2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)mbuf.h 8.5 (Berkeley) 2/19/95
34 * $FreeBSD: src/sys/sys/mbuf.h,v 1.44.2.17 2003/04/15 06:15:02 silby Exp $
35 * $DragonFly: src/sys/sys/mbuf.h,v 1.14 2004/07/31 07:52:50 dillon Exp $
41 #include <sys/queue.h>
44 * Mbufs are of a single size, MSIZE (machine/param.h), which
45 * includes overhead. An mbuf may add a single "mbuf cluster" of size
46 * MCLBYTES (also in machine/param.h), which has no additional overhead
47 * and is used instead of the internal data area; this is done when
48 * at least MINCLSIZE of data must be stored.
50 #define MLEN (MSIZE - sizeof(struct m_hdr)) /* normal data len */
51 #define MHLEN (MLEN - sizeof(struct pkthdr)) /* data len w/pkthdr */
52 #define MINCLSIZE (MHLEN + 1) /* smallest amount to put in cluster */
53 #define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */
56 * Macros for type conversion:
57 * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type.
58 * mtocl(x) - convert pointer within cluster to cluster index #
59 * cltom(x) - convert cluster # to ptr to beginning of cluster
61 #define mtod(m, t) ((t)((m)->m_data))
64 * Header present at the beginning of every mbuf.
67 struct mbuf *mh_next; /* next buffer in chain */
68 struct mbuf *mh_nextpkt; /* next chain in queue/record */
69 caddr_t mh_data; /* location of data */
70 int mh_len; /* amount of data in this mbuf */
71 short mh_type; /* type of data in this mbuf */
72 short mh_flags; /* flags; see below */
76 * Packet tag structure (see below for details).
79 SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */
80 u_int16_t m_tag_id; /* Tag ID */
81 u_int16_t m_tag_len; /* Length of data */
82 u_int32_t m_tag_cookie; /* ABI/Module ID */
86 * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set.
89 struct ifnet *rcvif; /* rcv interface */
90 int len; /* total packet length */
91 /* variables for ip and tcp reassembly */
92 void *header; /* pointer to packet header */
93 /* variables for hardware checksum */
94 int csum_flags; /* flags regarding checksum */
95 int csum_data; /* data field used by csum routines */
96 SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */
100 * Description of external storage mapped into mbuf; valid only if M_EXT is set.
103 caddr_t ext_buf; /* start of buffer */
105 void (*old)(caddr_t, u_int);
106 void (*new)(void *arg);
109 u_int ext_size; /* size of buffer, for ext_nfree */
111 void (*old)(caddr_t, u_int);
112 void (*new)(void *arg);
119 * The core of the mbuf object along with some shortcut defines for
120 * practical purposes.
126 struct pkthdr MH_pkthdr; /* M_PKTHDR set */
128 struct m_ext MH_ext; /* M_EXT set */
129 char MH_databuf[MHLEN];
132 char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */
135 #define m_next m_hdr.mh_next
136 #define m_len m_hdr.mh_len
137 #define m_data m_hdr.mh_data
138 #define m_type m_hdr.mh_type
139 #define m_flags m_hdr.mh_flags
140 #define m_nextpkt m_hdr.mh_nextpkt
141 #define m_pkthdr M_dat.MH.MH_pkthdr
142 #define m_ext M_dat.MH.MH_dat.MH_ext
143 #define m_pktdat M_dat.MH.MH_dat.MH_databuf
144 #define m_dat M_dat.M_databuf
147 * Code that uses m_act should be converted to use m_nextpkt
148 * instead; m_act is historical and deprecated.
150 #define m_act m_nextpkt
155 #define M_EXT 0x0001 /* has associated external storage */
156 #define M_PKTHDR 0x0002 /* start of record */
157 #define M_EOR 0x0004 /* end of record */
158 #define M_PROTO1 0x0008 /* protocol-specific */
159 #define M_PROTO2 0x0010 /* protocol-specific */
160 #define M_PROTO3 0x0020 /* protocol-specific */
161 #define M_PROTO4 0x0040 /* protocol-specific */
162 #define M_PROTO5 0x0080 /* protocol-specific */
165 * mbuf pkthdr flags (also stored in m_flags).
