[linux.git] / net / ipv4 / syncookies.c

/*
 *  Syncookies implementation for the Linux kernel
 *
 *  Copyright (C) 1997 Andi Kleen
 *  Based on ideas by D.J.Bernstein and Eric Schenk.
 *
 *	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; either version
 *      2 of the License, or (at your option) any later version.
 */

#include <linux/tcp.h>
#include <linux/slab.h>
#include <linux/random.h>
#include <linux/cryptohash.h>
#include <linux/kernel.h>
#include <net/tcp.h>
#include <net/route.h>

/* Timestamps: lowest 9 bits store TCP options */
#define TSBITS 9
#define TSMASK (((__u32)1 << TSBITS) - 1)

extern int sysctl_tcp_syncookies;

__u32 syncookie_secret[2][16-4+SHA_DIGEST_WORDS];
EXPORT_SYMBOL(syncookie_secret);

static __init int init_syncookies(void)
{
	get_random_bytes(syncookie_secret, sizeof(syncookie_secret));
	return 0;
}
__initcall(init_syncookies);

#define COOKIEBITS 24	/* Upper bits store count */
#define COOKIEMASK (((__u32)1 << COOKIEBITS) - 1)

static DEFINE_PER_CPU(__u32, cookie_scratch)[16 + 5 + SHA_WORKSPACE_WORDS];

static u32 cookie_hash(__be32 saddr, __be32 daddr, __be16 sport, __be16 dport,
		       u32 count, int c)
{
	__u32 *tmp = __get_cpu_var(cookie_scratch);

	memcpy(tmp + 4, syncookie_secret[c], sizeof(syncookie_secret[c]));
	tmp[0] = (__force u32)saddr;
	tmp[1] = (__force u32)daddr;
	tmp[2] = ((__force u32)sport << 16) + (__force u32)dport;
	tmp[3] = count;
	sha_transform(tmp + 16, (__u8 *)tmp, tmp + 16 + 5);

	return tmp[17];
}


/*
 * when syncookies are in effect and tcp timestamps are enabled we encode
 * tcp options in the lowest 9 bits of the timestamp value that will be
 * sent in the syn-ack.
 * Since subsequent timestamps use the normal tcp_time_stamp value, we
 * must make sure that the resulting initial timestamp is <= tcp_time_stamp.
 */
__u32 cookie_init_timestamp(struct request_sock *req)
{
	struct inet_request_sock *ireq;
	u32 ts, ts_now = tcp_time_stamp;
	u32 options = 0;

	ireq = inet_rsk(req);
	if (ireq->wscale_ok) {
		options = ireq->snd_wscale;
		options |= ireq->rcv_wscale << 4;
	}
	options |= ireq->sack_ok << 8;

	ts = ts_now & ~TSMASK;
	ts |= options;
	if (ts > ts_now) {
		ts >>= TSBITS;
		ts--;
		ts <<= TSBITS;
		ts |= options;
	}
	return ts;
}


static __u32 secure_tcp_syn_cookie(__be32 saddr, __be32 daddr, __be16 sport,
				   __be16 dport, __u32 sseq, __u32 count,
				   __u32 data)
{
	/*
	 * Compute the secure sequence number.
	 * The output should be:
	 *   HASH(sec1,saddr,sport,daddr,dport,sec1) + sseq + (count * 2^24)
	 *      + (HASH(sec2,saddr,sport,daddr,dport,count,sec2) % 2^24).
	 * Where sseq is their sequence number and count increases every
	 * minute by 1.
	 * As an extra hack, we add a small "data" value that encodes the
	 * MSS into the second hash value.
	 */

	return (cookie_hash(saddr, daddr, sport, dport, 0, 0) +
		sseq + (count << COOKIEBITS) +
		((cookie_hash(saddr, daddr, sport, dport, count, 1) + data)
		 & COOKIEMASK));
}

