Merge us.ibm.com:/home/sridhar/BK/linux-2.5.62

into us.ibm.com:/home/sridhar/BK/lksctp-2.5.62

include/net/sctp/command.h

@@ -86,6 +86,7 @@ typedef enum {
 	SCTP_CMD_PURGE_OUTQUEUE, /* Purge all data waiting to be sent. */
 	SCTP_CMD_SETUP_T2,       /* Hi-level, setup T2-shutdown parms.  */
 	SCTP_CMD_RTO_PENDING,	 /* Set transport's rto_pending. */
+	SCTP_CMD_CHUNK_PD,	 /* Partial data delivery considerations. */
 
 	SCTP_CMD_LAST
 } sctp_verb_t;
@@ -115,7 +116,7 @@ typedef union {
 	struct sctp_transport *transport;
 	sctp_bind_addr_t  *bp;
 	sctp_init_chunk_t *init;
-	sctp_ulpevent_t *ulpevent;
+	struct sctp_ulpevent *ulpevent;
 	sctp_packet_t *packet;
 	sctp_sackhdr_t *sackh;
 } sctp_arg_t;
@@ -163,7 +164,7 @@ SCTP_ARG_CONSTRUCTOR(ASOC,	sctp_association_t *, asoc)
 SCTP_ARG_CONSTRUCTOR(TRANSPORT,	struct sctp_transport *, transport)
 SCTP_ARG_CONSTRUCTOR(BA,	sctp_bind_addr_t *, bp)
 SCTP_ARG_CONSTRUCTOR(PEER_INIT,	sctp_init_chunk_t *, init)
-SCTP_ARG_CONSTRUCTOR(ULPEVENT,  sctp_ulpevent_t *, ulpevent)
+SCTP_ARG_CONSTRUCTOR(ULPEVENT,  struct sctp_ulpevent *, ulpevent)
 SCTP_ARG_CONSTRUCTOR(PACKET,	sctp_packet_t *, packet)
 SCTP_ARG_CONSTRUCTOR(SACKH,	sctp_sackhdr_t *, sackh)
 

include/net/sctp/sctp.h

@@ -214,6 +214,7 @@ DECLARE_SNMP_STAT(struct sctp_mib, sctp_statistics);
 #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)
+#define SCTP_DEC_STATS(field)		SNMP_DEC_STATS(sctp_statistics, field)
 
 /* Determine if this is a valid kernel address.  */
 static inline int sctp_is_valid_kaddr(unsigned long addr)

include/net/sctp/structs.h

@@ -2,7 +2,7 @@
  * Copyright (c) 1999-2000 Cisco, Inc.
  * Copyright (c) 1999-2001 Motorola, Inc.
  * Copyright (c) 2001 Intel Corp.
- * Copyright (c) 2001-2002 International Business Machines Corp.
+ * Copyright (c) 2001-2003 International Business Machines Corp.
  *
  * This file is part of the SCTP kernel reference Implementation
  *
@@ -105,27 +105,25 @@ union sctp_addr {
 
 /* Forward declarations for data structures. */
 struct sctp_protocol;
-struct SCTP_endpoint;
-struct SCTP_association;
+struct sctp_endpoint;
+struct sctp_association;
 struct sctp_transport;
-struct SCTP_packet;
-struct SCTP_chunk;
-struct SCTP_inqueue;
+struct sctp_packet;
+struct sctp_chunk;
+struct sctp_inq;
 struct sctp_outq;
-struct SCTP_bind_addr;
+struct sctp_bind_addr;
 struct sctp_ulpq;
 struct sctp_opt;
 struct sctp_endpoint_common;
 struct sctp_ssnmap;
 
 typedef struct sctp_protocol sctp_protocol_t;
-typedef struct SCTP_endpoint sctp_endpoint_t;
-typedef struct SCTP_association sctp_association_t;
-typedef struct SCTP_packet sctp_packet_t;
-typedef struct SCTP_chunk sctp_chunk_t;
-typedef struct SCTP_inqueue sctp_inqueue_t;
-typedef struct SCTP_bind_addr sctp_bind_addr_t;
-typedef struct sctp_opt sctp_opt_t;
+typedef struct sctp_endpoint sctp_endpoint_t;
+typedef struct sctp_association sctp_association_t;
+typedef struct sctp_packet sctp_packet_t;
+typedef struct sctp_chunk sctp_chunk_t;
+typedef struct sctp_bind_addr sctp_bind_addr_t;
 typedef struct sctp_endpoint_common sctp_endpoint_common_t;
 
 #include <net/sctp/tsnmap.h>
@@ -250,10 +248,10 @@ struct sctp_af {
 					 int optname,
 					 char *optval,
 					 int *optlen);
-	struct dst_entry *(*get_dst)	(sctp_association_t *asoc,
+	struct dst_entry *(*get_dst)	(struct sctp_association *asoc,
 					 union sctp_addr *daddr,
 					 union sctp_addr *saddr);
-	void 		(*get_saddr)	(sctp_association_t *asoc,
+	void 		(*get_saddr)	(struct sctp_association *asoc,
 					 struct dst_entry *dst,
 					 union sctp_addr *daddr,
 				 	 union sctp_addr *saddr);	 
@@ -289,7 +287,7 @@ int sctp_register_af(struct sctp_af *);
 
 /* Protocol family functions. */
 struct sctp_pf {
-	void (*event_msgname)(sctp_ulpevent_t *, char *, int *);
+	void (*event_msgname)(struct sctp_ulpevent *, char *, int *);
 	void (*skb_msgname)  (struct sk_buff *, char *, int *);
 	int  (*af_supported) (sa_family_t);
 	int  (*cmp_addr) (const union sctp_addr *,
@@ -311,6 +309,9 @@ struct sctp_opt {
 	/* What kind of a socket is this? */
 	sctp_socket_type_t type;
 
+	/* PF_ family specific functions.  */
+	struct sctp_pf *pf;
+
 	/* What is our base endpointer? */
 	sctp_endpoint_t *ep;
 
@@ -324,7 +325,10 @@ struct sctp_opt {
 	__u32 autoclose;
 	__u8 nodelay;
 	__u8 disable_fragments;
-	struct sctp_pf *pf;
+	__u8 pd_mode;
+
+	/* Receive to here while partial delivery is in effect. */
+	struct sk_buff_head pd_lobby;
 };
 
 
@@ -484,7 +488,7 @@ static inline __u16 sctp_ssn_next(struct sctp_stream *stream, __u16 id)
  * As a matter of convenience, we remember the SCTP common header for
  * each chunk as well as a few other header pointers...
  */
-struct SCTP_chunk {
+struct sctp_chunk {
 	/* These first three elements MUST PRECISELY match the first
 	 * three elements of struct sk_buff.  This allows us to reuse
 	 * all the skb_* queue management functions.
@@ -594,7 +598,7 @@ typedef sctp_chunk_t *(sctp_packet_phandler_t)(sctp_association_t *);
 /* This structure holds lists of chunks as we are assembling for
  * transmission.
  */
-struct SCTP_packet {
+struct sctp_packet {
 	/* These are the SCTP header values (host order) for the packet. */
 	__u16 source_port;
 	__u16 destination_port;
@@ -846,8 +850,8 @@ unsigned long sctp_transport_timeout(struct sctp_transport *);
 /* This is the structure we use to queue packets as they come into
  * SCTP.  We write packets to it and read chunks from it.
  */
-struct SCTP_inqueue {
-	/* This is actually a queue of sctp_chunk_t each
+struct sctp_inq {
+	/* This is actually a queue of sctp_chunk each
 	 * containing a partially decoded packet.
 	 */
 	struct sk_buff_head in;
@@ -864,13 +868,12 @@ struct SCTP_inqueue {
 	int malloced;        /* Is this structure kfree()able?  */
 };
 
-sctp_inqueue_t *sctp_inqueue_new(void);
-void sctp_inqueue_init(sctp_inqueue_t *);
-void sctp_inqueue_free(sctp_inqueue_t *);
-void sctp_push_inqueue(sctp_inqueue_t *, sctp_chunk_t *packet);
-sctp_chunk_t *sctp_pop_inqueue(sctp_inqueue_t *);
-void sctp_inqueue_set_th_handler(sctp_inqueue_t *,
-                                 void (*)(void *), void *);
+struct sctp_inq *sctp_inq_new(void);
+void sctp_inq_init(struct sctp_inq *);
+void sctp_inq_free(struct sctp_inq *);
+void sctp_inq_push(struct sctp_inq *, sctp_chunk_t *packet);
+struct sctp_chunk *sctp_inq_pop(struct sctp_inq *);
+void sctp_inq_set_th_handler(struct sctp_inq *, void (*)(void *), void *);
 
 /* This is the structure we use to hold outbound chunks.  You push
  * chunks in and they automatically pop out the other end as bundled
@@ -954,7 +957,7 @@ void sctp_retransmit_mark(struct sctp_outq *, struct sctp_transport *, __u8);
 
 
 /* These bind address data fields common between endpoints and associations */
-struct SCTP_bind_addr {
+struct sctp_bind_addr {
 
 	/* RFC 2960 12.1 Parameters necessary for the SCTP instance
 	 *
@@ -1043,7 +1046,7 @@ struct sctp_endpoint_common {
 	struct sock *sk;
 
 	/* This is where we receive inbound chunks.  */
-	sctp_inqueue_t   inqueue;
+	struct sctp_inq   inqueue;
 
 	/* This substructure includes the defining parameters of the
 	 * endpoint:
@@ -1076,7 +1079,7 @@ struct sctp_endpoint_common {
  * off one of these.
  */
 
-struct SCTP_endpoint {
+struct sctp_endpoint {
 	/* Common substructure for endpoint and association. */
 	sctp_endpoint_common_t base;
 
@@ -1172,7 +1175,7 @@ __u32 sctp_generate_tsn(const sctp_endpoint_t *ep);
 
 
 /* Here we have information about each individual association. */
-struct SCTP_association {
+struct sctp_association {
 
 	/* A base structure common to endpoint and association.
 	 * In this context, it represents the associations's view
@@ -1288,19 +1291,11 @@ struct SCTP_association {
 		 * used in the bulk of the text.  This value is hidden
 		 * in tsn_map--we get it by calling sctp_tsnmap_get_ctsn().
 		 */
-		sctp_tsnmap_t tsn_map;
+		struct sctp_tsnmap tsn_map;
 		__u8 _map[sctp_tsnmap_storage_size(SCTP_TSN_MAP_SIZE)];
 
-		/* We record duplicate TSNs here.  We clear this after
-		 * every SACK.
-		 * FIXME: We should move this into the tsnmap? --jgrimm
-		 */
-		sctp_dup_tsn_t dup_tsns[SCTP_MAX_DUP_TSNS];
-		int next_dup_tsn;
-
 		/* Do we need to sack the peer? */
-		uint8_t sack_needed;
-
+		__u8    sack_needed;
 		/* These are capabilities which our peer advertised.  */
 		__u8	ecn_capable;     /* Can peer do ECN? */
 		__u8	ipv4_address;    /* Peer understands IPv4 addresses? */
@@ -1457,7 +1452,10 @@ struct SCTP_association {
 
 	struct {
 		__u16 stream;
+		__u16 flags;
 		__u32 ppid;
+		__u32 context;
+		__u32 timetolive;
 	} defaults;
 
 	/* This tracks outbound ssn for a given stream.  */
@@ -1615,6 +1613,7 @@ void sctp_association_put(sctp_association_t *);
 void sctp_association_hold(sctp_association_t *);
 
 struct sctp_transport *sctp_assoc_choose_shutdown_transport(sctp_association_t *);
+void sctp_assoc_update_retran_path(sctp_association_t *);
 struct sctp_transport *sctp_assoc_lookup_paddr(const sctp_association_t *,
 					  const union sctp_addr *);
 struct sctp_transport *sctp_assoc_add_peer(sctp_association_t *,

include/net/sctp/tsnmap.h

 /* SCTP kernel reference Implementation Copyright (C) 1999-2001
  * Cisco, Motorola, Intel, and International Business Machines Corp.
- * 
+ *
  * This file is part of the SCTP kernel reference Implementation
- * 
+ *
  * These are the definitions needed for the tsnmap type.  The tsnmap is used
  * to track out of order TSNs received.
- * 
- * 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 one of the
  * following email addresses:
- * 
+ *
  * Jon Grimm <jgrimm@us.ibm.com>
  * La Monte H.P. Yarroll <piggy@acm.org>
  * Karl Knutson <karl@athena.chicago.il.us>
- * 
+ *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
  */
@@ -38,8 +38,6 @@
 #ifndef __sctp_tsnmap_h__
 #define __sctp_tsnmap_h__
 
-
-
 /* RFC 2960 12.2 Parameters necessary per association (i.e. the TCB)
  * Mapping  An array of bits or bytes indicating which out of
  * Array    order TSN's have been received (relative to the
@@ -48,9 +46,7 @@
  *          will be set to all zero. This structure may be
  *          in the form of a circular buffer or bit array.
  */
-typedef struct sctp_tsnmap {
-
-
+struct sctp_tsnmap {
 	/* This array counts the number of chunks with each TSN.
 	 * It points at one of the two buffers with which we will
 	 * ping-pong between.
@@ -93,25 +89,30 @@ typedef struct sctp_tsnmap {
 	/* This is the highest TSN we've marked.  */
 	__u32 max_tsn_seen;
 
-	/* No. of data chunks pending receipt. used by SCTP_STATUS sockopt */
+	/* Data chunks pending receipt. used by SCTP_STATUS sockopt */
 	__u16 pending_data;
 
+	/* We record duplicate TSNs here.  We clear this after
+	 * every SACK.  Store up to SCTP_MAX_DUP_TSNS worth of
+	 * information.
+	 */
+	__u32 dup_tsns[SCTP_MAX_DUP_TSNS];
+	__u16 num_dup_tsns;
+
 	int malloced;
 
 	__u8 raw_map[0];
-} sctp_tsnmap_t;
+};
 
-typedef struct sctp_tsnmap_iter {
+struct sctp_tsnmap_iter {
 	__u32 start;
-} sctp_tsnmap_iter_t;
-
+};
 
 /* Create a new tsnmap.  */
-sctp_tsnmap_t *sctp_tsnmap_new(__u16 len, __u32 initial_tsn,
-			       int priority);
+struct sctp_tsnmap *sctp_tsnmap_new(__u16 len, __u32 init_tsn, int priority);
 
 /* Dispose of a tsnmap.  */
-void sctp_tsnmap_free(sctp_tsnmap_t *map);
+void sctp_tsnmap_free(struct sctp_tsnmap *);
 
 /* This macro assists in creation of external storage for variable length
  * internal buffers.  We double allocate so the overflow map works.
@@ -119,9 +120,8 @@ void sctp_tsnmap_free(sctp_tsnmap_t *map);
 #define sctp_tsnmap_storage_size(count) (sizeof(__u8) * (count) * 2)
 
 /* Initialize a block of memory as a tsnmap.  */
-sctp_tsnmap_t *sctp_tsnmap_init(sctp_tsnmap_t *map, __u16 len, __u32 initial_tsn);
-
-
+struct sctp_tsnmap *sctp_tsnmap_init(struct sctp_tsnmap *, __u16 len,
+				     __u32 initial_tsn);
 
 /* Test the tracking state of this TSN.
  * Returns:
@@ -129,29 +129,51 @@ sctp_tsnmap_t *sctp_tsnmap_init(sctp_tsnmap_t *map, __u16 len, __u32 initial_tsn
  *  >0 if the TSN has been seen (duplicate)
  *  <0 if the TSN is invalid (too large to track)
  */
-int sctp_tsnmap_check(const sctp_tsnmap_t *map, __u32 tsn);
+int sctp_tsnmap_check(const struct sctp_tsnmap *, __u32 tsn);
 
 /* Mark this TSN as seen.  */
-void sctp_tsnmap_mark(sctp_tsnmap_t *map, __u32 tsn);
+void sctp_tsnmap_mark(struct sctp_tsnmap *, __u32 tsn);
 
 /* Retrieve the Cumulative TSN ACK Point.  */
-__u32 sctp_tsnmap_get_ctsn(const sctp_tsnmap_t *map);
+__u32 sctp_tsnmap_get_ctsn(const struct sctp_tsnmap *);
 
 /* Retrieve the highest TSN we've seen.  */
-__u32 sctp_tsnmap_get_max_tsn_seen(const sctp_tsnmap_t *map);
+__u32 sctp_tsnmap_get_max_tsn_seen(const struct sctp_tsnmap *);
+
+/* How many Duplicate TSNs are stored? */
+static inline __u16 sctp_tsnmap_num_dups(struct sctp_tsnmap *map)
+{
+	return map->num_dup_tsns;
+}
+
+/* Return pointer to duplicate tsn array as needed by SACK. */
+static inline __u32 *sctp_tsnmap_get_dups(struct sctp_tsnmap *map)
+{
+	map->num_dup_tsns = 0;
+	return map->dup_tsns;
+}
+
+/* Mark a duplicate TSN.  Note:  we limit how many we are willing to
+ * store and consequently report.
+ */
+static inline void sctp_tsnmap_mark_dup(struct sctp_tsnmap *map, __u32 tsn)
+{
+	if (map->num_dup_tsns < SCTP_MAX_DUP_TSNS)
+		map->dup_tsns[map->num_dup_tsns++] = tsn;
+}
 
 /* Is there a gap in the TSN map? */
-int sctp_tsnmap_has_gap(const sctp_tsnmap_t *map);
+int sctp_tsnmap_has_gap(const struct sctp_tsnmap *);
 
 /* Initialize a gap ack block interator from user-provided memory.  */
-void sctp_tsnmap_iter_init(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter);
+void sctp_tsnmap_iter_init(const struct sctp_tsnmap *,
+			   struct sctp_tsnmap_iter *);
 
 /* Get the next gap ack blocks.  We return 0 if there are no more
  * gap ack blocks.
  */
-int sctp_tsnmap_next_gap_ack(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter,
-			     __u16 *start, __u16 *end);
-
+int sctp_tsnmap_next_gap_ack(const struct sctp_tsnmap *,
+	struct sctp_tsnmap_iter *,__u16 *start, __u16 *end);
 
