1 // SPDX-License-Identifier: CDDL-1.0
2 /*
3 * CDDL HEADER START
4 *
5 * The contents of this file are subject to the terms of the
6 * Common Development and Distribution License (the "License").
7 * You may not use this file except in compliance with the License.
8 *
9 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10 * or https://opensource.org/licenses/CDDL-1.0.
11 * See the License for the specific language governing permissions
12 * and limitations under the License.
13 *
14 * When distributing Covered Code, include this CDDL HEADER in each
15 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16 * If applicable, add the following below this CDDL HEADER, with the
17 * fields enclosed by brackets "[]" replaced with your own identifying
18 * information: Portions Copyright [yyyy] [name of copyright owner]
19 *
20 * CDDL HEADER END
21 */
22 /*
23 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
25 * Copyright 2016 Nexenta Systems, Inc. All rights reserved.
26 */
27
28 #ifndef _SYS_FS_ZFS_ZNODE_H
29 #define _SYS_FS_ZFS_ZNODE_H
30
31 #include <sys/zfs_acl.h>
32 #include <sys/zil.h>
33 #include <sys/zfs_project.h>
34
35 #ifdef __cplusplus
36 extern "C" {
37 #endif
38
39 /*
40 * Additional file level attributes, that are stored
41 * in the upper half of z_pflags
42 */
43 #define ZFS_READONLY 0x0000000100000000ull
44 #define ZFS_HIDDEN 0x0000000200000000ull
45 #define ZFS_SYSTEM 0x0000000400000000ull
46 #define ZFS_ARCHIVE 0x0000000800000000ull
47 #define ZFS_IMMUTABLE 0x0000001000000000ull
48 #define ZFS_NOUNLINK 0x0000002000000000ull
49 #define ZFS_APPENDONLY 0x0000004000000000ull
50 #define ZFS_NODUMP 0x0000008000000000ull
51 #define ZFS_OPAQUE 0x0000010000000000ull
52 #define ZFS_AV_QUARANTINED 0x0000020000000000ull
53 #define ZFS_AV_MODIFIED 0x0000040000000000ull
54 #define ZFS_REPARSE 0x0000080000000000ull
55 #define ZFS_OFFLINE 0x0000100000000000ull
56 #define ZFS_SPARSE 0x0000200000000000ull
57
58 /*
59 * PROJINHERIT attribute is used to indicate that the child object under the
60 * directory which has the PROJINHERIT attribute needs to inherit its parent
61 * project ID that is used by project quota.
62 */
63 #define ZFS_PROJINHERIT 0x0000400000000000ull
64
65 /*
66 * PROJID attr is used internally to indicate that the object has project ID.
67 */
68 #define ZFS_PROJID 0x0000800000000000ull
69
70 #define ZFS_ATTR_SET(zp, attr, value, pflags, tx) \
71 { \
72 if (value) \
73 pflags |= attr; \
74 else \
75 pflags &= ~attr; \
76 VERIFY0(sa_update(zp->z_sa_hdl, SA_ZPL_FLAGS(ZTOZSB(zp)), \
77 &pflags, sizeof (pflags), tx)); \
78 }
79
80 /*
81 * Define special zfs pflags
82 */
83 #define ZFS_XATTR 0x1 /* is an extended attribute */
84 #define ZFS_INHERIT_ACE 0x2 /* ace has inheritable ACEs */
85 #define ZFS_ACL_TRIVIAL 0x4 /* files ACL is trivial */
86 #define ZFS_ACL_OBJ_ACE 0x8 /* ACL has CMPLX Object ACE */
87 #define ZFS_ACL_PROTECTED 0x10 /* ACL protected */
88 #define ZFS_ACL_DEFAULTED 0x20 /* ACL should be defaulted */
89 #define ZFS_ACL_AUTO_INHERIT 0x40 /* ACL should be inherited */
90 #define ZFS_BONUS_SCANSTAMP 0x80 /* Scanstamp in bonus area */
91 #define ZFS_NO_EXECS_DENIED 0x100 /* exec was given to everyone */
92
93 #define SA_ZPL_ATIME(z) z->z_attr_table[ZPL_ATIME]
94 #define SA_ZPL_MTIME(z) z->z_attr_table[ZPL_MTIME]
