1 /* SPDX-License-Identifier: BSD-3-Clause
2 * Copyright(c) 2010-2014 Intel Corporation
12 #include <rte_branch_prediction.h>
13 #include <rte_common.h>
14 #include <rte_memory.h>
15 #include <rte_malloc.h>
16 #include <rte_memcpy.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>
26 #include <rte_jhash.h>
27 #include <rte_tailq.h>
31 #define RTE_LPM6_TBL24_NUM_ENTRIES (1 << 24)
32 #define RTE_LPM6_TBL8_GROUP_NUM_ENTRIES 256
33 #define RTE_LPM6_TBL8_MAX_NUM_GROUPS (1 << 21)
35 #define RTE_LPM6_VALID_EXT_ENTRY_BITMASK 0xA0000000
36 #define RTE_LPM6_LOOKUP_SUCCESS 0x20000000
37 #define RTE_LPM6_TBL8_BITMASK 0x001FFFFF
39 #define ADD_FIRST_BYTE 3
40 #define LOOKUP_FIRST_BYTE 4
42 #define BYTES2_SIZE 16
44 #define RULE_HASH_TABLE_EXTRA_SPACE 64
45 #define TBL24_IND UINT32_MAX
47 #define lpm6_tbl8_gindex next_hop
49 /** Flags for setting an entry as valid/invalid. */
55 TAILQ_HEAD(rte_lpm6_list, rte_tailq_entry);
57 static struct rte_tailq_elem rte_lpm6_tailq = {
60 EAL_REGISTER_TAILQ(rte_lpm6_tailq)
62 /** Tbl entry structure. It is the same for both tbl24 and tbl8 */
63 struct rte_lpm6_tbl_entry {
64 uint32_t next_hop: 21; /**< Next hop / next table to be checked. */
65 uint32_t depth :8; /**< Rule depth. */
68 uint32_t valid :1; /**< Validation flag. */
69 uint32_t valid_group :1; /**< Group validation flag. */
70 uint32_t ext_entry :1; /**< External entry. */
73 /** Rules tbl entry structure. */
74 struct rte_lpm6_rule {
75 uint8_t ip[RTE_LPM6_IPV6_ADDR_SIZE]; /**< Rule IP address. */
76 uint32_t next_hop; /**< Rule next hop. */
77 uint8_t depth; /**< Rule depth. */
80 /** Rules tbl entry key. */
81 struct rte_lpm6_rule_key {
82 uint8_t ip[RTE_LPM6_IPV6_ADDR_SIZE]; /**< Rule IP address. */
83 uint32_t depth; /**< Rule depth. */
87 struct rte_lpm_tbl8_hdr {
88 uint32_t owner_tbl_ind; /**< owner table: TBL24_IND if owner is tbl24,
89 * otherwise index of tbl8
91 uint32_t owner_entry_ind; /**< index of the owner table entry where
92 * pointer to the tbl8 is stored
94 uint32_t ref_cnt; /**< table reference counter */
97 /** LPM6 structure. */
100 char name[RTE_LPM6_NAMESIZE]; /**< Name of the lpm. */
101 uint32_t max_rules; /**< Max number of rules. */
102 uint32_t used_rules; /**< Used rules so far. */
103 uint32_t number_tbl8s; /**< Number of tbl8s to allocate. */
106 struct rte_hash *rules_tbl; /**< LPM rules. */
107 struct rte_lpm6_tbl_entry tbl24[RTE_LPM6_TBL24_NUM_ENTRIES]
108 __rte_cache_aligned; /**< LPM tbl24 table. */
110 uint32_t *tbl8_pool; /**< pool of indexes of free tbl8s */
111 uint32_t tbl8_pool_pos; /**< current position in the tbl8 pool */
113 struct rte_lpm_tbl8_hdr *tbl8_hdrs; /* array of tbl8 headers */
115 struct rte_lpm6_tbl_entry tbl8[0]
116 __rte_cache_aligned; /**< LPM tbl8 table. */
120 * Takes an array of uint8_t (IPv6 address) and masks it using the depth.
121 * It leaves untouched one bit per unit in the depth variable
122 * and set the rest to 0.