 #endif /* __sctp_tsnmap_h__ */
 

include/net/sctp/ulpevent.h

@@ -6,34 +6,34 @@
  * Copyright (c) 2001 Nokia, Inc.
  * Copyright (c) 2001 La Monte H.P. Yarroll
  *
- * These are the definitions needed for the sctp_ulpevent type.  The 
+ * These are the definitions needed for the sctp_ulpevent type.  The
  * sctp_ulpevent type is used to carry information from the state machine
- * upwards to the ULP. 
- * 
- * The SCTP reference implementation  is free software; 
- * you can redistribute it and/or modify it under the terms of 
+ * upwards to the ULP.
+ *
+ * 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 one of the
  * following email addresses:
- * 
+ *
  * Jon Grimm <jgrimm@us.ibm.com>
  * La Monte H.P. Yarroll <piggy@acm.org>
  * Karl Knutson <karl@athena.chicago.il.us>
- * 
+ *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
  */
@@ -46,85 +46,97 @@
 /* Warning: This sits inside an skb.cb[] area.  Be very careful of
  * growing this structure as it is at the maximum limit now.
  */
-typedef struct sctp_ulpevent {
-	int malloced;
-	sctp_association_t *asoc;
-	struct sk_buff *parent;
+struct sctp_ulpevent {
+	struct sctp_association *asoc;
 	struct sctp_sndrcvinfo sndrcvinfo;
-	int chunk_flags; /* Temp. until we get a new chunk_t */
 	int msg_flags;
-} sctp_ulpevent_t;
-
-
-sctp_ulpevent_t *sctp_ulpevent_new(int size, int msg_flags, int priority);
-
-sctp_ulpevent_t *sctp_ulpevent_init(sctp_ulpevent_t *event, struct sk_buff *skb, int msg_flags);
-
-void sctp_ulpevent_free(sctp_ulpevent_t *event);
-
-int sctp_ulpevent_is_notification(const sctp_ulpevent_t *event);
-
-sctp_ulpevent_t *sctp_ulpevent_make_assoc_change(
-				const struct SCTP_association *asoc,
-				__u16 flags,
-				__u16 state,
-				__u16 error,
-				__u16 outbound,
-				__u16 inbound,
-				int priority);
-
-sctp_ulpevent_t *sctp_ulpevent_make_peer_addr_change(
-				const struct SCTP_association *asoc,
-				const struct sockaddr_storage *aaddr,
-				int flags,
-				int state,
-				int error,
-				int priority);
-
-sctp_ulpevent_t *sctp_ulpevent_make_remote_error(
-				const struct SCTP_association *asoc,
-				struct SCTP_chunk *chunk,
-				__u16 flags,
-				int priority);
-sctp_ulpevent_t *sctp_ulpevent_make_send_failed(
-				const struct SCTP_association *asoc,
-				struct SCTP_chunk *chunk,
-				__u16 flags,
-				__u32 error,
-				int priority);
-
-sctp_ulpevent_t *sctp_ulpevent_make_shutdown_event(
-				const struct SCTP_association *asoc,
-				__u16 flags,
-				int priority);
-
-sctp_ulpevent_t *sctp_ulpevent_make_rcvmsg(struct SCTP_association *asoc,
-					   struct SCTP_chunk *chunk,
-					   int priority);
-
-void sctp_ulpevent_read_sndrcvinfo(const sctp_ulpevent_t *event,
-				   struct msghdr *msghdr);
-
-__u16 sctp_ulpevent_get_notification_type(const sctp_ulpevent_t *event);
+};
 
+/* Retrieve the skb this event sits inside of. */
+static inline struct sk_buff *sctp_event2skb(struct sctp_ulpevent *ev)
+{
+	return container_of((void *)ev, struct sk_buff, cb);
+}
 
+/* Retrieve & cast the event sitting inside the skb. */
+static inline struct sctp_ulpevent *sctp_skb2event(struct sk_buff *skb)
+{
+	return (struct sctp_ulpevent *)skb->cb; 
+}
+	
+struct sctp_ulpevent *sctp_ulpevent_new(int size, int flags, int priority);
+struct sctp_ulpevent *sctp_ulpevent_init(struct sctp_ulpevent *, int flags);
+void sctp_ulpevent_free(struct sctp_ulpevent *);
+int sctp_ulpevent_is_notification(const struct sctp_ulpevent *);
+
+struct sctp_ulpevent *sctp_ulpevent_make_assoc_change(
+	const struct sctp_association *asoc,
+	__u16 flags,
+	__u16 state,
+	__u16 error,
+	__u16 outbound,
+	__u16 inbound,
+	int priority);
+
+struct sctp_ulpevent *sctp_ulpevent_make_peer_addr_change(
+	const struct sctp_association *asoc,
+	const struct sockaddr_storage *aaddr,
+	int flags,
+	int state,
+	int error,
+	int priority);
+
+struct sctp_ulpevent *sctp_ulpevent_make_remote_error(
+	const struct sctp_association *asoc,
+	struct sctp_chunk *chunk,
+	__u16 flags,
+	int priority);
+struct sctp_ulpevent *sctp_ulpevent_make_send_failed(
+	const struct sctp_association *asoc,
+	struct sctp_chunk *chunk,
+	__u16 flags,
+	__u32 error,
+	int priority);
+
+struct sctp_ulpevent *sctp_ulpevent_make_shutdown_event(
+	const struct sctp_association *asoc,
+	__u16 flags,
+	int priority);
+
+struct sctp_ulpevent *sctp_ulpevent_make_pdapi(
+	const struct sctp_association *asoc,
+	__u32 indication, int priority);
+
+struct sctp_ulpevent *sctp_ulpevent_make_rcvmsg(struct sctp_association *asoc,
+	struct sctp_chunk *chunk,
+	int priority);
+
+void sctp_ulpevent_read_sndrcvinfo(const struct sctp_ulpevent *event,
+	struct msghdr *);
+__u16 sctp_ulpevent_get_notification_type(const struct sctp_ulpevent *event);
+
+/* Is this event type enabled? */
+static inline int sctp_ulpevent_type_enabled(__u16 sn_type,
+					     struct sctp_event_subscribe *mask)
+{
+	char *amask = (char *) mask;
+	return amask[sn_type - SCTP_SN_TYPE_BASE];
+}
 
 /* Given an event subscription, is this event enabled? */
-static inline int sctp_ulpevent_is_enabled(const sctp_ulpevent_t *event,
-					   const struct sctp_event_subscribe *mask)
+static inline int sctp_ulpevent_is_enabled(const struct sctp_ulpevent *event,
+					   struct sctp_event_subscribe *mask)
 {
-	const char *amask = (const char *) mask;
 	__u16 sn_type;
 	int enabled = 1;
 
 	if (sctp_ulpevent_is_notification(event)) {
 		sn_type = sctp_ulpevent_get_notification_type(event);
-		enabled = amask[sn_type - SCTP_SN_TYPE_BASE];
+		enabled = sctp_ulpevent_type_enabled(sn_type, mask);
 	}
 	return enabled;
 }
 
-
 #endif /* __sctp_ulpevent_h__ */
 
 

include/net/sctp/ulpqueue.h

@@ -48,7 +48,8 @@
 
 /* A structure to carry information to the ULP (e.g. Sockets API) */
 struct sctp_ulpq {
-	int malloced;
+	char malloced;
+	char pd_mode;
 	sctp_association_t *asoc;
 	struct sk_buff_head reasm;
 	struct sk_buff_head lobby;
@@ -60,13 +61,19 @@ struct sctp_ulpq *sctp_ulpq_init(struct sctp_ulpq *, sctp_association_t *);
 void sctp_ulpq_free(struct sctp_ulpq *);
 
 /* Add a new DATA chunk for processing. */
-int sctp_ulpq_tail_data(struct sctp_ulpq *, sctp_chunk_t *chunk, int priority);
+int sctp_ulpq_tail_data(struct sctp_ulpq *, struct sctp_chunk *, int);
 
 /* Add a new event for propogation to the ULP. */
 int sctp_ulpq_tail_event(struct sctp_ulpq *, struct sctp_ulpevent *ev);
 
-/* Is the ulpqueue empty. */
-int sctp_ulpqueue_is_empty(struct sctp_ulpq *);
+/* Perform partial delivery. */
+void sctp_ulpq_partial_delivery(struct sctp_ulpq *, struct sctp_chunk *, int);
+
+/* Abort the partial delivery. */
+void sctp_ulpq_abort_pd(struct sctp_ulpq *, int);
+
+/* Clear the partial data delivery condition on this socket. */
+int sctp_clear_pd(struct sock *sk);
 
 #endif /* __sctp_ulpqueue_h__ */
 

include/net/sctp/user.h

@@ -166,6 +166,7 @@ struct sctp_sndrcvinfo {
 	__u32 sinfo_context;
 	__u32 sinfo_timetolive;
 	__u32 sinfo_tsn;
+	__u32 sinfo_cumtsn;
 	sctp_assoc_t sinfo_assoc_id;
 };
 
@@ -367,6 +368,7 @@ struct sctp_rcv_pdapi_event {
 	sctp_assoc_t pdapi_assoc_id;
 };
 
+enum { SCTP_PARTIAL_DELIVERY_ABORTED=0, };
 
 /*
  * Described in Section 7.3
@@ -414,8 +416,8 @@ enum sctp_sn_type {
 	SCTP_SN_TYPE_BASE     = (1<<15),
 	SCTP_ASSOC_CHANGE,
 	SCTP_PEER_ADDR_CHANGE,
-	SCTP_REMOTE_ERROR,
 	SCTP_SEND_FAILED,
+	SCTP_REMOTE_ERROR,
 	SCTP_SHUTDOWN_EVENT,
 	SCTP_PARTIAL_DELIVERY_EVENT,
 	SCTP_ADAPTION_INDICATION,

net/sctp/Makefile

@@ -10,7 +10,7 @@ sctp-y := sm_statetable.o sm_statefuns.o sm_sideeffect.o \
 	  inqueue.o outqueue.o ulpqueue.o command.o \
 	  tsnmap.o bind_addr.o socket.o primitive.o \
 	  output.o input.o hashdriver.o sla1.o \
-	  debug.o ssnmap.o
+	  debug.o ssnmap.o proc.o
 
 ifeq ($(CONFIG_SCTP_ADLER32), y)
 sctp-y += adler32.o

net/sctp/associola.c

@@ -95,7 +95,7 @@ sctp_association_t *sctp_association_init(sctp_association_t *asoc,
 					  sctp_scope_t scope,
 					  int priority)
 {
-	sctp_opt_t *sp;
+	struct sctp_opt *sp;
 	int i;
 
 	/* Retrieve the SCTP per socket area.  */
@@ -241,8 +241,8 @@ sctp_association_t *sctp_association_init(sctp_association_t *asoc,
 	asoc->peer.sack_needed = 1;
 
 	/* Create an input queue.  */
-	sctp_inqueue_init(&asoc->base.inqueue);
-	sctp_inqueue_set_th_handler(&asoc->base.inqueue,
+	sctp_inq_init(&asoc->base.inqueue);
+	sctp_inq_set_th_handler(&asoc->base.inqueue,
 				    (void (*)(void *))sctp_assoc_bh_rcv,
 				    asoc);
 
@@ -260,7 +260,6 @@ sctp_association_t *sctp_association_init(sctp_association_t *asoc,
 
 	/* Set up the tsn tracking. */
 	sctp_tsnmap_init(&asoc->peer.tsn_map, SCTP_TSN_MAP_SIZE, 0);
-	asoc->peer.next_dup_tsn = 0;
 
 	skb_queue_head_init(&asoc->addip_chunks);
 
@@ -311,7 +310,7 @@ void sctp_association_free(sctp_association_t *asoc)
 	sctp_ulpq_free(&asoc->ulpq);
 
 	/* Dispose of any pending chunks on the inqueue. */
-	sctp_inqueue_free(&asoc->base.inqueue);
+	sctp_inq_free(&asoc->base.inqueue);
 
 	/* Free ssnmap storage. */
 	sctp_ssnmap_free(asoc->ssnmap);
@@ -368,7 +367,7 @@ struct sctp_transport *sctp_assoc_add_peer(sctp_association_t *asoc,
 					   int priority)
 {
 	struct sctp_transport *peer;
-	sctp_opt_t *sp;
+	struct sctp_opt *sp;
 	unsigned short port;
 
 	/* AF_INET and AF_INET6 share common port field. */
@@ -505,7 +504,7 @@ void sctp_assoc_control_transport(sctp_association_t *asoc,
 	struct sctp_transport *t = NULL;
 	struct sctp_transport *first;
 	struct sctp_transport *second;
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct list_head *pos;
 	int spc_state = 0;
 
@@ -776,7 +775,7 @@ static void sctp_assoc_bh_rcv(sctp_association_t *asoc)
 	sctp_endpoint_t *ep;
 	sctp_chunk_t *chunk;
 	struct sock *sk;
-	sctp_inqueue_t *inqueue;
+	struct sctp_inq *inqueue;
 	int state, subtype;
 	sctp_assoc_t associd = sctp_assoc2id(asoc);
 	int error = 0;
@@ -786,7 +785,7 @@ static void sctp_assoc_bh_rcv(sctp_association_t *asoc)
 	sk = asoc->base.sk;
 
 	inqueue = &asoc->base.inqueue;
-	while (NULL != (chunk = sctp_pop_inqueue(inqueue))) {
+	while (NULL != (chunk = sctp_inq_pop(inqueue))) {
 		state = asoc->state;
 		subtype = chunk->chunk_hdr->type;
 
@@ -795,6 +794,8 @@ static void sctp_assoc_bh_rcv(sctp_association_t *asoc)
 		 */
 		if (sctp_chunk_is_data(chunk))
 			asoc->peer.last_data_from = chunk->transport;
+		else
+			SCTP_INC_STATS(SctpInCtrlChunks);
 
 		if (chunk->transport)
 			chunk->transport->last_time_heard = jiffies;
@@ -819,7 +820,7 @@ static void sctp_assoc_bh_rcv(sctp_association_t *asoc)
 /* This routine moves an association from its old sk to a new sk.  */
 void sctp_assoc_migrate(sctp_association_t *assoc, struct sock *newsk)
 {
-	sctp_opt_t *newsp = sctp_sk(newsk);
+	struct sctp_opt *newsp = sctp_sk(newsk);
 
 	/* Delete the association from the old endpoint's list of
 	 * associations.
@@ -848,7 +849,6 @@ void sctp_assoc_update(sctp_association_t *asoc, sctp_association_t *new)
 	/* Copy in new parameters of peer. */
 	asoc->c = new->c;
 	asoc->peer.rwnd = new->peer.rwnd;
-	asoc->peer.next_dup_tsn = new->peer.next_dup_tsn;
 	asoc->peer.sack_needed = new->peer.sack_needed;
 	asoc->peer.i = new->peer.i;
 	sctp_tsnmap_init(&asoc->peer.tsn_map, SCTP_TSN_MAP_SIZE,
@@ -887,26 +887,19 @@ void sctp_assoc_update(sctp_association_t *asoc, sctp_association_t *new)
 
 }
 
-/* Choose the transport for sending a shutdown packet.
+/* Update the retran path for sending a retransmitted packet.
  * Round-robin through the active transports, else round-robin
  * through the inactive transports as this is the next best thing
  * we can try.
  */
-struct sctp_transport *sctp_assoc_choose_shutdown_transport(sctp_association_t *asoc)
+void sctp_assoc_update_retran_path(sctp_association_t *asoc)
 {
 	struct sctp_transport *t, *next;
 	struct list_head *head = &asoc->peer.transport_addr_list;
 	struct list_head *pos;
 
-	/* If this is the first time SHUTDOWN is sent, use the active
-	 * path.
-	 */
-	if (!asoc->shutdown_last_sent_to)
-		return asoc->peer.active_path;
-
-	/* Otherwise, find the next transport in a round-robin fashion. */
-
-	t = asoc->shutdown_last_sent_to;
+	/* Find the next transport in a round-robin fashion. */
+	t = asoc->peer.retran_path;
 	pos = &t->transports;
 	next = NULL;
 
@@ -935,13 +928,30 @@ struct sctp_transport *sctp_assoc_choose_shutdown_transport(sctp_association_t *
 		 * other active transports.  If so, use the next
 		 * transport.
 		 */
-		if (t == asoc->shutdown_last_sent_to) {
+		if (t == asoc->peer.retran_path) {
 			t = next;
 			break;
 		}
 	}
 
-	return t;
+	asoc->peer.retran_path = t;
+}
+
+/* Choose the transport for sending a SHUTDOWN packet.  */
+struct sctp_transport *sctp_assoc_choose_shutdown_transport(sctp_association_t *asoc)
+{
+	/* If this is the first time SHUTDOWN is sent, use the active path,
+	 * else use the retran path. If the last SHUTDOWN was sent over the
+	 * retran path, update the retran path and use it. 
+	 */
+	if (!asoc->shutdown_last_sent_to)
+		return asoc->peer.active_path;
+	else {
+		if (asoc->shutdown_last_sent_to == asoc->peer.retran_path)
+			sctp_assoc_update_retran_path(asoc);
+		return asoc->peer.retran_path;
+	}
+
 }
 
 /* Update the association's pmtu and frag_point by going through all the
@@ -990,13 +1000,13 @@ void sctp_assoc_rwnd_increase(sctp_association_t *asoc, int len)
 		asoc->rwnd += len;
 	}
 
-	SCTP_DEBUG_PRINTK("%s: asoc %p rwnd increased by %d to (%u, %u) - %u\n",
-			  __FUNCTION__, asoc, len, asoc->rwnd, asoc->rwnd_over,
-			  asoc->a_rwnd);
+	SCTP_DEBUG_PRINTK("%s: asoc %p rwnd increased by %d to (%u, %u) "
+			  "- %u\n", __FUNCTION__, asoc, len, asoc->rwnd,
+			  asoc->rwnd_over, asoc->a_rwnd);
 
 	/* Send a window update SACK if the rwnd has increased by at least the
 	 * minimum of the association's PMTU and half of the receive buffer.
-	 * The algorithm used is similar to the one described in 
+	 * The algorithm used is similar to the one described in
 	 * Section 4.2.3.3 of RFC 1122.
 	 */
 	if ((asoc->state == SCTP_STATE_ESTABLISHED) &&
@@ -1006,15 +1016,14 @@ void sctp_assoc_rwnd_increase(sctp_association_t *asoc, int len)
 		SCTP_DEBUG_PRINTK("%s: Sending window update SACK- asoc: %p "
 				  "rwnd: %u a_rwnd: %u\n",
 				  __FUNCTION__, asoc, asoc->rwnd, asoc->a_rwnd);
-		sack = sctp_make_sack(asoc); 
+		sack = sctp_make_sack(asoc);
 		if (!sack)
-			return;	
+			return;
 
 		/* Update the last advertised rwnd value. */
 		asoc->a_rwnd = asoc->rwnd;
 
 		asoc->peer.sack_needed = 0;
-		asoc->peer.next_dup_tsn = 0;
 
 		sctp_outq_tail(&asoc->outqueue, sack);
 
@@ -1022,7 +1031,7 @@ void sctp_assoc_rwnd_increase(sctp_association_t *asoc, int len)
 		timer = &asoc->timers[SCTP_EVENT_TIMEOUT_SACK];
 		if (timer_pending(timer) && del_timer(timer))
 			sctp_association_put(asoc);
-	} 
+	}
 }
 
 /* Decrease asoc's rwnd by len. */

net/sctp/endpointola.c

@@ -92,7 +92,7 @@ sctp_endpoint_t *sctp_endpoint_new(sctp_protocol_t *proto,
 sctp_endpoint_t *sctp_endpoint_init(sctp_endpoint_t *ep, sctp_protocol_t *proto,
 				    struct sock *sk, int priority)
 {
-	sctp_opt_t *sp = sctp_sk(sk);
+	struct sctp_opt *sp = sctp_sk(sk);
 	memset(ep, 0, sizeof(sctp_endpoint_t));
 
 	/* Initialize the base structure. */
@@ -105,10 +105,10 @@ sctp_endpoint_t *sctp_endpoint_init(sctp_endpoint_t *ep, sctp_protocol_t *proto,
 	ep->base.malloced = 1;
 
 	/* Create an input queue.  */
-	sctp_inqueue_init(&ep->base.inqueue);
+	sctp_inq_init(&ep->base.inqueue);
 
 	/* Set its top-half handler */
-	sctp_inqueue_set_th_handler(&ep->base.inqueue,
+	sctp_inq_set_th_handler(&ep->base.inqueue,
 				    (void (*)(void *))sctp_endpoint_bh_rcv,
 				    ep);
 
@@ -198,7 +198,7 @@ void sctp_endpoint_destroy(sctp_endpoint_t *ep)
 	sctp_unhash_endpoint(ep);
 
 	/* Cleanup the inqueue. */
-	sctp_inqueue_free(&ep->base.inqueue);
+	sctp_inq_free(&ep->base.inqueue);
 
 	sctp_bind_addr_free(&ep->base.bind_addr);
 
