Merge branch 'dccp' of git://eden-feed.erg.abdn.ac.uk/net-next-2.6

include/linux/dccp.h

@@ -165,8 +165,10 @@ enum {
 	DCCPO_TIMESTAMP_ECHO = 42,
 	DCCPO_ELAPSED_TIME = 43,
 	DCCPO_MAX = 45,
-	DCCPO_MIN_CCID_SPECIFIC = 128,
+	DCCPO_MIN_RX_CCID_SPECIFIC = 128,	/* from sender to receiver */
-	DCCPO_MAX_CCID_SPECIFIC = 255,
+	DCCPO_MAX_RX_CCID_SPECIFIC = 191,
+	DCCPO_MIN_TX_CCID_SPECIFIC = 192,	/* from receiver to sender */
+	DCCPO_MAX_TX_CCID_SPECIFIC = 255,
 };
 /* maximum size of a single TLV-encoded DCCP option (sans type/len bytes) */
 #define DCCP_SINGLE_OPT_MAXLEN	253

net/dccp/ccid.h

@@ -62,18 +62,14 @@ struct ccid_operations {
 	void		(*ccid_hc_tx_exit)(struct sock *sk);
 	void		(*ccid_hc_rx_packet_recv)(struct sock *sk,
 						  struct sk_buff *skb);
-	int		(*ccid_hc_rx_parse_options)(struct sock *sk,
+	int		(*ccid_hc_rx_parse_options)(struct sock *sk, u8 pkt,
-						    unsigned char option,
+						    u8 opt, u8 *val, u8 len);
-						    unsigned char len, u16 idx,
-						    unsigned char* value);
 	int		(*ccid_hc_rx_insert_options)(struct sock *sk,
 						     struct sk_buff *skb);
 	void		(*ccid_hc_tx_packet_recv)(struct sock *sk,
 						  struct sk_buff *skb);
-	int		(*ccid_hc_tx_parse_options)(struct sock *sk,
+	int		(*ccid_hc_tx_parse_options)(struct sock *sk, u8 pkt,
-						    unsigned char option,
+						    u8 opt, u8 *val, u8 len);
-						    unsigned char len, u16 idx,
-						    unsigned char* value);
 	int		(*ccid_hc_tx_send_packet)(struct sock *sk,
 						  struct sk_buff *skb);
 	void		(*ccid_hc_tx_packet_sent)(struct sock *sk,
@@ -168,27 +164,31 @@ static inline void ccid_hc_tx_packet_recv(struct ccid *ccid, struct sock *sk,
 		ccid->ccid_ops->ccid_hc_tx_packet_recv(sk, skb);
 }
 
+/**
+ * ccid_hc_tx_parse_options  -  Parse CCID-specific options sent by the receiver
+ * @pkt: type of packet that @opt appears on (RFC 4340, 5.1)
+ * @opt: the CCID-specific option type (RFC 4340, 5.8 and 10.3)
+ * @val: value of @opt
+ * @len: length of @val in bytes
+ */
 static inline int ccid_hc_tx_parse_options(struct ccid *ccid, struct sock *sk,
-					   unsigned char option,
+					   u8 pkt, u8 opt, u8 *val, u8 len)
-					   unsigned char len, u16 idx,
-					   unsigned char* value)
 {
-	int rc = 0;
+	if (ccid->ccid_ops->ccid_hc_tx_parse_options == NULL)
-	if (ccid->ccid_ops->ccid_hc_tx_parse_options != NULL)
+		return 0;
-		rc = ccid->ccid_ops->ccid_hc_tx_parse_options(sk, option, len, idx,
+	return ccid->ccid_ops->ccid_hc_tx_parse_options(sk, pkt, opt, val, len);
-						    value);
-	return rc;
 }
 
