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 3 bits are reserved for generic mbuf flags
79 * - TX flags therefore start at bit position 60 (i.e. 63-3), and new flags get
80 * added to the right of the previously defined flags i.e. they should count
81 * downwards, not upwards.
83 * Keep these flags synchronized with rte_get_rx_ol_flag_name() and
84 * rte_get_tx_ol_flag_name().
86 #define PKT_RX_VLAN_PKT (1ULL << 0) /**< RX packet is a 802.1q VLAN packet. */
87 #define PKT_RX_RSS_HASH (1ULL << 1) /**< RX packet with RSS hash result. */
88 #define PKT_RX_FDIR (1ULL << 2) /**< RX packet with FDIR match indicate. */
89 #define PKT_RX_L4_CKSUM_BAD (1ULL << 3) /**< L4 cksum of RX pkt. is not OK. */
90 #define PKT_RX_IP_CKSUM_BAD (1ULL << 4) /**< IP cksum of RX pkt. is not OK. */
91 #define PKT_RX_EIP_CKSUM_BAD (0ULL << 0) /**< External IP header checksum error. */
92 #define PKT_RX_OVERSIZE (0ULL << 0) /**< Num of desc of an RX pkt oversize. */
93 #define PKT_RX_HBUF_OVERFLOW (0ULL << 0) /**< Header buffer overflow. */
94 #define PKT_RX_RECIP_ERR (0ULL << 0) /**< Hardware processing error. */
95 #define PKT_RX_MAC_ERR (0ULL << 0) /**< MAC error. */
97 #define PKT_RX_IPV4_HDR (1ULL << 5) /**< RX packet with IPv4 header. */
98 #define PKT_RX_IPV4_HDR_EXT (1ULL << 6) /**< RX packet with extended IPv4 header. */
99 #define PKT_RX_IPV6_HDR (1ULL << 7) /**< RX packet with IPv6 header. */
100 #define PKT_RX_IPV6_HDR_EXT (1ULL << 8) /**< RX packet with extended IPv6 header. */
101 #endif /* RTE_NEXT_ABI */
102 #define PKT_RX_IEEE1588_PTP (1ULL << 9) /**< RX IEEE1588 L2 Ethernet PT Packet. */
103 #define PKT_RX_IEEE1588_TMST (1ULL << 10) /**< RX IEEE1588 L2/L4 timestamped packet.*/
105 #define PKT_RX_TUNNEL_IPV4_HDR (1ULL << 11) /**< RX tunnel packet with IPv4 header.*/
106 #define PKT_RX_TUNNEL_IPV6_HDR (1ULL << 12) /**< RX tunnel packet with IPv6 header. */
107 #endif /* RTE_NEXT_ABI */
108 #define PKT_RX_FDIR_ID (1ULL << 13) /**< FD id reported if FDIR match. */
109 #define PKT_RX_FDIR_FLX (1ULL << 14) /**< Flexible bytes reported if FDIR match. */
110 #define PKT_RX_QINQ_PKT (1ULL << 15) /**< RX packet with double VLAN stripped. */
111 /* add new RX flags here */
113 /* add new TX flags here */
116 * Second VLAN insertion (QinQ) flag.
118 #define PKT_TX_QINQ_PKT (1ULL << 49) /**< TX packet with double VLAN inserted. */
121 * TCP segmentation offload. To enable this offload feature for a
122 * packet to be transmitted on hardware supporting TSO:
123 * - set the PKT_TX_TCP_SEG flag in mbuf->ol_flags (this flag implies
125 * - set the flag PKT_TX_IPV4 or PKT_TX_IPV6
126 * - if it's IPv4, set the PKT_TX_IP_CKSUM flag and write the IP checksum
128 * - fill the mbuf offload information: l2_len, l3_len, l4_len, tso_segsz
129 * - calculate the pseudo header checksum without taking ip_len in account,
130 * and set it in the TCP header. Refer to rte_ipv4_phdr_cksum() and
131 * rte_ipv6_phdr_cksum() that can be used as helpers.
133 #define PKT_TX_TCP_SEG (1ULL << 50)
135 #define PKT_TX_IEEE1588_TMST (1ULL << 51) /**< TX IEEE1588 packet to timestamp. */
138 * Bits 52+53 used for L4 packet type with checksum enabled: 00: Reserved,
139 * 01: TCP checksum, 10: SCTP checksum, 11: UDP checksum. To use hardware
140 * L4 checksum offload, the user needs to:
141 * - fill l2_len and l3_len in mbuf
142 * - set the flags PKT_TX_TCP_CKSUM, PKT_TX_SCTP_CKSUM or PKT_TX_UDP_CKSUM
143 * - set the flag PKT_TX_IPV4 or PKT_TX_IPV6
144 * - calculate the pseudo header checksum and set it in the L4 header (only
145 * for TCP or UDP). See rte_ipv4_phdr_cksum() and rte_ipv6_phdr_cksum().
146 * For SCTP, set the crc field to 0.
148 #define PKT_TX_L4_NO_CKSUM (0ULL << 52) /**< Disable L4 cksum of TX pkt. */
149 #define PKT_TX_TCP_CKSUM (1ULL << 52) /**< TCP cksum of TX pkt. computed by NIC. */
150 #define PKT_TX_SCTP_CKSUM (2ULL << 52) /**< SCTP cksum of TX pkt. computed by NIC. */
151 #define PKT_TX_UDP_CKSUM (3ULL << 52) /**< UDP cksum of TX pkt. computed by NIC. */
152 #define PKT_TX_L4_MASK (3ULL << 52) /**< Mask for L4 cksum offload request. */
155 * Offload the IP checksum in the hardware. The flag PKT_TX_IPV4 should
156 * also be set by the application, although a PMD will only check
158 * - set the IP checksum field in the packet to 0
159 * - fill the mbuf offload information: l2_len, l3_len
161 #define PKT_TX_IP_CKSUM (1ULL << 54)
164 * Packet is IPv4. This flag must be set when using any offload feature
165 * (TSO, L3 or L4 checksum) to tell the NIC that the packet is an IPv4
166 * packet. If the packet is a tunneled packet, this flag is related to
169 #define PKT_TX_IPV4 (1ULL << 55)
172 * Packet is IPv6. This flag must be set when using an offload feature
173 * (TSO or L4 checksum) to tell the NIC that the packet is an IPv6
174 * packet. If the packet is a tunneled packet, this flag is related to
177 #define PKT_TX_IPV6 (1ULL << 56)
179 #define PKT_TX_VLAN_PKT (1ULL << 57) /**< TX packet is a 802.1q VLAN packet. */
182 * Offload the IP checksum of an external header in the hardware. The
183 * flag PKT_TX_OUTER_IPV4 should also be set by the application, alto ugh
184 * a PMD will only check PKT_TX_IP_CKSUM. The IP checksum field in the
185 * packet must be set to 0.
186 * - set the outer IP checksum field in the packet to 0
187 * - fill the mbuf offload information: outer_l2_len, outer_l3_len
189 #define PKT_TX_OUTER_IP_CKSUM (1ULL << 58)
192 * Packet outer header is IPv4. This flag must be set when using any
193 * outer offload feature (L3 or L4 checksum) to tell the NIC that the
194 * outer header of the tunneled packet is an IPv4 packet.
