1 /*- 2 * Copyright (c) 2004-2006 Pawel Jakub Dawidek <pjd@FreeBSD.org> 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 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 */ 26 27 #include <sys/cdefs.h> 28 __FBSDID("$FreeBSD$"); 29 30 #include <sys/param.h> 31 #include <sys/systm.h> 32 #include <sys/kernel.h> 33 #include <sys/module.h> 34 #include <sys/limits.h> 35 #include <sys/lock.h> 36 #include <sys/mutex.h> 37 #include <sys/bio.h> 38 #include <sys/sbuf.h> 39 #include <sys/sysctl.h> 40 #include <sys/malloc.h> 41 #include <sys/eventhandler.h> 42 #include <vm/uma.h> 43 #include <geom/geom.h> 44 #include <sys/proc.h> 45 #include <sys/kthread.h> 46 #include <sys/sched.h> 47 #include <geom/raid3/g_raid3.h> 48 49 FEATURE(geom_raid3, "GEOM RAID-3 functionality"); 50 51 static MALLOC_DEFINE(M_RAID3, "raid3_data", "GEOM_RAID3 Data"); 52 53 SYSCTL_DECL(_kern_geom); 54 static SYSCTL_NODE(_kern_geom, OID_AUTO, raid3, CTLFLAG_RW, 0, 55 "GEOM_RAID3 stuff"); 56 u_int g_raid3_debug = 0; 57 TUNABLE_INT("kern.geom.raid3.debug", &g_raid3_debug); 58 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, debug, CTLFLAG_RW, &g_raid3_debug, 0, 59 "Debug level"); 60 static u_int g_raid3_timeout = 4; 61 TUNABLE_INT("kern.geom.raid3.timeout", &g_raid3_timeout); 62 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, timeout, CTLFLAG_RW, &g_raid3_timeout, 63 0, "Time to wait on all raid3 components"); 64 static u_int g_raid3_idletime = 5; 65 TUNABLE_INT("kern.geom.raid3.idletime", &g_raid3_idletime); 66 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, idletime, CTLFLAG_RW, 67 &g_raid3_idletime, 0, "Mark components as clean when idling"); 68 static u_int g_raid3_disconnect_on_failure = 1; 69 TUNABLE_INT("kern.geom.raid3.disconnect_on_failure", 70 &g_raid3_disconnect_on_failure); 71 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, disconnect_on_failure, CTLFLAG_RW, 72 &g_raid3_disconnect_on_failure, 0, "Disconnect component on I/O failure."); 73 static u_int g_raid3_syncreqs = 2; 74 TUNABLE_INT("kern.geom.raid3.sync_requests", &g_raid3_syncreqs); 75 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, sync_requests, CTLFLAG_RDTUN, 76 &g_raid3_syncreqs, 0, "Parallel synchronization I/O requests."); 77 static u_int g_raid3_use_malloc = 0; 78 TUNABLE_INT("kern.geom.raid3.use_malloc", &g_raid3_use_malloc); 79 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, use_malloc, CTLFLAG_RDTUN, 80 &g_raid3_use_malloc, 0, "Use malloc(9) instead of uma(9)."); 81 82 static u_int g_raid3_n64k = 50; 83 TUNABLE_INT("kern.geom.raid3.n64k", &g_raid3_n64k); 84 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, n64k, CTLFLAG_RD, &g_raid3_n64k, 0, 85 "Maximum number of 64kB allocations"); 86 static u_int g_raid3_n16k = 200; 87 TUNABLE_INT("kern.geom.raid3.n16k", &g_raid3_n16k); 88 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, n16k, CTLFLAG_RD, &g_raid3_n16k, 0, 89 "Maximum number of 16kB allocations"); 90 static u_int g_raid3_n4k = 1200; 91 TUNABLE_INT("kern.geom.raid3.n4k", &g_raid3_n4k); 92 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, n4k, CTLFLAG_RD, &g_raid3_n4k, 0, 93 "Maximum number of 4kB allocations"); 94 95 static SYSCTL_NODE(_kern_geom_raid3, OID_AUTO, stat, CTLFLAG_RW, 0, 96 "GEOM_RAID3 statistics"); 97 static u_int g_raid3_parity_mismatch = 0; 98 SYSCTL_UINT(_kern_geom_raid3_stat, OID_AUTO, parity_mismatch, CTLFLAG_RD, 99 &g_raid3_parity_mismatch, 0, "Number of failures in VERIFY mode"); 100 101 #define MSLEEP(ident, mtx, priority, wmesg, timeout) do { \ 102 G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, (ident)); \ 103 msleep((ident), (mtx), (priority), (wmesg), (timeout)); \ 104 G_RAID3_DEBUG(4, "%s: Woken up %p.", __func__, (ident)); \ 105 } while (0) 106 107 static eventhandler_tag g_raid3_pre_sync = NULL; 108 109 static int g_raid3_destroy_geom(struct gctl_req *req, struct g_class *mp, 110 struct g_geom *gp); 111 static g_taste_t g_raid3_taste; 112 static void g_raid3_init(struct g_class *mp); 113 static void g_raid3_fini(struct g_class *mp); 114 115 struct g_class g_raid3_class = { 116 .name = G_RAID3_CLASS_NAME, 117 .version = G_VERSION, 118 .ctlreq = g_raid3_config, 119 .taste = g_raid3_taste, 120 .destroy_geom = g_raid3_destroy_geom, 121 .init = g_raid3_init, 122 .fini = g_raid3_fini 123 }; 124 125 126 static void g_raid3_destroy_provider(struct g_raid3_softc *sc); 127 static int g_raid3_update_disk(struct g_raid3_disk *disk, u_int state); 128 static void g_raid3_update_device(struct g_raid3_softc *sc, boolean_t force); 129 static void g_raid3_dumpconf(struct sbuf *sb, const char *indent, 130 struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp); 131 static void g_raid3_sync_stop(struct g_raid3_softc *sc, int type); 132 static int g_raid3_register_request(struct bio *pbp); 133 static void g_raid3_sync_release(struct g_raid3_softc *sc); 134 135 136 static const char * 137 g_raid3_disk_state2str(int state) 138 { 139 140 switch (state) { 141 case G_RAID3_DISK_STATE_NODISK: 142 return ("NODISK"); 143 case G_RAID3_DISK_STATE_NONE: 144 return ("NONE"); 145 case G_RAID3_DISK_STATE_NEW: 146 return ("NEW"); 147 case G_RAID3_DISK_STATE_ACTIVE: 148 return ("ACTIVE"); 149 case G_RAID3_DISK_STATE_STALE: 150 return ("STALE"); 151 case G_RAID3_DISK_STATE_SYNCHRONIZING: 152 return ("SYNCHRONIZING"); 153 case G_RAID3_DISK_STATE_DISCONNECTED: 154 return ("DISCONNECTED"); 155 default: 156 return ("INVALID"); 157 } 158 } 159 160 static const char * 161 g_raid3_device_state2str(int state) 162 { 163 164 switch (state) { 165 case G_RAID3_DEVICE_STATE_STARTING: 166 return ("STARTING"); 167 case G_RAID3_DEVICE_STATE_DEGRADED: 168 return ("DEGRADED"); 169 case G_RAID3_DEVICE_STATE_COMPLETE: 170 return ("COMPLETE"); 171 default: 172 return ("INVALID"); 173 } 174 } 175 176 const char * 177 g_raid3_get_diskname(struct g_raid3_disk *disk) 178 { 179 180 if (disk->d_consumer == NULL || disk->d_consumer->provider == NULL) 181 return ("[unknown]"); 182 return (disk->d_name); 183 } 184 185 static void * 186 g_raid3_alloc(struct g_raid3_softc *sc, size_t size, int flags) 187 { 188 void *ptr; 189 enum g_raid3_zones zone; 190 191 if (g_raid3_use_malloc || 192 (zone = g_raid3_zone(size)) == G_RAID3_NUM_ZONES) 193 ptr = malloc(size, M_RAID3, flags); 194 else { 195 ptr = uma_zalloc_arg(sc->sc_zones[zone].sz_zone, 196 &sc->sc_zones[zone], flags); 197 sc->sc_zones[zone].sz_requested++; 198 if (ptr == NULL) 199 sc->sc_zones[zone].sz_failed++; 200 } 201 return (ptr); 202 } 203 204 static void 205 g_raid3_free(struct g_raid3_softc *sc, void *ptr, size_t size) 206 { 207 enum g_raid3_zones zone; 208 209 if (g_raid3_use_malloc || 210 (zone = g_raid3_zone(size)) == G_RAID3_NUM_ZONES) 211 free(ptr, M_RAID3); 212 else { 213 uma_zfree_arg(sc->sc_zones[zone].sz_zone, 214 ptr, &sc->sc_zones[zone]); 215 } 216 } 217 218 static int 219 g_raid3_uma_ctor(void *mem, int size, void *arg, int flags) 220 { 221 struct g_raid3_zone *sz = arg; 222 223 if (sz->sz_max > 0 && sz->sz_inuse == sz->sz_max) 224 return (ENOMEM); 225 sz->sz_inuse++; 226 return (0); 227 } 228 229 static void 230 g_raid3_uma_dtor(void *mem, int size, void *arg) 231 { 232 struct g_raid3_zone *sz = arg; 233 234 sz->sz_inuse--; 235 } 236 237 #define g_raid3_xor(src, dst, size) \ 238 _g_raid3_xor((uint64_t *)(src), \ 239 (uint64_t *)(dst), (size_t)size) 240 static void 241 _g_raid3_xor(uint64_t *src, uint64_t *dst, size_t size) 242 { 243 244 KASSERT((size % 128) == 0, ("Invalid size: %zu.", size)); 245 for (; size > 0; size -= 128) { 246 *dst++ ^= (*src++); 247 *dst++ ^= (*src++); 248 *dst++ ^= (*src++); 249 *dst++ ^= (*src++); 250 *dst++ ^= (*src++); 251 *dst++ ^= (*src++); 252 *dst++ ^= (*src++); 253 *dst++ ^= (*src++); 254 *dst++ ^= (*src++); 255 *dst++ ^= (*src++); 256 *dst++ ^= (*src++); 257 *dst++ ^= (*src++); 258 *dst++ ^= (*src++); 259 *dst++ ^= (*src++); 260 *dst++ ^= (*src++); 261 *dst++ ^= (*src++); 262 } 263 } 264 265 static int 266 g_raid3_is_zero(struct bio *bp) 267 { 268 static const uint64_t zeros[] = { 269 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 270 }; 271 u_char *addr; 272 ssize_t size; 273 274 size = bp->bio_length; 275 addr = (u_char *)bp->bio_data; 276 for (; size > 0; size -= sizeof(zeros), addr += sizeof(zeros)) { 277 if (bcmp(addr, zeros, sizeof(zeros)) != 0) 278 return (0); 279 } 280 return (1); 281 } 282 283 /* 284 * --- Events handling functions --- 285 * Events in geom_raid3 are used to maintain disks and device status 286 * from one thread to simplify locking. 287 */ 288 static void 289 g_raid3_event_free(struct g_raid3_event *ep) 290 { 291 292 free(ep, M_RAID3); 293 } 294 295 int 296 g_raid3_event_send(void *arg, int state, int flags) 297 { 298 struct g_raid3_softc *sc; 299 struct g_raid3_disk *disk; 300 struct g_raid3_event *ep; 301 int error; 302 303 ep = malloc(sizeof(*ep), M_RAID3, M_WAITOK); 304 G_RAID3_DEBUG(4, "%s: Sending event %p.", __func__, ep); 305 if ((flags & G_RAID3_EVENT_DEVICE) != 0) { 306 disk = NULL; 307 sc = arg; 308 } else { 309 disk = arg; 310 sc = disk->d_softc; 311 } 312 ep->e_disk = disk; 313 ep->e_state = state; 314 ep->e_flags = flags; 315 ep->e_error = 0; 316 mtx_lock(&sc->sc_events_mtx); 317 TAILQ_INSERT_TAIL(&sc->sc_events, ep, e_next); 318 mtx_unlock(&sc->sc_events_mtx); 319 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc); 320 mtx_lock(&sc->sc_queue_mtx); 321 wakeup(sc); 322 wakeup(&sc->sc_queue); 323 mtx_unlock(&sc->sc_queue_mtx); 324 if ((flags & G_RAID3_EVENT_DONTWAIT) != 0) 325 return (0); 326 sx_assert(&sc->sc_lock, SX_XLOCKED); 327 G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, ep); 328 sx_xunlock(&sc->sc_lock); 329 while ((ep->e_flags & G_RAID3_EVENT_DONE) == 0) { 330 mtx_lock(&sc->sc_events_mtx); 331 MSLEEP(ep, &sc->sc_events_mtx, PRIBIO | PDROP, "r3:event", 332 hz * 5); 333 } 334 error = ep->e_error; 335 g_raid3_event_free(ep); 336 sx_xlock(&sc->sc_lock); 337 return (error); 338 } 339 340 static struct g_raid3_event * 341 g_raid3_event_get(struct g_raid3_softc *sc) 342 { 343 struct g_raid3_event *ep; 344 345 mtx_lock(&sc->sc_events_mtx); 346 ep = TAILQ_FIRST(&sc->sc_events); 347 mtx_unlock(&sc->sc_events_mtx); 348 return (ep); 349 } 350 351 static void 352 g_raid3_event_remove(struct g_raid3_softc *sc, struct g_raid3_event *ep) 353 { 354 355 mtx_lock(&sc->sc_events_mtx); 356 TAILQ_REMOVE(&sc->sc_events, ep, e_next); 357 mtx_unlock(&sc->sc_events_mtx); 358 } 359 360 static void 361 g_raid3_event_cancel(struct g_raid3_disk *disk) 362 { 363 struct g_raid3_softc *sc; 364 struct g_raid3_event *ep, *tmpep; 365 366 sc = disk->d_softc; 367 sx_assert(&sc->sc_lock, SX_XLOCKED); 368 369 mtx_lock(&sc->sc_events_mtx); 370 TAILQ_FOREACH_SAFE(ep, &sc->sc_events, e_next, tmpep) { 371 if ((ep->e_flags & G_RAID3_EVENT_DEVICE) != 0) 372 continue; 373 if (ep->e_disk != disk) 374 continue; 375 TAILQ_REMOVE(&sc->sc_events, ep, e_next); 376 if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) 377 g_raid3_event_free(ep); 378 else { 379 ep->e_error = ECANCELED; 380 wakeup(ep); 381 } 382 } 383 mtx_unlock(&sc->sc_events_mtx); 384 } 385 386 /* 387 * Return the number of disks in the given state. 388 * If state is equal to -1, count all connected disks. 389 */ 390 u_int 391 g_raid3_ndisks(struct g_raid3_softc *sc, int state) 392 { 393 struct g_raid3_disk *disk; 394 u_int n, ndisks; 395 396 sx_assert(&sc->sc_lock, SX_LOCKED); 397 398 for (n = ndisks = 0; n < sc->sc_ndisks; n++) { 399 disk = &sc->sc_disks[n]; 400 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 401 continue; 402 if (state == -1 || disk->d_state == state) 403 ndisks++; 404 } 405 return (ndisks); 406 } 407 408 static u_int 409 g_raid3_nrequests(struct g_raid3_softc *sc, struct g_consumer *cp) 410 { 411 struct bio *bp; 412 u_int nreqs = 0; 413 414 mtx_lock(&sc->sc_queue_mtx); 415 TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { 416 if (bp->bio_from == cp) 417 nreqs++; 418 } 419 mtx_unlock(&sc->sc_queue_mtx); 420 return (nreqs); 421 } 422 423 static int 424 g_raid3_is_busy(struct g_raid3_softc *sc, struct g_consumer *cp) 425 { 426 427 if (cp->index > 0) { 428 G_RAID3_DEBUG(2, 429 "I/O requests for %s exist, can't destroy it now.", 430 cp->provider->name); 431 return (1); 432 } 433 if (g_raid3_nrequests(sc, cp) > 0) { 434 G_RAID3_DEBUG(2, 435 "I/O requests for %s in queue, can't destroy it now.", 436 cp->provider->name); 437 return (1); 438 } 439 return (0); 440 } 441 442 static void 443 g_raid3_destroy_consumer(void *arg, int flags __unused) 444 { 445 struct g_consumer *cp; 446 447 g_topology_assert(); 448 449 cp = arg; 450 G_RAID3_DEBUG(1, "Consumer %s destroyed.", cp->provider->name); 451 g_detach(cp); 452 g_destroy_consumer(cp); 453 } 454 455 static void 456 g_raid3_kill_consumer(struct g_raid3_softc *sc, struct g_consumer *cp) 457 { 458 struct g_provider *pp; 459 int retaste_wait; 460 461 g_topology_assert(); 462 463 cp->private = NULL; 464 if (g_raid3_is_busy(sc, cp)) 465 return; 466 G_RAID3_DEBUG(2, "Consumer %s destroyed.", cp->provider->name); 467 pp = cp->provider; 468 retaste_wait = 0; 469 if (cp->acw == 1) { 470 if ((pp->geom->flags & G_GEOM_WITHER) == 0) 471 retaste_wait = 1; 472 } 473 G_RAID3_DEBUG(2, "Access %s r%dw%de%d = %d", pp->name, -cp->acr, 474 -cp->acw, -cp->ace, 0); 475 if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) 476 g_access(cp, -cp->acr, -cp->acw, -cp->ace); 477 if (retaste_wait) { 478 /* 479 * After retaste event was send (inside g_access()), we can send 480 * event to detach and destroy consumer. 