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 rawread(dev_t dev, struct uio *uio);
  82 int rawwrite(dev_t dev, struct uio *uio);
  83
  84 int
  85 physio( void (*f_strategy)(buf_t),
  86         buf_t bp,
  87         dev_t dev,
  88         int flags,
  89         u_int (*f_minphys)(buf_t),
  90         struct uio *uio,
  91         int blocksize)
  92 {
  93         struct proc *p = current_proc();
  94         int error, i, buf_allocated, todo, iosize;
  95         int orig_bflags = 0;
  96 #if LP64KERN
  97         int64_t done;
  98 #else
  99         int done;
 100 #endif
 101
 102         error = 0;
 103         flags &= B_READ | B_WRITE;
 104         buf_allocated = 0;
 105
 106         /*
 107          * [check user read/write access to the data buffer]
 108          *
 109          * Check each iov one by one.  Note that we know if we're reading or
 110          * writing, so we ignore the uio's rw parameter.  Also note that if
 111          * we're doing a read, that's a *write* to user-space.
 112          */
 113         for (i = 0; i < uio->uio_iovcnt; i++) {
 114                 if (UIO_SEG_IS_USER_SPACE(uio->uio_segflg)) {
 115                         if (!useracc(uio_iov_base_at(uio, i),
 116                                         uio_iov_len_at(uio, i),
 117                                 (flags == B_READ) ? B_WRITE : B_READ))
 118                         return (EFAULT);
 119                 }
 120         }
 121         /*
 122          * Make sure we have a buffer, creating one if necessary.
 123          */
 124         if (bp == NULL) {
 125                 bp = buf_alloc((vnode_t)0);
 126                 buf_allocated = 1;
 127         } else
 128                 orig_bflags = buf_flags(bp);
 129         /*
 130          * at this point we should have a buffer
 131          * that is marked BL_BUSY... we either
 132          * acquired it via buf_alloc, or it was
 133          * passed into us... if it was passed
 134          * in, it needs to already be owned by
 135          * the caller (i.e. BL_BUSY is set)
 136          */
 137         assert(bp->b_lflags & BL_BUSY);
 138
 139         /*
 140          * [set up the fixed part of the buffer for a transfer]
 141          */
 142         bp->b_dev = dev;
 143         bp->b_proc = p;
 144
 145         /*
 146          * [mark the buffer busy for physical I/O]
 147          * (i.e. set B_PHYS (because it's an I/O to user
 148          * memory, and B_RAW, because B_RAW is to be
 149          * "Set by physio for raw transfers.", in addition
 150          * to the read/write flag.)
 151          */
 152         buf_setflags(bp, B_PHYS | B_RAW);
 153
 154         /*
 155          * [while there is data to transfer and no I/O error]
 156          * Note that I/O errors are handled with a 'goto' at the bottom
 157          * of the 'while' loop.
 158          */
 159         for (i = 0; i < uio->uio_iovcnt; i++) {
 160                 while (uio_iov_len_at(uio, i) > 0) {
 161
 162                         if ( (iosize = uio_iov_len_at(uio, i)) > MAXPHYSIO_WIRED)
 163                                 iosize = MAXPHYSIO_WIRED;
 164                         /*
 165                          * make sure we're set to issue a fresh I/O
 166                          * in the right direction
 167                          */
 168                         buf_reset(bp, flags);
 169
 170                         /* [set up the buffer for a maximum-sized transfer] */
 171                         buf_setblkno(bp, uio->uio_offset / blocksize);
 172                         buf_setcount(bp, iosize);
 173                         // LP64todo - fix this!
 174                         buf_setdataptr(bp, (uintptr_t)CAST_DOWN(caddr_t, uio_iov_base_at(uio, i)));
 175
 176                         /*
 177                          * [call f_minphys to bound the tranfer size]
 178                          * and remember the amount of data to transfer,
 179                          * for later comparison.
 180                          */
 181                         (*f_minphys)(bp);
 182                         todo = buf_count(bp);
 183
 184                         /*
 185                          * [lock the part of the user address space involved
 186                          *    in the transfer]
 187                          */
 188
 189                         if(UIO_SEG_IS_USER_SPACE(uio->uio_segflg)) {
 190                                 error = vslock(CAST_USER_ADDR_T(buf_dataptr(bp)),
 191                                                (user_size_t)todo);
 192                                 if (error)
 193                                         goto done;
 194                         }
 195
 196                         /* [call f_strategy to start the transfer] */
 197                         (*f_strategy)(bp);
 198
 199
 200                         /* [wait for the transfer to complete] */
 201                         error = (int)buf_biowait(bp);
 202
 203                         /*
 204                          * [unlock the part of the address space previously
 205                          *    locked]
 206                          */
 207                         if(UIO_SEG_IS_USER_SPACE(uio->uio_segflg))
 208                                 vsunlock(CAST_USER_ADDR_T(buf_dataptr(bp)),
 209                                          (user_size_t)todo,
 210                                          (flags & B_READ));
 211
 212                         /*
 213                          * [deduct the transfer size from the total number
 214                          *    of data to transfer]
 215                          */
 216                         done = buf_count(bp) - buf_resid(bp);
 217                         uio_iov_len_add_at(uio, -done, i);
 218                         uio_iov_base_add_at(uio, done, i);
 219                         uio->uio_offset += done;
 220                         uio_setresid(uio, (uio_resid(uio) - done));
 221
 222                         /*
 223                          * Now, check for an error.
 224                          * Also, handle weird end-of-disk semantics.
 225                          */
 226                         if (error || done < todo)
 227                                 goto done;
 228                 }
 229         }
 230
 231 done:
 232         if (buf_allocated)
 233                 buf_free(bp);
 234         else
 235                 buf_setflags(bp, orig_bflags);
 236
 237         return (error);
 238 }
 239
 240 /*
 241  * Leffler, et al., says on p. 231:
 242  * "The minphys() routine is called by physio() to adjust the
 243  * size of each I/O transfer before the latter is passed to
 244  * the strategy routine..."
 245  *
 246  * so, just adjust the buffer's count accounting to MAXPHYS here,
 247  * and return the new count;
 248  */
 249 u_int
 250 minphys(struct buf *bp)
 251 {
 252
 253         buf_setcount(bp, min(MAXPHYS, buf_count(bp)));
 254         return buf_count(bp);
 255 }
 256
 257 /*
 258  * Do a read on a device for a user process.
 259  */
 260 int
 261 rawread(dev_t dev, struct uio *uio)
 262 {
 263         return (physio(cdevsw[major(dev)].d_strategy, (struct buf *)NULL,
 264             dev, B_READ, minphys, uio, DEV_BSIZE));
 265 }
 266
 267 /*
 268  * Do a write on a device for a user process.
 269  */
 270 int
 271 rawwrite(dev_t dev, struct uio *uio)
 272 {
 273         return (physio(cdevsw[major(dev)].d_strategy, (struct buf *)NULL,
 274             dev, B_WRITE, minphys, uio, DEV_BSIZE));
 275 }