net/openvswitch/flow_table.c

   1 /*
   2  * Copyright (c) 2007-2013 Nicira, Inc.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of version 2 of the GNU General Public
   6  * License as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful, but
   9  * WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License
  14  * along with this program; if not, write to the Free Software
  15  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
  16  * 02110-1301, USA
  17  */
  18
  19 #include "flow.h"
  20 #include "datapath.h"
  21 #include <linux/uaccess.h>
  22 #include <linux/netdevice.h>
  23 #include <linux/etherdevice.h>
  24 #include <linux/if_ether.h>
  25 #include <linux/if_vlan.h>
  26 #include <net/llc_pdu.h>
  27 #include <linux/kernel.h>
  28 #include <linux/jhash.h>
  29 #include <linux/jiffies.h>
  30 #include <linux/llc.h>
  31 #include <linux/module.h>
  32 #include <linux/in.h>
  33 #include <linux/rcupdate.h>
  34 #include <linux/if_arp.h>
  35 #include <linux/ip.h>
  36 #include <linux/ipv6.h>
  37 #include <linux/sctp.h>
  38 #include <linux/tcp.h>
  39 #include <linux/udp.h>
  40 #include <linux/icmp.h>
  41 #include <linux/icmpv6.h>
  42 #include <linux/rculist.h>
  43 #include <net/ip.h>
  44 #include <net/ipv6.h>
  45 #include <net/ndisc.h>
  46
  47 #include "datapath.h"
  48
  49 #define TBL_MIN_BUCKETS         1024
  50 #define REHASH_INTERVAL         (10 * 60 * HZ)
  51
  52 static struct kmem_cache *flow_cache;
  53
  54 static u16 range_n_bytes(const struct sw_flow_key_range *range)
  55 {
  56         return range->end - range->start;
  57 }
  58
  59 void ovs_flow_mask_key(struct sw_flow_key *dst, const struct sw_flow_key *src,
  60                        const struct sw_flow_mask *mask)
  61 {
  62         const long *m = (long *)((u8 *)&mask->key + mask->range.start);
  63         const long *s = (long *)((u8 *)src + mask->range.start);
  64         long *d = (long *)((u8 *)dst + mask->range.start);
  65         int i;
  66
  67         /* The memory outside of the 'mask->range' are not set since
  68          * further operations on 'dst' only uses contents within
  69          * 'mask->range'.
  70          */
  71         for (i = 0; i < range_n_bytes(&mask->range); i += sizeof(long))
  72                 *d++ = *s++ & *m++;
  73 }
  74
  75 struct sw_flow *ovs_flow_alloc(void)
  76 {
  77         struct sw_flow *flow;
  78
  79         flow = kmem_cache_alloc(flow_cache, GFP_KERNEL);
  80         if (!flow)
  81                 return ERR_PTR(-ENOMEM);
  82
  83         spin_lock_init(&flow->lock);
  84         flow->sf_acts = NULL;
  85         flow->mask = NULL;
  86
  87         return flow;
  88 }
  89
  90 int ovs_flow_tbl_count(struct flow_table *table)
  91 {
  92         return table->count;
  93 }
  94
  95 static struct flex_array *alloc_buckets(unsigned int n_buckets)
  96 {
  97         struct flex_array *buckets;
  98         int i, err;
  99
 100         buckets = flex_array_alloc(sizeof(struct hlist_head),
 101                                    n_buckets, GFP_KERNEL);
 102         if (!buckets)
 103                 return NULL;
 104
 105         err = flex_array_prealloc(buckets, 0, n_buckets, GFP_KERNEL);
 106         if (err) {
 107                 flex_array_free(buckets);
 108                 return NULL;
 109         }
 110
 111         for (i = 0; i < n_buckets; i++)
 112                 INIT_HLIST_HEAD((struct hlist_head *)
 113                                         flex_array_get(buckets, i));
 114
 115         return buckets;
 116 }
 117
 118 static void flow_free(struct sw_flow *flow)
 119 {
 120         kfree((struct sf_flow_acts __force *)flow->sf_acts);
 121         kmem_cache_free(flow_cache, flow);
 122 }
 123
 124 static void rcu_free_flow_callback(struct rcu_head *rcu)
 125 {
 126         struct sw_flow *flow = container_of(rcu, struct sw_flow, rcu);
