* added 0.99 linux version
[mascara-docs.git] / i386 / linux / linux-2.3.21 / net / ipv4 / fib_rules.c
blob5ee1bfd78dbf35be826c37bc3ce22dfc3fd10087
1 /*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
6 * IPv4 Forwarding Information Base: policy rules.
8 * Version: $Id: fib_rules.c,v 1.14 1999/08/31 07:03:29 davem Exp $
10 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
12 * This program is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU General Public License
14 * as published by the Free Software Foundation; either version
15 * 2 of the License, or (at your option) any later version.
17 * Fixes:
18 * Rani Assaf : local_rule cannot be deleted
19 * Marc Boucher : routing by fwmark
22 #include <linux/config.h>
23 #include <asm/uaccess.h>
24 #include <asm/system.h>
25 #include <asm/bitops.h>
26 #include <linux/types.h>
27 #include <linux/kernel.h>
28 #include <linux/sched.h>
29 #include <linux/mm.h>
30 #include <linux/string.h>
31 #include <linux/socket.h>
32 #include <linux/sockios.h>
33 #include <linux/errno.h>
34 #include <linux/in.h>
35 #include <linux/inet.h>
36 #include <linux/netdevice.h>
37 #include <linux/if_arp.h>
38 #include <linux/proc_fs.h>
39 #include <linux/skbuff.h>
40 #include <linux/netlink.h>
41 #include <linux/init.h>
43 #include <net/ip.h>
44 #include <net/protocol.h>
45 #include <net/route.h>
46 #include <net/tcp.h>
47 #include <net/sock.h>
48 #include <net/ip_fib.h>
50 #define FRprintk(a...)
52 struct fib_rule
54 struct fib_rule *r_next;
55 atomic_t r_clntref;
56 u32 r_preference;
57 unsigned char r_table;
58 unsigned char r_action;
59 unsigned char r_dst_len;
60 unsigned char r_src_len;
61 u32 r_src;
62 u32 r_srcmask;
63 u32 r_dst;
64 u32 r_dstmask;
65 u32 r_srcmap;
66 u8 r_flags;
67 u8 r_tos;
68 #ifdef CONFIG_IP_ROUTE_FWMARK
69 u32 r_fwmark;
70 #endif
71 int r_ifindex;
72 #ifdef CONFIG_NET_CLS_ROUTE
73 __u32 r_tclassid;
74 #endif
75 char r_ifname[IFNAMSIZ];
76 int r_dead;
79 static struct fib_rule default_rule = { NULL, ATOMIC_INIT(2), 0x7FFF, RT_TABLE_DEFAULT, RTN_UNICAST, };
80 static struct fib_rule main_rule = { &default_rule, ATOMIC_INIT(2), 0x7FFE, RT_TABLE_MAIN, RTN_UNICAST, };
81 static struct fib_rule local_rule = { &main_rule, ATOMIC_INIT(2), 0, RT_TABLE_LOCAL, RTN_UNICAST, };
83 static struct fib_rule *fib_rules = &local_rule;
84 static rwlock_t fib_rules_lock = RW_LOCK_UNLOCKED;
86 int inet_rtm_delrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
88 struct rtattr **rta = arg;
89 struct rtmsg *rtm = NLMSG_DATA(nlh);
90 struct fib_rule *r, **rp;
91 int err = -ESRCH;
93 for (rp=&fib_rules; (r=*rp) != NULL; rp=&r->r_next) {
94 if ((!rta[RTA_SRC-1] || memcmp(RTA_DATA(rta[RTA_SRC-1]), &r->r_src, 4) == 0) &&
95 rtm->rtm_src_len == r->r_src_len &&
96 rtm->rtm_dst_len == r->r_dst_len &&
97 (!rta[RTA_DST-1] || memcmp(RTA_DATA(rta[RTA_DST-1]), &r->r_dst, 4) == 0) &&
98 rtm->rtm_tos == r->r_tos &&
99 #ifdef CONFIG_IP_ROUTE_FWMARK
100 (!rta[RTA_PROTOINFO-1] || memcmp(RTA_DATA(rta[RTA_PROTOINFO-1]), &r->r_fwmark, 4) == 0) &&