125 ip6_mask_addr(uint8_t *ip, uint8_t depth)
127 int16_t part_depth, mask;
132 for (i = 0; i < RTE_LPM6_IPV6_ADDR_SIZE; i++) {
133 if (part_depth < BYTE_SIZE && part_depth >= 0) {
134 mask = (uint16_t)(~(UINT8_MAX >> part_depth));
135 ip[i] = (uint8_t)(ip[i] & mask);
136 } else if (part_depth < 0)
139 part_depth -= BYTE_SIZE;
143 /* copy ipv6 address */
145 ip6_copy_addr(uint8_t *dst, const uint8_t *src)
147 rte_memcpy(dst, src, RTE_LPM6_IPV6_ADDR_SIZE);
151 * LPM6 rule hash function
153 * It's used as a hash function for the rte_hash
156 static inline uint32_t
157 rule_hash(const void *data, __rte_unused uint32_t data_len,
160 return rte_jhash(data, sizeof(struct rte_lpm6_rule_key), init_val);
164 * Init pool of free tbl8 indexes
167 tbl8_pool_init(struct rte_lpm6 *lpm)
171 /* put entire range of indexes to the tbl8 pool */
172 for (i = 0; i < lpm->number_tbl8s; i++)
173 lpm->tbl8_pool[i] = i;
175 lpm->tbl8_pool_pos = 0;
179 * Get an index of a free tbl8 from the pool
181 static inline uint32_t
182 tbl8_get(struct rte_lpm6 *lpm, uint32_t *tbl8_ind)
184 if (lpm->tbl8_pool_pos == lpm->number_tbl8s)
185 /* no more free tbl8 */
189 *tbl8_ind = lpm->tbl8_pool[lpm->tbl8_pool_pos++];
194 * Put an index of a free tbl8 back to the pool
196 static inline uint32_t
197 tbl8_put(struct rte_lpm6 *lpm, uint32_t tbl8_ind)
199 if (lpm->tbl8_pool_pos == 0)
203 lpm->tbl8_pool[--lpm->tbl8_pool_pos] = tbl8_ind;
208 * Returns number of tbl8s available in the pool
210 static inline uint32_t
211 tbl8_available(struct rte_lpm6 *lpm)
213 return lpm->number_tbl8s - lpm->tbl8_pool_pos;
218 * note that ip must be already masked
221 rule_key_init(struct rte_lpm6_rule_key *key, uint8_t *ip, uint8_t depth)
223 ip6_copy_addr(key->ip, ip);
228 * Rebuild the entire LPM tree by reinserting all rules
231 rebuild_lpm(struct rte_lpm6 *lpm)
234 struct rte_lpm6_rule_key *rule_key;
237 while (rte_hash_iterate(lpm->rules_tbl, (void *) &rule_key,
238 (void **) &next_hop, &iter) >= 0)
239 rte_lpm6_add(lpm, rule_key->ip, rule_key->depth,
240 (uint32_t) next_hop);
244 * Allocates memory for LPM object
247 rte_lpm6_create(const char *name, int socket_id,
248 const struct rte_lpm6_config *config)
250 char mem_name[RTE_LPM6_NAMESIZE];
251 struct rte_lpm6 *lpm = NULL;
252 struct rte_tailq_entry *te;
254 struct rte_lpm6_list *lpm_list;
255 struct rte_hash *rules_tbl = NULL;
256 uint32_t *tbl8_pool = NULL;
257 struct rte_lpm_tbl8_hdr *tbl8_hdrs = NULL;
259 lpm_list = RTE_TAILQ_CAST(rte_lpm6_tailq.head, rte_lpm6_list);
261 RTE_BUILD_BUG_ON(sizeof(struct rte_lpm6_tbl_entry) != sizeof(uint32_t));
262 RTE_BUILD_BUG_ON(sizeof(struct rte_lpm6_rule_key) %
263 sizeof(uint32_t) != 0);
265 /* Check user arguments. */
266 if ((name == NULL) || (socket_id < -1) || (config == NULL) ||
267 (config->max_rules == 0) ||
268 config->number_tbl8s > RTE_LPM6_TBL8_MAX_NUM_GROUPS) {
273 /* create rules hash table */
274 snprintf(mem_name, sizeof(mem_name), "LRH_%s", name);
275 struct rte_hash_parameters rule_hash_tbl_params = {
276 .entries = config->max_rules * 1.2 +
277 RULE_HASH_TABLE_EXTRA_SPACE,
278 .key_len = sizeof(struct rte_lpm6_rule_key),
279 .hash_func = rule_hash,
280 .hash_func_init_val = 0,
283 .socket_id = socket_id,
287 rules_tbl = rte_hash_create(&rule_hash_tbl_params);
288 if (rules_tbl == NULL) {
289 RTE_LOG(ERR, LPM, "LPM rules hash table allocation failed: %s (%d)",
290 rte_strerror(rte_errno), rte_errno);
294 /* allocate tbl8 indexes pool */
295 tbl8_pool = rte_malloc(NULL,
296 sizeof(uint32_t) * config->number_tbl8s,
297 RTE_CACHE_LINE_SIZE);
298 if (tbl8_pool == NULL) {
299 RTE_LOG(ERR, LPM, "LPM tbl8 pool allocation failed: %s (%d)",
300 rte_strerror(rte_errno), rte_errno);
305 /* allocate tbl8 headers */
306 tbl8_hdrs = rte_malloc(NULL,
307 sizeof(struct rte_lpm_tbl8_hdr) * config->number_tbl8s,
308 RTE_CACHE_LINE_SIZE);
309 if (tbl8_hdrs == NULL) {
310 RTE_LOG(ERR, LPM, "LPM tbl8 headers allocation failed: %s (%d)",
311 rte_strerror(rte_errno), rte_errno);
316 snprintf(mem_name, sizeof(mem_name), "LPM_%s", name);
318 /* Determine the amount of memory to allocate. */
319 mem_size = sizeof(*lpm) + (sizeof(lpm->tbl8[0]) *
320 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES * config->number_tbl8s);
322 rte_mcfg_tailq_write_lock();
324 /* Guarantee there's no existing */
325 TAILQ_FOREACH(te, lpm_list, next) {
326 lpm = (struct rte_lpm6 *) te->data;
327 if (strncmp(name, lpm->name, RTE_LPM6_NAMESIZE) == 0)
336 /* allocate tailq entry */
337 te = rte_zmalloc("LPM6_TAILQ_ENTRY", sizeof(*te), 0);
339 RTE_LOG(ERR, LPM, "Failed to allocate tailq entry!\n");
344 /* Allocate memory to store the LPM data structures. */
345 lpm = rte_zmalloc_socket(mem_name, (size_t)mem_size,
346 RTE_CACHE_LINE_SIZE, socket_id);
349 RTE_LOG(ERR, LPM, "LPM memory allocation failed\n");
355 /* Save user arguments. */
356 lpm->max_rules = config->max_rules;
357 lpm->number_tbl8s = config->number_tbl8s;
358 strlcpy(lpm->name, name, sizeof(lpm->name));
359 lpm->rules_tbl = rules_tbl;
360 lpm->tbl8_pool = tbl8_pool;
361 lpm->tbl8_hdrs = tbl8_hdrs;
366 te->data = (void *) lpm;
368 TAILQ_INSERT_TAIL(lpm_list, te, next);
369 rte_mcfg_tailq_write_unlock();
373 rte_mcfg_tailq_write_unlock();
378 rte_hash_free(rules_tbl);
384 * Find an existing lpm table and return a pointer to it.