/*
 * This retrieves the small "data" value from the syncookie.
 * If the syncookie is bad, the data returned will be out of
 * range.  This must be checked by the caller.
 *
 * The count value used to generate the cookie must be within
 * "maxdiff" if the current (passed-in) "count".  The return value
 * is (__u32)-1 if this test fails.
 */
static __u32 check_tcp_syn_cookie(__u32 cookie, __be32 saddr, __be32 daddr,
				  __be16 sport, __be16 dport, __u32 sseq,
				  __u32 count, __u32 maxdiff)
{
	__u32 diff;

	/* Strip away the layers from the cookie */
	cookie -= cookie_hash(saddr, daddr, sport, dport, 0, 0) + sseq;

	/* Cookie is now reduced to (count * 2^24) ^ (hash % 2^24) */
	diff = (count - (cookie >> COOKIEBITS)) & ((__u32) - 1 >> COOKIEBITS);
	if (diff >= maxdiff)
		return (__u32)-1;

	return (cookie -
		cookie_hash(saddr, daddr, sport, dport, count - diff, 1))
		& COOKIEMASK;	/* Leaving the data behind */
}

/*
 * This table has to be sorted and terminated with (__u16)-1.
 * XXX generate a better table.
 * Unresolved Issues: HIPPI with a 64k MSS is not well supported.
 */
static __u16 const msstab[] = {
	64 - 1,
	256 - 1,
	512 - 1,
	536 - 1,
	1024 - 1,
	1440 - 1,
	1460 - 1,
	4312 - 1,
	(__u16)-1
};
/* The number doesn't include the -1 terminator */
#define NUM_MSS (ARRAY_SIZE(msstab) - 1)

/*
 * Generate a syncookie.  mssp points to the mss, which is returned
 * rounded down to the value encoded in the cookie.
 */
__u32 cookie_v4_init_sequence(struct sock *sk, struct sk_buff *skb, __u16 *mssp)
{
	struct tcp_sock *tp = tcp_sk(sk);
	const struct iphdr *iph = ip_hdr(skb);
	const struct tcphdr *th = tcp_hdr(skb);
	int mssind;
	const __u16 mss = *mssp;

	tp->last_synq_overflow = jiffies;

	/* XXX sort msstab[] by probability?  Binary search? */
	for (mssind = 0; mss > msstab[mssind + 1]; mssind++)
		;
	*mssp = msstab[mssind] + 1;

	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESSENT);

	return secure_tcp_syn_cookie(iph->saddr, iph->daddr,
				     th->source, th->dest, ntohl(th->seq),
				     jiffies / (HZ * 60), mssind);
}

/*
 * This (misnamed) value is the age of syncookie which is permitted.
 * Its ideal value should be dependent on TCP_TIMEOUT_INIT and
 * sysctl_tcp_retries1. It's a rather complicated formula (exponential
 * backoff) to compute at runtime so it's currently hardcoded here.
 */
#define COUNTER_TRIES 4
/*
 * Check if a ack sequence number is a valid syncookie.
 * Return the decoded mss if it is, or 0 if not.
 */
static inline int cookie_check(struct sk_buff *skb, __u32 cookie)
{
	const struct iphdr *iph = ip_hdr(skb);
	const struct tcphdr *th = tcp_hdr(skb);
	__u32 seq = ntohl(th->seq) - 1;
	__u32 mssind = check_tcp_syn_cookie(cookie, iph->saddr, iph->daddr,
					    th->source, th->dest, seq,
					    jiffies / (HZ * 60),
					    COUNTER_TRIES);

	return mssind < NUM_MSS ? msstab[mssind] + 1 : 0;
}

static inline struct sock *get_cookie_sock(struct sock *sk, struct sk_buff *skb,
					   struct request_sock *req,
					   struct dst_entry *dst)
{
	struct inet_connection_sock *icsk = inet_csk(sk);
	struct sock *child;

