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 2007 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 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 #ifdef _KERNEL 32 #include <sys/isa_defs.h> 33 #include <sys/types32.h> 34 #include <sys/list.h> 35 #include <sys/dmu.h> 36 #include <sys/zfs_vfsops.h> 37 #include <sys/rrwlock.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 #define ZFS_FSID "FSID" 59 #define ZFS_UNLINKED_SET "DELETE_QUEUE" 60 #define ZFS_ROOT_OBJ "ROOT" 61 #define ZPL_VERSION_STR "VERSION" 62 63 64 #define ZFS_MAX_BLOCKSIZE (SPA_MAXBLOCKSIZE) 65 66 /* Path component length */ 67 /* 68 * The generic fs code uses MAXNAMELEN to represent 69 * what the largest component length is. Unfortunately, 70 * this length includes the terminating NULL. ZFS needs 71 * to tell the users via pathconf() and statvfs() what the 72 * true maximum length of a component is, excluding the NULL. 73 */ 74 #define ZFS_MAXNAMELEN (MAXNAMELEN - 1) 75 76 /* 77 * Convert mode bits (zp_mode) to BSD-style DT_* values for storing in 78 * the directory entries. 79 */ 80 #define IFTODT(mode) (((mode) & S_IFMT) >> 12) 81 82 /* 83 * The directory entry has the type (currently unused on Solaris) in the 84 * top 4 bits, and the object number in the low 48 bits. The "middle" 85 * 12 bits are unused. 86 */ 87 #define ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4) 88 #define ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48) 89 90 /* 91 * This is the persistent portion of the znode. It is stored 92 * in the "bonus buffer" of the file. Short symbolic links 93 * are also stored in the bonus buffer. 94 */ 95 typedef struct znode_phys { 96 uint64_t zp_atime[2]; /* 0 - last file access time */ 97 uint64_t zp_mtime[2]; /* 16 - last file modification time */ 98 uint64_t zp_ctime[2]; /* 32 - last file change time */ 99 uint64_t zp_crtime[2]; /* 48 - creation time */ 100 uint64_t zp_gen; /* 64 - generation (txg of creation) */ 101 uint64_t zp_mode; /* 72 - file mode bits */ 102 uint64_t zp_size; /* 80 - size of file */ 103 uint64_t zp_parent; /* 88 - directory parent (`..') */ 104 uint64_t zp_links; /* 96 - number of links to file */ 105 uint64_t zp_xattr; /* 104 - DMU object for xattrs */ 106 uint64_t zp_rdev; /* 112 - dev_t for VBLK & VCHR files */ 107 uint64_t zp_flags; /* 120 - persistent flags */ 108 uint64_t zp_uid; /* 128 - file owner */ 109 uint64_t zp_gid; /* 136 - owning group */ 110 uint64_t zp_pad[4]; /* 144 - future */ 111 zfs_znode_acl_t zp_acl; /* 176 - 263 ACL */ 112 /* 113 * Data may pad out any remaining bytes in the znode buffer, eg: 114 * 115 * |<---------------------- dnode_phys (512) ------------------------>| 116 * |<-- dnode (192) --->|<----------- "bonus" buffer (320) ---------->| 117 * |<---- znode (264) ---->|<---- data (56) ---->| 118 * 119 * At present, we only use this space to store symbolic links. 120 */ 121 } znode_phys_t; 122 123 /* 124 * Directory entry locks control access to directory entries. 125 * They are used to protect creates, deletes, and renames. 126 * Each directory znode has a mutex and a list of locked names. 