examples/ip_reassembly: initial import
authorIntel <intel.com>
Mon, 3 Jun 2013 00:00:00 +0000 (00:00 +0000)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Tue, 17 Sep 2013 12:16:09 +0000 (14:16 +0200)
Signed-off-by: Intel
examples/ip_reassembly/Makefile [new file with mode: 0644]
examples/ip_reassembly/ipv4_frag_tbl.h [new file with mode: 0644]
examples/ip_reassembly/ipv4_rsmbl.h [new file with mode: 0644]
examples/ip_reassembly/main.c [new file with mode: 0644]
examples/ip_reassembly/main.h [new file with mode: 0644]
lib/librte_net/rte_ip.h

diff --git a/examples/ip_reassembly/Makefile b/examples/ip_reassembly/Makefile
new file mode 100644 (file)
index 0000000..6661b50
--- /dev/null
@@ -0,0 +1,59 @@
+#   BSD LICENSE
+# 
+#   Copyright(c) 2010-2013 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.
+# 
+#
+
+ifeq ($(RTE_SDK),)
+$(error "Please define RTE_SDK environment variable")
+endif
+
+# Default target, can be overriden by command line or environment
+RTE_TARGET ?= x86_64-default-linuxapp-gcc
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+# binary name
+APP = ip_reassembly
+
+# all source are stored in SRCS-y
+SRCS-y := main.c
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+# workaround for a gcc bug with noreturn attribute
+# http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12603
+ifeq ($(CONFIG_RTE_TOOLCHAIN_GCC),y)
+CFLAGS_main.o += -Wno-return-type
+CFLAGS_main.o += -DIPV4_FRAG_TBL_STAT
+endif
+
+include $(RTE_SDK)/mk/rte.extapp.mk
diff --git a/examples/ip_reassembly/ipv4_frag_tbl.h b/examples/ip_reassembly/ipv4_frag_tbl.h
new file mode 100644 (file)
index 0000000..33b493b
--- /dev/null
@@ -0,0 +1,400 @@
+/*-
+ *   BSD LICENSE
+ * 
+ *   Copyright(c) 2010-2013 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.
+ * 
+ */
+
+#ifndef _IPV4_FRAG_TBL_H_
+#define _IPV4_FRAG_TBL_H_
+
+/**
+ * @file
+ * IPv4 fragments table.
+ *
+ * Implementation of IPv4 fragment table create/destroy/find/update.
+ *
+ */
+
+/*
+ * The ipv4_frag_tbl is a simple hash table:
+ * The basic idea is to use two hash functions and <bucket_entries>
+ * associativity. This provides 2 * <bucket_entries> possible locations in
+ * the hash table for each key. Sort of simplified Cuckoo hashing,
+ * when the collision occurs and all 2 * <bucket_entries> are occupied,
+ * instead of resinserting existing keys into alternative locations, we just
+ * return a faiure.
+ * Another thing timing: entries that resides in the table longer then
+ * <max_cycles> are considered as invalid, and could be removed/replaced
+ * byt the new ones. 
+ * <key, data> pair is stored together, all add/update/lookup opearions are not
+ * MT safe.
+ */
+
+#include <rte_jhash.h>
+#ifdef RTE_MACHINE_CPUFLAG_SSE4_2
+#include <rte_hash_crc.h>
+#endif /* RTE_MACHINE_CPUFLAG_SSE4_2 */
+
+#define        PRIME_VALUE     0xeaad8405
+
+TAILQ_HEAD(ipv4_pkt_list, ipv4_frag_pkt);
+
+struct ipv4_frag_tbl_stat {
+       uint64_t find_num;      /* total # of find/insert attempts. */
+       uint64_t add_num;       /* # of add ops. */
+       uint64_t del_num;       /* # of del ops. */
+       uint64_t reuse_num;     /* # of reuse (del/add) ops. */
+       uint64_t fail_total;    /* total # of add failures. */
+       uint64_t fail_nospace;  /* # of 'no space' add failures. */
+} __rte_cache_aligned;
+
+struct ipv4_frag_tbl {
+       uint64_t             max_cycles;      /* ttl for table entries. */
+       uint32_t             entry_mask;      /* hash value mask. */
+       uint32_t             max_entries;     /* max entries allowed. */
+       uint32_t             use_entries;     /* entries in use. */
+       uint32_t             bucket_entries;  /* hash assocaitivity. */
+       uint32_t             nb_entries;      /* total size of the table. */
+       uint32_t             nb_buckets;      /* num of associativity lines. */
+       struct ipv4_frag_pkt *last;           /* last used entry. */
+       struct ipv4_pkt_list lru;             /* LRU list for table entries. */
+       struct ipv4_frag_tbl_stat stat;       /* statistics counters. */
+       struct ipv4_frag_pkt pkt[0];          /* hash table. */
+};
+
+#define        IPV4_FRAG_TBL_POS(tbl, sig)     \
+       ((tbl)->pkt + ((sig) & (tbl)->entry_mask))
+
+#define        IPV4_FRAG_HASH_FNUM     2
+
+#ifdef IPV4_FRAG_TBL_STAT
+#define        IPV4_FRAG_TBL_STAT_UPDATE(s, f, v)      ((s)->f += (v))
+#else
+#define        IPV4_FRAG_TBL_STAT_UPDATE(s, f, v)      do {} while (0)
+#endif /* IPV4_FRAG_TBL_STAT */
+
+static inline void
+ipv4_frag_hash(const struct ipv4_frag_key *key, uint32_t *v1, uint32_t *v2)
+{
+       uint32_t v;
+       const uint32_t *p;
+
+       p = (const uint32_t *)&key->src_dst;
+
+#ifdef RTE_MACHINE_CPUFLAG_SSE4_2
+       v = rte_hash_crc_4byte(p[0], PRIME_VALUE);
+       v = rte_hash_crc_4byte(p[1], v);
+       v = rte_hash_crc_4byte(key->id, v);
+#else
+
+       v = rte_jhash_3words(p[0], p[1], key->id, PRIME_VALUE);
+#endif /* RTE_MACHINE_CPUFLAG_SSE4_2 */
+
+       *v1 =  v;
+       *v2 = (v << 7) + (v >> 14);
+}
+
+/*
+ * Update the table, after we finish processing it's entry.
+ */
+static inline void
+ipv4_frag_inuse(struct ipv4_frag_tbl *tbl, const struct  ipv4_frag_pkt *fp)
+{
+       if (IPV4_FRAG_KEY_EMPTY(&fp->key)) {
+               TAILQ_REMOVE(&tbl->lru, fp, lru);
+               tbl->use_entries--;
+       }
+}
+
+/*
+ * For the given key, try to find an existing entry.
+ * If such entry doesn't exist, will return free and/or timed-out entry,
+ * that can be used for that key.
+ */
+static inline struct  ipv4_frag_pkt *
+ipv4_frag_lookup(struct ipv4_frag_tbl *tbl,
+       const struct ipv4_frag_key *key, uint64_t tms,
+       struct ipv4_frag_pkt **free, struct ipv4_frag_pkt **stale)
+{
+       struct ipv4_frag_pkt *p1, *p2;
+       struct ipv4_frag_pkt *empty, *old;
+       uint64_t max_cycles;
+       uint32_t i, assoc, sig1, sig2;
+
+       empty = NULL;
+       old = NULL;
+
+       max_cycles = tbl->max_cycles;
+       assoc = tbl->bucket_entries;
+
+       if (tbl->last != NULL && IPV4_FRAG_KEY_CMP(&tbl->last->key, key) == 0)
+               return (tbl->last);
+
+       ipv4_frag_hash(key, &sig1, &sig2);
+       p1 = IPV4_FRAG_TBL_POS(tbl, sig1);
+       p2 = IPV4_FRAG_TBL_POS(tbl, sig2);
+
+       for (i = 0; i != assoc; i++) {
+
+               IPV4_FRAG_LOG(DEBUG, "%s:%d:\n"
+                "tbl: %p, max_entries: %u, use_entries: %u\n"
+                "ipv4_frag_pkt line0: %p, index: %u from %u\n"
+               "key: <%" PRIx64 ", %#x>, start: %" PRIu64 "\n",
+                __func__, __LINE__,
+                tbl, tbl->max_entries, tbl->use_entries,
+                p1, i, assoc,
+               p1[i].key.src_dst, p1[i].key.id, p1[i].start);
+
+               if (IPV4_FRAG_KEY_CMP(&p1[i].key, key) == 0)
+                       return (p1 + i);
+               else if (IPV4_FRAG_KEY_EMPTY(&p1[i].key))
+                       empty = (empty == NULL) ? (p1 + i) : empty;
+               else if (max_cycles + p1[i].start < tms)
+                       old = (old == NULL) ? (p1 + i) : old;
+
+               IPV4_FRAG_LOG(DEBUG, "%s:%d:\n"
+                "tbl: %p, max_entries: %u, use_entries: %u\n"
+                "ipv4_frag_pkt line1: %p, index: %u from %u\n"
+               "key: <%" PRIx64 ", %#x>, start: %" PRIu64 "\n",
+                __func__, __LINE__,
+                tbl, tbl->max_entries, tbl->use_entries,
+                p2, i, assoc,
+               p2[i].key.src_dst, p2[i].key.id, p2[i].start);
+
+               if (IPV4_FRAG_KEY_CMP(&p2[i].key, key) == 0)
+                       return (p2 + i);
+               else if (IPV4_FRAG_KEY_EMPTY(&p2[i].key))
+                       empty = (empty == NULL) ?( p2 + i) : empty;
+               else if (max_cycles + p2[i].start < tms)
+                       old = (old == NULL) ? (p2 + i) : old;
+       }
+
+       *free = empty;
+       *stale = old;
+       return (NULL);
+}
+
+static inline void
+ipv4_frag_tbl_del(struct ipv4_frag_tbl *tbl,  struct ipv4_frag_pkt *fp)
+{
+       ipv4_frag_free(fp);
+       IPV4_FRAG_KEY_INVALIDATE(&fp->key);
+       TAILQ_REMOVE(&tbl->lru, fp, lru);
+       tbl->use_entries--;
+       IPV4_FRAG_TBL_STAT_UPDATE(&tbl->stat, del_num, 1);
+}
+
+static inline void
+ipv4_frag_tbl_add(struct ipv4_frag_tbl *tbl,  struct ipv4_frag_pkt *fp,
+       const struct ipv4_frag_key *key, uint64_t tms)
+{
+       fp->key = key[0];
+       ipv4_frag_reset(fp, tms);
+       TAILQ_INSERT_TAIL(&tbl->lru, fp, lru);
+       tbl->use_entries++;
+       IPV4_FRAG_TBL_STAT_UPDATE(&tbl->stat, add_num, 1);
+}
+
+static inline void
+ipv4_frag_tbl_reuse(struct ipv4_frag_tbl *tbl,  struct ipv4_frag_pkt *fp,
+       uint64_t tms)
+{
+       ipv4_frag_free(fp);
+       ipv4_frag_reset(fp, tms);
+       TAILQ_REMOVE(&tbl->lru, fp, lru);
+       TAILQ_INSERT_TAIL(&tbl->lru, fp, lru);
+       IPV4_FRAG_TBL_STAT_UPDATE(&tbl->stat, reuse_num, 1);
+}
+
+/*
+ * Find an entry in the table for the corresponding fragment.
+ * If such entry is not present, then allocate a new one.
+ * If the entry is stale, then free and reuse it.
+ */
+static inline struct ipv4_frag_pkt *
+ipv4_frag_find(struct ipv4_frag_tbl *tbl, const struct ipv4_frag_key *key,
+       uint64_t tms)
+{
+       struct ipv4_frag_pkt *pkt, *free, *stale, *lru;
+       uint64_t max_cycles;
+
+       /*
+        * Actually the two line below are totally redundant.
+        * they are here, just to make gcc 4.6 happy.
