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 * @(#)mount.h 8.21 (Berkeley) 5/20/95 32 * $FreeBSD$ 33 */ 34 35 #ifndef _SYS_MOUNT_H_ 36 #define _SYS_MOUNT_H_ 37 38 #include <sys/ucred.h> 39 #include <sys/queue.h> 40 #ifdef _KERNEL 41 #include <sys/types.h> 42 #include <sys/lock.h> 43 #include <sys/lockmgr.h> 44 #include <sys/tslog.h> 45 #include <sys/_mutex.h> 46 #include <sys/_sx.h> 47 #endif 48 49 /* 50 * NOTE: When changing statfs structure, mount structure, MNT_* flags or 51 * MNTK_* flags also update DDB show mount command in vfs_subr.c. 52 */ 53 54 typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */ 55 56 #define fsidcmp(a, b) memcmp((a), (b), sizeof(fsid_t)) 57 58 /* 59 * File identifier. 60 * These are unique per filesystem on a single machine. 61 * 62 * Note that the offset of fid_data is 4 bytes, so care must be taken to avoid 63 * undefined behavior accessing unaligned fields within an embedded struct. 64 */ 65 #define MAXFIDSZ 16 66 67 struct fid { 68 u_short fid_len; /* length of data in bytes */ 69 u_short fid_data0; /* force longword alignment */ 70 char fid_data[MAXFIDSZ]; /* data (variable length) */ 71 }; 72 73 /* 74 * filesystem statistics 75 */ 76 #define MFSNAMELEN 16 /* length of type name including null */ 77 #define MNAMELEN 1024 /* size of on/from name bufs */ 78 #define STATFS_VERSION 0x20140518 /* current version number */ 79 struct statfs { 80 uint32_t f_version; /* structure version number */ 81 uint32_t f_type; /* type of filesystem */ 82 uint64_t f_flags; /* copy of mount exported flags */ 83 uint64_t f_bsize; /* filesystem fragment size */ 84 uint64_t f_iosize; /* optimal transfer block size */ 85 uint64_t f_blocks; /* total data blocks in filesystem */ 86 uint64_t f_bfree; /* free blocks in filesystem */ 87 int64_t f_bavail; /* free blocks avail to non-superuser */ 88 uint64_t f_files; /* total file nodes in filesystem */ 89 int64_t f_ffree; /* free nodes avail to non-superuser */ 90 uint64_t f_syncwrites; /* count of sync writes since mount */ 91 uint64_t f_asyncwrites; /* count of async writes since mount */ 92 uint64_t f_syncreads; /* count of sync reads since mount */ 93 uint64_t f_asyncreads; /* count of async reads since mount */ 94 uint32_t f_nvnodelistsize; /* # of vnodes */ 95 uint32_t f_spare0; /* unused spare */ 96 uint64_t f_spare[9]; /* unused spare */ 97 uint32_t f_namemax; /* maximum filename length */ 98 uid_t f_owner; /* user that mounted the filesystem */ 99 fsid_t f_fsid; /* filesystem id */ 100 char f_charspare[80]; /* spare string space */ 101 char f_fstypename[MFSNAMELEN]; /* filesystem type name */ 102 char f_mntfromname[MNAMELEN]; /* mounted filesystem */ 103 char f_mntonname[MNAMELEN]; /* directory on which mounted */ 104 }; 105 106 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL) 107 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */ 108 struct freebsd11_statfs { 109 uint32_t f_version; /* structure version number */ 110 uint32_t f_type; /* type of filesystem */ 111 uint64_t f_flags; /* copy of mount exported flags */ 112 uint64_t f_bsize; /* filesystem fragment size */ 113 uint64_t f_iosize; /* optimal transfer block size */ 114 uint64_t f_blocks; /* total data blocks in filesystem */ 115 uint64_t f_bfree; /* free blocks in filesystem */ 116 int64_t f_bavail; /* free blocks avail to non-superuser */ 117 uint64_t f_files; /* total file nodes in filesystem */ 118 int64_t f_ffree; /* free nodes avail to non-superuser */ 119 uint64_t f_syncwrites; /* count of sync writes since mount */ 120 uint64_t f_asyncwrites; /* count of async writes since mount */ 121 uint64_t f_syncreads; /* count of sync reads since mount */ 122 uint64_t f_asyncreads; /* count of async reads since mount */ 123 uint64_t f_spare[10]; /* unused spare */ 124 uint32_t f_namemax; /* maximum filename length */ 125 uid_t f_owner; /* user that mounted the filesystem */ 126 fsid_t f_fsid; /* filesystem id */ 127 char f_charspare[80]; /* spare string space */ 128 char f_fstypename[16]; /* filesystem type name */ 129 char f_mntfromname[88]; /* mounted filesystem */ 130 char f_mntonname[88]; /* directory on which mounted */ 131 }; 132 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */ 133 134 #ifdef _KERNEL 135 #define OMFSNAMELEN 16 /* length of fs type name, including null */ 136 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */ 137 138 /* XXX getfsstat.2 is out of date with write and read counter changes here. */ 139 /* XXX statfs.2 is out of date with read counter changes here. */ 140 struct ostatfs { 141 long f_spare2; /* placeholder */ 142 long f_bsize; /* fundamental filesystem block size */ 143 long f_iosize; /* optimal transfer block size */ 144 long f_blocks; /* total data blocks in filesystem */ 145 long f_bfree; /* free blocks in fs */ 146 long f_bavail; /* free blocks avail to non-superuser */ 147 long f_files; /* total file nodes in filesystem */ 148 long f_ffree; /* free file nodes in fs */ 149 fsid_t f_fsid; /* filesystem id */ 150 uid_t f_owner; /* user that mounted the filesystem */ 151 int f_type; /* type of filesystem */ 152 int f_flags; /* copy of mount exported flags */ 153 long f_syncwrites; /* count of sync writes since mount */ 154 long f_asyncwrites; /* count of async writes since mount */ 155 char f_fstypename[OMFSNAMELEN]; /* fs type name */ 156 char f_mntonname[OMNAMELEN]; /* directory on which mounted */ 157 long f_syncreads; /* count of sync reads since mount */ 158 long f_asyncreads; /* count of async reads since mount */ 159 short f_spares1; /* unused spare */ 160 char f_mntfromname[OMNAMELEN];/* mounted filesystem */ 161 short f_spares2; /* unused spare */ 162 /* 163 * XXX on machines where longs are aligned to 8-byte boundaries, there 164 * is an unnamed int32_t here. This spare was after the apparent end 165 * of the struct until we bit off the read counters from f_mntonname. 