196 #define PKT_TX_OUTER_IPV4 (1ULL << 59)
199 * Packet outer header is IPv6. This flag must be set when using any
200 * outer offload feature (L4 checksum) to tell the NIC that the outer
201 * header of the tunneled packet is an IPv6 packet.
203 #define PKT_TX_OUTER_IPV6 (1ULL << 60)
205 #define __RESERVED (1ULL << 61) /**< reserved for future mbuf use */
207 #define IND_ATTACHED_MBUF (1ULL << 62) /**< Indirect attached mbuf */
209 /* Use final bit of flags to indicate a control mbuf */
210 #define CTRL_MBUF_FLAG (1ULL << 63) /**< Mbuf contains control data */
214 * 32 bits are divided into several fields to mark packet types. Note that
215 * each field is indexical.
216 * - Bit 3:0 is for L2 types.
217 * - Bit 7:4 is for L3 or outer L3 (for tunneling case) types.
218 * - Bit 11:8 is for L4 or outer L4 (for tunneling case) types.
219 * - Bit 15:12 is for tunnel types.
220 * - Bit 19:16 is for inner L2 types.
221 * - Bit 23:20 is for inner L3 types.
222 * - Bit 27:24 is for inner L4 types.
223 * - Bit 31:28 is reserved.
225 * To be compatible with Vector PMD, RTE_PTYPE_L3_IPV4, RTE_PTYPE_L3_IPV4_EXT,
226 * RTE_PTYPE_L3_IPV6, RTE_PTYPE_L3_IPV6_EXT, RTE_PTYPE_L4_TCP, RTE_PTYPE_L4_UDP
227 * and RTE_PTYPE_L4_SCTP should be kept as below in a contiguous 7 bits.
229 * Note that L3 types values are selected for checking IPV4/IPV6 header from
230 * performance point of view. Reading annotations of RTE_ETH_IS_IPV4_HDR and
231 * RTE_ETH_IS_IPV6_HDR is needed for any future changes of L3 type values.
233 * Note that the packet types of the same packet recognized by different
234 * hardware may be different, as different hardware may have different
235 * capability of packet type recognition.
238 * <'ether type'=0x0800
239 * | 'version'=4, 'protocol'=0x29
240 * | 'version'=6, 'next header'=0x3A
242 * will be recognized on i40e hardware as packet type combination of,
243 * RTE_PTYPE_L2_ETHER |
244 * RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
245 * RTE_PTYPE_TUNNEL_IP |
246 * RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
247 * RTE_PTYPE_INNER_L4_ICMP.
249 * <'ether type'=0x86DD
250 * | 'version'=6, 'next header'=0x2F
252 * | 'version'=6, 'next header'=0x11
254 * will be recognized on i40e hardware as packet type combination of,
255 * RTE_PTYPE_L2_ETHER |
256 * RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
257 * RTE_PTYPE_TUNNEL_GRENAT |
258 * RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
259 * RTE_PTYPE_INNER_L4_UDP.
261 #define RTE_PTYPE_UNKNOWN 0x00000000
263 * Ethernet packet type.
264 * It is used for outer packet for tunneling cases.
267 * <'ether type'=[0x0800|0x86DD]>
269 #define RTE_PTYPE_L2_ETHER 0x00000001
271 * Ethernet packet type for time sync.
274 * <'ether type'=0x88F7>
276 #define RTE_PTYPE_L2_ETHER_TIMESYNC 0x00000002
278 * ARP (Address Resolution Protocol) packet type.
281 * <'ether type'=0x0806>
283 #define RTE_PTYPE_L2_ETHER_ARP 0x00000003
285 * LLDP (Link Layer Discovery Protocol) packet type.
288 * <'ether type'=0x88CC>
290 #define RTE_PTYPE_L2_ETHER_LLDP 0x00000004
292 * Mask of layer 2 packet types.
293 * It is used for outer packet for tunneling cases.
295 #define RTE_PTYPE_L2_MASK 0x0000000f
297 * IP (Internet Protocol) version 4 packet type.
298 * It is used for outer packet for tunneling cases, and does not contain any
302 * <'ether type'=0x0800
303 * | 'version'=4, 'ihl'=5>
305 #define RTE_PTYPE_L3_IPV4 0x00000010
307 * IP (Internet Protocol) version 4 packet type.
308 * It is used for outer packet for tunneling cases, and contains header
312 * <'ether type'=0x0800
313 * | 'version'=4, 'ihl'=[6-15], 'options'>
315 #define RTE_PTYPE_L3_IPV4_EXT 0x00000030
317 * IP (Internet Protocol) version 6 packet type.
318 * It is used for outer packet for tunneling cases, and does not contain any
322 * <'ether type'=0x86DD
323 * | 'version'=6, 'next header'=0x3B>
325 #define RTE_PTYPE_L3_IPV6 0x00000040
327 * IP (Internet Protocol) version 4 packet type.
328 * It is used for outer packet for tunneling cases, and may or maynot contain
332 * <'ether type'=0x0800
333 * | 'version'=4, 'ihl'=[5-15], <'options'>>
335 #define RTE_PTYPE_L3_IPV4_EXT_UNKNOWN 0x00000090
337 * IP (Internet Protocol) version 6 packet type.
338 * It is used for outer packet for tunneling cases, and contains extension
342 * <'ether type'=0x86DD
343 * | 'version'=6, 'next header'=[0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
344 * 'extension headers'>
346 #define RTE_PTYPE_L3_IPV6_EXT 0x000000c0
348 * IP (Internet Protocol) version 6 packet type.
349 * It is used for outer packet for tunneling cases, and may or maynot contain
353 * <'ether type'=0x86DD
354 * | 'version'=6, 'next header'=[0x3B|0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
355 * <'extension headers'>>
357 #define RTE_PTYPE_L3_IPV6_EXT_UNKNOWN 0x000000e0
359 * Mask of layer 3 packet types.
360 * It is used for outer packet for tunneling cases.
362 #define RTE_PTYPE_L3_MASK 0x000000f0
364 * TCP (Transmission Control Protocol) packet type.
365 * It is used for outer packet for tunneling cases.
368 * <'ether type'=0x0800
369 * | 'version'=4, 'protocol'=6, 'MF'=0>
371 * <'ether type'=0x86DD
372 * | 'version'=6, 'next header'=6>
374 #define RTE_PTYPE_L4_TCP 0x00000100
376 * UDP (User Datagram Protocol) packet type.
377 * It is used for outer packet for tunneling cases.
380 * <'ether type'=0x0800
381 * | 'version'=4, 'protocol'=17, 'MF'=0>
383 * <'ether type'=0x86DD
384 * | 'version'=6, 'next header'=17>
386 #define RTE_PTYPE_L4_UDP 0x00000200
388 * Fragmented IP (Internet Protocol) packet type.
389 * It is used for outer packet for tunneling cases.
391 * It refers to those packets of any IP types, which can be recognized as
392 * fragmented. A fragmented packet cannot be recognized as any other L4 types
393 * (RTE_PTYPE_L4_TCP, RTE_PTYPE_L4_UDP, RTE_PTYPE_L4_SCTP, RTE_PTYPE_L4_ICMP,
394 * RTE_PTYPE_L4_NONFRAG).
