xref: /freebsd/sys/geom/raid3/g_raid3.c (revision ba3c1f5972d7b90feb6e6da47905ff2757e0fe57)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
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 static void g_raid3_timeout_drain(struct g_raid3_softc *sc);
129 
130 static const char *
131 g_raid3_disk_state2str(int state)
132 {
133 
134 	switch (state) {
135 	case G_RAID3_DISK_STATE_NODISK:
136 		return ("NODISK");
137 	case G_RAID3_DISK_STATE_NONE:
138 		return ("NONE");
139 	case G_RAID3_DISK_STATE_NEW:
140 		return ("NEW");
141 	case G_RAID3_DISK_STATE_ACTIVE:
142 		return ("ACTIVE");
143 	case G_RAID3_DISK_STATE_STALE:
144 		return ("STALE");
145 	case G_RAID3_DISK_STATE_SYNCHRONIZING:
146 		return ("SYNCHRONIZING");
147 	case G_RAID3_DISK_STATE_DISCONNECTED:
148 		return ("DISCONNECTED");
149 	default:
150 		return ("INVALID");
151 	}
152 }
153 
154 static const char *
155 g_raid3_device_state2str(int state)
156 {
157 
158 	switch (state) {
159 	case G_RAID3_DEVICE_STATE_STARTING:
160 		return ("STARTING");
161 	case G_RAID3_DEVICE_STATE_DEGRADED:
162 		return ("DEGRADED");
163 	case G_RAID3_DEVICE_STATE_COMPLETE:
164 		return ("COMPLETE");
165 	default:
166 		return ("INVALID");
167 	}
168 }
169 
170 const char *
171 g_raid3_get_diskname(struct g_raid3_disk *disk)
172 {
173 
174 	if (disk->d_consumer == NULL || disk->d_consumer->provider == NULL)
175 		return ("[unknown]");
176 	return (disk->d_name);
177 }
178 
179 static void *
180 g_raid3_alloc(struct g_raid3_softc *sc, size_t size, int flags)
181 {
182 	void *ptr;
183 	enum g_raid3_zones zone;
184 
185 	if (g_raid3_use_malloc ||
186 	    (zone = g_raid3_zone(size)) == G_RAID3_NUM_ZONES)
187 		ptr = malloc(size, M_RAID3, flags);
188 	else {
189 		ptr = uma_zalloc_arg(sc->sc_zones[zone].sz_zone,
190 		   &sc->sc_zones[zone], flags);
191 		sc->sc_zones[zone].sz_requested++;
192 		if (ptr == NULL)
193 			sc->sc_zones[zone].sz_failed++;
194 	}
195 	return (ptr);
196 }
197 
198 static void
199 g_raid3_free(struct g_raid3_softc *sc, void *ptr, size_t size)
200 {
201 	enum g_raid3_zones zone;
202 
203 	if (g_raid3_use_malloc ||
204 	    (zone = g_raid3_zone(size)) == G_RAID3_NUM_ZONES)
205 		free(ptr, M_RAID3);
206 	else {
207 		uma_zfree_arg(sc->sc_zones[zone].sz_zone,
208 		    ptr, &sc->sc_zones[zone]);
209 	}
210 }
211 
212 static int
213 g_raid3_uma_ctor(void *mem, int size, void *arg, int flags)
214 {
215 	struct g_raid3_zone *sz = arg;
216 
217 	if (sz->sz_max > 0 && sz->sz_inuse == sz->sz_max)
218 		return (ENOMEM);
219 	sz->sz_inuse++;
220 	return (0);
221 }
222 
223 static void
224 g_raid3_uma_dtor(void *mem, int size, void *arg)
225 {
226 	struct g_raid3_zone *sz = arg;
227 
228 	sz->sz_inuse--;
229 }
230 
231 #define	g_raid3_xor(src, dst, size)					\
232 	_g_raid3_xor((uint64_t *)(src),					\
233 	    (uint64_t *)(dst), (size_t)size)
234 static void
235 _g_raid3_xor(uint64_t *src, uint64_t *dst, size_t size)
236 {
237 
238 	KASSERT((size % 128) == 0, ("Invalid size: %zu.", size));
239 	for (; size > 0; size -= 128) {
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 		*dst++ ^= (*src++);
256 	}
257 }
258 
259 static int
260 g_raid3_is_zero(struct bio *bp)
261 {
262 	static const uint64_t zeros[] = {
263 	    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
264 	};
265 	u_char *addr;
266 	ssize_t size;
267 
268 	size = bp->bio_length;
269 	addr = (u_char *)bp->bio_data;
270 	for (; size > 0; size -= sizeof(zeros), addr += sizeof(zeros)) {
271 		if (bcmp(addr, zeros, sizeof(zeros)) != 0)
272 			return (0);
273 	}
274 	return (1);
275 }
276 
277 /*
278  * --- Events handling functions ---
279  * Events in geom_raid3 are used to maintain disks and device status
280  * from one thread to simplify locking.
281  */
282 static void
283 g_raid3_event_free(struct g_raid3_event *ep)
284 {
285 
286 	free(ep, M_RAID3);
287 }
288 
289 static int
290 g_raid3_event_dispatch(struct g_raid3_event *ep, void *arg, int state,
291     int flags)
292 {
293 	struct g_raid3_softc *sc;
294 	struct g_raid3_disk *disk;
295 	int error;
296 
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 int
334 g_raid3_event_send(void *arg, int state, int flags)
335 {
336 	struct g_raid3_event *ep;
337 
338 	ep = malloc(sizeof(*ep), M_RAID3, M_WAITOK);
339 	return (g_raid3_event_dispatch(ep, arg, state, flags));
340 }
341 
342 static struct g_raid3_event *
343 g_raid3_event_get(struct g_raid3_softc *sc)
344 {
345 	struct g_raid3_event *ep;
346 
347 	mtx_lock(&sc->sc_events_mtx);
348 	ep = TAILQ_FIRST(&sc->sc_events);
349 	mtx_unlock(&sc->sc_events_mtx);
350 	return (ep);
351 }
352 
353 static void
354 g_raid3_event_remove(struct g_raid3_softc *sc, struct g_raid3_event *ep)
355 {
356 
357 	mtx_lock(&sc->sc_events_mtx);
358 	TAILQ_REMOVE(&sc->sc_events, ep, e_next);
359 	mtx_unlock(&sc->sc_events_mtx);
360 }
361 
362 static void
363 g_raid3_event_cancel(struct g_raid3_disk *disk)
364 {
365 	struct g_raid3_softc *sc;
366 	struct g_raid3_event *ep, *tmpep;
367 
368 	sc = disk->d_softc;
369 	sx_assert(&sc->sc_lock, SX_XLOCKED);
370 
371 	mtx_lock(&sc->sc_events_mtx);
372 	TAILQ_FOREACH_SAFE(ep, &sc->sc_events, e_next, tmpep) {
373 		if ((ep->e_flags & G_RAID3_EVENT_DEVICE) != 0)
374 			continue;
375 		if (ep->e_disk != disk)
376 			continue;
377 		TAILQ_REMOVE(&sc->sc_events, ep, e_next);
378 		if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0)
379 			g_raid3_event_free(ep);
380 		else {
381 			ep->e_error = ECANCELED;
382 			wakeup(ep);
383 		}
384 	}
385 	mtx_unlock(&sc->sc_events_mtx);
386 }
387 
388 /*
389  * Return the number of disks in the given state.
390  * If state is equal to -1, count all connected disks.
391  */
392 u_int
393 g_raid3_ndisks(struct g_raid3_softc *sc, int state)
394 {
395 	struct g_raid3_disk *disk;
396 	u_int n, ndisks;
397 
398 	sx_assert(&sc->sc_lock, SX_LOCKED);
399 
400 	for (n = ndisks = 0; n < sc->sc_ndisks; n++) {
401 		disk = &sc->sc_disks[n];
402 		if (disk->d_state == G_RAID3_DISK_STATE_NODISK)
403 			continue;
404 		if (state == -1 || disk->d_state == state)
405 			ndisks++;
406 	}
407 	return (ndisks);
408 }
409 
410 static u_int
411 g_raid3_nrequests(struct g_raid3_softc *sc, struct g_consumer *cp)
412 {
413 	struct bio *bp;
414 	u_int nreqs = 0;
415 
416 	mtx_lock(&sc->sc_queue_mtx);
417 	TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) {
418 		if (bp->bio_from == cp)
419 			nreqs++;
420 	}
421 	mtx_unlock(&sc->sc_queue_mtx);
422 	return (nreqs);
423 }
424 
425 static int
426 g_raid3_is_busy(struct g_raid3_softc *sc, struct g_consumer *cp)
427 {
428 
429 	if (cp->index > 0) {
430 		G_RAID3_DEBUG(2,
431 		    "I/O requests for %s exist, can't destroy it now.",
432 		    cp->provider->name);
433 		return (1);
434 	}
435 	if (g_raid3_nrequests(sc, cp) > 0) {
436 		G_RAID3_DEBUG(2,
437 		    "I/O requests for %s in queue, can't destroy it now.",
438 		    cp->provider->name);
439 		return (1);
440 	}
441 	return (0);
442 }
443 
444 static void
445 g_raid3_destroy_consumer(void *arg, int flags __unused)
446 {
447 	struct g_consumer *cp;
448 
449 	g_topology_assert();
450 
451 	cp = arg;
452 	G_RAID3_DEBUG(1, "Consumer %s destroyed.", cp->provider->name);
453 	g_detach(cp);
454 	g_destroy_consumer(cp);
455 }
456 
457 static void
458 g_raid3_kill_consumer(struct g_raid3_softc *sc, struct g_consumer *cp)
459 {
460 	struct g_provider *pp;
461 	int retaste_wait;
462 
463 	g_topology_assert();
464 
465 	cp->private = NULL;
466 	if (g_raid3_is_busy(sc, cp))
467 		return;
468 	G_RAID3_DEBUG(2, "Consumer %s destroyed.", cp->provider->name);
469 	pp = cp->provider;
470 	retaste_wait = 0;
471 	if (cp->acw == 1) {
472 		if ((pp->geom->flags & G_GEOM_WITHER) == 0)
473 			retaste_wait = 1;
474 	}
475 	G_RAID3_DEBUG(2, "Access %s r%dw%de%d = %d", pp->name, -cp->acr,
476 	    -cp->acw, -cp->ace, 0);
477 	if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0)
478 		g_access(cp, -cp->acr, -cp->acw, -cp->ace);
479 	if (retaste_wait) {
480 		/*
481 		 * After retaste event was send (inside g_access()), we can send
482 		 * event to detach and destroy consumer.
483 		 * A class, which has consumer to the given provider connected
484 		 * will not receive retaste event for the provider.
485 		 * This is the way how I ignore retaste events when I close
486 		 * consumers opened for write: I detach and destroy consumer
487 		 * after retaste event is sent.
488 		 */
489 		g_post_event(g_raid3_destroy_consumer, cp, M_WAITOK, NULL);
490 		return;
491 	}
492 	G_RAID3_DEBUG(1, "Consumer %s destroyed.", pp->name);
493 	g_detach(cp);
494 	g_destroy_consumer(cp);
495 }
496 
497 static int
498 g_raid3_connect_disk(struct g_raid3_disk *disk, struct g_provider *pp)
499 {
500 	struct g_consumer *cp;
501 	int error;
502 
503 	g_topology_assert_not();
504 	KASSERT(disk->d_consumer == NULL,
505 	    ("Disk already connected (device %s).", disk->d_softc->sc_name));
506 
507 	g_topology_lock();
508 	cp = g_new_consumer(disk->d_softc->sc_geom);
509 	error = g_attach(cp, pp);
510 	if (error != 0) {
511 		g_destroy_consumer(cp);
512 		g_topology_unlock();
513 		return (error);
514 	}
515 	error = g_access(cp, 1, 1, 1);
516 		g_topology_unlock();
517 	if (error != 0) {
518 		g_detach(cp);
519 		g_destroy_consumer(cp);
520 		G_RAID3_DEBUG(0, "Cannot open consumer %s (error=%d).",
521 		    pp->name, error);
522 		return (error);
523 	}
524 	disk->d_consumer = cp;
525 	disk->d_consumer->private = disk;
526 	disk->d_consumer->index = 0;
527 	G_RAID3_DEBUG(2, "Disk %s connected.", g_raid3_get_diskname(disk));
528 	return (0);
529 }
530 
531 static void
532 g_raid3_disconnect_consumer(struct g_raid3_softc *sc, struct g_consumer *cp)
533 {
534 
535 	g_topology_assert();
536 
537 	if (cp == NULL)
538 		return;
539 	if (cp->provider != NULL)
540 		g_raid3_kill_consumer(sc, cp);
541 	else
542 		g_destroy_consumer(cp);
543 }
544 
545 /*
546  * Initialize disk. This means allocate memory, create consumer, attach it
547  * to the provider and open access (r1w1e1) to it.
