Merge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [SCTP]: Allow linger to abort 1-N style sockets.
  [SCTP]: Validate the parameter length in HB-ACK chunk.
  [SCTP]: Respect the real chunk length when walking parameters.
  [SCTP]: A better solution to fix the race between sctp_peeloff() and
  [SCTP]: Set sk_err so that poll wakes up after a non-blocking connect failure.

include/net/sctp/command.h

@@ -99,6 +99,7 @@ typedef enum {
 	SCTP_CMD_DEL_NON_PRIMARY, /* Removes non-primary peer transports. */
 	SCTP_CMD_T3_RTX_TIMERS_STOP, /* Stops T3-rtx pending timers */
 	SCTP_CMD_FORCE_PRIM_RETRAN,  /* Forces retrans. over primary path. */
+	SCTP_CMD_SET_SK_ERR,	 /* Set sk_err */
 	SCTP_CMD_LAST
 } sctp_verb_t;
 

include/net/sctp/sctp.h

@@ -461,12 +461,12 @@ static inline int sctp_frag_point(const struct sctp_sock *sp, int pmtu)
  * there is room for a param header too.
  */
 #define sctp_walk_params(pos, chunk, member)\
-_sctp_walk_params((pos), (chunk), WORD_ROUND(ntohs((chunk)->chunk_hdr.length)), member)
+_sctp_walk_params((pos), (chunk), ntohs((chunk)->chunk_hdr.length), member)
 
 #define _sctp_walk_params(pos, chunk, end, member)\
 for (pos.v = chunk->member;\
      pos.v <= (void *)chunk + end - sizeof(sctp_paramhdr_t) &&\
-     pos.v <= (void *)chunk + end - WORD_ROUND(ntohs(pos.p->length)) &&\
+     pos.v <= (void *)chunk + end - ntohs(pos.p->length) &&\
      ntohs(pos.p->length) >= sizeof(sctp_paramhdr_t);\
      pos.v += WORD_ROUND(ntohs(pos.p->length)))
 
@@ -477,7 +477,7 @@ _sctp_walk_errors((err), (chunk_hdr), ntohs((chunk_hdr)->length))
 for (err = (sctp_errhdr_t *)((void *)chunk_hdr + \
 	    sizeof(sctp_chunkhdr_t));\
      (void *)err <= (void *)chunk_hdr + end - sizeof(sctp_errhdr_t) &&\
-     (void *)err <= (void *)chunk_hdr + end - WORD_ROUND(ntohs(err->length)) &&\
+     (void *)err <= (void *)chunk_hdr + end - ntohs(err->length) &&\
      ntohs(err->length) >= sizeof(sctp_errhdr_t); \
      err = (sctp_errhdr_t *)((void *)err + WORD_ROUND(ntohs(err->length))))
 

net/sctp/input.c

@@ -73,6 +73,8 @@ static struct sctp_association *__sctp_lookup_association(
 					const union sctp_addr *peer,
 					struct sctp_transport **pt);
 
+static void sctp_add_backlog(struct sock *sk, struct sk_buff *skb);
+
 
 /* Calculate the SCTP checksum of an SCTP packet.  */
 static inline int sctp_rcv_checksum(struct sk_buff *skb)
@@ -186,7 +188,6 @@ int sctp_rcv(struct sk_buff *skb)
 	 */
 	if (sk->sk_bound_dev_if && (sk->sk_bound_dev_if != af->skb_iif(skb)))
 	{
-		sock_put(sk);
 		if (asoc) {
 			sctp_association_put(asoc);
 			asoc = NULL;
@@ -197,7 +198,6 @@ int sctp_rcv(struct sk_buff *skb)
 		sk = sctp_get_ctl_sock();
 		ep = sctp_sk(sk)->ep;
 		sctp_endpoint_hold(ep);
-		sock_hold(sk);
 		rcvr = &ep->base;
 	}
 
@@ -253,25 +253,18 @@ int sctp_rcv(struct sk_buff *skb)
 	 */
 	sctp_bh_lock_sock(sk);
 
-	/* It is possible that the association could have moved to a different
-	 * socket if it is peeled off. If so, update the sk.
-	 */ 
-	if (sk != rcvr->sk) {
-		sctp_bh_lock_sock(rcvr->sk);
-		sctp_bh_unlock_sock(sk);
-		sk = rcvr->sk;
-	}
-
 	if (sock_owned_by_user(sk))
-		sk_add_backlog(sk, skb);
+		sctp_add_backlog(sk, skb);
 	else
-		sctp_backlog_rcv(sk, skb);
+		sctp_inq_push(&chunk->rcvr->inqueue, chunk);
 