481 * A class, which has consumer to the given provider connected 482 * will not receive retaste event for the provider. 483 * This is the way how I ignore retaste events when I close 484 * consumers opened for write: I detach and destroy consumer 485 * after retaste event is sent. 486 */ 487 g_post_event(g_raid3_destroy_consumer, cp, M_WAITOK, NULL); 488 return; 489 } 490 G_RAID3_DEBUG(1, "Consumer %s destroyed.", pp->name); 491 g_detach(cp); 492 g_destroy_consumer(cp); 493 } 494 495 static int 496 g_raid3_connect_disk(struct g_raid3_disk *disk, struct g_provider *pp) 497 { 498 struct g_consumer *cp; 499 int error; 500 501 g_topology_assert_not(); 502 KASSERT(disk->d_consumer == NULL, 503 ("Disk already connected (device %s).", disk->d_softc->sc_name)); 504 505 g_topology_lock(); 506 cp = g_new_consumer(disk->d_softc->sc_geom); 507 error = g_attach(cp, pp); 508 if (error != 0) { 509 g_destroy_consumer(cp); 510 g_topology_unlock(); 511 return (error); 512 } 513 error = g_access(cp, 1, 1, 1); 514 g_topology_unlock(); 515 if (error != 0) { 516 g_detach(cp); 517 g_destroy_consumer(cp); 518 G_RAID3_DEBUG(0, "Cannot open consumer %s (error=%d).", 519 pp->name, error); 520 return (error); 521 } 522 disk->d_consumer = cp; 523 disk->d_consumer->private = disk; 524 disk->d_consumer->index = 0; 525 G_RAID3_DEBUG(2, "Disk %s connected.", g_raid3_get_diskname(disk)); 526 return (0); 527 } 528 529 static void 530 g_raid3_disconnect_consumer(struct g_raid3_softc *sc, struct g_consumer *cp) 531 { 532 533 g_topology_assert(); 534 535 if (cp == NULL) 536 return; 537 if (cp->provider != NULL) 538 g_raid3_kill_consumer(sc, cp); 539 else 540 g_destroy_consumer(cp); 541 } 542 543 /* 544 * Initialize disk. This means allocate memory, create consumer, attach it 545 * to the provider and open access (r1w1e1) to it. 546 */ 547 static struct g_raid3_disk * 548 g_raid3_init_disk(struct g_raid3_softc *sc, struct g_provider *pp, 549 struct g_raid3_metadata *md, int *errorp) 550 { 551 struct g_raid3_disk *disk; 552 int error; 553 554 disk = &sc->sc_disks[md->md_no]; 555 error = g_raid3_connect_disk(disk, pp); 556 if (error != 0) { 557 if (errorp != NULL) 558 *errorp = error; 559 return (NULL); 560 } 561 disk->d_state = G_RAID3_DISK_STATE_NONE; 562 disk->d_flags = md->md_dflags; 563 if (md->md_provider[0] != '\0') 564 disk->d_flags |= G_RAID3_DISK_FLAG_HARDCODED; 565 disk->d_sync.ds_consumer = NULL; 566 disk->d_sync.ds_offset = md->md_sync_offset; 567 disk->d_sync.ds_offset_done = md->md_sync_offset; 568 disk->d_genid = md->md_genid; 569 disk->d_sync.ds_syncid = md->md_syncid; 570 if (errorp != NULL) 571 *errorp = 0; 572 return (disk); 573 } 574 575 static void 576 g_raid3_destroy_disk(struct g_raid3_disk *disk) 577 { 578 struct g_raid3_softc *sc; 579 580 g_topology_assert_not(); 581 sc = disk->d_softc; 582 sx_assert(&sc->sc_lock, SX_XLOCKED); 583 584 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 585 return; 586 g_raid3_event_cancel(disk); 587 switch (disk->d_state) { 588 case G_RAID3_DISK_STATE_SYNCHRONIZING: 589 if (sc->sc_syncdisk != NULL) 590 g_raid3_sync_stop(sc, 1); 591 /* FALLTHROUGH */ 592 case G_RAID3_DISK_STATE_NEW: 593 case G_RAID3_DISK_STATE_STALE: 594 case G_RAID3_DISK_STATE_ACTIVE: 595 g_topology_lock(); 596 g_raid3_disconnect_consumer(sc, disk->d_consumer); 597 g_topology_unlock(); 598 disk->d_consumer = NULL; 599 break; 600 default: 601 KASSERT(0 == 1, ("Wrong disk state (%s, %s).", 602 g_raid3_get_diskname(disk), 603 g_raid3_disk_state2str(disk->d_state))); 604 } 605 disk->d_state = G_RAID3_DISK_STATE_NODISK; 606 } 607 608 static void 609 g_raid3_destroy_device(struct g_raid3_softc *sc) 610 { 611 struct g_raid3_event *ep; 612 struct g_raid3_disk *disk; 613 struct g_geom *gp; 614 struct g_consumer *cp; 615 u_int n; 616 617 g_topology_assert_not(); 618 sx_assert(&sc->sc_lock, SX_XLOCKED); 619 620 gp = sc->sc_geom; 621 if (sc->sc_provider != NULL) 622 g_raid3_destroy_provider(sc); 623 for (n = 0; n < sc->sc_ndisks; n++) { 624 disk = &sc->sc_disks[n]; 625 if (disk->d_state != G_RAID3_DISK_STATE_NODISK) { 626 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 627 g_raid3_update_metadata(disk); 628 g_raid3_destroy_disk(disk); 629 } 630 } 631 while ((ep = g_raid3_event_get(sc)) != NULL) { 632 g_raid3_event_remove(sc, ep); 633 if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) 634 g_raid3_event_free(ep); 635 else { 636 ep->e_error = ECANCELED; 637 ep->e_flags |= G_RAID3_EVENT_DONE; 638 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, ep); 639 mtx_lock(&sc->sc_events_mtx); 640 wakeup(ep); 641 mtx_unlock(&sc->sc_events_mtx); 642 } 643 } 644 callout_drain(&sc->sc_callout); 645 cp = LIST_FIRST(&sc->sc_sync.ds_geom->consumer); 646 g_topology_lock(); 647 if (cp != NULL) 648 g_raid3_disconnect_consumer(sc, cp); 649 g_wither_geom(sc->sc_sync.ds_geom, ENXIO); 650 G_RAID3_DEBUG(0, "Device %s destroyed.", gp->name); 651 g_wither_geom(gp, ENXIO); 652 g_topology_unlock(); 653 if (!g_raid3_use_malloc) { 654 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_64K].sz_zone); 655 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_16K].sz_zone); 656 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_4K].sz_zone); 657 } 658 mtx_destroy(&sc->sc_queue_mtx); 659 mtx_destroy(&sc->sc_events_mtx); 660 sx_xunlock(&sc->sc_lock); 661 sx_destroy(&sc->sc_lock); 662 } 663 664 static void 665 g_raid3_orphan(struct g_consumer *cp) 666 { 667 struct g_raid3_disk *disk; 668 669 g_topology_assert(); 670 671 disk = cp->private; 672 if (disk == NULL) 673 return; 674 disk->d_softc->sc_bump_id = G_RAID3_BUMP_SYNCID; 675 g_raid3_event_send(disk, G_RAID3_DISK_STATE_DISCONNECTED, 676 G_RAID3_EVENT_DONTWAIT); 677 } 678 679 static int 680 g_raid3_write_metadata(struct g_raid3_disk *disk, struct g_raid3_metadata *md) 681 { 682 struct g_raid3_softc *sc; 683 struct g_consumer *cp; 684 off_t offset, length; 685 u_char *sector; 686 int error = 0; 687 688 g_topology_assert_not(); 689 sc = disk->d_softc; 690 sx_assert(&sc->sc_lock, SX_LOCKED); 691 692 cp = disk->d_consumer; 693 KASSERT(cp != NULL, ("NULL consumer (%s).", sc->sc_name)); 694 KASSERT(cp->provider != NULL, ("NULL provider (%s).", sc->sc_name)); 695 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 696 ("Consumer %s closed? (r%dw%de%d).", cp->provider->name, cp->acr, 697 cp->acw, cp->ace)); 698 length = cp->provider->sectorsize; 699 offset = cp->provider->mediasize - length; 700 sector = malloc((size_t)length, M_RAID3, M_WAITOK | M_ZERO); 701 if (md != NULL) 702 raid3_metadata_encode(md, sector); 703 error = g_write_data(cp, offset, sector, length); 704 free(sector, M_RAID3); 705 if (error != 0) { 706 if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) { 707 G_RAID3_DEBUG(0, "Cannot write metadata on %s " 708 "(device=%s, error=%d).", 709 g_raid3_get_diskname(disk), sc->sc_name, error); 710 disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN; 711 } else { 712 G_RAID3_DEBUG(1, "Cannot write metadata on %s " 713 "(device=%s, error=%d).", 714 g_raid3_get_diskname(disk), sc->sc_name, error); 715 } 716 if (g_raid3_disconnect_on_failure && 717 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 718 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 719 g_raid3_event_send(disk, 720 G_RAID3_DISK_STATE_DISCONNECTED, 721 G_RAID3_EVENT_DONTWAIT); 722 } 723 } 724 return (error); 725 } 726 727 int 728 g_raid3_clear_metadata(struct g_raid3_disk *disk) 729 { 730 int error; 731 732 g_topology_assert_not(); 733 sx_assert(&disk->d_softc->sc_lock, SX_LOCKED); 734 735 error = g_raid3_write_metadata(disk, NULL); 736 if (error == 0) { 737 G_RAID3_DEBUG(2, "Metadata on %s cleared.", 738 g_raid3_get_diskname(disk)); 739 } else { 740 G_RAID3_DEBUG(0, 741 "Cannot clear metadata on disk %s (error=%d).", 742 g_raid3_get_diskname(disk), error); 743 } 744 return (error); 745 } 746 747 void 748 g_raid3_fill_metadata(struct g_raid3_disk *disk, struct g_raid3_metadata *md) 749 { 750 struct g_raid3_softc *sc; 751 struct g_provider *pp; 752 753 sc = disk->d_softc; 754 strlcpy(md->md_magic, G_RAID3_MAGIC, sizeof(md->md_magic)); 755 md->md_version = G_RAID3_VERSION; 756 strlcpy(md->md_name, sc->sc_name, sizeof(md->md_name)); 757 md->md_id = sc->sc_id; 758 md->md_all = sc->sc_ndisks; 759 md->md_genid = sc->sc_genid; 760 md->md_mediasize = sc->sc_mediasize; 761 md->md_sectorsize = sc->sc_sectorsize; 762 md->md_mflags = (sc->sc_flags & G_RAID3_DEVICE_FLAG_MASK); 763 md->md_no = disk->d_no; 764 md->md_syncid = disk->d_sync.ds_syncid; 765 md->md_dflags = (disk->d_flags & G_RAID3_DISK_FLAG_MASK); 766 if (disk->d_state != G_RAID3_DISK_STATE_SYNCHRONIZING) 767 md->md_sync_offset = 0; 768 else { 769 md->md_sync_offset = 770 disk->d_sync.ds_offset_done / (sc->sc_ndisks - 1); 771 } 772 if (disk->d_consumer != NULL && disk->d_consumer->provider != NULL) 773 pp = disk->d_consumer->provider; 774 else 775 pp = NULL; 776 if ((disk->d_flags & G_RAID3_DISK_FLAG_HARDCODED) != 0 && pp != NULL) 777 strlcpy(md->md_provider, pp->name, sizeof(md->md_provider)); 778 else 779 bzero(md->md_provider, sizeof(md->md_provider)); 780 if (pp != NULL) 781 md->md_provsize = pp->mediasize; 782 else 783 md->md_provsize = 0; 784 } 785 786 void 787 g_raid3_update_metadata(struct g_raid3_disk *disk) 788 { 789 struct g_raid3_softc *sc; 790 struct g_raid3_metadata md; 791 int error; 792 793 g_topology_assert_not(); 794 sc = disk->d_softc; 795 sx_assert(&sc->sc_lock, SX_LOCKED); 796 797 g_raid3_fill_metadata(disk, &md); 798 error = g_raid3_write_metadata(disk, &md); 799 if (error == 0) { 800 G_RAID3_DEBUG(2, "Metadata on %s updated.", 801 g_raid3_get_diskname(disk)); 802 } else { 803 G_RAID3_DEBUG(0, 804 "Cannot update metadata on disk %s (error=%d).", 805 g_raid3_get_diskname(disk), error); 806 } 807 } 808 809 static void 810 g_raid3_bump_syncid(struct g_raid3_softc *sc) 811 { 812 struct g_raid3_disk *disk; 813 u_int n; 814 815 g_topology_assert_not(); 816 sx_assert(&sc->sc_lock, SX_XLOCKED); 817 KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) > 0, 818 ("%s called with no active disks (device=%s).", __func__, 819 sc->sc_name)); 820 821 sc->sc_syncid++; 822 G_RAID3_DEBUG(1, "Device %s: syncid bumped to %u.", sc->sc_name, 823 sc->sc_syncid); 824 for (n = 0; n < sc->sc_ndisks; n++) { 825 disk = &sc->sc_disks[n]; 826 if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 827 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 828 disk->d_sync.ds_syncid = sc->sc_syncid; 829 g_raid3_update_metadata(disk); 830 } 831 } 832 } 833 834 static void 835 g_raid3_bump_genid(struct g_raid3_softc *sc) 836 { 837 struct g_raid3_disk *disk; 838 u_int n; 839 840 g_topology_assert_not(); 841 sx_assert(&sc->sc_lock, SX_XLOCKED); 842 KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) > 0, 843 ("%s called with no active disks (device=%s).", __func__, 844 sc->sc_name)); 845 846 sc->sc_genid++; 847 G_RAID3_DEBUG(1, "Device %s: genid bumped to %u.", sc->sc_name, 848 sc->sc_genid); 849 for (n = 0; n < sc->sc_ndisks; n++) { 850 disk = &sc->sc_disks[n]; 851 if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 852 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 853 disk->d_genid = sc->sc_genid; 854 g_raid3_update_metadata(disk); 855 } 856 } 857 } 858 859 static int 860 g_raid3_idle(struct g_raid3_softc *sc, int acw) 861 { 862 struct g_raid3_disk *disk; 863 u_int i; 864 int timeout; 865 866 g_topology_assert_not(); 867 sx_assert(&sc->sc_lock, SX_XLOCKED); 868 869 if (sc->sc_provider == NULL) 870 return (0); 871 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0) 872 return (0); 873 if (sc->sc_idle) 874 return (0); 875 if (sc->sc_writes > 0) 876 return (0); 877 if (acw > 0 || (acw == -1 && sc->sc_provider->acw > 0)) { 878 timeout = g_raid3_idletime - (time_uptime - sc->sc_last_write); 879 if (timeout > 0) 880 return (timeout); 881 } 882 sc->sc_idle = 1; 883 for (i = 0; i < sc->sc_ndisks; i++) { 884 disk = &sc->sc_disks[i]; 885 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) 886 continue; 887 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as clean.", 888 g_raid3_get_diskname(disk), sc->sc_name); 889 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 890 g_raid3_update_metadata(disk); 891 } 892 return (0); 893 } 894 895 static void 896 g_raid3_unidle(struct g_raid3_softc *sc) 897 { 898 struct g_raid3_disk *disk; 899 u_int i; 900 901 g_topology_assert_not(); 902 sx_assert(&sc->sc_lock, SX_XLOCKED); 903 904 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0) 905 return; 906 sc->sc_idle = 0; 907 sc->sc_last_write = time_uptime; 908 for (i = 0; i < sc->sc_ndisks; i++) { 909 disk = &sc->sc_disks[i]; 910 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) 911 continue; 912 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as dirty.", 913 g_raid3_get_diskname(disk), sc->sc_name); 914 disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY; 915 g_raid3_update_metadata(disk); 916 } 917 } 918 919 /* 920 * Treat bio_driver1 field in parent bio as list head and field bio_caller1 921 * in child bio as pointer to the next element on the list. 922 */ 923 #define G_RAID3_HEAD_BIO(pbp) (pbp)->bio_driver1 924 925 #define G_RAID3_NEXT_BIO(cbp) (cbp)->bio_caller1 926 927 #define G_RAID3_FOREACH_BIO(pbp, bp) \ 928 for ((bp) = G_RAID3_HEAD_BIO(pbp); (bp) != NULL; \ 929 (bp) = G_RAID3_NEXT_BIO(bp)) 930 931 #define G_RAID3_FOREACH_SAFE_BIO(pbp, bp, tmpbp) \ 932 for ((bp) = G_RAID3_HEAD_BIO(pbp); \ 933 (bp) != NULL && ((tmpbp) = G_RAID3_NEXT_BIO(bp), 1); \ 934 (bp) = (tmpbp)) 935 936 static void 937 g_raid3_init_bio(struct bio *pbp) 938 { 939 940 