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