1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1989, 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 * @(#)vnode.h 8.7 (Berkeley) 2/4/94 32 * $FreeBSD$ 33 */ 34 35 #ifndef _SYS_VNODE_H_ 36 #define _SYS_VNODE_H_ 37 38 #include <sys/bufobj.h> 39 #include <sys/queue.h> 40 #include <sys/lock.h> 41 #include <sys/lockmgr.h> 42 #include <sys/mutex.h> 43 #include <sys/rangelock.h> 44 #include <sys/selinfo.h> 45 #include <sys/uio.h> 46 #include <sys/acl.h> 47 #include <sys/ktr.h> 48 49 /* 50 * The vnode is the focus of all file activity in UNIX. There is a 51 * unique vnode allocated for each active file, each current directory, 52 * each mounted-on file, text file, and the root. 53 */ 54 55 /* 56 * Vnode types. VNON means no type. 57 */ 58 enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD, 59 VMARKER }; 60 61 enum vgetstate { VGET_HOLDCNT, VGET_USECOUNT }; 62 /* 63 * Each underlying filesystem allocates its own private area and hangs 64 * it from v_data. If non-null, this area is freed in getnewvnode(). 65 */ 66 67 struct namecache; 68 69 struct vpollinfo { 70 struct mtx vpi_lock; /* lock to protect below */ 71 struct selinfo vpi_selinfo; /* identity of poller(s) */ 72 short vpi_events; /* what they are looking for */ 73 short vpi_revents; /* what has happened */ 74 }; 75 76 /* 77 * Reading or writing any of these items requires holding the appropriate lock. 78 * 79 * Lock reference: 80 * c - namecache mutex 81 * i - interlock 82 * l - mp mnt_listmtx or freelist mutex 83 * I - updated with atomics, 0->1 and 1->0 transitions with interlock held 84 * m - mount point interlock 85 * p - pollinfo lock 86 * u - Only a reference to the vnode is needed to read. 87 * v - vnode lock 88 * 89 * Vnodes may be found on many lists. The general way to deal with operating 90 * on a vnode that is on a list is: 91 * 1) Lock the list and find the vnode. 92 * 2) Lock interlock so that the vnode does not go away. 93 * 3) Unlock the list to avoid lock order reversals. 94 * 4) vget with LK_INTERLOCK and check for ENOENT, or 95 * 5) Check for DOOMED if the vnode lock is not required. 96 * 6) Perform your operation, then vput(). 97 */ 98 99 #if defined(_KERNEL) || defined(_KVM_VNODE) 100 101 struct vnode { 102 /* 103 * Fields which define the identity of the vnode. These fields are 104 * owned by the filesystem (XXX: and vgone() ?) 105 */ 106 enum vtype v_type:8; /* u vnode type */ 107 short v_irflag; /* i frequently read flags */ 108 struct vop_vector *v_op; /* u vnode operations vector */ 109 void *v_data; /* u private data for fs */ 110 111 /* 112 * Filesystem instance stuff 113 */ 114 struct mount *v_mount; /* u ptr to vfs we are in */ 115 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */ 116 117 /* 118 * Type specific fields, only one applies to any given vnode. 119 */ 120 union { 121 struct mount *v_mountedhere; /* v ptr to mountpoint (VDIR) */ 122 struct unpcb *v_unpcb; /* v unix domain net (VSOCK) */ 123 struct cdev *v_rdev; /* v device (VCHR, VBLK) */ 124 struct fifoinfo *v_fifoinfo; /* v fifo (VFIFO) */ 125 }; 126 127 /* 128 * vfs_hash: (mount + inode) -> vnode hash. The hash value 129 * itself is grouped with other int fields, to avoid padding. 130 */ 131 LIST_ENTRY(vnode) v_hashlist; 132 133 /* 134 * VFS_namecache stuff 135 */ 136 LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */ 137 TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */ 138 struct namecache *v_cache_dd; /* c Cache entry for .. vnode */ 139 140 /* 141 * Locking 142 */ 143 struct lock v_lock; /* u (if fs don't have one) */ 144 struct mtx v_interlock; /* lock for "i" things */ 145 struct lock *v_vnlock; /* u pointer to vnode lock */ 146 147 /* 148 * The machinery of being a vnode 149 */ 150 TAILQ_ENTRY(vnode) v_vnodelist; /* l vnode lists */ 151 TAILQ_ENTRY(vnode) v_lazylist; /* l vnode lazy list */ 152 struct bufobj v_bufobj; /* * Buffer cache object */ 153 154 /* 155 * Hooks for various subsystems and features. 