1616593 - [Ubuntu] Scrolling is broken on “Onscroll latency” test page

Status: NEW

(Core :: Panning and Zooming, defect, P3)

Description

[Maria Berlinger [:mberlinger]]

Attachment: scrolling.gif

Affected versions

• 75.0a1 (2020-02-19)
• 74.0b5
• 73.0.1

Affected platforms

• Ubuntu 18.04x64

Steps to reproduce

1. Launch Firefox.
2. Go to the website https://bug1209970.bmoattachments.org/attachment.cgi?id=8667908 and scroll fast up and down using the mouse wheel.

Expected result

• Scrolling works properly.

Actual result

• Scrolling on the website causes a delay of the red circle compared to the target.

Regression range

• This issue is not a recent regression since I was able to reproduce it on Firefox 73.0.1. I’ll come with a regression range asap.

Additional notes

• I’ll attach a screen recording with the issue.

Comment 1

[Daniel Holbert [:dholbert]]

Thanks. I was able to reproduce, in a fresh profile, with current Nightly on Ubuntu 19.10 (on my work desktop machine).

The issue goes away for me if I set the about:config pref layers.async-pan-zoom.enabled to false and restart Firefox. Could you see if that fixes it for you, too?

Comment 2

[Kartikaya Gupta (email:[e-mail])]

We'll probably need a Gecko performance profile of this to investigate. I'm not really seeing it locally on Nightly on Linux.

Comment 3

[Botond Ballo [:botond]]

(In reply to Daniel Holbert [:dholbert] from comment #1)

The issue goes away for me if I set the about:config pref layers.async-pan-zoom.enabled to false and restart Firefox. Could you see if that fixes it for you, too?

(Based on what that page is designed to test, I have no doubt that disabling APZ resolves the issue.)

Comment 4

[Release mgmt bot [:sylvestre / :calixte / :marco for bugbug]]

Bugbug thinks this bug is a regression, but please revert this change in case of error.

Comment 5

[Daniel Holbert [:dholbert]]

Adding dependency on bug 1209970, which is the bug that the testcase is from. (This seems kind of like a new incarnation of that bug.)

Comment 6

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #5)

Adding dependency on bug 1209970, which is the bug that the testcase is from. (This seems kind of like a new incarnation of that bug.)

...though actually, I tested the following nightlies using mozregression and I'm seeing this bug's symptoms in all of them:

• Nightly 2015-12-22, -23, and -25 (which are 1, 2, and 4 days after bug 1209970's patch hit mozilla-central)
• Nightly 2015-12-16 (which is 1 day before bug 1209970's patch first landed, i.e. hit an inbound branch for the first time)

So it doesn't seem like bug 1209970's patch appreciably influenced the symptoms described in this bug here, after all. So I don't think this is a case where bug 1209970 was fixed and then later regressed.

Comment 7

[Maria Berlinger [:mberlinger]]

**Regression range **

Last good revision: 891ee0d0ba3ec42b6484cf0205b3c95e21c58f74 (2015-09-30)
First bad revision: 096c0f407f8ba3ef7cfe4e0b831761993cac38b1 (2015-10-01)
Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=891ee0d0ba3ec42b6484cf0205b3c95e21c58f74&tochange=096c0f407f8ba3ef7cfe4e0b831761993cac38b1

I'm not sure which one of the listed bugs could have caused the regression.

Comment 8

[Release mgmt bot [:sylvestre / :calixte / :marco for bugbug]]

:maria_berlinger, since this bug is a regression, could you fill (if possible) the regressed_by field?
For more information, please visit auto_nag documentation.

Comment 9

[Botond Ballo [:botond]]

(In reply to Maria Berlinger [:maria_berlinger], Release Desktop QA from comment #7)

**Regression range **

Last good revision: 891ee0d0ba3ec42b6484cf0205b3c95e21c58f74 (2015-09-30)
First bad revision: 096c0f407f8ba3ef7cfe4e0b831761993cac38b1 (2015-10-01)
Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=891ee0d0ba3ec42b6484cf0205b3c95e21c58f74&tochange=096c0f407f8ba3ef7cfe4e0b831761993cac38b1

I'm not sure which one of the listed bugs could have caused the regression.

"Bug 1143856 - Enable APZ on Linux desktop nightly builds" seems likely in that range.

Comment 10

[Petruta Horea [:phorea]]

Reproduced this issue with latest Nightly 76.0a1 under Win 7 32-bit.

Comment 11

[Kris Taeleman (:ktaeleman)]

@ktaeleman: try repro on other platforms

Comment 12

[Kris Taeleman (:ktaeleman)]

Attachment: scroll-delay.gif

I don't see this nearly as obvious, but with and without webrender I can spot small alignment issues when scrolling.
I've slowed down the GIF considerably to show the effect. This is on a high performance machine running Windows 10.