1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2004-2006 Pawel Jakub Dawidek <pjd@FreeBSD.org> 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND 17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE 20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 */ 28 29 #include <sys/cdefs.h> 30 __FBSDID("$FreeBSD$"); 31 32 #include <sys/param.h> 33 #include <sys/systm.h> 34 #include <sys/kernel.h> 35 #include <sys/module.h> 36 #include <sys/limits.h> 37 #include <sys/lock.h> 38 #include <sys/mutex.h> 39 #include <sys/bio.h> 40 #include <sys/sbuf.h> 41 #include <sys/sysctl.h> 42 #include <sys/malloc.h> 43 #include <sys/eventhandler.h> 44 #include <vm/uma.h> 45 #include <geom/geom.h> 46 #include <geom/geom_dbg.h> 47 #include <sys/proc.h> 48 #include <sys/kthread.h> 49 #include <sys/sched.h> 50 #include <geom/raid3/g_raid3.h> 51 52 FEATURE(geom_raid3, "GEOM RAID-3 functionality"); 53 54 static MALLOC_DEFINE(M_RAID3, "raid3_data", "GEOM_RAID3 Data"); 55 56 SYSCTL_DECL(_kern_geom); 57 static SYSCTL_NODE(_kern_geom, OID_AUTO, raid3, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 58 "GEOM_RAID3 stuff"); 59 u_int g_raid3_debug = 0; 60 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, debug, CTLFLAG_RWTUN, &g_raid3_debug, 0, 61 "Debug level"); 62 static u_int g_raid3_timeout = 4; 63 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, timeout, CTLFLAG_RWTUN, &g_raid3_timeout, 64 0, "Time to wait on all raid3 components"); 65 static u_int g_raid3_idletime = 5; 66 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, idletime, CTLFLAG_RWTUN, 67 &g_raid3_idletime, 0, "Mark components as clean when idling"); 68 static u_int g_raid3_disconnect_on_failure = 1; 69 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, disconnect_on_failure, CTLFLAG_RWTUN, 70 &g_raid3_disconnect_on_failure, 0, "Disconnect component on I/O failure."); 71 static u_int g_raid3_syncreqs = 2; 72 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, sync_requests, CTLFLAG_RDTUN, 73 &g_raid3_syncreqs, 0, "Parallel synchronization I/O requests."); 74 static u_int g_raid3_use_malloc = 0; 75 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, use_malloc, CTLFLAG_RDTUN, 76 &g_raid3_use_malloc, 0, "Use malloc(9) instead of uma(9)."); 77 78 static u_int g_raid3_n64k = 50; 79 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, n64k, CTLFLAG_RDTUN, &g_raid3_n64k, 0, 80 "Maximum number of 64kB allocations"); 81 static u_int g_raid3_n16k = 200; 82 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, n16k, CTLFLAG_RDTUN, &g_raid3_n16k, 0, 83 "Maximum number of 16kB allocations"); 84 static u_int g_raid3_n4k = 1200; 85 SYSCTL_UINT(_kern_geom_raid3, OID_AUTO, n4k, CTLFLAG_RDTUN, &g_raid3_n4k, 0, 86 "Maximum number of 4kB allocations"); 87 88 static SYSCTL_NODE(_kern_geom_raid3, OID_AUTO, stat, 89 CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 90 "GEOM_RAID3 statistics"); 91 static u_int g_raid3_parity_mismatch = 0; 92 SYSCTL_UINT(_kern_geom_raid3_stat, OID_AUTO, parity_mismatch, CTLFLAG_RD, 93 &g_raid3_parity_mismatch, 0, "Number of failures in VERIFY mode"); 94 95 #define MSLEEP(ident, mtx, priority, wmesg, timeout) do { \ 96 G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, (ident)); \ 97 msleep((ident), (mtx), (priority), (wmesg), (timeout)); \ 98 G_RAID3_DEBUG(4, "%s: Woken up %p.", __func__, (ident)); \ 99 } while (0) 100 101 static eventhandler_tag g_raid3_post_sync = NULL; 102 static int g_raid3_shutdown = 0; 103 104 static int g_raid3_destroy_geom(struct gctl_req *req, struct g_class *mp, 105 struct g_geom *gp); 106 static g_taste_t g_raid3_taste; 107 static void g_raid3_init(struct g_class *mp); 108 static void g_raid3_fini(struct g_class *mp); 109 110 struct g_class g_raid3_class = { 111 .name = G_RAID3_CLASS_NAME, 112 .version = G_VERSION, 113 .ctlreq = g_raid3_config, 114 .taste = g_raid3_taste, 115 .destroy_geom = g_raid3_destroy_geom, 116 .init = g_raid3_init, 117 .fini = g_raid3_fini 118 }; 119 120 static void g_raid3_destroy_provider(struct g_raid3_softc *sc); 121 static int g_raid3_update_disk(struct g_raid3_disk *disk, u_int state); 122 static void g_raid3_update_device(struct g_raid3_softc *sc, boolean_t force); 123 static void g_raid3_dumpconf(struct sbuf *sb, const char *indent, 124 struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp); 125 static void g_raid3_sync_stop(struct g_raid3_softc *sc, int type); 126 static int g_raid3_register_request(struct bio *pbp); 127 static void g_raid3_sync_release(struct g_raid3_softc *sc); 128 129 static const char * 130 g_raid3_disk_state2str(int state) 131 { 132 133 switch (state) { 134 case G_RAID3_DISK_STATE_NODISK: 135 return ("NODISK"); 136 case G_RAID3_DISK_STATE_NONE: 137 return ("NONE"); 138 case G_RAID3_DISK_STATE_NEW: 139 return ("NEW"); 140 case G_RAID3_DISK_STATE_ACTIVE: 141 return ("ACTIVE"); 142 case G_RAID3_DISK_STATE_STALE: 143 return ("STALE"); 144 case G_RAID3_DISK_STATE_SYNCHRONIZING: 145 return ("SYNCHRONIZING"); 146 case G_RAID3_DISK_STATE_DISCONNECTED: 147 return ("DISCONNECTED"); 148 default: 149 return ("INVALID"); 150 } 151 } 152 153 static const char * 154 g_raid3_device_state2str(int state) 155 { 156 157 switch (state) { 158 case G_RAID3_DEVICE_STATE_STARTING: 159 return ("STARTING"); 160 case G_RAID3_DEVICE_STATE_DEGRADED: 161 return ("DEGRADED"); 162 case G_RAID3_DEVICE_STATE_COMPLETE: 163 return ("COMPLETE"); 164 default: 165 return ("INVALID"); 166 } 167 } 168 169 const char * 170 g_raid3_get_diskname(struct g_raid3_disk *disk) 171 { 172 173 if (disk->d_consumer == NULL || disk->d_consumer->provider == NULL) 174 return ("[unknown]"); 175 return (disk->d_name); 176 } 177 178 static void * 179 g_raid3_alloc(struct g_raid3_softc *sc, size_t size, int flags) 180 { 181 void *ptr; 182 enum g_raid3_zones zone; 183 184 if (g_raid3_use_malloc || 185 (zone = g_raid3_zone(size)) == G_RAID3_NUM_ZONES) 186 ptr = malloc(size, M_RAID3, flags); 187 else { 188 ptr = uma_zalloc_arg(sc->sc_zones[zone].sz_zone, 189 &sc->sc_zones[zone], flags); 190 sc->sc_zones[zone].sz_requested++; 191 if (ptr == NULL) 192 sc->sc_zones[zone].sz_failed++; 193 } 194 return (ptr); 195 } 196 197 static void 198 g_raid3_free(struct g_raid3_softc *sc, void *ptr, size_t size) 199 { 200 enum g_raid3_zones zone; 201 202 if (g_raid3_use_malloc || 203 (zone = g_raid3_zone(size)) == G_RAID3_NUM_ZONES) 204 free(ptr, M_RAID3); 205 else { 206 uma_zfree_arg(sc->sc_zones[zone].sz_zone, 207 ptr, &sc->sc_zones[zone]); 208 } 209 } 210 211 static int 212 g_raid3_uma_ctor(void *mem, int size, void *arg, int flags) 213 { 214 struct g_raid3_zone *sz = arg; 215 216 if (sz->sz_max > 0 && sz->sz_inuse == sz->sz_max) 217 return (ENOMEM); 218 sz->sz_inuse++; 219 return (0); 220 } 221 222 static void 223 g_raid3_uma_dtor(void *mem, int size, void *arg) 224 { 225 struct g_raid3_zone *sz = arg; 226 227 sz->sz_inuse--; 228 } 229 230 #define g_raid3_xor(src, dst, size) \ 231 _g_raid3_xor((uint64_t *)(src), \ 232 (uint64_t *)(dst), (size_t)size) 233 static void 234 _g_raid3_xor(uint64_t *src, uint64_t *dst, size_t size) 235 { 236 237 KASSERT((size % 128) == 0, ("Invalid size: %zu.", size)); 238 for (; size > 0; size -= 128) { 239 *dst++ ^= (*src++); 240 *dst++ ^= (*src++); 241 *dst++ ^= (*src++); 242 *dst++ ^= (*src++); 243 *dst++ ^= (*src++); 244 *dst++ ^= (*src++); 245 *dst++ ^= (*src++); 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 } 256 } 257 258 static int 259 g_raid3_is_zero(struct bio *bp) 260 { 261 static const uint64_t zeros[] = { 262 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 263 }; 264 u_char *addr; 265 ssize_t size; 266 267 size = bp->bio_length; 268 addr = (u_char *)bp->bio_data; 269 for (; size > 0; size -= sizeof(zeros), addr += sizeof(zeros)) { 270 if (bcmp(addr, zeros, sizeof(zeros)) != 0) 271 return (0); 272 } 273 return (1); 274 } 275 276 /* 277 * --- Events handling functions --- 278 * Events in geom_raid3 are used to maintain disks and device status 279 * from one thread to simplify locking. 280 */ 281 static void 282 g_raid3_event_free(struct g_raid3_event *ep) 283 { 284 285 free(ep, M_RAID3); 286 } 287 288 int 289 g_raid3_event_send(void *arg, int state, int flags) 290 { 291 struct g_raid3_softc *sc; 292 struct g_raid3_disk *disk; 293 struct g_raid3_event *ep; 294 int error; 295 296 ep = malloc(sizeof(*ep), M_RAID3, M_WAITOK); 297 G_RAID3_DEBUG(4, "%s: Sending event %p.", __func__, ep); 298 if ((flags & G_RAID3_EVENT_DEVICE) != 0) { 299 disk = NULL; 300 sc = arg; 301 } else { 302 disk = arg; 303 sc = disk->d_softc; 304 } 305 ep->e_disk = disk; 306 ep->e_state = state; 307 ep->e_flags = flags; 308 ep->e_error = 0; 309 mtx_lock(&sc->sc_events_mtx); 310 TAILQ_INSERT_TAIL(&sc->sc_events, ep, e_next); 311 mtx_unlock(&sc->sc_events_mtx); 312 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc); 313 mtx_lock(&sc->sc_queue_mtx); 314 wakeup(sc); 315 wakeup(&sc->sc_queue); 316 mtx_unlock(&sc->sc_queue_mtx); 317 if ((flags & G_RAID3_EVENT_DONTWAIT) != 0) 318 return (0); 319 sx_assert(&sc->sc_lock, SX_XLOCKED); 320 G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, ep); 321 sx_xunlock(&sc->sc_lock); 322 while ((ep->e_flags & G_RAID3_EVENT_DONE) == 0) { 323 mtx_lock(&sc->sc_events_mtx); 324 MSLEEP(ep, &sc->sc_events_mtx, PRIBIO | PDROP, "r3:event", 325 hz * 5); 326 } 327 error = ep->e_error; 328 g_raid3_event_free(ep); 329 sx_xlock(&sc->sc_lock); 330 return (error); 331 } 332 333 static struct g_raid3_event * 334 g_raid3_event_get(struct g_raid3_softc *sc) 335 { 336 struct g_raid3_event *ep; 337 338 mtx_lock(&sc->sc_events_mtx); 339 ep = TAILQ_FIRST(&sc->sc_events); 340 mtx_unlock(&sc->sc_events_mtx); 341 return (ep); 342 } 343 344 static void 345 g_raid3_event_remove(struct g_raid3_softc *sc, struct g_raid3_event *ep) 346 { 347 348 mtx_lock(&sc->sc_events_mtx); 349 TAILQ_REMOVE(&sc->sc_events, ep, e_next); 350 mtx_unlock(&sc->sc_events_mtx); 351 } 352 353 static void 354 g_raid3_event_cancel(struct g_raid3_disk *disk) 355 { 356 struct g_raid3_softc *sc; 357 struct g_raid3_event *ep, *tmpep; 358 359 sc = disk->d_softc; 360 sx_assert(&sc->sc_lock, SX_XLOCKED); 361 362 mtx_lock(&sc->sc_events_mtx); 363 TAILQ_FOREACH_SAFE(ep, &sc->sc_events, e_next, tmpep) { 364 if ((ep->e_flags & G_RAID3_EVENT_DEVICE) != 0) 365 continue; 366 if (ep->e_disk != disk) 367 continue; 368 TAILQ_REMOVE(&sc->sc_events, ep, e_next); 369 if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) 370 g_raid3_event_free(ep); 371 else { 372 ep->e_error = ECANCELED; 373 wakeup(ep); 374 } 375 } 376 mtx_unlock(&sc->sc_events_mtx); 377 } 378 379 /* 380 * Return the number of disks in the given state. 381 * If state is equal to -1, count all connected disks. 382 */ 383 u_int 384 g_raid3_ndisks(struct g_raid3_softc *sc, int state) 385 { 386 struct g_raid3_disk *disk; 387 u_int n, ndisks; 388 389 sx_assert(&sc->sc_lock, SX_LOCKED); 390 391 for (n = ndisks = 0; n < sc->sc_ndisks; n++) { 392 disk = &sc->sc_disks[n]; 393 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 394 continue; 395 if (state == -1 || disk->d_state == state) 396 ndisks++; 397 } 398 return (ndisks); 399 } 400 401 static u_int 402 g_raid3_nrequests(struct g_raid3_softc *sc, struct g_consumer *cp) 403 { 404 struct bio *bp; 405 u_int nreqs = 0; 406 407 mtx_lock(&sc->sc_queue_mtx); 408 TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { 409 if (bp->bio_from == cp) 410 nreqs++; 411 } 412 mtx_unlock(&sc->sc_queue_mtx); 413 return (nreqs); 414 } 415 416 static int 417 g_raid3_is_busy(struct g_raid3_softc *sc, struct g_consumer *cp) 418 { 419 420 if (cp->index > 0) { 421 G_RAID3_DEBUG(2, 422 "I/O requests for %s exist, can't destroy it now.", 423 cp->provider->name); 424 return (1); 425 } 426 if (g_raid3_nrequests(sc, cp) > 0) { 427 G_RAID3_DEBUG(2, 428 "I/O requests for %s in queue, can't destroy it now.", 429 cp->provider->name); 430 return (1); 431 } 432 return (0); 433 } 434 435 static void 436 g_raid3_destroy_consumer(void *arg, int flags __unused) 437 { 438 struct g_consumer *cp; 439 440 g_topology_assert(); 441 442 cp = arg; 443 G_RAID3_DEBUG(1, "Consumer %s destroyed.", cp->provider->name); 444 g_detach(cp); 445 g_destroy_consumer(cp); 446 } 447 448 static void 449 g_raid3_kill_consumer(struct g_raid3_softc *sc, struct g_consumer *cp) 450 { 451 struct g_provider *pp; 452 int retaste_wait; 453 454 g_topology_assert(); 455 456 cp->private = NULL; 457 if (g_raid3_is_busy(sc, cp)) 458 return; 459 G_RAID3_DEBUG(2, "Consumer %s destroyed.", cp->provider->name); 460 pp = cp->provider; 461 retaste_wait = 0; 462 if (cp->acw == 1) { 463 if ((pp->geom->flags & G_GEOM_WITHER) == 0) 464 retaste_wait = 1; 465 } 466 G_RAID3_DEBUG(2, "Access %s r%dw%de%d = %d", pp->name, -cp->acr, 467 -cp->acw, -cp->ace, 0); 468 if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) 469 g_access(cp, -cp->acr, -cp->acw, -cp->ace); 470 if (retaste_wait) { 471 /* 472 * After retaste event was send (inside g_access()), we can send 473 * event to detach and destroy consumer. 474 * A class, which has consumer to the given provider connected 475 * will not receive retaste event for the provider. 