397 * <'ether type'=0x0800
398 * | 'version'=4, 'MF'=1>
400 * <'ether type'=0x86DD
401 * | 'version'=6, 'next header'=44>
403 #define RTE_PTYPE_L4_FRAG 0x00000300
405 * SCTP (Stream Control Transmission Protocol) packet type.
406 * It is used for outer packet for tunneling cases.
409 * <'ether type'=0x0800
410 * | 'version'=4, 'protocol'=132, 'MF'=0>
412 * <'ether type'=0x86DD
413 * | 'version'=6, 'next header'=132>
415 #define RTE_PTYPE_L4_SCTP 0x00000400
417 * ICMP (Internet Control Message Protocol) packet type.
418 * It is used for outer packet for tunneling cases.
421 * <'ether type'=0x0800
422 * | 'version'=4, 'protocol'=1, 'MF'=0>
424 * <'ether type'=0x86DD
425 * | 'version'=6, 'next header'=1>
427 #define RTE_PTYPE_L4_ICMP 0x00000500
429 * Non-fragmented IP (Internet Protocol) packet type.
430 * It is used for outer packet for tunneling cases.
432 * It refers to those packets of any IP types, while cannot be recognized as
433 * any of above L4 types (RTE_PTYPE_L4_TCP, RTE_PTYPE_L4_UDP,
434 * RTE_PTYPE_L4_FRAG, RTE_PTYPE_L4_SCTP, RTE_PTYPE_L4_ICMP).
437 * <'ether type'=0x0800
438 * | 'version'=4, 'protocol'!=[6|17|132|1], 'MF'=0>
440 * <'ether type'=0x86DD
441 * | 'version'=6, 'next header'!=[6|17|44|132|1]>
443 #define RTE_PTYPE_L4_NONFRAG 0x00000600
445 * Mask of layer 4 packet types.
446 * It is used for outer packet for tunneling cases.
448 #define RTE_PTYPE_L4_MASK 0x00000f00
450 * IP (Internet Protocol) in IP (Internet Protocol) tunneling packet type.
453 * <'ether type'=0x0800
454 * | 'version'=4, 'protocol'=[4|41]>
456 * <'ether type'=0x86DD
457 * | 'version'=6, 'next header'=[4|41]>
459 #define RTE_PTYPE_TUNNEL_IP 0x00001000
461 * GRE (Generic Routing Encapsulation) tunneling packet type.
464 * <'ether type'=0x0800
465 * | 'version'=4, 'protocol'=47>
467 * <'ether type'=0x86DD
468 * | 'version'=6, 'next header'=47>
470 #define RTE_PTYPE_TUNNEL_GRE 0x00002000
472 * VXLAN (Virtual eXtensible Local Area Network) tunneling packet type.
475 * <'ether type'=0x0800
476 * | 'version'=4, 'protocol'=17
477 * | 'destination port'=4798>
479 * <'ether type'=0x86DD
480 * | 'version'=6, 'next header'=17
481 * | 'destination port'=4798>
483 #define RTE_PTYPE_TUNNEL_VXLAN 0x00003000
485 * NVGRE (Network Virtualization using Generic Routing Encapsulation) tunneling
489 * <'ether type'=0x0800
490 * | 'version'=4, 'protocol'=47
491 * | 'protocol type'=0x6558>
493 * <'ether type'=0x86DD
494 * | 'version'=6, 'next header'=47
495 * | 'protocol type'=0x6558'>
497 #define RTE_PTYPE_TUNNEL_NVGRE 0x00004000
499 * GENEVE (Generic Network Virtualization Encapsulation) tunneling packet type.
502 * <'ether type'=0x0800
503 * | 'version'=4, 'protocol'=17
504 * | 'destination port'=6081>
506 * <'ether type'=0x86DD
507 * | 'version'=6, 'next header'=17
508 * | 'destination port'=6081>
510 #define RTE_PTYPE_TUNNEL_GENEVE 0x00005000
512 * Tunneling packet type of Teredo, VXLAN (Virtual eXtensible Local Area
513 * Network) or GRE (Generic Routing Encapsulation) could be recognized as this
514 * packet type, if they can not be recognized independently as of hardware
517 #define RTE_PTYPE_TUNNEL_GRENAT 0x00006000
519 * Mask of tunneling packet types.
521 #define RTE_PTYPE_TUNNEL_MASK 0x0000f000
523 * Ethernet packet type.
524 * It is used for inner packet type only.
526 * Packet format (inner only):
527 * <'ether type'=[0x800|0x86DD]>
529 #define RTE_PTYPE_INNER_L2_ETHER 0x00010000
531 * Ethernet packet type with VLAN (Virtual Local Area Network) tag.
533 * Packet format (inner only):
534 * <'ether type'=[0x800|0x86DD], vlan=[1-4095]>
536 #define RTE_PTYPE_INNER_L2_ETHER_VLAN 0x00020000
538 * Mask of inner layer 2 packet types.
540 #define RTE_PTYPE_INNER_L2_MASK 0x000f0000
542 * IP (Internet Protocol) version 4 packet type.
543 * It is used for inner packet only, and does not contain any header option.
545 * Packet format (inner only):
546 * <'ether type'=0x0800
547 * | 'version'=4, 'ihl'=5>
549 #define RTE_PTYPE_INNER_L3_IPV4 0x00100000
551 * IP (Internet Protocol) version 4 packet type.
552 * It is used for inner packet only, and contains header options.
554 * Packet format (inner only):
555 * <'ether type'=0x0800
556 * | 'version'=4, 'ihl'=[6-15], 'options'>
558 #define RTE_PTYPE_INNER_L3_IPV4_EXT 0x00200000
560 * IP (Internet Protocol) version 6 packet type.
561 * It is used for inner packet only, and does not contain any extension header.
563 * Packet format (inner only):
564 * <'ether type'=0x86DD
565 * | 'version'=6, 'next header'=0x3B>
567 #define RTE_PTYPE_INNER_L3_IPV6 0x00300000
569 * IP (Internet Protocol) version 4 packet type.
570 * It is used for inner packet only, and may or maynot contain header options.
572 * Packet format (inner only):
573 * <'ether type'=0x0800
574 * | 'version'=4, 'ihl'=[5-15], <'options'>>
576 #define RTE_PTYPE_INNER_L3_IPV4_EXT_UNKNOWN 0x00400000
578 * IP (Internet Protocol) version 6 packet type.
579 * It is used for inner packet only, and contains extension headers.
581 * Packet format (inner only):
582 * <'ether type'=0x86DD
583 * | 'version'=6, 'next header'=[0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
584 * 'extension headers'>
586 #define RTE_PTYPE_INNER_L3_IPV6_EXT 0x00500000
588 * IP (Internet Protocol) version 6 packet type.
589 * It is used for inner packet only, and may or maynot contain extension
592 * Packet format (inner only):
593 * <'ether type'=0x86DD
594 * | 'version'=6, 'next header'=[0x3B|0x0|0x2B|0x2C|0x32|0x33|0x3C|0x87],
595 * <'extension headers'>>
597 #define RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN 0x00600000
599 * Mask of inner layer 3 packet types.