95 #define SA_ZPL_CTIME(z) z->z_attr_table[ZPL_CTIME]
96 #define SA_ZPL_CRTIME(z) z->z_attr_table[ZPL_CRTIME]
97 #define SA_ZPL_GEN(z) z->z_attr_table[ZPL_GEN]
98 #define SA_ZPL_DACL_ACES(z) z->z_attr_table[ZPL_DACL_ACES]
99 #define SA_ZPL_XATTR(z) z->z_attr_table[ZPL_XATTR]
100 #define SA_ZPL_SYMLINK(z) z->z_attr_table[ZPL_SYMLINK]
101 #define SA_ZPL_RDEV(z) z->z_attr_table[ZPL_RDEV]
102 #define SA_ZPL_SCANSTAMP(z) z->z_attr_table[ZPL_SCANSTAMP]
103 #define SA_ZPL_UID(z) z->z_attr_table[ZPL_UID]
104 #define SA_ZPL_GID(z) z->z_attr_table[ZPL_GID]
105 #define SA_ZPL_PARENT(z) z->z_attr_table[ZPL_PARENT]
106 #define SA_ZPL_LINKS(z) z->z_attr_table[ZPL_LINKS]
107 #define SA_ZPL_MODE(z) z->z_attr_table[ZPL_MODE]
108 #define SA_ZPL_DACL_COUNT(z) z->z_attr_table[ZPL_DACL_COUNT]
109 #define SA_ZPL_FLAGS(z) z->z_attr_table[ZPL_FLAGS]
110 #define SA_ZPL_SIZE(z) z->z_attr_table[ZPL_SIZE]
111 #define SA_ZPL_ZNODE_ACL(z) z->z_attr_table[ZPL_ZNODE_ACL]
112 #define SA_ZPL_DXATTR(z) z->z_attr_table[ZPL_DXATTR]
113 #define SA_ZPL_PAD(z) z->z_attr_table[ZPL_PAD]
114 #define SA_ZPL_PROJID(z) z->z_attr_table[ZPL_PROJID]
115 #define SA_ZPL_SEQ(z) z->z_attr_table[ZPL_SEQ]
116
117 /*
118 * may_grow for a dmu_tx_hold_sa() that may persist z_seq: the SA layout
119 * grows the first time SA_ZPL_SEQ is added, so grow until z_has_seq is
120 * set. z_has_seq is an in-core only marker (see znode_t) and is never
121 * persisted, so no global pflag bit is consumed.
122 */
123 #define ZFS_SEQ_MAY_GROW(zp) \
124 ((zp)->z_has_seq ? B_FALSE : B_TRUE)
125
126 /*
127 * Persist zp->z_seq: mark z_has_seq and add SA_ZPL_SEQ to the caller's
128 * bulk. No-op for legacy (non-SA-native) znodes. Chunked writers add
129 * SA_ZPL_SEQ once before their loop and set z_has_seq per chunk instead.
130 */
131 #define ZFS_PERSIST_SEQ(zp, bulk, count) \
132 { \
133 if ((zp)->z_is_sa) { \
134 (zp)->z_has_seq = B_TRUE; \
135 SA_ADD_BULK_ATTR((bulk), (count), SA_ZPL_SEQ(ZTOZSB(zp)), \
136 NULL, &(zp)->z_seq, 8); \
137 } \
138 }
139
140 /*
141 * Is ID ephemeral?
142 */
143 #define IS_EPHEMERAL(x) (x > MAXUID)
144
145 /*
146 * Should we use FUIDs?
147 */
148 #define USE_FUIDS(version, os) (version >= ZPL_VERSION_FUID && \
149 spa_version(dmu_objset_spa(os)) >= SPA_VERSION_FUID)
150 #define USE_SA(version, os) (version >= ZPL_VERSION_SA && \
151 spa_version(dmu_objset_spa(os)) >= SPA_VERSION_SA)
152
153 #define MASTER_NODE_OBJ 1
154
155 /*
156 * Special attributes for master node.
157 * "userquota@", "groupquota@" and "projectquota@" are also valid (from
158 * zfs_userquota_prop_prefixes[]).
159 */
160 #define ZFS_FSID "FSID"
161 #define ZFS_UNLINKED_SET "DELETE_QUEUE"
162 #define ZFS_ROOT_OBJ "ROOT"
163 #define ZPL_VERSION_STR "VERSION"
164 #define ZFS_FUID_TABLES "FUID"
165 #define ZFS_SHARES_DIR "SHARES"
166 #define ZFS_SA_ATTRS "SA_ATTRS"
167
168 /*
169 * Convert mode bits (zp_mode) to BSD-style DT_* values for storing in
170 * the directory entries. On Linux systems this value is already
171 * defined correctly as part of the /usr/include/dirent.h header file.
172 */
173 #ifndef IFTODT
174 #define IFTODT(mode) (((mode) & S_IFMT) >> 12)
175 #endif
176
177 /*
178 * The directory entry has the type (currently unused on Solaris) in the
179 * top 4 bits, and the object number in the low 48 bits. The "middle"
180 * 12 bits are unused.