+        */
+       free = NULL;
+       stale = NULL;
+       max_cycles = tbl->max_cycles;
+
+       IPV4_FRAG_TBL_STAT_UPDATE(&tbl->stat, find_num, 1);
+
+       if ((pkt = ipv4_frag_lookup(tbl, key, tms, &free, &stale)) == NULL) {
+
+               /*timed-out entry, free and invalidate it*/
+               if (stale != NULL) {
+                       ipv4_frag_tbl_del(tbl, stale);
+                       free = stale;
+
+               /*
+                * we found a free entry, check if we can use it.
+                * If we run out of free entries in the table, then
+                * check if we have a timed out entry to delete. 
+                */
+               } else if (free != NULL &&
+                               tbl->max_entries <= tbl->use_entries) {
+                       lru = TAILQ_FIRST(&tbl->lru);
+                       if (max_cycles + lru->start < tms) {
+                               ipv4_frag_tbl_del(tbl,  lru);
+                       } else {
+                               free = NULL;
+                               IPV4_FRAG_TBL_STAT_UPDATE(&tbl->stat,
+                                       fail_nospace, 1);
+                       }
+               }
+
+               /* found a free entry to reuse. */
+               if (free != NULL) {
+                       ipv4_frag_tbl_add(tbl,  free, key, tms);
+                       pkt = free;
+               }
+
+       /*
+        * we found the flow, but it is already timed out,
+        * so free associated resources, reposition it in the LRU list,
+        * and reuse it.
+        */
+       } else if (max_cycles + pkt->start < tms) {
+               ipv4_frag_tbl_reuse(tbl,  pkt, tms);
+       }
+
+       IPV4_FRAG_TBL_STAT_UPDATE(&tbl->stat, fail_total, (pkt == NULL));
+
+       tbl->last = pkt;
+       return (pkt);
+}
+
+/*
+ * Create a new IPV4 Frag table.
+ * @param bucket_num
+ *  Number of buckets in the hash table.
+ * @param bucket_entries
+ *  Number of entries per bucket (e.g. hash associativity).
+ *  Should be power of two.
+ * @param max_entries
+ *   Maximum number of entries that could be stored in the table.
+ *   The value should be less or equal then bucket_num * bucket_entries.
+ * @param max_cycles
+ *   Maximum TTL in cycles for each fragmented packet.
+ * @param socket_id
+ *  The *socket_id* argument is the socket identifier in the case of
+ *  NUMA. The value can be *SOCKET_ID_ANY* if there is no NUMA constraints.
+ * @return
+ *   The pointer to the new allocated mempool, on success. NULL on error.
+ */
+static struct ipv4_frag_tbl *
+ipv4_frag_tbl_create(uint32_t bucket_num, uint32_t bucket_entries,
+       uint32_t max_entries, uint64_t max_cycles, int socket_id)
+{
+       struct ipv4_frag_tbl *tbl;
+       size_t sz;
+       uint64_t nb_entries;
+
+       nb_entries = rte_align32pow2(bucket_num);
+       nb_entries *= bucket_entries;
+       nb_entries *= IPV4_FRAG_HASH_FNUM;
+
+       /* check input parameters. */
+       if (rte_is_power_of_2(bucket_entries) == 0 ||
+                       nb_entries > UINT32_MAX || nb_entries == 0 ||
+                       nb_entries < max_entries) {
+               RTE_LOG(ERR, USER1, "%s: invalid input parameter\n", __func__);
+               return (NULL);
+       }
+
+       sz = sizeof (*tbl) + nb_entries * sizeof (tbl->pkt[0]);
+       if ((tbl = rte_zmalloc_socket(__func__, sz, CACHE_LINE_SIZE,
+                       socket_id)) == NULL) {
+               RTE_LOG(ERR, USER1,
+                       "%s: allocation of %zu bytes at socket %d failed do\n",
+                       __func__, sz, socket_id);
+               return (NULL);
+       }
+
+       RTE_LOG(INFO, USER1, "%s: allocated of %zu bytes at socket %d\n",
+               __func__, sz, socket_id); 
+
+       tbl->max_cycles = max_cycles;
+       tbl->max_entries = max_entries;
+       tbl->nb_entries = (uint32_t)nb_entries;
+       tbl->nb_buckets = bucket_num;
+       tbl->bucket_entries = bucket_entries;
+       tbl->entry_mask = (tbl->nb_entries - 1) & ~(tbl->bucket_entries  - 1);
+
+       TAILQ_INIT(&(tbl->lru));
+       return (tbl);
+}
+
+static inline void
+ipv4_frag_tbl_destroy( struct ipv4_frag_tbl *tbl)
+{
+       rte_free(tbl);
+}
+
+static void
+ipv4_frag_tbl_dump_stat(FILE *f, const struct ipv4_frag_tbl *tbl)
+{
+       uint64_t fail_total, fail_nospace;
+
+       fail_total = tbl->stat.fail_total;
+       fail_nospace = tbl->stat.fail_nospace;
+
+       fprintf(f, "max entries:\t%u;\n"
+               "entries in use:\t%u;\n"
+               "finds/inserts:\t%" PRIu64 ";\n"
+               "entries added:\t%" PRIu64 ";\n"
+               "entries deleted by timeout:\t%" PRIu64 ";\n"
+               "entries reused by timeout:\t%" PRIu64 ";\n"
+               "total add failures:\t%" PRIu64 ";\n"
+               "add no-space failures:\t%" PRIu64 ";\n"
+               "add hash-collisions failures:\t%" PRIu64 ";\n",
+               tbl->max_entries,
+               tbl->use_entries,
+               tbl->stat.find_num,
+               tbl->stat.add_num,
+               tbl->stat.del_num,
+               tbl->stat.reuse_num,
+               fail_total,
+               fail_nospace,
+               fail_total - fail_nospace);
+}
+
+
+#endif /* _IPV4_FRAG_TBL_H_ */
diff --git a/examples/ip_reassembly/ipv4_rsmbl.h b/examples/ip_reassembly/ipv4_rsmbl.h
new file mode 100644 (file)
index 0000000..bc94030
--- /dev/null
@@ -0,0 +1,393 @@
+/*-
+ *   BSD LICENSE
+ * 
+ *   Copyright(c) 2010-2013 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.
+ * 
+ */
+
+#ifndef _IPV4_RSMBL_H_
+#define _IPV4_RSMBL_H_
+
+/**
+ * @file
+ * IPv4 reassemble
+ *
+ * Implementation of IPv4 reassemble.
+ *
+ */
+
+enum {
+       LAST_FRAG_IDX,
+       FIRST_FRAG_IDX,
+       MIN_FRAG_NUM,
+       MAX_FRAG_NUM = 4,
+};
+
+struct ipv4_frag {
+       uint16_t ofs;
+       uint16_t len;
+       struct rte_mbuf *mb;
+};
+
+/*
+ * Use <src addr, dst_addr, id> to uniquely indetify fragmented datagram.
+ */
+struct ipv4_frag_key {
+       uint64_t  src_dst;
+       uint32_t  id;
+};
+
+#define        IPV4_FRAG_KEY_INVALIDATE(k)     ((k)->src_dst = 0)
+#define        IPV4_FRAG_KEY_EMPTY(k)          ((k)->src_dst == 0)
+
+#define        IPV4_FRAG_KEY_CMP(k1, k2)       \
+       (((k1)->src_dst ^ (k2)->src_dst) | ((k1)->id ^ (k2)->id))
+       
+
+/*
+ * Fragmented packet to reassemble.
+ * First two entries in the frags[] array are for the last and first fragments.
+ */
+struct ipv4_frag_pkt {
+       TAILQ_ENTRY(ipv4_frag_pkt) lru;   /* LRU list */
+       struct ipv4_frag_key key;
+       uint64_t             start;       /* creation timestamp */
+       uint32_t             total_size;  /* expected reassembled size */
+       uint32_t             frag_size;   /* size of fragments received */
+       uint32_t             last_idx;    /* index of next entry to fill */
+       struct ipv4_frag     frags[MAX_FRAG_NUM];
+} __rte_cache_aligned;
+
+/* logging macros. */
+
+#ifdef IPV4_FRAG_DEBUG
+#define        IPV4_FRAG_LOG(lvl, fmt, args...)        RTE_LOG(lvl, USER1, fmt, ##args)
+#else
+#define        IPV4_FRAG_LOG(lvl, fmt, args...)        do {} while(0)
+#endif /* IPV4_FRAG_DEBUG */
+
+
+static inline void
+ipv4_frag_reset(struct ipv4_frag_pkt *fp, uint64_t tms)
+{
+       static const struct ipv4_frag zero_frag = {
+               .ofs = 0,
+               .len = 0,
+               .mb = NULL,
+       };
+
+       fp->start = tms;
+       fp->total_size = UINT32_MAX;
+       fp->frag_size = 0;
+       fp->last_idx = MIN_FRAG_NUM;
+       fp->frags[LAST_FRAG_IDX] = zero_frag;
+       fp->frags[FIRST_FRAG_IDX] = zero_frag;
+}
+
+static inline void
+ipv4_frag_free(struct ipv4_frag_pkt *fp)
+{
+       uint32_t i;
+
+       for (i = 0; i != fp->last_idx; i++) {
+               if (fp->frags[i].mb != NULL) {
+                       rte_pktmbuf_free(fp->frags[i].mb);
+                       fp->frags[i].mb = NULL;
+               }
+       }
+
+       fp->last_idx = 0;
+}
+
+/*
+ * Helper function.
+ * Takes 2 mbufs that represents two framents of the same packet and
+ * chains them into one mbuf.
+ */
+static inline void
+ipv4_frag_chain(struct rte_mbuf *mn, struct rte_mbuf *mp)
+{
+       struct rte_mbuf *ms;
+
+       /* adjust start of the last fragment data. */
+       rte_pktmbuf_adj(mp, (uint16_t)(mp->pkt.vlan_macip.f.l2_len +
+               mp->pkt.vlan_macip.f.l3_len));
+                               
+       /* chain two fragments. */
+       ms = rte_pktmbuf_lastseg(mn);
+       ms->pkt.next = mp;
+
+       /* accumulate number of segments and total length. */
+       mn->pkt.nb_segs = (uint8_t)(mn->pkt.nb_segs + mp->pkt.nb_segs);
+       mn->pkt.pkt_len += mp->pkt.pkt_len;
+                                       
+       /* reset pkt_len and nb_segs for chained fragment. */
+       mp->pkt.pkt_len = mp->pkt.data_len;
+       mp->pkt.nb_segs = 1;
+}
+
+/*
+ * Reassemble fragments into one packet.
+ */
+static inline struct rte_mbuf *
+ipv4_frag_reassemble(const struct ipv4_frag_pkt *fp)
+{
+       struct ipv4_hdr *ip_hdr;
+       struct rte_mbuf *m, *prev;
+       uint32_t i, n, ofs, first_len;
+
+       first_len = fp->frags[FIRST_FRAG_IDX].len;
+       n = fp->last_idx - 1;
+
+       /*start from the last fragment. */
+       m = fp->frags[LAST_FRAG_IDX].mb;
+       ofs = fp->frags[LAST_FRAG_IDX].ofs;
+
+       while (ofs != first_len) {
+
+               prev = m;
+
+               for (i = n; i != FIRST_FRAG_IDX && ofs != first_len; i--) {
+
+                       /* previous fragment found. */
+                       if(fp->frags[i].ofs + fp->frags[i].len == ofs) {
+
+                               ipv4_frag_chain(fp->frags[i].mb, m);
+
+                               /* update our last fragment and offset. */
+                               m = fp->frags[i].mb;
+                               ofs = fp->frags[i].ofs;
+                       }
+               }
+
+               /* error - hole in the packet. */
+               if (m == prev) {
+                       return (NULL);
+               }
+       }
+
+       /* chain with the first fragment. */
+       ipv4_frag_chain(fp->frags[FIRST_FRAG_IDX].mb, m);
+       m = fp->frags[FIRST_FRAG_IDX].mb;
+
+       /* update mbuf fields for reassembled packet. */
+       m->ol_flags |= PKT_TX_IP_CKSUM;
+
+       /* update ipv4 header for the reassmebled packet */
+       ip_hdr = (struct ipv4_hdr*)(rte_pktmbuf_mtod(m, uint8_t *) +
+               m->pkt.vlan_macip.f.l2_len);
+
+       ip_hdr->total_length = rte_cpu_to_be_16((uint16_t)(fp->total_size +
+               m->pkt.vlan_macip.f.l3_len));
+       ip_hdr->fragment_offset = (uint16_t)(ip_hdr->fragment_offset &
+               rte_cpu_to_be_16(IPV4_HDR_DF_FLAG));
+       ip_hdr->hdr_checksum = 0;
+
+       return (m);
+}
+
+static inline struct rte_mbuf *
+ipv4_frag_process(struct ipv4_frag_pkt *fp, struct rte_mbuf *mb,
+       uint16_t ofs, uint16_t len, uint16_t more_frags)
+{
+       uint32_t idx;
+
+       fp->frag_size += len;
+
+       /* this is the first fragment. */
+       if (ofs == 0) {
+               idx = (fp->frags[FIRST_FRAG_IDX].mb == NULL) ?