601 #define RTE_PTYPE_INNER_INNER_L3_MASK 0x00f00000
603 * TCP (Transmission Control Protocol) packet type.
604 * It is used for inner packet only.
606 * Packet format (inner only):
607 * <'ether type'=0x0800
608 * | 'version'=4, 'protocol'=6, 'MF'=0>
610 * <'ether type'=0x86DD
611 * | 'version'=6, 'next header'=6>
613 #define RTE_PTYPE_INNER_L4_TCP 0x01000000
615 * UDP (User Datagram Protocol) packet type.
616 * It is used for inner packet only.
618 * Packet format (inner only):
619 * <'ether type'=0x0800
620 * | 'version'=4, 'protocol'=17, 'MF'=0>
622 * <'ether type'=0x86DD
623 * | 'version'=6, 'next header'=17>
625 #define RTE_PTYPE_INNER_L4_UDP 0x02000000
627 * Fragmented IP (Internet Protocol) packet type.
628 * It is used for inner packet only, and may or maynot have layer 4 packet.
630 * Packet format (inner only):
631 * <'ether type'=0x0800
632 * | 'version'=4, 'MF'=1>
634 * <'ether type'=0x86DD
635 * | 'version'=6, 'next header'=44>
637 #define RTE_PTYPE_INNER_L4_FRAG 0x03000000
639 * SCTP (Stream Control Transmission Protocol) packet type.
640 * It is used for inner packet only.
642 * Packet format (inner only):
643 * <'ether type'=0x0800
644 * | 'version'=4, 'protocol'=132, 'MF'=0>
646 * <'ether type'=0x86DD
647 * | 'version'=6, 'next header'=132>
649 #define RTE_PTYPE_INNER_L4_SCTP 0x04000000
651 * ICMP (Internet Control Message Protocol) packet type.
652 * It is used for inner packet only.
654 * Packet format (inner only):
655 * <'ether type'=0x0800
656 * | 'version'=4, 'protocol'=1, 'MF'=0>
658 * <'ether type'=0x86DD
659 * | 'version'=6, 'next header'=1>
661 #define RTE_PTYPE_INNER_L4_ICMP 0x05000000
663 * Non-fragmented IP (Internet Protocol) packet type.
664 * It is used for inner packet only, and may or maynot have other unknown layer
667 * Packet format (inner only):
668 * <'ether type'=0x0800
669 * | 'version'=4, 'protocol'!=[6|17|132|1], 'MF'=0>
671 * <'ether type'=0x86DD
672 * | 'version'=6, 'next header'!=[6|17|44|132|1]>
674 #define RTE_PTYPE_INNER_L4_NONFRAG 0x06000000
676 * Mask of inner layer 4 packet types.
678 #define RTE_PTYPE_INNER_L4_MASK 0x0f000000
681 * Check if the (outer) L3 header is IPv4. To avoid comparing IPv4 types one by
682 * one, bit 4 is selected to be used for IPv4 only. Then checking bit 4 can
683 * determin if it is an IPV4 packet.
685 #define RTE_ETH_IS_IPV4_HDR(ptype) ((ptype) & RTE_PTYPE_L3_IPV4)
688 * Check if the (outer) L3 header is IPv4. To avoid comparing IPv4 types one by
689 * one, bit 6 is selected to be used for IPv4 only. Then checking bit 6 can
690 * determin if it is an IPV4 packet.
692 #define RTE_ETH_IS_IPV6_HDR(ptype) ((ptype) & RTE_PTYPE_L3_IPV6)
694 /* Check if it is a tunneling packet */
695 #define RTE_ETH_IS_TUNNEL_PKT(ptype) ((ptype) & RTE_PTYPE_TUNNEL_MASK)
696 #endif /* RTE_NEXT_ABI */
699 * Get the name of a RX offload flag
702 * The mask describing the flag.
704 * The name of this flag, or NULL if it's not a valid RX flag.
706 const char *rte_get_rx_ol_flag_name(uint64_t mask);
709 * Get the name of a TX offload flag
712 * The mask describing the flag. Usually only one bit must be set.
713 * Several bits can be given if they belong to the same mask.
714 * Ex: PKT_TX_L4_MASK.
716 * The name of this flag, or NULL if it's not a valid TX flag.
718 const char *rte_get_tx_ol_flag_name(uint64_t mask);
721 * Some NICs need at least 2KB buffer to RX standard Ethernet frame without
722 * splitting it into multiple segments.
723 * So, for mbufs that planned to be involved into RX/TX, the recommended
724 * minimal buffer length is 2KB + RTE_PKTMBUF_HEADROOM.
726 #define RTE_MBUF_DEFAULT_DATAROOM 2048
727 #define RTE_MBUF_DEFAULT_BUF_SIZE \
728 (RTE_MBUF_DEFAULT_DATAROOM + RTE_PKTMBUF_HEADROOM)
730 /* define a set of marker types that can be used to refer to set points in the
732 typedef void *MARKER[0]; /**< generic marker for a point in a structure */
733 typedef uint8_t MARKER8[0]; /**< generic marker with 1B alignment */
734 typedef uint64_t MARKER64[0]; /**< marker that allows us to overwrite 8 bytes
735 * with a single assignment */
738 * The generic rte_mbuf, containing a packet mbuf.
743 void *buf_addr; /**< Virtual address of segment buffer. */
744 phys_addr_t buf_physaddr; /**< Physical address of segment buffer. */
746 uint16_t buf_len; /**< Length of segment buffer. */
748 /* next 6 bytes are initialised on RX descriptor rearm */
753 * 16-bit Reference counter.
