xref: /illumos-gate/usr/src/uts/common/fs/zfs/sys/zfs_znode.h (revision 9adfa60d484ce2435f5af77cc99dcd4e692b6660)
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, 2015 by Delphix. All rights reserved.
24  * Copyright (c) 2014 Integros [integros.com]
25  */
26 
27 #ifndef	_SYS_FS_ZFS_ZNODE_H
28 #define	_SYS_FS_ZFS_ZNODE_H
29 
30 #ifdef _KERNEL
31 #include <sys/isa_defs.h>
32 #include <sys/types32.h>
33 #include <sys/attr.h>
34 #include <sys/list.h>
35 #include <sys/dmu.h>
36 #include <sys/sa.h>
37 #include <sys/zfs_vfsops.h>
38 #include <sys/rrwlock.h>
39 #include <sys/zfs_sa.h>
40 #include <sys/zfs_stat.h>
41 #endif
42 #include <sys/zfs_acl.h>
43 #include <sys/zil.h>
44 
45 #ifdef	__cplusplus
46 extern "C" {
47 #endif
48 
49 /*
50  * Additional file level attributes, that are stored
51  * in the upper half of zp_flags
52  */
53 #define	ZFS_READONLY		0x0000000100000000
54 #define	ZFS_HIDDEN		0x0000000200000000
55 #define	ZFS_SYSTEM		0x0000000400000000
56 #define	ZFS_ARCHIVE		0x0000000800000000
57 #define	ZFS_IMMUTABLE		0x0000001000000000
58 #define	ZFS_NOUNLINK		0x0000002000000000
59 #define	ZFS_APPENDONLY		0x0000004000000000
60 #define	ZFS_NODUMP		0x0000008000000000
61 #define	ZFS_OPAQUE		0x0000010000000000
62 #define	ZFS_AV_QUARANTINED 	0x0000020000000000
63 #define	ZFS_AV_MODIFIED 	0x0000040000000000
64 #define	ZFS_REPARSE		0x0000080000000000
65 #define	ZFS_OFFLINE		0x0000100000000000
66 #define	ZFS_SPARSE		0x0000200000000000
67 
68 #define	ZFS_ATTR_SET(zp, attr, value, pflags, tx) \
69 { \
70 	if (value) \
71 		pflags |= attr; \
72 	else \
73 		pflags &= ~attr; \
74 	VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_FLAGS(zp->z_zfsvfs), \
75 	    &pflags, sizeof (pflags), tx)); \
76 }
77 
78 /*
79  * Define special zfs pflags
80  */
81 #define	ZFS_XATTR		0x1		/* is an extended attribute */
82 #define	ZFS_INHERIT_ACE		0x2		/* ace has inheritable ACEs */
83 #define	ZFS_ACL_TRIVIAL 	0x4		/* files ACL is trivial */
84 #define	ZFS_ACL_OBJ_ACE 	0x8		/* ACL has CMPLX Object ACE */
85 #define	ZFS_ACL_PROTECTED	0x10		/* ACL protected */
86 #define	ZFS_ACL_DEFAULTED	0x20		/* ACL should be defaulted */
87 #define	ZFS_ACL_AUTO_INHERIT	0x40		/* ACL should be inherited */
88 #define	ZFS_BONUS_SCANSTAMP	0x80		/* Scanstamp in bonus area */
89 #define	ZFS_NO_EXECS_DENIED	0x100		/* exec was given to everyone */
90 
91 #define	SA_ZPL_ATIME(z)		z->z_attr_table[ZPL_ATIME]
92 #define	SA_ZPL_MTIME(z)		z->z_attr_table[ZPL_MTIME]
93 #define	SA_ZPL_CTIME(z)		z->z_attr_table[ZPL_CTIME]
