1727815 - [Wayland] Firefox becomes unresponsive if the wayland compositor refuses a fullscreen request

Status: UNCONFIRMED

(Core :: Widget: Gtk, defect, P3)

Description

[alexander]

Attachment: sway_refuse_fullscreen.patch

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Firefox/91.0

Steps to reproduce:

Run firefox under a wayland compositor that does not make firefox fullscreen in response to xdg-shell fullscreen requests. Compositors with this behaviour include:

• Vivarium (https://github.com/inclement/vivarium) as of commit d7074de0bfbd2b247cd981f3dfbb5e639152ae00 (master at time of writing)
• Sway 1.6.1 (https://github.com/swaywm/sway/releases/tag/1.6.1) with the attached patch "sway_refuse_fullscreen.patch" applied

Try to make firefox fullscreen by clicking "View->Full Screen" or pressing F11.

I reproduced this under firefox 91.0.1 and nightly, under both sway and vivarium as above.

Actual results:

Firefox does not transition to fullscreen and the firefox UI components become unresponsive, but it remains possible to interact with the currently-visible webpage. Firefox's non-visible state is still changing though, e.g. clicking the new tab button has no visual effect but the firefox window title changes to indicate the new tab page is showing.

Some actions cause the gui to refresh, showing e.g. popup menus that have been clicked on but did not previously appear.

Expected results:

Firefox should have gone into fullscreen view mode.

Comment 1

[alexander]

I think I understand what is wrong - GTK makes an xdg-shell request to go fullscreen, which is denied, but then firefox never receives/acts-on any event that would trigger completion of the fullscreen transition. It remains stuck in what is supposed to be a transitional state.

I believe the correct behaviour would be for firefox to wait for a fullscreen event, then transition anyway if it doesn't receive one within a short timeout. This is because the xdg-shell fullscreen state is places very specific requirements on the compositor, and in practice the firefox window may be fullscreen in a reasonable sense even if the xdg-shell requirements are not met. Specifically, xdg-shell says:

If the surface doesn't cover the whole output, the compositor will position the surface in the center of the output and compensate with with border fill covering the rest of the output. The content of the border fill is undefined, but should be assumed to be in some way that attempts to blend into the surrounding area (e.g. solid black).

If the fullscreened surface is not opaque, the compositor must make sure that other screen content not part of the same surface tree (made up of subsurfaces, popups or similarly coupled surfaces) are not visible below the fullscreened surface.

Wayland compositors may make the window fullscreen in some reasonable sense (e.g. filling a workspace even though there are multiple workspaces displayed on an output) even if they don't meet the xdg-shell requirements, so firefox should still go fullscreen internally to support this.

I have a code fix to propose that would fix the issue in this way.

Comment 2

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

The Bugbug bot thinks this bug should belong to the 'Core::Widget: Gtk' component, and is moving the bug to that component. Please revert this change in case you think the bot is wrong.

Comment 3

[alexander]

Attachment: Bug 1727815 Toggle internal fullscreen even if window widget fullscreen not achieved r=emilio,stransky

Comment 4

[Martin Stránský [:stransky] (ni? me)]

This may also help with Bug 1636168

Comment 5

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

There's a r+ patch which didn't land and no activity in this bug for 2 weeks.
:alexander, could you have a look please?
For more information, please visit auto_nag documentation.

Comment 6

[alexander]

There's a r+ patch which didn't land and no activity in this bug for 2 weeks.
:alexander, could you have a look please?

I'm not sure what I'm supposed to do now to progress this, so I have asked on the phabricator review.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/da5a7c4478d9
Toggle internal fullscreen even if window widget fullscreen not achieved r=stransky

Comment 8

[Cristina Cozmuta (:CrissCozmuta)]

Backed out for causing assertion failures on nsGlobalWindowOuter.cpp. CLOSED TREE
Backout link : https://hg.mozilla.org/integration/autoland/rev/8ca704c6073987818c631194e39acfb5489294ca
Push with failures : https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=pending%2Crunning%2Csuccess%2Ctestfailed%2Cbusted%2Cexception%2Crunnable&searchStr=linux%2C18.04%2Cx64%2Cwebrender%2Cdebug%2Cmochitests%2Cwith%2Csoftware%2Cwebrender%2Cenabled%2Ctest-linux1804-64-qr%2Fdebug-mochitest-browser-chrome-swr-e10s%2Cbc5&revision=da5a7c4478d92ad2b50afadaee8ecb5fc1c54f42
Multiple failures :
https://treeherder.mozilla.org/logviewer?job_id=354467973&repo=autoland&lineNumber=26951
https://treeherder.mozilla.org/logviewer?job_id=354485318&repo=autoland&lineNumber=144193
https://treeherder.mozilla.org/logviewer?job_id=354477855&repo=autoland&lineNumber=27231
and build bustage :
https://treeherder.mozilla.org/logviewer?job_id=354469808&repo=autoland&lineNumber=51658

Comment 9

[alexander]

The assertion that has failed is now out of date and should probably be removed, I will confirm and update the patch. I'll also work out how to test more thoroughly with DEBUG enabled, I didn't realise I didn't have that already (which would have hit the assert and caught the issue sooner).

Comment 10

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

The bug assignee didn't login in Bugzilla in the last 7 months, so the assignee is being reset.

Comment 11

[Martin Stránský [:stransky] (ni? me)]

Alexander, do you still see the bug?
Thanks.

Comment 12

[alexander]

I'm sorry, I don't have the development setup to easily test this any more. I will try to revisit it but I can't honestly guarantee I will get to it.