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, 2017 by Delphix. All rights reserved.
24 * Copyright (c) 2013, Joyent, Inc. All rights reserved.
25 * Copyright (c) 2013 Steven Hartland. All rights reserved.
26 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
27 * Copyright (c) 2014 Integros [integros.com]
28 * Copyright 2016 Nexenta Systems, Inc. All rights reserved.
29 */
30
31 #ifndef _SYS_DSL_DATASET_H
32 #define _SYS_DSL_DATASET_H
33
34 #include <sys/dmu.h>
35 #include <sys/spa.h>
36 #include <sys/txg.h>
37 #include <sys/zio.h>
38 #include <sys/bplist.h>
39 #include <sys/dsl_synctask.h>
40 #include <sys/zfs_context.h>
41 #include <sys/dsl_deadlist.h>
42 #include <sys/refcount.h>
43 #include <sys/rrwlock.h>
44 #include <zfeature_common.h>
45
46 #ifdef __cplusplus
47 extern "C" {
48 #endif
49
50 struct dsl_dataset;
51 struct dsl_dir;
52 struct dsl_pool;
53
54 #define DS_FLAG_INCONSISTENT (1ULL<<0)
55 #define DS_IS_INCONSISTENT(ds) \
56 (dsl_dataset_phys(ds)->ds_flags & DS_FLAG_INCONSISTENT)
57
58 /*
59 * Do not allow this dataset to be promoted.
60 */
61 #define DS_FLAG_NOPROMOTE (1ULL<<1)
62
63 /*
64 * DS_FLAG_UNIQUE_ACCURATE is set if ds_unique_bytes has been correctly
65 * calculated for head datasets (starting with SPA_VERSION_UNIQUE_ACCURATE,
66 * refquota/refreservations).
67 */
68 #define DS_FLAG_UNIQUE_ACCURATE (1ULL<<2)
69
70 /*
71 * DS_FLAG_DEFER_DESTROY is set after 'zfs destroy -d' has been called
72 * on a dataset. This allows the dataset to be destroyed using 'zfs release'.
73 */
74 #define DS_FLAG_DEFER_DESTROY (1ULL<<3)
75 #define DS_IS_DEFER_DESTROY(ds) \
76 (dsl_dataset_phys(ds)->ds_flags & DS_FLAG_DEFER_DESTROY)
77
78 /*
79 * DS_FIELD_* are strings that are used in the "extensified" dataset zap object.
80 * They should be of the format <reverse-dns>:<field>.
81 */
82
83 /*
84 * This field's value is the object ID of a zap object which contains the
85 * bookmarks of this dataset. If it is present, then this dataset is counted
86 * in the refcount of the SPA_FEATURES_BOOKMARKS feature.
87 */
88 #define DS_FIELD_BOOKMARK_NAMES "com.delphix:bookmarks"
89
90 /*
91 * These fields are set on datasets that are in the middle of a resumable
92 * receive, and allow the sender to resume the send if it is interrupted.
93 */
94 #define DS_FIELD_RESUME_FROMGUID "com.delphix:resume_fromguid"
95 #define DS_FIELD_RESUME_TONAME "com.delphix:resume_toname"
96 #define DS_FIELD_RESUME_TOGUID "com.delphix:resume_toguid"
97 #define DS_FIELD_RESUME_OBJECT "com.delphix:resume_object"
98 #define DS_FIELD_RESUME_OFFSET "com.delphix:resume_offset"
99 #define DS_FIELD_RESUME_BYTES "com.delphix:resume_bytes"
100 #define DS_FIELD_RESUME_LARGEBLOCK "com.delphix:resume_largeblockok"
101 #define DS_FIELD_RESUME_EMBEDOK "com.delphix:resume_embedok"
102 #define DS_FIELD_RESUME_COMPRESSOK "com.delphix:resume_compressok"
103
104 /*
105 * DS_FLAG_CI_DATASET is set if the dataset contains a file system whose
106 * name lookups should be performed case-insensitively.