166 */ 167 long f_spare[2]; /* unused spare */ 168 }; 169 #endif /* _KERNEL */ 170 171 #if defined(_WANT_MOUNT) || defined(_KERNEL) 172 TAILQ_HEAD(vnodelst, vnode); 173 174 /* Mount options list */ 175 TAILQ_HEAD(vfsoptlist, vfsopt); 176 struct vfsopt { 177 TAILQ_ENTRY(vfsopt) link; 178 char *name; 179 void *value; 180 int len; 181 int pos; 182 int seen; 183 }; 184 185 struct mount_pcpu { 186 int mntp_thread_in_ops; 187 int mntp_ref; 188 int mntp_lockref; 189 int mntp_writeopcount; 190 }; 191 192 _Static_assert(sizeof(struct mount_pcpu) == 16, 193 "the struct is allocated from pcpu 16 zone"); 194 195 /* 196 * Structure for tracking a stacked filesystem mounted above another 197 * filesystem. This is expected to be stored in the upper FS' per-mount data. 198 * 199 * Lock reference: 200 * i - lower mount interlock 201 * c - constant from node initialization 202 */ 203 struct mount_upper_node { 204 struct mount *mp; /* (c) mount object for upper FS */ 205 TAILQ_ENTRY(mount_upper_node) mnt_upper_link; /* (i) position in uppers list */ 206 }; 207 208 /* 209 * Structure per mounted filesystem. Each mounted filesystem has an 210 * array of operations and an instance record. The filesystems are 211 * put on a doubly linked list. 212 * 213 * Lock reference: 214 * l - mnt_listmtx 215 * m - mountlist_mtx 216 * i - interlock 217 * v - vnode freelist mutex 218 * d - deferred unmount list mutex 219 * e - mnt_explock 220 * 221 * Unmarked fields are considered stable as long as a ref is held. 222 * 223 */ 224 struct mount { 225 int mnt_vfs_ops; /* (i) pending vfs ops */ 226 int mnt_kern_flag; /* (i) kernel only flags */ 227 uint64_t mnt_flag; /* (i) flags shared with user */ 228 struct mount_pcpu *mnt_pcpu; /* per-CPU data */ 229 struct vnode *mnt_rootvnode; 230 struct vnode *mnt_vnodecovered; /* vnode we mounted on */ 231 struct vfsops *mnt_op; /* operations on fs */ 232 struct vfsconf *mnt_vfc; /* configuration info */ 233 struct mtx __aligned(CACHE_LINE_SIZE) mnt_mtx; /* mount structure interlock */ 234 int mnt_gen; /* struct mount generation */ 235 #define mnt_startzero mnt_list 236 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */ 237 struct vnode *mnt_syncer; /* syncer vnode */ 238 int mnt_ref; /* (i) Reference count */ 239 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */ 240 int mnt_nvnodelistsize; /* (i) # of vnodes */ 241 int mnt_writeopcount; /* (i) write syscalls pending */ 242 struct vfsoptlist *mnt_opt; /* current mount options */ 243 struct vfsoptlist *mnt_optnew; /* new options passed to fs */ 244 struct statfs mnt_stat; /* cache of filesystem stats */ 245 struct ucred *mnt_cred; /* credentials of mounter */ 246 void * mnt_data; /* private data */ 247 time_t mnt_time; /* last time written*/ 248 int mnt_iosize_max; /* max size for clusters, etc */ 249 struct netexport *mnt_export; /* (e) export list */ 250 struct label *mnt_label; /* MAC label for the fs */ 251 u_int mnt_hashseed; /* Random seed for vfs_hash */ 252 int mnt_lockref; /* (i) Lock reference count */ 253 int mnt_secondary_writes; /* (i) # of secondary writes */ 254 int mnt_secondary_accwrites;/* (i) secondary wr. starts */ 255 struct thread *mnt_susp_owner; /* (i) thread owning suspension */ 256 struct ucred *mnt_exjail; /* (i) jail which did exports */ 257 #define mnt_endzero mnt_gjprovider 258 char *mnt_gjprovider; /* gjournal provider name */ 259 struct mtx mnt_listmtx; 260 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */ 261 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */ 262 int mnt_upper_pending; /* (i) # of pending ops on mnt_uppers */ 263 struct lock mnt_explock; /* vfs_export walkers lock */ 264 TAILQ_HEAD(, mount_upper_node) mnt_uppers; /* (i) upper mounts over us */ 265 TAILQ_HEAD(, mount_upper_node) mnt_notify; /* (i) upper mounts for notification */ 266 STAILQ_ENTRY(mount) mnt_taskqueue_link; /* (d) our place in deferred unmount list */ 267 uint64_t mnt_taskqueue_flags; /* (d) unmount flags passed from taskqueue */ 268 unsigned int mnt_unmount_retries; /* (d) # of failed deferred unmount attempts */ 269 }; 270 #endif /* _WANT_MOUNT || _KERNEL */ 271 272 #ifdef _KERNEL 273 /* 274 * Definitions for MNT_VNODE_FOREACH_ALL. 275 */ 276 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp); 277 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp); 278 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp); 279 280 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \ 281 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \ 282 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp))) 283 284 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \ 285 do { \ 286 MNT_ILOCK(mp); \ 287 __mnt_vnode_markerfree_all(&(mvp), (mp)); \ 288 /* MNT_IUNLOCK(mp); -- done in above function */ \ 289 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \ 290 } while (0) 291 292 /* 293 * Definitions for MNT_VNODE_FOREACH_LAZY. 