Random pages go blank

RESOLVED FIXED in Firefox 55

Status

()

--
critical
RESOLVED FIXED
a year ago
a year ago

People

(Reporter: caspy77, Unassigned)

Tracking

({dataloss, regression})

55 Branch
Firefox 56
All
Windows
dataloss, regression
Points:
---

Firefox Tracking Flags

(firefox-esr45 unaffected, firefox-esr52 unaffected, firefox54 unaffected, firefox55+ fixed, firefox56+ fixed)

Details

Attachments

(1 attachment)

(Reporter)

Description

a year ago
(On Nightly)

I have random tabs to go blank.
It's happened to pages that loaded after restoring a session, unloaded tabs that load, and I think just regular pages during browsing.

The total content of the tab is white. Reloading, Shift + Reload or attempting to navigate to a new page (or back in history) does nothing. The tab is useless. Other tabs are fine.

I've seen this for a several days now and wondered if it was just a broken addon, but today I saw two reports from reddit users describing the same phenomenon.
https://www.reddit.com/r/firefox/comments/6dbfzk/firefox_nightly_single_tab_breaks/
(the poster and one of the commenters)

The poster listed their addons and we share none in common.
Any errors in the browser console when this happens, or when you try to load content into these dead tabs? Anything else that might help narrow this down, or explicit steps to reproduce?
Component: General → Untriaged
Flags: needinfo?(caspy77)
Something like that is happened also to me, sometimes happen that all the tabs are halt with content and the rest is blank and there is interaction. Like if it is lagging, after a while restart working everything.
The next time I will check in the browser console :-)
(Reporter)

Comment 3

a year ago
(In reply to :Gijs from comment #1)
> Any errors in the browser console when this happens, or when you try to load
> content into these dead tabs? Anything else that might help narrow this
> down, or explicit steps to reproduce?

Using such tools is not a common occurrence, but I will try and check this next time this happens and see what's up.

Also, no, I know of no steps to reproduce.
It's quite possible that this is a dupe of bug 1367964, which should be fixed in today's Nightly.
(Reporter)

Comment 5

a year ago
(In reply to Mike Conley (:mconley) from comment #4)
> It's quite possible that this is a dupe of bug 1367964, which should be
> fixed in today's Nightly.

I just experienced this again and, per that bug title, detaching the tab from the window fixed the issue.
Seems likely we can dup it soon.
(Reporter)

Comment 6

a year ago
(In reply to :Gijs from comment #1)
> Any errors in the browser console when this happens, or when you try to load
> content into these dead tabs? Anything else that might help narrow this
> down, or explicit steps to reproduce?

Probably moot given my last comment, but I experienced the issue, reloaded the page and copied all the entries that showed up in the Browser Console.
here they are:

====
XML Parsing Error: no root element found
Location: https://e.reddit.com/v1?key=RedditFrontend1&mac=b539a8080e39f09df69f799618610a693abfeefb64a424e5042196449219a724
Line Number 1, Column 1:  v1:1:1
[pulse.webext.lib.page-action] Destroying page action for tab 291.  background.js:93:8
[pulse.webext.lib.page-action] Showing for tab 291  background.js:93:8
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/require/topFrame.js
Line Number 1, Column 1:  topFrame.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/Clipper.js
Line Number 1, Column 1:  Clipper.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/ContentPreview.js
Line Number 1, Column 1:  ContentPreview.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/Coordinator.js
Line Number 1, Column 1:  Coordinator.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/js/GlobalUtils.js
Line Number 1, Column 1:  GlobalUtils.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/Promotion.js
Line Number 1, Column 1:  Promotion.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/tooltips/CustomTooltipEligibility.js
Line Number 1, Column 1:  CustomTooltipEligibility.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/require/checkSimSearch.js
Line Number 1, Column 1:  checkSimSearch.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/require/pageVisible.js
Line Number 1, Column 1:  pageVisible.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/PageInfo.js
Line Number 1, Column 1:  PageInfo.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/content/require/isTest.js
Line Number 1, Column 1:  isTest.js:1:1
XML Parsing Error: not well-formed
Location: jar:file:///C:/Users/<fake>/AppData/Roaming/Mozilla/Firefox/Profiles/6kkhzmzd.default/extensions/%7BE0B8C461-F8FB-49b4-8373-FE32E9252800%7D.xpi!/third_party/require/domReady.js
Line Number 1, Column 1:  domReady.js:1:1
moz-extension://384e4268-a110-4f6e-9a8a-3b186720820a/background.html: chrome tabs query did not return a result while changing window focus

