Merge touki.austin.ibm.com:/home/jgrimm/bk/linux-2.5.66

into touki.austin.ibm.com:/home/jgrimm/bk/lksctp-2.5.work

include/net/sctp/constants.h

@@ -138,12 +138,10 @@ typedef enum {
  */
 
 typedef union {
-
 	sctp_cid_t chunk;
 	sctp_event_timeout_t timeout;
 	sctp_event_other_t other;
 	sctp_event_primitive_t primitive;
-
 } sctp_subtype_t;
 
 #define SCTP_SUBTYPE_CONSTRUCTOR(_name, _type, _elt) \
@@ -421,9 +419,9 @@ typedef enum {
 
 /* Reasons to retransmit. */
 typedef enum {
-	SCTP_RETRANSMIT_T3_RTX,
-	SCTP_RETRANSMIT_FAST_RTX,
-	SCTP_RETRANSMIT_PMTU_DISCOVERY,
+	SCTP_RTXR_T3_RTX,
+	SCTP_RTXR_FAST_RTX,
+	SCTP_RTXR_PMTUD,
 } sctp_retransmit_reason_t;
 
 /* Reasons to lower cwnd. */

include/net/sctp/sctp.h

@@ -123,14 +123,14 @@
  */
 extern struct sctp_protocol sctp_proto;
 extern struct sock *sctp_get_ctl_sock(void);
-extern int sctp_copy_local_addr_list(struct sctp_protocol  *, 
+extern int sctp_copy_local_addr_list(struct sctp_protocol  *,
 				     struct sctp_bind_addr *,
 				     sctp_scope_t, int priority, int flags);
 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
+ * sctp/socket.c
  */
 extern int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb);
 extern int sctp_inet_listen(struct socket *sock, int backlog);
@@ -139,7 +139,7 @@ extern unsigned int sctp_poll(struct file *file, struct socket *sock,
 		poll_table *wait);
 
 /*
- * sctp_primitive.c
+ * sctp/primitive.c
  */
 extern int sctp_primitive_ASSOCIATE(sctp_association_t *, void *arg);
 extern int sctp_primitive_SHUTDOWN(sctp_association_t *, void *arg);
@@ -148,14 +148,14 @@ extern int sctp_primitive_SEND(sctp_association_t *, void *arg);
 extern int sctp_primitive_REQUESTHEARTBEAT(sctp_association_t *, void *arg);
 
 /*
- * sctp_crc32c.c
+ * sctp/crc32c.c
  */
 extern __u32 sctp_start_cksum(__u8 *ptr, __u16 count);
 extern __u32 sctp_update_cksum(__u8 *ptr, __u16 count, __u32 cksum);
 extern __u32 sctp_end_cksum(__u32 cksum);
 
 /*
- * sctp_input.c
+ * sctp/input.c
  */
 extern int sctp_rcv(struct sk_buff *skb);
 extern void sctp_v4_err(struct sk_buff *skb, u32 info);
@@ -170,9 +170,16 @@ extern void __sctp_unhash_endpoint(sctp_endpoint_t *);
 extern sctp_association_t *__sctp_lookup_association(const union sctp_addr *,
 						     const union sctp_addr *,
 						     struct sctp_transport **);
-
+extern struct sock *sctp_err_lookup(int family, struct sk_buff *,
+				    struct sctphdr *, struct sctp_endpoint **, 
+				    struct sctp_association **, 
+				    struct sctp_transport **);
+extern void sctp_err_finish(struct sock *, struct sctp_endpoint *, 
+			    struct sctp_association *);
+extern void sctp_icmp_frag_needed(struct sock *, struct sctp_association *,
+				  struct sctp_transport *t, __u32 pmtu);
 /*
- * sctp_hashdriver.c
+ * sctp/hashdriver.c
  */
 extern void sctp_hash_digest(const char *secret, const int secret_len,
 			     const char *text, const int text_len,
@@ -184,9 +191,7 @@ extern void sctp_hash_digest(const char *secret, const int secret_len,
 
 
 #ifdef TEST_FRAME
-
 #include <test_frame.h>
-
 #else
 
 /* spin lock wrappers. */
@@ -312,7 +317,6 @@ static inline void sctp_sysctl_register(void) { return; }
 static inline void sctp_sysctl_unregister(void) { return; }
 #endif
 
-
 /* Size of Supported Address Parameter for 'x' address types. */
 #define SCTP_SAT_LEN(x) (sizeof(struct sctp_paramhdr) + (x) * sizeof(__u16))
 
@@ -320,19 +324,15 @@ static inline void sctp_sysctl_unregister(void) { return; }
 
 extern int sctp_v6_init(void);
 extern void sctp_v6_exit(void);
-static inline int sctp_ipv6_addr_type(const struct in6_addr *addr)
-{
-	return ipv6_addr_type((struct in6_addr*) addr);
-}
+extern void sctp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
+			int type, int code, int offset, __u32 info);
 
-#else /* #ifdef defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
+#else /* #ifdef defined(CONFIG_IPV6) */
 
-#define sctp_ipv6_addr_type(a) 0
 static inline int sctp_v6_init(void) { return 0; }
 static inline void sctp_v6_exit(void) { return; }
 
-#endif /* #ifdef defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
-
+#endif /* #if defined(CONFIG_IPV6) */
 
 /* Map an association to an assoc_id. */
 static inline sctp_assoc_t sctp_assoc2id(const sctp_association_t *asoc)
@@ -414,13 +414,22 @@ static inline __s32 sctp_jitter(__u32 rto)
 	sctp_rand ^= (sctp_rand << 12);
 	sctp_rand ^= (sctp_rand >> 20);
 
-	/* Choose random number from 0 to rto, then move to -50% ~ +50% 
-	 * of rto. 
+	/* Choose random number from 0 to rto, then move to -50% ~ +50%
+	 * of rto.
 	 */
 	ret = sctp_rand % rto - (rto >> 1);
 	return ret;
 }
 
+/* Break down data chunks at this point.  */
+static inline int sctp_frag_point(int pmtu)
+{
+	pmtu -= SCTP_IP_OVERHEAD + sizeof(struct sctp_data_chunk);
+	pmtu -= sizeof(struct sctp_sack_chunk);
+
+	return pmtu;
+}
+
 /* Walk through a list of TLV parameters.  Don't trust the
  * individual parameter lengths and instead depend on
  * the chunk length to indicate when to stop.  Make sure
@@ -537,7 +546,7 @@ struct sctp_sock {
 	struct sock	  sk;
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 	struct ipv6_pinfo *pinet6;
-#endif /* CONFIG_IPV6 || CONFIG_IPV6_MODULE */
+#endif /* CONFIG_IPV6 */
 	struct inet_opt	  inet;
 	struct sctp_opt	  sctp;
 };
@@ -550,7 +559,7 @@ struct sctp6_sock {
 	struct sctp_opt	  sctp;
 	struct ipv6_pinfo inet6;
 };
-#endif /* CONFIG_IPV6 || CONFIG_IPV6_MODULE */
+#endif /* CONFIG_IPV6 */
 
 #define sctp_sk(__sk) (&((struct sctp_sock *)__sk)->sctp)
 

include/net/sctp/structs.h

@@ -590,13 +590,16 @@ struct sctp_packet {
 	/* This packet should advertise ECN capability to the network
 	 * via the ECT bit.
 	 */
-	int ecn_capable;
+	char ecn_capable;
 
 	/* This packet contains a COOKIE-ECHO chunk. */
-	int has_cookie_echo;
+	char has_cookie_echo;
+
+	/* This packet containsa SACK chunk. */
+	char has_sack; 
 
 	/* SCTP cannot fragment this packet. So let ip fragment it. */
-	int ipfragok;
+	char ipfragok;
 
 	int malloced;
 };

net/sctp/associola.c

@@ -421,8 +421,7 @@ struct sctp_transport *sctp_assoc_add_peer(struct sctp_association *asoc,
 	SCTP_DEBUG_PRINTK("sctp_assoc_add_peer:association %p PMTU set to "
 			  "%d\n", asoc, asoc->pmtu);
 
-	asoc->frag_point = asoc->pmtu;
-	asoc->frag_point -= SCTP_IP_OVERHEAD + sizeof(struct sctp_data_chunk);
+	asoc->frag_point = sctp_frag_point(asoc->pmtu);
 
 	/* The asoc->peer.port might not be meaningful yet, but
 	 * initialize the packet structure anyway.
@@ -658,32 +657,21 @@ int sctp_cmp_addr_exact(const union sctp_addr *ss1,
 }
 
 /* Return an ecne chunk to get prepended to a packet.
- * Note:  We are sly and return a shared, prealloced chunk.
+ * Note:  We are sly and return a shared, prealloced chunk.  FIXME:
+ * No we don't, but we could/should.
  */
-sctp_chunk_t *sctp_get_ecne_prepend(sctp_association_t *asoc)
+sctp_chunk_t *sctp_get_ecne_prepend(struct sctp_association *asoc)
 {
-	sctp_chunk_t *chunk;
-	int need_ecne;
-	__u32 lowest_tsn;
+	struct sctp_chunk *chunk;
 
-	/* Can be called from task or bh.   Both need_ecne and
-	 * last_ecne_tsn are written during bh.
+	/* Send ECNE if needed. 
+	 * Not being able to allocate a chunk here is not deadly. 
 	 */
-	need_ecne = asoc->need_ecne;
-	lowest_tsn = asoc->last_ecne_tsn;
-
-	if (need_ecne) {
-		chunk = sctp_make_ecne(asoc, lowest_tsn);
-
-		/* ECNE is not mandatory to the flow.  Being unable to
-		 * alloc mem is not deadly.  We are just unable to help
-		 * out the network.  If we run out of memory, just return
-		 * NULL.
-		 */
-	} else {
+	if (asoc->need_ecne)
+		chunk = sctp_make_ecne(asoc, asoc->last_ecne_tsn);
+	else
 		chunk = NULL;
-	}
-
+	
 	return chunk;
 }
 
@@ -986,8 +974,7 @@ void sctp_assoc_sync_pmtu(sctp_association_t *asoc)
 
 	if (pmtu) {
 		asoc->pmtu = pmtu;
-		asoc->frag_point = pmtu - (SCTP_IP_OVERHEAD +
-					   sizeof(sctp_data_chunk_t));
+		asoc->frag_point = sctp_frag_point(pmtu);
 	}
 
 	SCTP_DEBUG_PRINTK("%s: asoc:%p, pmtu:%d, frag_point:%d\n",

net/sctp/input.c

@@ -207,21 +207,19 @@ int sctp_rcv(struct sk_buff *skb)
 	 */
 	sctp_bh_lock_sock(sk);
 
-	if (sock_owned_by_user(sk)) {
+	if (sock_owned_by_user(sk))
 		sk_add_backlog(sk, (struct sk_buff *) chunk);
-	} else {
+	else
 		sctp_backlog_rcv(sk, (struct sk_buff *) chunk);
-	}
 
 	/* Release the sock and any reference counts we took in the
 	 * lookup calls.
 	 */
 	sctp_bh_unlock_sock(sk);
-	if (asoc) {
+	if (asoc)
 		sctp_association_put(asoc);
-	} else {
+	else
 		sctp_endpoint_put(ep);
-	}
 	sock_put(sk);
 	return ret;
 
@@ -268,10 +266,8 @@ int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb)
 }
 
 /* Handle icmp frag needed error. */
-static inline void sctp_icmp_frag_needed(struct sock *sk,
-					 sctp_association_t *asoc,
-					 struct sctp_transport *transport,
-					 __u32 pmtu)
+void sctp_icmp_frag_needed(struct sock *sk, struct sctp_association *asoc,
+			   struct sctp_transport *t, __u32 pmtu)
 {
 	if (unlikely(pmtu < SCTP_DEFAULT_MINSEGMENT)) {
 		printk(KERN_WARNING "%s: Reported pmtu %d too low, "
@@ -280,54 +276,38 @@ static inline void sctp_icmp_frag_needed(struct sock *sk,
 		pmtu = SCTP_DEFAULT_MINSEGMENT;
 	}
 
-	if (!sock_owned_by_user(sk) && transport && (transport->pmtu != pmtu)) {
-		transport->pmtu = pmtu;
+	if (!sock_owned_by_user(sk) && t && (t->pmtu != pmtu)) {
+		t->pmtu = pmtu;
 		sctp_assoc_sync_pmtu(asoc);
-		sctp_retransmit(&asoc->outqueue, transport,
-				SCTP_RETRANSMIT_PMTU_DISCOVERY );
+		sctp_retransmit(&asoc->outqueue, t, SCTP_RTXR_PMTUD);
 	}
 }
 
-/*
- * This routine is called by the ICMP module when it gets some
- * sort of error condition.  If err < 0 then the socket should
- * be closed and the error returned to the user.  If err > 0
- * it's just the icmp type << 8 | icmp code.  After adjustment
- * header points to the first 8 bytes of the sctp header.  We need
- * to find the appropriate port.
- *
- * The locking strategy used here is very "optimistic". When
- * someone else accesses the socket the ICMP is just dropped
- * and for some paths there is no check at all.
- * A more general error queue to queue errors for later handling
- * is probably better.
- *
- */
-void sctp_v4_err(struct sk_buff *skb, __u32 info)
+/* Common lookup code for icmp/icmpv6 error handler. */
+struct sock *sctp_err_lookup(int family, struct sk_buff *skb,
+			     struct sctphdr *sctphdr,
+			     struct sctp_endpoint **epp, 
+			     struct sctp_association **app, 
+			     struct sctp_transport **tpp)
 {
-	struct iphdr *iph = (struct iphdr *)skb->data;
-	struct sctphdr *sh = (struct sctphdr *)(skb->data + (iph->ihl <<2));
-	int type = skb->h.icmph->type;
-	int code = skb->h.icmph->code;
-	union sctp_addr saddr, daddr;
-	struct inet_opt *inet;
+	union sctp_addr saddr;
+	union sctp_addr daddr;
+	struct sctp_af *af;
 	struct sock *sk = NULL;
-	sctp_endpoint_t *ep = NULL;
-	sctp_association_t *asoc = NULL;
-	struct sctp_transport *transport;
-	int err;
+	struct sctp_endpoint *ep = NULL;
+	struct sctp_association *asoc = NULL;
+	struct sctp_transport *transport = NULL;
 
