SEO

Here is an updated version of the diff which fixes problems with 'make
build' and 'make release'.

- in ip_carp.h, define CARPDEVNAMSIZ instead of using IFNAMSIZ, so that
random applications don't have to include if.h
- get rid of carp_sendif(), store the pointer to carpdev directly in
struct ifnet, so random kernel code doesn't have to include ip_carp.h
- some files missing from the diff.
- other misc. cleanup.

As an aside, I've not recieved a single reports from a user about the
above problems, so I have to imagine that nobody cares enough to test
it! IF YOU USE CARP, YOU SHOULD TEST THIS CODE!

On Mon, Dec 06, 2004 at 02:10:10AM +0000, Ryan McBride wrote:
> Here's the long-promised diff which allows you to specify a physical
> interface 'carpdev' which the carp interface will bind to, and that
> interface does not have to have an IP address on the same subnet (or at
> all, in fact).
>
> $ sudo ifconfig carp3 192.168.10.10 vhid 10 carpdev hme0
>
> If you don't specify it, the attachment behaves as the current code, ie.
> it figures out the correct carpdev on the address. This means that
> existing configurations should work correctly with this diff.
>
> Pleas test this - it is important as it allows CARP to work on small
> subnets or other situations with address shortage. I'm interested in
> hearing reports of this code running on existing configurations as well
> as the new functionality that it provides.


Index: sys/net/if_ethersubr.c
================================================== =================
RCS file: /cvs/src/sys/net/if_ethersubr.c,v
retrieving revision 1.81
diff -u -r1.81 if_ethersubr.c
--- sys/net/if_ethersubr.c 28 Nov 2004 23:39:45 -0000 1.81
+++ sys/net/if_ethersubr.c 6 Dec 2004 20:35:59 -0000
@@ -238,8 +238,8 @@
* Assumes that ifp is actually pointer to arpcom structure.
*/
int
-ether_output(ifp, m0, dst, rt0)
- struct ifnet *ifp;
+ether_output(ifp0, m0, dst, rt0)
+ struct ifnet *ifp0;
struct mbuf *m0;
struct sockaddr *dst;
struct rtentry *rt0;
@@ -251,8 +251,28 @@
struct rtentry *rt;
struct mbuf *mcopy = (struct mbuf *)0;
struct ether_header *eh;
- struct arpcom *ac = (struct arpcom *)ifp;
+ struct arpcom *ac = (struct arpcom *)ifp0;
short mflags;
+ struct ifnet *ifp = ifp0;
+
+#if NCARP > 0
+ if (ifp->if_type == IFT_CARP) {
+ struct ifaddr *ifa;
+
+ /* loop back if this is going to the carp interface */
+ if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
+ (ifa = ifa_ifwithaddr(dst)) != NULL &&
+ ifa->ifa_ifp == ifp0)
+ return (looutput(ifp0, m, dst, rt0));
+
+ ifp = ifp->if_carpdev;
+ ac = (struct arpcom *)ifp;
+
+ if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
+ (IFF_UP|IFF_RUNNING))
+ senderr(ENETDOWN);
+ }
+#endif /* NCARP > 0 */

if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
senderr(ENETDOWN);
@@ -513,12 +533,12 @@
#endif

#if NCARP > 0
- if (ifp->if_carp) {
- int error;
- error = carp_output(ifp, m, dst, NULL);
- if (error)
- goto bad;
- }
+ if (ifp->if_carp) {
+ int error;
+ error = carp_output(ifp0, m, dst, NULL);
+ if (error)
+ goto bad;
+ }
#endif

mflags = m->m_flags;
@@ -535,6 +555,10 @@
return (error);
}
ifp->if_obytes += len + ETHER_HDR_LEN;
+#if NCARP > 0
+ if (ifp != ifp0)
+ ifp0->if_obytes += len + ETHER_HDR_LEN;
+#endif /* NCARP > 0 */
if (mflags & M_MCAST)
ifp->if_omcasts++;
if ((ifp->if_flags & IFF_OACTIVE) == 0)
@@ -650,9 +674,9 @@
#endif /* NVLAN > 0 */

#if NCARP > 0
- if (ifp->if_carp &&
- carp_forus(ifp->if_carp, eh->ether_dhost))
- goto decapsulate;
+ if (ifp->if_carp && ifp->if_type != IFT_CARP &&
+ (carp_input(eh, m) == 0))
+ return;
#endif /* NCARP > 0 */

ac = (struct arpcom *)ifp;
Index: sys/netinet/if_ether.c
================================================== =================
RCS file: /cvs/src/sys/netinet/if_ether.c,v
retrieving revision 1.54
diff -u -r1.54 if_ether.c
--- sys/netinet/if_ether.c 21 Jun 2004 23:50:37 -0000 1.54
+++ sys/netinet/if_ether.c 6 Dec 2004 21:03:45 -0000
@@ -557,7 +557,7 @@
#endif

#if NCARP > 0
- if (ac->ac_if.if_carp) {
+ if (ac->ac_if.if_carp && ac->ac_if.if_type != IFT_CARP) {
if (carp_iamatch(ac->ac_if.if_carp, ia,
&isaddr, &enaddr))
break;
@@ -645,7 +645,11 @@
}
}
} else if (rt->rt_ifp != &ac->ac_if && !(ac->ac_if.if_bridge &&
- (rt->rt_ifp->if_bridge == ac->ac_if.if_bridge))) {
+ (rt->rt_ifp->if_bridge == ac->ac_if.if_bridge)) &&
+ !(rt->rt_ifp->if_type == IFT_CARP &&
+ rt->rt_ifp->if_carpdev == &ac->ac_if) &&
+ !(ac->ac_if.if_type == IFT_CARP &&
+ ac->ac_if.if_carpdev == rt->rt_ifp)) {
log(LOG_WARNING,
"arp: attempt to add entry for %s "
"on %s by %s on %s\n",
Index: sys/netinet/ip_carp.c
================================================== =================
RCS file: /cvs/src/sys/netinet/ip_carp.c,v
retrieving revision 1.72
diff -u -r1.72 ip_carp.c
--- sys/netinet/ip_carp.c 30 Nov 2004 00:17:18 -0000 1.72
+++ sys/netinet/ip_carp.c 6 Dec 2004 23:17:38 -0000
@@ -29,7 +29,6 @@
/*
* TODO:
* - iface reconfigure
- * - track iface ip address changes;
* - support for hardware checksum calculations;
*
*/
@@ -41,6 +40,7 @@
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
+#include <sys/socketvar.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/device.h>
@@ -92,10 +92,19 @@

#include <netinet/ip_carp.h>

+struct carp_mc_entry {
+ LIST_ENTRY(carp_mc_entry) mc_entries;
+ union {
+ struct ether_multi *mcu_enm;
+ } mc_u;
+ struct sockaddr_storage mc_addr;
+};
+#define mc_enm mc_u.mcu_enm
+
struct carp_softc {
struct arpcom sc_ac;
+#define sc_carpdev sc_ac.ac_if.if_carpdev
int if_flags; /* current flags to treat UP/DOWN */
- struct ifnet *sc_ifp;
struct in_ifaddr *sc_ia; /* primary iface address */
struct ip_moptions sc_imo;
#ifdef INET6
@@ -132,6 +141,7 @@
struct timeout sc_md_tmo; /* master down timeout */
struct timeout sc_md6_tmo; /* master down timeout */

+ LIST_HEAD(__carp_mchead, carp_mc_entry) carp_mc_listhead;
};

int carp_suppress_preempt = 0;
@@ -161,7 +171,7 @@
int carp_hmac_verify(struct carp_softc *, u_int32_t *,
unsigned char *);
void carp_setroute(struct carp_softc *, int);
-void carp_input_c(struct mbuf *, struct carp_header *, sa_family_t);
+void carp_proto_input_c(struct mbuf *, struct carp_header *, sa_family_t);
void carpattach(int);
void carpdetach(struct carp_softc *);
int carp_prepare_ad(struct mbuf *, struct carp_softc *,
@@ -177,15 +187,20 @@
int carp_addrcount(struct carp_if *, struct in_ifaddr *, int);
enum { CARP_COUNT_MASTER, CARP_COUNT_RUNNING };

