[PATCH next-2.6] [SCTP] New SCTP functionality

[PATCH next-2.6] [SCTP] New SCTP functionality

[Vlad Yasevich]

The following is a short summary of the new SCTP functionality

* Update implementation of SCTP_DELAYD_ACK to conform to the latest
  spec.  We now implement SACK frequency in the number of packets as well
  as time.

* Support the new sctp_connectx() functionality that returns the
  established association id.

* Implement the remote address table OID for the SCTP MIB.

-vlad

[PATCH next-2.6 1/3] [SCTP]: Bring SCTP_DELAYED_ACK socket option into API compliance

[Vlad Yasevich]

From: Wei Yongjun <yjwei@cn.fujitsu.com>

Brings delayed_ack socket option set/get into line with the latest ietf
socket extensions API draft, while maintaining backwards compatibility.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
---
 include/net/sctp/structs.h |    4 +
 include/net/sctp/user.h    |   28 ++++--
 net/sctp/associola.c       |    3 +
 net/sctp/sm_sideeffect.c   |   17 +++-
 net/sctp/socket.c          |  242 ++++++++++++++++++++++++++++----------------
 5 files changed, 194 insertions(+), 100 deletions(-)

diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h
index 0ce0443..e111517 100644
--- a/include/net/sctp/structs.h
+++ b/include/net/sctp/structs.h
@@ -300,6 +300,7 @@ struct sctp_sock {
 
 	/* The default SACK delay timeout for new associations. */
 	__u32 sackdelay;
+	__u32 sackfreq;
 
 	/* Flags controlling Heartbeat, SACK delay, and Path MTU Discovery. */
 	__u32 param_flags;
@@ -938,6 +939,7 @@ struct sctp_transport {
 
 	/* SACK delay timeout */
 	unsigned long sackdelay;
+	__u32 sackfreq;
 
 	/* When was the last time (in jiffies) that we heard from this
 	 * transport?  We use this to pick new active and retran paths.
@@ -1542,6 +1544,7 @@ struct sctp_association {
 		 *             : SACK's are not delayed (see Section 6).
 		 */
 		__u8    sack_needed;     /* Do we need to sack the peer? */
+		__u32	sack_cnt;
 
 		/* These are capabilities which our peer advertised.  */
 		__u8	ecn_capable;	 /* Can peer do ECN? */
@@ -1651,6 +1654,7 @@ struct sctp_association {
 
 	/* SACK delay timeout */
 	unsigned long sackdelay;
+	__u32 sackfreq;
 
 
 	unsigned long timeouts[SCTP_NUM_TIMEOUT_TYPES];
diff --git a/include/net/sctp/user.h b/include/net/sctp/user.h
index 9619b9d..31371d2 100644
--- a/include/net/sctp/user.h
+++ b/include/net/sctp/user.h
@@ -93,8 +93,9 @@ enum sctp_optname {
 #define SCTP_STATUS SCTP_STATUS
 	SCTP_GET_PEER_ADDR_INFO,
 #define SCTP_GET_PEER_ADDR_INFO SCTP_GET_PEER_ADDR_INFO
-	SCTP_DELAYED_ACK_TIME,
-#define SCTP_DELAYED_ACK_TIME SCTP_DELAYED_ACK_TIME
+	SCTP_DELAYED_ACK,
+#define SCTP_DELAYED_ACK_TIME SCTP_DELAYED_ACK
+#define SCTP_DELAYED_ACK SCTP_DELAYED_ACK
 	SCTP_CONTEXT,	/* Receive Context */
 #define SCTP_CONTEXT SCTP_CONTEXT
 	SCTP_FRAGMENT_INTERLEAVE,
@@ -618,13 +619,26 @@ struct sctp_authkeyid {
 };
 
 
-/* 7.1.23. Delayed Ack Timer (SCTP_DELAYED_ACK_TIME)
+/*
+ * 7.1.23.  Get or set delayed ack timer (SCTP_DELAYED_SACK)
  *
- *   This options will get or set the delayed ack timer.  The time is set
- *   in milliseconds.  If the assoc_id is 0, then this sets or gets the
- *   endpoints default delayed ack timer value.  If the assoc_id field is
- *   non-zero, then the set or get effects the specified association.
+ * This option will effect the way delayed acks are performed.  This
+ * option allows you to get or set the delayed ack time, in
+ * milliseconds.  It also allows changing the delayed ack frequency.
+ * Changing the frequency to 1 disables the delayed sack algorithm.  If
+ * the assoc_id is 0, then this sets or gets the endpoints default
+ * values.  If the assoc_id field is non-zero, then the set or get
+ * effects the specified association for the one to many model (the
+ * assoc_id field is ignored by the one to one model).  Note that if
+ * sack_delay or sack_freq are 0 when setting this option, then the
+ * current values will remain unchanged.
  */
+struct sctp_sack_info {
+	sctp_assoc_t	sack_assoc_id;
+	uint32_t	sack_delay;
+	uint32_t	sack_freq;
+};
+
 struct sctp_assoc_value {
     sctp_assoc_t            assoc_id;
     uint32_t                assoc_value;
diff --git a/net/sctp/associola.c b/net/sctp/associola.c
index b4cd2b7..7b79d1e 100644
--- a/net/sctp/associola.c
+++ b/net/sctp/associola.c
@@ -136,6 +136,7 @@ static struct sctp_association *sctp_association_init(struct sctp_association *a
 