156 */ 157 struct vpollinfo *v_pollinfo; /* i Poll events, p for *v_pi */ 158 struct label *v_label; /* MAC label for vnode */ 159 struct lockf *v_lockf; /* Byte-level advisory lock list */ 160 struct rangelock v_rl; /* Byte-range lock */ 161 162 /* 163 * clustering stuff 164 */ 165 daddr_t v_cstart; /* v start block of cluster */ 166 daddr_t v_lasta; /* v last allocation */ 167 daddr_t v_lastw; /* v last write */ 168 int v_clen; /* v length of cur. cluster */ 169 170 u_int v_holdcnt; /* I prevents recycling. */ 171 u_int v_usecount; /* I ref count of users */ 172 u_int v_iflag; /* i vnode flags (see below) */ 173 u_int v_vflag; /* v vnode flags */ 174 u_short v_mflag; /* l mnt-specific vnode flags */ 175 short v_dbatchcpu; /* i LRU requeue deferral batch */ 176 int v_writecount; /* I ref count of writers or 177 (negative) text users */ 178 u_int v_hash; 179 }; 180 181 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */ 182 183 #define bo2vnode(bo) __containerof((bo), struct vnode, v_bufobj) 184 185 /* XXX: These are temporary to avoid a source sweep at this time */ 186 #define v_object v_bufobj.bo_object 187 188 /* 189 * Userland version of struct vnode, for sysctl. 190 */ 191 struct xvnode { 192 size_t xv_size; /* sizeof(struct xvnode) */ 193 void *xv_vnode; /* address of real vnode */ 194 u_long xv_flag; /* vnode vflags */ 195 int xv_usecount; /* reference count of users */ 196 int xv_writecount; /* reference count of writers */ 197 int xv_holdcnt; /* page & buffer references */ 198 u_long xv_id; /* capability identifier */ 199 void *xv_mount; /* address of parent mount */ 200 long xv_numoutput; /* num of writes in progress */ 201 enum vtype xv_type; /* vnode type */ 202 union { 203 void *xvu_socket; /* unpcb, if VSOCK */ 204 void *xvu_fifo; /* fifo, if VFIFO */ 205 dev_t xvu_rdev; /* maj/min, if VBLK/VCHR */ 206 struct { 207 dev_t xvu_dev; /* device, if VDIR/VREG/VLNK */ 208 ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */ 209 } xv_uns; 210 } xv_un; 211 }; 212 #define xv_socket xv_un.xvu_socket 213 #define xv_fifo xv_un.xvu_fifo 214 #define xv_rdev xv_un.xvu_rdev 215 #define xv_dev xv_un.xv_uns.xvu_dev 216 #define xv_ino xv_un.xv_uns.xvu_ino 217 218 /* We don't need to lock the knlist */ 219 #define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \ 220 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note)) 221 222 #define VN_KNOTE(vp, b, a) \ 223 do { \ 224 if (!VN_KNLIST_EMPTY(vp)) \ 225 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \ 226 (a) | KNF_NOKQLOCK); \ 227 } while (0) 228 #define VN_KNOTE_LOCKED(vp, b) VN_KNOTE(vp, b, KNF_LISTLOCKED) 229 #define VN_KNOTE_UNLOCKED(vp, b) VN_KNOTE(vp, b, 0) 230 231 /* 232 * Vnode flags. 233 * VI flags are protected by interlock and live in v_iflag 234 * VV flags are protected by the vnode lock and live in v_vflag 235 * 236 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both 237 * are required for writing but the status may be checked with either. 238 */ 239 #define VIRF_DOOMED 0x0001 /* This vnode is being recycled */ 240 241 #define VI_TEXT_REF 0x0001 /* Text ref grabbed use ref */ 242 #define VI_MOUNT 0x0020 /* Mount in progress */ 243 #define VI_DOINGINACT 0x0800 /* VOP_INACTIVE is in progress */ 244 #define VI_OWEINACT 0x1000 /* Need to call inactive */ 245 #define VI_DEFINACT 0x2000 /* deferred inactive */ 246 247 #define VV_ROOT 0x0001 /* root of its filesystem */ 248 #define VV_ISTTY 0x0002 /* vnode represents a tty */ 249 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */ 250 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */ 251 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */ 252 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */ 253 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */ 254 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */ 255 #define VV_PROCDEP 0x0100 /* vnode is process dependent */ 256 #define VV_NOKNOTE 0x0200 /* don't activate knotes on this vnode */ 257 #define VV_DELETED 0x0400 /* should be removed */ 258 #define VV_MD 0x0800 /* vnode backs the md device */ 259 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */ 260 #define VV_READLINK 0x2000 /* fdescfs linux vnode */ 261 262 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */ 263 264 /* 265 * Vnode attributes. A field value of VNOVAL represents a field whose value 266 * is unavailable (getattr) or which is not to be changed (setattr). 267 */ 268 struct vattr { 269 enum vtype va_type; /* vnode type (for create) */ 270 u_short va_mode; /* files access mode and type */ 271 u_short va_padding0; 272 uid_t va_uid; /* owner user id */ 273 gid_t va_gid; /* owner group id */ 274 nlink_t va_nlink; /* number of references to file */ 275 dev_t va_fsid; /* filesystem id */ 276 ino_t va_fileid; /* file id */ 277 u_quad_t va_size; /* file size in bytes */ 278 long va_blocksize; /* blocksize preferred for i/o */ 279 struct timespec va_atime; /* time of last access */ 280 struct timespec va_mtime; /* time of last modification */ 281 struct timespec va_ctime; /* time file changed */ 282 struct timespec va_birthtime; /* time file created */ 283 u_long va_gen; /* generation number of file */ 284 u_long va_flags; /* flags defined for file */ 285 dev_t va_rdev; /* device the special file represents */ 286 u_quad_t va_bytes; /* bytes of disk space held by file */ 287 u_quad_t va_filerev; /* file modification number */ 288 u_int va_vaflags; /* operations flags, see below */ 289 long va_spare; /* remain quad aligned */ 290 }; 291 292 /* 293 * Flags for va_vaflags. 