  Log.js:2:3854
Use of Mutation Events is deprecated. Use MutationObserver instead.  ContentVeil.js:2:4726
[pulse.webext.lib.page-action] Showing for tab 291  background.js:93:8
Content Security Policy: The page’s settings blocked the loading of a resource at data:application/javascript;base64,KGZ1b... (“script-src https://connect.facebook.net https://cm.g.doubleclick.net https://ssl.google-analytics.com https://graph.facebook.com https://twitter.com 'unsafe-eval' https://*.twimg.com https://api.twitter.com https://analytics.twitter.com https://publish.twitter.com https://ton.twitter.com https://syndication.twitter.com https://www.google.com https://t.tellapart.com 'nonce-3Fdc5fXLZBl8WFWsFgKmsw==' https://platform.twitter.com https://www.google-analytics.com https://twitter.com”).  (unknown)
Loading failed for the <script> with source “https://www.google-analytics.com/analytics.js”.  868319364855980032:1
<unavailable>  core.js:100
Error: call to Function() blocked by CSP (unknown)
[pulse.webext.lib.page-action] Showing for tab 291
Flags: needinfo?(caspy77)
Was this in today's Nightly?
Flags: needinfo?(caspy77)
(Reporter)

Comment 8

a year ago
(In reply to Mike Conley (:mconley) from comment #7)
> Was this in today's Nightly?

Yes.
Flags: needinfo?(caspy77)
In that case, this is likely not a dupe of bug 1367964 - that fix is in today's Nightly.

The browser console data is, unfortunately, not very helpful here. What would be lovely are some reliable STR.

Are there any steps to reproduce this issue?
Flags: needinfo?(caspy77)
Alternatively, one thing that _might_ draw out some more useful information in the console is to add a new bool pref "browser.tabs.remote.logSwitchTiming", and set it to true. After a restart, this will log tab switch data to the browser console.

Once you experience this bug again, that log might be very illuminating.
(Reporter)

Comment 11

a year ago
> The browser console data is, unfortunately, not very helpful here. What
> would be lovely are some reliable STR.
> 
> Are there any steps to reproduce this issue?

Not that I know of. When I just tested I could get it on multiple unloaded twitter pages, after restarting again, unloaded twitter pages seem to load fine.

I'll keep an eye out and play around for STR.

(In reply to Mike Conley (:mconley) from comment #9)(In reply to Mike Conley (:mconley) from comment #10)
> Alternatively, one thing that _might_ draw out some more useful information
> in the console is to add a new bool pref
> "browser.tabs.remote.logSwitchTiming", and set it to true. After a restart,
> this will log tab switch data to the browser console.
> 
> Once you experience this bug again, that log might be very illuminating.

Will do.
Flags: needinfo?(caspy77)
The other bit of confusion here is that this bug _might_ be related to bug 1362866. It might also be related to bug 1364563, which is a completely different component.

The next time someone is in this state, can you please open up the Browser Toolbox[1], switch to the Console, and (with the blank tab selected), type in:

gBrowser.selectedBrowser

and paste the resulting XUL node description to this bug?

If the node has the "blank" attribute set to "true", then this is indeed related to bug 1362866. If, however, it doesn't, then the next thing to check is about:support, and to look in the Graphics section for things like:

"CP+[GFX1-]: LayerManager::EndTransaction skip RenderLayer()"

If that's there, chances are, this is more related to bug 1364563.

