xref: /illumos-gate/usr/src/uts/common/fs/zfs/dsl_dataset.c (revision fb2a9bae0030340ad72b9c26ba1ffee2ee3cafec)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*
22  * Copyright 2010 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 #include <sys/dmu_objset.h>
27 #include <sys/dsl_dataset.h>
28 #include <sys/dsl_dir.h>
29 #include <sys/dsl_prop.h>
30 #include <sys/dsl_synctask.h>
31 #include <sys/dmu_traverse.h>
32 #include <sys/dmu_tx.h>
33 #include <sys/arc.h>
34 #include <sys/zio.h>
35 #include <sys/zap.h>
36 #include <sys/unique.h>
37 #include <sys/zfs_context.h>
38 #include <sys/zfs_ioctl.h>
39 #include <sys/spa.h>
40 #include <sys/zfs_znode.h>
41 #include <sys/zvol.h>
42 
43 static char *dsl_reaper = "the grim reaper";
44 
45 static dsl_checkfunc_t dsl_dataset_destroy_begin_check;
46 static dsl_syncfunc_t dsl_dataset_destroy_begin_sync;
47 static dsl_syncfunc_t dsl_dataset_set_reservation_sync;
48 
49 #define	DS_REF_MAX	(1ULL << 62)
50 
51 #define	DSL_DEADLIST_BLOCKSIZE	SPA_MAXBLOCKSIZE
52 
53 #define	DSL_DATASET_IS_DESTROYED(ds)	((ds)->ds_owner == dsl_reaper)
54 
55 
56 /*
57  * Figure out how much of this delta should be propogated to the dsl_dir
58  * layer.  If there's a refreservation, that space has already been
59  * partially accounted for in our ancestors.
60  */
61 static int64_t
62 parent_delta(dsl_dataset_t *ds, int64_t delta)
63 {
64 	uint64_t old_bytes, new_bytes;
65 
66 	if (ds->ds_reserved == 0)
67 		return (delta);
68 
69 	old_bytes = MAX(ds->ds_phys->ds_unique_bytes, ds->ds_reserved);
70 	new_bytes = MAX(ds->ds_phys->ds_unique_bytes + delta, ds->ds_reserved);
71 
72 	ASSERT3U(ABS((int64_t)(new_bytes - old_bytes)), <=, ABS(delta));
73 	return (new_bytes - old_bytes);
74 }
75 
76 void
77 dsl_dataset_block_born(dsl_dataset_t *ds, const blkptr_t *bp, dmu_tx_t *tx)
78 {
79 	int used = bp_get_dsize_sync(tx->tx_pool->dp_spa, bp);
80 	int compressed = BP_GET_PSIZE(bp);
81 	int uncompressed = BP_GET_UCSIZE(bp);
82 	int64_t delta;
83 
84 	dprintf_bp(bp, "born, ds=%p\n", ds);
85 
86 	ASSERT(dmu_tx_is_syncing(tx));
87 	/* It could have been compressed away to nothing */
88 	if (BP_IS_HOLE(bp))
89 		return;
90 	ASSERT(BP_GET_TYPE(bp) != DMU_OT_NONE);
91 	ASSERT3U(BP_GET_TYPE(bp), <, DMU_OT_NUMTYPES);
92 	if (ds == NULL) {
93 		/*
94 		 * Account for the meta-objset space in its placeholder
95 		 * dsl_dir.
96 		 */
97 		ASSERT3U(compressed, ==, uncompressed); /* it's all metadata */
98 		dsl_dir_diduse_space(tx->tx_pool->dp_mos_dir, DD_USED_HEAD,
99 		    used, compressed, uncompressed, tx);
100 		dsl_dir_dirty(tx->tx_pool->dp_mos_dir, tx);
101 		return;
102 	}
103 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
104 	mutex_enter(&ds->ds_dir->dd_lock);
105 	mutex_enter(&ds->ds_lock);
106 	delta = parent_delta(ds, used);
107 	ds->ds_phys->ds_used_bytes += used;
108 	ds->ds_phys->ds_compressed_bytes += compressed;
109 	ds->ds_phys->ds_uncompressed_bytes += uncompressed;
110 	ds->ds_phys->ds_unique_bytes += used;
111 	mutex_exit(&ds->ds_lock);
112 	dsl_dir_diduse_space(ds->ds_dir, DD_USED_HEAD, delta,
113 	    compressed, uncompressed, tx);
114 	dsl_dir_transfer_space(ds->ds_dir, used - delta,
115 	    DD_USED_REFRSRV, DD_USED_HEAD, tx);
116 	mutex_exit(&ds->ds_dir->dd_lock);
117 }
118 
119 int
120 dsl_dataset_block_kill(dsl_dataset_t *ds, const blkptr_t *bp, dmu_tx_t *tx,
121     boolean_t async)
122 {
123 	if (BP_IS_HOLE(bp))
124 		return (0);
125 
126 	ASSERT(dmu_tx_is_syncing(tx));
127 	ASSERT(bp->blk_birth <= tx->tx_txg);
128 
129 	int used = bp_get_dsize_sync(tx->tx_pool->dp_spa, bp);
130 	int compressed = BP_GET_PSIZE(bp);
131 	int uncompressed = BP_GET_UCSIZE(bp);
132 
133 	ASSERT(used > 0);
134 	if (ds == NULL) {
135 		/*
136 		 * Account for the meta-objset space in its placeholder
137 		 * dataset.
138 		 */
139 		dsl_free(tx->tx_pool, tx->tx_txg, bp);
140 
141 		dsl_dir_diduse_space(tx->tx_pool->dp_mos_dir, DD_USED_HEAD,
142 		    -used, -compressed, -uncompressed, tx);
143 		dsl_dir_dirty(tx->tx_pool->dp_mos_dir, tx);
144 		return (used);
145 	}
146 	ASSERT3P(tx->tx_pool, ==, ds->ds_dir->dd_pool);
147 
148 	ASSERT(!dsl_dataset_is_snapshot(ds));
149 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
150 
151 	if (bp->blk_birth > ds->ds_phys->ds_prev_snap_txg) {
152 		int64_t delta;
153 
154 		dprintf_bp(bp, "freeing: %s", "");
155 		dsl_free(tx->tx_pool, tx->tx_txg, bp);
156 
157 		mutex_enter(&ds->ds_dir->dd_lock);
158 		mutex_enter(&ds->ds_lock);
159 		ASSERT(ds->ds_phys->ds_unique_bytes >= used ||
160 		    !DS_UNIQUE_IS_ACCURATE(ds));
161 		delta = parent_delta(ds, -used);
162 		ds->ds_phys->ds_unique_bytes -= used;
163 		mutex_exit(&ds->ds_lock);
164 		dsl_dir_diduse_space(ds->ds_dir, DD_USED_HEAD,
165 		    delta, -compressed, -uncompressed, tx);
166 		dsl_dir_transfer_space(ds->ds_dir, -used - delta,
167 		    DD_USED_REFRSRV, DD_USED_HEAD, tx);
168 		mutex_exit(&ds->ds_dir->dd_lock);
169 	} else {
170 		dprintf_bp(bp, "putting on dead list: %s", "");
171 		if (async) {
172 			/*
173 			 * We are here as part of zio's write done callback,
174 			 * which means we're a zio interrupt thread.  We can't
175 			 * call bplist_enqueue() now because it may block
176 			 * waiting for I/O.  Instead, put bp on the deferred
177 			 * queue and let dsl_pool_sync() finish the job.
178 			 */
179 			bplist_enqueue_deferred(&ds->ds_deadlist, bp);
180 		} else {
181 			VERIFY(0 == bplist_enqueue(&ds->ds_deadlist, bp, tx));
182 		}
183 		ASSERT3U(ds->ds_prev->ds_object, ==,
184 		    ds->ds_phys->ds_prev_snap_obj);
185 		ASSERT(ds->ds_prev->ds_phys->ds_num_children > 0);
186 		/* if (bp->blk_birth > prev prev snap txg) prev unique += bs */
187 		if (ds->ds_prev->ds_phys->ds_next_snap_obj ==
188 		    ds->ds_object && bp->blk_birth >
189 		    ds->ds_prev->ds_phys->ds_prev_snap_txg) {
190 			dmu_buf_will_dirty(ds->ds_prev->ds_dbuf, tx);
191 			mutex_enter(&ds->ds_prev->ds_lock);
192 			ds->ds_prev->ds_phys->ds_unique_bytes += used;
193 			mutex_exit(&ds->ds_prev->ds_lock);
194 		}
195 		if (bp->blk_birth > ds->ds_origin_txg) {
196 			dsl_dir_transfer_space(ds->ds_dir, used,
197 			    DD_USED_HEAD, DD_USED_SNAP, tx);
198 		}
199 	}
200 	mutex_enter(&ds->ds_lock);
201 	ASSERT3U(ds->ds_phys->ds_used_bytes, >=, used);
202 	ds->ds_phys->ds_used_bytes -= used;
203 	ASSERT3U(ds->ds_phys->ds_compressed_bytes, >=, compressed);
204 	ds->ds_phys->ds_compressed_bytes -= compressed;
205 	ASSERT3U(ds->ds_phys->ds_uncompressed_bytes, >=, uncompressed);
206 	ds->ds_phys->ds_uncompressed_bytes -= uncompressed;
207 	mutex_exit(&ds->ds_lock);
208 
209 	return (used);
210 }
211 
212 uint64_t
213 dsl_dataset_prev_snap_txg(dsl_dataset_t *ds)
214 {
215 	uint64_t trysnap = 0;
216 
217 	if (ds == NULL)
218 		return (0);
219 	/*
220 	 * The snapshot creation could fail, but that would cause an
221 	 * incorrect FALSE return, which would only result in an
222 	 * overestimation of the amount of space that an operation would
223 	 * consume, which is OK.
224 	 *
225 	 * There's also a small window where we could miss a pending
226 	 * snapshot, because we could set the sync task in the quiescing
227 	 * phase.  So this should only be used as a guess.
228 	 */
229 	if (ds->ds_trysnap_txg >
230 	    spa_last_synced_txg(ds->ds_dir->dd_pool->dp_spa))
231 		trysnap = ds->ds_trysnap_txg;
232 	return (MAX(ds->ds_phys->ds_prev_snap_txg, trysnap));
233 }
234 
235 boolean_t
236 dsl_dataset_block_freeable(dsl_dataset_t *ds, uint64_t blk_birth)
237 {
238 	return (blk_birth > dsl_dataset_prev_snap_txg(ds));
239 }
240 
241 /* ARGSUSED */
242 static void
243 dsl_dataset_evict(dmu_buf_t *db, void *dsv)
244 {
245 	dsl_dataset_t *ds = dsv;
246 
247 	ASSERT(ds->ds_owner == NULL || DSL_DATASET_IS_DESTROYED(ds));
248 
249 	unique_remove(ds->ds_fsid_guid);
250 
251 	if (ds->ds_objset != NULL)
252 		dmu_objset_evict(ds->ds_objset);
253 
254 	if (ds->ds_prev) {
255 		dsl_dataset_drop_ref(ds->ds_prev, ds);
256 		ds->ds_prev = NULL;
257 	}
258 
259 	bplist_close(&ds->ds_deadlist);
260 	if (ds->ds_dir)
261 		dsl_dir_close(ds->ds_dir, ds);
262 
263 	ASSERT(!list_link_active(&ds->ds_synced_link));
264 
265 	mutex_destroy(&ds->ds_lock);
266 	mutex_destroy(&ds->ds_recvlock);
267 	mutex_destroy(&ds->ds_opening_lock);
268 	rw_destroy(&ds->ds_rwlock);
269 	cv_destroy(&ds->ds_exclusive_cv);
270 	bplist_fini(&ds->ds_deadlist);
271 
272 	kmem_free(ds, sizeof (dsl_dataset_t));
273 }
274 
275 static int
276 dsl_dataset_get_snapname(dsl_dataset_t *ds)
277 {
278 	dsl_dataset_phys_t *headphys;
279 	int err;
280 	dmu_buf_t *headdbuf;
281 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
282 	objset_t *mos = dp->dp_meta_objset;
283 
284 	if (ds->ds_snapname[0])
285 		return (0);
286 	if (ds->ds_phys->ds_next_snap_obj == 0)
287 		return (0);
288 
289 	err = dmu_bonus_hold(mos, ds->ds_dir->dd_phys->dd_head_dataset_obj,
290 	    FTAG, &headdbuf);
291 	if (err)
292 		return (err);
293 	headphys = headdbuf->db_data;
294 	err = zap_value_search(dp->dp_meta_objset,
295 	    headphys->ds_snapnames_zapobj, ds->ds_object, 0, ds->ds_snapname);
296 	dmu_buf_rele(headdbuf, FTAG);
297 	return (err);
298 }
299 
300 static int
301 dsl_dataset_snap_lookup(dsl_dataset_t *ds, const char *name, uint64_t *value)
302 {
303 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
304 	uint64_t snapobj = ds->ds_phys->ds_snapnames_zapobj;
305 	matchtype_t mt;
306 	int err;
307 
308 	if (ds->ds_phys->ds_flags & DS_FLAG_CI_DATASET)
309 		mt = MT_FIRST;
310 	else
311 		mt = MT_EXACT;
312 
313 	err = zap_lookup_norm(mos, snapobj, name, 8, 1,
314 	    value, mt, NULL, 0, NULL);
315 	if (err == ENOTSUP && mt == MT_FIRST)
316 		err = zap_lookup(mos, snapobj, name, 8, 1, value);
317 	return (err);
318 }
319 
320 static int
321 dsl_dataset_snap_remove(dsl_dataset_t *ds, char *name, dmu_tx_t *tx)
322 {
323 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
324 	uint64_t snapobj = ds->ds_phys->ds_snapnames_zapobj;
325 	matchtype_t mt;
326 	int err;
327 
328 	dsl_dir_snap_cmtime_update(ds->ds_dir);
329 
330 	if (ds->ds_phys->ds_flags & DS_FLAG_CI_DATASET)
331 		mt = MT_FIRST;
332 	else
333 		mt = MT_EXACT;
334 
335 	err = zap_remove_norm(mos, snapobj, name, mt, tx);
336 	if (err == ENOTSUP && mt == MT_FIRST)
337 		err = zap_remove(mos, snapobj, name, tx);
338 	return (err);
339 }
340 
341 static int
342 dsl_dataset_get_ref(dsl_pool_t *dp, uint64_t dsobj, void *tag,
343     dsl_dataset_t **dsp)
344 {
345 	objset_t *mos = dp->dp_meta_objset;
346 	dmu_buf_t *dbuf;
347 	dsl_dataset_t *ds;
348 	int err;
349 
350 	ASSERT(RW_LOCK_HELD(&dp->dp_config_rwlock) ||
351 	    dsl_pool_sync_context(dp));
352 
353 	err = dmu_bonus_hold(mos, dsobj, tag, &dbuf);
354 	if (err)
355 		return (err);
356 	ds = dmu_buf_get_user(dbuf);
357 	if (ds == NULL) {
358 		dsl_dataset_t *winner;
359 
360 		ds = kmem_zalloc(sizeof (dsl_dataset_t), KM_SLEEP);
361 		ds->ds_dbuf = dbuf;
362 		ds->ds_object = dsobj;
363 		ds->ds_phys = dbuf->db_data;
364 
365 		mutex_init(&ds->ds_lock, NULL, MUTEX_DEFAULT, NULL);
366 		mutex_init(&ds->ds_recvlock, NULL, MUTEX_DEFAULT, NULL);
367 		mutex_init(&ds->ds_opening_lock, NULL, MUTEX_DEFAULT, NULL);
368 		rw_init(&ds->ds_rwlock, 0, 0, 0);
369 		cv_init(&ds->ds_exclusive_cv, NULL, CV_DEFAULT, NULL);
370 		bplist_init(&ds->ds_deadlist);
371 
372 		err = bplist_open(&ds->ds_deadlist,
373 		    mos, ds->ds_phys->ds_deadlist_obj);
374 		if (err == 0) {
375 			err = dsl_dir_open_obj(dp,
376 			    ds->ds_phys->ds_dir_obj, NULL, ds, &ds->ds_dir);
377 		}
378 		if (err) {
379 			/*
380 			 * we don't really need to close the blist if we
381 			 * just opened it.
382 			 */
383 			mutex_destroy(&ds->ds_lock);
384 			mutex_destroy(&ds->ds_recvlock);
385 			mutex_destroy(&ds->ds_opening_lock);
386 			rw_destroy(&ds->ds_rwlock);
387 			cv_destroy(&ds->ds_exclusive_cv);
388 			bplist_fini(&ds->ds_deadlist);
389 			kmem_free(ds, sizeof (dsl_dataset_t));
390 			dmu_buf_rele(dbuf, tag);
391 			return (err);
392 		}
393 
394 		if (!dsl_dataset_is_snapshot(ds)) {
395 			ds->ds_snapname[0] = '\0';
396 			if (ds->ds_phys->ds_prev_snap_obj) {
397 				err = dsl_dataset_get_ref(dp,
398 				    ds->ds_phys->ds_prev_snap_obj,
399 				    ds, &ds->ds_prev);
400 			}
401 
402 			if (err == 0 && dsl_dir_is_clone(ds->ds_dir)) {
403 				dsl_dataset_t *origin;
404 
405 				err = dsl_dataset_hold_obj(dp,
406 				    ds->ds_dir->dd_phys->dd_origin_obj,
407 				    FTAG, &origin);
408 				if (err == 0) {
409 					ds->ds_origin_txg =
410 					    origin->ds_phys->ds_creation_txg;
411 					dsl_dataset_rele(origin, FTAG);
412 				}
413 			}
414 		} else {
415 			if (zfs_flags & ZFS_DEBUG_SNAPNAMES)
416 				err = dsl_dataset_get_snapname(ds);
417 			if (err == 0 && ds->ds_phys->ds_userrefs_obj != 0) {
418 				err = zap_count(
419 				    ds->ds_dir->dd_pool->dp_meta_objset,
420 				    ds->ds_phys->ds_userrefs_obj,
421 				    &ds->ds_userrefs);
422 			}
423 		}
424 
425 		if (err == 0 && !dsl_dataset_is_snapshot(ds)) {
426 			/*
427 			 * In sync context, we're called with either no lock
428 			 * or with the write lock.  If we're not syncing,
429 			 * we're always called with the read lock held.
430 			 */
431 			boolean_t need_lock =
432 			    !RW_WRITE_HELD(&dp->dp_config_rwlock) &&
433 			    dsl_pool_sync_context(dp);
434 
435 			if (need_lock)
436 				rw_enter(&dp->dp_config_rwlock, RW_READER);
437 
438 			err = dsl_prop_get_ds(ds,
439 			    "refreservation", sizeof (uint64_t), 1,
440 			    &ds->ds_reserved, NULL);
441 			if (err == 0) {
442 				err = dsl_prop_get_ds(ds,
443 				    "refquota", sizeof (uint64_t), 1,
444 				    &ds->ds_quota, NULL);
445 			}
446 
447 			if (need_lock)
448 				rw_exit(&dp->dp_config_rwlock);
449 		} else {
450 			ds->ds_reserved = ds->ds_quota = 0;
451 		}
452 
453 		if (err == 0) {
454 			winner = dmu_buf_set_user_ie(dbuf, ds, &ds->ds_phys,
455 			    dsl_dataset_evict);
456 		}
457 		if (err || winner) {
458 			bplist_close(&ds->ds_deadlist);
459 			if (ds->ds_prev)
460 				dsl_dataset_drop_ref(ds->ds_prev, ds);
461 			dsl_dir_close(ds->ds_dir, ds);
462 			mutex_destroy(&ds->ds_lock);
463 			mutex_destroy(&ds->ds_recvlock);
464 			mutex_destroy(&ds->ds_opening_lock);
465 			rw_destroy(&ds->ds_rwlock);
466 			cv_destroy(&ds->ds_exclusive_cv);
467 			bplist_fini(&ds->ds_deadlist);
468 			kmem_free(ds, sizeof (dsl_dataset_t));
469 			if (err) {
470 				dmu_buf_rele(dbuf, tag);
471 				return (err);
472 			}
473 			ds = winner;
474 		} else {
475 			ds->ds_fsid_guid =
476 			    unique_insert(ds->ds_phys->ds_fsid_guid);
477 		}
478 	}
479 	ASSERT3P(ds->ds_dbuf, ==, dbuf);
480 	ASSERT3P(ds->ds_phys, ==, dbuf->db_data);
481 	ASSERT(ds->ds_phys->ds_prev_snap_obj != 0 ||
482 	    spa_version(dp->dp_spa) < SPA_VERSION_ORIGIN ||
483 	    dp->dp_origin_snap == NULL || ds == dp->dp_origin_snap);
484 	mutex_enter(&ds->ds_lock);
485 	if (!dsl_pool_sync_context(dp) && DSL_DATASET_IS_DESTROYED(ds)) {
486 		mutex_exit(&ds->ds_lock);
487 		dmu_buf_rele(ds->ds_dbuf, tag);
488 		return (ENOENT);
489 	}
490 	mutex_exit(&ds->ds_lock);
491 	*dsp = ds;
492 	return (0);
493 }
494 
495 static int
496 dsl_dataset_hold_ref(dsl_dataset_t *ds, void *tag)
497 {
498 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
499 
500 	/*
501 	 * In syncing context we don't want the rwlock lock: there
502 	 * may be an existing writer waiting for sync phase to
503 	 * finish.  We don't need to worry about such writers, since
504 	 * sync phase is single-threaded, so the writer can't be
505 	 * doing anything while we are active.
