[net-next-2.6.git] / net / dccp / ackvec.c

/*
 *  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/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_record *avr =
			kmem_cache_alloc(dccp_ackvec_record_slab, GFP_ATOMIC);

	if (avr != NULL)
		INIT_LIST_HEAD(&avr->avr_node);

	return avr;
}

static void dccp_ackvec_record_delete(struct dccp_ackvec_record *avr)
{
	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);
}

static void dccp_ackvec_insert_avr(struct dccp_ackvec *av,
				   struct dccp_ackvec_record *avr)
{
	/*
	 * 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));
	}

	list_add(&avr->avr_node, &av->av_records);
}

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;
	/* Figure out how many options do we need to represent the ackvec */
	const u8 nr_opts = DIV_ROUND_UP(av->av_vec_len, DCCP_SINGLE_OPT_MAXLEN);
	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(skb, elapsed_time))
		return -1;

	avr = dccp_ackvec_record_new();
	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_SINGLE_OPT_MAXLEN)
			copylen = DCCP_SINGLE_OPT_MAXLEN;

		*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;
	}

	/*
	 *	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.
	 */
	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);

	dccp_pr_debug("%s ACK Vector 0, len=%d, ack_seqno=%llu, "
		      "ack_ackno=%llu\n",
		      dccp_role(sk), avr->avr_sent_len,
		      (unsigned long long)avr->avr_ack_seqno,
		      (unsigned long long)avr->avr_ack_ackno);
	return 0;
}

struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority)
{
	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);
	}

	return av;
}

void dccp_ackvec_free(struct dccp_ackvec *av)
{
	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);
}

static inline u8 dccp_ackvec_state(const struct dccp_ackvec *av,
				   const u32 index)
{
	return av->av_buf[index] & DCCP_ACKVEC_STATE_MASK;
}

static inline u8 dccp_ackvec_len(const struct dccp_ackvec *av,
				 const u32 index)
{
	return av->av_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)
{
	long gap;
	long new_head;

	if (av->av_vec_len + packets > DCCP_MAX_ACKVEC_LEN)
		return -ENOBUFS;

	gap	 = packets - 1;
	new_head = av->av_buf_head - packets;

	if (new_head < 0) {
		if (gap > 0) {
			memset(av->av_buf, DCCP_ACKVEC_STATE_NOT_RECEIVED,
			       gap + new_head + 1);
			gap = -new_head;
		}
		new_head += DCCP_MAX_ACKVEC_LEN;
	}

	av->av_buf_head = new_head;

	if (gap > 0)
		memset(av->av_buf + av->av_buf_head + 1,
		       DCCP_ACKVEC_STATE_NOT_RECEIVED, gap);

	av->av_buf[av->av_buf_head] = state;
	av->av_vec_len += packets;
	return 0;
}

/*
 * Implements the RFC 4340, 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 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.
	 */

	/* 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);

		/*
		 * 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->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;

		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_ackno = ackno;
	av->av_time = ktime_get_real();
out:
	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;
}

static void dccp_ackvec_throw_record(struct dccp_ackvec *av,
				     struct dccp_ackvec_record *avr)
{
	struct dccp_ackvec_record *next;

	/* 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;

	/* 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);
	}
}

void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, struct sock *sk,
				 const u64 ackno)
{
	struct dccp_ackvec_record *avr;

	/*
	 * 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 */
	}
}

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;

	/* Check if we actually sent an ACK vector */
	if (list_empty(&av->av_records))
		return;

	i = len;
	/*
	 * XXX
	 * I think it might be more efficient to work backwards. See comment on
	 * rcv_ackno. -sorbo.
	 */
	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;

		dccp_set_seqno(&ackno_end_rl, *ackno - rl);

