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

Conflicts:

	net/dccp/input.c
	net/dccp/options.c

Documentation/networking/dccp.txt

@@ -45,6 +45,25 @@ http://linux-net.osdl.org/index.php/DCCP_Testing#Experimental_DCCP_source_tree
 
 Socket options
 ==============
+DCCP_SOCKOPT_QPOLICY_ID sets the dequeuing policy for outgoing packets. It takes
+a policy ID as argument and can only be set before the connection (i.e. changes
+during an established connection are not supported). Currently, two policies are
+defined: the "simple" policy (DCCPQ_POLICY_SIMPLE), which does nothing special,
+and a priority-based variant (DCCPQ_POLICY_PRIO). The latter allows to pass an
+u32 priority value as ancillary data to sendmsg(), where higher numbers indicate
+a higher packet priority (similar to SO_PRIORITY). This ancillary data needs to
+be formatted using a cmsg(3) message header filled in as follows:
+	cmsg->cmsg_level = SOL_DCCP;
+	cmsg->cmsg_type	 = DCCP_SCM_PRIORITY;
+	cmsg->cmsg_len	 = CMSG_LEN(sizeof(uint32_t));	/* or CMSG_LEN(4) */
+
+DCCP_SOCKOPT_QPOLICY_TXQLEN sets the maximum length of the output queue. A zero
+value is always interpreted as unbounded queue length. If different from zero,
+the interpretation of this parameter depends on the current dequeuing policy
+(see above): the "simple" policy will enforce a fixed queue size by returning
+EAGAIN, whereas the "prio" policy enforces a fixed queue length by dropping the
+lowest-priority packet first. The default value for this parameter is
+initialised from /proc/sys/net/dccp/default/tx_qlen.
 
 DCCP_SOCKOPT_SERVICE sets the service. The specification mandates use of
 service codes (RFC 4340, sec. 8.1.2); if this socket option is not set,
@@ -57,6 +76,24 @@ can be set before calling bind().
 DCCP_SOCKOPT_GET_CUR_MPS is read-only and retrieves the current maximum packet
 size (application payload size) in bytes, see RFC 4340, section 14.
 
+DCCP_SOCKOPT_AVAILABLE_CCIDS is also read-only and returns the list of CCIDs
+supported by the endpoint (see include/linux/dccp.h for symbolic constants).
+The caller needs to provide a sufficiently large (> 2) array of type uint8_t.
+
+DCCP_SOCKOPT_CCID is write-only and sets both the TX and RX CCIDs at the same
+time, combining the operation of the next two socket options. This option is
+preferrable over the latter two, since often applications will use the same
+type of CCID for both directions; and mixed use of CCIDs is not currently well
+understood. This socket option takes as argument at least one uint8_t value, or
+an array of uint8_t values, which must match available CCIDS (see above). CCIDs
+must be registered on the socket before calling connect() or listen().
+
+DCCP_SOCKOPT_TX_CCID is read/write. It returns the current CCID (if set) or sets
+the preference list for the TX CCID, using the same format as DCCP_SOCKOPT_CCID.
+Please note that the getsockopt argument type here is `int', not uint8_t.
+
+DCCP_SOCKOPT_RX_CCID is analogous to DCCP_SOCKOPT_TX_CCID, but for the RX CCID.
+
 DCCP_SOCKOPT_SERVER_TIMEWAIT enables the server (listening socket) to hold
 timewait state when closing the connection (RFC 4340, 8.3). The usual case is
 that the closing server sends a CloseReq, whereupon the client holds timewait
@@ -115,23 +152,16 @@ retries2
 	importance for retransmitted acknowledgments and feature negotiation,
 	data packets are never retransmitted. Analogue of tcp_retries2.
 
-send_ndp = 1
-	Whether or not to send NDP count options (sec. 7.7.2).
-
-send_ackvec = 1
-	Whether or not to send Ack Vector options (sec. 11.5).
-
-ack_ratio = 2
-	The default Ack Ratio (sec. 11.3) to use.
-
 tx_ccid = 2
-	Default CCID for the sender-receiver half-connection.
+	Default CCID for the sender-receiver half-connection. Depending on the
+	choice of CCID, the Send Ack Vector feature is enabled automatically.
 
 rx_ccid = 2
-	Default CCID for the receiver-sender half-connection.
+	Default CCID for the receiver-sender half-connection; see tx_ccid.
 
 seq_window = 100
-	The initial sequence window (sec. 7.5.2).
+	The initial sequence window (sec. 7.5.2) of the sender. This influences
+	the local ackno validity and the remote seqno validity windows (7.5.1).
 
 tx_qlen = 5
 	The size of the transmit buffer in packets. A value of 0 corresponds

include/linux/dccp.h

@@ -165,9 +165,13 @@ enum {
 	DCCPO_TIMESTAMP_ECHO = 42,
 	DCCPO_ELAPSED_TIME = 43,
 	DCCPO_MAX = 45,
-	DCCPO_MIN_CCID_SPECIFIC = 128,
-	DCCPO_MAX_CCID_SPECIFIC = 255,
+	DCCPO_MIN_RX_CCID_SPECIFIC = 128,	/* from sender to receiver */
+	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
 
 /* DCCP CCIDS */
 enum {
@@ -176,27 +180,36 @@ enum {
 };
 
 /* DCCP features (RFC 4340 section 6.4) */
-enum {
+enum dccp_feature_numbers {
 	DCCPF_RESERVED = 0,
 	DCCPF_CCID = 1,
-	DCCPF_SHORT_SEQNOS = 2,		/* XXX: not yet implemented */
+	DCCPF_SHORT_SEQNOS = 2,
 	DCCPF_SEQUENCE_WINDOW = 3,
-	DCCPF_ECN_INCAPABLE = 4,	/* XXX: not yet implemented */
+	DCCPF_ECN_INCAPABLE = 4,
 	DCCPF_ACK_RATIO = 5,
 	DCCPF_SEND_ACK_VECTOR = 6,
 	DCCPF_SEND_NDP_COUNT = 7,
 	DCCPF_MIN_CSUM_COVER = 8,
-	DCCPF_DATA_CHECKSUM = 9,	/* XXX: not yet implemented */
+	DCCPF_DATA_CHECKSUM = 9,
 	/* 10-127 reserved */
 	DCCPF_MIN_CCID_SPECIFIC = 128,
+	DCCPF_SEND_LEV_RATE = 192,	/* RFC 4342, sec. 8.4 */
 	DCCPF_MAX_CCID_SPECIFIC = 255,
 };
 
-/* this structure is argument to DCCP_SOCKOPT_CHANGE_X */
-struct dccp_so_feat {
-	__u8 dccpsf_feat;
-	__u8 __user *dccpsf_val;
-	__u8 dccpsf_len;
+/* DCCP socket control message types for cmsg */
+enum dccp_cmsg_type {
+	DCCP_SCM_PRIORITY = 1,
+	DCCP_SCM_QPOLICY_MAX = 0xFFFF,
+	/* ^-- Up to here reserved exclusively for qpolicy parameters */
+	DCCP_SCM_MAX
+};
+
+/* DCCP priorities for outgoing/queued packets */
+enum dccp_packet_dequeueing_policy {
+	DCCPQ_POLICY_SIMPLE,
+	DCCPQ_POLICY_PRIO,
+	DCCPQ_POLICY_MAX
 };
 
 /* DCCP socket options */
@@ -208,6 +221,12 @@ struct dccp_so_feat {
 #define DCCP_SOCKOPT_SERVER_TIMEWAIT	6
 #define DCCP_SOCKOPT_SEND_CSCOV		10
 #define DCCP_SOCKOPT_RECV_CSCOV		11
+#define DCCP_SOCKOPT_AVAILABLE_CCIDS	12
+#define DCCP_SOCKOPT_CCID		13
+#define DCCP_SOCKOPT_TX_CCID		14
+#define DCCP_SOCKOPT_RX_CCID		15
+#define DCCP_SOCKOPT_QPOLICY_ID		16
+#define DCCP_SOCKOPT_QPOLICY_TXQLEN	17
 #define DCCP_SOCKOPT_CCID_RX_INFO	128
 #define DCCP_SOCKOPT_CCID_TX_INFO	192
 
@@ -355,62 +374,13 @@ static inline unsigned int dccp_hdr_len(const struct sk_buff *skb)
 	return __dccp_hdr_len(dccp_hdr(skb));
 }
 
-
-/* initial values for each feature */
-#define DCCPF_INITIAL_SEQUENCE_WINDOW		100
-#define DCCPF_INITIAL_ACK_RATIO			2
-#define DCCPF_INITIAL_CCID			DCCPC_CCID2
-#define DCCPF_INITIAL_SEND_ACK_VECTOR		1
-/* FIXME: for now we're default to 1 but it should really be 0 */
-#define DCCPF_INITIAL_SEND_NDP_COUNT		1
-
-/**
-  * struct dccp_minisock - Minimal DCCP connection representation
-  *
-  * Will be used to pass the state from dccp_request_sock to dccp_sock.
-  *
-  * @dccpms_sequence_window - Sequence Window Feature (section 7.5.2)
-  * @dccpms_ccid - Congestion Control Id (CCID) (section 10)
-  * @dccpms_send_ack_vector - Send Ack Vector Feature (section 11.5)
-  * @dccpms_send_ndp_count - Send NDP Count Feature (7.7.2)
-  * @dccpms_ack_ratio - Ack Ratio Feature (section 11.3)
-  * @dccpms_pending - List of features being negotiated
-  * @dccpms_conf -
-  */
-struct dccp_minisock {
-	__u64			dccpms_sequence_window;
-	__u8			dccpms_rx_ccid;
-	__u8			dccpms_tx_ccid;
-	__u8			dccpms_send_ack_vector;
-	__u8			dccpms_send_ndp_count;
-	__u8			dccpms_ack_ratio;
-	struct list_head	dccpms_pending;
-	struct list_head	dccpms_conf;
-};
-
-struct dccp_opt_conf {
-	__u8			*dccpoc_val;
-	__u8			dccpoc_len;
-};
-
-struct dccp_opt_pend {
-	struct list_head	dccpop_node;
-	__u8			dccpop_type;
-	__u8			dccpop_feat;
-	__u8		        *dccpop_val;
-	__u8			dccpop_len;
-	int			dccpop_conf;
-	struct dccp_opt_conf    *dccpop_sc;
-};
-
-extern void dccp_minisock_init(struct dccp_minisock *dmsk);
-
 /**
  * struct dccp_request_sock  -  represent DCCP-specific connection request
  * @dreq_inet_rsk: structure inherited from
  * @dreq_iss: initial sequence number sent on the Response (RFC 4340, 7.1)
  * @dreq_isr: initial sequence number received on the Request
  * @dreq_service: service code present on the Request (there is just one)
+ * @dreq_featneg: feature negotiation options for this connection
  * The following two fields are analogous to the ones in dccp_sock:
  * @dreq_timestamp_echo: last received timestamp to echo (13.1)
  * @dreq_timestamp_echo: the time of receiving the last @dreq_timestamp_echo
@@ -420,6 +390,7 @@ struct dccp_request_sock {
 	__u64			 dreq_iss;
 	__u64			 dreq_isr;
 	__be32			 dreq_service;
+	struct list_head	 dreq_featneg;
 	__u32			 dreq_timestamp_echo;
 	__u32			 dreq_timestamp_time;
 };
@@ -491,21 +462,28 @@ struct dccp_ackvec;
  * @dccps_timestamp_time - time of receiving latest @dccps_timestamp_echo
  * @dccps_l_ack_ratio - feature-local Ack Ratio
  * @dccps_r_ack_ratio - feature-remote Ack Ratio
+ * @dccps_l_seq_win - local Sequence Window (influences ack number validity)
+ * @dccps_r_seq_win - remote Sequence Window (influences seq number validity)
  * @dccps_pcslen - sender   partial checksum coverage (via sockopt)
  * @dccps_pcrlen - receiver partial checksum coverage (via sockopt)
+ * @dccps_send_ndp_count - local Send NDP Count feature (7.7.2)
  * @dccps_ndp_count - number of Non Data Packets since last data packet
  * @dccps_mss_cache - current value of MSS (path MTU minus header sizes)
  * @dccps_rate_last - timestamp for rate-limiting DCCP-Sync (RFC 4340, 7.5.4)
- * @dccps_minisock - associated minisock (accessed via dccp_msk)
+ * @dccps_featneg - tracks feature-negotiation state (mostly during handshake)
  * @dccps_hc_rx_ackvec - rx half connection ack vector
  * @dccps_hc_rx_ccid - CCID used for the receiver (or receiving half-connection)
  * @dccps_hc_tx_ccid - CCID used for the sender (or sending half-connection)
  * @dccps_options_received - parsed set of retrieved options
+ * @dccps_qpolicy - TX dequeueing policy, one of %dccp_packet_dequeueing_policy
+ * @dccps_tx_qlen - maximum length of the TX queue
  * @dccps_role - role of this sock, one of %dccp_role
  * @dccps_hc_rx_insert_options - receiver wants to add options when acking
  * @dccps_hc_tx_insert_options - sender wants to add options when sending
  * @dccps_server_timewait - server holds timewait state on close (RFC 4340, 8.3)
- * @dccps_xmit_timer - timer for when CCID is not ready to send
+ * @dccps_sync_scheduled - flag which signals "send out-of-band message soon"
+ * @dccps_xmitlet - tasklet scheduled by the TX CCID to dequeue data packets
+ * @dccps_xmit_timer - used by the TX CCID to delay sending (rate-based pacing)
  * @dccps_syn_rtt - RTT sample from Request/Response exchange (in usecs)
  */
 struct dccp_sock {
@@ -529,19 +507,26 @@ struct dccp_sock {
 	__u32				dccps_timestamp_time;
 	__u16				dccps_l_ack_ratio;
 	__u16				dccps_r_ack_ratio;
-	__u16				dccps_pcslen;
-	__u16				dccps_pcrlen;
+	__u64				dccps_l_seq_win:48;
+	__u64				dccps_r_seq_win:48;
+	__u8				dccps_pcslen:4;
+	__u8				dccps_pcrlen:4;
+	__u8				dccps_send_ndp_count:1;
 	__u64				dccps_ndp_count:48;
 	unsigned long			dccps_rate_last;
-	struct dccp_minisock		dccps_minisock;
+	struct list_head		dccps_featneg;
 	struct dccp_ackvec		*dccps_hc_rx_ackvec;
 	struct ccid			*dccps_hc_rx_ccid;
 	struct ccid			*dccps_hc_tx_ccid;
 	struct dccp_options_received	dccps_options_received;
+	__u8				dccps_qpolicy;
+	__u32				dccps_tx_qlen;
 	enum dccp_role			dccps_role:2;
 	__u8				dccps_hc_rx_insert_options:1;
 	__u8				dccps_hc_tx_insert_options:1;
 	__u8				dccps_server_timewait:1;
+	__u8				dccps_sync_scheduled:1;
+	struct tasklet_struct		dccps_xmitlet;
 	struct timer_list		dccps_xmit_timer;
 };
 
@@ -550,11 +535,6 @@ static inline struct dccp_sock *dccp_sk(const struct sock *sk)
 	return (struct dccp_sock *)sk;
 }
 
-static inline struct dccp_minisock *dccp_msk(const struct sock *sk)
-{
-	return (struct dccp_minisock *)&dccp_sk(sk)->dccps_minisock;
-}
-
 static inline const char *dccp_role(const struct sock *sk)
 {
 	switch (dccp_sk(sk)->dccps_role) {

include/net/tcp.h

@@ -782,6 +782,21 @@ static inline __u32 tcp_current_ssthresh(const struct sock *sk)
 /* Use define here intentionally to get WARN_ON location shown at the caller */
 #define tcp_verify_left_out(tp)	WARN_ON(tcp_left_out(tp) > tp->packets_out)
 
+/*
+ * Convert RFC3390 larger initial windows into an equivalent number of packets.
+ *
+ * John Heffner states:
+ *
+ *	The RFC specifies a window of no more than 4380 bytes
+ *	unless 2*MSS > 4380.  Reading the pseudocode in the RFC
+ *	is a bit misleading because they use a clamp at 4380 bytes
+ *	rather than a multiplier in the relevant range.
+ */
+static inline u32 rfc3390_bytes_to_packets(const u32 bytes)
+{
+	return bytes <= 1095 ? 4 : (bytes > 1460 ? 2 : 3);
+}
+
 extern void tcp_enter_cwr(struct sock *sk, const int set_ssthresh);
 extern __u32 tcp_init_cwnd(struct tcp_sock *tp, struct dst_entry *dst);
 

net/dccp/Kconfig

@@ -25,9 +25,6 @@ config INET_DCCP_DIAG
 	def_tristate y if (IP_DCCP = y && INET_DIAG = y)
 	def_tristate m
 
-config IP_DCCP_ACKVEC
-	bool
-
 source "net/dccp/ccids/Kconfig"
 
 menu "DCCP Kernel Hacking"

net/dccp/Makefile

@@ -1,6 +1,7 @@
 obj-$(CONFIG_IP_DCCP) += dccp.o dccp_ipv4.o
 
-dccp-y := ccid.o feat.o input.o minisocks.o options.o output.o proto.o timer.o
+dccp-y := ccid.o feat.o input.o minisocks.o options.o \
+	  qpolicy.o output.o proto.o timer.o ackvec.o
 
 dccp_ipv4-y := ipv4.o
 
@@ -8,8 +9,6 @@ dccp_ipv4-y := ipv4.o
 obj-$(subst y,$(CONFIG_IP_DCCP),$(CONFIG_IPV6)) += dccp_ipv6.o
 dccp_ipv6-y := ipv6.o
 
-dccp-$(CONFIG_IP_DCCP_ACKVEC) += ackvec.o
-
 obj-$(CONFIG_INET_DCCP_DIAG) += dccp_diag.o
 obj-$(CONFIG_NET_DCCPPROBE) += dccp_probe.o
 

net/dccp/ackvec.c

@@ -1,445 +1,375 @@
 /*
  *  net/dccp/ackvec.c
  *
- *  An implementation of the DCCP protocol
+ *  An implementation of Ack Vectors for the DCCP protocol
+ *  Copyright (c) 2007 University of Aberdeen, Scotland, UK
  *  Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
  *
  *      This program 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; version 2 of the License;
  */
-
-#include "ackvec.h"
 #include "dccp.h"
-
-#include <linux/dccp.h>
-#include <linux/init.h>
-#include <linux/errno.h>
 #include <linux/kernel.h>
-#include <linux/skbuff.h>
 #include <linux/slab.h>
 
-#include <net/sock.h>
-
 static struct kmem_cache *dccp_ackvec_slab;
 static struct kmem_cache *dccp_ackvec_record_slab;
 
-static struct dccp_ackvec_record *dccp_ackvec_record_new(void)
+struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority)
 {
-	struct dccp_ackvec_record *avr =
-			kmem_cache_alloc(dccp_ackvec_record_slab, GFP_ATOMIC);
+	struct dccp_ackvec *av = kmem_cache_zalloc(dccp_ackvec_slab, priority);
 
-	if (avr != NULL)
-		INIT_LIST_HEAD(&avr->avr_node);
-
-	return avr;
+	if (av != NULL) {
+		av->av_buf_head	= av->av_buf_tail = DCCPAV_MAX_ACKVEC_LEN - 1;
+		INIT_LIST_HEAD(&av->av_records);
+	}
+	return av;
 }
 
-static void dccp_ackvec_record_delete(struct dccp_ackvec_record *avr)
+static void dccp_ackvec_purge_records(struct dccp_ackvec *av)
 {
-	if (unlikely(avr == NULL))
-		return;
-	/* Check if deleting a linked record */
-	WARN_ON(!list_empty(&avr->avr_node));
-	kmem_cache_free(dccp_ackvec_record_slab, avr);
+	struct dccp_ackvec_record *cur, *next;
+
+	list_for_each_entry_safe(cur, next, &av->av_records, avr_node)
+		kmem_cache_free(dccp_ackvec_record_slab, cur);
+	INIT_LIST_HEAD(&av->av_records);
 }
 
-static void dccp_ackvec_insert_avr(struct dccp_ackvec *av,
-				   struct dccp_ackvec_record *avr)
+void dccp_ackvec_free(struct dccp_ackvec *av)
 {
-	/*
-	 * AVRs are sorted by seqno. Since we are sending them in order, we
-	 * just add the AVR at the head of the list.
-	 * -sorbo.
-	 */
-	if (!list_empty(&av->av_records)) {
-		const struct dccp_ackvec_record *head =
-					list_entry(av->av_records.next,
-						   struct dccp_ackvec_record,
-						   avr_node);
-		BUG_ON(before48(avr->avr_ack_seqno, head->avr_ack_seqno));
+	if (likely(av != NULL)) {
+		dccp_ackvec_purge_records(av);
+		kmem_cache_free(dccp_ackvec_slab, av);
 	}
-
-	list_add(&avr->avr_node, &av->av_records);
 }
 
-int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb)
+/**
+ * dccp_ackvec_update_records  -  Record information about sent Ack Vectors
+ * @av:		Ack Vector records to update
+ * @seqno:	Sequence number of the packet carrying the Ack Vector just sent
+ * @nonce_sum:	The sum of all buffer nonces contained in the Ack Vector
+ */
+int dccp_ackvec_update_records(struct dccp_ackvec *av, u64 seqno, u8 nonce_sum)
 {
-	struct dccp_sock *dp = dccp_sk(sk);
-	struct dccp_ackvec *av = dp->dccps_hc_rx_ackvec;
-	/* Figure out how many options do we need to represent the ackvec */
-	const u16 nr_opts = DIV_ROUND_UP(av->av_vec_len, DCCP_MAX_ACKVEC_OPT_LEN);
-	u16 len = av->av_vec_len + 2 * nr_opts, i;
-	u32 elapsed_time;
-	const unsigned char *tail, *from;
-	unsigned char *to;
 	struct dccp_ackvec_record *avr;
-	suseconds_t delta;
-
-	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN)
-		return -1;
-
-	delta = ktime_us_delta(ktime_get_real(), av->av_time);
-	elapsed_time = delta / 10;
 
-	if (elapsed_time != 0 &&
-	    dccp_insert_option_elapsed_time(sk, skb, elapsed_time))
-		return -1;
-
-	avr = dccp_ackvec_record_new();
+	avr = kmem_cache_alloc(dccp_ackvec_record_slab, GFP_ATOMIC);
 	if (avr == NULL)
-		return -1;
-
-	DCCP_SKB_CB(skb)->dccpd_opt_len += len;
-
-	to   = skb_push(skb, len);
-	len  = av->av_vec_len;
-	from = av->av_buf + av->av_buf_head;
-	tail = av->av_buf + DCCP_MAX_ACKVEC_LEN;
-
-	for (i = 0; i < nr_opts; ++i) {
-		int copylen = len;
-
-		if (len > DCCP_MAX_ACKVEC_OPT_LEN)
-			copylen = DCCP_MAX_ACKVEC_OPT_LEN;
-
-		*to++ = DCCPO_ACK_VECTOR_0;
-		*to++ = copylen + 2;
-
-		/* Check if buf_head wraps */
-		if (from + copylen > tail) {
-			const u16 tailsize = tail - from;
-
-			memcpy(to, from, tailsize);
-			to	+= tailsize;
-			len	-= tailsize;
-			copylen	-= tailsize;
-			from	= av->av_buf;
-		}
-
-		memcpy(to, from, copylen);
-		from += copylen;
-		to   += copylen;
-		len  -= copylen;
-	}
+		return -ENOBUFS;
 
+	avr->avr_ack_seqno  = seqno;
+	avr->avr_ack_ptr    = av->av_buf_head;
+	avr->avr_ack_ackno  = av->av_buf_ackno;
+	avr->avr_ack_nonce  = nonce_sum;
+	avr->avr_ack_runlen = dccp_ackvec_runlen(av->av_buf + av->av_buf_head);
 	/*
-	 *	From RFC 4340, A.2:
-	 *
-	 *	For each acknowledgement it sends, the HC-Receiver will add an
-	 *	acknowledgement record.  ack_seqno will equal the HC-Receiver
-	 *	sequence number it used for the ack packet; ack_ptr will equal
-	 *	buf_head; ack_ackno will equal buf_ackno; and ack_nonce will
-	 *	equal buf_nonce.
+	 * When the buffer overflows, we keep no more than one record. This is
+	 * the simplest way of disambiguating sender-Acks dating from before the
+	 * overflow from sender-Acks which refer to after the overflow; a simple
+	 * solution is preferable here since we are handling an exception.
 	 */
-	avr->avr_ack_seqno = DCCP_SKB_CB(skb)->dccpd_seq;
-	avr->avr_ack_ptr   = av->av_buf_head;
-	avr->avr_ack_ackno = av->av_buf_ackno;
-	avr->avr_ack_nonce = av->av_buf_nonce;
-	avr->avr_sent_len  = av->av_vec_len;
-
-	dccp_ackvec_insert_avr(av, avr);
+	if (av->av_overflow)
+		dccp_ackvec_purge_records(av);
+	/*
+	 * Since GSS is incremented for each packet, the list is automatically
+	 * arranged in descending order of @ack_seqno.
+	 */
+	list_add(&avr->avr_node, &av->av_records);
 
-	dccp_pr_debug("%s ACK Vector 0, len=%d, ack_seqno=%llu, "
-		      "ack_ackno=%llu\n",
-		      dccp_role(sk), avr->avr_sent_len,
+	dccp_pr_debug("Added Vector, ack_seqno=%llu, ack_ackno=%llu (rl=%u)\n",
 		      (unsigned long long)avr->avr_ack_seqno,
-		      (unsigned long long)avr->avr_ack_ackno);
+		      (unsigned long long)avr->avr_ack_ackno,
+		      avr->avr_ack_runlen);
 	return 0;
 }
 
-struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority)
+static struct dccp_ackvec_record *dccp_ackvec_lookup(struct list_head *av_list,
+						     const u64 ackno)
 {
-	struct dccp_ackvec *av = kmem_cache_alloc(dccp_ackvec_slab, priority);
-
-	if (av != NULL) {
-		av->av_buf_head	 = DCCP_MAX_ACKVEC_LEN - 1;
-		av->av_buf_ackno = UINT48_MAX + 1;
-		av->av_buf_nonce = 0;
-		av->av_time	 = ktime_set(0, 0);
-		av->av_vec_len	 = 0;
-		INIT_LIST_HEAD(&av->av_records);
+	struct dccp_ackvec_record *avr;
+	/*
+	 * Exploit that records are inserted in descending order of sequence
+	 * number, start with the oldest record first. If @ackno is `before'
+	 * the earliest ack_ackno, the packet is too old to be considered.
+	 */
+	list_for_each_entry_reverse(avr, av_list, avr_node) {
+		if (avr->avr_ack_seqno == ackno)
+			return avr;
+		if (before48(ackno, avr->avr_ack_seqno))
+			break;
 	}
-
-	return av;
+	return NULL;
 }
 
-void dccp_ackvec_free(struct dccp_ackvec *av)
+/*
+ * Buffer index and length computation using modulo-buffersize arithmetic.
+ * Note that, as pointers move from right to left, head is `before' tail.
+ */
+static inline u16 __ackvec_idx_add(const u16 a, const u16 b)
 {
-	if (unlikely(av == NULL))
-		return;
-
-	if (!list_empty(&av->av_records)) {
-		struct dccp_ackvec_record *avr, *next;
-
-		list_for_each_entry_safe(avr, next, &av->av_records, avr_node) {
-			list_del_init(&avr->avr_node);
-			dccp_ackvec_record_delete(avr);
-		}
-	}
-
-	kmem_cache_free(dccp_ackvec_slab, av);
+	return (a + b) % DCCPAV_MAX_ACKVEC_LEN;
 }
 
-static inline u8 dccp_ackvec_state(const struct dccp_ackvec *av,
-				   const u32 index)
+static inline u16 __ackvec_idx_sub(const u16 a, const u16 b)
 {
-	return av->av_buf[index] & DCCP_ACKVEC_STATE_MASK;
+	return __ackvec_idx_add(a, DCCPAV_MAX_ACKVEC_LEN - b);
 }
 
-static inline u8 dccp_ackvec_len(const struct dccp_ackvec *av,
-				 const u32 index)
+u16 dccp_ackvec_buflen(const struct dccp_ackvec *av)
 {
-	return av->av_buf[index] & DCCP_ACKVEC_LEN_MASK;
+	if (unlikely(av->av_overflow))
+		return DCCPAV_MAX_ACKVEC_LEN;
+	return __ackvec_idx_sub(av->av_buf_tail, av->av_buf_head);
 }
 
-/*
- * If several packets are missing, the HC-Receiver may prefer to enter multiple
- * bytes with run length 0, rather than a single byte with a larger run length;
- * this simplifies table updates if one of the missing packets arrives.
+/**
+ * dccp_ackvec_update_old  -  Update previous state as per RFC 4340, 11.4.1
+ * @av:		non-empty buffer to update
+ * @distance:   negative or zero distance of @seqno from buf_ackno downward
+ * @seqno:	the (old) sequence number whose record is to be updated
+ * @state:	state in which packet carrying @seqno was received
  */
-static inline int dccp_ackvec_set_buf_head_state(struct dccp_ackvec *av,
-						 const unsigned int packets,
-						 const unsigned char state)
+static void dccp_ackvec_update_old(struct dccp_ackvec *av, s64 distance,
+				   u64 seqno, enum dccp_ackvec_states state)
 {
-	unsigned int gap;
-	long new_head;
+	u16 ptr = av->av_buf_head;
 
-	if (av->av_vec_len + packets > DCCP_MAX_ACKVEC_LEN)
-		return -ENOBUFS;
+	BUG_ON(distance > 0);
+	if (unlikely(dccp_ackvec_is_empty(av)))
+		return;
 
-	gap	 = packets - 1;
-	new_head = av->av_buf_head - packets;
+	do {
+		u8 runlen = dccp_ackvec_runlen(av->av_buf + ptr);
 
-	if (new_head < 0) {
-		if (gap > 0) {
-			memset(av->av_buf, DCCP_ACKVEC_STATE_NOT_RECEIVED,
-			       gap + new_head + 1);
-			gap = -new_head;
+		if (distance + runlen >= 0) {
+			/*
+			 * Only update the state if packet has not been received
+			 * yet. This is OK as per the second table in RFC 4340,
+			 * 11.4.1; i.e. here we are using the following table:
+			 *                     RECEIVED
+			 *                      0   1   3
+			 *              S     +---+---+---+
+			 *              T   0 | 0 | 0 | 0 |
+			 *              O     +---+---+---+
+			 *              R   1 | 1 | 1 | 1 |
+			 *              E     +---+---+---+
+			 *              D   3 | 0 | 1 | 3 |
+			 *                    +---+---+---+
+			 * The "Not Received" state was set by reserve_seats().
+			 */
+			if (av->av_buf[ptr] == DCCPAV_NOT_RECEIVED)
+				av->av_buf[ptr] = state;
+			else
+				dccp_pr_debug("Not changing %llu state to %u\n",
+					      (unsigned long long)seqno, state);
+			break;
 		}
-		new_head += DCCP_MAX_ACKVEC_LEN;
-	}
 
-	av->av_buf_head = new_head;
+		distance += runlen + 1;
+		ptr	  = __ackvec_idx_add(ptr, 1);
 
-	if (gap > 0)
-		memset(av->av_buf + av->av_buf_head + 1,
-		       DCCP_ACKVEC_STATE_NOT_RECEIVED, gap);
+	} while (ptr != av->av_buf_tail);
+}
 
-	av->av_buf[av->av_buf_head] = state;
-	av->av_vec_len += packets;
-	return 0;
+/* Mark @num entries after buf_head as "Not yet received". */
+static void dccp_ackvec_reserve_seats(struct dccp_ackvec *av, u16 num)
+{
+	u16 start = __ackvec_idx_add(av->av_buf_head, 1),
+	    len	  = DCCPAV_MAX_ACKVEC_LEN - start;
+
+	/* check for buffer wrap-around */
+	if (num > len) {
+		memset(av->av_buf + start, DCCPAV_NOT_RECEIVED, len);
+		start = 0;
+		num  -= len;
+	}
+	if (num)
+		memset(av->av_buf + start, DCCPAV_NOT_RECEIVED, num);
 }
 
-/*
- * Implements the RFC 4340, Appendix A
+/**
+ * dccp_ackvec_add_new  -  Record one or more new entries in Ack Vector buffer
+ * @av:		 container of buffer to update (can be empty or non-empty)
+ * @num_packets: number of packets to register (must be >= 1)
+ * @seqno:	 sequence number of the first packet in @num_packets
+ * @state:	 state in which packet carrying @seqno was received
  */
-int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk,
-		    const u64 ackno, const u8 state)
+static void dccp_ackvec_add_new(struct dccp_ackvec *av, u32 num_packets,
+				u64 seqno, enum dccp_ackvec_states state)
 {
-	/*
-	 * Check at the right places if the buffer is full, if it is, tell the
-	 * caller to start dropping packets till the HC-Sender acks our ACK
-	 * vectors, when we will free up space in av_buf.
-	 *
-	 * We may well decide to do buffer compression, etc, but for now lets
-	 * just drop.
-	 *
-	 * From Appendix A.1.1 (`New Packets'):
-	 *
-	 *	Of course, the circular buffer may overflow, either when the
-	 *	HC-Sender is sending data at a very high rate, when the
-	 *	HC-Receiver's acknowledgements are not reaching the HC-Sender,
-	 *	or when the HC-Sender is forgetting to acknowledge those acks
-	 *	(so the HC-Receiver is unable to clean up old state). In this
-	 *	case, the HC-Receiver should either compress the buffer (by
-	 *	increasing run lengths when possible), transfer its state to
-	 *	a larger buffer, or, as a last resort, drop all received
-	 *	packets, without processing them whatsoever, until its buffer
-	 *	shrinks again.
-	 */
+	u32 num_cells = num_packets;
 
-	/* See if this is the first ackno being inserted */
-	if (av->av_vec_len == 0) {
-		av->av_buf[av->av_buf_head] = state;
-		av->av_vec_len = 1;
-	} else if (after48(ackno, av->av_buf_ackno)) {
-		const u64 delta = dccp_delta_seqno(av->av_buf_ackno, ackno);
+	if (num_packets > DCCPAV_BURST_THRESH) {
+		u32 lost_packets = num_packets - 1;
 
+		DCCP_WARN("Warning: large burst loss (%u)\n", lost_packets);
 		/*
-		 * Look if the state of this packet is the same as the
-		 * previous ackno and if so if we can bump the head len.
+		 * We received 1 packet and have a loss of size "num_packets-1"
+		 * which we squeeze into num_cells-1 rather than reserving an
+		 * entire byte for each lost packet.
+		 * The reason is that the vector grows in O(burst_length); when
+		 * it grows too large there will no room left for the payload.
+		 * This is a trade-off: if a few packets out of the burst show
+		 * up later, their state will not be changed; it is simply too
+		 * costly to reshuffle/reallocate/copy the buffer each time.
+		 * Should such problems persist, we will need to switch to a
+		 * different underlying data structure.
 		 */
-		if (delta == 1 &&
-		    dccp_ackvec_state(av, av->av_buf_head) == state &&
-		    dccp_ackvec_len(av, av->av_buf_head) < DCCP_ACKVEC_LEN_MASK)
-			av->av_buf[av->av_buf_head]++;
-		else if (dccp_ackvec_set_buf_head_state(av, delta, state))
-			return -ENOBUFS;
-	} else {
-		/*
-		 * A.1.2.  Old Packets
-		 *
-		 *	When a packet with Sequence Number S <= buf_ackno
-		 *	arrives, the HC-Receiver will scan the table for
-		 *	the byte corresponding to S. (Indexing structures
-		 *	could reduce the complexity of this scan.)
-		 */
-		u64 delta = dccp_delta_seqno(ackno, av->av_buf_ackno);
-		u32 index = av->av_buf_head;
+		for (num_packets = num_cells = 1; lost_packets; ++num_cells) {
+			u8 len = min(lost_packets, (u32)DCCPAV_MAX_RUNLEN);
 
-		while (1) {
-			const u8 len = dccp_ackvec_len(av, index);
-			const u8 av_state = dccp_ackvec_state(av, index);
-			/*
-			 * valid packets not yet in av_buf have a reserved
-			 * entry, with a len equal to 0.
-			 */
-			if (av_state == DCCP_ACKVEC_STATE_NOT_RECEIVED &&
-			    len == 0 && delta == 0) { /* Found our
-							 reserved seat! */
-				dccp_pr_debug("Found %llu reserved seat!\n",
-					      (unsigned long long)ackno);
-				av->av_buf[index] = state;
-				goto out;
-			}
-			/* len == 0 means one packet */
-			if (delta < len + 1)
-				goto out_duplicate;
-
-			delta -= len + 1;
-			if (++index == DCCP_MAX_ACKVEC_LEN)
-				index = 0;
+			av->av_buf_head = __ackvec_idx_sub(av->av_buf_head, 1);
+			av->av_buf[av->av_buf_head] = DCCPAV_NOT_RECEIVED | len;
+
+			lost_packets -= len;
 		}
 	}
 
-	av->av_buf_ackno = ackno;
-	av->av_time = ktime_get_real();
-out:
-	return 0;
+	if (num_cells + dccp_ackvec_buflen(av) >= DCCPAV_MAX_ACKVEC_LEN) {
+		DCCP_CRIT("Ack Vector buffer overflow: dropping old entries\n");
+		av->av_overflow = true;
+	}
+
+	av->av_buf_head = __ackvec_idx_sub(av->av_buf_head, num_packets);
+	if (av->av_overflow)
+		av->av_buf_tail = av->av_buf_head;
 
-out_duplicate:
-	/* Duplicate packet */
-	dccp_pr_debug("Received a dup or already considered lost "
-		      "packet: %llu\n", (unsigned long long)ackno);
-	return -EILSEQ;
+	av->av_buf[av->av_buf_head] = state;
+	av->av_buf_ackno	    = seqno;
+
+	if (num_packets > 1)
+		dccp_ackvec_reserve_seats(av, num_packets - 1);
 }
 
-static void dccp_ackvec_throw_record(struct dccp_ackvec *av,
-				     struct dccp_ackvec_record *avr)
+/**
+ * dccp_ackvec_input  -  Register incoming packet in the buffer
+ */
+void dccp_ackvec_input(struct dccp_ackvec *av, struct sk_buff *skb)
 {
-	struct dccp_ackvec_record *next;
+	u64 seqno = DCCP_SKB_CB(skb)->dccpd_seq;
+	enum dccp_ackvec_states state = DCCPAV_RECEIVED;
 
-	/* sort out vector length */
-	if (av->av_buf_head <= avr->avr_ack_ptr)
-		av->av_vec_len = avr->avr_ack_ptr - av->av_buf_head;
-	else
-		av->av_vec_len = DCCP_MAX_ACKVEC_LEN - 1 -
-				 av->av_buf_head + avr->avr_ack_ptr;
+	if (dccp_ackvec_is_empty(av)) {
+		dccp_ackvec_add_new(av, 1, seqno, state);
+		av->av_tail_ackno = seqno;
 
-	/* free records */
-	list_for_each_entry_safe_from(avr, next, &av->av_records, avr_node) {
-		list_del_init(&avr->avr_node);
-		dccp_ackvec_record_delete(avr);
-	}
-}
+	} else {
+		s64 num_packets = dccp_delta_seqno(av->av_buf_ackno, seqno);
+		u8 *current_head = av->av_buf + av->av_buf_head;
 
-void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, struct sock *sk,
-				 const u64 ackno)
-{
-	struct dccp_ackvec_record *avr;
+		if (num_packets == 1 &&
+		    dccp_ackvec_state(current_head) == state &&
+		    dccp_ackvec_runlen(current_head) < DCCPAV_MAX_RUNLEN) {
 
-	/*
-	 * If we traverse backwards, it should be faster when we have large
-	 * windows. We will be receiving ACKs for stuff we sent a while back
-	 * -sorbo.
-	 */
-	list_for_each_entry_reverse(avr, &av->av_records, avr_node) {
-		if (ackno == avr->avr_ack_seqno) {
-			dccp_pr_debug("%s ACK packet 0, len=%d, ack_seqno=%llu, "
-				      "ack_ackno=%llu, ACKED!\n",
-				      dccp_role(sk), 1,
-				      (unsigned long long)avr->avr_ack_seqno,
-				      (unsigned long long)avr->avr_ack_ackno);
-			dccp_ackvec_throw_record(av, avr);
-			break;
-		} else if (avr->avr_ack_seqno > ackno)
-			break; /* old news */
+			*current_head   += 1;
+			av->av_buf_ackno = seqno;
+
+		} else if (num_packets > 0) {
+			dccp_ackvec_add_new(av, num_packets, seqno, state);
+		} else {
+			dccp_ackvec_update_old(av, num_packets, seqno, state);
+		}
 	}
 }
 
-static void dccp_ackvec_check_rcv_ackvector(struct dccp_ackvec *av,
-					    struct sock *sk, u64 *ackno,
-					    const unsigned char len,
-					    const unsigned char *vector)
-{
-	unsigned char i;
-	struct dccp_ackvec_record *avr;
+/**
+ * dccp_ackvec_clear_state  -  Perform house-keeping / garbage-collection
+ * This routine is called when the peer acknowledges the receipt of Ack Vectors
+ * up to and including @ackno. While based on on section A.3 of RFC 4340, here
+ * are additional precautions to prevent corrupted buffer state. In particular,
+ * we use tail_ackno to identify outdated records; it always marks the earliest
+ * packet of group (2) in 11.4.2.
+ */
+void dccp_ackvec_clear_state(struct dccp_ackvec *av, const u64 ackno)
+ {
+	struct dccp_ackvec_record *avr, *next;
+	u8 runlen_now, eff_runlen;
+	s64 delta;
 
-	/* Check if we actually sent an ACK vector */
-	if (list_empty(&av->av_records))
+	avr = dccp_ackvec_lookup(&av->av_records, ackno);
+	if (avr == NULL)
 		return;
+	/*
+	 * Deal with outdated acknowledgments: this arises when e.g. there are
+	 * several old records and the acks from the peer come in slowly. In
+	 * that case we may still have records that pre-date tail_ackno.
+	 */
+	delta = dccp_delta_seqno(av->av_tail_ackno, avr->avr_ack_ackno);
+	if (delta < 0)
+		goto free_records;
+	/*
+	 * Deal with overlapping Ack Vectors: don't subtract more than the
+	 * number of packets between tail_ackno and ack_ackno.
+	 */
+	eff_runlen = delta < avr->avr_ack_runlen ? delta : avr->avr_ack_runlen;
 
-	i = len;
+	runlen_now = dccp_ackvec_runlen(av->av_buf + avr->avr_ack_ptr);
 	/*
-	 * XXX
-	 * I think it might be more efficient to work backwards. See comment on
-	 * rcv_ackno. -sorbo.
+	 * The run length of Ack Vector cells does not decrease over time. If
+	 * the run length is the same as at the time the Ack Vector was sent, we
+	 * free the ack_ptr cell. That cell can however not be freed if the run
+	 * length has increased: in this case we need to move the tail pointer
+	 * backwards (towards higher indices), to its next-oldest neighbour.
 	 */
-	avr = list_entry(av->av_records.next, struct dccp_ackvec_record, avr_node);
-	while (i--) {
-		const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
-		u64 ackno_end_rl;
+	if (runlen_now > eff_runlen) {
 