[1]: https://developer.mozilla.org/en-US/docs/Tools/Browser_Toolbox
Comment hidden (obsolete)
In fact, the information in comment 12 will be more useful than comment 10, if at least to confirm whether or not bug 1362866 is involved. If it's not, the data I requested from comment 10 will not be useful.
(Reporter)

Comment 15

a year ago
(In reply to Mike Conley (:mconley) from comment #12)
> The next time someone is in this state, can you please open up the Browser
> Toolbox[1], switch to the Console, and (with the blank tab selected), type
> in:
> 
> gBrowser.selectedBrowser
> 
> and paste the resulting XUL node description to this bug?

For other's benefit...
Had a conversation in IRC and it seems using the Console built into the Browser Toolbox does not give the right results, while using the Browser Console window does.
Comment hidden (off-topic)
(In reply to ajfhajf from comment #16)
> I've been using Nightly non-stop all week so I opened/restored and loaded
> more than 3000 tabs and never came across this bug. I have uBlock Origin,
> NoScript, Tab Mix Plus and Session Manager installed. However, I often see
> white flashes when closing or switching tabs.

Thanks. The flashes are an understood problem, and will be likely addressed in bug 1367596. This bug is specifically about tabs that go blank and stay blank.

Comment 18

a year ago
Prior to the landing of bug 1367964 I had this issue crop up at least once every 15-30 minutes, but since then I haven't seen it appear again. 

I have several addons installed, but as far as I know, the only ones that can possible change the content of the page would be: Decentraleyes, uMatrix, uBlock, Reddit Enhancement Suite. 

I saw it happen mostly on Patreon or reddit.
(In reply to ajfhajf from comment #16)
> [...] However, I often see white flashes when closing [...]
FYI it's bug #1237654.
(In reply to Virtual_ManPL [:Virtual] - (please needinfo? me - so I will see your comment/reply/question/etc.) from comment #19)
> (In reply to ajfhajf from comment #16)
> > [...] However, I often see white flashes when closing [...]
> FYI it's bug #1237654.

To be clear, the flashes that were introduced by bug 1362866 will be reduced in bug 1367596.

Comment 21

a year ago
(In reply to Mike Conley (:mconley) from comment #12)
> The other bit of confusion here is that this bug _might_ be related to bug
> 1362866. It might also be related to bug 1364563, which is a completely
> different component.
> 
> The next time someone is in this state, can you please open up the Browser
> Toolbox[1], switch to the Console, and (with the blank tab selected), type
> in:
> 
> gBrowser.selectedBrowser
> 
> and paste the resulting XUL node description to this bug?
> 
> If the node has the "blank" attribute set to "true", then this is indeed
> related to bug 1362866. If, however, it doesn't, then the next thing to
> check is about:support, and to look in the Graphics section for things like:
> 
> "CP+[GFX1-]: LayerManager::EndTransaction skip RenderLayer()"
> 
> If that's there, chances are, this is more related to bug 1364563.
> 
> [1]: https://developer.mozilla.org/en-US/docs/Tools/Browser_Toolbox

It Just happened to me and here is the output of the "gBrowser.selectedBrowser" for the blank tab:

<browser type="content" message="true" messagemanagergroup="browsers" contextmenu="contentAreaContextMenu" tooltip="aHTMLTooltip" remoteType="web" remote="true" autocompletepopup="PopupAutoComplete" selectmenulist="ContentSelectDropdown" datetimepicker="DateTimePickerPanel" autoscrollpopup="autoscroller" clickthrough="never" primary="true">
(In reply to António Pereira from comment #21)
> <browser type="content" message="true" messagemanagergroup="browsers"
> contextmenu="contentAreaContextMenu" tooltip="aHTMLTooltip" remoteType="web"
> remote="true" autocompletepopup="PopupAutoComplete"
> selectmenulist="ContentSelectDropdown" datetimepicker="DateTimePickerPanel"
> autoscrollpopup="autoscroller" clickthrough="never" primary="true">