-	if (skb->len < ((iph->ihl << 2) + 8)) {
-		ICMP_INC_STATS_BH(IcmpInErrors);
-		return;
+	*app = NULL; *epp = NULL; *tpp = NULL;
+
+	af = sctp_get_af_specific(family);
+	if (unlikely(!af)) {
+		return NULL;
 	}
 
-	saddr.v4.sin_family = AF_INET;
-	saddr.v4.sin_port = ntohs(sh->source);
-	memcpy(&saddr.v4.sin_addr.s_addr, &iph->saddr, sizeof(struct in_addr));	
-	daddr.v4.sin_family = AF_INET;
-	daddr.v4.sin_port = ntohs(sh->dest);
-	memcpy(&daddr.v4.sin_addr.s_addr, &iph->daddr, sizeof(struct in_addr));	
+	/* Initialize local addresses for lookups. */
+	af->from_skb(&saddr, skb, 1);
+	af->from_skb(&daddr, skb, 0);
 
 	/* Look for an association that matches the incoming ICMP error 
 	 * packet.
@@ -340,13 +320,12 @@ void sctp_v4_err(struct sk_buff *skb, __u32 info)
 		 */ 
 		ep = __sctp_rcv_lookup_endpoint(&daddr);
 		if (!ep) {
-			ICMP_INC_STATS_BH(IcmpInErrors);
-			return;
+			return NULL;
 		}
 	}
 
 	if (asoc) {
-		if (ntohl(sh->vtag) != asoc->c.peer_vtag) {
+		if (ntohl(sctphdr->vtag) != asoc->c.peer_vtag) {
 			ICMP_INC_STATS_BH(IcmpInErrors);
 			goto out;
 		}
@@ -355,12 +334,90 @@ void sctp_v4_err(struct sk_buff *skb, __u32 info)
 		sk = ep->base.sk;
 
 	sctp_bh_lock_sock(sk);
+
 	/* If too many ICMPs get dropped on busy
 	 * servers this needs to be solved differently.
 	 */
 	if (sock_owned_by_user(sk))
 		NET_INC_STATS_BH(LockDroppedIcmps);
 
+	*epp = ep;
+	*app = asoc;
+	*tpp = transport;
+	return sk;
+	
+out:
+	sock_put(sk);
+	if (asoc)
+		sctp_association_put(asoc);
+	if (ep)	
+		sctp_endpoint_put(ep);
+	return NULL;
+}
+
+/* Common cleanup code for icmp/icmpv6 error handler. */
+void sctp_err_finish(struct sock *sk, struct sctp_endpoint *ep, 
+		     struct sctp_association *asoc)
+{
+	sctp_bh_unlock_sock(sk);
+	sock_put(sk);
+	if (asoc)
+		sctp_association_put(asoc);
+	if (ep)	
+		sctp_endpoint_put(ep);
+}
+
+/*
+ * This routine is called by the ICMP module when it gets some
+ * sort of error condition.  If err < 0 then the socket should
+ * be closed and the error returned to the user.  If err > 0
+ * it's just the icmp type << 8 | icmp code.  After adjustment
+ * header points to the first 8 bytes of the sctp header.  We need
+ * to find the appropriate port.
+ *
+ * The locking strategy used here is very "optimistic". When
+ * someone else accesses the socket the ICMP is just dropped
+ * and for some paths there is no check at all.
+ * A more general error queue to queue errors for later handling
+ * is probably better.
+ *
+ */
+void sctp_v4_err(struct sk_buff *skb, __u32 info)
+{
+	struct iphdr *iph = (struct iphdr *)skb->data;
+	struct sctphdr *sh = (struct sctphdr *)(skb->data + (iph->ihl <<2));
+	int type = skb->h.icmph->type;
+	int code = skb->h.icmph->code;
+	struct sock *sk;
+	sctp_endpoint_t *ep;
+	sctp_association_t *asoc;
+	struct sctp_transport *transport;
+	struct inet_opt *inet;
+	char *saveip, *savesctp;
+	int err;
+
+	if (skb->len < ((iph->ihl << 2) + 8)) {
+		ICMP_INC_STATS_BH(IcmpInErrors);
+		return;
+	}
+
+	/* Fix up skb to look at the embedded net header. */
+	saveip = skb->nh.raw;
+	savesctp  = skb->h.raw;
+	skb->nh.iph = iph;
+	skb->h.raw = (char *)sh;
+	sk = sctp_err_lookup(AF_INET, skb, sh, &ep, &asoc, &transport);
+	/* Put back, the original pointers. */
+	skb->nh.raw = saveip;
+	skb->h.raw = savesctp;
+	if (!sk) {
+		ICMP_INC_STATS_BH(IcmpInErrors);
+		return;
+	}
+	/* Warning:  The sock lock is held.  Remember to call 
+	 * sctp_err_finish!
+	 */
+
 	switch (type) {
 	case ICMP_PARAMETERPROB:
 		err = EPROTO;
@@ -399,13 +456,7 @@ void sctp_v4_err(struct sk_buff *skb, __u32 info)
 	}
 
 out_unlock:
-	sctp_bh_unlock_sock(sk);
-out:
-	sock_put(sk);
-	if (asoc)
-		sctp_association_put(asoc);
-	if (ep)	
-		sctp_endpoint_put(ep);
+	sctp_err_finish(sk, ep, asoc);
 }
 
 /*
@@ -782,8 +833,3 @@ sctp_association_t *__sctp_rcv_lookup(struct sk_buff *skb,
 
 	return asoc;
 }
-
-
-
-
-

net/sctp/ipv6.c

 /* SCTP kernel reference Implementation
  * Copyright (c) 2001 Nokia, Inc.
  * Copyright (c) 2001 La Monte H.P. Yarroll
- * Copyright (c) 2002 International Business Machines, Corp.
+ * Copyright (c) 2002-2003 International Business Machines, Corp.
  *
  * This file is part of the SCTP kernel reference Implementation
  *
@@ -88,17 +88,62 @@ extern struct notifier_block sctp_inetaddr_notifier;
         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,
-			       int type, int code, int offset, __u32 info)
+/* ICMP error handler. */
+void sctp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
+		 int type, int code, int offset, __u32 info)
 {
-	/* BUG.  WRITE ME.  */
+	struct ipv6hdr *iph = (struct ipv6hdr *)skb->data;
+	struct sctphdr *sh = (struct sctphdr *)(skb->data + offset);
+	struct sock *sk;
+	sctp_endpoint_t *ep;
+	sctp_association_t *asoc;
+	struct sctp_transport *transport;
+	struct ipv6_pinfo *np;
+	char *saveip, *savesctp;
+	int err;
+
+	/* Fix up skb to look at the embedded net header. */
+	saveip = skb->nh.raw;
+	savesctp  = skb->h.raw;
+	skb->nh.ipv6h = iph;
+	skb->h.raw = (char *)sh;
+	sk = sctp_err_lookup(AF_INET6, skb, sh, &ep, &asoc, &transport);
+	/* Put back, the original pointers. */
+	skb->nh.raw = saveip;
+	skb->h.raw = savesctp;
+	if (!sk) {
+		ICMP6_INC_STATS_BH(Icmp6InErrors);
+		return;
+	}
+
+	/* Warning:  The sock lock is held.  Remember to call 
+	 * sctp_err_finish!
+	 */
+
+	switch (type) {
+	case ICMPV6_PKT_TOOBIG:
+		sctp_icmp_frag_needed(sk, asoc, transport, ntohl(info));
+		goto out_unlock;
+	default:
+		break;
+	}
+
+	np = inet6_sk(sk);
+	icmpv6_err_convert(type, code, &err);
+	if (!sock_owned_by_user(sk) && np->recverr) {
+		sk->err = err;
+		sk->error_report(sk);
+	} else {  /* Only an error on timeout */
+		sk->err_soft = err;
+	}
+
+out_unlock:
+	sctp_err_finish(sk, ep, asoc);
 }
 
 /* Based on tcp_v6_xmit() in tcp_ipv6.c. */
-static inline int sctp_v6_xmit(struct sk_buff *skb,
-			       struct sctp_transport *transport, int ipfragok)
+static int sctp_v6_xmit(struct sk_buff *skb, struct sctp_transport *transport, 
+			int ipfragok)
 {
 	struct sock *sk = skb->sk;
 	struct ipv6_pinfo *np = inet6_sk(sk);
@@ -110,9 +155,9 @@ static inline int sctp_v6_xmit(struct sk_buff *skb,
 
 	/* Fill in the dest address from the route entry passed with the skb
 	 * and the source address from the transport.
-	 */ 
+	 */
 	fl.fl6_dst = &rt6->rt6i_dst.addr;
-	fl.fl6_src = &transport->saddr.v6.sin6_addr; 
+	fl.fl6_src = &transport->saddr.v6.sin6_addr;
 
 	fl.fl6_flowlabel = np->flow_label;
 	IP6_ECN_flow_xmit(sk, fl.fl6_flowlabel);
@@ -174,7 +219,7 @@ struct dst_entry *sctp_v6_get_dst(sctp_association_t *asoc,
 
 /* Returns the number of consecutive initial bits that match in the 2 ipv6
  * addresses.
- */ 
+ */
 static inline int sctp_v6_addr_match_len(union sctp_addr *s1,
 					 union sctp_addr *s2)
 {
@@ -186,7 +231,7 @@ static inline int sctp_v6_addr_match_len(union sctp_addr *s1,
 		__u32 a1xora2;
 
 		a1xora2 = a1->s6_addr32[i] ^ a2->s6_addr32[i];
-		
+
 		if ((j = fls(ntohl(a1xora2))))
 			return (i * 32 + 32 - j);
 	}
@@ -196,7 +241,7 @@ static inline int sctp_v6_addr_match_len(union sctp_addr *s1,
 
 /* Fills in the source address(saddr) based on the destination address(daddr)
  * and asoc's bind address list.
- */   
+ */
 void sctp_v6_get_saddr(sctp_association_t *asoc, struct dst_entry *dst,
 		       union sctp_addr *daddr, union sctp_addr *saddr)
 {
@@ -432,7 +477,7 @@ static sctp_scope_t sctp_v6_scope(union sctp_addr *addr)
 	return retval;
 }
 
-/* Create and initialize a new sk for the socket to be returned by accept(). */ 
+/* Create and initialize a new sk for the socket to be returned by accept(). */
 struct sock *sctp_v6_create_accept_sk(struct sock *sk,
 				      struct sctp_association *asoc)
 {
@@ -469,11 +514,11 @@ struct sock *sctp_v6_create_accept_sk(struct sock *sk,
 
 	memcpy(newnp, np, sizeof(struct ipv6_pinfo));
 
-	ipv6_addr_copy(&newnp->daddr, &asoc->peer.primary_addr.v6.sin6_addr); 
+	ipv6_addr_copy(&newnp->daddr, &asoc->peer.primary_addr.v6.sin6_addr);
 
 	newinet->sport = inet->sport;
 	newinet->dport = asoc->peer.port;
-	
+
 #ifdef INET_REFCNT_DEBUG
 	atomic_inc(&inet6_sock_nr);
 	atomic_inc(&inet_sock_nr);
@@ -623,11 +668,11 @@ static int sctp_inet6_bind_verify(struct sctp_opt *opt, union sctp_addr *addr)
 /* Fill in Supported Address Type information for INIT and INIT-ACK
  * chunks.   Note: In the future, we may want to look at sock options
  * to determine whether a PF_INET6 socket really wants to have IPV4
- * addresses.  
+ * addresses.
  * Returns number of addresses supported.
  */
 static int sctp_inet6_supported_addrs(const struct sctp_opt *opt,
-				      __u16 *types) 
+				      __u16 *types)
 {
 	types[0] = SCTP_PARAM_IPV4_ADDRESS;
 	types[1] = SCTP_PARAM_IPV6_ADDRESS;

net/sctp/output.c

@@ -79,6 +79,7 @@ struct sctp_packet *sctp_packet_config(struct sctp_packet *packet,
 	packet->ecn_capable = ecn_capable;
 	packet->get_prepend_chunk = prepend_handler;
 	packet->has_cookie_echo = 0;
+	packet->has_sack = 0;
 	packet->ipfragok = 0;
 
 	/* We might need to call the prepend_handler right away.  */
@@ -100,6 +101,7 @@ struct sctp_packet *sctp_packet_init(struct sctp_packet *packet,
 	packet->ecn_capable = 0;
 	packet->get_prepend_chunk = NULL;
 	packet->has_cookie_echo = 0;
+	packet->has_sack = 0;
 	packet->ipfragok = 0;
 	packet->malloced = 0;
 	sctp_packet_reset(packet);
@@ -155,6 +157,37 @@ sctp_xmit_t sctp_packet_transmit_chunk(struct sctp_packet *packet,
 	return retval;
 }
 
+/* Try to bundle a SACK with the packet. */
+static sctp_xmit_t sctp_packet_bundle_sack(struct sctp_packet *pkt,
+					   struct sctp_chunk *chunk)
+{
+	sctp_xmit_t retval = SCTP_XMIT_OK;
+
+	/* If sending DATA and haven't aleady bundled a SACK, try to
+	 * bundle one in to the packet.
+	 */
+	if (sctp_chunk_is_data(chunk) && !pkt->has_sack && 
+	    !pkt->has_cookie_echo) {
+		struct sctp_association *asoc;
+		asoc = pkt->transport->asoc;
+			
+		if (asoc->a_rwnd > asoc->rwnd) {
+			struct sctp_chunk *sack;
+			asoc->a_rwnd = asoc->rwnd;
+			sack = sctp_make_sack(asoc);
+			if (sack) {
+				struct timer_list *timer;
+				retval = sctp_packet_append_chunk(pkt, sack);
+				asoc->peer.sack_needed = 0;
+				timer = &asoc->timers[SCTP_EVENT_TIMEOUT_SACK];
+				if (timer_pending(timer) && del_timer(timer))
+					sctp_association_put(asoc);
+			}
+		}
+	}
+	return retval;
+}
+
 /* Append a chunk to the offered packet reporting back any inability to do
  * so.
  */
@@ -167,6 +200,10 @@ sctp_xmit_t sctp_packet_append_chunk(struct sctp_packet *packet,
 	size_t pmtu;
 	int too_big;
 
