4 * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * * Neither the name of Intel Corporation nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
39 * RTE Longest Prefix Match (LPM)
43 #include <sys/queue.h>
46 #include <rte_branch_prediction.h>
47 #include <rte_byteorder.h>
48 #include <rte_memory.h>
49 #include <rte_common.h>
51 #include <rte_compat.h>
57 /** Max number of characters in LPM name. */
58 #define RTE_LPM_NAMESIZE 32
60 /** Maximum depth value possible for IPv4 LPM. */
61 #define RTE_LPM_MAX_DEPTH 32
63 /** @internal Total number of tbl24 entries. */
64 #define RTE_LPM_TBL24_NUM_ENTRIES (1 << 24)
66 /** @internal Number of entries in a tbl8 group. */
67 #define RTE_LPM_TBL8_GROUP_NUM_ENTRIES 256
69 /** @internal Max number of tbl8 groups in the tbl8. */
70 #define RTE_LPM_MAX_TBL8_NUM_GROUPS (1 << 24)
72 /** @internal Total number of tbl8 groups in the tbl8. */
73 #define RTE_LPM_TBL8_NUM_GROUPS 256
75 /** @internal Total number of tbl8 entries. */
76 #define RTE_LPM_TBL8_NUM_ENTRIES (RTE_LPM_TBL8_NUM_GROUPS * \
77 RTE_LPM_TBL8_GROUP_NUM_ENTRIES)
79 /** @internal Macro to enable/disable run-time checks. */
80 #if defined(RTE_LIBRTE_LPM_DEBUG)
81 #define RTE_LPM_RETURN_IF_TRUE(cond, retval) do { \
82 if (cond) return (retval); \
85 #define RTE_LPM_RETURN_IF_TRUE(cond, retval)
88 /** @internal bitmask with valid and valid_group fields set */
89 #define RTE_LPM_VALID_EXT_ENTRY_BITMASK 0x03000000
91 /** Bitmask used to indicate successful lookup */
92 #define RTE_LPM_LOOKUP_SUCCESS 0x01000000
94 #if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
95 /** @internal Tbl24 entry structure. */
96 struct rte_lpm_tbl_entry_v20 {
98 * Stores Next hop (tbl8 or tbl24 when valid_group is not set) or
99 * a group index pointing to a tbl8 structure (tbl24 only, when
100 * valid_group is set)
106 /* Using single uint8_t to store 3 values. */
107 uint8_t valid :1; /**< Validation flag. */
110 * - valid_group == 0: entry stores a next hop
111 * - valid_group == 1: entry stores a group_index pointing to a tbl8
113 * - valid_group indicates whether the current tbl8 is in use or not
115 uint8_t valid_group :1;
116 uint8_t depth :6; /**< Rule depth. */
119 struct rte_lpm_tbl_entry {
121 * Stores Next hop (tbl8 or tbl24 when valid_group is not set) or
122 * a group index pointing to a tbl8 structure (tbl24 only, when
123 * valid_group is set)
125 uint32_t next_hop :24;
126 /* Using single uint8_t to store 3 values. */
127 uint32_t valid :1; /**< Validation flag. */
130 * - valid_group == 0: entry stores a next hop
131 * - valid_group == 1: entry stores a group_index pointing to a tbl8
133 * - valid_group indicates whether the current tbl8 is in use or not
135 uint32_t valid_group :1;
136 uint32_t depth :6; /**< Rule depth. */
140 struct rte_lpm_tbl_entry_v20 {
142 uint8_t valid_group :1;
150 struct rte_lpm_tbl_entry {
152 uint32_t valid_group :1;
154 uint32_t next_hop :24;
160 /** LPM configuration structure. */
161 struct rte_lpm_config {
162 uint32_t max_rules; /**< Max number of rules. */
163 uint32_t number_tbl8s; /**< Number of tbl8s to allocate. */
164 int flags; /**< This field is currently unused. */
167 /** @internal Rule structure. */
168 struct rte_lpm_rule_v20 {
169 uint32_t ip; /**< Rule IP address. */
170 uint8_t next_hop; /**< Rule next hop. */
173 struct rte_lpm_rule {
174 uint32_t ip; /**< Rule IP address. */
175 uint32_t next_hop; /**< Rule next hop. */
178 /** @internal Contains metadata about the rules table. */
179 struct rte_lpm_rule_info {
180 uint32_t used_rules; /**< Used rules so far. */
181 uint32_t first_rule; /**< Indexes the first rule of a given depth. */
184 /** @internal LPM structure. */
187 char name[RTE_LPM_NAMESIZE]; /**< Name of the lpm. */
188 uint32_t max_rules; /**< Max. balanced rules per lpm. */
189 struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**< Rule info table. */
