bsd/kern/kern_physio.c

   1 /*
   2  * Copyright (c) 2000-2004 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /*-
  29  * Copyright (c) 1982, 1986, 1990, 1993
  30  *      The Regents of the University of California.  All rights reserved.
  31  * (c) UNIX System Laboratories, Inc.
  32  * All or some portions of this file are derived from material licensed
  33  * to the University of California by American Telephone and Telegraph
  34  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
  35  * the permission of UNIX System Laboratories, Inc.
  36  *
  37  * Redistribution and use in source and binary forms, with or without
  38  * modification, are permitted provided that the following conditions
  39  * are met:
  40  * 1. Redistributions of source code must retain the above copyright
  41  *    notice, this list of conditions and the following disclaimer.
  42  * 2. Redistributions in binary form must reproduce the above copyright
  43  *    notice, this list of conditions and the following disclaimer in the
  44  *    documentation and/or other materials provided with the distribution.
  45  * 3. All advertising materials mentioning features or use of this software
  46  *    must display the following acknowledgement:
  47  *      This product includes software developed by the University of
  48  *      California, Berkeley and its contributors.
  49  * 4. Neither the name of the University nor the names of its contributors
  50  *    may be used to endorse or promote products derived from this software
  51  *    without specific prior written permission.
  52  *
  53  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  54  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  55  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  56  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  57  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  58  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  59  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  61  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  62  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  63  * SUCH DAMAGE.
  64  *
  65  *      from: @(#)kern_physio.c 8.1 (Berkeley) 6/10/93
  66  */
  67 /*
  68  * HISTORY
  69  * 27-July-97  Umesh Vaishampayan  (umeshv@apple.com)
  70  *      Allow physio() to kernel space.
  71  */
  72
  73 #include <sys/param.h>
  74 #include <sys/systm.h>
  75 #include <sys/buf_internal.h>
  76 #include <sys/conf.h>
  77 #include <sys/proc_internal.h>
  78 #include <sys/uio_internal.h>
  79 #include <kern/assert.h>
  80
  81 int
  82 physio( void (*f_strategy)(buf_t),
  83     buf_t bp,
  84     dev_t dev,
  85     int flags,
  86     u_int (*f_minphys)(buf_t),
  87     struct uio *uio,
  88     int blocksize)
  89 {
  90         struct proc *p = current_proc();
  91         int error, i, buf_allocated, todo, iosize;
  92         int orig_bflags = 0;
  93         int64_t done;
  94
  95         error = 0;
  96         flags &= B_READ | B_WRITE;
  97         buf_allocated = 0;
  98
  99         /*
 100          * [check user read/write access to the data buffer]
 101          *
 102          * Check each iov one by one.  Note that we know if we're reading or
 103          * writing, so we ignore the uio's rw parameter.  Also note that if
 104          * we're doing a read, that's a *write* to user-space.
 105          */
 106         for (i = 0; i < uio->uio_iovcnt; i++) {
 107                 if (UIO_SEG_IS_USER_SPACE(uio->uio_segflg)) {
 108                         user_addr_t base;
 109                         user_size_t len;
 110
 111                         if (uio_getiov(uio, i, &base, &len) ||
 112                             !useracc(base,
 113                             len,
 114                             (flags == B_READ) ? B_WRITE : B_READ)) {
 115                                 return EFAULT;
 116                         }
 117                 }
 118         }
 119         /*
 120          * Make sure we have a buffer, creating one if necessary.
 121          */
 122         if (bp == NULL) {
 123                 bp = buf_alloc((vnode_t)0);
 124                 buf_allocated = 1;
 125         } else {
 126                 orig_bflags = buf_flags(bp);
 127         }
 128         /*
 129          * at this point we should have a buffer
 130          * that is marked BL_BUSY... we either
 131          * acquired it via buf_alloc, or it was
 132          * passed into us... if it was passed
 133          * in, it needs to already be owned by
 134          * the caller (i.e. BL_BUSY is set)
 135          */
 136         assert(bp->b_lflags & BL_BUSY);
 137
 138         /*
 139          * [set up the fixed part of the buffer for a transfer]
 140          */
 141         bp->b_dev = dev;
 142         bp->b_proc = p;
 143
 144         /*
 145          * [mark the buffer busy for physical I/O]
 146          * (i.e. set B_PHYS (because it's an I/O to user
 147          * memory, and B_RAW, because B_RAW is to be
 148          * "Set by physio for raw transfers.", in addition
 149          * to the read/write flag.)
