2 * net/dst.h Protocol independent destination cache definitions.
4 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
11 #include <net/dst_ops.h>
12 #include <linux/netdevice.h>
13 #include <linux/rtnetlink.h>
14 #include <linux/rcupdate.h>
15 #include <linux/bug.h>
16 #include <linux/jiffies.h>
17 #include <net/neighbour.h>
18 #include <asm/processor.h>
20 #define DST_GC_MIN (HZ/10)
21 #define DST_GC_INC (HZ/2)
22 #define DST_GC_MAX (120*HZ)
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.
34 struct rcu_head rcu_head
;
35 struct dst_entry
*child
;
36 struct net_device
*dev
;
38 unsigned long _metrics
;
39 unsigned long expires
;
40 struct dst_entry
*path
;
41 struct dst_entry
*from
;
43 struct xfrm_state
*xfrm
;
47 int (*input
)(struct sk_buff
*);
48 int (*output
)(struct sock
*sk
, struct sk_buff
*skb
);
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
57 #define DST_FAKE_RTABLE 0x0040
58 #define DST_XFRM_TUNNEL 0x0080
59 #define DST_XFRM_QUEUE 0x0100
60 #define DST_METADATA 0x0200
62 unsigned short pending_confirm
;
66 /* A non-zero value of dst->obsolete forces by-hand validation
67 * of the route entry. Positive values are set by the generic
68 * dst layer to indicate that the entry has been forcefully
71 * Negative values are used by the implementation layer code to
72 * force invocation of the dst_ops->check() method.
75 #define DST_OBSOLETE_NONE 0
76 #define DST_OBSOLETE_DEAD 2
77 #define DST_OBSOLETE_FORCE_CHK -1
78 #define DST_OBSOLETE_KILL -2
79 unsigned short header_len
; /* more space at head required */
80 unsigned short trailer_len
; /* space to reserve at tail */
81 #ifdef CONFIG_IP_ROUTE_CLASSID
88 * Align __refcnt to a 64 bytes alignment
89 * (L1_CACHE_SIZE would be too much)
92 long __pad_to_align_refcnt
[2];
95 * __refcnt wants to be on a different cache line from
96 * input/output/ops or performance tanks badly
98 atomic_t __refcnt
; /* client references */
100 unsigned long lastuse
;
102 struct dst_entry
*next
;
103 struct rtable __rcu
*rt_next
;
104 struct rt6_info
*rt6_next
;
105 struct dn_route __rcu
*dn_next
;
109 u32
*dst_cow_metrics_generic(struct dst_entry
*dst
, unsigned long old
);
110 extern const u32 dst_default_metrics
[];
112 #define DST_METRICS_READ_ONLY 0x1UL
113 #define DST_METRICS_FLAGS 0x3UL
114 #define __DST_METRICS_PTR(Y) \
115 ((u32 *)((Y) & ~DST_METRICS_FLAGS))
116 #define DST_METRICS_PTR(X) __DST_METRICS_PTR((X)->_metrics)
118 static inline bool dst_metrics_read_only(const struct dst_entry
*dst
)
120 return dst
->_metrics
& DST_METRICS_READ_ONLY
;
123 void __dst_destroy_metrics_generic(struct dst_entry
*dst
, unsigned long old
);
125 static inline void dst_destroy_metrics_generic(struct dst_entry
*dst
)
127 unsigned long val
= dst
->_metrics
;
128 if (!(val
& DST_METRICS_READ_ONLY
))
129 __dst_destroy_metrics_generic(dst
, val
);
132 static inline u32
*dst_metrics_write_ptr(struct dst_entry
*dst
)
134 unsigned long p
= dst
->_metrics
;
138 if (p
& DST_METRICS_READ_ONLY
)
139 return dst
->ops
->cow_metrics(dst
, p
);
140 return __DST_METRICS_PTR(p
);
143 /* This may only be invoked before the entry has reached global
146 static inline void dst_init_metrics(struct dst_entry
*dst
,
147 const u32
*src_metrics
,
150 dst
->_metrics
= ((unsigned long) src_metrics
) |
151 (read_only
? DST_METRICS_READ_ONLY
: 0);
154 static inline void dst_copy_metrics(struct dst_entry
*dest
, const struct dst_entry
*src
)
156 u32
*dst_metrics
= dst_metrics_write_ptr(dest
);
159 u32
*src_metrics
= DST_METRICS_PTR(src
);
161 memcpy(dst_metrics
, src_metrics
, RTAX_MAX
* sizeof(u32
));
165 static inline u32
*dst_metrics_ptr(struct dst_entry
