4 * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
5 * Copyright 2014 6WIND S.A.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
12 * * Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * * Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in
16 * the documentation and/or other materials provided with the
18 * * Neither the name of Intel Corporation nor the names of its
19 * contributors may be used to endorse or promote products derived
20 * from this software without specific prior written permission.
22 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
23 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
24 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
25 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
26 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
27 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
28 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
29 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
30 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
31 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
32 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
42 * The mbuf library provides the ability to create and destroy buffers
43 * that may be used by the RTE application to store message
44 * buffers. The message buffers are stored in a mempool, using the
45 * RTE mempool library.
47 * This library provide an API to allocate/free packet mbufs, which are
48 * used to carry network packets.
50 * To understand the concepts of packet buffers or mbufs, you
51 * should read "TCP/IP Illustrated, Volume 2: The Implementation,
52 * Addison-Wesley, 1995, ISBN 0-201-63354-X from Richard Stevens"
53 * http://www.kohala.com/start/tcpipiv2.html
57 #include <rte_common.h>
58 #include <rte_mempool.h>
59 #include <rte_memory.h>
60 #include <rte_atomic.h>
61 #include <rte_prefetch.h>
62 #include <rte_branch_prediction.h>
68 /* deprecated options */
69 #pragma GCC poison RTE_MBUF_SCATTER_GATHER
70 #pragma GCC poison RTE_MBUF_REFCNT
73 * Packet Offload Features Flags. It also carry packet type information.
74 * Critical resources. Both rx/tx shared these bits. Be cautious on any change
76 * - RX flags start at bit position zero, and get added to the left of previous
78 * - The most-significant 8 bits are reserved for generic mbuf flags
79 * - TX flags therefore start at bit position 55 (i.e. 63-8), and new flags get
80 * added to the right of the previously defined flags
82 * Keep these flags synchronized with rte_get_rx_ol_flag_name() and
83 * rte_get_tx_ol_flag_name().
85 #define PKT_RX_VLAN_PKT (1ULL << 0) /**< RX packet is a 802.1q VLAN packet. */
86 #define PKT_RX_RSS_HASH (1ULL << 1) /**< RX packet with RSS hash result. */
87 #define PKT_RX_FDIR (1ULL << 2) /**< RX packet with FDIR match indicate. */
88 #define PKT_RX_L4_CKSUM_BAD (1ULL << 3) /**< L4 cksum of RX pkt. is not OK. */
89 #define PKT_RX_IP_CKSUM_BAD (1ULL << 4) /**< IP cksum of RX pkt. is not OK. */
90 #define PKT_RX_EIP_CKSUM_BAD (0ULL << 0) /**< External IP header checksum error. */
91 #define PKT_RX_OVERSIZE (0ULL << 0) /**< Num of desc of an RX pkt oversize. */
92 #define PKT_RX_HBUF_OVERFLOW (0ULL << 0) /**< Header buffer overflow. */
93 #define PKT_RX_RECIP_ERR (0ULL << 0) /**< Hardware processing error. */
94 #define PKT_RX_MAC_ERR (0ULL << 0) /**< MAC error. */
96 #define PKT_RX_IPV4_HDR (1ULL << 5) /**< RX packet with IPv4 header. */
97 #define PKT_RX_IPV4_HDR_EXT (1ULL << 6) /**< RX packet with extended IPv4 header. */
98 #define PKT_RX_IPV6_HDR (1ULL << 7) /**< RX packet with IPv6 header. */
99 #define PKT_RX_IPV6_HDR_EXT (1ULL << 8) /**< RX packet with extended IPv6 header. */
100 #endif /* RTE_NEXT_ABI */
101 #define PKT_RX_IEEE1588_PTP (1ULL << 9) /**< RX IEEE1588 L2 Ethernet PT Packet. */
102 #define PKT_RX_IEEE1588_TMST (1ULL << 10) /**< RX IEEE1588 L2/L4 timestamped packet.*/
104 #define PKT_RX_TUNNEL_IPV4_HDR (1ULL << 11) /**< RX tunnel packet with IPv4 header.*/
105 #define PKT_RX_TUNNEL_IPV6_HDR (1ULL << 12) /**< RX tunnel packet with IPv6 header. */
106 #endif /* RTE_NEXT_ABI */
107 #define PKT_RX_FDIR_ID (1ULL << 13) /**< FD id reported if FDIR match. */
108 #define PKT_RX_FDIR_FLX (1ULL << 14) /**< Flexible bytes reported if FDIR match. */
109 #define PKT_RX_QINQ_PKT (1ULL << 15) /**< RX packet with double VLAN stripped. */
110 /* add new RX flags here */
112 /* add new TX flags here */
115 * Second VLAN insertion (QinQ) flag.
117 #define PKT_TX_QINQ_PKT (1ULL << 49) /**< TX packet with double VLAN inserted. */
120 * TCP segmentation offload. To enable this offload feature for a
121 * packet to be transmitted on hardware supporting TSO:
122 * - set the PKT_TX_TCP_SEG flag in mbuf->ol_flags (this flag implies
124 * - set the flag PKT_TX_IPV4 or PKT_TX_IPV6
125 * - if it's IPv4, set the PKT_TX_IP_CKSUM flag and write the IP checksum
127 * - fill the mbuf offload information: l2_len, l3_len, l4_len, tso_segsz
128 * - calculate the pseudo header checksum without taking ip_len in account,
129 * and set it in the TCP header. Refer to rte_ipv4_phdr_cksum() and
130 * rte_ipv6_phdr_cksum() that can be used as helpers.
132 #define PKT_TX_TCP_SEG (1ULL << 50)
134 #define PKT_TX_IEEE1588_TMST (1ULL << 51) /**< TX IEEE1588 packet to timestamp. */
137 * Bits 52+53 used for L4 packet type with checksum enabled: 00: Reserved,
138 * 01: TCP checksum, 10: SCTP checksum, 11: UDP checksum. To use hardware
139 * L4 checksum offload, the user needs to:
140 * - fill l2_len and l3_len in mbuf
141 * - set the flags PKT_TX_TCP_CKSUM, PKT_TX_SCTP_CKSUM or PKT_TX_UDP_CKSUM
142 * - set the flag PKT_TX_IPV4 or PKT_TX_IPV6
143 * - calculate the pseudo header checksum and set it in the L4 header (only
144 * for TCP or UDP). See rte_ipv4_phdr_cksum() and rte_ipv6_phdr_cksum().
145 * For SCTP, set the crc field to 0.
