.printer_Microsoft_XPS_Document_Writer.print_colorspace:
print.printer_Microsoft_XPS_Document_Writer.print_command:
print.printer_Microsoft_XPS_Document_Writer.print_downloadfonts: true
print.printer_Microsoft_XPS_Document_Writer.print_duplex: 7929961
print.printer_Microsoft_XPS_Document_Writer.print_edge_bottom: 0
print.printer_Microsoft_XPS_Document_Writer.print_edge_left: 0
print.printer_Microsoft_XPS_Document_Writer.print_edge_right: 0
print.printer_Microsoft_XPS_Document_Writer.print_edge_top: 0
print.printer_Microsoft_XPS_Document_Writer.print_evenpages: true
print.printer_Microsoft_XPS_Document_Writer.print_footercenter:
print.printer_Microsoft_XPS_Document_Writer.print_footerleft: &PT
print.printer_Microsoft_XPS_Document_Writer.print_footerright: &D
print.printer_Microsoft_XPS_Document_Writer.print_headercenter:
print.printer_Microsoft_XPS_Document_Writer.print_headerleft: &T
print.printer_Microsoft_XPS_Document_Writer.print_headerright: &U
print.printer_Microsoft_XPS_Document_Writer.print_in_color: true
print.printer_Microsoft_XPS_Document_Writer.print_margin_bottom: 0.5
print.printer_Microsoft_XPS_Document_Writer.print_margin_left: 0.5
print.printer_Microsoft_XPS_Document_Writer.print_margin_right: 0.5
print.printer_Microsoft_XPS_Document_Writer.print_margin_top: 0.5
print.printer_Microsoft_XPS_Document_Writer.print_oddpages: true
print.printer_Microsoft_XPS_Document_Writer.print_orientation: 0
print.printer_Microsoft_XPS_Document_Writer.print_page_delay: 50
print.printer_Microsoft_XPS_Document_Writer.print_pagedelay: 500
print.printer_Microsoft_XPS_Document_Writer.print_paper_data: 0
print.printer_Microsoft_XPS_Document_Writer.print_paper_height: 11,00
print.printer_Microsoft_XPS_Document_Writer.print_paper_name:
print.printer_Microsoft_XPS_Document_Writer.print_paper_size_type: 0
print.printer_Microsoft_XPS_Document_Writer.print_paper_size_unit: 1
print.printer_Microsoft_XPS_Document_Writer.print_paper_width: 8,50
print.printer_Microsoft_XPS_Document_Writer.print_plex_name:
print.printer_Microsoft_XPS_Document_Writer.print_resolution: 6488161
print.printer_Microsoft_XPS_Document_Writer.print_resolution_name:
print.printer_Microsoft_XPS_Document_Writer.print_reversed: false
print.printer_Microsoft_XPS_Document_Writer.print_scaling: 1,00
print.printer_Microsoft_XPS_Document_Writer.print_shrink_to_fit: true
print.printer_Microsoft_XPS_Document_Writer.print_to_file: false
print.printer_Microsoft_XPS_Document_Writer.print_unwriteable_margin_bottom: 0
print.printer_Microsoft_XPS_Document_Writer.print_unwriteable_margin_left: 0
print.printer_Microsoft_XPS_Document_Writer.print_unwriteable_margin_right: 0
print.printer_Microsoft_XPS_Document_Writer.print_unwriteable_margin_top: 0
privacy.clearOnShutdown.cache: false
privacy.clearOnShutdown.cookies: false
privacy.cpd.cache: false
privacy.cpd.cookies: false
privacy.donottrackheader.enabled: true
privacy.item.cache: false
privacy.sanitize.migrateClearSavedPwdsOnExit: true
privacy.sanitize.migrateFx3Prefs: true
security.crl.autoupdate.dayCnt.: 1
security.crl.autoupdate.enable.: false
security.crl.autoupdate.freqCnt.: 1
security.crl.autoupdate.nextInstant.: 1370817941000000
security.crl.autoupdate.timingType.: 1
security.crl.autoupdate.url.: http://crl.usertrust.com/USERTrustLegacySecureServerCA.crl
security.disable_button.openCertManager: false
security.OCSP.signingCA: Builtin Object Token:IPS CLASE1 root
security.OCSP.URL:
security.sandbox.content.tempDirSuffix: {e10a82f6-3085-44e6-a3fc-90b072c8d55a}
security.ssl.errorReporting.automatic: true
security.warn_viewing_mixed: false
security.warn_viewing_mixed.show_once: false
services.sync.declinedEngines:
services.sync.engine.greasemonkey: true
storage.vacuum.last.index: 1
storage.vacuum.last.places.sqlite: 1458352681