476 * This is the way how I ignore retaste events when I close 477 * consumers opened for write: I detach and destroy consumer 478 * after retaste event is sent. 479 */ 480 g_post_event(g_raid3_destroy_consumer, cp, M_WAITOK, NULL); 481 return; 482 } 483 G_RAID3_DEBUG(1, "Consumer %s destroyed.", pp->name); 484 g_detach(cp); 485 g_destroy_consumer(cp); 486 } 487 488 static int 489 g_raid3_connect_disk(struct g_raid3_disk *disk, struct g_provider *pp) 490 { 491 struct g_consumer *cp; 492 int error; 493 494 g_topology_assert_not(); 495 KASSERT(disk->d_consumer == NULL, 496 ("Disk already connected (device %s).", disk->d_softc->sc_name)); 497 498 g_topology_lock(); 499 cp = g_new_consumer(disk->d_softc->sc_geom); 500 error = g_attach(cp, pp); 501 if (error != 0) { 502 g_destroy_consumer(cp); 503 g_topology_unlock(); 504 return (error); 505 } 506 error = g_access(cp, 1, 1, 1); 507 g_topology_unlock(); 508 if (error != 0) { 509 g_detach(cp); 510 g_destroy_consumer(cp); 511 G_RAID3_DEBUG(0, "Cannot open consumer %s (error=%d).", 512 pp->name, error); 513 return (error); 514 } 515 disk->d_consumer = cp; 516 disk->d_consumer->private = disk; 517 disk->d_consumer->index = 0; 518 G_RAID3_DEBUG(2, "Disk %s connected.", g_raid3_get_diskname(disk)); 519 return (0); 520 } 521 522 static void 523 g_raid3_disconnect_consumer(struct g_raid3_softc *sc, struct g_consumer *cp) 524 { 525 526 g_topology_assert(); 527 528 if (cp == NULL) 529 return; 530 if (cp->provider != NULL) 531 g_raid3_kill_consumer(sc, cp); 532 else 533 g_destroy_consumer(cp); 534 } 535 536 /* 537 * Initialize disk. This means allocate memory, create consumer, attach it 538 * to the provider and open access (r1w1e1) to it. 539 */ 540 static struct g_raid3_disk * 541 g_raid3_init_disk(struct g_raid3_softc *sc, struct g_provider *pp, 542 struct g_raid3_metadata *md, int *errorp) 543 { 544 struct g_raid3_disk *disk; 545 int error; 546 547 disk = &sc->sc_disks[md->md_no]; 548 error = g_raid3_connect_disk(disk, pp); 549 if (error != 0) { 550 if (errorp != NULL) 551 *errorp = error; 552 return (NULL); 553 } 554 disk->d_state = G_RAID3_DISK_STATE_NONE; 555 disk->d_flags = md->md_dflags; 556 if (md->md_provider[0] != '\0') 557 disk->d_flags |= G_RAID3_DISK_FLAG_HARDCODED; 558 disk->d_sync.ds_consumer = NULL; 559 disk->d_sync.ds_offset = md->md_sync_offset; 560 disk->d_sync.ds_offset_done = md->md_sync_offset; 561 disk->d_genid = md->md_genid; 562 disk->d_sync.ds_syncid = md->md_syncid; 563 if (errorp != NULL) 564 *errorp = 0; 565 return (disk); 566 } 567 568 static void 569 g_raid3_destroy_disk(struct g_raid3_disk *disk) 570 { 571 struct g_raid3_softc *sc; 572 573 g_topology_assert_not(); 574 sc = disk->d_softc; 575 sx_assert(&sc->sc_lock, SX_XLOCKED); 576 577 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 578 return; 579 g_raid3_event_cancel(disk); 580 switch (disk->d_state) { 581 case G_RAID3_DISK_STATE_SYNCHRONIZING: 582 if (sc->sc_syncdisk != NULL) 583 g_raid3_sync_stop(sc, 1); 584 /* FALLTHROUGH */ 585 case G_RAID3_DISK_STATE_NEW: 586 case G_RAID3_DISK_STATE_STALE: 587 case G_RAID3_DISK_STATE_ACTIVE: 588 g_topology_lock(); 589 g_raid3_disconnect_consumer(sc, disk->d_consumer); 590 g_topology_unlock(); 591 disk->d_consumer = NULL; 592 break; 593 default: 594 KASSERT(0 == 1, ("Wrong disk state (%s, %s).", 595 g_raid3_get_diskname(disk), 596 g_raid3_disk_state2str(disk->d_state))); 597 } 598 disk->d_state = G_RAID3_DISK_STATE_NODISK; 599 } 600 601 static void 602 g_raid3_destroy_device(struct g_raid3_softc *sc) 603 { 604 struct g_raid3_event *ep; 605 struct g_raid3_disk *disk; 606 struct g_geom *gp; 607 struct g_consumer *cp; 608 u_int n; 609 610 g_topology_assert_not(); 611 sx_assert(&sc->sc_lock, SX_XLOCKED); 612 613 gp = sc->sc_geom; 614 if (sc->sc_provider != NULL) 615 g_raid3_destroy_provider(sc); 616 for (n = 0; n < sc->sc_ndisks; n++) { 617 disk = &sc->sc_disks[n]; 618 if (disk->d_state != G_RAID3_DISK_STATE_NODISK) { 619 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 620 g_raid3_update_metadata(disk); 621 g_raid3_destroy_disk(disk); 622 } 623 } 624 while ((ep = g_raid3_event_get(sc)) != NULL) { 625 g_raid3_event_remove(sc, ep); 626 if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) 627 g_raid3_event_free(ep); 628 else { 629 ep->e_error = ECANCELED; 630 ep->e_flags |= G_RAID3_EVENT_DONE; 631 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, ep); 632 mtx_lock(&sc->sc_events_mtx); 633 wakeup(ep); 634 mtx_unlock(&sc->sc_events_mtx); 635 } 636 } 637 callout_drain(&sc->sc_callout); 638 cp = LIST_FIRST(&sc->sc_sync.ds_geom->consumer); 639 g_topology_lock(); 640 if (cp != NULL) 641 g_raid3_disconnect_consumer(sc, cp); 642 g_wither_geom(sc->sc_sync.ds_geom, ENXIO); 643 G_RAID3_DEBUG(0, "Device %s destroyed.", gp->name); 644 g_wither_geom(gp, ENXIO); 645 g_topology_unlock(); 646 if (!g_raid3_use_malloc) { 647 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_64K].sz_zone); 648 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_16K].sz_zone); 649 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_4K].sz_zone); 650 } 651 mtx_destroy(&sc->sc_queue_mtx); 652 mtx_destroy(&sc->sc_events_mtx); 653 sx_xunlock(&sc->sc_lock); 654 sx_destroy(&sc->sc_lock); 655 } 656 657 static void 658 g_raid3_orphan(struct g_consumer *cp) 659 { 660 struct g_raid3_disk *disk; 661 662 g_topology_assert(); 663 664 disk = cp->private; 665 if (disk == NULL) 666 return; 667 disk->d_softc->sc_bump_id = G_RAID3_BUMP_SYNCID; 668 g_raid3_event_send(disk, G_RAID3_DISK_STATE_DISCONNECTED, 669 G_RAID3_EVENT_DONTWAIT); 670 } 671 672 static int 673 g_raid3_write_metadata(struct g_raid3_disk *disk, struct g_raid3_metadata *md) 674 { 675 struct g_raid3_softc *sc; 676 struct g_consumer *cp; 677 off_t offset, length; 678 u_char *sector; 679 int error = 0; 680 681 g_topology_assert_not(); 682 sc = disk->d_softc; 683 sx_assert(&sc->sc_lock, SX_LOCKED); 684 685 cp = disk->d_consumer; 686 KASSERT(cp != NULL, ("NULL consumer (%s).", sc->sc_name)); 687 KASSERT(cp->provider != NULL, ("NULL provider (%s).", sc->sc_name)); 688 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 689 ("Consumer %s closed? (r%dw%de%d).", cp->provider->name, cp->acr, 690 cp->acw, cp->ace)); 691 length = cp->provider->sectorsize; 692 offset = cp->provider->mediasize - length; 693 sector = malloc((size_t)length, M_RAID3, M_WAITOK | M_ZERO); 694 if (md != NULL) 695 raid3_metadata_encode(md, sector); 696 error = g_write_data(cp, offset, sector, length); 697 free(sector, M_RAID3); 698 if (error != 0) { 699 if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) { 700 G_RAID3_DEBUG(0, "Cannot write metadata on %s " 701 "(device=%s, error=%d).", 702 g_raid3_get_diskname(disk), sc->sc_name, error); 703 disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN; 704 } else { 705 G_RAID3_DEBUG(1, "Cannot write metadata on %s " 706 "(device=%s, error=%d).", 707 g_raid3_get_diskname(disk), sc->sc_name, error); 708 } 709 if (g_raid3_disconnect_on_failure && 710 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 711 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 712 g_raid3_event_send(disk, 713 G_RAID3_DISK_STATE_DISCONNECTED, 714 G_RAID3_EVENT_DONTWAIT); 715 } 716 } 717 return (error); 718 } 719 720 int 721 g_raid3_clear_metadata(struct g_raid3_disk *disk) 722 { 723 int error; 724 725 g_topology_assert_not(); 726 sx_assert(&disk->d_softc->sc_lock, SX_LOCKED); 727 728 error = g_raid3_write_metadata(disk, NULL); 729 if (error == 0) { 730 G_RAID3_DEBUG(2, "Metadata on %s cleared.", 731 g_raid3_get_diskname(disk)); 732 } else { 733 G_RAID3_DEBUG(0, 734 "Cannot clear metadata on disk %s (error=%d).", 735 g_raid3_get_diskname(disk), error); 736 } 737 return (error); 738 } 739 740 void 741 g_raid3_fill_metadata(struct g_raid3_disk *disk, struct g_raid3_metadata *md) 742 { 743 struct g_raid3_softc *sc; 744 struct g_provider *pp; 745 746 sc = disk->d_softc; 747 strlcpy(md->md_magic, G_RAID3_MAGIC, sizeof(md->md_magic)); 748 md->md_version = G_RAID3_VERSION; 749 strlcpy(md->md_name, sc->sc_name, sizeof(md->md_name)); 750 md->md_id = sc->sc_id; 751 md->md_all = sc->sc_ndisks; 752 md->md_genid = sc->sc_genid; 753 md->md_mediasize = sc->sc_mediasize; 754 md->md_sectorsize = sc->sc_sectorsize; 755 md->md_mflags = (sc->sc_flags & G_RAID3_DEVICE_FLAG_MASK); 756 md->md_no = disk->d_no; 757 md->md_syncid = disk->d_sync.ds_syncid; 758 md->md_dflags = (disk->d_flags & G_RAID3_DISK_FLAG_MASK); 759 if (disk->d_state != G_RAID3_DISK_STATE_SYNCHRONIZING) 760 md->md_sync_offset = 0; 761 else { 762 md->md_sync_offset = 763 disk->d_sync.ds_offset_done / (sc->sc_ndisks - 1); 764 } 765 if (disk->d_consumer != NULL && disk->d_consumer->provider != NULL) 766 pp = disk->d_consumer->provider; 767 else 768 pp = NULL; 769 if ((disk->d_flags & G_RAID3_DISK_FLAG_HARDCODED) != 0 && pp != NULL) 770 strlcpy(md->md_provider, pp->name, sizeof(md->md_provider)); 771 else 772 bzero(md->md_provider, sizeof(md->md_provider)); 773 if (pp != NULL) 774 md->md_provsize = pp->mediasize; 775 else 776 md->md_provsize = 0; 777 } 778 779 void 780 g_raid3_update_metadata(struct g_raid3_disk *disk) 781 { 782 struct g_raid3_softc *sc; 783 struct g_raid3_metadata md; 784 int error; 785 786 g_topology_assert_not(); 787 sc = disk->d_softc; 788 sx_assert(&sc->sc_lock, SX_LOCKED); 789 790 g_raid3_fill_metadata(disk, &md); 791 error = g_raid3_write_metadata(disk, &md); 792 if (error == 0) { 793 G_RAID3_DEBUG(2, "Metadata on %s updated.", 794 g_raid3_get_diskname(disk)); 795 } else { 796 G_RAID3_DEBUG(0, 797 "Cannot update metadata on disk %s (error=%d).", 798 g_raid3_get_diskname(disk), error); 799 } 800 } 801 802 static void 803 g_raid3_bump_syncid(struct g_raid3_softc *sc) 804 { 805 struct g_raid3_disk *disk; 806 u_int n; 807 808 g_topology_assert_not(); 809 sx_assert(&sc->sc_lock, SX_XLOCKED); 810 KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) > 0, 811 ("%s called with no active disks (device=%s).", __func__, 812 sc->sc_name)); 813 814 sc->sc_syncid++; 815 G_RAID3_DEBUG(1, "Device %s: syncid bumped to %u.", sc->sc_name, 816 sc->sc_syncid); 817 for (n = 0; n < sc->sc_ndisks; n++) { 818 disk = &sc->sc_disks[n]; 819 if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 820 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 821 disk->d_sync.ds_syncid = sc->sc_syncid; 822 g_raid3_update_metadata(disk); 823 } 824 } 825 } 826 827 static void 828 g_raid3_bump_genid(struct g_raid3_softc *sc) 829 { 830 struct g_raid3_disk *disk; 831 u_int n; 832 833 g_topology_assert_not(); 834 sx_assert(&sc->sc_lock, SX_XLOCKED); 835 KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) > 0, 836 ("%s called with no active disks (device=%s).", __func__, 837 sc->sc_name)); 838 839 sc->sc_genid++; 840 G_RAID3_DEBUG(1, "Device %s: genid bumped to %u.", sc->sc_name, 841 sc->sc_genid); 842 for (n = 0; n < sc->sc_ndisks; n++) { 843 disk = &sc->sc_disks[n]; 844 if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 845 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 846 disk->d_genid = sc->sc_genid; 847 g_raid3_update_metadata(disk); 848 } 849 } 850 } 851 852 static int 853 g_raid3_idle(struct g_raid3_softc *sc, int acw) 854 { 855 struct g_raid3_disk *disk; 856 u_int i; 857 int timeout; 858 859 g_topology_assert_not(); 860 sx_assert(&sc->sc_lock, SX_XLOCKED); 861 862 if (sc->sc_provider == NULL) 863 return (0); 864 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0) 865 return (0); 866 if (sc->sc_idle) 867 return (0); 868 if (sc->sc_writes > 0) 869 return (0); 870 if (acw > 0 || (acw == -1 && sc->sc_provider->acw > 0)) { 871 timeout = g_raid3_idletime - (time_uptime - sc->sc_last_write); 872 if (!g_raid3_shutdown && timeout > 0) 873 return (timeout); 874 } 875 sc->sc_idle = 1; 876 for (i = 0; i < sc->sc_ndisks; i++) { 877 disk = &sc->sc_disks[i]; 878 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) 879 continue; 880 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as clean.", 881 g_raid3_get_diskname(disk), sc->sc_name); 882 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 883 g_raid3_update_metadata(disk); 884 } 885 return (0); 886 } 887 888 static void 889 g_raid3_unidle(struct g_raid3_softc *sc) 890 { 891 struct g_raid3_disk *disk; 892 u_int i; 893 894 g_topology_assert_not(); 895 sx_assert(&sc->sc_lock, SX_XLOCKED); 896 897 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0) 898 return; 899 sc->sc_idle = 0; 900 sc->sc_last_write = time_uptime; 901 for (i = 0; i < sc->sc_ndisks; i++) { 902 disk = &sc->sc_disks[i]; 903 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) 904 continue; 905 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as dirty.", 906 g_raid3_get_diskname(disk), sc->sc_name); 907 disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY; 908 g_raid3_update_metadata(disk); 909 } 910 } 911 912 /* 913 * Treat bio_driver1 field in parent bio as list head and field bio_caller1 914 * in child bio as pointer to the next element on the list. 915 */ 916 #define G_RAID3_HEAD_BIO(pbp) (pbp)->bio_driver1 917 918 #define G_RAID3_NEXT_BIO(cbp) (cbp)->bio_caller1 919 920 #define G_RAID3_FOREACH_BIO(pbp, bp) \ 921 for ((bp) = G_RAID3_HEAD_BIO(pbp); (bp) != NULL; \ 922 (bp) = G_RAID3_NEXT_BIO(bp)) 923 924 #define G_RAID3_FOREACH_SAFE_BIO(pbp, bp, tmpbp) \ 925 for ((bp) = G_RAID3_HEAD_BIO(pbp); \ 926 (bp) != NULL && ((tmpbp) = G_RAID3_NEXT_BIO(bp), 1); \ 927 (bp) = (tmpbp)) 928 929 static void 930 g_raid3_init_bio(struct bio *pbp) 931 { 932 933 G_RAID3_HEAD_BIO(pbp) = NULL; 934 } 935 936 static void 937 g_raid3_remove_bio(struct bio *cbp) 938 { 939 struct bio *pbp, *bp; 940 941 pbp = cbp->bio_parent; 942 if (G_RAID3_HEAD_BIO(pbp) == cbp) 943 G_RAID3_HEAD_BIO(pbp) = G_RAID3_NEXT_BIO(cbp); 944 else { 945 G_RAID3_FOREACH_BIO(pbp, bp) { 946 if (G_RAID3_NEXT_BIO(bp) == cbp) { 947 G_RAID3_NEXT_BIO(bp) = G_RAID3_NEXT_BIO(cbp); 