blob: f3717407e14aafc56399b2dc28eabc8784b80145 [file] [log] [blame]
tuexen829d1a72012-05-23 09:39:05 +00001/*-
2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California.
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of the University nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 * SUCH DAMAGE.
29 *
30 */
31
tuexenbca1dae2011-11-01 23:02:02 +000032#ifndef _USER_MBUF_H_
33#define _USER_MBUF_H_
34
35/* __Userspace__ header file for mbufs */
36#include <stdio.h>
37#if !defined(SCTP_SIMPLE_ALLOCATOR)
38#include "umem.h"
39#endif
40#include "user_malloc.h"
41#include "netinet/sctp_os_userspace.h"
tuexenbca1dae2011-11-01 23:02:02 +000042
tuexenbca1dae2011-11-01 23:02:02 +000043#define USING_MBUF_CONSTRUCTOR 0
44
45/* For Linux */
46#ifndef MSIZE
47#define MSIZE 256
48/* #define MSIZE 1024 */
49#endif
50#ifndef MCLBYTES
51#define MCLBYTES 2048
52#endif
53
54struct mbuf * m_gethdr(int how, short type);
55struct mbuf * m_get(int how, short type);
56struct mbuf * m_free(struct mbuf *m);
57void m_clget(struct mbuf *m, int how);
58
59
60/* mbuf initialization function */
61void mbuf_init(void *);
62
63#define M_MOVE_PKTHDR(to, from) m_move_pkthdr((to), (from))
64#define MGET(m, how, type) ((m) = m_get((how), (type)))
65#define MGETHDR(m, how, type) ((m) = m_gethdr((how), (type)))
66#define MCLGET(m, how) m_clget((m), (how))
67
68
69#define M_HDR_PAD ((sizeof(intptr_t)==4) ? 2 : 6) /* modified for __Userspace__ */
70
71/* Length to m_copy to copy all. */
72#define M_COPYALL 1000000000
73
tuexenc7108432012-05-21 18:53:50 +000074/* umem_cache_t is defined in user_include/umem.h as
tuexenbca1dae2011-11-01 23:02:02 +000075 * typedef struct umem_cache umem_cache_t;
76 * Note:umem_zone_t is a pointer.
77 */
tuexenc7108432012-05-21 18:53:50 +000078#if defined(SCTP_SIMPLE_ALLOCATOR)
tuexenbca1dae2011-11-01 23:02:02 +000079typedef size_t sctp_zone_t;
80#else
81typedef umem_cache_t *sctp_zone_t;
82#endif
83
84extern sctp_zone_t zone_mbuf;
85extern sctp_zone_t zone_clust;
86extern sctp_zone_t zone_ext_refcnt;
87
88/*-
89 * Macros for type conversion:
90 * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type.
91 * dtom(x) -- Convert data pointer within mbuf to mbuf pointer (XXX).
92 */
93#define mtod(m, t) ((t)((m)->m_data))
94#define dtom(x) ((struct mbuf *)((intptr_t)(x) & ~(MSIZE-1)))
95
tuexenbca1dae2011-11-01 23:02:02 +000096struct mb_args {
97 int flags; /* Flags for mbuf being allocated */
98 short type; /* Type of mbuf being allocated */
99};
100
101struct clust_args {
t00fcxen9b332c22013-02-28 19:01:36 +0000102 struct mbuf * parent_mbuf;
tuexenbca1dae2011-11-01 23:02:02 +0000103};
104
tuexenbca1dae2011-11-01 23:02:02 +0000105struct mbuf * m_split(struct mbuf *, int, int);
106void m_cat(struct mbuf *m, struct mbuf *n);
107void m_adj(struct mbuf *, int);
108void mb_free_ext(struct mbuf *);
109void m_freem(struct mbuf *);
110struct m_tag *m_tag_alloc(u_int32_t, int, int, int);
111struct mbuf *m_copym(struct mbuf *, int, int, int);
tuexena89a6762011-11-03 15:57:29 +0000112void m_copyback(struct mbuf *, int, int, caddr_t);
tuexenbca1dae2011-11-01 23:02:02 +0000113struct mbuf *m_pullup(struct mbuf *, int);
tuexencd78deb2011-12-23 18:06:30 +0000114struct mbuf *m_pulldown(struct mbuf *, int off, int len, int *offp);
tuexenbca1dae2011-11-01 23:02:02 +0000115int m_dup_pkthdr(struct mbuf *, struct mbuf *, int);
116struct m_tag *m_tag_copy(struct m_tag *, int);
117int m_tag_copy_chain(struct mbuf *, struct mbuf *, int);
118struct mbuf *m_prepend(struct mbuf *, int, int);
119void m_copydata(const struct mbuf *, int, int, caddr_t);
120
121#define MBUF_MEM_NAME "mbuf"
122#define MBUF_CLUSTER_MEM_NAME "mbuf_cluster"
123#define MBUF_EXTREFCNT_MEM_NAME "mbuf_ext_refcnt"
124
125#define MT_NOINIT 255 /* Not a type but a flag to allocate
126 a non-initialized mbuf */
tuexenbca1dae2011-11-01 23:02:02 +0000127
128/*
129 * General mbuf allocator statistics structure.
