vhost: update packed ring dequeue
[dpdk.git] / lib / librte_lpm / rte_lpm.c
index ec67765..3a929a1 100644 (file)
@@ -1,34 +1,5 @@
-/*-
- *   BSD LICENSE
- *
- *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
- *   All rights reserved.
- *
- *   Redistribution and use in source and binary forms, with or without
- *   modification, are permitted provided that the following conditions
- *   are met:
- *
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in
- *       the documentation and/or other materials provided with the
- *       distribution.
- *     * Neither the name of Intel Corporation nor the names of its
- *       contributors may be used to endorse or promote products derived
- *       from this software without specific prior written permission.
- *
- *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2014 Intel Corporation
  */
 
 #include <string.h>
@@ -36,7 +7,6 @@
 #include <errno.h>
 #include <stdarg.h>
 #include <stdio.h>
-#include <errno.h>
 #include <sys/queue.h>
 
 #include <rte_log.h>
@@ -44,7 +14,6 @@
 #include <rte_common.h>
 #include <rte_memory.h>        /* for definition of RTE_CACHE_LINE_SIZE */
 #include <rte_malloc.h>
-#include <rte_memzone.h>
 #include <rte_eal.h>
 #include <rte_eal_memconfig.h>
 #include <rte_per_lcore.h>
@@ -52,6 +21,7 @@
 #include <rte_errno.h>
 #include <rte_rwlock.h>
 #include <rte_spinlock.h>
+#include <rte_tailq.h>
 
 #include "rte_lpm.h"
 
@@ -101,7 +71,7 @@ depth_to_mask(uint8_t depth)
 /*
  * Converts given depth value to its corresponding range value.
  */
-static inline uint32_t __attribute__((pure))
+static uint32_t __attribute__((pure))
 depth_to_range(uint8_t depth)
 {
        VERIFY_DEPTH(depth);
@@ -128,13 +98,13 @@ rte_lpm_find_existing_v20(const char *name)
 
        lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 
-       rte_rwlock_read_lock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_read_lock();
        TAILQ_FOREACH(te, lpm_list, next) {
-               l = (struct rte_lpm_v20 *) te->data;
+               l = te->data;
                if (strncmp(name, l->name, RTE_LPM_NAMESIZE) == 0)
                        break;
        }
-       rte_rwlock_read_unlock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_read_unlock();
 
        if (te == NULL) {
                rte_errno = ENOENT;
@@ -154,13 +124,13 @@ rte_lpm_find_existing_v1604(const char *name)
 
        lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 
-       rte_rwlock_read_lock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_read_lock();
        TAILQ_FOREACH(te, lpm_list, next) {
-               l = (struct rte_lpm *) te->data;
+               l = te->data;
                if (strncmp(name, l->name, RTE_LPM_NAMESIZE) == 0)
                        break;
        }
-       rte_rwlock_read_unlock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_read_unlock();
 
        if (te == NULL) {
                rte_errno = ENOENT;
@@ -201,16 +171,17 @@ rte_lpm_create_v20(const char *name, int socket_id, int max_rules,
        /* Determine the amount of memory to allocate. */
        mem_size = sizeof(*lpm) + (sizeof(lpm->rules_tbl[0]) * max_rules);
 
-       rte_rwlock_write_lock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_lock();
 
        /* guarantee there's no existing */
        TAILQ_FOREACH(te, lpm_list, next) {
-               lpm = (struct rte_lpm_v20 *) te->data;
+               lpm = te->data;
                if (strncmp(name, lpm->name, RTE_LPM_NAMESIZE) == 0)
                        break;
        }
-       lpm = NULL;
+
        if (te != NULL) {
+               lpm = NULL;
                rte_errno = EEXIST;
                goto exit;
        }
@@ -219,28 +190,30 @@ rte_lpm_create_v20(const char *name, int socket_id, int max_rules,
        te = rte_zmalloc("LPM_TAILQ_ENTRY", sizeof(*te), 0);
        if (te == NULL) {
                RTE_LOG(ERR, LPM, "Failed to allocate tailq entry\n");
+               rte_errno = ENOMEM;
                goto exit;
        }
 
        /* Allocate memory to store the LPM data structures. */
-       lpm = (struct rte_lpm_v20 *)rte_zmalloc_socket(mem_name, mem_size,
+       lpm = rte_zmalloc_socket(mem_name, mem_size,
                        RTE_CACHE_LINE_SIZE, socket_id);
        if (lpm == NULL) {
                RTE_LOG(ERR, LPM, "LPM memory allocation failed\n");
                rte_free(te);
+               rte_errno = ENOMEM;
                goto exit;
        }
 