@@ -333,7 +333,7 @@ static void sctp_endpoint_bh_rcv(sctp_endpoint_t *ep)
 	struct sock *sk;
 	struct sctp_transport *transport;
 	sctp_chunk_t *chunk;
-	sctp_inqueue_t *inqueue;
+	struct sctp_inq *inqueue;
 	sctp_subtype_t subtype;
 	sctp_state_t state;
 	int error = 0;
@@ -345,7 +345,7 @@ static void sctp_endpoint_bh_rcv(sctp_endpoint_t *ep)
 	inqueue = &ep->base.inqueue;
 	sk = ep->base.sk;
 
-	while (NULL != (chunk = sctp_pop_inqueue(inqueue))) {		
+	while (NULL != (chunk = sctp_inq_pop(inqueue))) {		
 		subtype.chunk = chunk->chunk_hdr->type;
 
 		/* We might have grown an association since last we
@@ -369,6 +369,8 @@ static void sctp_endpoint_bh_rcv(sctp_endpoint_t *ep)
 		 */
 		if (asoc && sctp_chunk_is_data(chunk))
 			asoc->peer.last_data_from = chunk->transport;
+		else
+			SCTP_INC_STATS(SctpInCtrlChunks);
 
 		if (chunk->transport)
 			chunk->transport->last_time_heard = jiffies;

net/sctp/input.c

@@ -90,6 +90,7 @@ static inline int sctp_rcv_checksum(struct sk_buff *skb)
 
 	if (val != cmp) {
 		/* CRC failure, dump it. */
+		SCTP_INC_STATS_BH(SctpChecksumErrors);
 		return -1;
 	}
 	return 0;
@@ -115,6 +116,8 @@ int sctp_rcv(struct sk_buff *skb)
 	if (skb->pkt_type!=PACKET_HOST)
 		goto discard_it;
 
+	SCTP_INC_STATS_BH(SctpInSCTPPacks);
+
 	sh = (struct sctphdr *) skb->h.raw;
 
 	/* Pull up the IP and SCTP headers. */
@@ -160,8 +163,10 @@ int sctp_rcv(struct sk_buff *skb)
 	 */
 	if (!asoc) {
 		ep = __sctp_rcv_lookup_endpoint(&dest);
-		if (sctp_rcv_ootb(skb))
+		if (sctp_rcv_ootb(skb)) {
+			SCTP_INC_STATS_BH(SctpOutOfBlues);
 			goto discard_release;
+		}
 	}
 
 	/* Retrieve the common input handling substructure. */
@@ -248,7 +253,7 @@ int sctp_rcv(struct sk_buff *skb)
 int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb)
 {
 	sctp_chunk_t *chunk;
-	sctp_inqueue_t *inqueue;
+	struct sctp_inq *inqueue;
 
 	/* One day chunk will live inside the skb, but for
 	 * now this works.
@@ -256,7 +261,7 @@ int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb)
 	chunk = (sctp_chunk_t *) skb;
 	inqueue = &chunk->rcvr->inqueue;
 
-	sctp_push_inqueue(inqueue, chunk);
+	sctp_inq_push(inqueue, chunk);
         return 0;
 }
 

net/sctp/inqueue.c

@@ -47,8 +47,8 @@
 #include <net/sctp/sm.h>
 #include <linux/interrupt.h>
 
-/* Initialize an SCTP_inqueue.  */
-void sctp_inqueue_init(sctp_inqueue_t *queue)
+/* Initialize an SCTP inqueue.  */
+void sctp_inq_init(struct sctp_inq *queue)
 {
 	skb_queue_head_init(&queue->in);
 	queue->in_progress = NULL;
@@ -59,21 +59,21 @@ void sctp_inqueue_init(sctp_inqueue_t *queue)
 	queue->malloced = 0;
 }
 
-/* Create an initialized SCTP_inqueue.  */
-sctp_inqueue_t *sctp_inqueue_new(void)
+/* Create an initialized sctp_inq.  */
+struct sctp_inq *sctp_inq_new(void)
 {
-	sctp_inqueue_t *retval;
+	struct sctp_inq *retval;
 
-	retval = t_new(sctp_inqueue_t, GFP_ATOMIC);
+	retval = t_new(struct sctp_inq, GFP_ATOMIC);
 	if (retval) {
-		sctp_inqueue_init(retval);
+		sctp_inq_init(retval);
 		retval->malloced = 1;
 	}
         return retval;
 }
 
 /* Release the memory associated with an SCTP inqueue.  */
-void sctp_inqueue_free(sctp_inqueue_t *queue)
+void sctp_inq_free(struct sctp_inq *queue)
 {
 	sctp_chunk_t *chunk;
 
@@ -96,7 +96,7 @@ void sctp_inqueue_free(sctp_inqueue_t *queue)
 /* Put a new packet in an SCTP inqueue.
  * We assume that packet->sctp_hdr is set and in host byte order.
  */
-void sctp_push_inqueue(sctp_inqueue_t *q, sctp_chunk_t *packet)
+void sctp_inq_push(struct sctp_inq *q, sctp_chunk_t *packet)
 {
 	/* Directly call the packet handling routine. */
 
@@ -114,7 +114,7 @@ void sctp_push_inqueue(sctp_inqueue_t *q, sctp_chunk_t *packet)
  * WARNING:  If you need to put the chunk on another queue, you need to
  * make a shallow copy (clone) of it.
  */
-sctp_chunk_t *sctp_pop_inqueue(sctp_inqueue_t *queue)
+sctp_chunk_t *sctp_inq_pop(struct sctp_inq *queue)
 {
 	sctp_chunk_t *chunk;
 	sctp_chunkhdr_t *ch = NULL;
@@ -172,7 +172,7 @@ sctp_chunk_t *sctp_pop_inqueue(sctp_inqueue_t *queue)
 		chunk->end_of_packet = 1;
 	}
 
-	SCTP_DEBUG_PRINTK("+++sctp_pop_inqueue+++ chunk %p[%s],"
+	SCTP_DEBUG_PRINTK("+++sctp_inq_pop+++ chunk %p[%s],"
 			  " length %d, skb->len %d\n",chunk,
 			  sctp_cname(SCTP_ST_CHUNK(chunk->chunk_hdr->type)),
 			  ntohs(chunk->chunk_hdr->length), chunk->skb->len);
@@ -182,12 +182,12 @@ sctp_chunk_t *sctp_pop_inqueue(sctp_inqueue_t *queue)
 /* Set a top-half handler.
  *
  * Originally, we the top-half handler was scheduled as a BH.  We now
- * call the handler directly in sctp_push_inqueue() at a time that
+ * call the handler directly in sctp_inq_push() at a time that
  * we know we are lock safe.
  * The intent is that this routine will pull stuff out of the
  * inqueue and process it.
  */
-void sctp_inqueue_set_th_handler(sctp_inqueue_t *q,
+void sctp_inq_set_th_handler(struct sctp_inq *q,
 				 void (*callback)(void *), void *arg)
 {
 	INIT_WORK(&q->immediate, callback, arg);

net/sctp/ipv6.c

@@ -132,6 +132,8 @@ static inline int sctp_v6_xmit(struct sk_buff *skb,
 			  __FUNCTION__, skb, skb->len, NIP6(fl.fl6_src),
 			  NIP6(fl.fl6_dst));
 
+	SCTP_INC_STATS(SctpOutSCTPPacks);
+
 	return ip6_xmit(sk, skb, &fl, np->opt);
 }
 
@@ -444,7 +446,7 @@ static void sctp_inet6_msgname(char *msgname, int *addr_len)
 }
 
 /* Initialize a PF_INET msgname from a ulpevent. */
-static void sctp_inet6_event_msgname(sctp_ulpevent_t *event, char *msgname,
+static void sctp_inet6_event_msgname(struct sctp_ulpevent *event, char *msgname,
 				     int *addrlen)
 {
 	struct sockaddr_in6 *sin6, *sin6from;

net/sctp/output.c

@@ -419,6 +419,7 @@ int sctp_packet_transmit(sctp_packet_t *packet)
 	dst = transport->dst;
 	/* The 'obsolete' field of dst is set to 2 when a dst is freed. */
 	if (!dst || (dst->obsolete > 1)) {
+		dst_release(dst);
 		sctp_transport_route(transport, NULL, sctp_sk(sk));
 		sctp_assoc_sync_pmtu(asoc);
 	}

net/sctp/outqueue.c

@@ -193,11 +193,17 @@ int sctp_outq_tail(struct sctp_outq *q, sctp_chunk_t *chunk)
 			  : "Illegal Chunk");
 
 			skb_queue_tail(&q->out, (struct sk_buff *) chunk);
+			if (chunk->chunk_hdr->flags & SCTP_DATA_UNORDERED)
+				SCTP_INC_STATS(SctpOutUnorderChunks);
+			else
+				SCTP_INC_STATS(SctpOutOrderChunks);
 			q->empty = 0;
 			break;
 		};
-	} else
+	} else {
 		skb_queue_tail(&q->control, (struct sk_buff *) chunk);
+		SCTP_INC_STATS(SctpOutCtrlChunks);
+	}
 
 	if (error < 0)
 		return error;
@@ -315,6 +321,11 @@ void sctp_retransmit(struct sctp_outq *q, struct sctp_transport *transport,
 	switch(reason) {
 	case SCTP_RETRANSMIT_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.
+		 */
+		if (transport == transport->asoc->peer.retran_path)
+			sctp_assoc_update_retran_path(transport->asoc);
 		break;
 	case SCTP_RETRANSMIT_FAST_RTX:
 		sctp_transport_lower_cwnd(transport, SCTP_LOWER_CWND_FAST_RTX);

net/sctp/proc.c

+/* SCTP kernel reference Implementation
+ * Copyright (c) 2003 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 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
+ * 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.
+ *
+ * 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:
+ *    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.
+ */
+
+#include <linux/types.h>
+#include <linux/seq_file.h>
+#include <net/sctp/sctp.h>
+
+static char *sctp_snmp_list[] = {
+#define SCTP_SNMP_ENTRY(x) #x 
+	SCTP_SNMP_ENTRY(SctpCurrEstab),
+	SCTP_SNMP_ENTRY(SctpActiveEstabs),
+	SCTP_SNMP_ENTRY(SctpPassiveEstabs),
+	SCTP_SNMP_ENTRY(SctpAborteds),
+	SCTP_SNMP_ENTRY(SctpShutdowns),
+	SCTP_SNMP_ENTRY(SctpOutOfBlues),
+	SCTP_SNMP_ENTRY(SctpChecksumErrors),
+	SCTP_SNMP_ENTRY(SctpOutCtrlChunks),
+	SCTP_SNMP_ENTRY(SctpOutOrderChunks),
+	SCTP_SNMP_ENTRY(SctpOutUnorderChunks),
+	SCTP_SNMP_ENTRY(SctpInCtrlChunks),
+	SCTP_SNMP_ENTRY(SctpInOrderChunks),
+	SCTP_SNMP_ENTRY(SctpInUnorderChunks),
+	SCTP_SNMP_ENTRY(SctpFragUsrMsgs),
+	SCTP_SNMP_ENTRY(SctpReasmUsrMsgs),
+	SCTP_SNMP_ENTRY(SctpOutSCTPPacks),
+	SCTP_SNMP_ENTRY(SctpInSCTPPacks),
+#undef SCTP_SNMP_ENTRY
+};
+
+/* Return the current value of a particular entry in the mib by adding its
+ * per cpu counters.
+ */ 
+static unsigned long
+fold_field(void *mib[], int nr)
+{
+	unsigned long res = 0;
+	int i;
+
+	for (i = 0; i < NR_CPUS; i++) {
+		if (!cpu_possible(i))
+			continue;
+		res +=
+		    *((unsigned long *) (((void *) per_cpu_ptr(mib[0], i)) +
+					 sizeof (unsigned long) * nr));
+		res +=
+		    *((unsigned long *) (((void *) per_cpu_ptr(mib[1], i)) +
+					 sizeof (unsigned long) * nr));
+	}
+	return res;
+}
+
+/* Display sctp snmp mib statistics(/proc/net/sctp/snmp). */
+static int sctp_snmp_seq_show(struct seq_file *seq, void *v)
+{
+	int i;
+
+	for (i = 0; i < sizeof(sctp_snmp_list) / sizeof(char *); i++)
+		seq_printf(seq, "%-32s\t%ld\n", sctp_snmp_list[i],
+			   fold_field((void **)sctp_statistics, i));
+
+	return 0;
+}
+
+/* Initialize the seq file operations for 'snmp' object. */
+static int sctp_snmp_seq_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, sctp_snmp_seq_show, NULL);
+}
+
+static struct file_operations sctp_snmp_seq_fops = {
+	.open	 = sctp_snmp_seq_open,
+	.read	 = seq_read,
+	.llseek	 = seq_lseek,
+	.release = single_release,
+};
+
+/* Set up the proc fs entry for 'snmp' object. */
+int __init sctp_snmp_proc_init(void)
+{
+	struct proc_dir_entry *p;
+
+	p = create_proc_entry("snmp", S_IRUGO, proc_net_sctp);
+	if (!p)
+		return -ENOMEM;
+
+	p->proc_fops = &sctp_snmp_seq_fops;
+
+	return 0;
+}
+
+/* Cleanup the proc fs entry for 'snmp' object. */
+void sctp_snmp_proc_exit(void)
+{
+	remove_proc_entry("snmp", proc_net_sctp);
+}

net/sctp/protocol.c

@@ -75,6 +75,9 @@ static struct sctp_af *sctp_af_v6_specific;
 
 extern struct net_proto_family inet_family_ops;
 
+extern int sctp_snmp_proc_init(void);
+extern int sctp_snmp_proc_exit(void);
+
 /* Return the address of the control sock. */
 struct sock *sctp_get_ctl_sock(void)
 {
@@ -82,21 +85,32 @@ struct sock *sctp_get_ctl_sock(void)
 }
 
 /* Set up the proc fs entry for the SCTP protocol. */
-__init void sctp_proc_init(void)
+__init int sctp_proc_init(void)
 {
+	int rc = 0;
+
 	if (!proc_net_sctp) {
 		struct proc_dir_entry *ent;
 		ent = proc_mkdir("net/sctp", 0);
 		if (ent) {
 			ent->owner = THIS_MODULE;
 			proc_net_sctp = ent;
-		}
+		} else
+			rc = -ENOMEM;
 	}
+
+	if (sctp_snmp_proc_init())
+		rc = -ENOMEM;
+
+	return rc;
 }
 
 /* Clean up the proc fs entry for the SCTP protocol. */
 void sctp_proc_exit(void)
 {
+
+	sctp_snmp_proc_exit();
+
 	if (proc_net_sctp) {
 		proc_net_sctp = NULL;
 		remove_proc_entry("net/sctp", 0);
@@ -124,7 +138,6 @@ static void sctp_v4_copy_addrlist(struct list_head *addrlist,
 		/* Add the address to the local list.  */
 		addr = t_new(struct sockaddr_storage_list, GFP_ATOMIC);
 		if (addr) {
-			INIT_LIST_HEAD(&addr->list);
 			addr->a.v4.sin_family = AF_INET;
 			addr->a.v4.sin_port = 0;
 			addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
@@ -436,7 +449,6 @@ struct dst_entry *sctp_v4_get_dst(sctp_association_t *asoc,
 
 		if (AF_INET == laddr->a.sa.sa_family) {
 			fl.fl4_src = laddr->a.v4.sin_addr.s_addr;
-			dst = sctp_v4_get_dst(asoc, daddr, &laddr->a);
 			if (!ip_route_output_key(&rt, &fl)) {
 				dst = &rt->u.dst;
 				goto out_unlock;
@@ -557,7 +569,7 @@ static void sctp_inet_msgname(char *msgname, int *addr_len)
 }
 
 /* Copy the primary address of the peer primary address as the msg_name. */
-static void sctp_inet_event_msgname(sctp_ulpevent_t *event, char *msgname,
+static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
 				    int *addr_len)
 {
 	struct sockaddr_in *sin, *sinfrom;
@@ -628,6 +640,7 @@ static inline int sctp_v4_xmit(struct sk_buff *skb,
 			  NIPQUAD(((struct rtable *)skb->dst)->rt_src),
 			  NIPQUAD(((struct rtable *)skb->dst)->rt_dst));
 
+	SCTP_INC_STATS(SctpOutSCTPPacks);
 	return ip_queue_xmit(skb, ipfragok);
 }
 

net/sctp/sm_make_chunk.c

@@ -222,9 +222,7 @@ sctp_chunk_t *sctp_make_init(const sctp_association_t *asoc,
 
 	sctp_addto_chunk(retval, sizeof(sctp_paramhdr_t), &sat_param);
 	sctp_addto_chunk(retval, sizeof(sat_addr_types), sat_addr_types);
-
 	sctp_addto_chunk(retval, sizeof(ecap_param), &ecap_param);
-
 nodata:
 	if (addrs.v)
 		kfree(addrs.v);
@@ -245,7 +243,8 @@ sctp_chunk_t *sctp_make_init_ack(const sctp_association_t *asoc,
 
 	retval = NULL;
 
-	addrs = sctp_bind_addrs_to_raw(&asoc->base.bind_addr, &addrs_len, priority);
+	addrs = sctp_bind_addrs_to_raw(&asoc->base.bind_addr, &addrs_len, 
+				       priority);
 	if (!addrs.v)
 		goto nomem_rawaddr;
 
@@ -586,14 +585,12 @@ sctp_chunk_t *sctp_make_sack(const sctp_association_t *asoc)
 	sctp_gap_ack_block_t gab;
 	int length;
 	__u32 ctsn;
-	sctp_tsnmap_iter_t iter;
-	__u16 num_gabs;
-	__u16 num_dup_tsns = asoc->peer.next_dup_tsn;
-	const sctp_tsnmap_t *map = &asoc->peer.tsn_map;
+	struct sctp_tsnmap_iter iter;
+	__u16 num_gabs, num_dup_tsns;
+	struct sctp_tsnmap *map = (struct sctp_tsnmap *)&asoc->peer.tsn_map;
 
 	ctsn = sctp_tsnmap_get_ctsn(map);
-	SCTP_DEBUG_PRINTK("make_sack: sackCTSNAck sent is 0x%x.\n",
-			  ctsn);
+	SCTP_DEBUG_PRINTK("sackCTSNAck sent is 0x%x.\n", ctsn);
 
 	/* Count the number of Gap Ack Blocks.  */
 	sctp_tsnmap_iter_init(map, &iter);
@@ -603,15 +600,17 @@ sctp_chunk_t *sctp_make_sack(const sctp_association_t *asoc)
 		/* Do nothing. */
 	}
 
+	num_dup_tsns = sctp_tsnmap_num_dups(map);
+
 	/* Initialize the SACK header.  */
 	sack.cum_tsn_ack	    = htonl(ctsn);
 	sack.a_rwnd 		    = htonl(asoc->rwnd);
 	sack.num_gap_ack_blocks     = htons(num_gabs);
-	sack.num_dup_tsns  = htons(num_dup_tsns);
+	sack.num_dup_tsns           = htons(num_dup_tsns);
 
 	length = sizeof(sack)
 		+ sizeof(sctp_gap_ack_block_t) * num_gabs
-		+ sizeof(sctp_dup_tsn_t) * num_dup_tsns;
+		+ sizeof(__u32) * num_dup_tsns;
 
 	/* Create the chunk.  */
 	retval = sctp_make_chunk(asoc, SCTP_CID_SACK, 0, length);
@@ -658,21 +657,18 @@ sctp_chunk_t *sctp_make_sack(const sctp_association_t *asoc)
 	while(sctp_tsnmap_next_gap_ack(map, &iter, &gab.start, &gab.end)) {
 		gab.start = htons(gab.start);
 		gab.end = htons(gab.end);
-		sctp_addto_chunk(retval,
-				 sizeof(sctp_gap_ack_block_t),
-				 &gab);
+		sctp_addto_chunk(retval, sizeof(sctp_gap_ack_block_t), &gab);
 	}
 