Важные заблокированные настройки
--------------------------------

JavaScript
----------

Инкрементальный GC: true

Поддержка доступности
---------------------

Активирована: false
Отключение поддержки доступности: 1

Версии библиотек
----------------

NSPR
Ожидаемая минимальная версия: 4.12
Используемая версия: 4.12

NSS
Ожидаемая минимальная версия: 3.22.3 Basic ECC
Используемая версия: 3.22.3 Basic ECC

NSSSMIME
Ожидаемая минимальная версия: 3.22.3 Basic ECC
Используемая версия: 3.22.3 Basic ECC

NSSSSL
Ожидаемая минимальная версия: 3.22.3 Basic ECC
Используемая версия: 3.22.3 Basic ECC

NSSUTIL
Ожидаемая минимальная версия: 3.22.3
Используемая версия: 3.22.3

Экспериментальные возможности
-----------------------------

Имя: Multi-process Firefox A/B Test 46.1
ID: [e-mail]
Описание: Measure the effect of multi-process Firefox
Активно: false
Конечная дата: 1458432146045
Домашняя страница:
Ветка: user-disabled

Имя: Multi-process Firefox A/B Test 45.2
ID: [e-mail]
Описание: Measure the effect of multi-process Firefox
Активно: false
Конечная дата: 1455978937449
Домашняя страница:
Ветка: user-disabled

Имя: Flash Protected-Mode Testing
ID: [e-mail]
Описание: Measuring the effect of Flash protected mode on crashes, hangs, and other browser jank.
Активно: false
Конечная дата: 1419961702706
Домашняя страница:
Ветка: control

Имя: New Tab Data
ID: [e-mail]
Описание: An experiment to analyze the data on about:newtab, see bug 1062708.
Активно: false
Конечная дата: 1419358913487
Домашняя страница:
Ветка: null

Имя: tile switcher
ID: [e-mail]
Описание: An add-on which switches the location of NewTab tiles.
Активно: false
Конечная дата: 1399221372238
Домашняя страница:
Ветка: null

Comment 21

[Jim Mathies [:jimm]]

For some indication of which of these addons might be causing problems you can check about:performance.

Maybe you could check that and disable anything dragging your system down to see if it helps with the memory consumption problems.

Comment 22

[User Dderss]

I have ran "about:performance" for hours, but it did not expose any issues with extensions.

My question is, why FireFox.exe does not register some types of crashes like those two types as described here. It would be much easier to diagnose, if it could, and there were crash reports. Running profiler with such giant memory footprint and not quite predictable time of crash other than PluginContainer reaching about 8 GB would generate giant file, and the time of second type of crash is not predictable, so Profiler is useless there, unless I had a terabyte to spare just for Profiler's output data.

Comment 23

[Jim Mathies [:jimm]]

(In reply to User Dderss from comment #22)
> I have ran "about:performance" for hours, but it did not expose any issues
> with extensions.
> 
> My question is, why FireFox.exe does not register some types of crashes like
> those two types as described here. It would be much easier to diagnose, if
> it could, and there were crash reports. 

It's possible the application is running into problems generating the crash reports due to memory constraint issues.

I would suggest disabling some of these remaining addons to see if it improves the situation. My guess is your memory consumption problems are primarily tied to one of these addons.

> Имя: 1C:Enterprise Extension
> Имя: Доступ к Рутрекеру
> Имя: About sessionstore
> Имя: Flash and Video Download
> Имя: Prevent Out Of Virtual Memory Crashes
> Имя: Quick Context Search
> Имя: stealthy
> Имя: Tab Mix Plus
> Имя: Tab Scope

Comment 24

[Jim Mathies [:jimm]]

Also, I wonder if 

> Имя: Prevent Out Of Virtual Memory Crashes

might be getting in the way of generating crash reports when the browser ooms?

