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/geom_int.h> 46#include <geom/part/g_part.h> 47 48#include "g_part_if.h" 49 50FEATURE(geom_part_mbr, "GEOM partitioning class for MBR support"); 51 52#define MBRSIZE 512 53 54struct g_part_mbr_table { 55 struct g_part_table base; 56 u_char mbr[MBRSIZE]; 57}; 58 59struct g_part_mbr_entry { 60 struct g_part_entry base; 61 struct dos_partition ent; 62}; 63 64static int g_part_mbr_add(struct g_part_table *, struct g_part_entry *, 65 struct g_part_parms *); 66static int g_part_mbr_bootcode(struct g_part_table *, struct g_part_parms *); 67static int g_part_mbr_create(struct g_part_table *, struct g_part_parms *); 68static int g_part_mbr_destroy(struct g_part_table *, struct g_part_parms *); 69static void g_part_mbr_dumpconf(struct g_part_table *, struct g_part_entry *, 70 struct sbuf *, const char *); 71static int g_part_mbr_dumpto(struct g_part_table *, struct g_part_entry *); 72static int g_part_mbr_modify(struct g_part_table *, struct g_part_entry *, 73 struct g_part_parms *); 74static const char *g_part_mbr_name(struct g_part_table *, struct g_part_entry *, 75 char *, size_t); 76static int g_part_mbr_probe(struct g_part_table *, struct g_consumer *); 77static int g_part_mbr_read(struct g_part_table *, struct g_consumer *); 78static int g_part_mbr_setunset(struct g_part_table *, struct g_part_entry *, 79 const char *, unsigned int); 80static const char *g_part_mbr_type(struct g_part_table *, struct g_part_entry *, 81 char *, size_t); 82static int g_part_mbr_write(struct g_part_table *, struct g_consumer *); 83static int g_part_mbr_resize(struct g_part_table *, struct g_part_entry *, 84 struct g_part_parms *); 85 86static kobj_method_t g_part_mbr_methods[] = { 87 KOBJMETHOD(g_part_add, g_part_mbr_add), 88 KOBJMETHOD(g_part_bootcode, g_part_mbr_bootcode), 89 KOBJMETHOD(g_part_create, g_part_mbr_create), 90 KOBJMETHOD(g_part_destroy, g_part_mbr_destroy), 91 KOBJMETHOD(g_part_dumpconf, g_part_mbr_dumpconf), 92 KOBJMETHOD(g_part_dumpto, g_part_mbr_dumpto), 93 KOBJMETHOD(g_part_modify, g_part_mbr_modify), 94 KOBJMETHOD(g_part_resize, g_part_mbr_resize), 95 KOBJMETHOD(g_part_name, g_part_mbr_name), 96 KOBJMETHOD(g_part_probe, g_part_mbr_probe), 97 KOBJMETHOD(g_part_read, g_part_mbr_read), 98 KOBJMETHOD(g_part_setunset, g_part_mbr_setunset), 99 KOBJMETHOD(g_part_type, g_part_mbr_type), 100 KOBJMETHOD(g_part_write, g_part_mbr_write), 101 { 0, 0 } 102}; 103 104static struct g_part_scheme g_part_mbr_scheme = { 105 "MBR", 106 g_part_mbr_methods, 107 sizeof(struct g_part_mbr_table), 108 .gps_entrysz = sizeof(struct g_part_mbr_entry), 109 .gps_minent = NDOSPART, 110 .gps_maxent = NDOSPART, 111 .gps_bootcodesz = MBRSIZE, 112}; 113G_PART_SCHEME_DECLARE(g_part_mbr); 114 115static struct g_part_mbr_alias { 116 u_char typ; 117 int alias; 118} mbr_alias_match[] = { 119 { DOSPTYP_386BSD, G_PART_ALIAS_FREEBSD }, 120 { DOSPTYP_EXT, G_PART_ALIAS_EBR }, 121 { DOSPTYP_NTFS, G_PART_ALIAS_MS_NTFS }, 122 { DOSPTYP_FAT16, G_PART_ALIAS_MS_FAT16 }, 123 { DOSPTYP_FAT32, G_PART_ALIAS_MS_FAT32 }, 124 { DOSPTYP_EXTLBA, G_PART_ALIAS_EBR }, 125 { DOSPTYP_LDM, G_PART_ALIAS_MS_LDM_DATA }, 126 { DOSPTYP_LINSWP, G_PART_ALIAS_LINUX_SWAP }, 127 { DOSPTYP_LINUX, G_PART_ALIAS_LINUX_DATA }, 128 { DOSPTYP_LINLVM, G_PART_ALIAS_LINUX_LVM }, 129 { DOSPTYP_LINRAID, G_PART_ALIAS_LINUX_RAID }, 130 { DOSPTYP_VMFS, G_PART_ALIAS_VMFS }, 131 { DOSPTYP_VMKDIAG, G_PART_ALIAS_VMKDIAG }, 132}; 133 134static int 135mbr_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 160static int 161mbr_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 178static void 179mbr_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 197static int 198mbr_align(struct g_part_table *basetable, uint32_t *start, uint32_t *size) 199{ 200 uint32_t sectors; 201 202 sectors = basetable->gpt_sectors; 203 if (*size < sectors) 204 return (EINVAL); 205 if (start != NULL && (*start % sectors)) { 206 *size += (*start % sectors) - sectors; 207 *start -= (*start % sectors) - sectors; 208 } 209 if (*size % sectors) 210 *size -= (*size % sectors); 211 if (*size < sectors) 212 return (EINVAL); 213 return (0); 214} 215 216static int 217g_part_mbr_add(struct g_part_table *basetable, struct g_part_entry *baseentry, 218 struct g_part_parms *gpp) 219{ 220 struct g_part_mbr_entry *entry; 221 uint32_t start, size; 222 223 if (gpp->gpp_parms & G_PART_PARM_LABEL) 224 return (EINVAL); 225 226 entry = (struct g_part_mbr_entry *)baseentry; 227 start = gpp->gpp_start; 228 size = gpp->gpp_size; 229 if (mbr_align(basetable, &start, &size) != 0) 230 return (EINVAL); 231 if (baseentry->gpe_deleted) 232 bzero(&entry->ent, sizeof(entry->ent)); 233 234 KASSERT(baseentry->gpe_start <= start, ("%s", __func__)); 235 KASSERT(baseentry->gpe_end >= start + size - 1, ("%s", __func__)); 236 baseentry->gpe_start = start; 237 baseentry->gpe_end = start + size - 1; 238 entry->ent.dp_start = start; 239 entry->ent.dp_size = size; 240 mbr_set_chs(basetable, baseentry->gpe_start, &entry->ent.dp_scyl, 241 &entry->ent.dp_shd, &entry->ent.dp_ssect); 242 mbr_set_chs(basetable, baseentry->gpe_end, &entry->ent.dp_ecyl, 243 &entry->ent.dp_ehd, &entry->ent.dp_esect); 244 return (mbr_parse_type(gpp->gpp_type, &entry->ent.dp_typ)); 245} 246 247static int 248g_part_mbr_bootcode(struct g_part_table *basetable, struct g_part_parms *gpp) 249{ 250 struct g_part_mbr_table *table; 251 uint32_t dsn; 252 253 if (gpp->gpp_codesize != MBRSIZE) 254 return (ENODEV); 255 256 table = (struct g_part_mbr_table *)basetable; 257 dsn = *(uint32_t *)(table->mbr + DOSDSNOFF); 258 bcopy(gpp->gpp_codeptr, table->mbr, DOSPARTOFF); 259 if (dsn != 0) 260 *(uint32_t *)(table->mbr + DOSDSNOFF) = dsn; 261 return (0); 262} 263 264static int 265g_part_mbr_create(struct g_part_table *basetable, struct g_part_parms *gpp) 266{ 267 struct g_provider *pp; 268 struct g_part_mbr_table *table; 269 270 pp = gpp->gpp_provider; 271 if (pp->sectorsize < MBRSIZE) 272 return (ENOSPC); 273 274 basetable->gpt_first = basetable->gpt_sectors; 275 basetable->gpt_last = MIN(pp->mediasize / pp->sectorsize, 276 