Fix a long-standing limitation in IPv4 multicast group membership.

By making the imo_membership array a dynamically allocated vector,
this minimizes disruption to existing IPv4 multicast code. This
change breaks the ABI for the kernel module ip_mroute.ko, and may
cause a small amount of churn for folks working on the IGMPv3 merge.

Previously, sockets were subject to a compile-time limitation on
the number of IPv4 group memberships, which was hard-coded to 20.
The imo_membership relationship, however, is 1:1 with regards to
a tuple of multicast group address and interface address. Users who
ran routing protocols such as OSPF ran into this limitation on machines
with a large system interface tree.

3 files changed, 40 insertions, 4 deletions

diff --git a/sys/netinet/in.h b/sys/netinet/in.h
index 1476bf4..570ed33 100644
--- a/sys/netinet/in.h
+++ b/sys/netinet/in.h
@@ -424,7 +424,14 @@ __END_DECLS
  */
 #define	IP_DEFAULT_MULTICAST_TTL  1	/* normally limit m'casts to 1 hop  */
 #define	IP_DEFAULT_MULTICAST_LOOP 1	/* normally hear sends if a member  */
-#define	IP_MAX_MEMBERSHIPS	20	/* per socket */
+
+/*
+ * The imo_membership vector for each socket is now dynamically allocated at
+ * run-time, bounded by USHRT_MAX, and is reallocated when needed, sized
+ * according to a power-of-two increment.
+ */
+#define	IP_MIN_MEMBERSHIPS	31
+#define	IP_MAX_MEMBERSHIPS	4095
 
 /*
  * Argument structure for IP_ADD_MEMBERSHIP and IP_DROP_MEMBERSHIP.
diff --git a/sys/netinet/ip_output.c b/sys/netinet/ip_output.c
index 915512e..9cb9871 100644
--- a/sys/netinet/ip_output.c
+++ b/sys/netinet/ip_output.c
@@ -1138,6 +1138,7 @@ static struct ip_moptions *
 ip_findmoptions(struct inpcb *inp)
 {
 	struct ip_moptions *imo;
+	struct in_multi **immp;
 
 	INP_LOCK(inp);
 	if (inp->inp_moptions != NULL)
@@ -1146,6 +1147,8 @@ ip_findmoptions(struct inpcb *inp)
 	INP_UNLOCK(inp);
 
 	imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, M_WAITOK);
+	immp = (struct in_multi **)malloc((sizeof(*immp) * IP_MIN_MEMBERSHIPS),
+					  M_IPMOPTS, M_WAITOK);
 
 	imo->imo_multicast_ifp = NULL;
 	imo->imo_multicast_addr.s_addr = INADDR_ANY;
@@ -1153,9 +1156,12 @@ ip_findmoptions(struct inpcb *inp)
 	imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL;
 	imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP;
 	imo->imo_num_memberships = 0;
+	imo->imo_max_memberships = IP_MIN_MEMBERSHIPS;
+	imo->imo_membership = immp;
 
 	INP_LOCK(inp);
 	if (inp->inp_moptions != NULL) {
+		free(immp, M_IPMOPTS);
 		free(imo, M_IPMOPTS);
 		return (inp->inp_moptions);
 	}
@@ -1360,11 +1366,32 @@ ip_setmoptions(struct inpcb *inp, struct sockopt *sopt)
 			splx(s);
 			break;
 		}
-		if (i == IP_MAX_MEMBERSHIPS) {
+		if (imo->imo_num_memberships == imo->imo_max_memberships) {
+		    struct in_multi **nmships, **omships;
+		    size_t newmax;
+		    /*
+		     * Resize the vector to next power-of-two minus 1. If the
+		     * size would exceed the maximum then we know we've really
+		     * run out of entries. Otherwise, we realloc() the vector
+		     * with the INP lock held to avoid introducing a race.
+		     */
+		    nmships = NULL;
+		    omships = imo->imo_membership;
+		    newmax = ((imo->imo_max_memberships + 1) * 2) - 1;
+		    if (newmax <= IP_MAX_MEMBERSHIPS) {
+			nmships = (struct in_multi **)realloc(omships,
+sizeof(*nmships) * newmax, M_IPMOPTS, M_NOWAIT);
+			if (nmships != NULL) {
+			    imo->imo_membership = nmships;
+			    imo->imo_max_memberships = newmax;
+			}
+		    }
+		    if (nmships == NULL) {
 			INP_UNLOCK(inp);
 			error = ETOOMANYREFS;
 			splx(s);
 			break;
+		    }
 		}
 		/*
 		 * Everything looks good; add a new record to the multicast
@@ -1538,6 +1565,7 @@ ip_freemoptions(imo)
 	if (imo != NULL) {
 		for (i = 0; i < imo->imo_num_memberships; ++i)
 			in_delmulti(imo->imo_membership[i]);
+		free(imo->imo_membership, M_IPMOPTS);
 		free(imo, M_IPMOPTS);
 	}
 }
diff --git a/sys/netinet/ip_var.h b/sys/netinet/ip_var.h
index 6d0a122..a051825 100644
--- a/sys/netinet/ip_var.h
+++ b/sys/netinet/ip_var.h
@@ -85,11 +85,12 @@ struct ipoption {
 struct ip_moptions {
 	struct	ifnet *imo_multicast_ifp; /* ifp for outgoing multicasts */
 	struct in_addr imo_multicast_addr; /* ifindex/addr on MULTICAST_IF */
+	u_long	imo_multicast_vif;	/* vif num outgoing multicasts */
 	u_char	imo_multicast_ttl;	/* TTL for outgoing multicasts */
 	u_char	imo_multicast_loop;	/* 1 => hear sends if a member */
 	u_short	imo_num_memberships;	/* no. memberships this socket */
-	struct	in_multi *imo_membership[IP_MAX_MEMBERSHIPS];
-	u_long	imo_multicast_vif;	/* vif num outgoing multicasts */
+	u_short	imo_max_memberships;	/* max memberships this socket */
+	struct	in_multi **imo_membership;	/* group memberships */
 };
 
 struct	ipstat {