cvs commit: src/sys/net if_tap.c

Gleb Smirnoff glebius at FreeBSD.org
Mon Feb 5 12:43:36 UTC 2007


On Sat, Feb 03, 2007 at 02:57:46AM +0000, Bruce M Simpson wrote:
B> bms         2007-02-03 02:57:46 UTC
B> 
B>   FreeBSD src repository
B> 
B>   Modified files:
B>     sys/net              if_tap.c 
B>   Log:
B>   Drop unicast Ethernet frames not destined for the configured address
B>   of a tap(4) instance, if IFF_PROMISC is not set.
B>   
B>   In tap(4), we should emulate the effect IFF_PROMISC would have on
B>   hardware, otherwise we risk introducing layer 2 loops if tap(4) is
B>   used with bridges. This means not even bpf(4) gets to see them.
B>   
B>   This patch has been tested in a variety of situations. Multicast and
B>   broadcast frames are correctly allowed through. I have observed this
B>   behaviour causing problems with multiple QEMU instances hosted on
B>   the same FreeBSD machine.
B>   
B>   The checks in in ether_demux() [if_ethersubr.c, rev 1.222, line 638]
B>   are insufficient to prevent this bug from occurring, as ifp->if_vlantrunk
B>   will always be NULL for the non-vlan case.

btw, we have a strange situation with handling off non-broadcast
frames not destined to us in general.

In a case, when we have generic Ethernet in promisc, and received foreign
packets (on a hub, or a buggy switch), they are forwarded to ether_demux()
and to upper layers respectively. Isn't this incorrect?

Moreover, in case of promisc + net.inet.ip.forwarding=1 our box will resend
all the received foreign packets :(

What do you think about checking that (ether_dhost == ours || BCAST || MCAST)
before sending packet to ether_demux?

-- 
Totus tuus, Glebius.
GLEBIUS-RIPN GLEB-RIPE


More information about the cvs-src mailing list