1 files changed, 99 insertions, 129 deletions

diff --git a/net/dccp/ccids/ccid2.c b/net/dccp/ccids/ccid2.c
index d694656b880..b5b52ebb269 100644
--- a/net/dccp/ccids/ccid2.c
+++ b/net/dccp/ccids/ccid2.c
@@ -24,9 +24,6 @@
 
 /*
  * This implementation should follow RFC 4341
- *
- * BUGS:
- * - sequence number wrapping
  */
 
 #include "../ccid.h"
@@ -129,50 +126,35 @@ static int ccid2_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
 {
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
 
-	ccid2_pr_debug("pipe=%d cwnd=%d\n", hctx->ccid2hctx_pipe,
-		       hctx->ccid2hctx_cwnd);
-
-	if (hctx->ccid2hctx_pipe < hctx->ccid2hctx_cwnd) {
-		/* OK we can send... make sure previous packet was sent off */
-		if (!hctx->ccid2hctx_sendwait) {
-			hctx->ccid2hctx_sendwait = 1;
-			return 0;
-		}
-	}
+	if (hctx->ccid2hctx_pipe < hctx->ccid2hctx_cwnd)
+		return 0;
 
 	return 1; /* XXX CCID should dequeue when ready instead of polling */
 }
 
-static void ccid2_change_l_ack_ratio(struct sock *sk, int val)
+static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
+	u32 max_ratio = DIV_ROUND_UP(ccid2_hc_tx_sk(sk)->ccid2hctx_cwnd, 2);
+
 	/*
-	 * XXX I don't really agree with val != 2.  If cwnd is 1, ack ratio
-	 * should be 1... it shouldn't be allowed to become 2.
-	 * -sorbo.
+	 * Ensure that Ack Ratio does not exceed ceil(cwnd/2), which is (2) from
+	 * RFC 4341, 6.1.2. We ignore the statement that Ack Ratio 2 is always
+	 * acceptable since this causes starvation/deadlock whenever cwnd < 2.
+	 * The same problem arises when Ack Ratio is 0 (ie. Ack Ratio disabled).
 	 */
-	if (val != 2) {
-		const struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
-		int max = hctx->ccid2hctx_cwnd / 2;
-
-		/* round up */
-		if (hctx->ccid2hctx_cwnd & 1)
-			max++;
-
-		if (val > max)
-			val = max;
+	if (val == 0 || val > max_ratio) {
+		DCCP_WARN("Limiting Ack Ratio (%u) to %u\n", val, max_ratio);
+		val = max_ratio;
 	}
+	if (val > 0xFFFF)		/* RFC 4340, 11.3 */
+		val = 0xFFFF;
 
-	ccid2_pr_debug("changing local ack ratio to %d\n", val);
-	WARN_ON(val <= 0);
-	dp->dccps_l_ack_ratio = val;
-}
+	if (val == dp->dccps_l_ack_ratio)
+		return;
 
-static void ccid2_change_cwnd(struct ccid2_hc_tx_sock *hctx, u32 val)
-{
-	/* XXX do we need to change ack ratio? */
-	hctx->ccid2hctx_cwnd = val? : 1;
-	ccid2_pr_debug("changed cwnd to %u\n", hctx->ccid2hctx_cwnd);
+	ccid2_pr_debug("changing local ack ratio to %u\n", val);
+	dp->dccps_l_ack_ratio = val;
 }
 
 static void ccid2_change_srtt(struct ccid2_hc_tx_sock *hctx, long val)
@@ -181,11 +163,6 @@ static void ccid2_change_srtt(struct ccid2_hc_tx_sock *hctx, long val)
 	hctx->ccid2hctx_srtt = val;
 }
 
-static void ccid2_change_pipe(struct ccid2_hc_tx_sock *hctx, long val)
-{
-	hctx->ccid2hctx_pipe = val;
-}
-
 static void ccid2_start_rto_timer(struct sock *sk);
 
 static void ccid2_hc_tx_rto_expire(unsigned long data)
@@ -215,21 +192,17 @@ static void ccid2_hc_tx_rto_expire(unsigned long data)
 	ccid2_start_rto_timer(sk);
 