+	retval = sctp_packet_bundle_sack(packet, chunk);
+	if (retval != SCTP_XMIT_OK)
+		goto finish;
+		
 	pmtu  = ((packet->transport->asoc) ?
 		 (packet->transport->asoc->pmtu) :
 		 (packet->transport->pmtu));
@@ -216,9 +253,10 @@ sctp_xmit_t sctp_packet_append_chunk(struct sctp_packet *packet,
 		retval = sctp_packet_append_data(packet, chunk);
 		if (SCTP_XMIT_OK != retval)
 			goto finish;
-	} else if (SCTP_CID_COOKIE_ECHO == chunk->chunk_hdr->type) {
+	} else if (SCTP_CID_COOKIE_ECHO == chunk->chunk_hdr->type)
 		packet->has_cookie_echo = 1;
-	}
+	else if (SCTP_CID_SACK == chunk->chunk_hdr->type) 
+		packet->has_sack = 1;
 
 	/* It is OK to send this chunk.  */
 	__skb_queue_tail(&packet->chunks, (struct sk_buff *)chunk);

net/sctp/outqueue.c

@@ -357,7 +357,7 @@ void sctp_retransmit(struct sctp_outq *q, struct sctp_transport *transport,
 	__u8 fast_retransmit = 0;
 
 	switch(reason) {
-	case SCTP_RETRANSMIT_T3_RTX:
+	case SCTP_RTXR_T3_RTX:
 		sctp_transport_lower_cwnd(transport, SCTP_LOWER_CWND_T3_RTX);
 		/* Update the retran path if the T3-rtx timer has expired for
 		 * the current retran path.
@@ -365,10 +365,11 @@ void sctp_retransmit(struct sctp_outq *q, struct sctp_transport *transport,
 		if (transport == transport->asoc->peer.retran_path)
 			sctp_assoc_update_retran_path(transport->asoc);
 		break;
-	case SCTP_RETRANSMIT_FAST_RTX:
+	case SCTP_RTXR_FAST_RTX:
 		sctp_transport_lower_cwnd(transport, SCTP_LOWER_CWND_FAST_RTX);
 		fast_retransmit = 1;
 		break;
+	case SCTP_RTXR_PMTUD:
 	default:
 		break;
 	}
@@ -876,7 +877,7 @@ int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout)
 		start_timer = 0;
 		queue = &q->out;
 
-		while (NULL != (chunk = sctp_outq_dequeue_data(q))) {
+		while ((chunk = sctp_outq_dequeue_data(q))) {
 			/* RFC 2960 6.5 Every DATA chunk MUST carry a valid
 			 * stream identifier.
 			 */
@@ -891,9 +892,7 @@ int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout)
 				if (ev)
 					sctp_ulpq_tail_event(&asoc->ulpq, ev);
 
-				/* Free the chunk. This chunk is not on any
-				 * list yet, just free it.
-				 */
+				/* Free the chunk. */
 				sctp_free_chunk(chunk);
 				continue;
 			}
@@ -1572,7 +1571,7 @@ static void sctp_check_transmitted(struct sctp_outq *q,
 
 	if (transport) {
 		if (do_fast_retransmit)
-			sctp_retransmit(q, transport, SCTP_RETRANSMIT_FAST_RTX);
+			sctp_retransmit(q, transport, SCTP_RTXR_FAST_RTX);
 
 		SCTP_DEBUG_PRINTK("%s: transport: %p, cwnd: %d, "
 				  "ssthresh: %d, flight_size: %d, pba: %d\n",

net/sctp/sm_sideeffect.c

@@ -55,1202 +55,1128 @@
 #include <net/sctp/sctp.h>
 #include <net/sctp/sm.h>
 
-/* Do forward declarations of static functions.  */
-static void sctp_do_ecn_ce_work(sctp_association_t *,__u32 lowest_tsn);
-static sctp_chunk_t *sctp_do_ecn_ecne_work(sctp_association_t *asoc,
-					   __u32 lowest_tsn,
-					   sctp_chunk_t *);
-static void sctp_do_ecn_cwr_work(sctp_association_t *,__u32 lowest_tsn);
-static void sctp_do_8_2_transport_strike(sctp_association_t *,
-					 struct sctp_transport *);
-static void sctp_cmd_init_failed(sctp_cmd_seq_t *, sctp_association_t *);
-static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *, sctp_association_t *,
-				  sctp_event_t, sctp_subtype_t,
-				  sctp_chunk_t *chunk);
-static int sctp_cmd_process_init(sctp_cmd_seq_t *, sctp_association_t *,
-				 sctp_chunk_t *chunk,
-				 sctp_init_chunk_t *peer_init,
-				 int priority);
-static void sctp_cmd_hb_timers_start(sctp_cmd_seq_t *, sctp_association_t *);
-static void sctp_cmd_hb_timers_stop(sctp_cmd_seq_t *, sctp_association_t *);
-static void sctp_cmd_hb_timer_update(sctp_cmd_seq_t *, sctp_association_t *,
-				     struct sctp_transport *);
-static void sctp_cmd_transport_reset(sctp_cmd_seq_t *, sctp_association_t *,
-				     struct sctp_transport *);
-static void sctp_cmd_transport_on(sctp_cmd_seq_t *, sctp_association_t *,
-				  struct sctp_transport *, sctp_chunk_t *);
-static int sctp_cmd_process_sack(sctp_cmd_seq_t *, sctp_association_t *,
-				 sctp_sackhdr_t *);
-static void sctp_cmd_setup_t2(sctp_cmd_seq_t *, sctp_association_t *,
-			      sctp_chunk_t *);
-static void sctp_cmd_new_state(sctp_cmd_seq_t *, sctp_association_t *,
-			       sctp_state_t);
-
-/* These three macros allow us to pull the debugging code out of the
- * main flow of sctp_do_sm() to keep attention focused on the real
- * functionality there.
- */
-#define DEBUG_PRE \
-	SCTP_DEBUG_PRINTK("sctp_do_sm prefn: " \
-			  "ep %p, %s, %s, asoc %p[%s], %s\n", \
-			  ep, sctp_evttype_tbl[event_type], \
-			  (*debug_fn)(subtype), asoc, \
-			  sctp_state_tbl[state], state_fn->name)
+/********************************************************************
+ * Helper functions
+ ********************************************************************/
 
-#define DEBUG_POST \
-	SCTP_DEBUG_PRINTK("sctp_do_sm postfn: " \
-			  "asoc %p, status: %s\n", \
-			  asoc, sctp_status_tbl[status])
+/* A helper function for delayed processing of INET ECN CE bit. */
+static void sctp_do_ecn_ce_work(sctp_association_t *asoc, __u32 lowest_tsn)
+{
+	/* Save the TSN away for comparison when we receive CWR */
 
-#define DEBUG_POST_SFX \
-	SCTP_DEBUG_PRINTK("sctp_do_sm post sfx: error %d, asoc %p[%s]\n", \
-			  error, asoc, \
-			  sctp_state_tbl[(asoc && sctp_id2assoc(ep->base.sk, \
-			  sctp_assoc2id(asoc)))?asoc->state:SCTP_STATE_CLOSED])
+	asoc->last_ecne_tsn = lowest_tsn;
+	asoc->need_ecne = 1;
+}
 
-/*
- * This is the master state machine processing function.
+/* Helper function for delayed processing of SCTP ECNE chunk.  */
+/* RFC 2960 Appendix A
  *
- * If you want to understand all of lksctp, this is a
- * good place to start.
+ * RFC 2481 details a specific bit for a sender to send in
+ * the header of its next outbound TCP segment to indicate to
+ * its peer that it has reduced its congestion window.  This
+ * is termed the CWR bit.  For SCTP the same indication is made
+ * by including the CWR chunk.  This chunk contains one data
+ * element, i.e. the TSN number that was sent in the ECNE chunk.
+ * This element represents the lowest TSN number in the datagram
+ * that was originally marked with the CE bit.
  */
-int sctp_do_sm(sctp_event_t event_type, sctp_subtype_t subtype,
-	       sctp_state_t state,
-	       sctp_endpoint_t *ep,
-	       sctp_association_t *asoc,
-	       void *event_arg,
-	       int priority)
+static sctp_chunk_t *sctp_do_ecn_ecne_work(sctp_association_t *asoc,
+					   __u32 lowest_tsn,
+					   sctp_chunk_t *chunk)
 {
-	sctp_cmd_seq_t commands;
-	sctp_sm_table_entry_t *state_fn;
-	sctp_disposition_t status;
-	int error = 0;
-	typedef const char *(printfn_t)(sctp_subtype_t);
+	sctp_chunk_t *repl;
 
-	static printfn_t *table[] = {
-		NULL, sctp_cname, sctp_tname, sctp_oname, sctp_pname,
-	};
-	printfn_t *debug_fn  __attribute__ ((unused)) = table[event_type];
+	/* Our previously transmitted packet ran into some congestion
+	 * so we should take action by reducing cwnd and ssthresh
+	 * and then ACK our peer that we we've done so by
+	 * sending a CWR.
+	 */
 
-	/* Look up the state function, run it, and then process the
-	 * side effects.  These three steps are the heart of lksctp.
+	/* First, try to determine if we want to actually lower
+	 * our cwnd variables.  Only lower them if the ECNE looks more
+	 * recent than the last response.
 	 */
-	state_fn = sctp_sm_lookup_event(event_type, state, subtype);
+	if (TSN_lt(asoc->last_cwr_tsn, lowest_tsn)) {
+		struct sctp_transport *transport;
 
-	sctp_init_cmd_seq(&commands);
+		/* Find which transport's congestion variables
+		 * need to be adjusted.
+		 */
+		transport = sctp_assoc_lookup_tsn(asoc, lowest_tsn);
 
-	DEBUG_PRE;
-	status = (*state_fn->fn)(ep, asoc, subtype, event_arg, &commands);
-	DEBUG_POST;
+		/* Update the congestion variables. */
+		if (transport)
+			sctp_transport_lower_cwnd(transport,
+						  SCTP_LOWER_CWND_ECNE);
+		asoc->last_cwr_tsn = lowest_tsn;
+	}
 
-	error = sctp_side_effects(event_type, subtype, state,
-				  ep, asoc, event_arg,
-				  status, &commands,
-				  priority);
-	DEBUG_POST_SFX;
+	/* Always try to quiet the other end.  In case of lost CWR,
+	 * resend last_cwr_tsn.
+	 */
+	repl = sctp_make_cwr(asoc, asoc->last_cwr_tsn, chunk);
 
-	return error;
+	/* If we run out of memory, it will look like a lost CWR.  We'll
+	 * get back in sync eventually.
+	 */
+	return repl;
 }
 
-#undef DEBUG_PRE
-#undef DEBUG_POST
-
-/*****************************************************************
- * This the master state function side effect processing function.
- *****************************************************************/
-int sctp_side_effects(sctp_event_t event_type, sctp_subtype_t subtype,
-		      sctp_state_t state,
-		      sctp_endpoint_t *ep,
-		      sctp_association_t *asoc,
-		      void *event_arg,
-		      sctp_disposition_t status,
-		      sctp_cmd_seq_t *commands,
-		      int priority)
+/* Helper function to do delayed processing of ECN CWR chunk.  */
+static void sctp_do_ecn_cwr_work(sctp_association_t *asoc,
+				 __u32 lowest_tsn)
 {
-	int error;
-
-	/* FIXME - Most of the dispositions left today would be categorized
-	 * as "exceptional" dispositions.  For those dispositions, it
-	 * may not be proper to run through any of the commands at all.
-	 * For example, the command interpreter might be run only with
-	 * disposition SCTP_DISPOSITION_CONSUME.
+	/* Turn off ECNE getting auto-prepended to every outgoing
+	 * packet
 	 */
-	if (0 != (error = sctp_cmd_interpreter(event_type, subtype, state,
-					       ep, asoc,
-					       event_arg, status,
-					       commands, priority)))
-		goto bail;
-
-	switch (status) {
-	case SCTP_DISPOSITION_DISCARD:
-		SCTP_DEBUG_PRINTK("Ignored sctp protocol event - state %d, "
-				  "event_type %d, event_id %d\n",
-				  state, event_type, subtype.chunk);
-		break;
+	asoc->need_ecne = 0;
+}
 
-	case SCTP_DISPOSITION_NOMEM:
-		/* We ran out of memory, so we need to discard this
-		 * packet.
-		 */
-		/* BUG--we should now recover some memory, probably by
-		 * reneging...
-		 */
-		error = -ENOMEM;
-		break;
+/* Generate SACK if necessary.  We call this at the end of a packet.  */
+int sctp_gen_sack(struct sctp_association *asoc, int force, 
+		  sctp_cmd_seq_t *commands)
+{
+	__u32 ctsn, max_tsn_seen;
+	struct sctp_chunk *sack;
+	int error = 0;
 
-        case SCTP_DISPOSITION_DELETE_TCB:
-		/* This should now be a command. */
-		break;
+	if (force)
+		asoc->peer.sack_needed = 1;
 
-	case SCTP_DISPOSITION_CONSUME:
-	case SCTP_DISPOSITION_ABORT:
-		/*
-		 * We should no longer have much work to do here as the
-		 * real work has been done as explicit commands above.
-		 */
-		break;
+	ctsn = sctp_tsnmap_get_ctsn(&asoc->peer.tsn_map);
+	max_tsn_seen = sctp_tsnmap_get_max_tsn_seen(&asoc->peer.tsn_map);
 
-	case SCTP_DISPOSITION_VIOLATION:
-		printk(KERN_ERR "sctp protocol violation state %d "
-		       "chunkid %d\n", state, subtype.chunk);
-		break;
+	/* From 12.2 Parameters necessary per association (i.e. the TCB):
+	 *
+	 * Ack State : This flag indicates if the next received packet
+	 * 	     : is to be responded to with a SACK. ...
+	 *	     : When DATA chunks are out of order, SACK's
+	 *           : are not delayed (see Section 6).
+	 *
+	 * [This is actually not mentioned in Section 6, but we
+	 * implement it here anyway. --piggy]
+	 */
+        if (max_tsn_seen != ctsn)
+		asoc->peer.sack_needed = 1;
 
-	case SCTP_DISPOSITION_NOT_IMPL:
-		printk(KERN_WARNING "sctp unimplemented feature in state %d, "
-		       "event_type %d, event_id %d\n",
-		       state, event_type, subtype.chunk);
-		break;
+	/* From 6.2  Acknowledgement on Reception of DATA Chunks:
+	 *
+	 * Section 4.2 of [RFC2581] SHOULD be followed. Specifically,
+	 * an acknowledgement SHOULD be generated for at least every
+	 * second packet (not every second DATA chunk) received, and
+	 * SHOULD be generated within 200 ms of the arrival of any
+	 * unacknowledged DATA chunk. ...
+	 */
+	if (!asoc->peer.sack_needed) {
+		/* We will need a SACK for the next packet.  */
+		asoc->peer.sack_needed = 1;
+		goto out;
+	} else {
+		if (asoc->a_rwnd > asoc->rwnd)
+			asoc->a_rwnd = asoc->rwnd;
+		sack = sctp_make_sack(asoc);
+		if (!sack)
+			goto nomem;
 