387 rte_lpm6_find_existing(const char *name)
389 struct rte_lpm6 *l = NULL;
390 struct rte_tailq_entry *te;
391 struct rte_lpm6_list *lpm_list;
393 lpm_list = RTE_TAILQ_CAST(rte_lpm6_tailq.head, rte_lpm6_list);
395 rte_mcfg_tailq_read_lock();
396 TAILQ_FOREACH(te, lpm_list, next) {
397 l = (struct rte_lpm6 *) te->data;
398 if (strncmp(name, l->name, RTE_LPM6_NAMESIZE) == 0)
401 rte_mcfg_tailq_read_unlock();
412 * Deallocates memory for given LPM table.
415 rte_lpm6_free(struct rte_lpm6 *lpm)
417 struct rte_lpm6_list *lpm_list;
418 struct rte_tailq_entry *te;
420 /* Check user arguments. */
424 lpm_list = RTE_TAILQ_CAST(rte_lpm6_tailq.head, rte_lpm6_list);
426 rte_mcfg_tailq_write_lock();
428 /* find our tailq entry */
429 TAILQ_FOREACH(te, lpm_list, next) {
430 if (te->data == (void *) lpm)
435 TAILQ_REMOVE(lpm_list, te, next);
437 rte_mcfg_tailq_write_unlock();
439 rte_free(lpm->tbl8_hdrs);
440 rte_free(lpm->tbl8_pool);
441 rte_hash_free(lpm->rules_tbl);
448 rule_find_with_key(struct rte_lpm6 *lpm,
449 const struct rte_lpm6_rule_key *rule_key,
455 /* lookup for a rule */
456 ret = rte_hash_lookup_data(lpm->rules_tbl, (const void *) rule_key,
457 (void **) &hash_val);
459 *next_hop = (uint32_t) hash_val;
468 rule_find(struct rte_lpm6 *lpm, uint8_t *ip, uint8_t depth,
471 struct rte_lpm6_rule_key rule_key;
473 /* init a rule key */
474 rule_key_init(&rule_key, ip, depth);
476 return rule_find_with_key(lpm, &rule_key, next_hop);
480 * Checks if a rule already exists in the rules table and updates
481 * the nexthop if so. Otherwise it adds a new rule if enough space is available.
484 * 0 - next hop of existed rule is updated
485 * 1 - new rule successfully added
489 rule_add(struct rte_lpm6 *lpm, uint8_t *ip, uint8_t depth, uint32_t next_hop)
492 struct rte_lpm6_rule_key rule_key;
495 /* init a rule key */
496 rule_key_init(&rule_key, ip, depth);
498 /* Scan through rule list to see if rule already exists. */
499 rule_exist = rule_find_with_key(lpm, &rule_key, &unused);
502 * If rule does not exist check if there is space to add a new rule to
503 * this rule group. If there is no space return error.
505 if (!rule_exist && lpm->used_rules == lpm->max_rules)
508 /* add the rule or update rules next hop */
509 ret = rte_hash_add_key_data(lpm->rules_tbl, &rule_key,
510 (void *)(uintptr_t) next_hop);
514 /* Increment the used rules counter for this rule group. */
524 * Function that expands a rule across the data structure when a less-generic
525 * one has been added before. It assures that every possible combination of bits
526 * in the IP address returns a match.
