zfs_znode.h revision 276081
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 by Delphix. All rights reserved. 24 */ 25 26#ifndef _SYS_FS_ZFS_ZNODE_H 27#define _SYS_FS_ZFS_ZNODE_H 28 29#ifdef _KERNEL 30#include <sys/list.h> 31#include <sys/dmu.h> 32#include <sys/sa.h> 33#include <sys/zfs_vfsops.h> 34#include <sys/rrwlock.h> 35#include <sys/zfs_sa.h> 36#include <sys/zfs_stat.h> 37#endif 38#include <sys/zfs_acl.h> 39#include <sys/zil.h> 40 41#ifdef __cplusplus 42extern "C" { 43#endif 44 45/* 46 * Additional file level attributes, that are stored 47 * in the upper half of zp_flags 48 */ 49#define ZFS_READONLY 0x0000000100000000 50#define ZFS_HIDDEN 0x0000000200000000 51#define ZFS_SYSTEM 0x0000000400000000 52#define ZFS_ARCHIVE 0x0000000800000000 53#define ZFS_IMMUTABLE 0x0000001000000000 54#define ZFS_NOUNLINK 0x0000002000000000 55#define ZFS_APPENDONLY 0x0000004000000000 56#define ZFS_NODUMP 0x0000008000000000 57#define ZFS_OPAQUE 0x0000010000000000 58#define ZFS_AV_QUARANTINED 0x0000020000000000 59#define ZFS_AV_MODIFIED 0x0000040000000000 60#define ZFS_REPARSE 0x0000080000000000 61#define ZFS_OFFLINE 0x0000100000000000 62#define ZFS_SPARSE 0x0000200000000000 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/* 137 * Path component length 138 * 139 * The generic fs code uses MAXNAMELEN to represent 140 * what the largest component length is. Unfortunately, 141 * this length includes the terminating NULL. ZFS needs 142 * to tell the users via pathconf() and statvfs() what the 143 * true maximum length of a component is, excluding the NULL. 144 */ 145#define ZFS_MAXNAMELEN (MAXNAMELEN - 1) 146 147/* 148 * Convert mode bits (zp_mode) to BSD-style DT_* values for storing in 149 * the directory entries. 150 */ 151#ifndef IFTODT 152#define IFTODT(mode) (((mode) & S_IFMT) >> 12) 153#endif 154 155/* 156 * The directory entry has the type (currently unused on Solaris) in the 157 * top 4 bits, and the object number in the low 48 bits. The "middle" 158 * 12 bits are unused. 159 */ 160#define ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4) 161#define ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48) 162 163/* 164 * Directory entry locks control access to directory entries. 165 * They are used to protect creates, deletes, and renames. 166 * Each directory znode has a mutex and a list of locked names. 167 */ 168#ifdef _KERNEL 169typedef struct zfs_dirlock { 170 char *dl_name; /* directory entry being locked */ 171 uint32_t dl_sharecnt; /* 0 if exclusive, > 0 if shared */ 172 uint8_t dl_namelock; /* 1 if z_name_lock is NOT held */ 173 uint16_t dl_namesize; /* set if dl_name was allocated */ 174 kcondvar_t dl_cv; /* wait for entry to be unlocked */ 175 struct znode *dl_dzp; /* directory znode */ 176 struct zfs_dirlock *dl_next; /* next in z_dirlocks list */ 177} zfs_dirlock_t; 178 179typedef struct znode { 180 struct zfsvfs *z_zfsvfs; 181 vnode_t *z_vnode; 182 uint64_t z_id; /* object ID for this znode */ 183 kmutex_t z_lock; /* znode modification lock */ 184 krwlock_t z_parent_lock; /* parent lock for directories */ 185 krwlock_t z_name_lock; /* "master" lock for dirent locks */ 186 zfs_dirlock_t *z_dirlocks; /* directory entry lock list */ 187 kmutex_t z_range_lock; /* protects changes to z_range_avl */ 188 avl_tree_t z_range_avl; /* avl tree of file range locks */ 189 uint8_t z_unlinked; /* file has been unlinked */ 190 uint8_t z_atime_dirty; /* atime needs to be synced */ 191 uint8_t z_zn_prefetch; /* Prefetch znodes? */ 192 uint8_t z_moved; /* Has this znode been moved? */ 193 uint_t z_blksz; /* block