Blame view
net/ipv4/syncookies.c
10.5 KB
1da177e4c
|
1 2 3 4 |
/* * Syncookies implementation for the Linux kernel * * Copyright (C) 1997 Andi Kleen |
e905a9eda
|
5 |
* Based on ideas by D.J.Bernstein and Eric Schenk. |
1da177e4c
|
6 7 8 9 10 |
* * 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. |
1da177e4c
|
11 12 13 14 15 16 17 |
*/ #include <linux/tcp.h> #include <linux/slab.h> #include <linux/random.h> #include <linux/cryptohash.h> #include <linux/kernel.h> |
bc3b2d7fb
|
18 |
#include <linux/export.h> |
1da177e4c
|
19 |
#include <net/tcp.h> |
86b08d867
|
20 |
#include <net/route.h> |
1da177e4c
|
21 |
|
734f614bc
|
22 |
/* Timestamps: lowest bits store TCP options */ |
172d69e63
|
23 |
#define TSBITS 6 |
4dfc28170
|
24 |
#define TSMASK (((__u32)1 << TSBITS) - 1) |
1da177e4c
|
25 |
extern int sysctl_tcp_syncookies; |
b23a002fc
|
26 |
static u32 syncookie_secret[2][16-4+SHA_DIGEST_WORDS]; |
1da177e4c
|
27 28 29 |
#define COOKIEBITS 24 /* Upper bits store count */ #define COOKIEMASK (((__u32)1 << COOKIEBITS) - 1) |
245b2e70e
|
30 31 |
static DEFINE_PER_CPU(__u32 [16 + 5 + SHA_WORKSPACE_WORDS], ipv4_cookie_scratch); |
11baab7ac
|
32 |
|
714e85be3
|
33 |
static u32 cookie_hash(__be32 saddr, __be32 daddr, __be16 sport, __be16 dport, |
1da177e4c
|
34 35 |
u32 count, int c) { |
b23a002fc
|
36 37 38 |
__u32 *tmp; net_get_random_once(syncookie_secret, sizeof(syncookie_secret)); |
1da177e4c
|
39 |
|
b23a002fc
|
40 |
tmp = __get_cpu_var(ipv4_cookie_scratch); |
2051f11fb
|
41 |
memcpy(tmp + 4, syncookie_secret[c], sizeof(syncookie_secret[c])); |
714e85be3
|
42 43 44 |
tmp[0] = (__force u32)saddr; tmp[1] = (__force u32)daddr; tmp[2] = ((__force u32)sport << 16) + (__force u32)dport; |
1da177e4c
|
45 46 47 48 49 |
tmp[3] = count; sha_transform(tmp + 16, (__u8 *)tmp, tmp + 16 + 5); return tmp[17]; } |
4dfc28170
|
50 51 52 |
/* * when syncookies are in effect and tcp timestamps are enabled we encode |
734f614bc
|
53 |
* tcp options in the lower bits of the timestamp value that will be |
4dfc28170
|
54 55 56 57 58 59 60 61 62 63 64 |
* 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); |
734f614bc
|
65 66 67 |
options = ireq->wscale_ok ? ireq->snd_wscale : 0xf; options |= ireq->sack_ok << 4; |
172d69e63
|
68 |
options |= ireq->ecn_ok << 5; |
4dfc28170
|
69 70 71 72 73 74 75 76 77 78 79 |
ts = ts_now & ~TSMASK; ts |= options; if (ts > ts_now) { ts >>= TSBITS; ts--; ts <<= TSBITS; ts |= options; } return ts; } |
714e85be3
|
80 |
static __u32 secure_tcp_syn_cookie(__be32 saddr, __be32 daddr, __be16 sport, |
8c27bd75f
|
81 |
__be16 dport, __u32 sseq, __u32 data) |
1da177e4c
|
82 83 84 85 |
{ /* * Compute the secure sequence number. * The output should be: |
e905a9eda
|
86 |
* HASH(sec1,saddr,sport,daddr,dport,sec1) + sseq + (count * 2^24) |
1da177e4c
|
87 88 89 90 91 92 |
* + (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. */ |
8c27bd75f
|
93 |
u32 count = tcp_cookie_time(); |
1da177e4c
|
94 95 96 97 98 99 100 101 102 103 104 |
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. * |
8c27bd75f
|
105 106 107 |
* The count value used to generate the cookie must be less than * MAX_SYNCOOKIE_AGE minutes in the past. * The return value (__u32)-1 if this test fails. |
1da177e4c
|
108 |
