xref: /illumos-gate/usr/src/uts/common/fs/zfs/sys/zfs_znode.h (revision c43efa7f6b109f90d7f4962df8c0e1a94008d2d1)
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) 2012, 2018 by Delphix. All rights reserved.
24  * Copyright (c) 2014 Integros [integros.com]
25  * Copyright 2016 Nexenta Systems, Inc. All rights reserved.
26  * Copyright 2022-2023 Racktop Systems, Inc.
27  */
28 
29 #ifndef	_SYS_FS_ZFS_ZNODE_H
30 #define	_SYS_FS_ZFS_ZNODE_H
31 
32 #ifdef _KERNEL
33 #include <sys/isa_defs.h>
34 #include <sys/types32.h>
35 #include <sys/attr.h>
36 #include <sys/list.h>
37 #include <sys/dmu.h>
38 #include <sys/sa.h>
39 #include <sys/zfs_vfsops.h>
40 #include <sys/rrwlock.h>
41 #include <sys/zfs_sa.h>
42 #include <sys/zfs_stat.h>
43 #include <sys/zfs_rlock.h>
44 #endif
45 #include <sys/zfs_acl.h>
46 #include <sys/zil.h>
47 #include <sys/zfs_project.h>
48 
49 #ifdef	__cplusplus
50 extern "C" {
51 #endif
52 
53 /*
54  * Additional file level attributes, that are stored
55  * in the upper half of z_pflags
56  */
57 #define	ZFS_READONLY		0x0000000100000000
58 #define	ZFS_HIDDEN		0x0000000200000000
59 #define	ZFS_SYSTEM		0x0000000400000000
60 #define	ZFS_ARCHIVE		0x0000000800000000
61 #define	ZFS_IMMUTABLE		0x0000001000000000
62 #define	ZFS_NOUNLINK		0x0000002000000000
63 #define	ZFS_APPENDONLY		0x0000004000000000
64 #define	ZFS_NODUMP		0x0000008000000000
65 #define	ZFS_OPAQUE		0x0000010000000000
66 #define	ZFS_AV_QUARANTINED	0x0000020000000000
67 #define	ZFS_AV_MODIFIED		0x0000040000000000
68 #define	ZFS_REPARSE		0x0000080000000000
69 #define	ZFS_OFFLINE		0x0000100000000000
70 #define	ZFS_SPARSE		0x0000200000000000
71 /*
72  * PROJINHERIT attribute is used to indicate that the child object under the
73  * directory which has the PROJINHERIT attribute needs to inherit its parent
74  * project ID that is used by project quota.
75  */
76 #define	ZFS_PROJINHERIT		0x0000400000000000
77 /*
78  * PROJID attr is used internally to indicate that the object has project ID.
79  */
80 #define	ZFS_PROJID		0x0000800000000000
81 
82 #define	ZFS_ATTR_SET(zp, attr, value, pflags, tx) \
83 { \
84 	if (value) \
85 		pflags |= attr; \
86 	else \
87 		pflags &= ~attr; \
88 	VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_FLAGS(zp->z_zfsvfs), \
89 	    &pflags, sizeof (pflags), tx)); \
90 }
91 
92 /*
93  * Define special zfs pflags
94  */
95 #define	ZFS_XATTR		0x1		/* is an extended attribute */
96 #define	ZFS_INHERIT_ACE		0x2		/* ace has inheritable ACEs */
97 #define	ZFS_ACL_TRIVIAL		0x4		/* files ACL is trivial */
98 #define	ZFS_ACL_OBJ_ACE		0x8		/* ACL has CMPLX Object ACE */
99 #define	ZFS_ACL_PROTECTED	0x10		/* ACL protected */
100 #define	ZFS_ACL_DEFAULTED	0x20		/* ACL should be defaulted */
101 #define	ZFS_ACL_AUTO_INHERIT	0x40		/* ACL should be inherited */
102 #define	ZFS_BONUS_SCANSTAMP	0x80		/* Scanstamp in bonus area */
