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 1.29 2005/06/08 22:22:58 dillon Exp $
43 #include <sys/queue.h>
46 * Mbufs are of a single size, MSIZE (machine/param.h), which
47 * includes overhead. An mbuf may add a single "mbuf cluster" of size
48 * MCLBYTES (also in machine/param.h), which has no additional overhead
49 * and is used instead of the internal data area; this is done when
50 * at least MINCLSIZE of data must be stored.
52 #define MLEN (MSIZE - sizeof(struct m_hdr)) /* normal data len */
53 #define MHLEN (MLEN - sizeof(struct pkthdr)) /* data len w/pkthdr */
54 #define MINCLSIZE (MHLEN + 1) /* smallest amount to put in cluster */
55 #define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */
58 * Macros for type conversion:
59 * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type.
60 * mtocl(x) - convert pointer within cluster to cluster index #
61 * cltom(x) - convert cluster # to ptr to beginning of cluster
63 #define mtod(m, t) ((t)((m)->m_data))
66 * Header present at the beginning of every mbuf.
69 struct mbuf *mh_next; /* next buffer in chain */
70 struct mbuf *mh_nextpkt; /* next chain in queue/record */
71 caddr_t mh_data; /* location of data */
72 int mh_len; /* amount of data in this mbuf */
73 short mh_type; /* type of data in this mbuf */
74 short mh_flags; /* flags; see below */
78 * Packet tag structure (see below for details).
81 SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */
82 u_int16_t m_tag_id; /* Tag ID */
83 u_int16_t m_tag_len; /* Length of data */
84 u_int32_t m_tag_cookie; /* ABI/Module ID */
87 SLIST_HEAD(packet_tags, m_tag);
90 * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set.
92 * Be careful: The fields have been carefully ordered to avoid hidden padding.
93 * Keep this in mind, when adding or removing fields!
96 struct ifnet *rcvif; /* rcv interface */
97 int len; /* total packet length */
98 struct packet_tags tags; /* list of packet tags */
100 /* variables for ip and tcp reassembly */
101 void *header; /* pointer to packet header */
103 /* variables for hardware checksum */
104 int csum_flags; /* flags regarding checksum */
105 int csum_data; /* data field used by csum routines */
108 uint32_t fw_flags; /* flags for PF */
110 /* variables for PF processing */
111 uint16_t pf_tag; /* PF tag id */
112 uint8_t pf_routed; /* PF routing counter */
114 /* variables for ALTQ processing */
115 uint8_t ecn_af; /* address family for ECN */
116 uint32_t altq_qid; /* queue id */
120 * Description of external storage mapped into mbuf; valid only if M_EXT is set.
123 caddr_t ext_buf; /* start of buffer */
124 void (*ext_free)(void *);
125 u_int ext_size; /* size of buffer, for ext_free */
126 void (*ext_ref)(void *);
131 * The core of the mbuf object along with some shortcut defines for
132 * practical purposes.
138 struct pkthdr MH_pkthdr; /* M_PKTHDR set */
140 struct m_ext MH_ext; /* M_EXT set */
141 char MH_databuf[MHLEN];
144 char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */
147 #define m_next m_hdr.mh_next
148 #define m_len m_hdr.mh_len
149 #define m_data m_hdr.mh_data
150 #define m_type m_hdr.mh_type
151 #define m_flags m_hdr.mh_flags
152 #define m_nextpkt m_hdr.mh_nextpkt
153 #define m_pkthdr M_dat.MH.MH_pkthdr
154 #define m_ext M_dat.MH.MH_dat.MH_ext
155 #define m_pktdat M_dat.MH.MH_dat.MH_databuf
156 #define m_dat M_dat.M_databuf
159 * Code that uses m_act should be converted to use m_nextpkt
160 * instead; m_act is historical and deprecated.
162 #define m_act m_nextpkt
167 #define M_EXT 0x0001 /* has associated external storage */
168 #define M_PKTHDR 0x0002 /* start of record */
169 #define M_EOR 0x0004 /* end of record */
170 #define M_PROTO1 0x0008 /* protocol-specific */
171 #define M_PROTO2 0x0010 /* protocol-specific */
172 #define M_PROTO3 0x0020 /* protocol-specific */
173 #define M_PROTO4 0x0040 /* protocol-specific */
174 #define M_PROTO5 0x0080 /* protocol-specific */
177 * mbuf pkthdr flags (also stored in m_flags).
