| .. | .. |
|---|
| 1 | +// SPDX-License-Identifier: GPL-2.0-only |
|---|
| 1 | 2 | /* |
|---|
| 2 | 3 | * net/core/dst.c Protocol independent destination cache. |
|---|
| 3 | 4 | * |
|---|
| .. | .. |
|---|
| 26 | 27 | #include <net/dst.h> |
|---|
| 27 | 28 | #include <net/dst_metadata.h> |
|---|
| 28 | 29 | |
|---|
| 29 | | -/* |
|---|
| 30 | | - * Theory of operations: |
|---|
| 31 | | - * 1) We use a list, protected by a spinlock, to add |
|---|
| 32 | | - * new entries from both BH and non-BH context. |
|---|
| 33 | | - * 2) In order to keep spinlock held for a small delay, |
|---|
| 34 | | - * we use a second list where are stored long lived |
|---|
| 35 | | - * entries, that are handled by the garbage collect thread |
|---|
| 36 | | - * fired by a workqueue. |
|---|
| 37 | | - * 3) This list is guarded by a mutex, |
|---|
| 38 | | - * so that the gc_task and dst_dev_event() can be synchronized. |
|---|
| 39 | | - */ |
|---|
| 40 | | - |
|---|
| 41 | | -/* |
|---|
| 42 | | - * We want to keep lock & list close together |
|---|
| 43 | | - * to dirty as few cache lines as possible in __dst_free(). |
|---|
| 44 | | - * As this is not a very strong hint, we dont force an alignment on SMP. |
|---|
| 45 | | - */ |
|---|
| 46 | 30 | int dst_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb) |
|---|
| 47 | 31 | { |
|---|
| 48 | 32 | kfree_skb(skb); |
|---|
| .. | .. |
|---|
| 97 | 81 | { |
|---|
| 98 | 82 | struct dst_entry *dst; |
|---|
| 99 | 83 | |
|---|
| 100 | | - if (ops->gc && dst_entries_get_fast(ops) > ops->gc_thresh) { |
|---|
| 101 | | - if (ops->gc(ops)) |
|---|
| 84 | + if (ops->gc && |
|---|
| 85 | + !(flags & DST_NOCOUNT) && |
|---|
| 86 | + dst_entries_get_fast(ops) > ops->gc_thresh) { |
|---|
| 87 | + if (ops->gc(ops)) { |
|---|
| 88 | + pr_notice_ratelimited("Route cache is full: consider increasing sysctl net.ipv6.route.max_size.\n"); |
|---|
| 102 | 89 | return NULL; |
|---|
| 90 | + } |
|---|
| 103 | 91 | } |
|---|
| 104 | 92 | |
|---|
| 105 | 93 | dst = kmem_cache_alloc(ops->kmem_cachep, GFP_ATOMIC); |
|---|
| .. | .. |
|---|
| 156 | 144 | |
|---|
| 157 | 145 | /* Operations to mark dst as DEAD and clean up the net device referenced |
|---|
| 158 | 146 | * by dst: |
|---|
| 159 | | - * 1. put the dst under loopback interface and discard all tx/rx packets |
|---|
| 147 | + * 1. put the dst under blackhole interface and discard all tx/rx packets |
|---|
| 160 | 148 | * on this route. |
|---|
| 161 | 149 | * 2. release the net_device |
|---|
| 162 | 150 | * This function should be called when removing routes from the fib tree |
|---|
| .. | .. |
|---|
| 172 | 160 | dst->ops->ifdown(dst, dev, true); |
|---|
| 173 | 161 | dst->input = dst_discard; |
|---|
| 174 | 162 | dst->output = dst_discard_out; |
|---|
| 175 | | - dst->dev = dev_net(dst->dev)->loopback_dev; |
|---|
| 163 | + dst->dev = blackhole_netdev; |
|---|
| 176 | 164 | dev_hold(dst->dev); |
|---|
| 177 | 165 | dev_put(dev); |
|---|
| 178 | 166 | } |
|---|
| .. | .. |
|---|
| 184 | 172 | int newrefcnt; |
|---|
| 185 | 173 | |
|---|
| 186 | 174 | newrefcnt = atomic_dec_return(&dst->__refcnt); |
|---|
| 187 | | - if (unlikely(newrefcnt < 0)) |
|---|
| 175 | + if (WARN_ONCE(newrefcnt < 0, "dst_release underflow")) |
|---|
| 188 | 176 | net_warn_ratelimited("%s: dst:%p refcnt:%d\n", |
|---|
| 189 | 177 | __func__, dst, newrefcnt); |
|---|
| 190 | 178 | if (!newrefcnt) |
|---|
| .. | .. |
|---|
| 199 | 187 | int newrefcnt; |
|---|
| 200 | 188 | |
|---|
| 201 | 189 | newrefcnt = atomic_dec_return(&dst->__refcnt); |
|---|
| 202 | | - if (unlikely(newrefcnt < 0)) |
|---|
| 190 | + if (WARN_ONCE(newrefcnt < 0, "dst_release_immediate underflow")) |
|---|
