[DCCP]: Move the ack vector code to net/dccp/ackvec.[ch]

Isolating it, that will be used when we introduce a CCID2 (TCP-Like)
implementation.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/dccp.h

@@ -353,14 +353,8 @@ static inline struct dccp_request_sock *dccp_rsk(const struct request_sock *req)
 
 extern struct inet_timewait_death_row dccp_death_row;
 
-/* Read about the ECN nonce to see why it is 253 */
-#define DCCP_MAX_ACK_VECTOR_LEN 253
-
 struct dccp_options_received {
-	u32	dccpor_ndp:24,
-		dccpor_ack_vector_len:8;
-	u32	dccpor_ack_vector_idx:10;
-	/* 22 bits hole, try to pack */
+	u32	dccpor_ndp; /* only 24 bits */
 	u32	dccpor_timestamp;
 	u32	dccpor_timestamp_echo;
 	u32	dccpor_elapsed_time;
@@ -394,6 +388,8 @@ static inline int dccp_list_has_service(const struct dccp_service_list *sl,
 	return 0;
 }
 
+struct dccp_ackvec;
+
 /**
  * struct dccp_sock - DCCP socket state
  *
@@ -414,7 +410,7 @@ static inline int dccp_list_has_service(const struct dccp_service_list *sl,
  * @dccps_packet_size - Set thru setsockopt
  * @dccps_role - Role of this sock, one of %dccp_role
  * @dccps_ndp_count - number of Non Data Packets since last data packet
- * @dccps_hc_rx_ackpkts - receiver half connection acked packets
+ * @dccps_hc_rx_ackvec - rx half connection ack vector
  */
 struct dccp_sock {
 	/* inet_connection_sock has to be the first member of dccp_sock */
@@ -439,7 +435,7 @@ struct dccp_sock {
 	__u32				dccps_pmtu_cookie;
 	__u32				dccps_mss_cache;
 	struct dccp_options		dccps_options;
-	struct dccp_ackpkts		*dccps_hc_rx_ackpkts;
+	struct dccp_ackvec		*dccps_hc_rx_ackvec;
 	void				*dccps_hc_rx_ccid_private;
 	void				*dccps_hc_tx_ccid_private;
 	struct ccid			*dccps_hc_rx_ccid;

net/dccp/Makefile

@@ -3,6 +3,8 @@ obj-$(CONFIG_IP_DCCP) += dccp.o
 dccp-y := ccid.o input.o ipv4.o minisocks.o options.o output.o proto.o \
 	  timer.o
 
+dccp-$(CONFIG_IP_DCCP_ACKVEC) += ackvec.o
+
 obj-$(CONFIG_INET_DCCP_DIAG) += dccp_diag.o
 