147 #define PKT_TX_L4_NO_CKSUM (0ULL << 52) /**< Disable L4 cksum of TX pkt. */
148 #define PKT_TX_TCP_CKSUM (1ULL << 52) /**< TCP cksum of TX pkt. computed by NIC. */
149 #define PKT_TX_SCTP_CKSUM (2ULL << 52) /**< SCTP cksum of TX pkt. computed by NIC. */
150 #define PKT_TX_UDP_CKSUM (3ULL << 52) /**< UDP cksum of TX pkt. computed by NIC. */
151 #define PKT_TX_L4_MASK (3ULL << 52) /**< Mask for L4 cksum offload request. */
154 * Offload the IP checksum in the hardware. The flag PKT_TX_IPV4 should
155 * also be set by the application, although a PMD will only check
157 * - set the IP checksum field in the packet to 0
158 * - fill the mbuf offload information: l2_len, l3_len
160 #define PKT_TX_IP_CKSUM (1ULL << 54)
163 * Packet is IPv4. This flag must be set when using any offload feature
164 * (TSO, L3 or L4 checksum) to tell the NIC that the packet is an IPv4
165 * packet. If the packet is a tunneled packet, this flag is related to
168 #define PKT_TX_IPV4 (1ULL << 55)
171 * Packet is IPv6. This flag must be set when using an offload feature
172 * (TSO or L4 checksum) to tell the NIC that the packet is an IPv6
173 * packet. If the packet is a tunneled packet, this flag is related to
176 #define PKT_TX_IPV6 (1ULL << 56)
178 #define PKT_TX_VLAN_PKT (1ULL << 57) /**< TX packet is a 802.1q VLAN packet. */
181 * Offload the IP checksum of an external header in the hardware. The
182 * flag PKT_TX_OUTER_IPV4 should also be set by the application, alto ugh
183 * a PMD will only check PKT_TX_IP_CKSUM. The IP checksum field in the
184 * packet must be set to 0.
185 * - set the outer IP checksum field in the packet to 0
186 * - fill the mbuf offload information: outer_l2_len, outer_l3_len
188 #define PKT_TX_OUTER_IP_CKSUM (1ULL << 58)
191 * Packet outer header is IPv4. This flag must be set when using any
192 * outer offload feature (L3 or L4 checksum) to tell the NIC that the
193 * outer header of the tunneled packet is an IPv4 packet.
195 #define PKT_TX_OUTER_IPV4 (1ULL << 59)
198 * Packet outer header is IPv6. This flag must be set when using any
199 * outer offload feature (L4 checksum) to tell the NIC that the outer
200 * header of the tunneled packet is an IPv6 packet.
202 #define PKT_TX_OUTER_IPV6 (1ULL << 60)
204 #define IND_ATTACHED_MBUF (1ULL << 62) /**< Indirect attached mbuf */
206 /* Use final bit of flags to indicate a control mbuf */
207 #define CTRL_MBUF_FLAG (1ULL << 63) /**< Mbuf contains control data */
211 * 32 bits are divided into several fields to mark packet types. Note that
212 * each field is indexical.
213 * - Bit 3:0 is for L2 types.
214 * - Bit 7:4 is for L3 or outer L3 (for tunneling case) types.
215 * - Bit 11:8 is for L4 or outer L4 (for tunneling case) types.
216 * - Bit 15:12 is for tunnel types.
217 * - Bit 19:16 is for inner L2 types.
218 * - Bit 23:20 is for inner L3 types.
219 * - Bit 27:24 is for inner L4 types.
220 * - Bit 31:28 is reserved.
222 * To be compatible with Vector PMD, RTE_PTYPE_L3_IPV4, RTE_PTYPE_L3_IPV4_EXT,
223 * RTE_PTYPE_L3_IPV6, RTE_PTYPE_L3_IPV6_EXT, RTE_PTYPE_L4_TCP, RTE_PTYPE_L4_UDP
224 * and RTE_PTYPE_L4_SCTP should be kept as below in a contiguous 7 bits.
226 * Note that L3 types values are selected for checking IPV4/IPV6 header from
227 * performance point of view. Reading annotations of RTE_ETH_IS_IPV4_HDR and
228 * RTE_ETH_IS_IPV6_HDR is needed for any future changes of L3 type values.
230 * Note that the packet types of the same packet recognized by different
231 * hardware may be different, as different hardware may have different
232 * capability of packet type recognition.
235 * <'ether type'=0x0800
236 * | 'version'=4, 'protocol'=0x29
237 * | 'version'=6, 'next header'=0x3A
239 * will be recognized on i40e hardware as packet type combination of,
240 * RTE_PTYPE_L2_ETHER |
241 * RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
242 * RTE_PTYPE_TUNNEL_IP |
243 * RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
244 * RTE_PTYPE_INNER_L4_ICMP.
246 * <'ether type'=0x86DD
247 * | 'version'=6, 'next header'=0x2F
249 * | 'version'=6, 'next header'=0x11
251 * will be recognized on i40e hardware as packet type combination of,
252 * RTE_PTYPE_L2_ETHER |
253 * RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
254 * RTE_PTYPE_TUNNEL_GRENAT |
255 * RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
256 * RTE_PTYPE_INNER_L4_UDP.
258 #define RTE_PTYPE_UNKNOWN 0x00000000
260 * Ethernet packet type.
261 * It is used for outer packet for tunneling cases.
264 * <'ether type'=[0x0800|0x86DD]>
266 #define RTE_PTYPE_L2_ETHER 0x00000001
268 * Ethernet packet type for time sync.
271 * <'ether type'=0x88F7>
273 #define RTE_PTYPE_L2_ETHER_TIMESYNC 0x00000002
275 * ARP (Address Resolution Protocol) packet type.
278 * <'ether type'=0x0806>
280 #define RTE_PTYPE_L2_ETHER_ARP 0x00000003
282 * LLDP (Link Layer Discovery Protocol) packet type.
285 * <'ether type'=0x88CC>
287 #define RTE_PTYPE_L2_ETHER_LLDP 0x00000004
289 * Mask of layer 2 packet types.
290 * It is used for outer packet for tunneling cases.
292 #define RTE_PTYPE_L2_MASK 0x0000000f
294 * IP (Internet Protocol) version 4 packet type.
295 * It is used for outer packet for tunneling cases, and does not contain any
299 * <'ether type'=0x0800
300 * | 'version'=4, 'ihl'=5>
302 #define RTE_PTYPE_L3_IPV4 0x00000010
304 * IP (Internet Protocol) version 4 packet type.
305 * It is used for outer packet for tunneling cases, and contains header
309 * <'ether type'=0x0800
310 * | 'version'=4, 'ihl'=[6-15], 'options'>
312 #define RTE_PTYPE_L3_IPV4_EXT 0x00000030
314 * IP (Internet Protocol) version 6 packet type.
315 * It is used for outer packet for tunneling cases, and does not contain any
319 * <'ether type'=0x86DD
320 * | 'version'=6, 'next header'=0x3B>
322 #define RTE_PTYPE_L3_IPV6 0x00000040
324 * IP (Internet Protocol) version 4 packet type.
325 * It is used for outer packet for tunneling cases, and may or maynot contain
329 * <'ether type'=0x0800
330 * | 'version'=4, 'ihl'=[5-15], <'options'>>
332 #define RTE_PTYPE_L3_IPV4_EXT_UNKNOWN 0x00000090
334 * IP (Internet Protocol) version 6 packet type.
335 * It is used for outer packet for tunneling cases, and contains extension
339 * <'ether type'=0x86DD
340 * | 'version'=6, 'next header'=[0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
341 * 'extension headers'>
343 #define RTE_PTYPE_L3_IPV6_EXT 0x000000c0
345 * IP (Internet Protocol) version 6 packet type.
346 * It is used for outer packet for tunneling cases, and may or maynot contain
350 * <'ether type'=0x86DD
351 * | 'version'=6, 'next header'=[0x3B|0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
352 * <'extension headers'>>
354 #define RTE_PTYPE_L3_IPV6_EXT_UNKNOWN 0x000000e0
356 * Mask of layer 3 packet types.
357 * It is used for outer packet for tunneling cases.
359 #define RTE_PTYPE_L3_MASK 0x000000f0
361 * TCP (Transmission Control Protocol) packet type.
