]>
Commit | Line | Data |
---|---|---|
34dc7c2f BB |
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 | /* | |
428870ff | 22 | * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. |
34dc7c2f BB |
23 | */ |
24 | ||
428870ff BB |
25 | /* Portions Copyright 2010 Robert Milkowski */ |
26 | ||
34dc7c2f BB |
27 | #ifndef _SYS_DMU_H |
28 | #define _SYS_DMU_H | |
29 | ||
34dc7c2f BB |
30 | /* |
31 | * This file describes the interface that the DMU provides for its | |
32 | * consumers. | |
33 | * | |
34 | * The DMU also interacts with the SPA. That interface is described in | |
35 | * dmu_spa.h. | |
36 | */ | |
37 | ||
38 | #include <sys/inttypes.h> | |
39 | #include <sys/types.h> | |
40 | #include <sys/param.h> | |
41 | #include <sys/cred.h> | |
428870ff | 42 | #include <sys/time.h> |
59e6e7ca | 43 | #include <sys/uio.h> |
34dc7c2f BB |
44 | |
45 | #ifdef __cplusplus | |
46 | extern "C" { | |
47 | #endif | |
48 | ||
34dc7c2f BB |
49 | struct page; |
50 | struct vnode; | |
51 | struct spa; | |
52 | struct zilog; | |
53 | struct zio; | |
54 | struct blkptr; | |
55 | struct zap_cursor; | |
56 | struct dsl_dataset; | |
57 | struct dsl_pool; | |
58 | struct dnode; | |
59 | struct drr_begin; | |
60 | struct drr_end; | |
61 | struct zbookmark; | |
62 | struct spa; | |
63 | struct nvlist; | |
9babb374 | 64 | struct arc_buf; |
428870ff BB |
65 | struct zio_prop; |
66 | struct sa_handle; | |
34dc7c2f BB |
67 | |
68 | typedef struct objset objset_t; | |
69 | typedef struct dmu_tx dmu_tx_t; | |
70 | typedef struct dsl_dir dsl_dir_t; | |
71 | ||
72 | typedef enum dmu_object_type { | |
73 | DMU_OT_NONE, | |
74 | /* general: */ | |
75 | DMU_OT_OBJECT_DIRECTORY, /* ZAP */ | |
76 | DMU_OT_OBJECT_ARRAY, /* UINT64 */ | |
77 | DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ | |
78 | DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ | |
428870ff BB |
79 | DMU_OT_BPOBJ, /* UINT64 */ |
80 | DMU_OT_BPOBJ_HDR, /* UINT64 */ | |
34dc7c2f BB |
81 | /* spa: */ |
82 | DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ | |
83 | DMU_OT_SPACE_MAP, /* UINT64 */ | |
84 | /* zil: */ | |
85 | DMU_OT_INTENT_LOG, /* UINT64 */ | |
86 | /* dmu: */ | |
87 | DMU_OT_DNODE, /* DNODE */ | |
88 | DMU_OT_OBJSET, /* OBJSET */ | |
89 | /* dsl: */ | |
90 | DMU_OT_DSL_DIR, /* UINT64 */ | |
91 | DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ | |
92 | DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ | |
93 | DMU_OT_DSL_PROPS, /* ZAP */ | |
94 | DMU_OT_DSL_DATASET, /* UINT64 */ | |
95 | /* zpl: */ | |
96 | DMU_OT_ZNODE, /* ZNODE */ | |
97 | DMU_OT_OLDACL, /* Old ACL */ | |
98 | DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ | |
99 | DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ | |
100 | DMU_OT_MASTER_NODE, /* ZAP */ | |
101 | DMU_OT_UNLINKED_SET, /* ZAP */ | |
102 | /* zvol: */ | |
103 | DMU_OT_ZVOL, /* UINT8 */ | |
104 | DMU_OT_ZVOL_PROP, /* ZAP */ | |
105 | /* other; for testing only! */ | |
106 | DMU_OT_PLAIN_OTHER, /* UINT8 */ | |
107 | DMU_OT_UINT64_OTHER, /* UINT64 */ | |
108 | DMU_OT_ZAP_OTHER, /* ZAP */ | |
109 | /* new object types: */ | |
110 | DMU_OT_ERROR_LOG, /* ZAP */ | |
111 | DMU_OT_SPA_HISTORY, /* UINT8 */ | |
112 | DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ | |
113 | DMU_OT_POOL_PROPS, /* ZAP */ | |
114 | DMU_OT_DSL_PERMS, /* ZAP */ | |
115 | DMU_OT_ACL, /* ACL */ | |
116 | DMU_OT_SYSACL, /* SYSACL */ | |
117 | DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ | |
118 | DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ | |
b128c09f | 119 | DMU_OT_NEXT_CLONES, /* ZAP */ |
428870ff | 120 | DMU_OT_SCAN_QUEUE, /* ZAP */ |
9babb374 BB |
121 | DMU_OT_USERGROUP_USED, /* ZAP */ |
122 | DMU_OT_USERGROUP_QUOTA, /* ZAP */ | |
45d1cae3 | 123 | DMU_OT_USERREFS, /* ZAP */ |
428870ff BB |
124 | DMU_OT_DDT_ZAP, /* ZAP */ |
125 | DMU_OT_DDT_STATS, /* ZAP */ | |
126 | DMU_OT_SA, /* System attr */ | |
127 | DMU_OT_SA_MASTER_NODE, /* ZAP */ | |
128 | DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ | |
129 | DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ | |
130 | DMU_OT_SCAN_XLATE, /* ZAP */ | |