-	case SCTP_DISPOSITION_BUG:
-		printk(KERN_ERR "sctp bug in state %d, "
-		       "event_type %d, event_id %d\n",
-		       state, event_type, subtype.chunk);
-		BUG();
-		break;
+		asoc->peer.sack_needed = 0;
 
-	default:
-		printk(KERN_ERR "sctp impossible disposition %d "
-		       "in state %d, event_type %d, event_id %d\n",
-		       status, state, event_type, subtype.chunk);
-		BUG();
-		break;
-	};
+		error = sctp_outq_tail(&asoc->outqueue, sack);
 
-bail:
+		/* Stop the SACK timer.  */
+		sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_STOP,
+				SCTP_TO(SCTP_EVENT_TIMEOUT_SACK));
+	}
+out:
+	return error;
+nomem:
+	error = -ENOMEM;
 	return error;
 }
 
-/********************************************************************
- * 2nd Level Abstractions
- ********************************************************************/
-
-/* This is the side-effect interpreter.  */
-int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
-			 sctp_state_t state, sctp_endpoint_t *ep,
-			 sctp_association_t *asoc, void *event_arg,
-			 sctp_disposition_t status, sctp_cmd_seq_t *commands,
-			 int priority)
+/* When the T3-RTX timer expires, it calls this function to create the
+ * relevant state machine event.
+ */
+void sctp_generate_t3_rtx_event(unsigned long peer)
 {
-	int error = 0;
-	int force;
-	sctp_cmd_t *cmd;
-	sctp_chunk_t *new_obj;
-	sctp_chunk_t *chunk = NULL;
-	struct sctp_packet *packet;
-	struct list_head *pos;
-	struct timer_list *timer;
-	unsigned long timeout;
-	struct sctp_transport *t;
-	sctp_sackhdr_t sackh;
+	int error;
+	struct sctp_transport *transport = (struct sctp_transport *) peer;
+	sctp_association_t *asoc = transport->asoc;
 
-	if(SCTP_EVENT_T_TIMEOUT != event_type)
-		chunk = (sctp_chunk_t *) event_arg;
+	/* Check whether a task is in the sock.  */
 
-	/* Note:  This whole file is a huge candidate for rework.
-	 * For example, each command could either have its own handler, so
-	 * the loop would look like:
-	 *     while (cmds)
-	 *         cmd->handle(x, y, z)
-	 * --jgrimm
-	 */
-	while (NULL != (cmd = sctp_next_cmd(commands))) {
-		switch (cmd->verb) {
-		case SCTP_CMD_NOP:
-			/* Do nothing. */
-			break;
+	sctp_bh_lock_sock(asoc->base.sk);
+	if (sock_owned_by_user(asoc->base.sk)) {
+		SCTP_DEBUG_PRINTK("%s:Sock is busy.\n", __FUNCTION__);
 
-		case SCTP_CMD_NEW_ASOC:
-			/* Register a new association.  */
-			asoc = cmd->obj.ptr;
-			/* Register with the endpoint.  */
-			sctp_endpoint_add_asoc(ep, asoc);
-			sctp_hash_established(asoc);
-			break;
+		/* Try again later.  */
+		if (!mod_timer(&transport->T3_rtx_timer, jiffies + (HZ/20)))
+			sctp_transport_hold(transport);
+		goto out_unlock;
+	}
 
-		case SCTP_CMD_UPDATE_ASSOC:
-		       sctp_assoc_update(asoc, cmd->obj.ptr);
-		       break;
+	/* Is this transport really dead and just waiting around for
+	 * the timer to let go of the reference?
+	 */
+	if (transport->dead)
+		goto out_unlock;
 
-		case SCTP_CMD_PURGE_OUTQUEUE:
-		       sctp_outq_teardown(&asoc->outqueue);
-		       break;
+	/* Run through the state machine.  */
+	error = sctp_do_sm(SCTP_EVENT_T_TIMEOUT,
+			   SCTP_ST_TIMEOUT(SCTP_EVENT_TIMEOUT_T3_RTX),
+			   asoc->state,
+			   asoc->ep, asoc,
+			   transport, GFP_ATOMIC);
 
-		case SCTP_CMD_DELETE_TCB:
-			/* Delete the current association.  */
-			sctp_unhash_established(asoc);
-			sctp_association_free(asoc);
-			asoc = NULL;
-			break;
+	if (error)
+		asoc->base.sk->err = -error;
 
-		case SCTP_CMD_NEW_STATE:
-			/* Enter a new state.  */
-			sctp_cmd_new_state(commands, asoc, cmd->obj.state);
-			break;
+out_unlock:
+	sctp_bh_unlock_sock(asoc->base.sk);
+	sctp_transport_put(transport);
+}
 
-		case SCTP_CMD_REPORT_TSN:
-			/* Record the arrival of a TSN.  */
-			sctp_tsnmap_mark(&asoc->peer.tsn_map, cmd->obj.u32);
-			break;
+/* This is a sa interface for producing timeout events.  It works
+ * for timeouts which use the association as their parameter.
+ */
+static void sctp_generate_timeout_event(sctp_association_t *asoc,
+					sctp_event_timeout_t timeout_type)
+{
+	int error = 0;
 
-		case SCTP_CMD_GEN_SACK:
-			/* Generate a Selective ACK.
-			 * The argument tells us whether to just count
-			 * the packet and MAYBE generate a SACK, or
-			 * force a SACK out.
-			 */
-			force = cmd->obj.i32;
-			error = sctp_gen_sack(asoc, force, commands);
-			break;
+	sctp_bh_lock_sock(asoc->base.sk);
+	if (sock_owned_by_user(asoc->base.sk)) {
+		SCTP_DEBUG_PRINTK("%s:Sock is busy: timer %d\n",
+				  __FUNCTION__,
+				  timeout_type);
 
-		case SCTP_CMD_PROCESS_SACK:
-			/* Process an inbound SACK.  */
-			error = sctp_cmd_process_sack(commands, asoc,
-						      cmd->obj.ptr);
-			break;
+		/* Try again later.  */
+		if (!mod_timer(&asoc->timers[timeout_type], jiffies + (HZ/20)))
+			sctp_association_hold(asoc);
+		goto out_unlock;
+	}
 
-		case SCTP_CMD_GEN_INIT_ACK:
-			/* Generate an INIT ACK chunk.  */
-			new_obj = sctp_make_init_ack(asoc, chunk, GFP_ATOMIC,
-						     0);
-			if (!new_obj)
-				goto nomem;
+	/* Is this association really dead and just waiting around for
+	 * the timer to let go of the reference?
+	 */
+	if (asoc->base.dead)
+		goto out_unlock;
 
-			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
-					SCTP_CHUNK(new_obj));
-			break;
+	/* Run through the state machine.  */
+	error = sctp_do_sm(SCTP_EVENT_T_TIMEOUT,
+			   SCTP_ST_TIMEOUT(timeout_type),
+			   asoc->state, asoc->ep, asoc,
+			   (void *)timeout_type,
+			   GFP_ATOMIC);
 
-		case SCTP_CMD_PEER_INIT:
-			/* Process a unified INIT from the peer.
-			 * Note: Only used during INIT-ACK processing.  If
-			 * there is an error just return to the outter
-			 * layer which will bail.
-			 */
-			error = sctp_cmd_process_init(commands, asoc, chunk,
-						      cmd->obj.ptr, priority);
-			break;
+	if (error)
+		asoc->base.sk->err = -error;
 
-		case SCTP_CMD_GEN_COOKIE_ECHO:
-			/* Generate a COOKIE ECHO chunk.  */
-			new_obj = sctp_make_cookie_echo(asoc, chunk);
-			if (!new_obj) {
-				if (cmd->obj.ptr)
-					sctp_free_chunk(cmd->obj.ptr);
-				goto nomem;
-			}
-			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
-					SCTP_CHUNK(new_obj));
+out_unlock:
+	sctp_bh_unlock_sock(asoc->base.sk);
+	sctp_association_put(asoc);
+}
 
-			/* If there is an ERROR chunk to be sent along with
-			 * the COOKIE_ECHO, send it, too.
-			 */
-			if (cmd->obj.ptr)
-				sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
-						SCTP_CHUNK(cmd->obj.ptr));
-			break;
+void sctp_generate_t1_cookie_event(unsigned long data)
+{
+	sctp_association_t *asoc = (sctp_association_t *) data;
+	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_T1_COOKIE);
+}
 
-		case SCTP_CMD_GEN_SHUTDOWN:
-			/* Generate SHUTDOWN when in SHUTDOWN_SENT state.
-			 * Reset error counts.
-			 */
-			asoc->overall_error_count = 0;
+void sctp_generate_t1_init_event(unsigned long data)
+{
+	sctp_association_t *asoc = (sctp_association_t *) data;
+	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_T1_INIT);
+}
 
-			/* Generate a SHUTDOWN chunk.  */
-			new_obj = sctp_make_shutdown(asoc);
-			if (!new_obj)
-				goto nomem;
-			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
-					SCTP_CHUNK(new_obj));
-			break;
+void sctp_generate_t2_shutdown_event(unsigned long data)
+{
+	sctp_association_t *asoc = (sctp_association_t *) data;
+	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_T2_SHUTDOWN);
+}
 
-		case SCTP_CMD_CHUNK_ULP:
-			/* Send a chunk to the sockets layer.  */
-			SCTP_DEBUG_PRINTK("sm_sideff: %s %p, %s %p.\n",
-					  "chunk_up:", cmd->obj.ptr,
-					  "ulpq:", &asoc->ulpq);
-			sctp_ulpq_tail_data(&asoc->ulpq, cmd->obj.ptr,
-					    GFP_ATOMIC);
-			break;
+void sctp_generate_t5_shutdown_guard_event(unsigned long data)
+{
+        sctp_association_t *asoc = (sctp_association_t *)data;
+        sctp_generate_timeout_event(asoc,
+				    SCTP_EVENT_TIMEOUT_T5_SHUTDOWN_GUARD);
 
-		case SCTP_CMD_EVENT_ULP:
-			/* Send a notification to the sockets layer.  */
-			SCTP_DEBUG_PRINTK("sm_sideff: %s %p, %s %p.\n",
-					  "event_up:",cmd->obj.ptr,
-					  "ulpq:",&asoc->ulpq);
-			sctp_ulpq_tail_event(&asoc->ulpq, cmd->obj.ptr);
-			break;
+} /* sctp_generate_t5_shutdown_guard_event() */
 
-		case SCTP_CMD_REPLY:
-			/* Send a chunk to our peer.  */
-			error = sctp_outq_tail(&asoc->outqueue,
-					       cmd->obj.ptr);
-			break;
+void sctp_generate_autoclose_event(unsigned long data)
+{
+	sctp_association_t *asoc = (sctp_association_t *) data;
+	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_AUTOCLOSE);
+}
 
-		case SCTP_CMD_SEND_PKT:
-			/* Send a full packet to our peer.  */
-			packet = cmd->obj.ptr;
-			sctp_packet_transmit(packet);
-			sctp_ootb_pkt_free(packet);
-			break;
+/* Generate a heart beat event.  If the sock is busy, reschedule.   Make
+ * sure that the transport is still valid.
+ */
+void sctp_generate_heartbeat_event(unsigned long data)
+{
+	int error = 0;
+	struct sctp_transport *transport = (struct sctp_transport *) data;
+	sctp_association_t *asoc = transport->asoc;
 
-		case SCTP_CMD_RETRAN:
-			/* Mark a transport for retransmission.  */
-			sctp_retransmit(&asoc->outqueue, cmd->obj.transport,
-					SCTP_RETRANSMIT_T3_RTX);
-			break;
+	sctp_bh_lock_sock(asoc->base.sk);
+	if (sock_owned_by_user(asoc->base.sk)) {
+		SCTP_DEBUG_PRINTK("%s:Sock is busy.\n", __FUNCTION__);
 
-		case SCTP_CMD_TRANSMIT:
-			/* Kick start transmission. */
-			error = sctp_outq_flush(&asoc->outqueue, 0);
-			break;
+		/* Try again later.  */
+		if (!mod_timer(&transport->hb_timer, jiffies + (HZ/20)))
+			sctp_transport_hold(transport);
+		goto out_unlock;
+	}
 
-		case SCTP_CMD_ECN_CE:
-			/* Do delayed CE processing.   */
-			sctp_do_ecn_ce_work(asoc, cmd->obj.u32);
-			break;
+	/* Is this structure just waiting around for us to actually
+	 * get destroyed?
+	 */
+	if (transport->dead)
+		goto out_unlock;
 
-		case SCTP_CMD_ECN_ECNE:
-			/* Do delayed ECNE processing. */
-			new_obj = sctp_do_ecn_ecne_work(asoc, cmd->obj.u32,
-							chunk);
-			if (new_obj)
-				sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
-						SCTP_CHUNK(new_obj));
-			break;
+	error = sctp_do_sm(SCTP_EVENT_T_TIMEOUT,
+			   SCTP_ST_TIMEOUT(SCTP_EVENT_TIMEOUT_HEARTBEAT),
+			   asoc->state,
+			   asoc->ep, asoc,
+			   transport, GFP_ATOMIC);
 
-		case SCTP_CMD_ECN_CWR:
-			/* Do delayed CWR processing.  */
-			sctp_do_ecn_cwr_work(asoc, cmd->obj.u32);
-			break;
+         if (error)
+		 asoc->base.sk->err = -error;
 
-		case SCTP_CMD_SETUP_T2:
-			sctp_cmd_setup_t2(commands, asoc, cmd->obj.ptr);
-			break;
+out_unlock:
+	sctp_bh_unlock_sock(asoc->base.sk);
+	sctp_transport_put(transport);
+}
 
-		case SCTP_CMD_TIMER_START:
-			timer = &asoc->timers[cmd->obj.to];
-			timeout = asoc->timeouts[cmd->obj.to];
-			if (!timeout)
-				BUG();
+/* Inject a SACK Timeout event into the state machine.  */
+void sctp_generate_sack_event(unsigned long data)
+{
+	sctp_association_t *asoc = (sctp_association_t *) data;
+	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_SACK);
+}
 
