1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1989, 1991, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 #ifndef _SYS_MOUNT_H_ 33 #define _SYS_MOUNT_H_ 34 35 #include <sys/types.h> 36 #include <sys/ucred.h> 37 #include <sys/queue.h> 38 #ifdef _KERNEL 39 #include <sys/systm.h> 40 #include <sys/lock.h> 41 #include <sys/lockmgr.h> 42 #include <sys/proc.h> 43 #include <sys/tslog.h> 44 #include <sys/_mutex.h> 45 #include <sys/_sx.h> 46 #elif defined(_WANT_MOUNT) 47 #include <sys/_lock.h> 48 #include <sys/_lockmgr.h> 49 #include <sys/_mutex.h> 50 #endif 51 52 /* 53 * NOTE: When changing statfs structure, mount structure, MNT_* flags or 54 * MNTK_* flags also update DDB show mount command in vfs_subr.c. 55 */ 56 57 typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */ 58 59 /* Returns non-zero if fsids are different. */ 60 static __inline int 61 fsidcmp(const fsid_t *a, const fsid_t *b) 62 { 63 return (a->val[0] != b->val[0] || a->val[1] != b->val[1]); 64 } 65 66 /* 67 * File identifier. 68 * These are unique per filesystem on a single machine. 69 * 70 * Note that the offset of fid_data is 4 bytes, so care must be taken to avoid 71 * undefined behavior accessing unaligned fields within an embedded struct. 72 */ 73 #define MAXFIDSZ 16 74 75 struct fid { 76 u_short fid_len; /* length of data in bytes */ 77 u_short fid_data0; /* force longword alignment */ 78 char fid_data[MAXFIDSZ]; /* data (variable length) */ 79 }; 80 81 /* 82 * filesystem statistics 83 */ 84 #define MFSNAMELEN 16 /* length of type name including null */ 85 #define MNAMELEN 1024 /* size of on/from name bufs */ 86 #define STATFS_VERSION 0x20140518 /* current version number */ 87 struct statfs { 88 uint32_t f_version; /* structure version number */ 89 uint32_t f_type; /* type of filesystem */ 90 uint64_t f_flags; /* copy of mount exported flags */ 91 uint64_t f_bsize; /* filesystem fragment size */ 92 uint64_t f_iosize; /* optimal transfer block size */ 93 uint64_t f_blocks; /* total data blocks in filesystem */ 94 uint64_t f_bfree; /* free blocks in filesystem */ 95 int64_t f_bavail; /* free blocks avail to non-superuser */ 96 uint64_t f_files; /* total file nodes in filesystem */ 97 int64_t f_ffree; /* free nodes avail to non-superuser */ 98 uint64_t f_syncwrites; /* count of sync writes since mount */ 99 uint64_t f_asyncwrites; /* count of async writes since mount */ 100 uint64_t f_syncreads; /* count of sync reads since mount */ 101 uint64_t f_asyncreads; /* count of async reads since mount */ 102 uint32_t f_nvnodelistsize; /* # of vnodes */ 103 uint32_t f_spare0; /* unused spare */ 104 uint64_t f_spare[9]; /* unused spare */ 105 uint32_t f_namemax; /* maximum filename length */ 106 uid_t f_owner; /* user that mounted the filesystem */ 107 fsid_t f_fsid; /* filesystem id */ 108 char f_charspare[80]; /* spare string space */ 109 char f_fstypename[MFSNAMELEN]; /* filesystem type name */ 110 char f_mntfromname[MNAMELEN]; /* mounted filesystem */ 111 char f_mntonname[MNAMELEN]; /* directory on which mounted */ 112 }; 113 114 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL) 115 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */ 116 struct freebsd11_statfs { 117 uint32_t f_version; /* structure version number */ 118 uint32_t f_type; /* type of filesystem */ 119 uint64_t f_flags; /* copy of mount exported flags */ 120 uint64_t f_bsize; /* filesystem fragment size */ 121 uint64_t f_iosize; /* optimal transfer block size */ 122 uint64_t f_blocks; /* total data blocks in filesystem */ 123 uint64_t f_bfree; /* free blocks in filesystem */ 124 int64_t f_bavail; /* free blocks avail to non-superuser */ 125 uint64_t f_files; /* total file nodes in filesystem */ 126 int64_t f_ffree; /* free nodes avail to non-superuser */ 127 uint64_t f_syncwrites; /* count of sync writes since mount */ 128 uint64_t f_asyncwrites; /* count of async writes since mount */ 129 uint64_t f_syncreads; /* count of sync reads since mount */ 130 uint64_t f_asyncreads; /* count of async reads since mount */ 131 uint64_t f_spare[10]; /* unused spare */ 132 uint32_t f_namemax; /* maximum filename length */ 133 uid_t f_owner; /* user that mounted the filesystem */ 134 fsid_t f_fsid; /* filesystem id */ 135 char f_charspare[80]; /* spare string space */ 136 char f_fstypename[16]; /* filesystem type name */ 137 char f_mntfromname[88]; /* mounted filesystem */ 138 char f_mntonname[88]; /* directory on which mounted */ 139 }; 140 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */ 141 142 #ifdef _KERNEL 143 #define OMFSNAMELEN 16 /* length of fs type name, including null */ 144 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */ 145 146 /* XXX getfsstat.2 is out of date with write and read counter changes here. */ 147 /* XXX statfs.2 is out of date with read counter changes here. */ 148 struct ostatfs { 149 long f_spare2; /* placeholder */ 150 long f_bsize; /* fundamental filesystem block size */ 151 long f_iosize; /* optimal transfer block size */ 152 long f_blocks; /* total data blocks in filesystem */ 153 long f_bfree; /* free blocks in fs */ 154 long f_bavail; /* free blocks avail to non-superuser */ 155 long f_files; /* total file nodes in filesystem */ 156 long f_ffree; /* free file nodes in fs */ 157 fsid_t f_fsid; /* filesystem id */ 158 uid_t f_owner; /* user that mounted the filesystem */ 159 int f_type; /* type of filesystem */ 160 int f_flags; /* copy of mount exported flags */ 161 long f_syncwrites; /* count of sync writes since mount */ 162 long f_asyncwrites; /* count of async writes since mount */ 163 char f_fstypename[OMFSNAMELEN]; /* fs type name */ 164 char f_mntonname[OMNAMELEN]; /* directory on which mounted */ 165 long f_syncreads; /* count of sync reads since mount */ 166 long f_asyncreads; /* count of async reads since mount */ 167 short f_spares1; /* unused spare */ 168 char f_mntfromname[OMNAMELEN];/* mounted filesystem */ 169 short f_spares2; /* unused spare */ 170 /* 171 * XXX on machines where longs are aligned to 8-byte boundaries, there 172 * is an unnamed int32_t here. This spare was after the apparent end 173 * of the struct until we bit off the read counters from f_mntonname. 