1 /*- 2 * Copyright (c) 2007, 2008 Marcel Moolenaar 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 17 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 18 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 19 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 20 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 24 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 25 */ 26 27 #include <sys/cdefs.h> 28 __FBSDID("$FreeBSD$"); 29 30 #include <sys/param.h> 31 #include <sys/bio.h> 32 #include <sys/diskmbr.h> 33 #include <sys/endian.h> 34 #include <sys/kernel.h> 35 #include <sys/kobj.h> 36 #include <sys/limits.h> 37 #include <sys/lock.h> 38 #include <sys/malloc.h> 39 #include <sys/mutex.h> 40 #include <sys/queue.h> 41 #include <sys/sbuf.h> 42 #include <sys/systm.h> 43 #include <sys/sysctl.h> 44 #include <geom/geom.h> 45 #include <geom/part/g_part.h> 46 47 #include "g_part_if.h" 48 49 FEATURE(geom_part_mbr, "GEOM partitioning class for MBR support"); 50 51 #define MBRSIZE 512 52 53 struct g_part_mbr_table { 54 struct g_part_table base; 55 u_char mbr[MBRSIZE]; 56 }; 57 58 struct g_part_mbr_entry { 59 struct g_part_entry base; 60 struct dos_partition ent; 61 }; 62 63 static int g_part_mbr_add(struct g_part_table *, struct g_part_entry *, 64 struct g_part_parms *); 65 static int g_part_mbr_bootcode(struct g_part_table *, struct g_part_parms *); 66 static int g_part_mbr_create(struct g_part_table *, struct g_part_parms *); 67 static int g_part_mbr_destroy(struct g_part_table *, struct g_part_parms *); 68 static void g_part_mbr_dumpconf(struct g_part_table *, struct g_part_entry *, 69 struct sbuf *, const char *); 70 static int g_part_mbr_dumpto(struct g_part_table *, struct g_part_entry *); 71 static int g_part_mbr_modify(struct g_part_table *, struct g_part_entry *, 72 struct g_part_parms *); 73 static const char *g_part_mbr_name(struct g_part_table *, struct g_part_entry *, 74 char *, size_t); 75 static int g_part_mbr_probe(struct g_part_table *, struct g_consumer *); 76 static int g_part_mbr_read(struct g_part_table *, struct g_consumer *); 77 static int g_part_mbr_setunset(struct g_part_table *, struct g_part_entry *, 78 const char *, unsigned int); 79 static const char *g_part_mbr_type(struct g_part_table *, struct g_part_entry *, 80 char *, size_t); 81 static int g_part_mbr_write(struct g_part_table *, struct g_consumer *); 82 static int g_part_mbr_resize(struct g_part_table *, struct g_part_entry *, 83 struct g_part_parms *); 84 85 static kobj_method_t g_part_mbr_methods[] = { 86 KOBJMETHOD(g_part_add, g_part_mbr_add), 87 KOBJMETHOD(g_part_bootcode, g_part_mbr_bootcode), 88 KOBJMETHOD(g_part_create, g_part_mbr_create), 89 KOBJMETHOD(g_part_destroy, g_part_mbr_destroy), 90 KOBJMETHOD(g_part_dumpconf, g_part_mbr_dumpconf), 91 KOBJMETHOD(g_part_dumpto, g_part_mbr_dumpto), 92 KOBJMETHOD(g_part_modify, g_part_mbr_modify), 93 