xref: /illumos-gate/usr/src/uts/common/fs/zfs/sys/zfs_znode.h (revision b1d7ec75953cd517f5b7c3d9cb427ff8ec5d7d07)
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 2010 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 #ifndef	_SYS_FS_ZFS_ZNODE_H
27 #define	_SYS_FS_ZFS_ZNODE_H
28 
29 #ifdef _KERNEL
30 #include <sys/isa_defs.h>
31 #include <sys/types32.h>
32 #include <sys/attr.h>
33 #include <sys/list.h>
34 #include <sys/dmu.h>
35 #include <sys/sa.h>
36 #include <sys/zfs_vfsops.h>
37 #include <sys/rrwlock.h>
38 #include <sys/zfs_sa.h>
39 #endif
40 #include <sys/zfs_acl.h>
41 #include <sys/zil.h>
42 
43 #ifdef	__cplusplus
44 extern "C" {
45 #endif
46 
47 /*
48  * Additional file level attributes, that are stored
49  * in the upper half of zp_flags
50  */
51 #define	ZFS_READONLY		0x0000000100000000
52 #define	ZFS_HIDDEN		0x0000000200000000
53 #define	ZFS_SYSTEM		0x0000000400000000
54 #define	ZFS_ARCHIVE		0x0000000800000000
55 #define	ZFS_IMMUTABLE		0x0000001000000000
56 #define	ZFS_NOUNLINK		0x0000002000000000
57 #define	ZFS_APPENDONLY		0x0000004000000000
58 #define	ZFS_NODUMP		0x0000008000000000
59 #define	ZFS_OPAQUE		0x0000010000000000
60 #define	ZFS_AV_QUARANTINED 	0x0000020000000000
61 #define	ZFS_AV_MODIFIED 	0x0000040000000000
62 #define	ZFS_REPARSE		0x0000080000000000
63 
64 #define	ZFS_ATTR_SET(zp, attr, value, pflags, tx) \
65 { \
66 	if (value) \
67 		pflags |= attr; \
68 	else \
69 		pflags &= ~attr; \
70 	VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_FLAGS(zp->z_zfsvfs), \
71 	    &pflags, sizeof (pflags), tx)); \
72 }
73 
74 /*
75  * Define special zfs pflags
76  */
77 #define	ZFS_XATTR		0x1		/* is an extended attribute */
78 #define	ZFS_INHERIT_ACE		0x2		/* ace has inheritable ACEs */
79 #define	ZFS_ACL_TRIVIAL 	0x4		/* files ACL is trivial */
80 #define	ZFS_ACL_OBJ_ACE 	0x8		/* ACL has CMPLX Object ACE */
81 #define	ZFS_ACL_PROTECTED	0x10		/* ACL protected */
82 #define	ZFS_ACL_DEFAULTED	0x20		/* ACL should be defaulted */
83 #define	ZFS_ACL_AUTO_INHERIT	0x40		/* ACL should be inherited */
84 #define	ZFS_BONUS_SCANSTAMP	0x80		/* Scanstamp in bonus area */
85 #define	ZFS_NO_EXECS_DENIED	0x100		/* exec was given to everyone */
86 
87 #define	SA_ZPL_ATIME(z)		z->z_attr_table[ZPL_ATIME]
88 #define	SA_ZPL_MTIME(z)		z->z_attr_table[ZPL_MTIME]
89 #define	SA_ZPL_CTIME(z)		z->z_attr_table[ZPL_CTIME]
90 #define	SA_ZPL_CRTIME(z)	z->z_attr_table[ZPL_CRTIME]
91 #define	SA_ZPL_GEN(z)		z->z_attr_table[ZPL_GEN]
92 #define	SA_ZPL_DACL_ACES(z)	z->z_attr_table[ZPL_DACL_ACES]
93 #define	SA_ZPL_XATTR(z)		z->z_attr_table[ZPL_XATTR]
94 #define	SA_ZPL_SYMLINK(z)	z->z_attr_table[ZPL_SYMLINK]
95 #define	SA_ZPL_RDEV(z)		z->z_attr_table[ZPL_RDEV]
96 #define	SA_ZPL_SCANSTAMP(z)	z->z_attr_table[ZPL_SCANSTAMP]
97 #define	SA_ZPL_UID(z)		z->z_attr_table[ZPL_UID]
98 #define	SA_ZPL_GID(z)		z->z_attr_table[ZPL_GID]
99 #define	SA_ZPL_PARENT(z)	z->z_attr_table[ZPL_PARENT]
100 #define	SA_ZPL_LINKS(z)		z->z_attr_table[ZPL_LINKS]
101 #define	SA_ZPL_MODE(z)		z->z_attr_table[ZPL_MODE]
102 #define	SA_ZPL_DACL_COUNT(z)	z->z_attr_table[ZPL_DACL_COUNT]
103 #define	SA_ZPL_FLAGS(z)		z->z_attr_table[ZPL_FLAGS]
104 #define	SA_ZPL_SIZE(z)		z->z_attr_table[ZPL_SIZE]
105 #define	SA_ZPL_ZNODE_ACL(z)	z->z_attr_table[ZPL_ZNODE_ACL]
106 #define	SA_ZPL_PAD(z)		z->z_attr_table[ZPL_PAD]
107 
108 /*
109  * Is ID ephemeral?
