1438409 - Firefox won't load page properly when in background

Status: UNCONFIRMED

(Core :: DOM: Core & HTML, defect, P3)

Description

[shapiro125]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.13; rv:58.0) Gecko/20100101 Firefox/58.0
Build ID: 20180206200532

Steps to reproduce:

Open a tab in the background, specifically through an RSS reader like Feedly. Most notably, this happens on all Kinja sites (Gawker, Gizmodo, Lifehacker) and intermittently with Reddit.

Great conversation from the Ublock Origin Github here: https://github.com/gorhill/uBlock/issues/3447


Actual results:

Certain CSS or javascript elements will not load properly. A refresh is required to load the page properly.




Expected results:

The page should have loaded properly in the first place. 

As reference, Chrome 64 handles this behavior fine.

Comment 1

[shapiro125]

Attachment: Page loaded in the background

Comment 2

[shapiro125]

Attachment: Page loaded normally

Comment 3

[u554753]

Please correct the component if this isn't the correct one. Thanks!

Comment 4

[Ben Kelly [:bkelly, not reviewing]]

My guess is this is either DOM (timers/events), networking (http cache), or tech evangelism.  Its probably not layout, though.

Moving to DOM for now.

Comment 5

[Ben Kelly [:bkelly, not reviewing]]

Honza, does any of the network prioritization stuff depend on foreground vs background?

Comment 6

[Honza Bambas (:mayhemer)]

(In reply to Ben Kelly [:bkelly] from comment #5)
> Honza, does any of the network prioritization stuff depend on foreground vs
> background?

Yes, we lower h2 streams prio when coming from bck tabs, we lower per-origin concurrency for background tabs to 1 during activity in the active tab (but when the active tab finishes, we allow full parallelism again - unless there is some bug in it, unlikely).  

we had bugs when parallelism was limit in the past when a response was dependent on a request to a different URI (css generation on xhr, for instance.)

I didn't look at this bug, just replying, if there is a STR I can grab a log.  Or if you can, provide one. (https://developer.mozilla.org/en-US/docs/Mozilla/Debugging/HTTP_logging)

Comment 7

[Honza Bambas (:mayhemer)]

shapiro125, is this related to uBlock?  if so, have you tried w/o it?

also, it would be great if you could collect a log for the issue: https://developer.mozilla.org/en-US/docs/Mozilla/Debugging/HTTP_logging

thanks.

could also be related to bug 1438505.

Comment 8

[Honza Bambas (:mayhemer)]

According what is told in the github issue, seems like this can be reproduced (with some effort) w/o any extension, but I'd like this to be 100% confirmed first.

Looks like this css fails to load from some reason, loading via h2:
https://x.kinja-static.com/assets/stylesheets/blog-9b9642c53a824b088cc15adf0ede74ac.css

Comment 9

[shapiro125]

:mayhemer - I originally thought it was a uBlock issue and posted the bug to there, but it occurred regardless. I also tried it with a new profile to see if there was an issue with my setup and reproduced it both times.

The bug occurs every time with Kinja sites, but I have also experienced the issue on sites like Reddit, though it's difficult to reproduce.

For a non-developer like myself, using uBlock's refresh tool in the logger was a great way to test reproducing it.

Log uploaded -- for reference, it is loading the site in the background with errors and then refreshing to load properly.

Comment 10

[shapiro125]

(In reply to shapiro125 from comment #9)
> :mayhemer - I originally thought it was a uBlock issue and posted the bug to
> there, but it occurred regardless. I also tried it with a new profile to see
> if there was an issue with my setup and reproduced it both times.
> 
> The bug occurs every time with Kinja sites, but I have also experienced the
> issue on sites like Reddit, though it's difficult to reproduce.
> 
> For a non-developer like myself, using uBlock's refresh tool in the logger
> was a great way to test reproducing it.
> 
> Log uploaded -- for reference, it is loading the site in the background with
> errors and then refreshing to load properly.

The log is 24MB - do you have any advice for uploading it?

Comment 11

[shapiro125]

Attachment: Log of bug

Comment 12

[Honza Bambas (:mayhemer)]

(In reply to shapiro125 from comment #10)
> The log is 24MB - do you have any advice for uploading it?

thanks!  I wanted to say zip it, but you figured that out.  note that xzip (.xz) has the best ratio.  I'll take a look.

Comment 13

[Honza Bambas (:mayhemer)]

I don't see any networking related problem in the log with that exact css file (comment 8).  Can you please tell me what exactly have you done when capturing the log, what page (URL) has failed for you and how exactly?

thanks.

Comment 14

[shapiro125]

Sure -- I went through the process detailed in this comment: https://github.com/gorhill/uBlock/issues/3447#issuecomment-358702797

I used this URL: https://gizmodo.com/first-evidence-that-microplastics-travel-up-the-food-ch-1823262468

The exact failure was that all page elements would not load properly in the background (e.g. style elements never loaded). Only when refreshed in the foreground did the page load correctly.

Comment 15

[Honza Bambas (:mayhemer)]

It's very hard to find out what's wrong from the log when I don't know which URL failed to load.

Note that it's also hard to compare the two page loads, because the first one (failing) is apparently a bookmark or address go-to (click or typing/pasting to address bar and pressing enter) while the second (succeeded) is a reload via F5.  We do a different HTTP cache use decisions.

I don't say it's a caching issue (may be, tho).  I just say it's hard to find out something from the log w/o more time spent on it.

Comment 16

[Honza Bambas (:mayhemer)]

OK, looking more closely (took some effort), there are two css files we don't event start a load for (necko is not instructed to make the load):


https://x.kinja-static.com/assets/stylesheets/blog-9b9642c53a824b088cc15adf0ede74ac.css
https://x.kinja-static.com/assets/stylesheets/insets-4ee1eb4bdf042aa24d594df458c5b075.css


W/o a child log it's impossible to find out why these two are not requested.  But I think this is a layout/styling content caching or similar bug.

Comment 17

[Henri Sivonen (:hsivonen)]

smaug, can this be fallout from bug 734015?

Comment 18

[Olli Pettay [:smaug][[e-mail]]]

that is quite old one, but sure, possible, though I'd guess some more recent change to background tab hanlding.

Comment 19

[Olli Pettay [:smaug][[e-mail]]]

(there has been many changes to timeouts and refreshdriver and networking etc.)