362 * It is used for outer packet for tunneling cases.
365 * <'ether type'=0x0800
366 * | 'version'=4, 'protocol'=6, 'MF'=0>
368 * <'ether type'=0x86DD
369 * | 'version'=6, 'next header'=6>
371 #define RTE_PTYPE_L4_TCP 0x00000100
373 * UDP (User Datagram Protocol) packet type.
374 * It is used for outer packet for tunneling cases.
377 * <'ether type'=0x0800
378 * | 'version'=4, 'protocol'=17, 'MF'=0>
380 * <'ether type'=0x86DD
381 * | 'version'=6, 'next header'=17>
383 #define RTE_PTYPE_L4_UDP 0x00000200
385 * Fragmented IP (Internet Protocol) packet type.
386 * It is used for outer packet for tunneling cases.
388 * It refers to those packets of any IP types, which can be recognized as
389 * fragmented. A fragmented packet cannot be recognized as any other L4 types
390 * (RTE_PTYPE_L4_TCP, RTE_PTYPE_L4_UDP, RTE_PTYPE_L4_SCTP, RTE_PTYPE_L4_ICMP,
391 * RTE_PTYPE_L4_NONFRAG).
394 * <'ether type'=0x0800
395 * | 'version'=4, 'MF'=1>
397 * <'ether type'=0x86DD
398 * | 'version'=6, 'next header'=44>
400 #define RTE_PTYPE_L4_FRAG 0x00000300
402 * SCTP (Stream Control Transmission Protocol) packet type.
403 * It is used for outer packet for tunneling cases.
406 * <'ether type'=0x0800
407 * | 'version'=4, 'protocol'=132, 'MF'=0>
409 * <'ether type'=0x86DD
410 * | 'version'=6, 'next header'=132>
412 #define RTE_PTYPE_L4_SCTP 0x00000400
414 * ICMP (Internet Control Message Protocol) packet type.
415 * It is used for outer packet for tunneling cases.
418 * <'ether type'=0x0800
419 * | 'version'=4, 'protocol'=1, 'MF'=0>
421 * <'ether type'=0x86DD
422 * | 'version'=6, 'next header'=1>
424 #define RTE_PTYPE_L4_ICMP 0x00000500
426 * Non-fragmented IP (Internet Protocol) packet type.
427 * It is used for outer packet for tunneling cases.
429 * It refers to those packets of any IP types, while cannot be recognized as
430 * any of above L4 types (RTE_PTYPE_L4_TCP, RTE_PTYPE_L4_UDP,
431 * RTE_PTYPE_L4_FRAG, RTE_PTYPE_L4_SCTP, RTE_PTYPE_L4_ICMP).
434 * <'ether type'=0x0800
435 * | 'version'=4, 'protocol'!=[6|17|132|1], 'MF'=0>
437 * <'ether type'=0x86DD
438 * | 'version'=6, 'next header'!=[6|17|44|132|1]>
440 #define RTE_PTYPE_L4_NONFRAG 0x00000600
442 * Mask of layer 4 packet types.
443 * It is used for outer packet for tunneling cases.
445 #define RTE_PTYPE_L4_MASK 0x00000f00
447 * IP (Internet Protocol) in IP (Internet Protocol) tunneling packet type.
450 * <'ether type'=0x0800
451 * | 'version'=4, 'protocol'=[4|41]>
453 * <'ether type'=0x86DD
454 * | 'version'=6, 'next header'=[4|41]>
456 #define RTE_PTYPE_TUNNEL_IP 0x00001000
458 * GRE (Generic Routing Encapsulation) tunneling packet type.
461 * <'ether type'=0x0800
462 * | 'version'=4, 'protocol'=47>
464 * <'ether type'=0x86DD
465 * | 'version'=6, 'next header'=47>
467 #define RTE_PTYPE_TUNNEL_GRE 0x00002000
469 * VXLAN (Virtual eXtensible Local Area Network) tunneling packet type.
472 * <'ether type'=0x0800
473 * | 'version'=4, 'protocol'=17
474 * | 'destination port'=4798>
476 * <'ether type'=0x86DD
477 * | 'version'=6, 'next header'=17
478 * | 'destination port'=4798>
480 #define RTE_PTYPE_TUNNEL_VXLAN 0x00003000
482 * NVGRE (Network Virtualization using Generic Routing Encapsulation) tunneling
486 * <'ether type'=0x0800
487 * | 'version'=4, 'protocol'=47
488 * | 'protocol type'=0x6558>
490 * <'ether type'=0x86DD
491 * | 'version'=6, 'next header'=47
492 * | 'protocol type'=0x6558'>
494 #define RTE_PTYPE_TUNNEL_NVGRE 0x00004000
496 * GENEVE (Generic Network Virtualization Encapsulation) tunneling packet type.
499 * <'ether type'=0x0800
500 * | 'version'=4, 'protocol'=17
501 * | 'destination port'=6081>
503 * <'ether type'=0x86DD
504 * | 'version'=6, 'next header'=17
505 * | 'destination port'=6081>
507 #define RTE_PTYPE_TUNNEL_GENEVE 0x00005000
509 * Tunneling packet type of Teredo, VXLAN (Virtual eXtensible Local Area
510 * Network) or GRE (Generic Routing Encapsulation) could be recognized as this
511 * packet type, if they can not be recognized independently as of hardware
514 #define RTE_PTYPE_TUNNEL_GRENAT 0x00006000
516 * Mask of tunneling packet types.
518 #define RTE_PTYPE_TUNNEL_MASK 0x0000f000
520 * Ethernet packet type.
521 * It is used for inner packet type only.
523 * Packet format (inner only):
524 * <'ether type'=[0x800|0x86DD]>
526 #define RTE_PTYPE_INNER_L2_ETHER 0x00010000
528 * Ethernet packet type with VLAN (Virtual Local Area Network) tag.
530 * Packet format (inner only):
531 * <'ether type'=[0x800|0x86DD], vlan=[1-4095]>
533 #define RTE_PTYPE_INNER_L2_ETHER_VLAN 0x00020000
535 * Mask of inner layer 2 packet types.
537 #define RTE_PTYPE_INNER_L2_MASK 0x000f0000
539 * IP (Internet Protocol) version 4 packet type.
540 * It is used for inner packet only, and does not contain any header option.
542 * Packet format (inner only):
543 * <'ether type'=0x0800
544 * | 'version'=4, 'ihl'=5>
546 #define RTE_PTYPE_INNER_L3_IPV4 0x00100000
548 * IP (Internet Protocol) version 4 packet type.
549 * It is used for inner packet only, and contains header options.
551 * Packet format (inner only):
552 * <'ether type'=0x0800
553 * | 'version'=4, 'ihl'=[6-15], 'options'>
555 #define RTE_PTYPE_INNER_L3_IPV4_EXT 0x00200000
557 * IP (Internet Protocol) version 6 packet type.
558 * It is used for inner packet only, and does not contain any extension header.
560 * Packet format (inner only):
561 * <'ether type'=0x86DD
562 * | 'version'=6, 'next header'=0x3B>
564 #define RTE_PTYPE_INNER_L3_IPV6 0x00300000
566 * IP (Internet Protocol) version 4 packet type.
567 * It is used for inner packet only, and may or maynot contain header options.
