According to this post:
“The Qubes Installer automatically detects the use of a USB keyboard and adds the correct boot option.”
I take this to mean that if, as in my case, there is only a USB keyboard, then the installer will configure grub with
usbcore.authorized_default=0
and the appropriate bypass privilege granted to the attached USB controller.
However, when I try to boot, I just get the error message about rd.qubes.hide_all_usb due to USB being open. The boot process halts with this being the only message displayed, so I have no opportunity to run grub2-config in order to correct the problem with an updated config file (and wouldn’t have a usable keyboard even if I did).
I tried to boot from another live OS in order to get into the Qubes partition and manually correct this but then I ran into other problems. Intuitively, I’m off in the weeds and this should just work out of the box. Is there a better way to do this? I’m sure I’m not the only user to ever try Qubes without a PS/2 keyboard.
My understanding is that the installer would need to find the controller to which the USB input device(s) are attached, punch a hole in the security policy for this controller, and use grub2-mkconfig to modify the grub config accordingly, all automatically during install. This could get complicated if there is more than one input device or they happen to be sitting deep in the hub tree or distributed across separate trees. As a practical matter, it might therefore be acceptable to ignore input devices behind hubs, with the lower PCI controller address being used to choose which controller gets the bypass privilege in the event that there are multiple instances of this configuration discovered. But is this happening under the installer hood? I have no idea if it’s trying and failing, or just not trying.