2 Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions
9 * Redistributions of source code must retain the above copyright
10 notice, this list of conditions and the following disclaimer.
11 * Redistributions in binary form must reproduce the above copyright
12 notice, this list of conditions and the following disclaimer in
13 the documentation and/or other materials provided with the
15 * Neither the name of Intel Corporation nor the names of its
16 contributors may be used to endorse or promote products derived
17 from this software without specific prior written permission.
19 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
31 IP Reassembly Sample Application
32 ================================
34 The L3 Forwarding application is a simple example of packet processing using the DPDK.
35 The application performs L3 forwarding with reassembly for fragmented IPv4 and IPv6 packets.
40 The application demonstrates the use of the DPDK libraries to implement packet forwarding
41 with reassembly for IPv4 and IPv6 fragmented packets.
42 The initialization and run- time paths are very similar to those of the :doc:`l2_forward_real_virtual`.
43 The main difference from the L2 Forwarding sample application is that
44 it reassembles fragmented IPv4 and IPv6 packets before forwarding.
45 The maximum allowed size of reassembled packet is 9.5 KB.
47 There are two key differences from the L2 Forwarding sample application:
49 * The first difference is that the forwarding decision is taken based on information read from the input packet's IP header.
51 * The second difference is that the application differentiates between IP and non-IP traffic by means of offload flags.
53 The Longest Prefix Match (LPM for IPv4, LPM6 for IPv6) table is used to store/lookup an outgoing port number, associated with that IPv4 address. Any unmatched packets are forwarded to the originating port.Compiling the Application
54 --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
56 To compile the application:
58 #. Go to the sample application directory:
60 .. code-block:: console
62 export RTE_SDK=/path/to/rte_sdk
63 cd ${RTE_SDK}/examples/ip_reassembly
65 #. Set the target (a default target is used if not specified). For example:
67 .. code-block:: console
69 export RTE_TARGET=x86_64-native-linuxapp-gcc
71 See the *DPDK Getting Started Guide* for possible RTE_TARGET values.
73 #. Build the application:
75 .. code-block:: console
79 Running the Application
80 -----------------------
82 The application has a number of command line options:
84 .. code-block:: console
86 ./build/ip_reassembly [EAL options] -- -p PORTMASK [-q NQ] [--maxflows=FLOWS>] [--flowttl=TTL[(s|ms)]]
90 * -p PORTMASK: Hexadecimal bitmask of ports to configure
92 * -q NQ: Number of RX queues per lcore
94 * --maxflows=FLOWS: determines maximum number of active fragmented flows (1-65535). Default value: 4096.
96 * --flowttl=TTL[(s|ms)]: determines maximum Time To Live for fragmented packet.
97 If all fragments of the packet wouldn't appear within given time-out,
98 then they are considered as invalid and will be dropped.
99 Valid range is 1ms - 3600s. Default value: 1s.
101 To run the example in linuxapp environment with 2 lcores (2,4) over 2 ports(0,2) with 1 RX queue per lcore:
103 .. code-block:: console
105 ./build/ip_reassembly -l 2,4 -n 3 -- -p 5
106 EAL: coremask set to 14
107 EAL: Detected lcore 0 on socket 0
108 EAL: Detected lcore 1 on socket 1
109 EAL: Detected lcore 2 on socket 0
110 EAL: Detected lcore 3 on socket 1
111 EAL: Detected lcore 4 on socket 0
114 Initializing port 0 on lcore 2... Address:00:1B:21:76:FA:2C, rxq=0 txq=2,0 txq=4,1
115 done: Link Up - speed 10000 Mbps - full-duplex
116 Skipping disabled port 1
117 Initializing port 2 on lcore 4... Address:00:1B:21:5C:FF:54, rxq=0 txq=2,0 txq=4,1
118 done: Link Up - speed 10000 Mbps - full-duplex
119 Skipping disabled port 3IP_FRAG: Socket 0: adding route 100.10.0.0/16 (port 0)
120 IP_RSMBL: Socket 0: adding route 100.20.0.0/16 (port 1)
123 IP_RSMBL: Socket 0: adding route 0101:0101:0101:0101:0101:0101:0101:0101/48 (port 0)
124 IP_RSMBL: Socket 0: adding route 0201:0101:0101:0101:0101:0101:0101:0101/48 (port 1)
127 IP_RSMBL: entering main loop on lcore 4
128 IP_RSMBL: -- lcoreid=4 portid=2
129 IP_RSMBL: entering main loop on lcore 2
130 IP_RSMBL: -- lcoreid=2 portid=0
132 To run the example in linuxapp environment with 1 lcore (4) over 2 ports(0,2) with 2 RX queues per lcore:
134 .. code-block:: console
136 ./build/ip_reassembly -l 4 -n 3 -- -p 5 -q 2
138 To test the application, flows should be set up in the flow generator that match the values in the
139 l3fwd_ipv4_route_array and/or l3fwd_ipv6_route_array table.
