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 1.8 2003/12/28 06:11:33 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 * dtom(x) -- Convert data pointer within mbuf to mbuf pointer (XXX).
59 * mtocl(x) - convert pointer within cluster to cluster index #
60 * cltom(x) - convert cluster # to ptr to beginning of cluster
62 #define mtod(m, t) ((t)((m)->m_data))
63 #define dtom(x) ((struct mbuf *)((intptr_t)(x) & ~(MSIZE-1)))
64 #define mtocl(x) (((uintptr_t)(x) - (uintptr_t)mbutl) >> MCLSHIFT)
65 #define cltom(x) ((caddr_t)((uintptr_t)mbutl + \
66 ((uintptr_t)(x) << MCLSHIFT)))
69 * Header present at the beginning of every mbuf.
72 struct mbuf *mh_next; /* next buffer in chain */
73 struct mbuf *mh_nextpkt; /* next chain in queue/record */
74 caddr_t mh_data; /* location of data */
75 int mh_len; /* amount of data in this mbuf */
76 short mh_type; /* type of data in this mbuf */
77 short mh_flags; /* flags; see below */
81 * Packet tag structure (see below for details).
84 SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */
85 u_int16_t m_tag_id; /* Tag ID */
86 u_int16_t m_tag_len; /* Length of data */
87 u_int32_t m_tag_cookie; /* ABI/Module ID */
91 * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set.
94 struct ifnet *rcvif; /* rcv interface */
95 int len; /* total packet length */
96 /* variables for ip and tcp reassembly */
97 void *header; /* pointer to packet header */
98 /* variables for hardware checksum */
99 int csum_flags; /* flags regarding checksum */
100 int csum_data; /* data field used by csum routines */
101 SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */
105 * Description of external storage mapped into mbuf; valid only if M_EXT is set.
108 caddr_t ext_buf; /* start of buffer */
109 void (*ext_free) /* free routine if not the usual */
111 u_int ext_size; /* size of buffer, for ext_free */
112 void (*ext_ref) /* add a reference to the ext object */
117 * The core of the mbuf object along with some shortcut defines for
118 * practical purposes.
124 struct pkthdr MH_pkthdr; /* M_PKTHDR set */
126 struct m_ext MH_ext; /* M_EXT set */
127 char MH_databuf[MHLEN];
130 char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */
133 #define m_next m_hdr.mh_next
134 #define m_len m_hdr.mh_len
135 #define m_data m_hdr.mh_data
136 #define m_type m_hdr.mh_type
137 #define m_flags m_hdr.mh_flags
138 #define m_nextpkt m_hdr.mh_nextpkt
139 #define m_act m_nextpkt
140 #define m_pkthdr M_dat.MH.MH_pkthdr
141 #define m_ext M_dat.MH.MH_dat.MH_ext
142 #define m_pktdat M_dat.MH.MH_dat.MH_databuf
143 #define m_dat M_dat.M_databuf
148 #define M_EXT 0x0001 /* has associated external storage */
149 #define M_PKTHDR 0x0002 /* start of record */
150 #define M_EOR 0x0004 /* end of record */
151 #define M_PROTO1 0x0008 /* protocol-specific */
152 #define M_PROTO2 0x0010 /* protocol-specific */
153 #define M_PROTO3 0x0020 /* protocol-specific */
154 #define M_PROTO4 0x0040 /* protocol-specific */
155 #define M_PROTO5 0x0080 /* protocol-specific */
158 * mbuf pkthdr flags (also stored in m_flags).
160 #define M_BCAST 0x0100 /* send/received as link-level broadcast */
161 #define M_MCAST 0x0200 /* send/received as link-level multicast */
162 #define M_FRAG 0x0400 /* packet is a fragment of a larger packet */
163 #define M_FIRSTFRAG 0x0800 /* packet is first fragment */
164 #define M_LASTFRAG 0x1000 /* packet is last fragment */
167 * Flags copied when copying m_pkthdr.
169 #define M_COPYFLAGS (M_PKTHDR|M_EOR|M_PROTO1|M_PROTO1|M_PROTO2|M_PROTO3 | \
170 M_PROTO4|M_PROTO5|M_BCAST|M_MCAST|M_FRAG | \
171 M_FIRSTFRAG|M_LASTFRAG)
174 * Flags indicating hw checksum support and sw checksum requirements.