548  */
549 static struct g_raid3_disk *
550 g_raid3_init_disk(struct g_raid3_softc *sc, struct g_provider *pp,
551     struct g_raid3_metadata *md, int *errorp)
552 {
553 	struct g_raid3_disk *disk;
554 	int error;
555 
556 	disk = &sc->sc_disks[md->md_no];
557 	error = g_raid3_connect_disk(disk, pp);
558 	if (error != 0) {
559 		if (errorp != NULL)
560 			*errorp = error;
561 		return (NULL);
562 	}
563 	disk->d_state = G_RAID3_DISK_STATE_NONE;
564 	disk->d_flags = md->md_dflags;
565 	if (md->md_provider[0] != '\0')
566 		disk->d_flags |= G_RAID3_DISK_FLAG_HARDCODED;
567 	disk->d_sync.ds_consumer = NULL;
568 	disk->d_sync.ds_offset = md->md_sync_offset;
569 	disk->d_sync.ds_offset_done = md->md_sync_offset;
570 	disk->d_genid = md->md_genid;
571 	disk->d_sync.ds_syncid = md->md_syncid;
572 	if (errorp != NULL)
573 		*errorp = 0;
574 	return (disk);
575 }
576 
577 static void
578 g_raid3_destroy_disk(struct g_raid3_disk *disk)
579 {
580 	struct g_raid3_softc *sc;
581 
582 	g_topology_assert_not();
583 	sc = disk->d_softc;
584 	sx_assert(&sc->sc_lock, SX_XLOCKED);
585 
586 	if (disk->d_state == G_RAID3_DISK_STATE_NODISK)
587 		return;
588 	g_raid3_event_cancel(disk);
589 	switch (disk->d_state) {
590 	case G_RAID3_DISK_STATE_SYNCHRONIZING:
591 		if (sc->sc_syncdisk != NULL)
592 			g_raid3_sync_stop(sc, 1);
593 		/* FALLTHROUGH */
594 	case G_RAID3_DISK_STATE_NEW:
595 	case G_RAID3_DISK_STATE_STALE:
596 	case G_RAID3_DISK_STATE_ACTIVE:
597 		g_topology_lock();
598 		g_raid3_disconnect_consumer(sc, disk->d_consumer);
599 		g_topology_unlock();
600 		disk->d_consumer = NULL;
601 		break;
602 	default:
603 		KASSERT(0 == 1, ("Wrong disk state (%s, %s).",
604 		    g_raid3_get_diskname(disk),
605 		    g_raid3_disk_state2str(disk->d_state)));
606 	}
607 	disk->d_state = G_RAID3_DISK_STATE_NODISK;
608 }
609 
610 static void
611 g_raid3_destroy_device(struct g_raid3_softc *sc)
612 {
613 	struct g_raid3_event *ep;
614 	struct g_raid3_disk *disk;
615 	struct g_geom *gp;
616 	struct g_consumer *cp;
617 	u_int n;
618 
619 	g_topology_assert_not();
620 	sx_assert(&sc->sc_lock, SX_XLOCKED);
621 
622 	gp = sc->sc_geom;
623 	if (sc->sc_provider != NULL)
624 		g_raid3_destroy_provider(sc);
625 	for (n = 0; n < sc->sc_ndisks; n++) {
626 		disk = &sc->sc_disks[n];
627 		if (disk->d_state != G_RAID3_DISK_STATE_NODISK) {
628 			disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY;
629 			g_raid3_update_metadata(disk);
630 			g_raid3_destroy_disk(disk);
631 		}
632 	}
633 	while ((ep = g_raid3_event_get(sc)) != NULL) {
634 		g_raid3_event_remove(sc, ep);
635 		if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0)
636 			g_raid3_event_free(ep);
637 		else {
638 			ep->e_error = ECANCELED;
639 			ep->e_flags |= G_RAID3_EVENT_DONE;
640 			G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, ep);
641 			mtx_lock(&sc->sc_events_mtx);
642 			wakeup(ep);
643 			mtx_unlock(&sc->sc_events_mtx);
644 		}
645 	}
646 	g_raid3_timeout_drain(sc);
647 	cp = LIST_FIRST(&sc->sc_sync.ds_geom->consumer);
648 	g_topology_lock();
649 	if (cp != NULL)
650 		g_raid3_disconnect_consumer(sc, cp);
651 	g_wither_geom(sc->sc_sync.ds_geom, ENXIO);
652 	G_RAID3_DEBUG(0, "Device %s destroyed.", gp->name);
653 	g_wither_geom(gp, ENXIO);
654 	g_topology_unlock();
655 	if (!g_raid3_use_malloc) {
656 		uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_64K].sz_zone);
657 		uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_16K].sz_zone);
658 		uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_4K].sz_zone);
659 	}
660 	mtx_destroy(&sc->sc_queue_mtx);
661 	mtx_destroy(&sc->sc_events_mtx);
662 	sx_xunlock(&sc->sc_lock);
663 	sx_destroy(&sc->sc_lock);
664 }
665 
666 static void
667 g_raid3_orphan(struct g_consumer *cp)
668 {
669 	struct g_raid3_disk *disk;
670 
671 	g_topology_assert();
672 
673 	disk = cp->private;
674 	if (disk == NULL)
675 		return;
676 	disk->d_softc->sc_bump_id = G_RAID3_BUMP_SYNCID;
677 	g_raid3_event_send(disk, G_RAID3_DISK_STATE_DISCONNECTED,
678 	    G_RAID3_EVENT_DONTWAIT);
679 }
680 
681 static int
682 g_raid3_write_metadata(struct g_raid3_disk *disk, struct g_raid3_metadata *md)
683 {
684 	struct g_raid3_softc *sc;
685 	struct g_consumer *cp;
686 	off_t offset, length;
687 	u_char *sector;
688 	int error = 0;
689 
690 	g_topology_assert_not();
691 	sc = disk->d_softc;
692 	sx_assert(&sc->sc_lock, SX_LOCKED);
693 
694 	cp = disk->d_consumer;
695 	KASSERT(cp != NULL, ("NULL consumer (%s).", sc->sc_name));
696 	KASSERT(cp->provider != NULL, ("NULL provider (%s).", sc->sc_name));
697 	KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1,
698 	    ("Consumer %s closed? (r%dw%de%d).", cp->provider->name, cp->acr,
699 	    cp->acw, cp->ace));
700 	length = cp->provider->sectorsize;
701 	offset = cp->provider->mediasize - length;
702 	sector = malloc((size_t)length, M_RAID3, M_WAITOK | M_ZERO);
703 	if (md != NULL)
704 		raid3_metadata_encode(md, sector);
705 	error = g_write_data(cp, offset, sector, length);
706 	free(sector, M_RAID3);
707 	if (error != 0) {
708 		if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) {
709 			G_RAID3_DEBUG(0, "Cannot write metadata on %s "
710 			    "(device=%s, error=%d).",
711 			    g_raid3_get_diskname(disk), sc->sc_name, error);
712 			disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN;
713 		} else {
714 			G_RAID3_DEBUG(1, "Cannot write metadata on %s "
715 			    "(device=%s, error=%d).",
716 			    g_raid3_get_diskname(disk), sc->sc_name, error);
717 		}
718 		if (g_raid3_disconnect_on_failure &&
719 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) {
720 			sc->sc_bump_id |= G_RAID3_BUMP_GENID;
721 			g_raid3_event_send(disk,
722 			    G_RAID3_DISK_STATE_DISCONNECTED,
723 			    G_RAID3_EVENT_DONTWAIT);
724 		}
725 	}
726 	return (error);
727 }
728 
729 int
730 g_raid3_clear_metadata(struct g_raid3_disk *disk)
731 {
732 	int error;
733 
734 	g_topology_assert_not();
735 	sx_assert(&disk->d_softc->sc_lock, SX_LOCKED);
736 
737 	error = g_raid3_write_metadata(disk, NULL);
738 	if (error == 0) {
739 		G_RAID3_DEBUG(2, "Metadata on %s cleared.",
740 		    g_raid3_get_diskname(disk));
741 	} else {
742 		G_RAID3_DEBUG(0,
743 		    "Cannot clear metadata on disk %s (error=%d).",
744 		    g_raid3_get_diskname(disk), error);
745 	}
746 	return (error);
747 }
748 
749 void
750 g_raid3_fill_metadata(struct g_raid3_disk *disk, struct g_raid3_metadata *md)
751 {
752 	struct g_raid3_softc *sc;
753 	struct g_provider *pp;
754 
755 	bzero(md, sizeof(*md));
756 	sc = disk->d_softc;
757 	strlcpy(md->md_magic, G_RAID3_MAGIC, sizeof(md->md_magic));
758 	md->md_version = G_RAID3_VERSION;
759 	strlcpy(md->md_name, sc->sc_name, sizeof(md->md_name));
760 	md->md_id = sc->sc_id;
761 	md->md_all = sc->sc_ndisks;
762 	md->md_genid = sc->sc_genid;
763 	md->md_mediasize = sc->sc_mediasize;
764 	md->md_sectorsize = sc->sc_sectorsize;
765 	md->md_mflags = (sc->sc_flags & G_RAID3_DEVICE_FLAG_MASK);
766 	md->md_no = disk->d_no;
767 	md->md_syncid = disk->d_sync.ds_syncid;
768 	md->md_dflags = (disk->d_flags & G_RAID3_DISK_FLAG_MASK);
769 	if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) {
770 		md->md_sync_offset =
771 		    disk->d_sync.ds_offset_done / (sc->sc_ndisks - 1);
772 	}
773 	if (disk->d_consumer != NULL && disk->d_consumer->provider != NULL)
774 		pp = disk->d_consumer->provider;
775 	else
776 		pp = NULL;
777 	if ((disk->d_flags & G_RAID3_DISK_FLAG_HARDCODED) != 0 && pp != NULL)
778 		strlcpy(md->md_provider, pp->name, sizeof(md->md_provider));
779 	if (pp != NULL)
780 		md->md_provsize = pp->mediasize;
781 }
782 
783 void
784 g_raid3_update_metadata(struct g_raid3_disk *disk)
785 {
786 	struct g_raid3_softc *sc __diagused;
787 	struct g_raid3_metadata md;
788 	int error;
789 
790 	g_topology_assert_not();
791 	sc = disk->d_softc;
792 	sx_assert(&sc->sc_lock, SX_LOCKED);
793 
794 	g_raid3_fill_metadata(disk, &md);
795 	error = g_raid3_write_metadata(disk, &md);
796 	if (error == 0) {
797 		G_RAID3_DEBUG(2, "Metadata on %s updated.",
798 		    g_raid3_get_diskname(disk));
799 	} else {
800 		G_RAID3_DEBUG(0,
801 		    "Cannot update metadata on disk %s (error=%d).",
802 		    g_raid3_get_diskname(disk), error);
803 	}
804 }
805 
806 static void
807 g_raid3_bump_syncid(struct g_raid3_softc *sc)
808 {
809 	struct g_raid3_disk *disk;
810 	u_int n;
811 
812 	g_topology_assert_not();
813 	sx_assert(&sc->sc_lock, SX_XLOCKED);
814 	KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) > 0,
815 	    ("%s called with no active disks (device=%s).", __func__,
816 	    sc->sc_name));
817 
818 	sc->sc_syncid++;
819 	G_RAID3_DEBUG(1, "Device %s: syncid bumped to %u.", sc->sc_name,
820 	    sc->sc_syncid);
821 	for (n = 0; n < sc->sc_ndisks; n++) {
822 		disk = &sc->sc_disks[n];
823 		if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE ||
824 		    disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) {
825 			disk->d_sync.ds_syncid = sc->sc_syncid;
826 			g_raid3_update_metadata(disk);
827 		}
828 	}
829 }
830 
831 static void
832 g_raid3_bump_genid(struct g_raid3_softc *sc)
833 {
834 	struct g_raid3_disk *disk;
835 	u_int n;
836 
837 	g_topology_assert_not();
838 	sx_assert(&sc->sc_lock, SX_XLOCKED);
839 	KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) > 0,
840 	    ("%s called with no active disks (device=%s).", __func__,
841 	    sc->sc_name));
842 
843 	sc->sc_genid++;
844 	G_RAID3_DEBUG(1, "Device %s: genid bumped to %u.", sc->sc_name,
845 	    sc->sc_genid);
846 	for (n = 0; n < sc->sc_ndisks; n++) {
847 		disk = &sc->sc_disks[n];
848 		if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE ||
849 		    disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) {
850 			disk->d_genid = sc->sc_genid;
851 			g_raid3_update_metadata(disk);
852 		}
853 	}
854 }
855 
856 static int
857 g_raid3_idle(struct g_raid3_softc *sc, int acw)
858 {
859 	struct g_raid3_disk *disk;
860 	u_int i;
861 	int timeout;
862 
863 	g_topology_assert_not();
864 	sx_assert(&sc->sc_lock, SX_XLOCKED);
865 
866 	if (sc->sc_provider == NULL)
867 		return (0);
868 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0)
869 		return (0);
870 	if (sc->sc_idle)
871 		return (0);
872 	if (sc->sc_writes > 0)
873 		return (0);
874 	if (acw > 0 || (acw == -1 && sc->sc_provider->acw > 0)) {
875 		timeout = g_raid3_idletime - (time_uptime - sc->sc_last_write);
876 		if (!g_raid3_shutdown && timeout > 0)
877 			return (timeout);
878 	}
879 	sc->sc_idle = 1;
880 	for (i = 0; i < sc->sc_ndisks; i++) {
881 		disk = &sc->sc_disks[i];
882 		if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE)
883 			continue;
884 		G_RAID3_DEBUG(1, "Disk %s (device %s) marked as clean.",
885 		    g_raid3_get_diskname(disk), sc->sc_name);
886 		disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY;
887 		g_raid3_update_metadata(disk);
888 	}
889 	return (0);
890 }
891 
892 static void
893 g_raid3_unidle(struct g_raid3_softc *sc)
894 {
895 	struct g_raid3_disk *disk;
896 	u_int i;
897 
898 	g_topology_assert_not();
899 	sx_assert(&sc->sc_lock, SX_XLOCKED);
900 
901 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0)
902 		return;
903 	sc->sc_idle = 0;
904 	sc->sc_last_write = time_uptime;
905 	for (i = 0; i < sc->sc_ndisks; i++) {
906 		disk = &sc->sc_disks[i];
907 		if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE)
908 			continue;
909 		G_RAID3_DEBUG(1, "Disk %s (device %s) marked as dirty.",
910 		    g_raid3_get_diskname(disk), sc->sc_name);
911 		disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY;
912 		g_raid3_update_metadata(disk);
913 	}
914 }
915 
916 /*
917  * Treat bio_driver1 field in parent bio as list head and field bio_caller1
918  * in child bio as pointer to the next element on the list.
919  */
920 #define	G_RAID3_HEAD_BIO(pbp)	(pbp)->bio_driver1
921 
922 #define	G_RAID3_NEXT_BIO(cbp)	(cbp)->bio_caller1
923 
924 #define	G_RAID3_FOREACH_BIO(pbp, bp)					\
925 	for ((bp) = G_RAID3_HEAD_BIO(pbp); (bp) != NULL;		\
926 	    (bp) = G_RAID3_NEXT_BIO(bp))
927 
928 #define	G_RAID3_FOREACH_SAFE_BIO(pbp, bp, tmpbp)			\
929 	for ((bp) = G_RAID3_HEAD_BIO(pbp);				\
930 	    (bp) != NULL && ((tmpbp) = G_RAID3_NEXT_BIO(bp), 1);	\
931 	    (bp) = (tmpbp))
932 
933 static void
934 g_raid3_init_bio(struct bio *pbp)
935 {
936 
937 	G_RAID3_HEAD_BIO(pbp) = NULL;
938 }
939 
940 static void
941 g_raid3_remove_bio(struct bio *cbp)
942 {
943 	struct bio *pbp, *bp;
944 
945 	pbp = cbp->bio_parent;
946 	if (G_RAID3_HEAD_BIO(pbp) == cbp)
947 		G_RAID3_HEAD_BIO(pbp) = G_RAID3_NEXT_BIO(cbp);
948 	else {
949 		G_RAID3_FOREACH_BIO(pbp, bp) {
950 			if (G_RAID3_NEXT_BIO(bp) == cbp) {
951 				G_RAID3_NEXT_BIO(bp) = G_RAID3_NEXT_BIO(cbp);
952 				break;
953 			}
954 		}
955 	}
956 	G_RAID3_NEXT_BIO(cbp) = NULL;
957 }
958 
959 static void
960 g_raid3_replace_bio(struct bio *sbp, struct bio *dbp)
961 {
962 	struct bio *pbp, *bp;
963 
964 	g_raid3_remove_bio(sbp);
965 	pbp = dbp->bio_parent;
966 	G_RAID3_NEXT_BIO(sbp) = G_RAID3_NEXT_BIO(dbp);
967 	if (G_RAID3_HEAD_BIO(pbp) == dbp)
968 		G_RAID3_HEAD_BIO(pbp) = sbp;
969 	else {
970 		G_RAID3_FOREACH_BIO(pbp, bp) {
971 			if (G_RAID3_NEXT_BIO(bp) == dbp) {
972 				G_RAID3_NEXT_BIO(bp) = sbp;
973 				break;
974 			}
975 		}
976 	}
977 	G_RAID3_NEXT_BIO(dbp) = NULL;
978 }
979 
980 static void
981 g_raid3_destroy_bio(struct g_raid3_softc *sc, struct bio *cbp)
982 {
983 	struct bio *bp, *pbp;
984 	size_t size;
985 
986 	pbp = cbp->bio_parent;
987 	pbp->bio_children--;
988 	KASSERT(cbp->bio_data != NULL, ("NULL bio_data"));
989 	size = pbp->bio_length / (sc->sc_ndisks - 1);
990 	g_raid3_free(sc, cbp->bio_data, size);
991 	if (G_RAID3_HEAD_BIO(pbp) == cbp) {
992 		G_RAID3_HEAD_BIO(pbp) = G_RAID3_NEXT_BIO(cbp);
993 		G_RAID3_NEXT_BIO(cbp) = NULL;
994 		g_destroy_bio(cbp);
995 	} else {
996 		G_RAID3_FOREACH_BIO(pbp, bp) {
997 			if (G_RAID3_NEXT_BIO(bp) == cbp)
998 				break;
999 		}
1000 		if (bp != NULL) {
1001 			KASSERT(G_RAID3_NEXT_BIO(bp) != NULL,
1002 			    ("NULL bp->bio_driver1"));
1003 			G_RAID3_NEXT_BIO(bp) = G_RAID3_NEXT_BIO(cbp);
1004 			G_RAID3_NEXT_BIO(cbp) = NULL;
1005 		}
1006 		g_destroy_bio(cbp);
1007 	}
1008 }
1009 
1010 static struct bio *
1011 g_raid3_clone_bio(struct g_raid3_softc *sc, struct bio *pbp)
1012 {
1013 	struct bio *bp, *cbp;
1014 	size_t size;
1015 	int memflag;
1016 
1017 	cbp = g_clone_bio(pbp);
1018 	if (cbp == NULL)
1019 		return (NULL);
1020 	size = pbp->bio_length / (sc->sc_ndisks - 1);
1021 	if ((pbp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) != 0)
1022 		memflag = M_WAITOK;
1023 	else
1024 		memflag = M_NOWAIT;
1025 	cbp->bio_data = g_raid3_alloc(sc, size, memflag);
1026 	if (cbp->bio_data == NULL) {
1027 		pbp->bio_children--;
1028 		g_destroy_bio(cbp);
1029 		return (NULL);
1030 	}
1031 	G_RAID3_NEXT_BIO(cbp) = NULL;
1032 	if (G_RAID3_HEAD_BIO(pbp) == NULL)
1033 		G_RAID3_HEAD_BIO(pbp) = cbp;
1034 	else {
1035 		G_RAID3_FOREACH_BIO(pbp, bp) {
1036 			if (G_RAID3_NEXT_BIO(bp) == NULL) {
1037 				G_RAID3_NEXT_BIO(bp) = cbp;
1038 				break;
1039 			}
1040 		}
1041 	}
1042 	return (cbp);
1043 }
1044 
1045 static void
1046 g_raid3_scatter(struct bio *pbp)
1047 {
1048 	struct g_raid3_softc *sc;
1049 	struct g_raid3_disk *disk;
1050 	struct bio *bp, *cbp, *tmpbp;
1051 	off_t atom, cadd, padd, left;
1052 	int first;
1053 
1054 	sc = pbp->bio_to->geom->softc;
1055 	bp = NULL;
1056 	if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_NOPARITY) == 0) {
1057 		/*
1058 		 * Find bio for which we should calculate data.
1059 		 */
1060 		G_RAID3_FOREACH_BIO(pbp, cbp) {
1061 			if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) {
1062 				bp = cbp;
1063 				break;
1064 			}
1065 		}
1066 		KASSERT(bp != NULL, ("NULL parity bio."));
1067 	}
1068 	atom = sc->sc_sectorsize / (sc->sc_ndisks - 1);
1069 	cadd = padd = 0;
1070 	for (left = pbp->bio_length; left > 0; left -= sc->sc_sectorsize) {
1071 		G_RAID3_FOREACH_BIO(pbp, cbp) {
1072 			if (cbp == bp)
1073 				continue;
1074 			bcopy(pbp->bio_data + padd, cbp->bio_data + cadd, atom);
1075 			padd += atom;
1076 		}
1077 		cadd += atom;
1078 	}
1079 	if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_NOPARITY) == 0) {
1080 		/*
1081 		 * Calculate parity.
1082 		 */
1083 		first = 1;
1084 		G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) {
1085 			if (cbp == bp)
1086 				continue;
1087 			if (first) {
1088 				bcopy(cbp->bio_data, bp->bio_data,
1089 				    bp->bio_length);
1090 				first = 0;
1091 			} else {
1092 				g_raid3_xor(cbp->bio_data, bp->bio_data,
1093 				    bp->bio_length);
1094 			}
1095 			if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_NODISK) != 0)
1096 				g_raid3_destroy_bio(sc, cbp);
1097 		}
1098 	}
1099 	G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) {
1100 		struct g_consumer *cp;
1101 
1102 		disk = cbp->bio_caller2;
1103 		cp = disk->d_consumer;
1104 		cbp->bio_to = cp->provider;
1105 		G_RAID3_LOGREQ(3, cbp, "Sending request.");
1106 		KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1,
1107 		    ("Consumer %s not opened (r%dw%de%d).", cp->provider->name,
1108 		    cp->acr, cp->acw, cp->ace));
1109 		cp->index++;
1110 		sc->sc_writes++;
1111 		g_io_request(cbp, cp);
1112 	}
1113 }
1114 
1115 static void
1116 g_raid3_gather(struct bio *pbp)
1117 {
1118 	struct g_raid3_softc *sc;
1119 	struct g_raid3_disk *disk;
1120 	struct bio *xbp, *fbp, *cbp;
1121 	off_t atom, cadd, padd, left;
1122 
1123 	sc = pbp->bio_to->geom->softc;
1124 	/*
1125 	 * Find bio for which we have to calculate data.
1126 	 * While going through this path, check if all requests
1127 	 * succeeded, if not, deny whole request.
1128 	 * If we're in COMPLETE mode, we allow one request to fail,
1129 	 * so if we find one, we're sending it to the parity consumer.