Okay, this tells me that bug 1362866 is not to blame for what you're seeing. If you're still in this state, can you check the Graphics section of about:support to see if there are any errors in the Graphics log?
Flags: needinfo?(toni_to_1993)

Comment 23

a year ago
(In reply to Mike Conley (:mconley) from comment #22)
> (In reply to António Pereira from comment #21)
> > <browser type="content" message="true" messagemanagergroup="browsers"
> > contextmenu="contentAreaContextMenu" tooltip="aHTMLTooltip" remoteType="web"
> > remote="true" autocompletepopup="PopupAutoComplete"
> > selectmenulist="ContentSelectDropdown" datetimepicker="DateTimePickerPanel"
> > autoscrollpopup="autoscroller" clickthrough="never" primary="true">
> 
> Okay, this tells me that bug 1362866 is not to blame for what you're seeing.
> If you're still in this state, can you check the Graphics section of
> about:support to see if there are any errors in the Graphics log?

I found that this blank tab was caused by me moving a "normal" tab to a private window which cause the tab to became blank and had to refresh it to show the content. And I can reproduce it consistently. So I'm not sure if this is the same bug as reported here.

I had "blank tabs" before when I only had one normal window open. Better wait fo it to happen on its own to find out.
Flags: needinfo?(toni_to_1993)
(In reply to António Pereira from comment #23)
> 
> I had "blank tabs" before when I only had one normal window open. Better
> wait fo it to happen on its own to find out.

When did that last happen?
Flags: needinfo?(toni_to_1993)

Comment 25

a year ago
Oh, you might not believe it, but it just happened!
On a "normal" tab in a normal window, here is the "gBrowser.selectedBrowser" output for the said tab:

<browser type="content" message="true" messagemanagergroup="browsers" contextmenu="contentAreaContextMenu" tooltip="aHTMLTooltip" remoteType="web" remote="true" autocompletepopup="PopupAutoComplete" selectmenulist="ContentSelectDropdown" datetimepicker="DateTimePickerPanel" autoscrollpopup="autoscroller" clickthrough="never" primary="true">

And I can't find any sort of error in about:support.
Flags: needinfo?(toni_to_1993)
(In reply to António Pereira from comment #25)

Would you mind going to about:support, clicking "Copy Text to Clipboard" and pasting the contents into this bug as a new attachment?

Also, do you have an IRC client? If you have the time, I'd be interested in working with you directly to investigate this.
Flags: needinfo?(toni_to_1993)
(Reporter)

Comment 27

a year ago
Created attachment 8872557 [details]
about:support text of an affected profile

I experienced it again now. It was an older unloaded Twitter tab.

The readout from Browser Console looks identical to António's in comment 25

I am attaching my about:support

Updated

a year ago
Flags: needinfo?(toni_to_1993)
[Tracking Requested - why for this release]: Dataloss regression

I can still reproduce it on Firefox 56.
It's mostly reproducible on unloaded/purged tabs from memory, while duplicating them or moving them to new window without opening, kinda similar in frontend to bug #610357.
Severity: normal → critical
Has Regression Range: --- → yes
Has STR: --- → yes
status-firefox54: --- → unaffected
status-firefox55: --- → affected
status-firefox56: --- → affected
status-firefox-esr45: --- → unaffected
status-firefox-esr52: --- → unaffected
tracking-firefox55: --- → ?
tracking-firefox56: --- → ?
Component: Untriaged → Tabbed Browser
Keywords: dataloss, regression
OS: Unspecified → Windows
Hardware: Unspecified → All
See Also: bug 1367596
Summary: Random pages go blank (Nightly) → Random pages go blank

Comment 29

a year ago
I get the exact same symptoms but a 100% percent of the time if I mass load unladed tabs via Tab Mix Plus' load left/right/other/all tabs features, or via All Tabs Helper's "load selected tabs" feature.