529 expand_rule(struct rte_lpm6 *lpm, uint32_t tbl8_gindex, uint8_t old_depth,
530 uint8_t new_depth, uint32_t next_hop, uint8_t valid)
532 uint32_t tbl8_group_end, tbl8_gindex_next, j;
534 tbl8_group_end = tbl8_gindex + RTE_LPM6_TBL8_GROUP_NUM_ENTRIES;
536 struct rte_lpm6_tbl_entry new_tbl8_entry = {
538 .valid_group = valid,
540 .next_hop = next_hop,
544 for (j = tbl8_gindex; j < tbl8_group_end; j++) {
545 if (!lpm->tbl8[j].valid || (lpm->tbl8[j].ext_entry == 0
546 && lpm->tbl8[j].depth <= old_depth)) {
548 lpm->tbl8[j] = new_tbl8_entry;
550 } else if (lpm->tbl8[j].ext_entry == 1) {
552 tbl8_gindex_next = lpm->tbl8[j].lpm6_tbl8_gindex
553 * RTE_LPM6_TBL8_GROUP_NUM_ENTRIES;
554 expand_rule(lpm, tbl8_gindex_next, old_depth, new_depth,
564 init_tbl8_header(struct rte_lpm6 *lpm, uint32_t tbl_ind,
565 uint32_t owner_tbl_ind, uint32_t owner_entry_ind)
567 struct rte_lpm_tbl8_hdr *tbl_hdr = &lpm->tbl8_hdrs[tbl_ind];
568 tbl_hdr->owner_tbl_ind = owner_tbl_ind;
569 tbl_hdr->owner_entry_ind = owner_entry_ind;
570 tbl_hdr->ref_cnt = 0;
574 * Calculate index to the table based on the number and position
575 * of the bytes being inspected in this step.
578 get_bitshift(const uint8_t *ip, uint8_t first_byte, uint8_t bytes)
580 uint32_t entry_ind, i;
584 for (i = first_byte; i < (uint32_t)(first_byte + bytes); i++) {
585 bitshift = (int8_t)((bytes - i)*BYTE_SIZE);
589 entry_ind = entry_ind | ip[i-1] << bitshift;
596 * Simulate adding a new route to the LPM counting number
597 * of new tables that will be needed
599 * It returns 0 on success, or 1 if
600 * the process needs to be continued by calling the function again.
603 simulate_add_step(struct rte_lpm6 *lpm, struct rte_lpm6_tbl_entry *tbl,
604 struct rte_lpm6_tbl_entry **next_tbl, const uint8_t *ip,
605 uint8_t bytes, uint8_t first_byte, uint8_t depth,
606 uint32_t *need_tbl_nb)
609 uint8_t bits_covered;
610 uint32_t next_tbl_ind;
613 * Calculate index to the table based on the number and position
614 * of the bytes being inspected in this step.
616 entry_ind = get_bitshift(ip, first_byte, bytes);
618 /* Number of bits covered in this step */
619 bits_covered = (uint8_t)((bytes+first_byte-1)*BYTE_SIZE);
621 if (depth <= bits_covered) {
626 if (tbl[entry_ind].valid == 0 || tbl[entry_ind].ext_entry == 0) {
627 /* from this point on a new table is needed on each level
628 * that is not covered yet
630 depth -= bits_covered;
631 uint32_t cnt = depth >> 3; /* depth / BYTE_SIZE */
632 if (depth & 7) /* 0b00000111 */
633 /* if depth % 8 > 0 then one more table is needed
634 * for those last bits
642 next_tbl_ind = tbl[entry_ind].lpm6_tbl8_gindex;
643 *next_tbl = &(lpm->tbl8[next_tbl_ind *
644 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES]);
650 * Partially adds a new route to the data structure (tbl24+tbl8s).
651 * It returns 0 on success, a negative number on failure, or 1 if
652 * the process needs to be continued by calling the function again.
655 add_step(struct rte_lpm6 *lpm, struct rte_lpm6_tbl_entry *tbl,
656 uint32_t tbl_ind, struct rte_lpm6_tbl_entry **next_tbl,
657 uint32_t *next_tbl_ind, uint8_t *ip, uint8_t bytes,
658 uint8_t first_byte, uint8_t depth, uint32_t next_hop,
661 uint32_t entry_ind, tbl_range, tbl8_group_start, tbl8_group_end, i;
662 uint32_t tbl8_gindex;
663 uint8_t bits_covered;
667 * Calculate index to the table based on the number and position
668 * of the bytes being inspected in this step.
670 entry_ind = get_bitshift(ip, first_byte, bytes);
672 /* Number of bits covered in this step */
673 bits_covered = (uint8_t)((bytes+first_byte-1)*BYTE_SIZE);
676 * If depth if smaller than this number (ie this is the last step)
677 * expand the rule across the relevant positions in the table.
679 if (depth <= bits_covered) {
680 tbl_range = 1 << (bits_covered - depth);
682 for (i = entry_ind; i < (entry_ind + tbl_range); i++) {
683 if (!tbl[i].valid || (tbl[i].ext_entry == 0 &&
684 tbl[i].depth <= depth)) {
686 struct rte_lpm6_tbl_entry new_tbl_entry = {
687 .next_hop = next_hop,
690 .valid_group = VALID,
694 tbl[i] = new_tbl_entry;
696 } else if (tbl[i].ext_entry == 1) {
699 * If tbl entry is valid and extended calculate the index
700 * into next tbl8 and expand the rule across the data structure.