181 */
182 #define ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4)
183 #define ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48)
184
185 extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len);
186 extern int zfs_obj_to_pobj(objset_t *osp, sa_handle_t *hdl,
187 sa_attr_type_t *sa_table, uint64_t *pobjp, int *is_xattrdir);
188 extern int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value);
189
190 #if defined(_KERNEL) || defined(_WANT_ZNODE)
191 #include <sys/zfs_znode_impl.h>
192 #include <sys/zfs_rlock.h>
193
194 /*
195 * Directory entry locks control access to directory entries.
196 * They are used to protect creates, deletes, and renames.
197 * Each directory znode has a mutex and a list of locked names.
198 */
199 typedef struct zfs_dirlock {
200 char *dl_name; /* directory entry being locked */
201 uint32_t dl_sharecnt; /* 0 if exclusive, > 0 if shared */
202 uint8_t dl_namelock; /* 1 if z_name_lock is NOT held */
203 uint16_t dl_namesize; /* set if dl_name was allocated */
204 kcondvar_t dl_cv; /* wait for entry to be unlocked */
205 struct znode *dl_dzp; /* directory znode */
206 struct zfs_dirlock *dl_next; /* next in z_dirlocks list */
207 } zfs_dirlock_t;
208
209 typedef struct znode {
210 uint64_t z_id; /* object ID for this znode */
211 kmutex_t z_lock; /* znode modification lock */
212 krwlock_t z_parent_lock; /* parent lock for directories */
213 krwlock_t z_name_lock; /* "master" lock for dirent locks */
214 zfs_dirlock_t *z_dirlocks; /* directory entry lock list */
215 zfs_rangelock_t z_rangelock; /* file range locks */
216 boolean_t z_unlinked; /* file has been unlinked */
217 boolean_t z_atime_dirty; /* atime needs to be synced */
218 boolean_t z_zn_prefetch; /* Prefetch znodes? */
219 boolean_t z_is_sa; /* are we native sa? */
220 boolean_t z_is_ctldir; /* are we .zfs entry */
221 boolean_t z_suspended; /* extra ref from a suspend? */
222 boolean_t z_xattr_dir_absent; /* no xattr dir (cached) */
223 boolean_t z_has_seq; /* SA_ZPL_SEQ present (in-core only) */
224 uint_t z_blksz; /* block size in bytes */
225 uint64_t z_seq; /* modification sequence number */
226 uint64_t z_mapcnt; /* number of pages mapped to file */
227 uint64_t z_dnodesize; /* dnode size */
228 uint64_t z_size; /* file size (cached) */
229 uint64_t z_pflags; /* pflags (cached) */
230 uint32_t z_sync_cnt; /* synchronous open count */
231 mode_t z_mode; /* mode (cached) */
232 kmutex_t z_acl_lock; /* acl data lock */
233 zfs_acl_t *z_acl_cached; /* cached acl */
234 krwlock_t z_xattr_lock; /* xattr data lock */
235 nvlist_t *z_xattr_cached; /* cached xattrs */
236 uint64_t z_xattr_parent; /* parent obj for this xattr */
237 uint64_t z_projid; /* project ID */
238 list_node_t z_link_node; /* all znodes in fs link */
239 sa_handle_t *z_sa_hdl; /* handle to sa data */
240
241 /*
242 * Platform specific field, defined by each platform and only
243 * accessible from platform specific code.