103 #define	ZFS_NO_EXECS_DENIED	0x100		/* exec was given to everyone */
104 
105 #define	SA_ZPL_ATIME(z)		z->z_attr_table[ZPL_ATIME]
106 #define	SA_ZPL_MTIME(z)		z->z_attr_table[ZPL_MTIME]
107 #define	SA_ZPL_CTIME(z)		z->z_attr_table[ZPL_CTIME]
108 #define	SA_ZPL_CRTIME(z)	z->z_attr_table[ZPL_CRTIME]
109 #define	SA_ZPL_GEN(z)		z->z_attr_table[ZPL_GEN]
110 #define	SA_ZPL_DACL_ACES(z)	z->z_attr_table[ZPL_DACL_ACES]
111 #define	SA_ZPL_XATTR(z)		z->z_attr_table[ZPL_XATTR]
112 #define	SA_ZPL_SYMLINK(z)	z->z_attr_table[ZPL_SYMLINK]
113 #define	SA_ZPL_RDEV(z)		z->z_attr_table[ZPL_RDEV]
114 #define	SA_ZPL_SCANSTAMP(z)	z->z_attr_table[ZPL_SCANSTAMP]
115 #define	SA_ZPL_UID(z)		z->z_attr_table[ZPL_UID]
116 #define	SA_ZPL_GID(z)		z->z_attr_table[ZPL_GID]
117 #define	SA_ZPL_PARENT(z)	z->z_attr_table[ZPL_PARENT]
118 #define	SA_ZPL_LINKS(z)		z->z_attr_table[ZPL_LINKS]
119 #define	SA_ZPL_MODE(z)		z->z_attr_table[ZPL_MODE]
120 #define	SA_ZPL_DACL_COUNT(z)	z->z_attr_table[ZPL_DACL_COUNT]
121 #define	SA_ZPL_FLAGS(z)		z->z_attr_table[ZPL_FLAGS]
122 #define	SA_ZPL_SIZE(z)		z->z_attr_table[ZPL_SIZE]
123 #define	SA_ZPL_ZNODE_ACL(z)	z->z_attr_table[ZPL_ZNODE_ACL]
124 #define	SA_ZPL_PAD(z)		z->z_attr_table[ZPL_PAD]
125 #define	SA_ZPL_PROJID(z)	z->z_attr_table[ZPL_PROJID]
126 
127 /*
128  * Is ID ephemeral?
129  */
130 #define	IS_EPHEMERAL(x)		(x > MAXUID)
131 
132 /*
133  * Should we use FUIDs?
134  */
135 #define	USE_FUIDS(version, os)	(version >= ZPL_VERSION_FUID && \
136     spa_version(dmu_objset_spa(os)) >= SPA_VERSION_FUID)
137 #define	USE_SA(version, os) (version >= ZPL_VERSION_SA && \
138     spa_version(dmu_objset_spa(os)) >= SPA_VERSION_SA)
139 
140 #define	MASTER_NODE_OBJ	1
141 
142 /*
143  * Special attributes for master node.
144  * "userquota@", "groupquota@" and "projectquota@" are also valid (from
145  * zfs_userquota_prop_prefixes[]).
146  */
147 #define	ZFS_FSID		"FSID"
148 #define	ZFS_UNLINKED_SET	"DELETE_QUEUE"
149 #define	ZFS_ROOT_OBJ		"ROOT"
150 #define	ZPL_VERSION_STR		"VERSION"
151 #define	ZFS_FUID_TABLES		"FUID"
152 #define	ZFS_SHARES_DIR		"SHARES"
153 #define	ZFS_SA_ATTRS		"SA_ATTRS"
154 
155 /*
156  * Convert mode bits (zp_mode) to BSD-style DT_* values for storing in
157  * the directory entries.
158  */
159 #define	IFTODT(mode) (((mode) & S_IFMT) >> 12)
160 
161 /*
162  * The directory entry has the type (currently unused on Solaris) in the
163  * top 4 bits, and the object number in the low 48 bits.  The "middle"
164  * 12 bits are unused.
165  */
166 #define	ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4)
167 #define	ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48)
168 
169 /*
170  * Directory entry locks control access to directory entries.
171  * They are used to protect creates, deletes, and renames.
172  * Each directory znode has a mutex and a list of locked names.
