Screen rotated to landscape: mouse stays in portrait mode

[sven]

If I rotate the screen to landscape, the mouse (bluetooth) seems to respect the original dimensions, so that the right half of the screen cannot be reached with the mouse. (Current OTA-4, focal, GS5pro)

[kugiigi]

@sven Sadly this is a known issue and some people already tried to fix it. Not sure if it's already fixed but in Ubuntu 24.04, it's still there.

[sven]

I see.

Is there a workaround so that the orientation could be fixed after/during a boot? I am asking because I am using rotation lock (staying in landscape), so a one-time fix would be enough for me

[kugiigi]

@sven I'm not aware of any workaround. If you know QML, you can check the code.and maybe a simple rotation of the Cursor can do the trick or something.

[arubislander]

@kugiigi If there were such a 'simple' work-around, then there would already have been a PR for that, no?

[kugiigi]

@arubislander Well I'm just talking about a workaround like perhaps only make it work in a specific orientation which seems to be what the OP needs

[hilbert]

I keep this thread watched, because I am very interested in it!
And I add an information: it is not just a visualization issue, but the pointing is affected too. If I rotate the phone to the left, the mouse pointer works as the application bar was along the top edge (where on the screen the notification bar is shown).
The pointer probably could operate on the notification bar, if it could reach the right edge (but the movement stop at a middle point, probably corresponding to the device screen width)

Rotating it to the right, the application bar is found (by the mouse pointer) on the virtual bottom (which is much farther than the device screen, probably corresponding to the screen height), while the left edge allows to operate on the notification area.

[benjamin]

Howdy y'all from Florida.

I've got this same issue; but if you ask me it's an issue with the touch screen overlay not being configured at all for landscape.

Same exact thing happens on a real PC if you never set up rotation on a manually configured screen.

I literally just installed this morning on my N10 and immediately got my heart broken over this.

I'm going to poke around and see what I can find.

I have not read anything about root status on this OS; is the default user already admin? If so, then poking around might be worth it...

[ikoz]

@benjamin said in Screen rotated to landscape: mouse stays in portrait mode:

have not read anything about root status on this OS; is the default user already admin?

The default user is phablet and it is a member of the sudo group so yes, you can run commands that require root privileges. Keep in mind that the filesystem is read-only like on immutable distros.

[Kamil]

@sven yes.It happens too on my oneplus nord n100 on OTA-7. Maybe there is a command for fixing the mouse orientation?

[uxes]

for the record I would like to attach a transcript from today's UBports chat [ENGLISH]

Don Fredl, [26.02.25 11:34]
the only thing I can tell is that Mir and QtMir are not negotiating geometry changes based on orientation yet

Don Fredl, [26.02.25 11:35]
but I am unsure as to how that is supposed to happen (maybe @alan_griffiths knows something)

Don Fredl, [26.02.25 11:35]
since we want to avoid resolution changes

Don Fredl, [26.02.25 11:38]
I will give my Laptop running Mir 2.x a shot later and see if something has changed

Don Fredl, [26.02.25 11:38]
this thing can rotate

Alan, [26.02.25 11:44]
No more than you've stated. When Mir handles both input and compositing everything is consistent: Other Mir based compositors like Frame, Miriway and Miracle-WM work fine.

But QtMir takes over both input and compositing and routes everything through the Qt compositor. There things are inconsistent. I have not looked, but it sounds as though Mir isn't informed about the rotation and therefore constraints input to the the output it thinks is there.

Don Fredl, [26.02.25 11:44]
could you point me to the place where it applies those constraints?

Don Fredl, [26.02.25 11:45]
in Mir that is

Alan, [26.02.25 11:59]
Not somethin I know well, but SeatInputDeviceTracker::confine* looks likely

Don Fredl, [26.02.25 12:00]
Thanks, will take a look when I can

TheKit, [26.02.25 16:19]
that sounds about right to me. I've checked that previously and another issue is that due to the mentioned above XWayland isn't getting screen size updated when the display is rotated

TheKit, [26.02.25 16:20]
and X11 calculates scaled relative input coordinates based on the screen size, so the touches in rotated mode end up out of range

TheKit, [26.02.25 16:26]
what we need to figure out is the way to handle screen rotation so Lomiri can still do the animation when it's changed, but let the Mir server be aware of the updated geometry

Alan, [26.02.25 18:50]
It shouldn't be hard (unless input is handled during rotation - and that is madness). Mir has an internal notification system for display changes and that just needs to be fed the new layout.

TheKit, [26.02.25 19:16]
One thing that makes me wonder. Which component handles the actual content rotation? Is it done in software in QML renderer?

[kugiigi]

Makes me wonder what changed because as far as I know, this wasn't an issue before. Could it be related to the big changes with multi-monitor support in Lomiri? I believe that's also when the scrolling issue started to happen on external displays