        /* Save user arguments. */
        lpm->max_rules = max_rules;
-       snprintf(lpm->name, sizeof(lpm->name), "%s", name);
+       strlcpy(lpm->name, name, sizeof(lpm->name));
 
-       te->data = (void *) lpm;
+       te->data = lpm;
 
        TAILQ_INSERT_TAIL(lpm_list, te, next);
 
 exit:
-       rte_rwlock_write_unlock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_unlock();
 
        return lpm;
 }
@@ -275,16 +248,17 @@ rte_lpm_create_v1604(const char *name, int socket_id,
        tbl8s_size = (sizeof(struct rte_lpm_tbl_entry) *
                        RTE_LPM_TBL8_GROUP_NUM_ENTRIES * config->number_tbl8s);
 
-       rte_rwlock_write_lock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_lock();
 
        /* guarantee there's no existing */
        TAILQ_FOREACH(te, lpm_list, next) {
-               lpm = (struct rte_lpm *) te->data;
+               lpm = te->data;
                if (strncmp(name, lpm->name, RTE_LPM_NAMESIZE) == 0)
                        break;
        }
-       lpm = NULL;
+
        if (te != NULL) {
+               lpm = NULL;
                rte_errno = EEXIST;
                goto exit;
        }
@@ -293,19 +267,21 @@ rte_lpm_create_v1604(const char *name, int socket_id,
        te = rte_zmalloc("LPM_TAILQ_ENTRY", sizeof(*te), 0);
        if (te == NULL) {
                RTE_LOG(ERR, LPM, "Failed to allocate tailq entry\n");
+               rte_errno = ENOMEM;
                goto exit;
        }
 
        /* Allocate memory to store the LPM data structures. */
-       lpm = (struct rte_lpm *)rte_zmalloc_socket(mem_name, mem_size,
+       lpm = rte_zmalloc_socket(mem_name, mem_size,
                        RTE_CACHE_LINE_SIZE, socket_id);
        if (lpm == NULL) {
                RTE_LOG(ERR, LPM, "LPM memory allocation failed\n");
                rte_free(te);
+               rte_errno = ENOMEM;
                goto exit;
        }
 
-       lpm->rules_tbl = (struct rte_lpm_rule *)rte_zmalloc_socket(NULL,
+       lpm->rules_tbl = rte_zmalloc_socket(NULL,
                        (size_t)rules_size, RTE_CACHE_LINE_SIZE, socket_id);
 
        if (lpm->rules_tbl == NULL) {
@@ -313,31 +289,34 @@ rte_lpm_create_v1604(const char *name, int socket_id,
                rte_free(lpm);
                lpm = NULL;
                rte_free(te);
+               rte_errno = ENOMEM;
                goto exit;
        }
 
-       lpm->tbl8 = (struct rte_lpm_tbl_entry *)rte_zmalloc_socket(NULL,
+       lpm->tbl8 = rte_zmalloc_socket(NULL,
                        (size_t)tbl8s_size, RTE_CACHE_LINE_SIZE, socket_id);
 
        if (lpm->tbl8 == NULL) {
                RTE_LOG(ERR, LPM, "LPM tbl8 memory allocation failed\n");
+               rte_free(lpm->rules_tbl);
                rte_free(lpm);
                lpm = NULL;
                rte_free(te);
+               rte_errno = ENOMEM;
                goto exit;
        }
 
        /* Save user arguments. */
        lpm->max_rules = config->max_rules;
        lpm->number_tbl8s = config->number_tbl8s;
-       snprintf(lpm->name, sizeof(lpm->name), "%s", name);
+       strlcpy(lpm->name, name, sizeof(lpm->name));
 
-       te->data = (void *) lpm;
+       te->data = lpm;
 
        TAILQ_INSERT_TAIL(lpm_list, te, next);
 
 exit:
-       rte_rwlock_write_unlock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_unlock();
 
        return lpm;
 }
@@ -361,7 +340,7 @@ rte_lpm_free_v20(struct rte_lpm_v20 *lpm)
 
        lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 
-       rte_rwlock_write_lock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_lock();
 
        /* find our tailq entry */
        TAILQ_FOREACH(te, lpm_list, next) {
@@ -371,7 +350,7 @@ rte_lpm_free_v20(struct rte_lpm_v20 *lpm)
        if (te != NULL)
                TAILQ_REMOVE(lpm_list, te, next);
 