G_RAID3_HEAD_BIO(pbp) = NULL; 941 } 942 943 static void 944 g_raid3_remove_bio(struct bio *cbp) 945 { 946 struct bio *pbp, *bp; 947 948 pbp = cbp->bio_parent; 949 if (G_RAID3_HEAD_BIO(pbp) == cbp) 950 G_RAID3_HEAD_BIO(pbp) = G_RAID3_NEXT_BIO(cbp); 951 else { 952 G_RAID3_FOREACH_BIO(pbp, bp) { 953 if (G_RAID3_NEXT_BIO(bp) == cbp) { 954 G_RAID3_NEXT_BIO(bp) = G_RAID3_NEXT_BIO(cbp); 955 break; 956 } 957 } 958 } 959 G_RAID3_NEXT_BIO(cbp) = NULL; 960 } 961 962 static void 963 g_raid3_replace_bio(struct bio *sbp, struct bio *dbp) 964 { 965 struct bio *pbp, *bp; 966 967 g_raid3_remove_bio(sbp); 968 pbp = dbp->bio_parent; 969 G_RAID3_NEXT_BIO(sbp) = G_RAID3_NEXT_BIO(dbp); 970 if (G_RAID3_HEAD_BIO(pbp) == dbp) 971 G_RAID3_HEAD_BIO(pbp) = sbp; 972 else { 973 G_RAID3_FOREACH_BIO(pbp, bp) { 974 if (G_RAID3_NEXT_BIO(bp) == dbp) { 975 G_RAID3_NEXT_BIO(bp) = sbp; 976 break; 977 } 978 } 979 } 980 G_RAID3_NEXT_BIO(dbp) = NULL; 981 } 982 983 static void 984 g_raid3_destroy_bio(struct g_raid3_softc *sc, struct bio *cbp) 985 { 986 struct bio *bp, *pbp; 987 size_t size; 988 989 pbp = cbp->bio_parent; 990 pbp->bio_children--; 991 KASSERT(cbp->bio_data != NULL, ("NULL bio_data")); 992 size = pbp->bio_length / (sc->sc_ndisks - 1); 993 g_raid3_free(sc, cbp->bio_data, size); 994 if (G_RAID3_HEAD_BIO(pbp) == cbp) { 995 G_RAID3_HEAD_BIO(pbp) = G_RAID3_NEXT_BIO(cbp); 996 G_RAID3_NEXT_BIO(cbp) = NULL; 997 g_destroy_bio(cbp); 998 } else { 999 G_RAID3_FOREACH_BIO(pbp, bp) { 1000 if (G_RAID3_NEXT_BIO(bp) == cbp) 1001 break; 1002 } 1003 if (bp != NULL) { 1004 KASSERT(G_RAID3_NEXT_BIO(bp) != NULL, 1005 ("NULL bp->bio_driver1")); 1006 G_RAID3_NEXT_BIO(bp) = G_RAID3_NEXT_BIO(cbp); 1007 G_RAID3_NEXT_BIO(cbp) = NULL; 1008 } 1009 g_destroy_bio(cbp); 1010 } 1011 } 1012 1013 static struct bio * 1014 g_raid3_clone_bio(struct g_raid3_softc *sc, struct bio *pbp) 1015 { 1016 struct bio *bp, *cbp; 1017 size_t size; 1018 int memflag; 1019 1020 cbp = g_clone_bio(pbp); 1021 if (cbp == NULL) 1022 return (NULL); 1023 size = pbp->bio_length / (sc->sc_ndisks - 1); 1024 if ((pbp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) != 0) 1025 memflag = M_WAITOK; 1026 else 1027 memflag = M_NOWAIT; 1028 cbp->bio_data = g_raid3_alloc(sc, size, memflag); 1029 if (cbp->bio_data == NULL) { 1030 pbp->bio_children--; 1031 g_destroy_bio(cbp); 1032 return (NULL); 1033 } 1034 G_RAID3_NEXT_BIO(cbp) = NULL; 1035 if (G_RAID3_HEAD_BIO(pbp) == NULL) 1036 G_RAID3_HEAD_BIO(pbp) = cbp; 1037 else { 1038 G_RAID3_FOREACH_BIO(pbp, bp) { 1039 if (G_RAID3_NEXT_BIO(bp) == NULL) { 1040 G_RAID3_NEXT_BIO(bp) = cbp; 1041 break; 1042 } 1043 } 1044 } 1045 return (cbp); 1046 } 1047 1048 static void 1049 g_raid3_scatter(struct bio *pbp) 1050 { 1051 struct g_raid3_softc *sc; 1052 struct g_raid3_disk *disk; 1053 struct bio *bp, *cbp, *tmpbp; 1054 off_t atom, cadd, padd, left; 1055 int first; 1056 1057 sc = pbp->bio_to->geom->softc; 1058 bp = NULL; 1059 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_NOPARITY) == 0) { 1060 /* 1061 * Find bio for which we should calculate data. 1062 */ 1063 G_RAID3_FOREACH_BIO(pbp, cbp) { 1064 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) { 1065 bp = cbp; 1066 break; 1067 } 1068 } 1069 KASSERT(bp != NULL, ("NULL parity bio.")); 1070 } 1071 atom = sc->sc_sectorsize / (sc->sc_ndisks - 1); 1072 cadd = padd = 0; 1073 for (left = pbp->bio_length; left > 0; left -= sc->sc_sectorsize) { 1074 G_RAID3_FOREACH_BIO(pbp, cbp) { 1075 if (cbp == bp) 1076 continue; 1077 bcopy(pbp->bio_data + padd, cbp->bio_data + cadd, atom); 1078 padd += atom; 1079 } 1080 cadd += atom; 1081 } 1082 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_NOPARITY) == 0) { 1083 /* 1084 * Calculate parity. 1085 */ 1086 first = 1; 1087 G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) { 1088 if (cbp == bp) 1089 continue; 1090 if (first) { 1091 bcopy(cbp->bio_data, bp->bio_data, 1092 bp->bio_length); 1093 first = 0; 1094 } else { 1095 g_raid3_xor(cbp->bio_data, bp->bio_data, 1096 bp->bio_length); 1097 } 1098 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_NODISK) != 0) 1099 g_raid3_destroy_bio(sc, cbp); 1100 } 1101 } 1102 G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) { 1103 struct g_consumer *cp; 1104 1105 disk = cbp->bio_caller2; 1106 cp = disk->d_consumer; 1107 cbp->bio_to = cp->provider; 1108 G_RAID3_LOGREQ(3, cbp, "Sending request."); 1109 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1110 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1111 cp->acr, cp->acw, cp->ace)); 1112 cp->index++; 1113 sc->sc_writes++; 1114 g_io_request(cbp, cp); 1115 } 1116 } 1117 1118 static void 1119 g_raid3_gather(struct bio *pbp) 1120 { 1121 struct g_raid3_softc *sc; 1122 struct g_raid3_disk *disk; 1123 struct bio *xbp, *fbp, *cbp; 1124 off_t atom, cadd, padd, left; 1125 1126 sc = pbp->bio_to->geom->softc; 1127 /* 1128 * Find bio for which we have to calculate data. 1129 * While going through this path, check if all requests 1130 * succeeded, if not, deny whole request. 1131 * If we're in COMPLETE mode, we allow one request to fail, 1132 * so if we find one, we're sending it to the parity consumer. 1133 * If there are more failed requests, we deny whole request. 1134 */ 1135 xbp = fbp = NULL; 1136 G_RAID3_FOREACH_BIO(pbp, cbp) { 1137 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) { 1138 KASSERT(xbp == NULL, ("More than one parity bio.")); 1139 xbp = cbp; 1140 } 1141 if (cbp->bio_error == 0) 1142 continue; 1143 /* 1144 * Found failed request. 1145 */ 1146 if (fbp == NULL) { 1147 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_DEGRADED) != 0) { 1148 /* 1149 * We are already in degraded mode, so we can't 1150 * accept any failures. 1151 */ 1152 if (pbp->bio_error == 0) 1153 pbp->bio_error = cbp->bio_error; 1154 } else { 1155 fbp = cbp; 1156 } 1157 } else { 1158 /* 1159 * Next failed request, that's too many. 1160 */ 1161 if (pbp->bio_error == 0) 1162 pbp->bio_error = fbp->bio_error; 1163 } 1164 disk = cbp->bio_caller2; 1165 if (disk == NULL) 1166 continue; 1167 if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) { 1168 disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN; 1169 G_RAID3_LOGREQ(0, cbp, "Request failed (error=%d).", 1170 cbp->bio_error); 1171 } else { 1172 G_RAID3_LOGREQ(1, cbp, "Request failed (error=%d).", 1173 cbp->bio_error); 1174 } 1175 if (g_raid3_disconnect_on_failure && 1176 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1177 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 1178 g_raid3_event_send(disk, 1179 G_RAID3_DISK_STATE_DISCONNECTED, 1180 G_RAID3_EVENT_DONTWAIT); 1181 } 1182 } 1183 if (pbp->bio_error != 0) 1184 goto finish; 1185 if (fbp != NULL && (pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) { 1186 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_VERIFY; 1187 if (xbp != fbp) 1188 g_raid3_replace_bio(xbp, fbp); 1189 g_raid3_destroy_bio(sc, fbp); 1190 } else if (fbp != NULL) { 1191 struct g_consumer *cp; 1192 1193 /* 1194 * One request failed, so send the same request to 1195 * the parity consumer. 1196 */ 1197 disk = pbp->bio_driver2; 1198 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) { 1199 pbp->bio_error = fbp->bio_error; 1200 goto finish; 1201 } 1202 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED; 1203 pbp->bio_inbed--; 1204 fbp->bio_flags &= ~(BIO_DONE | BIO_ERROR); 1205 if (disk->d_no == sc->sc_ndisks - 1) 1206 fbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1207 fbp->bio_error = 0; 1208 fbp->bio_completed = 0; 1209 fbp->bio_children = 0; 1210 fbp->bio_inbed = 0; 1211 cp = disk->d_consumer; 1212 fbp->bio_caller2 = disk; 1213 fbp->bio_to = cp->provider; 1214 G_RAID3_LOGREQ(3, fbp, "Sending request (recover)."); 1215 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1216 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1217 cp->acr, cp->acw, cp->ace)); 1218 cp->index++; 1219 g_io_request(fbp, cp); 1220 return; 1221 } 1222 if (xbp != NULL) { 1223 /* 1224 * Calculate parity. 1225 */ 1226 G_RAID3_FOREACH_BIO(pbp, cbp) { 1227 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) 1228 continue; 1229 g_raid3_xor(cbp->bio_data, xbp->bio_data, 1230 xbp->bio_length); 1231 } 1232 xbp->bio_cflags &= ~G_RAID3_BIO_CFLAG_PARITY; 1233 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) { 1234 if (!g_raid3_is_zero(xbp)) { 1235 g_raid3_parity_mismatch++; 1236 pbp->bio_error = EIO; 1237 goto finish; 1238 } 1239 g_raid3_destroy_bio(sc, xbp); 1240 } 1241 } 1242 atom = sc->sc_sectorsize / (sc->sc_ndisks - 1); 1243 cadd = padd = 0; 1244 for (left = pbp->bio_length; left > 0; left -= sc->sc_sectorsize) { 1245 G_RAID3_FOREACH_BIO(pbp, cbp) { 1246 bcopy(cbp->bio_data + cadd, pbp->bio_data + padd, atom); 1247 pbp->bio_completed += atom; 1248 padd += atom; 1249 } 1250 cadd += atom; 1251 } 1252 finish: 1253 if (pbp->bio_error == 0) 1254 G_RAID3_LOGREQ(3, pbp, "Request finished."); 1255 else { 1256 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) 1257 G_RAID3_LOGREQ(1, pbp, "Verification error."); 1258 else 1259 G_RAID3_LOGREQ(0, pbp, "Request failed."); 1260 } 1261 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_MASK; 1262 while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) 1263 g_raid3_destroy_bio(sc, cbp); 1264 g_io_deliver(pbp, pbp->bio_error); 1265 } 1266 1267 static void 1268 g_raid3_done(struct bio *bp) 1269 { 1270 struct g_raid3_softc *sc; 1271 1272 sc = bp->bio_from->geom->softc; 1273 bp->bio_cflags |= G_RAID3_BIO_CFLAG_REGULAR; 1274 G_RAID3_LOGREQ(3, bp, "Regular request done (error=%d).", bp->bio_error); 1275 mtx_lock(&sc->sc_queue_mtx); 1276 bioq_insert_head(&sc->sc_queue, bp); 1277 mtx_unlock(&sc->sc_queue_mtx); 1278 wakeup(sc); 1279 wakeup(&sc->sc_queue); 1280 } 1281 1282 static void 1283 g_raid3_regular_request(struct bio *cbp) 1284 { 1285 struct g_raid3_softc *sc; 1286 struct g_raid3_disk *disk; 1287 struct bio *pbp; 1288 1289 g_topology_assert_not(); 1290 1291 pbp = cbp->bio_parent; 1292 sc = pbp->bio_to->geom->softc; 1293 cbp->bio_from->index--; 1294 if (cbp->bio_cmd == BIO_WRITE) 1295 sc->sc_writes--; 1296 disk = cbp->bio_from->private; 1297 if (disk == NULL) { 1298 g_topology_lock(); 1299 g_raid3_kill_consumer(sc, cbp->bio_from); 1300 g_topology_unlock(); 1301 } 1302 1303 G_RAID3_LOGREQ(3, cbp, "Request finished."); 1304 pbp->bio_inbed++; 1305 KASSERT(pbp->bio_inbed <= pbp->bio_children, 1306 ("bio_inbed (%u) is bigger than bio_children (%u).", pbp->bio_inbed, 1307 pbp->bio_children)); 1308 if (pbp->bio_inbed != pbp->bio_children) 1309 return; 1310 switch (pbp->bio_cmd) { 1311 case BIO_READ: 1312 g_raid3_gather(pbp); 1313 break; 1314 case BIO_WRITE: 1315 case BIO_DELETE: 1316 { 1317 int error = 0; 1318 1319 pbp->bio_completed = pbp->bio_length; 1320 while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) { 1321 if (cbp->bio_error == 0) { 1322 g_raid3_destroy_bio(sc, cbp); 1323 continue; 1324 } 1325 1326 if (error == 0) 1327 error = cbp->bio_error; 1328 else if (pbp->bio_error == 0) { 1329 /* 1330 * Next failed request, that's too many. 1331 */ 1332 pbp->bio_error = error; 1333 } 1334 1335 disk = cbp->bio_caller2; 1336 if (disk == NULL) { 1337 g_raid3_destroy_bio(sc, cbp); 1338 continue; 1339 } 1340 1341 if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) { 1342 disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN; 1343 G_RAID3_LOGREQ(0, cbp, 1344 "Request failed (error=%d).", 1345 cbp->bio_error); 1346 } else { 1347 G_RAID3_LOGREQ(1, cbp, 1348 "Request failed (error=%d).", 1349 cbp->bio_error); 1350 } 1351 if (g_raid3_disconnect_on_failure && 1352 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1353 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 1354 g_raid3_event_send(disk, 1355 G_RAID3_DISK_STATE_DISCONNECTED, 1356 G_RAID3_EVENT_DONTWAIT); 1357 } 1358 g_raid3_destroy_bio(sc, cbp); 1359 } 1360 if (pbp->bio_error == 0) 1361 G_RAID3_LOGREQ(3, pbp, "Request finished."); 1362 else 1363 G_RAID3_LOGREQ(0, pbp, "Request failed."); 1364 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_DEGRADED; 1365 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_NOPARITY; 1366 bioq_remove(&sc->sc_inflight, pbp); 1367 /* Release delayed sync requests if possible. */ 1368 g_raid3_sync_release(sc); 1369 g_io_deliver(pbp, pbp->bio_error); 1370 break; 1371 } 1372 } 1373 } 1374 1375 static void 1376 g_raid3_sync_done(struct bio *bp) 1377 { 1378 struct g_raid3_softc *sc; 1379 1380 G_RAID3_LOGREQ(3, bp, "Synchronization request delivered."); 1381 sc = bp->bio_from->geom->softc; 1382 bp->bio_cflags |= G_RAID3_BIO_CFLAG_SYNC; 1383 mtx_lock(&sc->sc_queue_mtx); 1384 bioq_insert_head(&sc->sc_queue, bp); 1385 mtx_unlock(&sc->sc_queue_mtx); 1386 wakeup(sc); 1387 wakeup(&sc->sc_queue); 1388 } 1389 1390 static void 1391 g_raid3_flush(struct g_raid3_softc *sc, struct bio *bp) 1392 { 1393 struct bio_queue_head queue; 1394 struct g_raid3_disk *disk; 1395 struct g_consumer *cp; 1396 struct bio *cbp; 1397 u_int i; 1398 1399 bioq_init(&queue); 1400 for (i = 0; i < sc->sc_ndisks; i++) { 1401 disk = &sc->sc_disks[i]; 1402 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) 1403 continue; 1404 cbp = g_clone_bio(bp); 1405 if (cbp == NULL) { 1406 for (cbp = bioq_first(&queue); cbp != NULL; 1407 cbp = bioq_first(&queue)) { 1408 bioq_remove(&queue, cbp); 1409 g_destroy_bio(cbp); 1410 } 1411 if (bp->bio_error == 0) 1412 bp->bio_error = ENOMEM; 1413 g_io_deliver(bp, bp->bio_error); 1414 return; 1415 } 1416 bioq_insert_tail(&queue, cbp); 1417 cbp->bio_done = g_std_done; 1418 cbp->bio_caller1 = disk; 1419 cbp->bio_to = disk->d_consumer->provider; 1420 } 1421 for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { 1422 bioq_remove(&queue, cbp); 1423 G_RAID3_LOGREQ(3, cbp, "Sending request."); 1424 disk = cbp->bio_caller1; 1425 cbp->bio_caller1 = NULL; 1426 cp = disk->d_consumer; 1427 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1428 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1429 cp->acr, cp->acw, cp->ace)); 1430 g_io_request(cbp, disk->d_consumer); 1431 } 1432 } 1433 1434 static void 1435 g_raid3_start(struct bio *bp) 1436 { 1437 struct g_raid3_softc *sc; 1438 1439 sc = bp->bio_to->geom->softc; 1440 /* 1441 * If sc == NULL or there are no valid disks, provider's error 1442 * should be set and g_raid3_start() should not be called at all. 1443 */ 1444 KASSERT(sc != NULL && (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 1445 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE), 1446 ("Provider's error should be set (error=%d)(device=%s).", 1447 bp->bio_to->error, bp->bio_to->name)); 1448 G_RAID3_LOGREQ(3, bp, "Request received."); 1449 1450 switch (bp->bio_cmd) { 1451 case BIO_READ: 1452 case BIO_WRITE: 1453 case BIO_DELETE: 1454 break; 1455 case BIO_FLUSH: 1456 g_raid3_flush(sc, bp); 1457 return; 1458 case BIO_GETATTR: 1459 default: 1460 g_io_deliver(bp, EOPNOTSUPP); 1461 return; 1462 } 1463 mtx_lock(&sc->sc_queue_mtx); 1464 bioq_insert_tail(&sc->sc_queue, bp); 1465 mtx_unlock(&sc->sc_queue_mtx); 1466 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc); 1467 wakeup(sc); 1468 } 1469 1470 /* 1471 * Return TRUE if the given request is colliding with a in-progress 1472 * synchronization request. 