101 #endif
102 (!rtm->rtm_type || rtm->rtm_type == r->r_action) &&
103 (!rta[RTA_PRIORITY-1] || memcmp(RTA_DATA(rta[RTA_PRIORITY-1]), &r->r_preference, 4) == 0) &&
104 (!rta[RTA_IIF-1] || strcmp(RTA_DATA(rta[RTA_IIF-1]), r->r_ifname) == 0) &&
105 (!rtm->rtm_table || (r && rtm->rtm_table == r->r_table))) {
106 err = -EPERM;
107 if (r == &local_rule)
108 break;
110 write_lock_bh(&fib_rules_lock);
111 *rp = r->r_next;
112 r->r_dead = 1;
113 write_unlock_bh(&fib_rules_lock);
114 fib_rule_put(r);
115 err = 0;
116 break;
119 return err;
122 /* Allocate new unique table id */
124 static struct fib_table *fib_empty_table(void)
126 int id;
128 for (id = 1; id <= RT_TABLE_MAX; id++)
129 if (fib_tables[id] == NULL)
130 return __fib_new_table(id);
131 return NULL;
134 void fib_rule_put(struct fib_rule *r)
136 if (atomic_dec_and_test(&r->r_clntref)) {
137 if (r->r_dead)
138 kfree(r);
139 else
140 printk("Freeing alive rule %p\n", r);
144 int inet_rtm_newrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
146 struct rtattr **rta = arg;
147 struct rtmsg *rtm = NLMSG_DATA(nlh);
148 struct fib_rule *r, *new_r, **rp;
149 unsigned char table_id;
151 if (rtm->rtm_src_len > 32 || rtm->rtm_dst_len > 32 ||
152 (rtm->rtm_tos & ~IPTOS_TOS_MASK))
153 return -EINVAL;
155 if (rta[RTA_IIF-1] && RTA_PAYLOAD(rta[RTA_IIF-1]) > IFNAMSIZ)
156 return -EINVAL;
158 table_id = rtm->rtm_table;
159 if (table_id == RT_TABLE_UNSPEC) {
160 struct fib_table *table;
161 if (rtm->rtm_type == RTN_UNICAST || rtm->rtm_type == RTN_NAT) {
162 if ((table = fib_empty_table()) == NULL)
163 return -ENOBUFS;
164 table_id = table->tb_id;
168 new_r = kmalloc(sizeof(*new_r), GFP_KERNEL);
169 if (!new_r)
170 return -ENOMEM;
171 memset(new_r, 0, sizeof(*new_r));
172 if (rta[RTA_SRC-1])
173 memcpy(&new_r->r_src, RTA_DATA(rta[RTA_SRC-1]), 4);
174 if (rta[RTA_DST-1])
175 memcpy(&new_r->r_dst, RTA_DATA(rta[RTA_DST-1]), 4);
176 if (rta[RTA_GATEWAY-1])
177 memcpy(&new_r->r_srcmap, RTA_DATA(rta[RTA_GATEWAY-1]), 4);
178 new_r->r_src_len = rtm->rtm_src_len;
179 new_r->r_dst_len = rtm->rtm_dst_len;
180 new_r->r_srcmask = inet_make_mask(rtm->rtm_src_len);
181 new_r->r_dstmask = inet_make_mask(rtm->rtm_dst_len);
182 new_r->r_tos = rtm->rtm_tos;
183 #ifdef CONFIG_IP_ROUTE_FWMARK
184 if (rta[RTA_PROTOINFO-1])
185 memcpy(&new_r->r_fwmark, RTA_DATA(rta[RTA_PROTOINFO-1]), 4);
186 #endif
187 new_r->r_action = rtm->rtm_type;
188 new_r->r_flags = rtm->rtm_flags;
189 if (rta[RTA_PRIORITY-1])
190 memcpy(&new_r->r_preference, RTA_DATA(rta[RTA_PRIORITY-1]), 4);
191 new_r->r_table = table_id;
192 if (rta[RTA_IIF-1]) {
193 struct net_device *dev;
194 memcpy(new_r->r_ifname, RTA_DATA(rta[RTA_IIF-1]), IFNAMSIZ);
195 new_r->r_ifname[IFNAMSIZ-1] = 0;
196 new_r->r_ifindex = -1;
197 dev = __dev_get_by_name(new_r->r_ifname);
198 if (dev)
199 new_r->r_ifindex = dev->ifindex;
201 #ifdef CONFIG_NET_CLS_ROUTE
202 if (rta[RTA_FLOW-1])
203 memcpy(&new_r->r_tclassid, RTA_DATA(rta[RTA_FLOW-1]), 4);
