bsd/kern/uipc_mbuf2.c

   1 /*
   2  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * The contents of this file constitute Original Code as defined in and
   7  * are subject to the Apple Public Source License Version 1.1 (the
   8  * "License").  You may not use this file except in compliance with the
   9  * License.  Please obtain a copy of the License at
  10  * http://www.apple.com/publicsource and read it before using this file.
  11  *
  12  * This Original Code and all software distributed under the License are
  13  * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  14  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  15  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  16  * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
  17  * License for the specific language governing rights and limitations
  18  * under the License.
  19  *
  20  * @APPLE_LICENSE_HEADER_END@
  21  */
  22 /*      $NetBSD: uipc_mbuf.c,v 1.40 1999/04/01 00:23:25 thorpej Exp $   */
  23
  24 /*
  25  * Copyright (C) 1999 WIDE Project.
  26  * All rights reserved.
  27  *
  28  * Redistribution and use in source and binary forms, with or without
  29  * modification, are permitted provided that the following conditions
  30  * are met:
  31  * 1. Redistributions of source code must retain the above copyright
  32  *    notice, this list of conditions and the following disclaimer.
  33  * 2. Redistributions in binary form must reproduce the above copyright
  34  *    notice, this list of conditions and the following disclaimer in the
  35  *    documentation and/or other materials provided with the distribution.
  36  * 3. Neither the name of the project nor the names of its contributors
  37  *    may be used to endorse or promote products derived from this software
  38  *    without specific prior written permission.
  39  *
  40  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
  41  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  42  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  43  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
  44  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  45  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  46  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  47  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  48  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  49  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  50  * SUCH DAMAGE.
  51  */
  52
  53 /*
  54  * Copyright (c) 1982, 1986, 1988, 1991, 1993
  55  *      The Regents of the University of California.  All rights reserved.
  56  *
  57  * Redistribution and use in source and binary forms, with or without
  58  * modification, are permitted provided that the following conditions
  59  * are met:
  60  * 1. Redistributions of source code must retain the above copyright
  61  *    notice, this list of conditions and the following disclaimer.
  62  * 2. Redistributions in binary form must reproduce the above copyright
  63  *    notice, this list of conditions and the following disclaimer in the
  64  *    documentation and/or other materials provided with the distribution.
  65  * 3. All advertising materials mentioning features or use of this software
  66  *    must display the following acknowledgement:
  67  *      This product includes software developed by the University of
  68  *      California, Berkeley and its contributors.
  69  * 4. Neither the name of the University nor the names of its contributors
  70  *    may be used to endorse or promote products derived from this software
  71  *    without specific prior written permission.
  72  *
  73  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  74  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  75  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  76  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  77  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  78  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  79  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  80  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  81  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  82  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  83  * SUCH DAMAGE.
  84  *
  85  *      @(#)uipc_mbuf.c 8.4 (Berkeley) 2/14/95
  86  */
  87
  88
  89 /*#define PULLDOWN_DEBUG*/
  90
  91 #include <sys/param.h>
  92 #include <sys/systm.h>
  93 #include <sys/proc_internal.h>
  94 #include <sys/malloc.h>
  95 #include <sys/mbuf.h>
  96 #if defined(PULLDOWN_STAT) && defined(INET6)
  97 #include <netinet/in.h>
  98 #include <netinet/ip6.h>
  99 #include <netinet6/ip6_var.h>
 100 #endif
 101
 102 /*
 103  * ensure that [off, off + len) is contiguous on the mbuf chain "m".
 104  * packet chain before "off" is kept untouched.
 105  * if offp == NULL, the target will start at <retval, 0> on resulting chain.
 106  * if offp != NULL, the target will start at <retval, *offp> on resulting chain.
 107  *
 108  * on error return (NULL return value), original "m" will be freed.
