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 #include <sys/_seqc.h> 49 50 /* 51 * The vnode is the focus of all file activity in UNIX. There is a 52 * unique vnode allocated for each active file, each current directory, 53 * each mounted-on file, text file, and the root. 54 */ 55 56 /* 57 * Vnode types. VNON means no type. 58 */ 59 enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD, 60 VMARKER }; 61 62 enum vgetstate { VGET_NONE, VGET_HOLDCNT, VGET_USECOUNT }; 63 /* 64 * Each underlying filesystem allocates its own private area and hangs 65 * it from v_data. If non-null, this area is freed in getnewvnode(). 66 */ 67 68 struct namecache; 69 struct cache_fpl; 70 71 struct vpollinfo { 72 struct mtx vpi_lock; /* lock to protect below */ 73 struct selinfo vpi_selinfo; /* identity of poller(s) */ 74 short vpi_events; /* what they are looking for */ 75 short vpi_revents; /* what has happened */ 76 }; 77 78 /* 79 * Reading or writing any of these items requires holding the appropriate lock. 80 * 81 * Lock reference: 82 * c - namecache mutex 83 * i - interlock 84 * l - mp mnt_listmtx or freelist mutex 85 * I - updated with atomics, 0->1 and 1->0 transitions with interlock held 86 * m - mount point interlock 87 * p - pollinfo lock 88 * u - Only a reference to the vnode is needed to read. 89 * v - vnode lock 90 * 91 * Vnodes may be found on many lists. The general way to deal with operating 92 * on a vnode that is on a list is: 93 * 1) Lock the list and find the vnode. 94 * 2) Lock interlock so that the vnode does not go away. 95 * 3) Unlock the list to avoid lock order reversals. 96 * 4) vget with LK_INTERLOCK and check for ENOENT, or 97 * 5) Check for DOOMED if the vnode lock is not required. 98 * 6) Perform your operation, then vput(). 99 */ 100 101 #if defined(_KERNEL) || defined(_KVM_VNODE) 102 103 struct vnode { 104 /* 105 * Fields which define the identity of the vnode. These fields are 106 * owned by the filesystem (XXX: and vgone() ?) 107 */ 108 enum vtype v_type:8; /* u vnode type */ 109 short v_irflag; /* i frequently read flags */ 110 seqc_t v_seqc; /* i modification count */ 111 uint32_t v_nchash; /* u namecache hash */ 112 u_int v_hash; 113 struct vop_vector *v_op; /* u vnode operations vector */ 114 void *v_data; /* u private data for fs */ 115 116 /* 117 * Filesystem instance stuff 118 */ 119 struct mount *v_mount; /* u ptr to vfs we are in */ 120 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */ 121 122 /* 123 * Type specific fields, only one applies to any given vnode. 124 */ 125 union { 126 struct mount *v_mountedhere; /* v ptr to mountpoint (VDIR) */ 127 struct unpcb *v_unpcb; /* v unix domain net (VSOCK) */ 128 struct cdev *v_rdev; /* v device (VCHR, VBLK) */ 129 struct fifoinfo *v_fifoinfo; /* v fifo (VFIFO) */ 130 }; 131 132 /* 133 * vfs_hash: (mount + inode) -> vnode hash. The hash value 134 * itself is grouped with other int fields, to avoid padding. 135 */ 136 LIST_ENTRY(vnode) v_hashlist; 137 138 /* 139 * VFS_namecache stuff 140 */ 141 LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */ 142 TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */ 143 struct namecache *v_cache_dd; /* c Cache entry for .. vnode */ 144 145 /* 146 * Locking 147 */ 148 struct lock v_lock; /* u (if fs don't have one) */ 149 struct mtx v_interlock; /* lock for "i" things */ 150 struct lock *v_vnlock; /* u pointer to vnode lock */ 151 152 /* 153 * The machinery of being a vnode 154 */ 155 TAILQ_ENTRY(vnode) v_vnodelist; /* l vnode lists */ 156 TAILQ_ENTRY(vnode) v_lazylist; /* l vnode lazy list */ 157 struct bufobj v_bufobj; /* * Buffer cache object */ 158 159 /* 160 * Hooks for various subsystems and features. 161 */ 162 struct vpollinfo *v_pollinfo; /* i Poll events, p for *v_pi */ 163 struct label *v_label; /* MAC label for vnode */ 164 struct lockf *v_lockf; /* Byte-level advisory lock list */ 165 struct rangelock v_rl; /* Byte-range lock */ 166 167 u_int v_holdcnt; /* I prevents recycling. */ 168 u_int v_usecount; /* I ref count of users */ 169 u_short v_iflag; /* i vnode flags (see below) */ 170 u_short v_vflag; /* v vnode flags */ 171 u_short v_mflag; /* l mnt-specific vnode flags */ 172 short v_dbatchcpu; /* i LRU requeue deferral batch */ 173 int v_writecount; /* I ref count of writers or 174 (negative) text users */ 175 int v_seqc_users; /* i modifications pending */ 176 }; 177 178 #ifndef DEBUG_LOCKS 179 #ifdef _LP64 180 /* 181 * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields 182 * to the structure and there is nothing which can be done to prevent growth 183 * then so be it. But don't grow it without a good reason. 