176 #define CSUM_IP 0x0001 /* will csum IP */
177 #define CSUM_TCP 0x0002 /* will csum TCP */
178 #define CSUM_UDP 0x0004 /* will csum UDP */
179 #define CSUM_IP_FRAGS 0x0008 /* will csum IP fragments */
180 #define CSUM_FRAGMENT 0x0010 /* will do IP fragmentation */
182 #define CSUM_IP_CHECKED 0x0100 /* did csum IP */
183 #define CSUM_IP_VALID 0x0200 /* ... the csum is valid */
184 #define CSUM_DATA_VALID 0x0400 /* csum_data field is valid */
185 #define CSUM_PSEUDO_HDR 0x0800 /* csum_data has pseudo hdr */
187 #define CSUM_DELAY_DATA (CSUM_TCP | CSUM_UDP)
188 #define CSUM_DELAY_IP (CSUM_IP) /* XXX add ipv6 here too? */
193 #define MT_FREE 0 /* should be on free list */
194 #define MT_DATA 1 /* dynamic (data) allocation */
195 #define MT_HEADER 2 /* packet header */
197 #define MT_SOCKET 3 /* socket structure */
198 #define MT_PCB 4 /* protocol control block */
199 #define MT_RTABLE 5 /* routing tables */
200 #define MT_HTABLE 6 /* IMP host tables */
201 #define MT_ATABLE 7 /* address resolution tables */
203 #define MT_SONAME 8 /* socket name */
205 #define MT_SOOPTS 10 /* socket options */
207 #define MT_FTABLE 11 /* fragment reassembly header */
209 #define MT_RIGHTS 12 /* access rights */
210 #define MT_IFADDR 13 /* interface address */
212 #define MT_TAG 13 /* volatile metadata associated to pkts */
213 #define MT_CONTROL 14 /* extra-data protocol message */
214 #define MT_OOBDATA 15 /* expedited data */
215 #define MT_NTYPES 16 /* number of mbuf types for mbtypes[] */
218 * General mbuf allocator statistics structure.
221 u_long m_mbufs; /* mbufs obtained from page pool */
222 u_long m_clusters; /* clusters obtained from page pool */
223 u_long m_spare; /* spare field */
224 u_long m_clfree; /* free clusters */
225 u_long m_drops; /* times failed to find space */
226 u_long m_wait; /* times waited for space */
227 u_long m_drain; /* times drained protocols for space */
228 u_long m_mcfail; /* times m_copym failed */
229 u_long m_mpfail; /* times m_pullup failed */
230 u_long m_msize; /* length of an mbuf */
231 u_long m_mclbytes; /* length of an mbuf cluster */
232 u_long m_minclsize; /* min length of data to allocate a cluster */
233 u_long m_mlen; /* length of data in an mbuf */
234 u_long m_mhlen; /* length of data in a header mbuf */
238 * Flags specifying how an allocation should be made.
241 #define M_DONTWAIT 0x4
242 #define M_TRYWAIT 0x8
243 #define M_WAIT M_TRYWAIT
246 * Mbuf to Malloc Flag Conversion.
248 #define MBTOM(how) ((how) & M_TRYWAIT ? M_WAITOK : M_NOWAIT)
252 * Normal mbuf clusters are normally treated as character arrays
253 * after allocation, but use the first word of the buffer as a free list
254 * pointer while on the free list.
257 union mcluster *mcl_next;
258 char mcl_buf[MCLBYTES];
263 * These are identifying numbers passed to the m_mballoc_wait function,
264 * allowing us to determine whether the call came from an MGETHDR or
271 * Wake up the next instance (if any) of m_mballoc_wait() which is
272 * waiting for an mbuf to be freed. This should be called at splimp().
274 * XXX: If there is another free mbuf, this routine will be called [again]
275 * from the m_mballoc_wait routine in order to wake another sleep instance.
277 #define MMBWAKEUP() do { \
278 if (m_mballoc_wid) { \
280 wakeup_one(&m_mballoc_wid); \
285 * Same as above, but for mbuf cluster(s).
