GNOME Bugzilla – Bug 775869
"Logout" shortcut causes Power Off dialog to open twice and lose keyboard accessibility
Last modified: 2021-06-14 18:20:52 UTC
Pressing Control-Alt-Delete (default gnome shortcut) launches the shutdown dialog and focuses the Shutdown entry. That's convenient, since I have to only press the enter key to initiate shutdown. But 2 seconds later, another identical dialog pops up *under* it, which draws the focus away from the top dialog, and requires me to use the mouse. Furthermore, I have to click the cancel button twice in order for the dialog to go away. This happens with the following applications: * Chrome * Firefox * Dropbox * VLC (but also closes it) It does NOT happen with most applications, including these: * gnome-control-center * Nautilus * Rhythmbox * LibreOffice Writer, Impress, Calc * gedit How reproducible: Every time Steps to Reproduce: 1. Open Firefox or Chrome or one of the other programs mentioned above 2. Press Control-Alt-Delete 3. Wait 2 seconds 4. Observe a brief flicker and that the focus goes away 5. Click "Cancel" Actual results: Two shutdown dialogs on top of each other are shown. The brief flicker in between the dialogs is the expanded dialog with the message: "Some applications are busy or have unsaved work", this can only be seen by stopping the screen recording at the right time. Expected results: Only one dialog is shown, focus is on the Shutdown button. Additional info: Note that this happens only with the default GNOME keyboard binding. Only one dialog is shown when: - I run `gnome-session-quit --power-off` a terminal or from the Alt-F2 prompt - From the gnome menu shutdown button A workaround is to unbind the Ctrl-Alt-Delete shortcut, create a custom shortcut that runs `gnome-session-quit --power-off` and bind Ctrl-Alt-Delete to that.
Created attachment 341667 [details] Screen recording of the behaviour after pressing control-alt-delete
This is presumably related to Joanmarie's accessibility problem noted here: https://bugzilla.gnome.org/show_bug.cgi?id=728151#c3 I'm adding the accessibility keyword because, like that, this report sounds like it could seriously hinder users for whom mouse is not a option.
This was tracked down to gnome-session and supposedly fixed in 3.22. If it still happens for you under some circumstances, it's still highly unlikely that it's now a gnome-shell issue. *** This bug has been marked as a duplicate of bug 728151 ***
(In reply to Florian Müllner from comment #3) > This was tracked down to gnome-session and supposedly fixed in 3.22. If it > still happens for you under some circumstances, it's still highly unlikely > that it's now a gnome-shell issue. Yes, it still happens to me under GNOME Version 3.22.2. It happens under clearly defined circumstances (described above), under a fresh install of Fedora 25 in a VM, 100 % of the time. Well, if it's not a gnome-shell issue, then what is it? The keyboard handling is in gnome-shell, the shutdown code is also in gnome-shell, isn't it? Anyway, feel free to reassign the component if you are sure that the bug is not in gnome-shell. > *** This bug has been marked as a duplicate of bug 728151 *** Not quite sure why, though. On the other bug, you commented: > No, that bug [#775869] is not about the short delay mentioned in the original report, it's about the dialog appearing twice which the patch on this bug was supposed to fix. Doesn't that mean that this (#775869) bug is separate from the other one (#728151), according to you? Sorry, I'm a bit confused right now.
(In reply to Alexander Korsunsky from comment #4) > Well, if it's not a gnome-shell issue, then what is it? The keyboard > handling is in gnome-shell, the shutdown code is also in gnome-shell, isn't > it? It's complicated: gnome-settings-daemon asks gnome-shell to grab the key combination gnome-shell signals to gnome-settings-daemon when the key combination is activated gnome-settings-daemon calls gnome-session's Shutdown method gnome-session asks gnome-shell to display the end session dialog gnome-shell reports the user's response back to gnome-session So gnome-shell is clearly involved here, but not the only possible source of bugs - unless gnome-shell signals to gnome-settings-daemon that the shortcut was activated a 2nd time after the shortcut was actually activated (unlikely in response to the end-session dialog plus some delay), or gnome-shell shows the end session dialog without having been asked to do so, the bug is elsewhere. In bug 728151 it turned out that gnome-session was actually asking for a second dialog after the first one was dismissed. It's much more likely that this bug wasn't entirely fixed than having a different bug in gnome-shell with identical symptoms, so I reopened the old bug. It's still a guess though, so if you run "dbus-monitor path=/org/gnome/SessionManager/EndSessionDialog" when reproducing the issue and only see a single call to "Open", then the guess was wrong and this bug should be reopened instead. > > *** This bug has been marked as a duplicate of bug 728151 *** > > Not quite sure why, though. On the other bug, you commented: > > > No, that bug [#775869] is not about the short delay mentioned in the original report, it's about the dialog appearing twice which the patch on this bug was supposed to fix. > > Doesn't that mean that this (#775869) bug is separate from the other one > (#728151), according to you? Bug 728151 originally reported three issues (a keynav issue, a delay between activating the shortcut and seeing the dialog, and the dialog appearing twice). The bug that was supposedly fixed was the last issue, and Michael suggested to file separate bugs for the other ones in case people can still reproduce them. But this bug is about the dialog being shown twice, which is precisely the one issue that bug 728151 ended up being about ...