+/**
+ * ccid_hc_rx_parse_options  -  Parse CCID-specific options sent by the sender
+ * Arguments are analogous to ccid_hc_tx_parse_options()
+ */
 static inline int ccid_hc_rx_parse_options(struct ccid *ccid, struct sock *sk,
-					   unsigned char option,
+					   u8 pkt, u8 opt, u8 *val, u8 len)
-					   unsigned char len, u16 idx,
-					   unsigned char* value)
 {
-	int rc = 0;
+	if (ccid->ccid_ops->ccid_hc_rx_parse_options == NULL)
-	if (ccid->ccid_ops->ccid_hc_rx_parse_options != NULL)
+		return 0;
-		rc = ccid->ccid_ops->ccid_hc_rx_parse_options(sk, option, len, idx, value);
+	return ccid->ccid_ops->ccid_hc_rx_parse_options(sk, pkt, opt, val, len);
-	return rc;
 }
 
 static inline int ccid_hc_rx_insert_options(struct ccid *ccid, struct sock *sk,

net/dccp/ccids/ccid3.c

@@ -54,7 +54,6 @@ static const char *ccid3_tx_state_name(enum ccid3_hc_tx_states state)
 	[TFRC_SSTATE_NO_SENT]  = "NO_SENT",
 	[TFRC_SSTATE_NO_FBACK] = "NO_FBACK",
 	[TFRC_SSTATE_FBACK]    = "FBACK",
-	[TFRC_SSTATE_TERM]     = "TERM",
 	};
 
 	return ccid3_state_names[state];
@@ -208,10 +207,13 @@ static void ccid3_hc_tx_no_feedback_timer(unsigned long data)
 	ccid3_pr_debug("%s(%p, state=%s) - entry\n", dccp_role(sk), sk,
 		       ccid3_tx_state_name(hc->tx_state));
 
+	/* Ignore and do not restart after leaving the established state */
+	if ((1 << sk->sk_state) & ~(DCCPF_OPEN | DCCPF_PARTOPEN))
+		goto out;
+
+	/* Reset feedback state to "no feedback received" */
 	if (hc->tx_state == TFRC_SSTATE_FBACK)
 		ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK);
-	else if (hc->tx_state != TFRC_SSTATE_NO_FBACK)
-		goto out;
 
 	/*
 	 * Determine new allowed sending rate X as per draft rfc3448bis-00, 4.4
@@ -287,8 +289,7 @@ static int ccid3_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
 	if (unlikely(skb->len == 0))
 		return -EBADMSG;
 
-	switch (hc->tx_state) {
+	if (hc->tx_state == TFRC_SSTATE_NO_SENT) {
-	case TFRC_SSTATE_NO_SENT:
 		sk_reset_timer(sk, &hc->tx_no_feedback_timer, (jiffies +
 			       usecs_to_jiffies(TFRC_INITIAL_TIMEOUT)));
 		hc->tx_last_win_count	= 0;
@@ -323,9 +324,8 @@ static int ccid3_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
 		ccid3_update_send_interval(hc);
 
 		ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK);
-		break;
+
-	case TFRC_SSTATE_NO_FBACK:
+	} else {
-	case TFRC_SSTATE_FBACK:
 		delay = ktime_us_delta(hc->tx_t_nom, now);
 		ccid3_pr_debug("delay=%ld\n", (long)delay);
 		/*
@@ -340,10 +340,6 @@ static int ccid3_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
 			return (u32)delay / USEC_PER_MSEC;
 
 		ccid3_hc_tx_update_win_count(hc, now);
-		break;
-	case TFRC_SSTATE_TERM:
-		DCCP_BUG("%s(%p) - Illegal state TERM", dccp_role(sk), sk);
-		return -EINVAL;
 	}
 
 	/* prepare to send now (add options etc.) */
@@ -369,21 +365,15 @@ static void ccid3_hc_tx_packet_sent(struct sock *sk, int more,
 static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 {
 	struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk);
-	struct ccid3_options_received *opt_recv = &hc->tx_options_received;
 	struct tfrc_tx_hist_entry *acked;
 	ktime_t now;
 	unsigned long t_nfb;
-	u32 pinv, r_sample;
+	u32 r_sample;
 
