]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - include/net/dst.h
xfrm: Generate blackhole routes only from route lookup functions
[mirror_ubuntu-bionic-kernel.git] / include / net / dst.h
CommitLineData
1da177e4
LT
1/*
2 * net/dst.h Protocol independent destination cache definitions.
3 *
4 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
5 *
6 */
7
8#ifndef _NET_DST_H
9#define _NET_DST_H
10
86393e52 11#include <net/dst_ops.h>
14c85021 12#include <linux/netdevice.h>
1da177e4
LT
13#include <linux/rtnetlink.h>
14#include <linux/rcupdate.h>
187f1882 15#include <linux/bug.h>
1da177e4
LT
16#include <linux/jiffies.h>
17#include <net/neighbour.h>
18#include <asm/processor.h>
19
1da177e4
LT
20#define DST_GC_MIN (HZ/10)
21#define DST_GC_INC (HZ/2)
22#define DST_GC_MAX (120*HZ)
23
24/* Each dst_entry has reference count and sits in some parent list(s).
25 * When it is removed from parent list, it is "freed" (dst_free).
26 * After this it enters dead state (dst->obsolete > 0) and if its refcnt
27 * is zero, it can be destroyed immediately, otherwise it is added
28 * to gc list and garbage collector periodically checks the refcnt.
29 */
30
31struct sk_buff;
32
fd2c3ef7 33struct dst_entry {
1e19e02c 34 struct rcu_head rcu_head;
1da177e4
LT
35 struct dst_entry *child;
36 struct net_device *dev;
62fa8a84
DM
37 struct dst_ops *ops;
38 unsigned long _metrics;
ecd98837 39 unsigned long expires;
f1dd9c37 40 struct dst_entry *path;
ecd98837 41 struct dst_entry *from;
def8b4fa 42#ifdef CONFIG_XFRM
1da177e4 43 struct xfrm_state *xfrm;
5635c10d
ED
44#else
45 void *__pad1;
def8b4fa 46#endif
7f95e188 47 int (*input)(struct sk_buff *);
aad88724 48 int (*output)(struct sock *sk, struct sk_buff *skb);
1da177e4 49
5110effe 50 unsigned short flags;
f6b72b62
DM
51#define DST_HOST 0x0001
52#define DST_NOXFRM 0x0002
53#define DST_NOPOLICY 0x0004
54#define DST_NOHASH 0x0008
55#define DST_NOCACHE 0x0010
56#define DST_NOCOUNT 0x0020
e90c1483
HFS
57#define DST_FAKE_RTABLE 0x0040
58#define DST_XFRM_TUNNEL 0x0080
59#define DST_XFRM_QUEUE 0x0100
f6b72b62 60
5110effe
DM
61 unsigned short pending_confirm;
62
62fa8a84 63 short error;
f5b0a874
DM
64
65 /* A non-zero value of dst->obsolete forces by-hand validation
66 * of the route entry. Positive values are set by the generic
67 * dst layer to indicate that the entry has been forcefully
68 * destroyed.
69 *
70 * Negative values are used by the implementation layer code to
71 * force invocation of the dst_ops->check() method.
