zfs_znode.h revision 297112
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 * 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/list.h> 32#include <sys/dmu.h> 33#include <sys/sa.h> 34#include <sys/zfs_vfsops.h> 35#include <sys/rrwlock.h> 36#include <sys/zfs_sa.h> 37#include <sys/zfs_stat.h> 38#endif 39#include <sys/zfs_acl.h> 40#include <sys/zil.h> 41 42#ifdef __cplusplus 43extern "C" { 44#endif 45 46/* 47 * Additional file level attributes, that are stored 48 * in the upper half of zp_flags 49 */ 50#define ZFS_READONLY 0x0000000100000000 51#define ZFS_HIDDEN 0x0000000200000000 52#define ZFS_SYSTEM 0x0000000400000000 53#define ZFS_ARCHIVE 0x0000000800000000 54#define ZFS_IMMUTABLE 0x0000001000000000 55#define ZFS_NOUNLINK 0x0000002000000000 56#define ZFS_APPENDONLY 0x0000004000000000 57#define ZFS_NODUMP 0x0000008000000000 58#define ZFS_OPAQUE 0x0000010000000000 59#define ZFS_AV_QUARANTINED 0x0000020000000000 60#define ZFS_AV_MODIFIED 0x0000040000000000 61#define ZFS_REPARSE 0x0000080000000000 62#define ZFS_OFFLINE 0x0000100000000000 63#define ZFS_SPARSE 0x0000200000000000 64 65#define ZFS_ATTR_SET(zp, attr, value, pflags, tx) \ 66{ \ 67 if (value) \ 68 pflags |= attr; \ 69 else \ 70 pflags &= ~attr; \ 71 VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_FLAGS(zp->z_zfsvfs), \ 72 &pflags, sizeof (pflags), tx)); \ 73} 74 75/* 76 * Define special zfs pflags 77 */ 78#define ZFS_XATTR 0x1 /* is an extended attribute */ 79#define ZFS_INHERIT_ACE 0x2 /* ace has inheritable ACEs */ 80#define ZFS_ACL_TRIVIAL 0x4 /* files ACL is trivial */ 81#define ZFS_ACL_OBJ_ACE 0x8 /* ACL has CMPLX Object ACE */ 82#define ZFS_ACL_PROTECTED 0x10 /* ACL protected */ 83#define ZFS_ACL_DEFAULTED 0x20 /* ACL should be defaulted */ 84#define ZFS_ACL_AUTO_INHERIT 0x40 /* ACL should be inherited */ 85#define ZFS_BONUS_SCANSTAMP 0x80 /* Scanstamp in bonus area */ 86#define ZFS_NO_EXECS_DENIED 0x100 /* exec was given to everyone */ 87 88#define SA_ZPL_ATIME(z) z->z_attr_table[ZPL_ATIME] 89#define SA_ZPL_MTIME(z) z->z_attr_table[ZPL_MTIME] 90#define SA_ZPL_CTIME(z) z->z_attr_table[ZPL_CTIME] 91#define SA_ZPL_CRTIME(z) z->z_attr_table[ZPL_CRTIME] 92#define SA_ZPL_GEN(z) z->z_attr_table[ZPL_GEN] 93#define SA_ZPL_DACL_ACES(z) z->z_attr_table[ZPL_DACL_ACES] 94#define SA_ZPL_XATTR(z) z->z_attr_table[ZPL_XATTR] 95#define SA_ZPL_SYMLINK(z) z->z_attr_table[ZPL_SYMLINK] 96#define SA_ZPL_RDEV(z) z->z_attr_table[ZPL_RDEV] 97#define SA_ZPL_SCANSTAMP(z) z->z_attr_table[ZPL_SCANSTAMP] 98#define SA_ZPL_UID(z) z->z_attr_table[ZPL_UID] 99#define SA_ZPL_GID(z) z->z_attr_table[ZPL_GID] 100#define SA_ZPL_PARENT(z) z->z_attr_table[ZPL_PARENT] 101#define SA_ZPL_LINKS(z) z->z_attr_table[ZPL_LINKS] 102#define SA_ZPL_MODE(z) z->z_attr_table[ZPL_MODE] 103#define SA_ZPL_DACL_COUNT(z) z->z_attr_table[ZPL_DACL_COUNT] 104#define SA_ZPL_FLAGS(z) z->z_attr_table[ZPL_FLAGS] 105#define SA_ZPL_SIZE(z) z->z_attr_table[ZPL_SIZE] 106#define SA_ZPL_ZNODE_ACL(z) z->z_attr_table[ZPL_ZNODE_ACL] 107#define SA_ZPL_PAD(z) z->z_attr_table[ZPL_PAD] 108 109/* 110 * Is ID ephemeral? 111 */ 112#define IS_EPHEMERAL(x) (x > MAXUID) 113 114/* 115 * Should we use FUIDs? 