1130 	 * If there are more failed requests, we deny whole request.
1131 	 */
1132 	xbp = fbp = NULL;
1133 	G_RAID3_FOREACH_BIO(pbp, cbp) {
1134 		if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0) {
1135 			KASSERT(xbp == NULL, ("More than one parity bio."));
1136 			xbp = cbp;
1137 		}
1138 		if (cbp->bio_error == 0)
1139 			continue;
1140 		/*
1141 		 * Found failed request.
1142 		 */
1143 		if (fbp == NULL) {
1144 			if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_DEGRADED) != 0) {
1145 				/*
1146 				 * We are already in degraded mode, so we can't
1147 				 * accept any failures.
1148 				 */
1149 				if (pbp->bio_error == 0)
1150 					pbp->bio_error = cbp->bio_error;
1151 			} else {
1152 				fbp = cbp;
1153 			}
1154 		} else {
1155 			/*
1156 			 * Next failed request, that's too many.
1157 			 */
1158 			if (pbp->bio_error == 0)
1159 				pbp->bio_error = fbp->bio_error;
1160 		}
1161 		disk = cbp->bio_caller2;
1162 		if (disk == NULL)
1163 			continue;
1164 		if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) {
1165 			disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN;
1166 			G_RAID3_LOGREQ(0, cbp, "Request failed (error=%d).",
1167 			    cbp->bio_error);
1168 		} else {
1169 			G_RAID3_LOGREQ(1, cbp, "Request failed (error=%d).",
1170 			    cbp->bio_error);
1171 		}
1172 		if (g_raid3_disconnect_on_failure &&
1173 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) {
1174 			sc->sc_bump_id |= G_RAID3_BUMP_GENID;
1175 			g_raid3_event_send(disk,
1176 			    G_RAID3_DISK_STATE_DISCONNECTED,
1177 			    G_RAID3_EVENT_DONTWAIT);
1178 		}
1179 	}
1180 	if (pbp->bio_error != 0)
1181 		goto finish;
1182 	if (fbp != NULL && (pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) {
1183 		pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_VERIFY;
1184 		if (xbp != fbp)
1185 			g_raid3_replace_bio(xbp, fbp);
1186 		g_raid3_destroy_bio(sc, fbp);
1187 	} else if (fbp != NULL) {
1188 		struct g_consumer *cp;
1189 
1190 		/*
1191 		 * One request failed, so send the same request to
1192 		 * the parity consumer.
1193 		 */
1194 		disk = pbp->bio_driver2;
1195 		if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) {
1196 			pbp->bio_error = fbp->bio_error;
1197 			goto finish;
1198 		}
1199 		pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED;
1200 		pbp->bio_inbed--;
1201 		fbp->bio_flags &= ~(BIO_DONE | BIO_ERROR);
1202 		if (disk->d_no == sc->sc_ndisks - 1)
1203 			fbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY;
1204 		fbp->bio_error = 0;
1205 		fbp->bio_completed = 0;
1206 		fbp->bio_children = 0;
1207 		fbp->bio_inbed = 0;
1208 		cp = disk->d_consumer;
1209 		fbp->bio_caller2 = disk;
1210 		fbp->bio_to = cp->provider;
1211 		G_RAID3_LOGREQ(3, fbp, "Sending request (recover).");
1212 		KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1,
1213 		    ("Consumer %s not opened (r%dw%de%d).", cp->provider->name,
1214 		    cp->acr, cp->acw, cp->ace));
1215 		cp->index++;
1216 		g_io_request(fbp, cp);
1217 		return;
1218 	}
1219 	if (xbp != NULL) {
1220 		/*
1221 		 * Calculate parity.
1222 		 */
1223 		G_RAID3_FOREACH_BIO(pbp, cbp) {
1224 			if ((cbp->bio_cflags & G_RAID3_BIO_CFLAG_PARITY) != 0)
1225 				continue;
1226 			g_raid3_xor(cbp->bio_data, xbp->bio_data,
1227 			    xbp->bio_length);
1228 		}
1229 		xbp->bio_cflags &= ~G_RAID3_BIO_CFLAG_PARITY;
1230 		if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0) {
1231 			if (!g_raid3_is_zero(xbp)) {
1232 				g_raid3_parity_mismatch++;
1233 				pbp->bio_error = EIO;
1234 				goto finish;
1235 			}
1236 			g_raid3_destroy_bio(sc, xbp);
1237 		}
1238 	}
1239 	atom = sc->sc_sectorsize / (sc->sc_ndisks - 1);
1240 	cadd = padd = 0;
1241 	for (left = pbp->bio_length; left > 0; left -= sc->sc_sectorsize) {
1242 		G_RAID3_FOREACH_BIO(pbp, cbp) {
1243 			bcopy(cbp->bio_data + cadd, pbp->bio_data + padd, atom);
1244 			pbp->bio_completed += atom;
1245 			padd += atom;
1246 		}
1247 		cadd += atom;
1248 	}
1249 finish:
1250 	if (pbp->bio_error == 0)
1251 		G_RAID3_LOGREQ(3, pbp, "Request finished.");
1252 	else {
1253 		if ((pbp->bio_pflags & G_RAID3_BIO_PFLAG_VERIFY) != 0)
1254 			G_RAID3_LOGREQ(1, pbp, "Verification error.");
1255 		else
1256 			G_RAID3_LOGREQ(0, pbp, "Request failed.");
1257 	}
1258 	pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_MASK;
1259 	while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL)
1260 		g_raid3_destroy_bio(sc, cbp);
1261 	g_io_deliver(pbp, pbp->bio_error);
1262 }
1263 
1264 static void
1265 g_raid3_done(struct bio *bp)
1266 {
1267 	struct g_raid3_softc *sc;
1268 
1269 	sc = bp->bio_from->geom->softc;
1270 	bp->bio_cflags |= G_RAID3_BIO_CFLAG_REGULAR;
1271 	G_RAID3_LOGREQ(3, bp, "Regular request done (error=%d).", bp->bio_error);
1272 	mtx_lock(&sc->sc_queue_mtx);
1273 	bioq_insert_head(&sc->sc_queue, bp);
1274 	mtx_unlock(&sc->sc_queue_mtx);
1275 	wakeup(sc);
1276 	wakeup(&sc->sc_queue);
1277 }
1278 
1279 static void
1280 g_raid3_regular_request(struct bio *cbp)
1281 {
1282 	struct g_raid3_softc *sc;
1283 	struct g_raid3_disk *disk;
1284 	struct bio *pbp;
1285 
1286 	g_topology_assert_not();
1287 
1288 	pbp = cbp->bio_parent;
1289 	sc = pbp->bio_to->geom->softc;
1290 	cbp->bio_from->index--;
1291 	if (cbp->bio_cmd == BIO_WRITE)
1292 		sc->sc_writes--;
1293 	disk = cbp->bio_from->private;
1294 	if (disk == NULL) {
1295 		g_topology_lock();
1296 		g_raid3_kill_consumer(sc, cbp->bio_from);
1297 		g_topology_unlock();
1298 	}
1299 
1300 	G_RAID3_LOGREQ(3, cbp, "Request finished.");
1301 	pbp->bio_inbed++;
1302 	KASSERT(pbp->bio_inbed <= pbp->bio_children,
1303 	    ("bio_inbed (%u) is bigger than bio_children (%u).", pbp->bio_inbed,
1304 	    pbp->bio_children));
1305 	if (pbp->bio_inbed != pbp->bio_children)
1306 		return;
1307 	switch (pbp->bio_cmd) {
1308 	case BIO_READ:
1309 		g_raid3_gather(pbp);
1310 		break;
1311 	case BIO_WRITE:
1312 	case BIO_DELETE:
1313 	    {
1314 		int error = 0;
1315 
1316 		pbp->bio_completed = pbp->bio_length;
1317 		while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL) {
1318 			if (cbp->bio_error == 0) {
1319 				g_raid3_destroy_bio(sc, cbp);
1320 				continue;
1321 			}
1322 
1323 			if (error == 0)
1324 				error = cbp->bio_error;
1325 			else if (pbp->bio_error == 0) {
1326 				/*
1327 				 * Next failed request, that's too many.
1328 				 */
1329 				pbp->bio_error = error;
1330 			}
1331 
1332 			disk = cbp->bio_caller2;
1333 			if (disk == NULL) {
1334 				g_raid3_destroy_bio(sc, cbp);
1335 				continue;
1336 			}
1337 
1338 			if ((disk->d_flags & G_RAID3_DISK_FLAG_BROKEN) == 0) {
1339 				disk->d_flags |= G_RAID3_DISK_FLAG_BROKEN;
1340 				G_RAID3_LOGREQ(0, cbp,
1341 				    "Request failed (error=%d).",
1342 				    cbp->bio_error);
1343 			} else {
1344 				G_RAID3_LOGREQ(1, cbp,
1345 				    "Request failed (error=%d).",
1346 				    cbp->bio_error);
1347 			}
1348 			if (g_raid3_disconnect_on_failure &&
1349 			    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) {
1350 				sc->sc_bump_id |= G_RAID3_BUMP_GENID;
1351 				g_raid3_event_send(disk,
1352 				    G_RAID3_DISK_STATE_DISCONNECTED,
1353 				    G_RAID3_EVENT_DONTWAIT);
1354 			}
1355 			g_raid3_destroy_bio(sc, cbp);
1356 		}
1357 		if (pbp->bio_error == 0)
1358 			G_RAID3_LOGREQ(3, pbp, "Request finished.");
1359 		else
1360 			G_RAID3_LOGREQ(0, pbp, "Request failed.");
1361 		pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_DEGRADED;
1362 		pbp->bio_pflags &= ~G_RAID3_BIO_PFLAG_NOPARITY;
1363 		bioq_remove(&sc->sc_inflight, pbp);
1364 		/* Release delayed sync requests if possible. */
1365 		g_raid3_sync_release(sc);
1366 		g_io_deliver(pbp, pbp->bio_error);
1367 		break;
1368 	    }
1369 	}
1370 }
1371 
1372 static void
1373 g_raid3_sync_done(struct bio *bp)
1374 {
1375 	struct g_raid3_softc *sc;
1376 
1377 	G_RAID3_LOGREQ(3, bp, "Synchronization request delivered.");
1378 	sc = bp->bio_from->geom->softc;
1379 	bp->bio_cflags |= G_RAID3_BIO_CFLAG_SYNC;
1380 	mtx_lock(&sc->sc_queue_mtx);
1381 	bioq_insert_head(&sc->sc_queue, bp);
1382 	mtx_unlock(&sc->sc_queue_mtx);
1383 	wakeup(sc);
1384 	wakeup(&sc->sc_queue);
1385 }
1386 
1387 static void
1388 g_raid3_flush(struct g_raid3_softc *sc, struct bio *bp)
1389 {
1390 	struct bio_queue_head queue;
1391 	struct g_raid3_disk *disk;
1392 	struct g_consumer *cp __diagused;
1393 	struct bio *cbp;
1394 	u_int i;
1395 
1396 	bioq_init(&queue);
1397 	for (i = 0; i < sc->sc_ndisks; i++) {
1398 		disk = &sc->sc_disks[i];
1399 		if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE)
1400 			continue;
1401 		cbp = g_clone_bio(bp);
1402 		if (cbp == NULL) {
1403 			for (cbp = bioq_first(&queue); cbp != NULL;
1404 			    cbp = bioq_first(&queue)) {
1405 				bioq_remove(&queue, cbp);
1406 				g_destroy_bio(cbp);
1407 			}
1408 			if (bp->bio_error == 0)
1409 				bp->bio_error = ENOMEM;
1410 			g_io_deliver(bp, bp->bio_error);
1411 			return;
1412 		}
1413 		bioq_insert_tail(&queue, cbp);
1414 		cbp->bio_done = g_std_done;
1415 		cbp->bio_caller1 = disk;
1416 		cbp->bio_to = disk->d_consumer->provider;
1417 	}
1418 	for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) {
1419 		bioq_remove(&queue, cbp);
1420 		G_RAID3_LOGREQ(3, cbp, "Sending request.");
1421 		disk = cbp->bio_caller1;
1422 		cbp->bio_caller1 = NULL;
1423 		cp = disk->d_consumer;
1424 		KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1,
1425 		    ("Consumer %s not opened (r%dw%de%d).", cp->provider->name,
1426 		    cp->acr, cp->acw, cp->ace));
1427 		g_io_request(cbp, disk->d_consumer);
1428 	}
1429 }
1430 
1431 static void
1432 g_raid3_start(struct bio *bp)
1433 {
1434 	struct g_raid3_softc *sc;
1435 
1436 	sc = bp->bio_to->geom->softc;
1437 	/*
1438 	 * If sc == NULL or there are no valid disks, provider's error
1439 	 * should be set and g_raid3_start() should not be called at all.
1440 	 */
1441 	KASSERT(sc != NULL && (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED ||
1442 	    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE),
1443 	    ("Provider's error should be set (error=%d)(device=%s).",
1444 	    bp->bio_to->error, bp->bio_to->name));
1445 	G_RAID3_LOGREQ(3, bp, "Request received.");
1446 
1447 	switch (bp->bio_cmd) {
1448 	case BIO_READ:
1449 	case BIO_WRITE:
1450 	case BIO_DELETE:
1451 		break;
1452 	case BIO_SPEEDUP:
1453 	case BIO_FLUSH:
1454 		g_raid3_flush(sc, bp);
1455 		return;
1456 	case BIO_GETATTR:
1457 	default:
1458 		g_io_deliver(bp, EOPNOTSUPP);
1459 		return;
1460 	}
1461 	mtx_lock(&sc->sc_queue_mtx);
1462 	bioq_insert_tail(&sc->sc_queue, bp);
1463 	mtx_unlock(&sc->sc_queue_mtx);
1464 	G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc);
1465 	wakeup(sc);
1466 }
1467 
1468 /*
1469  * Return TRUE if the given request is colliding with a in-progress
1470  * synchronization request.
1471  */
1472 static int
1473 g_raid3_sync_collision(struct g_raid3_softc *sc, struct bio *bp)
1474 {
1475 	struct g_raid3_disk *disk;
1476 	struct bio *sbp;
1477 	off_t rstart, rend, sstart, send;
1478 	int i;
1479 
1480 	disk = sc->sc_syncdisk;
1481 	if (disk == NULL)
1482 		return (0);
1483 	rstart = bp->bio_offset;
1484 	rend = bp->bio_offset + bp->bio_length;
1485 	for (i = 0; i < g_raid3_syncreqs; i++) {
1486 		sbp = disk->d_sync.ds_bios[i];
1487 		if (sbp == NULL)
1488 			continue;
1489 		sstart = sbp->bio_offset;
1490 		send = sbp->bio_length;
1491 		if (sbp->bio_cmd == BIO_WRITE) {
1492 			sstart *= sc->sc_ndisks - 1;
1493 			send *= sc->sc_ndisks - 1;
1494 		}
1495 		send += sstart;
1496 		if (rend > sstart && rstart < send)
1497 			return (1);
1498 	}
1499 	return (0);
1500 }
1501 
1502 /*
1503  * Return TRUE if the given sync request is colliding with a in-progress regular
1504  * request.
1505  */
1506 static int
1507 g_raid3_regular_collision(struct g_raid3_softc *sc, struct bio *sbp)
1508 {
1509 	off_t rstart, rend, sstart, send;
1510 	struct bio *bp;
1511 
1512 	if (sc->sc_syncdisk == NULL)
1513 		return (0);
1514 	sstart = sbp->bio_offset;
1515 	send = sstart + sbp->bio_length;
1516 	TAILQ_FOREACH(bp, &sc->sc_inflight.queue, bio_queue) {
1517 		rstart = bp->bio_offset;
1518 		rend = bp->bio_offset + bp->bio_length;
1519 		if (rend > sstart && rstart < send)
1520 			return (1);
1521 	}
1522 	return (0);
1523 }
1524 
1525 /*
1526  * Puts request onto delayed queue.
1527  */
1528 static void
1529 g_raid3_regular_delay(struct g_raid3_softc *sc, struct bio *bp)
1530 {
1531 
1532 	G_RAID3_LOGREQ(2, bp, "Delaying request.");
1533 	bioq_insert_head(&sc->sc_regular_delayed, bp);
1534 }
1535 
1536 /*
1537  * Puts synchronization request onto delayed queue.
1538  */
1539 static void
1540 g_raid3_sync_delay(struct g_raid3_softc *sc, struct bio *bp)
1541 {
1542 
1543 	G_RAID3_LOGREQ(2, bp, "Delaying synchronization request.");
1544 	bioq_insert_tail(&sc->sc_sync_delayed, bp);
1545 }
1546 
1547 /*
1548  * Releases delayed regular requests which don't collide anymore with sync
1549  * requests.
1550  */
1551 static void
1552 g_raid3_regular_release(struct g_raid3_softc *sc)
1553 {
1554 	struct bio *bp, *bp2;
1555 
1556 	TAILQ_FOREACH_SAFE(bp, &sc->sc_regular_delayed.queue, bio_queue, bp2) {
1557 		if (g_raid3_sync_collision(sc, bp))
1558 			continue;
1559 		bioq_remove(&sc->sc_regular_delayed, bp);
1560 		G_RAID3_LOGREQ(2, bp, "Releasing delayed request (%p).", bp);
1561 		mtx_lock(&sc->sc_queue_mtx);
1562 		bioq_insert_head(&sc->sc_queue, bp);
1563 #if 0
1564 		/*
1565 		 * wakeup() is not needed, because this function is called from
1566 		 * the worker thread.
1567 		 */
1568 		wakeup(&sc->sc_queue);
1569 #endif
1570 		mtx_unlock(&sc->sc_queue_mtx);
1571 	}
1572 }
1573 
1574 /*
1575  * Releases delayed sync requests which don't collide anymore with regular
1576  * requests.