110  */
111 #define	IS_EPHEMERAL(x)		(x > MAXUID)
112 
113 /*
114  * Should we use FUIDs?
115  */
116 #define	USE_FUIDS(version, os)	(version >= ZPL_VERSION_FUID && \
117     spa_version(dmu_objset_spa(os)) >= SPA_VERSION_FUID)
118 #define	USE_SA(version, os) (version >= ZPL_VERSION_SA && \
119     spa_version(dmu_objset_spa(os)) >= SPA_VERSION_SA)
120 
121 #define	MASTER_NODE_OBJ	1
122 
123 /*
124  * Special attributes for master node.
125  * "userquota@" and "groupquota@" are also valid (from
126  * zfs_userquota_prop_prefixes[]).
127  */
128 #define	ZFS_FSID		"FSID"
129 #define	ZFS_UNLINKED_SET	"DELETE_QUEUE"
130 #define	ZFS_ROOT_OBJ		"ROOT"
131 #define	ZPL_VERSION_STR		"VERSION"
132 #define	ZFS_FUID_TABLES		"FUID"
133 #define	ZFS_SHARES_DIR		"SHARES"
134 #define	ZFS_SA_ATTRS		"SA_ATTRS"
135 
136 #define	ZFS_MAX_BLOCKSIZE	(SPA_MAXBLOCKSIZE)
137 
138 /* Path component length */
139 /*
140  * The generic fs code uses MAXNAMELEN to represent
141  * what the largest component length is.  Unfortunately,
142  * this length includes the terminating NULL.  ZFS needs
143  * to tell the users via pathconf() and statvfs() what the
144  * true maximum length of a component is, excluding the NULL.
145  */
146 #define	ZFS_MAXNAMELEN	(MAXNAMELEN - 1)
147 
148 /*
149  * Convert mode bits (zp_mode) to BSD-style DT_* values for storing in
150  * the directory entries.
151  */
152 #define	IFTODT(mode) (((mode) & S_IFMT) >> 12)
153 
154 /*
155  * The directory entry has the type (currently unused on Solaris) in the
156  * top 4 bits, and the object number in the low 48 bits.  The "middle"
157  * 12 bits are unused.
158  */
159 #define	ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4)
160 #define	ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48)
161 
162 /*
163  * Directory entry locks control access to directory entries.
164  * They are used to protect creates, deletes, and renames.
165  * Each directory znode has a mutex and a list of locked names.