-       rte_rwlock_write_unlock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_unlock();
 
        rte_free(lpm);
        rte_free(te);
@@ -390,7 +369,7 @@ rte_lpm_free_v1604(struct rte_lpm *lpm)
 
        lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
 
-       rte_rwlock_write_lock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_lock();
 
        /* find our tailq entry */
        TAILQ_FOREACH(te, lpm_list, next) {
@@ -400,8 +379,9 @@ rte_lpm_free_v1604(struct rte_lpm *lpm)
        if (te != NULL)
                TAILQ_REMOVE(lpm_list, te, next);
 
-       rte_rwlock_write_unlock(RTE_EAL_TAILQ_RWLOCK);
+       rte_mcfg_tailq_write_unlock();
 
+       rte_free(lpm->tbl8);
        rte_free(lpm->rules_tbl);
        rte_free(lpm);
        rte_free(te);
@@ -420,7 +400,7 @@ MAP_STATIC_SYMBOL(void rte_lpm_free(struct rte_lpm *lpm),
  * are stored in the rule table from 0 - 31.
  * NOTE: Valid range for depth parameter is 1 .. 32 inclusive.
  */
-static inline int32_t
+static int32_t
 rule_add_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
        uint8_t next_hop)
 {
@@ -492,7 +472,7 @@ rule_add_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
        return rule_index;
 }
 
-static inline int32_t
+static int32_t
 rule_add_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
        uint32_t next_hop)
 {
@@ -568,7 +548,7 @@ rule_add_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
  * Delete a rule from the rule table.
  * NOTE: Valid range for depth parameter is 1 .. 32 inclusive.
  */
-static inline void
+static void
 rule_delete_v20(struct rte_lpm_v20 *lpm, int32_t rule_index, uint8_t depth)
 {
        int i;
@@ -591,7 +571,7 @@ rule_delete_v20(struct rte_lpm_v20 *lpm, int32_t rule_index, uint8_t depth)
        lpm->rule_info[depth - 1].used_rules--;
 }
 
-static inline void
+static void
 rule_delete_v1604(struct rte_lpm *lpm, int32_t rule_index, uint8_t depth)
 {
        int i;
@@ -618,7 +598,7 @@ rule_delete_v1604(struct rte_lpm *lpm, int32_t rule_index, uint8_t depth)
  * Finds a rule in rule table.
  * NOTE: Valid range for depth parameter is 1 .. 32 inclusive.
  */
-static inline int32_t
+static int32_t
 rule_find_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth)
 {
        uint32_t rule_gindex, last_rule, rule_index;
@@ -639,7 +619,7 @@ rule_find_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth)
        return -EINVAL;
 }
 
-static inline int32_t
+static int32_t
 rule_find_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth)
 {
        uint32_t rule_gindex, last_rule, rule_index;
@@ -663,7 +643,7 @@ rule_find_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth)
 /*
  * Find, clean and allocate a tbl8.
  */
-static inline int32_t
+static int32_t
 tbl8_alloc_v20(struct rte_lpm_tbl_entry_v20 *tbl8)
 {
        uint32_t group_idx; /* tbl8 group index. */
@@ -675,11 +655,19 @@ tbl8_alloc_v20(struct rte_lpm_tbl_entry_v20 *tbl8)
                tbl8_entry = &tbl8[group_idx * RTE_LPM_TBL8_GROUP_NUM_ENTRIES];
                /* If a free tbl8 group is found clean it and set as VALID. */
                if (!tbl8_entry->valid_group) {
+                       struct rte_lpm_tbl_entry_v20 new_tbl8_entry = {
+                               .valid = INVALID,
+                               .depth = 0,
+                               .valid_group = VALID,
+                       };
+                       new_tbl8_entry.next_hop = 0;
+
                        memset(&tbl8_entry[0], 0,
                                        RTE_LPM_TBL8_GROUP_NUM_ENTRIES *
                                        sizeof(tbl8_entry[0]));
 
-                       tbl8_entry->valid_group = VALID;
+                       __atomic_store(tbl8_entry, &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
 
                        /* Return group index for allocated tbl8 group. */
                        return group_idx;
@@ -690,7 +678,7 @@ tbl8_alloc_v20(struct rte_lpm_tbl_entry_v20 *tbl8)
        return -ENOSPC;
 }
 