		/*
		 * If our AVR sequence number is greater than the ack, go
		 * forward in the AVR list until it is not so.
		 */
		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.
			 */
		}

		dccp_set_seqno(ackno, ackno_end_rl - 1);
		++vector;
	}
}

int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb,
		      u64 *ackno, const u8 opt, const u8 *value, const u8 len)
{
	if (len > DCCP_SINGLE_OPT_MAXLEN)
		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,
					ackno, len, value);
	return 0;
}

int __init dccp_ackvec_init(void)
{
	dccp_ackvec_slab = kmem_cache_create("dccp_ackvec",
					     sizeof(struct dccp_ackvec), 0,
					     SLAB_HWCACHE_ALIGN, NULL);
	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);
	if (dccp_ackvec_record_slab == NULL)
		goto out_destroy_slab;

	return 0;

out_destroy_slab:
	kmem_cache_destroy(dccp_ackvec_slab);
	dccp_ackvec_slab = NULL;
out_err:
	DCCP_CRIT("Unable to create Ack Vector slab cache");
	return -ENOBUFS;
}

void dccp_ackvec_exit(void)
{
	if (dccp_ackvec_slab != NULL) {
		kmem_cache_destroy(dccp_ackvec_slab);
		dccp_ackvec_slab = NULL;
	}
	if (dccp_ackvec_record_slab != NULL) {
		kmem_cache_destroy(dccp_ackvec_record_slab);
		dccp_ackvec_record_slab = NULL;
	}
}
Commit	Line	Data
ae31c339 ACM	1	/*
	2	* net/dccp/ackvec.c
	3	*
	4	* An implementation of the DCCP protocol
	5	* Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
	6	*
	7	* This program is free software; you can redistribute it and/or modify it
	8	* under the terms of the GNU General Public License as published by the
	9	* Free Software Foundation; version 2 of the License;
	10	*/
	11
	12	#include "ackvec.h"
	13	#include "dccp.h"
	14
9b07ef5d ACM	15	#include <linux/init.h>
	16	#include <linux/errno.h>
	17	#include <linux/kernel.h>
ae31c339	18	#include <linux/skbuff.h>
9b07ef5d	19	#include <linux/slab.h>
ae31c339 ACM	20
	21	#include <net/sock.h>
	22
e18b890b CL	23	static struct kmem_cache *dccp_ackvec_slab;
e18b890b CL	24	static struct kmem_cache *dccp_ackvec_record_slab;
02bcf28c AB	25
	26	static struct dccp_ackvec_record *dccp_ackvec_record_new(void)
	27	{
	28	struct dccp_ackvec_record *avr =
	29	kmem_cache_alloc(dccp_ackvec_record_slab, GFP_ATOMIC);
	30
	31	if (avr != NULL)
a47c5104	32	INIT_LIST_HEAD(&avr->avr_node);
02bcf28c AB	33
	34	return avr;
	35	}
	36
	37	static void dccp_ackvec_record_delete(struct dccp_ackvec_record *avr)
	38	{
	39	if (unlikely(avr == NULL))
	40	return;
	41	/* Check if deleting a linked record */
a47c5104	42	WARN_ON(!list_empty(&avr->avr_node));
02bcf28c AB	43	kmem_cache_free(dccp_ackvec_record_slab, avr);
	44	}
	45
	46	static void dccp_ackvec_insert_avr(struct dccp_ackvec *av,
	47	struct dccp_ackvec_record *avr)
	48	{
	49	/*
	50	* AVRs are sorted by seqno. Since we are sending them in order, we
	51	* just add the AVR at the head of the list.
	52	* -sorbo.
	53	*/
a47c5104	54	if (!list_empty(&av->av_records)) {
02bcf28c	55	const struct dccp_ackvec_record *head =
a47c5104	56	list_entry(av->av_records.next,
