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 1.27 2007/01/29 20:44:02 tgen 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 <sys/sfbuf.h>
  56 #include <sys/thread2.h>
  57 #include <machine/limits.h>
  58
  59 #include <vm/vm.h>
  60 #include <vm/vm_page.h>
  61 #include <vm/vm_map.h>
  62
  63 SYSCTL_INT(_kern, KERN_IOV_MAX, iov_max, CTLFLAG_RD, NULL, UIO_MAXIOV,
  64         "Maximum number of elements in an I/O vector; sysconf(_SC_IOV_MAX)");
  65
  66 /*
  67  * UIO_READ:    copy the kernelspace cp to the user or kernelspace UIO
  68  * UIO_WRITE:   copy the user or kernelspace UIO to the kernelspace cp
  69  *
  70  * For userspace UIO's, uio_td must be the current thread.
  71  *
  72  * The syscall interface is responsible for limiting the length to
  73  * ssize_t for things like read() or write() which return the bytes
  74  * read or written as ssize_t.  These functions work with unsigned
  75  * lengths.
  76  */
  77 int
  78 uiomove(caddr_t cp, size_t n, struct uio *uio)
  79 {
  80         struct iovec *iov;
  81         size_t cnt;
  82         int error = 0;
  83         int save = 0;
  84         int baseticks = ticks;
  85
  86         KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
  87             ("uiomove: mode"));
  88         KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
  89             ("uiomove proc"));
  90
  91         if (curproc) {
  92                 save = curproc->p_flag & P_DEADLKTREAT;
  93                 curproc->p_flag |= P_DEADLKTREAT;
  94         }
  95
  96         while (n > 0 && uio->uio_resid) {
  97                 iov = uio->uio_iov;
  98                 cnt = iov->iov_len;
  99                 if (cnt == 0) {
 100                         uio->uio_iov++;
 101                         uio->uio_iovcnt--;
 102                         continue;
 103                 }
 104                 if (cnt > n)
 105                         cnt = n;
 106
 107                 switch (uio->uio_segflg) {
 108
 109                 case UIO_USERSPACE:
 110                         if (ticks - baseticks >= hogticks) {
 111                                 uio_yield();
 112                                 baseticks = ticks;
 113                         }
 114                         if (uio->uio_rw == UIO_READ)
 115                                 error = copyout(cp, iov->iov_base, cnt);
 116                         else
 117                                 error = copyin(iov->iov_base, cp, cnt);
 118                         if (error)
 119                                 break;
 120                         break;
 121
 122                 case UIO_SYSSPACE:
 123                         if (uio->uio_rw == UIO_READ)
 124                                 bcopy((caddr_t)cp, iov->iov_base, cnt);
 125                         else
 126                                 bcopy(iov->iov_base, (caddr_t)cp, cnt);
 127                         break;
 128                 case UIO_NOCOPY:
 129                         break;
 130                 }
 131                 iov->iov_base = (char *)iov->iov_base + cnt;
 132                 iov->iov_len -= cnt;
 133                 uio->uio_resid -= cnt;
 134                 uio->uio_offset += cnt;
 135                 cp += cnt;
 136                 n -= cnt;
 137         }
 138         if (curproc)
 139                 curproc->p_flag = (curproc->p_flag & ~P_DEADLKTREAT) | save;
 140         return (error);
 141 }
 142 /*
 143  * Wrapper for uiomove() that validates the arguments against a known-good
 144  * kernel buffer.
 145  */
 146 int
 147 uiomove_frombuf(void *buf, size_t buflen, struct uio *uio)
 148 {
 149         size_t offset;
 150
 151         offset = (size_t)uio->uio_offset;
 152         if ((off_t)offset != uio->uio_offset)
 153                 return (EINVAL);
 154         if (buflen == 0 || offset >= buflen)
 155                 return (0);
 156         return (uiomove((char *)buf + offset, buflen - offset, uio));
 157 }
 158
 159 /*
 160  * Give next character to user as result of read.