 	/* Set association default SACK delay */
 	asoc->sackdelay = msecs_to_jiffies(sp->sackdelay);
+	asoc->sackfreq = sp->sackfreq;
 
 	/* Set the association default flags controlling
 	 * Heartbeat, SACK delay, and Path MTU Discovery.
@@ -261,6 +262,7 @@ static struct sctp_association *sctp_association_init(struct sctp_association *a
 	 * already received one packet.]
 	 */
 	asoc->peer.sack_needed = 1;
+	asoc->peer.sack_cnt = 0;
 
 	/* Assume that the peer will tell us if he recognizes ASCONF
 	 * as part of INIT exchange.
@@ -615,6 +617,7 @@ struct sctp_transport *sctp_assoc_add_peer(struct sctp_association *asoc,
 	 * association configured value.
 	 */
 	peer->sackdelay = asoc->sackdelay;
+	peer->sackfreq = asoc->sackfreq;
 
 	/* Enable/disable heartbeat, SACK delay, and path MTU discovery
 	 * based on association setting.
diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
index 23a9f1a..b083312 100644
--- a/net/sctp/sm_sideeffect.c
+++ b/net/sctp/sm_sideeffect.c
@@ -190,20 +190,28 @@ static int sctp_gen_sack(struct sctp_association *asoc, int force,
 	 * unacknowledged DATA chunk. ...
 	 */
 	if (!asoc->peer.sack_needed) {
-		/* We will need a SACK for the next packet.  */
-		asoc->peer.sack_needed = 1;
+		asoc->peer.sack_cnt++;
 
 		/* Set the SACK delay timeout based on the
 		 * SACK delay for the last transport
 		 * data was received from, or the default
 		 * for the association.
 		 */
-		if (trans)
+		if (trans) {
+			/* We will need a SACK for the next packet.  */
+			if (asoc->peer.sack_cnt >= trans->sackfreq - 1)
+				asoc->peer.sack_needed = 1;
+
 			asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] =
 				trans->sackdelay;
-		else
+		} else {
+			/* We will need a SACK for the next packet.  */
+			if (asoc->peer.sack_cnt >= asoc->sackfreq - 1)
+				asoc->peer.sack_needed = 1;
+
 			asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] =
 				asoc->sackdelay;
+		}
 
 		/* Restart the SACK timer. */
 		sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_RESTART,
@@ -216,6 +224,7 @@ static int sctp_gen_sack(struct sctp_association *asoc, int force,
 			goto nomem;
 
 		asoc->peer.sack_needed = 0;
+		asoc->peer.sack_cnt = 0;
 
 		sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(sack));
 
diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index e7e3baf..6698587 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -2305,74 +2305,98 @@ static int sctp_setsockopt_peer_addr_params(struct sock *sk,
 	return 0;
 }
 
-/* 7.1.23. Delayed Ack Timer (SCTP_DELAYED_ACK_TIME)
- *
- *   This options will get or set the delayed ack timer.  The time is set
- *   in milliseconds.  If the assoc_id is 0, then this sets or gets the
- *   endpoints default delayed ack timer value.  If the assoc_id field is
- *   non-zero, then the set or get effects the specified association.
- *
- *   struct sctp_assoc_value {
- *       sctp_assoc_t            assoc_id;
- *       uint32_t                assoc_value;
- *   };
- *
- *     assoc_id    - This parameter, indicates which association the
- *                   user is preforming an action upon. Note that if
- *                   this field's value is zero then the endpoints
- *                   default value is changed (effecting future
- *                   associations only).
- *
- *     assoc_value - This parameter contains the number of milliseconds
- *                   that the user is requesting the delayed ACK timer
- *                   be set to. Note that this value is defined in
- *                   the standard to be between 200 and 500 milliseconds.
- *
- *                   Note: a value of zero will leave the value alone,
- *                   but disable SACK delay. A non-zero value will also
- *                   enable SACK delay.
+/*
+ * 7.1.23.  Get or set delayed ack timer (SCTP_DELAYED_SACK)
+ *
+ * This option will effect the way delayed acks are performed.  This
+ * option allows you to get or set the delayed ack time, in
+ * milliseconds.  It also allows changing the delayed ack frequency.
+ * Changing the frequency to 1 disables the delayed sack algorithm.  If
+ * the assoc_id is 0, then this sets or gets the endpoints default
+ * values.  If the assoc_id field is non-zero, then the set or get
+ * effects the specified association for the one to many model (the
+ * assoc_id field is ignored by the one to one model).  Note that if
+ * sack_delay or sack_freq are 0 when setting this option, then the
+ * current values will remain unchanged.
+ *
+ * struct sctp_sack_info {
+ *     sctp_assoc_t            sack_assoc_id;
+ *     uint32_t                sack_delay;
+ *     uint32_t                sack_freq;
+ * };
+ *
+ * sack_assoc_id -  This parameter, indicates which association the user
+ *    is performing an action upon.  Note that if this field's value is
+ *    zero then the endpoints default value is changed (effecting future
+ *    associations only).
+ *
+ * sack_delay -  This parameter contains the number of milliseconds that
+ *    the user is requesting the delayed ACK timer be set to.  Note that
+ *    this value is defined in the standard to be between 200 and 500
+ *    milliseconds.
+ *
+ * sack_freq -  This parameter contains the number of packets that must
+ *    be received before a sack is sent without waiting for the delay
+ *    timer to expire.  The default value for this is 2, setting this
+ *    value to 1 will disable the delayed sack algorithm.
  */
 
