Merge dyn9-47-18-140.beaverton.ibm.com:/home/sridhar/BK/linux-2.5.51

into dyn9-47-18-140.beaverton.ibm.com:/home/sridhar/BK/lksctp-2.5.51

include/linux/sysctl.h

@@ -543,7 +543,8 @@ enum {
 	NET_SCTP_PATH_MAX_RETRANS        =  8,
 	NET_SCTP_MAX_INIT_RETRANSMITS    =  9,
 	NET_SCTP_HB_INTERVAL             = 10,
-	NET_SCTP_MAX_BURST               = 11,
+	NET_SCTP_PRESERVE_ENABLE         = 11,
+	NET_SCTP_MAX_BURST               = 12,
 };
 
 /* CTL_PROC names: */

include/net/sctp/sctp.h

@@ -123,8 +123,8 @@ extern sctp_protocol_t sctp_proto;
 extern struct sock *sctp_get_ctl_sock(void);
 extern int sctp_copy_local_addr_list(sctp_protocol_t *, sctp_bind_addr_t *,
 				     sctp_scope_t, int priority, int flags);
-extern sctp_pf_t *sctp_get_pf_specific(int family);
-extern void sctp_set_pf_specific(int family, sctp_pf_t *);
+extern struct sctp_pf *sctp_get_pf_specific(sa_family_t family);
+extern int sctp_register_pf(struct sctp_pf *, sa_family_t);
 
 /*
  * sctp_socket.c

include/net/sctp/sm.h

@@ -140,6 +140,8 @@ sctp_state_fn_t sctp_sf_do_5_2_2_dupinit;
 sctp_state_fn_t sctp_sf_do_5_2_4_dupcook;
 sctp_state_fn_t sctp_sf_unk_chunk;
 sctp_state_fn_t sctp_sf_do_8_5_1_E_sa;
+sctp_state_fn_t sctp_sf_cookie_echoed_err;
+sctp_state_fn_t sctp_sf_do_5_2_6_stale;
 
 /* Prototypes for primitive event state functions.  */
 sctp_state_fn_t sctp_sf_do_prm_asoc;
@@ -175,7 +177,6 @@ sctp_state_fn_t sctp_sf_autoclose_timer_expire;
  */
 
 /* Prototypes for chunk state functions.  Not in use. */
-sctp_state_fn_t sctp_sf_do_5_2_6_stale;
 sctp_state_fn_t sctp_sf_do_9_2_reshutack;
 sctp_state_fn_t sctp_sf_do_9_2_reshut;
 sctp_state_fn_t sctp_sf_do_9_2_shutack;
@@ -211,7 +212,7 @@ void sctp_populate_tie_tags(__u8 *cookie, __u32 curTag, __u32 hisTag);
 /* Prototypes for chunk-building functions.  */
 sctp_chunk_t *sctp_make_init(const sctp_association_t *,
 			     const sctp_bind_addr_t *,
-			     int priority);
+			     int priority, int vparam_len);
 sctp_chunk_t *sctp_make_init_ack(const sctp_association_t *,
 				 const sctp_chunk_t *,
 				 const int priority,
@@ -322,9 +323,15 @@ sctp_pack_cookie(const sctp_endpoint_t *, const sctp_association_t *,
 		 const __u8 *, int addrs_len);
 sctp_association_t *sctp_unpack_cookie(const sctp_endpoint_t *,
 				       const sctp_association_t *,
-				       sctp_chunk_t *, int priority, int *err);
+				       sctp_chunk_t *, int priority, int *err,
+				       sctp_chunk_t **err_chk_p);
 int sctp_addip_addr_config(sctp_association_t *, sctp_param_t,
 			   struct sockaddr_storage*, int);
+void sctp_send_stale_cookie_err(const sctp_endpoint_t *ep,
+				const sctp_association_t *asoc,
+				const sctp_chunk_t *chunk,
+				sctp_cmd_seq_t *commands,
+				sctp_chunk_t *err_chunk);
 
 /* 3rd level prototypes */
 __u32 sctp_generate_tag(const sctp_endpoint_t *);

include/net/sctp/structs.h

@@ -42,6 +42,7 @@
  *    Sridhar Samudrala     <sri@us.ibm.com>
  *    Daisy Chang	    <daisyc@us.ibm.com>
  *    Dajiang Zhang         <dajiang.zhang@nokia.com>
+ *    Ardelle Fan           <ardelle.fan@intel.com> 
  *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
@@ -182,6 +183,9 @@ struct SCTP_protocol {
 
 	/* Valid.Cookie.Life        - 60  seconds  */
 	int valid_cookie_life;
+	
+	/* Whether Cookie Preservative is enabled(1) or not(0) */ 
+	int cookie_preserve_enable;
 
 	/* Association.Max.Retrans  - 10 attempts
 	 * Path.Max.Retrans         - 5  attempts (per destination address)
@@ -234,7 +238,7 @@ struct SCTP_protocol {
  * Pointers to address related SCTP functions.
  * (i.e. things that depend on the address family.)
  */
-typedef struct sctp_func {
+struct sctp_af {
 	int		(*queue_xmit)	(struct sk_buff *skb);
 	int 		(*setsockopt)	(struct sock *sk,
 					 int level,
@@ -259,27 +263,34 @@ typedef struct sctp_func {
 	void            (*from_skb)     (union sctp_addr *,
 					 struct sk_buff *skb,
 					 int saddr);
+	void            (*from_sk)      (union sctp_addr *,
+					 struct sock *sk);
+	void            (*to_sk)        (union sctp_addr *,
+					 struct sock *sk);
 	int             (*addr_valid)   (union sctp_addr *);
 	sctp_scope_t    (*scope) (union sctp_addr *);
 	void            (*inaddr_any)   (union sctp_addr *, unsigned short);
 	int             (*is_any)       (const union sctp_addr *);
+	int             (*available)    (const union sctp_addr *);
 	__u16		net_header_len;
 	int		sockaddr_len;
 	sa_family_t	sa_family;
 	struct list_head list;
-} sctp_func_t;
+};
 
-sctp_func_t *sctp_get_af_specific(sa_family_t);
+struct sctp_af *sctp_get_af_specific(sa_family_t);
+int sctp_register_af(struct sctp_af *);
 
 /* Protocol family functions. */
 typedef struct sctp_pf {
 	void (*event_msgname)(sctp_ulpevent_t *, char *, int *);
-	void (*skb_msgname)(struct sk_buff *, char *, int *);
-	int  (*af_supported)(sa_family_t);
+	void (*skb_msgname)  (struct sk_buff *, char *, int *);
+	int  (*af_supported) (sa_family_t);
 	int  (*cmp_addr) (const union sctp_addr *,
 			  const union sctp_addr *,
 			  struct sctp_opt *);
-	struct sctp_func *af;
+	int  (*bind_verify) (struct sctp_opt *, union sctp_addr *);
+	struct sctp_af *af;
 } sctp_pf_t;
 
 /* SCTP Socket type: UDP or TCP style. */
@@ -623,7 +634,7 @@ struct SCTP_transport {
 	union sctp_addr ipaddr;
 
 	/* These are the functions we call to handle LLP stuff.  */
-	sctp_func_t *af_specific;
+	struct sctp_af *af_specific;
 
 	/* Which association do we belong to?  */
 	sctp_association_t *asoc;
@@ -1271,7 +1282,6 @@ struct SCTP_association {
 
 	/* The cookie life I award for any cookie.  */
 	struct timeval cookie_life;
-	__u32 cookie_preserve;
 
 	/* Overall     : The overall association error count.
 	 * Error Count : [Clear this any time I get something.]

net/sctp/associola.c

@@ -128,8 +128,9 @@ sctp_association_t *sctp_association_init(sctp_association_t *asoc,
 	asoc->state_timestamp = jiffies;
 
 	/* Set things that have constant value.  */
-	asoc->cookie_life.tv_sec = SCTP_DEFAULT_COOKIE_LIFE_SEC;
-	asoc->cookie_life.tv_usec = SCTP_DEFAULT_COOKIE_LIFE_USEC;
+	asoc->cookie_life.tv_sec = sctp_proto.valid_cookie_life / HZ;
+	asoc->cookie_life.tv_usec = (sctp_proto.valid_cookie_life % HZ) *
+					1000000L / HZ;
 
 	asoc->pmtu = 0;
 	asoc->frag_point = 0;
@@ -642,7 +643,7 @@ __u16 __sctp_association_get_next_ssn(sctp_association_t *asoc, __u16 sid)
 int sctp_cmp_addr_exact(const union sctp_addr *ss1,
 			const union sctp_addr *ss2)
 {
-	struct sctp_func *af;
+	struct sctp_af *af;
 
 	af = sctp_get_af_specific(ss1->sa.sa_family);
 	if (!af)

net/sctp/bind_addr.c

@@ -327,7 +327,7 @@ static int sctp_copy_one_addr(sctp_bind_addr_t *dest, union sctp_addr *addr,
 /* Is this a wildcard address?  */
 int sctp_is_any(const union sctp_addr *addr)
 {
-	struct sctp_func *af = sctp_get_af_specific(addr->sa.sa_family);
+	struct sctp_af *af = sctp_get_af_specific(addr->sa.sa_family);
 	if (!af)
 		return 0;
 	return af->is_any(addr);
@@ -362,7 +362,7 @@ int sctp_in_scope(const union sctp_addr *addr, sctp_scope_t scope)
 /* What is the scope of 'addr'?  */
 sctp_scope_t sctp_scope(const union sctp_addr *addr)
 {
-	struct sctp_func *af;
+	struct sctp_af *af;
 
 	af = sctp_get_af_specific(addr->sa.sa_family);
 	if (!af)

net/sctp/input.c

@@ -42,6 +42,7 @@
  *    Hui Huang <hui.huang@nokia.com>
  *    Daisy Chang <daisyc@us.ibm.com>
  *    Sridhar Samudrala <sri@us.ibm.com>
+ *    Ardelle Fan <ardelle.fan@intel.com>
  *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
@@ -96,7 +97,7 @@ int sctp_rcv(struct sk_buff *skb)
 	struct sctphdr *sh;
 	union sctp_addr src;
 	union sctp_addr dest;
-	struct sctp_func *af;
+	struct sctp_af *af;
 	int ret = 0;
 
 	if (skb->pkt_type!=PACKET_HOST)
@@ -279,6 +280,7 @@ int sctp_rcv_ootb(struct sk_buff *skb)
 {
 	sctp_chunkhdr_t *ch;
 	__u8 *ch_end;
+	sctp_errhdr_t *err;
 
 	ch = (sctp_chunkhdr_t *) skb->data;
 
