Does anybody else see the "pm" framebuffer gettting fbconnect'ed twice in
-current (once for console initialization, once for attach)? Also I want
an open() of /dev/fb? to save the colormap and close() to restore it since
the new rcons stuff (not yet committed) uses color. Oh and fb.c should
really have a dependancy line in the Makefile that includes fb_usrreq.c.

I'll commit fixes for these unless somebody has objections. Those crashes
have stopped since I took off that IBM disk. It's a little unnerving,
since I never saw ULTRIX crashing when I was running it off that disk.

-ad