	child = icsk->icsk_af_ops->syn_recv_sock(sk, skb, req, dst);
	if (child)
		inet_csk_reqsk_queue_add(sk, req, child);
	else
		reqsk_free(req);

	return child;
}


/*
 * when syncookies are in effect and tcp timestamps are enabled we stored
 * additional tcp options in the timestamp.
 * This extracts these options from the timestamp echo.
 *
 * The lowest 4 bits are for snd_wscale
 * The next 4 lsb are for rcv_wscale
 * The next lsb is for sack_ok
 */
void cookie_check_timestamp(struct tcp_options_received *tcp_opt)
{
	/* echoed timestamp, 9 lowest bits contain options */
	u32 options = tcp_opt->rcv_tsecr & TSMASK;

	tcp_opt->snd_wscale = options & 0xf;
	options >>= 4;
	tcp_opt->rcv_wscale = options & 0xf;

	tcp_opt->sack_ok = (options >> 4) & 0x1;

	if (tcp_opt->sack_ok)
		tcp_sack_reset(tcp_opt);

	if (tcp_opt->snd_wscale || tcp_opt->rcv_wscale)
		tcp_opt->wscale_ok = 1;
}
EXPORT_SYMBOL(cookie_check_timestamp);

struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
			     struct ip_options *opt)
{
	struct inet_request_sock *ireq;
	struct tcp_request_sock *treq;
	struct tcp_sock *tp = tcp_sk(sk);
	const struct tcphdr *th = tcp_hdr(skb);
	__u32 cookie = ntohl(th->ack_seq) - 1;
	struct sock *ret = sk;
	struct request_sock *req;
	int mss;
	struct rtable *rt;
	__u8 rcv_wscale;
	struct tcp_options_received tcp_opt;

	if (!sysctl_tcp_syncookies || !th->ack)
		goto out;

	if (time_after(jiffies, tp->last_synq_overflow + TCP_TIMEOUT_INIT) ||
	    (mss = cookie_check(skb, cookie)) == 0) {
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESFAILED);
		goto out;
	}

	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESRECV);

	/* check for timestamp cookie support */
	memset(&tcp_opt, 0, sizeof(tcp_opt));
	tcp_parse_options(skb, &tcp_opt, 0);

	if (tcp_opt.saw_tstamp)
		cookie_check_timestamp(&tcp_opt);

	ret = NULL;
	req = inet_reqsk_alloc(&tcp_request_sock_ops); /* for safety */
	if (!req)
		goto out;

	ireq = inet_rsk(req);
	treq = tcp_rsk(req);
	treq->rcv_isn		= ntohl(th->seq) - 1;
	treq->snt_isn		= cookie;
	req->mss		= mss;
	ireq->loc_port		= th->dest;
	ireq->rmt_port		= th->source;
	ireq->loc_addr		= ip_hdr(skb)->daddr;
	ireq->rmt_addr		= ip_hdr(skb)->saddr;
	ireq->ecn_ok		= 0;
	ireq->snd_wscale	= tcp_opt.snd_wscale;
	ireq->rcv_wscale	= tcp_opt.rcv_wscale;
	ireq->sack_ok		= tcp_opt.sack_ok;
	ireq->wscale_ok		= tcp_opt.wscale_ok;
	ireq->tstamp_ok		= tcp_opt.saw_tstamp;
	req->ts_recent		= tcp_opt.saw_tstamp ? tcp_opt.rcv_tsval : 0;

	/* We throwed the options of the initial SYN away, so we hope
	 * the ACK carries the same options again (see RFC1122 4.2.3.8)
	 */
	if (opt && opt->optlen) {
		int opt_size = sizeof(struct ip_options) + opt->optlen;

		ireq->opt = kmalloc(opt_size, GFP_ATOMIC);
		if (ireq->opt != NULL && ip_options_echo(ireq->opt, skb)) {
			kfree(ireq->opt);
			ireq->opt = NULL;
		}
	}

	if (security_inet_conn_request(sk, skb, req)) {
		reqsk_free(req);
		goto out;
	}

	req->expires	= 0UL;
	req->retrans	= 0;

