On Tue, Mar 31, 2020 at 05:33:24PM +0200, Mark Kettenis wrote: >> > I didn't write *that* code. Just the code that figures out which > /dev/tty[C-J]0 is the actual console corresponding to /dev/console. > > What seems to be happening here is that xenodm(1) and getty(8) are > racing eachother. Xorg looks for a free VT, chooses /dev/ttyC1 but > after it actually opens that VT, getty(8) is already running on it and > blocks us from switching into raw mode?
On other architectures with VT, there is one (/dev/ttyC4) without a getty process, and xenodm is using an /etc/X11/xenodm/Xservers files with an explicit parameter to X to pin it to ttyC4. I haven't checked which part of the puzzle is missing on arm64. -- Matthieu Herrb