506 	 */
507 	if (dsl_pool_sync_context(dp)) {
508 		ASSERT(!DSL_DATASET_IS_DESTROYED(ds));
509 		return (0);
510 	}
511 
512 	/*
513 	 * Normal users will hold the ds_rwlock as a READER until they
514 	 * are finished (i.e., call dsl_dataset_rele()).  "Owners" will
515 	 * drop their READER lock after they set the ds_owner field.
516 	 *
517 	 * If the dataset is being destroyed, the destroy thread will
518 	 * obtain a WRITER lock for exclusive access after it's done its
519 	 * open-context work and then change the ds_owner to
520 	 * dsl_reaper once destruction is assured.  So threads
521 	 * may block here temporarily, until the "destructability" of
522 	 * the dataset is determined.
523 	 */
524 	ASSERT(!RW_WRITE_HELD(&dp->dp_config_rwlock));
525 	mutex_enter(&ds->ds_lock);
526 	while (!rw_tryenter(&ds->ds_rwlock, RW_READER)) {
527 		rw_exit(&dp->dp_config_rwlock);
528 		cv_wait(&ds->ds_exclusive_cv, &ds->ds_lock);
529 		if (DSL_DATASET_IS_DESTROYED(ds)) {
530 			mutex_exit(&ds->ds_lock);
531 			dsl_dataset_drop_ref(ds, tag);
532 			rw_enter(&dp->dp_config_rwlock, RW_READER);
533 			return (ENOENT);
534 		}
535 		/*
536 		 * The dp_config_rwlock lives above the ds_lock. And
537 		 * we need to check DSL_DATASET_IS_DESTROYED() while
538 		 * holding the ds_lock, so we have to drop and reacquire
539 		 * the ds_lock here.
540 		 */
541 		mutex_exit(&ds->ds_lock);
542 		rw_enter(&dp->dp_config_rwlock, RW_READER);
543 		mutex_enter(&ds->ds_lock);
544 	}
545 	mutex_exit(&ds->ds_lock);
546 	return (0);
547 }
548 
549 int
550 dsl_dataset_hold_obj(dsl_pool_t *dp, uint64_t dsobj, void *tag,
551     dsl_dataset_t **dsp)
552 {
553 	int err = dsl_dataset_get_ref(dp, dsobj, tag, dsp);
554 
555 	if (err)
556 		return (err);
557 	return (dsl_dataset_hold_ref(*dsp, tag));
558 }
559 
560 int
561 dsl_dataset_own_obj(dsl_pool_t *dp, uint64_t dsobj, boolean_t inconsistentok,
562     void *tag, dsl_dataset_t **dsp)
563 {
564 	int err = dsl_dataset_hold_obj(dp, dsobj, tag, dsp);
565 	if (err)
566 		return (err);
567 	if (!dsl_dataset_tryown(*dsp, inconsistentok, tag)) {
568 		dsl_dataset_rele(*dsp, tag);
569 		*dsp = NULL;
570 		return (EBUSY);
571 	}
572 	return (0);
573 }
574 
575 int
576 dsl_dataset_hold(const char *name, void *tag, dsl_dataset_t **dsp)
577 {
578 	dsl_dir_t *dd;
579 	dsl_pool_t *dp;
580 	const char *snapname;
581 	uint64_t obj;
582 	int err = 0;
583 
584 	err = dsl_dir_open_spa(NULL, name, FTAG, &dd, &snapname);
585 	if (err)
586 		return (err);
587 
588 	dp = dd->dd_pool;
589 	obj = dd->dd_phys->dd_head_dataset_obj;
590 	rw_enter(&dp->dp_config_rwlock, RW_READER);
591 	if (obj)
592 		err = dsl_dataset_get_ref(dp, obj, tag, dsp);
593 	else
594 		err = ENOENT;
595 	if (err)
596 		goto out;
597 
598 	err = dsl_dataset_hold_ref(*dsp, tag);
599 
600 	/* we may be looking for a snapshot */
601 	if (err == 0 && snapname != NULL) {
602 		dsl_dataset_t *ds = NULL;
603 
604 		if (*snapname++ != '@') {
605 			dsl_dataset_rele(*dsp, tag);
606 			err = ENOENT;
607 			goto out;
608 		}
609 
610 		dprintf("looking for snapshot '%s'\n", snapname);
611 		err = dsl_dataset_snap_lookup(*dsp, snapname, &obj);
612 		if (err == 0)
613 			err = dsl_dataset_get_ref(dp, obj, tag, &ds);
614 		dsl_dataset_rele(*dsp, tag);
615 
616 		ASSERT3U((err == 0), ==, (ds != NULL));
617 
618 		if (ds) {
619 			mutex_enter(&ds->ds_lock);
620 			if (ds->ds_snapname[0] == 0)
621 				(void) strlcpy(ds->ds_snapname, snapname,
622 				    sizeof (ds->ds_snapname));
623 			mutex_exit(&ds->ds_lock);
624 			err = dsl_dataset_hold_ref(ds, tag);
625 			*dsp = err ? NULL : ds;
626 		}
627 	}
628 out:
629 	rw_exit(&dp->dp_config_rwlock);
630 	dsl_dir_close(dd, FTAG);
631 	return (err);
632 }
633 
634 int
635 dsl_dataset_own(const char *name, boolean_t inconsistentok,
636     void *tag, dsl_dataset_t **dsp)
637 {
638 	int err = dsl_dataset_hold(name, tag, dsp);
639 	if (err)
640 		return (err);
641 	if (!dsl_dataset_tryown(*dsp, inconsistentok, tag)) {
642 		dsl_dataset_rele(*dsp, tag);
643 		return (EBUSY);
644 	}
645 	return (0);
646 }
647 
648 void
649 dsl_dataset_name(dsl_dataset_t *ds, char *name)
650 {
651 	if (ds == NULL) {
652 		(void) strcpy(name, "mos");
653 	} else {
654 		dsl_dir_name(ds->ds_dir, name);
655 		VERIFY(0 == dsl_dataset_get_snapname(ds));
656 		if (ds->ds_snapname[0]) {
657 			(void) strcat(name, "@");
658 			/*
659 			 * We use a "recursive" mutex so that we
660 			 * can call dprintf_ds() with ds_lock held.
661 			 */
662 			if (!MUTEX_HELD(&ds->ds_lock)) {
663 				mutex_enter(&ds->ds_lock);
664 				(void) strcat(name, ds->ds_snapname);
665 				mutex_exit(&ds->ds_lock);
666 			} else {
667 				(void) strcat(name, ds->ds_snapname);
668 			}
669 		}
670 	}
671 }
672 
673 static int
674 dsl_dataset_namelen(dsl_dataset_t *ds)
675 {
676 	int result;
677 
678 	if (ds == NULL) {
679 		result = 3;	/* "mos" */
680 	} else {
681 		result = dsl_dir_namelen(ds->ds_dir);
682 		VERIFY(0 == dsl_dataset_get_snapname(ds));
683 		if (ds->ds_snapname[0]) {
684 			++result;	/* adding one for the @-sign */
685 			if (!MUTEX_HELD(&ds->ds_lock)) {
686 				mutex_enter(&ds->ds_lock);
687 				result += strlen(ds->ds_snapname);
688 				mutex_exit(&ds->ds_lock);
689 			} else {
690 				result += strlen(ds->ds_snapname);
691 			}
692 		}
693 	}
694 
695 	return (result);
696 }
697 
698 void
699 dsl_dataset_drop_ref(dsl_dataset_t *ds, void *tag)
700 {
701 	dmu_buf_rele(ds->ds_dbuf, tag);
702 }
703 
704 void
705 dsl_dataset_rele(dsl_dataset_t *ds, void *tag)
706 {
707 	if (!dsl_pool_sync_context(ds->ds_dir->dd_pool)) {
708 		rw_exit(&ds->ds_rwlock);
709 	}
710 	dsl_dataset_drop_ref(ds, tag);
711 }
712 
713 void
714 dsl_dataset_disown(dsl_dataset_t *ds, void *tag)
715 {
716 	ASSERT((ds->ds_owner == tag && ds->ds_dbuf) ||
717 	    (DSL_DATASET_IS_DESTROYED(ds) && ds->ds_dbuf == NULL));
718 
719 	mutex_enter(&ds->ds_lock);
720 	ds->ds_owner = NULL;
721 	if (RW_WRITE_HELD(&ds->ds_rwlock)) {
722 		rw_exit(&ds->ds_rwlock);
723 		cv_broadcast(&ds->ds_exclusive_cv);
724 	}
725 	mutex_exit(&ds->ds_lock);
726 	if (ds->ds_dbuf)
727 		dsl_dataset_drop_ref(ds, tag);
728 	else
729 		dsl_dataset_evict(ds->ds_dbuf, ds);
730 }
731 
732 boolean_t
733 dsl_dataset_tryown(dsl_dataset_t *ds, boolean_t inconsistentok, void *tag)
734 {
735 	boolean_t gotit = FALSE;
736 
737 	mutex_enter(&ds->ds_lock);
738 	if (ds->ds_owner == NULL &&
739 	    (!DS_IS_INCONSISTENT(ds) || inconsistentok)) {
740 		ds->ds_owner = tag;
741 		if (!dsl_pool_sync_context(ds->ds_dir->dd_pool))
742 			rw_exit(&ds->ds_rwlock);
743 		gotit = TRUE;
744 	}
745 	mutex_exit(&ds->ds_lock);
746 	return (gotit);
747 }
748 
749 void
750 dsl_dataset_make_exclusive(dsl_dataset_t *ds, void *owner)
751 {
752 	ASSERT3P(owner, ==, ds->ds_owner);
753 	if (!RW_WRITE_HELD(&ds->ds_rwlock))
754 		rw_enter(&ds->ds_rwlock, RW_WRITER);
755 }
756 
757 uint64_t
758 dsl_dataset_create_sync_dd(dsl_dir_t *dd, dsl_dataset_t *origin,
759     uint64_t flags, dmu_tx_t *tx)
760 {
761 	dsl_pool_t *dp = dd->dd_pool;
762 	dmu_buf_t *dbuf;
763 	dsl_dataset_phys_t *dsphys;
764 	uint64_t dsobj;
765 	objset_t *mos = dp->dp_meta_objset;
766 
767 	if (origin == NULL)
768 		origin = dp->dp_origin_snap;
769 
770 	ASSERT(origin == NULL || origin->ds_dir->dd_pool == dp);
771 	ASSERT(origin == NULL || origin->ds_phys->ds_num_children > 0);
772 	ASSERT(dmu_tx_is_syncing(tx));
773 	ASSERT(dd->dd_phys->dd_head_dataset_obj == 0);
774 
775 	dsobj = dmu_object_alloc(mos, DMU_OT_DSL_DATASET, 0,
776 	    DMU_OT_DSL_DATASET, sizeof (dsl_dataset_phys_t), tx);
777 	VERIFY(0 == dmu_bonus_hold(mos, dsobj, FTAG, &dbuf));
778 	dmu_buf_will_dirty(dbuf, tx);
779 	dsphys = dbuf->db_data;
780 	bzero(dsphys, sizeof (dsl_dataset_phys_t));
781 	dsphys->ds_dir_obj = dd->dd_object;
782 	dsphys->ds_flags = flags;
783 	dsphys->ds_fsid_guid = unique_create();
784 	(void) random_get_pseudo_bytes((void*)&dsphys->ds_guid,
785 	    sizeof (dsphys->ds_guid));
786 	dsphys->ds_snapnames_zapobj =
787 	    zap_create_norm(mos, U8_TEXTPREP_TOUPPER, DMU_OT_DSL_DS_SNAP_MAP,
788 	    DMU_OT_NONE, 0, tx);
789 	dsphys->ds_creation_time = gethrestime_sec();
790 	dsphys->ds_creation_txg = tx->tx_txg == TXG_INITIAL ? 1 : tx->tx_txg;
791 	dsphys->ds_deadlist_obj =
792 	    bplist_create(mos, DSL_DEADLIST_BLOCKSIZE, tx);
793 
794 	if (origin) {
795 		dsphys->ds_prev_snap_obj = origin->ds_object;
796 		dsphys->ds_prev_snap_txg =
797 		    origin->ds_phys->ds_creation_txg;
798 		dsphys->ds_used_bytes =
799 		    origin->ds_phys->ds_used_bytes;
800 		dsphys->ds_compressed_bytes =
801 		    origin->ds_phys->ds_compressed_bytes;
802 		dsphys->ds_uncompressed_bytes =
803 		    origin->ds_phys->ds_uncompressed_bytes;
804 		dsphys->ds_bp = origin->ds_phys->ds_bp;
805 		dsphys->ds_flags |= origin->ds_phys->ds_flags;
806 
807 		dmu_buf_will_dirty(origin->ds_dbuf, tx);
808 		origin->ds_phys->ds_num_children++;
809 
810 		if (spa_version(dp->dp_spa) >= SPA_VERSION_NEXT_CLONES) {
811 			if (origin->ds_phys->ds_next_clones_obj == 0) {
812 				origin->ds_phys->ds_next_clones_obj =
813 				    zap_create(mos,
814 				    DMU_OT_NEXT_CLONES, DMU_OT_NONE, 0, tx);
815 			}
816 			VERIFY(0 == zap_add_int(mos,
817 			    origin->ds_phys->ds_next_clones_obj,
818 			    dsobj, tx));
819 		}
820 
821 		dmu_buf_will_dirty(dd->dd_dbuf, tx);
822 		dd->dd_phys->dd_origin_obj = origin->ds_object;
823 	}
824 
825 	if (spa_version(dp->dp_spa) >= SPA_VERSION_UNIQUE_ACCURATE)
826 		dsphys->ds_flags |= DS_FLAG_UNIQUE_ACCURATE;
827 
828 	dmu_buf_rele(dbuf, FTAG);
829 
830 	dmu_buf_will_dirty(dd->dd_dbuf, tx);
831 	dd->dd_phys->dd_head_dataset_obj = dsobj;
832 
833 	return (dsobj);
834 }
835 
836 uint64_t
837 dsl_dataset_create_sync(dsl_dir_t *pdd, const char *lastname,
838     dsl_dataset_t *origin, uint64_t flags, cred_t *cr, dmu_tx_t *tx)
839 {
840 	dsl_pool_t *dp = pdd->dd_pool;
841 	uint64_t dsobj, ddobj;
842 	dsl_dir_t *dd;
843 
844 	ASSERT(lastname[0] != '@');
845 
846 	ddobj = dsl_dir_create_sync(dp, pdd, lastname, tx);
847 	VERIFY(0 == dsl_dir_open_obj(dp, ddobj, lastname, FTAG, &dd));
848 
849 	dsobj = dsl_dataset_create_sync_dd(dd, origin, flags, tx);
850 
851 	dsl_deleg_set_create_perms(dd, tx, cr);
852 
853 	dsl_dir_close(dd, FTAG);
854 
855 	return (dsobj);
856 }
857 
858 struct destroyarg {
859 	dsl_sync_task_group_t *dstg;
860 	char *snapname;
861 	char *failed;
862 	boolean_t defer;
863 };
864 
865 static int
866 dsl_snapshot_destroy_one(const char *name, void *arg)
867 {
868 	struct destroyarg *da = arg;
869 	dsl_dataset_t *ds;
870 	int err;
871 	char *dsname;
872 
873 	dsname = kmem_asprintf("%s@%s", name, da->snapname);
874 	err = dsl_dataset_own(dsname, B_TRUE, da->dstg, &ds);
875 	strfree(dsname);
876 	if (err == 0) {
877 		struct dsl_ds_destroyarg *dsda;
878 
879 		dsl_dataset_make_exclusive(ds, da->dstg);
880 		if (ds->ds_objset != NULL) {
881 			dmu_objset_evict(ds->ds_objset);
882 			ds->ds_objset = NULL;
883 		}
884 		dsda = kmem_zalloc(sizeof (struct dsl_ds_destroyarg), KM_SLEEP);
885 		dsda->ds = ds;
886 		dsda->defer = da->defer;
887 		dsl_sync_task_create(da->dstg, dsl_dataset_destroy_check,
888 		    dsl_dataset_destroy_sync, dsda, da->dstg, 0);
889 	} else if (err == ENOENT) {
890 		err = 0;
891 	} else {
892 		(void) strcpy(da->failed, name);
893 	}
894 	return (err);
895 }
896 
897 /*
898  * Destroy 'snapname' in all descendants of 'fsname'.
899  */
900 #pragma weak dmu_snapshots_destroy = dsl_snapshots_destroy
901 int
902 dsl_snapshots_destroy(char *fsname, char *snapname, boolean_t defer)
903 {
904 	int err;
905 	struct destroyarg da;
906 	dsl_sync_task_t *dst;
907 	spa_t *spa;
908 
909 	err = spa_open(fsname, &spa, FTAG);
910 	if (err)
911 		return (err);
912 	da.dstg = dsl_sync_task_group_create(spa_get_dsl(spa));
913 	da.snapname = snapname;
914 	da.failed = fsname;
915 	da.defer = defer;
916 
917 	err = dmu_objset_find(fsname,
918 	    dsl_snapshot_destroy_one, &da, DS_FIND_CHILDREN);
919 
920 	if (err == 0)
921 		err = dsl_sync_task_group_wait(da.dstg);
922 
923 	for (dst = list_head(&da.dstg->dstg_tasks); dst;
924 	    dst = list_next(&da.dstg->dstg_tasks, dst)) {
925 		struct dsl_ds_destroyarg *dsda = dst->dst_arg1;
926 		dsl_dataset_t *ds = dsda->ds;
927 
928 		/*
929 		 * Return the file system name that triggered the error
930 		 */
931 		if (dst->dst_err) {
932 			dsl_dataset_name(ds, fsname);
933 			*strchr(fsname, '@') = '\0';
934 		}
935 		ASSERT3P(dsda->rm_origin, ==, NULL);
936 		dsl_dataset_disown(ds, da.dstg);
937 		kmem_free(dsda, sizeof (struct dsl_ds_destroyarg));
938 	}
939 
940 	dsl_sync_task_group_destroy(da.dstg);
941 	spa_close(spa, FTAG);
942 	return (err);
943 }
944 
945 static boolean_t
946 dsl_dataset_might_destroy_origin(dsl_dataset_t *ds)
947 {
948 	boolean_t might_destroy = B_FALSE;
949 
950 	mutex_enter(&ds->ds_lock);
951 	if (ds->ds_phys->ds_num_children == 2 && ds->ds_userrefs == 0 &&
952 	    DS_IS_DEFER_DESTROY(ds))
953 		might_destroy = B_TRUE;
954 	mutex_exit(&ds->ds_lock);
955 
956 	return (might_destroy);
957 }
958 
959 /*
960  * If we're removing a clone, and these three conditions are true:
961  *	1) the clone's origin has no other children
962  *	2) the clone's origin has no user references
963  *	3) the clone's origin has been marked for deferred destruction
964  * Then, prepare to remove the origin as part of this sync task group.