-static int sctp_setsockopt_delayed_ack_time(struct sock *sk,
+static int sctp_setsockopt_delayed_ack(struct sock *sk,
 					    char __user *optval, int optlen)
 {
-	struct sctp_assoc_value  params;
+	struct sctp_sack_info    params;
 	struct sctp_transport   *trans = NULL;
 	struct sctp_association *asoc = NULL;
 	struct sctp_sock        *sp = sctp_sk(sk);
 
-	if (optlen != sizeof(struct sctp_assoc_value))
-		return - EINVAL;
+	if (optlen == sizeof(struct sctp_sack_info)) {
+		if (copy_from_user(&params, optval, optlen))
+			return -EFAULT;
 
-	if (copy_from_user(&params, optval, optlen))
-		return -EFAULT;
+		if (params.sack_delay == 0 && params.sack_freq == 0)
+			return 0;
+	} else if (optlen == sizeof(struct sctp_assoc_value)) {
+		printk(KERN_WARNING "SCTP: Use of struct sctp_sack_info "
+		       "in delayed_ack socket option deprecated\n");
+		printk(KERN_WARNING "SCTP: struct sctp_sack_info instead\n");
+		if (copy_from_user(&params, optval, optlen))
+			return -EFAULT;
+
+		if (params.sack_delay == 0)
+			params.sack_freq = 1;
+		else
+			params.sack_freq = 0;
+	} else
+		return - EINVAL;
 
 	/* Validate value parameter. */
-	if (params.assoc_value > 500)
+	if (params.sack_delay > 500)
 		return -EINVAL;
 
-	/* Get association, if assoc_id != 0 and the socket is a one
+	/* Get association, if sack_assoc_id != 0 and the socket is a one
 	 * to many style socket, and an association was not found, then
 	 * the id was invalid.
 	 */
-	asoc = sctp_id2assoc(sk, params.assoc_id);
-	if (!asoc && params.assoc_id && sctp_style(sk, UDP))
+	asoc = sctp_id2assoc(sk, params.sack_assoc_id);
+	if (!asoc && params.sack_assoc_id && sctp_style(sk, UDP))
 		return -EINVAL;
 
-	if (params.assoc_value) {
+	if (params.sack_delay) {
 		if (asoc) {
 			asoc->sackdelay =
-				msecs_to_jiffies(params.assoc_value);
+				msecs_to_jiffies(params.sack_delay);
 			asoc->param_flags =
 				(asoc->param_flags & ~SPP_SACKDELAY) |
 				SPP_SACKDELAY_ENABLE;
 		} else {
-			sp->sackdelay = params.assoc_value;
+			sp->sackdelay = params.sack_delay;
 			sp->param_flags =
 				(sp->param_flags & ~SPP_SACKDELAY) |
 				SPP_SACKDELAY_ENABLE;
 		}
-	} else {
+	}
+
+	if (params.sack_freq == 1) {
 		if (asoc) {
 			asoc->param_flags =
 				(asoc->param_flags & ~SPP_SACKDELAY) |
@@ -2382,22 +2406,40 @@ static int sctp_setsockopt_delayed_ack_time(struct sock *sk,
 				(sp->param_flags & ~SPP_SACKDELAY) |
 				SPP_SACKDELAY_DISABLE;
 		}
+	} else if (params.sack_freq > 1) {
+		if (asoc) {
+			asoc->sackfreq = params.sack_freq;
+			asoc->param_flags =
+				(asoc->param_flags & ~SPP_SACKDELAY) |
+				SPP_SACKDELAY_ENABLE;
+		} else {
+			sp->sackfreq = params.sack_freq;
+			sp->param_flags =
+				(sp->param_flags & ~SPP_SACKDELAY) |
+				SPP_SACKDELAY_ENABLE;
+		}
 	}
 
 	/* If change is for association, also apply to each transport. */
 	if (asoc) {
 		list_for_each_entry(trans, &asoc->peer.transport_addr_list,
 				transports) {
-			if (params.assoc_value) {
+			if (params.sack_delay) {
 				trans->sackdelay =
-					msecs_to_jiffies(params.assoc_value);
+					msecs_to_jiffies(params.sack_delay);
 				trans->param_flags =
 					(trans->param_flags & ~SPP_SACKDELAY) |
 					SPP_SACKDELAY_ENABLE;
-			} else {
+			}
+			if (params.sack_delay == 1) {
 				trans->param_flags =
 					(trans->param_flags & ~SPP_SACKDELAY) |
 					SPP_SACKDELAY_DISABLE;
+			} else if (params.sack_freq > 1) {
+				trans->sackfreq = params.sack_freq;
+				trans->param_flags =
+					(trans->param_flags & ~SPP_SACKDELAY) |
+					SPP_SACKDELAY_ENABLE;
 			}
 		}
 	}
@@ -3186,8 +3228,8 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 		retval = sctp_setsockopt_peer_addr_params(sk, optval, optlen);
 		break;
 