166  */
167 #ifdef _KERNEL
168 typedef struct zfs_dirlock {
169 	char		*dl_name;	/* directory entry being locked */
170 	uint32_t	dl_sharecnt;	/* 0 if exclusive, > 0 if shared */
171 	uint8_t		dl_namelock;	/* 1 if z_name_lock is NOT held */
172 	uint16_t	dl_namesize;	/* set if dl_name was allocated */
173 	kcondvar_t	dl_cv;		/* wait for entry to be unlocked */
174 	struct znode	*dl_dzp;	/* directory znode */
175 	struct zfs_dirlock *dl_next;	/* next in z_dirlocks list */
176 } zfs_dirlock_t;
177 
178 typedef struct znode {
179 	struct zfsvfs	*z_zfsvfs;
180 	vnode_t		*z_vnode;
181 	uint64_t	z_id;		/* object ID for this znode */
182 	kmutex_t	z_lock;		/* znode modification lock */
183 	krwlock_t	z_parent_lock;	/* parent lock for directories */
184 	krwlock_t	z_name_lock;	/* "master" lock for dirent locks */
185 	zfs_dirlock_t	*z_dirlocks;	/* directory entry lock list */
186 	kmutex_t	z_range_lock;	/* protects changes to z_range_avl */
187 	avl_tree_t	z_range_avl;	/* avl tree of file range locks */
188 	uint8_t		z_unlinked;	/* file has been unlinked */
189 	uint8_t		z_atime_dirty;	/* atime needs to be synced */
190 	uint8_t		z_zn_prefetch;	/* Prefetch znodes? */
191 	uint_t		z_blksz;	/* block size in bytes */
192 	uint_t		z_seq;		/* modification sequence number */
193 	uint64_t	z_mapcnt;	/* number of pages mapped to file */
194 	uint64_t	z_last_itx;	/* last ZIL itx on this znode */
195 	uint64_t	z_gen;		/* generation (cached) */
196 	uint64_t	z_size;		/* file size (cached) */
197 	uint64_t	z_atime[2];	/* atime (cached) */
198 	uint64_t	z_links;	/* file links (cached) */
199 	uint64_t	z_pflags;	/* pflags (cached) */
200 	uid_t		z_uid;		/* uid mapped (cached) */
201 	uid_t		z_gid;		/* gid mapped (cached) */
202 	mode_t		z_mode;		/* mode (cached) */
203 	uint32_t	z_sync_cnt;	/* synchronous open count */
204 	kmutex_t	z_acl_lock;	/* acl data lock */
205 	zfs_acl_t	*z_acl_cached;	/* cached acl */
206 	list_node_t	z_link_node;	/* all znodes in fs link */
207 	sa_handle_t	*z_sa_hdl;	/* handle to sa data */
208 	boolean_t	z_is_sa;	/* are we native sa? */
209 } znode_t;
210 
211 
212 /*
213  * Range locking rules
214  * --------------------
215  * 1. When truncating a file (zfs_create, zfs_setattr, zfs_space) the whole
216  *    file range needs to be locked as RL_WRITER. Only then can the pages be
217  *    freed etc and zp_size reset. zp_size must be set within range lock.
218  * 2. For writes and punching holes (zfs_write & zfs_space) just the range
219  *    being written or freed needs to be locked as RL_WRITER.
220  *    Multiple writes at the end of the file must coordinate zp_size updates
221  *    to ensure data isn't lost. A compare and swap loop is currently used
222  *    to ensure the file size is at least the offset last written.
223  * 3. For reads (zfs_read, zfs_get_data & zfs_putapage) just the range being
224  *    read needs to be locked as RL_READER. A check against zp_size can then
225  *    be made for reading beyond end of file.
226  */
227 
228 /*
229  * Convert between znode pointers and vnode pointers
230  */
231 #define	ZTOV(ZP)	((ZP)->z_vnode)
232 #define	VTOZ(VP)	((znode_t *)(VP)->v_data)
233 
234 /*
235  * ZFS_ENTER() is called on entry to each ZFS vnode and vfs operation.
236  * ZFS_EXIT() must be called before exitting the vop.
237  * ZFS_VERIFY_ZP() verifies the znode is valid.