294 */ 295 typedef int mnt_lazy_cb_t(struct vnode *, void *); 296 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp, 297 mnt_lazy_cb_t *cb, void *cbarg); 298 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp, 299 mnt_lazy_cb_t *cb, void *cbarg); 300 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp); 301 302 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \ 303 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \ 304 (vp) != NULL; \ 305 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg))) 306 307 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \ 308 __mnt_vnode_markerfree_lazy(&(mvp), (mp)) 309 310 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx) 311 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx) 312 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx) 313 #define MNT_MTX(mp) (&(mp)->mnt_mtx) 314 315 #define MNT_REF(mp) do { \ 316 mtx_assert(MNT_MTX(mp), MA_OWNED); \ 317 mp->mnt_ref++; \ 318 } while (0) 319 #define MNT_REL(mp) do { \ 320 mtx_assert(MNT_MTX(mp), MA_OWNED); \ 321 (mp)->mnt_ref--; \ 322 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \ 323 vfs_dump_mount_counters(mp); \ 324 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \ 325 wakeup((mp)); \ 326 } while (0) 327 328 #endif /* _KERNEL */ 329 330 #if defined(_WANT_MNTOPTNAMES) || defined(_KERNEL) 331 struct mntoptnames { 332 uint64_t o_opt; 333 const char *o_name; 334 }; 335 #define MNTOPT_NAMES \ 336 { MNT_ASYNC, "asynchronous" }, \ 337 { MNT_EXPORTED, "NFS exported" }, \ 338 { MNT_LOCAL, "local" }, \ 339 { MNT_NOATIME, "noatime" }, \ 340 { MNT_NOEXEC, "noexec" }, \ 341 { MNT_NOSUID, "nosuid" }, \ 342 { MNT_NOSYMFOLLOW, "nosymfollow" }, \ 343 { MNT_QUOTA, "with quotas" }, \ 344 { MNT_RDONLY, "read-only" }, \ 345 { MNT_SYNCHRONOUS, "synchronous" }, \ 346 { MNT_UNION, "union" }, \ 347 { MNT_NOCLUSTERR, "noclusterr" }, \ 348 { MNT_NOCLUSTERW, "noclusterw" }, \ 349 { MNT_SUIDDIR, "suiddir" }, \ 350 { MNT_SOFTDEP, "soft-updates" }, \ 351 { MNT_SUJ, "journaled soft-updates" }, \ 352 { MNT_MULTILABEL, "multilabel" }, \ 353 { MNT_ACLS, "acls" }, \ 354 { MNT_NFS4ACLS, "nfsv4acls" }, \ 355 { MNT_GJOURNAL, "gjournal" }, \ 356 { MNT_AUTOMOUNTED, "automounted" }, \ 357 { MNT_VERIFIED, "verified" }, \ 358 { MNT_UNTRUSTED, "untrusted" }, \ 359 { MNT_NOCOVER, "nocover" }, \ 360 { MNT_EMPTYDIR, "emptydir" }, \ 361 { MNT_UPDATE, "update" }, \ 362 { MNT_DELEXPORT, "delexport" }, \ 363 { MNT_RELOAD, "reload" }, \ 364 { MNT_FORCE, "force" }, \ 365 { MNT_SNAPSHOT, "snapshot" }, \ 366 { 0, NULL } 367 #endif 368 369 /* 370 * User specifiable flags, stored in mnt_flag. 371 */ 372 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */ 373 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */ 374 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */ 375 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */ 376 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */ 377 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */ 378 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */ 379 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */ 380 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */ 381 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */ 382 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */ 383 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */ 384 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */ 385 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */ 386 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */ 387 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */ 388 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */ 389 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */ 390 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */ 391 392 /* 393 * NFS export related mount flags. 394 */ 395 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */ 396 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */ 397 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */ 398 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */ 399 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */ 400 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */ 401 #define MNT_EXTLS 0x0000004000000000ULL /* require TLS */ 402 #define MNT_EXTLSCERT 0x0000008000000000ULL /* require TLS with client cert */ 403 #define MNT_EXTLSCERTUSER 0x0000010000000000ULL /* require TLS with user cert */ 404 405 /* 406 * Flags set by internal operations, but visible to the user. 407 */ 408 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */ 409 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */ 410 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */ 411 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */ 412 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */ 413 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */ 414 415 /* 416 * Mask of flags that are visible to statfs(). 