1// SPDX-License-Identifier: GPL-2.0-or-later 2/* 3 * IPv6 Syncookies implementation for the Linux kernel 4 * 5 * Authors: 6 * Glenn Griffin <ggriffin.kernel@gmail.com> 7 * 8 * Based on IPv4 implementation by Andi Kleen 9 * linux/net/ipv4/syncookies.c 10 */ 11 12#include <linux/tcp.h> 13#include <linux/random.h> 14#include <linux/siphash.h> 15#include <linux/kernel.h> 16#include <net/secure_seq.h> 17#include <net/ipv6.h> 18#include <net/tcp.h> 19 20#define COOKIEBITS 24 /* Upper bits store count */ 21#define COOKIEMASK (((__u32)1 << COOKIEBITS) - 1) 22 23static siphash_aligned_key_t syncookie6_secret[2]; 24 25/* RFC 2460, Section 8.3: 26 * [ipv6 tcp] MSS must be computed as the maximum packet size minus 60 [..] 27 * 28 * Due to IPV6_MIN_MTU=1280 the lowest possible MSS is 1220, which allows 29 * using higher values than ipv4 tcp syncookies. 30 * The other values are chosen based on ethernet (1500 and 9k MTU), plus 31 * one that accounts for common encap (PPPoe) overhead. Table must be sorted. 32 */ 33static __u16 const msstab[] = { 34 1280 - 60, /* IPV6_MIN_MTU - 60 */ 35 1480 - 60, 36 1500 - 60, 37 9000 - 60, 38}; 39 40static u32 cookie_hash(const struct in6_addr *saddr, 41 const struct in6_addr *daddr, 42 __be16 sport, __be16 dport, u32 count, int c) 43{ 44 const struct { 45 struct in6_addr saddr; 46 struct in6_addr daddr; 47 u32 count; 48 __be16 sport; 49 __be16 dport; 50 } __aligned(SIPHASH_ALIGNMENT) combined = { 51 .saddr = *saddr, 52 .daddr = *daddr, 53 .count = count, 54 .sport = sport, 55 .dport = dport 56 }; 57 58 net_get_random_once(syncookie6_secret, sizeof(syncookie6_secret)); 59 return siphash(&combined, offsetofend(typeof(combined), dport), 60 &syncookie6_secret[c]); 61} 62 63static __u32 secure_tcp_syn_cookie(const struct in6_addr *saddr, 64 const struct in6_addr *daddr, 65 __be16 sport, __be16 dport, __u32 sseq, 66 __u32 data) 67{ 68 u32 count = tcp_cookie_time(); 69 return (cookie_hash(saddr, daddr, sport, dport, 0, 0) + 70 sseq + (count << COOKIEBITS) + 71 ((cookie_hash(saddr, daddr, sport, dport, count, 1) + data) 72 & COOKIEMASK)); 73} 74 75static __u32 check_tcp_syn_cookie(__u32 cookie, const struct in6_addr *saddr, 76 const struct in6_addr *daddr, __be16 sport, 77 __be16 dport, __u32 sseq) 78{ 79 __u32 diff, count = tcp_cookie_time(); 80 81 cookie -= cookie_hash(saddr, daddr, sport, dport, 0, 0) + sseq; 82 83 diff = (count - (cookie >> COOKIEBITS)) & ((__u32) -1 >> COOKIEBITS); 84 if (diff >= MAX_SYNCOOKIE_AGE) 85 return (__u32)-1; 86 87 return (cookie - 88 cookie_hash(saddr, daddr, sport, dport, count - diff, 1)) 89 & COOKIEMASK; 90} 91 92u32 __cookie_v6_init_sequence(const struct ipv6hdr *iph, 93 const struct tcphdr *th, __u16 *mssp) 94{ 95 int mssind; 96 const __u16 mss = *mssp; 97 98 for (mssind = ARRAY_SIZE(msstab) - 1; mssind ; mssind--) 99 if (mss >= msstab[mssind]) 100 break; 101 102 *mssp = msstab[mssind]; 103 104 return secure_tcp_syn_cookie(&iph->saddr, &iph->daddr, th->source, 105 th->dest, ntohl(th->seq), mssind); 106} 107EXPORT_SYMBOL_GPL(__cookie_v6_init_sequence); 108 109__u32 cookie_v6_init_sequence(const struct sk_buff *skb, __u16 *mssp) 110{ 111 const struct ipv6hdr *iph = ipv6_hdr(skb); 112 const struct tcphdr *th = tcp_hdr(skb); 113 114 return __cookie_v6_init_sequence(iph, th, mssp); 115} 116 117int __cookie_v6_check(const struct ipv6hdr *iph, const struct tcphdr *th, 118 __u32 cookie) 119{ 120 __u32 seq = ntohl(th->seq) - 1; 121 __u32 mssind = check_tcp_syn_cookie(cookie, &iph->saddr, &iph->daddr, 122 th->source, th->dest, seq); 123 124 return mssind < ARRAY_SIZE(msstab) ? msstab[mssind] : 0; 125} 126EXPORT_SYMBOL_GPL(__cookie_v6_check); 127 128struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb) 129{ 130 struct tcp_options_received tcp_opt; 131 struct inet_request_sock *ireq; 132 struct tcp_request_sock *treq; 133 struct ipv6_pinfo *np = inet6_sk(sk); 134 struct tcp_sock *tp = tcp_sk(sk); 135 const struct tcphdr *th = tcp_hdr(skb); 136 __u32 cookie = ntohl(th->ack_seq) - 1; 137 struct sock *ret = sk; 138 struct request_sock *req; 139 int