-	case SCTP_DELAYED_ACK_TIME:
-		retval = sctp_setsockopt_delayed_ack_time(sk, optval, optlen);
+	case SCTP_DELAYED_ACK:
+		retval = sctp_setsockopt_delayed_ack(sk, optval, optlen);
 		break;
 	case SCTP_PARTIAL_DELIVERY_POINT:
 		retval = sctp_setsockopt_partial_delivery_point(sk, optval, optlen);
@@ -3446,6 +3488,7 @@ SCTP_STATIC int sctp_init_sock(struct sock *sk)
 	sp->pathmaxrxt  = sctp_max_retrans_path;
 	sp->pathmtu     = 0; // allow default discovery
 	sp->sackdelay   = sctp_sack_timeout;
+	sp->sackfreq	= 3;
 	sp->param_flags = SPP_HB_ENABLE |
 			  SPP_PMTUD_ENABLE |
 			  SPP_SACKDELAY_ENABLE;
@@ -3999,70 +4042,91 @@ static int sctp_getsockopt_peer_addr_params(struct sock *sk, int len,
 	return 0;
 }
 
-/* 7.1.23. Delayed Ack Timer (SCTP_DELAYED_ACK_TIME)
- *
- *   This options will get or set the delayed ack timer.  The time is set
- *   in milliseconds.  If the assoc_id is 0, then this sets or gets the
- *   endpoints default delayed ack timer value.  If the assoc_id field is
- *   non-zero, then the set or get effects the specified association.
- *
- *   struct sctp_assoc_value {
- *       sctp_assoc_t            assoc_id;
- *       uint32_t                assoc_value;
- *   };
+/*
+ * 7.1.23.  Get or set delayed ack timer (SCTP_DELAYED_SACK)
+ *
+ * This option will effect the way delayed acks are performed.  This
+ * option allows you to get or set the delayed ack time, in
+ * milliseconds.  It also allows changing the delayed ack frequency.
+ * Changing the frequency to 1 disables the delayed sack algorithm.  If
+ * the assoc_id is 0, then this sets or gets the endpoints default
+ * values.  If the assoc_id field is non-zero, then the set or get
+ * effects the specified association for the one to many model (the
+ * assoc_id field is ignored by the one to one model).  Note that if
+ * sack_delay or sack_freq are 0 when setting this option, then the
+ * current values will remain unchanged.
+ *
+ * struct sctp_sack_info {
+ *     sctp_assoc_t            sack_assoc_id;
+ *     uint32_t                sack_delay;
+ *     uint32_t                sack_freq;
+ * };
  *
- *     assoc_id    - This parameter, indicates which association the
- *                   user is preforming an action upon. Note that if
- *                   this field's value is zero then the endpoints
- *                   default value is changed (effecting future
- *                   associations only).
+ * sack_assoc_id -  This parameter, indicates which association the user
+ *    is performing an action upon.  Note that if this field's value is
+ *    zero then the endpoints default value is changed (effecting future
+ *    associations only).
  *
- *     assoc_value - This parameter contains the number of milliseconds
- *                   that the user is requesting the delayed ACK timer
- *                   be set to. Note that this value is defined in
- *                   the standard to be between 200 and 500 milliseconds.
+ * sack_delay -  This parameter contains the number of milliseconds that
+ *    the user is requesting the delayed ACK timer be set to.  Note that
+ *    this value is defined in the standard to be between 200 and 500
+ *    milliseconds.
  *
- *                   Note: a value of zero will leave the value alone,
- *                   but disable SACK delay. A non-zero value will also
- *                   enable SACK delay.
+ * sack_freq -  This parameter contains the number of packets that must
+ *    be received before a sack is sent without waiting for the delay
+ *    timer to expire.  The default value for this is 2, setting this
+ *    value to 1 will disable the delayed sack algorithm.
  */
-static int sctp_getsockopt_delayed_ack_time(struct sock *sk, int len,
+static int sctp_getsockopt_delayed_ack(struct sock *sk, int len,
 					    char __user *optval,
 					    int __user *optlen)
 {
-	struct sctp_assoc_value  params;
+	struct sctp_sack_info    params;
 	struct sctp_association *asoc = NULL;
 	struct sctp_sock        *sp = sctp_sk(sk);
 
-	if (len < sizeof(struct sctp_assoc_value))
-		return - EINVAL;
+	if (len >= sizeof(struct sctp_sack_info)) {
+		len = sizeof(struct sctp_sack_info);
 
-	len = sizeof(struct sctp_assoc_value);
-
-	if (copy_from_user(&params, optval, len))
-		return -EFAULT;
+		if (copy_from_user(&params, optval, len))
+			return -EFAULT;
+	} else if (len == sizeof(struct sctp_assoc_value)) {
+		printk(KERN_WARNING "SCTP: Use of struct sctp_sack_info "
+		       "in delayed_ack socket option deprecated\n");
+		printk(KERN_WARNING "SCTP: struct sctp_sack_info instead\n");
+		if (copy_from_user(&params, optval, len))
+			return -EFAULT;
+	} else
+		return - EINVAL;
 