 dccp_diag-y := diag.o

net/dccp/ackvec.c

@@ -0,0 +1,419 @@
+/*
+ *  net/dccp/ackvec.c
+ *
+ *  An implementation of the DCCP protocol
+ *  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/skbuff.h>
+
+#include <net/sock.h>
+
+int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb)
+{
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_ackvec *av = dp->dccps_hc_rx_ackvec;
+	int len = av->dccpav_vec_len + 2;
+	struct timeval now;
+	u32 elapsed_time;
+	unsigned char *to, *from;
+
+	dccp_timestamp(sk, &now);
+	elapsed_time = timeval_delta(&now, &av->dccpav_time) / 10;
+
+	if (elapsed_time != 0)
+		dccp_insert_option_elapsed_time(sk, skb, elapsed_time);
+
+	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN)
+		return -1;
+
+	/*
+	 * XXX: now we have just one ack vector sent record, so
+	 * we have to wait for it to be cleared.
+	 *
+	 * Of course this is not acceptable, but this is just for
+	 * basic testing now.
+	 */
+	if (av->dccpav_ack_seqno != DCCP_MAX_SEQNO + 1)
+		return -1;
+
+	DCCP_SKB_CB(skb)->dccpd_opt_len += len;
+
+	to    = skb_push(skb, len);
+	*to++ = DCCPO_ACK_VECTOR_0;
+	*to++ = len;
+
+	len  = av->dccpav_vec_len;
+	from = av->dccpav_buf + av->dccpav_buf_head;
+
+	/* Check if buf_head wraps */
+	if (av->dccpav_buf_head + len > av->dccpav_vec_len) {
+		const u32 tailsize = (av->dccpav_vec_len - av->dccpav_buf_head);
+
+		memcpy(to, from, tailsize);
+		to   += tailsize;
+		len  -= tailsize;
+		from = av->dccpav_buf;
+	}
+
+	memcpy(to, from, len);
+	/*
+	 *	From draft-ietf-dccp-spec-11.txt:
+	 *
+	 *	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.
+	 *
+	 * This implemention uses just one ack record for now.
+	 */
+	av->dccpav_ack_seqno = DCCP_SKB_CB(skb)->dccpd_seq;
+	av->dccpav_ack_ptr   = av->dccpav_buf_head;
+	av->dccpav_ack_ackno = av->dccpav_buf_ackno;
+	av->dccpav_ack_nonce = av->dccpav_buf_nonce;
+	av->dccpav_sent_len  = av->dccpav_vec_len;
+
+	dccp_pr_debug("%sACK Vector 0, len=%d, ack_seqno=%llu, "
+		      "ack_ackno=%llu\n",
+		      debug_prefix, av->dccpav_sent_len,
+		      (unsigned long long)av->dccpav_ack_seqno,
+		      (unsigned long long)av->dccpav_ack_ackno);
+	return -1;
+}
+
+struct dccp_ackvec *dccp_ackvec_alloc(const unsigned int len,
+				      const unsigned int __nocast priority)
+{
+	struct dccp_ackvec *av = kmalloc(sizeof(*av) + len, priority);
+
+	if (av != NULL) {
+		av->dccpav_buf_len	= len;
+		av->dccpav_buf_head	=
+			av->dccpav_buf_tail = av->dccpav_buf_len - 1;
+		av->dccpav_buf_ackno	=
+			av->dccpav_ack_ackno = av->dccpav_ack_seqno = ~0LLU;
+		av->dccpav_buf_nonce = av->dccpav_buf_nonce = 0;
+		av->dccpav_ack_ptr	= 0;
+		av->dccpav_time.tv_sec	= 0;
+		av->dccpav_time.tv_usec	= 0;
+		av->dccpav_sent_len	= av->dccpav_vec_len = 0;
+	}
+
+	return av;
+}
+
+void dccp_ackvec_free(struct dccp_ackvec *av)
+{
+	kfree(av);
+}
+
+static inline u8 dccp_ackvec_state(const struct dccp_ackvec *av,
+				   const unsigned int index)
+{
+	return av->dccpav_buf[index] & DCCP_ACKVEC_STATE_MASK;
+}
+
+static inline u8 dccp_ackvec_len(const struct dccp_ackvec *av,
+				 const unsigned int index)
+{
+	return av->dccpav_buf[index] & DCCP_ACKVEC_LEN_MASK;
+}
+
+/*
+ * 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.
+ */
+static inline int dccp_ackvec_set_buf_head_state(struct dccp_ackvec *av,
+						 const unsigned int packets,
+						  const unsigned char state)
+{
+	unsigned int gap;
+	signed long new_head;
+
+	if (av->dccpav_vec_len + packets > av->dccpav_buf_len)
+		return -ENOBUFS;
+
+	gap	 = packets - 1;
+	new_head = av->dccpav_buf_head - packets;
+
+	if (new_head < 0) {
+		if (gap > 0) {
+			memset(av->dccpav_buf, DCCP_ACKVEC_STATE_NOT_RECEIVED,
+			       gap + new_head + 1);
+			gap = -new_head;
+		}
+		new_head += av->dccpav_buf_len;
+	} 
+
+	av->dccpav_buf_head = new_head;
+
+	if (gap > 0)
+		memset(av->dccpav_buf + av->dccpav_buf_head + 1,
+		       DCCP_ACKVEC_STATE_NOT_RECEIVED, gap);
+
+	av->dccpav_buf[av->dccpav_buf_head] = state;
+	av->dccpav_vec_len += packets;
+	return 0;
+}
+
+/*
+ * Implements the draft-ietf-dccp-spec-11.txt Appendix A
+ */
+int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk,
+		    const u64 ackno, const u8 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 dccpav_buf.
+	 *
+	 * We may well decide to do buffer compression, etc, but for now lets
+	 * just drop.
+	 *
+	 * From Appendix A:
+	 *
+	 *	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.
+	 */
+
+	/* See if this is the first ackno being inserted */
+	if (av->dccpav_vec_len == 0) {
+		av->dccpav_buf[av->dccpav_buf_head] = state;
+		av->dccpav_vec_len = 1;
+	} else if (after48(ackno, av->dccpav_buf_ackno)) {
+		const u64 delta = dccp_delta_seqno(av->dccpav_buf_ackno,
+						   ackno);
+
+		/*
+		 * Look if the state of this packet is the same as the
+		 * previous ackno and if so if we can bump the head len.
+		 */
+		if (delta == 1 &&
+		    dccp_ackvec_state(av, av->dccpav_buf_head) == state &&
+		    (dccp_ackvec_len(av, av->dccpav_buf_head) <
+		     DCCP_ACKVEC_LEN_MASK))
+			av->dccpav_buf[av->dccpav_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 arrives, and
+		 *	S <= buf_ackno, 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->dccpav_buf_ackno);
+		unsigned int index = av->dccpav_buf_head;
+
+		while (1) {
+			const u8 len = dccp_ackvec_len(av, index);
+			const u8 state = dccp_ackvec_state(av, index);
+			/*
+			 * valid packets not yet in dccpav_buf have a reserved
+			 * entry, with a len equal to 0.
+			 */
+			if (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->dccpav_buf[index] = state;
+				goto out;
+			}
+			/* len == 0 means one packet */
+			if (delta < len + 1)
+				goto out_duplicate;
+
+			delta -= len + 1;
+			if (++index == av->dccpav_buf_len)
+				index = 0;
+		}
+	}
+
+	av->dccpav_buf_ackno = ackno;
+	dccp_timestamp(sk, &av->dccpav_time);
+out:
+	dccp_pr_debug("");
+	return 0;
+
+out_duplicate:
+	/* Duplicate packet */
+	dccp_pr_debug("Received a dup or already considered lost "
+		      "packet: %llu\n", (unsigned long long)ackno);
+	return -EILSEQ;
+}
+
+#ifdef CONFIG_IP_DCCP_DEBUG
+void dccp_ackvector_print(const u64 ackno, const unsigned char *vector, int len)
+{
+	if (!dccp_debug)
+		return;
+
+	printk("ACK vector len=%d, ackno=%llu |", len,
+	       (unsigned long long)ackno);
+
+	while (len--) {
+		const u8 state = (*vector & DCCP_ACKVEC_STATE_MASK) >> 6;
+		const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
+
+		printk("%d,%d|", state, rl);
+		++vector;
+	}
+
+	printk("\n");
+}
+
+void dccp_ackvec_print(const struct dccp_ackvec *av)
+{
+	dccp_ackvector_print(av->dccpav_buf_ackno,
+			     av->dccpav_buf + av->dccpav_buf_head,
+			     av->dccpav_vec_len);
+}
+#endif
+
+static void dccp_ackvec_trow_away_ack_record(struct dccp_ackvec *av)
+{
+	/*
+	 * As we're keeping track of the ack vector size (dccpav_vec_len) and
+	 * the sent ack vector size (dccpav_sent_len) we don't need
+	 * dccpav_buf_tail at all, but keep this code here as in the future
+	 * we'll implement a vector of ack records, as suggested in
+	 * draft-ietf-dccp-spec-11.txt Appendix A. -acme
+	 */
+#if 0
+	av->dccpav_buf_tail = av->dccpav_ack_ptr + 1;
+	if (av->dccpav_buf_tail >= av->dccpav_vec_len)
+		av->dccpav_buf_tail -= av->dccpav_vec_len;
+#endif
+	av->dccpav_vec_len -= av->dccpav_sent_len;
+}
+
+void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, struct sock *sk,
+				 const u64 ackno)
+{
+	/* Check if we actually sent an ACK vector */
+	if (av->dccpav_ack_seqno == DCCP_MAX_SEQNO + 1)
+		return;
+
+	if (ackno == av->dccpav_ack_seqno) {
+#ifdef CONFIG_IP_DCCP_DEBUG
+		struct dccp_sock *dp = dccp_sk(sk);
+		const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ?
+					"CLIENT rx ack: " : "server rx ack: ";
+#endif
+		dccp_pr_debug("%sACK packet 0, len=%d, ack_seqno=%llu, "
+			      "ack_ackno=%llu, ACKED!\n",
+			      debug_prefix, 1,
+			      (unsigned long long)av->dccpav_ack_seqno,
+			      (unsigned long long)av->dccpav_ack_ackno);
+		dccp_ackvec_trow_away_ack_record(av);
+		av->dccpav_ack_seqno = DCCP_MAX_SEQNO + 1;
+	}
+}
+
+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;
+
+	/* Check if we actually sent an ACK vector */
+	if (av->dccpav_ack_seqno == DCCP_MAX_SEQNO + 1)
+		return;
+	/*
+	 * We're in the receiver half connection, so if the received an ACK
+	 * vector ackno (e.g. 50) before dccpav_ack_seqno (e.g. 52), we're
+	 * not interested.
+	 *
+	 * Extra explanation with example:
+	 * 
+	 * if we received an ACK vector with ackno 50, it can only be acking
+	 * 50, 49, 48, etc, not 52 (the seqno for the ACK vector we sent).
+	 */
+	/* dccp_pr_debug("is %llu < %llu? ", ackno, av->dccpav_ack_seqno); */
+	if (before48(ackno, av->dccpav_ack_seqno)) {
+		/* dccp_pr_debug_cat("yes\n"); */
+		return;
+	}
+	/* dccp_pr_debug_cat("no\n"); */
+
+	i = len;
+	while (i--) {
+		const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
+		u64 ackno_end_rl;
+
+		dccp_set_seqno(&ackno_end_rl, ackno - rl);
+
+		/*
+		 * dccp_pr_debug("is %llu <= %llu <= %llu? ", ackno_end_rl,
+		 * av->dccpav_ack_seqno, ackno);
+		 */
+		if (between48(av->dccpav_ack_seqno, ackno_end_rl, ackno)) {
+			const u8 state = (*vector &
+					  DCCP_ACKVEC_STATE_MASK) >> 6;
+			/* dccp_pr_debug_cat("yes\n"); */
+
+			if (state != DCCP_ACKVEC_STATE_NOT_RECEIVED) {
+#ifdef CONFIG_IP_DCCP_DEBUG
+				struct dccp_sock *dp = dccp_sk(sk);
+				const char *debug_prefix =
+					dp->dccps_role == DCCP_ROLE_CLIENT ?
+					"CLIENT rx ack: " : "server rx ack: ";
+#endif
+				dccp_pr_debug("%sACK vector 0, len=%d, "
+					      "ack_seqno=%llu, ack_ackno=%llu, "
+					      "ACKED!\n",
+					      debug_prefix, len,
+					      (unsigned long long)
+					      av->dccpav_ack_seqno,
+					      (unsigned long long)
+					      av->dccpav_ack_ackno);
+				dccp_ackvec_trow_away_ack_record(av);
+			}
+			/*
+			 * If dccpav_ack_seqno was not received, no problem
+			 * we'll send another ACK vector.
+			 */
+			av->dccpav_ack_seqno = DCCP_MAX_SEQNO + 1;
+			break;
+		}
+		/* dccp_pr_debug_cat("no\n"); */
+
+		dccp_set_seqno(&ackno, ackno_end_rl - 1);
+		++vector;
+	}
+}
+
+int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb,
+		      const u8 opt, const u8 *value, const u8 len)
+{
+	if (len > DCCP_MAX_ACKVEC_LEN)
+		return -1;
+
+	/* 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,
+					DCCP_SKB_CB(skb)->dccpd_ack_seq,
+				        len, value);
+	return 0;
+}