94 #define	SA_ZPL_CRTIME(z)	z->z_attr_table[ZPL_CRTIME]
95 #define	SA_ZPL_GEN(z)		z->z_attr_table[ZPL_GEN]
96 #define	SA_ZPL_DACL_ACES(z)	z->z_attr_table[ZPL_DACL_ACES]
97 #define	SA_ZPL_XATTR(z)		z->z_attr_table[ZPL_XATTR]
98 #define	SA_ZPL_SYMLINK(z)	z->z_attr_table[ZPL_SYMLINK]
99 #define	SA_ZPL_RDEV(z)		z->z_attr_table[ZPL_RDEV]
100 #define	SA_ZPL_SCANSTAMP(z)	z->z_attr_table[ZPL_SCANSTAMP]
101 #define	SA_ZPL_UID(z)		z->z_attr_table[ZPL_UID]
102 #define	SA_ZPL_GID(z)		z->z_attr_table[ZPL_GID]
103 #define	SA_ZPL_PARENT(z)	z->z_attr_table[ZPL_PARENT]
104 #define	SA_ZPL_LINKS(z)		z->z_attr_table[ZPL_LINKS]
105 #define	SA_ZPL_MODE(z)		z->z_attr_table[ZPL_MODE]
106 #define	SA_ZPL_DACL_COUNT(z)	z->z_attr_table[ZPL_DACL_COUNT]
107 #define	SA_ZPL_FLAGS(z)		z->z_attr_table[ZPL_FLAGS]
108 #define	SA_ZPL_SIZE(z)		z->z_attr_table[ZPL_SIZE]
109 #define	SA_ZPL_ZNODE_ACL(z)	z->z_attr_table[ZPL_ZNODE_ACL]
110 #define	SA_ZPL_PAD(z)		z->z_attr_table[ZPL_PAD]
111 
112 /*
113  * Is ID ephemeral?
114  */
115 #define	IS_EPHEMERAL(x)		(x > MAXUID)
116 
117 /*
118  * Should we use FUIDs?
119  */
120 #define	USE_FUIDS(version, os)	(version >= ZPL_VERSION_FUID && \
121     spa_version(dmu_objset_spa(os)) >= SPA_VERSION_FUID)
122 #define	USE_SA(version, os) (version >= ZPL_VERSION_SA && \
123     spa_version(dmu_objset_spa(os)) >= SPA_VERSION_SA)
124 
125 #define	MASTER_NODE_OBJ	1
126 
127 /*
128  * Special attributes for master node.
129  * "userquota@" and "groupquota@" are also valid (from
130  * zfs_userquota_prop_prefixes[]).
131  */
132 #define	ZFS_FSID		"FSID"
133 #define	ZFS_UNLINKED_SET	"DELETE_QUEUE"
134 #define	ZFS_ROOT_OBJ		"ROOT"
135 #define	ZPL_VERSION_STR		"VERSION"
136 #define	ZFS_FUID_TABLES		"FUID"
137 #define	ZFS_SHARES_DIR		"SHARES"
138 #define	ZFS_SA_ATTRS		"SA_ATTRS"
139 
140 /*
141  * Convert mode bits (zp_mode) to BSD-style DT_* values for storing in
142  * the directory entries.
143  */
144 #define	IFTODT(mode) (((mode) & S_IFMT) >> 12)
145 
146 /*
147  * The directory entry has the type (currently unused on Solaris) in the
148  * top 4 bits, and the object number in the low 48 bits.  The "middle"
149  * 12 bits are unused.
150  */
151 #define	ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4)
152 #define	ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48)
153 
154 /*
155  * Directory entry locks control access to directory entries.
156  * They are used to protect creates, deletes, and renames.
157  * Each directory znode has a mutex and a list of locked names.
