1969829 - Blue overlay screen appears on top of the game during fullscreen and back on Gardens of Time game

Status: VERIFIED FIXED

(Core :: DOM: Selection, defect)

Description

[Willow Martin]

Attachment: Blue screen issue appearing in full screen

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:139.0) Gecko/20100101 Firefox/139.0

Steps to reproduce:

Steps to Reproduce:
Launch the game via Facebook in firefox browser (tested on Windows 10).
Proceed through any game play.
Enter fullscreen and come back.
Observe the screen.

Note: this is happening on build version 139.

Actual results:

Game enters fullscreen, but a blue overlay appears, partially covering gameplay content and after coming back to normal mode also the blue overlay still appears.

Expected results:

Game should enter fullscreen and back without any blue overlay or visual obstructions.

Comment 1

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

The Bugbug bot thinks this bug should belong to the 'Core::Audio/Video: Playback' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 2

[Willow Martin]

Hi Team,
Just checking in to see if there are any updates regarding the issue I reported earlier (blue overlay during fullscreen). Please let me know if you need any additional information or if there's anything I can help with to assist the investigation.

Thanks!

Comment 3

[Ru Cindrea]

This seems to be the same problem as described in https://bugzilla.mozilla.org/show_bug.cgi?id=1969493 The other ticket is reported for Android, but the problem is more general with the Firefox 139 version.

Facebook WebGL games seems to exhibit this when entering full screen, and there is a Unity Forum discussion on this with a suggested workaround: https://discussions.unity.com/t/odd-issue-with-focus-in-firefox/1651226

From the thread: "Firefox wants to consider the canvas as selectable and as it goes into fullscreen, Firefox is erroneously thinking it should select the canvas element when you click on it."

I reproduced this on 139.0.1 (aarch64) on MacOS 15.5 (24F74).

Comment 4

[Jim Mathies [:jimm]]

Looks like a dev tools selector problem.

Comment 5

[Nicolas Chevobbe [:nchevobbe]]

(In reply to Ru Cindrea from comment #3)

This seems to be the same problem as described in https://bugzilla.mozilla.org/show_bug.cgi?id=1969493 The other ticket is reported for Android, but the problem is more general with the Firefox 139 version.

Facebook WebGL games seems to exhibit this when entering full screen, and there is a Unity Forum discussion on this with a suggested workaround: https://discussions.unity.com/t/odd-issue-with-focus-in-firefox/1651226

From the thread: "Firefox wants to consider the canvas as selectable and as it goes into fullscreen, Firefox is erroneously thinking it should select the canvas element when you click on it."

I reproduced this on 139.0.1 (aarch64) on MacOS 15.5 (24F74).

Unfortunately I can't reproduce in 139 myself. Ru, could you run https://mozilla.github.io/mozregression/ to see if we can get a regression window?

Comment 6

[Nicolas Chevobbe [:nchevobbe]]

Bug 1954956 landed in 138 (and was uplifted to 137) and was about fullscreen, so maybe we should check that too

Comment 7

[Ru Cindrea]

I ran the mozgression tool, I attached the full log and a screenshot with the results. Here is the last part:

