[mirror_ubuntu-artful-kernel.git] / net / netfilter / nf_conntrack_seqadj.c

#include <linux/types.h>
#include <linux/netfilter.h>
#include <net/tcp.h>

#include <net/netfilter/nf_conntrack.h>
#include <net/netfilter/nf_conntrack_extend.h>
#include <net/netfilter/nf_conntrack_seqadj.h>

int nf_ct_seqadj_init(struct nf_conn *ct, enum ip_conntrack_info ctinfo,
		      s32 off)
{
	enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
	struct nf_conn_seqadj *seqadj;
	struct nf_ct_seqadj *this_way;

	if (off == 0)
		return 0;

	set_bit(IPS_SEQ_ADJUST_BIT, &ct->status);

	seqadj = nfct_seqadj(ct);
	this_way = &seqadj->seq[dir];
	this_way->offset_before	 = off;
	this_way->offset_after	 = off;
	return 0;
}
EXPORT_SYMBOL_GPL(nf_ct_seqadj_init);

int nf_ct_seqadj_set(struct nf_conn *ct, enum ip_conntrack_info ctinfo,
		     __be32 seq, s32 off)
{
	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
	struct nf_ct_seqadj *this_way;

	if (off == 0)
		return 0;

	if (unlikely(!seqadj)) {
		WARN_ONCE(1, "Missing nfct_seqadj_ext_add() setup call\n");
		return 0;
	}

	set_bit(IPS_SEQ_ADJUST_BIT, &ct->status);

	spin_lock_bh(&ct->lock);
	this_way = &seqadj->seq[dir];
	if (this_way->offset_before == this_way->offset_after ||
	    before(this_way->correction_pos, ntohl(seq))) {
		this_way->correction_pos = ntohl(seq);
		this_way->offset_before	 = this_way->offset_after;
		this_way->offset_after	+= off;
	}
	spin_unlock_bh(&ct->lock);
	return 0;
}
EXPORT_SYMBOL_GPL(nf_ct_seqadj_set);

void nf_ct_tcp_seqadj_set(struct sk_buff *skb,
			  struct nf_conn *ct, enum ip_conntrack_info ctinfo,
			  s32 off)
{
	const struct tcphdr *th;

	if (nf_ct_protonum(ct) != IPPROTO_TCP)
		return;

	th = (struct tcphdr *)(skb_network_header(skb) + ip_hdrlen(skb));
	nf_ct_seqadj_set(ct, ctinfo, th->seq, off);
}
EXPORT_SYMBOL_GPL(nf_ct_tcp_seqadj_set);

/* Adjust one found SACK option including checksum correction */
static void nf_ct_sack_block_adjust(struct sk_buff *skb,
				    struct tcphdr *tcph,
				    unsigned int sackoff,
				    unsigned int sackend,
				    struct nf_ct_seqadj *seq)
{
	while (sackoff < sackend) {
		struct tcp_sack_block_wire *sack;
		__be32 new_start_seq, new_end_seq;

		sack = (void *)skb->data + sackoff;
		if (after(ntohl(sack->start_seq) - seq->offset_before,
			  seq->correction_pos))
			new_start_seq = htonl(ntohl(sack->start_seq) -
					seq->offset_after);
		else
			new_start_seq = htonl(ntohl(sack->start_seq) -
					seq->offset_before);

		if (after(ntohl(sack->end_seq) - seq->offset_before,
			  seq->correction_pos))
			new_end_seq = htonl(ntohl(sack->end_seq) -
				      seq->offset_after);
		else
			new_end_seq = htonl(ntohl(sack->end_seq) -
				      seq->offset_before);

		pr_debug("sack_adjust: start_seq: %u->%u, end_seq: %u->%u\n",
			 ntohl(sack->start_seq), ntohl(new_start_seq),
			 ntohl(sack->end_seq), ntohl(new_end_seq));

		inet_proto_csum_replace4(&tcph->check, skb,
					 sack->start_seq, new_start_seq, false);
		inet_proto_csum_replace4(&tcph->check, skb,
					 sack->end_seq, new_end_seq, false);
		sack->start_seq = new_start_seq;
		sack->end_seq = new_end_seq;
		sackoff += sizeof(*sack);
	}
}

