4 * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * * Neither the name of Intel Corporation nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
34 #ifndef _RTE_IP_FRAG_H_
35 #define _RTE_IP_FRAG_H_
39 * RTE IP Fragmentation and Reassembly
41 * Implementation of IP packet fragmentation and reassembly.
51 #include <rte_malloc.h>
52 #include <rte_memory.h>
54 #include <rte_byteorder.h>
59 IP_LAST_FRAG_IDX, /**< index of last fragment */
60 IP_FIRST_FRAG_IDX, /**< index of first fragment */
61 IP_MIN_FRAG_NUM, /**< minimum number of fragments */
62 IP_MAX_FRAG_NUM = RTE_LIBRTE_IP_FRAG_MAX_FRAG,
63 /**< maximum number of fragments per packet */
66 /** @internal fragmented mbuf */
68 uint16_t ofs; /**< offset into the packet */
69 uint16_t len; /**< length of fragment */
70 struct rte_mbuf *mb; /**< fragment mbuf */
73 /** @internal <src addr, dst_addr, id> to uniquely identify fragmented datagram. */
75 uint64_t src_dst[4]; /**< src address, first 8 bytes used for IPv4 */
76 uint32_t id; /**< dst address */
77 uint32_t key_len; /**< src/dst key length */
81 * @internal Fragmented packet to reassemble.
82 * First two entries in the frags[] array are for the last and first fragments.
85 TAILQ_ENTRY(ip_frag_pkt) lru; /**< LRU list */
86 struct ip_frag_key key; /**< fragmentation key */
87 uint64_t start; /**< creation timestamp */
88 uint32_t total_size; /**< expected reassembled size */
89 uint32_t frag_size; /**< size of fragments received */
90 uint32_t last_idx; /**< index of next entry to fill */
91 struct ip_frag frags[IP_MAX_FRAG_NUM]; /**< fragments */
92 } __rte_cache_aligned;
94 #define IP_FRAG_DEATH_ROW_LEN 32 /**< death row size (in packets) */
96 /** mbuf death row (packets to be freed) */
97 struct rte_ip_frag_death_row {
98 uint32_t cnt; /**< number of mbufs currently on death row */
99 struct rte_mbuf *row[IP_FRAG_DEATH_ROW_LEN * (IP_MAX_FRAG_NUM + 1)];
100 /**< mbufs to be freed */
103 TAILQ_HEAD(ip_pkt_list, ip_frag_pkt); /**< @internal fragments tailq */
105 /** fragmentation table statistics */
106 struct ip_frag_tbl_stat {
107 uint64_t find_num; /**< total # of find/insert attempts. */
108 uint64_t add_num; /**< # of add ops. */
109 uint64_t del_num; /**< # of del ops. */
110 uint64_t reuse_num; /**< # of reuse (del/add) ops. */
111 uint64_t fail_total; /**< total # of add failures. */
112 uint64_t fail_nospace; /**< # of 'no space' add failures. */
113 } __rte_cache_aligned;
115 /** fragmentation table */
116 struct rte_ip_frag_tbl {
117 uint64_t max_cycles; /**< ttl for table entries. */
118 uint32_t entry_mask; /**< hash value mask. */
119 uint32_t max_entries; /**< max entries allowed. */
120 uint32_t use_entries; /**< entries in use. */
121 uint32_t bucket_entries; /**< hash associativity. */
122 uint32_t nb_entries; /**< total size of the table. */
123 uint32_t nb_buckets; /**< num of associativity lines. */
124 struct ip_frag_pkt *last; /**< last used entry. */
125 struct ip_pkt_list lru; /**< LRU list for table entries. */
126 struct ip_frag_tbl_stat stat; /**< statistics counters. */
127 __extension__ struct ip_frag_pkt pkt[0]; /**< hash table. */
130 /** IPv6 fragment extension header */
131 #define RTE_IPV6_EHDR_MF_SHIFT 0
132 #define RTE_IPV6_EHDR_MF_MASK 1
133 #define RTE_IPV6_EHDR_FO_SHIFT 3
134 #define RTE_IPV6_EHDR_FO_MASK (~((1 << RTE_IPV6_EHDR_FO_SHIFT) - 1))
136 #define RTE_IPV6_FRAG_USED_MASK \
137 (RTE_IPV6_EHDR_MF_MASK | RTE_IPV6_EHDR_FO_MASK)
139 #define RTE_IPV6_GET_MF(x) ((x) & RTE_IPV6_EHDR_MF_MASK)
140 #define RTE_IPV6_GET_FO(x) ((x) >> RTE_IPV6_EHDR_FO_SHIFT)
142 #define RTE_IPV6_SET_FRAG_DATA(fo, mf) \
143 (((fo) & RTE_IPV6_EHDR_FO_MASK) | ((mf) & RTE_IPV6_EHDR_MF_MASK))
145 struct ipv6_extension_fragment {
146 uint8_t next_header; /**< Next header type */
147 uint8_t reserved; /**< Reserved */
148 uint16_t frag_data; /**< All fragmentation data */
149 uint32_t id; /**< Packet ID */
150 } __attribute__((__packed__));
155 * Create a new IP fragmentation table.
158 * Number of buckets in the hash table.
159 * @param bucket_entries
160 * Number of entries per bucket (e.g. hash associativity).
161 * Should be power of two.
163 * Maximum number of entries that could be stored in the table.
164 * The value should be less or equal then bucket_num * bucket_entries.
166 * Maximum TTL in cycles for each fragmented packet.
