installboot is broken

[Johnny Billquist <bqt%softjar.se@localhost>]

I just went through some fiddling around with simh, since it recently 
started implementing 86x0 emulation.
Of course I wanted to test all my code that I've done for NetBSD on this 
simulation, to see what happens,

I failed. And the reason is that installboot is broken. installboot does 
not use the data from the primary boot block that it is given, for data 
in block 0. Instead installboot creates its own block 0 content. This 
works on the MicroVAXen, because of how VMB on those machine work. It 
does not work an older VAXen, because VMB acts in a different way.
Block 0 of the primary bootstrap is correct, but block 0 as done by 
installboot is not.
In the past, disklabel was used to write the boot block, and disklabel 
didn't try to be clever. We need to go back to this solution again.

Does anyone feel like fixing, or should I go and try figuring out 
exactly which parts of block 0 to pick from disk and from primary 
bootstrap to merge, to actually get a correct block 0?

        Johnny