1273091 - Mouse cursor does not disappear in html5 fullscreen video on Windows

Status: VERIFIED FIXED

(Core :: Widget: Win32, defect, P1)

Description

[algkmn]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:46.0) Gecko/20100101 Firefox/46.0
Build ID: 20160502172042

Steps to reproduce:

when I go fullscreen video on youtube, mouse cursor doesn't disappear. if I move mouse cursor a little bit in fullscreen video mouse cursor disappears

fyi: 
I tried disable all add-ons, uninstall and clean install after all of this I reinstall fresh windows10 then install firefox (without any addons or settings) it were still happening.

it's not happening on chrome, opera, or edge.

sorry my bad english

Gif explains issue: https://vgy.me/PrzI48.gif


Actual results:

mouse corsor does not disappear when go fullscreen


Expected results:

mouse corsor should disappear when go fullscreen

Comment 1

[Loic]

Are you using a software to manage your mouse?

Comment 2

[algkmn]

no, I am using mac os x mouse cursor theme. (problem already happening before using this theme)

Comment 3

[Liviu Cirdei]

User Agent 	Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:49.0) Gecko/20100101 Firefox/49.0
Build ID 	20160519030232

The issue is reproducible on Windows. I tested on Windows 10 and Windows 7, and on both systems this is reproducible. I used the current release (46.0.1) and latest Nightly (49.0a1).

I tried to make a regression range and found this pushlog:

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=8021382da9722db0ad97ebd93698b69a74f0d9b0&tochange=5efb9a11196ba113b830cf4717d85513e3468783

Note: On Mac OS X 10.11 and Ubuntu 15.04 this is not reproducible.

Comment 4

[Loic]

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=0288a0a7003ffba272dc8040567e225d7115d9bc&tochange=5efb9a11196ba113b830cf4717d85513e3468783

Comment 5

[Chris H-C :chutten|PTO (back March 23)]

Wow, they weren't kidding when they tried to warn me that fixing bug 506815 permanently was fraught with peril.

I don't know what code is responsible for hiding the cursor. Can anyone point me in the right direction?

Random speculation: we're waiting for a mouseleave or change in the mouse tracking window which, in pre-506815, would occur on the next 200ms timer edge when we hittested. Post-506815, Windows may wait for the first mouse event over a non-app window... or something. Not sure.

Comment 6

[Marcia Knous [:marcia]]

Let's track this for 48/49 since the mouse cursor should disappear.

Comment 7

[Liz Henry (:lizzard) (privacy/fingerprinting team)]

jonathan, can you help shed some light here or help find an owner for this bug? You have won the award for fixing a bunch of recent bugs in this component. The prize is a needinfo!

Comment 8

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

FWIW I am using win10 and I cannot repro this issue.

Comment 9

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

this is a valid bug but not a release blocker for Fx47. Too late for uplifting a fix.

Comment 10

[Jonathan Kew [:jfkthame]]

(In reply to Liz Henry (:lizzard) (needinfo? me) from comment #7)
> jonathan, can you help shed some light here or help find an owner for this
> bug? You have won the award for fixing a bunch of recent bugs in this
> component. The prize is a needinfo!

Gee, thanks! I should've known better than to touch it... ;)

I'm afraid I don't really know anything about the event handling / cursor management going on here. Maybe Masayuki would be a bit more familiar with this?

Comment 11

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

I guess that opening fullscreen window doesn't cause synthesized mousemove event or it's fired before setting CSS cursor property on the fullscreen window.

Comment 12

[Liz Henry (:lizzard) (privacy/fingerprinting team)]

David can you help find an owner for this bug? Thanks!

Comment 13

[David Bolter [:davidb] (NeedInfo me for attention)]

Jim any ideas who could take a look? (Masayuki, feel free to take this one if you want)

Comment 14

[Jim Mathies [:jimm]]

I'm not able to reproduce. Tracy can you work up good str here? Maybe the trick is to use the keyboard to go into fullscreen without moving the mouse.

Comment 15

[Loic]

(In reply to Jim Mathies [:jimm] from comment #14)
> I'm not able to reproduce. Tracy can you work up good str here? Maybe the
> trick is to use the keyboard to go into fullscreen without moving the mouse.

Nope, just load any YT video like https://www.youtube.com/watch?v=Ex8kDMKnLaE then click on the fullscreen icon and don't touch the mouse anymore. You'll see the mouse pointer doesn't disappear.