	/*
	 * We need to lookup the route here to get at the correct
	 * window size. We should better make sure that the window size
	 * hasn't changed since we received the original syn, but I see
	 * no easy way to do this.
	 */
	{
		struct flowi fl = { .nl_u = { .ip4_u =
					      { .daddr = ((opt && opt->srr) ?
							  opt->faddr :
							  ireq->rmt_addr),
						.saddr = ireq->loc_addr,
						.tos = RT_CONN_FLAGS(sk) } },
				    .proto = IPPROTO_TCP,
				    .flags = inet_sk_flowi_flags(sk),
				    .uli_u = { .ports =
					       { .sport = th->dest,
						 .dport = th->source } } };
		security_req_classify_flow(req, &fl);
		if (ip_route_output_key(&init_net, &rt, &fl)) {
			reqsk_free(req);
			goto out;
		}
	}

	/* Try to redo what tcp_v4_send_synack did. */
	req->window_clamp = tp->window_clamp ? :dst_metric(&rt->u.dst, RTAX_WINDOW);

	tcp_select_initial_window(tcp_full_space(sk), req->mss,
				  &req->rcv_wnd, &req->window_clamp,
				  ireq->wscale_ok, &rcv_wscale);

	ireq->rcv_wscale  = rcv_wscale;

	ret = get_cookie_sock(sk, skb, req, &rt->u.dst);
out:	return ret;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Syncookies implementation for the Linux kernel
	3	*
	4	* Copyright (C) 1997 Andi Kleen
e905a9ed	5	* Based on ideas by D.J.Bernstein and Eric Schenk.
1da177e4 LT	6	*
	7	* This program is free software; you can redistribute it and/or
	8	* modify it under the terms of the GNU General Public License
	9	* as published by the Free Software Foundation; either version
	10	* 2 of the License, or (at your option) any later version.
1da177e4 LT	11	*/
	12
	13	#include <linux/tcp.h>
	14	#include <linux/slab.h>
	15	#include <linux/random.h>
	16	#include <linux/cryptohash.h>
	17	#include <linux/kernel.h>
	18	#include <net/tcp.h>
86b08d86	19	#include <net/route.h>
1da177e4	20
4dfc2817 FW	21	/* Timestamps: lowest 9 bits store TCP options */
	22	#define TSBITS 9
	23	#define TSMASK (((__u32)1 << TSBITS) - 1)
	24
1da177e4 LT	25	extern int sysctl_tcp_syncookies;
1da177e4 LT	26
2051f11f	27	__u32 syncookie_secret[2][16-4+SHA_DIGEST_WORDS];
c6aefafb	28	EXPORT_SYMBOL(syncookie_secret);
1da177e4 LT	29
	30	static __init int init_syncookies(void)
	31	{
	32	get_random_bytes(syncookie_secret, sizeof(syncookie_secret));
	33	return 0;
	34	}
c6aefafb	35	__initcall(init_syncookies);
1da177e4 LT	36
	37	#define COOKIEBITS 24 /* Upper bits store count */
	38	#define COOKIEMASK (((__u32)1 << COOKIEBITS) - 1)
	39
11baab7a ED	40	static DEFINE_PER_CPU(__u32, cookie_scratch)[16 + 5 + SHA_WORKSPACE_WORDS];
11baab7a ED	41
714e85be	42	static u32 cookie_hash(__be32 saddr, __be32 daddr, __be16 sport, __be16 dport,
1da177e4 LT	43	u32 count, int c)
1da177e4 LT	44	{
11baab7a	45	__u32 *tmp = __get_cpu_var(cookie_scratch);
1da177e4	46
