1361635 - 1.19ms uninterruptible reflow from window.innerWidth at getContentSize@resource://gre/modules/PageThumbUtils.jsm:112:9

Status: NEW

(Core :: DOM: Core & HTML, defect, P3)

Description

[hazel ward]

Here's the stack:

getContentSize@resource://gre/modules/PageThumbUtils.jsm:112:9
createSnapshotThumbnail@resource://gre/modules/PageThumbUtils.jsm:144:41
_captureToCanvas@resource://gre/modules/PageThumbs.jsm:292:5
captureToCanvas@resource://gre/modules/PageThumbs.jsm:222:5
refreshThumbAndLabel@chrome://tabcenter/content/vertical-tabbrowser.xml:827:17
refreshAllTabs@resource://gre/modules/commonjs/toolkit/loader.js -> resource://tabcentertest1-at-mozilla-dot-com/verticaltabs.js:1291:7
actuallyResizeTabs@resource://gre/modules/commonjs/toolkit/loader.js -> resource://tabcentertest1-at-mozilla-dot-com/verticaltabs.js:1277:9
delayResizeTabs/this.resizeTimeout<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://tabcentertest1-at-mozilla-dot-com/verticaltabs.js:1172:55

Comment 1

[Florian Quèze [:florian]]

http://searchfox.org/mozilla-central/rev/abe68d5dad139e376d5521ca1d4b7892e1e7f1ba/toolkit/components/thumbnails/PageThumbUtils.jsm#112

    let width = aWindow.innerWidth - sbWidth.value;
    let height = aWindow.innerHeight - sbHeight.value;

Do we know why this need flushing? Could it be fixed in a way similar to bug 1358809? The above stack looks like this is running off a timer after a resize event, so using getBoundsWithoutFlushing (or similar) would probably give correct results here.

Comment 2

[Dão Gottwald [:dao]]

(In reply to Florian Quèze [:florian] [:flo] from comment #1)
> http://searchfox.org/mozilla-central/rev/
> abe68d5dad139e376d5521ca1d4b7892e1e7f1ba/toolkit/components/thumbnails/
> PageThumbUtils.jsm#112
> 
>     let width = aWindow.innerWidth - sbWidth.value;
>     let height = aWindow.innerHeight - sbHeight.value;
> 
> Do we know why this need flushing? Could it be fixed in a way similar to bug
> 1358809?

Note that aWindow is a content window here.

This might be more similar to bug 1307134, i.e. accessing an iframe's innerWidth should flush layout in ancestor content windows, but accessing a top-level content window's innerWidth shouldn't flush layout.

Comment 3

[Hsin-Yi Tsai (she/her)[:hsinyi]]

Hi Marco, we don't use P4 for DOM bugs. We use P3 (backlog) and P5 (not plan to fix it but happy to take a patch). Do you think we could remap your P4 into either of them? Thank you.

Comment 4

[Marco Mucci [:MarcoM]]

(In reply to Hsin-Yi Tsai [:hsinyi] from comment #3)
> Hi Marco, we don't use P4 for DOM bugs. We use P3 (backlog) and P5 (not plan
> to fix it but happy to take a patch). Do you think we could remap your P4
> into either of them? Thank you.

I'll remove from Photon Performance and you can adjust the priority level as you wish.