 127
 128         flow_free(flow);
 129 }
 130
 131 static void rcu_free_sw_flow_mask_cb(struct rcu_head *rcu)
 132 {
 133         struct sw_flow_mask *mask = container_of(rcu, struct sw_flow_mask, rcu);
 134
 135         kfree(mask);
 136 }
 137
 138 static void flow_mask_del_ref(struct sw_flow_mask *mask, bool deferred)
 139 {
 140         if (!mask)
 141                 return;
 142
 143         BUG_ON(!mask->ref_count);
 144         mask->ref_count--;
 145
 146         if (!mask->ref_count) {
 147                 list_del_rcu(&mask->list);
 148                 if (deferred)
 149                         call_rcu(&mask->rcu, rcu_free_sw_flow_mask_cb);
 150                 else
 151                         kfree(mask);
 152         }
 153 }
 154
 155 void ovs_flow_free(struct sw_flow *flow, bool deferred)
 156 {
 157         if (!flow)
 158                 return;
 159
 160         flow_mask_del_ref(flow->mask, deferred);
 161
 162         if (deferred)
 163                 call_rcu(&flow->rcu, rcu_free_flow_callback);
 164         else
 165                 flow_free(flow);
 166 }
 167
 168 static void free_buckets(struct flex_array *buckets)
 169 {
 170         flex_array_free(buckets);
 171 }
 172
 173 static void __table_instance_destroy(struct table_instance *ti)
 174 {
 175         int i;
 176
 177         if (ti->keep_flows)
 178                 goto skip_flows;
 179
 180         for (i = 0; i < ti->n_buckets; i++) {
 181                 struct sw_flow *flow;
 182                 struct hlist_head *head = flex_array_get(ti->buckets, i);
 183                 struct hlist_node *n;
 184                 int ver = ti->node_ver;
 185
 186                 hlist_for_each_entry_safe(flow, n, head, hash_node[ver]) {
 187                         hlist_del(&flow->hash_node[ver]);
 188                         ovs_flow_free(flow, false);
 189                 }
 190         }
 191
 192 skip_flows:
 193         free_buckets(ti->buckets);
 194         kfree(ti);
 195 }
 196
 197 static struct table_instance *table_instance_alloc(int new_size)
 198 {
 199         struct table_instance *ti = kmalloc(sizeof(*ti), GFP_KERNEL);
 200
 201         if (!ti)
 202                 return NULL;
 203
 204         ti->buckets = alloc_buckets(new_size);
 205
 206         if (!ti->buckets) {
 207                 kfree(ti);
 208                 return NULL;
 209         }
 210         ti->n_buckets = new_size;
 211         ti->node_ver = 0;
 212         ti->keep_flows = false;
 213         get_random_bytes(&ti->hash_seed, sizeof(u32));
 214
 215         return ti;
 216 }
 217
 218 int ovs_flow_tbl_init(struct flow_table *table)
 219 {
 220         struct table_instance *ti;
 221
 222         ti = table_instance_alloc(TBL_MIN_BUCKETS);
 223
 224         if (!ti)
 225                 return -ENOMEM;
 226
 227         rcu_assign_pointer(table->ti, ti);
 228         INIT_LIST_HEAD(&table->mask_list);
 229         table->last_rehash = jiffies;
 230         table->count = 0;
 231         return 0;
 232 }
 233
 234 static void flow_tbl_destroy_rcu_cb(struct rcu_head *rcu)
 235 {
 236         struct table_instance *ti = container_of(rcu, struct table_instance, rcu);
 237
 238         __table_instance_destroy(ti);
 239 }
 240
 241 static void table_instance_destroy(struct table_instance *ti, bool deferred)
 242 {
 243         if (!ti)
 244                 return;
 245
 246         if (deferred)
 247                 call_rcu(&ti->rcu, flow_tbl_destroy_rcu_cb);
 248         else
 249                 __table_instance_destroy(ti);
 250 }
 251
 252 void ovs_flow_tbl_destroy(struct flow_table *table)
 253 {
 254         struct table_instance *ti = ovsl_dereference(table->ti);
 255
 256         table_instance_destroy(ti, false);
 257 }
 258
 259 struct sw_flow *ovs_flow_tbl_dump_next(struct table_instance *ti,
 260                                        u32 *bucket, u32 *last)