131 | DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ | |
132 | DMU_OT_DEADLIST, /* ZAP */ | |
133 | DMU_OT_DEADLIST_HDR, /* UINT64 */ | |
134 | DMU_OT_DSL_CLONES, /* ZAP */ | |
135 | DMU_OT_BPOBJ_SUBOBJ, /* UINT64 */ | |
34dc7c2f BB |
136 | DMU_OT_NUMTYPES |
137 | } dmu_object_type_t; | |
138 | ||
139 | typedef enum dmu_objset_type { | |
140 | DMU_OST_NONE, | |
141 | DMU_OST_META, | |
142 | DMU_OST_ZFS, | |
143 | DMU_OST_ZVOL, | |
144 | DMU_OST_OTHER, /* For testing only! */ | |
145 | DMU_OST_ANY, /* Be careful! */ | |
146 | DMU_OST_NUMTYPES | |
147 | } dmu_objset_type_t; | |
148 | ||
149 | void byteswap_uint64_array(void *buf, size_t size); | |
150 | void byteswap_uint32_array(void *buf, size_t size); | |
151 | void byteswap_uint16_array(void *buf, size_t size); | |
152 | void byteswap_uint8_array(void *buf, size_t size); | |
153 | void zap_byteswap(void *buf, size_t size); | |
154 | void zfs_oldacl_byteswap(void *buf, size_t size); | |
155 | void zfs_acl_byteswap(void *buf, size_t size); | |
156 | void zfs_znode_byteswap(void *buf, size_t size); | |
157 | ||
34dc7c2f BB |
158 | #define DS_FIND_SNAPSHOTS (1<<0) |
159 | #define DS_FIND_CHILDREN (1<<1) | |
160 | ||
161 | /* | |
162 | * The maximum number of bytes that can be accessed as part of one | |
163 | * operation, including metadata. | |
164 | */ | |
165 | #define DMU_MAX_ACCESS (10<<20) /* 10MB */ | |
b128c09f | 166 | #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ |
34dc7c2f | 167 | |
9babb374 BB |
168 | #define DMU_USERUSED_OBJECT (-1ULL) |
169 | #define DMU_GROUPUSED_OBJECT (-2ULL) | |
428870ff | 170 | #define DMU_DEADLIST_OBJECT (-3ULL) |
9babb374 | 171 | |
428870ff BB |
172 | /* |
173 | * artificial blkids for bonus buffer and spill blocks | |
174 | */ | |
175 | #define DMU_BONUS_BLKID (-1ULL) | |
176 | #define DMU_SPILL_BLKID (-2ULL) | |
34dc7c2f BB |
177 | /* |
178 | * Public routines to create, destroy, open, and close objsets. | |
179 | */ | |
428870ff BB |
180 | int dmu_objset_hold(const char *name, void *tag, objset_t **osp); |
181 | int dmu_objset_own(const char *name, dmu_objset_type_t type, | |
182 | boolean_t readonly, void *tag, objset_t **osp); | |
183 | void dmu_objset_rele(objset_t *os, void *tag); | |
184 | void dmu_objset_disown(objset_t *os, void *tag); | |
185 | int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); | |
186 | ||
34dc7c2f | 187 | int dmu_objset_evict_dbufs(objset_t *os); |
428870ff | 188 | int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, |
34dc7c2f | 189 | void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); |
428870ff BB |
190 | int dmu_objset_clone(const char *name, struct dsl_dataset *clone_origin, |
191 | uint64_t flags); | |
45d1cae3 BB |
192 | int dmu_objset_destroy(const char *name, boolean_t defer); |
193 | int dmu_snapshots_destroy(char *fsname, char *snapname, boolean_t defer); | |
572e2857 BB |
194 | int dmu_objset_snapshot(char *fsname, char *snapname, char *tag, |
195 | struct nvlist *props, boolean_t recursive, boolean_t temporary, int fd); | |
34dc7c2f BB |
196 | int dmu_objset_rename(const char *name, const char *newname, |
197 | boolean_t recursive); | |
428870ff | 198 | int dmu_objset_find(char *name, int func(const char *, void *), void *arg, |
34dc7c2f BB |
199 | int flags); |
200 | void dmu_objset_byteswap(void *buf, size_t size); | |
201 | ||
202 | typedef struct dmu_buf { | |
203 | uint64_t db_object; /* object that this buffer is part of */ | |
204 | uint64_t db_offset; /* byte offset in this object */ | |
205 | uint64_t db_size; /* size of buffer in bytes */ | |
206 | void *db_data; /* data in buffer */ | |
207 | } dmu_buf_t; | |
208 | ||
209 | typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); | |
210 | ||
211 | /* | |
212 | * The names of zap entries in the DIRECTORY_OBJECT of the MOS. | |
213 | */ | |
214 | #define DMU_POOL_DIRECTORY_OBJECT 1 | |
215 | #define DMU_POOL_CONFIG "config" | |
216 | #define DMU_POOL_ROOT_DATASET "root_dataset" | |
428870ff | 217 | #define DMU_POOL_SYNC_BPOBJ "sync_bplist" |
34dc7c2f BB |