948 break; 949 } 950 } 951 } 952 G_RAID3_NEXT_BIO(cbp) = NULL; 953 } 954 955 static void 956 g_raid3_replace_bio(struct bio *sbp, struct bio *dbp) 957 { 958 struct bio *pbp, *bp; 959 960 g_raid3_remove_bio(sbp); 961 pbp = dbp->bio_parent; 962 G_RAID3_NEXT_BIO(sbp) = G_RAID3_NEXT_BIO(dbp); 963 if (G_RAID3_HEAD_BIO(pbp) == dbp) 964 G_RAID3_HEAD_BIO(pbp) = sbp; 965 else { 966 G_RAID3_FOREACH_BIO(pbp, bp) { 967 if (G_RAID3_NEXT_BIO(bp) == dbp) { 968 G_RAID3_NEXT_BIO(bp) = sbp; 969 break; 970 } 971 } 972 } 973 G_RAID3_NEXT_BIO(dbp) = NULL; 974 } 975 976 static void 977 g_raid3_destroy_bio(struct g_raid3_softc *sc, struct bio *cbp) 978 { 979 struct bio *bp, *pbp; 980 size_t size; 981 982 pbp = cbp->bio_parent; 983 pbp->bio_children--; 984 KASSERT(cbp->bio_data != NULL, ("NULL bio_data")); 985 size = pbp->bio_length / (sc->sc_ndisks - 1); 986 g_raid3_free(sc, cbp->bio_data, size); 987 if (G_RAID3_HEAD_BIO(pbp) == cbp) { 988 G_RAID3_HEAD_BIO(pbp) = G_RAID3_NEXT_BIO(cbp); 989 G_RAID3_NEXT_BIO(cbp) = NULL; 990 g_destroy_bio(cbp); 991 } else { 992 G_RAID3_FOREACH_BIO(pbp, bp) { 993 if (G_RAID3_NEXT_BIO(bp) == cbp) 994 break; 995 } 996 if (bp != NULL) { 997 KASSERT(G_RAID3_NEXT_BIO(bp) != NULL, 998 ("NULL bp->bio_driver1")); 999 G_RAID3_NEXT_BIO(bp) = G_RAID3_NEXT_BIO(cbp); 1000 G_RAID3_NEXT_BIO(cbp) = NULL; 1001 } 1002 g_destroy_bio(cbp); 1003 } 1004 } 1005 1006 static struct bio * 1007 g_raid3_clone_bio(struct g_raid3_softc *sc, struct bio *pbp) 1008 { 1009 struct bio *bp, *cbp; 1010 size_t size; 1011 int memflag; 1012 1013 cbp = g_clone_bio(pbp); 1014 if (cbp == NULL) 1015 return (NULL); 1016 size = pbp->bio_length / (sc->sc_ndisks - 1); 1017 if ((pbp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) != 0) 1018 memflag = M_WAITOK; 1019 else 1020 memflag = M_NOWAIT; 1021 cbp->bio_data = g_raid3_alloc(sc, size, memflag); 1022 if (cbp->bio_data == NULL) { 1023 pbp->bio_children--; 1024 g_destroy_bio(cbp); 1025 return (NULL); 1026 } 1027 G_RAID3_NEXT_BIO(cbp) = NULL; 1028 if (G_RAID3_HEAD_BIO(pbp) == NULL) 1029 G_RAID3_HEAD_BIO(pbp) = cbp; 1030 else { 1031 G_RAID3_FOREACH_BIO(pbp, bp) { 1032 if (G_RAID3_NEXT_BIO(bp) == NULL) { 1033 G_RAID3_NEXT_BIO(bp) = cbp; 1034 break; 1035 } 1036 } 1037 } 1038 return (cbp); 1039 } 1040 1041 static void 1042 g_raid3_scatter(struct bio *pbp) 1043 { 1044 struct g_raid3_softc *sc; 1045 struct g_raid3_disk *disk; 1046 struct bio *bp, *cbp, *tmpbp; 1047 off_t atom, cadd, padd, left; 1048 int first; 1049 1050 sc = pbp->bio_to->geom->softc; 1051 bp = NULL; 1052 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_NOPARITY) == 0) { 1053 /* 1054 * Find bio for which we should calculate data. 1055 */ 1056 G_RAID3_FOREACH_BIO(pbp, cbp) { 1057 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) { 1058 bp = cbp; 1059 break; 1060 } 1061 } 1062 KASSERT(bp != NULL, ("NULL parity bio.")); 1063 } 1064 atom = sc->sc_sectorsize / (sc->sc_ndisks - 1); 1065 cadd = padd = 0; 1066 for (left = pbp->bio_length; left > 0; left -= sc->sc_sectorsize) { 1067 G_RAID3_FOREACH_BIO(pbp, cbp) { 1068 if (cbp == bp) 1069 continue; 1070 bcopy(pbp->bio_data + padd, cbp->bio_data + cadd, atom); 1071 padd += atom; 1072 } 1073 cadd += atom; 1074 } 1075 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_NOPARITY) == 0) { 1076 /* 1077 * Calculate parity. 1078 */ 1079 first = 1; 1080 G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) { 1081 if (cbp == bp) 1082 continue; 1083 if (first) { 1084 bcopy(cbp->bio_data, bp->bio_data, 1085 bp->bio_length); 1086 first = 0; 1087 } else { 1088 g_raid3_xor(cbp->bio_data, bp->bio_data, 1089 bp->bio_length); 1090 } 1091 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_NODISK) != 0) 1092 g_raid3_destroy_bio(sc, cbp); 1093 } 1094 } 1095 G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) { 1096 struct g_consumer *cp; 1097 1098 disk = cbp->bio_caller2; 1099 cp = disk->d_consumer; 1100 cbp->bio_to = cp->provider; 1101 G_RAID3_LOGREQ(3, cbp, "Sending request."); 1102 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1103 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1104 cp->acr, cp->acw, cp->ace)); 1105 cp->index++; 1106 sc->sc_writes++; 1107 g_io_request(cbp, cp); 1108 } 1109 } 1110 1111 static void 1112 g_raid3_gather(struct bio *pbp) 1113 { 1114 struct g_raid3_softc *sc; 1115 struct g_raid3_disk *disk; 1116 struct bio *xbp, *fbp, *cbp; 1117 off_t atom, cadd, padd, left; 1118 1119 sc = pbp->bio_to->geom->softc; 1120 /* 1121 * Find bio for which we have to calculate data. 1122 * While going through this path, check if all requests 1123 * succeeded, if not, deny whole request. 1124 * If we're in COMPLETE mode, we allow one request to fail, 1125 * so if we find one, we're sending it to the parity consumer. 1126 * If there are more failed requests, we deny whole request. 1127 */ 1128 xbp = fbp = NULL; 1129 G_RAID3_FOREACH_BIO(pbp, cbp) { 1130 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) { 1131 KASSERT(xbp == NULL, ("More than one parity bio.")); 1132 xbp = cbp; 1133 } 1134 if (cbp->bio_error == 0) 1135 continue; 1136 /* 1137 * Found failed request. 1138 */ 1139 if (fbp == NULL) { 1140 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_DEGRADED) != 0) { 1141 /* 1142 * We are already in degraded mode, so we can't 1143 * accept any failures. 1144 */ 1145 if (pbp->bio_error == 0) 1146 pbp->bio_error = cbp->bio_error; 1147 } else { 1148 fbp = cbp; 1149 } 1150 } else { 1151 /* 1152 * Next failed request, that's too many. 1153 */ 1154 if (pbp->bio_error == 0) 1155 pbp->bio_error = fbp->bio_error; 1156 } 1157 disk = cbp->bio_caller2; 1158 if (disk == NULL) 1159 continue; 1160 if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) { 1161 disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN; 1162 G_RAID3_LOGREQ(0, cbp, "Request failed (error=%d).", 1163 cbp->bio_error); 1164 } else { 1165 G_RAID3_LOGREQ(1, cbp, "Request failed (error=%d).", 1166 cbp->bio_error); 1167 } 1168 if (g_raid3_disconnect_on_failure && 1169 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1170 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 1171 g_raid3_event_send(disk, 1172 G_RAID3_DISK_STATE_DISCONNECTED, 1173 G_RAID3_EVENT_DONTWAIT); 1174 } 1175 } 1176 if (pbp->bio_error != 0) 1177 goto finish; 1178 if (fbp != NULL && (pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) { 1179 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_VERIFY; 1180 if (xbp != fbp) 1181 g_raid3_replace_bio(xbp, fbp); 1182 g_raid3_destroy_bio(sc, fbp); 1183 } else if (fbp != NULL) { 1184 struct g_consumer *cp; 1185 1186 /* 1187 * One request failed, so send the same request to 1188 * the parity consumer. 1189 */ 1190 disk = pbp->bio_driver2; 1191 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) { 1192 pbp->bio_error = fbp->bio_error; 1193 goto finish; 1194 } 1195 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED; 1196 pbp->bio_inbed--; 1197 fbp->bio_flags &= ~(BIO_DONE | BIO_ERROR); 1198 if (disk->d_no == sc->sc_ndisks - 1) 1199 fbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1200 fbp->bio_error = 0; 1201 fbp->bio_completed = 0; 1202 fbp->bio_children = 0; 1203 fbp->bio_inbed = 0; 1204 cp = disk->d_consumer; 1205 fbp->bio_caller2 = disk; 1206 fbp->bio_to = cp->provider; 1207 G_RAID3_LOGREQ(3, fbp, "Sending request (recover)."); 1208 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1209 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1210 cp->acr, cp->acw, cp->ace)); 1211 cp->index++; 1212 g_io_request(fbp, cp); 1213 return; 1214 } 1215 if (xbp != NULL) { 1216 /* 1217 * Calculate parity. 1218 */ 1219 G_RAID3_FOREACH_BIO(pbp, cbp) { 1220 if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) 1221 continue; 1222 g_raid3_xor(cbp->bio_data, xbp->bio_data, 1223 xbp->bio_length); 1224 } 1225 xbp->bio_cflags &= ~G_RAID3_BIO_CFLAG_PARITY; 1226 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) { 1227 if (!g_raid3_is_zero(xbp)) { 1228 g_raid3_parity_mismatch++; 1229 pbp->bio_error = EIO; 1230 goto finish; 1231 } 1232 g_raid3_destroy_bio(sc, xbp); 1233 } 1234 } 1235 atom = sc->sc_sectorsize / (sc->sc_ndisks - 1); 1236 cadd = padd = 0; 1237 for (left = pbp->bio_length; left > 0; left -= sc->sc_sectorsize) { 1238 G_RAID3_FOREACH_BIO(pbp, cbp) { 1239 bcopy(cbp->bio_data + cadd, pbp->bio_data + padd, atom); 1240 pbp->bio_completed += atom; 1241 padd += atom; 1242 } 1243 cadd += atom; 1244 } 1245 finish: 1246 if (pbp->bio_error == 0) 1247 G_RAID3_LOGREQ(3, pbp, "Request finished."); 1248 else { 1249 if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) 1250 G_RAID3_LOGREQ(1, pbp, "Verification error."); 1251 else 1252 G_RAID3_LOGREQ(0, pbp, "Request failed."); 1253 } 1254 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_MASK; 1255 while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) 1256 g_raid3_destroy_bio(sc, cbp); 1257 g_io_deliver(pbp, pbp->bio_error); 1258 } 1259 1260 static void 1261 g_raid3_done(struct bio *bp) 1262 { 1263 struct g_raid3_softc *sc; 1264 1265 sc = bp->bio_from->geom->softc; 1266 bp->bio_cflags |= G_RAID3_BIO_CFLAG_REGULAR; 1267 G_RAID3_LOGREQ(3, bp, "Regular request done (error=%d).", bp->bio_error); 1268 mtx_lock(&sc->sc_queue_mtx); 1269 bioq_insert_head(&sc->sc_queue, bp); 1270 mtx_unlock(&sc->sc_queue_mtx); 1271 wakeup(sc); 1272 wakeup(&sc->sc_queue); 1273 } 1274 1275 static void 1276 g_raid3_regular_request(struct bio *cbp) 1277 { 1278 struct g_raid3_softc *sc; 1279 struct g_raid3_disk *disk; 1280 struct bio *pbp; 1281 1282 g_topology_assert_not(); 1283 1284 pbp = cbp->bio_parent; 1285 sc = pbp->bio_to->geom->softc; 1286 cbp->bio_from->index--; 1287 if (cbp->bio_cmd == BIO_WRITE) 1288 sc->sc_writes--; 1289 disk = cbp->bio_from->private; 1290 if (disk == NULL) { 1291 g_topology_lock(); 1292 g_raid3_kill_consumer(sc, cbp->bio_from); 1293 g_topology_unlock(); 1294 } 1295 1296 G_RAID3_LOGREQ(3, cbp, "Request finished."); 1297 pbp->bio_inbed++; 1298 KASSERT(pbp->bio_inbed <= pbp->bio_children, 1299 ("bio_inbed (%u) is bigger than bio_children (%u).", pbp->bio_inbed, 1300 pbp->bio_children)); 1301 if (pbp->bio_inbed != pbp->bio_children) 1302 return; 1303 switch (pbp->bio_cmd) { 1304 case BIO_READ: 1305 g_raid3_gather(pbp); 1306 break; 1307 case BIO_WRITE: 1308 case BIO_DELETE: 1309 { 1310 int error = 0; 1311 1312 pbp->bio_completed = pbp->bio_length; 1313 while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) { 1314 if (cbp->bio_error == 0) { 1315 g_raid3_destroy_bio(sc, cbp); 1316 continue; 1317 } 1318 1319 if (error == 0) 1320 error = cbp->bio_error; 1321 else if (pbp->bio_error == 0) { 1322 /* 1323 * Next failed request, that's too many. 1324 */ 1325 pbp->bio_error = error; 1326 } 1327 1328 disk = cbp->bio_caller2; 1329 if (disk == NULL) { 1330 g_raid3_destroy_bio(sc, cbp); 1331 continue; 1332 } 1333 1334 if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) { 1335 disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN; 1336 G_RAID3_LOGREQ(0, cbp, 1337 "Request failed (error=%d).", 1338 cbp->bio_error); 1339 } else { 1340 G_RAID3_LOGREQ(1, cbp, 1341 "Request failed (error=%d).", 1342 cbp->bio_error); 1343 } 1344 if (g_raid3_disconnect_on_failure && 1345 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1346 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 1347 g_raid3_event_send(disk, 1348 G_RAID3_DISK_STATE_DISCONNECTED, 1349 G_RAID3_EVENT_DONTWAIT); 1350 } 1351 g_raid3_destroy_bio(sc, cbp); 1352 } 1353 if (pbp->bio_error == 0) 1354 G_RAID3_LOGREQ(3, pbp, "Request finished."); 1355 else 1356 G_RAID3_LOGREQ(0, pbp, "Request failed."); 1357 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_DEGRADED; 1358 pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_NOPARITY; 1359 bioq_remove(&sc->sc_inflight, pbp); 1360 /* Release delayed sync requests if possible. */ 1361 g_raid3_sync_release(sc); 1362 g_io_deliver(pbp, pbp->bio_error); 1363 break; 1364 } 1365 } 1366 } 1367 1368 static void 1369 g_raid3_sync_done(struct bio *bp) 1370 { 1371 struct g_raid3_softc *sc; 1372 1373 G_RAID3_LOGREQ(3, bp, "Synchronization request delivered."); 1374 sc = bp->bio_from->geom->softc; 1375 bp->bio_cflags |= G_RAID3_BIO_CFLAG_SYNC; 1376 mtx_lock(&sc->sc_queue_mtx); 1377 bioq_insert_head(&sc->sc_queue, bp); 1378 mtx_unlock(&sc->sc_queue_mtx); 1379 wakeup(sc); 1380 wakeup(&sc->sc_queue); 1381 } 1382 1383 static void 1384 g_raid3_flush(struct g_raid3_softc *sc, struct bio *bp) 1385 { 1386 struct bio_queue_head queue; 1387 struct g_raid3_disk *disk; 1388 struct g_consumer *cp; 1389 struct bio *cbp; 1390 u_int i; 1391 1392 bioq_init(&queue); 1393 for (i = 0; i < sc->sc_ndisks; i++) { 1394 disk = &sc->sc_disks[i]; 1395 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) 1396 continue; 1397 cbp = g_clone_bio(bp); 1398 if (cbp == NULL) { 1399 for (cbp = bioq_first(&queue); cbp != NULL; 1400 cbp = bioq_first(&queue)) { 1401 bioq_remove(&queue, cbp); 1402 g_destroy_bio(cbp); 1403 } 1404 if (bp->bio_error == 0) 1405 bp->bio_error = ENOMEM; 1406 g_io_deliver(bp, bp->bio_error); 1407 return; 1408 } 1409 bioq_insert_tail(&queue, cbp); 1410 cbp->bio_done = g_std_done; 1411 cbp->bio_caller1 = disk; 1412 cbp->bio_to = disk->d_consumer->provider; 1413 } 1414 for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { 1415 bioq_remove(&queue, cbp); 1416 G_RAID3_LOGREQ(3, cbp, "Sending request."); 1417 disk = cbp->bio_caller1; 1418 cbp->bio_caller1 = NULL; 1419 cp = disk->d_consumer; 1420 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1421 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1422 cp->acr, cp->acw, cp->ace)); 1423 g_io_request(cbp, disk->d_consumer); 1424 } 1425 } 1426 1427 static void 1428 g_raid3_start(struct bio *bp) 1429 { 1430 struct g_raid3_softc *sc; 1431 1432 sc = bp->bio_to->geom->softc; 1433 /* 1434 * If sc == NULL or there are no valid disks, provider's error 1435 * should be set and g_raid3_start() should not be called at all. 1436 */ 1437 KASSERT(sc != NULL && (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 1438 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE), 1439 ("Provider's error should be set (error=%d)(device=%s).", 1440 bp->bio_to->error, bp->bio_to->name)); 1441 G_RAID3_LOGREQ(3, bp, "Request received."); 1442 1443 switch (bp->bio_cmd) { 1444 case BIO_READ: 1445 case BIO_WRITE: 1446 case BIO_DELETE: 1447 break; 1448 case BIO_SPEEDUP: 1449 case BIO_FLUSH: 1450 g_raid3_flush(sc, bp); 1451 return; 1452 case BIO_GETATTR: 1453 default: 1454 g_io_deliver(bp, EOPNOTSUPP); 1455 return; 1456 } 1457 mtx_lock(&sc->sc_queue_mtx); 1458 bioq_insert_tail(&sc->sc_queue, bp); 1459 mtx_unlock(&sc->sc_queue_mtx); 1460 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc); 1461 wakeup(sc); 1462 } 1463 1464 /* 1465 * Return TRUE if the given request is colliding with a in-progress 1466 * synchronization request. 