xref: /netbsd-src/external/cddl/osnet/dist/uts/common/fs/zfs/dsl_pool.c (revision ba2539a9805a0544ff82c0003cc02fe1eee5603d)
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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23  * Copyright (c) 2011, 2016 by Delphix. All rights reserved.
24  * Copyright (c) 2013 Steven Hartland. All rights reserved.
25  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
26  * Copyright (c) 2014 Integros [integros.com]
27  * Copyright 2016 Nexenta Systems, Inc.  All rights reserved.
28  */
29 
30 #include <sys/dsl_pool.h>
31 #include <sys/dsl_dataset.h>
32 #include <sys/dsl_prop.h>
33 #include <sys/dsl_dir.h>
34 #include <sys/dsl_synctask.h>
35 #include <sys/dsl_scan.h>
36 #include <sys/dnode.h>
37 #include <sys/dmu_tx.h>
38 #include <sys/dmu_objset.h>
39 #include <sys/arc.h>
40 #include <sys/zap.h>
41 #include <sys/zio.h>
42 #include <sys/zfs_context.h>
43 #include <sys/fs/zfs.h>
44 #include <sys/zfs_znode.h>
45 #include <sys/spa_impl.h>
46 #include <sys/dsl_deadlist.h>
47 #include <sys/bptree.h>
48 #include <sys/zfeature.h>
49 #include <sys/zil_impl.h>
50 #include <sys/dsl_userhold.h>
51 
52 #if defined(__FreeBSD__) && defined(_KERNEL)
53 #include <sys/types.h>
54 #include <sys/sysctl.h>
55 #endif
56 
57 /*
58  * ZFS Write Throttle
59  * ------------------
60  *
61  * ZFS must limit the rate of incoming writes to the rate at which it is able
62  * to sync data modifications to the backend storage. Throttling by too much
63  * creates an artificial limit; throttling by too little can only be sustained
64  * for short periods and would lead to highly lumpy performance. On a per-pool
65  * basis, ZFS tracks the amount of modified (dirty) data. As operations change
66  * data, the amount of dirty data increases; as ZFS syncs out data, the amount
67  * of dirty data decreases. When the amount of dirty data exceeds a
68  * predetermined threshold further modifications are blocked until the amount
69  * of dirty data decreases (as data is synced out).
70  *
71  * The limit on dirty data is tunable, and should be adjusted according to
72  * both the IO capacity and available memory of the system. The larger the
73  * window, the more ZFS is able to aggregate and amortize metadata (and data)
74  * changes. However, memory is a limited resource, and allowing for more dirty
75  * data comes at the cost of keeping other useful data in memory (for example
76  * ZFS data cached by the ARC).
77  *
78  * Implementation
79  *
80  * As buffers are modified dsl_pool_willuse_space() increments both the per-
81  * txg (dp_dirty_pertxg[]) and poolwide (dp_dirty_total) accounting of
82  * dirty space used; dsl_pool_dirty_space() decrements those values as data
83  * is synced out from dsl_pool_sync(). While only the poolwide value is
84  * relevant, the per-txg value is useful for debugging. The tunable
85  * zfs_dirty_data_max determines the dirty space limit. Once that value is
86  * exceeded, new writes are halted until space frees up.
87  *
88  * The zfs_dirty_data_sync tunable dictates the threshold at which we
89  * ensure that there is a txg syncing (see the comment in txg.c for a full
90  * description of transaction group stages).
91  *
92  * The IO scheduler uses both the dirty space limit and current amount of
93  * dirty data as inputs. Those values affect the number of concurrent IOs ZFS
94  * issues. See the comment in vdev_queue.c for details of the IO scheduler.
95  *
96  * The delay is also calculated based on the amount of dirty data.  See the
97  * comment above dmu_tx_delay() for details.
98  */
99 
100 /*
101  * zfs_dirty_data_max will be set to zfs_dirty_data_max_percent% of all memory,
102  * capped at zfs_dirty_data_max_max.  It can also be overridden in /etc/system.
103  */
104 uint64_t zfs_dirty_data_max;
105 uint64_t zfs_dirty_data_max_max = 4ULL * 1024 * 1024 * 1024;
106 int zfs_dirty_data_max_percent = 10;
107 
108 /*
109  * If there is at least this much dirty data, push out a txg.
110  */
111 uint64_t zfs_dirty_data_sync = 64 * 1024 * 1024;
112 
113 /*
114  * Once there is this amount of dirty data, the dmu_tx_delay() will kick in
115  * and delay each transaction.
116  * This value should be >= zfs_vdev_async_write_active_max_dirty_percent.
117  */
118 int zfs_delay_min_dirty_percent = 60;
119 
120 /*
121  * This controls how quickly the delay approaches infinity.
122  * Larger values cause it to delay more for a given amount of dirty data.
123  * Therefore larger values will cause there to be less dirty data for a
124  * given throughput.
125  *
126  * For the smoothest delay, this value should be about 1 billion divided
127  * by the maximum number of operations per second.  This will smoothly
128  * handle between 10x and 1/10th this number.
129  *
130  * Note: zfs_delay_scale * zfs_dirty_data_max must be < 2^64, due to the
131  * multiply in dmu_tx_delay().