116 */ 117#define USE_FUIDS(version, os) (version >= ZPL_VERSION_FUID && \ 118 spa_version(dmu_objset_spa(os)) >= SPA_VERSION_FUID) 119#define USE_SA(version, os) (version >= ZPL_VERSION_SA && \ 120 spa_version(dmu_objset_spa(os)) >= SPA_VERSION_SA) 121 122#define MASTER_NODE_OBJ 1 123 124/* 125 * Special attributes for master node. 126 * "userquota@" and "groupquota@" are also valid (from 127 * zfs_userquota_prop_prefixes[]). 128 */ 129#define ZFS_FSID "FSID" 130#define ZFS_UNLINKED_SET "DELETE_QUEUE" 131#define ZFS_ROOT_OBJ "ROOT" 132#define ZPL_VERSION_STR "VERSION" 133#define ZFS_FUID_TABLES "FUID" 134#define ZFS_SHARES_DIR "SHARES" 135#define ZFS_SA_ATTRS "SA_ATTRS" 136 137/* 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#ifndef IFTODT 153#define IFTODT(mode) (((mode) & S_IFMT) >> 12) 154#endif 155 156/* 157 * The directory entry has the type (currently unused on Solaris) in the 158 * top 4 bits, and the object number in the low 48 bits. The "middle" 159 * 12 bits are unused. 160 */ 161#define ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4) 162#define ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48) 163 164/* 165 * Directory entry locks control access to directory entries. 166 * They are used to protect creates, deletes, and renames. 167 * Each directory znode has a mutex and a list of locked names. 168 */ 169#ifdef _KERNEL 170typedef struct zfs_dirlock { 171 char *dl_name; /* directory entry being locked */ 172 uint32_t dl_sharecnt; /* 0 if exclusive, > 0 if shared */ 173 uint8_t dl_namelock; /* 1 if z_name_lock is NOT held */ 174 uint16_t dl_namesize; /* set if dl_name was allocated */ 175 kcondvar_t dl_cv; /* wait for entry to be unlocked */ 176 struct znode *dl_dzp; /* directory znode */ 177 struct zfs_dirlock *dl_next; /* next in z_dirlocks list */ 178} zfs_dirlock_t; 179 180typedef struct znode { 181 struct zfsvfs *z_zfsvfs; 182 vnode_t *z_vnode; 183 uint64_t z_id; /* object ID for this znode */ 184 kmutex_t z_lock; /* znode modification lock */ 185 krwlock_t z_parent_lock; /* parent lock for directories */ 186 krwlock_t z_name_lock; /* "master" lock for dirent locks */ 187 zfs_dirlock_t *z_dirlocks; /* directory entry lock list */ 188 kmutex_t z_range_lock; /* protects changes to z_range_avl */ 189 avl_tree_t z_range_avl; /* avl tree of file range locks */ 190 uint8_t z_unlinked; /* file has been unlinked */ 191 uint8_t z_atime_dirty; /* atime needs to be synced */ 192 uint8_t z_zn_prefetch; /* Prefetch znodes? */ 193 uint8_t z_moved; /* Has this znode been moved? */ 194 uint_t z_blksz; /* block size in bytes */ 195 uint_t z_seq; /* modification sequence number */ 196 uint64_t z_mapcnt; /* number of pages mapped to file */ 197 uint64_t z_gen; /* generation (cached) */ 198 uint64_t z_size; /* file size (cached) */ 199 uint64_t z_atime[2]; /* atime (cached) */ 200 uint64_t z_links; /* file links (cached) */ 201 uint64_t z_pflags; /* pflags (cached) */ 202 uint64_t z_uid; /* uid fuid (cached) */ 203 uint64_t z_gid; /* gid fuid (cached) */ 204 mode_t z_mode; /* mode (cached) */ 205 uint32_t z_sync_cnt; /* synchronous open count */ 206 kmutex_t z_acl_lock; /* acl data lock */ 207 zfs_acl_t *z_acl_cached; /* cached acl */ 208 