 	/* adjust pipe, cwnd etc */
-	ccid2_change_pipe(hctx, 0);
-	hctx->ccid2hctx_ssthresh = hctx->ccid2hctx_cwnd >> 1;
+	hctx->ccid2hctx_ssthresh = hctx->ccid2hctx_cwnd / 2;
 	if (hctx->ccid2hctx_ssthresh < 2)
 		hctx->ccid2hctx_ssthresh = 2;
-	ccid2_change_cwnd(hctx, 1);
+	hctx->ccid2hctx_cwnd	 = 1;
+	hctx->ccid2hctx_pipe	 = 0;
 
 	/* clear state about stuff we sent */
-	hctx->ccid2hctx_seqt	= hctx->ccid2hctx_seqh;
-	hctx->ccid2hctx_ssacks	= 0;
-	hctx->ccid2hctx_acks	= 0;
-	hctx->ccid2hctx_sent	= 0;
+	hctx->ccid2hctx_seqt = hctx->ccid2hctx_seqh;
+	hctx->ccid2hctx_packets_acked = 0;
 
 	/* clear ack ratio state. */
-	hctx->ccid2hctx_arsent	 = 0;
-	hctx->ccid2hctx_ackloss  = 0;
 	hctx->ccid2hctx_rpseq	 = 0;
 	hctx->ccid2hctx_rpdupack = -1;
 	ccid2_change_l_ack_ratio(sk, 1);
@@ -255,23 +228,10 @@ static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
 	struct dccp_sock *dp = dccp_sk(sk);
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
 	struct ccid2_seq *next;
-	u64 seq;
-
-	ccid2_hc_tx_check_sanity(hctx);
 
-	BUG_ON(!hctx->ccid2hctx_sendwait);
-	hctx->ccid2hctx_sendwait = 0;
-	ccid2_change_pipe(hctx, hctx->ccid2hctx_pipe + 1);
-	BUG_ON(hctx->ccid2hctx_pipe < 0);
+	hctx->ccid2hctx_pipe++;
 
-	/* There is an issue.  What if another packet is sent between
-	 * packet_send() and packet_sent().  Then the sequence number would be
-	 * wrong.
-	 * -sorbo.
-	 */
-	seq = dp->dccps_gss;
-
-	hctx->ccid2hctx_seqh->ccid2s_seq   = seq;
+	hctx->ccid2hctx_seqh->ccid2s_seq   = dp->dccps_gss;
 	hctx->ccid2hctx_seqh->ccid2s_acked = 0;
 	hctx->ccid2hctx_seqh->ccid2s_sent  = jiffies;
 
@@ -291,8 +251,26 @@ static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
 	ccid2_pr_debug("cwnd=%d pipe=%d\n", hctx->ccid2hctx_cwnd,
 		       hctx->ccid2hctx_pipe);
 
-	hctx->ccid2hctx_sent++;
-
+	/*
+	 * FIXME: The code below is broken and the variables have been removed
+	 * from the socket struct. The `ackloss' variable was always set to 0,
+	 * and with arsent there are several problems:
+	 *  (i) it doesn't just count the number of Acks, but all sent packets;
+	 *  (ii) it is expressed in # of packets, not # of windows, so the
+	 *  comparison below uses the wrong formula: Appendix A of RFC 4341
+	 *  comes up with the number K = cwnd / (R^2 - R) of consecutive windows
+	 *  of data with no lost or marked Ack packets. If arsent were the # of
+	 *  consecutive Acks received without loss, then Ack Ratio needs to be
+	 *  decreased by 1 when
+	 *	      arsent >=  K * cwnd / R  =  cwnd^2 / (R^3 - R^2)
+	 *  where cwnd / R is the number of Acks received per window of data
+	 *  (cf. RFC 4341, App. A). The problems are that
+	 *  - arsent counts other packets as well;
+	 *  - the comparison uses a formula different from RFC 4341;
+	 *  - computing a cubic/quadratic equation each time is too complicated.
+	 *  Hence a different algorithm is needed.
+	 */
+#if 0
 	/* Ack Ratio.  Need to maintain a concept of how many windows we sent */
 	hctx->ccid2hctx_arsent++;
 	/* We had an ack loss in this window... */
@@ -320,14 +298,13 @@ static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
 			hctx->ccid2hctx_arsent = 0; /* or maybe set it to cwnd*/
 		}
 	}
+#endif
 
