include/net/dst.h

   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
  11 #include <linux/netdevice.h>
  12 #include <linux/rtnetlink.h>
  13 #include <linux/rcupdate.h>
  14 #include <linux/jiffies.h>
  15 #include <net/neighbour.h>
  16 #include <asm/processor.h>
  17
  18 /*
  19  * 0 - no debugging messages
  20  * 1 - rare events and bugs (default)
  21  * 2 - trace mode.
  22  */
  23 #define RT_CACHE_DEBUG          0
  24
  25 #define DST_GC_MIN      (HZ/10)
  26 #define DST_GC_INC      (HZ/2)
  27 #define DST_GC_MAX      (120*HZ)
  28
  29 /* Each dst_entry has reference count and sits in some parent list(s).
  30  * When it is removed from parent list, it is "freed" (dst_free).
  31  * After this it enters dead state (dst->obsolete > 0) and if its refcnt
  32  * is zero, it can be destroyed immediately, otherwise it is added
  33  * to gc list and garbage collector periodically checks the refcnt.
  34  */
  35
  36 struct sk_buff;
  37
  38 struct dst_entry
  39 {
  40         struct rcu_head         rcu_head;
  41         struct dst_entry        *child;
  42         struct net_device       *dev;
  43         short                   error;
  44         short                   obsolete;
  45         int                     flags;
  46 #define DST_HOST                1
  47 #define DST_NOXFRM              2
  48 #define DST_NOPOLICY            4
  49 #define DST_NOHASH              8
  50         unsigned long           expires;
  51
  52         unsigned short          header_len;     /* more space at head required */
  53         unsigned short          trailer_len;    /* space to reserve at tail */
  54
  55         unsigned int            rate_tokens;
  56         unsigned long           rate_last;      /* rate limiting for ICMP */
  57
  58         struct dst_entry        *path;
  59
  60         struct neighbour        *neighbour;
  61         struct hh_cache         *hh;
  62         struct xfrm_state       *xfrm;
  63
  64         int                     (*input)(struct sk_buff*);
  65         int                     (*output)(struct sk_buff*);
  66
  67         struct  dst_ops         *ops;
  68
  69         u32                     metrics[RTAX_MAX];
  70
  71 #ifdef CONFIG_NET_CLS_ROUTE
  72         __u32                   tclassid;
  73 #endif
  74
  75         /*
  76          * __refcnt wants to be on a different cache line from
  77          * input/output/ops or performance tanks badly
  78          */
  79         atomic_t                __refcnt;       /* client references    */
  80         int                     __use;
  81         unsigned long           lastuse;
  82         union {
  83                 struct dst_entry *next;
  84                 struct rtable    *rt_next;
  85                 struct rt6_info   *rt6_next;
  86                 struct dn_route  *dn_next;
  87         };
  88 };
  89
  90
  91 struct dst_ops
  92 {
  93         unsigned short          family;
  94         __be16                  protocol;
  95         unsigned                gc_thresh;
  96
  97         int                     (*gc)(struct dst_ops *ops);
  98         struct dst_entry *      (*check)(struct dst_entry *, __u32 cookie);
  99         void                    (*destroy)(struct dst_entry *);
 100         void                    (*ifdown)(struct dst_entry *,
 101                                           struct net_device *dev, int how);
 102         struct dst_entry *      (*negative_advice)(struct dst_entry *);
 103         void                    (*link_failure)(struct sk_buff *);
 104         void                    (*update_pmtu)(struct dst_entry *dst, u32 mtu);
 105         int                     (*local_out)(struct sk_buff *skb);
 106         int                     entry_size;
 107
 108         atomic_t                entries;
 109         struct kmem_cache               *kmem_cachep;
 110         struct net              *dst_net;
 111 };
 112
 113 #ifdef __KERNEL__
 114
 115 static inline u32
 116 dst_metric(const struct dst_entry *dst, int metric)
 117 {
 118         return dst->metrics[metric-1];
 119 }
 120
 121 static inline u32 dst_mtu(const struct dst_entry *dst)
 122 {
 123         u32 mtu = dst_metric(dst, RTAX_MTU);
 124         /*
 125          * Alexey put it here, so ask him about it :)
 126          */
 127         barrier();
 128         return mtu;
 129 }
 130
 131 /* RTT metrics are stored in milliseconds for user ABI, but used as jiffies */
 132 static inline unsigned long dst_metric_rtt(const struct dst_entry *dst, int metric)
 133 {
 134         return msecs_to_jiffies(dst_metric(dst, metric));
 135 }
 136
 137 static inline void set_dst_metric_rtt(struct dst_entry *dst, int metric,
 138                                       unsigned long rtt)
 139 {
 140         dst->metrics[metric-1] = jiffies_to_msecs(rtt);
 141 }
 142
 143 static inline u32
 144 dst_allfrag(const struct dst_entry *dst)
 145 {
 146         int ret = dst_metric(dst, RTAX_FEATURES) & RTAX_FEATURE_ALLFRAG;
 147         /* Yes, _exactly_. This is paranoia. */
 148         barrier();
 149         return ret;
 150 }
 151
 152 static inline int
 153 dst_metric_locked(struct dst_entry *dst, int metric)
 154 {
 155         return dst_metric(dst, RTAX_LOCK) & (1<<metric);
 156 }
 157
 158 static inline void dst_hold(struct dst_entry * dst)
 159 {
 160         atomic_inc(&dst->__refcnt);
 161 }
 162
 163 static inline void dst_use(struct dst_entry *dst, unsigned long time)
 164 {
 165         dst_hold(dst);
 166         dst->__use++;
 167         dst->lastuse = time;
 168 }
 169
 170 static inline
 171 struct dst_entry * dst_clone(struct dst_entry * dst)
 172 {
 173         if (dst)
 174                 atomic_inc(&dst->__refcnt);
 175         return dst;
 176 }
 177
 178 extern void dst_release(struct dst_entry *dst);
 179
 180 /* Children define the path of the packet through the
 181  * Linux networking.  Thus, destinations are stackable.