-static inline int32_t
+static int32_t
 tbl8_alloc_v1604(struct rte_lpm_tbl_entry *tbl8, uint32_t number_tbl8s)
 {
        uint32_t group_idx; /* tbl8 group index. */
@@ -701,11 +689,19 @@ tbl8_alloc_v1604(struct rte_lpm_tbl_entry *tbl8, uint32_t number_tbl8s)
                tbl8_entry = &tbl8[group_idx * RTE_LPM_TBL8_GROUP_NUM_ENTRIES];
                /* If a free tbl8 group is found clean it and set as VALID. */
                if (!tbl8_entry->valid_group) {
+                       struct rte_lpm_tbl_entry new_tbl8_entry = {
+                               .next_hop = 0,
+                               .valid = INVALID,
+                               .depth = 0,
+                               .valid_group = VALID,
+                       };
+
                        memset(&tbl8_entry[0], 0,
                                        RTE_LPM_TBL8_GROUP_NUM_ENTRIES *
                                        sizeof(tbl8_entry[0]));
 
-                       tbl8_entry->valid_group = VALID;
+                       __atomic_store(tbl8_entry, &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
 
                        /* Return group index for allocated tbl8 group. */
                        return group_idx;
@@ -716,21 +712,32 @@ tbl8_alloc_v1604(struct rte_lpm_tbl_entry *tbl8, uint32_t number_tbl8s)
        return -ENOSPC;
 }
 
-static inline void
+static void
 tbl8_free_v20(struct rte_lpm_tbl_entry_v20 *tbl8, uint32_t tbl8_group_start)
 {
        /* Set tbl8 group invalid*/
-       tbl8[tbl8_group_start].valid_group = INVALID;
+       struct rte_lpm_tbl_entry_v20 zero_tbl8_entry = {
+               .valid = INVALID,
+               .depth = 0,
+               .valid_group = INVALID,
+       };
+       zero_tbl8_entry.next_hop = 0;
+
+       __atomic_store(&tbl8[tbl8_group_start], &zero_tbl8_entry,
+                       __ATOMIC_RELAXED);
 }
 
-static inline void
+static void
 tbl8_free_v1604(struct rte_lpm_tbl_entry *tbl8, uint32_t tbl8_group_start)
 {
        /* Set tbl8 group invalid*/
-       tbl8[tbl8_group_start].valid_group = INVALID;
+       struct rte_lpm_tbl_entry zero_tbl8_entry = {0};
+
+       __atomic_store(&tbl8[tbl8_group_start], &zero_tbl8_entry,
+                       __ATOMIC_RELAXED);
 }
 
-static inline int32_t
+static __rte_noinline int32_t
 add_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
                uint8_t next_hop)
 {
@@ -758,7 +765,8 @@ add_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
                        /* Setting tbl24 entry in one go to avoid race
                         * conditions
                         */
-                       lpm->tbl24[i] = new_tbl24_entry;
+                       __atomic_store(&lpm->tbl24[i], &new_tbl24_entry,
+                                       __ATOMIC_RELEASE);
 
                        continue;
                }
@@ -787,7 +795,9 @@ add_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
                                         * Setting tbl8 entry in one go to avoid
                                         * race conditions
                                         */
-                                       lpm->tbl8[j] = new_tbl8_entry;
+                                       __atomic_store(&lpm->tbl8[j],
+                                               &new_tbl8_entry,
+                                               __ATOMIC_RELAXED);
 
                                        continue;
                                }
@@ -798,7 +808,7 @@ add_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
        return 0;
 }
 
-static inline int32_t
+static __rte_noinline int32_t
 add_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
                uint32_t next_hop)
 {
@@ -827,7 +837,8 @@ add_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
                        /* Setting tbl24 entry in one go to avoid race
                         * conditions
                         */
-                       lpm->tbl24[i] = new_tbl24_entry;
+                       __atomic_store(&lpm->tbl24[i], &new_tbl24_entry,
+                                       __ATOMIC_RELEASE);
 
                        continue;
                }
@@ -856,7 +867,9 @@ add_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
                                         * Setting tbl8 entry in one go to avoid
                                         * race conditions
                                         */
-                                       lpm->tbl8[j] = new_tbl8_entry;
+                                       __atomic_store(&lpm->tbl8[j],
+                                               &new_tbl8_entry,
+                                               __ATOMIC_RELAXED);
 
                                        continue;
                                }
@@ -867,7 +880,7 @@ add_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
        return 0;
 }
 
-static inline int32_t
+static __rte_noinline int32_t
 add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
                uint8_t next_hop)
 {
@@ -894,9 +907,14 @@ add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
 
