Merge dyn9-47-18-140.beaverton.ibm.com:/home/sridhar/BK/linux-2.5.46

into dyn9-47-18-140.beaverton.ibm.com:/home/sridhar/BK/lksctp-2.5.46

include/net/sctp/command.h

@@ -27,6 +27,7 @@
  * 
  * La Monte H.P. Yarroll <piggy@acm.org>
  * Karl Knutson <karl@athena.chicago.il.us>
+ * 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.
@@ -72,6 +73,7 @@ typedef enum {
 	SCTP_CMD_STRIKE,	/* Mark a strike against a transport.  */
 	SCTP_CMD_TRANSMIT,      /* Transmit the outqueue. */
 	SCTP_CMD_HB_TIMERS_START,    /* Start the heartbeat timers.  */
+	SCTP_CMD_HB_TIMERS_UPDATE,   /* Update the heartbeat timers.  */
 	SCTP_CMD_TRANSPORT_RESET,    /* Reset the status of a transport. */
 	SCTP_CMD_TRANSPORT_ON,       /* Mark the transport as active. */
 	SCTP_CMD_REPORT_ERROR,   /* Pass this error back out of the sm. */

include/net/sctp/sctp.h

@@ -3,41 +3,41 @@
  * Copyright (c) 1999-2001 Motorola, Inc.
  * Copyright (c) 2001 International Business Machines, Corp.
  * Copyright (c) 2001 Intel Corp.
- * 
+ *
  * This file is part of the SCTP kernel reference Implementation
- * 
- * The base lksctp header. 
- * 
- * The SCTP reference implementation is free software; 
- * you can redistribute it and/or modify it under the terms of 
+ *
+ * The base lksctp header.
+ *
+ * The SCTP reference implementation is free software;
+ * you can redistribute it and/or modify it under the terms of
  * the GNU General Public License as published by
  * the Free Software Foundation; either version 2, or (at your option)
  * any later version.
- * 
- * The SCTP reference implementation is distributed in the hope that it 
+ *
+ * The SCTP reference implementation is distributed in the hope that it
  * will be useful, but WITHOUT ANY WARRANTY; without even the implied
  *                 ************************
  * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  * See the GNU General Public License for more details.
- * 
+ *
  * You should have received a copy of the GNU General Public License
  * along with GNU CC; see the file COPYING.  If not, write to
  * the Free Software Foundation, 59 Temple Place - Suite 330,
- * Boston, MA 02111-1307, USA.  
- * 
+ * Boston, MA 02111-1307, USA.
+ *
  * Please send any bug reports or fixes you make to the
  * email address(es):
  *    lksctp developers <lksctp-developers@lists.sourceforge.net>
- * 
+ *
  * Or submit a bug report through the following website:
  *    http://www.sf.net/projects/lksctp
  *
- * Written or modified by: 
+ * Written or modified by:
  *    La Monte H.P. Yarroll <piggy@acm.org>
  *    Xingang Guo           <xingang.guo@intel.com>
  *    Jon Grimm             <jgrimm@us.ibm.com>
  *    Daisy Chang	    <daisyc@us.ibm.com>
- * 
+ *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
  */
@@ -52,10 +52,10 @@
  *       structs
  *       prototypes
  *       macros, externs, and inlines
- * 
- *   Move test_frame specific items out of the kernel headers 
+ *
+ *   Move test_frame specific items out of the kernel headers
  *   and into the test frame headers.   This is not perfect in any sense
- *   and will continue to evolve.  
+ *   and will continue to evolve.
  */
 
 
@@ -78,11 +78,12 @@
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 #include <net/ipv6.h>
 #include <net/ip6_route.h>
-#endif 
+#endif
 
 #include <asm/uaccess.h>
 #include <asm/page.h>
 #include <net/sock.h>
+#include <net/snmp.h>
 #include <net/sctp/structs.h>
 #include <net/sctp/constants.h>
 #include <net/sctp/sm.h>
@@ -104,19 +105,19 @@
 #endif
 
 
-/* Certain internal static functions need to be exported when 
+/* Certain internal static functions need to be exported when
  * compiled into the test frame.
  */
 #ifndef SCTP_STATIC
 #define SCTP_STATIC static
 #endif
 
-/* 
- * Function declarations. 
+/*
+ * Function declarations.
  */
 
 /*
- * sctp_protocol.c 
+ * sctp_protocol.c
  */
 extern sctp_protocol_t sctp_proto;
 extern struct sock *sctp_get_ctl_sock(void);
@@ -141,7 +142,7 @@ extern int sctp_primitive_ASSOCIATE(sctp_association_t *, void *arg);
 extern int sctp_primitive_SHUTDOWN(sctp_association_t *, void *arg);
 extern int sctp_primitive_ABORT(sctp_association_t *, void *arg);
 extern int sctp_primitive_SEND(sctp_association_t *, void *arg);
-
+extern int sctp_primitive_REQUESTHEARTBEAT(sctp_association_t *, void *arg);
 
 /*
  * sctp_crc32c.c
@@ -201,6 +202,11 @@ extern void sctp_hash_digest(const char *secret, const int secret_len,
 #define SCTP_SOCK_SLEEP_PRE(sk) SOCK_SLEEP_PRE(sk)
 #define SCTP_SOCK_SLEEP_POST(sk) SOCK_SLEEP_POST(sk)
 
+/* SCTP SNMP MIB stats handlers */
+extern struct sctp_mib sctp_statistics[NR_CPUS * 2];
+#define SCTP_INC_STATS(field)		SNMP_INC_STATS(sctp_statistics, field)
+#define SCTP_INC_STATS_BH(field)	SNMP_INC_STATS_BH(sctp_statistics, field)
+#define SCTP_INC_STATS_USER(field)	SNMP_INC_STATS_USER(sctp_statistics, field)
 
 /* Determine if this is a valid kernel address.  */
 static inline int sctp_is_valid_kaddr(unsigned long addr)
@@ -412,6 +418,19 @@ static inline size_t get_user_iov_size(struct iovec *iov, int iovlen)
 	return retval;
 }
 
+/* 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
+ * there is room for a param header too.
+ */
+#define sctp_walk_params(pos, chunk, member)\
+_sctp_walk_params((pos), (chunk), ntohs((chunk)->chunk_hdr.length), member)
+
+#define _sctp_walk_params(pos, chunk, end, member)\
+for (pos.v = chunk->member;\
+     pos.v <= (void *)chunk + end - sizeof(sctp_paramhdr_t) &&\
+     pos.v <= (void *)chunk + end - WORD_ROUND(ntohs(pos.p->length)); \
+     pos.v += WORD_ROUND(ntohs(pos.p->length)))
 
 /* Round an int up to the next multiple of 4.  */
 #define WORD_ROUND(s) (((s)+3)&~3)

include/net/sctp/sm.h

@@ -44,6 +44,7 @@
  *    Dajiang Zhang <dajiang.zhang@nokia.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.
@@ -156,6 +157,7 @@ sctp_state_fn_t sctp_sf_shutdown_ack_sent_prm_abort;
 sctp_state_fn_t sctp_sf_error_closed;
 sctp_state_fn_t sctp_sf_error_shutdown;
 sctp_state_fn_t sctp_sf_ignore_primitive;
+sctp_state_fn_t sctp_sf_do_prm_requestheartbeat;
 
 /* Prototypes for other event state functions.  */
 sctp_state_fn_t sctp_sf_do_9_2_start_shutdown;
@@ -205,9 +207,6 @@ sctp_association_t *sctp_make_temp_asoc(const sctp_endpoint_t *,
 					sctp_chunk_t *,
 					const int priority);
 __u32 sctp_generate_verification_tag(void);
-sctpParam_t sctp_get_my_addrs_raw(const sctp_association_t *,
-				  const int priority, int *addrs_len);
-
 void sctp_populate_tie_tags(__u8 *cookie, __u32 curTag, __u32 hisTag);
 
 /* Prototypes for chunk-building functions.  */
@@ -253,6 +252,9 @@ sctp_chunk_t *sctp_make_abort(const sctp_association_t *,
 sctp_chunk_t *sctp_make_abort_no_data(const sctp_association_t *,
 				      const sctp_chunk_t *,
 				      __u32 tsn);
+sctp_chunk_t *sctp_make_abort_user(const sctp_association_t *,
+				   const sctp_chunk_t *,
+				   const struct msghdr *);
 sctp_chunk_t *sctp_make_heartbeat(const sctp_association_t *,
 				  const sctp_transport_t *,
 				  const void *payload,
@@ -332,7 +334,7 @@ __u32 sctp_generate_tsn(const sctp_endpoint_t *);
 /* 4th level prototypes */
 void sctp_param2sockaddr(sockaddr_storage_t *addr, sctp_addr_param_t *,
 			 __u16 port);
-int sctp_addr2sockaddr(const sctpParam_t, sockaddr_storage_t *);
+int sctp_addr2sockaddr(const union sctp_params, sockaddr_storage_t *);
 int sockaddr2sctp_addr(const sockaddr_storage_t *, sctp_addr_param_t *);
 