1473 */ 1474 static int 1475 g_raid3_sync_collision(struct g_raid3_softc *sc, struct bio *bp) 1476 { 1477 struct g_raid3_disk *disk; 1478 struct bio *sbp; 1479 off_t rstart, rend, sstart, send; 1480 int i; 1481 1482 disk = sc->sc_syncdisk; 1483 if (disk == NULL) 1484 return (0); 1485 rstart = bp->bio_offset; 1486 rend = bp->bio_offset + bp->bio_length; 1487 for (i = 0; i < g_raid3_syncreqs; i++) { 1488 sbp = disk->d_sync.ds_bios[i]; 1489 if (sbp == NULL) 1490 continue; 1491 sstart = sbp->bio_offset; 1492 send = sbp->bio_length; 1493 if (sbp->bio_cmd == BIO_WRITE) { 1494 sstart *= sc->sc_ndisks - 1; 1495 send *= sc->sc_ndisks - 1; 1496 } 1497 send += sstart; 1498 if (rend > sstart && rstart < send) 1499 return (1); 1500 } 1501 return (0); 1502 } 1503 1504 /* 1505 * Return TRUE if the given sync request is colliding with a in-progress regular 1506 * request. 1507 */ 1508 static int 1509 g_raid3_regular_collision(struct g_raid3_softc *sc, struct bio *sbp) 1510 { 1511 off_t rstart, rend, sstart, send; 1512 struct bio *bp; 1513 1514 if (sc->sc_syncdisk == NULL) 1515 return (0); 1516 sstart = sbp->bio_offset; 1517 send = sstart + sbp->bio_length; 1518 TAILQ_FOREACH(bp, &sc->sc_inflight.queue, bio_queue) { 1519 rstart = bp->bio_offset; 1520 rend = bp->bio_offset + bp->bio_length; 1521 if (rend > sstart && rstart < send) 1522 return (1); 1523 } 1524 return (0); 1525 } 1526 1527 /* 1528 * Puts request onto delayed queue. 1529 */ 1530 static void 1531 g_raid3_regular_delay(struct g_raid3_softc *sc, struct bio *bp) 1532 { 1533 1534 G_RAID3_LOGREQ(2, bp, "Delaying request."); 1535 bioq_insert_head(&sc->sc_regular_delayed, bp); 1536 } 1537 1538 /* 1539 * Puts synchronization request onto delayed queue. 1540 */ 1541 static void 1542 g_raid3_sync_delay(struct g_raid3_softc *sc, struct bio *bp) 1543 { 1544 1545 G_RAID3_LOGREQ(2, bp, "Delaying synchronization request."); 1546 bioq_insert_tail(&sc->sc_sync_delayed, bp); 1547 } 1548 1549 /* 1550 * Releases delayed regular requests which don't collide anymore with sync 1551 * requests. 1552 */ 1553 static void 1554 g_raid3_regular_release(struct g_raid3_softc *sc) 1555 { 1556 struct bio *bp, *bp2; 1557 1558 TAILQ_FOREACH_SAFE(bp, &sc->sc_regular_delayed.queue, bio_queue, bp2) { 1559 if (g_raid3_sync_collision(sc, bp)) 1560 continue; 1561 bioq_remove(&sc->sc_regular_delayed, bp); 1562 G_RAID3_LOGREQ(2, bp, "Releasing delayed request (%p).", bp); 1563 mtx_lock(&sc->sc_queue_mtx); 1564 bioq_insert_head(&sc->sc_queue, bp); 1565 #if 0 1566 /* 1567 * wakeup() is not needed, because this function is called from 1568 * the worker thread. 1569 */ 1570 wakeup(&sc->sc_queue); 1571 #endif 1572 mtx_unlock(&sc->sc_queue_mtx); 1573 } 1574 } 1575 1576 /* 1577 * Releases delayed sync requests which don't collide anymore with regular 1578 * requests. 1579 */ 1580 static void 1581 g_raid3_sync_release(struct g_raid3_softc *sc) 1582 { 1583 struct bio *bp, *bp2; 1584 1585 TAILQ_FOREACH_SAFE(bp, &sc->sc_sync_delayed.queue, bio_queue, bp2) { 1586 if (g_raid3_regular_collision(sc, bp)) 1587 continue; 1588 bioq_remove(&sc->sc_sync_delayed, bp); 1589 G_RAID3_LOGREQ(2, bp, 1590 "Releasing delayed synchronization request."); 1591 g_io_request(bp, bp->bio_from); 1592 } 1593 } 1594 1595 /* 1596 * Handle synchronization requests. 1597 * Every synchronization request is two-steps process: first, READ request is 1598 * send to active provider and then WRITE request (with read data) to the provider 1599 * beeing synchronized. When WRITE is finished, new synchronization request is 1600 * send. 1601 */ 1602 static void 1603 g_raid3_sync_request(struct bio *bp) 1604 { 1605 struct g_raid3_softc *sc; 1606 struct g_raid3_disk *disk; 1607 1608 bp->bio_from->index--; 1609 sc = bp->bio_from->geom->softc; 1610 disk = bp->bio_from->private; 1611 if (disk == NULL) { 1612 sx_xunlock(&sc->sc_lock); /* Avoid recursion on sc_lock. */ 1613 g_topology_lock(); 1614 g_raid3_kill_consumer(sc, bp->bio_from); 1615 g_topology_unlock(); 1616 free(bp->bio_data, M_RAID3); 1617 g_destroy_bio(bp); 1618 sx_xlock(&sc->sc_lock); 1619 return; 1620 } 1621 1622 /* 1623 * Synchronization request. 1624 */ 1625 switch (bp->bio_cmd) { 1626 case BIO_READ: 1627 { 1628 struct g_consumer *cp; 1629 u_char *dst, *src; 1630 off_t left; 1631 u_int atom; 1632 1633 if (bp->bio_error != 0) { 1634 G_RAID3_LOGREQ(0, bp, 1635 "Synchronization request failed (error=%d).", 1636 bp->bio_error); 1637 g_destroy_bio(bp); 1638 return; 1639 } 1640 G_RAID3_LOGREQ(3, bp, "Synchronization request finished."); 1641 atom = sc->sc_sectorsize / (sc->sc_ndisks - 1); 1642 dst = src = bp->bio_data; 1643 if (disk->d_no == sc->sc_ndisks - 1) { 1644 u_int n; 1645 1646 /* Parity component. */ 1647 for (left = bp->bio_length; left > 0; 1648 left -= sc->sc_sectorsize) { 1649 bcopy(src, dst, atom); 1650 src += atom; 1651 for (n = 1; n < sc->sc_ndisks - 1; n++) { 1652 g_raid3_xor(src, dst, atom); 1653 src += atom; 1654 } 1655 dst += atom; 1656 } 1657 } else { 1658 /* Regular component. */ 1659 src += atom * disk->d_no; 1660 for (left = bp->bio_length; left > 0; 1661 left -= sc->sc_sectorsize) { 1662 bcopy(src, dst, atom); 1663 src += sc->sc_sectorsize; 1664 dst += atom; 1665 } 1666 } 1667 bp->bio_driver1 = bp->bio_driver2 = NULL; 1668 bp->bio_pflags = 0; 1669 bp->bio_offset /= sc->sc_ndisks - 1; 1670 bp->bio_length /= sc->sc_ndisks - 1; 1671 bp->bio_cmd = BIO_WRITE; 1672 bp->bio_cflags = 0; 1673 bp->bio_children = bp->bio_inbed = 0; 1674 cp = disk->d_consumer; 1675 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1676 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1677 cp->acr, cp->acw, cp->ace)); 1678 cp->index++; 1679 g_io_request(bp, cp); 1680 return; 1681 } 1682 case BIO_WRITE: 1683 { 1684 struct g_raid3_disk_sync *sync; 1685 off_t boffset, moffset; 1686 void *data; 1687 int i; 1688 1689 if (bp->bio_error != 0) { 1690 G_RAID3_LOGREQ(0, bp, 1691 "Synchronization request failed (error=%d).", 1692 bp->bio_error); 1693 g_destroy_bio(bp); 1694 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 1695 g_raid3_event_send(disk, 1696 G_RAID3_DISK_STATE_DISCONNECTED, 1697 G_RAID3_EVENT_DONTWAIT); 1698 return; 1699 } 1700 G_RAID3_LOGREQ(3, bp, "Synchronization request finished."); 1701 sync = &disk->d_sync; 1702 if (sync->ds_offset == sc->sc_mediasize / (sc->sc_ndisks - 1) || 1703 sync->ds_consumer == NULL || 1704 (sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 1705 /* Don't send more synchronization requests. */ 1706 sync->ds_inflight--; 1707 if (sync->ds_bios != NULL) { 1708 i = (int)(uintptr_t)bp->bio_caller1; 1709 sync->ds_bios[i] = NULL; 1710 } 1711 free(bp->bio_data, M_RAID3); 1712 g_destroy_bio(bp); 1713 if (sync->ds_inflight > 0) 1714 return; 1715 if (sync->ds_consumer == NULL || 1716 (sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 1717 return; 1718 } 1719 /* 1720 * Disk up-to-date, activate it. 1721 */ 1722 g_raid3_event_send(disk, G_RAID3_DISK_STATE_ACTIVE, 1723 G_RAID3_EVENT_DONTWAIT); 1724 return; 1725 } 1726 1727 /* Send next synchronization request. */ 1728 data = bp->bio_data; 1729 bzero(bp, sizeof(*bp)); 1730 bp->bio_cmd = BIO_READ; 1731 bp->bio_offset = sync->ds_offset * (sc->sc_ndisks - 1); 1732 bp->bio_length = MIN(MAXPHYS, sc->sc_mediasize - bp->bio_offset); 1733 sync->ds_offset += bp->bio_length / (sc->sc_ndisks - 1); 1734 bp->bio_done = g_raid3_sync_done; 1735 bp->bio_data = data; 1736 bp->bio_from = sync->ds_consumer; 1737 bp->bio_to = sc->sc_provider; 1738 G_RAID3_LOGREQ(3, bp, "Sending synchronization request."); 1739 sync->ds_consumer->index++; 1740 /* 1741 * Delay the request if it is colliding with a regular request. 1742 */ 1743 if (g_raid3_regular_collision(sc, bp)) 1744 g_raid3_sync_delay(sc, bp); 1745 else 1746 g_io_request(bp, sync->ds_consumer); 1747 1748 /* Release delayed requests if possible. */ 1749 g_raid3_regular_release(sc); 1750 1751 /* Find the smallest offset. */ 1752 moffset = sc->sc_mediasize; 1753 for (i = 0; i < g_raid3_syncreqs; i++) { 1754 bp = sync->ds_bios[i]; 1755 boffset = bp->bio_offset; 1756 if (bp->bio_cmd == BIO_WRITE) 1757 boffset *= sc->sc_ndisks - 1; 1758 if (boffset < moffset) 1759 moffset = boffset; 1760 } 1761 if (sync->ds_offset_done + (MAXPHYS * 100) < moffset) { 1762 /* Update offset_done on every 100 blocks. */ 1763 sync->ds_offset_done = moffset; 1764 g_raid3_update_metadata(disk); 1765 } 1766 return; 1767 } 1768 default: 1769 KASSERT(1 == 0, ("Invalid command here: %u (device=%s)", 1770 bp->bio_cmd, sc->sc_name)); 1771 break; 1772 } 1773 } 1774 1775 static int 1776 g_raid3_register_request(struct bio *pbp) 1777 { 1778 struct g_raid3_softc *sc; 1779 struct g_raid3_disk *disk; 1780 struct g_consumer *cp; 1781 struct bio *cbp, *tmpbp; 1782 off_t offset, length; 1783 u_int n, ndisks; 1784 int round_robin, verify; 1785 1786 ndisks = 0; 1787 sc = pbp->bio_to->geom->softc; 1788 if ((pbp->bio_cflags & G_RAID3_BIO_CFLAG_REGSYNC) != 0 && 1789 sc->sc_syncdisk == NULL) { 1790 g_io_deliver(pbp, EIO); 1791 return (0); 1792 } 1793 g_raid3_init_bio(pbp); 1794 length = pbp->bio_length / (sc->sc_ndisks - 1); 1795 offset = pbp->bio_offset / (sc->sc_ndisks - 1); 1796 round_robin = verify = 0; 1797 switch (pbp->bio_cmd) { 1798 case BIO_READ: 1799 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_VERIFY) != 0 && 1800 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1801 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_VERIFY; 1802 verify = 1; 1803 ndisks = sc->sc_ndisks; 1804 } else { 1805 verify = 0; 1806 ndisks = sc->sc_ndisks - 1; 1807 } 1808 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0 && 1809 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1810 round_robin = 1; 1811 } else { 1812 round_robin = 0; 1813 } 1814 KASSERT(!round_robin || !verify, 1815 ("ROUND-ROBIN and VERIFY are mutually exclusive.")); 1816 pbp->bio_driver2 = &sc->sc_disks[sc->sc_ndisks - 1]; 1817 break; 1818 case BIO_WRITE: 1819 case BIO_DELETE: 1820 /* 1821 * Delay the request if it is colliding with a synchronization 1822 * request. 1823 */ 1824 if (g_raid3_sync_collision(sc, pbp)) { 1825 g_raid3_regular_delay(sc, pbp); 1826 return (0); 1827 } 1828 1829 if (sc->sc_idle) 1830 g_raid3_unidle(sc); 1831 else 1832 sc->sc_last_write = time_uptime; 1833 1834 ndisks = sc->sc_ndisks; 1835 break; 1836 } 1837 for (n = 0; n < ndisks; n++) { 1838 disk = &sc->sc_disks[n]; 1839 cbp = g_raid3_clone_bio(sc, pbp); 1840 if (cbp == NULL) { 1841 while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) 1842 g_raid3_destroy_bio(sc, cbp); 1843 /* 1844 * To prevent deadlock, we must run back up 1845 * with the ENOMEM for failed requests of any 1846 * of our consumers. Our own sync requests 1847 * can stick around, as they are finite. 1848 */ 1849 if ((pbp->bio_cflags & 1850 G_RAID3_BIO_CFLAG_REGULAR) != 0) { 1851 g_io_deliver(pbp, ENOMEM); 1852 return (0); 1853 } 1854 return (ENOMEM); 1855 } 1856 cbp->bio_offset = offset; 1857 cbp->bio_length = length; 1858 cbp->bio_done = g_raid3_done; 1859 switch (pbp->bio_cmd) { 1860 case BIO_READ: 1861 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) { 1862 /* 1863 * Replace invalid component with the parity 1864 * component. 1865 */ 1866 disk = &sc->sc_disks[sc->sc_ndisks - 1]; 1867 cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1868 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED; 1869 } else if (round_robin && 1870 disk->d_no == sc->sc_round_robin) { 1871 /* 1872 * In round-robin mode skip one data component 1873 * and use parity component when reading. 