2051f11f	47	memcpy(tmp + 4, syncookie_secret[c], sizeof(syncookie_secret[c]));
714e85be AV	48	tmp[0] = (__force u32)saddr;
	49	tmp[1] = (__force u32)daddr;
	50	tmp[2] = ((__force u32)sport << 16) + (__force u32)dport;
1da177e4 LT	51	tmp[3] = count;
	52	sha_transform(tmp + 16, (__u8 *)tmp, tmp + 16 + 5);
	53
	54	return tmp[17];
	55	}
	56
4dfc2817 FW	57
	58	/*
	59	* when syncookies are in effect and tcp timestamps are enabled we encode
	60	* tcp options in the lowest 9 bits of the timestamp value that will be
	61	* sent in the syn-ack.
	62	* Since subsequent timestamps use the normal tcp_time_stamp value, we
	63	* must make sure that the resulting initial timestamp is <= tcp_time_stamp.
	64	*/
	65	__u32 cookie_init_timestamp(struct request_sock *req)
	66	{
	67	struct inet_request_sock *ireq;
	68	u32 ts, ts_now = tcp_time_stamp;
	69	u32 options = 0;
	70
	71	ireq = inet_rsk(req);
	72	if (ireq->wscale_ok) {
	73	options = ireq->snd_wscale;
	74	options \|= ireq->rcv_wscale << 4;
	75	}
	76	options \|= ireq->sack_ok << 8;
	77
	78	ts = ts_now & ~TSMASK;
	79	ts \|= options;
	80	if (ts > ts_now) {
	81	ts >>= TSBITS;
	82	ts--;
	83	ts <<= TSBITS;
	84	ts \|= options;
	85	}
	86	return ts;
	87	}
	88
	89
714e85be AV	90	static __u32 secure_tcp_syn_cookie(__be32 saddr, __be32 daddr, __be16 sport,
714e85be AV	91	__be16 dport, __u32 sseq, __u32 count,
1da177e4 LT	92	__u32 data)
	93	{
	94	/*
	95	* Compute the secure sequence number.
	96	* The output should be:
e905a9ed	97	* HASH(sec1,saddr,sport,daddr,dport,sec1) + sseq + (count * 2^24)
1da177e4 LT	98	* + (HASH(sec2,saddr,sport,daddr,dport,count,sec2) % 2^24).
	99	* Where sseq is their sequence number and count increases every
	100	* minute by 1.
	101	* As an extra hack, we add a small "data" value that encodes the
	102	* MSS into the second hash value.
	103	*/
	104
	105	return (cookie_hash(saddr, daddr, sport, dport, 0, 0) +
	106	sseq + (count << COOKIEBITS) +
	107	((cookie_hash(saddr, daddr, sport, dport, count, 1) + data)
	108	& COOKIEMASK));
	109	}
	110
	111	/*
	112	* This retrieves the small "data" value from the syncookie.
	113	* If the syncookie is bad, the data returned will be out of
	114	* range. This must be checked by the caller.
	115	*
	116	* The count value used to generate the cookie must be within
	117	* "maxdiff" if the current (passed-in) "count". The return value
	118	* is (__u32)-1 if this test fails.
	119	*/
714e85be AV	120	static __u32 check_tcp_syn_cookie(__u32 cookie, __be32 saddr, __be32 daddr,
714e85be AV	121	__be16 sport, __be16 dport, __u32 sseq,
1da177e4 LT	122	__u32 count, __u32 maxdiff)
	123	{
	124	__u32 diff;
	125
	126	/* Strip away the layers from the cookie */
	127	cookie -= cookie_hash(saddr, daddr, sport, dport, 0, 0) + sseq;
	128
	129	/* Cookie is now reduced to (count * 2^24) ^ (hash % 2^24) */