 /* Extern declarations for major data structures.  */

include/net/sctp/structs.h

@@ -3,34 +3,34 @@
  * Copyright (c) 1999-2001 Motorola, Inc.
  * Copyright (c) 2001 Intel Corp.
  * Copyright (c) 2001 International Business Machines Corp.
- * 
+ *
  * This file is part of the SCTP kernel reference Implementation
- * 
- * The SCTP reference implementation is free software; 
- * you can redistribute it and/or modify it under the terms of 
+ *
+ * The SCTP reference implementation is free software;
+ * you can redistribute it and/or modify it under the terms of
  * the GNU General Public License as published by
  * the Free Software Foundation; either version 2, or (at your option)
  * any later version.
- * 
- * The SCTP reference implementation is distributed in the hope that it 
+ *
+ * The SCTP reference implementation is distributed in the hope that it
  * will be useful, but WITHOUT ANY WARRANTY; without even the implied
  *                 ************************
  * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  * See the GNU General Public License for more details.
- * 
+ *
  * You should have received a copy of the GNU General Public License
  * along with GNU CC; see the file COPYING.  If not, write to
  * the Free Software Foundation, 59 Temple Place - Suite 330,
- * Boston, MA 02111-1307, USA.  
- * 
+ * Boston, MA 02111-1307, USA.
+ *
  * Please send any bug reports or fixes you make to the
  * email addresses:
  *    lksctp developers <lksctp-developers@lists.sourceforge.net>
- * 
+ *
  * Or submit a bug report through the following website:
  *    http://www.sf.net/projects/lksctp
  *
- * Written or modified by: 
+ * Written or modified by:
  *    Randall Stewart       <randall@sctp.chicago.il.us>
  *    Ken Morneau           <kmorneau@cisco.com>
  *    Qiaobing Xie          <qxie1@email.mot.com>
@@ -41,8 +41,8 @@
  *    Hui Huang             <hui.huang@nokia.com>
  *    Sridhar Samudrala     <sri@us.ibm.com>
  *    Daisy Chang	    <daisyc@us.ibm.com>
- *    Dajiang Zhang         <dajiang.zhang@nokia.com> 
- * 
+ *    Dajiang Zhang         <dajiang.zhang@nokia.com>
+ *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
  */
@@ -246,8 +246,11 @@ typedef struct sctp_func {
 					 int optname,
 					 char *optval,
 					 int *optlen);
-	int		(*get_dst_mtu)	(const sockaddr_storage_t *address);
-	__u16		net_header_len;	
+	struct dst_entry *(*get_dst)	(sockaddr_storage_t *daddr,
+					 sockaddr_storage_t *saddr);
+	int		(*cmp_saddr)	(struct dst_entry *dst,
+					 sockaddr_storage_t *saddr);
+	__u16		net_header_len;
 	int		sockaddr_len;
 	sa_family_t	sa_family;
 	struct list_head list;
@@ -356,20 +359,6 @@ typedef struct sctp_signed_cookie {
 } sctp_signed_cookie_t;
 
 
-/* This convenience type allows us to avoid casting when walking
- * through a parameter list.
- */
-typedef union {
-	__u8 *v;
-	sctp_paramhdr_t *p;
-
-	sctp_cookie_preserve_param_t *bht;
-	sctp_hostname_param_t *dns;
-	sctp_cookie_param_t *cookie;
-	sctp_supported_addrs_param_t *sat;
-	sctp_ipv4addr_param_t *v4;
-	sctp_ipv6addr_param_t *v6;
-} sctpParam_t;
 
 /* This is another convenience type to allocate memory for address
  * params for the maximum size and pass such structures around
@@ -380,6 +369,21 @@ typedef union {
 	sctp_ipv6addr_param_t v6;
 } sctp_addr_param_t;
 
+/* A convenience type to allow walking through the various
+ * parameters and avoid casting all over the place.
+ */
+union sctp_params {
+	void *v;
+	sctp_paramhdr_t *p;
+	sctp_cookie_preserve_param_t *life;
+	sctp_hostname_param_t *dns;
+	sctp_cookie_param_t *cookie;
+	sctp_supported_addrs_param_t *sat;
+	sctp_ipv4addr_param_t *v4;
+	sctp_ipv6addr_param_t *v6;
+	sctp_addr_param_t *addr;
+};
+
 /* RFC 2960.  Section 3.3.5 Heartbeat.
  *    Heartbeat Information: variable length
  *    The Sender-specific Heartbeat Info field should normally include
@@ -430,7 +434,7 @@ struct SCTP_chunk {
 	 */
 
 	/* We point this at the FIRST TLV parameter to chunk_hdr.  */
-	sctpParam_t param_hdr;
+	union sctp_params param_hdr;
 	union {
 		__u8 *v;
 		sctp_datahdr_t *data_hdr;
@@ -471,11 +475,13 @@ struct SCTP_chunk {
 	__u8 ecn_ce_done;      /* Have we processed the ECN CE bit? */
 	__u8 pdiscard;	  /* Discard the whole packet now? */
 	__u8 tsn_gap_acked;	  /* Is this chunk acked by a GAP ACK? */
-	__u8 fast_retransmit;    /* Is this chunk fast retransmitted? */ 
+	__u8 fast_retransmit;    /* Is this chunk fast retransmitted? */
 	__u8 tsn_missing_report; /* Data chunk missing counter. */
 
 	/* What is the origin IP address for this chunk?  */
 	sockaddr_storage_t source;
+	/* Destination address for this chunk. */
+	sockaddr_storage_t dest;
 
 	/* For an inbound chunk, this tells us where it came from.
 	 * For an outbound chunk, it tells us where we'd like it to
@@ -492,7 +498,7 @@ void  *sctp_addto_chunk(sctp_chunk_t *chunk, int len, const void *data);
 int sctp_user_addto_chunk(sctp_chunk_t *chunk, int len, struct iovec *data);
 sctp_chunk_t *sctp_chunkify(struct sk_buff *, const sctp_association_t *,
 			    struct sock *);
-void sctp_init_source(sctp_chunk_t *chunk);
+void sctp_init_addrs(sctp_chunk_t *chunk);
 const sockaddr_storage_t *sctp_source(const sctp_chunk_t *chunk);
 
 /* This is a structure for holding either an IPv6 or an IPv4 address.  */
@@ -655,6 +661,9 @@ struct SCTP_transport {
 	/* PMTU       : The current known path MTU.  */
 	__u32 pmtu;
 
+	/* Destination */
+	struct dst_entry *dst;
+
 	/* When was the last time(in jiffies) that a data packet was sent on
 	 * this transport?  This is used to adjust the cwnd when the transport
 	 * becomes inactive.
@@ -676,13 +685,15 @@ struct SCTP_transport {
 	 */
 	unsigned long last_time_ecne_reduced;
 
-	/* state       : The current state of this destination,
-	 *             :  i.e. DOWN, UP, ALLOW-HB, NO-HEARTBEAT, etc.
+	/* active      : The current active state of this destination,
+	 *             :  i.e. DOWN, UP, etc.
 	 */
-	struct {
-		int active;
-		int hb_allowed;
-	} state;
+	int active;
+
+	/* hb_allowed  : The current heartbeat state of this destination,
+	 *             :  i.e. ALLOW-HB, NO-HEARTBEAT, etc.
+	 */
+	int hb_allowed;
 
 	/* These are the error stats for this destination.  */
 
@@ -735,6 +746,7 @@ extern sctp_transport_t *sctp_transport_new(const sockaddr_storage_t *, int);
 extern sctp_transport_t *sctp_transport_init(sctp_transport_t *,
 					     const sockaddr_storage_t *, int);
 extern void sctp_transport_set_owner(sctp_transport_t *, sctp_association_t *);
+extern void sctp_transport_route(sctp_transport_t *, sockaddr_storage_t *);
 extern void sctp_transport_free(sctp_transport_t *);
 extern void sctp_transport_destroy(sctp_transport_t *);
 extern void sctp_transport_reset_timers(sctp_transport_t *);
@@ -885,9 +897,9 @@ int sctp_add_bind_addr(sctp_bind_addr_t *, sockaddr_storage_t *,
 		       int priority);
 int sctp_del_bind_addr(sctp_bind_addr_t *, sockaddr_storage_t *);
 int sctp_bind_addr_has_addr(sctp_bind_addr_t *, const sockaddr_storage_t *);
-sctpParam_t sctp_bind_addrs_to_raw(const sctp_bind_addr_t *bp,
-				   int *addrs_len,
-				   int priority);
+union sctp_params sctp_bind_addrs_to_raw(const sctp_bind_addr_t *bp,
+					 int *addrs_len,
+					 int priority);
 int sctp_raw_to_bind_addrs(sctp_bind_addr_t *bp,
 			   __u8 *raw_addr_list,
 			   int addrs_len,
@@ -904,7 +916,7 @@ int sctp_addr_is_valid(const sockaddr_storage_t *addr);
 typedef enum {
 	SCTP_EP_TYPE_SOCKET,
 	SCTP_EP_TYPE_ASSOCIATION,
-} sctp_endpoint_type_t; 
+} sctp_endpoint_type_t;
 