+void carp_multicast_cleanup(struct carp_softc *);
+int carp_set_ifp(struct carp_softc *, struct ifnet *);
int carp_set_addr(struct carp_softc *, struct sockaddr_in *);
-int carp_del_addr(struct carp_softc *, struct sockaddr_in *);
+int carp_join_multicast(struct carp_softc *, struct ifnet *);
#ifdef INET6
void carp_send_na(struct carp_softc *);
int carp_set_addr6(struct carp_softc *, struct sockaddr_in6 *);
-int carp_del_addr6(struct carp_softc *, struct sockaddr_in6 *);
+int carp_join_multicast6(struct carp_softc *, struct ifnet *);
#endif
int carp_clone_create(struct if_clone *, int);
int carp_clone_destroy(struct ifnet *);
+int carp_ether_addmulti(struct carp_softc *, struct ifreq *);
+int carp_ether_delmulti(struct carp_softc *, struct ifreq *);
+void carp_ether_purgemulti(struct carp_softc *);

struct if_clone carp_cloner =
IF_CLONE_INITIALIZER("carp", carp_clone_create, carp_clone_destroy);
@@ -279,25 +294,91 @@
struct ifaddr *ifa;
int s;

- s = splnet();
+ s = splsoftnet();
TAILQ_FOREACH(ifa, &sc->sc_ac.ac_if.if_addrlist, ifa_list) {
- if (ifa->ifa_addr->sa_family == AF_INET && sc->sc_ifp != NULL) {
- int count = carp_addrcount(
- (struct carp_if *)sc->sc_ifp->if_carp,
- ifatoia(ifa), CARP_COUNT_MASTER);
-
- if ((cmd == RTM_ADD && count == 1) ||
- (cmd == RTM_DELETE && count == 0))
- rtinit(ifa, cmd, RTF_UP | RTF_HOST);
+ switch (ifa->ifa_addr->sa_family) {
+ case AF_INET: {
+ int count = 0;
+ struct sockaddr sa;
+ struct rtentry *rt;
+ struct radix_node_head *rnh =
+ rt_tables[ifa->ifa_addr->sa_family];
+ struct radix_node *rn;
+
+ /*
+ * Avoid screwing with the routes if there are other
+ * carp interfaces which are master and have the same
+ * address.
+ */
+ if (sc->sc_carpdev != NULL &&
+ sc->sc_carpdev->if_carp != NULL) {
+ count = carp_addrcount(
+ (struct carp_if *)sc->sc_carpdev->if_carp,
+ ifatoia(ifa), CARP_COUNT_MASTER);
+ if ((cmd == RTM_ADD && count != 1) ||
+ (cmd == RTM_DELETE && count != 0))
+ continue;
+ }
+
+ /* Remove the existing host route, if any */
+ rtrequest(RTM_DELETE, ifa->ifa_addr,
+ ifa->ifa_addr, ifa->ifa_netmask,
+ RTF_HOST, NULL);
+
+ /* Check for a route on a physical interface */
+ rn = rnh->rnh_matchaddr(ifa->ifa_addr, rnh);
+ rt = (struct rtentry *)rn;
+
+ switch (cmd) {
+ case RTM_ADD:
+ if (rt && rt->rt_ifp != &sc->sc_ac.ac_if &&
+
+ rt->rt_flags & (RTF_CLONING|RTF_CLONED)) {
+ rtrequest(RTM_ADD, ifa->ifa_addr,
+ ifa->ifa_addr, ifa->ifa_netmask,
+ RTF_UP | RTF_HOST, NULL);
+ } else {
+ ifa->ifa_rtrequest = arp_rtrequest;
+ ifa->ifa_flags |= RTF_CLONING;
+
+ rtrequest(RTM_ADD, ifa->ifa_addr,
+ ifa->ifa_addr, ifa->ifa_netmask,
+ 0, NULL);
+ }
+ break;
+ case RTM_DELETE:
+ ifa->ifa_rtrequest = NULL;
+ ifa->ifa_flags &= ~RTF_CLONING;
+
+ if (!(rt && rt->rt_ifp != &sc->sc_ac.ac_if &&
+ rt->rt_flags & (RTF_CLONING|RTF_CLONED))) {
+ bcopy(ifa->ifa_addr, &sa, sizeof(sa));
+ satosin(&sa)->sin_addr.s_addr =
+ satosin(ifa->ifa_netmask
+ )->sin_addr.s_addr &
+ satosin(&sa)->sin_addr.s_addr;
+
+ rtrequest(cmd, &sa, ifa->ifa_addr,
+ ifa->ifa_netmask, 0, NULL);
+ }
+ break;
+ default:
+ break;
+ }
+ break;
}
+
#ifdef INET6
- if (ifa->ifa_addr->sa_family == AF_INET6) {
+ case AF_INET6:
if (cmd == RTM_ADD)
in6_ifaddloop(ifa);
else
in6_ifremloop(ifa);
- }
+ break;
#endif /* INET6 */
+ default:
+ break;
+ }
}
splx(s);
}
@@ -308,7 +389,7 @@
* but it seems more efficient this way or not possible otherwise.
*/
void
-carp_input(struct mbuf *m, ...)
+carp_proto_input(struct mbuf *m, ...)
{
struct ip *ip = mtod(m, struct ip *);
struct carp_softc *sc = NULL;
@@ -328,7 +409,7 @@
}

/* check if received on a valid carp interface */
- if (m->m_pkthdr.rcvif->if_carp == NULL) {
+ if (m->m_pkthdr.rcvif->if_type != IFT_CARP) {
carpstats.carps_badif++;
CARP_LOG(sc, ("packet received on non-carp interface: %s",
m->m_pkthdr.rcvif->if_xname));
@@ -396,12 +477,12 @@
}
m->m_data -= iplen;

- carp_input_c(m, ch, AF_INET);
+ carp_proto_input_c(m, ch, AF_INET);
}

#ifdef INET6
int
-carp6_input(struct mbuf **mp, int *offp, int proto)
+carp6_proto_input(struct mbuf **mp, int *offp, int proto)
{
struct mbuf *m = *mp;
struct carp_softc *sc = NULL;
@@ -417,7 +498,7 @@
}

/* check if received on a valid carp interface */
- if (m->m_pkthdr.rcvif->if_carp == NULL) {
+ if (m->m_pkthdr.rcvif->if_type != IFT_CARP) {
carpstats.carps_badif++;
CARP_LOG(sc, ("packet received on non-carp interface: %s",
m->m_pkthdr.rcvif->if_xname));
@@ -456,23 +537,23 @@
}
m->m_data -= *offp;

- carp_input_c(m, ch, AF_INET6);
+ carp_proto_input_c(m, ch, AF_INET6);
return (IPPROTO_DONE);
}
#endif /* INET6 */

void
-carp_input_c(struct mbuf *m, struct carp_header *ch, sa_family_t af)
+carp_proto_input_c(struct mbuf *m, struct carp_header *ch, sa_family_t af)
{
struct carp_softc *sc;
- struct ifnet *ifp = m->m_pkthdr.rcvif;
u_int64_t tmp_counter;
struct timeval sc_tv, ch_tv;

- /* verify that the VHID is valid on the receiving interface */
- TAILQ_FOREACH(sc, &((struct carp_if *)ifp->if_carp)->vhif_vrs, sc_list)
+ TAILQ_FOREACH(sc, &((struct carp_if *)
+ m->m_pkthdr.rcvif->if_carpdev->if_carp)->vhif_vrs, sc_list)
if (sc->sc_vhid == ch->carp_vhid)
break;
+
if (!sc || (sc->sc_ac.ac_if.if_flags & (IFF_UP|IFF_RUNNING)) !=
(IFF_UP|IFF_RUNNING)) {
carpstats.carps_badvhid++;
@@ -484,25 +565,6 @@
sc->sc_ac.ac_if.if_ipackets++;
sc->sc_ac.ac_if.if_ibytes += m->m_pkthdr.len;

-#if NBPFILTER > 0
- if (sc->sc_ac.ac_if.if_bpf) {
- /*
- * We need to prepend the address family as
- * a four byte field. Cons up a dummy header
- * to pacify bpf. This is safe because bpf
- * will only read from the mbuf (i.e., it won't
- * try to free it or keep a pointer to it).
- */
- struct mbuf m0;
- u_int32_t af0 = htonl(af);
-
- m0.m_next = m;
- m0.m_len = sizeof(af0);
- m0.m_data = (char *)&af0;
- bpf_mtap(sc->sc_ac.ac_if.if_bpf, &m0);
- }
-#endif
-
/* verify the CARP version. */
if (ch->carp_version != CARP_VERSION) {
carpstats.carps_badver++;
@@ -642,22 +704,28 @@
timeout_set(&sc->sc_md_tmo, carp_master_down, sc);
timeout_set(&sc->sc_md6_tmo, carp_master_down, sc);