-	/* Release the sock and the sock ref we took in the lookup calls.
-	 * The asoc/ep ref will be released in sctp_backlog_rcv.
-	 */
 	sctp_bh_unlock_sock(sk);
-	sock_put(sk);
+
+	/* Release the asoc/ep ref we took in the lookup calls. */
+	if (asoc)
+		sctp_association_put(asoc);
+	else
+		sctp_endpoint_put(ep);
 
 	return 0;
 
@@ -280,8 +273,7 @@ int sctp_rcv(struct sk_buff *skb)
 	return 0;
 
 discard_release:
-	/* Release any structures we may be holding. */
-	sock_put(sk);
+	/* Release the asoc/ep ref we took in the lookup calls. */
 	if (asoc)
 		sctp_association_put(asoc);
 	else
@@ -290,56 +282,87 @@ int sctp_rcv(struct sk_buff *skb)
 	goto discard_it;
 }
 
-/* Handle second half of inbound skb processing.  If the sock was busy,
- * we may have need to delay processing until later when the sock is
- * released (on the backlog).   If not busy, we call this routine
- * directly from the bottom half.
+/* Process the backlog queue of the socket.  Every skb on
+ * the backlog holds a ref on an association or endpoint.
+ * We hold this ref throughout the state machine to make
+ * sure that the structure we need is still around.
  */
 int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb)
 {
 	struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk;
- 	struct sctp_inq *inqueue = NULL;
+ 	struct sctp_inq *inqueue = &chunk->rcvr->inqueue;
  	struct sctp_ep_common *rcvr = NULL;
+	int backloged = 0;
 
  	rcvr = chunk->rcvr;
 
-	BUG_TRAP(rcvr->sk == sk);
-
- 	if (rcvr->dead) {
- 		sctp_chunk_free(chunk);
- 	} else {
- 		inqueue = &chunk->rcvr->inqueue;
- 		sctp_inq_push(inqueue, chunk);
- 	}
-
-	/* Release the asoc/ep ref we took in the lookup calls in sctp_rcv. */ 
- 	if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type)
- 		sctp_association_put(sctp_assoc(rcvr));
- 	else
- 		sctp_endpoint_put(sctp_ep(rcvr));
-  
+	/* If the rcvr is dead then the association or endpoint
+	 * has been deleted and we can safely drop the chunk
+	 * and refs that we are holding.
+	 */
+	if (rcvr->dead) {
+		sctp_chunk_free(chunk);
+		goto done;
+	}
+
+	if (unlikely(rcvr->sk != sk)) {
+		/* In this case, the association moved from one socket to
+		 * another.  We are currently sitting on the backlog of the
+		 * old socket, so we need to move.
+		 * However, since we are here in the process context we
+		 * need to take make sure that the user doesn't own
+		 * the new socket when we process the packet.
+		 * If the new socket is user-owned, queue the chunk to the
+		 * backlog of the new socket without dropping any refs.
+		 * Otherwise, we can safely push the chunk on the inqueue.
+		 */
+
+		sk = rcvr->sk;
+		sctp_bh_lock_sock(sk);
+
+		if (sock_owned_by_user(sk)) {
+			sk_add_backlog(sk, skb);
+			backloged = 1;
+		} else
+			sctp_inq_push(inqueue, chunk);
+
+		sctp_bh_unlock_sock(sk);
+
+		/* If the chunk was backloged again, don't drop refs */
+		if (backloged)
+			return 0;
+	} else {
+		sctp_inq_push(inqueue, chunk);
+	}
+
+done:
+	/* Release the refs we took in sctp_add_backlog */
+	if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type)
+		sctp_association_put(sctp_assoc(rcvr));
+	else if (SCTP_EP_TYPE_SOCKET == rcvr->type)
+		sctp_endpoint_put(sctp_ep(rcvr));
+	else
+		BUG();
+
         return 0;
 }
 