 /*
  * A common base class to bridge the implmentation view of a
@@ -1052,22 +1064,11 @@ int sctp_verify_init(const sctp_association_t *asoc,
 		     sctp_init_chunk_t *peer_init,
 		     sctp_chunk_t *chunk,
 		     sctp_chunk_t **err_chunk);
-int sctp_verify_param(const sctp_association_t *asoc,
-		      sctpParam_t param,
-		      sctp_cid_t cid,
-		      sctp_chunk_t *chunk,
-		      sctp_chunk_t **err_chunk);
-int sctp_process_unk_param(const sctp_association_t *asoc,
-			   sctpParam_t param,
-			   sctp_chunk_t *chunk,
-			   sctp_chunk_t **err_chunk);
-void sctp_process_init(sctp_association_t *asoc, sctp_cid_t cid,
-		       const sockaddr_storage_t *peer_addr,
-		       sctp_init_chunk_t  *peer_init, int priority);
-int sctp_process_param(sctp_association_t *asoc,
-		       sctpParam_t param,
-		       const sockaddr_storage_t *peer_addr,
-		       sctp_cid_t cid, int priority);
+int sctp_process_init(sctp_association_t *asoc, sctp_cid_t cid,
+		      const sockaddr_storage_t *peer_addr,
+		      sctp_init_chunk_t *peer_init, int priority);
+int sctp_process_param(sctp_association_t *asoc, union sctp_params param,
+		       const sockaddr_storage_t *peer_addr, int priority);
 __u32 sctp_generate_tag(const sctp_endpoint_t *ep);
 __u32 sctp_generate_tsn(const sctp_endpoint_t *ep);
 
@@ -1154,10 +1155,10 @@ struct SCTP_association {
 		sctp_transport_t *primary_path;
 
 		/* Cache the primary path address here, when we
-		 * need a an address for msg_name. 
+		 * need a an address for msg_name.
 		 */
 		sockaddr_storage_t primary_addr;
-		
+
 		/* active_path
 		 *   The path that we are currently using to
 		 *   transmit new data and most control chunks.
@@ -1258,7 +1259,7 @@ struct SCTP_association {
 
 	/* Overall     : The threshold for this association that if
 	 * Error       : the Overall Error Count reaches will cause
-	 * Threshold   : this association to be torn down. 
+	 * Threshold   : this association to be torn down.
 	 */
 	int overall_error_threshold;
 
@@ -1304,13 +1305,13 @@ struct SCTP_association {
 	 */
 	__u32 next_tsn;
 
-	/* 
+	/*
 	 * Last Rcvd   : This is the last TSN received in sequence.  This value
 	 * TSN         : is set initially by taking the peer's Initial TSN,
 	 *             : received in the INIT or INIT ACK chunk, and
 	 *             : subtracting one from it.
 	 *
-	 * Most of RFC 2960 refers to this as the Cumulative TSN Ack Point. 
+	 * Most of RFC 2960 refers to this as the Cumulative TSN Ack Point.
 	 */
 
 	__u32 ctsn_ack_point;

include/net/snmp.h

@@ -189,6 +189,35 @@ struct udp_mib
 	unsigned long   __pad[0];
 } ____cacheline_aligned; 
 
+/* draft-ietf-sigtran-sctp-mib-07.txt */
+struct sctp_mib
+{
+	unsigned long   SctpCurrEstab;
+	unsigned long   SctpActiveEstabs;
+	unsigned long   SctpPassiveEstabs;
+	unsigned long   SctpAborteds;
+	unsigned long   SctpShutdowns;
+	unsigned long   SctpOutOfBlues;
+	unsigned long   SctpChecksumErrors;
+	unsigned long   SctpOutCtrlChunks;
+	unsigned long   SctpOutOrderChunks;
+	unsigned long   SctpOutUnorderChunks;
+	unsigned long   SctpInCtrlChunks;
+	unsigned long   SctpInOrderChunks;
+	unsigned long   SctpInUnorderChunks;
+	unsigned long   SctpFragUsrMsgs;
+	unsigned long   SctpReasmUsrMsgs;
+	unsigned long   SctpOutSCTPPacks;
+	unsigned long   SctpInSCTPPacks;
+	unsigned long   SctpRtoAlgorithm;
+	unsigned long   SctpRtoMin;
+	unsigned long   SctpRtoMax;
+	unsigned long   SctpRtoInitial;
+	unsigned long   SctpValCookieLife;
+	unsigned long   SctpMaxInitRetr;
+	unsigned long   __pad[0];
+} ____cacheline_aligned;
+
 struct linux_mib 
 {
 	unsigned long	SyncookiesSent;

net/sctp/associola.c

@@ -208,7 +208,7 @@ sctp_association_t *sctp_association_init(sctp_association_t *asoc,
 
 	asoc->ctsn_ack_point = asoc->next_tsn - 1;
 	asoc->highest_sacked = asoc->ctsn_ack_point;
-
+	asoc->last_cwr_tsn = asoc->ctsn_ack_point;
 	asoc->unack_data = 0;
 
 	SCTP_DEBUG_PRINTK("myctsnap for %s INIT as 0x%x.\n",
@@ -404,6 +404,9 @@ sctp_transport_t *sctp_assoc_add_peer(sctp_association_t *asoc,
 
 	sctp_transport_set_owner(peer, asoc);
 
+	/* Cache a route for the transport. */
+	sctp_transport_route(peer, NULL);
+
 	/* If this is the first transport addr on this association,
 	 * initialize the association PMTU to the peer's PMTU.
 	 * If not and the current association PMTU is higher than the new
@@ -458,6 +461,9 @@ sctp_transport_t *sctp_assoc_add_peer(sctp_association_t *asoc,
 		min(asoc->overall_error_threshold + peer->error_threshold,
 		    asoc->max_retrans);
 
+	/* By default, enable heartbeat for peer address. */
+	peer->hb_allowed = 1;
+
 	/* Initialize the peer's heartbeat interval based on the
 	 * sock configured value.
 	 */
@@ -520,12 +526,12 @@ void sctp_assoc_control_transport(sctp_association_t *asoc,
 	/* Record the transition on the transport.  */
 	switch (command) {
 	case SCTP_TRANSPORT_UP:
-		transport->state.active = 1;
+		transport->active = 1;
 		spc_state = ADDRESS_AVAILABLE;
 		break;
 
 	case SCTP_TRANSPORT_DOWN:
-		transport->state.active = 0;
+		transport->active = 0;
 		spc_state = ADDRESS_UNREACHABLE;
 		break;
 
@@ -555,7 +561,7 @@ void sctp_assoc_control_transport(sctp_association_t *asoc,
 	list_for_each(pos, &asoc->peer.transport_addr_list) {
 		t = list_entry(pos, sctp_transport_t, transports);
 
-		if (!t->state.active)
+		if (!t->active)
 			continue;
 		if (!first || t->last_time_heard > first->last_time_heard) {
 			second = first;
@@ -575,7 +581,7 @@ void sctp_assoc_control_transport(sctp_association_t *asoc,
 	 * [If the primary is active but not most recent, bump the most
 	 * recently used transport.]
 	 */
-	if (asoc->peer.primary_path->state.active &&
+	if (asoc->peer.primary_path->active &&
 	    first != asoc->peer.primary_path) {
 		second = first;
 		first = asoc->peer.primary_path;
@@ -1015,7 +1021,7 @@ sctp_transport_t *sctp_assoc_choose_shutdown_transport(sctp_association_t *asoc)
 
 		/* Try to find an active transport. */
 
-		if (t->state.active) {
+		if (t->active) {
 			break;
 		} else {
 			/* Keep track of the next transport in case

net/sctp/bind_addr.c

@@ -196,18 +196,16 @@ int sctp_del_bind_addr(sctp_bind_addr_t *bp, sockaddr_storage_t *del_addr)
  *
  * The second argument is the return value for the length.
  */
-sctpParam_t sctp_bind_addrs_to_raw(const sctp_bind_addr_t *bp, int *addrs_len,
-				   int priority)
+union sctp_params sctp_bind_addrs_to_raw(const sctp_bind_addr_t *bp, 
+					 int *addrs_len, int priority)
 {
-	sctpParam_t addrparms;
-	sctpParam_t retval;
+	union sctp_params addrparms;
+	union sctp_params retval;
 	int addrparms_len;
 	sctp_addr_param_t rawaddr;
 	int len;
 	struct sockaddr_storage_list *addr;
 	struct list_head *pos;
-
-	retval.v = NULL;
 	addrparms_len = 0;
 	len = 0;
 
