1 /*- 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 32 #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" 42 43 #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 54 struct mbuf * m_gethdr(int how, short type); 55 struct mbuf * m_get(int how, short type); 56 struct mbuf * m_free(struct mbuf *m); 57 void m_clget(struct mbuf *m, int how); 58 struct mbuf * m_getm2(struct mbuf *m, int len, int how, short type, int flags, int allonebuf); 59 struct mbuf *m_uiotombuf(struct uio *uio, int how, int len, int align, int flags); 60 u_int m_length(struct mbuf *m0, struct mbuf **last); 61 struct mbuf *m_last(struct mbuf *m); 62 63 /* mbuf initialization function */ 64 void mbuf_initialize(void *); 65 66 #define M_MOVE_PKTHDR(to, from) m_move_pkthdr((to), (from)) 67 #define MGET(m, how, type) ((m) = m_get((how), (type))) 68 #define MGETHDR(m, how, type) ((m) = m_gethdr((how), (type))) 69 #define MCLGET(m, how) m_clget((m), (how)) 70 71 72 #define M_HDR_PAD ((sizeof(intptr_t)==4) ? 2 : 6) /* modified for __Userspace__ */ 73 74 /* Length to m_copy to copy all. */ 75 #define M_COPYALL 1000000000 76 77 /* umem_cache_t is defined in user_include/umem.h as 78 * typedef struct umem_cache umem_cache_t; 79 * Note:umem_zone_t is a pointer. 80 */ 81 #if defined(SCTP_SIMPLE_ALLOCATOR) 82 typedef size_t sctp_zone_t; 83 #else 84 typedef umem_cache_t *sctp_zone_t; 85 #endif 86 87 extern sctp_zone_t zone_mbuf; 88 extern sctp_zone_t zone_clust; 89 extern sctp_zone_t zone_ext_refcnt; 90 91 /*- 92 * Macros for type conversion: 93 * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type. 94 * dtom(x) -- Convert data pointer within mbuf to mbuf pointer (XXX). 95 */ 96 #define mtod(m, t) ((t)((m)->m_data)) 97 #define dtom(x) ((struct mbuf *)((intptr_t)(x) & ~(MSIZE-1))) 98 99 struct mb_args { 100 int flags; /* Flags for mbuf being allocated */ 101 short type; /* Type of mbuf being allocated */ 102 }; 103 104 struct clust_args { 105 struct mbuf * parent_mbuf; 106 }; 107 108 struct mbuf * m_split(struct mbuf *, int, int); 109 void m_cat(struct mbuf *m, struct mbuf *n); 110 void m_adj(struct mbuf *, int); 111 void mb_free_ext(struct mbuf *); 112 void m_freem(struct mbuf *); 113 struct m_tag *m_tag_alloc(u_int32_t, int, int, int); 114 struct mbuf *m_copym(struct mbuf *, int, int, int); 115 void m_copyback(struct mbuf *, int, int, caddr_t); 116 struct mbuf *m_pullup(struct mbuf *, int); 117 struct mbuf *m_pulldown(struct mbuf *, int off, int len, int *offp); 118 int m_dup_pkthdr(struct mbuf *, struct mbuf *, int); 119 struct m_tag *m_tag_copy(struct m_tag *, int); 120 int m_tag_copy_chain(struct mbuf *, struct mbuf *, int); 121 struct mbuf *m_prepend(struct mbuf *, int, int); 122 void m_copydata(const struct mbuf *, int, int, caddr_t); 123 124 #define MBUF_MEM_NAME "mbuf" 125 #define MBUF_CLUSTER_MEM_NAME "mbuf_cluster" 126 #define MBUF_EXTREFCNT_MEM_NAME "mbuf_ext_refcnt" 127 128 #define MT_NOINIT 255 /* Not a type but a flag to allocate 129 a non-initialized mbuf */ 130 131 /* 132 * General mbuf allocator statistics structure. 133 * __Userspace__ mbstat may be useful for gathering statistics. 134 * In the kernel many of these statistics are no longer used as 135 * they track allocator statistics through kernel UMA's built in statistics mechanism. 