127 */ 128 #ifdef _KERNEL 129 typedef struct zfs_dirlock { 130 char *dl_name; /* directory entry being locked */ 131 uint32_t dl_sharecnt; /* 0 if exclusive, > 0 if shared */ 132 uint16_t dl_namesize; /* set if dl_name was allocated */ 133 kcondvar_t dl_cv; /* wait for entry to be unlocked */ 134 struct znode *dl_dzp; /* directory znode */ 135 struct zfs_dirlock *dl_next; /* next in z_dirlocks list */ 136 } zfs_dirlock_t; 137 138 typedef struct znode { 139 struct zfsvfs *z_zfsvfs; 140 vnode_t *z_vnode; 141 uint64_t z_id; /* object ID for this znode */ 142 kmutex_t z_lock; /* znode modification lock */ 143 krwlock_t z_map_lock; /* page map lock */ 144 krwlock_t z_parent_lock; /* parent lock for directories */ 145 krwlock_t z_name_lock; /* "master" lock for dirent locks */ 146 zfs_dirlock_t *z_dirlocks; /* directory entry lock list */ 147 kmutex_t z_range_lock; /* protects changes to z_range_avl */ 148 avl_tree_t z_range_avl; /* avl tree of file range locks */ 149 uint8_t z_unlinked; /* file has been unlinked */ 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_blksz; /* block size in bytes */ 154 uint_t z_seq; /* modification sequence number */ 155 uint64_t z_mapcnt; /* number of pages mapped to file */ 156 uint64_t z_last_itx; /* last ZIL itx on this znode */ 157 uint64_t z_gen; /* generation (same as zp_gen) */ 158 uint32_t z_sync_cnt; /* synchronous open count */ 159 kmutex_t z_acl_lock; /* acl data lock */ 160 list_node_t z_link_node; /* all znodes in fs link */ 161 /* 162 * These are dmu managed fields. 163 */ 164 znode_phys_t *z_phys; /* pointer to persistent znode */ 165 dmu_buf_t *z_dbuf; /* buffer containing the z_phys */ 166 } znode_t; 167 168 169 /* 170 * Range locking rules 171 * -------------------- 172 * 1. When truncating a file (zfs_create, zfs_setattr, zfs_space) the whole 173 * file range needs to be locked as RL_WRITER. Only then can the pages be 174 * freed etc and zp_size reset. zp_size must be set within range lock. 175 * 2. For writes and punching holes (zfs_write & zfs_space) just the range 176 * being written or freed needs to be locked as RL_WRITER. 177 * Multiple writes at the end of the file must coordinate zp_size updates 178 * to ensure data isn't lost. A compare and swap loop is currently used 179 * to ensure the file size is at least the offset last written. 180 * 3. For reads (zfs_read, zfs_get_data & zfs_putapage) just the range being 181 * read needs to be locked as RL_READER. A check against zp_size can then 182 * be made for reading beyond end of file. 183 */ 184 185 /* 186 * Convert between znode pointers and vnode pointers 187 */ 188 #define ZTOV(ZP) ((ZP)->z_vnode) 189 #define VTOZ(VP) ((znode_t *)(VP)->v_data) 190 191 /* 192 * ZFS_ENTER() is called on entry to each ZFS vnode and vfs operation. 193 * ZFS_EXIT() must be called before exitting the vop. 194 * ZFS_ENTER_VERIFY_ZP() does ZFS_ENTER plus verifies the znode is valid. 195 */ 196 #define ZFS_ENTER(zfsvfs) \ 197 { \ 198 rrw_enter(&(zfsvfs)->z_teardown_lock, RW_READER, FTAG); \ 199 if ((zfsvfs)->z_unmounted) { \ 200 ZFS_EXIT(zfsvfs); \ 201 return (EIO); \ 202 } \ 203 } 204 205 #define ZFS_EXIT(zfsvfs) rrw_exit(&(zfsvfs)->z_teardown_lock, FTAG) 206 207 #define ZFS_ENTER_VERIFY_ZP(zfsvfs, zp) \ 208 { \ 209 ZFS_ENTER((zfsvfs)); \ 210 if (!