218 | #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" |
219 | #define DMU_POOL_ERRLOG_LAST "errlog_last" | |
220 | #define DMU_POOL_SPARES "spares" | |
221 | #define DMU_POOL_DEFLATE "deflate" | |
222 | #define DMU_POOL_HISTORY "history" | |
223 | #define DMU_POOL_PROPS "pool_props" | |
224 | #define DMU_POOL_L2CACHE "l2cache" | |
428870ff BB |
225 | #define DMU_POOL_TMP_USERREFS "tmp_userrefs" |
226 | #define DMU_POOL_DDT "DDT-%s-%s-%s" | |
227 | #define DMU_POOL_DDT_STATS "DDT-statistics" | |
228 | #define DMU_POOL_CREATION_VERSION "creation_version" | |
229 | #define DMU_POOL_SCAN "scan" | |
230 | #define DMU_POOL_FREE_BPOBJ "free_bpobj" | |
b128c09f | 231 | |
34dc7c2f BB |
232 | /* |
233 | * Allocate an object from this objset. The range of object numbers | |
234 | * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. | |
235 | * | |
236 | * The transaction must be assigned to a txg. The newly allocated | |
237 | * object will be "held" in the transaction (ie. you can modify the | |
238 | * newly allocated object in this transaction). | |
239 | * | |
240 | * dmu_object_alloc() chooses an object and returns it in *objectp. | |
241 | * | |
242 | * dmu_object_claim() allocates a specific object number. If that | |
243 | * number is already allocated, it fails and returns EEXIST. | |
244 | * | |
245 | * Return 0 on success, or ENOSPC or EEXIST as specified above. | |
246 | */ | |
247 | uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, | |
248 | int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); | |
249 | int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, | |
250 | int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); | |
251 | int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, | |
9babb374 | 252 | int blocksize, dmu_object_type_t bonustype, int bonuslen); |
34dc7c2f BB |
253 | |
254 | /* | |
255 | * Free an object from this objset. | |
256 | * | |
257 | * The object's data will be freed as well (ie. you don't need to call | |
258 | * dmu_free(object, 0, -1, tx)). | |
259 | * | |
260 | * The object need not be held in the transaction. | |
261 | * | |
262 | * If there are any holds on this object's buffers (via dmu_buf_hold()), | |
263 | * or tx holds on the object (via dmu_tx_hold_object()), you can not | |
264 | * free it; it fails and returns EBUSY. | |
265 | * | |
266 | * If the object is not allocated, it fails and returns ENOENT. | |
267 | * | |
268 | * Return 0 on success, or EBUSY or ENOENT as specified above. | |
269 | */ | |
270 | int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); | |
271 | ||
272 | /* | |
273 | * Find the next allocated or free object. | |
274 | * | |
275 | * The objectp parameter is in-out. It will be updated to be the next | |
276 | * object which is allocated. Ignore objects which have not been | |
277 | * modified since txg. | |
278 | * | |
279 | * XXX Can only be called on a objset with no dirty data. | |
280 | * | |
281 | * Returns 0 on success, or ENOENT if there are no more objects. | |
282 | */ | |
283 | int dmu_object_next(objset_t *os, uint64_t *objectp, | |
284 | boolean_t hole, uint64_t txg); | |
285 | ||
286 | /* | |
287 | * Set the data blocksize for an object. | |
288 | * | |
289 | * The object cannot have any blocks allcated beyond the first. If | |
290 | * the first block is allocated already, the new size must be greater | |
291 | * than the current block size. If these conditions are not met, | |
292 | * ENOTSUP will be returned. | |
293 | * | |
294 | * Returns 0 on success, or EBUSY if there are any holds on the object | |
295 | * contents, or ENOTSUP as described above. | |
296 | */ | |
297 | int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, | |
298 | int ibs, dmu_tx_t *tx); | |
299 | ||
300 | /* | |
301 | * Set the checksum property on a dnode. The new checksum algorithm will | |
302 | * apply to all newly written blocks; existing blocks will not be affected. | |
303 | */ | |
304 | void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, | |
305 | dmu_tx_t *tx); | |
306 | ||
307 | /* | |