KOBJMETHOD(g_part_resize, g_part_mbr_resize), 94 KOBJMETHOD(g_part_name, g_part_mbr_name), 95 KOBJMETHOD(g_part_probe, g_part_mbr_probe), 96 KOBJMETHOD(g_part_read, g_part_mbr_read), 97 KOBJMETHOD(g_part_setunset, g_part_mbr_setunset), 98 KOBJMETHOD(g_part_type, g_part_mbr_type), 99 KOBJMETHOD(g_part_write, g_part_mbr_write), 100 { 0, 0 } 101 }; 102 103 static struct g_part_scheme g_part_mbr_scheme = { 104 "MBR", 105 g_part_mbr_methods, 106 sizeof(struct g_part_mbr_table), 107 .gps_entrysz = sizeof(struct g_part_mbr_entry), 108 .gps_minent = NDOSPART, 109 .gps_maxent = NDOSPART, 110 .gps_bootcodesz = MBRSIZE, 111 }; 112 G_PART_SCHEME_DECLARE(g_part_mbr); 113 114 static struct g_part_mbr_alias { 115 u_char typ; 116 int alias; 117 } mbr_alias_match[] = { 118 { DOSPTYP_386BSD, G_PART_ALIAS_FREEBSD }, 119 { DOSPTYP_EXT, G_PART_ALIAS_EBR }, 120 { DOSPTYP_NTFS, G_PART_ALIAS_MS_NTFS }, 121 { DOSPTYP_FAT16, G_PART_ALIAS_MS_FAT16 }, 122 { DOSPTYP_FAT32, G_PART_ALIAS_MS_FAT32 }, 123 { DOSPTYP_EXTLBA, G_PART_ALIAS_EBR }, 124 { DOSPTYP_LDM, G_PART_ALIAS_MS_LDM_DATA }, 125 { DOSPTYP_LINSWP, G_PART_ALIAS_LINUX_SWAP }, 126 { DOSPTYP_LINUX, G_PART_ALIAS_LINUX_DATA }, 127 { DOSPTYP_LINLVM, G_PART_ALIAS_LINUX_LVM }, 128 { DOSPTYP_LINRAID, G_PART_ALIAS_LINUX_RAID }, 129 { DOSPTYP_PPCBOOT, G_PART_ALIAS_FREEBSD_BOOT }, 130 { DOSPTYP_VMFS, G_PART_ALIAS_VMFS }, 131 { DOSPTYP_VMKDIAG, G_PART_ALIAS_VMKDIAG }, 132 }; 133 134 static int 135 mbr_parse_type(const char *type, u_char *dp_typ) 136 { 137 const char *alias; 138 char *endp; 139 long lt; 140 int i; 141 142 if (type[0] == '!') { 143 lt = strtol(type + 1, &endp, 0); 144 if (type[1] == '\0' || *endp != '\0' || lt <= 0 || lt >= 256) 145 return (EINVAL); 146 *dp_typ = (u_char)lt; 147 return (0); 148 } 149 for (i = 0; 150 i < sizeof(mbr_alias_match) / sizeof(mbr_alias_match[0]); i++) { 151 alias = g_part_alias_name(mbr_alias_match[i].alias); 152 if (strcasecmp(type, alias) == 0) { 153 *dp_typ = mbr_alias_match[i].typ; 154 return (0); 155 } 156 } 157 return (EINVAL); 158 } 159 160 static int 161 mbr_probe_bpb(u_char *bpb) 162 { 163 uint16_t secsz; 164 uint8_t clstsz; 165 166 #define PO2(x) ((x & (x - 1)) == 0) 167 secsz = le16dec(bpb); 168 if (secsz < 512 || secsz > 4096 || !PO2(secsz)) 169 return (0); 170 clstsz = bpb[2]; 171 if (clstsz < 1 || clstsz > 128 || !PO2(clstsz)) 172 return (0); 173 #undef PO2 174 175 return (1); 176 } 177 178 static void 179 mbr_set_chs(struct g_part_table *table, uint32_t lba, u_char *cylp, u_char *hdp, 180 u_char *secp) 181 { 182 uint32_t cyl, hd, sec; 183 184 sec = lba % table->gpt_sectors + 1; 185 lba /= table->gpt_sectors; 186 hd = lba % table->gpt_heads; 187 lba /= table->gpt_heads; 188 cyl = lba; 189 if (cyl > 1023) 190 sec = hd = cyl = ~0; 191 192 *cylp = cyl & 0xff; 193 *hdp = hd & 0xff; 194 *secp = (sec & 0x3f) | ((cyl >> 2) & 0xc0); 195 } 196 197 static int 198 g_part_mbr_add(struct