(In reply to Florian Müllner from comment #5) > It's complicated: > > ... > > So gnome-shell is clearly involved here, but not the only possible source of > bugs - unless gnome-shell signals to gnome-settings-daemon that the shortcut > was activated a 2nd time after the shortcut was actually activated (unlikely > in response to the end-session dialog plus some delay), or gnome-shell shows > the end session dialog without having been asked to do so, the bug is > elsewhere. That is indeed complicated. Mind you that I am a mere end-user, have no idea about GNOME project structure, and was just trying to help. Bugzilla asked a component from me, so I guessed and put in a component that seemed sensible. That's why I was kind of hoping that you, having more insight, would reassign the bug to the component that's more appropriate. > It's still a guess > though, so if you run "dbus-monitor > path=/org/gnome/SessionManager/EndSessionDialog" when reproducing the issue > and only see a single call to "Open" Here is my output: dbus-monitor path=/org/gnome/SessionManager/EndSessionDialog signal time=1481305857.619364 sender=org.freedesktop.DBus -> destination=:1.94 serial=2 path=/org/freedesktop/DBus; interface=org.freedesktop.DBus; member=NameAcquired string ":1.94" signal time=1481305857.619403 sender=org.freedesktop.DBus -> destination=:1.94 serial=4 path=/org/freedesktop/DBus; interface=org.freedesktop.DBus; member=NameLost string ":1.94" method call time=1481305859.946122 sender=:1.9 -> destination=:1.14 serial=231 path=/org/gnome/SessionManager/EndSessionDialog; interface=org.gnome.SessionManager.EndSessionDialog; member=Open uint32 1 uint32 0 uint32 60 array [ ] method call time=1481305861.560728 sender=:1.9 -> destination=:1.14 serial=235 path=/org/gnome/SessionManager/EndSessionDialog; interface=org.gnome.SessionManager.EndSessionDialog; member=Open uint32 1 uint32 0 uint32 60 array [ object path "/org/gnome/SessionManager/Inhibitor7" ] So no, it's not a single one, it's two. Not sure what exactly that means, but I hope it helps in debugging. > The bug that was supposedly fixed was the last issue, and Michael > suggested to file separate bugs for the other ones in case people can still > reproduce them. Which is exactly what this bug was supposed to be. I can still reproduce it, so I filed a new, separate one, so I don't quite get why you closed it as duplicate, when people were specifically instructed to create a new bug if the issue persists. Or to quote Michael from bug #728151, comment #32 (In reply to Michael Catanzaro from comment #32) > I see that there is still a bug, or multiple remaining bugs, but it's > definitely a different issue so please create a new bug report.
(In reply to Alexander Korsunsky from comment #6) > So no, it's not a single one, it's two. Not sure what exactly that means, > but I hope it helps in debugging. It means that gnome-shell is showing two dialogs because gnome-session requests two dialogs. Or in other words: Showing the dialog when asked to is not a bug, gnome-session requesting a dialog when it shouldn't is.
Has anyone from the gnome-session had a chance to ponder this? Whereas I couldn't replicate the accessibility problem that Joanmarie mentioned before - after the subsequent attempt to fix this, I now can confirm it: There are still 2 dialogs, and the 2nd is not keyboard accessible. https://bugzilla.gnome.org/show_bug.cgi?id=728151#c3
And at the risk of making this even more of a dupe of the closed one, I'm attempting to make the title more inclusive of the various issues here.
I can confirm this on GNOME 3.22.2 on Fedora 25.
GNOME is going to shut down bugzilla.gnome.org in favor of gitlab.gnome.org. As part of that, we are mass-closing older open tickets in bugzilla.gnome.org which have not seen updates for a longer time (resources are unfortunately quite limited so not every ticket can get handled). If you can still reproduce the situation described in this ticket in a recent and supported software version of gnome-session, then please follow https://wiki.gnome.org/GettingInTouch/BugReportingGuidelines and create a new ticket at https://gitlab.gnome.org/GNOME/gnome-session/-/issues/ Thank you for your understanding and your help.