Comment 16

[Tracy Walker [:tracy]]

Yes, Loic is correct on STR:

1) load https://www.youtube.com/watch?v=Ex8kDMKnLaE 
2) click on the fullscreen icon
3) do no move mouse after step 2 click

Expected results:  mouse pointer disappears

Tested results: finger point mouse pointer remains

This is Windows only.

Win 7 (VM), Nightly, 51.0a1, 20160802030437 - FAIL (e10s and non-e10s window)
Mac 10.11, Nightly, 51.0a1, 20160802030437 - PASS
Ubuntu 14.04(VM), Nightly 51.0a1, 20160802030437 - PASS

Comment 17

[algkmn]

by the way I guess it's happening every single html5 full screen content. not only videos. (I'm not sure)

thank you

Comment 18

[Gerry Chang [:gchang]]

Hi Jim,
Do you have any idea who can help on this important issue?

Comment 19

[Jim Mathies [:jimm]]

Aaron is going to poke at this when he has a chance.

Comment 20

[David Bolter [:davidb] (NeedInfo me for attention)]

Marking fix-optional for 49 because late in beta, but we really really hope for a fix here.

Comment 21

[algkmn]

is there any hope to fix this issue? :)

Comment 22

[Andrew Overholt [:overholt]]

Aaron, have you had any time to take a look?

Comment 23

[Andrew Overholt [:overholt]]

Jim, do you know who could investigate?

Comment 24

[Jim Mathies [:jimm]]

Yep, spoke with davidp, he's going to take a look at this.

Comment 25

[David Parks [:handyman]]

Attachment: Send an eMouseEnterIntoWidget immediately after entering fullscreen

Tests:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=9a72a7ba47b1bd1994684e3da11af0c467714326

Comment 26

[David Parks [:handyman]]

Comment on attachment 8812873 [details] [diff] [review]
Send an eMouseEnterIntoWidget immediately after entering fullscreen

Hi, Masayuki.  Hopefully, you're a good pick to review this.

The deal was that the OS sends a WM_MOUSELEAVE, which triggers a eMouseExitFromWidget.  This the mouse position stored in the PresShell is cleared by this message:
https://dxr.mozilla.org/mozilla-central/rev/0534254e9a40b4bade2577c631fe4cfa0b5db41d/layout/base/nsPresShell.cpp#6787

There are a number of synthetic mouse moves sent after that but they are all ignored since the mouse position is invalid:
https://dxr.mozilla.org/mozilla-central/rev/0534254e9a40b4bade2577c631fe4cfa0b5db41d/layout/base/nsPresShell.cpp#5495

By sending the eMouseEnterIntoWidget when we are done entering fullscreen instead of waiting for the next mouse move, we restore the presshell's mouse position cache:
https://dxr.mozilla.org/mozilla-central/rev/0534254e9a40b4bade2577c631fe4cfa0b5db41d/layout/base/nsPresShell.cpp#6762

Comment 27

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

Comment on attachment 8812873 [details] [diff] [review]
Send an eMouseEnterIntoWidget immediately after entering fullscreen

># HG changeset patch
># User David Parks <[e-mail]>
># Date 1479755237 28800
>#      Mon Nov 21 11:07:17 2016 -0800
># Node ID 022089979908a86fbd83a0181ffb2f5df06104f4
># Parent  79feeed4293336089590320a9f30a813fade8e3c
>Bug 1273091 - Mouse cursor does not disappear in html5 fullscreen video on Windows
>When we jump to fullscreen, the OS sends a WM_MOUSELEAVE, which we turn into a eMouseExitFromWidget.  The eMouseEnterIntoWidget was previously only sent when the mouse moved into the widget, which required the mouse to move.  When entering fullscreen, we want the eMouseEnterIntoWidget to happen right away.

Did you test this patch with multi-display?

>+// Used to remember the wParam (i.e. currently pressed modifier keys)
>+// and lParam (i.e. last mouse position) in screen coordinates from
>+// the previous mouse-exit.  This is static since it is not
>+// associated with a particular widget (since we exited the widget).
>+static WPARAM sMouseExitwParam = 0;
>+static LPARAM sMouseExitlParamScreen = 0;

I'd be happy if these are static member of nsWindow because they're not referred from outside of nsWindow.

r=masayuki if this works fine even in multi-display.

Comment 28

[David Parks [:handyman]]

Attachment: Send an eMouseEnterIntoWidget immediately after entering fullscreen (r=masayuki)

Checked with multiple displays -- the behavior is correct.

Also made the static values members of nsWindow.

