summaryrefslogtreecommitdiffstats
path: root/proto/rip
diff options
context:
space:
mode:
Diffstat (limited to 'proto/rip')
-rw-r--r--proto/rip/rip.c42
-rw-r--r--proto/rip/rip.h1
2 files changed, 22 insertions, 21 deletions
diff --git a/proto/rip/rip.c b/proto/rip/rip.c
index 233ded6..1ba1a24 100644
--- a/proto/rip/rip.c
+++ b/proto/rip/rip.c
@@ -65,7 +65,7 @@ rip_tx_err( sock *s, int err )
}
static void
-rip_tx_prepare(struct proto *p, ip_addr daddr, struct rip_block *b, struct rip_entry *e )
+rip_tx_prepare(struct proto *p, ip_addr daddr, struct rip_block *b, struct rip_entry *e, struct rip_interface *rif )
{
DBG( "." );
b->family = htons( 2 ); /* AF_INET */
@@ -83,6 +83,10 @@ rip_tx_prepare(struct proto *p, ip_addr daddr, struct rip_block *b, struct rip_e
* test to avoid dereferencing NULL pointer. --mj
*/
#if 0
+ /* IS this really neccessary?
+ If nexthop and destination of this packet are on same interface, split horizont applies and *nothing* should be sent....
+ */
+
neighbor *n1, *n2;
n1 = neigh_find( p, &daddr, 0 ); /* FIXME, mj: this is neccessary for responses, still it is too complicated for common case */
n2 = neigh_find( p, &e->nexthop, 0 );
@@ -90,14 +94,14 @@ rip_tx_prepare(struct proto *p, ip_addr daddr, struct rip_block *b, struct rip_e
b->nexthop = e->nexthop;
else
#endif
- b->nexthop = IPA_NONE;
+ b->nexthop = IPA_NONE;
}
ipa_hton( b->nexthop );
#else
b->pxlen = e->n.pxlen;
#endif
b->metric = htonl( e->metric );
- if (ipa_equal(e->whotoldme, daddr)) { /* FIXME: ouch, daddr is some kind of broadcast address. How am I expected to do split horizon?!?!? */
+ if (if_connected(e->whotoldme, rif->iface)) {
DBG( "(split horizon)" );
b->metric = htonl( P_CF->infinity );
}
@@ -131,7 +135,7 @@ rip_tx( sock *s )
if (!rif->triggered || (!(e->updated < now-5))) {
- rip_tx_prepare( p, s->daddr, packet->block + i, e );
+ rip_tx_prepare( p, s->daddr, packet->block + i, e, rif );
if (i++ == ((P_CF->authtype == AT_MD5) ? PACKET_MD5_MAX : PACKET_MAX)) {
FIB_ITERATE_PUT(&c->iter, z);
goto break_loop;
@@ -145,11 +149,7 @@ rip_tx( sock *s )
packetlen = rip_outgoing_authentication(p, (void *) &packet->block[0], packet, i);
DBG( ", sending %d blocks, ", i );
-#if 0 /* FIXME: enable this for production! */
- if (i == !!P_CF->authtype)
- continue;
-#endif
- if (!i) {
+ if (i == !!P_CF->authtype) {
DBG( "not sending NULL update\n" );
c->done = 1;
goto done;
@@ -180,14 +180,14 @@ static void
rip_sendto( struct proto *p, ip_addr daddr, int dport, struct rip_interface *rif )
{
struct iface *iface = rif->iface;
- struct rip_connection *c = mb_alloc( p->pool, sizeof( struct rip_connection ));
+ struct rip_connection *c;
static int num = 0;
if (rif->busy) {
log (L_WARN "Interface %s is much too slow, dropping request", iface->name);
- /* FIXME: memory leak */
return;
}
+ c = mb_alloc( p->pool, sizeof( struct rip_connection ));
rif->busy = c;
c->addr = daddr;
@@ -367,7 +367,7 @@ rip_process_packet( struct proto *p, struct rip_packet *packet, int num, ip_addr
#ifndef IPV6
ipa_ntoh( block->netmask );
ipa_ntoh( block->nexthop );