*/ |
714e85be3
|
109 |
static __u32 check_tcp_syn_cookie(__u32 cookie, __be32 saddr, __be32 daddr, |
8c27bd75f
|
110 |
__be16 sport, __be16 dport, __u32 sseq) |
1da177e4c
|
111 |
{ |
8c27bd75f
|
112 |
u32 diff, count = tcp_cookie_time(); |
1da177e4c
|
113 114 115 116 117 |
/* 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) */ |
0c9a67d2e
|
118 |
diff = (count - (cookie >> COOKIEBITS)) & ((__u32) -1 >> COOKIEBITS); |
8c27bd75f
|
119 |
if (diff >= MAX_SYNCOOKIE_AGE) |
1da177e4c
|
120 121 122 123 124 125 |
return (__u32)-1; return (cookie - cookie_hash(saddr, daddr, sport, dport, count - diff, 1)) & COOKIEMASK; /* Leaving the data behind */ } |
e905a9eda
|
126 |
/* |
086293542
|
127 128 129 130 131 132 133 |
* MSS Values are chosen based on the 2011 paper * 'An Analysis of TCP Maximum Segement Sizes' by S. Alcock and R. Nelson. * Values .. * .. lower than 536 are rare (< 0.2%) * .. between 537 and 1299 account for less than < 1.5% of observed values * .. in the 1300-1349 range account for about 15 to 20% of observed mss values * .. exceeding 1460 are very rare (< 0.04%) |
5918e2fb9
|
134 |
* |
086293542
|
135 136 |
* 1460 is the single most frequently announced mss value (30 to 46% depending * on monitor location). Table must be sorted. |
1da177e4c
|
137 138 |
*/ static __u16 const msstab[] = { |
5918e2fb9
|
139 |
536, |
086293542
|
140 141 |
1300, 1440, /* 1440, 1452: PPPoE */ |
5918e2fb9
|
142 |
1460, |
1da177e4c
|
143 |
}; |
1da177e4c
|
144 145 146 147 148 |
/* * Generate a syncookie. mssp points to the mss, which is returned * rounded down to the value encoded in the cookie. */ |
0198230b7
|
149 150 |
u32 __cookie_v4_init_sequence(const struct iphdr *iph, const struct tcphdr *th, u16 *mssp) |
1da177e4c
|
151 |
{ |
1da177e4c
|
152 153 |
int mssind; const __u16 mss = *mssp; |
5918e2fb9
|
154 155 156 157 |
for (mssind = ARRAY_SIZE(msstab) - 1; mssind ; mssind--) if (mss >= msstab[mssind]) break; *mssp = msstab[mssind]; |
1da177e4c
|
158 |
|
aa8223c7b
|
159 160 |
return secure_tcp_syn_cookie(iph->saddr, iph->daddr, th->source, th->dest, ntohl(th->seq), |
8c27bd75f
|
161 |
mssind); |
1da177e4c
|
162 |
} |
0198230b7
|
163 164 165 166 167 168 169 170 171 172 173 174 |
EXPORT_SYMBOL_GPL(__cookie_v4_init_sequence); __u32 cookie_v4_init_sequence(struct sock *sk, struct sk_buff *skb, __u16 *mssp) { const struct iphdr *iph = ip_hdr(skb); const struct tcphdr *th = tcp_hdr(skb); tcp_synq_overflow(sk); NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESSENT); return __cookie_v4_init_sequence(iph, th, mssp); } |
1da177e4c
|
175 |
|
e905a9eda
|
176 |
/* |
e905a9eda
|
177 |
* Check if a ack sequence number is a valid syncookie. |
1da177e4c
|
178 179 |
* Return the decoded mss if it is, or 0 if not. */ |
0198230b7
|
180 181 |
int __cookie_v4_check(const struct iphdr *iph, const struct tcphdr *th, u32 cookie) |
1da177e4c
|
182 |
{ |
aa8223c7b
|
183 184 |
__u32 seq = ntohl(th->seq) - 1; __u32 mssind = check_tcp_syn_cookie(cookie, iph->saddr, iph->daddr, |
8c27bd75f
|
185 |
th->source, th->dest, seq); |
1da177e4c
|
186 |
|
5918e2fb9
|
187 |
return mssind < ARRAY_SIZE(msstab) ? msstab[mssind] : 0; |
1da177e4c
|
188 |
} |
0198230b7
|
189 |
EXPORT_SYMBOL_GPL(__cookie_v4_check); |
1da177e4c
|
190 |