184 */ 185 _Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes"); 186 #endif 187 #endif 188 189 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */ 190 191 #define bo2vnode(bo) __containerof((bo), struct vnode, v_bufobj) 192 193 /* XXX: These are temporary to avoid a source sweep at this time */ 194 #define v_object v_bufobj.bo_object 195 196 /* 197 * Userland version of struct vnode, for sysctl. 198 */ 199 struct xvnode { 200 size_t xv_size; /* sizeof(struct xvnode) */ 201 void *xv_vnode; /* address of real vnode */ 202 u_long xv_flag; /* vnode vflags */ 203 int xv_usecount; /* reference count of users */ 204 int xv_writecount; /* reference count of writers */ 205 int xv_holdcnt; /* page & buffer references */ 206 u_long xv_id; /* capability identifier */ 207 void *xv_mount; /* address of parent mount */ 208 long xv_numoutput; /* num of writes in progress */ 209 enum vtype xv_type; /* vnode type */ 210 union { 211 void *xvu_socket; /* unpcb, if VSOCK */ 212 void *xvu_fifo; /* fifo, if VFIFO */ 213 dev_t xvu_rdev; /* maj/min, if VBLK/VCHR */ 214 struct { 215 dev_t xvu_dev; /* device, if VDIR/VREG/VLNK */ 216 ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */ 217 } xv_uns; 218 } xv_un; 219 }; 220 #define xv_socket xv_un.xvu_socket 221 #define xv_fifo xv_un.xvu_fifo 222 #define xv_rdev xv_un.xvu_rdev 223 #define xv_dev xv_un.xv_uns.xvu_dev 224 #define xv_ino xv_un.xv_uns.xvu_ino 225 226 /* We don't need to lock the knlist */ 227 #define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \ 228 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note)) 229 230 #define VN_KNOTE(vp, b, a) \ 231 do { \ 232 if (!VN_KNLIST_EMPTY(vp)) \ 233 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \ 234 (a) | KNF_NOKQLOCK); \ 235 } while (0) 236 #define VN_KNOTE_LOCKED(vp, b) VN_KNOTE(vp, b, KNF_LISTLOCKED) 237 #define VN_KNOTE_UNLOCKED(vp, b) VN_KNOTE(vp, b, 0) 238 239 /* 240 * Vnode flags. 241 * VI flags are protected by interlock and live in v_iflag 242 * VV flags are protected by the vnode lock and live in v_vflag 243 * 244 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both 245 * are required for writing but the status may be checked with either. 246 */ 247 #define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */ 248 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR) 249 250 #define VIRF_DOOMED 0x0001 /* This vnode is being recycled */ 251 #define VIRF_PGREAD 0x0002 /* Direct reads from the page cache are permitted, 252 never cleared once set */ 253 #define VIRF_MOUNTPOINT 0x0004 /* This vnode is mounted on */ 254 #define VIRF_TEXT_REF 0x0008 /* Executable mappings ref the vnode */ 255 256 #define VI_UNUSED0 0x0001 /* unused */ 257 #define VI_MOUNT 0x0002 /* Mount in progress */ 258 #define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */ 259 #define VI_OWEINACT 0x0008 /* Need to call inactive */ 260 #define VI_DEFINACT 0x0010 /* deferred inactive */ 261 #define VI_FOPENING 0x0020 /* In open, with opening process having the 262 first right to advlock file */ 263 264 #define VV_ROOT 0x0001 /* root of its filesystem */ 265 #define VV_ISTTY 0x0002 /* vnode represents a tty */ 266 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */ 267 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */ 268 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */ 269 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */ 270 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */ 271 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */ 272 #define VV_PROCDEP 0x0100 /* vnode is process dependent */ 273 /* UNUSED 0x0200 */ 274 #define VV_DELETED 0x0400 /* should be removed */ 275 #define VV_MD 0x0800 /* vnode backs the md device */ 276 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */ 277 #define VV_READLINK 0x2000 /* fdescfs linux vnode */ 278 #define VV_UNREF 0x4000 /* vunref, do not drop lock in inactive() */ 279 280 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */ 281 282 /* 283 * Vnode attributes. A field value of VNOVAL represents a field whose value 284 * is unavailable (getattr) or which is not to be changed (setattr). 285 */ 286 struct vattr { 287 enum vtype va_type; /* vnode type (for create) */ 288 u_short va_mode; /* files access mode and type */ 289 u_short va_padding0; 290 uid_t va_uid; /* owner user id */ 291 gid_t va_gid; /* owner group id */ 292 nlink_t va_nlink; /* number of references to file */ 293 dev_t va_fsid; /* filesystem id */ 294 ino_t va_fileid; /* file id */ 295 u_quad_t va_size; /* file size in bytes */ 296 long va_blocksize; /* blocksize preferred for i/o */ 297 struct timespec va_atime; /* time of last access */ 298 struct timespec va_mtime; /* time of last modification */ 299 struct timespec va_ctime; /* time file changed */ 300 struct timespec va_birthtime; /* time file created */ 301 u_long va_gen; /* generation number of file */ 302 u_long va_flags; /* flags defined for file */ 303 dev_t va_rdev; /* device the special file represents */ 304 u_quad_t va_bytes; /* bytes of disk space held by file */ 305 u_quad_t va_filerev; /* file modification number */ 306 u_int va_vaflags; /* operations flags, see below */ 307 long va_spare; /* remain quad aligned */ 308 }; 309 310 /* 311 * Flags for va_vaflags. 