754 * It should only be accessed using the following functions:
755 * rte_mbuf_refcnt_update(), rte_mbuf_refcnt_read(), and
756 * rte_mbuf_refcnt_set(). The functionality of these functions (atomic,
757 * or non-atomic) is controlled by the CONFIG_RTE_MBUF_REFCNT_ATOMIC
761 rte_atomic16_t refcnt_atomic; /**< Atomically accessed refcnt */
762 uint16_t refcnt; /**< Non-atomically accessed refcnt */
764 uint8_t nb_segs; /**< Number of segments. */
765 uint8_t port; /**< Input port. */
767 uint64_t ol_flags; /**< Offload features. */
769 /* remaining bytes are set on RX when pulling packet from descriptor */
770 MARKER rx_descriptor_fields1;
774 * The packet type, which is the combination of outer/inner L2, L3, L4
778 uint32_t packet_type; /**< L2/L3/L4 and tunnel information. */
780 uint32_t l2_type:4; /**< (Outer) L2 type. */
781 uint32_t l3_type:4; /**< (Outer) L3 type. */
782 uint32_t l4_type:4; /**< (Outer) L4 type. */
783 uint32_t tun_type:4; /**< Tunnel type. */
784 uint32_t inner_l2_type:4; /**< Inner L2 type. */
785 uint32_t inner_l3_type:4; /**< Inner L3 type. */
786 uint32_t inner_l4_type:4; /**< Inner L4 type. */
790 uint32_t pkt_len; /**< Total pkt len: sum of all segments. */
791 uint16_t data_len; /**< Amount of data in segment buffer. */
792 uint16_t vlan_tci; /**< VLAN Tag Control Identifier (CPU order) */
793 #else /* RTE_NEXT_ABI */
795 * The packet type, which is used to indicate ordinary packet and also
796 * tunneled packet format, i.e. each number is represented a type of
799 uint16_t packet_type;
801 uint16_t data_len; /**< Amount of data in segment buffer. */
802 uint32_t pkt_len; /**< Total pkt len: sum of all segments. */
803 uint16_t vlan_tci; /**< VLAN Tag Control Identifier (CPU order) */
804 uint16_t vlan_tci_outer; /**< Outer VLAN Tag Control Identifier (CPU order) */
805 #endif /* RTE_NEXT_ABI */
807 uint32_t rss; /**< RSS hash result if RSS enabled */
815 /**< Second 4 flexible bytes */
818 /**< First 4 flexible bytes or FD ID, dependent on
819 PKT_RX_FDIR_* flag in ol_flags. */
820 } fdir; /**< Filter identifier if FDIR enabled */
821 uint32_t sched; /**< Hierarchical scheduler */
822 uint32_t usr; /**< User defined tags. See rte_distributor_process() */
823 } hash; /**< hash information */
825 uint32_t seqn; /**< Sequence number. See also rte_reorder_insert() */
827 uint16_t vlan_tci_outer; /**< Outer VLAN Tag Control Identifier (CPU order) */
828 #endif /* RTE_NEXT_ABI */
830 /* second cache line - fields only used in slow path or on TX */
831 MARKER cacheline1 __rte_cache_aligned;
834 void *userdata; /**< Can be used for external metadata */
835 uint64_t udata64; /**< Allow 8-byte userdata on 32-bit */
838 struct rte_mempool *pool; /**< Pool from which mbuf was allocated. */
839 struct rte_mbuf *next; /**< Next segment of scattered packet. */
841 /* fields to support TX offloads */
843 uint64_t tx_offload; /**< combined for easy fetch */
845 uint64_t l2_len:7; /**< L2 (MAC) Header Length. */
846 uint64_t l3_len:9; /**< L3 (IP) Header Length. */
847 uint64_t l4_len:8; /**< L4 (TCP/UDP) Header Length. */
848 uint64_t tso_segsz:16; /**< TCP TSO segment size */
850 /* fields for TX offloading of tunnels */
851 uint64_t outer_l3_len:9; /**< Outer L3 (IP) Hdr Length. */
852 uint64_t outer_l2_len:7; /**< Outer L2 (MAC) Hdr Length. */
854 /* uint64_t unused:8; */
858 /** Size of the application private data. In case of an indirect
859 * mbuf, it stores the direct mbuf private data size. */
862 /** Timesync flags for use with IEEE1588. */
864 } __rte_cache_aligned;
866 static inline uint16_t rte_pktmbuf_priv_size(struct rte_mempool *mp);
869 * Return the mbuf owning the data buffer address of an indirect mbuf.
872 * The pointer to the indirect mbuf.
874 * The address of the direct mbuf corresponding to buffer_addr.
876 static inline struct rte_mbuf *
877 rte_mbuf_from_indirect(struct rte_mbuf *mi)
879 return RTE_PTR_SUB(mi->buf_addr, sizeof(*mi) + mi->priv_size);
883 * Return the buffer address embedded in the given mbuf.
886 * The pointer to the mbuf.
888 * The address of the data buffer owned by the mbuf.
891 rte_mbuf_to_baddr(struct rte_mbuf *md)
894 buffer_addr = (char *)md + sizeof(*md) + rte_pktmbuf_priv_size(md->pool);
899 * Returns TRUE if given mbuf is indirect, or FALSE otherwise.
901 #define RTE_MBUF_INDIRECT(mb) ((mb)->ol_flags & IND_ATTACHED_MBUF)
904 * Returns TRUE if given mbuf is direct, or FALSE otherwise.
906 #define RTE_MBUF_DIRECT(mb) (!RTE_MBUF_INDIRECT(mb))
909 * Private data in case of pktmbuf pool.
911 * A structure that contains some pktmbuf_pool-specific data that are
912 * appended after the mempool structure (in private data).
914 struct rte_pktmbuf_pool_private {
915 uint16_t mbuf_data_room_size; /**< Size of data space in each mbuf. */
916 uint16_t mbuf_priv_size; /**< Size of private area in each mbuf. */
919 #ifdef RTE_LIBRTE_MBUF_DEBUG
921 /** check mbuf type in debug mode */
922 #define __rte_mbuf_sanity_check(m, is_h) rte_mbuf_sanity_check(m, is_h)
924 /** check mbuf type in debug mode if mbuf pointer is not null */
925 #define __rte_mbuf_sanity_check_raw(m, is_h) do { \
927 rte_mbuf_sanity_check(m, is_h); \
930 /** MBUF asserts in debug mode */
931 #define RTE_MBUF_ASSERT(exp) \
933 rte_panic("line%d\tassert \"" #exp "\" failed\n", __LINE__); \
936 #else /* RTE_LIBRTE_MBUF_DEBUG */
938 /** check mbuf type in debug mode */
939 #define __rte_mbuf_sanity_check(m, is_h) do { } while (0)
941 /** check mbuf type in debug mode if mbuf pointer is not null */
942 #define __rte_mbuf_sanity_check_raw(m, is_h) do { } while (0)
944 /** MBUF asserts in debug mode */
945 #define RTE_MBUF_ASSERT(exp) do { } while (0)
947 #endif /* RTE_LIBRTE_MBUF_DEBUG */
949 #ifdef RTE_MBUF_REFCNT_ATOMIC
952 * Reads the value of an mbuf's refcnt.
956 * Reference count number.
958 static inline uint16_t
959 rte_mbuf_refcnt_read(const struct rte_mbuf *m)
961 return (uint16_t)(rte_atomic16_read(&m->refcnt_atomic));
965 * Sets an mbuf's refcnt to a defined value.
972 rte_mbuf_refcnt_set(struct rte_mbuf *m, uint16_t new_value)
974 rte_atomic16_set(&m->refcnt_atomic, new_value);
978 * Adds given value to an mbuf's refcnt and returns its new value.
982 * Value to add/subtract
986 static inline uint16_t
987 rte_mbuf_refcnt_update(struct rte_mbuf *m, int16_t value)
990 * The atomic_add is an expensive operation, so we don't want to
991 * call it in the case where we know we are the uniq holder of
992 * this mbuf (i.e. ref_cnt == 1). Otherwise, an atomic
993 * operation has to be used because concurrent accesses on the
994 * reference counter can occur.
996 if (likely(rte_mbuf_refcnt_read(m) == 1)) {
997 rte_mbuf_refcnt_set(m, 1 + value);
1001 return (uint16_t)(rte_atomic16_add_return(&m->refcnt_atomic, value));
1004 #else /* ! RTE_MBUF_REFCNT_ATOMIC */
1007 * Adds given value to an mbuf's refcnt and returns its new value.
1009 static inline uint16_t
1010 rte_mbuf_refcnt_update(struct rte_mbuf *m, int16_t value)
1012 m->refcnt = (uint16_t)(m->refcnt + value);
1017 * Reads the value of an mbuf's refcnt.
1019 static inline uint16_t
1020 rte_mbuf_refcnt_read(const struct rte_mbuf *m)
1026 * Sets an mbuf's refcnt to the defined value.
