2 * Copyright (c) 2004 The DragonFly Project. All rights reserved.
4 * Copyright (c) 1982, 1986, 1988, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * @(#)mbuf.h 8.5 (Berkeley) 2/19/95
36 * $FreeBSD: src/sys/sys/mbuf.h,v 1.44.2.17 2003/04/15 06:15:02 silby Exp $
37 * $DragonFly: src/sys/sys/mbuf.h,v 1.33 2006/05/20 02:42:13 dillon Exp $
44 #include <sys/types.h>
47 #include <sys/param.h>
50 #include <sys/queue.h>
54 * Mbufs are of a single size, MSIZE (machine/param.h), which
55 * includes overhead. An mbuf may add a single "mbuf cluster" of size
56 * MCLBYTES (also in machine/param.h), which has no additional overhead
57 * and is used instead of the internal data area; this is done when
58 * at least MINCLSIZE of data must be stored.
60 #define MLEN (MSIZE - sizeof(struct m_hdr)) /* normal data len */
61 #define MHLEN (MLEN - sizeof(struct pkthdr)) /* data len w/pkthdr */
62 #define MINCLSIZE (MHLEN + 1) /* smallest amount to put in cluster */
63 #define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */
66 * Macros for type conversion:
67 * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type.
68 * mtocl(x) - convert pointer within cluster to cluster index #
69 * cltom(x) - convert cluster # to ptr to beginning of cluster
71 #define mtod(m, t) ((t)((m)->m_data))
74 * Header present at the beginning of every mbuf.
77 struct mbuf *mh_next; /* next buffer in chain */
78 struct mbuf *mh_nextpkt; /* next chain in queue/record */
79 caddr_t mh_data; /* location of data */
80 int mh_len; /* amount of data in this mbuf */
81 short mh_type; /* type of data in this mbuf */
82 short mh_flags; /* flags; see below */
86 * Packet tag structure (see below for details).
89 SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */
90 u_int16_t m_tag_id; /* Tag ID */
91 u_int16_t m_tag_len; /* Length of data */
92 u_int32_t m_tag_cookie; /* ABI/Module ID */
95 SLIST_HEAD(packet_tags, m_tag);
98 * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set.
100 * Be careful: The fields have been carefully ordered to avoid hidden padding.
101 * Keep this in mind, when adding or removing fields!
104 struct ifnet *rcvif; /* rcv interface */
105 int len; /* total packet length */
106 struct packet_tags tags; /* list of packet tags */
108 /* variables for ip and tcp reassembly */
109 void *header; /* pointer to packet header */
111 /* variables for hardware checksum */
112 int csum_flags; /* flags regarding checksum */
113 int csum_data; /* data field used by csum routines */
116 uint32_t fw_flags; /* flags for PF */
118 /* variables for PF processing */
119 uint16_t pf_tag; /* PF tag id */
120 uint8_t pf_routed; /* PF routing counter */
122 /* variables for ALTQ processing */
123 uint8_t ecn_af; /* address family for ECN */
124 uint32_t altq_qid; /* queue id */
128 * Description of external storage mapped into mbuf; valid only if M_EXT is set.
131 caddr_t ext_buf; /* start of buffer */
132 void (*ext_free)(void *);
133 u_int ext_size; /* size of buffer, for ext_free */
134 void (*ext_ref)(void *);
139 * The core of the mbuf object along with some shortcut defines for
140 * practical purposes.
146 struct pkthdr MH_pkthdr; /* M_PKTHDR set */
148 struct m_ext MH_ext; /* M_EXT set */
149 char MH_databuf[MHLEN];
152 char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */
155 #define m_next m_hdr.mh_next
156 #define m_len m_hdr.mh_len
157 #define m_data m_hdr.mh_data
158 #define m_type m_hdr.mh_type
159 #define m_flags m_hdr.mh_flags
160 #define m_nextpkt m_hdr.mh_nextpkt
161 #define m_pkthdr M_dat.MH.MH_pkthdr
162 #define m_ext M_dat.MH.MH_dat.MH_ext
163 #define m_pktdat M_dat.MH.MH_dat.MH_databuf
164 #define m_dat M_dat.M_databuf
167 * Code that uses m_act should be converted to use m_nextpkt
168 * instead; m_act is historical and deprecated.