312 */ 313 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */ 314 #define VA_EXCLUSIVE 0x02 /* exclusive create request */ 315 #define VA_SYNC 0x04 /* O_SYNC truncation */ 316 317 /* 318 * Flags for ioflag. (high 16 bits used to ask for read-ahead and 319 * help with write clustering) 320 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h 321 */ 322 #define IO_UNIT 0x0001 /* do I/O as atomic unit */ 323 #define IO_APPEND 0x0002 /* append write to end */ 324 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */ 325 #define IO_NODELOCKED 0x0008 /* underlying node already locked */ 326 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */ 327 #define IO_VMIO 0x0020 /* data already in VMIO space */ 328 #define IO_INVAL 0x0040 /* invalidate after I/O */ 329 #define IO_SYNC 0x0080 /* do I/O synchronously */ 330 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */ 331 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */ 332 #define IO_EXT 0x0400 /* operate on external attributes */ 333 #define IO_NORMAL 0x0800 /* operate on regular data */ 334 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */ 335 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */ 336 #define IO_RANGELOCKED 0x4000 /* range locked */ 337 #define IO_DATASYNC 0x8000 /* do only data I/O synchronously */ 338 339 #define IO_SEQMAX 0x7F /* seq heuristic max value */ 340 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */ 341 342 /* 343 * Flags for accmode_t. 344 */ 345 #define VEXEC 000000000100 /* execute/search permission */ 346 #define VWRITE 000000000200 /* write permission */ 347 #define VREAD 000000000400 /* read permission */ 348 #define VADMIN 000000010000 /* being the file owner */ 349 #define VAPPEND 000000040000 /* permission to write/append */ 350 /* 351 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only 352 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL, 353 * and 0 otherwise. This never happens with ordinary unix access rights 354 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with 355 * some other V* constant. 356 */ 357 #define VEXPLICIT_DENY 000000100000 358 #define VREAD_NAMED_ATTRS 000000200000 /* not used */ 359 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */ 360 #define VDELETE_CHILD 000001000000 361 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */ 362 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */ 363 #define VDELETE 000010000000 364 #define VREAD_ACL 000020000000 /* read ACL and file mode */ 365 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */ 366 #define VWRITE_OWNER 000100000000 /* change file owner */ 367 #define VSYNCHRONIZE 000200000000 /* not used */ 368 #define VCREAT 000400000000 /* creating new file */ 369 #define VVERIFY 001000000000 /* verification required */ 370 371 /* 372 * Permissions that were traditionally granted only to the file owner. 373 */ 374 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \ 375 VWRITE_OWNER) 376 377 /* 378 * Permissions that were traditionally granted to everyone. 379 */ 380 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL) 381 382 /* 383 * Permissions that allow to change the state of the file in any way. 384 */ 385 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \ 386 VDELETE) 387 388 /* 389 * Token indicating no attribute value yet assigned. 390 */ 391 #define VNOVAL (-1) 392 393 /* 394 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon) 395 */ 396 #define VLKTIMEOUT (hz / 20 + 1) 397 398 #ifdef _KERNEL 399 400 #ifdef MALLOC_DECLARE 401 MALLOC_DECLARE(M_VNODE); 402 #endif 403 404 extern u_int ncsizefactor; 405 extern const u_int io_hold_cnt; 406 407 /* 408 * Convert between vnode types and inode formats (since POSIX.1 409 * defines mode word of stat structure in terms of inode formats). 410 */ 411 extern enum vtype iftovt_tab[]; 412 extern int vttoif_tab[]; 413 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12]) 414 #define VTTOIF(indx) (vttoif_tab[(int)(indx)]) 415 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode)) 416 417 /* 418 * Flags to various vnode functions. 