+ LIST_INIT(&sc->carp_mc_listhead);
ifp = &sc->sc_ac.ac_if;
ifp->if_softc = sc;
snprintf(ifp->if_xname, sizeof ifp->if_xname, "%s%d", ifc->ifc_name,
unit);
- ifp->if_mtu = ETHERMTU;
- ifp->if_flags = 0;
+ ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
ifp->if_ioctl = carp_ioctl;
- ifp->if_output = looutput;
+ ifp->if_output = ether_output;
ifp->if_start = carp_start;
ifp->if_type = IFT_CARP;
- ifp->if_snd.ifq_maxlen = ifqmaxlen;
- ifp->if_hdrlen = 0;
- if_attach(ifp);
+ ifp->if_addrlen = ETHER_ADDR_LEN;
+ ifp->if_hdrlen = ETHER_HDR_LEN;
+ ifp->if_mtu = ETHERMTU;
+ IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
+ IFQ_SET_READY(&ifp->if_snd);
+ if_attachhead(ifp);
+
if_alloc_sadl(ifp);
+ bcopy(&sc->sc_ac.ac_enaddr, LLADDR(ifp->if_sadl), ifp->if_addrlen);
+ LIST_INIT(&sc->sc_ac.ac_multiaddrs);
#if NBPFILTER > 0
- bpfattach(&ifp->if_bpf, ifp, DLT_LOOP, sizeof(u_int32_t));
+ bpfattach(&ifp->if_bpf, ifp, DLT_EN10MB, ETHER_HDR_LEN);
#endif
return (0);
}
@@ -667,44 +735,22 @@
{
struct carp_softc *sc = ifp->if_softc;
struct carp_if *cif;
- struct ip_moptions *imo = &sc->sc_imo;
-#ifdef INET6
- struct ip6_moptions *im6o = &sc->sc_im6o;
-#endif

timeout_del(&sc->sc_ad_tmo);
timeout_del(&sc->sc_md_tmo);
timeout_del(&sc->sc_md6_tmo);

- if (sc->sc_ifp != NULL) {
- cif = (struct carp_if *)sc->sc_ifp->if_carp;
+ if (sc->sc_carpdev != NULL) {
+ cif = (struct carp_if *)sc->sc_carpdev->if_carp;
TAILQ_REMOVE(&cif->vhif_vrs, sc, sc_list);
if (cif->vhif_nvrs) {
- ifpromisc(sc->sc_ifp, 0);

- /* Clear IPv4 multicast */
- in_delmulti(imo->imo_membership[--imo->imo_num_memberships]);
- imo->imo_multicast_ifp = NULL;
-
- /* Clear IPv6 multicast */
-#ifdef INET6
- while (!LIST_EMPTY(&im6o->im6o_memberships)) {
- struct in6_multi_mship *imm =
- LIST_FIRST(&im6o->im6o_memberships);
-
- LIST_REMOVE(imm, i6mm_chain);
- in6_leavegroup(imm);
- }
-#endif
-
- sc->sc_ifp->if_carp = NULL;
+ sc->sc_carpdev->if_carp = NULL;
FREE(cif, M_IFADDR);
}
}

-#if NBPFILTER > 0
- bpfdetach(ifp);
-#endif
+ ether_ifdetach(ifp);
if_detach(ifp);
free(sc, M_DEVBUF);

@@ -714,25 +760,27 @@
void
carpdetach(struct carp_softc *sc)
{
- struct ifaddr *ifa;
+ struct carp_if *cif;

timeout_del(&sc->sc_ad_tmo);
timeout_del(&sc->sc_md_tmo);
timeout_del(&sc->sc_md6_tmo);

- while ((ifa = TAILQ_FIRST(&sc->sc_ac.ac_if.if_addrlist)) != NULL)
- if (ifa->ifa_addr->sa_family == AF_INET) {
- struct in_ifaddr *ia = ifatoia(ifa);
-
- carp_del_addr(sc, &ia->ia_addr);
-
- /* ripped screaming from in_control(SIOCDIFADDR) */
- in_ifscrub(&sc->sc_ac.ac_if, ia);
- TAILQ_REMOVE(&sc->sc_ac.ac_if.if_addrlist,
- ifa, ifa_list);
- TAILQ_REMOVE(&in_ifaddr, ia, ia_list);
- IFAFREE((&ia->ia_ifa));
+ carp_set_state(sc, INIT);
+ sc->sc_ac.ac_if.if_flags &= ~IFF_UP;
+ carp_setrun(sc, 0);
+ carp_multicast_cleanup(sc);
+
+ if (sc->sc_carpdev != NULL) {
+ cif = (struct carp_if *)sc->sc_carpdev->if_carp;
+ TAILQ_REMOVE(&cif->vhif_vrs, sc, sc_list);
+ if (!--cif->vhif_nvrs) {
+ ifpromisc(sc->sc_carpdev, 0);
+ sc->sc_carpdev->if_carp = NULL;
+ FREE(cif, M_IFADDR);
}
+ }
+ sc->sc_carpdev = NULL;
}

/* Detach an interface from the carp. */
@@ -786,7 +834,7 @@
struct carp_softc *vh;

TAILQ_FOREACH(ifp, &ifnet, if_list) {
- if (ifp->if_carp == NULL)
+ if (ifp->if_carp == NULL || ifp->if_type == IFT_CARP)
continue;

cif = (struct carp_if *)ifp->if_carp;
@@ -808,9 +856,16 @@
struct carp_header *ch_ptr;
struct mbuf *m;
int error, len, advbase, advskew, s;
+ struct ifaddr *ifa;
+ struct sockaddr sa;

s = splsoftnet();

+ if (sc->sc_carpdev == NULL) {
+ sc->sc_ac.ac_if.if_oerrors++;
+ goto retry_later;
+ }
+
/* bow out if we've lost our UPness or RUNNINGuiness */
if ((sc->sc_ac.ac_if.if_flags &
(IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING)) {
@@ -837,7 +892,7 @@


#ifdef INET
- if (sc->sc_ia) {
+ if (sc->sc_naddrs) {
struct ip *ip;

MGETHDR(m, M_DONTWAIT, MT_HEADER);
@@ -863,7 +918,15 @@
ip->ip_ttl = CARP_DFLTTL;
ip->ip_p = IPPROTO_CARP;
ip->ip_sum = 0;
- ip->ip_src.s_addr = sc->sc_ia->ia_addr.sin_addr.s_addr;
+
+ bzero(&sa, sizeof(sa));
+ sa.sa_family = AF_INET;
+ ifa = ifaof_ifpforaddr(&sa, sc->sc_carpdev);
+ if (ifa == NULL)
+ ip->ip_src.s_addr = 0;
+ else
+ ip->ip_src.s_addr =
+ ifatoia(ifa)->ia_addr.sin_addr.s_addr;
ip->ip_dst.s_addr = INADDR_CARP_GROUP;

ch_ptr = (void *)ip + sizeof(*ip);
@@ -909,7 +972,7 @@
}
#endif /* INET */
#ifdef INET6
- if (sc->sc_ia6) {
+ if (sc->sc_naddrs6) {
struct ip6_hdr *ip6;