569 * Packet format (inner only):
570 * <'ether type'=0x0800
571 * | 'version'=4, 'ihl'=[5-15], <'options'>>
573 #define RTE_PTYPE_INNER_L3_IPV4_EXT_UNKNOWN 0x00400000
575 * IP (Internet Protocol) version 6 packet type.
576 * It is used for inner packet only, and contains extension headers.
578 * Packet format (inner only):
579 * <'ether type'=0x86DD
580 * | 'version'=6, 'next header'=[0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
581 * 'extension headers'>
583 #define RTE_PTYPE_INNER_L3_IPV6_EXT 0x00500000
585 * IP (Internet Protocol) version 6 packet type.
586 * It is used for inner packet only, and may or maynot contain extension
589 * Packet format (inner only):
590 * <'ether type'=0x86DD
591 * | 'version'=6, 'next header'=[0x3B|0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
592 * <'extension headers'>>
594 #define RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN 0x00600000
596 * Mask of inner layer 3 packet types.
598 #define RTE_PTYPE_INNER_INNER_L3_MASK 0x00f00000
600 * TCP (Transmission Control Protocol) packet type.
601 * It is used for inner packet only.
603 * Packet format (inner only):
604 * <'ether type'=0x0800
605 * | 'version'=4, 'protocol'=6, 'MF'=0>
607 * <'ether type'=0x86DD
608 * | 'version'=6, 'next header'=6>
610 #define RTE_PTYPE_INNER_L4_TCP 0x01000000
612 * UDP (User Datagram Protocol) packet type.
613 * It is used for inner packet only.
615 * Packet format (inner only):
616 * <'ether type'=0x0800
617 * | 'version'=4, 'protocol'=17, 'MF'=0>
619 * <'ether type'=0x86DD
620 * | 'version'=6, 'next header'=17>
622 #define RTE_PTYPE_INNER_L4_UDP 0x02000000
624 * Fragmented IP (Internet Protocol) packet type.
625 * It is used for inner packet only, and may or maynot have layer 4 packet.
627 * Packet format (inner only):
628 * <'ether type'=0x0800
629 * | 'version'=4, 'MF'=1>
631 * <'ether type'=0x86DD
632 * | 'version'=6, 'next header'=44>
634 #define RTE_PTYPE_INNER_L4_FRAG 0x03000000
636 * SCTP (Stream Control Transmission Protocol) packet type.
637 * It is used for inner packet only.
639 * Packet format (inner only):
640 * <'ether type'=0x0800
641 * | 'version'=4, 'protocol'=132, 'MF'=0>
643 * <'ether type'=0x86DD
644 * | 'version'=6, 'next header'=132>
646 #define RTE_PTYPE_INNER_L4_SCTP 0x04000000
648 * ICMP (Internet Control Message Protocol) packet type.
649 * It is used for inner packet only.
651 * Packet format (inner only):
652 * <'ether type'=0x0800
653 * | 'version'=4, 'protocol'=1, 'MF'=0>
655 * <'ether type'=0x86DD
656 * | 'version'=6, 'next header'=1>
658 #define RTE_PTYPE_INNER_L4_ICMP 0x05000000
660 * Non-fragmented IP (Internet Protocol) packet type.
661 * It is used for inner packet only, and may or maynot have other unknown layer
664 * Packet format (inner only):
665 * <'ether type'=0x0800
666 * | 'version'=4, 'protocol'!=[6|17|132|1], 'MF'=0>
668 * <'ether type'=0x86DD
669 * | 'version'=6, 'next header'!=[6|17|44|132|1]>
671 #define RTE_PTYPE_INNER_L4_NONFRAG 0x06000000
673 * Mask of inner layer 4 packet types.
675 #define RTE_PTYPE_INNER_L4_MASK 0x0f000000
678 * Check if the (outer) L3 header is IPv4. To avoid comparing IPv4 types one by
679 * one, bit 4 is selected to be used for IPv4 only. Then checking bit 4 can
680 * determin if it is an IPV4 packet.
682 #define RTE_ETH_IS_IPV4_HDR(ptype) ((ptype) & RTE_PTYPE_L3_IPV4)
685 * Check if the (outer) L3 header is IPv4. To avoid comparing IPv4 types one by
686 * one, bit 6 is selected to be used for IPv4 only. Then checking bit 6 can
687 * determin if it is an IPV4 packet.
689 #define RTE_ETH_IS_IPV6_HDR(ptype) ((ptype) & RTE_PTYPE_L3_IPV6)
691 /* Check if it is a tunneling packet */
692 #define RTE_ETH_IS_TUNNEL_PKT(ptype) ((ptype) & RTE_PTYPE_TUNNEL_MASK)
693 #endif /* RTE_NEXT_ABI */
696 * Get the name of a RX offload flag
699 * The mask describing the flag.
701 * The name of this flag, or NULL if it's not a valid RX flag.
703 const char *rte_get_rx_ol_flag_name(uint64_t mask);
706 * Get the name of a TX offload flag
709 * The mask describing the flag. Usually only one bit must be set.
710 * Several bits can be given if they belong to the same mask.
711 * Ex: PKT_TX_L4_MASK.
713 * The name of this flag, or NULL if it's not a valid TX flag.
715 const char *rte_get_tx_ol_flag_name(uint64_t mask);
718 * Some NICs need at least 2KB buffer to RX standard Ethernet frame without
719 * splitting it into multiple segments.
720 * So, for mbufs that planned to be involved into RX/TX, the recommended
721 * minimal buffer length is 2KB + RTE_PKTMBUF_HEADROOM.
723 #define RTE_MBUF_DEFAULT_DATAROOM 2048
724 #define RTE_MBUF_DEFAULT_BUF_SIZE \
725 (RTE_MBUF_DEFAULT_DATAROOM + RTE_PKTMBUF_HEADROOM)
727 /* define a set of marker types that can be used to refer to set points in the
729 typedef void *MARKER[0]; /**< generic marker for a point in a structure */
730 typedef uint8_t MARKER8[0]; /**< generic marker with 1B alignment */
731 typedef uint64_t MARKER64[0]; /**< marker that allows us to overwrite 8 bytes
732 * with a single assignment */
735 * The generic rte_mbuf, containing a packet mbuf.
740 void *buf_addr; /**< Virtual address of segment buffer. */
741 phys_addr_t buf_physaddr; /**< Physical address of segment buffer. */
743 uint16_t buf_len; /**< Length of segment buffer. */
745 /* next 6 bytes are initialised on RX descriptor rearm */
750 * 16-bit Reference counter.