 	/* Register the duplicates.  */
-	sctp_addto_chunk(retval,
-			 sizeof(sctp_dup_tsn_t) * num_dup_tsns,
-			 &asoc->peer.dup_tsns);
+	sctp_addto_chunk(retval, sizeof(__u32) * num_dup_tsns,
+			 sctp_tsnmap_get_dups(map));
 
 nodata:
 	return retval;
 }
 
-/* FIXME: Comments. */
+/* Make a SHUTDOWN chunk. */
 sctp_chunk_t *sctp_make_shutdown(const sctp_association_t *asoc)
 {
 	sctp_chunk_t *retval;
@@ -689,7 +685,6 @@ sctp_chunk_t *sctp_make_shutdown(const sctp_association_t *asoc)
 
 	retval->subh.shutdown_hdr =
 		sctp_addto_chunk(retval, sizeof(shut), &shut);
-
 nodata:
 	return retval;
 }
@@ -1180,6 +1175,9 @@ int sctp_datachunks_from_user(sctp_association_t *asoc,
 	over = msg_len % max;
 	offset = 0;
 
+	if (whole && over)
+		SCTP_INC_STATS_USER(SctpFragUsrMsgs);
+
 	/* Create chunks for all the full sized DATA chunks. */
 	for (i=0, len=first_len; i < whole; i++) {
 		frag = SCTP_DATA_MIDDLE_FRAG;

net/sctp/sm_sideeffect.c

@@ -527,10 +527,8 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 			break;
 
 		case SCTP_CMD_REPORT_DUP:
-			if (asoc->peer.next_dup_tsn < SCTP_MAX_DUP_TSNS) {
-				asoc->peer.dup_tsns[asoc->peer.next_dup_tsn++] =
-					ntohl(command->obj.u32);
-			}
+			sctp_tsnmap_mark_dup(&asoc->peer.tsn_map,
+					     ntohl(command->obj.u32));
 			break;
 
 		case SCTP_CMD_REPORT_BIGGAP:
@@ -598,6 +596,13 @@ int sctp_cmd_interpreter(sctp_event_t event_type, sctp_subtype_t subtype,
 			t->rto_pending = 1;
 			break;
 
+		case SCTP_CMD_CHUNK_PD:
+			/* Send a chunk to the sockets layer.  */
+			sctp_ulpq_partial_delivery(&asoc->ulpq,
+						   command->obj.ptr,
+						   GFP_ATOMIC);
+			break;
+
 		default:
 			printk(KERN_WARNING "Impossible command: %u, %p\n",
 			       command->verb, command->obj.ptr);
@@ -737,7 +742,6 @@ int sctp_gen_sack(sctp_association_t *asoc, int force, sctp_cmd_seq_t *commands)
 		asoc->a_rwnd = asoc->rwnd;
 
 		asoc->peer.sack_needed = 0;
-		asoc->peer.next_dup_tsn = 0;
 
 		error = sctp_outq_tail(&asoc->outqueue, sack);
 
@@ -1014,7 +1018,7 @@ static void sctp_do_8_2_transport_strike(sctp_association_t *asoc,
 static void sctp_cmd_init_failed(sctp_cmd_seq_t *commands,
 				 sctp_association_t *asoc)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 
 	event = sctp_ulpevent_make_assoc_change(asoc,
 						0,
@@ -1041,7 +1045,7 @@ static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *commands,
 				  sctp_subtype_t subtype,
 				  sctp_chunk_t *chunk)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	__u16 error = 0;
 
 	switch(event_type) {
@@ -1061,12 +1065,11 @@ static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *commands,
 		break;
 	}
 
-	event = sctp_ulpevent_make_assoc_change(asoc,
-						0,
-						SCTP_COMM_LOST,
-						error, 0, 0,
-						GFP_ATOMIC);
+	/* Cancel any partial delivery in progress. */
+	sctp_ulpq_abort_pd(&asoc->ulpq, GFP_ATOMIC);
 
+	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));
@@ -1141,7 +1144,7 @@ static void sctp_cmd_hb_timers_stop(sctp_cmd_seq_t *cmds,
 		if (del_timer(&t->hb_timer))
 			sctp_transport_put(t);
 	}
-} 
+}
 
 /* Helper function to update the heartbeat timer. */
 static void sctp_cmd_hb_timer_update(sctp_cmd_seq_t *cmds,

net/sctp/sm_statefuns.c

@@ -102,7 +102,7 @@ sctp_disposition_t sctp_sf_do_4_C(const sctp_endpoint_t *ep,
 				  sctp_cmd_seq_t *commands)
 {
 	sctp_chunk_t *chunk = arg;
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 
 	/* RFC 2960 6.10 Bundling
 	 *
@@ -145,6 +145,9 @@ sctp_disposition_t sctp_sf_do_4_C(const sctp_endpoint_t *ep,
 
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_CLOSED));
+	
+	SCTP_INC_STATS(SctpShutdowns);
+	SCTP_DEC_STATS(SctpCurrEstab);
 
 	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
 
@@ -223,6 +226,7 @@ sctp_disposition_t sctp_sf_do_5_1B_init(const sctp_endpoint_t *ep,
 			if (packet) {
 				sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
 						SCTP_PACKET(packet));
+				SCTP_INC_STATS(SctpOutCtrlChunks);
 				return SCTP_DISPOSITION_CONSUME;
 			} else {
 				return SCTP_DISPOSITION_NOMEM;
@@ -264,7 +268,7 @@ sctp_disposition_t sctp_sf_do_5_1B_init(const sctp_endpoint_t *ep,
 
 	if (sctp_assoc_set_bind_addr_from_ep(new_asoc, GFP_ATOMIC) < 0)
 		goto nomem_ack;
-	
+
 	repl = sctp_make_init_ack(new_asoc, chunk, GFP_ATOMIC, len);
 	if (!repl)
 		goto nomem_ack;
@@ -379,6 +383,7 @@ sctp_disposition_t sctp_sf_do_5_1C_ack(const sctp_endpoint_t *ep,
 		sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(reply));
 		sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 				SCTP_STATE(SCTP_STATE_CLOSED));
+		SCTP_INC_STATS(SctpAborteds);
 		sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
 		return SCTP_DISPOSITION_DELETE_TCB;
 	}
@@ -388,6 +393,9 @@ sctp_disposition_t sctp_sf_do_5_1C_ack(const sctp_endpoint_t *ep,
 	if (!sctp_verify_init(asoc, chunk->chunk_hdr->type,
 			      (sctp_init_chunk_t *)chunk->chunk_hdr, chunk,
 			      &err_chunk)) {
+
+		SCTP_INC_STATS(SctpAborteds);
+
 		/* This chunk contains fatal error. It is to be discarded.
 		 * Send an ABORT, with causes if there is any.
 		 */
@@ -403,6 +411,7 @@ sctp_disposition_t sctp_sf_do_5_1C_ack(const sctp_endpoint_t *ep,
 			if (packet) {
 				sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
 						SCTP_PACKET(packet));
+				SCTP_INC_STATS(SctpOutCtrlChunks);
 				sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 						SCTP_STATE(SCTP_STATE_CLOSED));
 				sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB,
@@ -504,7 +513,7 @@ sctp_disposition_t sctp_sf_do_5_1D_ce(const sctp_endpoint_t *ep,
 	sctp_association_t *new_asoc;
 	sctp_init_chunk_t *peer_init;
 	sctp_chunk_t *repl;
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 	int error = 0;
 	sctp_chunk_t *err_chk_p;
 
@@ -557,6 +566,8 @@ sctp_disposition_t sctp_sf_do_5_1D_ce(const sctp_endpoint_t *ep,
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_ASOC, SCTP_ASOC(new_asoc));
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_ESTABLISHED));
+	SCTP_INC_STATS(SctpCurrEstab);
+	SCTP_INC_STATS(SctpPassiveEstabs);
 	sctp_add_cmd_sf(commands, SCTP_CMD_HB_TIMERS_START, SCTP_NULL());
 
 	if (new_asoc->autoclose)
@@ -636,7 +647,7 @@ sctp_disposition_t sctp_sf_do_5_1E_ca(const sctp_endpoint_t *ep,
 				      const sctp_subtype_t type, void *arg,
 				      sctp_cmd_seq_t *commands)
 {
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 
 	/* RFC 2960 5.1 Normal Establishment of an Association
 	 *
@@ -648,6 +659,8 @@ sctp_disposition_t sctp_sf_do_5_1E_ca(const sctp_endpoint_t *ep,
 			SCTP_TO(SCTP_EVENT_TIMEOUT_T1_COOKIE));
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_ESTABLISHED));
+	SCTP_INC_STATS(SctpCurrEstab);
+	SCTP_INC_STATS(SctpActiveEstabs);
 	sctp_add_cmd_sf(commands, SCTP_CMD_HB_TIMERS_START, SCTP_NULL());
 	if (asoc->autoclose)
 		sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_START,
@@ -719,6 +732,8 @@ sctp_disposition_t sctp_sf_sendbeat_8_3(const sctp_endpoint_t *ep,
 	if (asoc->overall_error_count >= asoc->overall_error_threshold) {
 		/* CMD_ASSOC_FAILED calls CMD_DELETE_TCB. */
 		sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+		SCTP_INC_STATS(SctpAborteds);
+		SCTP_DEC_STATS(SctpCurrEstab);
 		return SCTP_DISPOSITION_DELETE_TCB;
 	}
 
@@ -929,6 +944,8 @@ static int sctp_sf_send_restart_abort(union sctp_addr *ssa,
 		goto out;
 	sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT, SCTP_PACKET(pkt));
 
+	SCTP_INC_STATS(SctpOutCtrlChunks);
+
 	/* Discard the rest of the inbound packet. */
 	sctp_add_cmd_sf(commands, SCTP_CMD_DISCARD_PACKET, SCTP_NULL());
 
@@ -1125,6 +1142,7 @@ static sctp_disposition_t sctp_sf_do_unexpected_init(
 			if (packet) {
 				sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
 						SCTP_PACKET(packet));
+				SCTP_INC_STATS(SctpOutCtrlChunks);
 				retval = SCTP_DISPOSITION_CONSUME;
 			} else {
 				retval = SCTP_DISPOSITION_NOMEM;
@@ -1355,7 +1373,7 @@ static sctp_disposition_t sctp_sf_do_dupcook_a(const sctp_endpoint_t *ep,
 					       sctp_association_t *new_asoc)
 {
 	sctp_init_chunk_t *peer_init;
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 	sctp_chunk_t *repl;
 
 	/* new_asoc is a brand-new association, so these are not yet
@@ -1421,7 +1439,7 @@ static sctp_disposition_t sctp_sf_do_dupcook_b(const sctp_endpoint_t *ep,
 					       sctp_association_t *new_asoc)
 {
 	sctp_init_chunk_t *peer_init;
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 	sctp_chunk_t *repl;
 
 	/* new_asoc is a brand-new association, so these are not yet
@@ -1436,6 +1454,7 @@ static sctp_disposition_t sctp_sf_do_dupcook_b(const sctp_endpoint_t *ep,
 	sctp_add_cmd_sf(commands, SCTP_CMD_UPDATE_ASSOC, SCTP_ASOC(new_asoc));
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_ESTABLISHED));
+	SCTP_INC_STATS(SctpCurrEstab);
 	sctp_add_cmd_sf(commands, SCTP_CMD_HB_TIMERS_START, SCTP_NULL());
 
 	repl = sctp_make_cookie_ack(new_asoc, chunk);
@@ -1503,7 +1522,7 @@ static sctp_disposition_t sctp_sf_do_dupcook_d(const sctp_endpoint_t *ep,
 					       sctp_cmd_seq_t *commands,
 					       sctp_association_t *new_asoc)
 {
-	sctp_ulpevent_t *ev = NULL;
+	struct sctp_ulpevent *ev = NULL;
 	sctp_chunk_t *repl;
 
 	/* Clarification from Implementor's Guide:
@@ -1519,6 +1538,7 @@ static sctp_disposition_t sctp_sf_do_dupcook_d(const sctp_endpoint_t *ep,
 				SCTP_TO(SCTP_EVENT_TIMEOUT_T1_COOKIE));
 		sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 				SCTP_STATE(SCTP_STATE_ESTABLISHED));
+		SCTP_INC_STATS(SctpCurrEstab);
 		sctp_add_cmd_sf(commands, SCTP_CMD_HB_TIMERS_START,
 				SCTP_NULL());
 
@@ -1540,11 +1560,11 @@ static sctp_disposition_t sctp_sf_do_dupcook_d(const sctp_endpoint_t *ep,
 				SCTP_ULPEVENT(ev));
 	}
 	sctp_add_cmd_sf(commands, SCTP_CMD_TRANSMIT, SCTP_NULL());
-	
+
 	repl = sctp_make_cookie_ack(new_asoc, chunk);
 	if (!repl)
 		goto nomem;
-	
+
 	sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(repl));
 	sctp_add_cmd_sf(commands, SCTP_CMD_TRANSMIT, SCTP_NULL());
 
@@ -1925,6 +1945,8 @@ sctp_disposition_t sctp_sf_do_9_1_abort(const sctp_endpoint_t *ep,
 
  	/* ASSOC_FAILED will DELETE_TCB. */
 	sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+	SCTP_INC_STATS(SctpAborteds);
+	SCTP_DEC_STATS(SctpCurrEstab);
 
 	/* BUG?  This does not look complete... */
 	return SCTP_DISPOSITION_ABORT;
@@ -1948,6 +1970,7 @@ sctp_disposition_t sctp_sf_cookie_wait_abort(const sctp_endpoint_t *ep,
 
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_CLOSED));
+	SCTP_INC_STATS(SctpAborteds);
 	sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_STOP,
 			SCTP_TO(SCTP_EVENT_TIMEOUT_T1_INIT));
 
@@ -2241,6 +2264,7 @@ sctp_disposition_t sctp_sf_eat_data_6_2(const sctp_endpoint_t *ep,
 	sctp_datahdr_t *data_hdr;
 	sctp_chunk_t *err;
 	size_t datalen;
+	sctp_verb_t deliver;
 	int tmp;
 	__u32 tsn;
 
@@ -2307,10 +2331,32 @@ sctp_disposition_t sctp_sf_eat_data_6_2(const sctp_endpoint_t *ep,
 	datalen = ntohs(chunk->chunk_hdr->length);
 	datalen -= sizeof(sctp_data_chunk_t);
 
+	deliver = SCTP_CMD_CHUNK_ULP;
+
+	/* Think about partial delivery. */
+	if ((datalen >= asoc->rwnd) && (!asoc->ulpq.pd_mode)) {
+
+		/* Even if we don't accept this chunk there is
+		 * memory pressure.
+		 */
+		sctp_add_cmd_sf(commands, SCTP_CMD_CHUNK_PD, SCTP_NULL());
+	}
+
 	if (asoc->rwnd_over || (datalen > asoc->rwnd + asoc->frag_point)) {
-		SCTP_DEBUG_PRINTK("Discarding tsn: %u datalen: %Zd, "
-				  "rwnd: %d\n", tsn, datalen, asoc->rwnd);
-		goto discard_force;
+	
+
+		/* There is absolutely no room, but this is the most
+		 * important tsn that we are waiting on, try to 
+		 * to partial deliver or renege to make room. 
+		 */
+		if ((sctp_tsnmap_get_ctsn(&asoc->peer.tsn_map) + 1) == tsn) {
+			deliver = SCTP_CMD_CHUNK_PD;
+		} else {
+			SCTP_DEBUG_PRINTK("Discard tsn: %u len: %Zd, "
+					  "rwnd: %d\n", tsn, datalen, 
+					  asoc->rwnd);
+			goto discard_force;
+		}
 	}
 
 	/*
@@ -2332,13 +2378,22 @@ sctp_disposition_t sctp_sf_eat_data_6_2(const sctp_endpoint_t *ep,
 		 */
 		sctp_add_cmd_sf(commands, SCTP_CMD_DISCARD_PACKET,SCTP_NULL());
 		sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+		SCTP_INC_STATS(SctpAborteds);
+		SCTP_INC_STATS(SctpCurrEstab);
 		return SCTP_DISPOSITION_CONSUME;
 	}
 
-	/* We are accepting this DATA chunk.  */
+	/* If definately accepting the DATA chunk, record its TSN, otherwise
+	 * wait for renege processing. 
+	 */
+	if (deliver != SCTP_CMD_CHUNK_PD) {
+		sctp_add_cmd_sf(commands, SCTP_CMD_REPORT_TSN, SCTP_U32(tsn));
 
-	/* Record the fact that we have received this TSN.  */
-	sctp_add_cmd_sf(commands, SCTP_CMD_REPORT_TSN, SCTP_U32(tsn));
+		if (chunk->chunk_hdr->flags & SCTP_DATA_UNORDERED)
+			SCTP_INC_STATS(SctpInUnorderChunks);
+		else
+			SCTP_INC_STATS(SctpInOrderChunks);
+	}
 
 	/* RFC 2960 6.5 Stream Identifier and Stream Sequence Number
 	 *
@@ -2352,10 +2407,9 @@ sctp_disposition_t sctp_sf_eat_data_6_2(const sctp_endpoint_t *ep,
 		err = sctp_make_op_error(asoc, chunk, SCTP_ERROR_INV_STRM,
 					 &data_hdr->stream,
 					 sizeof(data_hdr->stream));
-		if (err) {
+		if (err)
 			sctp_add_cmd_sf(commands, SCTP_CMD_REPLY,
 					SCTP_CHUNK(err));
-		}
 		goto discard_noforce;
 	}
 
@@ -2363,7 +2417,8 @@ sctp_disposition_t sctp_sf_eat_data_6_2(const sctp_endpoint_t *ep,
 	 * SCTP_CMD_CHUNK_ULP cmd before the SCTP_CMD_GEN_SACK, as the SACK
 	 * chunk needs the updated rwnd.
 	 */
-	sctp_add_cmd_sf(commands, SCTP_CMD_CHUNK_ULP, SCTP_CHUNK(chunk));
+	sctp_add_cmd_sf(commands, deliver, SCTP_CHUNK(chunk));
+
 	if (asoc->autoclose) {
 		sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_RESTART,
 				SCTP_TO(SCTP_EVENT_TIMEOUT_AUTOCLOSE));
@@ -2536,6 +2591,8 @@ sctp_disposition_t sctp_sf_eat_data_fast_4_4(const sctp_endpoint_t *ep,
 		 */
 		sctp_add_cmd_sf(commands, SCTP_CMD_DISCARD_PACKET,SCTP_NULL());
 		sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+		SCTP_INC_STATS(SctpAborteds);
+		SCTP_INC_STATS(SctpCurrEstab);
 		return SCTP_DISPOSITION_CONSUME;
 	}
 
@@ -2544,6 +2601,11 @@ sctp_disposition_t sctp_sf_eat_data_fast_4_4(const sctp_endpoint_t *ep,
 	/* Record the fact that we have received this TSN.  */
 	sctp_add_cmd_sf(commands, SCTP_CMD_REPORT_TSN, SCTP_U32(tsn));
 
+	if (chunk->chunk_hdr->flags & SCTP_DATA_UNORDERED)
+		SCTP_INC_STATS(SctpInUnorderChunks);
+	else
+		SCTP_INC_STATS(SctpInOrderChunks);
+
 	/* RFC 2960 6.5 Stream Identifier and Stream Sequence Number
 	 *
 	 * If an endpoint receive a DATA chunk with an invalid stream
@@ -2705,6 +2767,8 @@ sctp_disposition_t sctp_sf_tabort_8_4_8(const sctp_endpoint_t *ep,
 		sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
 				SCTP_PACKET(packet));
 
