bsd/kern/kern_physio.c

   1 /*
   2  * Copyright (c) 2000-2004 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_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. Please obtain a copy of the License at
  10  * http://www.opensource.apple.com/apsl/ and read it before using this
  11  * file.
  12  *
  13  * The Original Code and all software distributed under the License are
  14  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  15  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  16  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  18  * Please see the License for the specific language governing rights and
  19  * limitations under the License.
  20  *
  21  * @APPLE_LICENSE_HEADER_END@
  22  */
  23 /*-
  24  * Copyright (c) 1982, 1986, 1990, 1993
  25  *      The Regents of the University of California.  All rights reserved.
  26  * (c) UNIX System Laboratories, Inc.
  27  * All or some portions of this file are derived from material licensed
  28  * to the University of California by American Telephone and Telegraph
  29  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
  30  * the permission of UNIX System Laboratories, Inc.
  31  *
  32  * Redistribution and use in source and binary forms, with or without
  33  * modification, are permitted provided that the following conditions
  34  * are met:
  35  * 1. Redistributions of source code must retain the above copyright
  36  *    notice, this list of conditions and the following disclaimer.
  37  * 2. Redistributions in binary form must reproduce the above copyright
  38  *    notice, this list of conditions and the following disclaimer in the
  39  *    documentation and/or other materials provided with the distribution.
  40  * 3. All advertising materials mentioning features or use of this software
  41  *    must display the following acknowledgement:
  42  *      This product includes software developed by the University of
  43  *      California, Berkeley and its contributors.
  44  * 4. Neither the name of the University nor the names of its contributors
  45  *    may be used to endorse or promote products derived from this software
  46  *    without specific prior written permission.
  47  *
  48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  58  * SUCH DAMAGE.
  59  *
  60  *      from: @(#)kern_physio.c 8.1 (Berkeley) 6/10/93
  61  */
  62 /*
  63  * HISTORY
  64  * 27-July-97  Umesh Vaishampayan  (umeshv@apple.com)
  65  *      Allow physio() to kernel space.
  66  */
  67
  68 #include <sys/param.h>
  69 #include <sys/systm.h>
  70 #include <sys/buf_internal.h>
  71 #include <sys/conf.h>
  72 #include <sys/proc_internal.h>
  73 #include <sys/uio_internal.h>
  74
  75 int
  76 physio(strategy, bp, dev, flags, minphys, uio, blocksize)
  77         void (*strategy)();
  78         buf_t bp;
  79         dev_t dev;
  80         int flags;
  81         u_int (*minphys)();
  82         struct uio *uio;
  83         int blocksize;
  84 {
  85         struct proc *p = current_proc();
  86         int error, i, nobuf, todo, iosize;
  87 #if LP64KERN
  88         int64_t done;
  89 #else
  90         int done;
  91 #endif
  92
  93         error = 0;
  94         flags &= B_READ | B_WRITE;
  95
  96         /*
  97          * [check user read/write access to the data buffer]
  98          *
  99          * Check each iov one by one.  Note that we know if we're reading or
 100          * writing, so we ignore the uio's rw parameter.  Also note that if
 101          * we're doing a read, that's a *write* to user-space.
 102          */
 103         for (i = 0; i < uio->uio_iovcnt; i++) {
 104                 if(UIO_SEG_IS_USER_SPACE(uio->uio_segflg)) {
 105                         if (!useracc(uio_iov_base_at(uio, i),
 106                                         uio_iov_len_at(uio, i),
 107                                 (flags == B_READ) ? B_WRITE : B_READ))
 108                         return (EFAULT);
 109                 }
 110         }
 111         /* Make sure we have a buffer, creating one if necessary. */
 112         if (nobuf = (bp == NULL)) {
 113                         bp = buf_alloc((vnode_t)0);
 114                 }
 115
 116         /* [while the buffer is marked busy] */
 117         while (((error = (int)buf_acquire(bp, 0, 0, 0)) == EAGAIN));
 118
 119         if (error) {
 120                         if (nobuf)
 121                                 buf_free(bp);
 122                 return (error);
 123         }
 124
 125         /* [set up the fixed part of the buffer for a transfer] */
 126         bp->b_dev = dev;
 127         bp->b_proc = p;
 128
 129         buf_seterror(bp, 0);
 130         /*
 131          * [while there is data to transfer and no I/O error]
 132          * Note that I/O errors are handled with a 'goto' at the bottom
 133          * of the 'while' loop.