751 * It should only be accessed using the following functions:
752 * rte_mbuf_refcnt_update(), rte_mbuf_refcnt_read(), and
753 * rte_mbuf_refcnt_set(). The functionality of these functions (atomic,
754 * or non-atomic) is controlled by the CONFIG_RTE_MBUF_REFCNT_ATOMIC
758 rte_atomic16_t refcnt_atomic; /**< Atomically accessed refcnt */
759 uint16_t refcnt; /**< Non-atomically accessed refcnt */
761 uint8_t nb_segs; /**< Number of segments. */
762 uint8_t port; /**< Input port. */
764 uint64_t ol_flags; /**< Offload features. */
766 /* remaining bytes are set on RX when pulling packet from descriptor */
767 MARKER rx_descriptor_fields1;
771 * The packet type, which is the combination of outer/inner L2, L3, L4
775 uint32_t packet_type; /**< L2/L3/L4 and tunnel information. */
777 uint32_t l2_type:4; /**< (Outer) L2 type. */
778 uint32_t l3_type:4; /**< (Outer) L3 type. */
779 uint32_t l4_type:4; /**< (Outer) L4 type. */
780 uint32_t tun_type:4; /**< Tunnel type. */
781 uint32_t inner_l2_type:4; /**< Inner L2 type. */
782 uint32_t inner_l3_type:4; /**< Inner L3 type. */
783 uint32_t inner_l4_type:4; /**< Inner L4 type. */
787 uint32_t pkt_len; /**< Total pkt len: sum of all segments. */
788 uint16_t data_len; /**< Amount of data in segment buffer. */
789 uint16_t vlan_tci; /**< VLAN Tag Control Identifier (CPU order) */
790 #else /* RTE_NEXT_ABI */
792 * The packet type, which is used to indicate ordinary packet and also
793 * tunneled packet format, i.e. each number is represented a type of
796 uint16_t packet_type;
798 uint16_t data_len; /**< Amount of data in segment buffer. */
799 uint32_t pkt_len; /**< Total pkt len: sum of all segments. */
800 uint16_t vlan_tci; /**< VLAN Tag Control Identifier (CPU order) */
801 uint16_t vlan_tci_outer; /**< Outer VLAN Tag Control Identifier (CPU order) */
802 #endif /* RTE_NEXT_ABI */
804 uint32_t rss; /**< RSS hash result if RSS enabled */
812 /**< Second 4 flexible bytes */
815 /**< First 4 flexible bytes or FD ID, dependent on
816 PKT_RX_FDIR_* flag in ol_flags. */
817 } fdir; /**< Filter identifier if FDIR enabled */
818 uint32_t sched; /**< Hierarchical scheduler */
819 uint32_t usr; /**< User defined tags. See rte_distributor_process() */
820 } hash; /**< hash information */
822 uint32_t seqn; /**< Sequence number. See also rte_reorder_insert() */
824 uint16_t vlan_tci_outer; /**< Outer VLAN Tag Control Identifier (CPU order) */
825 #endif /* RTE_NEXT_ABI */
827 /* second cache line - fields only used in slow path or on TX */
828 MARKER cacheline1 __rte_cache_aligned;
831 void *userdata; /**< Can be used for external metadata */
832 uint64_t udata64; /**< Allow 8-byte userdata on 32-bit */
835 struct rte_mempool *pool; /**< Pool from which mbuf was allocated. */
836 struct rte_mbuf *next; /**< Next segment of scattered packet. */
838 /* fields to support TX offloads */
840 uint64_t tx_offload; /**< combined for easy fetch */
842 uint64_t l2_len:7; /**< L2 (MAC) Header Length. */
843 uint64_t l3_len:9; /**< L3 (IP) Header Length. */
844 uint64_t l4_len:8; /**< L4 (TCP/UDP) Header Length. */
845 uint64_t tso_segsz:16; /**< TCP TSO segment size */
847 /* fields for TX offloading of tunnels */
848 uint64_t outer_l3_len:9; /**< Outer L3 (IP) Hdr Length. */
849 uint64_t outer_l2_len:7; /**< Outer L2 (MAC) Hdr Length. */
851 /* uint64_t unused:8; */
855 /** Size of the application private data. In case of an indirect
856 * mbuf, it stores the direct mbuf private data size. */
859 /** Timesync flags for use with IEEE1588. */
861 } __rte_cache_aligned;
863 static inline uint16_t rte_pktmbuf_priv_size(struct rte_mempool *mp);
866 * Return the mbuf owning the data buffer address of an indirect mbuf.
869 * The pointer to the indirect mbuf.
871 * The address of the direct mbuf corresponding to buffer_addr.
873 static inline struct rte_mbuf *
874 rte_mbuf_from_indirect(struct rte_mbuf *mi)
876 return RTE_PTR_SUB(mi->buf_addr, sizeof(*mi) + mi->priv_size);
880 * Return the buffer address embedded in the given mbuf.
883 * The pointer to the mbuf.
885 * The address of the data buffer owned by the mbuf.
888 rte_mbuf_to_baddr(struct rte_mbuf *md)
891 buffer_addr = (char *)md + sizeof(*md) + rte_pktmbuf_priv_size(md->pool);
896 * Returns TRUE if given mbuf is indirect, or FALSE otherwise.
898 #define RTE_MBUF_INDIRECT(mb) ((mb)->ol_flags & IND_ATTACHED_MBUF)
901 * Returns TRUE if given mbuf is direct, or FALSE otherwise.
903 #define RTE_MBUF_DIRECT(mb) (!RTE_MBUF_INDIRECT(mb))
906 * Private data in case of pktmbuf pool.
908 * A structure that contains some pktmbuf_pool-specific data that are
909 * appended after the mempool structure (in private data).
911 struct rte_pktmbuf_pool_private {
912 uint16_t mbuf_data_room_size; /**< Size of data space in each mbuf. */
913 uint16_t mbuf_priv_size; /**< Size of private area in each mbuf. */
916 #ifdef RTE_LIBRTE_MBUF_DEBUG
918 /** check mbuf type in debug mode */
919 #define __rte_mbuf_sanity_check(m, is_h) rte_mbuf_sanity_check(m, is_h)
921 /** check mbuf type in debug mode if mbuf pointer is not null */
922 #define __rte_mbuf_sanity_check_raw(m, is_h) do { \
924 rte_mbuf_sanity_check(m, is_h); \
927 /** MBUF asserts in debug mode */
928 #define RTE_MBUF_ASSERT(exp) \
930 rte_panic("line%d\tassert \"" #exp "\" failed\n", __LINE__); \
933 #else /* RTE_LIBRTE_MBUF_DEBUG */
935 /** check mbuf type in debug mode */
936 #define __rte_mbuf_sanity_check(m, is_h) do { } while (0)
938 /** check mbuf type in debug mode if mbuf pointer is not null */
939 #define __rte_mbuf_sanity_check_raw(m, is_h) do { } while (0)
941 /** MBUF asserts in debug mode */
942 #define RTE_MBUF_ASSERT(exp) do { } while (0)
944 #endif /* RTE_LIBRTE_MBUF_DEBUG */
946 #ifdef RTE_MBUF_REFCNT_ATOMIC
949 * Reads the value of an mbuf's refcnt.
953 * Reference count number.
955 static inline uint16_t
956 rte_mbuf_refcnt_read(const struct rte_mbuf *m)
958 return (uint16_t)(rte_atomic16_read(&m->refcnt_atomic));
962 * Sets an mbuf's refcnt to a defined value.
969 rte_mbuf_refcnt_set(struct rte_mbuf *m, uint16_t new_value)
971 rte_atomic16_set(&m->refcnt_atomic, new_value);
975 * Adds given value to an mbuf's refcnt and returns its new value.
979 * Value to add/subtract
983 static inline uint16_t
984 rte_mbuf_refcnt_update(struct rte_mbuf *m, int16_t value)
987 * The atomic_add is an expensive operation, so we don't want to
988 * call it in the case where we know we are the uniq holder of
989 * this mbuf (i.e. ref_cnt == 1). Otherwise, an atomic
990 * operation has to be used because concurrent accesses on the
991 * reference counter can occur.
993 if (likely(rte_mbuf_refcnt_read(m) == 1)) {
994 rte_mbuf_refcnt_set(m, 1 + value);
998 return (uint16_t)(rte_atomic16_add_return(&m->refcnt_atomic, value));
1001 #else /* ! RTE_MBUF_REFCNT_ATOMIC */
1004 * Adds given value to an mbuf's refcnt and returns its new value.