107 */
108 #define DS_FLAG_CI_DATASET (1ULL<<16)
109
110 #define DS_CREATE_FLAG_NODIRTY (1ULL<<24)
111
112 typedef struct dsl_dataset_snapshot_arg {
113 nvlist_t *ddsa_snaps;
114 nvlist_t *ddsa_props;
115 nvlist_t *ddsa_errors;
116 cred_t *ddsa_cr;
117 boolean_t ddsa_autosnap;
118 } dsl_dataset_snapshot_arg_t;
119
120 typedef struct dsl_dataset_phys {
121 uint64_t ds_dir_obj; /* DMU_OT_DSL_DIR */
122 uint64_t ds_prev_snap_obj; /* DMU_OT_DSL_DATASET */
123 uint64_t ds_prev_snap_txg;
124 uint64_t ds_next_snap_obj; /* DMU_OT_DSL_DATASET */
125 uint64_t ds_snapnames_zapobj; /* DMU_OT_DSL_DS_SNAP_MAP 0 for snaps */
126 uint64_t ds_num_children; /* clone/snap children; ==0 for head */
127 uint64_t ds_creation_time; /* seconds since 1970 */
128 uint64_t ds_creation_txg;
129 uint64_t ds_deadlist_obj; /* DMU_OT_DEADLIST */
130 /*
131 * ds_referenced_bytes, ds_compressed_bytes, and ds_uncompressed_bytes
132 * include all blocks referenced by this dataset, including those
133 * shared with any other datasets.
134 */
135 uint64_t ds_referenced_bytes;
136 uint64_t ds_compressed_bytes;
137 uint64_t ds_uncompressed_bytes;
138 uint64_t ds_unique_bytes; /* only relevant to snapshots */
139 /*
140 * The ds_fsid_guid is a 56-bit ID that can change to avoid
141 * collisions. The ds_guid is a 64-bit ID that will never
142 * change, so there is a small probability that it will collide.
143 */
144 uint64_t ds_fsid_guid;
145 uint64_t ds_guid;
146 uint64_t ds_flags; /* DS_FLAG_* */
147 blkptr_t ds_bp;
148 uint64_t ds_next_clones_obj; /* DMU_OT_DSL_CLONES */
149 uint64_t ds_props_obj; /* DMU_OT_DSL_PROPS for snaps */
150 uint64_t ds_userrefs_obj; /* DMU_OT_USERREFS */
151 uint64_t ds_pad[5]; /* pad out to 320 bytes for good measure */
152 } dsl_dataset_phys_t;
153
154 typedef struct dsl_dataset {
155 dmu_buf_user_t ds_dbu;
156 rrwlock_t ds_bp_rwlock; /* Protects ds_phys->ds_bp */
157
158 /* Immutable: */
159 struct dsl_dir *ds_dir;
160 dmu_buf_t *ds_dbuf;
161 uint64_t ds_object;
162 uint64_t ds_fsid_guid;
163 boolean_t ds_is_snapshot;
164
165 /* only used in syncing context, only valid for non-snapshots: */
166 struct dsl_dataset *ds_prev;
167 uint64_t ds_bookmarks; /* DMU_OTN_ZAP_METADATA */
168
169 /* has internal locking: */
170 dsl_deadlist_t ds_deadlist;
171 bplist_t ds_pending_deadlist;
172
173 /* protected by lock on pool's dp_dirty_datasets list */
174 txg_node_t ds_dirty_link;
175 list_node_t ds_synced_link;
176
177 /*
178 * ds_phys->ds_<accounting> is also protected by ds_lock.
179 * Protected by ds_lock:
180 */
181 kmutex_t ds_lock;
182 objset_t *ds_objset;
183 uint64_t ds_userrefs;
184 void *ds_owner;
185
186 /*
187 * NOTE: for pool with special device only
188 * Last transaction data was moved from special devices
189 * to other (regular) devices in a pool.
190 */
191 uint64_t ds_lstxg;
192
193 /*
194 * NOTE: for pool with special device only
195 * Level 0 bookmark we remembered when we were interrupted
196 * during traversal process. Used to resume traversal
197 * later.
198 */
199 zbookmark_phys_t ds_lszb;
200
201 /*
202 * Long holds prevent the ds from being destroyed; they allow the
203 * ds to remain held even after dropping the dp_config_rwlock.
204 * Owning counts as a long hold. See the comments above
205 * dsl_pool_hold() for details.
206 */
207
208 refcount_t ds_longholds;
209
210 /* no locking; only for making guesses */
211 uint64_t ds_trysnap_txg;
212
213 /* for objset_open() */
214 kmutex_t ds_opening_lock;
215
216 uint64_t ds_reserved; /* cached refreservation */
217 uint64_t ds_quota; /* cached refquota */
218
219 kmutex_t ds_sendstream_lock;
220 list_t ds_sendstreams;
221
222 /*
223 * When in the middle of a resumable receive, tracks how much
224 * progress we have made.