965  */
966 static int
967 dsl_dataset_origin_rm_prep(struct dsl_ds_destroyarg *dsda, void *tag)
968 {
969 	dsl_dataset_t *ds = dsda->ds;
970 	dsl_dataset_t *origin = ds->ds_prev;
971 
972 	if (dsl_dataset_might_destroy_origin(origin)) {
973 		char *name;
974 		int namelen;
975 		int error;
976 
977 		namelen = dsl_dataset_namelen(origin) + 1;
978 		name = kmem_alloc(namelen, KM_SLEEP);
979 		dsl_dataset_name(origin, name);
980 #ifdef _KERNEL
981 		error = zfs_unmount_snap(name, NULL);
982 		if (error) {
983 			kmem_free(name, namelen);
984 			return (error);
985 		}
986 #endif
987 		error = dsl_dataset_own(name, B_TRUE, tag, &origin);
988 		kmem_free(name, namelen);
989 		if (error)
990 			return (error);
991 		dsda->rm_origin = origin;
992 		dsl_dataset_make_exclusive(origin, tag);
993 
994 		if (origin->ds_objset != NULL) {
995 			dmu_objset_evict(origin->ds_objset);
996 			origin->ds_objset = NULL;
997 		}
998 	}
999 
1000 	return (0);
1001 }
1002 
1003 /*
1004  * ds must be opened as OWNER.  On return (whether successful or not),
1005  * ds will be closed and caller can no longer dereference it.
1006  */
1007 int
1008 dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
1009 {
1010 	int err;
1011 	dsl_sync_task_group_t *dstg;
1012 	objset_t *os;
1013 	dsl_dir_t *dd;
1014 	uint64_t obj;
1015 	struct dsl_ds_destroyarg dsda = { 0 };
1016 	dsl_dataset_t dummy_ds = { 0 };
1017 
1018 	dsda.ds = ds;
1019 
1020 	if (dsl_dataset_is_snapshot(ds)) {
1021 		/* Destroying a snapshot is simpler */
1022 		dsl_dataset_make_exclusive(ds, tag);
1023 
1024 		if (ds->ds_objset != NULL) {
1025 			dmu_objset_evict(ds->ds_objset);
1026 			ds->ds_objset = NULL;
1027 		}
1028 		dsda.defer = defer;
1029 		err = dsl_sync_task_do(ds->ds_dir->dd_pool,
1030 		    dsl_dataset_destroy_check, dsl_dataset_destroy_sync,
1031 		    &dsda, tag, 0);
1032 		ASSERT3P(dsda.rm_origin, ==, NULL);
1033 		goto out;
1034 	} else if (defer) {
1035 		err = EINVAL;
1036 		goto out;
1037 	}
1038 
1039 	dd = ds->ds_dir;
1040 	dummy_ds.ds_dir = dd;
1041 	dummy_ds.ds_object = ds->ds_object;
1042 
1043 	/*
1044 	 * Check for errors and mark this ds as inconsistent, in
1045 	 * case we crash while freeing the objects.
1046 	 */
1047 	err = dsl_sync_task_do(dd->dd_pool, dsl_dataset_destroy_begin_check,
1048 	    dsl_dataset_destroy_begin_sync, ds, NULL, 0);
1049 	if (err)
1050 		goto out;
1051 
1052 	err = dmu_objset_from_ds(ds, &os);
1053 	if (err)
1054 		goto out;
1055 
1056 	/*
1057 	 * remove the objects in open context, so that we won't
1058 	 * have too much to do in syncing context.
1059 	 */
1060 	for (obj = 0; err == 0; err = dmu_object_next(os, &obj, FALSE,
1061 	    ds->ds_phys->ds_prev_snap_txg)) {
1062 		/*
1063 		 * Ignore errors, if there is not enough disk space
1064 		 * we will deal with it in dsl_dataset_destroy_sync().
1065 		 */
1066 		(void) dmu_free_object(os, obj);
1067 	}
1068 
1069 	/*
1070 	 * We need to sync out all in-flight IO before we try to evict
1071 	 * (the dataset evict func is trying to clear the cached entries
1072 	 * for this dataset in the ARC).
1073 	 */
1074 	txg_wait_synced(dd->dd_pool, 0);
1075 
1076 	/*
1077 	 * If we managed to free all the objects in open
1078 	 * context, the user space accounting should be zero.
1079 	 */
1080 	if (ds->ds_phys->ds_bp.blk_fill == 0 &&
1081 	    dmu_objset_userused_enabled(os)) {
1082 		uint64_t count;
1083 
1084 		ASSERT(zap_count(os, DMU_USERUSED_OBJECT, &count) != 0 ||
1085 		    count == 0);
1086 		ASSERT(zap_count(os, DMU_GROUPUSED_OBJECT, &count) != 0 ||
1087 		    count == 0);
1088 	}
1089 
1090 	if (err != ESRCH)
1091 		goto out;
1092 
1093 	rw_enter(&dd->dd_pool->dp_config_rwlock, RW_READER);
1094 	err = dsl_dir_open_obj(dd->dd_pool, dd->dd_object, NULL, FTAG, &dd);
1095 	rw_exit(&dd->dd_pool->dp_config_rwlock);
1096 
1097 	if (err)
1098 		goto out;
1099 
1100 	if (ds->ds_objset) {
1101 		/*
1102 		 * We need to sync out all in-flight IO before we try
1103 		 * to evict (the dataset evict func is trying to clear
1104 		 * the cached entries for this dataset in the ARC).
1105 		 */
1106 		txg_wait_synced(dd->dd_pool, 0);
1107 	}
1108 
1109 	/*
1110 	 * Blow away the dsl_dir + head dataset.
1111 	 */
1112 	dsl_dataset_make_exclusive(ds, tag);
1113 	if (ds->ds_objset) {
1114 		dmu_objset_evict(ds->ds_objset);
1115 		ds->ds_objset = NULL;
1116 	}
1117 
1118 	/*
1119 	 * If we're removing a clone, we might also need to remove its
1120 	 * origin.
1121 	 */
1122 	do {
1123 		dsda.need_prep = B_FALSE;
1124 		if (dsl_dir_is_clone(dd)) {
1125 			err = dsl_dataset_origin_rm_prep(&dsda, tag);
1126 			if (err) {
1127 				dsl_dir_close(dd, FTAG);
1128 				goto out;
1129 			}
1130 		}
1131 
1132 		dstg = dsl_sync_task_group_create(ds->ds_dir->dd_pool);
1133 		dsl_sync_task_create(dstg, dsl_dataset_destroy_check,
1134 		    dsl_dataset_destroy_sync, &dsda, tag, 0);
1135 		dsl_sync_task_create(dstg, dsl_dir_destroy_check,
1136 		    dsl_dir_destroy_sync, &dummy_ds, FTAG, 0);
1137 		err = dsl_sync_task_group_wait(dstg);
1138 		dsl_sync_task_group_destroy(dstg);
1139 
1140 		/*
1141 		 * We could be racing against 'zfs release' or 'zfs destroy -d'
1142 		 * on the origin snap, in which case we can get EBUSY if we
1143 		 * needed to destroy the origin snap but were not ready to
1144 		 * do so.
1145 		 */
1146 		if (dsda.need_prep) {
1147 			ASSERT(err == EBUSY);
1148 			ASSERT(dsl_dir_is_clone(dd));
1149 			ASSERT(dsda.rm_origin == NULL);
1150 		}
1151 	} while (dsda.need_prep);
1152 
1153 	if (dsda.rm_origin != NULL)
1154 		dsl_dataset_disown(dsda.rm_origin, tag);
1155 
1156 	/* if it is successful, dsl_dir_destroy_sync will close the dd */
1157 	if (err)
1158 		dsl_dir_close(dd, FTAG);
1159 out:
1160 	dsl_dataset_disown(ds, tag);
1161 	return (err);
1162 }
1163 
1164 blkptr_t *
1165 dsl_dataset_get_blkptr(dsl_dataset_t *ds)
1166 {
1167 	return (&ds->ds_phys->ds_bp);
1168 }
1169 
1170 void
1171 dsl_dataset_set_blkptr(dsl_dataset_t *ds, blkptr_t *bp, dmu_tx_t *tx)
1172 {
1173 	ASSERT(dmu_tx_is_syncing(tx));
1174 	/* If it's the meta-objset, set dp_meta_rootbp */
1175 	if (ds == NULL) {
1176 		tx->tx_pool->dp_meta_rootbp = *bp;
1177 	} else {
1178 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
1179 		ds->ds_phys->ds_bp = *bp;
1180 	}
1181 }
1182 
1183 spa_t *
1184 dsl_dataset_get_spa(dsl_dataset_t *ds)
1185 {
1186 	return (ds->ds_dir->dd_pool->dp_spa);
1187 }
1188 
1189 void
1190 dsl_dataset_dirty(dsl_dataset_t *ds, dmu_tx_t *tx)
1191 {
1192 	dsl_pool_t *dp;
1193 
1194 	if (ds == NULL) /* this is the meta-objset */
1195 		return;
1196 
1197 	ASSERT(ds->ds_objset != NULL);
1198 
1199 	if (ds->ds_phys->ds_next_snap_obj != 0)
1200 		panic("dirtying snapshot!");
1201 
1202 	dp = ds->ds_dir->dd_pool;
1203 
1204 	if (txg_list_add(&dp->dp_dirty_datasets, ds, tx->tx_txg) == 0) {
1205 		/* up the hold count until we can be written out */
1206 		dmu_buf_add_ref(ds->ds_dbuf, ds);
1207 	}
1208 }
1209 
1210 /*
1211  * The unique space in the head dataset can be calculated by subtracting
1212  * the space used in the most recent snapshot, that is still being used
1213  * in this file system, from the space currently in use.  To figure out
1214  * the space in the most recent snapshot still in use, we need to take
1215  * the total space used in the snapshot and subtract out the space that
1216  * has been freed up since the snapshot was taken.
1217  */
1218 static void
1219 dsl_dataset_recalc_head_uniq(dsl_dataset_t *ds)
1220 {
1221 	uint64_t mrs_used;
1222 	uint64_t dlused, dlcomp, dluncomp;
1223 
1224 	ASSERT(ds->ds_object == ds->ds_dir->dd_phys->dd_head_dataset_obj);
1225 
1226 	if (ds->ds_phys->ds_prev_snap_obj != 0)
1227 		mrs_used = ds->ds_prev->ds_phys->ds_used_bytes;
1228 	else
1229 		mrs_used = 0;
1230 
1231 	VERIFY(0 == bplist_space(&ds->ds_deadlist, &dlused, &dlcomp,
1232 	    &dluncomp));
1233 
1234 	ASSERT3U(dlused, <=, mrs_used);
1235 	ds->ds_phys->ds_unique_bytes =
1236 	    ds->ds_phys->ds_used_bytes - (mrs_used - dlused);
1237 
1238 	if (!DS_UNIQUE_IS_ACCURATE(ds) &&
1239 	    spa_version(ds->ds_dir->dd_pool->dp_spa) >=
1240 	    SPA_VERSION_UNIQUE_ACCURATE)
1241 		ds->ds_phys->ds_flags |= DS_FLAG_UNIQUE_ACCURATE;
1242 }
1243 
1244 static uint64_t
1245 dsl_dataset_unique(dsl_dataset_t *ds)
1246 {
1247 	if (!DS_UNIQUE_IS_ACCURATE(ds) && !dsl_dataset_is_snapshot(ds))
1248 		dsl_dataset_recalc_head_uniq(ds);
1249 
1250 	return (ds->ds_phys->ds_unique_bytes);
1251 }
1252 
1253 struct killarg {
1254 	dsl_dataset_t *ds;
1255 	dmu_tx_t *tx;
1256 };
1257 
1258 /* ARGSUSED */
1259 static int
1260 kill_blkptr(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
1261     const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg)
1262 {
1263 	struct killarg *ka = arg;
1264 	dmu_tx_t *tx = ka->tx;
1265 
1266 	if (bp == NULL)
1267 		return (0);
1268 
1269 	if (zb->zb_level == ZB_ZIL_LEVEL) {
1270 		ASSERT(zilog != NULL);
1271 		/*
1272 		 * It's a block in the intent log.  It has no
1273 		 * accounting, so just free it.
1274 		 */
1275 		dsl_free(ka->tx->tx_pool, ka->tx->tx_txg, bp);
1276 	} else {
1277 		ASSERT(zilog == NULL);
1278 		ASSERT3U(bp->blk_birth, >, ka->ds->ds_phys->ds_prev_snap_txg);
1279 		(void) dsl_dataset_block_kill(ka->ds, bp, tx, B_FALSE);
1280 	}
1281 
1282 	return (0);
1283 }
1284 
1285 /* ARGSUSED */
1286 static int
1287 dsl_dataset_destroy_begin_check(void *arg1, void *arg2, dmu_tx_t *tx)
1288 {
1289 	dsl_dataset_t *ds = arg1;
1290 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
1291 	uint64_t count;
1292 	int err;
1293 
1294 	/*
1295 	 * Can't delete a head dataset if there are snapshots of it.
1296 	 * (Except if the only snapshots are from the branch we cloned
1297 	 * from.)
1298 	 */
1299 	if (ds->ds_prev != NULL &&
1300 	    ds->ds_prev->ds_phys->ds_next_snap_obj == ds->ds_object)
1301 		return (EBUSY);
1302 
1303 	/*
1304 	 * This is really a dsl_dir thing, but check it here so that
1305 	 * we'll be less likely to leave this dataset inconsistent &
1306 	 * nearly destroyed.
1307 	 */
1308 	err = zap_count(mos, ds->ds_dir->dd_phys->dd_child_dir_zapobj, &count);
1309 	if (err)
1310 		return (err);
1311 	if (count != 0)
1312 		return (EEXIST);
1313 
1314 	return (0);
1315 }
1316 
1317 /* ARGSUSED */
1318 static void
1319 dsl_dataset_destroy_begin_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
1320 {
1321 	dsl_dataset_t *ds = arg1;
1322 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
1323 
1324 	/* Mark it as inconsistent on-disk, in case we crash */
1325 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
1326 	ds->ds_phys->ds_flags |= DS_FLAG_INCONSISTENT;
1327 
1328 	spa_history_internal_log(LOG_DS_DESTROY_BEGIN, dp->dp_spa, tx,
1329 	    cr, "dataset = %llu", ds->ds_object);
1330 }
1331 
1332 static int
1333 dsl_dataset_origin_check(struct dsl_ds_destroyarg *dsda, void *tag,
1334     dmu_tx_t *tx)
1335 {
1336 	dsl_dataset_t *ds = dsda->ds;
1337 	dsl_dataset_t *ds_prev = ds->ds_prev;
1338 
1339 	if (dsl_dataset_might_destroy_origin(ds_prev)) {
1340 		struct dsl_ds_destroyarg ndsda = {0};
1341 
1342 		/*
1343 		 * If we're not prepared to remove the origin, don't remove
1344 		 * the clone either.
1345 		 */
1346 		if (dsda->rm_origin == NULL) {
1347 			dsda->need_prep = B_TRUE;
1348 			return (EBUSY);
1349 		}
1350 
1351 		ndsda.ds = ds_prev;
1352 		ndsda.is_origin_rm = B_TRUE;
1353 		return (dsl_dataset_destroy_check(&ndsda, tag, tx));
1354 	}
1355 
1356 	/*
1357 	 * If we're not going to remove the origin after all,
1358 	 * undo the open context setup.
1359 	 */
1360 	if (dsda->rm_origin != NULL) {
1361 		dsl_dataset_disown(dsda->rm_origin, tag);
1362 		dsda->rm_origin = NULL;
1363 	}
1364 
1365 	return (0);
1366 }
1367 
1368 /* ARGSUSED */
1369 int
1370 dsl_dataset_destroy_check(void *arg1, void *arg2, dmu_tx_t *tx)
1371 {
1372 	struct dsl_ds_destroyarg *dsda = arg1;
1373 	dsl_dataset_t *ds = dsda->ds;
1374 
1375 	/* we have an owner hold, so noone else can destroy us */
1376 	ASSERT(!DSL_DATASET_IS_DESTROYED(ds));
1377 
1378 	/*
1379 	 * Only allow deferred destroy on pools that support it.
1380 	 * NOTE: deferred destroy is only supported on snapshots.
1381 	 */
1382 	if (dsda->defer) {
1383 		if (spa_version(ds->ds_dir->dd_pool->dp_spa) <
1384 		    SPA_VERSION_USERREFS)
1385 			return (ENOTSUP);
1386 		ASSERT(dsl_dataset_is_snapshot(ds));
1387 		return (0);
1388 	}
1389 
1390 	/*
1391 	 * Can't delete a head dataset if there are snapshots of it.
1392 	 * (Except if the only snapshots are from the branch we cloned
1393 	 * from.)
1394 	 */
1395 	if (ds->ds_prev != NULL &&
1396 	    ds->ds_prev->ds_phys->ds_next_snap_obj == ds->ds_object)
1397 		return (EBUSY);
1398 
1399 	/*
1400 	 * If we made changes this txg, traverse_dsl_dataset won't find
1401 	 * them.  Try again.
1402 	 */
1403 	if (ds->ds_phys->ds_bp.blk_birth >= tx->tx_txg)
1404 		return (EAGAIN);
1405 
1406 	if (dsl_dataset_is_snapshot(ds)) {
1407 		/*
1408 		 * If this snapshot has an elevated user reference count,
1409 		 * we can't destroy it yet.
1410 		 */
1411 		if (ds->ds_userrefs > 0 && !dsda->releasing)
1412 			return (EBUSY);
1413 
1414 		mutex_enter(&ds->ds_lock);
1415 		/*
1416 		 * Can't delete a branch point. However, if we're destroying
1417 		 * a clone and removing its origin due to it having a user
1418 		 * hold count of 0 and having been marked for deferred destroy,
1419 		 * it's OK for the origin to have a single clone.
1420 		 */
1421 		if (ds->ds_phys->ds_num_children >
1422 		    (dsda->is_origin_rm ? 2 : 1)) {
1423 			mutex_exit(&ds->ds_lock);
1424 			return (EEXIST);
1425 		}
1426 		mutex_exit(&ds->ds_lock);
1427 	} else if (dsl_dir_is_clone(ds->ds_dir)) {
1428 		return (dsl_dataset_origin_check(dsda, arg2, tx));
1429 	}
1430 
1431 	/* XXX we should do some i/o error checking... */
1432 	return (0);
1433 }
1434 
1435 struct refsarg {
1436 	kmutex_t lock;
1437 	boolean_t gone;
1438 	kcondvar_t cv;
1439 };
1440 
1441 /* ARGSUSED */
1442 static void
1443 dsl_dataset_refs_gone(dmu_buf_t *db, void *argv)
1444 {
1445 	struct refsarg *arg = argv;
1446 
1447 	mutex_enter(&arg->lock);
1448 	arg->gone = TRUE;
1449 	cv_signal(&arg->cv);
1450 	mutex_exit(&arg->lock);
1451 }
1452 
1453 static void
1454 dsl_dataset_drain_refs(dsl_dataset_t *ds, void *tag)
1455 {
1456 	struct refsarg arg;
1457 
1458 	mutex_init(&arg.lock, NULL, MUTEX_DEFAULT, NULL);
1459 	cv_init(&arg.cv, NULL, CV_DEFAULT, NULL);
1460 	arg.gone = FALSE;
1461 	(void) dmu_buf_update_user(ds->ds_dbuf, ds, &arg, &ds->ds_phys,
1462 	    dsl_dataset_refs_gone);
1463 	dmu_buf_rele(ds->ds_dbuf, tag);
1464 	mutex_enter(&arg.lock);
1465 	while (!arg.gone)
1466 		cv_wait(&arg.cv, &arg.lock);
1467 	ASSERT(arg.gone);
1468 	mutex_exit(&arg.lock);
1469 	ds->ds_dbuf = NULL;
1470 	ds->ds_phys = NULL;
1471 	mutex_destroy(&arg.lock);
1472 	cv_destroy(&arg.cv);
1473 }
1474 
1475 static void
1476 remove_from_next_clones(dsl_dataset_t *ds, uint64_t obj, dmu_tx_t *tx)
1477 {
1478 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
1479 	uint64_t count;
1480 	int err;
1481 
1482 	ASSERT(ds->ds_phys->ds_num_children >= 2);
1483 	err = zap_remove_int(mos, ds->ds_phys->ds_next_clones_obj, obj, tx);
1484 	/*
1485 	 * The err should not be ENOENT, but a bug in a previous version
1486 	 * of the code could cause upgrade_clones_cb() to not set
1487 	 * ds_next_snap_obj when it should, leading to a missing entry.