Comment 25

[Benjamin Smedberg]

As noted earlier: we're not going to block on this without evidence that it can be reproduced with no addons.

Comment 26

[User Dderss]

(In reply to Jim Mathies [:jimm] from comment #24)
> Also, I wonder if 
> 
> > Имя: Prevent Out Of Virtual Memory Crashes
> 
> might be getting in the way of generating crash reports when the browser
> ooms?

This extension only works in 32-bit mode (as can be seen in install.rdf).

I will try to run PluginContainer.exe to 9 GB size if it will be yanked or not; as far as I understand, this should not happen. It that will be the case, then one extensions will be culprit; lets see.

Comment 27

[User Dderss]

Attachment: memory-report.json.gz

Thanks everybody for help.

With a different set of extensions (and slightly different version of FireFox) I now managed to surpass the 8 GB  size of PluginContainer. 

Now it has 14.5 GB size, plus 1.5 GB is size of FireFox.exe.

I wonder, though, is this size normal? I have tabs with three videos open, but the buggering only goes that far, and even fully buffered those videos should not take those crazy gigabytes. 

Can somebody please look at the memory report to see if it normal or not? Thank you in advance.

Comment 28

[User Dderss]

Attachment: memory-reportTwoHoursLater.json.gz

Couple more hours of browsing/video viewing *without* total number of active tabs increasing -- I always close opened tabs -- and PluginContainer has grown to 18.5 GB (and rapidly increasing), plus the usual 1.5 GB size of FireFox.exe.

Why FireFox blows up? Can it be understood from the memory report?

Comment 29

[The 8472]

I see hundreds of detached/ghost windows and gigabytes of media data - probably belonging to the detached windows - in that output. So something is holding onto window objects, probably an addon.

Sadly content processes don't break down the usage per addon like the parent process does.

maybe you could try with e10s off and see if the about:memory output shows anything suspicious under tiny -> addons.

Alternatively you could try binary search, disable 50% of your addons, see if that fixes the issue, if yes then the flaw is in the disabled half, if no then it's still in the active half. enable another 50% of the suspect 50%, rinse, repeat until you have found the culprit.

And you could also try nightly, it would be a wasted effort if the issue has been fixed in the meantime.

Comment 30

[Benjamin Smedberg]

> Sadly content processes don't break down the usage per addon like the parent process does.

mccr8, njn, do you know if there's a bug on file for this?

Comment 31

[Nathan Froyd [:froydnj]]

(In reply to Benjamin Smedberg  [:bsmedberg] from comment #30)
> > Sadly content processes don't break down the usage per addon like the parent process does.
> 
> mccr8, njn, do you know if there's a bug on file for this?

I believe this is bug 1268148.

Comment 32

[Nicholas Nethercote [inactive]]

(In reply to Nathan Froyd [:froydnj] from comment #31)
> > > Sadly content processes don't break down the usage per addon like the parent process does.
> 
> I believe this is bug 1268148.

That's correct. I heard about this shortcoming for the first time only yesterday.

Comment 33

[User Dderss]

Attachment: FF 47 -- plugincontained has died.png

(In reply to The 8472 from comment #29)
> I see hundreds of detached/ghost windows and gigabytes of media data -
> probably belonging to the detached windows - in that output. So something is
> holding onto window objects, probably an addon.

Thanks, The 8472.

I never use windows, so the giant heap of detached garbage you saw belongs to (closed) tabs.

Now I have new version of FireFox, but still the same issue.

This time I have opened dozens of Facebook videos (versus lots of YouTube videos), had PluginContainer blow up and die, and, after a while, it has died second time with only relatively small size -- as seen in the screenshot, all tabs' labels and content is lost after this, and task manager shows that PluginContainer's size at the time of its death was relatively small.

The difference with FireFox 46 is that the browser did not die altogether during such second PluginContainer crash.

As to extensions that I use, the only probably culprit that can hold on to the huge left-overs is Tab Mix Plus, because it has backward-forward page history management.