179 #define M_BCAST 0x0100 /* send/received as link-level broadcast */
180 #define M_MCAST 0x0200 /* send/received as link-level multicast */
181 #define M_FRAG 0x0400 /* packet is a fragment of a larger packet */
182 #define M_FIRSTFRAG 0x0800 /* packet is first fragment */
183 #define M_LASTFRAG 0x1000 /* packet is last fragment */
184 #define M_CLCACHE 0x2000 /* mbuf allocated from the cluster cache */
185 #define M_EXT_CLUSTER 0x4000 /* standard cluster else special */
186 #define M_PHCACHE 0x8000 /* mbuf allocated from the pkt header cache */
189 * Flags copied when copying m_pkthdr.
191 #define M_COPYFLAGS (M_PKTHDR|M_EOR|M_PROTO1|M_PROTO1|M_PROTO2|M_PROTO3 | \
192 M_PROTO4|M_PROTO5|M_BCAST|M_MCAST|M_FRAG | \
193 M_FIRSTFRAG|M_LASTFRAG)
196 * Flags indicating hw checksum support and sw checksum requirements.
198 #define CSUM_IP 0x0001 /* will csum IP */
199 #define CSUM_TCP 0x0002 /* will csum TCP */
200 #define CSUM_UDP 0x0004 /* will csum UDP */
201 #define CSUM_IP_FRAGS 0x0008 /* will csum IP fragments */
202 #define CSUM_FRAGMENT 0x0010 /* will do IP fragmentation */
204 #define CSUM_IP_CHECKED 0x0100 /* did csum IP */
205 #define CSUM_IP_VALID 0x0200 /* ... the csum is valid */
206 #define CSUM_DATA_VALID 0x0400 /* csum_data field is valid */
207 #define CSUM_PSEUDO_HDR 0x0800 /* csum_data has pseudo hdr */
209 #define CSUM_DELAY_DATA (CSUM_TCP | CSUM_UDP)
210 #define CSUM_DELAY_IP (CSUM_IP) /* XXX add ipv6 here too? */
213 * Flags indicating PF processing status
215 #define PF_MBUF_GENERATED 0x00000001
216 #define PF_MBUF_TAGGED 0x00000002 /* pf_tag field is valid */
217 #define PF_MBUF_ROUTED 0x00000004 /* pf_routed field is valid */
218 #define PF_MBUF_TRANSLATE_LOCALHOST \
220 #define PF_MBUF_FRAGCACHE 0x00000010
221 #define ALTQ_MBUF_TAGGED 0x00000020 /* altq_qid is valid */
226 #define MT_FREE 0 /* should be on free list */
227 #define MT_DATA 1 /* dynamic (data) allocation */
228 #define MT_HEADER 2 /* packet header */
229 #define MT_SONAME 3 /* socket name */
230 #define MT_TAG 4 /* volatile metadata associated to pkts */
231 #define MT_CONTROL 5 /* extra-data protocol message */
232 #define MT_OOBDATA 6 /* expedited data */
233 #define MT_NTYPES 7 /* number of mbuf types for mbtypes[] */
236 * General mbuf allocator statistics structure.
239 u_long m_mbufs; /* mbufs obtained from page pool */
240 u_long m_clusters; /* clusters obtained from page pool */
241 u_long m_spare; /* spare field */
242 u_long m_clfree; /* free clusters */
243 u_long m_drops; /* times failed to find space */
244 u_long m_wait; /* times waited for space */
245 u_long m_drain; /* times drained protocols for space */
246 u_long m_mcfail; /* times m_copym failed */
247 u_long m_mpfail; /* times m_pullup failed */
248 u_long m_msize; /* length of an mbuf */
249 u_long m_mclbytes; /* length of an mbuf cluster */
250 u_long m_minclsize; /* min length of data to allocate a cluster */
251 u_long m_mlen; /* length of data in an mbuf */
252 u_long m_mhlen; /* length of data in a header mbuf */
256 * Flags specifying how an allocation should be made.
259 #define MB_DONTWAIT 0x4
260 #define MB_TRYWAIT 0x8
261 #define MB_WAIT MB_TRYWAIT
264 * Mbuf to Malloc Flag Conversion.