238  */
239 #define	ZFS_ENTER(zfsvfs) \
240 	{ \
241 		rrw_enter(&(zfsvfs)->z_teardown_lock, RW_READER, FTAG); \
242 		if ((zfsvfs)->z_unmounted) { \
243 			ZFS_EXIT(zfsvfs); \
244 			return (EIO); \
245 		} \
246 	}
247 
248 #define	ZFS_EXIT(zfsvfs) rrw_exit(&(zfsvfs)->z_teardown_lock, FTAG)
249 
250 #define	ZFS_VERIFY_ZP(zp) \
251 	if ((zp)->z_sa_hdl == NULL) { \
252 		ZFS_EXIT((zp)->z_zfsvfs); \
253 		return (EIO); \
254 	} \
255 
256 /*
257  * Macros for dealing with dmu_buf_hold
258  */
259 #define	ZFS_OBJ_HASH(obj_num)	((obj_num) & (ZFS_OBJ_MTX_SZ - 1))
260 #define	ZFS_OBJ_MUTEX(zfsvfs, obj_num)	\
261 	(&(zfsvfs)->z_hold_mtx[ZFS_OBJ_HASH(obj_num)])
262 #define	ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \
263 	mutex_enter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
264 #define	ZFS_OBJ_HOLD_TRYENTER(zfsvfs, obj_num) \
265 	mutex_tryenter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
266 #define	ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \
267 	mutex_exit(ZFS_OBJ_MUTEX((zfsvfs), (obj_num)))
268 
269 /*
270  * Macros to encode/decode ZFS stored time values from/to struct timespec
271  */
272 #define	ZFS_TIME_ENCODE(tp, stmp)		\
273 {						\
274 	(stmp)[0] = (uint64_t)(tp)->tv_sec;	\
275 	(stmp)[1] = (uint64_t)(tp)->tv_nsec;	\
276 }
277 
278 #define	ZFS_TIME_DECODE(tp, stmp)		\
279 {						\
280 	(tp)->tv_sec = (time_t)(stmp)[0];		\
281 	(tp)->tv_nsec = (long)(stmp)[1];		\
282 }
283 
284 /*
285  * Timestamp defines
286  */
287 #define	ACCESSED		(AT_ATIME)
288 #define	STATE_CHANGED		(AT_CTIME)
289 #define	CONTENT_MODIFIED	(AT_MTIME | AT_CTIME)
290 
291 #define	ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \
292 	if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \
293 		zfs_tstamp_update_setup(zp, ACCESSED, NULL, NULL, B_FALSE);
294 
295 extern int	zfs_init_fs(zfsvfs_t *, znode_t **);
296 extern void	zfs_set_dataprop(objset_t *);
297 extern void	zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *,
298     dmu_tx_t *tx);
299 extern void	zfs_tstamp_update_setup(znode_t *, uint_t, uint64_t [2],
300     uint64_t [2], boolean_t);
301 extern void	zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *);
302 extern int	zfs_freesp(znode_t *, uint64_t, uint64_t, int, boolean_t);
303 extern void	zfs_znode_init(void);
304 extern void	zfs_znode_fini(void);
305 extern int	zfs_zget(zfsvfs_t *, uint64_t, znode_t **);
306 extern int	zfs_rezget(znode_t *);
307 extern void	zfs_zinactive(znode_t *);
308 extern void	zfs_znode_delete(znode_t *, dmu_tx_t *);
309 extern void	zfs_znode_free(znode_t *);
310 extern void	zfs_remove_op_tables();
311 extern int	zfs_create_op_tables();
312 extern int	zfs_sync(vfs_t *vfsp, short flag, cred_t *cr);
313 extern dev_t	zfs_cmpldev(uint64_t);
314 extern int	zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value);
315 extern int	zfs_get_stats(objset_t *os, nvlist_t *nv);
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, char *name, vsecattr_t *, zfs_fuid_info_t *,
320     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, char *name);
325 extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
326     znode_t *dzp, znode_t *zp, char *name);
327 extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
328     znode_t *dzp, znode_t *zp, char *name, char *link);
329 extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
330     znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp);
331 extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
332     znode_t *zp, offset_t off, ssize_t len, int ioflag);
333 extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype,
334     znode_t *zp, uint64_t off, uint64_t len);
335 extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
336     znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp);
337 extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp,
338     vsecattr_t *vsecp, zfs_fuid_info_t *fuidp);
339 extern void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx);
340 extern void zfs_upgrade(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
341 extern int zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx);
342 
343 extern caddr_t zfs_map_page(page_t *, enum seg_rw);
344 extern void zfs_unmap_page(page_t *, caddr_t);
345 
346 extern zil_get_data_t zfs_get_data;
347 extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE];
348 extern int zfsfstype;
349 
350 #endif /* _KERNEL */
351 
352 extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len);
353 
354 #ifdef	__cplusplus
355 }
356 #endif
357 
358 #endif	/* _SYS_FS_ZFS_ZNODE_H */
359