419 */ 420 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */ 421 #define FORCECLOSE 0x0002 /* vflush: force file closure */ 422 #define WRITECLOSE 0x0004 /* vflush: only close writable files */ 423 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */ 424 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */ 425 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */ 426 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */ 427 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */ 428 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */ 429 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */ 430 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */ 431 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */ 432 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */ 433 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */ 434 #define V_MNTREF 0x0010 /* vn_start_write: mp is already ref-ed */ 435 436 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */ 437 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */ 438 439 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the 440 filesystem is being unmounted */ 441 442 #define VREF(vp) vref(vp) 443 444 #ifdef DIAGNOSTIC 445 #define VATTR_NULL(vap) vattr_null(vap) 446 #else 447 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */ 448 #endif /* DIAGNOSTIC */ 449 450 #define NULLVP ((struct vnode *)NULL) 451 452 /* 453 * Global vnode data. 454 */ 455 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */ 456 extern struct mount *rootdevmp; /* "/dev" mount */ 457 extern u_long desiredvnodes; /* number of vnodes desired */ 458 extern struct uma_zone *namei_zone; 459 extern struct vattr va_null; /* predefined null vattr structure */ 460 461 extern u_int vn_lock_pair_pause_max; 462 463 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock) 464 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags)) 465 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock) 466 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock) 467 #define VI_MTX(vp) (&(vp)->v_interlock) 468 469 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock) 470 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock) 471 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock) 472 473 #endif /* _KERNEL */ 474 475 /* 476 * Mods for extensibility. 477 */ 478 479 /* 480 * Flags for vdesc_flags: 481 */ 482 #define VDESC_MAX_VPS 16 483 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */ 484 #define VDESC_VP0_WILLRELE 0x0001 485 #define VDESC_VP1_WILLRELE 0x0002 486 #define VDESC_VP2_WILLRELE 0x0004 487 #define VDESC_VP3_WILLRELE 0x0008 488 489 /* 490 * A generic structure. 491 * This can be used by bypass routines to identify generic arguments. 492 */ 493 struct vop_generic_args { 494 struct vnodeop_desc *a_desc; 495 /* other random data follows, presumably */ 496 }; 497 498 typedef int vop_bypass_t(struct vop_generic_args *); 499 500 /* 501 * VDESC_NO_OFFSET is used to identify the end of the offset list 502 * and in places where no such field exists. 503 */ 504 #define VDESC_NO_OFFSET -1 505 506 /* 507 * This structure describes the vnode operation taking place. 508 */ 509 struct vnodeop_desc { 510 char *vdesc_name; /* a readable name for debugging */ 511 int vdesc_flags; /* VDESC_* flags */ 512 int vdesc_vop_offset; 513 vop_bypass_t *vdesc_call; /* Function to call */ 514 515 /* 516 * These ops are used by bypass routines to map and locate arguments. 517 * Creds and procs are not needed in bypass routines, but sometimes 518 * they are useful to (for example) transport layers. 519 * Nameidata is useful because it has a cred in it. 520 */ 521 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */ 522 int vdesc_vpp_offset; /* return vpp location */ 523 int vdesc_cred_offset; /* cred location, if any */ 524 int vdesc_thread_offset; /* thread location, if any */ 525 int vdesc_componentname_offset; /* if any */ 526 }; 527 528 #ifdef _KERNEL 529 /* 530 * A list of all the operation descs. 531 */ 532 extern struct vnodeop_desc *vnodeop_descs[]; 533 534 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field) 535 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \ 536 ((s_type)(((char*)(struct_p)) + (s_offset))) 537 538 #ifdef DEBUG_VFS_LOCKS 539 /* 540 * Support code to aid in debugging VFS locking problems. Not totally 541 * reliable since if the thread sleeps between changing the lock 542 * state and checking it with the assert, some other thread could 543 * change the state. They are good enough for debugging a single 544 * filesystem using a single-threaded test. Note that the unreliability is 545 * limited to false negatives; efforts were made to ensure that false 546 * positives cannot occur. 