158  */
159 #ifdef _KERNEL
160 typedef struct zfs_dirlock {
161 	char		*dl_name;	/* directory entry being locked */
162 	uint32_t	dl_sharecnt;	/* 0 if exclusive, > 0 if shared */
163 	uint8_t		dl_namelock;	/* 1 if z_name_lock is NOT held */
164 	uint16_t	dl_namesize;	/* set if dl_name was allocated */
165 	kcondvar_t	dl_cv;		/* wait for entry to be unlocked */
166 	struct znode	*dl_dzp;	/* directory znode */
167 	struct zfs_dirlock *dl_next;	/* next in z_dirlocks list */
168 } zfs_dirlock_t;
169 
170 typedef struct znode {
171 	struct zfsvfs	*z_zfsvfs;
172 	vnode_t		*z_vnode;
173 	uint64_t	z_id;		/* object ID for this znode */
174 	kmutex_t	z_lock;		/* znode modification lock */
175 	krwlock_t	z_parent_lock;	/* parent lock for directories */
176 	krwlock_t	z_name_lock;	/* "master" lock for dirent locks */
177 	zfs_dirlock_t	*z_dirlocks;	/* directory entry lock list */
178 	kmutex_t	z_range_lock;	/* protects changes to z_range_avl */
179 	avl_tree_t	z_range_avl;	/* avl tree of file range locks */
180 	uint8_t		z_unlinked;	/* file has been unlinked */
181 	uint8_t		z_atime_dirty;	/* atime needs to be synced */
182 	uint8_t		z_zn_prefetch;	/* Prefetch znodes? */
183 	uint8_t		z_moved;	/* Has this znode been moved? */
184 	uint_t		z_blksz;	/* block size in bytes */
185 	uint_t		z_seq;		/* modification sequence number */
186 	uint64_t	z_mapcnt;	/* number of pages mapped to file */
187 	uint64_t	z_gen;		/* generation (cached) */
188 	uint64_t	z_size;		/* file size (cached) */
189 	uint64_t	z_atime[2];	/* atime (cached) */
190 	uint64_t	z_links;	/* file links (cached) */
191 	uint64_t	z_pflags;	/* pflags (cached) */
192 	uint64_t	z_uid;		/* uid fuid (cached) */
193 	uint64_t	z_gid;		/* gid fuid (cached) */
194 	mode_t		z_mode;		/* mode (cached) */
195 	uint32_t	z_sync_cnt;	/* synchronous open count */
196 	kmutex_t	z_acl_lock;	/* acl data lock */
197 	zfs_acl_t	*z_acl_cached;	/* cached acl */
198 	list_node_t	z_link_node;	/* all znodes in fs link */
199 	sa_handle_t	*z_sa_hdl;	/* handle to sa data */
200 	boolean_t	z_is_sa;	/* are we native sa? */
201 } znode_t;
202 
203 
204 /*
205  * Range locking rules
206  * --------------------
207  * 1. When truncating a file (zfs_create, zfs_setattr, zfs_space) the whole
208  *    file range needs to be locked as RL_WRITER. Only then can the pages be
209  *    freed etc and zp_size reset. zp_size must be set within range lock.
210  * 2. For writes and punching holes (zfs_write & zfs_space) just the range
211  *    being written or freed needs to be locked as RL_WRITER.
212  *    Multiple writes at the end of the file must coordinate zp_size updates
213  *    to ensure data isn't lost. A compare and swap loop is currently used
214  *    to ensure the file size is at least the offset last written.
215  * 3. For reads (zfs_read, zfs_get_data & zfs_putapage) just the range being
216  *    read needs to be locked as RL_READER. A check against zp_size can then
217  *    be made for reading beyond end of file.