net/dccp/ackvec.h

@@ -0,0 +1,133 @@
+#ifndef _ACKVEC_H
+#define _ACKVEC_H
+/*
+ *  net/dccp/ackvec.h
+ *
+ *  An implementation of the DCCP protocol
+ *  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/config.h>
+#include <linux/compiler.h>
+#include <linux/time.h>
+#include <linux/types.h>
+
+/* Read about the ECN nonce to see why it is 253 */
+#define DCCP_MAX_ACKVEC_LEN 253
+
+#define DCCP_ACKVEC_STATE_RECEIVED	0
+#define DCCP_ACKVEC_STATE_ECN_MARKED	(1 << 6)
+#define DCCP_ACKVEC_STATE_NOT_RECEIVED	(3 << 6)
+
+#define DCCP_ACKVEC_STATE_MASK		0xC0 /* 11000000 */
+#define DCCP_ACKVEC_LEN_MASK		0x3F /* 00111111 */
+
+/** struct dccp_ackvec - ack vector
+ *
+ * This data structure is the one defined in the DCCP draft
+ * Appendix A.
+ *
+ * @dccpav_buf_head - circular buffer head
+ * @dccpav_buf_tail - circular buffer tail
+ * @dccpav_buf_ackno - ack # of the most recent packet acknowledgeable in the
+ * 		       buffer (i.e. %dccpav_buf_head)
+ * @dccpav_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:
+ *
+ * @dccpav_ack_seqno - the Sequence Number used for the packet
+ * 		       (HC-Receiver seqno)
+ * @dccpav_ack_ptr - the value of buf_head at the time of acknowledgement.
+ * @dccpav_ack_ackno - the Acknowledgement Number used for the packet
+ * 		       (HC-Sender seqno)
+ * @dccpav_ack_nonce - the one-bit sum of the ECN Nonces for all State 0.
+ *
+ * @dccpav_buf_len	- circular buffer length
+ * @dccpav_time		- the time in usecs
+ * @dccpav_buf - circular buffer of acknowledgeable packets
+ */
+struct dccp_ackvec {
+	unsigned int	dccpav_buf_head;
+	unsigned int	dccpav_buf_tail;
+	u64		dccpav_buf_ackno;
+	u64		dccpav_ack_seqno;
+	u64		dccpav_ack_ackno;
+	unsigned int	dccpav_ack_ptr;
+	unsigned int	dccpav_sent_len;
+	unsigned int	dccpav_vec_len;
+	unsigned int	dccpav_buf_len;
+	struct timeval	dccpav_time;
+	u8		dccpav_buf_nonce;
+	u8		dccpav_ack_nonce;
+	u8		dccpav_buf[0];
+};
+
+struct sock;
+struct sk_buff;
+
+#ifdef CONFIG_IP_DCCP_ACKVEC
+extern struct dccp_ackvec *dccp_ackvec_alloc(unsigned int len,
+					  const unsigned int __nocast 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,
+			     const u8 opt, const u8 *value, const u8 len);
+
+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->dccpav_sent_len != av->dccpav_vec_len;
+}
+#else /* CONFIG_IP_DCCP_ACKVEC */
+static inline struct dccp_ackvec *dccp_ackvec_alloc(unsigned int len,
+					   const unsigned int __nocast 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 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;
+}
+
+static inline int dccp_ackvec_pending(const struct dccp_ackvec *av)
+{
+	return 0;
+}
+#endif /* CONFIG_IP_DCCP_ACKVEC */
+#endif /* _ACKVEC_H */

net/dccp/dccp.h

@@ -17,6 +17,7 @@
 #include <net/snmp.h>
 #include <net/sock.h>
 #include <net/tcp.h>
+#include "ackvec.h"
 
