svn commit: r260224 - head/sys/netinet

Gleb Smirnoff glebius at FreeBSD.org
Thu Jan 9 00:13:51 UTC 2014


  Bryan,

On Mon, Jan 06, 2014 at 09:09:45AM -0600, Bryan Venteicher wrote:
B> > On Fri, Jan 03, 2014 at 11:03:12AM +0000, Gleb Smirnoff wrote:
B> > > Author: glebius
B> > > Date: Fri Jan  3 11:03:12 2014
B> > > New Revision: 260224
B> > > URL: http://svnweb.freebsd.org/changeset/base/260224
B> > >
B> > > Log:
B> > >   Make failure of ifpromisc() a non-fatal error. This makes it possible
B> > to
B> > >   run carp(4) on vtnet(4).
B> > >
B> >
B> > vtnet(4) is the only device that doesn't correctly support
B> > promiscuous mode?  I don't know details of vtnet(4) but it seems
B> > it's not hard to mimic promiscuous mode.  I'm not sure why the
B> > driver returns ENOTSUP to user land given that vtnet(4) defaults
B> > to promiscuous mode for backwards compatibility.  It also does
B> > not handle multicast filter configuration when VTNET_FLAG_CTRL_RX
B> > flag is not set.  If vtnet(4) does not support multicast filter,
B> > it shouldn't announce IFF_MULTICAST. I wonder how vtnet(4) can work
B> > with carp(4) given that its multicast handling is ignored.
B> >
B> 
B> I've talked to Gleb off-list about this. I intent to remove the default to
B> promiscuous mode hack. Previous versions of the specification had a
B> footnote about this, but it has since been removed. Note that both promisc
B> and multicast likely require some host configuration on the
B> tap/bridge/physical interfaces as well.
B> 
B> I'll look at the multicast handling. If carp(4) already works with bhyve,
B> we might want to add a minimalist control virtqueue support to bhyve.

If you decide that vtnet(4) should provide promiscuous mode (or pretend to),
and implement that, and carp(4) runs okay, then feel free to back out r260224.

-- 
Totus tuus, Glebius.


More information about the svn-src-all mailing list