 109  *
 110  * XXX M_TRAILINGSPACE/M_LEADINGSPACE on shared cluster (sharedcluster)
 111  */
 112 struct mbuf *
 113 m_pulldown(m, off, len, offp)
 114         struct mbuf *m;
 115         int off, len;
 116         int *offp;
 117 {
 118         struct mbuf *n, *o;
 119         int hlen, tlen, olen;
 120         int sharedcluster;
 121 #if defined(PULLDOWN_STAT) && defined(INET6)
 122         static struct mbuf *prev = NULL;
 123         int prevlen = 0, prevmlen = 0;
 124 #endif
 125
 126         /* check invalid arguments. */
 127         if (m == NULL)
 128                 panic("m == NULL in m_pulldown()");
 129         if (len > MCLBYTES) {
 130                 m_freem(m);
 131                 return NULL;    /* impossible */
 132         }
 133
 134 #if defined(PULLDOWN_STAT) && defined(INET6)
 135         ip6stat.ip6s_pulldown++;
 136 #endif
 137
 138 #if defined(PULLDOWN_STAT) && defined(INET6)
 139         /* statistics for m_pullup */
 140         ip6stat.ip6s_pullup++;
 141         if (off + len > MHLEN)
 142                 ip6stat.ip6s_pullup_fail++;
 143         else {
 144                 int dlen, mlen;
 145
 146                 dlen = (prev == m) ? prevlen : m->m_len;
 147                 mlen = (prev == m) ? prevmlen : m->m_len + M_TRAILINGSPACE(m);
 148
 149                 if (dlen >= off + len)
 150                         ip6stat.ip6s_pullup--; /* call will not be made! */
 151                 else if ((m->m_flags & M_EXT) != 0) {
 152                         ip6stat.ip6s_pullup_alloc++;
 153                         ip6stat.ip6s_pullup_copy++;
 154                 } else {
 155                         if (mlen >= off + len)
 156                                 ip6stat.ip6s_pullup_copy++;
 157                         else {
 158                                 ip6stat.ip6s_pullup_alloc++;
 159                                 ip6stat.ip6s_pullup_copy++;
 160                         }
 161                 }
 162
 163                 prevlen = off + len;
 164                 prevmlen = MHLEN;
 165         }
 166
 167         /* statistics for m_pullup2 */
 168         ip6stat.ip6s_pullup2++;
 169         if (off + len > MCLBYTES)
 170                 ip6stat.ip6s_pullup2_fail++;
 171         else {
 172                 int dlen, mlen;
 173
 174                 dlen = (prev == m) ? prevlen : m->m_len;
 175                 mlen = (prev == m) ? prevmlen : m->m_len + M_TRAILINGSPACE(m);
 176                 prevlen = off + len;
 177                 prevmlen = mlen;
 178
 179                 if (dlen >= off + len)
 180                         ip6stat.ip6s_pullup2--; /* call will not be made! */
 181                 else if ((m->m_flags & M_EXT) != 0) {
 182                         ip6stat.ip6s_pullup2_alloc++;
 183                         ip6stat.ip6s_pullup2_copy++;
 184                         prevmlen = (off + len > MHLEN) ? MCLBYTES : MHLEN;
 185                 } else {
 186                         if (mlen >= off + len)
 187                                 ip6stat.ip6s_pullup2_copy++;
 188                         else {
 189                                 ip6stat.ip6s_pullup2_alloc++;
 190                                 ip6stat.ip6s_pullup2_copy++;
 191                                 prevmlen = (off + len > MHLEN) ? MCLBYTES
 192                                                                : MHLEN;
 193                         }
 194                 }
 195         }
 196
 197         prev = m;
 198 #endif
 199
 200 #ifdef PULLDOWN_DEBUG
 201     {
 202         struct mbuf *t;
 203         printf("before:");
 204         for (t = m; t; t = t->m_next)
 205                 printf(" %d", t->m_len);
 206         printf("\n");
 207     }
 208 #endif
 209         n = m;
 210         while (n != NULL && off > 0) {
 211                 if (n->m_len > off)
 212                         break;
 213                 off -= n->m_len;
 214                 n = n->m_next;
 215         }
 216         /* be sure to point non-empty mbuf */
 217         while (n != NULL && n->m_len == 0)
 218                 n = n->m_next;
 219         if (!n) {
 220                 m_freem(m);
 221                 return NULL;    /* mbuf chain too short */
 222         }
 223
 224         /*
 225          * the target data is on <n, off>.
 226          * if we got enough data on the mbuf "n", we're done.
 227          */
 228         if ((off == 0 || offp) && len <= n->m_len - off)
 229                 goto ok;
 230
 231 #if defined(PULLDOWN_STAT) && defined(INET6)
 232         ip6stat.ip6s_pulldown_copy++;
 233 #endif
 234
 235         /*
 236          * when len < n->m_len - off and off != 0, it is a special case.
 237          * len bytes from <n, off> sits in single mbuf, but the caller does
 238          * not like the starting position (off).
 239          * chop the current mbuf into two pieces, set off to 0.