132  */
133 uint64_t zfs_delay_scale = 1000 * 1000 * 1000 / 2000;
134 
135 
136 #if defined(__FreeBSD__) && defined(_KERNEL)
137 
138 extern int zfs_vdev_async_write_active_max_dirty_percent;
139 
140 SYSCTL_DECL(_vfs_zfs);
141 
142 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, dirty_data_max, CTLFLAG_RWTUN,
143     &zfs_dirty_data_max, 0,
144     "The maximum amount of dirty data in bytes after which new writes are "
145     "halted until space becomes available");
146 
147 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, dirty_data_max_max, CTLFLAG_RDTUN,
148     &zfs_dirty_data_max_max, 0,
149     "The absolute cap on dirty_data_max when auto calculating");
150 
151 static int sysctl_zfs_dirty_data_max_percent(SYSCTL_HANDLER_ARGS);
152 SYSCTL_PROC(_vfs_zfs, OID_AUTO, dirty_data_max_percent,
153     CTLTYPE_INT | CTLFLAG_MPSAFE | CTLFLAG_RWTUN, 0, sizeof(int),
154     sysctl_zfs_dirty_data_max_percent, "I",
155     "The percent of physical memory used to auto calculate dirty_data_max");
156 
157 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, dirty_data_sync, CTLFLAG_RWTUN,
158     &zfs_dirty_data_sync, 0,
159     "Force a txg if the number of dirty buffer bytes exceed this value");
160 
161 static int sysctl_zfs_delay_min_dirty_percent(SYSCTL_HANDLER_ARGS);
162 /* No zfs_delay_min_dirty_percent tunable due to limit requirements */
163 SYSCTL_PROC(_vfs_zfs, OID_AUTO, delay_min_dirty_percent,
164     CTLTYPE_INT | CTLFLAG_MPSAFE | CTLFLAG_RW, 0, sizeof(int),
165     sysctl_zfs_delay_min_dirty_percent, "I",
166     "The limit of outstanding dirty data before transations are delayed");
167 
168 static int sysctl_zfs_delay_scale(SYSCTL_HANDLER_ARGS);
169 /* No zfs_delay_scale tunable due to limit requirements */
170 SYSCTL_PROC(_vfs_zfs, OID_AUTO, delay_scale,
171     CTLTYPE_U64 | CTLFLAG_MPSAFE | CTLFLAG_RW, 0, sizeof(uint64_t),
172     sysctl_zfs_delay_scale, "QU",
173     "Controls how quickly the delay approaches infinity");
174 
175 static int
sysctl_zfs_dirty_data_max_percent(SYSCTL_HANDLER_ARGS)176 sysctl_zfs_dirty_data_max_percent(SYSCTL_HANDLER_ARGS)
177 {
178 	int val, err;
179 
180 	val = zfs_dirty_data_max_percent;
181 	err = sysctl_handle_int(oidp, &val, 0, req);
182 	if (err != 0 || req->newptr == NULL)
183 		return (err);
184 
185 	if (val < 0 || val > 100)
186 		return (EINVAL);
187 
188 	zfs_dirty_data_max_percent = val;
189 
190 	return (0);
191 }
192 
193 static int
sysctl_zfs_delay_min_dirty_percent(SYSCTL_HANDLER_ARGS)194 sysctl_zfs_delay_min_dirty_percent(SYSCTL_HANDLER_ARGS)
195 {
196 	int val, err;
197 
198 	val = zfs_delay_min_dirty_percent;
199 	err = sysctl_handle_int(oidp, &val, 0, req);
200 	if (err != 0 || req->newptr == NULL)
201 		return (err);
202 
203 	if (val < zfs_vdev_async_write_active_max_dirty_percent)
204 		return (EINVAL);
205 
206 	zfs_delay_min_dirty_percent = val;
207 
208 	return (0);
209 }
210 
211 static int
sysctl_zfs_delay_scale(SYSCTL_HANDLER_ARGS)212 sysctl_zfs_delay_scale(SYSCTL_HANDLER_ARGS)
213 {
214 	uint64_t val;
215 	int err;
216 
217 	val = zfs_delay_scale;
218 	err = sysctl_handle_64(oidp, &val, 0, req);
219 	if (err != 0 || req->newptr == NULL)
220 		return (err);
221 
222 	if (val > UINT64_MAX / zfs_dirty_data_max)
223 		return (EINVAL);
224 
225 	zfs_delay_scale = val;
226 
227 	return (0);
228 }
229 #endif
230 
231 hrtime_t zfs_throttle_delay = MSEC2NSEC(10);
232 hrtime_t zfs_throttle_resolution = MSEC2NSEC(10);
233 
234 int
dsl_pool_open_special_dir(dsl_pool_t * dp,const char * name,dsl_dir_t ** ddp)235 dsl_pool_open_special_dir(dsl_pool_t *dp, const char *name, dsl_dir_t **ddp)
236 {
237 	uint64_t obj;
238 	int err;
239 
240 	err = zap_lookup(dp->dp_meta_objset,
241 	    dsl_dir_phys(dp->dp_root_dir)->dd_child_dir_zapobj,
242 	    name, sizeof (obj), 1, &obj);
243 	if (err)
244 		return (err);
245 
246 	return (dsl_dir_hold_obj(dp, obj, name, dp, ddp));
247 }
248 
249 static dsl_pool_t *
dsl_pool_open_impl(spa_t * spa,uint64_t txg)250 dsl_pool_open_impl(spa_t *spa, uint64_t txg)
251 {
252 	dsl_pool_t *dp;
253 	blkptr_t *bp = spa_get_rootblkptr(spa);
254 
255 	dp = kmem_zalloc(sizeof (dsl_pool_t), KM_SLEEP);
256 	dp->dp_spa = spa;
257 	dp->dp_meta_rootbp = *bp;
258 	rrw_init(&dp->dp_config_rwlock, B_TRUE);
259 	txg_init(dp, txg);
260 
261 	txg_list_create(&dp->dp_dirty_datasets,
262 	    offsetof(dsl_dataset_t, ds_dirty_link));
263 	txg_list_create(&dp->dp_dirty_zilogs,
264 	    offsetof(zilog_t, zl_dirty_link));
265 	txg_list_create(&dp->dp_dirty_dirs,
266 	    offsetof(dsl_dir_t, dd_dirty_link));
267 	txg_list_create(&dp->dp_sync_tasks,
268 	    offsetof(dsl_sync_task_t, dst_node));
269 
270 	mutex_init(&dp->dp_lock, NULL, MUTEX_DEFAULT, NULL);
271 	cv_init(&dp->dp_spaceavail_cv, NULL, CV_DEFAULT, NULL);
272 
273 	dp->dp_vnrele_taskq = taskq_create("zfs_vn_rele_taskq", 1, minclsyspri,
274 	    1, 4, 0);
275 
276 	return (dp);
277 }
278 
279 int
dsl_pool_init(spa_t * spa,uint64_t txg,dsl_pool_t ** dpp)280 dsl_pool_init(spa_t *spa, uint64_t txg, dsl_pool_t **dpp)
281 {
282 	int err;
283 	dsl_pool_t *dp = dsl_pool_open_impl(spa, txg);
284 
285 	err = dmu_objset_open_impl(spa, NULL, &dp->dp_meta_rootbp,
286 	    &dp->dp_meta_objset);
287 	if (err != 0)
288 		dsl_pool_close(dp);
289 	else
290 		*dpp = dp;
291 
292 	return (err);
293 }
294 
295 int
dsl_pool_open(dsl_pool_t * dp)296 dsl_pool_open(dsl_pool_t *dp)
297 {
298 	int err;
299 	dsl_dir_t *dd;
300 	dsl_dataset_t *ds;
301 	uint64_t obj;
302 
303 	rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG);
304 	err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
305 	    DMU_POOL_ROOT_DATASET, sizeof (uint64_t), 1,
306 	    &dp->dp_root_dir_obj);
307 	if (err)
308 		goto out;
309 
310 	err = dsl_dir_hold_obj(dp, dp->dp_root_dir_obj,
311 	    NULL, dp, &dp->dp_root_dir);
312 	if (err)
313 		goto out;
314 
315 	err = dsl_pool_open_special_dir(dp, MOS_DIR_NAME, &dp->dp_mos_dir);
316 	if (err)
317 		goto out;
318 
319 	if (spa_version(dp->dp_spa) >= SPA_VERSION_ORIGIN) {
320 		err = dsl_pool_open_special_dir(dp, ORIGIN_DIR_NAME, &dd);
321 		if (err)
322 			goto out;
323 		err = dsl_dataset_hold_obj(dp,
324 		    dsl_dir_phys(dd)->dd_head_dataset_obj, FTAG, &ds);
325 		if (err == 0) {
326 			err = dsl_dataset_hold_obj(dp,
327 			    dsl_dataset_phys(ds)->ds_prev_snap_obj, dp,
328 			    &dp->dp_origin_snap);
329 			dsl_dataset_rele(ds, FTAG);
330 		}
331 		dsl_dir_rele(dd, dp);
332 		if (err)
333 			goto out;
334 	}
335 
336 	if (spa_version(dp->dp_spa) >= SPA_VERSION_DEADLISTS) {
337 		err = dsl_pool_open_special_dir(dp, FREE_DIR_NAME,
338 		    &dp->dp_free_dir);
339 		if (err)
340 			goto out;
341 
342 		err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
343 		    DMU_POOL_FREE_BPOBJ, sizeof (uint64_t), 1, &obj);
344 		if (err)
345 			goto out;
346 		VERIFY0(bpobj_open(&dp->dp_free_bpobj,
347 		    dp->dp_meta_objset, obj));
348 	}
349 
350 	/*
351 	 * Note: errors ignored, because the leak dir will not exist if we
352 	 * have not encountered a leak yet.