 	/* we are only interested in ACKs */
 	if (!(DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_ACK ||
 	      DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_DATAACK))
 		return;
-	/* ... and only in the established state */
-	if (hc->tx_state != TFRC_SSTATE_FBACK &&
-	    hc->tx_state != TFRC_SSTATE_NO_FBACK)
-		return;
-
 	/*
 	 * Locate the acknowledged packet in the TX history.
 	 *
@@ -403,17 +393,6 @@ static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	r_sample  = dccp_sample_rtt(sk, ktime_us_delta(now, acked->stamp));
 	hc->tx_rtt = tfrc_ewma(hc->tx_rtt, r_sample, 9);
 
-	/* Update receive rate in units of 64 * bytes/second */
-	hc->tx_x_recv = opt_recv->ccid3or_receive_rate;
-	hc->tx_x_recv <<= 6;
-
-	/* Update loss event rate (which is scaled by 1e6) */
-	pinv = opt_recv->ccid3or_loss_event_rate;
-	if (pinv == ~0U || pinv == 0)	       /* see RFC 4342, 8.5   */
-		hc->tx_p = 0;
-	else				       /* can not exceed 100% */
-		hc->tx_p = scaled_div(1, pinv);
-
 	/*
 	 * Update allowed sending rate X as per draft rfc3448bis-00, 4.2/3
 	 */
@@ -481,30 +460,36 @@ done_computing_x:
 			   jiffies + usecs_to_jiffies(t_nfb));
 }
 
-static int ccid3_hc_tx_parse_options(struct sock *sk, unsigned char option,
+static int ccid3_hc_tx_parse_options(struct sock *sk, u8 packet_type,
-				     unsigned char len, u16 idx,
+				     u8 option, u8 *optval, u8 optlen)
-				     unsigned char *value)
 {
 	struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk);
-	struct ccid3_options_received *opt_recv = &hc->tx_options_received;
 	__be32 opt_val;
 
 	switch (option) {
 	case TFRC_OPT_RECEIVE_RATE:
 	case TFRC_OPT_LOSS_EVENT_RATE:
-		if (unlikely(len != 4)) {
+		/* Must be ignored on Data packets, cf. RFC 4342 8.3 and 8.5 */
+		if (packet_type == DCCP_PKT_DATA)
+			break;
+		if (unlikely(optlen != 4)) {
 			DCCP_WARN("%s(%p), invalid len %d for %u\n",
-				  dccp_role(sk), sk, len, option);
+				  dccp_role(sk), sk, optlen, option);
 			return -EINVAL;
 		}
-		opt_val = ntohl(get_unaligned((__be32 *)value));
+		opt_val = ntohl(get_unaligned((__be32 *)optval));
 
 		if (option == TFRC_OPT_RECEIVE_RATE) {
-			opt_recv->ccid3or_receive_rate = opt_val;
+			/* Receive Rate is kept in units of 64 bytes/second */
+			hc->tx_x_recv = opt_val;
+			hc->tx_x_recv <<= 6;
+
 			ccid3_pr_debug("%s(%p), RECEIVE_RATE=%u\n",
 				       dccp_role(sk), sk, opt_val);
 		} else {
-			opt_recv->ccid3or_loss_event_rate = opt_val;
+			/* Update the fixpoint Loss Event Rate fraction */
+			hc->tx_p = tfrc_invert_loss_event_rate(opt_val);
+
 			ccid3_pr_debug("%s(%p), LOSS_EVENT_RATE=%u\n",
 				       dccp_role(sk), sk, opt_val);
 		}
@@ -527,9 +512,7 @@ static void ccid3_hc_tx_exit(struct sock *sk)
 {
 	struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk);
 