702 tbl8_gindex = tbl[i].lpm6_tbl8_gindex *
703 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES;
704 expand_rule(lpm, tbl8_gindex, depth, depth,
709 /* update tbl8 rule reference counter */
710 if (tbl_ind != TBL24_IND && is_new_rule)
711 lpm->tbl8_hdrs[tbl_ind].ref_cnt++;
716 * If this is not the last step just fill one position
717 * and calculate the index to the next table.
720 /* If it's invalid a new tbl8 is needed */
721 if (!tbl[entry_ind].valid) {
722 /* get a new table */
723 ret = tbl8_get(lpm, &tbl8_gindex);
727 /* invalidate all new tbl8 entries */
728 tbl8_group_start = tbl8_gindex *
729 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES;
730 memset(&lpm->tbl8[tbl8_group_start], 0,
731 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES *
732 sizeof(struct rte_lpm6_tbl_entry));
734 /* init the new table's header:
735 * save the reference to the owner table
737 init_tbl8_header(lpm, tbl8_gindex, tbl_ind, entry_ind);
739 /* reference to a new tbl8 */
740 struct rte_lpm6_tbl_entry new_tbl_entry = {
741 .lpm6_tbl8_gindex = tbl8_gindex,
744 .valid_group = VALID,
748 tbl[entry_ind] = new_tbl_entry;
750 /* update the current table's reference counter */
751 if (tbl_ind != TBL24_IND)
752 lpm->tbl8_hdrs[tbl_ind].ref_cnt++;
755 * If it's valid but not extended the rule that was stored
756 * here needs to be moved to the next table.
758 else if (tbl[entry_ind].ext_entry == 0) {
759 /* get a new tbl8 index */
760 ret = tbl8_get(lpm, &tbl8_gindex);
764 tbl8_group_start = tbl8_gindex *
765 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES;
766 tbl8_group_end = tbl8_group_start +
767 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES;
769 struct rte_lpm6_tbl_entry tbl_entry = {
770 .next_hop = tbl[entry_ind].next_hop,
771 .depth = tbl[entry_ind].depth,
773 .valid_group = VALID,
777 /* Populate new tbl8 with tbl value. */
778 for (i = tbl8_group_start; i < tbl8_group_end; i++)
779 lpm->tbl8[i] = tbl_entry;
781 /* init the new table's header:
782 * save the reference to the owner table
784 init_tbl8_header(lpm, tbl8_gindex, tbl_ind, entry_ind);
787 * Update tbl entry to point to new tbl8 entry. Note: The
788 * ext_flag and tbl8_index need to be updated simultaneously,
789 * so assign whole structure in one go.
791 struct rte_lpm6_tbl_entry new_tbl_entry = {
792 .lpm6_tbl8_gindex = tbl8_gindex,
795 .valid_group = VALID,
799 tbl[entry_ind] = new_tbl_entry;
801 /* update the current table's reference counter */
802 if (tbl_ind != TBL24_IND)
803 lpm->tbl8_hdrs[tbl_ind].ref_cnt++;
806 *next_tbl_ind = tbl[entry_ind].lpm6_tbl8_gindex;
807 *next_tbl = &(lpm->tbl8[*next_tbl_ind *
808 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES]);
815 * Simulate adding a route to LPM
819 * -ENOSPC not enough tbl8 left
822 simulate_add(struct rte_lpm6 *lpm, const uint8_t *masked_ip, uint8_t depth)
824 struct rte_lpm6_tbl_entry *tbl;
825 struct rte_lpm6_tbl_entry *tbl_next = NULL;
828 /* number of new tables needed for a step */
829 uint32_t need_tbl_nb;
830 /* total number of new tables needed */
831 uint32_t total_need_tbl_nb;
833 /* Inspect the first three bytes through tbl24 on the first step. */
834 ret = simulate_add_step(lpm, lpm->tbl24, &tbl_next, masked_ip,
835 ADD_FIRST_BYTE, 1, depth, &need_tbl_nb);
836 total_need_tbl_nb = need_tbl_nb;
838 * Inspect one by one the rest of the bytes until
839 * the process is completed.
841 for (i = ADD_FIRST_BYTE; i < RTE_LPM6_IPV6_ADDR_SIZE && ret == 1; i++) {
843 ret = simulate_add_step(lpm, tbl, &tbl_next, masked_ip, 1,
844 (uint8_t)(i + 1), depth, &need_tbl_nb);
845 total_need_tbl_nb += need_tbl_nb;
848 if (tbl8_available(lpm) < total_need_tbl_nb)
849 /* not enough tbl8 to add a rule */
859 rte_lpm6_add(struct rte_lpm6 *lpm, const uint8_t *ip, uint8_t depth,
862 struct rte_lpm6_tbl_entry *tbl;
863 struct rte_lpm6_tbl_entry *tbl_next = NULL;
864 /* init to avoid compiler warning */
865 uint32_t tbl_next_num = 123456;
867 uint8_t masked_ip[RTE_LPM6_IPV6_ADDR_SIZE];
870 /* Check user arguments. */
871 if ((lpm == NULL) || (depth < 1) || (depth > RTE_LPM6_MAX_DEPTH))
874 /* Copy the IP and mask it to avoid modifying user's input data. */
875 ip6_copy_addr(masked_ip, ip);
876 ip6_mask_addr(masked_ip, depth);
878 /* Simulate adding a new route */
879 int ret = simulate_add(lpm, masked_ip, depth);
883 /* Add the rule to the rule table. */
884 int is_new_rule = rule_add(lpm, masked_ip, depth, next_hop);
885 /* If there is no space available for new rule return error. */
889 /* Inspect the first three bytes through tbl24 on the first step. */
891 status = add_step(lpm, tbl, TBL24_IND, &tbl_next, &tbl_next_num,
892 masked_ip, ADD_FIRST_BYTE, 1, depth, next_hop,
897 * Inspect one by one the rest of the bytes until
898 * the process is completed.