136 */ 137 struct mbstat { 138 u_long m_mbufs; /* XXX */ 139 u_long m_mclusts; /* XXX */ 140 141 u_long m_drain; /* times drained protocols for space */ 142 u_long m_mcfail; /* XXX: times m_copym failed */ 143 u_long m_mpfail; /* XXX: times m_pullup failed */ 144 u_long m_msize; /* length of an mbuf */ 145 u_long m_mclbytes; /* length of an mbuf cluster */ 146 u_long m_minclsize; /* min length of data to allocate a cluster */ 147 u_long m_mlen; /* length of data in an mbuf */ 148 u_long m_mhlen; /* length of data in a header mbuf */ 149 150 /* Number of mbtypes (gives # elems in mbtypes[] array: */ 151 short m_numtypes; 152 153 /* XXX: Sendfile stats should eventually move to their own struct */ 154 u_long sf_iocnt; /* times sendfile had to do disk I/O */ 155 u_long sf_allocfail; /* times sfbuf allocation failed */ 156 u_long sf_allocwait; /* times sfbuf allocation had to wait */ 157 }; 158 159 160 /* 161 * Mbufs are of a single size, MSIZE (sys/param.h), which includes overhead. 162 * An mbuf may add a single "mbuf cluster" of size MCLBYTES (also in 163 * sys/param.h), which has no additional overhead and is used instead of the 164 * internal data area; this is done when at least MINCLSIZE of data must be 165 * stored. Additionally, it is possible to allocate a separate buffer 166 * externally and attach it to the mbuf in a way similar to that of mbuf 167 * clusters. 168 */ 169 #define MLEN ((int)(MSIZE - sizeof(struct m_hdr))) /* normal data len */ 170 #define MHLEN ((int)(MLEN - sizeof(struct pkthdr))) /* data len w/pkthdr */ 171 #define MINCLSIZE ((int)(MHLEN + 1)) /* smallest amount to put in cluster */ 172 #define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */ 173 174 175 /* 176 * Header present at the beginning of every mbuf. 177 */ 178 struct m_hdr { 179 struct mbuf *mh_next; /* next buffer in chain */ 180 struct mbuf *mh_nextpkt; /* next chain in queue/record */ 181 caddr_t mh_data; /* location of data */ 182 int mh_len; /* amount of data in this mbuf */ 183 int mh_flags; /* flags; see below */ 184 short mh_type; /* type of data in this mbuf */ 185 uint8_t pad[M_HDR_PAD];/* word align */ 186 }; 187 188 /* 189 * Packet tag structure (see below for details). 190 */ 191 struct m_tag { 192 SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */ 193 u_int16_t m_tag_id; /* Tag ID */ 194 u_int16_t m_tag_len; /* Length of data */ 195 u_int32_t m_tag_cookie; /* ABI/Module ID */ 196 void (*m_tag_free)(struct m_tag *); 197 }; 198 199 /* 200 * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set. 201 */ 202 struct pkthdr { 203 struct ifnet *rcvif; /* rcv interface */ 204 /* variables for ip and tcp reassembly */ 205 void *header; /* pointer to packet header */ 206 int len; /* total packet length */ 207 /* variables for hardware checksum */ 208 int csum_flags; /* flags regarding checksum */ 209 int csum_data; /* data field used by csum routines */ 210 u_int16_t tso_segsz; /* TSO segment size */ 211 u_int16_t ether_vtag; /* Ethernet 802.1p+q vlan tag */ 212 SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */ 213 }; 214 215 /* 216 * Description of external storage mapped into mbuf; valid only if M_EXT is 217 * set. 218 */ 219 struct m_ext { 220 caddr_t ext_buf; /* start of buffer */ 221 void (*ext_free) /* free routine if not the usual */ 222 (void *, void *); 223 void *ext_args; /* optional argument pointer */ 224 u_int ext_size; /* size of buffer, for ext_free */ 225 volatile u_int *ref_cnt; /* pointer to ref count info */ 226 int ext_type; /* type of external storage */ 227 }; 228 229 230 /* 231 * The core of the mbuf object along with some shortcut defined for practical 232 * purposes. 233 */ 234 struct mbuf { 235 struct m_hdr m_hdr; 236 union { 237 struct { 238 struct pkthdr MH_pkthdr; /* M_PKTHDR set */ 239 union { 240 struct m_ext MH_ext; /* M_EXT set */ 241 char MH_databuf[MHLEN]; 242 } MH_dat; 243 } MH; 244 char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */ 245 } M_dat; 246 }; 247 248 #define m_next m_hdr.mh_next 249 #define m_len m_hdr.mh_len 250 #define m_data m_hdr.mh_data 251 #define m_type m_hdr.mh_type 252 #define m_flags m_hdr.mh_flags 253 #define m_nextpkt m_hdr.mh_nextpkt 254 #define m_act m_nextpkt 255 #define m_pkthdr M_dat.MH.MH_pkthdr 256 #define m_ext M_dat.MH.MH_dat.MH_ext 257 #define m_pktdat M_dat.MH.MH_dat.MH_databuf 258 #define m_dat M_dat.M_databuf 259 260 261 /* 262 * mbuf flags. 