353 	 */
354 	(void) dsl_pool_open_special_dir(dp, LEAK_DIR_NAME,
355 	    &dp->dp_leak_dir);
356 
357 	if (spa_feature_is_active(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY)) {
358 		err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
359 		    DMU_POOL_BPTREE_OBJ, sizeof (uint64_t), 1,
360 		    &dp->dp_bptree_obj);
361 		if (err != 0)
362 			goto out;
363 	}
364 
365 	if (spa_feature_is_active(dp->dp_spa, SPA_FEATURE_EMPTY_BPOBJ)) {
366 		err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
367 		    DMU_POOL_EMPTY_BPOBJ, sizeof (uint64_t), 1,
368 		    &dp->dp_empty_bpobj);
369 		if (err != 0)
370 			goto out;
371 	}
372 
373 	err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
374 	    DMU_POOL_TMP_USERREFS, sizeof (uint64_t), 1,
375 	    &dp->dp_tmp_userrefs_obj);
376 	if (err == ENOENT)
377 		err = 0;
378 	if (err)
379 		goto out;
380 
381 	err = dsl_scan_init(dp, dp->dp_tx.tx_open_txg);
382 
383 out:
384 	rrw_exit(&dp->dp_config_rwlock, FTAG);
385 	return (err);
386 }
387 
388 void
dsl_pool_close(dsl_pool_t * dp)389 dsl_pool_close(dsl_pool_t *dp)
390 {
391 	/*
392 	 * Drop our references from dsl_pool_open().
393 	 *
394 	 * Since we held the origin_snap from "syncing" context (which
395 	 * includes pool-opening context), it actually only got a "ref"
396 	 * and not a hold, so just drop that here.
397 	 */
398 	if (dp->dp_origin_snap)
399 		dsl_dataset_rele(dp->dp_origin_snap, dp);
400 	if (dp->dp_mos_dir)
401 		dsl_dir_rele(dp->dp_mos_dir, dp);
402 	if (dp->dp_free_dir)
403 		dsl_dir_rele(dp->dp_free_dir, dp);
404 	if (dp->dp_leak_dir)
405 		dsl_dir_rele(dp->dp_leak_dir, dp);
406 	if (dp->dp_root_dir)
407 		dsl_dir_rele(dp->dp_root_dir, dp);
408 
409 	bpobj_close(&dp->dp_free_bpobj);
410 
411 	/* undo the dmu_objset_open_impl(mos) from dsl_pool_open() */
412 	if (dp->dp_meta_objset)
413 		dmu_objset_evict(dp->dp_meta_objset);
414 
415 	txg_list_destroy(&dp->dp_dirty_datasets);
416 	txg_list_destroy(&dp->dp_dirty_zilogs);
417 	txg_list_destroy(&dp->dp_sync_tasks);
418 	txg_list_destroy(&dp->dp_dirty_dirs);
419 
420 	/*
421 	 * We can't set retry to TRUE since we're explicitly specifying
422 	 * a spa to flush. This is good enough; any missed buffers for
423 	 * this spa won't cause trouble, and they'll eventually fall
424 	 * out of the ARC just like any other unused buffer.
425 	 */
426 	arc_flush(dp->dp_spa, FALSE);
427 
428 	txg_fini(dp);
429 	dsl_scan_fini(dp);
430 	dmu_buf_user_evict_wait();
431 
432 	rrw_destroy(&dp->dp_config_rwlock);
433 	mutex_destroy(&dp->dp_lock);
434 	cv_destroy(&dp->dp_spaceavail_cv);
435 	taskq_destroy(dp->dp_vnrele_taskq);
436 	if (dp->dp_blkstats)
437 		kmem_free(dp->dp_blkstats, sizeof (zfs_all_blkstats_t));
438 	kmem_free(dp, sizeof (dsl_pool_t));
439 }
440 
441 dsl_pool_t *
dsl_pool_create(spa_t * spa,nvlist_t * zplprops,uint64_t txg)442 dsl_pool_create(spa_t *spa, nvlist_t *zplprops, uint64_t txg)
443 {
444 	int err;
445 	dsl_pool_t *dp = dsl_pool_open_impl(spa, txg);
446 	dmu_tx_t *tx = dmu_tx_create_assigned(dp, txg);
447 	objset_t *os;
448 	dsl_dataset_t *ds;
449 	uint64_t obj;
450 
451 	rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG);
452 
453 	/* create and open the MOS (meta-objset) */
454 	dp->dp_meta_objset = dmu_objset_create_impl(spa,
455 	    NULL, &dp->dp_meta_rootbp, DMU_OST_META, tx);
456 
457 	/* create the pool directory */
458 	err = zap_create_claim(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
459 	    DMU_OT_OBJECT_DIRECTORY, DMU_OT_NONE, 0, tx);
460 	ASSERT0(err);
461 
462 	/* Initialize scan structures */
463 	VERIFY0(dsl_scan_init(dp, txg));
464 
465 	/* create and open the root dir */
466 	dp->dp_root_dir_obj = dsl_dir_create_sync(dp, NULL, NULL, tx);
467 	VERIFY0(dsl_dir_hold_obj(dp, dp->dp_root_dir_obj,
468 	    NULL, dp, &dp->dp_root_dir));
469 
470 	/* create and open the meta-objset dir */
471 	(void) dsl_dir_create_sync(dp, dp->dp_root_dir, MOS_DIR_NAME, tx);
472 	VERIFY0(dsl_pool_open_special_dir(dp,
473 	    MOS_DIR_NAME, &dp->dp_mos_dir));
474 
475 	if (spa_version(spa) >= SPA_VERSION_DEADLISTS) {
476 		/* create and open the free dir */
477 		(void) dsl_dir_create_sync(dp, dp->dp_root_dir,
478 		    FREE_DIR_NAME, tx);
479 		VERIFY0(dsl_pool_open_special_dir(dp,
480 		    FREE_DIR_NAME, &dp->dp_free_dir));
481 
482 		/* create and open the free_bplist */
483 		obj = bpobj_alloc(dp->dp_meta_objset, SPA_OLD_MAXBLOCKSIZE, tx);
484 		VERIFY(zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
485 		    DMU_POOL_FREE_BPOBJ, sizeof (uint64_t), 1, &obj, tx) == 0);
486 		VERIFY0(bpobj_open(&dp->dp_free_bpobj,
487 		    dp->dp_meta_objset, obj));
488 	}
489 
490 	if (spa_version(spa) >= SPA_VERSION_DSL_SCRUB)
491 		dsl_pool_create_origin(dp, tx);
492 
493 	/* create the root dataset */
494 	obj = dsl_dataset_create_sync_dd(dp->dp_root_dir, NULL, 0, tx);
495 
496 	/* create the root objset */
497 	VERIFY0(dsl_dataset_hold_obj(dp, obj, FTAG, &ds));
498 	rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
499 	os = dmu_objset_create_impl(dp->dp_spa, ds,
500 	    dsl_dataset_get_blkptr(ds), DMU_OST_ZFS, tx);
501 	rrw_exit(&ds->ds_bp_rwlock, FTAG);
502 #ifdef _KERNEL
503 	zfs_create_fs(os, kcred, zplprops, tx);
504 #endif
505 	dsl_dataset_rele(ds, FTAG);
506 
507 	dmu_tx_commit(tx);
508 
509 	rrw_exit(&dp->dp_config_rwlock, FTAG);
510 
511 	return (dp);
512 }
513 
514 /*
515  * Account for the meta-objset space in its placeholder dsl_dir.