- if (packet->heading.version == RIP_V1) /* FIXME: switch to disable this? (nonurgent) */
+ if (packet->heading.version == RIP_V1) /* FIXME (nonurgent): switch to disable this? */
block->netmask = ipa_class_mask(block->network);
#endif
process_block( p, block, whotoldme );
@@ -546,7 +546,6 @@ static struct rip_interface *
new_iface(struct proto *p, struct iface *new, unsigned long flags, struct iface_patt *patt )
{
struct rip_interface *rif;
- int want_multicast = 0;
rif = mb_allocz(p->pool, sizeof( struct rip_interface ));
rif->iface = new;
@@ -555,14 +554,14 @@ new_iface(struct proto *p, struct iface *new, unsigned long flags, struct iface_
rif->patt = (struct rip_patt *) patt;
if (rif->patt)
- want_multicast = (!(rif->patt->mode & IM_BROADCAST)) && (flags & IF_MULTICAST);
+ rif->multicast = (!(rif->patt->mode & IM_BROADCAST)) && (flags & IF_MULTICAST);
/* lookup multicasts over unnumbered links - no: rip is not defined over unnumbered links */
- if (want_multicast)
+ if (rif->multicast)
DBG( "Doing multicasts!\n" );
rif->sock = sk_new( p->pool );
- rif->sock->type = want_multicast?SK_UDP_MC:SK_UDP;
+ rif->sock->type = rif->multicast?SK_UDP_MC:SK_UDP;
rif->sock->sport = P_CF->port;
rif->sock->rx_hook = rip_rx;
rif->sock->data = rif;
@@ -580,13 +579,13 @@ new_iface(struct proto *p, struct iface *new, unsigned long flags, struct iface_
rif->sock->tos = IP_PREC_INTERNET_CONTROL;
if (new) {
- rif->sock->daddr = new->addr->brd;
if (new->addr->flags & IA_UNNUMBERED)
log( L_WARN "%s: rip is not defined over unnumbered links", P_NAME );
- if (want_multicast) {
+ if (rif->multicast) {
rif->sock->daddr = ipa_from_u32(0xe0000009);
rif->sock->saddr = ipa_from_u32(0xe0000009);
- }
+ } else
+ rif->sock->daddr = new->addr->brd;
}
if (!ipa_nonzero(rif->sock->daddr)) {
@@ -598,7 +597,7 @@ new_iface(struct proto *p, struct iface *new, unsigned long flags, struct iface_
/* Don't try to transmit into this one? Well, why not? This should not happen, anyway :-) */
}
- TRACE(D_EVENTS, "%s: listening on %s, port %d, mode %s (%I)", P_NAME, rif->iface ? rif->iface->name : "(dummy)", P_CF->port, want_multicast ? "multicast" : "broadcast", rif->sock->daddr );
+ TRACE(D_EVENTS, "%s: listening on %s, port %d, mode %s (%I)", P_NAME, rif->iface ? rif->iface->name : "(dummy)", P_CF->port, rif->multicast ? "multicast" : "broadcast", rif->sock->daddr );
return rif;
}
@@ -642,11 +641,12 @@ rip_if_notify(struct proto *p, unsigned c, struct iface *iface)
if (!k) return; /* We are not interested in this interface */
lock = olock_new( p->pool );
- lock->addr = IPA_NONE; /* FIXME: how to set this? */
+ lock->addr = ipa_from_u32(0xe0000009);
lock->port = P_CF->port;
lock->iface = iface;
lock->hook = rip_real_if_add;
lock->data = p;
+ lock->type = OBJLOCK_UDP;
olock_acquire(lock);
}
}
diff --git a/proto/rip/rip.h b/proto/rip/rip.h
index 465e4d3..96c5646 100644
--- a/proto/rip/rip.h
+++ b/proto/rip/rip.h
@@ -104,6 +104,7 @@ struct rip_interface {
struct rip_patt *patt;
int triggered;
struct object_lock *lock;
+ int multicast;
};
struct rip_patt {