Input device settings not applied on hotplug/reconnect/resume in Xorg sessions

https://answers.launchpad.net/ubuntu/+question/693359

I did a bit more digging (based on Bernard Stafford's suggestion that USB keyboard unplug/reconnect) is causing the reconnected keyboard to go back to the default keyboard config. I was able to get my keyboard customizations to apply to the reconnected USB keyboard by explicitly adding them to /etc/default/keyboard:XKBOPTIONS.

When they are set only by Tweak tool (and not in /etc/default/keyboard:XKBOPTIONS), the options only seem to be defined in dconf in "org.gnome.desktop.input-sources xkb-options ['shift:both_capslock', 'ctrl:swap_lalt_lctl_lwin']". These settings seem to stop working when a USB keyboard is reconnected. This is the behavior which seems to have changed in the past month or so. (Thanks to Bernard for suggesting I even look into default keyboard configuration. This should help some.)

In addition to the keyboard settings not getting applied to the reconnected USB keyboard, it also appears that the mouse tracking speed setting (established in Gnome settings) is also lost on USB reconnect. This is also new behavior (almost certainly related to the reconnect issue I'm seeing with the USB keyboard).

I don't think it's very ergonomic to need to restart my Gnome session just to get my expected keyboard and mouse settings established when i reconnect my input devices back into my laptop.

I set gnome-shell as the affected package for now, since a theory would be that the changed behavior is related to some upgrade of that package.

@Weston: To find out, can you please downgrade gnome-shell temporarily:

sudo apt install gnome-shell=3.36.1-5ubuntu1 gnome-shell-common=3.36.1-5ubuntu1

Then reboot and let us know if the issue is still present.

Maybe related:

https://gitlab.gnome.org/GNOME/mutter/-/issues/1466

had the same problem with my apple keyboard settings, downgrading solved it. thanks.

Thanks for testing. That helps us narrow down the cause of the issue.

@Gunnar I downgraded to 3.36.1-5ubuntu1 (for gnome-shell and gnome-shell-common). Downgrading to 3.36.1-5ubuntu1 did not fix the issue for me (like it fixed things for Detlef).

Even with 3.36.1-5ubuntu1, I still lose my customized keyboard tweaks (stored in `org.gnome.desktop.input-sources -> xkb-options`). It always defaults back to whatever is in `/etc/default/keyboard -> XKBOPTIONS`. I also lose my mouse speed setting.

With 3.36.1-5ubuntu1, after doing Alt-F2 -> "r", my desired settings are correctly applied.

Are there any other tests I should try since my observations are different from Detlef?

@Detlef: Do you ever experience the loss of the mouse speed setting (with either version of gnome-shell)?

@Weston: Just to be sure: Did you reboot after having downgraded?

@Gunnar. Yes, I rebooted after downgrade. After rebooting, I confirmed (`dpkg -l` and `gnome-shell --version`) that I was using the downgraded version.

I'm wondering if I should be going back to versions which are even older? Is it safe to downgrade to even earlier versions of gnome-shell and gnome-shell-common without messing up other software on my system?

Is there a list of even older downgrade versions you recommend I should try?

Thanks!

Thanks for clarifying.

On 2020-10-12 19:58, Weston Hunter wrote:
> I'm wondering if I should be going back to versions which are even
> older?

That wouldn't make much sense considering that you saw the changed behavior only the past few weeks.

Maybe it's not really a gnome-shell issue. The upstream similar bug, which Daniel linked to, was moved from gnome-shell to mutter.

So if you want to make a similar attempt with mutter, you would do:

sudo apt install mutter=3.36.1-3ubuntu3 mutter-common=3.36.1-3ubuntu3 libmutter-6-0=3.36.1-3ubuntu3 gir1.2-mutter-6=3.36.1-3ubuntu3

@Gunnar: I only first recognized the issue last week -- it's possible I had been affected by the issue for longer, but did not realize it. After reading the description in the linked mutter bug, it does sound like I'm very likely affected by the same bug.

Tonight, I'll try the downgrade for the mutter packages as well.

Thanks for working through this with me!