+		SCTP_INC_STATS(SctpOutCtrlChunks);
+
 		return SCTP_DISPOSITION_CONSUME;
 	}
 
@@ -2726,7 +2790,7 @@ sctp_disposition_t sctp_sf_operr_notify(const sctp_endpoint_t *ep,
 					sctp_cmd_seq_t *commands)
 {
 	sctp_chunk_t *chunk = arg;
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 
 	while (chunk->chunk_end > chunk->skb->data) {
 		ev = sctp_ulpevent_make_remote_error(asoc, chunk, 0,
@@ -2764,7 +2828,7 @@ sctp_disposition_t sctp_sf_do_9_2_final(const sctp_endpoint_t *ep,
 {
 	sctp_chunk_t *chunk = arg;
 	sctp_chunk_t *reply;
-	sctp_ulpevent_t *ev;
+	struct sctp_ulpevent *ev;
 
 	/* 10.2 H) SHUTDOWN COMPLETE notification
 	 *
@@ -2794,6 +2858,8 @@ sctp_disposition_t sctp_sf_do_9_2_final(const sctp_endpoint_t *ep,
 
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_CLOSED));
+	SCTP_INC_STATS(SctpShutdowns);
+	SCTP_DEC_STATS(SctpCurrEstab);
 	sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(reply));
 
 	/* ...and remove all record of the association. */
@@ -2834,6 +2900,8 @@ sctp_disposition_t sctp_sf_ootb(const sctp_endpoint_t *ep,
 	__u8 *ch_end;
 	int ootb_shut_ack = 0;
 
+	SCTP_INC_STATS(SctpOutOfBlues);
+
 	ch = (sctp_chunkhdr_t *) chunk->chunk_hdr;
 	do {
 		ch_end = ((__u8 *)ch) + WORD_ROUND(ntohs(ch->length));
@@ -2901,6 +2969,8 @@ sctp_disposition_t sctp_sf_shut_8_4_5(const sctp_endpoint_t *ep,
 		sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
 				SCTP_PACKET(packet));
 
+		SCTP_INC_STATS(SctpOutCtrlChunks);
+
 		return SCTP_DISPOSITION_CONSUME;
 	}
 
@@ -3472,6 +3542,10 @@ sctp_disposition_t sctp_sf_do_9_1_prm_abort(const sctp_endpoint_t *ep,
 
 	/* Delete the established association. */
 	sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+
+	SCTP_INC_STATS(SctpAborteds);
+	SCTP_DEC_STATS(SctpCurrEstab);
+
 	return retval;
 }
 
@@ -3527,6 +3601,8 @@ sctp_disposition_t sctp_sf_cookie_wait_prm_shutdown(
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_CLOSED));
 
+	SCTP_INC_STATS(SctpShutdowns);
+
 	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
 
 	return SCTP_DISPOSITION_DELETE_TCB;
@@ -3597,6 +3673,8 @@ sctp_disposition_t sctp_sf_cookie_wait_prm_abort(const sctp_endpoint_t *ep,
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_CLOSED));
 
+	SCTP_INC_STATS(SctpAborteds);
+
 	/* Even if we can't send the ABORT due to low memory delete the
 	 * TCB.  This is a departure from our typical NOMEM handling.
 	 */
@@ -3929,6 +4007,8 @@ sctp_disposition_t sctp_sf_do_6_3_3_rtx(const sctp_endpoint_t *ep,
 	if (asoc->overall_error_count >= asoc->overall_error_threshold) {
 		/* CMD_ASSOC_FAILED calls CMD_DELETE_TCB. */
 		sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+		SCTP_INC_STATS(SctpAborteds);
+		SCTP_DEC_STATS(SctpCurrEstab);
 		return SCTP_DISPOSITION_DELETE_TCB;
 	}
 
@@ -4096,6 +4176,8 @@ sctp_disposition_t sctp_sf_t2_timer_expire(const sctp_endpoint_t *ep,
 	if (asoc->overall_error_count >= asoc->overall_error_threshold) {
 		/* Note:  CMD_ASSOC_FAILED calls CMD_DELETE_TCB. */
 		sctp_add_cmd_sf(commands, SCTP_CMD_ASSOC_FAILED, SCTP_NULL());
+		SCTP_INC_STATS(SctpAborteds);
+		SCTP_DEC_STATS(SctpCurrEstab);
 		return SCTP_DISPOSITION_DELETE_TCB;
 	}
 
@@ -4271,6 +4353,9 @@ sctp_sackhdr_t *sctp_sm_pull_sack(sctp_chunk_t *chunk)
 	__u16 num_blocks;
 	__u16 num_dup_tsns;
 
+	/* FIXME:  Protect ourselves from reading too far into
+	 * the skb from a bogus sender.
+	 */
 	sack = (sctp_sackhdr_t *) chunk->skb->data;
 	skb_pull(chunk->skb, sizeof(sctp_sackhdr_t));
 
@@ -4401,6 +4486,7 @@ void sctp_send_stale_cookie_err(const sctp_endpoint_t *ep,
 			sctp_packet_append_chunk(packet, err_chunk);
 			sctp_add_cmd_sf(commands, SCTP_CMD_SEND_PKT,
 					SCTP_PACKET(packet));
+			SCTP_INC_STATS(SctpOutCtrlChunks);
 		} else
 			sctp_free_chunk (err_chunk);
 	}

net/sctp/socket.c

@@ -81,13 +81,13 @@
 
 /* Forward declarations for internal helper functions. */
 static int sctp_writeable(struct sock *sk);
-static inline int sctp_wspace(sctp_association_t *asoc);
+static inline int sctp_wspace(struct sctp_association *asoc);
 static inline void sctp_set_owner_w(sctp_chunk_t *chunk);
 static void sctp_wfree(struct sk_buff *skb);
-static int sctp_wait_for_sndbuf(sctp_association_t *asoc, long *timeo_p,
+static int sctp_wait_for_sndbuf(struct sctp_association *, long *timeo_p,
 				int msg_len);
 static int sctp_wait_for_packet(struct sock * sk, int *err, long *timeo_p);
-static int sctp_wait_for_connect(sctp_association_t *asoc, long *timeo_p);
+static int sctp_wait_for_connect(struct sctp_association *, long *timeo_p);
 static inline int sctp_verify_addr(struct sock *, union sctp_addr *, int);
 static int sctp_bindx_add(struct sock *, struct sockaddr_storage *, int);
 static int sctp_bindx_rem(struct sock *, struct sockaddr_storage *, int);
@@ -158,7 +158,7 @@ static struct sctp_af *sctp_sockaddr_af(struct sctp_opt *opt,
 /* Bind a local address either to an endpoint or to an association.  */
 SCTP_STATIC int sctp_do_bind(struct sock *sk, union sctp_addr *addr, int len)
 {
-	sctp_opt_t *sp = sctp_sk(sk);
+	struct sctp_opt *sp = sctp_sk(sk);
 	sctp_endpoint_t *ep = sp->ep;
 	sctp_bind_addr_t *bp = &ep->base.bind_addr;
 	struct sctp_af *af;
@@ -454,7 +454,7 @@ int sctp_bindx_add(struct sock *sk, struct sockaddr_storage *addrs, int addrcnt)
  */
 int sctp_bindx_rem(struct sock *sk, struct sockaddr_storage *addrs, int addrcnt)
 {
-	sctp_opt_t *sp = sctp_sk(sk);
+	struct sctp_opt *sp = sctp_sk(sk);
 	sctp_endpoint_t *ep = sp->ep;
 	int cnt;
 	sctp_bind_addr_t *bp = &ep->base.bind_addr;
@@ -662,6 +662,7 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout)
 
 	/* Clean up any skbs sitting on the receive queue.  */
 	skb_queue_purge(&sk->receive_queue);
+	skb_queue_purge(&sctp_sk(sk)->pd_lobby);
 
 	/* This will run the backlog queue.  */
 	sctp_release_sock(sk);
@@ -714,7 +715,7 @@ SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *, sctp_cmsgs_t *);
 SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 			     struct msghdr *msg, int msg_len)
 {
-	sctp_opt_t *sp;
+	struct sctp_opt *sp;
 	sctp_endpoint_t *ep;
 	sctp_association_t *new_asoc=NULL, *asoc=NULL;
 	struct sctp_transport *transport;
@@ -939,6 +940,19 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 	/* ASSERT: we have a valid association at this point.  */
 	SCTP_DEBUG_PRINTK("We have a valid association.\n");
 
+	if (!sinfo) {
+		/* If the user didn't specify SNDRCVINFO, make up one with
+		 * some defaults.
+		 */
+		default_sinfo.sinfo_stream = asoc->defaults.stream;
+		default_sinfo.sinfo_flags = asoc->defaults.flags;
+		default_sinfo.sinfo_ppid = asoc->defaults.ppid;
+		default_sinfo.sinfo_context = asoc->defaults.context;
+		default_sinfo.sinfo_timetolive = asoc->defaults.timetolive;
+		default_sinfo.sinfo_assoc_id = sctp_assoc2id(asoc);
+		sinfo = &default_sinfo;
+	}
+
 	/* API 7.1.7, the sndbuf size per association bounds the
 	 * maximum size of data that can be sent in a single send call.
 	 */
@@ -963,13 +977,6 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 			err = -EINVAL;
 			goto out_free;
 		}
-	} else {
-		/* If the user didn't specify SNDRCVINFO, make up one with
-		 * some defaults.
-		 */
-		default_sinfo.sinfo_stream = asoc->defaults.stream;
-		default_sinfo.sinfo_ppid = asoc->defaults.ppid;
-		sinfo = &default_sinfo;
 	}
 
 	timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
@@ -979,21 +986,6 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 			goto out_free;
 	}
 
-#if 0
-	/* FIXME: This looks wrong so I'll comment out.
-	 * We should be able to use this same technique for
-	 * primary address override!  --jgrimm
-	 */
-	/* If the user gave us an address, copy it in.  */
-	if (msg->msg_name) {
-		chunk->transport = sctp_assoc_lookup_paddr(asoc, &to);
-		if (!chunk->transport) {
-			err = -EINVAL;
-			goto out_free;
-		}
-	}
-#endif /* 0 */
-
 	/* Break the message into multiple chunks of maximum size. */
 	skb_queue_head_init(&chunks);
 	err = sctp_datachunks_from_user(asoc, sinfo, msg, msg_len, &chunks);
@@ -1013,6 +1005,23 @@ SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
 
 		/* Do accounting for the write space.  */
 		sctp_set_owner_w(chunk);
+
+		/* This flag, in the UDP model, requests the SCTP stack to
+		 * override the primary destination address with the 
+		 * address found with the sendto/sendmsg call.
+		 */
+		if (sinfo_flags & MSG_ADDR_OVER) {
+			if (!msg->msg_name) {
+				err = -EINVAL;
+				goto out_free;
+			}
+			chunk->transport = sctp_assoc_lookup_paddr(asoc, &to);
+			if (!chunk->transport) {
+				err = -EINVAL;
+				goto out_free;
+			}
+		}
+
 		/* Send it to the lower layers.  */
 		sctp_primitive_SEND(asoc, chunk);
 		SCTP_DEBUG_PRINTK("We sent primitively.\n");
@@ -1110,8 +1119,8 @@ static struct sk_buff *sctp_skb_recv_datagram(struct sock *, int, int, int *);
 SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 			     int len, int noblock, int flags, int *addr_len)
 {
-	sctp_ulpevent_t *event = NULL;
-	sctp_opt_t *sp = sctp_sk(sk);
+	struct sctp_ulpevent *event = NULL;
+	struct sctp_opt *sp = sctp_sk(sk);
 	struct sk_buff *skb;
 	int copied;
 	int err = 0;
@@ -1143,7 +1152,7 @@ SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr
 
 	err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 
-	event = (sctp_ulpevent_t *) skb->cb;
+	event = sctp_skb2event(skb);
 
 	if (err)
 		goto out_free;
@@ -1170,7 +1179,6 @@ SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr
 	/* If skb's length exceeds the user's buffer, update the skb and
 	 * push it back to the receive_queue so that the next call to
 	 * recvmsg() will return the remaining data. Don't set MSG_EOR.
-	 * Otherwise, set MSG_EOR indicating the end of a message.
 	 */
 	if (skb_len > copied) {
 		msg->msg_flags &= ~MSG_EOR;
@@ -1178,6 +1186,7 @@ SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr
 			goto out_free;
 		sctp_skb_pull(skb, copied);
 		skb_queue_head(&sk->receive_queue, skb);
+
 		/* When only partial message is copied to the user, increase
 		 * rwnd by that amount. If all the data in the skb is read,
 		 * rwnd is updated when the skb's destructor is called via
@@ -1185,9 +1194,11 @@ SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr
 		 */
 		sctp_assoc_rwnd_increase(event->asoc, copied);
 		goto out;
-	} else {
-		 msg->msg_flags |= MSG_EOR;
-	}
+	} else if ((event->msg_flags & MSG_NOTIFICATION) ||
+		   (event->msg_flags & MSG_EOR))
+		msg->msg_flags |= MSG_EOR;
+	else
+		msg->msg_flags &= ~MSG_EOR;
 
 out_free:
 	sctp_ulpevent_free(event); /* Free the skb. */
@@ -1225,7 +1236,7 @@ static inline int sctp_setsockopt_set_events(struct sock *sk, char *optval,
 static inline int sctp_setsockopt_autoclose(struct sock *sk, char *optval,
 					    int optlen)
 {
-	sctp_opt_t *sp = sctp_sk(sk);
+	struct sctp_opt *sp = sctp_sk(sk);
 
 	/* Applicable to UDP-style socket only */
 	if (SCTP_SOCKET_TCP == sp->type)
@@ -1310,6 +1321,44 @@ static inline int sctp_setsockopt_initmsg(struct sock *sk, char *optval,
 	return 0;
 }
 
+/*
+ *
+ * 7.1.15 Set default send parameters (SET_DEFAULT_SEND_PARAM)
+ *
+ *   Applications that wish to use the sendto() system call may wish to
+ *   specify a default set of parameters that would normally be supplied
+ *   through the inclusion of ancillary data.  This socket option allows
+ *   such an application to set the default sctp_sndrcvinfo structure.
+ *   The application that wishes to use this socket option simply passes
+ *   in to this call the sctp_sndrcvinfo structure defined in Section
+ *   5.2.2) The input parameters accepted by this call include
+ *   sinfo_stream, sinfo_flags, sinfo_ppid, sinfo_context,
+ *   sinfo_timetolive.  The user must provide the sinfo_assoc_id field in
+ *   to this call if the caller is using the UDP model.
+ */
+static inline int sctp_setsockopt_set_default_send_param(struct sock *sk,
+						char *optval, int optlen)
+{
+	struct sctp_sndrcvinfo info;
+	sctp_association_t *asoc;
+
+	if (optlen != sizeof(struct sctp_sndrcvinfo))
+		return -EINVAL;
+	if (copy_from_user(&info, optval, optlen))
+		return -EFAULT;
+
+	asoc = sctp_id2assoc(sk, info.sinfo_assoc_id);
+	if (!asoc)
+		return -EINVAL;
+
+	asoc->defaults.stream = info.sinfo_stream;
+	asoc->defaults.flags = info.sinfo_flags;
+	asoc->defaults.ppid = info.sinfo_ppid;
+	asoc->defaults.context = info.sinfo_context;
+	asoc->defaults.timetolive = info.sinfo_timetolive;
+	return 0;
+}
+
 /* API 6.2 setsockopt(), getsockopt()
  *
  * Applications use setsockopt() and getsockopt() to set or retrieve
@@ -1401,6 +1450,11 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 		retval = sctp_setsockopt_initmsg(sk, optval, optlen);
 		break;
 
+	case SCTP_SET_DEFAULT_SEND_PARAM:
+		retval = sctp_setsockopt_set_default_send_param(sk,
+							optval, optlen);
+		break;
+
 	default:
 		retval = -ENOPROTOOPT;
 		break;
@@ -1432,7 +1486,7 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 SCTP_STATIC int sctp_connect(struct sock *sk, struct sockaddr *uaddr,
 			     int addr_len)
 {
-	sctp_opt_t *sp;
+	struct sctp_opt *sp;
 	sctp_endpoint_t *ep;
 	sctp_association_t *asoc;
 	struct sctp_transport *transport;
@@ -1554,7 +1608,7 @@ SCTP_STATIC int sctp_init_sock(struct sock *sk)
 {
 	sctp_endpoint_t *ep;
 	sctp_protocol_t *proto;
-	sctp_opt_t *sp;
+	struct sctp_opt *sp;
 
 	SCTP_DEBUG_PRINTK("sctp_init_sock(sk: %p)\n", sk);
 
@@ -1583,7 +1637,7 @@ SCTP_STATIC int sctp_init_sock(struct sock *sk)
 
 	/* Initialize default RTO related parameters.  These parameters can
 	 * be modified for with the SCTP_RTOINFO socket option.
-	 * FIXME: This are not used yet.
+	 * FIXME: These are not used yet.
 	 */
 	sp->rtoinfo.srto_initial = proto->rto_initial;
 	sp->rtoinfo.srto_max     = proto->rto_max;
@@ -1620,6 +1674,11 @@ SCTP_STATIC int sctp_init_sock(struct sock *sk)
 	 */
 	sp->autoclose         = 0;
 	sp->pf = sctp_get_pf_specific(sk->family);
+
+	/* Control variables for partial data delivery. */
+	sp->pd_mode           = 0;
+	skb_queue_head_init(&sp->pd_lobby);
+
 	/* Create a per socket endpoint structure.  Even if we
 	 * change the data structure relationships, this may still
 	 * be useful for storing pre-connect address information.
@@ -1774,10 +1833,10 @@ SCTP_STATIC int sctp_do_peeloff(sctp_association_t *assoc, struct socket **newso
 	struct sock *newsk;
 	struct socket *tmpsock;
 	sctp_endpoint_t *newep;
-	sctp_opt_t *oldsp = sctp_sk(oldsk);
-	sctp_opt_t *newsp;
+	struct sctp_opt *oldsp = sctp_sk(oldsk);
+	struct sctp_opt *newsp;
 	struct sk_buff *skb, *tmp;
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	int err = 0;
 
 	/* An association cannot be branched off from an already peeled-off
@@ -1811,13 +1870,50 @@ SCTP_STATIC int sctp_do_peeloff(sctp_association_t *assoc, struct socket **newso
 	 * peeled off association to the new socket's receive queue.
 	 */
 	sctp_skb_for_each(skb, &oldsk->receive_queue, tmp) {
-		event = (sctp_ulpevent_t *)skb->cb;
+		event = sctp_skb2event(skb);
 		if (event->asoc == assoc) {
 			__skb_unlink(skb, skb->list);
 			__skb_queue_tail(&newsk->receive_queue, skb);
 		}
 	}
 