MGETHDR(m, M_DONTWAIT, MT_HEADER);
@@ -930,8 +993,16 @@
ip6->ip6_vfc |= IPV6_VERSION;
ip6->ip6_hlim = CARP_DFLTTL;
ip6->ip6_nxt = IPPROTO_CARP;
- bcopy(&sc->sc_ia6->ia_addr.sin6_addr, &ip6->ip6_src,
- sizeof(struct in6_addr));
+
+ /* set the source address */
+ bzero(&sa, sizeof(sa));
+ sa.sa_family = AF_INET6;
+ ifa = ifaof_ifpforaddr(&sa, sc->sc_carpdev);
+ if (ifa == NULL) /* This should never happen with IPv6 */
+ bzero(&ip6->ip6_src, sizeof(struct in6_addr));
+ else
+ bcopy(ifatoia6(ifa)->ia_addr.sin6_addr.s6_addr,
+ &ip6->ip6_src, sizeof(struct in6_addr));
/* set the multicast destination */

ip6->ip6_dst.s6_addr8[0] = 0xff;
@@ -1004,7 +1075,7 @@
continue;

in = ifatoia(ifa)->ia_addr.sin_addr.s_addr;
- arprequest(sc->sc_ifp, &in, &in, sc->sc_ac.ac_enaddr);
+ arprequest(sc->sc_carpdev, &in, &in, sc->sc_ac.ac_enaddr);
DELAY(1000); /* XXX */
}
splx(s);
@@ -1025,7 +1096,7 @@
continue;

in6 = &ifatoia6(ifa)->ia_addr.sin6_addr;
- nd6_na_output(sc->sc_ifp, &mcast, in6,
+ nd6_na_output(sc->sc_carpdev, &mcast, in6,
ND_NA_FLAG_OVERRIDE, 1, NULL);
DELAY(1000); /* XXX */
}
@@ -1178,23 +1249,67 @@
}
#endif /* INET6 */

-struct ifnet *
-carp_forus(void *v, void *dhost)
+int
+carp_input(struct ether_header *eh, struct mbuf *m)
{
- struct carp_if *cif = v;
+ struct carp_if *cif = (struct carp_if *)m->m_pkthdr.rcvif->if_carp;
struct carp_softc *vh;
- u_int8_t *ena = dhost;
+ u_int8_t *ena = (u_int8_t *)&eh->ether_dhost;
+ struct ifnet *ifp;
+
+ if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
+ struct mbuf *m0;
+
+ /*
+ * XXX Should reall check the list of multicast addresses
+ * for each CARP interface _before_ copying.
+ */
+ TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list) {
+ m0 = m_copym2(m, 0, M_COPYALL, M_DONTWAIT);
+ if (m0 == NULL)
+ continue;
+ m0->m_pkthdr.rcvif = &vh->sc_ac.ac_if;
+ ether_input(&vh->sc_ac.ac_if, eh, m0);
+ }
+ return (1);
+ }

if (ena[0] || ena[1] || ena[2] != 0x5e || ena[3] || ena[4] != 1)
- return (NULL);
+ return (1);

TAILQ_FOREACH(vh, &cif->vhif_vrs, sc_list)
if ((vh->sc_ac.ac_if.if_flags & (IFF_UP|IFF_RUNNING)) ==
(IFF_UP|IFF_RUNNING) && vh->sc_state == MASTER &&
- !bcmp(dhost, vh->sc_ac.ac_enaddr, ETHER_ADDR_LEN))
- return (&vh->sc_ac.ac_if);
+ !bcmp(ena, vh->sc_ac.ac_enaddr,
+ ETHER_ADDR_LEN))
+ break;

- return (NULL);
+ if (vh == NULL)
+ return (1);
+
+ m->m_pkthdr.rcvif = ifp = &vh->sc_ac.ac_if;
+
+#if NBPFILTER > 0
+ if (ifp->if_bpf) {
+ /*
+ * Do the usual BPF fakery. Note that we don't support
+ * promiscuous mode here, since it would require the
+ * drivers to know about CARP and we're not ready for
+ * that yet.
+ */
+ struct mbuf m0;
+
+ m0.m_flags = 0;
+ m0.m_next = m;
+ m0.m_len = ETHER_HDR_LEN;
+ m0.m_data = (char *)eh;
+ bpf_mtap(ifp->if_bpf, &m0);
+ }
+#endif
+ ifp->if_ipackets++;
+ ether_input(ifp, eh, m);
+
+ return (0);
}

void
@@ -1231,10 +1346,16 @@
{
struct timeval tv;

+ if (sc->sc_carpdev == NULL) {
+ sc->sc_ac.ac_if.if_flags &= ~IFF_RUNNING;
+ carp_set_state(sc, INIT);
+ return;
+ }
+
if (sc->sc_ac.ac_if.if_flags & IFF_UP &&
- sc->sc_vhid > 0 && (sc->sc_naddrs || sc->sc_naddrs6))
+ sc->sc_vhid > 0 && (sc->sc_naddrs || sc->sc_naddrs6)) {
sc->sc_ac.ac_if.if_flags |= IFF_RUNNING;
- else {
+ } else {
sc->sc_ac.ac_if.if_flags &= ~IFF_RUNNING;
carp_setroute(sc, RTM_DELETE);
return;
@@ -1287,15 +1408,120 @@
}
}

+void
+carp_multicast_cleanup(struct carp_softc *sc) {
+ struct ip_moptions *imo = &sc->sc_imo;
+ struct ip6_moptions *im6o = &sc->sc_im6o;
+
+ /* Clean up our own multicast memberships */
+ while (imo->imo_num_memberships > 0) {
+ if (imo->imo_membership[--imo->imo_num_memberships] == NULL)
+ in_delmulti(imo->imo_membership[imo->imo_num_memberships]);
+ imo->imo_membership[imo->imo_num_memberships] = NULL;
+ }
+ imo->imo_multicast_ifp = NULL;
+
+ while (!LIST_EMPTY(&im6o->im6o_memberships)) {
+ struct in6_multi_mship *imm =
+ LIST_FIRST(&im6o->im6o_memberships);
+
+ LIST_REMOVE(imm, i6mm_chain);
+ in6_leavegroup(imm);
+ }
+ im6o->im6o_multicast_ifp = NULL;
+
+ /* And any other multicast memberships */
+ carp_ether_purgemulti(sc);
+}
+
+int
+carp_set_ifp(struct carp_softc *sc, struct ifnet *ifp)
+{
+ struct carp_if *cif, *ncif = NULL;
+ struct carp_softc *vr, *after = NULL;
+ int myself = 0, error = 0;
+
+ if (ifp == sc->sc_carpdev)
+ return (0);
+
+ if (ifp != NULL) {
+ if ((ifp->if_flags & IFF_MULTICAST) == 0)
+ return (EADDRNOTAVAIL);
+
+ if (ifp->if_carp == NULL) {
+ MALLOC(ncif, struct carp_if *, sizeof(*cif),
+ M_IFADDR, M_WAITOK);
+ if (!ncif)
+ return (ENOBUFS);
+ if ((error = ifpromisc(ifp, 1))) {
+ FREE(ncif, M_IFADDR);
+ return (error);
+ }
+
+ ncif->vhif_ifp = ifp;
+ TAILQ_INIT(&ncif->vhif_vrs);
+ } else {
+ cif = (struct carp_if *)ifp->if_carp;
+ TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list)
+ if (vr != sc && vr->sc_vhid == sc->sc_vhid)
+ return (EINVAL);
+ }
+
+ /* detach from old interface */
+ if (sc->sc_carpdev != NULL)
+ carpdetach(sc);
+
+ /* join multicast groups */
+ if (sc->sc_naddrs < 0 &&
+ (error = carp_join_multicast(sc, ifp)) != 0) {
+ FREE(ncif, M_IFADDR);
+ return (error);
+ }
+
+ if (sc->sc_naddrs6 < 0 &&
+ (error = carp_join_multicast(sc, ifp)) != 0) {
+ FREE(ncif, M_IFADDR);
+ return (error);
+ }
+
+ /* attach carp interface to physical interface */
+ if (ncif != NULL)
+ ifp->if_carp = (caddr_t)ncif;
+ sc->sc_carpdev = ifp;
+ cif = (struct carp_if *)ifp->if_carp;
+ TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list) {
+ if (vr == sc)
+ myself = 1;
+ if (vr->sc_vhid < sc->sc_vhid)
+ after = vr;
+ }
+
+ if (!myself) {
+ /* We're trying to keep things in order */
+ if (after == NULL) {
+ TAILQ_INSERT_TAIL(&cif->vhif_vrs, sc, sc_list);
+ } else {
+ TAILQ_INSERT_AFTER(&cif->vhif_vrs, after,
+ sc, sc_list);
+ }
+ cif->vhif_nvrs++;
+ }
+ if (sc->sc_naddrs || sc->sc_naddrs6)
+ sc->sc_ac.ac_if.if_flags |= IFF_UP;
+ } else {
+ carpdetach(sc);
+ sc->sc_ac.ac_if.if_flags &= ~(IFF_UP|IFF_RUNNING);
+ }
+ return (0);
+
+}
+
int
carp_set_addr(struct carp_softc *sc, struct sockaddr_in *sin)
{
- struct ifnet *ifp;
- struct carp_if *cif;
+ struct ifnet *ifp = sc->sc_carpdev;
struct in_ifaddr *ia, *ia_if;
- struct ip_moptions *imo = &sc->sc_imo;
- struct in_addr addr;
- int own, error;
+ int own, error = 0;

if (sin->sin_addr.s_addr == 0) {
if (!(sc->sc_ac.ac_if.if_flags & IFF_UP))
@@ -1308,7 +1534,8 @@