g_part_table *basetable, struct g_part_entry *baseentry, 199 struct g_part_parms *gpp) 200 { 201 struct g_part_mbr_entry *entry; 202 struct g_part_mbr_table *table; 203 uint32_t start, size, sectors; 204 205 if (gpp->gpp_parms & G_PART_PARM_LABEL) 206 return (EINVAL); 207 208 sectors = basetable->gpt_sectors; 209 210 entry = (struct g_part_mbr_entry *)baseentry; 211 table = (struct g_part_mbr_table *)basetable; 212 213 start = gpp->gpp_start; 214 size = gpp->gpp_size; 215 if (size < sectors) 216 return (EINVAL); 217 if (start % sectors) { 218 size = size - sectors + (start % sectors); 219 start = start - (start % sectors) + sectors; 220 } 221 if (size % sectors) 222 size = size - (size % sectors); 223 if (size < sectors) 224 return (EINVAL); 225 226 if (baseentry->gpe_deleted) 227 bzero(&entry->ent, sizeof(entry->ent)); 228 229 KASSERT(baseentry->gpe_start <= start, ("%s", __func__)); 230 KASSERT(baseentry->gpe_end >= start + size - 1, ("%s", __func__)); 231 baseentry->gpe_start = start; 232 baseentry->gpe_end = start + size - 1; 233 entry->ent.dp_start = start; 234 entry->ent.dp_size = size; 235 mbr_set_chs(basetable, baseentry->gpe_start, &entry->ent.dp_scyl, 236 &entry->ent.dp_shd, &entry->ent.dp_ssect); 237 mbr_set_chs(basetable, baseentry->gpe_end, &entry->ent.dp_ecyl, 238 &entry->ent.dp_ehd, &entry->ent.dp_esect); 239 return (mbr_parse_type(gpp->gpp_type, &entry->ent.dp_typ)); 240 } 241 242 static int 243 g_part_mbr_bootcode(struct g_part_table *basetable, struct g_part_parms *gpp) 244 { 245 struct g_part_mbr_table *table; 246 uint32_t dsn; 247 248 if (gpp->gpp_codesize != MBRSIZE) 249 return (ENODEV); 250 251 table = (struct g_part_mbr_table *)basetable; 252 dsn = *(uint32_t *)(table->mbr + DOSDSNOFF); 253 bcopy(gpp->gpp_codeptr, table->mbr, DOSPARTOFF); 254 if (dsn != 0) 255 *(uint32_t *)(table->mbr + DOSDSNOFF) = dsn; 256 return (0); 257 } 258 259 static int 260 g_part_mbr_create(struct g_part_table *basetable, struct g_part_parms *gpp) 261 { 262 struct g_provider *pp; 263 struct g_part_mbr_table *table; 264 265 pp = gpp->gpp_provider; 266 if (pp->sectorsize < MBRSIZE) 267 return (ENOSPC); 268 269 basetable->gpt_first = basetable->gpt_sectors; 270 basetable->gpt_last = MIN(pp->mediasize / pp->sectorsize, 271 UINT32_MAX) - 1; 272 273 table = (struct g_part_mbr_table *)basetable; 274 le16enc(table->mbr + DOSMAGICOFFSET, DOSMAGIC); 275 return (0); 276 } 277 278 static int 279 g_part_mbr_destroy(struct g_part_table *basetable, struct g_part_parms *gpp) 280 { 281 282 /* Wipe the first sector to clear the partitioning. */ 283 basetable->gpt_smhead |= 1; 284 return (0); 285 } 286 287 static void 288 g_part_mbr_dumpconf(struct g_part_table *table, struct g_part_entry *baseentry, 289 struct sbuf *sb, const char *indent) 290 { 291 struct g_part_mbr_entry *entry; 292 293 entry = (struct g_part_mbr_entry *)baseentry; 294 if (indent == NULL) { 295 /* conftxt: libdisk compatibility */ 296 sbuf_printf(sb, " xs MBR xt %u", entry->ent.dp_typ); 297 } else if (entry != NULL) { 298 /* confxml: partition entry information */ 299 sbuf_printf(sb, "%s<rawtype>%u</rawtype>\n", indent, 300 entry->ent.dp_typ); 301 if (entry->ent.dp_flag & 0x80) 302 sbuf_printf(sb, "%s<attrib>active</attrib>\n", indent); 303 } else { 304 /* confxml: scheme information */ 305 } 306 } 307 308 static int 309 g_part_mbr_dumpto(struct g_part_table *table, struct g_part_entry *baseentry) 310 { 311 struct g_part_mbr_entry *entry; 312 313 /* Allow dumping to a FreeBSD partition or Linux swap partition only. */ 314 entry = (struct g_part_mbr_entry *)baseentry; 315 return ((entry->ent.dp_typ == DOSPTYP_386BSD || 316 entry->ent.dp_typ == DOSPTYP_LINSWP) ? 1 : 0); 317 } 318 319 static int 320 g_part_mbr_modify(struct g_part_table *basetable, 321 struct g_part_entry *baseentry, struct g_part_parms *gpp) 322 { 323 struct g_part_mbr_entry *entry; 324 325 if (gpp->gpp_parms & G_PART_PARM_LABEL) 326 return (EINVAL); 327 328 entry = (struct g_part_mbr_entry *)baseentry; 329 if (gpp->gpp_parms & G_PART_PARM_TYPE) 330 return (mbr_parse_type(gpp->gpp_type, &entry->ent.dp_typ)); 331 return (0); 332 } 333 334 static int 335 g_part_mbr_resize(struct g_part_table *basetable, 336 struct g_part_entry *baseentry, struct g_part_parms *gpp) 337 { 338 struct g_part_mbr_entry *entry; 339 struct g_provider *pp; 340 uint32_t size, sectors; 341 342 if (baseentry == NULL) { 343 pp = LIST_FIRST(&basetable->gpt_gp->consumer)->provider; 344 basetable->gpt_last = MIN(pp->mediasize / pp->sectorsize, 345 UINT32_MAX) - 1; 346 return (0); 347 } 348 sectors = basetable->gpt_sectors; 349 size = gpp->gpp_size; 350 351 if (size < sectors) 352 return (EINVAL); 353 if (size % sectors) 354 size = size - (size % sectors); 355 if (size < sectors) 356 return (EINVAL); 357 358 entry = (struct g_part_mbr_entry *)baseentry; 359 baseentry->gpe_end = baseentry->gpe_start + size - 1; 360 entry->ent.dp_size = size; 361 mbr_set_chs(basetable, baseentry->gpe_end, &entry->ent.dp_ecyl, 362 &entry->ent.dp_ehd, &entry->ent.dp_esect); 363 return (0); 364 } 365 366 static const char * 367 g_part_mbr_name(struct g_part_table *table, struct g_part_entry *baseentry, 368 char *buf, size_t bufsz) 369 { 370 371 snprintf(buf, bufsz, "s%d", baseentry->gpe_index); 372 return (buf); 373 } 374 375 static int 376 g_part_mbr_probe(struct g_part_table *table, struct g_consumer *cp) 377 { 378 char psn[8]; 379 struct g_provider *pp; 380 u_char *buf, *p; 381 int error, index, res, sum; 382 uint16_t magic; 383 384 pp = cp->provider; 385 386 /* Sanity-check the provider. */ 387 if (pp->sectorsize < MBRSIZE || pp->mediasize < pp->sectorsize) 388 return (ENOSPC); 389 if (pp->sectorsize > 4096) 390 return (ENXIO); 391 392 /* We don't nest under an MBR (see EBR instead). */ 393 error = g_getattr("PART::scheme", cp, &psn); 394 if (error == 0 && strcmp(psn, g_part_mbr_scheme.name) == 0) 395 return (ELOOP); 396 397 /* Check that there's a MBR. */ 398 buf = g_read_data(cp, 0L, pp->sectorsize, &error); 399 if (buf == NULL) 400 return (error); 401 402 /* We goto out on mismatch. */ 403 res = ENXIO; 404 405 magic = le16dec(buf + DOSMAGICOFFSET); 406 if (magic != DOSMAGIC) 407 goto out; 408 409 for (index = 0; index < NDOSPART; index++) { 410 p = buf + DOSPARTOFF + index * DOSPARTSIZE; 411 if (p[0] != 0 && p[0] != 0x80) 412 goto out; 413 } 414 415 /* 416 * If the partition table does not consist of all zeroes, 417 * assume we have a MBR. If it's all zeroes, we could have 418 * a boot sector. For example, a boot sector that doesn't 419 * have boot code -- common on non-i386 hardware. In that 420 * case we check if we have a possible BPB. If so, then we 421 * assume we have a boot sector instead. 422 */ 423 sum = 0; 424 for (index = 0; index < NDOSPART * DOSPARTSIZE; index++) 425 sum += buf[DOSPARTOFF + index]; 426 if (sum != 0 || !mbr_probe_bpb(buf + 0x0b)) 427 res = G_PART_PROBE_PRI_NORM; 428 429 out: 430 g_free(buf); 431 return (res); 432 } 433 434 static int 435 g_part_mbr_read(struct g_part_table *basetable, struct g_consumer *cp) 436 { 437 struct dos_partition ent; 438 struct g_provider *pp; 439 struct g_part_mbr_table *table; 440 struct g_part_mbr_entry *entry; 441 u_char *buf, *p; 442 off_t chs, msize, first; 443 u_int sectors, heads; 444 int error, index; 445 446 pp = cp->provider; 447 table = (struct g_part_mbr_table *)basetable; 448 first = basetable->gpt_sectors; 449 msize = MIN(pp->mediasize / pp->sectorsize, UINT32_MAX); 450 451 buf = g_read_data(cp, 0L, pp->sectorsize, &error); 452 if (buf == NULL) 453 return (error); 454 455 bcopy(buf, table->mbr, sizeof(table->mbr)); 456 for (index = NDOSPART - 1; index >= 0; index--) { 457 p = buf + DOSPARTOFF + index * DOSPARTSIZE; 458 ent.dp_flag = p[0]; 459 ent.dp_shd = p[1]; 460 ent.dp_ssect = p[2]; 461 ent.dp_scyl = p[3]; 462 ent.dp_typ = p[4]; 463 ent.dp_ehd = p[5]; 464 ent.dp_esect = p[6]; 465 ent.dp_ecyl = p[7]; 466 ent.dp_start = le32dec(p + 8); 467 ent.dp_size = le32dec(p + 12); 468 if (ent.dp_typ == 0 || ent.dp_typ == DOSPTYP_PMBR) 469 continue; 470 if (ent.dp_start == 0 || ent.dp_size == 0) 471 continue; 472 sectors = ent.dp_esect & 0x3f; 473 if (sectors > basetable->gpt_sectors && 474 !basetable->gpt_fixgeom) { 475 g_part_geometry_heads(msize, sectors, &chs, &heads); 476 if (chs != 0) { 477 basetable->gpt_sectors = sectors; 478 basetable->gpt_heads = heads; 479 } 480 } 481 if (ent.dp_start < first) 482 first = ent.dp_start; 483 entry = (struct g_part_mbr_entry *)g_part_new_entry(basetable, 484 index + 1, ent.dp_start, ent.dp_start + ent.dp_size - 1); 485 entry->ent = ent; 486 } 487 488 basetable->gpt_entries = NDOSPART; 489 basetable->gpt_first = basetable->gpt_sectors; 490 