/* TCP SACK sequence number adjustment */
static unsigned int nf_ct_sack_adjust(struct sk_buff *skb,
				      unsigned int protoff,
				      struct tcphdr *tcph,
				      struct nf_conn *ct,
				      enum ip_conntrack_info ctinfo)
{
	unsigned int dir, optoff, optend;
	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);

	optoff = protoff + sizeof(struct tcphdr);
	optend = protoff + tcph->doff * 4;

	if (!skb_make_writable(skb, optend))
		return 0;

	dir = CTINFO2DIR(ctinfo);

	while (optoff < optend) {
		/* Usually: option, length. */
		unsigned char *op = skb->data + optoff;

		switch (op[0]) {
		case TCPOPT_EOL:
			return 1;
		case TCPOPT_NOP:
			optoff++;
			continue;
		default:
			/* no partial options */
			if (optoff + 1 == optend ||
			    optoff + op[1] > optend ||
			    op[1] < 2)
				return 0;
			if (op[0] == TCPOPT_SACK &&
			    op[1] >= 2+TCPOLEN_SACK_PERBLOCK &&
			    ((op[1] - 2) % TCPOLEN_SACK_PERBLOCK) == 0)
				nf_ct_sack_block_adjust(skb, tcph, optoff + 2,
							optoff+op[1],
							&seqadj->seq[!dir]);
			optoff += op[1];
		}
	}
	return 1;
}

/* TCP sequence number adjustment.  Returns 1 on success, 0 on failure */
int nf_ct_seq_adjust(struct sk_buff *skb,
		     struct nf_conn *ct, enum ip_conntrack_info ctinfo,
		     unsigned int protoff)
{
	enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
	struct tcphdr *tcph;
	__be32 newseq, newack;
	s32 seqoff, ackoff;
	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	struct nf_ct_seqadj *this_way, *other_way;
	int res = 1;

	this_way  = &seqadj->seq[dir];
	other_way = &seqadj->seq[!dir];

	if (!skb_make_writable(skb, protoff + sizeof(*tcph)))
		return 0;

	tcph = (void *)skb->data + protoff;
	spin_lock_bh(&ct->lock);
	if (after(ntohl(tcph->seq), this_way->correction_pos))
		seqoff = this_way->offset_after;
	else
		seqoff = this_way->offset_before;

	newseq = htonl(ntohl(tcph->seq) + seqoff);
	inet_proto_csum_replace4(&tcph->check, skb, tcph->seq, newseq, false);
	pr_debug("Adjusting sequence number from %u->%u\n",
		 ntohl(tcph->seq), ntohl(newseq));
	tcph->seq = newseq;

	if (!tcph->ack)
		goto out;

	if (after(ntohl(tcph->ack_seq) - other_way->offset_before,
		  other_way->correction_pos))
		ackoff = other_way->offset_after;
	else
		ackoff = other_way->offset_before;

	newack = htonl(ntohl(tcph->ack_seq) - ackoff);
	inet_proto_csum_replace4(&tcph->check, skb, tcph->ack_seq, newack,
				 false);
	pr_debug("Adjusting ack number from %u->%u, ack from %u->%u\n",
		 ntohl(tcph->seq), ntohl(newseq), ntohl(tcph->ack_seq),
		 ntohl(newack));
	tcph->ack_seq = newack;

	res = nf_ct_sack_adjust(skb, protoff, tcph, ct, ctinfo);
out:
	spin_unlock_bh(&ct->lock);

	return res;
}
EXPORT_SYMBOL_GPL(nf_ct_seq_adjust);