72 */
62fa8a84 73 short obsolete;
f5b0a874
DM
74#define DST_OBSOLETE_NONE 0
75#define DST_OBSOLETE_DEAD 2
76#define DST_OBSOLETE_FORCE_CHK -1
ceb33206 77#define DST_OBSOLETE_KILL -2
62fa8a84
DM
78 unsigned short header_len; /* more space at head required */
79 unsigned short trailer_len; /* space to reserve at tail */
c7066f70 80#ifdef CONFIG_IP_ROUTE_CLASSID
f1dd9c37 81 __u32 tclassid;
5635c10d
ED
82#else
83 __u32 __pad2;
f1dd9c37
ZY
84#endif
85
5635c10d
ED
86 /*
87 * Align __refcnt to a 64 bytes alignment
88 * (L1_CACHE_SIZE would be too much)
89 */
90#ifdef CONFIG_64BIT
f6b72b62 91 long __pad_to_align_refcnt[2];
5635c10d 92#endif
f1dd9c37
ZY
93 /*
94 * __refcnt wants to be on a different cache line from
95 * input/output/ops or performance tanks badly
96 */
1e19e02c
ED
97 atomic_t __refcnt; /* client references */
98 int __use;
f1dd9c37 99 unsigned long lastuse;
1e19e02c 100 union {
fc766e4c
ED
101 struct dst_entry *next;
102 struct rtable __rcu *rt_next;
103 struct rt6_info *rt6_next;
104 struct dn_route __rcu *dn_next;
1e19e02c 105 };
1da177e4
LT
106};
107
a4023dd0 108u32 *dst_cow_metrics_generic(struct dst_entry *dst, unsigned long old);
a37e6e34 109extern const u32 dst_default_metrics[];
62fa8a84 110
e5fd387a
MK
111#define DST_METRICS_READ_ONLY 0x1UL
112#define DST_METRICS_FORCE_OVERWRITE 0x2UL
113#define DST_METRICS_FLAGS 0x3UL
62fa8a84 114#define __DST_METRICS_PTR(Y) \
e5fd387a 115 ((u32 *)((Y) & ~DST_METRICS_FLAGS))
62fa8a84
DM
116#define DST_METRICS_PTR(X) __DST_METRICS_PTR((X)->_metrics)
117
118static inline bool dst_metrics_read_only(const struct dst_entry *dst)
119{
120 return dst->_metrics & DST_METRICS_READ_ONLY;
121}
122
e5fd387a
MK
123static inline void dst_metrics_set_force_overwrite(struct dst_entry *dst)
124{
125 dst->_metrics |= DST_METRICS_FORCE_OVERWRITE;
126}
127
a4023dd0 128void __dst_destroy_metrics_generic(struct dst_entry *dst, unsigned long old);
62fa8a84
DM
129
130static inline void dst_destroy_metrics_generic(struct dst_entry *dst)
131{
132 unsigned long val = dst->_metrics;
133 if (!(val & DST_METRICS_READ_ONLY))
134 __dst_destroy_metrics_generic(dst, val);
135}
136
137static inline u32 *dst_metrics_write_ptr(struct dst_entry *dst)
138{
139 unsigned long p = dst->_metrics;
140
1f37070d
SH
141 BUG_ON(!p);
142
62fa8a84
DM
143 if (p & DST_METRICS_READ_ONLY)
144 return dst->ops->cow_metrics(dst, p);
145 return __DST_METRICS_PTR(p);
146}
147
148/* This may only be invoked before the entry has reached global
149 * visibility.
150 */
151static inline void dst_init_metrics(struct dst_entry *dst,
152 const u32 *src_metrics,
153 bool read_only)
154{
155 dst->_metrics = ((unsigned long) src_metrics) |
156 (read_only ? DST_METRICS_READ_ONLY : 0);
157}
158
159static inline void dst_copy_metrics(struct dst_entry *dest, const struct dst_entry *src)
160{
161 u32 *dst_metrics = dst_metrics_write_ptr(dest);
162
163 if (dst_metrics) {
164 u32 *src_metrics = DST_METRICS_PTR(src);
165
166 memcpy(dst_metrics, src_metrics, RTAX_MAX * sizeof(u32));
167 }
168}
169
170static inline u32 *dst_metrics_ptr(struct dst_entry *dst)
171{
172 return DST_METRICS_PTR(dst);
173}
174
1da177e4 175static inline u32
5170ae82 176dst_metric_raw(const struct dst_entry *dst, const int metric)
1da177e4 177{
62fa8a84
DM
178 u32 *p = DST_METRICS_PTR(dst);
179
180 return p[metric-1];
defb3519
DM
181}
182
5170ae82
DM
183static inline u32
184dst_metric(const struct dst_entry *dst, const int metric)
185{
0dbaee3b 186 WARN_ON_ONCE(metric == RTAX_HOPLIMIT ||
d33e4553
DM
187 metric == RTAX_ADVMSS ||
188 metric == RTAX_MTU);
5170ae82
DM
189 return dst_metric_raw(dst, metric);
190}
191
0dbaee3b
DM
192static inline u32
193dst_metric_advmss(const struct dst_entry *dst)
194{
195 u32 advmss = dst_metric_raw(dst, RTAX_ADVMSS);
196
197 if (!advmss)
198 advmss = dst->ops->default_advmss(dst);