1488 	 * If we knew that the pool was created after
1489 	 * SPA_VERSION_NEXT_CLONES, we could assert that it isn't
1490 	 * ENOENT.  However, at least we can check that we don't have
1491 	 * too many entries in the next_clones_obj even after failing to
1492 	 * remove this one.
1493 	 */
1494 	if (err != ENOENT) {
1495 		VERIFY3U(err, ==, 0);
1496 	}
1497 	ASSERT3U(0, ==, zap_count(mos, ds->ds_phys->ds_next_clones_obj,
1498 	    &count));
1499 	ASSERT3U(count, <=, ds->ds_phys->ds_num_children - 2);
1500 }
1501 
1502 void
1503 dsl_dataset_destroy_sync(void *arg1, void *tag, cred_t *cr, dmu_tx_t *tx)
1504 {
1505 	struct dsl_ds_destroyarg *dsda = arg1;
1506 	dsl_dataset_t *ds = dsda->ds;
1507 	int err;
1508 	int after_branch_point = FALSE;
1509 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
1510 	objset_t *mos = dp->dp_meta_objset;
1511 	dsl_dataset_t *ds_prev = NULL;
1512 	uint64_t obj;
1513 
1514 	ASSERT(ds->ds_owner);
1515 	ASSERT(dsda->defer || ds->ds_phys->ds_num_children <= 1);
1516 	ASSERT(ds->ds_prev == NULL ||
1517 	    ds->ds_prev->ds_phys->ds_next_snap_obj != ds->ds_object);
1518 	ASSERT3U(ds->ds_phys->ds_bp.blk_birth, <=, tx->tx_txg);
1519 
1520 	if (dsda->defer) {
1521 		ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS);
1522 		if (ds->ds_userrefs > 0 || ds->ds_phys->ds_num_children > 1) {
1523 			dmu_buf_will_dirty(ds->ds_dbuf, tx);
1524 			ds->ds_phys->ds_flags |= DS_FLAG_DEFER_DESTROY;
1525 			return;
1526 		}
1527 	}
1528 
1529 	/* signal any waiters that this dataset is going away */
1530 	mutex_enter(&ds->ds_lock);
1531 	ds->ds_owner = dsl_reaper;
1532 	cv_broadcast(&ds->ds_exclusive_cv);
1533 	mutex_exit(&ds->ds_lock);
1534 
1535 	/* Remove our reservation */
1536 	if (ds->ds_reserved != 0) {
1537 		dsl_prop_setarg_t psa;
1538 		uint64_t value = 0;
1539 
1540 		dsl_prop_setarg_init_uint64(&psa, "refreservation",
1541 		    (ZPROP_SRC_NONE | ZPROP_SRC_LOCAL | ZPROP_SRC_RECEIVED),
1542 		    &value);
1543 		psa.psa_effective_value = 0;	/* predict default value */
1544 
1545 		dsl_dataset_set_reservation_sync(ds, &psa, cr, tx);
1546 		ASSERT3U(ds->ds_reserved, ==, 0);
1547 	}
1548 
1549 	ASSERT(RW_WRITE_HELD(&dp->dp_config_rwlock));
1550 
1551 	dsl_pool_ds_destroyed(ds, tx);
1552 
1553 	obj = ds->ds_object;
1554 
1555 	if (ds->ds_phys->ds_prev_snap_obj != 0) {
1556 		if (ds->ds_prev) {
1557 			ds_prev = ds->ds_prev;
1558 		} else {
1559 			VERIFY(0 == dsl_dataset_hold_obj(dp,
1560 			    ds->ds_phys->ds_prev_snap_obj, FTAG, &ds_prev));
1561 		}
1562 		after_branch_point =
1563 		    (ds_prev->ds_phys->ds_next_snap_obj != obj);
1564 
1565 		dmu_buf_will_dirty(ds_prev->ds_dbuf, tx);
1566 		if (after_branch_point &&
1567 		    ds_prev->ds_phys->ds_next_clones_obj != 0) {
1568 			remove_from_next_clones(ds_prev, obj, tx);
1569 			if (ds->ds_phys->ds_next_snap_obj != 0) {
1570 				VERIFY(0 == zap_add_int(mos,
1571 				    ds_prev->ds_phys->ds_next_clones_obj,
1572 				    ds->ds_phys->ds_next_snap_obj, tx));
1573 			}
1574 		}
1575 		if (after_branch_point &&
1576 		    ds->ds_phys->ds_next_snap_obj == 0) {
1577 			/* This clone is toast. */
1578 			ASSERT(ds_prev->ds_phys->ds_num_children > 1);
1579 			ds_prev->ds_phys->ds_num_children--;
1580 
1581 			/*
1582 			 * If the clone's origin has no other clones, no
1583 			 * user holds, and has been marked for deferred
1584 			 * deletion, then we should have done the necessary
1585 			 * destroy setup for it.
1586 			 */
1587 			if (ds_prev->ds_phys->ds_num_children == 1 &&
1588 			    ds_prev->ds_userrefs == 0 &&
1589 			    DS_IS_DEFER_DESTROY(ds_prev)) {
1590 				ASSERT3P(dsda->rm_origin, !=, NULL);
1591 			} else {
1592 				ASSERT3P(dsda->rm_origin, ==, NULL);
1593 			}
1594 		} else if (!after_branch_point) {
1595 			ds_prev->ds_phys->ds_next_snap_obj =
1596 			    ds->ds_phys->ds_next_snap_obj;
1597 		}
1598 	}
1599 
1600 	if (ds->ds_phys->ds_next_snap_obj != 0) {
1601 		blkptr_t bp;
1602 		dsl_dataset_t *ds_next;
1603 		uint64_t itor = 0;
1604 		uint64_t old_unique;
1605 		int64_t used = 0, compressed = 0, uncompressed = 0;
1606 
1607 		VERIFY(0 == dsl_dataset_hold_obj(dp,
1608 		    ds->ds_phys->ds_next_snap_obj, FTAG, &ds_next));
1609 		ASSERT3U(ds_next->ds_phys->ds_prev_snap_obj, ==, obj);
1610 
1611 		old_unique = dsl_dataset_unique(ds_next);
1612 
1613 		dmu_buf_will_dirty(ds_next->ds_dbuf, tx);
1614 		ds_next->ds_phys->ds_prev_snap_obj =
1615 		    ds->ds_phys->ds_prev_snap_obj;
1616 		ds_next->ds_phys->ds_prev_snap_txg =
1617 		    ds->ds_phys->ds_prev_snap_txg;
1618 		ASSERT3U(ds->ds_phys->ds_prev_snap_txg, ==,
1619 		    ds_prev ? ds_prev->ds_phys->ds_creation_txg : 0);
1620 
1621 		/*
1622 		 * Transfer to our deadlist (which will become next's
1623 		 * new deadlist) any entries from next's current
1624 		 * deadlist which were born before prev, and free the
1625 		 * other entries.
1626 		 *
1627 		 * XXX we're doing this long task with the config lock held
1628 		 */
1629 		while (bplist_iterate(&ds_next->ds_deadlist, &itor, &bp) == 0) {
1630 			if (bp.blk_birth <= ds->ds_phys->ds_prev_snap_txg) {
1631 				VERIFY(0 == bplist_enqueue(&ds->ds_deadlist,
1632 				    &bp, tx));
1633 				if (ds_prev && !after_branch_point &&
1634 				    bp.blk_birth >
1635 				    ds_prev->ds_phys->ds_prev_snap_txg) {
1636 					ds_prev->ds_phys->ds_unique_bytes +=
1637 					    bp_get_dsize_sync(dp->dp_spa, &bp);
1638 				}
1639 			} else {
1640 				used += bp_get_dsize_sync(dp->dp_spa, &bp);
1641 				compressed += BP_GET_PSIZE(&bp);
1642 				uncompressed += BP_GET_UCSIZE(&bp);
1643 				dsl_free(dp, tx->tx_txg, &bp);
1644 			}
1645 		}
1646 
1647 		ASSERT3U(used, ==, ds->ds_phys->ds_unique_bytes);
1648 
1649 		/* change snapused */
1650 		dsl_dir_diduse_space(ds->ds_dir, DD_USED_SNAP,
1651 		    -used, -compressed, -uncompressed, tx);
1652 
1653 		/* free next's deadlist */
1654 		bplist_close(&ds_next->ds_deadlist);
1655 		bplist_destroy(mos, ds_next->ds_phys->ds_deadlist_obj, tx);
1656 
1657 		/* set next's deadlist to our deadlist */
1658 		bplist_close(&ds->ds_deadlist);
1659 		ds_next->ds_phys->ds_deadlist_obj =
1660 		    ds->ds_phys->ds_deadlist_obj;
1661 		VERIFY(0 == bplist_open(&ds_next->ds_deadlist, mos,
1662 		    ds_next->ds_phys->ds_deadlist_obj));
1663 		ds->ds_phys->ds_deadlist_obj = 0;
1664 
1665 		if (ds_next->ds_phys->ds_next_snap_obj != 0) {
1666 			/*
1667 			 * Update next's unique to include blocks which
1668 			 * were previously shared by only this snapshot
1669 			 * and it.  Those blocks will be born after the
1670 			 * prev snap and before this snap, and will have
1671 			 * died after the next snap and before the one
1672 			 * after that (ie. be on the snap after next's
1673 			 * deadlist).
1674 			 *
1675 			 * XXX we're doing this long task with the
1676 			 * config lock held
1677 			 */
1678 			dsl_dataset_t *ds_after_next;
1679 			uint64_t space;
1680 
1681 			VERIFY(0 == dsl_dataset_hold_obj(dp,
1682 			    ds_next->ds_phys->ds_next_snap_obj,
1683 			    FTAG, &ds_after_next));
1684 
1685 			VERIFY(0 ==
1686 			    bplist_space_birthrange(&ds_after_next->ds_deadlist,
1687 			    ds->ds_phys->ds_prev_snap_txg,
1688 			    ds->ds_phys->ds_creation_txg, &space));
1689 			ds_next->ds_phys->ds_unique_bytes += space;
1690 
1691 			dsl_dataset_rele(ds_after_next, FTAG);
1692 			ASSERT3P(ds_next->ds_prev, ==, NULL);
1693 		} else {
1694 			ASSERT3P(ds_next->ds_prev, ==, ds);
1695 			dsl_dataset_drop_ref(ds_next->ds_prev, ds_next);
1696 			ds_next->ds_prev = NULL;
1697 			if (ds_prev) {
1698 				VERIFY(0 == dsl_dataset_get_ref(dp,
1699 				    ds->ds_phys->ds_prev_snap_obj,
1700 				    ds_next, &ds_next->ds_prev));
1701 			}
1702 
1703 			dsl_dataset_recalc_head_uniq(ds_next);
1704 
1705 			/*
1706 			 * Reduce the amount of our unconsmed refreservation
1707 			 * being charged to our parent by the amount of
1708 			 * new unique data we have gained.
1709 			 */
1710 			if (old_unique < ds_next->ds_reserved) {
1711 				int64_t mrsdelta;
1712 				uint64_t new_unique =
1713 				    ds_next->ds_phys->ds_unique_bytes;
1714 
1715 				ASSERT(old_unique <= new_unique);
1716 				mrsdelta = MIN(new_unique - old_unique,
1717 				    ds_next->ds_reserved - old_unique);
1718 				dsl_dir_diduse_space(ds->ds_dir,
1719 				    DD_USED_REFRSRV, -mrsdelta, 0, 0, tx);
1720 			}
1721 		}
1722 		dsl_dataset_rele(ds_next, FTAG);
1723 	} else {
1724 		/*
1725 		 * There's no next snapshot, so this is a head dataset.
1726 		 * Destroy the deadlist.  Unless it's a clone, the
1727 		 * deadlist should be empty.  (If it's a clone, it's
1728 		 * safe to ignore the deadlist contents.)
1729 		 */
1730 		struct killarg ka;
1731 
1732 		ASSERT(after_branch_point || bplist_empty(&ds->ds_deadlist));
1733 		bplist_close(&ds->ds_deadlist);
1734 		bplist_destroy(mos, ds->ds_phys->ds_deadlist_obj, tx);
1735 		ds->ds_phys->ds_deadlist_obj = 0;
1736 
1737 		/*
1738 		 * Free everything that we point to (that's born after
1739 		 * the previous snapshot, if we are a clone)
1740 		 *
1741 		 * NB: this should be very quick, because we already
1742 		 * freed all the objects in open context.
1743 		 */
1744 		ka.ds = ds;
1745 		ka.tx = tx;
1746 		err = traverse_dataset(ds, ds->ds_phys->ds_prev_snap_txg,
1747 		    TRAVERSE_POST, kill_blkptr, &ka);
1748 		ASSERT3U(err, ==, 0);
1749 		ASSERT(!DS_UNIQUE_IS_ACCURATE(ds) ||
1750 		    ds->ds_phys->ds_unique_bytes == 0);
1751 
1752 		if (ds->ds_prev != NULL) {
1753 			dsl_dataset_rele(ds->ds_prev, ds);
1754 			ds->ds_prev = ds_prev = NULL;
1755 		}
1756 	}
1757 
1758 	if (ds->ds_dir->dd_phys->dd_head_dataset_obj == ds->ds_object) {
1759 		/* Erase the link in the dir */
1760 		dmu_buf_will_dirty(ds->ds_dir->dd_dbuf, tx);
1761 		ds->ds_dir->dd_phys->dd_head_dataset_obj = 0;
1762 		ASSERT(ds->ds_phys->ds_snapnames_zapobj != 0);
1763 		err = zap_destroy(mos, ds->ds_phys->ds_snapnames_zapobj, tx);
1764 		ASSERT(err == 0);
1765 	} else {
1766 		/* remove from snapshot namespace */
1767 		dsl_dataset_t *ds_head;
1768 		ASSERT(ds->ds_phys->ds_snapnames_zapobj == 0);
1769 		VERIFY(0 == dsl_dataset_hold_obj(dp,
1770 		    ds->ds_dir->dd_phys->dd_head_dataset_obj, FTAG, &ds_head));
1771 		VERIFY(0 == dsl_dataset_get_snapname(ds));
1772 #ifdef ZFS_DEBUG
1773 		{
1774 			uint64_t val;
1775 
1776 			err = dsl_dataset_snap_lookup(ds_head,
1777 			    ds->ds_snapname, &val);
1778 			ASSERT3U(err, ==, 0);
1779 			ASSERT3U(val, ==, obj);
1780 		}
1781 #endif
1782 		err = dsl_dataset_snap_remove(ds_head, ds->ds_snapname, tx);
1783 		ASSERT(err == 0);
1784 		dsl_dataset_rele(ds_head, FTAG);
1785 	}
1786 
1787 	if (ds_prev && ds->ds_prev != ds_prev)
1788 		dsl_dataset_rele(ds_prev, FTAG);
1789 
1790 	spa_prop_clear_bootfs(dp->dp_spa, ds->ds_object, tx);
1791 	spa_history_internal_log(LOG_DS_DESTROY, dp->dp_spa, tx,
1792 	    cr, "dataset = %llu", ds->ds_object);
1793 
1794 	if (ds->ds_phys->ds_next_clones_obj != 0) {
1795 		uint64_t count;
1796 		ASSERT(0 == zap_count(mos,
1797 		    ds->ds_phys->ds_next_clones_obj, &count) && count == 0);
1798 		VERIFY(0 == dmu_object_free(mos,
1799 		    ds->ds_phys->ds_next_clones_obj, tx));
1800 	}
1801 	if (ds->ds_phys->ds_props_obj != 0)
1802 		VERIFY(0 == zap_destroy(mos, ds->ds_phys->ds_props_obj, tx));
1803 	if (ds->ds_phys->ds_userrefs_obj != 0)
1804 		VERIFY(0 == zap_destroy(mos, ds->ds_phys->ds_userrefs_obj, tx));
1805 	dsl_dir_close(ds->ds_dir, ds);
1806 	ds->ds_dir = NULL;
1807 	dsl_dataset_drain_refs(ds, tag);
1808 	VERIFY(0 == dmu_object_free(mos, obj, tx));
1809 
1810 	if (dsda->rm_origin) {
1811 		/*
1812 		 * Remove the origin of the clone we just destroyed.
1813 		 */
1814 		struct dsl_ds_destroyarg ndsda = {0};
1815 
1816 		ndsda.ds = dsda->rm_origin;
1817 		dsl_dataset_destroy_sync(&ndsda, tag, cr, tx);
1818 	}
1819 }
1820 
1821 static int
1822 dsl_dataset_snapshot_reserve_space(dsl_dataset_t *ds, dmu_tx_t *tx)
1823 {
1824 	uint64_t asize;
1825 
1826 	if (!dmu_tx_is_syncing(tx))
1827 		return (0);
1828 
1829 	/*
1830 	 * If there's an fs-only reservation, any blocks that might become
1831 	 * owned by the snapshot dataset must be accommodated by space
1832 	 * outside of the reservation.
1833 	 */
1834 	asize = MIN(dsl_dataset_unique(ds), ds->ds_reserved);
1835 	if (asize > dsl_dir_space_available(ds->ds_dir, NULL, 0, FALSE))
1836 		return (ENOSPC);
1837 
1838 	/*
1839 	 * Propogate any reserved space for this snapshot to other
1840 	 * snapshot checks in this sync group.
1841 	 */
1842 	if (asize > 0)
1843 		dsl_dir_willuse_space(ds->ds_dir, asize, tx);
1844 
1845 	return (0);
1846 }
1847 
1848 /* ARGSUSED */
1849 int
1850 dsl_dataset_snapshot_check(void *arg1, void *arg2, dmu_tx_t *tx)
1851 {
1852 	dsl_dataset_t *ds = arg1;
1853 	const char *snapname = arg2;
1854 	int err;
1855 	uint64_t value;
1856 
1857 	/*
1858 	 * We don't allow multiple snapshots of the same txg.  If there
1859 	 * is already one, try again.
1860 	 */
1861 	if (ds->ds_phys->ds_prev_snap_txg >= tx->tx_txg)
1862 		return (EAGAIN);
1863 
1864 	/*
1865 	 * Check for conflicting name snapshot name.