|
1da177e4c
|
191 |
static inline struct sock *get_cookie_sock(struct sock *sk, struct sk_buff *skb, |
60236fdd0
|
192 |
struct request_sock *req, |
1da177e4c
|
193 194 |
struct dst_entry *dst) { |
8292a17a3
|
195 |
struct inet_connection_sock *icsk = inet_csk(sk); |
1da177e4c
|
196 |
struct sock *child; |
8292a17a3
|
197 |
child = icsk->icsk_af_ops->syn_recv_sock(sk, skb, req, dst); |
1da177e4c
|
198 |
if (child) |
463c84b97
|
199 |
inet_csk_reqsk_queue_add(sk, req, child); |
1da177e4c
|
200 |
else |
60236fdd0
|
201 |
reqsk_free(req); |
1da177e4c
|
202 203 204 |
return child; } |
4dfc28170
|
205 206 207 208 209 210 |
/* * 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. * |
734f614bc
|
211 |
* The lowest 4 bits store snd_wscale. |
172d69e63
|
212 |
* next 2 bits indicate SACK and ECN support. |
8c7636817
|
213 214 |
* * return false if we decode an option that should not be. |
4dfc28170
|
215 |
*/ |
5d134f1c1
|
216 217 |
bool cookie_check_timestamp(struct tcp_options_received *tcp_opt, struct net *net, bool *ecn_ok) |
4dfc28170
|
218 |
{ |
734f614bc
|
219 |
/* echoed timestamp, lowest bits contain options */ |
4dfc28170
|
220 |
u32 options = tcp_opt->rcv_tsecr & TSMASK; |
8c7636817
|
221 222 223 224 225 226 227 |
if (!tcp_opt->saw_tstamp) { tcp_clear_options(tcp_opt); return true; } if (!sysctl_tcp_timestamps) return false; |
ab56222a3
|
228 |
tcp_opt->sack_ok = (options & (1 << 4)) ? TCP_SACK_SEEN : 0; |
172d69e63
|
229 |
*ecn_ok = (options >> 5) & 1; |
5d134f1c1
|
230 |
if (*ecn_ok && !net->ipv4.sysctl_tcp_ecn) |
172d69e63
|
231 |
return false; |
4dfc28170
|
232 |
|
8c7636817
|
233 234 |
if (tcp_opt->sack_ok && !sysctl_tcp_sack) return false; |
4dfc28170
|
235 |
|
734f614bc
|
236 237 238 239 240 241 |
if ((options & 0xf) == 0xf) return true; /* no window scaling */ tcp_opt->wscale_ok = 1; tcp_opt->snd_wscale = options & 0xf; return sysctl_tcp_window_scaling != 0; |
4dfc28170
|
242 243 |
} EXPORT_SYMBOL(cookie_check_timestamp); |
1da177e4c
|
244 245 246 |
struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb, struct ip_options *opt) { |
4957faade
|
247 |
struct tcp_options_received tcp_opt; |
2e6599cb8
|
248 249 |
struct inet_request_sock *ireq; struct tcp_request_sock *treq; |
1da177e4c
|
250 |
struct tcp_sock *tp = tcp_sk(sk); |
aa8223c7b
|
251 252 |
const struct tcphdr *th = tcp_hdr(skb); __u32 cookie = ntohl(th->ack_seq) - 1; |
1da177e4c
|
253 |
struct sock *ret = sk; |
e905a9eda
|
254 255 256 |
struct request_sock *req; int mss; struct rtable *rt; |
1da177e4c
|
257 |
__u8 rcv_wscale; |
f0e3d0689
|
258 |
bool ecn_ok = false; |
dfd25ffff
|
259 |
struct flowi4 fl4; |
1da177e4c
|
260 |
|
af9b47385
|
261 |
if (!sysctl_tcp_syncookies || !th->ack || th->rst) |
1da177e4c
|
262 |
goto out; |
a0f82f64e
|
263 |
if (tcp_synq_no_recent_overflow(sk) || |
0198230b7
|
264 |
(mss = __cookie_v4_check(ip_hdr(skb), th, cookie)) == 0) { |
de0744af1
|
265 |
NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESFAILED); |
1da177e4c
|
266 267 |
goto out; } |
de0744af1
|
268 |
NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_SYNCOOKIESRECV); |
1da177e4c
|
269 |
|
bb5b7c112
|
270 271 |
/* check for timestamp cookie support */ memset(&tcp_opt, 0, sizeof(tcp_opt)); |
1a2c6181c
|
272 |