@@ -308,8 +310,9 @@ int sctp_rcv_ootb(struct sk_buff *skb)
 			goto discard;
 
 		if (ch->type == SCTP_CID_ERROR) {
-			/* FIXME - Need to check the "Stale cookie" ERROR. */
-			goto discard;
+			err = (sctp_errhdr_t *)(ch + sizeof(sctp_chunkhdr_t));
+			if (SCTP_ERROR_STALE_COOKIE == err->cause)
+				goto discard;
 		}
 
 		ch = (sctp_chunkhdr_t *) ch_end;

net/sctp/ipv6.c

@@ -76,8 +76,19 @@
 
 #include <asm/uaccess.h>
 
-/* FIXME: Cleanup so we don't need TEST_FRAME here. */
-#ifndef TEST_FRAME
+extern struct notifier_block sctp_inetaddr_notifier;
+
+/* FIXME: This macro needs to be moved to a common header file. */
+#define NIP6(addr) \
+        ntohs((addr)->s6_addr16[0]), \
+        ntohs((addr)->s6_addr16[1]), \
+        ntohs((addr)->s6_addr16[2]), \
+        ntohs((addr)->s6_addr16[3]), \
+        ntohs((addr)->s6_addr16[4]), \
+        ntohs((addr)->s6_addr16[5]), \
+        ntohs((addr)->s6_addr16[6]), \
+        ntohs((addr)->s6_addr16[7])
+
 /* FIXME: Comments. */
 static inline void sctp_v6_err(struct sk_buff *skb,
 			       struct inet6_skb_parm *opt,
@@ -92,13 +103,38 @@ static inline int sctp_v6_xmit(struct sk_buff *skb)
 	struct sock *sk = skb->sk;
 	struct ipv6_pinfo *np = inet6_sk(sk);
 	struct flowi fl;
-	struct dst_entry *dst;
+	struct dst_entry *dst = skb->dst;
+	struct rt6_info *rt6 = (struct rt6_info *)dst;
 	struct in6_addr saddr;
-	int err = 0;
+	int err;
 
 	fl.proto = sk->protocol;
-	fl.fl6_dst = &np->daddr;
-	fl.fl6_src = NULL;
+	fl.fl6_dst = &rt6->rt6i_dst.addr;
+
+	/* FIXME: Currently, ip6_route_output() doesn't fill in the source
+	 * address in the returned route entry. So we call ipv6_get_saddr()
+	 * to get an appropriate source address. It is possible that this address
+	 * may not be part of the bind address list of the association.
+	 * Once ip6_route_ouput() is fixed so that it returns a route entry
+	 * with an appropriate source address, the following if condition can
+	 * be removed. With ip6_route_output() returning a source address filled
+	 * route entry, sctp_transport_route() can do real source address 
+	 * selection for v6.
+	 */ 
+	if (ipv6_addr_any(&rt6->rt6i_src.addr)) {
+		err = ipv6_get_saddr(dst, fl.fl6_dst, &saddr);
+
+		if (err) {
+			printk(KERN_ERR "%s: No saddr available for "
+			       "DST=%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x\n",
+			       __FUNCTION__, NIP6(fl.fl6_src));
+			return err;
+		}
+
+		fl.fl6_src = &saddr;
+	} else {
+		fl.fl6_src = &rt6->rt6i_src.addr;
+	}
 
 	fl.fl6_flowlabel = np->flow_label;
 	IP6_ECN_flow_xmit(sk, fl.fl6_flowlabel);
@@ -111,63 +147,8 @@ static inline int sctp_v6_xmit(struct sk_buff *skb)
 		fl.nl_u.ip6_u.daddr = rt0->addr;
 	}
 
-	dst = __sk_dst_check(sk, np->dst_cookie);
-
-	if (dst == NULL) {
-		dst = ip6_route_output(sk, &fl);
-
-		if (dst->error) {
-			sk->err_soft = -dst->error;
-			dst_release(dst);
-			return -sk->err_soft;
-		}
-		ip6_dst_store(sk, dst, NULL);
-	}
-
-	skb->dst = dst_clone(dst);
-
-	/* FIXME: This is all temporary until real source address
-	 * selection is done.
-	 */
-	if (ipv6_addr_any(&np->saddr)) {
-		err = ipv6_get_saddr(dst, fl.fl6_dst, &saddr);
-
-		if (err)
-			printk(KERN_ERR "sctp_v6_xmit: no saddr available\n");
-
-		/* FIXME: This is a workaround until we get
-		 * real source address selection done.  This is here
-		 * to disallow loopback when the scoping rules have
-		 * not bound loopback to the endpoint.
-		 */
-		if (sctp_ipv6_addr_type(&saddr) & IPV6_ADDR_LOOPBACK) {
-			if (!(sctp_ipv6_addr_type(&np->daddr) &
-			      IPV6_ADDR_LOOPBACK)) {
-				ipv6_addr_copy(&saddr, &np->daddr);
-			}
-		}
-		fl.fl6_src = &saddr;
-	} else {
-		fl.fl6_src = &np->saddr;
-	}
-
-	/* Restore final destination back after routing done */
-	fl.nl_u.ip6_u.daddr = &np->daddr;
-
 	return ip6_xmit(sk, skb, &fl, np->opt);
 }
-#endif /* TEST_FRAME */
-
-/* FIXME: This macro needs to be moved to a common header file. */
-#define NIP6(addr) \
-        ntohs((addr)->s6_addr16[0]), \
-        ntohs((addr)->s6_addr16[1]), \
-        ntohs((addr)->s6_addr16[2]), \
-        ntohs((addr)->s6_addr16[3]), \
-        ntohs((addr)->s6_addr16[4]), \
-        ntohs((addr)->s6_addr16[5]), \
-        ntohs((addr)->s6_addr16[6]), \
-        ntohs((addr)->s6_addr16[7])
 
 /* Returns the dst cache entry for the given source and destination ip
  * addresses.
@@ -176,7 +157,7 @@ struct dst_entry *sctp_v6_get_dst(union sctp_addr *daddr,
 				  union sctp_addr *saddr)
 {
 	struct dst_entry *dst;
-	struct flowi fl = { 
+	struct flowi fl = {
 		.nl_u = { .ip6_u = { .daddr = &daddr->v6.sin6_addr, } } };
 
 
@@ -261,6 +242,20 @@ static void sctp_v6_from_skb(union sctp_addr *addr,struct sk_buff *skb,
 	ipv6_addr_copy(&addr->v6.sin6_addr, from);
 }
 
+/* Initialize an sctp_addr from a socket. */
+static void sctp_v6_from_sk(union sctp_addr *addr, struct sock *sk)
+{
+	addr->v6.sin6_family = AF_INET6;
+	addr->v6.sin6_port = inet_sk(sk)->num;
+	addr->v6.sin6_addr = inet6_sk(sk)->rcv_saddr;
+}
+
+/* Initialize sk->rcv_saddr from sctp_addr. */
+static void sctp_v6_to_sk(union sctp_addr *addr, struct sock *sk)
+{
+	inet6_sk(sk)->rcv_saddr = addr->v6.sin6_addr;
+}
+
 /* Initialize a sctp_addr from a dst_entry. */
 static void sctp_v6_dst_saddr(union sctp_addr *addr, struct dst_entry *dst)
 {
@@ -270,15 +265,15 @@ static void sctp_v6_dst_saddr(union sctp_addr *addr, struct dst_entry *dst)
 }
 
 /* Compare addresses exactly.  Well.. almost exactly; ignore scope_id
- * for now.  FIXME.
+ * for now.  FIXME: v4-mapped-v6.
  */
-static int sctp_v6_cmp_addr(const union sctp_addr *addr1, 
+static int sctp_v6_cmp_addr(const union sctp_addr *addr1,
 			    const union sctp_addr *addr2)
 {
 	int match;
 	if (addr1->sa.sa_family != addr2->sa.sa_family)
 		return 0;
-	match = !ipv6_addr_cmp((struct in6_addr *)&addr1->v6.sin6_addr, 
+	match = !ipv6_addr_cmp((struct in6_addr *)&addr1->v6.sin6_addr,
 			       (struct in6_addr *)&addr2->v6.sin6_addr);
 
 	return match;
@@ -300,6 +295,22 @@ static int sctp_v6_is_any(const union sctp_addr *addr)
 	return IPV6_ADDR_ANY == type;
 }
 
+/* Should this be available for binding?   */
+static int sctp_v6_available(const union sctp_addr *addr)
+{
+	int type;
+	struct in6_addr *in6 = (struct in6_addr *)&addr->v6.sin6_addr;
+
+	type = ipv6_addr_type(in6);
+	if (IPV6_ADDR_ANY == type)
+		return 1;
+	if (!(type & IPV6_ADDR_UNICAST))
+		return 0;
+
+	return ipv6_chk_addr(in6, NULL);
+}
+
+
 /* This function checks if the address is a valid address to be used for
  * SCTP.
  *
@@ -309,7 +320,7 @@ static int sctp_v6_is_any(const union sctp_addr *addr)
  */
 static int sctp_v6_addr_valid(union sctp_addr *addr)
 {
-	int ret = sctp_ipv6_addr_type(&addr->v6.sin6_addr);
+	int ret = ipv6_addr_type(&addr->v6.sin6_addr);
 
 	/* FIXME:  v4-mapped-v6 address support. */
 
@@ -442,14 +453,14 @@ static int sctp_inet6_af_supported(sa_family_t family)
 
 /* Address matching with wildcards allowed.  This extra level
  * of indirection lets us choose whether a PF_INET6 should
- * disallow any v4 addresses if we so choose. 
+ * disallow any v4 addresses if we so choose.
  */
-static int sctp_inet6_cmp_addr(const union sctp_addr *addr1, 
+static int sctp_inet6_cmp_addr(const union sctp_addr *addr1,
 			       const union sctp_addr *addr2,
 			       struct sctp_opt *opt)
 {
-	struct sctp_func *af1, *af2;
-	
+	struct sctp_af *af1, *af2;
+
 	af1 = sctp_get_af_specific(addr1->sa.sa_family);
 	af2 = sctp_get_af_specific(addr2->sa.sa_family);
 
@@ -461,11 +472,25 @@ static int sctp_inet6_cmp_addr(const union sctp_addr *addr1,
 
 	if (addr1->sa.sa_family != addr2->sa.sa_family)
 		return 0;
-	
+
 	return af1->cmp_addr(addr1, addr2);
 }
 