 240          */
 241         if (len < n->m_len - off) {
 242                 o = m_copym(n, off, n->m_len - off, M_DONTWAIT);
 243                 if (o == NULL) {
 244                         m_freem(m);
 245                         return NULL;    /* ENOBUFS */
 246                 }
 247                 n->m_len = off;
 248                 o->m_next = n->m_next;
 249                 n->m_next = o;
 250                 n = n->m_next;
 251                 off = 0;
 252                 goto ok;
 253         }
 254
 255         /*
 256          * we need to take hlen from <n, off> and tlen from <n->m_next, 0>,
 257          * and construct contiguous mbuf with m_len == len.
 258          * note that hlen + tlen == len, and tlen > 0.
 259          */
 260         hlen = n->m_len - off;
 261         tlen = len - hlen;
 262
 263         /*
 264          * ensure that we have enough trailing data on mbuf chain.
 265          * if not, we can do nothing about the chain.
 266          */
 267         olen = 0;
 268         for (o = n->m_next; o != NULL; o = o->m_next)
 269                 olen += o->m_len;
 270         if (hlen + olen < len) {
 271                 m_freem(m);
 272                 return NULL;    /* mbuf chain too short */
 273         }
 274
 275         /*
 276          * easy cases first.
 277          * we need to use m_copydata() to get data from <n->m_next, 0>.
 278          */
 279         if ((n->m_flags & M_EXT) == 0)
 280                 sharedcluster = 0;
 281         else {
 282                 if (n->m_ext.ext_free)
 283                         sharedcluster = 1;
 284                 else if (m_mclhasreference(n))
 285                         sharedcluster = 1;
 286                 else
 287                         sharedcluster = 0;
 288         }
 289         if ((off == 0 || offp) && M_TRAILINGSPACE(n) >= tlen
 290          && !sharedcluster) {
 291                 m_copydata(n->m_next, 0, tlen, mtod(n, caddr_t) + n->m_len);
 292                 n->m_len += tlen;
 293                 m_adj(n->m_next, tlen);
 294                 goto ok;
 295         }
 296         if ((off == 0 || offp) && M_LEADINGSPACE(n->m_next) >= hlen
 297          && !sharedcluster) {
 298                 n->m_next->m_data -= hlen;
 299                 n->m_next->m_len += hlen;
 300                 bcopy(mtod(n, caddr_t) + off, mtod(n->m_next, caddr_t), hlen);
 301                 n->m_len -= hlen;
 302                 n = n->m_next;
 303                 off = 0;
 304                 goto ok;
 305         }
 306
 307         /*
 308          * now, we need to do the hard way.  don't m_copy as there's no room
 309          * on both end.
 310          */
 311 #if defined(PULLDOWN_STAT) && defined(INET6)
 312         ip6stat.ip6s_pulldown_alloc++;
 313 #endif
 314         MGET(o, M_DONTWAIT, m->m_type);
 315         if (o == NULL) {
 316                 m_freem(m);
 317                 return NULL;    /* ENOBUFS */
 318         }
 319         if (len > MHLEN) {      /* use MHLEN just for safety */
 320                 MCLGET(o, M_DONTWAIT);
 321                 if ((o->m_flags & M_EXT) == 0) {
 322                         m_freem(m);
 323                         m_free(o);
 324                         return NULL;    /* ENOBUFS */
 325                 }
 326         }
 327         /* get hlen from <n, off> into <o, 0> */
 328         o->m_len = hlen;
 329         bcopy(mtod(n, caddr_t) + off, mtod(o, caddr_t), hlen);
 330         n->m_len -= hlen;
 331         /* get tlen from <n->m_next, 0> into <o, hlen> */
 332         m_copydata(n->m_next, 0, tlen, mtod(o, caddr_t) + o->m_len);
 333         o->m_len += tlen;
 334         m_adj(n->m_next, tlen);
 335         o->m_next = n->m_next;
 336         n->m_next = o;
 337         n = o;
 338         off = 0;
 339
 340 ok:
 341 #ifdef PULLDOWN_DEBUG
 342     {
 343         struct mbuf *t;
 344         printf("after:");
 345         for (t = m; t; t = t->m_next)
 346                 printf("%c%d", t == n ? '*' : ' ', t->m_len);
 347         printf(" (off=%d)\n", off);
 348     }
 349 #endif
 350         if (offp)
 351                 *offp = off;
 352         return n;
 353 }
 354
 355 /*
 356  * pkthdr.aux chain manipulation.
 357  * we don't allow clusters at this moment.