1006 static inline uint16_t
1007 rte_mbuf_refcnt_update(struct rte_mbuf *m, int16_t value)
1009 m->refcnt = (uint16_t)(m->refcnt + value);
1014 * Reads the value of an mbuf's refcnt.
1016 static inline uint16_t
1017 rte_mbuf_refcnt_read(const struct rte_mbuf *m)
1023 * Sets an mbuf's refcnt to the defined value.
1026 rte_mbuf_refcnt_set(struct rte_mbuf *m, uint16_t new_value)
1028 m->refcnt = new_value;
1031 #endif /* RTE_MBUF_REFCNT_ATOMIC */
1033 /** Mbuf prefetch */
1034 #define RTE_MBUF_PREFETCH_TO_FREE(m) do { \
1041 * Sanity checks on an mbuf.
1043 * Check the consistency of the given mbuf. The function will cause a
1044 * panic if corruption is detected.
1047 * The mbuf to be checked.
1049 * True if the mbuf is a packet header, false if it is a sub-segment
1050 * of a packet (in this case, some fields like nb_segs are not checked)
1053 rte_mbuf_sanity_check(const struct rte_mbuf *m, int is_header);
1056 * @internal Allocate a new mbuf from mempool *mp*.
1057 * The use of that function is reserved for RTE internal needs.
1058 * Please use rte_pktmbuf_alloc().
1061 * The mempool from which mbuf is allocated.
1063 * - The pointer to the new mbuf on success.
1064 * - NULL if allocation failed.
1066 static inline struct rte_mbuf *__rte_mbuf_raw_alloc(struct rte_mempool *mp)
1070 if (rte_mempool_get(mp, &mb) < 0)
1072 m = (struct rte_mbuf *)mb;
1073 RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(m) == 0);
1074 rte_mbuf_refcnt_set(m, 1);
1079 * @internal Put mbuf back into its original mempool.
1080 * The use of that function is reserved for RTE internal needs.
1081 * Please use rte_pktmbuf_free().
1084 * The mbuf to be freed.
1086 static inline void __attribute__((always_inline))
1087 __rte_mbuf_raw_free(struct rte_mbuf *m)
1089 RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(m) == 0);
1090 rte_mempool_put(m->pool, m);
1093 /* Operations on ctrl mbuf */
1096 * The control mbuf constructor.
1098 * This function initializes some fields in an mbuf structure that are
1099 * not modified by the user once created (mbuf type, origin pool, buffer
1100 * start address, and so on). This function is given as a callback function
1101 * to rte_mempool_create() at pool creation time.
1104 * The mempool from which the mbuf is allocated.
1106 * A pointer that can be used by the user to retrieve useful information
1107 * for mbuf initialization. This pointer comes from the ``init_arg``
1108 * parameter of rte_mempool_create().
1110 * The mbuf to initialize.
1112 * The index of the mbuf in the pool table.
1114 void rte_ctrlmbuf_init(struct rte_mempool *mp, void *opaque_arg,
1115 void *m, unsigned i);
1118 * Allocate a new mbuf (type is ctrl) from mempool *mp*.
1120 * This new mbuf is initialized with data pointing to the beginning of
1121 * buffer, and with a length of zero.
1124 * The mempool from which the mbuf is allocated.
1126 * - The pointer to the new mbuf on success.
1127 * - NULL if allocation failed.
1129 #define rte_ctrlmbuf_alloc(mp) rte_pktmbuf_alloc(mp)
1132 * Free a control mbuf back into its original mempool.
1135 * The control mbuf to be freed.
1137 #define rte_ctrlmbuf_free(m) rte_pktmbuf_free(m)
1140 * A macro that returns the pointer to the carried data.
1142 * The value that can be read or assigned.
1147 #define rte_ctrlmbuf_data(m) ((char *)((m)->buf_addr) + (m)->data_off)
1150 * A macro that returns the length of the carried data.
1152 * The value that can be read or assigned.
1157 #define rte_ctrlmbuf_len(m) rte_pktmbuf_data_len(m)
1160 * Tests if an mbuf is a control mbuf
1163 * The mbuf to be tested
1165 * - True (1) if the mbuf is a control mbuf
1166 * - False(0) otherwise
1169 rte_is_ctrlmbuf(struct rte_mbuf *m)
1171 return !!(m->ol_flags & CTRL_MBUF_FLAG);
1174 /* Operations on pkt mbuf */
1177 * The packet mbuf constructor.
1179 * This function initializes some fields in the mbuf structure that are
1180 * not modified by the user once created (origin pool, buffer start
1181 * address, and so on). This function is given as a callback function to
1182 * rte_mempool_create() at pool creation time.
1185 * The mempool from which mbufs originate.
1187 * A pointer that can be used by the user to retrieve useful information
1188 * for mbuf initialization. This pointer comes from the ``init_arg``
1189 * parameter of rte_mempool_create().
1191 * The mbuf to initialize.
1193 * The index of the mbuf in the pool table.
1195 void rte_pktmbuf_init(struct rte_mempool *mp, void *opaque_arg,
1196 void *m, unsigned i);
1200 * A packet mbuf pool constructor.
1202 * This function initializes the mempool private data in the case of a
1203 * pktmbuf pool. This private data is needed by the driver. The
1204 * function is given as a callback function to rte_mempool_create() at
1205 * pool creation. It can be extended by the user, for example, to
1206 * provide another packet size.
1209 * The mempool from which mbufs originate.
1211 * A pointer that can be used by the user to retrieve useful information
1212 * for mbuf initialization. This pointer comes from the ``init_arg``
1213 * parameter of rte_mempool_create().
1215 void rte_pktmbuf_pool_init(struct rte_mempool *mp, void *opaque_arg);
1218 * Create a mbuf pool.
1220 * This function creates and initializes a packet mbuf pool. It is
1221 * a wrapper to rte_mempool_create() with the proper packet constructor
1222 * and mempool constructor.
1225 * The name of the mbuf pool.
1227 * The number of elements in the mbuf pool. The optimum size (in terms
1228 * of memory usage) for a mempool is when n is a power of two minus one:
1231 * Size of the per-core object cache. See rte_mempool_create() for
1234 * Size of application private are between the rte_mbuf structure
1235 * and the data buffer.
1236 * @param data_room_size
1237 * Size of data buffer in each mbuf, including RTE_PKTMBUF_HEADROOM.
1239 * The socket identifier where the memory should be allocated. The
1240 * value can be *SOCKET_ID_ANY* if there is no NUMA constraint for the
1243 * The pointer to the new allocated mempool, on success. NULL on error
1244 * with rte_errno set appropriately. Possible rte_errno values include:
1245 * - E_RTE_NO_CONFIG - function could not get pointer to rte_config structure
1246 * - E_RTE_SECONDARY - function was called from a secondary process instance
1247 * - EINVAL - cache size provided is too large
1248 * - ENOSPC - the maximum number of memzones has already been allocated
1249 * - EEXIST - a memzone with the same name already exists
1250 * - ENOMEM - no appropriate memory area found in which to create memzone
1252 struct rte_mempool *
1253 rte_pktmbuf_pool_create(const char *name, unsigned n,
1254 unsigned cache_size, uint16_t priv_size, uint16_t data_room_size,
1258 * Get the data room size of mbufs stored in a pktmbuf_pool
1260 * The data room size is the amount of data that can be stored in a
1261 * mbuf including the headroom (RTE_PKTMBUF_HEADROOM).