@@ -216,11 +214,11 @@ sctpParam_t sctp_bind_addrs_to_raw(const sctp_bind_addr_t *bp, int *addrs_len,
 		len += sizeof(sctp_addr_param_t);
 	}
 
-	addrparms.v = kmalloc(len, priority);
-	if (!addrparms.v)
+	retval.v = kmalloc(len, priority);
+	if (!retval.v)
 		goto end_raw;
 
-	retval = addrparms;
+	addrparms = retval;
 
 	list_for_each(pos, &bp->address_list) {
 		addr = list_entry(pos, struct sockaddr_storage_list, list);

net/sctp/input.c

@@ -41,6 +41,7 @@
  *    Jon Grimm <jgrimm@us.ibm.com>
  *    Hui Huang <hui.huang@nokia.com>
  *    Daisy Chang <daisyc@us.ibm.com>
+ *    Sridhar Samudrala <sri@us.ibm.com>
  *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
@@ -217,8 +218,8 @@ int sctp_rcv(struct sk_buff *skb)
 	/* Remember the SCTP header. */
 	chunk->sctp_hdr = sh;
 
-	/* Set the source address.  */
-	sctp_init_source(chunk);
+	/* Set the source and destination addresses of the incoming chunk.  */
+	sctp_init_addrs(chunk);
 
 	/* Remember where we came from.  */
 	chunk->transport = transport;
@@ -567,7 +568,7 @@ sctp_association_t *sctp_lookup_association(const sockaddr_storage_t *laddr,
 	sctp_local_bh_disable();
 	asoc = __sctp_lookup_association(laddr, paddr, transportp);
 	sctp_local_bh_enable();
-	
+
 	return asoc;
 }
 
@@ -613,13 +614,11 @@ static sctp_association_t *__sctp_rcv_initack_lookup(struct sk_buff *skb,
 	sockaddr_storage_t *paddr = &addr;
 	struct sctphdr *sh = (struct sctphdr *) skb->h.raw;
 	sctp_chunkhdr_t *ch;
-	__u8 *ch_end, *data;
-	sctp_paramhdr_t *parm;
+	union sctp_params params;
+	sctp_init_chunk_t *init;
 
 	ch = (sctp_chunkhdr_t *) skb->data;
 
-	ch_end = ((__u8 *) ch) + WORD_ROUND(ntohs(ch->length));
-
 	/* If this is INIT/INIT-ACK look inside the chunk too. */
 	switch (ch->type) {
 	case SCTP_CID_INIT:
@@ -645,24 +644,17 @@ static sctp_association_t *__sctp_rcv_initack_lookup(struct sk_buff *skb,
 	/* Find the start of the TLVs and the end of the chunk.  This is
 	 * the region we search for address parameters.
 	 */
-	data = skb->data + sizeof(sctp_init_chunk_t);
-
-	/* See sctp_process_init() for how to go thru TLVs. */
-	while (data < ch_end) {
-		parm = (sctp_paramhdr_t *)data;
-
-		if (!parm->length)
-			break;
+	init = (sctp_init_chunk_t *)skb->data;
 
-		data += WORD_ROUND(ntohs(parm->length));
+	/* Walk the parameters looking for embedded addresses. */
+	sctp_walk_params(params, init, init_hdr.params) {
 
 		/* Note: Ignoring hostname addresses. */
-		if ((SCTP_PARAM_IPV4_ADDRESS != parm->type) &&
-		    (SCTP_PARAM_IPV6_ADDRESS != parm->type))
+		if ((SCTP_PARAM_IPV4_ADDRESS != params.p->type) &&
+		    (SCTP_PARAM_IPV6_ADDRESS != params.p->type))
 			continue;
 
-		sctp_param2sockaddr(paddr, (sctp_addr_param_t *)parm, 
-				    ntohs(sh->source));
+		sctp_param2sockaddr(paddr, params.addr, ntohs(sh->source));
 		asoc = __sctp_lookup_association(laddr, paddr, transportp);
 		if (asoc)
 			return asoc;

net/sctp/ipv6.c

@@ -158,35 +158,60 @@ static inline int sctp_v6_xmit(struct sk_buff *skb)
 }
 #endif /* TEST_FRAME */
 
-/* Returns the mtu for the given v6 destination address. */
-int sctp_v6_get_dst_mtu(const sockaddr_storage_t *address)
+/* 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.
+ */
+struct dst_entry *sctp_v6_get_dst(sockaddr_storage_t *daddr,
+				  sockaddr_storage_t *saddr)
 {
 	struct dst_entry *dst;
-	struct flowi fl;
-	int dst_mtu = SCTP_DEFAULT_MAXSEGMENT;
+	struct flowi fl = { .nl_u = { .ip6_u = { .daddr = &daddr->v6.sin6_addr,
+					       } } };
 
-	fl.proto = 0;
-	fl.fl6_dst = (struct in6_addr *)&address->v6.sin6_addr;
-	fl.fl6_src = NULL;
-	fl.fl6_flowlabel = 0;
-	fl.oif = 0;
-	fl.uli_u.ports.sport = 0;
-	fl.uli_u.ports.dport = 0;
+
+	SCTP_DEBUG_PRINTK("%s: DST=%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x ",
+			  __FUNCTION__, NIP6(fl.fl6_dst)); 
+
+	if (saddr) {
+		fl.fl6_src = &saddr->v6.sin6_addr;
+		SCTP_DEBUG_PRINTK(
+			"SRC=%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x - ",
+			NIP6(fl.fl6_src));
+	}
 
 	dst = ip6_route_output(NULL, &fl);
 	if (dst) {
-		dst_mtu = dst_pmtu(dst);
-		SCTP_DEBUG_PRINTK("sctp_v6_get_dst_mtu: "
-				  "ip6_route_output: dev:%s pmtu:%d\n",
-				  dst->dev->name, dst_mtu);
-		dst_release(dst);
+		struct rt6_info *rt;
+		rt = (struct rt6_info *)dst;
+		SCTP_DEBUG_PRINTK(
+			"rt6_dst:%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x "
+			"rt6_src:%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x\n",
+			NIP6(&rt->rt6i_dst.addr), NIP6(&rt->rt6i_src.addr));
 	} else {
-		SCTP_DEBUG_PRINTK("sctp_v6_get_dst_mtu: "
-				  "ip6_route_output failed, returning "
-				  "%d as dst_mtu\n", dst_mtu);
+		SCTP_DEBUG_PRINTK("NO ROUTE\n");
+		return NULL;
 	}
 
-	return dst_mtu;
+	return dst;
+}
+
+/* Check if the dst entry's source addr matches the given source addr. */
+int sctp_v6_cmp_saddr(struct dst_entry *dst, sockaddr_storage_t *saddr)
+{
+	struct rt6_info *rt = (struct rt6_info *)dst;
+
+	return ipv6_addr_cmp(&rt->rt6i_src.addr, &saddr->v6.sin6_addr); 
 }
 
 /* Initialize a PF_INET6 socket msg_name. */
@@ -301,7 +326,8 @@ static sctp_func_t sctp_ipv6_specific = {
 	.queue_xmit      = sctp_v6_xmit,
 	.setsockopt      = ipv6_setsockopt,
 	.getsockopt      = ipv6_getsockopt,
-	.get_dst_mtu     = sctp_v6_get_dst_mtu,
+	.get_dst	 = sctp_v6_get_dst,
+	.cmp_saddr	 = sctp_v6_cmp_saddr,
 	.net_header_len  = sizeof(struct ipv6hdr),
 	.sockaddr_len    = sizeof(struct sockaddr_in6),
 	.sa_family       = AF_INET6,

net/sctp/output.c

@@ -245,6 +245,7 @@ int sctp_packet_transmit(sctp_packet_t *packet)
 	int err = 0;
 	int padding;		/* How much padding do we need?  */
 	__u8 packet_has_data = 0;
+	struct dst_entry *dst;
 
 	/* Do NOT generate a chunkless packet... */
 	if (skb_queue_empty(&packet->chunks))
@@ -410,13 +411,6 @@ int sctp_packet_transmit(sctp_packet_t *packet)
 		asoc->peer.last_sent_to = transport;
 	}
 
-	/* Hey, before Linux changes, here's what we have to
-	 * do to force IP routing to recognize the change of
-	 * dest addr.				--xguo
-	 */
-	if (sk->dst_cache)
-		sk->dst_cache->obsolete = 1;
-
 	if (packet_has_data) {
 		struct timer_list *timer;
 		unsigned long timeout;
@@ -434,6 +428,13 @@ int sctp_packet_transmit(sctp_packet_t *packet)
 		}
 	}
 
+	dst = transport->dst;
+	if (!dst || dst->obsolete) {
+		sctp_transport_route(transport, NULL);
+	}
+
+	nskb->dst = dst_clone(transport->dst);
+
 	SCTP_DEBUG_PRINTK("***sctp_transmit_packet*** skb length %d\n",
 			  nskb->len);
 	(*transport->af_specific->queue_xmit)(nskb);

net/sctp/outqueue.c

@@ -204,18 +204,43 @@ int sctp_push_outqueue(sctp_outqueue_t *q, sctp_chunk_t *chunk)
 	return error;
 }
 