1874 */ 1875 pbp->bio_driver2 = disk; 1876 disk = &sc->sc_disks[sc->sc_ndisks - 1]; 1877 cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1878 sc->sc_round_robin++; 1879 round_robin = 0; 1880 } else if (verify && disk->d_no == sc->sc_ndisks - 1) { 1881 cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1882 } 1883 break; 1884 case BIO_WRITE: 1885 case BIO_DELETE: 1886 if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 1887 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 1888 if (n == ndisks - 1) { 1889 /* 1890 * Active parity component, mark it as such. 1891 */ 1892 cbp->bio_cflags |= 1893 G_RAID3_BIO_CFLAG_PARITY; 1894 } 1895 } else { 1896 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED; 1897 if (n == ndisks - 1) { 1898 /* 1899 * Parity component is not connected, 1900 * so destroy its request. 1901 */ 1902 pbp->bio_pflags |= 1903 G_RAID3_BIO_PFLAG_NOPARITY; 1904 g_raid3_destroy_bio(sc, cbp); 1905 cbp = NULL; 1906 } else { 1907 cbp->bio_cflags |= 1908 G_RAID3_BIO_CFLAG_NODISK; 1909 disk = NULL; 1910 } 1911 } 1912 break; 1913 } 1914 if (cbp != NULL) 1915 cbp->bio_caller2 = disk; 1916 } 1917 switch (pbp->bio_cmd) { 1918 case BIO_READ: 1919 if (round_robin) { 1920 /* 1921 * If we are in round-robin mode and 'round_robin' is 1922 * still 1, it means, that we skipped parity component 1923 * for this read and must reset sc_round_robin field. 1924 */ 1925 sc->sc_round_robin = 0; 1926 } 1927 G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) { 1928 disk = cbp->bio_caller2; 1929 cp = disk->d_consumer; 1930 cbp->bio_to = cp->provider; 1931 G_RAID3_LOGREQ(3, cbp, "Sending request."); 1932 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1933 ("Consumer %s not opened (r%dw%de%d).", 1934 cp->provider->name, cp->acr, cp->acw, cp->ace)); 1935 cp->index++; 1936 g_io_request(cbp, cp); 1937 } 1938 break; 1939 case BIO_WRITE: 1940 case BIO_DELETE: 1941 /* 1942 * Put request onto inflight queue, so we can check if new 1943 * synchronization requests don't collide with it. 1944 */ 1945 bioq_insert_tail(&sc->sc_inflight, pbp); 1946 1947 /* 1948 * Bump syncid on first write. 1949 */ 1950 if ((sc->sc_bump_id & G_RAID3_BUMP_SYNCID) != 0) { 1951 sc->sc_bump_id &= ~G_RAID3_BUMP_SYNCID; 1952 g_raid3_bump_syncid(sc); 1953 } 1954 g_raid3_scatter(pbp); 1955 break; 1956 } 1957 return (0); 1958 } 1959 1960 static int 1961 g_raid3_can_destroy(struct g_raid3_softc *sc) 1962 { 1963 struct g_geom *gp; 1964 struct g_consumer *cp; 1965 1966 g_topology_assert(); 1967 gp = sc->sc_geom; 1968 if (gp->softc == NULL) 1969 return (1); 1970 LIST_FOREACH(cp, &gp->consumer, consumer) { 1971 if (g_raid3_is_busy(sc, cp)) 1972 return (0); 1973 } 1974 gp = sc->sc_sync.ds_geom; 1975 LIST_FOREACH(cp, &gp->consumer, consumer) { 1976 if (g_raid3_is_busy(sc, cp)) 1977 return (0); 1978 } 1979 G_RAID3_DEBUG(2, "No I/O requests for %s, it can be destroyed.", 1980 sc->sc_name); 1981 return (1); 1982 } 1983 1984 static int 1985 g_raid3_try_destroy(struct g_raid3_softc *sc) 1986 { 1987 1988 g_topology_assert_not(); 1989 sx_assert(&sc->sc_lock, SX_XLOCKED); 1990 1991 if (sc->sc_rootmount != NULL) { 1992 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__, 1993 sc->sc_rootmount); 1994 root_mount_rel(sc->sc_rootmount); 1995 sc->sc_rootmount = NULL; 1996 } 1997 1998 g_topology_lock(); 1999 if (!g_raid3_can_destroy(sc)) { 2000 g_topology_unlock(); 2001 return (0); 2002 } 2003 sc->sc_geom->softc = NULL; 2004 sc->sc_sync.ds_geom->softc = NULL; 2005 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_WAIT) != 0) { 2006 g_topology_unlock(); 2007 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, 2008 &sc->sc_worker); 2009 /* Unlock sc_lock here, as it can be destroyed after wakeup. */ 2010 sx_xunlock(&sc->sc_lock); 2011 wakeup(&sc->sc_worker); 2012 sc->sc_worker = NULL; 2013 } else { 2014 g_topology_unlock(); 2015 g_raid3_destroy_device(sc); 2016 free(sc->sc_disks, M_RAID3); 2017 free(sc, M_RAID3); 2018 } 2019 return (1); 2020 } 2021 2022 /* 2023 * Worker thread. 2024 */ 2025 static void 2026 g_raid3_worker(void *arg) 2027 { 2028 struct g_raid3_softc *sc; 2029 struct g_raid3_event *ep; 2030 struct bio *bp; 2031 int timeout; 2032 2033 sc = arg; 2034 thread_lock(curthread); 2035 sched_prio(curthread, PRIBIO); 2036 thread_unlock(curthread); 2037 2038 sx_xlock(&sc->sc_lock); 2039 for (;;) { 2040 G_RAID3_DEBUG(5, "%s: Let's see...", __func__); 2041 /* 2042 * First take a look at events. 2043 * This is important to handle events before any I/O requests. 2044 */ 2045 ep = g_raid3_event_get(sc); 2046 if (ep != NULL) { 2047 g_raid3_event_remove(sc, ep); 2048 if ((ep->e_flags & G_RAID3_EVENT_DEVICE) != 0) { 2049 /* Update only device status. */ 2050 G_RAID3_DEBUG(3, 2051 "Running event for device %s.", 2052 sc->sc_name); 2053 ep->e_error = 0; 2054 g_raid3_update_device(sc, 1); 2055 } else { 2056 /* Update disk status. */ 2057 G_RAID3_DEBUG(3, "Running event for disk %s.", 2058 g_raid3_get_diskname(ep->e_disk)); 2059 ep->e_error = g_raid3_update_disk(ep->e_disk, 2060 ep->e_state); 2061 if (ep->e_error == 0) 2062 g_raid3_update_device(sc, 0); 2063 } 2064 if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) { 2065 KASSERT(ep->e_error == 0, 2066 ("Error cannot be handled.")); 2067 g_raid3_event_free(ep); 2068 } else { 2069 ep->e_flags |= G_RAID3_EVENT_DONE; 2070 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, 2071 ep); 2072 mtx_lock(&sc->sc_events_mtx); 2073 wakeup(ep); 2074 mtx_unlock(&sc->sc_events_mtx); 2075 } 2076 if ((sc->sc_flags & 2077 G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 2078 if (g_raid3_try_destroy(sc)) { 2079 curthread->td_pflags &= ~TDP_GEOM; 2080 G_RAID3_DEBUG(1, "Thread exiting."); 2081 kproc_exit(0); 2082 } 2083 } 2084 G_RAID3_DEBUG(5, "%s: I'm here 1.", __func__); 2085 continue; 2086 } 2087 /* 2088 * Check if we can mark array as CLEAN and if we can't take 2089 * how much seconds should we wait. 2090 */ 2091 timeout = g_raid3_idle(sc, -1); 2092 /* 2093 * Now I/O requests. 2094 */ 2095 /* Get first request from the queue. */ 2096 mtx_lock(&sc->sc_queue_mtx); 2097 bp = bioq_first(&sc->sc_queue); 2098 if (bp == NULL) { 2099 if ((sc->sc_flags & 2100 G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 2101 mtx_unlock(&sc->sc_queue_mtx); 2102 if (g_raid3_try_destroy(sc)) { 2103 curthread->td_pflags &= ~TDP_GEOM; 2104 G_RAID3_DEBUG(1, "Thread exiting."); 2105 kproc_exit(0); 2106 } 2107 mtx_lock(&sc->sc_queue_mtx); 2108 } 2109 sx_xunlock(&sc->sc_lock); 2110 /* 2111 * XXX: We can miss an event here, because an event 2112 * can be added without sx-device-lock and without 2113 * mtx-queue-lock. Maybe I should just stop using 2114 * dedicated mutex for events synchronization and 2115 * stick with the queue lock? 2116 * The event will hang here until next I/O request 2117 * or next event is received. 2118 */ 2119 MSLEEP(sc, &sc->sc_queue_mtx, PRIBIO | PDROP, "r3:w1", 2120 timeout * hz); 2121 sx_xlock(&sc->sc_lock); 2122 G_RAID3_DEBUG(5, "%s: I'm here 4.", __func__); 2123 continue; 2124 } 2125 process: 2126 bioq_remove(&sc->sc_queue, bp); 2127 mtx_unlock(&sc->sc_queue_mtx); 2128 2129 if (bp->bio_from->geom == sc->sc_sync.ds_geom && 2130 (bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) != 0) { 2131 g_raid3_sync_request(bp); /* READ */ 2132 } else if (bp->bio_to != sc->sc_provider) { 2133 if ((bp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) != 0) 2134 g_raid3_regular_request(bp); 2135 else if ((bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) != 0) 2136 g_raid3_sync_request(bp); /* WRITE */ 2137 else { 2138 KASSERT(0, 2139 ("Invalid request cflags=0x%hhx to=%s.", 2140 bp->bio_cflags, bp->bio_to->name)); 2141 } 2142 } else if (g_raid3_register_request(bp) != 0) { 2143 mtx_lock(&sc->sc_queue_mtx); 2144 bioq_insert_head(&sc->sc_queue, bp); 2145 /* 2146 * We are short in memory, let see if there are finished 2147 * request we can free. 2148 */ 2149 TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { 2150 if (bp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) 2151 goto process; 2152 } 2153 /* 2154 * No finished regular request, so at least keep 2155 * synchronization running. 2156 */ 2157 TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { 2158 if (bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) 2159 goto process; 2160 } 2161 sx_xunlock(&sc->sc_lock); 2162 MSLEEP(&sc->sc_queue, &sc->sc_queue_mtx, PRIBIO | PDROP, 2163 "r3:lowmem", hz / 10); 2164 sx_xlock(&sc->sc_lock); 2165 } 2166 G_RAID3_DEBUG(5, "%s: I'm here 9.", __func__); 2167 } 2168 } 2169 2170 static void 2171 g_raid3_update_idle(struct g_raid3_softc *sc, struct g_raid3_disk *disk) 2172 { 2173 2174 sx_assert(&sc->sc_lock, SX_LOCKED); 2175 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0) 2176 return; 2177 if (!sc->sc_idle && (disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) == 0) { 2178 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as dirty.", 2179 g_raid3_get_diskname(disk), sc->sc_name); 2180 disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY; 2181 } else if (sc->sc_idle && 2182 (disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) != 0) { 2183 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as clean.", 2184 g_raid3_get_diskname(disk), sc->sc_name); 2185 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2186 } 2187 } 2188 2189 static void 2190 g_raid3_sync_start(struct g_raid3_softc *sc) 2191 { 2192 struct g_raid3_disk *disk; 2193 struct g_consumer *cp; 2194 struct bio *bp; 2195 int error; 2196 u_int n; 2197 2198 g_topology_assert_not(); 2199 sx_assert(&sc->sc_lock, SX_XLOCKED); 2200 2201 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED, 2202 ("Device not in DEGRADED state (%s, %u).", sc->sc_name, 2203 sc->sc_state)); 2204 KASSERT(sc->sc_syncdisk == NULL, ("Syncdisk is not NULL (%s, %u).", 2205 sc->sc_name, sc->sc_state)); 2206 disk = NULL; 2207 for (n = 0; n < sc->sc_ndisks; n++) { 2208 if (sc->sc_disks[n].d_state != G_RAID3_DISK_STATE_SYNCHRONIZING) 2209 continue; 2210 disk = &sc->sc_disks[n]; 2211 break; 2212 } 2213 if (disk == NULL) 2214 return; 2215 2216 sx_xunlock(&sc->sc_lock); 2217 g_topology_lock(); 2218 cp = g_new_consumer(sc->sc_sync.ds_geom); 2219 error = g_attach(cp, sc->sc_provider); 2220 KASSERT(error == 0, 2221 ("Cannot attach to %s (error=%d).", sc->sc_name, error)); 2222 error = g_access(cp, 1, 0, 0); 2223 KASSERT(error == 0, ("Cannot open %s (error=%d).", sc->sc_name, error)); 2224 g_topology_unlock(); 2225 sx_xlock(&sc->sc_lock); 2226 2227 G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s.", sc->sc_name, 2228 g_raid3_get_diskname(disk)); 2229 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) == 0) 2230 disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY; 2231 KASSERT(disk->d_sync.ds_consumer == NULL, 2232 ("Sync consumer already exists (device=%s, disk=%s).", 2233 sc->sc_name, g_raid3_get_diskname(disk))); 2234 2235 disk->d_sync.ds_consumer = cp; 2236 disk->d_sync.ds_consumer->private = disk; 2237 disk->d_sync.ds_consumer->index = 0; 2238 sc->sc_syncdisk = disk; 2239 2240 /* 2241 * Allocate memory for synchronization bios and initialize them. 2242 */ 2243 disk->d_sync.ds_bios = malloc(sizeof(struct bio *) * g_raid3_syncreqs, 2244 M_RAID3, M_WAITOK); 2245 for (n = 0; n < g_raid3_syncreqs; n++) { 2246 bp = g_alloc_bio(); 2247 disk->d_sync.ds_bios[n] = bp; 2248 bp->bio_parent = NULL; 2249 bp->bio_cmd = BIO_READ; 2250 bp->bio_data = malloc(MAXPHYS, M_RAID3, M_WAITOK); 2251 bp->bio_cflags = 0; 2252 bp->bio_offset = disk->d_sync.ds_offset * (sc->sc_ndisks - 1); 2253 bp->bio_length = MIN(MAXPHYS, sc->sc_mediasize - bp->bio_offset); 2254 disk->d_sync.ds_offset += bp->bio_length / (sc->sc_ndisks - 1); 2255 bp->bio_done = g_raid3_sync_done; 2256 bp->bio_from = disk->d_sync.ds_consumer; 2257 bp->bio_to = sc->sc_provider; 2258 bp->bio_caller1 = (void *)(uintptr_t)n; 2259 } 2260 2261 /* Set the number of in-flight synchronization requests. */ 2262 disk->d_sync.ds_inflight = g_raid3_syncreqs; 2263 2264 /* 2265 * Fire off first synchronization requests. 2266 */ 2267 for (n = 0; n < g_raid3_syncreqs; n++) { 2268 bp = disk->d_sync.ds_bios[n]; 2269 G_RAID3_LOGREQ(3, bp, "Sending synchronization request."); 2270 disk->d_sync.ds_consumer->index++; 2271 /* 2272 * Delay the request if it is colliding with a regular request. 2273 */ 2274 if (g_raid3_regular_collision(sc, bp)) 2275 g_raid3_sync_delay(sc, bp); 2276 else 2277 g_io_request(bp, disk->d_sync.ds_consumer); 2278 } 2279 } 2280 2281 /* 2282 * Stop synchronization process. 