1467 */ 1468 static int 1469 g_raid3_sync_collision(struct g_raid3_softc *sc, struct bio *bp) 1470 { 1471 struct g_raid3_disk *disk; 1472 struct bio *sbp; 1473 off_t rstart, rend, sstart, send; 1474 int i; 1475 1476 disk = sc->sc_syncdisk; 1477 if (disk == NULL) 1478 return (0); 1479 rstart = bp->bio_offset; 1480 rend = bp->bio_offset + bp->bio_length; 1481 for (i = 0; i < g_raid3_syncreqs; i++) { 1482 sbp = disk->d_sync.ds_bios[i]; 1483 if (sbp == NULL) 1484 continue; 1485 sstart = sbp->bio_offset; 1486 send = sbp->bio_length; 1487 if (sbp->bio_cmd == BIO_WRITE) { 1488 sstart *= sc->sc_ndisks - 1; 1489 send *= sc->sc_ndisks - 1; 1490 } 1491 send += sstart; 1492 if (rend > sstart && rstart < send) 1493 return (1); 1494 } 1495 return (0); 1496 } 1497 1498 /* 1499 * Return TRUE if the given sync request is colliding with a in-progress regular 1500 * request. 1501 */ 1502 static int 1503 g_raid3_regular_collision(struct g_raid3_softc *sc, struct bio *sbp) 1504 { 1505 off_t rstart, rend, sstart, send; 1506 struct bio *bp; 1507 1508 if (sc->sc_syncdisk == NULL) 1509 return (0); 1510 sstart = sbp->bio_offset; 1511 send = sstart + sbp->bio_length; 1512 TAILQ_FOREACH(bp, &sc->sc_inflight.queue, bio_queue) { 1513 rstart = bp->bio_offset; 1514 rend = bp->bio_offset + bp->bio_length; 1515 if (rend > sstart && rstart < send) 1516 return (1); 1517 } 1518 return (0); 1519 } 1520 1521 /* 1522 * Puts request onto delayed queue. 1523 */ 1524 static void 1525 g_raid3_regular_delay(struct g_raid3_softc *sc, struct bio *bp) 1526 { 1527 1528 G_RAID3_LOGREQ(2, bp, "Delaying request."); 1529 bioq_insert_head(&sc->sc_regular_delayed, bp); 1530 } 1531 1532 /* 1533 * Puts synchronization request onto delayed queue. 1534 */ 1535 static void 1536 g_raid3_sync_delay(struct g_raid3_softc *sc, struct bio *bp) 1537 { 1538 1539 G_RAID3_LOGREQ(2, bp, "Delaying synchronization request."); 1540 bioq_insert_tail(&sc->sc_sync_delayed, bp); 1541 } 1542 1543 /* 1544 * Releases delayed regular requests which don't collide anymore with sync 1545 * requests. 1546 */ 1547 static void 1548 g_raid3_regular_release(struct g_raid3_softc *sc) 1549 { 1550 struct bio *bp, *bp2; 1551 1552 TAILQ_FOREACH_SAFE(bp, &sc->sc_regular_delayed.queue, bio_queue, bp2) { 1553 if (g_raid3_sync_collision(sc, bp)) 1554 continue; 1555 bioq_remove(&sc->sc_regular_delayed, bp); 1556 G_RAID3_LOGREQ(2, bp, "Releasing delayed request (%p).", bp); 1557 mtx_lock(&sc->sc_queue_mtx); 1558 bioq_insert_head(&sc->sc_queue, bp); 1559 #if 0 1560 /* 1561 * wakeup() is not needed, because this function is called from 1562 * the worker thread. 1563 */ 1564 wakeup(&sc->sc_queue); 1565 #endif 1566 mtx_unlock(&sc->sc_queue_mtx); 1567 } 1568 } 1569 1570 /* 1571 * Releases delayed sync requests which don't collide anymore with regular 1572 * requests. 1573 */ 1574 static void 1575 g_raid3_sync_release(struct g_raid3_softc *sc) 1576 { 1577 struct bio *bp, *bp2; 1578 1579 TAILQ_FOREACH_SAFE(bp, &sc->sc_sync_delayed.queue, bio_queue, bp2) { 1580 if (g_raid3_regular_collision(sc, bp)) 1581 continue; 1582 bioq_remove(&sc->sc_sync_delayed, bp); 1583 G_RAID3_LOGREQ(2, bp, 1584 "Releasing delayed synchronization request."); 1585 g_io_request(bp, bp->bio_from); 1586 } 1587 } 1588 1589 /* 1590 * Handle synchronization requests. 1591 * Every synchronization request is two-steps process: first, READ request is 1592 * send to active provider and then WRITE request (with read data) to the provider 1593 * being synchronized. When WRITE is finished, new synchronization request is 1594 * send. 1595 */ 1596 static void 1597 g_raid3_sync_request(struct bio *bp) 1598 { 1599 struct g_raid3_softc *sc; 1600 struct g_raid3_disk *disk; 1601 1602 bp->bio_from->index--; 1603 sc = bp->bio_from->geom->softc; 1604 disk = bp->bio_from->private; 1605 if (disk == NULL) { 1606 sx_xunlock(&sc->sc_lock); /* Avoid recursion on sc_lock. */ 1607 g_topology_lock(); 1608 g_raid3_kill_consumer(sc, bp->bio_from); 1609 g_topology_unlock(); 1610 free(bp->bio_data, M_RAID3); 1611 g_destroy_bio(bp); 1612 sx_xlock(&sc->sc_lock); 1613 return; 1614 } 1615 1616 /* 1617 * Synchronization request. 1618 */ 1619 switch (bp->bio_cmd) { 1620 case BIO_READ: 1621 { 1622 struct g_consumer *cp; 1623 u_char *dst, *src; 1624 off_t left; 1625 u_int atom; 1626 1627 if (bp->bio_error != 0) { 1628 G_RAID3_LOGREQ(0, bp, 1629 "Synchronization request failed (error=%d).", 1630 bp->bio_error); 1631 g_destroy_bio(bp); 1632 return; 1633 } 1634 G_RAID3_LOGREQ(3, bp, "Synchronization request finished."); 1635 atom = sc->sc_sectorsize / (sc->sc_ndisks - 1); 1636 dst = src = bp->bio_data; 1637 if (disk->d_no == sc->sc_ndisks - 1) { 1638 u_int n; 1639 1640 /* Parity component. */ 1641 for (left = bp->bio_length; left > 0; 1642 left -= sc->sc_sectorsize) { 1643 bcopy(src, dst, atom); 1644 src += atom; 1645 for (n = 1; n < sc->sc_ndisks - 1; n++) { 1646 g_raid3_xor(src, dst, atom); 1647 src += atom; 1648 } 1649 dst += atom; 1650 } 1651 } else { 1652 /* Regular component. */ 1653 src += atom * disk->d_no; 1654 for (left = bp->bio_length; left > 0; 1655 left -= sc->sc_sectorsize) { 1656 bcopy(src, dst, atom); 1657 src += sc->sc_sectorsize; 1658 dst += atom; 1659 } 1660 } 1661 bp->bio_driver1 = bp->bio_driver2 = NULL; 1662 bp->bio_pflags = 0; 1663 bp->bio_offset /= sc->sc_ndisks - 1; 1664 bp->bio_length /= sc->sc_ndisks - 1; 1665 bp->bio_cmd = BIO_WRITE; 1666 bp->bio_cflags = 0; 1667 bp->bio_children = bp->bio_inbed = 0; 1668 cp = disk->d_consumer; 1669 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1670 ("Consumer %s not opened (r%dw%de%d).", cp->provider->name, 1671 cp->acr, cp->acw, cp->ace)); 1672 cp->index++; 1673 g_io_request(bp, cp); 1674 return; 1675 } 1676 case BIO_WRITE: 1677 { 1678 struct g_raid3_disk_sync *sync; 1679 off_t boffset, moffset; 1680 void *data; 1681 int i; 1682 1683 if (bp->bio_error != 0) { 1684 G_RAID3_LOGREQ(0, bp, 1685 "Synchronization request failed (error=%d).", 1686 bp->bio_error); 1687 g_destroy_bio(bp); 1688 sc->sc_bump_id |= G_RAID3_BUMP_GENID; 1689 g_raid3_event_send(disk, 1690 G_RAID3_DISK_STATE_DISCONNECTED, 1691 G_RAID3_EVENT_DONTWAIT); 1692 return; 1693 } 1694 G_RAID3_LOGREQ(3, bp, "Synchronization request finished."); 1695 sync = &disk->d_sync; 1696 if (sync->ds_offset == sc->sc_mediasize / (sc->sc_ndisks - 1) || 1697 sync->ds_consumer == NULL || 1698 (sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 1699 /* Don't send more synchronization requests. */ 1700 sync->ds_inflight--; 1701 if (sync->ds_bios != NULL) { 1702 i = (int)(uintptr_t)bp->bio_caller1; 1703 sync->ds_bios[i] = NULL; 1704 } 1705 free(bp->bio_data, M_RAID3); 1706 g_destroy_bio(bp); 1707 if (sync->ds_inflight > 0) 1708 return; 1709 if (sync->ds_consumer == NULL || 1710 (sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 1711 return; 1712 } 1713 /* 1714 * Disk up-to-date, activate it. 1715 */ 1716 g_raid3_event_send(disk, G_RAID3_DISK_STATE_ACTIVE, 1717 G_RAID3_EVENT_DONTWAIT); 1718 return; 1719 } 1720 1721 /* Send next synchronization request. */ 1722 data = bp->bio_data; 1723 g_reset_bio(bp); 1724 bp->bio_cmd = BIO_READ; 1725 bp->bio_offset = sync->ds_offset * (sc->sc_ndisks - 1); 1726 bp->bio_length = MIN(MAXPHYS, sc->sc_mediasize - bp->bio_offset); 1727 sync->ds_offset += bp->bio_length / (sc->sc_ndisks - 1); 1728 bp->bio_done = g_raid3_sync_done; 1729 bp->bio_data = data; 1730 bp->bio_from = sync->ds_consumer; 1731 bp->bio_to = sc->sc_provider; 1732 G_RAID3_LOGREQ(3, bp, "Sending synchronization request."); 1733 sync->ds_consumer->index++; 1734 /* 1735 * Delay the request if it is colliding with a regular request. 1736 */ 1737 if (g_raid3_regular_collision(sc, bp)) 1738 g_raid3_sync_delay(sc, bp); 1739 else 1740 g_io_request(bp, sync->ds_consumer); 1741 1742 /* Release delayed requests if possible. */ 1743 g_raid3_regular_release(sc); 1744 1745 /* Find the smallest offset. */ 1746 moffset = sc->sc_mediasize; 1747 for (i = 0; i < g_raid3_syncreqs; i++) { 1748 bp = sync->ds_bios[i]; 1749 boffset = bp->bio_offset; 1750 if (bp->bio_cmd == BIO_WRITE) 1751 boffset *= sc->sc_ndisks - 1; 1752 if (boffset < moffset) 1753 moffset = boffset; 1754 } 1755 if (sync->ds_offset_done + (MAXPHYS * 100) < moffset) { 1756 /* Update offset_done on every 100 blocks. */ 1757 sync->ds_offset_done = moffset; 1758 g_raid3_update_metadata(disk); 1759 } 1760 return; 1761 } 1762 default: 1763 KASSERT(1 == 0, ("Invalid command here: %u (device=%s)", 1764 bp->bio_cmd, sc->sc_name)); 1765 break; 1766 } 1767 } 1768 1769 static int 1770 g_raid3_register_request(struct bio *pbp) 1771 { 1772 struct g_raid3_softc *sc; 1773 struct g_raid3_disk *disk; 1774 struct g_consumer *cp; 1775 struct bio *cbp, *tmpbp; 1776 off_t offset, length; 1777 u_int n, ndisks; 1778 int round_robin, verify; 1779 1780 ndisks = 0; 1781 sc = pbp->bio_to->geom->softc; 1782 if ((pbp->bio_cflags & G_RAID3_BIO_CFLAG_REGSYNC) != 0 && 1783 sc->sc_syncdisk == NULL) { 1784 g_io_deliver(pbp, EIO); 1785 return (0); 1786 } 1787 g_raid3_init_bio(pbp); 1788 length = pbp->bio_length / (sc->sc_ndisks - 1); 1789 offset = pbp->bio_offset / (sc->sc_ndisks - 1); 1790 round_robin = verify = 0; 1791 switch (pbp->bio_cmd) { 1792 case BIO_READ: 1793 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_VERIFY) != 0 && 1794 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1795 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_VERIFY; 1796 verify = 1; 1797 ndisks = sc->sc_ndisks; 1798 } else { 1799 verify = 0; 1800 ndisks = sc->sc_ndisks - 1; 1801 } 1802 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0 && 1803 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 1804 round_robin = 1; 1805 } else { 1806 round_robin = 0; 1807 } 1808 KASSERT(!round_robin || !verify, 1809 ("ROUND-ROBIN and VERIFY are mutually exclusive.")); 1810 pbp->bio_driver2 = &sc->sc_disks[sc->sc_ndisks - 1]; 1811 break; 1812 case BIO_WRITE: 1813 case BIO_DELETE: 1814 /* 1815 * Delay the request if it is colliding with a synchronization 1816 * request. 1817 */ 1818 if (g_raid3_sync_collision(sc, pbp)) { 1819 g_raid3_regular_delay(sc, pbp); 1820 return (0); 1821 } 1822 1823 if (sc->sc_idle) 1824 g_raid3_unidle(sc); 1825 else 1826 sc->sc_last_write = time_uptime; 1827 1828 ndisks = sc->sc_ndisks; 1829 break; 1830 } 1831 for (n = 0; n < ndisks; n++) { 1832 disk = &sc->sc_disks[n]; 1833 cbp = g_raid3_clone_bio(sc, pbp); 1834 if (cbp == NULL) { 1835 while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) 1836 g_raid3_destroy_bio(sc, cbp); 1837 /* 1838 * To prevent deadlock, we must run back up 1839 * with the ENOMEM for failed requests of any 1840 * of our consumers. Our own sync requests 1841 * can stick around, as they are finite. 1842 */ 1843 if ((pbp->bio_cflags & 1844 G_RAID3_BIO_CFLAG_REGULAR) != 0) { 1845 g_io_deliver(pbp, ENOMEM); 1846 return (0); 1847 } 1848 return (ENOMEM); 1849 } 1850 cbp->bio_offset = offset; 1851 cbp->bio_length = length; 1852 cbp->bio_done = g_raid3_done; 1853 switch (pbp->bio_cmd) { 1854 case BIO_READ: 1855 if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) { 1856 /* 1857 * Replace invalid component with the parity 1858 * component. 1859 */ 1860 disk = &sc->sc_disks[sc->sc_ndisks - 1]; 1861 cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1862 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED; 1863 } else if (round_robin && 1864 disk->d_no == sc->sc_round_robin) { 1865 /* 1866 * In round-robin mode skip one data component 1867 * and use parity component when reading. 1868 */ 1869 pbp->bio_driver2 = disk; 1870 disk = &sc->sc_disks[sc->sc_ndisks - 1]; 1871 cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1872 sc->sc_round_robin++; 1873 round_robin = 0; 1874 } else if (verify && disk->d_no == sc->sc_ndisks - 1) { 1875 cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY; 1876 } 1877 break; 1878 case BIO_WRITE: 1879 case BIO_DELETE: 1880 if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 1881 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 1882 if (n == ndisks - 1) { 1883 /* 1884 * Active parity component, mark it as such. 1885 */ 1886 cbp->bio_cflags |= 1887 G_RAID3_BIO_CFLAG_PARITY; 1888 } 1889 } else { 1890 pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED; 1891 if (n == ndisks - 1) { 1892 /* 1893 * Parity component is not connected, 1894 * so destroy its request. 