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null */ 134 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */ 135 136 /* XXX getfsstat.2 is out of date with write and read counter changes here. */ 137 /* XXX statfs.2 is out of date with read counter changes here.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