1738585 - Main process is "live hanging" when I opening the site

Status: RESOLVED WONTFIX

(Core :: Networking, defect, P3)

Description

[Elle Solomina (she/her)]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:93.0) Gecko/20100101 Firefox/93.0

Steps to reproduce:

I open the site https://techporn.club/@punkiepie

Actual results:

The browser didn't response some long time

Expected results:

I want the hang to be localized in the site window. I report additional bug here https://webcompat.com/issues/91910 but it's a problem of browser core.

Comment 1

[Negritas Sergiu, Desktop QA]

Hello! I have managed to reproduce the issue with firefox 93.0 the browser is freezing for a few moments it also happens with firefox 95.0a1 (2021-11-01) but with the 95 version the hanging of the browser happens after the page is loaded, with firefox 94.0 is the same behaviour as for firefox 95. I only managed to reproduce the issue with Ubuntu 20.04. On macOS 10.15 and Windows 10 I wasn't able to reproduce it.

Marking this issue as NEW and assigning to a component in order for our developers to look more into it. If it's not the right component please feel free to change it to an appropriate one.

Have a nice day!

Comment 2

[Andrew McCreight [:mccr8]]

There's a profile in the linked webcompat discussion. It isn't obviously DOM related.

Comment 3

[Markus Stange [:mstange]]

That profile being https://share.firefox.dev/3mz0Khe .

In the network chart you can see that the page loads thousands upon thousands of SVG files for emojis. (6692 network request, to be precise.)

We seem to have an unnecessarily high overhead per network request, and with numbers in the thousands, that overhead seems to show.

Comment 4

[Markus Stange [:mstange]]

The unresponsiveness of the parent process is probably caused by the fact that its event queue is flooded with events.
Example: https://share.firefox.dev/3GQJlIM

You can see a huge number of "PNecko::Msg_PHttpChannelConstructor" runnables, for example.

Comment 5

[Sean Feng [:sefeng211]]

It looks like Chrome doesn't trigger these many SVG requests. I'll try to do some investigation here.

Comment 6

[Bas Schouten (:bas.schouten)]

I do see the large number of SVG network requests in Chrome.

This is very bad site design, but what makes this bug particularly bad is that it actually impacts the parent process responsiveness rather than just the child.

This is evidence that Chrome is much better able to keep their main thread responsive during large numbers of network requests. Something to keep in mind during the future OMT networking work.

Comment 7

[Sean Feng [:sefeng211]]

I tried to see if we can still reproduce the problem but the site was down. So I think it's probably impossible for us to move this bug forward. I am going to close this as a WONTFIX, please feel free to reopen it.