1895 */ 1896 pbp->bio_pflags |= 1897 G_RAID3_BIO_PFLAG_NOPARITY; 1898 g_raid3_destroy_bio(sc, cbp); 1899 cbp = NULL; 1900 } else { 1901 cbp->bio_cflags |= 1902 G_RAID3_BIO_CFLAG_NODISK; 1903 disk = NULL; 1904 } 1905 } 1906 break; 1907 } 1908 if (cbp != NULL) 1909 cbp->bio_caller2 = disk; 1910 } 1911 switch (pbp->bio_cmd) { 1912 case BIO_READ: 1913 if (round_robin) { 1914 /* 1915 * If we are in round-robin mode and 'round_robin' is 1916 * still 1, it means, that we skipped parity component 1917 * for this read and must reset sc_round_robin field. 1918 */ 1919 sc->sc_round_robin = 0; 1920 } 1921 G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) { 1922 disk = cbp->bio_caller2; 1923 cp = disk->d_consumer; 1924 cbp->bio_to = cp->provider; 1925 G_RAID3_LOGREQ(3, cbp, "Sending request."); 1926 KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1, 1927 ("Consumer %s not opened (r%dw%de%d).", 1928 cp->provider->name, cp->acr, cp->acw, cp->ace)); 1929 cp->index++; 1930 g_io_request(cbp, cp); 1931 } 1932 break; 1933 case BIO_WRITE: 1934 case BIO_DELETE: 1935 /* 1936 * Put request onto inflight queue, so we can check if new 1937 * synchronization requests don't collide with it. 1938 */ 1939 bioq_insert_tail(&sc->sc_inflight, pbp); 1940 1941 /* 1942 * Bump syncid on first write. 1943 */ 1944 if ((sc->sc_bump_id & G_RAID3_BUMP_SYNCID) != 0) { 1945 sc->sc_bump_id &= ~G_RAID3_BUMP_SYNCID; 1946 g_raid3_bump_syncid(sc); 1947 } 1948 g_raid3_scatter(pbp); 1949 break; 1950 } 1951 return (0); 1952 } 1953 1954 static int 1955 g_raid3_can_destroy(struct g_raid3_softc *sc) 1956 { 1957 struct g_geom *gp; 1958 struct g_consumer *cp; 1959 1960 g_topology_assert(); 1961 gp = sc->sc_geom; 1962 if (gp->softc == NULL) 1963 return (1); 1964 LIST_FOREACH(cp, &gp->consumer, consumer) { 1965 if (g_raid3_is_busy(sc, cp)) 1966 return (0); 1967 } 1968 gp = sc->sc_sync.ds_geom; 1969 LIST_FOREACH(cp, &gp->consumer, consumer) { 1970 if (g_raid3_is_busy(sc, cp)) 1971 return (0); 1972 } 1973 G_RAID3_DEBUG(2, "No I/O requests for %s, it can be destroyed.", 1974 sc->sc_name); 1975 return (1); 1976 } 1977 1978 static int 1979 g_raid3_try_destroy(struct g_raid3_softc *sc) 1980 { 1981 1982 g_topology_assert_not(); 1983 sx_assert(&sc->sc_lock, SX_XLOCKED); 1984 1985 if (sc->sc_rootmount != NULL) { 1986 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__, 1987 sc->sc_rootmount); 1988 root_mount_rel(sc->sc_rootmount); 1989 sc->sc_rootmount = NULL; 1990 } 1991 1992 g_topology_lock(); 1993 if (!g_raid3_can_destroy(sc)) { 1994 g_topology_unlock(); 1995 return (0); 1996 } 1997 sc->sc_geom->softc = NULL; 1998 sc->sc_sync.ds_geom->softc = NULL; 1999 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_WAIT) != 0) { 2000 g_topology_unlock(); 2001 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, 2002 &sc->sc_worker); 2003 /* Unlock sc_lock here, as it can be destroyed after wakeup. */ 2004 sx_xunlock(&sc->sc_lock); 2005 wakeup(&sc->sc_worker); 2006 sc->sc_worker = NULL; 2007 } else { 2008 g_topology_unlock(); 2009 g_raid3_destroy_device(sc); 2010 free(sc->sc_disks, M_RAID3); 2011 free(sc, M_RAID3); 2012 } 2013 return (1); 2014 } 2015 2016 /* 2017 * Worker thread. 2018 */ 2019 static void 2020 g_raid3_worker(void *arg) 2021 { 2022 struct g_raid3_softc *sc; 2023 struct g_raid3_event *ep; 2024 struct bio *bp; 2025 int timeout; 2026 2027 sc = arg; 2028 thread_lock(curthread); 2029 sched_prio(curthread, PRIBIO); 2030 thread_unlock(curthread); 2031 2032 sx_xlock(&sc->sc_lock); 2033 for (;;) { 2034 G_RAID3_DEBUG(5, "%s: Let's see...", __func__); 2035 /* 2036 * First take a look at events. 2037 * This is important to handle events before any I/O requests. 2038 */ 2039 ep = g_raid3_event_get(sc); 2040 if (ep != NULL) { 2041 g_raid3_event_remove(sc, ep); 2042 if ((ep->e_flags & G_RAID3_EVENT_DEVICE) != 0) { 2043 /* Update only device status. */ 2044 G_RAID3_DEBUG(3, 2045 "Running event for device %s.", 2046 sc->sc_name); 2047 ep->e_error = 0; 2048 g_raid3_update_device(sc, 1); 2049 } else { 2050 /* Update disk status. */ 2051 G_RAID3_DEBUG(3, "Running event for disk %s.", 2052 g_raid3_get_diskname(ep->e_disk)); 2053 ep->e_error = g_raid3_update_disk(ep->e_disk, 2054 ep->e_state); 2055 if (ep->e_error == 0) 2056 g_raid3_update_device(sc, 0); 2057 } 2058 if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) { 2059 KASSERT(ep->e_error == 0, 2060 ("Error cannot be handled.")); 2061 g_raid3_event_free(ep); 2062 } else { 2063 ep->e_flags |= G_RAID3_EVENT_DONE; 2064 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, 2065 ep); 2066 mtx_lock(&sc->sc_events_mtx); 2067 wakeup(ep); 2068 mtx_unlock(&sc->sc_events_mtx); 2069 } 2070 if ((sc->sc_flags & 2071 G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 2072 if (g_raid3_try_destroy(sc)) { 2073 curthread->td_pflags &= ~TDP_GEOM; 2074 G_RAID3_DEBUG(1, "Thread exiting."); 2075 kproc_exit(0); 2076 } 2077 } 2078 G_RAID3_DEBUG(5, "%s: I'm here 1.", __func__); 2079 continue; 2080 } 2081 /* 2082 * Check if we can mark array as CLEAN and if we can't take 2083 * how much seconds should we wait. 2084 */ 2085 timeout = g_raid3_idle(sc, -1); 2086 /* 2087 * Now I/O requests. 2088 */ 2089 /* Get first request from the queue. */ 2090 mtx_lock(&sc->sc_queue_mtx); 2091 bp = bioq_first(&sc->sc_queue); 2092 if (bp == NULL) { 2093 if ((sc->sc_flags & 2094 G_RAID3_DEVICE_FLAG_DESTROY) != 0) { 2095 mtx_unlock(&sc->sc_queue_mtx); 2096 if (g_raid3_try_destroy(sc)) { 2097 curthread->td_pflags &= ~TDP_GEOM; 2098 G_RAID3_DEBUG(1, "Thread exiting."); 2099 kproc_exit(0); 2100 } 2101 mtx_lock(&sc->sc_queue_mtx); 2102 } 2103 sx_xunlock(&sc->sc_lock); 2104 /* 2105 * XXX: We can miss an event here, because an event 2106 * can be added without sx-device-lock and without 2107 * mtx-queue-lock. Maybe I should just stop using 2108 * dedicated mutex for events synchronization and 2109 * stick with the queue lock? 2110 * The event will hang here until next I/O request 2111 * or next event is received. 2112 */ 2113 MSLEEP(sc, &sc->sc_queue_mtx, PRIBIO | PDROP, "r3:w1", 2114 timeout * hz); 2115 sx_xlock(&sc->sc_lock); 2116 G_RAID3_DEBUG(5, "%s: I'm here 4.", __func__); 2117 continue; 2118 } 2119 process: 2120 bioq_remove(&sc->sc_queue, bp); 2121 mtx_unlock(&sc->sc_queue_mtx); 2122 2123 if (bp->bio_from->geom == sc->sc_sync.ds_geom && 2124 (bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) != 0) { 2125 g_raid3_sync_request(bp); /* READ */ 2126 } else if (bp->bio_to != sc->sc_provider) { 2127 if ((bp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) != 0) 2128 g_raid3_regular_request(bp); 2129 else if ((bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) != 0) 2130 g_raid3_sync_request(bp); /* WRITE */ 2131 else { 2132 KASSERT(0, 2133 ("Invalid request cflags=0x%hx to=%s.", 2134 bp->bio_cflags, bp->bio_to->name)); 2135 } 2136 } else if (g_raid3_register_request(bp) != 0) { 2137 mtx_lock(&sc->sc_queue_mtx); 2138 bioq_insert_head(&sc->sc_queue, bp); 2139 /* 2140 * We are short in memory, let see if there are finished 2141 * request we can free. 2142 */ 2143 TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { 2144 if (bp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) 2145 goto process; 2146 } 2147 /* 2148 * No finished regular request, so at least keep 2149 * synchronization running. 2150 */ 2151 TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { 2152 if (bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) 2153 goto process; 2154 } 2155 sx_xunlock(&sc->sc_lock); 2156 MSLEEP(&sc->sc_queue, &sc->sc_queue_mtx, PRIBIO | PDROP, 2157 "r3:lowmem", hz / 10); 2158 sx_xlock(&sc->sc_lock); 2159 } 2160 G_RAID3_DEBUG(5, "%s: I'm here 9.", __func__); 2161 } 2162 } 2163 2164 static void 2165 g_raid3_update_idle(struct g_raid3_softc *sc, struct g_raid3_disk *disk) 2166 { 2167 2168 sx_assert(&sc->sc_lock, SX_LOCKED); 2169 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0) 2170 return; 2171 if (!sc->sc_idle && (disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) == 0) { 2172 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as dirty.", 2173 g_raid3_get_diskname(disk), sc->sc_name); 2174 disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY; 2175 } else if (sc->sc_idle && 2176 (disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) != 0) { 2177 G_RAID3_DEBUG(1, "Disk %s (device %s) marked as clean.", 2178 g_raid3_get_diskname(disk), sc->sc_name); 2179 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2180 } 2181 } 2182 2183 static void 2184 g_raid3_sync_start(struct g_raid3_softc *sc) 2185 { 2186 struct g_raid3_disk *disk; 2187 struct g_consumer *cp; 2188 struct bio *bp; 2189 int error; 2190 u_int n; 2191 2192 g_topology_assert_not(); 2193 sx_assert(&sc->sc_lock, SX_XLOCKED); 2194 2195 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED, 2196 ("Device not in DEGRADED state (%s, %u).", sc->sc_name, 2197 sc->sc_state)); 2198 KASSERT(sc->sc_syncdisk == NULL, ("Syncdisk is not NULL (%s, %u).", 2199 sc->sc_name, sc->sc_state)); 2200 disk = NULL; 2201 for (n = 0; n < sc->sc_ndisks; n++) { 2202 if (sc->sc_disks[n].d_state != G_RAID3_DISK_STATE_SYNCHRONIZING) 2203 continue; 2204 disk = &sc->sc_disks[n]; 2205 break; 2206 } 2207 if (disk == NULL) 2208 return; 2209 2210 sx_xunlock(&sc->sc_lock); 2211 g_topology_lock(); 2212 cp = g_new_consumer(sc->sc_sync.ds_geom); 2213 error = g_attach(cp, sc->sc_provider); 2214 KASSERT(error == 0, 2215 ("Cannot attach to %s (error=%d).", sc->sc_name, error)); 2216 error = g_access(cp, 1, 0, 0); 2217 KASSERT(error == 0, ("Cannot open %s (error=%d).", sc->sc_name, error)); 2218 g_topology_unlock(); 2219 sx_xlock(&sc->sc_lock); 2220 2221 G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s.", sc->sc_name, 2222 g_raid3_get_diskname(disk)); 2223 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) == 0) 2224 disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY; 2225 KASSERT(disk->d_sync.ds_consumer == NULL, 2226 ("Sync consumer already exists (device=%s, disk=%s).", 2227 sc->sc_name, g_raid3_get_diskname(disk))); 2228 2229 disk->d_sync.ds_consumer = cp; 2230 disk->d_sync.ds_consumer->private = disk; 2231 disk->d_sync.ds_consumer->index = 0; 2232 sc->sc_syncdisk = disk; 2233 2234 /* 2235 * Allocate memory for synchronization bios and initialize them. 2236 */ 2237 disk->d_sync.ds_bios = malloc(sizeof(struct bio *) * g_raid3_syncreqs, 2238 M_RAID3, M_WAITOK); 2239 for (n = 0; n < g_raid3_syncreqs; n++) { 2240 bp = g_alloc_bio(); 2241 disk->d_sync.ds_bios[n] = bp; 2242 bp->bio_parent = NULL; 2243 bp->bio_cmd = BIO_READ; 2244 bp->bio_data = malloc(MAXPHYS, M_RAID3, M_WAITOK); 2245 bp->bio_cflags = 0; 2246 bp->bio_offset = disk->d_sync.ds_offset * (sc->sc_ndisks - 1); 2247 bp->bio_length = MIN(MAXPHYS, sc->sc_mediasize - bp->bio_offset); 2248 disk->d_sync.ds_offset += bp->bio_length / (sc->sc_ndisks - 1); 2249 bp->bio_done = g_raid3_sync_done; 2250 bp->bio_from = disk->d_sync.ds_consumer; 2251 bp->bio_to = sc->sc_provider; 2252 bp->bio_caller1 = (void *)(uintptr_t)n; 2253 } 2254 2255 /* Set the number of in-flight synchronization requests. */ 2256 disk->d_sync.ds_inflight = g_raid3_syncreqs; 2257 2258 /* 2259 * Fire off first synchronization requests. 2260 */ 2261 for (n = 0; n < g_raid3_syncreqs; n++) { 2262 bp = disk->d_sync.ds_bios[n]; 2263 G_RAID3_LOGREQ(3, bp, "Sending synchronization request."); 2264 disk->d_sync.ds_consumer->index++; 2265 /* 2266 * Delay the request if it is colliding with a regular request. 2267 */ 2268 if (g_raid3_regular_collision(sc, bp)) 2269 g_raid3_sync_delay(sc, bp); 2270 else 2271 g_io_request(bp, disk->d_sync.ds_consumer); 2272 } 2273 } 2274 2275 /* 2276 * Stop synchronization process. 