 161  */
 162 int
 163 ureadc(int c, struct uio *uio)
 164 {
 165         struct iovec *iov;
 166         char *iov_base;
 167
 168 again:
 169         if (uio->uio_iovcnt == 0 || uio->uio_resid == 0)
 170                 panic("ureadc");
 171         iov = uio->uio_iov;
 172         if (iov->iov_len == 0) {
 173                 uio->uio_iovcnt--;
 174                 uio->uio_iov++;
 175                 goto again;
 176         }
 177         switch (uio->uio_segflg) {
 178
 179         case UIO_USERSPACE:
 180                 if (subyte(iov->iov_base, c) < 0)
 181                         return (EFAULT);
 182                 break;
 183
 184         case UIO_SYSSPACE:
 185                 iov_base = iov->iov_base;
 186                 *iov_base = c;
 187                 iov->iov_base = iov_base;
 188                 break;
 189
 190         case UIO_NOCOPY:
 191                 break;
 192         }
 193         iov->iov_base = (char *)iov->iov_base + 1;
 194         iov->iov_len--;
 195         uio->uio_resid--;
 196         uio->uio_offset++;
 197         return (0);
 198 }
 199
 200 /*
 201  * General routine to allocate a hash table.  Make the hash table size a
 202  * power of 2 greater or equal to the number of elements requested, and
 203  * store the masking value in *hashmask.
 204  */
 205 void *
 206 hashinit(int elements, struct malloc_type *type, u_long *hashmask)
 207 {
 208         long hashsize;
 209         LIST_HEAD(generic, generic) *hashtbl;
 210         int i;
 211
 212         if (elements <= 0)
 213                 panic("hashinit: bad elements");
 214         for (hashsize = 2; hashsize < elements; hashsize <<= 1)
 215                 continue;
 216         hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
 217         for (i = 0; i < hashsize; i++)
 218                 LIST_INIT(&hashtbl[i]);
 219         *hashmask = hashsize - 1;
 220         return (hashtbl);
 221 }
 222
 223 static int primes[] = { 1, 13, 31, 61, 127, 251, 509, 761, 1021, 1531, 2039,
 224                         2557, 3067, 3583, 4093, 4603, 5119, 5623, 6143, 6653,
 225                         7159, 7673, 8191, 12281, 16381, 24571, 32749 };
 226 #define NPRIMES (sizeof(primes) / sizeof(primes[0]))
 227
 228 /*
 229  * General routine to allocate a prime number sized hash table.
 230  */
 231 void *
 232 phashinit(int elements, struct malloc_type *type, u_long *nentries)
 233 {
 234         long hashsize;
 235         LIST_HEAD(generic, generic) *hashtbl;
 236         int i;
 237
 238         if (elements <= 0)
 239                 panic("phashinit: bad elements");
 240         for (i = 1, hashsize = primes[1]; hashsize <= elements;) {
 241                 i++;
 242                 if (i == NPRIMES)
 243                         break;
 244                 hashsize = primes[i];
 245         }
 246         hashsize = primes[i - 1];
 247         hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
 248         for (i = 0; i < hashsize; i++)
 249                 LIST_INIT(&hashtbl[i]);
 250         *nentries = hashsize;
 251         return (hashtbl);
 252 }
 253
 254 /*
 255  * Copyin an iovec.  If the iovec array fits, use the preallocated small
 256  * iovec structure.  If it is too big, dynamically allocate an iovec array
 257  * of sufficient size.
 258  *
 259  * MPSAFE
 260  */
 261 int
 262 iovec_copyin(struct iovec *uiov, struct iovec **kiov, struct iovec *siov,
 263              size_t iov_cnt, size_t *iov_len)
 264 {
 265         struct iovec *iovp;
 266         int error, i;
 267         size_t len;
 268
 269         if (iov_cnt > UIO_MAXIOV)
 270                 return EMSGSIZE;
 271         if (iov_cnt > UIO_SMALLIOV) {
 272                 MALLOC(*kiov, struct iovec *, sizeof(struct iovec) * iov_cnt,
 273                     M_IOV, M_WAITOK);
 274         } else {
 275                 *kiov = siov;
 276         }
 277         error = copyin(uiov, *kiov, iov_cnt * sizeof(struct iovec));
 278         if (error == 0) {
 279                 *iov_len = 0;
 280                 for (i = 0, iovp = *kiov; i < iov_cnt; i++, iovp++) {
 281                         /*
 282                          * Check for both *iov_len overflows and out of
 283                          * range iovp->iov_len's.  We limit to the
 284                          * capabilities of signed integers.
 285                          *
 286                          * GCC4 - overflow check opt requires assign/test.
 287                          */
 288                         len = *iov_len + iovp->iov_len;
 289                         if (len < *iov_len)
 290                                 error = EINVAL;
 291                         *iov_len = len;
 292                 }
 293         }
 294
 295         /*
 296          * From userland disallow iovec's which exceed the sized size
 297          * limit as the system calls return ssize_t.
 298          *
 299          * NOTE: Internal kernel interfaces can handle the unsigned
 300          *       limit.