-		dccp_set_seqno(&ackno_end_rl, *ackno - rl);
+		av->av_buf[avr->avr_ack_ptr] -= eff_runlen + 1;
+		av->av_buf_tail = __ackvec_idx_add(avr->avr_ack_ptr, 1);
 
+		/* This move may not have cleared the overflow flag. */
+		if (av->av_overflow)
+			av->av_overflow = (av->av_buf_head == av->av_buf_tail);
+	} else {
+		av->av_buf_tail	= avr->avr_ack_ptr;
 		/*
-		 * If our AVR sequence number is greater than the ack, go
-		 * forward in the AVR list until it is not so.
+		 * We have made sure that avr points to a valid cell within the
+		 * buffer. This cell is either older than head, or equals head
+		 * (empty buffer): in both cases we no longer have any overflow.
 		 */
-		list_for_each_entry_from(avr, &av->av_records, avr_node) {
-			if (!after48(avr->avr_ack_seqno, *ackno))
-				goto found;
-		}
-		/* End of the av_records list, not found, exit */
-		break;
-found:
-		if (between48(avr->avr_ack_seqno, ackno_end_rl, *ackno)) {
-			const u8 state = *vector & DCCP_ACKVEC_STATE_MASK;
-			if (state != DCCP_ACKVEC_STATE_NOT_RECEIVED) {
-				dccp_pr_debug("%s ACK vector 0, len=%d, "
-					      "ack_seqno=%llu, ack_ackno=%llu, "
-					      "ACKED!\n",
-					      dccp_role(sk), len,
-					      (unsigned long long)
-					      avr->avr_ack_seqno,
-					      (unsigned long long)
-					      avr->avr_ack_ackno);
-				dccp_ackvec_throw_record(av, avr);
-				break;
-			}
-			/*
-			 * If it wasn't received, continue scanning... we might
-			 * find another one.
-			 */
-		}
+		av->av_overflow	= 0;
+	}
 
-		dccp_set_seqno(ackno, ackno_end_rl - 1);
-		++vector;
+	/*
+	 * The peer has acknowledged up to and including ack_ackno. Hence the
+	 * first packet in group (2) of 11.4.2 is the successor of ack_ackno.
+	 */
+	av->av_tail_ackno = ADD48(avr->avr_ack_ackno, 1);
+
+free_records:
+	list_for_each_entry_safe_from(avr, next, &av->av_records, avr_node) {
+		list_del(&avr->avr_node);
+		kmem_cache_free(dccp_ackvec_record_slab, avr);
 	}
 }
 
-int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb,
-		      u64 *ackno, const u8 opt, const u8 *value, const u8 len)
+/*
+ *	Routines to keep track of Ack Vectors received in an skb
+ */
+int dccp_ackvec_parsed_add(struct list_head *head, u8 *vec, u8 len, u8 nonce)
 {
-	if (len > DCCP_MAX_ACKVEC_OPT_LEN)
-		return -1;
+	struct dccp_ackvec_parsed *new = kmalloc(sizeof(*new), GFP_ATOMIC);
+
+	if (new == NULL)
+		return -ENOBUFS;
+	new->vec   = vec;
+	new->len   = len;
+	new->nonce = nonce;
 
-	/* dccp_ackvector_print(DCCP_SKB_CB(skb)->dccpd_ack_seq, value, len); */
-	dccp_ackvec_check_rcv_ackvector(dccp_sk(sk)->dccps_hc_rx_ackvec, sk,
-					ackno, len, value);
+	list_add_tail(&new->node, head);
 	return 0;
 }
+EXPORT_SYMBOL_GPL(dccp_ackvec_parsed_add);
+
+void dccp_ackvec_parsed_cleanup(struct list_head *parsed_chunks)
+{
+	struct dccp_ackvec_parsed *cur, *next;
+
+	list_for_each_entry_safe(cur, next, parsed_chunks, node)
+		kfree(cur);
+	INIT_LIST_HEAD(parsed_chunks);
+}
+EXPORT_SYMBOL_GPL(dccp_ackvec_parsed_cleanup);
 
 int __init dccp_ackvec_init(void)
 {
@@ -449,10 +379,9 @@ int __init dccp_ackvec_init(void)
 	if (dccp_ackvec_slab == NULL)
 		goto out_err;
 
-	dccp_ackvec_record_slab =
-			kmem_cache_create("dccp_ackvec_record",
-					  sizeof(struct dccp_ackvec_record),
-					  0, SLAB_HWCACHE_ALIGN, NULL);
+	dccp_ackvec_record_slab = kmem_cache_create("dccp_ackvec_record",
+					     sizeof(struct dccp_ackvec_record),
+					     0, SLAB_HWCACHE_ALIGN, NULL);
 	if (dccp_ackvec_record_slab == NULL)
 		goto out_destroy_slab;
 

net/dccp/ackvec.h

@@ -3,156 +3,134 @@
 /*
  *  net/dccp/ackvec.h
  *
- *  An implementation of the DCCP protocol
+ *  An implementation of Ack Vectors for the DCCP protocol
+ *  Copyright (c) 2007 University of Aberdeen, Scotland, UK
  *  Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@mandriva.com>
- *
  *	This program is free software; you can redistribute it and/or modify it
  *	under the terms of the GNU General Public License version 2 as
  *	published by the Free Software Foundation.
  */
 
+#include <linux/dccp.h>
 #include <linux/compiler.h>
-#include <linux/ktime.h>
 #include <linux/list.h>
 #include <linux/types.h>
 
-/* Read about the ECN nonce to see why it is 253 */
-#define DCCP_MAX_ACKVEC_OPT_LEN 253
-/* We can spread an ack vector across multiple options */
-#define DCCP_MAX_ACKVEC_LEN (DCCP_MAX_ACKVEC_OPT_LEN * 2)
+/*
+ * Ack Vector buffer space is static, in multiples of %DCCP_SINGLE_OPT_MAXLEN,
+ * the maximum size of a single Ack Vector. Setting %DCCPAV_NUM_ACKVECS to 1
+ * will be sufficient for most cases of low Ack Ratios, using a value of 2 gives
+ * more headroom if Ack Ratio is higher or when the sender acknowledges slowly.
+ * The maximum value is bounded by the u16 types for indices and functions.
+ */
+#define DCCPAV_NUM_ACKVECS	2
+#define DCCPAV_MAX_ACKVEC_LEN	(DCCP_SINGLE_OPT_MAXLEN * DCCPAV_NUM_ACKVECS)
 
-#define DCCP_ACKVEC_STATE_RECEIVED	0
-#define DCCP_ACKVEC_STATE_ECN_MARKED	(1 << 6)
-#define DCCP_ACKVEC_STATE_NOT_RECEIVED	(3 << 6)
+/* Estimated minimum average Ack Vector length - used for updating MPS */
+#define DCCPAV_MIN_OPTLEN	16
 
-#define DCCP_ACKVEC_STATE_MASK		0xC0 /* 11000000 */
-#define DCCP_ACKVEC_LEN_MASK		0x3F /* 00111111 */
+/* Threshold for coping with large bursts of losses */
+#define DCCPAV_BURST_THRESH	(DCCPAV_MAX_ACKVEC_LEN / 8)
 
-/** struct dccp_ackvec - ack vector
- *
- * This data structure is the one defined in RFC 4340, Appendix A.
- *
- * @av_buf_head - circular buffer head
- * @av_buf_tail - circular buffer tail
- * @av_buf_ackno - ack # of the most recent packet acknowledgeable in the
- *		       buffer (i.e. %av_buf_head)
- * @av_buf_nonce - the one-bit sum of the ECN Nonces on all packets acked
- * 		       by the buffer with State 0
- *
- * Additionally, the HC-Receiver must keep some information about the
- * Ack Vectors it has recently sent. For each packet sent carrying an
- * Ack Vector, it remembers four variables:
+enum dccp_ackvec_states {
+	DCCPAV_RECEIVED =	0x00,
+	DCCPAV_ECN_MARKED =	0x40,
+	DCCPAV_RESERVED =	0x80,
+	DCCPAV_NOT_RECEIVED =	0xC0
+};
+#define DCCPAV_MAX_RUNLEN	0x3F
+
+static inline u8 dccp_ackvec_runlen(const u8 *cell)
+{
+	return *cell & DCCPAV_MAX_RUNLEN;
+}
+
+static inline u8 dccp_ackvec_state(const u8 *cell)
+{
+	return *cell & ~DCCPAV_MAX_RUNLEN;
+}
+
+/** struct dccp_ackvec - Ack Vector main data structure
  *
- * @av_records - list of dccp_ackvec_record
- * @av_ack_nonce - the one-bit sum of the ECN Nonces for all State 0.
+ * This implements a fixed-size circular buffer within an array and is largely
+ * based on Appendix A of RFC 4340.
  *
- * @av_time - the time in usecs
- * @av_buf - circular buffer of acknowledgeable packets
+ * @av_buf:	   circular buffer storage area
+ * @av_buf_head:   head index; begin of live portion in @av_buf
+ * @av_buf_tail:   tail index; first index _after_ the live portion in @av_buf
+ * @av_buf_ackno:  highest seqno of acknowledgeable packet recorded in @av_buf
+ * @av_tail_ackno: lowest  seqno of acknowledgeable packet recorded in @av_buf
+ * @av_buf_nonce:  ECN nonce sums, each covering subsequent segments of up to
+ *		   %DCCP_SINGLE_OPT_MAXLEN cells in the live portion of @av_buf
+ * @av_overflow:   if 1 then buf_head == buf_tail indicates buffer wraparound
+ * @av_records:	   list of %dccp_ackvec_record (Ack Vectors sent previously)
  */
 struct dccp_ackvec {
-	u64			av_buf_ackno;
-	struct list_head	av_records;
-	ktime_t			av_time;
+	u8			av_buf[DCCPAV_MAX_ACKVEC_LEN];
 	u16			av_buf_head;
-	u16			av_vec_len;
-	u8			av_buf_nonce;
-	u8			av_ack_nonce;
-	u8			av_buf[DCCP_MAX_ACKVEC_LEN];
+	u16			av_buf_tail;
+	u64			av_buf_ackno:48;
+	u64			av_tail_ackno:48;
+	bool			av_buf_nonce[DCCPAV_NUM_ACKVECS];
+	u8			av_overflow:1;
+	struct list_head	av_records;
 };
 
-/** struct dccp_ackvec_record - ack vector record
+/** struct dccp_ackvec_record - Records information about sent Ack Vectors
  *
- * ACK vector record as defined in Appendix A of spec.
+ * These list entries define the additional information which the HC-Receiver
+ * keeps about recently-sent Ack Vectors; again refer to RFC 4340, Appendix A.
  *
- * The list is sorted by avr_ack_seqno
+ * @avr_node:	    the list node in @av_records
+ * @avr_ack_seqno:  sequence number of the packet the Ack Vector was sent on
+ * @avr_ack_ackno:  the Ack number that this record/Ack Vector refers to
+ * @avr_ack_ptr:    pointer into @av_buf where this record starts
+ * @avr_ack_runlen: run length of @avr_ack_ptr at the time of sending
+ * @avr_ack_nonce:  the sum of @av_buf_nonce's at the time this record was sent
  *
- * @avr_node - node in av_records
- * @avr_ack_seqno - sequence number of the packet this record was sent on
- * @avr_ack_ackno - sequence number being acknowledged
- * @avr_ack_ptr - pointer into av_buf where this record starts
- * @avr_ack_nonce - av_ack_nonce at the time this record was sent
- * @avr_sent_len - lenght of the record in av_buf
+ * The list as a whole is sorted in descending order by @avr_ack_seqno.
  */
 struct dccp_ackvec_record {
 	struct list_head avr_node;
-	u64		 avr_ack_seqno;
-	u64		 avr_ack_ackno;
+	u64		 avr_ack_seqno:48;
+	u64		 avr_ack_ackno:48;
 	u16		 avr_ack_ptr;
-	u16		 avr_sent_len;
-	u8		 avr_ack_nonce;
+	u8		 avr_ack_runlen;
+	u8		 avr_ack_nonce:1;
 };
 
-struct sock;
-struct sk_buff;
-
-#ifdef CONFIG_IP_DCCP_ACKVEC
-extern int dccp_ackvec_init(void);
+extern int  dccp_ackvec_init(void);
 extern void dccp_ackvec_exit(void);
 
 extern struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority);
 extern void dccp_ackvec_free(struct dccp_ackvec *av);
 
-extern int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk,
-			   const u64 ackno, const u8 state);
-
-extern void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av,
-					struct sock *sk, const u64 ackno);
-extern int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb,
-			     u64 *ackno, const u8 opt,
-			     const u8 *value, const u8 len);
+extern void dccp_ackvec_input(struct dccp_ackvec *av, struct sk_buff *skb);
+extern int  dccp_ackvec_update_records(struct dccp_ackvec *av, u64 seq, u8 sum);
+extern void dccp_ackvec_clear_state(struct dccp_ackvec *av, const u64 ackno);
+extern u16  dccp_ackvec_buflen(const struct dccp_ackvec *av);
 
-extern int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb);
-
-static inline int dccp_ackvec_pending(const struct dccp_ackvec *av)
-{
-	return av->av_vec_len;
-}
-#else /* CONFIG_IP_DCCP_ACKVEC */
-static inline int dccp_ackvec_init(void)
+static inline bool dccp_ackvec_is_empty(const struct dccp_ackvec *av)
 {
-	return 0;
+	return av->av_overflow == 0 && av->av_buf_head == av->av_buf_tail;
 }
 
-static inline void dccp_ackvec_exit(void)
-{
-}
-
-static inline struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority)
-{
-	return NULL;
-}
-
-static inline void dccp_ackvec_free(struct dccp_ackvec *av)
-{
-}
-
-static inline int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk,
-				  const u64 ackno, const u8 state)
-{
-	return -1;
-}
-
-static inline void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av,
-					       struct sock *sk, const u64 ackno)
-{
-}
-
-static inline int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb,
-				    const u64 *ackno, const u8 opt,
-				    const u8 *value, const u8 len)
-{
-	return -1;
-}
-
-static inline int dccp_insert_option_ackvec(const struct sock *sk,
-					    const struct sk_buff *skb)
-{
-	return -1;
-}
+/**
+ * struct dccp_ackvec_parsed  -  Record offsets of Ack Vectors in skb
+ * @vec:	start of vector (offset into skb)
+ * @len:	length of @vec
+ * @nonce:	whether @vec had an ECN nonce of 0 or 1
+ * @node:	FIFO - arranged in descending order of ack_ackno
+ * This structure is used by CCIDs to access Ack Vectors in a received skb.
+ */
+struct dccp_ackvec_parsed {
+	u8		 *vec,
+			 len,
+			 nonce:1;
+	struct list_head node;
+};
 
-static inline int dccp_ackvec_pending(const struct dccp_ackvec *av)
-{
-	return 0;
-}
-#endif /* CONFIG_IP_DCCP_ACKVEC */
+extern int dccp_ackvec_parsed_add(struct list_head *head,
+				  u8 *vec, u8 len, u8 nonce);
+extern void dccp_ackvec_parsed_cleanup(struct list_head *parsed_chunks);
 #endif /* _ACKVEC_H */

net/dccp/ccid.c

@@ -13,6 +13,13 @@
 
 #include "ccid.h"
 
+static u8 builtin_ccids[] = {
+	DCCPC_CCID2,		/* CCID2 is supported by default */
+#if defined(CONFIG_IP_DCCP_CCID3) || defined(CONFIG_IP_DCCP_CCID3_MODULE)
+	DCCPC_CCID3,
+#endif
+};
+
 static struct ccid_operations *ccids[CCID_MAX];
 #if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT)
 static atomic_t ccids_lockct = ATOMIC_INIT(0);
@@ -86,6 +93,47 @@ static void ccid_kmem_cache_destroy(struct kmem_cache *slab)
 	}
 }
 
+/* check that up to @array_len members in @ccid_array are supported */
+bool ccid_support_check(u8 const *ccid_array, u8 array_len)
+{
+	u8 i, j, found;
+
+	for (i = 0, found = 0; i < array_len; i++, found = 0) {
+		for (j = 0; !found && j < ARRAY_SIZE(builtin_ccids); j++)
+			found = (ccid_array[i] == builtin_ccids[j]);
+		if (!found)
+			return false;
+	}
+	return true;
+}
+
+/**
+ * ccid_get_builtin_ccids  -  Provide copy of `builtin' CCID array
+ * @ccid_array: pointer to copy into
+ * @array_len: value to return length into
+ * This function allocates memory - caller must see that it is freed after use.
+ */
+int ccid_get_builtin_ccids(u8 **ccid_array, u8 *array_len)
+{
+	*ccid_array = kmemdup(builtin_ccids, sizeof(builtin_ccids), gfp_any());
+	if (*ccid_array == NULL)
+		return -ENOBUFS;
+	*array_len = ARRAY_SIZE(builtin_ccids);
+	return 0;
+}
+
+int ccid_getsockopt_builtin_ccids(struct sock *sk, int len,
+				    char __user *optval, int __user *optlen)
+{
+	if (len < sizeof(builtin_ccids))
+		return -EINVAL;
+
+	if (put_user(sizeof(builtin_ccids), optlen) ||
+	    copy_to_user(optval, builtin_ccids, sizeof(builtin_ccids)))
+		return -EFAULT;
+	return 0;
+}
+
 int ccid_register(struct ccid_operations *ccid_ops)
 {
 	int err = -ENOBUFS;
@@ -148,22 +196,41 @@ int ccid_unregister(struct ccid_operations *ccid_ops)
 
 EXPORT_SYMBOL_GPL(ccid_unregister);
 
+/**
+ * ccid_request_module  -  Pre-load CCID module for later use
+ * This should be called only from process context (e.g. during connection
+ * setup) and is necessary for later calls to ccid_new (typically in software
+ * interrupt), so that it has the modules available when they are needed.
+ */
+static int ccid_request_module(u8 id)
+{
+	if (!in_atomic()) {
+		ccids_read_lock();
+		if (ccids[id] == NULL) {
+			ccids_read_unlock();
+			return request_module("net-dccp-ccid-%d", id);
+		}
+		ccids_read_unlock();
+	}
+	return 0;
+}
+
+int ccid_request_modules(u8 const *ccid_array, u8 array_len)
+{
+#ifdef CONFIG_KMOD
+	while (array_len--)
+		if (ccid_request_module(ccid_array[array_len]))
+			return -1;
+#endif
+	return 0;
+}
+
 struct ccid *ccid_new(unsigned char id, struct sock *sk, int rx, gfp_t gfp)
 {
 	struct ccid_operations *ccid_ops;
 	struct ccid *ccid = NULL;
 
 	ccids_read_lock();
-#ifdef CONFIG_KMOD
-	if (ccids[id] == NULL) {
-		/* We only try to load if in process context */
-		ccids_read_unlock();
-		if (gfp & GFP_ATOMIC)
-			goto out;
-		request_module("net-dccp-ccid-%d", id);
-		ccids_read_lock();
-	}
-#endif
 	ccid_ops = ccids[id];
 	if (ccid_ops == NULL)
 		goto out_unlock;
@@ -205,20 +272,6 @@ out_module_put:
 
 EXPORT_SYMBOL_GPL(ccid_new);
 
-struct ccid *ccid_hc_rx_new(unsigned char id, struct sock *sk, gfp_t gfp)
-{
-	return ccid_new(id, sk, 1, gfp);
-}
-
-EXPORT_SYMBOL_GPL(ccid_hc_rx_new);
-
-struct ccid *ccid_hc_tx_new(unsigned char id,struct sock *sk,  gfp_t gfp)
-{
-	return ccid_new(id, sk, 0, gfp);
-}
-
-EXPORT_SYMBOL_GPL(ccid_hc_tx_new);
-
 static void ccid_delete(struct ccid *ccid, struct sock *sk, int rx)
 {
 	struct ccid_operations *ccid_ops;

net/dccp/ccid.h

@@ -60,22 +60,18 @@ 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,
-						    unsigned char option,
-						    unsigned char len, u16 idx,
-						    unsigned char* value);
+	int		(*ccid_hc_rx_parse_options)(struct sock *sk, u8 pkt,
+						    u8 opt, u8 *val, u8 len);
 	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,
-						    unsigned char option,
-						    unsigned char len, u16 idx,
-						    unsigned char* value);
+	int		(*ccid_hc_tx_parse_options)(struct sock *sk, u8 pkt,
+						    u8 opt, u8 *val, u8 len);
 	int		(*ccid_hc_tx_send_packet)(struct sock *sk,
 						  struct sk_buff *skb);
 	void		(*ccid_hc_tx_packet_sent)(struct sock *sk,
-						  int more, unsigned int len);
+						  unsigned int len);
 	void		(*ccid_hc_rx_get_info)(struct sock *sk,
 					       struct tcp_info *info);
 	void		(*ccid_hc_tx_get_info)(struct sock *sk,
@@ -103,31 +99,78 @@ static inline void *ccid_priv(const struct ccid *ccid)
 	return (void *)ccid->ccid_priv;
 }
 
+extern bool ccid_support_check(u8 const *ccid_array, u8 array_len);
+extern int  ccid_get_builtin_ccids(u8 **ccid_array, u8 *array_len);
+extern int  ccid_getsockopt_builtin_ccids(struct sock *sk, int len,
+					  char __user *, int __user *);
+
+extern int    ccid_request_modules(u8 const *ccid_array, u8 array_len);
 extern struct ccid *ccid_new(unsigned char id, struct sock *sk, int rx,
 			     gfp_t gfp);
 
-extern struct ccid *ccid_hc_rx_new(unsigned char id, struct sock *sk,
-				   gfp_t gfp);
-extern struct ccid *ccid_hc_tx_new(unsigned char id, struct sock *sk,
-				   gfp_t gfp);
+static inline int ccid_get_current_rx_ccid(struct dccp_sock *dp)
+{
+	struct ccid *ccid = dp->dccps_hc_rx_ccid;
+
+	if (ccid == NULL || ccid->ccid_ops == NULL)
+		return -1;
+	return ccid->ccid_ops->ccid_id;
+}
+
+static inline int ccid_get_current_tx_ccid(struct dccp_sock *dp)
+{
+	struct ccid *ccid = dp->dccps_hc_tx_ccid;
+
+	if (ccid == NULL || ccid->ccid_ops == NULL)
+		return -1;
+	return ccid->ccid_ops->ccid_id;
+}
 
 extern void ccid_hc_rx_delete(struct ccid *ccid, struct sock *sk);
 extern void ccid_hc_tx_delete(struct ccid *ccid, struct sock *sk);
 
+/*
+ * Congestion control of queued data packets via CCID decision.
+ *
+ * The TX CCID performs its congestion-control by indicating whether and when a
+ * queued packet may be sent, using the return code of ccid_hc_tx_send_packet().
+ * The following modes are supported via the symbolic constants below:
+ * - timer-based pacing    (CCID returns a delay value in milliseconds);
+ * - autonomous dequeueing (CCID internally schedules dccps_xmitlet).
+ */
+
+enum ccid_dequeueing_decision {
+	CCID_PACKET_SEND_AT_ONCE =	 0x00000,  /* "green light": no delay */
+	CCID_PACKET_DELAY_MAX =		 0x0FFFF,  /* maximum delay in msecs  */
+	CCID_PACKET_DELAY =		 0x10000,  /* CCID msec-delay mode */
+	CCID_PACKET_WILL_DEQUEUE_LATER = 0x20000,  /* CCID autonomous mode */
+	CCID_PACKET_ERR =		 0xF0000,  /* error condition */
+};
+
+static inline int ccid_packet_dequeue_eval(const int return_code)
+{
+	if (return_code < 0)
+		return CCID_PACKET_ERR;
+	if (return_code == 0)
+		return CCID_PACKET_SEND_AT_ONCE;
+	if (return_code <= CCID_PACKET_DELAY_MAX)
+		return CCID_PACKET_DELAY;
+	return return_code;
+}
+
 static inline int ccid_hc_tx_send_packet(struct ccid *ccid, struct sock *sk,
 					 struct sk_buff *skb)
 {
-	int rc = 0;
 	if (ccid->ccid_ops->ccid_hc_tx_send_packet != NULL)
-		rc = ccid->ccid_ops->ccid_hc_tx_send_packet(sk, skb);
-	return rc;
+		return ccid->ccid_ops->ccid_hc_tx_send_packet(sk, skb);
+	return CCID_PACKET_SEND_AT_ONCE;
 }
 
 static inline void ccid_hc_tx_packet_sent(struct ccid *ccid, struct sock *sk,
-					  int more, unsigned int len)
+					  unsigned int len)
 {
 	if (ccid->ccid_ops->ccid_hc_tx_packet_sent != NULL)
-		ccid->ccid_ops->ccid_hc_tx_packet_sent(sk, more, len);
+		ccid->ccid_ops->ccid_hc_tx_packet_sent(sk, len);
 }
 
 static inline void ccid_hc_rx_packet_recv(struct ccid *ccid, struct sock *sk,
@@ -144,27 +187,31 @@ 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,
-					   unsigned char len, u16 idx,
-					   unsigned char* value)
+					   u8 pkt, u8 opt, u8 *val, u8 len)
 {
-	int rc = 0;
-	if (ccid->ccid_ops->ccid_hc_tx_parse_options != NULL)
-		rc = ccid->ccid_ops->ccid_hc_tx_parse_options(sk, option, len, idx,
-						    value);
-	return rc;
+	if (ccid->ccid_ops->ccid_hc_tx_parse_options == NULL)
+		return 0;
+	return ccid->ccid_ops->ccid_hc_tx_parse_options(sk, pkt, opt, val, len);
 }
 
+/**
+ * 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,
-					   unsigned char len, u16 idx,
-					   unsigned char* value)
+					   u8 pkt, u8 opt, u8 *val, u8 len)
 {
-	int rc = 0;
-	if (ccid->ccid_ops->ccid_hc_rx_parse_options != NULL)
-		rc = ccid->ccid_ops->ccid_hc_rx_parse_options(sk, option, len, idx, value);
-	return rc;
+	if (ccid->ccid_ops->ccid_hc_rx_parse_options == NULL)
+		return 0;
+	return ccid->ccid_ops->ccid_hc_rx_parse_options(sk, pkt, opt, val, len);
 }
 
 static inline int ccid_hc_rx_insert_options(struct ccid *ccid, struct sock *sk,

net/dccp/ccids/Kconfig

@@ -1,10 +1,8 @@
 menu "DCCP CCIDs Configuration (EXPERIMENTAL)"
-	depends on EXPERIMENTAL
 
 config IP_DCCP_CCID2
-	tristate "CCID2 (TCP-Like) (EXPERIMENTAL)"
+	tristate "CCID2 (TCP-Like)"
 	def_tristate IP_DCCP
-	select IP_DCCP_ACKVEC
 	---help---
 	  CCID 2, TCP-like Congestion Control, denotes Additive Increase,
 	  Multiplicative Decrease (AIMD) congestion control with behavior
@@ -36,7 +34,7 @@ config IP_DCCP_CCID2_DEBUG
 	    If in doubt, say N.
 
 config IP_DCCP_CCID3
-	tristate "CCID3 (TCP-Friendly) (EXPERIMENTAL)"
+	tristate "CCID3 (TCP-Friendly)"
 	def_tristate IP_DCCP
 	select IP_DCCP_TFRC_LIB
 	---help---
@@ -64,9 +62,9 @@ config IP_DCCP_CCID3
 
 	  If in doubt, say M.
 
+if IP_DCCP_CCID3
 config IP_DCCP_CCID3_DEBUG
 	  bool "CCID3 debugging messages"
-	  depends on IP_DCCP_CCID3
 	  ---help---
 	    Enable CCID3-specific debugging messages.
 
@@ -76,10 +74,29 @@ config IP_DCCP_CCID3_DEBUG
 
 	    If in doubt, say N.
 
+choice
+	  prompt "Select method for measuring the packet size s"
+	  default IP_DCCP_CCID3_MEASURE_S_AS_MPS
+
+config IP_DCCP_CCID3_MEASURE_S_AS_MPS
+	  bool "Always use MPS in place of s"
+	  ---help---
+	    This use is recommended as it is consistent with the initialisation
+	    of X and suggested when s varies (rfc3448bis, (1) in section 4.1).
+config IP_DCCP_CCID3_MEASURE_S_AS_AVG
+	  bool "Use moving average"
+	  ---help---
+	    An alternative way of tracking s, also supported by rfc3448bis.
+	    This used to be the default for CCID-3 in previous kernels.
+config IP_DCCP_CCID3_MEASURE_S_AS_MAX
+	  bool "Track the maximum payload length"
+	  ---help---
+	    An experimental method based on tracking the maximum packet size.
+endchoice
+
 config IP_DCCP_CCID3_RTO
 	  int "Use higher bound for nofeedback timer"
 	  default 100
-	  depends on IP_DCCP_CCID3 && EXPERIMENTAL
 	  ---help---
 	    Use higher lower bound for nofeedback timer expiration.
 
@@ -106,6 +123,7 @@ config IP_DCCP_CCID3_RTO
 	    The purpose of the nofeedback timer is to slow DCCP down when there
 	    is serious network congestion: experimenting with larger values should
 	    therefore not be performed on WANs.
+endif	# IP_DCCP_CCID3
 
 config IP_DCCP_TFRC_LIB
 	tristate

net/dccp/ccids/ccid2.c

@@ -25,7 +25,7 @@
 /*
  * This implementation should follow RFC 4341
  */
-
+#include "../feat.h"
 #include "../ccid.h"
 #include "../dccp.h"
 #include "ccid2.h"
@@ -34,51 +34,8 @@
 #ifdef CONFIG_IP_DCCP_CCID2_DEBUG
 static int ccid2_debug;
 #define ccid2_pr_debug(format, a...)	DCCP_PR_DEBUG(ccid2_debug, format, ##a)
-
-static void ccid2_hc_tx_check_sanity(const struct ccid2_hc_tx_sock *hctx)
-{
-	int len = 0;
-	int pipe = 0;
-	struct ccid2_seq *seqp = hctx->ccid2hctx_seqh;
-
-	/* there is data in the chain */
-	if (seqp != hctx->ccid2hctx_seqt) {
-		seqp = seqp->ccid2s_prev;
-		len++;
-		if (!seqp->ccid2s_acked)
-			pipe++;
-
-		while (seqp != hctx->ccid2hctx_seqt) {
-			struct ccid2_seq *prev = seqp->ccid2s_prev;
-
-			len++;
-			if (!prev->ccid2s_acked)
-				pipe++;
-
-			/* packets are sent sequentially */
-			BUG_ON(dccp_delta_seqno(seqp->ccid2s_seq,
-						prev->ccid2s_seq ) >= 0);
-			BUG_ON(time_before(seqp->ccid2s_sent,
-					   prev->ccid2s_sent));
-
-			seqp = prev;
-		}
-	}
-
-	BUG_ON(pipe != hctx->ccid2hctx_pipe);
-	ccid2_pr_debug("len of chain=%d\n", len);
-
-	do {
-		seqp = seqp->ccid2s_prev;
-		len++;
-	} while (seqp != hctx->ccid2hctx_seqh);
-
-	ccid2_pr_debug("total len=%d\n", len);
-	BUG_ON(len != hctx->ccid2hctx_seqbufc * CCID2_SEQBUF_LEN);
-}
 #else
 #define ccid2_pr_debug(format, a...)
-#define ccid2_hc_tx_check_sanity(hctx)
 #endif
 
 static int ccid2_hc_tx_alloc_seq(struct ccid2_hc_tx_sock *hctx)
@@ -87,8 +44,7 @@ static int ccid2_hc_tx_alloc_seq(struct ccid2_hc_tx_sock *hctx)
 	int i;
 
 	/* check if we have space to preserve the pointer to the buffer */
-	if (hctx->ccid2hctx_seqbufc >= (sizeof(hctx->ccid2hctx_seqbuf) /
-					sizeof(struct ccid2_seq*)))
+	if (hctx->seqbufc >= sizeof(hctx->seqbuf) / sizeof(struct ccid2_seq *))
 		return -ENOMEM;
 
 	/* allocate buffer and initialize linked list */
@@ -104,38 +60,35 @@ static int ccid2_hc_tx_alloc_seq(struct ccid2_hc_tx_sock *hctx)
 	seqp->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1];
 
 	/* This is the first allocation.  Initiate the head and tail.  */
-	if (hctx->ccid2hctx_seqbufc == 0)
-		hctx->ccid2hctx_seqh = hctx->ccid2hctx_seqt = seqp;
+	if (hctx->seqbufc == 0)
+		hctx->seqh = hctx->seqt = seqp;
 	else {
 		/* link the existing list with the one we just created */
-		hctx->ccid2hctx_seqh->ccid2s_next = seqp;
-		seqp->ccid2s_prev = hctx->ccid2hctx_seqh;
+		hctx->seqh->ccid2s_next = seqp;
+		seqp->ccid2s_prev = hctx->seqh;
 
-		hctx->ccid2hctx_seqt->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1];
-		seqp[CCID2_SEQBUF_LEN - 1].ccid2s_next = hctx->ccid2hctx_seqt;
+		hctx->seqt->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1];
+		seqp[CCID2_SEQBUF_LEN - 1].ccid2s_next = hctx->seqt;
 	}
 
 	/* store the original pointer to the buffer so we can free it */
-	hctx->ccid2hctx_seqbuf[hctx->ccid2hctx_seqbufc] = seqp;
-	hctx->ccid2hctx_seqbufc++;
+	hctx->seqbuf[hctx->seqbufc] = seqp;
+	hctx->seqbufc++;
 
 	return 0;
 }
 
 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);
-
-	if (hctx->ccid2hctx_pipe < hctx->ccid2hctx_cwnd)
-		return 0;
-
-	return 1; /* XXX CCID should dequeue when ready instead of polling */
+	if (ccid2_cwnd_network_limited(ccid2_hc_tx_sk(sk)))
+		return CCID_PACKET_WILL_DEQUEUE_LATER;
+	return CCID_PACKET_SEND_AT_ONCE;
 }
 
 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);
+	u32 max_ratio = DIV_ROUND_UP(ccid2_hc_tx_sk(sk)->cwnd, 2);
 
 	/*
 	 * Ensure that Ack Ratio does not exceed ceil(cwnd/2), which is (2) from
@@ -147,8 +100,8 @@ static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val)
 		DCCP_WARN("Limiting Ack Ratio (%u) to %u\n", val, max_ratio);
 		val = max_ratio;
 	}
-	if (val > 0xFFFF)		/* RFC 4340, 11.3 */
-		val = 0xFFFF;
+	if (val > DCCPF_ACK_RATIO_MAX)
+		val = DCCPF_ACK_RATIO_MAX;
 
 	if (val == dp->dccps_l_ack_ratio)
 		return;
@@ -157,99 +110,77 @@ static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val)
 	dp->dccps_l_ack_ratio = val;
 }
 
-static void ccid2_change_srtt(struct ccid2_hc_tx_sock *hctx, long val)
-{
-	ccid2_pr_debug("change SRTT to %ld\n", val);
-	hctx->ccid2hctx_srtt = val;
-}
-
-static void ccid2_start_rto_timer(struct sock *sk);
-
 static void ccid2_hc_tx_rto_expire(unsigned long data)
 {
 	struct sock *sk = (struct sock *)data;
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
-	long s;
+	const bool sender_was_blocked = ccid2_cwnd_network_limited(hctx);
 
 	bh_lock_sock(sk);
 	if (sock_owned_by_user(sk)) {
-		sk_reset_timer(sk, &hctx->ccid2hctx_rtotimer,
-			       jiffies + HZ / 5);
+		sk_reset_timer(sk, &hctx->rtotimer, jiffies + HZ / 5);
 		goto out;
 	}
 
 	ccid2_pr_debug("RTO_EXPIRE\n");
 
-	ccid2_hc_tx_check_sanity(hctx);
-
 	/* back-off timer */
-	hctx->ccid2hctx_rto <<= 1;
-
-	s = hctx->ccid2hctx_rto / HZ;
-	if (s > 60)
-		hctx->ccid2hctx_rto = 60 * HZ;
-
-	ccid2_start_rto_timer(sk);
+	hctx->rto <<= 1;
+	if (hctx->rto > DCCP_RTO_MAX)
+		hctx->rto = DCCP_RTO_MAX;
 
 	/* adjust pipe, cwnd etc */
-	hctx->ccid2hctx_ssthresh = hctx->ccid2hctx_cwnd / 2;
-	if (hctx->ccid2hctx_ssthresh < 2)
-		hctx->ccid2hctx_ssthresh = 2;
-	hctx->ccid2hctx_cwnd	 = 1;
-	hctx->ccid2hctx_pipe	 = 0;
+	hctx->ssthresh = hctx->cwnd / 2;
+	if (hctx->ssthresh < 2)
+		hctx->ssthresh = 2;
+	hctx->cwnd = 1;
+	hctx->pipe = 0;
 
 	/* clear state about stuff we sent */
-	hctx->ccid2hctx_seqt = hctx->ccid2hctx_seqh;
-	hctx->ccid2hctx_packets_acked = 0;
+	hctx->seqt = hctx->seqh;
+	hctx->packets_acked = 0;
 
 	/* clear ack ratio state. */
-	hctx->ccid2hctx_rpseq	 = 0;
-	hctx->ccid2hctx_rpdupack = -1;
+	hctx->rpseq    = 0;
+	hctx->rpdupack = -1;
 	ccid2_change_l_ack_ratio(sk, 1);
-	ccid2_hc_tx_check_sanity(hctx);
+
+	/* if we were blocked before, we may now send cwnd=1 packet */
+	if (sender_was_blocked)
+		tasklet_schedule(&dccp_sk(sk)->dccps_xmitlet);
+	/* restart backed-off timer */
+	sk_reset_timer(sk, &hctx->rtotimer, jiffies + hctx->rto);
 out:
 	bh_unlock_sock(sk);
 	sock_put(sk);
 }
 
-static void ccid2_start_rto_timer(struct sock *sk)
-{
-	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
-
-	ccid2_pr_debug("setting RTO timeout=%ld\n", hctx->ccid2hctx_rto);
-
-	BUG_ON(timer_pending(&hctx->ccid2hctx_rtotimer));
-	sk_reset_timer(sk, &hctx->ccid2hctx_rtotimer,
-		       jiffies + hctx->ccid2hctx_rto);
-}
-
-static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
+static void ccid2_hc_tx_packet_sent(struct sock *sk, 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;
 
-	hctx->ccid2hctx_pipe++;
+	hctx->pipe++;
 
-	hctx->ccid2hctx_seqh->ccid2s_seq   = dp->dccps_gss;
-	hctx->ccid2hctx_seqh->ccid2s_acked = 0;
-	hctx->ccid2hctx_seqh->ccid2s_sent  = jiffies;
+	hctx->seqh->ccid2s_seq   = dp->dccps_gss;
+	hctx->seqh->ccid2s_acked = 0;
+	hctx->seqh->ccid2s_sent  = jiffies;
 
-	next = hctx->ccid2hctx_seqh->ccid2s_next;
+	next = hctx->seqh->ccid2s_next;
 	/* check if we need to alloc more space */
-	if (next == hctx->ccid2hctx_seqt) {
+	if (next == hctx->seqt) {
 		if (ccid2_hc_tx_alloc_seq(hctx)) {
 			DCCP_CRIT("packet history - out of memory!");
 			/* FIXME: find a more graceful way to bail out */
 			return;
 		}
-		next = hctx->ccid2hctx_seqh->ccid2s_next;
-		BUG_ON(next == hctx->ccid2hctx_seqt);
+		next = hctx->seqh->ccid2s_next;
+		BUG_ON(next == hctx->seqt);
 	}
-	hctx->ccid2hctx_seqh = next;
+	hctx->seqh = next;
 
-	ccid2_pr_debug("cwnd=%d pipe=%d\n", hctx->ccid2hctx_cwnd,
-		       hctx->ccid2hctx_pipe);
+	ccid2_pr_debug("cwnd=%d pipe=%d\n", hctx->cwnd, hctx->pipe);
 
 	/*
 	 * FIXME: The code below is broken and the variables have been removed
@@ -272,12 +203,12 @@ static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
 	 */
 #if 0
 	/* Ack Ratio.  Need to maintain a concept of how many windows we sent */
-	hctx->ccid2hctx_arsent++;
+	hctx->arsent++;
 	/* We had an ack loss in this window... */
-	if (hctx->ccid2hctx_ackloss) {
-		if (hctx->ccid2hctx_arsent >= hctx->ccid2hctx_cwnd) {
-			hctx->ccid2hctx_arsent	= 0;
-			hctx->ccid2hctx_ackloss	= 0;
+	if (hctx->ackloss) {
+		if (hctx->arsent >= hctx->cwnd) {
+			hctx->arsent  = 0;
+			hctx->ackloss = 0;
 		}
 	} else {
 		/* No acks lost up to now... */
@@ -287,28 +218,28 @@ static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
 			int denom = dp->dccps_l_ack_ratio * dp->dccps_l_ack_ratio -
 				    dp->dccps_l_ack_ratio;
 
-			denom = hctx->ccid2hctx_cwnd * hctx->ccid2hctx_cwnd / denom;
+			denom = hctx->cwnd * hctx->cwnd / denom;
 
-			if (hctx->ccid2hctx_arsent >= denom) {
+			if (hctx->arsent >= denom) {
 				ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio - 1);
-				hctx->ccid2hctx_arsent = 0;
+				hctx->arsent = 0;
 			}
 		} else {
 			/* we can't increase ack ratio further [1] */
-			hctx->ccid2hctx_arsent = 0; /* or maybe set it to cwnd*/
+			hctx->arsent = 0; /* or maybe set it to cwnd*/
 		}
 	}
 #endif
 
 	/* setup RTO timer */
-	if (!timer_pending(&hctx->ccid2hctx_rtotimer))
-		ccid2_start_rto_timer(sk);
+	if (!timer_pending(&hctx->rtotimer))
+		sk_reset_timer(sk, &hctx->rtotimer, jiffies + hctx->rto);
 
 #ifdef CONFIG_IP_DCCP_CCID2_DEBUG
 	do {
-		struct ccid2_seq *seqp = hctx->ccid2hctx_seqt;
+		struct ccid2_seq *seqp = hctx->seqt;
 
-		while (seqp != hctx->ccid2hctx_seqh) {
+		while (seqp != hctx->seqh) {
 			ccid2_pr_debug("out seq=%llu acked=%d time=%lu\n",
 				       (unsigned long long)seqp->ccid2s_seq,
 				       seqp->ccid2s_acked, seqp->ccid2s_sent);
@@ -316,205 +247,158 @@ static void ccid2_hc_tx_packet_sent(struct sock *sk, int more, unsigned int len)
 		}
 	} while (0);
 	ccid2_pr_debug("=========\n");
-	ccid2_hc_tx_check_sanity(hctx);
 #endif
 }
 