2283 * type: 0 - synchronization finished 2284 * 1 - synchronization stopped 2285 */ 2286 static void 2287 g_raid3_sync_stop(struct g_raid3_softc *sc, int type) 2288 { 2289 struct g_raid3_disk *disk; 2290 struct g_consumer *cp; 2291 2292 g_topology_assert_not(); 2293 sx_assert(&sc->sc_lock, SX_LOCKED); 2294 2295 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED, 2296 ("Device not in DEGRADED state (%s, %u).", sc->sc_name, 2297 sc->sc_state)); 2298 disk = sc->sc_syncdisk; 2299 sc->sc_syncdisk = NULL; 2300 KASSERT(disk != NULL, ("No disk was synchronized (%s).", sc->sc_name)); 2301 KASSERT(disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING, 2302 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2303 g_raid3_disk_state2str(disk->d_state))); 2304 if (disk->d_sync.ds_consumer == NULL) 2305 return; 2306 2307 if (type == 0) { 2308 G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s finished.", 2309 sc->sc_name, g_raid3_get_diskname(disk)); 2310 } else /* if (type == 1) */ { 2311 G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s stopped.", 2312 sc->sc_name, g_raid3_get_diskname(disk)); 2313 } 2314 free(disk->d_sync.ds_bios, M_RAID3); 2315 disk->d_sync.ds_bios = NULL; 2316 cp = disk->d_sync.ds_consumer; 2317 disk->d_sync.ds_consumer = NULL; 2318 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2319 sx_xunlock(&sc->sc_lock); /* Avoid recursion on sc_lock. */ 2320 g_topology_lock(); 2321 g_raid3_kill_consumer(sc, cp); 2322 g_topology_unlock(); 2323 sx_xlock(&sc->sc_lock); 2324 } 2325 2326 static void 2327 g_raid3_launch_provider(struct g_raid3_softc *sc) 2328 { 2329 struct g_provider *pp; 2330 struct g_raid3_disk *disk; 2331 int n; 2332 2333 sx_assert(&sc->sc_lock, SX_LOCKED); 2334 2335 g_topology_lock(); 2336 pp = g_new_providerf(sc->sc_geom, "raid3/%s", sc->sc_name); 2337 pp->mediasize = sc->sc_mediasize; 2338 pp->sectorsize = sc->sc_sectorsize; 2339 pp->stripesize = 0; 2340 pp->stripeoffset = 0; 2341 for (n = 0; n < sc->sc_ndisks; n++) { 2342 disk = &sc->sc_disks[n]; 2343 if (disk->d_consumer && disk->d_consumer->provider && 2344 disk->d_consumer->provider->stripesize > pp->stripesize) { 2345 pp->stripesize = disk->d_consumer->provider->stripesize; 2346 pp->stripeoffset = disk->d_consumer->provider->stripeoffset; 2347 } 2348 } 2349 pp->stripesize *= sc->sc_ndisks - 1; 2350 pp->stripeoffset *= sc->sc_ndisks - 1; 2351 sc->sc_provider = pp; 2352 g_error_provider(pp, 0); 2353 g_topology_unlock(); 2354 G_RAID3_DEBUG(0, "Device %s launched (%u/%u).", pp->name, 2355 g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE), sc->sc_ndisks); 2356 2357 if (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED) 2358 g_raid3_sync_start(sc); 2359 } 2360 2361 static void 2362 g_raid3_destroy_provider(struct g_raid3_softc *sc) 2363 { 2364 struct bio *bp; 2365 2366 g_topology_assert_not(); 2367 KASSERT(sc->sc_provider != NULL, ("NULL provider (device=%s).", 2368 sc->sc_name)); 2369 2370 g_topology_lock(); 2371 g_error_provider(sc->sc_provider, ENXIO); 2372 mtx_lock(&sc->sc_queue_mtx); 2373 while ((bp = bioq_first(&sc->sc_queue)) != NULL) { 2374 bioq_remove(&sc->sc_queue, bp); 2375 g_io_deliver(bp, ENXIO); 2376 } 2377 mtx_unlock(&sc->sc_queue_mtx); 2378 G_RAID3_DEBUG(0, "Device %s: provider %s destroyed.", sc->sc_name, 2379 sc->sc_provider->name); 2380 sc->sc_provider->flags |= G_PF_WITHER; 2381 g_orphan_provider(sc->sc_provider, ENXIO); 2382 g_topology_unlock(); 2383 sc->sc_provider = NULL; 2384 if (sc->sc_syncdisk != NULL) 2385 g_raid3_sync_stop(sc, 1); 2386 } 2387 2388 static void 2389 g_raid3_go(void *arg) 2390 { 2391 struct g_raid3_softc *sc; 2392 2393 sc = arg; 2394 G_RAID3_DEBUG(0, "Force device %s start due to timeout.", sc->sc_name); 2395 g_raid3_event_send(sc, 0, 2396 G_RAID3_EVENT_DONTWAIT | G_RAID3_EVENT_DEVICE); 2397 } 2398 2399 static u_int 2400 g_raid3_determine_state(struct g_raid3_disk *disk) 2401 { 2402 struct g_raid3_softc *sc; 2403 u_int state; 2404 2405 sc = disk->d_softc; 2406 if (sc->sc_syncid == disk->d_sync.ds_syncid) { 2407 if ((disk->d_flags & 2408 G_RAID3_DISK_FLAG_SYNCHRONIZING) == 0) { 2409 /* Disk does not need synchronization. */ 2410 state = G_RAID3_DISK_STATE_ACTIVE; 2411 } else { 2412 if ((sc->sc_flags & 2413 G_RAID3_DEVICE_FLAG_NOAUTOSYNC) == 0 || 2414 (disk->d_flags & 2415 G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) { 2416 /* 2417 * We can start synchronization from 2418 * the stored offset. 2419 */ 2420 state = G_RAID3_DISK_STATE_SYNCHRONIZING; 2421 } else { 2422 state = G_RAID3_DISK_STATE_STALE; 2423 } 2424 } 2425 } else if (disk->d_sync.ds_syncid < sc->sc_syncid) { 2426 /* 2427 * Reset all synchronization data for this disk, 2428 * because if it even was synchronized, it was 2429 * synchronized to disks with different syncid. 2430 */ 2431 disk->d_flags |= G_RAID3_DISK_FLAG_SYNCHRONIZING; 2432 disk->d_sync.ds_offset = 0; 2433 disk->d_sync.ds_offset_done = 0; 2434 disk->d_sync.ds_syncid = sc->sc_syncid; 2435 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) == 0 || 2436 (disk->d_flags & G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) { 2437 state = G_RAID3_DISK_STATE_SYNCHRONIZING; 2438 } else { 2439 state = G_RAID3_DISK_STATE_STALE; 2440 } 2441 } else /* if (sc->sc_syncid < disk->d_sync.ds_syncid) */ { 2442 /* 2443 * Not good, NOT GOOD! 2444 * It means that device was started on stale disks 2445 * and more fresh disk just arrive. 2446 * If there were writes, device is broken, sorry. 2447 * I think the best choice here is don't touch 2448 * this disk and inform the user loudly. 2449 */ 2450 G_RAID3_DEBUG(0, "Device %s was started before the freshest " 2451 "disk (%s) arrives!! It will not be connected to the " 2452 "running device.", sc->sc_name, 2453 g_raid3_get_diskname(disk)); 2454 g_raid3_destroy_disk(disk); 2455 state = G_RAID3_DISK_STATE_NONE; 2456 /* Return immediately, because disk was destroyed. */ 2457 return (state); 2458 } 2459 G_RAID3_DEBUG(3, "State for %s disk: %s.", 2460 g_raid3_get_diskname(disk), g_raid3_disk_state2str(state)); 2461 return (state); 2462 } 2463 2464 /* 2465 * Update device state. 2466 */ 2467 static void 2468 g_raid3_update_device(struct g_raid3_softc *sc, boolean_t force) 2469 { 2470 struct g_raid3_disk *disk; 2471 u_int state; 2472 2473 sx_assert(&sc->sc_lock, SX_XLOCKED); 2474 2475 switch (sc->sc_state) { 2476 case G_RAID3_DEVICE_STATE_STARTING: 2477 { 2478 u_int n, ndirty, ndisks, genid, syncid; 2479 2480 KASSERT(sc->sc_provider == NULL, 2481 ("Non-NULL provider in STARTING state (%s).", sc->sc_name)); 2482 /* 2483 * Are we ready? We are, if all disks are connected or 2484 * one disk is missing and 'force' is true. 2485 */ 2486 if (g_raid3_ndisks(sc, -1) + force == sc->sc_ndisks) { 2487 if (!force) 2488 callout_drain(&sc->sc_callout); 2489 } else { 2490 if (force) { 2491 /* 2492 * Timeout expired, so destroy device. 2493 */ 2494 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 2495 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", 2496 __LINE__, sc->sc_rootmount); 2497 root_mount_rel(sc->sc_rootmount); 2498 sc->sc_rootmount = NULL; 2499 } 2500 return; 2501 } 2502 2503 /* 2504 * Find the biggest genid. 2505 */ 2506 genid = 0; 2507 for (n = 0; n < sc->sc_ndisks; n++) { 2508 disk = &sc->sc_disks[n]; 2509 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2510 continue; 2511 if (disk->d_genid > genid) 2512 genid = disk->d_genid; 2513 } 2514 sc->sc_genid = genid; 2515 /* 2516 * Remove all disks without the biggest genid. 2517 */ 2518 for (n = 0; n < sc->sc_ndisks; n++) { 2519 disk = &sc->sc_disks[n]; 2520 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2521 continue; 2522 if (disk->d_genid < genid) { 2523 G_RAID3_DEBUG(0, 2524 "Component %s (device %s) broken, skipping.", 2525 g_raid3_get_diskname(disk), sc->sc_name); 2526 g_raid3_destroy_disk(disk); 2527 } 2528 } 2529 2530 /* 2531 * There must be at least 'sc->sc_ndisks - 1' components 2532 * with the same syncid and without SYNCHRONIZING flag. 2533 */ 2534 2535 /* 2536 * Find the biggest syncid, number of valid components and 2537 * number of dirty components. 2538 */ 2539 ndirty = ndisks = syncid = 0; 2540 for (n = 0; n < sc->sc_ndisks; n++) { 2541 disk = &sc->sc_disks[n]; 2542 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2543 continue; 2544 if ((disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) != 0) 2545 ndirty++; 2546 if (disk->d_sync.ds_syncid > syncid) { 2547 syncid = disk->d_sync.ds_syncid; 2548 ndisks = 0; 2549 } else if (disk->d_sync.ds_syncid < syncid) { 2550 continue; 2551 } 2552 if ((disk->d_flags & 2553 G_RAID3_DISK_FLAG_SYNCHRONIZING) != 0) { 2554 continue; 2555 } 2556 ndisks++; 2557 } 2558 /* 2559 * Do we have enough valid components? 2560 */ 2561 if (ndisks + 1 < sc->sc_ndisks) { 2562 G_RAID3_DEBUG(0, 2563 "Device %s is broken, too few valid components.", 2564 sc->sc_name); 2565 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 2566 return; 2567 } 2568 /* 2569 * If there is one DIRTY component and all disks are present, 2570 * mark it for synchronization. If there is more than one DIRTY 2571 * component, mark parity component for synchronization. 2572 */ 2573 if (ndisks == sc->sc_ndisks && ndirty == 1) { 2574 for (n = 0; n < sc->sc_ndisks; n++) { 2575 disk = &sc->sc_disks[n]; 2576 if ((disk->d_flags & 2577 G_RAID3_DISK_FLAG_DIRTY) == 0) { 2578 continue; 2579 } 2580 disk->d_flags |= 2581 G_RAID3_DISK_FLAG_SYNCHRONIZING; 2582 } 2583 } else if (ndisks == sc->sc_ndisks && ndirty > 1) { 2584 disk = &sc->sc_disks[sc->sc_ndisks - 1]; 2585 disk->d_flags |= G_RAID3_DISK_FLAG_SYNCHRONIZING; 2586 } 2587 2588 sc->sc_syncid = syncid; 2589 if (force) { 2590 /* Remember to bump syncid on first write. */ 2591 sc->sc_bump_id |= G_RAID3_BUMP_SYNCID; 2592 } 2593 if (ndisks == sc->sc_ndisks) 2594 state = G_RAID3_DEVICE_STATE_COMPLETE; 2595 else /* if (ndisks == sc->sc_ndisks - 1) */ 2596 state = G_RAID3_DEVICE_STATE_DEGRADED; 2597 G_RAID3_DEBUG(1, "Device %s state changed from %s to %s.", 2598 sc->sc_name, g_raid3_device_state2str(sc->sc_state), 2599 g_raid3_device_state2str(state)); 2600 sc->sc_state = state; 2601 for (n = 0; n < sc->sc_ndisks; n++) { 2602 disk = &sc->sc_disks[n]; 2603 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2604 continue; 2605 state = g_raid3_determine_state(disk); 2606 g_raid3_event_send(disk, state, G_RAID3_EVENT_DONTWAIT); 2607 if (state == G_RAID3_DISK_STATE_STALE) 2608 sc->sc_bump_id |= G_RAID3_BUMP_SYNCID; 2609 } 2610 break; 2611 } 2612 case G_RAID3_DEVICE_STATE_DEGRADED: 2613 /* 2614 * Genid need to be bumped immediately, so do it here. 2615 */ 2616 if ((sc->sc_bump_id & G_RAID3_BUMP_GENID) != 0) { 2617 sc->sc_bump_id &= ~G_RAID3_BUMP_GENID; 2618 g_raid3_bump_genid(sc); 2619 } 2620 2621 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NEW) > 0) 2622 return; 2623 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) < 2624 sc->sc_ndisks - 1) { 2625 if (sc->sc_provider != NULL) 2626 g_raid3_destroy_provider(sc); 2627 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 2628 return; 2629 } 2630 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) == 2631 sc->sc_ndisks) { 2632 state = G_RAID3_DEVICE_STATE_COMPLETE; 2633 G_RAID3_DEBUG(1, 2634 "Device %s state changed from %s to %s.", 2635 sc->sc_name, g_raid3_device_state2str(sc->sc_state), 2636 g_raid3_device_state2str(state)); 2637 sc->sc_state = state; 2638 } 2639 if (sc->sc_provider == NULL) 2640 g_raid3_launch_provider(sc); 2641 if (sc->sc_rootmount != NULL) { 2642 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__, 2643 sc->sc_rootmount); 2644 root_mount_rel(sc->sc_rootmount); 2645 sc->sc_rootmount = NULL; 2646 } 2647 break; 2648 case G_RAID3_DEVICE_STATE_COMPLETE: 2649 /* 2650 * Genid need to be bumped immediately, so do it here. 2651 */ 2652 if ((sc->sc_bump_id & G_RAID3_BUMP_GENID) != 0) { 2653 sc->sc_bump_id &= ~G_RAID3_BUMP_GENID; 2654 g_raid3_bump_genid(sc); 2655 } 2656 2657 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NEW) > 0) 2658 return; 2659 KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) >= 2660 sc->sc_ndisks - 1, 2661 ("Too few ACTIVE components in COMPLETE state (device %s).", 2662 sc->sc_name)); 2663 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) == 2664 sc->sc_ndisks - 1) { 2665 state = G_RAID3_DEVICE_STATE_DEGRADED; 2666 G_RAID3_DEBUG(1, 2667 "Device %s state changed from %s to %s.", 2668 sc->sc_name, g_raid3_device_state2str(sc->sc_state), 2669 g_raid3_device_state2str(state)); 2670 sc->sc_state = state; 2671 } 2672 if (sc->sc_provider == NULL) 2673 g_raid3_launch_provider(sc); 2674 if (sc->sc_rootmount != NULL) { 2675 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__, 2676 sc->sc_rootmount); 2677 root_mount_rel(sc->sc_rootmount); 2678 sc->sc_rootmount = NULL; 2679 } 2680 break; 2681 default: 2682 KASSERT(1 == 0, ("Wrong device state (%s, %s).", sc->sc_name, 2683 g_raid3_device_state2str(sc->sc_state))); 2684 break; 2685 } 2686 } 2687 2688 /* 2689 * Update disk state and device state if needed. 2690 */ 2691 #define DISK_STATE_CHANGED() G_RAID3_DEBUG(1, \ 2692 "Disk %s state changed from %s to %s (device %s).", \ 2693 g_raid3_get_diskname(disk), \ 2694 g_raid3_disk_state2str(disk->d_state), \ 2695 g_raid3_disk_state2str(state), sc->sc_name) 2696 static int 2697 g_raid3_update_disk(struct g_raid3_disk *disk, u_int state) 2698 { 2699 struct g_raid3_softc *sc; 2700 2701 sc = disk->d_softc; 2702 sx_assert(&sc->sc_lock, SX_XLOCKED); 2703 2704 again: 2705 G_RAID3_DEBUG(3, "Changing disk %s state from %s to %s.", 2706 g_raid3_get_diskname(disk), g_raid3_disk_state2str(disk->d_state), 2707 g_raid3_disk_state2str(state)); 2708 switch (state) { 2709 case G_RAID3_DISK_STATE_NEW: 2710 /* 2711 * Possible scenarios: 2712 * 1. New disk arrive. 2713 */ 2714 /* Previous state should be NONE. */ 2715 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NONE, 2716 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2717 g_raid3_disk_state2str(disk->d_state))); 2718 DISK_STATE_CHANGED(); 2719 2720 disk->d_state = state; 2721 G_RAID3_DEBUG(1, "Device %s: provider %s detected.", 2722 sc->sc_name, g_raid3_get_diskname(disk)); 2723 if (sc->sc_state == G_RAID3_DEVICE_STATE_STARTING) 2724 break; 2725 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2726 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2727 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2728 g_raid3_device_state2str(sc->sc_state), 2729 g_raid3_get_diskname(disk), 2730 g_raid3_disk_state2str(disk->d_state))); 2731 state = g_raid3_determine_state(disk); 2732 if (state != G_RAID3_DISK_STATE_NONE) 2733 goto again; 2734 break; 2735 case G_RAID3_DISK_STATE_ACTIVE: 2736 /* 2737 * Possible scenarios: 2738 * 1. New disk does not need synchronization. 2739 * 2. Synchronization process finished successfully. 