@Gunnar: Confirmed! The mutter downgrade fixes it: `sudo apt install mutter=3.36.1-3ubuntu3 mutter-common=3.36.1-3ubuntu3 libmutter-6-0=3.36.1-3ubuntu3 gir1.2-mutter-6=3.36.1-3ubuntu3`.

Thanks so much!

This looks like the bug was introduced in mutter between 3.36.1-3ubuntu3 and 3.36.6-1ubuntu0.20.04.2.

I posted this observation in the linked mutter bug as well (in case it might help them narrow things).

Thanks Weston, that was useful info.

Now let's await input from some desktop engineer. I'm just a volunteer who tries to help. ;)

@Gunnar: as a "volunteer who tries to help", you have been immensely helpful!!

I think this may be related to https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1371 I will look into this

Status changed to 'Confirmed' because the bug affects multiple users.

This bug not only happens when I unplug my Apple magic keyboard, but also after I unlock the screen when it has been locked for a short period. Imagine having to restart the Gnome session everytime you unlock the screen so the customized settings are applied. This a new behaviour related to the latest updates.

@mhalsharif: Interesting observation! Just for the record, although I experience the original issue, I don't think I'm losing anything by doing lock/unlock. I frequently lock/unlock my desktop and the lock duration can be anywhere from a few seconds to many hours. I think I would notice pretty quickly if I had this problem.

Could you clarify if you lose the customized settings only in the lock-screen (where you type your password) or are your settings lost even after you successfully get back into your own desktop?

@westonhunter: I use gnome tweasks to swap the left Win with left Ctrl on my Apple keyboard. To clarify, that setting is lost even after I type in my password and get back into the desktop. As a temporary solution, I set "blank screen delay" to never and disabled "automatic screen lock", so I don't have to reset the gnome session everytime my screen goes blank.

Do you use Apple Keyboard? I am not sure if this behaviour have something to do with that. I should also mention that I use Apple Magic Mouse II and noticed that "Natural Scrolling" does not work when the mouse is reconnected or the system is restarted. So I would have to reset the gnome session for the "Natural Scrolling" to work. All these settings were working as expected before I installed the ubuntu update earlier this week.

@mhalsharif: I'm currently not using any Apple hardware, but I'm seeing all the same behavior you are (except I don't lose settings on lock/unlock).

I have currently done the following things:
* I moved all my keyboard customizations from dconf settings [/org/gnome/desktop/input-sources/xkb-options] (which is where Tweak Tool puts the config) and put them in /etc/default/keyboard:XKBOPTIONS. Since these settings are default, when the keyboard resets to these defaults, it resets back to the config I want. (This fixes the keyboard reset, but still will not fix the mouse resetting issue.) If you want to try this approach, you can just add the default configuration options like this:
XKBOPTIONS="shift:both_capslock,ctrl:swap_lalt_lctl_lwin,caps:escape"

* I have also downgraded mutter (`sudo apt install mutter=3.36.1-3ubuntu3 mutter-common=3.36.1-3ubuntu3 libmutter-6-0=3.36.1-3ubuntu3 gir1.2-mutter-6=3.36.1-3ubuntu3`). This downgraded version fixes all the resetting issues (although I'm stuck on the downgraded mutter version until a fix gets released).

Hi,

I would like to report what appears to be a similar issue. I have a Logitech bluetooth keyboard. I used gnome-tweak-tools to set CapsLock to change layout. After reboot, the key binding didn't work and I had to open gnome-tweaks-tools again and just check and then uncheck some box. Downgrading gnome-shell didn't work and in fact removed also gnome-tweaks. Downgrading mutter, and then installing gnome-tweask again solved the issue.

This might turn out to be the same issue as bug 1899509.

Hi, I've upgraded my machine to 20.10. This has updated my mutter version to the buggy one. Now my keyboard layout configurations are cancelled randomly not only after boot/sleep. Even worse, I cannot install the 3.36.1 anymore.

This seems similar to bug 1902001.

Possibly fixed in mutter 3.38.2 by:

https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1553

Actually the upstream bug link says it's definitely fixed in 3.38.2 :)

Can this be fixed as some kind of "extras" in connection with the update to 3.36.7? I have seen several questions at Ask Ubuntu which seem related to this bug.

