bridge and ng_ether deja vu
Gleb Smirnoff
glebius at FreeBSD.org
Wed Oct 5 04:49:09 PDT 2005
Andrew, Max,
exactly after one year I see if_bridge(4) having the same problem
with ng_ether(4) as old bridge(4) had. The problem is that packets
flowed thru ng_ether miss bridge processing. The problem is explained
well in this mail:
http://lists.freebsd.org/mailman/htdig/freebsd-net/2004-May/003881.html
It applies to if_bridge, too.
There was two ways to fix the problem:
1) Split ether_input() into ether_input()+ether_input_frame(), like
to ether_output is split. In this case ether_input() must end up
with ng_ether call and ether_input_frame(). ether_input_frame()
does all bridge(4) processing and calls ether_demux().
ng_ether_rcv_upper() simply calls ether_input_frame().
2) Copy and paste the bridge processing into ng_ether_rcv_upper().
We (me, sam, andre) have choosed the second way:
http://lists.freebsd.org/pipermail/cvs-src/2004-October/033496.html
However, I have cut all the bridge stuff from if_ethersubr.c and moved
it into bridge.c. This is how it looked like:
http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/net/if_ethersubr.c.diff?r1=1.178&r2=1.179&f=h
Now I notice the same problem with if_bridge. And I decided to go second way
again, but noticed that unfortunately I again have to copy and paste > 20 lines
from if_ethersubr.c to ng_ether.c.
So, the question is: is it possible to push the block about ETHER_IS_MULTICAST
into bridge_input()? This will make copy and paste region smaller.
--
Totus tuus, Glebius.
GLEBIUS-RIPN GLEB-RIPE
More information about the freebsd-net
mailing list