225 */
226 uint64_t ds_resume_object[TXG_SIZE];
227 uint64_t ds_resume_offset[TXG_SIZE];
228 uint64_t ds_resume_bytes[TXG_SIZE];
229
230 /* Protected by our dsl_dir's dd_lock */
231 list_t ds_prop_cbs;
232
233 /*
234 * For ZFEATURE_FLAG_PER_DATASET features, set if this dataset
235 * uses this feature.
236 */
237 uint8_t ds_feature_inuse[SPA_FEATURES];
238
239 /*
240 * Set if we need to activate the feature on this dataset this txg
241 * (used only in syncing context).
242 */
243 uint8_t ds_feature_activation_needed[SPA_FEATURES];
244
245 /* Protected by ds_lock; keep at end of struct for better locality */
246 char ds_snapname[ZFS_MAX_DATASET_NAME_LEN];
247 } dsl_dataset_t;
248
249 inline dsl_dataset_phys_t *
250 dsl_dataset_phys(dsl_dataset_t *ds)
251 {
252 return (ds->ds_dbuf->db_data);
253 }
254
255 typedef struct dsl_dataset_promote_arg {
256 const char *ddpa_clonename;
257 dsl_dataset_t *ddpa_clone;
258 list_t shared_snaps, origin_snaps, clone_snaps;
259 dsl_dataset_t *origin_origin; /* origin of the origin */
260 uint64_t used, comp, uncomp, unique, cloneusedsnap, originusedsnap;
261 nvlist_t *err_ds;
262 cred_t *cr;
263 } dsl_dataset_promote_arg_t;
264
265 typedef struct dsl_dataset_rollback_arg {
266 const char *ddra_fsname;
267 const char *ddra_tosnap;
268 void *ddra_owner;
269 nvlist_t *ddra_result;
270 } dsl_dataset_rollback_arg_t;
271
272 /*
273 * The max length of a temporary tag prefix is the number of hex digits
274 * required to express UINT64_MAX plus one for the hyphen.
275 */
276 #define MAX_TAG_PREFIX_LEN 17
277
278 #define dsl_dataset_is_snapshot(ds) \
279 (dsl_dataset_phys(ds)->ds_num_children != 0)
280
281 #define DS_UNIQUE_IS_ACCURATE(ds) \
282 ((dsl_dataset_phys(ds)->ds_flags & DS_FLAG_UNIQUE_ACCURATE) != 0)
283
284 int dsl_dataset_hold(struct dsl_pool *dp, const char *name, void *tag,
285 dsl_dataset_t **dsp);
286 boolean_t dsl_dataset_try_add_ref(struct dsl_pool *dp, dsl_dataset_t *ds,
287 void *tag);
288 int dsl_dataset_hold_obj(struct dsl_pool *dp, uint64_t dsobj, void *tag,
289 dsl_dataset_t **);
290 void dsl_dataset_rele(dsl_dataset_t *ds, void *tag);
291 int dsl_dataset_own(struct dsl_pool *dp, const char *name,
292 void *tag, dsl_dataset_t **dsp);
293 int dsl_dataset_own_obj(struct dsl_pool *dp, uint64_t dsobj,
294 void *tag, dsl_dataset_t **dsp);
295 void dsl_dataset_disown(dsl_dataset_t *ds, void *tag);
296 void dsl_dataset_name(dsl_dataset_t *ds, char *name);
297 boolean_t dsl_dataset_tryown(dsl_dataset_t *ds, void *tag);
298 int dsl_dataset_namelen(dsl_dataset_t *ds);
299 boolean_t dsl_dataset_has_owner(dsl_dataset_t *ds);
300 uint64_t dsl_dataset_create_sync(dsl_dir_t *pds, const char *lastname,
301 dsl_dataset_t *origin, uint64_t flags, cred_t *, dmu_tx_t *);
302 uint64_t dsl_dataset_create_sync_dd(dsl_dir_t *dd, dsl_dataset_t *origin,
303 uint64_t flags, dmu_tx_t *tx);
304 void dsl_dataset_snapshot_sync(void *arg, dmu_tx_t *tx);
305 int dsl_dataset_snapshot_check(void *arg, dmu_tx_t *tx);
306 int dsl_dataset_snapshot(nvlist_t *snaps, nvlist_t *props, nvlist_t *errors);
307 void dsl_dataset_promote_sync(void *arg, dmu_tx_t *tx);
308 int dsl_dataset_promote_check(void *arg, dmu_tx_t *tx);
309 int dsl_dataset_promote(const char *name, char *conflsnap);