-			timer->expires = jiffies + timeout;
-			sctp_association_hold(asoc);
-			add_timer(timer);
-			break;
+sctp_timer_event_t *sctp_timer_events[SCTP_NUM_TIMEOUT_TYPES] = {
+	NULL,
+	sctp_generate_t1_cookie_event,
+	sctp_generate_t1_init_event,
+	sctp_generate_t2_shutdown_event,
+	NULL,
+	sctp_generate_t5_shutdown_guard_event,
+	sctp_generate_heartbeat_event,
+	sctp_generate_sack_event,
+	sctp_generate_autoclose_event,
+};
 
-		case SCTP_CMD_TIMER_RESTART:
-			timer = &asoc->timers[cmd->obj.to];
-			timeout = asoc->timeouts[cmd->obj.to];
-			if (!mod_timer(timer, jiffies + timeout))
-				sctp_association_hold(asoc);
-			break;
 
-		case SCTP_CMD_TIMER_STOP:
-			timer = &asoc->timers[cmd->obj.to];
-			if (timer_pending(timer) && del_timer(timer))
-				sctp_association_put(asoc);
-			break;
+/* RFC 2960 8.2 Path Failure Detection
+ *
+ * When its peer endpoint is multi-homed, an endpoint should keep a
+ * error counter for each of the destination transport addresses of the
+ * peer endpoint.
+ *
+ * Each time the T3-rtx timer expires on any address, or when a
+ * HEARTBEAT sent to an idle address is not acknowledged within a RTO,
+ * the error counter of that destination address will be incremented.
+ * When the value in the error counter exceeds the protocol parameter
+ * 'Path.Max.Retrans' of that destination address, the endpoint should
+ * mark the destination transport address as inactive, and a
+ * notification SHOULD be sent to the upper layer.
+ *
+ */
+static void sctp_do_8_2_transport_strike(sctp_association_t *asoc,
+					 struct sctp_transport *transport)
+{
+	/* The check for association's overall error counter exceeding the
+	 * threshold is done in the state function.
+	 */
+	asoc->overall_error_count++;
 
-		case SCTP_CMD_INIT_RESTART:
-			/* Do the needed accounting and updates
-			 * associated with restarting an initialization
-			 * timer.
-			 */
-			asoc->counters[SCTP_COUNTER_INIT_ERROR]++;
-			asoc->timeouts[cmd->obj.to] *= 2;
-			if (asoc->timeouts[cmd->obj.to] >
-			    asoc->max_init_timeo) {
-				asoc->timeouts[cmd->obj.to] =
-					asoc->max_init_timeo;
-			}
+	if (transport->active &&
+	    (transport->error_count++ >= transport->error_threshold)) {
+		SCTP_DEBUG_PRINTK("transport_strike: transport "
+				  "IP:%d.%d.%d.%d failed.\n",
+				  NIPQUAD(transport->ipaddr.v4.sin_addr));
+		sctp_assoc_control_transport(asoc, transport,
+					     SCTP_TRANSPORT_DOWN,
+					     SCTP_FAILED_THRESHOLD);
+	}
 
-			/* 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, struct sctp_transport,
-					       transports);
-				sctp_retransmit_mark(&asoc->outqueue, t, 0);
-			}
+	/* E2) For the destination address for which the timer
+	 * expires, set RTO <- RTO * 2 ("back off the timer").  The
+	 * maximum value discussed in rule C7 above (RTO.max) may be
+	 * used to provide an upper bound to this doubling operation.
+	 */
+	transport->rto = min((transport->rto * 2), transport->asoc->rto_max);
+}
 
-			sctp_add_cmd_sf(commands,
-					SCTP_CMD_TIMER_RESTART,
-					SCTP_TO(cmd->obj.to));
-			break;
+/* Worker routine to handle INIT command failure.  */
+static void sctp_cmd_init_failed(sctp_cmd_seq_t *commands,
+				 sctp_association_t *asoc)
+{
+	struct sctp_ulpevent *event;
 
-		case SCTP_CMD_INIT_FAILED:
-			sctp_cmd_init_failed(commands, asoc);
-			break;
+	event = sctp_ulpevent_make_assoc_change(asoc,0, SCTP_CANT_STR_ASSOC,
+						0, 0, 0, GFP_ATOMIC);
 
-		case SCTP_CMD_ASSOC_FAILED:
-			sctp_cmd_assoc_failed(commands, asoc, event_type,
-					      subtype, chunk);
-			break;
+	if (event)
+		sctp_add_cmd_sf(commands, SCTP_CMD_EVENT_ULP,
+				SCTP_ULPEVENT(event));
 
-		case SCTP_CMD_COUNTER_INC:
-			asoc->counters[cmd->obj.counter]++;
-			break;
+	/* FIXME:  We need to handle data possibly either
+	 * sent via COOKIE-ECHO bundling or just waiting in
+	 * the transmit queue, if the user has enabled
+	 * SEND_FAILED notifications.
+	 */
+	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
+}
 
-		case SCTP_CMD_COUNTER_RESET:
-			asoc->counters[cmd->obj.counter] = 0;
-			break;
+/* Worker routine to handle SCTP_CMD_ASSOC_FAILED.  */
+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)
+{
+	struct sctp_ulpevent *event;
+	__u16 error = 0;
 
-		case SCTP_CMD_REPORT_DUP:
-			sctp_tsnmap_mark_dup(&asoc->peer.tsn_map, 
-					     cmd->obj.u32);
-			break;
+	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;
+	}
 
-		case SCTP_CMD_REPORT_BAD_TAG:
-			SCTP_DEBUG_PRINTK("vtag mismatch!\n");
-			break;
+	/* Cancel any partial delivery in progress. */
+	sctp_ulpq_abort_pd(&asoc->ulpq, GFP_ATOMIC);
 
-		case SCTP_CMD_STRIKE:
-			/* Mark one strike against a transport.  */
-			sctp_do_8_2_transport_strike(asoc, cmd->obj.transport);
-			break;
+	event = sctp_ulpevent_make_assoc_change(asoc, 0, SCTP_COMM_LOST,
+						error, 0, 0, GFP_ATOMIC);
+	if (event)
+		sctp_add_cmd_sf(commands, SCTP_CMD_EVENT_ULP,
+				SCTP_ULPEVENT(event));
 
-		case SCTP_CMD_TRANSPORT_RESET:
-			t = cmd->obj.transport;
-			sctp_cmd_transport_reset(commands, asoc, t);
-			break;
+	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
+			SCTP_STATE(SCTP_STATE_CLOSED));
 
-		case SCTP_CMD_TRANSPORT_ON:
-			t = cmd->obj.transport;
-			sctp_cmd_transport_on(commands, asoc, t, chunk);
-			break;
+	/* FIXME:  We need to handle data that could not be sent or was not
+	 * acked, if the user has enabled SEND_FAILED notifications.
+	 */
+	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
+}
 
-		case SCTP_CMD_HB_TIMERS_START:
-			sctp_cmd_hb_timers_start(commands, asoc);
-			break;
+/* Process an init chunk (may be real INIT/INIT-ACK or an embedded INIT
+ * inside the cookie.  In reality, this is only used for INIT-ACK processing
+ * since all other cases use "temporary" associations and can do all
+ * their work in statefuns directly.
+ */
+static int sctp_cmd_process_init(sctp_cmd_seq_t *commands,
+				 sctp_association_t *asoc,
+				 sctp_chunk_t *chunk,
+				 sctp_init_chunk_t *peer_init,
+				 int priority)
+{
+	int error;
 
-		case SCTP_CMD_HB_TIMER_UPDATE:
-			t = cmd->obj.transport;
-			sctp_cmd_hb_timer_update(commands, asoc, t);
-			break;
+	/* We only process the init as a sideeffect in a single
+	 * case.   This is when we process the INIT-ACK.   If we
+	 * fail during INIT processing (due to malloc problems),
+	 * just return the error and stop processing the stack.
+	 */
 
-		case SCTP_CMD_HB_TIMERS_STOP:
-			sctp_cmd_hb_timers_stop(commands, asoc);
-			break;
+	if (!sctp_process_init(asoc, chunk->chunk_hdr->type,
+			       sctp_source(chunk), peer_init,
+			       priority))
+		error = -ENOMEM;
+	else
+		error = 0;
 
-		case SCTP_CMD_REPORT_ERROR:
-			error = cmd->obj.error;
-			break;
+	return error;
+}
 
-		case SCTP_CMD_PROCESS_CTSN:
-			/* Dummy up a SACK for processing. */
-			sackh.cum_tsn_ack = cmd->obj.u32;
-			sackh.a_rwnd = 0;
-			sackh.num_gap_ack_blocks = 0;
-			sackh.num_dup_tsns = 0;
-			sctp_add_cmd_sf(commands, SCTP_CMD_PROCESS_SACK,
-					SCTP_SACKH(&sackh));
-			break;
+/* Helper function to break out starting up of heartbeat timers.  */
+static void sctp_cmd_hb_timers_start(sctp_cmd_seq_t *cmds,
+				     sctp_association_t *asoc)
+{
+	struct sctp_transport *t;
+	struct list_head *pos;
 
-		case SCTP_CMD_DISCARD_PACKET:
-			/* We need to discard the whole packet.  */
-			chunk->pdiscard = 1;
-			break;
+	/* Start a heartbeat timer for each transport on the association.
+	 * hold a reference on the transport to make sure none of
+	 * the needed data structures go away.
+	 */
+	list_for_each(pos, &asoc->peer.transport_addr_list) {
+		t = list_entry(pos, struct sctp_transport, transports);
 
-		case SCTP_CMD_RTO_PENDING:
-			t = cmd->obj.transport;
-			t->rto_pending = 1;
-			break;
+		if (!mod_timer(&t->hb_timer, sctp_transport_timeout(t)))
+			sctp_transport_hold(t);
+	}
+}
 
-		case SCTP_CMD_PART_DELIVER:
-			sctp_ulpq_partial_delivery(&asoc->ulpq, cmd->obj.ptr,
-						   GFP_ATOMIC);
-			break;
+static void sctp_cmd_hb_timers_stop(sctp_cmd_seq_t *cmds,
+				    sctp_association_t *asoc)
+{
+	struct sctp_transport *t;
+	struct list_head *pos;
 
-		case SCTP_CMD_RENEGE:
-			sctp_ulpq_renege(&asoc->ulpq, cmd->obj.ptr,
-					 GFP_ATOMIC);
-			break;
+	/* Stop all heartbeat timers. */
 
-		default:
-			printk(KERN_WARNING "Impossible command: %u, %p\n",
-			       cmd->verb, cmd->obj.ptr);
-			break;
-		};
-		if (error)
-			return error;
+	list_for_each(pos, &asoc->peer.transport_addr_list) {
+		t = list_entry(pos, struct sctp_transport, transports);
+		if (del_timer(&t->hb_timer))
+			sctp_transport_put(t);
 	}
-
-	return error;
-
-nomem:
-	error = -ENOMEM;
-	return error;
 }
 
-/* A helper function for delayed processing of INET ECN CE bit. */
-static void sctp_do_ecn_ce_work(sctp_association_t *asoc, __u32 lowest_tsn)
+/* Helper function to update the heartbeat timer. */
+static void sctp_cmd_hb_timer_update(sctp_cmd_seq_t *cmds,
+				     sctp_association_t *asoc,
+				     struct sctp_transport *t)
 {
-	/* Save the TSN away for comparison when we receive CWR */
-
-	asoc->last_ecne_tsn = lowest_tsn;
-	asoc->need_ecne = 1;
+	/* Update the heartbeat timer.  */
+	if (!mod_timer(&t->hb_timer, sctp_transport_timeout(t)))
+		sctp_transport_hold(t);
 }
 
-/* Helper function for delayed processing of SCTP ECNE chunk.  */
-/* RFC 2960 Appendix A
- *
- * RFC 2481 details a specific bit for a sender to send in
- * the header of its next outbound TCP segment to indicate to
- * its peer that it has reduced its congestion window.  This
- * is termed the CWR bit.  For SCTP the same indication is made
- * by including the CWR chunk.  This chunk contains one data
- * element, i.e. the TSN number that was sent in the ECNE chunk.
- * This element represents the lowest TSN number in the datagram
- * that was originally marked with the CE bit.
- */
-static sctp_chunk_t *sctp_do_ecn_ecne_work(sctp_association_t *asoc,
-					   __u32 lowest_tsn,
-					   sctp_chunk_t *chunk)
+/* Helper function to handle the reception of an HEARTBEAT ACK.  */
+static void sctp_cmd_transport_on(sctp_cmd_seq_t *cmds,
+				  sctp_association_t *asoc,
+				  struct sctp_transport *t,
+				  sctp_chunk_t *chunk)
 {
-	sctp_chunk_t *repl;
-
-	/* Our previously transmitted packet ran into some congestion
-	 * so we should take action by reducing cwnd and ssthresh
-	 * and then ACK our peer that we we've done so by
-	 * sending a CWR.
-	 */
+	sctp_sender_hb_info_t *hbinfo;
 
-	/* First, try to determine if we want to actually lower
-	 * our cwnd variables.  Only lower them if the ECNE looks more
-	 * recent than the last response.
+	/* 8.3 Upon the receipt of the HEARTBEAT ACK, the sender of the
+	 * HEARTBEAT should clear the error counter of the destination
+	 * transport address to which the HEARTBEAT was sent.
+	 * The association's overall error count is also cleared.
 	 */
-	if (TSN_lt(asoc->last_cwr_tsn, lowest_tsn)) {
-		struct sctp_transport *transport;
-
-		/* Find which transport's congestion variables
-		 * need to be adjusted.
-		 */
-		transport = sctp_assoc_lookup_tsn(asoc, lowest_tsn);
-
-		/* Update the congestion variables. */
-		if (transport)
-			sctp_transport_lower_cwnd(transport,
-						  SCTP_LOWER_CWND_ECNE);
-		asoc->last_cwr_tsn = lowest_tsn;
-	}
+	t->error_count = 0;
+	t->asoc->overall_error_count = 0;
 
