sys/kern/kern_subr.c

   1 /*
   2  * Copyright (c) 1982, 1986, 1991, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  * (c) UNIX System Laboratories, Inc.
   5  * All or some portions of this file are derived from material licensed
   6  * to the University of California by American Telephone and Telegraph
   7  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
   8  * the permission of UNIX System Laboratories, Inc.
   9  *
  10  * Redistribution and use in source and binary forms, with or without
  11  * modification, are permitted provided that the following conditions
  12  * are met:
  13  * 1. Redistributions of source code must retain the above copyright
  14  *    notice, this list of conditions and the following disclaimer.
  15  * 2. Redistributions in binary form must reproduce the above copyright
  16  *    notice, this list of conditions and the following disclaimer in the
  17  *    documentation and/or other materials provided with the distribution.
  18  * 3. All advertising materials mentioning features or use of this software
  19  *    must display the following acknowledgement:
  20  *      This product includes software developed by the University of
  21  *      California, Berkeley and its contributors.
  22  * 4. Neither the name of the University nor the names of its contributors
  23  *    may be used to endorse or promote products derived from this software
  24  *    without specific prior written permission.
  25  *
  26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  36  * SUCH DAMAGE.
  37  *
  38  *      @(#)kern_subr.c 8.3 (Berkeley) 1/21/94
  39  * $FreeBSD: src/sys/kern/kern_subr.c,v 1.31.2.2 2002/04/21 08:09:37 bde Exp $
  40  * $DragonFly: src/sys/kern/kern_subr.c,v 1.20 2004/11/09 17:52:45 joerg Exp $
  41  */
  42
  43 #include "opt_ddb.h"
  44
  45 #include <sys/param.h>
  46 #include <sys/systm.h>
  47 #include <sys/kernel.h>
  48 #include <sys/proc.h>
  49 #include <sys/malloc.h>
  50 #include <sys/lock.h>
  51 #include <sys/resourcevar.h>
  52 #include <sys/sysctl.h>
  53 #include <sys/uio.h>
  54 #include <sys/vnode.h>
  55 #include <machine/limits.h>
  56
  57 #include <vm/vm.h>
  58 #include <vm/vm_page.h>
  59 #include <vm/vm_map.h>
  60
  61 SYSCTL_INT(_kern, KERN_IOV_MAX, iov_max, CTLFLAG_RD, NULL, UIO_MAXIOV,
  62         "Maximum number of elements in an I/O vector; sysconf(_SC_IOV_MAX)");
  63
  64 /*
  65  * UIO_READ:    copy the kernelspace cp to the user or kernelspace UIO
  66  * UIO_WRITE:   copy the user or kernelspace UIO to cp
  67  *
  68  * For userspace UIO's, uio_td must be the current thread.
  69  */
  70 int
  71 uiomove(caddr_t cp, int n, struct uio *uio)
  72 {
  73         struct iovec *iov;
  74         u_int cnt;
  75         int error = 0;
  76         int save = 0;
  77         int baseticks = ticks;
  78
  79         KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
  80             ("uiomove: mode"));
  81         KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
  82             ("uiomove proc"));
  83
  84         if (curproc) {
  85                 save = curproc->p_flag & P_DEADLKTREAT;
  86                 curproc->p_flag |= P_DEADLKTREAT;
  87         }
  88
  89         while (n > 0 && uio->uio_resid) {
  90                 iov = uio->uio_iov;
  91                 cnt = iov->iov_len;
  92                 if (cnt == 0) {
  93                         uio->uio_iov++;
  94                         uio->uio_iovcnt--;
  95                         continue;
  96                 }
  97                 if (cnt > n)
  98                         cnt = n;
  99
 100                 switch (uio->uio_segflg) {
 101
 102                 case UIO_USERSPACE:
 103                         if (ticks - baseticks >= hogticks) {
 104                                 uio_yield();
 105                                 baseticks = ticks;
 106                         }
 107                         if (uio->uio_rw == UIO_READ)
 108                                 error = copyout(cp, iov->iov_base, cnt);
 109                         else
 110                                 error = copyin(iov->iov_base, cp, cnt);
 111                         if (error)
 112                                 break;
 113                         break;
 114
 115                 case UIO_SYSSPACE:
 116                         if (uio->uio_rw == UIO_READ)
 117                                 bcopy((caddr_t)cp, iov->iov_base, cnt);
 118                         else
 119                                 bcopy(iov->iov_base, (caddr_t)cp, cnt);
 120                         break;
 121                 case UIO_NOCOPY:
 122                         break;
 123                 }
 124                 iov->iov_base += cnt;
 125                 iov->iov_len -= cnt;
 126                 uio->uio_resid -= cnt;
 127                 uio->uio_offset += cnt;
 128                 cp += cnt;
 129                 n -= cnt;
 130         }
 131         if (curproc)
 132                 curproc->p_flag = (curproc->p_flag & ~P_DEADLKTREAT) | save;
 133         return (error);
 134 }
 135 /*
 136  * Wrapper for uiomove() that validates the arguments against a known-good
 137  * kernel buffer.  Currently, uiomove accepts a signed (n) argument, which
 138  * is almost definitely a bad thing, so we catch that here as well.  We
 139  * return a runtime failure, but it might be desirable to generate a runtime
 140  * assertion failure instead.