310 int dsl_dataset_clone_swap(dsl_dataset_t *clone, dsl_dataset_t *origin_head,
311 boolean_t force);
312 int dsl_dataset_rename_snapshot(const char *fsname,
313 const char *oldsnapname, const char *newsnapname, boolean_t recursive);
314 int dsl_dataset_snapshot_tmp(const char *fsname, const char *snapname,
315 minor_t cleanup_minor, const char *htag);
316
317 blkptr_t *dsl_dataset_get_blkptr(dsl_dataset_t *ds);
318
319 spa_t *dsl_dataset_get_spa(dsl_dataset_t *ds);
320
321 boolean_t dsl_dataset_modified_since_snap(dsl_dataset_t *ds,
322 dsl_dataset_t *snap);
323
324 void dsl_dataset_sync(dsl_dataset_t *os, zio_t *zio, dmu_tx_t *tx);
325 void dsl_dataset_sync_done(dsl_dataset_t *os, dmu_tx_t *tx);
326
327 void dsl_dataset_block_born(dsl_dataset_t *ds, const blkptr_t *bp,
328 dmu_tx_t *tx);
329 int dsl_dataset_block_kill(dsl_dataset_t *ds, const blkptr_t *bp,
330 dmu_tx_t *tx, boolean_t async);
331
332 void dsl_dataset_dirty(dsl_dataset_t *ds, dmu_tx_t *tx);
333
334 int get_clones_stat_impl(dsl_dataset_t *ds, nvlist_t *val);
335 char *get_receive_resume_stats_impl(dsl_dataset_t *ds);
336 char *get_child_receive_stats(dsl_dataset_t *ds);
337 uint64_t dsl_get_refratio(dsl_dataset_t *ds);
338 uint64_t dsl_get_logicalreferenced(dsl_dataset_t *ds);
339 uint64_t dsl_get_compressratio(dsl_dataset_t *ds);
340 uint64_t dsl_get_used(dsl_dataset_t *ds);
341 uint64_t dsl_get_creation(dsl_dataset_t *ds);
342 uint64_t dsl_get_creationtxg(dsl_dataset_t *ds);
343 uint64_t dsl_get_refquota(dsl_dataset_t *ds);
344 uint64_t dsl_get_refreservation(dsl_dataset_t *ds);
345 uint64_t dsl_get_guid(dsl_dataset_t *ds);
346 uint64_t dsl_get_unique(dsl_dataset_t *ds);
347 uint64_t dsl_get_objsetid(dsl_dataset_t *ds);
348 uint64_t dsl_get_userrefs(dsl_dataset_t *ds);
349 uint64_t dsl_get_defer_destroy(dsl_dataset_t *ds);
350 uint64_t dsl_get_referenced(dsl_dataset_t *ds);
351 uint64_t dsl_get_numclones(dsl_dataset_t *ds);
352 uint64_t dsl_get_inconsistent(dsl_dataset_t *ds);
353 uint64_t dsl_get_available(dsl_dataset_t *ds);
354 int dsl_get_written(dsl_dataset_t *ds, uint64_t *written);
355 int dsl_get_prev_snap(dsl_dataset_t *ds, char *snap);
356 int dsl_get_mountpoint(dsl_dataset_t *ds, const char *dsname, char *value,
357 char *source);
358
359 void get_clones_stat(dsl_dataset_t *ds, nvlist_t *nv);
360
361 void dsl_dataset_stats(dsl_dataset_t *os, nvlist_t *nv);
362
363 void dsl_dataset_fast_stat(dsl_dataset_t *ds, dmu_objset_stats_t *stat);
364 void dsl_dataset_space(dsl_dataset_t *ds,
365 uint64_t *refdbytesp, uint64_t *availbytesp,
366 uint64_t *usedobjsp, uint64_t *availobjsp);
367 uint64_t dsl_dataset_fsid_guid(dsl_dataset_t *ds);
368 int dsl_dataset_space_written(dsl_dataset_t *oldsnap, dsl_dataset_t *new,
369 uint64_t *usedp, uint64_t *compp, uint64_t *uncompp);
370 int dsl_dataset_space_wouldfree(dsl_dataset_t *firstsnap, dsl_dataset_t *last,
371 uint64_t *usedp, uint64_t *compp, uint64_t *uncompp);
372 boolean_t dsl_dataset_is_dirty(dsl_dataset_t *ds);
373
374 int dsl_dsobj_to_dsname(char *pname, uint64_t obj, char *buf);
375
376 int dsl_dataset_check_quota(dsl_dataset_t *ds, boolean_t check_quota,
377 uint64_t asize, uint64_t inflight, uint64_t *used,
378 uint64_t *ref_rsrv);