130 * __Userspace__ mbstat may be useful for gathering statistics.
131 * In the kernel many of these statistics are no longer used as
132 * they track allocator statistics through kernel UMA's built in statistics mechanism.
133 */
134struct mbstat {
135 u_long m_mbufs; /* XXX */
136 u_long m_mclusts; /* XXX */
137
138 u_long m_drain; /* times drained protocols for space */
139 u_long m_mcfail; /* XXX: times m_copym failed */
140 u_long m_mpfail; /* XXX: times m_pullup failed */
141 u_long m_msize; /* length of an mbuf */
142 u_long m_mclbytes; /* length of an mbuf cluster */
143 u_long m_minclsize; /* min length of data to allocate a cluster */
144 u_long m_mlen; /* length of data in an mbuf */
145 u_long m_mhlen; /* length of data in a header mbuf */
146
147 /* Number of mbtypes (gives # elems in mbtypes[] array: */
148 short m_numtypes;
149
150 /* XXX: Sendfile stats should eventually move to their own struct */
151 u_long sf_iocnt; /* times sendfile had to do disk I/O */
152 u_long sf_allocfail; /* times sfbuf allocation failed */
153 u_long sf_allocwait; /* times sfbuf allocation had to wait */
154};
155
156
157/*
158 * Mbufs are of a single size, MSIZE (sys/param.h), which includes overhead.
159 * An mbuf may add a single "mbuf cluster" of size MCLBYTES (also in
160 * sys/param.h), which has no additional overhead and is used instead of the
161 * internal data area; this is done when at least MINCLSIZE of data must be
162 * stored. Additionally, it is possible to allocate a separate buffer
163 * externally and attach it to the mbuf in a way similar to that of mbuf
164 * clusters.
165 */
t00fcxen702e2862013-12-20 18:51:38 +0000166#define MLEN ((int)(MSIZE - sizeof(struct m_hdr))) /* normal data len */
167#define MHLEN ((int)(MLEN - sizeof(struct pkthdr))) /* data len w/pkthdr */
t00fcxen95bfa382014-02-20 18:56:07 +0000168#define MINCLSIZE ((int)(MHLEN + 1)) /* smallest amount to put in cluster */
tuexenbca1dae2011-11-01 23:02:02 +0000169#define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */
170
171
172/*
173 * Header present at the beginning of every mbuf.
174 */
175struct m_hdr {
176 struct mbuf *mh_next; /* next buffer in chain */
177 struct mbuf *mh_nextpkt; /* next chain in queue/record */
178 caddr_t mh_data; /* location of data */
179 int mh_len; /* amount of data in this mbuf */
180 int mh_flags; /* flags; see below */
181 short mh_type; /* type of data in this mbuf */
182 uint8_t pad[M_HDR_PAD];/* word align */
183};
184
185/*
186 * Packet tag structure (see below for details).
187 */
188struct m_tag {
189 SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */
190 u_int16_t m_tag_id; /* Tag ID */
191 u_int16_t m_tag_len; /* Length of data */
192 u_int32_t m_tag_cookie; /* ABI/Module ID */
193 void (*m_tag_free)(struct m_tag *);
194};
195
196/*
197 * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set.
198 */
199struct pkthdr {
200 struct ifnet *rcvif; /* rcv interface */
201 /* variables for ip and tcp reassembly */
202 void *header; /* pointer to packet header */
203 int len; /* total packet length */
204 /* variables for hardware checksum */
205 int csum_flags; /* flags regarding checksum */
206 int csum_data; /* data field used by csum routines */
207 u_int16_t tso_segsz; /* TSO segment size */
208 u_int16_t ether_vtag; /* Ethernet 802.1p+q vlan tag */
209 SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */
210};
211
212/*
213 * Description of external storage mapped into mbuf; valid only if M_EXT is
214 * set.
215 */
216struct m_ext {
217 caddr_t ext_buf; /* start of buffer */
218 void (*ext_free) /* free routine if not the usual */
219 (void *, void *);
220 void *ext_args; /* optional argument pointer */
221 u_int ext_size; /* size of buffer, for ext_free */
222 volatile u_int *ref_cnt; /* pointer to ref count info */
223 int ext_type; /* type of external storage */
224};
225
226
227/*
228 * The core of the mbuf object along with some shortcut defined for practical
229 * purposes.