02bcf28c	57	struct dccp_ackvec_record,
a47c5104 GR	58	avr_node);
a47c5104 GR	59	BUG_ON(before48(avr->avr_ack_seqno, head->avr_ack_seqno));
02bcf28c AB	60	}
02bcf28c AB	61
a47c5104	62	list_add(&avr->avr_node, &av->av_records);
02bcf28c	63	}
9b07ef5d	64
ae31c339 ACM	65	int dccp_insert_option_ackvec(struct sock sk, struct sk_buff skb)
	66	{
	67	struct dccp_sock *dp = dccp_sk(sk);
	68	struct dccp_ackvec *av = dp->dccps_hc_rx_ackvec;
522f1d09	69	/* Figure out how many options do we need to represent the ackvec */
b20a9c24	70	const u8 nr_opts = DIV_ROUND_UP(av->av_vec_len, DCCP_SINGLE_OPT_MAXLEN);
a47c5104	71	u16 len = av->av_vec_len + 2 * nr_opts, i;
ae31c339	72	u32 elapsed_time;
522f1d09 AB	73	const unsigned char tail, from;
522f1d09 AB	74	unsigned char *to;
02bcf28c	75	struct dccp_ackvec_record *avr;
b8bda9d7	76	suseconds_t delta;
02bcf28c AB	77
	78	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN)
	79	return -1;
	80
a47c5104	81	delta = ktime_us_delta(ktime_get_real(), av->av_time);
b8bda9d7	82	elapsed_time = delta / 10;
ae31c339	83
2d0817d1	84	if (elapsed_time != 0 &&
a7d13fbf	85	dccp_insert_option_elapsed_time(skb, elapsed_time))
2d0817d1 ACM	86	return -1;
	87
	88	avr = dccp_ackvec_record_new();
	89	if (avr == NULL)
	90	return -1;
ae31c339	91
ae31c339 ACM	92	DCCP_SKB_CB(skb)->dccpd_opt_len += len;
ae31c339 ACM	93
522f1d09	94	to = skb_push(skb, len);
a47c5104 GR	95	len = av->av_vec_len;
	96	from = av->av_buf + av->av_buf_head;
	97	tail = av->av_buf + DCCP_MAX_ACKVEC_LEN;
522f1d09 AB	98
	99	for (i = 0; i < nr_opts; ++i) {
	100	int copylen = len;
	101
b20a9c24 GR	102	if (len > DCCP_SINGLE_OPT_MAXLEN)
b20a9c24 GR	103	copylen = DCCP_SINGLE_OPT_MAXLEN;
522f1d09 AB	104
	105	*to++ = DCCPO_ACK_VECTOR_0;
	106	*to++ = copylen + 2;
	107
	108	/* Check if buf_head wraps */
	109	if (from + copylen > tail) {
	110	const u16 tailsize = tail - from;
ae31c339	111
522f1d09 AB	112	memcpy(to, from, tailsize);
	113	to += tailsize;
	114	len -= tailsize;
	115	copylen -= tailsize;
a47c5104	116	from = av->av_buf;
522f1d09	117	}
ae31c339	118
522f1d09 AB	119	memcpy(to, from, copylen);
	120	from += copylen;
	121	to += copylen;
	122	len -= copylen;
ae31c339 ACM	123	}
ae31c339 ACM	124
ae31c339	125	/*
0e64e94e	126	* From RFC 4340, A.2:
ae31c339 ACM	127	*
	128	* For each acknowledgement it sends, the HC-Receiver will add an
	129	* acknowledgement record. ack_seqno will equal the HC-Receiver
	130	* sequence number it used for the ack packet; ack_ptr will equal
	131	* buf_head; ack_ackno will equal buf_ackno; and ack_nonce will
	132	* equal buf_nonce.
ae31c339	133	*/
a47c5104 GR	134	avr->avr_ack_seqno = DCCP_SKB_CB(skb)->dccpd_seq;
	135	avr->avr_ack_ptr = av->av_buf_head;
	136	avr->avr_ack_ackno = av->av_buf_ackno;
	137	avr->avr_ack_nonce = av->av_buf_nonce;
	138	avr->avr_sent_len = av->av_vec_len;
02bcf28c AB	139
02bcf28c AB	140	dccp_ackvec_insert_avr(av, avr);
ae31c339	141
09dbc389	142	dccp_pr_debug("%s ACK Vector 0, len=%d, ack_seqno=%llu, "
ae31c339	143	"ack_ackno=%llu\n",