-	ccid3_hc_tx_set_state(sk, TFRC_SSTATE_TERM);
 	sk_stop_timer(sk, &hc->tx_no_feedback_timer);
-
 	tfrc_tx_hist_purge(&hc->tx_hist);
 }
 
@@ -588,7 +571,6 @@ static const char *ccid3_rx_state_name(enum ccid3_hc_rx_states state)
 	static const char *const ccid3_rx_state_names[] = {
 	[TFRC_RSTATE_NO_DATA] = "NO_DATA",
 	[TFRC_RSTATE_DATA]    = "DATA",
-	[TFRC_RSTATE_TERM]    = "TERM",
 	};
 
 	return ccid3_rx_state_names[state];
@@ -614,14 +596,9 @@ static void ccid3_hc_rx_send_feedback(struct sock *sk,
 {
 	struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk);
 	struct dccp_sock *dp = dccp_sk(sk);
-	ktime_t now;
+	ktime_t now = ktime_get_real();
 	s64 delta = 0;
 
-	if (unlikely(hc->rx_state == TFRC_RSTATE_TERM))
-		return;
-
-	now = ktime_get_real();
-
 	switch (fbtype) {
 	case CCID3_FBACK_INITIAL:
 		hc->rx_x_recv = 0;
@@ -825,8 +802,6 @@ static void ccid3_hc_rx_exit(struct sock *sk)
 {
 	struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk);
 
-	ccid3_hc_rx_set_state(sk, TFRC_RSTATE_TERM);
-
 	tfrc_rx_hist_purge(&hc->rx_hist);
 	tfrc_lh_cleanup(&hc->rx_li_hist);
 }
@@ -851,8 +826,7 @@ static int ccid3_hc_rx_getsockopt(struct sock *sk, const int optname, int len,
 			return -EINVAL;
 		rx_info.tfrcrx_x_recv = hc->rx_x_recv;
 		rx_info.tfrcrx_rtt    = hc->rx_rtt;
-		rx_info.tfrcrx_p      = hc->rx_pinv == 0 ? ~0U :
+		rx_info.tfrcrx_p      = tfrc_invert_loss_event_rate(hc->rx_pinv);
-					   scaled_div(1, hc->rx_pinv);
 		len = sizeof(rx_info);
 		val = &rx_info;
 		break;

net/dccp/ccids/ccid3.h

@@ -67,17 +67,11 @@ enum ccid3_options {
 	TFRC_OPT_RECEIVE_RATE	 = 194,
 };
 
-struct ccid3_options_received {
-	u32 ccid3or_loss_event_rate;
-	u32 ccid3or_receive_rate;
-};
-
 /* TFRC sender states */
 enum ccid3_hc_tx_states {
 	TFRC_SSTATE_NO_SENT = 1,
 	TFRC_SSTATE_NO_FBACK,
 	TFRC_SSTATE_FBACK,
-	TFRC_SSTATE_TERM,
 };
 
 /**
@@ -98,7 +92,6 @@ enum ccid3_hc_tx_states {
  * @tx_t_ld:		  Time last doubled during slow start
  * @tx_t_nom:		  Nominal send time of next packet
  * @tx_hist:		  Packet history
- * @tx_options_received:  Parsed set of retrieved options
  */
 struct ccid3_hc_tx_sock {
 	u64				tx_x;
@@ -116,7 +109,6 @@ struct ccid3_hc_tx_sock {
 	ktime_t				tx_t_ld;
 	ktime_t				tx_t_nom;
 	struct tfrc_tx_hist_entry	*tx_hist;
-	struct ccid3_options_received	tx_options_received;
 };
 
 static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk)
@@ -130,7 +122,6 @@ static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk)
 enum ccid3_hc_rx_states {
 	TFRC_RSTATE_NO_DATA = 1,
 	TFRC_RSTATE_DATA,
-	TFRC_RSTATE_TERM    = 127,
 };
 
