~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/net/netfilter/xt_TCPMSS.c

Version: ~ [ linux-5.15-rc5 ] ~ [ linux-5.14.11 ] ~ [ linux-5.13.19 ] ~ [ linux-5.12.19 ] ~ [ linux-5.11.22 ] ~ [ linux-5.10.72 ] ~ [ linux-5.9.16 ] ~ [ linux-5.8.18 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.152 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.210 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.250 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.286 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.288 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.18.140 ] ~ [ linux-3.16.85 ] ~ [ linux-3.14.79 ] ~ [ linux-3.12.74 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  1 /*
  2  * This is a module which is used for setting the MSS option in TCP packets.
  3  *
  4  * Copyright (C) 2000 Marc Boucher <marc@mbsi.ca>
  5  * Copyright (C) 2007 Patrick McHardy <kaber@trash.net>
  6  *
  7  * This program is free software; you can redistribute it and/or modify
  8  * it under the terms of the GNU General Public License version 2 as
  9  * published by the Free Software Foundation.
 10  */
 11 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
 12 #include <linux/module.h>
 13 #include <linux/skbuff.h>
 14 #include <linux/ip.h>
 15 #include <linux/gfp.h>
 16 #include <linux/ipv6.h>
 17 #include <linux/tcp.h>
 18 #include <net/dst.h>
 19 #include <net/flow.h>
 20 #include <net/ipv6.h>
 21 #include <net/route.h>
 22 #include <net/tcp.h>
 23 
 24 #include <linux/netfilter_ipv4/ip_tables.h>
 25 #include <linux/netfilter_ipv6/ip6_tables.h>
 26 #include <linux/netfilter/x_tables.h>
 27 #include <linux/netfilter/xt_tcpudp.h>
 28 #include <linux/netfilter/xt_TCPMSS.h>
 29 
 30 MODULE_LICENSE("GPL");
 31 MODULE_AUTHOR("Marc Boucher <marc@mbsi.ca>");
 32 MODULE_DESCRIPTION("Xtables: TCP Maximum Segment Size (MSS) adjustment");
 33 MODULE_ALIAS("ipt_TCPMSS");
 34 MODULE_ALIAS("ip6t_TCPMSS");
 35 
 36 static inline unsigned int
 37 optlen(const u_int8_t *opt, unsigned int offset)
 38 {
 39         /* Beware zero-length options: make finite progress */
 40         if (opt[offset] <= TCPOPT_NOP || opt[offset+1] == 0)
 41                 return 1;
 42         else
 43                 return opt[offset+1];
 44 }
 45 
 46 static u_int32_t tcpmss_reverse_mtu(struct net *net,
 47                                     const struct sk_buff *skb,
 48                                     unsigned int family)
 49 {
 50         struct flowi fl;
 51         const struct nf_afinfo *ai;
 52         struct rtable *rt = NULL;
 53         u_int32_t mtu     = ~0U;
 54 
 55         if (family == PF_INET) {
 56                 struct flowi4 *fl4 = &fl.u.ip4;
 57                 memset(fl4, 0, sizeof(*fl4));
 58                 fl4->daddr = ip_hdr(skb)->saddr;
 59         } else {
 60                 struct flowi6 *fl6 = &fl.u.ip6;
 61 
 62                 memset(fl6, 0, sizeof(*fl6));
 63                 fl6->daddr = ipv6_hdr(skb)->saddr;
 64         }
 65         rcu_read_lock();
 66         ai = nf_get_afinfo(family);
 67         if (ai != NULL)
 68                 ai->route(net, (struct dst_entry **)&rt, &fl, false);
 69         rcu_read_unlock();
 70 
 71         if (rt != NULL) {
 72                 mtu = dst_mtu(&rt->dst);
 73                 dst_release(&rt->dst);
 74         }
 75         return mtu;
 76 }
 77 
 78 static int
 79 tcpmss_mangle_packet(struct sk_buff *skb,
 80                      const struct xt_action_param *par,
 81                      unsigned int family,
 82                      unsigned int tcphoff,
 83                      unsigned int minlen)
 84 {
 85         const struct xt_tcpmss_info *info = par->targinfo;
 86         struct tcphdr *tcph;
 87         int len, tcp_hdrlen;
 88         unsigned int i;
 89         __be16 oldval;
 90         u16 newmss;
 91         u8 *opt;
 92 
 93         /* This is a fragment, no TCP header is available */
 94         if (par->fragoff != 0)
 95                 return 0;
 96 
 97         if (!skb_make_writable(skb, skb->len))
 98                 return -1;
 99 
100         len = skb->len - tcphoff;
101         if (len < (int)sizeof(struct tcphdr))
102                 return -1;
103 
104         tcph = (struct tcphdr *)(skb_network_header(skb) + tcphoff);
105         tcp_hdrlen = tcph->doff * 4;
106 
107         if (len < tcp_hdrlen || tcp_hdrlen < sizeof(struct tcphdr))
108                 return -1;
109 
110         if (info->mss == XT_TCPMSS_CLAMP_PMTU) {
111                 struct net *net = dev_net(par->in ? par->in : par->out);
112                 unsigned int in_mtu = tcpmss_reverse_mtu(net, skb, family);
113 
114                 if (dst_mtu(skb_dst(skb)) <= minlen) {
115                         net_err_ratelimited("unknown or invalid path-MTU (%u)\n",
116                                             dst_mtu(skb_dst(skb)));
117                         return -1;
118                 }
119                 if (in_mtu <= minlen) {
120                         net_err_ratelimited("unknown or invalid path-MTU (%u)\n",
121                                             in_mtu);
122                         return -1;
123                 }
124                 newmss = min(dst_mtu(skb_dst(skb)), in_mtu) - minlen;
125         } else
126                 newmss = info->mss;
127 
128         opt = (u_int8_t *)tcph;
129         for (i = sizeof(struct tcphdr); i <= tcp_hdrlen - TCPOLEN_MSS; i += optlen(opt, i)) {
130                 if (opt[i] == TCPOPT_MSS && opt[i+1] == TCPOLEN_MSS) {
131                         u_int16_t oldmss;
132 
133                         oldmss = (opt[i+2] << 8) | opt[i+3];
134 
135                         /* Never increase MSS, even when setting it, as
136                          * doing so results in problems for hosts that rely
137                          * on MSS being set correctly.
