xref: /illumos-gate/usr/src/uts/common/fs/zfs/sys/zfs_znode.h (revision a07094369b21309434206d9b3601d162693466fc)
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, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
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 2005 Sun Microsystems, Inc.  All rights reserved.
24  * Use is subject to license terms.
25  */
26 
27 #ifndef	_SYS_FS_ZFS_ZNODE_H
28 #define	_SYS_FS_ZFS_ZNODE_H
29 
30 #pragma ident	"%Z%%M%	%I%	%E% SMI"
31 
32 #ifdef _KERNEL
33 #include <sys/isa_defs.h>
34 #include <sys/types32.h>
35 #include <sys/list.h>
36 #include <sys/dmu.h>
37 #include <sys/zfs_vfsops.h>
38 #endif
39 #include <sys/zfs_acl.h>
40 #include <sys/zil.h>
41 
42 #ifdef	__cplusplus
43 extern "C" {
44 #endif
45 
46 /*
47  * Define special zfs pflags
48  */
49 #define	ZFS_XATTR	0x1		/* is an extended attribute */
50 #define	ZFS_INHERIT_ACE	0x2		/* ace has inheritable ACEs */
51 #define	ZFS_ACL_TRIVIAL 0x4		/* files ACL is trivial */
52 
53 #define	MASTER_NODE_OBJ	1
54 
55 /*
56  * special attributes for master node.
57  */
58 
59 #define	ZFS_FSID		"FSID"
60 #define	ZFS_DELETE_QUEUE	"DELETE_QUEUE"
61 #define	ZFS_ROOT_OBJ		"ROOT"
62 #define	ZFS_VERSION_OBJ		"VERSION"
63 #define	ZFS_PROP_BLOCKPERPAGE	"BLOCKPERPAGE"
64 #define	ZFS_PROP_NOGROWBLOCKS	"NOGROWBLOCKS"
65 
66 #define	ZFS_FLAG_BLOCKPERPAGE	0x1
67 #define	ZFS_FLAG_NOGROWBLOCKS	0x2
68 
69 /*
70  * ZFS version - rev'd whenever an incompatible on-disk format change
71  * occurs.  Independent of SPA/DMU/ZAP versioning.
72  */
73 
74 #define	ZFS_VERSION		1ULL
75 
76 #define	ZFS_MAX_BLOCKSIZE	(SPA_MAXBLOCKSIZE)
77 
78 /* Path component length */
79 /*
80  * The generic fs code uses MAXNAMELEN to represent
81  * what the largest component length is.  Unfortunately,
82  * this length includes the terminating NULL.  ZFS needs
83  * to tell the users via pathconf() and statvfs() what the
84  * true maximum length of a component is, excluding the NULL.
85  */
86 #define	ZFS_MAXNAMELEN	(MAXNAMELEN - 1)
87 
88 /*
89  * This is the persistent portion of the znode.  It is stored
90  * in the "bonus buffer" of the file.  Short symbolic links
91  * are also stored in the bonus buffer.
92  */
93 typedef struct znode_phys {
94 	uint64_t zp_atime[2];		/*  0 - last file access time */
95 	uint64_t zp_mtime[2];		/* 16 - last file modification time */
96 	uint64_t zp_ctime[2];		/* 32 - last file change time */
97 	uint64_t zp_crtime[2];		/* 48 - creation time */
98 	uint64_t zp_gen;		/* 64 - generation (txg of creation) */
99 	uint64_t zp_mode;		/* 72 - file mode bits */
100 	uint64_t zp_size;		/* 80 - size of file */
101 	uint64_t zp_parent;		/* 88 - directory parent (`..') */
102 	uint64_t zp_links;		/* 96 - number of links to file */
103 	uint64_t zp_xattr;		/* 104 - DMU object for xattrs */
104 	uint64_t zp_rdev;		/* 112 - dev_t for VBLK & VCHR files */
105 	uint64_t zp_flags;		/* 120 - persistent flags */
106 	uint64_t zp_uid;		/* 128 - file owner */
107 	uint64_t zp_gid;		/* 136 - owning group */
108 	uint64_t zp_pad[4];		/* 144 - future */
109 	zfs_znode_acl_t zp_acl;		/* 176 - 263 ACL */
110 	/*
111 	 * Data may pad out any remaining bytes in the znode buffer, eg:
112 	 *
113 	 * |<---------------------- dnode_phys (512) ------------------------>|
114 	 * |<-- dnode (192) --->|<----------- "bonus" buffer (320) ---------->|
115 	 *			|<---- znode (264) ---->|<---- data (56) ---->|
116 	 *
117 	 * At present, we only use this space to store symbolic links.