                /* Set tbl8 entry. */
                for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
-                       lpm->tbl8[i].depth = depth;
-                       lpm->tbl8[i].next_hop = next_hop;
-                       lpm->tbl8[i].valid = VALID;
+                       struct rte_lpm_tbl_entry_v20 new_tbl8_entry = {
+                               .valid = VALID,
+                               .depth = depth,
+                               .valid_group = lpm->tbl8[i].valid_group,
+                       };
+                       new_tbl8_entry.next_hop = next_hop;
+                       __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
                }
 
                /*
@@ -906,13 +924,14 @@ add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
                 */
 
                struct rte_lpm_tbl_entry_v20 new_tbl24_entry = {
-                       { .group_idx = (uint8_t)tbl8_group_index, },
+                       .group_idx = (uint8_t)tbl8_group_index,
                        .valid = VALID,
                        .valid_group = 1,
                        .depth = 0,
                };
 
-               lpm->tbl24[tbl24_index] = new_tbl24_entry;
+               __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
+                               __ATOMIC_RELEASE);
 
        } /* If valid entry but not extended calculate the index into Table8. */
        else if (lpm->tbl24[tbl24_index].valid_group == 0) {
@@ -930,19 +949,29 @@ add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
 
                /* Populate new tbl8 with tbl24 value. */
                for (i = tbl8_group_start; i < tbl8_group_end; i++) {
-                       lpm->tbl8[i].valid = VALID;
-                       lpm->tbl8[i].depth = lpm->tbl24[tbl24_index].depth;
-                       lpm->tbl8[i].next_hop =
-                                       lpm->tbl24[tbl24_index].next_hop;
+                       struct rte_lpm_tbl_entry_v20 new_tbl8_entry = {
+                               .valid = VALID,
+                               .depth = lpm->tbl24[tbl24_index].depth,
+                               .valid_group = lpm->tbl8[i].valid_group,
+                       };
+                       new_tbl8_entry.next_hop =
+                               lpm->tbl24[tbl24_index].next_hop;
+                       __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
                }
 
                tbl8_index = tbl8_group_start + (ip_masked & 0xFF);
 
                /* Insert new rule into the tbl8 entry. */
                for (i = tbl8_index; i < tbl8_index + tbl8_range; i++) {
-                       lpm->tbl8[i].valid = VALID;
-                       lpm->tbl8[i].depth = depth;
-                       lpm->tbl8[i].next_hop = next_hop;
+                       struct rte_lpm_tbl_entry_v20 new_tbl8_entry = {
+                               .valid = VALID,
+                               .depth = depth,
+                               .valid_group = lpm->tbl8[i].valid_group,
+                       };
+                       new_tbl8_entry.next_hop = next_hop;
+                       __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
                }
 
                /*
@@ -952,13 +981,14 @@ add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
                 */
 
                struct rte_lpm_tbl_entry_v20 new_tbl24_entry = {
-                               { .group_idx = (uint8_t)tbl8_group_index, },
+                               .group_idx = (uint8_t)tbl8_group_index,
                                .valid = VALID,
                                .valid_group = 1,
                                .depth = 0,
                };
 
-               lpm->tbl24[tbl24_index] = new_tbl24_entry;
+               __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
+                               __ATOMIC_RELEASE);
 
        } else { /*
                * If it is valid, extended entry calculate the index into tbl8.
@@ -982,7 +1012,8 @@ add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
                                 * Setting tbl8 entry in one go to avoid race
                                 * condition
                                 */
-                               lpm->tbl8[i] = new_tbl8_entry;
+                               __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                               __ATOMIC_RELAXED);
 
                                continue;
                        }
@@ -992,7 +1023,7 @@ add_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked, uint8_t depth,
        return 0;
 }
 
-static inline int32_t
+static __rte_noinline int32_t
 add_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
                uint32_t next_hop)
 {
@@ -1020,9 +1051,14 @@ add_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
 
                /* Set tbl8 entry. */
                for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
-                       lpm->tbl8[i].depth = depth;
-                       lpm->tbl8[i].next_hop = next_hop;
-                       lpm->tbl8[i].valid = VALID;
+                       struct rte_lpm_tbl_entry new_tbl8_entry = {
+                               .valid = VALID,
+                               .depth = depth,
+                               .valid_group = lpm->tbl8[i].valid_group,
+                               .next_hop = next_hop,
+                       };
+                       __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
                }
 
                /*
@@ -1032,13 +1068,17 @@ add_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
                 */
 
                struct rte_lpm_tbl_entry new_tbl24_entry = {
-                       .group_idx = (uint8_t)tbl8_group_index,
+                       .group_idx = tbl8_group_index,
                        .valid = VALID,
                        .valid_group = 1,
                        .depth = 0,
                };
 
