lib/librte_lpm/rte_lpm.c

   1 /* SPDX-License-Identifier: BSD-3-Clause
   2  * Copyright(c) 2010-2014 Intel Corporation
   3  */
   4
   5 #include <string.h>
   6 #include <stdint.h>
   7 #include <errno.h>
   8 #include <stdarg.h>
   9 #include <stdio.h>
  10 #include <sys/queue.h>
  11
  12 #include <rte_log.h>
  13 #include <rte_branch_prediction.h>
  14 #include <rte_common.h>
  15 #include <rte_memory.h>        /* for definition of RTE_CACHE_LINE_SIZE */
  16 #include <rte_malloc.h>
  17 #include <rte_eal.h>
  18 #include <rte_eal_memconfig.h>
  19 #include <rte_per_lcore.h>
  20 #include <rte_string_fns.h>
  21 #include <rte_errno.h>
  22 #include <rte_rwlock.h>
  23 #include <rte_spinlock.h>
  24 #include <rte_tailq.h>
  25
  26 #include "rte_lpm.h"
  27
  28 TAILQ_HEAD(rte_lpm_list, rte_tailq_entry);
  29
  30 static struct rte_tailq_elem rte_lpm_tailq = {
  31         .name = "RTE_LPM",
  32 };
  33 EAL_REGISTER_TAILQ(rte_lpm_tailq)
  34
  35 #define MAX_DEPTH_TBL24 24
  36
  37 enum valid_flag {
  38         INVALID = 0,
  39         VALID
  40 };
  41
  42 /* Macro to enable/disable run-time checks. */
  43 #if defined(RTE_LIBRTE_LPM_DEBUG)
  44 #include <rte_debug.h>
  45 #define VERIFY_DEPTH(depth) do {                                \
  46         if ((depth == 0) || (depth > RTE_LPM_MAX_DEPTH))        \
  47                 rte_panic("LPM: Invalid depth (%u) at line %d", \
  48                                 (unsigned)(depth), __LINE__);   \
  49 } while (0)
  50 #else
  51 #define VERIFY_DEPTH(depth)
  52 #endif
  53
  54 /*
  55  * Converts a given depth value to its corresponding mask value.
  56  *
  57  * depth  (IN)          : range = 1 - 32
  58  * mask   (OUT)         : 32bit mask
  59  */
  60 static uint32_t __attribute__((pure))
  61 depth_to_mask(uint8_t depth)
  62 {
  63         VERIFY_DEPTH(depth);
  64
  65         /* To calculate a mask start with a 1 on the left hand side and right
  66          * shift while populating the left hand side with 1's
  67          */
  68         return (int)0x80000000 >> (depth - 1);
  69 }
  70
  71 /*
  72  * Converts given depth value to its corresponding range value.
  73  */
  74 static uint32_t __attribute__((pure))
  75 depth_to_range(uint8_t depth)
  76 {
  77         VERIFY_DEPTH(depth);
  78
  79         /*
  80          * Calculate tbl24 range. (Note: 2^depth = 1 << depth)
  81          */
  82         if (depth <= MAX_DEPTH_TBL24)
  83                 return 1 << (MAX_DEPTH_TBL24 - depth);
  84
  85         /* Else if depth is greater than 24 */
  86         return 1 << (RTE_LPM_MAX_DEPTH - depth);
  87 }
  88
  89 /*
  90  * Find an existing lpm table and return a pointer to it.
  91  */
  92 struct rte_lpm *
  93 rte_lpm_find_existing(const char *name)
  94 {
  95         struct rte_lpm *l = NULL;
  96         struct rte_tailq_entry *te;
  97         struct rte_lpm_list *lpm_list;
  98
  99         lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 100
 101         rte_mcfg_tailq_read_lock();
 102         TAILQ_FOREACH(te, lpm_list, next) {
 103                 l = te->data;
 104                 if (strncmp(name, l->name, RTE_LPM_NAMESIZE) == 0)
 105                         break;
 106         }
 107         rte_mcfg_tailq_read_unlock();
 108
 109         if (te == NULL) {
 110                 rte_errno = ENOENT;
 111                 return NULL;
 112         }
 113
 114         return l;
 115 }
 116
 117 /*
 118  * Allocates memory for LPM object
 119  */
 120 struct rte_lpm *
 121 rte_lpm_create(const char *name, int socket_id,
 122                 const struct rte_lpm_config *config)
 123 {
 124         char mem_name[RTE_LPM_NAMESIZE];
 125         struct rte_lpm *lpm = NULL;
 126         struct rte_tailq_entry *te;
 127         uint32_t mem_size, rules_size, tbl8s_size;
 128         struct rte_lpm_list *lpm_list;
 129
 130         lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 131
 132         RTE_BUILD_BUG_ON(sizeof(struct rte_lpm_tbl_entry) != 4);
 133
 134         /* Check user arguments. */
 135         if ((name == NULL) || (socket_id < -1) || (config->max_rules == 0)
 136                         || config->number_tbl8s > RTE_LPM_MAX_TBL8_NUM_GROUPS) {
 137                 rte_errno = EINVAL;
 138                 return NULL;
 139         }
 140
