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)
107 /* Using single uint8_t to store 3 values. */
108 uint8_t valid :1; /**< Validation flag. */
111 * - valid_group == 0: entry stores a next hop
112 * - valid_group == 1: entry stores a group_index pointing to a tbl8
114 * - valid_group indicates whether the current tbl8 is in use or not
116 uint8_t valid_group :1;
117 uint8_t depth :6; /**< Rule depth. */
121 struct rte_lpm_tbl_entry {
123 * Stores Next hop (tbl8 or tbl24 when valid_group is not set) or
124 * a group index pointing to a tbl8 structure (tbl24 only, when
125 * valid_group is set)
127 uint32_t next_hop :24;
128 /* Using single uint8_t to store 3 values. */
129 uint32_t valid :1; /**< Validation flag. */
132 * - valid_group == 0: entry stores a next hop
133 * - valid_group == 1: entry stores a group_index pointing to a tbl8
135 * - valid_group indicates whether the current tbl8 is in use or not
137 uint32_t valid_group :1;
138 uint32_t depth :6; /**< Rule depth. */
143 struct rte_lpm_tbl_entry_v20 {
145 uint8_t valid_group :1;
154 struct rte_lpm_tbl_entry {
156 uint32_t valid_group :1;
158 uint32_t next_hop :24;
164 /** LPM configuration structure. */
165 struct rte_lpm_config {
166 uint32_t max_rules; /**< Max number of rules. */
167 uint32_t number_tbl8s; /**< Number of tbl8s to allocate. */
168 int flags; /**< This field is currently unused. */
171 /** @internal Rule structure. */
172 struct rte_lpm_rule_v20 {
173 uint32_t ip; /**< Rule IP address. */
174 uint8_t next_hop; /**< Rule next hop. */
177 struct rte_lpm_rule {
178 uint32_t ip; /**< Rule IP address. */
179 uint32_t next_hop; /**< Rule next hop. */
182 /** @internal Contains metadata about the rules table. */
183 struct rte_lpm_rule_info {
184 uint32_t used_rules; /**< Used rules so far. */
185 uint32_t first_rule; /**< Indexes the first rule of a given depth. */
188 /** @internal LPM structure. */
191 char name[RTE_LPM_NAMESIZE]; /**< Name of the lpm. */
192 uint32_t max_rules; /**< Max. balanced rules per lpm. */
193 struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**< Rule info table. */
196 struct rte_lpm_tbl_entry_v20 tbl24[RTE_LPM_TBL24_NUM_ENTRIES]
197 __rte_cache_aligned; /**< LPM tbl24 table. */
198 struct rte_lpm_tbl_entry_v20 tbl8[RTE_LPM_TBL8_NUM_ENTRIES]
199 __rte_cache_aligned; /**< LPM tbl8 table. */
200 struct rte_lpm_rule_v20 rules_tbl[]
201 __rte_cache_aligned; /**< LPM rules. */
206 char name[RTE_LPM_NAMESIZE]; /**< Name of the lpm. */
207 uint32_t max_rules; /**< Max. balanced rules per lpm. */
208 uint32_t number_tbl8s; /**< Number of tbl8s. */
209 struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**< Rule info table. */
212 struct rte_lpm_tbl_entry tbl24[RTE_LPM_TBL24_NUM_ENTRIES]
213 __rte_cache_aligned; /**< LPM tbl24 table. */
214 struct rte_lpm_tbl_entry *tbl8; /**< LPM tbl8 table. */
215 struct rte_lpm_rule *rules_tbl; /**< LPM rules. */
219 * Create an LPM object.
224 * NUMA socket ID for LPM table memory allocation
226 * Structure containing the configuration
228 * Handle to LPM object on success, NULL otherwise with rte_errno set
229 * to an appropriate values. Possible rte_errno values include:
230 * - E_RTE_NO_CONFIG - function could not get pointer to rte_config structure
231 * - E_RTE_SECONDARY - function was called from a secondary process instance
232 * - EINVAL - invalid parameter passed to function
233 * - ENOSPC - the maximum number of memzones has already been allocated
234 * - EEXIST - a memzone with the same name already exists
235 * - ENOMEM - no appropriate memory area found in which to create memzone
238 rte_lpm_create(const char *name, int socket_id,
239 const struct rte_lpm_config *config);
241 rte_lpm_create_v20(const char *name, int socket_id, int max_rules, int flags);
243 rte_lpm_create_v1604(const char *name, int socket_id,
244 const struct rte_lpm_config *config);
247 * Find an existing LPM object and return a pointer to it.
250 * Name of the lpm object as passed to rte_lpm_create()
252 * Pointer to lpm object or NULL if object not found with rte_errno
253 * set appropriately. Possible rte_errno values include:
254 * - ENOENT - required entry not available to return.
257 rte_lpm_find_existing(const char *name);
259 rte_lpm_find_existing_v20(const char *name);
261 rte_lpm_find_existing_v1604(const char *name);
264 * Free an LPM object.
272 rte_lpm_free(struct rte_lpm *lpm);
274 rte_lpm_free_v20(struct rte_lpm_v20 *lpm);
276 rte_lpm_free_v1604(struct rte_lpm *lpm);
279 * Add a rule to the LPM table.
284 * IP of the rule to be added to the LPM table
286 * Depth of the rule to be added to the LPM table
288 * Next hop of the rule to be added to the LPM table
290 * 0 on success, negative value otherwise
293 rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth, uint32_t next_hop);
295 rte_lpm_add_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
298 rte_lpm_add_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
302 * Check if a rule is present in the LPM table,
303 * and provide its next hop if it is.