 #ifdef CONFIG_IP_DCCP_DEBUG
 extern int dccp_debug;
@@ -358,6 +359,17 @@ static inline void dccp_update_gss(struct sock *sk, u64 seq)
 		       (dp->dccps_gss -
 			dp->dccps_options.dccpo_sequence_window + 1));
 }
+				
+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
+	       (dp->dccps_options.dccpo_send_ack_vector &&
+		dccp_ackvec_pending(dp->dccps_hc_rx_ackvec)) ||
+#endif
+	       inet_csk_ack_scheduled(sk);
+}
 
 extern void dccp_insert_options(struct sock *sk, struct sk_buff *skb);
 extern void dccp_insert_option_elapsed_time(struct sock *sk,
@@ -371,65 +383,6 @@ extern void dccp_insert_option(struct sock *sk, struct sk_buff *skb,
 
 extern struct socket *dccp_ctl_socket;
 
-#define DCCP_ACKPKTS_STATE_RECEIVED	0
-#define DCCP_ACKPKTS_STATE_ECN_MARKED	(1 << 6)
-#define DCCP_ACKPKTS_STATE_NOT_RECEIVED	(3 << 6)
-
-#define DCCP_ACKPKTS_STATE_MASK		0xC0 /* 11000000 */
-#define DCCP_ACKPKTS_LEN_MASK		0x3F /* 00111111 */
-
-/** struct dccp_ackpkts - acknowledgeable packets
- *
- * This data structure is the one defined in the DCCP draft
- * Appendix A.
- *
- * @dccpap_buf_head - circular buffer head
- * @dccpap_buf_tail - circular buffer tail
- * @dccpap_buf_ackno - ack # of the most recent packet acknowledgeable in the
- * 		       buffer (i.e. %dccpap_buf_head)
- * @dccpap_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:
- *
- * @dccpap_ack_seqno - the Sequence Number used for the packet
- * 		       (HC-Receiver seqno)
- * @dccpap_ack_ptr - the value of buf_head at the time of acknowledgement.
- * @dccpap_ack_ackno - the Acknowledgement Number used for the packet
- * 		       (HC-Sender seqno)
- * @dccpap_ack_nonce - the one-bit sum of the ECN Nonces for all State 0.
- *
- * @dccpap_buf_len - circular buffer length
- * @dccpap_time		- the time in usecs
- * @dccpap_buf - circular buffer of acknowledgeable packets
- */
-struct dccp_ackpkts {
-	unsigned int		dccpap_buf_head;
-	unsigned int		dccpap_buf_tail;
-	u64			dccpap_buf_ackno;
-	u64			dccpap_ack_seqno;
-	u64			dccpap_ack_ackno;
-	unsigned int		dccpap_ack_ptr;
-	unsigned int		dccpap_buf_vector_len;
-	unsigned int		dccpap_ack_vector_len;
-	unsigned int		dccpap_buf_len;
-	struct timeval		dccpap_time;
-	u8			dccpap_buf_nonce;
-	u8			dccpap_ack_nonce;
-	u8			dccpap_buf[0];
-};
-
-extern struct dccp_ackpkts *
-		dccp_ackpkts_alloc(unsigned int len,
-				  const unsigned int __nocast priority);
-extern void dccp_ackpkts_free(struct dccp_ackpkts *ap);
-extern int dccp_ackpkts_add(struct dccp_ackpkts *ap, const struct sock *sk,
-			    u64 ackno, u8 state);
-extern void dccp_ackpkts_check_rcv_ackno(struct dccp_ackpkts *ap,
-					 struct sock *sk, u64 ackno);
-
 extern void dccp_timestamp(const struct sock *sk, struct timeval *tv);
 
 static inline suseconds_t timeval_usecs(const struct timeval *tv)
@@ -470,15 +423,4 @@ static inline void timeval_sub_usecs(struct timeval *tv,
 	}
 }
 
-#ifdef CONFIG_IP_DCCP_DEBUG
-extern void dccp_ackvector_print(const u64 ackno,
-				 const unsigned char *vector, int len);
-extern void dccp_ackpkts_print(const struct dccp_ackpkts *ap);
-#else
-static inline void dccp_ackvector_print(const u64 ackno,
-					const unsigned char *vector,
-					int len) { }
-static inline void dccp_ackpkts_print(const struct dccp_ackpkts *ap) { }
-#endif
-
 #endif /* _DCCP_H */

net/dccp/input.c

@@ -16,6 +16,7 @@
 
 #include <net/sock.h>
 
+#include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
 
@@ -60,8 +61,8 @@ static inline void dccp_event_ack_recv(struct sock *sk, struct sk_buff *skb)
 	struct dccp_sock *dp = dccp_sk(sk);
 
 	if (dp->dccps_options.dccpo_send_ack_vector)
-		dccp_ackpkts_check_rcv_ackno(dp->dccps_hc_rx_ackpkts, sk,
-					     DCCP_SKB_CB(skb)->dccpd_ack_seq);
+		dccp_ackvec_check_rcv_ackno(dp->dccps_hc_rx_ackvec, sk,
+					    DCCP_SKB_CB(skb)->dccpd_ack_seq);
 }
 
 static int dccp_check_seqno(struct sock *sk, struct sk_buff *skb)
@@ -164,37 +165,11 @@ int dccp_rcv_established(struct sock *sk, struct sk_buff *skb,
 	if (DCCP_SKB_CB(skb)->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
 		dccp_event_ack_recv(sk, skb);
 
-	/*
-	 * FIXME: check ECN to see if we should use
-	 * DCCP_ACKPKTS_STATE_ECN_MARKED
-	 */
-	if (dp->dccps_options.dccpo_send_ack_vector) {
-		struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts;
-
-		if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk,
-				     DCCP_SKB_CB(skb)->dccpd_seq,
-				     DCCP_ACKPKTS_STATE_RECEIVED)) {
-			LIMIT_NETDEBUG(KERN_WARNING "DCCP: acknowledgeable "
-						    "packets buffer full!\n");
-			ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1;
-			inet_csk_schedule_ack(sk);
-			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
-						  TCP_DELACK_MIN,
-						  DCCP_RTO_MAX);
-			goto discard;
-		}
-
-		/*
-		 * FIXME: this activation is probably wrong, have to study more
-		 * TCP delack machinery and how it fits into DCCP draft, but
-		 * for now it kinda "works" 8)
-		 */
-		if (!inet_csk_ack_scheduled(sk)) {
-			inet_csk_schedule_ack(sk);
-			inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, 5 * HZ,
-						  DCCP_RTO_MAX);
-		}
-	}
+	if (dp->dccps_options.dccpo_send_ack_vector &&
+	    dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk,
+			    DCCP_SKB_CB(skb)->dccpd_seq,
+			    DCCP_ACKVEC_STATE_RECEIVED))
+		goto discard;
 
 	ccid_hc_rx_packet_recv(dp->dccps_hc_rx_ccid, sk, skb);
 	ccid_hc_tx_packet_recv(dp->dccps_hc_tx_ccid, sk, skb);
@@ -495,29 +470,11 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		ccid_hc_rx_packet_recv(dp->dccps_hc_rx_ccid, sk, skb);
 		ccid_hc_tx_packet_recv(dp->dccps_hc_tx_ccid, sk, skb);
 