 141         snprintf(mem_name, sizeof(mem_name), "LPM_%s", name);
 142
 143         /* Determine the amount of memory to allocate. */
 144         mem_size = sizeof(*lpm);
 145         rules_size = sizeof(struct rte_lpm_rule) * config->max_rules;
 146         tbl8s_size = (sizeof(struct rte_lpm_tbl_entry) *
 147                         RTE_LPM_TBL8_GROUP_NUM_ENTRIES * config->number_tbl8s);
 148
 149         rte_mcfg_tailq_write_lock();
 150
 151         /* guarantee there's no existing */
 152         TAILQ_FOREACH(te, lpm_list, next) {
 153                 lpm = te->data;
 154                 if (strncmp(name, lpm->name, RTE_LPM_NAMESIZE) == 0)
 155                         break;
 156         }
 157
 158         if (te != NULL) {
 159                 lpm = NULL;
 160                 rte_errno = EEXIST;
 161                 goto exit;
 162         }
 163
 164         /* allocate tailq entry */
 165         te = rte_zmalloc("LPM_TAILQ_ENTRY", sizeof(*te), 0);
 166         if (te == NULL) {
 167                 RTE_LOG(ERR, LPM, "Failed to allocate tailq entry\n");
 168                 rte_errno = ENOMEM;
 169                 goto exit;
 170         }
 171
 172         /* Allocate memory to store the LPM data structures. */
 173         lpm = rte_zmalloc_socket(mem_name, mem_size,
 174                         RTE_CACHE_LINE_SIZE, socket_id);
 175         if (lpm == NULL) {
 176                 RTE_LOG(ERR, LPM, "LPM memory allocation failed\n");
 177                 rte_free(te);
 178                 rte_errno = ENOMEM;
 179                 goto exit;
 180         }
 181
 182         lpm->rules_tbl = rte_zmalloc_socket(NULL,
 183                         (size_t)rules_size, RTE_CACHE_LINE_SIZE, socket_id);
 184
 185         if (lpm->rules_tbl == NULL) {
 186                 RTE_LOG(ERR, LPM, "LPM rules_tbl memory allocation failed\n");
 187                 rte_free(lpm);
 188                 lpm = NULL;
 189                 rte_free(te);
 190                 rte_errno = ENOMEM;
 191                 goto exit;
 192         }
 193
 194         lpm->tbl8 = rte_zmalloc_socket(NULL,
 195                         (size_t)tbl8s_size, RTE_CACHE_LINE_SIZE, socket_id);
 196
 197         if (lpm->tbl8 == NULL) {
 198                 RTE_LOG(ERR, LPM, "LPM tbl8 memory allocation failed\n");
 199                 rte_free(lpm->rules_tbl);
 200                 rte_free(lpm);
 201                 lpm = NULL;
 202                 rte_free(te);
 203                 rte_errno = ENOMEM;
 204                 goto exit;
 205         }
 206
 207         /* Save user arguments. */
 208         lpm->max_rules = config->max_rules;
 209         lpm->number_tbl8s = config->number_tbl8s;
 210         strlcpy(lpm->name, name, sizeof(lpm->name));
 211
 212         te->data = lpm;
 213
 214         TAILQ_INSERT_TAIL(lpm_list, te, next);
 215
 216 exit:
 217         rte_mcfg_tailq_write_unlock();
 218
 219         return lpm;
 220 }
 221
 222 /*
 223  * Deallocates memory for given LPM table.
 224  */
 225 void
 226 rte_lpm_free(struct rte_lpm *lpm)
 227 {
 228         struct rte_lpm_list *lpm_list;
 229         struct rte_tailq_entry *te;
 230
 231         /* Check user arguments. */
 232         if (lpm == NULL)
 233                 return;
 234
 235         lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 236
 237         rte_mcfg_tailq_write_lock();
 238
 239         /* find our tailq entry */
 240         TAILQ_FOREACH(te, lpm_list, next) {
 241                 if (te->data == (void *) lpm)
 242                         break;
 243         }
 244         if (te != NULL)
 245                 TAILQ_REMOVE(lpm_list, te, next);
 246
 247         rte_mcfg_tailq_write_unlock();
 248
 249         rte_free(lpm->tbl8);
 250         rte_free(lpm->rules_tbl);
 251         rte_free(lpm);
 252         rte_free(te);
 253 }
 254
 255 /*
 256  * Adds a rule to the rule table.
 257  *
 258  * NOTE: The rule table is split into 32 groups. Each group contains rules that
 259  * apply to a specific prefix depth (i.e. group 1 contains rules that apply to
 260  * prefixes with a depth of 1 etc.). In the following code (depth - 1) is used
 261  * to refer to depth 1 because even though the depth range is 1 - 32, depths
 262  * are stored in the rule table from 0 - 31.
 263  * NOTE: Valid range for depth parameter is 1 .. 32 inclusive.