173  */
174 #ifdef _KERNEL
175 typedef struct zfs_dirlock {
176 	char		*dl_name;	/* directory entry being locked */
177 	uint32_t	dl_sharecnt;	/* 0 if exclusive, > 0 if shared */
178 	uint8_t		dl_namelock;	/* 1 if z_name_lock is NOT held */
179 	uint16_t	dl_namesize;	/* set if dl_name was allocated */
180 	kcondvar_t	dl_cv;		/* wait for entry to be unlocked */
181 	struct znode	*dl_dzp;	/* directory znode */
182 	struct zfs_dirlock *dl_next;	/* next in z_dirlocks list */
183 } zfs_dirlock_t;
184 
185 typedef struct znode {
186 	struct zfsvfs	*z_zfsvfs;
187 	vnode_t		*z_vnode;
188 	uint64_t	z_id;		/* object ID for this znode */
189 	kmutex_t	z_lock;		/* znode modification lock */
190 	krwlock_t	z_parent_lock;	/* parent lock for directories */
191 	krwlock_t	z_name_lock;	/* "master" lock for dirent locks */
192 	zfs_dirlock_t	*z_dirlocks;	/* directory entry lock list */
193 	rangelock_t	z_rangelock;	/* file range locks */
194 	uint8_t		z_unlinked;	/* file has been unlinked */
195 	uint8_t		z_atime_dirty;	/* atime needs to be synced */
196 	uint8_t		z_zn_prefetch;	/* Prefetch znodes? */
197 	uint8_t		z_moved;	/* Has this znode been moved? */
198 	uint_t		z_blksz;	/* block size in bytes */
199 	uint_t		z_seq;		/* modification sequence number */
200 	uint64_t	z_mapcnt;	/* number of pages mapped to file */
201 	uint64_t	z_dnodesize;	/* dnode size */
202 	uint64_t	z_gen;		/* generation (cached) */
203 	uint64_t	z_size;		/* file size (cached) */
204 	uint64_t	z_atime[2];	/* atime (cached) */
205 	uint64_t	z_links;	/* file links (cached) */
206 	uint64_t	z_pflags;	/* pflags (cached) */
207 	uint64_t	z_uid;		/* uid fuid (cached) */
208 	uint64_t	z_gid;		/* gid fuid (cached) */
209 	mode_t		z_mode;		/* mode (cached) */
210 	uint32_t	z_sync_cnt;	/* synchronous open count */
211 	krwlock_t	z_acl_lock;	/* acl data lock */
212 	zfs_acl_t	*z_acl_cached;	/* cached acl */
213 	uint64_t	z_projid;	/* project ID */
214 	list_node_t	z_link_node;	/* all znodes in fs link */
215 	sa_handle_t	*z_sa_hdl;	/* handle to sa data */
216 	boolean_t	z_is_sa;	/* are we native sa? */
217 } znode_t;
218 
219 static inline uint64_t
220 zfs_inherit_projid(znode_t *dzp)
221 {
222 	return ((dzp->z_pflags & ZFS_PROJINHERIT) ? dzp->z_projid :
223 	    ZFS_DEFAULT_PROJID);
224 }
225 
226 /*
227  * Range locking rules
228  * --------------------
229  * 1. When truncating a file (zfs_create, zfs_setattr, zfs_space) the whole
230  *    file range needs to be locked as RL_WRITER. Only then can the pages be
231  *    freed etc and zp_size reset. zp_size must be set within range lock.
232  * 2. For writes and punching holes (zfs_write & zfs_space) just the range
233  *    being written or freed needs to be locked as RL_WRITER.
234  *    Multiple writes at the end of the file must coordinate zp_size updates
235  *    to ensure data isn't lost. A compare and swap loop is currently used
236  *    to ensure the file size is at least the offset last written.
237  * 3. For reads (zfs_read, zfs_get_data & zfs_putapage) just the range being
238  *    read needs to be locked as RL_READER. A check against zp_size can then
239  *    be made for reading beyond end of file.
