subr_disk.c revision 1.85.2.2
1 /* $NetBSD: subr_disk.c,v 1.85.2.2 2007/08/19 19:24:54 ad Exp $ */ 2 3 /*- 4 * Copyright (c) 1996, 1997, 1999, 2000 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, 9 * NASA Ames Research Center. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the NetBSD 22 * Foundation, Inc. and its contributors. 23 * 4. Neither the name of The NetBSD Foundation nor the names of its 24 * contributors may be used to endorse or promote products derived 25 * from this software without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 37 * POSSIBILITY OF SUCH DAMAGE. 38 */ 39 40 /* 41 * Copyright (c) 1982, 1986, 1988, 1993 42 * The Regents of the University of California. All rights reserved. 43 * (c) UNIX System Laboratories, Inc. 44 * All or some portions of this file are derived from material licensed 45 * to the University of California by American Telephone and Telegraph 46 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 47 * the permission of UNIX System Laboratories, Inc. 48 * 49 * Redistribution and use in source and binary forms, with or without 50 * modification, are permitted provided that the following conditions 51 * are met: 52 * 1. Redistributions of source code must retain the above copyright 53 * notice, this list of conditions and the following disclaimer. 54 * 2. Redistributions in binary form must reproduce the above copyright 55 * notice, this list of conditions and the following disclaimer in the 56 * documentation and/or other materials provided with the distribution. 57 * 3. Neither the name of the University nor the names of its contributors 58 * may be used to endorse or promote products derived from this software 59 * without specific prior written permission. 60 * 61 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 62 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 63 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 64 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 65 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 66 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 67 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 68 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 69 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 70 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 71 * SUCH DAMAGE. 72 * 73 * @(#)ufs_disksubr.c 8.5 (Berkeley) 1/21/94 74 */ 75 76 #include <sys/cdefs.h> 77 __KERNEL_RCSID(0, "$NetBSD: subr_disk.c,v 1.85.2.2 2007/08/19 19:24:54 ad Exp $"); 78 79 #include <sys/param.h> 80 #include <sys/kernel.h> 81 #include <sys/malloc.h> 82 #include <sys/buf.h> 83 #include <sys/syslog.h> 84 #include <sys/disklabel.h> 85 #include <sys/disk.h> 86 #include <sys/sysctl.h> 87 #include <lib/libkern/libkern.h> 88 89 /* 90 * Compute checksum for disk label. 91 */ 92 u_int 93 dkcksum(struct disklabel *lp) 94 { 95 return dkcksum_sized(lp, lp->d_npartitions); 96 } 97 98 u_int 99 dkcksum_sized(struct disklabel *lp, size_t npartitions) 100 { 101 u_short *start, *end; 102 u_short sum = 0; 103 104 start = (u_short *)lp; 105 end = (u_short *)&lp->d_partitions[npartitions]; 106 while (start < end) 107 sum ^= *start++; 108 return (sum); 109 } 110 111 /* 112 * Disk error is the preface to plaintive error messages 113 * about failing disk transfers. It prints messages of the form 114 115 hp0g: hard error reading fsbn 12345 of 12344-12347 (hp0 bn %d cn %d tn %d sn %d) 116 117 * if the offset of the error in the transfer and a disk label 118 * are both available. blkdone should be -1 if the position of the error 119 * is unknown; the disklabel pointer may be null from drivers that have not 120 * been converted to use them. The message is printed with printf 121 * if pri is LOG_PRINTF, otherwise it uses log at the specified priority. 