118 	 */
119 } znode_phys_t;
120 
121 /*
122  * Directory entry locks control access to directory entries.
123  * They are used to protect creates, deletes, and renames.
124  * Each directory znode has a mutex and a list of locked names.
125  */
126 #ifdef _KERNEL
127 typedef struct zfs_dirlock {
128 	char		*dl_name;	/* directory entry being locked */
129 	uint32_t	dl_sharecnt;	/* 0 if exclusive, > 0 if shared */
130 	uint16_t	dl_namesize;	/* set if dl_name was allocated */
131 	kcondvar_t	dl_cv;		/* wait for entry to be unlocked */
132 	struct znode	*dl_dzp;	/* directory znode */
133 	struct zfs_dirlock *dl_next;	/* next in z_dirlocks list */
134 } zfs_dirlock_t;
135 
136 struct zcache_state;
137 
138 typedef struct znode {
139 	struct zfsvfs	*z_zfsvfs;
140 	vnode_t		*z_vnode;
141 	list_node_t 	z_list_node;	/* deleted znodes */
142 	uint64_t	z_id;		/* object ID for this znode */
143 	kmutex_t	z_lock;		/* znode modification lock */
144 	krwlock_t	z_map_lock;	/* page map lock */
145 	krwlock_t	z_grow_lock;	/* grow block size lock */
146 	krwlock_t	z_append_lock;	/* append-mode lock */
147 	zfs_dirlock_t	*z_dirlocks;	/* directory entry lock list */
148 	uint8_t		z_active;	/* znode is in use */
149 	uint8_t		z_reap;		/* reap file at last reference */
150 	uint8_t		z_atime_dirty;	/* atime needs to be synced */
151 	uint8_t		z_dbuf_held;	/* Is z_dbuf already held? */
152 	uint8_t		z_zn_prefetch;	/* Prefetch znodes? */
153 	uint_t		z_mapcnt;	/* number of memory maps to file */
154 	uint_t		z_blksz;	/* block size in bytes */
155 	uint_t		z_seq;		/* modification sequence number */
156 	uint64_t	z_last_itx;	/* last ZIL itx on this znode */
157 	kmutex_t	z_acl_lock;	/* acl data lock */
158 	list_node_t	z_link_node;	/* all znodes in fs link */
159 	list_node_t	z_zcache_node;
160 	struct zcache_state *z_zcache_state;
161 	uint64_t	z_zcache_access;
162 
163 	/*
164 	 * These are dmu managed fields.
165 	 */
166 	znode_phys_t	*z_phys;	/* pointer to persistent znode */
167 	dmu_buf_t	*z_dbuf;	/* buffer containing the z_phys */
168 } znode_t;
169 
170 /*
171  * The grow_lock is only applicable to "regular" files.
172  * The parent_lock is only applicable to directories.
173  */
174 #define	z_parent_lock	z_grow_lock
175 
176 /*
177  * Convert between znode pointers and vnode pointers
178  */
179 #define	ZTOV(ZP)	((ZP)->z_vnode)
180 #define	VTOZ(VP)	((znode_t *)(VP)->v_data)
181 
182 /*
183  * ZFS_ENTER() is called on entry to each ZFS vnode and vfs operation.
184  * ZFS_EXIT() must be called before exitting the vop.