+                       FIRST_FRAG_IDX : UINT32_MAX;
+
+       /* this is the last fragment. */
+       } else if (more_frags == 0) {
+               fp->total_size = ofs + len;
+               idx = (fp->frags[LAST_FRAG_IDX].mb == NULL) ?
+                       LAST_FRAG_IDX : UINT32_MAX;
+
+       /* this is the intermediate fragment. */
+       } else if ((idx = fp->last_idx) <
+               sizeof (fp->frags) / sizeof (fp->frags[0])) {
+               fp->last_idx++;
+       }
+
+       /*
+        * errorneous packet: either exceeed max allowed number of fragments,
+        * or duplicate first/last fragment encountered.
+        */
+       if (idx >= sizeof (fp->frags) / sizeof (fp->frags[0])) {
+
+               /* report an error. */
+               IPV4_FRAG_LOG(DEBUG, "%s:%d invalid fragmented packet:\n"
+                       "ipv4_frag_pkt: %p, key: <%" PRIx64 ", %#x>, "
+                       "total_size: %u, frag_size: %u, last_idx: %u\n"
+                       "first fragment: ofs: %u, len: %u\n"
+                       "last fragment: ofs: %u, len: %u\n\n",
+                       __func__, __LINE__,
+                       fp, fp->key.src_dst, fp->key.id,
+                       fp->total_size, fp->frag_size, fp->last_idx,
+                       fp->frags[FIRST_FRAG_IDX].ofs,
+                       fp->frags[FIRST_FRAG_IDX].len,
+                       fp->frags[LAST_FRAG_IDX].ofs,
+                       fp->frags[LAST_FRAG_IDX].len);
+
+               /* free all fragments, invalidate the entry. */
+               ipv4_frag_free(fp);
+               IPV4_FRAG_KEY_INVALIDATE(&fp->key);
+               rte_pktmbuf_free(mb);
+
+               return (NULL);
+       }
+
+       fp->frags[idx].ofs = ofs;
+       fp->frags[idx].len = len;
+       fp->frags[idx].mb = mb;
+
+       mb = NULL;
+
+       /* not all fragments are collected yet. */
+       if (likely (fp->frag_size < fp->total_size)) {
+               return (mb);
+
+       /* if we collected all fragments, then try to reassemble. */
+       } else if (fp->frag_size == fp->total_size &&
+                       fp->frags[FIRST_FRAG_IDX].mb != NULL) {
+               mb = ipv4_frag_reassemble(fp);
+       }
+
+       /* errorenous set of fragments. */
+       if (mb == NULL) {
+
+               /* report an error. */
+               IPV4_FRAG_LOG(DEBUG, "%s:%d invalid fragmented packet:\n"
+                       "ipv4_frag_pkt: %p, key: <%" PRIx64 ", %#x>, "
+                       "total_size: %u, frag_size: %u, last_idx: %u\n"
+                       "first fragment: ofs: %u, len: %u\n"
+                       "last fragment: ofs: %u, len: %u\n\n",
+                       __func__, __LINE__,
+                       fp, fp->key.src_dst, fp->key.id,
+                       fp->total_size, fp->frag_size, fp->last_idx,
+                       fp->frags[FIRST_FRAG_IDX].ofs,
+                       fp->frags[FIRST_FRAG_IDX].len,
+                       fp->frags[LAST_FRAG_IDX].ofs,
+                       fp->frags[LAST_FRAG_IDX].len);
+
+               /* free associated resources. */
+               ipv4_frag_free(fp);
+       }
+
+       /* we are done with that entry, invalidate it. */
+       IPV4_FRAG_KEY_INVALIDATE(&fp->key);
+       return (mb);
+}
+
+#include "ipv4_frag_tbl.h"
+
+/*
+ * Process new mbuf with fragment of IPV4 packet.
+ * Incoming mbuf should have it's l2_len/l3_len fields setuped correclty.
+ * @param tbl
+ *   Table where to lookup/add the fragmented packet.
+ * @param mb
+ *   Incoming mbuf with IPV4 fragment.
+ * @param tms
+ *   Fragment arrival timestamp.
+ * @param ip_hdr
+ *   Pointer to the IPV4 header inside the fragment.
+ * @param ip_ofs
+ *   Fragment's offset (as extracted from the header).
+ * @param ip_flag
+ *   Fragment's MF flag.
+ * @return  
+ *   Pointer to mbuf for reassebled packet, or NULL if:
+ *   - an error occured.
+ *   - not all fragments of the packet are collected yet.
+ */
+static inline struct rte_mbuf *
+ipv4_frag_mbuf(struct ipv4_frag_tbl *tbl, struct rte_mbuf *mb, uint64_t tms,
+       struct ipv4_hdr *ip_hdr, uint16_t ip_ofs, uint16_t ip_flag)
+{
+       struct ipv4_frag_pkt *fp;
+       struct ipv4_frag_key key;
+       const uint64_t *psd;
+       uint16_t ip_len;
+
+       psd = (uint64_t *)&ip_hdr->src_addr;
+       key.src_dst = psd[0];
+       key.id = ip_hdr->packet_id;
+
+       ip_ofs *= IPV4_HDR_OFFSET_UNITS;
+       ip_len = (uint16_t)(rte_be_to_cpu_16(ip_hdr->total_length) -
+               mb->pkt.vlan_macip.f.l3_len);
+
+       IPV4_FRAG_LOG(DEBUG, "%s:%d:\n"
+               "mbuf: %p, tms: %" PRIu64
+               ", key: <%" PRIx64 ", %#x>, ofs: %u, len: %u, flags: %#x\n"
+               "tbl: %p, max_cycles: %" PRIu64 ", entry_mask: %#x, "
+               "max_entries: %u, use_entries: %u\n\n",
+               __func__, __LINE__,
+               mb, tms, key.src_dst, key.id, ip_ofs, ip_len, ip_flag,
+               tbl, tbl->max_cycles, tbl->entry_mask, tbl->max_entries,
+               tbl->use_entries);
+
+       /* try to find/add entry into the fragment's table. */
+       if ((fp = ipv4_frag_find(tbl, &key, tms)) == NULL) {
+               rte_pktmbuf_free(mb);
+               return (NULL);
+       }
+
+       IPV4_FRAG_LOG(DEBUG, "%s:%d:\n"
+               "tbl: %p, max_entries: %u, use_entries: %u\n"
+               "ipv4_frag_pkt: %p, key: <%" PRIx64 ", %#x>, start: %" PRIu64
+               ", total_size: %u, frag_size: %u, last_idx: %u\n\n",
+               __func__, __LINE__,
+               tbl, tbl->max_entries, tbl->use_entries,
+               fp, fp->key.src_dst, fp->key.id, fp->start,
+               fp->total_size, fp->frag_size, fp->last_idx);
+               
+
+       /* process the fragmented packet. */
+       mb = ipv4_frag_process(fp, mb, ip_ofs, ip_len, ip_flag);
+       ipv4_frag_inuse(tbl, fp);
+
+       IPV4_FRAG_LOG(DEBUG, "%s:%d:\n"
+               "mbuf: %p\n"
+               "tbl: %p, max_entries: %u, use_entries: %u\n"
+               "ipv4_frag_pkt: %p, key: <%" PRIx64 ", %#x>, start: %" PRIu64
+               ", total_size: %u, frag_size: %u, last_idx: %u\n\n",
+               __func__, __LINE__, mb,
+               tbl, tbl->max_entries, tbl->use_entries,
+               fp, fp->key.src_dst, fp->key.id, fp->start,
+               fp->total_size, fp->frag_size, fp->last_idx);
+
+       return (mb);
+}
+
+#endif /* _IPV4_RSMBL_H_ */
diff --git a/examples/ip_reassembly/main.c b/examples/ip_reassembly/main.c
new file mode 100644 (file)
index 0000000..e18dfe8
--- /dev/null
@@ -0,0 +1,1624 @@
+/*-
+ *   BSD LICENSE
+ * 
+ *   Copyright(c) 2010-2013 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.
+ * 
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <inttypes.h>
+#include <sys/types.h>
+#include <string.h>
+#include <sys/queue.h>
+#include <stdarg.h>
+#include <errno.h>
+#include <getopt.h>
+#include <signal.h>
+
+#include <rte_common.h>
+#include <rte_byteorder.h>
+#include <rte_log.h>
+#include <rte_memory.h>
+#include <rte_memcpy.h>
+#include <rte_memzone.h>
+#include <rte_tailq.h>
+#include <rte_eal.h>
+#include <rte_per_lcore.h>
+#include <rte_launch.h>
+#include <rte_atomic.h>
+#include <rte_cycles.h>
+#include <rte_prefetch.h>
+#include <rte_lcore.h>
+#include <rte_per_lcore.h>
+#include <rte_branch_prediction.h>
+#include <rte_interrupts.h>
+#include <rte_pci.h>
+#include <rte_random.h>
+#include <rte_debug.h>
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#include <rte_ring.h>
+#include <rte_mempool.h>
+#include <rte_mbuf.h>
+#include <rte_malloc.h>
+#include <rte_ip.h>
+#include <rte_tcp.h>
+#include <rte_udp.h>
+#include <rte_string_fns.h>
+
+#include "main.h"
+#include "ipv4_rsmbl.h"
+
+#define APP_LOOKUP_EXACT_MATCH          0
+#define APP_LOOKUP_LPM                  1
+#define DO_RFC_1812_CHECKS
+
+#ifndef APP_LOOKUP_METHOD
+#define APP_LOOKUP_METHOD             APP_LOOKUP_LPM
+#endif
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_EXACT_MATCH)
+#include <rte_hash.h>
+#elif (APP_LOOKUP_METHOD == APP_LOOKUP_LPM)
+#include <rte_lpm.h>
+#include <rte_lpm6.h>
+#else
+#error "APP_LOOKUP_METHOD set to incorrect value"
+#endif
+
+#ifndef IPv6_BYTES
+#define IPv6_BYTES_FMT "%02x%02x:%02x%02x:%02x%02x:%02x%02x:"\
+                       "%02x%02x:%02x%02x:%02x%02x:%02x%02x"
+#define IPv6_BYTES(addr) \
+       addr[0],  addr[1], addr[2],  addr[3], \
+       addr[4],  addr[5], addr[6],  addr[7], \
+       addr[8],  addr[9], addr[10], addr[11],\
+       addr[12], addr[13],addr[14], addr[15]
+#endif
+
+
+#define RTE_LOGTYPE_L3FWD RTE_LOGTYPE_USER1
+
+#define MAX_PORTS      RTE_MAX_ETHPORTS
+
+#define MAX_JUMBO_PKT_LEN  9600
+
+#define IPV6_ADDR_LEN 16
+
+#define MEMPOOL_CACHE_SIZE 256
+
+#define        BUF_SIZE        2048
+#define MBUF_SIZE      \
+       (BUF_SIZE + sizeof(struct rte_mbuf) + RTE_PKTMBUF_HEADROOM)
+
+#define        MAX_FLOW_NUM    UINT16_MAX
+#define        MIN_FLOW_NUM    1
+#define        DEF_FLOW_NUM    0x1000
+
+/* TTL numbers are in ms. */
+#define        MAX_FLOW_TTL    (3600 * MS_PER_S)
+#define        MIN_FLOW_TTL    1
+#define        DEF_FLOW_TTL    MS_PER_S
+
+#define        DEF_MBUF_NUM    0x400
+
+/* Should be power of two. */
+#define        IPV4_FRAG_TBL_BUCKET_ENTRIES    2
+
+static uint32_t max_flow_num = DEF_FLOW_NUM;
+static uint32_t max_flow_ttl = DEF_FLOW_TTL;
+
+/*
+ * RX and TX Prefetch, Host, and Write-back threshold values should be
+ * carefully set for optimal performance. Consult the network
+ * controller's datasheet and supporting DPDK documentation for guidance
+ * on how these parameters should be set.