 141  */
 142 int
 143 uiomove_frombuf(void *buf, int buflen, struct uio *uio)
 144 {
 145         unsigned int offset, n;
 146
 147         if (uio->uio_offset < 0 || uio->uio_resid < 0 ||
 148             (offset = uio->uio_offset) != uio->uio_offset)
 149                 return (EINVAL);
 150         if (buflen <= 0 || offset >= buflen)
 151                 return (0);
 152         if ((n = buflen - offset) > INT_MAX)
 153                 return (EINVAL);
 154         return (uiomove((char *)buf + offset, n, uio));
 155 }
 156
 157
 158 int
 159 uiomoveco(cp, n, uio, obj)
 160         caddr_t cp;
 161         int n;
 162         struct uio *uio;
 163         struct vm_object *obj;
 164 {
 165         struct iovec *iov;
 166         u_int cnt;
 167         int error;
 168         int baseticks = ticks;
 169
 170         KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
 171             ("uiomoveco: mode"));
 172         KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
 173             ("uiomoveco proc"));
 174
 175         while (n > 0 && uio->uio_resid) {
 176                 iov = uio->uio_iov;
 177                 cnt = iov->iov_len;
 178                 if (cnt == 0) {
 179                         uio->uio_iov++;
 180                         uio->uio_iovcnt--;
 181                         continue;
 182                 }
 183                 if (cnt > n)
 184                         cnt = n;
 185
 186                 switch (uio->uio_segflg) {
 187
 188                 case UIO_USERSPACE:
 189                         if (ticks - baseticks >= hogticks) {
 190                                 uio_yield();
 191                                 baseticks = ticks;
 192                         }
 193                         if (uio->uio_rw == UIO_READ) {
 194                                 error = copyout(cp, iov->iov_base, cnt);
 195                         } else {
 196                                 error = copyin(iov->iov_base, cp, cnt);
 197                         }
 198                         if (error)
 199                                 return (error);
 200                         break;
 201
 202                 case UIO_SYSSPACE:
 203                         if (uio->uio_rw == UIO_READ)
 204                                 bcopy((caddr_t)cp, iov->iov_base, cnt);
 205                         else
 206                                 bcopy(iov->iov_base, (caddr_t)cp, cnt);
 207                         break;
 208                 case UIO_NOCOPY:
 209                         break;
 210                 }
 211                 iov->iov_base += cnt;
 212                 iov->iov_len -= cnt;
 213                 uio->uio_resid -= cnt;
 214                 uio->uio_offset += cnt;
 215                 cp += cnt;
 216                 n -= cnt;
 217         }
 218         return (0);
 219 }
 220
 221 /*
 222  * Give next character to user as result of read.