 261 {
 262         struct sw_flow *flow;
 263         struct hlist_head *head;
 264         int ver;
 265         int i;
 266
 267         ver = ti->node_ver;
 268         while (*bucket < ti->n_buckets) {
 269                 i = 0;
 270                 head = flex_array_get(ti->buckets, *bucket);
 271                 hlist_for_each_entry_rcu(flow, head, hash_node[ver]) {
 272                         if (i < *last) {
 273                                 i++;
 274                                 continue;
 275                         }
 276                         *last = i + 1;
 277                         return flow;
 278                 }
 279                 (*bucket)++;
 280                 *last = 0;
 281         }
 282
 283         return NULL;
 284 }
 285
 286 static struct hlist_head *find_bucket(struct table_instance *ti, u32 hash)
 287 {
 288         hash = jhash_1word(hash, ti->hash_seed);
 289         return flex_array_get(ti->buckets,
 290                                 (hash & (ti->n_buckets - 1)));
 291 }
 292
 293 static void table_instance_insert(struct table_instance *ti, struct sw_flow *flow)
 294 {
 295         struct hlist_head *head;
 296
 297         head = find_bucket(ti, flow->hash);
 298         hlist_add_head_rcu(&flow->hash_node[ti->node_ver], head);
 299 }
 300
 301 static void flow_table_copy_flows(struct table_instance *old,
 302                                   struct table_instance *new)
 303 {
 304         int old_ver;
 305         int i;
 306
 307         old_ver = old->node_ver;
 308         new->node_ver = !old_ver;
 309
 310         /* Insert in new table. */
 311         for (i = 0; i < old->n_buckets; i++) {
 312                 struct sw_flow *flow;
 313                 struct hlist_head *head;
 314
 315                 head = flex_array_get(old->buckets, i);
 316
 317                 hlist_for_each_entry(flow, head, hash_node[old_ver])
 318                         table_instance_insert(new, flow);
 319         }
 320
 321         old->keep_flows = true;
 322 }
 323
 324 static struct table_instance *table_instance_rehash(struct table_instance *ti,
 325                                             int n_buckets)
 326 {
 327         struct table_instance *new_ti;
 328
 329         new_ti = table_instance_alloc(n_buckets);
 330         if (!new_ti)
 331                 return NULL;
 332
 333         flow_table_copy_flows(ti, new_ti);
 334
 335         return new_ti;
 336 }
 337
 338 int ovs_flow_tbl_flush(struct flow_table *flow_table)
 339 {
 340         struct table_instance *old_ti;
 341         struct table_instance *new_ti;
 342
 343         old_ti = ovsl_dereference(flow_table->ti);
 344         new_ti = table_instance_alloc(TBL_MIN_BUCKETS);
 345         if (!new_ti)
 346                 return -ENOMEM;
 347
 348         rcu_assign_pointer(flow_table->ti, new_ti);
 349         flow_table->last_rehash = jiffies;
 350         flow_table->count = 0;
 351
 352         table_instance_destroy(old_ti, true);
 353         return 0;
 354 }
 355
 356 static u32 flow_hash(const struct sw_flow_key *key, int key_start,
 357                      int key_end)
 358 {
 359         u32 *hash_key = (u32 *)((u8 *)key + key_start);
 360         int hash_u32s = (key_end - key_start) >> 2;
 361
 362         /* Make sure number of hash bytes are multiple of u32. */
 363         BUILD_BUG_ON(sizeof(long) % sizeof(u32));
 364
 365         return jhash2(hash_key, hash_u32s, 0);
 366 }
 367
 368 static int flow_key_start(const struct sw_flow_key *key)
 369 {
 370         if (key->tun_key.ipv4_dst)
 371                 return 0;
 372         else
 373                 return rounddown(offsetof(struct sw_flow_key, phy),
 374                                           sizeof(long));
 375 }
 376
 377 static bool cmp_key(const struct sw_flow_key *key1,