185  */
186 #define	ZFS_ENTER(zfsvfs) \
187 	{ \
188 		atomic_add_32(&(zfsvfs)->z_op_cnt, 1); \
189 		if ((zfsvfs)->z_unmounted1) { \
190 			ZFS_EXIT(zfsvfs); \
191 			return (EIO); \
192 		} \
193 	}
194 #define	ZFS_EXIT(zfsvfs) atomic_add_32(&(zfsvfs)->z_op_cnt, -1)
195 
196 /*
197  * Macros for dealing with dmu_buf_hold
198  */
199 #define	ZFS_OBJ_HASH(obj_num)	(obj_num & (ZFS_OBJ_MTX_SZ - 1))
200 #define	ZFS_OBJ_MUTEX(zp)	\
201 	(&zp->z_zfsvfs->z_hold_mtx[ZFS_OBJ_HASH(zp->z_id)])
202 #define	ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \
203 	mutex_enter(&zfsvfs->z_hold_mtx[ZFS_OBJ_HASH(obj_num)]);
204 
205 #define	ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \
206 	mutex_exit(&zfsvfs->z_hold_mtx[ZFS_OBJ_HASH(obj_num)])
207 
208 /*
209  * Macros to encode/decode ZFS stored time values from/to struct timespec
210  */
211 #define	ZFS_TIME_ENCODE(tp, stmp)		\
212 {						\
213 	stmp[0] = (uint64_t)(tp)->tv_sec; 	\
214 	stmp[1] = (uint64_t)(tp)->tv_nsec;	\
215 }
216 
217 #define	ZFS_TIME_DECODE(tp, stmp)		\
218 {						\
219 	(tp)->tv_sec = (time_t)stmp[0];		\
220 	(tp)->tv_nsec = (long)stmp[1];		\
221 }
222 
223 /*
224  * Timestamp defines
225  */
226 #define	ACCESSED		(AT_ATIME)
227 #define	STATE_CHANGED		(AT_CTIME)
228 #define	CONTENT_MODIFIED	(AT_MTIME | AT_CTIME)
229 
230 #define	ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \
231 	if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \
232 		zfs_time_stamper(zp, ACCESSED, NULL)
233 
234 extern int	zfs_init_fs(zfsvfs_t *, znode_t **, cred_t *);
235 extern void	zfs_set_dataprop(objset_t *);
236 extern void	zfs_create_fs(objset_t *os, cred_t *cr, dmu_tx_t *tx);
237 extern void	zfs_time_stamper(znode_t *, uint_t, dmu_tx_t *);
238 extern void	zfs_time_stamper_locked(znode_t *, uint_t, dmu_tx_t *);
239 extern int	zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *);
240 extern int	zfs_freesp(znode_t *, uint64_t, uint64_t, int, dmu_tx_t *,
241     cred_t *cr);
242 extern void	zfs_znode_init(void);
243 extern void	zfs_znode_fini(void);
244 extern znode_t	*zfs_znode_alloc(zfsvfs_t *, dmu_buf_t *, uint64_t, int);
245 extern int	zfs_zget(zfsvfs_t *, uint64_t, znode_t **);
246 extern void	zfs_zinactive(znode_t *);
247 extern void	zfs_znode_delete(znode_t *, dmu_tx_t *);
248 extern void	zfs_znode_free(znode_t *);
249 extern int	zfs_delete_thread_target(zfsvfs_t *zfsvfs, int nthreads);
250 extern void	zfs_delete_wait_empty(zfsvfs_t *zfsvfs);
251 extern void	zfs_zcache_flush(zfsvfs_t *zfsvf);
252 extern void	zfs_remove_op_tables();
253 extern int	zfs_create_op_tables();
254 extern int	zfs_sync(vfs_t *vfsp, short flag, cred_t *cr);
255 
256 extern uint64_t zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, int txtype,
257     znode_t *dzp, znode_t *zp, char *name);
258 extern uint64_t zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, int txtype,
259     znode_t *dzp, char *name);
260 extern uint64_t zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, int txtype,
261     znode_t *dzp, znode_t *zp, char *name);
262 extern uint64_t zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, int txtype,
263     znode_t *dzp, znode_t *zp, char *name, char *link);
264 extern uint64_t zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, int txtype,
265     znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp);
266 extern uint64_t zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
267     znode_t *zp, offset_t off, ssize_t len, int ioflag, uio_t *uio);
268 extern uint64_t zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype,
269     znode_t *zp, uint64_t off, uint64_t len);
270 extern uint64_t zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
271     znode_t *zp, vattr_t *vap, uint_t mask_applied);
272 extern uint64_t zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, int txtype,
273     znode_t *zp, int aclcnt, ace_t *z_ace);
274 
275 extern zil_get_data_t zfs_get_data;
276 extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE];
277 extern int zfsfstype;
278 
279 #endif /* _KERNEL */
280 
281 #ifdef	__cplusplus
282 }
283 #endif
284 
285 #endif	/* _SYS_FS_ZFS_ZNODE_H */
286