+/* Insert a chunk into the retransmit queue.  Chunks on the retransmit
+ * queue are kept in order, based on the TSNs.
+ */
+void sctp_retransmit_insert(struct list_head *tlchunk, sctp_outqueue_t *q)
+{
+	struct list_head *rlchunk;
+	sctp_chunk_t *tchunk, *rchunk;
+	__u32 ttsn, rtsn;
+	int done = 0;
+
+	tchunk = list_entry(tlchunk, sctp_chunk_t, transmitted_list);
+	ttsn = ntohl(tchunk->subh.data_hdr->tsn);
+
+	list_for_each(rlchunk, &q->retransmit) {
+		rchunk = list_entry(rlchunk, sctp_chunk_t, transmitted_list);
+		rtsn = ntohl(rchunk->subh.data_hdr->tsn);
+		if (TSN_lt(ttsn, rtsn)) {
+			list_add(tlchunk, rlchunk->prev);
+			done = 1;
+			break;
+		}
+	}
+	if (!done) {
+		list_add_tail(tlchunk, &q->retransmit);
+	}
+}
+			
 /* Mark all the eligible packets on a transport for retransmission.  */
 void sctp_retransmit_mark(sctp_outqueue_t *q, sctp_transport_t *transport,
 			  __u8 fast_retransmit)
 {
-	struct list_head *lchunk;
+	struct list_head *lchunk, *ltemp;
 	sctp_chunk_t *chunk;
-	struct list_head tlist;
 
-	INIT_LIST_HEAD(&tlist);
 
-	while (!list_empty(&transport->transmitted)) {
-		lchunk = sctp_list_dequeue(&transport->transmitted);
+	/* Walk through the specified transmitted queue.  */
+	list_for_each_safe(lchunk, ltemp, &transport->transmitted) {
 		chunk = list_entry(lchunk, sctp_chunk_t, transmitted_list);
 
 		/* If we are doing retransmission due to a fast retransmit,
@@ -224,10 +249,8 @@ void sctp_retransmit_mark(sctp_outqueue_t *q, sctp_transport_t *transport,
 		 * retransmission due to a timeout, only the chunks that are
 		 * not yet acked should be added to the retransmit queue.
 		 */
-		if ((fast_retransmit && !chunk->fast_retransmit) ||
-		    (!fast_retransmit && chunk->tsn_gap_acked)) {
-			list_add_tail(lchunk, &tlist);
-		} else {
+		if ((fast_retransmit && chunk->fast_retransmit) ||
+		   (!fast_retransmit && !chunk->tsn_gap_acked)) {
 			/* RFC 2960 6.2.1 Processing a Received SACK
 			 *
 			 * C) Any time a DATA chunk is marked for
@@ -257,16 +280,15 @@ void sctp_retransmit_mark(sctp_outqueue_t *q, sctp_transport_t *transport,
 				chunk->rtt_in_progress = 0;
 				transport->rto_pending = 0;
 			}
-			list_add_tail(lchunk, &q->retransmit);
+
+			/* Move the chunk to the retransmit queue. The chunks
+			 * on the retransmit queue is always kept in order.
+			 */
+			list_del(lchunk);
+			sctp_retransmit_insert(lchunk, q);
 		}
 	}
 
-	/* Reconstruct the transmitted queue with chunks that are not
-	 * eligible for retransmission.
-	 */
-	while (NULL != (lchunk = sctp_list_dequeue(&tlist)))
-		list_add_tail(lchunk, &transport->transmitted);
-
 	SCTP_DEBUG_PRINTK("%s: transport: %p, fast_retransmit: %d, "
 			  "cwnd: %d, ssthresh: %d, flight_size: %d, "
 			  "pba: %d\n", __FUNCTION__,
@@ -656,7 +678,7 @@ int sctp_flush_outqueue(sctp_outqueue_t *q, int rtx_timeout)
 
 		if (!new_transport) {
 			new_transport = asoc->peer.active_path;
-		} else if (!new_transport->state.active) {
+		} else if (!new_transport->active) {
 			/* If the chunk is Heartbeat, send it to
 			 * chunk->transport, even it's inactive.
 			 */
@@ -813,7 +835,7 @@ int sctp_flush_outqueue(sctp_outqueue_t *q, int rtx_timeout)
 			 */
 			new_transport = chunk->transport;
 			if (new_transport == NULL ||
-			    !new_transport->state.active)
+			    !new_transport->active)
 				new_transport = asoc->peer.active_path;
 
 			/* Change packets if necessary.  */
@@ -1382,7 +1404,7 @@ static void sctp_check_transmitted(sctp_outqueue_t *q,
 			/* Mark the destination transport address as
 			 * active if it is not so marked.
 			 */
-			if (!transport->state.active) {
+			if (!transport->active) {
 				sctp_assoc_control_transport(
 					transport->asoc,
 					transport,

net/sctp/primitive.c

@@ -38,6 +38,7 @@
  *    La Monte H.P. Yarroll <piggy@acm.org>
  *    Narasimha Budihal     <narasimha@refcode.org>
  *    Karl Knutson          <karl@athena.chicago.il.us>
+ *    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.
@@ -181,6 +182,28 @@ DECLARE_PRIMITIVE(ABORT);
 
 DECLARE_PRIMITIVE(SEND);
 
+/* 10.1 ULP-to-SCTP
+ * J) Request Heartbeat
+ *
+ * Format: REQUESTHEARTBEAT(association id, destination transport address)
+ *
+ * -> result
+ *
+ * Instructs the local endpoint to perform a HeartBeat on the specified
+ * destination transport address of the given association. The returned
+ * result should indicate whether the transmission of the HEARTBEAT
+ * chunk to the destination address is successful.
+ *
+ * Mandatory attributes:
+ *
+ * o association id - local handle to the SCTP association
+ *
+ * o destination transport address - the transport address of the
+ *   asociation on which a heartbeat should be issued.
+ */
+
+DECLARE_PRIMITIVE(REQUESTHEARTBEAT);
+
 /* COMMENT BUG.  Find out where this is mentioned in the spec.  */
 int sctp_other_icmp_unreachfrag(sctp_association_t *asoc, void *arg)
 {

net/sctp/protocol.c

@@ -59,6 +59,7 @@
 /* Global data structures. */
 sctp_protocol_t sctp_proto;
 struct proc_dir_entry	*proc_net_sctp;
+struct sctp_mib sctp_statistics[NR_CPUS * 2];
 
 /* This is the global socket data structure used for responding to
  * the Out-of-the-blue (OOTB) packets.  A control sock will be created
@@ -255,27 +256,41 @@ int sctp_copy_local_addr_list(sctp_protocol_t *proto, sctp_bind_addr_t *bp,
 	return error;
 }
 
-/* Returns the mtu for the given v4 destination address.  */
-int sctp_v4_get_dst_mtu(const sockaddr_storage_t *address)
+/* Returns the dst cache entry for the given source and destination ip
+ * addresses.
+ */
+struct dst_entry *sctp_v4_get_dst(sockaddr_storage_t *daddr,
+				  sockaddr_storage_t *saddr)
 {
-	int dst_mtu = SCTP_DEFAULT_MAXSEGMENT;
 	struct rtable *rt;
-	struct flowi fl = { .nl_u = { .ip4_u =
-				      { .daddr = address->v4.sin_addr.s_addr } } };
+	struct flowi fl = { .nl_u = { .ip4_u = { .daddr =
+						    daddr->v4.sin_addr.s_addr,
+					       } } };
+
+	if (saddr)
+		fl.fl4_src = saddr->v4.sin_addr.s_addr;
+
+	SCTP_DEBUG_PRINTK("%s: DST:%u.%u.%u.%u, SRC:%u.%u.%u.%u - ",
+			  __FUNCTION__, NIPQUAD(fl.fl4_dst),
+			  NIPQUAD(fl.fl4_src));
 
 	if (ip_route_output_key(&rt, &fl)) {
-		SCTP_DEBUG_PRINTK("sctp_v4_get_dst_mtu:ip_route_output_key"
-				  " failed, returning %d as dst_mtu\n",
-				  dst_mtu);
-	} else {
-		dst_mtu = dst_pmtu(&rt->u.dst);
-		SCTP_DEBUG_PRINTK("sctp_v4_get_dst_mtu: "
-				  "ip_route_output_key: dev:%s pmtu:%d\n",
-				  rt->u.dst.dev->name, dst_mtu);
-		ip_rt_put(rt);
+		SCTP_DEBUG_PRINTK("NO ROUTE\n");
+		return NULL;
 	}
 
