Open Bug 1601394 Opened 4 years ago Updated 2 years ago

FF 70.0.1 buggy and can not load various sites

Categories

(Core :: Storage: localStorage & sessionStorage, defect)

70 Branch
x86_64
macOS
defect

Tracking

()

UNCONFIRMED

People

(Reporter: joerg, Unassigned)

Details

Attachments

(4 files)

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:70.0) Gecko/20100101 Firefox/70.0

Steps to reproduce:

As I had to do a fresh installation I now have the latest FF 70.0.1.
However, this browser versions seems to be very buggy:

I also tried to start FF in safe mode with all extensions/plugins disabled but still get the same bugs.

Actual results:

See above (firefox renders or behaves wrong)

Expected results:

I should be able to see the website normal and be able to scroll and interact with it.

Forgot to mention: I am using MacOS Catalina. Might be related if the errors are not cross-plattform.

Screenshot shows FF 70.0.1 on the left and Chrome on the right.
As you can see, the toolbar of JIRA is somehow messed in FF and nothing can be clicked.

And here is FF on the left with https://travis-ci.org and Chrome with same URL on the right.

Vertical Scrolling issue can be reproduced with my FF 70.0.1 in this site: https://bugs.openjdk.java.net/browse/JDK-4935355 (does not happen in any JIRA). However, same URL works with vertical scrolling in Chrome without issues.

Hi Jorg,

I cannot reproduce this issue on either Firefox 71.0 or Firefox Nightly 73.0a1 using MacOS 10.14 / MacOS 10.15
I've tried every link you've mentioned, and they work as intended.

Could you check if this also happens while browsing in safe-mode? Here is a link that can help you do that:
https://support.mozilla.org/en-US/kb/troubleshoot-firefox-issues-using-safe-mode

If it does occur, please create a fresh profile, and check it again, you can find the steps to do that below:
https://support.mozilla.org/en-US/kb/profile-manager-create-and-remove-firefox-profiles?redirectlocale=en-US&redirectslug=Managing-profiles#w_starting-the-profile-manager

Thanks!

Flags: needinfo?(joerg)

I can confirm that something seems messed with my profile.
This means that the priority for this issue is not so bug as I may only affect me or very few users.
You might even close it.
However, I do not easily want to dump my profile as there are tons of valuable settings, cookies, plugin configs and data in it.
For the travis-ci problem I opened the debugger and got this:
NS_ERROR_FILE_CORRUPTED tracer.js:98

To me this looks as if firefox has cached some data locally that it does not reload from the server that is however corrupted in my profile data.

Flags: needinfo?(joerg)

Just as a final feedback. The profile of a (power) user contains lots of work and valuable data. Treating this as junk and saying "oh from time to time you simply delete your profile and start from scratch" is not always the right sign. As I have to invest a lot to get all running again, I better switch to a different browser that is more reliable.

Hi Jorg,

Please go to https://firefox-storage-test.glitch.me/ and check if all the subsystem statuses are fully operational.
If one or more are busted, could you check if setting dom.storage.next_gen in about:config solves your issue?

Thanks!

Flags: needinfo?(joerg)

Hi Peter,
thanks for not giving up on my exotic case.
All firefox storages are fully functional:

Overview:
Storage is working. This is the same version (72) as the last time you loaded this page.
Specific Subsystem Statuses:

LocalStorage
Good: Totally Working. (fullyOperational)
QuotaManager
Good: Totally Working. (fullyOperational)
IndexedDB
Good: Totally Working. (fullyOperational)
Cache API
Good: Totally Working. (fullyOperational)

Debug Info:

{
"v": 1,
"curVersion": 72,
"prevVersion": 72,
"ls": {},
"qm": {
"lastWorkedIn": 72
},
"idb": {
"persistentCreatedIn": 70,
"persistentLastOpenedIn": 72,
"clearDetectedIn": 0
},
"cache": {
"firstCacheCreatedIn": 70,
"unpaddedOpaqueCreatedIn": 0,
"paddedOpaqueCreatedIn": 70
}
}

I also toggled the dom.storage.next_gen property (set to true).
However, still I have the same effects. I assume something is cached deep inside my firefox profile that can not be cleared with purging local-storage, index-db, cookies, or cached content (as I deleted those for the sites producing the errors including involved 3rd party sites). IMHO the web has become a complex beast and even though I am a developer that can deal with debugger and JS this is too much black magic for me to trace down the root of the problem...

Flags: needinfo?(joerg)

Is this still happening in FFxx 72 or the current nightly (74)?

Another thing worth trying would be to do a refresh, which will save your current profile to the desktop, then diffing that with fresh profile and looking for a glitch.

https://support.mozilla.org/en-US/kb/refresh-firefox-reset-add-ons-and-settings

Flags: needinfo?(joerg)

I just updated to FF 73.0 but the problems still persist.

Flags: needinfo?(joerg)

I guess it should go there...