 264  */
 265 static int32_t
 266 rule_add(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
 267         uint32_t next_hop)
 268 {
 269         uint32_t rule_gindex, rule_index, last_rule;
 270         int i;
 271
 272         VERIFY_DEPTH(depth);
 273
 274         /* Scan through rule group to see if rule already exists. */
 275         if (lpm->rule_info[depth - 1].used_rules > 0) {
 276
 277                 /* rule_gindex stands for rule group index. */
 278                 rule_gindex = lpm->rule_info[depth - 1].first_rule;
 279                 /* Initialise rule_index to point to start of rule group. */
 280                 rule_index = rule_gindex;
 281                 /* Last rule = Last used rule in this rule group. */
 282                 last_rule = rule_gindex + lpm->rule_info[depth - 1].used_rules;
 283
 284                 for (; rule_index < last_rule; rule_index++) {
 285
 286                         /* If rule already exists update its next_hop and return. */
 287                         if (lpm->rules_tbl[rule_index].ip == ip_masked) {
 288                                 lpm->rules_tbl[rule_index].next_hop = next_hop;
 289
 290                                 return rule_index;
 291                         }
 292                 }
 293
 294                 if (rule_index == lpm->max_rules)
 295                         return -ENOSPC;
 296         } else {
 297                 /* Calculate the position in which the rule will be stored. */
 298                 rule_index = 0;
 299
 300                 for (i = depth - 1; i > 0; i--) {
 301                         if (lpm->rule_info[i - 1].used_rules > 0) {
 302                                 rule_index = lpm->rule_info[i - 1].first_rule
 303                                                 + lpm->rule_info[i - 1].used_rules;
 304                                 break;
 305                         }
 306                 }
 307                 if (rule_index == lpm->max_rules)
 308                         return -ENOSPC;
 309
 310                 lpm->rule_info[depth - 1].first_rule = rule_index;
 311         }
 312
 313         /* Make room for the new rule in the array. */
 314         for (i = RTE_LPM_MAX_DEPTH; i > depth; i--) {
 315                 if (lpm->rule_info[i - 1].first_rule
 316                                 + lpm->rule_info[i - 1].used_rules == lpm->max_rules)
 317                         return -ENOSPC;
 318
 319                 if (lpm->rule_info[i - 1].used_rules > 0) {
 320                         lpm->rules_tbl[lpm->rule_info[i - 1].first_rule
 321                                 + lpm->rule_info[i - 1].used_rules]
 322                                         = lpm->rules_tbl[lpm->rule_info[i - 1].first_rule];
 323                         lpm->rule_info[i - 1].first_rule++;
 324                 }
 325         }
 326
 327         /* Add the new rule. */
 328         lpm->rules_tbl[rule_index].ip = ip_masked;
 329         lpm->rules_tbl[rule_index].next_hop = next_hop;
 330
 331         /* Increment the used rules counter for this rule group. */
 332         lpm->rule_info[depth - 1].used_rules++;
 333
 334         return rule_index;
 335 }
 336
 337 /*
 338  * Delete a rule from the rule table.
 339  * NOTE: Valid range for depth parameter is 1 .. 32 inclusive.
 340  */
 341 static void
 342 rule_delete(struct rte_lpm *lpm, int32_t rule_index, uint8_t depth)
 343 {
 344         int i;
 345
 346         VERIFY_DEPTH(depth);
 347
 348         lpm->rules_tbl[rule_index] =
 349                         lpm->rules_tbl[lpm->rule_info[depth - 1].first_rule
 350                         + lpm->rule_info[depth - 1].used_rules - 1];
 351
 352         for (i = depth; i < RTE_LPM_MAX_DEPTH; i++) {
 353                 if (lpm->rule_info[i].used_rules > 0) {
 354                         lpm->rules_tbl[lpm->rule_info[i].first_rule - 1] =
 355                                         lpm->rules_tbl[lpm->rule_info[i].first_rule
 356                                                 + lpm->rule_info[i].used_rules - 1];
 357                         lpm->rule_info[i].first_rule--;
 358                 }
 359         }
 360
 361         lpm->rule_info[depth - 1].used_rules--;
 362 }
 363
 364 /*
 365  * Finds a rule in rule table.
 366  * NOTE: Valid range for depth parameter is 1 .. 32 inclusive.
 367  */
 368 static int32_t
 369 rule_find(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth)
 370 {
 371         uint32_t rule_gindex, last_rule, rule_index;
 372
 373         VERIFY_DEPTH(depth);
 374
 375         rule_gindex = lpm->rule_info[depth - 1].first_rule;
 376         last_rule = rule_gindex + lpm->rule_info[depth - 1].used_rules;
 377
 378         /* Scan used rules at given depth to find rule. */
 379         for (rule_index = rule_gindex; rule_index < last_rule; rule_index++) {
 380                 /* If rule is found return the rule index. */
 381                 if (lpm->rules_tbl[rule_index].ip == ip_masked)
 382                         return rule_index;
 383         }
 384
 385         /* If rule is not found return -EINVAL. */
 386         return -EINVAL;
 387 }
 388
 389 /*
 390  * Find, clean and allocate a tbl8.