900 for (i = ADD_FIRST_BYTE; i < RTE_LPM6_IPV6_ADDR_SIZE && status == 1; i++) {
902 status = add_step(lpm, tbl, tbl_next_num, &tbl_next,
903 &tbl_next_num, masked_ip, 1, (uint8_t)(i + 1),
904 depth, next_hop, is_new_rule);
912 * Takes a pointer to a table entry and inspect one level.
913 * The function returns 0 on lookup success, ENOENT if no match was found
914 * or 1 if the process needs to be continued by calling the function again.
917 lookup_step(const struct rte_lpm6 *lpm, const struct rte_lpm6_tbl_entry *tbl,
918 const struct rte_lpm6_tbl_entry **tbl_next, const uint8_t *ip,
919 uint8_t first_byte, uint32_t *next_hop)
921 uint32_t tbl8_index, tbl_entry;
923 /* Take the integer value from the pointer. */
924 tbl_entry = *(const uint32_t *)tbl;
926 /* If it is valid and extended we calculate the new pointer to return. */
927 if ((tbl_entry & RTE_LPM6_VALID_EXT_ENTRY_BITMASK) ==
928 RTE_LPM6_VALID_EXT_ENTRY_BITMASK) {
930 tbl8_index = ip[first_byte-1] +
931 ((tbl_entry & RTE_LPM6_TBL8_BITMASK) *
932 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES);
934 *tbl_next = &lpm->tbl8[tbl8_index];
938 /* If not extended then we can have a match. */
939 *next_hop = ((uint32_t)tbl_entry & RTE_LPM6_TBL8_BITMASK);
940 return (tbl_entry & RTE_LPM6_LOOKUP_SUCCESS) ? 0 : -ENOENT;
948 rte_lpm6_lookup(const struct rte_lpm6 *lpm, const uint8_t *ip,
951 const struct rte_lpm6_tbl_entry *tbl;
952 const struct rte_lpm6_tbl_entry *tbl_next = NULL;
955 uint32_t tbl24_index;
957 /* DEBUG: Check user input arguments. */
958 if ((lpm == NULL) || (ip == NULL) || (next_hop == NULL))
961 first_byte = LOOKUP_FIRST_BYTE;
962 tbl24_index = (ip[0] << BYTES2_SIZE) | (ip[1] << BYTE_SIZE) | ip[2];
964 /* Calculate pointer to the first entry to be inspected */
965 tbl = &lpm->tbl24[tbl24_index];
968 /* Continue inspecting following levels until success or failure */
969 status = lookup_step(lpm, tbl, &tbl_next, ip, first_byte++, next_hop);
971 } while (status == 1);
977 * Looks up a group of IP addresses
980 rte_lpm6_lookup_bulk_func(const struct rte_lpm6 *lpm,
981 uint8_t ips[][RTE_LPM6_IPV6_ADDR_SIZE],
982 int32_t *next_hops, unsigned int n)
985 const struct rte_lpm6_tbl_entry *tbl;
986 const struct rte_lpm6_tbl_entry *tbl_next = NULL;
987 uint32_t tbl24_index, next_hop;
991 /* DEBUG: Check user input arguments. */
992 if ((lpm == NULL) || (ips == NULL) || (next_hops == NULL))
995 for (i = 0; i < n; i++) {
996 first_byte = LOOKUP_FIRST_BYTE;
997 tbl24_index = (ips[i][0] << BYTES2_SIZE) |
998 (ips[i][1] << BYTE_SIZE) | ips[i][2];
1000 /* Calculate pointer to the first entry to be inspected */
1001 tbl = &lpm->tbl24[tbl24_index];
1004 /* Continue inspecting following levels
1005 * until success or failure
1007 status = lookup_step(lpm, tbl, &tbl_next, ips[i],
1008 first_byte++, &next_hop);
1010 } while (status == 1);
1015 next_hops[i] = (int32_t)next_hop;
1022 * Look for a rule in the high-level rules table
1025 rte_lpm6_is_rule_present(struct rte_lpm6 *lpm, const uint8_t *ip, uint8_t depth,
1028 uint8_t masked_ip[RTE_LPM6_IPV6_ADDR_SIZE];
1030 /* Check user arguments. */
1031 if ((lpm == NULL) || next_hop == NULL || ip == NULL ||
1032 (depth < 1) || (depth > RTE_LPM6_MAX_DEPTH))
1035 /* Copy the IP and mask it to avoid modifying user's input data. */
1036 ip6_copy_addr(masked_ip, ip);
1037 ip6_mask_addr(masked_ip, depth);
1039 return rule_find(lpm, masked_ip, depth, next_hop);
1043 * Delete a rule from the rule table.