266 #define MBTOM(how) ((how) & MB_TRYWAIT ? M_WAITOK : M_NOWAIT)
269 * These are identifying numbers passed to the m_mballoc_wait function,
270 * allowing us to determine whether the call came from an MGETHDR or
277 * Wake up the next instance (if any) of m_mballoc_wait() which is
278 * waiting for an mbuf to be freed. This should be called at splimp().
280 * XXX: If there is another free mbuf, this routine will be called [again]
281 * from the m_mballoc_wait routine in order to wake another sleep instance.
283 #define MMBWAKEUP() do { \
284 if (m_mballoc_wid) { \
286 wakeup_one(&m_mballoc_wid); \
291 * Same as above, but for mbuf cluster(s).
293 #define MCLWAKEUP() do { \
294 if (m_clalloc_wid) { \
296 wakeup_one(&m_clalloc_wid); \
301 * mbuf utility macros:
304 * prevents a section of code from from being interrupted by network
307 #define MBUFLOCK(code) do { \
308 int _ms = splimp(); \
315 * mbuf allocation/deallocation macros (YYY deprecated, too big):
317 * MGET(struct mbuf *m, int how, int type)
318 * allocates an mbuf and initializes it to contain internal data.
320 * MGETHDR(struct mbuf *m, int how, int type)
321 * allocates an mbuf and initializes it to contain a packet header
324 #define MGET(m, how, type) do { \
325 (m) = m_get((how), (type)); \
328 #define MGETHDR(m, how, type) do { \
329 (m) = m_gethdr((how), (type)); \
333 * MCLGET adds such clusters to a normal mbuf. The flag M_EXT is set upon
335 * Deprecated. Use m_getcl() or m_getl() instead.
337 #define MCLGET(m, how) do { \
338 m_mclget((m), (how)); \
342 * NB: M_COPY_PKTHDR is deprecated; use either M_MOVE_PKTHDR
346 * Move mbuf pkthdr from "from" to "to".
347 * from should have M_PKTHDR set, and to must be empty.
348 * from no longer has a pkthdr after this operation.
350 #define M_MOVE_PKTHDR(_to, _from) m_move_pkthdr((_to), (_from))
353 * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place
354 * an object of the specified size at the end of the mbuf, longword aligned.
356 #define M_ALIGN(m, len) do { \
357 (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \
361 * As above, for mbufs allocated with m_gethdr/MGETHDR
362 * or initialized by M_COPY_PKTHDR.
364 #define MH_ALIGN(m, len) do { \
365 (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \
369 * Check if we can write to an mbuf.
371 #define M_EXT_WRITABLE(m) (m_sharecount(m) == 1)
372 #define M_WRITABLE(m) (!((m)->m_flags & M_EXT) || M_EXT_WRITABLE(m))
375 * Check if the supplied mbuf has a packet header, or else panic.
377 #define M_ASSERTPKTHDR(m) \
378 KASSERT(m != NULL && m->m_flags & M_PKTHDR, \
379 ("%s: invalid mbuf or no mbuf packet header!", __func__))
382 * Compute the amount of space available before the current start of data.
383 * The M_EXT_WRITABLE() is a temporary, conservative safety measure: the burden
384 * of checking writability of the mbuf data area rests solely with the caller.
386 #define M_LEADINGSPACE(m) \
387 ((m)->m_flags & M_EXT ? \
388 (M_EXT_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \
389 (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \
390 (m)->m_data - (m)->m_dat)
393 * Compute the amount of space available after the end of data in an mbuf.
394 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
395 * of checking writability of the mbuf data area rests solely with the caller.
397 #define M_TRAILINGSPACE(m) \
398 ((m)->m_flags & M_EXT ? \
399 (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \
400 - ((m)->m_data + (m)->m_len) : 0) : \
401 &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len))
404 * Arrange to prepend space of size plen to mbuf m.
405 * If a new mbuf must be allocated, how specifies whether to wait.
406 * If how is MB_DONTWAIT and allocation fails, the original mbuf chain
407 * is freed and m is set to NULL.