 /**

net/dccp/ccids/lib/tfrc.h

@@ -57,6 +57,7 @@ static inline u32 tfrc_ewma(const u32 avg, const u32 newval, const u8 weight)
 
 extern u32  tfrc_calc_x(u16 s, u32 R, u32 p);
 extern u32  tfrc_calc_x_reverse_lookup(u32 fvalue);
+extern u32  tfrc_invert_loss_event_rate(u32 loss_event_rate);
 
 extern int  tfrc_tx_packet_history_init(void);
 extern void tfrc_tx_packet_history_exit(void);

net/dccp/ccids/lib/tfrc_equation.c

@@ -687,3 +687,17 @@ u32 tfrc_calc_x_reverse_lookup(u32 fvalue)
 	index = tfrc_binsearch(fvalue, 0);
 	return (index + 1) * 1000000 / TFRC_CALC_X_ARRSIZE;
 }
+
+/**
+ * tfrc_invert_loss_event_rate  -  Compute p so that 10^6 corresponds to 100%
+ * When @loss_event_rate is large, there is a chance that p is truncated to 0.
+ * To avoid re-entering slow-start in that case, we set p = TFRC_SMALLEST_P > 0.
+ */
+u32 tfrc_invert_loss_event_rate(u32 loss_event_rate)
+{
+	if (loss_event_rate == UINT_MAX)		/* see RFC 4342, 8.5 */
+		return 0;
+	if (unlikely(loss_event_rate == 0))		/* map 1/0 into 100% */
+		return 1000000;
+	return max_t(u32, scaled_div(1, loss_event_rate), TFRC_SMALLEST_P);
+}

net/dccp/options.c

@@ -96,18 +96,11 @@ int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
 		}
 
 		/*
-		 * CCID-Specific Options (from RFC 4340, sec. 10.3):
-		 *
-		 * Option numbers 128 through 191 are for options sent from the
-		 * HC-Sender to the HC-Receiver; option numbers 192 through 255
-		 * are for options sent from the HC-Receiver to	the HC-Sender.
-		 *
 		 * CCID-specific options are ignored during connection setup, as
 		 * negotiation may still be in progress (see RFC 4340, 10.3).
 		 * The same applies to Ack Vectors, as these depend on the CCID.
-		 *
 		 */
-		if (dreq != NULL && (opt >= 128 ||
+		if (dreq != NULL && (opt >= DCCPO_MIN_RX_CCID_SPECIFIC ||
 		    opt == DCCPO_ACK_VECTOR_0 || opt == DCCPO_ACK_VECTOR_1))
 			goto ignore_option;
 
@@ -226,23 +219,15 @@ int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
 			dccp_pr_debug("%s rx opt: ELAPSED_TIME=%d\n",
 				      dccp_role(sk), elapsed_time);
 			break;
-		case 128 ... 191: {
+		case DCCPO_MIN_RX_CCID_SPECIFIC ... DCCPO_MAX_RX_CCID_SPECIFIC:
-			const u16 idx = value - options;
-
 			if (ccid_hc_rx_parse_options(dp->dccps_hc_rx_ccid, sk,
-						     opt, len, idx,
+						     pkt_type, opt, value, len))
-						     value) != 0)
 				goto out_invalid_option;
-		}
 			break;
-		case 192 ... 255: {
+		case DCCPO_MIN_TX_CCID_SPECIFIC ... DCCPO_MAX_TX_CCID_SPECIFIC:
-			const u16 idx = value - options;
-
 			if (ccid_hc_tx_parse_options(dp->dccps_hc_tx_ccid, sk,
-						     opt, len, idx,
+						     pkt_type, opt, value, len))
-						     value) != 0)
 				goto out_invalid_option;
-		}
 			break;
 		default:
 			DCCP_CRIT("DCCP(%p): option %d(len=%d) not "