-		/*
-		 * FIXME: check ECN to see if we should use
-		 * DCCP_ACKPKTS_STATE_ECN_MARKED
-		 */
-		if (dp->dccps_options.dccpo_send_ack_vector) {
-			if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk,
-					     dcb->dccpd_seq,
-					     DCCP_ACKPKTS_STATE_RECEIVED))
-				goto discard;
-			/*
-			 * FIXME: this activation is probably wrong, have to
-			 * study more TCP delack machinery and how it fits into
-			 * DCCP draft, but for now it kinda "works" 8)
-			 */
-			if ((dp->dccps_hc_rx_ackpkts->dccpap_ack_seqno ==
-			     DCCP_MAX_SEQNO + 1) &&
-			    !inet_csk_ack_scheduled(sk)) {
-				inet_csk_schedule_ack(sk);
-				inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
-							  TCP_DELACK_MIN,
-							  DCCP_RTO_MAX);
-			}
-		}
+ 		if (dp->dccps_options.dccpo_send_ack_vector &&
+		    dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk,
+ 				    DCCP_SKB_CB(skb)->dccpd_seq,
+ 				    DCCP_ACKVEC_STATE_RECEIVED))
+ 			goto discard;
 	}
 
 	/*

net/dccp/ipv4.c

@@ -23,6 +23,7 @@
 #include <net/tcp_states.h>
 #include <net/xfrm.h>
 
+#include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
 
@@ -1112,45 +1113,7 @@ int dccp_v4_rcv(struct sk_buff *skb)
 		goto discard_it;
 
 	dh = dccp_hdr(skb);
-#if 0
-	/*
-	 * Use something like this to simulate some DATA/DATAACK loss to test
-	 * dccp_ackpkts_add, you'll get something like this on a session that
-	 * sends 10 DATA/DATAACK packets:
-	 *
-	 * ackpkts_print: 281473596467422 |0,0|3,0|0,0|3,0|0,0|3,0|0,0|3,0|0,1|
-	 *
-	 * 0, 0 means: DCCP_ACKPKTS_STATE_RECEIVED, RLE == just this packet
-	 * 0, 1 means: DCCP_ACKPKTS_STATE_RECEIVED, RLE == two adjacent packets
-	 * 						   with the same state
-	 * 3, 0 means: DCCP_ACKPKTS_STATE_NOT_RECEIVED, RLE == just this packet
-	 *
-	 * So...
-	 *
-	 * 281473596467422 was received
-	 * 281473596467421 was not received
-	 * 281473596467420 was received
-	 * 281473596467419 was not received
-	 * 281473596467418 was received
-	 * 281473596467417 was not received
-	 * 281473596467416 was received
-	 * 281473596467415 was not received
-	 * 281473596467414 was received
-	 * 281473596467413 was received (this one was the 3way handshake
-	 * 				 RESPONSE)
-	 *
-	 */
-	if (dh->dccph_type == DCCP_PKT_DATA ||
-	    dh->dccph_type == DCCP_PKT_DATAACK) {
-		static int discard = 0;
 
-		if (discard) {
-			discard = 0;
-			goto discard_it;
-		}
-		discard = 1;
-	}
-#endif
 	DCCP_SKB_CB(skb)->dccpd_seq  = dccp_hdr_seq(skb);
 	DCCP_SKB_CB(skb)->dccpd_type = dh->dccph_type;
 
@@ -1264,11 +1227,9 @@ static int dccp_v4_init_sock(struct sock *sk)
 	do_gettimeofday(&dp->dccps_epoch);
 
 	if (dp->dccps_options.dccpo_send_ack_vector) {
-		dp->dccps_hc_rx_ackpkts =
-			dccp_ackpkts_alloc(DCCP_MAX_ACK_VECTOR_LEN,
-					   GFP_KERNEL);
-
-		if (dp->dccps_hc_rx_ackpkts == NULL)
+		dp->dccps_hc_rx_ackvec = dccp_ackvec_alloc(DCCP_MAX_ACKVEC_LEN,
+							   GFP_KERNEL);
+		if (dp->dccps_hc_rx_ackvec == NULL)
 			return -ENOMEM;
 	}
 
@@ -1288,8 +1249,10 @@ static int dccp_v4_init_sock(struct sock *sk)
 		    dp->dccps_hc_tx_ccid == NULL) {
 			ccid_exit(dp->dccps_hc_rx_ccid, sk);
 			ccid_exit(dp->dccps_hc_tx_ccid, sk);
-			dccp_ackpkts_free(dp->dccps_hc_rx_ackpkts);
-			dp->dccps_hc_rx_ackpkts = NULL;
+			if (dp->dccps_options.dccpo_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;
 		}
@@ -1331,8 +1294,10 @@ static int dccp_v4_destroy_sock(struct sock *sk)
 
 	ccid_hc_rx_exit(dp->dccps_hc_rx_ccid, sk);
 	ccid_hc_tx_exit(dp->dccps_hc_tx_ccid, sk);
-	dccp_ackpkts_free(dp->dccps_hc_rx_ackpkts);
-	dp->dccps_hc_rx_ackpkts = NULL;
+	if (dp->dccps_options.dccpo_send_ack_vector) {
+		dccp_ackvec_free(dp->dccps_hc_rx_ackvec);
+		dp->dccps_hc_rx_ackvec = NULL;
+	}
 	ccid_exit(dp->dccps_hc_rx_ccid, sk);
 	ccid_exit(dp->dccps_hc_tx_ccid, sk);
 	dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL;

net/dccp/minisocks.c

@@ -19,6 +19,7 @@
 #include <net/xfrm.h>
 #include <net/inet_timewait_sock.h>
 
+#include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
 
@@ -94,23 +95,23 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 		struct dccp_sock *newdp = dccp_sk(newsk);
 
 		newdp->dccps_role	   = DCCP_ROLE_SERVER;
-		newdp->dccps_hc_rx_ackpkts = NULL;
+		newdp->dccps_hc_rx_ackvec  = NULL;
 		newdp->dccps_service_list  = NULL;
 		newdp->dccps_service	   = dreq->dreq_service;
 		newicsk->icsk_rto	   = DCCP_TIMEOUT_INIT;
 		do_gettimeofday(&newdp->dccps_epoch);
 
 		if (newdp->dccps_options.dccpo_send_ack_vector) {
-			newdp->dccps_hc_rx_ackpkts =
-				dccp_ackpkts_alloc(DCCP_MAX_ACK_VECTOR_LEN,
-						   GFP_ATOMIC);
+			newdp->dccps_hc_rx_ackvec =
+				dccp_ackvec_alloc(DCCP_MAX_ACKVEC_LEN,
+						  GFP_ATOMIC);
 			/*
 			 * XXX: We're using the same CCIDs set on the parent,
 			 * i.e. sk_clone copied the master sock and left the
 			 * CCID pointers for this child, that is why we do the
 			 * __ccid_get calls.
 			 */
-			if (unlikely(newdp->dccps_hc_rx_ackpkts == NULL))
+			if (unlikely(newdp->dccps_hc_rx_ackvec == NULL))
 				goto out_free;
 		}
 