 	/* setup RTO timer */
 	if (!timer_pending(&hctx->ccid2hctx_rtotimer))
 		ccid2_start_rto_timer(sk);
 
 #ifdef CONFIG_IP_DCCP_CCID2_DEBUG
-	ccid2_pr_debug("pipe=%d\n", hctx->ccid2hctx_pipe);
-	ccid2_pr_debug("Sent: seq=%llu\n", (unsigned long long)seq);
 	do {
 		struct ccid2_seq *seqp = hctx->ccid2hctx_seqt;
 
@@ -419,31 +396,15 @@ static inline void ccid2_new_ack(struct sock *sk,
 {
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
 
-	/* slow start */
 	if (hctx->ccid2hctx_cwnd < hctx->ccid2hctx_ssthresh) {
-		hctx->ccid2hctx_acks = 0;
-
-		/* We can increase cwnd at most maxincr [ack_ratio/2] */
-		if (*maxincr) {
-			/* increase every 2 acks */
-			hctx->ccid2hctx_ssacks++;
-			if (hctx->ccid2hctx_ssacks == 2) {
-				ccid2_change_cwnd(hctx, hctx->ccid2hctx_cwnd+1);
-				hctx->ccid2hctx_ssacks = 0;
-				*maxincr = *maxincr - 1;
-			}
-		} else {
-			/* increased cwnd enough for this single ack */
-			hctx->ccid2hctx_ssacks = 0;
-		}
-	} else {
-		hctx->ccid2hctx_ssacks = 0;
-		hctx->ccid2hctx_acks++;
-
-		if (hctx->ccid2hctx_acks >= hctx->ccid2hctx_cwnd) {
-			ccid2_change_cwnd(hctx, hctx->ccid2hctx_cwnd + 1);
-			hctx->ccid2hctx_acks = 0;
+		if (*maxincr > 0 && ++hctx->ccid2hctx_packets_acked == 2) {
+			hctx->ccid2hctx_cwnd += 1;
+			*maxincr	     -= 1;
+			hctx->ccid2hctx_packets_acked = 0;
 		}
+	} else if (++hctx->ccid2hctx_packets_acked >= hctx->ccid2hctx_cwnd) {
+			hctx->ccid2hctx_cwnd += 1;
+			hctx->ccid2hctx_packets_acked = 0;
 	}
 
 	/* update RTO */
@@ -502,7 +463,6 @@ static inline void ccid2_new_ack(struct sock *sk,
 		ccid2_pr_debug("srtt: %ld rttvar: %ld rto: %ld (HZ=%d) R=%lu\n",
 			       hctx->ccid2hctx_srtt, hctx->ccid2hctx_rttvar,
 			       hctx->ccid2hctx_rto, HZ, r);
-		hctx->ccid2hctx_sent = 0;
 	}
 
 	/* we got a new ack, so re-start RTO timer */
@@ -514,16 +474,19 @@ static void ccid2_hc_tx_dec_pipe(struct sock *sk)
 {
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
 