-/* XXX Lame code duplication!
- * returns -1 if none was found.
- * else returns the next offset to use in the function call.
+/**
+ * ccid2_rtt_estimator - Sample RTT and compute RTO using RFC2988 algorithm
+ * This code is almost identical with TCP's tcp_rtt_estimator(), since
+ * - it has a higher sampling frequency (recommended by RFC 1323),
+ * - the RTO does not collapse into RTT due to RTTVAR going towards zero,
+ * - it is simple (cf. more complex proposals such as Eifel timer or research
+ *   which suggests that the gain should be set according to window size),
+ * - in tests it was found to work well with CCID2 [gerrit].
  */
-static int ccid2_ackvector(struct sock *sk, struct sk_buff *skb, int offset,
-			   unsigned char **vec, unsigned char *veclen)
+static void ccid2_rtt_estimator(struct sock *sk, const long mrtt)
 {
-	const struct dccp_hdr *dh = dccp_hdr(skb);
-	unsigned char *options = (unsigned char *)dh + dccp_hdr_len(skb);
-	unsigned char *opt_ptr;
-	const unsigned char *opt_end = (unsigned char *)dh +
-					(dh->dccph_doff * 4);
-	unsigned char opt, len;
-	unsigned char *value;
-
-	BUG_ON(offset < 0);
-	options += offset;
-	opt_ptr = options;
-	if (opt_ptr >= opt_end)
-		return -1;
-
-	while (opt_ptr != opt_end) {
-		opt   = *opt_ptr++;
-		len   = 0;
-		value = NULL;
-
-		/* Check if this isn't a single byte option */
-		if (opt > DCCPO_MAX_RESERVED) {
-			if (opt_ptr == opt_end)
-				goto out_invalid_option;
-
-			len = *opt_ptr++;
-			if (len < 3)
-				goto out_invalid_option;
+	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
+	long m = mrtt ? : 1;
+
+	if (hctx->srtt == 0) {
+		/* First measurement m */
+		hctx->srtt = m << 3;
+		hctx->mdev = m << 1;
+
+		hctx->mdev_max = max(TCP_RTO_MIN, hctx->mdev);
+		hctx->rttvar   = hctx->mdev_max;
+		hctx->rtt_seq  = dccp_sk(sk)->dccps_gss;
+	} else {
+		/* Update scaled SRTT as SRTT += 1/8 * (m - SRTT) */
+		m -= (hctx->srtt >> 3);
+		hctx->srtt += m;
+
+		/* Similarly, update scaled mdev with regard to |m| */
+		if (m < 0) {
+			m = -m;
+			m -= (hctx->mdev >> 2);
 			/*
-			 * Remove the type and len fields, leaving
-			 * just the value size
+			 * This neutralises RTO increase when RTT < SRTT - mdev
+			 * (see P. Sarolahti, A. Kuznetsov,"Congestion Control
+			 * in Linux TCP", USENIX 2002, pp. 49-62).
 			 */
-			len     -= 2;
-			value   = opt_ptr;
-			opt_ptr += len;
+			if (m > 0)
+				m >>= 3;
+		} else {
+			m -= (hctx->mdev >> 2);
+		}
+		hctx->mdev += m;
 
-			if (opt_ptr > opt_end)
-				goto out_invalid_option;
+		if (hctx->mdev > hctx->mdev_max) {
+			hctx->mdev_max = hctx->mdev;
+			if (hctx->mdev_max > hctx->rttvar)
+				hctx->rttvar = hctx->mdev_max;
 		}
 
-		switch (opt) {
-		case DCCPO_ACK_VECTOR_0:
-		case DCCPO_ACK_VECTOR_1:
-			*vec	= value;
-			*veclen = len;
-			return offset + (opt_ptr - options);
+		/*
+		 * Decay RTTVAR at most once per flight, exploiting that
+		 *  1) pipe <= cwnd <= Sequence_Window = W  (RFC 4340, 7.5.2)
+		 *  2) AWL = GSS-W+1 <= GAR <= GSS          (RFC 4340, 7.5.1)
+		 * GAR is a useful bound for FlightSize = pipe, AWL is probably
+		 * too low as it over-estimates pipe.
+		 */
+		if (after48(dccp_sk(sk)->dccps_gar, hctx->rtt_seq)) {
+			if (hctx->mdev_max < hctx->rttvar)
+				hctx->rttvar -= (hctx->rttvar -
+						 hctx->mdev_max) >> 2;
+			hctx->rtt_seq  = dccp_sk(sk)->dccps_gss;
+			hctx->mdev_max = TCP_RTO_MIN;
 		}
 	}
 
-	return -1;
-
-out_invalid_option:
-	DCCP_BUG("Invalid option - this should not happen (previous parsing)!");
-	return -1;
-}
-
-static void ccid2_hc_tx_kill_rto_timer(struct sock *sk)
-{
-	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
+	/*
+	 * Set RTO from SRTT and RTTVAR
+	 * Clock granularity is ignored since the minimum error for RTTVAR is
+	 * clamped to 50msec (corresponding to HZ=20). This leads to a minimum
+	 * RTO of 200msec. This agrees with TCP and RFC 4341, 5.: "Because DCCP
+	 * does not retransmit data, DCCP does not require TCP's recommended
+	 * minimum timeout of one second".
+	 */
+	hctx->rto = (hctx->srtt >> 3) + hctx->rttvar;
 
-	sk_stop_timer(sk, &hctx->ccid2hctx_rtotimer);
-	ccid2_pr_debug("deleted RTO timer\n");
+	if (hctx->rto > DCCP_RTO_MAX)
+		hctx->rto = DCCP_RTO_MAX;
 }
 
-static inline void ccid2_new_ack(struct sock *sk,
-				 struct ccid2_seq *seqp,
-				 unsigned int *maxincr)
+static void ccid2_new_ack(struct sock *sk, struct ccid2_seq *seqp,
+			  unsigned int *maxincr)
 {
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
 
-	if (hctx->ccid2hctx_cwnd < hctx->ccid2hctx_ssthresh) {
-		if (*maxincr > 0 && ++hctx->ccid2hctx_packets_acked == 2) {
-			hctx->ccid2hctx_cwnd += 1;
-			*maxincr	     -= 1;
-			hctx->ccid2hctx_packets_acked = 0;
+	if (hctx->cwnd < hctx->ssthresh) {
+		if (*maxincr > 0 && ++hctx->packets_acked == 2) {
+			hctx->cwnd += 1;
+			*maxincr   -= 1;
+			hctx->packets_acked = 0;
 		}
-	} else if (++hctx->ccid2hctx_packets_acked >= hctx->ccid2hctx_cwnd) {
-			hctx->ccid2hctx_cwnd += 1;
-			hctx->ccid2hctx_packets_acked = 0;
+	} else if (++hctx->packets_acked >= hctx->cwnd) {
+			hctx->cwnd += 1;
+			hctx->packets_acked = 0;
 	}
-
-	/* update RTO */
-	if (hctx->ccid2hctx_srtt == -1 ||
-	    time_after(jiffies, hctx->ccid2hctx_lastrtt + hctx->ccid2hctx_srtt)) {
-		unsigned long r = (long)jiffies - (long)seqp->ccid2s_sent;
-		int s;
-
-		/* first measurement */
-		if (hctx->ccid2hctx_srtt == -1) {
-			ccid2_pr_debug("R: %lu Time=%lu seq=%llu\n",
-				       r, jiffies,
-				       (unsigned long long)seqp->ccid2s_seq);
-			ccid2_change_srtt(hctx, r);
-			hctx->ccid2hctx_rttvar = r >> 1;
-		} else {
-			/* RTTVAR */
-			long tmp = hctx->ccid2hctx_srtt - r;
-			long srtt;
-
-			if (tmp < 0)
-				tmp *= -1;
-
-			tmp >>= 2;
-			hctx->ccid2hctx_rttvar *= 3;
-			hctx->ccid2hctx_rttvar >>= 2;
-			hctx->ccid2hctx_rttvar += tmp;
-
-			/* SRTT */
-			srtt = hctx->ccid2hctx_srtt;
-			srtt *= 7;
-			srtt >>= 3;
-			tmp = r >> 3;
-			srtt += tmp;
-			ccid2_change_srtt(hctx, srtt);
-		}
-		s = hctx->ccid2hctx_rttvar << 2;
-		/* clock granularity is 1 when based on jiffies */
-		if (!s)
-			s = 1;
-		hctx->ccid2hctx_rto = hctx->ccid2hctx_srtt + s;
-
-		/* must be at least a second */
-		s = hctx->ccid2hctx_rto / HZ;
-		/* DCCP doesn't require this [but I like it cuz my code sux] */
-#if 1
-		if (s < 1)
-			hctx->ccid2hctx_rto = HZ;
-#endif
-		/* max 60 seconds */
-		if (s > 60)
-			hctx->ccid2hctx_rto = HZ * 60;
-
-		hctx->ccid2hctx_lastrtt = jiffies;
-
-		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);
-	}
-
-	/* we got a new ack, so re-start RTO timer */
-	ccid2_hc_tx_kill_rto_timer(sk);
-	ccid2_start_rto_timer(sk);
-}
-
-static void ccid2_hc_tx_dec_pipe(struct sock *sk)
-{
-	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
-
-	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);
+	/*
+	 * FIXME: RTT is sampled several times per acknowledgment (for each
+	 * entry in the Ack Vector), instead of once per Ack (as in TCP SACK).
+	 * This causes the RTT to be over-estimated, since the older entries
+	 * in the Ack Vector have earlier sending times.
+	 * The cleanest solution is to not use the ccid2s_sent field at all
+	 * and instead use DCCP timestamps - need to be resolved at some time.
+	 */
+	ccid2_rtt_estimator(sk, jiffies - seqp->ccid2s_sent);
 }
 
 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)) {
+	if (time_before(seqp->ccid2s_sent, hctx->last_cong)) {
 		ccid2_pr_debug("Multiple losses in an RTT---treating as one\n");
 		return;
 	}
 
-	hctx->ccid2hctx_last_cong = jiffies;
+	hctx->last_cong = jiffies;
 
-	hctx->ccid2hctx_cwnd     = hctx->ccid2hctx_cwnd / 2 ? : 1U;
-	hctx->ccid2hctx_ssthresh = max(hctx->ccid2hctx_cwnd, 2U);
+	hctx->cwnd     = hctx->cwnd / 2 ? : 1U;
+	hctx->ssthresh = max(hctx->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);
+	if (dccp_sk(sk)->dccps_l_ack_ratio > hctx->cwnd)
+		ccid2_change_l_ack_ratio(sk, hctx->cwnd);
+}
+
+static int ccid2_hc_tx_parse_options(struct sock *sk, u8 packet_type,
+				     u8 option, u8 *optval, u8 optlen)
+{
+	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
+
+	switch (option) {
+	case DCCPO_ACK_VECTOR_0:
+	case DCCPO_ACK_VECTOR_1:
+		return dccp_ackvec_parsed_add(&hctx->av_chunks, optval, optlen,
+					      option - DCCPO_ACK_VECTOR_0);
+	}
+	return 0;
 }
 
 static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
+	const bool sender_was_blocked = ccid2_cwnd_network_limited(hctx);
+	struct dccp_ackvec_parsed *avp;
 	u64 ackno, seqno;
 	struct ccid2_seq *seqp;
-	unsigned char *vector;
-	unsigned char veclen;
-	int offset = 0;
 	int done = 0;
 	unsigned int maxincr = 0;
 
-	ccid2_hc_tx_check_sanity(hctx);
 	/* check reverse path congestion */
 	seqno = DCCP_SKB_CB(skb)->dccpd_seq;
 
@@ -523,21 +407,21 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	 * -sorbo.
 	 */
 	/* need to bootstrap */
-	if (hctx->ccid2hctx_rpdupack == -1) {
-		hctx->ccid2hctx_rpdupack = 0;
-		hctx->ccid2hctx_rpseq = seqno;
+	if (hctx->rpdupack == -1) {
+		hctx->rpdupack = 0;
+		hctx->rpseq = seqno;
 	} else {
 		/* check if packet is consecutive */
-		if (dccp_delta_seqno(hctx->ccid2hctx_rpseq, seqno) == 1)
-			hctx->ccid2hctx_rpseq = seqno;
+		if (dccp_delta_seqno(hctx->rpseq, seqno) == 1)
+			hctx->rpseq = seqno;
 		/* it's a later packet */
-		else if (after48(seqno, hctx->ccid2hctx_rpseq)) {
-			hctx->ccid2hctx_rpdupack++;
+		else if (after48(seqno, hctx->rpseq)) {
+			hctx->rpdupack++;
 
 			/* check if we got enough dupacks */
-			if (hctx->ccid2hctx_rpdupack >= NUMDUPACK) {
-				hctx->ccid2hctx_rpdupack = -1; /* XXX lame */
-				hctx->ccid2hctx_rpseq = 0;
+			if (hctx->rpdupack >= NUMDUPACK) {
+				hctx->rpdupack = -1; /* XXX lame */
+				hctx->rpseq = 0;
 
 				ccid2_change_l_ack_ratio(sk, 2 * dp->dccps_l_ack_ratio);
 			}
@@ -545,27 +429,22 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	}
 
 	/* check forward path congestion */
-	/* still didn't send out new data packets */
-	if (hctx->ccid2hctx_seqh == hctx->ccid2hctx_seqt)
+	if (dccp_packet_without_ack(skb))
 		return;
 
-	switch (DCCP_SKB_CB(skb)->dccpd_type) {
-	case DCCP_PKT_ACK:
-	case DCCP_PKT_DATAACK:
-		break;
-	default:
-		return;
-	}
+	/* still didn't send out new data packets */
+	if (hctx->seqh == hctx->seqt)
+		goto done;
 
 	ackno = DCCP_SKB_CB(skb)->dccpd_ack_seq;
-	if (after48(ackno, hctx->ccid2hctx_high_ack))
-		hctx->ccid2hctx_high_ack = ackno;
+	if (after48(ackno, hctx->high_ack))
+		hctx->high_ack = ackno;
 
-	seqp = hctx->ccid2hctx_seqt;
+	seqp = hctx->seqt;
 	while (before48(seqp->ccid2s_seq, ackno)) {
 		seqp = seqp->ccid2s_next;
-		if (seqp == hctx->ccid2hctx_seqh) {
-			seqp = hctx->ccid2hctx_seqh->ccid2s_prev;
+		if (seqp == hctx->seqh) {
+			seqp = hctx->seqh->ccid2s_prev;
 			break;
 		}
 	}
@@ -575,26 +454,26 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	 * packets per acknowledgement. Rounding up avoids that cwnd is not
 	 * advanced when Ack Ratio is 1 and gives a slight edge otherwise.
 	 */
-	if (hctx->ccid2hctx_cwnd < hctx->ccid2hctx_ssthresh)
+	if (hctx->cwnd < hctx->ssthresh)
 		maxincr = DIV_ROUND_UP(dp->dccps_l_ack_ratio, 2);
 
 	/* go through all ack vectors */
-	while ((offset = ccid2_ackvector(sk, skb, offset,
-					 &vector, &veclen)) != -1) {
+	list_for_each_entry(avp, &hctx->av_chunks, node) {
 		/* go through this ack vector */
-		while (veclen--) {
-			const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
-			u64 ackno_end_rl = SUB48(ackno, rl);
+		for (; avp->len--; avp->vec++) {
+			u64 ackno_end_rl = SUB48(ackno,
+						 dccp_ackvec_runlen(avp->vec));
 
-			ccid2_pr_debug("ackvec start:%llu end:%llu\n",
+			ccid2_pr_debug("ackvec %llu |%u,%u|\n",
 				       (unsigned long long)ackno,
-				       (unsigned long long)ackno_end_rl);
+				       dccp_ackvec_state(avp->vec) >> 6,
+				       dccp_ackvec_runlen(avp->vec));
 			/* if the seqno we are analyzing is larger than the
 			 * current ackno, then move towards the tail of our
 			 * seqnos.
 			 */
 			while (after48(seqp->ccid2s_seq, ackno)) {
-				if (seqp == hctx->ccid2hctx_seqt) {
+				if (seqp == hctx->seqt) {
 					done = 1;
 					break;
 				}
@@ -607,26 +486,24 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 			 * run length
 			 */
 			while (between48(seqp->ccid2s_seq,ackno_end_rl,ackno)) {
-				const u8 state = *vector &
-						 DCCP_ACKVEC_STATE_MASK;
+				const u8 state = dccp_ackvec_state(avp->vec);
 
 				/* new packet received or marked */
-				if (state != DCCP_ACKVEC_STATE_NOT_RECEIVED &&
+				if (state != DCCPAV_NOT_RECEIVED &&
 				    !seqp->ccid2s_acked) {
-					if (state ==
-					    DCCP_ACKVEC_STATE_ECN_MARKED) {
+					if (state == DCCPAV_ECN_MARKED)
 						ccid2_congestion_event(sk,
 								       seqp);
-					} else
+					else
 						ccid2_new_ack(sk, seqp,
 							      &maxincr);
 
 					seqp->ccid2s_acked = 1;
 					ccid2_pr_debug("Got ack for %llu\n",
 						       (unsigned long long)seqp->ccid2s_seq);
-					ccid2_hc_tx_dec_pipe(sk);
+					hctx->pipe--;
 				}
-				if (seqp == hctx->ccid2hctx_seqt) {
+				if (seqp == hctx->seqt) {
 					done = 1;
 					break;
 				}
@@ -636,7 +513,6 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 				break;
 
 			ackno = SUB48(ackno_end_rl, 1);
-			vector++;
 		}
 		if (done)
 			break;
@@ -645,11 +521,11 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	/* The state about what is acked should be correct now
 	 * Check for NUMDUPACK
 	 */
-	seqp = hctx->ccid2hctx_seqt;
-	while (before48(seqp->ccid2s_seq, hctx->ccid2hctx_high_ack)) {
+	seqp = hctx->seqt;
+	while (before48(seqp->ccid2s_seq, hctx->high_ack)) {
 		seqp = seqp->ccid2s_next;
-		if (seqp == hctx->ccid2hctx_seqh) {
-			seqp = hctx->ccid2hctx_seqh->ccid2s_prev;
+		if (seqp == hctx->seqh) {
+			seqp = hctx->seqh->ccid2s_prev;
 			break;
 		}
 	}
@@ -660,7 +536,7 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 			if (done == NUMDUPACK)
 				break;
 		}
-		if (seqp == hctx->ccid2hctx_seqt)
+		if (seqp == hctx->seqt)
 			break;
 		seqp = seqp->ccid2s_prev;
 	}
@@ -681,25 +557,34 @@ static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
 				 * one ack vector.
 				 */
 				ccid2_congestion_event(sk, seqp);
-				ccid2_hc_tx_dec_pipe(sk);
+				hctx->pipe--;
 			}
-			if (seqp == hctx->ccid2hctx_seqt)
+			if (seqp == hctx->seqt)
 				break;
 			seqp = seqp->ccid2s_prev;
 		}
 
-		hctx->ccid2hctx_seqt = last_acked;
+		hctx->seqt = last_acked;
 	}
 
 	/* trim acked packets in tail */
-	while (hctx->ccid2hctx_seqt != hctx->ccid2hctx_seqh) {
-		if (!hctx->ccid2hctx_seqt->ccid2s_acked)
+	while (hctx->seqt != hctx->seqh) {
+		if (!hctx->seqt->ccid2s_acked)
 			break;
 
-		hctx->ccid2hctx_seqt = hctx->ccid2hctx_seqt->ccid2s_next;
+		hctx->seqt = hctx->seqt->ccid2s_next;
 	}
 
-	ccid2_hc_tx_check_sanity(hctx);
+	/* restart RTO timer if not all outstanding data has been acked */
+	if (hctx->pipe == 0)
+		sk_stop_timer(sk, &hctx->rtotimer);
+	else
+		sk_reset_timer(sk, &hctx->rtotimer, jiffies + hctx->rto);
+done:
+	/* check if incoming Acks allow pending packets to be sent */
+	if (sender_was_blocked && !ccid2_cwnd_network_limited(hctx))
+		tasklet_schedule(&dccp_sk(sk)->dccps_xmitlet);
+	dccp_ackvec_parsed_cleanup(&hctx->av_chunks);
 }
 
 static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
@@ -709,17 +594,13 @@ static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
 	u32 max_ratio;
 
 	/* RFC 4341, 5: initialise ssthresh to arbitrarily high (max) value */
-	hctx->ccid2hctx_ssthresh  = ~0U;
+	hctx->ssthresh = ~0U;
 
-	/*
-	 * 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_cwnd = clamp(4380U / dp->dccps_mss_cache, 2U, 4U);
+	/* Use larger initial windows (RFC 3390, rfc2581bis) */
+	hctx->cwnd = rfc3390_bytes_to_packets(dp->dccps_mss_cache);
 
 	/* Make sure that Ack Ratio is enabled and within bounds. */
-	max_ratio = DIV_ROUND_UP(hctx->ccid2hctx_cwnd, 2);
+	max_ratio = DIV_ROUND_UP(hctx->cwnd, 2);
 	if (dp->dccps_l_ack_ratio == 0 || dp->dccps_l_ack_ratio > max_ratio)
 		dp->dccps_l_ack_ratio = max_ratio;
 
@@ -727,15 +608,11 @@ static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
 	if (ccid2_hc_tx_alloc_seq(hctx))
 		return -ENOMEM;
 
-	hctx->ccid2hctx_rto	 = 3 * HZ;
-	ccid2_change_srtt(hctx, -1);
-	hctx->ccid2hctx_rttvar	 = -1;
-	hctx->ccid2hctx_rpdupack = -1;
-	hctx->ccid2hctx_last_cong = jiffies;
-	setup_timer(&hctx->ccid2hctx_rtotimer, ccid2_hc_tx_rto_expire,
-			(unsigned long)sk);
-
-	ccid2_hc_tx_check_sanity(hctx);
+	hctx->rto	= DCCP_TIMEOUT_INIT;
+	hctx->rpdupack  = -1;
+	hctx->last_cong = jiffies;
+	setup_timer(&hctx->rtotimer, ccid2_hc_tx_rto_expire, (unsigned long)sk);
+	INIT_LIST_HEAD(&hctx->av_chunks);
 	return 0;
 }
 
@@ -744,11 +621,11 @@ static void ccid2_hc_tx_exit(struct sock *sk)
 	struct ccid2_hc_tx_sock *hctx = ccid2_hc_tx_sk(sk);
 	int i;
 
-	ccid2_hc_tx_kill_rto_timer(sk);
+	sk_stop_timer(sk, &hctx->rtotimer);
 
-	for (i = 0; i < hctx->ccid2hctx_seqbufc; i++)
-		kfree(hctx->ccid2hctx_seqbuf[i]);
-	hctx->ccid2hctx_seqbufc = 0;
+	for (i = 0; i < hctx->seqbufc; i++)
+		kfree(hctx->seqbuf[i]);
+	hctx->seqbufc = 0;
 }
 
 static void ccid2_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
@@ -759,27 +636,28 @@ static void ccid2_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
 	switch (DCCP_SKB_CB(skb)->dccpd_type) {
 	case DCCP_PKT_DATA:
 	case DCCP_PKT_DATAACK:
-		hcrx->ccid2hcrx_data++;
-		if (hcrx->ccid2hcrx_data >= dp->dccps_r_ack_ratio) {
+		hcrx->data++;
+		if (hcrx->data >= dp->dccps_r_ack_ratio) {
 			dccp_send_ack(sk);
-			hcrx->ccid2hcrx_data = 0;
+			hcrx->data = 0;
 		}
 		break;
 	}
 }
 
 static struct ccid_operations ccid2 = {
-	.ccid_id		= DCCPC_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,
-	.ccid_hc_tx_exit	= ccid2_hc_tx_exit,
-	.ccid_hc_tx_send_packet	= ccid2_hc_tx_send_packet,
-	.ccid_hc_tx_packet_sent	= ccid2_hc_tx_packet_sent,
-	.ccid_hc_tx_packet_recv	= ccid2_hc_tx_packet_recv,
-	.ccid_hc_rx_obj_size	= sizeof(struct ccid2_hc_rx_sock),
-	.ccid_hc_rx_packet_recv	= ccid2_hc_rx_packet_recv,
+	.ccid_id		  = DCCPC_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,
+	.ccid_hc_tx_exit	  = ccid2_hc_tx_exit,
+	.ccid_hc_tx_send_packet	  = ccid2_hc_tx_send_packet,
+	.ccid_hc_tx_packet_sent	  = ccid2_hc_tx_packet_sent,
+	.ccid_hc_tx_parse_options = ccid2_hc_tx_parse_options,
+	.ccid_hc_tx_packet_recv	  = ccid2_hc_tx_packet_recv,
+	.ccid_hc_rx_obj_size	  = sizeof(struct ccid2_hc_rx_sock),
+	.ccid_hc_rx_packet_recv	  = ccid2_hc_rx_packet_recv,
 };
 
 #ifdef CONFIG_IP_DCCP_CCID2_DEBUG

net/dccp/ccids/ccid2.h

@@ -42,34 +42,49 @@ struct ccid2_seq {
 
 /** struct ccid2_hc_tx_sock - CCID2 TX half connection
  *
- * @ccid2hctx_{cwnd,ssthresh,pipe}: as per RFC 4341, section 5
- * @ccid2hctx_packets_acked - Ack counter for deriving cwnd growth (RFC 3465)
- * @ccid2hctx_lastrtt -time RTT was last measured
- * @ccid2hctx_rpseq - last consecutive seqno
- * @ccid2hctx_rpdupack - dupacks since rpseq
-*/
+ * @{cwnd,ssthresh,pipe}: as per RFC 4341, section 5
+ * @packets_acked: Ack counter for deriving cwnd growth (RFC 3465)
+ * @srtt: smoothed RTT estimate, scaled by 2^3
+ * @mdev: smoothed RTT variation, scaled by 2^2
+ * @mdev_max: maximum of @mdev during one flight
+ * @rttvar: moving average/maximum of @mdev_max
+ * @rto: RTO value deriving from SRTT and RTTVAR (RFC 2988)
+ * @rtt_seq: to decay RTTVAR at most once per flight
+ * @rpseq: last consecutive seqno
+ * @rpdupack: dupacks since rpseq
+ * @av_chunks: list of Ack Vectors received on current skb
+ */
 struct ccid2_hc_tx_sock {
-	u32			ccid2hctx_cwnd;
-	u32			ccid2hctx_ssthresh;
-	u32			ccid2hctx_pipe;
-	u32			ccid2hctx_packets_acked;
-	struct ccid2_seq	*ccid2hctx_seqbuf[CCID2_SEQBUF_MAX];
-	int			ccid2hctx_seqbufc;
-	struct ccid2_seq	*ccid2hctx_seqh;
-	struct ccid2_seq	*ccid2hctx_seqt;
-	long			ccid2hctx_rto;
-	long			ccid2hctx_srtt;
-	long			ccid2hctx_rttvar;
-	unsigned long		ccid2hctx_lastrtt;
-	struct timer_list	ccid2hctx_rtotimer;
-	u64			ccid2hctx_rpseq;
-	int			ccid2hctx_rpdupack;
-	unsigned long		ccid2hctx_last_cong;
-	u64			ccid2hctx_high_ack;
+	u32			cwnd;
+	u32			ssthresh;
+	u32			pipe;
+	u32			packets_acked;
+	struct ccid2_seq	*seqbuf[CCID2_SEQBUF_MAX];
+	int			seqbufc;
+	struct ccid2_seq	*seqh;
+	struct ccid2_seq	*seqt;
+	/* RTT measurement: variables/principles are the same as in TCP */
+	u32			srtt,
+				mdev,
+				mdev_max,
+				rttvar,
+				rto;
+	u64			rtt_seq:48;
+	struct timer_list	rtotimer;
+	u64			rpseq;
+	int			rpdupack;
+	unsigned long		last_cong;
+	u64			high_ack;
+	struct list_head	av_chunks;
 };
 
+static inline bool ccid2_cwnd_network_limited(struct ccid2_hc_tx_sock *hctx)
+{
+	return (hctx->pipe >= hctx->cwnd);
+}
+
 struct ccid2_hc_rx_sock {
-	int	ccid2hcrx_data;
+	int			data;
 };
 
 static inline struct ccid2_hc_tx_sock *ccid2_hc_tx_sk(const struct sock *sk)

net/dccp/ccids/ccid3.h

@@ -47,11 +47,22 @@
 /* Two seconds as per RFC 3448 4.2 */
 #define TFRC_INITIAL_TIMEOUT	   (2 * USEC_PER_SEC)
 
-/* In usecs - half the scheduling granularity as per RFC3448 4.6 */
-#define TFRC_OPSYS_HALF_TIME_GRAN  (USEC_PER_SEC / (2 * HZ))
+/* Maximum backoff interval t_mbi (RFC 3448, 4.3) */
+#define TFRC_T_MBI		   (64 * USEC_PER_SEC)
 
-/* Parameter t_mbi from [RFC 3448, 4.3]: backoff interval in seconds */
-#define TFRC_T_MBI		   64
+/*
+ * The t_delta parameter (RFC 3448, 4.6): delays of less than %USEC_PER_MSEC are
+ * rounded down to 0, since sk_reset_timer() here uses millisecond granularity.
+ * Hence we can use a constant t_delta = %USEC_PER_MSEC when HZ >= 500. A coarse
+ * resolution of HZ < 500 means that the error is below one timer tick (t_gran)
+ * when using the constant t_delta  =  t_gran / 2  =  %USEC_PER_SEC / (2 * HZ).
+ */
+#if (HZ >= 500)
+# define TFRC_T_DELTA		   USEC_PER_MSEC
+#else
+# define TFRC_T_DELTA		   (USEC_PER_SEC / (2 * HZ))
+#warning Coarse CONFIG_HZ resolution -- higher value recommended for TFRC.
+#endif
 
 enum ccid3_options {
 	TFRC_OPT_LOSS_EVENT_RATE = 192,
@@ -59,62 +70,43 @@ enum ccid3_options {
 	TFRC_OPT_RECEIVE_RATE	 = 194,
 };
 
-struct ccid3_options_received {
-	u64 ccid3or_seqno:48,
-	    ccid3or_loss_intervals_idx:16;
-	u16 ccid3or_loss_intervals_len;
-	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,
-};
-
 /** struct ccid3_hc_tx_sock - CCID3 sender half-connection socket
  *
- * @ccid3hctx_x - Current sending rate in 64 * bytes per second
- * @ccid3hctx_x_recv - Receive rate    in 64 * bytes per second
- * @ccid3hctx_x_calc - Calculated rate in bytes per second
- * @ccid3hctx_rtt - Estimate of current round trip time in usecs
- * @ccid3hctx_p - Current loss event rate (0-1) scaled by 1000000
- * @ccid3hctx_s - Packet size in bytes
- * @ccid3hctx_t_rto - Nofeedback Timer setting in usecs
- * @ccid3hctx_t_ipi - Interpacket (send) interval (RFC 3448, 4.6) in usecs
- * @ccid3hctx_state - Sender state, one of %ccid3_hc_tx_states
- * @ccid3hctx_last_win_count - Last window counter sent
- * @ccid3hctx_t_last_win_count - Timestamp of earliest packet
- *				 with last_win_count value sent
- * @ccid3hctx_no_feedback_timer - Handle to no feedback timer
- * @ccid3hctx_t_ld - Time last doubled during slow start
- * @ccid3hctx_t_nom - Nominal send time of next packet
- * @ccid3hctx_delta - Send timer delta (RFC 3448, 4.6) in usecs
- * @ccid3hctx_hist - Packet history
- * @ccid3hctx_options_received - Parsed set of retrieved options
+ * @x - Current sending rate in 64 * bytes per second
+ * @x_recv - Receive rate    in 64 * bytes per second
+ * @x_calc - Calculated rate in bytes per second
+ * @rtt - Estimate of current round trip time in usecs
+ * @r_sqmean - Estimate of long-term RTT (RFC 3448, 4.5)
+ * @p - Current loss event rate (0-1) scaled by 1000000
+ * @s - Packet size in bytes
+ * @t_rto - Nofeedback Timer setting in usecs
+ * @t_ipi - Interpacket (send) interval (RFC 3448, 4.6) in usecs
+ * @feedback - Whether feedback has been received or not
+ * @last_win_count - Last window counter sent
+ * @t_last_win_count - Timestamp of earliest packet with
+ *                     last_win_count value sent
+ * @no_feedback_timer - Handle to no feedback timer
+ * @t_ld - Time last doubled during slow start
+ * @t_nom - Nominal send time of next packet
+ * @hist - Packet history
  */
 struct ccid3_hc_tx_sock {
-	struct tfrc_tx_info		ccid3hctx_tfrc;
-#define ccid3hctx_x			ccid3hctx_tfrc.tfrctx_x
-#define ccid3hctx_x_recv		ccid3hctx_tfrc.tfrctx_x_recv
-#define ccid3hctx_x_calc		ccid3hctx_tfrc.tfrctx_x_calc
-#define ccid3hctx_rtt			ccid3hctx_tfrc.tfrctx_rtt
-#define ccid3hctx_p			ccid3hctx_tfrc.tfrctx_p
-#define ccid3hctx_t_rto			ccid3hctx_tfrc.tfrctx_rto
-#define ccid3hctx_t_ipi			ccid3hctx_tfrc.tfrctx_ipi
-	u16				ccid3hctx_s;
-	enum ccid3_hc_tx_states		ccid3hctx_state:8;
-	u8				ccid3hctx_last_win_count;
-	ktime_t				ccid3hctx_t_last_win_count;
-	struct timer_list		ccid3hctx_no_feedback_timer;
-	ktime_t				ccid3hctx_t_ld;
-	ktime_t				ccid3hctx_t_nom;
-	u32				ccid3hctx_delta;
-	struct tfrc_tx_hist_entry	*ccid3hctx_hist;
-	struct ccid3_options_received	ccid3hctx_options_received;
+	u64				x;
+	u64				x_recv;
+	u32				x_calc;
+	u32				rtt;
+	u16				r_sqmean;
+	u32				p;
+	u32				t_rto;
+	u32				t_ipi;
+	u16				s;
+	bool				feedback:1;
+	u8				last_win_count;
+	ktime_t				t_last_win_count;
+	struct timer_list		no_feedback_timer;
+	ktime_t				t_ld;
+	ktime_t				t_nom;
+	struct tfrc_tx_hist_entry	*hist;
 };
 
 static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk)
@@ -124,41 +116,32 @@ static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk)
     return hctx;
 }
 
-/* TFRC receiver states */
-enum ccid3_hc_rx_states {
-	TFRC_RSTATE_NO_DATA = 1,
-	TFRC_RSTATE_DATA,
-	TFRC_RSTATE_TERM    = 127,
+
+enum ccid3_fback_type {
+	CCID3_FBACK_NONE = 0,
+	CCID3_FBACK_INITIAL,
+	CCID3_FBACK_PERIODIC,
+	CCID3_FBACK_PARAM_CHANGE
 };
 
 /** struct ccid3_hc_rx_sock - CCID3 receiver half-connection socket
  *
- *  @ccid3hcrx_x_recv  -  Receiver estimate of send rate (RFC 3448 4.3)
- *  @ccid3hcrx_rtt  -  Receiver estimate of rtt (non-standard)
- *  @ccid3hcrx_p  -  Current loss event rate (RFC 3448 5.4)
- *  @ccid3hcrx_last_counter  -  Tracks window counter (RFC 4342, 8.1)
- *  @ccid3hcrx_state  -  Receiver state, one of %ccid3_hc_rx_states
- *  @ccid3hcrx_bytes_recv  -  Total sum of DCCP payload bytes
- *  @ccid3hcrx_x_recv  -  Receiver estimate of send rate (RFC 3448, sec. 4.3)
- *  @ccid3hcrx_rtt  -  Receiver estimate of RTT
- *  @ccid3hcrx_tstamp_last_feedback  -  Time at which last feedback was sent
- *  @ccid3hcrx_tstamp_last_ack  -  Time at which last feedback was sent
- *  @ccid3hcrx_hist  -  Packet history (loss detection + RTT sampling)
- *  @ccid3hcrx_li_hist  -  Loss Interval database
- *  @ccid3hcrx_s  -  Received packet size in bytes
- *  @ccid3hcrx_pinv  -  Inverse of Loss Event Rate (RFC 4342, sec. 8.5)
+ *  @last_counter  -  Tracks window counter (RFC 4342, 8.1)
+ *  @feedback  -  The type of the feedback last sent
+ *  @x_recv  -  Receiver estimate of send rate (RFC 3448, sec. 4.3)
+ *  @tstamp_last_feedback  -  Time at which last feedback was sent
+ *  @hist  -  Packet history (loss detection + RTT sampling)
+ *  @li_hist  -  Loss Interval database
+ *  @p_inverse  -  Inverse of Loss Event Rate (RFC 4342, sec. 8.5)
  */
 struct ccid3_hc_rx_sock {
-	u8				ccid3hcrx_last_counter:4;
-	enum ccid3_hc_rx_states		ccid3hcrx_state:8;
-	u32				ccid3hcrx_bytes_recv;
-	u32				ccid3hcrx_x_recv;
-	u32				ccid3hcrx_rtt;
-	ktime_t				ccid3hcrx_tstamp_last_feedback;
-	struct tfrc_rx_hist		ccid3hcrx_hist;
-	struct tfrc_loss_hist		ccid3hcrx_li_hist;
-	u16				ccid3hcrx_s;
-#define ccid3hcrx_pinv			ccid3hcrx_li_hist.i_mean
+	u8				last_counter:4;
+	enum ccid3_fback_type		feedback:4;
+	u32				x_recv;
+	ktime_t				tstamp_last_feedback;
+	struct tfrc_rx_hist		hist;
+	struct tfrc_loss_hist		li_hist;
+#define p_inverse			li_hist.i_mean
 };
 
 static inline struct ccid3_hc_rx_sock *ccid3_hc_rx_sk(const struct sock *sk)

net/dccp/ccids/lib/loss_interval.c

@@ -86,21 +86,26 @@ static void tfrc_lh_calc_i_mean(struct tfrc_loss_hist *lh)
 
 /**
  * tfrc_lh_update_i_mean  -  Update the `open' loss interval I_0
- * For recomputing p: returns `true' if p > p_prev  <=>  1/p < 1/p_prev
+ * This updates I_mean as the sequence numbers increase. As a consequence, the
+ * open loss interval I_0 increases, hence p = W_tot/max(I_tot0, I_tot1)
+ * decreases, and thus there is no need to send renewed feedback.
  */
-u8 tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *skb)
+void tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *skb)
 {
 	struct tfrc_loss_interval *cur = tfrc_lh_peek(lh);
-	u32 old_i_mean = lh->i_mean;
 	s64 len;
 
 	if (cur == NULL)			/* not initialised */
-		return 0;
+		return;
+
+	/* FIXME: should probably also count non-data packets (RFC 4342, 6.1) */
+	if (!dccp_data_packet(skb))
+		return;
 
 	len = dccp_delta_seqno(cur->li_seqno, DCCP_SKB_CB(skb)->dccpd_seq) + 1;
 
 	if (len - (s64)cur->li_length <= 0)	/* duplicate or reordered */
-		return 0;
+		return;
 
 	if (SUB16(dccp_hdr(skb)->dccph_ccval, cur->li_ccval) > 4)
 		/*
@@ -114,14 +119,11 @@ u8 tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *skb)
 		cur->li_is_closed = 1;
 
 	if (tfrc_lh_length(lh) == 1)		/* due to RFC 3448, 6.3.1 */
-		return 0;
+		return;
 
 	cur->li_length = len;
 	tfrc_lh_calc_i_mean(lh);
-
-	return (lh->i_mean < old_i_mean);
 }
-EXPORT_SYMBOL_GPL(tfrc_lh_update_i_mean);
 
 /* Determine if `new_loss' does begin a new loss interval [RFC 4342, 10.2] */
 static inline u8 tfrc_lh_is_new_loss(struct tfrc_loss_interval *cur,
@@ -138,18 +140,18 @@ static inline u8 tfrc_lh_is_new_loss(struct tfrc_loss_interval *cur,
  * @sk:		   Used by @calc_first_li in caller-specific way (subtyping)
  * Updates I_mean and returns 1 if a new interval has in fact been added to @lh.
  */
-int tfrc_lh_interval_add(struct tfrc_loss_hist *lh, struct tfrc_rx_hist *rh,
-			 u32 (*calc_first_li)(struct sock *), struct sock *sk)
+bool tfrc_lh_interval_add(struct tfrc_loss_hist *lh, struct tfrc_rx_hist *rh,
+			  u32 (*calc_first_li)(struct sock *), struct sock *sk)
 {
 	struct tfrc_loss_interval *cur = tfrc_lh_peek(lh), *new;
 
 	if (cur != NULL && !tfrc_lh_is_new_loss(cur, tfrc_rx_hist_loss_prev(rh)))
-		return 0;
+		return false;
 
 	new = tfrc_lh_demand_next(lh);
 	if (unlikely(new == NULL)) {
 		DCCP_CRIT("Cannot allocate/add loss record.");
-		return 0;
+		return false;
 	}
 
 	new->li_seqno	  = tfrc_rx_hist_loss_prev(rh)->tfrchrx_seqno;
@@ -167,7 +169,7 @@ int tfrc_lh_interval_add(struct tfrc_loss_hist *lh, struct tfrc_rx_hist *rh,
 
 		tfrc_lh_calc_i_mean(lh);
 	}
-	return 1;
+	return true;
 }
 EXPORT_SYMBOL_GPL(tfrc_lh_interval_add);
 

net/dccp/ccids/lib/loss_interval.h

@@ -67,9 +67,9 @@ static inline u8 tfrc_lh_length(struct tfrc_loss_hist *lh)
 
 struct tfrc_rx_hist;
 
-extern int  tfrc_lh_interval_add(struct tfrc_loss_hist *, struct tfrc_rx_hist *,
+extern bool tfrc_lh_interval_add(struct tfrc_loss_hist *, struct tfrc_rx_hist *,
 				 u32 (*first_li)(struct sock *), struct sock *);
-extern u8   tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *);
+extern void tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *);
 extern void tfrc_lh_cleanup(struct tfrc_loss_hist *lh);
 
 #endif /* _DCCP_LI_HIST_ */

net/dccp/ccids/lib/packet_history.c

@@ -40,18 +40,6 @@
 #include "packet_history.h"
 #include "../../dccp.h"
 
-/**
- *  tfrc_tx_hist_entry  -  Simple singly-linked TX history list
- *  @next:  next oldest entry (LIFO order)
- *  @seqno: sequence number of this entry
- *  @stamp: send time of packet with sequence number @seqno
- */
-struct tfrc_tx_hist_entry {
-	struct tfrc_tx_hist_entry *next;
-	u64			  seqno;
-	ktime_t			  stamp;
-};
-
 /*
  * Transmitter History Routines
  */
@@ -73,15 +61,6 @@ void tfrc_tx_packet_history_exit(void)
 	}
 }
 
