On Mon, Mar 26, 2007 at 01:10:00PM +0400, Gleb Smirnoff wrote: > On Mon, Mar 26, 2007 at 04:50:06AM +0000, Yar Tikhiy wrote: > Y> yar 2007-03-26 04:50:06 UTC > Y> > Y> FreeBSD src repository > Y> > Y> Added files: > Y> sys/net if_edsc.c > Y> Log: > Y> Introduce a new toy interface, edsc(4). It's a discard interface > Y> imitating an Ethernet device, so vlan(4) and if_bridge(4) can be > Y> attached to it for testing and benchmarking purposes. Its source > Y> can be an introduction to the anatomy of a network interface driver > Y> due to its simplicity as well as to a bunch of comments in it. > Y> > Y> (The rest of needed changes were in my previous commit, which got > Y> interrupted in the middle. Alas, CVS commits are not atomic.) > > I am surprised that you didn't use ng_eiface(4) for the same purpose.
No reason for surprise, I tried to. > If you attach ng_hole(4) to the ng_eiface, then you'll get the same > functionality as edsc(4). But if you attach with ngctl(8) to the > interface via ng_socket(4) you will get a hex dump of the frames > sent to the interface. > > AFAIK, the edsc(4)'s functionality is already present in the system. Functionality, but not speed. I think I explained on -net why I needed edsc(4): ng_eiface+ng_hole just couldn't consume a gigabit flow for me. Perhaps I didn't mention an important detail: I was interested in pps rate, so I used minimal packets, and the callback mechanism of ng_eiface(4) was falling far behind rates as high as 1Mpps. -- Yar _______________________________________________ [email protected] mailing list http://lists.freebsd.org/mailman/listinfo/cvs-all To unsubscribe, send any mail to "[EMAIL PROTECTED]"