1029 rte_mbuf_refcnt_set(struct rte_mbuf *m, uint16_t new_value)
1031 m->refcnt = new_value;
1034 #endif /* RTE_MBUF_REFCNT_ATOMIC */
1036 /** Mbuf prefetch */
1037 #define RTE_MBUF_PREFETCH_TO_FREE(m) do { \
1044 * Sanity checks on an mbuf.
1046 * Check the consistency of the given mbuf. The function will cause a
1047 * panic if corruption is detected.
1050 * The mbuf to be checked.
1052 * True if the mbuf is a packet header, false if it is a sub-segment
1053 * of a packet (in this case, some fields like nb_segs are not checked)
1056 rte_mbuf_sanity_check(const struct rte_mbuf *m, int is_header);
1059 * @internal Allocate a new mbuf from mempool *mp*.
1060 * The use of that function is reserved for RTE internal needs.
1061 * Please use rte_pktmbuf_alloc().
1064 * The mempool from which mbuf is allocated.
1066 * - The pointer to the new mbuf on success.
1067 * - NULL if allocation failed.
1069 static inline struct rte_mbuf *__rte_mbuf_raw_alloc(struct rte_mempool *mp)
1073 if (rte_mempool_get(mp, &mb) < 0)
1075 m = (struct rte_mbuf *)mb;
1076 RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(m) == 0);
1077 rte_mbuf_refcnt_set(m, 1);
1082 * @internal Put mbuf back into its original mempool.
1083 * The use of that function is reserved for RTE internal needs.
1084 * Please use rte_pktmbuf_free().
1087 * The mbuf to be freed.
1089 static inline void __attribute__((always_inline))
1090 __rte_mbuf_raw_free(struct rte_mbuf *m)
1092 RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(m) == 0);
1093 rte_mempool_put(m->pool, m);
1096 /* Operations on ctrl mbuf */
1099 * The control mbuf constructor.
1101 * This function initializes some fields in an mbuf structure that are
1102 * not modified by the user once created (mbuf type, origin pool, buffer
1103 * start address, and so on). This function is given as a callback function
1104 * to rte_mempool_create() at pool creation time.
1107 * The mempool from which the mbuf is allocated.
1109 * A pointer that can be used by the user to retrieve useful information
1110 * for mbuf initialization. This pointer comes from the ``init_arg``
1111 * parameter of rte_mempool_create().
1113 * The mbuf to initialize.
1115 * The index of the mbuf in the pool table.
1117 void rte_ctrlmbuf_init(struct rte_mempool *mp, void *opaque_arg,
1118 void *m, unsigned i);
1121 * Allocate a new mbuf (type is ctrl) from mempool *mp*.
1123 * This new mbuf is initialized with data pointing to the beginning of
1124 * buffer, and with a length of zero.
1127 * The mempool from which the mbuf is allocated.
1129 * - The pointer to the new mbuf on success.
1130 * - NULL if allocation failed.
1132 #define rte_ctrlmbuf_alloc(mp) rte_pktmbuf_alloc(mp)
1135 * Free a control mbuf back into its original mempool.
1138 * The control mbuf to be freed.
1140 #define rte_ctrlmbuf_free(m) rte_pktmbuf_free(m)
1143 * A macro that returns the pointer to the carried data.
1145 * The value that can be read or assigned.
1150 #define rte_ctrlmbuf_data(m) ((char *)((m)->buf_addr) + (m)->data_off)
1153 * A macro that returns the length of the carried data.
1155 * The value that can be read or assigned.
1160 #define rte_ctrlmbuf_len(m) rte_pktmbuf_data_len(m)
1163 * Tests if an mbuf is a control mbuf
1166 * The mbuf to be tested
1168 * - True (1) if the mbuf is a control mbuf
1169 * - False(0) otherwise
1172 rte_is_ctrlmbuf(struct rte_mbuf *m)
1174 return !!(m->ol_flags & CTRL_MBUF_FLAG);
1177 /* Operations on pkt mbuf */
1180 * The packet mbuf constructor.
1182 * This function initializes some fields in the mbuf structure that are
1183 * not modified by the user once created (origin pool, buffer start
1184 * address, and so on). This function is given as a callback function to
1185 * rte_mempool_create() at pool creation time.
1188 * The mempool from which mbufs originate.
1190 * A pointer that can be used by the user to retrieve useful information
1191 * for mbuf initialization. This pointer comes from the ``init_arg``
1192 * parameter of rte_mempool_create().
1194 * The mbuf to initialize.
1196 * The index of the mbuf in the pool table.
1198 void rte_pktmbuf_init(struct rte_mempool *mp, void *opaque_arg,
1199 void *m, unsigned i);
1203 * A packet mbuf pool constructor.
1205 * This function initializes the mempool private data in the case of a
1206 * pktmbuf pool. This private data is needed by the driver. The
1207 * function is given as a callback function to rte_mempool_create() at
1208 * pool creation. It can be extended by the user, for example, to
1209 * provide another packet size.
1212 * The mempool from which mbufs originate.
1214 * A pointer that can be used by the user to retrieve useful information
1215 * for mbuf initialization. This pointer comes from the ``init_arg``
1216 * parameter of rte_mempool_create().
1218 void rte_pktmbuf_pool_init(struct rte_mempool *mp, void *opaque_arg);
1221 * Create a mbuf pool.
1223 * This function creates and initializes a packet mbuf pool. It is
1224 * a wrapper to rte_mempool_create() with the proper packet constructor
1225 * and mempool constructor.
1228 * The name of the mbuf pool.
1230 * The number of elements in the mbuf pool. The optimum size (in terms
1231 * of memory usage) for a mempool is when n is a power of two minus one:
1234 * Size of the per-core object cache. See rte_mempool_create() for
1237 * Size of application private are between the rte_mbuf structure
1238 * and the data buffer.
1239 * @param data_room_size
1240 * Size of data buffer in each mbuf, including RTE_PKTMBUF_HEADROOM.
1242 * The socket identifier where the memory should be allocated. The
1243 * value can be *SOCKET_ID_ANY* if there is no NUMA constraint for the
1246 * The pointer to the new allocated mempool, on success. NULL on error
1247 * with rte_errno set appropriately. Possible rte_errno values include:
1248 * - E_RTE_NO_CONFIG - function could not get pointer to rte_config structure
1249 * - E_RTE_SECONDARY - function was called from a secondary process instance
1250 * - EINVAL - cache size provided is too large
1251 * - ENOSPC - the maximum number of memzones has already been allocated
1252 * - EEXIST - a memzone with the same name already exists
1253 * - ENOMEM - no appropriate memory area found in which to create memzone
1255 struct rte_mempool *
1256 rte_pktmbuf_pool_create(const char *name, unsigned n,
1257 unsigned cache_size, uint16_t priv_size, uint16_t data_room_size,
1261 * Get the data room size of mbufs stored in a pktmbuf_pool
1263 * The data room size is the amount of data that can be stored in a
1264 * mbuf including the headroom (RTE_PKTMBUF_HEADROOM).
1267 * The packet mbuf pool.
1269 * The data room size of mbufs stored in this mempool.
