1359259 - 0.92ms uninterruptible reflow at PopupNotifications_showPanel@resource://gre/modules/PopupNotifications.jsm:932:9

Status: NEW

(Toolkit :: Notifications and Alerts, defect, P4)

Description

[Hazel Ward]

Here's the stack:

PopupNotifications_showPanel@resource://gre/modules/PopupNotifications.jsm:932:9
PopupNotifications_update@resource://gre/modules/PopupNotifications.jsm:1097:9
anchorVisibilityChange@resource://gre/modules/PopupNotifications.jsm:579:7
PopupNotifications_locationChange@resource://gre/modules/PopupNotifications.jsm:560:7
onLocationChange@chrome://browser/content/browser.js:5084:7
callListeners@chrome://browser/content/tabbrowser.xml:492:24
_callProgressListeners@chrome://browser/content/tabbrowser.xml:513:13
_callProgressListeners@chrome://browser/content/tabbrowser.xml:563:22
onLocationChange@chrome://browser/content/tabbrowser.xml:845:17
_callProgressListeners@resource://gre/modules/RemoteWebProgress.jsm:179:11
receiveMessage@resource://gre/modules/RemoteWebProgress.jsm:265:7

Comment 1

[Florian Quèze [:florian]]

http://searchfox.org/mozilla-central/rev/baf47b352e873d4516d7656186d6d7c7447d3873/toolkit/modules/PopupNotifications.jsm#931

    // If the anchor element is hidden or null, fall back to the identity icon.
    if (!anchorElement || (anchorElement.boxObject.height == 0 &&
                           anchorElement.boxObject.width == 0)) {

I hope this test would still work with a getBoundsWithoutFlushing call instead, but I'm not sure how to verify that there's no edge case this would break.

Comment 3

[Johann Hofmann [:johannh]]

(In reply to Florian Quèze [:florian] [:flo] from comment #1)
> http://searchfox.org/mozilla-central/rev/
> baf47b352e873d4516d7656186d6d7c7447d3873/toolkit/modules/PopupNotifications.
> jsm#931
> 
>     // If the anchor element is hidden or null, fall back to the identity
> icon.
>     if (!anchorElement || (anchorElement.boxObject.height == 0 &&
>                            anchorElement.boxObject.width == 0)) {
> 
> I hope this test would still work with a getBoundsWithoutFlushing call
> instead, but I'm not sure how to verify that there's no edge case this would
> break.

FWIW I think this needs the flushing to check if the anchor elements are visible. Maybe there's another way to do it but I don't think we can simply substitute with getBoundsWithoutFlushing here.

Comment 5

[Mike Conley (:mconley) (:⚙️)]

I can still reproduce this by visiting a page that opens up a permission panel - example, GeoLocation.

Comment 6

[:Felipe Gomes (needinfo for replies!)]

(not startup related: fxperf:p2 -> fxperf:p3)