+/* Verify that the provided sockaddr looks bindable.   Common verification,
+ * has already been taken care of.
+ */
+static int sctp_inet6_bind_verify(struct sctp_opt *opt, union sctp_addr *addr)
+{
+	struct sctp_af *af;
+
+	/* ASSERT: address family has already been verified. */
+	if (addr->sa.sa_family != AF_INET6) {
+		af = sctp_get_af_specific(addr->sa.sa_family);
+	} else
+		af = opt->pf->af;
 
+	return af->available(addr);
+}
 
 static struct proto_ops inet6_seqpacket_ops = {
 	.family     = PF_INET6,
@@ -501,29 +526,33 @@ static struct inet6_protocol sctpv6_protocol = {
 	.err_handler  = sctp_v6_err,
 };
 
-static sctp_func_t sctp_ipv6_specific = {
+static struct sctp_af sctp_ipv6_specific = {
 	.queue_xmit      = sctp_v6_xmit,
-	.setsockopt      = ipv6_setsockopt,
+	.setsockopt    = ipv6_setsockopt,
 	.getsockopt      = ipv6_getsockopt,
 	.get_dst	 = sctp_v6_get_dst,
 	.copy_addrlist   = sctp_v6_copy_addrlist,
 	.from_skb        = sctp_v6_from_skb,
+	.from_sk         = sctp_v6_from_sk,
+	.to_sk           = sctp_v6_to_sk,
 	.dst_saddr       = sctp_v6_dst_saddr,
 	.cmp_addr        = sctp_v6_cmp_addr,
 	.scope           = sctp_v6_scope,
 	.addr_valid      = sctp_v6_addr_valid,
 	.inaddr_any      = sctp_v6_inaddr_any,
 	.is_any          = sctp_v6_is_any,
+	.available       = sctp_v6_available,
 	.net_header_len  = sizeof(struct ipv6hdr),
 	.sockaddr_len    = sizeof(struct sockaddr_in6),
 	.sa_family       = AF_INET6,
 };
 
-static sctp_pf_t sctp_pf_inet6_specific = {
+static struct sctp_pf sctp_pf_inet6_specific = {
 	.event_msgname = sctp_inet6_event_msgname,
 	.skb_msgname   = sctp_inet6_skb_msgname,
 	.af_supported  = sctp_inet6_af_supported,
 	.cmp_addr      = sctp_inet6_cmp_addr,
+	.bind_verify   = sctp_inet6_bind_verify,
 	.af            = &sctp_ipv6_specific,
 };
 
@@ -538,11 +567,13 @@ int sctp_v6_init(void)
 	inet6_register_protosw(&sctpv6_protosw);
 
 	/* Register the SCTP specfic PF_INET6 functions. */
-	sctp_set_pf_specific(PF_INET6, &sctp_pf_inet6_specific);
+	sctp_register_pf(&sctp_pf_inet6_specific, PF_INET6);
+
+	/* Register the SCTP specfic AF_INET6 functions. */
+	sctp_register_af(&sctp_ipv6_specific);
 
-	/* Fill in address family info.  */
-	INIT_LIST_HEAD(&sctp_ipv6_specific.list);
-	list_add_tail(&sctp_ipv6_specific.list, &sctp_proto.address_families);
+	/* Register notifier for inet6 address additions/deletions. */ 
+	register_inet6addr_notifier(&sctp_inetaddr_notifier);
 
 	return 0;
 }
@@ -553,4 +584,5 @@ void sctp_v6_exit(void)
 	list_del(&sctp_ipv6_specific.list);
 	inet6_del_protocol(&sctpv6_protocol, IPPROTO_SCTP);
 	inet6_unregister_protosw(&sctpv6_protosw);
+	unregister_inet6addr_notifier(&sctp_inetaddr_notifier);
 }

net/sctp/protocol.c

@@ -40,6 +40,7 @@
  *    Jon Grimm <jgrimm@us.ibm.com>
  *    Sridhar Samudrala <sri@us.ibm.com>
  *    Daisy Chang <daisyc@us.ibm.com>
+ *    Ardelle Fan <ardelle.fan@intel.com>
  *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
@@ -67,8 +68,10 @@ struct sctp_mib sctp_statistics[NR_CPUS * 2];
  */
 static struct socket *sctp_ctl_socket;
 
-static sctp_pf_t *sctp_pf_inet6_specific;
-static sctp_pf_t *sctp_pf_inet_specific;
+static struct sctp_pf *sctp_pf_inet6_specific;
+static struct sctp_pf *sctp_pf_inet_specific;
+static struct sctp_af *sctp_af_v4_specific;
+static struct sctp_af *sctp_af_v6_specific;
 
 extern struct net_proto_family inet_family_ops;
 
@@ -140,12 +143,12 @@ static void __sctp_get_local_addr_list(sctp_protocol_t *proto)
 {
 	struct net_device *dev;
 	struct list_head *pos;
-	struct sctp_func *af;
+	struct sctp_af *af;
 
 	read_lock(&dev_base_lock);
 	for (dev = dev_base; dev; dev = dev->next) {
 		list_for_each(pos, &proto->address_families) {
-			af = list_entry(pos, sctp_func_t, list);
+			af = list_entry(pos, struct sctp_af, list);
 			af->copy_addrlist(&proto->local_addr_list, dev);
 		}
 	}
@@ -251,7 +254,6 @@ struct dst_entry *sctp_v4_get_dst(union sctp_addr *daddr,
 	return &rt->u.dst;
 }
 
-
 /* Initialize a sctp_addr from in incoming skb.  */
 static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
 			     int is_saddr)
@@ -274,6 +276,21 @@ static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
 	memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr));
 }
 
+/* Initialize an sctp_addr from a socket. */
+static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
+{
+	addr->v4.sin_family = AF_INET;
+	addr->v4.sin_port = inet_sk(sk)->num;
+	addr->v4.sin_addr.s_addr = inet_sk(sk)->rcv_saddr;
+}
+
+/* Initialize sk->rcv_saddr from sctp_addr. */
+static void sctp_v4_to_sk(union sctp_addr *addr, struct sock *sk)
+{
+	inet_sk(sk)->rcv_saddr = addr->v4.sin_addr.s_addr;
+}
+
+
 /* Initialize a sctp_addr from a dst_entry. */
 static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct dst_entry *dst)
 {
@@ -311,7 +328,7 @@ static int sctp_v4_is_any(const union sctp_addr *addr)
 }
 
 /* This function checks if the address is a valid address to be used for
- * SCTP.
+ * SCTP binding.
  *
  * Output:
  * Return 0 - If the address is a non-unicast or an illegal address.
@@ -326,6 +343,18 @@ static int sctp_v4_addr_valid(union sctp_addr *addr)
 	return 1;
 }
 
+/* Should this be available for binding?   */
+static int sctp_v4_available(const union sctp_addr *addr)
+{
+	int ret = inet_addr_type(addr->v4.sin_addr.s_addr);
+	
+	/* FIXME: ip_nonlocal_bind sysctl support. */
+
+	if (addr->v4.sin_addr.s_addr != INADDR_ANY && ret != RTN_LOCAL)
+		return 0;
+	return 1;
+}
+
 /* Checking the loopback, private and other address scopes as defined in
  * RFC 1918.   The IPv4 scoping is based on the draft for SCTP IPv4
  * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
@@ -365,11 +394,11 @@ static sctp_scope_t sctp_v4_scope(union sctp_addr *addr)
 	return retval;
 }
 
-/* Event handler for inet device events.
+/* Event handler for inet address addition/deletion events.
  * Basically, whenever there is an event, we re-build our local address list.
  */
-static int sctp_netdev_event(struct notifier_block *this, unsigned long event,
-			     void *ptr)
+static int sctp_inetaddr_event(struct notifier_block *this, unsigned long event,
+			       void *ptr)
 {
 	long flags __attribute__ ((unused));
 
@@ -405,29 +434,42 @@ int sctp_ctl_sock_init(void)
 	return 0;
 }
 
+/* Register address family specific functions. */
+int sctp_register_af(struct sctp_af *af)
+{
+	switch (af->sa_family) {
+	case AF_INET:
+		if (sctp_af_v4_specific)
+			return 0;
+		sctp_af_v4_specific = af;
+		break;
+	case AF_INET6:
+		if (sctp_af_v6_specific)
+			return 0;
+		sctp_af_v6_specific = af;
+		break;
+	default:
+		return 0;
+	}
+
+	INIT_LIST_HEAD(&af->list);
+	list_add_tail(&af->list, &sctp_proto.address_families);
+	return 1;
+}
+
 /* Get the table of functions for manipulating a particular address
  * family.
  */
-sctp_func_t *sctp_get_af_specific(sa_family_t family)
+struct sctp_af *sctp_get_af_specific(sa_family_t family)
 {
-	struct list_head *pos;
-	sctp_protocol_t *proto = sctp_get_protocol();
-	struct sctp_func *retval, *af;
-
-	retval = NULL;
-
-	/* Cycle through all AF specific functions looking for a
-	 * match.
-	 */
-	list_for_each(pos, &proto->address_families) {
-		af = list_entry(pos, sctp_func_t, list);
-		if (family == af->sa_family) {
-			retval = af;
-			break;
-		}
+	switch (family) {
+	case AF_INET:
+		return sctp_af_v4_specific;
+	case AF_INET6:
+		return sctp_af_v6_specific;
+	default:
+		return NULL;
 	}
-
-	return retval;
 }
 
 /* Common code to initialize a AF_INET msg_name. */
@@ -495,21 +537,28 @@ static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
 	return 0;
 }
 
+/* Verify that provided sockaddr looks bindable.  Common verification has
+ * already been taken care of.
+ */
+static int sctp_inet_bind_verify(struct sctp_opt *opt, union sctp_addr *addr)
+{
+	return sctp_v4_available(addr);
+}
 
-struct sctp_func sctp_ipv4_specific;
+struct sctp_af sctp_ipv4_specific;
 
-static sctp_pf_t sctp_pf_inet = {
+static struct sctp_pf sctp_pf_inet = {
 	.event_msgname = sctp_inet_event_msgname,
 	.skb_msgname   = sctp_inet_skb_msgname,
 	.af_supported  = sctp_inet_af_supported,
 	.cmp_addr      = sctp_inet_cmp_addr,
+	.bind_verify   = sctp_inet_bind_verify,
 	.af            = &sctp_ipv4_specific,
 };
 
-
-/* Registration for netdev events.  */
-struct notifier_block sctp_netdev_notifier = {
-	.notifier_call = sctp_netdev_event,
+/* Notifier for inetaddr addition/deletion events.  */
+struct notifier_block sctp_inetaddr_notifier = {
+	.notifier_call = sctp_inetaddr_event,
 };
 