-	/* Always try to quiet the other end.  In case of lost CWR,
-	 * resend last_cwr_tsn.
+	/* Mark the destination transport address as active if it is not so
+	 * marked.
 	 */
-	repl = sctp_make_cwr(asoc, asoc->last_cwr_tsn, chunk);
+	if (!t->active)
+		sctp_assoc_control_transport(asoc, t, SCTP_TRANSPORT_UP,
+					     SCTP_HEARTBEAT_SUCCESS);
 
-	/* If we run out of memory, it will look like a lost CWR.  We'll
-	 * get back in sync eventually.
+	/* The receiver of the HEARTBEAT ACK should also perform an
+	 * RTT measurement for that destination transport address
+	 * using the time value carried in the HEARTBEAT ACK chunk.
 	 */
-	return repl;
+	hbinfo = (sctp_sender_hb_info_t *) chunk->skb->data;
+	sctp_transport_update_rto(t, (jiffies - hbinfo->sent_at));
 }
 
-/* Helper function to do delayed processing of ECN CWR chunk.  */
-static void sctp_do_ecn_cwr_work(sctp_association_t *asoc,
-				 __u32 lowest_tsn)
+/* Helper function to do a transport reset at the expiry of the hearbeat
+ * timer.
+ */
+static void sctp_cmd_transport_reset(sctp_cmd_seq_t *cmds,
+				     sctp_association_t *asoc,
+				     struct sctp_transport *t)
 {
-	/* Turn off ECNE getting auto-prepended to every outgoing
-	 * packet
-	 */
-	asoc->need_ecne = 0;
-}
+	sctp_transport_lower_cwnd(t, SCTP_LOWER_CWND_INACTIVE);
 
-/* This macro is to compress the text a bit...  */
-#define AP(v) asoc->peer.v
+	/* Mark one strike against a transport.  */
+	sctp_do_8_2_transport_strike(asoc, t);
+}
 
-/* Generate SACK if necessary.  We call this at the end of a packet.  */
-int sctp_gen_sack(sctp_association_t *asoc, int force, sctp_cmd_seq_t *commands)
+/* Helper function to process the process SACK command.  */
+static int sctp_cmd_process_sack(sctp_cmd_seq_t *cmds,
+				 sctp_association_t *asoc,
+				 sctp_sackhdr_t *sackh)
 {
-	__u32 ctsn, max_tsn_seen;
-	sctp_chunk_t *sack;
-	int error = 0;
-
-	if (force)
-		asoc->peer.sack_needed = 1;
-
-	ctsn = sctp_tsnmap_get_ctsn(&asoc->peer.tsn_map);
-	max_tsn_seen = sctp_tsnmap_get_max_tsn_seen(&asoc->peer.tsn_map);
-
-	/* From 12.2 Parameters necessary per association (i.e. the TCB):
-	 *
-	 * Ack State : This flag indicates if the next received packet
-	 * 	     : is to be responded to with a SACK. ...
-	 *	     : When DATA chunks are out of order, SACK's
-	 *           : are not delayed (see Section 6).
-	 *
-	 * [This is actually not mentioned in Section 6, but we
-	 * implement it here anyway. --piggy]
-	 */
-        if (max_tsn_seen != ctsn)
-		asoc->peer.sack_needed = 1;
+	int err;
 
-	/* From 6.2  Acknowledgement on Reception of DATA Chunks:
-	 *
-	 * Section 4.2 of [RFC2581] SHOULD be followed. Specifically,
-	 * an acknowledgement SHOULD be generated for at least every
-	 * second packet (not every second DATA chunk) received, and
-	 * SHOULD be generated within 200 ms of the arrival of any
-	 * unacknowledged DATA chunk. ...
-	 */
-	if (!asoc->peer.sack_needed) {
-		/* We will need a SACK for the next packet.  */
-		asoc->peer.sack_needed = 1;
-		goto out;
+	if (sctp_outq_sack(&asoc->outqueue, sackh)) {
+		/* There are no more TSNs awaiting SACK.  */
+		err = sctp_do_sm(SCTP_EVENT_T_OTHER,
+				 SCTP_ST_OTHER(SCTP_EVENT_NO_PENDING_TSN),
+				 asoc->state, asoc->ep, asoc, NULL,
+				 GFP_ATOMIC);
 	} else {
-		if (asoc->a_rwnd > asoc->rwnd)
-			asoc->a_rwnd = asoc->rwnd;
-		sack = sctp_make_sack(asoc);
-		if (!sack)
-			goto nomem;
-
-		asoc->peer.sack_needed = 0;
-
-		error = sctp_outq_tail(&asoc->outqueue, sack);
-
-		/* Stop the SACK timer.  */
-		sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_STOP,
-				SCTP_TO(SCTP_EVENT_TIMEOUT_SACK));
+		/* Windows may have opened, so we need
+		 * to check if we have DATA to transmit
+		 */
+		err = sctp_outq_flush(&asoc->outqueue, 0);
 	}
 
-out:
-	return error;
-
-nomem:
-	error = -ENOMEM;
-	return error;
+	return err;
 }
 
-/* Handle a duplicate TSN.  */
-void sctp_do_TSNdup(sctp_association_t *asoc, sctp_chunk_t *chunk, long gap)
+/* Helper function to set the timeout value for T2-SHUTDOWN timer and to set
+ * the transport for a shutdown chunk.
+ */
+static void sctp_cmd_setup_t2(sctp_cmd_seq_t *cmds, sctp_association_t *asoc,
+			      sctp_chunk_t *chunk)
 {
-#if 0
-	sctp_chunk_t *sack;
+	struct sctp_transport *t;
 
-	/* Caution:  gap < 2 * SCTP_TSN_MAP_SIZE
-	 * 	so gap can be negative.
-	 *
-	 *		--xguo
-	 */
+	t = sctp_assoc_choose_shutdown_transport(asoc);
+	asoc->shutdown_last_sent_to = t;
+	asoc->timeouts[SCTP_EVENT_TIMEOUT_T2_SHUTDOWN] = t->rto;
+	chunk->transport = t;
+}
 
-	/* Count this TSN.  */
-	if (gap < SCTP_TSN_MAP_SIZE) {
-		asoc->peer.tsn_map[gap]++;
-	} else {
-		asoc->peer.tsn_map_overflow[gap - SCTP_TSN_MAP_SIZE]++;
-	}
+/* Helper function to change the state of an association. */
+static void sctp_cmd_new_state(sctp_cmd_seq_t *cmds, sctp_association_t *asoc,
+			       sctp_state_t state)
+{
 
-	/* From 6.2  Acknowledgement on Reception of DATA Chunks
-	 *
-	 * When a packet arrives with duplicate DATA chunk(s)
-	 * and with no new DATA chunk(s), the endpoint MUST
-	 * immediately send a SACK with no delay. If a packet
-	 * arrives with duplicate DATA chunk(s) bundled with
-	 * new DATA chunks, the endpoint MAY immediately send a
-	 * SACK.  Normally receipt of duplicate DATA chunks
-	 * will occur when the original SACK chunk was lost and
-	 * the peer's RTO has expired. The duplicate TSN
-	 * number(s) SHOULD be reported in the SACK as
-	 * duplicate.
-	 */
-	asoc->counters[SctpCounterAckState] = 2;
-#endif /* 0 */
-} /* sctp_do_TSNdup() */
+	struct sock *sk = asoc->base.sk;
+	struct sctp_opt *sp = sctp_sk(sk);
 
-#undef AP
+	asoc->state = state;
+	asoc->state_timestamp = jiffies;
 
-/* When the T3-RTX timer expires, it calls this function to create the
- * relevant state machine event.
- */
-void sctp_generate_t3_rtx_event(unsigned long peer)
-{
-	int error;
-	struct sctp_transport *transport = (struct sctp_transport *) peer;
-	sctp_association_t *asoc = transport->asoc;
-
-	/* Check whether a task is in the sock.  */
-
-	sctp_bh_lock_sock(asoc->base.sk);
-	if (sock_owned_by_user(asoc->base.sk)) {
-		SCTP_DEBUG_PRINTK("%s:Sock is busy.\n", __FUNCTION__);
+	if ((SCTP_STATE_ESTABLISHED == asoc->state) ||
+	    (SCTP_STATE_CLOSED == asoc->state)) { 
+		/* Wake up any processes waiting in the asoc's wait queue in
+		 * sctp_wait_for_connect() or sctp_wait_for_sndbuf(). 
+	 	 */
+		if (waitqueue_active(&asoc->wait))
+			wake_up_interruptible(&asoc->wait);
 
-		/* Try again later.  */
-		if (!mod_timer(&transport->T3_rtx_timer, jiffies + (HZ/20)))
-			sctp_transport_hold(transport);
-		goto out_unlock;
+		/* Wake up any processes waiting in the sk's sleep queue of
+		 * a TCP-style or UDP-style peeled-off socket in
+		 * sctp_wait_for_accept() or sctp_wait_for_packet().
+		 * For a UDP-style socket, the waiters are woken up by the
+		 * notifications.
+		 */
+		if (SCTP_SOCKET_UDP != sp->type)
+			sk->state_change(sk);
 	}
 
-	/* Is this transport really dead and just waiting around for
-	 * the timer to let go of the reference?
+	/* Change the sk->state of a TCP-style socket that has sucessfully
+	 * completed a connect() call.
 	 */
-	if (transport->dead)
-		goto out_unlock;
+	if ((SCTP_STATE_ESTABLISHED == asoc->state) &&
+	    (SCTP_SOCKET_TCP == sp->type) && (SCTP_SS_CLOSED == sk->state))
+	    sk->state = SCTP_SS_ESTABLISHED;
+}
 
-	/* Run through the state machine.  */
-	error = sctp_do_sm(SCTP_EVENT_T_TIMEOUT,
-			   SCTP_ST_TIMEOUT(SCTP_EVENT_TIMEOUT_T3_RTX),
-			   asoc->state,
-			   asoc->ep, asoc,
-			   transport, GFP_ATOMIC);
+/* These three macros allow us to pull the debugging code out of the
+ * main flow of sctp_do_sm() to keep attention focused on the real
+ * functionality there.
+ */
+#define DEBUG_PRE \
+	SCTP_DEBUG_PRINTK("sctp_do_sm prefn: " \
+			  "ep %p, %s, %s, asoc %p[%s], %s\n", \
+			  ep, sctp_evttype_tbl[event_type], \
+			  (*debug_fn)(subtype), asoc, \
+			  sctp_state_tbl[state], state_fn->name)
 
-	if (error)
-		asoc->base.sk->err = -error;
+#define DEBUG_POST \
+	SCTP_DEBUG_PRINTK("sctp_do_sm postfn: " \
+			  "asoc %p, status: %s\n", \
+			  asoc, sctp_status_tbl[status])
 
-out_unlock:
-	sctp_bh_unlock_sock(asoc->base.sk);
-	sctp_transport_put(transport);
-}
+#define DEBUG_POST_SFX \
+	SCTP_DEBUG_PRINTK("sctp_do_sm post sfx: error %d, asoc %p[%s]\n", \
+			  error, asoc, \
+			  sctp_state_tbl[(asoc && sctp_id2assoc(ep->base.sk, \
+			  sctp_assoc2id(asoc)))?asoc->state:SCTP_STATE_CLOSED])
 
-/* This is a sa interface for producing timeout events.  It works
- * for timeouts which use the association as their parameter.
+/*
+ * This is the master state machine processing function.
+ *
+ * If you want to understand all of lksctp, this is a
+ * good place to start.
  */
-static void sctp_generate_timeout_event(sctp_association_t *asoc,
-					sctp_event_timeout_t timeout_type)
+int sctp_do_sm(sctp_event_t event_type, sctp_subtype_t subtype,
+	       sctp_state_t state,
+	       sctp_endpoint_t *ep,
+	       sctp_association_t *asoc,
+	       void *event_arg,
+	       int priority)
 {
+	sctp_cmd_seq_t commands;
+	sctp_sm_table_entry_t *state_fn;
+	sctp_disposition_t status;
 	int error = 0;
+	typedef const char *(printfn_t)(sctp_subtype_t);
 
-	sctp_bh_lock_sock(asoc->base.sk);
-	if (sock_owned_by_user(asoc->base.sk)) {
-		SCTP_DEBUG_PRINTK("%s:Sock is busy: timer %d\n",
-				  __FUNCTION__,
-				  timeout_type);
-
-		/* Try again later.  */
-		if (!mod_timer(&asoc->timers[timeout_type], jiffies + (HZ/20)))
-			sctp_association_hold(asoc);
-		goto out_unlock;
-	}
+	static printfn_t *table[] = {
+		NULL, sctp_cname, sctp_tname, sctp_oname, sctp_pname,
+	};
+	printfn_t *debug_fn  __attribute__ ((unused)) = table[event_type];
 
-	/* Is this association really dead and just waiting around for
-	 * the timer to let go of the reference?
+	/* Look up the state function, run it, and then process the
+	 * side effects.  These three steps are the heart of lksctp.
 	 */
-	if (asoc->base.dead)
-		goto out_unlock;
+	state_fn = sctp_sm_lookup_event(event_type, state, subtype);
 
-	/* Run through the state machine.  */
-	error = sctp_do_sm(SCTP_EVENT_T_TIMEOUT,
-			   SCTP_ST_TIMEOUT(timeout_type),
-			   asoc->state, asoc->ep, asoc,
-			   (void *)timeout_type,
-			   GFP_ATOMIC);
+	sctp_init_cmd_seq(&commands);
 
-	if (error)
-		asoc->base.sk->err = -error;
+	DEBUG_PRE;
+	status = (*state_fn->fn)(ep, asoc, subtype, event_arg, &commands);
+	DEBUG_POST;
 
-out_unlock:
-	sctp_bh_unlock_sock(asoc->base.sk);
-	sctp_association_put(asoc);
-}
+	error = sctp_side_effects(event_type, subtype, state,
+				  ep, asoc, event_arg,
+				  status, &commands,
+				  priority);
+	DEBUG_POST_SFX;
 
-void sctp_generate_t1_cookie_event(unsigned long data)
-{
-	sctp_association_t *asoc = (sctp_association_t *) data;
-	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_T1_COOKIE);
+	return error;
 }
 
-void sctp_generate_t1_init_event(unsigned long data)
-{
-	sctp_association_t *asoc = (sctp_association_t *) data;
-	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_T1_INIT);
-}
+#undef DEBUG_PRE
+#undef DEBUG_POST
 