308 | * Set the compress property on a dnode. The new compression algorithm will | |
309 | * apply to all newly written blocks; existing blocks will not be affected. | |
310 | */ | |
311 | void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, | |
312 | dmu_tx_t *tx); | |
313 | ||
314 | /* | |
428870ff | 315 | * Decide how to write a block: checksum, compression, number of copies, etc. |
34dc7c2f | 316 | */ |
428870ff BB |
317 | #define WP_NOFILL 0x1 |
318 | #define WP_DMU_SYNC 0x2 | |
319 | #define WP_SPILL 0x4 | |
320 | ||
321 | void dmu_write_policy(objset_t *os, struct dnode *dn, int level, int wp, | |
322 | struct zio_prop *zp); | |
34dc7c2f BB |
323 | /* |
324 | * The bonus data is accessed more or less like a regular buffer. | |
325 | * You must dmu_bonus_hold() to get the buffer, which will give you a | |
326 | * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus | |
327 | * data. As with any normal buffer, you must call dmu_buf_read() to | |
328 | * read db_data, dmu_buf_will_dirty() before modifying it, and the | |
329 | * object must be held in an assigned transaction before calling | |
330 | * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus | |
331 | * buffer as well. You must release your hold with dmu_buf_rele(). | |
332 | */ | |
333 | int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); | |
334 | int dmu_bonus_max(void); | |
335 | int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); | |
428870ff | 336 | int dmu_set_bonustype(dmu_buf_t *, dmu_object_type_t, dmu_tx_t *); |
572e2857 | 337 | dmu_object_type_t dmu_get_bonustype(dmu_buf_t *); |
428870ff BB |
338 | int dmu_rm_spill(objset_t *, uint64_t, dmu_tx_t *); |
339 | ||
340 | /* | |
341 | * Special spill buffer support used by "SA" framework | |
342 | */ | |
343 | ||
344 | int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); | |
345 | int dmu_spill_hold_by_dnode(struct dnode *dn, uint32_t flags, | |
346 | void *tag, dmu_buf_t **dbp); | |
347 | int dmu_spill_hold_existing(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); | |
34dc7c2f BB |
348 | |
349 | /* | |
350 | * Obtain the DMU buffer from the specified object which contains the | |
351 | * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so | |
352 | * that it will remain in memory. You must release the hold with | |
353 | * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your | |
354 | * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. | |
355 | * | |
356 | * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill | |
357 | * on the returned buffer before reading or writing the buffer's | |
358 | * db_data. The comments for those routines describe what particular | |
359 | * operations are valid after calling them. | |
360 | * | |
361 | * The object number must be a valid, allocated object number. | |
362 | */ | |
363 | int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, | |
428870ff | 364 | void *tag, dmu_buf_t **, int flags); |
34dc7c2f BB |
365 | void dmu_buf_add_ref(dmu_buf_t *db, void* tag); |
366 | void dmu_buf_rele(dmu_buf_t *db, void *tag); | |
367 | uint64_t dmu_buf_refcount(dmu_buf_t *db); | |
368 | ||
369 | /* | |
370 | * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a | |
371 | * range of an object. A pointer to an array of dmu_buf_t*'s is | |
372 | * returned (in *dbpp). | |
373 | * | |
374 | * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and | |
375 | * frees the array. The hold on the array of buffers MUST be released | |
376 | * with dmu_buf_rele_array. You can NOT release the hold on each buffer | |
377 | * individually with dmu_buf_rele. | |
378 | */ | |
379 | int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, | |
380 | uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); | |
381 | void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); | |
382 | ||
383 | /* | |
384 | * Returns NULL on success, or the existing user ptr if it's already | |
385 | * been set. | |
386 | * | |
387 | * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). | |
388 | * | |
389 | * user_data_ptr_ptr should be NULL, or a pointer to a pointer which | |