294 */ 295 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */ 296 #define VA_EXCLUSIVE 0x02 /* exclusive create request */ 297 #define VA_SYNC 0x04 /* O_SYNC truncation */ 298 299 /* 300 * Flags for ioflag. (high 16 bits used to ask for read-ahead and 301 * help with write clustering) 302 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h 303 */ 304 #define IO_UNIT 0x0001 /* do I/O as atomic unit */ 305 #define IO_APPEND 0x0002 /* append write to end */ 306 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */ 307 #define IO_NODELOCKED 0x0008 /* underlying node already locked */ 308 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */ 309 #define IO_VMIO 0x0020 /* data already in VMIO space */ 310 #define IO_INVAL 0x0040 /* invalidate after I/O */ 311 #define IO_SYNC 0x0080 /* do I/O synchronously */ 312 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */ 313 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */ 314 #define IO_EXT 0x0400 /* operate on external attributes */ 315 #define IO_NORMAL 0x0800 /* operate on regular data */ 316 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */ 317 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */ 318 #define IO_RANGELOCKED 0x4000 /* range locked */ 319 320 #define IO_SEQMAX 0x7F /* seq heuristic max value */ 321 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */ 322 323 /* 324 * Flags for accmode_t. 325 */ 326 #define VEXEC 000000000100 /* execute/search permission */ 327 #define VWRITE 000000000200 /* write permission */ 328 #define VREAD 000000000400 /* read permission */ 329 #define VADMIN 000000010000 /* being the file owner */ 330 #define VAPPEND 000000040000 /* permission to write/append */ 331 /* 332 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only 333 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL, 334 * and 0 otherwise. This never happens with ordinary unix access rights 335 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with 336 * some other V* constant. 337 */ 338 #define VEXPLICIT_DENY 000000100000 339 #define VREAD_NAMED_ATTRS 000000200000 /* not used */ 340 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */ 341 #define VDELETE_CHILD 000001000000 342 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */ 343 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */ 344 #define VDELETE 000010000000 345 #define VREAD_ACL 000020000000 /* read ACL and file mode */ 346 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */ 347 #define VWRITE_OWNER 000100000000 /* change file owner */ 348 #define VSYNCHRONIZE 000200000000 /* not used */ 349 #define VCREAT 000400000000 /* creating new file */ 350 #define VVERIFY 001000000000 /* verification required */ 351 352 /* 353 * Permissions that were traditionally granted only to the file owner. 354 */ 355 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \ 356 VWRITE_OWNER) 357 358 /* 359 * Permissions that were traditionally granted to everyone. 360 */ 361 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL) 362 363 /* 364 * Permissions that allow to change the state of the file in any way. 365 */ 366 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \ 367 VDELETE) 368 369 /* 370 * Token indicating no attribute value yet assigned. 371 */ 372 #define VNOVAL (-1) 373 374 /* 375 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon) 376 */ 377 #define VLKTIMEOUT (hz / 20 + 1) 378 379 #ifdef _KERNEL 380 381 #ifdef MALLOC_DECLARE 382 MALLOC_DECLARE(M_VNODE); 383 #endif 384 385 extern u_int ncsizefactor; 386 387 /* 388 * Convert between vnode types and inode formats (since POSIX.1 389 * defines mode word of stat structure in terms of inode formats). 390 */ 391 extern enum vtype iftovt_tab[]; 392 extern int vttoif_tab[]; 393 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12]) 394 #define VTTOIF(indx) (vttoif_tab[(int)(indx)]) 395 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode)) 396 397 /* 398 * Flags to various vnode functions. 