516  */
517 void
dsl_pool_mos_diduse_space(dsl_pool_t * dp,int64_t used,int64_t comp,int64_t uncomp)518 dsl_pool_mos_diduse_space(dsl_pool_t *dp,
519     int64_t used, int64_t comp, int64_t uncomp)
520 {
521 	ASSERT3U(comp, ==, uncomp); /* it's all metadata */
522 	mutex_enter(&dp->dp_lock);
523 	dp->dp_mos_used_delta += used;
524 	dp->dp_mos_compressed_delta += comp;
525 	dp->dp_mos_uncompressed_delta += uncomp;
526 	mutex_exit(&dp->dp_lock);
527 }
528 
529 static void
dsl_pool_sync_mos(dsl_pool_t * dp,dmu_tx_t * tx)530 dsl_pool_sync_mos(dsl_pool_t *dp, dmu_tx_t *tx)
531 {
532 	zio_t *zio = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED);
533 	dmu_objset_sync(dp->dp_meta_objset, zio, tx);
534 	VERIFY0(zio_wait(zio));
535 	dprintf_bp(&dp->dp_meta_rootbp, "meta objset rootbp is %s", "");
536 	spa_set_rootblkptr(dp->dp_spa, &dp->dp_meta_rootbp);
537 }
538 
539 static void
dsl_pool_dirty_delta(dsl_pool_t * dp,int64_t delta)540 dsl_pool_dirty_delta(dsl_pool_t *dp, int64_t delta)
541 {
542 	ASSERT(MUTEX_HELD(&dp->dp_lock));
543 
544 	if (delta < 0)
545 		ASSERT3U(-delta, <=, dp->dp_dirty_total);
546 
547 	dp->dp_dirty_total += delta;
548 
549 	/*
550 	 * Note: we signal even when increasing dp_dirty_total.
551 	 * This ensures forward progress -- each thread wakes the next waiter.
552 	 */
553 	if (dp->dp_dirty_total <= zfs_dirty_data_max)
554 		cv_signal(&dp->dp_spaceavail_cv);
555 }
556 
557 void
dsl_pool_sync(dsl_pool_t * dp,uint64_t txg)558 dsl_pool_sync(dsl_pool_t *dp, uint64_t txg)
559 {
560 	zio_t *zio;
561 	dmu_tx_t *tx;
562 	dsl_dir_t *dd;
563 	dsl_dataset_t *ds;
564 	objset_t *mos = dp->dp_meta_objset;
565 	list_t synced_datasets;
566 
567 	list_create(&synced_datasets, sizeof (dsl_dataset_t),
568 	    offsetof(dsl_dataset_t, ds_synced_link));
569 
570 	tx = dmu_tx_create_assigned(dp, txg);
571 
572 	/*
573 	 * Write out all dirty blocks of dirty datasets.
574 	 */
575 	zio = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED);
576 	while ((ds = txg_list_remove(&dp->dp_dirty_datasets, txg)) != NULL) {
577 		/*
578 		 * We must not sync any non-MOS datasets twice, because
579 		 * we may have taken a snapshot of them.  However, we
580 		 * may sync newly-created datasets on pass 2.
581 		 */
582 		ASSERT(!list_link_active(&ds->ds_synced_link));
583 		list_insert_tail(&synced_datasets, ds);
584 		dsl_dataset_sync(ds, zio, tx);
585 	}
586 	VERIFY0(zio_wait(zio));
587 
588 	/*
589 	 * We have written all of the accounted dirty data, so our
590 	 * dp_space_towrite should now be zero.  However, some seldom-used
591 	 * code paths do not adhere to this (e.g. dbuf_undirty(), also
592 	 * rounding error in dbuf_write_physdone).
593 	 * Shore up the accounting of any dirtied space now.
594 	 */
595 	dsl_pool_undirty_space(dp, dp->dp_dirty_pertxg[txg & TXG_MASK], txg);
596 
597 	/*
598 	 * Update the long range free counter after
599 	 * we're done syncing user data
600 	 */
601 	mutex_enter(&dp->dp_lock);
602 	ASSERT(spa_sync_pass(dp->dp_spa) == 1 ||
603 	    dp->dp_long_free_dirty_pertxg[txg & TXG_MASK] == 0);
604 	dp->dp_long_free_dirty_pertxg[txg & TXG_MASK] = 0;
605 	mutex_exit(&dp->dp_lock);
606 
607 	/*
608 	 * After the data blocks have been written (ensured by the zio_wait()
609 	 * above), update the user/group space accounting.
610 	 */
611 	for (ds = list_head(&synced_datasets); ds != NULL;
612 	    ds = list_next(&synced_datasets, ds)) {
613 		dmu_objset_do_userquota_updates(ds->ds_objset, tx);
614 	}
615 
616 	/*
617 	 * Sync the datasets again to push out the changes due to
618 	 * userspace updates.  This must be done before we process the
619 	 * sync tasks, so that any snapshots will have the correct
620 	 * user accounting information (and we won't get confused
621 	 * about which blocks are part of the snapshot).