170 #define m_act m_nextpkt
175 #define M_EXT 0x0001 /* has associated external storage */
176 #define M_PKTHDR 0x0002 /* start of record */
177 #define M_EOR 0x0004 /* end of record */
178 #define M_PROTO1 0x0008 /* protocol-specific */
179 #define M_PROTO2 0x0010 /* protocol-specific */
180 #define M_PROTO3 0x0020 /* protocol-specific */
181 #define M_PROTO4 0x0040 /* protocol-specific */
182 #define M_PROTO5 0x0080 /* protocol-specific */
185 * mbuf pkthdr flags (also stored in m_flags).
187 #define M_BCAST 0x0100 /* send/received as link-level broadcast */
188 #define M_MCAST 0x0200 /* send/received as link-level multicast */
189 #define M_FRAG 0x0400 /* packet is a fragment of a larger packet */
190 #define M_FIRSTFRAG 0x0800 /* packet is first fragment */
191 #define M_LASTFRAG 0x1000 /* packet is last fragment */
192 #define M_CLCACHE 0x2000 /* mbuf allocated from the cluster cache */
193 #define M_EXT_CLUSTER 0x4000 /* standard cluster else special */
194 #define M_PHCACHE 0x8000 /* mbuf allocated from the pkt header cache */
195 #define M_NOTIFICATION 0x10000 /* notification event */
198 * Flags copied when copying m_pkthdr.
200 #define M_COPYFLAGS (M_PKTHDR|M_EOR|M_PROTO1|M_PROTO1|M_PROTO2|M_PROTO3 | \
201 M_PROTO4|M_PROTO5|M_BCAST|M_MCAST|M_FRAG | \
202 M_FIRSTFRAG|M_LASTFRAG)
205 * Flags indicating hw checksum support and sw checksum requirements.
207 #define CSUM_IP 0x0001 /* will csum IP */
208 #define CSUM_TCP 0x0002 /* will csum TCP */
209 #define CSUM_UDP 0x0004 /* will csum UDP */
210 #define CSUM_IP_FRAGS 0x0008 /* will csum IP fragments */
211 #define CSUM_FRAGMENT 0x0010 /* will do IP fragmentation */
213 #define CSUM_IP_CHECKED 0x0100 /* did csum IP */
214 #define CSUM_IP_VALID 0x0200 /* ... the csum is valid */
215 #define CSUM_DATA_VALID 0x0400 /* csum_data field is valid */
216 #define CSUM_PSEUDO_HDR 0x0800 /* csum_data has pseudo hdr */
218 #define CSUM_DELAY_DATA (CSUM_TCP | CSUM_UDP)
219 #define CSUM_DELAY_IP (CSUM_IP) /* XXX add ipv6 here too? */
222 * Flags indicating PF processing status
224 #define PF_MBUF_GENERATED 0x00000001
225 #define PF_MBUF_TAGGED 0x00000002 /* pf_tag field is valid */
226 #define PF_MBUF_ROUTED 0x00000004 /* pf_routed field is valid */
227 #define PF_MBUF_TRANSLATE_LOCALHOST \
229 #define PF_MBUF_FRAGCACHE 0x00000010
230 #define ALTQ_MBUF_TAGGED 0x00000020 /* altq_qid is valid */
235 #define MT_FREE 0 /* should be on free list */
236 #define MT_DATA 1 /* dynamic (data) allocation */
237 #define MT_HEADER 2 /* packet header */
238 #define MT_SONAME 3 /* socket name */
239 #define MT_TAG 4 /* volatile metadata associated to pkts */
240 #define MT_CONTROL 5 /* extra-data protocol message */
241 #define MT_OOBDATA 6 /* expedited data */
242 #define MT_NTYPES 7 /* number of mbuf types for mbtypes[] */
245 * General mbuf allocator statistics structure.
248 u_long m_mbufs; /* mbufs obtained from page pool */
249 u_long m_clusters; /* clusters obtained from page pool */
250 u_long m_spare; /* spare field */
251 u_long m_clfree; /* free clusters */
252 u_long m_drops; /* times failed to find space */
253 u_long m_wait; /* times waited for space */
254 u_long m_drain; /* times drained protocols for space */
255 u_long m_mcfail; /* times m_copym failed */
256 u_long m_mpfail; /* times m_pullup failed */
257 u_long m_msize; /* length of an mbuf */
258 u_long m_mclbytes; /* length of an mbuf cluster */
259 u_long m_minclsize; /* min length of data to allocate a cluster */
260 u_long m_mlen; /* length of data in an mbuf */
261 u_long m_mhlen; /* length of data in a header mbuf */
265 * Flags specifying how an allocation should be made.