-static struct tfrc_tx_hist_entry *
-	tfrc_tx_hist_find_entry(struct tfrc_tx_hist_entry *head, u64 seqno)
-{
-	while (head != NULL && head->seqno != seqno)
-		head = head->next;
-
-	return head;
-}
-
 int tfrc_tx_hist_add(struct tfrc_tx_hist_entry **headp, u64 seqno)
 {
 	struct tfrc_tx_hist_entry *entry = kmem_cache_alloc(tfrc_tx_hist_slab, gfp_any());
@@ -111,25 +90,6 @@ void tfrc_tx_hist_purge(struct tfrc_tx_hist_entry **headp)
 }
 EXPORT_SYMBOL_GPL(tfrc_tx_hist_purge);
 
-u32 tfrc_tx_hist_rtt(struct tfrc_tx_hist_entry *head, const u64 seqno,
-		     const ktime_t now)
-{
-	u32 rtt = 0;
-	struct tfrc_tx_hist_entry *packet = tfrc_tx_hist_find_entry(head, seqno);
-
-	if (packet != NULL) {
-		rtt = ktime_us_delta(now, packet->stamp);
-		/*
-		 * Garbage-collect older (irrelevant) entries:
-		 */
-		tfrc_tx_hist_purge(&packet->next);
-	}
-
-	return rtt;
-}
-EXPORT_SYMBOL_GPL(tfrc_tx_hist_rtt);
-
-
 /*
  * 	Receiver History Routines
  */
@@ -191,14 +151,31 @@ int tfrc_rx_hist_duplicate(struct tfrc_rx_hist *h, struct sk_buff *skb)
 }
 EXPORT_SYMBOL_GPL(tfrc_rx_hist_duplicate);
 
+
+static void __tfrc_rx_hist_swap(struct tfrc_rx_hist *h, const u8 a, const u8 b)
+{
+	struct tfrc_rx_hist_entry *tmp = h->ring[a];
+
+	h->ring[a] = h->ring[b];
+	h->ring[b] = tmp;
+}
+
 static void tfrc_rx_hist_swap(struct tfrc_rx_hist *h, const u8 a, const u8 b)
 {
-	const u8 idx_a = tfrc_rx_hist_index(h, a),
-		 idx_b = tfrc_rx_hist_index(h, b);
-	struct tfrc_rx_hist_entry *tmp = h->ring[idx_a];
+	__tfrc_rx_hist_swap(h, tfrc_rx_hist_index(h, a),
+			       tfrc_rx_hist_index(h, b));
+}
 
-	h->ring[idx_a] = h->ring[idx_b];
-	h->ring[idx_b] = tmp;
+/**
+ * tfrc_rx_hist_resume_rtt_sampling  -  Prepare RX history for RTT sampling
+ * This is called after loss detection has finished, when the history entry
+ * with the index of `loss_count' holds the highest-received sequence number.
+ * RTT sampling requires this information at ring[0] (tfrc_rx_hist_sample_rtt).
+ */
+static inline void tfrc_rx_hist_resume_rtt_sampling(struct tfrc_rx_hist *h)
+{
+	__tfrc_rx_hist_swap(h, 0, tfrc_rx_hist_index(h, h->loss_count));
+	h->loss_count = h->loss_start = 0;
 }
 
 /*
@@ -215,10 +192,8 @@ static void __do_track_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u64 n1)
 	u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno,
 	    s1 = DCCP_SKB_CB(skb)->dccpd_seq;
 
-	if (!dccp_loss_free(s0, s1, n1)) {	/* gap between S0 and S1 */
+	if (!dccp_loss_free(s0, s1, n1))	/* gap between S0 and S1 */
 		h->loss_count = 1;
-		tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n1);
-	}
 }
 
 static void __one_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n2)
@@ -240,8 +215,7 @@ static void __one_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n2
 
 		if (dccp_loss_free(s2, s1, n1)) {
 			/* hole is filled: S0, S2, and S1 are consecutive */
-			h->loss_count = 0;
-			h->loss_start = tfrc_rx_hist_index(h, 1);
+			tfrc_rx_hist_resume_rtt_sampling(h);
 		} else
 			/* gap between S2 and S1: just update loss_prev */
 			tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_loss_prev(h), skb, n2);
@@ -294,8 +268,7 @@ static int __two_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n3)
 
 			if (dccp_loss_free(s1, s2, n2)) {
 				/* entire hole filled by S0, S3, S1, S2 */
-				h->loss_start = tfrc_rx_hist_index(h, 2);
-				h->loss_count = 0;
+				tfrc_rx_hist_resume_rtt_sampling(h);
 			} else {
 				/* gap remains between S1 and S2 */
 				h->loss_start = tfrc_rx_hist_index(h, 1);
@@ -339,8 +312,7 @@ static void __three_after_loss(struct tfrc_rx_hist *h)
 
 		if (dccp_loss_free(s2, s3, n3)) {
 			/* no gap between S2 and S3: entire hole is filled */
-			h->loss_start = tfrc_rx_hist_index(h, 3);
-			h->loss_count = 0;
+			tfrc_rx_hist_resume_rtt_sampling(h);
 		} else {
 			/* gap between S2 and S3 */
 			h->loss_start = tfrc_rx_hist_index(h, 2);
@@ -354,13 +326,13 @@ static void __three_after_loss(struct tfrc_rx_hist *h)
 }
 
 /**
- *  tfrc_rx_handle_loss  -  Loss detection and further processing
- *  @h:		    The non-empty RX history object
- *  @lh:	    Loss Intervals database to update
- *  @skb:	    Currently received packet
- *  @ndp:	    The NDP count belonging to @skb
- *  @calc_first_li: Caller-dependent computation of first loss interval in @lh
- *  @sk:	    Used by @calc_first_li (see tfrc_lh_interval_add)
+ *  tfrc_rx_congestion_event  -  Loss detection and further processing
+ *  @h:		The non-empty RX history object
+ *  @lh:	Loss Intervals database to update
+ *  @skb:	Currently received packet
+ *  @ndp:	The NDP count belonging to @skb
+ *  @first_li:	Caller-dependent computation of first loss interval in @lh
+ *  @sk:	Used by @calc_first_li (see tfrc_lh_interval_add)
  *  Chooses action according to pending loss, updates LI database when a new
  *  loss was detected, and does required post-processing. Returns 1 when caller
  *  should send feedback, 0 otherwise.
@@ -368,15 +340,20 @@ static void __three_after_loss(struct tfrc_rx_hist *h)
  *  records accordingly, the caller should not perform any more RX history
  *  operations when loss_count is greater than 0 after calling this function.
  */
-int tfrc_rx_handle_loss(struct tfrc_rx_hist *h,
-			struct tfrc_loss_hist *lh,
-			struct sk_buff *skb, const u64 ndp,
-			u32 (*calc_first_li)(struct sock *), struct sock *sk)
+bool tfrc_rx_congestion_event(struct tfrc_rx_hist *h,
+			      struct tfrc_loss_hist *lh,
+			      struct sk_buff *skb, const u64 ndp,
+			      u32 (*first_li)(struct sock *), struct sock *sk)
 {
-	int is_new_loss = 0;
+	bool new_event = false;
+
+	if (tfrc_rx_hist_duplicate(h, skb))
+		return 0;
 
 	if (h->loss_count == 0) {
 		__do_track_loss(h, skb, ndp);
+		tfrc_rx_hist_sample_rtt(h, skb);
+		tfrc_rx_hist_add_packet(h, skb, ndp);
 	} else if (h->loss_count == 1) {
 		__one_after_loss(h, skb, ndp);
 	} else if (h->loss_count != 2) {
@@ -385,34 +362,57 @@ int tfrc_rx_handle_loss(struct tfrc_rx_hist *h,
 		/*
 		 * Update Loss Interval database and recycle RX records
 		 */
-		is_new_loss = tfrc_lh_interval_add(lh, h, calc_first_li, sk);
+		new_event = tfrc_lh_interval_add(lh, h, first_li, sk);
 		__three_after_loss(h);
 	}
-	return is_new_loss;
+
+	/*
+	 * Update moving-average of `s' and the sum of received payload bytes.
+	 */
+	if (dccp_data_packet(skb)) {
+		const u32 payload = skb->len - dccp_hdr(skb)->dccph_doff * 4;
+
+		h->packet_size = tfrc_ewma(h->packet_size, payload, 9);
+		h->bytes_recvd += payload;
+	}
+
+	/* RFC 3448, 6.1: update I_0, whose growth implies p <= p_prev */
+	if (!new_event)
+		tfrc_lh_update_i_mean(lh, skb);
+
+	return new_event;
 }
-EXPORT_SYMBOL_GPL(tfrc_rx_handle_loss);
+EXPORT_SYMBOL_GPL(tfrc_rx_congestion_event);
 
-int tfrc_rx_hist_alloc(struct tfrc_rx_hist *h)
+/* Compute the sending rate X_recv measured between feedback intervals */
+u32 tfrc_rx_hist_x_recv(struct tfrc_rx_hist *h, const u32 last_x_recv)
 {
-	int i;
+	u64 bytes = h->bytes_recvd, last_rtt = h->rtt_estimate;
+	s64 delta = ktime_to_us(net_timedelta(h->bytes_start));
 
-	for (i = 0; i <= TFRC_NDUPACK; i++) {
-		h->ring[i] = kmem_cache_alloc(tfrc_rx_hist_slab, GFP_ATOMIC);
-		if (h->ring[i] == NULL)
-			goto out_free;
-	}
+	WARN_ON(delta <= 0);
+	/*
+	 * Ensure that the sampling interval for X_recv is at least one RTT,
+	 * by extending the sampling interval backwards in time, over the last
+	 * R_(m-1) seconds, as per rfc3448bis-06, 6.2.
+	 * To reduce noise (e.g. when the RTT changes often), this is only
+	 * done when delta is smaller than RTT/2.
+	 */
+	if (last_x_recv > 0 && delta < last_rtt/2) {
+		tfrc_pr_debug("delta < RTT ==> %ld us < %u us\n",
+			      (long)delta, (unsigned)last_rtt);
 
-	h->loss_count = h->loss_start = 0;
-	return 0;
+		delta = (bytes ? delta : 0) + last_rtt;
+		bytes += div_u64((u64)last_x_recv * last_rtt, USEC_PER_SEC);
+	}
 
-out_free:
-	while (i-- != 0) {
-		kmem_cache_free(tfrc_rx_hist_slab, h->ring[i]);
-		h->ring[i] = NULL;
+	if (unlikely(bytes == 0)) {
+		DCCP_WARN("X_recv == 0, using old value of %u\n", last_x_recv);
+		return last_x_recv;
 	}
-	return -ENOBUFS;
+	return scaled_div32(bytes, delta);
 }
-EXPORT_SYMBOL_GPL(tfrc_rx_hist_alloc);
+EXPORT_SYMBOL_GPL(tfrc_rx_hist_x_recv);
 
 void tfrc_rx_hist_purge(struct tfrc_rx_hist *h)
 {
@@ -426,73 +426,81 @@ void tfrc_rx_hist_purge(struct tfrc_rx_hist *h)
 }
 EXPORT_SYMBOL_GPL(tfrc_rx_hist_purge);
 
-/**
- * tfrc_rx_hist_rtt_last_s - reference entry to compute RTT samples against
- */
-static inline struct tfrc_rx_hist_entry *
-			tfrc_rx_hist_rtt_last_s(const struct tfrc_rx_hist *h)
+static int tfrc_rx_hist_alloc(struct tfrc_rx_hist *h)
 {
-	return h->ring[0];
+	int i;
+
+	memset(h, 0, sizeof(*h));
+
+	for (i = 0; i <= TFRC_NDUPACK; i++) {
+		h->ring[i] = kmem_cache_alloc(tfrc_rx_hist_slab, GFP_ATOMIC);
+		if (h->ring[i] == NULL) {
+			tfrc_rx_hist_purge(h);
+			return -ENOBUFS;
+		}
+	}
+	return 0;
 }
 
-/**
- * tfrc_rx_hist_rtt_prev_s: previously suitable (wrt rtt_last_s) RTT-sampling entry
- */
-static inline struct tfrc_rx_hist_entry *
-			tfrc_rx_hist_rtt_prev_s(const struct tfrc_rx_hist *h)
+int tfrc_rx_hist_init(struct tfrc_rx_hist *h, struct sock *sk)
 {
-	return h->ring[h->rtt_sample_prev];
+	if (tfrc_rx_hist_alloc(h))
+		return -ENOBUFS;
+	/*
+	 * Initialise first entry with GSR to start loss detection as early as
+	 * possible. Code using this must not use any other fields. The entry
+	 * will be overwritten once the CCID updates its received packets.
+	 */
+	tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno = dccp_sk(sk)->dccps_gsr;
+	return 0;
 }
+EXPORT_SYMBOL_GPL(tfrc_rx_hist_init);
 
 /**
  * tfrc_rx_hist_sample_rtt  -  Sample RTT from timestamp / CCVal
- * Based on ideas presented in RFC 4342, 8.1. Returns 0 if it was not able
- * to compute a sample with given data - calling function should check this.
+ * Based on ideas presented in RFC 4342, 8.1. This function expects that no loss
+ * is pending and uses the following history entries (via rtt_sample_prev):
+ * - h->ring[0]  contains the most recent history entry prior to @skb;
+ * - h->ring[1]  is an unused `dummy' entry when the current difference is 0;
  */
-u32 tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h, const struct sk_buff *skb)
+void tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h, const struct sk_buff *skb)
 {
-	u32 sample = 0,
-	    delta_v = SUB16(dccp_hdr(skb)->dccph_ccval,
-			    tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval);
-
-	if (delta_v < 1 || delta_v > 4) {	/* unsuitable CCVal delta */
-		if (h->rtt_sample_prev == 2) {	/* previous candidate stored */
-			sample = SUB16(tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_ccval,
-				       tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval);
-			if (sample)
-				sample = 4 / sample *
-				         ktime_us_delta(tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_tstamp,
-							tfrc_rx_hist_rtt_last_s(h)->tfrchrx_tstamp);
-			else    /*
-				 * FIXME: This condition is in principle not
-				 * possible but occurs when CCID is used for
-				 * two-way data traffic. I have tried to trace
-				 * it, but the cause does not seem to be here.
-				 */
-				DCCP_BUG("please report to dccp@vger.kernel.org"
-					 " => prev = %u, last = %u",
-					 tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_ccval,
-					 tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval);
-		} else if (delta_v < 1) {
-			h->rtt_sample_prev = 1;
-			goto keep_ref_for_next_time;
-		}
+	struct tfrc_rx_hist_entry *last = h->ring[0];
+	u32 sample, delta_v;
 
-	} else if (delta_v == 4) /* optimal match */
-		sample = ktime_to_us(net_timedelta(tfrc_rx_hist_rtt_last_s(h)->tfrchrx_tstamp));
-	else {			 /* suboptimal match */
-		h->rtt_sample_prev = 2;
-		goto keep_ref_for_next_time;
-	}
+	/*
+	 * When not to sample:
+	 * - on non-data packets
+	 *   (RFC 4342, 8.1: CCVal only fully defined for data packets);
+	 * - when no data packets have been received yet
+	 *   (FIXME: using sampled packet size as indicator here);
+	 * - as long as there are gaps in the sequence space (pending loss).
+	 */
+	if (!dccp_data_packet(skb) || h->packet_size == 0 ||
+	    tfrc_rx_hist_loss_pending(h))
+		return;
 
-	if (unlikely(sample > DCCP_SANE_RTT_MAX)) {
-		DCCP_WARN("RTT sample %u too large, using max\n", sample);
-		sample = DCCP_SANE_RTT_MAX;
+	h->rtt_sample_prev = 0;		/* reset previous candidate */
+
+	delta_v = SUB16(dccp_hdr(skb)->dccph_ccval, last->tfrchrx_ccval);
+	if (delta_v == 0) {		/* less than RTT/4 difference */
+		h->rtt_sample_prev = 1;
+		return;
 	}
+	sample = dccp_sane_rtt(ktime_to_us(net_timedelta(last->tfrchrx_tstamp)));
 
-	h->rtt_sample_prev = 0;	       /* use current entry as next reference */
-keep_ref_for_next_time:
+	if (delta_v <= 4)		/* between RTT/4 and RTT */
+		sample *= 4 / delta_v;
+	else if (!(sample < h->rtt_estimate && sample > h->rtt_estimate/2))
+		/*
+		* Optimisation: CCVal difference is greater than 1 RTT, yet the
+		* sample is less than the local RTT estimate; which means that
+		* the RTT estimate is too high.
+		* To avoid noise, it is not done if the sample is below RTT/2.
+		*/
+		return;
 
-	return sample;
+	/* Use a lower weight than usual to increase responsiveness */
+	h->rtt_estimate = tfrc_ewma(h->rtt_estimate, sample, 5);
 }
 EXPORT_SYMBOL_GPL(tfrc_rx_hist_sample_rtt);

net/dccp/ccids/lib/packet_history.h

@@ -40,12 +40,28 @@
 #include <linux/slab.h>
 #include "tfrc.h"
 
-struct tfrc_tx_hist_entry;
+/**
+ *  tfrc_tx_hist_entry  -  Simple singly-linked TX history list
+ *  @next:  next oldest entry (LIFO order)
+ *  @seqno: sequence number of this entry
+ *  @stamp: send time of packet with sequence number @seqno
+ */
+struct tfrc_tx_hist_entry {
+	struct tfrc_tx_hist_entry *next;
+	u64			  seqno;
+	ktime_t			  stamp;
+};
+
+static inline struct tfrc_tx_hist_entry *
+	tfrc_tx_hist_find_entry(struct tfrc_tx_hist_entry *head, u64 seqno)
+{
+	while (head != NULL && head->seqno != seqno)
+		head = head->next;
+	return head;
+}
 
 extern int  tfrc_tx_hist_add(struct tfrc_tx_hist_entry **headp, u64 seqno);
 extern void tfrc_tx_hist_purge(struct tfrc_tx_hist_entry **headp);
-extern u32  tfrc_tx_hist_rtt(struct tfrc_tx_hist_entry *head,
-			     const u64 seqno, const ktime_t now);
 
 /* Subtraction a-b modulo-16, respects circular wrap-around */
 #define SUB16(a, b) (((a) + 16 - (b)) & 0xF)
@@ -75,12 +91,22 @@ struct tfrc_rx_hist_entry {
  * @loss_count:		Number of entries in circular history
  * @loss_start:		Movable index (for loss detection)
  * @rtt_sample_prev:	Used during RTT sampling, points to candidate entry
+ * @rtt_estimate:	Receiver RTT estimate
+ * @packet_size:	Packet size in bytes (as per RFC 3448, 3.1)
+ * @bytes_recvd:	Number of bytes received since @bytes_start
+ * @bytes_start:	Start time for counting @bytes_recvd
  */
 struct tfrc_rx_hist {
 	struct tfrc_rx_hist_entry *ring[TFRC_NDUPACK + 1];
 	u8			  loss_count:2,
 				  loss_start:2;
+	/* Receiver RTT sampling */
 #define rtt_sample_prev		  loss_start
+	u32			  rtt_estimate;
+	/* Receiver sampling of application payload lengths */
+	u32			  packet_size,
+				  bytes_recvd;
+	ktime_t			  bytes_start;
 };
 
 /**
@@ -124,20 +150,50 @@ static inline bool tfrc_rx_hist_loss_pending(const struct tfrc_rx_hist *h)
 	return h->loss_count > 0;
 }
 
+/*
+ * Accessor functions to retrieve parameters sampled by the RX history
+ */
+static inline u32 tfrc_rx_hist_packet_size(const struct tfrc_rx_hist *h)
+{
+	if (h->packet_size == 0) {
+		DCCP_WARN("No sample for s, using fallback\n");
+		return TCP_MIN_RCVMSS;
+	}
+	return h->packet_size;
+
+}
+static inline u32 tfrc_rx_hist_rtt(const struct tfrc_rx_hist *h)
+{
+	if (h->rtt_estimate == 0) {
+		DCCP_WARN("No RTT estimate available, using fallback RTT\n");
+		return  DCCP_FALLBACK_RTT;
+	}
+	return h->rtt_estimate;
+}
+
+static inline void tfrc_rx_hist_restart_byte_counter(struct tfrc_rx_hist *h)
+{
+	h->bytes_recvd = 0;
+	h->bytes_start = ktime_get_real();
+}
+
+extern u32  tfrc_rx_hist_x_recv(struct tfrc_rx_hist *h, const u32 last_x_recv);
+
+
 extern void tfrc_rx_hist_add_packet(struct tfrc_rx_hist *h,
 				    const struct sk_buff *skb, const u64 ndp);
 
 extern int tfrc_rx_hist_duplicate(struct tfrc_rx_hist *h, struct sk_buff *skb);
 
 struct tfrc_loss_hist;
-extern int  tfrc_rx_handle_loss(struct tfrc_rx_hist *h,
-				struct tfrc_loss_hist *lh,
-				struct sk_buff *skb, const u64 ndp,
-				u32 (*first_li)(struct sock *sk),
-				struct sock *sk);
-extern u32 tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h,
-				   const struct sk_buff *skb);
-extern int tfrc_rx_hist_alloc(struct tfrc_rx_hist *h);
+extern bool tfrc_rx_congestion_event(struct tfrc_rx_hist *h,
+				     struct tfrc_loss_hist *lh,
+				     struct sk_buff *skb, const u64 ndp,
+				     u32 (*first_li)(struct sock *sk),
+				     struct sock *sk);
+extern void tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h,
+				    const struct sk_buff *skb);
+extern int  tfrc_rx_hist_init(struct tfrc_rx_hist *h, struct sock *sk);
 extern void tfrc_rx_hist_purge(struct tfrc_rx_hist *h);
 
 #endif /* _DCCP_PKT_HIST_ */

net/dccp/ccids/lib/tfrc.h

@@ -47,6 +47,21 @@ static inline u32 scaled_div32(u64 a, u64 b)
 	return result;
 }
 
+/**
+ * tfrc_scaled_sqrt  -  Compute scaled integer sqrt(x) for 0 < x < 2^22-1
+ * Uses scaling to improve accuracy of the integer approximation of sqrt(). The
+ * scaling factor of 2^10 limits the maximum @sample to 4e6; this is okay for
+ * clamped RTT samples (dccp_sample_rtt).
+ * Should best be used for expressions of type sqrt(x)/sqrt(y), since then the
+ * scaling factor is neutralised. For this purpose, it avoids returning zero.
+ */
+static inline u16 tfrc_scaled_sqrt(const u32 sample)
+{
+	const unsigned long non_zero_sample = sample ? : 1;
+
+	return int_sqrt(non_zero_sample << 10);
+}
+
 /**
  * tfrc_ewma  -  Exponentially weighted moving average
  * @weight: Weight to be used as damping factor, in units of 1/10
@@ -58,6 +73,7 @@ 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

@@ -632,8 +632,16 @@ u32 tfrc_calc_x(u16 s, u32 R, u32 p)
 
 	if (p <= TFRC_CALC_X_SPLIT) 		{     /* 0.0000 < p <= 0.05   */
 		if (p < TFRC_SMALLEST_P) {	      /* 0.0000 < p <  0.0001 */
-			DCCP_WARN("Value of p (%d) below resolution. "
-				  "Substituting %d\n", p, TFRC_SMALLEST_P);
+			/*
+			 * In the congestion-avoidance phase p decays towards 0
+			 * when there are no further losses, so this case is
+			 * natural. Truncating to p_min = 0.01% means that the
+			 * maximum achievable throughput is limited to about
+			 * X_calc_max = 122.4 * s/RTT (see RFC 3448, 3.1); e.g.
+			 * with s=1500 bytes, RTT=0.01 s: X_calc_max = 147 Mbps.
+			 */
+			tfrc_pr_debug("Value of p (%d) below resolution. "
+				      "Substituting %d\n", p, TFRC_SMALLEST_P);
 			index = 0;
 		} else 				      /* 0.0001 <= p <= 0.05  */
 			index =  p/TFRC_SMALLEST_P - 1;
@@ -658,7 +666,6 @@ u32 tfrc_calc_x(u16 s, u32 R, u32 p)
 	result = scaled_div(s, R);
 	return scaled_div32(result, f);
 }
-
 EXPORT_SYMBOL_GPL(tfrc_calc_x);
 
 /**
@@ -693,5 +700,19 @@ u32 tfrc_calc_x_reverse_lookup(u32 fvalue)
 	index = tfrc_binsearch(fvalue, 0);
 	return (index + 1) * 1000000 / TFRC_CALC_X_ARRSIZE;
 }
-
 EXPORT_SYMBOL_GPL(tfrc_calc_x_reverse_lookup);
+
+/**
+ * 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);
+}
+EXPORT_SYMBOL_GPL(tfrc_invert_loss_event_rate);

net/dccp/dccp.h

@@ -42,9 +42,11 @@
 extern int dccp_debug;
 #define dccp_pr_debug(format, a...)	  DCCP_PR_DEBUG(dccp_debug, format, ##a)
 #define dccp_pr_debug_cat(format, a...)   DCCP_PRINTK(dccp_debug, format, ##a)
+#define dccp_debug(fmt, a...)		  dccp_pr_debug_cat(KERN_DEBUG fmt, ##a)
 #else
 #define dccp_pr_debug(format, a...)
 #define dccp_pr_debug_cat(format, a...)
+#define dccp_debug(format, a...)
 #endif
 
 extern struct inet_hashinfo dccp_hashinfo;
@@ -61,11 +63,14 @@ extern void dccp_time_wait(struct sock *sk, int state, int timeo);
  *    - DCCP-Reset    with ACK Subheader and 4 bytes of Reset Code fields
  *  Hence a safe upper bound for the maximum option length is 1020-28 = 992
  */
-#define MAX_DCCP_SPECIFIC_HEADER (255 * sizeof(int))
+#define MAX_DCCP_SPECIFIC_HEADER (255 * sizeof(uint32_t))
 #define DCCP_MAX_PACKET_HDR 28
 #define DCCP_MAX_OPT_LEN (MAX_DCCP_SPECIFIC_HEADER - DCCP_MAX_PACKET_HDR)
 #define MAX_DCCP_HEADER (MAX_DCCP_SPECIFIC_HEADER + MAX_HEADER)
 
+/* Upper bound for initial feature-negotiation overhead (padded to 32 bits) */
+#define DCCP_FEATNEG_OVERHEAD	 (32 * sizeof(uint32_t))
+
 #define DCCP_TIMEWAIT_LEN (60 * HZ) /* how long to wait to destroy TIME-WAIT
 				     * state, about 60 seconds */
 
@@ -81,10 +86,13 @@ extern void dccp_time_wait(struct sock *sk, int state, int timeo);
  */
 #define DCCP_RTO_MAX ((unsigned)(64 * HZ))
 
+/* DCCP base time resolution - 10 microseconds (RFC 4340, 13.1 ... 13.3) */
+#define DCCP_TIME_RESOLUTION	10
+
 /*
  * RTT sampling: sanity bounds and fallback RTT value from RFC 4340, section 3.4
  */
-#define DCCP_SANE_RTT_MIN	100
+#define DCCP_SANE_RTT_MIN	(10 * DCCP_TIME_RESOLUTION)
 #define DCCP_FALLBACK_RTT	(USEC_PER_SEC / 5)
 #define DCCP_SANE_RTT_MAX	(3 * USEC_PER_SEC)
 
@@ -95,12 +103,6 @@ extern void dccp_time_wait(struct sock *sk, int state, int timeo);
 extern int  sysctl_dccp_request_retries;
 extern int  sysctl_dccp_retries1;
 extern int  sysctl_dccp_retries2;
-extern int  sysctl_dccp_feat_sequence_window;
-extern int  sysctl_dccp_feat_rx_ccid;
-extern int  sysctl_dccp_feat_tx_ccid;
-extern int  sysctl_dccp_feat_ack_ratio;
-extern int  sysctl_dccp_feat_send_ack_vector;
-extern int  sysctl_dccp_feat_send_ndp_count;
 extern int  sysctl_dccp_tx_qlen;
 extern int  sysctl_dccp_sync_ratelimit;
 
@@ -235,8 +237,22 @@ extern void dccp_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
 extern void dccp_send_sync(struct sock *sk, const u64 seq,
 			   const enum dccp_pkt_type pkt_type);
 
-extern void dccp_write_xmit(struct sock *sk, int block);
+/*
+ * TX Packet Dequeueing Interface
+ */
+extern void		dccp_qpolicy_push(struct sock *sk, struct sk_buff *skb);
+extern bool		dccp_qpolicy_full(struct sock *sk);
+extern void		dccp_qpolicy_drop(struct sock *sk, struct sk_buff *skb);
+extern struct sk_buff	*dccp_qpolicy_top(struct sock *sk);
+extern struct sk_buff	*dccp_qpolicy_pop(struct sock *sk);
+extern bool		dccp_qpolicy_param_ok(struct sock *sk, __be32 param);
+
+/*
+ * TX Packet Output and TX Timers
+ */
+extern void dccp_write_xmit(struct sock *sk);
 extern void dccp_write_space(struct sock *sk);
+extern void dccp_flush_write_queue(struct sock *sk, long *time_budget);
 
 extern void dccp_init_xmit_timers(struct sock *sk);
 static inline void dccp_clear_xmit_timers(struct sock *sk)
@@ -252,7 +268,8 @@ extern const char *dccp_state_name(const int state);
 extern void dccp_set_state(struct sock *sk, const int state);
 extern void dccp_done(struct sock *sk);
 
-extern void dccp_reqsk_init(struct request_sock *req, struct sk_buff *skb);
+extern int  dccp_reqsk_init(struct request_sock *rq, struct dccp_sock const *dp,
+			    struct sk_buff const *skb);
 
 extern int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb);
 
@@ -317,7 +334,14 @@ extern struct sk_buff *dccp_ctl_make_reset(struct sock *sk,
 extern int	   dccp_send_reset(struct sock *sk, enum dccp_reset_codes code);
 extern void	   dccp_send_close(struct sock *sk, const int active);
 extern int	   dccp_invalid_packet(struct sk_buff *skb);
-extern u32	   dccp_sample_rtt(struct sock *sk, long delta);
+
+static inline u32  dccp_sane_rtt(long usec_sample)
+{
+	if (unlikely(usec_sample <= 0 || usec_sample > DCCP_SANE_RTT_MAX))
+		DCCP_WARN("RTT sample %ld out of bounds!\n", usec_sample);
+	return clamp_val(usec_sample, DCCP_SANE_RTT_MIN, DCCP_SANE_RTT_MAX);
+}
+extern u32 dccp_sample_rtt(struct sock *sk, long delta);
 
 static inline int dccp_bad_service_code(const struct sock *sk,
 					const __be32 service)
@@ -411,36 +435,62 @@ static inline void dccp_hdr_set_ack(struct dccp_hdr_ack_bits *dhack,
 static inline void dccp_update_gsr(struct sock *sk, u64 seq)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
-	const struct dccp_minisock *dmsk = dccp_msk(sk);
 
 	dp->dccps_gsr = seq;
-	dccp_set_seqno(&dp->dccps_swl,
-		       dp->dccps_gsr + 1 - (dmsk->dccpms_sequence_window / 4));
-	dccp_set_seqno(&dp->dccps_swh,
-		       dp->dccps_gsr + (3 * dmsk->dccpms_sequence_window) / 4);
+	/* Sequence validity window depends on remote Sequence Window (7.5.1) */
+	dp->dccps_swl = SUB48(ADD48(dp->dccps_gsr, 1), dp->dccps_r_seq_win / 4);
+	/*
+	 * Adjust SWL so that it is not below ISR. In contrast to RFC 4340,
+	 * 7.5.1 we perform this check beyond the initial handshake: W/W' are
+	 * always > 32, so for the first W/W' packets in the lifetime of a
+	 * connection we always have to adjust SWL.
+	 * A second reason why we are doing this is that the window depends on
+	 * the feature-remote value of Sequence Window: nothing stops the peer
+	 * from updating this value while we are busy adjusting SWL for the
+	 * first W packets (we would have to count from scratch again then).
+	 * Therefore it is safer to always make sure that the Sequence Window
+	 * is not artificially extended by a peer who grows SWL downwards by
+	 * continually updating the feature-remote Sequence-Window.
+	 * If sequence numbers wrap it is bad luck. But that will take a while
+	 * (48 bit), and this measure prevents Sequence-number attacks.
+	 */
+	if (before48(dp->dccps_swl, dp->dccps_isr))
+		dp->dccps_swl = dp->dccps_isr;
+	dp->dccps_swh = ADD48(dp->dccps_gsr, (3 * dp->dccps_r_seq_win) / 4);
 }
 
 static inline void dccp_update_gss(struct sock *sk, u64 seq)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
 
-	dp->dccps_awh = dp->dccps_gss = seq;
-	dccp_set_seqno(&dp->dccps_awl,
-		       (dp->dccps_gss -
-			dccp_msk(sk)->dccpms_sequence_window + 1));
+	dp->dccps_gss = seq;
+	/* Ack validity window depends on local Sequence Window value (7.5.1) */
+	dp->dccps_awl = SUB48(ADD48(dp->dccps_gss, 1), dp->dccps_l_seq_win);
+	/* Adjust AWL so that it is not below ISS - see comment above for SWL */
+	if (before48(dp->dccps_awl, dp->dccps_iss))
+		dp->dccps_awl = dp->dccps_iss;
+	dp->dccps_awh = dp->dccps_gss;
+}
+
+static inline int dccp_ackvec_pending(const struct sock *sk)
+{
+	return dccp_sk(sk)->dccps_hc_rx_ackvec != NULL &&
+	       !dccp_ackvec_is_empty(dccp_sk(sk)->dccps_hc_rx_ackvec);
 }
 
 static inline int dccp_ack_pending(const struct sock *sk)
 {
-	const struct dccp_sock *dp = dccp_sk(sk);
-	return dp->dccps_timestamp_echo != 0 ||
-#ifdef CONFIG_IP_DCCP_ACKVEC
-	       (dccp_msk(sk)->dccpms_send_ack_vector &&
-		dccp_ackvec_pending(dp->dccps_hc_rx_ackvec)) ||
-#endif
-	       inet_csk_ack_scheduled(sk);
+	return dccp_ackvec_pending(sk) || inet_csk_ack_scheduled(sk);
 }
 
+extern int  dccp_feat_signal_nn_change(struct sock *sk, u8 feat, u64 nn_val);
+extern int  dccp_feat_finalise_settings(struct dccp_sock *dp);
+extern int  dccp_feat_server_ccid_dependencies(struct dccp_request_sock *dreq);
+extern int  dccp_feat_insert_opts(struct dccp_sock*, struct dccp_request_sock*,
+				  struct sk_buff *skb);
+extern int  dccp_feat_activate_values(struct sock *sk, struct list_head *fn);
+extern void dccp_feat_list_purge(struct list_head *fn_list);
+
 extern int dccp_insert_options(struct sock *sk, struct sk_buff *skb);
 extern int dccp_insert_options_rsk(struct dccp_request_sock*, struct sk_buff*);
 extern int dccp_insert_option_elapsed_time(struct sock *sk,

net/dccp/diag.c

@@ -29,7 +29,7 @@ static void dccp_get_info(struct sock *sk, struct tcp_info *info)
 	info->tcpi_backoff	= icsk->icsk_backoff;
 	info->tcpi_pmtu		= icsk->icsk_pmtu_cookie;
 
-	if (dccp_msk(sk)->dccpms_send_ack_vector)
+	if (dp->dccps_hc_rx_ackvec != NULL)
 		info->tcpi_options |= TCPI_OPT_SACK;
 
 	ccid_hc_rx_get_info(dp->dccps_hc_rx_ccid, sk, info);

net/dccp/feat.c

@@ -1,11 +1,19 @@
 /*
  *  net/dccp/feat.c
  *
- *  An implementation of the DCCP protocol
- *  Andrea Bittau <a.bittau@cs.ucl.ac.uk>
+ *  Feature negotiation for the DCCP protocol (RFC 4340, section 6)
+ *
+ *  Copyright (c) 2008 The University of Aberdeen, Scotland, UK
+ *  Copyright (c) 2008 Gerrit Renker <gerrit@erg.abdn.ac.uk>
+ *  Rewrote from scratch, some bits from earlier code by
+ *  Copyright (c) 2005 Andrea Bittau <a.bittau@cs.ucl.ac.uk>
+ *
  *
  *  ASSUMPTIONS
  *  -----------
+ *  o Feature negotiation is coordinated with connection setup (as in TCP), wild
+ *    changes of parameters of an established connection are not supported.
+ *  o Changing NN values (Ack Ratio only) is supported in state OPEN/PARTOPEN.
  *  o All currently known SP features have 1-byte quantities. If in the future
  *    extensions of RFCs 4340..42 define features with item lengths larger than
  *    one byte, a feature-specific extension of the code will be required.
@@ -15,635 +23,1510 @@
  *  as published by the Free Software Foundation; either version
  *  2 of the License, or (at your option) any later version.
  */
-
 #include <linux/module.h>
-
 #include "ccid.h"
 #include "feat.h"
 
-#define DCCP_FEAT_SP_NOAGREE (-123)
-
-int dccp_feat_change(struct dccp_minisock *dmsk, u8 type, u8 feature,
-		     u8 *val, u8 len, gfp_t gfp)
-{
-	struct dccp_opt_pend *opt;
-
-	dccp_feat_debug(type, feature, *val);
-
-	if (len > 3) {
-		DCCP_WARN("invalid length %d\n", len);
-		return -EINVAL;
-	}
-	/* XXX add further sanity checks */
-
-	/* check if that feature is already being negotiated */
-	list_for_each_entry(opt, &dmsk->dccpms_pending, dccpop_node) {
-		/* ok we found a negotiation for this option already */
-		if (opt->dccpop_feat == feature && opt->dccpop_type == type) {
-			dccp_pr_debug("Replacing old\n");
-			/* replace */
-			BUG_ON(opt->dccpop_val == NULL);
-			kfree(opt->dccpop_val);
-			opt->dccpop_val	 = val;
-			opt->dccpop_len	 = len;
-			opt->dccpop_conf = 0;
-			return 0;
-		}
-	}
-
-	/* negotiation for a new feature */
-	opt = kmalloc(sizeof(*opt), gfp);
-	if (opt == NULL)
-		return -ENOMEM;
-
-	opt->dccpop_type = type;
-	opt->dccpop_feat = feature;
-	opt->dccpop_len	 = len;
-	opt->dccpop_val	 = val;
-	opt->dccpop_conf = 0;
-	opt->dccpop_sc	 = NULL;
-
-	BUG_ON(opt->dccpop_val == NULL);
-
-	list_add_tail(&opt->dccpop_node, &dmsk->dccpms_pending);
-	return 0;
-}
+/* feature-specific sysctls - initialised to the defaults from RFC 4340, 6.4 */
+unsigned long	sysctl_dccp_sequence_window __read_mostly = 100;
+int		sysctl_dccp_rx_ccid	    __read_mostly = 2,
+		sysctl_dccp_tx_ccid	    __read_mostly = 2;
 
-EXPORT_SYMBOL_GPL(dccp_feat_change);
-
-static int dccp_feat_update_ccid(struct sock *sk, u8 type, u8 new_ccid_nr)
+/*
+ * Feature activation handlers.
+ *
+ * These all use an u64 argument, to provide enough room for NN/SP features. At
+ * this stage the negotiated values have been checked to be within their range.
+ */
+static int dccp_hdlr_ccid(struct sock *sk, u64 ccid, bool rx)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
-	struct dccp_minisock *dmsk = dccp_msk(sk);
-	/* figure out if we are changing our CCID or the peer's */
-	const int rx = type == DCCPO_CHANGE_R;
-	const u8 ccid_nr = rx ? dmsk->dccpms_rx_ccid : dmsk->dccpms_tx_ccid;
-	struct ccid *new_ccid;
-
-	/* Check if nothing is being changed. */
-	if (ccid_nr == new_ccid_nr)
-		return 0;
+	struct ccid *new_ccid = ccid_new(ccid, sk, rx, gfp_any());
 
-	new_ccid = ccid_new(new_ccid_nr, sk, rx, GFP_ATOMIC);
 	if (new_ccid == NULL)
 		return -ENOMEM;
 
 	if (rx) {
 		ccid_hc_rx_delete(dp->dccps_hc_rx_ccid, sk);
 		dp->dccps_hc_rx_ccid = new_ccid;
-		dmsk->dccpms_rx_ccid = new_ccid_nr;
 	} else {
 		ccid_hc_tx_delete(dp->dccps_hc_tx_ccid, sk);
 		dp->dccps_hc_tx_ccid = new_ccid;
-		dmsk->dccpms_tx_ccid = new_ccid_nr;
 	}
-
 	return 0;
 }
 
-static int dccp_feat_update(struct sock *sk, u8 type, u8 feat, u8 val)
+static int dccp_hdlr_seq_win(struct sock *sk, u64 seq_win, bool rx)
 {
-	dccp_feat_debug(type, feat, val);
+	struct dccp_sock *dp = dccp_sk(sk);
 
-	switch (feat) {
-	case DCCPF_CCID:
-		return dccp_feat_update_ccid(sk, type, val);
-	default:
-		dccp_pr_debug("UNIMPLEMENTED: %s(%d, ...)\n",
-			      dccp_feat_typename(type), feat);
-		break;
+	if (rx) {
+		dp->dccps_r_seq_win = seq_win;
+		/* propagate changes to update SWL/SWH */
+		dccp_update_gsr(sk, dp->dccps_gsr);
+	} else {
+		dp->dccps_l_seq_win = seq_win;
+		/* propagate changes to update AWL */
+		dccp_update_gss(sk, dp->dccps_gss);
 	}
 	return 0;
 }
 
-static int dccp_feat_reconcile(struct sock *sk, struct dccp_opt_pend *opt,
-			       u8 *rpref, u8 rlen)
+static int dccp_hdlr_ack_ratio(struct sock *sk, u64 ratio, bool rx)
+{
+#ifndef __CCID2_COPES_GRACEFULLY_WITH_DYNAMIC_ACK_RATIO_UPDATES__
+	/*
+	 * FIXME: This is required until several problems in the CCID-2 code are
+	 * resolved. The CCID-2 code currently does not cope well; using dynamic
+	 * Ack Ratios greater than 1 caused instabilities. These were manifest
+	 * in hangups and long RTO timeouts (1...3 seconds). Until this has been
+	 * stabilised, it is safer not to activate dynamic Ack Ratio changes.
+	 */
+	dccp_pr_debug("Not changing %s Ack Ratio from 1 to %u\n",
+		      rx ? "RX" : "TX", (u16)ratio);
+	ratio = 1;
+#endif
+	if (rx)
+		dccp_sk(sk)->dccps_r_ack_ratio = ratio;
+	else
+		dccp_sk(sk)->dccps_l_ack_ratio = ratio;
+	return 0;
+}
+
+static int dccp_hdlr_ackvec(struct sock *sk, u64 enable, bool rx)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
-	u8 *spref, slen, *res = NULL;
-	int i, j, rc, agree = 1;
 
-	BUG_ON(rpref == NULL);
+	if (rx) {
+		if (enable && dp->dccps_hc_rx_ackvec == NULL) {
+			dp->dccps_hc_rx_ackvec = dccp_ackvec_alloc(gfp_any());
+			if (dp->dccps_hc_rx_ackvec == NULL)
+				return -ENOMEM;
+		} else if (!enable) {
+			dccp_ackvec_free(dp->dccps_hc_rx_ackvec);
+			dp->dccps_hc_rx_ackvec = NULL;
+		}
+	}
+	return 0;
+}
 