308 * IP of the rule to be searched
310 * Depth of the rule to searched
312 * Next hop of the rule (valid only if it is found)
314 * 1 if the rule exists, 0 if it does not, a negative value on failure
317 rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
320 rte_lpm_is_rule_present_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
323 rte_lpm_is_rule_present_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
327 * Delete a rule from the LPM table.
332 * IP of the rule to be deleted from the LPM table
334 * Depth of the rule to be deleted from the LPM table
336 * 0 on success, negative value otherwise
339 rte_lpm_delete(struct rte_lpm *lpm, uint32_t ip, uint8_t depth);
341 rte_lpm_delete_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth);
343 rte_lpm_delete_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth);
346 * Delete all rules from the LPM table.
352 rte_lpm_delete_all(struct rte_lpm *lpm);
354 rte_lpm_delete_all_v20(struct rte_lpm_v20 *lpm);
356 rte_lpm_delete_all_v1604(struct rte_lpm *lpm);
359 * Lookup an IP into the LPM table.
364 * IP to be looked up in the LPM table
366 * Next hop of the most specific rule found for IP (valid on lookup hit only)
368 * -EINVAL for incorrect arguments, -ENOENT on lookup miss, 0 on lookup hit
371 rte_lpm_lookup(struct rte_lpm *lpm, uint32_t ip, uint32_t *next_hop)
373 unsigned tbl24_index = (ip >> 8);
375 const uint32_t *ptbl;
377 /* DEBUG: Check user input arguments. */
378 RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (next_hop == NULL)), -EINVAL);
380 /* Copy tbl24 entry */
381 ptbl = (const uint32_t *)(&lpm->tbl24[tbl24_index]);
384 /* Copy tbl8 entry (only if needed) */
385 if (unlikely((tbl_entry & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
386 RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
388 unsigned tbl8_index = (uint8_t)ip +
389 (((uint32_t)tbl_entry & 0x00FFFFFF) *
390 RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
392 ptbl = (const uint32_t *)&lpm->tbl8[tbl8_index];
396 *next_hop = ((uint32_t)tbl_entry & 0x00FFFFFF);
397 return (tbl_entry & RTE_LPM_LOOKUP_SUCCESS) ? 0 : -ENOENT;
401 * Lookup multiple IP addresses in an LPM table. This may be implemented as a
402 * macro, so the address of the function should not be used.
407 * Array of IPs to be looked up in the LPM table
409 * Next hop of the most specific rule found for IP (valid on lookup hit only).
410 * This is an array of two byte values. The most significant byte in each
411 * value says whether the lookup was successful (bitmask
412 * RTE_LPM_LOOKUP_SUCCESS is set). The least significant byte is the
415 * Number of elements in ips (and next_hops) array to lookup. This should be a
416 * compile time constant, and divisible by 8 for best performance.
418 * -EINVAL for incorrect arguments, otherwise 0
420 #define rte_lpm_lookup_bulk(lpm, ips, next_hops, n) \
421 rte_lpm_lookup_bulk_func(lpm, ips, next_hops, n)
424 rte_lpm_lookup_bulk_func(const struct rte_lpm *lpm, const uint32_t *ips,
425 uint32_t *next_hops, const unsigned n)
428 unsigned tbl24_indexes[n];
429 const uint32_t *ptbl;
431 /* DEBUG: Check user input arguments. */
432 RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (ips == NULL) ||
433 (next_hops == NULL)), -EINVAL);
435 for (i = 0; i < n; i++) {
436 tbl24_indexes[i] = ips[i] >> 8;
439 for (i = 0; i < n; i++) {
440 /* Simply copy tbl24 entry to output */
441 ptbl = (const uint32_t *)&lpm->tbl24[tbl24_indexes[i]];
442 next_hops[i] = *ptbl;
444 /* Overwrite output with tbl8 entry if needed */
445 if (unlikely((next_hops[i] & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
446 RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
448 unsigned tbl8_index = (uint8_t)ips[i] +
449 (((uint32_t)next_hops[i] & 0x00FFFFFF) *
450 RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
452 ptbl = (const uint32_t *)&lpm->tbl8[tbl8_index];
453 next_hops[i] = *ptbl;
459 /* Mask four results. */
460 #define RTE_LPM_MASKX4_RES UINT64_C(0x00ffffff00ffffff)
463 * Lookup four IP addresses in an LPM table.
468 * Four IPs to be looked up in the LPM table
470 * Next hop of the most specific rule found for IP (valid on lookup hit only).
471 * This is an 4 elements array of two byte values.
472 * If the lookup was succesfull for the given IP, then least significant byte
473 * of the corresponding element is the actual next hop and the most
474 * significant byte is zero.
475 * If the lookup for the given IP failed, then corresponding element would
476 * contain default value, see description of then next parameter.
478 * Default value to populate into corresponding element of hop[] array,
479 * if lookup would fail.
482 rte_lpm_lookupx4(const struct rte_lpm *lpm, xmm_t ip, uint32_t hop[4],
485 #if defined(RTE_ARCH_ARM) || defined(RTE_ARCH_ARM64)
486 #include "rte_lpm_neon.h"
487 #elif defined(RTE_ARCH_PPC_64)
488 #include "rte_lpm_altivec.h"
490 #include "rte_lpm_sse.h"
497 #endif /* _RTE_LPM_H_ */