-	ccid2_change_pipe(hctx, hctx->ccid2hctx_pipe-1);
-	BUG_ON(hctx->ccid2hctx_pipe < 0);
+	if (hctx->ccid2hctx_pipe == 0)
+		DCCP_BUG("pipe == 0");
+	else
+		hctx->ccid2hctx_pipe--;
 
 	if (hctx->ccid2hctx_pipe == 0)
 		ccid2_hc_tx_kill_rto_timer(sk);
 }
 
-static void ccid2_congestion_event(struct ccid2_hc_tx_sock *hctx,
-				   struct ccid2_seq *seqp)
+static void ccid2_congestion_event(struct sock *sk, struct ccid2_seq *seqp)
 {
+	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
+
 	if (time_before(seqp->ccid2s_sent, hctx->ccid2hctx_last_cong)) {
 		ccid2_pr_debug("Multiple losses in an RTT---treating as one\n");
 		return;
@@ -531,10 +494,12 @@ static void ccid2_congestion_event(struct ccid2_hc_tx_sock *hctx,
 
 	hctx->ccid2hctx_last_cong = jiffies;
 
-	ccid2_change_cwnd(hctx, hctx->ccid2hctx_cwnd >> 1);
-	hctx->ccid2hctx_ssthresh = hctx->ccid2hctx_cwnd;
-	if (hctx->ccid2hctx_ssthresh < 2)
-		hctx->ccid2hctx_ssthresh = 2;
+	hctx->ccid2hctx_cwnd     = hctx->ccid2hctx_cwnd / 2 ? : 1U;
+	hctx->ccid2hctx_ssthresh = max(hctx->ccid2hctx_cwnd, 2U);
+
+	/* Avoid spurious timeouts resulting from Ack Ratio > cwnd */
+	if (dccp_sk(sk)->dccps_l_ack_ratio > hctx->ccid2hctx_cwnd)
+		ccid2_change_l_ack_ratio(sk, hctx->ccid2hctx_cwnd);
 }
 
 static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
@@ -570,12 +535,11 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 			hctx->ccid2hctx_rpdupack++;
 
 			/* check if we got enough dupacks */
-			if (hctx->ccid2hctx_rpdupack >=
-			    hctx->ccid2hctx_numdupack) {
+			if (hctx->ccid2hctx_rpdupack >= NUMDUPACK) {
 				hctx->ccid2hctx_rpdupack = -1; /* XXX lame */
 				hctx->ccid2hctx_rpseq = 0;
 
-				ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio << 1);
+				ccid2_change_l_ack_ratio(sk, 2 * dp->dccps_l_ack_ratio);
 			}
 		}
 	}
@@ -606,12 +570,13 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 		}
 	}
 
-	/* If in slow-start, cwnd can increase at most Ack Ratio / 2 packets for
-	 * this single ack.  I round up.
-	 * -sorbo.
+	/*
+	 * In slow-start, cwnd can increase up to a maximum of Ack Ratio/2
+	 * packets per acknowledgement. Rounding up avoids that cwnd is not
+	 * advanced when Ack Ratio is 1 and gives a slight edge otherwise.
 	 */
-	maxincr = dp->dccps_l_ack_ratio >> 1;
-	maxincr++;
+	if (hctx->ccid2hctx_cwnd < hctx->ccid2hctx_ssthresh)
+		maxincr = DIV_ROUND_UP(dp->dccps_l_ack_ratio, 2);
 
 	/* go through all ack vectors */
 	while ((offset = ccid2_ackvector(sk, skb, offset,
@@ -619,9 +584,8 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 		/* go through this ack vector */
 		while (veclen--) {
 			const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
-			u64 ackno_end_rl;
+			u64 ackno_end_rl = SUB48(ackno, rl);
 
-			dccp_set_seqno(&ackno_end_rl, ackno - rl);
 			ccid2_pr_debug("ackvec start:%llu end:%llu\n",
 				       (unsigned long long)ackno,
 				       (unsigned long long)ackno_end_rl);
@@ -651,7 +615,7 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 				    !seqp->ccid2s_acked) {
 					if (state ==
 					    DCCP_ACKVEC_STATE_ECN_MARKED) {
-						ccid2_congestion_event(hctx,
+						ccid2_congestion_event(sk,
 								       seqp);
 					} else
 						ccid2_new_ack(sk, seqp,
@@ -666,13 +630,12 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 					done = 1;
 					break;
 				}
-				seqp = seqp->ccid2s_next;
+				seqp = seqp->ccid2s_prev;
 			}
 			if (done)
 				break;
 