417 * XXX I think that this could now become (~(MNT_CMDFLAGS)) 418 * but the 'mount' program may need changing to handle this. 419 */ 420 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \ 421 MNT_NOSUID | MNT_UNION | MNT_SUJ | \ 422 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \ 423 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \ 424 MNT_LOCAL | MNT_USER | MNT_QUOTA | \ 425 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \ 426 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \ 427 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \ 428 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \ 429 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \ 430 MNT_UNTRUSTED) 431 432 /* Mask of flags that can be updated. */ 433 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \ 434 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \ 435 MNT_NOATIME | \ 436 MNT_NOSYMFOLLOW | MNT_IGNORE | \ 437 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \ 438 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \ 439 MNT_AUTOMOUNTED | MNT_UNTRUSTED) 440 441 /* 442 * External filesystem command modifier flags. 443 * Unmount can use the MNT_FORCE flag. 444 * XXX: These are not STATES and really should be somewhere else. 445 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL, 446 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2), 447 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2), 448 * it's harmless. 449 */ 450 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */ 451 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */ 452 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */ 453 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */ 454 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */ 455 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */ 456 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */ 457 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */ 458 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */ 459 #define MNT_RECURSE 0x0000100000000000ULL /* recursively unmount uppers */ 460 #define MNT_DEFERRED 0x0000200000000000ULL /* unmount in async context */ 461 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \ 462 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \ 463 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR | \ 464 MNT_RECURSE | MNT_DEFERRED) 465 466 /* 467 * Internal filesystem control flags stored in mnt_kern_flag. 468 * 469 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot 470 * proceed past the mount point. This keeps the subtree stable during 471 * mounts and unmounts. When non-forced unmount flushes all vnodes 472 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from 473 * queueing new vnodes. 474 * 475 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while 476 * dounmount() is still waiting to lock the mountpoint. This allows 477 * the filesystem to cancel operations that might otherwise deadlock 478 * with the unmount attempt (used by NFS). 479 */ 480 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */ 481 #define MNTK_ASYNC 0x00000002 /* filtered async flag */ 482 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */ 483 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */ 484 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */ 485 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */ 486 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */ 487 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */ 488 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads 489 and writes. Filesystem shall properly 490 handle i/o state on EFAULT. */ 491 #define MNTK_RECURSE 0x00000200 /* pending recursive unmount */ 492 #define MNTK_UPPER_WAITER 0x00000400 /* waiting to drain MNTK_UPPER_PENDING */ 493 /* UNUSED 0x00000800 */ 494 #define MNTK_UNLOCKED_INSMNTQUE 0x00001000 /* fs does not lock the vnode for insmntque */ 495 #define MNTK_UNMAPPED_BUFS 0x00002000 496 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */ 497 /* UNUSED 0x00008000 */ 498 #define MNTK_VMSETSIZE_BUG 0x00010000 499 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */ 500 #define MNTK_FPLOOKUP 0x00040000 /* fast path lookup is supported */ 501 #define MNTK_SUSPEND_ALL 0x00080000 /* Suspended by all-fs suspension */ 502 #define MNTK_TASKQUEUE_WAITER 0x00100000 /* Waiting on unmount taskqueue */ 503 /* UNUSED 0x00200000 */ 504 /* UNUSED 0x00400000 */ 505 #define MNTK_NOASYNC 0x00800000 /* disable async */ 506 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */ 507 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */ 508 #define MNTK_SUSPEND 0x08000000 /* request write suspension */ 509 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */ 510 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */ 511 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs 512 mounts over this fs */ 513 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */ 514 /* UNUSED 0x80000000 */ 515 516 #ifdef _KERNEL 517 static inline int 518 MNT_SHARED_WRITES(struct mount *mp) 519 { 520 521 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0); 522 } 523 524 static inline int 525 MNT_EXTENDED_SHARED(struct mount *mp) 526 { 527 528 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0); 529 } 530 #endif 531 532 /* 533 * Sysctl CTL_VFS definitions. 534 * 535 * Second level identifier specifies which filesystem. Second level 536 * identifier VFS_VFSCONF returns information about all filesystems. 537 * Second level identifier VFS_GENERIC is non-terminal. 538 */ 539 #define VFS_VFSCONF 0 /* get configured filesystems */ 540 #define VFS_GENERIC 0 /* generic filesystem information */ 541 /* 542 * Third level identifiers for VFS_GENERIC are given below; third 543 * level identifiers for specific filesystems are given in their 544 * mount specific header files. 