2740 */ 2741 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2742 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2743 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2744 g_raid3_device_state2str(sc->sc_state), 2745 g_raid3_get_diskname(disk), 2746 g_raid3_disk_state2str(disk->d_state))); 2747 /* Previous state should be NEW or SYNCHRONIZING. */ 2748 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW || 2749 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING, 2750 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2751 g_raid3_disk_state2str(disk->d_state))); 2752 DISK_STATE_CHANGED(); 2753 2754 if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 2755 disk->d_flags &= ~G_RAID3_DISK_FLAG_SYNCHRONIZING; 2756 disk->d_flags &= ~G_RAID3_DISK_FLAG_FORCE_SYNC; 2757 g_raid3_sync_stop(sc, 0); 2758 } 2759 disk->d_state = state; 2760 disk->d_sync.ds_offset = 0; 2761 disk->d_sync.ds_offset_done = 0; 2762 g_raid3_update_idle(sc, disk); 2763 g_raid3_update_metadata(disk); 2764 G_RAID3_DEBUG(1, "Device %s: provider %s activated.", 2765 sc->sc_name, g_raid3_get_diskname(disk)); 2766 break; 2767 case G_RAID3_DISK_STATE_STALE: 2768 /* 2769 * Possible scenarios: 2770 * 1. Stale disk was connected. 2771 */ 2772 /* Previous state should be NEW. */ 2773 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW, 2774 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2775 g_raid3_disk_state2str(disk->d_state))); 2776 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2777 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2778 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2779 g_raid3_device_state2str(sc->sc_state), 2780 g_raid3_get_diskname(disk), 2781 g_raid3_disk_state2str(disk->d_state))); 2782 /* 2783 * STALE state is only possible if device is marked 2784 * NOAUTOSYNC. 2785 */ 2786 KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) != 0, 2787 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2788 g_raid3_device_state2str(sc->sc_state), 2789 g_raid3_get_diskname(disk), 2790 g_raid3_disk_state2str(disk->d_state))); 2791 DISK_STATE_CHANGED(); 2792 2793 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2794 disk->d_state = state; 2795 g_raid3_update_metadata(disk); 2796 G_RAID3_DEBUG(0, "Device %s: provider %s is stale.", 2797 sc->sc_name, g_raid3_get_diskname(disk)); 2798 break; 2799 case G_RAID3_DISK_STATE_SYNCHRONIZING: 2800 /* 2801 * Possible scenarios: 2802 * 1. Disk which needs synchronization was connected. 2803 */ 2804 /* Previous state should be NEW. */ 2805 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW, 2806 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2807 g_raid3_disk_state2str(disk->d_state))); 2808 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2809 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2810 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2811 g_raid3_device_state2str(sc->sc_state), 2812 g_raid3_get_diskname(disk), 2813 g_raid3_disk_state2str(disk->d_state))); 2814 DISK_STATE_CHANGED(); 2815 2816 if (disk->d_state == G_RAID3_DISK_STATE_NEW) 2817 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2818 disk->d_state = state; 2819 if (sc->sc_provider != NULL) { 2820 g_raid3_sync_start(sc); 2821 g_raid3_update_metadata(disk); 2822 } 2823 break; 2824 case G_RAID3_DISK_STATE_DISCONNECTED: 2825 /* 2826 * Possible scenarios: 2827 * 1. Device wasn't running yet, but disk disappear. 2828 * 2. Disk was active and disapppear. 2829 * 3. Disk disappear during synchronization process. 2830 */ 2831 if (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2832 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 2833 /* 2834 * Previous state should be ACTIVE, STALE or 2835 * SYNCHRONIZING. 2836 */ 2837 KASSERT(disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 2838 disk->d_state == G_RAID3_DISK_STATE_STALE || 2839 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING, 2840 ("Wrong disk state (%s, %s).", 2841 g_raid3_get_diskname(disk), 2842 g_raid3_disk_state2str(disk->d_state))); 2843 } else if (sc->sc_state == G_RAID3_DEVICE_STATE_STARTING) { 2844 /* Previous state should be NEW. */ 2845 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW, 2846 ("Wrong disk state (%s, %s).", 2847 g_raid3_get_diskname(disk), 2848 g_raid3_disk_state2str(disk->d_state))); 2849 /* 2850 * Reset bumping syncid if disk disappeared in STARTING 2851 * state. 2852 */ 2853 if ((sc->sc_bump_id & G_RAID3_BUMP_SYNCID) != 0) 2854 sc->sc_bump_id &= ~G_RAID3_BUMP_SYNCID; 2855 #ifdef INVARIANTS 2856 } else { 2857 KASSERT(1 == 0, ("Wrong device state (%s, %s, %s, %s).", 2858 sc->sc_name, 2859 g_raid3_device_state2str(sc->sc_state), 2860 g_raid3_get_diskname(disk), 2861 g_raid3_disk_state2str(disk->d_state))); 2862 #endif 2863 } 2864 DISK_STATE_CHANGED(); 2865 G_RAID3_DEBUG(0, "Device %s: provider %s disconnected.", 2866 sc->sc_name, g_raid3_get_diskname(disk)); 2867 2868 g_raid3_destroy_disk(disk); 2869 break; 2870 default: 2871 KASSERT(1 == 0, ("Unknown state (%u).", state)); 2872 break; 2873 } 2874 return (0); 2875 } 2876 #undef DISK_STATE_CHANGED 2877 2878 int 2879 g_raid3_read_metadata(struct g_consumer *cp, struct g_raid3_metadata *md) 2880 { 2881 struct g_provider *pp; 2882 u_char *buf; 2883 int error; 2884 2885 g_topology_assert(); 2886 2887 error = g_access(cp, 1, 0, 0); 2888 if (error != 0) 2889 return (error); 2890 pp = cp->provider; 2891 g_topology_unlock(); 2892 /* Metadata are stored on last sector. */ 2893 buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, 2894 &error); 2895 g_topology_lock(); 2896 g_access(cp, -1, 0, 0); 2897 if (buf == NULL) { 2898 G_RAID3_DEBUG(1, "Cannot read metadata from %s (error=%d).", 2899 cp->provider->name, error); 2900 return (error); 2901 } 2902 2903 /* Decode metadata. */ 2904 error = raid3_metadata_decode(buf, md); 2905 g_free(buf); 2906 if (strcmp(md->md_magic, G_RAID3_MAGIC) != 0) 2907 return (EINVAL); 2908 if (md->md_version > G_RAID3_VERSION) { 2909 G_RAID3_DEBUG(0, 2910 "Kernel module is too old to handle metadata from %s.", 2911 cp->provider->name); 2912 return (EINVAL); 2913 } 2914 if (error != 0) { 2915 G_RAID3_DEBUG(1, "MD5 metadata hash mismatch for provider %s.", 2916 cp->provider->name); 2917 return (error); 2918 } 2919 if (md->md_sectorsize > MAXPHYS) { 2920 G_RAID3_DEBUG(0, "The blocksize is too big."); 2921 return (EINVAL); 2922 } 2923 2924 return (0); 2925 } 2926 2927 static int 2928 g_raid3_check_metadata(struct g_raid3_softc *sc, struct g_provider *pp, 2929 struct g_raid3_metadata *md) 2930 { 2931 2932 if (md->md_no >= sc->sc_ndisks) { 2933 G_RAID3_DEBUG(1, "Invalid disk %s number (no=%u), skipping.", 2934 pp->name, md->md_no); 2935 return (EINVAL); 2936 } 2937 if (sc->sc_disks[md->md_no].d_state != G_RAID3_DISK_STATE_NODISK) { 2938 G_RAID3_DEBUG(1, "Disk %s (no=%u) already exists, skipping.", 2939 pp->name, md->md_no); 2940 return (EEXIST); 2941 } 2942 if (md->md_all != sc->sc_ndisks) { 2943 G_RAID3_DEBUG(1, 2944 "Invalid '%s' field on disk %s (device %s), skipping.", 2945 "md_all", pp->name, sc->sc_name); 2946 return (EINVAL); 2947 } 2948 if ((md->md_mediasize % md->md_sectorsize) != 0) { 2949 G_RAID3_DEBUG(1, "Invalid metadata (mediasize %% sectorsize != " 2950 "0) on disk %s (device %s), skipping.", pp->name, 2951 sc->sc_name); 2952 return (EINVAL); 2953 } 2954 if (md->md_mediasize != sc->sc_mediasize) { 2955 G_RAID3_DEBUG(1, 2956 "Invalid '%s' field on disk %s (device %s), skipping.", 2957 "md_mediasize", pp->name, sc->sc_name); 2958 return (EINVAL); 2959 } 2960 if ((md->md_mediasize % (sc->sc_ndisks - 1)) != 0) { 2961 G_RAID3_DEBUG(1, 2962 "Invalid '%s' field on disk %s (device %s), skipping.", 2963 "md_mediasize", pp->name, sc->sc_name); 2964 return (EINVAL); 2965 } 2966 if ((sc->sc_mediasize / (sc->sc_ndisks - 1)) > pp->mediasize) { 2967 G_RAID3_DEBUG(1, 2968 "Invalid size of disk %s (device %s), skipping.", pp->name, 2969 sc->sc_name); 2970 return (EINVAL); 2971 } 2972 if ((md->md_sectorsize / pp->sectorsize) < sc->sc_ndisks - 1) { 2973 G_RAID3_DEBUG(1, 2974 "Invalid '%s' field on disk %s (device %s), skipping.", 2975 "md_sectorsize", pp->name, sc->sc_name); 2976 return (EINVAL); 2977 } 2978 if (md->md_sectorsize != sc->sc_sectorsize) { 2979 G_RAID3_DEBUG(1, 2980 "Invalid '%s' field on disk %s (device %s), skipping.", 2981 "md_sectorsize", pp->name, sc->sc_name); 2982 return (EINVAL); 2983 } 2984 if ((sc->sc_sectorsize % pp->sectorsize) != 0) { 2985 G_RAID3_DEBUG(1, 2986 "Invalid sector size of disk %s (device %s), skipping.", 2987 pp->name, sc->sc_name); 2988 return (EINVAL); 2989 } 2990 if ((md->md_mflags & ~G_RAID3_DEVICE_FLAG_MASK) != 0) { 2991 G_RAID3_DEBUG(1, 2992 "Invalid device flags on disk %s (device %s), skipping.", 2993 pp->name, sc->sc_name); 2994 return (EINVAL); 2995 } 2996 if ((md->md_mflags & G_RAID3_DEVICE_FLAG_VERIFY) != 0 && 2997 (md->md_mflags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0) { 2998 /* 2999 * VERIFY and ROUND-ROBIN options are mutally exclusive. 3000 */ 3001 G_RAID3_DEBUG(1, "Both VERIFY and ROUND-ROBIN flags exist on " 3002 "disk %s (device %s), skipping.", pp->name, sc->sc_name); 3003 return (EINVAL); 3004 } 3005 if ((md->md_dflags & ~G_RAID3_DISK_FLAG_MASK) != 0) { 3006 G_RAID3_DEBUG(1, 3007 "Invalid disk flags on disk %s (device %s), skipping.", 3008 pp->name, sc->sc_name); 3009 return (EINVAL); 3010 } 3011 return (0); 3012 } 3013 3014 int 3015 g_raid3_add_disk(struct g_raid3_softc *sc, struct g_provider *pp, 3016 struct g_raid3_metadata *md) 3017 { 3018 struct g_raid3_disk *disk; 3019 int error; 3020 3021 g_topology_assert_not(); 3022 G_RAID3_DEBUG(2, "Adding disk %s.", pp->name); 3023 3024 error = g_raid3_check_metadata(sc, pp, md); 3025 if (error != 0) 3026 return (error); 3027 if (sc->sc_state != G_RAID3_DEVICE_STATE_STARTING && 3028 md->md_genid < sc->sc_genid) { 3029 G_RAID3_DEBUG(0, "Component %s (device %s) broken, skipping.", 3030 pp->name, sc->sc_name); 3031 return (EINVAL); 3032 } 3033 disk = g_raid3_init_disk(sc, pp, md, &error); 3034 if (disk == NULL) 3035 return (error); 3036 error = g_raid3_event_send(disk, G_RAID3_DISK_STATE_NEW, 3037 G_RAID3_EVENT_WAIT); 3038 if (error != 0) 3039 return (error); 3040 if (md->md_version < G_RAID3_VERSION) { 3041 G_RAID3_DEBUG(0, "Upgrading metadata on %s (v%d->v%d).", 3042 pp->name, md->md_version, G_RAID3_VERSION); 3043 g_raid3_update_metadata(disk); 3044 } 3045 return (0); 3046 } 3047 3048 static void 3049 g_raid3_destroy_delayed(void *arg, int flag) 3050 { 3051 struct g_raid3_softc *sc; 3052 int error; 3053 3054 if (flag == EV_CANCEL) { 3055 G_RAID3_DEBUG(1, "Destroying canceled."); 3056 return; 3057 } 3058 sc = arg; 3059 g_topology_unlock(); 3060 sx_xlock(&sc->sc_lock); 3061 KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) == 0, 3062 ("DESTROY flag set on %s.", sc->sc_name)); 3063 KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROYING) != 0, 3064 ("DESTROYING flag not set on %s.", sc->sc_name)); 3065 G_RAID3_DEBUG(0, "Destroying %s (delayed).", sc->sc_name); 3066 error = g_raid3_destroy(sc, G_RAID3_DESTROY_SOFT); 3067 if (error != 0) { 3068 G_RAID3_DEBUG(0, "Cannot destroy %s.", sc->sc_name); 3069 sx_xunlock(&sc->sc_lock); 3070 } 3071 g_topology_lock(); 3072 } 3073 3074 static int 3075 g_raid3_access(struct g_provider *pp, int acr, int acw, int ace) 3076 { 3077 struct g_raid3_softc *sc; 3078 int dcr, dcw, dce, error = 0; 3079 3080 g_topology_assert(); 3081 G_RAID3_DEBUG(2, "Access request for %s: r%dw%de%d.", pp->name, acr, 3082 acw, ace); 3083 3084 sc = pp->geom->softc; 3085 if (sc == NULL && acr <= 0 && acw <= 0 && ace <= 0) 3086 return (0); 3087 KASSERT(sc != NULL, ("NULL softc (provider=%s).", pp->name)); 3088 3089 dcr = pp->acr + acr; 3090 dcw = pp->acw + acw; 3091 dce = pp->ace + ace; 3092 3093 g_topology_unlock(); 3094 sx_xlock(&sc->sc_lock); 3095 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0 || 3096 g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) < sc->sc_ndisks - 1) { 3097 if (acr > 0 || acw > 0 || ace > 0) 3098 error = ENXIO; 3099 goto end; 3100 } 3101 if (dcw == 0 && !sc->sc_idle) 3102 g_raid3_idle(sc, dcw); 3103 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROYING) != 0) { 3104 if (acr > 0 || acw > 0 || ace > 0) { 3105 error = ENXIO; 3106 goto end; 3107 } 3108 if (dcr == 0 && dcw == 0 && dce == 0) { 3109 g_post_event(g_raid3_destroy_delayed, sc, M_WAITOK, 3110 sc, NULL); 3111 } 3112 } 3113 end: 3114 sx_xunlock(&sc->sc_lock); 3115 g_topology_lock(); 3116 return (error); 3117 } 3118 3119 static struct g_geom * 3120 g_raid3_create(struct g_class *mp, const struct g_raid3_metadata *md) 3121 { 3122 struct g_raid3_softc *sc; 3123 struct g_geom *gp; 3124 int error, timeout; 3125 u_int n; 3126 3127 g_topology_assert(); 3128 G_RAID3_DEBUG(1, "Creating device %s (id=%u).", md->md_name, md->md_id); 3129 3130 /* One disk is minimum. */ 3131 if (md->md_all < 1) 3132 return (NULL); 3133 /* 3134 * Action geom. 3135 */ 3136 gp = g_new_geomf(mp, "%s", md->md_name); 3137 sc = malloc(sizeof(*sc), M_RAID3, M_WAITOK | M_ZERO); 3138 sc->sc_disks = malloc(sizeof(struct g_raid3_disk) * md->md_all, M_RAID3, 3139 M_WAITOK | M_ZERO); 3140 gp->start = g_raid3_start; 3141 gp->orphan = g_raid3_orphan; 3142 gp->access = g_raid3_access; 3143 gp->dumpconf = g_raid3_dumpconf; 3144 3145 sc->sc_id = md->md_id; 3146 sc->sc_mediasize = md->md_mediasize; 3147 sc->sc_sectorsize = md->md_sectorsize; 3148 sc->sc_ndisks = md->md_all; 3149 sc->sc_round_robin = 0; 3150 sc->sc_flags = md->md_mflags; 3151 sc->sc_bump_id = 0; 3152 sc->sc_idle = 1; 3153 sc->sc_last_write = time_uptime; 3154 sc->sc_writes = 0; 3155 for (n = 0; n < sc->sc_ndisks; n++) { 3156 sc->sc_disks[n].d_softc = sc; 3157 