@@ -118,7 +119,7 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 					     newsk) != 0 ||
 			     ccid_hc_tx_init(newdp->dccps_hc_tx_ccid,
 				     	     newsk) != 0)) {
-			dccp_ackpkts_free(newdp->dccps_hc_rx_ackpkts);
+			dccp_ackvec_free(newdp->dccps_hc_rx_ackvec);
 			ccid_hc_rx_exit(newdp->dccps_hc_rx_ccid, newsk);
 			ccid_hc_tx_exit(newdp->dccps_hc_tx_ccid, newsk);
 out_free:

net/dccp/options.c

@@ -18,15 +18,10 @@
 #include <linux/kernel.h>
 #include <linux/skbuff.h>
 
+#include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
 
-static void dccp_ackpkts_check_rcv_ackvector(struct dccp_ackpkts *ap,
-					     struct sock *sk,
-					     const u64 ackno,
-					     const unsigned char len,
-					     const unsigned char *vector);
-
 /* stores the default values for new connection. may be changed with sysctl */
 static const struct dccp_options dccpo_default_values = {
 	.dccpo_sequence_window	  = DCCPF_INITIAL_SEQUENCE_WINDOW,
@@ -113,25 +108,13 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
 				      opt_recv->dccpor_ndp);
 			break;
 		case DCCPO_ACK_VECTOR_0:
-			if (len > DCCP_MAX_ACK_VECTOR_LEN)
-				goto out_invalid_option;
-
+		case DCCPO_ACK_VECTOR_1:
 			if (pkt_type == DCCP_PKT_DATA)
 				continue;
 
-			opt_recv->dccpor_ack_vector_len = len;
-			opt_recv->dccpor_ack_vector_idx = value - options;
-
-			dccp_pr_debug("%sACK vector 0, len=%d, ack_ackno=%llu\n",
-				      debug_prefix, len,
-				      (unsigned long long)
-				      DCCP_SKB_CB(skb)->dccpd_ack_seq);
-			dccp_ackvector_print(DCCP_SKB_CB(skb)->dccpd_ack_seq,
-					     value, len);
-			dccp_ackpkts_check_rcv_ackvector(dp->dccps_hc_rx_ackpkts,
-							 sk,
-						 DCCP_SKB_CB(skb)->dccpd_ack_seq,
-							 len, value);
+			if (dp->dccps_options.dccpo_send_ack_vector &&
+			    dccp_ackvec_parse(sk, skb, opt, value, len))
+				goto out_invalid_option;
 			break;
 		case DCCPO_TIMESTAMP:
 			if (len != 4)
@@ -352,86 +335,6 @@ void dccp_insert_option_elapsed_time(struct sock *sk,
 
 EXPORT_SYMBOL_GPL(dccp_insert_option_elapsed_time);
 
-static void dccp_insert_option_ack_vector(struct sock *sk, struct sk_buff *skb)
-{
-	struct dccp_sock *dp = dccp_sk(sk);
-#ifdef CONFIG_IP_DCCP_DEBUG
-	const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ?
-					"CLIENT TX opt: " : "server TX opt: ";
-#endif
-	struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts;
-	int len = ap->dccpap_buf_vector_len + 2;
-	struct timeval now;
-	u32 elapsed_time;
-	unsigned char *to, *from;
-
-	dccp_timestamp(sk, &now);
-	elapsed_time = timeval_delta(&now, &ap->dccpap_time) / 10;
-
-	if (elapsed_time != 0)
-		dccp_insert_option_elapsed_time(sk, skb, elapsed_time);
-
-	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) {
-		LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to "
-					 "insert ACK Vector!\n");
-		return;
-	}
-
-	/*
-	 * XXX: now we have just one ack vector sent record, so
-	 * we have to wait for it to be cleared.
-	 *
-	 * Of course this is not acceptable, but this is just for
-	 * basic testing now.
-	 */
-	if (ap->dccpap_ack_seqno != DCCP_MAX_SEQNO + 1)
-		return;
-
-	DCCP_SKB_CB(skb)->dccpd_opt_len += len;
-
-	to    = skb_push(skb, len);
-	*to++ = DCCPO_ACK_VECTOR_0;
-	*to++ = len;
-
-	len  = ap->dccpap_buf_vector_len;
-	from = ap->dccpap_buf + ap->dccpap_buf_head;
-
-	/* Check if buf_head wraps */
-	if (ap->dccpap_buf_head + len > ap->dccpap_buf_len) {
-		const unsigned int tailsize = (ap->dccpap_buf_len -
-					       ap->dccpap_buf_head);
-
-		memcpy(to, from, tailsize);
-		to   += tailsize;
-		len  -= tailsize;
-		from = ap->dccpap_buf;
-	}
-
-	memcpy(to, from, len);
-	/*
-	 *	From draft-ietf-dccp-spec-11.txt:
-	 *
-	 *	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.
-	 *
-	 * This implemention uses just one ack record for now.
-	 */
-	ap->dccpap_ack_seqno	  = DCCP_SKB_CB(skb)->dccpd_seq;
-	ap->dccpap_ack_ptr	  = ap->dccpap_buf_head;
-	ap->dccpap_ack_ackno	  = ap->dccpap_buf_ackno;
-	ap->dccpap_ack_nonce	  = ap->dccpap_buf_nonce;
-	ap->dccpap_ack_vector_len = ap->dccpap_buf_vector_len;
-
-	dccp_pr_debug("%sACK Vector 0, len=%d, ack_seqno=%llu, "
-		      "ack_ackno=%llu\n",
-		      debug_prefix, ap->dccpap_ack_vector_len,
-		      (unsigned long long) ap->dccpap_ack_seqno,
-		      (unsigned long long) ap->dccpap_ack_ackno);
-}
-
 void dccp_timestamp(const struct sock *sk, struct timeval *tv)
 {
 	const struct dccp_sock *dp = dccp_sk(sk);
@@ -528,9 +431,8 @@ void dccp_insert_options(struct sock *sk, struct sk_buff *skb)
 