 301          */
 302         if (error == 0 && (ssize_t)*iov_len < 0)
 303                 error = EINVAL;
 304
 305         if (error)
 306                 iovec_free(kiov, siov);
 307         return (error);
 308 }
 309
 310
 311 /*
 312  * Copyright (c) 2004 Alan L. Cox <alc@cs.rice.edu>
 313  * Copyright (c) 1982, 1986, 1991, 1993
 314  *      The Regents of the University of California.  All rights reserved.
 315  * (c) UNIX System Laboratories, Inc.
 316  * All or some portions of this file are derived from material licensed
 317  * to the University of California by American Telephone and Telegraph
 318  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
 319  * the permission of UNIX System Laboratories, Inc.
 320  *
 321  * Redistribution and use in source and binary forms, with or without
 322  * modification, are permitted provided that the following conditions
 323  * are met:
 324  * 1. Redistributions of source code must retain the above copyright
 325  *    notice, this list of conditions and the following disclaimer.
 326  * 2. Redistributions in binary form must reproduce the above copyright
 327  *    notice, this list of conditions and the following disclaimer in the
 328  *    documentation and/or other materials provided with the distribution.
 329  * 4. Neither the name of the University nor the names of its contributors
 330  *    may be used to endorse or promote products derived from this software
 331  *    without specific prior written permission.
 332  *
 333  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 334  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 335  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 336  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 337  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 338  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 339  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 340  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 341  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 342  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 343  * SUCH DAMAGE.
 344  *
 345  * @(#)kern_subr.c      8.3 (Berkeley) 1/21/94
 346  * $FreeBSD: src/sys/i386/i386/uio_machdep.c,v 1.1 2004/03/21 20:28:36 alc Exp $
 347  */
 348
 349 /*
 350  * Implement uiomove(9) from physical memory using sf_bufs to reduce
 351  * the creation and destruction of ephemeral mappings.
 352  */
 353 int
 354 uiomove_fromphys(vm_page_t *ma, vm_offset_t offset, size_t n, struct uio *uio)
 355 {
 356         struct sf_buf *sf;
 357         struct thread *td = curthread;
 358         struct iovec *iov;
 359         void *cp;
 360         vm_offset_t page_offset;
 361         vm_page_t m;
 362         size_t cnt;
 363         int error = 0;
 364         int save = 0;
 365
 366         KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
 367             ("uiomove_fromphys: mode"));
 368         KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
 369             ("uiomove_fromphys proc"));
 370
 371         crit_enter();
 372         save = td->td_flags & TDF_DEADLKTREAT;
 373         td->td_flags |= TDF_DEADLKTREAT;
 374         crit_exit();
 375
 376         while (n > 0 && uio->uio_resid) {
 377                 iov = uio->uio_iov;
 378                 cnt = iov->iov_len;
 379                 if (cnt == 0) {
 380                         uio->uio_iov++;
 381                         uio->uio_iovcnt--;
 382                         continue;
 383                 }
 384                 if (cnt > n)
 385                         cnt = n;
 386                 page_offset = offset & PAGE_MASK;
 387                 cnt = min(cnt, PAGE_SIZE - page_offset);
 388                 m = ma[offset >> PAGE_SHIFT];
 389                 sf = sf_buf_alloc(m, SFB_CPUPRIVATE);
 390                 cp = (char *)sf_buf_kva(sf) + page_offset;
 391                 switch (uio->uio_segflg) {
 392                 case UIO_USERSPACE:
 393                         /*
 394                          * note: removed uioyield (it was the wrong place to
 395                          * put it).
 396                          */
 397                         if (uio->uio_rw == UIO_READ)
 398                                 error = copyout(cp, iov->iov_base, cnt);
 399                         else
 400                                 error = copyin(iov->iov_base, cp, cnt);
 401                         if (error) {
 402                                 sf_buf_free(sf);
 403                                 goto out;
 404                         }
 405                         break;
 406                 case UIO_SYSSPACE:
 407                         if (uio->uio_rw == UIO_READ)
 408                                 bcopy(cp, iov->iov_base, cnt);
 409                         else
 410                                 bcopy(iov->iov_base, cp, cnt);
 411                         break;
 412                 case UIO_NOCOPY:
 413                         break;
 414                 }
 415                 sf_buf_free(sf);
 416                 iov->iov_base = (char *)iov->iov_base + cnt;
 417                 iov->iov_len -= cnt;
 418                 uio->uio_resid -= cnt;
 419                 uio->uio_offset += cnt;
 420                 offset += cnt;
 421                 n -= cnt;
 422         }
 423 out:
 424         if (save == 0) {
 425                 crit_enter();
 426                 td->td_flags &= ~TDF_DEADLKTREAT;
 427                 crit_exit();
 428         }
 429         return (error);
 430 }
 431