141 Please note that in order to test this application,
142 the traffic generator should be generating valid fragmented IP packets.
143 For IPv6, the only supported case is when no other extension headers other than
144 fragment extension header are present in the packet.
146 The default l3fwd_ipv4_route_array table is:
150 struct l3fwd_ipv4_route l3fwd_ipv4_route_array[] = {
151 {IPv4(100, 10, 0, 0), 16, 0},
152 {IPv4(100, 20, 0, 0), 16, 1},
153 {IPv4(100, 30, 0, 0), 16, 2},
154 {IPv4(100, 40, 0, 0), 16, 3},
155 {IPv4(100, 50, 0, 0), 16, 4},
156 {IPv4(100, 60, 0, 0), 16, 5},
157 {IPv4(100, 70, 0, 0), 16, 6},
158 {IPv4(100, 80, 0, 0), 16, 7},
161 The default l3fwd_ipv6_route_array table is:
165 struct l3fwd_ipv6_route l3fwd_ipv6_route_array[] = {
166 {{1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 0},
167 {{2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 1},
168 {{3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 2},
169 {{4, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 3},
170 {{5, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 4},
171 {{6, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 5},
172 {{7, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 6},
173 {{8, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1}, 48, 7},
176 For example, for the fragmented input IPv4 packet with destination address: 100.10.1.1,
177 a reassembled IPv4 packet be sent out from port #0 to the destination address 100.10.1.1
178 once all the fragments are collected.
183 The following sections provide some explanation of the sample application code.
184 As mentioned in the overview section, the initialization and run-time paths are very similar to those of the :doc:`l2_forward_real_virtual`.
185 The following sections describe aspects that are specific to the IP reassemble sample application.
187 IPv4 Fragment Table Initialization
188 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
190 This application uses the rte_ip_frag library. Please refer to Programmer's Guide for more detailed explanation of how to use this library.
191 Fragment table maintains information about already received fragments of the packet.
192 Each IP packet is uniquely identified by triple <Source IP address>, <Destination IP address>, <ID>.
193 To avoid lock contention, each RX queue has its own Fragment Table,
194 e.g. the application can't handle the situation when different fragments of the same packet arrive through different RX queues.
195 Each table entry can hold information about packet consisting of up to RTE_LIBRTE_IP_FRAG_MAX_FRAGS fragments.
199 frag_cycles = (rte_get_tsc_hz() + MS_PER_S - 1) / MS_PER_S * max_flow_ttl;
201 if ((qconf->frag_tbl[queue] = rte_ip_frag_tbl_create(max_flow_num, IPV4_FRAG_TBL_BUCKET_ENTRIES, max_flow_num, frag_cycles, socket)) == NULL)
203 RTE_LOG(ERR, IP_RSMBL, "ip_frag_tbl_create(%u) on " "lcore: %u for queue: %u failed\n", max_flow_num, lcore, queue);
207 Mempools Initialization
208 ~~~~~~~~~~~~~~~~~~~~~~~
210 The reassembly application demands a lot of mbuf's to be allocated.
211 At any given time up to (2 \* max_flow_num \* RTE_LIBRTE_IP_FRAG_MAX_FRAGS \* <maximum number of mbufs per packet>)
212 can be stored inside Fragment Table waiting for remaining fragments.