-               lpm->tbl24[tbl24_index] = new_tbl24_entry;
+               /* The tbl24 entry must be written only after the
+                * tbl8 entries are written.
+                */
+               __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
+                               __ATOMIC_RELEASE);
 
        } /* If valid entry but not extended calculate the index into Table8. */
        else if (lpm->tbl24[tbl24_index].valid_group == 0) {
@@ -1056,19 +1096,28 @@ add_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
 
                /* Populate new tbl8 with tbl24 value. */
                for (i = tbl8_group_start; i < tbl8_group_end; i++) {
-                       lpm->tbl8[i].valid = VALID;
-                       lpm->tbl8[i].depth = lpm->tbl24[tbl24_index].depth;
-                       lpm->tbl8[i].next_hop =
-                                       lpm->tbl24[tbl24_index].next_hop;
+                       struct rte_lpm_tbl_entry new_tbl8_entry = {
+                               .valid = VALID,
+                               .depth = lpm->tbl24[tbl24_index].depth,
+                               .valid_group = lpm->tbl8[i].valid_group,
+                               .next_hop = lpm->tbl24[tbl24_index].next_hop,
+                       };
+                       __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
                }
 
                tbl8_index = tbl8_group_start + (ip_masked & 0xFF);
 
                /* Insert new rule into the tbl8 entry. */
                for (i = tbl8_index; i < tbl8_index + tbl8_range; i++) {
-                       lpm->tbl8[i].valid = VALID;
-                       lpm->tbl8[i].depth = depth;
-                       lpm->tbl8[i].next_hop = next_hop;
+                       struct rte_lpm_tbl_entry new_tbl8_entry = {
+                               .valid = VALID,
+                               .depth = depth,
+                               .valid_group = lpm->tbl8[i].valid_group,
+                               .next_hop = next_hop,
+                       };
+                       __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                       __ATOMIC_RELAXED);
                }
 
                /*
@@ -1078,13 +1127,17 @@ add_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
                 */
 
                struct rte_lpm_tbl_entry new_tbl24_entry = {
-                               .group_idx = (uint8_t)tbl8_group_index,
+                               .group_idx = tbl8_group_index,
                                .valid = VALID,
                                .valid_group = 1,
                                .depth = 0,
                };
 
-               lpm->tbl24[tbl24_index] = new_tbl24_entry;
+               /* The tbl24 entry must be written only after the
+                * tbl8 entries are written.
+                */
+               __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
+                               __ATOMIC_RELEASE);
 
        } else { /*
                * If it is valid, extended entry calculate the index into tbl8.
@@ -1109,7 +1162,8 @@ add_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
                                 * Setting tbl8 entry in one go to avoid race
                                 * condition
                                 */
-                               lpm->tbl8[i] = new_tbl8_entry;
+                               __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                               __ATOMIC_RELAXED);
 
                                continue;
                        }
@@ -1265,7 +1319,7 @@ BIND_DEFAULT_SYMBOL(rte_lpm_is_rule_present, _v1604, 16.04);
 MAP_STATIC_SYMBOL(int rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip,
                uint8_t depth, uint32_t *next_hop), rte_lpm_is_rule_present_v1604);
 
-static inline int32_t
+static int32_t
 find_previous_rule_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
                uint8_t *sub_rule_depth)
 {
@@ -1287,7 +1341,7 @@ find_previous_rule_v20(struct rte_lpm_v20 *lpm, uint32_t ip, uint8_t depth,
        return -1;
 }
 
-static inline int32_t
+static int32_t
 find_previous_rule_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
                uint8_t *sub_rule_depth)
 {
@@ -1309,7 +1363,7 @@ find_previous_rule_v1604(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
        return -1;
 }
 
-static inline int32_t
+static int32_t
 delete_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
        uint8_t depth, int32_t sub_rule_index, uint8_t sub_rule_depth)
 {
@@ -1332,7 +1386,15 @@ delete_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
 