1271 static inline uint16_t
1272 rte_pktmbuf_data_room_size(struct rte_mempool *mp)
1274 struct rte_pktmbuf_pool_private *mbp_priv;
1276 mbp_priv = (struct rte_pktmbuf_pool_private *)rte_mempool_get_priv(mp);
1277 return mbp_priv->mbuf_data_room_size;
1281 * Get the application private size of mbufs stored in a pktmbuf_pool
1283 * The private size of mbuf is a zone located between the rte_mbuf
1284 * structure and the data buffer where an application can store data
1285 * associated to a packet.
1288 * The packet mbuf pool.
1290 * The private size of mbufs stored in this mempool.
1292 static inline uint16_t
1293 rte_pktmbuf_priv_size(struct rte_mempool *mp)
1295 struct rte_pktmbuf_pool_private *mbp_priv;
1297 mbp_priv = (struct rte_pktmbuf_pool_private *)rte_mempool_get_priv(mp);
1298 return mbp_priv->mbuf_priv_size;
1302 * Reset the fields of a packet mbuf to their default values.
1304 * The given mbuf must have only one segment.
1307 * The packet mbuf to be resetted.
1309 static inline void rte_pktmbuf_reset(struct rte_mbuf *m)
1315 m->vlan_tci_outer = 0;
1321 m->data_off = (RTE_PKTMBUF_HEADROOM <= m->buf_len) ?
1322 RTE_PKTMBUF_HEADROOM : m->buf_len;
1325 __rte_mbuf_sanity_check(m, 1);
1329 * Allocate a new mbuf from a mempool.
1331 * This new mbuf contains one segment, which has a length of 0. The pointer
1332 * to data is initialized to have some bytes of headroom in the buffer
1333 * (if buffer size allows).
1336 * The mempool from which the mbuf is allocated.
1338 * - The pointer to the new mbuf on success.
1339 * - NULL if allocation failed.
1341 static inline struct rte_mbuf *rte_pktmbuf_alloc(struct rte_mempool *mp)
1344 if ((m = __rte_mbuf_raw_alloc(mp)) != NULL)
1345 rte_pktmbuf_reset(m);
1350 * Attach packet mbuf to another packet mbuf.
1352 * After attachment we refer the mbuf we attached as 'indirect',
1353 * while mbuf we attached to as 'direct'.
1354 * Right now, not supported:
1355 * - attachment for already indirect mbuf (e.g. - mi has to be direct).
1356 * - mbuf we trying to attach (mi) is used by someone else
1357 * e.g. it's reference counter is greater then 1.
1360 * The indirect packet mbuf.
1362 * The packet mbuf we're attaching to.
1364 static inline void rte_pktmbuf_attach(struct rte_mbuf *mi, struct rte_mbuf *m)
1366 struct rte_mbuf *md;
1368 RTE_MBUF_ASSERT(RTE_MBUF_DIRECT(mi) &&
1369 rte_mbuf_refcnt_read(mi) == 1);
1371 /* if m is not direct, get the mbuf that embeds the data */
1372 if (RTE_MBUF_DIRECT(m))
1375 md = rte_mbuf_from_indirect(m);
1377 rte_mbuf_refcnt_update(md, 1);
1378 mi->priv_size = m->priv_size;
1379 mi->buf_physaddr = m->buf_physaddr;
1380 mi->buf_addr = m->buf_addr;
1381 mi->buf_len = m->buf_len;
1384 mi->data_off = m->data_off;
1385 mi->data_len = m->data_len;
1387 mi->vlan_tci = m->vlan_tci;
1388 mi->vlan_tci_outer = m->vlan_tci_outer;
1389 mi->tx_offload = m->tx_offload;
1393 mi->pkt_len = mi->data_len;
1395 mi->ol_flags = m->ol_flags | IND_ATTACHED_MBUF;
1396 mi->packet_type = m->packet_type;
1398 __rte_mbuf_sanity_check(mi, 1);
1399 __rte_mbuf_sanity_check(m, 0);
1403 * Detach an indirect packet mbuf.
1405 * - restore original mbuf address and length values.
1406 * - reset pktmbuf data and data_len to their default values.
1407 * All other fields of the given packet mbuf will be left intact.
1410 * The indirect attached packet mbuf.
1412 static inline void rte_pktmbuf_detach(struct rte_mbuf *m)
1414 struct rte_mempool *mp = m->pool;
1415 uint32_t mbuf_size, buf_len, priv_size;
1417 priv_size = rte_pktmbuf_priv_size(mp);
1418 mbuf_size = sizeof(struct rte_mbuf) + priv_size;
1419 buf_len = rte_pktmbuf_data_room_size(mp);
1421 m->priv_size = priv_size;
1422 m->buf_addr = (char *)m + mbuf_size;
1423 m->buf_physaddr = rte_mempool_virt2phy(mp, m) + mbuf_size;
1424 m->buf_len = (uint16_t)buf_len;
1425 m->data_off = RTE_MIN(RTE_PKTMBUF_HEADROOM, (uint16_t)m->buf_len);
1430 static inline struct rte_mbuf* __attribute__((always_inline))
1431 __rte_pktmbuf_prefree_seg(struct rte_mbuf *m)
1433 __rte_mbuf_sanity_check(m, 0);
1435 if (likely(rte_mbuf_refcnt_update(m, -1) == 0)) {
1437 /* if this is an indirect mbuf, then
1439 * - free attached mbuf segment
1441 if (RTE_MBUF_INDIRECT(m)) {
1442 struct rte_mbuf *md = rte_mbuf_from_indirect(m);
1443 rte_pktmbuf_detach(m);
1444 if (rte_mbuf_refcnt_update(md, -1) == 0)
1445 __rte_mbuf_raw_free(md);
1453 * Free a segment of a packet mbuf into its original mempool.
1455 * Free an mbuf, without parsing other segments in case of chained
1459 * The packet mbuf segment to be freed.
1461 static inline void __attribute__((always_inline))
1462 rte_pktmbuf_free_seg(struct rte_mbuf *m)
1464 if (likely(NULL != (m = __rte_pktmbuf_prefree_seg(m)))) {
1466 __rte_mbuf_raw_free(m);
1471 * Free a packet mbuf back into its original mempool.
1473 * Free an mbuf, and all its segments in case of chained buffers. Each
1474 * segment is added back into its original mempool.
1477 * The packet mbuf to be freed.
1479 static inline void rte_pktmbuf_free(struct rte_mbuf *m)
1481 struct rte_mbuf *m_next;
1483 __rte_mbuf_sanity_check(m, 1);
1487 rte_pktmbuf_free_seg(m);
1493 * Creates a "clone" of the given packet mbuf.
1495 * Walks through all segments of the given packet mbuf, and for each of them:
1496 * - Creates a new packet mbuf from the given pool.
1497 * - Attaches newly created mbuf to the segment.
1498 * Then updates pkt_len and nb_segs of the "clone" packet mbuf to match values
1499 * from the original packet mbuf.
1502 * The packet mbuf to be cloned.
1504 * The mempool from which the "clone" mbufs are allocated.
1506 * - The pointer to the new "clone" mbuf on success.
1507 * - NULL if allocation fails.