-	/* check if we are the black sheep */
-	if (dp->dccps_role == DCCP_ROLE_CLIENT) {
-		spref = rpref;
-		slen  = rlen;
-		rpref = opt->dccpop_val;
-		rlen  = opt->dccpop_len;
-	} else {
-		spref = opt->dccpop_val;
-		slen  = opt->dccpop_len;
+static int dccp_hdlr_ndp(struct sock *sk, u64 enable, bool rx)
+{
+	if (!rx)
+		dccp_sk(sk)->dccps_send_ndp_count = (enable > 0);
+	return 0;
+}
+
+/*
+ * Minimum Checksum Coverage is located at the RX side (9.2.1). This means that
+ * `rx' holds when the sending peer informs about his partial coverage via a
+ * ChangeR() option. In the other case, we are the sender and the receiver
+ * announces its coverage via ChangeL() options. The policy here is to honour
+ * such communication by enabling the corresponding partial coverage - but only
+ * if it has not been set manually before; the warning here means that all
+ * packets will be dropped.
+ */
+static int dccp_hdlr_min_cscov(struct sock *sk, u64 cscov, bool rx)
+{
+	struct dccp_sock *dp = dccp_sk(sk);
+
+	if (rx)
+		dp->dccps_pcrlen = cscov;
+	else {
+		if (dp->dccps_pcslen == 0)
+			dp->dccps_pcslen = cscov;
+		else if (cscov > dp->dccps_pcslen)
+			DCCP_WARN("CsCov %u too small, peer requires >= %u\n",
+				  dp->dccps_pcslen, (u8)cscov);
 	}
+	return 0;
+}
+
+static const struct {
+	u8			feat_num;		/* DCCPF_xxx */
+	enum dccp_feat_type	rxtx;			/* RX or TX  */
+	enum dccp_feat_type	reconciliation;		/* SP or NN  */
+	u8			default_value;		/* as in 6.4 */
+	int (*activation_hdlr)(struct sock *sk, u64 val, bool rx);
+/*
+ *    Lookup table for location and type of features (from RFC 4340/4342)
+ *  +--------------------------+----+-----+----+----+---------+-----------+
+ *  | Feature                  | Location | Reconc. | Initial |  Section  |
+ *  |                          | RX | TX  | SP | NN |  Value  | Reference |
+ *  +--------------------------+----+-----+----+----+---------+-----------+
+ *  | DCCPF_CCID               |    |  X  | X  |    |   2     | 10        |
+ *  | DCCPF_SHORT_SEQNOS       |    |  X  | X  |    |   0     |  7.6.1    |
+ *  | DCCPF_SEQUENCE_WINDOW    |    |  X  |    | X  | 100     |  7.5.2    |
+ *  | DCCPF_ECN_INCAPABLE      | X  |     | X  |    |   0     | 12.1      |
+ *  | DCCPF_ACK_RATIO          |    |  X  |    | X  |   2     | 11.3      |
+ *  | DCCPF_SEND_ACK_VECTOR    | X  |     | X  |    |   0     | 11.5      |
+ *  | DCCPF_SEND_NDP_COUNT     |    |  X  | X  |    |   0     |  7.7.2    |
+ *  | DCCPF_MIN_CSUM_COVER     | X  |     | X  |    |   0     |  9.2.1    |
+ *  | DCCPF_DATA_CHECKSUM      | X  |     | X  |    |   0     |  9.3.1    |
+ *  | DCCPF_SEND_LEV_RATE      | X  |     | X  |    |   0     | 4342/8.4  |
+ *  +--------------------------+----+-----+----+----+---------+-----------+
+ */
+} dccp_feat_table[] = {
+	{ DCCPF_CCID,		 FEAT_AT_TX, FEAT_SP, 2,   dccp_hdlr_ccid     },
+	{ DCCPF_SHORT_SEQNOS,	 FEAT_AT_TX, FEAT_SP, 0,   NULL },
+	{ DCCPF_SEQUENCE_WINDOW, FEAT_AT_TX, FEAT_NN, 100, dccp_hdlr_seq_win  },
+	{ DCCPF_ECN_INCAPABLE,	 FEAT_AT_RX, FEAT_SP, 0,   NULL },
+	{ DCCPF_ACK_RATIO,	 FEAT_AT_TX, FEAT_NN, 2,   dccp_hdlr_ack_ratio},
+	{ DCCPF_SEND_ACK_VECTOR, FEAT_AT_RX, FEAT_SP, 0,   dccp_hdlr_ackvec   },
+	{ DCCPF_SEND_NDP_COUNT,  FEAT_AT_TX, FEAT_SP, 0,   dccp_hdlr_ndp      },
+	{ DCCPF_MIN_CSUM_COVER,  FEAT_AT_RX, FEAT_SP, 0,   dccp_hdlr_min_cscov},
+	{ DCCPF_DATA_CHECKSUM,	 FEAT_AT_RX, FEAT_SP, 0,   NULL },
+	{ DCCPF_SEND_LEV_RATE,	 FEAT_AT_RX, FEAT_SP, 0,   NULL },
+};
+#define DCCP_FEAT_SUPPORTED_MAX		ARRAY_SIZE(dccp_feat_table)
+
+/**
+ * dccp_feat_index  -  Hash function to map feature number into array position
+ * Returns consecutive array index or -1 if the feature is not understood.
+ */
+static int dccp_feat_index(u8 feat_num)
+{
+	/* The first 9 entries are occupied by the types from RFC 4340, 6.4 */
+	if (feat_num > DCCPF_RESERVED && feat_num <= DCCPF_DATA_CHECKSUM)
+		return feat_num - 1;
+
 	/*
-	 * Now we have server preference list in spref and client preference in
-	 * rpref
+	 * Other features: add cases for new feature types here after adding
+	 * them to the above table.
 	 */
-	BUG_ON(spref == NULL);
-	BUG_ON(rpref == NULL);
+	switch (feat_num) {
+	case DCCPF_SEND_LEV_RATE:
+			return DCCP_FEAT_SUPPORTED_MAX - 1;
+	}
+	return -1;
+}
 
-	/* FIXME sanity check vals */
+static u8 dccp_feat_type(u8 feat_num)
+{
+	int idx = dccp_feat_index(feat_num);
 
-	/* Are values in any order?  XXX Lame "algorithm" here */
-	for (i = 0; i < slen; i++) {
-		for (j = 0; j < rlen; j++) {
-			if (spref[i] == rpref[j]) {
-				res = &spref[i];
-				break;
-			}
-		}
-		if (res)
-			break;
-	}
+	if (idx < 0)
+		return FEAT_UNKNOWN;
+	return dccp_feat_table[idx].reconciliation;
+}
 
-	/* we didn't agree on anything */
-	if (res == NULL) {
-		/* confirm previous value */
-		switch (opt->dccpop_feat) {
-		case DCCPF_CCID:
-			/* XXX did i get this right? =P */
-			if (opt->dccpop_type == DCCPO_CHANGE_L)
-				res = &dccp_msk(sk)->dccpms_tx_ccid;
-			else
-				res = &dccp_msk(sk)->dccpms_rx_ccid;
-			break;
+static int dccp_feat_default_value(u8 feat_num)
+{
+	int idx = dccp_feat_index(feat_num);
 
-		default:
-			DCCP_BUG("Fell through, feat=%d", opt->dccpop_feat);
-			/* XXX implement res */
-			return -EFAULT;
-		}
+	return idx < 0 ? : dccp_feat_table[idx].default_value;
+}
 
-		dccp_pr_debug("Don't agree... reconfirming %d\n", *res);
-		agree = 0; /* this is used for mandatory options... */
+/*
+ *	Debugging and verbose-printing section
+ */
+static const char *dccp_feat_fname(const u8 feat)
+{
+	static const char *feature_names[] = {
+		[DCCPF_RESERVED]	= "Reserved",
+		[DCCPF_CCID]		= "CCID",
+		[DCCPF_SHORT_SEQNOS]	= "Allow Short Seqnos",
+		[DCCPF_SEQUENCE_WINDOW]	= "Sequence Window",
+		[DCCPF_ECN_INCAPABLE]	= "ECN Incapable",
+		[DCCPF_ACK_RATIO]	= "Ack Ratio",
+		[DCCPF_SEND_ACK_VECTOR]	= "Send ACK Vector",
+		[DCCPF_SEND_NDP_COUNT]	= "Send NDP Count",
+		[DCCPF_MIN_CSUM_COVER]	= "Min. Csum Coverage",
+		[DCCPF_DATA_CHECKSUM]	= "Send Data Checksum",
+	};
+	if (feat > DCCPF_DATA_CHECKSUM && feat < DCCPF_MIN_CCID_SPECIFIC)
+		return feature_names[DCCPF_RESERVED];
+
+	if (feat ==  DCCPF_SEND_LEV_RATE)
+		return "Send Loss Event Rate";
+	if (feat >= DCCPF_MIN_CCID_SPECIFIC)
+		return "CCID-specific";
+
+	return feature_names[feat];
+}
+
+static const char *dccp_feat_sname[] = { "DEFAULT", "INITIALISING", "CHANGING",
+					 "UNSTABLE", "STABLE" };
+
+#ifdef CONFIG_IP_DCCP_DEBUG
+static const char *dccp_feat_oname(const u8 opt)
+{
+	switch (opt) {
+	case DCCPO_CHANGE_L:  return "Change_L";
+	case DCCPO_CONFIRM_L: return "Confirm_L";
+	case DCCPO_CHANGE_R:  return "Change_R";
+	case DCCPO_CONFIRM_R: return "Confirm_R";
 	}
+	return NULL;
+}
 
-	/* need to put result and our preference list */
-	rlen = 1 + opt->dccpop_len;
-	rpref = kmalloc(rlen, GFP_ATOMIC);
-	if (rpref == NULL)
-		return -ENOMEM;
+static void dccp_feat_printval(u8 feat_num, dccp_feat_val const *val)
+{
+	u8 i, type = dccp_feat_type(feat_num);
+
+	if (val == NULL || (type == FEAT_SP && val->sp.vec == NULL))
+		dccp_pr_debug_cat("(NULL)");
+	else if (type == FEAT_SP)
+		for (i = 0; i < val->sp.len; i++)
+			dccp_pr_debug_cat("%s%u", i ? " " : "", val->sp.vec[i]);
+	else if (type == FEAT_NN)
+		dccp_pr_debug_cat("%llu", (unsigned long long)val->nn);
+	else
+		dccp_pr_debug_cat("unknown type %u", type);
+}
+
+static void dccp_feat_printvals(u8 feat_num, u8 *list, u8 len)
+{
+	u8 type = dccp_feat_type(feat_num);
+	dccp_feat_val fval = { .sp.vec = list, .sp.len = len };
+
+	if (type == FEAT_NN)
+		fval.nn = dccp_decode_value_var(list, len);
+	dccp_feat_printval(feat_num, &fval);
+}
+
+static void dccp_feat_print_entry(struct dccp_feat_entry const *entry)
+{
+	dccp_debug("   * %s %s = ", entry->is_local ? "local" : "remote",
+				    dccp_feat_fname(entry->feat_num));
+	dccp_feat_printval(entry->feat_num, &entry->val);
+	dccp_pr_debug_cat(", state=%s %s\n", dccp_feat_sname[entry->state],
+			  entry->needs_confirm ? "(Confirm pending)" : "");
+}
+
+#define dccp_feat_print_opt(opt, feat, val, len, mandatory)	do {	      \
+	dccp_pr_debug("%s(%s, ", dccp_feat_oname(opt), dccp_feat_fname(feat));\
+	dccp_feat_printvals(feat, val, len);				      \
+	dccp_pr_debug_cat(") %s\n", mandatory ? "!" : "");	} while (0)
+
+#define dccp_feat_print_fnlist(fn_list)  {		\
+	const struct dccp_feat_entry *___entry;		\
+							\
+	dccp_pr_debug("List Dump:\n");			\
+	list_for_each_entry(___entry, fn_list, node)	\
+		dccp_feat_print_entry(___entry);	\
+}
+#else	/* ! CONFIG_IP_DCCP_DEBUG */
+#define dccp_feat_print_opt(opt, feat, val, len, mandatory)
+#define dccp_feat_print_fnlist(fn_list)
+#endif
 
-	*rpref = *res;
-	memcpy(&rpref[1], opt->dccpop_val, opt->dccpop_len);
+static int __dccp_feat_activate(struct sock *sk, const int idx,
+				const bool is_local, dccp_feat_val const *fval)
+{
+	bool rx;
+	u64 val;
+
+	if (idx < 0 || idx >= DCCP_FEAT_SUPPORTED_MAX)
+		return -1;
+	if (dccp_feat_table[idx].activation_hdlr == NULL)
+		return 0;
 
-	/* put it in the "confirm queue" */
-	if (opt->dccpop_sc == NULL) {
-		opt->dccpop_sc = kmalloc(sizeof(*opt->dccpop_sc), GFP_ATOMIC);
-		if (opt->dccpop_sc == NULL) {
-			kfree(rpref);
-			return -ENOMEM;
+	if (fval == NULL) {
+		val = dccp_feat_table[idx].default_value;
+	} else if (dccp_feat_table[idx].reconciliation == FEAT_SP) {
+		if (fval->sp.vec == NULL) {
+			/*
+			 * This can happen when an empty Confirm is sent
+			 * for an SP (i.e. known) feature. In this case
+			 * we would be using the default anyway.
+			 */
+			DCCP_CRIT("Feature #%d undefined: using default", idx);
+			val = dccp_feat_table[idx].default_value;
+		} else {
+			val = fval->sp.vec[0];
 		}
 	} else {
-		/* recycle the confirm slot */
-		BUG_ON(opt->dccpop_sc->dccpoc_val == NULL);
-		kfree(opt->dccpop_sc->dccpoc_val);
-		dccp_pr_debug("recycling confirm slot\n");
-	}
-	memset(opt->dccpop_sc, 0, sizeof(*opt->dccpop_sc));
-
-	opt->dccpop_sc->dccpoc_val = rpref;
-	opt->dccpop_sc->dccpoc_len = rlen;
-
-	/* update the option on our side [we are about to send the confirm] */
-	rc = dccp_feat_update(sk, opt->dccpop_type, opt->dccpop_feat, *res);
-	if (rc) {
-		kfree(opt->dccpop_sc->dccpoc_val);
-		kfree(opt->dccpop_sc);
-		opt->dccpop_sc = NULL;
-		return rc;
+		val = fval->nn;
 	}
 
-	dccp_pr_debug("Will confirm %d\n", *rpref);
+	/* Location is RX if this is a local-RX or remote-TX feature */
+	rx = (is_local == (dccp_feat_table[idx].rxtx == FEAT_AT_RX));
 
-	/* say we want to change to X but we just got a confirm X, suppress our
-	 * change
-	 */
-	if (!opt->dccpop_conf) {
-		if (*opt->dccpop_val == *res)
-			opt->dccpop_conf = 1;
-		dccp_pr_debug("won't ask for change of same feature\n");
+	dccp_debug("   -> activating %s %s, %sval=%llu\n", rx ? "RX" : "TX",
+		   dccp_feat_fname(dccp_feat_table[idx].feat_num),
+		   fval ? "" : "default ",  (unsigned long long)val);
+
+	return dccp_feat_table[idx].activation_hdlr(sk, val, rx);
+}
+
+/**
+ * dccp_feat_activate  -  Activate feature value on socket
+ * @sk: fully connected DCCP socket (after handshake is complete)
+ * @feat_num: feature to activate, one of %dccp_feature_numbers
+ * @local: whether local (1) or remote (0) @feat_num is meant
+ * @fval: the value (SP or NN) to activate, or NULL to use the default value
+ * For general use this function is preferable over __dccp_feat_activate().
+ */
+static int dccp_feat_activate(struct sock *sk, u8 feat_num, bool local,
+			      dccp_feat_val const *fval)
+{
+	return __dccp_feat_activate(sk, dccp_feat_index(feat_num), local, fval);
+}
+
+/* Test for "Req'd" feature (RFC 4340, 6.4) */
+static inline int dccp_feat_must_be_understood(u8 feat_num)
+{
+	return	feat_num == DCCPF_CCID || feat_num == DCCPF_SHORT_SEQNOS ||
+		feat_num == DCCPF_SEQUENCE_WINDOW;
+}
+
+/* copy constructor, fval must not already contain allocated memory */
+static int dccp_feat_clone_sp_val(dccp_feat_val *fval, u8 const *val, u8 len)
+{
+	fval->sp.len = len;
+	if (fval->sp.len > 0) {
+		fval->sp.vec = kmemdup(val, len, gfp_any());
+		if (fval->sp.vec == NULL) {
+			fval->sp.len = 0;
+			return -ENOBUFS;
+		}
 	}
+	return 0;
+}
 
-	return agree ? 0 : DCCP_FEAT_SP_NOAGREE; /* used for mandatory opts */
+static void dccp_feat_val_destructor(u8 feat_num, dccp_feat_val *val)
+{
+	if (unlikely(val == NULL))
+		return;
+	if (dccp_feat_type(feat_num) == FEAT_SP)
+		kfree(val->sp.vec);
+	memset(val, 0, sizeof(*val));
 }
 
-static int dccp_feat_sp(struct sock *sk, u8 type, u8 feature, u8 *val, u8 len)
+static struct dccp_feat_entry *
+	      dccp_feat_clone_entry(struct dccp_feat_entry const *original)
 {
-	struct dccp_minisock *dmsk = dccp_msk(sk);
-	struct dccp_opt_pend *opt;
-	int rc = 1;
-	u8 t;
+	struct dccp_feat_entry *new;
+	u8 type = dccp_feat_type(original->feat_num);
 
-	/*
-	 * We received a CHANGE.  We gotta match it against our own preference
-	 * list.  If we got a CHANGE_R it means it's a change for us, so we need
-	 * to compare our CHANGE_L list.
-	 */
-	if (type == DCCPO_CHANGE_L)
-		t = DCCPO_CHANGE_R;
-	else
-		t = DCCPO_CHANGE_L;
+	if (type == FEAT_UNKNOWN)
+		return NULL;
 
-	/* find our preference list for this feature */
-	list_for_each_entry(opt, &dmsk->dccpms_pending, dccpop_node) {
-		if (opt->dccpop_type != t || opt->dccpop_feat != feature)
-			continue;
+	new = kmemdup(original, sizeof(struct dccp_feat_entry), gfp_any());
+	if (new == NULL)
+		return NULL;
 
-		/* find the winner from the two preference lists */
-		rc = dccp_feat_reconcile(sk, opt, val, len);
-		break;
+	if (type == FEAT_SP && dccp_feat_clone_sp_val(&new->val,
+						      original->val.sp.vec,
+						      original->val.sp.len)) {
+		kfree(new);
+		return NULL;
 	}
+	return new;
+}
 
-	/* We didn't deal with the change.  This can happen if we have no
-	 * preference list for the feature.  In fact, it just shouldn't
-	 * happen---if we understand a feature, we should have a preference list
-	 * with at least the default value.
-	 */
-	BUG_ON(rc == 1);
+static void dccp_feat_entry_destructor(struct dccp_feat_entry *entry)
+{
+	if (entry != NULL) {
+		dccp_feat_val_destructor(entry->feat_num, &entry->val);
+		kfree(entry);
+	}
+}
 
-	return rc;
+/*
+ * List management functions
+ *
+ * Feature negotiation lists rely on and maintain the following invariants:
+ * - each feat_num in the list is known, i.e. we know its type and default value
+ * - each feat_num/is_local combination is unique (old entries are overwritten)
+ * - SP values are always freshly allocated
+ * - list is sorted in increasing order of feature number (faster lookup)
+ */
+static struct dccp_feat_entry *dccp_feat_list_lookup(struct list_head *fn_list,
+						     u8 feat_num, bool is_local)
+{
+	struct dccp_feat_entry *entry;
+
+	list_for_each_entry(entry, fn_list, node)
+		if (entry->feat_num == feat_num && entry->is_local == is_local)
+			return entry;
+		else if (entry->feat_num > feat_num)
+			break;
+	return NULL;
 }
 
-static int dccp_feat_nn(struct sock *sk, u8 type, u8 feature, u8 *val, u8 len)
+/**
+ * dccp_feat_entry_new  -  Central list update routine (called by all others)
+ * @head:  list to add to
+ * @feat:  feature number
+ * @local: whether the local (1) or remote feature with number @feat is meant
+ * This is the only constructor and serves to ensure the above invariants.
+ */
+static struct dccp_feat_entry *
+	      dccp_feat_entry_new(struct list_head *head, u8 feat, bool local)
 {
-	struct dccp_opt_pend *opt;
-	struct dccp_minisock *dmsk = dccp_msk(sk);
-	u8 *copy;
-	int rc;
+	struct dccp_feat_entry *entry;
+
+	list_for_each_entry(entry, head, node)
+		if (entry->feat_num == feat && entry->is_local == local) {
+			dccp_feat_val_destructor(entry->feat_num, &entry->val);
+			return entry;
+		} else if (entry->feat_num > feat) {
+			head = &entry->node;
+			break;
+		}
 
-	/* NN features must be Change L (sec. 6.3.2) */
-	if (type != DCCPO_CHANGE_L) {
-		dccp_pr_debug("received %s for NN feature %d\n",
-				dccp_feat_typename(type), feature);
-		return -EFAULT;
+	entry = kmalloc(sizeof(*entry), gfp_any());
+	if (entry != NULL) {
+		entry->feat_num = feat;
+		entry->is_local = local;
+		list_add_tail(&entry->node, head);
 	}
+	return entry;
+}
 
-	/* XXX sanity check opt val */
+/**
+ * dccp_feat_push_change  -  Add/overwrite a Change option in the list
+ * @fn_list: feature-negotiation list to update
+ * @feat: one of %dccp_feature_numbers
+ * @local: whether local (1) or remote (0) @feat_num is meant
+ * @needs_mandatory: whether to use Mandatory feature negotiation options
+ * @fval: pointer to NN/SP value to be inserted (will be copied)
+ */
+static int dccp_feat_push_change(struct list_head *fn_list, u8 feat, u8 local,
+				 u8 mandatory, dccp_feat_val *fval)
+{
+	struct dccp_feat_entry *new = dccp_feat_entry_new(fn_list, feat, local);
 
-	/* copy option so we can confirm it */
-	opt = kzalloc(sizeof(*opt), GFP_ATOMIC);
-	if (opt == NULL)
+	if (new == NULL)
 		return -ENOMEM;
 
-	copy = kmemdup(val, len, GFP_ATOMIC);
-	if (copy == NULL) {
-		kfree(opt);
-		return -ENOMEM;
-	}
+	new->feat_num	     = feat;
+	new->is_local	     = local;
+	new->state	     = FEAT_INITIALISING;
+	new->needs_confirm   = 0;
+	new->empty_confirm   = 0;
+	new->val	     = *fval;
+	new->needs_mandatory = mandatory;
 
-	opt->dccpop_type = DCCPO_CONFIRM_R; /* NN can only confirm R */
-	opt->dccpop_feat = feature;
-	opt->dccpop_val	 = copy;
-	opt->dccpop_len	 = len;
+	return 0;
+}
 
-	/* change feature */
-	rc = dccp_feat_update(sk, type, feature, *val);
-	if (rc) {
-		kfree(opt->dccpop_val);
-		kfree(opt);
-		return rc;
-	}
+/**
+ * dccp_feat_push_confirm  -  Add a Confirm entry to the FN list
+ * @fn_list: feature-negotiation list to add to
+ * @feat: one of %dccp_feature_numbers
+ * @local: whether local (1) or remote (0) @feat_num is being confirmed
+ * @fval: pointer to NN/SP value to be inserted or NULL
+ * Returns 0 on success, a Reset code for further processing otherwise.
+ */
+static int dccp_feat_push_confirm(struct list_head *fn_list, u8 feat, u8 local,
+				  dccp_feat_val *fval)
+{
+	struct dccp_feat_entry *new = dccp_feat_entry_new(fn_list, feat, local);
 
-	dccp_feat_debug(type, feature, *copy);
+	if (new == NULL)
+		return DCCP_RESET_CODE_TOO_BUSY;
 
-	list_add_tail(&opt->dccpop_node, &dmsk->dccpms_conf);
+	new->feat_num	     = feat;
+	new->is_local	     = local;
+	new->state	     = FEAT_STABLE;	/* transition in 6.6.2 */
+	new->needs_confirm   = 1;
+	new->empty_confirm   = (fval == NULL);
+	new->val.nn	     = 0;		/* zeroes the whole structure */
+	if (!new->empty_confirm)
+		new->val     = *fval;
+	new->needs_mandatory = 0;
 
 	return 0;
 }
 
-static void dccp_feat_empty_confirm(struct dccp_minisock *dmsk,
-				    u8 type, u8 feature)
+static int dccp_push_empty_confirm(struct list_head *fn_list, u8 feat, u8 local)
 {
-	/* XXX check if other confirms for that are queued and recycle slot */
-	struct dccp_opt_pend *opt = kzalloc(sizeof(*opt), GFP_ATOMIC);
+	return dccp_feat_push_confirm(fn_list, feat, local, NULL);
+}
 
-	if (opt == NULL) {
-		/* XXX what do we do?  Ignoring should be fine.  It's a change
-		 * after all =P
-		 */
-		return;
-	}
+static inline void dccp_feat_list_pop(struct dccp_feat_entry *entry)
+{
+	list_del(&entry->node);
+	dccp_feat_entry_destructor(entry);
+}
 
-	switch (type) {
-	case DCCPO_CHANGE_L:
-		opt->dccpop_type = DCCPO_CONFIRM_R;
-		break;
-	case DCCPO_CHANGE_R:
-		opt->dccpop_type = DCCPO_CONFIRM_L;
-		break;
-	default:
-		DCCP_WARN("invalid type %d\n", type);
-		kfree(opt);
-		return;
+void dccp_feat_list_purge(struct list_head *fn_list)
+{
+	struct dccp_feat_entry *entry, *next;
+
+	list_for_each_entry_safe(entry, next, fn_list, node)
+		dccp_feat_entry_destructor(entry);
+	INIT_LIST_HEAD(fn_list);
+}
+EXPORT_SYMBOL_GPL(dccp_feat_list_purge);
+
+/* generate @to as full clone of @from - @to must not contain any nodes */
+int dccp_feat_clone_list(struct list_head const *from, struct list_head *to)
+{
+	struct dccp_feat_entry *entry, *new;
+
+	INIT_LIST_HEAD(to);
+	list_for_each_entry(entry, from, node) {
+		new = dccp_feat_clone_entry(entry);
+		if (new == NULL)
+			goto cloning_failed;
+		list_add_tail(&new->node, to);
 	}
-	opt->dccpop_feat = feature;
-	opt->dccpop_val	 = NULL;
-	opt->dccpop_len	 = 0;
+	return 0;
 
-	/* change feature */
-	dccp_pr_debug("Empty %s(%d)\n", dccp_feat_typename(type), feature);
+cloning_failed:
+	dccp_feat_list_purge(to);
+	return -ENOMEM;
+}
 
-	list_add_tail(&opt->dccpop_node, &dmsk->dccpms_conf);
+/**
+ * dccp_feat_valid_nn_length  -  Enforce length constraints on NN options
+ * Length is between 0 and %DCCP_OPTVAL_MAXLEN. Used for outgoing packets only,
+ * incoming options are accepted as long as their values are valid.
+ */
+static u8 dccp_feat_valid_nn_length(u8 feat_num)
+{
+	if (feat_num == DCCPF_ACK_RATIO)	/* RFC 4340, 11.3 and 6.6.8 */
+		return 2;
+	if (feat_num == DCCPF_SEQUENCE_WINDOW)	/* RFC 4340, 7.5.2 and 6.5  */
+		return 6;
+	return 0;
 }
 
-static void dccp_feat_flush_confirm(struct sock *sk)
+static u8 dccp_feat_is_valid_nn_val(u8 feat_num, u64 val)
 {
-	struct dccp_minisock *dmsk = dccp_msk(sk);
-	/* Check if there is anything to confirm in the first place */
-	int yes = !list_empty(&dmsk->dccpms_conf);
+	switch (feat_num) {
+	case DCCPF_ACK_RATIO:
+		return val <= DCCPF_ACK_RATIO_MAX;
+	case DCCPF_SEQUENCE_WINDOW:
+		return val >= DCCPF_SEQ_WMIN && val <= DCCPF_SEQ_WMAX;
+	}
+	return 0;	/* feature unknown - so we can't tell */
+}
 
-	if (!yes) {
-		struct dccp_opt_pend *opt;
+/* check that SP values are within the ranges defined in RFC 4340 */
+static u8 dccp_feat_is_valid_sp_val(u8 feat_num, u8 val)
+{
+	switch (feat_num) {
+	case DCCPF_CCID:
+		return val == DCCPC_CCID2 || val == DCCPC_CCID3;
+	/* Type-check Boolean feature values: */
+	case DCCPF_SHORT_SEQNOS:
+	case DCCPF_ECN_INCAPABLE:
+	case DCCPF_SEND_ACK_VECTOR:
+	case DCCPF_SEND_NDP_COUNT:
+	case DCCPF_DATA_CHECKSUM:
+	case DCCPF_SEND_LEV_RATE:
+		return val < 2;
+	case DCCPF_MIN_CSUM_COVER:
+		return val < 16;
+	}
+	return 0;			/* feature unknown */
+}
+
+static u8 dccp_feat_sp_list_ok(u8 feat_num, u8 const *sp_list, u8 sp_len)
+{
+	if (sp_list == NULL || sp_len < 1)
+		return 0;
+	while (sp_len--)
+		if (!dccp_feat_is_valid_sp_val(feat_num, *sp_list++))
+			return 0;
+	return 1;
+}
 
-		list_for_each_entry(opt, &dmsk->dccpms_pending, dccpop_node) {
-			if (opt->dccpop_conf) {
-				yes = 1;
-				break;
+/**
+ * dccp_feat_insert_opts  -  Generate FN options from current list state
+ * @skb: next sk_buff to be sent to the peer
+ * @dp: for client during handshake and general negotiation
+ * @dreq: used by the server only (all Changes/Confirms in LISTEN/RESPOND)
+ */
+int dccp_feat_insert_opts(struct dccp_sock *dp, struct dccp_request_sock *dreq,
+			  struct sk_buff *skb)
+{
+	struct list_head *fn = dreq ? &dreq->dreq_featneg : &dp->dccps_featneg;
+	struct dccp_feat_entry *pos, *next;
+	u8 opt, type, len, *ptr, nn_in_nbo[DCCP_OPTVAL_MAXLEN];
+	bool rpt;
+
+	/* put entries into @skb in the order they appear in the list */
+	list_for_each_entry_safe_reverse(pos, next, fn, node) {
+		opt  = dccp_feat_genopt(pos);
+		type = dccp_feat_type(pos->feat_num);
+		rpt  = false;
+
+		if (pos->empty_confirm) {
+			len = 0;
+			ptr = NULL;
+		} else {
+			if (type == FEAT_SP) {
+				len = pos->val.sp.len;
+				ptr = pos->val.sp.vec;
+				rpt = pos->needs_confirm;
+			} else if (type == FEAT_NN) {
+				len = dccp_feat_valid_nn_length(pos->feat_num);
+				ptr = nn_in_nbo;
+				dccp_encode_value_var(pos->val.nn, ptr, len);
+			} else {
+				DCCP_BUG("unknown feature %u", pos->feat_num);
+				return -1;
 			}
 		}
+		dccp_feat_print_opt(opt, pos->feat_num, ptr, len, 0);
+
+		if (dccp_insert_fn_opt(skb, opt, pos->feat_num, ptr, len, rpt))
+			return -1;
+		if (pos->needs_mandatory && dccp_insert_option_mandatory(skb))
+			return -1;
+		/*
+		 * Enter CHANGING after transmitting the Change option (6.6.2).
+		 */
+		if (pos->state == FEAT_INITIALISING)
+			pos->state = FEAT_CHANGING;
 	}
+	return 0;
+}
 
-	if (!yes)
-		return;
+/**
+ * __feat_register_nn  -  Register new NN value on socket
+ * @fn: feature-negotiation list to register with
+ * @feat: an NN feature from %dccp_feature_numbers
+ * @mandatory: use Mandatory option if 1
+ * @nn_val: value to register (restricted to 4 bytes)
+ * Note that NN features are local by definition (RFC 4340, 6.3.2).
+ */
+static int __feat_register_nn(struct list_head *fn, u8 feat,
+			      u8 mandatory, u64 nn_val)
+{
+	dccp_feat_val fval = { .nn = nn_val };
 
-	/* OK there is something to confirm... */
-	/* XXX check if packet is in flight?  Send delayed ack?? */
-	if (sk->sk_state == DCCP_OPEN)
-		dccp_send_ack(sk);
+	if (dccp_feat_type(feat) != FEAT_NN ||
+	    !dccp_feat_is_valid_nn_val(feat, nn_val))
+		return -EINVAL;
+
+	/* Don't bother with default values, they will be activated anyway. */
+	if (nn_val - (u64)dccp_feat_default_value(feat) == 0)
+		return 0;
+
+	return dccp_feat_push_change(fn, feat, 1, mandatory, &fval);
 }
 
-int dccp_feat_change_recv(struct sock *sk, u8 type, u8 feature, u8 *val, u8 len)
+/**
+ * __feat_register_sp  -  Register new SP value/list on socket
+ * @fn: feature-negotiation list to register with
+ * @feat: an SP feature from %dccp_feature_numbers
+ * @is_local: whether the local (1) or the remote (0) @feat is meant
+ * @mandatory: use Mandatory option if 1
+ * @sp_val: SP value followed by optional preference list
+ * @sp_len: length of @sp_val in bytes
+ */
+static int __feat_register_sp(struct list_head *fn, u8 feat, u8 is_local,
+			      u8 mandatory, u8 const *sp_val, u8 sp_len)
 {
-	int rc;
+	dccp_feat_val fval;
 
-	dccp_feat_debug(type, feature, *val);
+	if (dccp_feat_type(feat) != FEAT_SP ||
+	    !dccp_feat_sp_list_ok(feat, sp_val, sp_len))
+		return -EINVAL;
 
-	/* figure out if it's SP or NN feature */
-	switch (feature) {
-	/* deal with SP features */
-	case DCCPF_CCID:
-		rc = dccp_feat_sp(sk, type, feature, val, len);
-		break;
+	/* Avoid negotiating alien CCIDs by only advertising supported ones */
+	if (feat == DCCPF_CCID && !ccid_support_check(sp_val, sp_len))
+		return -EOPNOTSUPP;
 
-	/* deal with NN features */
-	case DCCPF_ACK_RATIO:
-		rc = dccp_feat_nn(sk, type, feature, val, len);
-		break;
+	if (dccp_feat_clone_sp_val(&fval, sp_val, sp_len))
+		return -ENOMEM;
 
-	/* XXX implement other features */
-	default:
-		dccp_pr_debug("UNIMPLEMENTED: not handling %s(%d, ...)\n",
-			      dccp_feat_typename(type), feature);
-		rc = -EFAULT;
-		break;
+	return dccp_feat_push_change(fn, feat, is_local, mandatory, &fval);
+}
+
+/**
+ * dccp_feat_register_sp  -  Register requests to change SP feature values
+ * @sk: client or listening socket
+ * @feat: one of %dccp_feature_numbers
+ * @is_local: whether the local (1) or remote (0) @feat is meant
+ * @list: array of preferred values, in descending order of preference
+ * @len: length of @list in bytes
+ */
+int dccp_feat_register_sp(struct sock *sk, u8 feat, u8 is_local,
+			  u8 const *list, u8 len)
+{	 /* any changes must be registered before establishing the connection */
+	if (sk->sk_state != DCCP_CLOSED)
+		return -EISCONN;
+	if (dccp_feat_type(feat) != FEAT_SP)
+		return -EINVAL;
+	return __feat_register_sp(&dccp_sk(sk)->dccps_featneg, feat, is_local,
+				  0, list, len);
+}
+
+/* Analogous to dccp_feat_register_sp(), but for non-negotiable values */
+int dccp_feat_register_nn(struct sock *sk, u8 feat, u64 val)
+{
+	/* any changes must be registered before establishing the connection */
+	if (sk->sk_state != DCCP_CLOSED)
+		return -EISCONN;
+	if (dccp_feat_type(feat) != FEAT_NN)
+		return -EINVAL;
+	return __feat_register_nn(&dccp_sk(sk)->dccps_featneg, feat, 0, val);
+}
+
+/**
+ * dccp_feat_signal_nn_change  -  Update NN values for an established connection
+ * @sk: DCCP socket of an established connection
+ * @feat: NN feature number from %dccp_feature_numbers
+ * @nn_val: the new value to use
+ * This function is used to communicate NN updates out-of-band. The difference
+ * to feature negotiation during connection setup is that values are activated
+ * immediately after validation, i.e. we don't wait for the Confirm: either the
+ * value is accepted by the peer (and then the waiting is futile), or it is not
+ * (Reset or empty Confirm). We don't accept empty Confirms - transmitted values
+ * are validated, and the peer "MUST accept any valid value" (RFC 4340, 6.3.2).
+ */
+int dccp_feat_signal_nn_change(struct sock *sk, u8 feat, u64 nn_val)
+{
+	struct list_head *fn = &dccp_sk(sk)->dccps_featneg;
+	dccp_feat_val fval = { .nn = nn_val };
+	struct dccp_feat_entry *entry;
+
+	if (sk->sk_state != DCCP_OPEN && sk->sk_state != DCCP_PARTOPEN)
+		return 0;
+
+	if (dccp_feat_type(feat) != FEAT_NN ||
+	    !dccp_feat_is_valid_nn_val(feat, nn_val))
+		return -EINVAL;
+
+	entry = dccp_feat_list_lookup(fn, feat, 1);
+	if (entry != NULL) {
+		dccp_pr_debug("Ignoring %llu, entry %llu exists in state %s\n",
+			      (unsigned long long)nn_val,
+			      (unsigned long long)entry->val.nn,
+			      dccp_feat_sname[entry->state]);
+		return 0;
 	}
 
-	/* check if there were problems changing features */
-	if (rc) {
-		/* If we don't agree on SP, we sent a confirm for old value.
-		 * However we propagate rc to caller in case option was
-		 * mandatory
+	if (dccp_feat_activate(sk, feat, 1, &fval))
+		return -EADV;
+
+	inet_csk_schedule_ack(sk);
+	return dccp_feat_push_change(fn, feat, 1, 0, &fval);
+}
+EXPORT_SYMBOL_GPL(dccp_feat_signal_nn_change);
+
+/*
+ *	Tracking features whose value depend on the choice of CCID
+ *
+ * This is designed with an extension in mind so that a list walk could be done
+ * before activating any features. However, the existing framework was found to
+ * work satisfactorily up until now, the automatic verification is left open.
+ * When adding new CCIDs, add a corresponding dependency table here.
+ */
+static const struct ccid_dependency *dccp_feat_ccid_deps(u8 ccid, bool is_local)
+{
+	static const struct ccid_dependency ccid2_dependencies[2][2] = {
+		/*
+		 * CCID2 mandates Ack Vectors (RFC 4341, 4.): as CCID is a TX
+		 * feature and Send Ack Vector is an RX feature, `is_local'
+		 * needs to be reversed.
 		 */
-		if (rc != DCCP_FEAT_SP_NOAGREE)
-			dccp_feat_empty_confirm(dccp_msk(sk), type, feature);
+		{	/* Dependencies of the receiver-side (remote) CCID2 */
+			{
+				.dependent_feat	= DCCPF_SEND_ACK_VECTOR,
+				.is_local	= true,
+				.is_mandatory	= true,
+				.val		= 1
+			},
+			{ 0, 0, 0, 0 }
+		},
+		{	/* Dependencies of the sender-side (local) CCID2 */
+			{
+				.dependent_feat	= DCCPF_SEND_ACK_VECTOR,
+				.is_local	= false,
+				.is_mandatory	= true,
+				.val		= 1
+			},
+			{ 0, 0, 0, 0 }
+		}
+	};
+	static const struct ccid_dependency ccid3_dependencies[2][5] = {
+		{	/*
+			 * Dependencies of the receiver-side CCID3
+			 */
+			{	/* locally disable Ack Vectors */
+				.dependent_feat	= DCCPF_SEND_ACK_VECTOR,
+				.is_local	= true,
+				.is_mandatory	= false,
+				.val		= 0
+			},
+			{	/* see below why Send Loss Event Rate is on */
+				.dependent_feat	= DCCPF_SEND_LEV_RATE,
+				.is_local	= true,
+				.is_mandatory	= true,
+				.val		= 1
+			},
+			{	/* NDP Count is needed as per RFC 4342, 6.1.1 */
+				.dependent_feat	= DCCPF_SEND_NDP_COUNT,
+				.is_local	= false,
+				.is_mandatory	= true,
+				.val		= 1
+			},
+			{ 0, 0, 0, 0 },
+		},
+		{	/*
+			 * CCID3 at the TX side: we request that the HC-receiver
+			 * will not send Ack Vectors (they will be ignored, so
+			 * Mandatory is not set); we enable Send Loss Event Rate
+			 * (Mandatory since the implementation does not support
+			 * the Loss Intervals option of RFC 4342, 8.6).
+			 * The last two options are for peer's information only.
+			*/
+			{
+				.dependent_feat	= DCCPF_SEND_ACK_VECTOR,
+				.is_local	= false,
+				.is_mandatory	= false,
+				.val		= 0
+			},
+			{
+				.dependent_feat	= DCCPF_SEND_LEV_RATE,
+				.is_local	= false,
+				.is_mandatory	= true,
+				.val		= 1
+			},
+			{	/* this CCID does not support Ack Ratio */
+				.dependent_feat	= DCCPF_ACK_RATIO,
+				.is_local	= true,
+				.is_mandatory	= false,
+				.val		= 0
+			},
+			{	/* tell receiver we are sending NDP counts */
+				.dependent_feat	= DCCPF_SEND_NDP_COUNT,
+				.is_local	= true,
+				.is_mandatory	= false,
+				.val		= 1
+			},
+			{ 0, 0, 0, 0 }
+		}
+	};
+	switch (ccid) {
+	case DCCPC_CCID2:
+		return ccid2_dependencies[is_local];
+	case DCCPC_CCID3:
+		return ccid3_dependencies[is_local];
+	default:
+		return NULL;
 	}
+}
 
-	/* generate the confirm [if required] */
-	dccp_feat_flush_confirm(sk);
-
+/**
+ * dccp_feat_propagate_ccid - Resolve dependencies of features on choice of CCID
+ * @fn: feature-negotiation list to update
+ * @id: CCID number to track
+ * @is_local: whether TX CCID (1) or RX CCID (0) is meant
+ * This function needs to be called after registering all other features.
+ */
+static int dccp_feat_propagate_ccid(struct list_head *fn, u8 id, bool is_local)
+{
+	const struct ccid_dependency *table = dccp_feat_ccid_deps(id, is_local);
+	int i, rc = (table == NULL);
+
+	for (i = 0; rc == 0 && table[i].dependent_feat != DCCPF_RESERVED; i++)
+		if (dccp_feat_type(table[i].dependent_feat) == FEAT_SP)
+			rc = __feat_register_sp(fn, table[i].dependent_feat,
+						    table[i].is_local,
+						    table[i].is_mandatory,
+						    &table[i].val, 1);
+		else
+			rc = __feat_register_nn(fn, table[i].dependent_feat,
+						    table[i].is_mandatory,
+						    table[i].val);
 	return rc;
 }
 