1577  */
1578 static void
1579 g_raid3_sync_release(struct g_raid3_softc *sc)
1580 {
1581 	struct bio *bp, *bp2;
1582 
1583 	TAILQ_FOREACH_SAFE(bp, &sc->sc_sync_delayed.queue, bio_queue, bp2) {
1584 		if (g_raid3_regular_collision(sc, bp))
1585 			continue;
1586 		bioq_remove(&sc->sc_sync_delayed, bp);
1587 		G_RAID3_LOGREQ(2, bp,
1588 		    "Releasing delayed synchronization request.");
1589 		g_io_request(bp, bp->bio_from);
1590 	}
1591 }
1592 
1593 /*
1594  * Handle synchronization requests.
1595  * Every synchronization request is two-steps process: first, READ request is
1596  * send to active provider and then WRITE request (with read data) to the provider
1597  * being synchronized. When WRITE is finished, new synchronization request is
1598  * send.
1599  */
1600 static void
1601 g_raid3_sync_request(struct bio *bp)
1602 {
1603 	struct g_raid3_softc *sc;
1604 	struct g_raid3_disk *disk;
1605 
1606 	bp->bio_from->index--;
1607 	sc = bp->bio_from->geom->softc;
1608 	disk = bp->bio_from->private;
1609 	if (disk == NULL) {
1610 		sx_xunlock(&sc->sc_lock); /* Avoid recursion on sc_lock. */
1611 		g_topology_lock();
1612 		g_raid3_kill_consumer(sc, bp->bio_from);
1613 		g_topology_unlock();
1614 		free(bp->bio_data, M_RAID3);
1615 		g_destroy_bio(bp);
1616 		sx_xlock(&sc->sc_lock);
1617 		return;
1618 	}
1619 
1620 	/*
1621 	 * Synchronization request.
1622 	 */
1623 	switch (bp->bio_cmd) {
1624 	case BIO_READ:
1625 	    {
1626 		struct g_consumer *cp;
1627 		u_char *dst, *src;
1628 		off_t left;
1629 		u_int atom;
1630 
1631 		if (bp->bio_error != 0) {
1632 			G_RAID3_LOGREQ(0, bp,
1633 			    "Synchronization request failed (error=%d).",
1634 			    bp->bio_error);
1635 			g_destroy_bio(bp);
1636 			return;
1637 		}
1638 		G_RAID3_LOGREQ(3, bp, "Synchronization request finished.");
1639 		atom = sc->sc_sectorsize / (sc->sc_ndisks - 1);
1640 		dst = src = bp->bio_data;
1641 		if (disk->d_no == sc->sc_ndisks - 1) {
1642 			u_int n;
1643 
1644 			/* Parity component. */
1645 			for (left = bp->bio_length; left > 0;
1646 			    left -= sc->sc_sectorsize) {
1647 				bcopy(src, dst, atom);
1648 				src += atom;
1649 				for (n = 1; n < sc->sc_ndisks - 1; n++) {
1650 					g_raid3_xor(src, dst, atom);
1651 					src += atom;
1652 				}
1653 				dst += atom;
1654 			}
1655 		} else {
1656 			/* Regular component. */
1657 			src += atom * disk->d_no;
1658 			for (left = bp->bio_length; left > 0;
1659 			    left -= sc->sc_sectorsize) {
1660 				bcopy(src, dst, atom);
1661 				src += sc->sc_sectorsize;
1662 				dst += atom;
1663 			}
1664 		}
1665 		bp->bio_driver1 = bp->bio_driver2 = NULL;
1666 		bp->bio_pflags = 0;
1667 		bp->bio_offset /= sc->sc_ndisks - 1;
1668 		bp->bio_length /= sc->sc_ndisks - 1;
1669 		bp->bio_cmd = BIO_WRITE;
1670 		bp->bio_cflags = 0;
1671 		bp->bio_children = bp->bio_inbed = 0;
1672 		cp = disk->d_consumer;
1673 		KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1,
1674 		    ("Consumer %s not opened (r%dw%de%d).", cp->provider->name,
1675 		    cp->acr, cp->acw, cp->ace));
1676 		cp->index++;
1677 		g_io_request(bp, cp);
1678 		return;
1679 	    }
1680 	case BIO_WRITE:
1681 	    {
1682 		struct g_raid3_disk_sync *sync;
1683 		off_t boffset, moffset;
1684 		void *data;
1685 		int i;
1686 
1687 		if (bp->bio_error != 0) {
1688 			G_RAID3_LOGREQ(0, bp,
1689 			    "Synchronization request failed (error=%d).",
1690 			    bp->bio_error);
1691 			g_destroy_bio(bp);
1692 			sc->sc_bump_id |= G_RAID3_BUMP_GENID;
1693 			g_raid3_event_send(disk,
1694 			    G_RAID3_DISK_STATE_DISCONNECTED,
1695 			    G_RAID3_EVENT_DONTWAIT);
1696 			return;
1697 		}
1698 		G_RAID3_LOGREQ(3, bp, "Synchronization request finished.");
1699 		sync = &disk->d_sync;
1700 		if (sync->ds_offset == sc->sc_mediasize / (sc->sc_ndisks - 1) ||
1701 		    sync->ds_consumer == NULL ||
1702 		    (sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0) {
1703 			/* Don't send more synchronization requests. */
1704 			sync->ds_inflight--;
1705 			if (sync->ds_bios != NULL) {
1706 				i = (int)(uintptr_t)bp->bio_caller1;
1707 				sync->ds_bios[i] = NULL;
1708 			}
1709 			free(bp->bio_data, M_RAID3);
1710 			g_destroy_bio(bp);
1711 			if (sync->ds_inflight > 0)
1712 				return;
1713 			if (sync->ds_consumer == NULL ||
1714 			    (sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0) {
1715 				return;
1716 			}
1717 			/*
1718 			 * Disk up-to-date, activate it.
1719 			 */
1720 			g_raid3_event_send(disk, G_RAID3_DISK_STATE_ACTIVE,
1721 			    G_RAID3_EVENT_DONTWAIT);
1722 			return;
1723 		}
1724 
1725 		/* Send next synchronization request. */
1726 		data = bp->bio_data;
1727 		g_reset_bio(bp);
1728 		bp->bio_cmd = BIO_READ;
1729 		bp->bio_offset = sync->ds_offset * (sc->sc_ndisks - 1);
1730 		bp->bio_length = MIN(maxphys, sc->sc_mediasize - bp->bio_offset);
1731 		sync->ds_offset += bp->bio_length / (sc->sc_ndisks - 1);
1732 		bp->bio_done = g_raid3_sync_done;
1733 		bp->bio_data = data;
1734 		bp->bio_from = sync->ds_consumer;
1735 		bp->bio_to = sc->sc_provider;
1736 		G_RAID3_LOGREQ(3, bp, "Sending synchronization request.");
1737 		sync->ds_consumer->index++;
1738 		/*
1739 		 * Delay the request if it is colliding with a regular request.
1740 		 */
1741 		if (g_raid3_regular_collision(sc, bp))
1742 			g_raid3_sync_delay(sc, bp);
1743 		else
1744 			g_io_request(bp, sync->ds_consumer);
1745 
1746 		/* Release delayed requests if possible. */
1747 		g_raid3_regular_release(sc);
1748 
1749 		/* Find the smallest offset. */
1750 		moffset = sc->sc_mediasize;
1751 		for (i = 0; i < g_raid3_syncreqs; i++) {
1752 			bp = sync->ds_bios[i];
1753 			boffset = bp->bio_offset;
1754 			if (bp->bio_cmd == BIO_WRITE)
1755 				boffset *= sc->sc_ndisks - 1;
1756 			if (boffset < moffset)
1757 				moffset = boffset;
1758 		}
1759 		if (sync->ds_offset_done + maxphys * 100 < moffset) {
1760 			/* Update offset_done on every 100 blocks. */
1761 			sync->ds_offset_done = moffset;
1762 			g_raid3_update_metadata(disk);
1763 		}
1764 		return;
1765 	    }
1766 	default:
1767 		KASSERT(1 == 0, ("Invalid command here: %u (device=%s)",
1768 		    bp->bio_cmd, sc->sc_name));
1769 		break;
1770 	}
1771 }
1772 
1773 static int
1774 g_raid3_register_request(struct bio *pbp)
1775 {
1776 	struct g_raid3_softc *sc;
1777 	struct g_raid3_disk *disk;
1778 	struct g_consumer *cp;
1779 	struct bio *cbp, *tmpbp;
1780 	off_t offset, length;
1781 	u_int n, ndisks;
1782 	int round_robin, verify;
1783 
1784 	ndisks = 0;
1785 	sc = pbp->bio_to->geom->softc;
1786 	if ((pbp->bio_cflags & G_RAID3_BIO_CFLAG_REGSYNC) != 0 &&
1787 	    sc->sc_syncdisk == NULL) {
1788 		g_io_deliver(pbp, EIO);
1789 		return (0);
1790 	}
1791 	g_raid3_init_bio(pbp);
1792 	length = pbp->bio_length / (sc->sc_ndisks - 1);
1793 	offset = pbp->bio_offset / (sc->sc_ndisks - 1);
1794 	round_robin = verify = 0;
1795 	switch (pbp->bio_cmd) {
1796 	case BIO_READ:
1797 		if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_VERIFY) != 0 &&
1798 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) {
1799 			pbp->bio_pflags |= G_RAID3_BIO_PFLAG_VERIFY;
1800 			verify = 1;
1801 			ndisks = sc->sc_ndisks;
1802 		} else {
1803 			verify = 0;
1804 			ndisks = sc->sc_ndisks - 1;
1805 		}
1806 		if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0 &&
1807 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) {
1808 			round_robin = 1;
1809 		} else {
1810 			round_robin = 0;
1811 		}
1812 		KASSERT(!round_robin || !verify,
1813 		    ("ROUND-ROBIN and VERIFY are mutually exclusive."));
1814 		pbp->bio_driver2 = &sc->sc_disks[sc->sc_ndisks - 1];
1815 		break;
1816 	case BIO_WRITE:
1817 	case BIO_DELETE:
1818 		/*
1819 		 * Delay the request if it is colliding with a synchronization
1820 		 * request.
1821 		 */
1822 		if (g_raid3_sync_collision(sc, pbp)) {
1823 			g_raid3_regular_delay(sc, pbp);
1824 			return (0);
1825 		}
1826 
1827 		if (sc->sc_idle)
1828 			g_raid3_unidle(sc);
1829 		else
1830 			sc->sc_last_write = time_uptime;
1831 
1832 		ndisks = sc->sc_ndisks;
1833 		break;
1834 	}
1835 	for (n = 0; n < ndisks; n++) {
1836 		disk = &sc->sc_disks[n];
1837 		cbp = g_raid3_clone_bio(sc, pbp);
1838 		if (cbp == NULL) {
1839 			while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL)
1840 				g_raid3_destroy_bio(sc, cbp);
1841 			/*
1842 			 * To prevent deadlock, we must run back up
1843 			 * with the ENOMEM for failed requests of any
1844 			 * of our consumers.  Our own sync requests
1845 			 * can stick around, as they are finite.
1846 			 */
1847 			if ((pbp->bio_cflags &
1848 			    G_RAID3_BIO_CFLAG_REGULAR) != 0) {
1849 				g_io_deliver(pbp, ENOMEM);
1850 				return (0);
1851 			}
1852 			return (ENOMEM);
1853 		}
1854 		cbp->bio_offset = offset;
1855 		cbp->bio_length = length;
1856 		cbp->bio_done = g_raid3_done;
1857 		switch (pbp->bio_cmd) {
1858 		case BIO_READ:
1859 			if (disk->d_state != G_RAID3_DISK_STATE_ACTIVE) {
1860 				/*
1861 				 * Replace invalid component with the parity
1862 				 * component.
1863 				 */
1864 				disk = &sc->sc_disks[sc->sc_ndisks - 1];
1865 				cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY;
1866 				pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED;
1867 			} else if (round_robin &&
1868 			    disk->d_no == sc->sc_round_robin) {
1869 				/*
1870 				 * In round-robin mode skip one data component
1871 				 * and use parity component when reading.
1872 				 */
1873 				pbp->bio_driver2 = disk;
1874 				disk = &sc->sc_disks[sc->sc_ndisks - 1];
1875 				cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY;
1876 				sc->sc_round_robin++;
1877 				round_robin = 0;
1878 			} else if (verify && disk->d_no == sc->sc_ndisks - 1) {
1879 				cbp->bio_cflags |= G_RAID3_BIO_CFLAG_PARITY;
1880 			}
1881 			break;
1882 		case BIO_WRITE:
1883 		case BIO_DELETE:
1884 			if (disk->d_state == G_RAID3_DISK_STATE_ACTIVE ||
1885 			    disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) {
1886 				if (n == ndisks - 1) {
1887 					/*
1888 					 * Active parity component, mark it as such.
1889 					 */
1890 					cbp->bio_cflags |=
1891 					    G_RAID3_BIO_CFLAG_PARITY;
1892 				}
1893 			} else {
1894 				pbp->bio_pflags |= G_RAID3_BIO_PFLAG_DEGRADED;
1895 				if (n == ndisks - 1) {
1896 					/*
1897 					 * Parity component is not connected,
1898 					 * so destroy its request.
1899 					 */
1900 					pbp->bio_pflags |=
1901 					    G_RAID3_BIO_PFLAG_NOPARITY;
1902 					g_raid3_destroy_bio(sc, cbp);
1903 					cbp = NULL;
1904 				} else {
1905 					cbp->bio_cflags |=
1906 					    G_RAID3_BIO_CFLAG_NODISK;
1907 					disk = NULL;
1908 				}
1909 			}
1910 			break;
1911 		}
1912 		if (cbp != NULL)
1913 			cbp->bio_caller2 = disk;
1914 	}
1915 	switch (pbp->bio_cmd) {
1916 	case BIO_READ:
1917 		if (round_robin) {
1918 			/*
1919 			 * If we are in round-robin mode and 'round_robin' is
1920 			 * still 1, it means, that we skipped parity component
1921 			 * for this read and must reset sc_round_robin field.
1922 			 */
1923 			sc->sc_round_robin = 0;
1924 		}
1925 		G_RAID3_FOREACH_SAFE_BIO(pbp, cbp, tmpbp) {
1926 			disk = cbp->bio_caller2;
1927 			cp = disk->d_consumer;
1928 			cbp->bio_to = cp->provider;
1929 			G_RAID3_LOGREQ(3, cbp, "Sending request.");
1930 			KASSERT(cp->acr >= 1 && cp->acw >= 1 && cp->ace >= 1,
1931 			    ("Consumer %s not opened (r%dw%de%d).",
1932 			    cp->provider->name, cp->acr, cp->acw, cp->ace));
1933 			cp->index++;
1934 			g_io_request(cbp, cp);
1935 		}
1936 		break;
1937 	case BIO_WRITE:
1938 	case BIO_DELETE:
1939 		/*
1940 		 * Put request onto inflight queue, so we can check if new
1941 		 * synchronization requests don't collide with it.
1942 		 */
1943 		bioq_insert_tail(&sc->sc_inflight, pbp);
1944 
1945 		/*
1946 		 * Bump syncid on first write.
1947 		 */
1948 		if ((sc->sc_bump_id & G_RAID3_BUMP_SYNCID) != 0) {
1949 			sc->sc_bump_id &= ~G_RAID3_BUMP_SYNCID;
1950 			g_raid3_bump_syncid(sc);
1951 		}
1952 		g_raid3_scatter(pbp);
1953 		break;
1954 	}
1955 	return (0);
1956 }
1957 
1958 static int
1959 g_raid3_can_destroy(struct g_raid3_softc *sc)
1960 {
1961 	struct g_geom *gp;
1962 	struct g_consumer *cp;
1963 
1964 	g_topology_assert();
1965 	gp = sc->sc_geom;
1966 	if (gp->softc == NULL)
1967 		return (1);
1968 	LIST_FOREACH(cp, &gp->consumer, consumer) {
1969 		if (g_raid3_is_busy(sc, cp))
1970 			return (0);
1971 	}
1972 	gp = sc->sc_sync.ds_geom;
1973 	LIST_FOREACH(cp, &gp->consumer, consumer) {
1974 		if (g_raid3_is_busy(sc, cp))
1975 			return (0);
1976 	}
1977 	G_RAID3_DEBUG(2, "No I/O requests for %s, it can be destroyed.",
1978 	    sc->sc_name);
1979 	return (1);
1980 }
1981 
1982 static int
1983 g_raid3_try_destroy(struct g_raid3_softc *sc)
1984 {
1985 
1986 	g_topology_assert_not();
1987 	sx_assert(&sc->sc_lock, SX_XLOCKED);
1988 
1989 	if (sc->sc_rootmount != NULL) {
1990 		G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__,
1991 		    sc->sc_rootmount);
1992 		root_mount_rel(sc->sc_rootmount);
1993 		sc->sc_rootmount = NULL;
1994 	}
1995 
1996 	g_topology_lock();
1997 	if (!g_raid3_can_destroy(sc)) {
1998 		g_topology_unlock();
1999 		return (0);
2000 	}
2001 	sc->sc_geom->softc = NULL;
2002 	sc->sc_sync.ds_geom->softc = NULL;
2003 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_WAIT) != 0) {
2004 		g_topology_unlock();
2005 		G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__,
2006 		    &sc->sc_worker);
2007 		/* Unlock sc_lock here, as it can be destroyed after wakeup. */
2008 		sx_xunlock(&sc->sc_lock);
2009 		wakeup(&sc->sc_worker);
2010 		sc->sc_worker = NULL;
2011 	} else {
2012 		g_topology_unlock();
2013 		g_raid3_destroy_device(sc);
2014 		free(sc->sc_disks, M_RAID3);
2015 		free(sc, M_RAID3);
2016 	}
2017 	return (1);
2018 }
2019 
2020 /*
2021  * Worker thread.
2022  */
2023 static void
2024 g_raid3_worker(void *arg)
2025 {
2026 	struct g_raid3_softc *sc;
2027 	struct g_raid3_event *ep;
2028 	struct bio *bp;
2029 	int timeout;
2030 
2031 	sc = arg;
2032 	thread_lock(curthread);
2033 	sched_prio(curthread, PRIBIO);
2034 	thread_unlock(curthread);
2035 
2036 	sx_xlock(&sc->sc_lock);
2037 	for (;;) {
2038 		G_RAID3_DEBUG(5, "%s: Let's see...", __func__);
2039 		/*
2040 		 * First take a look at events.