As to plugins I have, there are only three active: Open H264 by Cisco (but the last time I used was years ago), PrimeTime content management by Adobe (I do not use Adobe products online for a long time already), and Widevine CDM by Google (I used Google products online, but FF blows-ups happen even if I do not at all).

Comment 34

[User Dderss]

By the way, this PluginContainer's death might be or might be not depend on Bug #1130968 (a confirmed bug), which is about FF's graphics renderer failing in case of huge pages -- like Facebook page with big number tiled videos icons that I was browsing through before PluginContainer has died in the latest case.

This is not a firm connection, though, since previous deaths of PluginContainers, mentioned earlier in discussion of this bug, were not connected to huge single pages (and, hence, to graphics rendering of them).

Comment 35

[Abe - QA (:Abe_LV)]

Nathan/Nicholas, any comment ? thanks

Comment 36

[Nathan Froyd [:froydnj]]

(In reply to Abe - QA from comment #35)
> Nathan/Nicholas, any comment ? thanks

You will have to be more specific than that.  I don't have much to add beyond what's already been said and especially comment 25.

Comment 37

[User Dderss]

Nathaniel, #comment 25 does not help in this case since if the issue was with a pure clean FF in perfect condition, then it would have been noticed and fixed right away, much earlier. Besides FF becoming really, really slow (reacting to input with delays, showing spinner for a long time) blowing up the size of Plugin Container in "safe mode" does not cause crashes.

This is why the discussion went to profiler, though the debugging has turned out to be useless. 

Also FireFox' PluginContainer dying -- and, of course, without CrashReporter having captured anything -- supposedly to an (approved) extension like TabMixPlus might be a bug in itself. Why and how an extension can corrupt the session so badly? Should not it be impossible, the more so in e10s mode?

Thank you in advance.

Comment 38

[Nathan Froyd [:froydnj]]

(In reply to User Dderss from comment #37)
> Also FireFox' PluginContainer dying -- and, of course, without CrashReporter
> having captured anything -- supposedly to an (approved) extension like
> TabMixPlus might be a bug in itself. Why and how an extension can corrupt
> the session so badly? Should not it be impossible, the more so in e10s mode?

Extensions run with the same privileges as the UI itself (this is called "chrome" or "browser chrome").  That means extensions are free to do a whole host of useful things, but they are also free to do a whole host of dangerous things, like leaking windows and everything inside of those windows, which is what sounds like is going on here.  We can't prevent things like that, e10s or not.

I misread comment 25; it says that we're not going to block the release of e10s based on this bug unless we can show that it happens without any addons.  I read it as we're not going to concern ourselves with this bug unless we can show that it happens without any addons...my mistake.

That being said, since the problem doesn't reproduce in safe mode (which disables addons), it seems likely that addons are at fault.  Comment 33 suggests that Tab Mix Plus might be the cause of this leak, but it's not clear to me that the set of addons has been whittled down to reproduce the problem *only* with Tab Mix Plus.  Are you able to do that?

Comment 39

[Nathan Froyd [:froydnj]]

(In reply to Nathan Froyd [:froydnj] from comment #38)
> That being said, since the problem doesn't reproduce in safe mode (which
> disables addons), it seems likely that addons are at fault.  Comment 33
> suggests that Tab Mix Plus might be the cause of this leak, but it's not
> clear to me that the set of addons has been whittled down to reproduce the
> problem *only* with Tab Mix Plus.  Are you able to do that?

If we are able to reproduce the problem with a single addon, would the GC or CC log at the point when things are becoming sluggish tell us anything useful about what's creating ghost windows, or what's holding on to the ghost windows?

Comment 40

[The 8472]

(In reply to User Dderss from comment #33)
> I never use windows, so the giant heap of detached garbage you saw belongs
> to (closed) tabs.

They should not exist. It's likely an extension causing a leak.


> This time I have opened dozens of Facebook videos (versus lots of YouTube
> videos), had PluginContainer blow up and die, and, after a while, it has
> died second time with only relatively small size -- as seen in the
> screenshot, all tabs' labels and content is lost after this, and task
> manager shows that PluginContainer's size at the time of its death was
> relatively small.

That crash might be an unrelated issue to the leak. Check about:crashes if you have a crash report for it.