287 #define MCLWAKEUP() do { \
288 if (m_clalloc_wid) { \
290 wakeup_one(&m_clalloc_wid); \
295 * mbuf utility macros:
298 * prevents a section of code from from being interrupted by network
301 #define MBUFLOCK(code) do { \
302 int _ms = splimp(); \
309 * mbuf allocation/deallocation macros (YYY deprecated, too big):
311 * MGET(struct mbuf *m, int how, int type)
312 * allocates an mbuf and initializes it to contain internal data.
314 * MGETHDR(struct mbuf *m, int how, int type)
315 * allocates an mbuf and initializes it to contain a packet header
318 #define MGET(m, how, type) do { \
319 (m) = m_get((how), (type)); \
322 #define MGETHDR(m, how, type) do { \
323 (m) = m_gethdr((how), (type)); \
327 * Mbuf cluster macros.
328 * MCLALLOC(caddr_t p, int how) allocates an mbuf cluster.
329 * MCLGET adds such clusters to a normal mbuf;
330 * the flag M_EXT is set upon success.
331 * MCLFREE releases a reference to a cluster allocated by MCLALLOC,
332 * freeing the cluster if the reference count has reached 0.
334 #define MCLALLOC(p, how) do { \
335 (p) = m_mclalloc(how); \
338 #define MCLGET(m, how) do { \
339 m_mclget((m), (how)); \
343 * NB: M_COPY_PKTHDR is deprecated; use either M_MOVE_PKTHDR
347 * Move mbuf pkthdr from "from" to "to".
348 * from should have M_PKTHDR set, and to must be empty.
349 * from no longer has a pkthdr after this operation.
351 #define M_MOVE_PKTHDR(_to, _from) m_move_pkthdr((_to), (_from))
354 * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place
355 * an object of the specified size at the end of the mbuf, longword aligned.
357 #define M_ALIGN(m, len) do { \
358 (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \
362 * As above, for mbufs allocated with m_gethdr/MGETHDR
363 * or initialized by M_COPY_PKTHDR.
365 #define MH_ALIGN(m, len) do { \
366 (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \
370 * Check if we can write to an mbuf.
372 #define M_EXT_WRITABLE(m) \
373 ((m)->m_ext.ext_free == NULL && mclrefcnt[mtocl((m)->m_ext.ext_buf)] == 1)
375 #define M_WRITABLE(m) (!((m)->m_flags & M_EXT) || \
379 * Compute the amount of space available
380 * before the current start of data in an mbuf.
382 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
383 * of checking writability of the mbuf data area rests solely with the caller.
385 #define M_LEADINGSPACE(m) \
386 ((m)->m_flags & M_EXT ? \
387 (M_EXT_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \
388 (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \
389 (m)->m_data - (m)->m_dat)
392 * Compute the amount of space available
393 * after the end of data in an mbuf.
395 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
396 * of checking writability of the mbuf data area rests solely with the caller.
398 #define M_TRAILINGSPACE(m) \
399 ((m)->m_flags & M_EXT ? \
400 (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \
401 - ((m)->m_data + (m)->m_len) : 0) : \
402 &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len))
405 * Arrange to prepend space of size plen to mbuf m.
406 * If a new mbuf must be allocated, how specifies whether to wait.
407 * If how is M_DONTWAIT and allocation fails, the original mbuf chain
408 * is freed and m is set to NULL.