168 * The *socket_id* argument is the socket identifier in the case of
169 * NUMA. The value can be *SOCKET_ID_ANY* if there is no NUMA constraints.
171 * The pointer to the new allocated fragmentation table, on success. NULL on error.
173 struct rte_ip_frag_tbl * rte_ip_frag_table_create(uint32_t bucket_num,
174 uint32_t bucket_entries, uint32_t max_entries,
175 uint64_t max_cycles, int socket_id);
178 * Free allocated IP fragmentation table.
181 * Fragmentation table to free.
184 rte_ip_frag_table_destroy(struct rte_ip_frag_tbl *tbl);
187 * This function implements the fragmentation of IPv6 packets.
192 * Array storing the output fragments.
194 * Number of fragments.
196 * Size in bytes of the Maximum Transfer Unit (MTU) for the outgoing IPv6
197 * datagrams. This value includes the size of the IPv6 header.
199 * MBUF pool used for allocating direct buffers for the output fragments.
200 * @param pool_indirect
201 * MBUF pool used for allocating indirect buffers for the output fragments.
203 * Upon successful completion - number of output fragments placed
204 * in the pkts_out array.
205 * Otherwise - (-1) * errno.
208 rte_ipv6_fragment_packet(struct rte_mbuf *pkt_in,
209 struct rte_mbuf **pkts_out,
210 uint16_t nb_pkts_out,
212 struct rte_mempool *pool_direct,
213 struct rte_mempool *pool_indirect);
216 * This function implements reassembly of fragmented IPv6 packets.
217 * Incoming mbuf should have its l2_len/l3_len fields setup correctly.
220 * Table where to lookup/add the fragmented packet.
222 * Death row to free buffers to
224 * Incoming mbuf with IPv6 fragment.
226 * Fragment arrival timestamp.
228 * Pointer to the IPv6 header.
230 * Pointer to the IPv6 fragment extension header.
232 * Pointer to mbuf for reassembled packet, or NULL if:
233 * - an error occurred.
234 * - not all fragments of the packet are collected yet.
236 struct rte_mbuf *rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
237 struct rte_ip_frag_death_row *dr,
238 struct rte_mbuf *mb, uint64_t tms, struct ipv6_hdr *ip_hdr,
239 struct ipv6_extension_fragment *frag_hdr);
242 * Return a pointer to the packet's fragment header, if found.
243 * It only looks at the extension header that's right after the fixed IPv6
244 * header, and doesn't follow the whole chain of extension headers.
247 * Pointer to the IPv6 header.
249 * Pointer to the IPv6 fragment extension header, or NULL if it's not
252 static inline struct ipv6_extension_fragment *
253 rte_ipv6_frag_get_ipv6_fragment_header(struct ipv6_hdr *hdr)
255 if (hdr->proto == IPPROTO_FRAGMENT) {
256 return (struct ipv6_extension_fragment *) ++hdr;
263 * IPv4 fragmentation.
265 * This function implements the fragmentation of IPv4 packets.
270 * Array storing the output fragments.
272 * Number of fragments.
274 * Size in bytes of the Maximum Transfer Unit (MTU) for the outgoing IPv4
275 * datagrams. This value includes the size of the IPv4 header.
277 * MBUF pool used for allocating direct buffers for the output fragments.
278 * @param pool_indirect
279 * MBUF pool used for allocating indirect buffers for the output fragments.
281 * Upon successful completion - number of output fragments placed
282 * in the pkts_out array.
283 * Otherwise - (-1) * errno.
285 int32_t rte_ipv4_fragment_packet(struct rte_mbuf *pkt_in,
286 struct rte_mbuf **pkts_out,
287 uint16_t nb_pkts_out, uint16_t mtu_size,
288 struct rte_mempool *pool_direct,
289 struct rte_mempool *pool_indirect);
292 * This function implements reassembly of fragmented IPv4 packets.
293 * Incoming mbufs should have its l2_len/l3_len fields setup correclty.
296 * Table where to lookup/add the fragmented packet.
298 * Death row to free buffers to
300 * Incoming mbuf with IPv4 fragment.
302 * Fragment arrival timestamp.
304 * Pointer to the IPV4 header inside the fragment.
306 * Pointer to mbuf for reassembled packet, or NULL if:
307 * - an error occurred.
308 * - not all fragments of the packet are collected yet.
310 struct rte_mbuf * rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
311 struct rte_ip_frag_death_row *dr,
312 struct rte_mbuf *mb, uint64_t tms, struct ipv4_hdr *ip_hdr);
315 * Check if the IPv4 packet is fragmented
318 * IPv4 header of the packet
320 * 1 if fragmented, 0 if not fragmented
323 rte_ipv4_frag_pkt_is_fragmented(const struct ipv4_hdr * hdr) {
324 uint16_t flag_offset, ip_flag, ip_ofs;
326 flag_offset = rte_be_to_cpu_16(hdr->fragment_offset);
327 ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
328 ip_flag = (uint16_t)(flag_offset & IPV4_HDR_MF_FLAG);
330 return ip_flag != 0 || ip_ofs != 0;
334 * Free mbufs on a given death row.
337 * Death row to free mbufs in.
339 * How many buffers to prefetch before freeing.
341 void rte_ip_frag_free_death_row(struct rte_ip_frag_death_row *dr,
346 * Dump fragmentation table statistics to file.
349 * File to dump statistics to
351 * Fragmentation table to dump statistics from
354 rte_ip_frag_table_statistics_dump(FILE * f, const struct rte_ip_frag_tbl *tbl);
360 #endif /* _RTE_IP_FRAG_H_ */