Component: Untriaged → Startup and Profile System
Product: Firefox → Toolkit

Nothing here indicates any problem relating to startup.

Component: Startup and Profile System → Untriaged
Product: Toolkit → Firefox

The Bugbug bot thinks this bug should belong to the 'Core::Panning and Zooming' component, and is moving the bug to that component. Please revert this change in case you think the bot is wrong.

Component: Untriaged → Panning and Zooming
Product: Firefox → Core

As this bug is over a year old and mentions Firefox 70, it would be useful to get a confirmation that the issue still occurs in the latest version (currently Firefox 85).

Component: Panning and Zooming → Untriaged
Product: Core → Firefox

Can you upload your about:support?

Flags: needinfo?(joerg)
Attached file about-support.json

Sure, I replaced my profile ID with abcdef.default and changed 2 senstive settings (last printed file, and a base64 blob for the printing details) due to data-privacy concerns. I hope that is fine.

Flags: needinfo?(joerg)

Thanks. There are several prefs in there that seem odd

"dom.security.https_only_mode": true,
"dom.security.https_only_mode_ever_enabled": true,
"dom.storage.next_gen": true,
"network.automatic-ntlm-auth.trusted-uris": ".company.com",
"network.negotiate-auth.trusted-uris": ".company.com",
"privacy.purge_trackers.last_purge": "1613646724398",
"privacy.donottrackheader.enabled": true,
"privacy.socialtracking.notification.counter": 2,
"privacy.socialtracking.notification.lastShown": "1572541951586",
"privacy.userContext.enabled": true,
"privacy.userContext.extension": "simple-tab-groups@drive4ik",
"privacy.userContext.longPressBehavior": 2,
"privacy.userContext.ui.enabled": true,

You might try setting some of these back to default to see if that affects it.

Did you try running in a fresh profile to see if these problems happened there? I'm not suggesting that as a solution but as a step to debugging the problem, if the problem still happens in a fresh profile then it's not your profile, if it does then we know it is something in your profile and we can figure out what it is.

thanks for your hints.
I did reset all these settings with the result that now SSO and SimpleTabGroups (FF extension) stopped working.
However, the actual error persists.

Did you try running in a fresh profile to see if these problems happened there?

Yes we have already analyzed this if you read through the entire history of the issue.
Indeed with a fresh profile the error goes away.
However, I loved FF due to the many things I configured to make it work nicely.
For all broken pages I already started using other browsers.
If I have to dump all my invest into my FF profile I can also switch to anther browser completely.
The benefits of FF are in the customization of my profile with bookmark-keywords for JIRA projects, SimpleTabGroups for my different project contexts with lots of custom regex logic, etc. (I am a developer).
I fully understand that some of these "tweaks" might be the reason for the error.
However, I already step by step disabled all plugins and extensions, etc. but could not trace down the error.
I rather assume the error does not come from some configuration in my profile but rather from some "dynamic data" like local storage or whatever. What I could purge (caches, etc.) I did already. So I am still stuck.
Debugging what is going on under the hood of FF seems to be rather hard...

I just went to yammer.com and purged all storage things via FF debugger including cookies, etc.
After logging back into yammer, I still get an entirely empty page.
So if that sounds frank, but if the firefox profile contains things that prevent FF from downloading HTML,JS,CSS,... and instead getting something from years ago or from fundamentally breaking browser standard behaviour then something is designed odd.
I did not give up on my profile as I still hope that we might be able to trace down the error to the root and probably find a server bug in FF that might be very rare and unlikely to happen but if it happens it can have really bad and odd effects...
And surely I do not use user-scripts with greasemonkey/tampermonkey or anything like this.
I observed the DOM of the emtpy Yammer page and really only the core structure is there with empty content.
JS Console in Debugger says (maybe there pre-load warnings can shed some light of what is going wrong):