tcp_parse_options(skb, &tcp_opt, 0, NULL); |
bb5b7c112
|
273 |
|
5d134f1c1
|
274 |
if (!cookie_check_timestamp(&tcp_opt, sock_net(sk), &ecn_ok)) |
8c7636817
|
275 |
goto out; |
bb5b7c112
|
276 |
|
1da177e4c
|
277 |
ret = NULL; |
ce4a7d0d4
|
278 |
req = inet_reqsk_alloc(&tcp_request_sock_ops); /* for safety */ |
1da177e4c
|
279 280 |
if (!req) goto out; |
2e6599cb8
|
281 282 |
ireq = inet_rsk(req); treq = tcp_rsk(req); |
aa8223c7b
|
283 |
treq->rcv_isn = ntohl(th->seq) - 1; |
e905a9eda
|
284 |
treq->snt_isn = cookie; |
1da177e4c
|
285 |
req->mss = mss; |
b44084c2c
|
286 287 288 289 |
ireq->ir_num = ntohs(th->dest); ireq->ir_rmt_port = th->source; ireq->ir_loc_addr = ip_hdr(skb)->daddr; ireq->ir_rmt_addr = ip_hdr(skb)->saddr; |
84f39b08d
|
290 |
ireq->ir_mark = inet_request_mark(sk, skb); |
172d69e63
|
291 |
ireq->ecn_ok = ecn_ok; |
bb5b7c112
|
292 |
ireq->snd_wscale = tcp_opt.snd_wscale; |
bb5b7c112
|
293 294 295 296 |
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; |
9ad7c049f
|
297 |
treq->snt_synack = tcp_opt.saw_tstamp ? tcp_opt.rcv_tsecr : 0; |
8336886f7
|
298 |
treq->listener = NULL; |
1da177e4c
|
299 300 301 302 303 |
/* 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) { |
f6d8bd051
|
304 |
int opt_size = sizeof(struct ip_options_rcu) + opt->optlen; |
1da177e4c
|
305 |
|
2e6599cb8
|
306 |
ireq->opt = kmalloc(opt_size, GFP_ATOMIC); |
f6d8bd051
|
307 |
if (ireq->opt != NULL && ip_options_echo(&ireq->opt->opt, skb)) { |
2e6599cb8
|
308 309 |
kfree(ireq->opt); ireq->opt = NULL; |
1da177e4c
|
310 311 |
} } |
284904aa7
|
312 313 314 315 |
if (security_inet_conn_request(sk, skb, req)) { reqsk_free(req); goto out; } |
e905a9eda
|
316 |
req->expires = 0UL; |
e6c022a4f
|
317 |
req->num_retrans = 0; |
e905a9eda
|
318 |
|
1da177e4c
|
319 320 321 322 |
/* * 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 |
e905a9eda
|
323 |
* no easy way to do this. |
1da177e4c
|
324 |
*/ |
84f39b08d
|
325 |
flowi4_init_output(&fl4, sk->sk_bound_dev_if, ireq->ir_mark, |
d66954a06
|
326 |
RT_CONN_FLAGS(sk), RT_SCOPE_UNIVERSE, IPPROTO_TCP, |
dfd25ffff
|
327 |
inet_sk_flowi_flags(sk), |
634fb979e
|
328 329 |
(opt && opt->srr) ? opt->faddr : ireq->ir_rmt_addr, ireq->ir_loc_addr, th->source, th->dest); |
dfd25ffff
|
330 331 332 333 334 |
security_req_classify_flow(req, flowi4_to_flowi(&fl4)); rt = ip_route_output_key(sock_net(sk), &fl4); if (IS_ERR(rt)) { reqsk_free(req); goto out; |
1da177e4c
|
335 336 337 |
} /* Try to redo what tcp_v4_send_synack did. */ |
d8d1f30b9
|
338 |
req->window_clamp = tp->window_clamp ? :dst_metric(&rt->dst, RTAX_WINDOW); |
4dfc28170
|
339 |
|
1da177e4c
|
340 |
tcp_select_initial_window(tcp_full_space(sk), req->mss, |
e905a9eda
|
341 |
&req->rcv_wnd, &req->window_clamp, |
31d12926e
|
342 |
ireq->wscale_ok, &rcv_wscale, |
d8d1f30b9
|
343 |
dst_metric(&rt->dst, RTAX_INITRWND)); |
4dfc28170
|
344 |
|
e905a9eda
|
345 |
ireq->rcv_wscale = rcv_wscale; |
1da177e4c
|
346 |
|
d8d1f30b9
|
347 |
ret = get_cookie_sock(sk, skb, req, &rt->dst); |
dfd25ffff
|
348 349 350 351 352 |
/* ip_queue_xmit() depends on our flow being setup * Normal sockets get it right from inet_csk_route_child_sock() */ if (ret) inet_sk(ret)->cork.fl.u.ip4 = fl4; |
1da177e4c
|
353 354 |
out: return ret; } |