Maybe someone needs to test the fix first to make sure it really is relevant to this bug :)

https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1553.patch

Ok, the assumed fix is available for testing in this PPA:

https://launchpad.net/~gunnarhj/+archive/ubuntu/mutter

Would be great if some users who are affected by this bug could install the mutter packages from the PPA and let us know if it helps.

I might be experiencing this issue too, but I am unsure. I am posting this in case it helps.

After upgrading from Ubuntu 18.04 to 20.04, whenever I remove my external keyboard and plug it back in, the external keyboard's layout switches to US English instead of whatever it was set to (a custom layout in my case, but I confirmed it with other, built-in layouts too). The laptop's built-in keyboard layout remains correct.

I applied @gunnarhj's fix, but the behavior did not change. (Again, I am not sure I if the problem I'm experiencing has the same root cause as the OP's problem.)

However, I was able to work around the problem by switching to Wayland.

To be sure you are really testing the fix please run:

  dpkg -l | grep libmutter-6-0

and tell us what it displays.

This is my output:

$ dpkg -l | grep libmutter-6-0
ii libmutter-6-0:amd64 3.36.6-1ubuntu0.20.04.3~ppa2 amd64 window manager library from the Mutter window manager

I've added PPA gunnarhj/mutter and updated
porozhnetalv@porozhnetalv-m50sv:~$ dpkg -l | grep libmutter-6-0
ii libmutter-6-0:amd64 3.36.6-1ubuntu0.20.04.2 amd64 window manager library from the Mutter window manager

This fix is definitely relevant to the bug however it have not really fixed it.

I'm using "Additional Layout Options" to enable "Switching to another layout" with Caps Lock.
With this update on USB keyboard unplug/reconnect laptop embedded keyboard's CapsLock changes layout and external keyboard's CapsLock works as real Caps.
Even more to that: after external keyboard's Caps pressing first press of internal keyboard's Caps is ignored.

Sorry, i obviously don't understand something (or anything) about updates.
Now it's fixed

porozhnetalv@porozhnetalv-m50sv:~$ dpkg -l | grep libmutter-6-0
ii libmutter-6-0:amd64 3.36.6-1ubuntu0.20.04.3~ppa2 amd64 window manager library from the Mutter window manager

Cannot confirm behavior described at https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/1899206/comments/30

That sounds promising, Leonid. The submitter of <https://askubuntu.com/q/1291175> also confirmed that the PPA packages fix a similar issue.

Maybe the fix we are testing only addresses issues with xkb-options. scheleaap's case seems to be about using some other keyboard layout but the system one. While it's indeed a similar case, it's not identical and the mechanisms behind the scenes differ.

After booting my computer this morning I tested again (using X) and the problem seems to be gone. When I tested the patch yesterday, I logged out and back in again. Could it be that that was not enough to apply the patch, and a reboot was necessary?

On 2020-11-12 09:44, scheleaap wrote:
> After booting my computer this morning I tested again (using X) and
> the problem seems to be gone.

Great news!

> When I tested the patch yesterday, I logged out and back in again.
> Could it be that that was not enough to apply the patch, and a reboot
> was necessary?

Could be. Myself always reboot before testing changes to such core components as mutter is.

Yet another Ask Ubuntu confirmation: https://askubuntu.com/a/1291705/159370

I committed the patch to the ubuntu/focal branch.

Thank you for investigating and solving the problem, everyone!

Could the fix be included in 20.04 too?

Any rough idea on if / when we will see this in 20.04?

I can also confirm the work around via @Gunnar's PPA also fixes this problem for me: https://askubuntu.com/questions/1294774/swapping-control-and-caps-lock-on-dell-xps-on-20-04-used-to-work-with-gnome-tw . Thanks, people!