                        if (lpm->tbl24[i].valid_group == 0 &&
                                        lpm->tbl24[i].depth <= depth) {
-                               lpm->tbl24[i].valid = INVALID;
+                               struct rte_lpm_tbl_entry_v20
+                                       zero_tbl24_entry = {
+                                               .valid = INVALID,
+                                               .depth = 0,
+                                               .valid_group = 0,
+                                       };
+                                       zero_tbl24_entry.next_hop = 0;
+                               __atomic_store(&lpm->tbl24[i],
+                                       &zero_tbl24_entry, __ATOMIC_RELEASE);
                        } else if (lpm->tbl24[i].valid_group == 1) {
                                /*
                                 * If TBL24 entry is extended, then there has
@@ -1359,7 +1421,7 @@ delete_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
                 */
 
                struct rte_lpm_tbl_entry_v20 new_tbl24_entry = {
-                       {.next_hop = lpm->rules_tbl[sub_rule_index].next_hop,},
+                       .next_hop = lpm->rules_tbl[sub_rule_index].next_hop,
                        .valid = VALID,
                        .valid_group = 0,
                        .depth = sub_rule_depth,
@@ -1377,7 +1439,8 @@ delete_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
 
                        if (lpm->tbl24[i].valid_group == 0 &&
                                        lpm->tbl24[i].depth <= depth) {
-                               lpm->tbl24[i] = new_tbl24_entry;
+                               __atomic_store(&lpm->tbl24[i], &new_tbl24_entry,
+                                               __ATOMIC_RELEASE);
                        } else  if (lpm->tbl24[i].valid_group == 1) {
                                /*
                                 * If TBL24 entry is extended, then there has
@@ -1393,7 +1456,9 @@ delete_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
                                        RTE_LPM_TBL8_GROUP_NUM_ENTRIES); j++) {
 
                                        if (lpm->tbl8[j].depth <= depth)
-                                               lpm->tbl8[j] = new_tbl8_entry;
+                                               __atomic_store(&lpm->tbl8[j],
+                                                       &new_tbl8_entry,
+                                                       __ATOMIC_RELAXED);
                                }
                        }
                }
@@ -1402,7 +1467,7 @@ delete_depth_small_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
        return 0;
 }
 
-static inline int32_t
+static int32_t
 delete_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
        uint8_t depth, int32_t sub_rule_index, uint8_t sub_rule_depth)
 {
@@ -1412,6 +1477,7 @@ delete_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
        /* Calculate the range and index into Table24. */
        tbl24_range = depth_to_range(depth);
        tbl24_index = (ip_masked >> 8);
+       struct rte_lpm_tbl_entry zero_tbl24_entry = {0};
 
        /*
         * Firstly check the sub_rule_index. A -1 indicates no replacement rule
@@ -1426,7 +1492,8 @@ delete_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
 
                        if (lpm->tbl24[i].valid_group == 0 &&
                                        lpm->tbl24[i].depth <= depth) {
-                               lpm->tbl24[i].valid = INVALID;
+                               __atomic_store(&lpm->tbl24[i],
+                                       &zero_tbl24_entry, __ATOMIC_RELEASE);
                        } else if (lpm->tbl24[i].valid_group == 1) {
                                /*
                                 * If TBL24 entry is extended, then there has
@@ -1471,7 +1538,8 @@ delete_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
 
                        if (lpm->tbl24[i].valid_group == 0 &&
                                        lpm->tbl24[i].depth <= depth) {
-                               lpm->tbl24[i] = new_tbl24_entry;
+                               __atomic_store(&lpm->tbl24[i], &new_tbl24_entry,
+                                               __ATOMIC_RELEASE);
                        } else  if (lpm->tbl24[i].valid_group == 1) {
                                /*
                                 * If TBL24 entry is extended, then there has
@@ -1487,7 +1555,9 @@ delete_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
                                        RTE_LPM_TBL8_GROUP_NUM_ENTRIES); j++) {
 
                                        if (lpm->tbl8[j].depth <= depth)
-                                               lpm->tbl8[j] = new_tbl8_entry;
+                                               __atomic_store(&lpm->tbl8[j],
+                                                       &new_tbl8_entry,
+                                                       __ATOMIC_RELAXED);
                                }
                        }
                }
@@ -1504,7 +1574,7 @@ delete_depth_small_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
  * Return of value > -1 means tbl8 is in use but has all the same values and
  * thus can be recycled
  */
-static inline int32_t
+static int32_t
 tbl8_recycle_check_v20(struct rte_lpm_tbl_entry_v20 *tbl8,
                uint32_t tbl8_group_start)
 {
@@ -1551,7 +1621,7 @@ tbl8_recycle_check_v20(struct rte_lpm_tbl_entry_v20 *tbl8,
        return -EINVAL;
 }
 
-static inline int32_t
+static int32_t
 tbl8_recycle_check_v1604(struct rte_lpm_tbl_entry *tbl8,
                uint32_t tbl8_group_start)
 {
@@ -1598,7 +1668,7 @@ tbl8_recycle_check_v1604(struct rte_lpm_tbl_entry *tbl8,
        return -EINVAL;
 }
 