622 	 */
623 	zio = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED);
624 	while ((ds = txg_list_remove(&dp->dp_dirty_datasets, txg)) != NULL) {
625 		ASSERT(list_link_active(&ds->ds_synced_link));
626 		dmu_buf_rele(ds->ds_dbuf, ds);
627 		dsl_dataset_sync(ds, zio, tx);
628 	}
629 	VERIFY0(zio_wait(zio));
630 
631 	/*
632 	 * Now that the datasets have been completely synced, we can
633 	 * clean up our in-memory structures accumulated while syncing:
634 	 *
635 	 *  - move dead blocks from the pending deadlist to the on-disk deadlist
636 	 *  - release hold from dsl_dataset_dirty()
637 	 */
638 	while ((ds = list_remove_head(&synced_datasets)) != NULL) {
639 		dsl_dataset_sync_done(ds, tx);
640 	}
641 	while ((dd = txg_list_remove(&dp->dp_dirty_dirs, txg)) != NULL) {
642 		dsl_dir_sync(dd, tx);
643 	}
644 
645 	/*
646 	 * The MOS's space is accounted for in the pool/$MOS
647 	 * (dp_mos_dir).  We can't modify the mos while we're syncing
648 	 * it, so we remember the deltas and apply them here.
649 	 */
650 	if (dp->dp_mos_used_delta != 0 || dp->dp_mos_compressed_delta != 0 ||
651 	    dp->dp_mos_uncompressed_delta != 0) {
652 		dsl_dir_diduse_space(dp->dp_mos_dir, DD_USED_HEAD,
653 		    dp->dp_mos_used_delta,
654 		    dp->dp_mos_compressed_delta,
655 		    dp->dp_mos_uncompressed_delta, tx);
656 		dp->dp_mos_used_delta = 0;
657 		dp->dp_mos_compressed_delta = 0;
658 		dp->dp_mos_uncompressed_delta = 0;
659 	}
660 
661 	if (list_head(&mos->os_dirty_dnodes[txg & TXG_MASK]) != NULL ||
662 	    list_head(&mos->os_free_dnodes[txg & TXG_MASK]) != NULL) {
663 		dsl_pool_sync_mos(dp, tx);
664 	}
665 
666 	/*
667 	 * If we modify a dataset in the same txg that we want to destroy it,
668 	 * its dsl_dir's dd_dbuf will be dirty, and thus have a hold on it.
669 	 * dsl_dir_destroy_check() will fail if there are unexpected holds.
670 	 * Therefore, we want to sync the MOS (thus syncing the dd_dbuf
671 	 * and clearing the hold on it) before we process the sync_tasks.
672 	 * The MOS data dirtied by the sync_tasks will be synced on the next
673 	 * pass.
674 	 */
675 	if (!txg_list_empty(&dp->dp_sync_tasks, txg)) {
676 		dsl_sync_task_t *dst;
677 		/*
678 		 * No more sync tasks should have been added while we
679 		 * were syncing.
680 		 */
681 		ASSERT3U(spa_sync_pass(dp->dp_spa), ==, 1);
682 		while ((dst = txg_list_remove(&dp->dp_sync_tasks, txg)) != NULL)
683 			dsl_sync_task_sync(dst, tx);
684 	}
685 
686 	dmu_tx_commit(tx);
687 
688 	DTRACE_PROBE2(dsl_pool_sync__done, dsl_pool_t *dp, dp, uint64_t, txg);
689 }
690 
691 void
dsl_pool_sync_done(dsl_pool_t * dp,uint64_t txg)692 dsl_pool_sync_done(dsl_pool_t *dp, uint64_t txg)
693 {
694 	zilog_t *zilog;
695 
696 	while (zilog = txg_list_head(&dp->dp_dirty_zilogs, txg)) {
697 		dsl_dataset_t *ds = dmu_objset_ds(zilog->zl_os);
698 		/*
699 		 * We don't remove the zilog from the dp_dirty_zilogs
700 		 * list until after we've cleaned it. This ensures that
701 		 * callers of zilog_is_dirty() receive an accurate
702 		 * answer when they are racing with the spa sync thread.
703 		 */
704 		zil_clean(zilog, txg);
705 		(void) txg_list_remove_this(&dp->dp_dirty_zilogs, zilog, txg);
706 		ASSERT(!dmu_objset_is_dirty(zilog->zl_os, txg));
707 		dmu_buf_rele(ds->ds_dbuf, zilog);
708 	}
709 	ASSERT(!dmu_objset_is_dirty(dp->dp_meta_objset, txg));
710 }
711 
712 /*
713  * TRUE if the current thread is the tx_sync_thread or if we
714  * are being called from SPA context during pool initialization.
715  */
716 int
dsl_pool_sync_context(dsl_pool_t * dp)717 dsl_pool_sync_context(dsl_pool_t *dp)
718 {
719 	return (curthread == dp->dp_tx.tx_sync_thread ||
720 	    spa_is_initializing(dp->dp_spa));
721 }
722 
723 uint64_t
dsl_pool_adjustedsize(dsl_pool_t * dp,boolean_t netfree)724 dsl_pool_adjustedsize(dsl_pool_t *dp, boolean_t netfree)
725 {
726 	uint64_t space, resv;
727 
728 	/*
729 	 * If we're trying to assess whether it's OK to do a free,
730 	 * cut the reservation in half to allow forward progress
731 	 * (e.g. make it possible to rm(1) files from a full pool).