1866 	 */
1867 	err = dsl_dataset_snap_lookup(ds, snapname, &value);
1868 	if (err == 0)
1869 		return (EEXIST);
1870 	if (err != ENOENT)
1871 		return (err);
1872 
1873 	/*
1874 	 * Check that the dataset's name is not too long.  Name consists
1875 	 * of the dataset's length + 1 for the @-sign + snapshot name's length
1876 	 */
1877 	if (dsl_dataset_namelen(ds) + 1 + strlen(snapname) >= MAXNAMELEN)
1878 		return (ENAMETOOLONG);
1879 
1880 	err = dsl_dataset_snapshot_reserve_space(ds, tx);
1881 	if (err)
1882 		return (err);
1883 
1884 	ds->ds_trysnap_txg = tx->tx_txg;
1885 	return (0);
1886 }
1887 
1888 void
1889 dsl_dataset_snapshot_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
1890 {
1891 	dsl_dataset_t *ds = arg1;
1892 	const char *snapname = arg2;
1893 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
1894 	dmu_buf_t *dbuf;
1895 	dsl_dataset_phys_t *dsphys;
1896 	uint64_t dsobj, crtxg;
1897 	objset_t *mos = dp->dp_meta_objset;
1898 	int err;
1899 
1900 	ASSERT(RW_WRITE_HELD(&dp->dp_config_rwlock));
1901 
1902 	/*
1903 	 * The origin's ds_creation_txg has to be < TXG_INITIAL
1904 	 */
1905 	if (strcmp(snapname, ORIGIN_DIR_NAME) == 0)
1906 		crtxg = 1;
1907 	else
1908 		crtxg = tx->tx_txg;
1909 
1910 	dsobj = dmu_object_alloc(mos, DMU_OT_DSL_DATASET, 0,
1911 	    DMU_OT_DSL_DATASET, sizeof (dsl_dataset_phys_t), tx);
1912 	VERIFY(0 == dmu_bonus_hold(mos, dsobj, FTAG, &dbuf));
1913 	dmu_buf_will_dirty(dbuf, tx);
1914 	dsphys = dbuf->db_data;
1915 	bzero(dsphys, sizeof (dsl_dataset_phys_t));
1916 	dsphys->ds_dir_obj = ds->ds_dir->dd_object;
1917 	dsphys->ds_fsid_guid = unique_create();
1918 	(void) random_get_pseudo_bytes((void*)&dsphys->ds_guid,
1919 	    sizeof (dsphys->ds_guid));
1920 	dsphys->ds_prev_snap_obj = ds->ds_phys->ds_prev_snap_obj;
1921 	dsphys->ds_prev_snap_txg = ds->ds_phys->ds_prev_snap_txg;
1922 	dsphys->ds_next_snap_obj = ds->ds_object;
1923 	dsphys->ds_num_children = 1;
1924 	dsphys->ds_creation_time = gethrestime_sec();
1925 	dsphys->ds_creation_txg = crtxg;
1926 	dsphys->ds_deadlist_obj = ds->ds_phys->ds_deadlist_obj;
1927 	dsphys->ds_used_bytes = ds->ds_phys->ds_used_bytes;
1928 	dsphys->ds_compressed_bytes = ds->ds_phys->ds_compressed_bytes;
1929 	dsphys->ds_uncompressed_bytes = ds->ds_phys->ds_uncompressed_bytes;
1930 	dsphys->ds_flags = ds->ds_phys->ds_flags;
1931 	dsphys->ds_bp = ds->ds_phys->ds_bp;
1932 	dmu_buf_rele(dbuf, FTAG);
1933 
1934 	ASSERT3U(ds->ds_prev != 0, ==, ds->ds_phys->ds_prev_snap_obj != 0);
1935 	if (ds->ds_prev) {
1936 		uint64_t next_clones_obj =
1937 		    ds->ds_prev->ds_phys->ds_next_clones_obj;
1938 		ASSERT(ds->ds_prev->ds_phys->ds_next_snap_obj ==
1939 		    ds->ds_object ||
1940 		    ds->ds_prev->ds_phys->ds_num_children > 1);
1941 		if (ds->ds_prev->ds_phys->ds_next_snap_obj == ds->ds_object) {
1942 			dmu_buf_will_dirty(ds->ds_prev->ds_dbuf, tx);
1943 			ASSERT3U(ds->ds_phys->ds_prev_snap_txg, ==,
1944 			    ds->ds_prev->ds_phys->ds_creation_txg);
1945 			ds->ds_prev->ds_phys->ds_next_snap_obj = dsobj;
1946 		} else if (next_clones_obj != 0) {
1947 			remove_from_next_clones(ds->ds_prev,
1948 			    dsphys->ds_next_snap_obj, tx);
1949 			VERIFY3U(0, ==, zap_add_int(mos,
1950 			    next_clones_obj, dsobj, tx));
1951 		}
1952 	}
1953 
1954 	/*
1955 	 * If we have a reference-reservation on this dataset, we will
1956 	 * need to increase the amount of refreservation being charged
1957 	 * since our unique space is going to zero.
1958 	 */
1959 	if (ds->ds_reserved) {
1960 		int64_t add = MIN(dsl_dataset_unique(ds), ds->ds_reserved);
1961 		dsl_dir_diduse_space(ds->ds_dir, DD_USED_REFRSRV,
1962 		    add, 0, 0, tx);
1963 	}
1964 
1965 	bplist_close(&ds->ds_deadlist);
1966 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
1967 	ASSERT3U(ds->ds_phys->ds_prev_snap_txg, <, tx->tx_txg);
1968 	ds->ds_phys->ds_prev_snap_obj = dsobj;
1969 	ds->ds_phys->ds_prev_snap_txg = crtxg;
1970 	ds->ds_phys->ds_unique_bytes = 0;
1971 	if (spa_version(dp->dp_spa) >= SPA_VERSION_UNIQUE_ACCURATE)
1972 		ds->ds_phys->ds_flags |= DS_FLAG_UNIQUE_ACCURATE;
1973 	ds->ds_phys->ds_deadlist_obj =
1974 	    bplist_create(mos, DSL_DEADLIST_BLOCKSIZE, tx);
1975 	VERIFY(0 == bplist_open(&ds->ds_deadlist, mos,
1976 	    ds->ds_phys->ds_deadlist_obj));
1977 
1978 	dprintf("snap '%s' -> obj %llu\n", snapname, dsobj);
1979 	err = zap_add(mos, ds->ds_phys->ds_snapnames_zapobj,
1980 	    snapname, 8, 1, &dsobj, tx);
1981 	ASSERT(err == 0);
1982 
1983 	if (ds->ds_prev)
1984 		dsl_dataset_drop_ref(ds->ds_prev, ds);
1985 	VERIFY(0 == dsl_dataset_get_ref(dp,
1986 	    ds->ds_phys->ds_prev_snap_obj, ds, &ds->ds_prev));
1987 
1988 	dsl_pool_ds_snapshotted(ds, tx);
1989 
1990 	dsl_dir_snap_cmtime_update(ds->ds_dir);
1991 
1992 	spa_history_internal_log(LOG_DS_SNAPSHOT, dp->dp_spa, tx, cr,
1993 	    "dataset = %llu", dsobj);
1994 }
1995 
1996 void
1997 dsl_dataset_sync(dsl_dataset_t *ds, zio_t *zio, dmu_tx_t *tx)
1998 {
1999 	ASSERT(dmu_tx_is_syncing(tx));
2000 	ASSERT(ds->ds_objset != NULL);
2001 	ASSERT(ds->ds_phys->ds_next_snap_obj == 0);
2002 
2003 	/*
2004 	 * in case we had to change ds_fsid_guid when we opened it,
2005 	 * sync it out now.
2006 	 */
2007 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
2008 	ds->ds_phys->ds_fsid_guid = ds->ds_fsid_guid;
2009 
2010 	dsl_dir_dirty(ds->ds_dir, tx);
2011 	dmu_objset_sync(ds->ds_objset, zio, tx);
2012 }
2013 
2014 void
2015 dsl_dataset_stats(dsl_dataset_t *ds, nvlist_t *nv)
2016 {
2017 	uint64_t refd, avail, uobjs, aobjs;
2018 
2019 	dsl_dir_stats(ds->ds_dir, nv);
2020 
2021 	dsl_dataset_space(ds, &refd, &avail, &uobjs, &aobjs);
2022 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_AVAILABLE, avail);
2023 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFERENCED, refd);
2024 
2025 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_CREATION,
2026 	    ds->ds_phys->ds_creation_time);
2027 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_CREATETXG,
2028 	    ds->ds_phys->ds_creation_txg);
2029 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFQUOTA,
2030 	    ds->ds_quota);
2031 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFRESERVATION,
2032 	    ds->ds_reserved);
2033 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_GUID,
2034 	    ds->ds_phys->ds_guid);
2035 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_UNIQUE,
2036 	    dsl_dataset_unique(ds));
2037 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_OBJSETID,
2038 	    ds->ds_object);
2039 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_USERREFS,
2040 	    ds->ds_userrefs);
2041 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_DEFER_DESTROY,
2042 	    DS_IS_DEFER_DESTROY(ds) ? 1 : 0);
2043 
2044 	if (ds->ds_phys->ds_next_snap_obj) {
2045 		/*
2046 		 * This is a snapshot; override the dd's space used with
2047 		 * our unique space and compression ratio.
2048 		 */
2049 		dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_USED,
2050 		    ds->ds_phys->ds_unique_bytes);
2051 		dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_COMPRESSRATIO,
2052 		    ds->ds_phys->ds_compressed_bytes == 0 ? 100 :
2053 		    (ds->ds_phys->ds_uncompressed_bytes * 100 /
2054 		    ds->ds_phys->ds_compressed_bytes));
2055 	}
2056 }
2057 
2058 void
2059 dsl_dataset_fast_stat(dsl_dataset_t *ds, dmu_objset_stats_t *stat)
2060 {
2061 	stat->dds_creation_txg = ds->ds_phys->ds_creation_txg;
2062 	stat->dds_inconsistent = ds->ds_phys->ds_flags & DS_FLAG_INCONSISTENT;
2063 	stat->dds_guid = ds->ds_phys->ds_guid;
2064 	if (ds->ds_phys->ds_next_snap_obj) {
2065 		stat->dds_is_snapshot = B_TRUE;
2066 		stat->dds_num_clones = ds->ds_phys->ds_num_children - 1;
2067 	} else {
2068 		stat->dds_is_snapshot = B_FALSE;
2069 		stat->dds_num_clones = 0;
2070 	}
2071 
2072 	/* clone origin is really a dsl_dir thing... */
2073 	rw_enter(&ds->ds_dir->dd_pool->dp_config_rwlock, RW_READER);
2074 	if (dsl_dir_is_clone(ds->ds_dir)) {
2075 		dsl_dataset_t *ods;
2076 
2077 		VERIFY(0 == dsl_dataset_get_ref(ds->ds_dir->dd_pool,
2078 		    ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &ods));
2079 		dsl_dataset_name(ods, stat->dds_origin);
2080 		dsl_dataset_drop_ref(ods, FTAG);
2081 	} else {
2082 		stat->dds_origin[0] = '\0';
2083 	}
2084 	rw_exit(&ds->ds_dir->dd_pool->dp_config_rwlock);
2085 }
2086 
2087 uint64_t
2088 dsl_dataset_fsid_guid(dsl_dataset_t *ds)
2089 {
2090 	return (ds->ds_fsid_guid);
2091 }
2092 
2093 void
2094 dsl_dataset_space(dsl_dataset_t *ds,
2095     uint64_t *refdbytesp, uint64_t *availbytesp,
2096     uint64_t *usedobjsp, uint64_t *availobjsp)
2097 {
2098 	*refdbytesp = ds->ds_phys->ds_used_bytes;
2099 	*availbytesp = dsl_dir_space_available(ds->ds_dir, NULL, 0, TRUE);
2100 	if (ds->ds_reserved > ds->ds_phys->ds_unique_bytes)
2101 		*availbytesp += ds->ds_reserved - ds->ds_phys->ds_unique_bytes;
2102 	if (ds->ds_quota != 0) {
2103 		/*
2104 		 * Adjust available bytes according to refquota
2105 		 */
2106 		if (*refdbytesp < ds->ds_quota)
2107 			*availbytesp = MIN(*availbytesp,
2108 			    ds->ds_quota - *refdbytesp);
2109 		else
2110 			*availbytesp = 0;
2111 	}
2112 	*usedobjsp = ds->ds_phys->ds_bp.blk_fill;
2113 	*availobjsp = DN_MAX_OBJECT - *usedobjsp;
2114 }
2115 
2116 boolean_t
2117 dsl_dataset_modified_since_lastsnap(dsl_dataset_t *ds)
2118 {
2119 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
2120 
2121 	ASSERT(RW_LOCK_HELD(&dp->dp_config_rwlock) ||
2122 	    dsl_pool_sync_context(dp));
2123 	if (ds->ds_prev == NULL)
2124 		return (B_FALSE);
2125 	if (ds->ds_phys->ds_bp.blk_birth >
2126 	    ds->ds_prev->ds_phys->ds_creation_txg)
2127 		return (B_TRUE);
2128 	return (B_FALSE);
2129 }
2130 
2131 /* ARGSUSED */
2132 static int
2133 dsl_dataset_snapshot_rename_check(void *arg1, void *arg2, dmu_tx_t *tx)
2134 {
2135 	dsl_dataset_t *ds = arg1;
2136 	char *newsnapname = arg2;
2137 	dsl_dir_t *dd = ds->ds_dir;
2138 	dsl_dataset_t *hds;
2139 	uint64_t val;
2140 	int err;
2141 
2142 	err = dsl_dataset_hold_obj(dd->dd_pool,
2143 	    dd->dd_phys->dd_head_dataset_obj, FTAG, &hds);
2144 	if (err)
2145 		return (err);
2146 
2147 	/* new name better not be in use */
2148 	err = dsl_dataset_snap_lookup(hds, newsnapname, &val);
2149 	dsl_dataset_rele(hds, FTAG);
2150 
2151 	if (err == 0)
2152 		err = EEXIST;
2153 	else if (err == ENOENT)
2154 		err = 0;
2155 
2156 	/* dataset name + 1 for the "@" + the new snapshot name must fit */
2157 	if (dsl_dir_namelen(ds->ds_dir) + 1 + strlen(newsnapname) >= MAXNAMELEN)
2158 		err = ENAMETOOLONG;
2159 
2160 	return (err);
2161 }
2162 
2163 static void
2164 dsl_dataset_snapshot_rename_sync(void *arg1, void *arg2,
2165     cred_t *cr, dmu_tx_t *tx)
2166 {
2167 	dsl_dataset_t *ds = arg1;
2168 	const char *newsnapname = arg2;
2169 	dsl_dir_t *dd = ds->ds_dir;
2170 	objset_t *mos = dd->dd_pool->dp_meta_objset;
2171 	dsl_dataset_t *hds;
2172 	int err;
2173 
2174 	ASSERT(ds->ds_phys->ds_next_snap_obj != 0);
2175 
2176 	VERIFY(0 == dsl_dataset_hold_obj(dd->dd_pool,
2177 	    dd->dd_phys->dd_head_dataset_obj, FTAG, &hds));
2178 
2179 	VERIFY(0 == dsl_dataset_get_snapname(ds));
2180 	err = dsl_dataset_snap_remove(hds, ds->ds_snapname, tx);
2181 	ASSERT3U(err, ==, 0);
2182 	mutex_enter(&ds->ds_lock);
2183 	(void) strcpy(ds->ds_snapname, newsnapname);
2184 	mutex_exit(&ds->ds_lock);
2185 	err = zap_add(mos, hds->ds_phys->ds_snapnames_zapobj,
2186 	    ds->ds_snapname, 8, 1, &ds->ds_object, tx);
2187 	ASSERT3U(err, ==, 0);
2188 
2189 	spa_history_internal_log(LOG_DS_RENAME, dd->dd_pool->dp_spa, tx,
2190 	    cr, "dataset = %llu", ds->ds_object);
2191 	dsl_dataset_rele(hds, FTAG);
2192 }
2193 
2194 struct renamesnaparg {
2195 	dsl_sync_task_group_t *dstg;
2196 	char failed[MAXPATHLEN];
2197 	char *oldsnap;
2198 	char *newsnap;
2199 };
2200 
2201 static int
2202 dsl_snapshot_rename_one(const char *name, void *arg)
2203 {
2204 	struct renamesnaparg *ra = arg;
2205 	dsl_dataset_t *ds = NULL;
2206 	char *snapname;
2207 	int err;
2208 
2209 	snapname = kmem_asprintf("%s@%s", name, ra->oldsnap);
2210 	(void) strlcpy(ra->failed, snapname, sizeof (ra->failed));
2211 
2212 	/*
2213 	 * For recursive snapshot renames the parent won't be changing
2214 	 * so we just pass name for both the to/from argument.
2215 	 */
2216 	err = zfs_secpolicy_rename_perms(snapname, snapname, CRED());
2217 	if (err != 0) {
2218 		strfree(snapname);
2219 		return (err == ENOENT ? 0 : err);
2220 	}
2221 
2222 #ifdef _KERNEL
2223 	/*
2224 	 * For all filesystems undergoing rename, we'll need to unmount it.
2225 	 */
2226 	(void) zfs_unmount_snap(snapname, NULL);
2227 #endif
2228 	err = dsl_dataset_hold(snapname, ra->dstg, &ds);
2229 	strfree(snapname);
2230 	if (err != 0)
2231 		return (err == ENOENT ? 0 : err);
2232 
2233 	dsl_sync_task_create(ra->dstg, dsl_dataset_snapshot_rename_check,
2234 	    dsl_dataset_snapshot_rename_sync, ds, ra->newsnap, 0);
2235 
2236 	return (0);
2237 }
2238 
2239 static int
2240 dsl_recursive_rename(char *oldname, const char *newname)
2241 {
2242 	int err;
2243 	struct renamesnaparg *ra;
2244 	dsl_sync_task_t *dst;
2245 	spa_t *spa;
2246 	char *cp, *fsname = spa_strdup(oldname);
2247 	int len = strlen(oldname) + 1;
2248 
2249 	/* truncate the snapshot name to get the fsname */
2250 	cp = strchr(fsname, '@');
2251 	*cp = '\0';
2252 
2253 	err = spa_open(fsname, &spa, FTAG);
2254 	if (err) {
2255 		kmem_free(fsname, len);
2256 		return (err);
2257 	}
2258 	ra = kmem_alloc(sizeof (struct renamesnaparg), KM_SLEEP);
2259 	ra->dstg = dsl_sync_task_group_create(spa_get_dsl(spa));
2260 
2261 	ra->oldsnap = strchr(oldname, '@') + 1;
2262 	ra->newsnap = strchr(newname, '@') + 1;
2263 	*ra->failed = '\0';
2264 
2265 	err = dmu_objset_find(fsname, dsl_snapshot_rename_one, ra,
2266 	    DS_FIND_CHILDREN);
2267 	kmem_free(fsname, len);
2268 
2269 	if (err == 0) {
2270 		err = dsl_sync_task_group_wait(ra->dstg);
2271 	}
2272 
2273 	for (dst = list_head(&ra->dstg->dstg_tasks); dst;
2274 	    dst = list_next(&ra->dstg->dstg_tasks, dst)) {
2275 		dsl_dataset_t *ds = dst->dst_arg1;
2276 		if (dst->dst_err) {
2277 			dsl_dir_name(ds->ds_dir, ra->failed);
2278 			(void) strlcat(ra->failed, "@", sizeof (ra->failed));
2279 			(void) strlcat(ra->failed, ra->newsnap,
2280 			    sizeof (ra->failed));
2281 		}
2282 		dsl_dataset_rele(ds, ra->dstg);
2283 	}
2284 
2285 	if (err)
2286 		(void) strlcpy(oldname, ra->failed, sizeof (ra->failed));
2287 
2288 	dsl_sync_task_group_destroy(ra->dstg);
2289 	kmem_free(ra, sizeof (struct renamesnaparg));
2290 	spa_close(spa, FTAG);
2291 	return (err);
2292 }
2293 
2294 static int
2295 dsl_valid_rename(const char *oldname, void *arg)
2296 {
2297 	int delta = *(int *)arg;
2298 
2299 	if (strlen(oldname) + delta >= MAXNAMELEN)
2300 		return (ENAMETOOLONG);
2301 
2302 	return (0);
2303 }
2304 
2305 #pragma weak dmu_objset_rename = dsl_dataset_rename
2306 int
2307 dsl_dataset_rename(char *oldname, const char *newname, boolean_t recursive)
2308 {
2309 	dsl_dir_t *dd;
2310 	dsl_dataset_t *ds;
2311 	const char *tail;
2312 	int err;
2313 
2314 	err = dsl_dir_open(oldname, FTAG, &dd, &tail);
2315 	if (err)
2316 		return (err);
2317 
2318 	if (tail == NULL) {
2319 		int delta = strlen(newname) - strlen(oldname);
2320 
2321 		/* if we're growing, validate child name lengths */
2322 		if (delta > 0)
2323 			err = dmu_objset_find(oldname, dsl_valid_rename,
2324 			    &delta, DS_FIND_CHILDREN | DS_FIND_SNAPSHOTS);
2325 
2326 		if (err == 0)
2327 			err = dsl_dir_rename(dd, newname);
2328 		dsl_dir_close(dd, FTAG);
2329 		return (err);
2330 	}
2331 
2332 	if (tail[0] != '@') {
2333 		/* the name ended in a nonexistent component */
2334 		dsl_dir_close(dd, FTAG);
2335 		return (ENOENT);
2336 	}
2337 
2338 	dsl_dir_close(dd, FTAG);
2339 
2340 	/* new name must be snapshot in same filesystem */
2341 	tail = strchr(newname, '@');
2342 	if (tail == NULL)
2343 		return (EINVAL);
2344 	tail++;
2345 	if (strncmp(oldname, newname, tail - newname) != 0)
2346 		return (EXDEV);
2347 
2348 	if (recursive) {
2349 		err = dsl_recursive_rename(oldname, newname);
2350 	} else {
2351 		err = dsl_dataset_hold(oldname, FTAG, &ds);
2352 		if (err)
2353 			return (err);
2354 
2355 		err = dsl_sync_task_do(ds->ds_dir->dd_pool,
2356 		    dsl_dataset_snapshot_rename_check,
2357 		    dsl_dataset_snapshot_rename_sync, ds, (char *)tail, 1);
2358 
2359 		dsl_dataset_rele(ds, FTAG);
2360 	}
2361 
2362 	return (err);
2363 }
2364 
2365 struct promotenode {
2366 	list_node_t link;
2367 	dsl_dataset_t *ds;
2368 };
2369 
2370 struct promotearg {
2371 	list_t shared_snaps, origin_snaps, clone_snaps;
2372 	dsl_dataset_t *origin_origin, *origin_head;
2373 	uint64_t used, comp, uncomp, unique, cloneusedsnap, originusedsnap;
2374 	char *err_ds;
2375 };
2376 
2377 static int snaplist_space(list_t *l, uint64_t mintxg, uint64_t *spacep);
2378 
2379 /* ARGSUSED */
2380 static int
2381 dsl_dataset_promote_check(void *arg1, void *arg2, dmu_tx_t *tx)
2382 {
2383 	dsl_dataset_t *hds = arg1;
2384 	struct promotearg *pa = arg2;
2385 	struct promotenode *snap = list_head(&pa->shared_snaps);
2386 	dsl_dataset_t *origin_ds = snap->ds;
2387 	int err;
2388 
2389 	/* Check that it is a real clone */
2390 	if (!dsl_dir_is_clone(hds->ds_dir))
2391 		return (EINVAL);
2392 
2393 	/* Since this is so expensive, don't do the preliminary check */
2394 	if (!dmu_tx_is_syncing(tx))
2395 		return (0);
2396 
2397 	if (hds->ds_phys->ds_flags & DS_FLAG_NOPROMOTE)
2398 		return (EXDEV);
2399 
2400 	/* compute origin's new unique space */
2401 	snap = list_tail(&pa->clone_snaps);
2402 	ASSERT3U(snap->ds->ds_phys->ds_prev_snap_obj, ==, origin_ds->ds_object);
2403 	err = bplist_space_birthrange(&snap->ds->ds_deadlist,
2404 	    origin_ds->ds_phys->ds_prev_snap_txg, UINT64_MAX, &pa->unique);
2405 	if (err)
2406 		return (err);
2407 
2408 	/*
2409 	 * Walk the snapshots that we are moving
2410 	 *
2411 	 * Compute space to transfer.  Consider the incremental changes
2412 	 * to used for each snapshot:
2413 	 * (my used) = (prev's used) + (blocks born) - (blocks killed)
2414 	 * So each snapshot gave birth to:
2415 	 * (blocks born) = (my used) - (prev's used) + (blocks killed)
2416 	 * So a sequence would look like:
2417 	 * (uN - u(N-1) + kN) + ... + (u1 - u0 + k1) + (u0 - 0 + k0)
2418 	 * Which simplifies to:
2419 	 * uN + kN + kN-1 + ... + k1 + k0
2420 	 * Note however, if we stop before we reach the ORIGIN we get:
2421 	 * uN + kN + kN-1 + ... + kM - uM-1
2422 	 */
2423 	pa->used = origin_ds->ds_phys->ds_used_bytes;
2424 	pa->comp = origin_ds->ds_phys->ds_compressed_bytes;
2425 	pa->uncomp = origin_ds->ds_phys->ds_uncompressed_bytes;
2426 	for (snap = list_head(&pa->shared_snaps); snap;
2427 	    snap = list_next(&pa->shared_snaps, snap)) {
2428 		uint64_t val, dlused, dlcomp, dluncomp;
2429 		dsl_dataset_t *ds = snap->ds;
2430 
2431 		/* Check that the snapshot name does not conflict */
2432 		VERIFY(0 == dsl_dataset_get_snapname(ds));
2433 		err = dsl_dataset_snap_lookup(hds, ds->ds_snapname, &val);
2434 		if (err == 0) {
2435 			err = EEXIST;
2436 			goto out;
2437 		}
2438 		if (err != ENOENT)
2439 			goto out;
2440 
2441 		/* The very first snapshot does not have a deadlist */
2442 		if (ds->ds_phys->ds_prev_snap_obj == 0)
2443 			continue;
2444 
2445 		if (err = bplist_space(&ds->ds_deadlist,
2446 		    &dlused, &dlcomp, &dluncomp))
2447 			goto out;
2448 		pa->used += dlused;
2449 		pa->comp += dlcomp;
2450 		pa->uncomp += dluncomp;
2451 	}
2452 
2453 	/*
2454 	 * If we are a clone of a clone then we never reached ORIGIN,
2455 	 * so we need to subtract out the clone origin's used space.