 /* Socket operations.  */
@@ -551,25 +600,28 @@ static struct inet_protocol sctp_protocol = {
 };
 
 /* IPv4 address related functions.  */
-struct sctp_func sctp_ipv4_specific = {
+struct sctp_af sctp_ipv4_specific = {
 	.queue_xmit     = ip_queue_xmit,
 	.setsockopt     = ip_setsockopt,
 	.getsockopt     = ip_getsockopt,
 	.get_dst	= sctp_v4_get_dst,
 	.copy_addrlist  = sctp_v4_copy_addrlist,
 	.from_skb       = sctp_v4_from_skb,
+	.from_sk        = sctp_v4_from_sk,
+	.to_sk          = sctp_v4_to_sk,
 	.dst_saddr      = sctp_v4_dst_saddr,
 	.cmp_addr       = sctp_v4_cmp_addr,
 	.addr_valid     = sctp_v4_addr_valid,
 	.inaddr_any     = sctp_v4_inaddr_any,
 	.is_any         = sctp_v4_is_any,
+	.available      = sctp_v4_available,
 	.scope          = sctp_v4_scope,
 	.net_header_len = sizeof(struct iphdr),
 	.sockaddr_len   = sizeof(struct sockaddr_in),
 	.sa_family      = AF_INET,
 };
 
-sctp_pf_t *sctp_get_pf_specific(int family) {
+struct sctp_pf *sctp_get_pf_specific(sa_family_t family) {
 
 	switch (family) {
 	case PF_INET:
@@ -581,20 +633,24 @@ sctp_pf_t *sctp_get_pf_specific(int family) {
 	}
 }
 
-/* Set the PF specific function table.  */
-void sctp_set_pf_specific(int family, sctp_pf_t *pf)
+/* Register the PF specific function table.  */
+int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
 {
 	switch (family) {
 	case PF_INET:
+		if (sctp_pf_inet_specific)
+			return 0;
 		sctp_pf_inet_specific = pf;
 		break;
 	case PF_INET6:
+		if (sctp_pf_inet6_specific)
+			return 0;
 		sctp_pf_inet6_specific = pf;
 		break;
 	default:
-		BUG();
-		break;
+		return 0;
 	}
+	return 1;
 }
 
 /* Initialize the universe into something sensible.  */
@@ -617,7 +673,7 @@ int sctp_init(void)
 	sctp_dbg_objcnt_init();
 
 	/* Initialize the SCTP specific PF functions. */
-	sctp_set_pf_specific(PF_INET, &sctp_pf_inet);
+	sctp_register_pf(&sctp_pf_inet, PF_INET);
 	/*
 	 * 14. Suggested SCTP Protocol Parameter Values
 	 */
@@ -636,6 +692,9 @@ int sctp_init(void)
 	/* Valid.Cookie.Life        - 60  seconds */
 	sctp_proto.valid_cookie_life	= 60 * HZ;
 
+	/* Whether Cookie Preservative is enabled(1) or not(0) */ 
+	sctp_proto.cookie_preserve_enable = 1;
+
 	/* Max.Burst		    - 4 */
 	sctp_proto.max_burst = SCTP_MAX_BURST;
 
@@ -709,8 +768,7 @@ int sctp_init(void)
 	sctp_sysctl_register();
 
 	INIT_LIST_HEAD(&sctp_proto.address_families);
-	INIT_LIST_HEAD(&sctp_ipv4_specific.list);
-	list_add_tail(&sctp_ipv4_specific.list, &sctp_proto.address_families);
+	sctp_register_af(&sctp_ipv4_specific);
 
 	status = sctp_v6_init();
 	if (status)
@@ -727,7 +785,9 @@ int sctp_init(void)
 	INIT_LIST_HEAD(&sctp_proto.local_addr_list);
 	sctp_proto.local_addr_lock = SPIN_LOCK_UNLOCKED;
 
-	register_inetaddr_notifier(&sctp_netdev_notifier);
+	/* Register notifier for inet address additions/deletions. */ 
+	register_inetaddr_notifier(&sctp_inetaddr_notifier);
+
 	sctp_get_local_addr_list(&sctp_proto);
 
 	return 0;
@@ -757,8 +817,10 @@ void sctp_exit(void)
 	 * up all the remaining associations and all that memory.
 	 */
 
+	/* Unregister notifier for inet address additions/deletions. */
+	unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
+
 	/* Free the local address list.  */
-	unregister_inetaddr_notifier(&sctp_netdev_notifier);
 	sctp_free_local_addr_list(&sctp_proto);
 
 	/* Free the control endpoint.  */

net/sctp/sm_make_chunk.c

@@ -77,13 +77,18 @@ static const sctp_supported_addrs_param_t sat_param = {
 	{
 		SCTP_PARAM_SUPPORTED_ADDRESS_TYPES,
 		__constant_htons(SCTP_SAT_LEN),
-	},
-	{               /* types[] */
-		SCTP_PARAM_IPV4_ADDRESS,
-		SCTP_V6(SCTP_PARAM_IPV6_ADDRESS,)
 	}
 };
 
+/* gcc 3.2 doesn't allow initialization of zero-length arrays. So the above
+ * structure is split and the address types array is initialized using a
+ * fixed length array. 
+ */
+static const __u16 sat_addr_types[2] = {
+	SCTP_PARAM_IPV4_ADDRESS,
+	SCTP_V6(SCTP_PARAM_IPV6_ADDRESS,)
+};			
+
 /* RFC 2960 3.3.2 Initiation (INIT) (1)
  *
  * Note 2: The ECN capable field is reserved for future use of
@@ -163,7 +168,7 @@ void  sctp_init_cause(sctp_chunk_t *chunk, __u16 cause_code,
  */
 sctp_chunk_t *sctp_make_init(const sctp_association_t *asoc,
 			     const sctp_bind_addr_t *bp,
-			     int priority)
+			     int priority, int vparam_len)
 {
 	sctp_inithdr_t init;
 	union sctp_params addrs;
@@ -192,6 +197,7 @@ sctp_chunk_t *sctp_make_init(const sctp_association_t *asoc,
 
 	chunksize = sizeof(init) + addrs_len + SCTP_SAT_LEN;
 	chunksize += sizeof(ecap_param);
+	chunksize += vparam_len;
 
 	/* RFC 2960 3.3.2 Initiation (INIT) (1)
 	 *
@@ -213,7 +219,10 @@ sctp_chunk_t *sctp_make_init(const sctp_association_t *asoc,
 		sctp_addto_chunk(retval, sizeof(init), &init);
 	retval->param_hdr.v =
 		sctp_addto_chunk(retval, addrs_len, addrs.v);
-	sctp_addto_chunk(retval, SCTP_SAT_LEN, &sat_param);
+
+	sctp_addto_chunk(retval, sizeof(sctp_paramhdr_t), &sat_param);
+	sctp_addto_chunk(retval, sizeof(sat_addr_types), sat_addr_types);
+
 	sctp_addto_chunk(retval, sizeof(ecap_param), &ecap_param);
 
 nodata:
@@ -1337,7 +1346,7 @@ sctp_cookie_param_t *sctp_pack_cookie(const sctp_endpoint_t *ep,
 sctp_association_t *sctp_unpack_cookie(const sctp_endpoint_t *ep,
 				       const sctp_association_t *asoc,
 				       sctp_chunk_t *chunk, int priority,
-				       int *error)
+				       int *error, sctp_chunk_t **err_chk_p)
 {
 	sctp_association_t *retval = NULL;
 	sctp_signed_cookie_t *cookie;
@@ -1394,7 +1403,29 @@ sctp_association_t *sctp_unpack_cookie(const sctp_endpoint_t *ep,
 	 * for init collision case of lost COOKIE ACK.
 	 */
 	if (!asoc && tv_lt(bear_cookie->expiration, chunk->skb->stamp)) {
-		*error = -SCTP_IERROR_STALE_COOKIE;
+		/*
+		 * Section 3.3.10.3 Stale Cookie Error (3)
+		 *
+		 * Cause of error
+		 * ---------------
+		 * Stale Cookie Error:  Indicates the receipt of a valid State
+		 * Cookie that has expired.
+		 */
+		*err_chk_p = sctp_make_op_error_space(asoc, chunk,
+						ntohs(chunk->chunk_hdr->length));
+		if (*err_chk_p) {
+			suseconds_t usecs = (chunk->skb->stamp.tv_sec -
+				bear_cookie->expiration.tv_sec) * 1000000L +
+				chunk->skb->stamp.tv_usec -
+				bear_cookie->expiration.tv_usec;
+
+			usecs = htonl(usecs);
+			sctp_init_cause(*err_chk_p, SCTP_ERROR_STALE_COOKIE,
+					&usecs, sizeof(usecs));
+			*error = -SCTP_IERROR_STALE_COOKIE;
+		} else
+			*error = -SCTP_IERROR_NOMEM;
+
 		goto fail;
 	}
 
@@ -1751,6 +1782,7 @@ int sctp_process_param(sctp_association_t *asoc, union sctp_params param,
 	__u16 sat;
 	int retval = 1;
 	sctp_scope_t scope;
+	time_t stale;
 
 	/* We maintain all INIT parameters in network byte order all the
 	 * time.  This allows us to not worry about whether the parameters
@@ -1770,8 +1802,16 @@ int sctp_process_param(sctp_association_t *asoc, union sctp_params param,
 		break;
 
 	case SCTP_PARAM_COOKIE_PRESERVATIVE:
-		asoc->cookie_preserve =
-			ntohl(param.life->lifespan_increment);
+		if (!sctp_proto.cookie_preserve_enable)
+			break;
+
+		stale = ntohl(param.life->lifespan_increment);
+
+		/* Suggested Cookie Life span increment's unit is msec,
+		 * (1/1000sec).
+		 */
+		asoc->cookie_life.tv_sec += stale / 1000;
+		asoc->cookie_life.tv_usec += (stale % 1000) * 1000;
 		break;
 
 	case SCTP_PARAM_HOST_NAME_ADDRESS:

net/sctp/sm_sideeffect.c

@@ -68,7 +68,8 @@ static void sctp_do_8_2_transport_strike(sctp_association_t *asoc,
 					 sctp_transport_t *transport);
 static void sctp_cmd_init_failed(sctp_cmd_seq_t *, sctp_association_t *asoc);
 static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *, sctp_association_t *asoc,
-				  sctp_event_t event_type, sctp_chunk_t *chunk);
+				  sctp_event_t event_type, sctp_subtype_t stype,
+				  sctp_chunk_t *chunk);
 static int sctp_cmd_process_init(sctp_cmd_seq_t *, sctp_association_t *asoc,
 				 sctp_chunk_t *chunk,
 				 sctp_init_chunk_t *peer_init,
@@ -517,7 +518,7 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 
 		case SCTP_CMD_ASSOC_FAILED:
 			sctp_cmd_assoc_failed(commands, asoc, event_type,
-					      chunk);
+					      subtype, chunk);
 			break;
 
 		case SCTP_CMD_COUNTER_INC:
@@ -1046,18 +1047,27 @@ static void sctp_cmd_init_failed(sctp_cmd_seq_t *commands,
 static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *commands,
 				  sctp_association_t *asoc,
 				  sctp_event_t event_type,
+				  sctp_subtype_t subtype,
 				  sctp_chunk_t *chunk)
 {
 	sctp_ulpevent_t *event;
 	__u16 error = 0;
 