 378                     const struct sw_flow_key *key2,
 379                     int key_start, int key_end)
 380 {
 381         const long *cp1 = (long *)((u8 *)key1 + key_start);
 382         const long *cp2 = (long *)((u8 *)key2 + key_start);
 383         long diffs = 0;
 384         int i;
 385
 386         for (i = key_start; i < key_end;  i += sizeof(long))
 387                 diffs |= *cp1++ ^ *cp2++;
 388
 389         return diffs == 0;
 390 }
 391
 392 static bool flow_cmp_masked_key(const struct sw_flow *flow,
 393                                 const struct sw_flow_key *key,
 394                                 int key_start, int key_end)
 395 {
 396         return cmp_key(&flow->key, key, key_start, key_end);
 397 }
 398
 399 bool ovs_flow_cmp_unmasked_key(const struct sw_flow *flow,
 400                                struct sw_flow_match *match)
 401 {
 402         struct sw_flow_key *key = match->key;
 403         int key_start = flow_key_start(key);
 404         int key_end = match->range.end;
 405
 406         return cmp_key(&flow->unmasked_key, key, key_start, key_end);
 407 }
 408
 409 static struct sw_flow *masked_flow_lookup(struct table_instance *ti,
 410                                           const struct sw_flow_key *unmasked,
 411                                           struct sw_flow_mask *mask)
 412 {
 413         struct sw_flow *flow;
 414         struct hlist_head *head;
 415         int key_start = mask->range.start;
 416         int key_end = mask->range.end;
 417         u32 hash;
 418         struct sw_flow_key masked_key;
 419
 420         ovs_flow_mask_key(&masked_key, unmasked, mask);
 421         hash = flow_hash(&masked_key, key_start, key_end);
 422         head = find_bucket(ti, hash);
 423         hlist_for_each_entry_rcu(flow, head, hash_node[ti->node_ver]) {
 424                 if (flow->mask == mask && flow->hash == hash &&
 425                     flow_cmp_masked_key(flow, &masked_key,
 426                                           key_start, key_end))
 427                         return flow;
 428         }
 429         return NULL;
 430 }
 431
 432 struct sw_flow *ovs_flow_tbl_lookup_stats(struct flow_table *tbl,
 433                                     const struct sw_flow_key *key,
 434                                     u32 *n_mask_hit)
 435 {
 436         struct table_instance *ti = rcu_dereference(tbl->ti);
 437         struct sw_flow_mask *mask;
 438         struct sw_flow *flow;
 439
 440         *n_mask_hit = 0;
 441         list_for_each_entry_rcu(mask, &tbl->mask_list, list) {
 442                 (*n_mask_hit)++;
 443                 flow = masked_flow_lookup(ti, key, mask);
 444                 if (flow)  /* Found */
 445                         return flow;
 446         }
 447         return NULL;
 448 }
 449
 450 struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *tbl,
 451                                     const struct sw_flow_key *key)
 452 {
 453         u32 __always_unused n_mask_hit;
 454
 455         return ovs_flow_tbl_lookup_stats(tbl, key, &n_mask_hit);
 456 }
 457
 458 int ovs_flow_tbl_num_masks(const struct flow_table *table)
 459 {
 460         struct sw_flow_mask *mask;
 461         int num = 0;
 462
 463         list_for_each_entry(mask, &table->mask_list, list)
 464                 num++;
 465
 466         return num;
 467 }
 468
 469 static struct table_instance *table_instance_expand(struct table_instance *ti)
 470 {
 471         return table_instance_rehash(ti, ti->n_buckets * 2);
 472 }
 473
 474 void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow)
 475 {
 476         struct table_instance *ti = ovsl_dereference(table->ti);
 477
 478         BUG_ON(table->count == 0);
 479         hlist_del_rcu(&flow->hash_node[ti->node_ver]);
 480         table->count--;
 481 }
 482
 483 static struct sw_flow_mask *mask_alloc(void)
 484 {