This bug was fixed in the package mutter - 3.38.2-1ubuntu1

---------------
mutter (3.38.2-1ubuntu1) hirsute; urgency=medium

  * Merge with debian, containing new upstream release (LP: #1908158):
    - Apply input device settings on hotplug / reconnect and resume
      (LP: #1899206)
  * d/p/x11-Add-support-for-fractional-scaling-using-Randr.patch:
    - Refreshed as per upstream changes

mutter (3.38.2-1) unstable; urgency=medium

  * Team upload
  * New upstream release
    - Don't crash during screencasting if cursor is not available
    - Don't crash if a kernel bug results in appearing to have multiple
      builtin panels
    - Log warnings instead of crashing if user-defined keymap is
      misconfigured
    - Don't forget we have a touchscreen if a non-touchscreen input device
      is connected
    - Log warnings instead of crashing if invalid barriers are created
    - Otherwise functionally equivalent to 3.38.1-4
  * Drop patches cherry-picked from upstream or applied upstream
  * d/gbp.conf: Use upstream 3.38.x branch.
    Version 40~alpha was already released, so it's misleading to say that
    3.38.x is the latest.
  * Standards-Version: 4.5.1 (no changes required)
  * d/rules: Don't run tests if built with nocheck option
  * Upload to unstable

mutter (3.38.1-4) experimental; urgency=medium

  * Team upload
  * Update to upstream gnome-3-38 branch, commit 3.38.1-50-gc70610c43
    - If cursor theme is missing, draw a grey semi-transparent square
      instead of crashing
    - Allow primary GPU to be overridden with a udev rule
    - Fix fullscreen toggle for some applications (Closes: #975453)
    - Only schedule frame drawing for each surface (window) according to
      one stage view (monitor)
    - Fix size hints for client-side-decorated windows
    - Avoid adding input devices too soon
    - Performance optimizations
    - Do not disable the X Security extension if Xwayland was built with it
  * Don't depend on transitional libgdk-pixbuf2.0-dev
  * Explicitly build-depend on gdk-pixbuf.
    Previously it was pulled in by some other dependency (presumably
    gnome-desktop3 and gtk), but mutter's build system checks for it, so
    we should build-depend on it here too.
  * d/p/clutter-frame-clock-Schedule-a-frame-at-least-once-per-se.patch:
    Add proposed patch to ensure frame clock does not stop (Closes: #974172)

mutter (3.38.1-3) experimental; urgency=medium

  * Team upload
  * Update to upstream gnome-3-38 branch, commit 3.38.1-33-g067af969c
    - Prevent a use-after-free on (virtual) device removal
    - Prevent crash on startup caused by dispatching libinput events too
      early
    - If Wayland subsurfaces have circular relationships, disconnect the
      client instead of crashing
    - When resizing terminals interactively, don't offset the position
    - Fix broken timestamp behaviour if X11 timestamps overflow
    - Fix a memory leak
    - Ensure clock is updated when going from fullscreen app to overview
    - Fix artifacts around GNOME Shell magnifier cursor
    - Fix monitor tiling on X11
    - Configure input devices correctly in X11
      (LP: #1899206, LP: #1899509)
    - Avoid spurious warnings in lo...

Are there plans to backport this fix to 20.04 and 20.10? Or do we have to put up with broken mice until 21.04?

On 2020-12-19 18:48, Matthew Wilson wrote:
> Are there plans to backport this fix to 20.04 and 20.10?

Yes. The fix has been uploaded to the focal and groovy queues for stable release updates.

very irritating bug, relates to bluetooth keyboards too.

Hello Weston, or anyone else affected,

Accepted mutter into groovy-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/mutter/3.38.2-1ubuntu1~20.10.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-groovy to verification-done-groovy. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-groovy. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

SRU header is missing, please fix

Hello Weston, or anyone else affected,

Accepted mutter into focal-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/mutter/3.36.7+git20201123-0.20.04.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-focal to verification-done-focal. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-focal. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

Tested with my bluetooth mouse and it seems to work correctly now. Thanks!

On 2020-12-24 19:35, Matthew Wilson wrote:
> Tested with my bluetooth mouse and it seems to work correctly now.

Did you test on 20.04 or 20.10?

Sorry, yes I should have said. I tested on both 20.04 and 20.10.

Excellent, thanks! Then let's consider both the uploads verified.

Tested with USB keyboard on 20.10.
Layout settings are now correctly retained after reconnecting keyboard.
Thank you!
(I had reported bug #1902001 marked as a duplicate of this bug.)

Will this be available in 20.04? Or is the recommendation for everyone to upgrade mutter manually?

@Mark, my understanding from the previous messages is that it's in "proposed" at present. Once it's gone through the processes explained above, it is released to 20.04 main. I.e. we all get the update normally.

