Flash of white when scrolling to start/end (with Home/End) on https://www.redwoodcity.org/Home/Components/Calendar/Event/88868/4760
Categories
(Core :: Panning and Zooming, defect, P3)
Tracking
()
People
(Reporter: dholbert, Unassigned)
References
(Regression, )
Details
(Keywords: regression)
Attachments
(2 files)
STR:
- Turn off smooth scrolling in Firefox preferences ("about:preferences", "General" sidebar-topic, all the way at the bottom under "Browsing")
- Visit https://www.redwoodcity.org/Home/Components/Calendar/Event/88868/4760
- Press
Endkey on your keyboard to scroll to the end. - Press
Homekey to scroll back to the top. - Repeat steps 3-4, watching carefully.
ACTUAL RESULTS:
Each time I scroll to the opposite extreme, the page mostly flashes solid-white for a frame.
EXPECTED RESULTS:
No such flash.
Chrome gives EXPECTED RESULTS (you can turn off smooth scrolling there at chrome://flags/#smooth-scrolling)
I'm using a fresh user profile in Nightly 143.0a1 (2025-08-01) on Ubuntu 25.04.
Here's a performance profile, using "Graphics" preset with screenshots added:
https://share.firefox.dev/4msq6tk
Here's the profile zoomed in to show one of the white flashes in the screenshots track:
https://share.firefox.dev/45l7lkB
| Reporter | ||
Comment 1•3 months ago
|
||
| Reporter | ||
Comment 2•3 months ago
|
||
| Reporter | ||
Updated•3 months ago
|
| Reporter | ||
Comment 3•3 months ago
•
|
||
I think there were two distinct regressions here.
(A) No flicker at all.
(B) Occasional brief/subtle flicker of white, only for some of the Home/End keypresses. (These are often only at the extreme end of the page where I'm scrolling to, particularly for the first few Home/End keypresses (feels subjectively like it goes away after I've cycled a few times)
(C) More noticeable flickers (flash easier to see since it's typically full-viewport, and happens nearly every time, and I think lasts a bit longer)
Here's the regression-range for when we switched from (A) "No Flicker" to (B) "Brief/Subtle Flicker"
INFO: Last good revision: c25899d7b631470b983650ee254177381a62eaad (2020-10-02)
INFO: First bad revision: 33139800aca52aac192973fe24c9b74df7c87e00 (2020-10-03)
INFO: Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=c25899d7b631470b983650ee254177381a62eaad&tochange=33139800aca52aac192973fe24c9b74df7c87e00
Suspected: "Bug 1648799 - Enable partial present for webrender on linux"
Here's the regression range for when we switched From (B) "Occasional Brief/Subtle Flicker" to (C) "More noticeable flicker":
INFO: Last good revision: c15abadbc3c7008549ecb562326483749698ce46 (2022-03-21)
INFO: First bad revision: 3649f9875f3ded66f3f50b8ecfd21a0a7ea438f2 (2022-03-22)
INFO: Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=c15abadbc3c7008549ecb562326483749698ce46&tochange=3649f9875f3ded66f3f50b8ecfd21a0a7ea438f2
Suspected: not sure.
| Reporter | ||
Comment 4•3 months ago
|
||
(In reply to Daniel Holbert [:dholbert] from comment #3)
Here's the regression-range for when we switched from (A) "No Flicker" to (B) "Brief/Subtle Flicker"
[...]
Suspected: "Bug 1648799 - Enable partial present for webrender on linux"
Classifying under WR to match this^ bug's component. (This happens to be triggerable by scrolling, but it's probably not a bug in our scrolling code, but rather it's a case where we're failing to repaint in time, or something along those lines.)
Comment 5•3 months ago
|
||
I can reproduce on macos. I get the same 2022-03-21 regression range. I wonder if it could be https://hg-edge.mozilla.org/mozilla-central/rev/fb898843abd189664e7ae574c58dfba866015591 in that range. It does make changes to PresShell::CompleteScroll and smooth scroll code, so it could have had some effect her perhaps.
I think this is more likely to be a pan zoom issue. We've had several similar issues with checkerboarding (drawing white) when smooth scroll is disabled, and the fact that we see this on linux and mac with the same regression range suggests it is not a gfx issue. The smaller flicker that happens before 2022-03-21 we can look into after fixing the larger flicker that started happening then.
Comment 6•3 months ago
|
||
I have a different regression window on Windows11.
Regression window with general.smoothScroll = false:
https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=20ab625c5204936d33b883fd395e70a83b96552e&tochange=f92971683f01f9eb09cd7702abf355a059b85f11
Suspect: Bug 1756529 - Honor general.smoothScroll.[line|page|other] for keyboard scrolls with MSD physics r=botond,hiro
Updated•3 months ago
|
Comment 7•3 months ago
|
||
(In reply to Alice0775 White from comment #6)
I have a different regression window on Windows11.
Regression window with general.smoothScroll = false:
https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=20ab625c5204936d33b883fd395e70a83b96552e&tochange=f92971683f01f9eb09cd7702abf355a059b85f11Suspect: Bug 1756529 - Honor general.smoothScroll.[line|page|other] for keyboard scrolls with MSD physics r=botond,hiro
That range is a subset of the ranges dholbert and I found. Your local cache of builds is amazing! Thanks!
| Reporter | ||
Updated•3 months ago
|
Comment 8•3 months ago
|
||
Set release status flags based on info from the regressing bug 1756529
| Reporter | ||
Comment 9•3 months ago
|
||
(Let's call this 'wontfix' for release branches; any fix here can presumably ride the trains, given we've been shipping with this for quite a while before the bug was even noticed.)
| Reporter | ||
Updated•3 months ago
|
Updated•2 months ago
|
Updated•2 months ago
|
Description
•