 485         struct sw_flow_mask *mask;
 486
 487         mask = kmalloc(sizeof(*mask), GFP_KERNEL);
 488         if (mask)
 489                 mask->ref_count = 0;
 490
 491         return mask;
 492 }
 493
 494 static void mask_add_ref(struct sw_flow_mask *mask)
 495 {
 496         mask->ref_count++;
 497 }
 498
 499 static bool mask_equal(const struct sw_flow_mask *a,
 500                        const struct sw_flow_mask *b)
 501 {
 502         u8 *a_ = (u8 *)&a->key + a->range.start;
 503         u8 *b_ = (u8 *)&b->key + b->range.start;
 504
 505         return  (a->range.end == b->range.end)
 506                 && (a->range.start == b->range.start)
 507                 && (memcmp(a_, b_, range_n_bytes(&a->range)) == 0);
 508 }
 509
 510 static struct sw_flow_mask *flow_mask_find(const struct flow_table *tbl,
 511                                            const struct sw_flow_mask *mask)
 512 {
 513         struct list_head *ml;
 514
 515         list_for_each(ml, &tbl->mask_list) {
 516                 struct sw_flow_mask *m;
 517                 m = container_of(ml, struct sw_flow_mask, list);
 518                 if (mask_equal(mask, m))
 519                         return m;
 520         }
 521
 522         return NULL;
 523 }
 524
 525 /* Add 'mask' into the mask list, if it is not already there. */
 526 static int flow_mask_insert(struct flow_table *tbl, struct sw_flow *flow,
 527                             struct sw_flow_mask *new)
 528 {
 529         struct sw_flow_mask *mask;
 530         mask = flow_mask_find(tbl, new);
 531         if (!mask) {
 532                 /* Allocate a new mask if none exsits. */
 533                 mask = mask_alloc();
 534                 if (!mask)
 535                         return -ENOMEM;
 536                 mask->key = new->key;
 537                 mask->range = new->range;
 538                 list_add_rcu(&mask->list, &tbl->mask_list);
 539         }
 540
 541         mask_add_ref(mask);
 542         flow->mask = mask;
 543         return 0;
 544 }
 545
 546 int ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow,
 547                         struct sw_flow_mask *mask)
 548 {
 549         struct table_instance *new_ti = NULL;
 550         struct table_instance *ti;
 551         int err;
 552
 553         err = flow_mask_insert(table, flow, mask);
 554         if (err)
 555                 return err;
 556
 557         flow->hash = flow_hash(&flow->key, flow->mask->range.start,
 558                         flow->mask->range.end);
 559         ti = ovsl_dereference(table->ti);
 560         table_instance_insert(ti, flow);
 561         table->count++;
 562
 563         /* Expand table, if necessary, to make room. */
 564         if (table->count > ti->n_buckets)
 565                 new_ti = table_instance_expand(ti);
 566         else if (time_after(jiffies, table->last_rehash + REHASH_INTERVAL))
 567                 new_ti = table_instance_rehash(ti, ti->n_buckets);
 568
 569         if (new_ti) {
 570                 rcu_assign_pointer(table->ti, new_ti);
 571                 table_instance_destroy(ti, true);
 572                 table->last_rehash = jiffies;
 573         }
 574         return 0;
 575 }
 576
 577 /* Initializes the flow module.
 578  * Returns zero if successful or a negative error code. */
 579 int ovs_flow_init(void)
 580 {
 581         BUILD_BUG_ON(__alignof__(struct sw_flow_key) % __alignof__(long));
 582         BUILD_BUG_ON(sizeof(struct sw_flow_key) % sizeof(long));
 583
 584         flow_cache = kmem_cache_create("sw_flow", sizeof(struct sw_flow), 0,
 585                                         0, NULL);
 586         if (flow_cache == NULL)
 587                 return -ENOMEM;
 588
 589         return 0;
 590 }
 591
 592 /* Uninitializes the flow module. */
 593 void ovs_flow_exit(void)
 594 {
 595         kmem_cache_destroy(flow_cache);
 596 }