167 #define M_BCAST 0x0100 /* send/received as link-level broadcast */
168 #define M_MCAST 0x0200 /* send/received as link-level multicast */
169 #define M_FRAG 0x0400 /* packet is a fragment of a larger packet */
170 #define M_FIRSTFRAG 0x0800 /* packet is first fragment */
171 #define M_LASTFRAG 0x1000 /* packet is last fragment */
172 #define M_EXT_OLD 0x2000 /* new ext function format */
173 #define M_EXT_CLUSTER 0x4000 /* standard cluster else special */
176 * Flags copied when copying m_pkthdr.
178 #define M_COPYFLAGS (M_PKTHDR|M_EOR|M_PROTO1|M_PROTO1|M_PROTO2|M_PROTO3 | \
179 M_PROTO4|M_PROTO5|M_BCAST|M_MCAST|M_FRAG | \
180 M_FIRSTFRAG|M_LASTFRAG)
183 * Flags indicating hw checksum support and sw checksum requirements.
185 #define CSUM_IP 0x0001 /* will csum IP */
186 #define CSUM_TCP 0x0002 /* will csum TCP */
187 #define CSUM_UDP 0x0004 /* will csum UDP */
188 #define CSUM_IP_FRAGS 0x0008 /* will csum IP fragments */
189 #define CSUM_FRAGMENT 0x0010 /* will do IP fragmentation */
191 #define CSUM_IP_CHECKED 0x0100 /* did csum IP */
192 #define CSUM_IP_VALID 0x0200 /* ... the csum is valid */
193 #define CSUM_DATA_VALID 0x0400 /* csum_data field is valid */
194 #define CSUM_PSEUDO_HDR 0x0800 /* csum_data has pseudo hdr */
196 #define CSUM_DELAY_DATA (CSUM_TCP | CSUM_UDP)
197 #define CSUM_DELAY_IP (CSUM_IP) /* XXX add ipv6 here too? */
202 #define MT_FREE 0 /* should be on free list */
203 #define MT_DATA 1 /* dynamic (data) allocation */
204 #define MT_HEADER 2 /* packet header */
206 #define MT_SOCKET 3 /* socket structure */
207 #define MT_PCB 4 /* protocol control block */
208 #define MT_RTABLE 5 /* routing tables */
209 #define MT_HTABLE 6 /* IMP host tables */
210 #define MT_ATABLE 7 /* address resolution tables */
212 #define MT_SONAME 8 /* socket name */
214 #define MT_SOOPTS 10 /* socket options */
216 #define MT_FTABLE 11 /* fragment reassembly header */
218 #define MT_RIGHTS 12 /* access rights */
219 #define MT_IFADDR 13 /* interface address */
221 #define MT_TAG 13 /* volatile metadata associated to pkts */
222 #define MT_CONTROL 14 /* extra-data protocol message */
223 #define MT_OOBDATA 15 /* expedited data */
224 #define MT_NTYPES 16 /* number of mbuf types for mbtypes[] */
227 * General mbuf allocator statistics structure.
230 u_long m_mbufs; /* mbufs obtained from page pool */
231 u_long m_clusters; /* clusters obtained from page pool */
232 u_long m_spare; /* spare field */
233 u_long m_clfree; /* free clusters */
234 u_long m_drops; /* times failed to find space */
235 u_long m_wait; /* times waited for space */
236 u_long m_drain; /* times drained protocols for space */
237 u_long m_mcfail; /* times m_copym failed */
238 u_long m_mpfail; /* times m_pullup failed */
239 u_long m_msize; /* length of an mbuf */
240 u_long m_mclbytes; /* length of an mbuf cluster */
241 u_long m_minclsize; /* min length of data to allocate a cluster */
242 u_long m_mlen; /* length of data in an mbuf */
243 u_long m_mhlen; /* length of data in a header mbuf */
247 * Flags specifying how an allocation should be made.
250 #define MB_DONTWAIT 0x4
251 #define MB_TRYWAIT 0x8
252 #define MB_WAIT MB_TRYWAIT
255 * Mbuf to Malloc Flag Conversion.