410 #define M_PREPEND(m, plen, how) do { \
411 struct mbuf **_mmp = &(m); \
412 struct mbuf *_mm = *_mmp; \
413 int _mplen = (plen); \
414 int __mhow = (how); \
416 if (M_LEADINGSPACE(_mm) >= _mplen) { \
417 _mm->m_data -= _mplen; \
418 _mm->m_len += _mplen; \
420 _mm = m_prepend(_mm, _mplen, __mhow); \
421 if (_mm != NULL && _mm->m_flags & M_PKTHDR) \
422 _mm->m_pkthdr.len += _mplen; \
426 /* change mbuf to new type */
427 #define MCHTYPE(m, t) do { \
428 struct mbuf *_mm = (m); \
430 int _ms = splimp(); \
432 mbtypes[_mm->m_type]--; \
435 _mm->m_type = (_mt); \
438 /* Length to m_copy to copy all. */
439 #define M_COPYALL 1000000000
441 /* Compatibility with 4.3 */
442 #define m_copy(m, o, l) m_copym((m), (o), (l), M_DONTWAIT)
445 extern u_int m_clalloc_wid; /* mbuf cluster wait count */
446 extern u_int m_mballoc_wid; /* mbuf wait count */
447 extern int max_linkhdr; /* largest link-level header */
448 extern int max_protohdr; /* largest protocol header */
449 extern int max_hdr; /* largest link+protocol header */
450 extern int max_datalen; /* MHLEN - max_hdr */
451 extern struct mbstat mbstat;
452 extern u_long mbtypes[MT_NTYPES]; /* per-type mbuf allocations */
453 extern int mbuf_wait; /* mbuf sleep time */
454 extern struct mbuf *mbutl; /* virtual address of mclusters */
455 extern struct mbuf *mbute; /* ending VA of mclusters */
456 extern char *mclrefcnt; /* cluster reference counts */
457 extern union mcluster *mclfree;
458 extern struct mbuf *mmbfree;
459 extern int nmbclusters;
462 void m_adj(struct mbuf *, int);
463 void m_cat(struct mbuf *, struct mbuf *);
464 int m_clalloc(int, int);
465 caddr_t m_clalloc_wait(void);
466 void m_copyback(struct mbuf *, int, int, caddr_t);
467 void m_copydata(const struct mbuf *, int, int, caddr_t);
468 struct mbuf *m_copym(const struct mbuf *, int, int, int);
469 struct mbuf *m_copypacket(struct mbuf *, int);
470 struct mbuf *m_defrag(struct mbuf *, int);
471 struct mbuf *m_devget(char *, int, int, struct ifnet *,
472 void (*copy)(char *, caddr_t, u_int));
473 struct mbuf *m_dup(struct mbuf *, int);
474 int m_dup_pkthdr(struct mbuf *, const struct mbuf *, int);
475 struct mbuf *m_free(struct mbuf *);
476 void m_freem(struct mbuf *);
477 struct mbuf *m_get(int, int);
478 struct mbuf *m_getcl(int how, short type, int flags);
479 struct mbuf *m_getclr(int, int);
480 struct mbuf *m_gethdr(int, int);
481 struct mbuf *m_getm(struct mbuf *, int, int, int);
482 int m_mballoc(int, int);
483 struct mbuf *m_mballoc_wait(int, int);
484 void m_move_pkthdr(struct mbuf *, struct mbuf *);
485 struct mbuf *m_prepend(struct mbuf *, int, int);
486 void m_print(const struct mbuf *m);
487 struct mbuf *m_pulldown(struct mbuf *, int, int, int *);
488 struct mbuf *m_pullup(struct mbuf *, int);
489 struct mbuf *m_retry(int, int);
490 struct mbuf *m_retryhdr(int, int);
491 struct mbuf *m_split(struct mbuf *, int, int);
492 caddr_t m_mclalloc(int how);
493 void m_mclget(struct mbuf *m, int how);
494 void m_mclfree(caddr_t data);
497 * Packets may have annotations attached by affixing a list
498 * of "packet tags" to the pkthdr structure. Packet tags are
499 * dynamically allocated semi-opaque data structures that have
500 * a fixed header (struct m_tag) that specifies the size of the
501 * memory block and a <cookie,type> pair that identifies it.
502 * The cookie is a 32-bit unique unsigned value used to identify
503 * a module or ABI. By convention this value is chose as the
504 * date+time that the module is created, expressed as the number of
505 * seconds since the epoch (e.g. using date -u +'%s'). The type value
506 * is an ABI/module-specific value that identifies a particular annotation
507 * and is private to the module. For compatibility with systems
508 * like openbsd that define packet tags w/o an ABI/module cookie,
509 * the value PACKET_ABI_COMPAT is used to implement m_tag_get and
510 * m_tag_find compatibility shim functions and several tag types are
511 * defined below. Users that do not require compatibility should use
512 * a private cookie value so that packet tag-related definitions
513 * can be maintained privately.