390 | * will be set to db->db_data when you are allowed to access it. Note | |
391 | * that db->db_data (the pointer) can change when you do dmu_buf_read(), | |
392 | * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). | |
393 | * *user_data_ptr_ptr will be set to the new value when it changes. | |
394 | * | |
395 | * If non-NULL, pageout func will be called when this buffer is being | |
396 | * excised from the cache, so that you can clean up the data structure | |
397 | * pointed to by user_ptr. | |
398 | * | |
399 | * dmu_evict_user() will call the pageout func for all buffers in a | |
400 | * objset with a given pageout func. | |
401 | */ | |
402 | void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, | |
403 | dmu_buf_evict_func_t *pageout_func); | |
404 | /* | |
405 | * set_user_ie is the same as set_user, but request immediate eviction | |
406 | * when hold count goes to zero. | |
407 | */ | |
408 | void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, | |
409 | void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); | |
410 | void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, | |
411 | void *user_ptr, void *user_data_ptr_ptr, | |
412 | dmu_buf_evict_func_t *pageout_func); | |
413 | void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); | |
414 | ||
415 | /* | |
416 | * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. | |
417 | */ | |
418 | void *dmu_buf_get_user(dmu_buf_t *db); | |
419 | ||
420 | /* | |
421 | * Indicate that you are going to modify the buffer's data (db_data). | |
422 | * | |
423 | * The transaction (tx) must be assigned to a txg (ie. you've called | |
424 | * dmu_tx_assign()). The buffer's object must be held in the tx | |
425 | * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). | |
426 | */ | |
427 | void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); | |
428 | ||
9babb374 BB |
429 | /* |
430 | * Tells if the given dbuf is freeable. | |
431 | */ | |
432 | boolean_t dmu_buf_freeable(dmu_buf_t *); | |
433 | ||
34dc7c2f BB |
434 | /* |
435 | * You must create a transaction, then hold the objects which you will | |
436 | * (or might) modify as part of this transaction. Then you must assign | |
437 | * the transaction to a transaction group. Once the transaction has | |
438 | * been assigned, you can modify buffers which belong to held objects as | |
439 | * part of this transaction. You can't modify buffers before the | |
440 | * transaction has been assigned; you can't modify buffers which don't | |
441 | * belong to objects which this transaction holds; you can't hold | |
442 | * objects once the transaction has been assigned. You may hold an | |
443 | * object which you are going to free (with dmu_object_free()), but you | |
444 | * don't have to. | |
445 | * | |
446 | * You can abort the transaction before it has been assigned. | |
447 | * | |
448 | * Note that you may hold buffers (with dmu_buf_hold) at any time, | |
449 | * regardless of transaction state. | |
450 | */ | |
451 | ||
452 | #define DMU_NEW_OBJECT (-1ULL) | |
453 | #define DMU_OBJECT_END (-1ULL) | |
454 | ||
455 | dmu_tx_t *dmu_tx_create(objset_t *os); | |
456 | void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); | |
457 | void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, | |
458 | uint64_t len); | |
9babb374 | 459 | void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); |
34dc7c2f | 460 | void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); |
428870ff BB |
461 | void dmu_tx_hold_spill(dmu_tx_t *tx, uint64_t object); |
462 | void dmu_tx_hold_sa(dmu_tx_t *tx, struct sa_handle *hdl, boolean_t may_grow); | |
463 | void dmu_tx_hold_sa_create(dmu_tx_t *tx, int total_size); | |
34dc7c2f BB |
464 | void dmu_tx_abort(dmu_tx_t *tx); |
465 | int dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how); | |
466 | void dmu_tx_wait(dmu_tx_t *tx); | |
467 | void dmu_tx_commit(dmu_tx_t *tx); | |
468 | ||
428870ff BB |
469 | /* |
470 | * To register a commit callback, dmu_tx_callback_register() must be called. | |
471 | * | |