174 */ 175 long f_spare[2]; /* unused spare */ 176 }; 177 #endif /* _KERNEL */ 178 179 #if defined(_WANT_MOUNT) || defined(_KERNEL) 180 TAILQ_HEAD(vnodelst, vnode); 181 182 /* Mount options list */ 183 TAILQ_HEAD(vfsoptlist, vfsopt); 184 struct vfsopt { 185 TAILQ_ENTRY(vfsopt) link; 186 char *name; 187 void *value; 188 int len; 189 int pos; 190 int seen; 191 }; 192 193 struct mount_pcpu { 194 int mntp_thread_in_ops; 195 int mntp_ref; 196 int mntp_lockref; 197 int mntp_writeopcount; 198 }; 199 200 _Static_assert(sizeof(struct mount_pcpu) == 16, 201 "the struct is allocated from pcpu 16 zone"); 202 203 /* 204 * Structure for tracking a stacked filesystem mounted above another 205 * filesystem. This is expected to be stored in the upper FS' per-mount data. 206 * 207 * Lock reference: 208 * i - lower mount interlock 209 * c - constant from node initialization 210 */ 211 struct mount_upper_node { 212 struct mount *mp; /* (c) mount object for upper FS */ 213 TAILQ_ENTRY(mount_upper_node) mnt_upper_link; /* (i) position in uppers list */ 214 }; 215 216 /* 217 * Structure per mounted filesystem. Each mounted filesystem has an 218 * array of operations and an instance record. The filesystems are 219 * put on a doubly linked list. 220 * 221 * Lock reference: 222 * l - mnt_listmtx 223 * m - mountlist_mtx 224 * i - interlock 225 * v - vnode freelist mutex 226 * d - deferred unmount list mutex 227 * e - mnt_explock 228 * 229 * Unmarked fields are considered stable as long as a ref is held. 230 * 231 */ 232 struct mount { 233 int mnt_vfs_ops; /* (i) pending vfs ops */ 234 int mnt_kern_flag; /* (i) kernel only flags */ 235 uint64_t mnt_flag; /* (i) flags shared with user */ 236 struct mount_pcpu *mnt_pcpu; /* per-CPU data */ 237 struct vnode *mnt_rootvnode; 238 struct vnode *mnt_vnodecovered; /* vnode we mounted on */ 239 struct vfsops *mnt_op; /* operations on fs */ 240 struct vfsconf *mnt_vfc; /* configuration info */ 241 struct mtx __aligned(CACHE_LINE_SIZE) mnt_mtx; /* mount structure interlock */ 242 int mnt_gen; /* struct mount generation */ 243 #define mnt_startzero mnt_list 244 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */ 245 struct vnode *mnt_syncer; /* syncer vnode */ 246 int mnt_ref; /* (i) Reference count */ 247 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */ 248 int mnt_nvnodelistsize; /* (i) # of vnodes */ 249 int mnt_writeopcount; /* (i) write syscalls pending */ 250 struct vfsoptlist *mnt_opt; /* current mount options */ 251 struct vfsoptlist *mnt_optnew; /* new options passed to fs */ 252 struct statfs mnt_stat; /* cache of filesystem stats */ 253 struct ucred *mnt_cred; /* credentials of mounter */ 254 void * mnt_data; /* private data */ 255 time_t mnt_time; /* last time written*/ 256 int mnt_iosize_max; /* max size for clusters, etc */ 257 struct netexport *mnt_export; /* (e) export list */ 258 struct label *mnt_label; /* MAC label for the fs */ 259 u_int mnt_hashseed; /* Random seed for vfs_hash */ 260 int mnt_lockref; /* (i) Lock reference count */ 261 int mnt_secondary_writes; /* (i) # of secondary writes */ 262 int mnt_secondary_accwrites;/* (i) secondary wr. starts */ 263 struct thread *mnt_susp_owner; /* (i) thread owning suspension */ 264 struct ucred *mnt_exjail; /* (i) jail which did exports */ 265 #define mnt_endzero mnt_gjprovider 266 char *mnt_gjprovider; /* gjournal provider name */ 267 struct mtx mnt_listmtx; 268 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */ 269 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */ 270 int mnt_upper_pending; /* (i) # of pending ops on mnt_uppers */ 271 struct lock mnt_explock; /* vfs_export walkers lock */ 272 struct lock mnt_renamelock; /* renames and O_RESOLVE_BENEATH */ 273 TAILQ_HEAD(, mount_upper_node) mnt_uppers; /* (i) upper mounts over us */ 274 TAILQ_HEAD(, mount_upper_node) mnt_notify; /* (i) upper mounts for notification */ 275 STAILQ_ENTRY(mount) mnt_taskqueue_link; /* (d) our place in deferred unmount list */ 276 uint64_t mnt_taskqueue_flags; /* (d) unmount flags passed from taskqueue */ 277 unsigned int mnt_unmount_retries; /* (d) # of failed deferred unmount attempts */ 278 }; 279 #endif /* _WANT_MOUNT || _KERNEL */ 280 281 #ifdef _KERNEL 282 /* 283 * Definitions for MNT_VNODE_FOREACH_ALL. 284 */ 285 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp); 286 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp); 287 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp); 288 289 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \ 290 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \ 291 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp))) 292 293 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \ 294 do { \ 295 MNT_ILOCK(mp); \ 296 __mnt_vnode_markerfree_all(&(mvp), (mp)); \ 297 /* MNT_IUNLOCK(mp); -- done in above function */ \ 298 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \ 299 } while (0) 300 301 /* 302 * Definitions for MNT_VNODE_FOREACH_LAZY. 303 */ 304 typedef int mnt_lazy_cb_t(struct vnode *, void *); 305 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp, 306 mnt_lazy_cb_t *cb, void *cbarg); 307 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp, 308 mnt_lazy_cb_t *cb, void *cbarg); 309 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp); 310 311 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \ 312 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \ 313 (vp) != NULL; \ 314 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg))) 315 316 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \ 317 __mnt_vnode_markerfree_lazy(&(mvp), (mp)) 318 319 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx) 320 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx) 321 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx) 322 #define MNT_MTX(mp) (&(mp)->mnt_mtx) 323 324 #define MNT_REF(mp) do { \ 325 mtx_assert(MNT_MTX(mp), MA_OWNED); \ 326 mp->mnt_ref++; \ 327 } while (0) 328 #define MNT_REL(mp) do { \ 329 mtx_assert(MNT_MTX(mp), MA_OWNED); \ 330 (mp)->mnt_ref--; \ 331 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \ 332 vfs_dump_mount_counters(mp); \ 333 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \ 334 wakeup((mp)); \ 335 } while (0) 336 337 #endif /* _KERNEL */ 338 339 #if defined(_WANT_MNTOPTNAMES) || defined(_KERNEL) 340 struct mntoptnames { 341 uint64_t o_opt; 342 const char *o_name; 343 }; 344 #define MNTOPT_NAMES \ 345 { MNT_ASYNC, "asynchronous" }, \ 346 { MNT_EXPORTED, "NFS exported" }, \ 347 { MNT_LOCAL, "local" }, \ 348 { MNT_NOATIME, "noatime" }, \ 349 { MNT_NOEXEC, "noexec" }, \ 350 { MNT_NOSUID, "nosuid" }, \ 351 { MNT_NOSYMFOLLOW, "nosymfollow" }, \ 352 { MNT_QUOTA, "with quotas" }, \ 353 { MNT_RDONLY, "read-only" }, \ 354 { MNT_SYNCHRONOUS, "synchronous" }, \ 355 { MNT_UNION, "union" }, \ 356 { MNT_NOCLUSTERR, "noclusterr" }, \ 357 { MNT_NOCLUSTERW, "noclusterw" }, \ 358 { MNT_SUIDDIR, "suiddir" }, \ 359 { MNT_SOFTDEP, "soft-updates" }, \ 360 { MNT_SUJ, "journaled soft-updates" }, \ 361 { MNT_MULTILABEL, "multilabel" }, \ 362 { MNT_ACLS, "acls" }, \ 363 { MNT_NFS4ACLS, "nfsv4acls" }, \ 364 { MNT_GJOURNAL, "gjournal" }, \ 365 { MNT_AUTOMOUNTED, "automounted" }, \ 366 { MNT_VERIFIED, "verified" }, \ 367 { MNT_UNTRUSTED, "untrusted" }, \ 368 { MNT_NOCOVER, "nocover" }, \ 369 { MNT_EMPTYDIR, "emptydir" }, \ 370 { MNT_UPDATE, "update" }, \ 371 { MNT_DELEXPORT, "delexport" }, \ 372 { MNT_RELOAD, "reload" }, \ 373 { MNT_FORCE, "force" }, \ 374 { MNT_SNAPSHOT, "snapshot" }, \ 375 { MNT_NAMEDATTR, "named attributes" }, \ 376 { 0, NULL } 377 #endif 378 379 /* 380 * User specifiable flags, stored in mnt_flag. 381 */ 382 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */ 383 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */ 384 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */ 385 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */ 386 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */ 387 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */ 388 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */ 389 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */ 390 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */ 391 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */ 392 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */ 393 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */ 394 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */ 395 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */ 396 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */ 397 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */ 398 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */ 399 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */ 400 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */ 401 #define MNT_NAMEDATTR 0x0000020000000000ULL /* named attributes enabled */ 402 403 /* 404 * NFS export related mount flags. 405 */ 406 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */ 407 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */ 408 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */ 409 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */ 410 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */ 411 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */ 412 #define MNT_EXTLS 0x0000004000000000ULL /* require TLS */ 413 #define MNT_EXTLSCERT 0x0000008000000000ULL /* require TLS with client cert */ 414 #define MNT_EXTLSCERTUSER 0x0000010000000000ULL /* require TLS with user cert */ 415 416 /* 417 * Flags set by internal operations, but visible to the user. 418 */ 419 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */ 420 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */ 421 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */ 422 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */ 423 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */ 424 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */ 425 426 /* 427 * Mask of flags that are visible to statfs(). 428 * XXX I think that this could now become (~(MNT_CMDFLAGS)) 429 * but the 'mount' program may need changing to handle this. 430 */ 431 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \ 432 MNT_NOSUID | MNT_UNION | MNT_SUJ | \ 433 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \ 434 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \ 435 MNT_LOCAL | MNT_USER | MNT_QUOTA | \ 436 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \ 437 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \ 438 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \ 439 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \ 440 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \ 441 MNT_UNTRUSTED | MNT_NAMEDATTR) 442 443 /* Mask of flags that can be updated. */ 444 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \ 445 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \ 446 MNT_NOATIME | \ 447 MNT_NOSYMFOLLOW | MNT_IGNORE | \ 448 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \ 449 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \ 450 MNT_AUTOMOUNTED | MNT_UNTRUSTED) 451 452 /* 453 * External filesystem command modifier flags. 454 * Unmount can use the MNT_FORCE flag. 455 * XXX: These are not STATES and really should be somewhere else. 456 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL, 457 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2), 458 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2), 459 * it's harmless. 