 358  */
 359 struct mbuf *
 360 m_aux_add(m, af, type)
 361         struct mbuf *m;
 362         int af, type;
 363 {
 364         struct mbuf *n;
 365         struct mauxtag *t;
 366
 367         if ((m->m_flags & M_PKTHDR) == 0)
 368                 return NULL;
 369
 370         n = m_aux_find(m, af, type);
 371         if (n)
 372                 return n;
 373
 374         MGET(n, M_DONTWAIT, m->m_type);
 375         if (n == NULL)
 376                 return NULL;
 377
 378         t = mtod(n, struct mauxtag *);
 379         t->af = af;
 380         t->type = type;
 381         n->m_data += sizeof(struct mauxtag);
 382         n->m_len = 0;
 383         n->m_next = m->m_pkthdr.aux;
 384         m->m_pkthdr.aux = n;
 385         return n;
 386 }
 387
 388 struct mbuf *
 389 m_aux_find(m, af, type)
 390         struct mbuf *m;
 391         int af, type;
 392 {
 393         struct mbuf *n;
 394         struct mauxtag *t;
 395
 396         if ((m->m_flags & M_PKTHDR) == 0)
 397                 return NULL;
 398
 399         for (n = m->m_pkthdr.aux; n; n = n->m_next) {
 400                 t = (struct mauxtag *)n->m_dat;
 401                 if (t->af == af && t->type == type)
 402                         return n;
 403         }
 404         return NULL;
 405 }
 406
 407 void
 408 m_aux_delete(m, victim)
 409         struct mbuf *m;
 410         struct mbuf *victim;
 411 {
 412         struct mbuf *n, *prev, *next;
 413         struct mauxtag *t;
 414
 415         if ((m->m_flags & M_PKTHDR) == 0)
 416                 return;
 417
 418         prev = NULL;
 419         n = m->m_pkthdr.aux;
 420         while (n) {
 421                 t = (struct mauxtag *)n->m_dat;
 422                 next = n->m_next;
 423                 if (n == victim) {
 424                         if (prev)
 425                                 prev->m_next = n->m_next;
 426                         else
 427                                 m->m_pkthdr.aux = n->m_next;
 428                         n->m_next = NULL;
 429                         m_free(n);
 430                 } else
 431                         prev = n;
 432                 n = next;
 433         }
 434 }
 435
 436 /* Get a packet tag structure along with specified data following. */
 437 struct m_tag *
 438 m_tag_alloc(u_int32_t id, u_int16_t type, int len, int wait)
 439 {
 440         struct m_tag *t;
 441
 442         if (len < 0)
 443                 return NULL;
 444 #ifndef __APPLE__
 445         t = malloc(len + sizeof(struct m_tag), M_PACKET_TAGS, wait);
 446 #else
 447         /*MALLOC(t, struct m_tag *, len + sizeof(struct m_tag), M_TEMP, M_WAITOK);*/
 448         if (len + sizeof(struct m_tag) <= MLEN) {
 449                 struct mbuf *m = m_get(wait, MT_TAG);
 450                 if (m == NULL)
 451                         return NULL;
 452                 t = (struct m_tag *) m->m_dat;
 453         } else if (len + sizeof(struct m_tag) <= MCLBYTES) {
 454                 MCLALLOC((caddr_t)t, wait);
 455         } else
 456                 t = NULL;
 457 #endif
 458         if (t == NULL)
 459                 return NULL;
 460         t->m_tag_type = type;
 461         t->m_tag_len = len;
 462         t->m_tag_id = id;
 463         return t;
 464 }
 465
 466
 467 /* Free a packet tag. */
 468 void
 469 m_tag_free(struct m_tag *t)
 470 {
 471 #ifndef __APPLE__
 472         free(t, M_PACKET_TAGS);
 473 #else
 474         /* FREE(t, M_TEMP); */
 475         if (t == NULL)
 476                 return;
 477         if (t->m_tag_len <= MLEN) {
 478                 struct mbuf * m = m_dtom(t);
 479                 m_free(m);
 480         } else {
 481                 MCLFREE((caddr_t)t);
 482         }
 483 #endif
 484 }
 485
 486 /* Prepend a packet tag. */
 487 void
 488 m_tag_prepend(struct mbuf *m, struct m_tag *t)
 489 {
 490         KASSERT(m && t, ("m_tag_prepend: null argument, m %p t %p", m, t));