+ */
+#define RX_PTHRESH 8 /**< Default values of RX prefetch threshold reg. */
+#define RX_HTHRESH 8 /**< Default values of RX host threshold reg. */
+#define RX_WTHRESH 4 /**< Default values of RX write-back threshold reg. */
+
+/*
+ * These default values are optimized for use with the Intel(R) 82599 10 GbE
+ * Controller and the DPDK ixgbe PMD. Consider using other values for other
+ * network controllers and/or network drivers.
+ */
+#define TX_PTHRESH 36 /**< Default values of TX prefetch threshold reg. */
+#define TX_HTHRESH 0  /**< Default values of TX host threshold reg. */
+#define TX_WTHRESH 0  /**< Default values of TX write-back threshold reg. */
+
+#define MAX_PKT_BURST 32
+#define BURST_TX_DRAIN_US 100 /* TX drain every ~100us */
+
+#define NB_SOCKETS 8
+
+/* Configure how many packets ahead to prefetch, when reading packets */
+#define PREFETCH_OFFSET        3
+
+/*
+ * Configurable number of RX/TX ring descriptors
+ */
+#define RTE_TEST_RX_DESC_DEFAULT 128
+#define RTE_TEST_TX_DESC_DEFAULT 512
+static uint16_t nb_rxd = RTE_TEST_RX_DESC_DEFAULT;
+static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT;
+
+/* ethernet addresses of ports */
+static struct ether_addr ports_eth_addr[MAX_PORTS];
+
+/* mask of enabled ports */
+static uint32_t enabled_port_mask = 0;
+static int promiscuous_on = 0; /**< Ports set in promiscuous mode off by default. */
+static int numa_on = 1; /**< NUMA is enabled by default. */
+
+struct mbuf_table {
+       uint16_t len;
+       struct rte_mbuf *m_table[MAX_PKT_BURST];
+};
+
+struct lcore_rx_queue {
+       uint8_t port_id;
+       uint8_t queue_id;
+} __rte_cache_aligned;
+
+#define MAX_RX_QUEUE_PER_LCORE 16
+#define MAX_TX_QUEUE_PER_PORT MAX_PORTS
+#define MAX_RX_QUEUE_PER_PORT 128
+
+#define MAX_LCORE_PARAMS 1024
+struct lcore_params {
+       uint8_t port_id;
+       uint8_t queue_id;
+       uint8_t lcore_id;
+} __rte_cache_aligned;
+
+static struct lcore_params lcore_params_array[MAX_LCORE_PARAMS];
+static struct lcore_params lcore_params_array_default[] = {
+       {0, 0, 2},
+       {0, 1, 2},
+       {0, 2, 2},
+       {1, 0, 2},
+       {1, 1, 2},
+       {1, 2, 2},
+       {2, 0, 2},
+       {3, 0, 3},
+       {3, 1, 3},
+};
+
+static struct lcore_params * lcore_params = lcore_params_array_default;
+static uint16_t nb_lcore_params = sizeof(lcore_params_array_default) /
+                               sizeof(lcore_params_array_default[0]);
+
+static struct rte_eth_conf port_conf = {
+       .rxmode = {
+               .max_rx_pkt_len = ETHER_MAX_LEN,
+               .split_hdr_size = 0,
+               .header_split   = 0, /**< Header Split disabled */
+               .hw_ip_checksum = 1, /**< IP checksum offload enabled */
+               .hw_vlan_filter = 0, /**< VLAN filtering disabled */
+               .jumbo_frame    = 0, /**< Jumbo Frame Support disabled */
+               .hw_strip_crc   = 0, /**< CRC stripped by hardware */
+       },
+       .rx_adv_conf = {
+               .rss_conf = {
+                       .rss_key = NULL,
+                       .rss_hf = ETH_RSS_IPV4 | ETH_RSS_IPV6,
+               },
+       },
+       .txmode = {
+               .mq_mode = ETH_MQ_TX_NONE,
+       },
+};
+
+static const struct rte_eth_rxconf rx_conf = {
+       .rx_thresh = {
+               .pthresh = RX_PTHRESH,
+               .hthresh = RX_HTHRESH,
+               .wthresh = RX_WTHRESH,
+       },
+       .rx_free_thresh = 32,
+};
+
+static const struct rte_eth_txconf tx_conf = {
+       .tx_thresh = {
+               .pthresh = TX_PTHRESH,
+               .hthresh = TX_HTHRESH,
+               .wthresh = TX_WTHRESH,
+       },
+       .tx_free_thresh = 0, /* Use PMD default values */
+       .tx_rs_thresh = 0, /* Use PMD default values */
+       .txq_flags = 0x0,
+};
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_EXACT_MATCH)
+
+#ifdef RTE_MACHINE_CPUFLAG_SSE4_2
+#include <rte_hash_crc.h>
+#define DEFAULT_HASH_FUNC       rte_hash_crc
+#else
+#include <rte_jhash.h>
+#define DEFAULT_HASH_FUNC       rte_jhash
+#endif
+
+struct ipv4_5tuple {
+       uint32_t ip_dst;
+       uint32_t ip_src;
+       uint16_t port_dst;
+       uint16_t port_src;
+       uint8_t  proto;
+} __attribute__((__packed__));
+
+struct ipv6_5tuple {
+       uint8_t  ip_dst[IPV6_ADDR_LEN];
+       uint8_t  ip_src[IPV6_ADDR_LEN];
+       uint16_t port_dst;
+       uint16_t port_src;
+       uint8_t  proto;
+} __attribute__((__packed__));
+
+struct ipv4_l3fwd_route {
+       struct ipv4_5tuple key;
+       uint8_t if_out;
+};
+
+struct ipv6_l3fwd_route {
+       struct ipv6_5tuple key;
+       uint8_t if_out;
+};
+
+static struct ipv4_l3fwd_route ipv4_l3fwd_route_array[] = {
+       {{IPv4(100,10,0,1), IPv4(200,10,0,1), 101, 11, IPPROTO_TCP}, 0},
+       {{IPv4(100,20,0,2), IPv4(200,20,0,2), 102, 12, IPPROTO_TCP}, 1},
+       {{IPv4(100,30,0,3), IPv4(200,30,0,3), 103, 13, IPPROTO_TCP}, 2},
+       {{IPv4(100,40,0,4), IPv4(200,40,0,4), 104, 14, IPPROTO_TCP}, 3},
+};
+
+static struct ipv6_l3fwd_route ipv6_l3fwd_route_array[] = {
+       {
+               {
+                       {0xfe, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+                        0x02, 0x1b, 0x21, 0xff, 0xfe, 0x91, 0x38, 0x05},
+                       {0xfe, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+                        0x02, 0x1e, 0x67, 0xff, 0xfe, 0x0d, 0xb6, 0x0a},
+                        1, 10, IPPROTO_UDP
+               }, 4
+       },
+};
+
+typedef struct rte_hash lookup_struct_t;
+static lookup_struct_t *ipv4_l3fwd_lookup_struct[NB_SOCKETS];
+static lookup_struct_t *ipv6_l3fwd_lookup_struct[NB_SOCKETS];
+
+#define L3FWD_HASH_ENTRIES     1024
+
+#define IPV4_L3FWD_NUM_ROUTES \
+       (sizeof(ipv4_l3fwd_route_array) / sizeof(ipv4_l3fwd_route_array[0]))
+
+#define IPV6_L3FWD_NUM_ROUTES \
+       (sizeof(ipv6_l3fwd_route_array) / sizeof(ipv6_l3fwd_route_array[0]))
+
+static uint8_t ipv4_l3fwd_out_if[L3FWD_HASH_ENTRIES] __rte_cache_aligned;
+static uint8_t ipv6_l3fwd_out_if[L3FWD_HASH_ENTRIES] __rte_cache_aligned;
+#endif
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_LPM)
+struct ipv4_l3fwd_route {
+       uint32_t ip;
+       uint8_t  depth;
+       uint8_t  if_out;
+};
+
+struct ipv6_l3fwd_route {
+       uint8_t ip[16];
+       uint8_t  depth;
+       uint8_t  if_out;
+};
+
+static struct ipv4_l3fwd_route ipv4_l3fwd_route_array[] = {
+       {IPv4(1,1,1,0), 24, 0},
+       {IPv4(2,1,1,0), 24, 1},
+       {IPv4(3,1,1,0), 24, 2},
+       {IPv4(4,1,1,0), 24, 3},
+       {IPv4(5,1,1,0), 24, 4},
+       {IPv4(6,1,1,0), 24, 5},
+       {IPv4(7,1,1,0), 24, 6},
+       {IPv4(8,1,1,0), 24, 7},
+};
+
+static struct ipv6_l3fwd_route ipv6_l3fwd_route_array[] = {
+       {{1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 0},
+       {{2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 1},
+       {{3,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 2},
+       {{4,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 3},
+       {{5,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 4},
+       {{6,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 5},
+       {{7,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 6},
+       {{8,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, 48, 7},
+};
+
+#define IPV4_L3FWD_NUM_ROUTES \
+       (sizeof(ipv4_l3fwd_route_array) / sizeof(ipv4_l3fwd_route_array[0]))
+#define IPV6_L3FWD_NUM_ROUTES \
+       (sizeof(ipv6_l3fwd_route_array) / sizeof(ipv6_l3fwd_route_array[0]))
+
+#define IPV4_L3FWD_LPM_MAX_RULES         1024
+#define IPV6_L3FWD_LPM_MAX_RULES         1024
+#define IPV6_L3FWD_LPM_NUMBER_TBL8S (1 << 16)
+
+typedef struct rte_lpm lookup_struct_t;
+typedef struct rte_lpm6 lookup6_struct_t;
+static lookup_struct_t *ipv4_l3fwd_lookup_struct[NB_SOCKETS];
+static lookup6_struct_t *ipv6_l3fwd_lookup_struct[NB_SOCKETS];
+#endif
+
+struct lcore_conf {
+       uint16_t n_rx_queue;
+       struct lcore_rx_queue rx_queue_list[MAX_RX_QUEUE_PER_LCORE];
+       uint16_t tx_queue_id[MAX_PORTS];
+       struct mbuf_table tx_mbufs[MAX_PORTS];
+       lookup_struct_t * ipv4_lookup_struct;
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_LPM)
+       lookup6_struct_t * ipv6_lookup_struct;
+#else
+       lookup_struct_t * ipv6_lookup_struct;
+#endif
+       struct ipv4_frag_tbl *frag_tbl[MAX_RX_QUEUE_PER_LCORE];
+       struct rte_mempool *pool[MAX_RX_QUEUE_PER_LCORE];
+} __rte_cache_aligned;
+
+static struct lcore_conf lcore_conf[RTE_MAX_LCORE];
+
+/* Send burst of packets on an output interface */
+static inline int
+send_burst(struct lcore_conf *qconf, uint16_t n, uint8_t port)
+{
+       struct rte_mbuf **m_table;
+       int ret;
+       uint16_t queueid;
+
+       queueid = qconf->tx_queue_id[port];
+       m_table = (struct rte_mbuf **)qconf->tx_mbufs[port].m_table;
+
+       ret = rte_eth_tx_burst(port, queueid, m_table, n);
+       if (unlikely(ret < n)) {
+               do {
+                       rte_pktmbuf_free(m_table[ret]);
+               } while (++ret < n);
+       }
+
+       return 0;
+}
+
+/* Enqueue a single packet, and send burst if queue is filled */
+static inline int
+send_single_packet(struct rte_mbuf *m, uint8_t port)
+{
+       uint32_t lcore_id;
+       uint16_t len;
+       struct lcore_conf *qconf;
+
+       lcore_id = rte_lcore_id();
+
+       qconf = &lcore_conf[lcore_id];
+       len = qconf->tx_mbufs[port].len;
+       qconf->tx_mbufs[port].m_table[len] = m;
+       len++;
+
+       /* enough pkts to be sent */
+       if (unlikely(len == MAX_PKT_BURST)) {
+               send_burst(qconf, MAX_PKT_BURST, port);
+               len = 0;
+       }
+
+       qconf->tx_mbufs[port].len = len;
+       return 0;
+}
+
+#ifdef DO_RFC_1812_CHECKS
+static inline int
+is_valid_ipv4_pkt(struct ipv4_hdr *pkt, uint32_t link_len)
+{
+       /* From http://www.rfc-editor.org/rfc/rfc1812.txt section 5.2.2 */
+       /*
+        * 1. The packet length reported by the Link Layer must be large
+        * enough to hold the minimum length legal IP datagram (20 bytes).