460 */ 461 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */ 462 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */ 463 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */ 464 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */ 465 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */ 466 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */ 467 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */ 468 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */ 469 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */ 470 #define MNT_RECURSE 0x0000100000000000ULL /* recursively unmount uppers */ 471 #define MNT_DEFERRED 0x0000200000000000ULL /* unmount in async context */ 472 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \ 473 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \ 474 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR | \ 475 MNT_RECURSE | MNT_DEFERRED) 476 477 /* 478 * Internal filesystem control flags stored in mnt_kern_flag. 479 * 480 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot 481 * proceed past the mount point. This keeps the subtree stable during 482 * mounts and unmounts. When non-forced unmount flushes all vnodes 483 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from 484 * queueing new vnodes. 485 * 486 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while 487 * dounmount() is still waiting to lock the mountpoint. This allows 488 * the filesystem to cancel operations that might otherwise deadlock 489 * with the unmount attempt (used by NFS). 490 */ 491 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */ 492 #define MNTK_ASYNC 0x00000002 /* filtered async flag */ 493 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */ 494 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */ 495 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */ 496 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */ 497 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */ 498 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */ 499 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads 500 and writes. Filesystem shall 501 properly handle i/o state on 502 EFAULT. */ 503 #define MNTK_RECURSE 0x00000200 /* pending recursive unmount */ 504 #define MNTK_UPPER_WAITER 0x00000400 /* waiting to drain MNTK_UPPER_PENDING */ 505 /* UNUSED 0x00000800 */ 506 #define MNTK_UNLOCKED_INSMNTQUE 0x00001000 /* fs does not lock the vnode for 507 insmntque */ 508 #define MNTK_UNMAPPED_BUFS 0x00002000 509 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */ 510 /* UNUSED 0x00008000 */ 511 #define MNTK_VMSETSIZE_BUG 0x00010000 512 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */ 513 #define MNTK_FPLOOKUP 0x00040000 /* fast path lookup is supported */ 514 #define MNTK_SUSPEND_ALL 0x00080000 /* Suspended by all-fs suspension */ 515 #define MNTK_TASKQUEUE_WAITER 0x00100000 /* Waiting on unmount taskqueue */ 516 /* UNUSED 0x00200000 */ 517 /* UNUSED 0x00400000 */ 518 #define MNTK_NOASYNC 0x00800000 /* disable async */ 519 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */ 520 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */ 521 #define MNTK_SUSPEND 0x08000000 /* request write suspension */ 522 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */ 523 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */ 524 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs 525 mounts over this fs */ 526 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */ 527 /* UNUSED 0x80000000 */ 528 529 #ifdef _KERNEL 530 static inline int 531 MNT_SHARED_WRITES(struct mount *mp) 532 { 533 534 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0); 535 } 536 537 static inline int 538 MNT_EXTENDED_SHARED(struct mount *mp) 539 { 540 541 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0); 542 } 543 #endif 544 545 /* 546 * Sysctl CTL_VFS definitions. 547 * 548 * Second level identifier specifies which filesystem. Second level 549 * identifier VFS_VFSCONF returns information about all filesystems. 550 * Second level identifier VFS_GENERIC is non-terminal. 551 */ 552 #define VFS_VFSCONF 0 /* get configured filesystems */ 553 #define VFS_GENERIC 0 /* generic filesystem information */ 554 /* 555 * Third level identifiers for VFS_GENERIC are given below; third 556 * level identifiers for specific filesystems are given in their 557 * mount specific header files. 558 */ 559 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */ 560 #define VFS_CONF 2 /* struct: vfsconf for filesystem given 561 as next argument */ 562 563 /* 564 * Flags for various system call interfaces. 565 * 566 * waitfor flags to vfs_sync() and getfsstat() 567 */ 568 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */ 569 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */ 570 #define MNT_LAZY 3 /* push data not written by filesystem syncer */ 571 #define MNT_SUSPEND 4 /* Suspend file system after sync */ 572 573 /* 574 * Generic file handle 575 */ 576 struct fhandle { 577 fsid_t fh_fsid; /* Filesystem id of mount point */ 578 struct fid fh_fid; /* Filesys specific id */ 579 }; 580 typedef struct fhandle fhandle_t; 581 582 /* 583 * Old export arguments without security flavor list 584 */ 585 struct oexport_args { 586 int ex_flags; /* export related flags */ 587 uid_t ex_root; /* mapping for root uid */ 588 struct xucred ex_anon; /* mapping for anonymous user */ 589 struct sockaddr *ex_addr; /* net address to which exported */ 590 u_char ex_addrlen; /* and the net address length */ 591 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 592 u_char ex_masklen; /* and the smask length */ 593 char *ex_indexfile; /* index file for WebNFS URLs */ 594 }; 595 596 /* 597 * Not quite so old export arguments with 32bit ex_flags and xucred ex_anon. 598 */ 599 #define MAXSECFLAVORS 5 600 struct o2export_args { 601 int ex_flags; /* export related flags */ 602 uid_t ex_root; /* mapping for root uid */ 603 struct xucred ex_anon; /* mapping for anonymous user */ 604 struct sockaddr *ex_addr; /* net address to which exported */ 605 u_char ex_addrlen; /* and the net address length */ 606 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 607 u_char ex_masklen; /* and the smask length */ 608 char *ex_indexfile; /* index file for WebNFS URLs */ 609 int ex_numsecflavors; /* security flavor count */ 610 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ 611 }; 612 613 /* 614 * Export arguments for local filesystem mount calls. 