| 203 | 191 | net_warn_ratelimited("%s: dst:%p refcnt:%d\n", |
|---|
| 204 | 192 | __func__, dst, newrefcnt); |
|---|
| 205 | 193 | if (!newrefcnt) |
|---|
| .. | .. |
|---|
| 249 | 237 | } |
|---|
| 250 | 238 | EXPORT_SYMBOL(__dst_destroy_metrics_generic); |
|---|
| 251 | 239 | |
|---|
| 252 | | -static struct dst_ops md_dst_ops = { |
|---|
| 253 | | - .family = AF_UNSPEC, |
|---|
| 240 | +struct dst_entry *dst_blackhole_check(struct dst_entry *dst, u32 cookie) |
|---|
| 241 | +{ |
|---|
| 242 | + return NULL; |
|---|
| 243 | +} |
|---|
| 244 | + |
|---|
| 245 | +u32 *dst_blackhole_cow_metrics(struct dst_entry *dst, unsigned long old) |
|---|
| 246 | +{ |
|---|
| 247 | + return NULL; |
|---|
| 248 | +} |
|---|
| 249 | + |
|---|
| 250 | +struct neighbour *dst_blackhole_neigh_lookup(const struct dst_entry *dst, |
|---|
| 251 | + struct sk_buff *skb, |
|---|
| 252 | + const void *daddr) |
|---|
| 253 | +{ |
|---|
| 254 | + return NULL; |
|---|
| 255 | +} |
|---|
| 256 | + |
|---|
| 257 | +void dst_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk, |
|---|
| 258 | + struct sk_buff *skb, u32 mtu, |
|---|
| 259 | + bool confirm_neigh) |
|---|
| 260 | +{ |
|---|
| 261 | +} |
|---|
| 262 | +EXPORT_SYMBOL_GPL(dst_blackhole_update_pmtu); |
|---|
| 263 | + |
|---|
| 264 | +void dst_blackhole_redirect(struct dst_entry *dst, struct sock *sk, |
|---|
| 265 | + struct sk_buff *skb) |
|---|
| 266 | +{ |
|---|
| 267 | +} |
|---|
| 268 | +EXPORT_SYMBOL_GPL(dst_blackhole_redirect); |
|---|
| 269 | + |
|---|
| 270 | +unsigned int dst_blackhole_mtu(const struct dst_entry *dst) |
|---|
| 271 | +{ |
|---|
| 272 | + unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
|---|
| 273 | + |
|---|
| 274 | + return mtu ? : dst->dev->mtu; |
|---|
| 275 | +} |
|---|
| 276 | +EXPORT_SYMBOL_GPL(dst_blackhole_mtu); |
|---|
| 277 | + |
|---|
| 278 | +static struct dst_ops dst_blackhole_ops = { |
|---|
| 279 | + .family = AF_UNSPEC, |
|---|
| 280 | + .neigh_lookup = dst_blackhole_neigh_lookup, |
|---|
| 281 | + .check = dst_blackhole_check, |
|---|
| 282 | + .cow_metrics = dst_blackhole_cow_metrics, |
|---|
| 283 | + .update_pmtu = dst_blackhole_update_pmtu, |
|---|
| 284 | + .redirect = dst_blackhole_redirect, |
|---|
| 285 | + .mtu = dst_blackhole_mtu, |
|---|
| 254 | 286 | }; |
|---|
| 255 | | - |
|---|
| 256 | | -static int dst_md_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb) |
|---|
| 257 | | -{ |
|---|
| 258 | | - WARN_ONCE(1, "Attempting to call output on metadata dst\n"); |
|---|
| 259 | | - kfree_skb(skb); |
|---|
| 260 | | - return 0; |
|---|
| 261 | | -} |
|---|
| 262 | | - |
|---|
| 263 | | -static int dst_md_discard(struct sk_buff *skb) |
|---|
| 264 | | -{ |
|---|
| 265 | | - WARN_ONCE(1, "Attempting to call input on metadata dst\n"); |
|---|
| 266 | | - kfree_skb(skb); |
|---|
| 267 | | - return 0; |
|---|
| 268 | | -} |
|---|
| 269 | 287 | |
|---|
| 270 | 288 | static void __metadata_dst_init(struct metadata_dst *md_dst, |
|---|
| 271 | 289 | enum metadata_type type, u8 optslen) |
|---|
| 272 | | - |
|---|
| 273 | 290 | { |
|---|
| 274 | 291 | struct dst_entry *dst; |
|---|
| 275 | 292 | |
|---|
| 276 | 293 | dst = &md_dst->dst; |
|---|
| 277 | | - dst_init(dst, &md_dst_ops, NULL, 1, DST_OBSOLETE_NONE, |
|---|
| 294 | + dst_init(dst, &dst_blackhole_ops, NULL, 1, DST_OBSOLETE_NONE, |
|---|
| 278 | 295 | DST_METADATA | DST_NOCOUNT); |
|---|
| 279 | | - |
|---|
| 280 | | - dst->input = dst_md_discard; |
|---|
| 281 | | - dst->output = dst_md_discard_out; |
|---|
| 282 | | - |
|---|
| 283 | 296 | memset(dst + 1, 0, sizeof(*md_dst) + optslen - sizeof(*dst)); |
|---|
| 284 | 297 | md_dst->type = type; |
|---|
| 285 | 298 | } |
|---|