(zp)->z_dbuf_held) { \ 211 ZFS_EXIT(zfsvfs); \ 212 return (EIO); \ 213 } \ 214 } 215 216 /* 217 * Macros for dealing with dmu_buf_hold 218 */ 219 #define ZFS_OBJ_HASH(obj_num) (obj_num & (ZFS_OBJ_MTX_SZ - 1)) 220 #define ZFS_OBJ_MUTEX(zp) \ 221 (&zp->z_zfsvfs->z_hold_mtx[ZFS_OBJ_HASH(zp->z_id)]) 222 #define ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \ 223 mutex_enter(&zfsvfs->z_hold_mtx[ZFS_OBJ_HASH(obj_num)]); 224 225 #define ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \ 226 mutex_exit(&zfsvfs->z_hold_mtx[ZFS_OBJ_HASH(obj_num)]) 227 228 /* 229 * Macros to encode/decode ZFS stored time values from/to struct timespec 230 */ 231 #define ZFS_TIME_ENCODE(tp, stmp) \ 232 { \ 233 stmp[0] = (uint64_t)(tp)->tv_sec; \ 234 stmp[1] = (uint64_t)(tp)->tv_nsec; \ 235 } 236 237 #define ZFS_TIME_DECODE(tp, stmp) \ 238 { \ 239 (tp)->tv_sec = (time_t)stmp[0]; \ 240 (tp)->tv_nsec = (long)stmp[1]; \ 241 } 242 243 /* 244 * Timestamp defines 245 */ 246 #define ACCESSED (AT_ATIME) 247 #define STATE_CHANGED (AT_CTIME) 248 #define CONTENT_MODIFIED (AT_MTIME | AT_CTIME) 249 250 #define ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \ 251 if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \ 252 zfs_time_stamper(zp, ACCESSED, NULL) 253 254 extern int zfs_init_fs(zfsvfs_t *, znode_t **, cred_t *); 255 extern void zfs_set_dataprop(objset_t *); 256 extern void zfs_create_fs(objset_t *os, cred_t *cr, uint64_t, dmu_tx_t *tx); 257 extern void zfs_time_stamper(znode_t *, uint_t, dmu_tx_t *); 258 extern void zfs_time_stamper_locked(znode_t *, uint_t, dmu_tx_t *); 259 extern void zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *); 260 extern int zfs_freesp(znode_t *, uint64_t, uint64_t, int, boolean_t); 261 extern void zfs_znode_init(void); 262 extern void zfs_znode_fini(void); 263 extern int zfs_zget(zfsvfs_t *, uint64_t, znode_t **); 264 extern int zfs_rezget(znode_t *); 265 extern void zfs_zinactive(znode_t *); 266 extern void zfs_znode_delete(znode_t *, dmu_tx_t *); 267 extern void zfs_znode_free(znode_t *); 268 extern void zfs_remove_op_tables(); 269 extern int zfs_create_op_tables(); 270 extern int zfs_sync(vfs_t *vfsp, short flag, cred_t *cr); 271 extern dev_t zfs_cmpldev(uint64_t); 272 extern int zfs_get_version(objset_t *os, uint64_t *version); 273 extern int zfs_set_version(const char *name, uint64_t newvers); 274 275 extern void zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, int txtype, 276 znode_t *dzp, znode_t *zp, char *name); 277 extern void zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, int txtype, 278 znode_t *dzp, char *name); 279 extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, int txtype, 280 znode_t *dzp, znode_t *zp, char *name); 281 extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, int txtype, 282 znode_t *dzp, znode_t *zp, char *name, char *link); 283 extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, int txtype, 284 znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp); 285 extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype, 286 znode_t *zp, offset_t off, ssize_t len, int ioflag); 287 extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype, 288 znode_t *zp, uint64_t off, uint64_t len); 289 extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype, 290 znode_t *zp, vattr_t *vap, uint_t mask_applied); 291 extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, int txtype, 292 znode_t *zp, int aclcnt, ace_t *z_ace); 293 294 extern zil_get_data_t zfs_get_data; 295 extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE]; 296 extern int zfsfstype; 297 298 #endif /* _KERNEL */ 299 300 extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len); 301 302 #ifdef __cplusplus 303 } 304 #endif 305 306 #endif /* _SYS_FS_ZFS_ZNODE_H */ 307