268 #define MB_DONTWAIT 0x4
269 #define MB_TRYWAIT 0x8
270 #define MB_WAIT MB_TRYWAIT
273 * Mbuf to Malloc Flag Conversion.
275 #define MBTOM(how) ((how) & MB_TRYWAIT ? M_WAITOK : M_NOWAIT)
278 * These are identifying numbers passed to the m_mballoc_wait function,
279 * allowing us to determine whether the call came from an MGETHDR or
286 * mbuf allocation/deallocation macros (YYY deprecated, too big):
288 * MGET(struct mbuf *m, int how, int type)
289 * allocates an mbuf and initializes it to contain internal data.
291 * MGETHDR(struct mbuf *m, int how, int type)
292 * allocates an mbuf and initializes it to contain a packet header
295 #define MGET(m, how, type) do { \
296 (m) = m_get((how), (type)); \
299 #define MGETHDR(m, how, type) do { \
300 (m) = m_gethdr((how), (type)); \
304 * MCLGET adds such clusters to a normal mbuf. The flag M_EXT is set upon
306 * Deprecated. Use m_getcl() or m_getl() instead.
308 #define MCLGET(m, how) do { \
309 m_mclget((m), (how)); \
313 * NB: M_COPY_PKTHDR is deprecated; use either M_MOVE_PKTHDR
317 * Move mbuf pkthdr from "from" to "to".
318 * from should have M_PKTHDR set, and to must be empty.
319 * from no longer has a pkthdr after this operation.
321 #define M_MOVE_PKTHDR(_to, _from) m_move_pkthdr((_to), (_from))
324 * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place
325 * an object of the specified size at the end of the mbuf, longword aligned.
327 #define M_ALIGN(m, len) do { \
328 (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \
332 * As above, for mbufs allocated with m_gethdr/MGETHDR
333 * or initialized by M_COPY_PKTHDR.
335 #define MH_ALIGN(m, len) do { \
336 (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \
340 * Check if we can write to an mbuf.
342 #define M_EXT_WRITABLE(m) (m_sharecount(m) == 1)
343 #define M_WRITABLE(m) (!((m)->m_flags & M_EXT) || M_EXT_WRITABLE(m))
346 * Check if the supplied mbuf has a packet header, or else panic.
348 #define M_ASSERTPKTHDR(m) \
349 KASSERT(m != NULL && m->m_flags & M_PKTHDR, \
350 ("%s: invalid mbuf or no mbuf packet header!", __func__))
353 * Compute the amount of space available before the current start of data.
354 * The M_EXT_WRITABLE() is a temporary, conservative safety measure: the burden
355 * of checking writability of the mbuf data area rests solely with the caller.
357 #define M_LEADINGSPACE(m) \
358 ((m)->m_flags & M_EXT ? \
359 (M_EXT_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \
360 (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \
361 (m)->m_data - (m)->m_dat)
364 * Compute the amount of space available after the end of data in an mbuf.
365 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
366 * of checking writability of the mbuf data area rests solely with the caller.
368 #define M_TRAILINGSPACE(m) \
369 ((m)->m_flags & M_EXT ? \
370 (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \
371 - ((m)->m_data + (m)->m_len) : 0) : \
372 &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len))
375 * Arrange to prepend space of size plen to mbuf m.
376 * If a new mbuf must be allocated, how specifies whether to wait.
377 * If how is MB_DONTWAIT and allocation fails, the original mbuf chain
378 * is freed and m is set to NULL.