2041 		 * This is important to handle events before any I/O requests.
2042 		 */
2043 		ep = g_raid3_event_get(sc);
2044 		if (ep != NULL) {
2045 			g_raid3_event_remove(sc, ep);
2046 			if ((ep->e_flags & G_RAID3_EVENT_DEVICE) != 0) {
2047 				/* Update only device status. */
2048 				G_RAID3_DEBUG(3,
2049 				    "Running event for device %s.",
2050 				    sc->sc_name);
2051 				ep->e_error = 0;
2052 				g_raid3_update_device(sc, 1);
2053 			} else {
2054 				/* Update disk status. */
2055 				G_RAID3_DEBUG(3, "Running event for disk %s.",
2056 				     g_raid3_get_diskname(ep->e_disk));
2057 				ep->e_error = g_raid3_update_disk(ep->e_disk,
2058 				    ep->e_state);
2059 				if (ep->e_error == 0)
2060 					g_raid3_update_device(sc, 0);
2061 			}
2062 			if ((ep->e_flags & G_RAID3_EVENT_DONTWAIT) != 0) {
2063 				KASSERT(ep->e_error == 0,
2064 				    ("Error cannot be handled."));
2065 				g_raid3_event_free(ep);
2066 			} else {
2067 				ep->e_flags |= G_RAID3_EVENT_DONE;
2068 				G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__,
2069 				    ep);
2070 				mtx_lock(&sc->sc_events_mtx);
2071 				wakeup(ep);
2072 				mtx_unlock(&sc->sc_events_mtx);
2073 			}
2074 			if ((sc->sc_flags &
2075 			    G_RAID3_DEVICE_FLAG_DESTROY) != 0) {
2076 				if (g_raid3_try_destroy(sc)) {
2077 					curthread->td_pflags &= ~TDP_GEOM;
2078 					G_RAID3_DEBUG(1, "Thread exiting.");
2079 					kproc_exit(0);
2080 				}
2081 			}
2082 			G_RAID3_DEBUG(5, "%s: I'm here 1.", __func__);
2083 			continue;
2084 		}
2085 		/*
2086 		 * Check if we can mark array as CLEAN and if we can't take
2087 		 * how much seconds should we wait.
2088 		 */
2089 		timeout = g_raid3_idle(sc, -1);
2090 		/*
2091 		 * Now I/O requests.
2092 		 */
2093 		/* Get first request from the queue. */
2094 		mtx_lock(&sc->sc_queue_mtx);
2095 		bp = bioq_first(&sc->sc_queue);
2096 		if (bp == NULL) {
2097 			if ((sc->sc_flags &
2098 			    G_RAID3_DEVICE_FLAG_DESTROY) != 0) {
2099 				mtx_unlock(&sc->sc_queue_mtx);
2100 				if (g_raid3_try_destroy(sc)) {
2101 					curthread->td_pflags &= ~TDP_GEOM;
2102 					G_RAID3_DEBUG(1, "Thread exiting.");
2103 					kproc_exit(0);
2104 				}
2105 				mtx_lock(&sc->sc_queue_mtx);
2106 			}
2107 			sx_xunlock(&sc->sc_lock);
2108 			/*
2109 			 * XXX: We can miss an event here, because an event
2110 			 *      can be added without sx-device-lock and without
2111 			 *      mtx-queue-lock. Maybe I should just stop using
2112 			 *      dedicated mutex for events synchronization and
2113 			 *      stick with the queue lock?
2114 			 *      The event will hang here until next I/O request
2115 			 *      or next event is received.
2116 			 */
2117 			MSLEEP(sc, &sc->sc_queue_mtx, PRIBIO | PDROP, "r3:w1",
2118 			    timeout * hz);
2119 			sx_xlock(&sc->sc_lock);
2120 			G_RAID3_DEBUG(5, "%s: I'm here 4.", __func__);
2121 			continue;
2122 		}
2123 process:
2124 		bioq_remove(&sc->sc_queue, bp);
2125 		mtx_unlock(&sc->sc_queue_mtx);
2126 
2127 		if (bp->bio_from->geom == sc->sc_sync.ds_geom &&
2128 		    (bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) != 0) {
2129 			g_raid3_sync_request(bp);	/* READ */
2130 		} else if (bp->bio_to != sc->sc_provider) {
2131 			if ((bp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR) != 0)
2132 				g_raid3_regular_request(bp);
2133 			else if ((bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC) != 0)
2134 				g_raid3_sync_request(bp);	/* WRITE */
2135 			else {
2136 				KASSERT(0,
2137 				    ("Invalid request cflags=0x%hx to=%s.",
2138 				    bp->bio_cflags, bp->bio_to->name));
2139 			}
2140 		} else if (g_raid3_register_request(bp) != 0) {
2141 			mtx_lock(&sc->sc_queue_mtx);
2142 			bioq_insert_head(&sc->sc_queue, bp);
2143 			/*
2144 			 * We are short in memory, let see if there are finished
2145 			 * request we can free.
2146 			 */
2147 			TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) {
2148 				if (bp->bio_cflags & G_RAID3_BIO_CFLAG_REGULAR)
2149 					goto process;
2150 			}
2151 			/*
2152 			 * No finished regular request, so at least keep
2153 			 * synchronization running.
2154 			 */
2155 			TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) {
2156 				if (bp->bio_cflags & G_RAID3_BIO_CFLAG_SYNC)
2157 					goto process;
2158 			}
2159 			sx_xunlock(&sc->sc_lock);
2160 			MSLEEP(&sc->sc_queue, &sc->sc_queue_mtx, PRIBIO | PDROP,
2161 			    "r3:lowmem", hz / 10);
2162 			sx_xlock(&sc->sc_lock);
2163 		}
2164 		G_RAID3_DEBUG(5, "%s: I'm here 9.", __func__);
2165 	}
2166 }
2167 
2168 static void
2169 g_raid3_update_idle(struct g_raid3_softc *sc, struct g_raid3_disk *disk)
2170 {
2171 
2172 	sx_assert(&sc->sc_lock, SX_LOCKED);
2173 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) != 0)
2174 		return;
2175 	if (!sc->sc_idle && (disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) == 0) {
2176 		G_RAID3_DEBUG(1, "Disk %s (device %s) marked as dirty.",
2177 		    g_raid3_get_diskname(disk), sc->sc_name);
2178 		disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY;
2179 	} else if (sc->sc_idle &&
2180 	    (disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) != 0) {
2181 		G_RAID3_DEBUG(1, "Disk %s (device %s) marked as clean.",
2182 		    g_raid3_get_diskname(disk), sc->sc_name);
2183 		disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY;
2184 	}
2185 }
2186 
2187 static void
2188 g_raid3_sync_start(struct g_raid3_softc *sc)
2189 {
2190 	struct g_raid3_disk *disk;
2191 	struct g_consumer *cp;
2192 	struct bio *bp;
2193 	int error __diagused;
2194 	u_int n;
2195 
2196 	g_topology_assert_not();
2197 	sx_assert(&sc->sc_lock, SX_XLOCKED);
2198 
2199 	KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED,
2200 	    ("Device not in DEGRADED state (%s, %u).", sc->sc_name,
2201 	    sc->sc_state));
2202 	KASSERT(sc->sc_syncdisk == NULL, ("Syncdisk is not NULL (%s, %u).",
2203 	    sc->sc_name, sc->sc_state));
2204 	disk = NULL;
2205 	for (n = 0; n < sc->sc_ndisks; n++) {
2206 		if (sc->sc_disks[n].d_state != G_RAID3_DISK_STATE_SYNCHRONIZING)
2207 			continue;
2208 		disk = &sc->sc_disks[n];
2209 		break;
2210 	}
2211 	if (disk == NULL)
2212 		return;
2213 
2214 	sx_xunlock(&sc->sc_lock);
2215 	g_topology_lock();
2216 	cp = g_new_consumer(sc->sc_sync.ds_geom);
2217 	error = g_attach(cp, sc->sc_provider);
2218 	KASSERT(error == 0,
2219 	    ("Cannot attach to %s (error=%d).", sc->sc_name, error));
2220 	error = g_access(cp, 1, 0, 0);
2221 	KASSERT(error == 0, ("Cannot open %s (error=%d).", sc->sc_name, error));
2222 	g_topology_unlock();
2223 	sx_xlock(&sc->sc_lock);
2224 
2225 	G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s.", sc->sc_name,
2226 	    g_raid3_get_diskname(disk));
2227 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOFAILSYNC) == 0)
2228 		disk->d_flags |= G_RAID3_DISK_FLAG_DIRTY;
2229 	KASSERT(disk->d_sync.ds_consumer == NULL,
2230 	    ("Sync consumer already exists (device=%s, disk=%s).",
2231 	    sc->sc_name, g_raid3_get_diskname(disk)));
2232 
2233 	disk->d_sync.ds_consumer = cp;
2234 	disk->d_sync.ds_consumer->private = disk;
2235 	disk->d_sync.ds_consumer->index = 0;
2236 	sc->sc_syncdisk = disk;
2237 
2238 	/*
2239 	 * Allocate memory for synchronization bios and initialize them.
2240 	 */
2241 	disk->d_sync.ds_bios = malloc(sizeof(struct bio *) * g_raid3_syncreqs,
2242 	    M_RAID3, M_WAITOK);
2243 	for (n = 0; n < g_raid3_syncreqs; n++) {
2244 		bp = g_alloc_bio();
2245 		disk->d_sync.ds_bios[n] = bp;
2246 		bp->bio_parent = NULL;
2247 		bp->bio_cmd = BIO_READ;
2248 		bp->bio_data = malloc(maxphys, M_RAID3, M_WAITOK);
2249 		bp->bio_cflags = 0;
2250 		bp->bio_offset = disk->d_sync.ds_offset * (sc->sc_ndisks - 1);
2251 		bp->bio_length = MIN(maxphys, sc->sc_mediasize - bp->bio_offset);
2252 		disk->d_sync.ds_offset += bp->bio_length / (sc->sc_ndisks - 1);
2253 		bp->bio_done = g_raid3_sync_done;
2254 		bp->bio_from = disk->d_sync.ds_consumer;
2255 		bp->bio_to = sc->sc_provider;
2256 		bp->bio_caller1 = (void *)(uintptr_t)n;
2257 	}
2258 
2259 	/* Set the number of in-flight synchronization requests. */
2260 	disk->d_sync.ds_inflight = g_raid3_syncreqs;
2261 
2262 	/*
2263 	 * Fire off first synchronization requests.
2264 	 */
2265 	for (n = 0; n < g_raid3_syncreqs; n++) {
2266 		bp = disk->d_sync.ds_bios[n];
2267 		G_RAID3_LOGREQ(3, bp, "Sending synchronization request.");
2268 		disk->d_sync.ds_consumer->index++;
2269 		/*
2270 		 * Delay the request if it is colliding with a regular request.
2271 		 */
2272 		if (g_raid3_regular_collision(sc, bp))
2273 			g_raid3_sync_delay(sc, bp);
2274 		else
2275 			g_io_request(bp, disk->d_sync.ds_consumer);
2276 	}
2277 }
2278 
2279 /*
2280  * Stop synchronization process.
2281  * type: 0 - synchronization finished
2282  *       1 - synchronization stopped
2283  */
2284 static void
2285 g_raid3_sync_stop(struct g_raid3_softc *sc, int type)
2286 {
2287 	struct g_raid3_disk *disk;
2288 	struct g_consumer *cp;
2289 
2290 	g_topology_assert_not();
2291 	sx_assert(&sc->sc_lock, SX_LOCKED);
2292 
2293 	KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED,
2294 	    ("Device not in DEGRADED state (%s, %u).", sc->sc_name,
2295 	    sc->sc_state));
2296 	disk = sc->sc_syncdisk;
2297 	sc->sc_syncdisk = NULL;
2298 	KASSERT(disk != NULL, ("No disk was synchronized (%s).", sc->sc_name));
2299 	KASSERT(disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING,
2300 	    ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk),
2301 	    g_raid3_disk_state2str(disk->d_state)));
2302 	if (disk->d_sync.ds_consumer == NULL)
2303 		return;
2304 
2305 	if (type == 0) {
2306 		G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s finished.",
2307 		    sc->sc_name, g_raid3_get_diskname(disk));
2308 	} else /* if (type == 1) */ {
2309 		G_RAID3_DEBUG(0, "Device %s: rebuilding provider %s stopped.",
2310 		    sc->sc_name, g_raid3_get_diskname(disk));
2311 	}
2312 	free(disk->d_sync.ds_bios, M_RAID3);
2313 	disk->d_sync.ds_bios = NULL;
2314 	cp = disk->d_sync.ds_consumer;
2315 	disk->d_sync.ds_consumer = NULL;
2316 	disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY;
2317 	sx_xunlock(&sc->sc_lock); /* Avoid recursion on sc_lock. */
2318 	g_topology_lock();
2319 	g_raid3_kill_consumer(sc, cp);
2320 	g_topology_unlock();
2321 	sx_xlock(&sc->sc_lock);
2322 }
2323 
2324 static void
2325 g_raid3_launch_provider(struct g_raid3_softc *sc)
2326 {
2327 	struct g_provider *pp;
2328 	struct g_raid3_disk *disk;
2329 	int n;
2330 
2331 	sx_assert(&sc->sc_lock, SX_LOCKED);
2332 
2333 	g_topology_lock();
2334 	pp = g_new_providerf(sc->sc_geom, "raid3/%s", sc->sc_name);
2335 	pp->mediasize = sc->sc_mediasize;
2336 	pp->sectorsize = sc->sc_sectorsize;
2337 	pp->stripesize = 0;
2338 	pp->stripeoffset = 0;
2339 	for (n = 0; n < sc->sc_ndisks; n++) {
2340 		disk = &sc->sc_disks[n];
2341 		if (disk->d_consumer && disk->d_consumer->provider &&
2342 		    disk->d_consumer->provider->stripesize > pp->stripesize) {
2343 			pp->stripesize = disk->d_consumer->provider->stripesize;
2344 			pp->stripeoffset = disk->d_consumer->provider->stripeoffset;
2345 		}
2346 	}
2347 	pp->stripesize *= sc->sc_ndisks - 1;
2348 	pp->stripeoffset *= sc->sc_ndisks - 1;
2349 	sc->sc_provider = pp;
2350 	g_error_provider(pp, 0);
2351 	g_topology_unlock();
2352 	G_RAID3_DEBUG(0, "Device %s launched (%u/%u).", pp->name,
2353 	    g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE), sc->sc_ndisks);
2354 
2355 	if (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED)
2356 		g_raid3_sync_start(sc);
2357 }
2358 
2359 static void
2360 g_raid3_destroy_provider(struct g_raid3_softc *sc)
2361 {
2362 	struct bio *bp;
2363 
2364 	g_topology_assert_not();
2365 	KASSERT(sc->sc_provider != NULL, ("NULL provider (device=%s).",
2366 	    sc->sc_name));
2367 
2368 	g_topology_lock();
2369 	g_error_provider(sc->sc_provider, ENXIO);
2370 	mtx_lock(&sc->sc_queue_mtx);
2371 	while ((bp = bioq_first(&sc->sc_queue)) != NULL) {
2372 		bioq_remove(&sc->sc_queue, bp);
2373 		g_io_deliver(bp, ENXIO);
2374 	}
2375 	mtx_unlock(&sc->sc_queue_mtx);
2376 	G_RAID3_DEBUG(0, "Device %s: provider %s destroyed.", sc->sc_name,
2377 	    sc->sc_provider->name);
2378 	g_wither_provider(sc->sc_provider, ENXIO);
2379 	g_topology_unlock();
2380 	sc->sc_provider = NULL;
2381 	if (sc->sc_syncdisk != NULL)
2382 		g_raid3_sync_stop(sc, 1);
2383 }
2384 
2385 static void
2386 g_raid3_go(void *arg)
2387 {
2388 	struct g_raid3_softc *sc;
2389 	struct g_raid3_event *ep;
2390 
2391 	sc = arg;
2392 	G_RAID3_DEBUG(0, "Force device %s start due to timeout.", sc->sc_name);
2393 	ep = sc->sc_timeout_event;
2394 	sc->sc_timeout_event = NULL;
2395 	g_raid3_event_dispatch(ep, sc, 0,
2396 	    G_RAID3_EVENT_DONTWAIT | G_RAID3_EVENT_DEVICE);
2397 }
2398 
2399 static void
2400 g_raid3_timeout_drain(struct g_raid3_softc *sc)
2401 {
2402 	sx_assert(&sc->sc_lock, SX_XLOCKED);
2403 
2404 	callout_drain(&sc->sc_callout);
2405 	g_raid3_event_free(sc->sc_timeout_event);
2406 	sc->sc_timeout_event = NULL;
2407 }
2408 
2409 static u_int
2410 g_raid3_determine_state(struct g_raid3_disk *disk)
2411 {
2412 	struct g_raid3_softc *sc;
2413 	u_int state;
2414 
2415 	sc = disk->d_softc;
2416 	if (sc->sc_syncid == disk->d_sync.ds_syncid) {
2417 		if ((disk->d_flags &
2418 		    G_RAID3_DISK_FLAG_SYNCHRONIZING) == 0) {
2419 			/* Disk does not need synchronization. */
2420 			state = G_RAID3_DISK_STATE_ACTIVE;
2421 		} else {
2422 			if ((sc->sc_flags &
2423 			     G_RAID3_DEVICE_FLAG_NOAUTOSYNC) == 0 ||
2424 			    (disk->d_flags &
2425 			     G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) {
2426 				/*
2427 				 * We can start synchronization from
2428 				 * the stored offset.
2429 				 */
2430 				state = G_RAID3_DISK_STATE_SYNCHRONIZING;
2431 			} else {
2432 				state = G_RAID3_DISK_STATE_STALE;
2433 			}
2434 		}
2435 	} else if (disk->d_sync.ds_syncid < sc->sc_syncid) {
2436 		/*
2437 		 * Reset all synchronization data for this disk,
2438 		 * because if it even was synchronized, it was
2439 		 * synchronized to disks with different syncid.