/* we have to do it by hands to check we won't match on us */
ia_if = NULL; own = 0;
- for (ia = TAILQ_FIRST(&in_ifaddr); ia; ia = TAILQ_NEXT(ia, ia_list)) {
+ for (ia = TAILQ_FIRST(&in_ifaddr); ia;
+ ia = TAILQ_NEXT(ia, ia_list)) {

/* and, yeah, we need a multicast-capable iface too */
if (ia->ia_ifp != &sc->sc_ac.ac_if &&
@@ -1322,127 +1549,66 @@
}
}

- if (!ia_if)
- return (EADDRNOTAVAIL);
- ia = ia_if;
- ifp = ia->ia_ifp;
-
- if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0 ||
- (imo->imo_multicast_ifp && imo->imo_multicast_ifp != ifp))
- return (EADDRNOTAVAIL);
-
- if (imo->imo_num_memberships == 0) {
- addr.s_addr = INADDR_CARP_GROUP;
- if ((imo->imo_membership[0] = in_addmulti(&addr, ifp)) == NULL)
- return (ENOBUFS);
- imo->imo_num_memberships++;
- imo->imo_multicast_ifp = ifp;
- imo->imo_multicast_ttl = CARP_DFLTTL;
- imo->imo_multicast_loop = 0;
- }
-
- if (!ifp->if_carp) {
-
- MALLOC(cif, struct carp_if *, sizeof(*cif), M_IFADDR, M_WAITOK);
- if (!cif) {
- error = ENOBUFS;
- goto cleanup;
- }
- if ((error = ifpromisc(ifp, 1))) {
- FREE(cif, M_IFADDR);
- goto cleanup;
- }
-
- bzero(cif, sizeof(*cif));
- cif->vhif_ifp = ifp;
- TAILQ_INIT(&cif->vhif_vrs);
- ifp->if_carp = (caddr_t)cif;
-
- } else {
- struct carp_softc *vr;
+ if (ia_if) {
+ ia = ia_if;
+ if (ifp) {
+ if (ifp != ia->ia_ifp)
+ return (EADDRNOTAVAIL);
+ } else {
+ ifp = ia->ia_ifp;
+ }
+ }

- cif = (struct carp_if *)ifp->if_carp;
- TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list)
- if (vr != sc && vr->sc_vhid == sc->sc_vhid) {
- error = EINVAL;
- goto cleanup;
- }
- }
- sc->sc_ia = ia;
- sc->sc_ifp = ifp;
+ if ((error = carp_set_ifp(sc, ifp)))
+ return (error);

- { /* XXX prevent endless loop if already in queue */
- struct carp_softc *vr, *after = NULL;
- int myself = 0;
- cif = (struct carp_if *)ifp->if_carp;
-
- TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list) {
- if (vr == sc)
- myself = 1;
- if (vr->sc_vhid < sc->sc_vhid)
- after = vr;
- }
+ if (sc->sc_carpdev == NULL)
+ return (EADDRNOTAVAIL);

- if (!myself) {
- /* We're trying to keep things in order */
- if (after == NULL) {
- TAILQ_INSERT_TAIL(&cif->vhif_vrs, sc, sc_list);
- } else {
- TAILQ_INSERT_AFTER(&cif->vhif_vrs, after, sc, sc_list);
- }
- cif->vhif_nvrs++;
- }
- }
+ if (sc->sc_naddrs == 0 && (error = carp_join_multicast(sc, ifp)) != 0)
+ return (error);

sc->sc_naddrs++;
- sc->sc_ac.ac_if.if_flags |= IFF_UP;
+ if (sc->sc_carpdev != NULL)
+ sc->sc_ac.ac_if.if_flags |= IFF_UP;
+
if (own)
sc->sc_advskew = 0;
carp_set_state(sc, INIT);
carp_setrun(sc, 0);

return (0);
-
-cleanup:
- in_delmulti(imo->imo_membership[--imo->imo_num_memberships]);
- return (error);
}

int
-carp_del_addr(struct carp_softc *sc, struct sockaddr_in *sin)
+carp_join_multicast(struct carp_softc *sc, struct ifnet *ifp)
{
- int error = 0;
-
- if (!--sc->sc_naddrs) {
- struct carp_if *cif = (struct carp_if *)sc->sc_ifp->if_carp;
- struct ip_moptions *imo = &sc->sc_imo;
+ struct ip_moptions *imo = &sc->sc_imo, tmpimo;
+ struct in_addr addr;

- timeout_del(&sc->sc_ad_tmo);
- sc->sc_ac.ac_if.if_flags &= ~(IFF_UP|IFF_RUNNING);
- sc->sc_vhid = -1;
- in_delmulti(imo->imo_membership[--imo->imo_num_memberships]);
- imo->imo_multicast_ifp = NULL;
- TAILQ_REMOVE(&cif->vhif_vrs, sc, sc_list);
- if (!--cif->vhif_nvrs) {
- sc->sc_ifp->if_carp = NULL;
- FREE(cif, M_IFADDR);
- }
+ bzero(&tmpimo, sizeof(tmpimo));
+ addr.s_addr = INADDR_CARP_GROUP;
+ if ((tmpimo.imo_membership[0] =
+ in_addmulti(&addr, &sc->sc_ac.ac_if)) == NULL) {
+ return (ENOBUFS);
}

- return (error);
+ imo->imo_membership[0] = tmpimo.imo_membership[0];
+ imo->imo_num_memberships = 1;
+ imo->imo_multicast_ifp = &sc->sc_ac.ac_if;
+ imo->imo_multicast_ttl = CARP_DFLTTL;
+ imo->imo_multicast_loop = 0;
+ return (0);
}

+
#ifdef INET6
int
carp_set_addr6(struct carp_softc *sc, struct sockaddr_in6 *sin6)
{
- struct ifnet *ifp;
- struct carp_if *cif;
+ struct ifnet *ifp = sc->sc_carpdev;
struct in6_ifaddr *ia, *ia_if;
- struct ip6_moptions *im6o = &sc->sc_im6o;
- struct in6_multi_mship *imm;
- struct sockaddr_in6 addr;
- int own, error;
+ int own, error = 0;

if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
if (!(sc->sc_ac.ac_if.if_flags & IFF_UP))
@@ -1477,143 +1643,84 @@
}
}

- if (!ia_if)
- return (EADDRNOTAVAIL);
- ia = ia_if;
- ifp = ia->ia_ifp;
-
- if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0 ||
- (im6o->im6o_multicast_ifp && im6o->im6o_multicast_ifp != ifp))
- return (EADDRNOTAVAIL);
-
- if (!sc->sc_naddrs6) {
- im6o->im6o_multicast_ifp = ifp;
-
- /* join CARP multicast address */
- bzero(&addr, sizeof(addr));
- addr.sin6_family = AF_INET6;
- addr.sin6_len = sizeof(addr);
- addr.sin6_addr.s6_addr16[0] = htons(0xff02);
- addr.sin6_addr.s6_addr16[1] = htons(ifp->if_index);
- addr.sin6_addr.s6_addr8[15] = 0x12;
- if ((imm = in6_joingroup(ifp, &addr.sin6_addr, &error)) == NULL)
- goto cleanup;
- LIST_INSERT_HEAD(&im6o->im6o_memberships, imm, i6mm_chain);
-
- /* join solicited multicast address */
- bzero(&addr.sin6_addr, sizeof(addr.sin6_addr));
- addr.sin6_addr.s6_addr16[0] = htons(0xff02);
- addr.sin6_addr.s6_addr16[1] = htons(ifp->if_index);
- addr.sin6_addr.s6_addr32[1] = 0;
- addr.sin6_addr.s6_addr32[2] = htonl(1);
- addr.sin6_addr.s6_addr32[3] = sin6->sin6_addr.s6_addr32[3];
- addr.sin6_addr.s6_addr8[12] = 0xff;
- if ((imm = in6_joingroup(ifp, &addr.sin6_addr, &error)) == NULL)
- goto cleanup;
- LIST_INSERT_HEAD(&im6o->im6o_memberships, imm, i6mm_chain);
- }
-
- if (!ifp->if_carp) {
- MALLOC(cif, struct carp_if *, sizeof(*cif), M_IFADDR, M_WAITOK);
- if (!cif) {
- error = ENOBUFS;
- goto cleanup;
- }
- if ((error = ifpromisc(ifp, 1))) {
- FREE(cif, M_IFADDR);
- goto cleanup;
+ if (ia_if) {
+ ia = ia_if;
+ if (sc->sc_carpdev) {
+ if (sc->sc_carpdev != ia->ia_ifp)
+ return (EADDRNOTAVAIL);
+ } else {
+ ifp = ia->ia_ifp;
}
-
- bzero(cif, sizeof(*cif));
- cif->vhif_ifp = ifp;
- TAILQ_INIT(&cif->vhif_vrs);
- ifp->if_carp = (caddr_t)cif;
-
- } else {
- struct carp_softc *vr;
-
- cif = (struct carp_if *)ifp->if_carp;
- TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list)
- if (vr != sc && vr->sc_vhid == sc->sc_vhid) {
- error = EINVAL;
- goto cleanup;
- }
}
- sc->sc_ia6 = ia;
- sc->sc_ifp = ifp;

