Bug 1622000

*** Bug 1594392 has been marked as a duplicate of this bug. ***

I did not manage to reproduce it on RHEL 7.5 even with affected version of mutter - 3.26.2-13.el7.x86_64.

After setting some screen resolution in /etc/X11/xorg.conf and restarting GUI, the resolution is being set, so xorg.conf is not overriden by mutter.

Created attachment 1480856 [details]
file /etc/X11/xorg.conf

xorg.conf generated with
# Xorg :3 -configure
and added line 'Modes "1024x768"' to Screen section
I did not manage to reproduce the bug with that xorg.conf.

I don't believe this is the proper test for this issue.  From my experience the issue is with display layout not display resolution.

A simple test would be create a 2x1 display output then see if mutter reconfigures this to be a 1xN (1x2 in this case).

(In reply to Derek Schrock from comment #7)
> From my experience the issue is with display layout not display resolution.

That's correct, thanks for the hint.

Ok, thanks for the hint. I have set some monitor layout in xorg.conf. It works and it is not overriden by mutter-3.26.2-16 if there is no file ~/.config/monitors.xml (from gnome-control-center).

But if there exists file ~/.config/monitors.xml, layout in the xorg.conf is present only in gdm and in desktop is overriden by that in monitors.xml.

Is it correct behaviour? Should ~/.config/monitors.xml have priority before /etc/X11/xorg.conf regarding monitors layout?

(In reply to Tomas Hudziec from comment #9)
> Is it correct behaviour? Should ~/.config/monitors.xml have priority before
> /etc/X11/xorg.conf regarding monitors layout?

Yes, if the user uses the display settings to change the layout, that should have precedence over the default layout.

(In reply to Florian Müllner from comment #10)
> Yes, if the user uses the display settings to change the layout, that should
> have precedence over the default layout.

Ok, then it works correctly with mutter-3.26.2-16.

But I had no luck to reproduce the overriding layout in xorg.conf by mutter-3.26.2-13.
I have following sections in my xorg.conf:
Section "Monitor"
    Identifier  "DP-2"
EndSection

Section "Monitor"
    Identifier  "DP-3"
    Option      "Primary" "true"
    Option      "LeftOf" "DP-2"
EndSection

How should I reproduce the bug? Do I have to use some Nvidia proprietary module mentioned in the original bug 1585230?

(In reply to Tomas Hudziec from comment #11)
> (In reply to Florian Müllner from comment #10)
> > Yes, if the user uses the display settings to change the layout, that should
> > have precedence over the default layout.
> 
> Ok, then it works correctly with mutter-3.26.2-16.
> 
> But I had no luck to reproduce the overriding layout in xorg.conf by
> mutter-3.26.2-13.
> I have following sections in my xorg.conf:
> Section "Monitor"
>     Identifier  "DP-2"
> EndSection
> 
> Section "Monitor"
>     Identifier  "DP-3"
>     Option      "Primary" "true"
>     Option      "LeftOf" "DP-2"
> EndSection
> 
> How should I reproduce the bug? Do I have to use some Nvidia proprietary
> module mentioned in the original bug 1585230?

Tomas if I may. We might be able to verify as it is working now, who knows under what conditions it was reproduced originally. Moreover the hotfix was confirmed by customer if I'm not mistaken.

Objections Florian?

(In reply to Tomas Hudziec from comment #11)
> How should I reproduce the bug? Do I have to use some Nvidia proprietary
> module mentioned in the original bug 1585230?

No, this shouldn't be Nvidia-specific - for example I tested on intel when writing the original patch.

It's simply possible that your test configuration happens to match the layout that mutter uses as fallback if no configuration is found. As mutter's fallback is always a horizontal layout, I found that a vertical layout makes for a good test case.

(In reply to Tomas Pelka from comment #12)
> Objections Florian?

No, not at all - the important thing is that the issue is fixed for the customer, which seems to be the case.

(In reply to Florian Müllner from comment #13)
> It's simply possible that your test configuration happens to match the
> layout that mutter uses as fallback if no configuration is found.

Yes, it was the case.

> As mutter's fallback is always a horizontal layout, I found that a vertical
> layout makes for a good test case.

With vertical layout I reproduced the issue, mutter-3.26.2-13 sets the layout back to horizontal. Vertical layout stays with mutter-3.26.2-16, which is correct. Switching to verified finally. Apologies for upset.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2018:2765