random(4) related panic: sleeping without a mutex

Mark Murray markm at FreeBSD.ORG
Mon Apr 12 15:20:21 PDT 2004

"Simon L. Nielsen" writes:
> > Yeah. Part of the startup does a "kickstart" of the device by writing
> > garbage to it. I wonder why it didn't do that this time. Is your
> > /etc/rc* completely up to date? What is in your rc.conf?
> The problem is that I use initdiskless which is run as the very first
> rc.d script, even before initrandom.  It has to run first since /etc is
> replaced by that script.

Hmmm. Methinks initdiskless should depend on initrandom.

> As a workaround I added the hack from initrandom (ps -fauxww; sysctl...
> >/dev/random and so on) and the system boots again.  So I'm not really
> sure if this should be fixed in rc.d or in the kernel.

Short term; rc.d. Long term, kernel might not be a bad idea. There are
some very nasty initial conditions to fix first. The device MUST start
up secure, for one.

> Is it intentional that random(4) isn't seeded by the kernel itself now?

Unintentional, but hard to fix.

Mark Murray
iumop ap!sdn w,I idlaH

More information about the freebsd-current mailing list