2440 		 */
2441 		disk->d_flags |= G_RAID3_DISK_FLAG_SYNCHRONIZING;
2442 		disk->d_sync.ds_offset = 0;
2443 		disk->d_sync.ds_offset_done = 0;
2444 		disk->d_sync.ds_syncid = sc->sc_syncid;
2445 		if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) == 0 ||
2446 		    (disk->d_flags & G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) {
2447 			state = G_RAID3_DISK_STATE_SYNCHRONIZING;
2448 		} else {
2449 			state = G_RAID3_DISK_STATE_STALE;
2450 		}
2451 	} else /* if (sc->sc_syncid < disk->d_sync.ds_syncid) */ {
2452 		/*
2453 		 * Not good, NOT GOOD!
2454 		 * It means that device was started on stale disks
2455 		 * and more fresh disk just arrive.
2456 		 * If there were writes, device is broken, sorry.
2457 		 * I think the best choice here is don't touch
2458 		 * this disk and inform the user loudly.
2459 		 */
2460 		G_RAID3_DEBUG(0, "Device %s was started before the freshest "
2461 		    "disk (%s) arrives!! It will not be connected to the "
2462 		    "running device.", sc->sc_name,
2463 		    g_raid3_get_diskname(disk));
2464 		g_raid3_destroy_disk(disk);
2465 		state = G_RAID3_DISK_STATE_NONE;
2466 		/* Return immediately, because disk was destroyed. */
2467 		return (state);
2468 	}
2469 	G_RAID3_DEBUG(3, "State for %s disk: %s.",
2470 	    g_raid3_get_diskname(disk), g_raid3_disk_state2str(state));
2471 	return (state);
2472 }
2473 
2474 /*
2475  * Update device state.
2476  */
2477 static void
2478 g_raid3_update_device(struct g_raid3_softc *sc, boolean_t force)
2479 {
2480 	struct g_raid3_disk *disk;
2481 	u_int state;
2482 
2483 	sx_assert(&sc->sc_lock, SX_XLOCKED);
2484 
2485 	switch (sc->sc_state) {
2486 	case G_RAID3_DEVICE_STATE_STARTING:
2487 	    {
2488 		u_int n, ndirty, ndisks, genid, syncid;
2489 
2490 		KASSERT(sc->sc_provider == NULL,
2491 		    ("Non-NULL provider in STARTING state (%s).", sc->sc_name));
2492 		/*
2493 		 * Are we ready? We are, if all disks are connected or
2494 		 * one disk is missing and 'force' is true.
2495 		 */
2496 		if (g_raid3_ndisks(sc, -1) + force == sc->sc_ndisks) {
2497 			if (!force)
2498 				g_raid3_timeout_drain(sc);
2499 		} else {
2500 			if (force) {
2501 				/*
2502 				 * Timeout expired, so destroy device.
2503 				 */
2504 				sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY;
2505 				G_RAID3_DEBUG(1, "root_mount_rel[%u] %p",
2506 				    __LINE__, sc->sc_rootmount);
2507 				root_mount_rel(sc->sc_rootmount);
2508 				sc->sc_rootmount = NULL;
2509 			}
2510 			return;
2511 		}
2512 
2513 		/*
2514 		 * Find the biggest genid.
2515 		 */
2516 		genid = 0;
2517 		for (n = 0; n < sc->sc_ndisks; n++) {
2518 			disk = &sc->sc_disks[n];
2519 			if (disk->d_state == G_RAID3_DISK_STATE_NODISK)
2520 				continue;
2521 			if (disk->d_genid > genid)
2522 				genid = disk->d_genid;
2523 		}
2524 		sc->sc_genid = genid;
2525 		/*
2526 		 * Remove all disks without the biggest genid.
2527 		 */
2528 		for (n = 0; n < sc->sc_ndisks; n++) {
2529 			disk = &sc->sc_disks[n];
2530 			if (disk->d_state == G_RAID3_DISK_STATE_NODISK)
2531 				continue;
2532 			if (disk->d_genid < genid) {
2533 				G_RAID3_DEBUG(0,
2534 				    "Component %s (device %s) broken, skipping.",
2535 				    g_raid3_get_diskname(disk), sc->sc_name);
2536 				g_raid3_destroy_disk(disk);
2537 			}
2538 		}
2539 
2540 		/*
2541 		 * There must be at least 'sc->sc_ndisks - 1' components
2542 		 * with the same syncid and without SYNCHRONIZING flag.
2543 		 */
2544 
2545 		/*
2546 		 * Find the biggest syncid, number of valid components and
2547 		 * number of dirty components.
2548 		 */
2549 		ndirty = ndisks = syncid = 0;
2550 		for (n = 0; n < sc->sc_ndisks; n++) {
2551 			disk = &sc->sc_disks[n];
2552 			if (disk->d_state == G_RAID3_DISK_STATE_NODISK)
2553 				continue;
2554 			if ((disk->d_flags & G_RAID3_DISK_FLAG_DIRTY) != 0)
2555 				ndirty++;
2556 			if (disk->d_sync.ds_syncid > syncid) {
2557 				syncid = disk->d_sync.ds_syncid;
2558 				ndisks = 0;
2559 			} else if (disk->d_sync.ds_syncid < syncid) {
2560 				continue;
2561 			}
2562 			if ((disk->d_flags &
2563 			    G_RAID3_DISK_FLAG_SYNCHRONIZING) != 0) {
2564 				continue;
2565 			}
2566 			ndisks++;
2567 		}
2568 		/*
2569 		 * Do we have enough valid components?
2570 		 */
2571 		if (ndisks + 1 < sc->sc_ndisks) {
2572 			G_RAID3_DEBUG(0,
2573 			    "Device %s is broken, too few valid components.",
2574 			    sc->sc_name);
2575 			sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY;
2576 			return;
2577 		}
2578 		/*
2579 		 * If there is one DIRTY component and all disks are present,
2580 		 * mark it for synchronization. If there is more than one DIRTY
2581 		 * component, mark parity component for synchronization.
2582 		 */
2583 		if (ndisks == sc->sc_ndisks && ndirty == 1) {
2584 			for (n = 0; n < sc->sc_ndisks; n++) {
2585 				disk = &sc->sc_disks[n];
2586 				if ((disk->d_flags &
2587 				    G_RAID3_DISK_FLAG_DIRTY) == 0) {
2588 					continue;
2589 				}
2590 				disk->d_flags |=
2591 				    G_RAID3_DISK_FLAG_SYNCHRONIZING;
2592 			}
2593 		} else if (ndisks == sc->sc_ndisks && ndirty > 1) {
2594 			disk = &sc->sc_disks[sc->sc_ndisks - 1];
2595 			disk->d_flags |= G_RAID3_DISK_FLAG_SYNCHRONIZING;
2596 		}
2597 
2598 		sc->sc_syncid = syncid;
2599 		if (force) {
2600 			/* Remember to bump syncid on first write. */
2601 			sc->sc_bump_id |= G_RAID3_BUMP_SYNCID;
2602 		}
2603 		if (ndisks == sc->sc_ndisks)
2604 			state = G_RAID3_DEVICE_STATE_COMPLETE;
2605 		else /* if (ndisks == sc->sc_ndisks - 1) */
2606 			state = G_RAID3_DEVICE_STATE_DEGRADED;
2607 		G_RAID3_DEBUG(1, "Device %s state changed from %s to %s.",
2608 		    sc->sc_name, g_raid3_device_state2str(sc->sc_state),
2609 		    g_raid3_device_state2str(state));
2610 		sc->sc_state = state;
2611 		for (n = 0; n < sc->sc_ndisks; n++) {
2612 			disk = &sc->sc_disks[n];
2613 			if (disk->d_state == G_RAID3_DISK_STATE_NODISK)
2614 				continue;
2615 			state = g_raid3_determine_state(disk);
2616 			g_raid3_event_send(disk, state, G_RAID3_EVENT_DONTWAIT);
2617 			if (state == G_RAID3_DISK_STATE_STALE)
2618 				sc->sc_bump_id |= G_RAID3_BUMP_SYNCID;
2619 		}
2620 		break;
2621 	    }
2622 	case G_RAID3_DEVICE_STATE_DEGRADED:
2623 		/*
2624 		 * Genid need to be bumped immediately, so do it here.
2625 		 */
2626 		if ((sc->sc_bump_id & G_RAID3_BUMP_GENID) != 0) {
2627 			sc->sc_bump_id &= ~G_RAID3_BUMP_GENID;
2628 			g_raid3_bump_genid(sc);
2629 		}
2630 
2631 		if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NEW) > 0)
2632 			return;
2633 		if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) <
2634 		    sc->sc_ndisks - 1) {
2635 			if (sc->sc_provider != NULL)
2636 				g_raid3_destroy_provider(sc);
2637 			sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY;
2638 			return;
2639 		}
2640 		if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) ==
2641 		    sc->sc_ndisks) {
2642 			state = G_RAID3_DEVICE_STATE_COMPLETE;
2643 			G_RAID3_DEBUG(1,
2644 			    "Device %s state changed from %s to %s.",
2645 			    sc->sc_name, g_raid3_device_state2str(sc->sc_state),
2646 			    g_raid3_device_state2str(state));
2647 			sc->sc_state = state;
2648 		}
2649 		if (sc->sc_provider == NULL)
2650 			g_raid3_launch_provider(sc);
2651 		if (sc->sc_rootmount != NULL) {
2652 			G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__,
2653 			    sc->sc_rootmount);
2654 			root_mount_rel(sc->sc_rootmount);
2655 			sc->sc_rootmount = NULL;
2656 		}
2657 		break;
2658 	case G_RAID3_DEVICE_STATE_COMPLETE:
2659 		/*
2660 		 * Genid need to be bumped immediately, so do it here.
2661 		 */
2662 		if ((sc->sc_bump_id & G_RAID3_BUMP_GENID) != 0) {
2663 			sc->sc_bump_id &= ~G_RAID3_BUMP_GENID;
2664 			g_raid3_bump_genid(sc);
2665 		}
2666 
2667 		if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NEW) > 0)
2668 			return;
2669 		KASSERT(g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) >=
2670 		    sc->sc_ndisks - 1,
2671 		    ("Too few ACTIVE components in COMPLETE state (device %s).",
2672 		    sc->sc_name));
2673 		if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) ==
2674 		    sc->sc_ndisks - 1) {
2675 			state = G_RAID3_DEVICE_STATE_DEGRADED;
2676 			G_RAID3_DEBUG(1,
2677 			    "Device %s state changed from %s to %s.",
2678 			    sc->sc_name, g_raid3_device_state2str(sc->sc_state),
2679 			    g_raid3_device_state2str(state));
2680 			sc->sc_state = state;
2681 		}
2682 		if (sc->sc_provider == NULL)
2683 			g_raid3_launch_provider(sc);
2684 		if (sc->sc_rootmount != NULL) {
2685 			G_RAID3_DEBUG(1, "root_mount_rel[%u] %p", __LINE__,
2686 			    sc->sc_rootmount);
2687 			root_mount_rel(sc->sc_rootmount);
2688 			sc->sc_rootmount = NULL;
2689 		}
2690 		break;
2691 	default:
2692 		KASSERT(1 == 0, ("Wrong device state (%s, %s).", sc->sc_name,
2693 		    g_raid3_device_state2str(sc->sc_state)));
2694 		break;
2695 	}
2696 }
2697 
2698 /*
2699  * Update disk state and device state if needed.
2700  */
2701 #define	DISK_STATE_CHANGED()	G_RAID3_DEBUG(1,			\
2702 	"Disk %s state changed from %s to %s (device %s).",		\
2703 	g_raid3_get_diskname(disk),					\
2704 	g_raid3_disk_state2str(disk->d_state),				\
2705 	g_raid3_disk_state2str(state), sc->sc_name)
2706 static int
2707 g_raid3_update_disk(struct g_raid3_disk *disk, u_int state)
2708 {
2709 	struct g_raid3_softc *sc;
2710 
2711 	sc = disk->d_softc;
2712 	sx_assert(&sc->sc_lock, SX_XLOCKED);
2713 
2714 again:
2715 	G_RAID3_DEBUG(3, "Changing disk %s state from %s to %s.",
2716 	    g_raid3_get_diskname(disk), g_raid3_disk_state2str(disk->d_state),
2717 	    g_raid3_disk_state2str(state));
2718 	switch (state) {
2719 	case G_RAID3_DISK_STATE_NEW:
2720 		/*
2721 		 * Possible scenarios:
2722 		 * 1. New disk arrive.
2723 		 */
2724 		/* Previous state should be NONE. */
2725 		KASSERT(disk->d_state == G_RAID3_DISK_STATE_NONE,
2726 		    ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk),
2727 		    g_raid3_disk_state2str(disk->d_state)));
2728 		DISK_STATE_CHANGED();
2729 
2730 		disk->d_state = state;
2731 		G_RAID3_DEBUG(1, "Device %s: provider %s detected.",
2732 		    sc->sc_name, g_raid3_get_diskname(disk));
2733 		if (sc->sc_state == G_RAID3_DEVICE_STATE_STARTING)
2734 			break;
2735 		KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED ||
2736 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE,
2737 		    ("Wrong device state (%s, %s, %s, %s).", sc->sc_name,
2738 		    g_raid3_device_state2str(sc->sc_state),
2739 		    g_raid3_get_diskname(disk),
2740 		    g_raid3_disk_state2str(disk->d_state)));
2741 		state = g_raid3_determine_state(disk);
2742 		if (state != G_RAID3_DISK_STATE_NONE)
2743 			goto again;
2744 		break;
2745 	case G_RAID3_DISK_STATE_ACTIVE:
2746 		/*
2747 		 * Possible scenarios:
2748 		 * 1. New disk does not need synchronization.
2749 		 * 2. Synchronization process finished successfully.
2750 		 */
2751 		KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED ||
2752 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE,
2753 		    ("Wrong device state (%s, %s, %s, %s).", sc->sc_name,
2754 		    g_raid3_device_state2str(sc->sc_state),
2755 		    g_raid3_get_diskname(disk),
2756 		    g_raid3_disk_state2str(disk->d_state)));
2757 		/* Previous state should be NEW or SYNCHRONIZING. */
2758 		KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW ||
2759 		    disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING,
2760 		    ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk),
2761 		    g_raid3_disk_state2str(disk->d_state)));
2762 		DISK_STATE_CHANGED();
2763 
2764 		if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) {
2765 			disk->d_flags &= ~G_RAID3_DISK_FLAG_SYNCHRONIZING;
2766 			disk->d_flags &= ~G_RAID3_DISK_FLAG_FORCE_SYNC;
2767 			g_raid3_sync_stop(sc, 0);
2768 		}
2769 		disk->d_state = state;
2770 		disk->d_sync.ds_offset = 0;
2771 		disk->d_sync.ds_offset_done = 0;
2772 		g_raid3_update_idle(sc, disk);
2773 		g_raid3_update_metadata(disk);
2774 		G_RAID3_DEBUG(1, "Device %s: provider %s activated.",
2775 		    sc->sc_name, g_raid3_get_diskname(disk));
2776 		break;
2777 	case G_RAID3_DISK_STATE_STALE:
2778 		/*
2779 		 * Possible scenarios:
2780 		 * 1. Stale disk was connected.
2781 		 */
2782 		/* Previous state should be NEW. */
2783 		KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW,
2784 		    ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk),
2785 		    g_raid3_disk_state2str(disk->d_state)));
2786 		KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED ||
2787 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE,
2788 		    ("Wrong device state (%s, %s, %s, %s).", sc->sc_name,
2789 		    g_raid3_device_state2str(sc->sc_state),
2790 		    g_raid3_get_diskname(disk),
2791 		    g_raid3_disk_state2str(disk->d_state)));
2792 		/*
2793 		 * STALE state is only possible if device is marked
2794 		 * NOAUTOSYNC.
2795 		 */
2796 		KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) != 0,
2797 		    ("Wrong device state (%s, %s, %s, %s).", sc->sc_name,
2798 		    g_raid3_device_state2str(sc->sc_state),
2799 		    g_raid3_get_diskname(disk),
2800 		    g_raid3_disk_state2str(disk->d_state)));
2801 		DISK_STATE_CHANGED();
2802 
2803 		disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY;
2804 		disk->d_state = state;
2805 		g_raid3_update_metadata(disk);
2806 		G_RAID3_DEBUG(0, "Device %s: provider %s is stale.",
2807 		    sc->sc_name, g_raid3_get_diskname(disk));
2808 		break;
2809 	case G_RAID3_DISK_STATE_SYNCHRONIZING:
2810 		/*
2811 		 * Possible scenarios:
2812 		 * 1. Disk which needs synchronization was connected.
2813 		 */
2814 		/* Previous state should be NEW. */
2815 		KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW,
2816 		    ("Wrong disk state (%s, %s).", g_raid3_get_diskname(disk),
2817 		    g_raid3_disk_state2str(disk->d_state)));
2818 		KASSERT(sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED ||
2819 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE,
2820 		    ("Wrong device state (%s, %s, %s, %s).", sc->sc_name,
2821 		    g_raid3_device_state2str(sc->sc_state),
2822 		    g_raid3_get_diskname(disk),
2823 		    g_raid3_disk_state2str(disk->d_state)));
2824 		DISK_STATE_CHANGED();
2825 
2826 		if (disk->d_state == G_RAID3_DISK_STATE_NEW)
2827 			disk->d_flags &= ~G_RAID3_DISK_FLAG_DIRTY;
2828 		disk->d_state = state;
2829 		if (sc->sc_provider != NULL) {
2830 			g_raid3_sync_start(sc);
2831 			g_raid3_update_metadata(disk);
2832 		}
2833 		break;
2834 	case G_RAID3_DISK_STATE_DISCONNECTED:
2835 		/*
2836 		 * Possible scenarios:
2837 		 * 1. Device wasn't running yet, but disk disappear.
2838 		 * 2. Disk was active and disapppear.
2839 		 * 3. Disk disappear during synchronization process.