 391  */
 392 static int32_t
 393 tbl8_alloc(struct rte_lpm_tbl_entry *tbl8, uint32_t number_tbl8s)
 394 {
 395         uint32_t group_idx; /* tbl8 group index. */
 396         struct rte_lpm_tbl_entry *tbl8_entry;
 397
 398         /* Scan through tbl8 to find a free (i.e. INVALID) tbl8 group. */
 399         for (group_idx = 0; group_idx < number_tbl8s; group_idx++) {
 400                 tbl8_entry = &tbl8[group_idx * RTE_LPM_TBL8_GROUP_NUM_ENTRIES];
 401                 /* If a free tbl8 group is found clean it and set as VALID. */
 402                 if (!tbl8_entry->valid_group) {
 403                         struct rte_lpm_tbl_entry new_tbl8_entry = {
 404                                 .next_hop = 0,
 405                                 .valid = INVALID,
 406                                 .depth = 0,
 407                                 .valid_group = VALID,
 408                         };
 409
 410                         memset(&tbl8_entry[0], 0,
 411                                         RTE_LPM_TBL8_GROUP_NUM_ENTRIES *
 412                                         sizeof(tbl8_entry[0]));
 413
 414                         __atomic_store(tbl8_entry, &new_tbl8_entry,
 415                                         __ATOMIC_RELAXED);
 416
 417                         /* Return group index for allocated tbl8 group. */
 418                         return group_idx;
 419                 }
 420         }
 421
 422         /* If there are no tbl8 groups free then return error. */
 423         return -ENOSPC;
 424 }
 425
 426 static void
 427 tbl8_free(struct rte_lpm_tbl_entry *tbl8, uint32_t tbl8_group_start)
 428 {
 429         /* Set tbl8 group invalid*/
 430         struct rte_lpm_tbl_entry zero_tbl8_entry = {0};
 431
 432         __atomic_store(&tbl8[tbl8_group_start], &zero_tbl8_entry,
 433                         __ATOMIC_RELAXED);
 434 }
 435
 436 static __rte_noinline int32_t
 437 add_depth_small(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
 438                 uint32_t next_hop)
 439 {
 440 #define group_idx next_hop
 441         uint32_t tbl24_index, tbl24_range, tbl8_index, tbl8_group_end, i, j;
 442
 443         /* Calculate the index into Table24. */
 444         tbl24_index = ip >> 8;
 445         tbl24_range = depth_to_range(depth);
 446
 447         for (i = tbl24_index; i < (tbl24_index + tbl24_range); i++) {
 448                 /*
 449                  * For invalid OR valid and non-extended tbl 24 entries set
 450                  * entry.
 451                  */
 452                 if (!lpm->tbl24[i].valid || (lpm->tbl24[i].valid_group == 0 &&
 453                                 lpm->tbl24[i].depth <= depth)) {
 454
 455                         struct rte_lpm_tbl_entry new_tbl24_entry = {
 456                                 .next_hop = next_hop,
 457                                 .valid = VALID,
 458                                 .valid_group = 0,
 459                                 .depth = depth,
 460                         };
 461
 462                         /* Setting tbl24 entry in one go to avoid race
 463                          * conditions
 464                          */
 465                         __atomic_store(&lpm->tbl24[i], &new_tbl24_entry,
 466                                         __ATOMIC_RELEASE);
 467
 468                         continue;
 469                 }
 470
 471                 if (lpm->tbl24[i].valid_group == 1) {
 472                         /* If tbl24 entry is valid and extended calculate the
 473                          *  index into tbl8.
 474                          */
 475                         tbl8_index = lpm->tbl24[i].group_idx *
 476                                         RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 477                         tbl8_group_end = tbl8_index +
 478                                         RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 479
 480                         for (j = tbl8_index; j < tbl8_group_end; j++) {
 481                                 if (!lpm->tbl8[j].valid ||
 482                                                 lpm->tbl8[j].depth <= depth) {
 483                                         struct rte_lpm_tbl_entry
 484                                                 new_tbl8_entry = {
 485                                                 .valid = VALID,
 486                                                 .valid_group = VALID,
 487                                                 .depth = depth,
 488                                                 .next_hop = next_hop,
 489                                         };
 490
 491                                         /*
 492                                          * Setting tbl8 entry in one go to avoid
 493                                          * race conditions
 494                                          */
 495                                         __atomic_store(&lpm->tbl8[j],
 496                                                 &new_tbl8_entry,
 497                                                 __ATOMIC_RELAXED);
 498
 499                                         continue;
 500                                 }
 501                         }
 502                 }
 503         }
 504 #undef group_idx
 505         return 0;
 506 }
 507
 508 static __rte_noinline int32_t
 509 add_depth_big(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
 510                 uint32_t next_hop)
 511 {
 512 #define group_idx next_hop
 513         uint32_t tbl24_index;
 514         int32_t tbl8_group_index, tbl8_group_start, tbl8_group_end, tbl8_index,
 515                 tbl8_range, i;
 516
 517         tbl24_index = (ip_masked >> 8);
 518         tbl8_range = depth_to_range(depth);
 519
 520         if (!lpm->tbl24[tbl24_index].valid) {
 521                 /* Search for a free tbl8 group. */
 522                 tbl8_group_index = tbl8_alloc(lpm->tbl8, lpm->number_tbl8s);
 523
 524                 /* Check tbl8 allocation was successful. */
 525                 if (tbl8_group_index < 0) {
 526                         return tbl8_group_index;
 527                 }
 528
 529                 /* Find index into tbl8 and range. */
 530                 tbl8_index = (tbl8_group_index *
 531                                 RTE_LPM_TBL8_GROUP_NUM_ENTRIES) +
 532                                 (ip_masked & 0xFF);
 533
 534                 /* Set tbl8 entry. */
 535                 for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
 536                         struct rte_lpm_tbl_entry new_tbl8_entry = {
 537                                 .valid = VALID,
 538                                 .depth = depth,
 539                                 .valid_group = lpm->tbl8[i].valid_group,
 540                                 .next_hop = next_hop,
 541                         };
 542                         __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
 543                                         __ATOMIC_RELAXED);
 544                 }
 545
 546                 /*
 547                  * Update tbl24 entry to point to new tbl8 entry. Note: The
 548                  * ext_flag and tbl8_index need to be updated simultaneously,
 549                  * so assign whole structure in one go
 550                  */
 551
 552                 struct rte_lpm_tbl_entry new_tbl24_entry = {
 553                         .group_idx = tbl8_group_index,
 554                         .valid = VALID,
 555                         .valid_group = 1,
 556                         .depth = 0,
 557                 };
 558
 559                 /* The tbl24 entry must be written only after the
 560                  * tbl8 entries are written.