+	/* Clean up an messages pending delivery due to partial
+	 * delivery.   Three cases:
+	 * 1) No partial deliver;  no work.
+	 * 2) Peeling off partial delivery; keep pd_lobby in new pd_lobby.
+	 * 3) Peeling off non-partial delivery; move pd_lobby to recieve_queue.
+	 */
+	skb_queue_head_init(&newsp->pd_lobby);
+	sctp_sk(newsk)->pd_mode = assoc->ulpq.pd_mode;;
+
+	if (sctp_sk(oldsk)->pd_mode) {
+		struct sk_buff_head *queue;
+
+		/* Decide which queue to move pd_lobby skbs to. */
+		if (assoc->ulpq.pd_mode) {
+			queue = &newsp->pd_lobby;
+		} else
+			queue = &newsk->receive_queue;
+
+		/* Walk through the pd_lobby, looking for skbs that
+		 * need moved to the new socket.
+		 */
+		sctp_skb_for_each(skb, &oldsp->pd_lobby, tmp) {
+			event = sctp_skb2event(skb);
+			if (event->asoc == assoc) {
+				__skb_unlink(skb, skb->list);
+				__skb_queue_tail(queue, skb);
+			}
+		}
+
+		/* Clear up any skbs waiting for the partial
+		 * delivery to finish.
+		 */
+		if (assoc->ulpq.pd_mode)
+			sctp_clear_pd(oldsk);
+
+	}
+
 	/* Set the type of socket to indicate that it is peeled off from the
 	 * original socket.
 	 */
@@ -2389,7 +2485,7 @@ static int sctp_get_port(struct sock *sk, unsigned short snum)
  */
 SCTP_STATIC int sctp_seqpacket_listen(struct sock *sk, int backlog)
 {
-	sctp_opt_t *sp = sctp_sk(sk);
+	struct sctp_opt *sp = sctp_sk(sk);
 	sctp_endpoint_t *ep = sp->ep;
 
 	/* Only UDP style sockets that are not peeled off are allowed to

net/sctp/tsnmap.c

@@ -3,40 +3,40 @@
  * 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
- * 
+ *
  * These functions manipulate sctp tsn mapping array.
- * 
- * 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 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>
  *    Jon Grimm             <jgrimm@us.ibm.com>
  *    Karl Knutson          <karl@athena.chicago.il.us>
- * 
+ *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
  */
@@ -45,21 +45,21 @@
 #include <net/sctp/sctp.h>
 #include <net/sctp/sm.h>
 
-static void _sctp_tsnmap_update(sctp_tsnmap_t *map);
-static void _sctp_tsnmap_update_pending_data(sctp_tsnmap_t *map);
-static void _sctp_tsnmap_find_gap_ack(__u8 *map, __u16 off,
-				      __u16 len, __u16 base,
-				      int *started, __u16 *start,
-				      int *ended, __u16 *end);
+static void sctp_tsnmap_update(struct sctp_tsnmap *map);
+static void sctp_tsnmap_update_pending_data(struct sctp_tsnmap *map);
+static void sctp_tsnmap_find_gap_ack(__u8 *map, __u16 off,
+				     __u16 len, __u16 base,
+				     int *started, __u16 *start,
+				     int *ended, __u16 *end);
 
 /* Create a new sctp_tsnmap.
  * Allocate room to store at least 'len' contiguous TSNs.
  */
-sctp_tsnmap_t *sctp_tsnmap_new(__u16 len, __u32 initial_tsn, int priority)
+struct sctp_tsnmap *sctp_tsnmap_new(__u16 len, __u32 initial_tsn, int priority)
 {
-	sctp_tsnmap_t *retval;
+	struct sctp_tsnmap *retval;
 
-	retval = kmalloc(sizeof(sctp_tsnmap_t) +
+	retval = kmalloc(sizeof(struct sctp_tsnmap) +
 			 sctp_tsnmap_storage_size(len),
 			 priority);
 	if (!retval)
@@ -72,13 +72,13 @@ sctp_tsnmap_t *sctp_tsnmap_new(__u16 len, __u32 initial_tsn, int priority)
 
 fail_map:
 	kfree(retval);
-
 fail:
 	return NULL;
 }
 
 /* Initialize a block of memory as a tsnmap.  */
-sctp_tsnmap_t *sctp_tsnmap_init(sctp_tsnmap_t *map, __u16 len, __u32 initial_tsn)
+struct sctp_tsnmap *sctp_tsnmap_init(struct sctp_tsnmap *map, __u16 len,
+				     __u32 initial_tsn)
 {
 	map->tsn_map = map->raw_map;
 	map->overflow_map = map->tsn_map + len;
@@ -94,6 +94,7 @@ sctp_tsnmap_t *sctp_tsnmap_init(sctp_tsnmap_t *map, __u16 len, __u32 initial_tsn
 	map->max_tsn_seen = map->cumulative_tsn_ack_point;
 	map->malloced = 0;
 	map->pending_data = 0;
+	map->num_dup_tsns = 0;
 
 	return map;
 }
@@ -104,7 +105,7 @@ sctp_tsnmap_t *sctp_tsnmap_init(sctp_tsnmap_t *map, __u16 len, __u32 initial_tsn
  *  >0 if the TSN has been seen (duplicate)
  *  <0 if the TSN is invalid (too large to track)
  */
-int sctp_tsnmap_check(const sctp_tsnmap_t *map, __u32 tsn)
+int sctp_tsnmap_check(const struct sctp_tsnmap *map, __u32 tsn)
 {
 	__s32 gap;
 	int dup;
@@ -136,7 +137,7 @@ int sctp_tsnmap_check(const sctp_tsnmap_t *map, __u32 tsn)
 }
 
 /* Is there a gap in the TSN map?  */
-int sctp_tsnmap_has_gap(const sctp_tsnmap_t *map)
+int sctp_tsnmap_has_gap(const struct sctp_tsnmap *map)
 {
 	int has_gap;
 
@@ -145,7 +146,7 @@ int sctp_tsnmap_has_gap(const sctp_tsnmap_t *map)
 }
 
 /* Mark this TSN as seen.  */
-void sctp_tsnmap_mark(sctp_tsnmap_t *map, __u32 tsn)
+void sctp_tsnmap_mark(struct sctp_tsnmap *map, __u32 tsn)
 {
 	__s32 gap;
 
@@ -173,40 +174,45 @@ void sctp_tsnmap_mark(sctp_tsnmap_t *map, __u32 tsn)
 	/* Go fixup any internal TSN mapping variables including
 	 * cumulative_tsn_ack_point.
 	 */
-	_sctp_tsnmap_update(map);
+	sctp_tsnmap_update(map);
+}
+
+void sctp_tsnmap_report_dup(struct sctp_tsnmap *map, __u32 tsn)
+{
 }
 
 /* Retrieve the Cumulative TSN Ack Point. */
-__u32 sctp_tsnmap_get_ctsn(const sctp_tsnmap_t *map)
+__u32 sctp_tsnmap_get_ctsn(const struct sctp_tsnmap *map)
 {
 	return map->cumulative_tsn_ack_point;
 }
 
 /* Retrieve the highest TSN we've seen.  */
-__u32 sctp_tsnmap_get_max_tsn_seen(const sctp_tsnmap_t *map)
+__u32 sctp_tsnmap_get_max_tsn_seen(const struct sctp_tsnmap *map)
 {
 	return map->max_tsn_seen;
 }
 
 /* Dispose of a tsnmap.  */
-void sctp_tsnmap_free(sctp_tsnmap_t *map)
+void sctp_tsnmap_free(struct sctp_tsnmap *map)
 {
 	if (map->malloced)
 		kfree(map);
 }
 
 /* Initialize a Gap Ack Block iterator from memory being provided.  */
-void sctp_tsnmap_iter_init(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter)
+void sctp_tsnmap_iter_init(const struct sctp_tsnmap *map,
+			   struct sctp_tsnmap_iter *iter)
 {
 	/* Only start looking one past the Cumulative TSN Ack Point.  */
 	iter->start = map->cumulative_tsn_ack_point + 1;
 }
 
-/* Get the next Gap Ack Blocks. Returns 0 if there was not
- * another block to get.
+/* Get the next Gap Ack Blocks. Returns 0 if there was not another block
+ * to get.
  */
-int sctp_tsnmap_next_gap_ack(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter,
-			     __u16 *start, __u16 *end)
+int sctp_tsnmap_next_gap_ack(const struct sctp_tsnmap *map,
+	struct sctp_tsnmap_iter *iter, __u16 *start, __u16 *end)
 {
 	int started, ended;
 	__u16 _start, _end, offset;
@@ -216,12 +222,10 @@ int sctp_tsnmap_next_gap_ack(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter,
 
 	/* Search the first mapping array.  */
 	if (iter->start - map->base_tsn < map->len) {
+
 		offset = iter->start - map->base_tsn;
-		_sctp_tsnmap_find_gap_ack(map->tsn_map,
-					  offset,
-					  map->len, 0,
-					  &started, &_start,
-					  &ended, &_end);
+		sctp_tsnmap_find_gap_ack(map->tsn_map, offset, map->len, 0,
+					 &started, &_start, &ended, &_end);
 	}
 
 	/* Do we need to check the overflow map? */
@@ -235,12 +239,12 @@ int sctp_tsnmap_next_gap_ack(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter,
 			offset = iter->start - map->base_tsn - map->len;
 
 		/* Search the overflow map.  */
-		_sctp_tsnmap_find_gap_ack(map->overflow_map,
-					  offset,
-					  map->len,
-					  map->len,
-					  &started, &_start,
-					  &ended, &_end);
+		sctp_tsnmap_find_gap_ack(map->overflow_map,
+					 offset,
+					 map->len,
+					 map->len,
+					 &started, &_start,
+					 &ended, &_end);
 	}
 
 	/* The Gap Ack Block happens to end at the end of the
@@ -278,7 +282,7 @@ int sctp_tsnmap_next_gap_ack(const sctp_tsnmap_t *map, sctp_tsnmap_iter_t *iter,
 /* This private helper function updates the tsnmap buffers and
  * the Cumulative TSN Ack Point.
  */
-static void _sctp_tsnmap_update(sctp_tsnmap_t *map)
+static void sctp_tsnmap_update(struct sctp_tsnmap *map)
 {
 	__u32 ctsn;
 
@@ -301,10 +305,10 @@ static void _sctp_tsnmap_update(sctp_tsnmap_t *map)
 	} while (map->tsn_map[ctsn - map->base_tsn]);
 
 	map->cumulative_tsn_ack_point = ctsn - 1; /* Back up one. */
-	_sctp_tsnmap_update_pending_data(map);
+	sctp_tsnmap_update_pending_data(map);
 }
 
-static void _sctp_tsnmap_update_pending_data(sctp_tsnmap_t *map)
+static void sctp_tsnmap_update_pending_data(struct sctp_tsnmap *map)
 {
 	__u32 cum_tsn = map->cumulative_tsn_ack_point;
 	__u32 max_tsn = map->max_tsn_seen;
@@ -324,7 +328,7 @@ static void _sctp_tsnmap_update_pending_data(sctp_tsnmap_t *map)
 	for (i = start; i < end; i++) {
 		if (map->tsn_map[i])
 			pending_data--;
-	} 
+	}
 
 	if (gap >= map->len) {
 		start = 0;
@@ -345,14 +349,14 @@ static void _sctp_tsnmap_update_pending_data(sctp_tsnmap_t *map)
  * The flags "started" and "ended" tell is if we found the beginning
  * or (respectively) the end of a Gap Ack Block.
  */
-static void _sctp_tsnmap_find_gap_ack(__u8 *map, __u16 off,
-				      __u16 len, __u16 base,
-				      int *started, __u16 *start,
-				      int *ended, __u16 *end)
+static void sctp_tsnmap_find_gap_ack(__u8 *map, __u16 off,
+				     __u16 len, __u16 base,
+				     int *started, __u16 *start,
+				     int *ended, __u16 *end)
 {
 	int i = off;
 
-	/* Let's look through the entire array, but break out
+	/* Look through the entire array, but break out
 	 * early if we have found the end of the Gap Ack Block.
 	 */
 

net/sctp/ulpevent.c

@@ -5,37 +5,37 @@
  * Copyright (c) 2001 Intel Corp.
  * Copyright (c) 2001 Nokia, Inc.
  * Copyright (c) 2001 La Monte H.P. Yarroll
- * 
- * These functions manipulate an sctp event.   The sctp_ulpevent_t is used
- * to carry notifications and data to the ULP (sockets).  
- * The SCTP reference implementation is free software; 
- * you can redistribute it and/or modify it under the terms of 
+ *
+ * These functions manipulate an sctp event.   The struct ulpevent is used
+ * to carry notifications and data to the ULP (sockets).
+ * 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:
  *    Jon Grimm             <jgrimm@us.ibm.com>
  *    La Monte H.P. Yarroll <piggy@acm.org>
- * 
+ *
  * Any bugs reported given to us we will try to fix... any fixes shared will
  * be incorporated into the next SCTP release.
  */
@@ -47,58 +47,51 @@
 #include <net/sctp/sm.h>
 
 static void sctp_ulpevent_set_owner_r(struct sk_buff *skb,
-				      sctp_association_t *asoc);
-static void
-sctp_ulpevent_set_owner(struct sk_buff *skb, const sctp_association_t *asoc);
+				      struct sctp_association *asoc);
+static void sctp_ulpevent_set_owner(struct sk_buff *skb,
+				    const struct sctp_association *asoc);
 
 /* Create a new sctp_ulpevent.  */
-sctp_ulpevent_t *sctp_ulpevent_new(int size, int msg_flags, int priority)
+struct sctp_ulpevent *sctp_ulpevent_new(int size, int msg_flags, int priority)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct sk_buff *skb;
 
 	skb = alloc_skb(size, priority);
 	if (!skb)
 		goto fail;
 
-	event = (sctp_ulpevent_t *) skb->cb;
-	event = sctp_ulpevent_init(event, skb, msg_flags);
+	event = sctp_skb2event(skb);
+	event = sctp_ulpevent_init(event, msg_flags);
 	if (!event)
 		goto fail_init;
-
-	event->malloced = 1;
 	return event;
 
 fail_init:
-	kfree_skb(event->parent);
-
+	kfree_skb(skb);
 fail:
 	return NULL;
 }
 
 /* Initialize an ULP event from an given skb.  */
-sctp_ulpevent_t *sctp_ulpevent_init(sctp_ulpevent_t *event,
-				    struct sk_buff *parent,
-				    int msg_flags)
+struct sctp_ulpevent *sctp_ulpevent_init(struct sctp_ulpevent *event,
+					 int msg_flags)
 {
-	memset(event, sizeof(sctp_ulpevent_t), 0x00);
+	memset(event, sizeof(struct sctp_ulpevent), 0x00);
 	event->msg_flags = msg_flags;
-	event->parent = parent;
-	event->malloced = 0;
 	return event;
 }
 
 /* Dispose of an event.  */
-void sctp_ulpevent_free(sctp_ulpevent_t *event)
+void sctp_ulpevent_free(struct sctp_ulpevent *event)
 {
-	if (event->malloced)
-		kfree_skb(event->parent);
+	kfree_skb(sctp_event2skb(event));
 }
 
 /* Is this a MSG_NOTIFICATION?  */
-int sctp_ulpevent_is_notification(const sctp_ulpevent_t *event)
+int sctp_ulpevent_is_notification(const struct sctp_ulpevent *event)
 {
-	return event->msg_flags & MSG_NOTIFICATION;
+	return MSG_NOTIFICATION == (event->msg_flags & MSG_NOTIFICATION);
 }
 
 /* Create and initialize an SCTP_ASSOC_CHANGE event.
@@ -112,24 +105,22 @@ int sctp_ulpevent_is_notification(const sctp_ulpevent_t *event)
  * Note: There is no field checking here.  If a field is unused it will be
  * zero'd out.
  */
-sctp_ulpevent_t *sctp_ulpevent_make_assoc_change(const sctp_association_t *asoc,
-						 __u16 flags,
-						 __u16 state,
-						 __u16 error,
-						 __u16 outbound,
-						 __u16 inbound,
-						 int priority)
+struct sctp_ulpevent  *sctp_ulpevent_make_assoc_change(
+	const sctp_association_t *asoc,
+	__u16 flags, __u16 state, __u16 error, __u16 outbound,
+	__u16 inbound, int priority)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct sctp_assoc_change *sac;
+	struct sk_buff *skb;
 
 	event = sctp_ulpevent_new(sizeof(struct sctp_assoc_change),
 				  MSG_NOTIFICATION, priority);
 	if (!event)
 		goto fail;
-
+	skb = sctp_event2skb(event);
 	sac = (struct sctp_assoc_change *)
-		skb_put(event->parent, sizeof(struct sctp_assoc_change));
+		skb_put(skb, sizeof(struct sctp_assoc_change));
 
 	/* Socket Extensions for SCTP
 	 * 5.3.1.1 SCTP_ASSOC_CHANGE
@@ -198,13 +189,13 @@ sctp_ulpevent_t *sctp_ulpevent_make_assoc_change(const sctp_association_t *asoc,
 	 * All notifications for a given association have the same association
 	 * identifier.  For TCP style socket, this field is ignored.
 	 */
-	sctp_ulpevent_set_owner(event->parent, asoc);
+	sctp_ulpevent_set_owner(skb, asoc);
 	sac->sac_assoc_id = sctp_assoc2id(asoc);
 
 	return event;
 
 fail:
-	return NULL;	
+	return NULL;
 }
 
 /* Create and initialize an SCTP_PEER_ADDR_CHANGE event.
@@ -215,24 +206,22 @@ sctp_ulpevent_t *sctp_ulpevent_make_assoc_change(const sctp_association_t *asoc,
  * When a destination address on a multi-homed peer encounters a change
  * an interface details event is sent.
  */
-sctp_ulpevent_t *sctp_ulpevent_make_peer_addr_change(
-				const sctp_association_t *asoc,
-				const struct sockaddr_storage *aaddr,
-				int flags,
-				int state,
-				int error,
-				int priority)
+struct sctp_ulpevent *sctp_ulpevent_make_peer_addr_change(
+	const sctp_association_t *asoc, const struct sockaddr_storage *aaddr,
+	int flags, int state, int error, int priority)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct sctp_paddr_change  *spc;
+	struct sk_buff *skb;
 
 	event = sctp_ulpevent_new(sizeof(struct sctp_paddr_change),
 				  MSG_NOTIFICATION, priority);
 	if (!event)
 		goto fail;
 
+	skb = sctp_event2skb(event);
 	spc = (struct sctp_paddr_change *)
-		skb_put(event->parent, sizeof(struct sctp_paddr_change));
+		skb_put(skb, sizeof(struct sctp_paddr_change));
 
 	/* Sockets API Extensions for SCTP
 	 * Section 5.3.1.2 SCTP_PEER_ADDR_CHANGE
@@ -265,7 +254,7 @@ sctp_ulpevent_t *sctp_ulpevent_make_peer_addr_change(
 	 * Section 5.3.1.2 SCTP_PEER_ADDR_CHANGE
 	 *
 	 * spc_state:  32 bits (signed integer)
-	 * 
+	 *
 	 * This field holds one of a number of values that communicate the
 	 * event that happened to the address.
 	 */
@@ -291,7 +280,7 @@ sctp_ulpevent_t *sctp_ulpevent_make_peer_addr_change(
 	 * All notifications for a given association have the same association
 	 * identifier.  For TCP style socket, this field is ignored.
 	 */
-	sctp_ulpevent_set_owner(event->parent, asoc);
+	sctp_ulpevent_set_owner(skb, asoc);
 	spc->spc_assoc_id = sctp_assoc2id(asoc);
 
 	/* Sockets API Extensions for SCTP
@@ -325,12 +314,11 @@ sctp_ulpevent_t *sctp_ulpevent_make_peer_addr_change(
  * specification [SCTP] and any extensions for a list of possible
  * error formats.
  */
-sctp_ulpevent_t *sctp_ulpevent_make_remote_error(const sctp_association_t *asoc,
-						 sctp_chunk_t *chunk,
-						 __u16 flags,
-						 int priority)
+struct sctp_ulpevent *sctp_ulpevent_make_remote_error(
+	const sctp_association_t *asoc, sctp_chunk_t *chunk,
+	__u16 flags, int priority)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct sctp_remote_error *sre;
 	struct sk_buff *skb;
 	sctp_errhdr_t *ch;
@@ -358,13 +346,12 @@ sctp_ulpevent_t *sctp_ulpevent_make_remote_error(const sctp_association_t *asoc,
 		goto fail;
 