-void sctp_generate_t2_shutdown_event(unsigned long data)
+/*****************************************************************
+ * This the master state function side effect processing function.
+ *****************************************************************/
+int sctp_side_effects(sctp_event_t event_type, sctp_subtype_t subtype,
+		      sctp_state_t state,
+		      sctp_endpoint_t *ep,
+		      sctp_association_t *asoc,
+		      void *event_arg,
+		      sctp_disposition_t status,
+		      sctp_cmd_seq_t *commands,
+		      int priority)
 {
-	sctp_association_t *asoc = (sctp_association_t *) data;
-	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_T2_SHUTDOWN);
-}
+	int error;
 
-void sctp_generate_t5_shutdown_guard_event(unsigned long data)
-{
-        sctp_association_t *asoc = (sctp_association_t *)data;
-        sctp_generate_timeout_event(asoc,
-				    SCTP_EVENT_TIMEOUT_T5_SHUTDOWN_GUARD);
+	/* FIXME - Most of the dispositions left today would be categorized
+	 * as "exceptional" dispositions.  For those dispositions, it
+	 * may not be proper to run through any of the commands at all.
+	 * For example, the command interpreter might be run only with
+	 * disposition SCTP_DISPOSITION_CONSUME.
+	 */
+	if (0 != (error = sctp_cmd_interpreter(event_type, subtype, state,
+					       ep, asoc,
+					       event_arg, status,
+					       commands, priority)))
+		goto bail;
 
-} /* sctp_generate_t5_shutdown_guard_event() */
+	switch (status) {
+	case SCTP_DISPOSITION_DISCARD:
+		SCTP_DEBUG_PRINTK("Ignored sctp protocol event - state %d, "
+				  "event_type %d, event_id %d\n",
+				  state, event_type, subtype.chunk);
+		break;
 
-void sctp_generate_autoclose_event(unsigned long data)
-{
-	sctp_association_t *asoc = (sctp_association_t *) data;
-	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_AUTOCLOSE);
-}
+	case SCTP_DISPOSITION_NOMEM:
+		/* We ran out of memory, so we need to discard this
+		 * packet.
+		 */
+		/* BUG--we should now recover some memory, probably by
+		 * reneging...
+		 */
+		error = -ENOMEM;
+		break;
 
-/* Generate a heart beat event.  If the sock is busy, reschedule.   Make
- * sure that the transport is still valid.
- */
-void sctp_generate_heartbeat_event(unsigned long data)
-{
-	int error = 0;
-	struct sctp_transport *transport = (struct sctp_transport *) data;
-	sctp_association_t *asoc = transport->asoc;
+        case SCTP_DISPOSITION_DELETE_TCB:
+		/* This should now be a command. */
+		break;
 
-	sctp_bh_lock_sock(asoc->base.sk);
-	if (sock_owned_by_user(asoc->base.sk)) {
-		SCTP_DEBUG_PRINTK("%s:Sock is busy.\n", __FUNCTION__);
+	case SCTP_DISPOSITION_CONSUME:
+	case SCTP_DISPOSITION_ABORT:
+		/*
+		 * We should no longer have much work to do here as the
+		 * real work has been done as explicit commands above.
+		 */
+		break;
 
-		/* Try again later.  */
-		if (!mod_timer(&transport->hb_timer, jiffies + (HZ/20)))
-			sctp_transport_hold(transport);
-		goto out_unlock;
-	}
+	case SCTP_DISPOSITION_VIOLATION:
+		printk(KERN_ERR "sctp protocol violation state %d "
+		       "chunkid %d\n", state, subtype.chunk);
+		break;
 
-	/* Is this structure just waiting around for us to actually
-	 * get destroyed?
-	 */
-	if (transport->dead)
-		goto out_unlock;
+	case SCTP_DISPOSITION_NOT_IMPL:
+		printk(KERN_WARNING "sctp unimplemented feature in state %d, "
+		       "event_type %d, event_id %d\n",
+		       state, event_type, subtype.chunk);
+		break;
 
-	error = sctp_do_sm(SCTP_EVENT_T_TIMEOUT,
-			   SCTP_ST_TIMEOUT(SCTP_EVENT_TIMEOUT_HEARTBEAT),
-			   asoc->state,
-			   asoc->ep, asoc,
-			   transport, GFP_ATOMIC);
+	case SCTP_DISPOSITION_BUG:
+		printk(KERN_ERR "sctp bug in state %d, "
+		       "event_type %d, event_id %d\n",
+		       state, event_type, subtype.chunk);
+		BUG();
+		break;
 
-         if (error)
-		 asoc->base.sk->err = -error;
+	default:
+		printk(KERN_ERR "sctp impossible disposition %d "
+		       "in state %d, event_type %d, event_id %d\n",
+		       status, state, event_type, subtype.chunk);
+		BUG();
+		break;
+	};
 
-out_unlock:
-	sctp_bh_unlock_sock(asoc->base.sk);
-	sctp_transport_put(transport);
-}
-
-/* Inject a SACK Timeout event into the state machine.  */
-void sctp_generate_sack_event(unsigned long data)
-{
-	sctp_association_t *asoc = (sctp_association_t *) data;
-	sctp_generate_timeout_event(asoc, SCTP_EVENT_TIMEOUT_SACK);
+bail:
+	return error;
 }
 
-sctp_timer_event_t *sctp_timer_events[SCTP_NUM_TIMEOUT_TYPES] = {
-	NULL,
-	sctp_generate_t1_cookie_event,
-	sctp_generate_t1_init_event,
-	sctp_generate_t2_shutdown_event,
-	NULL,
-	sctp_generate_t5_shutdown_guard_event,
-	sctp_generate_heartbeat_event,
-	sctp_generate_sack_event,
-	sctp_generate_autoclose_event,
-};
-
 /********************************************************************
- * 3rd Level Abstractions
+ * 2nd Level Abstractions
  ********************************************************************/
 
-/* RFC 2960 8.2 Path Failure Detection
- *
- * When its peer endpoint is multi-homed, an endpoint should keep a
- * error counter for each of the destination transport addresses of the
- * peer endpoint.
- *
- * Each time the T3-rtx timer expires on any address, or when a
- * HEARTBEAT sent to an idle address is not acknowledged within a RTO,
- * the error counter of that destination address will be incremented.
- * When the value in the error counter exceeds the protocol parameter
- * 'Path.Max.Retrans' of that destination address, the endpoint should
- * mark the destination transport address as inactive, and a
- * notification SHOULD be sent to the upper layer.
- *
- */
-static void sctp_do_8_2_transport_strike(sctp_association_t *asoc,
-					 struct sctp_transport *transport)
+/* This is the side-effect interpreter.  */
+int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
+			 sctp_state_t state, sctp_endpoint_t *ep,
+			 sctp_association_t *asoc, void *event_arg,
+			 sctp_disposition_t status, sctp_cmd_seq_t *commands,
+			 int priority)
 {
-	/* The check for association's overall error counter exceeding the
-	 * threshold is done in the state function.
-	 */
-	asoc->overall_error_count++;
+	int error = 0;
+	int force;
+	sctp_cmd_t *cmd;
+	sctp_chunk_t *new_obj;
+	sctp_chunk_t *chunk = NULL;
+	struct sctp_packet *packet;
+	struct list_head *pos;
+	struct timer_list *timer;
+	unsigned long timeout;
+	struct sctp_transport *t;
+	sctp_sackhdr_t sackh;
 
-	if (transport->active &&
-	    (transport->error_count++ >= transport->error_threshold)) {
-		SCTP_DEBUG_PRINTK("transport_strike: transport "
-				  "IP:%d.%d.%d.%d failed.\n",
-				  NIPQUAD(transport->ipaddr.v4.sin_addr));
-		sctp_assoc_control_transport(asoc, transport,
-					     SCTP_TRANSPORT_DOWN,
-					     SCTP_FAILED_THRESHOLD);
-	}
+	if(SCTP_EVENT_T_TIMEOUT != event_type)
+		chunk = (sctp_chunk_t *) event_arg;
 
-	/* E2) For the destination address for which the timer
-	 * expires, set RTO <- RTO * 2 ("back off the timer").  The
-	 * maximum value discussed in rule C7 above (RTO.max) may be
-	 * used to provide an upper bound to this doubling operation.
+	/* Note:  This whole file is a huge candidate for rework.
+	 * For example, each command could either have its own handler, so
+	 * the loop would look like:
+	 *     while (cmds)
+	 *         cmd->handle(x, y, z)
+	 * --jgrimm
 	 */
-	transport->rto = min((transport->rto * 2), transport->asoc->rto_max);
-}
+	while (NULL != (cmd = sctp_next_cmd(commands))) {
+		switch (cmd->verb) {
+		case SCTP_CMD_NOP:
+			/* Do nothing. */
+			break;
 
-/* Worker routine to handle INIT command failure.  */
-static void sctp_cmd_init_failed(sctp_cmd_seq_t *commands,
-				 sctp_association_t *asoc)
-{
-	struct sctp_ulpevent *event;
+		case SCTP_CMD_NEW_ASOC:
+			/* Register a new association.  */
+			asoc = cmd->obj.ptr;
+			/* Register with the endpoint.  */
+			sctp_endpoint_add_asoc(ep, asoc);
+			sctp_hash_established(asoc);
+			break;
 
-	event = sctp_ulpevent_make_assoc_change(asoc,
-						0,
-						SCTP_CANT_STR_ASSOC,
-						0, 0, 0,
-						GFP_ATOMIC);
+		case SCTP_CMD_UPDATE_ASSOC:
+		       sctp_assoc_update(asoc, cmd->obj.ptr);
+		       break;
 
-	if (event)
-		sctp_add_cmd_sf(commands, SCTP_CMD_EVENT_ULP,
-				SCTP_ULPEVENT(event));
+		case SCTP_CMD_PURGE_OUTQUEUE:
+		       sctp_outq_teardown(&asoc->outqueue);
+		       break;
 
-	/* FIXME:  We need to handle data possibly either
-	 * sent via COOKIE-ECHO bundling or just waiting in
-	 * the transmit queue, if the user has enabled
-	 * SEND_FAILED notifications.
-	 */
-	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
-}
+		case SCTP_CMD_DELETE_TCB:
+			/* Delete the current association.  */
+			sctp_unhash_established(asoc);
+			sctp_association_free(asoc);
+			asoc = NULL;
+			break;
 
-/* Worker routine to handle SCTP_CMD_ASSOC_FAILED.  */
-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)
-{
-	struct sctp_ulpevent *event;
-	__u16 error = 0;
+		case SCTP_CMD_NEW_STATE:
+			/* Enter a new state.  */
+			sctp_cmd_new_state(commands, asoc, cmd->obj.state);
+			break;
 
-	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;
-	}
+		case SCTP_CMD_REPORT_TSN:
+			/* Record the arrival of a TSN.  */
+			sctp_tsnmap_mark(&asoc->peer.tsn_map, cmd->obj.u32);
+			break;
 
-	/* Cancel any partial delivery in progress. */
-	sctp_ulpq_abort_pd(&asoc->ulpq, GFP_ATOMIC);
+		case SCTP_CMD_GEN_SACK:
+			/* Generate a Selective ACK.
+			 * The argument tells us whether to just count
+			 * the packet and MAYBE generate a SACK, or
+			 * force a SACK out.
+			 */
+			force = cmd->obj.i32;
+			error = sctp_gen_sack(asoc, force, commands);
+			break;
 
-	event = sctp_ulpevent_make_assoc_change(asoc, 0, SCTP_COMM_LOST,
-						error, 0, 0, GFP_ATOMIC);
-	if (event)
-		sctp_add_cmd_sf(commands, SCTP_CMD_EVENT_ULP,
-				SCTP_ULPEVENT(event));
+		case SCTP_CMD_PROCESS_SACK:
+			/* Process an inbound SACK.  */
+			error = sctp_cmd_process_sack(commands, asoc,
+						      cmd->obj.ptr);
+			break;
 
-	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
-			SCTP_STATE(SCTP_STATE_CLOSED));
+		case SCTP_CMD_GEN_INIT_ACK:
+			/* Generate an INIT ACK chunk.  */
+			new_obj = sctp_make_init_ack(asoc, chunk, GFP_ATOMIC,
+						     0);
+			if (!new_obj)
+				goto nomem;
 
-	/* FIXME:  We need to handle data that could not be sent or was not
-	 * acked, if the user has enabled SEND_FAILED notifications.
-	 */
-	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
-}
+			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
+					SCTP_CHUNK(new_obj));
+			break;
 
-/* Process an init chunk (may be real INIT/INIT-ACK or an embedded INIT
- * inside the cookie.  In reality, this is only used for INIT-ACK processing
- * since all other cases use "temporary" associations and can do all
- * their work in statefuns directly.
- */
-static int sctp_cmd_process_init(sctp_cmd_seq_t *commands,
-				 sctp_association_t *asoc,
-				 sctp_chunk_t *chunk,
-				 sctp_init_chunk_t *peer_init,
-				 int priority)
-{
-	int error;
+		case SCTP_CMD_PEER_INIT:
+			/* Process a unified INIT from the peer.
+			 * Note: Only used during INIT-ACK processing.  If
+			 * there is an error just return to the outter
+			 * layer which will bail.
+			 */
+			error = sctp_cmd_process_init(commands, asoc, chunk,
+						      cmd->obj.ptr, priority);
+			break;
 
-	/* We only process the init as a sideeffect in a single
-	 * case.   This is when we process the INIT-ACK.   If we
-	 * fail during INIT processing (due to malloc problems),
-	 * just return the error and stop processing the stack.
-	 */
+		case SCTP_CMD_GEN_COOKIE_ECHO:
+			/* Generate a COOKIE ECHO chunk.  */
+			new_obj = sctp_make_cookie_echo(asoc, chunk);
+			if (!new_obj) {
+				if (cmd->obj.ptr)
+					sctp_free_chunk(cmd->obj.ptr);
+				goto nomem;
+			}
+			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
+					SCTP_CHUNK(new_obj));
 