 561                  */
 562                 __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
 563                                 __ATOMIC_RELEASE);
 564
 565         } /* If valid entry but not extended calculate the index into Table8. */
 566         else if (lpm->tbl24[tbl24_index].valid_group == 0) {
 567                 /* Search for free tbl8 group. */
 568                 tbl8_group_index = tbl8_alloc(lpm->tbl8, lpm->number_tbl8s);
 569
 570                 if (tbl8_group_index < 0) {
 571                         return tbl8_group_index;
 572                 }
 573
 574                 tbl8_group_start = tbl8_group_index *
 575                                 RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 576                 tbl8_group_end = tbl8_group_start +
 577                                 RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 578
 579                 /* Populate new tbl8 with tbl24 value. */
 580                 for (i = tbl8_group_start; i < tbl8_group_end; i++) {
 581                         struct rte_lpm_tbl_entry new_tbl8_entry = {
 582                                 .valid = VALID,
 583                                 .depth = lpm->tbl24[tbl24_index].depth,
 584                                 .valid_group = lpm->tbl8[i].valid_group,
 585                                 .next_hop = lpm->tbl24[tbl24_index].next_hop,
 586                         };
 587                         __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
 588                                         __ATOMIC_RELAXED);
 589                 }
 590
 591                 tbl8_index = tbl8_group_start + (ip_masked & 0xFF);
 592
 593                 /* Insert new rule into the tbl8 entry. */
 594                 for (i = tbl8_index; i < tbl8_index + tbl8_range; i++) {
 595                         struct rte_lpm_tbl_entry new_tbl8_entry = {
 596                                 .valid = VALID,
 597                                 .depth = depth,
 598                                 .valid_group = lpm->tbl8[i].valid_group,
 599                                 .next_hop = next_hop,
 600                         };
 601                         __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
 602                                         __ATOMIC_RELAXED);
 603                 }
 604
 605                 /*
 606                  * Update tbl24 entry to point to new tbl8 entry. Note: The
 607                  * ext_flag and tbl8_index need to be updated simultaneously,
 608                  * so assign whole structure in one go.
 609                  */
 610
 611                 struct rte_lpm_tbl_entry new_tbl24_entry = {
 612                                 .group_idx = tbl8_group_index,
 613                                 .valid = VALID,
 614                                 .valid_group = 1,
 615                                 .depth = 0,
 616                 };
 617
 618                 /* The tbl24 entry must be written only after the
 619                  * tbl8 entries are written.
 620                  */
 621                 __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
 622                                 __ATOMIC_RELEASE);
 623
 624         } else { /*
 625                 * If it is valid, extended entry calculate the index into tbl8.
 626                 */
 627                 tbl8_group_index = lpm->tbl24[tbl24_index].group_idx;
 628                 tbl8_group_start = tbl8_group_index *
 629                                 RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 630                 tbl8_index = tbl8_group_start + (ip_masked & 0xFF);
 631
 632                 for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
 633
 634                         if (!lpm->tbl8[i].valid ||
 635                                         lpm->tbl8[i].depth <= depth) {
 636                                 struct rte_lpm_tbl_entry new_tbl8_entry = {
 637                                         .valid = VALID,
 638                                         .depth = depth,
 639                                         .next_hop = next_hop,
 640                                         .valid_group = lpm->tbl8[i].valid_group,
 641                                 };
 642
 643                                 /*
 644                                  * Setting tbl8 entry in one go to avoid race
 645                                  * condition
 646                                  */
 647                                 __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
 648                                                 __ATOMIC_RELAXED);
 649
 650                                 continue;
 651                         }
 652                 }
 653         }
 654 #undef group_idx
 655         return 0;
 656 }
 657
 658 /*
 659  * Add a route
 660  */
 661 int
 662 rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
 663                 uint32_t next_hop)
 664 {
 665         int32_t rule_index, status = 0;
 666         uint32_t ip_masked;
 667
 668         /* Check user arguments. */
 669         if ((lpm == NULL) || (depth < 1) || (depth > RTE_LPM_MAX_DEPTH))
 670                 return -EINVAL;
 671
 672         ip_masked = ip & depth_to_mask(depth);
 673
 674         /* Add the rule to the rule table. */
 675         rule_index = rule_add(lpm, ip_masked, depth, next_hop);
 676
 677         /* If the is no space available for new rule return error. */
 678         if (rule_index < 0) {
 679                 return rule_index;
 680         }
 681
 682         if (depth <= MAX_DEPTH_TBL24) {
 683                 status = add_depth_small(lpm, ip_masked, depth, next_hop);
 684         } else { /* If depth > RTE_LPM_MAX_DEPTH_TBL24 */
 685                 status = add_depth_big(lpm, ip_masked, depth, next_hop);
 686
 687                 /*
 688                  * If add fails due to exhaustion of tbl8 extensions delete
 689                  * rule that was added to rule table.
