1360215 - 3.72ms uninterruptible reflow at get_lineScrollAmount@chrome://global/content/bindings/scrollbox.xml:172:13

Status: RESOLVED INCOMPLETE

(Toolkit :: UI Widgets, defect, P4)

Description

[Marco Castelluccio [:marco]]

Here's the stack:

get_lineScrollAmount@chrome://global/content/bindings/scrollbox.xml:172:13
onxblwheel@chrome://global/content/bindings/scrollbox.xml:685:15

Comment 2

[Florian Quèze [:florian]]

          var totalWidth =
            elements[high].getBoundingClientRect()[end] - elements[low].getBoundingClientRect()[start];
http://searchfox.org/mozilla-central/rev/ce5ccb6a8ca803271c946ccb8b43b7e7d8b64e7a/toolkit/content/widgets/scrollbox.xml#170

Comment 3

[Dão Gottwald [:dao]]

What UI does this happen with? Overflowing menus?

Comment 4

[Marco Castelluccio [:marco]]

I was able to reproduce this again, but I haven't noticed exactly when this happens.

Comment 5

[Dão Gottwald [:dao]]

I don't think this is related to tabs.

Comment 6

[Neil Deakin]

The code here no longer exists; it was removed by 1387701.

Dao, is there still something to do here?

Comment 7

[Dão Gottwald [:dao]]

(In reply to Neil Deakin from comment #6)
> The code here no longer exists; it was removed by 1387701.
> 
> Dao, is there still something to do here?

It likely still flushes layout via scrollSize. I don't know if there's a reasonable way around this, and if this is even worth much effort. AFAIK this only affects overflowing menus.

Comment 8

[Mike Conley (:mconley) (:⚙️)]

(In reply to Dão Gottwald [::dao] from comment #7)
> It likely still flushes layout via scrollSize. I don't know if there's a
> reasonable way around this, and if this is even worth much effort. AFAIK
> this only affects overflowing menus.

Yes, scrollSize can still cause us to flush here:

https://searchfox.org/mozilla-central/rev/7a8c667bdd2a4a32746c9862356e199627c0896d/toolkit/content/widgets/scrollbox.xml#134-140

Comment 9

[Mike Conley (:mconley) (:⚙️)]

Re-queuing for triage by fxperf to reconcile with qf evaluation.

Comment 10

[Alex Thayer [:alexical] (she/her)]

Keeping it [fxperf:p3] unless someone disagrees. We're not hitting this often and it's not clear that there's a good way around it.

Comment 11

[Mike Conley (:mconley) (:⚙️)]

Thanks, dthayer. I concur - I'm going to drop the qf priority, too.

Comment 12

[Mike Conley (:mconley) (:⚙️)]

This binding is gone now. I'm guessing the reflow has moved elsewhere, but I'm going to close this one out as INCOMPLETE.