545 */ 546 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */ 547 #define VFS_CONF 2 /* struct: vfsconf for filesystem given 548 as next argument */ 549 550 /* 551 * Flags for various system call interfaces. 552 * 553 * waitfor flags to vfs_sync() and getfsstat() 554 */ 555 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */ 556 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */ 557 #define MNT_LAZY 3 /* push data not written by filesystem syncer */ 558 #define MNT_SUSPEND 4 /* Suspend file system after sync */ 559 560 /* 561 * Generic file handle 562 */ 563 struct fhandle { 564 fsid_t fh_fsid; /* Filesystem id of mount point */ 565 struct fid fh_fid; /* Filesys specific id */ 566 }; 567 typedef struct fhandle fhandle_t; 568 569 /* 570 * Old export arguments without security flavor list 571 */ 572 struct oexport_args { 573 int ex_flags; /* export related flags */ 574 uid_t ex_root; /* mapping for root uid */ 575 struct xucred ex_anon; /* mapping for anonymous user */ 576 struct sockaddr *ex_addr; /* net address to which exported */ 577 u_char ex_addrlen; /* and the net address length */ 578 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 579 u_char ex_masklen; /* and the smask length */ 580 char *ex_indexfile; /* index file for WebNFS URLs */ 581 }; 582 583 /* 584 * Not quite so old export arguments with 32bit ex_flags and xucred ex_anon. 585 */ 586 #define MAXSECFLAVORS 5 587 struct o2export_args { 588 int ex_flags; /* export related flags */ 589 uid_t ex_root; /* mapping for root uid */ 590 struct xucred ex_anon; /* mapping for anonymous user */ 591 struct sockaddr *ex_addr; /* net address to which exported */ 592 u_char ex_addrlen; /* and the net address length */ 593 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 594 u_char ex_masklen; /* and the smask length */ 595 char *ex_indexfile; /* index file for WebNFS URLs */ 596 int ex_numsecflavors; /* security flavor count */ 597 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ 598 }; 599 600 /* 601 * Export arguments for local filesystem mount calls. 602 */ 603 struct export_args { 604 uint64_t ex_flags; /* export related flags */ 605 uid_t ex_root; /* mapping for root uid */ 606 uid_t ex_uid; /* mapping for anonymous user */ 607 int ex_ngroups; 608 gid_t *ex_groups; 609 struct sockaddr *ex_addr; /* net address to which exported */ 610 u_char ex_addrlen; /* and the net address length */ 611 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 612 u_char ex_masklen; /* and the smask length */ 613 char *ex_indexfile; /* index file for WebNFS URLs */ 614 int ex_numsecflavors; /* security flavor count */ 615 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ 616 }; 617 618 /* 619 * Structure holding information for a publicly exported filesystem 620 * (WebNFS). Currently the specs allow just for one such filesystem. 621 */ 622 struct nfs_public { 623 int np_valid; /* Do we hold valid information */ 624 fhandle_t np_handle; /* Filehandle for pub fs (internal) */ 625 struct mount *np_mount; /* Mountpoint of exported fs */ 626 char *np_index; /* Index file */ 627 }; 628 629 /* 630 * Filesystem configuration information. One of these exists for each 631 * type of filesystem supported by the kernel. These are searched at 632 * mount time to identify the requested filesystem. 633 * 634 * XXX: Never change the first two arguments! 635 */ 636 struct vfsconf { 637 u_int vfc_version; /* ABI version number */ 638 char vfc_name[MFSNAMELEN]; /* filesystem type name */ 639 struct vfsops *vfc_vfsops; /* filesystem operations vector */ 640 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */ 641 int vfc_typenum; /* historic filesystem type number */ 642 int vfc_refcount; /* number mounted of this type */ 643 int vfc_flags; /* permanent flags */ 644 int vfc_prison_flag; /* prison allow.mount.* flag */ 645 struct vfsoptdecl *vfc_opts; /* mount options */ 646 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */ 647 }; 648 649 /* Userland version of the struct vfsconf. */ 650 struct xvfsconf { 651 struct vfsops *vfc_vfsops; /* filesystem operations vector */ 652 char vfc_name[MFSNAMELEN]; /* filesystem type name */ 653 int vfc_typenum; /* historic filesystem type number */ 654 int vfc_refcount; /* number mounted of this type */ 655 int vfc_flags; /* permanent flags */ 656 struct vfsconf *vfc_next; /* next in list */ 657 }; 658 659 #ifndef BURN_BRIDGES 660 struct ovfsconf { 661 void *vfc_vfsops; 662 char vfc_name[32]; 663 int vfc_index; 664 int vfc_refcount; 665 int vfc_flags; 666 }; 667 #endif 668 669 /* 670 * NB: these flags refer to IMPLEMENTATION properties, not properties of 671 * any actual mounts; i.e., it does not make sense to change the flags. 672 */ 673 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */ 674 #define VFCF_NETWORK 0x00020000 /* may get data over the network */ 675 #define VFCF_READONLY 0x00040000 /* writes are not implemented */ 676 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */ 677 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */ 678 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */ 679 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */ 680 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */ 681 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests 682 to kernel->user (AST) transition */ 683 #define VFCF_FILEMOUNT 0x02000000 /* allow mounting files */ 684 685 typedef uint32_t fsctlop_t; 686 687 struct vfsidctl { 688 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */ 689 fsid_t vc_fsid; /* fsid to operate on */ 690 char vc_fstypename[MFSNAMELEN]; 691 /* type of fs 'nfs' or '*' */ 692 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */ 693 void *vc_ptr; /* pointer to data structure */ 694 size_t vc_len; /* sizeof said structure */ 695 u_int32_t vc_spare[12]; /* spare (must be zero) */ 696 }; 697 698 /* vfsidctl API version. */ 699 #define VFS_CTL_VERS1 0x01 700 701 /* 702 * New style VFS sysctls, do not reuse/conflict with the namespace for 703 * private sysctls. 