192 struct rte_lpm_tbl_entry_v20 tbl24[RTE_LPM_TBL24_NUM_ENTRIES]
193 __rte_cache_aligned; /**< LPM tbl24 table. */
194 struct rte_lpm_tbl_entry_v20 tbl8[RTE_LPM_TBL8_NUM_ENTRIES]
195 __rte_cache_aligned; /**< LPM tbl8 table. */
196 struct rte_lpm_rule_v20 rules_tbl[0] \
197 __rte_cache_aligned; /**< LPM rules. */
202 char name[RTE_LPM_NAMESIZE]; /**< Name of the lpm. */
203 uint32_t max_rules; /**< Max. balanced rules per lpm. */
204 uint32_t number_tbl8s; /**< Number of tbl8s. */
205 struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**< Rule info table. */
208 struct rte_lpm_tbl_entry tbl24[RTE_LPM_TBL24_NUM_ENTRIES]
209 __rte_cache_aligned; /**< LPM tbl24 table. */
210 struct rte_lpm_tbl_entry *tbl8; /**< LPM tbl8 table. */
211 struct rte_lpm_rule *rules_tbl; /**< LPM rules. */
215 * Create an LPM object.
220 * NUMA socket ID for LPM table memory allocation
222 * Structure containing the configuration
224 * Handle to LPM object on success, NULL otherwise with rte_errno set
225 * to an appropriate values. Possible rte_errno values include:
226 * - E_RTE_NO_CONFIG - function could not get pointer to rte_config structure
227 * - E_RTE_SECONDARY - function was called from a secondary process instance
228 * - EINVAL - invalid parameter passed to function
229 * - ENOSPC - the maximum number of memzones has already been allocated
230 * - EEXIST - a memzone with the same name already exists
231 * - ENOMEM - no appropriate memory area found in which to create memzone
234 rte_lpm_create(const char *name, int socket_id,
235 const struct rte_lpm_config *config);
237 rte_lpm_create_v20(const char *name, int socket_id, int max_rules, int flags);
239 rte_lpm_create_v1604(const char *name, int socket_id,
240 const struct rte_lpm_config *config);
243 * Find an existing LPM object and return a pointer to it.
246 * Name of the lpm object as passed to rte_lpm_create()
248 * Pointer to lpm object or NULL if object not found with rte_errno
249 * set appropriately. Possible rte_errno values include:
250 * - ENOENT - required entry not available to return.
253 rte_lpm_find_existing(const char *name);
255 rte_lpm_find_existing_v20(const char *name);
257 rte_lpm_find_existing_v1604(const char *name);
260 * Free an LPM object.
268 rte_lpm_free(struct rte_lpm *lpm);
270 rte_lpm_free_v20(struct rte_lpm_v20 *lpm);
272 rte_lpm_free_v1604(struct rte_lpm *lpm);
275 * Add a rule to the LPM table.
280 * IP of the rule to be added to the LPM table
282 * Depth of the rule to be added to the LPM table
284 * Next hop of the rule to be added to the LPM table
286 * 0 on success, negative value otherwise
289 rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth, uint32_t next_hop);
291 rte_lpm_add_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
294 rte_lpm_add_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
298 * Check if a rule is present in the LPM table,
299 * and provide its next hop if it is.
304 * IP of the rule to be searched
306 * Depth of the rule to searched
308 * Next hop of the rule (valid only if it is found)
310 * 1 if the rule exists, 0 if it does not, a negative value on failure
313 rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
316 rte_lpm_is_rule_present_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
319 rte_lpm_is_rule_present_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
323 * Delete a rule from the LPM table.
328 * IP of the rule to be deleted from the LPM table
330 * Depth of the rule to be deleted from the LPM table
332 * 0 on success, negative value otherwise
335 rte_lpm_delete(struct rte_lpm *lpm, uint32_t ip, uint8_t depth);
337 rte_lpm_delete_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth);
339 rte_lpm_delete_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth);
342 * Delete all rules from the LPM table.