	130	diff = (count - (cookie >> COOKIEBITS)) & ((__u32) - 1 >> COOKIEBITS);
	131	if (diff >= maxdiff)
	132	return (__u32)-1;
	133
	134	return (cookie -
	135	cookie_hash(saddr, daddr, sport, dport, count - diff, 1))
	136	& COOKIEMASK; /* Leaving the data behind */
	137	}
	138
e905a9ed	139	/*
1da177e4 LT	140	* This table has to be sorted and terminated with (__u16)-1.
	141	* XXX generate a better table.
	142	* Unresolved Issues: HIPPI with a 64k MSS is not well supported.
	143	*/
	144	static __u16 const msstab[] = {
	145	64 - 1,
e905a9ed	146	256 - 1,
1da177e4 LT	147	512 - 1,
1da177e4 LT	148	536 - 1,
e905a9ed	149	1024 - 1,
1da177e4 LT	150	1440 - 1,
	151	1460 - 1,
	152	4312 - 1,
	153	(__u16)-1
	154	};
	155	/* The number doesn't include the -1 terminator */
	156	#define NUM_MSS (ARRAY_SIZE(msstab) - 1)
	157
	158	/*
	159	* Generate a syncookie. mssp points to the mss, which is returned
	160	* rounded down to the value encoded in the cookie.
	161	*/
	162	__u32 cookie_v4_init_sequence(struct sock sk, struct sk_buff skb, __u16 *mssp)
	163	{
	164	struct tcp_sock *tp = tcp_sk(sk);
aa8223c7 ACM	165	const struct iphdr *iph = ip_hdr(skb);
aa8223c7 ACM	166	const struct tcphdr *th = tcp_hdr(skb);
1da177e4 LT	167	int mssind;
	168	const __u16 mss = *mssp;
	169
1da177e4 LT	170	tp->last_synq_overflow = jiffies;
	171
	172	/* XXX sort msstab[] by probability? Binary search? */
	173	for (mssind = 0; mss > msstab[mssind + 1]; mssind++)
	174	;
	175	*mssp = msstab[mssind] + 1;
	176
de0744af	177	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESSENT);
1da177e4	178
aa8223c7 ACM	179	return secure_tcp_syn_cookie(iph->saddr, iph->daddr,
aa8223c7 ACM	180	th->source, th->dest, ntohl(th->seq),
1da177e4 LT	181	jiffies / (HZ * 60), mssind);
	182	}
	183
e905a9ed	184	/*
1da177e4 LT	185	* This (misnamed) value is the age of syncookie which is permitted.
	186	* Its ideal value should be dependent on TCP_TIMEOUT_INIT and
	187	* sysctl_tcp_retries1. It's a rather complicated formula (exponential
	188	* backoff) to compute at runtime so it's currently hardcoded here.
	189	*/
	190	#define COUNTER_TRIES 4
e905a9ed YH	191	/*
e905a9ed YH	192	* Check if a ack sequence number is a valid syncookie.
1da177e4 LT	193	* Return the decoded mss if it is, or 0 if not.
	194	*/
	195	static inline int cookie_check(struct sk_buff *skb, __u32 cookie)
	196	{
aa8223c7 ACM	197	const struct iphdr *iph = ip_hdr(skb);
	198	const struct tcphdr *th = tcp_hdr(skb);
	199	__u32 seq = ntohl(th->seq) - 1;
	200	__u32 mssind = check_tcp_syn_cookie(cookie, iph->saddr, iph->daddr,
	201	th->source, th->dest, seq,
	202	jiffies / (HZ * 60),
	203	COUNTER_TRIES);
1da177e4 LT	204
	205	return mssind < NUM_MSS ? msstab[mssind] + 1 : 0;
	206	}
	207