704 * All "global" sysctl ops have the 33rd bit set: 705 * 0x...1.... 706 * Private sysctl ops should have the 33rd bit unset. 707 */ 708 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */ 709 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */ 710 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */ 711 712 struct vfsquery { 713 u_int32_t vq_flags; 714 u_int32_t vq_spare[31]; 715 }; 716 717 /* vfsquery flags */ 718 #define VQ_NOTRESP 0x0001 /* server down */ 719 #define VQ_NEEDAUTH 0x0002 /* server bad auth */ 720 #define VQ_LOWDISK 0x0004 /* we're low on space */ 721 #define VQ_MOUNT 0x0008 /* new filesystem arrived */ 722 #define VQ_UNMOUNT 0x0010 /* filesystem has left */ 723 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */ 724 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external 725 program */ 726 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */ 727 #define VQ_FLAG0100 0x0100 /* placeholder */ 728 #define VQ_FLAG0200 0x0200 /* placeholder */ 729 #define VQ_FLAG0400 0x0400 /* placeholder */ 730 #define VQ_FLAG0800 0x0800 /* placeholder */ 731 #define VQ_FLAG1000 0x1000 /* placeholder */ 732 #define VQ_FLAG2000 0x2000 /* placeholder */ 733 #define VQ_FLAG4000 0x4000 /* placeholder */ 734 #define VQ_FLAG8000 0x8000 /* placeholder */ 735 736 #ifdef _KERNEL 737 /* Point a sysctl request at a vfsidctl's data. */ 738 #define VCTLTOREQ(vc, req) \ 739 do { \ 740 (req)->newptr = (vc)->vc_ptr; \ 741 (req)->newlen = (vc)->vc_len; \ 742 (req)->newidx = 0; \ 743 } while (0) 744 #endif 745 746 struct iovec; 747 struct uio; 748 749 #ifdef _KERNEL 750 751 /* 752 * vfs_busy specific flags and mask. 753 */ 754 #define MBF_NOWAIT 0x01 755 #define MBF_MNTLSTLOCK 0x02 756 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK) 757 758 #ifdef MALLOC_DECLARE 759 MALLOC_DECLARE(M_MOUNT); 760 MALLOC_DECLARE(M_STATFS); 761 #endif 762 extern int maxvfsconf; /* highest defined filesystem type */ 763 764 TAILQ_HEAD(vfsconfhead, vfsconf); 765 extern struct vfsconfhead vfsconf; 766 767 /* 768 * Operations supported on mounted filesystem. 769 */ 770 struct mount_args; 771 struct nameidata; 772 struct sysctl_req; 773 struct mntarg; 774 775 /* 776 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall. 777 * The new way is vfs_mount. 778 * 779 * vfs_cmount implementations typically translate arguments from their 780 * respective old per-FS structures into the key-value list supported by 781 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace. 782 * 783 * Filesystems with mounters that use nmount(2) do not need to and should not 784 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and 785 * mount(2) entirely. 786 */ 787 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags); 788 typedef int vfs_unmount_t(struct mount *mp, int mntflags); 789 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp); 790 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg, 791 bool *mp_busy); 792 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp); 793 typedef int vfs_sync_t(struct mount *mp, int waitfor); 794 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags, 795 struct vnode **vpp); 796 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp, 797 int flags, struct vnode **vpp); 798 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam, 799 uint64_t *extflagsp, struct ucred **credanonp, 800 int *numsecflavors, int *secflavors); 801 typedef int vfs_init_t(struct vfsconf *); 802 typedef int vfs_uninit_t(struct vfsconf *); 803 typedef int vfs_extattrctl_t(struct mount *mp, int cmd, 804 struct vnode *filename_vp, int attrnamespace, 805 const char *attrname); 806 typedef int vfs_mount_t(struct mount *mp); 807 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op, 808 struct sysctl_req *req); 809 typedef void vfs_susp_clean_t(struct mount *mp); 810 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp); 811 typedef void vfs_purge_t(struct mount *mp); 812 struct sbuf; 813 typedef int vfs_report_lockf_t(struct mount *mp, struct sbuf *sb); 814 815 struct vfsops { 816 vfs_mount_t *vfs_mount; 817 vfs_cmount_t *vfs_cmount; 818 vfs_unmount_t *vfs_unmount; 819 vfs_root_t *vfs_root; 820 vfs_root_t *vfs_cachedroot; 821 vfs_quotactl_t *vfs_quotactl; 822 vfs_statfs_t *vfs_statfs; 823 vfs_sync_t *vfs_sync; 824 vfs_vget_t *vfs_vget; 825 vfs_fhtovp_t *vfs_fhtovp; 826 vfs_checkexp_t *vfs_checkexp; 827 vfs_init_t *vfs_init; 828 vfs_uninit_t *vfs_uninit; 829 vfs_extattrctl_t *vfs_extattrctl; 830 vfs_sysctl_t *vfs_sysctl; 831 vfs_susp_clean_t *vfs_susp_clean; 832 vfs_notify_lowervp_t *vfs_reclaim_lowervp; 833 vfs_notify_lowervp_t *vfs_unlink_lowervp; 834 vfs_purge_t *vfs_purge; 835 vfs_report_lockf_t *vfs_report_lockf; 836 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */ 837 }; 838 839 