s32 nf_ct_seq_offset(const struct nf_conn *ct,
		     enum ip_conntrack_dir dir,
		     u32 seq)
{
	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	struct nf_ct_seqadj *this_way;

	if (!seqadj)
		return 0;

	this_way = &seqadj->seq[dir];
	return after(seq, this_way->correction_pos) ?
		 this_way->offset_after : this_way->offset_before;
}
EXPORT_SYMBOL_GPL(nf_ct_seq_offset);

static const struct nf_ct_ext_type nf_ct_seqadj_extend = {
	.len	= sizeof(struct nf_conn_seqadj),
	.align	= __alignof__(struct nf_conn_seqadj),
	.id	= NF_CT_EXT_SEQADJ,
};

int nf_conntrack_seqadj_init(void)
{
	return nf_ct_extend_register(&nf_ct_seqadj_extend);
}

void nf_conntrack_seqadj_fini(void)
{
	nf_ct_extend_unregister(&nf_ct_seqadj_extend);
}
Commit	Line	Data
41d73ec0 PM	1	#include <linux/types.h>
	2	#include <linux/netfilter.h>
	3	#include <net/tcp.h>
	4
	5	#include <net/netfilter/nf_conntrack.h>
	6	#include <net/netfilter/nf_conntrack_extend.h>
	7	#include <net/netfilter/nf_conntrack_seqadj.h>
	8
48b1de4c PM	9	int nf_ct_seqadj_init(struct nf_conn *ct, enum ip_conntrack_info ctinfo,
	10	s32 off)
	11	{
	12	enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
	13	struct nf_conn_seqadj *seqadj;
	14	struct nf_ct_seqadj *this_way;
	15
	16	if (off == 0)
	17	return 0;
	18
	19	set_bit(IPS_SEQ_ADJUST_BIT, &ct->status);
	20
	21	seqadj = nfct_seqadj(ct);
	22	this_way = &seqadj->seq[dir];
	23	this_way->offset_before = off;
	24	this_way->offset_after = off;
	25	return 0;
	26	}
	27	EXPORT_SYMBOL_GPL(nf_ct_seqadj_init);
	28
41d73ec0 PM	29	int nf_ct_seqadj_set(struct nf_conn *ct, enum ip_conntrack_info ctinfo,
	30	__be32 seq, s32 off)
	31	{
	32	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	33	enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
	34	struct nf_ct_seqadj *this_way;
	35
	36	if (off == 0)
	37	return 0;
	38
db12cf27	39	if (unlikely(!seqadj)) {
f2661adc	40	WARN_ONCE(1, "Missing nfct_seqadj_ext_add() setup call\n");
db12cf27 JDB	41	return 0;
	42	}
	43
41d73ec0 PM	44	set_bit(IPS_SEQ_ADJUST_BIT, &ct->status);
	45
	46	spin_lock_bh(&ct->lock);
	47	this_way = &seqadj->seq[dir];
	48	if (this_way->offset_before == this_way->offset_after \|\|
23dfe136 PO	49	before(this_way->correction_pos, ntohl(seq))) {
23dfe136 PO	50	this_way->correction_pos = ntohl(seq);
41d73ec0 PM	51	this_way->offset_before = this_way->offset_after;
	52	this_way->offset_after += off;
	53	}
	54	spin_unlock_bh(&ct->lock);
	55	return 0;
	56	}
	57	EXPORT_SYMBOL_GPL(nf_ct_seqadj_set);
	58
	59	void nf_ct_tcp_seqadj_set(struct sk_buff *skb,
	60	struct nf_conn *ct, enum ip_conntrack_info ctinfo,
	61	s32 off)
	62	{
	63	const struct tcphdr *th;
	64
	65	if (nf_ct_protonum(ct) != IPPROTO_TCP)
	66	return;
	67
	68	th = (struct tcphdr *)(skb_network_header(skb) + ip_hdrlen(skb));