 134          */
 135         for (i = 0; i < uio->uio_iovcnt; i++) {
 136                 while (uio_iov_len_at(uio, i) > 0) {
 137                         /*
 138                          * [mark the buffer busy for physical I/O]
 139                          * (i.e. set B_PHYS (because it's an I/O to user
 140                          * memory, and B_RAW, because B_RAW is to be
 141                          * "Set by physio for raw transfers.", in addition
 142                          * to the read/write flag.)
 143                          */
 144                         buf_setflags(bp, B_PHYS | B_RAW | flags);
 145
 146                         if ( (iosize = uio_iov_len_at(uio, i)) > MAXPHYSIO_WIRED)
 147                                 iosize = MAXPHYSIO_WIRED;
 148
 149                         /* [set up the buffer for a maximum-sized transfer] */
 150                         buf_setblkno(bp, uio->uio_offset / blocksize);
 151                         buf_setcount(bp, iosize);
 152                         // LP64todo - fix this!
 153                         buf_setdataptr(bp, CAST_DOWN(caddr_t, uio_iov_base_at(uio, i)));
 154
 155                         /*
 156                          * [call minphys to bound the tranfer size]
 157                          * and remember the amount of data to transfer,
 158                          * for later comparison.
 159                          */
 160                         (*minphys)(bp);
 161                         todo = buf_count(bp);
 162
 163                         /*
 164                          * [lock the part of the user address space involved
 165                          *    in the transfer]
 166                          */
 167
 168                         if(UIO_SEG_IS_USER_SPACE(uio->uio_segflg))
 169                                 vslock(CAST_USER_ADDR_T(buf_dataptr(bp)),
 170                                        (user_size_t)todo);
 171
 172                         /* [call strategy to start the transfer] */
 173                         (*strategy)(bp);
 174
 175
 176                         /* [wait for the transfer to complete] */
 177                         error = (int)buf_biowait(bp);
 178
 179                         /*
 180                          * [unlock the part of the address space previously
 181                          *    locked]
 182                          */
 183                         if(UIO_SEG_IS_USER_SPACE(uio->uio_segflg))
 184                                 vsunlock(CAST_USER_ADDR_T(buf_dataptr(bp)),
 185                                          (user_size_t)todo,
 186                                          (flags & B_READ));
 187
 188                         /*
 189                          * [deduct the transfer size from the total number
 190                          *    of data to transfer]
 191                          */
 192                         done = buf_count(bp) - buf_resid(bp);
 193                         uio_iov_len_add_at(uio, -done, i);
 194                         uio_iov_base_add_at(uio, done, i);
 195                         uio->uio_offset += done;
 196                         uio_setresid(uio, (uio_resid(uio) - done));
 197
 198                         /*
 199                          * Now, check for an error.
 200                          * Also, handle weird end-of-disk semantics.
 201                          */
 202                         if (error || done < todo)
 203                                 goto done;
 204                 }
 205         }
 206
 207 done:
 208         /*
 209          * [clean up the state of the buffer]
 210          * Remember if somebody wants it, so we can wake them up below.
 211          * Also, if we had to steal it, give it back.
 212          */
 213
 214         buf_clearflags(bp, B_PHYS | B_RAW);
 215         if (nobuf)
 216                 buf_free(bp);
 217         else
 218                 {
 219                         buf_drop(bp);
 220                 }
 221
 222         return (error);
 223 }
 224
 225 /*
 226  * Leffler, et al., says on p. 231:
 227  * "The minphys() routine is called by physio() to adjust the
 228  * size of each I/O transfer before the latter is passed to
 229  * the strategy routine..."
 230  *
 231  * so, just adjust the buffer's count accounting to MAXPHYS here,
 232  * and return the new count;
 233  */
 234 u_int
 235 minphys(bp)
 236         struct buf *bp;
 237 {
 238
 239         buf_setcount(bp, min(MAXPHYS, buf_count(bp)));
 240         return buf_count(bp);
 241 }
 242
 243 /*
 244  * Do a read on a device for a user process.
 245  */
 246 rawread(dev, uio)
 247         dev_t dev;
 248         struct uio *uio;
 249 {
 250         return (physio(cdevsw[major(dev)].d_strategy, (struct buf *)NULL,
 251             dev, B_READ, minphys, uio, DEV_BSIZE));
 252 }
 253
 254 /*
 255  * Do a write on a device for a user process.
 256  */
 257 rawwrite(dev, uio)
 258         dev_t dev;
 259         struct uio *uio;
 260 {
 261         return (physio(cdevsw[major(dev)].d_strategy, (struct buf *)NULL,
 262             dev, B_WRITE, minphys, uio, DEV_BSIZE));
 263 }