230 */
231struct mbuf {
232 struct m_hdr m_hdr;
233 union {
234 struct {
235 struct pkthdr MH_pkthdr; /* M_PKTHDR set */
236 union {
237 struct m_ext MH_ext; /* M_EXT set */
238 char MH_databuf[MHLEN];
239 } MH_dat;
240 } MH;
241 char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */
242 } M_dat;
243};
244
245#define m_next m_hdr.mh_next
246#define m_len m_hdr.mh_len
247#define m_data m_hdr.mh_data
248#define m_type m_hdr.mh_type
249#define m_flags m_hdr.mh_flags
250#define m_nextpkt m_hdr.mh_nextpkt
251#define m_act m_nextpkt
252#define m_pkthdr M_dat.MH.MH_pkthdr
253#define m_ext M_dat.MH.MH_dat.MH_ext
254#define m_pktdat M_dat.MH.MH_dat.MH_databuf
255#define m_dat M_dat.M_databuf
256
257
258/*
259 * mbuf flags.
260 */
261#define M_EXT 0x0001 /* has associated external storage */
262#define M_PKTHDR 0x0002 /* start of record */
263#define M_EOR 0x0004 /* end of record */
264#define M_RDONLY 0x0008 /* associated data is marked read-only */
265#define M_PROTO1 0x0010 /* protocol-specific */
266#define M_PROTO2 0x0020 /* protocol-specific */
267#define M_PROTO3 0x0040 /* protocol-specific */
268#define M_PROTO4 0x0080 /* protocol-specific */
269#define M_PROTO5 0x0100 /* protocol-specific */
tuexenbca1dae2011-11-01 23:02:02 +0000270#define M_SKIP_FIREWALL 0x4000 /* skip firewall processing */
271#define M_FREELIST 0x8000 /* mbuf is on the free list */
272
273
274/*
275 * Flags copied when copying m_pkthdr.
276 */
277#define M_COPYFLAGS (M_PKTHDR|M_EOR|M_RDONLY|M_PROTO1|M_PROTO1|M_PROTO2|\
278 M_PROTO3|M_PROTO4|M_PROTO5|M_SKIP_FIREWALL|\
279 M_BCAST|M_MCAST|M_FRAG|M_FIRSTFRAG|M_LASTFRAG|\
280 M_VLANTAG|M_PROMISC)
281
282
283/*
284 * mbuf pkthdr flags (also stored in m_flags).
285 */
286#define M_BCAST 0x0200 /* send/received as link-level broadcast */
287#define M_MCAST 0x0400 /* send/received as link-level multicast */
288#define M_FRAG 0x0800 /* packet is a fragment of a larger packet */
289#define M_FIRSTFRAG 0x1000 /* packet is first fragment */
290#define M_LASTFRAG 0x2000 /* packet is last fragment */
291#define M_VLANTAG 0x10000 /* ether_vtag is valid */
292#define M_PROMISC 0x20000 /* packet was not for us */
293#define M_NOFREE 0x40000 /* do not free mbuf - it is embedded in the cluster */
294
295
296/*
297 * External buffer types: identify ext_buf type.
298 */
299#define EXT_CLUSTER 1 /* mbuf cluster */
300#define EXT_SFBUF 2 /* sendfile(2)'s sf_bufs */
301#define EXT_JUMBOP 3 /* jumbo cluster 4096 bytes */
302#define EXT_JUMBO9 4 /* jumbo cluster 9216 bytes */
303#define EXT_JUMBO16 5 /* jumbo cluster 16184 bytes */
304#define EXT_PACKET 6 /* mbuf+cluster from packet zone */
305#define EXT_MBUF 7 /* external mbuf reference (M_IOVEC) */
306#define EXT_NET_DRV 100 /* custom ext_buf provided by net driver(s) */
307#define EXT_MOD_TYPE 200 /* custom module's ext_buf type */
308#define EXT_DISPOSABLE 300 /* can throw this buffer away w/page flipping */
309#define EXT_EXTREF 400 /* has externally maintained ref_cnt ptr */
310
311
312/*
313 * mbuf types.
314 */
315#define MT_NOTMBUF 0 /* USED INTERNALLY ONLY! Object is not mbuf */
316#define MT_DATA 1 /* dynamic (data) allocation */
317#define MT_HEADER MT_DATA /* packet header, use M_PKTHDR instead */
318#define MT_SONAME 8 /* socket name */
319#define MT_CONTROL 14 /* extra-data protocol message */
320#define MT_OOBDATA 15 /* expedited data */
321#define MT_NTYPES 16 /* number of mbuf types for mbtypes[] */
322
323#define MT_NOINIT 255 /* Not a type but a flag to allocate
324 a non-initialized mbuf */
325
tuexenbca1dae2011-11-01 23:02:02 +0000326/*
327 * __Userspace__ flags like M_NOWAIT are defined in malloc.h
328 * Flags like these are used in functions like uma_zalloc()
329 * but don't have an equivalent in userland umem
330 * Flags specifying how an allocation should be made.