-	return dst_mtu;
+	SCTP_DEBUG_PRINTK("rt_dst:%u.%u.%u.%u, rt_src:%u.%u.%u.%u\n",
+			  NIPQUAD(rt->rt_src), NIPQUAD(rt->rt_dst));
+
+	return &rt->u.dst;
+}
+
+/* Check if the dst entry's source addr matches the given source addr. */
+int sctp_v4_cmp_saddr(struct dst_entry *dst, sockaddr_storage_t *saddr)
+{
+	struct rtable *rt = (struct rtable *)dst;
+
+	return (rt->rt_src == saddr->v4.sin_addr.s_addr); 
 }
 
 /* Event handler for inet device events.
@@ -437,7 +452,8 @@ sctp_func_t sctp_ipv4_specific = {
 	.queue_xmit     = ip_queue_xmit,
 	.setsockopt     = ip_setsockopt,
 	.getsockopt     = ip_getsockopt,
-	.get_dst_mtu    = sctp_v4_get_dst_mtu,
+	.get_dst	= sctp_v4_get_dst,
+	.cmp_saddr	= sctp_v4_cmp_saddr,
 	.net_header_len = sizeof(struct iphdr),
 	.sockaddr_len   = sizeof(struct sockaddr_in),
 	.sa_family      = AF_INET,

net/sctp/sm_sideeffect.c

@@ -41,6 +41,7 @@
  *    Dajiang Zhang	    <dajiang.zhang@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.
@@ -66,12 +67,15 @@ static void sctp_do_ecn_cwr_work(sctp_association_t *asoc,
 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);
+static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *, sctp_association_t *asoc,
+				  sctp_event_t event_type, sctp_chunk_t *chunk);
 static void sctp_cmd_process_init(sctp_cmd_seq_t *, sctp_association_t *asoc,
 				  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_update(sctp_cmd_seq_t *, sctp_association_t *,
+				      sctp_transport_t *);
 static void sctp_cmd_set_bind_addrs(sctp_cmd_seq_t *, sctp_association_t *,
 				    sctp_bind_addr_t *);
 static void sctp_cmd_transport_reset(sctp_cmd_seq_t *, sctp_association_t *,
@@ -192,6 +196,7 @@ int sctp_side_effects(sctp_event_t event_type, sctp_subtype_t subtype,
 		/* BUG--we should now recover some memory, probably by
 		 * reneging...
 		 */
+		error = -ENOMEM;
 		break;
 
         case SCTP_DISPOSITION_DELETE_TCB:
@@ -251,7 +256,7 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 	int force;
 	sctp_cmd_t *command;
 	sctp_chunk_t *new_obj;
-	sctp_chunk_t *chunk;
+	sctp_chunk_t *chunk = NULL;
 	sctp_packet_t *packet;
 	struct list_head *pos;
 	struct timer_list *timer;
@@ -259,7 +264,8 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 	sctp_transport_t *t;
 	sctp_sackhdr_t sackh;
 
-	chunk = (sctp_chunk_t *) event_arg;
+	if(SCTP_EVENT_T_TIMEOUT != event_type)
+		chunk = (sctp_chunk_t *) event_arg;
 
 	/* Note:  This whole file is a huge candidate for rework.
 	 * For example, each command could either have its own handler, so
@@ -504,7 +510,8 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 			break;
 
 		case SCTP_CMD_ASSOC_FAILED:
-			sctp_cmd_assoc_failed(commands, asoc);
+			sctp_cmd_assoc_failed(commands, asoc, event_type,
+					      chunk);
 			break;
 
 		case SCTP_CMD_COUNTER_INC:
@@ -557,6 +564,11 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 		case SCTP_CMD_HB_TIMERS_START:
 			sctp_cmd_hb_timers_start(commands, asoc);
 			break;
+	
+		case SCTP_CMD_HB_TIMERS_UPDATE:
+			t = command->obj.transport;
+			sctp_cmd_hb_timers_update(commands, asoc, t);
+			break;
 
 		case SCTP_CMD_REPORT_ERROR:
 			error = command->obj.error;
@@ -595,10 +607,7 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 /* 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
-	 * Note: dp->TSN is expected in host endian
-	 */
+	/* Save the TSN away for comparison when we receive CWR */
 
 	asoc->last_ecne_tsn = lowest_tsn;
 	asoc->need_ecne = 1;
@@ -621,7 +630,6 @@ static sctp_chunk_t *sctp_do_ecn_ecne_work(sctp_association_t *asoc,
 					   sctp_chunk_t *chunk)
 {
 	sctp_chunk_t *repl;
-	sctp_transport_t *transport;
 
 	/* Our previously transmitted packet ran into some congestion
 	 * so we should take action by reducing cwnd and ssthresh
@@ -629,43 +637,28 @@ static sctp_chunk_t *sctp_do_ecn_ecne_work(sctp_association_t *asoc,
 	 * sending a CWR.
 	 */
 
-	/* Find which transport's congestion variables
-	 * need to be adjusted.
+	/* 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.
 	 */
+	if (TSN_lt(asoc->last_cwr_tsn, lowest_tsn)) {
+		sctp_transport_t *transport;
 
-	transport = sctp_assoc_lookup_tsn(asoc, lowest_tsn);
+		/* 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);
+		/* Update the congestion variables. */
+		if (transport)
+			sctp_transport_lower_cwnd(transport,
+						  SCTP_LOWER_CWND_ECNE);
+		asoc->last_cwr_tsn = lowest_tsn;
+	}
 
-	/* Save away a rough idea of when we last sent out a CWR.
-	 * We compare against this value (see above) to decide if
-	 * this is a fairly new request.
-	 * Note that this is not a perfect solution.  We may
-	 * have moved beyond the window (several times) by the
-	 * next time we get an ECNE.  However, it is cute.  This idea
-	 * came from Randy's reference code.
-	 *
-	 * Here's what RFC 2960 has to say about CWR.  This is NOT
-	 * what we do.
-	 *
-	 * RFC 2960 Appendix A
-	 *
-	 *    CWR:
-	 *
-	 *    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.
+	/* Always try to quiet the other end.  In case of lost CWR,
+	 * resend last_cwr_tsn.  
 	 */
-	asoc->last_cwr_tsn = asoc->next_tsn - 1;
-
 	repl = sctp_make_cwr(asoc, asoc->last_cwr_tsn, chunk);
 
 	/* If we run out of memory, it will look like a lost CWR.  We'll
@@ -994,7 +987,7 @@ static void sctp_do_8_2_transport_strike(sctp_association_t *asoc,
 	 */
 	asoc->overall_error_count++;
 
-	if (transport->state.active &&
+	if (transport->active &&
 	    (transport->error_count++ >= transport->error_threshold)) {
 		SCTP_DEBUG_PRINTK("transport_strike: transport "
 				  "IP:%d.%d.%d.%d failed.\n",
@@ -1038,14 +1031,26 @@ static void sctp_cmd_init_failed(sctp_cmd_seq_t *commands,
 
 /* Worker routine to handle SCTP_CMD_ASSOC_FAILED.  */
 static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *commands,
-				  sctp_association_t *asoc)
+				  sctp_association_t *asoc,
+				  sctp_event_t event_type,
+				  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;
+	}
 
 	event = sctp_ulpevent_make_assoc_change(asoc,
 						0,
 						SCTP_COMM_LOST,
-						0, 0, 0,
+						error, 0, 0,
 						GFP_ATOMIC);
 
 	if (event)
@@ -1100,6 +1105,16 @@ static void sctp_cmd_hb_timers_start(sctp_cmd_seq_t *cmds,
 	}
 }
 
+/* Helper function to update the heartbeat timer. */
+static void sctp_cmd_hb_timers_update(sctp_cmd_seq_t *cmds,
+				   sctp_association_t *asoc,
+				   sctp_transport_t *t)
+{
+	/* Update the heartbeat timer.  */
+	if (!mod_timer(&t->hb_timer, t->hb_interval + t->rto + jiffies))
+		sctp_transport_hold(t);
+}
+
 /* Helper function to break out SCTP_CMD_SET_BIND_ADDR handling.  */
 void sctp_cmd_set_bind_addrs(sctp_cmd_seq_t *cmds, sctp_association_t *asoc,
 			     sctp_bind_addr_t *bp)
@@ -1135,7 +1150,7 @@ static void sctp_cmd_transport_on(sctp_cmd_seq_t *cmds,
 	/* Mark the destination transport address as active if it is not so
 	 * marked.
 	 */
-	if (!t->state.active)
+	if (!t->active)
 		sctp_assoc_control_transport(asoc, t, SCTP_TRANSPORT_UP,
 					     SCTP_HEARTBEAT_SUCCESS);
 
@@ -1158,10 +1173,6 @@ static void sctp_cmd_transport_reset(sctp_cmd_seq_t *cmds,
 
 	/* Mark one strike against a transport.  */
 	sctp_do_8_2_transport_strike(asoc, t);
-
-	/* Update the heartbeat timer.  */
-	if (!mod_timer(&t->hb_timer, t->hb_interval + t->rto + jiffies))
-		sctp_transport_hold(t);
 }
 