-	/* Get association, if assoc_id != 0 and the socket is a one
+	/* Get association, if sack_assoc_id != 0 and the socket is a one
 	 * to many style socket, and an association was not found, then
 	 * the id was invalid.
 	 */
-	asoc = sctp_id2assoc(sk, params.assoc_id);
-	if (!asoc && params.assoc_id && sctp_style(sk, UDP))
+	asoc = sctp_id2assoc(sk, params.sack_assoc_id);
+	if (!asoc && params.sack_assoc_id && sctp_style(sk, UDP))
 		return -EINVAL;
 
 	if (asoc) {
 		/* Fetch association values. */
-		if (asoc->param_flags & SPP_SACKDELAY_ENABLE)
-			params.assoc_value = jiffies_to_msecs(
+		if (asoc->param_flags & SPP_SACKDELAY_ENABLE) {
+			params.sack_delay = jiffies_to_msecs(
 				asoc->sackdelay);
-		else
-			params.assoc_value = 0;
+			params.sack_freq = asoc->sackfreq;
+
+		} else {
+			params.sack_delay = 0;
+			params.sack_freq = 1;
+		}
 	} else {
 		/* Fetch socket values. */
-		if (sp->param_flags & SPP_SACKDELAY_ENABLE)
-			params.assoc_value  = sp->sackdelay;
-		else
-			params.assoc_value  = 0;
+		if (sp->param_flags & SPP_SACKDELAY_ENABLE) {
+			params.sack_delay  = sp->sackdelay;
+			params.sack_freq = sp->sackfreq;
+		} else {
+			params.sack_delay  = 0;
+			params.sack_freq = 1;
+		}
 	}
 
 	if (copy_to_user(optval, &params, len))
@@ -5218,8 +5282,8 @@ SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
 		retval = sctp_getsockopt_peer_addr_params(sk, len, optval,
 							  optlen);
 		break;
-	case SCTP_DELAYED_ACK_TIME:
-		retval = sctp_getsockopt_delayed_ack_time(sk, len, optval,
+	case SCTP_DELAYED_ACK:
+		retval = sctp_getsockopt_delayed_ack(sk, len, optval,
 							  optlen);
 		break;
 	case SCTP_INITMSG:
-- 
1.5.3.5

Re: [PATCH next-2.6 1/3] [SCTP]: Bring SCTP_DELAYED_ACK socket option into API compliance

[David Miller]

From: Vlad Yasevich <vladislav.yasevich@hp.com>
Date: Fri,  9 May 2008 15:16:04 -0400

> Brings delayed_ack socket option set/get into line with the latest ietf
> socket extensions API draft, while maintaining backwards compatibility.
> 
> Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
> Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>

Applied.

[PATCH next-2.6 2/3] [SCTP]: Support the new specification of sctp_connectx()

[Vlad Yasevich]

The specification of sctp_connectx() has been changed to return
an association id.  We've added a new socket option that will
return the association id as the return value from the setsockopt()
call.  The library that implements sctp_connectx() interface will
implement both socket options.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
---
 include/net/sctp/user.h |    6 +++-
 net/sctp/socket.c       |   70 +++++++++++++++++++++++++++++++++++++++-------
 2 files changed, 63 insertions(+), 13 deletions(-)

diff --git a/include/net/sctp/user.h b/include/net/sctp/user.h
index 31371d2..f205b10 100644
--- a/include/net/sctp/user.h
+++ b/include/net/sctp/user.h
@@ -137,12 +137,14 @@ enum sctp_optname {
 #define SCTP_GET_LOCAL_ADDRS_NUM_OLD	SCTP_GET_LOCAL_ADDRS_NUM_OLD
 	SCTP_GET_LOCAL_ADDRS_OLD, 	/* Get all local addresss. */
 #define SCTP_GET_LOCAL_ADDRS_OLD	SCTP_GET_LOCAL_ADDRS_OLD
-	SCTP_SOCKOPT_CONNECTX, /* CONNECTX requests. */
-#define SCTP_SOCKOPT_CONNECTX	SCTP_SOCKOPT_CONNECTX
+	SCTP_SOCKOPT_CONNECTX_OLD, /* CONNECTX old requests. */
+#define SCTP_SOCKOPT_CONNECTX_OLD	SCTP_SOCKOPT_CONNECTX_OLD
 	SCTP_GET_PEER_ADDRS, 	/* Get all peer addresss. */
 #define SCTP_GET_PEER_ADDRS	SCTP_GET_PEER_ADDRS
 	SCTP_GET_LOCAL_ADDRS, 	/* Get all local addresss. */
 #define SCTP_GET_LOCAL_ADDRS	SCTP_GET_LOCAL_ADDRS
+	SCTP_SOCKOPT_CONNECTX, /* CONNECTX requests. */
+#define SCTP_SOCKOPT_CONNECTX	SCTP_SOCKOPT_CONNECTX
 };
 
 /*
diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index 6698587..81600ee 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -956,7 +956,8 @@ out:
  */
 static int __sctp_connect(struct sock* sk,
 			  struct sockaddr *kaddrs,
-			  int addrs_size)
+			  int addrs_size,
+			  sctp_assoc_t *assoc_id)
 {
 	struct sctp_sock *sp;
 	struct sctp_endpoint *ep;
@@ -1111,6 +1112,8 @@ static int __sctp_connect(struct sock* sk,
 	timeo = sock_sndtimeo(sk, f_flags & O_NONBLOCK);
 
 	err = sctp_wait_for_connect(asoc, &timeo);
+	if (!err && assoc_id)
+		*assoc_id = asoc->assoc_id;
 