a47c5104 GR	144	dccp_role(sk), avr->avr_sent_len,
	145	(unsigned long long)avr->avr_ack_seqno,
	146	(unsigned long long)avr->avr_ack_ackno);
02bcf28c	147	return 0;
ae31c339 ACM	148	}
ae31c339 ACM	149
7400d781	150	struct dccp_ackvec *dccp_ackvec_alloc(const gfp_t priority)
ae31c339	151	{
9b07ef5d	152	struct dccp_ackvec *av = kmem_cache_alloc(dccp_ackvec_slab, priority);
e4dfd449	153
ae31c339	154	if (av != NULL) {
a47c5104 GR	155	av->av_buf_head = DCCP_MAX_ACKVEC_LEN - 1;
	156	av->av_buf_ackno = UINT48_MAX + 1;
	157	av->av_buf_nonce = 0;
	158	av->av_time = ktime_set(0, 0);
	159	av->av_vec_len = 0;
	160	INIT_LIST_HEAD(&av->av_records);
ae31c339 ACM	161	}
	162
	163	return av;
	164	}
	165
	166	void dccp_ackvec_free(struct dccp_ackvec *av)
	167	{
02bcf28c AB	168	if (unlikely(av == NULL))
02bcf28c AB	169	return;
d5e9b2c7	170
a47c5104	171	if (!list_empty(&av->av_records)) {
d5e9b2c7 ACM	172	struct dccp_ackvec_record avr, next;
d5e9b2c7 ACM	173
a47c5104 GR	174	list_for_each_entry_safe(avr, next, &av->av_records, avr_node) {
a47c5104 GR	175	list_del_init(&avr->avr_node);
d5e9b2c7 ACM	176	dccp_ackvec_record_delete(avr);
	177	}
	178	}
	179
9b07ef5d	180	kmem_cache_free(dccp_ackvec_slab, av);
ae31c339 ACM	181	}
	182
	183	static inline u8 dccp_ackvec_state(const struct dccp_ackvec *av,
bdf13d20	184	const u32 index)
ae31c339	185	{
a47c5104	186	return av->av_buf[index] & DCCP_ACKVEC_STATE_MASK;
ae31c339 ACM	187	}
	188
	189	static inline u8 dccp_ackvec_len(const struct dccp_ackvec *av,
bdf13d20	190	const u32 index)
ae31c339	191	{
a47c5104	192	return av->av_buf[index] & DCCP_ACKVEC_LEN_MASK;
ae31c339 ACM	193	}
	194
	195	/*
	196	* If several packets are missing, the HC-Receiver may prefer to enter multiple
	197	* bytes with run length 0, rather than a single byte with a larger run length;
	198	* this simplifies table updates if one of the missing packets arrives.
	199	*/
	200	static inline int dccp_ackvec_set_buf_head_state(struct dccp_ackvec *av,
	201	const unsigned int packets,
e4dfd449	202	const unsigned char state)
ae31c339	203	{
8e64159d	204	long gap;
a8fc3d8d	205	long new_head;
ae31c339	206
a47c5104	207	if (av->av_vec_len + packets > DCCP_MAX_ACKVEC_LEN)
ae31c339 ACM	208	return -ENOBUFS;
	209
	210	gap = packets - 1;
a47c5104	211	new_head = av->av_buf_head - packets;
ae31c339 ACM	212
	213	if (new_head < 0) {
	214	if (gap > 0) {
a47c5104	215	memset(av->av_buf, DCCP_ACKVEC_STATE_NOT_RECEIVED,
ae31c339 ACM	216	gap + new_head + 1);
	217	gap = -new_head;
	218	}
7400d781	219	new_head += DCCP_MAX_ACKVEC_LEN;
8109b02b	220	}
ae31c339	221
a47c5104	222	av->av_buf_head = new_head;
ae31c339 ACM	223
ae31c339 ACM	224	if (gap > 0)
a47c5104	225	memset(av->av_buf + av->av_buf_head + 1,
ae31c339 ACM	226	DCCP_ACKVEC_STATE_NOT_RECEIVED, gap);
ae31c339 ACM	227
a47c5104 GR	228	av->av_buf[av->av_buf_head] = state;
a47c5104 GR	229	av->av_vec_len += packets;
ae31c339 ACM	230	return 0;
	231	}
	232
	233	/*
0e64e94e	234	* Implements the RFC 4340, Appendix A
ae31c339 ACM	235	*/