331 *
332 * The flag to use is as follows:
333 * - M_DONTWAIT or M_NOWAIT from an interrupt handler to not block allocation.
334 * - M_WAIT or M_WAITOK or M_TRYWAIT from wherever it is safe to block.
335 *
336 * M_DONTWAIT/M_NOWAIT means that we will not block the thread explicitly and
337 * if we cannot allocate immediately we may return NULL, whereas
338 * M_WAIT/M_WAITOK/M_TRYWAIT means that if we cannot allocate resources we
339 * will block until they are available, and thus never return NULL.
340 *
341 * XXX Eventually just phase this out to use M_WAITOK/M_NOWAIT.
342 */
343#define MBTOM(how) (how)
tuexenbca1dae2011-11-01 23:02:02 +0000344
345void m_tag_delete(struct mbuf *, struct m_tag *);
346void m_tag_delete_chain(struct mbuf *, struct m_tag *);
347void m_move_pkthdr(struct mbuf *, struct mbuf *);
348void m_tag_free_default(struct m_tag *);
349
350extern int max_linkhdr; /* Largest link-level header */
351extern int max_protohdr; /* Size of largest protocol layer header. See user_mbuf.c */
352
353extern struct mbstat mbstat; /* General mbuf stats/infos */
354
355
356/*
357 * Evaluate TRUE if it's safe to write to the mbuf m's data region (this can
358 * be both the local data payload, or an external buffer area, depending on
359 * whether M_EXT is set).
360 */
361#define M_WRITABLE(m) (!((m)->m_flags & M_RDONLY) && \
362 (!(((m)->m_flags & M_EXT)) || \
363 (*((m)->m_ext.ref_cnt) == 1)) ) \
364
365
366/*
367 * Compute the amount of space available before the current start of data in
368 * an mbuf.
369 *
370 * The M_WRITABLE() is a temporary, conservative safety measure: the burden
371 * of checking writability of the mbuf data area rests solely with the caller.
372 */
373#define M_LEADINGSPACE(m) \
374 ((m)->m_flags & M_EXT ? \
375 (M_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \
376 (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \
377 (m)->m_data - (m)->m_dat)
378
379/*
380 * Compute the amount of space available after the end of data in an mbuf.
381 *
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.
384 */
385#define M_TRAILINGSPACE(m) \
386 ((m)->m_flags & M_EXT ? \
387 (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \
388 - ((m)->m_data + (m)->m_len) : 0) : \
389 &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len))
390
391
392
393/*
394 * Arrange to prepend space of size plen to mbuf m. If a new mbuf must be
395 * allocated, how specifies whether to wait. If the allocation fails, the
396 * original mbuf chain is freed and m is set to NULL.
397 */
398#define M_PREPEND(m, plen, how) do { \
399 struct mbuf **_mmp = &(m); \
400 struct mbuf *_mm = *_mmp; \
401 int _mplen = (plen); \
402 int __mhow = (how); \
403 \
404 if (M_LEADINGSPACE(_mm) >= _mplen) { \
405 _mm->m_data -= _mplen; \
406 _mm->m_len += _mplen; \
407 } else \
408 _mm = m_prepend(_mm, _mplen, __mhow); \
409 if (_mm != NULL && _mm->m_flags & M_PKTHDR) \
410 _mm->m_pkthdr.len += _mplen; \
411 *_mmp = _mm; \
412} while (0)
413
414/*
415 * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place an
416 * object of the specified size at the end of the mbuf, longword aligned.
417 */
418#define M_ALIGN(m, len) do { \
tuexenc7108432012-05-21 18:53:50 +0000419 KASSERT(!((m)->m_flags & (M_PKTHDR|M_EXT)), \
420 ("%s: M_ALIGN not normal mbuf", __func__)); \
421 KASSERT((m)->m_data == (m)->m_dat, \
422 ("%s: M_ALIGN not a virgin mbuf", __func__)); \
tuexenbca1dae2011-11-01 23:02:02 +0000423 (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \
424} while (0)
425
426/*
427 * As above, for mbufs allocated with m_gethdr/MGETHDR or initialized by
428 * M_DUP/MOVE_PKTHDR.
429 */
430#define MH_ALIGN(m, len) do { \
tuexenc7108432012-05-21 18:53:50 +0000431 KASSERT((m)->m_flags & M_PKTHDR && !((m)->m_flags & M_EXT), \
432 ("%s: MH_ALIGN not PKTHDR mbuf", __func__)); \
433 KASSERT((m)->m_data == (m)->m_pktdat, \
434 ("%s: MH_ALIGN not a virgin mbuf", __func__)); \
tuexenbca1dae2011-11-01 23:02:02 +0000435 (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \
436} while (0)
437
438#endif