515 * Note that the packet tag returned by m_tag_allocate has the default
516 * memory alignment implemented by malloc. To reference private data
517 * one can use a construct like:
519 * struct m_tag *mtag = m_tag_allocate(...);
520 * struct foo *p = (struct foo *)(mtag+1);
522 * if the alignment of struct m_tag is sufficient for referencing members
523 * of struct foo. Otherwise it is necessary to embed struct m_tag within
524 * the private data structure to insure proper alignment; e.g.
530 * struct foo *p = (struct foo *) m_tag_allocate(...);
531 * struct m_tag *mtag = &p->tag;
534 #define PACKET_TAG_NONE 0 /* Nadda */
536 /* Packet tag for use with PACKET_ABI_COMPAT */
537 #define PACKET_TAG_IPSEC_IN_DONE 1 /* IPsec applied, in */
538 #define PACKET_TAG_IPSEC_OUT_DONE 2 /* IPsec applied, out */
539 #define PACKET_TAG_IPSEC_IN_CRYPTO_DONE 3 /* NIC IPsec crypto done */
540 #define PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED 4 /* NIC IPsec crypto req'ed */
541 #define PACKET_TAG_IPSEC_IN_COULD_DO_CRYPTO 5 /* NIC notifies IPsec */
542 #define PACKET_TAG_IPSEC_PENDING_TDB 6 /* Reminder to do IPsec */
543 #define PACKET_TAG_BRIDGE 7 /* Bridge processing done */
544 #define PACKET_TAG_GIF 8 /* GIF processing done */
545 #define PACKET_TAG_GRE 9 /* GRE processing done */
546 #define PACKET_TAG_IN_PACKET_CHECKSUM 10 /* NIC checksumming done */
547 #define PACKET_TAG_ENCAP 11 /* Encap. processing */
548 #define PACKET_TAG_IPSEC_SOCKET 12 /* IPSEC socket ref */
549 #define PACKET_TAG_IPSEC_HISTORY 13 /* IPSEC history */
550 #define PACKET_TAG_IPV6_INPUT 14 /* IPV6 input processing */
553 * As a temporary and low impact solution to replace the even uglier
554 * approach used so far in some parts of the network stack (which relies
555 * on global variables), packet tag-like annotations are stored in MT_TAG
556 * mbufs (or lookalikes) prepended to the actual mbuf chain.
560 * m_next = next buffer in chain.
562 * BE VERY CAREFUL not to pass these blocks to the mbuf handling routines.
564 #define _m_tag_id m_hdr.mh_flags
566 /* Packet tags used in the FreeBSD network stack */
567 #define PACKET_TAG_DUMMYNET 15 /* dummynet info */
568 #define PACKET_TAG_IPFW 16 /* ipfw classification */
569 #define PACKET_TAG_DIVERT 17 /* divert info */
570 #define PACKET_TAG_IPFORWARD 18 /* ipforward info */
572 /* Packet tag routines */
573 struct m_tag *m_tag_alloc(u_int32_t, int, int, int);
574 void m_tag_free(struct m_tag *);
575 void m_tag_prepend(struct mbuf *, struct m_tag *);
576 void m_tag_unlink(struct mbuf *, struct m_tag *);
577 void m_tag_delete(struct mbuf *, struct m_tag *);
578 void m_tag_delete_chain(struct mbuf *, struct m_tag *);
579 struct m_tag *m_tag_locate(struct mbuf *, u_int32_t, int, struct m_tag *);
580 struct m_tag *m_tag_copy(struct m_tag *, int);
581 int m_tag_copy_chain(struct mbuf *, const struct mbuf *, int);
582 void m_tag_init(struct mbuf *);
583 struct m_tag *m_tag_first(struct mbuf *);
584 struct m_tag *m_tag_next(struct mbuf *, struct m_tag *);
586 /* these are for openbsd compatibility */
587 #define MTAG_ABI_COMPAT 0 /* compatibility ABI */
589 static __inline struct m_tag *
590 m_tag_get(int type, int length, int wait)
592 return m_tag_alloc(MTAG_ABI_COMPAT, type, length, wait);
595 static __inline struct m_tag *
596 m_tag_find(struct mbuf *m, int type, struct m_tag *start)
598 return m_tag_locate(m, MTAG_ABI_COMPAT, type, start);
602 #endif /* !_SYS_MBUF_H_ */