-EXPORT_SYMBOL_GPL(dccp_feat_change_recv);
+/**
+ * dccp_feat_finalise_settings  -  Finalise settings before starting negotiation
+ * @dp: client or listening socket (settings will be inherited)
+ * This is called after all registrations (socket initialisation, sysctls, and
+ * sockopt calls), and before sending the first packet containing Change options
+ * (ie. client-Request or server-Response), to ensure internal consistency.
+ */
+int dccp_feat_finalise_settings(struct dccp_sock *dp)
+{
+	struct list_head *fn = &dp->dccps_featneg;
+	struct dccp_feat_entry *entry;
+	int i = 2, ccids[2] = { -1, -1 };
+
+	/*
+	 * Propagating CCIDs:
+	 * 1) not useful to propagate CCID settings if this host advertises more
+	 *    than one CCID: the choice of CCID  may still change - if this is
+	 *    the client, or if this is the server and the client sends
+	 *    singleton CCID values.
+	 * 2) since is that propagate_ccid changes the list, we defer changing
+	 *    the sorted list until after the traversal.
+	 */
+	list_for_each_entry(entry, fn, node)
+		if (entry->feat_num == DCCPF_CCID && entry->val.sp.len == 1)
+			ccids[entry->is_local] = entry->val.sp.vec[0];
+	while (i--)
+		if (ccids[i] > 0 && dccp_feat_propagate_ccid(fn, ccids[i], i))
+			return -1;
+	dccp_feat_print_fnlist(fn);
+	return 0;
+}
 
-int dccp_feat_confirm_recv(struct sock *sk, u8 type, u8 feature,
-			   u8 *val, u8 len)
+/**
+ * dccp_feat_server_ccid_dependencies  -  Resolve CCID-dependent features
+ * It is the server which resolves the dependencies once the CCID has been
+ * fully negotiated. If no CCID has been negotiated, it uses the default CCID.
+ */
+int dccp_feat_server_ccid_dependencies(struct dccp_request_sock *dreq)
 {
-	u8 t;
-	struct dccp_opt_pend *opt;
-	struct dccp_minisock *dmsk = dccp_msk(sk);
-	int found = 0;
-	int all_confirmed = 1;
+	struct list_head *fn = &dreq->dreq_featneg;
+	struct dccp_feat_entry *entry;
+	u8 is_local, ccid;
 
-	dccp_feat_debug(type, feature, *val);
+	for (is_local = 0; is_local <= 1; is_local++) {
+		entry = dccp_feat_list_lookup(fn, DCCPF_CCID, is_local);
 
-	/* locate our change request */
-	switch (type) {
-	case DCCPO_CONFIRM_L: t = DCCPO_CHANGE_R; break;
-	case DCCPO_CONFIRM_R: t = DCCPO_CHANGE_L; break;
-	default:	      DCCP_WARN("invalid type %d\n", type);
-			      return 1;
+		if (entry != NULL && !entry->empty_confirm)
+			ccid = entry->val.sp.vec[0];
+		else
+			ccid = dccp_feat_default_value(DCCPF_CCID);
 
+		if (dccp_feat_propagate_ccid(fn, ccid, is_local))
+			return -1;
 	}
-	/* XXX sanity check feature value */
+	return 0;
+}
 
-	list_for_each_entry(opt, &dmsk->dccpms_pending, dccpop_node) {
-		if (!opt->dccpop_conf && opt->dccpop_type == t &&
-		    opt->dccpop_feat == feature) {
-			found = 1;
-			dccp_pr_debug("feature %d found\n", opt->dccpop_feat);
+/* Select the first entry in @servlist that also occurs in @clilist (6.3.1) */
+static int dccp_feat_preflist_match(u8 *servlist, u8 slen, u8 *clilist, u8 clen)
+{
+	u8 c, s;
 
-			/* XXX do sanity check */
+	for (s = 0; s < slen; s++)
+		for (c = 0; c < clen; c++)
+			if (servlist[s] == clilist[c])
+				return servlist[s];
+	return -1;
+}
 
-			opt->dccpop_conf = 1;
+/**
+ * dccp_feat_prefer  -  Move preferred entry to the start of array
+ * Reorder the @array_len elements in @array so that @preferred_value comes
+ * first. Returns >0 to indicate that @preferred_value does occur in @array.
+ */
+static u8 dccp_feat_prefer(u8 preferred_value, u8 *array, u8 array_len)
+{
+	u8 i, does_occur = 0;
 
-			/* We got a confirmation---change the option */
-			dccp_feat_update(sk, opt->dccpop_type,
-					 opt->dccpop_feat, *val);
+	if (array != NULL) {
+		for (i = 0; i < array_len; i++)
+			if (array[i] == preferred_value) {
+				array[i] = array[0];
+				does_occur++;
+			}
+		if (does_occur)
+			array[0] = preferred_value;
+	}
+	return does_occur;
+}
 
-			/* XXX check the return value of dccp_feat_update */
-			break;
-		}
+/**
+ * dccp_feat_reconcile  -  Reconcile SP preference lists
+ *  @fval: SP list to reconcile into
+ *  @arr: received SP preference list
+ *  @len: length of @arr in bytes
+ *  @is_server: whether this side is the server (and @fv is the server's list)
+ *  @reorder: whether to reorder the list in @fv after reconciling with @arr
+ * When successful, > 0 is returned and the reconciled list is in @fval.
+ * A value of 0 means that negotiation failed (no shared entry).
+ */
+static int dccp_feat_reconcile(dccp_feat_val *fv, u8 *arr, u8 len,
+			       bool is_server, bool reorder)
+{
+	int rc;
 
-		if (!opt->dccpop_conf)
-			all_confirmed = 0;
+	if (!fv->sp.vec || !arr) {
+		DCCP_CRIT("NULL feature value or array");
+		return 0;
 	}
 
-	/* fix re-transmit timer */
-	/* XXX gotta make sure that no option negotiation occurs during
-	 * connection shutdown.  Consider that the CLOSEREQ is sent and timer is
-	 * on.  if all options are confirmed it might kill timer which should
-	 * remain alive until close is received.
-	 */
-	if (all_confirmed) {
-		dccp_pr_debug("clear feat negotiation timer %p\n", sk);
-		inet_csk_clear_xmit_timer(sk, ICSK_TIME_RETRANS);
-	}
+	if (is_server)
+		rc = dccp_feat_preflist_match(fv->sp.vec, fv->sp.len, arr, len);
+	else
+		rc = dccp_feat_preflist_match(arr, len, fv->sp.vec, fv->sp.len);
 
-	if (!found)
-		dccp_pr_debug("%s(%d, ...) never requested\n",
-			      dccp_feat_typename(type), feature);
-	return 0;
-}
+	if (!reorder)
+		return rc;
+	if (rc < 0)
+		return 0;
 
-EXPORT_SYMBOL_GPL(dccp_feat_confirm_recv);
+	/*
+	 * Reorder list: used for activating features and in dccp_insert_fn_opt.
+	 */
+	return dccp_feat_prefer(rc, fv->sp.vec, fv->sp.len);
+}
 
-void dccp_feat_clean(struct dccp_minisock *dmsk)
+/**
+ * dccp_feat_change_recv  -  Process incoming ChangeL/R options
+ * @fn: feature-negotiation list to update
+ * @is_mandatory: whether the Change was preceded by a Mandatory option
+ * @opt: %DCCPO_CHANGE_L or %DCCPO_CHANGE_R
+ * @feat: one of %dccp_feature_numbers
+ * @val: NN value or SP value/preference list
+ * @len: length of @val in bytes
+ * @server: whether this node is the server (1) or the client (0)
+ */
+static u8 dccp_feat_change_recv(struct list_head *fn, u8 is_mandatory, u8 opt,
+				u8 feat, u8 *val, u8 len, const bool server)
 {
-	struct dccp_opt_pend *opt, *next;
+	u8 defval, type = dccp_feat_type(feat);
+	const bool local = (opt == DCCPO_CHANGE_R);
+	struct dccp_feat_entry *entry;
+	dccp_feat_val fval;
+
+	if (len == 0 || type == FEAT_UNKNOWN)		/* 6.1 and 6.6.8 */
+		goto unknown_feature_or_value;
+
+	dccp_feat_print_opt(opt, feat, val, len, is_mandatory);
+
+	/*
+	 *	Negotiation of NN features: Change R is invalid, so there is no
+	 *	simultaneous negotiation; hence we do not look up in the list.
+	 */
+	if (type == FEAT_NN) {
+		if (local || len > sizeof(fval.nn))
+			goto unknown_feature_or_value;
 
-	list_for_each_entry_safe(opt, next, &dmsk->dccpms_pending,
-				 dccpop_node) {
-		BUG_ON(opt->dccpop_val == NULL);
-		kfree(opt->dccpop_val);
+		/* 6.3.2: "The feature remote MUST accept any valid value..." */
+		fval.nn = dccp_decode_value_var(val, len);
+		if (!dccp_feat_is_valid_nn_val(feat, fval.nn))
+			goto unknown_feature_or_value;
 
-		if (opt->dccpop_sc != NULL) {
-			BUG_ON(opt->dccpop_sc->dccpoc_val == NULL);
-			kfree(opt->dccpop_sc->dccpoc_val);
-			kfree(opt->dccpop_sc);
+		return dccp_feat_push_confirm(fn, feat, local, &fval);
+	}
+
+	/*
+	 *	Unidirectional/simultaneous negotiation of SP features (6.3.1)
+	 */
+	entry = dccp_feat_list_lookup(fn, feat, local);
+	if (entry == NULL) {
+		/*
+		 * No particular preferences have been registered. We deal with
+		 * this situation by assuming that all valid values are equally
+		 * acceptable, and apply the following checks:
+		 * - if the peer's list is a singleton, we accept a valid value;
+		 * - if we are the server, we first try to see if the peer (the
+		 *   client) advertises the default value. If yes, we use it,
+		 *   otherwise we accept the preferred value;
+		 * - else if we are the client, we use the first list element.
+		 */
+		if (dccp_feat_clone_sp_val(&fval, val, 1))
+			return DCCP_RESET_CODE_TOO_BUSY;
+
+		if (len > 1 && server) {
+			defval = dccp_feat_default_value(feat);
+			if (dccp_feat_preflist_match(&defval, 1, val, len) > -1)
+				fval.sp.vec[0] = defval;
+		} else if (!dccp_feat_is_valid_sp_val(feat, fval.sp.vec[0])) {
+			kfree(fval.sp.vec);
+			goto unknown_feature_or_value;
+		}
+
+		/* Treat unsupported CCIDs like invalid values */
+		if (feat == DCCPF_CCID && !ccid_support_check(fval.sp.vec, 1)) {
+			kfree(fval.sp.vec);
+			goto not_valid_or_not_known;
 		}
 
-		kfree(opt);
+		return dccp_feat_push_confirm(fn, feat, local, &fval);
+
+	} else if (entry->state == FEAT_UNSTABLE) {	/* 6.6.2 */
+		return 0;
 	}
-	INIT_LIST_HEAD(&dmsk->dccpms_pending);
 
-	list_for_each_entry_safe(opt, next, &dmsk->dccpms_conf, dccpop_node) {
-		BUG_ON(opt == NULL);
-		if (opt->dccpop_val != NULL)
-			kfree(opt->dccpop_val);
-		kfree(opt);
+	if (dccp_feat_reconcile(&entry->val, val, len, server, true)) {
+		entry->empty_confirm = 0;
+	} else if (is_mandatory) {
+		return DCCP_RESET_CODE_MANDATORY_ERROR;
+	} else if (entry->state == FEAT_INITIALISING) {
+		/*
+		 * Failed simultaneous negotiation (server only): try to `save'
+		 * the connection by checking whether entry contains the default
+		 * value for @feat. If yes, send an empty Confirm to signal that
+		 * the received Change was not understood - which implies using
+		 * the default value.
+		 * If this also fails, we use Reset as the last resort.
+		 */
+		WARN_ON(!server);
+		defval = dccp_feat_default_value(feat);
+		if (!dccp_feat_reconcile(&entry->val, &defval, 1, server, true))
+			return DCCP_RESET_CODE_OPTION_ERROR;
+		entry->empty_confirm = 1;
 	}
-	INIT_LIST_HEAD(&dmsk->dccpms_conf);
-}
+	entry->needs_confirm   = 1;
+	entry->needs_mandatory = 0;
+	entry->state	       = FEAT_STABLE;
+	return 0;
 
-EXPORT_SYMBOL_GPL(dccp_feat_clean);
+unknown_feature_or_value:
+	if (!is_mandatory)
+		return dccp_push_empty_confirm(fn, feat, local);
 
-/* this is to be called only when a listening sock creates its child.  It is
- * assumed by the function---the confirm is not duplicated, but rather it is
- * "passed on".
+not_valid_or_not_known:
+	return is_mandatory ? DCCP_RESET_CODE_MANDATORY_ERROR
+			    : DCCP_RESET_CODE_OPTION_ERROR;
+}
+
+/**
+ * dccp_feat_confirm_recv  -  Process received Confirm options
+ * @fn: feature-negotiation list to update
+ * @is_mandatory: whether @opt was preceded by a Mandatory option
+ * @opt: %DCCPO_CONFIRM_L or %DCCPO_CONFIRM_R
+ * @feat: one of %dccp_feature_numbers
+ * @val: NN value or SP value/preference list
+ * @len: length of @val in bytes
+ * @server: whether this node is server (1) or client (0)
  */
-int dccp_feat_clone(struct sock *oldsk, struct sock *newsk)
+static u8 dccp_feat_confirm_recv(struct list_head *fn, u8 is_mandatory, u8 opt,
+				 u8 feat, u8 *val, u8 len, const bool server)
 {
-	struct dccp_minisock *olddmsk = dccp_msk(oldsk);
-	struct dccp_minisock *newdmsk = dccp_msk(newsk);
-	struct dccp_opt_pend *opt;
-	int rc = 0;
+	u8 *plist, plen, type = dccp_feat_type(feat);
+	const bool local = (opt == DCCPO_CONFIRM_R);
+	struct dccp_feat_entry *entry = dccp_feat_list_lookup(fn, feat, local);
 
-	INIT_LIST_HEAD(&newdmsk->dccpms_pending);
-	INIT_LIST_HEAD(&newdmsk->dccpms_conf);
+	dccp_feat_print_opt(opt, feat, val, len, is_mandatory);
 
-	list_for_each_entry(opt, &olddmsk->dccpms_pending, dccpop_node) {
-		struct dccp_opt_pend *newopt;
-		/* copy the value of the option */
-		u8 *val = kmemdup(opt->dccpop_val, opt->dccpop_len, GFP_ATOMIC);
+	if (entry == NULL) {	/* nothing queued: ignore or handle error */
+		if (is_mandatory && type == FEAT_UNKNOWN)
+			return DCCP_RESET_CODE_MANDATORY_ERROR;
 
-		if (val == NULL)
-			goto out_clean;
-
-		newopt = kmemdup(opt, sizeof(*newopt), GFP_ATOMIC);
-		if (newopt == NULL) {
-			kfree(val);
-			goto out_clean;
-		}
+		if (!local && type == FEAT_NN)		/* 6.3.2 */
+			goto confirmation_failed;
+		return 0;
+	}
 
-		/* insert the option */
-		newopt->dccpop_val = val;
-		list_add_tail(&newopt->dccpop_node, &newdmsk->dccpms_pending);
+	if (entry->state != FEAT_CHANGING)		/* 6.6.2 */
+		return 0;
 
-		/* XXX what happens with backlogs and multiple connections at
-		 * once...
+	if (len == 0) {
+		if (dccp_feat_must_be_understood(feat))	/* 6.6.7 */
+			goto confirmation_failed;
+		/*
+		 * Empty Confirm during connection setup: this means reverting
+		 * to the `old' value, which in this case is the default. Since
+		 * we handle default values automatically when no other values
+		 * have been set, we revert to the old value by removing this
+		 * entry from the list.
 		 */
-		/* the master socket no longer needs to worry about confirms */
-		opt->dccpop_sc = NULL; /* it's not a memleak---new socket has it */
+		dccp_feat_list_pop(entry);
+		return 0;
+	}
+
+	if (type == FEAT_NN) {
+		if (len > sizeof(entry->val.nn))
+			goto confirmation_failed;
+
+		if (entry->val.nn == dccp_decode_value_var(val, len))
+			goto confirmation_succeeded;
+
+		DCCP_WARN("Bogus Confirm for non-existing value\n");
+		goto confirmation_failed;
+	}
 
-		/* reset state for a new socket */
-		opt->dccpop_conf = 0;
+	/*
+	 * Parsing SP Confirms: the first element of @val is the preferred
+	 * SP value which the peer confirms, the remainder depends on @len.
+	 * Note that only the confirmed value need to be a valid SP value.
+	 */
+	if (!dccp_feat_is_valid_sp_val(feat, *val))
+		goto confirmation_failed;
+
+	if (len == 1) {		/* peer didn't supply a preference list */
+		plist = val;
+		plen  = len;
+	} else {		/* preferred value + preference list */
+		plist = val + 1;
+		plen  = len - 1;
 	}
 
-	/* XXX not doing anything about the conf queue */
+	/* Check whether the peer got the reconciliation right (6.6.8) */
+	if (dccp_feat_reconcile(&entry->val, plist, plen, server, 0) != *val) {
+		DCCP_WARN("Confirm selected the wrong value %u\n", *val);
+		return DCCP_RESET_CODE_OPTION_ERROR;
+	}
+	entry->val.sp.vec[0] = *val;
 
-out:
-	return rc;
+confirmation_succeeded:
+	entry->state = FEAT_STABLE;
+	return 0;
 
-out_clean:
-	dccp_feat_clean(newdmsk);
-	rc = -ENOMEM;
-	goto out;
+confirmation_failed:
+	DCCP_WARN("Confirmation failed\n");
+	return is_mandatory ? DCCP_RESET_CODE_MANDATORY_ERROR
+			    : DCCP_RESET_CODE_OPTION_ERROR;
 }
 
-EXPORT_SYMBOL_GPL(dccp_feat_clone);
+/**
+ * dccp_feat_handle_nn_established  -  Fast-path reception of NN options
+ * @sk:		socket of an established DCCP connection
+ * @mandatory:	whether @opt was preceded by a Mandatory option
+ * @opt:	%DCCPO_CHANGE_L | %DCCPO_CONFIRM_R (NN only)
+ * @feat:	NN number, one of %dccp_feature_numbers
+ * @val:	NN value
+ * @len:	length of @val in bytes
+ * This function combines the functionality of change_recv/confirm_recv, with
+ * the following differences (reset codes are the same):
+ *    - cleanup after receiving the Confirm;
+ *    - values are directly activated after successful parsing;
+ *    - deliberately restricted to NN features.
+ * The restriction to NN features is essential since SP features can have non-
+ * predictable outcomes (depending on the remote configuration), and are inter-
+ * dependent (CCIDs for instance cause further dependencies).
+ */
+static u8 dccp_feat_handle_nn_established(struct sock *sk, u8 mandatory, u8 opt,
+					  u8 feat, u8 *val, u8 len)
+{
+	struct list_head *fn = &dccp_sk(sk)->dccps_featneg;
+	const bool local = (opt == DCCPO_CONFIRM_R);
+	struct dccp_feat_entry *entry;
+	u8 type = dccp_feat_type(feat);
+	dccp_feat_val fval;
+
+	dccp_feat_print_opt(opt, feat, val, len, mandatory);
+
+	/* Ignore non-mandatory unknown and non-NN features */
+	if (type == FEAT_UNKNOWN) {
+		if (local && !mandatory)
+			return 0;
+		goto fast_path_unknown;
+	} else if (type != FEAT_NN) {
+		return 0;
+	}
+
+	/*
+	 * We don't accept empty Confirms, since in fast-path feature
+	 * negotiation the values are enabled immediately after sending
+	 * the Change option.
+	 * Empty Changes on the other hand are invalid (RFC 4340, 6.1).
+	 */
+	if (len == 0 || len > sizeof(fval.nn))
+		goto fast_path_unknown;
+
+	if (opt == DCCPO_CHANGE_L) {
+		fval.nn = dccp_decode_value_var(val, len);
+		if (!dccp_feat_is_valid_nn_val(feat, fval.nn))
+			goto fast_path_unknown;
+
+		if (dccp_feat_push_confirm(fn, feat, local, &fval) ||
+		    dccp_feat_activate(sk, feat, local, &fval))
+			return DCCP_RESET_CODE_TOO_BUSY;
+
+		/* set the `Ack Pending' flag to piggyback a Confirm */
+		inet_csk_schedule_ack(sk);
+
+	} else if (opt == DCCPO_CONFIRM_R) {
+		entry = dccp_feat_list_lookup(fn, feat, local);
+		if (entry == NULL || entry->state != FEAT_CHANGING)
+			return 0;
+
+		fval.nn = dccp_decode_value_var(val, len);
+		if (fval.nn != entry->val.nn) {
+			DCCP_WARN("Bogus Confirm for non-existing value\n");
+			goto fast_path_failed;
+		}
+
+		/* It has been confirmed - so remove the entry */
+		dccp_feat_list_pop(entry);
+
+	} else {
+		DCCP_WARN("Received illegal option %u\n", opt);
+		goto fast_path_failed;
+	}
+	return 0;
+
+fast_path_unknown:
+	if (!mandatory)
+		return dccp_push_empty_confirm(fn, feat, local);
+
+fast_path_failed:
+	return mandatory ? DCCP_RESET_CODE_MANDATORY_ERROR
+			 : DCCP_RESET_CODE_OPTION_ERROR;
+}
 
-static int __dccp_feat_init(struct dccp_minisock *dmsk, u8 type, u8 feat,
-			    u8 *val, u8 len)
+/**
+ * dccp_feat_parse_options  -  Process Feature-Negotiation Options
+ * @sk: for general use and used by the client during connection setup
+ * @dreq: used by the server during connection setup
+ * @mandatory: whether @opt was preceded by a Mandatory option
+ * @opt: %DCCPO_CHANGE_L | %DCCPO_CHANGE_R | %DCCPO_CONFIRM_L | %DCCPO_CONFIRM_R
+ * @feat: one of %dccp_feature_numbers
+ * @val: value contents of @opt
+ * @len: length of @val in bytes
+ * Returns 0 on success, a Reset code for ending the connection otherwise.
+ */
+int dccp_feat_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
+			    u8 mandatory, u8 opt, u8 feat, u8 *val, u8 len)
 {
-	int rc = -ENOMEM;
-	u8 *copy = kmemdup(val, len, GFP_KERNEL);
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct list_head *fn = dreq ? &dreq->dreq_featneg : &dp->dccps_featneg;
+	bool server = false;
 
-	if (copy != NULL) {
-		rc = dccp_feat_change(dmsk, type, feat, copy, len, GFP_KERNEL);
-		if (rc)
-			kfree(copy);
+	switch (sk->sk_state) {
+	/*
+	 *	Negotiation during connection setup
+	 */
+	case DCCP_LISTEN:
+		server = true;			/* fall through */
+	case DCCP_REQUESTING:
+		switch (opt) {
+		case DCCPO_CHANGE_L:
+		case DCCPO_CHANGE_R:
+			return dccp_feat_change_recv(fn, mandatory, opt, feat,
+						     val, len, server);
+		case DCCPO_CONFIRM_R:
+		case DCCPO_CONFIRM_L:
+			return dccp_feat_confirm_recv(fn, mandatory, opt, feat,
+						      val, len, server);
+		}
+		break;
+	/*
+	 *	Support for exchanging NN options on an established connection
+	 *	This is currently restricted to Ack Ratio (RFC 4341, 6.1.2)
+	 */
+	case DCCP_OPEN:
+	case DCCP_PARTOPEN:
+		return dccp_feat_handle_nn_established(sk, mandatory, opt, feat,
+						       val, len);
 	}
-	return rc;
+	return 0;	/* ignore FN options in all other states */
 }
 
-int dccp_feat_init(struct dccp_minisock *dmsk)
+/**
+ * dccp_feat_init  -  Seed feature negotiation with host-specific defaults
+ * This initialises global defaults, depending on the value of the sysctls.
+ * These can later be overridden by registering changes via setsockopt calls.
+ * The last link in the chain is finalise_settings, to make sure that between
+ * here and the start of actual feature negotiation no inconsistencies enter.
+ *
+ * All features not appearing below use either defaults or are otherwise
+ * later adjusted through dccp_feat_finalise_settings().
+ */
+int dccp_feat_init(struct sock *sk)
 {
+	struct list_head *fn = &dccp_sk(sk)->dccps_featneg;
+	u8 on = 1, off = 0;
 	int rc;
+	struct {
+		u8 *val;
+		u8 len;
+	} tx, rx;
+
+	/* Non-negotiable (NN) features */
+	rc = __feat_register_nn(fn, DCCPF_SEQUENCE_WINDOW, 0,
+				    sysctl_dccp_sequence_window);
+	if (rc)
+		return rc;
 
-	INIT_LIST_HEAD(&dmsk->dccpms_pending);
-	INIT_LIST_HEAD(&dmsk->dccpms_conf);
+	/* Server-priority (SP) features */
+
+	/* Advertise that short seqnos are not supported (7.6.1) */
+	rc = __feat_register_sp(fn, DCCPF_SHORT_SEQNOS, true, true, &off, 1);
+	if (rc)
+		return rc;
 
-	/* CCID L */
-	rc = __dccp_feat_init(dmsk, DCCPO_CHANGE_L, DCCPF_CCID,
-			      &dmsk->dccpms_tx_ccid, 1);
+	/* RFC 4340 12.1: "If a DCCP is not ECN capable, ..." */
+	rc = __feat_register_sp(fn, DCCPF_ECN_INCAPABLE, true, true, &on, 1);
 	if (rc)
-		goto out;
+		return rc;
+
+	/*
+	 * We advertise the available list of CCIDs and reorder according to
+	 * preferences, to avoid failure resulting from negotiating different
+	 * singleton values (which always leads to failure).
+	 * These settings can still (later) be overridden via sockopts.
+	 */
+	if (ccid_get_builtin_ccids(&tx.val, &tx.len) ||
+	    ccid_get_builtin_ccids(&rx.val, &rx.len))
+		return -ENOBUFS;
+
+	/* Pre-load all CCID modules that are going to be advertised */
+	rc = -EUNATCH;
+	if (ccid_request_modules(tx.val, tx.len))
+		goto free_ccid_lists;
+
+	if (!dccp_feat_prefer(sysctl_dccp_tx_ccid, tx.val, tx.len) ||
+	    !dccp_feat_prefer(sysctl_dccp_rx_ccid, rx.val, rx.len))
+		goto free_ccid_lists;
 
-	/* CCID R */
-	rc = __dccp_feat_init(dmsk, DCCPO_CHANGE_R, DCCPF_CCID,
-			      &dmsk->dccpms_rx_ccid, 1);
+	rc = __feat_register_sp(fn, DCCPF_CCID, true, false, tx.val, tx.len);
 	if (rc)
-		goto out;
+		goto free_ccid_lists;
 
-	/* Ack ratio */
-	rc = __dccp_feat_init(dmsk, DCCPO_CHANGE_L, DCCPF_ACK_RATIO,
-			      &dmsk->dccpms_ack_ratio, 1);
-out:
+	rc = __feat_register_sp(fn, DCCPF_CCID, false, false, rx.val, rx.len);
+
+free_ccid_lists:
+	kfree(tx.val);
+	kfree(rx.val);
 	return rc;
 }
 
-EXPORT_SYMBOL_GPL(dccp_feat_init);
-
-#ifdef CONFIG_IP_DCCP_DEBUG
-const char *dccp_feat_typename(const u8 type)
+int dccp_feat_activate_values(struct sock *sk, struct list_head *fn_list)
 {
-	switch(type) {
-	case DCCPO_CHANGE_L:  return("ChangeL");
-	case DCCPO_CONFIRM_L: return("ConfirmL");
-	case DCCPO_CHANGE_R:  return("ChangeR");
-	case DCCPO_CONFIRM_R: return("ConfirmR");
-	/* the following case must not appear in feature negotation  */
-	default:	      dccp_pr_debug("unknown type %d [BUG!]\n", type);
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_feat_entry *cur, *next;
+	int idx;
+	dccp_feat_val *fvals[DCCP_FEAT_SUPPORTED_MAX][2] = {
+		 [0 ... DCCP_FEAT_SUPPORTED_MAX-1] = { NULL, NULL }
+	};
+
+	list_for_each_entry(cur, fn_list, node) {
+		/*
+		 * An empty Confirm means that either an unknown feature type
+		 * or an invalid value was present. In the first case there is
+		 * nothing to activate, in the other the default value is used.
+		 */
+		if (cur->empty_confirm)
+			continue;
+
+		idx = dccp_feat_index(cur->feat_num);
+		if (idx < 0) {
+			DCCP_BUG("Unknown feature %u", cur->feat_num);
+			goto activation_failed;
+		}
+		if (cur->state != FEAT_STABLE) {
+			DCCP_CRIT("Negotiation of %s %s failed in state %s",
+				  cur->is_local ? "local" : "remote",
+				  dccp_feat_fname(cur->feat_num),
+				  dccp_feat_sname[cur->state]);
+			goto activation_failed;
+		}
+		fvals[idx][cur->is_local] = &cur->val;
 	}
-	return NULL;
-}
 
-EXPORT_SYMBOL_GPL(dccp_feat_typename);
+	/*
+	 * Activate in decreasing order of index, so that the CCIDs are always
+	 * activated as the last feature. This avoids the case where a CCID
+	 * relies on the initialisation of one or more features that it depends
+	 * on (e.g. Send NDP Count, Send Ack Vector, and Ack Ratio features).
+	 */
+	for (idx = DCCP_FEAT_SUPPORTED_MAX; --idx >= 0;)
+		if (__dccp_feat_activate(sk, idx, 0, fvals[idx][0]) ||
+		    __dccp_feat_activate(sk, idx, 1, fvals[idx][1])) {
+			DCCP_CRIT("Could not activate %d", idx);
+			goto activation_failed;
+		}
 
-const char *dccp_feat_name(const u8 feat)
-{
-	static const char *feature_names[] = {
-		[DCCPF_RESERVED]	= "Reserved",
-		[DCCPF_CCID]		= "CCID",
-		[DCCPF_SHORT_SEQNOS]	= "Allow Short Seqnos",
-		[DCCPF_SEQUENCE_WINDOW]	= "Sequence Window",
-		[DCCPF_ECN_INCAPABLE]	= "ECN Incapable",
-		[DCCPF_ACK_RATIO]	= "Ack Ratio",
-		[DCCPF_SEND_ACK_VECTOR]	= "Send ACK Vector",
-		[DCCPF_SEND_NDP_COUNT]	= "Send NDP Count",
-		[DCCPF_MIN_CSUM_COVER]	= "Min. Csum Coverage",
-		[DCCPF_DATA_CHECKSUM]	= "Send Data Checksum",
-	};
-	if (feat > DCCPF_DATA_CHECKSUM && feat < DCCPF_MIN_CCID_SPECIFIC)
-		return feature_names[DCCPF_RESERVED];
+	/* Clean up Change options which have been confirmed already */
+	list_for_each_entry_safe(cur, next, fn_list, node)
+		if (!cur->needs_confirm)
+			dccp_feat_list_pop(cur);
 
-	if (feat >= DCCPF_MIN_CCID_SPECIFIC)
-		return "CCID-specific";
+	dccp_pr_debug("Activation OK\n");
+	return 0;
 
-	return feature_names[feat];
+activation_failed:
+	/*
+	 * We clean up everything that may have been allocated, since
+	 * it is difficult to track at which stage negotiation failed.
+	 * This is ok, since all allocation functions below are robust
+	 * against NULL arguments.
+	 */
+	ccid_hc_rx_delete(dp->dccps_hc_rx_ccid, sk);
+	ccid_hc_tx_delete(dp->dccps_hc_tx_ccid, sk);
+	dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL;
+	dccp_ackvec_free(dp->dccps_hc_rx_ackvec);
+	dp->dccps_hc_rx_ackvec = NULL;
+	return -1;
 }
-
-EXPORT_SYMBOL_GPL(dccp_feat_name);
-#endif /* CONFIG_IP_DCCP_DEBUG */

net/dccp/feat.h

@@ -3,38 +3,134 @@
 /*
  *  net/dccp/feat.h
  *
- *  An implementation of the DCCP protocol
+ *  Feature negotiation for the DCCP protocol (RFC 4340, section 6)
+ *  Copyright (c) 2008 Gerrit Renker <gerrit@erg.abdn.ac.uk>
  *  Copyright (c) 2005 Andrea Bittau <a.bittau@cs.ucl.ac.uk>
  *
- *	This program is free software; you can redistribute it and/or modify it
- *	under the terms of the GNU General Public License version 2 as
- *	published by the Free Software Foundation.
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms of the GNU General Public License version 2 as
+ *  published by the Free Software Foundation.
  */
-
 #include <linux/types.h>
 #include "dccp.h"
 
-#ifdef CONFIG_IP_DCCP_DEBUG
-extern const char *dccp_feat_typename(const u8 type);
-extern const char *dccp_feat_name(const u8 feat);
+/*
+ * Known limit values
+ */
+/* Ack Ratio takes 2-byte integer values (11.3) */
+#define DCCPF_ACK_RATIO_MAX	0xFFFF
+/* Wmin=32 and Wmax=2^46-1 from 7.5.2 */
+#define DCCPF_SEQ_WMIN		32
+#define DCCPF_SEQ_WMAX		0x3FFFFFFFFFFFull
+/* Maximum number of SP values that fit in a single (Confirm) option */
+#define DCCP_FEAT_MAX_SP_VALS	(DCCP_SINGLE_OPT_MAXLEN - 2)
+
+enum dccp_feat_type {
+	FEAT_AT_RX   = 1,	/* located at RX side of half-connection  */
+	FEAT_AT_TX   = 2,	/* located at TX side of half-connection  */
+	FEAT_SP      = 4,	/* server-priority reconciliation (6.3.1) */
+	FEAT_NN	     = 8,	/* non-negotiable reconciliation (6.3.2)  */
+	FEAT_UNKNOWN = 0xFF	/* not understood or invalid feature	  */
+};
+
+enum dccp_feat_state {
+	FEAT_DEFAULT = 0,	/* using default values from 6.4 */
+	FEAT_INITIALISING,	/* feature is being initialised  */
+	FEAT_CHANGING,		/* Change sent but not confirmed yet */
+	FEAT_UNSTABLE,		/* local modification in state CHANGING */
+	FEAT_STABLE		/* both ends (think they) agree */
+};
 
-static inline void dccp_feat_debug(const u8 type, const u8 feat, const u8 val)
+/**
+ * dccp_feat_val  -  Container for SP or NN feature values
+ * @nn:     single NN value
+ * @sp.vec: single SP value plus optional preference list
+ * @sp.len: length of @sp.vec in bytes
+ */
+typedef union {
+	u64 nn;
+	struct {
+		u8	*vec;
+		u8	len;
+	}   sp;
+} dccp_feat_val;
+
+/**
+ * struct feat_entry  -  Data structure to perform feature negotiation
+ * @feat_num: one of %dccp_feature_numbers
+ * @val: feature's current value (SP features may have preference list)
+ * @state: feature's current state
+ * @needs_mandatory: whether Mandatory options should be sent
+ * @needs_confirm: whether to send a Confirm instead of a Change
+ * @empty_confirm: whether to send an empty Confirm (depends on @needs_confirm)
+ * @is_local: feature location (1) or feature-remote (0)
+ * @node: list pointers, entries arranged in FIFO order
+ */
+struct dccp_feat_entry {
+	u8                      feat_num;
+	dccp_feat_val           val;
+	enum dccp_feat_state    state:8;
+	bool			needs_mandatory:1,
+				needs_confirm:1,
+				empty_confirm:1,
+				is_local:1;
+
+	struct list_head	node;
+};
+
+static inline u8 dccp_feat_genopt(struct dccp_feat_entry *entry)
 {
-	dccp_pr_debug("%s(%s (%d), %d)\n", dccp_feat_typename(type),
-					   dccp_feat_name(feat), feat, val);
+	if (entry->needs_confirm)
+		return entry->is_local ? DCCPO_CONFIRM_L : DCCPO_CONFIRM_R;
+	return entry->is_local ? DCCPO_CHANGE_L : DCCPO_CHANGE_R;
 }
-#else
-#define dccp_feat_debug(type, feat, val)
-#endif /* CONFIG_IP_DCCP_DEBUG */
-
-extern int  dccp_feat_change(struct dccp_minisock *dmsk, u8 type, u8 feature,
-			     u8 *val, u8 len, gfp_t gfp);
-extern int  dccp_feat_change_recv(struct sock *sk, u8 type, u8 feature,
-				  u8 *val, u8 len);
-extern int  dccp_feat_confirm_recv(struct sock *sk, u8 type, u8 feature,
-				   u8 *val, u8 len);
-extern void dccp_feat_clean(struct dccp_minisock *dmsk);
-extern int  dccp_feat_clone(struct sock *oldsk, struct sock *newsk);
-extern int  dccp_feat_init(struct dccp_minisock *dmsk);
 
+/**
+ * struct ccid_dependency  -  Track changes resulting from choosing a CCID
+ * @dependent_feat: one of %dccp_feature_numbers
+ * @is_local: local (1) or remote (0) @dependent_feat
+ * @is_mandatory: whether presence of @dependent_feat is mission-critical or not
+ * @val: corresponding default value for @dependent_feat (u8 is sufficient here)
+ */
+struct ccid_dependency {
+	u8	dependent_feat;
+	bool	is_local:1,
+		is_mandatory:1;
+	u8	val;
+};
+
+/*
+ * Sysctls to seed defaults for feature negotiation
+ */
+extern unsigned long sysctl_dccp_sequence_window;
+extern int	     sysctl_dccp_rx_ccid;
+extern int	     sysctl_dccp_tx_ccid;
+
+extern int  dccp_feat_init(struct sock *sk);
+extern void dccp_feat_initialise_sysctls(void);
+extern int  dccp_feat_register_sp(struct sock *sk, u8 feat, u8 is_local,
+				  u8 const *list, u8 len);
+extern int  dccp_feat_register_nn(struct sock *sk, u8 feat, u64 val);
+extern int  dccp_feat_parse_options(struct sock *, struct dccp_request_sock *,
+				    u8 mand, u8 opt, u8 feat, u8 *val, u8 len);
+extern int  dccp_feat_clone_list(struct list_head const *, struct list_head *);
+
+/*
+ * Encoding variable-length options and their maximum length.
+ *
+ * This affects NN options (SP options are all u8) and other variable-length
+ * options (see table 3 in RFC 4340). The limit is currently given the Sequence
+ * Window NN value (sec. 7.5.2) and the NDP count (sec. 7.7) option, all other
+ * options consume less than 6 bytes (timestamps are 4 bytes).
+ * When updating this constant (e.g. due to new internet drafts / RFCs), make
+ * sure that you also update all code which refers to it.
+ */
+#define DCCP_OPTVAL_MAXLEN	6
+
+extern void dccp_encode_value_var(const u64 value, u8 *to, const u8 len);
+extern u64  dccp_decode_value_var(const u8 *bf, const u8 len);
+
+extern int  dccp_insert_option_mandatory(struct sk_buff *skb);
+extern int  dccp_insert_fn_opt(struct sk_buff *skb, u8 type, u8 feat,
+			       u8 *val, u8 len, bool repeat_first);
 #endif /* _DCCP_FEAT_H */

net/dccp/input.c

@@ -159,13 +159,15 @@ static void dccp_rcv_reset(struct sock *sk, struct sk_buff *skb)
 	dccp_time_wait(sk, DCCP_TIME_WAIT, 0);
 }
 
-static void dccp_event_ack_recv(struct sock *sk, struct sk_buff *skb)
+static void dccp_handle_ackvec_processing(struct sock *sk, struct sk_buff *skb)
 {
-	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_ackvec *av = dccp_sk(sk)->dccps_hc_rx_ackvec;
 
-	if (dccp_msk(sk)->dccpms_send_ack_vector)
-		dccp_ackvec_check_rcv_ackno(dp->dccps_hc_rx_ackvec, sk,
-					    DCCP_SKB_CB(skb)->dccpd_ack_seq);
+	if (av == NULL)
+		return;
+	if (DCCP_SKB_CB(skb)->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
+		dccp_ackvec_clear_state(av, DCCP_SKB_CB(skb)->dccpd_ack_seq);
+	dccp_ackvec_input(av, skb);
 }
 
 static void dccp_deliver_input_to_ccids(struct sock *sk, struct sk_buff *skb)
@@ -364,22 +366,13 @@ discard:
 int dccp_rcv_established(struct sock *sk, struct sk_buff *skb,
 			 const struct dccp_hdr *dh, const unsigned len)
 {
-	struct dccp_sock *dp = dccp_sk(sk);
-
 	if (dccp_check_seqno(sk, skb))
 		goto discard;
 
 	if (dccp_parse_options(sk, NULL, skb))
 		return 1;
 
-	if (DCCP_SKB_CB(skb)->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
-		dccp_event_ack_recv(sk, skb);
-
-	if (dccp_msk(sk)->dccpms_send_ack_vector &&
-	    dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk,
-			    DCCP_SKB_CB(skb)->dccpd_seq,
-			    DCCP_ACKVEC_STATE_RECEIVED))
-		goto discard;
+	dccp_handle_ackvec_processing(sk, skb);
 	dccp_deliver_input_to_ccids(sk, skb);
 
 	return __dccp_rcv_established(sk, skb, dh, len);
@@ -421,40 +414,33 @@ static int dccp_rcv_request_sent_state_process(struct sock *sk,
 			goto out_invalid_packet;
 		}
 
+		/*
+		 * If option processing (Step 8) failed, return 1 here so that
+		 * dccp_v4_do_rcv() sends a Reset. The Reset code depends on
+		 * the option type and is set in dccp_parse_options().
+		 */
 		if (dccp_parse_options(sk, NULL, skb))
-			goto out_invalid_packet;
+			return 1;
 
 		/* Obtain usec RTT sample from SYN exchange (used by CCID 3) */
 		if (likely(dp->dccps_options_received.dccpor_timestamp_echo))
 			dp->dccps_syn_rtt = dccp_sample_rtt(sk, 10 * (tstamp -
 			    dp->dccps_options_received.dccpor_timestamp_echo));
 
-		if (dccp_msk(sk)->dccpms_send_ack_vector &&
-		    dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk,
-				    DCCP_SKB_CB(skb)->dccpd_seq,
-				    DCCP_ACKVEC_STATE_RECEIVED))
-			goto out_invalid_packet; /* FIXME: change error code */
-
 		/* Stop the REQUEST timer */
 		inet_csk_clear_xmit_timer(sk, ICSK_TIME_RETRANS);
 		WARN_ON(sk->sk_send_head == NULL);
 		kfree_skb(sk->sk_send_head);
 		sk->sk_send_head = NULL;
 
