Date: Sun, 30 Dec 2012 10:04:54 -0700 From: Peter Grehan <grehan@freebsd.org> To: Alfred Perlstein <bright@mu.org> Cc: neel@freebsd.org, freebsd-virtualization@freebsd.org Subject: Re: question on bhyve + 9-stable Message-ID: <50E07436.4040105@freebsd.org> In-Reply-To: <50E06701.7070409@mu.org> References: <50E06701.7070409@mu.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Alfred,
> I only have 1 really fast machine, so I've decided to give bhyve a whirl
> and try to do -stable development inside of a bhyve instance.
Great !
> The problem I'm having is that:
> 1) the 9.1-RELEASE image does not seem to boot because it can not mount
> the rootfs (iso). I think this may be due to missing virtualization
> modules.
Yep, that's right. Michael Dexter has pre-built virtio modules at
http://mirrors.nycbug.org/pub/BHyVe/r244024/
.. and I guess the long-term solution for running older FreeBSD
releases is to have bhyve implement block/network device models that
will be recognized. AHCI looks like a good pick for a block model: any
input on a network device would be appreciated (e1000 does seem to stick
out).
> 2) building release iso from 9-stable, while it has the klds needed, the
> image seems to causes the hypervisor to exit.
That's the new TSC sync detection code, which does a CPUID with
basically a random leaf value. bhyve should deal with that: a fix will
be in when I'm back from vacation. But, in the meantime, there are two
workarounds:
- use single CPU only (-c 1), or
- force the hard-coded hypervisor detection code in FreeBSD to trigger
with a loader variable
set smbios.bios.vendor="Bochs"
... and, since the TSC is actually OK, force the use of that as the
timecounter
set kern.timecounter.invariant_tsc=1
set kern.timecounter.smp_tsc=1
later,
Peter.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?50E07436.4040105>