 	/* Embed the event fields inside the cloned skb.  */
-	event = (sctp_ulpevent_t *) skb->cb;
-	event = sctp_ulpevent_init(event, skb, MSG_NOTIFICATION);
+	event = sctp_skb2event(skb);
+	event = sctp_ulpevent_init(event, MSG_NOTIFICATION);
 
 	if (!event)
 		goto fail;
 
-	event->malloced = 1;
 	sre = (struct sctp_remote_error *)
 		skb_push(skb, sizeof(struct sctp_remote_error));
 
@@ -416,7 +403,8 @@ sctp_ulpevent_t *sctp_ulpevent_make_remote_error(const sctp_association_t *asoc,
 	 * All notifications for a given association have the same association
 	 * identifier.  For TCP style socket, this field is ignored.
 	 */
-	sctp_ulpevent_set_owner(event->parent, asoc);
+	skb = sctp_event2skb(event);
+	sctp_ulpevent_set_owner(skb, asoc);
 	sre->sre_assoc_id = sctp_assoc2id(asoc);
 
 	return event;
@@ -430,13 +418,11 @@ sctp_ulpevent_t *sctp_ulpevent_make_remote_error(const sctp_association_t *asoc,
  * Socket Extensions for SCTP - draft-01
  * 5.3.1.4 SCTP_SEND_FAILED
  */
-sctp_ulpevent_t *sctp_ulpevent_make_send_failed(const sctp_association_t *asoc,
-						sctp_chunk_t *chunk,
-						__u16 flags,
-						__u32 error,
-						int priority)
+struct sctp_ulpevent *sctp_ulpevent_make_send_failed(
+	const sctp_association_t *asoc, sctp_chunk_t *chunk,
+	__u16 flags, __u32 error, int priority)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct sctp_send_failed *ssf;
 	struct sk_buff *skb;
 
@@ -452,16 +438,11 @@ sctp_ulpevent_t *sctp_ulpevent_make_send_failed(const sctp_association_t *asoc,
 	skb_pull(skb, sizeof(sctp_data_chunk_t));
 
 	/* Embed the event fields inside the cloned skb.  */
-	event = (sctp_ulpevent_t *) skb->cb;
-	event = sctp_ulpevent_init(event, skb, MSG_NOTIFICATION);
+	event = sctp_skb2event(skb);
+	event = sctp_ulpevent_init(event, MSG_NOTIFICATION);
 	if (!event)
 		goto fail;
 
-	/* Mark as malloced, even though the constructor was not
-	 * called.
-	 */
-	event->malloced = 1;
-
 	ssf = (struct sctp_send_failed *)
 		skb_push(skb, sizeof(struct sctp_send_failed));
 
@@ -525,7 +506,8 @@ sctp_ulpevent_t *sctp_ulpevent_make_send_failed(const sctp_association_t *asoc,
 	 * same association identifier.  For TCP style socket, this field is
 	 * ignored.
 	 */
-	sctp_ulpevent_set_owner(event->parent, asoc);
+	skb = sctp_event2skb(event);
+	sctp_ulpevent_set_owner(skb, asoc);
 	ssf->ssf_assoc_id = sctp_assoc2id(asoc);
 	return event;
 
@@ -538,21 +520,22 @@ sctp_ulpevent_t *sctp_ulpevent_make_send_failed(const sctp_association_t *asoc,
  * Socket Extensions for SCTP - draft-01
  * 5.3.1.5 SCTP_SHUTDOWN_EVENT
  */
-sctp_ulpevent_t *sctp_ulpevent_make_shutdown_event(
+struct sctp_ulpevent *sctp_ulpevent_make_shutdown_event(
 	const sctp_association_t *asoc,
-	__u16 flags,
-	int priority)
+	__u16 flags, int priority)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 	struct sctp_shutdown_event *sse;
+	struct sk_buff *skb;
 
 	event = sctp_ulpevent_new(sizeof(struct sctp_assoc_change),
 				  MSG_NOTIFICATION, priority);
 	if (!event)
 		goto fail;
 
+	skb = sctp_event2skb(event);
 	sse = (struct sctp_shutdown_event *)
-		skb_put(event->parent, sizeof(struct sctp_shutdown_event));
+		skb_put(skb, sizeof(struct sctp_shutdown_event));
 
 	/* Socket Extensions for SCTP
 	 * 5.3.1.5 SCTP_SHUTDOWN_EVENT
@@ -587,7 +570,7 @@ sctp_ulpevent_t *sctp_ulpevent_make_shutdown_event(
 	 * All notifications for a given association have the same association
 	 * identifier.  For TCP style socket, this field is ignored.
 	 */
-	sctp_ulpevent_set_owner(event->parent, asoc);
+	sctp_ulpevent_set_owner(skb, asoc);
 	sse->sse_assoc_id = sctp_assoc2id(asoc);
 
 	return event;
@@ -600,13 +583,13 @@ sctp_ulpevent_t *sctp_ulpevent_make_shutdown_event(
  * to pass it to the upper layers.  Go ahead and calculate the sndrcvinfo
  * even if filtered out later.
  *
- * Socket Extensions for SCTP - draft-01
+ * Socket Extensions for SCTP
  * 5.2.2 SCTP Header Information Structure (SCTP_SNDRCV)
  */
-sctp_ulpevent_t *sctp_ulpevent_make_rcvmsg(sctp_association_t *asoc,
-					   sctp_chunk_t *chunk, int priority)
+struct sctp_ulpevent *sctp_ulpevent_make_rcvmsg(sctp_association_t *asoc,
+		   sctp_chunk_t *chunk, int priority)
 {
-	sctp_ulpevent_t *event, *levent;
+	struct sctp_ulpevent *event;
 	struct sctp_sndrcvinfo *info;
 	struct sk_buff *skb, *list;
 	size_t padding, len;
@@ -638,24 +621,19 @@ sctp_ulpevent_t *sctp_ulpevent_make_rcvmsg(sctp_association_t *asoc,
 	sctp_ulpevent_set_owner_r(skb, asoc);
 
 	/* Embed the event fields inside the cloned skb.  */
-	event = (sctp_ulpevent_t *) skb->cb;
+	event = sctp_skb2event(skb);
 
 	/* Initialize event with flags 0.  */
-	event = sctp_ulpevent_init(event, skb, 0);
+	event = sctp_ulpevent_init(event, 0);
 	if (!event)
 		goto fail_init;
 
-	event->malloced = 1;
-
-	for (list = skb_shinfo(skb)->frag_list; list; list = list->next) {
+	/* Note:  Not clearing the entire event struct as
+	 * this is just a fragment of the real event.  However,
+	 * we still need to do rwnd accounting. 
+	 */
+	for (list = skb_shinfo(skb)->frag_list; list; list = list->next)
 		sctp_ulpevent_set_owner_r(list, asoc);
-		/* Initialize event with flags 0.  */
-		levent = sctp_ulpevent_init(event, skb, 0);
-		if (!levent)
-			goto fail_init;
-
-		levent->malloced = 1;
-	}
 
 	info = (struct sctp_sndrcvinfo *) &event->sndrcvinfo;
 
@@ -707,18 +685,26 @@ sctp_ulpevent_t *sctp_ulpevent_make_rcvmsg(sctp_association_t *asoc,
 	 * MSG_UNORDERED - This flag is present when the message was sent
 	 *                 non-ordered.
 	 */
-	if (chunk->chunk_hdr->flags & SCTP_DATA_UNORDERED)
+	if (chunk->chunk_hdr->flags & SCTP_DATA_UNORDERED) {
 		info->sinfo_flags |= MSG_UNORDERED;
 
-	/* FIXME:  For reassembly, we need to have the fragmentation bits.
-	 * This really does not belong in the event structure, but
-	 * its difficult to fix everything at the same time.   Eventually,
-	 * we should create and skb based chunk structure.   This structure
-	 * storage can be converted to an event.  --jgrimm
+		/* sinfo_cumtsn: 32 bit (unsigned integer)
+		 *  
+		 * This field will hold the current cumulative TSN as 
+		 * known by the underlying SCTP layer.  Note this field is 
+		 * ignored when sending and only valid for a receive 
+		 * operation when sinfo_flags are set to MSG_UNORDERED.
+		 */
+		info->sinfo_cumtsn = sctp_tsnmap_get_ctsn(&asoc->peer.tsn_map);
+	}
+
+	/* Note:  For reassembly, we need to have the fragmentation bits. 
+	 * For now, merge these into the msg_flags, since those bit
+	 * possitions are not used.
 	 */
-	event->chunk_flags = chunk->chunk_hdr->flags;
+	event->msg_flags |= chunk->chunk_hdr->flags;
 
-	/* With -04 draft, tsn moves into sndrcvinfo. */
+	/* With 04 draft, tsn moves into sndrcvinfo. */
 	info->sinfo_tsn = ntohl(chunk->subh.data_hdr->tsn);
 
 	/* Context is not used on receive. */
@@ -745,19 +731,79 @@ sctp_ulpevent_t *sctp_ulpevent_make_rcvmsg(sctp_association_t *asoc,
 	return NULL;
 }
 
+/* Create a partial delivery related event. 
+ *
+ * 5.3.1.7 SCTP_PARTIAL_DELIVERY_EVENT
+ *
+ *   When a reciever is engaged in a partial delivery of a
+ *   message this notification will be used to inidicate
+ *   various events.
+ */
+struct sctp_ulpevent *sctp_ulpevent_make_pdapi(
+	const sctp_association_t *asoc, __u32 indication, int priority)
+{
+	struct sctp_ulpevent *event;
+	struct sctp_rcv_pdapi_event *pd;
+	struct sk_buff *skb;
+
+	event = sctp_ulpevent_new(sizeof(struct sctp_assoc_change),
+				  MSG_NOTIFICATION, priority);
+	if (!event)
+		goto fail;
+
+	skb = sctp_event2skb(event);
+	pd = (struct sctp_rcv_pdapi_event *)
+		skb_put(skb, sizeof(struct sctp_rcv_pdapi_event));
+
+	/* pdapi_type
+	 *   It should be SCTP_PARTIAL_DELIVERY_EVENT
+	 *
+	 * pdapi_flags: 16 bits (unsigned integer)
+	 *   Currently unused.
+	 */
+	pd->pdapi_type = SCTP_PARTIAL_DELIVERY_EVENT;
+	pd->pdapi_flags = 0;
+
+	/* pdapi_length: 32 bits (unsigned integer)
+	 *
+	 * This field is the total length of the notification data, including
+	 * the notification header.  It will generally be sizeof (struct
+	 * sctp_rcv_pdapi_event).
+	 */
+	pd->pdapi_length = sizeof(struct sctp_rcv_pdapi_event);
+
+        /*  pdapi_indication: 32 bits (unsigned integer)
+	 *
+	 * This field holds the indication being sent to the application.
+	 */
+	pd->pdapi_indication = indication;
+
+	/*  pdapi_assoc_id: sizeof (sctp_assoc_t)
+	 * 
+	 * The association id field, holds the identifier for the association.
+	 */
+	pd->pdapi_assoc_id = sctp_assoc2id(asoc);
+
+	return event;
+fail:
+	return NULL;
+}
+
 /* Return the notification type, assuming this is a notification
  * event.
  */
-__u16 sctp_ulpevent_get_notification_type(const sctp_ulpevent_t *event)
+__u16 sctp_ulpevent_get_notification_type(const struct sctp_ulpevent *event)
 {
 	union sctp_notification *notification;
+	struct sk_buff *skb;
 
-	notification = (union sctp_notification *) event->parent->data;
+	skb = sctp_event2skb((struct sctp_ulpevent *)event);
+	notification = (union sctp_notification *) skb->data;
 	return notification->h.sn_type;
 }
 
 /* Copy out the sndrcvinfo into a msghdr.  */
-void sctp_ulpevent_read_sndrcvinfo(const sctp_ulpevent_t *event,
+void sctp_ulpevent_read_sndrcvinfo(const struct sctp_ulpevent *event,
 				   struct msghdr *msghdr)
 {
 	if (!sctp_ulpevent_is_notification(event)) {
@@ -771,7 +817,7 @@ void sctp_ulpevent_read_sndrcvinfo(const sctp_ulpevent_t *event,
 static void sctp_rcvmsg_rfree(struct sk_buff *skb)
 {
 	sctp_association_t *asoc;
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 
 	/* Current stack structures assume that the rcv buffer is
 	 * per socket.   For UDP style sockets this is not true as
@@ -779,16 +825,17 @@ static void sctp_rcvmsg_rfree(struct sk_buff *skb)
 	 * Use the local private area of the skb to track the owning
 	 * association.
 	 */
-	event = (sctp_ulpevent_t *) skb->cb;
+	event = sctp_skb2event(skb);
 	asoc = event->asoc;
 	sctp_assoc_rwnd_increase(asoc, skb_headlen(skb));
 	sctp_association_put(asoc);
 }
 
 /* Charge receive window for bytes recieved.  */
-static void sctp_ulpevent_set_owner_r(struct sk_buff *skb, sctp_association_t *asoc)
+static void sctp_ulpevent_set_owner_r(struct sk_buff *skb,
+				      sctp_association_t *asoc)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 
 	/* The current stack structures assume that the rcv buffer is
 	 * per socket.  For UDP-style sockets this is not true as
@@ -798,7 +845,7 @@ static void sctp_ulpevent_set_owner_r(struct sk_buff *skb, sctp_association_t *a
 	 */
 	sctp_association_hold(asoc);
 	skb->sk = asoc->base.sk;
-	event = (sctp_ulpevent_t *) skb->cb;
+	event = sctp_skb2event(skb);
 	event->asoc = asoc;
 
 	skb->destructor = sctp_rcvmsg_rfree;
@@ -809,26 +856,26 @@ static void sctp_ulpevent_set_owner_r(struct sk_buff *skb, sctp_association_t *a
 /* A simple destructor to give up the reference to the association. */
 static void sctp_ulpevent_rfree(struct sk_buff *skb)
 {
-	sctp_ulpevent_t *event;	
-	
-	event = (sctp_ulpevent_t *)skb->cb;
+	struct sctp_ulpevent *event;
+
+	event = sctp_skb2event(skb);
 	sctp_association_put(event->asoc);
 }
 
-/* Hold the association in case the msg_name needs read out of 
- * the association. 
+/* Hold the association in case the msg_name needs read out of
+ * the association.
  */
 static void sctp_ulpevent_set_owner(struct sk_buff *skb,
-				    const sctp_association_t *asoc)
+				    const struct sctp_association *asoc)
 {
-	sctp_ulpevent_t *event;
+	struct sctp_ulpevent *event;
 
 	/* Cast away the const, as we are just wanting to
 	 * bump the reference count.
 	 */
-	sctp_association_hold((sctp_association_t *)asoc);
+	sctp_association_hold((struct sctp_association *)asoc);
 	skb->sk = asoc->base.sk;
-	event = (sctp_ulpevent_t *)skb->cb;
-	event->asoc = (sctp_association_t *)asoc;
+	event = sctp_skb2event(skb);
+	event->asoc = (struct sctp_association *)asoc;
 	skb->destructor = sctp_ulpevent_rfree;
 }

net/sctp/ulpqueue.c

@@ -84,6 +84,7 @@ struct sctp_ulpq *sctp_ulpq_init(struct sctp_ulpq *ulpq,
 	ulpq->asoc = asoc;
 	skb_queue_head_init(&ulpq->reasm);
 	skb_queue_head_init(&ulpq->lobby);
+	ulpq->pd_mode  = 0;
 	ulpq->malloced = 0;
 
 	return ulpq;
@@ -96,15 +97,16 @@ void sctp_ulpq_flush(struct sctp_ulpq *ulpq)
 	struct sk_buff *skb;
 	struct sctp_ulpevent *event;
 
-	while ((skb = skb_dequeue(&ulpq->lobby))) {
-		event = (struct sctp_ulpevent *) skb->cb;
+	while ((skb = __skb_dequeue(&ulpq->lobby))) {
+		event = sctp_skb2event(skb);
 		sctp_ulpevent_free(event);
 	}
 
-	while ((skb = skb_dequeue(&ulpq->reasm))) {
-		event = (struct sctp_ulpevent *) skb->cb;
+	while ((skb = __skb_dequeue(&ulpq->reasm))) {
+		event = sctp_skb2event(skb);
 		sctp_ulpevent_free(event);
 	}
+
 }
 
 /* Dispose of a ulpqueue.  */
@@ -117,7 +119,7 @@ void sctp_ulpq_free(struct sctp_ulpq *ulpq)
 
 /* Process an incoming DATA chunk.  */
 int sctp_ulpq_tail_data(struct sctp_ulpq *ulpq, sctp_chunk_t *chunk,
-			    int priority)
+			int priority)
 {
 	struct sk_buff_head temp;
 	sctp_data_chunk_t *hdr;
@@ -125,12 +127,7 @@ int sctp_ulpq_tail_data(struct sctp_ulpq *ulpq, sctp_chunk_t *chunk,
 
 	hdr = (sctp_data_chunk_t *) chunk->chunk_hdr;
 
-	/* FIXME: Instead of event being the skb clone, we really should
-	 * have a new skb based chunk structure that we can convert to
-	 * an event.  Temporarily, I'm carrying a few chunk fields in
-	 * the event to allow reassembly.  Its too painful to change
-	 * everything at once.  --jgrimm
-	 */
+	/* Create an event from the incoming chunk. */
 	event = sctp_ulpevent_make_rcvmsg(chunk->asoc, chunk, priority);
 	if (!event)
 		return -ENOMEM;
@@ -139,10 +136,10 @@ int sctp_ulpq_tail_data(struct sctp_ulpq *ulpq, sctp_chunk_t *chunk,
 	event = sctp_ulpq_reasm(ulpq, event);
 
 	/* Do ordering if needed.  */
-	if (event) {
+	if ((event) && (event->msg_flags & MSG_EOR)){
 		/* Create a temporary list to collect chunks on.  */
 		skb_queue_head_init(&temp);
-		skb_queue_tail(&temp, event->parent);
+		__skb_queue_tail(&temp, sctp_event2skb(event));
 
 		event = sctp_ulpq_order(ulpq, event);
 	}
@@ -154,10 +151,40 @@ int sctp_ulpq_tail_data(struct sctp_ulpq *ulpq, sctp_chunk_t *chunk,
 	return 0;
 }
 
+/* Clear the partial delivery mode for this socket.   Note: This
+ * assumes that no association is currently in partial delivery mode.
+ */
+int sctp_clear_pd(struct sock *sk)
+{
+	struct sctp_opt *sp;
+	sp = sctp_sk(sk);
+
+	sp->pd_mode = 0;
+	if (!skb_queue_empty(&sp->pd_lobby)) {
+		struct list_head *list;
+		sctp_skb_list_tail(&sp->pd_lobby, &sk->receive_queue);
+		list = (struct list_head *)&sctp_sk(sk)->pd_lobby;
+		INIT_LIST_HEAD(list);
+		return 1;
+	}
+	return 0;
+}
+
+/* Clear the pd_mode and restart any pending messages waiting for delivery. */
+static int sctp_ulpq_clear_pd(struct sctp_ulpq *ulpq)
+{
+	ulpq->pd_mode = 0;
+	return sctp_clear_pd(ulpq->asoc->base.sk);
+}
+
+
+
 /* Add a new event for propogation to the ULP.  */
 int sctp_ulpq_tail_event(struct sctp_ulpq *ulpq, struct sctp_ulpevent *event)
 {
 	struct sock *sk = ulpq->asoc->base.sk;
+	struct sk_buff_head *queue;
+	int clear_pd = 0;
 