UINT32_MAX) - 1; 277 278 table = (struct g_part_mbr_table *)basetable; 279 le16enc(table->mbr + DOSMAGICOFFSET, DOSMAGIC); 280 return (0); 281} 282 283static int 284g_part_mbr_destroy(struct g_part_table *basetable, struct g_part_parms *gpp) 285{ 286 287 /* Wipe the first sector to clear the partitioning. */ 288 basetable->gpt_smhead |= 1; 289 return (0); 290} 291 292static void 293g_part_mbr_dumpconf(struct g_part_table *table, struct g_part_entry *baseentry, 294 struct sbuf *sb, const char *indent) 295{ 296 struct g_part_mbr_entry *entry; 297 298 entry = (struct g_part_mbr_entry *)baseentry; 299 if (indent == NULL) { 300 /* conftxt: libdisk compatibility */ 301 sbuf_printf(sb, " xs MBR xt %u", entry->ent.dp_typ); 302 } else if (entry != NULL) { 303 /* confxml: partition entry information */ 304 sbuf_printf(sb, "%s<rawtype>%u</rawtype>\n", indent, 305 entry->ent.dp_typ); 306 if (entry->ent.dp_flag & 0x80) 307 sbuf_printf(sb, "%s<attrib>active</attrib>\n", indent); 308 } else { 309 /* confxml: scheme information */ 310 } 311} 312 313static int 314g_part_mbr_dumpto(struct g_part_table *table, struct g_part_entry *baseentry) 315{ 316 struct g_part_mbr_entry *entry; 317 318 /* Allow dumping to a FreeBSD partition or Linux swap partition only. */ 319 entry = (struct g_part_mbr_entry *)baseentry; 320 return ((entry->ent.dp_typ == DOSPTYP_386BSD || 321 entry->ent.dp_typ == DOSPTYP_LINSWP) ? 1 : 0); 322} 323 324static int 325g_part_mbr_modify(struct g_part_table *basetable, 326 struct g_part_entry *baseentry, struct g_part_parms *gpp) 327{ 328 struct g_part_mbr_entry *entry; 329 330 if (gpp->gpp_parms & G_PART_PARM_LABEL) 331 return (EINVAL); 332 333 entry = (struct g_part_mbr_entry *)baseentry; 334 if (gpp->gpp_parms & G_PART_PARM_TYPE) 335 return (mbr_parse_type(gpp->gpp_type, &entry->ent.dp_typ)); 336 return (0); 337} 338 339static int 340g_part_mbr_resize(struct g_part_table *basetable, 341 struct g_part_entry *baseentry, struct g_part_parms *gpp) 342{ 343 struct g_part_mbr_entry *entry; 344 struct g_provider *pp; 345 uint32_t size; 346 347 size = gpp->gpp_size; 348 if (mbr_align(basetable, NULL, &size) != 0) 349 return (EINVAL); 350 /* XXX: prevent unexpected shrinking. */ 351 pp = baseentry->gpe_pp; 352 if ((g_debugflags & 0x10) == 0 && size < gpp->gpp_size && 353 pp->mediasize / pp->sectorsize > size) 354 return (EBUSY); 355 entry = (struct g_part_mbr_entry *)baseentry; 356 baseentry->gpe_end = baseentry->gpe_start + size - 1; 357 entry->ent.dp_size = size; 358 mbr_set_chs(basetable, baseentry->gpe_end, &entry->ent.dp_ecyl, 359 &entry->ent.dp_ehd, &entry->ent.dp_esect); 360 return (0); 361} 362 363static const char * 364g_part_mbr_name(struct g_part_table *table, struct g_part_entry *baseentry, 365 char *buf, size_t bufsz) 366{ 367 368 snprintf(buf, bufsz, "s%d", baseentry->gpe_index); 369 return (buf); 370} 371 372static int 373g_part_mbr_probe(struct g_part_table *table, struct g_consumer *cp) 374{ 375 char psn[8]; 376 struct g_provider *pp; 377 u_char *buf, *p; 378 int error, index, res, sum; 