- { /* XXX prevent endless loop if already in queue */
- struct carp_softc *vr, *after = NULL;
- int myself = 0;
- cif = (struct carp_if *)ifp->if_carp;
+ if ((error = carp_set_ifp(sc, ifp)))
+ return (error);

- TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list) {
- if (vr == sc)
- myself = 1;
- if (vr->sc_vhid < sc->sc_vhid)
- after = vr;
- }
+ if (sc->sc_carpdev == NULL)
+ return (EADDRNOTAVAIL);

- if (!myself) {
- /* We're trying to keep things in order */
- if (after == NULL) {
- TAILQ_INSERT_TAIL(&cif->vhif_vrs, sc, sc_list);
- } else {
- TAILQ_INSERT_AFTER(&cif->vhif_vrs, after, sc, sc_list);
- }
- cif->vhif_nvrs++;
- }
- }
+ if (sc->sc_naddrs6 == 0 && (error = carp_join_multicast6(sc, ifp)) != 0)
+ return (error);

sc->sc_naddrs6++;
- sc->sc_ac.ac_if.if_flags |= IFF_UP;
- if (own)
- sc->sc_advskew = 0;
+ if (sc->sc_carpdev != NULL)
+ sc->sc_ac.ac_if.if_flags |= IFF_UP;
carp_set_state(sc, INIT);
carp_setrun(sc, 0);

return (0);
-
-cleanup:
- /* clean up multicast memberships */
- if (!sc->sc_naddrs6) {
- while (!LIST_EMPTY(&im6o->im6o_memberships)) {
- imm = LIST_FIRST(&im6o->im6o_memberships);
- LIST_REMOVE(imm, i6mm_chain);
- in6_leavegroup(imm);
- }
- }
- return (error);
}

int
-carp_del_addr6(struct carp_softc *sc, struct sockaddr_in6 *sin6)
+carp_join_multicast6(struct carp_softc *sc, struct ifnet *ifp)
{
- int error = 0;
-
- if (!--sc->sc_naddrs6) {
- struct carp_if *cif = (struct carp_if *)sc->sc_ifp->if_carp;
- struct ip6_moptions *im6o = &sc->sc_im6o;
+ struct in6_multi_mship *imm, *imm2;
+ struct ip6_moptions *im6o = &sc->sc_im6o;
+ struct sockaddr_in6 addr6;
+ int error;
+
+ /*
+ * For IPv6, we can attach to the physical interface, as
+ * there will be a link-local address there.
+ * This way we don't need a link-local address on the
+ * CARP interface.
+ */

- timeout_del(&sc->sc_ad_tmo);
- sc->sc_ac.ac_if.if_flags &= ~(IFF_UP|IFF_RUNNING);
- sc->sc_vhid = -1;
- while (!LIST_EMPTY(&im6o->im6o_memberships)) {
- struct in6_multi_mship *imm =
- LIST_FIRST(&im6o->im6o_memberships);
+ /* Join IPv6 CARP multicast group */
+ bzero(&addr6, sizeof(addr6));
+ addr6.sin6_family = AF_INET6;
+ addr6.sin6_len = sizeof(addr6);
+ addr6.sin6_addr.s6_addr16[0] = htons(0xff02);
+ addr6.sin6_addr.s6_addr16[1] = htons(ifp->if_index);
+ addr6.sin6_addr.s6_addr8[15] = 0x12;
+ if ((imm = in6_joingroup(ifp,
+ &addr6.sin6_addr, &error)) == NULL) {
+ return (error);
+ }
+ /* join solicited multicast address */
+ bzero(&addr6.sin6_addr, sizeof(addr6.sin6_addr));
+ addr6.sin6_addr.s6_addr16[0] = htons(0xff02);
+ addr6.sin6_addr.s6_addr16[1] = htons(ifp->if_index);
+ addr6.sin6_addr.s6_addr32[1] = 0;
+ addr6.sin6_addr.s6_addr32[2] = htonl(1);
+ addr6.sin6_addr.s6_addr32[3] = 0;
+ addr6.sin6_addr.s6_addr8[12] = 0xff;
+ if ((imm2 = in6_joingroup(ifp,
+ &addr6.sin6_addr, &error)) == NULL) {
+ in6_leavegroup(imm);
+ return (error);
+ }
+
+ /* apply v6 multicast membership */
+ im6o->im6o_multicast_ifp = sc->sc_carpdev;
+ if (imm)
+ LIST_INSERT_HEAD(&im6o->im6o_memberships, imm,
+ i6mm_chain);
+ if (imm2)
+ LIST_INSERT_HEAD(&im6o->im6o_memberships, imm2,
+ i6mm_chain);

- LIST_REMOVE(imm, i6mm_chain);
- in6_leavegroup(imm);
- }
- im6o->im6o_multicast_ifp = NULL;
- TAILQ_REMOVE(&cif->vhif_vrs, sc, sc_list);
- if (!--cif->vhif_nvrs) {
- sc->sc_ifp->if_carp = NULL;
- FREE(cif, M_IFADDR);
- }
- }
-
- return (error);
+ return (0);
}

#endif /* INET6 */
@@ -1627,6 +1734,7 @@
struct ifaddr *ifa;
struct ifreq *ifr;
struct ifaliasreq *ifra;
+ struct ifnet *cdev = NULL;
int error = 0;

ifa = (struct ifaddr *)addr;
@@ -1683,12 +1791,12 @@
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
- error = carp_del_addr(sc, satosin(&ifra->ifra_addr));
+ sc->sc_naddrs--;
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
- error = carp_del_addr6(sc, satosin6(&ifra->ifra_addr));
+ sc->sc_naddrs6--;
break;
#endif /* INET6 */
default:
@@ -1719,6 +1827,11 @@
if ((error = copyin(ifr->ifr_data, &carpr, sizeof carpr)))
break;
error = 1;
+ if (carpr.carpr_carpdev[0] != '\0' &&
+ (cdev = ifunit(carpr.carpr_carpdev)) == NULL)
+ return (EINVAL);
+ if ((error = carp_set_ifp(sc, cdev)))
+ return (error);
if (sc->sc_state != INIT && carpr.carpr_state != sc->sc_state) {
switch (carpr.carpr_state) {
case BACKUP:
@@ -1739,9 +1852,9 @@
error = EINVAL;
break;
}
- if (sc->sc_ifp) {
+ if (sc->sc_carpdev) {
struct carp_if *cif;
- cif = (struct carp_if *)sc->sc_ifp->if_carp;
+ cif = (struct carp_if *)sc->sc_carpdev->if_carp;
TAILQ_FOREACH(vr, &cif->vhif_vrs, sc_list)
if (vr != sc &&
vr->sc_vhid == carpr.carpr_vhid)
@@ -1754,6 +1867,8 @@
sc->sc_ac.ac_enaddr[3] = 0;
sc->sc_ac.ac_enaddr[4] = 1;
sc->sc_ac.ac_enaddr[5] = sc->sc_vhid;
+ bcopy(&sc->sc_ac.ac_enaddr,
+ LLADDR(ifp->if_sadl), ifp->if_addrlen);
error--;
}
if (carpr.carpr_advbase > 0 || carpr.carpr_advskew > 0) {
@@ -1780,6 +1895,9 @@

case SIOCGVH:
bzero(&carpr, sizeof(carpr));
+ if (sc->sc_carpdev != NULL)
+ strlcpy(carpr.carpr_carpdev, sc->sc_carpdev->if_xname,
+ IFNAMSIZ);
carpr.carpr_state = sc->sc_state;
carpr.carpr_vhid = sc->sc_vhid;
carpr.carpr_advbase = sc->sc_advbase;
@@ -1790,6 +1908,14 @@
error = copyout(&carpr, ifr->ifr_data, sizeof(carpr));
break;