-		dp->dccps_isr = DCCP_SKB_CB(skb)->dccpd_seq;
-		dccp_update_gsr(sk, dp->dccps_isr);
 		/*
-		 * SWL and AWL are initially adjusted so that they are not less than
-		 * the initial Sequence Numbers received and sent, respectively:
-		 *	SWL := max(GSR + 1 - floor(W/4), ISR),
-		 *	AWL := max(GSS - W' + 1, ISS).
-		 * These adjustments MUST be applied only at the beginning of the
-		 * connection.
-		 *
-		 * AWL was adjusted in dccp_v4_connect -acme
+		 * Set ISR, GSR from packet. ISS was set in dccp_v{4,6}_connect
+		 * and GSS in dccp_transmit_skb(). Setting AWL/AWH and SWL/SWH
+		 * is done as part of activating the feature values below, since
+		 * these settings depend on the local/remote Sequence Window
+		 * features, which were undefined or not confirmed until now.
 		 */
-		dccp_set_seqno(&dp->dccps_swl,
-			       max48(dp->dccps_swl, dp->dccps_isr));
+		dp->dccps_gsr = dp->dccps_isr = DCCP_SKB_CB(skb)->dccpd_seq;
 
 		dccp_sync_mss(sk, icsk->icsk_pmtu_cookie);
 
@@ -475,6 +461,15 @@ static int dccp_rcv_request_sent_state_process(struct sock *sk,
 		 */
 		dccp_set_state(sk, DCCP_PARTOPEN);
 
+		/*
+		 * If feature negotiation was successful, activate features now;
+		 * an activation failure means that this host could not activate
+		 * one ore more features (e.g. insufficient memory), which would
+		 * leave at least one feature in an undefined state.
+		 */
+		if (dccp_feat_activate_values(sk, &dp->dccps_featneg))
+			goto unable_to_proceed;
+
 		/* Make sure socket is routed, for correct metrics. */
 		icsk->icsk_af_ops->rebuild_header(sk);
 
@@ -509,6 +504,16 @@ out_invalid_packet:
 	/* dccp_v4_do_rcv will send a reset */
 	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR;
 	return 1;
+
+unable_to_proceed:
+	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_ABORTED;
+	/*
+	 * We mark this socket as no longer usable, so that the loop in
+	 * dccp_sendmsg() terminates and the application gets notified.
+	 */
+	dccp_set_state(sk, DCCP_CLOSED);
+	sk->sk_err = ECOMM;
+	return 1;
 }
 
 static int dccp_rcv_respond_partopen_state_process(struct sock *sk,
@@ -590,8 +595,6 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 			if (inet_csk(sk)->icsk_af_ops->conn_request(sk,
 								    skb) < 0)
 				return 1;
-
-			/* FIXME: do congestion control initialization */
 			goto discard;
 		}
 		if (dh->dccph_type == DCCP_PKT_RESET)
@@ -600,30 +603,36 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		/* Caller (dccp_v4_do_rcv) will send Reset */
 		dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION;
 		return 1;
+	} else if (sk->sk_state == DCCP_CLOSED) {
+		dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION;
+		return 1;
 	}
 
-	if (sk->sk_state != DCCP_REQUESTING) {
-		if (dccp_check_seqno(sk, skb))
-			goto discard;
-
-		/*
-		 * Step 8: Process options and mark acknowledgeable
-		 */
-		if (dccp_parse_options(sk, NULL, skb))
-			return 1;
-
-		if (dcb->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
-			dccp_event_ack_recv(sk, skb);
-
-		if (dccp_msk(sk)->dccpms_send_ack_vector &&
-		    dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk,
-				    DCCP_SKB_CB(skb)->dccpd_seq,
-				    DCCP_ACKVEC_STATE_RECEIVED))
-			goto discard;
+	/* Step 6: Check sequence numbers (omitted in LISTEN/REQUEST state) */
+	if (sk->sk_state != DCCP_REQUESTING && dccp_check_seqno(sk, skb))
+		goto discard;
 
-		dccp_deliver_input_to_ccids(sk, skb);
+	/*
+	 *   Step 7: Check for unexpected packet types
+	 *      If (S.is_server and P.type == Response)
+	 *	    or (S.is_client and P.type == Request)
+	 *	    or (S.state == RESPOND and P.type == Data),
+	 *	  Send Sync packet acknowledging P.seqno
+	 *	  Drop packet and return
+	 */
+	if ((dp->dccps_role != DCCP_ROLE_CLIENT &&
+	     dh->dccph_type == DCCP_PKT_RESPONSE) ||
+	    (dp->dccps_role == DCCP_ROLE_CLIENT &&
+	     dh->dccph_type == DCCP_PKT_REQUEST) ||
+	    (sk->sk_state == DCCP_RESPOND && dh->dccph_type == DCCP_PKT_DATA)) {
+		dccp_send_sync(sk, dcb->dccpd_seq, DCCP_PKT_SYNC);
+		goto discard;
 	}
 
+	/*  Step 8: Process options */
+	if (dccp_parse_options(sk, NULL, skb))
+		return 1;
+
 	/*
 	 *  Step 9: Process Reset
 	 *	If P.type == Reset,
@@ -631,44 +640,22 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 	 *		S.state := TIMEWAIT
 	 *		Set TIMEWAIT timer
 	 *		Drop packet and return
-	*/
+	 */
 	if (dh->dccph_type == DCCP_PKT_RESET) {
 		dccp_rcv_reset(sk, skb);
 		return 0;
-		/*
-		 *   Step 7: Check for unexpected packet types
-		 *      If (S.is_server and P.type == Response)
-		 *	    or (S.is_client and P.type == Request)
-		 *	    or (S.state == RESPOND and P.type == Data),
-		 *	  Send Sync packet acknowledging P.seqno
-		 *	  Drop packet and return
-		 */
-	} else if ((dp->dccps_role != DCCP_ROLE_CLIENT &&
-		    dh->dccph_type == DCCP_PKT_RESPONSE) ||
-		    (dp->dccps_role == DCCP_ROLE_CLIENT &&
-		     dh->dccph_type == DCCP_PKT_REQUEST) ||
-		    (sk->sk_state == DCCP_RESPOND &&
-		     dh->dccph_type == DCCP_PKT_DATA)) {
-		dccp_send_sync(sk, dcb->dccpd_seq, DCCP_PKT_SYNC);
-		goto discard;
-	} else if (dh->dccph_type == DCCP_PKT_CLOSEREQ) {
+	} else if (dh->dccph_type == DCCP_PKT_CLOSEREQ) {	/* Step 13 */
 		if (dccp_rcv_closereq(sk, skb))
 			return 0;
 		goto discard;
-	} else if (dh->dccph_type == DCCP_PKT_CLOSE) {
+	} else if (dh->dccph_type == DCCP_PKT_CLOSE) {		/* Step 14 */
 		if (dccp_rcv_close(sk, skb))
 			return 0;
 		goto discard;
 	}
 
 	switch (sk->sk_state) {
-	case DCCP_CLOSED:
-		dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION;
-		return 1;
-
 	case DCCP_REQUESTING:
-		/* FIXME: do congestion control initialization */
-
 		queued = dccp_rcv_request_sent_state_process(sk, skb, dh, len);
 		if (queued >= 0)
 			return queued;
@@ -676,8 +663,12 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		__kfree_skb(skb);
 		return 0;
 
-	case DCCP_RESPOND:
 	case DCCP_PARTOPEN:
+		/* Step 8: if using Ack Vectors, mark packet acknowledgeable */
+		dccp_handle_ackvec_processing(sk, skb);
+		dccp_deliver_input_to_ccids(sk, skb);
+		/* fall through */
+	case DCCP_RESPOND:
 		queued = dccp_rcv_respond_partopen_state_process(sk, skb,
 								 dh, len);
 		break;
@@ -716,16 +707,7 @@ u32 dccp_sample_rtt(struct sock *sk, long delta)
 	/* dccpor_elapsed_time is either zeroed out or set and > 0 */
 	delta -= dccp_sk(sk)->dccps_options_received.dccpor_elapsed_time * 10;
 
-	if (unlikely(delta <= 0)) {
-		DCCP_WARN("unusable RTT sample %ld, using min\n", delta);
-		return DCCP_SANE_RTT_MIN;
-	}
-	if (unlikely(delta > DCCP_SANE_RTT_MAX)) {
-		DCCP_WARN("RTT sample %ld too large, using max\n", delta);
-		return DCCP_SANE_RTT_MAX;
-	}
-
-	return delta;
+	return dccp_sane_rtt(delta);
 }
 
 EXPORT_SYMBOL_GPL(dccp_sample_rtt);

net/dccp/ipv4.c

@@ -545,6 +545,7 @@ out:
 
 static void dccp_v4_reqsk_destructor(struct request_sock *req)
 {
+	dccp_feat_list_purge(&dccp_rsk(req)->dreq_featneg);
 	kfree(inet_rsk(req)->opt);
 }
 
@@ -595,7 +596,8 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 	if (req == NULL)
 		goto drop;
 
-	dccp_reqsk_init(req, skb);
+	if (dccp_reqsk_init(req, dccp_sk(sk), skb))
+		goto drop_and_free;
 
 	dreq = dccp_rsk(req);
 	if (dccp_parse_options(sk, dreq, skb))

net/dccp/ipv6.c

@@ -302,6 +302,7 @@ done:
 
 static void dccp_v6_reqsk_destructor(struct request_sock *req)
 {
+	dccp_feat_list_purge(&dccp_rsk(req)->dreq_featneg);
 	if (inet6_rsk(req)->pktopts != NULL)
 		kfree_skb(inet6_rsk(req)->pktopts);
 }
@@ -424,7 +425,8 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 	if (req == NULL)
 		goto drop;
 
-	dccp_reqsk_init(req, skb);
+	if (dccp_reqsk_init(req, dccp_sk(sk), skb))
+		goto drop_and_free;
 
 	dreq = dccp_rsk(req);
 	if (dccp_parse_options(sk, dreq, skb))

net/dccp/minisocks.c

@@ -42,16 +42,6 @@ struct inet_timewait_death_row dccp_death_row = {
 
 EXPORT_SYMBOL_GPL(dccp_death_row);
 
-void dccp_minisock_init(struct dccp_minisock *dmsk)
-{
-	dmsk->dccpms_sequence_window = sysctl_dccp_feat_sequence_window;
-	dmsk->dccpms_rx_ccid	     = sysctl_dccp_feat_rx_ccid;
-	dmsk->dccpms_tx_ccid	     = sysctl_dccp_feat_tx_ccid;
-	dmsk->dccpms_ack_ratio	     = sysctl_dccp_feat_ack_ratio;
-	dmsk->dccpms_send_ack_vector = sysctl_dccp_feat_send_ack_vector;
-	dmsk->dccpms_send_ndp_count  = sysctl_dccp_feat_send_ndp_count;
-}
-
 void dccp_time_wait(struct sock *sk, int state, int timeo)
 {
 	struct inet_timewait_sock *tw = NULL;
@@ -112,10 +102,9 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 	struct sock *newsk = inet_csk_clone(sk, req, GFP_ATOMIC);
 
 	if (newsk != NULL) {
-		const struct dccp_request_sock *dreq = dccp_rsk(req);
+		struct dccp_request_sock *dreq = dccp_rsk(req);
 		struct inet_connection_sock *newicsk = inet_csk(newsk);
 		struct dccp_sock *newdp = dccp_sk(newsk);
-		struct dccp_minisock *newdmsk = dccp_msk(newsk);
 
 		newdp->dccps_role	    = DCCP_ROLE_SERVER;
 		newdp->dccps_hc_rx_ackvec   = NULL;
@@ -125,65 +114,32 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 		newdp->dccps_timestamp_time = dreq->dreq_timestamp_time;
 		newicsk->icsk_rto	    = DCCP_TIMEOUT_INIT;
 
-		if (dccp_feat_clone(sk, newsk))
-			goto out_free;
-
-		if (newdmsk->dccpms_send_ack_vector) {
-			newdp->dccps_hc_rx_ackvec =
-						dccp_ackvec_alloc(GFP_ATOMIC);
-			if (unlikely(newdp->dccps_hc_rx_ackvec == NULL))
-				goto out_free;
-		}
-
-		newdp->dccps_hc_rx_ccid =
-			    ccid_hc_rx_new(newdmsk->dccpms_rx_ccid,
-					   newsk, GFP_ATOMIC);
-		newdp->dccps_hc_tx_ccid =
-			    ccid_hc_tx_new(newdmsk->dccpms_tx_ccid,
-					   newsk, GFP_ATOMIC);
-		if (unlikely(newdp->dccps_hc_rx_ccid == NULL ||
-			     newdp->dccps_hc_tx_ccid == NULL)) {
-			dccp_ackvec_free(newdp->dccps_hc_rx_ackvec);
-			ccid_hc_rx_delete(newdp->dccps_hc_rx_ccid, newsk);
-			ccid_hc_tx_delete(newdp->dccps_hc_tx_ccid, newsk);
-out_free:
-			/* It is still raw copy of parent, so invalidate
-			 * destructor and make plain sk_free() */
-			newsk->sk_destruct = NULL;
-			sk_free(newsk);
-			return NULL;
-		}
-
+		INIT_LIST_HEAD(&newdp->dccps_featneg);
 		/*
 		 * Step 3: Process LISTEN state
 		 *
 		 *    Choose S.ISS (initial seqno) or set from Init Cookies
 		 *    Initialize S.GAR := S.ISS
-		 *    Set S.ISR, S.GSR, S.SWL, S.SWH from packet or Init Cookies
+		 *    Set S.ISR, S.GSR from packet (or Init Cookies)
+		 *
+		 *    Setting AWL/AWH and SWL/SWH happens as part of the feature
+		 *    activation below, as these windows all depend on the local
+		 *    and remote Sequence Window feature values (7.5.2).
 		 */
-
-		/* See dccp_v4_conn_request */
-		newdmsk->dccpms_sequence_window = req->rcv_wnd;
-
-		newdp->dccps_gar = newdp->dccps_iss = dreq->dreq_iss;
-		dccp_update_gss(newsk, dreq->dreq_iss);
-
-		newdp->dccps_isr = dreq->dreq_isr;
-		dccp_update_gsr(newsk, dreq->dreq_isr);
+		newdp->dccps_gss = newdp->dccps_iss = dreq->dreq_iss;
+		newdp->dccps_gar = newdp->dccps_iss;
+		newdp->dccps_gsr = newdp->dccps_isr = dreq->dreq_isr;
 
 		/*
-		 * SWL and AWL are initially adjusted so that they are not less than
-		 * the initial Sequence Numbers received and sent, respectively:
-		 *	SWL := max(GSR + 1 - floor(W/4), ISR),
-		 *	AWL := max(GSS - W' + 1, ISS).
-		 * These adjustments MUST be applied only at the beginning of the
-		 * connection.
+		 * Activate features: initialise CCIDs, sequence windows etc.
 		 */
-		dccp_set_seqno(&newdp->dccps_swl,
-			       max48(newdp->dccps_swl, newdp->dccps_isr));
-		dccp_set_seqno(&newdp->dccps_awl,
-			       max48(newdp->dccps_awl, newdp->dccps_iss));
-
+		if (dccp_feat_activate_values(newsk, &dreq->dreq_featneg)) {
+			/* It is still raw copy of parent, so invalidate
+			 * destructor and make plain sk_free() */
+			newsk->sk_destruct = NULL;
+			sk_free(newsk);
+			return NULL;
+		}
 		dccp_init_xmit_timers(newsk);
 
 		DCCP_INC_STATS_BH(DCCP_MIB_PASSIVEOPENS);
@@ -304,14 +260,17 @@ void dccp_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
 
 EXPORT_SYMBOL_GPL(dccp_reqsk_send_ack);
 
-void dccp_reqsk_init(struct request_sock *req, struct sk_buff *skb)
+int dccp_reqsk_init(struct request_sock *req,
+		    struct dccp_sock const *dp, struct sk_buff const *skb)
 {
 	struct dccp_request_sock *dreq = dccp_rsk(req);
 
 	inet_rsk(req)->rmt_port	  = dccp_hdr(skb)->dccph_sport;
 	inet_rsk(req)->acked	  = 0;
-	req->rcv_wnd		  = sysctl_dccp_feat_sequence_window;
 	dreq->dreq_timestamp_echo = 0;
+
+	/* inherit feature negotiation options from listening socket */
+	return dccp_feat_clone_list(&dp->dccps_featneg, &dreq->dreq_featneg);
 }
 
 EXPORT_SYMBOL_GPL(dccp_reqsk_init);

net/dccp/options.c

@@ -23,23 +23,20 @@
 #include "dccp.h"
 #include "feat.h"
 
-int sysctl_dccp_feat_sequence_window = DCCPF_INITIAL_SEQUENCE_WINDOW;
-int sysctl_dccp_feat_rx_ccid	      = DCCPF_INITIAL_CCID;
-int sysctl_dccp_feat_tx_ccid	      = DCCPF_INITIAL_CCID;
-int sysctl_dccp_feat_ack_ratio	      = DCCPF_INITIAL_ACK_RATIO;
-int sysctl_dccp_feat_send_ack_vector = DCCPF_INITIAL_SEND_ACK_VECTOR;
-int sysctl_dccp_feat_send_ndp_count  = DCCPF_INITIAL_SEND_NDP_COUNT;
-
-static u32 dccp_decode_value_var(const unsigned char *bf, const u8 len)
+u64 dccp_decode_value_var(const u8 *bf, const u8 len)
 {
-	u32 value = 0;
+	u64 value = 0;
 
+	if (len >= DCCP_OPTVAL_MAXLEN)
+		value += ((u64)*bf++) << 40;
+	if (len > 4)
+		value += ((u64)*bf++) << 32;
 	if (len > 3)
-		value += *bf++ << 24;
+		value += ((u64)*bf++) << 24;
 	if (len > 2)
-		value += *bf++ << 16;
+		value += ((u64)*bf++) << 16;
 	if (len > 1)
-		value += *bf++ << 8;
+		value += ((u64)*bf++) << 8;
 	if (len > 0)
 		value += *bf;
 
@@ -57,7 +54,6 @@ int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
 	struct dccp_sock *dp = dccp_sk(sk);
 	const struct dccp_hdr *dh = dccp_hdr(skb);
 	const u8 pkt_type = DCCP_SKB_CB(skb)->dccpd_type;
-	u64 ackno = DCCP_SKB_CB(skb)->dccpd_ack_seq;
 	unsigned char *options = (unsigned char *)dh + dccp_hdr_len(skb);
 	unsigned char *opt_ptr = options;
 	const unsigned char *opt_end = (unsigned char *)dh +
@@ -99,18 +95,11 @@ 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;
 
@@ -131,43 +120,13 @@ int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
 			dccp_pr_debug("%s opt: NDP count=%llu\n", dccp_role(sk),
 				      (unsigned long long)opt_recv->dccpor_ndp);
 			break;
-		case DCCPO_CHANGE_L:
-			/* fall through */
-		case DCCPO_CHANGE_R:
-			if (pkt_type == DCCP_PKT_DATA)
+		case DCCPO_CHANGE_L ... DCCPO_CONFIRM_R:
+			if (pkt_type == DCCP_PKT_DATA)      /* RFC 4340, 6 */
 				break;
-			if (len < 2)
-				goto out_invalid_option;
-			rc = dccp_feat_change_recv(sk, opt, *value, value + 1,
-						   len - 1);
-			/*
-			 * When there is a change error, change_recv is
-			 * responsible for dealing with it.  i.e. reply with an
-			 * empty confirm.
-			 * If the change was mandatory, then we need to die.
-			 */
-			if (rc && mandatory)
-				goto out_invalid_option;
-			break;
-		case DCCPO_CONFIRM_L:
-			/* fall through */
-		case DCCPO_CONFIRM_R:
-			if (pkt_type == DCCP_PKT_DATA)
-				break;
-			if (len < 2)	/* FIXME this disallows empty confirm */
-				goto out_invalid_option;
-			if (dccp_feat_confirm_recv(sk, opt, *value,
-						   value + 1, len - 1))
-				goto out_invalid_option;
-			break;
-		case DCCPO_ACK_VECTOR_0:
-		case DCCPO_ACK_VECTOR_1:
-			if (dccp_packet_without_ack(skb))   /* RFC 4340, 11.4 */
-				break;
-
-			if (dccp_msk(sk)->dccpms_send_ack_vector &&
-			    dccp_ackvec_parse(sk, skb, &ackno, opt, value, len))
-				goto out_invalid_option;
+			rc = dccp_feat_parse_options(sk, dreq, mandatory, opt,
+						    *value, value + 1, len - 1);
+			if (rc)
+				goto out_featneg_failed;
 			break;
 		case DCCPO_TIMESTAMP:
 			if (len != 4)
@@ -195,6 +154,8 @@ int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
 				      dccp_role(sk), ntohl(opt_val),
 				      (unsigned long long)
 				      DCCP_SKB_CB(skb)->dccpd_ack_seq);
+			/* schedule an Ack in case this sender is quiescent */
+			inet_csk_schedule_ack(sk);
 			break;
 		case DCCPO_TIMESTAMP_ECHO:
 			if (len != 4 && len != 6 && len != 8)
@@ -251,23 +212,25 @@ 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: {
-			const u16 idx = value - options;
-
+		case DCCPO_MIN_RX_CCID_SPECIFIC ... DCCPO_MAX_RX_CCID_SPECIFIC:
 			if (ccid_hc_rx_parse_options(dp->dccps_hc_rx_ccid, sk,
-						     opt, len, idx,
-						     value) != 0)
+						     pkt_type, opt, value, len))
 				goto out_invalid_option;
-		}
 			break;
-		case 192 ... 255: {
-			const u16 idx = value - options;
-
+		case DCCPO_ACK_VECTOR_0:
+		case DCCPO_ACK_VECTOR_1:
+			if (dccp_packet_without_ack(skb))   /* RFC 4340, 11.4 */
+				break;
+			/*
+			 * Ack vectors are processed by the TX CCID if it is
+			 * interested. The RX CCID need not parse Ack Vectors,
+			 * since it is only interested in clearing old state.
+			 * Fall through.
+			 */
+		case DCCPO_MIN_TX_CCID_SPECIFIC ... DCCPO_MAX_TX_CCID_SPECIFIC:
 			if (ccid_hc_tx_parse_options(dp->dccps_hc_tx_ccid, sk,
-						     opt, len, idx,
-						     value) != 0)
+						     pkt_type, opt, value, len))
 				goto out_invalid_option;
-		}
 			break;
 		default:
 			DCCP_CRIT("DCCP(%p): option %d(len=%d) not "
@@ -289,8 +252,10 @@ out_nonsensical_length:
 
 out_invalid_option:
 	DCCP_INC_STATS_BH(DCCP_MIB_INVALIDOPT);
-	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_OPTION_ERROR;
-	DCCP_WARN("DCCP(%p): invalid option %d, len=%d", sk, opt, len);
+	rc = DCCP_RESET_CODE_OPTION_ERROR;
+out_featneg_failed:
+	DCCP_WARN("DCCP(%p): Option %d (len=%d) error=%u\n", sk, opt, len, rc);
+	DCCP_SKB_CB(skb)->dccpd_reset_code = rc;
 	DCCP_SKB_CB(skb)->dccpd_reset_data[0] = opt;
 	DCCP_SKB_CB(skb)->dccpd_reset_data[1] = len > 0 ? value[0] : 0;
 	DCCP_SKB_CB(skb)->dccpd_reset_data[2] = len > 1 ? value[1] : 0;
@@ -299,9 +264,12 @@ out_invalid_option:
 
 EXPORT_SYMBOL_GPL(dccp_parse_options);
 
-static void dccp_encode_value_var(const u32 value, unsigned char *to,
-				  const unsigned int len)
+void dccp_encode_value_var(const u64 value, u8 *to, const u8 len)
 {
+	if (len >= DCCP_OPTVAL_MAXLEN)
+		*to++ = (value & 0xFF0000000000ull) >> 40;
+	if (len > 4)
+		*to++ = (value & 0xFF00000000ull) >> 32;
 	if (len > 3)
 		*to++ = (value & 0xFF000000) >> 24;
 	if (len > 2)
@@ -461,92 +429,140 @@ static int dccp_insert_option_timestamp_echo(struct dccp_sock *dp,
 	return 0;
 }
 
-static int dccp_insert_feat_opt(struct sk_buff *skb, u8 type, u8 feat,
-				u8 *val, u8 len)
+static int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb)
 {
-	u8 *to;
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_ackvec *av = dp->dccps_hc_rx_ackvec;
+	struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
+	const u16 buflen = dccp_ackvec_buflen(av);
+	/* Figure out how many options do we need to represent the ackvec */
+	const u8 nr_opts = DIV_ROUND_UP(buflen, DCCP_SINGLE_OPT_MAXLEN);
+	u16 len = buflen + 2 * nr_opts;
+	u8 i, nonce = 0;
+	const unsigned char *tail, *from;
+	unsigned char *to;
 
-	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len + 3 > DCCP_MAX_OPT_LEN) {
-		DCCP_WARN("packet too small for feature %d option!\n", feat);
+	if (dcb->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) {
+		DCCP_WARN("Lacking space for %u bytes on %s packet\n", len,
+			  dccp_packet_name(dcb->dccpd_type));
 		return -1;
 	}
+	/*
+	 * Since Ack Vectors are variable-length, we can not always predict
+	 * their size. To catch exception cases where the space is running out
+	 * on the skb, a separate Sync is scheduled to carry the Ack Vector.
+	 */
+	if (len > DCCPAV_MIN_OPTLEN &&
+	    len + dcb->dccpd_opt_len + skb->len > dp->dccps_mss_cache) {
+		DCCP_WARN("No space left for Ack Vector (%u) on skb (%u+%u), "
+			  "MPS=%u ==> reduce payload size?\n", len, skb->len,
+			  dcb->dccpd_opt_len, dp->dccps_mss_cache);
+		dp->dccps_sync_scheduled = 1;
+		return 0;
+	}
+	dcb->dccpd_opt_len += len;
 
-	DCCP_SKB_CB(skb)->dccpd_opt_len += len + 3;
+	to   = skb_push(skb, len);
+	len  = buflen;
+	from = av->av_buf + av->av_buf_head;
+	tail = av->av_buf + DCCPAV_MAX_ACKVEC_LEN;
 
-	to    = skb_push(skb, len + 3);
-	*to++ = type;
-	*to++ = len + 3;
-	*to++ = feat;
+	for (i = 0; i < nr_opts; ++i) {
+		int copylen = len;
 
-	if (len)
-		memcpy(to, val, len);
+		if (len > DCCP_SINGLE_OPT_MAXLEN)
+			copylen = DCCP_SINGLE_OPT_MAXLEN;
+
+		/*
+		 * RFC 4340, 12.2: Encode the Nonce Echo for this Ack Vector via
+		 * its type; ack_nonce is the sum of all individual buf_nonce's.
+		 */
+		nonce ^= av->av_buf_nonce[i];
+
+		*to++ = DCCPO_ACK_VECTOR_0 + av->av_buf_nonce[i];
+		*to++ = copylen + 2;
 
-	dccp_pr_debug("%s(%s (%d), ...), length %d\n",
-		      dccp_feat_typename(type),
-		      dccp_feat_name(feat), feat, len);
+		/* Check if buf_head wraps */
+		if (from + copylen > tail) {
+			const u16 tailsize = tail - from;
+
+			memcpy(to, from, tailsize);
+			to	+= tailsize;
+			len	-= tailsize;
+			copylen	-= tailsize;
+			from	= av->av_buf;
+		}
+
+		memcpy(to, from, copylen);
+		from += copylen;
+		to   += copylen;
+		len  -= copylen;
+	}
+	/*
+	 * Each sent Ack Vector is recorded in the list, as per A.2 of RFC 4340.
+	 */
+	if (dccp_ackvec_update_records(av, dcb->dccpd_seq, nonce))
+		return -ENOBUFS;
 	return 0;
 }
 
-static int dccp_insert_options_feat(struct sock *sk, struct sk_buff *skb)
+/**
+ * dccp_insert_option_mandatory  -  Mandatory option (5.8.2)
+ * Note that since we are using skb_push, this function needs to be called
+ * _after_ inserting the option it is supposed to influence (stack order).
+ */
+int dccp_insert_option_mandatory(struct sk_buff *skb)
 {
-	struct dccp_sock *dp = dccp_sk(sk);
-	struct dccp_minisock *dmsk = dccp_msk(sk);
-	struct dccp_opt_pend *opt, *next;
-	int change = 0;
-
-	/* confirm any options [NN opts] */
-	list_for_each_entry_safe(opt, next, &dmsk->dccpms_conf, dccpop_node) {
-		dccp_insert_feat_opt(skb, opt->dccpop_type,
-				     opt->dccpop_feat, opt->dccpop_val,
-				     opt->dccpop_len);
-		/* fear empty confirms */
-		if (opt->dccpop_val)
-			kfree(opt->dccpop_val);
-		kfree(opt);
-	}
-	INIT_LIST_HEAD(&dmsk->dccpms_conf);
-
-	/* see which features we need to send */
-	list_for_each_entry(opt, &dmsk->dccpms_pending, dccpop_node) {
-		/* see if we need to send any confirm */
-		if (opt->dccpop_sc) {
-			dccp_insert_feat_opt(skb, opt->dccpop_type + 1,
-					     opt->dccpop_feat,
-					     opt->dccpop_sc->dccpoc_val,
-					     opt->dccpop_sc->dccpoc_len);
-
-			BUG_ON(!opt->dccpop_sc->dccpoc_val);
-			kfree(opt->dccpop_sc->dccpoc_val);
-			kfree(opt->dccpop_sc);
-			opt->dccpop_sc = NULL;
-		}
+	if (DCCP_SKB_CB(skb)->dccpd_opt_len >= DCCP_MAX_OPT_LEN)
+		return -1;
 
-		/* any option not confirmed, re-send it */
-		if (!opt->dccpop_conf) {
-			dccp_insert_feat_opt(skb, opt->dccpop_type,
-					     opt->dccpop_feat, opt->dccpop_val,
-					     opt->dccpop_len);
-			change++;
-		}
+	DCCP_SKB_CB(skb)->dccpd_opt_len++;
+	*skb_push(skb, 1) = DCCPO_MANDATORY;
+	return 0;
+}
+
+/**
+ * dccp_insert_fn_opt  -  Insert single Feature-Negotiation option into @skb
+ * @type: %DCCPO_CHANGE_L, %DCCPO_CHANGE_R, %DCCPO_CONFIRM_L, %DCCPO_CONFIRM_R
+ * @feat: one out of %dccp_feature_numbers
+ * @val: NN value or SP array (preferred element first) to copy
+ * @len: true length of @val in bytes (excluding first element repetition)
+ * @repeat_first: whether to copy the first element of @val twice
+ * The last argument is used to construct Confirm options, where the preferred
+ * value and the preference list appear separately (RFC 4340, 6.3.1). Preference
+ * lists are kept such that the preferred entry is always first, so we only need
+ * to copy twice, and avoid the overhead of cloning into a bigger array.
+ */
+int dccp_insert_fn_opt(struct sk_buff *skb, u8 type, u8 feat,
+		       u8 *val, u8 len, bool repeat_first)
+{
+	u8 tot_len, *to;
+
+	/* take the `Feature' field and possible repetition into account */
+	if (len > (DCCP_SINGLE_OPT_MAXLEN - 2)) {
+		DCCP_WARN("length %u for feature %u too large\n", len, feat);
+		return -1;
 	}
 
-	/* Retransmit timer.
-	 * If this is the master listening sock, we don't set a timer on it.  It
-	 * should be fine because if the dude doesn't receive our RESPONSE
-	 * [which will contain the CHANGE] he will send another REQUEST which
-	 * will "retrnasmit" the change.
-	 */
-	if (change && dp->dccps_role != DCCP_ROLE_LISTEN) {
-		dccp_pr_debug("reset feat negotiation timer %p\n", sk);
+	if (unlikely(val == NULL || len == 0))
+		len = repeat_first = 0;
+	tot_len = 3 + repeat_first + len;
 
-		/* XXX don't reset the timer on re-transmissions.  I.e. reset it
-		 * only when sending new stuff i guess.  Currently the timer
-		 * never backs off because on re-transmission it just resets it!
-		 */
-		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
-					  inet_csk(sk)->icsk_rto, DCCP_RTO_MAX);
+	if (DCCP_SKB_CB(skb)->dccpd_opt_len + tot_len > DCCP_MAX_OPT_LEN) {
+		DCCP_WARN("packet too small for feature %d option!\n", feat);
+		return -1;
 	}
+	DCCP_SKB_CB(skb)->dccpd_opt_len += tot_len;
+
+	to    = skb_push(skb, tot_len);
+	*to++ = type;
+	*to++ = tot_len;
+	*to++ = feat;
 
+	if (repeat_first)
+		*to++ = *val;
+	if (len)
+		memcpy(to, val, len);
 	return 0;
 }
 
@@ -565,19 +581,30 @@ static void dccp_insert_option_padding(struct sk_buff *skb)
 int dccp_insert_options(struct sock *sk, struct sk_buff *skb)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
-	struct dccp_minisock *dmsk = dccp_msk(sk);
 
 	DCCP_SKB_CB(skb)->dccpd_opt_len = 0;
 
-	if (dmsk->dccpms_send_ndp_count &&
-	    dccp_insert_option_ndp(sk, skb))
+	if (dp->dccps_send_ndp_count && dccp_insert_option_ndp(sk, skb))
 		return -1;
 
-	if (!dccp_packet_without_ack(skb)) {
-		if (dmsk->dccpms_send_ack_vector &&
-		    dccp_ackvec_pending(dp->dccps_hc_rx_ackvec) &&
-		    dccp_insert_option_ackvec(sk, skb))
+	if (DCCP_SKB_CB(skb)->dccpd_type != DCCP_PKT_DATA) {
+
+		/* Feature Negotiation */
+		if (dccp_feat_insert_opts(dp, NULL, skb))
 			return -1;
+
+		if (DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_REQUEST) {
+			/*
+			 * Obtain RTT sample from Request/Response exchange.
+			 * This is currently used in CCID 3 initialisation.
+			 */
+			if (dccp_insert_option_timestamp(sk, skb))
+				return -1;
+
+		} else if (dccp_ackvec_pending(sk) &&
+			   dccp_insert_option_ackvec(sk, skb)) {
+				return -1;
+		}
 	}
 
 	if (dp->dccps_hc_rx_insert_options) {
@@ -586,21 +613,6 @@ int dccp_insert_options(struct sock *sk, struct sk_buff *skb)
 		dp->dccps_hc_rx_insert_options = 0;
 	}
 
-	/* Feature negotiation */
-	/* Data packets can't do feat negotiation */
-	if (DCCP_SKB_CB(skb)->dccpd_type != DCCP_PKT_DATA &&
-	    DCCP_SKB_CB(skb)->dccpd_type != DCCP_PKT_DATAACK &&
-	    dccp_insert_options_feat(sk, skb))
-		return -1;
-
-	/*
-	 * Obtain RTT sample from Request/Response exchange.
-	 * This is currently used in CCID 3 initialisation.
-	 */
-	if (DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_REQUEST &&
-	    dccp_insert_option_timestamp(sk, skb))
-		return -1;
-
 	if (dp->dccps_timestamp_echo != 0 &&
 	    dccp_insert_option_timestamp_echo(dp, NULL, skb))
 		return -1;
@@ -613,6 +625,9 @@ int dccp_insert_options_rsk(struct dccp_request_sock *dreq, struct sk_buff *skb)
 {
 	DCCP_SKB_CB(skb)->dccpd_opt_len = 0;
 
+	if (dccp_feat_insert_opts(NULL, dreq, skb))
+		return -1;
+
 	if (dreq->dreq_timestamp_echo != 0 &&
 	    dccp_insert_option_timestamp_echo(NULL, dreq, skb))
 		return -1;

net/dccp/output.c

@@ -26,11 +26,13 @@ static inline void dccp_event_ack_sent(struct sock *sk)
 	inet_csk_clear_xmit_timer(sk, ICSK_TIME_DACK);
 }
 
-static void dccp_skb_entail(struct sock *sk, struct sk_buff *skb)
+/* enqueue @skb on sk_send_head for retransmission, return clone to send now */
+static struct sk_buff *dccp_skb_entail(struct sock *sk, struct sk_buff *skb)
 {
 	skb_set_owner_w(skb, sk);
 	WARN_ON(sk->sk_send_head);
 	sk->sk_send_head = skb;
+	return skb_clone(sk->sk_send_head, gfp_any());
 }
 
 /*
@@ -161,21 +163,27 @@ unsigned int dccp_sync_mss(struct sock *sk, u32 pmtu)
 	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct dccp_sock *dp = dccp_sk(sk);
 	u32 ccmps = dccp_determine_ccmps(dp);
-	int cur_mps = ccmps ? min(pmtu, ccmps) : pmtu;
+	u32 cur_mps = ccmps ? min(pmtu, ccmps) : pmtu;
 
 	/* Account for header lengths and IPv4/v6 option overhead */
 	cur_mps -= (icsk->icsk_af_ops->net_header_len + icsk->icsk_ext_hdr_len +
 		    sizeof(struct dccp_hdr) + sizeof(struct dccp_hdr_ext));
 
 	/*
-	 * FIXME: this should come from the CCID infrastructure, where, say,
-	 * TFRC will say it wants TIMESTAMPS, ELAPSED time, etc, for now lets
-	 * put a rough estimate for NDP + TIMESTAMP + TIMESTAMP_ECHO + ELAPSED
-	 * TIME + TFRC_OPT_LOSS_EVENT_RATE + TFRC_OPT_RECEIVE_RATE + padding to
-	 * make it a multiple of 4
+	 * Leave enough headroom for common DCCP header options.
+	 * This only considers options which may appear on DCCP-Data packets, as
+	 * per table 3 in RFC 4340, 5.8. When running out of space for other
+	 * options (eg. Ack Vector which can take up to 255 bytes), it is better
+	 * to schedule a separate Ack. Thus we leave headroom for the following:
+	 *  - 1 byte for Slow Receiver (11.6)
+	 *  - 6 bytes for Timestamp (13.1)
+	 *  - 10 bytes for Timestamp Echo (13.3)
+	 *  - 8 bytes for NDP count (7.7, when activated)
+	 *  - 6 bytes for Data Checksum (9.3)
+	 *  - %DCCPAV_MIN_OPTLEN bytes for Ack Vector size (11.4, when enabled)
 	 */
-
-	cur_mps -= ((5 + 6 + 10 + 6 + 6 + 6 + 3) / 4) * 4;
+	cur_mps -= roundup(1 + 6 + 10 + dp->dccps_send_ndp_count * 8 + 6 +
+			   (dp->dccps_hc_rx_ackvec ? DCCPAV_MIN_OPTLEN : 0), 4);
 
 	/* And store cached results */
 	icsk->icsk_pmtu_cookie = pmtu;
@@ -200,95 +208,158 @@ void dccp_write_space(struct sock *sk)
 }
 
 /**
- * dccp_wait_for_ccid - Wait for ccid to tell us we can send a packet
+ * dccp_wait_for_ccid  -  Await CCID send permission
  * @sk:    socket to wait for
- * @skb:   current skb to pass on for waiting
- * @delay: sleep timeout in milliseconds (> 0)
- * This function is called by default when the socket is closed, and
- * when a non-zero linger time is set on the socket. For consistency
+ * @delay: timeout in jiffies
+ * This is used by CCIDs which need to delay the send time in process context.
  */
-static int dccp_wait_for_ccid(struct sock *sk, struct sk_buff *skb, int delay)
+static int dccp_wait_for_ccid(struct sock *sk, unsigned long delay)
 {
-	struct dccp_sock *dp = dccp_sk(sk);
 	DEFINE_WAIT(wait);
-	unsigned long jiffdelay;
-	int rc;
+	long remaining;
 
-	do {
-		dccp_pr_debug("delayed send by %d msec\n", delay);
-		jiffdelay = msecs_to_jiffies(delay);
+	prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
+	sk->sk_write_pending++;
+	release_sock(sk);
 
-		prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
+	remaining = schedule_timeout(delay);
 
-		sk->sk_write_pending++;
-		release_sock(sk);
-		schedule_timeout(jiffdelay);
-		lock_sock(sk);
-		sk->sk_write_pending--;
+	lock_sock(sk);
+	sk->sk_write_pending--;
+	finish_wait(sk->sk_sleep, &wait);
 
-		if (sk->sk_err)
-			goto do_error;
-		if (signal_pending(current))
-			goto do_interrupted;
+	if (signal_pending(current) || sk->sk_err)
+		return -1;
+	return remaining;
+}
 
-		rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
-	} while ((delay = rc) > 0);
-out:
-	finish_wait(sk->sk_sleep, &wait);
-	return rc;
-
-do_error:
-	rc = -EPIPE;
-	goto out;
-do_interrupted:
-	rc = -EINTR;
-	goto out;
+/**
+ * dccp_xmit_packet  -  Send data packet under control of CCID
+ * Transmits next-queued payload and informs CCID to account for the packet.
+ */
+static void dccp_xmit_packet(struct sock *sk)
+{
+	int err, len;
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct sk_buff *skb = dccp_qpolicy_pop(sk);
+
+	if (unlikely(skb == NULL))
+		return;
+	len = skb->len;
+
+	if (sk->sk_state == DCCP_PARTOPEN) {
+		const u32 cur_mps = dp->dccps_mss_cache - DCCP_FEATNEG_OVERHEAD;
+		/*
+		 * See 8.1.5 - Handshake Completion.
+		 *
+		 * For robustness we resend Confirm options until the client has
+		 * entered OPEN. During the initial feature negotiation, the MPS
+		 * is smaller than usual, reduced by the Change/Confirm options.
+		 */
+		if (!list_empty(&dp->dccps_featneg) && len > cur_mps) {
+			DCCP_WARN("Payload too large (%d) for featneg.\n", len);
+			dccp_send_ack(sk);
+			dccp_feat_list_purge(&dp->dccps_featneg);
+		}
+
+		inet_csk_schedule_ack(sk);
+		inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
+					      inet_csk(sk)->icsk_rto,
+					      DCCP_RTO_MAX);
+		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_DATAACK;
+	} else if (dccp_ack_pending(sk)) {
+		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_DATAACK;
+	} else {
+		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_DATA;
+	}
+
+	err = dccp_transmit_skb(sk, skb);
+	if (err)
+		dccp_pr_debug("transmit_skb() returned err=%d\n", err);
+	/*
+	 * Register this one as sent even if an error occurred. To the remote
+	 * end a local packet drop is indistinguishable from network loss, i.e.
+	 * any local drop will eventually be reported via receiver feedback.
+	 */
+	ccid_hc_tx_packet_sent(dp->dccps_hc_tx_ccid, sk, len);
+
+	/*
+	 * If the CCID needs to transfer additional header options out-of-band
+	 * (e.g. Ack Vectors or feature-negotiation options), it activates this
+	 * flag to schedule a Sync. The Sync will automatically incorporate all
+	 * currently pending header options, thus clearing the backlog.
+	 */
+	if (dp->dccps_sync_scheduled)
+		dccp_send_sync(sk, dp->dccps_gsr, DCCP_PKT_SYNC);
 }
 