380 #define M_PREPEND(m, plen, how) do { \
381 struct mbuf **_mmp = &(m); \
382 struct mbuf *_mm = *_mmp; \
383 int _mplen = (plen); \
384 int __mhow = (how); \
386 if (M_LEADINGSPACE(_mm) >= _mplen) { \
387 _mm->m_data -= _mplen; \
388 _mm->m_len += _mplen; \
390 _mm = m_prepend(_mm, _mplen, __mhow); \
391 if (_mm != NULL && _mm->m_flags & M_PKTHDR) \
392 _mm->m_pkthdr.len += _mplen; \
396 /* Length to m_copy to copy all. */
397 #define M_COPYALL 1000000000
399 /* Compatibility with 4.3 */
400 #define m_copy(m, o, l) m_copym((m), (o), (l), MB_DONTWAIT)
403 extern u_int m_clalloc_wid; /* mbuf cluster wait count */
404 extern u_int m_mballoc_wid; /* mbuf wait count */
405 extern int max_linkhdr; /* largest link-level header */
406 extern int max_protohdr; /* largest protocol header */
407 extern int max_hdr; /* largest link+protocol header */
408 extern int max_datalen; /* MHLEN - max_hdr */
409 extern struct mbstat mbstat;
410 extern int mbuf_wait; /* mbuf sleep time */
411 extern int nmbclusters;
416 void m_adj(struct mbuf *, int);
417 void m_cat(struct mbuf *, struct mbuf *);
418 u_int m_countm(struct mbuf *m, struct mbuf **lastm, u_int *mbcnt);
419 void m_copyback(struct mbuf *, int, int, caddr_t);
420 void m_copydata(const struct mbuf *, int, int, caddr_t);
421 struct mbuf *m_copym(const struct mbuf *, int, int, int);
422 struct mbuf *m_copypacket(struct mbuf *, int);
423 struct mbuf *m_defrag(struct mbuf *, int);
424 struct mbuf *m_defrag_nofree(struct mbuf *, int);
425 struct mbuf *m_devget(char *, int, int, struct ifnet *,
426 void (*copy)(volatile const void *, volatile void *, size_t));
427 struct mbuf *m_dup(struct mbuf *, int);
428 int m_dup_pkthdr(struct mbuf *, const struct mbuf *, int);
429 struct mbuf *m_free(struct mbuf *);
430 void m_freem(struct mbuf *);
431 struct mbuf *m_get(int, int);
432 struct mbuf *m_getc(int len, int how, int type);
433 struct mbuf *m_getcl(int how, short type, int flags);
434 struct mbuf *m_getclr(int, int);
435 struct mbuf *m_gethdr(int, int);
436 struct mbuf *m_getm(struct mbuf *, int, int, int);
437 struct mbuf *m_last(struct mbuf *m);
438 u_int m_lengthm(struct mbuf *m, struct mbuf **lastm);
439 void m_move_pkthdr(struct mbuf *, struct mbuf *);
440 struct mbuf *m_prepend(struct mbuf *, int, int);
441 void m_print(const struct mbuf *m);
442 struct mbuf *m_pulldown(struct mbuf *, int, int, int *);
443 struct mbuf *m_pullup(struct mbuf *, int);
444 struct mbuf *m_split(struct mbuf *, int, int);
445 struct mbuf *m_uiomove(struct uio *);
446 void m_mclget(struct mbuf *m, int how);
447 int m_sharecount(struct mbuf *m);
448 void m_chtype(struct mbuf *m, int type);
451 * Allocate the right type of mbuf for the desired total length.
453 static __inline struct mbuf *
454 m_getl(int len, int how, int type, int flags, int *psize)
459 if (len >= MINCLSIZE) {
460 m = m_getcl(how, type, flags);
462 } else if (flags & M_PKTHDR) {
463 m = m_gethdr(how, type);
466 m = m_get(how, type);
475 * Packets may have annotations attached by affixing a list
476 * of "packet tags" to the pkthdr structure. Packet tags are
477 * dynamically allocated semi-opaque data structures that have
478 * a fixed header (struct m_tag) that specifies the size of the
479 * memory block and a <cookie,type> pair that identifies it.
480 * The cookie is a 32-bit unique unsigned value used to identify
481 * a module or ABI. By convention this value is chose as the
482 * date+time that the module is created, expressed as the number of
483 * seconds since the epoch (e.g. using date -u +'%s'). The type value
484 * is an ABI/module-specific value that identifies a particular annotation
485 * and is private to the module. For compatibility with systems
486 * like openbsd that define packet tags w/o an ABI/module cookie,
487 * the value PACKET_ABI_COMPAT is used to implement m_tag_get and
488 * m_tag_find compatibility shim functions and several tag types are
489 * defined below. Users that do not require compatibility should use
490 * a private cookie value so that packet tag-related definitions
491 * can be maintained privately.