 	/* Don't free association on exit. */
 	asoc = NULL;
@@ -1128,7 +1131,8 @@ out_free:
 /* Helper for tunneling sctp_connectx() requests through sctp_setsockopt()
  *
  * API 8.9
- * int sctp_connectx(int sd, struct sockaddr *addrs, int addrcnt);
+ * int sctp_connectx(int sd, struct sockaddr *addrs, int addrcnt,
+ * 			sctp_assoc_t *asoc);
  *
  * If sd is an IPv4 socket, the addresses passed must be IPv4 addresses.
  * If the sd is an IPv6 socket, the addresses passed can either be IPv4
@@ -1144,8 +1148,10 @@ out_free:
  * representation is termed a "packed array" of addresses). The caller
  * specifies the number of addresses in the array with addrcnt.
  *
- * On success, sctp_connectx() returns 0. On failure, sctp_connectx() returns
- * -1, and sets errno to the appropriate error code.
+ * On success, sctp_connectx() returns 0. It also sets the assoc_id to
+ * the association id of the new association.  On failure, sctp_connectx()
+ * returns -1, and sets errno to the appropriate error code.  The assoc_id
+ * is not touched by the kernel.
  *
  * For SCTP, the port given in each socket address must be the same, or
  * sctp_connectx() will fail, setting errno to EINVAL.
@@ -1182,11 +1188,12 @@ out_free:
  * addrs     The pointer to the addresses in user land
  * addrssize Size of the addrs buffer
  *
- * Returns 0 if ok, <0 errno code on error.
+ * Returns >=0 if ok, <0 errno code on error.
  */
-SCTP_STATIC int sctp_setsockopt_connectx(struct sock* sk,
+SCTP_STATIC int __sctp_setsockopt_connectx(struct sock* sk,
 				      struct sockaddr __user *addrs,
-				      int addrs_size)
+				      int addrs_size,
+				      sctp_assoc_t *assoc_id)
 {
 	int err = 0;
 	struct sockaddr *kaddrs;
@@ -1209,13 +1216,46 @@ SCTP_STATIC int sctp_setsockopt_connectx(struct sock* sk,
 	if (__copy_from_user(kaddrs, addrs, addrs_size)) {
 		err = -EFAULT;
 	} else {
-		err = __sctp_connect(sk, kaddrs, addrs_size);
+		err = __sctp_connect(sk, kaddrs, addrs_size, assoc_id);
 	}
 
 	kfree(kaddrs);
+
 	return err;
 }
 
+/*
+ * This is an older interface.  It's kept for backward compatibility
+ * to the option that doesn't provide association id.
+ */
+SCTP_STATIC int sctp_setsockopt_connectx_old(struct sock* sk,
+				      struct sockaddr __user *addrs,
+				      int addrs_size)
+{
+	return __sctp_setsockopt_connectx(sk, addrs, addrs_size, NULL);
+}
+
+/*
+ * New interface for the API.  The since the API is done with a socket
+ * option, to make it simple we feed back the association id is as a return
+ * indication to the call.  Error is always negative and association id is
+ * always positive.
+ */
+SCTP_STATIC int sctp_setsockopt_connectx(struct sock* sk,
+				      struct sockaddr __user *addrs,
+				      int addrs_size)
+{
+	sctp_assoc_t assoc_id = 0;
+	int err = 0;
+
+	err = __sctp_setsockopt_connectx(sk, addrs, addrs_size, &assoc_id);
+
+	if (err)
+		return err;
+	else
+		return assoc_id;
+}
+
 /* API 3.1.4 close() - UDP Style Syntax
  * Applications use close() to perform graceful shutdown (as described in
  * Section 10.1 of [SCTP]) on ALL the associations currently represented
@@ -3206,10 +3246,18 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
 					       optlen, SCTP_BINDX_REM_ADDR);
 		break;
 
+	case SCTP_SOCKOPT_CONNECTX_OLD:
+		/* 'optlen' is the size of the addresses buffer. */
+		retval = sctp_setsockopt_connectx_old(sk,
+					    (struct sockaddr __user *)optval,
+					    optlen);
+		break;
+
 	case SCTP_SOCKOPT_CONNECTX:
 		/* 'optlen' is the size of the addresses buffer. */
-		retval = sctp_setsockopt_connectx(sk, (struct sockaddr __user *)optval,
-					       optlen);
+		retval = sctp_setsockopt_connectx(sk,
+					    (struct sockaddr __user *)optval,
+					    optlen);
 		break;
 
 	case SCTP_DISABLE_FRAGMENTS:
@@ -3336,7 +3384,7 @@ SCTP_STATIC int sctp_connect(struct sock *sk, struct sockaddr *addr,
 		/* Pass correct addr len to common routine (so it knows there
 		 * is only one address being passed.
 		 */
-		err = __sctp_connect(sk, addr, af->sockaddr_len);
+		err = __sctp_connect(sk, addr, af->sockaddr_len, NULL);
 	}
 
 	sctp_release_sock(sk);
-- 
1.5.3.5

Re: [PATCH next-2.6 2/3] [SCTP]: Support the new specification of sctp_connectx()

[David Miller]

From: Vlad Yasevich <vladislav.yasevich@hp.com>
Date: Fri,  9 May 2008 15:16:05 -0400

> The specification of sctp_connectx() has been changed to return
> an association id.  We've added a new socket option that will
> return the association id as the return value from the setsockopt()
> call.  The library that implements sctp_connectx() interface will
> implement both socket options.
> 
> Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>

Applied.

[PATCH next-2.6 3/3] [SCTP]: add sctp/remaddr table to complete RFC remote address table OID

[Vlad Yasevich]

From: Neil Horman <nhorman@tuxdriver.com>

Add support for RFC3873 remote address table OID.

