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. */
97 struct rte_lpm_tbl_entry_v20 {
99 * Stores Next hop (tbl8 or tbl24 when valid_group is not set) or
100 * a group index pointing to a tbl8 structure (tbl24 only, when
101 * valid_group is set)
108 /* Using single uint8_t to store 3 values. */
109 uint8_t valid :1; /**< Validation flag. */
112 * - valid_group == 0: entry stores a next hop
113 * - valid_group == 1: entry stores a group_index pointing to a tbl8
115 * - valid_group indicates whether the current tbl8 is in use or not
117 uint8_t valid_group :1;
118 uint8_t depth :6; /**< Rule depth. */
122 struct rte_lpm_tbl_entry {
124 * Stores Next hop (tbl8 or tbl24 when valid_group is not set) or
125 * a group index pointing to a tbl8 structure (tbl24 only, when
126 * valid_group is set)
128 uint32_t next_hop :24;
129 /* Using single uint8_t to store 3 values. */
130 uint32_t valid :1; /**< Validation flag. */
133 * - valid_group == 0: entry stores a next hop
134 * - valid_group == 1: entry stores a group_index pointing to a tbl8
136 * - valid_group indicates whether the current tbl8 is in use or not
138 uint32_t valid_group :1;
139 uint32_t depth :6; /**< Rule depth. */
144 struct rte_lpm_tbl_entry_v20 {
146 uint8_t valid_group :1;
155 struct rte_lpm_tbl_entry {
157 uint32_t valid_group :1;
159 uint32_t next_hop :24;
165 /** LPM configuration structure. */
166 struct rte_lpm_config {
167 uint32_t max_rules; /**< Max number of rules. */
168 uint32_t number_tbl8s; /**< Number of tbl8s to allocate. */
169 int flags; /**< This field is currently unused. */
172 /** @internal Rule structure. */
173 struct rte_lpm_rule_v20 {
174 uint32_t ip; /**< Rule IP address. */
175 uint8_t next_hop; /**< Rule next hop. */
178 struct rte_lpm_rule {
179 uint32_t ip; /**< Rule IP address. */
180 uint32_t next_hop; /**< Rule next hop. */
183 /** @internal Contains metadata about the rules table. */
184 struct rte_lpm_rule_info {
185 uint32_t used_rules; /**< Used rules so far. */
186 uint32_t first_rule; /**< Indexes the first rule of a given depth. */
189 /** @internal LPM structure. */
192 char name[RTE_LPM_NAMESIZE]; /**< Name of the lpm. */
193 uint32_t max_rules; /**< Max. balanced rules per lpm. */
194 struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**< Rule info table. */
197 struct rte_lpm_tbl_entry_v20 tbl24[RTE_LPM_TBL24_NUM_ENTRIES]
198 __rte_cache_aligned; /**< LPM tbl24 table. */
199 struct rte_lpm_tbl_entry_v20 tbl8[RTE_LPM_TBL8_NUM_ENTRIES]
200 __rte_cache_aligned; /**< LPM tbl8 table. */
201 struct rte_lpm_rule_v20 rules_tbl[]
202 __rte_cache_aligned; /**< LPM rules. */
207 char name[RTE_LPM_NAMESIZE]; /**< Name of the lpm. */
208 uint32_t max_rules; /**< Max. balanced rules per lpm. */
209 uint32_t number_tbl8s; /**< Number of tbl8s. */
210 struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**< Rule info table. */
213 struct rte_lpm_tbl_entry tbl24[RTE_LPM_TBL24_NUM_ENTRIES]
214 __rte_cache_aligned; /**< LPM tbl24 table. */
215 struct rte_lpm_tbl_entry *tbl8; /**< LPM tbl8 table. */
216 struct rte_lpm_rule *rules_tbl; /**< LPM rules. */
220 * Create an LPM object.
225 * NUMA socket ID for LPM table memory allocation
227 * Structure containing the configuration
229 * Handle to LPM object on success, NULL otherwise with rte_errno set
230 * to an appropriate values. Possible rte_errno values include:
231 * - E_RTE_NO_CONFIG - function could not get pointer to rte_config structure
232 * - E_RTE_SECONDARY - function was called from a secondary process instance
233 * - EINVAL - invalid parameter passed to function
234 * - ENOSPC - the maximum number of memzones has already been allocated
235 * - EEXIST - a memzone with the same name already exists
236 * - ENOMEM - no appropriate memory area found in which to create memzone
239 rte_lpm_create(const char *name, int socket_id,
240 const struct rte_lpm_config *config);
242 rte_lpm_create_v20(const char *name, int socket_id, int max_rules, int flags);
244 rte_lpm_create_v1604(const char *name, int socket_id,
245 const struct rte_lpm_config *config);
248 * Find an existing LPM object and return a pointer to it.
251 * Name of the lpm object as passed to rte_lpm_create()
253 * Pointer to lpm object or NULL if object not found with rte_errno
254 * set appropriately. Possible rte_errno values include:
255 * - ENOENT - required entry not available to return.
258 rte_lpm_find_existing(const char *name);
260 rte_lpm_find_existing_v20(const char *name);
262 rte_lpm_find_existing_v1604(const char *name);
265 * Free an LPM object.
273 rte_lpm_free(struct rte_lpm *lpm);
275 rte_lpm_free_v20(struct rte_lpm_v20 *lpm);
277 rte_lpm_free_v1604(struct rte_lpm *lpm);
280 * Add a rule to the LPM table.