 /* Helper function to process the process SACK command.  */

net/sctp/sm_statetable.c

@@ -39,6 +39,7 @@
  *    Jon Grimm             <jgrimm@us.ibm.com>
  *    Hui Huang		    <hui.huang@nokia.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.
@@ -706,21 +707,28 @@ chunk_event_table_unknown[SCTP_STATE_NUM_STATES] = {
 	/* SCTP_STATE_EMPTY */ \
 	{.fn = sctp_sf_bug, .name = "sctp_sf_bug"}, \
 	/* SCTP_STATE_CLOSED */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_bug, .name = "sctp_sf_bug"}, \
 	/* SCTP_STATE_COOKIE_WAIT */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 	/* SCTP_STATE_COOKIE_ECHOED */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 	/* SCTP_STATE_ESTABLISHED */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 	/* SCTP_STATE_SHUTDOWN_PENDING */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 	/* SCTP_STATE_SHUTDOWN_SENT */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 	/* SCTP_STATE_SHUTDOWN_RECEIVED */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 	/* SCTP_STATE_SHUTDOWN_ACK_SENT */ \
-	{.fn = sctp_sf_not_impl, .name = "sctp_sf_not_impl"}, \
+	{.fn = sctp_sf_do_prm_requestheartbeat,		      \
+	 .name = "sctp_sf_do_prm_requestheartbeat"},          \
 } /* TYPE_SCTP_PRIMITIVE_REQUESTHEARTBEAT */
 
 #define TYPE_SCTP_PRIMITIVE_GETSRTTREPORT  { \

net/sctp/socket.c

@@ -804,10 +804,12 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 	SCTP_DEBUG_PRINTK("msg_len: %Zd, sinfo_flags: 0x%x\n",
 			  msg_len, sinfo_flags);
 
-	/* If MSG_EOF|MSG_ABORT is set, no data can be sent.  Disallow
-	 * sending 0-length messages when MSG_EOF|MSG_ABORT is not set.
-	 */
-	if (((sinfo_flags & (MSG_EOF|MSG_ABORT)) && (msg_len > 0)) ||
+	/* If MSG_EOF is set, no data can be sent. Disallow sending zero
+	 * length messages when MSG_EOF|MSG_ABORT is not set.
+	 * If MSG_ABORT is set, the message length could be non zero with
+	 * the msg_iov set to the user abort reason.
+ 	 */
+	if (((sinfo_flags & MSG_EOF) && (msg_len > 0)) ||
 	    (!(sinfo_flags & (MSG_EOF|MSG_ABORT)) && (msg_len == 0))) {
 		err = -EINVAL;
 		goto out_nounlock;
@@ -879,7 +881,7 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 		}
 		if (sinfo_flags & MSG_ABORT) {
 			SCTP_DEBUG_PRINTK("Aborting association: %p\n", asoc);
-			sctp_primitive_ABORT(asoc, NULL);
+			sctp_primitive_ABORT(asoc, msg);
 			err = 0;
 			goto out_unlock;
 		}
@@ -1238,6 +1240,9 @@ static inline int sctp_setsockopt_autoclose(struct sock *sk, char *optval,
 {
 	sctp_opt_t *sp = sctp_sk(sk);
 
+	/* Applicable to UDP-style socket only */
+	if (SCTP_SOCKET_TCP == sp->type)
+		return -EOPNOTSUPP;
 	if (optlen != sizeof(int))
 		return -EINVAL;
 	if (copy_from_user(&sp->autoclose, optval, optlen))
@@ -1247,6 +1252,67 @@ static inline int sctp_setsockopt_autoclose(struct sock *sk, char *optval,
 	return 0;
 }
 
+static inline int sctp_setsockopt_set_peer_addr_params(struct sock *sk,
+						       char *optval,
+						       int optlen)
+{
+	struct sctp_paddrparams params;
+	sctp_association_t *asoc;
+	sockaddr_storage_t *addr;
+	sctp_transport_t *trans;
+	int error;
+
+	if (optlen != sizeof(struct sctp_paddrparams))
+		return -EINVAL;
+	if (copy_from_user(&params, optval, optlen))
+		return -EFAULT;
+
+	asoc = sctp_id2assoc(sk, params.spp_assoc_id);
+	if (!asoc)
+		return -EINVAL;
+
+	addr = (sockaddr_storage_t *) &(params.spp_address);
+
+	trans = sctp_assoc_lookup_paddr(asoc, addr);
+	if (!trans)
+		return -ENOENT;
+
+	/* Applications can enable or disable heartbeats for any peer address
+	 * of an association, modify an address's heartbeat interval, force a
+	 * heartbeat to be sent immediately, and adjust the address's maximum
+	 * number of retransmissions sent before an address is considered
+	 * unreachable.
+	 *
+	 * The value of the heartbeat interval, in milliseconds. A value of
+	 * UINT32_MAX (4294967295), when modifying the parameter, specifies
+	 * that a heartbeat should be sent immediately to the peer address,
+	 * and the current interval should remain unchanged.
+	 */
+	if (0xffffffff == params.spp_hbinterval) {
+		error = sctp_primitive_REQUESTHEARTBEAT (asoc, trans);
+		if (error)
+			return error;
+	}
+	else {
+	/* The value of the heartbeat interval, in milliseconds. A value of 0,
+	 * when modifying the parameter, specifies that the heartbeat on this
+	 * address should be disabled.
+	 */
+		if (params.spp_hbinterval) {
+			trans->hb_allowed = 1;
+			trans->hb_interval = params.spp_hbinterval * HZ / 1000;
+		} else
+			trans->hb_allowed = 0;
+	}
+
+	/* spp_pathmaxrxt contains the maximum number of retransmissions
+	 * before this address shall be considered unreachable.
+	 */
+	trans->error_threshold = params.spp_pathmaxrxt;
+
+	return 0;
+}
+
 /* API 6.2 setsockopt(), getsockopt()
  *
  * Applications use setsockopt() and getsockopt() to set or retrieve
@@ -1337,6 +1403,11 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 		retval = sctp_setsockopt_autoclose(sk, optval, optlen);
 		break;
 
+	case SCTP_SET_PEER_ADDR_PARAMS:
+		retval = sctp_setsockopt_set_peer_addr_params(sk, optval,
+							      optlen);
+		break;
+
 	default:
 		retval = -ENOPROTOOPT;
 		break;
@@ -1498,28 +1569,26 @@ static int sctp_getsockopt_sctp_status(struct sock *sk, int len, char *optval,
 
 	if (len != sizeof(status)) {
 		retval = -EINVAL;
-		goto out_nounlock;
+		goto out;
 	}
 
 	if (copy_from_user(&status, optval, sizeof(status))) {
 		retval = -EFAULT;
-		goto out_nounlock;
+		goto out;
 	}
 
-	sctp_lock_sock(sk);
-
 	associd = status.sstat_assoc_id;
 	if ((SCTP_SOCKET_UDP_HIGH_BANDWIDTH != sctp_sk(sk)->type) && associd) {
 		assoc = sctp_id2assoc(sk, associd);
 		if (!assoc) {
 			retval = -EINVAL;
-			goto out_unlock;
+			goto out;
 		}
 	} else {
 		ep = sctp_sk(sk)->ep;
 		if (list_empty(&ep->asocs)) {
 			retval = -EINVAL;
-			goto out_unlock;
+			goto out;
 		}
 
 		assoc = list_entry(ep->asocs.next, sctp_association_t, asocs);
@@ -1538,7 +1607,7 @@ static int sctp_getsockopt_sctp_status(struct sock *sk, int len, char *optval,
 	status.sstat_primary.spinfo_assoc_id = sctp_assoc2id(transport->asoc);
 	memcpy(&status.sstat_primary.spinfo_address,
 	       &(transport->ipaddr), sizeof(sockaddr_storage_t));
-	status.sstat_primary.spinfo_state = transport->state.active;
+	status.sstat_primary.spinfo_state = transport->active;
 	status.sstat_primary.spinfo_cwnd = transport->cwnd;
 	status.sstat_primary.spinfo_srtt = transport->srtt;
 	status.sstat_primary.spinfo_rto = transport->rto;
@@ -1546,7 +1615,7 @@ static int sctp_getsockopt_sctp_status(struct sock *sk, int len, char *optval,
 
 	if (put_user(len, optlen)) {
 		retval = -EFAULT;
-		goto out_unlock;
+		goto out;
 	}
 
 	SCTP_DEBUG_PRINTK("sctp_getsockopt_sctp_status(%d): %d %d %p\n",
@@ -1555,13 +1624,10 @@ static int sctp_getsockopt_sctp_status(struct sock *sk, int len, char *optval,
 
 	if (copy_to_user(optval, &status, len)) {
 		retval = -EFAULT;
-		goto out_unlock;
+		goto out;
 	}
 
-out_unlock:
-	sctp_release_sock(sk);
-
-out_nounlock:
+out:
 	return (retval);
 }
 