basetable->gpt_last = msize - 1; 491 492 if (first < basetable->gpt_first) 493 basetable->gpt_first = 1; 494 495 g_free(buf); 496 return (0); 497 } 498 499 static int 500 g_part_mbr_setunset(struct g_part_table *table, struct g_part_entry *baseentry, 501 const char *attrib, unsigned int set) 502 { 503 struct g_part_entry *iter; 504 struct g_part_mbr_entry *entry; 505 int changed; 506 507 if (baseentry == NULL) 508 return (ENODEV); 509 if (strcasecmp(attrib, "active") != 0) 510 return (EINVAL); 511 512 /* Only one entry can have the active attribute. */ 513 LIST_FOREACH(iter, &table->gpt_entry, gpe_entry) { 514 if (iter->gpe_deleted) 515 continue; 516 changed = 0; 517 entry = (struct g_part_mbr_entry *)iter; 518 if (iter == baseentry) { 519 if (set && (entry->ent.dp_flag & 0x80) == 0) { 520 entry->ent.dp_flag |= 0x80; 521 changed = 1; 522 } else if (!set && (entry->ent.dp_flag & 0x80)) { 523 entry->ent.dp_flag &= ~0x80; 524 changed = 1; 525 } 526 } else { 527 if (set && (entry->ent.dp_flag & 0x80)) { 528 entry->ent.dp_flag &= ~0x80; 529 changed = 1; 530 } 531 } 532 if (changed && !iter->gpe_created) 533 iter->gpe_modified = 1; 534 } 535 return (0); 536 } 537 538 static const char * 539 g_part_mbr_type(struct g_part_table *basetable, struct g_part_entry *baseentry, 540 char *buf, size_t bufsz) 541 { 542 struct g_part_mbr_entry *entry; 543 int i; 544 545 entry = (struct g_part_mbr_entry *)baseentry; 546 for (i = 0; 547 i < sizeof(mbr_alias_match) / sizeof(mbr_alias_match[0]); i++) { 548 if (mbr_alias_match[i].typ == entry->ent.dp_typ) 549 return (g_part_alias_name(mbr_alias_match[i].alias)); 550 } 551 snprintf(buf, bufsz, "!%d", entry->ent.dp_typ); 552 return (buf); 553 } 554 555 static int 556 g_part_mbr_write(struct g_part_table *basetable, struct g_consumer *cp) 557 { 558 struct g_part_entry *baseentry; 559 struct g_part_mbr_entry *entry; 560 struct g_part_mbr_table *table; 561 u_char *p; 562 int error, index; 563 564 table = (struct g_part_mbr_table *)basetable; 565 baseentry = LIST_FIRST(&basetable->gpt_entry); 566 for (index = 1; index <= basetable->gpt_entries; index++) { 567 p = table->mbr + DOSPARTOFF + (index - 1) * DOSPARTSIZE; 568 entry = (baseentry != NULL && index == baseentry->gpe_index) 569 ? (struct g_part_mbr_entry *)baseentry : NULL; 570 if (entry != NULL && !baseentry->gpe_deleted) { 571 p[0] = entry->ent.dp_flag; 572 p[1] = entry->ent.dp_shd; 573 p[2] = entry->ent.dp_ssect; 574 p[3] = entry->ent.dp_scyl; 575 p[4] = entry->ent.dp_typ; 576 p[5] = entry->ent.dp_ehd; 577 p[6] = entry->ent.dp_esect; 578 p[7] = entry->ent.dp_ecyl; 579 le32enc(p + 8, entry->ent.dp_start); 580 le32enc(p + 12, entry->ent.dp_size); 581 } else 582 bzero(p, DOSPARTSIZE); 583 584 if (entry != NULL) 585 baseentry = LIST_NEXT(baseentry, gpe_entry); 586 } 587 588 error = g_write_data(cp, 0, table->mbr, cp->provider->sectorsize); 589 return (error); 590 } 591