204 #endif
206 rp = &fib_rules;
207 if (!new_r->r_preference) {
208 r = fib_rules;
209 if (r && (r = r->r_next) != NULL) {
210 rp = &fib_rules->r_next;
211 if (r->r_preference)
212 new_r->r_preference = r->r_preference - 1;
216 while ( (r = *rp) != NULL ) {
217 if (r->r_preference > new_r->r_preference)
218 break;
219 rp = &r->r_next;
222 new_r->r_next = r;
223 atomic_inc(&new_r->r_clntref);
224 write_lock_bh(&fib_rules_lock);
225 *rp = new_r;
226 write_unlock_bh(&fib_rules_lock);
227 return 0;
230 u32 fib_rules_map_destination(u32 daddr, struct fib_result *res)
232 u32 mask = inet_make_mask(res->prefixlen);
233 return (daddr&~mask)|res->fi->fib_nh->nh_gw;
236 u32 fib_rules_policy(u32 saddr, struct fib_result *res, unsigned *flags)
238 struct fib_rule *r = res->r;
240 if (r->r_action == RTN_NAT) {
241 int addrtype = inet_addr_type(r->r_srcmap);
243 if (addrtype == RTN_NAT) {
244 /* Packet is from translated source; remember it */
245 saddr = (saddr&~r->r_srcmask)|r->r_srcmap;
246 *flags |= RTCF_SNAT;
247 } else if (addrtype == RTN_LOCAL || r->r_srcmap == 0) {
248 /* Packet is from masqueraded source; remember it */
249 saddr = r->r_srcmap;
250 *flags |= RTCF_MASQ;
253 return saddr;
256 #ifdef CONFIG_NET_CLS_ROUTE
257 u32 fib_rules_tclass(struct fib_result *res)
259 if (res->r)
260 return res->r->r_tclassid;
261 return 0;
263 #endif
266 static void fib_rules_detach(struct net_device *dev)
268 struct fib_rule *r;
270 for (r=fib_rules; r; r=r->r_next) {
271 if (r->r_ifindex == dev->ifindex) {
272 write_lock_bh(&fib_rules_lock);
273 r->r_ifindex = -1;
274 write_unlock_bh(&fib_rules_lock);
279 static void fib_rules_attach(struct net_device *dev)
281 struct fib_rule *r;
283 for (r=fib_rules; r; r=r->r_next) {
284 if (r->r_ifindex == -1 && strcmp(dev->name, r->r_ifname) == 0) {
285 write_lock_bh(&fib_rules_lock);
286 r->r_ifindex = dev->ifindex;
287 write_unlock_bh(&fib_rules_lock);
292 int fib_lookup(const struct rt_key *key, struct fib_result *res)
294 int err;
295 struct fib_rule *r, *policy;
296 struct fib_table *tb;
298 u32 daddr = key->dst;
299 u32 saddr = key->src;
301 FRprintk("Lookup: %08x <- %08x ", key->dst, key->src);
302 read_lock(&fib_rules_lock);
303 for (r = fib_rules; r; r=r->r_next) {
304 if (((saddr^r->r_src) & r->r_srcmask) ||
305 ((daddr^r->r_dst) & r->r_dstmask) ||
306 #ifdef CONFIG_IP_ROUTE_TOS
307 (r->r_tos && r->r_tos != key->tos) ||
308 #endif
309 #ifdef CONFIG_IP_ROUTE_FWMARK
310 (r->r_fwmark && r->r_fwmark != key->fwmark) ||
311 #endif
312 (r->r_ifindex && r->r_ifindex != key->iif))
313 continue;
315 FRprintk("tb %d r %d ", r->r_table, r->r_action);
316 switch (r->r_action) {
317 case RTN_UNICAST:
318 case RTN_NAT:
319 policy = r;
320 break;
321 case RTN_UNREACHABLE:
322 read_unlock(&fib_rules_lock);
323 return -ENETUNREACH;
324 default:
325 case RTN_BLACKHOLE:
326 read_unlock(&fib_rules_lock);
327 return -EINVAL;
328 case RTN_PROHIBIT:
329 read_unlock(&fib_rules_lock);
330 return -EACCES;
333 if ((tb = fib_get_table(r->r_table)) == NULL)
334 continue;
335 err = tb->tb_lookup(tb, key, res);
336 if (err == 0) {
337 res->r = policy;
338 if (policy)
339 atomic_inc(&policy->r_clntref);
340 read_unlock(&fib_rules_lock);
341 return 0;
343 if (err < 0 && err != -EAGAIN) {
344 read_unlock(&fib_rules_lock);
345 return err;
348 FRprintk("FAILURE\n");
349 read_unlock(&fib_rules_lock);
350 return -ENETUNREACH;
353 void fib_select_default(const struct rt_key *key, struct fib_result *res)
355 if (res->r && res->r->r_action == RTN_UNICAST &&
356 FIB_RES_GW(*res) && FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK) {
357 struct fib_table *tb;
358 if ((tb = fib_get_table(res->r->r_table)) != NULL)
359 tb->tb_select_default(tb, key, res);
363 static int fib_rules_event(struct notifier_block *this, unsigned long event, void *ptr)
365 struct net_device *dev = ptr;
367 if (event == NETDEV_UNREGISTER)
368 fib_rules_detach(dev);
369 else if (event == NETDEV_REGISTER)
370 fib_rules_attach(dev);
371 return NOTIFY_DONE;
375 struct notifier_block fib_rules_notifier = {
376 fib_rules_event,
377 NULL,
381 #ifdef CONFIG_RTNETLINK
383 extern __inline__ int inet_fill_rule(struct sk_buff *skb,
384 struct fib_rule *r,
385 struct netlink_callback *cb)
387 struct rtmsg *rtm;
388 struct nlmsghdr *nlh;
389 unsigned char *b = skb->tail;
391 nlh = NLMSG_PUT(skb, NETLINK_CREDS(cb->skb)->pid, cb->nlh->nlmsg_seq, RTM_NEWRULE, sizeof(*rtm));
392 rtm = NLMSG_DATA(nlh);
393 rtm->rtm_family = AF_INET;
394 rtm->rtm_dst_len = r->r_dst_len;
395 rtm->rtm_src_len = r->r_src_len;
396 rtm->rtm_tos = r->r_tos;
397 #ifdef CONFIG_IP_ROUTE_FWMARK
398 if (r->r_fwmark)
399 RTA_PUT(skb, RTA_PROTOINFO, 4, &r->r_fwmark);
400 #endif
401 rtm->rtm_table = r->r_table;
402 rtm->rtm_protocol = 0;
403 rtm->rtm_scope = 0;
404 rtm->rtm_type = r->r_action;
405 rtm->rtm_flags = r->r_flags;
407 if (r->r_dst_len)
408 RTA_PUT(skb, RTA_DST, 4, &r->r_dst);
409 if (r->r_src_len)
410 RTA_PUT(skb, RTA_SRC, 4, &r->r_src);
411 if (r->r_ifname[0])
412 RTA_PUT(skb, RTA_IIF, IFNAMSIZ, &r->r_ifname);
413 if (r->r_preference)
414 RTA_PUT(skb, RTA_PRIORITY, 4, &r->r_preference);
415 if (r->r_srcmap)
416 RTA_PUT(skb, RTA_GATEWAY, 4, &r->r_srcmap);
417 #ifdef CONFIG_NET_CLS_ROUTE
418 if (r->r_tclassid)
419 RTA_PUT(skb, RTA_FLOW, 4, &r->r_tclassid);
420 #endif
421 nlh->nlmsg_len = skb->tail - b;
422 return skb->len;
424 nlmsg_failure:
425 rtattr_failure:
426 skb_put(skb, b - skb->tail);
427 return -1;
430 int inet_dump_rules(struct sk_buff *skb, struct netlink_callback *cb)
432 int idx;
433 int s_idx = cb->args[0];
434 struct fib_rule *r;
436 read_lock(&fib_rules_lock);
437 for (r=fib_rules, idx=0; r; r = r->r_next, idx++) {
438 if (idx < s_idx)
439 continue;
440 if (inet_fill_rule(skb, r, cb) < 0)
441 break;
443 read_unlock(&fib_rules_lock);
444 cb->args[0] = idx;
446 return skb->len;
449 #endif /* CONFIG_RTNETLINK */
451 void __init fib_rules_init(void)
453 register_netdevice_notifier(&fib_rules_notifier);