*dst
)
167 return DST_METRICS_PTR(dst
);
171 dst_metric_raw(const struct dst_entry
*dst
, const int metric
)
173 u32
*p
= DST_METRICS_PTR(dst
);
179 dst_metric(const struct dst_entry
*dst
, const int metric
)
181 WARN_ON_ONCE(metric
== RTAX_HOPLIMIT
||
182 metric
== RTAX_ADVMSS
||
184 return dst_metric_raw(dst
, metric
);
188 dst_metric_advmss(const struct dst_entry
*dst
)
190 u32 advmss
= dst_metric_raw(dst
, RTAX_ADVMSS
);
193 advmss
= dst
->ops
->default_advmss(dst
);
198 static inline void dst_metric_set(struct dst_entry
*dst
, int metric
, u32 val
)
200 u32
*p
= dst_metrics_write_ptr(dst
);
207 dst_feature(const struct dst_entry
*dst
, u32 feature
)
209 return dst_metric(dst
, RTAX_FEATURES
) & feature
;
212 static inline u32
dst_mtu(const struct dst_entry
*dst
)
214 return dst
->ops
->mtu(dst
);
217 /* RTT metrics are stored in milliseconds for user ABI, but used as jiffies */
218 static inline unsigned long dst_metric_rtt(const struct dst_entry
*dst
, int metric
)
220 return msecs_to_jiffies(dst_metric(dst
, metric
));
224 dst_allfrag(const struct dst_entry
*dst
)
226 int ret
= dst_feature(dst
, RTAX_FEATURE_ALLFRAG
);
231 dst_metric_locked(const struct dst_entry
*dst
, int metric
)
233 return dst_metric(dst
, RTAX_LOCK
) & (1<<metric
);
236 static inline void dst_hold(struct dst_entry
*dst
)
239 * If your kernel compilation stops here, please check
240 * __pad_to_align_refcnt declaration in struct dst_entry
242 BUILD_BUG_ON(offsetof(struct dst_entry
, __refcnt
) & 63);
243 atomic_inc(&dst
->__refcnt
);
246 static inline void dst_use(struct dst_entry
*dst
, unsigned long time
)
253 static inline void dst_use_noref(struct dst_entry
*dst
, unsigned long time
)
259 static inline struct dst_entry
*dst_clone(struct dst_entry
*dst
)
262 atomic_inc(&dst
->__refcnt
);
266 void dst_release(struct dst_entry
*dst
);
268 static inline void refdst_drop(unsigned long refdst
)
270 if (!(refdst
& SKB_DST_NOREF
))
271 dst_release((struct dst_entry
*)(refdst
& SKB_DST_PTRMASK
));
275 * skb_dst_drop - drops skb dst
278 * Drops dst reference count if a reference was taken.
280 static inline void skb_dst_drop(struct sk_buff
*skb
)
282 if (skb
->_skb_refdst
) {
283 refdst_drop(skb
->_skb_refdst
);
284 skb
->_skb_refdst
= 0UL;
288 static inline void skb_dst_copy(struct sk_buff
*nskb
, const struct sk_buff
*oskb
)
290 nskb
->_skb_refdst
= oskb
->_skb_refdst
;
291 if (!(nskb
->_skb_refdst
& SKB_DST_NOREF
))
292 dst_clone(skb_dst(nskb
));
296 * skb_dst_force - makes sure skb dst is refcounted
299 * If dst is not yet refcounted, let's do it
301 static inline void skb_dst_force(struct sk_buff
*skb
)
303 if (skb_dst_is_noref(skb
)) {
304 WARN_ON(!rcu_read_lock_held());
305 skb
->_skb_refdst
&= ~SKB_DST_NOREF
;
306 dst_clone(skb_dst(skb
));
312 * __skb_tunnel_rx - prepare skb for rx reinsert
314 * @dev: tunnel device
315 * @net: netns for packet i/o
317 * After decapsulation, packet is going to re-enter (netif_rx()) our stack,
318 * so make some cleanups. (no accounting done)
320 static inline void __skb_tunnel_rx(struct sk_buff
*skb
, struct net_device
*dev
,
326 * Clear hash so that we can recalulate the hash for the
327 * encapsulated packet, unless we have already determine the hash
328 * over the L4 4-tuple.
330 skb_clear_hash_if_not_l4(skb
);
331 skb_set_queue_mapping(skb
, 0);
332 skb_scrub_packet(skb
, !net_eq(net
, dev_net(dev
)));
336 * skb_tunnel_rx - prepare skb for rx reinsert
338 * @dev: tunnel device
340 * After decapsulation, packet is going to re-enter (netif_rx()) our stack,
341 * so make some cleanups, and perform accounting.
342 * Note: this accounting is not SMP safe.