218  */
219 
220 /*
221  * Convert between znode pointers and vnode pointers
222  */
223 #define	ZTOV(ZP)	((ZP)->z_vnode)
224 #define	VTOZ(VP)	((znode_t *)(VP)->v_data)
225 
226 /* Called on entry to each ZFS vnode and vfs operation  */
227 #define	ZFS_ENTER(zfsvfs) \
228 	{ \
229 		rrm_enter_read(&(zfsvfs)->z_teardown_lock, FTAG); \
230 		if ((zfsvfs)->z_unmounted) { \
231 			ZFS_EXIT(zfsvfs); \
232 			return (EIO); \
233 		} \
234 	}
235 
236 /* Must be called before exiting the vop */
237 #define	ZFS_EXIT(zfsvfs) rrm_exit(&(zfsvfs)->z_teardown_lock, FTAG)
238 
239 /* Verifies the znode is valid */
240 #define	ZFS_VERIFY_ZP(zp) \
241 	if ((zp)->z_sa_hdl == NULL) { \
242 		ZFS_EXIT((zp)->z_zfsvfs); \
243 		return (EIO); \
244 	} \
245 
246 /*
247  * Macros for dealing with dmu_buf_hold
248  */
249 #define	ZFS_OBJ_HASH(obj_num)	((obj_num) & (ZFS_OBJ_MTX_SZ - 1))
250 #define	ZFS_OBJ_MUTEX(zfsvfs, obj_num)	\
251 	(&(zfsvfs)->z_hold_mtx[ZFS_OBJ_HASH(obj_num)])
252 #define	ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \
253 	mutex_enter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
254 #define	ZFS_OBJ_HOLD_TRYENTER(zfsvfs, obj_num) \
255 	mutex_tryenter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
256 #define	ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \
257 	mutex_exit(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
258 
259 /* Encode ZFS stored time values from a struct timespec */
260 #define	ZFS_TIME_ENCODE(tp, stmp)		\
261 {						\
262 	(stmp)[0] = (uint64_t)(tp)->tv_sec;	\
263 	(stmp)[1] = (uint64_t)(tp)->tv_nsec;	\
264 }
265 
266 /* Decode ZFS stored time values to a struct timespec */
267 #define	ZFS_TIME_DECODE(tp, stmp)		\
268 {						\
269 	(tp)->tv_sec = (time_t)(stmp)[0];		\
270 	(tp)->tv_nsec = (long)(stmp)[1];		\
271 }
272 
273 /*
274  * Timestamp defines
275  */
276 #define	ACCESSED		(AT_ATIME)
277 #define	STATE_CHANGED		(AT_CTIME)
278 #define	CONTENT_MODIFIED	(AT_MTIME | AT_CTIME)
279 
280 #define	ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \
281 	if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \
282 		zfs_tstamp_update_setup(zp, ACCESSED, NULL, NULL, B_FALSE);
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], boolean_t);
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_zget(zfsvfs_t *, uint64_t, znode_t **);
295 extern int	zfs_rezget(znode_t *);
296 extern void	zfs_zinactive(znode_t *);
297 extern void	zfs_znode_delete(znode_t *, dmu_tx_t *);
298 extern void	zfs_znode_free(znode_t *);
299 extern void	zfs_remove_op_tables();
300 extern int	zfs_create_op_tables();
301 extern int	zfs_sync(vfs_t *vfsp, short flag, cred_t *cr);
302 extern dev_t	zfs_cmpldev(uint64_t);
303 extern int	zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value);
304 extern int	zfs_get_stats(objset_t *os, nvlist_t *nv);
305 extern void	zfs_znode_dmu_fini(znode_t *);
306 
307 extern void zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
308     znode_t *dzp, znode_t *zp, char *name, vsecattr_t *, zfs_fuid_info_t *,
309     vattr_t *vap);
310 extern int zfs_log_create_txtype(zil_create_t, vsecattr_t *vsecp,
311     vattr_t *vap);
312 extern void zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
313     znode_t *dzp, char *name, uint64_t foid);
314 #define	ZFS_NO_OBJECT	0	/* no object id */
315 extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
316     znode_t *dzp, znode_t *zp, char *name);
317 extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
318     znode_t *dzp, znode_t *zp, char *name, char *link);
319 extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
320     znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp);
321 extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
322     znode_t *zp, offset_t off, ssize_t len, int ioflag);
323 extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype,
324     znode_t *zp, uint64_t off, uint64_t len);
325 extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
326     znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp);
327 extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp,
328     vsecattr_t *vsecp, zfs_fuid_info_t *fuidp);
329 extern void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx);
330 extern void zfs_upgrade(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
331 extern int zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
332 
333 extern caddr_t zfs_map_page(page_t *, enum seg_rw);
334 extern void zfs_unmap_page(page_t *, caddr_t);
335 
336 extern zil_get_data_t zfs_get_data;
337 extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE];
338 extern int zfsfstype;
339 
340 #endif /* _KERNEL */
341 
342 extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len);
343 
344 #ifdef	__cplusplus
345 }
346 #endif
347 
348 #endif	/* _SYS_FS_ZFS_ZNODE_H */
349