257 #define MBTOM(how) ((how) & MB_TRYWAIT ? M_WAITOK : M_NOWAIT)
260 * These are identifying numbers passed to the m_mballoc_wait function,
261 * allowing us to determine whether the call came from an MGETHDR or
268 * Wake up the next instance (if any) of m_mballoc_wait() which is
269 * waiting for an mbuf to be freed. This should be called at splimp().
271 * XXX: If there is another free mbuf, this routine will be called [again]
272 * from the m_mballoc_wait routine in order to wake another sleep instance.
274 #define MMBWAKEUP() do { \
275 if (m_mballoc_wid) { \
277 wakeup_one(&m_mballoc_wid); \
282 * Same as above, but for mbuf cluster(s).
284 #define MCLWAKEUP() do { \
285 if (m_clalloc_wid) { \
287 wakeup_one(&m_clalloc_wid); \
292 * mbuf utility macros:
295 * prevents a section of code from from being interrupted by network
298 #define MBUFLOCK(code) do { \
299 int _ms = splimp(); \
306 * mbuf allocation/deallocation macros (YYY deprecated, too big):
308 * MGET(struct mbuf *m, int how, int type)
309 * allocates an mbuf and initializes it to contain internal data.
311 * MGETHDR(struct mbuf *m, int how, int type)
312 * allocates an mbuf and initializes it to contain a packet header
315 #define MGET(m, how, type) do { \
316 (m) = m_get((how), (type)); \
319 #define MGETHDR(m, how, type) do { \
320 (m) = m_gethdr((how), (type)); \
324 * MCLGET adds such clusters to a normal mbuf. The flag M_EXT is set upon
327 #define MCLGET(m, how) do { \
328 m_mclget((m), (how)); \
332 * NB: M_COPY_PKTHDR is deprecated; use either M_MOVE_PKTHDR
336 * Move mbuf pkthdr from "from" to "to".
337 * from should have M_PKTHDR set, and to must be empty.
338 * from no longer has a pkthdr after this operation.
340 #define M_MOVE_PKTHDR(_to, _from) m_move_pkthdr((_to), (_from))
343 * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place
344 * an object of the specified size at the end of the mbuf, longword aligned.
346 #define M_ALIGN(m, len) do { \
347 (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \
351 * As above, for mbufs allocated with m_gethdr/MGETHDR
352 * or initialized by M_COPY_PKTHDR.
354 #define MH_ALIGN(m, len) do { \
355 (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \
359 * Check if we can write to an mbuf.
361 #define M_EXT_WRITABLE(m) (m_sharecount(m) == 1)
362 #define M_WRITABLE(m) (!((m)->m_flags & M_EXT) || M_EXT_WRITABLE(m))
365 * Compute the amount of space available
366 * before the current start of data in an mbuf.
368 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
369 * of checking writability of the mbuf data area rests solely with the caller.
371 #define M_LEADINGSPACE(m) \
372 ((m)->m_flags & M_EXT ? \
373 (M_EXT_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \
374 (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \
375 (m)->m_data - (m)->m_dat)
378 * Compute the amount of space available
379 * after the end of data in an mbuf.
381 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
382 * of checking writability of the mbuf data area rests solely with the caller.
384 #define M_TRAILINGSPACE(m) \
385 ((m)->m_flags & M_EXT ? \
386 (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \
387 - ((m)->m_data + (m)->m_len) : 0) : \
388 &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len))
391 * Arrange to prepend space of size plen to mbuf m.
392 * If a new mbuf must be allocated, how specifies whether to wait.
393 * If how is MB_DONTWAIT and allocation fails, the original mbuf chain
394 * is freed and m is set to NULL.