122 * The message should be completed (with at least a newline) with printf 123 * or addlog, respectively. There is no trailing space. 124 */ 125 #ifndef PRIdaddr 126 #define PRIdaddr PRId64 127 #endif 128 void 129 diskerr(const struct buf *bp, const char *dname, const char *what, int pri, 130 int blkdone, const struct disklabel *lp) 131 { 132 int unit = DISKUNIT(bp->b_dev), part = DISKPART(bp->b_dev); 133 void (*pr)(const char *, ...); 134 char partname = 'a' + part; 135 daddr_t sn; 136 137 if (/*CONSTCOND*/0) 138 /* Compiler will error this is the format is wrong... */ 139 printf("%" PRIdaddr, bp->b_blkno); 140 141 if (pri != LOG_PRINTF) { 142 static const char fmt[] = ""; 143 log(pri, fmt); 144 pr = addlog; 145 } else 146 pr = printf; 147 (*pr)("%s%d%c: %s %sing fsbn ", dname, unit, partname, what, 148 bp->b_flags & B_READ ? "read" : "writ"); 149 sn = bp->b_blkno; 150 if (bp->b_bcount <= DEV_BSIZE) 151 (*pr)("%" PRIdaddr, sn); 152 else { 153 if (blkdone >= 0) { 154 sn += blkdone; 155 (*pr)("%" PRIdaddr " of ", sn); 156 } 157 (*pr)("%" PRIdaddr "-%" PRIdaddr "", bp->b_blkno, 158 bp->b_blkno + (bp->b_bcount - 1) / DEV_BSIZE); 159 } 160 if (lp && (blkdone >= 0 || bp->b_bcount <= lp->d_secsize)) { 161 sn += lp->d_partitions[part].p_offset; 162 (*pr)(" (%s%d bn %" PRIdaddr "; cn %" PRIdaddr "", 163 dname, unit, sn, sn / lp->d_secpercyl); 164 sn %= lp->d_secpercyl; 165 (*pr)(" tn %" PRIdaddr " sn %" PRIdaddr ")", 166 sn / lp->d_nsectors, sn % lp->d_nsectors); 167 } 168 } 169 170 /* 171 * Searches the iostatlist for the disk corresponding to the 172 * name provided. 173 */ 174 struct disk * 175 disk_find(const char *name) 176 { 177 struct io_stats *stat; 178 179 stat = iostat_find(name); 180 181 if ((stat != NULL) && (stat->io_type == IOSTAT_DISK)) 182 return stat->io_parent; 183 184 return (NULL); 185 } 186 187 static void 188 disk_init0(struct disk *diskp) 189 { 190 191 /* 192 * Initialize the wedge-related locks and other fields. 193 */ 194 lockinit(&diskp->dk_rawlock, PRIBIO, "dkrawlk", 0, 0); 195 lockinit(&diskp->dk_openlock, PRIBIO, "dkoplk", 0, 0); 196 LIST_INIT(&diskp->dk_wedges); 197 diskp->dk_nwedges = 0; 198 diskp->dk_labelsector = LABELSECTOR; 199 disk_blocksize(diskp, DEV_BSIZE); 200 } 201 202 static void 203 disk_attach0(struct disk *diskp) 204 { 205 206 /* 207 * Allocate and initialize the disklabel structures. Note that 208 * it's not safe to sleep here, since we're probably going to be 209 * called during autoconfiguration. 210 */ 211 diskp->dk_label = malloc(sizeof(struct disklabel), M_DEVBUF, M_NOWAIT); 212 diskp->dk_cpulabel = malloc(sizeof(struct cpu_disklabel), M_DEVBUF, 213 M_NOWAIT); 214 if ((diskp->dk_label == NULL) || (diskp->dk_cpulabel == NULL)) 215 panic("disk_attach: can't allocate storage for disklabel"); 216 217 memset(diskp->dk_label, 0, sizeof(struct disklabel)); 218 memset(diskp->dk_cpulabel, 0, sizeof(struct cpu_disklabel)); 219 220 /* 221 * Set up the stats collection. 222 */ 223 diskp->dk_stats = iostat_alloc(IOSTAT_DISK, diskp, diskp->dk_name); 224 } 225 226 static void 227 disk_detach0(struct disk *diskp) 228 { 229 230 lockdestroy(&diskp->dk_openlock); 231 lockdestroy(&diskp->dk_rawlock); 232 233 /* 234 * Remove from the drivelist. 235 */ 236 iostat_free(diskp->dk_stats); 237 238 /* 239 * Release the disk-info dictionary. 