 	/* If the socket is just going to throw this away, do not
 	 * even try to deliver it.
@@ -169,29 +196,55 @@ int sctp_ulpq_tail_event(struct sctp_ulpq *ulpq, struct sctp_ulpevent *event)
 	if (!sctp_ulpevent_is_enabled(event, &sctp_sk(sk)->subscribe))
 		goto out_free;
 
+	/* If we are in partial delivery mode, post to the lobby until
+	 * partial delivery is cleared, unless, of course _this_ is
+	 * the association the cause of the partial delivery.
+	 */
+
+	if (!sctp_sk(sk)->pd_mode) {
+		queue = &sk->receive_queue;
+	} else if (ulpq->pd_mode) {
+		if (event->msg_flags & MSG_NOTIFICATION)
+		       	queue = &sctp_sk(sk)->pd_lobby;
+		else {
+			clear_pd = event->msg_flags & MSG_EOR;
+			queue = &sk->receive_queue;
+		}
+	} else
+		queue = &sctp_sk(sk)->pd_lobby;
+
+
 	/* If we are harvesting multiple skbs they will be
 	 * collected on a list.
 	 */
-	if (event->parent->list)
-		sctp_skb_list_tail(event->parent->list, &sk->receive_queue);
+	if (sctp_event2skb(event)->list)
+		sctp_skb_list_tail(sctp_event2skb(event)->list, queue);
 	else
-		skb_queue_tail(&sk->receive_queue, event->parent);
+		skb_queue_tail(queue, sctp_event2skb(event));
 
-	wake_up_interruptible(sk->sleep);
+	/* Did we just complete partial delivery and need to get
+	 * rolling again?  Move pending data to the receive
+	 * queue.
+	 */
+	if (clear_pd)
+		sctp_ulpq_clear_pd(ulpq);
+
+	if (queue == &sk->receive_queue)
+		wake_up_interruptible(sk->sleep);
 	return 1;
 
 out_free:
-	if (event->parent->list)
-		skb_queue_purge(event->parent->list);
+	if (sctp_event2skb(event)->list)
+		skb_queue_purge(sctp_event2skb(event)->list);
 	else
-		kfree_skb(event->parent);
+		kfree_skb(sctp_event2skb(event));
 	return 0;
 }
 
 /* 2nd Level Abstractions */
 
 /* Helper function to store chunks that need to be reassembled.  */
-static inline void sctp_ulpq_store_reasm(struct sctp_ulpq *ulpq, 
+static inline void sctp_ulpq_store_reasm(struct sctp_ulpq *ulpq,
 					 struct sctp_ulpevent *event)
 {
 	struct sk_buff *pos, *tmp;
@@ -202,7 +255,7 @@ static inline void sctp_ulpq_store_reasm(struct sctp_ulpq *ulpq,
 
 	/* Find the right place in this list. We store them by TSN.  */
 	sctp_skb_for_each(pos, &ulpq->reasm, tmp) {
-		cevent = (struct sctp_ulpevent *)pos->cb;
+		cevent = sctp_skb2event(pos);
 		ctsn = cevent->sndrcvinfo.sinfo_tsn;
 
 		if (TSN_lt(tsn, ctsn))
@@ -211,9 +264,10 @@ static inline void sctp_ulpq_store_reasm(struct sctp_ulpq *ulpq,
 
 	/* If the queue is empty, we have a different function to call.  */
 	if (skb_peek(&ulpq->reasm))
-		__skb_insert(event->parent, pos->prev, pos, &ulpq->reasm);
+		__skb_insert(sctp_event2skb(event), pos->prev, pos,
+			     &ulpq->reasm);
 	else
-		__skb_queue_tail(&ulpq->reasm, event->parent);
+		__skb_queue_tail(&ulpq->reasm, sctp_event2skb(event));
 }
 
 /* Helper function to return an event corresponding to the reassembled
@@ -231,7 +285,10 @@ static inline struct sctp_ulpevent *sctp_make_reassembled_event(struct sk_buff *
 	struct sk_buff *list = skb_shinfo(f_frag)->frag_list;
 
 	/* Store the pointer to the 2nd skb */
-	pos = f_frag->next;
+	if (f_frag == l_frag)
+		pos = NULL;
+	else
+		pos = f_frag->next;
 
 	/* Get the last skb in the f_frag's frag_list if present. */
 	for (last = list; list; last = list, list = list->next);
@@ -246,7 +303,8 @@ static inline struct sctp_ulpevent *sctp_make_reassembled_event(struct sk_buff *
 
 	/* Remove the first fragment from the reassembly queue.  */
 	__skb_unlink(f_frag, f_frag->list);
-	do {
+	while (pos) {
+
 		pnext = pos->next;
 
 		/* Update the len and data_len fields of the first fragment. */
@@ -262,25 +320,27 @@ static inline struct sctp_ulpevent *sctp_make_reassembled_event(struct sk_buff *
 
 		pos->next = pnext;
 		pos = pnext;
-	} while (1);
+	};
 
-	event = (sctp_ulpevent_t *) f_frag->cb;
+	event = sctp_skb2event(f_frag);
+	SCTP_INC_STATS(SctpReasmUsrMsgs);
 
 	return event;
 }
 
+
 /* Helper function to check if an incoming chunk has filled up the last
  * missing fragment in a SCTP datagram and return the corresponding event.
  */
-static inline sctp_ulpevent_t *sctp_ulpq_retrieve_reassembled(struct sctp_ulpq *ulpq)
+static inline struct sctp_ulpevent *sctp_ulpq_retrieve_reassembled(struct sctp_ulpq *ulpq)
 {
 	struct sk_buff *pos, *tmp;
-	sctp_ulpevent_t *cevent;
+	struct sctp_ulpevent *cevent;
 	struct sk_buff *first_frag = NULL;
 	__u32 ctsn, next_tsn;
-	sctp_ulpevent_t *retval = NULL;
+	struct sctp_ulpevent *retval = NULL;
 
-	/* Initialized to 0 just to avoid compiler warning message. Will
+	/* Initialized to 0 just to avoid compiler warning message.  Will
 	 * never be used with this value. It is referenced only after it
 	 * is set when we find the first fragment of a message.
 	 */
@@ -296,10 +356,10 @@ static inline sctp_ulpevent_t *sctp_ulpq_retrieve_reassembled(struct sctp_ulpq *
 	 * start the next pass when we find another first fragment.
 	 */
 	sctp_skb_for_each(pos, &ulpq->reasm, tmp) {
-		cevent = (sctp_ulpevent_t *) pos->cb;
+		cevent = sctp_skb2event(pos);
 		ctsn = cevent->sndrcvinfo.sinfo_tsn;
 
-		switch (cevent->chunk_flags & SCTP_DATA_FRAG_MASK) {
+		switch (cevent->msg_flags & SCTP_DATA_FRAG_MASK) {
 		case SCTP_DATA_FIRST_FRAG:
 			first_frag = pos;
 			next_tsn = ctsn + 1;
@@ -313,7 +373,7 @@ static inline sctp_ulpevent_t *sctp_ulpq_retrieve_reassembled(struct sctp_ulpq *
 			break;
 
 		case SCTP_DATA_LAST_FRAG:
-			if ((first_frag) && (ctsn == next_tsn))
+			if (first_frag && (ctsn == next_tsn))
 				retval = sctp_make_reassembled_event(
 						first_frag, pos);
 			else
@@ -324,42 +384,170 @@ static inline sctp_ulpevent_t *sctp_ulpq_retrieve_reassembled(struct sctp_ulpq *
 		/* We have the reassembled event. There is no need to look
 		 * further.
 		 */
-		if (retval)
+		if (retval) {
+			retval->msg_flags |= MSG_EOR;
 			break;
+		}
 	}
 
 	return retval;
 }
 
-/* Helper function to reassemble chunks. Hold chunks on the reasm queue that
- * need reassembling.
- */
-static inline sctp_ulpevent_t *sctp_ulpq_reasm(struct sctp_ulpq *ulpq,
-						   sctp_ulpevent_t *event)
+/* Retrieve the next set of fragments of a partial message. */
+static inline struct sctp_ulpevent *sctp_ulpq_retrieve_partial(struct sctp_ulpq *ulpq)
 {
-	sctp_ulpevent_t *retval = NULL;
+	struct sk_buff *pos, *tmp, *last_frag, *first_frag;
+	struct sctp_ulpevent *cevent;
+	__u32 ctsn, next_tsn;
+	int is_last;
+	struct sctp_ulpevent *retval;
+
+	/* The chunks are held in the reasm queue sorted by TSN.
+	 * Walk through the queue sequentially and look for the first
+	 * sequence of fragmented chunks.
+	 */
+
+	if (skb_queue_empty(&ulpq->reasm))
+		return NULL;
+
+	last_frag = first_frag = NULL;
+	retval = NULL;
+	next_tsn = 0;
+	is_last = 0;
+
+	sctp_skb_for_each(pos, &ulpq->reasm, tmp) {
+		cevent = sctp_skb2event(pos);
+		ctsn = cevent->sndrcvinfo.sinfo_tsn;
+
+		switch (cevent->msg_flags & SCTP_DATA_FRAG_MASK) {
+		case SCTP_DATA_MIDDLE_FRAG:
+			if (!first_frag) {
+				first_frag = pos;
+				next_tsn = ctsn + 1;
+				last_frag = pos;
+			} else if (next_tsn == ctsn)
+				next_tsn++;
+			else
+				goto done;
+			break;
+		case SCTP_DATA_LAST_FRAG:
+			if (!first_frag)
+				first_frag = pos;
+			else if (ctsn != next_tsn)
+				goto done;
+			last_frag = pos;
+			is_last = 1;
+			goto done;
+		default:
+			return NULL;
+		};
+	}
 
-	/* FIXME: We should be using some new chunk structure here
-	 * instead of carrying chunk fields in the event structure.
-	 * This is temporary as it is too painful to change everything
-	 * at once.
+	/* We have the reassembled event. There is no need to look
+	 * further.
 	 */
+done:
+	retval = sctp_make_reassembled_event(first_frag, last_frag);
+	if (is_last)
+		retval->msg_flags |= MSG_EOR;
+
+	return retval;
+}
+
+
+/* Helper function to reassemble chunks.  Hold chunks on the reasm queue that
+ * need reassembling.
+ */
+static inline struct sctp_ulpevent *sctp_ulpq_reasm(struct sctp_ulpq *ulpq,
+						   struct sctp_ulpevent *event)
+{
+	struct sctp_ulpevent *retval = NULL;
 
 	/* Check if this is part of a fragmented message.  */
-	if (SCTP_DATA_NOT_FRAG == (event->chunk_flags & SCTP_DATA_FRAG_MASK))
+	if (SCTP_DATA_NOT_FRAG == (event->msg_flags & SCTP_DATA_FRAG_MASK)) {
+		event->msg_flags |= MSG_EOR;
 		return event;
+	}
 
 	sctp_ulpq_store_reasm(ulpq, event);
-	retval = sctp_ulpq_retrieve_reassembled(ulpq);
+	if (!ulpq->pd_mode)
+		retval = sctp_ulpq_retrieve_reassembled(ulpq);
+	else {
+		__u32 ctsn, ctsnap;
+
+		/* Do not even bother unless this is the next tsn to
+		 * be delivered.
+		 */
+		ctsn = event->sndrcvinfo.sinfo_tsn;
+		ctsnap = sctp_tsnmap_get_ctsn(&ulpq->asoc->peer.tsn_map);
+		if (TSN_lte(ctsn, ctsnap))
+			retval = sctp_ulpq_retrieve_partial(ulpq);
+	}
 
 	return retval;
 }
 
+/* Retrieve the first part (sequential fragments) for partial delivery.  */
+static inline struct sctp_ulpevent *sctp_ulpq_retrieve_first(struct sctp_ulpq *ulpq)
+{
+	struct sk_buff *pos, *tmp, *last_frag, *first_frag;
+	struct sctp_ulpevent *cevent;
+	__u32 ctsn, next_tsn;
+	struct sctp_ulpevent *retval;
+
+	/* The chunks are held in the reasm queue sorted by TSN.
+	 * Walk through the queue sequentially and look for a sequence of
+	 * fragmented chunks that start a datagram.
+	 */
+
+	if (skb_queue_empty(&ulpq->reasm))
+		return NULL;
+
+	last_frag = first_frag = NULL;
+	retval = NULL;
+	next_tsn = 0;
+
+	sctp_skb_for_each(pos, &ulpq->reasm, tmp) {
+		cevent = sctp_skb2event(pos);
+		ctsn = cevent->sndrcvinfo.sinfo_tsn;
+
+		switch (cevent->msg_flags & SCTP_DATA_FRAG_MASK) {
+		case SCTP_DATA_FIRST_FRAG:
+			if (!first_frag) {
+				first_frag = pos;
+				next_tsn = ctsn + 1;
+				last_frag = pos;
+			} else
+				goto done;
+			break;
+
+		case SCTP_DATA_MIDDLE_FRAG:
+			if (!first_frag)
+				return NULL;
+			if (ctsn == next_tsn) {
+				next_tsn++;
+				last_frag = pos;
+			} else
+				goto done;
+			break;
+		default:
+			return NULL;
+		};
+	}
+
+	/* We have the reassembled event. There is no need to look
+	 * further.
+	 */
+done:
+	retval = sctp_make_reassembled_event(first_frag, last_frag);
+	return retval;
+}
+
 /* Helper function to gather skbs that have possibly become
  * ordered by an an incoming chunk.
  */
 static inline void sctp_ulpq_retrieve_ordered(struct sctp_ulpq *ulpq,
-					      sctp_ulpevent_t *event)
+					      struct sctp_ulpevent *event)
 {
 	struct sk_buff *pos, *tmp;
 	struct sctp_ulpevent *cevent;
@@ -373,7 +561,7 @@ static inline void sctp_ulpq_retrieve_ordered(struct sctp_ulpq *ulpq,
 
 	/* We are holding the chunks by stream, by SSN.  */
 	sctp_skb_for_each(pos, &ulpq->lobby, tmp) {
-		cevent = (sctp_ulpevent_t *) pos->cb;
+		cevent = (struct sctp_ulpevent *) pos->cb;
 		csid = cevent->sndrcvinfo.sinfo_stream;
 		cssn = cevent->sndrcvinfo.sinfo_ssn;
 
@@ -390,32 +578,31 @@ static inline void sctp_ulpq_retrieve_ordered(struct sctp_ulpq *ulpq,
 
 		/* Found it, so mark in the ssnmap. */
 		sctp_ssn_next(in, sid);
-	       
+
 		__skb_unlink(pos, pos->list);
 
 		/* Attach all gathered skbs to the event.  */
-		__skb_queue_tail(event->parent->list, pos);
+		__skb_queue_tail(sctp_event2skb(event)->list, pos);
 	}
 }
 
 /* Helper function to store chunks needing ordering.  */
 static inline void sctp_ulpq_store_ordered(struct sctp_ulpq *ulpq,
-					   sctp_ulpevent_t *event)
+					   struct sctp_ulpevent *event)
 {
 	struct sk_buff *pos, *tmp;
-	sctp_ulpevent_t *cevent;
+	struct sctp_ulpevent *cevent;
 	__u16 sid, csid;
 	__u16 ssn, cssn;
 
 	sid = event->sndrcvinfo.sinfo_stream;
 	ssn = event->sndrcvinfo.sinfo_ssn;
 
-
 	/* Find the right place in this list.  We store them by
 	 * stream ID and then by SSN.
 	 */
 	sctp_skb_for_each(pos, &ulpq->lobby, tmp) {
-		cevent = (sctp_ulpevent_t *) pos->cb;
+		cevent = (struct sctp_ulpevent *) pos->cb;
 		csid = cevent->sndrcvinfo.sinfo_stream;
 		cssn = cevent->sndrcvinfo.sinfo_ssn;
 
@@ -427,25 +614,20 @@ static inline void sctp_ulpq_store_ordered(struct sctp_ulpq *ulpq,
 
 	/* If the queue is empty, we have a different function to call.  */
 	if (skb_peek(&ulpq->lobby))
-		__skb_insert(event->parent, pos->prev, pos, &ulpq->lobby);
+		__skb_insert(sctp_event2skb(event), pos->prev, pos,
+			     &ulpq->lobby);
 	else
-		__skb_queue_tail(&ulpq->lobby, event->parent);
+		__skb_queue_tail(&ulpq->lobby, sctp_event2skb(event));
 }
 
-static inline sctp_ulpevent_t *sctp_ulpq_order(struct sctp_ulpq *ulpq,
-					       sctp_ulpevent_t *event)
+static inline struct sctp_ulpevent *sctp_ulpq_order(struct sctp_ulpq *ulpq,
+					struct sctp_ulpevent *event)
 {
 	__u16 sid, ssn;
 	struct sctp_stream *in;
 
-	/* FIXME: We should be using some new chunk structure here
-	 * instead of carrying chunk fields in the event structure.
-	 * This is temporary as it is too painful to change everything
-	 * at once.
-	 */
-
 	/* Check if this message needs ordering.  */
-	if (SCTP_DATA_UNORDERED & event->chunk_flags)
+	if (SCTP_DATA_UNORDERED & event->msg_flags)
 		return event;
 
 	/* Note: The stream ID must be verified before this routine.  */
@@ -472,3 +654,54 @@ static inline sctp_ulpevent_t *sctp_ulpq_order(struct sctp_ulpq *ulpq,
 
 	return event;
 }
+
+/* Partial deliver the first message as there is pressure on rwnd. */
+void sctp_ulpq_partial_delivery(struct sctp_ulpq *ulpq,
+				struct sctp_chunk *chunk, int priority)
+{
+	struct sctp_ulpevent *event;
+
+	/* Are we already in partial delivery mode?  */
+	if (!sctp_sk(ulpq->asoc->base.sk)->pd_mode) {
+
+		/* Is partial delivery possible?  */
+		event = sctp_ulpq_retrieve_first(ulpq);
+		/* Send event to the ULP.   */
+		if (event) {
+			sctp_ulpq_tail_event(ulpq, event);
+			sctp_sk(ulpq->asoc->base.sk)->pd_mode = 1;
+			ulpq->pd_mode = 1;
+			return;
+		}
+	}
+
+	/* Assert:  Either already in partial delivery mode or partial
+	 * delivery wasn't possible, so now the only recourse is
+	 * to renege.  FIXME: Add renege support starts here.
+	 */
+}
+
+/* Notify the application if an association is aborted and in
+ * partial delivery mode.  Send up any pending received messages.
+ */
+void sctp_ulpq_abort_pd(struct sctp_ulpq *ulpq, int priority)
+{
+	struct sctp_ulpevent *ev = NULL;
+	struct sock *sk;
+
+	if (!ulpq->pd_mode)
+		return;
+
+	sk = ulpq->asoc->base.sk;
+	if (sctp_ulpevent_type_enabled(SCTP_PARTIAL_DELIVERY_EVENT,
+				       &sctp_sk(sk)->subscribe))
+		ev = sctp_ulpevent_make_pdapi(ulpq->asoc,
+					      SCTP_PARTIAL_DELIVERY_ABORTED,
+					      priority);
+	if (ev)
+		skb_queue_tail(&sk->receive_queue, sctp_event2skb(ev));
+
+	/* If there is data waiting, send it up the socket now. */
+	if (sctp_ulpq_clear_pd(ulpq) || ev)
+		wake_up_interruptible(sk->sleep);
+}