2025-06-12T09:07:02.761000: INFO : Launching /private/var/folders/84/drkv10r94n79d07nls2hpb700000gn/T/tmp8i_7prex/Firefox Nightly.app/Contents/MacOS/firefox
2025-06-12T09:07:02.762000: INFO : Application command: /private/var/folders/84/drkv10r94n79d07nls2hpb700000gn/T/tmp8i_7prex/Firefox Nightly.app/Contents/MacOS/firefox https://redsclub.com/reds-arcade-testing/ -foreground -profile /var/folders/84/drkv10r94n79d07nls2hpb700000gn/T/tmp2lsb2s4q.mozrunner
2025-06-12T09:07:02.770000: INFO : application_buildid: 20250413073515
2025-06-12T09:07:02.770000: INFO : application_changeset: c9fadced05b5db792c315e810a145b4b2926797f
2025-06-12T09:07:02.770000: INFO : application_display_name: Firefox Nightly
2025-06-12T09:07:02.770000: INFO : application_id: {ec8030f7-c20a-464f-9b0e-13a3a9e97384}
2025-06-12T09:07:02.771000: INFO : application_name: Firefox
2025-06-12T09:07:02.771000: INFO : application_remotingname: firefox-default
2025-06-12T09:07:02.771000: INFO : application_repository: https://hg.mozilla.org/integration/autoland
2025-06-12T09:07:02.771000: INFO : application_vendor: Mozilla
2025-06-12T09:07:02.771000: INFO : application_version: 139.0a1
2025-06-12T09:07:02.771000: INFO : platform_buildid: 20250413073515
2025-06-12T09:07:02.771000: INFO : platform_changeset: c9fadced05b5db792c315e810a145b4b2926797f
2025-06-12T09:07:02.771000: INFO : platform_repository: https://hg.mozilla.org/integration/autoland
2025-06-12T09:07:02.771000: INFO : platform_version: 139.0a1
2025-06-12T09:07:40.041000: INFO : b'2025-06-12 09:07:40.040 firefox[48563:52267735] WARNING: AVCaptureDeviceTypeExternal is deprecated for Continuity Cameras. Please use AVCaptureDeviceTypeContinuityCamera and add NSCameraUseContinuityCameraDeviceType to your Info.plist.'
2025-06-12T09:08:08.940000: INFO : Narrowed integration regression window from [2a6769a4, ec51400e] (3 builds) to [c9fadced, ec51400e] (2 builds) (~1 steps left)
2025-06-12T09:08:08.946000: DEBUG : Starting merge handling...
2025-06-12T09:08:08.946000: DEBUG : Using url: https://hg.mozilla.org/integration/autoland/json-pushes?changeset=ec51400e8dd7aad3c1347c0746018224e20e5d80&full=1
2025-06-12T09:08:08.946000: DEBUG : redo: attempt 1/3
2025-06-12T09:08:08.946000: DEBUG : redo: retry: calling _default_get with args: ('https://hg.mozilla.org/integration/autoland/json-pushes?changeset=ec51400e8dd7aad3c1347c0746018224e20e5d80&full=1',), kwargs: {}, attempt #1
2025-06-12T09:08:08.948000: DEBUG : urllib3.connectionpool: Resetting dropped connection: hg.mozilla.org
2025-06-12T09:08:09.674000: DEBUG : urllib3.connectionpool: https://hg.mozilla.org:443 "GET /integration/autoland/json-pushes?changeset=ec51400e8dd7aad3c1347c0746018224e20e5d80&full=1 HTTP/11" 302 0
2025-06-12T09:08:11.113000: DEBUG : urllib3.connectionpool: https://hg-edge.mozilla.org:443 "GET /integration/autoland/json-pushes?changeset=ec51400e8dd7aad3c1347c0746018224e20e5d80&full=1 HTTP/11" 200 None
2025-06-12T09:08:11.118000: DEBUG : Found commit message:
Bug 1959780 - Make GetRangeForFrame() always return the inside range for non-replaced elements, not just blocks and editing hosts. r=masayuki

Otherwise you can't insert stuff in something like an empty
inline-block, there are tests for that luckily, yay.

Differential Revision: https://phabricator.services.mozilla.com/D245266

2025-06-12T09:08:11.119000: DEBUG : Did not find a branch, checking all integration branches
2025-06-12T09:08:11.123000: INFO : The bisection is done.
2025-06-12T09:08:11.124000: INFO : Stopped

Comment 8

[Ru Cindrea]

Attachment: mozgression_full.log

Mozgression full log

Comment 9

[Ru Cindrea]

Attachment: mozgression_result.png

Comment 10

[Nicolas Chevobbe [:nchevobbe]]

Thanks a lot Ru
So it looks like Bug 1959780 is causing the issue.
Can you tell me which website you're testing this on? https://lande-audio.itch.io/lunar-crash-defense ? Do you need to have DevTools opened to see the issue? Can you post detailed steps to reproduce so we can investigate ourselves?

Thanks again

Comment 11

[Ru Cindrea]

I have been testing this with an internal Facebook game, so I cannot share that link as it's not available publicly. But the original reporter was using Gardens of Time game on Facebook and I just tried it and was able to reproduce the problem with that one too.