263 */ 264 #define M_EXT 0x0001 /* has associated external storage */ 265 #define M_PKTHDR 0x0002 /* start of record */ 266 #define M_EOR 0x0004 /* end of record */ 267 #define M_RDONLY 0x0008 /* associated data is marked read-only */ 268 #define M_PROTO1 0x0010 /* protocol-specific */ 269 #define M_PROTO2 0x0020 /* protocol-specific */ 270 #define M_PROTO3 0x0040 /* protocol-specific */ 271 #define M_PROTO4 0x0080 /* protocol-specific */ 272 #define M_PROTO5 0x0100 /* protocol-specific */ 273 #define M_FREELIST 0x8000 /* mbuf is on the free list */ 274 275 276 /* 277 * Flags copied when copying m_pkthdr. 278 */ 279 #define M_COPYFLAGS (M_PKTHDR|M_EOR|M_RDONLY|M_PROTO1|M_PROTO1|M_PROTO2|\ 280 M_PROTO3|M_PROTO4|M_PROTO5|\ 281 M_BCAST|M_MCAST|M_FRAG|M_FIRSTFRAG|M_LASTFRAG|\ 282 M_VLANTAG|M_PROMISC) 283 284 285 /* 286 * mbuf pkthdr flags (also stored in m_flags). 287 */ 288 #define M_BCAST 0x0200 /* send/received as link-level broadcast */ 289 #define M_MCAST 0x0400 /* send/received as link-level multicast */ 290 #define M_FRAG 0x0800 /* packet is a fragment of a larger packet */ 291 #define M_FIRSTFRAG 0x1000 /* packet is first fragment */ 292 #define M_LASTFRAG 0x2000 /* packet is last fragment */ 293 #define M_VLANTAG 0x10000 /* ether_vtag is valid */ 294 #define M_PROMISC 0x20000 /* packet was not for us */ 295 #define M_NOFREE 0x40000 /* do not free mbuf - it is embedded in the cluster */ 296 297 298 /* 299 * External buffer types: identify ext_buf type. 300 */ 301 #define EXT_CLUSTER 1 /* mbuf cluster */ 302 #define EXT_SFBUF 2 /* sendfile(2)'s sf_bufs */ 303 #define EXT_JUMBOP 3 /* jumbo cluster 4096 bytes */ 304 #define EXT_JUMBO9 4 /* jumbo cluster 9216 bytes */ 305 #define EXT_JUMBO16 5 /* jumbo cluster 16184 bytes */ 306 #define EXT_PACKET 6 /* mbuf+cluster from packet zone */ 307 #define EXT_MBUF 7 /* external mbuf reference (M_IOVEC) */ 308 #define EXT_NET_DRV 100 /* custom ext_buf provided by net driver(s) */ 309 #define EXT_MOD_TYPE 200 /* custom module's ext_buf type */ 310 #define EXT_DISPOSABLE 300 /* can throw this buffer away w/page flipping */ 311 #define EXT_EXTREF 400 /* has externally maintained ref_cnt ptr */ 312 313 314 /* 315 * mbuf types. 316 */ 317 #define MT_NOTMBUF 0 /* USED INTERNALLY ONLY! Object is not mbuf */ 318 #define MT_DATA 1 /* dynamic (data) allocation */ 319 #define MT_HEADER MT_DATA /* packet header, use M_PKTHDR instead */ 320 #define MT_SONAME 8 /* socket name */ 321 #define MT_CONTROL 14 /* extra-data protocol message */ 322 #define MT_OOBDATA 15 /* expedited data */ 323 #define MT_NTYPES 16 /* number of mbuf types for mbtypes[] */ 324 325 #define MT_NOINIT 255 /* Not a type but a flag to allocate 326 a non-initialized mbuf */ 327 328 /* 329 * __Userspace__ flags like M_NOWAIT are defined in malloc.h 330 * Flags like these are used in functions like uma_zalloc() 331 * but don't have an equivalent in userland umem 332 * Flags specifying how an allocation should be made. 333 * 334 * The flag to use is as follows: 335 * - M_DONTWAIT or M_NOWAIT from an interrupt handler to not block allocation. 336 * - M_WAIT or M_WAITOK or M_TRYWAIT from wherever it is safe to block. 