+        */
+       if (link_len < sizeof(struct ipv4_hdr))
+               return -1;
+
+       /* 2. The IP checksum must be correct. */
+       /* this is checked in H/W */
+
+       /*
+        * 3. The IP version number must be 4. If the version number is not 4
+        * then the packet may be another version of IP, such as IPng or
+        * ST-II.
+        */
+       if (((pkt->version_ihl) >> 4) != 4)
+               return -3;
+       /*
+        * 4. The IP header length field must be large enough to hold the
+        * minimum length legal IP datagram (20 bytes = 5 words).
+        */
+       if ((pkt->version_ihl & 0xf) < 5)
+               return -4;
+
+       /*
+        * 5. The IP total length field must be large enough to hold the IP
+        * datagram header, whose length is specified in the IP header length
+        * field.
+        */
+       if (rte_cpu_to_be_16(pkt->total_length) < sizeof(struct ipv4_hdr))
+               return -5;
+
+       return 0;
+}
+#endif
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_EXACT_MATCH)
+static void
+print_ipv4_key(struct ipv4_5tuple key)
+{
+       printf("IP dst = %08x, IP src = %08x, port dst = %d, port src = %d, proto = %d\n",
+                       (unsigned)key.ip_dst, (unsigned)key.ip_src, key.port_dst, key.port_src, key.proto);
+}
+static void
+print_ipv6_key(struct ipv6_5tuple key)
+{
+       printf( "IP dst = " IPv6_BYTES_FMT ", IP src = " IPv6_BYTES_FMT ", "
+               "port dst = %d, port src = %d, proto = %d\n",
+               IPv6_BYTES(key.ip_dst), IPv6_BYTES(key.ip_src),
+               key.port_dst, key.port_src, key.proto);
+}
+
+static inline uint8_t
+get_ipv4_dst_port(struct ipv4_hdr *ipv4_hdr,  uint8_t portid, lookup_struct_t * ipv4_l3fwd_lookup_struct)
+{
+       struct ipv4_5tuple key;
+       struct tcp_hdr *tcp;
+       struct udp_hdr *udp;
+       int ret = 0;
+
+       key.ip_dst = rte_be_to_cpu_32(ipv4_hdr->dst_addr);
+       key.ip_src = rte_be_to_cpu_32(ipv4_hdr->src_addr);
+       key.proto = ipv4_hdr->next_proto_id;
+
+       switch (ipv4_hdr->next_proto_id) {
+       case IPPROTO_TCP:
+               tcp = (struct tcp_hdr *)((unsigned char *) ipv4_hdr +
+                                       sizeof(struct ipv4_hdr));
+               key.port_dst = rte_be_to_cpu_16(tcp->dst_port);
+               key.port_src = rte_be_to_cpu_16(tcp->src_port);
+               break;
+
+       case IPPROTO_UDP:
+               udp = (struct udp_hdr *)((unsigned char *) ipv4_hdr +
+                                       sizeof(struct ipv4_hdr));
+               key.port_dst = rte_be_to_cpu_16(udp->dst_port);
+               key.port_src = rte_be_to_cpu_16(udp->src_port);
+               break;
+
+       default:
+               key.port_dst = 0;
+               key.port_src = 0;
+               break;
+       }
+
+       /* Find destination port */
+       ret = rte_hash_lookup(ipv4_l3fwd_lookup_struct, (const void *)&key);
+       return (uint8_t)((ret < 0)? portid : ipv4_l3fwd_out_if[ret]);
+}
+
+static inline uint8_t
+get_ipv6_dst_port(struct ipv6_hdr *ipv6_hdr,  uint8_t portid, lookup_struct_t * ipv6_l3fwd_lookup_struct)
+{
+       struct ipv6_5tuple key;
+       struct tcp_hdr *tcp;
+       struct udp_hdr *udp;
+       int ret = 0;
+
+       memcpy(key.ip_dst, ipv6_hdr->dst_addr, IPV6_ADDR_LEN);
+       memcpy(key.ip_src, ipv6_hdr->src_addr, IPV6_ADDR_LEN);
+
+       key.proto = ipv6_hdr->proto;
+
+       switch (ipv6_hdr->proto) {
+       case IPPROTO_TCP:
+               tcp = (struct tcp_hdr *)((unsigned char *) ipv6_hdr +
+                                       sizeof(struct ipv6_hdr));
+               key.port_dst = rte_be_to_cpu_16(tcp->dst_port);
+               key.port_src = rte_be_to_cpu_16(tcp->src_port);
+               break;
+
+       case IPPROTO_UDP:
+               udp = (struct udp_hdr *)((unsigned char *) ipv6_hdr +
+                                       sizeof(struct ipv6_hdr));
+               key.port_dst = rte_be_to_cpu_16(udp->dst_port);
+               key.port_src = rte_be_to_cpu_16(udp->src_port);
+               break;
+
+       default:
+               key.port_dst = 0;
+               key.port_src = 0;
+               break;
+       }
+
+       /* Find destination port */
+       ret = rte_hash_lookup(ipv6_l3fwd_lookup_struct, (const void *)&key);
+       return (uint8_t)((ret < 0)? portid : ipv6_l3fwd_out_if[ret]);
+}
+#endif
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_LPM)
+static inline uint8_t
+get_ipv4_dst_port(struct ipv4_hdr *ipv4_hdr,  uint8_t portid, lookup_struct_t * ipv4_l3fwd_lookup_struct)
+{
+       uint8_t next_hop;
+
+       return (uint8_t) ((rte_lpm_lookup(ipv4_l3fwd_lookup_struct,
+                       rte_be_to_cpu_32(ipv4_hdr->dst_addr), &next_hop) == 0)?
+                       next_hop : portid);
+}
+
+static inline uint8_t
+get_ipv6_dst_port(struct ipv6_hdr *ipv6_hdr,  uint8_t portid, lookup6_struct_t * ipv6_l3fwd_lookup_struct)
+{
+       uint8_t next_hop;
+
+       return (uint8_t) ((rte_lpm6_lookup(ipv6_l3fwd_lookup_struct,
+                       ipv6_hdr->dst_addr, &next_hop) == 0)?
+                       next_hop : portid);
+}
+#endif
+
+static inline void
+l3fwd_simple_forward(struct rte_mbuf *m, uint8_t portid, uint32_t queue,
+       struct lcore_conf *qconf, uint64_t tms)
+{
+       struct ether_hdr *eth_hdr;
+       struct ipv4_hdr *ipv4_hdr;
+       void *d_addr_bytes;
+       uint8_t dst_port;
+       uint16_t flag_offset, ip_flag, ip_ofs;
+
+       eth_hdr = rte_pktmbuf_mtod(m, struct ether_hdr *);
+
+       if (m->ol_flags & PKT_RX_IPV4_HDR) {
+               /* Handle IPv4 headers.*/
+               ipv4_hdr = (struct ipv4_hdr *)(eth_hdr + 1);
+
+#ifdef DO_RFC_1812_CHECKS
+               /* Check to make sure the packet is valid (RFC1812) */
+               if (is_valid_ipv4_pkt(ipv4_hdr, m->pkt.pkt_len) < 0) {
+                       rte_pktmbuf_free(m);
+                       return;
+               }
+
+               /* Update time to live and header checksum */
+               --(ipv4_hdr->time_to_live);
+               ++(ipv4_hdr->hdr_checksum);
+#endif
+
+               flag_offset = rte_be_to_cpu_16(ipv4_hdr->fragment_offset);
+               ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
+               ip_flag = (uint16_t)(flag_offset & IPV4_HDR_MF_FLAG);
+
+                /* if it is a fragmented packet, then try to reassemble. */
+               if (ip_flag != 0 || ip_ofs  != 0) {
+
+                       struct rte_mbuf *mo;
+                       struct ipv4_frag_tbl *tbl;
+
+                       tbl = qconf->frag_tbl[queue];
+
+                       /* prepare mbuf: setup l2_len/l3_len. */
+                       m->pkt.vlan_macip.f.l2_len = sizeof(*eth_hdr);
+                       m->pkt.vlan_macip.f.l3_len = sizeof(*ipv4_hdr);
+
+                       /* process this fragment. */
+                       if ((mo = ipv4_frag_mbuf(tbl, m, tms, ipv4_hdr,
+                                       ip_ofs, ip_flag)) == NULL) 
+                               /* no packet to send out. */
+                               return;
+
+                       /* we have our packet reassembled. */
+                       if (mo != m) {
+                               m = mo;
+                               eth_hdr = rte_pktmbuf_mtod(m,
+                                       struct ether_hdr *);
+                               ipv4_hdr = (struct ipv4_hdr *)(eth_hdr + 1);
+                       }
+               }
+
+               dst_port = get_ipv4_dst_port(ipv4_hdr, portid, qconf->ipv4_lookup_struct);
+               if (dst_port >= MAX_PORTS || (enabled_port_mask & 1 << dst_port) == 0)
+                       dst_port = portid;
+
+               /* 02:00:00:00:00:xx */
+               d_addr_bytes = &eth_hdr->d_addr.addr_bytes[0];
+               *((uint64_t *)d_addr_bytes) = 0x000000000002 + ((uint64_t)dst_port << 40);
+
+               /* src addr */
+               ether_addr_copy(&ports_eth_addr[dst_port], &eth_hdr->s_addr);
+
+               send_single_packet(m, dst_port);
+       }
+       else {
+               /* Handle IPv6 headers.*/
+               struct ipv6_hdr *ipv6_hdr;
+
+               ipv6_hdr = (struct ipv6_hdr *)(rte_pktmbuf_mtod(m, unsigned char *) +
+                               sizeof(struct ether_hdr));
+
+               dst_port = get_ipv6_dst_port(ipv6_hdr, portid, qconf->ipv6_lookup_struct);
+
+               if (dst_port >= MAX_PORTS || (enabled_port_mask & 1 << dst_port) == 0)
+                       dst_port = portid;
+
+               /* 02:00:00:00:00:xx */
+               d_addr_bytes = &eth_hdr->d_addr.addr_bytes[0];
+               *((uint64_t *)d_addr_bytes) = 0x000000000002 + ((uint64_t)dst_port << 40);
+
+               /* src addr */
+               ether_addr_copy(&ports_eth_addr[dst_port], &eth_hdr->s_addr);
+
+               send_single_packet(m, dst_port);
+       }
+
+}
+
+/* main processing loop */
+static int
+main_loop(__attribute__((unused)) void *dummy)
+{
+       struct rte_mbuf *pkts_burst[MAX_PKT_BURST];
+       unsigned lcore_id;
+       uint64_t diff_tsc, cur_tsc, prev_tsc;
+       int i, j, nb_rx;
+       uint8_t portid, queueid;
+       struct lcore_conf *qconf;
+       const uint64_t drain_tsc = (rte_get_tsc_hz() + US_PER_S - 1) / US_PER_S * BURST_TX_DRAIN_US;
+
+       prev_tsc = 0;
+
+       lcore_id = rte_lcore_id();
+       qconf = &lcore_conf[lcore_id];
+
+       if (qconf->n_rx_queue == 0) {
+               RTE_LOG(INFO, L3FWD, "lcore %u has nothing to do\n", lcore_id);
+               return 0;
+       }
+
+       RTE_LOG(INFO, L3FWD, "entering main loop on lcore %u\n", lcore_id);
+
+       for (i = 0; i < qconf->n_rx_queue; i++) {
+
+               portid = qconf->rx_queue_list[i].port_id;
+               queueid = qconf->rx_queue_list[i].queue_id;
+               RTE_LOG(INFO, L3FWD, " -- lcoreid=%u portid=%hhu rxqueueid=%hhu\n", lcore_id,
+                       portid, queueid);
+       }
+
+       while (1) {
+
+               cur_tsc = rte_rdtsc();
+
+               /*
+                * TX burst queue drain
+                */
+               diff_tsc = cur_tsc - prev_tsc;
+               if (unlikely(diff_tsc > drain_tsc)) {
+
+                       /*
+                        * This could be optimized (use queueid instead of
+                        * portid), but it is not called so often
+                        */
+                       for (portid = 0; portid < MAX_PORTS; portid++) {
+                               if (qconf->tx_mbufs[portid].len == 0)
+                                       continue;
+                               send_burst(&lcore_conf[lcore_id],
+                                       qconf->tx_mbufs[portid].len,
+                                       portid);
+                               qconf->tx_mbufs[portid].len = 0;
+                       }
+
+                       prev_tsc = cur_tsc;
+               }
+
+               /*
+                * Read packet from RX queues
+                */
+               for (i = 0; i < qconf->n_rx_queue; ++i) {
+
+                       portid = qconf->rx_queue_list[i].port_id;