-	if (event_type == SCTP_EVENT_T_PRIMITIVE)
-		error = SCTP_ERROR_USER_ABORT;
-
-	if (chunk && (SCTP_CID_ABORT == chunk->chunk_hdr->type) &&
-	    (ntohs(chunk->chunk_hdr->length) >= (sizeof(struct sctp_chunkhdr) +
-				 		 sizeof(struct sctp_errhdr)))) {
-		error = ((sctp_errhdr_t *)chunk->skb->data)->cause;
+	switch(event_type) {
+	case SCTP_EVENT_T_PRIMITIVE:
+		if (SCTP_PRIMITIVE_ABORT == subtype.primitive)
+			error = SCTP_ERROR_USER_ABORT;
+		break;
+	case SCTP_EVENT_T_CHUNK:
+		if (chunk && (SCTP_CID_ABORT == chunk->chunk_hdr->type) &&
+	    	    (ntohs(chunk->chunk_hdr->length) >=
+			(sizeof(struct sctp_chunkhdr) +
+				sizeof(struct sctp_errhdr)))) {
+			error = ((sctp_errhdr_t *)chunk->skb->data)->cause;
+		}
+		break;
+	default:
+		break;
 	}
 
 	event = sctp_ulpevent_make_assoc_change(asoc,

net/sctp/sm_statefuns.c

@@ -2,6 +2,7 @@
  * Copyright (c) 1999-2000 Cisco, Inc.
  * Copyright (c) 1999-2001 Motorola, Inc.
  * Copyright (c) 2001-2002 International Business Machines, Corp.
+ * Copyright (c) 2001-2002 Intel Corp.
  * Copyright (c) 2002      Nokia Corp.
  *
  * This file is part of the SCTP kernel reference Implementation
@@ -502,6 +503,7 @@ sctp_disposition_t sctp_sf_do_5_1D_ce(const sctp_endpoint_t *ep,
 	sctp_chunk_t *repl;
 	sctp_ulpevent_t *ev;
 	int error = 0;
+	sctp_chunk_t *err_chk_p;
 
 	/* If the packet is an OOTB packet which is temporarily on the
 	 * control endpoint, responding with an ABORT.
@@ -521,7 +523,8 @@ sctp_disposition_t sctp_sf_do_5_1D_ce(const sctp_endpoint_t *ep,
 	 * "Z" will reply with a COOKIE ACK chunk after building a TCB
 	 * and moving to the ESTABLISHED state.
 	 */
-	new_asoc = sctp_unpack_cookie(ep, asoc, chunk, GFP_ATOMIC, &error);
+	new_asoc = sctp_unpack_cookie(ep, asoc, chunk, GFP_ATOMIC, &error,
+				      &err_chk_p);
 
 	/* FIXME:
 	 * If the re-build failed, what is the proper error path
@@ -537,6 +540,11 @@ sctp_disposition_t sctp_sf_do_5_1D_ce(const sctp_endpoint_t *ep,
 		case -SCTP_IERROR_NOMEM:
 			goto nomem;
 
+		case -SCTP_IERROR_STALE_COOKIE:
+			sctp_send_stale_cookie_err(ep, asoc, chunk, commands,
+						   err_chk_p);
+			return sctp_sf_pdiscard(ep, asoc, type, arg, commands);
+
 		case -SCTP_IERROR_BAD_SIG:
 		default:
 			return sctp_sf_pdiscard(ep, asoc, type, arg, commands);
@@ -862,8 +870,8 @@ sctp_disposition_t sctp_sf_backbeat_8_3(const sctp_endpoint_t *ep,
 	/* Check if the timestamp looks valid.  */
 	if (time_after(hbinfo->sent_at, jiffies) ||
 	    time_after(jiffies, hbinfo->sent_at + max_interval)) {
-		SCTP_DEBUG_PRINTK("%s: HEARTBEAT ACK with invalid timestamp
-				   received for transport: %p\n",
+		SCTP_DEBUG_PRINTK("%s: HEARTBEAT ACK with invalid timestamp"
+				  "received for transport: %p\n",
 				   __FUNCTION__, link);
 		return SCTP_DISPOSITION_DISCARD;
 	}
@@ -1562,6 +1570,7 @@ sctp_disposition_t sctp_sf_do_5_2_4_dupcook(const sctp_endpoint_t *ep,
 	sctp_association_t *new_asoc;
 	int error = 0;
 	char action;
+	sctp_chunk_t *err_chk_p;
 
 	/* "Decode" the chunk.  We have no optional parameters so we
 	 * are in good shape.
@@ -1575,7 +1584,8 @@ sctp_disposition_t sctp_sf_do_5_2_4_dupcook(const sctp_endpoint_t *ep,
 	 * current association, consider the State Cookie valid even if
 	 * the lifespan is exceeded.
 	 */
-	new_asoc = sctp_unpack_cookie(ep, asoc, chunk, GFP_ATOMIC, &error);
+	new_asoc = sctp_unpack_cookie(ep, asoc, chunk, GFP_ATOMIC, &error,
+				      &err_chk_p);
 
 	/* FIXME:
 	 * If the re-build failed, what is the proper error path
@@ -1591,6 +1601,12 @@ sctp_disposition_t sctp_sf_do_5_2_4_dupcook(const sctp_endpoint_t *ep,
 		case -SCTP_IERROR_NOMEM:
 			goto nomem;
 
+		case -SCTP_IERROR_STALE_COOKIE:
+			sctp_send_stale_cookie_err(ep, asoc, chunk, commands,
+						   err_chk_p);
+			return sctp_sf_pdiscard(ep, asoc, type, arg, commands);
+
+			break;
 		case -SCTP_IERROR_BAD_SIG:
 		default:
 			return sctp_sf_pdiscard(ep, asoc, type, arg, commands);
@@ -1706,7 +1722,47 @@ sctp_disposition_t sctp_sf_shutdown_ack_sent_abort(const sctp_endpoint_t *ep,
 	return sctp_sf_shutdown_sent_abort(ep, asoc, type, arg, commands);
 }
 
-#if 0
+/*
+ * Handle an Error received in COOKIE_ECHOED state.
+ *
+ * Only handle the error type of stale COOKIE Error, the other errors will
+ * be ignored.
+ *
+ * Inputs
+ * (endpoint, asoc, chunk)
+ *
+ * Outputs
+ * (asoc, reply_msg, msg_up, timers, counters)
+ *
+ * The return value is the disposition of the chunk.
+ */
+sctp_disposition_t sctp_sf_cookie_echoed_err(const sctp_endpoint_t *ep,
+					     const sctp_association_t *asoc,
+					     const sctp_subtype_t type,
+					     void *arg,
+					     sctp_cmd_seq_t *commands)
+{
+	sctp_chunk_t *chunk = arg;
+	sctp_errhdr_t *err;
+
+	/* If we have gotten too many failures, give up.  */
+	if (1 + asoc->counters[SCTP_COUNTER_INIT_ERROR] >
+					 asoc->max_init_attempts) {
+		/* INIT_FAILED will issue an ulpevent.  */
+		sctp_add_cmd_sf(commands, SCTP_CMD_INIT_FAILED, SCTP_NULL());
+		return SCTP_DISPOSITION_DELETE_TCB;
+	}
+	err = (sctp_errhdr_t *)(chunk->skb->data);
+
+	/* Process the error here */
+	switch (err->cause) {
+	case SCTP_ERROR_STALE_COOKIE:
+		return sctp_sf_do_5_2_6_stale(ep, asoc, type, arg, commands);
+	default:
+		return sctp_sf_pdiscard(ep, asoc, type, arg, commands);
+	}
+}
+
 /*
  * Handle a Stale COOKIE Error
  *
@@ -1732,47 +1788,30 @@ sctp_disposition_t sctp_sf_shutdown_ack_sent_abort(const sctp_endpoint_t *ep,
  *
  * The return value is the disposition of the chunk.
  */
-sctp_disposition_t do_5_2_6_stale(const sctp_endpoint_t *ep,
-				  const sctp_association_t *asoc,
-				  const sctp_subtype_t type,
-				  void *arg,
-				  sctp_cmd_seq_t *commands)
+sctp_disposition_t sctp_sf_do_5_2_6_stale(const sctp_endpoint_t *ep,
+					  const sctp_association_t *asoc,
+					  const sctp_subtype_t type,
+					  void *arg,
+					  sctp_cmd_seq_t *commands)
 {
 	sctp_chunk_t *chunk = arg;
+	time_t stale;
+	sctp_cookie_preserve_param_t bht;
+	sctp_errhdr_t *err;
+	struct list_head *pos;
+	sctp_transport_t *t;
+	sctp_chunk_t *reply;
+	sctp_bind_addr_t *bp;
+	int attempts;
 
-	/* This is not a real chunk type.  It is a subtype of the
-	 * ERROR chunk type.  The ERROR chunk processing will bring us
-	 * here.
-	 */
-	sctp_chunk_t *in_packet;
-	stp_chunk_t *reply;
-	sctp_inithdr_t initack;
-	__u8 *addrs;
-	int addrs_len;
-	time_t rtt;
-	struct sctpCookiePreserve bht;
+	attempts = asoc->counters[SCTP_COUNTER_INIT_ERROR] + 1;
 
-	/* If we have gotten too many failures, give up.  */
-	if (1 + asoc->counters[SctpCounterInits] > asoc->max_init_attempts) {
-		/* FIXME: Move to new ulpevent.  */
-		retval->event_up = sctp_make_ulp_init_timeout(asoc);
-		if (!retval->event_up)
-			goto nomem;
-		sctp_add_cmd_sf(retval->commands, SCTP_CMD_DELETE_TCB,
-				SCTP_NULL());
+	if (attempts >= asoc->max_init_attempts) {
+		sctp_add_cmd_sf(commands, SCTP_CMD_INIT_FAILED, SCTP_NULL());
 		return SCTP_DISPOSITION_DELETE_TCB;
 	}
 