2277 * type: 0 - synchronization finished 2278 * 1 - synchronization stopped 2279 */ 2280 static void 2281 g_raid3_sync_stop(struct g_raid3_softc *sc, int type) 2282 { 2283 struct g_raid3_disk *disk; 2284 struct g_consumer *cp; 2285 2286 g_topology_assert_not(); 2287 sx_assert(&sc->sc_lock, SX_LOCKED); 2288 2289 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED, 2290 ("Device not in DEGRADED state (%s, %u).", sc->sc_name, 2291 sc->sc_state)); 2292 disk = sc->sc_syncdisk; 2293 sc->sc_syncdisk = NULL; 2294 KASSERT(disk != NULL, ("No disk was synchronized (%s).", sc->sc_name)); 2295 KASSERT(disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING, 2296 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2297 g_raid3_disk_state2str(disk->d_state))); 2298 if (disk->d_sync.ds_consumer == NULL) 2299 return; 2300 2301 if (type == 0) { 2302 G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s finished.", 2303 sc->sc_name, g_raid3_get_diskname(disk)); 2304 } else /* if (type == 1) */ { 2305 G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s stopped.", 2306 sc->sc_name, g_raid3_get_diskname(disk)); 2307 } 2308 free(disk->d_sync.ds_bios, M_RAID3); 2309 disk->d_sync.ds_bios = NULL; 2310 cp = disk->d_sync.ds_consumer; 2311 disk->d_sync.ds_consumer = NULL; 2312 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2313 sx_xunlock(&sc->sc_lock); /* Avoid recursion on sc_lock. */ 2314 g_topology_lock(); 2315 g_raid3_kill_consumer(sc, cp); 2316 g_topology_unlock(); 2317 sx_xlock(&sc->sc_lock); 2318 } 2319 2320 static void 2321 g_raid3_launch_provider(struct g_raid3_softc *sc) 2322 { 2323 struct g_provider *pp; 2324 struct g_raid3_disk *disk; 2325 int n; 2326 2327 sx_assert(&sc->sc_lock, SX_LOCKED); 2328 2329 g_topology_lock(); 2330 pp = g_new_providerf(sc->sc_geom, "raid3/%s", sc->sc_name); 2331 pp->mediasize = sc->sc_mediasize; 2332 pp->sectorsize = sc->sc_sectorsize; 2333 pp->stripesize = 0; 2334 pp->stripeoffset = 0; 2335 for (n = 0; n < sc->sc_ndisks; n++) { 2336 disk = &sc->sc_disks[n]; 2337 if (disk->d_consumer && disk->d_consumer->provider && 2338 disk->d_consumer->provider->stripesize > pp->stripesize) { 2339 pp->stripesize = disk->d_consumer->provider->stripesize; 2340 pp->stripeoffset = disk->d_consumer->provider->stripeoffset; 2341 } 2342 } 2343 pp->stripesize *= sc->sc_ndisks - 1; 2344 pp->stripeoffset *= sc->sc_ndisks - 1; 2345 sc->sc_provider = pp; 2346 g_error_provider(pp, 0); 2347 g_topology_unlock(); 2348 G_RAID3_DEBUG(0, "Device %s launched (%u/%u).", pp->name, 2349 g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE), sc->sc_ndisks); 2350 2351 if (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED) 2352 g_raid3_sync_start(sc); 2353 } 2354 2355 static void 2356 g_raid3_destroy_provider(struct g_raid3_softc *sc) 2357 { 2358 struct bio *bp; 2359 2360 g_topology_assert_not(); 2361 KASSERT(sc->sc_provider != NULL, ("NULL provider (device=%s).", 2362 sc->sc_name)); 2363 2364 g_topology_lock(); 2365 g_error_provider(sc->sc_provider, ENXIO); 2366 mtx_lock(&sc->sc_queue_mtx); 2367 while ((bp = bioq_first(&sc->sc_queue)) != NULL) { 2368 bioq_remove(&sc->sc_queue, bp); 2369 g_io_deliver(bp, ENXIO); 2370 } 2371 mtx_unlock(&sc->sc_queue_mtx); 2372 G_RAID3_DEBUG(0, "Device %s: provider %s destroyed.", sc->sc_name, 2373 sc->sc_provider->name); 2374 g_wither_provider(sc->sc_provider, ENXIO); 2375 g_topology_unlock(); 2376 sc->sc_provider = NULL; 2377 if (sc->sc_syncdisk != NULL) 2378 g_raid3_sync_stop(sc, 1); 2379 } 2380 2381 static void 2382 g_raid3_go(void *arg) 2383 { 2384 struct g_raid3_softc *sc; 2385 2386 sc = arg; 2387 G_RAID3_DEBUG(0, "Force device %s start due to timeout.", sc->sc_name); 2388 g_raid3_event_send(sc, 0, 2389 G_RAID3_EVENT_DONTWAIT | G_RAID3_EVENT_DEVICE); 2390 } 2391 2392 static u_int 2393 g_raid3_determine_state(struct g_raid3_disk *disk) 2394 { 2395 struct g_raid3_softc *sc; 2396 u_int state; 2397 2398 sc = disk->d_softc; 2399 if (sc->sc_syncid == disk->d_sync.ds_syncid) { 2400 if ((disk->d_flags & 2401 G_RAID3_DISK_FLAG_SYNCHRONIZING) == 0) { 2402 /* Disk does not need synchronization. */ 2403 state = G_RAID3_DISK_STATE_ACTIVE; 2404 } else { 2405 if ((sc->sc_flags & 2406 G_RAID3_DEVICE_FLAG_NOAUTOSYNC) == 0 || 2407 (disk->d_flags & 2408 G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) { 2409 /* 2410 * We can start synchronization from 2411 * the stored offset. 2412 */ 2413 state = G_RAID3_DISK_STATE_SYNCHRONIZING; 2414 } else { 2415 state = G_RAID3_DISK_STATE_STALE; 2416 } 2417 } 2418 } else if (disk->d_sync.ds_syncid < sc->sc_syncid) { 2419 /* 2420 * Reset all synchronization data for this disk, 2421 * because if it even was synchronized, it was 2422 * synchronized to disks with different syncid. 2423 */ 2424 disk->d_flags |= G_RAID3_DISK_FLAG_SYNCHRONIZING; 2425 disk->d_sync.ds_offset = 0; 2426 disk->d_sync.ds_offset_done = 0; 2427 disk->d_sync.ds_syncid = sc->sc_syncid; 2428 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) == 0 || 2429 (disk->d_flags & G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) { 2430 state = G_RAID3_DISK_STATE_SYNCHRONIZING; 2431 } else { 2432 state = G_RAID3_DISK_STATE_STALE; 2433 } 2434 } else /* if (sc->sc_syncid < disk->d_sync.ds_syncid) */ { 2435 /* 2436 * Not good, NOT GOOD! 2437 * It means that device was started on stale disks 2438 * and more fresh disk just arrive. 2439 * If there were writes, device is broken, sorry. 2440 * I think the best choice here is don't touch 2441 * this disk and inform the user loudly. 2442 */ 2443 G_RAID3_DEBUG(0, "Device %s was started before the freshest " 2444 "disk (%s) arrives!! It will not be connected to the " 2445 "running device.", sc->sc_name, 2446 g_raid3_get_diskname(disk)); 2447 g_raid3_destroy_disk(disk); 2448 state = G_RAID3_DISK_STATE_NONE; 2449 /* Return immediately, because disk was destroyed. */ 2450 return (state); 2451 } 2452 G_RAID3_DEBUG(3, "State for %s disk: %s.", 2453 g_raid3_get_diskname(disk), g_raid3_disk_state2str(state)); 2454 return (state); 2455 } 2456 2457 /* 2458 * Update device state. 2459 */ 2460 static void 2461 g_raid3_update_device(struct g_raid3_softc *sc, boolean_t force) 2462 { 2463 struct g_raid3_disk *disk; 2464 u_int state; 2465 2466 sx_assert(&sc->sc_lock, SX_XLOCKED); 2467 2468 switch (sc->sc_state) { 2469 case G_RAID3_DEVICE_STATE_STARTING: 2470 { 2471 u_int n, ndirty, ndisks, genid, syncid; 2472 2473 KASSERT(sc->sc_provider == NULL, 2474 ("Non-NULL provider in STARTING state (%s).", sc->sc_name)); 2475 /* 2476 * Are we ready? We are, if all disks are connected or 2477 * one disk is missing and 'force' is true. 2478 */ 2479 if (g_raid3_ndisks(sc, -1) + force == sc->sc_ndisks) { 2480 if (!force) 2481 callout_drain(&sc->sc_callout); 2482 } else { 2483 if (force) { 2484 /* 2485 * Timeout expired, so destroy device. 2486 */ 2487 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 2488 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", 2489 __LINE__, sc->sc_rootmount); 2490 root_mount_rel(sc->sc_rootmount); 2491 sc->sc_rootmount = NULL; 2492 } 2493 return; 2494 } 2495 2496 /* 2497 * Find the biggest genid. 2498 */ 2499 genid = 0; 2500 for (n = 0; n < sc->sc_ndisks; n++) { 2501 disk = &sc->sc_disks[n]; 2502 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2503 continue; 2504 if (disk->d_genid > genid) 2505 genid = disk->d_genid; 2506 } 2507 sc->sc_genid = genid; 2508 /* 2509 * Remove all disks without the biggest genid. 2510 */ 2511 for (n = 0; n < sc->sc_ndisks; n++) { 2512 disk = &sc->sc_disks[n]; 2513 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2514 continue; 2515 if (disk->d_genid < genid) { 2516 G_RAID3_DEBUG(0, 2517 "Component %s (device %s) broken, skipping.", 2518 g_raid3_get_diskname(disk), sc->sc_name); 2519 g_raid3_destroy_disk(disk); 2520 } 2521 } 2522 2523 /* 2524 * There must be at least 'sc->sc_ndisks - 1' components 2525 * with the same syncid and without SYNCHRONIZING flag. 2526 */ 2527 2528 /* 2529 * Find the biggest syncid, number of valid components and 2530 * number of dirty components. 2531 */ 2532 ndirty = ndisks = syncid = 0; 2533 for (n = 0; n < sc->sc_ndisks; n++) { 2534 disk = &sc->sc_disks[n]; 2535 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2536 continue; 2537 if ((disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) != 0) 2538 ndirty++; 2539 if (disk->d_sync.ds_syncid > syncid) { 2540 syncid = disk->d_sync.ds_syncid; 2541 ndisks = 0; 2542 } else if (disk->d_sync.ds_syncid < syncid) { 2543 continue; 2544 } 2545 if ((disk->d_flags & 2546 G_RAID3_DISK_FLAG_SYNCHRONIZING) != 0) { 2547 continue; 2548 } 2549 ndisks++; 2550 } 2551 /* 2552 * Do we have enough valid components? 2553 */ 2554 if (ndisks + 1 < sc->sc_ndisks) { 2555 G_RAID3_DEBUG(0, 2556 "Device %s is broken, too few valid components.", 2557 sc->sc_name); 2558 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 2559 return; 2560 } 2561 /* 2562 * If there is one DIRTY component and all disks are present, 2563 * mark it for synchronization. If there is more than one DIRTY 2564 * component, mark parity component for synchronization. 2565 */ 2566 if (ndisks == sc->sc_ndisks && ndirty == 1) { 2567 for (n = 0; n < sc->sc_ndisks; n++) { 2568 disk = &sc->sc_disks[n]; 2569 if ((disk->d_flags & 2570 G_RAID3_DISK_FLAG_DIRTY) == 0) { 2571 continue; 2572 } 2573 disk->d_flags |= 2574 G_RAID3_DISK_FLAG_SYNCHRONIZING; 2575 } 2576 } else if (ndisks == sc->sc_ndisks && ndirty > 1) { 2577 disk = &sc->sc_disks[sc->sc_ndisks - 1]; 2578 disk->d_flags |= G_RAID3_DISK_FLAG_SYNCHRONIZING; 2579 } 2580 2581 sc->sc_syncid = syncid; 2582 if (force) { 2583 /* Remember to bump syncid on first write. */ 2584 sc->sc_bump_id |= G_RAID3_BUMP_SYNCID; 2585 } 2586 if (ndisks == sc->sc_ndisks) 2587 state = G_RAID3_DEVICE_STATE_COMPLETE; 2588 else /* if (ndisks == sc->sc_ndisks - 1) */ 2589 state = G_RAID3_DEVICE_STATE_DEGRADED; 2590 G_RAID3_DEBUG(1, "Device %s state changed from %s to %s.", 2591 sc->sc_name, g_raid3_device_state2str(sc->sc_state), 2592 g_raid3_device_state2str(state)); 2593 sc->sc_state = state; 2594 for (n = 0; n < sc->sc_ndisks; n++) { 2595 disk = &sc->sc_disks[n]; 2596 if (disk->d_state == G_RAID3_DISK_STATE_NODISK) 2597 continue; 2598 state = g_raid3_determine_state(disk); 2599 g_raid3_event_send(disk, state, G_RAID3_EVENT_DONTWAIT); 2600 if (state == G_RAID3_DISK_STATE_STALE) 2601 sc->sc_bump_id |= G_RAID3_BUMP_SYNCID; 2602 } 2603 break; 2604 } 2605 case G_RAID3_DEVICE_STATE_DEGRADED: 2606 /* 2607 * Genid need to be bumped immediately, so do it here. 2608 */ 2609 if ((sc->sc_bump_id & G_RAID3_BUMP_GENID) != 0) { 2610 sc->sc_bump_id &= ~G_RAID3_BUMP_GENID; 2611 g_raid3_bump_genid(sc); 2612 } 2613 2614 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NEW) > 0) 2615 return; 2616 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) < 2617 sc->sc_ndisks - 1) { 2618 if (sc->sc_provider != NULL) 2619 g_raid3_destroy_provider(sc); 2620 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 2621 return; 2622 } 2623 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) == 2624 sc->sc_ndisks) { 2625 state = G_RAID3_DEVICE_STATE_COMPLETE; 2626 G_RAID3_DEBUG(1, 2627 "Device %s state changed from %s to %s.", 2628 sc->sc_name, g_raid3_device_state2str(sc->sc_state), 2629 g_raid3_device_state2str(state)); 2630 sc->sc_state = state; 2631 } 2632 if (sc->sc_provider == NULL) 2633 g_raid3_launch_provider(sc); 2634 if (sc->sc_rootmount != NULL) { 2635 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__, 2636 sc->sc_rootmount); 2637 root_mount_rel(sc->sc_rootmount); 2638 sc->sc_rootmount = NULL; 2639 } 2640 break; 2641 case G_RAID3_DEVICE_STATE_COMPLETE: 2642 /* 2643 * Genid need to be bumped immediately, so do it here. 2644 */ 2645 if ((sc->sc_bump_id & G_RAID3_BUMP_GENID) != 0) { 2646 sc->sc_bump_id &= ~G_RAID3_BUMP_GENID; 2647 g_raid3_bump_genid(sc); 2648 } 2649 2650 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NEW) > 0) 2651 return; 2652 KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) >= 2653 sc->sc_ndisks - 1, 2654 ("Too few ACTIVE components in COMPLETE state (device %s).", 2655 sc->sc_name)); 2656 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) == 2657 sc->sc_ndisks - 1) { 2658 state = G_RAID3_DEVICE_STATE_DEGRADED; 2659 G_RAID3_DEBUG(1, 2660 "Device %s state changed from %s to %s.", 2661 sc->sc_name, g_raid3_device_state2str(sc->sc_state), 2662 g_raid3_device_state2str(state)); 2663 sc->sc_state = state; 2664 } 2665 if (sc->sc_provider == NULL) 2666 g_raid3_launch_provider(sc); 2667 if (sc->sc_rootmount != NULL) { 2668 G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__, 2669 sc->sc_rootmount); 2670 root_mount_rel(sc->sc_rootmount); 2671 sc->sc_rootmount = NULL; 2672 } 2673 break; 2674 default: 2675 KASSERT(1 == 0, ("Wrong device state (%s, %s).", sc->sc_name, 2676 g_raid3_device_state2str(sc->sc_state))); 2677 break; 2678 } 2679 } 2680 2681 /* 2682 * Update disk state and device state if needed. 2683 */ 2684 #define DISK_STATE_CHANGED() G_RAID3_DEBUG(1, \ 2685 "Disk %s state changed from %s to %s (device %s).", \ 2686 g_raid3_get_diskname(disk), \ 2687 g_raid3_disk_state2str(disk->d_state), \ 2688 g_raid3_disk_state2str(state), sc->sc_name) 2689 static int 2690 g_raid3_update_disk(struct g_raid3_disk *disk, u_int state) 2691 { 2692 struct g_raid3_softc *sc; 2693 2694 sc = disk->d_softc; 2695 sx_assert(&sc->sc_lock, SX_XLOCKED); 2696 2697 again: 2698 G_RAID3_DEBUG(3, "Changing disk %s state from %s to %s.", 2699 g_raid3_get_diskname(disk), g_raid3_disk_state2str(disk->d_state), 2700 g_raid3_disk_state2str(state)); 2701 switch (state) { 2702 case G_RAID3_DISK_STATE_NEW: 2703 /* 2704 * Possible scenarios: 2705 * 1. New disk arrive. 2706 */ 2707 /* Previous state should be NONE. */ 2708 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NONE, 2709 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2710 g_raid3_disk_state2str(disk->d_state))); 2711 DISK_STATE_CHANGED(); 2712 2713 disk->d_state = state; 2714 G_RAID3_DEBUG(1, "Device %s: provider %s detected.", 2715 sc->sc_name, g_raid3_get_diskname(disk)); 2716 if (sc->sc_state == G_RAID3_DEVICE_STATE_STARTING) 2717 break; 2718 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2719 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2720 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2721 g_raid3_device_state2str(sc->sc_state), 2722 g_raid3_get_diskname(disk), 2723 g_raid3_disk_state2str(disk->d_state))); 2724 state = g_raid3_determine_state(disk); 2725 if (state != G_RAID3_DISK_STATE_NONE) 2726 goto again; 2727 break; 2728 case G_RAID3_DISK_STATE_ACTIVE: 2729 /* 2730 * Possible scenarios: 2731 * 1. New disk does not need synchronization. 2732 * 2. Synchronization process finished successfully. 