409 #define M_PREPEND(m, plen, how) do { \
410 struct mbuf **_mmp = &(m); \
411 struct mbuf *_mm = *_mmp; \
412 int _mplen = (plen); \
413 int __mhow = (how); \
415 if (M_LEADINGSPACE(_mm) >= _mplen) { \
416 _mm->m_data -= _mplen; \
417 _mm->m_len += _mplen; \
419 _mm = m_prepend(_mm, _mplen, __mhow); \
420 if (_mm != NULL && _mm->m_flags & M_PKTHDR) \
421 _mm->m_pkthdr.len += _mplen; \
425 /* Length to m_copy to copy all. */
426 #define M_COPYALL 1000000000
428 /* Compatibility with 4.3 */
429 #define m_copy(m, o, l) m_copym((m), (o), (l), MB_DONTWAIT)
432 extern u_int m_clalloc_wid; /* mbuf cluster wait count */
433 extern u_int m_mballoc_wid; /* mbuf wait count */
434 extern int max_linkhdr; /* largest link-level header */
435 extern int max_protohdr; /* largest protocol header */
436 extern int max_hdr; /* largest link+protocol header */
437 extern int max_datalen; /* MHLEN - max_hdr */
438 extern struct mbstat mbstat;
439 extern int mbuf_wait; /* mbuf sleep time */
440 extern int nmbclusters;
445 void m_adj(struct mbuf *, int);
446 void m_cat(struct mbuf *, struct mbuf *);
447 u_int m_countm(struct mbuf *m, struct mbuf **lastm, u_int *mbcnt);
448 void m_copyback(struct mbuf *, int, int, caddr_t);
449 void m_copydata(const struct mbuf *, int, int, caddr_t);
450 struct mbuf *m_copym(const struct mbuf *, int, int, int);
451 struct mbuf *m_copypacket(struct mbuf *, int);
452 struct mbuf *m_defrag(struct mbuf *, int);
453 struct mbuf *m_defrag_nofree(struct mbuf *, int);
454 struct mbuf *m_devget(char *, int, int, struct ifnet *,
455 void (*copy)(volatile const void *, volatile void *, size_t));
456 struct mbuf *m_dup(struct mbuf *, int);
457 int m_dup_pkthdr(struct mbuf *, const struct mbuf *, int);
458 struct mbuf *m_free(struct mbuf *);
459 void m_freem(struct mbuf *);
460 struct mbuf *m_get(int, int);
461 struct mbuf *m_getc(int len, int how, int type);
462 struct mbuf *m_getcl(int how, short type, int flags);
463 struct mbuf *m_getclr(int, int);
464 struct mbuf *m_gethdr(int, int);
465 struct mbuf *m_getm(struct mbuf *, int, int, int);
466 struct mbuf *m_last(struct mbuf *m);
467 u_int m_lengthm(struct mbuf *m, struct mbuf **lastm);
468 void m_move_pkthdr(struct mbuf *, struct mbuf *);
469 struct mbuf *m_prepend(struct mbuf *, int, int);
470 void m_print(const struct mbuf *m);
471 struct mbuf *m_pulldown(struct mbuf *, int, int, int *);
472 struct mbuf *m_pullup(struct mbuf *, int);
473 struct mbuf *m_split(struct mbuf *, int, int);
474 struct mbuf *m_uiomove(struct uio *);
475 void m_mclget(struct mbuf *m, int how);
476 int m_sharecount(struct mbuf *m);
477 void m_chtype(struct mbuf *m, int type);
480 * Allocate the right type of mbuf for the desired total length.
482 static __inline struct mbuf *
483 m_getl(int len, int how, int type, int flags, int *psize)
488 if (len >= MINCLSIZE) {
489 m = m_getcl(how, type, flags);
491 } else if (flags & M_PKTHDR) {
492 m = m_gethdr(how, type);
495 m = m_get(how, type);
504 * Packets may have annotations attached by affixing a list
505 * of "packet tags" to the pkthdr structure. Packet tags are
506 * dynamically allocated semi-opaque data structures that have
507 * a fixed header (struct m_tag) that specifies the size of the
508 * memory block and a <cookie,type> pair that identifies it.
509 * The cookie is a 32-bit unique unsigned value used to identify
510 * a module or ABI. By convention this value is chose as the
511 * date+time that the module is created, expressed as the number of
512 * seconds since the epoch (e.g. using date -u +'%s'). The type value
513 * is an ABI/module-specific value that identifies a particular annotation
514 * and is private to the module. For compatibility with systems
515 * like openbsd that define packet tags w/o an ABI/module cookie,
516 * the value PACKET_ABI_COMPAT is used to implement m_tag_get and
517 * m_tag_find compatibility shim functions and several tag types are
518 * defined below. Users that do not require compatibility should use
519 * a private cookie value so that packet tag-related definitions
520 * can be maintained privately.