2456 	 */
2457 	if (pa->origin_origin) {
2458 		pa->used -= pa->origin_origin->ds_phys->ds_used_bytes;
2459 		pa->comp -= pa->origin_origin->ds_phys->ds_compressed_bytes;
2460 		pa->uncomp -= pa->origin_origin->ds_phys->ds_uncompressed_bytes;
2461 	}
2462 
2463 	/* Check that there is enough space here */
2464 	err = dsl_dir_transfer_possible(origin_ds->ds_dir, hds->ds_dir,
2465 	    pa->used);
2466 	if (err)
2467 		return (err);
2468 
2469 	/*
2470 	 * Compute the amounts of space that will be used by snapshots
2471 	 * after the promotion (for both origin and clone).  For each,
2472 	 * it is the amount of space that will be on all of their
2473 	 * deadlists (that was not born before their new origin).
2474 	 */
2475 	if (hds->ds_dir->dd_phys->dd_flags & DD_FLAG_USED_BREAKDOWN) {
2476 		uint64_t space;
2477 
2478 		/*
2479 		 * Note, typically this will not be a clone of a clone,
2480 		 * so snap->ds->ds_origin_txg will be < TXG_INITIAL, so
2481 		 * these snaplist_space() -> bplist_space_birthrange()
2482 		 * calls will be fast because they do not have to
2483 		 * iterate over all bps.
2484 		 */
2485 		snap = list_head(&pa->origin_snaps);
2486 		err = snaplist_space(&pa->shared_snaps,
2487 		    snap->ds->ds_origin_txg, &pa->cloneusedsnap);
2488 		if (err)
2489 			return (err);
2490 
2491 		err = snaplist_space(&pa->clone_snaps,
2492 		    snap->ds->ds_origin_txg, &space);
2493 		if (err)
2494 			return (err);
2495 		pa->cloneusedsnap += space;
2496 	}
2497 	if (origin_ds->ds_dir->dd_phys->dd_flags & DD_FLAG_USED_BREAKDOWN) {
2498 		err = snaplist_space(&pa->origin_snaps,
2499 		    origin_ds->ds_phys->ds_creation_txg, &pa->originusedsnap);
2500 		if (err)
2501 			return (err);
2502 	}
2503 
2504 	return (0);
2505 out:
2506 	pa->err_ds =  snap->ds->ds_snapname;
2507 	return (err);
2508 }
2509 
2510 static void
2511 dsl_dataset_promote_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
2512 {
2513 	dsl_dataset_t *hds = arg1;
2514 	struct promotearg *pa = arg2;
2515 	struct promotenode *snap = list_head(&pa->shared_snaps);
2516 	dsl_dataset_t *origin_ds = snap->ds;
2517 	dsl_dataset_t *origin_head;
2518 	dsl_dir_t *dd = hds->ds_dir;
2519 	dsl_pool_t *dp = hds->ds_dir->dd_pool;
2520 	dsl_dir_t *odd = NULL;
2521 	uint64_t oldnext_obj;
2522 	int64_t delta;
2523 
2524 	ASSERT(0 == (hds->ds_phys->ds_flags & DS_FLAG_NOPROMOTE));
2525 
2526 	snap = list_head(&pa->origin_snaps);
2527 	origin_head = snap->ds;
2528 
2529 	/*
2530 	 * We need to explicitly open odd, since origin_ds's dd will be
2531 	 * changing.
2532 	 */
2533 	VERIFY(0 == dsl_dir_open_obj(dp, origin_ds->ds_dir->dd_object,
2534 	    NULL, FTAG, &odd));
2535 
2536 	/* change origin's next snap */
2537 	dmu_buf_will_dirty(origin_ds->ds_dbuf, tx);
2538 	oldnext_obj = origin_ds->ds_phys->ds_next_snap_obj;
2539 	snap = list_tail(&pa->clone_snaps);
2540 	ASSERT3U(snap->ds->ds_phys->ds_prev_snap_obj, ==, origin_ds->ds_object);
2541 	origin_ds->ds_phys->ds_next_snap_obj = snap->ds->ds_object;
2542 
2543 	/* change the origin's next clone */
2544 	if (origin_ds->ds_phys->ds_next_clones_obj) {
2545 		remove_from_next_clones(origin_ds, snap->ds->ds_object, tx);
2546 		VERIFY3U(0, ==, zap_add_int(dp->dp_meta_objset,
2547 		    origin_ds->ds_phys->ds_next_clones_obj,
2548 		    oldnext_obj, tx));
2549 	}
2550 
2551 	/* change origin */
2552 	dmu_buf_will_dirty(dd->dd_dbuf, tx);
2553 	ASSERT3U(dd->dd_phys->dd_origin_obj, ==, origin_ds->ds_object);
2554 	dd->dd_phys->dd_origin_obj = odd->dd_phys->dd_origin_obj;
2555 	hds->ds_origin_txg = origin_head->ds_origin_txg;
2556 	dmu_buf_will_dirty(odd->dd_dbuf, tx);
2557 	odd->dd_phys->dd_origin_obj = origin_ds->ds_object;
2558 	origin_head->ds_origin_txg = origin_ds->ds_phys->ds_creation_txg;
2559 
2560 	/* move snapshots to this dir */
2561 	for (snap = list_head(&pa->shared_snaps); snap;
2562 	    snap = list_next(&pa->shared_snaps, snap)) {
2563 		dsl_dataset_t *ds = snap->ds;
2564 
2565 		/* unregister props as dsl_dir is changing */
2566 		if (ds->ds_objset) {
2567 			dmu_objset_evict(ds->ds_objset);
2568 			ds->ds_objset = NULL;
2569 		}
2570 		/* move snap name entry */
2571 		VERIFY(0 == dsl_dataset_get_snapname(ds));
2572 		VERIFY(0 == dsl_dataset_snap_remove(origin_head,
2573 		    ds->ds_snapname, tx));
2574 		VERIFY(0 == zap_add(dp->dp_meta_objset,
2575 		    hds->ds_phys->ds_snapnames_zapobj, ds->ds_snapname,
2576 		    8, 1, &ds->ds_object, tx));
2577 		/* change containing dsl_dir */
2578 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
2579 		ASSERT3U(ds->ds_phys->ds_dir_obj, ==, odd->dd_object);
2580 		ds->ds_phys->ds_dir_obj = dd->dd_object;
2581 		ASSERT3P(ds->ds_dir, ==, odd);
2582 		dsl_dir_close(ds->ds_dir, ds);
2583 		VERIFY(0 == dsl_dir_open_obj(dp, dd->dd_object,
2584 		    NULL, ds, &ds->ds_dir));
2585 
2586 		ASSERT3U(dsl_prop_numcb(ds), ==, 0);
2587 	}
2588 
2589 	/*
2590 	 * Change space accounting.
2591 	 * Note, pa->*usedsnap and dd_used_breakdown[SNAP] will either
2592 	 * both be valid, or both be 0 (resulting in delta == 0).  This
2593 	 * is true for each of {clone,origin} independently.
2594 	 */
2595 
2596 	delta = pa->cloneusedsnap -
2597 	    dd->dd_phys->dd_used_breakdown[DD_USED_SNAP];
2598 	ASSERT3S(delta, >=, 0);
2599 	ASSERT3U(pa->used, >=, delta);
2600 	dsl_dir_diduse_space(dd, DD_USED_SNAP, delta, 0, 0, tx);
2601 	dsl_dir_diduse_space(dd, DD_USED_HEAD,
2602 	    pa->used - delta, pa->comp, pa->uncomp, tx);
2603 
2604 	delta = pa->originusedsnap -
2605 	    odd->dd_phys->dd_used_breakdown[DD_USED_SNAP];
2606 	ASSERT3S(delta, <=, 0);
2607 	ASSERT3U(pa->used, >=, -delta);
2608 	dsl_dir_diduse_space(odd, DD_USED_SNAP, delta, 0, 0, tx);
2609 	dsl_dir_diduse_space(odd, DD_USED_HEAD,
2610 	    -pa->used - delta, -pa->comp, -pa->uncomp, tx);
2611 
2612 	origin_ds->ds_phys->ds_unique_bytes = pa->unique;
2613 
2614 	/* log history record */
2615 	spa_history_internal_log(LOG_DS_PROMOTE, dd->dd_pool->dp_spa, tx,
2616 	    cr, "dataset = %llu", hds->ds_object);
2617 
2618 	dsl_dir_close(odd, FTAG);
2619 }
2620 
2621 static char *snaplist_tag = "snaplist";
2622 /*
2623  * Make a list of dsl_dataset_t's for the snapshots between first_obj
2624  * (exclusive) and last_obj (inclusive).  The list will be in reverse
2625  * order (last_obj will be the list_head()).  If first_obj == 0, do all
2626  * snapshots back to this dataset's origin.
2627  */
2628 static int
2629 snaplist_make(dsl_pool_t *dp, boolean_t own,
2630     uint64_t first_obj, uint64_t last_obj, list_t *l)
2631 {
2632 	uint64_t obj = last_obj;
2633 
2634 	ASSERT(RW_LOCK_HELD(&dp->dp_config_rwlock));
2635 
2636 	list_create(l, sizeof (struct promotenode),
2637 	    offsetof(struct promotenode, link));
2638 
2639 	while (obj != first_obj) {
2640 		dsl_dataset_t *ds;
2641 		struct promotenode *snap;
2642 		int err;
2643 
2644 		if (own) {
2645 			err = dsl_dataset_own_obj(dp, obj,
2646 			    0, snaplist_tag, &ds);
2647 			if (err == 0)
2648 				dsl_dataset_make_exclusive(ds, snaplist_tag);
2649 		} else {
2650 			err = dsl_dataset_hold_obj(dp, obj, snaplist_tag, &ds);
2651 		}
2652 		if (err == ENOENT) {
2653 			/* lost race with snapshot destroy */
2654 			struct promotenode *last = list_tail(l);
2655 			ASSERT(obj != last->ds->ds_phys->ds_prev_snap_obj);
2656 			obj = last->ds->ds_phys->ds_prev_snap_obj;
2657 			continue;
2658 		} else if (err) {
2659 			return (err);
2660 		}
2661 
2662 		if (first_obj == 0)
2663 			first_obj = ds->ds_dir->dd_phys->dd_origin_obj;
2664 
2665 		snap = kmem_alloc(sizeof (struct promotenode), KM_SLEEP);
2666 		snap->ds = ds;
2667 		list_insert_tail(l, snap);
2668 		obj = ds->ds_phys->ds_prev_snap_obj;
2669 	}
2670 
2671 	return (0);
2672 }
2673 
2674 static int
2675 snaplist_space(list_t *l, uint64_t mintxg, uint64_t *spacep)
2676 {
2677 	struct promotenode *snap;
2678 
2679 	*spacep = 0;
2680 	for (snap = list_head(l); snap; snap = list_next(l, snap)) {
2681 		uint64_t used;
2682 		int err = bplist_space_birthrange(&snap->ds->ds_deadlist,
2683 		    mintxg, UINT64_MAX, &used);
2684 		if (err)
2685 			return (err);
2686 		*spacep += used;
2687 	}
2688 	return (0);
2689 }
2690 
2691 static void
2692 snaplist_destroy(list_t *l, boolean_t own)
2693 {
2694 	struct promotenode *snap;
2695 
2696 	if (!l || !list_link_active(&l->list_head))
2697 		return;
2698 
2699 	while ((snap = list_tail(l)) != NULL) {
2700 		list_remove(l, snap);
2701 		if (own)
2702 			dsl_dataset_disown(snap->ds, snaplist_tag);
2703 		else
2704 			dsl_dataset_rele(snap->ds, snaplist_tag);
2705 		kmem_free(snap, sizeof (struct promotenode));
2706 	}
2707 	list_destroy(l);
2708 }
2709 
2710 /*
2711  * Promote a clone.  Nomenclature note:
2712  * "clone" or "cds": the original clone which is being promoted
2713  * "origin" or "ods": the snapshot which is originally clone's origin
2714  * "origin head" or "ohds": the dataset which is the head
2715  * (filesystem/volume) for the origin
2716  * "origin origin": the origin of the origin's filesystem (typically
2717  * NULL, indicating that the clone is not a clone of a clone).
2718  */
2719 int
2720 dsl_dataset_promote(const char *name, char *conflsnap)
2721 {
2722 	dsl_dataset_t *ds;
2723 	dsl_dir_t *dd;
2724 	dsl_pool_t *dp;
2725 	dmu_object_info_t doi;
2726 	struct promotearg pa = { 0 };
2727 	struct promotenode *snap;
2728 	int err;
2729 
2730 	err = dsl_dataset_hold(name, FTAG, &ds);
2731 	if (err)
2732 		return (err);
2733 	dd = ds->ds_dir;
2734 	dp = dd->dd_pool;
2735 
2736 	err = dmu_object_info(dp->dp_meta_objset,
2737 	    ds->ds_phys->ds_snapnames_zapobj, &doi);
2738 	if (err) {
2739 		dsl_dataset_rele(ds, FTAG);
2740 		return (err);
2741 	}
2742 
2743 	if (dsl_dataset_is_snapshot(ds) || dd->dd_phys->dd_origin_obj == 0) {
2744 		dsl_dataset_rele(ds, FTAG);
2745 		return (EINVAL);
2746 	}
2747 
2748 	/*
2749 	 * We are going to inherit all the snapshots taken before our
2750 	 * origin (i.e., our new origin will be our parent's origin).
2751 	 * Take ownership of them so that we can rename them into our
2752 	 * namespace.
2753 	 */
2754 	rw_enter(&dp->dp_config_rwlock, RW_READER);
2755 
2756 	err = snaplist_make(dp, B_TRUE, 0, dd->dd_phys->dd_origin_obj,
2757 	    &pa.shared_snaps);
2758 	if (err != 0)
2759 		goto out;
2760 
2761 	err = snaplist_make(dp, B_FALSE, 0, ds->ds_object, &pa.clone_snaps);
2762 	if (err != 0)
2763 		goto out;
2764 
2765 	snap = list_head(&pa.shared_snaps);
2766 	ASSERT3U(snap->ds->ds_object, ==, dd->dd_phys->dd_origin_obj);
2767 	err = snaplist_make(dp, B_FALSE, dd->dd_phys->dd_origin_obj,
2768 	    snap->ds->ds_dir->dd_phys->dd_head_dataset_obj, &pa.origin_snaps);
2769 	if (err != 0)
2770 		goto out;
2771 
2772 	if (dsl_dir_is_clone(snap->ds->ds_dir)) {
2773 		err = dsl_dataset_own_obj(dp,
2774 		    snap->ds->ds_dir->dd_phys->dd_origin_obj,
2775 		    0, FTAG, &pa.origin_origin);
2776 		if (err != 0)
2777 			goto out;
2778 	}
2779 
2780 out:
2781 	rw_exit(&dp->dp_config_rwlock);
2782 
2783 	/*
2784 	 * Add in 128x the snapnames zapobj size, since we will be moving
2785 	 * a bunch of snapnames to the promoted ds, and dirtying their
2786 	 * bonus buffers.