 690                  */
 691                 if (status < 0) {
 692                         rule_delete(lpm, rule_index, depth);
 693
 694                         return status;
 695                 }
 696         }
 697
 698         return 0;
 699 }
 700
 701 /*
 702  * Look for a rule in the high-level rules table
 703  */
 704 int
 705 rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
 706 uint32_t *next_hop)
 707 {
 708         uint32_t ip_masked;
 709         int32_t rule_index;
 710
 711         /* Check user arguments. */
 712         if ((lpm == NULL) ||
 713                 (next_hop == NULL) ||
 714                 (depth < 1) || (depth > RTE_LPM_MAX_DEPTH))
 715                 return -EINVAL;
 716
 717         /* Look for the rule using rule_find. */
 718         ip_masked = ip & depth_to_mask(depth);
 719         rule_index = rule_find(lpm, ip_masked, depth);
 720
 721         if (rule_index >= 0) {
 722                 *next_hop = lpm->rules_tbl[rule_index].next_hop;
 723                 return 1;
 724         }
 725
 726         /* If rule is not found return 0. */
 727         return 0;
 728 }
 729
 730 static int32_t
 731 find_previous_rule(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
 732                 uint8_t *sub_rule_depth)
 733 {
 734         int32_t rule_index;
 735         uint32_t ip_masked;
 736         uint8_t prev_depth;
 737
 738         for (prev_depth = (uint8_t)(depth - 1); prev_depth > 0; prev_depth--) {
 739                 ip_masked = ip & depth_to_mask(prev_depth);
 740
 741                 rule_index = rule_find(lpm, ip_masked, prev_depth);
 742
 743                 if (rule_index >= 0) {
 744                         *sub_rule_depth = prev_depth;
 745                         return rule_index;
 746                 }
 747         }
 748
 749         return -1;
 750 }
 751
 752 static int32_t
 753 delete_depth_small(struct rte_lpm *lpm, uint32_t ip_masked,
 754         uint8_t depth, int32_t sub_rule_index, uint8_t sub_rule_depth)
 755 {
 756 #define group_idx next_hop
 757         uint32_t tbl24_range, tbl24_index, tbl8_group_index, tbl8_index, i, j;
 758
 759         /* Calculate the range and index into Table24. */
 760         tbl24_range = depth_to_range(depth);
 761         tbl24_index = (ip_masked >> 8);
 762         struct rte_lpm_tbl_entry zero_tbl24_entry = {0};
 763
 764         /*
 765          * Firstly check the sub_rule_index. A -1 indicates no replacement rule
 766          * and a positive number indicates a sub_rule_index.
 767          */
 768         if (sub_rule_index < 0) {
 769                 /*
 770                  * If no replacement rule exists then invalidate entries
 771                  * associated with this rule.
 772                  */
 773                 for (i = tbl24_index; i < (tbl24_index + tbl24_range); i++) {
 774
 775                         if (lpm->tbl24[i].valid_group == 0 &&
 776                                         lpm->tbl24[i].depth <= depth) {
 777                                 __atomic_store(&lpm->tbl24[i],
 778                                         &zero_tbl24_entry, __ATOMIC_RELEASE);
 779                         } else if (lpm->tbl24[i].valid_group == 1) {
 780                                 /*
 781                                  * If TBL24 entry is extended, then there has
 782                                  * to be a rule with depth >= 25 in the
 783                                  * associated TBL8 group.
 784                                  */
 785
 786                                 tbl8_group_index = lpm->tbl24[i].group_idx;
 787                                 tbl8_index = tbl8_group_index *
 788                                                 RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 789
 790                                 for (j = tbl8_index; j < (tbl8_index +
 791                                         RTE_LPM_TBL8_GROUP_NUM_ENTRIES); j++) {
 792
 793                                         if (lpm->tbl8[j].depth <= depth)
 794                                                 lpm->tbl8[j].valid = INVALID;
 795                                 }
 796                         }
 797                 }
 798         } else {
 799                 /*
 800                  * If a replacement rule exists then modify entries
 801                  * associated with this rule.