522 * Note that the packet tag returned by m_tag_allocate has the default
523 * memory alignment implemented by malloc. To reference private data
524 * one can use a construct like:
526 * struct m_tag *mtag = m_tag_allocate(...);
527 * struct foo *p = (struct foo *)(mtag+1);
529 * if the alignment of struct m_tag is sufficient for referencing members
530 * of struct foo. Otherwise it is necessary to embed struct m_tag within
531 * the private data structure to insure proper alignment; e.g.
537 * struct foo *p = (struct foo *) m_tag_allocate(...);
538 * struct m_tag *mtag = &p->tag;
541 #define PACKET_TAG_NONE 0 /* Nadda */
543 /* Packet tag for use with PACKET_ABI_COMPAT */
544 #define PACKET_TAG_IPSEC_IN_DONE 1 /* IPsec applied, in */
545 /* struct tdb_indent */
546 #define PACKET_TAG_IPSEC_OUT_DONE 2 /* IPsec applied, out */
547 /* struct tdb_indent */
548 #define PACKET_TAG_IPSEC_IN_CRYPTO_DONE 3 /* NIC IPsec crypto done */
549 /* struct tdb_indent, never added */
550 #define PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED 4 /* NIC IPsec crypto req'ed */
551 /* struct tdb_indent, never added */
552 #define PACKET_TAG_IPSEC_PENDING_TDB 5 /* Reminder to do IPsec */
553 /* struct tdb_indent, never added */
554 #define PACKET_TAG_ENCAP 6 /* Encap. processing */
555 /* struct ifnet *, the GIF interface */
556 #define PACKET_TAG_IPSEC_HISTORY 7 /* IPSEC history */
557 /* struct ipsec_history */
558 #define PACKET_TAG_IPV6_INPUT 8 /* IPV6 input processing */
560 #define PACKET_TAG_IPFW_DIVERT 9 /* divert info */
564 * As a temporary and low impact solution to replace the even uglier
565 * approach used so far in some parts of the network stack (which relies
566 * on global variables), packet tag-like annotations are stored in MT_TAG
567 * mbufs (or lookalikes) prepended to the actual mbuf chain.
571 * m_next = next buffer in chain.
573 * BE VERY CAREFUL not to pass these blocks to the mbuf handling routines.
575 #define _m_tag_id m_hdr.mh_flags
577 /* Packet tags used in the FreeBSD network stack */
578 #define PACKET_TAG_DUMMYNET 15 /* dummynet info */
579 /* struct dn_pkt as fake mbuf */
580 #define PACKET_TAG_IPFORWARD 18 /* ipforward info */
581 /* struct sockaddr_in * as m_data */
583 /* Packet tag routines */
584 struct m_tag *m_tag_alloc(u_int32_t, int, int, int);
585 void m_tag_free(struct m_tag *);
586 void m_tag_prepend(struct mbuf *, struct m_tag *);
587 void m_tag_unlink(struct mbuf *, struct m_tag *);
588 void m_tag_delete(struct mbuf *, struct m_tag *);
589 void m_tag_delete_chain(struct mbuf *);
590 struct m_tag *m_tag_locate(struct mbuf *, u_int32_t, int, struct m_tag *);
591 struct m_tag *m_tag_copy(struct m_tag *, int);
592 int m_tag_copy_chain(struct mbuf *, const struct mbuf *, int);
593 void m_tag_init(struct mbuf *);
594 struct m_tag *m_tag_first(struct mbuf *);
595 struct m_tag *m_tag_next(struct mbuf *, struct m_tag *);
597 /* these are for openbsd compatibility */
598 #define MTAG_ABI_COMPAT 0 /* compatibility ABI */
600 static __inline struct m_tag *
601 m_tag_get(int type, int length, int wait)
603 return m_tag_alloc(MTAG_ABI_COMPAT, type, length, wait);
606 static __inline struct m_tag *
607 m_tag_find(struct mbuf *m, int type, struct m_tag *start)
609 return m_tag_locate(m, MTAG_ABI_COMPAT, type, start);
613 #endif /* !_SYS_MBUF_H_ */