Reading
[here], I see this:
6. Kernel parameters for changing SELinux modes at boot:
autorelabel=1 → forces the system to relabel
selinux=0 → kernel doesn't load any part of the SELinux infrastructure
enforcing=0 → boot in permissive mode
So, by logic, if selinux=0 fixed a boot problem, then the existing selinux infrastructure which worked for kernel 5.0.14 broke 5.1. That looks like some kernel development in 5.1 is a violation of the selinux infrastructe that was OK for 5.0.14 and many prior kernels. Maybe kernel devs need to talk with selinux devs.