379 uint16_t magic; 380 381 pp = cp->provider; 382 383 /* Sanity-check the provider. */ 384 if (pp->sectorsize < MBRSIZE || pp->mediasize < pp->sectorsize) 385 return (ENOSPC); 386 if (pp->sectorsize > 4096) 387 return (ENXIO); 388 389 /* We don't nest under an MBR (see EBR instead). */ 390 error = g_getattr("PART::scheme", cp, &psn); 391 if (error == 0 && strcmp(psn, g_part_mbr_scheme.name) == 0) 392 return (ELOOP); 393 394 /* Check that there's a MBR. */ 395 buf = g_read_data(cp, 0L, pp->sectorsize, &error); 396 if (buf == NULL) 397 return (error); 398 399 /* We goto out on mismatch. */ 400 res = ENXIO; 401 402 magic = le16dec(buf + DOSMAGICOFFSET); 403 if (magic != DOSMAGIC) 404 goto out; 405 406 for (index = 0; index < NDOSPART; index++) { 407 p = buf + DOSPARTOFF + index * DOSPARTSIZE; 408 if (p[0] != 0 && p[0] != 0x80) 409 goto out; 410 } 411 412 /* 413 * If the partition table does not consist of all zeroes, 414 * assume we have a MBR. If it's all zeroes, we could have 415 * a boot sector. For example, a boot sector that doesn't 416 * have boot code -- common on non-i386 hardware. In that 417 * case we check if we have a possible BPB. If so, then we 418 * assume we have a boot sector instead. 419 */ 420 sum = 0; 421 for (index = 0; index < NDOSPART * DOSPARTSIZE; index++) 422 sum += buf[DOSPARTOFF + index]; 423 if (sum != 0 || !mbr_probe_bpb(buf + 0x0b)) 424 res = G_PART_PROBE_PRI_NORM; 425 426 out: 427 g_free(buf); 428 return (res); 429} 430 431static int 432g_part_mbr_read(struct g_part_table *basetable, struct g_consumer *cp) 433{ 434 struct dos_partition ent; 435 struct g_provider *pp; 436 struct g_part_mbr_table *table; 437 struct g_part_mbr_entry *entry; 438 u_char *buf, *p; 439 off_t chs, msize, first; 440 u_int sectors, heads; 441 int error, index; 442 443 pp = cp->provider; 444 table = (struct g_part_mbr_table *)basetable; 445 first = basetable->gpt_sectors; 446 msize = MIN(pp->mediasize / pp->sectorsize, UINT32_MAX); 447 448 buf = g_read_data(cp, 0L, pp->sectorsize, &error); 449 if (buf == NULL) 450 return (error); 451 452 bcopy(buf, table->mbr, sizeof(table->mbr)); 453 for (index = NDOSPART - 1; index >= 0; index--) { 454 p = buf + DOSPARTOFF + index * DOSPARTSIZE; 455 ent.dp_flag = p[0]; 456 ent.dp_shd = p[1]; 457 ent.dp_ssect = p[2]; 458 ent.dp_scyl = p[3]; 459 ent.dp_typ = p[4]; 460 ent.dp_ehd = p[5]; 461 ent.dp_esect = p[6]; 462 ent.dp_ecyl = p[7]; 463 ent.dp_start = le32dec(p + 8); 464 ent.dp_size = le32dec(p + 12); 465 if (ent.dp_typ == 0 || ent.dp_typ == DOSPTYP_PMBR) 466 continue; 467 if (ent.dp_start == 0 || ent.dp_size == 0) 468 continue; 469 sectors = ent.dp_esect & 0x3f; 470 if (sectors > basetable->gpt_sectors && 471 !basetable->gpt_fixgeom) { 472 g_part_geometry_heads(msize, sectors, &chs, &heads); 473 if (chs != 0) { 474 basetable->gpt_sectors = sectors; 475 basetable->gpt_heads = heads; 476 } 477 } 478 if (ent.dp_start < first) 479 first = ent.dp_start; 480 entry = (struct g_part_mbr_entry *)g_part_new_entry(basetable, 481 index + 1, ent.dp_start, ent.dp_start + ent.dp_size - 1); 482 entry->ent = ent; 483 } 484 485 basetable->gpt_entries = NDOSPART; 486 basetable->gpt_first = basetable->gpt_sectors; 487 basetable->gpt_last = msize - 1; 488 489 if (first < basetable->gpt_first) 490 basetable->gpt_first = 1; 491 492 g_free(buf); 493 return (0); 494} 495 496static int 497g_part_mbr_setunset(struct g_part_table *table, struct g_part_entry *baseentry, 498 const char *attrib, unsigned int set) 499{ 500 struct g_part_entry *iter; 501 struct g_part_mbr_entry *entry; 502 int changed; 503 504 if (baseentry == NULL) 505 return (ENODEV); 506 if (strcasecmp(attrib, "active") != 0) 507 return (EINVAL); 508 509 /* Only one entry can have the active attribute. */ 510 LIST_FOREACH(iter, &table->gpt_entry, gpe_entry) { 511 if (iter->gpe_deleted) 512 continue; 513 changed = 0; 514 entry = (struct g_part_mbr_entry *)iter; 515 if (iter == baseentry) { 516 if (set && (entry->ent.dp_flag & 0x80) == 0) { 517 entry->ent.dp_flag |= 0x80; 518 changed = 1; 519 } else if (!set && (entry->ent.dp_flag & 0x80)) { 520 entry->ent.dp_flag &= ~0x80; 521 changed = 1; 522 } 523 } else { 524 if (set && (entry->ent.dp_flag & 0x80)) { 525 entry->ent.dp_flag &= ~0x80; 526 changed = 1; 527 } 528 } 529 if (changed && !iter->gpe_created) 530 iter->gpe_modified = 1; 531 } 532 return (0); 533} 534 535static const char * 536g_part_mbr_type(struct g_part_table *basetable, struct g_part_entry *baseentry, 537 char *buf, size_t bufsz) 538{ 539 struct g_part_mbr_entry *entry; 540 int i; 541 542 entry = (struct g_part_mbr_entry *)baseentry; 543 for (i = 0; 544 i < sizeof(mbr_alias_match) / sizeof(mbr_alias_match[0]); i++) { 545 if (mbr_alias_match[i].typ == entry->ent.dp_typ) 546 return (g_part_alias_name(mbr_alias_match[i].alias)); 547 } 548 snprintf(buf, bufsz, "!%d", entry->ent.dp_typ); 549 return (buf); 550} 551 552static int 553g_part_mbr_write(struct g_part_table *basetable, struct g_consumer *cp) 554{ 555 struct g_part_entry *baseentry; 556 struct g_part_mbr_entry *entry; 557 struct g_part_mbr_table *table; 558 u_char *p; 559 int error, index; 560 561 table = (struct g_part_mbr_table *)basetable; 562 baseentry = LIST_FIRST(&basetable->gpt_entry); 563 for (index = 1; index <= basetable->gpt_entries; index++) { 564 p = table->mbr + DOSPARTOFF + (index - 1) * DOSPARTSIZE; 565 entry = (baseentry != NULL && index == baseentry->gpe_index) 566 ? (struct g_part_mbr_entry *)baseentry : NULL; 567 if (entry != NULL && !baseentry->gpe_deleted) { 568 p[0] = entry->ent.dp_flag; 569 p[1] = entry->ent.dp_shd; 570 p[2] = entry->ent.dp_ssect; 571 p[3] = entry->ent.dp_scyl; 572 p[4] = entry->ent.dp_typ; 573 p[5] = entry->ent.dp_ehd; 574 p[6] = entry->ent.dp_esect; 575 p[7] = entry->ent.dp_ecyl; 576 le32enc(p + 8, entry->ent.dp_start); 577 le32enc(p + 12, entry->ent.dp_size); 578 } else 579 bzero(p, DOSPARTSIZE); 580 581 if (entry != NULL) 582 baseentry = LIST_NEXT(baseentry, gpe_entry); 583 } 584 585 error = g_write_data(cp, 0, table->mbr, cp->provider->sectorsize); 586 return (error); 587} 588