+ case SIOCADDMULTI:
+ error = carp_ether_addmulti(sc, ifr);
+ break;
+
+ case SIOCDELMULTI:
+ error = carp_ether_delmulti(sc, ifr);
+ break;
+
default:
error = EINVAL;
}
@@ -1841,7 +1967,7 @@
sc = carp_ifp->if_softc;

/* Set the source MAC address to Virtual Router MAC Address */
- switch (ifp->if_type) {
+ switch (sc->sc_carpdev->if_type) {
#if NETHER > 0
case IFT_ETHER: {
struct ether_header *eh;
@@ -1886,7 +2012,7 @@
#endif
default:
printf("%s: carp is not supported for this interface type\n",
- ifp->if_xname);
+ sc->sc_carpdev->if_xname);
return (EOPNOTSUPP);
}

@@ -1921,8 +2047,8 @@
struct carp_softc *sc;

TAILQ_FOREACH(sc, &cif->vhif_vrs, sc_list) {
- if (sc->sc_ifp->if_link_state == LINK_STATE_DOWN ||
- !(sc->sc_ifp->if_flags & IFF_UP)) {
+ if (sc->sc_carpdev->if_link_state == LINK_STATE_DOWN ||
+ !(sc->sc_carpdev->if_flags & IFF_UP)) {
sc->sc_flags_backup = sc->sc_ac.ac_if.if_flags;
sc->sc_ac.ac_if.if_flags &= ~(IFF_UP|IFF_RUNNING);
timeout_del(&sc->sc_ad_tmo);
@@ -1944,5 +2070,131 @@
carp_suppress_preempt--;
sc->sc_suppress = 0;
}
+ }
+}
+
+
+int
+carp_ether_addmulti(struct carp_softc *sc, struct ifreq *ifr)
+{
+ struct ifnet *ifp;
+ struct carp_mc_entry *mc;
+ u_int8_t addrlo[ETHER_ADDR_LEN], addrhi[ETHER_ADDR_LEN];
+ int error;
+
+ ifp = sc->sc_carpdev;
+ if (ifp == NULL)
+ return (EINVAL);
+
+ error = ether_addmulti(ifr, (struct arpcom *)&sc->sc_ac);
+ if (error != ENETRESET)
+ return (error);
+
+ /*
+ * This is new multicast address. We have to tell parent
+ * about it. Also, remember this multicast address so that
+ * we can delete them on unconfigure.
+ */
+ MALLOC(mc, struct carp_mc_entry *, sizeof(struct carp_mc_entry),
+ M_DEVBUF, M_NOWAIT);
+ if (mc == NULL) {
+ error = ENOMEM;
+ goto alloc_failed;
+ }
+
+ /*
+ * As ether_addmulti() returns ENETRESET, following two
+ * statement shouldn't fail.
+ */
+ (void)ether_multiaddr(&ifr->ifr_addr, addrlo, addrhi);
+ ETHER_LOOKUP_MULTI(addrlo, addrhi, &sc->sc_ac, mc->mc_enm);
+ memcpy(&mc->mc_addr, &ifr->ifr_addr, ifr->ifr_addr.sa_len);
+ LIST_INSERT_HEAD(&sc->carp_mc_listhead, mc, mc_entries);
+
+ error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)ifr);
+ if (error != 0)
+ goto ioctl_failed;
+
+ return (error);
+
+ ioctl_failed:
+ LIST_REMOVE(mc, mc_entries);
+ FREE(mc, M_DEVBUF);
+ alloc_failed:
+ (void)ether_delmulti(ifr, (struct arpcom *)&sc->sc_ac);
+
+ return (error);
+}
+
+int
+carp_ether_delmulti(struct carp_softc *sc, struct ifreq *ifr)
+{
+ struct ifnet *ifp;
+ struct ether_multi *enm;
+ struct carp_mc_entry *mc;
+ u_int8_t addrlo[ETHER_ADDR_LEN], addrhi[ETHER_ADDR_LEN];
+ int error;
+
+ ifp = sc->sc_carpdev;
+ if (ifp == NULL)
+ return (EINVAL);
+
+ /*
+ * Find a key to lookup carp_mc_entry. We have to do this
+ * before calling ether_delmulti for obvious reason.
+ */
+ if ((error = ether_multiaddr(&ifr->ifr_addr, addrlo, addrhi)) != 0)
+ return (error);
+ ETHER_LOOKUP_MULTI(addrlo, addrhi, &sc->sc_ac, enm);
+
+ error = ether_delmulti(ifr, (struct arpcom *)&sc->sc_ac);
+ if (error != ENETRESET)
+ return (error);
+
+ /* We no longer use this multicast address. Tell parent so. */
+ error = (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)ifr);
+ if (error == 0) {
+ /* And forget about this address. */
+ for (mc = LIST_FIRST(&sc->carp_mc_listhead); mc != NULL;
+ mc = LIST_NEXT(mc, mc_entries)) {
+ if (mc->mc_enm == enm) {
+ LIST_REMOVE(mc, mc_entries);
+ FREE(mc, M_DEVBUF);
+ break;
+ }
+ }
+ KASSERT(mc != NULL);
+ } else
+ (void)ether_addmulti(ifr, (struct arpcom *)&sc->sc_ac);
+ return (error);
+}
+
+/*
+ * Delete any multicast address we have asked to add from parent
+ * interface. Called when the carp is being unconfigured.
+ */
+void
+carp_ether_purgemulti(struct carp_softc *sc)
+{
+ struct ifnet *ifp = sc->sc_carpdev; /* Parent. */
+ struct carp_mc_entry *mc;
+ union {
+ struct ifreq ifreq;
+ struct {
+ char ifr_name[IFNAMSIZ];
+ struct sockaddr_storage ifr_ss;
+ } ifreq_storage;
+ } ifreq;
+ struct ifreq *ifr = &ifreq.ifreq;
+
+ if (ifp == NULL)
+ return;
+
+ memcpy(ifr->ifr_name, ifp->if_xname, IFNAMSIZ);
+ while ((mc = LIST_FIRST(&sc->carp_mc_listhead)) != NULL) {
+ memcpy(&ifr->ifr_addr, &mc->mc_addr, mc->mc_addr.ss_len);
+ (void)(*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)ifr);
+ LIST_REMOVE(mc, mc_entries);
+ FREE(mc, M_DEVBUF);
}
}
Index: sys/netinet/ip_carp.h
================================================== =================
RCS file: /cvs/src/sys/netinet/ip_carp.h,v
retrieving revision 1.8
diff -u -r1.8 ip_carp.h
--- sys/netinet/ip_carp.h 29 Jul 2004 22:12:15 -0000 1.8
+++ sys/netinet/ip_carp.h 6 Dec 2004 20:41:12 -0000
@@ -109,6 +109,13 @@
u_int64_t carps_preempt; /* if enabled, preemptions */
};

+#define CARPDEVNAMSIZ 16
+#ifdef IFNAMSIZ
+#if CARPDEVNAMSIZ != IFNAMSIZ
+#error
+#endif
+#endif
+
/*
* Configuration structure for SIOCSVH SIOCGVH
*/
@@ -116,6 +123,8 @@
int carpr_state;
#define CARP_STATES "INIT", "BACKUP", "MASTER"
#define CARP_MAXSTATE 2
+
+ char carpr_carpdev[CARPDEVNAMSIZ];
int carpr_vhid;
int carpr_advskew;
int carpr_advbase;
@@ -143,15 +152,15 @@