399 */ 400 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */ 401 #define FORCECLOSE 0x0002 /* vflush: force file closure */ 402 #define WRITECLOSE 0x0004 /* vflush: only close writable files */ 403 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */ 404 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */ 405 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */ 406 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */ 407 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */ 408 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */ 409 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */ 410 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */ 411 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */ 412 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */ 413 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */ 414 #define V_MNTREF 0x0010 /* vn_start_write: mp is already ref-ed */ 415 416 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */ 417 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */ 418 419 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the 420 filesystem is being unmounted */ 421 422 #define VREF(vp) vref(vp) 423 424 #ifdef DIAGNOSTIC 425 #define VATTR_NULL(vap) vattr_null(vap) 426 #else 427 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */ 428 #endif /* DIAGNOSTIC */ 429 430 #define NULLVP ((struct vnode *)NULL) 431 432 /* 433 * Global vnode data. 434 */ 435 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */ 436 extern struct mount *rootdevmp; /* "/dev" mount */ 437 extern u_long desiredvnodes; /* number of vnodes desired */ 438 extern struct uma_zone *namei_zone; 439 extern struct vattr va_null; /* predefined null vattr structure */ 440 441 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock) 442 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags)) 443 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock) 444 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock) 445 #define VI_MTX(vp) (&(vp)->v_interlock) 446 447 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock) 448 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock) 449 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock) 450 451 #endif /* _KERNEL */ 452 453 /* 454 * Mods for extensibility. 455 */ 456 457 /* 458 * Flags for vdesc_flags: 459 */ 460 #define VDESC_MAX_VPS 16 461 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */ 462 #define VDESC_VP0_WILLRELE 0x0001 463 #define VDESC_VP1_WILLRELE 0x0002 464 #define VDESC_VP2_WILLRELE 0x0004 465 #define VDESC_VP3_WILLRELE 0x0008 466 467 /* 468 * A generic structure. 469 * This can be used by bypass routines to identify generic arguments. 470 */ 471 struct vop_generic_args { 472 struct vnodeop_desc *a_desc; 473 /* other random data follows, presumably */ 474 }; 475 476 typedef int vop_bypass_t(struct vop_generic_args *); 477 478 /* 479 * VDESC_NO_OFFSET is used to identify the end of the offset list 480 * and in places where no such field exists. 481 */ 482 #define VDESC_NO_OFFSET -1 483 484 /* 485 * This structure describes the vnode operation taking place. 486 */ 487 struct vnodeop_desc { 488 char *vdesc_name; /* a readable name for debugging */ 489 int vdesc_flags; /* VDESC_* flags */ 490 int vdesc_vop_offset; 491 vop_bypass_t *vdesc_call; /* Function to call */ 492 493 /* 494 * These ops are used by bypass routines to map and locate arguments. 495 * Creds and procs are not needed in bypass routines, but sometimes 496 * they are useful to (for example) transport layers. 497 * Nameidata is useful because it has a cred in it. 498 */ 499 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */ 500 int vdesc_vpp_offset; /* return vpp location */ 501 int vdesc_cred_offset; /* cred location, if any */ 502 int vdesc_thread_offset; /* thread location, if any */ 503 int vdesc_componentname_offset; /* if any */ 504 }; 505 506 #ifdef _KERNEL 507 /* 508 * A list of all the operation descs. 509 */ 510 extern struct vnodeop_desc *vnodeop_descs[]; 511 512 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field) 513 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \ 514 ((s_type)(((char*)(struct_p)) + (s_offset))) 515 516 517 #ifdef DEBUG_VFS_LOCKS 518 /* 519 * Support code to aid in debugging VFS locking problems. Not totally 520 * reliable since if the thread sleeps between changing the lock 521 * state and checking it with the assert, some other thread could 522 * change the state. They are good enough for debugging a single 523 * filesystem using a single-threaded test. Note that the unreliability is 524 * limited to false negatives; efforts were made to ensure that false 525 * positives cannot occur. 