+                       queueid = qconf->rx_queue_list[i].queue_id;
+
+                       nb_rx = rte_eth_rx_burst(portid, queueid, pkts_burst,
+                               MAX_PKT_BURST);
+
+                       /* Prefetch first packets */
+                       for (j = 0; j < PREFETCH_OFFSET && j < nb_rx; j++) {
+                               rte_prefetch0(rte_pktmbuf_mtod(
+                                               pkts_burst[j], void *));
+                       }
+
+                       /* Prefetch and forward already prefetched packets */
+                       for (j = 0; j < (nb_rx - PREFETCH_OFFSET); j++) {
+                               rte_prefetch0(rte_pktmbuf_mtod(pkts_burst[
+                                       j + PREFETCH_OFFSET], void *));
+                               l3fwd_simple_forward(pkts_burst[j], portid,
+                                       i, qconf, cur_tsc);
+                       }
+
+                       /* Forward remaining prefetched packets */
+                       for (; j < nb_rx; j++) {
+                               l3fwd_simple_forward(pkts_burst[j], portid,
+                                       i, qconf, cur_tsc);
+                       }
+               }
+       }
+}
+
+static int
+check_lcore_params(void)
+{
+       uint8_t queue, lcore;
+       uint16_t i;
+       int socketid;
+
+       for (i = 0; i < nb_lcore_params; ++i) {
+               queue = lcore_params[i].queue_id;
+               if (queue >= MAX_RX_QUEUE_PER_PORT) {
+                       printf("invalid queue number: %hhu\n", queue);
+                       return -1;
+               }
+               lcore = lcore_params[i].lcore_id;
+               if (!rte_lcore_is_enabled(lcore)) {
+                       printf("error: lcore %hhu is not enabled in lcore mask\n", lcore);
+                       return -1;
+               }
+               if ((socketid = rte_lcore_to_socket_id(lcore) != 0) &&
+                       (numa_on == 0)) {
+                       printf("warning: lcore %hhu is on socket %d with numa off \n",
+                               lcore, socketid);
+               }
+       }
+       return 0;
+}
+
+static int
+check_port_config(const unsigned nb_ports)
+{
+       unsigned portid;
+       uint16_t i;
+
+       for (i = 0; i < nb_lcore_params; ++i) {
+               portid = lcore_params[i].port_id;
+               if ((enabled_port_mask & (1 << portid)) == 0) {
+                       printf("port %u is not enabled in port mask\n", portid);
+                       return -1;
+               }
+               if (portid >= nb_ports) {
+                       printf("port %u is not present on the board\n", portid);
+                       return -1;
+               }
+       }
+       return 0;
+}
+
+static uint8_t
+get_port_n_rx_queues(const uint8_t port)
+{
+       int queue = -1;
+       uint16_t i;
+
+       for (i = 0; i < nb_lcore_params; ++i) {
+               if (lcore_params[i].port_id == port && lcore_params[i].queue_id > queue)
+                       queue = lcore_params[i].queue_id;
+       }
+       return (uint8_t)(++queue);
+}
+
+static int
+init_lcore_rx_queues(void)
+{
+       uint16_t i, nb_rx_queue;
+       uint8_t lcore;
+
+       for (i = 0; i < nb_lcore_params; ++i) {
+               lcore = lcore_params[i].lcore_id;
+               nb_rx_queue = lcore_conf[lcore].n_rx_queue;
+               if (nb_rx_queue >= MAX_RX_QUEUE_PER_LCORE) {
+                       printf("error: too many queues (%u) for lcore: %u\n",
+                               (unsigned)nb_rx_queue + 1, (unsigned)lcore);
+                       return -1;
+               } else {
+                       lcore_conf[lcore].rx_queue_list[nb_rx_queue].port_id =
+                               lcore_params[i].port_id;
+                       lcore_conf[lcore].rx_queue_list[nb_rx_queue].queue_id =
+                               lcore_params[i].queue_id;
+                       lcore_conf[lcore].n_rx_queue++;
+               }
+       }
+       return 0;
+}
+
+/* display usage */
+static void
+print_usage(const char *prgname)
+{
+       printf ("%s [EAL options] -- -p PORTMASK -P"
+               "  [--config (port,queue,lcore)[,(port,queue,lcore]]"
+               "  [--enable-jumbo [--max-pkt-len PKTLEN]]"
+               "  [--maxflows=<flows>]  [--flowttl=<ttl>[(s|ms)]]\n"
+               "  -p PORTMASK: hexadecimal bitmask of ports to configure\n"
+               "  -P : enable promiscuous mode\n"
+               "  --config (port,queue,lcore): rx queues configuration\n"
+               "  --no-numa: optional, disable numa awareness\n"
+               "  --enable-jumbo: enable jumbo frame"
+               " which max packet len is PKTLEN in decimal (64-9600)\n"
+               "  --maxflows=<flows>: optional, maximum number of flows "
+               "supported\n"
+               "  --flowttl=<ttl>[(s|ms)]: optional, maximum TTL for each "
+               "flow\n",
+               prgname);
+}
+
+static uint32_t
+parse_flow_num(const char *str, uint32_t min, uint32_t max, uint32_t *val)
+{
+       char *end;
+       uint64_t v;
+
+       /* parse decimal string */
+       errno = 0;
+       v = strtoul(str, &end, 10);
+       if (errno != 0 || *end != '\0')
+               return (-EINVAL);
+
+       if (v < min || v > max)
+               return (-EINVAL);
+
+       *val = (uint32_t)v;
+       return (0);
+}
+
+static int
+parse_flow_ttl(const char *str, uint32_t min, uint32_t max, uint32_t *val)
+{
+       char *end;
+       uint64_t v;
+
+       static const char frmt_sec[] = "s"; 
+       static const char frmt_msec[] = "ms"; 
+
+       /* parse decimal string */
+       errno = 0;
+       v = strtoul(str, &end, 10);
+       if (errno != 0)
+               return (-EINVAL);
+
+       if (*end != '\0') {
+               if (strncmp(frmt_sec, end, sizeof(frmt_sec)) == 0)
+                       v *= MS_PER_S;
+               else if (strncmp(frmt_msec, end, sizeof (frmt_msec)) != 0)
+                       return (-EINVAL);
+       }
+
+       if (v < min || v > max)
+               return (-EINVAL);
+
+       *val = (uint32_t)v;
+       return (0);
+}
+
+
+static int parse_max_pkt_len(const char *pktlen)
+{
+       char *end = NULL;
+       unsigned long len;
+
+       /* parse decimal string */
+       len = strtoul(pktlen, &end, 10);
+       if ((pktlen[0] == '\0') || (end == NULL) || (*end != '\0'))
+               return -1;
+
+       if (len == 0)
+               return -1;
+
+       return len;
+}
+
+static int
+parse_portmask(const char *portmask)
+{
+       char *end = NULL;
+       unsigned long pm;
+
+       /* parse hexadecimal string */
+       pm = strtoul(portmask, &end, 16);
+       if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0'))
+               return -1;
+
+       if (pm == 0)
+               return -1;
+
+       return pm;
+}
+
+static int
+parse_config(const char *q_arg)
+{
+       char s[256];
+       const char *p, *p0 = q_arg;
+       char *end;
+       enum fieldnames {
+               FLD_PORT = 0,
+               FLD_QUEUE,
+               FLD_LCORE,
+               _NUM_FLD
+       };
+       unsigned long int_fld[_NUM_FLD];
+       char *str_fld[_NUM_FLD];
+       int i;
+       unsigned size;
+
+       nb_lcore_params = 0;
+
+       while ((p = strchr(p0,'(')) != NULL) {
+               ++p;
+               if((p0 = strchr(p,')')) == NULL)
+                       return -1;
+
+               size = p0 - p;
+               if(size >= sizeof(s))
+                       return -1;
+
+               rte_snprintf(s, sizeof(s), "%.*s", size, p);
+               if (rte_strsplit(s, sizeof(s), str_fld, _NUM_FLD, ',') != _NUM_FLD)
+                       return -1;
+               for (i = 0; i < _NUM_FLD; i++){
+                       errno = 0;
+                       int_fld[i] = strtoul(str_fld[i], &end, 0);
+                       if (errno != 0 || end == str_fld[i] || int_fld[i] > 255)
+                               return -1;
+               }
+               if (nb_lcore_params >= MAX_LCORE_PARAMS) {
+                       printf("exceeded max number of lcore params: %hu\n",
+                               nb_lcore_params);
+                       return -1;
+               }
+               lcore_params_array[nb_lcore_params].port_id = (uint8_t)int_fld[FLD_PORT];
+               lcore_params_array[nb_lcore_params].queue_id = (uint8_t)int_fld[FLD_QUEUE];
+               lcore_params_array[nb_lcore_params].lcore_id = (uint8_t)int_fld[FLD_LCORE];
+               ++nb_lcore_params;
+       }
+       lcore_params = lcore_params_array;
+       return 0;
+}
+
+/* Parse the argument given in the command line of the application */
+static int
+parse_args(int argc, char **argv)
+{
+       int opt, ret;
+       char **argvopt;
+       int option_index;
+       char *prgname = argv[0];
+       static struct option lgopts[] = {
+               {"config", 1, 0, 0},
+               {"no-numa", 0, 0, 0},
+               {"enable-jumbo", 0, 0, 0},
+               {"maxflows", 1, 0, 0},
+               {"flowttl", 1, 0, 0},
+               {NULL, 0, 0, 0}
+       };
+
+       argvopt = argv;
+
+       while ((opt = getopt_long(argc, argvopt, "p:P",
+                               lgopts, &option_index)) != EOF) {
+
+               switch (opt) {
+               /* portmask */
+               case 'p':
+                       enabled_port_mask = parse_portmask(optarg);
+                       if (enabled_port_mask == 0) {
+                               printf("invalid portmask\n");
+                               print_usage(prgname);
+                               return -1;
+                       }
+                       break;
+               case 'P':
+                       printf("Promiscuous mode selected\n");
+                       promiscuous_on = 1;
+                       break;
+
+               /* long options */
+               case 0:
+                       if (!strncmp(lgopts[option_index].name, "config", 6)) {
+                               ret = parse_config(optarg);
+                               if (ret) {
+                                       printf("invalid config\n");
+                                       print_usage(prgname);
+                                       return -1;
+                               }
+                       }
+
+                       if (!strncmp(lgopts[option_index].name, "no-numa", 7)) {
+                               printf("numa is disabled \n");
+                               numa_on = 0;
+                       }
+                       
+                       if (!strncmp(lgopts[option_index].name,
+                                       "maxflows", 8)) {
+                               if ((ret = parse_flow_num(optarg, MIN_FLOW_NUM,
+                                               MAX_FLOW_NUM,
+                                               &max_flow_num)) != 0) {
+                                       printf("invalid value: \"%s\" for "
+                                               "parameter %s\n",
+                                               optarg,
+                                               lgopts[option_index].name);