1044 * NOTE: Valid range for depth parameter is 1 .. 128 inclusive.
1050 rule_delete(struct rte_lpm6 *lpm, uint8_t *ip, uint8_t depth)
1053 struct rte_lpm6_rule_key rule_key;
1056 rule_key_init(&rule_key, ip, depth);
1058 /* delete the rule */
1059 ret = rte_hash_del_key(lpm->rules_tbl, (void *) &rule_key);
1067 * Deletes a group of rules
1069 * Note that the function rebuilds the lpm table,
1070 * rather than doing incremental updates like
1071 * the regular delete function
1074 rte_lpm6_delete_bulk_func(struct rte_lpm6 *lpm,
1075 uint8_t ips[][RTE_LPM6_IPV6_ADDR_SIZE], uint8_t *depths,
1078 uint8_t masked_ip[RTE_LPM6_IPV6_ADDR_SIZE];
1081 /* Check input arguments. */
1082 if ((lpm == NULL) || (ips == NULL) || (depths == NULL))
1085 for (i = 0; i < n; i++) {
1086 ip6_copy_addr(masked_ip, ips[i]);
1087 ip6_mask_addr(masked_ip, depths[i]);
1088 rule_delete(lpm, masked_ip, depths[i]);
1092 * Set all the table entries to 0 (ie delete every rule
1093 * from the data structure.
1095 memset(lpm->tbl24, 0, sizeof(lpm->tbl24));
1096 memset(lpm->tbl8, 0, sizeof(lpm->tbl8[0])
1097 * RTE_LPM6_TBL8_GROUP_NUM_ENTRIES * lpm->number_tbl8s);
1098 tbl8_pool_init(lpm);
1101 * Add every rule again (except for the ones that were removed from
1110 * Delete all rules from the LPM table.
1113 rte_lpm6_delete_all(struct rte_lpm6 *lpm)
1115 /* Zero used rules counter. */
1116 lpm->used_rules = 0;
1119 memset(lpm->tbl24, 0, sizeof(lpm->tbl24));
1122 memset(lpm->tbl8, 0, sizeof(lpm->tbl8[0]) *
1123 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES * lpm->number_tbl8s);
1125 /* init pool of free tbl8 indexes */
1126 tbl8_pool_init(lpm);
1128 /* Delete all rules form the rules table. */
1129 rte_hash_reset(lpm->rules_tbl);
1133 * Convert a depth to a one byte long mask
1134 * Example: 4 will be converted to 0xF0
1136 static uint8_t __attribute__((pure))
1137 depth_to_mask_1b(uint8_t depth)
1139 /* To calculate a mask start with a 1 on the left hand side and right
1140 * shift while populating the left hand side with 1's
1142 return (signed char)0x80 >> (depth - 1);
1146 * Find a less specific rule
1149 rule_find_less_specific(struct rte_lpm6 *lpm, uint8_t *ip, uint8_t depth,
1150 struct rte_lpm6_rule *rule)
1155 struct rte_lpm6_rule_key rule_key;
1160 rule_key_init(&rule_key, ip, depth);
1165 /* each iteration zero one more bit of the key */
1166 mask = depth & 7; /* depth % BYTE_SIZE */
1168 mask = depth_to_mask_1b(mask);
1170 rule_key.depth = depth;
1171 rule_key.ip[depth >> 3] &= mask;
1173 ret = rule_find_with_key(lpm, &rule_key, &next_hop);
1175 rule->depth = depth;
1176 ip6_copy_addr(rule->ip, rule_key.ip);
1177 rule->next_hop = next_hop;
1186 * Find range of tbl8 cells occupied by a rule
1189 rule_find_range(struct rte_lpm6 *lpm, const uint8_t *ip, uint8_t depth,
1190 struct rte_lpm6_tbl_entry **from,
1191 struct rte_lpm6_tbl_entry **to,
1192 uint32_t *out_tbl_ind)
1195 uint32_t first_3bytes = (uint32_t)ip[0] << 16 | ip[1] << 8 | ip[2];
1198 /* rule is within the top level */
1200 *from = &lpm->tbl24[ind];
1201 ind += (1 << (24 - depth)) - 1;
1202 *to = &lpm->tbl24[ind];
1203 *out_tbl_ind = TBL24_IND;
1205 /* top level entry */
1206 struct rte_lpm6_tbl_entry *tbl = &lpm->tbl24[first_3bytes];
1207 assert(tbl->ext_entry == 1);
1209 uint32_t tbl_ind = tbl->lpm6_tbl8_gindex;
1210 tbl = &lpm->tbl8[tbl_ind *
1211 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES];
1212 /* current ip byte, the top level is already behind */
1214 /* minus top level */
1217 /* iterate through levels (tbl8s)
1218 * until we reach the last one
1222 assert(tbl->ext_entry == 1);
1223 /* go to the next level/tbl8 */