 150          */
 151         buf_setflags(bp, B_PHYS | B_RAW);
 152
 153         /*
 154          * [while there is data to transfer and no I/O error]
 155          * Note that I/O errors are handled with a 'goto' at the bottom
 156          * of the 'while' loop.
 157          */
 158         while (uio_resid(uio) > 0) {
 159                 if ((iosize = uio_curriovlen(uio)) > MAXPHYSIO_WIRED) {
 160                         iosize = MAXPHYSIO_WIRED;
 161                 }
 162                 /*
 163                  * make sure we're set to issue a fresh I/O
 164                  * in the right direction
 165                  */
 166                 buf_reset(bp, flags);
 167
 168                 /* [set up the buffer for a maximum-sized transfer] */
 169                 buf_setblkno(bp, uio_offset(uio) / blocksize);
 170                 buf_setcount(bp, iosize);
 171                 buf_setdataptr(bp, (uintptr_t)CAST_DOWN(caddr_t, uio_curriovbase(uio)));
 172
 173                 /*
 174                  * [call f_minphys to bound the tranfer size]
 175                  * and remember the amount of data to transfer,
 176                  * for later comparison.
 177                  */
 178                 (*f_minphys)(bp);
 179                 todo = buf_count(bp);
 180
 181                 /*
 182                  * [lock the part of the user address space involved
 183                  *    in the transfer]
 184                  */
 185
 186                 if (UIO_SEG_IS_USER_SPACE(uio->uio_segflg)) {
 187                         error = vslock(CAST_USER_ADDR_T(buf_dataptr(bp)),
 188                             (user_size_t)todo);
 189                         if (error) {
 190                                 goto done;
 191                         }
 192                 }
 193
 194                 /* [call f_strategy to start the transfer] */
 195                 (*f_strategy)(bp);
 196
 197
 198                 /* [wait for the transfer to complete] */
 199                 error = (int)buf_biowait(bp);
 200
 201                 /*
 202                  * [unlock the part of the address space previously
 203                  *    locked]
 204                  */
 205                 if (UIO_SEG_IS_USER_SPACE(uio->uio_segflg)) {
 206                         vsunlock(CAST_USER_ADDR_T(buf_dataptr(bp)),
 207                             (user_size_t)todo,
 208                             (flags & B_READ));
 209                 }
 210
 211                 /*
 212                  * [deduct the transfer size from the total number
 213                  *    of data to transfer]
 214                  */
 215                 done = buf_count(bp) - buf_resid(bp);
 216                 uio_update(uio, done);
 217
 218                 /*
 219                  * Now, check for an error.
 220                  * Also, handle weird end-of-disk semantics.
 221                  */
 222                 if (error || done < todo) {
 223                         goto done;
 224                 }
 225         }
 226
 227 done:
 228         if (buf_allocated) {
 229                 buf_free(bp);
 230         } else {
 231                 buf_setflags(bp, orig_bflags);
 232         }
 233
 234         return error;
 235 }
 236
 237 /*
 238  * Leffler, et al., says on p. 231:
 239  * "The minphys() routine is called by physio() to adjust the
 240  * size of each I/O transfer before the latter is passed to
 241  * the strategy routine..."
 242  *
 243  * so, just adjust the buffer's count accounting to MAXPHYS here,
 244  * and return the new count;
 245  */
 246 u_int
 247 minphys(struct buf *bp)
 248 {
 249         buf_setcount(bp, min(MAXPHYS, buf_count(bp)));
 250         return buf_count(bp);
 251 }