full_space, mss; 140 struct dst_entry *dst; 141 __u8 rcv_wscale; 142 u32 tsoff = 0; 143 144 if (!READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_syncookies) || 145 !th->ack || th->rst) 146 goto out; 147 148 if (tcp_synq_no_recent_overflow(sk)) 149 goto out; 150 151 mss = __cookie_v6_check(ipv6_hdr(skb), th, cookie); 152 if (mss == 0) { 153 __NET_INC_STATS(sock_net(sk), LINUX_MIB_SYNCOOKIESFAILED); 154 goto out; 155 } 156 157 __NET_INC_STATS(sock_net(sk), LINUX_MIB_SYNCOOKIESRECV); 158 159 /* check for timestamp cookie support */ 160 memset(&tcp_opt, 0, sizeof(tcp_opt)); 161 tcp_parse_options(sock_net(sk), skb, &tcp_opt, 0, NULL); 162 163 if (tcp_opt.saw_tstamp && tcp_opt.rcv_tsecr) { 164 tsoff = secure_tcpv6_ts_off(sock_net(sk), 165 ipv6_hdr(skb)->daddr.s6_addr32, 166 ipv6_hdr(skb)->saddr.s6_addr32); 167 tcp_opt.rcv_tsecr -= tsoff; 168 } 169 170 if (!cookie_timestamp_decode(sock_net(sk), &tcp_opt)) 171 goto out; 172 173 ret = NULL; 174 req = cookie_tcp_reqsk_alloc(&tcp6_request_sock_ops, 175 &tcp_request_sock_ipv6_ops, sk, skb); 176 if (!req) 177 goto out; 178 179 ireq = inet_rsk(req); 180 treq = tcp_rsk(req); 181 treq->tfo_listener = false; 182 183 req->mss = mss; 184 ireq->ir_rmt_port = th->source; 185 ireq->ir_num = ntohs(th->dest); 186 ireq->ir_v6_rmt_addr = ipv6_hdr(skb)->saddr; 187 ireq->ir_v6_loc_addr = ipv6_hdr(skb)->daddr; 188 189 if (security_inet_conn_request(sk, skb, req)) 190 goto out_free; 191 192 if (ipv6_opt_accepted(sk, skb, &TCP_SKB_CB(skb)->header.h6) || 193 np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo || 194 np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim) { 195 refcount_inc(&skb->users); 196 ireq->pktopts = skb; 197 } 198 199 ireq->ir_iif = inet_request_bound_dev_if(sk, skb); 200 /* So that link locals have meaning */ 201 if (!sk->sk_bound_dev_if && 202 ipv6_addr_type(&ireq->ir_v6_rmt_addr) & IPV6_ADDR_LINKLOCAL) 203 ireq->ir_iif = tcp_v6_iif(skb); 204 205 ireq->ir_mark = inet_request_mark(sk, skb); 206 207 req->num_retrans = 0; 208 ireq->snd_wscale = tcp_opt.snd_wscale; 209 ireq->sack_ok = tcp_opt.sack_ok; 210 ireq->wscale_ok = tcp_opt.wscale_ok; 211 ireq->tstamp_ok = tcp_opt.saw_tstamp; 212 req->ts_recent = tcp_opt.saw_tstamp ? tcp_opt.rcv_tsval : 0; 213 treq->snt_synack = 0; 214 treq->rcv_isn = ntohl(th->seq) - 1; 215 treq->snt_isn = cookie; 216 treq->ts_off = 0; 217 treq->txhash = net_tx_rndhash(); 218 if (IS_ENABLED(CONFIG_SMC)) 219 ireq->smc_ok = 0; 220 221 /* 222 * We need to lookup the dst_entry to get the correct window size. 223 * This is taken from tcp_v6_syn_recv_sock. Somebody please enlighten 224 * me if there is a preferred way. 225 */ 226 { 227 struct in6_addr *final_p, final; 228 struct flowi6 fl6; 229 memset(&fl6, 0, sizeof(fl6)); 230 fl6.flowi6_proto = IPPROTO_TCP; 231 fl6.daddr = ireq->ir_v6_rmt_addr; 232 final_p = fl6_update_dst(&fl6, rcu_dereference(np->opt), &final); 233 fl6.saddr = ireq->ir_v6_loc_addr; 234 fl6.flowi6_oif = ireq->ir_iif; 235 fl6.flowi6_mark = ireq->ir_mark; 236 fl6.fl6_dport = ireq->ir_rmt_port; 237 fl6.fl6_sport = inet_sk(sk)->inet_sport; 238 fl6.flowi6_uid = sk->sk_uid; 239 security_req_classify_flow(req, flowi6_to_flowi_common(&fl6)); 240 241 dst = ip6_dst_lookup_flow(sock_net(sk), sk, &fl6, final_p); 242 if (IS_ERR(dst)) 243 goto out_free; 244 } 245 246 req->rsk_window_clamp = tp->window_clamp ? :dst_metric(dst, RTAX_WINDOW); 247 /* limit the window selection if the user enforce a smaller rx buffer */ 248 full_space = tcp_full_space(sk); 249 if (sk->sk_userlocks & SOCK_RCVBUF_LOCK && 250 (req->rsk_window_clamp > full_space || req->rsk_window_clamp == 0)) 251 req->rsk_window_clamp = full_space; 252 253 tcp_select_initial_window(sk, full_space, req->mss, 254 &req->rsk_rcv_wnd, &req->rsk_window_clamp, 255 ireq->wscale_ok, &rcv_wscale, 256 dst_metric(dst, RTAX_INITRWND)); 257 258 ireq->rcv_wscale = rcv_wscale; 259 ireq->ecn_ok = cookie_ecn_ok(&tcp_opt, sock_net(sk), dst); 260 261 ret = tcp_get_cookie_sock(sk, skb, req, dst, tsoff); 262out: 263 return ret; 264out_free: 265 reqsk_free(req); 266 return NULL; 267} 268