138                          */
139                         if (oldmss <= newmss)
140                                 return 0;
141 
142                         opt[i+2] = (newmss & 0xff00) >> 8;
143                         opt[i+3] = newmss & 0x00ff;
144 
145                         inet_proto_csum_replace2(&tcph->check, skb,
146                                                  htons(oldmss), htons(newmss),
147                                                  0);
148                         return 0;
149                 }
150         }
151 
152         /* There is data after the header so the option can't be added
153          * without moving it, and doing so may make the SYN packet
154          * itself too large. Accept the packet unmodified instead.
155          */
156         if (len > tcp_hdrlen)
157                 return 0;
158 
159         /* tcph->doff has 4 bits, do not wrap it to 0 */
160         if (tcp_hdrlen >= 15 * 4)
161                 return 0;
162 
163         /*
164          * MSS Option not found ?! add it..
165          */
166         if (skb_tailroom(skb) < TCPOLEN_MSS) {
167                 if (pskb_expand_head(skb, 0,
168                                      TCPOLEN_MSS - skb_tailroom(skb),
169                                      GFP_ATOMIC))
170                         return -1;
171                 tcph = (struct tcphdr *)(skb_network_header(skb) + tcphoff);
172         }
173 
174         skb_put(skb, TCPOLEN_MSS);
175 
176         /*
177          * IPv4: RFC 1122 states "If an MSS option is not received at
178          * connection setup, TCP MUST assume a default send MSS of 536".
179          * IPv6: RFC 2460 states IPv6 has a minimum MTU of 1280 and a minimum
180          * length IPv6 header of 60, ergo the default MSS value is 1220
181          * Since no MSS was provided, we must use the default values
182          */
183         if (par->family == NFPROTO_IPV4)
184                 newmss = min(newmss, (u16)536);
185         else
186                 newmss = min(newmss, (u16)1220);
187 
188         opt = (u_int8_t *)tcph + sizeof(struct tcphdr);
189         memmove(opt + TCPOLEN_MSS, opt, len - sizeof(struct tcphdr));
190 
191         inet_proto_csum_replace2(&tcph->check, skb,
192                                  htons(len), htons(len + TCPOLEN_MSS), 1);
193         opt[0] = TCPOPT_MSS;
194         opt[1] = TCPOLEN_MSS;
195         opt[2] = (newmss & 0xff00) >> 8;
196         opt[3] = newmss & 0x00ff;
197 
198         inet_proto_csum_replace4(&tcph->check, skb, 0, *((__be32 *)opt), 0);
199 
200         oldval = ((__be16 *)tcph)[6];
201         tcph->doff += TCPOLEN_MSS/4;
202         inet_proto_csum_replace2(&tcph->check, skb,
203                                  oldval, ((__be16 *)tcph)[6], 0);
204         return TCPOLEN_MSS;
205 }
206 
207 static unsigned int
208 tcpmss_tg4(struct sk_buff *skb, const struct xt_action_param *par)
209 {
210         struct iphdr *iph = ip_hdr(skb);
211         __be16 newlen;
212         int ret;
213 
214         ret = tcpmss_mangle_packet(skb, par,
215                                    PF_INET,
216                                    iph->ihl * 4,
217                                    sizeof(*iph) + sizeof(struct tcphdr));
218         if (ret < 0)
219                 return NF_DROP;
220         if (ret > 0) {
221                 iph = ip_hdr(skb);
222                 newlen = htons(ntohs(iph->tot_len) + ret);
223                 csum_replace2(&iph->check, iph->tot_len, newlen);
224                 iph->tot_len = newlen;
225         }
226         return XT_CONTINUE;
227 }
228 
229 #if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
230 static unsigned int
231 tcpmss_tg6(struct sk_buff *skb, const struct xt_action_param *par)
232 {
233         struct ipv6hdr *ipv6h = ipv6_hdr(skb);
234         u8 nexthdr;
235         __be16 frag_off;
236         int tcphoff;
237         int ret;
238 
239         nexthdr = ipv6h->nexthdr;
240         tcphoff = ipv6_skip_exthdr(skb, sizeof(*ipv6h), &nexthdr, &frag_off);
241         if (tcphoff < 0)
242                 return NF_DROP;