472 | * dcb_data is a pointer to caller private data that is passed on as a | |
473 | * callback parameter. The caller is responsible for properly allocating and | |
474 | * freeing it. | |
475 | * | |
476 | * When registering a callback, the transaction must be already created, but | |
477 | * it cannot be committed or aborted. It can be assigned to a txg or not. | |
478 | * | |
479 | * The callback will be called after the transaction has been safely written | |
480 | * to stable storage and will also be called if the dmu_tx is aborted. | |
481 | * If there is any error which prevents the transaction from being committed to | |
482 | * disk, the callback will be called with a value of error != 0. | |
483 | */ | |
484 | typedef void dmu_tx_callback_func_t(void *dcb_data, int error); | |
485 | ||
486 | void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, | |
487 | void *dcb_data); | |
488 | ||
34dc7c2f BB |
489 | /* |
490 | * Free up the data blocks for a defined range of a file. If size is | |
491 | * zero, the range from offset to end-of-file is freed. | |
492 | */ | |
493 | int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, | |
494 | uint64_t size, dmu_tx_t *tx); | |
b128c09f BB |
495 | int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, |
496 | uint64_t size); | |
497 | int dmu_free_object(objset_t *os, uint64_t object); | |
34dc7c2f BB |
498 | |
499 | /* | |
500 | * Convenience functions. | |
501 | * | |
502 | * Canfail routines will return 0 on success, or an errno if there is a | |
503 | * nonrecoverable I/O error. | |
504 | */ | |
9babb374 BB |
505 | #define DMU_READ_PREFETCH 0 /* prefetch */ |
506 | #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ | |
34dc7c2f | 507 | int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, |
9babb374 | 508 | void *buf, uint32_t flags); |
34dc7c2f BB |
509 | void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, |
510 | const void *buf, dmu_tx_t *tx); | |
b128c09f BB |
511 | void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, |
512 | dmu_tx_t *tx); | |
60101509 | 513 | #ifdef _KERNEL |
61e90960 | 514 | #include <linux/blkdev_compat.h> |
60101509 | 515 | int dmu_read_req(objset_t *os, uint64_t object, struct request *req); |
872e8d26 BB |
516 | int dmu_write_req(objset_t *os, uint64_t object, struct request *req, |
517 | dmu_tx_t *tx); | |
518 | int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); | |
519 | int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, | |
520 | dmu_tx_t *tx); | |
521 | int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size, | |
522 | dmu_tx_t *tx); | |
872e8d26 | 523 | #endif |
9babb374 BB |
524 | struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); |
525 | void dmu_return_arcbuf(struct arc_buf *buf); | |
526 | void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, | |
527 | dmu_tx_t *tx); | |
428870ff BB |
528 | int dmu_xuio_init(struct xuio *uio, int niov); |
529 | void dmu_xuio_fini(struct xuio *uio); | |
530 | int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, | |
531 | size_t n); | |
532 | int dmu_xuio_cnt(struct xuio *uio); | |
533 | struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); | |
534 | void dmu_xuio_clear(struct xuio *uio, int i); | |
0bc8fd78 BB |
535 | void xuio_stat_wbuf_copied(void); |
536 | void xuio_stat_wbuf_nocopy(void); | |
34dc7c2f BB |
537 | |
538 | extern int zfs_prefetch_disable; | |
539 | ||
540 | /* | |
541 | * Asynchronously try to read in the data. | |
542 | */ | |
543 | void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, | |
544 | uint64_t len); | |
545 | ||
546 | typedef struct dmu_object_info { | |
428870ff | 547 | /* All sizes are in bytes unless otherwise indicated. */ |
34dc7c2f BB |
548 | uint32_t doi_data_block_size; |
549 | uint32_t doi_metadata_block_size; | |
34dc7c2f BB |
550 | dmu_object_type_t doi_type; |
551 | dmu_object_type_t doi_bonus_type; | |
428870ff | 552 | uint64_t doi_bonus_size; |
34dc7c2f BB |
553 | uint8_t doi_indirection; /* 2 = dnode->indirect->data */ |
554 | uint8_t doi_checksum; | |
555 | uint8_t doi_compress; | |