	69	nf_ct_seqadj_set(ct, ctinfo, th->seq, off);
	70	}
	71	EXPORT_SYMBOL_GPL(nf_ct_tcp_seqadj_set);
	72
	73	/* Adjust one found SACK option including checksum correction */
	74	static void nf_ct_sack_block_adjust(struct sk_buff *skb,
	75	struct tcphdr *tcph,
	76	unsigned int sackoff,
	77	unsigned int sackend,
	78	struct nf_ct_seqadj *seq)
	79	{
	80	while (sackoff < sackend) {
	81	struct tcp_sack_block_wire *sack;
	82	__be32 new_start_seq, new_end_seq;
	83
	84	sack = (void *)skb->data + sackoff;
	85	if (after(ntohl(sack->start_seq) - seq->offset_before,
	86	seq->correction_pos))
	87	new_start_seq = htonl(ntohl(sack->start_seq) -
	88	seq->offset_after);
	89	else
	90	new_start_seq = htonl(ntohl(sack->start_seq) -
	91	seq->offset_before);
	92
	93	if (after(ntohl(sack->end_seq) - seq->offset_before,
	94	seq->correction_pos))
	95	new_end_seq = htonl(ntohl(sack->end_seq) -
	96	seq->offset_after);
	97	else
	98	new_end_seq = htonl(ntohl(sack->end_seq) -
	99	seq->offset_before);
	100
b44b565c G	101	pr_debug("sack_adjust: start_seq: %u->%u, end_seq: %u->%u\n",
	102	ntohl(sack->start_seq), ntohl(new_start_seq),
	103	ntohl(sack->end_seq), ntohl(new_end_seq));
41d73ec0 PM	104
41d73ec0 PM	105	inet_proto_csum_replace4(&tcph->check, skb,
4b048d6d	106	sack->start_seq, new_start_seq, false);
41d73ec0	107	inet_proto_csum_replace4(&tcph->check, skb,
4b048d6d	108	sack->end_seq, new_end_seq, false);
41d73ec0 PM	109	sack->start_seq = new_start_seq;
	110	sack->end_seq = new_end_seq;
	111	sackoff += sizeof(*sack);
	112	}
	113	}
	114
	115	/* TCP SACK sequence number adjustment */
	116	static unsigned int nf_ct_sack_adjust(struct sk_buff *skb,
	117	unsigned int protoff,
	118	struct tcphdr *tcph,
	119	struct nf_conn *ct,
	120	enum ip_conntrack_info ctinfo)
	121	{
	122	unsigned int dir, optoff, optend;
	123	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	124
	125	optoff = protoff + sizeof(struct tcphdr);
	126	optend = protoff + tcph->doff * 4;
	127
	128	if (!skb_make_writable(skb, optend))
	129	return 0;
	130
	131	dir = CTINFO2DIR(ctinfo);
	132
	133	while (optoff < optend) {
	134	/* Usually: option, length. */
	135	unsigned char *op = skb->data + optoff;
	136
	137	switch (op[0]) {
	138	case TCPOPT_EOL:
	139	return 1;
	140	case TCPOPT_NOP:
	141	optoff++;
	142	continue;
	143	default:
	144	/* no partial options */
	145	if (optoff + 1 == optend \|\|
	146	optoff + op[1] > optend \|\|
	147	op[1] < 2)
	148	return 0;
	149	if (op[0] == TCPOPT_SACK &&
	150	op[1] >= 2+TCPOLEN_SACK_PERBLOCK &&
	151	((op[1] - 2) % TCPOLEN_SACK_PERBLOCK) == 0)
	152	nf_ct_sack_block_adjust(skb, tcph, optoff + 2,
	153	optoff+op[1],
	154	&seqadj->seq[!dir]);
	155	optoff += op[1];
	156	}
	157	}
	158	return 1;
	159	}
	160
	161	/* TCP sequence number adjustment. Returns 1 on success, 0 on failure */