      +--(5) sctpAssocRemAddrTable
      |   |
      |   |--(-) sctpAssocId (shared index)
      |   |
      |   +--(1) sctpAssocRemAddrType (index)
      .   |
      .   +--(2) sctpAssocRemAddr (index)
      .   |
          +--(3) sctpAssocRemAddrActive
          |
          +--(4) sctpAssocRemAddrHBActive
          |
          +--(5) sctpAssocRemAddrRTO
          |
          +--(6) sctpAssocRemAddrMaxPathRtx
          |
          +--(7) sctpAssocRemAddrRtx
          |
          +--(8) sctpAssocRemAddrStartTime

This patch places all the requsite data in /proc/net/sctp/remaddr.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
---
 include/net/sctp/sctp.h |    2 +
 net/sctp/proc.c         |  141 +++++++++++++++++++++++++++++++++++++++++++++++
 net/sctp/protocol.c     |    3 +
 3 files changed, 146 insertions(+), 0 deletions(-)

diff --git a/include/net/sctp/sctp.h b/include/net/sctp/sctp.h
index 90b1e8d..5672d48 100644
--- a/include/net/sctp/sctp.h
+++ b/include/net/sctp/sctp.h
@@ -179,6 +179,8 @@ int sctp_eps_proc_init(void);
 void sctp_eps_proc_exit(void);
 int sctp_assocs_proc_init(void);
 void sctp_assocs_proc_exit(void);
+int sctp_remaddr_proc_init(void);
+void sctp_remaddr_proc_exit(void);
 
 
 /*
diff --git a/net/sctp/proc.c b/net/sctp/proc.c
index 0aba759..5dd8983 100644
--- a/net/sctp/proc.c
+++ b/net/sctp/proc.c
@@ -383,3 +383,144 @@ void sctp_assocs_proc_exit(void)
 {
 	remove_proc_entry("assocs", proc_net_sctp);
 }
+
+static void *sctp_remaddr_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	if (*pos >= sctp_assoc_hashsize)
+		return NULL;
+
+	if (*pos < 0)
+		*pos = 0;
+
+	if (*pos == 0)
+		seq_printf(seq, "ADDR ASSOC_ID HB_ACT RTO MAX_PATH_RTX "
+				"REM_ADDR_RTX  START\n");
+
+	return (void *)pos;
+}
+
+static void *sctp_remaddr_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	if (++*pos >= sctp_assoc_hashsize)
+		return NULL;
+
+	return pos;
+}
+
+static void sctp_remaddr_seq_stop(struct seq_file *seq, void *v)
+{
+	return;
+}
+
+static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
+{
+	struct sctp_hashbucket *head;
+	struct sctp_ep_common *epb;
+	struct sctp_association *assoc;
+	struct hlist_node *node;
+	struct sctp_transport *tsp;
+	int    hash = *(loff_t *)v;
+
+	if (hash >= sctp_assoc_hashsize)
+		return -ENOMEM;
+
+	head = &sctp_assoc_hashtable[hash];
+	sctp_local_bh_disable();
+	read_lock(&head->lock);
+	sctp_for_each_hentry(epb, node, &head->chain) {
+		assoc = sctp_assoc(epb);
+		list_for_each_entry(tsp, &assoc->peer.transport_addr_list,
+					transports) {
+			/*
+			 * The remote address (ADDR)
+			 */
+			tsp->af_specific->seq_dump_addr(seq, &tsp->ipaddr);
+			seq_printf(seq, " ");
+
+			/*
+			 * The association ID (ASSOC_ID)
+			 */
+			seq_printf(seq, "%d ", tsp->asoc->assoc_id);
+
+			/*
+			 * If the Heartbeat is active (HB_ACT)
+			 * Note: 1 = Active, 0 = Inactive
+			 */
+			seq_printf(seq, "%d ", timer_pending(&tsp->hb_timer));
+
+			/*
+			 * Retransmit time out (RTO)
+			 */
+			seq_printf(seq, "%lu ", tsp->rto);
+
+			/*
+			 * Maximum path retransmit count (PATH_MAX_RTX)
+			 */
+			seq_printf(seq, "%d ", tsp->pathmaxrxt);
+
+			/*
+			 * remote address retransmit count (REM_ADDR_RTX)
+			 * Note: We don't have a way to tally this at the moment
+			 * so lets just leave it as zero for the moment
+			 */
+			seq_printf(seq, "0 ");
+
+			/*
+			 * remote address start time (START).  This is also not
+			 * currently implemented, but we can record it with a
+			 * jiffies marker in a subsequent patch
+			 */
+			seq_printf(seq, "0");
+
+			seq_printf(seq, "\n");
+		}
+	}
+
+	read_unlock(&head->lock);
+	sctp_local_bh_enable();
+
+	return 0;
+
+}
+
+static const struct seq_operations sctp_remaddr_ops = {
+	.start = sctp_remaddr_seq_start,
+	.next  = sctp_remaddr_seq_next,
+	.stop  = sctp_remaddr_seq_stop,
+	.show  = sctp_remaddr_seq_show,
+};
+
+/* Cleanup the proc fs entry for 'remaddr' object. */
+void sctp_remaddr_proc_exit(void)
+{
+	remove_proc_entry("remaddr", proc_net_sctp);
+}
+
+static int sctp_remaddr_seq_open(struct inode *inode, struct file *file)
+{
+	return seq_open(file, &sctp_remaddr_ops);
+}
+
+static const struct file_operations sctp_remaddr_seq_fops = {
+	.open = sctp_remaddr_seq_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = seq_release,
+};
+
+int __init sctp_remaddr_proc_init(void)
+{
+	struct proc_dir_entry *p;
+
+	p = create_proc_entry("remaddr", S_IRUGO, proc_net_sctp);
+	if (!p)
+		return -ENOMEM;
+	p->proc_fops = &sctp_remaddr_seq_fops;
+
+	return 0;
+}
+
+void sctp_assoc_proc_exit(void)
+{
+	remove_proc_entry("remaddr", proc_net_sctp);
+}
diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c
index 0ec234b..b8bd9e0 100644
--- a/net/sctp/protocol.c
+++ b/net/sctp/protocol.c
@@ -113,6 +113,8 @@ static __init int sctp_proc_init(void)
 		goto out_nomem;
 	if (sctp_assocs_proc_init())
 		goto out_nomem;
+	if (sctp_remaddr_proc_init())
+		goto out_nomem;
 