556 | uint8_t doi_pad[5]; | |
428870ff BB |
557 | uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ |
558 | uint64_t doi_max_offset; | |
559 | uint64_t doi_fill_count; /* number of non-empty blocks */ | |
34dc7c2f BB |
560 | } dmu_object_info_t; |
561 | ||
562 | typedef void arc_byteswap_func_t(void *buf, size_t size); | |
563 | ||
564 | typedef struct dmu_object_type_info { | |
565 | arc_byteswap_func_t *ot_byteswap; | |
566 | boolean_t ot_metadata; | |
567 | char *ot_name; | |
568 | } dmu_object_type_info_t; | |
569 | ||
570 | extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; | |
571 | ||
572 | /* | |
573 | * Get information on a DMU object. | |
574 | * | |
575 | * Return 0 on success or ENOENT if object is not allocated. | |
576 | * | |
577 | * If doi is NULL, just indicates whether the object exists. | |
578 | */ | |
579 | int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); | |
580 | void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); | |
581 | void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); | |
582 | void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, | |
583 | u_longlong_t *nblk512); | |
584 | ||
585 | typedef struct dmu_objset_stats { | |
586 | uint64_t dds_num_clones; /* number of clones of this */ | |
587 | uint64_t dds_creation_txg; | |
588 | uint64_t dds_guid; | |
589 | dmu_objset_type_t dds_type; | |
590 | uint8_t dds_is_snapshot; | |
591 | uint8_t dds_inconsistent; | |
592 | char dds_origin[MAXNAMELEN]; | |
593 | } dmu_objset_stats_t; | |
594 | ||
595 | /* | |
596 | * Get stats on a dataset. | |
597 | */ | |
598 | void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); | |
599 | ||
600 | /* | |
601 | * Add entries to the nvlist for all the objset's properties. See | |
602 | * zfs_prop_table[] and zfs(1m) for details on the properties. | |
603 | */ | |
604 | void dmu_objset_stats(objset_t *os, struct nvlist *nv); | |
605 | ||
606 | /* | |
607 | * Get the space usage statistics for statvfs(). | |
608 | * | |
609 | * refdbytes is the amount of space "referenced" by this objset. | |
610 | * availbytes is the amount of space available to this objset, taking | |
611 | * into account quotas & reservations, assuming that no other objsets | |
612 | * use the space first. These values correspond to the 'referenced' and | |
613 | * 'available' properties, described in the zfs(1m) manpage. | |
614 | * | |
615 | * usedobjs and availobjs are the number of objects currently allocated, | |
616 | * and available. | |
617 | */ | |
618 | void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, | |
619 | uint64_t *usedobjsp, uint64_t *availobjsp); | |
620 | ||
621 | /* | |
622 | * The fsid_guid is a 56-bit ID that can change to avoid collisions. | |
623 | * (Contrast with the ds_guid which is a 64-bit ID that will never | |
624 | * change, so there is a small probability that it will collide.) | |
625 | */ | |
626 | uint64_t dmu_objset_fsid_guid(objset_t *os); | |
627 | ||
428870ff BB |
628 | /* |
629 | * Get the [cm]time for an objset's snapshot dir | |
630 | */ | |
631 | timestruc_t dmu_objset_snap_cmtime(objset_t *os); | |
632 | ||
34dc7c2f BB |
633 | int dmu_objset_is_snapshot(objset_t *os); |
634 | ||
635 | extern struct spa *dmu_objset_spa(objset_t *os); | |
636 | extern struct zilog *dmu_objset_zil(objset_t *os); | |
637 | extern struct dsl_pool *dmu_objset_pool(objset_t *os); | |
638 | extern struct dsl_dataset *dmu_objset_ds(objset_t *os); | |
639 | extern void dmu_objset_name(objset_t *os, char *buf); | |
640 | extern dmu_objset_type_t dmu_objset_type(objset_t *os); | |
641 | extern uint64_t dmu_objset_id(objset_t *os); | |
428870ff BB |
642 | extern uint64_t dmu_objset_syncprop(objset_t *os); |
643 | extern uint64_t dmu_objset_logbias(objset_t *os); | |
34dc7c2f BB |
644 | extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, |
645 | uint64_t *id, uint64_t *offp, boolean_t *case_conflict); | |
ebe7e575 | 646 | extern int dmu_snapshot_id(objset_t *os, const char *snapname, uint64_t *idp); |
34dc7c2f BB |
647 | extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, |
