Bug 1305003

I'd set this bug to block the tracker for the Wayland-by-default Change too:

https://fedoraproject.org/wiki/Changes/WaylandByDefault

only it says "Tracker bug: <will be assigned by the Wrangler>". Sigh.

The findings of https://bugzilla.redhat.com/show_bug.cgi?id=1281675#c47 apply likewise to a system on which the current Fedora development tree has been installed permanently using the network install media.
Thus, as of today, Xwayland belongs to the group of applications comprised of gnome-terminal, nautilus, sealert, etc. which do not start up correctly because of bug 1281675. After a "GNOME on Xorg" session, plain "GNOME" sessions will use Xwayland successfully as expected. Therefore, this bug is redundant.

As exhibited in comment https://bugzilla.redhat.com/show_bug.cgi?id=1281675#c53 of bug 1281675, GDM appears not set the session type unless it gets changed at the login panel. Simply toggling the session type forth and back allows for a working "GNOME (on Wayland)" session.
By disabling automatic login and setting a password for liveuser before booting into graphical mode, the live system behaves exactly like an installed system.

this should be fixed in rawhide

This bug appears to have been reported against 'rawhide' during the Fedora 24 development cycle.
Changing version to '24'.

More information and reason for this action is here:
https://fedoraproject.org/wiki/Fedora_Program_Management/HouseKeeping/Fedora24#Rawhide_Rebase

This does not seem to have been fixed. Current F24 Workstation nightlies have gdm-3.19.90-1.fc24, but still seem to be running X, not Wayland. The last F24 Workstation live we have is https://kojipkgs.fedoraproject.org/compose/branched/Fedora-24-20160225.n.0/compose/Workstation/x86_64/iso/Fedora-Live-Workstation-x86_64-Rawhide-20160225.0.iso , you'll have to boot with 'enforcing=0' to get anywhere. It boots to GDM instead of autologging in as it should; if you click 'Live System User' you get an X session. GDM also appears to be running on X.

(In reply to Adam Williamson from comment #6)
Here, a system with an RV620 based video card boots successfully from a Fedora-Live-Workstation-x86_64-Rawhide-20160225.0 live media with kernel option "enforcing=0" into the default session GNOME (on Wayland) without displaying the login panel.

Hmmm...perhaps there's a difference in behaviour between bare metal with an accelerated video card, and VMs. I *do* get a crash notification for the check-accelerated helper thingy, now I come to think of it. I'll check a bit further (see if I can reproduce Wayland on bare metal) and maybe file a separate bug, thanks.

(In reply to Adam Williamson from comment #8)
I have just tried the Fedora-Live-Workstation-x86_64-Rawhide-20160225.0 live media again inside gnome-boxes, and it actually boots with kernel option "enforcing=0" into the (flickering) liveuser session using GNOME (on Wayland) - just switch to a virtual console and check the running processes.
Your issue is related to the virtual QXL driver; see attached screenshot of the dmesg output.

Created attachment 1131659 [details]
Output of dmesg for GNOME (on Wayland) session inside gnome-boxes

Actually, I had already stated a week ago in https://bugzilla.redhat.com/show_bug.cgi?id=1308771#c28 in reply to your comment https://bugzilla.redhat.com/show_bug.cgi?id=1308771#c26 that

"1. Current live media boot correctly into GNOME (on Wayland) on bare metal after adding kernel option "enforcing=0". The steps suggested in comment 23 are unnecessary.
2. Current live media boot correctly into GNOME (on Wayland) in a -virtual machine- with kernel option "enforcing=0" but heavy flickering related to a QXL DRM issue (qxl 0000:00:02.0: ... unpin not necessary) makes the the session unusable."

That's nice, but it's not what I see.

(In reply to Adam Williamson from comment #12)
The host system is current Fedora 23 including packages:
 - gnome-boxes-3.18.1-1.fc23
 - qemu-*-2.4.1.7.fc23
 - libvirt-*-1.2.18.2-2.fc23
 - spice-*-0.30-1.fc23
If you are using an older/different flavour of Fedora/Red Hat this is likely to have an influence ..

I'm running Rawhide. And I don't use Boxes, I use virt-manager.

So indeed I get Wayland booting on real hardware (tested the 2016-02-25 Rawhide Workstation live). The general case is fixed here, looks like there's some specific case with (some) virtual machines, perhaps related to the check-accelerated crasher...let's close this and open a new bug.

Adam, when you file the new bug, can you please link it here? Thanks.