So here are the steps:

1. Go to https://apps.facebook.com/gardensoftime/
2. Sign-in with Facebook if you're not already signed in.
3. Go through the game tutorial up to the point where you are allowed to put the game in Full Screen mode from the top right corner button.
4. Once game is in full-screen, click anywhere on the screen -> this causes the canvas to get selected and so the screen gets the blue tint all over it. As you are in full-screen mode, you are not able to click outside the canvas, so you cannot get rid of the blue tint.

I have tried with the game you shared (https://lande-audio.itch.io/lunar-crash-defense), but I can't reproduce it there.

Comment 12

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

:emilio, since you are the author of the regressor, bug 1959780, could you take a look? Also, could you set the severity field?

For more information, please visit BugBot documentation.

Comment 13

[Nicolas Chevobbe [:nchevobbe]]

Attachment: selected canvas element after exiting fullscreen with escape

Okay Thanks Ru, I did try with gardensoftime yesterday but couldn't reproduce for some reason, now I can see it
I'm attaching a screenshot of the page after triggering the issue, and then hitting Escape to exit fullscreen. We can indeed see that the canvas element is "selected" (my OS selection color is pink-ish , not blue like the one from the original reporter)
I'm reproducing without DevTools opened, and I don't think DevTools are involved here.

Comment 14

[Emilio Cobos Álvarez [:emilio]]

I don't have a facebook account, would it be possible for someone to provide steps that don't require it? I'll look into borrowing one somehow... Last time I tried to create a throwaway account it got insta-closed.

Comment 15

[Nicolas Chevobbe [:nchevobbe]]

Attachment: simple test page

Emilio, I think the attached page should reproduce. Click the fullscreen button, then when in fullscreen, click on the rect -> canvas gets selected

Comment 16

[Nicolas Chevobbe [:nchevobbe]]

Attachment: Simple test page

Comment 17

[Emilio Cobos Álvarez [:emilio]]

Thanks! I can repro with that test-case. Masayuki, we're hitting this codepath extending the selection, even tho aFocusMode is kCollapseToNewPoint. That seems surprising to me?

Let me try adjusting that and see what breaks, but extending the selection when trying to collapse it seems backwards to me :)

Comment 18

[Emilio Cobos Álvarez [:emilio]]

Attachment: WIP: Bug 1969829 - Don't extend selection in kCollapseToNewPoint mode.

Comment 19

[Emilio Cobos Álvarez [:emilio]]

Attachment: reduced test-case.

This doesn't need fullscreen or anything.

Comment 20

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

Set release status flags based on info from the regressing bug 1959780

Comment 21

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

Well, I don't understand what's going on. Why aContentEndOffset != aContentOffset? One of them is adjusted only for user-select style by the caller? If so, it's the caller's bug. The bug keeps staying even after applying your patch once Ctrl (or Command) is pressed at clicking the <canvas>.

Comment 22

[Emilio Cobos Álvarez [:emilio]]

So... canvas usually has user-select: none since bug 294852. But if fullscreen (or otherwise overridden to user-select: text) then this issue happens. I need to dig on why it doesn't happen on, e.g. a regular <img> element, it's indeed a bit surprising.

Comment 23

[Emilio Cobos Álvarez [:emilio]]

Bug 1754459 is also relevant, and is the reason why we don't see this for non-draggable images...

Ok, so I think I have a low-risk fix to uplift.

Comment 24

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 1969829 - Non-editable <canvas> elements should use the frame edge for selection targetting. r=masayuki,saschanaz

Just like images.