240  */
241 
242 /*
243  * Convert between znode pointers and vnode pointers
244  */
245 #define	ZTOV(ZP)	((ZP)->z_vnode)
246 #define	VTOZ(VP)	((znode_t *)(VP)->v_data)
247 
248 /* Called on entry to each ZFS vnode and vfs operation  */
249 #define	ZFS_ENTER(zfsvfs) \
250 	{ \
251 		rrm_enter_read(&(zfsvfs)->z_teardown_lock, FTAG); \
252 		if ((zfsvfs)->z_unmounted) { \
253 			ZFS_EXIT(zfsvfs); \
254 			return (EIO); \
255 		} \
256 	}
257 
258 /* Must be called before exiting the vop */
259 #define	ZFS_EXIT(zfsvfs) rrm_exit(&(zfsvfs)->z_teardown_lock, FTAG)
260 
261 /* Verifies the znode is valid */
262 #define	ZFS_VERIFY_ZP(zp) \
263 	if ((zp)->z_sa_hdl == NULL) { \
264 		ZFS_EXIT((zp)->z_zfsvfs); \
265 		return (EIO); \
266 	} \
267 
268 /*
269  * Macros for dealing with dmu_buf_hold
270  */
271 #define	ZFS_OBJ_HASH(obj_num)	((obj_num) & (ZFS_OBJ_MTX_SZ - 1))
272 #define	ZFS_OBJ_MUTEX(zfsvfs, obj_num)	\
273 	(&(zfsvfs)->z_hold_mtx[ZFS_OBJ_HASH(obj_num)])
274 #define	ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \
275 	mutex_enter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
276 #define	ZFS_OBJ_HOLD_TRYENTER(zfsvfs, obj_num) \
277 	mutex_tryenter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
278 #define	ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \
279 	mutex_exit(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
280 
281 /* Encode ZFS stored time values from a struct timespec */
282 #define	ZFS_TIME_ENCODE(tp, stmp)		\
283 {						\
284 	(stmp)[0] = (uint64_t)(tp)->tv_sec;	\
285 	(stmp)[1] = (uint64_t)(tp)->tv_nsec;	\
286 }
287 
288 /* Decode ZFS stored time values to a struct timespec */
289 #define	ZFS_TIME_DECODE(tp, stmp)		\
290 {						\
291 	(tp)->tv_sec = (time_t)(stmp)[0];		\
292 	(tp)->tv_nsec = (long)(stmp)[1];		\
293 }
294 
295 /*
296  * Timestamp defines
297  */
298 #define	ACCESSED		(AT_ATIME)
299 #define	STATE_CHANGED		(AT_CTIME)
300 #define	CONTENT_MODIFIED	(AT_MTIME | AT_CTIME)
301 
302 #define	ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \
303 	if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \
304 		zfs_tstamp_update_setup(zp, ACCESSED, NULL, NULL, B_FALSE);
305 
306 extern int	zfs_init_fs(zfsvfs_t *, znode_t **);
307 extern void	zfs_set_dataprop(objset_t *);
308 extern void	zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *,
309     dmu_tx_t *tx);
310 extern void	zfs_tstamp_update_setup(znode_t *, uint_t, uint64_t [2],
311     uint64_t [2], boolean_t);
312 extern void	zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *);
313 extern int	zfs_freesp(znode_t *, uint64_t, uint64_t, int, boolean_t);
314 extern void	zfs_znode_init(void);
315 extern void	zfs_znode_fini(void);
316 extern int	zfs_zget(zfsvfs_t *, uint64_t, znode_t **);
317 extern int	zfs_rezget(znode_t *);
318 extern void	zfs_zinactive(znode_t *);
319 extern void	zfs_znode_delete(znode_t *, dmu_tx_t *);
320 extern void	zfs_znode_free(znode_t *);
321 extern void	zfs_remove_op_tables();
322 extern int	zfs_create_op_tables();
323 extern int	zfs_sync(vfs_t *vfsp, short flag, cred_t *cr);
324 extern dev_t	zfs_cmpldev(uint64_t);
325 extern int	zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value);
326 extern int	zfs_get_stats(objset_t *os, nvlist_t *nv);
327 extern boolean_t zfs_get_vfs_flag_unmounted(objset_t *os);
328 extern void	zfs_znode_dmu_fini(znode_t *);
329 
330 extern void zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
331     znode_t *dzp, znode_t *zp, char *name, vsecattr_t *, zfs_fuid_info_t *,
332     vattr_t *vap);
333 extern int zfs_log_create_txtype(zil_create_t, vsecattr_t *vsecp,
334     vattr_t *vap);
335 extern void zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
336     znode_t *dzp, char *name, uint64_t foid, boolean_t unlinked);
337 #define	ZFS_NO_OBJECT	0	/* no object id */
338 extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
339     znode_t *dzp, znode_t *zp, char *name);
340 extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
341     znode_t *dzp, znode_t *zp, char *name, char *link);
342 extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
343     znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp);
344 extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
345     znode_t *zp, offset_t off, ssize_t len, int ioflag);
346 extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype,
347     znode_t *zp, uint64_t off, uint64_t len);
348 extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
349     znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp);
350 extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp,
351     vsecattr_t *vsecp, zfs_fuid_info_t *fuidp);
352 extern void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx);
353 extern void zfs_upgrade(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
354 extern int zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
355 
356 extern caddr_t zfs_map_page(page_t *, enum seg_rw);
357 extern void zfs_unmap_page(page_t *, caddr_t);
358 
359 extern zil_get_data_t zfs_get_data;
360 extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE];
361 extern int zfsfstype;
362 
363 #endif /* _KERNEL */
364 
365 extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len);
366 
367 #ifdef	__cplusplus
368 }
369 #endif
370 
371 #endif	/* _SYS_FS_ZFS_ZNODE_H */
372