size in bytes */ 194 uint_t z_seq; /* modification sequence number */ 195 uint64_t z_mapcnt; /* number of pages mapped to file */ 196 uint64_t z_gen; /* generation (cached) */ 197 uint64_t z_size; /* file size (cached) */ 198 uint64_t z_atime[2]; /* atime (cached) */ 199 uint64_t z_links; /* file links (cached) */ 200 uint64_t z_pflags; /* pflags (cached) */ 201 uint64_t z_uid; /* uid fuid (cached) */ 202 uint64_t z_gid; /* gid fuid (cached) */ 203 mode_t z_mode; /* mode (cached) */ 204 uint32_t z_sync_cnt; /* synchronous open count */ 205 kmutex_t z_acl_lock; /* acl data lock */ 206 zfs_acl_t *z_acl_cached; /* cached acl */ 207 list_node_t z_link_node; /* all znodes in fs link */ 208 sa_handle_t *z_sa_hdl; /* handle to sa data */ 209 boolean_t z_is_sa; /* are we native sa? */ 210} znode_t; 211 212 213/* 214 * Range locking rules 215 * -------------------- 216 * 1. When truncating a file (zfs_create, zfs_setattr, zfs_space) the whole 217 * file range needs to be locked as RL_WRITER. Only then can the pages be 218 * freed etc and zp_size reset. zp_size must be set within range lock. 219 * 2. For writes and punching holes (zfs_write & zfs_space) just the range 220 * being written or freed needs to be locked as RL_WRITER. 221 * Multiple writes at the end of the file must coordinate zp_size updates 222 * to ensure data isn't lost. A compare and swap loop is currently used 223 * to ensure the file size is at least the offset last written. 224 * 3. For reads (zfs_read, zfs_get_data & zfs_putapage) just the range being 225 * read needs to be locked as RL_READER. A check against zp_size can then 226 * be made for reading beyond end of file. 227 */ 228 229/* 230 * Convert between znode pointers and vnode pointers 231 */ 232#ifdef DEBUG 233static __inline vnode_t * 234ZTOV(znode_t *zp) 235{ 236 vnode_t *vp = zp->z_vnode; 237 238 ASSERT(vp == NULL || vp->v_data == NULL || vp->v_data == zp); 239 return (vp); 240} 241static __inline znode_t * 242VTOZ(vnode_t *vp) 243{ 244 znode_t *zp = (znode_t *)vp->v_data; 245 246 ASSERT(zp == NULL || zp->z_vnode == NULL || zp->z_vnode == vp); 247 return (zp); 248} 249#else 250#define ZTOV(ZP) ((ZP)->z_vnode) 251#define VTOZ(VP) ((znode_t *)(VP)->v_data) 252#endif 253 254/* Called on entry to each ZFS vnode and vfs operation */ 255#define ZFS_ENTER(zfsvfs) \ 256 { \ 257 rrm_enter_read(&(zfsvfs)->z_teardown_lock, FTAG); \ 258 if ((zfsvfs)->z_unmounted) { \ 259 ZFS_EXIT(zfsvfs); \ 260 return (EIO); \ 261 } \ 262 } 263 264/* Called on entry to each ZFS vnode and vfs operation that can not return EIO */ 265#define ZFS_ENTER_NOERROR(zfsvfs) \ 266 rrm_enter(&(zfsvfs)->z_teardown_lock, RW_READER, FTAG) 267 268/* Must be called before exiting the vop */ 269#define ZFS_EXIT(zfsvfs) rrm_exit(&(zfsvfs)->z_teardown_lock, FTAG) 270 271/* Verifies the znode is valid */ 272#define ZFS_VERIFY_ZP(zp) \ 273 if ((zp)->z_sa_hdl == NULL) { \ 274 ZFS_EXIT((zp)->z_zfsvfs); \ 275 return (EIO); \ 276 } \ 277 278/* 279 * Macros for dealing with dmu_buf_hold 280 */ 281#define ZFS_OBJ_HASH(obj_num) ((obj_num) & (ZFS_OBJ_MTX_SZ - 1)) 282#define ZFS_OBJ_MUTEX(zfsvfs, obj_num) \ 283 (&(zfsvfs)->z_hold_mtx[ZFS_OBJ_HASH(obj_num)]) 284#define ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \ 285 mutex_enter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num))) 286#define ZFS_OBJ_HOLD_TRYENTER(zfsvfs, obj_num) \ 287 mutex_tryenter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num))) 288#define ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \ 289 mutex_exit(ZFS_OBJ_MUTEX((zfsvfs), (obj_num))) 