-	retval->counters[0] = SCTP_COUNTER_INCR;
-	retval->counters[0] = SctpCounterInits;
-	retval->counters[1] = 0;
-	retval->counters[1] = 0;
-
-	/* Calculate the RTT in ms.  */
-	/* BUG--we should get the send time of the HEARTBEAT REQUEST.  */
-	in_packet = chunk;
-	rtt = 1000 * timeval_sub(in_packet->skb->stamp,
-				 asoc->c.state_timestamp);
+	err = (sctp_errhdr_t *)(chunk->skb->data);
 
 	/* When calculating the time extension, an implementation
 	 * SHOULD use the RTT information measured based on the
@@ -1780,28 +1819,48 @@ sctp_disposition_t do_5_2_6_stale(const sctp_endpoint_t *ep,
 	 * more than 1 second beyond the measured RTT, due to long
 	 * State Cookie lifetimes making the endpoint more subject to
 	 * a replay attack.
+	 * Measure of Staleness's unit is usec. (1/1000000 sec)
+	 * Suggested Cookie Life-span Increment's unit is msec.
+	 * (1/1000 sec)
+	 * In general, if you use the suggested cookie life, the value
+	 * found in the field of measure of staleness should be doubled
+	 * to give ample time to retransmit the new cookie and thus
+	 * yield a higher probability of success on the reattempt.
 	 */
-	bht.p = {SCTP_COOKIE_PRESERVE, 8};
-	bht.extraTime = htonl(rtt + 1000);
+	stale = ntohl(*(suseconds_t *)((u8 *)err + sizeof(sctp_errhdr_t)));
+	stale = stale << 1 / 1000;
 
-	initack.init_tag		= htonl(asoc->c.my_vtag);
-	initack.a_rwnd 		        = htonl(atomic_read(&asoc->rnwd));
-	initack.num_outbound_streams    = htons(asoc->streamoutcnt);
-	initack.num_inbound_streams     = htons(asoc->streamincnt);
-	initack.initial_tsn             = htonl(asoc->c.initSeqNumber);
-
-	sctp_get_my_addrs(asoc, &addrs, &addrs_len);
+	bht.param_hdr.type = SCTP_PARAM_COOKIE_PRESERVATIVE;
+	bht.param_hdr.length = htons(sizeof(bht));
+	bht.lifespan_increment = htonl(stale);
 
 	/* Build that new INIT chunk.  */
-	reply = sctp_make_chunk(SCTP_INITIATION, 0,
-				sizeof(initack)
-				+ sizeof(bht)
-				+ addrs_len);
+	bp = (sctp_bind_addr_t *) &asoc->base.bind_addr;
+	reply = sctp_make_init(asoc, bp, GFP_ATOMIC, sizeof(bht));
 	if (!reply)
 		goto nomem;
-	sctp_addto_chunk(reply, sizeof(initack), &initack);
+
 	sctp_addto_chunk(reply, sizeof(bht), &bht);
-	sctp_addto_chunk(reply, addrs_len, addrs);
+
+	/* Cast away the const modifier, as we want to just
+	 * rerun it through as a sideffect.
+	 */
+	sctp_add_cmd_sf(commands, SCTP_CMD_COUNTER_INC,
+			SCTP_COUNTER(SCTP_COUNTER_INIT_ERROR));
+
+	/* If we've sent any data bundled with COOKIE-ECHO we need to resend. */
+	list_for_each(pos, &asoc->peer.transport_addr_list) {
+		t = list_entry(pos, sctp_transport_t, transports);
+		sctp_add_cmd_sf(commands, SCTP_CMD_RETRAN, SCTP_TRANSPORT(t));
+	}
+
+	sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_STOP,
+			SCTP_TO(SCTP_EVENT_TIMEOUT_T1_COOKIE));
+	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
+			SCTP_STATE(SCTP_STATE_COOKIE_WAIT));
+	sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_START,
+			SCTP_TO(SCTP_EVENT_TIMEOUT_T1_INIT));
+
 	sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(reply));
 
 	return SCTP_DISPOSITION_CONSUME;
@@ -1809,7 +1868,6 @@ sctp_disposition_t do_5_2_6_stale(const sctp_endpoint_t *ep,
 nomem:
 	return SCTP_DISPOSITION_NOMEM;
 }
-#endif /* 0 */
 
 /*
  * Process an ABORT.
@@ -3220,7 +3278,7 @@ sctp_disposition_t sctp_sf_do_prm_asoc(const sctp_endpoint_t *ep,
 	 * 1 to 4294967295 (see 5.3.1 for Tag value selection). ...
 	 */
 
-	repl = sctp_make_init(asoc, bp, GFP_ATOMIC);
+	repl = sctp_make_init(asoc, bp, GFP_ATOMIC, 0);
 	if (!repl)
 		goto nomem;
 
@@ -3992,7 +4050,7 @@ sctp_disposition_t sctp_sf_t1_timer_expire(const sctp_endpoint_t *ep,
 		switch (timer) {
 		case SCTP_EVENT_TIMEOUT_T1_INIT:
 			bp = (sctp_bind_addr_t *) &asoc->base.bind_addr;
-			repl = sctp_make_init(asoc, bp, GFP_ATOMIC);
+			repl = sctp_make_init(asoc, bp, GFP_ATOMIC, 0);
 			break;
 
 		case SCTP_EVENT_TIMEOUT_T1_COOKIE:
@@ -4334,3 +4392,25 @@ void sctp_ootb_pkt_free(sctp_packet_t *packet)
 	sctp_transport_free(packet->transport);
 	sctp_packet_free(packet);
 }
+
+/* Send a stale cookie error when a invalid COOKIE ECHO chunk is found  */
+void sctp_send_stale_cookie_err(const sctp_endpoint_t *ep,
+				const sctp_association_t *asoc,
+				const sctp_chunk_t *chunk,
+				sctp_cmd_seq_t *commands,
+				sctp_chunk_t *err_chunk)
+{
+	sctp_packet_t *packet;
+
+	if (err_chunk) {
+		packet = sctp_ootb_pkt_new(asoc, chunk);
+		if (packet) {
+			/* Set the skb to the belonging sock for accounting. */
+			err_chunk->skb->sk = ep->base.sk;
+			sctp_packet_append_chunk(packet, err_chunk);
+			sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
+					SCTP_PACKET(packet));
+		} else
+			sctp_free_chunk (err_chunk);
+	}
+}

net/sctp/sm_statetable.c

@@ -295,7 +295,7 @@ sctp_sm_table_entry_t *sctp_sm_lookup_event(sctp_event_t event_type,
 	/* SCTP_STATE_COOKIE_WAIT */ \
 	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
 	/* SCTP_STATE_COOKIE_ECHOED */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_cookie_echoed_err, .name = "sctp_sf_cookie_echoed_err"}, \
 	/* SCTP_STATE_ESTABLISHED */ \
 	{.fn = sctp_sf_operr_notify, .name = "sctp_sf_operr_notify"}, \
 	/* SCTP_STATE_SHUTDOWN_PENDING */ \

net/sctp/socket.c

@@ -87,12 +87,7 @@ static int sctp_wait_for_sndbuf(sctp_association_t *asoc, long *timeo_p,
 				int msg_len);
 static int sctp_wait_for_packet(struct sock * sk, int *err, long *timeo_p);
 static int sctp_wait_for_connect(sctp_association_t *asoc, long *timeo_p);
-static inline void sctp_sk_addr_set(struct sock *,
-				    const union sctp_addr *newaddr,
-				    union sctp_addr *saveaddr);
-static inline void sctp_sk_addr_restore(struct sock *,
-					const union sctp_addr *);
-static inline int sctp_verify_addr(struct sock *, struct sockaddr *, int);
+static inline int sctp_verify_addr(struct sock *, union sctp_addr *, int);
 static int sctp_bindx_add(struct sock *, struct sockaddr_storage *, int);
 static int sctp_bindx_rem(struct sock *, struct sockaddr_storage *, int);
 static int sctp_do_bind(struct sock *, union sctp_addr *, int);
@@ -133,101 +128,75 @@ int sctp_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 	return retval;
 }
 
-static long sctp_get_port_local(struct sock *, unsigned short);
+static long sctp_get_port_local(struct sock *, union sctp_addr *);
 
-/* Bind a local address either to an endpoint or to an association.  */
-SCTP_STATIC int sctp_do_bind(struct sock *sk, union sctp_addr *newaddr,
-			     int addr_len)
+/* Verify this is a valid sockaddr. */
+static struct sctp_af *sctp_sockaddr_af(struct sctp_opt *opt, 
+					union sctp_addr *addr, int len)
 {
-	sctp_opt_t *sp = sctp_sk(sk);
-	sctp_endpoint_t *ep = sp->ep;
-	sctp_bind_addr_t *bp = &ep->base.bind_addr;
-	unsigned short sa_family = newaddr->sa.sa_family;
-	union sctp_addr tmpaddr, saveaddr;
-	unsigned short *snum;
-	int ret = 0;
+	struct sctp_af *af;
 
-	SCTP_DEBUG_PRINTK("sctp_do_bind(sk: %p, newaddr: %p, addr_len: %d)\n",
-			  sk, newaddr, addr_len);
-
-	/* FIXME: This function needs to handle v4-mapped-on-v6
-	 * addresses!
-	 */
-	if (PF_INET == sk->family) {
-		if (sa_family != AF_INET)
-			return -EINVAL;
-	}
+	/* Check minimum size.  */
+	if (len < sizeof (struct sockaddr))
+		return NULL;
 
-	/* Make a local copy of the new address.  */
-	tmpaddr = *newaddr;
+	/* Does this PF support this AF? */
+	if (!opt->pf->af_supported(addr->sa.sa_family))
+		return NULL;
 
-	switch (sa_family) {
-	case AF_INET:
-		if (addr_len < sizeof(struct sockaddr_in))
-			return -EINVAL;
+	/* If we get this far, af is valid. */
+	af = sctp_get_af_specific(addr->sa.sa_family);
 
-		ret = inet_addr_type(newaddr->v4.sin_addr.s_addr);
+	if (len < af->sockaddr_len)
+		return NULL;
 
-		/* FIXME:
-		 * Should we allow apps to bind to non-local addresses by
-		 * checking the IP sysctl parameter "ip_nonlocal_bind"?
-		 */
-		if (newaddr->v4.sin_addr.s_addr != INADDR_ANY &&
-		    ret != RTN_LOCAL)
-			return -EADDRNOTAVAIL;
+	return af;
+}
 
-		tmpaddr.v4.sin_port = htons(tmpaddr.v4.sin_port);
-		snum = &tmpaddr.v4.sin_port;
-		break;
 