240 */ 241 if (diskp->dk_info) { 242 prop_object_release(diskp->dk_info); 243 diskp->dk_info = NULL; 244 } 245 246 /* 247 * Free the space used by the disklabel structures. 248 */ 249 free(diskp->dk_label, M_DEVBUF); 250 free(diskp->dk_cpulabel, M_DEVBUF); 251 } 252 253 /* 254 * Attach a disk. 255 */ 256 void 257 disk_attach(struct disk *diskp) 258 { 259 260 disk_init0(diskp); 261 disk_attach0(diskp); 262 } 263 264 /* 265 * Detach a disk. 266 */ 267 void 268 disk_detach(struct disk *diskp) 269 { 270 271 (void) lockmgr(&diskp->dk_openlock, LK_DRAIN, NULL); 272 (void) lockmgr(&diskp->dk_rawlock, LK_DRAIN, NULL); 273 disk_detach0(diskp); 274 } 275 276 /* 277 * Initialize a pseudo disk. 278 */ 279 void 280 pseudo_disk_init(struct disk *diskp) 281 { 282 283 disk_init0(diskp); 284 } 285 286 /* 287 * Attach a pseudo disk. 288 */ 289 void 290 pseudo_disk_attach(struct disk *diskp) 291 { 292 293 disk_attach0(diskp); 294 } 295 296 /* 297 * Detach a pseudo disk. 298 */ 299 void 300 pseudo_disk_detach(struct disk *diskp) 301 { 302 303 disk_detach0(diskp); 304 } 305 306 /* 307 * Mark the disk as busy for metrics collection. 308 */ 309 void 310 disk_busy(struct disk *diskp) 311 { 312 313 iostat_busy(diskp->dk_stats); 314 } 315 316 /* 317 * Finished disk operations, gather metrics. 318 */ 319 void 320 disk_unbusy(struct disk *diskp, long bcount, int read) 321 { 322 323 iostat_unbusy(diskp->dk_stats, bcount, read); 324 } 325 326 /* 327 * Set the physical blocksize of a disk, in bytes. 328 * Only necessary if blocksize != DEV_BSIZE. 329 */ 330 void 331 disk_blocksize(struct disk *diskp, int blocksize) 332 { 333 334 diskp->dk_blkshift = DK_BSIZE2BLKSHIFT(blocksize); 335 diskp->dk_byteshift = DK_BSIZE2BYTESHIFT(blocksize); 336 } 337 338 /* 339 * Bounds checking against the media size, used for the raw partition. 340 * The sector size passed in should currently always be DEV_BSIZE, 341 * and the media size the size of the device in DEV_BSIZE sectors. 342 */ 343 int 344 bounds_check_with_mediasize(struct buf *bp, int secsize, uint64_t mediasize) 345 { 346 int64_t sz; 347 348 sz = howmany(bp->b_bcount, secsize); 349 350 if (bp->b_blkno + sz > mediasize) { 351 sz = mediasize - bp->b_blkno; 352 if (sz == 0) { 353 /* If exactly at end of disk, return EOF. */ 354 bp->b_resid = bp->b_bcount; 355 return 0; 356 } 357 if (sz < 0) { 358 /* If past end of disk, return EINVAL. */ 359 bp->b_error = EINVAL; 360 return 0; 361 } 362 /* Otherwise, truncate request. */ 363 bp->b_bcount = sz << DEV_BSHIFT; 364 } 365 366 return 1; 367 } 368 369 /* 370 * Determine the size of the transfer, and make sure it is 371 * within the boundaries of the partition. Adjust transfer 372 * if needed, and signal errors or early completion. 373 */ 374 int 375 bounds_check_with_label(struct disk *dk, struct buf *bp, int wlabel) 376 { 377 struct disklabel *lp = dk->dk_label; 378 struct partition *p = lp->d_partitions + DISKPART(bp->b_dev); 379 uint64_t p_size, p_offset, labelsector; 380 int64_t sz; 381 382 /* Protect against division by zero. XXX: Should never happen?!?! */ 383 if (lp->d_secpercyl == 0) { 384 bp->b_error = EINVAL; 385 return -1; 386 } 387 388 p_size = p->p_size << dk->dk_blkshift; 389 p_offset = p->p_offset << dk->dk_blkshift; 390 #if RAW_PART == 3 391 labelsector = lp->d_partitions[2].p_offset; 392 #else 393 labelsector = lp->d_partitions[RAW_PART].p_offset; 394 #endif 395 labelsector = (labelsector + dk->dk_labelsector) << dk->dk_blkshift; 396 397 sz = howmany(bp->b_bcount, DEV_BSIZE); 398 if ((bp->b_blkno + sz) > p_size) { 399 sz = p_size - bp->b_blkno; 400 if (sz == 0) { 401 /* If exactly at end of disk, return EOF. */ 402 