199
200 return advmss;
201}
202
defb3519
DM
203static inline void dst_metric_set(struct dst_entry *dst, int metric, u32 val)
204{
62fa8a84 205 u32 *p = dst_metrics_write_ptr(dst);
defb3519 206
62fa8a84
DM
207 if (p)
208 p[metric-1] = val;
1da177e4
LT
209}
210
0c3adfb8
GBY
211static inline u32
212dst_feature(const struct dst_entry *dst, u32 feature)
213{
bb5b7c11 214 return dst_metric(dst, RTAX_FEATURES) & feature;
0c3adfb8
GBY
215}
216
1da177e4
LT
217static inline u32 dst_mtu(const struct dst_entry *dst)
218{
618f9bc7 219 return dst->ops->mtu(dst);
1da177e4
LT
220}
221
c1e20f7c
SH
222/* RTT metrics are stored in milliseconds for user ABI, but used as jiffies */
223static inline unsigned long dst_metric_rtt(const struct dst_entry *dst, int metric)
224{
225 return msecs_to_jiffies(dst_metric(dst, metric));
226}
227
1da177e4
LT
228static inline u32
229dst_allfrag(const struct dst_entry *dst)
230{
0c3adfb8 231 int ret = dst_feature(dst, RTAX_FEATURE_ALLFRAG);
1da177e4
LT
232 return ret;
233}
234
235static inline int
d33e4553 236dst_metric_locked(const struct dst_entry *dst, int metric)
1da177e4
LT
237{
238 return dst_metric(dst, RTAX_LOCK) & (1<<metric);
239}
240
7f95e188 241static inline void dst_hold(struct dst_entry *dst)
1da177e4 242{
5635c10d
ED
243 /*
244 * If your kernel compilation stops here, please check
245 * __pad_to_align_refcnt declaration in struct dst_entry
246 */
247 BUILD_BUG_ON(offsetof(struct dst_entry, __refcnt) & 63);
1da177e4
LT
248 atomic_inc(&dst->__refcnt);
249}
250
03f49f34
PE
251static inline void dst_use(struct dst_entry *dst, unsigned long time)
252{
253 dst_hold(dst);
254 dst->__use++;
255 dst->lastuse = time;
256}
257
7fee226a
ED
258static inline void dst_use_noref(struct dst_entry *dst, unsigned long time)
259{
260 dst->__use++;
261 dst->lastuse = time;
262}
263
7f95e188 264static inline struct dst_entry *dst_clone(struct dst_entry *dst)
1da177e4
LT
265{
266 if (dst)
267 atomic_inc(&dst->__refcnt);
268 return dst;
269}
270
a4023dd0 271void dst_release(struct dst_entry *dst);
7fee226a
ED
272
273static inline void refdst_drop(unsigned long refdst)
274{
275 if (!(refdst & SKB_DST_NOREF))
276 dst_release((struct dst_entry *)(refdst & SKB_DST_PTRMASK));
277}
278
279/**
280 * skb_dst_drop - drops skb dst
281 * @skb: buffer
282 *
283 * Drops dst reference count if a reference was taken.
284 */
adf30907
ED
285static inline void skb_dst_drop(struct sk_buff *skb)
286{
7fee226a
ED
287 if (skb->_skb_refdst) {
288 refdst_drop(skb->_skb_refdst);
289 skb->_skb_refdst = 0UL;
290 }
291}
292
293static inline void skb_dst_copy(struct sk_buff *nskb, const struct sk_buff *oskb)
294{
295 nskb->_skb_refdst = oskb->_skb_refdst;
296 if (!(nskb->_skb_refdst & SKB_DST_NOREF))
297 dst_clone(skb_dst(nskb));
298}
299
300/**
301 * skb_dst_force - makes sure skb dst is refcounted
302 * @skb: buffer
303 *
304 * If dst is not yet refcounted, let's do it
305 */
306static inline void skb_dst_force(struct sk_buff *skb)
307{
308 if (skb_dst_is_noref(skb)) {
309 WARN_ON(!rcu_read_lock_held());
310 skb->_skb_refdst &= ~SKB_DST_NOREF;
311 dst_clone(skb_dst(skb));
312 }
adf30907 313}
1da177e4 314
d19d56dd 315
290b895e
ED
316/**
317 * __skb_tunnel_rx - prepare skb for rx reinsert
318 * @skb: buffer
319 * @dev: tunnel device
ea23192e 320 * @net: netns for packet i/o
290b895e
ED
321 *
322 * After decapsulation, packet is going to re-enter (netif_rx()) our stack,
323 * so make some cleanups. (no accounting done)
324 */
ea23192e
ND
325static inline void __skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev,
326 struct net *net)
290b895e
ED
327{
328 skb->dev = dev;
bdeab991
TH
329
330 /*
7539fadc 331 * Clear hash so that we can recalulate the hash for the
bdeab991
TH
332 * encapsulated packet, unless we have already determine the hash
333 * over the L4 4-tuple.