+                                       print_usage(prgname);
+                                       return (ret);
+                               }
+                       }
+                       
+                       if (!strncmp(lgopts[option_index].name, "flowttl", 7)) {
+                               if ((ret = parse_flow_ttl(optarg, MIN_FLOW_TTL,
+                                               MAX_FLOW_TTL,
+                                               &max_flow_ttl)) != 0) {
+                                       printf("invalid value: \"%s\" for "
+                                               "parameter %s\n",
+                                               optarg,
+                                               lgopts[option_index].name);
+                                       print_usage(prgname);
+                                       return (ret);
+                               }
+                       }
+
+                       if (!strncmp(lgopts[option_index].name, "enable-jumbo", 12)) {
+                               struct option lenopts = {"max-pkt-len", required_argument, 0, 0};
+
+                               printf("jumbo frame is enabled \n");
+                               port_conf.rxmode.jumbo_frame = 1;
+       
+                               /* if no max-pkt-len set, use the default value ETHER_MAX_LEN */        
+                               if (0 == getopt_long(argc, argvopt, "", &lenopts, &option_index)) {
+                                       ret = parse_max_pkt_len(optarg);
+                                       if ((ret < 64) || (ret > MAX_JUMBO_PKT_LEN)){
+                                               printf("invalid packet length\n");
+                                               print_usage(prgname);
+                                               return -1;
+                                       }
+                                       port_conf.rxmode.max_rx_pkt_len = ret;
+                               }
+                               printf("set jumbo frame max packet length to %u\n", 
+                                               (unsigned int)port_conf.rxmode.max_rx_pkt_len);
+                       }
+                       
+                       break;
+
+               default:
+                       print_usage(prgname);
+                       return -1;
+               }
+       }
+
+       if (optind >= 0)
+               argv[optind-1] = prgname;
+
+       ret = optind-1;
+       optind = 0; /* reset getopt lib */
+       return ret;
+}
+
+static void
+print_ethaddr(const char *name, const struct ether_addr *eth_addr)
+{
+       printf ("%s%02X:%02X:%02X:%02X:%02X:%02X", name,
+               eth_addr->addr_bytes[0],
+               eth_addr->addr_bytes[1],
+               eth_addr->addr_bytes[2],
+               eth_addr->addr_bytes[3],
+               eth_addr->addr_bytes[4],
+               eth_addr->addr_bytes[5]);
+}
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_EXACT_MATCH)
+static void
+setup_hash(int socketid)
+{
+       struct rte_hash_parameters ipv4_l3fwd_hash_params = {
+               .name = NULL,
+               .entries = L3FWD_HASH_ENTRIES,
+               .bucket_entries = 4,
+               .key_len = sizeof(struct ipv4_5tuple),
+               .hash_func = DEFAULT_HASH_FUNC,
+               .hash_func_init_val = 0,
+       };
+
+       struct rte_hash_parameters ipv6_l3fwd_hash_params = {
+               .name = NULL,
+               .entries = L3FWD_HASH_ENTRIES,
+               .bucket_entries = 4,
+               .key_len = sizeof(struct ipv6_5tuple),
+               .hash_func = DEFAULT_HASH_FUNC,
+               .hash_func_init_val = 0,
+       };
+
+       unsigned i;
+       int ret;
+       char s[64];
+
+       /* create ipv4 hash */
+       rte_snprintf(s, sizeof(s), "ipv4_l3fwd_hash_%d", socketid);
+       ipv4_l3fwd_hash_params.name = s;
+       ipv4_l3fwd_hash_params.socket_id = socketid;
+       ipv4_l3fwd_lookup_struct[socketid] = rte_hash_create(&ipv4_l3fwd_hash_params);
+       if (ipv4_l3fwd_lookup_struct[socketid] == NULL)
+               rte_exit(EXIT_FAILURE, "Unable to create the l3fwd hash on "
+                               "socket %d\n", socketid);
+
+       /* create ipv6 hash */
+       rte_snprintf(s, sizeof(s), "ipv6_l3fwd_hash_%d", socketid);
+       ipv6_l3fwd_hash_params.name = s;
+       ipv6_l3fwd_hash_params.socket_id = socketid;
+       ipv6_l3fwd_lookup_struct[socketid] = rte_hash_create(&ipv6_l3fwd_hash_params);
+       if (ipv6_l3fwd_lookup_struct[socketid] == NULL)
+               rte_exit(EXIT_FAILURE, "Unable to create the l3fwd hash on "
+                               "socket %d\n", socketid);
+
+
+       /* populate the ipv4 hash */
+       for (i = 0; i < IPV4_L3FWD_NUM_ROUTES; i++) {
+               ret = rte_hash_add_key (ipv4_l3fwd_lookup_struct[socketid],
+                               (void *) &ipv4_l3fwd_route_array[i].key);
+               if (ret < 0) {
+                       rte_exit(EXIT_FAILURE, "Unable to add entry %u to the"
+                               "l3fwd hash on socket %d\n", i, socketid);
+               }
+               ipv4_l3fwd_out_if[ret] = ipv4_l3fwd_route_array[i].if_out;
+               printf("Hash: Adding key\n");
+               print_ipv4_key(ipv4_l3fwd_route_array[i].key);
+       }
+
+       /* populate the ipv6 hash */
+       for (i = 0; i < IPV6_L3FWD_NUM_ROUTES; i++) {
+               ret = rte_hash_add_key (ipv6_l3fwd_lookup_struct[socketid],
+                               (void *) &ipv6_l3fwd_route_array[i].key);
+               if (ret < 0) {
+                       rte_exit(EXIT_FAILURE, "Unable to add entry %u to the"
+                               "l3fwd hash on socket %d\n", i, socketid);
+               }
+               ipv6_l3fwd_out_if[ret] = ipv6_l3fwd_route_array[i].if_out;
+               printf("Hash: Adding key\n");
+               print_ipv6_key(ipv6_l3fwd_route_array[i].key);
+       }
+}
+#endif
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_LPM)
+static void
+setup_lpm(int socketid)
+{
+       struct rte_lpm6_config config;
+       unsigned i;
+       int ret;
+       char s[64];
+
+       /* create the LPM table */
+       rte_snprintf(s, sizeof(s), "IPV4_L3FWD_LPM_%d", socketid);
+       ipv4_l3fwd_lookup_struct[socketid] = rte_lpm_create(s, socketid,
+                               IPV4_L3FWD_LPM_MAX_RULES, 0);
+       if (ipv4_l3fwd_lookup_struct[socketid] == NULL)
+               rte_exit(EXIT_FAILURE, "Unable to create the l3fwd LPM table"
+                               " on socket %d\n", socketid);
+
+       /* populate the LPM table */
+       for (i = 0; i < IPV4_L3FWD_NUM_ROUTES; i++) {
+               ret = rte_lpm_add(ipv4_l3fwd_lookup_struct[socketid],
+                       ipv4_l3fwd_route_array[i].ip,
+                       ipv4_l3fwd_route_array[i].depth,
+                       ipv4_l3fwd_route_array[i].if_out);
+
+               if (ret < 0) {
+                       rte_exit(EXIT_FAILURE, "Unable to add entry %u to the "
+                               "l3fwd LPM table on socket %d\n",
+                               i, socketid);
+               }
+
+               printf("LPM: Adding route 0x%08x / %d (%d)\n",
+                       (unsigned)ipv4_l3fwd_route_array[i].ip,
+                       ipv4_l3fwd_route_array[i].depth,
+                       ipv4_l3fwd_route_array[i].if_out);
+       }
+       
+       /* create the LPM6 table */
+       rte_snprintf(s, sizeof(s), "IPV6_L3FWD_LPM_%d", socketid);
+       
+       config.max_rules = IPV6_L3FWD_LPM_MAX_RULES;
+       config.number_tbl8s = IPV6_L3FWD_LPM_NUMBER_TBL8S;
+       config.flags = 0;
+       ipv6_l3fwd_lookup_struct[socketid] = rte_lpm6_create(s, socketid,
+                               &config);
+       if (ipv6_l3fwd_lookup_struct[socketid] == NULL)
+               rte_exit(EXIT_FAILURE, "Unable to create the l3fwd LPM table"
+                               " on socket %d\n", socketid);
+
+       /* populate the LPM table */
+       for (i = 0; i < IPV6_L3FWD_NUM_ROUTES; i++) {
+               ret = rte_lpm6_add(ipv6_l3fwd_lookup_struct[socketid],
+                       ipv6_l3fwd_route_array[i].ip,
+                       ipv6_l3fwd_route_array[i].depth,
+                       ipv6_l3fwd_route_array[i].if_out);
+
+               if (ret < 0) {
+                       rte_exit(EXIT_FAILURE, "Unable to add entry %u to the "
+                               "l3fwd LPM table on socket %d\n",
+                               i, socketid);
+               }
+
+               printf("LPM: Adding route %s / %d (%d)\n",
+                       "IPV6",
+                       ipv6_l3fwd_route_array[i].depth,
+                       ipv6_l3fwd_route_array[i].if_out);
+       }
+}
+#endif
+
+static int
+init_mem(void)
+{
+       struct lcore_conf *qconf;
+       int socketid;
+       unsigned lcore_id;
+
+       for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) {
+               if (rte_lcore_is_enabled(lcore_id) == 0)
+                       continue;
+
+               if (numa_on)
+                       socketid = rte_lcore_to_socket_id(lcore_id);
+               else
+                       socketid = 0;
+
+               if (socketid >= NB_SOCKETS) {
+                       rte_exit(EXIT_FAILURE,
+                               "Socket %d of lcore %u is out of range %d\n",
+                               socketid, lcore_id, NB_SOCKETS);
+               }
+
+#if (APP_LOOKUP_METHOD == APP_LOOKUP_LPM)
+                       setup_lpm(socketid);
+#else
+                       setup_hash(socketid);
+#endif
+               qconf = &lcore_conf[lcore_id];
+               qconf->ipv4_lookup_struct = ipv4_l3fwd_lookup_struct[socketid];
+               qconf->ipv6_lookup_struct = ipv6_l3fwd_lookup_struct[socketid];
+       }
+       return 0;
+}
+
+/* Check the link status of all ports in up to 9s, and print them finally */
+static void
+check_all_ports_link_status(uint8_t port_num, uint32_t port_mask)
+{
+#define CHECK_INTERVAL 100 /* 100ms */
+#define MAX_CHECK_TIME 90 /* 9s (90 * 100ms) in total */
+       uint8_t portid, count, all_ports_up, print_flag = 0;
+       struct rte_eth_link link;
+
+       printf("\nChecking link status");
+       fflush(stdout);
+       for (count = 0; count <= MAX_CHECK_TIME; count++) {
+               all_ports_up = 1;
+               for (portid = 0; portid < port_num; portid++) {
+                       if ((port_mask & (1 << portid)) == 0)
+                               continue;
+                       memset(&link, 0, sizeof(link));
+                       rte_eth_link_get_nowait(portid, &link);
+                       /* print link status if flag set */
+                       if (print_flag == 1) {
+                               if (link.link_status)
+                                       printf("Port %d Link Up - speed %u "
+                                               "Mbps - %s\n", (uint8_t)portid,
+                                               (unsigned)link.link_speed,
+                               (link.link_duplex == ETH_LINK_FULL_DUPLEX) ?