396 #define M_PREPEND(m, plen, how) do { \
397 struct mbuf **_mmp = &(m); \
398 struct mbuf *_mm = *_mmp; \
399 int _mplen = (plen); \
400 int __mhow = (how); \
402 if (M_LEADINGSPACE(_mm) >= _mplen) { \
403 _mm->m_data -= _mplen; \
404 _mm->m_len += _mplen; \
406 _mm = m_prepend(_mm, _mplen, __mhow); \
407 if (_mm != NULL && _mm->m_flags & M_PKTHDR) \
408 _mm->m_pkthdr.len += _mplen; \
412 /* Length to m_copy to copy all. */
413 #define M_COPYALL 1000000000
415 /* Compatibility with 4.3 */
416 #define m_copy(m, o, l) m_copym((m), (o), (l), MB_DONTWAIT)
419 extern u_int m_clalloc_wid; /* mbuf cluster wait count */
420 extern u_int m_mballoc_wid; /* mbuf wait count */
421 extern int max_linkhdr; /* largest link-level header */
422 extern int max_protohdr; /* largest protocol header */
423 extern int max_hdr; /* largest link+protocol header */
424 extern int max_datalen; /* MHLEN - max_hdr */
425 extern struct mbstat mbstat;
426 extern int mbuf_wait; /* mbuf sleep time */
427 extern int nmbclusters;
432 void m_adj(struct mbuf *, int);
433 void m_cat(struct mbuf *, struct mbuf *);
434 void m_copyback(struct mbuf *, int, int, caddr_t);
435 void m_copydata(const struct mbuf *, int, int, caddr_t);
436 struct mbuf *m_copym(const struct mbuf *, int, int, int);
437 struct mbuf *m_copypacket(struct mbuf *, int);
438 struct mbuf *m_defrag(struct mbuf *, int);
439 struct mbuf *m_devget(char *, int, int, struct ifnet *,
440 void (*copy)(char *, caddr_t, u_int));
441 struct mbuf *m_dup(struct mbuf *, int);
442 int m_dup_pkthdr(struct mbuf *, const struct mbuf *, int);
443 struct mbuf *m_free(struct mbuf *);
444 void m_freem(struct mbuf *);
445 struct mbuf *m_get(int, int);
446 struct mbuf *m_getcl(int how, short type, int flags);
447 struct mbuf *m_getclr(int, int);
448 struct mbuf *m_gethdr(int, int);
449 struct mbuf *m_getm(struct mbuf *, int, int, int);
450 void m_move_pkthdr(struct mbuf *, struct mbuf *);
451 struct mbuf *m_prepend(struct mbuf *, int, int);
452 void m_print(const struct mbuf *m);
453 struct mbuf *m_pulldown(struct mbuf *, int, int, int *);
454 struct mbuf *m_pullup(struct mbuf *, int);
455 struct mbuf *m_retry(int, int);
456 struct mbuf *m_retryhdr(int, int);
457 struct mbuf *m_split(struct mbuf *, int, int);
458 struct mbuf *m_uiomove(struct uio *, int, int);
459 void m_mclget(struct mbuf *m, int how);
460 int m_sharecount(struct mbuf *m);
461 void m_chtype(struct mbuf *m, int type);
465 * Packets may have annotations attached by affixing a list
466 * of "packet tags" to the pkthdr structure. Packet tags are
467 * dynamically allocated semi-opaque data structures that have
468 * a fixed header (struct m_tag) that specifies the size of the
469 * memory block and a <cookie,type> pair that identifies it.
470 * The cookie is a 32-bit unique unsigned value used to identify
471 * a module or ABI. By convention this value is chose as the
472 * date+time that the module is created, expressed as the number of
473 * seconds since the epoch (e.g. using date -u +'%s'). The type value
474 * is an ABI/module-specific value that identifies a particular annotation
475 * and is private to the module. For compatibility with systems
476 * like openbsd that define packet tags w/o an ABI/module cookie,
477 * the value PACKET_ABI_COMPAT is used to implement m_tag_get and
478 * m_tag_find compatibility shim functions and several tag types are
479 * defined below. Users that do not require compatibility should use
480 * a private cookie value so that packet tag-related definitions
481 * can be maintained privately.
483 * Note that the packet tag returned by m_tag_allocate has the default
484 * memory alignment implemented by malloc. To reference private data
485 * one can use a construct like:
487 * struct m_tag *mtag = m_tag_allocate(...);
488 * struct foo *p = (struct foo *)(mtag+1);
490 * if the alignment of struct m_tag is sufficient for referencing members
491 * of struct foo. Otherwise it is necessary to embed struct m_tag within
492 * the private data structure to insure proper alignment; e.g.