337 * 338 * M_DONTWAIT/M_NOWAIT means that we will not block the thread explicitly and 339 * if we cannot allocate immediately we may return NULL, whereas 340 * M_WAIT/M_WAITOK/M_TRYWAIT means that if we cannot allocate resources we 341 * will block until they are available, and thus never return NULL. 342 * 343 * XXX Eventually just phase this out to use M_WAITOK/M_NOWAIT. 344 */ 345 #define MBTOM(how) (how) 346 347 void m_tag_delete(struct mbuf *, struct m_tag *); 348 void m_tag_delete_chain(struct mbuf *, struct m_tag *); 349 void m_move_pkthdr(struct mbuf *, struct mbuf *); 350 void m_tag_free_default(struct m_tag *); 351 352 extern int max_linkhdr; /* Largest link-level header */ 353 extern int max_protohdr; /* Size of largest protocol layer header. See user_mbuf.c */ 354 355 extern struct mbstat mbstat; /* General mbuf stats/infos */ 356 357 358 /* 359 * Evaluate TRUE if it's safe to write to the mbuf m's data region (this can 360 * be both the local data payload, or an external buffer area, depending on 361 * whether M_EXT is set). 362 */ 363 #define M_WRITABLE(m) (!((m)->m_flags & M_RDONLY) && \ 364 (!(((m)->m_flags & M_EXT)) || \ 365 (*((m)->m_ext.ref_cnt) == 1)) ) \ 366 367 368 /* 369 * Compute the amount of space available before the current start of data in 370 * an mbuf. 371 * 372 * The M_WRITABLE() is a temporary, conservative safety measure: the burden 373 * of checking writability of the mbuf data area rests solely with the caller. 374 */ 375 #define M_LEADINGSPACE(m) \ 376 ((m)->m_flags & M_EXT ? \ 377 (M_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \ 378 (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \ 379 (m)->m_data - (m)->m_dat) 380 381 /* 382 * Compute the amount of space available after the end of data in an mbuf. 383 * 384 * The M_WRITABLE() is a temporary, conservative safety measure: the burden 385 * of checking writability of the mbuf data area rests solely with the caller. 386 */ 387 #define M_TRAILINGSPACE(m) \ 388 ((m)->m_flags & M_EXT ? \ 389 (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \ 390 - ((m)->m_data + (m)->m_len) : 0) : \ 391 &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len)) 392 393 394 395 /* 396 * Arrange to prepend space of size plen to mbuf m. If a new mbuf must be 397 * allocated, how specifies whether to wait. If the allocation fails, the 398 * original mbuf chain is freed and m is set to NULL. 399 */ 400 #define M_PREPEND(m, plen, how) do { \ 401 struct mbuf **_mmp = &(m); \ 402 struct mbuf *_mm = *_mmp; \ 403 int _mplen = (plen); \ 404 int __mhow = (how); \ 405 \ 406 if (M_LEADINGSPACE(_mm) >= _mplen) { \ 407 _mm->m_data -= _mplen; \ 408 _mm->m_len += _mplen; \ 409 } else \ 410 _mm = m_prepend(_mm, _mplen, __mhow); \ 411 if (_mm != NULL && _mm->m_flags & M_PKTHDR) \ 412 _mm->m_pkthdr.len += _mplen; \ 413 *_mmp = _mm; \ 414 } while (0) 415 416 /* 417 * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place an 418 * object of the specified size at the end of the mbuf, longword aligned. 419 */ 420 #define M_ALIGN(m, len) do { \ 421 KASSERT(!((m)->m_flags & (M_PKTHDR|M_EXT)), \ 422 ("%s: M_ALIGN not normal mbuf", __func__)); \ 423 KASSERT((m)->m_data == (m)->m_dat, \ 424 ("%s: M_ALIGN not a virgin mbuf", __func__)); \ 425 (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \ 426 } while (0) 427 428 /* 429 * As above, for mbufs allocated with m_gethdr/MGETHDR or initialized by 430 * M_DUP/MOVE_PKTHDR. 431 */ 432 #define MH_ALIGN(m, len) do { \ 433 KASSERT((m)->m_flags & M_PKTHDR && !((m)->m_flags & M_EXT), \ 434 ("%s: MH_ALIGN not PKTHDR mbuf", __func__)); \ 435 KASSERT((m)->m_data == (m)->m_pktdat, \ 436 ("%s: MH_ALIGN not a virgin mbuf", __func__)); \ 437 (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \ 438 } while (0) 439 440 #endif 441