 182  */
 183
 184 static inline struct dst_entry *dst_pop(struct dst_entry *dst)
 185 {
 186         struct dst_entry *child = dst_clone(dst->child);
 187
 188         dst_release(dst);
 189         return child;
 190 }
 191
 192 extern int dst_discard(struct sk_buff *skb);
 193 extern void * dst_alloc(struct dst_ops * ops);
 194 extern void __dst_free(struct dst_entry * dst);
 195 extern struct dst_entry *dst_destroy(struct dst_entry * dst);
 196
 197 static inline void dst_free(struct dst_entry * dst)
 198 {
 199         if (dst->obsolete > 1)
 200                 return;
 201         if (!atomic_read(&dst->__refcnt)) {
 202                 dst = dst_destroy(dst);
 203                 if (!dst)
 204                         return;
 205         }
 206         __dst_free(dst);
 207 }
 208
 209 static inline void dst_rcu_free(struct rcu_head *head)
 210 {
 211         struct dst_entry *dst = container_of(head, struct dst_entry, rcu_head);
 212         dst_free(dst);
 213 }
 214
 215 static inline void dst_confirm(struct dst_entry *dst)
 216 {
 217         if (dst)
 218                 neigh_confirm(dst->neighbour);
 219 }
 220
 221 static inline void dst_negative_advice(struct dst_entry **dst_p)
 222 {
 223         struct dst_entry * dst = *dst_p;
 224         if (dst && dst->ops->negative_advice)
 225                 *dst_p = dst->ops->negative_advice(dst);
 226 }
 227
 228 static inline void dst_link_failure(struct sk_buff *skb)
 229 {
 230         struct dst_entry * dst = skb->dst;
 231         if (dst && dst->ops && dst->ops->link_failure)
 232                 dst->ops->link_failure(skb);
 233 }
 234
 235 static inline void dst_set_expires(struct dst_entry *dst, int timeout)
 236 {
 237         unsigned long expires = jiffies + timeout;
 238
 239         if (expires == 0)
 240                 expires = 1;
 241
 242         if (dst->expires == 0 || time_before(expires, dst->expires))
 243                 dst->expires = expires;
 244 }
 245
 246 /* Output packet to network from transport.  */
 247 static inline int dst_output(struct sk_buff *skb)
 248 {
 249         return skb->dst->output(skb);
 250 }
 251
 252 /* Input packet from network to transport.  */
 253 static inline int dst_input(struct sk_buff *skb)
 254 {
 255         int err;
 256
 257         for (;;) {
 258                 err = skb->dst->input(skb);
 259
 260                 if (likely(err == 0))
 261                         return err;
 262                 /* Oh, Jamal... Seems, I will not forgive you this mess. :-) */
 263                 if (unlikely(err != NET_XMIT_BYPASS))
 264                         return err;
 265         }
 266 }
 267
 268 static inline struct dst_entry *dst_check(struct dst_entry *dst, u32 cookie)
 269 {
 270         if (dst->obsolete)
 271                 dst = dst->ops->check(dst, cookie);
 272         return dst;
 273 }
 274
 275 extern void             dst_init(void);
 276
 277 /* Flags for xfrm_lookup flags argument. */
 278 enum {
 279         XFRM_LOOKUP_WAIT = 1 << 0,
 280         XFRM_LOOKUP_ICMP = 1 << 1,
 281 };
 282
 283 struct flowi;
 284 #ifndef CONFIG_XFRM
 285 static inline int xfrm_lookup(struct dst_entry **dst_p, struct flowi *fl,
 286                        struct sock *sk, int flags)
 287 {
 288         return 0;
 289 }
 290 static inline int __xfrm_lookup(struct dst_entry **dst_p, struct flowi *fl,
 291                                 struct sock *sk, int flags)
 292 {
 293         return 0;
 294 }
 295 #else
 296 extern int xfrm_lookup(struct dst_entry **dst_p, struct flowi *fl,
 297                        struct sock *sk, int flags);
 298 extern int __xfrm_lookup(struct dst_entry **dst_p, struct flowi *fl,
 299                          struct sock *sk, int flags);
 300 #endif
 301 #endif
 302
 303 #endif /* _NET_DST_H */