379 int dsl_dataset_set_refquota(const char *dsname, zprop_source_t source,
380 uint64_t quota);
381 int dsl_dataset_set_refreservation(const char *dsname, zprop_source_t source,
382 uint64_t reservation);
383
384 boolean_t dsl_dataset_is_before(dsl_dataset_t *later, dsl_dataset_t *earlier,
385 uint64_t earlier_txg);
386 void dsl_dataset_long_hold(dsl_dataset_t *ds, void *tag);
387 void dsl_dataset_long_rele(dsl_dataset_t *ds, void *tag);
388 boolean_t dsl_dataset_long_held(dsl_dataset_t *ds);
389
390 int dsl_dataset_clone_swap_check_impl(dsl_dataset_t *clone,
391 dsl_dataset_t *origin_head, boolean_t force, void *owner, dmu_tx_t *tx);
392 void dsl_dataset_clone_swap_sync_impl(dsl_dataset_t *clone,
393 dsl_dataset_t *origin_head, dmu_tx_t *tx);
394 int dsl_dataset_snapshot_check_impl(dsl_dataset_t *ds, const char *snapname,
395 dmu_tx_t *tx, boolean_t recv, uint64_t cnt, cred_t *cr);
396 void dsl_dataset_snapshot_sync_impl(dsl_dataset_t *ds, const char *snapname,
397 dmu_tx_t *tx);
398
399 void dsl_dataset_remove_from_next_clones(dsl_dataset_t *ds, uint64_t obj,
400 dmu_tx_t *tx);
401 void dsl_dataset_recalc_head_uniq(dsl_dataset_t *ds);
402 int dsl_dataset_get_snapname(dsl_dataset_t *ds);
403 int dsl_dataset_snap_lookup(dsl_dataset_t *ds, const char *name,
404 uint64_t *value);
405 int dsl_dataset_snap_remove(dsl_dataset_t *ds, const char *name, dmu_tx_t *tx,
406 boolean_t adj_cnt);
407 void dsl_dataset_set_refreservation_sync_impl(dsl_dataset_t *ds,
408 zprop_source_t source, uint64_t value, dmu_tx_t *tx);
409 void dsl_dataset_zapify(dsl_dataset_t *ds, dmu_tx_t *tx);
410 boolean_t dsl_dataset_is_zapified(dsl_dataset_t *ds);
411 boolean_t dsl_dataset_has_resume_receive_state(dsl_dataset_t *ds);
412
413 int dsl_dataset_rollback_check(void *arg, dmu_tx_t *tx);
414 void dsl_dataset_rollback_sync(void *arg, dmu_tx_t *tx);
415 int dsl_dataset_rollback(const char *fsname, const char *tosnap, void *owner,
416 nvlist_t *result);
417
418 int dsl_dataset_snapshot_check(void *arg, dmu_tx_t *tx);
419 void dsl_dataset_snapshot_sync(void *arg, dmu_tx_t *tx);
420
421 typedef struct {
422 char name[ZFS_MAX_DATASET_NAME_LEN];
423 char origin_name[ZFS_MAX_DATASET_NAME_LEN];
424 uint64_t origin_guid;
425 uint64_t cookie;
426 boolean_t cookie_is_snap;
427 boolean_t top_level_ds;
428 dsl_dataset_t *ds;
429 list_node_t node;
430 avl_tree_t snapshots;
431 } zfs_ds_collector_entry_t;
432
433 uint64_t dsl_dataset_creation_txg(const char *name);
434
435 boolean_t dataset_name_hidden(const char *name);
436
437 zfs_ds_collector_entry_t *dsl_dataset_collector_cache_alloc();
438 void dsl_dataset_collector_cache_free(zfs_ds_collector_entry_t *entry);
439
440 void dsl_dataset_deactivate_feature(uint64_t dsobj,
441 spa_feature_t f, dmu_tx_t *tx);
442
443 #ifdef ZFS_DEBUG
444 #define dprintf_ds(ds, fmt, ...) do { \
445 if (zfs_flags & ZFS_DEBUG_DPRINTF) { \
446 char *__ds_name = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP); \
447 dsl_dataset_name(ds, __ds_name); \
448 dprintf("ds=%s " fmt, __ds_name, __VA_ARGS__); \
449 kmem_free(__ds_name, ZFS_MAX_DATASET_NAME_LEN); \
450 } \
451 _NOTE(CONSTCOND) } while (0)
452 #else
453 #define dprintf_ds(dd, fmt, ...)
454 #endif
455
456 #ifdef __cplusplus
457 }
458 #endif
459
460 #endif /* _SYS_DSL_DATASET_H */