list_node_t z_link_node; /* all znodes in fs link */ 209 sa_handle_t *z_sa_hdl; /* handle to sa data */ 210 boolean_t z_is_sa; /* are we native sa? */ 211} znode_t; 212 213 214/* 215 * Range locking rules 216 * -------------------- 217 * 1. When truncating a file (zfs_create, zfs_setattr, zfs_space) the whole 218 * file range needs to be locked as RL_WRITER. Only then can the pages be 219 * freed etc and zp_size reset. zp_size must be set within range lock. 220 * 2. For writes and punching holes (zfs_write & zfs_space) just the range 221 * being written or freed needs to be locked as RL_WRITER. 222 * Multiple writes at the end of the file must coordinate zp_size updates 223 * to ensure data isn't lost. A compare and swap loop is currently used 224 * to ensure the file size is at least the offset last written. 225 * 3. For reads (zfs_read, zfs_get_data & zfs_putapage) just the range being 226 * read needs to be locked as RL_READER. A check against zp_size can then 227 * be made for reading beyond end of file. 228 */ 229 230/* 231 * Convert between znode pointers and vnode pointers 232 */ 233#ifdef DEBUG 234static __inline vnode_t * 235ZTOV(znode_t *zp) 236{ 237 vnode_t *vp = zp->z_vnode; 238 239 ASSERT(vp == NULL || vp->v_data == NULL || vp->v_data == zp); 240 return (vp); 241} 242static __inline znode_t * 243VTOZ(vnode_t *vp) 244{ 245 znode_t *zp = (znode_t *)vp->v_data; 246 247 ASSERT(zp == NULL || zp->z_vnode == NULL || zp->z_vnode == vp); 248 return (zp); 249} 250#else 251#define ZTOV(ZP) ((ZP)->z_vnode) 252#define VTOZ(VP) ((znode_t *)(VP)->v_data) 253#endif 254 255/* Called on entry to each ZFS vnode and vfs operation */ 256#define ZFS_ENTER(zfsvfs) \ 257 { \ 258 rrm_enter_read(&(zfsvfs)->z_teardown_lock, FTAG); \ 259 if ((zfsvfs)->z_unmounted) { \ 260 ZFS_EXIT(zfsvfs); \ 261 return (EIO); \ 262 } \ 263 } 264 265/* Must be called before exiting the vop */ 266#define ZFS_EXIT(zfsvfs) rrm_exit(&(zfsvfs)->z_teardown_lock, FTAG) 267 268/* Verifies the znode is valid */ 269#define ZFS_VERIFY_ZP(zp) \ 270 if ((zp)->z_sa_hdl == NULL) { \ 271 ZFS_EXIT((zp)->z_zfsvfs); \ 272 return (EIO); \ 273 } \ 274 275/* 276 * Macros for dealing with dmu_buf_hold 277 */ 278#define ZFS_OBJ_HASH(obj_num) ((obj_num) & (ZFS_OBJ_MTX_SZ - 1)) 279#define ZFS_OBJ_MUTEX(zfsvfs, obj_num) \ 280 (&(zfsvfs)->z_hold_mtx[ZFS_OBJ_HASH(obj_num)]) 281#define ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num) \ 282 mutex_enter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num))) 283#define ZFS_OBJ_HOLD_TRYENTER(zfsvfs, obj_num) \ 284 mutex_tryenter(ZFS_OBJ_MUTEX((zfsvfs), (obj_num))) 285#define ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num) \ 286 mutex_exit(ZFS_OBJ_MUTEX((zfsvfs), (obj_num))) 287 288/* Encode ZFS stored time values from a struct timespec */ 289#define ZFS_TIME_ENCODE(tp, stmp) \ 290{ \ 291 (stmp)[0] = (uint64_t)(tp)->tv_sec; \ 292 (stmp)[1] = (uint64_t)(tp)->tv_nsec; \ 293} 294 295/* Decode ZFS stored time values to a struct timespec */ 296#define ZFS_TIME_DECODE(tp, stmp) \ 297{ \ 298 (tp)->tv_sec = (time_t)(stmp)[0]; \ 299 (tp)->tv_nsec = (long)(stmp)[1]; \ 300} 301 302/* 303 * Timestamp defines 304 */ 305#define ACCESSED (AT_ATIME) 306#define STATE_CHANGED (AT_CTIME) 307#define CONTENT_MODIFIED (AT_MTIME | AT_CTIME) 308 309#define ZFS_ACCESSTIME_STAMP(zfsvfs, zp) \ 310 if ((zfsvfs)->z_atime && !