> As to extensions that I use, the only probably culprit that can hold on to
> the huge left-overs is Tab Mix Plus, because it has backward-forward page
> history management.

Any addon can cause a leak in subtle ways, it does not have to be tab-related in an obvious way.

If you can reproduce the ghost windows in a reliable way you could do a population bisection (testing 1/2, then 1/4th, then 1/8th ...) of your addons to figure out which causes the issue.

Comment 41

[User Dderss]

Thanks for all the replies. I will try to test with only TabMixPlus and other combinations to single out the extension that might cause this. Another culprit maybe Greasemonkey, but lets see. In general I have noticed that those crashes happen with lots of video tabs opened and closed -- be it YouTube or Facebook videos. I have no extensions that would target both of those sites, so it might be some common one like TabMixPlus or Greasemonkey.


(In reply to The 8472 from comment #40)
> > This time I have opened dozens of Facebook videos (versus lots of YouTube
> > videos), had PluginContainer blow up and die, and, after a while, it has
> > died second time with only relatively small size -- as seen in the
> > screenshot, all tabs' labels and content is lost after this, and task
> > manager shows that PluginContainer's size at the time of its death was
> > relatively small.
> 
> That crash might be an unrelated issue to the leak. Check about:crashes if
> you have a crash report for it.

Those crashes do not leave anything for crash reporter. And the crashes somehow always happen during session store write in a way that the current tab always dies even after restart, exclamation mark in red circle appears on tab that was active during the crash, and there is no URL or anything left, it is wiped out forever. 


(In reply to Nathan Froyd [:froydnj] from comment #38)
> Extensions run with the same privileges as the UI itself (this is called
> "chrome" or "browser chrome").  That means extensions are free to do a whole
> host of useful things, but they are also free to do a whole host of
> dangerous things, like leaking windows and everything inside of those
> windows, which is what sounds like is going on here.  We can't prevent
> things like that, e10s or not.

Thanks; it is an off-topic regarding this specific bug, excuse me, but do you lads plan to re-design FF core to separate GUI from everything else such as GC and make the GUI separate process (with a higher priority)? 

It would solve quite a few serious UI slow-down issues like the input lag for text fields and context/main menu operations like I described above (in Safe Mode) in big sessions or wicked garbage generators such as TweetDeck page? 

It turns out that just having FireFox.exe and PluginContainer.exe is not enough for UI to be absolute priority. In few cases where I had to use profiler the slowdowns that killed usability were GC synchronous operations. e10s mode did not solve the slowdowns fundamentally due to this; another, third process might be needed. Or there would be some serious drawbacks from totally detaching GUI and making it a separate process, and hence there are no plans for that even for the future?

Comment 42

[Wayne Mery (:wsmwk)]

footprint keyword is about excessive code or memory usage, not limits :)
  https://bugzilla.mozilla.org/describekeywords.cgi#footprint
also, this isn't security related

Comment 43

[Andrew McCreight [:mccr8]]

(In reply to Nathan Froyd [:froydnj] from comment #39)
> If we are able to reproduce the problem with a single addon, would the GC or
> CC log at the point when things are becoming sluggish tell us anything
> useful about what's creating ghost windows, or what's holding on to the
> ghost windows?

The CC log will tell you what object is keeping the window alive, but it is often difficult to figure out why that other object is still alive.

(In reply to User Dderss from comment #41)
> Or there would be some serious drawbacks from
> totally detaching GUI and making it a separate process, and hence there are
> no plans for that even for the future?

As you noted, e10s is basically this: the main process pretty much only deals with the UI. This keeps problems with webpages from interfering with the UI. However, the UI itself is written in JS, so it needs to have a GC. Addons currently run like anything other browser code, so they can cause issues. Once addons are moved to WebExtensions, they could potentially be run in a separate process, and thus it might be possible to isolate them more.

Comment 44

[:shell escalante]

did you find which extensions was causing this?

Comment 45

[Firefox Bug Husbandry Bot]

Moving from Core::Untriaged to Core::General https://bugzilla.mozilla.org/show_bug.cgi?id=1407598

Comment 46

[User Dderss]

Alas, no, I did not find the cause.