547 */ 548 void assert_vi_locked(struct vnode *vp, const char *str); 549 void assert_vi_unlocked(struct vnode *vp, const char *str); 550 void assert_vop_elocked(struct vnode *vp, const char *str); 551 void assert_vop_locked(struct vnode *vp, const char *str); 552 void assert_vop_unlocked(struct vnode *vp, const char *str); 553 554 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str)) 555 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str)) 556 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str)) 557 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str)) 558 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str)) 559 560 #define ASSERT_VOP_IN_SEQC(vp) do { \ 561 struct vnode *_vp = (vp); \ 562 \ 563 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \ 564 } while (0) 565 566 #define ASSERT_VOP_NOT_IN_SEQC(vp) do { \ 567 struct vnode *_vp = (vp); \ 568 \ 569 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \ 570 } while (0) 571 572 #else /* !DEBUG_VFS_LOCKS */ 573 574 #define ASSERT_VI_LOCKED(vp, str) ((void)0) 575 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0) 576 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0) 577 #define ASSERT_VOP_LOCKED(vp, str) ((void)0) 578 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0) 579 580 #define ASSERT_VOP_IN_SEQC(vp) ((void)0) 581 #define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0) 582 583 #endif /* DEBUG_VFS_LOCKS */ 584 585 /* 586 * This call works for vnodes in the kernel. 587 */ 588 #define VCALL(c) ((c)->a_desc->vdesc_call(c)) 589 590 #define DOINGASYNC(vp) \ 591 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \ 592 ((curthread->td_pflags & TDP_SYNCIO) == 0)) 593 594 /* 595 * VMIO support inline 596 */ 597 598 extern int vmiodirenable; 599 600 static __inline int 601 vn_canvmio(struct vnode *vp) 602 { 603 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR))) 604 return(TRUE); 605 return(FALSE); 606 } 607 608 /* 609 * Finally, include the default set of vnode operations. 610 */ 611 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int); 612 #include "vnode_if.h" 613 614 /* vn_open_flags */ 615 #define VN_OPEN_NOAUDIT 0x00000001 616 #define VN_OPEN_NOCAPCHECK 0x00000002 617 #define VN_OPEN_NAMECACHE 0x00000004 618 #define VN_OPEN_INVFS 0x00000008 619 620 /* copy_file_range kernel flags */ 621 #define COPY_FILE_RANGE_KFLAGS 0xff000000 622 #define COPY_FILE_RANGE_TIMEO1SEC 0x01000000 /* Return after 1sec. */ 623 624 /* 625 * Public vnode manipulation functions. 626 */ 627 struct componentname; 628 struct file; 629 struct mount; 630 struct nameidata; 631 struct ostat; 632 struct freebsd11_stat; 633 struct thread; 634 struct proc; 635 struct stat; 636 struct nstat; 637 struct ucred; 638 struct uio; 639 struct vattr; 640 struct vfsops; 641 struct vnode; 642 643 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **); 644 645 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn, 646 daddr_t endn); 647 /* cache_* may belong in namei.h. */ 648 void cache_changesize(u_long newhashsize); 649 650 #define VFS_CACHE_DROPOLD 0x1 651 652 void cache_enter_time_flags(struct vnode *dvp, struct vnode *vp, 653 struct componentname *cnp, struct timespec *tsp, 654 struct timespec *dtsp, int flags); 655 #define cache_enter(dvp, vp, cnp) \ 656 cache_enter_time(dvp, vp, cnp, NULL, NULL) 657 void cache_enter_time(struct vnode *dvp, struct vnode *vp, 658 struct componentname *cnp, struct timespec *tsp, 659 struct timespec *dtsp); 660 int cache_lookup(struct vnode *dvp, struct vnode **vpp, 661 struct componentname *cnp, struct timespec *tsp, int *ticksp); 662 void cache_vnode_init(struct vnode *vp); 663 void cache_purge(struct vnode *vp); 664 void cache_purge_vgone(struct vnode *vp); 665 void cache_purge_negative(struct vnode *vp); 666 void cache_purgevfs(struct mount *mp); 667 char *cache_symlink_alloc(size_t size, int flags); 668 void cache_symlink_free(char *string, size_t size); 669 int cache_symlink_resolve(struct cache_fpl *fpl, const char *string, 670 size_t len); 671 void cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp, 672 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp); 673 void cache_vop_rmdir(struct vnode *dvp, struct vnode *vp); 674 #ifdef INVARIANTS 675 void cache_validate(struct vnode *dvp, struct vnode *vp, 676 struct componentname *cnp); 677 #else 678 static inline void 679 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp) 680 { 681 } 682 #endif 683 void cache_fast_lookup_enabled_recalc(void); 684 int change_dir(struct vnode *vp, struct thread *td); 685 void cvtstat(struct stat *st, struct ostat *ost); 686 int freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb); 687 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost); 688 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops, 689 struct vnode **vpp); 690 void getnewvnode_reserve(void); 691 void getnewvnode_drop_reserve(void); 692 int insmntque(struct vnode *vp, struct mount *mp); 693 int insmntque1(struct vnode *vp, struct mount *mp); 694 u_quad_t init_va_filerev(void); 695 int speedup_syncer(void); 696 int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen); 697 int vn_getcwd(char *buf, char **retbuf, size_t *buflen); 698 int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf); 699 int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf); 700 int vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp, 701 const char *hdrl_name, size_t hrdl_name_length, char **retbuf, 702 char **freebuf, size_t *buflen); 703 struct vnode * 704 vn_dir_dd_ino(struct vnode *vp); 705 int vn_commname(struct vnode *vn, char *buf, u_int buflen); 706 int vn_path_to_global_path(struct thread *td, struct vnode *vp, 707 char *path, u_int pathlen); 708 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid, 709 gid_t file_gid, accmode_t accmode, struct ucred *cred); 710 int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid, 711 struct ucred *cred); 712 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid, 713 struct acl *aclp, accmode_t accmode, struct ucred *cred); 714 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid, 715 gid_t file_gid, struct acl *acl, accmode_t accmode, 716 struct ucred *cred); 717 void vattr_null(struct vattr *vap); 718 void vlazy(struct vnode *); 719 void vdrop(struct vnode *); 720 void vdropl(struct vnode *); 721 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td); 722 int vget(struct vnode *vp, int flags); 723 enum vgetstate vget_prep_smr(struct vnode *vp); 724 enum vgetstate vget_prep(struct vnode *vp); 725 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs); 726 void vget_finish_ref(struct vnode *vp, enum vgetstate vs); 727 void vget_abort(struct vnode *vp, enum vgetstate vs); 728 void vgone(struct vnode *vp); 729 void vhold(struct vnode *); 730 void vholdnz(struct vnode *); 731 bool vhold_smr(struct vnode *); 732 int vinactive(struct vnode *vp); 733 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo); 734 int vtruncbuf(struct vnode *vp, off_t length, int blksize); 735 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn, 736 int blksize); 737 void vunref(struct vnode *); 738 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3); 739 int vrecycle(struct vnode *vp); 740 int vrecyclel(struct vnode *vp); 741 int vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off, 742 struct ucred *cred); 743 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off, 744 struct ucred *cred); 745 int vn_close(struct vnode *vp, 746 int flags, struct ucred *file_cred, struct thread *td); 747 int vn_copy_file_range(struct vnode *invp, off_t *inoffp, 748 struct vnode *outvp, off_t *outoffp, size_t *lenp, 749 unsigned int flags, struct ucred *incred, struct ucred *outcred, 750 struct thread *fsize_td); 751 int vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags, 752 int ioflg, struct ucred *active_cred, struct ucred *file_cred); 753 void vn_finished_write(struct mount *mp); 754 void vn_finished_secondary_write(struct mount *mp); 755 int vn_fsync_buf(struct vnode *vp, int waitfor); 756 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp, 757 struct vnode *outvp, off_t *outoffp, size_t *lenp, 758 unsigned int flags, struct ucred *incred, struct ucred *outcred, 759 struct thread *fsize_td); 760 int vn_need_pageq_flush(struct vnode *vp); 761 bool vn_isdisk_error(struct vnode *vp, int *errp); 762 bool vn_isdisk(struct vnode *vp); 763 int _vn_lock(struct vnode *vp, int flags, const char *file, int line); 764 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__) 765 void vn_lock_pair(struct vnode *vp1, bool vp1_locked, struct vnode *vp2, 766 bool vp2_locked); 767 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp); 768 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode, 769 u_int vn_open_flags, struct ucred *cred, struct file *fp); 770 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred, 771 struct thread *td, struct file *fp); 772 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end); 773 void vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start, 774 vm_pindex_t end); 775 int vn_pollrecord(struct vnode *vp, struct thread *p, int events); 776 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, 777 int len, off_t offset, enum uio_seg segflg, int ioflg, 778 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid, 779 struct thread *td); 780 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base, 781 size_t len, off_t offset, enum uio_seg segflg, int ioflg, 782 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid, 783 struct thread *td); 784 int vn_read_from_obj(struct vnode *vp, struct uio *uio); 785 int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio, 786 struct thread *td); 787 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags); 788 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp, 789 int flags); 790 int vn_truncate_locked(struct vnode *vp, off_t length, bool sync, 791 struct ucred *cred); 792 int vn_writechk(struct vnode *vp); 793 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 794 const char *attrname, int *buflen, char *buf, struct thread *td); 795 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 796 const char *attrname, int buflen, char *buf, struct thread *td); 797 