FreeBSD as read-only firmware

Garrett Cooper yanegomi at
Sat Nov 3 21:08:35 UTC 2012

On Sat, Nov 3, 2012 at 6:34 AM, Alexander Yerenkow <yerenkow at>wrote:


Of course,  that's all IMHO and fit for my usage:


> 2) .vmdk simply deployed into Esxi/virtualbox (not sure nanobsd can produce
> that)

There's no reason why nanobsd couldn't (or shouldn't) do that. It's
something that I thought about when the "Google Code In" announcement was
sent out; honestly it's no more complicated than installing a port and
running a few commands on the full disk image produced with nanobsd.

> 3) Transparent /etc/ modifiying VS nanobsd approach (edit, don't forget
> mount /cfg, copy there;)

I agree that /cfg doesn't make things transparent for end-users, but that's
a feature (that could potentially be polished up with a one-liner
command/alias). rc.initdiskless is the one that does the shuffling/stomping
of files BTW.

> 4) Only OS, no packages included - e.g. I can upgrade/downgrade packages
> without touching any byte of OS. Except for symlinks :) nanobsd specified
> that if you want packages - you need built them in.

You can do that with nanobsd too. People that use FreeNAS 8 frequently
tried to do this :).

> Of course differences not so big, and I'm not saying that my way is more
> better.
> It just raised question deep in me - why OS still aren't modularized, and
> most of it not in RO (while it should).
> Something like this


More information about the freebsd-current mailing list