bp->b_resid = bp->b_bcount; 403 return 0; 404 } 405 if (sz < 0) { 406 /* If past end of disk, return EINVAL. */ 407 bp->b_error = EINVAL; 408 return -1; 409 } 410 /* Otherwise, truncate request. */ 411 bp->b_bcount = sz << DEV_BSHIFT; 412 } 413 414 /* Overwriting disk label? */ 415 if (bp->b_blkno + p_offset <= labelsector && 416 bp->b_blkno + p_offset + sz > labelsector && 417 (bp->b_flags & B_READ) == 0 && !wlabel) { 418 bp->b_error = EROFS; 419 return -1; 420 } 421 422 /* calculate cylinder for disksort to order transfers with */ 423 bp->b_cylinder = (bp->b_blkno + p->p_offset) / 424 (lp->d_secsize / DEV_BSIZE) / lp->d_secpercyl; 425 return 1; 426 } 427 428 int 429 disk_read_sectors(void (*strat)(struct buf *), const struct disklabel *lp, 430 struct buf *bp, unsigned int sector, int count) 431 { 432 bp->b_blkno = sector; 433 bp->b_bcount = count * lp->d_secsize; 434 bp->b_flags = (bp->b_flags & ~(B_WRITE | B_DONE)) | B_READ; 435 bp->b_cylinder = sector / lp->d_secpercyl; 436 (*strat)(bp); 437 return biowait(bp); 438 } 439 440 const char * 441 convertdisklabel(struct disklabel *lp, void (*strat)(struct buf *), 442 struct buf *bp, uint32_t secperunit) 443 { 444 struct partition rp, *altp, *p; 445 int geom_ok; 446 447 memset(&rp, 0, sizeof(rp)); 448 rp.p_size = secperunit; 449 rp.p_fstype = FS_UNUSED; 450 451 /* If we can seek to d_secperunit - 1, believe the disk geometry. */ 452 if (secperunit != 0 && 453 disk_read_sectors(strat, lp, bp, secperunit - 1, 1) == 0) 454 geom_ok = 1; 455 else 456 geom_ok = 0; 457 458 #if 0 459 printf("%s: secperunit (%" PRIu32 ") %s\n", __func__, 460 secperunit, geom_ok ? "ok" : "not ok"); 461 #endif 462 463 p = &lp->d_partitions[RAW_PART]; 464 if (RAW_PART == 'c' - 'a') 465 altp = &lp->d_partitions['d' - 'a']; 466 else 467 altp = &lp->d_partitions['c' - 'a']; 468 469 if (lp->d_npartitions > RAW_PART && p->p_offset == 0 && p->p_size != 0) 470 ; /* already a raw partition */ 471 else if (lp->d_npartitions > MAX('c', 'd') - 'a' && 472 altp->p_offset == 0 && altp->p_size != 0) { 473 /* alternate partition ('c' or 'd') is suitable for raw slot, 474 * swap with 'd' or 'c'. 475 */ 476 rp = *p; 477 *p = *altp; 478 *altp = rp; 479 } else if (lp->d_npartitions <= RAW_PART && 480 lp->d_npartitions > 'c' - 'a') { 481 /* No raw partition is present, but the alternate is present. 482 * Copy alternate to raw partition. 483 */ 484 lp->d_npartitions = RAW_PART + 1; 485 *p = *altp; 486 } else if (!geom_ok) 487 return "no raw partition and disk reports bad geometry"; 488 else if (lp->d_npartitions <= RAW_PART) { 489 memset(&lp->d_partitions[lp->d_npartitions], 0, 490 sizeof(struct partition) * (RAW_PART - lp->d_npartitions)); 491 *p = rp; 492 lp->d_npartitions = RAW_PART + 1; 493 } else if (lp->d_npartitions < MAXPARTITIONS) { 494 memmove(p + 1, p, 495 sizeof(struct partition) * (lp->d_npartitions - RAW_PART)); 496 *p = rp; 497 lp->d_npartitions++; 498 } else 499 return "no raw partition and partition table is full"; 500 return NULL; 501 } 502 503 /* 504 * disk_ioctl -- 505 * Generic disk ioctl handling. 506 */ 507 int 508 disk_ioctl(struct disk *diskp, u_long cmd, void *data, int flag, 509 struct lwp *l) 510 { 511 int error; 512 513 switch (cmd) { 514 case DIOCGDISKINFO: 515 { 516 struct plistref *pref = (struct plistref *) data; 517 518 if (diskp->dk_info == NULL) 519 error = ENOTSUP; 520 else 521 error = prop_dictionary_copyout_ioctl(pref, cmd, 522 diskp->dk_info); 523 break; 524 } 525 526 default: 527 error = EPASSTHROUGH; 528 } 529 530 return (error); 531 } 532
Indexes created Sat Sep 27 14:09:57 GMT 2025