 802                  */
 803
 804                 struct rte_lpm_tbl_entry new_tbl24_entry = {
 805                         .next_hop = lpm->rules_tbl[sub_rule_index].next_hop,
 806                         .valid = VALID,
 807                         .valid_group = 0,
 808                         .depth = sub_rule_depth,
 809                 };
 810
 811                 struct rte_lpm_tbl_entry new_tbl8_entry = {
 812                         .valid = VALID,
 813                         .valid_group = VALID,
 814                         .depth = sub_rule_depth,
 815                         .next_hop = lpm->rules_tbl
 816                         [sub_rule_index].next_hop,
 817                 };
 818
 819                 for (i = tbl24_index; i < (tbl24_index + tbl24_range); i++) {
 820
 821                         if (lpm->tbl24[i].valid_group == 0 &&
 822                                         lpm->tbl24[i].depth <= depth) {
 823                                 __atomic_store(&lpm->tbl24[i], &new_tbl24_entry,
 824                                                 __ATOMIC_RELEASE);
 825                         } else  if (lpm->tbl24[i].valid_group == 1) {
 826                                 /*
 827                                  * If TBL24 entry is extended, then there has
 828                                  * to be a rule with depth >= 25 in the
 829                                  * associated TBL8 group.
 830                                  */
 831
 832                                 tbl8_group_index = lpm->tbl24[i].group_idx;
 833                                 tbl8_index = tbl8_group_index *
 834                                                 RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 835
 836                                 for (j = tbl8_index; j < (tbl8_index +
 837                                         RTE_LPM_TBL8_GROUP_NUM_ENTRIES); j++) {
 838
 839                                         if (lpm->tbl8[j].depth <= depth)
 840                                                 __atomic_store(&lpm->tbl8[j],
 841                                                         &new_tbl8_entry,
 842                                                         __ATOMIC_RELAXED);
 843                                 }
 844                         }
 845                 }
 846         }
 847 #undef group_idx
 848         return 0;
 849 }
 850
 851 /*
 852  * Checks if table 8 group can be recycled.
 853  *
 854  * Return of -EEXIST means tbl8 is in use and thus can not be recycled.
 855  * Return of -EINVAL means tbl8 is empty and thus can be recycled
 856  * Return of value > -1 means tbl8 is in use but has all the same values and
 857  * thus can be recycled
 858  */
 859 static int32_t
 860 tbl8_recycle_check(struct rte_lpm_tbl_entry *tbl8,
 861                 uint32_t tbl8_group_start)
 862 {
 863         uint32_t tbl8_group_end, i;
 864         tbl8_group_end = tbl8_group_start + RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 865
 866         /*
 867          * Check the first entry of the given tbl8. If it is invalid we know
 868          * this tbl8 does not contain any rule with a depth < RTE_LPM_MAX_DEPTH
 869          *  (As they would affect all entries in a tbl8) and thus this table
 870          *  can not be recycled.
 871          */
 872         if (tbl8[tbl8_group_start].valid) {
 873                 /*
 874                  * If first entry is valid check if the depth is less than 24
 875                  * and if so check the rest of the entries to verify that they
 876                  * are all of this depth.
 877                  */
 878                 if (tbl8[tbl8_group_start].depth <= MAX_DEPTH_TBL24) {
 879                         for (i = (tbl8_group_start + 1); i < tbl8_group_end;
 880                                         i++) {
 881
 882                                 if (tbl8[i].depth !=
 883                                                 tbl8[tbl8_group_start].depth) {
 884
 885                                         return -EEXIST;
 886                                 }
 887                         }
 888                         /* If all entries are the same return the tb8 index */
 889                         return tbl8_group_start;
 890                 }
 891
 892                 return -EEXIST;
 893         }
 894         /*
 895          * If the first entry is invalid check if the rest of the entries in
 896          * the tbl8 are invalid.
 897          */
 898         for (i = (tbl8_group_start + 1); i < tbl8_group_end; i++) {
 899                 if (tbl8[i].valid)
 900                         return -EEXIST;
 901         }
 902         /* If no valid entries are found then return -EINVAL. */
 903         return -EINVAL;
 904 }
 905
 906 static int32_t
 907 delete_depth_big(struct rte_lpm *lpm, uint32_t ip_masked,
 908         uint8_t depth, int32_t sub_rule_index, uint8_t sub_rule_depth)
 909 {
 910 #define group_idx next_hop
 911         uint32_t tbl24_index, tbl8_group_index, tbl8_group_start, tbl8_index,
 912                         tbl8_range, i;
 913         int32_t tbl8_recycle_index;
 914
 915         /*
 916          * Calculate the index into tbl24 and range. Note: All depths larger
 917          * than MAX_DEPTH_TBL24 are associated with only one tbl24 entry.
 918          */
 919         tbl24_index = ip_masked >> 8;
 920
 921         /* Calculate the index into tbl8 and range. */
 922         tbl8_group_index = lpm->tbl24[tbl24_index].group_idx;
 923         tbl8_group_start = tbl8_group_index * RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
 924         tbl8_index = tbl8_group_start + (ip_masked & 0xFF);
 925         tbl8_range = depth_to_range(depth);
 926
 927         if (sub_rule_index < 0) {
 928                 /*
 929                  * Loop through the range of entries on tbl8 for which the
 930                  * rule_to_delete must be removed or modified.
 931                  */
 932                 for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
 933                         if (lpm->tbl8[i].depth <= depth)
 934                                 lpm->tbl8[i].valid = INVALID;
 935                 }
 936         } else {
 937                 /* Set new tbl8 entry. */
 938                 struct rte_lpm_tbl_entry new_tbl8_entry = {
 939                         .valid = VALID,
 940                         .depth = sub_rule_depth,
 941                         .valid_group = lpm->tbl8[tbl8_group_start].valid_group,
 942                         .next_hop = lpm->rules_tbl[sub_rule_index].next_hop,
 943                 };
 944
 945                 /*
 946                  * Loop through the range of entries on tbl8 for which the
 947                  * rule_to_delete must be modified.