"N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days."
- https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/1899206/comments/50

This bug was fixed in the package mutter - 3.36.7+git20201123-0.20.04.1

---------------
mutter (3.36.7+git20201123-0.20.04.1) focal; urgency=medium

  * New upstream release and git snapshot (LP: #1903230)
    - Fix Night Light updates after DPMS
    - Fix IM handling on X11 (LP: #1899206, thanks to Gunnar Hjalmarsson for
      the initial cherry-pick)
    - Fix resizing of attached modal dialogs on wayland
    - Fix jumps when resizing windows using discrete steps
    - Fixed crashes
    - Plugged Memory leaks
    - Misc. bug fixes and cleanups
  * debian/libmutter-6-0.symbols: Update to match upstream changes.
    No symbol is a public one

 -- Marco Trevisan (Treviño) <email address hidden> Tue, 15 Dec 2020 04:27:46 +0100

The verification of the Stable Release Update for mutter has completed successfully and the package is now being released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.

This bug was fixed in the package mutter - 3.38.2-1ubuntu1~20.10.1

---------------
mutter (3.38.2-1ubuntu1~20.10.1) groovy; urgency=medium

  * No-change backport to groovy
  * debian: Prepare gbp.conf and control references to groovy branching

mutter (3.38.2-1ubuntu1) hirsute; urgency=medium

  * Merge with debian, containing new upstream release (LP: #1908158):
    - Apply input device settings on hotplug / reconnect and resume
      (LP: #1899206)
  * d/p/x11-Add-support-for-fractional-scaling-using-Randr.patch:
    - Refreshed as per upstream changes

mutter (3.38.2-1) unstable; urgency=medium

  * Team upload
  * New upstream release
    - Don't crash during screencasting if cursor is not available
    - Don't crash if a kernel bug results in appearing to have multiple
      builtin panels
    - Log warnings instead of crashing if user-defined keymap is
      misconfigured
    - Don't forget we have a touchscreen if a non-touchscreen input device
      is connected
    - Log warnings instead of crashing if invalid barriers are created
    - Otherwise functionally equivalent to 3.38.1-4
  * Drop patches cherry-picked from upstream or applied upstream
  * d/gbp.conf: Use upstream 3.38.x branch.
    Version 40~alpha was already released, so it's misleading to say that
    3.38.x is the latest.
  * Standards-Version: 4.5.1 (no changes required)
  * d/rules: Don't run tests if built with nocheck option
  * Upload to unstable

mutter (3.38.1-4) experimental; urgency=medium

  * Team upload
  * Update to upstream gnome-3-38 branch, commit 3.38.1-50-gc70610c43
    - If cursor theme is missing, draw a grey semi-transparent square
      instead of crashing
    - Allow primary GPU to be overridden with a udev rule
    - Fix fullscreen toggle for some applications (Closes: #975453)
    - Only schedule frame drawing for each surface (window) according to
      one stage view (monitor)
    - Fix size hints for client-side-decorated windows
    - Avoid adding input devices too soon
    - Performance optimizations
    - Do not disable the X Security extension if Xwayland was built with it
  * Don't depend on transitional libgdk-pixbuf2.0-dev
  * Explicitly build-depend on gdk-pixbuf.
    Previously it was pulled in by some other dependency (presumably
    gnome-desktop3 and gtk), but mutter's build system checks for it, so
    we should build-depend on it here too.
  * d/p/clutter-frame-clock-Schedule-a-frame-at-least-once-per-se.patch:
    Add proposed patch to ensure frame clock does not stop (Closes: #974172)

mutter (3.38.1-3) experimental; urgency=medium

  * Team upload
  * Update to upstream gnome-3-38 branch, commit 3.38.1-33-g067af969c
    - Prevent a use-after-free on (virtual) device removal
    - Prevent crash on startup caused by dispatching libinput events too
      early
    - If Wayland subsurfaces have circular relationships, disconnect the
      client instead of crashing
    - When resizing terminals interactively, don't offset the position
    - Fix broken timestamp behaviour if X11 timestamps overflow
    - Fix a memory leak
    - Ensure clock is updated when going from fullscreen app to overview
    - Fix artifacts around GNOME S...