-static inline int32_t
+static int32_t
 delete_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
        uint8_t depth, int32_t sub_rule_index, uint8_t sub_rule_depth)
 {
@@ -1643,7 +1713,8 @@ delete_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
                 */
                for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
                        if (lpm->tbl8[i].depth <= depth)
-                               lpm->tbl8[i] = new_tbl8_entry;
+                               __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                               __ATOMIC_RELAXED);
                }
        }
 
@@ -1656,27 +1727,34 @@ delete_depth_big_v20(struct rte_lpm_v20 *lpm, uint32_t ip_masked,
        tbl8_recycle_index = tbl8_recycle_check_v20(lpm->tbl8, tbl8_group_start);
 
        if (tbl8_recycle_index == -EINVAL) {
-               /* Set tbl24 before freeing tbl8 to avoid race condition. */
+               /* Set tbl24 before freeing tbl8 to avoid race condition.
+                * Prevent the free of the tbl8 group from hoisting.
+                */
                lpm->tbl24[tbl24_index].valid = 0;
+               __atomic_thread_fence(__ATOMIC_RELEASE);
                tbl8_free_v20(lpm->tbl8, tbl8_group_start);
        } else if (tbl8_recycle_index > -1) {
                /* Update tbl24 entry. */
                struct rte_lpm_tbl_entry_v20 new_tbl24_entry = {
-                       { .next_hop = lpm->tbl8[tbl8_recycle_index].next_hop, },
+                       .next_hop = lpm->tbl8[tbl8_recycle_index].next_hop,
                        .valid = VALID,
                        .valid_group = 0,
                        .depth = lpm->tbl8[tbl8_recycle_index].depth,
                };
 
-               /* Set tbl24 before freeing tbl8 to avoid race condition. */
-               lpm->tbl24[tbl24_index] = new_tbl24_entry;
+               /* Set tbl24 before freeing tbl8 to avoid race condition.
+                * Prevent the free of the tbl8 group from hoisting.
+                */
+               __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
+                               __ATOMIC_RELAXED);
+               __atomic_thread_fence(__ATOMIC_RELEASE);
                tbl8_free_v20(lpm->tbl8, tbl8_group_start);
        }
 
        return 0;
 }
 
-static inline int32_t
+static int32_t
 delete_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
        uint8_t depth, int32_t sub_rule_index, uint8_t sub_rule_depth)
 {
@@ -1721,7 +1799,8 @@ delete_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
                 */
                for (i = tbl8_index; i < (tbl8_index + tbl8_range); i++) {
                        if (lpm->tbl8[i].depth <= depth)
-                               lpm->tbl8[i] = new_tbl8_entry;
+                               __atomic_store(&lpm->tbl8[i], &new_tbl8_entry,
+                                               __ATOMIC_RELAXED);
                }
        }
 
@@ -1734,8 +1813,11 @@ delete_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
        tbl8_recycle_index = tbl8_recycle_check_v1604(lpm->tbl8, tbl8_group_start);
 
        if (tbl8_recycle_index == -EINVAL) {
-               /* Set tbl24 before freeing tbl8 to avoid race condition. */
+               /* Set tbl24 before freeing tbl8 to avoid race condition.
+                * Prevent the free of the tbl8 group from hoisting.
+                */
                lpm->tbl24[tbl24_index].valid = 0;
+               __atomic_thread_fence(__ATOMIC_RELEASE);
                tbl8_free_v1604(lpm->tbl8, tbl8_group_start);
        } else if (tbl8_recycle_index > -1) {
                /* Update tbl24 entry. */
@@ -1746,8 +1828,12 @@ delete_depth_big_v1604(struct rte_lpm *lpm, uint32_t ip_masked,
                        .depth = lpm->tbl8[tbl8_recycle_index].depth,
                };
 
-               /* Set tbl24 before freeing tbl8 to avoid race condition. */
-               lpm->tbl24[tbl24_index] = new_tbl24_entry;
+               /* Set tbl24 before freeing tbl8 to avoid race condition.
+                * Prevent the free of the tbl8 group from hoisting.
+                */
+               __atomic_store(&lpm->tbl24[tbl24_index], &new_tbl24_entry,
+                               __ATOMIC_RELAXED);
+               __atomic_thread_fence(__ATOMIC_RELEASE);
                tbl8_free_v1604(lpm->tbl8, tbl8_group_start);
        }
 #undef group_idx