Adding a small delay (in my case 2 secs) in a setfont wrapper script makes the problem go away for KMS clients too.
Also, in /lib/udev/console-setup-tty and in /lib/udev/rules.d/85-console-setup some comments indicate that there are known race conditions, maybe this bug is caused by them.
Adding a small delay (in my case 2 secs) in a setfont wrapper script makes the problem go away for KMS clients too.
Also, in /lib/udev/ console- setup-tty and in /lib/udev/ rules.d/ 85-console- setup some comments indicate that there are known race conditions, maybe this bug is caused by them.