vfs_statfs_t __vfs_statfs; 840 841 #define VFS_MOUNT(MP) ({ \ 842 int _rc; \ 843 \ 844 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\ 845 _rc = (*(MP)->mnt_op->vfs_mount)(MP); \ 846 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\ 847 _rc; }) 848 849 #define VFS_UNMOUNT(MP, FORCE) ({ \ 850 int _rc; \ 851 \ 852 _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \ 853 _rc; }) 854 855 #define VFS_ROOT(MP, FLAGS, VPP) ({ \ 856 int _rc; \ 857 \ 858 _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \ 859 _rc; }) 860 861 #define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \ 862 int _rc; \ 863 \ 864 _rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \ 865 _rc; }) 866 867 #define VFS_QUOTACTL(MP, C, U, A, MP_BUSY) ({ \ 868 int _rc; \ 869 \ 870 _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A, MP_BUSY); \ 871 _rc; }) 872 873 #define VFS_STATFS(MP, SBP) ({ \ 874 int _rc; \ 875 \ 876 _rc = __vfs_statfs((MP), (SBP)); \ 877 _rc; }) 878 879 #define VFS_SYNC(MP, WAIT) ({ \ 880 int _rc; \ 881 \ 882 _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \ 883 _rc; }) 884 885 #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \ 886 int _rc; \ 887 \ 888 _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \ 889 _rc; }) 890 891 #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \ 892 int _rc; \ 893 \ 894 _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \ 895 _rc; }) 896 897 #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \ 898 int _rc; \ 899 \ 900 _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\ 901 SEC); \ 902 _rc; }) 903 904 #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \ 905 int _rc; \ 906 \ 907 _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \ 908 _rc; }) 909 910 #define VFS_SYSCTL(MP, OP, REQ) ({ \ 911 int _rc; \ 912 \ 913 _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \ 914 _rc; }) 915 916 #define VFS_SUSP_CLEAN(MP) do { \ 917 if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \ 918 (*(MP)->mnt_op->vfs_susp_clean)(MP); \ 919 } \ 920 } while (0) 921 922 #define VFS_RECLAIM_LOWERVP(MP, VP) do { \ 923 if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \ 924 (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \ 925 } \ 926 } while (0) 927 928 #define VFS_UNLINK_LOWERVP(MP, VP) do { \ 929 if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \ 930 (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \ 931 } \ 932 } while (0) 933 934 #define VFS_PURGE(MP) do { \ 935 if (*(MP)->mnt_op->vfs_purge != NULL) { \ 936 (*(MP)->mnt_op->vfs_purge)(MP); \ 937 } \ 938 } while (0) 939 940 #define VFS_KNOTE_LOCKED(vp, hint) do \ 941 { \ 942 VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \ 943 } while (0) 944 945 #define VFS_KNOTE_UNLOCKED(vp, hint) do \ 946 { \ 947 VN_KNOTE((vp), (hint), 0); \ 948 } while (0) 949 950 #include <sys/module.h> 951 952 /* 953 * Version numbers. 954 */ 955 #define VFS_VERSION_00 0x19660120 956 #define VFS_VERSION_01 0x20121030 957 #define VFS_VERSION_02 0x20180504 958 #define VFS_VERSION VFS_VERSION_02 959 960 #define VFS_SET(vfsops, fsname, flags) \ 961 static struct vfsconf fsname ## _vfsconf = { \ 962 .vfc_version = VFS_VERSION, \ 963 .vfc_name = #fsname, \ 964 .vfc_vfsops = &vfsops, \ 965 .vfc_typenum = -1, \ 966 .vfc_flags = flags, \ 967 }; \ 968 static moduledata_t fsname ## _mod = { \ 969 #fsname, \ 970 vfs_modevent, \ 971 & fsname ## _vfsconf \ 972 }; \ 973 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE) 974 975 enum vfs_notify_upper_type { 976 VFS_NOTIFY_UPPER_RECLAIM, 977 VFS_NOTIFY_UPPER_UNLINK, 978 }; 979 980 /* 981 * exported vnode operations 982 */ 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_delete(struct prison *); 1021 int vfs_export /* process mount export info */ 1022 (struct mount *, struct export_args *, bool); 1023 void vfs_free_addrlist(struct netexport *); 1024 void vfs_allocate_syncvnode(struct mount *); 1025 void vfs_deallocate_syncvnode(struct mount *); 1026 int vfs_donmount(struct thread *td, uint64_t fsflags, 1027 struct uio *fsoptions); 1028 void vfs_getnewfsid(struct mount *); 1029 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */ 1030 struct mount *vfs_busyfs(fsid_t *); 1031 int vfs_modevent(module_t, int, void *); 1032 void vfs_mount_error(struct mount *, const char *, ...); 1033 void vfs_mountroot(void); /* mount our root filesystem */ 1034 void vfs_mountedfrom(struct mount *, const char *from); 1035 void vfs_notify_upper(struct vnode *, enum vfs_notify_upper_type); 1036 struct mount *vfs_ref_from_vp(struct vnode *); 1037 void vfs_ref(struct mount *); 1038 void vfs_rel(struct mount *); 1039 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *, 1040 struct ucred *); 1041 int vfs_suser(struct mount *, struct thread *); 1042 void vfs_unbusy(struct mount *); 1043 void vfs_unmountall(void); 1044 struct mount *vfs_register_upper_from_vp(struct vnode *, 1045 struct mount *ump, struct mount_upper_node *); 1046 void vfs_register_for_notification(struct mount *, struct mount *, 1047 struct mount_upper_node *); 1048 void vfs_unregister_for_notification(struct mount *, 1049 struct mount_upper_node *); 1050 void vfs_unregister_upper(struct mount *, struct mount_upper_node *); 1051 int vfs_remount_ro(struct mount *mp); 1052 int vfs_report_lockf(struct mount *mp, struct sbuf *sb); 1053 1054 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */ 1055 extern struct mtx_padalign mountlist_mtx; 1056 extern struct nfs_public nfs_pub; 1057 extern struct sx vfsconf_sx; 1058 #define vfsconf_lock() sx_xlock(&vfsconf_sx) 1059 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx) 1060 #define vfsconf_slock() sx_slock(&vfsconf_sx) 1061 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx) 1062 struct vnode *mntfs_allocvp(struct mount *, struct vnode *); 1063 void mntfs_freevp(struct vnode *); 1064 1065 /* 1066 * Declarations for these vfs default operations are located in 1067 * kern/vfs_default.c. They will be automatically used to replace 1068 * null entries in VFS ops tables when registering a new filesystem 1069 * type in the global table. 