290 291/* Encode ZFS stored time values from a struct timespec */ 292#define ZFS_TIME_ENCODE(tp, stmp) \ 293{ \ 294 (stmp)[0] = (uint64_t)(tp)->tv_sec; \ 295 (stmp)[1] = (uint64_t)(tp)->tv_nsec; \ 296} 297 298/* Decode ZFS stored time values to a struct timespec */ 299#define ZFS_TIME_DECODE(tp, stmp) \ 300{ \ 301 (tp)->tv_sec = (time_t)(stmp)[0]; \ 302 (tp)->tv_nsec = (long)(stmp)[1]; \ 303} 304 305/* 306 * Timestamp defines 307 */ 308#define ACCESSED (AT_ATIME) 309#define STATE_CHANGED (AT_CTIME) 310#define CONTENT_MODIFIED (AT_MTIME | AT_CTIME) 311 312#define ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \ 313 if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \ 314 zfs_tstamp_update_setup(zp, ACCESSED, NULL, NULL, B_FALSE); 315 316extern int zfs_init_fs(zfsvfs_t *, znode_t **); 317extern void zfs_set_dataprop(objset_t *); 318extern void zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *, 319 dmu_tx_t *tx); 320extern void zfs_tstamp_update_setup(znode_t *, uint_t, uint64_t [2], 321 uint64_t [2], boolean_t); 322extern void zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *); 323extern int zfs_freesp(znode_t *, uint64_t, uint64_t, int, boolean_t); 324extern void zfs_znode_init(void); 325extern void zfs_znode_fini(void); 326extern int zfs_zget(zfsvfs_t *, uint64_t, znode_t **); 327extern int zfs_rezget(znode_t *); 328extern void zfs_zinactive(znode_t *); 329extern void zfs_znode_delete(znode_t *, dmu_tx_t *); 330extern void zfs_znode_free(znode_t *); 331extern void zfs_remove_op_tables(); 332extern int zfs_create_op_tables(); 333extern dev_t zfs_cmpldev(uint64_t); 334extern int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value); 335extern int zfs_get_stats(objset_t *os, nvlist_t *nv); 336extern void zfs_znode_dmu_fini(znode_t *); 337 338extern void zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 339 znode_t *dzp, znode_t *zp, char *name, vsecattr_t *, zfs_fuid_info_t *, 340 vattr_t *vap); 341extern int zfs_log_create_txtype(zil_create_t, vsecattr_t *vsecp, 342 vattr_t *vap); 343extern void zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 344 znode_t *dzp, char *name, uint64_t foid); 345#define ZFS_NO_OBJECT 0 /* no object id */ 346extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 347 znode_t *dzp, znode_t *zp, char *name); 348extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 349 znode_t *dzp, znode_t *zp, char *name, char *link); 350extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 351 znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp); 352extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype, 353 znode_t *zp, offset_t off, ssize_t len, int ioflag); 354extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype, 355 znode_t *zp, uint64_t off, uint64_t len); 356extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype, 357 znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp); 358#ifndef ZFS_NO_ACL 359extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp, 360 vsecattr_t *vsecp, zfs_fuid_info_t *fuidp); 361#endif 362extern void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx); 363extern void zfs_upgrade(zfsvfs_t *zfsvfs, dmu_tx_t *tx); 364extern int zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx); 365 366extern zil_get_data_t zfs_get_data; 367extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE]; 368extern int zfsfstype; 369 370#endif /* _KERNEL */ 371 372extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len); 373 374#ifdef __cplusplus 375} 376#endif 377 378#endif /* _SYS_FS_ZFS_ZNODE_H */ 379