-	case AF_INET6:
-		SCTP_V6(
-			/* FIXME: Hui, please verify this.   Looking at
-			 * the ipv6 code I see a SIN6_LEN_RFC2133 check.
-			 * I'm guessing that scope_id is a newer addition.
-			 */
-			if (addr_len < sizeof(struct sockaddr_in6))
-				return -EINVAL;
+/* Bind a local address either to an endpoint or to an association.  */
+SCTP_STATIC int sctp_do_bind(struct sock *sk, union sctp_addr *addr, int len)
+{
+	sctp_opt_t *sp = sctp_sk(sk);
+	sctp_endpoint_t *ep = sp->ep;
+	sctp_bind_addr_t *bp = &ep->base.bind_addr;
+	struct sctp_af *af;
+	unsigned short snum;
+	int ret = 0;
 
-			/* FIXME - The support for IPv6 multiple types
-			 * of addresses need to be added later.
-			 */
-			ret = sctp_ipv6_addr_type(&newaddr->v6.sin6_addr);
-			tmpaddr.v6.sin6_port = htons(tmpaddr.v6.sin6_port);
-			snum = &tmpaddr.v6.sin6_port;
-			break;
-		)
+	SCTP_DEBUG_PRINTK("sctp_do_bind(sk: %p, newaddr: %p, len: %d)\n",
+			  sk, addr, len);
 
-	default:
+	/* Common sockaddr verification. */
+	af = sctp_sockaddr_af(sp, addr, len);
+	if (!af)
 		return -EINVAL;
-	};
+
+	/* PF specific bind() address verification. */
+	if (!sp->pf->bind_verify(sp, addr))
+		return -EADDRNOTAVAIL;
+
+	snum= ntohs(addr->v4.sin_port);
 
 	SCTP_DEBUG_PRINTK("sctp_do_bind: port: %d, new port: %d\n",
-			  bp->port, *snum);
+			  bp->port, snum);
 
 	/* We must either be unbound, or bind to the same port.  */
-	if (bp->port && (*snum != bp->port)) {
+	if (bp->port && (snum != bp->port)) {
 		SCTP_DEBUG_PRINTK("sctp_do_bind:"
 				  " New port %d does not match existing port "
-				  "%d.\n", *snum, bp->port);
+				  "%d.\n", snum, bp->port);
 		return -EINVAL;
 	}
 
-	if (*snum && *snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE))
+	if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE))
 		return -EACCES;
 
-	/* FIXME - Make socket understand that there might be multiple bind
-	 * addresses and there will be multiple source addresses involved in
-	 * routing and failover decisions.
-	 */
-	sctp_sk_addr_set(sk, &tmpaddr, &saveaddr);
-
 	/* Make sure we are allowed to bind here.
 	 * The function sctp_get_port_local() does duplicate address
 	 * detection.
 	 */
-	if ((ret = sctp_get_port_local(sk, *snum))) {
-		sctp_sk_addr_restore(sk, &saveaddr);
+	if ((ret = sctp_get_port_local(sk, addr))) {
 		if (ret == (long) sk) {
 			/* This endpoint has a conflicting address. */
 			return -EINVAL;
@@ -237,25 +206,32 @@ SCTP_STATIC int sctp_do_bind(struct sock *sk, union sctp_addr *newaddr,
 	}
 
 	/* Refresh ephemeral port.  */
-	if (!*snum)
-		*snum = inet_sk(sk)->num;
+	if (!snum)
+		snum = inet_sk(sk)->num;
+
+
 
-	/* The getsockname() API depends on 'sport' being set.  */
-	inet_sk(sk)->sport = htons(inet_sk(sk)->num);
 
 	/* Add the address to the bind address list.  */
 	sctp_local_bh_disable();
 	sctp_write_lock(&ep->base.addr_lock);
 
 	/* Use GFP_ATOMIC since BHs are disabled.  */
-	if ((ret = sctp_add_bind_addr(bp, &tmpaddr, GFP_ATOMIC))) {
-		sctp_sk_addr_restore(sk, &saveaddr);
-	} else if (!bp->port) {
-		bp->port = *snum;
-	}
+	addr->v4.sin_port = ntohs(addr->v4.sin_port);
+	ret = sctp_add_bind_addr(bp, addr, GFP_ATOMIC);
+	addr->v4.sin_port = htons(addr->v4.sin_port);
+	if (!ret && !bp->port)
+		bp->port = snum;
 
 	sctp_write_unlock(&ep->base.addr_lock);
 	sctp_local_bh_enable();
+
+	/* Copy back into socket for getsockname() use. */
+	if (!ret) {
+		inet_sk(sk)->sport = htons(inet_sk(sk)->num);
+		af->to_sk(addr, sk);
+	}
+
 	return ret;
 }
 
@@ -735,7 +711,7 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout)
 SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *, sctp_cmsgs_t *);
 
 SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
-			     struct msghdr *msg, int size)
+			     struct msghdr *msg, int msg_len)
 {
 	sctp_opt_t *sp;
 	sctp_endpoint_t *ep;
@@ -750,13 +726,12 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 	sctp_assoc_t associd = NULL;
 	sctp_cmsgs_t cmsgs = { 0 };
 	int err;
-	size_t msg_len;
 	sctp_scope_t scope;
 	long timeo;
 	__u16 sinfo_flags = 0;
 
-	SCTP_DEBUG_PRINTK("sctp_sendmsg(sk: %p, msg: %p, "
-			  "size: %d)\n", sk, msg, size);
+	SCTP_DEBUG_PRINTK("sctp_sendmsg(sk: %p, msg: %p, msg_len: %d)\n",
+			  sk, msg, msg_len);
 
 	err = 0;
 	sp = sctp_sk(sk);
@@ -778,12 +753,16 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 	 * For a peeled-off socket, msg_name is ignored.
 	 */
 	if ((SCTP_SOCKET_UDP_HIGH_BANDWIDTH != sp->type) && msg->msg_name) {
-		err = sctp_verify_addr(sk, (struct sockaddr *)msg->msg_name,
-				       msg->msg_namelen);
+		int msg_namelen = msg->msg_namelen;
+	
+		err = sctp_verify_addr(sk, (union sctp_addr *)msg->msg_name, 
+				       msg_namelen);
 		if (err)
 			return err;
 
-		memcpy(&to, msg->msg_name, msg->msg_namelen);
+		if (msg_namelen > sizeof(to))
+			msg_namelen = sizeof(to);
+		memcpy(&to, msg->msg_name, msg_namelen);
 		SCTP_DEBUG_PRINTK("Just memcpy'd. msg_name is "
 				  "0x%x:%u.\n",
 				  to.v4.sin_addr.s_addr, to.v4.sin_port);
@@ -792,8 +771,6 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 		msg_name = msg->msg_name;
 	}
 
-	msg_len = get_user_iov_size(msg->msg_iov, msg->msg_iovlen);
-
 	sinfo = cmsgs.info;
 	sinit = cmsgs.init;
 
@@ -1216,9 +1193,11 @@ SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr
 	 * Otherwise, set MSG_EOR indicating the end of a message. 
 	 */
 	if (skb_len > copied) {
+		msg->msg_flags &= ~MSG_EOR;
+		if (flags & MSG_PEEK)
+			goto out_free;	
 		sctp_skb_pull(skb, copied);
 		skb_queue_head(&sk->receive_queue, skb);
-		msg->msg_flags &= ~MSG_EOR;
 		goto out;
 	} else {
 		 msg->msg_flags |= MSG_EOR;
@@ -1335,6 +1314,16 @@ static inline int sctp_setsockopt_set_peer_addr_params(struct sock *sk,
 	return 0;
 }
 
+static inline int sctp_setsockopt_initmsg(struct sock *sk, char *optval,
+					  int optlen)
+{
+	if (optlen != sizeof(struct sctp_initmsg))
+		return -EINVAL;
+	if (copy_from_user(&sctp_sk(sk)->initmsg, optval, optlen))
+		return -EFAULT;
+	return 0;
+}
+
 /* API 6.2 setsockopt(), getsockopt()
  *
  * Applications use setsockopt() and getsockopt() to set or retrieve
@@ -1355,13 +1344,10 @@ static inline int sctp_setsockopt_set_peer_addr_params(struct sock *sk,
  *   optlen  - the size of the buffer.
  */
 SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
-			   char *optval, int optlen)
+				char *optval, int optlen)
 {
 	int retval = 0;
 	char *tmp;
-	sctp_protocol_t *proto = sctp_get_protocol();
-	struct list_head *pos;
-	sctp_func_t *af;
 
 	SCTP_DEBUG_PRINTK("sctp_setsockopt(sk: %p... optname: %d)\n",
 			  sk, optname);
@@ -1373,14 +1359,9 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 	 * are at all well-founded.
 	 */
 	if (level != SOL_SCTP) {
-		list_for_each(pos, &proto->address_families) {
-			af = list_entry(pos, sctp_func_t, list);
-
-			retval = af->setsockopt(sk, level, optname, optval,
-						optlen);
-			if (retval < 0)
-				goto out_nounlock;
-		}
+		struct sctp_af *af = sctp_sk(sk)->pf->af;
+		retval = af->setsockopt(sk, level, optname, optval, optlen);
+		goto out_nounlock;
 	}
 
 	sctp_lock_sock(sk);
@@ -1430,6 +1411,10 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 							      optlen);
 		break;
 
+	case SCTP_INITMSG:
+		retval = sctp_setsockopt_initmsg(sk, optval, optlen);
+		break;
+
 	default:
 		retval = -ENOPROTOOPT;
 		break;
@@ -1484,7 +1469,7 @@ SCTP_STATIC int sctp_connect(struct sock *sk, struct sockaddr *uaddr,
 		goto out_unlock;
 	}
 
-	err = sctp_verify_addr(sk, uaddr, addr_len);
+	err = sctp_verify_addr(sk, (union sctp_addr *)uaddr, addr_len);
 	if (err)
 		goto out_unlock;	
 
@@ -1938,13 +1923,19 @@ static inline int sctp_getsockopt_get_peer_addr_params(struct sock *sk,
 	return 0;
 }
 
+static inline int sctp_getsockopt_initmsg(struct sock *sk, int len, char *optval, int *optlen)
+{
+	if (len != sizeof(struct sctp_initmsg))
+		return -EINVAL;
+	if (copy_to_user(optval, &sctp_sk(sk)->initmsg, len))
+		return -EFAULT;
+	return 0;
+}
+
 SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 				char *optval, int *optlen)
 {
 	int retval = 0;
-	sctp_protocol_t *proto = sctp_get_protocol();
-	sctp_func_t *af;
-	struct list_head *pos;
 	int len;
 