-void sctp_backlog_migrate(struct sctp_association *assoc, 
-			  struct sock *oldsk, struct sock *newsk)
+static void sctp_add_backlog(struct sock *sk, struct sk_buff *skb)
 {
-	struct sk_buff *skb;
-	struct sctp_chunk *chunk;
+	struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk;
+	struct sctp_ep_common *rcvr = chunk->rcvr;
 
-	skb = oldsk->sk_backlog.head;
-	oldsk->sk_backlog.head = oldsk->sk_backlog.tail = NULL;
-	while (skb != NULL) {
-		struct sk_buff *next = skb->next;
-
-		chunk = SCTP_INPUT_CB(skb)->chunk;
-		skb->next = NULL;
-		if (&assoc->base == chunk->rcvr)
-			sk_add_backlog(newsk, skb);
-		else
-			sk_add_backlog(oldsk, skb);
-		skb = next;
-	}
+	/* Hold the assoc/ep while hanging on the backlog queue.
+	 * This way, we know structures we need will not disappear from us
+	 */
+	if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type)
+		sctp_association_hold(sctp_assoc(rcvr));
+	else if (SCTP_EP_TYPE_SOCKET == rcvr->type)
+		sctp_endpoint_hold(sctp_ep(rcvr));
+	else
+		BUG();
+
+	sk_add_backlog(sk, skb);
 }
 
 /* Handle icmp frag needed error. */
@@ -412,7 +435,7 @@ struct sock *sctp_err_lookup(int family, struct sk_buff *skb,
 	union sctp_addr daddr;
 	struct sctp_af *af;
 	struct sock *sk = NULL;
-	struct sctp_association *asoc = NULL;
+	struct sctp_association *asoc;
 	struct sctp_transport *transport = NULL;
 
 	*app = NULL; *tpp = NULL;
@@ -453,7 +476,6 @@ struct sock *sctp_err_lookup(int family, struct sk_buff *skb,
 	return sk;
 
 out:
-	sock_put(sk);
 	if (asoc)
 		sctp_association_put(asoc);
 	return NULL;
@@ -463,7 +485,6 @@ struct sock *sctp_err_lookup(int family, struct sk_buff *skb,
 void sctp_err_finish(struct sock *sk, struct sctp_association *asoc)
 {
 	sctp_bh_unlock_sock(sk);
-	sock_put(sk);
 	if (asoc)
 		sctp_association_put(asoc);
 }
@@ -490,7 +511,7 @@ void sctp_v4_err(struct sk_buff *skb, __u32 info)
 	int type = skb->h.icmph->type;
 	int code = skb->h.icmph->code;
 	struct sock *sk;
-	struct sctp_association *asoc;
+	struct sctp_association *asoc = NULL;
 	struct sctp_transport *transport;
 	struct inet_sock *inet;
 	char *saveip, *savesctp;
@@ -716,7 +737,6 @@ static struct sctp_endpoint *__sctp_rcv_lookup_endpoint(const union sctp_addr *l
 
 hit:
 	sctp_endpoint_hold(ep);
-	sock_hold(epb->sk);
 	read_unlock(&head->lock);
 	return ep;
 }
@@ -818,7 +838,6 @@ static struct sctp_association *__sctp_lookup_association(
 hit:
 	*pt = transport;
 	sctp_association_hold(asoc);
-	sock_hold(epb->sk);
 	read_unlock(&head->lock);
 	return asoc;
 }
@@ -846,7 +865,6 @@ int sctp_has_association(const union sctp_addr *laddr,
 	struct sctp_transport *transport;
 
 	if ((asoc = sctp_lookup_association(laddr, paddr, &transport))) {
-		sock_put(asoc->base.sk);
 		sctp_association_put(asoc);
 		return 1;
 	}

net/sctp/sm_sideeffect.c

@@ -498,10 +498,6 @@ static void sctp_cmd_assoc_failed(sctp_cmd_seq_t *commands,
 	sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
 			SCTP_STATE(SCTP_STATE_CLOSED));
 
-	/* Set sk_err to ECONNRESET on a 1-1 style socket. */
-	if (!sctp_style(asoc->base.sk, UDP))
-		asoc->base.sk->sk_err = ECONNRESET; 
-
 	/* SEND_FAILED sent later when cleaning up the association. */
 	asoc->outqueue.error = error;
 	sctp_add_cmd_sf(commands, SCTP_CMD_DELETE_TCB, SCTP_NULL());
@@ -838,6 +834,15 @@ static void sctp_cmd_del_non_primary(struct sctp_association *asoc)
 	return;
 }
 