526 */ 527 void assert_vi_locked(struct vnode *vp, const char *str); 528 void assert_vi_unlocked(struct vnode *vp, const char *str); 529 void assert_vop_elocked(struct vnode *vp, const char *str); 530 void assert_vop_locked(struct vnode *vp, const char *str); 531 void assert_vop_unlocked(struct vnode *vp, const char *str); 532 533 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str)) 534 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str)) 535 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str)) 536 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str)) 537 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str)) 538 539 #else /* !DEBUG_VFS_LOCKS */ 540 541 #define ASSERT_VI_LOCKED(vp, str) ((void)0) 542 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0) 543 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0) 544 #define ASSERT_VOP_LOCKED(vp, str) ((void)0) 545 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0) 546 #endif /* DEBUG_VFS_LOCKS */ 547 548 549 /* 550 * This call works for vnodes in the kernel. 551 */ 552 #define VCALL(c) ((c)->a_desc->vdesc_call(c)) 553 554 #define DOINGASYNC(vp) \ 555 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \ 556 ((curthread->td_pflags & TDP_SYNCIO) == 0)) 557 558 /* 559 * VMIO support inline 560 */ 561 562 extern int vmiodirenable; 563 564 static __inline int 565 vn_canvmio(struct vnode *vp) 566 { 567 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR))) 568 return(TRUE); 569 return(FALSE); 570 } 571 572 /* 573 * Finally, include the default set of vnode operations. 574 */ 575 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int); 576 #include "vnode_if.h" 577 578 /* vn_open_flags */ 579 #define VN_OPEN_NOAUDIT 0x00000001 580 #define VN_OPEN_NOCAPCHECK 0x00000002 581 #define VN_OPEN_NAMECACHE 0x00000004 582 #define VN_OPEN_INVFS 0x00000008 583 584 /* 585 * Public vnode manipulation functions. 586 */ 587 struct componentname; 588 struct file; 589 struct mount; 590 struct nameidata; 591 struct ostat; 592 struct freebsd11_stat; 593 struct thread; 594 struct proc; 595 struct stat; 596 struct nstat; 597 struct ucred; 598 struct uio; 599 struct vattr; 600 struct vfsops; 601 struct vnode; 602 603 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **); 604 605 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn, 606 daddr_t endn); 607 /* cache_* may belong in namei.h. */ 608 void cache_changesize(u_long newhashsize); 609 #define cache_enter(dvp, vp, cnp) \ 610 cache_enter_time(dvp, vp, cnp, NULL, NULL) 611 void cache_enter_time(struct vnode *dvp, struct vnode *vp, 612 struct componentname *cnp, struct timespec *tsp, 613 struct timespec *dtsp); 614 int cache_lookup(struct vnode *dvp, struct vnode **vpp, 615 struct componentname *cnp, struct timespec *tsp, int *ticksp); 616 void cache_purge(struct vnode *vp); 617 void cache_purge_negative(struct vnode *vp); 618 void cache_purgevfs(struct mount *mp, bool force); 619 int change_dir(struct vnode *vp, struct thread *td); 620 void cvtstat(struct stat *st, struct ostat *ost); 621 void freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb); 622 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost); 623 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops, 624 struct vnode **vpp); 625 void getnewvnode_reserve(void); 626 void getnewvnode_drop_reserve(void); 627 int insmntque1(struct vnode *vp, struct mount *mp, 628 void (*dtr)(struct vnode *, void *), void *dtr_arg); 629 int insmntque(struct vnode *vp, struct mount *mp); 630 u_quad_t init_va_filerev(void); 631 int speedup_syncer(void); 632 int vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf, 633 size_t *buflen); 634 int vn_getcwd(struct thread *td, char *buf, char **retbuf, size_t *buflen); 635 int vn_fullpath(struct thread *td, struct vnode *vn, 636 char **retbuf, char **freebuf); 637 int vn_fullpath_global(struct thread *td, struct vnode *vn, 638 char **retbuf, char **freebuf); 639 struct vnode * 640 vn_dir_dd_ino(struct vnode *vp); 641 int vn_commname(struct vnode *vn, char *buf, u_int buflen); 642 int vn_path_to_global_path(struct thread *td, struct vnode *vp, 643 char *path, u_int pathlen); 644 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid, 645 gid_t file_gid, accmode_t accmode, struct ucred *cred, 646 int *privused); 647 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid, 648 struct acl *aclp, accmode_t accmode, struct ucred *cred, 649 int *privused); 650 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid, 651 gid_t file_gid, struct acl *acl, accmode_t accmode, 652 struct ucred *cred, int *privused); 653 void vattr_null(struct vattr *vap); 654 int vcount(struct vnode *vp); 655 void vlazy(struct vnode *); 656 void vdrop(struct vnode *); 657 void vdropl(struct vnode *); 658 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td); 659 int vget(struct vnode *vp, int flags, struct thread *td); 660 enum vgetstate vget_prep(struct vnode *vp); 661 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs); 662 void vgone(struct vnode *vp); 663 void vhold(struct vnode *); 664 void vholdl(struct vnode *); 665 void vholdnz(struct vnode *); 666 void vinactive(struct vnode *vp); 667 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo); 668 int vtruncbuf(struct vnode *vp, off_t length, int blksize); 669 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn, 670 int blksize); 671 void vunref(struct vnode *); 672 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3); 673 int vrecycle(struct vnode *vp); 674 int vrecyclel(struct vnode *vp); 675 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off, 676 struct ucred *cred); 677 int vn_close(struct vnode *vp, 678 int flags, struct ucred *file_cred, struct thread *td); 679 int vn_copy_file_range(struct vnode *invp, off_t *inoffp, 680 struct vnode *outvp, off_t *outoffp, size_t *lenp, 681 unsigned int flags, struct ucred *incred, struct ucred *outcred, 682 struct thread *fsize_td); 683 void vn_finished_write(struct mount *mp); 684 void vn_finished_secondary_write(struct mount *mp); 685 int vn_fsync_buf(struct vnode *vp, int waitfor); 686 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp, 687 struct vnode *outvp, off_t *outoffp, size_t *lenp, 688 unsigned int flags, struct ucred *incred, struct ucred *outcred, 689 struct thread *fsize_td); 690 int vn_need_pageq_flush(struct vnode *vp); 691 int vn_isdisk(struct vnode *vp, int *errp); 692 int _vn_lock(struct vnode *vp, int flags, const char *file, int line); 693 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__) 694 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp); 695 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode, 696 u_int vn_open_flags, struct ucred *cred, struct file *fp); 697 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred, 698 struct thread *td, struct file *fp); 699 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end); 700 int vn_pollrecord(struct vnode *vp, struct thread *p, int events); 701 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, 702 int len, off_t offset, enum uio_seg segflg, int ioflg, 703 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid, 704 struct thread *td); 705 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base, 706 size_t len, off_t offset, enum uio_seg segflg, int ioflg, 707 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid, 708 struct thread *td); 709 int vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio, 710 struct thread *td); 711 int vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred, 712 struct ucred *file_cred, struct thread *td); 713 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags); 714 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp, 715 int flags); 716 int vn_truncate_locked(struct vnode *vp, off_t length, bool sync, 717 struct ucred *cred); 718 int vn_writechk(struct vnode *vp); 719 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 720 const char *attrname, int *buflen, char *buf, struct thread *td); 721 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 722 const char *attrname, int buflen, char *buf, struct thread *td); 723 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 724 const char *attrname, struct thread *td); 725 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags, 726 struct vnode **rvp); 727 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc, 728 void *alloc_arg, int lkflags, struct vnode **rvp); 729 int vn_utimes_perm(struct vnode *vp, struct vattr *vap, 730 struct ucred *cred, struct thread *td); 731 732 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio); 733 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize, 734 struct uio *uio); 735 736 #define vn_rangelock_unlock(vp, cookie) \ 737 rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp)) 738 #define vn_rangelock_unlock_range(vp, cookie, start, end) \ 739 rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), \ 740 VI_MTX(vp)) 741 #define vn_rangelock_rlock(vp, start, end) \ 742 rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 743 #define vn_rangelock_tryrlock(vp, start, end) \ 744 rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 745 #define vn_rangelock_wlock(vp, start, end) \ 746 rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 747 #define vn_rangelock_trywlock(vp, start, end) \ 748 rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 749 750 int vfs_cache_lookup(struct vop_lookup_args *ap); 751 int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp); 752 void vfs_timestamp(struct timespec *); 753 void vfs_write_resume(struct mount *mp, int flags); 754 int vfs_write_suspend(struct mount *mp, int flags); 755 int vfs_write_suspend_umnt(struct mount *mp); 756 void vnlru_free(int, struct vfsops *); 757 int vop_stdbmap(struct vop_bmap_args *); 758 int vop_stdfdatasync_buf(struct vop_fdatasync_args *); 759 int vop_stdfsync(struct vop_fsync_args *); 760 int vop_stdgetwritemount(struct vop_getwritemount_args *); 761 int vop_stdgetpages(struct vop_getpages_args *); 762 int vop_stdinactive(struct vop_inactive_args *); 763 int vop_stdioctl(struct vop_ioctl_args *); 764 int vop_stdneed_inactive(struct vop_need_inactive_args *); 765 int vop_stdkqfilter(struct vop_kqfilter_args *); 766 int vop_stdlock(struct vop_lock1_args *); 767 int vop_stdunlock(struct vop_unlock_args *); 768 int vop_stdislocked(struct vop_islocked_args *); 769 int vop_lock(struct vop_lock1_args *); 770 int vop_unlock(struct vop_unlock_args *); 771 int vop_islocked(struct vop_islocked_args *); 772 int vop_stdputpages(struct vop_putpages_args *); 773 int vop_nopoll(struct vop_poll_args *); 774 int vop_stdaccess(struct vop_access_args *ap); 775 int vop_stdaccessx(struct vop_accessx_args *ap); 776 int vop_stdadvise(struct vop_advise_args *ap); 777 int vop_stdadvlock(struct vop_advlock_args *ap); 778 int vop_stdadvlockasync(struct vop_advlockasync_args *ap); 779 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap); 780 int vop_stdallocate(struct vop_allocate_args *ap); 781 int vop_stdset_text(struct vop_set_text_args *ap); 782 int vop_stdpathconf(struct vop_pathconf_args *); 783 int vop_stdpoll(struct vop_poll_args *); 784 int vop_stdvptocnp(struct vop_vptocnp_args *ap); 785 int vop_stdvptofh(struct vop_vptofh_args *ap); 786 int vop_stdunp_bind(struct vop_unp_bind_args *ap); 787 int vop_stdunp_connect(struct vop_unp_connect_args *ap); 788 int vop_stdunp_detach(struct vop_unp_detach_args *ap); 789 int vop_eopnotsupp(struct vop_generic_args *ap); 790 int vop_ebadf(struct vop_generic_args *ap); 791 int vop_einval(struct vop_generic_args *ap); 792 int vop_enoent(struct vop_generic_args *ap); 793 int vop_enotty(struct vop_generic_args *ap); 794 int vop_null(struct vop_generic_args *ap); 795 int vop_panic(struct vop_generic_args *ap); 796 int dead_poll(struct vop_poll_args *ap); 797 int dead_read(struct vop_read_args *ap); 798 int dead_write(struct vop_write_args *ap); 799 800 /* These are called from within the actual VOPS. */ 801 void vop_close_post(void *a, int rc); 802 void vop_create_post(void *a, int rc); 803 void vop_deleteextattr_post(void *a, int rc); 804 void vop_link_post(void *a, int rc); 805 void vop_lookup_post(void *a, int rc); 806 void vop_lookup_pre(void *a); 807 void vop_mkdir_post(void *a, int rc); 808 void vop_mknod_post(void *a, int rc); 809 void vop_open_post(void *a, int rc); 810 void vop_read_post(void *a, int rc); 811 void vop_readdir_post(void *a, int rc); 812 void vop_reclaim_post(void *a, int rc); 813 void vop_remove_post(void *a, int rc); 814 void vop_rename_post(void *a, int rc); 815 void vop_rename_pre(void *a); 816 void vop_rmdir_post(void *a, int rc); 817 void vop_setattr_post(void *a, int rc); 818 void vop_setextattr_post(void *a, int rc); 819 void vop_symlink_post(void *a, int rc); 820 int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a); 821 822 #ifdef DEBUG_VFS_LOCKS 823 void vop_strategy_pre(void *a); 824 void vop_lock_pre(void *a); 825 void vop_lock_post(void *a, int rc); 826 void vop_unlock_pre(void *a); 827 void vop_need_inactive_pre(void *a); 828 void vop_need_inactive_post(void *a, int rc); 829 #else 830 #define vop_strategy_pre(x) do { } while (0) 831 #define vop_lock_pre(x) do { } while (0) 832 #define vop_lock_post(x, y) do { } while (0) 833 #define vop_unlock_pre(x) do { } while (0) 834 #define vop_need_inactive_pre(x) do { } while (0) 835 #define vop_need_inactive_post(x, y) do { } while (0) 836 #endif 837 838 void vop_rename_fail(struct vop_rename_args *ap); 839 840 #define VOP_WRITE_PRE(ap) \ 841 struct vattr va; \ 842 int error; \ 843 off_t osize, ooffset, noffset; \ 844 \ 845 osize = ooffset = noffset = 0; \ 846 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \ 847 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \ 848 if (error) \ 849 return (error); \ 850 ooffset = (ap)->a_uio->uio_offset; \ 851 osize = (off_t)va.va_size; \ 852 } 853 854 #define VOP_WRITE_POST(ap, ret) \ 855 noffset = (ap)->a_uio->uio_offset; \ 856 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \ 857 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \ 858 | (noffset > osize ? NOTE_EXTEND : 0)); \ 859 } 860 861 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__) 862 863 #ifdef INVARIANTS 864 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \ 865 do { \ 866 int error_; \ 867 \ 868 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \ 869 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \ 870 error_)); \ 871 } while (0) 872 #define VOP_SET_TEXT_CHECKED(vp) \ 873 do { \ 874 int error_; \ 875 \ 876 error_ = VOP_SET_TEXT((vp)); \ 877 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \ 878 error_)); \ 879 } while (0) 880 #define VOP_UNSET_TEXT_CHECKED(vp) \ 881 do { \ 882 int error_; \ 883 \ 884 error_ = VOP_UNSET_TEXT((vp)); \ 885 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \ 886 error_)); \ 887 } while (0) 888 #else 889 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt)) 890 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp)) 891 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp)) 892 #endif 893 894 #define VN_IS_DOOMED(vp) __predict_false((vp)->v_irflag & VIRF_DOOMED) 895 896 void vput(struct vnode *vp); 897 void vrele(struct vnode *vp); 898 void vref(struct vnode *vp); 899 void vrefl(struct vnode *vp); 900 void vrefact(struct vnode *vp); 901 void vrefactn(struct vnode *vp, u_int n); 902 int vrefcnt(struct vnode *vp); 903 void v_addpollinfo(struct vnode *vp); 904 905 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td); 906 void vnode_destroy_vobject(struct vnode *vp); 907 908 extern struct vop_vector fifo_specops; 909 extern struct vop_vector dead_vnodeops; 910 extern struct vop_vector default_vnodeops; 911 912 #define VOP_PANIC ((void*)(uintptr_t)vop_panic) 913 #define VOP_NULL ((void*)(uintptr_t)vop_null) 914 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf) 915 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty) 916 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval) 917 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent) 918 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp) 919 920 /* fifo_vnops.c */ 921 int fifo_printinfo(struct vnode *); 922 923 /* vfs_hash.c */ 924 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg); 925 926 void vfs_hash_changesize(u_long newhashsize); 927 int vfs_hash_get(const struct mount *mp, u_int hash, int flags, 928 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 929 u_int vfs_hash_index(struct vnode *vp); 930 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, 931 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 932 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td, 933 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 934 void vfs_hash_rehash(struct vnode *vp, u_int hash); 935 void vfs_hash_remove(struct vnode *vp); 936 937 int vfs_kqfilter(struct vop_kqfilter_args *); 938 struct dirent; 939 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off); 940 int vfs_emptydir(struct vnode *vp); 941 942 int vfs_unixify_accmode(accmode_t *accmode); 943 944 void vfs_unp_reclaim(struct vnode *vp); 945 946 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode); 947 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid, 948 gid_t gid); 949 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, 950 struct thread *td); 951 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred, 952 struct thread *td); 953 954 void vn_fsid(struct vnode *vp, struct vattr *va); 955 956 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp); 957 958 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \ 959 struct vnode *_vp = (vp); \ 960 int _flags = (flags); \ 961 int _error; \ 962 \ 963 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \ 964 panic("%s: unsupported flags %x\n", __func__, flags); \ 965 _error = VOP_UNLOCK(_vp); \ 966 if (_flags & LK_INTERLOCK) \ 967 VI_UNLOCK(_vp); \ 968 _error; \ 969 }) 970 971 #include <sys/kernel.h> 972 973 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \ 974 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \ 975 vfs_vector_op_register, &vnodeops) 976 977 #endif /* _KERNEL */ 978 979 #endif /* !_SYS_VNODE_H_ */ 980