1224 tbl_ind = tbl->lpm6_tbl8_gindex;
1225 tbl = &lpm->tbl8[tbl_ind *
1226 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES];
1231 /* last level/tbl8 */
1232 ind = ip[byte] & depth_to_mask_1b(depth);
1234 ind += (1 << (8 - depth)) - 1;
1236 *out_tbl_ind = tbl_ind;
1241 * Remove a table from the LPM tree
1244 remove_tbl(struct rte_lpm6 *lpm, struct rte_lpm_tbl8_hdr *tbl_hdr,
1245 uint32_t tbl_ind, struct rte_lpm6_rule *lsp_rule)
1247 struct rte_lpm6_tbl_entry *owner_entry;
1249 if (tbl_hdr->owner_tbl_ind == TBL24_IND)
1250 owner_entry = &lpm->tbl24[tbl_hdr->owner_entry_ind];
1252 uint32_t owner_tbl_ind = tbl_hdr->owner_tbl_ind;
1253 owner_entry = &lpm->tbl8[
1254 owner_tbl_ind * RTE_LPM6_TBL8_GROUP_NUM_ENTRIES +
1255 tbl_hdr->owner_entry_ind];
1257 struct rte_lpm_tbl8_hdr *owner_tbl_hdr =
1258 &lpm->tbl8_hdrs[owner_tbl_ind];
1259 if (--owner_tbl_hdr->ref_cnt == 0)
1260 remove_tbl(lpm, owner_tbl_hdr, owner_tbl_ind, lsp_rule);
1263 assert(owner_entry->ext_entry == 1);
1265 /* unlink the table */
1266 if (lsp_rule != NULL) {
1267 struct rte_lpm6_tbl_entry new_tbl_entry = {
1268 .next_hop = lsp_rule->next_hop,
1269 .depth = lsp_rule->depth,
1271 .valid_group = VALID,
1275 *owner_entry = new_tbl_entry;
1277 struct rte_lpm6_tbl_entry new_tbl_entry = {
1281 .valid_group = INVALID,
1285 *owner_entry = new_tbl_entry;
1288 /* return the table to the pool */
1289 tbl8_put(lpm, tbl_ind);
1296 rte_lpm6_delete(struct rte_lpm6 *lpm, const uint8_t *ip, uint8_t depth)
1298 uint8_t masked_ip[RTE_LPM6_IPV6_ADDR_SIZE];
1299 struct rte_lpm6_rule lsp_rule_obj;
1300 struct rte_lpm6_rule *lsp_rule;
1303 struct rte_lpm6_tbl_entry *from, *to;
1305 /* Check input arguments. */
1306 if ((lpm == NULL) || (depth < 1) || (depth > RTE_LPM6_MAX_DEPTH))
1309 /* Copy the IP and mask it to avoid modifying user's input data. */
1310 ip6_copy_addr(masked_ip, ip);
1311 ip6_mask_addr(masked_ip, depth);
1313 /* Delete the rule from the rule table. */
1314 ret = rule_delete(lpm, masked_ip, depth);
1318 /* find rule cells */
1319 rule_find_range(lpm, masked_ip, depth, &from, &to, &tbl_ind);
1321 /* find a less specific rule (a rule with smaller depth)
1322 * note: masked_ip will be modified, don't use it anymore
1324 ret = rule_find_less_specific(lpm, masked_ip, depth,
1326 lsp_rule = ret ? &lsp_rule_obj : NULL;
1328 /* decrement the table rule counter,
1329 * note that tbl24 doesn't have a header
1331 if (tbl_ind != TBL24_IND) {
1332 struct rte_lpm_tbl8_hdr *tbl_hdr = &lpm->tbl8_hdrs[tbl_ind];
1333 if (--tbl_hdr->ref_cnt == 0) {
1334 /* remove the table */
1335 remove_tbl(lpm, tbl_hdr, tbl_ind, lsp_rule);
1340 /* iterate rule cells */
1341 for (; from <= to; from++)
1342 if (from->ext_entry == 1) {
1343 /* reference to a more specific space
1344 * of the prefix/rule. Entries in a more
1345 * specific space that are not used by
1346 * a more specific prefix must be occupied
1349 if (lsp_rule != NULL)
1351 from->lpm6_tbl8_gindex *
1352 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES,
1353 depth, lsp_rule->depth,
1354 lsp_rule->next_hop, VALID);
1356 /* since the prefix has no less specific prefix,
1357 * its more specific space must be invalidated
1360 from->lpm6_tbl8_gindex *
1361 RTE_LPM6_TBL8_GROUP_NUM_ENTRIES,
1362 depth, 0, 0, INVALID);
1363 } else if (from->depth == depth) {
1364 /* entry is not a reference and belongs to the prefix */
1365 if (lsp_rule != NULL) {
1366 struct rte_lpm6_tbl_entry new_tbl_entry = {
1367 .next_hop = lsp_rule->next_hop,
1368 .depth = lsp_rule->depth,
1370 .valid_group = VALID,
1374 *from = new_tbl_entry;
1376 struct rte_lpm6_tbl_entry new_tbl_entry = {
1380 .valid_group = INVALID,
1384 *from = new_tbl_entry;