+/* Helper function to set sk_err on a 1-1 style socket. */
+static void sctp_cmd_set_sk_err(struct sctp_association *asoc, int error)
+{
+	struct sock *sk = asoc->base.sk;
+
+	if (!sctp_style(sk, UDP))
+		sk->sk_err = error;
+}
+
 /* These three macros allow us to pull the debugging code out of the
  * main flow of sctp_do_sm() to keep attention focused on the real
  * functionality there.
@@ -1458,6 +1463,9 @@ static int sctp_cmd_interpreter(sctp_event_t event_type,
 			local_cork = 0;
 			asoc->peer.retran_path = t;
 			break;
+		case SCTP_CMD_SET_SK_ERR:
+			sctp_cmd_set_sk_err(asoc, cmd->obj.error);
+			break;
 		default:
 			printk(KERN_WARNING "Impossible command: %u, %p\n",
 			       cmd->verb, cmd->obj.ptr);

net/sctp/socket.c

@@ -1057,6 +1057,7 @@ static int __sctp_connect(struct sock* sk,
 	inet_sk(sk)->dport = htons(asoc->peer.port);
 	af = sctp_get_af_specific(to.sa.sa_family);
 	af->to_sk_daddr(&to, sk);
+	sk->sk_err = 0;
 
 	timeo = sock_sndtimeo(sk, sk->sk_socket->file->f_flags & O_NONBLOCK);
 	err = sctp_wait_for_connect(asoc, &timeo);
@@ -1228,7 +1229,7 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout)
 
 	ep = sctp_sk(sk)->ep;
 
-	/* Walk all associations on a socket, not on an endpoint.  */
+	/* Walk all associations on an endpoint.  */
 	list_for_each_safe(pos, temp, &ep->asocs) {
 		asoc = list_entry(pos, struct sctp_association, asocs);
 
@@ -1241,13 +1242,13 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout)
 			if (sctp_state(asoc, CLOSED)) {
 				sctp_unhash_established(asoc);
 				sctp_association_free(asoc);
+				continue;
+			}
+		}
 
-			} else if (sock_flag(sk, SOCK_LINGER) &&
-				   !sk->sk_lingertime)
-				sctp_primitive_ABORT(asoc, NULL);
-			else
-				sctp_primitive_SHUTDOWN(asoc, NULL);
-		} else
+		if (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime)
+			sctp_primitive_ABORT(asoc, NULL);
+		else
 			sctp_primitive_SHUTDOWN(asoc, NULL);
 	}
 
@@ -5317,6 +5318,7 @@ static int sctp_wait_for_sndbuf(struct sctp_association *asoc, long *timeo_p,
 		 */
 		sctp_release_sock(sk);
 		current_timeo = schedule_timeout(current_timeo);
+		BUG_ON(sk != asoc->base.sk);
 		sctp_lock_sock(sk);
 
 		*timeo_p = current_timeo;
@@ -5604,12 +5606,14 @@ static void sctp_sock_migrate(struct sock *oldsk, struct sock *newsk,
 	 */
 	newsp->type = type;
 
-	spin_lock_bh(&oldsk->sk_lock.slock);
-	/* Migrate the backlog from oldsk to newsk. */
-	sctp_backlog_migrate(assoc, oldsk, newsk);
-	/* Migrate the association to the new socket. */
+	/* Mark the new socket "in-use" by the user so that any packets
+	 * that may arrive on the association after we've moved it are
+	 * queued to the backlog.  This prevents a potential race between
+	 * backlog processing on the old socket and new-packet processing
+	 * on the new socket.
+	 */
+	sctp_lock_sock(newsk);
 	sctp_assoc_migrate(assoc, newsk);
-	spin_unlock_bh(&oldsk->sk_lock.slock);
 
 	/* If the association on the newsk is already closed before accept()
 	 * is called, set RCV_SHUTDOWN flag.
@@ -5618,6 +5622,7 @@ static void sctp_sock_migrate(struct sock *oldsk, struct sock *newsk,
 		newsk->sk_shutdown |= RCV_SHUTDOWN;
 
 	newsk->sk_state = SCTP_SS_ESTABLISHED;
+	sctp_release_sock(newsk);
 }
 
 /* This proto struct describes the ULP interface for SCTP.  */