648 | int maxlen, boolean_t *conflict); | |
649 | extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, | |
650 | uint64_t *idp, uint64_t *offp); | |
9babb374 | 651 | |
428870ff BB |
652 | typedef int objset_used_cb_t(dmu_object_type_t bonustype, |
653 | void *bonus, uint64_t *userp, uint64_t *groupp); | |
9babb374 BB |
654 | extern void dmu_objset_register_type(dmu_objset_type_t ost, |
655 | objset_used_cb_t *cb); | |
34dc7c2f BB |
656 | extern void dmu_objset_set_user(objset_t *os, void *user_ptr); |
657 | extern void *dmu_objset_get_user(objset_t *os); | |
658 | ||
659 | /* | |
660 | * Return the txg number for the given assigned transaction. | |
661 | */ | |
662 | uint64_t dmu_tx_get_txg(dmu_tx_t *tx); | |
663 | ||
664 | /* | |
665 | * Synchronous write. | |
666 | * If a parent zio is provided this function initiates a write on the | |
667 | * provided buffer as a child of the parent zio. | |
668 | * In the absence of a parent zio, the write is completed synchronously. | |
669 | * At write completion, blk is filled with the bp of the written block. | |
670 | * Note that while the data covered by this function will be on stable | |
671 | * storage when the write completes this new data does not become a | |
672 | * permanent part of the file until the associated transaction commits. | |
673 | */ | |
428870ff BB |
674 | |
675 | /* | |
676 | * {zfs,zvol,ztest}_get_done() args | |
677 | */ | |
678 | typedef struct zgd { | |
679 | struct zilog *zgd_zilog; | |
680 | struct blkptr *zgd_bp; | |
681 | dmu_buf_t *zgd_db; | |
682 | struct rl *zgd_rl; | |
683 | void *zgd_private; | |
684 | } zgd_t; | |
685 | ||
686 | typedef void dmu_sync_cb_t(zgd_t *arg, int error); | |
687 | int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); | |
34dc7c2f BB |
688 | |
689 | /* | |
690 | * Find the next hole or data block in file starting at *off | |
691 | * Return found offset in *off. Return ESRCH for end of file. | |
692 | */ | |
693 | int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, | |
694 | uint64_t *off); | |
695 | ||
696 | /* | |
697 | * Initial setup and final teardown. | |
698 | */ | |
699 | extern void dmu_init(void); | |
700 | extern void dmu_fini(void); | |
701 | ||
702 | typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, | |
703 | uint64_t object, uint64_t offset, int len); | |
704 | void dmu_traverse_objset(objset_t *os, uint64_t txg_start, | |
705 | dmu_traverse_cb_t cb, void *arg); | |
706 | ||
707 | int dmu_sendbackup(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorigin, | |
708 | struct vnode *vp, offset_t *off); | |
709 | ||
710 | typedef struct dmu_recv_cookie { | |
711 | /* | |
712 | * This structure is opaque! | |
713 | * | |
714 | * If logical and real are different, we are recving the stream | |
715 | * into the "real" temporary clone, and then switching it with | |
716 | * the "logical" target. | |
717 | */ | |
718 | struct dsl_dataset *drc_logical_ds; | |
719 | struct dsl_dataset *drc_real_ds; | |
720 | struct drr_begin *drc_drrb; | |
721 | char *drc_tosnap; | |
428870ff | 722 | char *drc_top_ds; |
34dc7c2f BB |
723 | boolean_t drc_newfs; |
724 | boolean_t drc_force; | |
8d35c149 | 725 | struct avl_tree *drc_guid_to_ds_map; |
34dc7c2f BB |
726 | } dmu_recv_cookie_t; |
727 | ||
428870ff | 728 | int dmu_recv_begin(char *tofs, char *tosnap, char *topds, struct drr_begin *, |
45d1cae3 | 729 | boolean_t force, objset_t *origin, dmu_recv_cookie_t *); |
572e2857 BB |
730 | int dmu_recv_stream(dmu_recv_cookie_t *drc, struct vnode *vp, offset_t *voffp, |
731 | int cleanup_fd, uint64_t *action_handlep); | |
34dc7c2f | 732 | int dmu_recv_end(dmu_recv_cookie_t *drc); |
34dc7c2f | 733 | |
572e2857 BB |
734 | int dmu_diff(objset_t *tosnap, objset_t *fromsnap, struct vnode *vp, |
735 | offset_t *off); | |
736 | ||
34dc7c2f BB |
737 | /* CRC64 table */ |
738 | #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ | |
739 | extern uint64_t zfs_crc64_table[256]; | |
740 | ||
741 | #ifdef __cplusplus | |
742 | } | |
743 | #endif | |
744 | ||
745 | #endif /* _SYS_DMU_H */ |