@@ -1593,6 +1659,9 @@ static inline int sctp_getsockopt_set_events(struct sock *sk, int len, char *opt
 
 static inline int sctp_getsockopt_autoclose(struct sock *sk, int len, char *optval, int *optlen)
 {
+	/* Applicable to UDP-style socket only */
+	if (SCTP_SOCKET_TCP == sctp_sk(sk)->type)
+		return -EOPNOTSUPP;
 	if (len != sizeof(int))
 		return -EINVAL;
 	if (copy_to_user(optval, &sctp_sk(sk)->autoclose, len))
@@ -1614,10 +1683,10 @@ SCTP_STATIC int sctp_do_peeloff(sctp_association_t *assoc, struct socket **newso
 	int err = 0;
 
 	/* An association cannot be branched off from an already peeled-off
-	 * socket.
+	 * socket, nor is this supported for tcp style sockets.
 	 */
-	if (SCTP_SOCKET_UDP_HIGH_BANDWIDTH == sctp_sk(oldsk)->type)
-		return -EINVAL;
+	if (SCTP_SOCKET_UDP != sctp_sk(oldsk)->type)
+		return -EOPNOTSUPP;
 
 	/* Create a new socket.  */
 	err = sock_create(PF_INET, SOCK_SEQPACKET, IPPROTO_SCTP, &tmpsock);
@@ -1676,25 +1745,23 @@ static inline int sctp_getsockopt_peeloff(struct sock *sk, int len, char *optval
 	if (copy_from_user(&peeloff, optval, len))
 		return -EFAULT;
 
-	sctp_lock_sock(sk);
-
 	assoc = sctp_id2assoc(sk, peeloff.associd);
 	if (NULL == assoc) {
 		retval = -EINVAL;
-		goto out_unlock;
+		goto out;
 	}
 
 	SCTP_DEBUG_PRINTK("%s: sk: %p assoc: %p\n", __FUNCTION__, sk, assoc);
 
 	retval = sctp_do_peeloff(assoc, &newsock);
 	if (retval < 0)
-		goto out_unlock;
+		goto out;
 
 	/* Map the socket to an unused fd that can be returned to the user.  */
 	retval = sock_map_fd(newsock);
 	if (retval < 0) {
 		sock_release(newsock);
-		goto out_unlock;
+		goto out;
 	}
 
 	SCTP_DEBUG_PRINTK("%s: sk: %p assoc: %p newsk: %p sd: %d\n",
@@ -1705,11 +1772,54 @@ static inline int sctp_getsockopt_peeloff(struct sock *sk, int len, char *optval
 	if (copy_to_user(optval, &peeloff, len))
 		retval = -EFAULT;
 
-out_unlock:
-	sctp_release_sock(sk);
+out:
 	return retval;
 }
 
+static inline int sctp_getsockopt_get_peer_addr_params(struct sock *sk,
+					int len, char *optval, int *optlen)
+{
+	struct sctp_paddrparams params;
+	sctp_association_t *asoc;
+	sockaddr_storage_t *addr;
+	sctp_transport_t *trans;
+
+	if (len != sizeof(struct sctp_paddrparams))
+		return -EINVAL;
+	if (copy_from_user(&params, optval, *optlen))
+		return -EFAULT;
+
+	asoc = sctp_id2assoc(sk, params.spp_assoc_id);
+	if (!asoc)
+		return -EINVAL;
+
+	addr = (sockaddr_storage_t *) &(params.spp_address);
+
+	trans = sctp_assoc_lookup_paddr(asoc, addr);
+	if (!trans)
+		return -ENOENT;
+
+	/* The value of the heartbeat interval, in milliseconds. A value of 0,
+	 * when modifying the parameter, specifies that the heartbeat on this
+	 * address should be disabled.
+	 */
+	if (!trans->hb_allowed)
+		params.spp_hbinterval = 0;
+	else
+		params.spp_hbinterval = trans->hb_interval * 1000 / HZ;
+
+	/* spp_pathmaxrxt contains the maximum number of retransmissions
+	 * before this address shall be considered unreachable.
+	 */
+	params.spp_pathmaxrxt = trans->error_threshold;
+	
+	if (copy_to_user(optval, &params, len))
+		return -EFAULT;
+	*optlen = len;
+
+	return 0;
+}
+
 SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 				char *optval, int *optlen)
 {
@@ -1740,6 +1850,8 @@ SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 	if (get_user(len, optlen))
 		return -EFAULT;
 
+	sctp_lock_sock(sk);
+
 	switch (optname) {
 	case SCTP_STATUS:
 		retval = sctp_getsockopt_sctp_status(sk, len, optval, optlen);
@@ -1762,11 +1874,17 @@ SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 		retval = sctp_getsockopt_peeloff(sk, len, optval, optlen);
 		break;
 
+	case SCTP_GET_PEER_ADDR_PARAMS:
+		retval = sctp_getsockopt_get_peer_addr_params(sk, len, optval,
+							      optlen);
+		break;
+
 	default:
 		retval = -ENOPROTOOPT;
 		break;
 	};
 
+	sctp_release_sock(sk);
 	return retval;
 }
 

net/sctp/transport.c

@@ -87,7 +87,6 @@ sctp_transport_t *sctp_transport_init(sctp_transport_t *peer,
 	peer->ipaddr = *addr;
 	peer->af_specific = sctp_get_af_specific(addr);
 	peer->asoc = NULL;
-	peer->pmtu = peer->af_specific->get_dst_mtu(addr);
 
 	/* From 6.3.1 RTO Calculation:
 	 *
@@ -105,8 +104,8 @@ sctp_transport_t *sctp_transport_init(sctp_transport_t *peer,
 	peer->last_time_used = jiffies;
 	peer->last_time_ecne_reduced = jiffies;
 
-	peer->state.active = 1;
-	peer->state.hb_allowed = 0;
+	peer->active = 1;
+	peer->hb_allowed = 0;
 
 	/* Initialize the default path max_retrans.  */
 	peer->max_retrans = proto->max_retrans_path;
@@ -161,6 +160,7 @@ void sctp_transport_destroy(sctp_transport_t *transport)
 	if (transport->asoc)
 		sctp_association_put(transport->asoc);
 
+	dst_release(transport->dst);
 	kfree(transport);
 	SCTP_DBG_OBJCNT_DEC(transport);
 }
@@ -200,6 +200,78 @@ void sctp_transport_set_owner(sctp_transport_t *transport,
 	sctp_association_hold(asoc);
 }
 
+/* Caches the dst entry for a transport's destination address and an optional
+ * souce address.
+ */ 
+void sctp_transport_route(sctp_transport_t *transport,
+			  sockaddr_storage_t *saddr)
+{
+	sctp_association_t *asoc = transport->asoc;
+	sctp_func_t *af = transport->af_specific;
+	sockaddr_storage_t *daddr = &transport->ipaddr;
+	sctp_bind_addr_t *bp;
+	rwlock_t *addr_lock;
+	struct sockaddr_storage_list *laddr;
+	struct list_head *pos;
+	struct dst_entry *dst;
+
+	dst = af->get_dst(daddr, saddr);
+
+	/* If there is no association or if a source address is passed,
+	 * no more validation is required.
+	 */
+	if (!asoc || saddr)
+		goto out;
+
+	if (SCTP_STATE_ESTABLISHED == asoc->state) {
+		bp = &asoc->base.bind_addr;
+		addr_lock = &asoc->base.addr_lock;
+	} else {
+		bp = &asoc->ep->base.bind_addr;
+		addr_lock = &asoc->ep->base.addr_lock;
+	}
+
+	if (dst) {
+		/* Walk through the bind address list and look for a bind
+		 * address that matches the source address of the returned dst.
+		 */
+		sctp_read_lock(addr_lock);
+		list_for_each(pos, &bp->address_list) {
+			laddr = list_entry(pos, struct sockaddr_storage_list,
+					   list);
+			if (af->cmp_saddr(dst, &laddr->a))
+				goto out_unlock;
+		}
+		sctp_read_unlock(addr_lock);
+	
+		/* None of the bound addresses match the source address of the
+		 * dst. So release it.
+		 */	
+		dst_release(dst);
+	}
+
+	/* Walk through the bind address list and try to get a dst that
+	 * matches a bind address as the source address.
+	 */
+	sctp_read_lock(addr_lock);
+	list_for_each(pos, &bp->address_list) {
+		laddr = list_entry(pos, struct sockaddr_storage_list, list);
+
+		dst = af->get_dst(daddr, &laddr->a);
+		if (dst)
+			goto out_unlock;
+	}
+
+out_unlock:
+	sctp_read_unlock(addr_lock);
+out:
+	transport->dst = dst;
+	if (dst)
+		transport->pmtu = dst_pmtu(dst);
+	else
+		transport->pmtu = SCTP_DEFAULT_MAXSEGMENT;
+}
+
 /* Hold a reference to a transport.  */
 void sctp_transport_hold(sctp_transport_t *transport)
 {