#ifdef _KERNEL
void carp_ifdetach (struct ifnet *);
-void carp_input (struct mbuf *, ...);
+void carp_proto_input (struct mbuf *, ...);
void carp_carpdev_state(void *);
-int carp6_input (struct mbuf **, int *, int);
-int carp_output (struct ifnet *, struct mbuf *, struct sockaddr *,
+int carp6_proto_input(struct mbuf **, int *, int);
+int carp_output(struct ifnet *, struct mbuf *, struct sockaddr *,
struct rtentry *);
-int carp_iamatch (void *, struct in_ifaddr *, struct in_addr *,
+int carp_iamatch(void *, struct in_ifaddr *, struct in_addr *,
u_int8_t **);
struct ifaddr *carp_iamatch6(void *, struct in6_addr *);
void *carp_macmatch6(void *, struct mbuf *, struct in6_addr *);
-struct ifnet *carp_forus (void *, void *);
-int carp_sysctl (int *, u_int, void *, size_t *, void *, size_t);
+int carp_input(struct ether_header *, struct mbuf *);
+int carp_sysctl(int *, u_int, void *, size_t *, void *, size_t);
#endif
Index: sys/netinet/in_proto.c
================================================== =================
RCS file: /cvs/src/sys/netinet/in_proto.c,v
retrieving revision 1.41
diff -u -r1.41 in_proto.c
--- sys/netinet/in_proto.c 17 Sep 2004 11:32:53 -0000 1.41
+++ sys/netinet/in_proto.c 20 Nov 2004 17:59:21 -0000
@@ -292,7 +292,7 @@
#endif /* NGRE > 0 */
#if NCARP > 0
{ SOCK_RAW, &inetdomain, IPPROTO_CARP, PR_ATOMIC|PR_ADDR,
- carp_input, rip_output, 0, rip_ctloutput,
+ carp_proto_input, rip_output, 0, rip_ctloutput,
rip_usrreq,
0, 0, 0, 0, carp_sysctl
},
Index: sys/netinet6/in6_proto.c
================================================== =================
RCS file: /cvs/src/sys/netinet6/in6_proto.c,v
retrieving revision 1.45
diff -u -r1.45 in6_proto.c
--- sys/netinet6/in6_proto.c 18 Oct 2004 03:59:33 -0000 1.45
+++ sys/netinet6/in6_proto.c 20 Nov 2004 17:59:21 -0000
@@ -220,7 +220,7 @@
},
#if NCARP > 0
{ SOCK_RAW, &inet6domain, IPPROTO_CARP, PR_ATOMIC|PR_ADDR,
- carp6_input, rip6_output, 0, rip6_ctloutput,
+ carp6_proto_input, rip6_output, 0, rip6_ctloutput,
rip6_usrreq,
0, 0, 0, 0, carp_sysctl
},
Index: sys/netinet6/in6_ifattach.c
================================================== =================
RCS file: /cvs/src/sys/netinet6/in6_ifattach.c,v
retrieving revision 1.36
diff -u -r1.36 in6_ifattach.c
--- sys/netinet6/in6_ifattach.c 7 May 2004 14:42:27 -0000 1.36
+++ sys/netinet6/in6_ifattach.c 20 Nov 2004 17:59:21 -0000
@@ -577,6 +577,7 @@
case IFT_ENC:
case IFT_PFLOG:
case IFT_PFSYNC:
+ case IFT_CARP:
return;
case IFT_PROPVIRTUAL:
if (strncmp("bridge", ifp->if_xname, sizeof("bridge")) == 0 &&
Index: sbin/ifconfig/ifconfig.c
================================================== =================
RCS file: /cvs/src/sbin/ifconfig/ifconfig.c,v
retrieving revision 1.121
diff -u -r1.121 ifconfig.c
--- sbin/ifconfig/ifconfig.c 1 Dec 2004 15:57:44 -0000 1.121
+++ sbin/ifconfig/ifconfig.c 6 Dec 2004 20:07:53 -0000
@@ -203,6 +203,8 @@
void setcarp_passwd(const char *, int);
void setcarp_vhid(const char *, int);
void setcarp_state(const char *, int);
+void setcarpdev(const char *, int);
+void unsetcarpdev(const char *, int);
void setpfsync_syncif(const char *, int);
void setpfsync_maxupd(const char *, int);
void unsetpfsync_syncif(const char *, int);
@@ -304,6 +306,8 @@
{ "pass", NEXTARG, 0, setcarp_passwd },
{ "vhid", NEXTARG, 0, setcarp_vhid },
{ "state", NEXTARG, 0, setcarp_state },
+ { "carpdev", NEXTARG, 0, setcarpdev },
+ { "-carpdev", 1, 0, unsetcarpdev },
{ "syncif", NEXTARG, 0, setpfsync_syncif },
{ "-syncif", 1, 0, unsetpfsync_syncif },
{ "syncpeer", NEXTARG, 0, setpfsync_syncpeer },
@@ -2875,9 +2879,10 @@
else
state = carp_states[carpr.carpr_state];

- printf("\tcarp: %s vhid %d advbase %d advskew %d\n",
- state, carpr.carpr_vhid, carpr.carpr_advbase,
- carpr.carpr_advskew);
+ printf("\tcarp: %s carpdev %s vhid %d advbase %d advskew %d\n",
+ state, carpr.carpr_carpdev[0] != '\0' ?
+ carpr.carpr_carpdev : "none", carpr.carpr_vhid,
+ carpr.carpr_advbase, carpr.carpr_advskew);
}
}

@@ -2997,6 +3002,40 @@
}

/* ARGSUSED */
+void
+setcarpdev(const char *val, int d)
+{
+ struct carpreq carpr;
+
+ bzero((char *)&carpr, sizeof(struct carpreq));
+ ifr.ifr_data = (caddr_t)&carpr;
+
+ if (ioctl(s, SIOCGVH, (caddr_t)&ifr) == -1)
+ err(1, "SIOCGVH");
+
+ strlcpy(carpr.carpr_carpdev, val, sizeof(carpr.carpr_carpdev));
+
+ if (ioctl(s, SIOCSVH, (caddr_t)&ifr) == -1)
+ err(1, "SIOCSVH");
+}
+
+void
+unsetcarpdev(const char *val, int d)
+{
+ struct carpreq carpr;
+
+ bzero((char *)&carpr, sizeof(struct carpreq));
+ ifr.ifr_data = (caddr_t)&carpr;
+
+ if (ioctl(s, SIOCGVH, (caddr_t)&ifr) == -1)
+ err(1, "SIOCGVH");
+
+ bzero((char *)&carpr.carpr_carpdev, sizeof(carpr.carpr_carpdev));
+
+ if (ioctl(s, SIOCSVH, (caddr_t)&ifr) == -1)
+ err(1, "SIOCSVH");
+}
+
void
setpfsync_syncif(const char *val, int d)
{
Index: sys/net/if.c
================================================== =================
RCS file: /cvs/src/sys/net/if.c,v
retrieving revision 1.95
diff -u -r1.95 if.c
--- sys/net/if.c 4 Dec 2004 16:16:45 -0000 1.95
+++ sys/net/if.c 5 Dec 2004 03:31:36 -0000
@@ -79,6 +79,7 @@

#include <net/if.h>
#include <net/if_dl.h>
+#include <net/if_types.h>
#include <net/route.h>

#ifdef INET
@@ -485,7 +486,7 @@

#if NCARP > 0
/* Remove the interface from any carp group it is a part of. */
- if (ifp->if_carp)
+ if (ifp->if_carp && ifp->if_type != IFT_CARP)
carp_ifdetach(ifp);
#endif

Index: sys/net/if.h
================================================== =================
RCS file: /cvs/src/sys/net/if.h,v
retrieving revision 1.60
diff -u -r1.60 if.h
--- sys/net/if.h 3 Dec 2004 17:31:03 -0000 1.60
+++ sys/net/if.h 6 Dec 2004 21:20:53 -0000
@@ -180,7 +180,12 @@
int if_pcount; /* number of promiscuous listeners */
caddr_t if_bpf; /* packet filter structure */
caddr_t if_bridge; /* bridge structure */
- caddr_t if_carp; /* carp structure */
+ union {
+ caddr_t carp_s; /* carp structure (used by !carp ifs) */
+ struct ifnet *carp_d; /* ptr to carpdev (used by carp ifs) */
+ } if_carp_ptr;
+#define if_carp if_carp_ptr.carp_s
+#define if_carpdev if_carp_ptr.carp_d
u_short if_index; /* numeric abbreviation for this if */
short if_timer; /* time 'til if_watchdog called */
short if_flags; /* up/down, broadcast, etc. */