244 */
245 ZNODE_OS_FIELDS;
246 } znode_t;
247 #endif
248
249 #ifdef _KERNEL
250 /* Verifies the znode is valid. */
251 static inline int
zfs_verify_zp(znode_t * zp)252 zfs_verify_zp(znode_t *zp)
253 {
254 if (unlikely(zp->z_sa_hdl == NULL))
255 return (SET_ERROR(EIO));
256 return (0);
257 }
258
259 /* zfs_enter and zfs_verify_zp together */
260 static inline int
zfs_enter_verify_zp(zfsvfs_t * zfsvfs,znode_t * zp,const char * tag)261 zfs_enter_verify_zp(zfsvfs_t *zfsvfs, znode_t *zp, const char *tag)
262 {
263 int error;
264 if ((error = zfs_enter(zfsvfs, tag)) != 0)
265 return (error);
266 if ((error = zfs_verify_zp(zp)) != 0) {
267 zfs_exit(zfsvfs, tag);
268 return (error);
269 }
270 return (0);
271 }
272
273 typedef struct znode_hold {
274 uint64_t zh_obj; /* object id */
275 avl_node_t zh_node; /* avl tree linkage */
276 kmutex_t zh_lock; /* lock serializing object access */
277 int zh_refcount; /* active consumer reference count */
278 } znode_hold_t;
279
280 static inline uint64_t
zfs_inherit_projid(znode_t * dzp)281 zfs_inherit_projid(znode_t *dzp)
282 {
283 return ((dzp->z_pflags & ZFS_PROJINHERIT) ? dzp->z_projid :
284 ZFS_DEFAULT_PROJID);
285 }
286
287 /*
288 * Timestamp defines
289 */
290 #define ACCESSED (ATTR_ATIME)
291 #define STATE_CHANGED (ATTR_CTIME)
292 #define CONTENT_MODIFIED (ATTR_MTIME | ATTR_CTIME)
293
294 extern int zfs_init_fs(zfsvfs_t *, znode_t **);
295 extern void zfs_set_dataprop(objset_t *);
296 extern void zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *,
297 dmu_tx_t *tx);
298 extern void zfs_tstamp_update_setup(znode_t *, uint_t, uint64_t [2],
299 uint64_t [2]);
300 extern void zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *);
301 extern int zfs_freesp(znode_t *, uint64_t, uint64_t, int, boolean_t);
302 extern void zfs_znode_init(void);
303 extern void zfs_znode_fini(void);
304 extern int zfs_znode_hold_compare(const void *, const void *);
305 extern znode_hold_t *zfs_znode_hold_enter(zfsvfs_t *, uint64_t);
306 extern void zfs_znode_hold_exit(zfsvfs_t *, znode_hold_t *);
307 extern int zfs_zget(zfsvfs_t *, uint64_t, znode_t **);
308 extern int zfs_rezget(znode_t *);
309 extern void zfs_zinactive(znode_t *);
310 extern void zfs_znode_delete(znode_t *, dmu_tx_t *);
311 extern void zfs_remove_op_tables(void);
312 extern int zfs_create_op_tables(void);
313 extern dev_t zfs_cmpldev(uint64_t);
314 extern int zfs_get_stats(objset_t *os, nvlist_t *nv);
315 extern boolean_t zfs_get_vfs_flag_unmounted(objset_t *os);
316 extern void zfs_znode_dmu_fini(znode_t *);
317
318 extern void zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
319 znode_t *dzp, znode_t *zp, const char *name, vsecattr_t *,
320 zfs_fuid_info_t *, vattr_t *vap);
321 extern int zfs_log_create_txtype(zil_create_t, vsecattr_t *vsecp,
322 vattr_t *vap);
323 extern void zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
324 znode_t *dzp, const char *name, uint64_t foid, boolean_t unlinked);
325 #define ZFS_NO_OBJECT 0 /* no object id */
326 extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
327 znode_t *dzp, znode_t *zp, const char *name);
328 extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
329 znode_t *dzp, znode_t *zp, const char *name, const char *link);
330 extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
331 znode_t *sdzp, const char *sname, znode_t *tdzp, const char *dname,
332 znode_t *szp);
333 extern void zfs_log_rename_exchange(zilog_t *zilog, dmu_tx_t *tx,
334 uint64_t txtype, znode_t *sdzp, const char *sname, znode_t *tdzp,
335 const char *dname, znode_t *szp);
336 extern void zfs_log_rename_whiteout(zilog_t *zilog, dmu_tx_t *tx,
337 uint64_t txtype, znode_t *sdzp, const char *sname, znode_t *tdzp,
338 const char *dname, znode_t *szp, znode_t *wzp);
339 extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
340 znode_t *zp, offset_t off, ssize_t len, boolean_t commit,
341 boolean_t o_direct, zil_callback_t callback, void *callback_data);
342 extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype,
343 znode_t *zp, uint64_t off, uint64_t len);
344 extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
345 znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp);
346 extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp,
347 vsecattr_t *vsecp, zfs_fuid_info_t *fuidp);
348 extern void zfs_log_clone_range(zilog_t *zilog, dmu_tx_t *tx, int txtype,
349 znode_t *zp, uint64_t offset, uint64_t length, uint64_t blksz,
350 const blkptr_t *bps, size_t nbps);
351 extern void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx);
352 extern void zfs_upgrade(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
353 extern void zfs_log_setsaxattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
354 znode_t *zp, const char *name, const void *value, size_t size);
355
356 extern void zfs_znode_update_vfs(struct znode *);
357
358 #endif
359 #ifdef __cplusplus
360 }
361 #endif
362
363 #endif /* _SYS_FS_ZFS_ZNODE_H */
364