2840 		 */
2841 		if (sc->sc_state == G_RAID3_DEVICE_STATE_DEGRADED ||
2842 		    sc->sc_state == G_RAID3_DEVICE_STATE_COMPLETE) {
2843 			/*
2844 			 * Previous state should be ACTIVE, STALE or
2845 			 * SYNCHRONIZING.
2846 			 */
2847 			KASSERT(disk->d_state == G_RAID3_DISK_STATE_ACTIVE ||
2848 			    disk->d_state == G_RAID3_DISK_STATE_STALE ||
2849 			    disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING,
2850 			    ("Wrong disk state (%s, %s).",
2851 			    g_raid3_get_diskname(disk),
2852 			    g_raid3_disk_state2str(disk->d_state)));
2853 		} else if (sc->sc_state == G_RAID3_DEVICE_STATE_STARTING) {
2854 			/* Previous state should be NEW. */
2855 			KASSERT(disk->d_state == G_RAID3_DISK_STATE_NEW,
2856 			    ("Wrong disk state (%s, %s).",
2857 			    g_raid3_get_diskname(disk),
2858 			    g_raid3_disk_state2str(disk->d_state)));
2859 			/*
2860 			 * Reset bumping syncid if disk disappeared in STARTING
2861 			 * state.
2862 			 */
2863 			if ((sc->sc_bump_id & G_RAID3_BUMP_SYNCID) != 0)
2864 				sc->sc_bump_id &= ~G_RAID3_BUMP_SYNCID;
2865 #ifdef	INVARIANTS
2866 		} else {
2867 			KASSERT(1 == 0, ("Wrong device state (%s, %s, %s, %s).",
2868 			    sc->sc_name,
2869 			    g_raid3_device_state2str(sc->sc_state),
2870 			    g_raid3_get_diskname(disk),
2871 			    g_raid3_disk_state2str(disk->d_state)));
2872 #endif
2873 		}
2874 		DISK_STATE_CHANGED();
2875 		G_RAID3_DEBUG(0, "Device %s: provider %s disconnected.",
2876 		    sc->sc_name, g_raid3_get_diskname(disk));
2877 
2878 		g_raid3_destroy_disk(disk);
2879 		break;
2880 	default:
2881 		KASSERT(1 == 0, ("Unknown state (%u).", state));
2882 		break;
2883 	}
2884 	return (0);
2885 }
2886 #undef	DISK_STATE_CHANGED
2887 
2888 int
2889 g_raid3_read_metadata(struct g_consumer *cp, struct g_raid3_metadata *md)
2890 {
2891 	struct g_provider *pp;
2892 	u_char *buf;
2893 	int error;
2894 
2895 	g_topology_assert();
2896 
2897 	error = g_access(cp, 1, 0, 0);
2898 	if (error != 0)
2899 		return (error);
2900 	pp = cp->provider;
2901 	g_topology_unlock();
2902 	/* Metadata are stored on last sector. */
2903 	buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize,
2904 	    &error);
2905 	g_topology_lock();
2906 	g_access(cp, -1, 0, 0);
2907 	if (buf == NULL) {
2908 		G_RAID3_DEBUG(1, "Cannot read metadata from %s (error=%d).",
2909 		    cp->provider->name, error);
2910 		return (error);
2911 	}
2912 
2913 	/* Decode metadata. */
2914 	error = raid3_metadata_decode(buf, md);
2915 	g_free(buf);
2916 	if (strcmp(md->md_magic, G_RAID3_MAGIC) != 0)
2917 		return (EINVAL);
2918 	if (md->md_version > G_RAID3_VERSION) {
2919 		G_RAID3_DEBUG(0,
2920 		    "Kernel module is too old to handle metadata from %s.",
2921 		    cp->provider->name);
2922 		return (EINVAL);
2923 	}
2924 	if (error != 0) {
2925 		G_RAID3_DEBUG(1, "MD5 metadata hash mismatch for provider %s.",
2926 		    cp->provider->name);
2927 		return (error);
2928 	}
2929 	if (md->md_sectorsize > maxphys) {
2930 		G_RAID3_DEBUG(0, "The blocksize is too big.");
2931 		return (EINVAL);
2932 	}
2933 
2934 	return (0);
2935 }
2936 
2937 static int
2938 g_raid3_check_metadata(struct g_raid3_softc *sc, struct g_provider *pp,
2939     struct g_raid3_metadata *md)
2940 {
2941 
2942 	if (md->md_no >= sc->sc_ndisks) {
2943 		G_RAID3_DEBUG(1, "Invalid disk %s number (no=%u), skipping.",
2944 		    pp->name, md->md_no);
2945 		return (EINVAL);
2946 	}
2947 	if (sc->sc_disks[md->md_no].d_state != G_RAID3_DISK_STATE_NODISK) {
2948 		G_RAID3_DEBUG(1, "Disk %s (no=%u) already exists, skipping.",
2949 		    pp->name, md->md_no);
2950 		return (EEXIST);
2951 	}
2952 	if (md->md_all != sc->sc_ndisks) {
2953 		G_RAID3_DEBUG(1,
2954 		    "Invalid '%s' field on disk %s (device %s), skipping.",
2955 		    "md_all", pp->name, sc->sc_name);
2956 		return (EINVAL);
2957 	}
2958 	if ((md->md_mediasize % md->md_sectorsize) != 0) {
2959 		G_RAID3_DEBUG(1, "Invalid metadata (mediasize %% sectorsize != "
2960 		    "0) on disk %s (device %s), skipping.", pp->name,
2961 		    sc->sc_name);
2962 		return (EINVAL);
2963 	}
2964 	if (md->md_mediasize != sc->sc_mediasize) {
2965 		G_RAID3_DEBUG(1,
2966 		    "Invalid '%s' field on disk %s (device %s), skipping.",
2967 		    "md_mediasize", pp->name, sc->sc_name);
2968 		return (EINVAL);
2969 	}
2970 	if ((md->md_mediasize % (sc->sc_ndisks - 1)) != 0) {
2971 		G_RAID3_DEBUG(1,
2972 		    "Invalid '%s' field on disk %s (device %s), skipping.",
2973 		    "md_mediasize", pp->name, sc->sc_name);
2974 		return (EINVAL);
2975 	}
2976 	if ((sc->sc_mediasize / (sc->sc_ndisks - 1)) > pp->mediasize) {
2977 		G_RAID3_DEBUG(1,
2978 		    "Invalid size of disk %s (device %s), skipping.", pp->name,
2979 		    sc->sc_name);
2980 		return (EINVAL);
2981 	}
2982 	if ((md->md_sectorsize / pp->sectorsize) < sc->sc_ndisks - 1) {
2983 		G_RAID3_DEBUG(1,
2984 		    "Invalid '%s' field on disk %s (device %s), skipping.",
2985 		    "md_sectorsize", pp->name, sc->sc_name);
2986 		return (EINVAL);
2987 	}
2988 	if (md->md_sectorsize != sc->sc_sectorsize) {
2989 		G_RAID3_DEBUG(1,
2990 		    "Invalid '%s' field on disk %s (device %s), skipping.",
2991 		    "md_sectorsize", pp->name, sc->sc_name);
2992 		return (EINVAL);
2993 	}
2994 	if ((sc->sc_sectorsize % pp->sectorsize) != 0) {
2995 		G_RAID3_DEBUG(1,
2996 		    "Invalid sector size of disk %s (device %s), skipping.",
2997 		    pp->name, sc->sc_name);
2998 		return (EINVAL);
2999 	}
3000 	if ((md->md_mflags & ~G_RAID3_DEVICE_FLAG_MASK) != 0) {
3001 		G_RAID3_DEBUG(1,
3002 		    "Invalid device flags on disk %s (device %s), skipping.",
3003 		    pp->name, sc->sc_name);
3004 		return (EINVAL);
3005 	}
3006 	if ((md->md_mflags & G_RAID3_DEVICE_FLAG_VERIFY) != 0 &&
3007 	    (md->md_mflags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0) {
3008 		/*
3009 		 * VERIFY and ROUND-ROBIN options are mutally exclusive.
3010 		 */
3011 		G_RAID3_DEBUG(1, "Both VERIFY and ROUND-ROBIN flags exist on "
3012 		    "disk %s (device %s), skipping.", pp->name, sc->sc_name);
3013 		return (EINVAL);
3014 	}
3015 	if ((md->md_dflags & ~G_RAID3_DISK_FLAG_MASK) != 0) {
3016 		G_RAID3_DEBUG(1,
3017 		    "Invalid disk flags on disk %s (device %s), skipping.",
3018 		    pp->name, sc->sc_name);
3019 		return (EINVAL);
3020 	}
3021 	return (0);
3022 }
3023 
3024 int
3025 g_raid3_add_disk(struct g_raid3_softc *sc, struct g_provider *pp,
3026     struct g_raid3_metadata *md)
3027 {
3028 	struct g_raid3_disk *disk;
3029 	int error;
3030 
3031 	g_topology_assert_not();
3032 	G_RAID3_DEBUG(2, "Adding disk %s.", pp->name);
3033 
3034 	error = g_raid3_check_metadata(sc, pp, md);
3035 	if (error != 0)
3036 		return (error);
3037 	if (sc->sc_state != G_RAID3_DEVICE_STATE_STARTING &&
3038 	    md->md_genid < sc->sc_genid) {
3039 		G_RAID3_DEBUG(0, "Component %s (device %s) broken, skipping.",
3040 		    pp->name, sc->sc_name);
3041 		return (EINVAL);
3042 	}
3043 	disk = g_raid3_init_disk(sc, pp, md, &error);
3044 	if (disk == NULL)
3045 		return (error);
3046 	error = g_raid3_event_send(disk, G_RAID3_DISK_STATE_NEW,
3047 	    G_RAID3_EVENT_WAIT);
3048 	if (error != 0)
3049 		return (error);
3050 	if (md->md_version < G_RAID3_VERSION) {
3051 		G_RAID3_DEBUG(0, "Upgrading metadata on %s (v%d->v%d).",
3052 		    pp->name, md->md_version, G_RAID3_VERSION);
3053 		g_raid3_update_metadata(disk);
3054 	}
3055 	return (0);
3056 }
3057 
3058 static void
3059 g_raid3_destroy_delayed(void *arg, int flag)
3060 {
3061 	struct g_raid3_softc *sc;
3062 	int error;
3063 
3064 	if (flag == EV_CANCEL) {
3065 		G_RAID3_DEBUG(1, "Destroying canceled.");
3066 		return;
3067 	}
3068 	sc = arg;
3069 	g_topology_unlock();
3070 	sx_xlock(&sc->sc_lock);
3071 	KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) == 0,
3072 	    ("DESTROY flag set on %s.", sc->sc_name));
3073 	KASSERT((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROYING) != 0,
3074 	    ("DESTROYING flag not set on %s.", sc->sc_name));
3075 	G_RAID3_DEBUG(0, "Destroying %s (delayed).", sc->sc_name);
3076 	error = g_raid3_destroy(sc, G_RAID3_DESTROY_SOFT);
3077 	if (error != 0) {
3078 		G_RAID3_DEBUG(0, "Cannot destroy %s.", sc->sc_name);
3079 		sx_xunlock(&sc->sc_lock);
3080 	}
3081 	g_topology_lock();
3082 }
3083 
3084 static int
3085 g_raid3_access(struct g_provider *pp, int acr, int acw, int ace)
3086 {
3087 	struct g_raid3_softc *sc;
3088 	int dcr, dcw, dce, error = 0;
3089 
3090 	g_topology_assert();
3091 	G_RAID3_DEBUG(2, "Access request for %s: r%dw%de%d.", pp->name, acr,
3092 	    acw, ace);
3093 
3094 	sc = pp->geom->softc;
3095 	if (sc == NULL && acr <= 0 && acw <= 0 && ace <= 0)
3096 		return (0);
3097 	KASSERT(sc != NULL, ("NULL softc (provider=%s).", pp->name));
3098 
3099 	dcr = pp->acr + acr;
3100 	dcw = pp->acw + acw;
3101 	dce = pp->ace + ace;
3102 
3103 	g_topology_unlock();
3104 	sx_xlock(&sc->sc_lock);
3105 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROY) != 0 ||
3106 	    g_raid3_ndisks(sc, G_RAID3_DISK_STATE_ACTIVE) < sc->sc_ndisks - 1) {
3107 		if (acr > 0 || acw > 0 || ace > 0)
3108 			error = ENXIO;
3109 		goto end;
3110 	}
3111 	if (dcw == 0)
3112 		g_raid3_idle(sc, dcw);
3113 	if ((sc->sc_flags & G_RAID3_DEVICE_FLAG_DESTROYING) != 0) {
3114 		if (acr > 0 || acw > 0 || ace > 0) {
3115 			error = ENXIO;
3116 			goto end;
3117 		}
3118 		if (dcr == 0 && dcw == 0 && dce == 0) {
3119 			g_post_event(g_raid3_destroy_delayed, sc, M_WAITOK,
3120 			    sc, NULL);
3121 		}
3122 	}
3123 end:
3124 	sx_xunlock(&sc->sc_lock);
3125 	g_topology_lock();
3126 	return (error);
3127 }
3128 
3129 static struct g_geom *
3130 g_raid3_create(struct g_class *mp, const struct g_raid3_metadata *md)
3131 {
3132 	struct g_raid3_softc *sc;
3133 	struct g_geom *gp;
3134 	int error, timeout;
3135 	u_int n;
3136 
3137 	g_topology_assert();
3138 	G_RAID3_DEBUG(1, "Creating device %s (id=%u).", md->md_name, md->md_id);
3139 
3140 	/* One disk is minimum. */
3141 	if (md->md_all < 1)
3142 		return (NULL);
3143 	/*
3144 	 * Action geom.
3145 	 */
3146 	gp = g_new_geomf(mp, "%s", md->md_name);
3147 	sc = malloc(sizeof(*sc), M_RAID3, M_WAITOK | M_ZERO);
3148 	sc->sc_disks = malloc(sizeof(struct g_raid3_disk) * md->md_all, M_RAID3,
3149 	    M_WAITOK | M_ZERO);
3150 	gp->start = g_raid3_start;
3151 	gp->orphan = g_raid3_orphan;
3152 	gp->access = g_raid3_access;
3153 	gp->dumpconf = g_raid3_dumpconf;
3154 
3155 	sc->sc_id = md->md_id;
3156 	sc->sc_mediasize = md->md_mediasize;
3157 	sc->sc_sectorsize = md->md_sectorsize;
3158 	sc->sc_ndisks = md->md_all;
3159 	sc->sc_round_robin = 0;
3160 	sc->sc_flags = md->md_mflags;
3161 	sc->sc_bump_id = 0;
3162 	sc->sc_idle = 1;
3163 	sc->sc_last_write = time_uptime;
3164 	sc->sc_writes = 0;
3165 	for (n = 0; n < sc->sc_ndisks; n++) {
3166 		sc->sc_disks[n].d_softc = sc;
3167 		sc->sc_disks[n].d_no = n;
3168 		sc->sc_disks[n].d_state = G_RAID3_DISK_STATE_NODISK;
3169 	}
3170 	sx_init(&sc->sc_lock, "graid3:lock");
3171 	bioq_init(&sc->sc_queue);
3172 	mtx_init(&sc->sc_queue_mtx, "graid3:queue", NULL, MTX_DEF);
3173 	bioq_init(&sc->sc_regular_delayed);
3174 	bioq_init(&sc->sc_inflight);
3175 	bioq_init(&sc->sc_sync_delayed);
3176 	TAILQ_INIT(&sc->sc_events);
3177 	mtx_init(&sc->sc_events_mtx, "graid3:events", NULL, MTX_DEF);
3178 	callout_init(&sc->sc_callout, 1);
3179 	sc->sc_state = G_RAID3_DEVICE_STATE_STARTING;
3180 	gp->softc = sc;
3181 	sc->sc_geom = gp;
3182 	sc->sc_provider = NULL;
3183 	/*
3184 	 * Synchronization geom.
3185 	 */
3186 	gp = g_new_geomf(mp, "%s.sync", md->md_name);
3187 	gp->softc = sc;
3188 	gp->orphan = g_raid3_orphan;
3189 	sc->sc_sync.ds_geom = gp;
3190 
3191 	if (!g_raid3_use_malloc) {
3192 		sc->sc_zones[G_RAID3_ZONE_64K].sz_zone = uma_zcreate("gr3:64k",
3193 		    65536, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL,
3194 		    UMA_ALIGN_PTR, 0);
3195 		sc->sc_zones[G_RAID3_ZONE_64K].sz_inuse = 0;
3196 		sc->sc_zones[G_RAID3_ZONE_64K].sz_max = g_raid3_n64k;
3197 		sc->sc_zones[G_RAID3_ZONE_64K].sz_requested =
3198 		    sc->sc_zones[G_RAID3_ZONE_64K].sz_failed = 0;
3199 		sc->sc_zones[G_RAID3_ZONE_16K].sz_zone = uma_zcreate("gr3:16k",
3200 		    16384, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL,
3201 		    UMA_ALIGN_PTR, 0);
3202 		sc->sc_zones[G_RAID3_ZONE_16K].sz_inuse = 0;
3203 		sc->sc_zones[G_RAID3_ZONE_16K].sz_max = g_raid3_n16k;
3204 		sc->sc_zones[G_RAID3_ZONE_16K].sz_requested =
3205 		    sc->sc_zones[G_RAID3_ZONE_16K].sz_failed = 0;
3206 		sc->sc_zones[G_RAID3_ZONE_4K].sz_zone = uma_zcreate("gr3:4k",
3207 		    4096, g_raid3_uma_ctor, g_raid3_uma_dtor, NULL, NULL,
3208 		    UMA_ALIGN_PTR, 0);
3209 		sc->sc_zones[G_RAID3_ZONE_4K].sz_inuse = 0;
3210 		sc->sc_zones[G_RAID3_ZONE_4K].sz_max = g_raid3_n4k;
3211 		sc->sc_zones[G_RAID3_ZONE_4K].sz_requested =
3212 		    sc->sc_zones[G_RAID3_ZONE_4K].sz_failed = 0;
3213 	}
3214 
3215 	error = kproc_create(g_raid3_worker, sc, &sc->sc_worker, 0, 0,
3216 	    "g_raid3 %s", md->md_name);
3217 	if (error != 0) {
3218 		G_RAID3_DEBUG(1, "Cannot create kernel thread for %s.",
3219 		    sc->sc_name);
3220 		if (!g_raid3_use_malloc) {
3221 			uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_64K].sz_zone);
3222 			uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_16K].sz_zone);
3223 			uma_zdestroy(sc->sc_zones[G_RAID3_ZONE_4K].sz_zone);
3224 		}
3225 		g_destroy_geom(sc->sc_sync.ds_geom);
3226 		mtx_destroy(&sc->sc_events_mtx);
3227 		mtx_destroy(&sc->sc_queue_mtx);
3228 		sx_destroy(&sc->sc_lock);
3229 		g_destroy_geom(sc->sc_geom);
3230 		free(sc->sc_disks, M_RAID3);
3231 		free(sc, M_RAID3);
3232 		return (NULL);
3233 	}
3234 
3235 	G_RAID3_DEBUG(1, "Device %s created (%u components, id=%u).",
3236 	    sc->sc_name, sc->sc_ndisks, sc->sc_id);
3237 
3238 	sc->sc_rootmount = root_mount_hold("GRAID3");
3239 	G_RAID3_DEBUG(1, "root_mount_hold %p", sc->sc_rootmount);
3240 
3241 	/*
3242 	 * Schedule startup timeout.