((zfsvfs)->z_vfs->vfs_flag & VFS_RDONLY)) \ 311 zfs_tstamp_update_setup(zp, ACCESSED, NULL, NULL, B_FALSE); 312 313extern int zfs_init_fs(zfsvfs_t *, znode_t **); 314extern void zfs_set_dataprop(objset_t *); 315extern void zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *, 316 dmu_tx_t *tx); 317extern void zfs_tstamp_update_setup(znode_t *, uint_t, uint64_t [2], 318 uint64_t [2], boolean_t); 319extern void zfs_grow_blocksize(znode_t *, uint64_t, dmu_tx_t *); 320extern int zfs_freesp(znode_t *, uint64_t, uint64_t, int, boolean_t); 321extern void zfs_znode_init(void); 322extern void zfs_znode_fini(void); 323extern int zfs_zget(zfsvfs_t *, uint64_t, znode_t **); 324extern int zfs_rezget(znode_t *); 325extern void zfs_zinactive(znode_t *); 326extern void zfs_znode_delete(znode_t *, dmu_tx_t *); 327extern void zfs_znode_free(znode_t *); 328extern void zfs_remove_op_tables(); 329extern int zfs_create_op_tables(); 330extern dev_t zfs_cmpldev(uint64_t); 331extern int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value); 332extern int zfs_get_stats(objset_t *os, nvlist_t *nv); 333extern void zfs_znode_dmu_fini(znode_t *); 334 335extern void zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 336 znode_t *dzp, znode_t *zp, char *name, vsecattr_t *, zfs_fuid_info_t *, 337 vattr_t *vap); 338extern int zfs_log_create_txtype(zil_create_t, vsecattr_t *vsecp, 339 vattr_t *vap); 340extern void zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 341 znode_t *dzp, char *name, uint64_t foid); 342#define ZFS_NO_OBJECT 0 /* no object id */ 343extern void zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 344 znode_t *dzp, znode_t *zp, char *name); 345extern void zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 346 znode_t *dzp, znode_t *zp, char *name, char *link); 347extern void zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype, 348 znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp); 349extern void zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype, 350 znode_t *zp, offset_t off, ssize_t len, int ioflag); 351extern void zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype, 352 znode_t *zp, uint64_t off, uint64_t len); 353extern void zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype, 354 znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp); 355#ifndef ZFS_NO_ACL 356extern void zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp, 357 vsecattr_t *vsecp, zfs_fuid_info_t *fuidp); 358#endif 359extern void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx); 360extern void zfs_upgrade(zfsvfs_t *zfsvfs, dmu_tx_t *tx); 361extern int zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx); 362 363extern zil_get_data_t zfs_get_data; 364extern zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE]; 365extern int zfsfstype; 366 367#endif /* _KERNEL */ 368 369extern int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len); 370 371#ifdef __cplusplus 372} 373#endif 374 375#endif /* _SYS_FS_ZFS_ZNODE_H */ 376