732 	 */
733 	space = spa_get_dspace(dp->dp_spa);
734 	resv = spa_get_slop_space(dp->dp_spa);
735 	if (netfree)
736 		resv >>= 1;
737 
738 	return (space - resv);
739 }
740 
741 boolean_t
dsl_pool_need_dirty_delay(dsl_pool_t * dp)742 dsl_pool_need_dirty_delay(dsl_pool_t *dp)
743 {
744 	uint64_t delay_min_bytes =
745 	    zfs_dirty_data_max * zfs_delay_min_dirty_percent / 100;
746 	boolean_t rv;
747 
748 	mutex_enter(&dp->dp_lock);
749 	if (dp->dp_dirty_total > zfs_dirty_data_sync)
750 		txg_kick(dp);
751 	rv = (dp->dp_dirty_total > delay_min_bytes);
752 	mutex_exit(&dp->dp_lock);
753 	return (rv);
754 }
755 
756 void
dsl_pool_dirty_space(dsl_pool_t * dp,int64_t space,dmu_tx_t * tx)757 dsl_pool_dirty_space(dsl_pool_t *dp, int64_t space, dmu_tx_t *tx)
758 {
759 	if (space > 0) {
760 		mutex_enter(&dp->dp_lock);
761 		dp->dp_dirty_pertxg[tx->tx_txg & TXG_MASK] += space;
762 		dsl_pool_dirty_delta(dp, space);
763 		mutex_exit(&dp->dp_lock);
764 	}
765 }
766 
767 void
dsl_pool_undirty_space(dsl_pool_t * dp,int64_t space,uint64_t txg)768 dsl_pool_undirty_space(dsl_pool_t *dp, int64_t space, uint64_t txg)
769 {
770 	ASSERT3S(space, >=, 0);
771 	if (space == 0)
772 		return;
773 	mutex_enter(&dp->dp_lock);
774 	if (dp->dp_dirty_pertxg[txg & TXG_MASK] < space) {
775 		/* XXX writing something we didn't dirty? */
776 		space = dp->dp_dirty_pertxg[txg & TXG_MASK];
777 	}
778 	ASSERT3U(dp->dp_dirty_pertxg[txg & TXG_MASK], >=, space);
779 	dp->dp_dirty_pertxg[txg & TXG_MASK] -= space;
780 	ASSERT3U(dp->dp_dirty_total, >=, space);
781 	dsl_pool_dirty_delta(dp, -space);
782 	mutex_exit(&dp->dp_lock);
783 }
784 
785 /* ARGSUSED */
786 static int
upgrade_clones_cb(dsl_pool_t * dp,dsl_dataset_t * hds,void * arg)787 upgrade_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg)
788 {
789 	dmu_tx_t *tx = arg;
790 	dsl_dataset_t *ds, *prev = NULL;
791 	int err;
792 
793 	err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds);
794 	if (err)
795 		return (err);
796 
797 	while (dsl_dataset_phys(ds)->ds_prev_snap_obj != 0) {
798 		err = dsl_dataset_hold_obj(dp,
799 		    dsl_dataset_phys(ds)->ds_prev_snap_obj, FTAG, &prev);
800 		if (err) {
801 			dsl_dataset_rele(ds, FTAG);
802 			return (err);
803 		}
804 
805 		if (dsl_dataset_phys(prev)->ds_next_snap_obj != ds->ds_object)
806 			break;
807 		dsl_dataset_rele(ds, FTAG);
808 		ds = prev;
809 		prev = NULL;
810 	}
811 
812 	if (prev == NULL) {
813 		prev = dp->dp_origin_snap;
814 
815 		/*
816 		 * The $ORIGIN can't have any data, or the accounting
817 		 * will be wrong.
818 		 */
819 		rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
820 		ASSERT0(dsl_dataset_phys(prev)->ds_bp.blk_birth);
821 		rrw_exit(&ds->ds_bp_rwlock, FTAG);
822 
823 		/* The origin doesn't get attached to itself */
824 		if (ds->ds_object == prev->ds_object) {
825 			dsl_dataset_rele(ds, FTAG);
826 			return (0);
827 		}
828 
829 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
830 		dsl_dataset_phys(ds)->ds_prev_snap_obj = prev->ds_object;
831 		dsl_dataset_phys(ds)->ds_prev_snap_txg =
832 		    dsl_dataset_phys(prev)->ds_creation_txg;
833 
834 		dmu_buf_will_dirty(ds->ds_dir->dd_dbuf, tx);
835 		dsl_dir_phys(ds->ds_dir)->dd_origin_obj = prev->ds_object;
836 
837 		dmu_buf_will_dirty(prev->ds_dbuf, tx);
838 		dsl_dataset_phys(prev)->ds_num_children++;
839 
840 		if (dsl_dataset_phys(ds)->ds_next_snap_obj == 0) {
841 			ASSERT(ds->ds_prev == NULL);
842 			VERIFY0(dsl_dataset_hold_obj(dp,
843 			    dsl_dataset_phys(ds)->ds_prev_snap_obj,
844 			    ds, &ds->ds_prev));
845 		}
846 	}
847 
848 	ASSERT3U(dsl_dir_phys(ds->ds_dir)->dd_origin_obj, ==, prev->ds_object);
849 	ASSERT3U(dsl_dataset_phys(ds)->ds_prev_snap_obj, ==, prev->ds_object);
850 
851 	if (dsl_dataset_phys(prev)->ds_next_clones_obj == 0) {
852 		dmu_buf_will_dirty(prev->ds_dbuf, tx);
853 		dsl_dataset_phys(prev)->ds_next_clones_obj =
854 		    zap_create(dp->dp_meta_objset,
855 		    DMU_OT_NEXT_CLONES, DMU_OT_NONE, 0, tx);
856 	}
857 	VERIFY0(zap_add_int(dp->dp_meta_objset,
858 	    dsl_dataset_phys(prev)->ds_next_clones_obj, ds->ds_object, tx));
859 
860 	dsl_dataset_rele(ds, FTAG);
861 	if (prev != dp->dp_origin_snap)
862 		dsl_dataset_rele(prev, FTAG);
863 	return (0);
864 }
865 
866 void
dsl_pool_upgrade_clones(dsl_pool_t * dp,dmu_tx_t * tx)867 dsl_pool_upgrade_clones(dsl_pool_t *dp, dmu_tx_t *tx)
868 {
869 	ASSERT(dmu_tx_is_syncing(tx));
870 	ASSERT(dp->dp_origin_snap != NULL);
871 
872 	VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, upgrade_clones_cb,
873 	    tx, DS_FIND_CHILDREN | DS_FIND_SERIALIZE));
874 }
875 
876 /* ARGSUSED */
877 static int
upgrade_dir_clones_cb(dsl_pool_t * dp,dsl_dataset_t * ds,void * arg)878 upgrade_dir_clones_cb(dsl_pool_t *dp, dsl_dataset_t *ds, void *arg)
879 {
880 	dmu_tx_t *tx = arg;
881 	objset_t *mos = dp->dp_meta_objset;
882 
883 	if (dsl_dir_phys(ds->ds_dir)->dd_origin_obj != 0) {
884 		dsl_dataset_t *origin;
885 
886 		VERIFY0(dsl_dataset_hold_obj(dp,
887 		    dsl_dir_phys(ds->ds_dir)->dd_origin_obj, FTAG, &origin));
888 
889 		if (dsl_dir_phys(origin->ds_dir)->dd_clones == 0) {
890 			dmu_buf_will_dirty(origin->ds_dir->dd_dbuf, tx);
891 			dsl_dir_phys(origin->ds_dir)->dd_clones =
892 			    zap_create(mos, DMU_OT_DSL_CLONES, DMU_OT_NONE,
893 			    0, tx);
894 		}
895 
896 		VERIFY0(zap_add_int(dp->dp_meta_objset,
897 		    dsl_dir_phys(origin->ds_dir)->dd_clones,
898 		    ds->ds_object, tx));
899 
900 		dsl_dataset_rele(origin, FTAG);
901 	}
902 	return (0);
903 }
904 
905 void
dsl_pool_upgrade_dir_clones(dsl_pool_t * dp,dmu_tx_t * tx)906 dsl_pool_upgrade_dir_clones(dsl_pool_t *dp, dmu_tx_t *tx)
907 {
908 	ASSERT(dmu_tx_is_syncing(tx));
909 	uint64_t obj;
910 
911 	(void) dsl_dir_create_sync(dp, dp->dp_root_dir, FREE_DIR_NAME, tx);
912 	VERIFY0(dsl_pool_open_special_dir(dp,
913 	    FREE_DIR_NAME, &dp->dp_free_dir));
914 
915 	/*
916 	 * We can't use bpobj_alloc(), because spa_version() still
917 	 * returns the old version, and we need a new-version bpobj with
918 	 * subobj support.  So call dmu_object_alloc() directly.