 	return 0;
 
@@ -129,6 +131,7 @@ static void sctp_proc_exit(void)
 	sctp_snmp_proc_exit();
 	sctp_eps_proc_exit();
 	sctp_assocs_proc_exit();
+	sctp_remaddr_proc_exit();
 
 	if (proc_net_sctp) {
 		proc_net_sctp = NULL;
-- 
1.5.3.5

Re: [PATCH next-2.6 3/3] [SCTP]: add sctp/remaddr table to complete RFC remote address table OID

[David Miller]

From: Vlad Yasevich <vladislav.yasevich@hp.com>
Date: Fri,  9 May 2008 15:16:06 -0400

> Add support for RFC3873 remote address table OID.
> 
>       +--(5) sctpAssocRemAddrTable
>       |   |
>       |   |--(-) sctpAssocId (shared index)
>       |   |
>       |   +--(1) sctpAssocRemAddrType (index)
>       .   |
>       .   +--(2) sctpAssocRemAddr (index)
>       .   |
>           +--(3) sctpAssocRemAddrActive
>           |
>           +--(4) sctpAssocRemAddrHBActive
>           |
>           +--(5) sctpAssocRemAddrRTO
>           |
>           +--(6) sctpAssocRemAddrMaxPathRtx
>           |
>           +--(7) sctpAssocRemAddrRtx
>           |
>           +--(8) sctpAssocRemAddrStartTime
> 
> This patch places all the requsite data in /proc/net/sctp/remaddr.
> 
> Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>

Applied.

Re: [PATCH next-2.6 3/3] [SCTP]: add sctp/remaddr table to complete RFC remote address table OID

[Neil Horman]

On Fri, May 09, 2008 at 03:15:18PM -0700, David Miller wrote:
> From: Vlad Yasevich <vladislav.yasevich@hp.com>
> Date: Fri,  9 May 2008 15:16:06 -0400
> 
> > Add support for RFC3873 remote address table OID.
> > 
> >       +--(5) sctpAssocRemAddrTable
> >       |   |
> >       |   |--(-) sctpAssocId (shared index)
> >       |   |
> >       |   +--(1) sctpAssocRemAddrType (index)
> >       .   |
> >       .   +--(2) sctpAssocRemAddr (index)
> >       .   |
> >           +--(3) sctpAssocRemAddrActive
> >           |
> >           +--(4) sctpAssocRemAddrHBActive
> >           |
> >           +--(5) sctpAssocRemAddrRTO
> >           |
> >           +--(6) sctpAssocRemAddrMaxPathRtx
> >           |
> >           +--(7) sctpAssocRemAddrRtx
> >           |
> >           +--(8) sctpAssocRemAddrStartTime
> > 
> > This patch places all the requsite data in /proc/net/sctp/remaddr.
> > 
> > Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> > Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
> 
> Applied.

Dave, where is this applied?  I don't see it in either your net or net-next
trees?

Thanks & Regards
Neil

-- 
/****************************************************
 * Neil Horman <nhorman@tuxdriver.com>
 * Software Engineer, Red Hat
 ****************************************************/

Re: [PATCH next-2.6 3/3] [SCTP]: add sctp/remaddr table to complete RFC remote address table OID

[David Miller]

From: Neil Horman <nhorman@tuxdriver.com>
Date: Mon, 12 May 2008 10:37:36 -0400

> Dave, where is this applied?  I don't see it in either your net or net-next
> trees?

My bad, sorry.

I didn't want to push this batch out until I did some sanity builds.

I did that just now and pushed everything out to net-next-2.6.git

Re: [PATCH next-2.6 3/3] [SCTP]: add sctp/remaddr table to complete RFC remote address table OID

[Neil Horman]

On Mon, May 12, 2008 at 02:22:25PM -0700, David Miller wrote:
> From: Neil Horman <nhorman@tuxdriver.com>
> Date: Mon, 12 May 2008 10:37:36 -0400
> 
> > Dave, where is this applied?  I don't see it in either your net or net-next
> > trees?
> 
> My bad, sorry.
> 
> I didn't want to push this batch out until I did some sanity builds.
> 
> I did that just now and pushed everything out to net-next-2.6.git

cool. thanks dave!
neil