2733 */ 2734 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2735 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2736 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2737 g_raid3_device_state2str(sc->sc_state), 2738 g_raid3_get_diskname(disk), 2739 g_raid3_disk_state2str(disk->d_state))); 2740 /* Previous state should be NEW or SYNCHRONIZING. */ 2741 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW || 2742 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING, 2743 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2744 g_raid3_disk_state2str(disk->d_state))); 2745 DISK_STATE_CHANGED(); 2746 2747 if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 2748 disk->d_flags &= ~G_RAID3_DISK_FLAG_SYNCHRONIZING; 2749 disk->d_flags &= ~G_RAID3_DISK_FLAG_FORCE_SYNC; 2750 g_raid3_sync_stop(sc, 0); 2751 } 2752 disk->d_state = state; 2753 disk->d_sync.ds_offset = 0; 2754 disk->d_sync.ds_offset_done = 0; 2755 g_raid3_update_idle(sc, disk); 2756 g_raid3_update_metadata(disk); 2757 G_RAID3_DEBUG(1, "Device %s: provider %s activated.", 2758 sc->sc_name, g_raid3_get_diskname(disk)); 2759 break; 2760 case G_RAID3_DISK_STATE_STALE: 2761 /* 2762 * Possible scenarios: 2763 * 1. Stale disk was connected. 2764 */ 2765 /* Previous state should be NEW. */ 2766 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW, 2767 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2768 g_raid3_disk_state2str(disk->d_state))); 2769 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2770 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2771 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2772 g_raid3_device_state2str(sc->sc_state), 2773 g_raid3_get_diskname(disk), 2774 g_raid3_disk_state2str(disk->d_state))); 2775 /* 2776 * STALE state is only possible if device is marked 2777 * NOAUTOSYNC. 2778 */ 2779 KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) != 0, 2780 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2781 g_raid3_device_state2str(sc->sc_state), 2782 g_raid3_get_diskname(disk), 2783 g_raid3_disk_state2str(disk->d_state))); 2784 DISK_STATE_CHANGED(); 2785 2786 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2787 disk->d_state = state; 2788 g_raid3_update_metadata(disk); 2789 G_RAID3_DEBUG(0, "Device %s: provider %s is stale.", 2790 sc->sc_name, g_raid3_get_diskname(disk)); 2791 break; 2792 case G_RAID3_DISK_STATE_SYNCHRONIZING: 2793 /* 2794 * Possible scenarios: 2795 * 1. Disk which needs synchronization was connected. 2796 */ 2797 /* Previous state should be NEW. */ 2798 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW, 2799 ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk), 2800 g_raid3_disk_state2str(disk->d_state))); 2801 KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2802 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE, 2803 ("Wrong device state (%s, %s, %s, %s).", sc->sc_name, 2804 g_raid3_device_state2str(sc->sc_state), 2805 g_raid3_get_diskname(disk), 2806 g_raid3_disk_state2str(disk->d_state))); 2807 DISK_STATE_CHANGED(); 2808 2809 if (disk->d_state == G_RAID3_DISK_STATE_NEW) 2810 disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY; 2811 disk->d_state = state; 2812 if (sc->sc_provider != NULL) { 2813 g_raid3_sync_start(sc); 2814 g_raid3_update_metadata(disk); 2815 } 2816 break; 2817 case G_RAID3_DISK_STATE_DISCONNECTED: 2818 /* 2819 * Possible scenarios: 2820 * 1. Device wasn't running yet, but disk disappear. 2821 * 2. Disk was active and disapppear. 2822 * 3. Disk disappear during synchronization process. 2823 */ 2824 if (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED || 2825 sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) { 2826 /* 2827 * Previous state should be ACTIVE, STALE or 2828 * SYNCHRONIZING. 2829 */ 2830 KASSERT(disk->d_state == G_RAID3_DISK_STATE_ACTIVE || 2831 disk->d_state == G_RAID3_DISK_STATE_STALE || 2832 disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING, 2833 ("Wrong disk state (%s, %s).", 2834 g_raid3_get_diskname(disk), 2835 g_raid3_disk_state2str(disk->d_state))); 2836 } else if (sc->sc_state == G_RAID3_DEVICE_STATE_STARTING) { 2837 /* Previous state should be NEW. */ 2838 KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW, 2839 ("Wrong disk state (%s, %s).", 2840 g_raid3_get_diskname(disk), 2841 g_raid3_disk_state2str(disk->d_state))); 2842 /* 2843 * Reset bumping syncid if disk disappeared in STARTING 2844 * state. 2845 */ 2846 if ((sc->sc_bump_id & G_RAID3_BUMP_SYNCID) != 0) 2847 sc->sc_bump_id &= ~G_RAID3_BUMP_SYNCID; 2848 #ifdef INVARIANTS 2849 } else { 2850 KASSERT(1 == 0, ("Wrong device state (%s, %s, %s, %s).", 2851 sc->sc_name, 2852 g_raid3_device_state2str(sc->sc_state), 2853 g_raid3_get_diskname(disk), 2854 g_raid3_disk_state2str(disk->d_state))); 2855 #endif 2856 } 2857 DISK_STATE_CHANGED(); 2858 G_RAID3_DEBUG(0, "Device %s: provider %s disconnected.", 2859 sc->sc_name, g_raid3_get_diskname(disk)); 2860 2861 g_raid3_destroy_disk(disk); 2862 break; 2863 default: 2864 KASSERT(1 == 0, ("Unknown state (%u).", state)); 2865 break; 2866 } 2867 return (0); 2868 } 2869 #undef DISK_STATE_CHANGED 2870 2871 int 2872 g_raid3_read_metadata(struct g_consumer *cp, struct g_raid3_metadata *md) 2873 { 2874 struct g_provider *pp; 2875 u_char *buf; 2876 int error; 2877 2878 g_topology_assert(); 2879 2880 error = g_access(cp, 1, 0, 0); 2881 if (error != 0) 2882 return (error); 2883 pp = cp->provider; 2884 g_topology_unlock(); 2885 /* Metadata are stored on last sector. */ 2886 buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, 2887 &error); 2888 g_topology_lock(); 2889 g_access(cp, -1, 0, 0); 2890 if (buf == NULL) { 2891 G_RAID3_DEBUG(1, "Cannot read metadata from %s (error=%d).", 2892 cp->provider->name, error); 2893 return (error); 2894 } 2895 2896 /* Decode metadata. */ 2897 error = raid3_metadata_decode(buf, md); 2898 g_free(buf); 2899 if (strcmp(md->md_magic, G_RAID3_MAGIC) != 0) 2900 return (EINVAL); 2901 if (md->md_version > G_RAID3_VERSION) { 2902 G_RAID3_DEBUG(0, 2903 "Kernel module is too old to handle metadata from %s.", 2904 cp->provider->name); 2905 return (EINVAL); 2906 } 2907 if (error != 0) { 2908 G_RAID3_DEBUG(1, "MD5 metadata hash mismatch for provider %s.", 2909 cp->provider->name); 2910 return (error); 2911 } 2912 if (md->md_sectorsize > MAXPHYS) { 2913 G_RAID3_DEBUG(0, "The blocksize is too big."); 2914 return (EINVAL); 2915 } 2916 2917 return (0); 2918 } 2919 2920 static int 2921 g_raid3_check_metadata(struct g_raid3_softc *sc, struct g_provider *pp, 2922 struct g_raid3_metadata *md) 2923 { 2924 2925 if (md->md_no >= sc->sc_ndisks) { 2926 G_RAID3_DEBUG(1, "Invalid disk %s number (no=%u), skipping.", 2927 pp->name, md->md_no); 2928 return (EINVAL); 2929 } 2930 if (sc->sc_disks[md->md_no].d_state != G_RAID3_DISK_STATE_NODISK) { 2931 G_RAID3_DEBUG(1, "Disk %s (no=%u) already exists, skipping.", 2932 pp->name, md->md_no); 2933 return (EEXIST); 2934 } 2935 if (md->md_all != sc->sc_ndisks) { 2936 G_RAID3_DEBUG(1, 2937 "Invalid '%s' field on disk %s (device %s), skipping.", 2938 "md_all", pp->name, sc->sc_name); 2939 return (EINVAL); 2940 } 2941 if ((md->md_mediasize % md->md_sectorsize) != 0) { 2942 G_RAID3_DEBUG(1, "Invalid metadata (mediasize %% sectorsize != " 2943 "0) on disk %s (device %s), skipping.", pp->name, 2944 sc->sc_name); 2945 return (EINVAL); 2946 } 2947 if (md->md_mediasize != sc->sc_mediasize) { 2948 G_RAID3_DEBUG(1, 2949 "Invalid '%s' field on disk %s (device %s), skipping.", 2950 "md_mediasize", pp->name, sc->sc_name); 2951 return (EINVAL); 2952 } 2953 if ((md->md_mediasize % (sc->sc_ndisks - 1)) != 0) { 2954 G_RAID3_DEBUG(1, 2955 "Invalid '%s' field on disk %s (device %s), skipping.", 2956 "md_mediasize", pp->name, sc->sc_name); 2957 return (EINVAL); 2958 } 2959 if ((sc->sc_mediasize / (sc->sc_ndisks - 1)) > pp->mediasize) { 2960 G_RAID3_DEBUG(1, 2961 "Invalid size of disk %s (device %s), skipping.", pp->name, 2962 sc->sc_name); 2963 return (EINVAL); 2964 } 2965 if ((md->md_sectorsize / pp->sectorsize) < sc->sc_ndisks - 1) { 2966 G_RAID3_DEBUG(1, 2967 "Invalid '%s' field on disk %s (device %s), skipping.", 2968 "md_sectorsize", pp->name, sc->sc_name); 2969 return (EINVAL); 2970 } 2971 if (md->md_sectorsize != sc->sc_sectorsize) { 2972 G_RAID3_DEBUG(1, 2973 "Invalid '%s' field on disk %s (device %s), skipping.", 2974 "md_sectorsize", pp->name, sc->sc_name); 2975 return (EINVAL); 2976 } 2977 if ((sc->sc_sectorsize % pp->sectorsize) != 0) { 2978 G_RAID3_DEBUG(1, 2979 "Invalid sector size of disk %s (device %s), skipping.", 2980 pp->name, sc->sc_name); 2981 return (EINVAL); 2982 } 2983 if ((md->md_mflags & ~G_RAID3_DEVICE_FLAG_MASK) != 0) { 2984 G_RAID3_DEBUG(1, 2985 "Invalid device flags on disk %s (device %s), skipping.", 2986 pp->name, sc->sc_name); 2987 return (EINVAL); 2988 } 2989 if ((md->md_mflags & G_RAID3_DEVICE_FLAG_VERIFY) != 0 && 2990 (md->md_mflags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0) { 2991 /* 2992 * VERIFY and ROUND-ROBIN options are mutally exclusive. 2993 */ 2994 G_RAID3_DEBUG(1, "Both VERIFY and ROUND-ROBIN flags exist on " 2995 "disk %s (device %s), skipping.", pp->name, sc->sc_name); 2996 return (EINVAL); 2997 } 2998 if ((md->md_dflags & ~G_RAID3_DISK_FLAG_MASK) != 0) { 2999 G_RAID3_DEBUG(1, 3000 "Invalid disk flags on disk %s (device %s), skipping.", 3001 pp->name, sc->sc_name); 3002 return (EINVAL); 3003 } 3004 return (0); 3005 } 3006 3007 int 3008 g_raid3_add_disk(struct g_raid3_softc *sc, struct g_provider *pp, 3009 struct g_raid3_metadata *md) 3010 { 3011 struct g_raid3_disk *disk; 3012 int error; 3013 3014 g_topology_assert_not(); 3015 G_RAID3_DEBUG(2, "Adding disk %s.", pp->name); 3016 3017 error = g_raid3_check_metadata(sc, pp, md); 3018 if (error != 0) 3019 return (error); 3020 if (sc->sc_state != G_RAID3_DEVICE_STATE_STARTING && 3021 md->md_genid < sc->sc_genid) { 3022 G_RAID3_DEBUG(0, "Component %s (device %s) broken, skipping.", 3023 pp->name, sc->sc_name); 3024 return (EINVAL); 3025 } 3026 disk = g_raid3_init_disk(sc, pp, md, &error); 3027 if (disk == NULL) 3028 return (error); 3029 error = g_raid3_event_send(disk, G_RAID3_DISK_STATE_NEW, 3030 G_RAID3_EVENT_WAIT); 3031 if (error != 0) 3032 return (error); 3033 if (md->md_version < G_RAID3_VERSION) { 3034 G_RAID3_DEBUG(0, "Upgrading metadata on %s (v%d->v%d).", 3035 pp->name, md->md_version, G_RAID3_VERSION); 3036 g_raid3_update_metadata(disk); 3037 } 3038 return (0); 3039 } 3040 3041 static void 3042 g_raid3_destroy_delayed(void *arg, int flag) 3043 { 3044 struct g_raid3_softc *sc; 3045 int error; 3046 3047 if (flag == EV_CANCEL) { 3048 G_RAID3_DEBUG(1, "Destroying canceled."); 3049 return; 3050 } 3051 sc = arg; 3052 g_topology_unlock(); 3053 sx_xlock(&sc->sc_lock); 3054 KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) == 0, 3055 ("DESTROY flag set on %s.", sc->sc_name)); 3056 KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROYING) != 0, 3057 ("DESTROYING flag not set on %s.", sc->sc_name)); 3058 G_RAID3_DEBUG(0, "Destroying %s (delayed).", sc->sc_name); 3059 error = g_raid3_destroy(sc, G_RAID3_DESTROY_SOFT); 3060 if (error != 0) { 3061 G_RAID3_DEBUG(0, "Cannot destroy %s.", sc->sc_name); 3062 sx_xunlock(&sc->sc_lock); 3063 } 3064 g_topology_lock(); 3065 } 3066 3067 static int 3068 g_raid3_access(struct g_provider *pp, int acr, int acw, int ace) 3069 { 3070 struct g_raid3_softc *sc; 3071 int dcr, dcw, dce, error = 0; 3072 3073 g_topology_assert(); 3074 G_RAID3_DEBUG(2, "Access request for %s: r%dw%de%d.", pp->name, acr, 3075 acw, ace); 3076 3077 sc = pp->geom->softc; 3078 if (sc == NULL && acr <= 0 && acw <= 0 && ace <= 0) 3079 return (0); 3080 KASSERT(sc != NULL, ("NULL softc (provider=%s).", pp->name)); 3081 3082 dcr = pp->acr + acr; 3083 dcw = pp->acw + acw; 3084 dce = pp->ace + ace; 3085 3086 g_topology_unlock(); 3087 sx_xlock(&sc->sc_lock); 3088 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0 || 3089 g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) < sc->sc_ndisks - 1) { 3090 if (acr > 0 || acw > 0 || ace > 0) 3091 error = ENXIO; 3092 goto end; 3093 } 3094 if (dcw == 0) 3095 g_raid3_idle(sc, dcw); 3096 if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROYING) != 0) { 3097 if (acr > 0 || acw > 0 || ace > 0) { 3098 error = ENXIO; 3099 goto end; 3100 } 3101 if (dcr == 0 && dcw == 0 && dce == 0) { 3102 g_post_event(g_raid3_destroy_delayed, sc, M_WAITOK, 3103 sc, NULL); 3104 } 3105 } 3106 end: 3107 sx_xunlock(&sc->sc_lock); 3108 g_topology_lock(); 3109 return (error); 3110 } 3111 3112 static struct g_geom * 3113 g_raid3_create(struct g_class *mp, const struct g_raid3_metadata *md) 3114 { 3115 struct g_raid3_softc *sc; 3116 struct g_geom *gp; 3117 int error, timeout; 3118 u_int n; 3119 3120 g_topology_assert(); 3121 G_RAID3_DEBUG(1, "Creating device %s (id=%u).", md->md_name, md->md_id); 3122 3123 /* One disk is minimum. */ 3124 if (md->md_all < 1) 3125 return (NULL); 3126 /* 3127 * Action geom. 3128 */ 3129 gp = g_new_geomf(mp, "%s", md->md_name); 3130 sc = malloc(sizeof(*sc), M_RAID3, M_WAITOK | M_ZERO); 3131 sc->sc_disks = malloc(sizeof(struct g_raid3_disk) * md->md_all, M_RAID3, 3132 M_WAITOK | M_ZERO); 3133 gp->start = g_raid3_start; 3134 gp->orphan = g_raid3_orphan; 3135 gp->access = g_raid3_access; 3136 gp->dumpconf = g_raid3_dumpconf; 3137 3138 sc->sc_id = md->md_id; 3139 sc->sc_mediasize = md->md_mediasize; 3140 sc->sc_sectorsize = md->md_sectorsize; 3141 sc->sc_ndisks = md->md_all; 3142 sc->sc_round_robin = 0; 3143 sc->sc_flags = md->md_mflags; 3144 sc->sc_bump_id = 0; 3145 sc->sc_idle = 1; 3146 sc->sc_last_write = time_uptime; 3147 sc->sc_writes = 0; 3148 for (n = 0; n < sc->sc_ndisks; n++) { 3149 sc->sc_disks[n].d_softc = sc; 3150 sc->sc_disks[n].d_no = n; 3151 sc->sc_disks[n].d_state = G_RAID3_DISK_STATE_NODISK; 3152 } 3153 sx_init(&sc->sc_lock, "graid3:lock"); 