1070 */ 1071 vfs_root_t vfs_stdroot; 1072 vfs_quotactl_t vfs_stdquotactl; 1073 vfs_statfs_t vfs_stdstatfs; 1074 vfs_sync_t vfs_stdsync; 1075 vfs_sync_t vfs_stdnosync; 1076 vfs_vget_t vfs_stdvget; 1077 vfs_fhtovp_t vfs_stdfhtovp; 1078 vfs_checkexp_t vfs_stdcheckexp; 1079 vfs_init_t vfs_stdinit; 1080 vfs_uninit_t vfs_stduninit; 1081 vfs_extattrctl_t vfs_stdextattrctl; 1082 vfs_sysctl_t vfs_stdsysctl; 1083 1084 void syncer_suspend(void); 1085 void syncer_resume(void); 1086 1087 struct vnode *vfs_cache_root_clear(struct mount *); 1088 void vfs_cache_root_set(struct mount *, struct vnode *); 1089 1090 void vfs_op_barrier_wait(struct mount *); 1091 void vfs_op_enter(struct mount *); 1092 void vfs_op_exit_locked(struct mount *); 1093 void vfs_op_exit(struct mount *); 1094 1095 #ifdef DIAGNOSTIC 1096 void vfs_assert_mount_counters(struct mount *); 1097 void vfs_dump_mount_counters(struct mount *); 1098 #else 1099 #define vfs_assert_mount_counters(mp) do { } while (0) 1100 #define vfs_dump_mount_counters(mp) do { } while (0) 1101 #endif 1102 1103 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT }; 1104 int vfs_mount_fetch_counter(struct mount *, enum mount_counter); 1105 1106 void suspend_all_fs(void); 1107 void resume_all_fs(void); 1108 1109 /* 1110 * Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes 1111 * all CPUs not executing code enclosed by thread_in_ops_pcpu variable. 1112 * 1113 * This provides an invariant that by the time the last CPU is observed not 1114 * executing, everyone else entering will see the counter > 0 and exit. 1115 * 1116 * Note there is no barrier between vfs_ops and the rest of the code in the 1117 * section. It is not necessary as the writer has to wait for everyone to drain 1118 * before making any changes or only make changes safe while the section is 1119 * executed. 1120 */ 1121 #define vfs_mount_pcpu(mp) zpcpu_get(mp->mnt_pcpu) 1122 #define vfs_mount_pcpu_remote(mp, cpu) zpcpu_get_cpu(mp->mnt_pcpu, cpu) 1123 1124 #define vfs_op_thread_entered(mp) ({ \ 1125 MPASS(curthread->td_critnest > 0); \ 1126 struct mount_pcpu *_mpcpu = vfs_mount_pcpu(mp); \ 1127 _mpcpu->mntp_thread_in_ops == 1; \ 1128 }) 1129 1130 #define vfs_op_thread_enter_crit(mp, _mpcpu) ({ \ 1131 bool _retval_crit = true; \ 1132 MPASS(curthread->td_critnest > 0); \ 1133 _mpcpu = vfs_mount_pcpu(mp); \ 1134 MPASS(mpcpu->mntp_thread_in_ops == 0); \ 1135 _mpcpu->mntp_thread_in_ops = 1; \ 1136 atomic_interrupt_fence(); \ 1137 if (__predict_false(mp->mnt_vfs_ops > 0)) { \ 1138 vfs_op_thread_exit_crit(mp, _mpcpu); \ 1139 _retval_crit = false; \ 1140 } \ 1141 _retval_crit; \ 1142 }) 1143 1144 #define vfs_op_thread_enter(mp, _mpcpu) ({ \ 1145 bool _retval; \ 1146 critical_enter(); \ 1147 _retval = vfs_op_thread_enter_crit(mp, _mpcpu); \ 1148 if (__predict_false(!_retval)) \ 1149 critical_exit(); \ 1150 _retval; \ 1151 }) 1152 1153 #define vfs_op_thread_exit_crit(mp, _mpcpu) do { \ 1154 MPASS(_mpcpu == vfs_mount_pcpu(mp)); \ 1155 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1156 atomic_interrupt_fence(); \ 1157 _mpcpu->mntp_thread_in_ops = 0; \ 1158 } while (0) 1159 1160 #define vfs_op_thread_exit(mp, _mpcpu) do { \ 1161 vfs_op_thread_exit_crit(mp, _mpcpu); \ 1162 critical_exit(); \ 1163 } while (0) 1164 1165 #define vfs_mp_count_add_pcpu(_mpcpu, count, val) do { \ 1166 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1167 _mpcpu->mntp_##count += val; \ 1168 } while (0) 1169 1170 #define vfs_mp_count_sub_pcpu(_mpcpu, count, val) do { \ 1171 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1172 _mpcpu->mntp_##count -= val; \ 1173 } while (0) 1174 1175 #else /* !_KERNEL */ 1176 1177 #include <sys/cdefs.h> 1178 1179 struct stat; 1180 1181 __BEGIN_DECLS 1182 int fhlink(struct fhandle *, const char *); 1183 int fhlinkat(struct fhandle *, int, const char *); 1184 int fhopen(const struct fhandle *, int); 1185 int fhreadlink(struct fhandle *, char *, size_t); 1186 int fhstat(const struct fhandle *, struct stat *); 1187 int fhstatfs(const struct fhandle *, struct statfs *); 1188 int fstatfs(int, struct statfs *); 1189 int getfh(const char *, fhandle_t *); 1190 int getfhat(int, char *, struct fhandle *, int); 1191 int getfsstat(struct statfs *, long, int); 1192 int getmntinfo(struct statfs **, int); 1193 int lgetfh(const char *, fhandle_t *); 1194 int mount(const char *, const char *, int, void *); 1195 int nmount(struct iovec *, unsigned int, int); 1196 int statfs(const char *, struct statfs *); 1197 int unmount(const char *, int); 1198 1199 /* C library stuff */ 1200 int getvfsbyname(const char *, struct xvfsconf *); 1201 __END_DECLS 1202 1203 #endif /* _KERNEL */ 1204 1205 #endif /* !_SYS_MOUNT_H_ */ 1206