3243 	 */
3244 	timeout = atomic_load_acq_int(&g_raid3_timeout);
3245 	sc->sc_timeout_event = malloc(sizeof(struct g_raid3_event), M_RAID3,
3246 	    M_WAITOK);
3247 	callout_reset(&sc->sc_callout, timeout * hz, g_raid3_go, sc);
3248 	return (sc->sc_geom);
3249 }
3250 
3251 int
3252 g_raid3_destroy(struct g_raid3_softc *sc, int how)
3253 {
3254 	struct g_provider *pp;
3255 
3256 	g_topology_assert_not();
3257 	if (sc == NULL)
3258 		return (ENXIO);
3259 	sx_assert(&sc->sc_lock, SX_XLOCKED);
3260 
3261 	pp = sc->sc_provider;
3262 	if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) {
3263 		switch (how) {
3264 		case G_RAID3_DESTROY_SOFT:
3265 			G_RAID3_DEBUG(1,
3266 			    "Device %s is still open (r%dw%de%d).", pp->name,
3267 			    pp->acr, pp->acw, pp->ace);
3268 			return (EBUSY);
3269 		case G_RAID3_DESTROY_DELAYED:
3270 			G_RAID3_DEBUG(1,
3271 			    "Device %s will be destroyed on last close.",
3272 			    pp->name);
3273 			if (sc->sc_syncdisk != NULL)
3274 				g_raid3_sync_stop(sc, 1);
3275 			sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROYING;
3276 			return (EBUSY);
3277 		case G_RAID3_DESTROY_HARD:
3278 			G_RAID3_DEBUG(1, "Device %s is still open, so it "
3279 			    "can't be definitely removed.", pp->name);
3280 			break;
3281 		}
3282 	}
3283 
3284 	g_topology_lock();
3285 	if (sc->sc_geom->softc == NULL) {
3286 		g_topology_unlock();
3287 		return (0);
3288 	}
3289 	sc->sc_geom->softc = NULL;
3290 	sc->sc_sync.ds_geom->softc = NULL;
3291 	g_topology_unlock();
3292 
3293 	sc->sc_flags |= G_RAID3_DEVICE_FLAG_DESTROY;
3294 	sc->sc_flags |= G_RAID3_DEVICE_FLAG_WAIT;
3295 	G_RAID3_DEBUG(4, "%s: Waking up %p.", __func__, sc);
3296 	sx_xunlock(&sc->sc_lock);
3297 	mtx_lock(&sc->sc_queue_mtx);
3298 	wakeup(sc);
3299 	wakeup(&sc->sc_queue);
3300 	mtx_unlock(&sc->sc_queue_mtx);
3301 	G_RAID3_DEBUG(4, "%s: Sleeping %p.", __func__, &sc->sc_worker);
3302 	while (sc->sc_worker != NULL)
3303 		tsleep(&sc->sc_worker, PRIBIO, "r3:destroy", hz / 5);
3304 	G_RAID3_DEBUG(4, "%s: Woken up %p.", __func__, &sc->sc_worker);
3305 	sx_xlock(&sc->sc_lock);
3306 	g_raid3_destroy_device(sc);
3307 	free(sc->sc_disks, M_RAID3);
3308 	free(sc, M_RAID3);
3309 	return (0);
3310 }
3311 
3312 static void
3313 g_raid3_taste_orphan(struct g_consumer *cp)
3314 {
3315 
3316 	KASSERT(1 == 0, ("%s called while tasting %s.", __func__,
3317 	    cp->provider->name));
3318 }
3319 
3320 static struct g_geom *
3321 g_raid3_taste(struct g_class *mp, struct g_provider *pp, int flags __unused)
3322 {
3323 	struct g_raid3_metadata md;
3324 	struct g_raid3_softc *sc;
3325 	struct g_consumer *cp;
3326 	struct g_geom *gp;
3327 	int error;
3328 
3329 	g_topology_assert();
3330 	g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name);
3331 	G_RAID3_DEBUG(2, "Tasting %s.", pp->name);
3332 
3333 	gp = g_new_geomf(mp, "raid3:taste");
3334 	/* This orphan function should be never called. */
3335 	gp->orphan = g_raid3_taste_orphan;
3336 	cp = g_new_consumer(gp);
3337 	cp->flags |= G_CF_DIRECT_SEND | G_CF_DIRECT_RECEIVE;
3338 	error = g_attach(cp, pp);
3339 	if (error == 0) {
3340 		error = g_raid3_read_metadata(cp, &md);
3341 		g_detach(cp);
3342 	}
3343 	g_destroy_consumer(cp);
3344 	g_destroy_geom(gp);
3345 	if (error != 0)
3346 		return (NULL);
3347 	gp = NULL;
3348 
3349 	if (md.md_provider[0] != '\0' &&
3350 	    !g_compare_names(md.md_provider, pp->name))
3351 		return (NULL);
3352 	if (md.md_provsize != 0 && md.md_provsize != pp->mediasize)
3353 		return (NULL);
3354 	if (g_raid3_debug >= 2)
3355 		raid3_metadata_dump(&md);
3356 
3357 	/*
3358 	 * Let's check if device already exists.
3359 	 */
3360 	sc = NULL;
3361 	LIST_FOREACH(gp, &mp->geom, geom) {
3362 		sc = gp->softc;
3363 		if (sc == NULL)
3364 			continue;
3365 		if (sc->sc_sync.ds_geom == gp)
3366 			continue;
3367 		if (strcmp(md.md_name, sc->sc_name) != 0)
3368 			continue;
3369 		if (md.md_id != sc->sc_id) {
3370 			G_RAID3_DEBUG(0, "Device %s already configured.",
3371 			    sc->sc_name);
3372 			return (NULL);
3373 		}
3374 		break;
3375 	}
3376 	if (gp == NULL) {
3377 		gp = g_raid3_create(mp, &md);
3378 		if (gp == NULL) {
3379 			G_RAID3_DEBUG(0, "Cannot create device %s.",
3380 			    md.md_name);
3381 			return (NULL);
3382 		}
3383 		sc = gp->softc;
3384 	}
3385 	G_RAID3_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name);
3386 	g_topology_unlock();
3387 	sx_xlock(&sc->sc_lock);
3388 	error = g_raid3_add_disk(sc, pp, &md);
3389 	if (error != 0) {
3390 		G_RAID3_DEBUG(0, "Cannot add disk %s to %s (error=%d).",
3391 		    pp->name, gp->name, error);
3392 		if (g_raid3_ndisks(sc, G_RAID3_DISK_STATE_NODISK) ==
3393 		    sc->sc_ndisks) {
3394 			g_cancel_event(sc);
3395 			g_raid3_destroy(sc, G_RAID3_DESTROY_HARD);
3396 			g_topology_lock();
3397 			return (NULL);
3398 		}
3399 		gp = NULL;
3400 	}
3401 	sx_xunlock(&sc->sc_lock);
3402 	g_topology_lock();
3403 	return (gp);
3404 }
3405 
3406 static int
3407 g_raid3_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused,
3408     struct g_geom *gp)
3409 {
3410 	struct g_raid3_softc *sc;
3411 	int error;
3412 
3413 	g_topology_unlock();
3414 	sc = gp->softc;
3415 	sx_xlock(&sc->sc_lock);
3416 	g_cancel_event(sc);
3417 	error = g_raid3_destroy(gp->softc, G_RAID3_DESTROY_SOFT);
3418 	if (error != 0)
3419 		sx_xunlock(&sc->sc_lock);
3420 	g_topology_lock();
3421 	return (error);
3422 }
3423 
3424 static void
3425 g_raid3_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp,
3426     struct g_consumer *cp, struct g_provider *pp)
3427 {
3428 	struct g_raid3_softc *sc;
3429 
3430 	g_topology_assert();
3431 
3432 	sc = gp->softc;
3433 	if (sc == NULL)
3434 		return;
3435 	/* Skip synchronization geom. */
3436 	if (gp == sc->sc_sync.ds_geom)
3437 		return;
3438 	if (pp != NULL) {
3439 		/* Nothing here. */
3440 	} else if (cp != NULL) {
3441 		struct g_raid3_disk *disk;
3442 
3443 		disk = cp->private;
3444 		if (disk == NULL)
3445 			return;
3446 		g_topology_unlock();
3447 		sx_xlock(&sc->sc_lock);
3448 		sbuf_printf(sb, "%s<Type>", indent);
3449 		if (disk->d_no == sc->sc_ndisks - 1)
3450 			sbuf_cat(sb, "PARITY");
3451 		else
3452 			sbuf_cat(sb, "DATA");
3453 		sbuf_cat(sb, "</Type>\n");
3454 		sbuf_printf(sb, "%s<Number>%u</Number>\n", indent,
3455 		    (u_int)disk->d_no);
3456 		if (disk->d_state == G_RAID3_DISK_STATE_SYNCHRONIZING) {
3457 			sbuf_printf(sb, "%s<Synchronized>", indent);
3458 			if (disk->d_sync.ds_offset == 0)
3459 				sbuf_cat(sb, "0%");
3460 			else {
3461 				sbuf_printf(sb, "%u%%",
3462 				    (u_int)((disk->d_sync.ds_offset * 100) /
3463 				    (sc->sc_mediasize / (sc->sc_ndisks - 1))));
3464 			}
3465 			sbuf_cat(sb, "</Synchronized>\n");
3466 			if (disk->d_sync.ds_offset > 0) {
3467 				sbuf_printf(sb, "%s<BytesSynced>%jd"
3468 				    "</BytesSynced>\n", indent,
3469 				    (intmax_t)disk->d_sync.ds_offset);
3470 			}
3471 		}
3472 		sbuf_printf(sb, "%s<SyncID>%u</SyncID>\n", indent,
3473 		    disk->d_sync.ds_syncid);
3474 		sbuf_printf(sb, "%s<GenID>%u</GenID>\n", indent, disk->d_genid);
3475 		sbuf_printf(sb, "%s<Flags>", indent);
3476 		if (disk->d_flags == 0)
3477 			sbuf_cat(sb, "NONE");
3478 		else {
3479 			int first = 1;
3480 
3481 #define	ADD_FLAG(flag, name)	do {					\
3482 	if ((disk->d_flags & (flag)) != 0) {				\
3483 		if (!first)						\
3484 			sbuf_cat(sb, ", ");				\
3485 		else							\
3486 			first = 0;					\
3487 		sbuf_cat(sb, name);					\
3488 	}								\
3489 } while (0)
3490 			ADD_FLAG(G_RAID3_DISK_FLAG_DIRTY, "DIRTY");
3491 			ADD_FLAG(G_RAID3_DISK_FLAG_HARDCODED, "HARDCODED");
3492 			ADD_FLAG(G_RAID3_DISK_FLAG_SYNCHRONIZING,
3493 			    "SYNCHRONIZING");
3494 			ADD_FLAG(G_RAID3_DISK_FLAG_FORCE_SYNC, "FORCE_SYNC");
3495 			ADD_FLAG(G_RAID3_DISK_FLAG_BROKEN, "BROKEN");
3496 #undef	ADD_FLAG
3497 		}
3498 		sbuf_cat(sb, "</Flags>\n");
3499 		sbuf_printf(sb, "%s<State>%s</State>\n", indent,
3500 		    g_raid3_disk_state2str(disk->d_state));
3501 		sx_xunlock(&sc->sc_lock);
3502 		g_topology_lock();
3503 	} else {
3504 		g_topology_unlock();
3505 		sx_xlock(&sc->sc_lock);
3506 		if (!g_raid3_use_malloc) {
3507 			sbuf_printf(sb,
3508 			    "%s<Zone4kRequested>%u</Zone4kRequested>\n", indent,
3509 			    sc->sc_zones[G_RAID3_ZONE_4K].sz_requested);
3510 			sbuf_printf(sb,
3511 			    "%s<Zone4kFailed>%u</Zone4kFailed>\n", indent,
3512 			    sc->sc_zones[G_RAID3_ZONE_4K].sz_failed);
3513 			sbuf_printf(sb,
3514 			    "%s<Zone16kRequested>%u</Zone16kRequested>\n", indent,
3515 			    sc->sc_zones[G_RAID3_ZONE_16K].sz_requested);
3516 			sbuf_printf(sb,
3517 			    "%s<Zone16kFailed>%u</Zone16kFailed>\n", indent,
3518 			    sc->sc_zones[G_RAID3_ZONE_16K].sz_failed);
3519 			sbuf_printf(sb,
3520 			    "%s<Zone64kRequested>%u</Zone64kRequested>\n", indent,
3521 			    sc->sc_zones[G_RAID3_ZONE_64K].sz_requested);
3522 			sbuf_printf(sb,
3523 			    "%s<Zone64kFailed>%u</Zone64kFailed>\n", indent,
3524 			    sc->sc_zones[G_RAID3_ZONE_64K].sz_failed);
3525 		}
3526 		sbuf_printf(sb, "%s<ID>%u</ID>\n", indent, (u_int)sc->sc_id);
3527 		sbuf_printf(sb, "%s<SyncID>%u</SyncID>\n", indent, sc->sc_syncid);
3528 		sbuf_printf(sb, "%s<GenID>%u</GenID>\n", indent, sc->sc_genid);
3529 		sbuf_printf(sb, "%s<Flags>", indent);
3530 		if (sc->sc_flags == 0)
3531 			sbuf_cat(sb, "NONE");
3532 		else {
3533 			int first = 1;
3534 
3535 #define	ADD_FLAG(flag, name)	do {					\
3536 	if ((sc->sc_flags & (flag)) != 0) {				\
3537 		if (!first)						\
3538 			sbuf_cat(sb, ", ");				\
3539 		else							\
3540 			first = 0;					\
3541 		sbuf_cat(sb, name);					\
3542 	}								\
3543 } while (0)
3544 			ADD_FLAG(G_RAID3_DEVICE_FLAG_NOFAILSYNC, "NOFAILSYNC");
3545 			ADD_FLAG(G_RAID3_DEVICE_FLAG_NOAUTOSYNC, "NOAUTOSYNC");
3546 			ADD_FLAG(G_RAID3_DEVICE_FLAG_ROUND_ROBIN,
3547 			    "ROUND-ROBIN");
3548 			ADD_FLAG(G_RAID3_DEVICE_FLAG_VERIFY, "VERIFY");
3549 #undef	ADD_FLAG
3550 		}
3551 		sbuf_cat(sb, "</Flags>\n");
3552 		sbuf_printf(sb, "%s<Components>%u</Components>\n", indent,
3553 		    sc->sc_ndisks);
3554 		sbuf_printf(sb, "%s<State>%s</State>\n", indent,
3555 		    g_raid3_device_state2str(sc->sc_state));
3556 		sx_xunlock(&sc->sc_lock);
3557 		g_topology_lock();
3558 	}
3559 }
3560 
3561 static void
3562 g_raid3_shutdown_post_sync(void *arg, int howto)
3563 {
3564 	struct g_class *mp;
3565 	struct g_geom *gp, *gp2;
3566 	struct g_raid3_softc *sc;
3567 	int error;
3568 
3569 	mp = arg;
3570 	g_topology_lock();
3571 	g_raid3_shutdown = 1;
3572 	LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) {
3573 		if ((sc = gp->softc) == NULL)
3574 			continue;
3575 		/* Skip synchronization geom. */
3576 		if (gp == sc->sc_sync.ds_geom)
3577 			continue;
3578 		g_topology_unlock();
3579 		sx_xlock(&sc->sc_lock);
3580 		g_raid3_idle(sc, -1);
3581 		g_cancel_event(sc);
3582 		error = g_raid3_destroy(sc, G_RAID3_DESTROY_DELAYED);
3583 		if (error != 0)
3584 			sx_xunlock(&sc->sc_lock);
3585 		g_topology_lock();
3586 	}
3587 	g_topology_unlock();
3588 }
3589 
3590 static void
3591 g_raid3_init(struct g_class *mp)
3592 {
3593 
3594 	g_raid3_post_sync = EVENTHANDLER_REGISTER(shutdown_post_sync,
3595 	    g_raid3_shutdown_post_sync, mp, SHUTDOWN_PRI_FIRST);
3596 	if (g_raid3_post_sync == NULL)
3597 		G_RAID3_DEBUG(0, "Warning! Cannot register shutdown event.");
3598 }
3599 
3600 static void
3601 g_raid3_fini(struct g_class *mp)
3602 {
3603 
3604 	if (g_raid3_post_sync != NULL)
3605 		EVENTHANDLER_DEREGISTER(shutdown_post_sync, g_raid3_post_sync);
3606 }
3607 
3608 DECLARE_GEOM_CLASS(g_raid3_class, g_raid3);
3609 MODULE_VERSION(geom_raid3, 0);
3610