213 To keep mempool size under reasonable limits and to avoid situation when one RX queue can starve other queues,
214 each RX queue uses its own mempool.
218 nb_mbuf = RTE_MAX(max_flow_num, 2UL * MAX_PKT_BURST) * RTE_LIBRTE_IP_FRAG_MAX_FRAGS;
219 nb_mbuf *= (port_conf.rxmode.max_rx_pkt_len + BUF_SIZE - 1) / BUF_SIZE;
220 nb_mbuf *= 2; /* ipv4 and ipv6 */
221 nb_mbuf += RTE_TEST_RX_DESC_DEFAULT + RTE_TEST_TX_DESC_DEFAULT;
222 nb_mbuf = RTE_MAX(nb_mbuf, (uint32_t)NB_MBUF);
224 snprintf(buf, sizeof(buf), "mbuf_pool_%u_%u", lcore, queue);
226 if ((rxq->pool = rte_mempool_create(buf, nb_mbuf, MBUF_SIZE, 0, sizeof(struct rte_pktmbuf_pool_private), rte_pktmbuf_pool_init, NULL,
227 rte_pktmbuf_init, NULL, socket, MEMPOOL_F_SP_PUT | MEMPOOL_F_SC_GET)) == NULL) {
229 RTE_LOG(ERR, IP_RSMBL, "mempool_create(%s) failed", buf);
233 Packet Reassembly and Forwarding
234 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
236 For each input packet, the packet forwarding operation is done by the l3fwd_simple_forward() function.
237 If the packet is an IPv4 or IPv6 fragment, then it calls rte_ipv4_reassemble_packet() for IPv4 packets,
238 or rte_ipv6_reassemble_packet() for IPv6 packets.
239 These functions either return a pointer to valid mbuf that contains reassembled packet,
240 or NULL (if the packet can't be reassembled for some reason).
241 Then l3fwd_simple_forward() continues with the code for the packet forwarding decision
242 (that is, the identification of the output interface for the packet) and
243 actual transmit of the packet.
245 The rte_ipv4_reassemble_packet() or rte_ipv6_reassemble_packet() are responsible for:
247 #. Searching the Fragment Table for entry with packet's <IP Source Address, IP Destination Address, Packet ID>
249 #. If the entry is found, then check if that entry already timed-out.
250 If yes, then free all previously received fragments,
251 and remove information about them from the entry.
253 #. If no entry with such key is found, then try to create a new one by one of two ways:
255 #. Use as empty entry
257 #. Delete a timed-out entry, free mbufs associated with it mbufs and store a new entry with specified key in it.
259 #. Update the entry with new fragment information and check
260 if a packet can be reassembled (the packet's entry contains all fragments).
262 #. If yes, then, reassemble the packet, mark table's entry as empty and return the reassembled mbuf to the caller.
264 #. If no, then just return a NULL to the caller.
266 If at any stage of packet processing a reassembly function encounters an error
267 (can't insert new entry into the Fragment table, or invalid/timed-out fragment),
268 then it will free all associated with the packet fragments,
269 mark the table entry as invalid and return NULL to the caller.
271 Debug logging and Statistics Collection
272 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
274 The RTE_LIBRTE_IP_FRAG_TBL_STAT controls statistics collection for the IP Fragment Table.
275 This macro is disabled by default.
276 To make ip_reassembly print the statistics to the standard output,
277 the user must send either an USR1, INT or TERM signal to the process.
278 For all of these signals, the ip_reassembly process prints Fragment table statistics for each RX queue,
279 plus the INT and TERM will cause process termination as usual.