2008277 - Pointerout and Mouseout behave differently over 3rd-party iframes (Pointerout is unreliable)

Status: RESOLVED FIXED

(Core :: DOM: UI Events & Focus Handling, defect)

Description

[Paul Rhomberg]

Attachment: index.html

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:146.0) Gecko/20100101 Firefox/146.0

Steps to reproduce:

I created a <div> above a third party iframe and attached an event listener for "pointerout". Then I noticed that it is unreliable compared to mouseout with this setup in Firefox.

In Google Chrome, this works reliably and Pointerout triggers 100% of the time.

Actual results:

When you move the cursor out of an element placed above a 3rd-party iframe, pointerout only triggers rarely (maybe ~10% of the time) while mouseout works reliably.

Expected results:

Pointerout should trigger reliably. Attached is a demo html file.

Comment 1

[Paul Rhomberg]

Attachment: Video showcasing the bug in Firefox

Comment 2

[Paul Rhomberg]

Attachment: Showcasing this in Chrome (works as it should)

Comment 3

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'Core::DOM: Core & HTML' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 4

[Alice0775 White]

Regression window:
https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=383986e2f5cb835a55c47bbd6e15d81035ca0784&tochange=d6e8528f0a936df88369c71f4e390e31a4d621a1

After Bug 1732358: the pointerout event never fired until mouse out the iframe.

And
https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=35eb9cf6b15a28d3c9954b12ad2eb54cd8a43d3b&tochange=fe63e89a0316a6a6cc84d02b0d7b98a0f1ba863f

After Bug 1966551 : The pointerout event fires significantly late and is overwritten by subsequent events.

Comment 5

[BugBot [:suhaib / :marco/ :calixte]]

:nika, since you are the author of the regressor, bug 1732358, could you take a look? Also, could you set the severity field?

For more information, please visit BugBot documentation.

Comment 6

[BugBot [:suhaib / :marco/ :calixte]]

The bug has a release status flag that shows some version of Firefox is affected, thus it will be considered confirmed.

Comment 7

[Nika Layzell [:nika] (ni? for response)]

This appears to be some kind of focus/input behaviour change which only started happening after Fission was enabled.

Redirecting to masayuki who may have more context as to what component this needs to be directed to.

Comment 8

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Looks like it's truly our bug. Unfortunately I don't have much time for this for a while. I'll be back later, or, edgar might be interested into this.

Comment 9

[Edgar Chen [:edgar]]

We do generate the pointerout event, but that happens when you stop moving the cursor on the iframe.
So I think this is something about the timing of generating the pointer boundary event. For mouse event, we generate additional IPC to notify the processes about the mouse leaving, see https://searchfox.org/firefox-main/rev/8e6b6cb1dd0fdd9838e2359219e2b8d3b84490b2/dom/events/EventStateManager.cpp#2154-2200, I guess target process doesn't generate the pointer boundary event from that for some reason.

Comment 10

[Edgar Chen [:edgar]]

https://treeherder.mozilla.org/jobs?repo=try&landoCommitID=173022

Comment 11

[Edgar Chen [:edgar]]

I also noticed that when cursor moves to frame that lives in different process, we dispatch the mouseover event first then the pointerover event, whereas other browsers dispatch pointerover event first. I think this is also something related to the additional IPC that is to notify about the mouse entering.

Comment 12

[Edgar Chen [:edgar]]

Attachment: Bug 2008277 - Part 1: Add tests for pointer boundary events when moving the mouse into and out of an iframe;

Comment 13

[Edgar Chen [:edgar]]

Attachment: Bug 2008277 - Part 2: Generate pointer boundary event when mouse moves to iframe that lives in different process;

Comment 14

[Edgar Chen [:edgar]]

Attachment: Bug 2008277 - Part 3: Store pointer location when receiving the eMouseEnterIntoWidget event;

Comment 15

[Edgar Chen [:edgar]]

https://treeherder.mozilla.org/jobs?repo=try&landoCommitID=174869

Comment 16

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/f79f6d094333
https://hg.mozilla.org/integration/autoland/rev/48607a060b2b
Part 1: Add tests for pointer boundary events when moving the mouse into and out of an iframe; r=masayuki
https://github.com/mozilla-firefox/firefox/commit/c907508b77d4
https://hg.mozilla.org/integration/autoland/rev/23c12f806687
Part 2: Generate pointer boundary event when mouse moves to iframe that lives in different process; r=masayuki
https://github.com/mozilla-firefox/firefox/commit/54735c546659
https://hg.mozilla.org/integration/autoland/rev/e13094c5a82e
Part 3: Store pointer location when receiving the eMouseEnterIntoWidget event; r=masayuki

Comment 17

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/59af38d29bbe
https://hg.mozilla.org/integration/autoland/rev/4e01810cc1ee
Revert "Bug 2008277 - Part 3: Store pointer location when receiving the eMouseEnterIntoWidget event; r=masayuki" for causing mochitest failures @ test_pointer_boundary_events_iframe.html

Comment 18

[amarc]

Backed out for causing mochitest failures @ test_pointer_boundary_events_iframe

• Backout link
• Push with failures
• Failure Log

Comment 19

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/57924745447c
https://hg.mozilla.org/integration/autoland/rev/8395b8919a1a
Part 1: Add tests for pointer boundary events when moving the mouse into and out of an iframe; r=masayuki
https://github.com/mozilla-firefox/firefox/commit/284741694d9b
https://hg.mozilla.org/integration/autoland/rev/e5f0a14a7726
Part 2: Generate pointer boundary event when mouse moves to iframe that lives in different process; r=masayuki
https://github.com/mozilla-firefox/firefox/commit/17d39e8a52c5
https://hg.mozilla.org/integration/autoland/rev/a77158ef95d1
Part 3: Store pointer location when receiving the eMouseEnterIntoWidget event; r=masayuki

Comment 20

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/b0d32bdee951
https://hg.mozilla.org/integration/autoland/rev/95d47659ea99
Revert "Bug 2008277 - Part 3: Store pointer location when receiving the eMouseEnterIntoWidget event; r=masayuki" for causing mochitest failures @ test_pointer_boundary_events_iframe

Comment 21

[amarc]

Backed out for causing mochitest failures

• Backout link
• Push with failures
• Failure Log

Comment 22

[Edgar Chen [:edgar]]

(In reply to amarc from comment #21)

• Push with failures

Oh, all failures are on GeckoView, the APZ controller of GeckoView lives in different thread, I guess the mIsSynthesizedForTests flag isn't propagated correctly, so the async synthesized mousemove event is always ignored by EventStateManager...

Err, hmm, probably something else...

Comment 23

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/e181f4cca6f6
https://hg.mozilla.org/integration/autoland/rev/00c06139023f
Part 1: Add tests for pointer boundary events when moving the mouse into and out of an iframe; r=masayuki
https://github.com/mozilla-firefox/firefox/commit/62427a03d334
https://hg.mozilla.org/integration/autoland/rev/5afa93c303d2
Part 2: Generate pointer boundary event when mouse moves to iframe that lives in different process; r=masayuki
https://github.com/mozilla-firefox/firefox/commit/26b89e9f01ef
https://hg.mozilla.org/integration/autoland/rev/1c06425763ba
Part 3: Store pointer location when receiving the eMouseEnterIntoWidget event; r=masayuki

Comment 24

[Iulian Moraru]

https://hg.mozilla.org/mozilla-central/rev/00c06139023f
https://hg.mozilla.org/mozilla-central/rev/5afa93c303d2
https://hg.mozilla.org/mozilla-central/rev/1c06425763ba