 	if (!dccp_packet_without_ack(skb)) {
 		if (dp->dccps_options.dccpo_send_ack_vector &&
-		    (dp->dccps_hc_rx_ackpkts->dccpap_buf_ackno !=
-		     DCCP_MAX_SEQNO + 1))
-			dccp_insert_option_ack_vector(sk, skb);
+		    dccp_ackvec_pending(dp->dccps_hc_rx_ackvec))
+			dccp_insert_option_ackvec(sk, skb);
 		if (dp->dccps_timestamp_echo != 0)
 			dccp_insert_option_timestamp_echo(sk, skb);
 	}
@@ -557,331 +459,3 @@ void dccp_insert_options(struct sock *sk, struct sk_buff *skb)
 		}
 	}
 }
-
-struct dccp_ackpkts *dccp_ackpkts_alloc(const unsigned int len,
-				        const unsigned int __nocast priority)
-{
-	struct dccp_ackpkts *ap = kmalloc(sizeof(*ap) + len, priority);
-
-	if (ap != NULL) {
-#ifdef CONFIG_IP_DCCP_DEBUG
-		memset(ap->dccpap_buf, 0xFF, len);
-#endif
-		ap->dccpap_buf_len   = len;
-		ap->dccpap_buf_head  =
-			ap->dccpap_buf_tail =
-				ap->dccpap_buf_len - 1;
-		ap->dccpap_buf_ackno =
-			ap->dccpap_ack_ackno =
-				ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1;
-		ap->dccpap_buf_nonce = ap->dccpap_buf_nonce = 0;
-		ap->dccpap_ack_ptr   = 0;
-		ap->dccpap_time.tv_sec = 0;
-		ap->dccpap_time.tv_usec = 0;
-		ap->dccpap_buf_vector_len = ap->dccpap_ack_vector_len = 0;
-	}
-
-	return ap;
-}
-
-void dccp_ackpkts_free(struct dccp_ackpkts *ap)
-{
-	if (ap != NULL) {
-#ifdef CONFIG_IP_DCCP_DEBUG
-		memset(ap, 0xFF, sizeof(*ap) + ap->dccpap_buf_len);
-#endif
-		kfree(ap);
-	}
-}
-
-static inline u8 dccp_ackpkts_state(const struct dccp_ackpkts *ap,
-				    const unsigned int index)
-{
-	return ap->dccpap_buf[index] & DCCP_ACKPKTS_STATE_MASK;
-}
-
-static inline u8 dccp_ackpkts_len(const struct dccp_ackpkts *ap,
-				  const unsigned int index)
-{
-	return ap->dccpap_buf[index] & DCCP_ACKPKTS_LEN_MASK;
-}
-
-/*
- * 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.
- */
-static inline int dccp_ackpkts_set_buf_head_state(struct dccp_ackpkts *ap,
-						  const unsigned int packets,
-						  const unsigned char state)
-{
-	unsigned int gap;
-	signed long new_head;
-
-	if (ap->dccpap_buf_vector_len + packets > ap->dccpap_buf_len)
-		return -ENOBUFS;
-
-	gap	 = packets - 1;
-	new_head = ap->dccpap_buf_head - packets;
-
-	if (new_head < 0) {
-		if (gap > 0) {
-			memset(ap->dccpap_buf, DCCP_ACKPKTS_STATE_NOT_RECEIVED,
-			       gap + new_head + 1);
-			gap = -new_head;
-		}
-		new_head += ap->dccpap_buf_len;
-	} 
-
-	ap->dccpap_buf_head = new_head;
-
-	if (gap > 0)
-		memset(ap->dccpap_buf + ap->dccpap_buf_head + 1,
-		       DCCP_ACKPKTS_STATE_NOT_RECEIVED, gap);
-
-	ap->dccpap_buf[ap->dccpap_buf_head] = state;
-	ap->dccpap_buf_vector_len += packets;
-	return 0;
-}
-
-/*
- * Implements the draft-ietf-dccp-spec-11.txt Appendix A
- */
-int dccp_ackpkts_add(struct dccp_ackpkts *ap, const struct sock *sk,
-		     u64 ackno, u8 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 dccpap_buf.
-	 *
-	 * We may well decide to do buffer compression, etc, but for now lets
-	 * just drop.
-	 *
-	 * From Appendix A:
-	 *
-	 *	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.
-	 */
-
-	/* See if this is the first ackno being inserted */
-	if (ap->dccpap_buf_vector_len == 0) {
-		ap->dccpap_buf[ap->dccpap_buf_head] = state;
-		ap->dccpap_buf_vector_len = 1;
-	} else if (after48(ackno, ap->dccpap_buf_ackno)) {
-		const u64 delta = dccp_delta_seqno(ap->dccpap_buf_ackno,
-						   ackno);
-
-		/*
-		 * Look if the state of this packet is the same as the
-		 * previous ackno and if so if we can bump the head len.
-		 */
-		if (delta == 1 &&
-		    dccp_ackpkts_state(ap, ap->dccpap_buf_head) == state &&
-		    (dccp_ackpkts_len(ap, ap->dccpap_buf_head) <
-		     DCCP_ACKPKTS_LEN_MASK))
-			ap->dccpap_buf[ap->dccpap_buf_head]++;
-		else if (dccp_ackpkts_set_buf_head_state(ap, delta, state))
-			return -ENOBUFS;
-	} else {
-		/*
-		 * A.1.2.  Old Packets
-		 *
-		 *	When a packet with Sequence Number S arrives, and
-		 *	S <= buf_ackno, 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, ap->dccpap_buf_ackno);
-		unsigned int index = ap->dccpap_buf_head;
-
-		while (1) {
-			const u8 len = dccp_ackpkts_len(ap, index);
-			const u8 state = dccp_ackpkts_state(ap, index);
-			/*
-			 * valid packets not yet in dccpap_buf have a reserved
-			 * entry, with a len equal to 0.
-			 */
-			if (state == DCCP_ACKPKTS_STATE_NOT_RECEIVED &&
-			    len == 0 && delta == 0) { /* Found our
-							 reserved seat! */
-				dccp_pr_debug("Found %llu reserved seat!\n",
-					      (unsigned long long) ackno);
-				ap->dccpap_buf[index] = state;
-				goto out;
-			}
-			/* len == 0 means one packet */
-			if (delta < len + 1)
-				goto out_duplicate;
-
-			delta -= len + 1;
-			if (++index == ap->dccpap_buf_len)
-				index = 0;
-		}
-	}
-
-	ap->dccpap_buf_ackno = ackno;
-	dccp_timestamp(sk, &ap->dccpap_time);
-out:
-	dccp_pr_debug("");
-	dccp_ackpkts_print(ap);
-	return 0;
-
-out_duplicate:
-	/* Duplicate packet */
-	dccp_pr_debug("Received a dup or already considered lost "
-		      "packet: %llu\n", (unsigned long long) ackno);
-	return -EILSEQ;
-}
-
-#ifdef CONFIG_IP_DCCP_DEBUG
-void dccp_ackvector_print(const u64 ackno, const unsigned char *vector,
-			  int len)
-{
-	if (!dccp_debug)
-		return;
-
-	printk("ACK vector len=%d, ackno=%llu |", len,
-	       (unsigned long long) ackno);
-
-	while (len--) {
-		const u8 state = (*vector & DCCP_ACKPKTS_STATE_MASK) >> 6;
-		const u8 rl = (*vector & DCCP_ACKPKTS_LEN_MASK);
-
-		printk("%d,%d|", state, rl);
-		++vector;
-	}
-
-	printk("\n");
-}
-
-void dccp_ackpkts_print(const struct dccp_ackpkts *ap)
-{
-	dccp_ackvector_print(ap->dccpap_buf_ackno,
-			     ap->dccpap_buf + ap->dccpap_buf_head,
-			     ap->dccpap_buf_vector_len);
-}
-#endif
-
-static void dccp_ackpkts_trow_away_ack_record(struct dccp_ackpkts *ap)
-{
-	/*
-	 * As we're keeping track of the ack vector size
-	 * (dccpap_buf_vector_len) and the sent ack vector size
-	 * (dccpap_ack_vector_len) we don't need dccpap_buf_tail at all, but
-	 * keep this code here as in the future we'll implement a vector of
-	 * ack records, as suggested in draft-ietf-dccp-spec-11.txt
-	 * Appendix A. -acme
-	 */
-#if 0
-	ap->dccpap_buf_tail = ap->dccpap_ack_ptr + 1;
-	if (ap->dccpap_buf_tail >= ap->dccpap_buf_len)
-		ap->dccpap_buf_tail -= ap->dccpap_buf_len;
-#endif
-	ap->dccpap_buf_vector_len -= ap->dccpap_ack_vector_len;
-}
-
-void dccp_ackpkts_check_rcv_ackno(struct dccp_ackpkts *ap, struct sock *sk,
-				 u64 ackno)
-{
-	/* Check if we actually sent an ACK vector */
-	if (ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1)
-		return;
-
-	if (ackno == ap->dccpap_ack_seqno) {
-#ifdef CONFIG_IP_DCCP_DEBUG
-		struct dccp_sock *dp = dccp_sk(sk);
-		const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ?
-					"CLIENT rx ack: " : "server rx ack: ";
-#endif
-		dccp_pr_debug("%sACK packet 0, len=%d, ack_seqno=%llu, "
-			      "ack_ackno=%llu, ACKED!\n",
-			      debug_prefix, 1,
-			      (unsigned long long) ap->dccpap_ack_seqno,
-			      (unsigned long long) ap->dccpap_ack_ackno);
-		dccp_ackpkts_trow_away_ack_record(ap);
-		ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1;
-	}
-}
-
-static void dccp_ackpkts_check_rcv_ackvector(struct dccp_ackpkts *ap,
-					     struct sock *sk, u64 ackno,
-					     const unsigned char len,
-					     const unsigned char *vector)
-{
-	unsigned char i;
-
-	/* Check if we actually sent an ACK vector */
-	if (ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1)
-		return;
-	/*
-	 * We're in the receiver half connection, so if the received an ACK
-	 * vector ackno (e.g. 50) before dccpap_ack_seqno (e.g. 52), we're
-	 * not interested.
-	 *
-	 * Extra explanation with example:
-	 * 
-	 * if we received an ACK vector with ackno 50, it can only be acking
-	 * 50, 49, 48, etc, not 52 (the seqno for the ACK vector we sent).
-	 */
-	/* dccp_pr_debug("is %llu < %llu? ", ackno, ap->dccpap_ack_seqno); */
-	if (before48(ackno, ap->dccpap_ack_seqno)) {
-		/* dccp_pr_debug_cat("yes\n"); */
-		return;
-	}
-	/* dccp_pr_debug_cat("no\n"); */
-
-	i = len;
-	while (i--) {
-		const u8 rl = (*vector & DCCP_ACKPKTS_LEN_MASK);
-		u64 ackno_end_rl;
-
-		dccp_set_seqno(&ackno_end_rl, ackno - rl);
-
-		/*
-		 * dccp_pr_debug("is %llu <= %llu <= %llu? ", ackno_end_rl,
-		 * ap->dccpap_ack_seqno, ackno);
-		 */
-		if (between48(ap->dccpap_ack_seqno, ackno_end_rl, ackno)) {
-			const u8 state = (*vector &
-					  DCCP_ACKPKTS_STATE_MASK) >> 6;
-			/* dccp_pr_debug_cat("yes\n"); */
-
-			if (state != DCCP_ACKPKTS_STATE_NOT_RECEIVED) {
-#ifdef CONFIG_IP_DCCP_DEBUG
-				struct dccp_sock *dp = dccp_sk(sk);
-				const char *debug_prefix =
-					dp->dccps_role == DCCP_ROLE_CLIENT ?
-					"CLIENT rx ack: " : "server rx ack: ";
-#endif
-				dccp_pr_debug("%sACK vector 0, len=%d, "
-					      "ack_seqno=%llu, ack_ackno=%llu, "
-					      "ACKED!\n",
-					      debug_prefix, len,
-					      (unsigned long long)
-					      ap->dccpap_ack_seqno,
-					      (unsigned long long)
-					      ap->dccpap_ack_ackno);
-				dccp_ackpkts_trow_away_ack_record(ap);
-			}
-			/*
-			 * If dccpap_ack_seqno was not received, no problem
-			 * we'll send another ACK vector.
-			 */
-			ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1;
-			break;
-		}
-		/* dccp_pr_debug_cat("no\n"); */
-
-		dccp_set_seqno(&ackno, ackno_end_rl - 1);
-		++vector;
-	}
-}

