In ipv4 a call to the setsockopt() function with the optname set to
IP_ADD_MEMBERSHIP and a raw ip socket adds the socket to the multicast
group specified by the user. As a result the tcpip stack code does the
following:
...
ip_setsockopt(_,_,IP_ADD_MEMBERSHIP,_,_) calls ip_mc_join_group(sock,_)
function, which adds a multicast entry to the sock->protinfo.af_inet.mc_list
and then adds a multicast entry to the bound device (if any) with a call
to the ip_mc_inc_group() function.
When remote multicast packets that belong to the multicast group we just
joined are received, the routing code recognizes the packets as multicast
packet going to a local receiver, but the __raw_v4_lookup() function only
examines the main sock address (sock->rcv_saddr) and totally forgets that
this raw sock can a member of multiple multicast groups.
The following patch fixes the problem:
---[ PATCH STARTS ]--------------------------------------------------------
--- linux-2.4.2/net/ipv4/raw.c Fri Feb 9 14:29:44 2001
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <linux/mroute.h>
+#include <linux/igmp.h>
#include <net/ip.h>
#include <net/protocol.h>
!(s->rcv_saddr && s->rcv_saddr != laddr) &&
!(s->bound_dev_if && s->bound_dev_if != dif))
break; /* gotcha */
+ if (LOCAL_MCAST(laddr)) {
+
+ struct ip_mc_socklist *iml;
+ struct ip_mreqn *imr;
+
+ for (iml=sk->protinfo.af_inet.mc_list; iml; iml=iml->next) {
+ imr = &(iml->multi);
+ if ((imr->imr_multiaddr.s_addr == laddr) && !(imr->imr_ifindex && imr->imr_ifindex != dif))
+ return s;
+ }
+
+ }
}
return s;
}
---[ PATCH ENDS ]-----------------------------------------------------------
Note: the same problem exists for UDP sockets also.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/