919 	 */
920 	obj = dmu_object_alloc(dp->dp_meta_objset, DMU_OT_BPOBJ,
921 	    SPA_OLD_MAXBLOCKSIZE, DMU_OT_BPOBJ_HDR, sizeof (bpobj_phys_t), tx);
922 	VERIFY0(zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
923 	    DMU_POOL_FREE_BPOBJ, sizeof (uint64_t), 1, &obj, tx));
924 	VERIFY0(bpobj_open(&dp->dp_free_bpobj, dp->dp_meta_objset, obj));
925 
926 	VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj,
927 	    upgrade_dir_clones_cb, tx, DS_FIND_CHILDREN | DS_FIND_SERIALIZE));
928 }
929 
930 void
dsl_pool_create_origin(dsl_pool_t * dp,dmu_tx_t * tx)931 dsl_pool_create_origin(dsl_pool_t *dp, dmu_tx_t *tx)
932 {
933 	uint64_t dsobj;
934 	dsl_dataset_t *ds;
935 
936 	ASSERT(dmu_tx_is_syncing(tx));
937 	ASSERT(dp->dp_origin_snap == NULL);
938 	ASSERT(rrw_held(&dp->dp_config_rwlock, RW_WRITER));
939 
940 	/* create the origin dir, ds, & snap-ds */
941 	dsobj = dsl_dataset_create_sync(dp->dp_root_dir, ORIGIN_DIR_NAME,
942 	    NULL, 0, kcred, tx);
943 	VERIFY0(dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds));
944 	dsl_dataset_snapshot_sync_impl(ds, ORIGIN_DIR_NAME, tx);
945 	VERIFY0(dsl_dataset_hold_obj(dp, dsl_dataset_phys(ds)->ds_prev_snap_obj,
946 	    dp, &dp->dp_origin_snap));
947 	dsl_dataset_rele(ds, FTAG);
948 }
949 
950 taskq_t *
dsl_pool_vnrele_taskq(dsl_pool_t * dp)951 dsl_pool_vnrele_taskq(dsl_pool_t *dp)
952 {
953 	return (dp->dp_vnrele_taskq);
954 }
955 
956 /*
957  * Walk through the pool-wide zap object of temporary snapshot user holds
958  * and release them.
959  */
960 void
dsl_pool_clean_tmp_userrefs(dsl_pool_t * dp)961 dsl_pool_clean_tmp_userrefs(dsl_pool_t *dp)
962 {
963 	zap_attribute_t za;
964 	zap_cursor_t zc;
965 	objset_t *mos = dp->dp_meta_objset;
966 	uint64_t zapobj = dp->dp_tmp_userrefs_obj;
967 	nvlist_t *holds;
968 
969 	if (zapobj == 0)
970 		return;
971 	ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS);
972 
973 	holds = fnvlist_alloc();
974 
975 	for (zap_cursor_init(&zc, mos, zapobj);
976 	    zap_cursor_retrieve(&zc, &za) == 0;
977 	    zap_cursor_advance(&zc)) {
978 		char *htag;
979 		nvlist_t *tags;
980 
981 		htag = strchr(za.za_name, '-');
982 		*htag = '\0';
983 		++htag;
984 		if (nvlist_lookup_nvlist(holds, za.za_name, &tags) != 0) {
985 			tags = fnvlist_alloc();
986 			fnvlist_add_boolean(tags, htag);
987 			fnvlist_add_nvlist(holds, za.za_name, tags);
988 			fnvlist_free(tags);
989 		} else {
990 			fnvlist_add_boolean(tags, htag);
991 		}
992 	}
993 	dsl_dataset_user_release_tmp(dp, holds);
994 	fnvlist_free(holds);
995 	zap_cursor_fini(&zc);
996 }
997 
998 /*
999  * Create the pool-wide zap object for storing temporary snapshot holds.
1000  */
1001 void
dsl_pool_user_hold_create_obj(dsl_pool_t * dp,dmu_tx_t * tx)1002 dsl_pool_user_hold_create_obj(dsl_pool_t *dp, dmu_tx_t *tx)
1003 {
1004 	objset_t *mos = dp->dp_meta_objset;
1005 
1006 	ASSERT(dp->dp_tmp_userrefs_obj == 0);
1007 	ASSERT(dmu_tx_is_syncing(tx));
1008 
1009 	dp->dp_tmp_userrefs_obj = zap_create_link(mos, DMU_OT_USERREFS,
1010 	    DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_TMP_USERREFS, tx);
1011 }
1012 
1013 static int
dsl_pool_user_hold_rele_impl(dsl_pool_t * dp,uint64_t dsobj,const char * tag,uint64_t now,dmu_tx_t * tx,boolean_t holding)1014 dsl_pool_user_hold_rele_impl(dsl_pool_t *dp, uint64_t dsobj,
1015     const char *tag, uint64_t now, dmu_tx_t *tx, boolean_t holding)
1016 {
1017 	objset_t *mos = dp->dp_meta_objset;
1018 	uint64_t zapobj = dp->dp_tmp_userrefs_obj;
1019 	char *name;
1020 	int error;
1021 
1022 	ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS);
1023 	ASSERT(dmu_tx_is_syncing(tx));
1024 
1025 	/*
1026 	 * If the pool was created prior to SPA_VERSION_USERREFS, the
1027 	 * zap object for temporary holds might not exist yet.
1028 	 */
1029 	if (zapobj == 0) {
1030 		if (holding) {
1031 			dsl_pool_user_hold_create_obj(dp, tx);
1032 			zapobj = dp->dp_tmp_userrefs_obj;
1033 		} else {
1034 			return (SET_ERROR(ENOENT));
1035 		}
1036 	}
1037 
1038 	name = kmem_asprintf("%llx-%s", (u_longlong_t)dsobj, tag);
1039 	if (holding)
1040 		error = zap_add(mos, zapobj, name, 8, 1, &now, tx);
1041 	else
1042 		error = zap_remove(mos, zapobj, name, tx);
1043 	strfree(name);
1044 
1045 	return (error);
1046 }
1047 
1048 /*
1049  * Add a temporary hold for the given dataset object and tag.
