On Sat, Aug 19, 2006 at 07:05:49PM +0530, Rajkumar S wrote: > On 8/19/06, Ruslan Ermilov <[EMAIL PROTECTED]> wrote: > >On Sat, Aug 19, 2006 at 05:44:39PM +0530, Rajkumar S wrote: > >> The man page states > >> > >> DESCRIPTION > >> The ip_input node type takes all received packets and queues them > >into > >> the IP in input processing subsystem. > > >It does this simple thing: > > > >> netisr_dispatch(NETISR_IP, m); > > > >which means putting the mbuf pointed to by "m" into the IP > >input queue and scheduling a call to ip_input(). > > Thanks a lot for your answer. > > So this is used to inject packets into the ip stack? > Yes.
> also, normally > packets received from ethernet is passed into the same point, right? > Normally, yes. By sys/net/if_ethersubr.c::ether_demux(). > Any points to docs to read about a packet's traversal in FreeBSD ip > stack? (especially wrt pf) > What level of detalization do you need? Filters, such as pf(4), are embedded into the normal processing using the pfil(9) API. Cheers, -- Ruslan Ermilov [EMAIL PROTECTED] FreeBSD committer
pgpR0D2PNkl4d.pgp
Description: PGP signature