1509 static inline struct rte_mbuf *rte_pktmbuf_clone(struct rte_mbuf *md,
1510 struct rte_mempool *mp)
1512 struct rte_mbuf *mc, *mi, **prev;
1516 if (unlikely ((mc = rte_pktmbuf_alloc(mp)) == NULL))
1521 pktlen = md->pkt_len;
1526 rte_pktmbuf_attach(mi, md);
1529 } while ((md = md->next) != NULL &&
1530 (mi = rte_pktmbuf_alloc(mp)) != NULL);
1534 mc->pkt_len = pktlen;
1536 /* Allocation of new indirect segment failed */
1537 if (unlikely (mi == NULL)) {
1538 rte_pktmbuf_free(mc);
1542 __rte_mbuf_sanity_check(mc, 1);
1547 * Adds given value to the refcnt of all packet mbuf segments.
1549 * Walks through all segments of given packet mbuf and for each of them
1550 * invokes rte_mbuf_refcnt_update().
1553 * The packet mbuf whose refcnt to be updated.
1555 * The value to add to the mbuf's segments refcnt.
1557 static inline void rte_pktmbuf_refcnt_update(struct rte_mbuf *m, int16_t v)
1559 __rte_mbuf_sanity_check(m, 1);
1562 rte_mbuf_refcnt_update(m, v);
1563 } while ((m = m->next) != NULL);
1567 * Get the headroom in a packet mbuf.
1572 * The length of the headroom.
1574 static inline uint16_t rte_pktmbuf_headroom(const struct rte_mbuf *m)
1576 __rte_mbuf_sanity_check(m, 1);
1581 * Get the tailroom of a packet mbuf.
1586 * The length of the tailroom.
1588 static inline uint16_t rte_pktmbuf_tailroom(const struct rte_mbuf *m)
1590 __rte_mbuf_sanity_check(m, 1);
1591 return (uint16_t)(m->buf_len - rte_pktmbuf_headroom(m) -
1596 * Get the last segment of the packet.
1601 * The last segment of the given mbuf.
1603 static inline struct rte_mbuf *rte_pktmbuf_lastseg(struct rte_mbuf *m)
1605 struct rte_mbuf *m2 = (struct rte_mbuf *)m;
1607 __rte_mbuf_sanity_check(m, 1);
1608 while (m2->next != NULL)
1614 * A macro that points to an offset into the data in the mbuf.
1616 * The returned pointer is cast to type t. Before using this
1617 * function, the user must ensure that the first segment is large
1618 * enough to accommodate its data.
1623 * The offset into the mbuf data.
1625 * The type to cast the result into.
1627 #define rte_pktmbuf_mtod_offset(m, t, o) \
1628 ((t)((char *)(m)->buf_addr + (m)->data_off + (o)))
1631 * A macro that points to the start of the data in the mbuf.
1633 * The returned pointer is cast to type t. Before using this
1634 * function, the user must ensure that the first segment is large
1635 * enough to accommodate its data.
1640 * The type to cast the result into.
1642 #define rte_pktmbuf_mtod(m, t) rte_pktmbuf_mtod_offset(m, t, 0)
1645 * A macro that returns the length of the packet.
1647 * The value can be read or assigned.
1652 #define rte_pktmbuf_pkt_len(m) ((m)->pkt_len)
1655 * A macro that returns the length of the segment.
1657 * The value can be read or assigned.
1662 #define rte_pktmbuf_data_len(m) ((m)->data_len)
1665 * Prepend len bytes to an mbuf data area.
1667 * Returns a pointer to the new
1668 * data start address. If there is not enough headroom in the first
1669 * segment, the function will return NULL, without modifying the mbuf.
1674 * The amount of data to prepend (in bytes).
1676 * A pointer to the start of the newly prepended data, or
1677 * NULL if there is not enough headroom space in the first segment
1679 static inline char *rte_pktmbuf_prepend(struct rte_mbuf *m,
1682 __rte_mbuf_sanity_check(m, 1);
1684 if (unlikely(len > rte_pktmbuf_headroom(m)))
1688 m->data_len = (uint16_t)(m->data_len + len);
1689 m->pkt_len = (m->pkt_len + len);
1691 return (char *)m->buf_addr + m->data_off;
1695 * Append len bytes to an mbuf.
1697 * Append len bytes to an mbuf and return a pointer to the start address
1698 * of the added data. If there is not enough tailroom in the last
1699 * segment, the function will return NULL, without modifying the mbuf.
1704 * The amount of data to append (in bytes).
1706 * A pointer to the start of the newly appended data, or
1707 * NULL if there is not enough tailroom space in the last segment
1709 static inline char *rte_pktmbuf_append(struct rte_mbuf *m, uint16_t len)
1712 struct rte_mbuf *m_last;
1714 __rte_mbuf_sanity_check(m, 1);
1716 m_last = rte_pktmbuf_lastseg(m);
1717 if (unlikely(len > rte_pktmbuf_tailroom(m_last)))
1720 tail = (char *)m_last->buf_addr + m_last->data_off + m_last->data_len;
1721 m_last->data_len = (uint16_t)(m_last->data_len + len);
1722 m->pkt_len = (m->pkt_len + len);
1723 return (char*) tail;
1727 * Remove len bytes at the beginning of an mbuf.
1729 * Returns a pointer to the start address of the new data area. If the
1730 * length is greater than the length of the first segment, then the
1731 * function will fail and return NULL, without modifying the mbuf.
1736 * The amount of data to remove (in bytes).
1738 * A pointer to the new start of the data.
1740 static inline char *rte_pktmbuf_adj(struct rte_mbuf *m, uint16_t len)
1742 __rte_mbuf_sanity_check(m, 1);
1744 if (unlikely(len > m->data_len))
1747 m->data_len = (uint16_t)(m->data_len - len);
1749 m->pkt_len = (m->pkt_len - len);
1750 return (char *)m->buf_addr + m->data_off;
1754 * Remove len bytes of data at the end of the mbuf.
1756 * If the length is greater than the length of the last segment, the
1757 * function will fail and return -1 without modifying the mbuf.
1762 * The amount of data to remove (in bytes).
1767 static inline int rte_pktmbuf_trim(struct rte_mbuf *m, uint16_t len)
1769 struct rte_mbuf *m_last;
1771 __rte_mbuf_sanity_check(m, 1);
1773 m_last = rte_pktmbuf_lastseg(m);
1774 if (unlikely(len > m_last->data_len))
1777 m_last->data_len = (uint16_t)(m_last->data_len - len);
1778 m->pkt_len = (m->pkt_len - len);
1783 * Test if mbuf data is contiguous.
1788 * - 1, if all data is contiguous (one segment).
1789 * - 0, if there is several segments.
1791 static inline int rte_pktmbuf_is_contiguous(const struct rte_mbuf *m)
1793 __rte_mbuf_sanity_check(m, 1);
1794 return !!(m->nb_segs == 1);
1798 * Dump an mbuf structure to the console.
1800 * Dump all fields for the given packet mbuf and all its associated
1801 * segments (in the case of a chained buffer).
1804 * A pointer to a file for output
1808 * If dump_len != 0, also dump the "dump_len" first data bytes of
1811 void rte_pktmbuf_dump(FILE *f, const struct rte_mbuf *m, unsigned dump_len);
1817 #endif /* _RTE_MBUF_H_ */