Diese Seite verwendet die nicht standardisierte Eigenschaft "zoom". Stattdessen sollte calc() in den entsprechenden Eigenschaftswerten oder "transform" zusammen mit "transform-origin: 0 0" verwendet werden. company.com
Uncaught
Exception { name: "NS_ERROR_FILE_CORRUPTED", message: "", result: 2152857611, filename: "https://www.yammer.com/company.com/", lineNumber: 25, columnNumber: 0, data: null, stack: "@https://www.yammer.com/company.com/:25:25\n" }
company.com:25
JQMIGRATE: Migrate is installed with logging active, version 3.0.0 yam-vendor-c501b46b6b8fd11d93f2.js:2:638461
Array [ Exception ]
yam-requirejs-home-main-d10402e080645e543165.js:2:223784
Die Ressource unter "https://s0-azure.assets-yammer.com/assets/fonts/yam-yamicon-958e5f62307a3e069f96ad5343649e8c.woff2" wurde vorab geladen ("link-preload"), aber nicht innerhalb der ersten Sekunden verwendet. Es wird eine Überprüfung empfohlen, ob alle preload-Attribute korrekt gesetzt sind. company.com
Uncaught TypeError: i.getCurrentUser() is undefined
repairToken https://s0-azure.assets-yammer.com/assets/releases/stable/yam-dfr-home-main-c2eaf44deef182c922fc.js:2
dispatch https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
handle https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
yam-dfr-home-main-c2eaf44deef182c922fc.js:2:1993251
Uncaught TypeError: i.getCurrentUser() is undefined
repairToken https://s0-azure.assets-yammer.com/assets/releases/stable/yam-dfr-home-main-c2eaf44deef182c922fc.js:2
dispatch https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
handle https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
yam-dfr-home-main-c2eaf44deef182c922fc.js:2:1993251
Uncaught TypeError: i.getCurrentUser() is undefined
repairToken https://s0-azure.assets-yammer.com/assets/releases/stable/yam-dfr-home-main-c2eaf44deef182c922fc.js:2
dispatch https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
handle https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
yam-dfr-home-main-c2eaf44deef182c922fc.js:2:1993251
Uncaught TypeError: i.getCurrentUser() is undefined
repairToken https://s0-azure.assets-yammer.com/assets/releases/stable/yam-dfr-home-main-c2eaf44deef182c922fc.js:2
dispatch https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
handle https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
yam-dfr-home-main-c2eaf44deef182c922fc.js:2:1993251
repairToken https://s0-azure.assets-yammer.com/assets/releases/stable/yam-dfr-home-main-c2eaf44deef182c922fc.js:2
dispatch https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2
handle https://s0-azure.assets-yammer.com/assets/releases/stable/yam-vendor-c501b46b6b8fd11d93f2.js:2

(In reply to Jörg Hohwiller from comment #20)

Yes we have already analyzed this if you read through the entire history of the issue.

I have read the entire history of the bug thanks for the suggestion. If you read the very comment you are replying to I take great pains to indicate that trying in a new profile is not my suggestion for you to abandon your old profile but rather a way to debug the issue so we can figure out what is wrong.

Yes, I fully agree and understand your comments.
Is there maybe a constructive way how to step by step insterpolate between a fresh profile and my current profile to trace down the bug (something analogue to "git bisect"). I am happy to assist but I am quite clueless how to do that. I could use a recursive diff tool to compare the two profiles and step by step merge changes into the new profile unless it breaks. Then those last changes must be related to the problem.
So as a preparation I will create a local git repo and commit my current buggy profile to it. Then I create a new profile and rename it so it has the exact same name and commit this as a separate branch. I will cherry pick changes from master branch and commit them as long as the error is not reproduced.
Does that make sense? Any better idea?

This seems to be exactly the issue I'm having with the Azure portal:

  • Go to portal.azure.com
  • Get shown login screen
  • Fill out details, submit
  • It eventually redirects to a blank page that has the URL https://portal.azure.com/signin/index/
  • Clear cookies, cache and even local/session storage stuff
  • Go back to the original portal.azure.com URL (not by pressing back but by re-typing the URL)
  • Once again clear cookies etc (since you're now on login.microsoftonline.com or something)
  • Try the same steps again, until the blank page
  • The developer console still/again shows:
Uncaught Exception { name: "NS_ERROR_FILE_CORRUPTED", message: "", result: 2152857611, filename: "https://portal.azure.com/signin/index/", lineNumber: 10, columnNumber: 0, data: null, stack: "t@https://portal.azure.com/signin/index/:10:129\nh@https://portal.azure.com/signin/index/:12:479\nf@https://portal.azure.com/signin/index/:13:394\n@https://portal.azure.com/signin/index/:14:14\n" }

Cookie “OpenIdConnect.nonce.%2Bsnipped01%2Fsnipped02%3D” has been rejected because it is already expired.
Cookie “OpenIdConnect.nonce.snipped03%2Fsnipped04%3D” has been rejected because it is already expired.
Cookie “Portal.ReqTenant” has been rejected because it is already expired.
Cookie “OpenIdConnect.nonce.snipped03%2Fsnipped04%3D” has been rejected because it is already expired.

(of course the snipped0x is to censor the token, I dunno what it could be used for)

The odd thing is, if I open the Azure portal in an incognito window then it does work. So I think there's a corrupt cookie or storage object somewhere that doesn't belong to the "regular" domains, but rather one of those intermediary ones that are only used during/for redirection.

My Firefox should be the latest from stable thanks to auto-update and is as bare as can be (zero add-ons etc). It only happens on Windows by the way, as far as I can tell.

I also noticed the suggestion of trying a fresh profile, but A) that's effectively almost the same as an incognito window so the issue probably won't occur B) it'd be very hard to reproduce it on-demand because something seems to get corrupted over time.

Component: Untriaged → Storage: localStorage & sessionStorage
OS: Unspecified → macOS
Product: Firefox → Core
Hardware: Unspecified → x86_64
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: