1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1989, 1991, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 #ifndef _SYS_MOUNT_H_ 33 #define _SYS_MOUNT_H_ 34 35 #include <sys/ucred.h> 36 #include <sys/queue.h> 37 #ifdef _KERNEL 38 #include <sys/types.h> 39 #include <sys/lock.h> 40 #include <sys/lockmgr.h> 41 #include <sys/tslog.h> 42 #include <sys/_mutex.h> 43 #include <sys/_sx.h> 44 #endif 45 46 /* 47 * NOTE: When changing statfs structure, mount structure, MNT_* flags or 48 * MNTK_* flags also update DDB show mount command in vfs_subr.c. 49 */ 50 51 typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */ 52 53 #define fsidcmp(a, b) memcmp((a), (b), sizeof(fsid_t)) 54 55 /* 56 * File identifier. 57 * These are unique per filesystem on a single machine. 58 * 59 * Note that the offset of fid_data is 4 bytes, so care must be taken to avoid 60 * undefined behavior accessing unaligned fields within an embedded struct. 61 */ 62 #define MAXFIDSZ 16 63 64 struct fid { 65 u_short fid_len; /* length of data in bytes */ 66 u_short fid_data0; /* force longword alignment */ 67 char fid_data[MAXFIDSZ]; /* data (variable length) */ 68 }; 69 70 /* 71 * filesystem statistics 72 */ 73 #define MFSNAMELEN 16 /* length of type name including null */ 74 #define MNAMELEN 1024 /* size of on/from name bufs */ 75 #define STATFS_VERSION 0x20140518 /* current version number */ 76 struct statfs { 77 uint32_t f_version; /* structure version number */ 78 uint32_t f_type; /* type of filesystem */ 79 uint64_t f_flags; /* copy of mount exported flags */ 80 uint64_t f_bsize; /* filesystem fragment size */ 81 uint64_t f_iosize; /* optimal transfer block size */ 82 uint64_t f_blocks; /* total data blocks in filesystem */ 83 uint64_t f_bfree; /* free blocks in filesystem */ 84 int64_t f_bavail; /* free blocks avail to non-superuser */ 85 uint64_t f_files; /* total file nodes in filesystem */ 86 int64_t f_ffree; /* free nodes avail to non-superuser */ 87 uint64_t f_syncwrites; /* count of sync writes since mount */ 88 uint64_t f_asyncwrites; /* count of async writes since mount */ 89 uint64_t f_syncreads; /* count of sync reads since mount */ 90 uint64_t f_asyncreads; /* count of async reads since mount */ 91 uint32_t f_nvnodelistsize; /* # of vnodes */ 92 uint32_t f_spare0; /* unused spare */ 93 uint64_t f_spare[9]; /* unused spare */ 94 uint32_t f_namemax; /* maximum filename length */ 95 uid_t f_owner; /* user that mounted the filesystem */ 96 fsid_t f_fsid; /* filesystem id */ 97 char f_charspare[80]; /* spare string space */ 98 char f_fstypename[MFSNAMELEN]; /* filesystem type name */ 99 char f_mntfromname[MNAMELEN]; /* mounted filesystem */ 100 char f_mntonname[MNAMELEN]; /* directory on which mounted */ 101 }; 102 103 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL) 104 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */ 105 struct freebsd11_statfs { 106 uint32_t f_version; /* structure version number */ 107 uint32_t f_type; /* type of filesystem */ 108 uint64_t f_flags; /* copy of mount exported flags */ 109 uint64_t f_bsize; /* filesystem fragment size */ 110 uint64_t f_iosize; /* optimal transfer block size */ 111 uint64_t f_blocks; /* total data blocks in filesystem */ 112 uint64_t f_bfree; /* free blocks in filesystem */ 113 int64_t f_bavail; /* free blocks avail to non-superuser */ 114 uint64_t f_files; /* total file nodes in filesystem */ 115 int64_t f_ffree; /* free nodes avail to non-superuser */ 116 uint64_t f_syncwrites; /* count of sync writes since mount */ 117 uint64_t f_asyncwrites; /* count of async writes since mount */ 118 uint64_t f_syncreads; /* count of sync reads since mount */ 119 uint64_t f_asyncreads; /* count of async reads since mount */ 120 uint64_t f_spare[10]; /* unused spare */ 121 uint32_t f_namemax; /* maximum filename length */ 122 uid_t f_owner; /* user that mounted the filesystem */ 123 fsid_t f_fsid; /* filesystem id */ 124 char f_charspare[80]; /* spare string space */ 125 char f_fstypename[16]; /* filesystem type name */ 126 char f_mntfromname[88]; /* mounted filesystem */ 127 char f_mntonname[88]; /* directory on which mounted */ 128 }; 129 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */ 130 131 #ifdef _KERNEL 132 #define OMFSNAMELEN 16 /* length of fs type name, including null */ 133 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */ 134 135 /* XXX getfsstat.2 is out of date with write and read counter changes here. */ 136 /* XXX statfs.2 is out of date with read counter changes here. */ 137 struct ostatfs { 138 long f_spare2; /* placeholder */ 139 long f_bsize; /* fundamental filesystem block size */ 140 long f_iosize; /* optimal transfer block size */ 141 long f_blocks; /* total data blocks in filesystem */ 142 long f_bfree; /* free blocks in fs */ 143 long f_bavail; /* free blocks avail to non-superuser */ 144 long f_files; /* total file nodes in filesystem */ 145 long f_ffree; /* free file nodes in fs */ 146 fsid_t f_fsid; /* filesystem id */ 147 uid_t f_owner; /* user that mounted the filesystem */ 148 int f_type; /* type of filesystem */ 149 int f_flags; /* copy of mount exported flags */ 150 long f_syncwrites; /* count of sync writes since mount */ 151 long f_asyncwrites; /* count of async writes since mount */ 152 char f_fstypename[OMFSNAMELEN]; /* fs type name */ 153 char f_mntonname[OMNAMELEN]; /* directory on which mounted */ 154 long f_syncreads; /* count of sync reads since mount */ 155 long f_asyncreads; /* count of async reads since mount */ 156 short f_spares1; /* unused spare */ 157 char f_mntfromname[OMNAMELEN];/* mounted filesystem */ 158 short f_spares2; /* unused spare */ 159 /* 160 * XXX on machines where longs are aligned to 8-byte boundaries, there 161 * is an unnamed int32_t here. This spare was after the apparent end 162 * of the struct until we bit off the read counters from f_mntonname. 163 */ 164 long f_spare[2]; /* unused spare */ 165 }; 166 #endif /* _KERNEL */ 167 168 #if defined(_WANT_MOUNT) || defined(_KERNEL) 169 TAILQ_HEAD(vnodelst, vnode); 170 171 /* Mount options list */ 172 TAILQ_HEAD(vfsoptlist, vfsopt); 173 struct vfsopt { 174 TAILQ_ENTRY(vfsopt) link; 175 char *name; 176 void *value; 177 int len; 178 int pos; 179 int seen; 180 }; 181 182 struct mount_pcpu { 183 int mntp_thread_in_ops; 184 int mntp_ref; 185 int mntp_lockref; 186 int mntp_writeopcount; 187 }; 188 189 _Static_assert(sizeof(struct mount_pcpu) == 16, 190 "the struct is allocated from pcpu 16 zone"); 191 192 /* 193 * Structure for tracking a stacked filesystem mounted above another 194 * filesystem. This is expected to be stored in the upper FS' per-mount data. 195 * 196 * Lock reference: 197 * i - lower mount interlock 198 * c - constant from node initialization 199 */ 200 struct mount_upper_node { 201 struct mount *mp; /* (c) mount object for upper FS */ 202 TAILQ_ENTRY(mount_upper_node) mnt_upper_link; /* (i) position in uppers list */ 203 }; 204 205 /* 206 * Structure per mounted filesystem. Each mounted filesystem has an 207 * array of operations and an instance record. The filesystems are 208 * put on a doubly linked list. 209 * 210 * Lock reference: 211 * l - mnt_listmtx 212 * m - mountlist_mtx 213 * i - interlock 214 * v - vnode freelist mutex 215 * d - deferred unmount list mutex 216 * e - mnt_explock 217 * 218 * Unmarked fields are considered stable as long as a ref is held. 219 * 220 */ 221 struct mount { 222 int mnt_vfs_ops; /* (i) pending vfs ops */ 223 int mnt_kern_flag; /* (i) kernel only flags */ 224 uint64_t mnt_flag; /* (i) flags shared with user */ 225 struct mount_pcpu *mnt_pcpu; /* per-CPU data */ 226 struct vnode *mnt_rootvnode; 227 struct vnode *mnt_vnodecovered; /* vnode we mounted on */ 228 struct vfsops *mnt_op; /* operations on fs */ 229 struct vfsconf *mnt_vfc; /* configuration info */ 230 struct mtx __aligned(CACHE_LINE_SIZE) mnt_mtx; /* mount structure interlock */ 231 int mnt_gen; /* struct mount generation */ 232 #define mnt_startzero mnt_list 233 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */ 234 struct vnode *mnt_syncer; /* syncer vnode */ 235 int mnt_ref; /* (i) Reference count */ 236 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */ 237 int mnt_nvnodelistsize; /* (i) # of vnodes */ 238 int mnt_writeopcount; /* (i) write syscalls pending */ 239 struct vfsoptlist *mnt_opt; /* current mount options */ 240 struct vfsoptlist *mnt_optnew; /* new options passed to fs */ 241 struct statfs mnt_stat; /* cache of filesystem stats */ 242 struct ucred *mnt_cred; /* credentials of mounter */ 243 void * mnt_data; /* private data */ 244 time_t mnt_time; /* last time written*/ 245 int mnt_iosize_max; /* max size for clusters, etc */ 246 struct netexport *mnt_export; /* (e) export list */ 247 struct label *mnt_label; /* MAC label for the fs */ 248 u_int mnt_hashseed; /* Random seed for vfs_hash */ 249 int mnt_lockref; /* (i) Lock reference count */ 250 int mnt_secondary_writes; /* (i) # of secondary writes */ 251 int mnt_secondary_accwrites;/* (i) secondary wr. starts */ 252 struct thread *mnt_susp_owner; /* (i) thread owning suspension */ 253 struct ucred *mnt_exjail; /* (i) jail which did exports */ 254 #define mnt_endzero mnt_gjprovider 255 char *mnt_gjprovider; /* gjournal provider name */ 256 struct mtx mnt_listmtx; 257 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */ 258 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */ 259 int mnt_upper_pending; /* (i) # of pending ops on mnt_uppers */ 260 struct lock mnt_explock; /* vfs_export walkers lock */ 261 TAILQ_HEAD(, mount_upper_node) mnt_uppers; /* (i) upper mounts over us */ 262 TAILQ_HEAD(, mount_upper_node) mnt_notify; /* (i) upper mounts for notification */ 263 STAILQ_ENTRY(mount) mnt_taskqueue_link; /* (d) our place in deferred unmount list */ 264 uint64_t mnt_taskqueue_flags; /* (d) unmount flags passed from taskqueue */ 265 unsigned int mnt_unmount_retries; /* (d) # of failed deferred unmount attempts */ 266 }; 267 #endif /* _WANT_MOUNT || _KERNEL */ 268 269 #ifdef _KERNEL 270 /* 271 * Definitions for MNT_VNODE_FOREACH_ALL. 272 */ 273 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp); 274 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp); 275 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp); 276 277 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \ 278 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \ 279 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp))) 280 281 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \ 282 do { \ 283 MNT_ILOCK(mp); \ 284 __mnt_vnode_markerfree_all(&(mvp), (mp)); \ 285 /* MNT_IUNLOCK(mp); -- done in above function */ \ 286 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \ 287 } while (0) 288 289 /* 290 * Definitions for MNT_VNODE_FOREACH_LAZY. 291 */ 292 typedef int mnt_lazy_cb_t(struct vnode *, void *); 293 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp, 294 mnt_lazy_cb_t *cb, void *cbarg); 295 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp, 296 mnt_lazy_cb_t *cb, void *cbarg); 297 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp); 298 299 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \ 300 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \ 301 (vp) != NULL; \ 302 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg))) 303 304 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \ 305 __mnt_vnode_markerfree_lazy(&(mvp), (mp)) 306 307 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx) 308 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx) 309 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx) 310 #define MNT_MTX(mp) (&(mp)->mnt_mtx) 311 312 #define MNT_REF(mp) do { \ 313 mtx_assert(MNT_MTX(mp), MA_OWNED); \ 314 mp->mnt_ref++; \ 315 } while (0) 316 #define MNT_REL(mp) do { \ 317 mtx_assert(MNT_MTX(mp), MA_OWNED); \ 318 (mp)->mnt_ref--; \ 319 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \ 320 vfs_dump_mount_counters(mp); \ 321 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \ 322 wakeup((mp)); \ 323 } while (0) 324 325 #endif /* _KERNEL */ 326 327 #if defined(_WANT_MNTOPTNAMES) || defined(_KERNEL) 328 struct mntoptnames { 329 uint64_t o_opt; 330 const char *o_name; 331 }; 332 #define MNTOPT_NAMES \ 333 { MNT_ASYNC, "asynchronous" }, \ 334 { MNT_EXPORTED, "NFS exported" }, \ 335 { MNT_LOCAL, "local" }, \ 336 { MNT_NOATIME, "noatime" }, \ 337 { MNT_NOEXEC, "noexec" }, \ 338 { MNT_NOSUID, "nosuid" }, \ 339 { MNT_NOSYMFOLLOW, "nosymfollow" }, \ 340 { MNT_QUOTA, "with quotas" }, \ 341 { MNT_RDONLY, "read-only" }, \ 342 { MNT_SYNCHRONOUS, "synchronous" }, \ 343 { MNT_UNION, "union" }, \ 344 { MNT_NOCLUSTERR, "noclusterr" }, \ 345 { MNT_NOCLUSTERW, "noclusterw" }, \ 346 { MNT_SUIDDIR, "suiddir" }, \ 347 { MNT_SOFTDEP, "soft-updates" }, \ 348 { MNT_SUJ, "journaled soft-updates" }, \ 349 { MNT_MULTILABEL, "multilabel" }, \ 350 { MNT_ACLS, "acls" }, \ 351 { MNT_NFS4ACLS, "nfsv4acls" }, \ 352 { MNT_GJOURNAL, "gjournal" }, \ 353 { MNT_AUTOMOUNTED, "automounted" }, \ 354 { MNT_VERIFIED, "verified" }, \ 355 { MNT_UNTRUSTED, "untrusted" }, \ 356 { MNT_NOCOVER, "nocover" }, \ 357 { MNT_EMPTYDIR, "emptydir" }, \ 358 { MNT_UPDATE, "update" }, \ 359 { MNT_DELEXPORT, "delexport" }, \ 360 { MNT_RELOAD, "reload" }, \ 361 { MNT_FORCE, "force" }, \ 362 { MNT_SNAPSHOT, "snapshot" }, \ 363 { 0, NULL } 364 #endif 365 366 /* 367 * User specifiable flags, stored in mnt_flag. 368 */ 369 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */ 370 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */ 371 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */ 372 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */ 373 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */ 374 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */ 375 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */ 376 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */ 377 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */ 378 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */ 379 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */ 380 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */ 381 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */ 382 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */ 383 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */ 384 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */ 385 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */ 386 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */ 387 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */ 388 389 /* 390 * NFS export related mount flags. 391 */ 392 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */ 393 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */ 394 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */ 395 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */ 396 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */ 397 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */ 398 #define MNT_EXTLS 0x0000004000000000ULL /* require TLS */ 399 #define MNT_EXTLSCERT 0x0000008000000000ULL /* require TLS with client cert */ 400 #define MNT_EXTLSCERTUSER 0x0000010000000000ULL /* require TLS with user cert */ 401 402 /* 403 * Flags set by internal operations, but visible to the user. 404 */ 405 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */ 406 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */ 407 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */ 408 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */ 409 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */ 410 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */ 411 412 /* 413 * Mask of flags that are visible to statfs(). 414 * XXX I think that this could now become (~(MNT_CMDFLAGS)) 415 * but the 'mount' program may need changing to handle this. 416 */ 417 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \ 418 MNT_NOSUID | MNT_UNION | MNT_SUJ | \ 419 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \ 420 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \ 421 MNT_LOCAL | MNT_USER | MNT_QUOTA | \ 422 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \ 423 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \ 424 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \ 425 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \ 426 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \ 427 MNT_UNTRUSTED) 428 429 /* Mask of flags that can be updated. */ 430 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \ 431 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \ 432 MNT_NOATIME | \ 433 MNT_NOSYMFOLLOW | MNT_IGNORE | \ 434 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \ 435 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \ 436 MNT_AUTOMOUNTED | MNT_UNTRUSTED) 437 438 /* 439 * External filesystem command modifier flags. 440 * Unmount can use the MNT_FORCE flag. 441 * XXX: These are not STATES and really should be somewhere else. 442 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL, 443 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2), 444 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2), 445 * it's harmless. 446 */ 447 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */ 448 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */ 449 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */ 450 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */ 451 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */ 452 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */ 453 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */ 454 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */ 455 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */ 456 #define MNT_RECURSE 0x0000100000000000ULL /* recursively unmount uppers */ 457 #define MNT_DEFERRED 0x0000200000000000ULL /* unmount in async context */ 458 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \ 459 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \ 460 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR | \ 461 MNT_RECURSE | MNT_DEFERRED) 462 463 /* 464 * Internal filesystem control flags stored in mnt_kern_flag. 465 * 466 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot 467 * proceed past the mount point. This keeps the subtree stable during 468 * mounts and unmounts. When non-forced unmount flushes all vnodes 469 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from 470 * queueing new vnodes. 471 * 472 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while 473 * dounmount() is still waiting to lock the mountpoint. This allows 474 * the filesystem to cancel operations that might otherwise deadlock 475 * with the unmount attempt (used by NFS). 476 */ 477 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */ 478 #define MNTK_ASYNC 0x00000002 /* filtered async flag */ 479 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */ 480 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */ 481 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */ 482 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */ 483 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */ 484 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */ 485 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads 486 and writes. Filesystem shall properly 487 handle i/o state on EFAULT. */ 488 #define MNTK_RECURSE 0x00000200 /* pending recursive unmount */ 489 #define MNTK_UPPER_WAITER 0x00000400 /* waiting to drain MNTK_UPPER_PENDING */ 490 /* UNUSED 0x00000800 */ 491 #define MNTK_UNLOCKED_INSMNTQUE 0x00001000 /* fs does not lock the vnode for insmntque */ 492 #define MNTK_UNMAPPED_BUFS 0x00002000 493 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */ 494 /* UNUSED 0x00008000 */ 495 #define MNTK_VMSETSIZE_BUG 0x00010000 496 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */ 497 #define MNTK_FPLOOKUP 0x00040000 /* fast path lookup is supported */ 498 #define MNTK_SUSPEND_ALL 0x00080000 /* Suspended by all-fs suspension */ 499 #define MNTK_TASKQUEUE_WAITER 0x00100000 /* Waiting on unmount taskqueue */ 500 /* UNUSED 0x00200000 */ 501 /* UNUSED 0x00400000 */ 502 #define MNTK_NOASYNC 0x00800000 /* disable async */ 503 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */ 504 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */ 505 #define MNTK_SUSPEND 0x08000000 /* request write suspension */ 506 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */ 507 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */ 508 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs 509 mounts over this fs */ 510 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */ 511 /* UNUSED 0x80000000 */ 512 513 #ifdef _KERNEL 514 static inline int 515 MNT_SHARED_WRITES(struct mount *mp) 516 { 517 518 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0); 519 } 520 521 static inline int 522 MNT_EXTENDED_SHARED(struct mount *mp) 523 { 524 525 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0); 526 } 527 #endif 528 529 /* 530 * Sysctl CTL_VFS definitions. 531 * 532 * Second level identifier specifies which filesystem. Second level 533 * identifier VFS_VFSCONF returns information about all filesystems. 534 * Second level identifier VFS_GENERIC is non-terminal. 535 */ 536 #define VFS_VFSCONF 0 /* get configured filesystems */ 537 #define VFS_GENERIC 0 /* generic filesystem information */ 538 /* 539 * Third level identifiers for VFS_GENERIC are given below; third 540 * level identifiers for specific filesystems are given in their 541 * mount specific header files. 542 */ 543 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */ 544 #define VFS_CONF 2 /* struct: vfsconf for filesystem given 545 as next argument */ 546 547 /* 548 * Flags for various system call interfaces. 549 * 550 * waitfor flags to vfs_sync() and getfsstat() 551 */ 552 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */ 553 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */ 554 #define MNT_LAZY 3 /* push data not written by filesystem syncer */ 555 #define MNT_SUSPEND 4 /* Suspend file system after sync */ 556 557 /* 558 * Generic file handle 559 */ 560 struct fhandle { 561 fsid_t fh_fsid; /* Filesystem id of mount point */ 562 struct fid fh_fid; /* Filesys specific id */ 563 }; 564 typedef struct fhandle fhandle_t; 565 566 /* 567 * Old export arguments without security flavor list 568 */ 569 struct oexport_args { 570 int ex_flags; /* export related flags */ 571 uid_t ex_root; /* mapping for root uid */ 572 struct xucred ex_anon; /* mapping for anonymous user */ 573 struct sockaddr *ex_addr; /* net address to which exported */ 574 u_char ex_addrlen; /* and the net address length */ 575 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 576 u_char ex_masklen; /* and the smask length */ 577 char *ex_indexfile; /* index file for WebNFS URLs */ 578 }; 579 580 /* 581 * Not quite so old export arguments with 32bit ex_flags and xucred ex_anon. 582 */ 583 #define MAXSECFLAVORS 5 584 struct o2export_args { 585 int ex_flags; /* export related flags */ 586 uid_t ex_root; /* mapping for root uid */ 587 struct xucred ex_anon; /* mapping for anonymous user */ 588 struct sockaddr *ex_addr; /* net address to which exported */ 589 u_char ex_addrlen; /* and the net address length */ 590 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 591 u_char ex_masklen; /* and the smask length */ 592 char *ex_indexfile; /* index file for WebNFS URLs */ 593 int ex_numsecflavors; /* security flavor count */ 594 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ 595 }; 596 597 /* 598 * Export arguments for local filesystem mount calls. 599 */ 600 struct export_args { 601 uint64_t ex_flags; /* export related flags */ 602 uid_t ex_root; /* mapping for root uid */ 603 uid_t ex_uid; /* mapping for anonymous user */ 604 int ex_ngroups; 605 gid_t *ex_groups; 606 struct sockaddr *ex_addr; /* net address to which exported */ 607 u_char ex_addrlen; /* and the net address length */ 608 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 609 u_char ex_masklen; /* and the smask length */ 610 char *ex_indexfile; /* index file for WebNFS URLs */ 611 int ex_numsecflavors; /* security flavor count */ 612 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ 613 }; 614 615 /* 616 * Structure holding information for a publicly exported filesystem 617 * (WebNFS). Currently the specs allow just for one such filesystem. 618 */ 619 struct nfs_public { 620 int np_valid; /* Do we hold valid information */ 621 fhandle_t np_handle; /* Filehandle for pub fs (internal) */ 622 struct mount *np_mount; /* Mountpoint of exported fs */ 623 char *np_index; /* Index file */ 624 }; 625 626 /* 627 * Filesystem configuration information. One of these exists for each 628 * type of filesystem supported by the kernel. These are searched at 629 * mount time to identify the requested filesystem. 630 * 631 * XXX: Never change the first two arguments! 632 */ 633 struct vfsconf { 634 u_int vfc_version; /* ABI version number */ 635 char vfc_name[MFSNAMELEN]; /* filesystem type name */ 636 struct vfsops *vfc_vfsops; /* filesystem operations vector */ 637 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */ 638 int vfc_typenum; /* historic filesystem type number */ 639 int vfc_refcount; /* number mounted of this type */ 640 int vfc_flags; /* permanent flags */ 641 int vfc_prison_flag; /* prison allow.mount.* flag */ 642 struct vfsoptdecl *vfc_opts; /* mount options */ 643 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */ 644 }; 645 646 /* Userland version of the struct vfsconf. */ 647 struct xvfsconf { 648 struct vfsops *vfc_vfsops; /* filesystem operations vector */ 649 char vfc_name[MFSNAMELEN]; /* filesystem type name */ 650 int vfc_typenum; /* historic filesystem type number */ 651 int vfc_refcount; /* number mounted of this type */ 652 int vfc_flags; /* permanent flags */ 653 struct vfsconf *vfc_next; /* next in list */ 654 }; 655 656 #ifndef BURN_BRIDGES 657 struct ovfsconf { 658 void *vfc_vfsops; 659 char vfc_name[32]; 660 int vfc_index; 661 int vfc_refcount; 662 int vfc_flags; 663 }; 664 #endif 665 666 /* 667 * NB: these flags refer to IMPLEMENTATION properties, not properties of 668 * any actual mounts; i.e., it does not make sense to change the flags. 669 */ 670 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */ 671 #define VFCF_NETWORK 0x00020000 /* may get data over the network */ 672 #define VFCF_READONLY 0x00040000 /* writes are not implemented */ 673 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */ 674 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */ 675 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */ 676 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */ 677 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */ 678 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests 679 to kernel->user (AST) transition */ 680 #define VFCF_FILEMOUNT 0x02000000 /* allow mounting files */ 681 682 typedef uint32_t fsctlop_t; 683 684 struct vfsidctl { 685 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */ 686 fsid_t vc_fsid; /* fsid to operate on */ 687 char vc_fstypename[MFSNAMELEN]; 688 /* type of fs 'nfs' or '*' */ 689 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */ 690 void *vc_ptr; /* pointer to data structure */ 691 size_t vc_len; /* sizeof said structure */ 692 u_int32_t vc_spare[12]; /* spare (must be zero) */ 693 }; 694 695 /* vfsidctl API version. */ 696 #define VFS_CTL_VERS1 0x01 697 698 /* 699 * New style VFS sysctls, do not reuse/conflict with the namespace for 700 * private sysctls. 701 * All "global" sysctl ops have the 33rd bit set: 702 * 0x...1.... 703 * Private sysctl ops should have the 33rd bit unset. 704 */ 705 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */ 706 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */ 707 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */ 708 709 struct vfsquery { 710 u_int32_t vq_flags; 711 u_int32_t vq_spare[31]; 712 }; 713 714 /* vfsquery flags */ 715 #define VQ_NOTRESP 0x0001 /* server down */ 716 #define VQ_NEEDAUTH 0x0002 /* server bad auth */ 717 #define VQ_LOWDISK 0x0004 /* we're low on space */ 718 #define VQ_MOUNT 0x0008 /* new filesystem arrived */ 719 #define VQ_UNMOUNT 0x0010 /* filesystem has left */ 720 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */ 721 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external 722 program */ 723 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */ 724 #define VQ_FLAG0100 0x0100 /* placeholder */ 725 #define VQ_FLAG0200 0x0200 /* placeholder */ 726 #define VQ_FLAG0400 0x0400 /* placeholder */ 727 #define VQ_FLAG0800 0x0800 /* placeholder */ 728 #define VQ_FLAG1000 0x1000 /* placeholder */ 729 #define VQ_FLAG2000 0x2000 /* placeholder */ 730 #define VQ_FLAG4000 0x4000 /* placeholder */ 731 #define VQ_FLAG8000 0x8000 /* placeholder */ 732 733 #ifdef _KERNEL 734 /* Point a sysctl request at a vfsidctl's data. */ 735 #define VCTLTOREQ(vc, req) \ 736 do { \ 737 (req)->newptr = (vc)->vc_ptr; \ 738 (req)->newlen = (vc)->vc_len; \ 739 (req)->newidx = 0; \ 740 } while (0) 741 #endif 742 743 struct iovec; 744 struct uio; 745 746 #ifdef _KERNEL 747 748 /* 749 * vfs_busy specific flags and mask. 750 */ 751 #define MBF_NOWAIT 0x01 752 #define MBF_MNTLSTLOCK 0x02 753 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK) 754 755 #ifdef MALLOC_DECLARE 756 MALLOC_DECLARE(M_MOUNT); 757 MALLOC_DECLARE(M_STATFS); 758 #endif 759 extern int maxvfsconf; /* highest defined filesystem type */ 760 761 TAILQ_HEAD(vfsconfhead, vfsconf); 762 extern struct vfsconfhead vfsconf; 763 764 /* 765 * Operations supported on mounted filesystem. 766 */ 767 struct mount_args; 768 struct nameidata; 769 struct sysctl_req; 770 struct mntarg; 771 772 /* 773 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall. 774 * The new way is vfs_mount. 775 * 776 * vfs_cmount implementations typically translate arguments from their 777 * respective old per-FS structures into the key-value list supported by 778 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace. 779 * 780 * Filesystems with mounters that use nmount(2) do not need to and should not 781 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and 782 * mount(2) entirely. 783 */ 784 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags); 785 typedef int vfs_unmount_t(struct mount *mp, int mntflags); 786 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp); 787 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg, 788 bool *mp_busy); 789 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp); 790 typedef int vfs_sync_t(struct mount *mp, int waitfor); 791 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags, 792 struct vnode **vpp); 793 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp, 794 int flags, struct vnode **vpp); 795 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam, 796 uint64_t *extflagsp, struct ucred **credanonp, 797 int *numsecflavors, int *secflavors); 798 typedef int vfs_init_t(struct vfsconf *); 799 typedef int vfs_uninit_t(struct vfsconf *); 800 typedef int vfs_extattrctl_t(struct mount *mp, int cmd, 801 struct vnode *filename_vp, int attrnamespace, 802 const char *attrname); 803 typedef int vfs_mount_t(struct mount *mp); 804 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op, 805 struct sysctl_req *req); 806 typedef void vfs_susp_clean_t(struct mount *mp); 807 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp); 808 typedef void vfs_purge_t(struct mount *mp); 809 struct sbuf; 810 typedef int vfs_report_lockf_t(struct mount *mp, struct sbuf *sb); 811 812 struct vfsops { 813 vfs_mount_t *vfs_mount; 814 vfs_cmount_t *vfs_cmount; 815 vfs_unmount_t *vfs_unmount; 816 vfs_root_t *vfs_root; 817 vfs_root_t *vfs_cachedroot; 818 vfs_quotactl_t *vfs_quotactl; 819 vfs_statfs_t *vfs_statfs; 820 vfs_sync_t *vfs_sync; 821 vfs_vget_t *vfs_vget; 822 vfs_fhtovp_t *vfs_fhtovp; 823 vfs_checkexp_t *vfs_checkexp; 824 vfs_init_t *vfs_init; 825 vfs_uninit_t *vfs_uninit; 826 vfs_extattrctl_t *vfs_extattrctl; 827 vfs_sysctl_t *vfs_sysctl; 828 vfs_susp_clean_t *vfs_susp_clean; 829 vfs_notify_lowervp_t *vfs_reclaim_lowervp; 830 vfs_notify_lowervp_t *vfs_unlink_lowervp; 831 vfs_purge_t *vfs_purge; 832 vfs_report_lockf_t *vfs_report_lockf; 833 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */ 834 }; 835 836 vfs_statfs_t __vfs_statfs; 837 838 #define VFS_MOUNT(MP) ({ \ 839 int _rc; \ 840 \ 841 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\ 842 _rc = (*(MP)->mnt_op->vfs_mount)(MP); \ 843 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\ 844 _rc; }) 845 846 #define VFS_UNMOUNT(MP, FORCE) ({ \ 847 int _rc; \ 848 \ 849 _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \ 850 _rc; }) 851 852 #define VFS_ROOT(MP, FLAGS, VPP) ({ \ 853 int _rc; \ 854 \ 855 _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \ 856 _rc; }) 857 858 #define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \ 859 int _rc; \ 860 \ 861 _rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \ 862 _rc; }) 863 864 #define VFS_QUOTACTL(MP, C, U, A, MP_BUSY) ({ \ 865 int _rc; \ 866 \ 867 _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A, MP_BUSY); \ 868 _rc; }) 869 870 #define VFS_STATFS(MP, SBP) ({ \ 871 int _rc; \ 872 \ 873 _rc = __vfs_statfs((MP), (SBP)); \ 874 _rc; }) 875 876 #define VFS_SYNC(MP, WAIT) ({ \ 877 int _rc; \ 878 \ 879 _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \ 880 _rc; }) 881 882 #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \ 883 int _rc; \ 884 \ 885 _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \ 886 _rc; }) 887 888 #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \ 889 int _rc; \ 890 \ 891 _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \ 892 _rc; }) 893 894 #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \ 895 int _rc; \ 896 \ 897 _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\ 898 SEC); \ 899 _rc; }) 900 901 #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \ 902 int _rc; \ 903 \ 904 _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \ 905 _rc; }) 906 907 #define VFS_SYSCTL(MP, OP, REQ) ({ \ 908 int _rc; \ 909 \ 910 _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \ 911 _rc; }) 912 913 #define VFS_SUSP_CLEAN(MP) do { \ 914 if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \ 915 (*(MP)->mnt_op->vfs_susp_clean)(MP); \ 916 } \ 917 } while (0) 918 919 #define VFS_RECLAIM_LOWERVP(MP, VP) do { \ 920 if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \ 921 (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \ 922 } \ 923 } while (0) 924 925 #define VFS_UNLINK_LOWERVP(MP, VP) do { \ 926 if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \ 927 (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \ 928 } \ 929 } while (0) 930 931 #define VFS_PURGE(MP) do { \ 932 if (*(MP)->mnt_op->vfs_purge != NULL) { \ 933 (*(MP)->mnt_op->vfs_purge)(MP); \ 934 } \ 935 } while (0) 936 937 #define VFS_KNOTE_LOCKED(vp, hint) do \ 938 { \ 939 VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \ 940 } while (0) 941 942 #define VFS_KNOTE_UNLOCKED(vp, hint) do \ 943 { \ 944 VN_KNOTE((vp), (hint), 0); \ 945 } while (0) 946 947 #include <sys/module.h> 948 949 /* 950 * Version numbers. 951 */ 952 #define VFS_VERSION_00 0x19660120 953 #define VFS_VERSION_01 0x20121030 954 #define VFS_VERSION_02 0x20180504 955 #define VFS_VERSION VFS_VERSION_02 956 957 #define VFS_SET(vfsops, fsname, flags) \ 958 static struct vfsconf fsname ## _vfsconf = { \ 959 .vfc_version = VFS_VERSION, \ 960 .vfc_name = #fsname, \ 961 .vfc_vfsops = &vfsops, \ 962 .vfc_typenum = -1, \ 963 .vfc_flags = flags, \ 964 }; \ 965 static moduledata_t fsname ## _mod = { \ 966 #fsname, \ 967 vfs_modevent, \ 968 & fsname ## _vfsconf \ 969 }; \ 970 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE) 971 972 enum vfs_notify_upper_type { 973 VFS_NOTIFY_UPPER_RECLAIM, 974 VFS_NOTIFY_UPPER_UNLINK, 975 }; 976 977 /* 978 * exported vnode operations 979 */ 980 981 /* Define this to indicate that vfs_exjail_clone() exists for ZFS to use. */ 982 #define VFS_SUPPORTS_EXJAIL_CLONE 1 983 984 int dounmount(struct mount *, uint64_t, struct thread *); 985 986 int kernel_mount(struct mntarg *ma, uint64_t flags); 987 struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len); 988 struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name); 989 struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...); 990 struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len); 991 void statfs_scale_blocks(struct statfs *sf, long max_size); 992 struct vfsconf *vfs_byname(const char *); 993 struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *); 994 void vfs_mount_destroy(struct mount *); 995 void vfs_event_signal(fsid_t *, u_int32_t, intptr_t); 996 void vfs_freeopts(struct vfsoptlist *opts); 997 void vfs_deleteopt(struct vfsoptlist *opts, const char *name); 998 int vfs_buildopts(struct uio *auio, struct vfsoptlist **options); 999 int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w, 1000 uint64_t val); 1001 int vfs_getopt(struct vfsoptlist *, const char *, void **, int *); 1002 int vfs_getopt_pos(struct vfsoptlist *opts, const char *name); 1003 int vfs_getopt_size(struct vfsoptlist *opts, const char *name, 1004 off_t *value); 1005 char *vfs_getopts(struct vfsoptlist *, const char *, int *error); 1006 int vfs_copyopt(struct vfsoptlist *, const char *, void *, int); 1007 int vfs_filteropt(struct vfsoptlist *, const char **legal); 1008 void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...); 1009 int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...); 1010 int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value, 1011 int len); 1012 int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value, 1013 int len); 1014 int vfs_setopts(struct vfsoptlist *opts, const char *name, 1015 const char *value); 1016 int vfs_setpublicfs /* set publicly exported fs */ 1017 (struct mount *, struct netexport *, struct export_args *); 1018 void vfs_periodic(struct mount *, int); 1019 int vfs_busy(struct mount *, int); 1020 void vfs_exjail_clone(struct mount *, struct mount *); 1021 void vfs_exjail_delete(struct prison *); 1022 int vfs_export /* process mount export info */ 1023 (struct mount *, struct export_args *, bool); 1024 void vfs_free_addrlist(struct netexport *); 1025 void vfs_allocate_syncvnode(struct mount *); 1026 void vfs_deallocate_syncvnode(struct mount *); 1027 int vfs_donmount(struct thread *td, uint64_t fsflags, 1028 struct uio *fsoptions); 1029 void vfs_getnewfsid(struct mount *); 1030 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */ 1031 struct mount *vfs_busyfs(fsid_t *); 1032 int vfs_modevent(module_t, int, void *); 1033 void vfs_mount_error(struct mount *, const char *, ...); 1034 void vfs_mountroot(void); /* mount our root filesystem */ 1035 void vfs_mountedfrom(struct mount *, const char *from); 1036 void vfs_notify_upper(struct vnode *, enum vfs_notify_upper_type); 1037 struct mount *vfs_ref_from_vp(struct vnode *); 1038 void vfs_ref(struct mount *); 1039 void vfs_rel(struct mount *); 1040 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *, 1041 struct ucred *); 1042 int vfs_suser(struct mount *, struct thread *); 1043 void vfs_unbusy(struct mount *); 1044 void vfs_unmountall(void); 1045 struct mount *vfs_register_upper_from_vp(struct vnode *, 1046 struct mount *ump, struct mount_upper_node *); 1047 void vfs_register_for_notification(struct mount *, struct mount *, 1048 struct mount_upper_node *); 1049 void vfs_unregister_for_notification(struct mount *, 1050 struct mount_upper_node *); 1051 void vfs_unregister_upper(struct mount *, struct mount_upper_node *); 1052 int vfs_remount_ro(struct mount *mp); 1053 int vfs_report_lockf(struct mount *mp, struct sbuf *sb); 1054 1055 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */ 1056 extern struct mtx_padalign mountlist_mtx; 1057 extern struct nfs_public nfs_pub; 1058 extern struct sx vfsconf_sx; 1059 #define vfsconf_lock() sx_xlock(&vfsconf_sx) 1060 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx) 1061 #define vfsconf_slock() sx_slock(&vfsconf_sx) 1062 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx) 1063 struct vnode *mntfs_allocvp(struct mount *, struct vnode *); 1064 void mntfs_freevp(struct vnode *); 1065 1066 /* 1067 * Declarations for these vfs default operations are located in 1068 * kern/vfs_default.c. They will be automatically used to replace 1069 * null entries in VFS ops tables when registering a new filesystem 1070 * type in the global table. 1071 */ 1072 vfs_root_t vfs_stdroot; 1073 vfs_quotactl_t vfs_stdquotactl; 1074 vfs_statfs_t vfs_stdstatfs; 1075 vfs_sync_t vfs_stdsync; 1076 vfs_sync_t vfs_stdnosync; 1077 vfs_vget_t vfs_stdvget; 1078 vfs_fhtovp_t vfs_stdfhtovp; 1079 vfs_checkexp_t vfs_stdcheckexp; 1080 vfs_init_t vfs_stdinit; 1081 vfs_uninit_t vfs_stduninit; 1082 vfs_extattrctl_t vfs_stdextattrctl; 1083 vfs_sysctl_t vfs_stdsysctl; 1084 1085 void syncer_suspend(void); 1086 void syncer_resume(void); 1087 1088 struct vnode *vfs_cache_root_clear(struct mount *); 1089 void vfs_cache_root_set(struct mount *, struct vnode *); 1090 1091 void vfs_op_barrier_wait(struct mount *); 1092 void vfs_op_enter(struct mount *); 1093 void vfs_op_exit_locked(struct mount *); 1094 void vfs_op_exit(struct mount *); 1095 1096 #ifdef DIAGNOSTIC 1097 void vfs_assert_mount_counters(struct mount *); 1098 void vfs_dump_mount_counters(struct mount *); 1099 #else 1100 #define vfs_assert_mount_counters(mp) do { } while (0) 1101 #define vfs_dump_mount_counters(mp) do { } while (0) 1102 #endif 1103 1104 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT }; 1105 int vfs_mount_fetch_counter(struct mount *, enum mount_counter); 1106 1107 void suspend_all_fs(void); 1108 void resume_all_fs(void); 1109 1110 /* 1111 * Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes 1112 * all CPUs not executing code enclosed by thread_in_ops_pcpu variable. 1113 * 1114 * This provides an invariant that by the time the last CPU is observed not 1115 * executing, everyone else entering will see the counter > 0 and exit. 1116 * 1117 * Note there is no barrier between vfs_ops and the rest of the code in the 1118 * section. It is not necessary as the writer has to wait for everyone to drain 1119 * before making any changes or only make changes safe while the section is 1120 * executed. 1121 */ 1122 #define vfs_mount_pcpu(mp) zpcpu_get(mp->mnt_pcpu) 1123 #define vfs_mount_pcpu_remote(mp, cpu) zpcpu_get_cpu(mp->mnt_pcpu, cpu) 1124 1125 #define vfs_op_thread_entered(mp) ({ \ 1126 MPASS(curthread->td_critnest > 0); \ 1127 struct mount_pcpu *_mpcpu = vfs_mount_pcpu(mp); \ 1128 _mpcpu->mntp_thread_in_ops == 1; \ 1129 }) 1130 1131 #define vfs_op_thread_enter_crit(mp, _mpcpu) ({ \ 1132 bool _retval_crit = true; \ 1133 MPASS(curthread->td_critnest > 0); \ 1134 _mpcpu = vfs_mount_pcpu(mp); \ 1135 MPASS(mpcpu->mntp_thread_in_ops == 0); \ 1136 _mpcpu->mntp_thread_in_ops = 1; \ 1137 atomic_interrupt_fence(); \ 1138 if (__predict_false(mp->mnt_vfs_ops > 0)) { \ 1139 vfs_op_thread_exit_crit(mp, _mpcpu); \ 1140 _retval_crit = false; \ 1141 } \ 1142 _retval_crit; \ 1143 }) 1144 1145 #define vfs_op_thread_enter(mp, _mpcpu) ({ \ 1146 bool _retval; \ 1147 critical_enter(); \ 1148 _retval = vfs_op_thread_enter_crit(mp, _mpcpu); \ 1149 if (__predict_false(!_retval)) \ 1150 critical_exit(); \ 1151 _retval; \ 1152 }) 1153 1154 #define vfs_op_thread_exit_crit(mp, _mpcpu) do { \ 1155 MPASS(_mpcpu == vfs_mount_pcpu(mp)); \ 1156 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1157 atomic_interrupt_fence(); \ 1158 _mpcpu->mntp_thread_in_ops = 0; \ 1159 } while (0) 1160 1161 #define vfs_op_thread_exit(mp, _mpcpu) do { \ 1162 vfs_op_thread_exit_crit(mp, _mpcpu); \ 1163 critical_exit(); \ 1164 } while (0) 1165 1166 #define vfs_mp_count_add_pcpu(_mpcpu, count, val) do { \ 1167 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1168 _mpcpu->mntp_##count += val; \ 1169 } while (0) 1170 1171 #define vfs_mp_count_sub_pcpu(_mpcpu, count, val) do { \ 1172 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1173 _mpcpu->mntp_##count -= val; \ 1174 } while (0) 1175 1176 #else /* !_KERNEL */ 1177 1178 #include <sys/cdefs.h> 1179 1180 struct stat; 1181 1182 __BEGIN_DECLS 1183 int fhlink(struct fhandle *, const char *); 1184 int fhlinkat(struct fhandle *, int, const char *); 1185 int fhopen(const struct fhandle *, int); 1186 int fhreadlink(struct fhandle *, char *, size_t); 1187 int fhstat(const struct fhandle *, struct stat *); 1188 int fhstatfs(const struct fhandle *, struct statfs *); 1189 int fstatfs(int, struct statfs *); 1190 int getfh(const char *, fhandle_t *); 1191 int getfhat(int, char *, struct fhandle *, int); 1192 int getfsstat(struct statfs *, long, int); 1193 int getmntinfo(struct statfs **, int); 1194 int lgetfh(const char *, fhandle_t *); 1195 int mount(const char *, const char *, int, void *); 1196 int nmount(struct iovec *, unsigned int, int); 1197 int statfs(const char *, struct statfs *); 1198 int unmount(const char *, int); 1199 1200 /* C library stuff */ 1201 int getvfsbyname(const char *, struct xvfsconf *); 1202 __END_DECLS 1203 1204 #endif /* _KERNEL */ 1205 1206 #endif /* !_SYS_MOUNT_H_ */ 1207