348 rte_lpm_delete_all(struct rte_lpm *lpm);
350 rte_lpm_delete_all_v20(struct rte_lpm_v20 *lpm);
352 rte_lpm_delete_all_v1604(struct rte_lpm *lpm);
355 * Lookup an IP into the LPM table.
360 * IP to be looked up in the LPM table
362 * Next hop of the most specific rule found for IP (valid on lookup hit only)
364 * -EINVAL for incorrect arguments, -ENOENT on lookup miss, 0 on lookup hit
367 rte_lpm_lookup(struct rte_lpm *lpm, uint32_t ip, uint32_t *next_hop)
369 unsigned tbl24_index = (ip >> 8);
371 const uint32_t *ptbl;
373 /* DEBUG: Check user input arguments. */
374 RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (next_hop == NULL)), -EINVAL);
376 /* Copy tbl24 entry */
377 ptbl = (const uint32_t *)(&lpm->tbl24[tbl24_index]);
380 /* Copy tbl8 entry (only if needed) */
381 if (unlikely((tbl_entry & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
382 RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
384 unsigned tbl8_index = (uint8_t)ip +
385 (((uint32_t)tbl_entry & 0x00FFFFFF) *
386 RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
388 ptbl = (const uint32_t *)&lpm->tbl8[tbl8_index];
392 *next_hop = ((uint32_t)tbl_entry & 0x00FFFFFF);
393 return (tbl_entry & RTE_LPM_LOOKUP_SUCCESS) ? 0 : -ENOENT;
397 * Lookup multiple IP addresses in an LPM table. This may be implemented as a
398 * macro, so the address of the function should not be used.
403 * Array of IPs to be looked up in the LPM table
405 * Next hop of the most specific rule found for IP (valid on lookup hit only).
406 * This is an array of two byte values. The most significant byte in each
407 * value says whether the lookup was successful (bitmask
408 * RTE_LPM_LOOKUP_SUCCESS is set). The least significant byte is the
411 * Number of elements in ips (and next_hops) array to lookup. This should be a
412 * compile time constant, and divisible by 8 for best performance.
414 * -EINVAL for incorrect arguments, otherwise 0
416 #define rte_lpm_lookup_bulk(lpm, ips, next_hops, n) \
417 rte_lpm_lookup_bulk_func(lpm, ips, next_hops, n)
420 rte_lpm_lookup_bulk_func(const struct rte_lpm *lpm, const uint32_t *ips,
421 uint32_t *next_hops, const unsigned n)
424 unsigned tbl24_indexes[n];
425 const uint32_t *ptbl;
427 /* DEBUG: Check user input arguments. */
428 RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (ips == NULL) ||
429 (next_hops == NULL)), -EINVAL);
431 for (i = 0; i < n; i++) {
432 tbl24_indexes[i] = ips[i] >> 8;
435 for (i = 0; i < n; i++) {
436 /* Simply copy tbl24 entry to output */
437 ptbl = (const uint32_t *)&lpm->tbl24[tbl24_indexes[i]];
438 next_hops[i] = *ptbl;
440 /* Overwrite output with tbl8 entry if needed */
441 if (unlikely((next_hops[i] & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
442 RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
444 unsigned tbl8_index = (uint8_t)ips[i] +
445 (((uint32_t)next_hops[i] & 0x00FFFFFF) *
446 RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
448 ptbl = (const uint32_t *)&lpm->tbl8[tbl8_index];
449 next_hops[i] = *ptbl;
455 /* Mask four results. */
456 #define RTE_LPM_MASKX4_RES UINT64_C(0x00ffffff00ffffff)
459 * Lookup four IP addresses in an LPM table.
464 * Four IPs to be looked up in the LPM table
466 * Next hop of the most specific rule found for IP (valid on lookup hit only).
467 * This is an 4 elements array of two byte values.
468 * If the lookup was succesfull for the given IP, then least significant byte
469 * of the corresponding element is the actual next hop and the most
470 * significant byte is zero.
471 * If the lookup for the given IP failed, then corresponding element would
472 * contain default value, see description of then next parameter.
474 * Default value to populate into corresponding element of hop[] array,
475 * if lookup would fail.
478 rte_lpm_lookupx4(const struct rte_lpm *lpm, xmm_t ip, uint32_t hop[4],
481 #if defined(RTE_ARCH_ARM) || defined(RTE_ARCH_ARM64)
482 #include "rte_lpm_neon.h"
484 #include "rte_lpm_sse.h"
491 #endif /* _RTE_LPM_H_ */