493 * Note that the packet tag returned by m_tag_allocate has the default
494 * memory alignment implemented by malloc. To reference private data
495 * one can use a construct like:
497 * struct m_tag *mtag = m_tag_allocate(...);
498 * struct foo *p = (struct foo *)(mtag+1);
500 * if the alignment of struct m_tag is sufficient for referencing members
501 * of struct foo. Otherwise it is necessary to embed struct m_tag within
502 * the private data structure to insure proper alignment; e.g.
508 * struct foo *p = (struct foo *) m_tag_allocate(...);
509 * struct m_tag *mtag = &p->tag;
512 #define PACKET_TAG_NONE 0 /* Nadda */
514 /* Packet tag for use with PACKET_ABI_COMPAT */
515 #define PACKET_TAG_IPSEC_IN_DONE 1 /* IPsec applied, in */
516 /* struct tdb_indent */
517 #define PACKET_TAG_IPSEC_OUT_DONE 2 /* IPsec applied, out */
518 /* struct tdb_indent */
519 #define PACKET_TAG_IPSEC_IN_CRYPTO_DONE 3 /* NIC IPsec crypto done */
520 /* struct tdb_indent, never added */
521 #define PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED 4 /* NIC IPsec crypto req'ed */
522 /* struct tdb_indent, never added */
523 #define PACKET_TAG_IPSEC_PENDING_TDB 5 /* Reminder to do IPsec */
524 /* struct tdb_indent, never added */
525 #define PACKET_TAG_ENCAP 6 /* Encap. processing */
526 /* struct ifnet *, the GIF interface */
527 #define PACKET_TAG_IPSEC_HISTORY 7 /* IPSEC history */
528 /* struct ipsec_history */
529 #define PACKET_TAG_IPV6_INPUT 8 /* IPV6 input processing */
531 #define PACKET_TAG_IPFW_DIVERT 9 /* divert info */
535 * As a temporary and low impact solution to replace the even uglier
536 * approach used so far in some parts of the network stack (which relies
537 * on global variables), packet tag-like annotations are stored in MT_TAG
538 * mbufs (or lookalikes) prepended to the actual mbuf chain.
542 * m_next = next buffer in chain.
544 * BE VERY CAREFUL not to pass these blocks to the mbuf handling routines.
546 #define _m_tag_id m_hdr.mh_flags
548 /* Packet tags used in the FreeBSD network stack */
549 #define PACKET_TAG_DUMMYNET 15 /* dummynet info */
550 /* struct dn_pkt as fake mbuf */
551 #define PACKET_TAG_IPFORWARD 18 /* ipforward info */
552 /* struct sockaddr_in * as m_data */
554 /* Packet tag routines */
555 struct m_tag *m_tag_alloc(u_int32_t, int, int, int);
556 void m_tag_free(struct m_tag *);
557 void m_tag_prepend(struct mbuf *, struct m_tag *);
558 void m_tag_unlink(struct mbuf *, struct m_tag *);
559 void m_tag_delete(struct mbuf *, struct m_tag *);
560 void m_tag_delete_chain(struct mbuf *);
561 struct m_tag *m_tag_locate(struct mbuf *, u_int32_t, int, struct m_tag *);
562 struct m_tag *m_tag_copy(struct m_tag *, int);
563 int m_tag_copy_chain(struct mbuf *, const struct mbuf *, int);
564 void m_tag_init(struct mbuf *);
565 struct m_tag *m_tag_first(struct mbuf *);
566 struct m_tag *m_tag_next(struct mbuf *, struct m_tag *);
568 /* these are for openbsd compatibility */
569 #define MTAG_ABI_COMPAT 0 /* compatibility ABI */
571 static __inline void *
572 m_tag_data(struct m_tag *tag)
574 return ((void *)(tag + 1));
577 static __inline struct m_tag *
578 m_tag_get(int type, int length, int wait)
580 return m_tag_alloc(MTAG_ABI_COMPAT, type, length, wait);
583 static __inline struct m_tag *
584 m_tag_find(struct mbuf *m, int type, struct m_tag *start)
586 return m_tag_locate(m, MTAG_ABI_COMPAT, type, start);
590 #endif /* !_SYS_MBUF_H_ */