-
-			dccp_set_seqno(&ackno, ackno_end_rl - 1);
+			ackno = SUB48(ackno_end_rl, 1);
 			vector++;
 		}
 		if (done)
@@ -694,7 +657,7 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	while (1) {
 		if (seqp->ccid2s_acked) {
 			done++;
-			if (done == hctx->ccid2hctx_numdupack)
+			if (done == NUMDUPACK)
 				break;
 		}
 		if (seqp == hctx->ccid2hctx_seqt)
@@ -705,7 +668,7 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	/* If there are at least 3 acknowledgements, anything unacknowledged
 	 * below the last sequence number is considered lost
 	 */
-	if (done == hctx->ccid2hctx_numdupack) {
+	if (done == NUMDUPACK) {
 		struct ccid2_seq *last_acked = seqp;
 
 		/* check for lost packets */
@@ -717,7 +680,7 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 				 * order to detect multiple congestion events in
 				 * one ack vector.
 				 */
-				ccid2_congestion_event(hctx, seqp);
+				ccid2_congestion_event(sk, seqp);
 				ccid2_hc_tx_dec_pipe(sk);
 			}
 			if (seqp == hctx->ccid2hctx_seqt)
@@ -742,14 +705,23 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
 {
 	struct ccid2_hc_tx_sock *hctx = ccid_priv(ccid);
+	struct dccp_sock *dp = dccp_sk(sk);
+	u32 max_ratio;
+
+	/* RFC 4341, 5: initialise ssthresh to arbitrarily high (max) value */
+	hctx->ccid2hctx_ssthresh  = ~0U;
 
-	ccid2_change_cwnd(hctx, 1);
-	/* Initialize ssthresh to infinity.  This means that we will exit the
-	 * initial slow-start after the first packet loss.  This is what we
-	 * want.
+	/*
+	 * RFC 4341, 5: "The cwnd parameter is initialized to at most four
+	 * packets for new connections, following the rules from [RFC3390]".
+	 * We need to convert the bytes of RFC3390 into the packets of RFC 4341.
 	 */
-	hctx->ccid2hctx_ssthresh  = ~0;
-	hctx->ccid2hctx_numdupack = 3;
+	hctx->ccid2hctx_cwnd = min(4U, max(2U, 4380U / dp->dccps_mss_cache));
+
+	/* Make sure that Ack Ratio is enabled and within bounds. */
+	max_ratio = DIV_ROUND_UP(hctx->ccid2hctx_cwnd, 2);
+	if (dp->dccps_l_ack_ratio == 0 || dp->dccps_l_ack_ratio > max_ratio)
+		dp->dccps_l_ack_ratio = max_ratio;
 
 	/* XXX init ~ to window size... */
 	if (ccid2_hc_tx_alloc_seq(hctx))
@@ -760,10 +732,8 @@ static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
 	hctx->ccid2hctx_rttvar	 = -1;
 	hctx->ccid2hctx_rpdupack = -1;
 	hctx->ccid2hctx_last_cong = jiffies;
-
-	hctx->ccid2hctx_rtotimer.function = &ccid2_hc_tx_rto_expire;
-	hctx->ccid2hctx_rtotimer.data	  = (unsigned long)sk;
-	init_timer(&hctx->ccid2hctx_rtotimer);
+	setup_timer(&hctx->ccid2hctx_rtotimer, ccid2_hc_tx_rto_expire,
+			(unsigned long)sk);
 
 	ccid2_hc_tx_check_sanity(hctx);
 	return 0;
@@ -800,7 +770,7 @@ static void ccid2_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
 
 static struct ccid_operations ccid2 = {
 	.ccid_id		= DCCPC_CCID2,
-	.ccid_name		= "ccid2",
+	.ccid_name		= "TCP-like",
 	.ccid_owner		= THIS_MODULE,
 	.ccid_hc_tx_obj_size	= sizeof(struct ccid2_hc_tx_sock),
 	.ccid_hc_tx_init	= ccid2_hc_tx_init,