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 798 const char *attrname, struct thread *td); 799 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags, 800 struct vnode **rvp); 801 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc, 802 void *alloc_arg, int lkflags, struct vnode **rvp); 803 int vn_utimes_perm(struct vnode *vp, struct vattr *vap, 804 struct ucred *cred, struct thread *td); 805 806 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio); 807 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize, 808 struct uio *uio); 809 810 void vn_seqc_write_begin_locked(struct vnode *vp); 811 void vn_seqc_write_begin(struct vnode *vp); 812 void vn_seqc_write_end_locked(struct vnode *vp); 813 void vn_seqc_write_end(struct vnode *vp); 814 #define vn_seqc_read_any(vp) seqc_read_any(&(vp)->v_seqc) 815 #define vn_seqc_read_notmodify(vp) seqc_read_notmodify(&(vp)->v_seqc) 816 #define vn_seqc_consistent(vp, seq) seqc_consistent(&(vp)->v_seqc, seq) 817 818 #define vn_rangelock_unlock(vp, cookie) \ 819 rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp)) 820 #define vn_rangelock_unlock_range(vp, cookie, start, end) \ 821 rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), \ 822 VI_MTX(vp)) 823 #define vn_rangelock_rlock(vp, start, end) \ 824 rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 825 #define vn_rangelock_tryrlock(vp, start, end) \ 826 rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 827 #define vn_rangelock_wlock(vp, start, end) \ 828 rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 829 #define vn_rangelock_trywlock(vp, start, end) \ 830 rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 831 832 #define vn_irflag_read(vp) atomic_load_short(&(vp)->v_irflag) 833 void vn_irflag_set_locked(struct vnode *vp, short toset); 834 void vn_irflag_set(struct vnode *vp, short toset); 835 void vn_irflag_set_cond_locked(struct vnode *vp, short toset); 836 void vn_irflag_set_cond(struct vnode *vp, short toset); 837 void vn_irflag_unset_locked(struct vnode *vp, short tounset); 838 void vn_irflag_unset(struct vnode *vp, short tounset); 839 840 int vfs_cache_lookup(struct vop_lookup_args *ap); 841 int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp); 842 void vfs_timestamp(struct timespec *); 843 void vfs_write_resume(struct mount *mp, int flags); 844 int vfs_write_suspend(struct mount *mp, int flags); 845 int vfs_write_suspend_umnt(struct mount *mp); 846 struct vnode *vnlru_alloc_marker(void); 847 void vnlru_free_marker(struct vnode *); 848 void vnlru_free_vfsops(int, struct vfsops *, struct vnode *); 849 int vop_stdbmap(struct vop_bmap_args *); 850 int vop_stdfdatasync_buf(struct vop_fdatasync_args *); 851 int vop_stdfsync(struct vop_fsync_args *); 852 int vop_stdgetwritemount(struct vop_getwritemount_args *); 853 int vop_stdgetpages(struct vop_getpages_args *); 854 int vop_stdinactive(struct vop_inactive_args *); 855 int vop_stdioctl(struct vop_ioctl_args *); 856 int vop_stdneed_inactive(struct vop_need_inactive_args *); 857 int vop_stdkqfilter(struct vop_kqfilter_args *); 858 int vop_stdlock(struct vop_lock1_args *); 859 int vop_stdunlock(struct vop_unlock_args *); 860 int vop_stdislocked(struct vop_islocked_args *); 861 int vop_lock(struct vop_lock1_args *); 862 int vop_unlock(struct vop_unlock_args *); 863 int vop_islocked(struct vop_islocked_args *); 864 int vop_stdputpages(struct vop_putpages_args *); 865 int vop_nopoll(struct vop_poll_args *); 866 int vop_stdaccess(struct vop_access_args *ap); 867 int vop_stdaccessx(struct vop_accessx_args *ap); 868 int vop_stdadvise(struct vop_advise_args *ap); 869 int vop_stdadvlock(struct vop_advlock_args *ap); 870 int vop_stdadvlockasync(struct vop_advlockasync_args *ap); 871 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap); 872 int vop_stdallocate(struct vop_allocate_args *ap); 873 int vop_stddeallocate(struct vop_deallocate_args *ap); 874 int vop_stdset_text(struct vop_set_text_args *ap); 875 int vop_stdpathconf(struct vop_pathconf_args *); 876 int vop_stdpoll(struct vop_poll_args *); 877 int vop_stdvptocnp(struct vop_vptocnp_args *ap); 878 int vop_stdvptofh(struct vop_vptofh_args *ap); 879 int vop_stdunp_bind(struct vop_unp_bind_args *ap); 880 int vop_stdunp_connect(struct vop_unp_connect_args *ap); 881 int vop_stdunp_detach(struct vop_unp_detach_args *ap); 882 int vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap); 883 int vop_eopnotsupp(struct vop_generic_args *ap); 884 int vop_ebadf(struct vop_generic_args *ap); 885 int vop_einval(struct vop_generic_args *ap); 886 int vop_enoent(struct vop_generic_args *ap); 887 int vop_enotty(struct vop_generic_args *ap); 888 int vop_eagain(struct vop_generic_args *ap); 889 int vop_null(struct vop_generic_args *ap); 890 int vop_panic(struct vop_generic_args *ap); 891 int dead_poll(struct vop_poll_args *ap); 