2787 	 */
2788 	if (err == 0) {
2789 		err = dsl_sync_task_do(dp, dsl_dataset_promote_check,
2790 		    dsl_dataset_promote_sync, ds, &pa,
2791 		    2 + 2 * doi.doi_physical_blocks_512);
2792 		if (err && pa.err_ds && conflsnap)
2793 			(void) strncpy(conflsnap, pa.err_ds, MAXNAMELEN);
2794 	}
2795 
2796 	snaplist_destroy(&pa.shared_snaps, B_TRUE);
2797 	snaplist_destroy(&pa.clone_snaps, B_FALSE);
2798 	snaplist_destroy(&pa.origin_snaps, B_FALSE);
2799 	if (pa.origin_origin)
2800 		dsl_dataset_disown(pa.origin_origin, FTAG);
2801 	dsl_dataset_rele(ds, FTAG);
2802 	return (err);
2803 }
2804 
2805 struct cloneswaparg {
2806 	dsl_dataset_t *cds; /* clone dataset */
2807 	dsl_dataset_t *ohds; /* origin's head dataset */
2808 	boolean_t force;
2809 	int64_t unused_refres_delta; /* change in unconsumed refreservation */
2810 };
2811 
2812 /* ARGSUSED */
2813 static int
2814 dsl_dataset_clone_swap_check(void *arg1, void *arg2, dmu_tx_t *tx)
2815 {
2816 	struct cloneswaparg *csa = arg1;
2817 
2818 	/* they should both be heads */
2819 	if (dsl_dataset_is_snapshot(csa->cds) ||
2820 	    dsl_dataset_is_snapshot(csa->ohds))
2821 		return (EINVAL);
2822 
2823 	/* the branch point should be just before them */
2824 	if (csa->cds->ds_prev != csa->ohds->ds_prev)
2825 		return (EINVAL);
2826 
2827 	/* cds should be the clone (unless they are unrelated) */
2828 	if (csa->cds->ds_prev != NULL &&
2829 	    csa->cds->ds_prev != csa->cds->ds_dir->dd_pool->dp_origin_snap &&
2830 	    csa->ohds->ds_object !=
2831 	    csa->cds->ds_prev->ds_phys->ds_next_snap_obj)
2832 		return (EINVAL);
2833 
2834 	/* the clone should be a child of the origin */
2835 	if (csa->cds->ds_dir->dd_parent != csa->ohds->ds_dir)
2836 		return (EINVAL);
2837 
2838 	/* ohds shouldn't be modified unless 'force' */
2839 	if (!csa->force && dsl_dataset_modified_since_lastsnap(csa->ohds))
2840 		return (ETXTBSY);
2841 
2842 	/* adjust amount of any unconsumed refreservation */
2843 	csa->unused_refres_delta =
2844 	    (int64_t)MIN(csa->ohds->ds_reserved,
2845 	    csa->ohds->ds_phys->ds_unique_bytes) -
2846 	    (int64_t)MIN(csa->ohds->ds_reserved,
2847 	    csa->cds->ds_phys->ds_unique_bytes);
2848 
2849 	if (csa->unused_refres_delta > 0 &&
2850 	    csa->unused_refres_delta >
2851 	    dsl_dir_space_available(csa->ohds->ds_dir, NULL, 0, TRUE))
2852 		return (ENOSPC);
2853 
2854 	if (csa->ohds->ds_quota != 0 &&
2855 	    csa->cds->ds_phys->ds_unique_bytes > csa->ohds->ds_quota)
2856 		return (EDQUOT);
2857 
2858 	return (0);
2859 }
2860 
2861 /* ARGSUSED */
2862 static void
2863 dsl_dataset_clone_swap_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
2864 {
2865 	struct cloneswaparg *csa = arg1;
2866 	dsl_pool_t *dp = csa->cds->ds_dir->dd_pool;
2867 
2868 	ASSERT(csa->cds->ds_reserved == 0);
2869 	ASSERT(csa->ohds->ds_quota == 0 ||
2870 	    csa->cds->ds_phys->ds_unique_bytes <= csa->ohds->ds_quota);
2871 
2872 	dmu_buf_will_dirty(csa->cds->ds_dbuf, tx);
2873 	dmu_buf_will_dirty(csa->ohds->ds_dbuf, tx);
2874 
2875 	if (csa->cds->ds_objset != NULL) {
2876 		dmu_objset_evict(csa->cds->ds_objset);
2877 		csa->cds->ds_objset = NULL;
2878 	}
2879 
2880 	if (csa->ohds->ds_objset != NULL) {
2881 		dmu_objset_evict(csa->ohds->ds_objset);
2882 		csa->ohds->ds_objset = NULL;
2883 	}
2884 
2885 	/*
2886 	 * Reset origin's unique bytes, if it exists.
2887 	 */
2888 	if (csa->cds->ds_prev) {
2889 		dsl_dataset_t *origin = csa->cds->ds_prev;
2890 		dmu_buf_will_dirty(origin->ds_dbuf, tx);
2891 		VERIFY(0 == bplist_space_birthrange(&csa->cds->ds_deadlist,
2892 		    origin->ds_phys->ds_prev_snap_txg, UINT64_MAX,
2893 		    &origin->ds_phys->ds_unique_bytes));
2894 	}
2895 
2896 	/* swap blkptrs */
2897 	{
2898 		blkptr_t tmp;
2899 		tmp = csa->ohds->ds_phys->ds_bp;
2900 		csa->ohds->ds_phys->ds_bp = csa->cds->ds_phys->ds_bp;
2901 		csa->cds->ds_phys->ds_bp = tmp;
2902 	}
2903 
2904 	/* set dd_*_bytes */
2905 	{
2906 		int64_t dused, dcomp, duncomp;
2907 		uint64_t cdl_used, cdl_comp, cdl_uncomp;
2908 		uint64_t odl_used, odl_comp, odl_uncomp;
2909 
2910 		ASSERT3U(csa->cds->ds_dir->dd_phys->
2911 		    dd_used_breakdown[DD_USED_SNAP], ==, 0);
2912 
2913 		VERIFY(0 == bplist_space(&csa->cds->ds_deadlist, &cdl_used,
2914 		    &cdl_comp, &cdl_uncomp));
2915 		VERIFY(0 == bplist_space(&csa->ohds->ds_deadlist, &odl_used,
2916 		    &odl_comp, &odl_uncomp));
2917 
2918 		dused = csa->cds->ds_phys->ds_used_bytes + cdl_used -
2919 		    (csa->ohds->ds_phys->ds_used_bytes + odl_used);
2920 		dcomp = csa->cds->ds_phys->ds_compressed_bytes + cdl_comp -
2921 		    (csa->ohds->ds_phys->ds_compressed_bytes + odl_comp);
2922 		duncomp = csa->cds->ds_phys->ds_uncompressed_bytes +
2923 		    cdl_uncomp -
2924 		    (csa->ohds->ds_phys->ds_uncompressed_bytes + odl_uncomp);
2925 
2926 		dsl_dir_diduse_space(csa->ohds->ds_dir, DD_USED_HEAD,
2927 		    dused, dcomp, duncomp, tx);
2928 		dsl_dir_diduse_space(csa->cds->ds_dir, DD_USED_HEAD,
2929 		    -dused, -dcomp, -duncomp, tx);
2930 
2931 		/*
2932 		 * The difference in the space used by snapshots is the
2933 		 * difference in snapshot space due to the head's
2934 		 * deadlist (since that's the only thing that's
2935 		 * changing that affects the snapused).
2936 		 */
2937 		VERIFY(0 == bplist_space_birthrange(&csa->cds->ds_deadlist,
2938 		    csa->ohds->ds_origin_txg, UINT64_MAX, &cdl_used));
2939 		VERIFY(0 == bplist_space_birthrange(&csa->ohds->ds_deadlist,
2940 		    csa->ohds->ds_origin_txg, UINT64_MAX, &odl_used));
2941 		dsl_dir_transfer_space(csa->ohds->ds_dir, cdl_used - odl_used,
2942 		    DD_USED_HEAD, DD_USED_SNAP, tx);
2943 	}
2944 
2945 #define	SWITCH64(x, y) \
2946 	{ \
2947 		uint64_t __tmp = (x); \
2948 		(x) = (y); \
2949 		(y) = __tmp; \
2950 	}
2951 
2952 	/* swap ds_*_bytes */
2953 	SWITCH64(csa->ohds->ds_phys->ds_used_bytes,
2954 	    csa->cds->ds_phys->ds_used_bytes);
2955 	SWITCH64(csa->ohds->ds_phys->ds_compressed_bytes,
2956 	    csa->cds->ds_phys->ds_compressed_bytes);
2957 	SWITCH64(csa->ohds->ds_phys->ds_uncompressed_bytes,
2958 	    csa->cds->ds_phys->ds_uncompressed_bytes);
2959 	SWITCH64(csa->ohds->ds_phys->ds_unique_bytes,
2960 	    csa->cds->ds_phys->ds_unique_bytes);
2961 
2962 	/* apply any parent delta for change in unconsumed refreservation */
2963 	dsl_dir_diduse_space(csa->ohds->ds_dir, DD_USED_REFRSRV,
2964 	    csa->unused_refres_delta, 0, 0, tx);
2965 
2966 	/* swap deadlists */
2967 	bplist_close(&csa->cds->ds_deadlist);
2968 	bplist_close(&csa->ohds->ds_deadlist);
2969 	SWITCH64(csa->ohds->ds_phys->ds_deadlist_obj,
2970 	    csa->cds->ds_phys->ds_deadlist_obj);
2971 	VERIFY(0 == bplist_open(&csa->cds->ds_deadlist, dp->dp_meta_objset,
2972 	    csa->cds->ds_phys->ds_deadlist_obj));
2973 	VERIFY(0 == bplist_open(&csa->ohds->ds_deadlist, dp->dp_meta_objset,
2974 	    csa->ohds->ds_phys->ds_deadlist_obj));
2975 
2976 	dsl_pool_ds_clone_swapped(csa->ohds, csa->cds, tx);
2977 }
2978 
2979 /*
2980  * Swap 'clone' with its origin head datasets.  Used at the end of "zfs
2981  * recv" into an existing fs to swizzle the file system to the new
2982  * version, and by "zfs rollback".  Can also be used to swap two
2983  * independent head datasets if neither has any snapshots.
2984  */
2985 int
2986 dsl_dataset_clone_swap(dsl_dataset_t *clone, dsl_dataset_t *origin_head,
2987     boolean_t force)
2988 {
2989 	struct cloneswaparg csa;
2990 	int error;
2991 
2992 	ASSERT(clone->ds_owner);
2993 	ASSERT(origin_head->ds_owner);
2994 retry:
2995 	/* Need exclusive access for the swap */
2996 	rw_enter(&clone->ds_rwlock, RW_WRITER);
2997 	if (!rw_tryenter(&origin_head->ds_rwlock, RW_WRITER)) {
2998 		rw_exit(&clone->ds_rwlock);
2999 		rw_enter(&origin_head->ds_rwlock, RW_WRITER);
3000 		if (!rw_tryenter(&clone->ds_rwlock, RW_WRITER)) {
3001 			rw_exit(&origin_head->ds_rwlock);
3002 			goto retry;
3003 		}
3004 	}
3005 	csa.cds = clone;
3006 	csa.ohds = origin_head;
3007 	csa.force = force;
3008 	error = dsl_sync_task_do(clone->ds_dir->dd_pool,
3009 	    dsl_dataset_clone_swap_check,
3010 	    dsl_dataset_clone_swap_sync, &csa, NULL, 9);
3011 	return (error);
3012 }
3013 
3014 /*
3015  * Given a pool name and a dataset object number in that pool,
3016  * return the name of that dataset.
3017  */
3018 int
3019 dsl_dsobj_to_dsname(char *pname, uint64_t obj, char *buf)
3020 {
3021 	spa_t *spa;
3022 	dsl_pool_t *dp;
3023 	dsl_dataset_t *ds;
3024 	int error;
3025 
3026 	if ((error = spa_open(pname, &spa, FTAG)) != 0)
3027 		return (error);
3028 	dp = spa_get_dsl(spa);
3029 	rw_enter(&dp->dp_config_rwlock, RW_READER);
3030 	if ((error = dsl_dataset_hold_obj(dp, obj, FTAG, &ds)) == 0) {
3031 		dsl_dataset_name(ds, buf);
3032 		dsl_dataset_rele(ds, FTAG);
3033 	}
3034 	rw_exit(&dp->dp_config_rwlock);
3035 	spa_close(spa, FTAG);
3036 
3037 	return (error);
3038 }
3039 
3040 int
3041 dsl_dataset_check_quota(dsl_dataset_t *ds, boolean_t check_quota,
3042     uint64_t asize, uint64_t inflight, uint64_t *used, uint64_t *ref_rsrv)
3043 {
3044 	int error = 0;
3045 
3046 	ASSERT3S(asize, >, 0);
3047 
3048 	/*
3049 	 * *ref_rsrv is the portion of asize that will come from any
3050 	 * unconsumed refreservation space.
3051 	 */
3052 	*ref_rsrv = 0;
3053 
3054 	mutex_enter(&ds->ds_lock);
3055 	/*
3056 	 * Make a space adjustment for reserved bytes.
3057 	 */
3058 	if (ds->ds_reserved > ds->ds_phys->ds_unique_bytes) {
3059 		ASSERT3U(*used, >=,
3060 		    ds->ds_reserved - ds->ds_phys->ds_unique_bytes);
3061 		*used -= (ds->ds_reserved - ds->ds_phys->ds_unique_bytes);
3062 		*ref_rsrv =
3063 		    asize - MIN(asize, parent_delta(ds, asize + inflight));
3064 	}
3065 
3066 	if (!check_quota || ds->ds_quota == 0) {
3067 		mutex_exit(&ds->ds_lock);
3068 		return (0);
3069 	}
3070 	/*
3071 	 * If they are requesting more space, and our current estimate
3072 	 * is over quota, they get to try again unless the actual
3073 	 * on-disk is over quota and there are no pending changes (which
3074 	 * may free up space for us).
3075 	 */
3076 	if (ds->ds_phys->ds_used_bytes + inflight >= ds->ds_quota) {
3077 		if (inflight > 0 || ds->ds_phys->ds_used_bytes < ds->ds_quota)
3078 			error = ERESTART;
3079 		else
3080 			error = EDQUOT;
3081 	}
3082 	mutex_exit(&ds->ds_lock);
3083 
3084 	return (error);
3085 }
3086 
3087 /* ARGSUSED */
3088 static int
3089 dsl_dataset_set_quota_check(void *arg1, void *arg2, dmu_tx_t *tx)
3090 {
3091 	dsl_dataset_t *ds = arg1;
3092 	dsl_prop_setarg_t *psa = arg2;
3093 	int err;
3094 
3095 	if (spa_version(ds->ds_dir->dd_pool->dp_spa) < SPA_VERSION_REFQUOTA)
3096 		return (ENOTSUP);
3097 
3098 	if ((err = dsl_prop_predict_sync(ds->ds_dir, psa)) != 0)
3099 		return (err);
3100 
3101 	if (psa->psa_effective_value == 0)
3102 		return (0);
3103 
3104 	if (psa->psa_effective_value < ds->ds_phys->ds_used_bytes ||
3105 	    psa->psa_effective_value < ds->ds_reserved)
3106 		return (ENOSPC);
3107 
3108 	return (0);
3109 }
3110 
3111 extern void dsl_prop_set_sync(void *, void *, cred_t *, dmu_tx_t *);
3112 
3113 void
3114 dsl_dataset_set_quota_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
3115 {
3116 	dsl_dataset_t *ds = arg1;
3117 	dsl_prop_setarg_t *psa = arg2;
3118 	uint64_t effective_value = psa->psa_effective_value;
3119 
3120 	dsl_prop_set_sync(ds, psa, cr, tx);
3121 	DSL_PROP_CHECK_PREDICTION(ds->ds_dir, psa);
3122 
3123 	if (ds->ds_quota != effective_value) {
3124 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
3125 		ds->ds_quota = effective_value;
3126 
3127 		spa_history_internal_log(LOG_DS_REFQUOTA,
3128 		    ds->ds_dir->dd_pool->dp_spa, tx, cr, "%lld dataset = %llu ",
3129 		    (longlong_t)ds->ds_quota, ds->ds_object);
3130 	}
3131 }
3132 
3133 int
3134 dsl_dataset_set_quota(const char *dsname, zprop_source_t source, uint64_t quota)
3135 {
3136 	dsl_dataset_t *ds;
3137 	dsl_prop_setarg_t psa;
3138 	int err;
3139 
3140 	dsl_prop_setarg_init_uint64(&psa, "refquota", source, &quota);
3141 
3142 	err = dsl_dataset_hold(dsname, FTAG, &ds);
3143 	if (err)
3144 		return (err);
3145 
3146 	/*
3147 	 * If someone removes a file, then tries to set the quota, we
3148 	 * want to make sure the file freeing takes effect.
3149 	 */
3150 	txg_wait_open(ds->ds_dir->dd_pool, 0);
3151 
3152 	err = dsl_sync_task_do(ds->ds_dir->dd_pool,
3153 	    dsl_dataset_set_quota_check, dsl_dataset_set_quota_sync,
3154 	    ds, &psa, 0);
3155 
3156 	dsl_dataset_rele(ds, FTAG);
3157 	return (err);
3158 }
3159 
3160 static int
3161 dsl_dataset_set_reservation_check(void *arg1, void *arg2, dmu_tx_t *tx)
3162 {
3163 	dsl_dataset_t *ds = arg1;
3164 	dsl_prop_setarg_t *psa = arg2;
3165 	uint64_t effective_value;
3166 	uint64_t unique;
3167 	int err;
3168 
3169 	if (spa_version(ds->ds_dir->dd_pool->dp_spa) <
3170 	    SPA_VERSION_REFRESERVATION)
3171 		return (ENOTSUP);
3172 
3173 	if (dsl_dataset_is_snapshot(ds))
3174 		return (EINVAL);
3175 
3176 	if ((err = dsl_prop_predict_sync(ds->ds_dir, psa)) != 0)
3177 		return (err);
3178 
3179 	effective_value = psa->psa_effective_value;
3180 
3181 	/*
3182 	 * If we are doing the preliminary check in open context, the
3183 	 * space estimates may be inaccurate.
3184 	 */
3185 	if (!dmu_tx_is_syncing(tx))
3186 		return (0);
3187 
3188 	mutex_enter(&ds->ds_lock);
3189 	unique = dsl_dataset_unique(ds);
3190 	mutex_exit(&ds->ds_lock);
3191 
3192 	if (MAX(unique, effective_value) > MAX(unique, ds->ds_reserved)) {
3193 		uint64_t delta = MAX(unique, effective_value) -
3194 		    MAX(unique, ds->ds_reserved);
3195 
3196 		if (delta > dsl_dir_space_available(ds->ds_dir, NULL, 0, TRUE))
3197 			return (ENOSPC);
3198 		if (ds->ds_quota > 0 &&
3199 		    effective_value > ds->ds_quota)
3200 			return (ENOSPC);
3201 	}
3202 
3203 	return (0);
3204 }
3205 
3206 /* ARGSUSED */
3207 static void
3208 dsl_dataset_set_reservation_sync(void *arg1, void *arg2, cred_t *cr,
3209     dmu_tx_t *tx)
3210 {
3211 	dsl_dataset_t *ds = arg1;
3212 	dsl_prop_setarg_t *psa = arg2;
3213 	uint64_t effective_value = psa->psa_effective_value;
3214 	uint64_t unique;
3215 	int64_t delta;
3216 
3217 	dsl_prop_set_sync(ds, psa, cr, tx);
3218 	DSL_PROP_CHECK_PREDICTION(ds->ds_dir, psa);
3219 
3220 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
3221 
3222 	mutex_enter(&ds->ds_dir->dd_lock);
3223 	mutex_enter(&ds->ds_lock);
3224 	unique = dsl_dataset_unique(ds);
3225 	delta = MAX(0, (int64_t)(effective_value - unique)) -
3226 	    MAX(0, (int64_t)(ds->ds_reserved - unique));
3227 	ds->ds_reserved = effective_value;
3228 	mutex_exit(&ds->ds_lock);
3229 
3230 	dsl_dir_diduse_space(ds->ds_dir, DD_USED_REFRSRV, delta, 0, 0, tx);
3231 	mutex_exit(&ds->ds_dir->dd_lock);
3232 
3233 	spa_history_internal_log(LOG_DS_REFRESERV,
3234 	    ds->ds_dir->dd_pool->dp_spa, tx, cr, "%lld dataset = %llu",
3235 	    (longlong_t)effective_value, ds->ds_object);
3236 }
3237 
3238 int
3239 dsl_dataset_set_reservation(const char *dsname, zprop_source_t source,
3240     uint64_t reservation)
3241 {
3242 	dsl_dataset_t *ds;
3243 	dsl_prop_setarg_t psa;
3244 	int err;
3245 
3246 	dsl_prop_setarg_init_uint64(&psa, "refreservation", source,
3247 	    &reservation);
3248 
3249 	err = dsl_dataset_hold(dsname, FTAG, &ds);
3250 	if (err)
3251 		return (err);
3252 
3253 	err = dsl_sync_task_do(ds->ds_dir->dd_pool,
3254 	    dsl_dataset_set_reservation_check,
3255 	    dsl_dataset_set_reservation_sync, ds, &psa, 0);
3256 
3257 	dsl_dataset_rele(ds, FTAG);
3258 	return (err);
3259 }
3260 
3261 struct dsl_ds_holdarg {
3262 	dsl_sync_task_group_t *dstg;
3263 	char *htag;
3264 	char *snapname;
3265 	boolean_t recursive;
3266 	boolean_t gotone;
3267 	boolean_t temphold;
3268 	char failed[MAXPATHLEN];
3269 };
3270 
3271 /*
3272  * The max length of a temporary tag prefix is the number of hex digits
3273  * required to express UINT64_MAX plus one for the hyphen.