 	SCTP_DEBUG_PRINTK("sctp_getsockopt(sk: %p, ...)\n", sk);
@@ -1956,13 +1947,10 @@ SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 	 * are at all well-founded.
 	 */
 	if (level != SOL_SCTP) {
-		list_for_each(pos, &proto->address_families) {
-			af = list_entry(pos, sctp_func_t, list);
-			retval = af->getsockopt(sk, level, optname,
-						optval, optlen);
-			if (retval < 0)
-				return retval;
-		}
+		struct sctp_af *af = sctp_sk(sk)->pf->af;
+
+		retval = af->getsockopt(sk, level, optname, optval, optlen);
+		return retval;
 	}
 
 	if (get_user(len, optlen))
@@ -1997,6 +1985,10 @@ SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 							      optlen);
 		break;
 
+	case SCTP_INITMSG:
+		retval = sctp_getsockopt_initmsg(sk, len, optval, optlen);
+		break;
+
 	default:
 		retval = -ENOPROTOOPT;
 		break;
@@ -2030,12 +2022,17 @@ static void sctp_unhash(struct sock *sk)
  */
 static sctp_bind_bucket_t *sctp_bucket_create(sctp_bind_hashbucket_t *head,
 					      unsigned short snum);
-static long sctp_get_port_local(struct sock *sk, unsigned short snum)
+static long sctp_get_port_local(struct sock *sk, union sctp_addr *addr)
 {
 	sctp_bind_hashbucket_t *head; /* hash list */
 	sctp_bind_bucket_t *pp; /* hash list port iterator */
 	sctp_protocol_t *sctp = sctp_get_protocol();
+	unsigned short snum;
 	int ret;
+	
+	/* NOTE:  Remember to put this back to net order. */
+	addr->v4.sin_port = ntohs(addr->v4.sin_port);
+	snum = addr->v4.sin_port;
 
 	SCTP_DEBUG_PRINTK("sctp_get_port() begins, snum=%d\n", snum);
 
@@ -2101,6 +2098,7 @@ static long sctp_get_port_local(struct sock *sk, unsigned short snum)
 		}
 	}
 
+	
 	if (pp != NULL && pp->sk != NULL) {
 		/* We had a port hash table hit - there is an
 		 * available port (pp != NULL) and it is being
@@ -2108,7 +2106,6 @@ static long sctp_get_port_local(struct sock *sk, unsigned short snum)
 		 * socket is going to be sk2.
 		 */
 		int sk_reuse = sk->reuse;
-		union sctp_addr tmpaddr;
 		struct sock *sk2 = pp->sk;
 
 		SCTP_DEBUG_PRINTK("sctp_get_port() found a "
@@ -2116,27 +2113,6 @@ static long sctp_get_port_local(struct sock *sk, unsigned short snum)
 		if (pp->fastreuse != 0 && sk->reuse != 0)
 			goto success;
 
-		/* FIXME - multiple addresses need to be supported
-		 * later.
-		 */
-		switch (sk->family) {
-		case PF_INET:
-			tmpaddr.v4.sin_family = AF_INET;
-			tmpaddr.v4.sin_port = snum;
-			tmpaddr.v4.sin_addr.s_addr = inet_sk(sk)->rcv_saddr;
-			break;
-
-		case PF_INET6:
-			SCTP_V6(tmpaddr.v6.sin6_family = AF_INET6;
-				tmpaddr.v6.sin6_port = snum;
-				tmpaddr.v6.sin6_addr = inet6_sk(sk)->rcv_saddr;
-			)
-			break;
-
-		default:
-			break;
-		};
-
 		/* Run through the list of sockets bound to the port
 		 * (pp->port) [via the pointers bind_next and
 		 * bind_pprev in the struct sock *sk2 (pp->sk)]. On each one,
@@ -2154,8 +2130,7 @@ static long sctp_get_port_local(struct sock *sk, unsigned short snum)
 			if (sk_reuse && sk2->reuse)
 				continue;
 			
-			if (sctp_bind_addr_match(&ep2->base.bind_addr, 
-						 &tmpaddr,
+			if (sctp_bind_addr_match(&ep2->base.bind_addr, addr,
 						 sctp_sk(sk)))
 				goto found;
 		}
@@ -2207,12 +2182,25 @@ static long sctp_get_port_local(struct sock *sk, unsigned short snum)
 	sctp_local_bh_enable();
 
 	SCTP_DEBUG_PRINTK("sctp_get_port() ends, ret=%d\n", ret);
+	addr->v4.sin_port = htons(addr->v4.sin_port);
 	return ret;
 }
 
+/* Assign a 'snum' port to the socket.  If snum == 0, an ephemeral
+ * port is requested.  
+ */
 static int sctp_get_port(struct sock *sk, unsigned short snum)
 {
-	long ret = sctp_get_port_local(sk, snum);
+	long ret;
+	union sctp_addr addr;
+	struct sctp_af *af = sctp_sk(sk)->pf->af;
+
+	/* Set up a dummy address struct from the sk. */
+	af->from_sk(&addr, sk);
+	addr.v4.sin_port = htons(snum);
+
+	/* Note: sk->num gets filled in if ephemeral port request. */
+	ret = sctp_get_port_local(sk, &addr);
 
 	return (ret ? 1 : 0);
 }
@@ -2413,7 +2401,7 @@ void sctp_put_port(struct sock *sk)
 static int sctp_autobind(struct sock *sk)
 {
 	union sctp_addr autoaddr;
-	struct sctp_func *af;
+	struct sctp_af *af;
 	unsigned short port;
 
 	/* Initialize a local sockaddr structure to INADDR_ANY. */
@@ -2537,58 +2525,6 @@ SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *msg,
 	return 0;
 }
 
-/* Setup sk->rcv_saddr before calling get_port().  */
-static inline void sctp_sk_addr_set(struct sock *sk,
-				    const union sctp_addr *newaddr,
-				    union sctp_addr *saveaddr)
-{
-	struct inet_opt *inet = inet_sk(sk);
-
-	saveaddr->sa.sa_family = newaddr->sa.sa_family;
-
-	switch (newaddr->sa.sa_family) {
-	case AF_INET:
-		saveaddr->v4.sin_addr.s_addr = inet->rcv_saddr;
-		inet->rcv_saddr = inet->saddr = newaddr->v4.sin_addr.s_addr;
-		break;
-
-	case AF_INET6:
-		SCTP_V6({
-			struct ipv6_pinfo *np = inet6_sk(sk);
-
-			saveaddr->v6.sin6_addr = np->rcv_saddr;
-			np->rcv_saddr = np->saddr = newaddr->v6.sin6_addr;
-			break;
-		})
-
-	default:
-		break;
-	};
-}
-
-/* Restore sk->rcv_saddr after failing get_port().  */
-static inline void sctp_sk_addr_restore(struct sock *sk, const union sctp_addr *addr)
-{
-	struct inet_opt *inet = inet_sk(sk);
-
-	switch (addr->sa.sa_family) {
-	case AF_INET:
-		inet->rcv_saddr = inet->saddr = addr->v4.sin_addr.s_addr;
-		break;
-
-	case AF_INET6:
-		SCTP_V6({
-			struct ipv6_pinfo *np = inet6_sk(sk);
-
-			np->rcv_saddr = np->saddr = addr->v6.sin6_addr;
-			break;
-		})
-
-	default:
-		break;
-	};
-}
-
 /*
  * Wait for a packet..
  * Note: This function is the same function as in core/datagram.c
@@ -2711,27 +2647,15 @@ static struct sk_buff *sctp_skb_recv_datagram(struct sock *sk, int flags, int no
 }
 
 /* Verify that this is a valid address. */
-static int sctp_verify_addr(struct sock *sk, struct sockaddr *addr, int len)
+static int sctp_verify_addr(struct sock *sk, union sctp_addr *addr, int len)
 {
-	struct sctp_func *af;
-
-	/* Check minimum size.  */
-	if (len < sizeof (struct sockaddr))
-		return -EINVAL;
+	struct sctp_af *af;
 
-	/* Do we support this address family in general? */
-	af = sctp_get_af_specific(addr->sa_family);
+	/* Verify basic sockaddr. */
+	af = sctp_sockaddr_af(sctp_sk(sk), addr, len);
 	if (!af)
 		return -EINVAL;
 
-	/* Does this PF support this AF? */
-	if (!sctp_sk(sk)->pf->af_supported(addr->sa_family))
-		return -EINVAL;
-     
-	/* Verify the minimum for this AF sockaddr. */
-	if (len < af->sockaddr_len)
-		return -EINVAL;
-
 	/* Is this a valid SCTP address?  */
 	if (!af->addr_valid((union sctp_addr *)addr)) 
 		return -EINVAL;

net/sctp/sysctl.c

 /* SCTP kernel reference Implementation 
  * Copyright (c) 2002 International Business Machines Corp.
+ * Copyright (c) 2002 Intel Corp.
  * 
  * This file is part of the SCTP kernel reference Implementation
  * 
@@ -32,6 +33,7 @@
  * Written or modified by: 
  *    Mingqin Liu           <liuming@us.ibm.com>
  *    Jon Grimm             <jgrimm@us.ibm.com>
+ *    Ardelle Fan           <ardelle.fan@intel.com>
  *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
@@ -70,6 +72,9 @@ static ctl_table sctp_table[] = {
 	{ NET_SCTP_HB_INTERVAL, "hb_interval",
 	  &sctp_proto.hb_interval, sizeof(int), 0644, NULL,
 	  &proc_dointvec_jiffies, &sysctl_jiffies },
+	{ NET_SCTP_PRESERVE_ENABLE, "cookie_preserve_enable",
+	  &sctp_proto.cookie_preserve_enable, sizeof(int), 0644, NULL,
+	  &proc_dointvec_jiffies, &sysctl_jiffies },
 	{ NET_SCTP_RTO_ALPHA, "rto_alpha_exp_divisor",
 	  &sctp_proto.rto_alpha, sizeof(int), 0644, NULL,
 	  &proc_dointvec },

net/sctp/transport.c

@@ -207,7 +207,7 @@ void sctp_transport_route(sctp_transport_t *transport, union sctp_addr *saddr,
 			  struct sctp_opt *opt)
 {
 	sctp_association_t *asoc = transport->asoc;
-	struct sctp_func *af = transport->af_specific;
+	struct sctp_af *af = transport->af_specific;
 	union sctp_addr *daddr = &transport->ipaddr;
 	sctp_bind_addr_t *bp;
 	rwlock_t *addr_lock;