1da177e4	208	static inline struct sock get_cookie_sock(struct sock sk, struct sk_buff *skb,
60236fdd	209	struct request_sock *req,
1da177e4 LT	210	struct dst_entry *dst)
1da177e4 LT	211	{
8292a17a	212	struct inet_connection_sock *icsk = inet_csk(sk);
1da177e4 LT	213	struct sock *child;
1da177e4 LT	214
8292a17a	215	child = icsk->icsk_af_ops->syn_recv_sock(sk, skb, req, dst);
1da177e4	216	if (child)
463c84b9	217	inet_csk_reqsk_queue_add(sk, req, child);
1da177e4	218	else
60236fdd	219	reqsk_free(req);
1da177e4 LT	220
	221	return child;
	222	}
	223
4dfc2817 FW	224
	225	/*
	226	* when syncookies are in effect and tcp timestamps are enabled we stored
	227	* additional tcp options in the timestamp.
	228	* This extracts these options from the timestamp echo.
	229	*
	230	* The lowest 4 bits are for snd_wscale
	231	* The next 4 lsb are for rcv_wscale
	232	* The next lsb is for sack_ok
	233	*/
	234	void cookie_check_timestamp(struct tcp_options_received *tcp_opt)
	235	{
	236	/* echoed timestamp, 9 lowest bits contain options */
	237	u32 options = tcp_opt->rcv_tsecr & TSMASK;
	238
	239	tcp_opt->snd_wscale = options & 0xf;
	240	options >>= 4;
	241	tcp_opt->rcv_wscale = options & 0xf;
	242
	243	tcp_opt->sack_ok = (options >> 4) & 0x1;
	244
	245	if (tcp_opt->sack_ok)
	246	tcp_sack_reset(tcp_opt);
	247
	248	if (tcp_opt->snd_wscale \|\| tcp_opt->rcv_wscale)
	249	tcp_opt->wscale_ok = 1;
	250	}
	251	EXPORT_SYMBOL(cookie_check_timestamp);
	252
1da177e4 LT	253	struct sock cookie_v4_check(struct sock sk, struct sk_buff *skb,
	254	struct ip_options *opt)
	255	{
2e6599cb ACM	256	struct inet_request_sock *ireq;
2e6599cb ACM	257	struct tcp_request_sock *treq;
1da177e4	258	struct tcp_sock *tp = tcp_sk(sk);
aa8223c7 ACM	259	const struct tcphdr *th = tcp_hdr(skb);
aa8223c7 ACM	260	__u32 cookie = ntohl(th->ack_seq) - 1;
1da177e4	261	struct sock *ret = sk;
e905a9ed YH	262	struct request_sock *req;
	263	int mss;
	264	struct rtable *rt;
1da177e4	265	__u8 rcv_wscale;
4dfc2817	266	struct tcp_options_received tcp_opt;
1da177e4	267
aa8223c7	268	if (!sysctl_tcp_syncookies \|\| !th->ack)
1da177e4 LT	269	goto out;
1da177e4 LT	270
e905a9ed	271	if (time_after(jiffies, tp->last_synq_overflow + TCP_TIMEOUT_INIT) \|\|
1da177e4	272	(mss = cookie_check(skb, cookie)) == 0) {
de0744af	273	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESFAILED);
1da177e4 LT	274	goto out;
	275	}
	276
de0744af	277	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESRECV);
1da177e4	278
4dfc2817 FW	279	/* check for timestamp cookie support */
	280	memset(&tcp_opt, 0, sizeof(tcp_opt));
	281	tcp_parse_options(skb, &tcp_opt, 0);
	282
	283	if (tcp_opt.saw_tstamp)
	284	cookie_check_timestamp(&tcp_opt);
	285
1da177e4	286	ret = NULL;
ce4a7d0d	287	req = inet_reqsk_alloc(&tcp_request_sock_ops); /* for safety */