	236	int dccp_ackvec_add(struct dccp_ackvec av, const struct sock sk,
	237	const u64 ackno, const u8 state)
	238	{
	239	/*
	240	* Check at the right places if the buffer is full, if it is, tell the
	241	* caller to start dropping packets till the HC-Sender acks our ACK
a47c5104	242	* vectors, when we will free up space in av_buf.
ae31c339 ACM	243	*
	244	* We may well decide to do buffer compression, etc, but for now lets
	245	* just drop.
	246	*
0e64e94e	247	* From Appendix A.1.1 (`New Packets'):
ae31c339 ACM	248	*
	249	* Of course, the circular buffer may overflow, either when the
	250	* HC-Sender is sending data at a very high rate, when the
	251	* HC-Receiver's acknowledgements are not reaching the HC-Sender,
	252	* or when the HC-Sender is forgetting to acknowledge those acks
	253	* (so the HC-Receiver is unable to clean up old state). In this
	254	* case, the HC-Receiver should either compress the buffer (by
	255	* increasing run lengths when possible), transfer its state to
	256	* a larger buffer, or, as a last resort, drop all received
	257	* packets, without processing them whatsoever, until its buffer
	258	* shrinks again.
	259	*/
	260
	261	/* See if this is the first ackno being inserted */
a47c5104 GR	262	if (av->av_vec_len == 0) {
	263	av->av_buf[av->av_buf_head] = state;
	264	av->av_vec_len = 1;
	265	} else if (after48(ackno, av->av_buf_ackno)) {
	266	const u64 delta = dccp_delta_seqno(av->av_buf_ackno, ackno);
ae31c339 ACM	267
	268	/*
	269	* Look if the state of this packet is the same as the
	270	* previous ackno and if so if we can bump the head len.
	271	*/
	272	if (delta == 1 &&
a47c5104 GR	273	dccp_ackvec_state(av, av->av_buf_head) == state &&
	274	dccp_ackvec_len(av, av->av_buf_head) < DCCP_ACKVEC_LEN_MASK)
	275	av->av_buf[av->av_buf_head]++;
ae31c339 ACM	276	else if (dccp_ackvec_set_buf_head_state(av, delta, state))
	277	return -ENOBUFS;
	278	} else {
	279	/*
	280	* A.1.2. Old Packets
	281	*
0e64e94e GR	282	* When a packet with Sequence Number S <= buf_ackno
	283	* arrives, the HC-Receiver will scan the table for
	284	* the byte corresponding to S. (Indexing structures
ae31c339 ACM	285	* could reduce the complexity of this scan.)
ae31c339 ACM	286	*/
a47c5104 GR	287	u64 delta = dccp_delta_seqno(ackno, av->av_buf_ackno);
a47c5104 GR	288	u32 index = av->av_buf_head;
ae31c339 ACM	289
	290	while (1) {
	291	const u8 len = dccp_ackvec_len(av, index);
1e2f0e5e	292	const u8 av_state = dccp_ackvec_state(av, index);
ae31c339	293	/*
a47c5104	294	* valid packets not yet in av_buf have a reserved
ae31c339 ACM	295	* entry, with a len equal to 0.
ae31c339 ACM	296	*/
1e2f0e5e	297	if (av_state == DCCP_ACKVEC_STATE_NOT_RECEIVED &&
ae31c339 ACM	298	len == 0 && delta == 0) { /* Found our
	299	reserved seat! */
	300	dccp_pr_debug("Found %llu reserved seat!\n",
	301	(unsigned long long)ackno);
a47c5104	302	av->av_buf[index] = state;
ae31c339 ACM	303	goto out;
	304	}
	305	/* len == 0 means one packet */
	306	if (delta < len + 1)