-void dccp_write_xmit(struct sock *sk, int block)
+/**
+ * dccp_flush_write_queue  -  Drain queue at end of connection
+ * Since dccp_sendmsg queues packets without waiting for them to be sent, it may
+ * happen that the TX queue is not empty at the end of a connection. We give the
+ * HC-sender CCID a grace period of up to @time_budget jiffies. If this function
+ * returns with a non-empty write queue, it will be purged later.
+ */
+void dccp_flush_write_queue(struct sock *sk, long *time_budget)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
 	struct sk_buff *skb;
+	long delay, rc;
+
+	while (*time_budget > 0 && (skb = skb_peek(&sk->sk_write_queue))) {
+		rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
 
-	while ((skb = skb_peek(&sk->sk_write_queue))) {
-		int err = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
-
-		if (err > 0) {
-			if (!block) {
-				sk_reset_timer(sk, &dp->dccps_xmit_timer,
-						msecs_to_jiffies(err)+jiffies);
-				break;
-			} else
-				err = dccp_wait_for_ccid(sk, skb, err);
-			if (err && err != -EINTR)
-				DCCP_BUG("err=%d after dccp_wait_for_ccid", err);
+		switch (ccid_packet_dequeue_eval(rc)) {
+		case CCID_PACKET_WILL_DEQUEUE_LATER:
+			/*
+			 * If the CCID determines when to send, the next sending
+			 * time is unknown or the CCID may not even send again
+			 * (e.g. remote host crashes or lost Ack packets).
+			 */
+			DCCP_WARN("CCID did not manage to send all packets\n");
+			return;
+		case CCID_PACKET_DELAY:
+			delay = msecs_to_jiffies(rc);
+			if (delay > *time_budget)
+				return;
+			rc = dccp_wait_for_ccid(sk, delay);
+			if (rc < 0)
+				return;
+			*time_budget -= (delay - rc);
+			/* check again if we can send now */
+			break;
+		case CCID_PACKET_SEND_AT_ONCE:
+			dccp_xmit_packet(sk);
+			break;
+		case CCID_PACKET_ERR:
+			skb_dequeue(&sk->sk_write_queue);
+			kfree_skb(skb);
+			dccp_pr_debug("packet discarded due to err=%ld\n", rc);
 		}
+	}
+}
 
-		skb_dequeue(&sk->sk_write_queue);
-		if (err == 0) {
-			struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
-			const int len = skb->len;
+void dccp_write_xmit(struct sock *sk)
+{
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct sk_buff *skb;
 
-			if (sk->sk_state == DCCP_PARTOPEN) {
-				/* See 8.1.5.  Handshake Completion */
-				inet_csk_schedule_ack(sk);
-				inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
-						  inet_csk(sk)->icsk_rto,
-						  DCCP_RTO_MAX);
-				dcb->dccpd_type = DCCP_PKT_DATAACK;
-			} else if (dccp_ack_pending(sk))
-				dcb->dccpd_type = DCCP_PKT_DATAACK;
-			else
-				dcb->dccpd_type = DCCP_PKT_DATA;
-
-			err = dccp_transmit_skb(sk, skb);
-			ccid_hc_tx_packet_sent(dp->dccps_hc_tx_ccid, sk, 0, len);
-			if (err)
-				DCCP_BUG("err=%d after ccid_hc_tx_packet_sent",
-					 err);
-		} else {
-			dccp_pr_debug("packet discarded due to err=%d\n", err);
-			kfree_skb(skb);
+	while ((skb = dccp_qpolicy_top(sk))) {
+		int rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb);
+
+		switch (ccid_packet_dequeue_eval(rc)) {
+		case CCID_PACKET_WILL_DEQUEUE_LATER:
+			return;
+		case CCID_PACKET_DELAY:
+			sk_reset_timer(sk, &dp->dccps_xmit_timer,
+				       jiffies + msecs_to_jiffies(rc));
+			return;
+		case CCID_PACKET_SEND_AT_ONCE:
+			dccp_xmit_packet(sk);
+			break;
+		case CCID_PACKET_ERR:
+			dccp_qpolicy_drop(sk, skb);
+			dccp_pr_debug("packet discarded due to err=%d\n", rc);
 		}
 	}
 }
@@ -339,10 +410,12 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
 	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE;
 	DCCP_SKB_CB(skb)->dccpd_seq  = dreq->dreq_iss;
 
-	if (dccp_insert_options_rsk(dreq, skb)) {
-		kfree_skb(skb);
-		return NULL;
-	}
+	/* Resolve feature dependencies resulting from choice of CCID */
+	if (dccp_feat_server_ccid_dependencies(dreq))
+		goto response_failed;
+
+	if (dccp_insert_options_rsk(dreq, skb))
+		goto response_failed;
 
 	/* Build and checksum header */
 	dh = dccp_zeroed_hdr(skb, dccp_header_size);
@@ -363,6 +436,9 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
 	inet_rsk(req)->acked = 1;
 	DCCP_INC_STATS(DCCP_MIB_OUTSEGS);
 	return skb;
+response_failed:
+	kfree_skb(skb);
+	return NULL;
 }
 
 EXPORT_SYMBOL_GPL(dccp_make_response);
@@ -447,8 +523,9 @@ int dccp_send_reset(struct sock *sk, enum dccp_reset_codes code)
 /*
  * Do all connect socket setups that can be done AF independent.
  */
-static inline void dccp_connect_init(struct sock *sk)
+int dccp_connect(struct sock *sk)
 {
+	struct sk_buff *skb;
 	struct dccp_sock *dp = dccp_sk(sk);
 	struct dst_entry *dst = __sk_dst_get(sk);
 	struct inet_connection_sock *icsk = inet_csk(sk);
@@ -458,19 +535,13 @@ static inline void dccp_connect_init(struct sock *sk)
 
 	dccp_sync_mss(sk, dst_mtu(dst));
 
+	/* do not connect if feature negotiation setup fails */
+	if (dccp_feat_finalise_settings(dccp_sk(sk)))
+		return -EPROTO;
+
 	/* Initialise GAR as per 8.5; AWL/AWH are set in dccp_transmit_skb() */
 	dp->dccps_gar = dp->dccps_iss;
 
-	icsk->icsk_retransmits = 0;
-}
-
-int dccp_connect(struct sock *sk)
-{
-	struct sk_buff *skb;
-	struct inet_connection_sock *icsk = inet_csk(sk);
-
-	dccp_connect_init(sk);
-
 	skb = alloc_skb(sk->sk_prot->max_header, sk->sk_allocation);
 	if (unlikely(skb == NULL))
 		return -ENOBUFS;
@@ -480,11 +551,11 @@ int dccp_connect(struct sock *sk)
 
 	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_REQUEST;
 
-	dccp_skb_entail(sk, skb);
-	dccp_transmit_skb(sk, skb_clone(skb, GFP_KERNEL));
+	dccp_transmit_skb(sk, dccp_skb_entail(sk, skb));
 	DCCP_INC_STATS(DCCP_MIB_ACTIVEOPENS);
 
 	/* Timer for repeating the REQUEST until an answer. */
+	icsk->icsk_retransmits = 0;
 	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
 				  icsk->icsk_rto, DCCP_RTO_MAX);
 	return 0;
@@ -571,6 +642,12 @@ void dccp_send_sync(struct sock *sk, const u64 ackno,
 	DCCP_SKB_CB(skb)->dccpd_type = pkt_type;
 	DCCP_SKB_CB(skb)->dccpd_ack_seq = ackno;
 
+	/*
+	 * Clear the flag in case the Sync was scheduled for out-of-band data,
+	 * such as carrying a long Ack Vector.
+	 */
+	dccp_sk(sk)->dccps_sync_scheduled = 0;
+
 	dccp_transmit_skb(sk, skb);
 }
 
@@ -599,9 +676,7 @@ void dccp_send_close(struct sock *sk, const int active)
 		DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_CLOSE;
 
 	if (active) {
-		dccp_write_xmit(sk, 1);
-		dccp_skb_entail(sk, skb);
-		dccp_transmit_skb(sk, skb_clone(skb, prio));
+		skb = dccp_skb_entail(sk, skb);
 		/*
 		 * Retransmission timer for active-close: RFC 4340, 8.3 requires
 		 * to retransmit the Close/CloseReq until the CLOSING/CLOSEREQ
@@ -614,6 +689,6 @@ void dccp_send_close(struct sock *sk, const int active)
 		 */
 		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
 					  DCCP_TIMEOUT_INIT, DCCP_RTO_MAX);
-	} else
-		dccp_transmit_skb(sk, skb);
+	}
+	dccp_transmit_skb(sk, skb);
 }

net/dccp/probe.c

@@ -46,75 +46,54 @@ static struct {
 	struct kfifo	  *fifo;
 	spinlock_t	  lock;
 	wait_queue_head_t wait;
-	struct timespec	  tstart;
+	ktime_t		  start;
 } dccpw;
 
-static void printl(const char *fmt, ...)
+static void jdccp_write_xmit(struct sock *sk)
 {
-	va_list args;
-	int len;
-	struct timespec now;
-	char tbuf[256];
-
-	va_start(args, fmt);
-	getnstimeofday(&now);
-
-	now = timespec_sub(now, dccpw.tstart);
-
-	len = sprintf(tbuf, "%lu.%06lu ",
-		      (unsigned long) now.tv_sec,
-		      (unsigned long) now.tv_nsec / NSEC_PER_USEC);
-	len += vscnprintf(tbuf+len, sizeof(tbuf)-len, fmt, args);
-	va_end(args);
-
-	kfifo_put(dccpw.fifo, tbuf, len);
-	wake_up(&dccpw.wait);
-}
-
-static int jdccp_sendmsg(struct kiocb *iocb, struct sock *sk,
-			 struct msghdr *msg, size_t size)
-{
-	const struct dccp_minisock *dmsk = dccp_msk(sk);
 	const struct inet_sock *inet = inet_sk(sk);
-	const struct ccid3_hc_tx_sock *hctx;
+	struct ccid3_hc_tx_sock *hctx = NULL;
+	struct timespec tv;
+	char buf[256];
+	int len, ccid = ccid_get_current_tx_ccid(dccp_sk(sk));
 
-	if (dmsk->dccpms_tx_ccid == DCCPC_CCID3)
+	if (ccid == DCCPC_CCID3)
 		hctx = ccid3_hc_tx_sk(sk);
-	else
-		hctx = NULL;
 
-	if (port == 0 || ntohs(inet->dport) == port ||
-	    ntohs(inet->sport) == port) {
-		if (hctx)
-			printl("%d.%d.%d.%d:%u %d.%d.%d.%d:%u %d %d %d %d %u "
-			       "%llu %llu %d\n",
-			       NIPQUAD(inet->saddr), ntohs(inet->sport),
-			       NIPQUAD(inet->daddr), ntohs(inet->dport), size,
-			       hctx->ccid3hctx_s, hctx->ccid3hctx_rtt,
-			       hctx->ccid3hctx_p, hctx->ccid3hctx_x_calc,
-			       hctx->ccid3hctx_x_recv >> 6,
-			       hctx->ccid3hctx_x >> 6, hctx->ccid3hctx_t_ipi);
-		else
-			printl("%d.%d.%d.%d:%u %d.%d.%d.%d:%u %d\n",
+	if (!port || ntohs(inet->dport) == port || ntohs(inet->sport) == port) {
+
+		tv  = ktime_to_timespec(ktime_sub(ktime_get(), dccpw.start));
+		len = sprintf(buf, "%lu.%09lu %d.%d.%d.%d:%u %d.%d.%d.%d:%u %d",
+			       (unsigned long)tv.tv_sec,
+			       (unsigned long)tv.tv_nsec,
 			       NIPQUAD(inet->saddr), ntohs(inet->sport),
-			       NIPQUAD(inet->daddr), ntohs(inet->dport), size);
+			       NIPQUAD(inet->daddr), ntohs(inet->dport), ccid);
+
+		if (hctx)
+			len += sprintf(buf + len, " %d %d %d %u %u %u %d",
+			       hctx->s, hctx->rtt, hctx->p, hctx->x_calc,
+			       (unsigned)(hctx->x_recv >> 6),
+			       (unsigned)(hctx->x >> 6), hctx->t_ipi);
+
+		len += sprintf(buf + len, "\n");
+		kfifo_put(dccpw.fifo, buf, len);
+		wake_up(&dccpw.wait);
 	}
 
 	jprobe_return();
-	return 0;
 }
 
 static struct jprobe dccp_send_probe = {
 	.kp	= {
-		.symbol_name = "dccp_sendmsg",
+		.symbol_name = "dccp_write_xmit",
 	},
-	.entry	= jdccp_sendmsg,
+	.entry	= jdccp_write_xmit,
 };
 
 static int dccpprobe_open(struct inode *inode, struct file *file)
 {
 	kfifo_reset(dccpw.fifo);
-	getnstimeofday(&dccpw.tstart);
+	dccpw.start = ktime_get();
 	return 0;
 }
 

net/dccp/proto.c

@@ -67,6 +67,9 @@ void dccp_set_state(struct sock *sk, const int state)
 	case DCCP_OPEN:
 		if (oldstate != DCCP_OPEN)
 			DCCP_INC_STATS(DCCP_MIB_CURRESTAB);
+		/* Client retransmits all Confirm options until entering OPEN */
+		if (oldstate == DCCP_PARTOPEN)
+			dccp_feat_list_purge(&dccp_sk(sk)->dccps_featneg);
 		break;
 
 	case DCCP_CLOSED:
@@ -175,63 +178,25 @@ EXPORT_SYMBOL_GPL(dccp_state_name);
 int dccp_init_sock(struct sock *sk, const __u8 ctl_sock_initialized)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
-	struct dccp_minisock *dmsk = dccp_msk(sk);
 	struct inet_connection_sock *icsk = inet_csk(sk);
 
-	dccp_minisock_init(&dp->dccps_minisock);
-
 	icsk->icsk_rto		= DCCP_TIMEOUT_INIT;
 	icsk->icsk_syn_retries	= sysctl_dccp_request_retries;
 	sk->sk_state		= DCCP_CLOSED;
 	sk->sk_write_space	= dccp_write_space;
 	icsk->icsk_sync_mss	= dccp_sync_mss;
-	dp->dccps_mss_cache	= 536;
+	dp->dccps_mss_cache	= TCP_MIN_RCVMSS;
 	dp->dccps_rate_last	= jiffies;
 	dp->dccps_role		= DCCP_ROLE_UNDEFINED;
 	dp->dccps_service	= DCCP_SERVICE_CODE_IS_ABSENT;
-	dp->dccps_l_ack_ratio	= dp->dccps_r_ack_ratio = 1;
+	dp->dccps_tx_qlen	= sysctl_dccp_tx_qlen;
 
 	dccp_init_xmit_timers(sk);
 
-	/*
-	 * FIXME: We're hardcoding the CCID, and doing this at this point makes
-	 * the listening (master) sock get CCID control blocks, which is not
-	 * necessary, but for now, to not mess with the test userspace apps,
-	 * lets leave it here, later the real solution is to do this in a
-	 * setsockopt(CCIDs-I-want/accept). -acme
-	 */
-	if (likely(ctl_sock_initialized)) {
-		int rc = dccp_feat_init(dmsk);
-
-		if (rc)
-			return rc;
-
-		if (dmsk->dccpms_send_ack_vector) {
-			dp->dccps_hc_rx_ackvec = dccp_ackvec_alloc(GFP_KERNEL);
-			if (dp->dccps_hc_rx_ackvec == NULL)
-				return -ENOMEM;
-		}
-		dp->dccps_hc_rx_ccid = ccid_hc_rx_new(dmsk->dccpms_rx_ccid,
-						      sk, GFP_KERNEL);
-		dp->dccps_hc_tx_ccid = ccid_hc_tx_new(dmsk->dccpms_tx_ccid,
-						      sk, GFP_KERNEL);
-		if (unlikely(dp->dccps_hc_rx_ccid == NULL ||
-			     dp->dccps_hc_tx_ccid == NULL)) {
-			ccid_hc_rx_delete(dp->dccps_hc_rx_ccid, sk);
-			ccid_hc_tx_delete(dp->dccps_hc_tx_ccid, sk);
-			if (dmsk->dccpms_send_ack_vector) {
-				dccp_ackvec_free(dp->dccps_hc_rx_ackvec);
-				dp->dccps_hc_rx_ackvec = NULL;
-			}
-			dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL;
-			return -ENOMEM;
-		}
-	} else {
-		/* control socket doesn't need feat nego */
-		INIT_LIST_HEAD(&dmsk->dccpms_pending);
-		INIT_LIST_HEAD(&dmsk->dccpms_conf);
-	}
-
+	INIT_LIST_HEAD(&dp->dccps_featneg);
+	/* control socket doesn't need feat nego */
+	if (likely(ctl_sock_initialized))
+		return dccp_feat_init(sk);
 	return 0;
 }
 
@@ -240,7 +205,6 @@ EXPORT_SYMBOL_GPL(dccp_init_sock);
 void dccp_destroy_sock(struct sock *sk)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
-	struct dccp_minisock *dmsk = dccp_msk(sk);
 
 	/*
 	 * DCCP doesn't use sk_write_queue, just sk_send_head
@@ -258,7 +222,7 @@ void dccp_destroy_sock(struct sock *sk)
 	kfree(dp->dccps_service_list);
 	dp->dccps_service_list = NULL;
 
-	if (dmsk->dccpms_send_ack_vector) {
+	if (dp->dccps_hc_rx_ackvec != NULL) {
 		dccp_ackvec_free(dp->dccps_hc_rx_ackvec);
 		dp->dccps_hc_rx_ackvec = NULL;
 	}
@@ -267,7 +231,7 @@ void dccp_destroy_sock(struct sock *sk)
 	dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL;
 
 	/* clean up feature negotiation state */
-	dccp_feat_clean(dmsk);
+	dccp_feat_list_purge(&dp->dccps_featneg);
 }
 
 EXPORT_SYMBOL_GPL(dccp_destroy_sock);
@@ -277,6 +241,9 @@ static inline int dccp_listen_start(struct sock *sk, int backlog)
 	struct dccp_sock *dp = dccp_sk(sk);
 
 	dp->dccps_role = DCCP_ROLE_LISTEN;
+	/* do not start to listen if feature negotiation setup fails */
+	if (dccp_feat_finalise_settings(dp))
+		return -EPROTO;
 	return inet_csk_listen_start(sk, backlog);
 }
 
@@ -466,42 +433,70 @@ static int dccp_setsockopt_service(struct sock *sk, const __be32 service,
 	return 0;
 }
 
-/* byte 1 is feature.  the rest is the preference list */
-static int dccp_setsockopt_change(struct sock *sk, int type,
-				  struct dccp_so_feat __user *optval)
+static int dccp_setsockopt_cscov(struct sock *sk, int cscov, bool rx)
 {
-	struct dccp_so_feat opt;
-	u8 *val;
-	int rc;
+	u8 *list, len;
+	int i, rc;
 
-	if (copy_from_user(&opt, optval, sizeof(opt)))
-		return -EFAULT;
+	if (cscov < 0 || cscov > 15)
+		return -EINVAL;
 	/*
-	 * rfc4340: 6.1. Change Options
+	 * Populate a list of permissible values, in the range cscov...15. This
+	 * is necessary since feature negotiation of single values only works if
+	 * both sides incidentally choose the same value. Since the list starts
+	 * lowest-value first, negotiation will pick the smallest shared value.
 	 */
-	if (opt.dccpsf_len < 1)
+	if (cscov == 0)
+		return 0;
+	len = 16 - cscov;
+
+	list = kmalloc(len, GFP_KERNEL);
+	if (list == NULL)
+		return -ENOBUFS;
+
+	for (i = 0; i < len; i++)
+		list[i] = cscov++;
+
+	rc = dccp_feat_register_sp(sk, DCCPF_MIN_CSUM_COVER, rx, list, len);
+
+	if (rc == 0) {
+		if (rx)
+			dccp_sk(sk)->dccps_pcrlen = cscov;
+		else
+			dccp_sk(sk)->dccps_pcslen = cscov;
+	}
+	kfree(list);
+	return rc;
+}
+
+static int dccp_setsockopt_ccid(struct sock *sk, int type,
+				char __user *optval, int optlen)
+{
+	u8 *val;
+	int rc = 0;
+
+	if (optlen < 1 || optlen > DCCP_FEAT_MAX_SP_VALS)
 		return -EINVAL;
 
-	val = kmalloc(opt.dccpsf_len, GFP_KERNEL);
-	if (!val)
+	val = kmalloc(optlen, GFP_KERNEL);
+	if (val == NULL)
 		return -ENOMEM;
 
-	if (copy_from_user(val, opt.dccpsf_val, opt.dccpsf_len)) {
-		rc = -EFAULT;
-		goto out_free_val;
+	if (copy_from_user(val, optval, optlen)) {
+		kfree(val);
+		return -EFAULT;
 	}
 
-	rc = dccp_feat_change(dccp_msk(sk), type, opt.dccpsf_feat,
-			      val, opt.dccpsf_len, GFP_KERNEL);
-	if (rc)
-		goto out_free_val;
+	lock_sock(sk);
+	if (type == DCCP_SOCKOPT_TX_CCID || type == DCCP_SOCKOPT_CCID)
+		rc = dccp_feat_register_sp(sk, DCCPF_CCID, 1, val, optlen);
 
-out:
-	return rc;
+	if (!rc && (type == DCCP_SOCKOPT_RX_CCID || type == DCCP_SOCKOPT_CCID))
+		rc = dccp_feat_register_sp(sk, DCCPF_CCID, 0, val, optlen);
+	release_sock(sk);
 
-out_free_val:
 	kfree(val);
-	goto out;
+	return rc;
 }
 
 static int do_dccp_setsockopt(struct sock *sk, int level, int optname,
@@ -510,7 +505,21 @@ static int do_dccp_setsockopt(struct sock *sk, int level, int optname,
 	struct dccp_sock *dp = dccp_sk(sk);
 	int val, err = 0;
 
-	if (optlen < sizeof(int))
+	switch (optname) {
+	case DCCP_SOCKOPT_PACKET_SIZE:
+		DCCP_WARN("sockopt(PACKET_SIZE) is deprecated: fix your app\n");
+		return 0;
+	case DCCP_SOCKOPT_CHANGE_L:
+	case DCCP_SOCKOPT_CHANGE_R:
+		DCCP_WARN("sockopt(CHANGE_L/R) is deprecated: fix your app\n");
+		return 0;
+	case DCCP_SOCKOPT_CCID:
+	case DCCP_SOCKOPT_RX_CCID:
+	case DCCP_SOCKOPT_TX_CCID:
+		return dccp_setsockopt_ccid(sk, optname, optval, optlen);
+	}
+
+	if (optlen < (int)sizeof(int))
 		return -EINVAL;
 
 	if (get_user(val, (int __user *)optval))
@@ -521,53 +530,38 @@ static int do_dccp_setsockopt(struct sock *sk, int level, int optname,
 
 	lock_sock(sk);
 	switch (optname) {
-	case DCCP_SOCKOPT_PACKET_SIZE:
-		DCCP_WARN("sockopt(PACKET_SIZE) is deprecated: fix your app\n");
-		err = 0;
-		break;
-	case DCCP_SOCKOPT_CHANGE_L:
-		if (optlen != sizeof(struct dccp_so_feat))
-			err = -EINVAL;
-		else
-			err = dccp_setsockopt_change(sk, DCCPO_CHANGE_L,
-						     (struct dccp_so_feat __user *)
-						     optval);
-		break;
-	case DCCP_SOCKOPT_CHANGE_R:
-		if (optlen != sizeof(struct dccp_so_feat))
-			err = -EINVAL;
-		else
-			err = dccp_setsockopt_change(sk, DCCPO_CHANGE_R,
-						     (struct dccp_so_feat __user *)
-						     optval);
-		break;
 	case DCCP_SOCKOPT_SERVER_TIMEWAIT:
 		if (dp->dccps_role != DCCP_ROLE_SERVER)
 			err = -EOPNOTSUPP;
 		else
 			dp->dccps_server_timewait = (val != 0);
 		break;
-	case DCCP_SOCKOPT_SEND_CSCOV:	/* sender side, RFC 4340, sec. 9.2 */
-		if (val < 0 || val > 15)
+	case DCCP_SOCKOPT_SEND_CSCOV:
+		err = dccp_setsockopt_cscov(sk, val, false);
+		break;
+	case DCCP_SOCKOPT_RECV_CSCOV:
+		err = dccp_setsockopt_cscov(sk, val, true);
+		break;
+	case DCCP_SOCKOPT_QPOLICY_ID:
+		if (sk->sk_state != DCCP_CLOSED)
+			err = -EISCONN;
+		else if (val < 0 || val >= DCCPQ_POLICY_MAX)
 			err = -EINVAL;
 		else
-			dp->dccps_pcslen = val;
+			dp->dccps_qpolicy = val;
 		break;
-	case DCCP_SOCKOPT_RECV_CSCOV:	/* receiver side, RFC 4340 sec. 9.2.1 */
-		if (val < 0 || val > 15)
+	case DCCP_SOCKOPT_QPOLICY_TXQLEN:
+		if (val < 0)
 			err = -EINVAL;
-		else {
-			dp->dccps_pcrlen = val;
-			/* FIXME: add feature negotiation,
-			 * ChangeL(MinimumChecksumCoverage, val) */
-		}
+		else
+			dp->dccps_tx_qlen = val;
 		break;
 	default:
 		err = -ENOPROTOOPT;
 		break;
 	}
-
 	release_sock(sk);
+
 	return err;
 }
 
@@ -648,6 +642,18 @@ static int do_dccp_getsockopt(struct sock *sk, int level, int optname,
 	case DCCP_SOCKOPT_GET_CUR_MPS:
 		val = dp->dccps_mss_cache;
 		break;
+	case DCCP_SOCKOPT_AVAILABLE_CCIDS:
+		return ccid_getsockopt_builtin_ccids(sk, len, optval, optlen);
+	case DCCP_SOCKOPT_TX_CCID:
+		val = ccid_get_current_tx_ccid(dp);
+		if (val < 0)
+			return -ENOPROTOOPT;
+		break;
+	case DCCP_SOCKOPT_RX_CCID:
+		val = ccid_get_current_rx_ccid(dp);
+		if (val < 0)
+			return -ENOPROTOOPT;
+		break;
 	case DCCP_SOCKOPT_SERVER_TIMEWAIT:
 		val = dp->dccps_server_timewait;
 		break;
@@ -657,6 +663,12 @@ static int do_dccp_getsockopt(struct sock *sk, int level, int optname,
 	case DCCP_SOCKOPT_RECV_CSCOV:
 		val = dp->dccps_pcrlen;
 		break;
+	case DCCP_SOCKOPT_QPOLICY_ID:
+		val = dp->dccps_qpolicy;
+		break;
+	case DCCP_SOCKOPT_QPOLICY_TXQLEN:
+		val = dp->dccps_tx_qlen;
+		break;
 	case 128 ... 191:
 		return ccid_hc_rx_getsockopt(dp->dccps_hc_rx_ccid, sk, optname,
 					     len, (u32 __user *)optval, optlen);
@@ -699,6 +711,47 @@ int compat_dccp_getsockopt(struct sock *sk, int level, int optname,
 EXPORT_SYMBOL_GPL(compat_dccp_getsockopt);
 #endif
 
+static int dccp_msghdr_parse(struct msghdr *msg, struct sk_buff *skb)
+{
+	struct cmsghdr *cmsg = CMSG_FIRSTHDR(msg);
+
+	/*
+	 * Assign an (opaque) qpolicy priority value to skb->priority.
+	 *
+	 * We are overloading this skb field for use with the qpolicy subystem.
+	 * The skb->priority is normally used for the SO_PRIORITY option, which
+	 * is initialised from sk_priority. Since the assignment of sk_priority
+	 * to skb->priority happens later (on layer 3), we overload this field
+	 * for use with queueing priorities as long as the skb is on layer 4.
+	 * The default priority value (if nothing is set) is 0.
+	 */
+	skb->priority = 0;
+
+	for (; cmsg != NULL; cmsg = CMSG_NXTHDR(msg, cmsg)) {
+
+		if (!CMSG_OK(msg, cmsg))
+			return -EINVAL;
+
+		if (cmsg->cmsg_level != SOL_DCCP)
+			continue;
+
+		if (cmsg->cmsg_type <= DCCP_SCM_QPOLICY_MAX &&
+		    !dccp_qpolicy_param_ok(skb->sk, cmsg->cmsg_type))
+			return -EINVAL;
+
+		switch (cmsg->cmsg_type) {
+		case DCCP_SCM_PRIORITY:
+			if (cmsg->cmsg_len != CMSG_LEN(sizeof(__u32)))
+				return -EINVAL;
+			skb->priority = *(__u32 *)CMSG_DATA(cmsg);
+			break;
+		default:
+			return -EINVAL;
+		}
+	}
+	return 0;
+}
+
 int dccp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 		 size_t len)
 {
@@ -714,8 +767,7 @@ int dccp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 
 	lock_sock(sk);
 
-	if (sysctl_dccp_tx_qlen &&
-	    (sk->sk_write_queue.qlen >= sysctl_dccp_tx_qlen)) {
+	if (dccp_qpolicy_full(sk)) {
 		rc = -EAGAIN;
 		goto out_release;
 	}
@@ -743,8 +795,12 @@ int dccp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 	if (rc != 0)
 		goto out_discard;
 
-	skb_queue_tail(&sk->sk_write_queue, skb);
-	dccp_write_xmit(sk,0);
+	rc = dccp_msghdr_parse(msg, skb);
+	if (rc != 0)
+		goto out_discard;
+
+	dccp_qpolicy_push(sk, skb);
+	dccp_write_xmit(sk);
 out_release:
 	release_sock(sk);
 	return rc ? : len;
@@ -967,9 +1023,22 @@ void dccp_close(struct sock *sk, long timeout)
 		/* Check zero linger _after_ checking for unread data. */
 		sk->sk_prot->disconnect(sk, 0);
 	} else if (sk->sk_state != DCCP_CLOSED) {
+		/*
+		 * Normal connection termination. May need to wait if there are
+		 * still packets in the TX queue that are delayed by the CCID.
+		 */
+		dccp_flush_write_queue(sk, &timeout);
 		dccp_terminate_connection(sk);
 	}
 
+	/*
+	 * Flush write queue. This may be necessary in several cases:
+	 * - we have been closed by the peer but still have application data;
+	 * - abortive termination (unread data or zero linger time),
+	 * - normal termination but queue could not be flushed within time limit
+	 */
+	__skb_queue_purge(&sk->sk_write_queue);
+
 	sk_stream_wait_close(sk, timeout);
 
 adjudge_to_death:

net/dccp/qpolicy.c

@@ -0,0 +1,137 @@
+/*
+ *  net/dccp/qpolicy.c
+ *
+ *  Policy-based packet dequeueing interface for DCCP.
+ *
+ *  Copyright (c) 2008 Tomasz Grobelny <tomasz@grobelny.oswiecenia.net>
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License v2
+ *  as published by the Free Software Foundation.
+ */
+#include "dccp.h"
+
+/*
+ *	Simple Dequeueing Policy:
+ *	If tx_qlen is different from 0, enqueue up to tx_qlen elements.
+ */
+static void qpolicy_simple_push(struct sock *sk, struct sk_buff *skb)
+{
+	skb_queue_tail(&sk->sk_write_queue, skb);
+}
+
+static bool qpolicy_simple_full(struct sock *sk)
+{
+	return dccp_sk(sk)->dccps_tx_qlen &&
+	       sk->sk_write_queue.qlen >= dccp_sk(sk)->dccps_tx_qlen;
+}
+
+static struct sk_buff *qpolicy_simple_top(struct sock *sk)
+{
+	return skb_peek(&sk->sk_write_queue);
+}
+
+/*
+ *	Priority-based Dequeueing Policy:
+ *	If tx_qlen is different from 0 and the queue has reached its upper bound
+ *	of tx_qlen elements, replace older packets lowest-priority-first.
+ */
+static struct sk_buff *qpolicy_prio_best_skb(struct sock *sk)
+{
+	struct sk_buff *skb, *best = NULL;
+
+	skb_queue_walk(&sk->sk_write_queue, skb)
+		if (best == NULL || skb->priority > best->priority)
+			best = skb;
+	return best;
+}
+
+static struct sk_buff *qpolicy_prio_worst_skb(struct sock *sk)
+{
+	struct sk_buff *skb, *worst = NULL;
+
+	skb_queue_walk(&sk->sk_write_queue, skb)
+		if (worst == NULL || skb->priority < worst->priority)
+			worst = skb;
+	return worst;
+}
+
+static bool qpolicy_prio_full(struct sock *sk)
+{
+	if (qpolicy_simple_full(sk))
+		dccp_qpolicy_drop(sk, qpolicy_prio_worst_skb(sk));
+	return false;
+}
+
+/**
+ * struct dccp_qpolicy_operations  -  TX Packet Dequeueing Interface
+ * @push: add a new @skb to the write queue
+ * @full: indicates that no more packets will be admitted
+ * @top:  peeks at whatever the queueing policy defines as its `top'
+ */
+static struct dccp_qpolicy_operations {
+	void		(*push)	(struct sock *sk, struct sk_buff *skb);
+	bool		(*full) (struct sock *sk);
+	struct sk_buff*	(*top)  (struct sock *sk);
+	__be32		params;
+
+} qpol_table[DCCPQ_POLICY_MAX] = {
+	[DCCPQ_POLICY_SIMPLE] = {
+		.push   = qpolicy_simple_push,
+		.full   = qpolicy_simple_full,
+		.top    = qpolicy_simple_top,
+		.params = 0,
+	},
+	[DCCPQ_POLICY_PRIO] = {
+		.push   = qpolicy_simple_push,
+		.full   = qpolicy_prio_full,
+		.top    = qpolicy_prio_best_skb,
+		.params = DCCP_SCM_PRIORITY,
+	},
+};
+
+/*
+ *	Externally visible interface
+ */
+void dccp_qpolicy_push(struct sock *sk, struct sk_buff *skb)
+{
+	qpol_table[dccp_sk(sk)->dccps_qpolicy].push(sk, skb);
+}
+
+bool dccp_qpolicy_full(struct sock *sk)
+{
+	return qpol_table[dccp_sk(sk)->dccps_qpolicy].full(sk);
+}
+
+void dccp_qpolicy_drop(struct sock *sk, struct sk_buff *skb)
+{
+	if (skb != NULL) {
+		skb_unlink(skb, &sk->sk_write_queue);
+		kfree_skb(skb);
+	}
+}
+
+struct sk_buff *dccp_qpolicy_top(struct sock *sk)
+{
+	return qpol_table[dccp_sk(sk)->dccps_qpolicy].top(sk);
+}
+
+struct sk_buff *dccp_qpolicy_pop(struct sock *sk)
+{
+	struct sk_buff *skb = dccp_qpolicy_top(sk);
+
+	/* Clear any skb fields that we used internally */
+	skb->priority = 0;
+
+	if (skb)
+		skb_unlink(skb, &sk->sk_write_queue);
+	return skb;
+}
+
+bool dccp_qpolicy_param_ok(struct sock *sk, __be32 param)
+{
+	/* check if exactly one bit is set */
+	if (!param || (param & (param - 1)))
+		return false;
+	return (qpol_table[dccp_sk(sk)->dccps_qpolicy].params & param) == param;
+}

net/dccp/sysctl.c

@@ -18,76 +18,72 @@
 #error This file should not be compiled without CONFIG_SYSCTL defined
 #endif
 
+/* Boundary values */
+static int		zero     = 0,
+			u8_max   = 0xFF;
+static unsigned long	seqw_min = 32;
+
 static struct ctl_table dccp_default_table[] = {
 	{
 		.procname	= "seq_window",
-		.data		= &sysctl_dccp_feat_sequence_window,
-		.maxlen		= sizeof(sysctl_dccp_feat_sequence_window),
+		.data		= &sysctl_dccp_sequence_window,
+		.maxlen		= sizeof(sysctl_dccp_sequence_window),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_doulongvec_minmax,
+		.extra1		= &seqw_min,		/* RFC 4340, 7.5.2 */
 	},
 	{
 		.procname	= "rx_ccid",
-		.data		= &sysctl_dccp_feat_rx_ccid,
-		.maxlen		= sizeof(sysctl_dccp_feat_rx_ccid),
+		.data		= &sysctl_dccp_rx_ccid,
+		.maxlen		= sizeof(sysctl_dccp_rx_ccid),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &u8_max,		/* RFC 4340, 10. */
 	},
 	{
 		.procname	= "tx_ccid",
-		.data		= &sysctl_dccp_feat_tx_ccid,
-		.maxlen		= sizeof(sysctl_dccp_feat_tx_ccid),
-		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
-	},
-	{
-		.procname	= "ack_ratio",
-		.data		= &sysctl_dccp_feat_ack_ratio,
-		.maxlen		= sizeof(sysctl_dccp_feat_ack_ratio),
-		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
-	},
-	{
-		.procname	= "send_ackvec",
-		.data		= &sysctl_dccp_feat_send_ack_vector,
-		.maxlen		= sizeof(sysctl_dccp_feat_send_ack_vector),
-		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
-	},
-	{
-		.procname	= "send_ndp",
-		.data		= &sysctl_dccp_feat_send_ndp_count,
-		.maxlen		= sizeof(sysctl_dccp_feat_send_ndp_count),
+		.data		= &sysctl_dccp_tx_ccid,
+		.maxlen		= sizeof(sysctl_dccp_tx_ccid),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &u8_max,		/* RFC 4340, 10. */
 	},
 	{
 		.procname	= "request_retries",
 		.data		= &sysctl_dccp_request_retries,
 		.maxlen		= sizeof(sysctl_dccp_request_retries),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &u8_max,
 	},
 	{
 		.procname	= "retries1",
 		.data		= &sysctl_dccp_retries1,
 		.maxlen		= sizeof(sysctl_dccp_retries1),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &u8_max,
 	},
 	{
 		.procname	= "retries2",
 		.data		= &sysctl_dccp_retries2,
 		.maxlen		= sizeof(sysctl_dccp_retries2),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &u8_max,
 	},
 	{
 		.procname	= "tx_qlen",
 		.data		= &sysctl_dccp_tx_qlen,
 		.maxlen		= sizeof(sysctl_dccp_tx_qlen),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
 	},
 	{
 		.procname	= "sync_ratelimit",

net/dccp/timer.c

@@ -87,17 +87,6 @@ static void dccp_retransmit_timer(struct sock *sk)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 
-	/* retransmit timer is used for feature negotiation throughout
-	 * connection.  In this case, no packet is re-transmitted, but rather an
-	 * ack is generated and pending changes are placed into its options.
-	 */
-	if (sk->sk_send_head == NULL) {
-		dccp_pr_debug("feat negotiation retransmit timeout %p\n", sk);
-		if (sk->sk_state == DCCP_OPEN)
-			dccp_send_ack(sk);
-		goto backoff;
-	}
-
 	/*
 	 * More than than 4MSL (8 minutes) has passed, a RESET(aborted) was
 	 * sent, no need to retransmit, this sock is dead.
@@ -126,7 +115,6 @@ static void dccp_retransmit_timer(struct sock *sk)
 		return;
 	}
 
-backoff:
 	icsk->icsk_backoff++;
 
 	icsk->icsk_rto = min(icsk->icsk_rto << 1, DCCP_RTO_MAX);
@@ -249,32 +237,35 @@ out:
 	sock_put(sk);
 }
 
-/* Transmit-delay timer: used by the CCIDs to delay actual send time */
-static void dccp_write_xmit_timer(unsigned long data)
+/**
+ * dccp_write_xmitlet  -  Workhorse for CCID packet dequeueing interface
+ * See the comments above %ccid_dequeueing_decision for supported modes.
+ */
+static void dccp_write_xmitlet(unsigned long data)
 {
 	struct sock *sk = (struct sock *)data;
-	struct dccp_sock *dp = dccp_sk(sk);
 
 	bh_lock_sock(sk);
 	if (sock_owned_by_user(sk))
-		sk_reset_timer(sk, &dp->dccps_xmit_timer, jiffies+1);
+		sk_reset_timer(sk, &dccp_sk(sk)->dccps_xmit_timer, jiffies + 1);
 	else
-		dccp_write_xmit(sk, 0);
+		dccp_write_xmit(sk);
 	bh_unlock_sock(sk);
-	sock_put(sk);
 }
 
-static void dccp_init_write_xmit_timer(struct sock *sk)
+static void dccp_write_xmit_timer(unsigned long data)
 {
-	struct dccp_sock *dp = dccp_sk(sk);
-
-	setup_timer(&dp->dccps_xmit_timer, dccp_write_xmit_timer,
-			(unsigned long)sk);
+	dccp_write_xmitlet(data);
+	sock_put((struct sock *)data);
 }
 
 void dccp_init_xmit_timers(struct sock *sk)
 {
-	dccp_init_write_xmit_timer(sk);
+	struct dccp_sock *dp = dccp_sk(sk);
+
+	tasklet_init(&dp->dccps_xmitlet, dccp_write_xmitlet, (unsigned long)sk);
+	setup_timer(&dp->dccps_xmit_timer, dccp_write_xmit_timer,
+							     (unsigned long)sk);
 	inet_csk_init_xmit_timers(sk, &dccp_write_timer, &dccp_delack_timer,
 				  &dccp_keepalive_timer);
 }
@@ -290,8 +281,7 @@ u32 dccp_timestamp(void)
 {
 	s64 delta = ktime_us_delta(ktime_get_real(), dccp_timestamp_seed);
 
-	do_div(delta, 10);
-	return delta;
+	return div_u64(delta, DCCP_TIME_RESOLUTION);
 }
 EXPORT_SYMBOL_GPL(dccp_timestamp);
 

net/ipv4/tcp_input.c

@@ -811,25 +811,12 @@ void tcp_update_metrics(struct sock *sk)
 	}
 }
 
-/* Numbers are taken from RFC3390.
- *
- * John Heffner states:
- *
- *	The RFC specifies a window of no more than 4380 bytes
- *	unless 2*MSS > 4380.  Reading the pseudocode in the RFC
- *	is a bit misleading because they use a clamp at 4380 bytes
- *	rather than use a multiplier in the relevant range.
- */
 __u32 tcp_init_cwnd(struct tcp_sock *tp, struct dst_entry *dst)
 {
 	__u32 cwnd = (dst ? dst_metric(dst, RTAX_INITCWND) : 0);
 
-	if (!cwnd) {
-		if (tp->mss_cache > 1460)
-			cwnd = 2;
-		else
-			cwnd = (tp->mss_cache > 1095) ? 3 : 4;
-	}
+	if (!cwnd)
+		cwnd = rfc3390_bytes_to_packets(tp->mss_cache);
 	return min_t(__u32, cwnd, tp->snd_cwnd_clamp);
 }