-	if (!sctp_process_init(asoc, chunk->chunk_hdr->type,
-			       sctp_source(chunk), peer_init,
-			       priority))
-		error = -ENOMEM;
-	else
-		error = 0;
+			/* If there is an ERROR chunk to be sent along with
+			 * the COOKIE_ECHO, send it, too.
+			 */
+			if (cmd->obj.ptr)
+				sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
+						SCTP_CHUNK(cmd->obj.ptr));
+			break;
+
+		case SCTP_CMD_GEN_SHUTDOWN:
+			/* Generate SHUTDOWN when in SHUTDOWN_SENT state.
+			 * Reset error counts.
+			 */
+			asoc->overall_error_count = 0;
+
+			/* Generate a SHUTDOWN chunk.  */
+			new_obj = sctp_make_shutdown(asoc);
+			if (!new_obj)
+				goto nomem;
+			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
+					SCTP_CHUNK(new_obj));
+			break;
+
+		case SCTP_CMD_CHUNK_ULP:
+			/* Send a chunk to the sockets layer.  */
+			SCTP_DEBUG_PRINTK("sm_sideff: %s %p, %s %p.\n",
+					  "chunk_up:", cmd->obj.ptr,
+					  "ulpq:", &asoc->ulpq);
+			sctp_ulpq_tail_data(&asoc->ulpq, cmd->obj.ptr,
+					    GFP_ATOMIC);
+			break;
+
+		case SCTP_CMD_EVENT_ULP:
+			/* Send a notification to the sockets layer.  */
+			SCTP_DEBUG_PRINTK("sm_sideff: %s %p, %s %p.\n",
+					  "event_up:",cmd->obj.ptr,
+					  "ulpq:",&asoc->ulpq);
+			sctp_ulpq_tail_event(&asoc->ulpq, cmd->obj.ptr);
+			break;
+
+		case SCTP_CMD_REPLY:
+			/* Send a chunk to our peer.  */
+			error = sctp_outq_tail(&asoc->outqueue,
+					       cmd->obj.ptr);
+			break;
+
+		case SCTP_CMD_SEND_PKT:
+			/* Send a full packet to our peer.  */
+			packet = cmd->obj.ptr;
+			sctp_packet_transmit(packet);
+			sctp_ootb_pkt_free(packet);
+			break;
+
+		case SCTP_CMD_RETRAN:
+			/* Mark a transport for retransmission.  */
+			sctp_retransmit(&asoc->outqueue, cmd->obj.transport,
+					SCTP_RTXR_T3_RTX);
+			break;
+
+		case SCTP_CMD_TRANSMIT:
+			/* Kick start transmission. */
+			error = sctp_outq_flush(&asoc->outqueue, 0);
+			break;
+
+		case SCTP_CMD_ECN_CE:
+			/* Do delayed CE processing.   */
+			sctp_do_ecn_ce_work(asoc, cmd->obj.u32);
+			break;
+
+		case SCTP_CMD_ECN_ECNE:
+			/* Do delayed ECNE processing. */
+			new_obj = sctp_do_ecn_ecne_work(asoc, cmd->obj.u32,
+							chunk);
+			if (new_obj)
+				sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
+						SCTP_CHUNK(new_obj));
+			break;
+
+		case SCTP_CMD_ECN_CWR:
+			/* Do delayed CWR processing.  */
+			sctp_do_ecn_cwr_work(asoc, cmd->obj.u32);
+			break;
+
+		case SCTP_CMD_SETUP_T2:
+			sctp_cmd_setup_t2(commands, asoc, cmd->obj.ptr);
+			break;
+
+		case SCTP_CMD_TIMER_START:
+			timer = &asoc->timers[cmd->obj.to];
+			timeout = asoc->timeouts[cmd->obj.to];
+			if (!timeout)
+				BUG();
+
+			timer->expires = jiffies + timeout;
+			sctp_association_hold(asoc);
+			add_timer(timer);
+			break;
 
-	return error;
-}
+		case SCTP_CMD_TIMER_RESTART:
+			timer = &asoc->timers[cmd->obj.to];
+			timeout = asoc->timeouts[cmd->obj.to];
+			if (!mod_timer(timer, jiffies + timeout))
+				sctp_association_hold(asoc);
+			break;
 
-/* Helper function to break out starting up of heartbeat timers.  */
-static void sctp_cmd_hb_timers_start(sctp_cmd_seq_t *cmds,
-				     sctp_association_t *asoc)
-{
-	struct sctp_transport *t;
-	struct list_head *pos;
+		case SCTP_CMD_TIMER_STOP:
+			timer = &asoc->timers[cmd->obj.to];
+			if (timer_pending(timer) && del_timer(timer))
+				sctp_association_put(asoc);
+			break;
 
-	/* Start a heartbeat timer for each transport on the association.
-	 * hold a reference on the transport to make sure none of
-	 * the needed data structures go away.
-	 */
-	list_for_each(pos, &asoc->peer.transport_addr_list) {
-		t = list_entry(pos, struct sctp_transport, transports);
+		case SCTP_CMD_INIT_RESTART:
+			/* Do the needed accounting and updates
+			 * associated with restarting an initialization
+			 * timer.
+			 */
+			asoc->counters[SCTP_COUNTER_INIT_ERROR]++;
+			asoc->timeouts[cmd->obj.to] *= 2;
+			if (asoc->timeouts[cmd->obj.to] >
+			    asoc->max_init_timeo) {
+				asoc->timeouts[cmd->obj.to] =
+					asoc->max_init_timeo;
+			}
 
-		if (!mod_timer(&t->hb_timer, sctp_transport_timeout(t)))
-			sctp_transport_hold(t);
-	}
-}
+			/* 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, struct sctp_transport,
+					       transports);
+				sctp_retransmit_mark(&asoc->outqueue, t, 0);
+			}
 
-static void sctp_cmd_hb_timers_stop(sctp_cmd_seq_t *cmds,
-				    sctp_association_t *asoc)
-{
-	struct sctp_transport *t;
-	struct list_head *pos;
+			sctp_add_cmd_sf(commands,
+					SCTP_CMD_TIMER_RESTART,
+					SCTP_TO(cmd->obj.to));
+			break;
 
-	/* Stop all heartbeat timers. */
+		case SCTP_CMD_INIT_FAILED:
+			sctp_cmd_init_failed(commands, asoc);
+			break;
 
-	list_for_each(pos, &asoc->peer.transport_addr_list) {
-		t = list_entry(pos, struct sctp_transport, transports);
-		if (del_timer(&t->hb_timer))
-			sctp_transport_put(t);
-	}
-}
+		case SCTP_CMD_ASSOC_FAILED:
+			sctp_cmd_assoc_failed(commands, asoc, event_type,
+					      subtype, chunk);
+			break;
 
-/* Helper function to update the heartbeat timer. */
-static void sctp_cmd_hb_timer_update(sctp_cmd_seq_t *cmds,
-				     sctp_association_t *asoc,
-				     struct sctp_transport *t)
-{
-	/* Update the heartbeat timer.  */
-	if (!mod_timer(&t->hb_timer, sctp_transport_timeout(t)))
-		sctp_transport_hold(t);
-}
+		case SCTP_CMD_COUNTER_INC:
+			asoc->counters[cmd->obj.counter]++;
+			break;
 
-/* Helper function to handle the reception of an HEARTBEAT ACK.  */
-static void sctp_cmd_transport_on(sctp_cmd_seq_t *cmds,
-				  sctp_association_t *asoc,
-				  struct sctp_transport *t,
-				  sctp_chunk_t *chunk)
-{
-	sctp_sender_hb_info_t *hbinfo;
+		case SCTP_CMD_COUNTER_RESET:
+			asoc->counters[cmd->obj.counter] = 0;
+			break;
 
-	/* 8.3 Upon the receipt of the HEARTBEAT ACK, the sender of the
-	 * HEARTBEAT should clear the error counter of the destination
-	 * transport address to which the HEARTBEAT was sent.
-	 * The association's overall error count is also cleared.
-	 */
-	t->error_count = 0;
-	t->asoc->overall_error_count = 0;
+		case SCTP_CMD_REPORT_DUP:
+			sctp_tsnmap_mark_dup(&asoc->peer.tsn_map, 
+					     cmd->obj.u32);
+			break;
 
-	/* Mark the destination transport address as active if it is not so
-	 * marked.
-	 */
-	if (!t->active)
-		sctp_assoc_control_transport(asoc, t, SCTP_TRANSPORT_UP,
-					     SCTP_HEARTBEAT_SUCCESS);
+		case SCTP_CMD_REPORT_BAD_TAG:
+			SCTP_DEBUG_PRINTK("vtag mismatch!\n");
+			break;
 
-	/* The receiver of the HEARTBEAT ACK should also perform an
-	 * RTT measurement for that destination transport address
-	 * using the time value carried in the HEARTBEAT ACK chunk.
-	 */
-	hbinfo = (sctp_sender_hb_info_t *) chunk->skb->data;
-	sctp_transport_update_rto(t, (jiffies - hbinfo->sent_at));
-}
+		case SCTP_CMD_STRIKE:
+			/* Mark one strike against a transport.  */
+			sctp_do_8_2_transport_strike(asoc, cmd->obj.transport);
+			break;
 
-/* Helper function to do a transport reset at the expiry of the hearbeat
- * timer.
- */
-static void sctp_cmd_transport_reset(sctp_cmd_seq_t *cmds,
-				     sctp_association_t *asoc,
-				     struct sctp_transport *t)
-{
-	sctp_transport_lower_cwnd(t, SCTP_LOWER_CWND_INACTIVE);
+		case SCTP_CMD_TRANSPORT_RESET:
+			t = cmd->obj.transport;
+			sctp_cmd_transport_reset(commands, asoc, t);
+			break;
 
-	/* Mark one strike against a transport.  */
-	sctp_do_8_2_transport_strike(asoc, t);
-}
+		case SCTP_CMD_TRANSPORT_ON:
+			t = cmd->obj.transport;
+			sctp_cmd_transport_on(commands, asoc, t, chunk);
+			break;
 
-/* Helper function to process the process SACK command.  */
-static int sctp_cmd_process_sack(sctp_cmd_seq_t *cmds,
-				 sctp_association_t *asoc,
-				 sctp_sackhdr_t *sackh)
-{
-	int err;
+		case SCTP_CMD_HB_TIMERS_START:
+			sctp_cmd_hb_timers_start(commands, asoc);
+			break;
 
-	if (sctp_outq_sack(&asoc->outqueue, sackh)) {
-		/* There are no more TSNs awaiting SACK.  */
-		err = sctp_do_sm(SCTP_EVENT_T_OTHER,
-				 SCTP_ST_OTHER(SCTP_EVENT_NO_PENDING_TSN),
-				 asoc->state, asoc->ep, asoc, NULL,
-				 GFP_ATOMIC);
-	} else {
-		/* Windows may have opened, so we need
-		 * to check if we have DATA to transmit
-		 */
-		err = sctp_outq_flush(&asoc->outqueue, 0);
-	}
+		case SCTP_CMD_HB_TIMER_UPDATE:
+			t = cmd->obj.transport;
+			sctp_cmd_hb_timer_update(commands, asoc, t);
+			break;
 
-	return err;
-}
+		case SCTP_CMD_HB_TIMERS_STOP:
+			sctp_cmd_hb_timers_stop(commands, asoc);
+			break;
 
-/* Helper function to set the timeout value for T2-SHUTDOWN timer and to set
- * the transport for a shutdown chunk.
- */
-static void sctp_cmd_setup_t2(sctp_cmd_seq_t *cmds, sctp_association_t *asoc,
-			      sctp_chunk_t *chunk)
-{
-	struct sctp_transport *t;
+		case SCTP_CMD_REPORT_ERROR:
+			error = cmd->obj.error;
+			break;
 
-	t = sctp_assoc_choose_shutdown_transport(asoc);
-	asoc->shutdown_last_sent_to = t;
-	asoc->timeouts[SCTP_EVENT_TIMEOUT_T2_SHUTDOWN] = t->rto;
-	chunk->transport = t;
-}
+		case SCTP_CMD_PROCESS_CTSN:
+			/* Dummy up a SACK for processing. */
+			sackh.cum_tsn_ack = cmd->obj.u32;
+			sackh.a_rwnd = 0;
+			sackh.num_gap_ack_blocks = 0;
+			sackh.num_dup_tsns = 0;
+			sctp_add_cmd_sf(commands, SCTP_CMD_PROCESS_SACK,
+					SCTP_SACKH(&sackh));
+			break;
 
-/* Helper function to change the state of an association. */
-static void sctp_cmd_new_state(sctp_cmd_seq_t *cmds, sctp_association_t *asoc,
-			       sctp_state_t state)
-{
+		case SCTP_CMD_DISCARD_PACKET:
+			/* We need to discard the whole packet.  */
+			chunk->pdiscard = 1;
+			break;
 
-	struct sock *sk = asoc->base.sk;
-	struct sctp_opt *sp = sctp_sk(sk);
+		case SCTP_CMD_RTO_PENDING:
+			t = cmd->obj.transport;
+			t->rto_pending = 1;
+			break;
 
-	asoc->state = state;
-	asoc->state_timestamp = jiffies;
+		case SCTP_CMD_PART_DELIVER:
+			sctp_ulpq_partial_delivery(&asoc->ulpq, cmd->obj.ptr,
+						   GFP_ATOMIC);
+			break;
 
-	if ((SCTP_STATE_ESTABLISHED == asoc->state) ||
-	    (SCTP_STATE_CLOSED == asoc->state)) { 
-		/* Wake up any processes waiting in the asoc's wait queue in
-		 * sctp_wait_for_connect() or sctp_wait_for_sndbuf(). 
-	 	 */
-		if (waitqueue_active(&asoc->wait))
-			wake_up_interruptible(&asoc->wait);
+		case SCTP_CMD_RENEGE:
+			sctp_ulpq_renege(&asoc->ulpq, cmd->obj.ptr,
+					 GFP_ATOMIC);
+			break;
 
-		/* Wake up any processes waiting in the sk's sleep queue of
-		 * a TCP-style or UDP-style peeled-off socket in
-		 * sctp_wait_for_accept() or sctp_wait_for_packet().
-		 * For a UDP-style socket, the waiters are woken up by the
-		 * notifications.
-		 */
-		if (SCTP_SOCKET_UDP != sp->type)
-			sk->state_change(sk);
+		default:
+			printk(KERN_WARNING "Impossible command: %u, %p\n",
+			       cmd->verb, cmd->obj.ptr);
+			break;
+		};
+		if (error)
+			return error;
 	}
 
-	/* Change the sk->state of a TCP-style socket that has sucessfully
-	 * completed a connect() call.
-	 */
-	if ((SCTP_STATE_ESTABLISHED == asoc->state) &&
-	    (SCTP_SOCKET_TCP == sp->type) && (SCTP_SS_CLOSED == sk->state))
-	    sk->state = SCTP_SS_ESTABLISHED;
+	return error;
+
+nomem:
+	error = -ENOMEM;
+	return error;
 }
+