	307	goto out_duplicate;
	308
	309	delta -= len + 1;
7400d781	310	if (++index == DCCP_MAX_ACKVEC_LEN)
ae31c339 ACM	311	index = 0;
	312	}
	313	}
	314
a47c5104 GR	315	av->av_buf_ackno = ackno;
a47c5104 GR	316	av->av_time = ktime_get_real();
ae31c339	317	out:
ae31c339 ACM	318	return 0;
	319
	320	out_duplicate:
	321	/* Duplicate packet */
	322	dccp_pr_debug("Received a dup or already considered lost "
	323	"packet: %llu\n", (unsigned long long)ackno);
	324	return -EILSEQ;
	325	}
	326
02bcf28c AB	327	static void dccp_ackvec_throw_record(struct dccp_ackvec *av,
02bcf28c AB	328	struct dccp_ackvec_record *avr)
ae31c339	329	{
02bcf28c AB	330	struct dccp_ackvec_record *next;
02bcf28c AB	331
23d06e3b	332	/* sort out vector length */
a47c5104 GR	333	if (av->av_buf_head <= avr->avr_ack_ptr)
a47c5104 GR	334	av->av_vec_len = avr->avr_ack_ptr - av->av_buf_head;
23d06e3b	335	else
a47c5104 GR	336	av->av_vec_len = DCCP_MAX_ACKVEC_LEN - 1 -
a47c5104 GR	337	av->av_buf_head + avr->avr_ack_ptr;
02bcf28c AB	338
02bcf28c AB	339	/* free records */
a47c5104 GR	340	list_for_each_entry_safe_from(avr, next, &av->av_records, avr_node) {
a47c5104 GR	341	list_del_init(&avr->avr_node);
02bcf28c AB	342	dccp_ackvec_record_delete(avr);
02bcf28c AB	343	}
ae31c339 ACM	344	}
	345
	346	void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec av, struct sock sk,
	347	const u64 ackno)
	348	{
02bcf28c	349	struct dccp_ackvec_record *avr;
ae31c339	350
02bcf28c AB	351	/*
	352	* If we traverse backwards, it should be faster when we have large
	353	* windows. We will be receiving ACKs for stuff we sent a while back
	354	* -sorbo.
	355	*/
a47c5104 GR	356	list_for_each_entry_reverse(avr, &av->av_records, avr_node) {
a47c5104 GR	357	if (ackno == avr->avr_ack_seqno) {
09dbc389	358	dccp_pr_debug("%s ACK packet 0, len=%d, ack_seqno=%llu, "
02bcf28c	359	"ack_ackno=%llu, ACKED!\n",
09dbc389	360	dccp_role(sk), 1,
a47c5104 GR	361	(unsigned long long)avr->avr_ack_seqno,
a47c5104 GR	362	(unsigned long long)avr->avr_ack_ackno);
02bcf28c AB	363	dccp_ackvec_throw_record(av, avr);
02bcf28c AB	364	break;
a47c5104	365	} else if (avr->avr_ack_seqno > ackno)
d23ca15a	366	break; /* old news */
ae31c339 ACM	367	}
	368	}
	369
	370	static void dccp_ackvec_check_rcv_ackvector(struct dccp_ackvec *av,
bdf13d20	371	struct sock sk, u64 ackno,
ae31c339 ACM	372	const unsigned char len,
	373	const unsigned char *vector)
	374	{
	375	unsigned char i;
02bcf28c	376	struct dccp_ackvec_record *avr;
ae31c339 ACM	377
ae31c339 ACM	378	/* Check if we actually sent an ACK vector */
a47c5104	379	if (list_empty(&av->av_records))
ae31c339	380	return;
ae31c339 ACM	381
ae31c339 ACM	382	i = len;
02bcf28c AB	383	/*
	384	* XXX
	385	* I think it might be more efficient to work backwards. See comment on
	386	* rcv_ackno. -sorbo.
	387	*/
a47c5104	388	avr = list_entry(av->av_records.next, struct dccp_ackvec_record, avr_node);
ae31c339 ACM	389	while (i--) {
	390	const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
	391	u64 ackno_end_rl;