344 static inline void skb_tunnel_rx(struct sk_buff
*skb
, struct net_device
*dev
,
347 /* TODO : stats should be SMP safe */
348 dev
->stats
.rx_packets
++;
349 dev
->stats
.rx_bytes
+= skb
->len
;
350 __skb_tunnel_rx(skb
, dev
, net
);
353 int dst_discard_sk(struct sock
*sk
, struct sk_buff
*skb
);
354 static inline int dst_discard(struct sk_buff
*skb
)
356 return dst_discard_sk(skb
->sk
, skb
);
358 void *dst_alloc(struct dst_ops
*ops
, struct net_device
*dev
, int initial_ref
,
359 int initial_obsolete
, unsigned short flags
);
360 void dst_init(struct dst_entry
*dst
, struct dst_ops
*ops
,
361 struct net_device
*dev
, int initial_ref
, int initial_obsolete
,
362 unsigned short flags
);
363 void __dst_free(struct dst_entry
*dst
);
364 struct dst_entry
*dst_destroy(struct dst_entry
*dst
);
366 static inline void dst_free(struct dst_entry
*dst
)
368 if (dst
->obsolete
> 0)
370 if (!atomic_read(&dst
->__refcnt
)) {
371 dst
= dst_destroy(dst
);
378 static inline void dst_rcu_free(struct rcu_head
*head
)
380 struct dst_entry
*dst
= container_of(head
, struct dst_entry
, rcu_head
);
384 static inline void dst_confirm(struct dst_entry
*dst
)
386 dst
->pending_confirm
= 1;
389 static inline int dst_neigh_output(struct dst_entry
*dst
, struct neighbour
*n
,
392 const struct hh_cache
*hh
;
394 if (dst
->pending_confirm
) {
395 unsigned long now
= jiffies
;
397 dst
->pending_confirm
= 0;
398 /* avoid dirtying neighbour */
399 if (n
->confirmed
!= now
)
404 if ((n
->nud_state
& NUD_CONNECTED
) && hh
->hh_len
)
405 return neigh_hh_output(hh
, skb
);
407 return n
->output(n
, skb
);
410 static inline struct neighbour
*dst_neigh_lookup(const struct dst_entry
*dst
, const void *daddr
)
412 struct neighbour
*n
= dst
->ops
->neigh_lookup(dst
, NULL
, daddr
);
413 return IS_ERR(n
) ? NULL
: n
;
416 static inline struct neighbour
*dst_neigh_lookup_skb(const struct dst_entry
*dst
,
419 struct neighbour
*n
= dst
->ops
->neigh_lookup(dst
, skb
, NULL
);
420 return IS_ERR(n
) ? NULL
: n
;
423 static inline void dst_link_failure(struct sk_buff
*skb
)
425 struct dst_entry
*dst
= skb_dst(skb
);
426 if (dst
&& dst
->ops
&& dst
->ops
->link_failure
)
427 dst
->ops
->link_failure(skb
);
430 static inline void dst_set_expires(struct dst_entry
*dst
, int timeout
)
432 unsigned long expires
= jiffies
+ timeout
;
437 if (dst
->expires
== 0 || time_before(expires
, dst
->expires
))
438 dst
->expires
= expires
;
441 /* Output packet to network from transport. */
442 static inline int dst_output_sk(struct sock
*sk
, struct sk_buff
*skb
)
444 return skb_dst(skb
)->output(sk
, skb
);
446 static inline int dst_output(struct sk_buff
*skb
)
448 return dst_output_sk(skb
->sk
, skb
);
451 /* Input packet from network to transport. */
452 static inline int dst_input(struct sk_buff
*skb
)
454 return skb_dst(skb
)->input(skb
);
457 static inline struct dst_entry
*dst_check(struct dst_entry
*dst
, u32 cookie
)
460 dst
= dst
->ops
->check(dst
, cookie
);
464 void dst_subsys_init(void);
466 /* Flags for xfrm_lookup flags argument. */
468 XFRM_LOOKUP_ICMP
= 1 << 0,
469 XFRM_LOOKUP_QUEUE
= 1 << 1,
470 XFRM_LOOKUP_KEEP_DST_REF
= 1 << 2,
475 static inline struct dst_entry
*xfrm_lookup(struct net
*net
,
476 struct dst_entry
*dst_orig
,
477 const struct flowi
*fl
, struct sock
*sk
,
483 static inline struct dst_entry
*xfrm_lookup_route(struct net
*net
,
484 struct dst_entry
*dst_orig
,
485 const struct flowi
*fl
,
492 static inline struct xfrm_state
*dst_xfrm(const struct dst_entry
*dst
)
498 struct dst_entry
*xfrm_lookup(struct net
*net
, struct dst_entry
*dst_orig
,
499 const struct flowi
*fl
, struct sock
*sk
,
502 struct dst_entry
*xfrm_lookup_route(struct net
*net
, struct dst_entry
*dst_orig
,
503 const struct flowi
*fl
, struct sock
*sk
,
506 /* skb attached with this dst needs transformation if dst->xfrm is valid */
507 static inline struct xfrm_state
*dst_xfrm(const struct dst_entry
*dst
)
513 #endif /* _NET_DST_H */
This page took 0.053266 seconds and 6 git commands to generate.