334 */
7539fadc 335 skb_clear_hash_if_not_l4(skb);
290b895e 336 skb_set_queue_mapping(skb, 0);
ea23192e 337 skb_scrub_packet(skb, !net_eq(net, dev_net(dev)));
290b895e
ED
338}
339
d19d56dd
ED
340/**
341 * skb_tunnel_rx - prepare skb for rx reinsert
342 * @skb: buffer
343 * @dev: tunnel device
344 *
345 * After decapsulation, packet is going to re-enter (netif_rx()) our stack,
346 * so make some cleanups, and perform accounting.
290b895e 347 * Note: this accounting is not SMP safe.
d19d56dd 348 */
ea23192e
ND
349static inline void skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev,
350 struct net *net)
d19d56dd 351{
d19d56dd
ED
352 /* TODO : stats should be SMP safe */
353 dev->stats.rx_packets++;
354 dev->stats.rx_bytes += skb->len;
ea23192e 355 __skb_tunnel_rx(skb, dev, net);
d19d56dd
ED
356}
357
1da177e4
LT
358/* Children define the path of the packet through the
359 * Linux networking. Thus, destinations are stackable.
360 */
361
8764ab2c 362static inline struct dst_entry *skb_dst_pop(struct sk_buff *skb)
1da177e4 363{
e433430a 364 struct dst_entry *child = dst_clone(skb_dst(skb)->child);
1da177e4 365
8764ab2c 366 skb_dst_drop(skb);
1da177e4
LT
367 return child;
368}
369
aad88724
ED
370int dst_discard_sk(struct sock *sk, struct sk_buff *skb);
371static inline int dst_discard(struct sk_buff *skb)
372{
373 return dst_discard_sk(skb->sk, skb);
374}
a4023dd0
JP
375void *dst_alloc(struct dst_ops *ops, struct net_device *dev, int initial_ref,
376 int initial_obsolete, unsigned short flags);
377void __dst_free(struct dst_entry *dst);
378struct dst_entry *dst_destroy(struct dst_entry *dst);
1da177e4 379
7f95e188 380static inline void dst_free(struct dst_entry *dst)
1da177e4 381{
f5b0a874 382 if (dst->obsolete > 0)
1da177e4
LT
383 return;
384 if (!atomic_read(&dst->__refcnt)) {
385 dst = dst_destroy(dst);
386 if (!dst)
387 return;
388 }
389 __dst_free(dst);
390}
391
392static inline void dst_rcu_free(struct rcu_head *head)
393{
394 struct dst_entry *dst = container_of(head, struct dst_entry, rcu_head);
395 dst_free(dst);
396}
397
398static inline void dst_confirm(struct dst_entry *dst)
399{
5110effe
DM
400 dst->pending_confirm = 1;
401}
f2c31e32 402
5110effe
DM
403static inline int dst_neigh_output(struct dst_entry *dst, struct neighbour *n,
404 struct sk_buff *skb)
405{
425f09ab
ED
406 const struct hh_cache *hh;
407
408 if (dst->pending_confirm) {
409 unsigned long now = jiffies;
5110effe 410
5110effe 411 dst->pending_confirm = 0;
425f09ab
ED
412 /* avoid dirtying neighbour */
413 if (n->confirmed != now)
414 n->confirmed = now;
69cce1d1 415 }
5110effe
DM
416
417 hh = &n->hh;
418 if ((n->nud_state & NUD_CONNECTED) && hh->hh_len)
419 return neigh_hh_output(hh, skb);
420 else
421 return n->output(n, skb);
1da177e4
LT
422}
423
d3aaeb38
DM
424static inline struct neighbour *dst_neigh_lookup(const struct dst_entry *dst, const void *daddr)
425{
aaa0c23c
ZZ