The only way to restore those tabs is to close them and then undo the close.
Tracked for 55+ as it's a bad regression. NI MConley, Avaida about STR in comment 29, in case that helps us repro this in-house.
tracking-firefox55: ? → +
tracking-firefox56: ? → +
Flags: needinfo?(mconley)
Flags: needinfo?(andrei.vaida)

Updated

a year ago
Duplicate of this bug: 1374696
Mike, want to take this on?  Caspy7 hangs out in #sumo if you want to try and find him.
(In reply to Ritu Kothari (:ritu) from comment #30)
> Tracked for 55+ as it's a bad regression. NI MConley, Avaida about STR in
> comment 29, in case that helps us repro this in-house.

Unfortunately, I'm not able to reproduce using these steps, despite installing the add-on. :(

I suspect this is an issue in our Graphics layer, to be honest. It might be better if somebody from Graphics attempted to address this.
Flags: needinfo?(mconley)
Thanks Mike. Milan, might you be able to help? :)
Flags: needinfo?(milan)
If this was graphics, in particularly related to bug 1364563 I would have expected to see messages in about:support, and we're not, so there goes that.

I do have one question, for those that can reproduce - does resizing the window help?

Sotaro, you've recently looked at moving tabs between windows in unrelated scenarios, and some of the STR at least involve that.  Are there any other questions we can ask the users that can reproduce to help us narrow the problem down?
Flags: needinfo?(milan) → needinfo?(sotaro.ikeda.g)
(In reply to Milan Sreckovic [:milan] from comment #35)
> Sotaro, you've recently looked at moving tabs between windows in unrelated
> scenarios, and some of the STR at least involve that.  Are there any other
> questions we can ask the users that can reproduce to help us narrow the
> problem down?

It might better to check what happens with resizing, reloading and opening a new tab. If bug 1364563 happens, resizing seems not help. If resizing works to fix the problem, it could be another problem.

One other possibility is that nsRefreshDriver stops to trigger new layer transaction. One possibility of it is that DidComposite() is not delivered to ClientLayerManager. Then ClientLayerManager calls nsRefreshDriver::NotifyTransactionCompleted(). nsRefreshDriver permits at most 2 pending transactions. If nsRefreshDriver did not receive pending transaction complete, content rendering is not update automatically. In this case, resizing or re-load seems to work. And it might happen during tab move between different windows. Then I created Bug 1375349.
Flags: needinfo?(sotaro.ikeda.g)
I've tried to reproduce this issue on Firefox 55.0a1 (2017-05-25) and on Firefox 56.0a1 (2017-06-26), under Windows 10x64, without success.

Virtual_ManPL or avada can you still reproduce this issue on Latest Nightly. If yes, can you please provide the exact STR?

Note that the tests were performed with and without Tab Mix Plus addon.
Flags: needinfo?(dqeswn)
Flags: needinfo?(andrei.vaida)
Flags: needinfo?(Virtual)
(Reporter)

Comment 38

a year ago
I'd been meaning to comment and say that I haven't seen this issue again, however I did end up having an OS reinstall and wasn't sure if that mucked up the factors - notably if it had something to do with graphics drivers somehow.
Comment hidden (off-topic)

Comment 40

a year ago
(In reply to Mihai Boldan, QA [:mboldan] from comment #37)
> Virtual_ManPL or avada can you still reproduce this issue on Latest Nightly.
> If yes, can you please provide the exact STR?
> 
> Note that the tests were performed with and without Tab Mix Plus addon.

It looks like it stopped happening to me on v55 beta.
Flags: needinfo?(dqeswn)
I'm on Mozilla Firefox Nightly 56.0a1 and I also didn't see this issue for some time, so I think it's safe to say that it was fixed by some other patch.
Status: NEW → RESOLVED
Last Resolved: a year ago
status-firefox55: affected → fixed
status-firefox56: affected → fixed
Flags: needinfo?(Virtual)
Resolution: --- → FIXED
Target Milestone: --- → Firefox 56
You need to log in before you can comment on or make changes to this bug.