3274  */
3275 #define	MAX_TAG_PREFIX_LEN	17
3276 
3277 static int
3278 dsl_dataset_user_hold_check(void *arg1, void *arg2, dmu_tx_t *tx)
3279 {
3280 	dsl_dataset_t *ds = arg1;
3281 	struct dsl_ds_holdarg *ha = arg2;
3282 	char *htag = ha->htag;
3283 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
3284 	int error = 0;
3285 
3286 	if (spa_version(ds->ds_dir->dd_pool->dp_spa) < SPA_VERSION_USERREFS)
3287 		return (ENOTSUP);
3288 
3289 	if (!dsl_dataset_is_snapshot(ds))
3290 		return (EINVAL);
3291 
3292 	/* tags must be unique */
3293 	mutex_enter(&ds->ds_lock);
3294 	if (ds->ds_phys->ds_userrefs_obj) {
3295 		error = zap_lookup(mos, ds->ds_phys->ds_userrefs_obj, htag,
3296 		    8, 1, tx);
3297 		if (error == 0)
3298 			error = EEXIST;
3299 		else if (error == ENOENT)
3300 			error = 0;
3301 	}
3302 	mutex_exit(&ds->ds_lock);
3303 
3304 	if (error == 0 && ha->temphold &&
3305 	    strlen(htag) + MAX_TAG_PREFIX_LEN >= MAXNAMELEN)
3306 		error = E2BIG;
3307 
3308 	return (error);
3309 }
3310 
3311 static void
3312 dsl_dataset_user_hold_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
3313 {
3314 	dsl_dataset_t *ds = arg1;
3315 	struct dsl_ds_holdarg *ha = arg2;
3316 	char *htag = ha->htag;
3317 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
3318 	objset_t *mos = dp->dp_meta_objset;
3319 	uint64_t now = gethrestime_sec();
3320 	uint64_t zapobj;
3321 
3322 	mutex_enter(&ds->ds_lock);
3323 	if (ds->ds_phys->ds_userrefs_obj == 0) {
3324 		/*
3325 		 * This is the first user hold for this dataset.  Create
3326 		 * the userrefs zap object.
3327 		 */
3328 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
3329 		zapobj = ds->ds_phys->ds_userrefs_obj =
3330 		    zap_create(mos, DMU_OT_USERREFS, DMU_OT_NONE, 0, tx);
3331 	} else {
3332 		zapobj = ds->ds_phys->ds_userrefs_obj;
3333 	}
3334 	ds->ds_userrefs++;
3335 	mutex_exit(&ds->ds_lock);
3336 
3337 	VERIFY(0 == zap_add(mos, zapobj, htag, 8, 1, &now, tx));
3338 
3339 	if (ha->temphold) {
3340 		VERIFY(0 == dsl_pool_user_hold(dp, ds->ds_object,
3341 		    htag, &now, tx));
3342 	}
3343 
3344 	spa_history_internal_log(LOG_DS_USER_HOLD,
3345 	    dp->dp_spa, tx, cr, "<%s> temp = %d dataset = %llu", htag,
3346 	    (int)ha->temphold, ds->ds_object);
3347 }
3348 
3349 static int
3350 dsl_dataset_user_hold_one(const char *dsname, void *arg)
3351 {
3352 	struct dsl_ds_holdarg *ha = arg;
3353 	dsl_dataset_t *ds;
3354 	int error;
3355 	char *name;
3356 
3357 	/* alloc a buffer to hold dsname@snapname plus terminating NULL */
3358 	name = kmem_asprintf("%s@%s", dsname, ha->snapname);
3359 	error = dsl_dataset_hold(name, ha->dstg, &ds);
3360 	strfree(name);
3361 	if (error == 0) {
3362 		ha->gotone = B_TRUE;
3363 		dsl_sync_task_create(ha->dstg, dsl_dataset_user_hold_check,
3364 		    dsl_dataset_user_hold_sync, ds, ha, 0);
3365 	} else if (error == ENOENT && ha->recursive) {
3366 		error = 0;
3367 	} else {
3368 		(void) strlcpy(ha->failed, dsname, sizeof (ha->failed));
3369 	}
3370 	return (error);
3371 }
3372 
3373 int
3374 dsl_dataset_user_hold(char *dsname, char *snapname, char *htag,
3375     boolean_t recursive, boolean_t temphold)
3376 {
3377 	struct dsl_ds_holdarg *ha;
3378 	dsl_sync_task_t *dst;
3379 	spa_t *spa;
3380 	int error;
3381 
3382 	ha = kmem_zalloc(sizeof (struct dsl_ds_holdarg), KM_SLEEP);
3383 
3384 	(void) strlcpy(ha->failed, dsname, sizeof (ha->failed));
3385 
3386 	error = spa_open(dsname, &spa, FTAG);
3387 	if (error) {
3388 		kmem_free(ha, sizeof (struct dsl_ds_holdarg));
3389 		return (error);
3390 	}
3391 
3392 	ha->dstg = dsl_sync_task_group_create(spa_get_dsl(spa));
3393 	ha->htag = htag;
3394 	ha->snapname = snapname;
3395 	ha->recursive = recursive;
3396 	ha->temphold = temphold;
3397 	if (recursive) {
3398 		error = dmu_objset_find(dsname, dsl_dataset_user_hold_one,
3399 		    ha, DS_FIND_CHILDREN);
3400 	} else {
3401 		error = dsl_dataset_user_hold_one(dsname, ha);
3402 	}
3403 	if (error == 0)
3404 		error = dsl_sync_task_group_wait(ha->dstg);
3405 
3406 	for (dst = list_head(&ha->dstg->dstg_tasks); dst;
3407 	    dst = list_next(&ha->dstg->dstg_tasks, dst)) {
3408 		dsl_dataset_t *ds = dst->dst_arg1;
3409 
3410 		if (dst->dst_err) {
3411 			dsl_dataset_name(ds, ha->failed);
3412 			*strchr(ha->failed, '@') = '\0';
3413 		}
3414 		dsl_dataset_rele(ds, ha->dstg);
3415 	}
3416 
3417 	if (error == 0 && recursive && !ha->gotone)
3418 		error = ENOENT;
3419 
3420 	if (error)
3421 		(void) strlcpy(dsname, ha->failed, sizeof (ha->failed));
3422 
3423 	dsl_sync_task_group_destroy(ha->dstg);
3424 	kmem_free(ha, sizeof (struct dsl_ds_holdarg));
3425 	spa_close(spa, FTAG);
3426 	return (error);
3427 }
3428 
3429 struct dsl_ds_releasearg {
3430 	dsl_dataset_t *ds;
3431 	const char *htag;
3432 	boolean_t own;		/* do we own or just hold ds? */
3433 };
3434 
3435 static int
3436 dsl_dataset_release_might_destroy(dsl_dataset_t *ds, const char *htag,
3437     boolean_t *might_destroy)
3438 {
3439 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
3440 	uint64_t zapobj;
3441 	uint64_t tmp;
3442 	int error;
3443 
3444 	*might_destroy = B_FALSE;
3445 
3446 	mutex_enter(&ds->ds_lock);
3447 	zapobj = ds->ds_phys->ds_userrefs_obj;
3448 	if (zapobj == 0) {
3449 		/* The tag can't possibly exist */
3450 		mutex_exit(&ds->ds_lock);
3451 		return (ESRCH);
3452 	}
3453 
3454 	/* Make sure the tag exists */
3455 	error = zap_lookup(mos, zapobj, htag, 8, 1, &tmp);
3456 	if (error) {
3457 		mutex_exit(&ds->ds_lock);
3458 		if (error == ENOENT)
3459 			error = ESRCH;
3460 		return (error);
3461 	}
3462 
3463 	if (ds->ds_userrefs == 1 && ds->ds_phys->ds_num_children == 1 &&
3464 	    DS_IS_DEFER_DESTROY(ds))
3465 		*might_destroy = B_TRUE;
3466 
3467 	mutex_exit(&ds->ds_lock);
3468 	return (0);
3469 }
3470 
3471 static int
3472 dsl_dataset_user_release_check(void *arg1, void *tag, dmu_tx_t *tx)
3473 {
3474 	struct dsl_ds_releasearg *ra = arg1;
3475 	dsl_dataset_t *ds = ra->ds;
3476 	boolean_t might_destroy;
3477 	int error;
3478 
3479 	if (spa_version(ds->ds_dir->dd_pool->dp_spa) < SPA_VERSION_USERREFS)
3480 		return (ENOTSUP);
3481 
3482 	error = dsl_dataset_release_might_destroy(ds, ra->htag, &might_destroy);
3483 	if (error)
3484 		return (error);
3485 
3486 	if (might_destroy) {
3487 		struct dsl_ds_destroyarg dsda = {0};
3488 
3489 		if (dmu_tx_is_syncing(tx)) {
3490 			/*
3491 			 * If we're not prepared to remove the snapshot,
3492 			 * we can't allow the release to happen right now.
3493 			 */
3494 			if (!ra->own)
3495 				return (EBUSY);
3496 			if (ds->ds_objset) {
3497 				dmu_objset_evict(ds->ds_objset);
3498 				ds->ds_objset = NULL;
3499 			}
3500 		}
3501 		dsda.ds = ds;
3502 		dsda.releasing = B_TRUE;
3503 		return (dsl_dataset_destroy_check(&dsda, tag, tx));
3504 	}
3505 
3506 	return (0);
3507 }
3508 
3509 static void
3510 dsl_dataset_user_release_sync(void *arg1, void *tag, cred_t *cr, dmu_tx_t *tx)
3511 {
3512 	struct dsl_ds_releasearg *ra = arg1;
3513 	dsl_dataset_t *ds = ra->ds;
3514 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
3515 	objset_t *mos = dp->dp_meta_objset;
3516 	uint64_t zapobj;
3517 	uint64_t dsobj = ds->ds_object;
3518 	uint64_t refs;
3519 	int error;
3520 
3521 	mutex_enter(&ds->ds_lock);
3522 	ds->ds_userrefs--;
3523 	refs = ds->ds_userrefs;
3524 	mutex_exit(&ds->ds_lock);
3525 	error = dsl_pool_user_release(dp, ds->ds_object, ra->htag, tx);
3526 	VERIFY(error == 0 || error == ENOENT);
3527 	zapobj = ds->ds_phys->ds_userrefs_obj;
3528 	VERIFY(0 == zap_remove(mos, zapobj, ra->htag, tx));
3529 	if (ds->ds_userrefs == 0 && ds->ds_phys->ds_num_children == 1 &&
3530 	    DS_IS_DEFER_DESTROY(ds)) {
3531 		struct dsl_ds_destroyarg dsda = {0};
3532 
3533 		ASSERT(ra->own);
3534 		dsda.ds = ds;
3535 		dsda.releasing = B_TRUE;
3536 		/* We already did the destroy_check */
3537 		dsl_dataset_destroy_sync(&dsda, tag, cr, tx);
3538 	}
3539 
3540 	spa_history_internal_log(LOG_DS_USER_RELEASE,
3541 	    dp->dp_spa, tx, cr, "<%s> %lld dataset = %llu",
3542 	    ra->htag, (longlong_t)refs, dsobj);
3543 }
3544 
3545 static int
3546 dsl_dataset_user_release_one(const char *dsname, void *arg)
3547 {
3548 	struct dsl_ds_holdarg *ha = arg;
3549 	struct dsl_ds_releasearg *ra;
3550 	dsl_dataset_t *ds;
3551 	int error;
3552 	void *dtag = ha->dstg;
3553 	char *name;
3554 	boolean_t own = B_FALSE;
3555 	boolean_t might_destroy;
3556 
3557 	/* alloc a buffer to hold dsname@snapname, plus the terminating NULL */
3558 	name = kmem_asprintf("%s@%s", dsname, ha->snapname);
3559 	error = dsl_dataset_hold(name, dtag, &ds);
3560 	strfree(name);
3561 	if (error == ENOENT && ha->recursive)
3562 		return (0);
3563 	(void) strlcpy(ha->failed, dsname, sizeof (ha->failed));
3564 	if (error)
3565 		return (error);
3566 
3567 	ha->gotone = B_TRUE;
3568 
3569 	ASSERT(dsl_dataset_is_snapshot(ds));
3570 
3571 	error = dsl_dataset_release_might_destroy(ds, ha->htag, &might_destroy);
3572 	if (error) {
3573 		dsl_dataset_rele(ds, dtag);
3574 		return (error);
3575 	}
3576 
3577 	if (might_destroy) {
3578 #ifdef _KERNEL
3579 		error = zfs_unmount_snap(name, NULL);
3580 		if (error) {
3581 			dsl_dataset_rele(ds, dtag);
3582 			return (error);
3583 		}
3584 #endif
3585 		if (!dsl_dataset_tryown(ds, B_TRUE, dtag)) {
3586 			dsl_dataset_rele(ds, dtag);
3587 			return (EBUSY);
3588 		} else {
3589 			own = B_TRUE;
3590 			dsl_dataset_make_exclusive(ds, dtag);
3591 		}
3592 	}
3593 
3594 	ra = kmem_alloc(sizeof (struct dsl_ds_releasearg), KM_SLEEP);
3595 	ra->ds = ds;
3596 	ra->htag = ha->htag;
3597 	ra->own = own;
3598 	dsl_sync_task_create(ha->dstg, dsl_dataset_user_release_check,
3599 	    dsl_dataset_user_release_sync, ra, dtag, 0);
3600 
3601 	return (0);
3602 }
3603 
3604 int
3605 dsl_dataset_user_release(char *dsname, char *snapname, char *htag,
3606     boolean_t recursive)
3607 {
3608 	struct dsl_ds_holdarg *ha;
3609 	dsl_sync_task_t *dst;
3610 	spa_t *spa;
3611 	int error;
3612 
3613 top:
3614 	ha = kmem_zalloc(sizeof (struct dsl_ds_holdarg), KM_SLEEP);
3615 
3616 	(void) strlcpy(ha->failed, dsname, sizeof (ha->failed));
3617 
3618 	error = spa_open(dsname, &spa, FTAG);
3619 	if (error) {
3620 		kmem_free(ha, sizeof (struct dsl_ds_holdarg));
3621 		return (error);
3622 	}
3623 
3624 	ha->dstg = dsl_sync_task_group_create(spa_get_dsl(spa));
3625 	ha->htag = htag;
3626 	ha->snapname = snapname;
3627 	ha->recursive = recursive;
3628 	if (recursive) {
3629 		error = dmu_objset_find(dsname, dsl_dataset_user_release_one,
3630 		    ha, DS_FIND_CHILDREN);
3631 	} else {
3632 		error = dsl_dataset_user_release_one(dsname, ha);
3633 	}
3634 	if (error == 0)
3635 		error = dsl_sync_task_group_wait(ha->dstg);
3636 
3637 	for (dst = list_head(&ha->dstg->dstg_tasks); dst;
3638 	    dst = list_next(&ha->dstg->dstg_tasks, dst)) {
3639 		struct dsl_ds_releasearg *ra = dst->dst_arg1;
3640 		dsl_dataset_t *ds = ra->ds;
3641 
3642 		if (dst->dst_err)
3643 			dsl_dataset_name(ds, ha->failed);
3644 
3645 		if (ra->own)
3646 			dsl_dataset_disown(ds, ha->dstg);
3647 		else
3648 			dsl_dataset_rele(ds, ha->dstg);
3649 
3650 		kmem_free(ra, sizeof (struct dsl_ds_releasearg));
3651 	}
3652 
3653 	if (error == 0 && recursive && !ha->gotone)
3654 		error = ENOENT;
3655 
3656 	if (error && error != EBUSY)
3657 		(void) strlcpy(dsname, ha->failed, sizeof (ha->failed));
3658 
3659 	dsl_sync_task_group_destroy(ha->dstg);
3660 	kmem_free(ha, sizeof (struct dsl_ds_holdarg));
3661 	spa_close(spa, FTAG);
3662 
3663 	/*
3664 	 * We can get EBUSY if we were racing with deferred destroy and
3665 	 * dsl_dataset_user_release_check() hadn't done the necessary
3666 	 * open context setup.  We can also get EBUSY if we're racing
3667 	 * with destroy and that thread is the ds_owner.  Either way
3668 	 * the busy condition should be transient, and we should retry
3669 	 * the release operation.
3670 	 */
3671 	if (error == EBUSY)
3672 		goto top;
3673 
3674 	return (error);
3675 }
3676 
3677 /*
3678  * Called at spa_load time to release a stale temporary user hold.
3679  */
3680 int
3681 dsl_dataset_user_release_tmp(dsl_pool_t *dp, uint64_t dsobj, char *htag)
3682 {
3683 	dsl_dataset_t *ds;
3684 	char *snap;
3685 	char *name;
3686 	int namelen;
3687 	int error;
3688 
3689 	rw_enter(&dp->dp_config_rwlock, RW_READER);
3690 	error = dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds);
3691 	rw_exit(&dp->dp_config_rwlock);
3692 	if (error)
3693 		return (error);
3694 	namelen = dsl_dataset_namelen(ds)+1;
3695 	name = kmem_alloc(namelen, KM_SLEEP);
3696 	dsl_dataset_name(ds, name);
3697 	dsl_dataset_rele(ds, FTAG);
3698 
3699 	snap = strchr(name, '@');
3700 	*snap = '\0';
3701 	++snap;
3702 	return (dsl_dataset_user_release(name, snap, htag, B_FALSE));
3703 }
3704 
3705 int
3706 dsl_dataset_get_holds(const char *dsname, nvlist_t **nvp)
3707 {
3708 	dsl_dataset_t *ds;
3709 	int err;
3710 
3711 	err = dsl_dataset_hold(dsname, FTAG, &ds);
3712 	if (err)
3713 		return (err);
3714 
3715 	VERIFY(0 == nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP));
3716 	if (ds->ds_phys->ds_userrefs_obj != 0) {
3717 		zap_attribute_t *za;
3718 		zap_cursor_t zc;
3719 
3720 		za = kmem_alloc(sizeof (zap_attribute_t), KM_SLEEP);
3721 		for (zap_cursor_init(&zc, ds->ds_dir->dd_pool->dp_meta_objset,
3722 		    ds->ds_phys->ds_userrefs_obj);
3723 		    zap_cursor_retrieve(&zc, za) == 0;
3724 		    zap_cursor_advance(&zc)) {
3725 			VERIFY(0 == nvlist_add_uint64(*nvp, za->za_name,
3726 			    za->za_first_integer));
3727 		}
3728 		zap_cursor_fini(&zc);
3729 		kmem_free(za, sizeof (zap_attribute_t));
3730 	}
3731 	dsl_dataset_rele(ds, FTAG);
3732 	return (0);
3733 }
3734 
3735 /*
3736  * Note, this fuction is used as the callback for dmu_objset_find().  We
3737  * always return 0 so that we will continue to find and process
3738  * inconsistent datasets, even if we encounter an error trying to
3739  * process one of them.
3740  */
3741 /* ARGSUSED */
3742 int
3743 dsl_destroy_inconsistent(const char *dsname, void *arg)
3744 {
3745 	dsl_dataset_t *ds;
3746 
3747 	if (dsl_dataset_own(dsname, B_TRUE, FTAG, &ds) == 0) {
3748 		if (DS_IS_INCONSISTENT(ds))
3749 			(void) dsl_dataset_destroy(ds, FTAG, B_FALSE);
3750 		else
3751 			dsl_dataset_disown(ds, FTAG);
3752 	}
3753 	return (0);
3754 }
3755