3154 bioq_init(&sc->sc_queue); 3155 mtx_init(&sc->sc_queue_mtx, "graid3:queue", NULL, MTX_DEF); 3156 bioq_init(&sc->sc_regular_delayed); 3157 bioq_init(&sc->sc_inflight); 3158 bioq_init(&sc->sc_sync_delayed); 3159 TAILQ_INIT(&sc->sc_events); 3160 mtx_init(&sc->sc_events_mtx, "graid3:events", NULL, MTX_DEF); 3161 callout_init(&sc->sc_callout, 1); 3162 sc->sc_state = G_RAID3_DEVICE_STATE_STARTING; 3163 gp->softc = sc; 3164 sc->sc_geom = gp; 3165 sc->sc_provider = NULL; 3166 /* 3167 * Synchronization geom. 3168 */ 3169 gp = g_new_geomf(mp, "%s.sync", md->md_name); 3170 gp->softc = sc; 3171 gp->orphan = g_raid3_orphan; 3172 sc->sc_sync.ds_geom = gp; 3173 3174 if (!g_raid3_use_malloc) { 3175 sc->sc_zones[G_RAID3_ZONE_64K].sz_zone = uma_zcreate("gr3:64k", 3176 65536, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL, 3177 UMA_ALIGN_PTR, 0); 3178 sc->sc_zones[G_RAID3_ZONE_64K].sz_inuse = 0; 3179 sc->sc_zones[G_RAID3_ZONE_64K].sz_max = g_raid3_n64k; 3180 sc->sc_zones[G_RAID3_ZONE_64K].sz_requested = 3181 sc->sc_zones[G_RAID3_ZONE_64K].sz_failed = 0; 3182 sc->sc_zones[G_RAID3_ZONE_16K].sz_zone = uma_zcreate("gr3:16k", 3183 16384, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL, 3184 UMA_ALIGN_PTR, 0); 3185 sc->sc_zones[G_RAID3_ZONE_16K].sz_inuse = 0; 3186 sc->sc_zones[G_RAID3_ZONE_16K].sz_max = g_raid3_n16k; 3187 sc->sc_zones[G_RAID3_ZONE_16K].sz_requested = 3188 sc->sc_zones[G_RAID3_ZONE_16K].sz_failed = 0; 3189 sc->sc_zones[G_RAID3_ZONE_4K].sz_zone = uma_zcreate("gr3:4k", 3190 4096, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL, 3191 UMA_ALIGN_PTR, 0); 3192 sc->sc_zones[G_RAID3_ZONE_4K].sz_inuse = 0; 3193 sc->sc_zones[G_RAID3_ZONE_4K].sz_max = g_raid3_n4k; 3194 sc->sc_zones[G_RAID3_ZONE_4K].sz_requested = 3195 sc->sc_zones[G_RAID3_ZONE_4K].sz_failed = 0; 3196 } 3197 3198 error = kproc_create(g_raid3_worker, sc, &sc->sc_worker, 0, 0, 3199 "g_raid3 %s", md->md_name); 3200 if (error != 0) { 3201 G_RAID3_DEBUG(1, "Cannot create kernel thread for %s.", 3202 sc->sc_name); 3203 if (!g_raid3_use_malloc) { 3204 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_64K].sz_zone); 3205 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_16K].sz_zone); 3206 uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_4K].sz_zone); 3207 } 3208 g_destroy_geom(sc->sc_sync.ds_geom); 3209 mtx_destroy(&sc->sc_events_mtx); 3210 mtx_destroy(&sc->sc_queue_mtx); 3211 sx_destroy(&sc->sc_lock); 3212 g_destroy_geom(sc->sc_geom); 3213 free(sc->sc_disks, M_RAID3); 3214 free(sc, M_RAID3); 3215 return (NULL); 3216 } 3217 3218 G_RAID3_DEBUG(1, "Device %s created (%u components, id=%u).", 3219 sc->sc_name, sc->sc_ndisks, sc->sc_id); 3220 3221 sc->sc_rootmount = root_mount_hold("GRAID3"); 3222 G_RAID3_DEBUG(1, "root_mount_hold %p", sc->sc_rootmount); 3223 3224 /* 3225 * Run timeout. 3226 */ 3227 timeout = atomic_load_acq_int(&g_raid3_timeout); 3228 callout_reset(&sc->sc_callout, timeout * hz, g_raid3_go, sc); 3229 return (sc->sc_geom); 3230 } 3231 3232 int 3233 g_raid3_destroy(struct g_raid3_softc *sc, int how) 3234 { 3235 struct g_provider *pp; 3236 3237 g_topology_assert_not(); 3238 if (sc == NULL) 3239 return (ENXIO); 3240 sx_assert(&sc->sc_lock, SX_XLOCKED); 3241 3242 pp = sc->sc_provider; 3243 if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { 3244 switch (how) { 3245 case G_RAID3_DESTROY_SOFT: 3246 G_RAID3_DEBUG(1, 3247 "Device %s is still open (r%dw%de%d).", pp->name, 3248 pp->acr, pp->acw, pp->ace); 3249 return (EBUSY); 3250 case G_RAID3_DESTROY_DELAYED: 3251 G_RAID3_DEBUG(1, 3252 "Device %s will be destroyed on last close.", 3253 pp->name); 3254 if (sc->sc_syncdisk != NULL) 3255 g_raid3_sync_stop(sc, 1); 3256 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROYING; 3257 return (EBUSY); 3258 case G_RAID3_DESTROY_HARD: 3259 G_RAID3_DEBUG(1, "Device %s is still open, so it " 3260 "can't be definitely removed.", pp->name); 3261 break; 3262 } 3263 } 3264 3265 g_topology_lock(); 3266 if (sc->sc_geom->softc == NULL) { 3267 g_topology_unlock(); 3268 return (0); 3269 } 3270 sc->sc_geom->softc = NULL; 3271 sc->sc_sync.ds_geom->softc = NULL; 3272 g_topology_unlock(); 3273 3274 sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY; 3275 sc->sc_flags |= G_RAID3_DEVICE_FLAG_WAIT; 3276 G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc); 3277 sx_xunlock(&sc->sc_lock); 3278 mtx_lock(&sc->sc_queue_mtx); 3279 wakeup(sc); 3280 wakeup(&sc->sc_queue); 3281 mtx_unlock(&sc->sc_queue_mtx); 3282 G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, &sc->sc_worker); 3283 while (sc->sc_worker != NULL) 3284 tsleep(&sc->sc_worker, PRIBIO, "r3:destroy", hz / 5); 3285 G_RAID3_DEBUG(4, "%s: Woken up %p.", __func__, &sc->sc_worker); 3286 sx_xlock(&sc->sc_lock); 3287 g_raid3_destroy_device(sc); 3288 free(sc->sc_disks, M_RAID3); 3289 free(sc, M_RAID3); 3290 return (0); 3291 } 3292 3293 static void 3294 g_raid3_taste_orphan(struct g_consumer *cp) 3295 { 3296 3297 KASSERT(1 == 0, ("%s called while tasting %s.", __func__, 3298 cp->provider->name)); 3299 } 3300 3301 static struct g_geom * 3302 g_raid3_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) 3303 { 3304 struct g_raid3_metadata md; 3305 struct g_raid3_softc *sc; 3306 struct g_consumer *cp; 3307 struct g_geom *gp; 3308 int error; 3309 3310 g_topology_assert(); 3311 g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); 3312 G_RAID3_DEBUG(2, "Tasting %s.", pp->name); 3313 3314 gp = g_new_geomf(mp, "raid3:taste"); 3315 /* This orphan function should be never called. */ 3316 gp->orphan = g_raid3_taste_orphan; 3317 cp = g_new_consumer(gp); 3318 error = g_attach(cp, pp); 3319 if (error == 0) { 3320 error = g_raid3_read_metadata(cp, &md); 3321 g_detach(cp); 3322 } 3323 g_destroy_consumer(cp); 3324 g_destroy_geom(gp); 3325 if (error != 0) 3326 return (NULL); 3327 gp = NULL; 3328 3329 if (md.md_provider[0] != '\0' && 3330 !g_compare_names(md.md_provider, pp->name)) 3331 return (NULL); 3332 if (md.md_provsize != 0 && md.md_provsize != pp->mediasize) 3333 return (NULL); 3334 if (g_raid3_debug >= 2) 3335 raid3_metadata_dump(&md); 3336 3337 /* 3338 * Let's check if device already exists. 3339 */ 3340 sc = NULL; 3341 LIST_FOREACH(gp, &mp->geom, geom) { 3342 sc = gp->softc; 3343 if (sc == NULL) 3344 continue; 3345 if (sc->sc_sync.ds_geom == gp) 3346 continue; 3347 if (strcmp(md.md_name, sc->sc_name) != 0) 3348 continue; 3349 if (md.md_id != sc->sc_id) { 3350 G_RAID3_DEBUG(0, "Device %s already configured.", 3351 sc->sc_name); 3352 return (NULL); 3353 } 3354 break; 3355 } 3356 if (gp == NULL) { 3357 gp = g_raid3_create(mp, &md); 3358 if (gp == NULL) { 3359 G_RAID3_DEBUG(0, "Cannot create device %s.", 3360 md.md_name); 3361 return (NULL); 3362 } 3363 sc = gp->softc; 3364 } 3365 G_RAID3_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); 3366 g_topology_unlock(); 3367 sx_xlock(&sc->sc_lock); 3368 error = g_raid3_add_disk(sc, pp, &md); 3369 if (error != 0) { 3370 G_RAID3_DEBUG(0, "Cannot add disk %s to %s (error=%d).", 3371 pp->name, gp->name, error); 3372 if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NODISK) == 3373 sc->sc_ndisks) { 3374 g_cancel_event(sc); 3375 g_raid3_destroy(sc, G_RAID3_DESTROY_HARD); 3376 g_topology_lock(); 3377 return (NULL); 3378 } 3379 gp = NULL; 3380 } 3381 sx_xunlock(&sc->sc_lock); 3382 g_topology_lock(); 3383 return (gp); 3384 } 3385 3386 static int 3387 g_raid3_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused, 3388 struct g_geom *gp) 3389 { 3390 struct g_raid3_softc *sc; 3391 int error; 3392 3393 g_topology_unlock(); 3394 sc = gp->softc; 3395 sx_xlock(&sc->sc_lock); 3396 g_cancel_event(sc); 3397 error = g_raid3_destroy(gp->softc, G_RAID3_DESTROY_SOFT); 3398 if (error != 0) 3399 sx_xunlock(&sc->sc_lock); 3400 g_topology_lock(); 3401 return (error); 3402 } 3403 3404 static void 3405 g_raid3_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, 3406 struct g_consumer *cp, struct g_provider *pp) 3407 { 3408 struct g_raid3_softc *sc; 3409 3410 g_topology_assert(); 3411 3412 sc = gp->softc; 3413 if (sc == NULL) 3414 return; 3415 /* Skip synchronization geom. */ 3416 if (gp == sc->sc_sync.ds_geom) 3417 return; 3418 if (pp != NULL) { 3419 /* Nothing here. */ 3420 } else if (cp != NULL) { 3421 struct g_raid3_disk *disk; 3422 3423 disk = cp->private; 3424 if (disk == NULL) 3425 return; 3426 g_topology_unlock(); 3427 sx_xlock(&sc->sc_lock); 3428 sbuf_printf(sb, "%s<Type>", indent); 3429 if (disk->d_no == sc->sc_ndisks - 1) 3430 sbuf_cat(sb, "PARITY"); 3431 else 3432 sbuf_cat(sb, "DATA"); 3433 sbuf_cat(sb, "</Type>\n"); 3434 sbuf_printf(sb, "%s<Number>%u</Number>\n", indent, 3435 (u_int)disk->d_no); 3436 if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) { 3437 sbuf_printf(sb, "%s<Synchronized>", indent); 3438 if (disk->d_sync.ds_offset == 0) 3439 sbuf_cat(sb, "0%"); 3440 else { 3441 sbuf_printf(sb, "%u%%", 3442 (u_int)((disk->d_sync.ds_offset * 100) / 3443 (sc->sc_mediasize / (sc->sc_ndisks - 1)))); 3444 } 3445 sbuf_cat(sb, "</Synchronized>\n"); 3446 if (disk->d_sync.ds_offset > 0) { 3447 sbuf_printf(sb, "%s<BytesSynced>%jd" 3448 "</BytesSynced>\n", indent, 3449 (intmax_t)disk->d_sync.ds_offset); 3450 } 3451 } 3452 sbuf_printf(sb, "%s<SyncID>%u</SyncID>\n", indent, 3453 disk->d_sync.ds_syncid); 3454 sbuf_printf(sb, "%s<GenID>%u</GenID>\n", indent, disk->d_genid); 3455 sbuf_printf(sb, "%s<Flags>", indent); 3456 if (disk->d_flags == 0) 3457 sbuf_cat(sb, "NONE"); 3458 else { 3459 int first = 1; 3460 3461 #define ADD_FLAG(flag, name) do { \ 3462 if ((disk->d_flags & (flag)) != 0) { \ 3463 if (!first) \ 3464 sbuf_cat(sb, ", "); \ 3465 else \ 3466 first = 0; \ 3467 sbuf_cat(sb, name); \ 3468 } \ 3469 } while (0) 3470 ADD_FLAG(G_RAID3_DISK_FLAG_DIRTY, "DIRTY"); 3471 ADD_FLAG(G_RAID3_DISK_FLAG_HARDCODED, "HARDCODED"); 3472 ADD_FLAG(G_RAID3_DISK_FLAG_SYNCHRONIZING, 3473 "SYNCHRONIZING"); 3474 ADD_FLAG(G_RAID3_DISK_FLAG_FORCE_SYNC, "FORCE_SYNC"); 3475 ADD_FLAG(G_RAID3_DISK_FLAG_BROKEN, "BROKEN"); 3476 #undef ADD_FLAG 3477 } 3478 sbuf_cat(sb, "</Flags>\n"); 3479 sbuf_printf(sb, "%s<State>%s</State>\n", indent, 3480 g_raid3_disk_state2str(disk->d_state)); 3481 sx_xunlock(&sc->sc_lock); 3482 g_topology_lock(); 3483 } else { 3484 g_topology_unlock(); 3485 sx_xlock(&sc->sc_lock); 3486 if (!g_raid3_use_malloc) { 3487 sbuf_printf(sb, 3488 "%s<Zone4kRequested>%u</Zone4kRequested>\n", indent, 3489 sc->sc_zones[G_RAID3_ZONE_4K].sz_requested); 3490 sbuf_printf(sb, 3491 "%s<Zone4kFailed>%u</Zone4kFailed>\n", indent, 3492 sc->sc_zones[G_RAID3_ZONE_4K].sz_failed); 3493 sbuf_printf(sb, 3494 "%s<Zone16kRequested>%u</Zone16kRequested>\n", indent, 3495 sc->sc_zones[G_RAID3_ZONE_16K].sz_requested); 3496 sbuf_printf(sb, 3497 "%s<Zone16kFailed>%u</Zone16kFailed>\n", indent, 3498 sc->sc_zones[G_RAID3_ZONE_16K].sz_failed); 3499 sbuf_printf(sb, 3500 "%s<Zone64kRequested>%u</Zone64kRequested>\n", indent, 3501 sc->sc_zones[G_RAID3_ZONE_64K].sz_requested); 3502 sbuf_printf(sb, 3503 "%s<Zone64kFailed>%u</Zone64kFailed>\n", indent, 3504 sc->sc_zones[G_RAID3_ZONE_64K].sz_failed); 3505 } 3506 sbuf_printf(sb, "%s<ID>%u</ID>\n", indent, (u_int)sc->sc_id); 3507 sbuf_printf(sb, "%s<SyncID>%u</SyncID>\n", indent, sc->sc_syncid); 3508 sbuf_printf(sb, "%s<GenID>%u</GenID>\n", indent, sc->sc_genid); 3509 sbuf_printf(sb, "%s<Flags>", indent); 3510 if (sc->sc_flags == 0) 3511 sbuf_cat(sb, "NONE"); 3512 else { 3513 int first = 1; 3514 3515 #define ADD_FLAG(flag, name) do { \ 3516 if ((sc->sc_flags & (flag)) != 0) { \ 3517 if (!first) \ 3518 sbuf_cat(sb, ", "); \ 3519 else \ 3520 first = 0; \ 3521 sbuf_cat(sb, name); \ 3522 } \ 3523 } while (0) 3524 ADD_FLAG(G_RAID3_DEVICE_FLAG_NOFAILSYNC, "NOFAILSYNC"); 3525 ADD_FLAG(G_RAID3_DEVICE_FLAG_NOAUTOSYNC, "NOAUTOSYNC"); 3526 ADD_FLAG(G_RAID3_DEVICE_FLAG_ROUND_ROBIN, 3527 "ROUND-ROBIN"); 3528 ADD_FLAG(G_RAID3_DEVICE_FLAG_VERIFY, "VERIFY"); 3529 #undef ADD_FLAG 3530 } 3531 sbuf_cat(sb, "</Flags>\n"); 3532 sbuf_printf(sb, "%s<Components>%u</Components>\n", indent, 3533 sc->sc_ndisks); 3534 sbuf_printf(sb, "%s<State>%s</State>\n", indent, 3535 g_raid3_device_state2str(sc->sc_state)); 3536 sx_xunlock(&sc->sc_lock); 3537 g_topology_lock(); 3538 } 3539 } 3540 3541 static void 3542 g_raid3_shutdown_post_sync(void *arg, int howto) 3543 { 3544 struct g_class *mp; 3545 struct g_geom *gp, *gp2; 3546 struct g_raid3_softc *sc; 3547 int error; 3548 3549 mp = arg; 3550 g_topology_lock(); 3551 g_raid3_shutdown = 1; 3552 LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) { 3553 if ((sc = gp->softc) == NULL) 3554 continue; 3555 /* Skip synchronization geom. */ 3556 if (gp == sc->sc_sync.ds_geom) 3557 continue; 3558 g_topology_unlock(); 3559 sx_xlock(&sc->sc_lock); 3560 g_raid3_idle(sc, -1); 3561 g_cancel_event(sc); 3562 error = g_raid3_destroy(sc, G_RAID3_DESTROY_DELAYED); 3563 if (error != 0) 3564 sx_xunlock(&sc->sc_lock); 3565 g_topology_lock(); 3566 } 3567 g_topology_unlock(); 3568 } 3569 3570 static void 3571 g_raid3_init(struct g_class *mp) 3572 { 3573 3574 g_raid3_post_sync = EVENTHANDLER_REGISTER(shutdown_post_sync, 3575 g_raid3_shutdown_post_sync, mp, SHUTDOWN_PRI_FIRST); 3576 if (g_raid3_post_sync == NULL) 3577 G_RAID3_DEBUG(0, "Warning! Cannot register shutdown event."); 3578 } 3579 3580 static void 3581 g_raid3_fini(struct g_class *mp) 3582 { 3583 3584 if (g_raid3_post_sync != NULL) 3585 EVENTHANDLER_DEREGISTER(shutdown_post_sync, g_raid3_post_sync); 3586 } 3587 3588 DECLARE_GEOM_CLASS(g_raid3_class, g_raid3); 3589 MODULE_VERSION(geom_raid3, 0); 3590