+                                       ("full-duplex") : ("half-duplex\n"));
+                               else
+                                       printf("Port %d Link Down\n",
+                                               (uint8_t)portid);
+                               continue;
+                       }
+                       /* clear all_ports_up flag if any link down */
+                       if (link.link_status == 0) {
+                               all_ports_up = 0;
+                               break;
+                       }
+               }
+               /* after finally printing all link status, get out */
+               if (print_flag == 1)
+                       break;
+
+               if (all_ports_up == 0) {
+                       printf(".");
+                       fflush(stdout);
+                       rte_delay_ms(CHECK_INTERVAL);
+               }
+
+               /* set the print_flag if all ports up or timeout */
+               if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) {
+                       print_flag = 1;
+                       printf("done\n");
+               }
+       }
+}
+
+static void
+setup_queue_frag_tbl(struct lcore_conf *qconf, uint32_t lcore, int socket,
+       uint32_t queue)
+{
+       uint32_t nb_mbuf;
+       uint64_t frag_cycles;
+       char buf[RTE_MEMPOOL_NAMESIZE];
+
+       frag_cycles = (rte_get_tsc_hz() + MS_PER_S - 1) / MS_PER_S *
+               max_flow_ttl;
+
+       if ((qconf->frag_tbl[queue] = ipv4_frag_tbl_create(max_flow_num,
+                       IPV4_FRAG_TBL_BUCKET_ENTRIES, max_flow_num, frag_cycles,
+                       socket)) == NULL)
+               rte_exit(EXIT_FAILURE, "ipv4_frag_tbl_create(%u) on "
+                       "lcore: %u for queue: %u failed\n",
+                       max_flow_num, lcore, queue);
+
+       nb_mbuf = max_flow_num * MAX_FRAG_NUM;
+       nb_mbuf *= (port_conf.rxmode.max_rx_pkt_len + BUF_SIZE - 1) / BUF_SIZE;
+       nb_mbuf += RTE_TEST_RX_DESC_DEFAULT + MAX_PKT_BURST +
+               RTE_TEST_TX_DESC_DEFAULT;
+       nb_mbuf = RTE_MAX(nb_mbuf, (uint32_t)DEF_MBUF_NUM);
+               
+       rte_snprintf(buf, sizeof(buf), "mbuf_pool_%u_%u", lcore, queue);
+
+       if ((qconf->pool[queue] = rte_mempool_create(buf, nb_mbuf, MBUF_SIZE, 0,
+                       sizeof(struct rte_pktmbuf_pool_private),
+                       rte_pktmbuf_pool_init, NULL, rte_pktmbuf_init, NULL,
+                       socket, MEMPOOL_F_SP_PUT | MEMPOOL_F_SC_GET)) == NULL)
+               rte_exit(EXIT_FAILURE, "mempool_create(%s) failed", buf);
+}
+
+static void
+queue_frag_tbl_dump_stat(void)
+{
+       uint32_t i, lcore;
+       const struct lcore_conf *qconf;
+
+       for (lcore = 0; lcore < RTE_MAX_LCORE; lcore++) {
+               if (rte_lcore_is_enabled(lcore) == 0)
+                       continue;
+
+               qconf = lcore_conf + lcore;
+               for (i = 0; i < qconf->n_rx_queue; i++) {
+
+                       fprintf(stdout, " -- lcoreid=%u portid=%hhu "
+                               "rxqueueid=%hhu frag tbl stat:\n",
+                               lcore,  qconf->rx_queue_list[i].port_id,
+                               qconf->rx_queue_list[i].queue_id);
+                       ipv4_frag_tbl_dump_stat(stdout, qconf->frag_tbl[i]);
+               }
+       }
+}
+
+static void
+signal_handler(int signum)
+{
+       queue_frag_tbl_dump_stat();
+       if (signum != SIGUSR1)
+               rte_exit(0, "received signal: %d, exiting\n", signum);
+}
+
+int
+MAIN(int argc, char **argv)
+{
+       struct lcore_conf *qconf;
+       int ret;
+       unsigned nb_ports;
+       uint16_t queueid;
+       unsigned lcore_id;
+       uint32_t n_tx_queue, nb_lcores;
+       uint8_t portid, nb_rx_queue, queue, socketid;
+
+       /* init EAL */
+       ret = rte_eal_init(argc, argv);
+       if (ret < 0)
+               rte_exit(EXIT_FAILURE, "Invalid EAL parameters\n");
+       argc -= ret;
+       argv += ret;
+
+       /* parse application arguments (after the EAL ones) */
+       ret = parse_args(argc, argv);
+       if (ret < 0)
+               rte_exit(EXIT_FAILURE, "Invalid L3FWD parameters\n");
+
+       if (check_lcore_params() < 0)
+               rte_exit(EXIT_FAILURE, "check_lcore_params failed\n");
+
+       ret = init_lcore_rx_queues();
+       if (ret < 0)
+               rte_exit(EXIT_FAILURE, "init_lcore_rx_queues failed\n");
+
+
+       /* init driver(s) */
+       if (rte_pmd_init_all() < 0)
+               rte_exit(EXIT_FAILURE, "Cannot init pmd\n");
+
+       if (rte_eal_pci_probe() < 0)
+               rte_exit(EXIT_FAILURE, "Cannot probe PCI\n");
+
+       nb_ports = rte_eth_dev_count();
+       if (nb_ports > MAX_PORTS)
+               nb_ports = MAX_PORTS;
+
+       if (check_port_config(nb_ports) < 0)
+               rte_exit(EXIT_FAILURE, "check_port_config failed\n");
+
+       nb_lcores = rte_lcore_count();
+
+       /* initialize all ports */
+       for (portid = 0; portid < nb_ports; portid++) {
+               /* skip ports that are not enabled */
+               if ((enabled_port_mask & (1 << portid)) == 0) {
+                       printf("\nSkipping disabled port %d\n", portid);
+                       continue;
+               }
+
+               /* init port */
+               printf("Initializing port %d ... ", portid );
+               fflush(stdout);
+
+               nb_rx_queue = get_port_n_rx_queues(portid);
+               n_tx_queue = nb_lcores;
+               if (n_tx_queue > MAX_TX_QUEUE_PER_PORT)
+                       n_tx_queue = MAX_TX_QUEUE_PER_PORT;
+               printf("Creating queues: nb_rxq=%d nb_txq=%u... ",
+                       nb_rx_queue, (unsigned)n_tx_queue );
+               ret = rte_eth_dev_configure(portid, nb_rx_queue,
+                                       (uint16_t)n_tx_queue, &port_conf);
+               if (ret < 0)
+                       rte_exit(EXIT_FAILURE, "Cannot configure device: err=%d, port=%d\n",
+                               ret, portid);
+
+               rte_eth_macaddr_get(portid, &ports_eth_addr[portid]);
+               print_ethaddr(" Address:", &ports_eth_addr[portid]);
+               printf(", ");
+
+               /* init memory */
+               ret = init_mem();
+               if (ret < 0)
+                       rte_exit(EXIT_FAILURE, "init_mem failed\n");
+
+               /* init one TX queue per couple (lcore,port) */
+               queueid = 0;
+               for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) {
+                       if (rte_lcore_is_enabled(lcore_id) == 0)
+                               continue;
+
+                       if (numa_on)
+                               socketid = (uint8_t)rte_lcore_to_socket_id(lcore_id);
+                       else
+                               socketid = 0;
+
+                       printf("txq=%u,%d,%d ", lcore_id, queueid, socketid);
+                       fflush(stdout);
+                       ret = rte_eth_tx_queue_setup(portid, queueid, nb_txd,
+                                                    socketid, &tx_conf);
+                       if (ret < 0)
+                               rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup: err=%d, "
+                                       "port=%d\n", ret, portid);
+
+                       qconf = &lcore_conf[lcore_id];
+                       qconf->tx_queue_id[portid] = queueid;
+                       queueid++;
+               }
+               printf("\n");
+       }
+
+       for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) {
+               if (rte_lcore_is_enabled(lcore_id) == 0)
+                       continue;
+               qconf = &lcore_conf[lcore_id];
+               printf("\nInitializing rx queues on lcore %u ... ", lcore_id );
+               fflush(stdout);
+               /* init RX queues */
+               for(queue = 0; queue < qconf->n_rx_queue; ++queue) {
+                       portid = qconf->rx_queue_list[queue].port_id;
+                       queueid = qconf->rx_queue_list[queue].queue_id;
+
+                       if (numa_on)
+                               socketid = (uint8_t)rte_lcore_to_socket_id(lcore_id);
+                       else
+                               socketid = 0;
+
+                       printf("rxq=%d,%d,%d ", portid, queueid, socketid);
+                       fflush(stdout);
+
+                       setup_queue_frag_tbl(qconf, lcore_id, socketid, queue);
+
+                       ret = rte_eth_rx_queue_setup(portid, queueid, nb_rxd,
+                                       socketid, &rx_conf, qconf->pool[queue]);
+                       if (ret < 0)
+                               rte_exit(EXIT_FAILURE,
+                                       "rte_eth_rx_queue_setup: err=%d,"
+                                       "port=%d\n", ret, portid);
+               }
+       }
+
+       printf("\n");
+
+       /* start ports */
+       for (portid = 0; portid < nb_ports; portid++) {
+               if ((enabled_port_mask & (1 << portid)) == 0) {
+                       continue;
+               }
+               /* Start device */
+               ret = rte_eth_dev_start(portid);
+               if (ret < 0)
+                       rte_exit(EXIT_FAILURE, "rte_eth_dev_start: err=%d, port=%d\n",
+                               ret, portid);
+
+               /*
+                * If enabled, put device in promiscuous mode.
+                * This allows IO forwarding mode to forward packets
+                * to itself through 2 cross-connected  ports of the
+                * target machine.
+                */
+               if (promiscuous_on)
+                       rte_eth_promiscuous_enable(portid);
+       }
+
+       check_all_ports_link_status((uint8_t)nb_ports, enabled_port_mask);
+
+       signal(SIGUSR1, signal_handler);
+       signal(SIGTERM, signal_handler);
+       signal(SIGINT, signal_handler);
+
+       /* launch per-lcore init on every lcore */
+       rte_eal_mp_remote_launch(main_loop, NULL, CALL_MASTER);
+       RTE_LCORE_FOREACH_SLAVE(lcore_id) {
+               if (rte_eal_wait_lcore(lcore_id) < 0)
+                       return -1;
+       }
+
+       return 0;
+}
diff --git a/examples/ip_reassembly/main.h b/examples/ip_reassembly/main.h
new file mode 100644 (file)
index 0000000..0534e95
--- /dev/null
@@ -0,0 +1,47 @@
+/*-
+ *   BSD LICENSE
+ * 
+ *   Copyright(c) 2010-2013 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.
+ * 
+ */
+
+#ifndef _MAIN_H_
+#define _MAIN_H_
+
+#ifdef RTE_EXEC_ENV_BAREMETAL
+#define MAIN _main
+#else
+#define MAIN main
+#endif
+
+
+int MAIN(int argc, char **argv);
+
+#endif /* _MAIN_H_ */
index dd159be..f42844e 100644 (file)
@@ -105,6 +105,18 @@ struct ipv4_hdr {
                                           (((c) & 0xff) << 8)  | \
                                           ((d) & 0xff))
 
+/* Fragment Offset * Flags. */
+#define        IPV4_HDR_DF_SHIFT       14
+#define        IPV4_HDR_MF_SHIFT       13
+#define        IPV4_HDR_FO_SHIFT       3
+
+#define        IPV4_HDR_DF_FLAG        (1 << IPV4_HDR_DF_SHIFT)
+#define        IPV4_HDR_MF_FLAG        (1 << IPV4_HDR_MF_SHIFT)
+
+#define        IPV4_HDR_OFFSET_MASK    ((1 << IPV4_HDR_MF_SHIFT) - 1)
+
+#define        IPV4_HDR_OFFSET_UNITS   8
+
 /* IPv4 protocols */
 #define IPPROTO_IP         0  /**< dummy for IP */
 #define IPPROTO_HOPOPTS    0  /**< IP6 hop-by-hop options */