sc->sc_disks[n].d_no = n; 3158 sc->sc_disks[n].d_state = G_RAID3_DISK_STATE_NODISK; 3159 } 3160 sx_init(&sc->sc_lock, "graid3:lock"); 3161 bioq_init(&sc->sc_queue); 3162 mtx_init(&sc->sc_queue_mtx, "graid3:queue", NULL, MTX_DEF); 3163 bioq_init(&sc->sc_regular_delayed); 3164 bioq_init(&sc->sc_inflight); 3165 bioq_init(&sc->sc_sync_delayed); 3166 TAILQ_INIT(&sc->sc_events); 3167 mtx_init(&sc->sc_events_mtx, "graid3:events", NULL, MTX_DEF); 3168 callout_init(&sc->sc_callout, CALLOUT_MPSAFE); 3169 sc->sc_state = G_RAID3_DEVICE_STATE_STARTING; 3170 gp->softc = sc; 3171 sc->sc_geom = gp; 3172 sc->sc_provider = NULL; 3173 /* 3174 * Synchronization geom. 3175 */ 3176 gp = g_new_geomf(mp, "%s.sync", md->md_name); 3177 gp->softc = sc; 3178 gp->orphan = g_raid3_orphan; 3179 sc->sc_sync.ds_geom = gp; 3180 3181 if (!g_raid3_use_malloc) { 3182 sc->sc_zones[G_RAID3_ZONE_64K].sz_zone = uma_zcreate("gr3:64k", 3183 65536, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL, 3184 UMA_ALIGN_PTR, 0); 3185 sc->sc_zones[G_RAID3_ZONE_64K].sz_inuse = 0; 3186 sc->sc_zones[G_RAID3_ZONE_64K].sz_max = g_raid3_n64k; 3187 sc->sc_zones[G_RAID3_ZONE_64K].sz_requested = 3188 sc->sc_zones[G_RAID3_ZONE_64K].sz_failed = 0; 3189 sc->sc_zones[G_RAID3_ZONE_16K].sz_zone = uma_zcreate("gr3:16k", 3190 16384, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL, 3191 UMA_ALIGN_PTR, 0); 3192 sc->sc_zones[G_RAID3_ZONE_16K].sz_inuse = 0; 3193 sc->sc_zones[G_RAID3_ZONE_16K].sz_max = g_raid3_n16k; 3194 sc->sc_zones[G_RAID3_ZONE_16K].sz_requested = 3195 sc->sc_zones[G_RAID3_ZONE_16K].sz_failed = 0; 3196 sc->sc_zones[G_RAID3_ZONE_4K].sz_zone = uma_zcreate("gr3:4k", 3197 4096, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL, 3198 UMA_ALIGN_PTR, 0); 3199 sc->sc_zones[G_RAID3_ZONE_4K].sz_inuse = 0; 3200 sc->sc_zones[G_RAID3_ZONE_4K].sz_max = g_raid3_n4k; 3201 sc->sc_zones[G_RAID3_ZONE_4K].sz_requested = 3202 sc->sc_zones[G_RAID3_ZONE_4K].sz_failed = 0; 3203 } 3204 3205 error = kproc_create(g_raid3_worker, sc, &sc->sc_worker, 0, 0, 3206 "g_raid3 %s", md->md_name); 3207 if (error != 0) { 3208 G_RAID3_DEBUG(1, "Cannot create kernel thread for %s.", 3209 sc->sc_name); 3210 if (!g_raid3_use_malloc) { 3211 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_64K].sz_zone); 3212 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_16K].sz_zone); 3213 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_4K].sz_zone); 3214 } 3215 g_destroy_geom(sc->sc_sync.ds_geom); 3216 mtx_destroy(&sc->sc_events_mtx); 3217 mtx_destroy(&sc->sc_queue_mtx); 3218 sx_destroy(&sc->sc_lock); 3219 g_destroy_geom(sc->sc_geom); 3220 free(sc->sc_disks, M_RAID3); 3221 free(sc, M_RAID3); 3222 return (NULL); 3223 } 3224 3225 G_RAID3_DEBUG(1, "Device %s created (%u components, id=%u).", 3226 sc->sc_name, sc->sc_ndisks, sc->sc_id); 3227 3228 sc->sc_rootmount = root_mount_hold("GRAID3"); 3229 G_RAID3_DEBUG(1, "root_mount_hold %p", sc->sc_rootmount); 3230 3231 /* 3232 * Run timeout. 3233 */ 3234 timeout = atomic_load_acq_int(&g_raid3_timeout); 3235 callout_reset(&sc->sc_callout, timeout * hz, g_raid3_go, sc); 3236 return (sc->sc_geom); 3237 } 3238 3239 int 3240 g_raid3_destroy(struct g_raid3_softc *sc, int how) 3241 { 3242 struct g_provider *pp; 3243 3244 g_topology_assert_not(); 3245 if (sc == NULL) 3246 return (ENXIO); 3247 sx_assert(&sc->sc_lock, SX_XLOCKED); 3248 3249 pp = sc->sc_provider; 3250 if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { 3251 switch (how) { 3252 case G_RAID3_DESTROY_SOFT: 3253 G_RAID3_DEBUG(1, 3254 "Device %s is still open (r%dw%de%d).", pp->name, 3255 pp->acr, pp->acw, pp->ace); 3256 return (EBUSY); 3257 case G_RAID3_DESTROY_DELAYED: 3258 G_RAID3_DEBUG(1, 3259 "Device %s will be destroyed on last close.", 3260 pp->name); 3261 if (sc->sc_syncdisk != NULL) 3262 g_raid3_sync_stop(sc, 1); 3263 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROYING; 3264 return (EBUSY); 3265 case G_RAID3_DESTROY_HARD: 3266 G_RAID3_DEBUG(1, "Device %s is still open, so it " 3267 "can't be definitely removed.", pp->name); 3268 break; 3269 } 3270 } 3271 3272 g_topology_lock(); 3273 if (sc->sc_geom->softc == NULL) { 3274 g_topology_unlock(); 3275 return (0); 3276 } 3277 sc->sc_geom->softc = NULL; 3278 sc->sc_sync.ds_geom->softc = NULL; 3279 g_topology_unlock(); 3280 3281 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 3282 sc->sc_flags |= G_RAID3_DEVICE_FLAG_WAIT; 3283 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc); 3284 sx_xunlock(&sc->sc_lock); 3285 mtx_lock(&sc->sc_queue_mtx); 3286 wakeup(sc); 3287 wakeup(&sc->sc_queue); 3288 mtx_unlock(&sc->sc_queue_mtx); 3289 G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, &sc->sc_worker); 3290 while (sc->sc_worker != NULL) 3291 tsleep(&sc->sc_worker, PRIBIO, "r3:destroy", hz / 5); 3292 G_RAID3_DEBUG(4, "%s: Woken up %p.", __func__, &sc->sc_worker); 3293 sx_xlock(&sc->sc_lock); 3294 g_raid3_destroy_device(sc); 3295 free(sc->sc_disks, M_RAID3); 3296 free(sc, M_RAID3); 3297 return (0); 3298 } 3299 3300 static void 3301 g_raid3_taste_orphan(struct g_consumer *cp) 3302 { 3303 3304 KASSERT(1 == 0, ("%s called while tasting %s.", __func__, 3305 cp->provider->name)); 3306 } 3307 3308 static struct g_geom * 3309 g_raid3_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) 3310 { 3311 struct g_raid3_metadata md; 3312 struct g_raid3_softc *sc; 3313 struct g_consumer *cp; 3314 struct g_geom *gp; 3315 int error; 3316 3317 g_topology_assert(); 3318 g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); 3319 G_RAID3_DEBUG(2, "Tasting %s.", pp->name); 3320 3321 gp = g_new_geomf(mp, "raid3:taste"); 3322 /* This orphan function should be never called. */ 3323 gp->orphan = g_raid3_taste_orphan; 3324 cp = g_new_consumer(gp); 3325 g_attach(cp, pp); 3326 error = g_raid3_read_metadata(cp, &md); 3327 g_detach(cp); 3328 g_destroy_consumer(cp); 3329 g_destroy_geom(gp); 3330 if (error != 0) 3331 return (NULL); 3332 gp = NULL; 3333 3334 if (md.md_provider[0] != '\0' && 3335 !g_compare_names(md.md_provider, pp->name)) 3336 return (NULL); 3337 if (md.md_provsize != 0 && md.md_provsize != pp->mediasize) 3338 return (NULL); 3339 if (g_raid3_debug >= 2) 3340 raid3_metadata_dump(&md); 3341 3342 /* 3343 * Let's check if device already exists. 3344 */ 3345 sc = NULL; 3346 LIST_FOREACH(gp, &mp->geom, geom) { 3347 sc = gp->softc; 3348 if (sc == NULL) 3349 continue; 3350 if (sc->sc_sync.ds_geom == gp) 3351 continue; 3352 if (strcmp(md.md_name, sc->sc_name) != 0) 3353 continue; 3354 if (md.md_id != sc->sc_id) { 3355 G_RAID3_DEBUG(0, "Device %s already configured.", 3356 sc->sc_name); 3357 return (NULL); 3358 } 3359 break; 3360 } 3361 if (gp == NULL) { 3362 gp = g_raid3_create(mp, &md); 3363 if (gp == NULL) { 3364 G_RAID3_DEBUG(0, "Cannot create device %s.", 3365 md.md_name); 3366 return (NULL); 3367 } 3368 sc = gp->softc; 3369 } 3370 G_RAID3_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); 3371 g_topology_unlock(); 3372 sx_xlock(&sc->sc_lock); 3373 error = g_raid3_add_disk(sc, pp, &md); 3374 if (error != 0) { 3375 G_RAID3_DEBUG(0, "Cannot add disk %s to %s (error=%d).", 3376 pp->name, gp->name, error); 3377 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NODISK) == 3378 sc->sc_ndisks) { 3379 g_cancel_event(sc); 3380 g_raid3_destroy(sc, G_RAID3_DESTROY_HARD); 3381 g_topology_lock(); 3382 return (NULL); 3383 } 3384 gp = NULL; 3385 } 3386 sx_xunlock(&sc->sc_lock); 3387 g_topology_lock(); 3388 return (gp); 3389 } 3390 3391 static int 3392 g_raid3_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused, 3393 struct g_geom *gp) 3394 { 3395 struct g_raid3_softc *sc; 3396 int error; 3397 3398 g_topology_unlock(); 3399 sc = gp->softc; 3400 sx_xlock(&sc->sc_lock); 3401 g_cancel_event(sc); 3402 error = g_raid3_destroy(gp->softc, G_RAID3_DESTROY_SOFT); 3403 if (error != 0) 3404 sx_xunlock(&sc->sc_lock); 3405 g_topology_lock(); 3406 return (error); 3407 } 3408 3409 static void 3410 g_raid3_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, 3411 struct g_consumer *cp, struct g_provider *pp) 3412 { 3413 struct g_raid3_softc *sc; 3414 3415 g_topology_assert(); 3416 3417 sc = gp->softc; 3418 if (sc == NULL) 3419 return; 3420 /* Skip synchronization geom. */ 3421 if (gp == sc->sc_sync.ds_geom) 3422 return; 3423 if (pp != NULL) { 3424 /* Nothing here. */ 3425 } else if (cp != NULL) { 3426 struct g_raid3_disk *disk; 3427 3428 disk = cp->private; 3429 if (disk == NULL) 3430 return; 3431 g_topology_unlock(); 3432 sx_xlock(&sc->sc_lock); 3433 sbuf_printf(sb, "%s<Type>", indent); 3434 if (disk->d_no == sc->sc_ndisks - 1) 3435 sbuf_printf(sb, "PARITY"); 3436 else 3437 sbuf_printf(sb, "DATA"); 3438 sbuf_printf(sb, "</Type>\n"); 3439 sbuf_printf(sb, "%s<Number>%u</Number>\n", indent, 3440 (u_int)disk->d_no); 3441 if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 3442 sbuf_printf(sb, "%s<Synchronized>", indent); 3443 if (disk->d_sync.ds_offset == 0) 3444 sbuf_printf(sb, "0%%"); 3445 else { 3446 sbuf_printf(sb, "%u%%", 3447 (u_int)((disk->d_sync.ds_offset * 100) / 3448 (sc->sc_mediasize / (sc->sc_ndisks - 1)))); 3449 } 3450 sbuf_printf(sb, "</Synchronized>\n"); 3451 if (disk->d_sync.ds_offset > 0) { 3452 sbuf_printf(sb, "%s<BytesSynced>%jd" 3453 "</BytesSynced>\n", indent, 3454 (intmax_t)disk->d_sync.ds_offset); 3455 } 3456 } 3457 sbuf_printf(sb, "%s<SyncID>%u</SyncID>\n", indent, 3458 disk->d_sync.ds_syncid); 3459 sbuf_printf(sb, "%s<GenID>%u</GenID>\n", indent, disk->d_genid); 3460 sbuf_printf(sb, "%s<Flags>", indent); 3461 if (disk->d_flags == 0) 3462 sbuf_printf(sb, "NONE"); 3463 else { 3464 int first = 1; 3465 3466 #define ADD_FLAG(flag, name) do { \ 3467 if ((disk->d_flags & (flag)) != 0) { \ 3468 if (!first) \ 3469 sbuf_printf(sb, ", "); \ 3470 else \ 3471 first = 0; \ 3472 sbuf_printf(sb, name); \ 3473 } \ 3474 } while (0) 3475 ADD_FLAG(G_RAID3_DISK_FLAG_DIRTY, "DIRTY"); 3476 ADD_FLAG(G_RAID3_DISK_FLAG_HARDCODED, "HARDCODED"); 3477 ADD_FLAG(G_RAID3_DISK_FLAG_SYNCHRONIZING, 3478 "SYNCHRONIZING"); 3479 ADD_FLAG(G_RAID3_DISK_FLAG_FORCE_SYNC, "FORCE_SYNC"); 3480 ADD_FLAG(G_RAID3_DISK_FLAG_BROKEN, "BROKEN"); 3481 #undef ADD_FLAG 3482 } 3483 sbuf_printf(sb, "</Flags>\n"); 3484 sbuf_printf(sb, "%s<State>%s</State>\n", indent, 3485 g_raid3_disk_state2str(disk->d_state)); 3486 sx_xunlock(&sc->sc_lock); 3487 g_topology_lock(); 3488 } else { 3489 g_topology_unlock(); 3490 sx_xlock(&sc->sc_lock); 3491 if (!g_raid3_use_malloc) { 3492 sbuf_printf(sb, 3493 "%s<Zone4kRequested>%u</Zone4kRequested>\n", indent, 3494 sc->sc_zones[G_RAID3_ZONE_4K].sz_requested); 3495 sbuf_printf(sb, 3496 "%s<Zone4kFailed>%u</Zone4kFailed>\n", indent, 3497 sc->sc_zones[G_RAID3_ZONE_4K].sz_failed); 3498 sbuf_printf(sb, 3499 "%s<Zone16kRequested>%u</Zone16kRequested>\n", indent, 3500 sc->sc_zones[G_RAID3_ZONE_16K].sz_requested); 3501 sbuf_printf(sb, 3502 "%s<Zone16kFailed>%u</Zone16kFailed>\n", indent, 3503 sc->sc_zones[G_RAID3_ZONE_16K].sz_failed); 3504 sbuf_printf(sb, 3505 "%s<Zone64kRequested>%u</Zone64kRequested>\n", indent, 3506 sc->sc_zones[G_RAID3_ZONE_64K].sz_requested); 3507 sbuf_printf(sb, 3508 "%s<Zone64kFailed>%u</Zone64kFailed>\n", indent, 3509 sc->sc_zones[G_RAID3_ZONE_64K].sz_failed); 3510 } 3511 sbuf_printf(sb, "%s<ID>%u</ID>\n", indent, (u_int)sc->sc_id); 3512 sbuf_printf(sb, "%s<SyncID>%u</SyncID>\n", indent, sc->sc_syncid); 3513 sbuf_printf(sb, "%s<GenID>%u</GenID>\n", indent, sc->sc_genid); 3514 sbuf_printf(sb, "%s<Flags>", indent); 3515 if (sc->sc_flags == 0) 3516 sbuf_printf(sb, "NONE"); 3517 else { 3518 int first = 1; 3519 3520 #define ADD_FLAG(flag, name) do { \ 3521 if ((sc->sc_flags & (flag)) != 0) { \ 3522 if (!first) \ 3523 sbuf_printf(sb, ", "); \ 3524 else \ 3525 first = 0; \ 3526 sbuf_printf(sb, name); \ 3527 } \ 3528 } while (0) 3529 ADD_FLAG(G_RAID3_DEVICE_FLAG_NOFAILSYNC, "NOFAILSYNC"); 3530 ADD_FLAG(G_RAID3_DEVICE_FLAG_NOAUTOSYNC, "NOAUTOSYNC"); 3531 ADD_FLAG(G_RAID3_DEVICE_FLAG_ROUND_ROBIN, 3532 "ROUND-ROBIN"); 3533 ADD_FLAG(G_RAID3_DEVICE_FLAG_VERIFY, "VERIFY"); 3534 #undef ADD_FLAG 3535 } 3536 sbuf_printf(sb, "</Flags>\n"); 3537 sbuf_printf(sb, "%s<Components>%u</Components>\n", indent, 3538 sc->sc_ndisks); 3539 sbuf_printf(sb, "%s<State>%s</State>\n", indent, 3540 g_raid3_device_state2str(sc->sc_state)); 3541 sx_xunlock(&sc->sc_lock); 3542 g_topology_lock(); 3543 } 3544 } 3545 3546 static void 3547 g_raid3_shutdown_pre_sync(void *arg, int howto) 3548 { 3549 struct g_class *mp; 3550 struct g_geom *gp, *gp2; 3551 struct g_raid3_softc *sc; 3552 int error; 3553 3554 mp = arg; 3555 DROP_GIANT(); 3556 g_topology_lock(); 3557 LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) { 3558 if ((sc = gp->softc) == NULL) 3559 continue; 3560 /* Skip synchronization geom. */ 3561 if (gp == sc->sc_sync.ds_geom) 3562 continue; 3563 g_topology_unlock(); 3564 sx_xlock(&sc->sc_lock); 3565 g_cancel_event(sc); 3566 error = g_raid3_destroy(sc, G_RAID3_DESTROY_DELAYED); 3567 if (error != 0) 3568 sx_xunlock(&sc->sc_lock); 3569 g_topology_lock(); 3570 } 3571 g_topology_unlock(); 3572 PICKUP_GIANT(); 3573 } 3574 3575 static void 3576 g_raid3_init(struct g_class *mp) 3577 { 3578 3579 g_raid3_pre_sync = EVENTHANDLER_REGISTER(shutdown_pre_sync, 3580 g_raid3_shutdown_pre_sync, mp, SHUTDOWN_PRI_FIRST); 3581 if (g_raid3_pre_sync == NULL) 3582 G_RAID3_DEBUG(0, "Warning! Cannot register shutdown event."); 3583 } 3584 3585 static void 3586 g_raid3_fini(struct g_class *mp) 3587 { 3588 3589 if (g_raid3_pre_sync != NULL) 3590 EVENTHANDLER_DEREGISTER(shutdown_pre_sync, g_raid3_pre_sync); 3591 } 3592 3593 DECLARE_GEOM_CLASS(g_raid3_class, g_raid3); 3594