426 struct neighbour *n = dst->ops->neigh_lookup(dst, NULL, daddr);
427 return IS_ERR(n) ? NULL : n;
f894cbf8
DM
428}
429
430static inline struct neighbour *dst_neigh_lookup_skb(const struct dst_entry *dst,
431 struct sk_buff *skb)
432{
aaa0c23c
ZZ
433 struct neighbour *n = dst->ops->neigh_lookup(dst, skb, NULL);
434 return IS_ERR(n) ? NULL : n;
d3aaeb38
DM
435}
436
1da177e4
LT
437static inline void dst_link_failure(struct sk_buff *skb)
438{
adf30907 439 struct dst_entry *dst = skb_dst(skb);
1da177e4
LT
440 if (dst && dst->ops && dst->ops->link_failure)
441 dst->ops->link_failure(skb);
442}
443
444static inline void dst_set_expires(struct dst_entry *dst, int timeout)
445{
446 unsigned long expires = jiffies + timeout;
447
448 if (expires == 0)
449 expires = 1;
450
451 if (dst->expires == 0 || time_before(expires, dst->expires))
452 dst->expires = expires;
453}
454
455/* Output packet to network from transport. */
aad88724
ED
456static inline int dst_output_sk(struct sock *sk, struct sk_buff *skb)
457{
458 return skb_dst(skb)->output(sk, skb);
459}
1da177e4
LT
460static inline int dst_output(struct sk_buff *skb)
461{
aad88724 462 return dst_output_sk(skb->sk, skb);
1da177e4
LT
463}
464
465/* Input packet from network to transport. */
466static inline int dst_input(struct sk_buff *skb)
467{
adf30907 468 return skb_dst(skb)->input(skb);
1da177e4
LT
469}
470
471static inline struct dst_entry *dst_check(struct dst_entry *dst, u32 cookie)
472{
473 if (dst->obsolete)
474 dst = dst->ops->check(dst, cookie);
475 return dst;
476}
477
a4023dd0 478void dst_init(void);
1da177e4 479
815f4e57
HX
480/* Flags for xfrm_lookup flags argument. */
481enum {
80c0bc9e 482 XFRM_LOOKUP_ICMP = 1 << 0,
815f4e57
HX
483};
484
1da177e4
LT
485struct flowi;
486#ifndef CONFIG_XFRM
452edd59
DM
487static inline struct dst_entry *xfrm_lookup(struct net *net,
488 struct dst_entry *dst_orig,
489 const struct flowi *fl, struct sock *sk,
490 int flags)
1da177e4 491{
452edd59 492 return dst_orig;
f92ee619
SK
493}
494
495static inline struct dst_entry *xfrm_lookup_route(struct net *net,
496 struct dst_entry *dst_orig,
497 const struct flowi *fl,
498 struct sock *sk,
499 int flags)
500{
501 return dst_orig;
502}
e87b3998
VY
503
504static inline struct xfrm_state *dst_xfrm(const struct dst_entry *dst)
505{
506 return NULL;
507}
508
1da177e4 509#else
a4023dd0
JP
510struct dst_entry *xfrm_lookup(struct net *net, struct dst_entry *dst_orig,
511 const struct flowi *fl, struct sock *sk,
512 int flags);
e87b3998 513
f92ee619
SK
514struct dst_entry *xfrm_lookup_route(struct net *net, struct dst_entry *dst_orig,
515 const struct flowi *fl, struct sock *sk,
516 int flags);
517
e87b3998
VY
518/* skb attached with this dst needs transformation if dst->xfrm is valid */
519static inline struct xfrm_state *dst_xfrm(const struct dst_entry *dst)
520{
521 return dst->xfrm;
522}
1da177e4 523#endif
1da177e4
LT
524
525#endif /* _NET_DST_H */