1264 * The packet mbuf pool.
1266 * The data room size of mbufs stored in this mempool.
1268 static inline uint16_t
1269 rte_pktmbuf_data_room_size(struct rte_mempool *mp)
1271 struct rte_pktmbuf_pool_private *mbp_priv;
1273 mbp_priv = (struct rte_pktmbuf_pool_private *)rte_mempool_get_priv(mp);
1274 return mbp_priv->mbuf_data_room_size;
1278 * Get the application private size of mbufs stored in a pktmbuf_pool
1280 * The private size of mbuf is a zone located between the rte_mbuf
1281 * structure and the data buffer where an application can store data
1282 * associated to a packet.
1285 * The packet mbuf pool.
1287 * The private size of mbufs stored in this mempool.
1289 static inline uint16_t
1290 rte_pktmbuf_priv_size(struct rte_mempool *mp)
1292 struct rte_pktmbuf_pool_private *mbp_priv;
1294 mbp_priv = (struct rte_pktmbuf_pool_private *)rte_mempool_get_priv(mp);
1295 return mbp_priv->mbuf_priv_size;
1299 * Reset the fields of a packet mbuf to their default values.
1301 * The given mbuf must have only one segment.
1304 * The packet mbuf to be resetted.
1306 static inline void rte_pktmbuf_reset(struct rte_mbuf *m)
1312 m->vlan_tci_outer = 0;
1318 m->data_off = (RTE_PKTMBUF_HEADROOM <= m->buf_len) ?
1319 RTE_PKTMBUF_HEADROOM : m->buf_len;
1322 __rte_mbuf_sanity_check(m, 1);
1326 * Allocate a new mbuf from a mempool.
1328 * This new mbuf contains one segment, which has a length of 0. The pointer
1329 * to data is initialized to have some bytes of headroom in the buffer
1330 * (if buffer size allows).
1333 * The mempool from which the mbuf is allocated.
1335 * - The pointer to the new mbuf on success.
1336 * - NULL if allocation failed.
1338 static inline struct rte_mbuf *rte_pktmbuf_alloc(struct rte_mempool *mp)
1341 if ((m = __rte_mbuf_raw_alloc(mp)) != NULL)
1342 rte_pktmbuf_reset(m);
1347 * Attach packet mbuf to another packet mbuf.
1349 * After attachment we refer the mbuf we attached as 'indirect',
1350 * while mbuf we attached to as 'direct'.
1351 * Right now, not supported:
1352 * - attachment for already indirect mbuf (e.g. - mi has to be direct).
1353 * - mbuf we trying to attach (mi) is used by someone else
1354 * e.g. it's reference counter is greater then 1.
1357 * The indirect packet mbuf.
1359 * The packet mbuf we're attaching to.
1361 static inline void rte_pktmbuf_attach(struct rte_mbuf *mi, struct rte_mbuf *m)
1363 struct rte_mbuf *md;
1365 RTE_MBUF_ASSERT(RTE_MBUF_DIRECT(mi) &&
1366 rte_mbuf_refcnt_read(mi) == 1);
1368 /* if m is not direct, get the mbuf that embeds the data */
1369 if (RTE_MBUF_DIRECT(m))
1372 md = rte_mbuf_from_indirect(m);
1374 rte_mbuf_refcnt_update(md, 1);
1375 mi->priv_size = m->priv_size;
1376 mi->buf_physaddr = m->buf_physaddr;
1377 mi->buf_addr = m->buf_addr;
1378 mi->buf_len = m->buf_len;
1381 mi->data_off = m->data_off;
1382 mi->data_len = m->data_len;
1384 mi->vlan_tci = m->vlan_tci;
1385 mi->vlan_tci_outer = m->vlan_tci_outer;
1386 mi->tx_offload = m->tx_offload;
1390 mi->pkt_len = mi->data_len;
1392 mi->ol_flags = m->ol_flags | IND_ATTACHED_MBUF;
1393 mi->packet_type = m->packet_type;
1395 __rte_mbuf_sanity_check(mi, 1);
1396 __rte_mbuf_sanity_check(m, 0);
1400 * Detach an indirect packet mbuf.
1402 * - restore original mbuf address and length values.
1403 * - reset pktmbuf data and data_len to their default values.
1404 * All other fields of the given packet mbuf will be left intact.
1407 * The indirect attached packet mbuf.
1409 static inline void rte_pktmbuf_detach(struct rte_mbuf *m)
1411 struct rte_mempool *mp = m->pool;
1412 uint32_t mbuf_size, buf_len, priv_size;
1414 priv_size = rte_pktmbuf_priv_size(mp);
1415 mbuf_size = sizeof(struct rte_mbuf) + priv_size;
1416 buf_len = rte_pktmbuf_data_room_size(mp);
1418 m->priv_size = priv_size;
1419 m->buf_addr = (char *)m + mbuf_size;
1420 m->buf_physaddr = rte_mempool_virt2phy(mp, m) + mbuf_size;
1421 m->buf_len = (uint16_t)buf_len;
1422 m->data_off = RTE_MIN(RTE_PKTMBUF_HEADROOM, (uint16_t)m->buf_len);
1427 static inline struct rte_mbuf* __attribute__((always_inline))
1428 __rte_pktmbuf_prefree_seg(struct rte_mbuf *m)
1430 __rte_mbuf_sanity_check(m, 0);
1432 if (likely(rte_mbuf_refcnt_update(m, -1) == 0)) {
1434 /* if this is an indirect mbuf, then
1436 * - free attached mbuf segment
1438 if (RTE_MBUF_INDIRECT(m)) {
1439 struct rte_mbuf *md = rte_mbuf_from_indirect(m);
1440 rte_pktmbuf_detach(m);
1441 if (rte_mbuf_refcnt_update(md, -1) == 0)
1442 __rte_mbuf_raw_free(md);
1450 * Free a segment of a packet mbuf into its original mempool.
1452 * Free an mbuf, without parsing other segments in case of chained
1456 * The packet mbuf segment to be freed.
1458 static inline void __attribute__((always_inline))
1459 rte_pktmbuf_free_seg(struct rte_mbuf *m)
1461 if (likely(NULL != (m = __rte_pktmbuf_prefree_seg(m)))) {
1463 __rte_mbuf_raw_free(m);
1468 * Free a packet mbuf back into its original mempool.
1470 * Free an mbuf, and all its segments in case of chained buffers. Each
1471 * segment is added back into its original mempool.
1474 * The packet mbuf to be freed.
1476 static inline void rte_pktmbuf_free(struct rte_mbuf *m)
1478 struct rte_mbuf *m_next;
1480 __rte_mbuf_sanity_check(m, 1);
1484 rte_pktmbuf_free_seg(m);
1490 * Creates a "clone" of the given packet mbuf.
1492 * Walks through all segments of the given packet mbuf, and for each of them:
1493 * - Creates a new packet mbuf from the given pool.
1494 * - Attaches newly created mbuf to the segment.
1495 * Then updates pkt_len and nb_segs of the "clone" packet mbuf to match values
1496 * from the original packet mbuf.
1499 * The packet mbuf to be cloned.
1501 * The mempool from which the "clone" mbufs are allocated.
1503 * - The pointer to the new "clone" mbuf on success.