 223  */
 224 int
 225 ureadc(c, uio)
 226         int c;
 227         struct uio *uio;
 228 {
 229         struct iovec *iov;
 230
 231 again:
 232         if (uio->uio_iovcnt == 0 || uio->uio_resid == 0)
 233                 panic("ureadc");
 234         iov = uio->uio_iov;
 235         if (iov->iov_len == 0) {
 236                 uio->uio_iovcnt--;
 237                 uio->uio_iov++;
 238                 goto again;
 239         }
 240         switch (uio->uio_segflg) {
 241
 242         case UIO_USERSPACE:
 243                 if (subyte(iov->iov_base, c) < 0)
 244                         return (EFAULT);
 245                 break;
 246
 247         case UIO_SYSSPACE:
 248                 *iov->iov_base = c;
 249                 break;
 250
 251         case UIO_NOCOPY:
 252                 break;
 253         }
 254         iov->iov_base++;
 255         iov->iov_len--;
 256         uio->uio_resid--;
 257         uio->uio_offset++;
 258         return (0);
 259 }
 260
 261 /*
 262  * General routine to allocate a hash table.  Make the hash table size a
 263  * power of 2 greater or equal to the number of elements requested, and
 264  * store the masking value in *hashmask.
 265  */
 266 void *
 267 hashinit(elements, type, hashmask)
 268         int elements;
 269         struct malloc_type *type;
 270         u_long *hashmask;
 271 {
 272         long hashsize;
 273         LIST_HEAD(generic, generic) *hashtbl;
 274         int i;
 275
 276         if (elements <= 0)
 277                 panic("hashinit: bad elements");
 278         for (hashsize = 2; hashsize < elements; hashsize <<= 1)
 279                 continue;
 280         hashtbl = malloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
 281         for (i = 0; i < hashsize; i++)
 282                 LIST_INIT(&hashtbl[i]);
 283         *hashmask = hashsize - 1;
 284         return (hashtbl);
 285 }
 286
 287 static int primes[] = { 1, 13, 31, 61, 127, 251, 509, 761, 1021, 1531, 2039,
 288                         2557, 3067, 3583, 4093, 4603, 5119, 5623, 6143, 6653,
 289                         7159, 7673, 8191, 12281, 16381, 24571, 32749 };
 290 #define NPRIMES (sizeof(primes) / sizeof(primes[0]))
 291
 292 /*
 293  * General routine to allocate a prime number sized hash table.
 294  */
 295 void *
 296 phashinit(elements, type, nentries)
 297         int elements;
 298         struct malloc_type *type;
 299         u_long *nentries;
 300 {
 301         long hashsize;
 302         LIST_HEAD(generic, generic) *hashtbl;
 303         int i;
 304
 305         if (elements <= 0)
 306                 panic("phashinit: bad elements");
 307         for (i = 1, hashsize = primes[1]; hashsize <= elements;) {
 308                 i++;
 309                 if (i == NPRIMES)
 310                         break;
 311                 hashsize = primes[i];
 312         }
 313         hashsize = primes[i - 1];
 314         hashtbl = malloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
 315         for (i = 0; i < hashsize; i++)
 316                 LIST_INIT(&hashtbl[i]);
 317         *nentries = hashsize;
 318         return (hashtbl);
 319 }
 320
 321 /*
 322  * Copyin an iovec.  If the iovec array fits, use the preallocated small
 323  * iovec structure.  If it is too big, dynamically allocate an iovec array
 324  * of sufficient size.
 325  */
 326 int
 327 iovec_copyin(struct iovec *uiov, struct iovec **kiov, struct iovec *siov,
 328     size_t iov_cnt, size_t *iov_len)
 329 {
 330         struct iovec *iovp;
 331         int error, i;
 332
 333         if (iov_cnt >= UIO_MAXIOV)
 334                 return EMSGSIZE;
 335         if (iov_cnt >= UIO_SMALLIOV) {
 336                 MALLOC(*kiov, struct iovec *, sizeof(struct iovec) * iov_cnt,
 337                     M_IOV, M_WAITOK);
 338         } else {
 339                 *kiov = siov;
 340         }
 341         error = copyin(uiov, *kiov, iov_cnt * sizeof(struct iovec));
 342         if (error)
 343                 goto cleanup;
 344         *iov_len = 0;
 345         for (i = 0, iovp = *kiov; i < iov_cnt; i++, iovp++)
 346                 *iov_len += iovp->iov_len;
 347
 348 cleanup:
 349         if (error)
 350                 iovec_free(kiov, siov);
 351         return (error);
 352 }