> In message <mtubrtzgs05.fsf@contents-vnder-pressvre.mit.edu>, "Nathan J. Willia
> ms" writes:
> >itojun@itojun.org (Jun-ichiro itojun Hagino) writes:
> >
> >> 	well, then, we should probably put some code into sysinst that warns
> >> 	user like "password entries are created with weak random number, you
> >> 	will not want to configure root password during this installation
> >> 	session" for kernels without rnd(4).
> >
> >This seems totally overwrought. All the random number is used for here
> >is generating a salt, whose goal is to make dictionary attacks on the
> >encrypted password difficult, right? I don't think that requires a
> >top-notch random-number generator.
> 
> I was about to post the same observation.  (The situation will be 
> different if, as itojun suggests, sysinst generates ssh keys, but 
> perhaps that should be done at first boot?)

	there are lots of other stuffs that is using random number source
	(TCP sequence number, ffs_dirpref() as discussed, fragment id, ...). 
	i think we really should add rnd(4) to installation kernel.

itojun