	392
bdf13d20	393	dccp_set_seqno(&ackno_end_rl, *ackno - rl);
ae31c339 ACM	394
ae31c339 ACM	395	/*
02bcf28c AB	396	* If our AVR sequence number is greater than the ack, go
02bcf28c AB	397	* forward in the AVR list until it is not so.
ae31c339	398	*/
a47c5104 GR	399	list_for_each_entry_from(avr, &av->av_records, avr_node) {
a47c5104 GR	400	if (!after48(avr->avr_ack_seqno, *ackno))
02bcf28c AB	401	goto found;
02bcf28c AB	402	}
a47c5104	403	/* End of the av_records list, not found, exit */
02bcf28c AB	404	break;
02bcf28c AB	405	found:
a47c5104	406	if (between48(avr->avr_ack_seqno, ackno_end_rl, *ackno)) {
8e27e465	407	const u8 state = *vector & DCCP_ACKVEC_STATE_MASK;
ae31c339	408	if (state != DCCP_ACKVEC_STATE_NOT_RECEIVED) {
09dbc389	409	dccp_pr_debug("%s ACK vector 0, len=%d, "
ae31c339 ACM	410	"ack_seqno=%llu, ack_ackno=%llu, "
ae31c339 ACM	411	"ACKED!\n",
09dbc389	412	dccp_role(sk), len,
ae31c339	413	(unsigned long long)
a47c5104	414	avr->avr_ack_seqno,
ae31c339	415	(unsigned long long)
a47c5104	416	avr->avr_ack_ackno);
02bcf28c	417	dccp_ackvec_throw_record(av, avr);
afec35e3	418	break;
ae31c339 ACM	419	}
ae31c339 ACM	420	/*
02bcf28c AB	421	* If it wasn't received, continue scanning... we might
02bcf28c AB	422	* find another one.
ae31c339	423	*/
ae31c339	424	}
ae31c339	425
bdf13d20	426	dccp_set_seqno(ackno, ackno_end_rl - 1);
ae31c339 ACM	427	++vector;
	428	}
	429	}
	430
	431	int dccp_ackvec_parse(struct sock sk, const struct sk_buff skb,
bdf13d20	432	u64 ackno, const u8 opt, const u8 value, const u8 len)
ae31c339	433	{
b20a9c24	434	if (len > DCCP_SINGLE_OPT_MAXLEN)
ae31c339 ACM	435	return -1;
	436
	437	/* dccp_ackvector_print(DCCP_SKB_CB(skb)->dccpd_ack_seq, value, len); */
	438	dccp_ackvec_check_rcv_ackvector(dccp_sk(sk)->dccps_hc_rx_ackvec, sk,
bdf13d20	439	ackno, len, value);
ae31c339 ACM	440	return 0;
ae31c339 ACM	441	}
9b07ef5d	442
9b07ef5d ACM	443	int __init dccp_ackvec_init(void)
	444	{
	445	dccp_ackvec_slab = kmem_cache_create("dccp_ackvec",
	446	sizeof(struct dccp_ackvec), 0,
20c2df83	447	SLAB_HWCACHE_ALIGN, NULL);
02bcf28c AB	448	if (dccp_ackvec_slab == NULL)
	449	goto out_err;
	450
	451	dccp_ackvec_record_slab =
	452	kmem_cache_create("dccp_ackvec_record",
	453	sizeof(struct dccp_ackvec_record),
20c2df83	454	0, SLAB_HWCACHE_ALIGN, NULL);
02bcf28c AB	455	if (dccp_ackvec_record_slab == NULL)
02bcf28c AB	456	goto out_destroy_slab;
9b07ef5d ACM	457
9b07ef5d ACM	458	return 0;
02bcf28c AB	459
	460	out_destroy_slab:
	461	kmem_cache_destroy(dccp_ackvec_slab);
	462	dccp_ackvec_slab = NULL;
	463	out_err:
59348b19	464	DCCP_CRIT("Unable to create Ack Vector slab cache");
02bcf28c	465	return -ENOBUFS;
9b07ef5d ACM	466	}
	467
	468	void dccp_ackvec_exit(void)
	469	{
	470	if (dccp_ackvec_slab != NULL) {
	471	kmem_cache_destroy(dccp_ackvec_slab);
	472	dccp_ackvec_slab = NULL;
	473	}
02bcf28c AB	474	if (dccp_ackvec_record_slab != NULL) {
	475	kmem_cache_destroy(dccp_ackvec_record_slab);
	476	dccp_ackvec_record_slab = NULL;
	477	}
9b07ef5d	478	}