498 * struct foo *p = (struct foo *) m_tag_allocate(...);
499 * struct m_tag *mtag = &p->tag;
502 #define PACKET_TAG_NONE 0 /* Nadda */
504 /* Packet tag for use with PACKET_ABI_COMPAT */
505 #define PACKET_TAG_IPSEC_IN_DONE 1 /* IPsec applied, in */
506 #define PACKET_TAG_IPSEC_OUT_DONE 2 /* IPsec applied, out */
507 #define PACKET_TAG_IPSEC_IN_CRYPTO_DONE 3 /* NIC IPsec crypto done */
508 #define PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED 4 /* NIC IPsec crypto req'ed */
509 #define PACKET_TAG_IPSEC_IN_COULD_DO_CRYPTO 5 /* NIC notifies IPsec */
510 #define PACKET_TAG_IPSEC_PENDING_TDB 6 /* Reminder to do IPsec */
511 #define PACKET_TAG_BRIDGE 7 /* Bridge processing done */
512 #define PACKET_TAG_GIF 8 /* GIF processing done */
513 #define PACKET_TAG_GRE 9 /* GRE processing done */
514 #define PACKET_TAG_IN_PACKET_CHECKSUM 10 /* NIC checksumming done */
515 #define PACKET_TAG_ENCAP 11 /* Encap. processing */
516 #define PACKET_TAG_IPSEC_SOCKET 12 /* IPSEC socket ref */
517 #define PACKET_TAG_IPSEC_HISTORY 13 /* IPSEC history */
518 #define PACKET_TAG_IPV6_INPUT 14 /* IPV6 input processing */
521 * As a temporary and low impact solution to replace the even uglier
522 * approach used so far in some parts of the network stack (which relies
523 * on global variables), packet tag-like annotations are stored in MT_TAG
524 * mbufs (or lookalikes) prepended to the actual mbuf chain.
528 * m_next = next buffer in chain.
530 * BE VERY CAREFUL not to pass these blocks to the mbuf handling routines.
532 #define _m_tag_id m_hdr.mh_flags
534 /* Packet tags used in the FreeBSD network stack */
535 #define PACKET_TAG_DUMMYNET 15 /* dummynet info */
536 #define PACKET_TAG_IPFW 16 /* ipfw classification */
537 #define PACKET_TAG_DIVERT 17 /* divert info */
538 #define PACKET_TAG_IPFORWARD 18 /* ipforward info */
540 /* Packet tag routines */
541 struct m_tag *m_tag_alloc(u_int32_t, int, int, int);
542 void m_tag_free(struct m_tag *);
543 void m_tag_prepend(struct mbuf *, struct m_tag *);
544 void m_tag_unlink(struct mbuf *, struct m_tag *);
545 void m_tag_delete(struct mbuf *, struct m_tag *);
546 void m_tag_delete_chain(struct mbuf *, struct m_tag *);
547 struct m_tag *m_tag_locate(struct mbuf *, u_int32_t, int, struct m_tag *);
548 struct m_tag *m_tag_copy(struct m_tag *, int);
549 int m_tag_copy_chain(struct mbuf *, const struct mbuf *, int);
550 void m_tag_init(struct mbuf *);
551 struct m_tag *m_tag_first(struct mbuf *);
552 struct m_tag *m_tag_next(struct mbuf *, struct m_tag *);
554 /* these are for openbsd compatibility */
555 #define MTAG_ABI_COMPAT 0 /* compatibility ABI */
557 static __inline struct m_tag *
558 m_tag_get(int type, int length, int wait)
560 return m_tag_alloc(MTAG_ABI_COMPAT, type, length, wait);
563 static __inline struct m_tag *
564 m_tag_find(struct mbuf *m, int type, struct m_tag *start)
566 return m_tag_locate(m, MTAG_ABI_COMPAT, type, start);
570 #endif /* !_SYS_MBUF_H_ */