1da177e4 LT	288	if (!req)
	289	goto out;
	290
2e6599cb ACM	291	ireq = inet_rsk(req);
2e6599cb ACM	292	treq = tcp_rsk(req);
aa8223c7	293	treq->rcv_isn = ntohl(th->seq) - 1;
e905a9ed	294	treq->snt_isn = cookie;
1da177e4	295	req->mss = mss;
a3116ac5	296	ireq->loc_port = th->dest;
aa8223c7	297	ireq->rmt_port = th->source;
eddc9ec5 ACM	298	ireq->loc_addr = ip_hdr(skb)->daddr;
eddc9ec5 ACM	299	ireq->rmt_addr = ip_hdr(skb)->saddr;
16df845f	300	ireq->ecn_ok = 0;
4dfc2817 FW	301	ireq->snd_wscale = tcp_opt.snd_wscale;
	302	ireq->rcv_wscale = tcp_opt.rcv_wscale;
	303	ireq->sack_ok = tcp_opt.sack_ok;
	304	ireq->wscale_ok = tcp_opt.wscale_ok;
	305	ireq->tstamp_ok = tcp_opt.saw_tstamp;
	306	req->ts_recent = tcp_opt.saw_tstamp ? tcp_opt.rcv_tsval : 0;
1da177e4 LT	307
	308	/* We throwed the options of the initial SYN away, so we hope
	309	* the ACK carries the same options again (see RFC1122 4.2.3.8)
	310	*/
	311	if (opt && opt->optlen) {
	312	int opt_size = sizeof(struct ip_options) + opt->optlen;
	313
2e6599cb ACM	314	ireq->opt = kmalloc(opt_size, GFP_ATOMIC);
	315	if (ireq->opt != NULL && ip_options_echo(ireq->opt, skb)) {
	316	kfree(ireq->opt);
	317	ireq->opt = NULL;
1da177e4 LT	318	}
	319	}
	320
284904aa PM	321	if (security_inet_conn_request(sk, skb, req)) {
	322	reqsk_free(req);
	323	goto out;
	324	}
	325
e905a9ed YH	326	req->expires = 0UL;
	327	req->retrans = 0;
	328
1da177e4 LT	329	/*
	330	* We need to lookup the route here to get at the correct
	331	* window size. We should better make sure that the window size
	332	* hasn't changed since we received the original syn, but I see
e905a9ed	333	* no easy way to do this.
1da177e4 LT	334	*/
	335	{
	336	struct flowi fl = { .nl_u = { .ip4_u =
	337	{ .daddr = ((opt && opt->srr) ?
	338	opt->faddr :
2e6599cb ACM	339	ireq->rmt_addr),
2e6599cb ACM	340	.saddr = ireq->loc_addr,
1da177e4 LT	341	.tos = RT_CONN_FLAGS(sk) } },
1da177e4 LT	342	.proto = IPPROTO_TCP,
86b08d86	343	.flags = inet_sk_flowi_flags(sk),
1da177e4	344	.uli_u = { .ports =
aa8223c7 ACM	345	{ .sport = th->dest,
aa8223c7 ACM	346	.dport = th->source } } };
4237c75c	347	security_req_classify_flow(req, &fl);
f206351a	348	if (ip_route_output_key(&init_net, &rt, &fl)) {
60236fdd	349	reqsk_free(req);
e905a9ed	350	goto out;
1da177e4 LT	351	}
	352	}
	353
	354	/* Try to redo what tcp_v4_send_synack did. */
4dfc2817 FW	355	req->window_clamp = tp->window_clamp ? :dst_metric(&rt->u.dst, RTAX_WINDOW);
4dfc2817 FW	356
1da177e4	357	tcp_select_initial_window(tcp_full_space(sk), req->mss,
e905a9ed	358	&req->rcv_wnd, &req->window_clamp,
4dfc2817 FW	359	ireq->wscale_ok, &rcv_wscale);
4dfc2817 FW	360
e905a9ed	361	ireq->rcv_wscale = rcv_wscale;
1da177e4 LT	362
	363	ret = get_cookie_sock(sk, skb, req, &rt->u.dst);
	364	out: return ret;
	365	}