1504 * - NULL if allocation fails.
1506 static inline struct rte_mbuf *rte_pktmbuf_clone(struct rte_mbuf *md,
1507 struct rte_mempool *mp)
1509 struct rte_mbuf *mc, *mi, **prev;
1513 if (unlikely ((mc = rte_pktmbuf_alloc(mp)) == NULL))
1518 pktlen = md->pkt_len;
1523 rte_pktmbuf_attach(mi, md);
1526 } while ((md = md->next) != NULL &&
1527 (mi = rte_pktmbuf_alloc(mp)) != NULL);
1531 mc->pkt_len = pktlen;
1533 /* Allocation of new indirect segment failed */
1534 if (unlikely (mi == NULL)) {
1535 rte_pktmbuf_free(mc);
1539 __rte_mbuf_sanity_check(mc, 1);
1544 * Adds given value to the refcnt of all packet mbuf segments.
1546 * Walks through all segments of given packet mbuf and for each of them
1547 * invokes rte_mbuf_refcnt_update().
1550 * The packet mbuf whose refcnt to be updated.
1552 * The value to add to the mbuf's segments refcnt.
1554 static inline void rte_pktmbuf_refcnt_update(struct rte_mbuf *m, int16_t v)
1556 __rte_mbuf_sanity_check(m, 1);
1559 rte_mbuf_refcnt_update(m, v);
1560 } while ((m = m->next) != NULL);
1564 * Get the headroom in a packet mbuf.
1569 * The length of the headroom.
1571 static inline uint16_t rte_pktmbuf_headroom(const struct rte_mbuf *m)
1573 __rte_mbuf_sanity_check(m, 1);
1578 * Get the tailroom of a packet mbuf.
1583 * The length of the tailroom.
1585 static inline uint16_t rte_pktmbuf_tailroom(const struct rte_mbuf *m)
1587 __rte_mbuf_sanity_check(m, 1);
1588 return (uint16_t)(m->buf_len - rte_pktmbuf_headroom(m) -
1593 * Get the last segment of the packet.
1598 * The last segment of the given mbuf.
1600 static inline struct rte_mbuf *rte_pktmbuf_lastseg(struct rte_mbuf *m)
1602 struct rte_mbuf *m2 = (struct rte_mbuf *)m;
1604 __rte_mbuf_sanity_check(m, 1);
1605 while (m2->next != NULL)
1611 * A macro that points to an offset into the data in the mbuf.
1613 * The returned pointer is cast to type t. Before using this
1614 * function, the user must ensure that the first segment is large
1615 * enough to accommodate its data.
1620 * The offset into the mbuf data.
1622 * The type to cast the result into.
1624 #define rte_pktmbuf_mtod_offset(m, t, o) \
1625 ((t)((char *)(m)->buf_addr + (m)->data_off + (o)))
1628 * A macro that points to the start of the data in the mbuf.
1630 * The returned pointer is cast to type t. Before using this
1631 * function, the user must ensure that the first segment is large
1632 * enough to accommodate its data.
1637 * The type to cast the result into.
1639 #define rte_pktmbuf_mtod(m, t) rte_pktmbuf_mtod_offset(m, t, 0)
1642 * A macro that returns the length of the packet.
1644 * The value can be read or assigned.
1649 #define rte_pktmbuf_pkt_len(m) ((m)->pkt_len)
1652 * A macro that returns the length of the segment.
1654 * The value can be read or assigned.
1659 #define rte_pktmbuf_data_len(m) ((m)->data_len)
1662 * Prepend len bytes to an mbuf data area.
1664 * Returns a pointer to the new
1665 * data start address. If there is not enough headroom in the first
1666 * segment, the function will return NULL, without modifying the mbuf.
1671 * The amount of data to prepend (in bytes).
1673 * A pointer to the start of the newly prepended data, or
1674 * NULL if there is not enough headroom space in the first segment
1676 static inline char *rte_pktmbuf_prepend(struct rte_mbuf *m,
1679 __rte_mbuf_sanity_check(m, 1);
1681 if (unlikely(len > rte_pktmbuf_headroom(m)))
1685 m->data_len = (uint16_t)(m->data_len + len);
1686 m->pkt_len = (m->pkt_len + len);
1688 return (char *)m->buf_addr + m->data_off;
1692 * Append len bytes to an mbuf.
1694 * Append len bytes to an mbuf and return a pointer to the start address
1695 * of the added data. If there is not enough tailroom in the last
1696 * segment, the function will return NULL, without modifying the mbuf.
1701 * The amount of data to append (in bytes).
1703 * A pointer to the start of the newly appended data, or
1704 * NULL if there is not enough tailroom space in the last segment
1706 static inline char *rte_pktmbuf_append(struct rte_mbuf *m, uint16_t len)
1709 struct rte_mbuf *m_last;
1711 __rte_mbuf_sanity_check(m, 1);
1713 m_last = rte_pktmbuf_lastseg(m);
1714 if (unlikely(len > rte_pktmbuf_tailroom(m_last)))
1717 tail = (char *)m_last->buf_addr + m_last->data_off + m_last->data_len;
1718 m_last->data_len = (uint16_t)(m_last->data_len + len);
1719 m->pkt_len = (m->pkt_len + len);
1720 return (char*) tail;
1724 * Remove len bytes at the beginning of an mbuf.
1726 * Returns a pointer to the start address of the new data area. If the
1727 * length is greater than the length of the first segment, then the
1728 * function will fail and return NULL, without modifying the mbuf.
1733 * The amount of data to remove (in bytes).
1735 * A pointer to the new start of the data.
1737 static inline char *rte_pktmbuf_adj(struct rte_mbuf *m, uint16_t len)
1739 __rte_mbuf_sanity_check(m, 1);
1741 if (unlikely(len > m->data_len))
1744 m->data_len = (uint16_t)(m->data_len - len);
1746 m->pkt_len = (m->pkt_len - len);
1747 return (char *)m->buf_addr + m->data_off;
1751 * Remove len bytes of data at the end of the mbuf.
1753 * If the length is greater than the length of the last segment, the
1754 * function will fail and return -1 without modifying the mbuf.
1759 * The amount of data to remove (in bytes).
1764 static inline int rte_pktmbuf_trim(struct rte_mbuf *m, uint16_t len)
1766 struct rte_mbuf *m_last;
1768 __rte_mbuf_sanity_check(m, 1);
1770 m_last = rte_pktmbuf_lastseg(m);
1771 if (unlikely(len > m_last->data_len))
1774 m_last->data_len = (uint16_t)(m_last->data_len - len);
1775 m->pkt_len = (m->pkt_len - len);
1780 * Test if mbuf data is contiguous.
1785 * - 1, if all data is contiguous (one segment).
1786 * - 0, if there is several segments.
1788 static inline int rte_pktmbuf_is_contiguous(const struct rte_mbuf *m)
1790 __rte_mbuf_sanity_check(m, 1);
1791 return !!(m->nb_segs == 1);
1795 * Dump an mbuf structure to the console.
1797 * Dump all fields for the given packet mbuf and all its associated
1798 * segments (in the case of a chained buffer).
1801 * A pointer to a file for output
1805 * If dump_len != 0, also dump the "dump_len" first data bytes of
1808 void rte_pktmbuf_dump(FILE *f, const struct rte_mbuf *m, unsigned dump_len);
1814 #endif /* _RTE_MBUF_H_ */