 491         SLIST_INSERT_HEAD(&m->m_pkthdr.tags, t, m_tag_link);
 492 }
 493
 494 /* Unlink a packet tag. */
 495 void
 496 m_tag_unlink(struct mbuf *m, struct m_tag *t)
 497 {
 498         KASSERT(m && t, ("m_tag_unlink: null argument, m %p t %p", m, t));
 499         SLIST_REMOVE(&m->m_pkthdr.tags, t, m_tag, m_tag_link);
 500 }
 501
 502 /* Unlink and free a packet tag. */
 503 void
 504 m_tag_delete(struct mbuf *m, struct m_tag *t)
 505 {
 506         KASSERT(m && t, ("m_tag_delete: null argument, m %p t %p", m, t));
 507         m_tag_unlink(m, t);
 508         m_tag_free(t);
 509 }
 510
 511 /* Unlink and free a packet tag chain, starting from given tag. */
 512 void
 513 m_tag_delete_chain(struct mbuf *m, struct m_tag *t)
 514 {
 515         struct m_tag *p, *q;
 516
 517         KASSERT(m, ("m_tag_delete_chain: null mbuf"));
 518         if (t != NULL)
 519                 p = t;
 520         else
 521                 p = SLIST_FIRST(&m->m_pkthdr.tags);
 522         if (p == NULL)
 523                 return;
 524         while ((q = SLIST_NEXT(p, m_tag_link)) != NULL)
 525                 m_tag_delete(m, q);
 526         m_tag_delete(m, p);
 527 }
 528
 529 /* Find a tag, starting from a given position. */
 530 struct m_tag *
 531 m_tag_locate(struct mbuf *m, u_int32_t id, u_int16_t type, struct m_tag *t)
 532 {
 533         struct m_tag *p;
 534
 535         KASSERT(m, ("m_tag_find: null mbuf"));
 536         if (t == NULL)
 537                 p = SLIST_FIRST(&m->m_pkthdr.tags);
 538         else
 539                 p = SLIST_NEXT(t, m_tag_link);
 540         while (p != NULL) {
 541                 if (p->m_tag_id == id && p->m_tag_type == type)
 542                         return p;
 543                 p = SLIST_NEXT(p, m_tag_link);
 544         }
 545         return NULL;
 546 }
 547
 548 /* Copy a single tag. */
 549 struct m_tag *
 550 m_tag_copy(struct m_tag *t, int how)
 551 {
 552         struct m_tag *p;
 553
 554         KASSERT(t, ("m_tag_copy: null tag"));
 555         p = m_tag_alloc(t->m_tag_id, t->m_tag_type, t->m_tag_len, how);
 556         if (p == NULL)
 557                 return (NULL);
 558         bcopy(t + 1, p + 1, t->m_tag_len); /* Copy the data */
 559         return p;
 560 }
 561
 562 /*
 563  * Copy two tag chains. The destination mbuf (to) loses any attached
 564  * tags even if the operation fails. This should not be a problem, as
 565  * m_tag_copy_chain() is typically called with a newly-allocated
 566  * destination mbuf.
 567  */
 568 int
 569 m_tag_copy_chain(struct mbuf *to, struct mbuf *from, int how)
 570 {
 571         struct m_tag *p, *t, *tprev = NULL;
 572
 573         KASSERT(to && from,
 574                 ("m_tag_copy: null argument, to %p from %p", to, from));
 575         m_tag_delete_chain(to, NULL);
 576         SLIST_FOREACH(p, &from->m_pkthdr.tags, m_tag_link) {
 577                 t = m_tag_copy(p, how);
 578                 if (t == NULL) {
 579                         m_tag_delete_chain(to, NULL);
 580                         return 0;
 581                 }
 582                 if (tprev == NULL)
 583                         SLIST_INSERT_HEAD(&to->m_pkthdr.tags, t, m_tag_link);
 584                 else {
 585                         SLIST_INSERT_AFTER(tprev, t, m_tag_link);
 586                         tprev = t;
 587                 }
 588         }
 589         return 1;
 590 }
 591
 592 /* Initialize tags on an mbuf. */
 593 void
 594 m_tag_init(struct mbuf *m)
 595 {
 596         SLIST_INIT(&m->m_pkthdr.tags);
 597 }
 598
 599 /* Get first tag in chain. */
 600 struct m_tag *
 601 m_tag_first(struct mbuf *m)
 602 {
 603         return SLIST_FIRST(&m->m_pkthdr.tags);
 604 }
 605
 606 /* Get next tag in chain. */
 607 struct m_tag *
 608 m_tag_next(__unused struct mbuf *m, struct m_tag *t)
 609 {
 610         return SLIST_NEXT(t, m_tag_link);
 611 }