 948                  */
 949                 for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
 950                         if (lpm->tbl8[i].depth <= depth)
 951                                 __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
 952                                                 __ATOMIC_RELAXED);
 953                 }
 954         }
 955
 956         /*
 957          * Check if there are any valid entries in this tbl8 group. If all
 958          * tbl8 entries are invalid we can free the tbl8 and invalidate the
 959          * associated tbl24 entry.
 960          */
 961
 962         tbl8_recycle_index = tbl8_recycle_check(lpm->tbl8, tbl8_group_start);
 963
 964         if (tbl8_recycle_index == -EINVAL) {
 965                 /* Set tbl24 before freeing tbl8 to avoid race condition.
 966                  * Prevent the free of the tbl8 group from hoisting.
 967                  */
 968                 lpm->tbl24[tbl24_index].valid = 0;
 969                 __atomic_thread_fence(__ATOMIC_RELEASE);
 970                 tbl8_free(lpm->tbl8, tbl8_group_start);
 971         } else if (tbl8_recycle_index > -1) {
 972                 /* Update tbl24 entry. */
 973                 struct rte_lpm_tbl_entry new_tbl24_entry = {
 974                         .next_hop = lpm->tbl8[tbl8_recycle_index].next_hop,
 975                         .valid = VALID,
 976                         .valid_group = 0,
 977                         .depth = lpm->tbl8[tbl8_recycle_index].depth,
 978                 };
 979
 980                 /* Set tbl24 before freeing tbl8 to avoid race condition.
 981                  * Prevent the free of the tbl8 group from hoisting.
 982                  */
 983                 __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
 984                                 __ATOMIC_RELAXED);
 985                 __atomic_thread_fence(__ATOMIC_RELEASE);
 986                 tbl8_free(lpm->tbl8, tbl8_group_start);
 987         }
 988 #undef group_idx
 989         return 0;
 990 }
 991
 992 /*
 993  * Deletes a rule
 994  */
 995 int
 996 rte_lpm_delete(struct rte_lpm *lpm, uint32_t ip, uint8_t depth)
 997 {
 998         int32_t rule_to_delete_index, sub_rule_index;
 999         uint32_t ip_masked;
1000         uint8_t sub_rule_depth;
1001         /*
1002          * Check input arguments. Note: IP must be a positive integer of 32
1003          * bits in length therefore it need not be checked.
1004          */
1005         if ((lpm == NULL) || (depth < 1) || (depth > RTE_LPM_MAX_DEPTH)) {
1006                 return -EINVAL;
1007         }
1008
1009         ip_masked = ip & depth_to_mask(depth);
1010
1011         /*
1012          * Find the index of the input rule, that needs to be deleted, in the
1013          * rule table.
1014          */
1015         rule_to_delete_index = rule_find(lpm, ip_masked, depth);
1016
1017         /*
1018          * Check if rule_to_delete_index was found. If no rule was found the
1019          * function rule_find returns -EINVAL.
1020          */
1021         if (rule_to_delete_index < 0)
1022                 return -EINVAL;
1023
1024         /* Delete the rule from the rule table. */
1025         rule_delete(lpm, rule_to_delete_index, depth);
1026
1027         /*
1028          * Find rule to replace the rule_to_delete. If there is no rule to
1029          * replace the rule_to_delete we return -1 and invalidate the table
1030          * entries associated with this rule.
1031          */
1032         sub_rule_depth = 0;
1033         sub_rule_index = find_previous_rule(lpm, ip, depth, &sub_rule_depth);
1034
1035         /*
1036          * If the input depth value is less than 25 use function
1037          * delete_depth_small otherwise use delete_depth_big.
1038          */
1039         if (depth <= MAX_DEPTH_TBL24) {
1040                 return delete_depth_small(lpm, ip_masked, depth,
1041                                 sub_rule_index, sub_rule_depth);
1042         } else { /* If depth > MAX_DEPTH_TBL24 */
1043                 return delete_depth_big(lpm, ip_masked, depth, sub_rule_index,
1044                                 sub_rule_depth);
1045         }
1046 }
1047
1048 /*
1049  * Delete all rules from the LPM table.
1050  */
1051 void
1052 rte_lpm_delete_all(struct rte_lpm *lpm)
1053 {
1054         /* Zero rule information. */
1055         memset(lpm->rule_info, 0, sizeof(lpm->rule_info));
1056
1057         /* Zero tbl24. */
1058         memset(lpm->tbl24, 0, sizeof(lpm->tbl24));
1059
1060         /* Zero tbl8. */
1061         memset(lpm->tbl8, 0, sizeof(lpm->tbl8[0])
1062                         * RTE_LPM_TBL8_GROUP_NUM_ENTRIES * lpm->number_tbl8s);
1063
1064         /* Delete all rules form the rules table. */
1065         memset(lpm->rules_tbl, 0, sizeof(lpm->rules_tbl[0]) * lpm->max_rules);
1066 }