[Ted: please forward as appropriate...]

Recent pmax kernels have been panicking early in the boot process.
I've used stack traceback to localize the cause of the panic
to exit1(), and  debugging printfs() to identify the problem. 

It seems that, somehow, p->p_p_session->s_ttyp ends up being NULL
in exit1() early on the boot process. (For me it's a process with
pid 5, but I have a non-standard /etc/rc.)
Due to dereferencing the null pointer, the kernel gets a trap
and panics.

I have no idea if this is some side-effect of recent changes
to the pmax, console  code, or something *I* did to my own
tree (I doubt it though, since other pmax are seeing the same thing),
or due to something in the machine-independent code.

--Jonathan Stone