Comment 25

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 1969829 - Remove now unnecessary user-select override on <canvas>. r=masayuki,#layout

Comment 26

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 1969829 - Use frame edge for all replaced elements consistently. r=masayuki

Comment 27

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/fd1e4fe85947
https://hg.mozilla.org/integration/autoland/rev/b47b552e50ed
Non-editable <canvas> elements should use the frame edge for selection targetting. r=masayuki

Comment 28

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/f1d70d947db4
https://hg.mozilla.org/integration/autoland/rev/4b3277962525
Remove now unnecessary user-select override on <canvas>. r=masayuki,dshin

Comment 29

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/53155 for changes under testing/web-platform/tests

Comment 30

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/cb3cdd175aef
https://hg.mozilla.org/integration/autoland/rev/c4e149968430
Use frame edge for all replaced elements consistently. r=masayuki

Comment 31

[Norisz Fay [:noriszfay]]

https://hg.mozilla.org/mozilla-central/rev/b47b552e50ed
https://hg.mozilla.org/mozilla-central/rev/4b3277962525
https://hg.mozilla.org/mozilla-central/rev/c4e149968430

Comment 32

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Upstream PR merged by moz-wptsync-bot

Comment 33

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

The patch landed in nightly and beta is affected.
:emilio, is this bug important enough to require an uplift?

• If yes, please nominate the patch for beta approval.
  • See https://wiki.mozilla.org/Release_Management/Requesting_an_Uplift for documentation on how to request an uplift.
• If no, please set status-firefox140 to wontfix.

For more information, please visit BugBot documentation.

Comment 34

[Emilio Cobos Álvarez [:emilio]]

Comment on attachment 9494679 [details]
Bug 1969829 - Non-editable <canvas> elements should use the frame edge for selection targetting. r=masayuki,saschanaz

Beta/Release Uplift Approval Request

• User impact if declined/Reason for urgency: comment 0
• Is this code covered by automated tests?: Yes
• Has the fix been verified in Nightly?: Yes
• Needs manual test from QE?: Yes
• If yes, steps to reproduce: comment 0
• List of other uplifts needed: none
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): Matches images, very targetted fix.
• String changes made/needed: none
• Is Android affected?: Yes

Comment 35

[Donal Meehan [:dmeehan]]

Comment on attachment 9494679 [details]
Bug 1969829 - Non-editable <canvas> elements should use the frame edge for selection targetting. r=masayuki,saschanaz

Approved for 140.0 RC1
Rejecting release uplift request since we are in RC week for

Comment 36

[Pulsebot]

https://github.com/mozilla-firefox/firefox/commit/4e9167a7b113
https://hg.mozilla.org/releases/mozilla-beta/rev/5741161b3e46

Comment 37

[Daniel Bodea [:danibodea]]

I can't seem to reproducethe issue using the original game mentioned in comment 11: https://apps.facebook.com/gardensoftime/ .
The attached demo video shows that the game is interrupted (frozen for a few seconds, then it shows the interrupt message) when the user clicks the "Exit Full Screen (ESC)" button from the "Full Screen notification" displayed when entering Full Screen. The blue overlay issue appears to be somewhat linked with that interrupt.

The issue could only be reproduced using the Reduced Test Case from comment 19, in Nightly v141.0a1 from 2025-06-14 and Beta v140.0b9; The reduced test case behaved as followed:

1. Clicking the green rectangle selects the rectangle.
2. Clicking outside of the rectangle, selects the rectangle.

The fix is confirmed in Beta v140.0 (RC) and Nightly v141.0a1 from 2025-06-17. The reduced test case behaved as followed:

1. Clicking the green rectangle, does NOT select it.
2. Clicking outside the green rectangle, does select the green rectangle.
3. Double-clicking the green rectangle, selects the green rectangle.

Assuming that the reduced test case behaves as expected, I can confirm this fix in Windows 10, MacOS 15 and Ubuntu 22.

Comment 38

[Pulsebot]

https://github.com/mozilla-firefox/firefox/commit/f18a70c3608c
https://hg.mozilla.org/releases/mozilla-beta/rev/1a1c782c259a

Revert "Bug 1969829 - Remove now unnecessary user-select override on <canvas>. r=masayuki,dshin" for causing bug 1978673 a=backout
Backed out of 142.0b5 for causing bug 1978673

Comment 39

[Dianna Smith [:diannaS]]

As im not familiar with the behavior of each patch, not sure what to change the tracking flag to for 142. (not all patches were reverted)

Comment 40

[Pulsebot]

https://github.com/mozilla-firefox/firefox/commit/2d31f673efdc
https://hg.mozilla.org/releases/mozilla-release/rev/f318c7723a0a