615 */ 616 struct export_args { 617 uint64_t ex_flags; /* export related flags */ 618 uid_t ex_root; /* mapping for root uid */ 619 uid_t ex_uid; /* mapping for anonymous user */ 620 int ex_ngroups; 621 gid_t *ex_groups; 622 struct sockaddr *ex_addr; /* net address to which exported */ 623 u_char ex_addrlen; /* and the net address length */ 624 struct sockaddr *ex_mask; /* mask of valid bits in saddr */ 625 u_char ex_masklen; /* and the smask length */ 626 char *ex_indexfile; /* index file for WebNFS URLs */ 627 int ex_numsecflavors; /* security flavor count */ 628 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ 629 }; 630 631 /* 632 * Structure holding information for a publicly exported filesystem 633 * (WebNFS). Currently the specs allow just for one such filesystem. 634 */ 635 struct nfs_public { 636 int np_valid; /* Do we hold valid information */ 637 fhandle_t np_handle; /* Filehandle for pub fs (internal) */ 638 struct mount *np_mount; /* Mountpoint of exported fs */ 639 char *np_index; /* Index file */ 640 }; 641 642 /* 643 * Filesystem configuration information. One of these exists for each 644 * type of filesystem supported by the kernel. These are searched at 645 * mount time to identify the requested filesystem. 646 * 647 * XXX: Never change the first two arguments! 648 */ 649 struct vfsconf { 650 u_int vfc_version; /* ABI version number */ 651 char vfc_name[MFSNAMELEN]; /* filesystem type name */ 652 struct vfsops *vfc_vfsops; /* filesystem operations vector */ 653 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */ 654 int vfc_typenum; /* historic filesystem type number */ 655 int vfc_refcount; /* number mounted of this type */ 656 int vfc_flags; /* permanent flags */ 657 int vfc_prison_flag; /* prison allow.mount.* flag */ 658 struct vfsoptdecl *vfc_opts; /* mount options */ 659 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */ 660 }; 661 662 /* Userland version of the struct vfsconf. */ 663 struct xvfsconf { 664 struct vfsops *vfc_vfsops; /* filesystem operations vector */ 665 char vfc_name[MFSNAMELEN]; /* filesystem type name */ 666 int vfc_typenum; /* historic filesystem type number */ 667 int vfc_refcount; /* number mounted of this type */ 668 int vfc_flags; /* permanent flags */ 669 struct vfsconf *vfc_next; /* next in list */ 670 }; 671 672 #ifndef BURN_BRIDGES 673 struct ovfsconf { 674 void *vfc_vfsops; 675 char vfc_name[32]; 676 int vfc_index; 677 int vfc_refcount; 678 int vfc_flags; 679 }; 680 #endif 681 682 /* 683 * NB: these flags refer to IMPLEMENTATION properties, not properties of 684 * any actual mounts; i.e., it does not make sense to change the flags. 685 */ 686 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */ 687 #define VFCF_NETWORK 0x00020000 /* may get data over the network */ 688 #define VFCF_READONLY 0x00040000 /* writes are not implemented */ 689 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */ 690 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */ 691 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */ 692 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */ 693 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */ 694 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests 695 to kernel->user (AST) transition */ 696 #define VFCF_FILEMOUNT 0x02000000 /* allow mounting files */ 697 #define VFCF_FILEREVINC 0x04000000 /* va_filerev is incr. by one */ 698 #define VFCF_FILEREVCT 0x08000000 /* va_filerev is set to ctime */ 699 700 typedef uint32_t fsctlop_t; 701 702 struct vfsidctl { 703 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */ 704 fsid_t vc_fsid; /* fsid to operate on */ 705 char vc_fstypename[MFSNAMELEN]; 706 /* type of fs 'nfs' or '*' */ 707 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */ 708 void *vc_ptr; /* pointer to data structure */ 709 size_t vc_len; /* sizeof said structure */ 710 u_int32_t vc_spare[12]; /* spare (must be zero) */ 711 }; 712 713 /* vfsidctl API version. */ 714 #define VFS_CTL_VERS1 0x01 715 716 /* 717 * New style VFS sysctls, do not reuse/conflict with the namespace for 718 * private sysctls. 719 * All "global" sysctl ops have the 33rd bit set: 720 * 0x...1.... 721 * Private sysctl ops should have the 33rd bit unset. 722 */ 723 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */ 724 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */ 725 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */ 726 727 struct vfsquery { 728 u_int32_t vq_flags; 729 u_int32_t vq_spare[31]; 730 }; 731 732 /* vfsquery flags */ 733 #define VQ_NOTRESP 0x0001 /* server down */ 734 #define VQ_NEEDAUTH 0x0002 /* server bad auth */ 735 #define VQ_LOWDISK 0x0004 /* we're low on space */ 736 #define VQ_MOUNT 0x0008 /* new filesystem arrived */ 737 #define VQ_UNMOUNT 0x0010 /* filesystem has left */ 738 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */ 739 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external 740 program */ 741 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */ 742 #define VQ_FLAG0100 0x0100 /* placeholder */ 743 #define VQ_FLAG0200 0x0200 /* placeholder */ 744 #define VQ_FLAG0400 0x0400 /* placeholder */ 745 #define VQ_FLAG0800 0x0800 /* placeholder */ 746 #define VQ_FLAG1000 0x1000 /* placeholder */ 747 #define VQ_FLAG2000 0x2000 /* placeholder */ 748 #define VQ_FLAG4000 0x4000 /* placeholder */ 749 #define VQ_FLAG8000 0x8000 /* placeholder */ 750 751 #ifdef _KERNEL 752 /* Point a sysctl request at a vfsidctl's data. */ 753 #define VCTLTOREQ(vc, req) \ 754 do { \ 755 (req)->newptr = (vc)->vc_ptr; \ 756 (req)->newlen = (vc)->vc_len; \ 757 (req)->newidx = 0; \ 758 } while (0) 759 #endif 760 761 struct iovec; 762 struct uio; 763 764 #ifdef _KERNEL 765 766 /* 767 * vfs_busy specific flags and mask. 768 */ 769 #define MBF_NOWAIT 0x01 770 #define MBF_MNTLSTLOCK 0x02 771 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK) 772 773 #ifdef MALLOC_DECLARE 774 MALLOC_DECLARE(M_MOUNT); 775 MALLOC_DECLARE(M_STATFS); 776 #endif 777 extern int maxvfsconf; /* highest defined filesystem type */ 778 779 TAILQ_HEAD(vfsconfhead, vfsconf); 780 extern struct vfsconfhead vfsconf; 781 782 /* 783 * Operations supported on mounted filesystem. 784 */ 785 struct mount_args; 786 struct nameidata; 787 struct sysctl_req; 788 struct mntarg; 789 790 /* 791 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall. 792 * The new way is vfs_mount. 793 * 794 * vfs_cmount implementations typically translate arguments from their 795 * respective old per-FS structures into the key-value list supported by 796 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace. 797 * 798 * Filesystems with mounters that use nmount(2) do not need to and should not 799 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and 800 * mount(2) entirely. 801 */ 802 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags); 803 typedef int vfs_unmount_t(struct mount *mp, int mntflags); 804 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp); 805 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg, 806 bool *mp_busy); 807 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp); 808 typedef int vfs_sync_t(struct mount *mp, int waitfor); 809 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags, 810 struct vnode **vpp); 811 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp, 812 int flags, struct vnode **vpp); 813 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam, 814 uint64_t *extflagsp, struct ucred **credanonp, 815 int *numsecflavors, int *secflavors); 816 typedef int vfs_init_t(struct vfsconf *); 817 typedef int vfs_uninit_t(struct vfsconf *); 818 typedef int vfs_extattrctl_t(struct mount *mp, int cmd, 819 struct vnode *filename_vp, int attrnamespace, 820 const char *attrname); 821 typedef int vfs_mount_t(struct mount *mp); 822 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op, 823 struct sysctl_req *req); 824 typedef void vfs_susp_clean_t(struct mount *mp); 825 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp); 826 typedef void vfs_purge_t(struct mount *mp); 827 struct sbuf; 828 typedef int vfs_report_lockf_t(struct mount *mp, struct sbuf *sb); 829 830 struct vfsops { 831 vfs_mount_t *vfs_mount; 832 vfs_cmount_t *vfs_cmount; 833 vfs_unmount_t *vfs_unmount; 834 vfs_root_t *vfs_root; 835 vfs_root_t *vfs_cachedroot; 836 vfs_quotactl_t *vfs_quotactl; 837 vfs_statfs_t *vfs_statfs; 838 vfs_sync_t *vfs_sync; 839 vfs_vget_t *vfs_vget; 840 vfs_fhtovp_t *vfs_fhtovp; 841 vfs_checkexp_t *vfs_checkexp; 842 vfs_init_t *vfs_init; 843 vfs_uninit_t *vfs_uninit; 844 vfs_extattrctl_t *vfs_extattrctl; 845 vfs_sysctl_t *vfs_sysctl; 846 vfs_susp_clean_t *vfs_susp_clean; 847 vfs_notify_lowervp_t *vfs_reclaim_lowervp; 848 vfs_notify_lowervp_t *vfs_unlink_lowervp; 849 vfs_purge_t *vfs_purge; 850 vfs_report_lockf_t *vfs_report_lockf; 851 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */ 852 }; 853 854 vfs_statfs_t __vfs_statfs; 855 856 static inline int 857 VFS_MOUNT(struct mount *mp) 858 { 859 int rc; 860 861 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", mp->mnt_vfc->vfc_name); 862 rc = mp->mnt_op->vfs_mount(mp); 863 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", mp->mnt_vfc->vfc_name); 864 return (rc); 865 } 866 867 static inline int 868 VFS_UNMOUNT(struct mount *mp, int force) 869 { 870 return (mp->mnt_op->vfs_unmount(mp, force)); 871 } 872 873 static inline int 874 VFS_ROOT(struct mount *mp, int flags, struct vnode **vpp) 875 { 876 return (mp->mnt_op->vfs_root(mp, flags, vpp)); 877 } 878 879 static inline int 880 VFS_CACHEDROOT(struct mount *mp, int flags, struct vnode **vpp) 881 { 882 return (mp->mnt_op->vfs_cachedroot(mp, flags, vpp)); 883 } 884 885 static inline int 886 VFS_QUOTACTL(struct mount *mp, int cmds, uid_t uid, void *arg, bool *mp_busy) 887 { 888 return (mp->mnt_op->vfs_quotactl(mp, cmds, uid, arg, mp_busy)); 889 } 890 891 static inline int 892 VFS_STATFS(struct mount *mp, struct statfs *sbp) 893 { 894 return (__vfs_statfs(mp, sbp)); 895 } 896 897 static inline int 898 VFS_SYNC(struct mount *mp, int waitfor) 899 { 900 return (mp->mnt_op->vfs_sync(mp, waitfor)); 901 } 902 903 static inline int 904 VFS_VGET(struct mount *mp, ino_t ino, int flags, struct vnode **vpp) 905 { 906 return (mp->mnt_op->vfs_vget(mp, ino, flags, vpp)); 907 } 908 909 static inline int 910 VFS_FHTOVP(struct mount *mp, struct fid *fidp, int flags, struct vnode **vpp) 911 { 912 return (mp->mnt_op->vfs_fhtovp(mp, fidp, flags, vpp)); 913 } 914 915 static inline int 916 VFS_CHECKEXP(struct mount *mp, struct sockaddr *nam, uint64_t *extflagsp, 917 struct ucred **credanonp, int *numsecflavors, int *secflavors) 918 { 919 return (mp->mnt_op->vfs_checkexp(mp, nam, extflagsp, credanonp, 920 numsecflavors, secflavors)); 921 } 922 923 static inline int 924 VFS_EXTATTRCTL(struct mount *mp, int cmd, struct vnode *filename_vp, 925 int attrnamespace, const char *attrname) 926 { 927 return (mp->mnt_op->vfs_extattrctl(mp, cmd, filename_vp, 928 attrnamespace, attrname)); 929 } 930 931 static inline int 932 VFS_SYSCTL(struct mount *mp, fsctlop_t op, struct sysctl_req *req) 933 { 934 return (mp->mnt_op->vfs_sysctl(mp, op, req)); 935 } 936 937 static inline void 938 VFS_SUSP_CLEAN(struct mount *mp) 939 { 940 if (mp->mnt_op->vfs_susp_clean != NULL) 941 mp->mnt_op->vfs_susp_clean(mp); 942 } 943 944 static inline void 945 VFS_RECLAIM_LOWERVP(struct mount *mp, struct vnode *vp) 946 { 947 if (mp->mnt_op->vfs_reclaim_lowervp != NULL) 948 mp->mnt_op->vfs_reclaim_lowervp(mp, vp); 949 } 950 951 static inline void 952 VFS_UNLINK_LOWERVP(struct mount *mp, struct vnode *vp) 953 { 954 if (mp->mnt_op->vfs_unlink_lowervp != NULL) 955 mp->mnt_op->vfs_unlink_lowervp(mp, vp); 956 } 957 958 static inline void 959 VFS_PURGE(struct mount *mp) 960 { 961 if (mp->mnt_op->vfs_purge != NULL) 962 mp->mnt_op->vfs_purge(mp); 963 } 964 965 #include <sys/vnode.h> 966 967 static inline void 968 VFS_KNOTE_LOCKED(struct vnode *vp, int hint) 969 { 970 if ((vp->v_v2flag & V2_KNOTE) != 0) { 971 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, 972 hint, KNF_LISTLOCKED | KNF_NOKQLOCK); 973 } 974 } 975 976 static inline void 977 VFS_KNOTE_UNLOCKED(struct vnode *vp, int hint) 978 { 979 if ((vp->v_v2flag & V2_KNOTE) != 0) { 980 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, 981 hint, KNF_NOKQLOCK); 982 } 983 } 984 985 #include <sys/module.h> 986 987 /* 988 * Version numbers. 989 */ 990 #define VFS_VERSION_00 0x19660120 991 #define VFS_VERSION_01 0x20121030 992 #define VFS_VERSION_02 0x20180504 993 #define VFS_VERSION VFS_VERSION_02 994 995 #define VFS_SET(vfsops, fsname, flags) \ 996 static struct vfsconf fsname ## _vfsconf = { \ 997 .vfc_version = VFS_VERSION, \ 998 .vfc_name = #fsname, \ 999 .vfc_vfsops = &vfsops, \ 1000 .vfc_typenum = -1, \ 1001 .vfc_flags = flags, \ 1002 }; \ 1003 static moduledata_t fsname ## _mod = { \ 1004 #fsname, \ 1005 vfs_modevent, \ 1006 & fsname ## _vfsconf \ 1007 }; \ 1008 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE) 1009 1010 enum vfs_notify_upper_type { 1011 VFS_NOTIFY_UPPER_RECLAIM, 1012 VFS_NOTIFY_UPPER_UNLINK, 1013 }; 1014 1015 /* 1016 * exported vnode operations 1017 */ 1018 1019 /* Define this to indicate that vfs_exjail_clone() exists for ZFS to use. */ 1020 #define VFS_SUPPORTS_EXJAIL_CLONE 1 1021 1022 int dounmount(struct mount *, uint64_t, struct thread *); 1023 1024 int kernel_mount(struct mntarg *ma, uint64_t flags); 1025 struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len); 1026 struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name); 1027 struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...); 1028 struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len); 1029 void statfs_scale_blocks(struct statfs *sf, long max_size); 1030 struct vfsconf *vfs_byname(const char *); 1031 struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *); 1032 void vfs_unref_vfsconf(struct vfsconf *vfsp); 1033 void vfs_mount_destroy(struct mount *); 1034 void vfs_event_signal(fsid_t *, u_int32_t, intptr_t); 1035 void vfs_freeopts(struct vfsoptlist *opts); 1036 void vfs_deleteopt(struct vfsoptlist *opts, const char *name); 1037 int vfs_buildopts(struct uio *auio, struct vfsoptlist **options); 1038 int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w, 1039 uint64_t val); 1040 int vfs_getopt(struct vfsoptlist *, const char *, void **, int *); 1041 int vfs_getopt_pos(struct vfsoptlist *opts, const char *name); 1042 int vfs_getopt_size(struct vfsoptlist *opts, const char *name, 1043 off_t *value); 1044 char *vfs_getopts(struct vfsoptlist *, const char *, int *error); 1045 int vfs_copyopt(struct vfsoptlist *, const char *, void *, int); 1046 int vfs_filteropt(struct vfsoptlist *, const char **legal); 1047 void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...); 1048 int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...); 1049 int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value, 1050 int len); 1051 int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value, 1052 int len); 1053 int vfs_setopts(struct vfsoptlist *opts, const char *name, 1054 const char *value); 1055 int vfs_setpublicfs /* set publicly exported fs */ 1056 (struct mount *, struct netexport *, struct export_args *); 1057 void vfs_periodic(struct mount *, int); 1058 int vfs_busy(struct mount *, int); 1059 void vfs_exjail_clone(struct mount *, struct mount *); 1060 void vfs_exjail_delete(struct prison *); 1061 int vfs_export /* process mount export info */ 1062 (struct mount *, struct export_args *, bool); 1063 void vfs_free_addrlist(struct netexport *); 1064 u_int vfs_netexport_acquire(struct netexport *); 1065 void vfs_netexport_release(struct netexport *); 1066 void vfs_allocate_syncvnode(struct mount *); 1067 void vfs_deallocate_syncvnode(struct mount *); 1068 int vfs_donmount(struct thread *td, uint64_t fsflags, 1069 struct uio *fsoptions); 1070 void vfs_getnewfsid(struct mount *); 1071 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */ 1072 struct mount *vfs_busyfs(fsid_t *); 1073 int vfs_modevent(module_t, int, void *); 1074 void vfs_mount_error(struct mount *, const char *, ...); 1075 void vfs_mountroot(void); /* mount our root filesystem */ 1076 void vfs_mountedfrom(struct mount *, const char *from); 1077 void vfs_notify_upper(struct vnode *, enum vfs_notify_upper_type); 1078 struct mount *vfs_ref_from_vp(struct vnode *); 1079 void vfs_ref(struct mount *); 1080 void vfs_rel(struct mount *); 1081 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *, 1082 struct ucred *); 1083 int vfs_suser(struct mount *, struct thread *); 1084 void vfs_unbusy(struct mount *); 1085 void vfs_unmountall(void); 1086 struct mount *vfs_register_upper_from_vp(struct vnode *, 1087 struct mount *ump, struct mount_upper_node *); 1088 void vfs_register_for_notification(struct mount *, struct mount *, 1089 struct mount_upper_node *); 1090 void vfs_unregister_for_notification(struct mount *, 1091 struct mount_upper_node *); 1092 void vfs_unregister_upper(struct mount *, struct mount_upper_node *); 1093 int vfs_remount_ro(struct mount *mp); 1094 int vfs_report_lockf(struct mount *mp, struct sbuf *sb); 1095 1096 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */ 1097 extern struct mtx_padalign mountlist_mtx; 1098 extern struct sx vfsconf_sx; 1099 #define vfsconf_lock() sx_xlock(&vfsconf_sx) 1100 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx) 1101 #define vfsconf_slock() sx_slock(&vfsconf_sx) 1102 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx) 1103 struct vnode *mntfs_allocvp(struct mount *, struct vnode *); 1104 void mntfs_freevp(struct vnode *); 1105 1106 /* 1107 * Declarations for these vfs default operations are located in 1108 * kern/vfs_default.c. They will be automatically used to replace 1109 * null entries in VFS ops tables when registering a new filesystem 1110 * type in the global table. 1111 */ 1112 vfs_root_t vfs_stdroot; 1113 vfs_quotactl_t vfs_stdquotactl; 1114 vfs_statfs_t vfs_stdstatfs; 1115 vfs_sync_t vfs_stdsync; 1116 vfs_sync_t vfs_stdnosync; 1117 vfs_vget_t vfs_stdvget; 1118 vfs_fhtovp_t vfs_stdfhtovp; 1119 vfs_checkexp_t vfs_stdcheckexp; 1120 vfs_init_t vfs_stdinit; 1121 vfs_uninit_t vfs_stduninit; 1122 vfs_extattrctl_t vfs_stdextattrctl; 1123 vfs_sysctl_t vfs_stdsysctl; 1124 1125 void syncer_suspend(void); 1126 void syncer_resume(void); 1127 1128 struct vnode *vfs_cache_root_clear(struct mount *); 1129 void vfs_cache_root_set(struct mount *, struct vnode *); 1130 1131 void vfs_op_barrier_wait(struct mount *); 1132 void vfs_op_enter(struct mount *); 1133 void vfs_op_exit_locked(struct mount *); 1134 void vfs_op_exit(struct mount *); 1135 1136 #ifdef DIAGNOSTIC 1137 void vfs_assert_mount_counters(struct mount *); 1138 void vfs_dump_mount_counters(struct mount *); 1139 #else 1140 #define vfs_assert_mount_counters(mp) do { } while (0) 1141 #define vfs_dump_mount_counters(mp) do { } while (0) 1142 #endif 1143 1144 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT }; 1145 int vfs_mount_fetch_counter(struct mount *, enum mount_counter); 1146 1147 void suspend_all_fs(void); 1148 void resume_all_fs(void); 1149 1150 /* 1151 * Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes 1152 * all CPUs not executing code enclosed by thread_in_ops_pcpu variable. 1153 * 1154 * This provides an invariant that by the time the last CPU is observed not 1155 * executing, everyone else entering will see the counter > 0 and exit. 1156 * 1157 * Note there is no barrier between vfs_ops and the rest of the code in the 1158 * section. It is not necessary as the writer has to wait for everyone to drain 1159 * before making any changes or only make changes safe while the section is 1160 * executed. 1161 */ 1162 #define vfs_mount_pcpu(mp) zpcpu_get(mp->mnt_pcpu) 1163 #define vfs_mount_pcpu_remote(mp, cpu) zpcpu_get_cpu(mp->mnt_pcpu, cpu) 1164 static void vfs_op_thread_exit_crit(struct mount *mp, struct mount_pcpu *mpcpu); 1165 1166 static inline bool 1167 vfs_op_thread_entered(struct mount *mp) 1168 { 1169 struct mount_pcpu *mpcpu = vfs_mount_pcpu(mp); 1170 1171 MPASS(curthread->td_critnest > 0); 1172 return (mpcpu->mntp_thread_in_ops == 1); 1173 } 1174 1175 static inline bool 1176 vfs_op_thread_enter_crit(struct mount *mp, struct mount_pcpu **mpcpup) 1177 { 1178 struct mount_pcpu *mpcpu; 1179 bool retval_crit = true; 1180 1181 MPASS(curthread->td_critnest > 0); 1182 mpcpu = vfs_mount_pcpu(mp); 1183 MPASS(mpcpu->mntp_thread_in_ops == 0); 1184 mpcpu->mntp_thread_in_ops = 1; 1185 atomic_interrupt_fence(); 1186 if (__predict_false(mp->mnt_vfs_ops > 0)) { 1187 vfs_op_thread_exit_crit(mp, mpcpu); 1188 retval_crit = false; 1189 } 1190 *mpcpup = mpcpu; 1191 return (retval_crit); 1192 } 1193 1194 static inline bool 1195 vfs_op_thread_enter(struct mount *mp, struct mount_pcpu **mpcpup) 1196 { 1197 bool retval; 1198 1199 critical_enter(); 1200 retval = vfs_op_thread_enter_crit(mp, mpcpup); 1201 if (__predict_false(!retval)) 1202 critical_exit(); 1203 return (retval); 1204 } 1205 1206 static inline void 1207 vfs_op_thread_exit_crit(struct mount *mp, struct mount_pcpu *mpcpu) 1208 { 1209 MPASS(mpcpu == vfs_mount_pcpu(mp)); 1210 MPASS(mpcpu->mntp_thread_in_ops == 1); 1211 1212 atomic_interrupt_fence(); 1213 mpcpu->mntp_thread_in_ops = 0; 1214 } 1215 1216 static inline void 1217 vfs_op_thread_exit(struct mount *mp, struct mount_pcpu *mpcpu) 1218 { 1219 vfs_op_thread_exit_crit(mp, mpcpu); 1220 critical_exit(); 1221 } 1222 1223 #define vfs_mp_count_add_pcpu(_mpcpu, count, val) do { \ 1224 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1225 _mpcpu->mntp_##count += val; \ 1226 } while (0) 1227 1228 #define vfs_mp_count_sub_pcpu(_mpcpu, count, val) do { \ 1229 MPASS(_mpcpu->mntp_thread_in_ops == 1); \ 1230 _mpcpu->mntp_##count -= val; \ 1231 } while (0) 1232 1233 #else /* !_KERNEL */ 1234 1235 #include <sys/cdefs.h> 1236 1237 struct stat; 1238 1239 __BEGIN_DECLS 1240 int fhlink(struct fhandle *, const char *); 1241 int fhlinkat(struct fhandle *, int, const char *); 1242 int fhopen(const struct fhandle *, int); 1243 int fhreadlink(struct fhandle *, char *, size_t); 1244 int fhstat(const struct fhandle *, struct stat *); 1245 int fhstatfs(const struct fhandle *, struct statfs *); 1246 int fstatfs(int, struct statfs *); 1247 int getfh(const char *, fhandle_t *); 1248 int getfhat(int, char *, struct fhandle *, int); 1249 int getfsstat(struct statfs *, long, int); 1250 int getmntinfo(struct statfs **, int); 1251 int lgetfh(const char *, fhandle_t *); 1252 int mount(const char *, const char *, int, void *); 1253 int nmount(struct iovec *, unsigned int, int); 1254 int statfs(const char *, struct statfs *); 1255 int unmount(const char *, int); 1256 1257 /* C library stuff */ 1258 int getvfsbyname(const char *, struct xvfsconf *); 1259 __END_DECLS 1260 1261 #endif /* _KERNEL */ 1262 1263 #endif /* !_SYS_MOUNT_H_ */ 1264