1050  */
1051 int
dsl_pool_user_hold(dsl_pool_t * dp,uint64_t dsobj,const char * tag,uint64_t now,dmu_tx_t * tx)1052 dsl_pool_user_hold(dsl_pool_t *dp, uint64_t dsobj, const char *tag,
1053     uint64_t now, dmu_tx_t *tx)
1054 {
1055 	return (dsl_pool_user_hold_rele_impl(dp, dsobj, tag, now, tx, B_TRUE));
1056 }
1057 
1058 /*
1059  * Release a temporary hold for the given dataset object and tag.
1060  */
1061 int
dsl_pool_user_release(dsl_pool_t * dp,uint64_t dsobj,const char * tag,dmu_tx_t * tx)1062 dsl_pool_user_release(dsl_pool_t *dp, uint64_t dsobj, const char *tag,
1063     dmu_tx_t *tx)
1064 {
1065 	return (dsl_pool_user_hold_rele_impl(dp, dsobj, tag, 0,
1066 	    tx, B_FALSE));
1067 }
1068 
1069 /*
1070  * DSL Pool Configuration Lock
1071  *
1072  * The dp_config_rwlock protects against changes to DSL state (e.g. dataset
1073  * creation / destruction / rename / property setting).  It must be held for
1074  * read to hold a dataset or dsl_dir.  I.e. you must call
1075  * dsl_pool_config_enter() or dsl_pool_hold() before calling
1076  * dsl_{dataset,dir}_hold{_obj}.  In most circumstances, the dp_config_rwlock
1077  * must be held continuously until all datasets and dsl_dirs are released.
1078  *
1079  * The only exception to this rule is that if a "long hold" is placed on
1080  * a dataset, then the dp_config_rwlock may be dropped while the dataset
1081  * is still held.  The long hold will prevent the dataset from being
1082  * destroyed -- the destroy will fail with EBUSY.  A long hold can be
1083  * obtained by calling dsl_dataset_long_hold(), or by "owning" a dataset
1084  * (by calling dsl_{dataset,objset}_{try}own{_obj}).
1085  *
1086  * Legitimate long-holders (including owners) should be long-running, cancelable
1087  * tasks that should cause "zfs destroy" to fail.  This includes DMU
1088  * consumers (i.e. a ZPL filesystem being mounted or ZVOL being open),
1089  * "zfs send", and "zfs diff".  There are several other long-holders whose
1090  * uses are suboptimal (e.g. "zfs promote", and zil_suspend()).
1091  *
1092  * The usual formula for long-holding would be:
1093  * dsl_pool_hold()
1094  * dsl_dataset_hold()
1095  * ... perform checks ...
1096  * dsl_dataset_long_hold()
1097  * dsl_pool_rele()
1098  * ... perform long-running task ...
1099  * dsl_dataset_long_rele()
1100  * dsl_dataset_rele()
1101  *
1102  * Note that when the long hold is released, the dataset is still held but
1103  * the pool is not held.  The dataset may change arbitrarily during this time
1104  * (e.g. it could be destroyed).  Therefore you shouldn't do anything to the
1105  * dataset except release it.
1106  *
1107  * User-initiated operations (e.g. ioctls, zfs_ioc_*()) are either read-only
1108  * or modifying operations.
1109  *
1110  * Modifying operations should generally use dsl_sync_task().  The synctask
1111  * infrastructure enforces proper locking strategy with respect to the
1112  * dp_config_rwlock.  See the comment above dsl_sync_task() for details.
1113  *
1114  * Read-only operations will manually hold the pool, then the dataset, obtain
1115  * information from the dataset, then release the pool and dataset.
1116  * dmu_objset_{hold,rele}() are convenience routines that also do the pool
1117  * hold/rele.
1118  */
1119 
1120 int
dsl_pool_hold(const char * name,void * tag,dsl_pool_t ** dp)1121 dsl_pool_hold(const char *name, void *tag, dsl_pool_t **dp)
1122 {
1123 	spa_t *spa;
1124 	int error;
1125 
1126 	error = spa_open(name, &spa, tag);
1127 	if (error == 0) {
1128 		*dp = spa_get_dsl(spa);
1129 		dsl_pool_config_enter(*dp, tag);
1130 	}
1131 	return (error);
1132 }
1133 
1134 void
dsl_pool_rele(dsl_pool_t * dp,void * tag)1135 dsl_pool_rele(dsl_pool_t *dp, void *tag)
1136 {
1137 	dsl_pool_config_exit(dp, tag);
1138 	spa_close(dp->dp_spa, tag);
1139 }
1140 
1141 void
dsl_pool_config_enter(dsl_pool_t * dp,void * tag)1142 dsl_pool_config_enter(dsl_pool_t *dp, void *tag)
1143 {
1144 	/*
1145 	 * We use a "reentrant" reader-writer lock, but not reentrantly.
1146 	 *
1147 	 * The rrwlock can (with the track_all flag) track all reading threads,
1148 	 * which is very useful for debugging which code path failed to release
1149 	 * the lock, and for verifying that the *current* thread does hold
1150 	 * the lock.
1151 	 *
1152 	 * (Unlike a rwlock, which knows that N threads hold it for
1153 	 * read, but not *which* threads, so rw_held(RW_READER) returns TRUE
1154 	 * if any thread holds it for read, even if this thread doesn't).
1155 	 */
1156 	ASSERT(!rrw_held(&dp->dp_config_rwlock, RW_READER));
1157 	rrw_enter(&dp->dp_config_rwlock, RW_READER, tag);
1158 }
1159 
1160 void
dsl_pool_config_enter_prio(dsl_pool_t * dp,void * tag)1161 dsl_pool_config_enter_prio(dsl_pool_t *dp, void *tag)
1162 {
1163 	ASSERT(!rrw_held(&dp->dp_config_rwlock, RW_READER));
1164 	rrw_enter_read_prio(&dp->dp_config_rwlock, tag);
1165 }
1166 
1167 void
dsl_pool_config_exit(dsl_pool_t * dp,void * tag)1168 dsl_pool_config_exit(dsl_pool_t *dp, void *tag)
1169 {
1170 	rrw_exit(&dp->dp_config_rwlock, tag);
1171 }
1172 
1173 boolean_t
dsl_pool_config_held(dsl_pool_t * dp)1174 dsl_pool_config_held(dsl_pool_t *dp)
1175 {
1176 	return (RRW_LOCK_HELD(&dp->dp_config_rwlock));
1177 }
1178 
1179 boolean_t
dsl_pool_config_held_writer(dsl_pool_t * dp)1180 dsl_pool_config_held_writer(dsl_pool_t *dp)
1181 {
1182 	return (RRW_WRITE_HELD(&dp->dp_config_rwlock));
1183 }
1184