Comment 29

[(No longer employed by Mozilla) Aaron Klotz]

Thanks for fixing this, David! \o/

Comment 30

[Loic]

Yes, this bug was so annoying when watching video in FS mode on YouTube.

Comment 31

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/efad76d74c06
Mouse cursor does not disappear in html5 fullscreen video on Windows. r=masayuki

Comment 32

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/efad76d74c06

Comment 33

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Too late to fix in 50.1.0 release

Comment 34

[Gerry Chang [:gchang]]

Hi :handyman, 
Is this worth uplifting to Beta51?

Comment 35

[David Parks [:handyman]]

Hey Gerry,

Jimm had some thoughts about pushing this to Aurora/Beta -- he wanted to wait a bit to see if any regressions came up.  Jim, do you think we're good to uplift this?  Aurora _and_ beta?

Comment 36

[Jim Mathies [:jimm]]

(In reply to David Parks [:handyman] from comment #35)
> Hey Gerry,
> 
> Jimm had some thoughts about pushing this to Aurora/Beta -- he wanted to
> wait a bit to see if any regressions came up.  Jim, do you think we're good
> to uplift this?  Aurora _and_ beta?

Yes, go ahead and flag for aurora and beta. You'll need to fill out a short uplift form that the request will populate in bugzilla.

Comment 37

[David Parks [:handyman]]

Comment on attachment 8815043 [details] [diff] [review]
Send an eMouseEnterIntoWidget immediately after entering fullscreen (r=masayuki)

Approval Request Comment
[Feature/Bug causing the regression]:
N/A
[User impact if declined]:
Mouse cursor lingers after taking video into fullscreen (until further mouse movement).  This is annoying for media consumers.
[Is this code covered by automated tests?]:
No, the only testing was manual visual confirmation against 32 and 64-bit builds, in single and multiple-monitor configurations.
[Has the fix been verified in Nightly?]:
Yes
[Needs manual test from QE? If yes, steps to reproduce]: 
No, but STR is in comment 16.
[List of other uplifts needed for the feature/fix]:
N/A
[Is the change risky?]:
No.
[Why is the change risky/not risky?]:
The patch simply sends a synthetic mouse move when going to fullscreen.  (Synthetic) mouse moves are highly unlikely to inspire any significant surprises.
[String changes made/needed]:
None

Comment 38

[Julien Cristau [:jcristau]]

Comment on attachment 8815043 [details] [diff] [review]
Send an eMouseEnterIntoWidget immediately after entering fullscreen (r=masayuki)

hide mouse for fullscreen video on windows, take in aurora52

Comment 39

[Gerry Chang [:gchang]]

Comment on attachment 8815043 [details] [diff] [review]
Send an eMouseEnterIntoWidget immediately after entering fullscreen (r=masayuki)

Fix a regression for mouse cursor lingers after taking video into fullscreen. Beta51+. Should be in 51 beta 8.

Comment 40

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/ee5a46acf5ce

Comment 41

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-beta/rev/4333d321b0de

Comment 42

[Camelia Badau [:cbadau], Desktop Test Engineering]

I've reproduced the initial issue on Windows 7 x64 using Nightly 51.0a1 (2016-08-23). 

Verified fixed on Windows 7 x64 using latest Nightly 53.0a1 (buildID: 20161227030213), latest Aurora 52.0a2 (buildID: 20161227004018) and Firefox 51 Beta 10 (buildID: 20161222080852): finger point mouse disappears.

Comment 43

[Stefano Mariani]

This seems NOT to be fixed when watching an Amazon Prime Video. It happens again. Usually the first video you see it's ok, but then from the second the cursor is still there, fullscreen or not, and there is no way to get rid of it.

FF 54

Comment 44

[(No longer employed by Mozilla) Aaron Klotz]

(In reply to Stefano Mariani from comment #43)
> This seems NOT to be fixed when watching an Amazon Prime Video. It happens
> again. Usually the first video you see it's ok, but then from the second the
> cursor is still there, fullscreen or not, and there is no way to get rid of
> it.
> 
> FF 54

Please file a new bug.

Comment 45

[m]

This is still regressing on OS X. I'm using Catalina 10.15.5.

Comment 46

[Jonathan Kew [:jfkthame]]

(In reply to m from comment #45)

This is still regressing on OS X. I'm using Catalina 10.15.5.

This bug was specifically about Windows, and was closed several years ago. If you're currently seeing a problem on macOS, please file a new bug with the details. Thanks.

Comment 47

[Ravi Abram]

Issue seems to have reappeared. On 82.0.