243         ret = tcpmss_mangle_packet(skb, par,
244                                    PF_INET6,
245                                    tcphoff,
246                                    sizeof(*ipv6h) + sizeof(struct tcphdr));
247         if (ret < 0)
248                 return NF_DROP;
249         if (ret > 0) {
250                 ipv6h = ipv6_hdr(skb);
251                 ipv6h->payload_len = htons(ntohs(ipv6h->payload_len) + ret);
252         }
253         return XT_CONTINUE;
254 }
255 #endif
256 
257 /* Must specify -p tcp --syn */
258 static inline bool find_syn_match(const struct xt_entry_match *m)
259 {
260         const struct xt_tcp *tcpinfo = (const struct xt_tcp *)m->data;
261 
262         if (strcmp(m->u.kernel.match->name, "tcp") == 0 &&
263             tcpinfo->flg_cmp & TCPHDR_SYN &&
264             !(tcpinfo->invflags & XT_TCP_INV_FLAGS))
265                 return true;
266 
267         return false;
268 }
269 
270 static int tcpmss_tg4_check(const struct xt_tgchk_param *par)
271 {
272         const struct xt_tcpmss_info *info = par->targinfo;
273         const struct ipt_entry *e = par->entryinfo;
274         const struct xt_entry_match *ematch;
275 
276         if (info->mss == XT_TCPMSS_CLAMP_PMTU &&
277             (par->hook_mask & ~((1 << NF_INET_FORWARD) |
278                            (1 << NF_INET_LOCAL_OUT) |
279                            (1 << NF_INET_POST_ROUTING))) != 0) {
280                 pr_info("path-MTU clamping only supported in "
281                         "FORWARD, OUTPUT and POSTROUTING hooks\n");
282                 return -EINVAL;
283         }
284         xt_ematch_foreach(ematch, e)
285                 if (find_syn_match(ematch))
286                         return 0;
287         pr_info("Only works on TCP SYN packets\n");
288         return -EINVAL;
289 }
290 
291 #if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
292 static int tcpmss_tg6_check(const struct xt_tgchk_param *par)
293 {
294         const struct xt_tcpmss_info *info = par->targinfo;
295         const struct ip6t_entry *e = par->entryinfo;
296         const struct xt_entry_match *ematch;
297 
298         if (info->mss == XT_TCPMSS_CLAMP_PMTU &&
299             (par->hook_mask & ~((1 << NF_INET_FORWARD) |
300                            (1 << NF_INET_LOCAL_OUT) |
301                            (1 << NF_INET_POST_ROUTING))) != 0) {
302                 pr_info("path-MTU clamping only supported in "
303                         "FORWARD, OUTPUT and POSTROUTING hooks\n");
304                 return -EINVAL;
305         }
306         xt_ematch_foreach(ematch, e)
307                 if (find_syn_match(ematch))
308                         return 0;
309         pr_info("Only works on TCP SYN packets\n");
310         return -EINVAL;
311 }
312 #endif
313 
314 static struct xt_target tcpmss_tg_reg[] __read_mostly = {
315         {
316                 .family         = NFPROTO_IPV4,
317                 .name           = "TCPMSS",
318                 .checkentry     = tcpmss_tg4_check,
319                 .target         = tcpmss_tg4,
320                 .targetsize     = sizeof(struct xt_tcpmss_info),
321                 .proto          = IPPROTO_TCP,
322                 .me             = THIS_MODULE,
323         },
324 #if IS_ENABLED(CONFIG_IP6_NF_IPTABLES)
325         {
326                 .family         = NFPROTO_IPV6,
327                 .name           = "TCPMSS",
328                 .checkentry     = tcpmss_tg6_check,
329                 .target         = tcpmss_tg6,
330                 .targetsize     = sizeof(struct xt_tcpmss_info),
331                 .proto          = IPPROTO_TCP,
332                 .me             = THIS_MODULE,
333         },
334 #endif
335 };
336 
337 static int __init tcpmss_tg_init(void)
338 {
339         return xt_register_targets(tcpmss_tg_reg, ARRAY_SIZE(tcpmss_tg_reg));
340 }
341 
342 static void __exit tcpmss_tg_exit(void)
343 {
344         xt_unregister_targets(tcpmss_tg_reg, ARRAY_SIZE(tcpmss_tg_reg));
345 }
346 
347 module_init(tcpmss_tg_init);
348 module_exit(tcpmss_tg_exit);
349 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp