aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorThomas Graf <tgraf@suug.ch>2006-08-04 06:39:22 -0400
committerDavid S. Miller <davem@sunset.davemloft.net>2006-09-22 17:53:42 -0400
commite1ef4bf23b1ced0bf78a1c98289f746486e5c912 (patch)
tree9085559a5bf39e2dffa33d2ae4548d7b15d27064
parent101367c2f8c464ea96643192673aa18d88e6336d (diff)
[IPV4]: Use Protocol Independant Policy Routing Rules Framework
Signed-off-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
-rw-r--r--include/net/ip_fib.h14
-rw-r--r--net/ipv4/Kconfig1
-rw-r--r--net/ipv4/devinet.c4
-rw-r--r--net/ipv4/fib_frontend.c2
-rw-r--r--net/ipv4/fib_rules.c605
5 files changed, 249 insertions, 377 deletions
diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
index a095d1dec7a4..14c82e611c95 100644
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h
@@ -18,6 +18,7 @@
18 18
19#include <net/flow.h> 19#include <net/flow.h>
20#include <linux/seq_file.h> 20#include <linux/seq_file.h>
21#include <net/fib_rules.h>
21 22
22/* WARNING: The ordering of these elements must match ordering 23/* WARNING: The ordering of these elements must match ordering
23 * of RTA_* rtnetlink attribute numbers. 24 * of RTA_* rtnetlink attribute numbers.
@@ -203,9 +204,8 @@ static inline void fib_select_default(const struct flowi *flp, struct fib_result
203#define ip_fib_main_table (fib_tables[RT_TABLE_MAIN]) 204#define ip_fib_main_table (fib_tables[RT_TABLE_MAIN])
204 205
205extern struct fib_table * fib_tables[RT_TABLE_MAX+1]; 206extern struct fib_table * fib_tables[RT_TABLE_MAX+1];
206extern int fib_lookup(const struct flowi *flp, struct fib_result *res); 207extern int fib_lookup(struct flowi *flp, struct fib_result *res);
207extern struct fib_table *__fib_new_table(int id); 208extern struct fib_table *__fib_new_table(int id);
208extern void fib_rule_put(struct fib_rule *r);
209 209
210static inline struct fib_table *fib_get_table(int id) 210static inline struct fib_table *fib_get_table(int id)
211{ 211{
@@ -251,15 +251,15 @@ extern u32 __fib_res_prefsrc(struct fib_result *res);
251extern struct fib_table *fib_hash_init(int id); 251extern struct fib_table *fib_hash_init(int id);
252 252
253#ifdef CONFIG_IP_MULTIPLE_TABLES 253#ifdef CONFIG_IP_MULTIPLE_TABLES
254/* Exported by fib_rules.c */ 254extern int fib4_rules_dump(struct sk_buff *skb, struct netlink_callback *cb);
255
256extern void __init fib4_rules_init(void);
257extern void __exit fib4_rules_cleanup(void);
255 258
256extern int inet_rtm_delrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg);
257extern int inet_rtm_newrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg);
258extern int inet_dump_rules(struct sk_buff *skb, struct netlink_callback *cb);
259#ifdef CONFIG_NET_CLS_ROUTE 259#ifdef CONFIG_NET_CLS_ROUTE
260extern u32 fib_rules_tclass(struct fib_result *res); 260extern u32 fib_rules_tclass(struct fib_result *res);
261#endif 261#endif
262extern void fib_rules_init(void); 262
263#endif 263#endif
264 264
265static inline void fib_combine_itag(u32 *itag, struct fib_result *res) 265static inline void fib_combine_itag(u32 *itag, struct fib_result *res)
diff --git a/net/ipv4/Kconfig b/net/ipv4/Kconfig
index 3b5d504a74be..1650b64415aa 100644
--- a/net/ipv4/Kconfig
+++ b/net/ipv4/Kconfig
@@ -88,6 +88,7 @@ config IP_FIB_HASH
88config IP_MULTIPLE_TABLES 88config IP_MULTIPLE_TABLES
89 bool "IP: policy routing" 89 bool "IP: policy routing"
90 depends on IP_ADVANCED_ROUTER 90 depends on IP_ADVANCED_ROUTER
91 select FIB_RULES
91 ---help--- 92 ---help---
92 Normally, a router decides what to do with a received packet based 93 Normally, a router decides what to do with a received packet based
93 solely on the packet's final destination address. If you say Y here, 94 solely on the packet's final destination address. If you say Y here,
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index a6cc31d911eb..9f3ffbec3296 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1151,9 +1151,7 @@ static struct rtnetlink_link inet_rtnetlink_table[RTM_NR_MSGTYPES] = {
1151 [RTM_GETROUTE - RTM_BASE] = { .doit = inet_rtm_getroute, 1151 [RTM_GETROUTE - RTM_BASE] = { .doit = inet_rtm_getroute,
1152 .dumpit = inet_dump_fib, }, 1152 .dumpit = inet_dump_fib, },
1153#ifdef CONFIG_IP_MULTIPLE_TABLES 1153#ifdef CONFIG_IP_MULTIPLE_TABLES
1154 [RTM_NEWRULE - RTM_BASE] = { .doit = inet_rtm_newrule, }, 1154 [RTM_GETRULE - RTM_BASE] = { .dumpit = fib4_rules_dump, },
1155 [RTM_DELRULE - RTM_BASE] = { .doit = inet_rtm_delrule, },
1156 [RTM_GETRULE - RTM_BASE] = { .dumpit = inet_dump_rules, },
1157#endif 1155#endif
1158}; 1156};
1159 1157
diff --git a/net/ipv4/fib_frontend.c b/net/ipv4/fib_frontend.c
index ba2a70745a63..fe4a53d4d10d 100644
--- a/net/ipv4/fib_frontend.c
+++ b/net/ipv4/fib_frontend.c
@@ -656,7 +656,7 @@ void __init ip_fib_init(void)
656 ip_fib_local_table = fib_hash_init(RT_TABLE_LOCAL); 656 ip_fib_local_table = fib_hash_init(RT_TABLE_LOCAL);
657 ip_fib_main_table = fib_hash_init(RT_TABLE_MAIN); 657 ip_fib_main_table = fib_hash_init(RT_TABLE_MAIN);
658#else 658#else
659 fib_rules_init(); 659 fib4_rules_init();
660#endif 660#endif
661 661
662 register_netdevice_notifier(&fib_netdev_notifier); 662 register_netdevice_notifier(&fib_netdev_notifier);
diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 79b04718bdfd..23ec6ae1a0f6 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c
@@ -5,9 +5,8 @@
5 * 5 *
6 * IPv4 Forwarding Information Base: policy rules. 6 * IPv4 Forwarding Information Base: policy rules.
7 * 7 *
8 * Version: $Id: fib_rules.c,v 1.17 2001/10/31 21:55:54 davem Exp $
9 *
10 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru> 8 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
9 * Thomas Graf <tgraf@suug.ch>
11 * 10 *
12 * This program is free software; you can redistribute it and/or 11 * This program is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU General Public License 12 * modify it under the terms of the GNU General Public License
@@ -19,129 +18,154 @@
19 * Marc Boucher : routing by fwmark 18 * Marc Boucher : routing by fwmark
20 */ 19 */
21 20
22#include <asm/uaccess.h>
23#include <asm/system.h>
24#include <linux/bitops.h>
25#include <linux/types.h> 21#include <linux/types.h>
26#include <linux/kernel.h> 22#include <linux/kernel.h>
27#include <linux/sched.h>
28#include <linux/mm.h>
29#include <linux/string.h>
30#include <linux/socket.h>
31#include <linux/sockios.h>
32#include <linux/errno.h>
33#include <linux/in.h>
34#include <linux/inet.h>
35#include <linux/inetdevice.h>
36#include <linux/netdevice.h> 23#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> 24#include <linux/netlink.h>
25#include <linux/inetdevice.h>
41#include <linux/init.h> 26#include <linux/init.h>
42#include <linux/list.h> 27#include <linux/list.h>
43#include <linux/rcupdate.h> 28#include <linux/rcupdate.h>
44
45#include <net/ip.h> 29#include <net/ip.h>
46#include <net/protocol.h>
47#include <net/route.h> 30#include <net/route.h>
48#include <net/tcp.h> 31#include <net/tcp.h>
49#include <net/sock.h>
50#include <net/ip_fib.h> 32#include <net/ip_fib.h>
33#include <net/fib_rules.h>
51 34
52#define FRprintk(a...) 35static struct fib_rules_ops fib4_rules_ops;
53 36
54struct fib_rule 37struct fib4_rule
55{ 38{
56 struct hlist_node hlist; 39 struct fib_rule common;
57 atomic_t r_clntref; 40 u8 dst_len;
58 u32 r_preference; 41 u8 src_len;
59 unsigned char r_table; 42 u8 tos;
60 unsigned char r_action; 43 u32 src;
61 unsigned char r_dst_len; 44 u32 srcmask;
62 unsigned char r_src_len; 45 u32 dst;
63 u32 r_src; 46 u32 dstmask;
64 u32 r_srcmask;
65 u32 r_dst;
66 u32 r_dstmask;
67 u32 r_srcmap;
68 u8 r_flags;
69 u8 r_tos;
70#ifdef CONFIG_IP_ROUTE_FWMARK 47#ifdef CONFIG_IP_ROUTE_FWMARK
71 u32 r_fwmark; 48 u32 fwmark;
72#endif 49#endif
73 int r_ifindex;
74#ifdef CONFIG_NET_CLS_ROUTE 50#ifdef CONFIG_NET_CLS_ROUTE
75 __u32 r_tclassid; 51 u32 tclassid;
76#endif 52#endif
77 char r_ifname[IFNAMSIZ];
78 int r_dead;
79 struct rcu_head rcu;
80}; 53};
81 54
82static struct fib_rule default_rule = { 55static struct fib4_rule default_rule = {
83 .r_clntref = ATOMIC_INIT(2), 56 .common = {
84 .r_preference = 0x7FFF, 57 .refcnt = ATOMIC_INIT(2),
85 .r_table = RT_TABLE_DEFAULT, 58 .pref = 0x7FFF,
86 .r_action = RTN_UNICAST, 59 .table = RT_TABLE_DEFAULT,
60 .action = FR_ACT_TO_TBL,
61 },
87}; 62};
88 63
89static struct fib_rule main_rule = { 64static struct fib4_rule main_rule = {
90 .r_clntref = ATOMIC_INIT(2), 65 .common = {
91 .r_preference = 0x7FFE, 66 .refcnt = ATOMIC_INIT(2),
92 .r_table = RT_TABLE_MAIN, 67 .pref = 0x7FFE,
93 .r_action = RTN_UNICAST, 68 .table = RT_TABLE_MAIN,
69 .action = FR_ACT_TO_TBL,
70 },
94}; 71};
95 72
96static struct fib_rule local_rule = { 73static struct fib4_rule local_rule = {
97 .r_clntref = ATOMIC_INIT(2), 74 .common = {
98 .r_table = RT_TABLE_LOCAL, 75 .refcnt = ATOMIC_INIT(2),
99 .r_action = RTN_UNICAST, 76 .table = RT_TABLE_LOCAL,
77 .action = FR_ACT_TO_TBL,
78 .flags = FIB_RULE_PERMANENT,
79 },
100}; 80};
101 81
102static struct hlist_head fib_rules; 82static LIST_HEAD(fib4_rules);
83
84#ifdef CONFIG_NET_CLS_ROUTE
85u32 fib_rules_tclass(struct fib_result *res)
86{
87 return res->r ? ((struct fib4_rule *) res->r)->tclassid : 0;
88}
89#endif
103 90
104/* writer func called from netlink -- rtnl_sem hold*/ 91int fib_lookup(struct flowi *flp, struct fib_result *res)
92{
93 struct fib_lookup_arg arg = {
94 .result = res,
95 };
96 int err;
105 97
106static void rtmsg_rule(int, struct fib_rule *); 98 err = fib_rules_lookup(&fib4_rules_ops, flp, 0, &arg);
99 res->r = arg.rule;
107 100
108int inet_rtm_delrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) 101 return err;
102}
103
104int fib4_rule_action(struct fib_rule *rule, struct flowi *flp, int flags,
105 struct fib_lookup_arg *arg)
109{ 106{
110 struct rtattr **rta = arg; 107 int err = -EAGAIN;
111 struct rtmsg *rtm = NLMSG_DATA(nlh); 108 struct fib_table *tbl;
112 struct fib_rule *r; 109
113 struct hlist_node *node; 110 switch (rule->action) {
114 int err = -ESRCH; 111 case FR_ACT_TO_TBL:
115 112 break;
116 hlist_for_each_entry(r, node, &fib_rules, hlist) { 113
117 if ((!rta[RTA_SRC-1] || memcmp(RTA_DATA(rta[RTA_SRC-1]), &r->r_src, 4) == 0) && 114 case FR_ACT_UNREACHABLE:
118 rtm->rtm_src_len == r->r_src_len && 115 err = -ENETUNREACH;
119 rtm->rtm_dst_len == r->r_dst_len && 116 goto errout;
120 (!rta[RTA_DST-1] || memcmp(RTA_DATA(rta[RTA_DST-1]), &r->r_dst, 4) == 0) && 117
121 rtm->rtm_tos == r->r_tos && 118 case FR_ACT_PROHIBIT:
122#ifdef CONFIG_IP_ROUTE_FWMARK 119 err = -EACCES;
123 (!rta[RTA_PROTOINFO-1] || memcmp(RTA_DATA(rta[RTA_PROTOINFO-1]), &r->r_fwmark, 4) == 0) && 120 goto errout;
124#endif 121
125 (!rtm->rtm_type || rtm->rtm_type == r->r_action) && 122 case FR_ACT_BLACKHOLE:
126 (!rta[RTA_PRIORITY-1] || memcmp(RTA_DATA(rta[RTA_PRIORITY-1]), &r->r_preference, 4) == 0) && 123 default:
127 (!rta[RTA_IIF-1] || rtattr_strcmp(rta[RTA_IIF-1], r->r_ifname) == 0) && 124 err = -EINVAL;
128 (!rtm->rtm_table || (r && rtm->rtm_table == r->r_table))) { 125 goto errout;
129 err = -EPERM;
130 if (r == &local_rule)
131 break;
132
133 hlist_del_rcu(&r->hlist);
134 r->r_dead = 1;
135 rtmsg_rule(RTM_DELRULE, r);
136 fib_rule_put(r);
137 err = 0;
138 break;
139 }
140 } 126 }
127
128 if ((tbl = fib_get_table(rule->table)) == NULL)
129 goto errout;
130
131 err = tbl->tb_lookup(tbl, flp, (struct fib_result *) arg->result);
132 if (err > 0)
133 err = -EAGAIN;
134errout:
141 return err; 135 return err;
142} 136}
143 137
144/* Allocate new unique table id */ 138
139void fib_select_default(const struct flowi *flp, struct fib_result *res)
140{
141 if (res->r && res->r->action == FR_ACT_TO_TBL &&
142 FIB_RES_GW(*res) && FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK) {
143 struct fib_table *tb;
144 if ((tb = fib_get_table(res->r->table)) != NULL)
145 tb->tb_select_default(tb, flp, res);
146 }
147}
148
149static int fib4_rule_match(struct fib_rule *rule, struct flowi *fl, int flags)
150{
151 struct fib4_rule *r = (struct fib4_rule *) rule;
152 u32 daddr = fl->fl4_dst;
153 u32 saddr = fl->fl4_src;
154
155 if (((saddr ^ r->src) & r->srcmask) ||
156 ((daddr ^ r->dst) & r->dstmask))
157 return 0;
158
159 if (r->tos && (r->tos != fl->fl4_tos))
160 return 0;
161
162#ifdef CONFIG_IP_ROUTE_FWMARK
163 if (r->fwmark && (r->fwmark != fl->fl4_fwmark))
164 return 0;
165#endif
166
167 return 1;
168}
145 169
146static struct fib_table *fib_empty_table(void) 170static struct fib_table *fib_empty_table(void)
147{ 171{
@@ -153,329 +177,178 @@ static struct fib_table *fib_empty_table(void)
153 return NULL; 177 return NULL;
154} 178}
155 179
156static inline void fib_rule_put_rcu(struct rcu_head *head) 180static struct nla_policy fib4_rule_policy[FRA_MAX+1] __read_mostly = {
157{ 181 [FRA_IFNAME] = { .type = NLA_STRING },
158 struct fib_rule *r = container_of(head, struct fib_rule, rcu); 182 [FRA_PRIORITY] = { .type = NLA_U32 },
159 kfree(r); 183 [FRA_SRC] = { .type = NLA_U32 },
160} 184 [FRA_DST] = { .type = NLA_U32 },
185 [FRA_FWMARK] = { .type = NLA_U32 },
186 [FRA_FLOW] = { .type = NLA_U32 },
187};
161 188
162void fib_rule_put(struct fib_rule *r) 189static int fib4_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
190 struct nlmsghdr *nlh, struct fib_rule_hdr *frh,
191 struct nlattr **tb)
163{ 192{
164 if (atomic_dec_and_test(&r->r_clntref)) { 193 int err = -EINVAL;
165 if (r->r_dead) 194 struct fib4_rule *rule4 = (struct fib4_rule *) rule;
166 call_rcu(&r->rcu, fib_rule_put_rcu);
167 else
168 printk("Freeing alive rule %p\n", r);
169 }
170}
171 195
172/* writer func called from netlink -- rtnl_sem hold*/ 196 if (frh->src_len > 32 || frh->dst_len > 32 ||
197 (frh->tos & ~IPTOS_TOS_MASK))
198 goto errout;
173 199
174int inet_rtm_newrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) 200 if (rule->table == RT_TABLE_UNSPEC) {
175{ 201 if (rule->action == FR_ACT_TO_TBL) {
176 struct rtattr **rta = arg; 202 struct fib_table *table;
177 struct rtmsg *rtm = NLMSG_DATA(nlh);
178 struct fib_rule *r, *new_r, *last = NULL;
179 struct hlist_node *node = NULL;
180 unsigned char table_id;
181
182 if (rtm->rtm_src_len > 32 || rtm->rtm_dst_len > 32 ||
183 (rtm->rtm_tos & ~IPTOS_TOS_MASK))
184 return -EINVAL;
185
186 if (rta[RTA_IIF-1] && RTA_PAYLOAD(rta[RTA_IIF-1]) > IFNAMSIZ)
187 return -EINVAL;
188
189 table_id = rtm->rtm_table;
190 if (table_id == RT_TABLE_UNSPEC) {
191 struct fib_table *table;
192 if (rtm->rtm_type == RTN_UNICAST) {
193 if ((table = fib_empty_table()) == NULL)
194 return -ENOBUFS;
195 table_id = table->tb_id;
196 }
197 }
198 203
199 new_r = kzalloc(sizeof(*new_r), GFP_KERNEL); 204 table = fib_empty_table();
200 if (!new_r) 205 if (table == NULL) {
201 return -ENOMEM; 206 err = -ENOBUFS;
202 207 goto errout;
203 if (rta[RTA_SRC-1]) 208 }
204 memcpy(&new_r->r_src, RTA_DATA(rta[RTA_SRC-1]), 4);
205 if (rta[RTA_DST-1])
206 memcpy(&new_r->r_dst, RTA_DATA(rta[RTA_DST-1]), 4);
207 if (rta[RTA_GATEWAY-1])
208 memcpy(&new_r->r_srcmap, RTA_DATA(rta[RTA_GATEWAY-1]), 4);
209 new_r->r_src_len = rtm->rtm_src_len;
210 new_r->r_dst_len = rtm->rtm_dst_len;
211 new_r->r_srcmask = inet_make_mask(rtm->rtm_src_len);
212 new_r->r_dstmask = inet_make_mask(rtm->rtm_dst_len);
213 new_r->r_tos = rtm->rtm_tos;
214#ifdef CONFIG_IP_ROUTE_FWMARK
215 if (rta[RTA_PROTOINFO-1])
216 memcpy(&new_r->r_fwmark, RTA_DATA(rta[RTA_PROTOINFO-1]), 4);
217#endif
218 new_r->r_action = rtm->rtm_type;
219 new_r->r_flags = rtm->rtm_flags;
220 if (rta[RTA_PRIORITY-1])
221 memcpy(&new_r->r_preference, RTA_DATA(rta[RTA_PRIORITY-1]), 4);
222 new_r->r_table = table_id;
223 if (rta[RTA_IIF-1]) {
224 struct net_device *dev;
225 rtattr_strlcpy(new_r->r_ifname, rta[RTA_IIF-1], IFNAMSIZ);
226 new_r->r_ifindex = -1;
227 dev = __dev_get_by_name(new_r->r_ifname);
228 if (dev)
229 new_r->r_ifindex = dev->ifindex;
230 }
231#ifdef CONFIG_NET_CLS_ROUTE
232 if (rta[RTA_FLOW-1])
233 memcpy(&new_r->r_tclassid, RTA_DATA(rta[RTA_FLOW-1]), 4);
234#endif
235 r = container_of(fib_rules.first, struct fib_rule, hlist);
236 209
237 if (!new_r->r_preference) { 210 rule->table = table->tb_id;
238 if (r && r->hlist.next != NULL) {
239 r = container_of(r->hlist.next, struct fib_rule, hlist);
240 if (r->r_preference)
241 new_r->r_preference = r->r_preference - 1;
242 } 211 }
243 } 212 }
244 213
245 hlist_for_each_entry(r, node, &fib_rules, hlist) { 214 if (tb[FRA_SRC])
246 if (r->r_preference > new_r->r_preference) 215 rule4->src = nla_get_u32(tb[FRA_SRC]);
247 break;
248 last = r;
249 }
250 atomic_inc(&new_r->r_clntref);
251 216
252 if (last) 217 if (tb[FRA_DST])
253 hlist_add_after_rcu(&last->hlist, &new_r->hlist); 218 rule4->dst = nla_get_u32(tb[FRA_DST]);
254 else
255 hlist_add_before_rcu(&new_r->hlist, &r->hlist);
256 219
257 rtmsg_rule(RTM_NEWRULE, new_r); 220#ifdef CONFIG_IP_ROUTE_FWMARK
258 return 0; 221 if (tb[FRA_FWMARK])
259} 222 rule4->fwmark = nla_get_u32(tb[FRA_FWMARK]);
223#endif
260 224
261#ifdef CONFIG_NET_CLS_ROUTE 225#ifdef CONFIG_NET_CLS_ROUTE
262u32 fib_rules_tclass(struct fib_result *res) 226 if (tb[FRA_FLOW])
263{ 227 rule4->tclassid = nla_get_u32(tb[FRA_FLOW]);
264 if (res->r)
265 return res->r->r_tclassid;
266 return 0;
267}
268#endif 228#endif
269 229
270/* callers should hold rtnl semaphore */ 230 rule4->src_len = frh->src_len;
271 231 rule4->srcmask = inet_make_mask(rule4->src_len);
272static void fib_rules_detach(struct net_device *dev) 232 rule4->dst_len = frh->dst_len;
273{ 233 rule4->dstmask = inet_make_mask(rule4->dst_len);
274 struct hlist_node *node; 234 rule4->tos = frh->tos;
275 struct fib_rule *r;
276
277 hlist_for_each_entry(r, node, &fib_rules, hlist) {
278 if (r->r_ifindex == dev->ifindex)
279 r->r_ifindex = -1;
280 235
281 } 236 err = 0;
282} 237errout:
283 238 return err;
284/* callers should hold rtnl semaphore */
285
286static void fib_rules_attach(struct net_device *dev)
287{
288 struct hlist_node *node;
289 struct fib_rule *r;
290
291 hlist_for_each_entry(r, node, &fib_rules, hlist) {
292 if (r->r_ifindex == -1 && strcmp(dev->name, r->r_ifname) == 0)
293 r->r_ifindex = dev->ifindex;
294 }
295} 239}
296 240
297int fib_lookup(const struct flowi *flp, struct fib_result *res) 241static int fib4_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
242 struct nlattr **tb)
298{ 243{
299 int err; 244 struct fib4_rule *rule4 = (struct fib4_rule *) rule;
300 struct fib_rule *r, *policy;
301 struct fib_table *tb;
302 struct hlist_node *node;
303 245
304 u32 daddr = flp->fl4_dst; 246 if (frh->src_len && (rule4->src_len != frh->src_len))
305 u32 saddr = flp->fl4_src; 247 return 0;
306 248
307FRprintk("Lookup: %u.%u.%u.%u <- %u.%u.%u.%u ", 249 if (frh->dst_len && (rule4->dst_len != frh->dst_len))
308 NIPQUAD(flp->fl4_dst), NIPQUAD(flp->fl4_src)); 250 return 0;
309 251
310 rcu_read_lock(); 252 if (frh->tos && (rule4->tos != frh->tos))
253 return 0;
311 254
312 hlist_for_each_entry_rcu(r, node, &fib_rules, hlist) {
313 if (((saddr^r->r_src) & r->r_srcmask) ||
314 ((daddr^r->r_dst) & r->r_dstmask) ||
315 (r->r_tos && r->r_tos != flp->fl4_tos) ||
316#ifdef CONFIG_IP_ROUTE_FWMARK 255#ifdef CONFIG_IP_ROUTE_FWMARK
317 (r->r_fwmark && r->r_fwmark != flp->fl4_fwmark) || 256 if (tb[FRA_FWMARK] && (rule4->fwmark != nla_get_u32(tb[FRA_FWMARK])))
257 return 0;
318#endif 258#endif
319 (r->r_ifindex && r->r_ifindex != flp->iif))
320 continue;
321
322FRprintk("tb %d r %d ", r->r_table, r->r_action);
323 switch (r->r_action) {
324 case RTN_UNICAST:
325 policy = r;
326 break;
327 case RTN_UNREACHABLE:
328 rcu_read_unlock();
329 return -ENETUNREACH;
330 default:
331 case RTN_BLACKHOLE:
332 rcu_read_unlock();
333 return -EINVAL;
334 case RTN_PROHIBIT:
335 rcu_read_unlock();
336 return -EACCES;
337 }
338 259
339 if ((tb = fib_get_table(r->r_table)) == NULL) 260#ifdef CONFIG_NET_CLS_ROUTE
340 continue; 261 if (tb[FRA_FLOW] && (rule4->tclassid != nla_get_u32(tb[FRA_FLOW])))
341 err = tb->tb_lookup(tb, flp, res); 262 return 0;
342 if (err == 0) { 263#endif
343 res->r = policy;
344 if (policy)
345 atomic_inc(&policy->r_clntref);
346 rcu_read_unlock();
347 return 0;
348 }
349 if (err < 0 && err != -EAGAIN) {
350 rcu_read_unlock();
351 return err;
352 }
353 }
354FRprintk("FAILURE\n");
355 rcu_read_unlock();
356 return -ENETUNREACH;
357}
358 264
359void fib_select_default(const struct flowi *flp, struct fib_result *res) 265 if (tb[FRA_SRC] && (rule4->src != nla_get_u32(tb[FRA_SRC])))
360{ 266 return 0;
361 if (res->r && res->r->r_action == RTN_UNICAST &&
362 FIB_RES_GW(*res) && FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK) {
363 struct fib_table *tb;
364 if ((tb = fib_get_table(res->r->r_table)) != NULL)
365 tb->tb_select_default(tb, flp, res);
366 }
367}
368 267
369static int fib_rules_event(struct notifier_block *this, unsigned long event, void *ptr) 268 if (tb[FRA_DST] && (rule4->dst != nla_get_u32(tb[FRA_DST])))
370{ 269 return 0;
371 struct net_device *dev = ptr;
372 270
373 if (event == NETDEV_UNREGISTER) 271 return 1;
374 fib_rules_detach(dev);
375 else if (event == NETDEV_REGISTER)
376 fib_rules_attach(dev);
377 return NOTIFY_DONE;
378} 272}
379 273
274static int fib4_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
275 struct nlmsghdr *nlh, struct fib_rule_hdr *frh)
276{
277 struct fib4_rule *rule4 = (struct fib4_rule *) rule;
380 278
381static struct notifier_block fib_rules_notifier = { 279 frh->family = AF_INET;
382 .notifier_call =fib_rules_event, 280 frh->dst_len = rule4->dst_len;
383}; 281 frh->src_len = rule4->src_len;
282 frh->tos = rule4->tos;
384 283
385static __inline__ int inet_fill_rule(struct sk_buff *skb,
386 struct fib_rule *r,
387 u32 pid, u32 seq, int event,
388 unsigned int flags)
389{
390 struct rtmsg *rtm;
391 struct nlmsghdr *nlh;
392 unsigned char *b = skb->tail;
393
394 nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*rtm), flags);
395 rtm = NLMSG_DATA(nlh);
396 rtm->rtm_family = AF_INET;
397 rtm->rtm_dst_len = r->r_dst_len;
398 rtm->rtm_src_len = r->r_src_len;
399 rtm->rtm_tos = r->r_tos;
400#ifdef CONFIG_IP_ROUTE_FWMARK 284#ifdef CONFIG_IP_ROUTE_FWMARK
401 if (r->r_fwmark) 285 if (rule4->fwmark)
402 RTA_PUT(skb, RTA_PROTOINFO, 4, &r->r_fwmark); 286 NLA_PUT_U32(skb, FRA_FWMARK, rule4->fwmark);
403#endif 287#endif
404 rtm->rtm_table = r->r_table; 288
405 rtm->rtm_protocol = 0; 289 if (rule4->dst_len)
406 rtm->rtm_scope = 0; 290 NLA_PUT_U32(skb, FRA_DST, rule4->dst);
407 rtm->rtm_type = r->r_action; 291
408 rtm->rtm_flags = r->r_flags; 292 if (rule4->src_len)
409 293 NLA_PUT_U32(skb, FRA_SRC, rule4->src);
410 if (r->r_dst_len) 294
411 RTA_PUT(skb, RTA_DST, 4, &r->r_dst);
412 if (r->r_src_len)
413 RTA_PUT(skb, RTA_SRC, 4, &r->r_src);
414 if (r->r_ifname[0])
415 RTA_PUT(skb, RTA_IIF, IFNAMSIZ, &r->r_ifname);
416 if (r->r_preference)
417 RTA_PUT(skb, RTA_PRIORITY, 4, &r->r_preference);
418 if (r->r_srcmap)
419 RTA_PUT(skb, RTA_GATEWAY, 4, &r->r_srcmap);
420#ifdef CONFIG_NET_CLS_ROUTE 295#ifdef CONFIG_NET_CLS_ROUTE
421 if (r->r_tclassid) 296 if (rule4->tclassid)
422 RTA_PUT(skb, RTA_FLOW, 4, &r->r_tclassid); 297 NLA_PUT_U32(skb, FRA_FLOW, rule4->tclassid);
423#endif 298#endif
424 nlh->nlmsg_len = skb->tail - b; 299 return 0;
425 return skb->len;
426 300
427nlmsg_failure: 301nla_put_failure:
428rtattr_failure: 302 return -ENOBUFS;
429 skb_trim(skb, b - skb->data);
430 return -1;
431} 303}
432 304
433/* callers should hold rtnl semaphore */ 305int fib4_rules_dump(struct sk_buff *skb, struct netlink_callback *cb)
306{
307 return fib_rules_dump(skb, cb, AF_INET);
308}
434 309
435static void rtmsg_rule(int event, struct fib_rule *r) 310static u32 fib4_rule_default_pref(void)
436{ 311{
437 int size = NLMSG_SPACE(sizeof(struct rtmsg) + 128); 312 struct list_head *pos;
438 struct sk_buff *skb = alloc_skb(size, GFP_KERNEL); 313 struct fib_rule *rule;
439 314
440 if (!skb) 315 if (!list_empty(&fib4_rules)) {
441 netlink_set_err(rtnl, 0, RTNLGRP_IPV4_RULE, ENOBUFS); 316 pos = fib4_rules.next;
442 else if (inet_fill_rule(skb, r, 0, 0, event, 0) < 0) { 317 if (pos->next != &fib4_rules) {
443 kfree_skb(skb); 318 rule = list_entry(pos->next, struct fib_rule, list);
444 netlink_set_err(rtnl, 0, RTNLGRP_IPV4_RULE, EINVAL); 319 if (rule->pref)
445 } else { 320 return rule->pref - 1;
446 netlink_broadcast(rtnl, skb, 0, RTNLGRP_IPV4_RULE, GFP_KERNEL); 321 }
447 } 322 }
323
324 return 0;
448} 325}
449 326
450int inet_dump_rules(struct sk_buff *skb, struct netlink_callback *cb) 327static struct fib_rules_ops fib4_rules_ops = {
328 .family = AF_INET,
329 .rule_size = sizeof(struct fib4_rule),
330 .action = fib4_rule_action,
331 .match = fib4_rule_match,
332 .configure = fib4_rule_configure,
333 .compare = fib4_rule_compare,
334 .fill = fib4_rule_fill,
335 .default_pref = fib4_rule_default_pref,
336 .nlgroup = RTNLGRP_IPV4_RULE,
337 .policy = fib4_rule_policy,
338 .rules_list = &fib4_rules,
339 .owner = THIS_MODULE,
340};
341
342void __init fib4_rules_init(void)
451{ 343{
452 int idx = 0; 344 list_add_tail(&local_rule.common.list, &fib4_rules);
453 int s_idx = cb->args[0]; 345 list_add_tail(&main_rule.common.list, &fib4_rules);
454 struct fib_rule *r; 346 list_add_tail(&default_rule.common.list, &fib4_rules);
455 struct hlist_node *node;
456
457 rcu_read_lock();
458 hlist_for_each_entry(r, node, &fib_rules, hlist) {
459 if (idx < s_idx)
460 goto next;
461 if (inet_fill_rule(skb, r, NETLINK_CB(cb->skb).pid,
462 cb->nlh->nlmsg_seq,
463 RTM_NEWRULE, NLM_F_MULTI) < 0)
464 break;
465next:
466 idx++;
467 }
468 rcu_read_unlock();
469 cb->args[0] = idx;
470 347
471 return skb->len; 348 fib_rules_register(&fib4_rules_ops);
472} 349}
473 350
474void __init fib_rules_init(void) 351void __exit fib4_rules_cleanup(void)
475{ 352{
476 INIT_HLIST_HEAD(&fib_rules); 353 fib_rules_unregister(&fib4_rules_ops);
477 hlist_add_head(&local_rule.hlist, &fib_rules);
478 hlist_add_after(&local_rule.hlist, &main_rule.hlist);
479 hlist_add_after(&main_rule.hlist, &default_rule.hlist);
480 register_netdevice_notifier(&fib_rules_notifier);
481} 354}