	162	int nf_ct_seq_adjust(struct sk_buff *skb,
	163	struct nf_conn *ct, enum ip_conntrack_info ctinfo,
	164	unsigned int protoff)
	165	{
	166	enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
	167	struct tcphdr *tcph;
	168	__be32 newseq, newack;
	169	s32 seqoff, ackoff;
	170	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	171	struct nf_ct_seqadj this_way, other_way;
8d11350f	172	int res = 1;
41d73ec0 PM	173
	174	this_way = &seqadj->seq[dir];
	175	other_way = &seqadj->seq[!dir];
	176
	177	if (!skb_make_writable(skb, protoff + sizeof(*tcph)))
	178	return 0;
	179
	180	tcph = (void *)skb->data + protoff;
	181	spin_lock_bh(&ct->lock);
	182	if (after(ntohl(tcph->seq), this_way->correction_pos))
	183	seqoff = this_way->offset_after;
	184	else
	185	seqoff = this_way->offset_before;
	186
8d11350f GF	187	newseq = htonl(ntohl(tcph->seq) + seqoff);
	188	inet_proto_csum_replace4(&tcph->check, skb, tcph->seq, newseq, false);
	189	pr_debug("Adjusting sequence number from %u->%u\n",
	190	ntohl(tcph->seq), ntohl(newseq));
	191	tcph->seq = newseq;
	192
	193	if (!tcph->ack)
	194	goto out;
	195
41d73ec0 PM	196	if (after(ntohl(tcph->ack_seq) - other_way->offset_before,
	197	other_way->correction_pos))
	198	ackoff = other_way->offset_after;
	199	else
	200	ackoff = other_way->offset_before;
	201
41d73ec0	202	newack = htonl(ntohl(tcph->ack_seq) - ackoff);
4b048d6d TH	203	inet_proto_csum_replace4(&tcph->check, skb, tcph->ack_seq, newack,
4b048d6d TH	204	false);
8d11350f	205	pr_debug("Adjusting ack number from %u->%u, ack from %u->%u\n",
41d73ec0 PM	206	ntohl(tcph->seq), ntohl(newseq), ntohl(tcph->ack_seq),
41d73ec0 PM	207	ntohl(newack));
41d73ec0 PM	208	tcph->ack_seq = newack;
	209
	210	res = nf_ct_sack_adjust(skb, protoff, tcph, ct, ctinfo);
8d11350f	211	out:
41d73ec0 PM	212	spin_unlock_bh(&ct->lock);
	213
	214	return res;
	215	}
	216	EXPORT_SYMBOL_GPL(nf_ct_seq_adjust);
	217
	218	s32 nf_ct_seq_offset(const struct nf_conn *ct,
	219	enum ip_conntrack_dir dir,
	220	u32 seq)
	221	{
	222	struct nf_conn_seqadj *seqadj = nfct_seqadj(ct);
	223	struct nf_ct_seqadj *this_way;
	224
	225	if (!seqadj)
	226	return 0;
	227
	228	this_way = &seqadj->seq[dir];
	229	return after(seq, this_way->correction_pos) ?
	230	this_way->offset_after : this_way->offset_before;
	231	}
	232	EXPORT_SYMBOL_GPL(nf_ct_seq_offset);
	233
23f671a1	234	static const struct nf_ct_ext_type nf_ct_seqadj_extend = {
41d73ec0 PM	235	.len = sizeof(struct nf_conn_seqadj),
	236	.align = __alignof__(struct nf_conn_seqadj),
	237	.id = NF_CT_EXT_SEQADJ,
	238	};
	239
	240	int nf_conntrack_seqadj_init(void)
	241	{
	242	return nf_ct_extend_register(&nf_ct_seqadj_extend);
	243	}
	244
	245	void nf_conntrack_seqadj_fini(void)
	246	{
	247	nf_ct_extend_unregister(&nf_ct_seqadj_extend);
	248	}