285 * IP of the rule to be added to the LPM table
287 * Depth of the rule to be added to the LPM table
289 * Next hop of the rule to be added to the LPM table
291 * 0 on success, negative value otherwise
294 rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth, uint32_t next_hop);
296 rte_lpm_add_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
299 rte_lpm_add_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
303 * Check if a rule is present in the LPM table,
304 * and provide its next hop if it is.
309 * IP of the rule to be searched
311 * Depth of the rule to searched
313 * Next hop of the rule (valid only if it is found)
315 * 1 if the rule exists, 0 if it does not, a negative value on failure
318 rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
321 rte_lpm_is_rule_present_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
324 rte_lpm_is_rule_present_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
328 * Delete a rule from the LPM table.
333 * IP of the rule to be deleted from the LPM table
335 * Depth of the rule to be deleted from the LPM table
337 * 0 on success, negative value otherwise
340 rte_lpm_delete(struct rte_lpm *lpm, uint32_t ip, uint8_t depth);
342 rte_lpm_delete_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth);
344 rte_lpm_delete_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth);
347 * Delete all rules from the LPM table.
353 rte_lpm_delete_all(struct rte_lpm *lpm);
355 rte_lpm_delete_all_v20(struct rte_lpm_v20 *lpm);
357 rte_lpm_delete_all_v1604(struct rte_lpm *lpm);
360 * Lookup an IP into the LPM table.
365 * IP to be looked up in the LPM table
367 * Next hop of the most specific rule found for IP (valid on lookup hit only)
369 * -EINVAL for incorrect arguments, -ENOENT on lookup miss, 0 on lookup hit
372 rte_lpm_lookup(struct rte_lpm *lpm, uint32_t ip, uint32_t *next_hop)
374 unsigned tbl24_index = (ip >> 8);
376 const uint32_t *ptbl;
378 /* DEBUG: Check user input arguments. */
379 RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (next_hop == NULL)), -EINVAL);
381 /* Copy tbl24 entry */
382 ptbl = (const uint32_t *)(&lpm->tbl24[tbl24_index]);
385 /* Copy tbl8 entry (only if needed) */
386 if (unlikely((tbl_entry & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
387 RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
389 unsigned tbl8_index = (uint8_t)ip +
390 (((uint32_t)tbl_entry & 0x00FFFFFF) *
391 RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
393 ptbl = (const uint32_t *)&lpm->tbl8[tbl8_index];
397 *next_hop = ((uint32_t)tbl_entry & 0x00FFFFFF);
398 return (tbl_entry & RTE_LPM_LOOKUP_SUCCESS) ? 0 : -ENOENT;
402 * Lookup multiple IP addresses in an LPM table. This may be implemented as a
403 * macro, so the address of the function should not be used.
408 * Array of IPs to be looked up in the LPM table
410 * Next hop of the most specific rule found for IP (valid on lookup hit only).
411 * This is an array of two byte values. The most significant byte in each
412 * value says whether the lookup was successful (bitmask
413 * RTE_LPM_LOOKUP_SUCCESS is set). The least significant byte is the
416 * Number of elements in ips (and next_hops) array to lookup. This should be a
417 * compile time constant, and divisible by 8 for best performance.
419 * -EINVAL for incorrect arguments, otherwise 0
421 #define rte_lpm_lookup_bulk(lpm, ips, next_hops, n) \
422 rte_lpm_lookup_bulk_func(lpm, ips, next_hops, n)
425 rte_lpm_lookup_bulk_func(const struct rte_lpm *lpm, const uint32_t *ips,
426 uint32_t *next_hops, const unsigned n)
429 unsigned tbl24_indexes[n];
430 const uint32_t *ptbl;
432 /* DEBUG: Check user input arguments. */
433 RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (ips == NULL) ||
434 (next_hops == NULL)), -EINVAL);
436 for (i = 0; i < n; i++) {
437 tbl24_indexes[i] = ips[i] >> 8;
440 for (i = 0; i < n; i++) {
441 /* Simply copy tbl24 entry to output */
442 ptbl = (const uint32_t *)&lpm->tbl24[tbl24_indexes[i]];
443 next_hops[i] = *ptbl;
445 /* Overwrite output with tbl8 entry if needed */
446 if (unlikely((next_hops[i] & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
447 RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
449 unsigned tbl8_index = (uint8_t)ips[i] +
450 (((uint32_t)next_hops[i] & 0x00FFFFFF) *
451 RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
453 ptbl = (const uint32_t *)&lpm->tbl8[tbl8_index];
454 next_hops[i] = *ptbl;
460 /* Mask four results. */
461 #define RTE_LPM_MASKX4_RES UINT64_C(0x00ffffff00ffffff)
464 * Lookup four IP addresses in an LPM table.
469 * Four IPs to be looked up in the LPM table
471 * Next hop of the most specific rule found for IP (valid on lookup hit only).
472 * This is an 4 elements array of two byte values.
473 * If the lookup was succesfull for the given IP, then least significant byte
474 * of the corresponding element is the actual next hop and the most
475 * significant byte is zero.
476 * If the lookup for the given IP failed, then corresponding element would
477 * contain default value, see description of then next parameter.
479 * Default value to populate into corresponding element of hop[] array,
480 * if lookup would fail.
483 rte_lpm_lookupx4(const struct rte_lpm *lpm, xmm_t ip, uint32_t hop[4],
486 #if defined(RTE_ARCH_ARM) || defined(RTE_ARCH_ARM64)
487 #include "rte_lpm_neon.h"
488 #elif defined(RTE_ARCH_PPC_64)
489 #include "rte_lpm_altivec.h"
491 #include "rte_lpm_sse.h"
498 #endif /* _RTE_LPM_H_ */