Open Bug 1705541 Opened 4 years ago Updated 4 years ago

Youtube janks on layout after navigating back

Categories

(Core :: Layout, defect)

Product:
Core
Component:
Layout
Type:
defect

Tracking

()

People

(Reporter: bryce, Unassigned)

References

Details

See https://bugzilla.mozilla.org/show_bug.cgi?id=1701808#c26 which contains a profile and a YT video of the reporter triggering the issue. From 75s in the profile there is a large jank that looks related to layout.

That bug was triaged into media because we similar janks on some media calls in profile runs earlier in the bug. However, it looks like there's a number of JS doing too many calls kind of janks which has me wondering if the actual cause is something else trigger hot JS loops.

Have we seen anything like this before?

17s restyle looks absolutely insane. Does it reproduce on a clean profile? I see that the reporter has a number of extensions.

Looks like this was discussed a bit more on bug 1701808.

In particular:

  • bug 1701808 comment 31: the reporter said this didn't reproduce in a fresh profile. To reproduce, they had to "add mini search and add youtube as search engine and then booom catch this error". (Note, I'm guessing by "mini search" they mean the Firefox toolbar's dedicated search-bar... At least, I don't see any add-on called "mini search" or anything like that in the list of extensions from their profile in https://share.firefox.dev/3mQamCu , so I don't think it's an add-on.)

  • bug 1701808 comment 32: emilio seems to have thought "mini search" was an add-on, and asked the reporter for links to the relevant add-ons. The reporter replied with a screenshot of their add-ons page; there's no "mini search" add-on, so that suggests they were indeed probably referring to the dedicated search bar.

  • bug 1701808 comment 34: the reporter mentions they're getting a special/uncommon YouTube UI, specifically in Firefox, and suspects they're getting some A/B beta test from YouTube. But it sounds like they were seeing this issue before receiving that YouTube UI, so this A/B testing is probably unrelated.

  • bug 1701808 comment 37: the reporter noted that ESR78 seems to be unaffected.

  • bug 1701808 comment 42: the reporter expressed a request not to be contacted further, and says they're not using Firefox anymore, though "89.a1 is now less visible slowdown" which sounds like things have improved somewhat.

I tried to reproduce this myself, using YouTube in the search bar and trying to follow the user's steps from their video in bug 1701808 comment 26 ( https://youtu.be/fsNqyZJ2baY ) but I wasn't able to reproduce any sort of hang on navigation (other than some minor paint delays while videos are being populated; but those don't seem to be restyle-related).

The user's hang seems to be just after this point in the video (this links to 1:20, and the hang is at 1:23 when the user clicks 'back'):
https://www.youtube.com/watch?v=fsNqyZJ2baY#t=1m20s

I'm triaging this as S3 for the moment, to take this out of the triage queue. It may become higher-priority if we find a way to reproduce or get more information about other factors that could be involved. (I'm not sure we can get any more information from the user who reported this, given their request to unsubscribe / lack of further interest expressed in bug 1701808 comment 42.)

Severity: -- → S3
You need to log in before you can comment on or make changes to this bug.