net/dccp/output.c

@@ -16,6 +16,7 @@
 
 #include <net/sock.h>
 
+#include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
 
@@ -225,7 +226,6 @@ int dccp_write_xmit(struct sock *sk, struct sk_buff *skb, long *timeo)
 		err = dccp_wait_for_ccid(sk, skb, timeo);
 
 	if (err == 0) {
-		const struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts;
 		struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
 		const int len = skb->len;
 
@@ -236,15 +236,7 @@ int dccp_write_xmit(struct sock *sk, struct sk_buff *skb, long *timeo)
 						  inet_csk(sk)->icsk_rto,
 						  DCCP_RTO_MAX);
 			dcb->dccpd_type = DCCP_PKT_DATAACK;
-			/*
-			 * FIXME: we really should have a
-			 * dccps_ack_pending or use icsk.
-			 */
-		} else if (inet_csk_ack_scheduled(sk) ||
-			   dp->dccps_timestamp_echo != 0 ||
-			   (dp->dccps_options.dccpo_send_ack_vector &&
-			    ap->dccpap_buf_ackno != DCCP_MAX_SEQNO + 1 &&
-			    ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1))
+		} else if (dccp_ack_pending(sk))
 			dcb->dccpd_type = DCCP_PKT_DATAACK;
 		else
 			dcb->dccpd_type = DCCP_PKT_DATA;