892 int dead_read(struct vop_read_args *ap); 893 int dead_write(struct vop_write_args *ap); 894 895 /* These are called from within the actual VOPS. */ 896 void vop_close_post(void *a, int rc); 897 void vop_create_pre(void *a); 898 void vop_create_post(void *a, int rc); 899 void vop_whiteout_pre(void *a); 900 void vop_whiteout_post(void *a, int rc); 901 void vop_deleteextattr_pre(void *a); 902 void vop_deleteextattr_post(void *a, int rc); 903 void vop_link_pre(void *a); 904 void vop_link_post(void *a, int rc); 905 void vop_lookup_post(void *a, int rc); 906 void vop_lookup_pre(void *a); 907 void vop_mkdir_pre(void *a); 908 void vop_mkdir_post(void *a, int rc); 909 void vop_mknod_pre(void *a); 910 void vop_mknod_post(void *a, int rc); 911 void vop_open_post(void *a, int rc); 912 void vop_read_post(void *a, int rc); 913 void vop_read_pgcache_post(void *ap, int rc); 914 void vop_readdir_post(void *a, int rc); 915 void vop_reclaim_post(void *a, int rc); 916 void vop_remove_pre(void *a); 917 void vop_remove_post(void *a, int rc); 918 void vop_rename_post(void *a, int rc); 919 void vop_rename_pre(void *a); 920 void vop_rmdir_pre(void *a); 921 void vop_rmdir_post(void *a, int rc); 922 void vop_setattr_pre(void *a); 923 void vop_setattr_post(void *a, int rc); 924 void vop_setacl_pre(void *a); 925 void vop_setacl_post(void *a, int rc); 926 void vop_setextattr_pre(void *a); 927 void vop_setextattr_post(void *a, int rc); 928 void vop_symlink_pre(void *a); 929 void vop_symlink_post(void *a, int rc); 930 int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a); 931 932 #ifdef DEBUG_VFS_LOCKS 933 void vop_fdatasync_debugpre(void *a); 934 void vop_fdatasync_debugpost(void *a, int rc); 935 void vop_fplookup_vexec_debugpre(void *a); 936 void vop_fplookup_vexec_debugpost(void *a, int rc); 937 void vop_fplookup_symlink_debugpre(void *a); 938 void vop_fplookup_symlink_debugpost(void *a, int rc); 939 void vop_fsync_debugpre(void *a); 940 void vop_fsync_debugpost(void *a, int rc); 941 void vop_strategy_debugpre(void *a); 942 void vop_lock_debugpre(void *a); 943 void vop_lock_debugpost(void *a, int rc); 944 void vop_unlock_debugpre(void *a); 945 void vop_need_inactive_debugpre(void *a); 946 void vop_need_inactive_debugpost(void *a, int rc); 947 void vop_mkdir_debugpost(void *a, int rc); 948 #else 949 #define vop_fdatasync_debugpre(x) do { } while (0) 950 #define vop_fdatasync_debugpost(x, y) do { } while (0) 951 #define vop_fplookup_vexec_debugpre(x) do { } while (0) 952 #define vop_fplookup_vexec_debugpost(x, y) do { } while (0) 953 #define vop_fplookup_symlink_debugpre(x) do { } while (0) 954 #define vop_fplookup_symlink_debugpost(x, y) do { } while (0) 955 #define vop_fsync_debugpre(x) do { } while (0) 956 #define vop_fsync_debugpost(x, y) do { } while (0) 957 #define vop_strategy_debugpre(x) do { } while (0) 958 #define vop_lock_debugpre(x) do { } while (0) 959 #define vop_lock_debugpost(x, y) do { } while (0) 960 #define vop_unlock_debugpre(x) do { } while (0) 961 #define vop_need_inactive_debugpre(x) do { } while (0) 962 #define vop_need_inactive_debugpost(x, y) do { } while (0) 963 #define vop_mkdir_debugpost(x, y) do { } while (0) 964 #endif 965 966 void vop_rename_fail(struct vop_rename_args *ap); 967 968 #define vop_stat_helper_pre(ap) ({ \ 969 int _error; \ 970 AUDIT_ARG_VNODE1(ap->a_vp); \ 971 _error = mac_vnode_check_stat(ap->a_active_cred, ap->a_file_cred, ap->a_vp);\ 972 if (__predict_true(_error == 0)) \ 973 bzero(ap->a_sb, sizeof(*ap->a_sb)); \ 974 _error; \ 975 }) 976 977 #define vop_stat_helper_post(ap, error) ({ \ 978 int _error = (error); \ 979 if (priv_check_cred_vfs_generation(ap->a_active_cred)) \ 980 ap->a_sb->st_gen = 0; \ 981 _error; \ 982 }) 983 984 #define VOP_WRITE_PRE(ap) \ 985 struct vattr va; \ 986 int error; \ 987 off_t osize, ooffset, noffset; \ 988 \ 989 osize = ooffset = noffset = 0; \ 990 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \ 991 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \ 992 if (error) \ 993 return (error); \ 994 ooffset = (ap)->a_uio->uio_offset; \ 995 osize = (off_t)va.va_size; \ 996 } 997 998 #define VOP_WRITE_POST(ap, ret) \ 999 noffset = (ap)->a_uio->uio_offset; \ 1000 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \ 1001 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \ 1002 | (noffset > osize ? NOTE_EXTEND : 0)); \ 1003 } 1004 1005 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__) 1006 1007 #ifdef INVARIANTS 1008 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \ 1009 do { \ 1010 int error_; \ 1011 \ 1012 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \ 1013 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \ 1014 error_)); \ 1015 } while (0) 1016 #define VOP_SET_TEXT_CHECKED(vp) \ 1017 do { \ 1018 int error_; \ 1019 \ 1020 error_ = VOP_SET_TEXT((vp)); \ 1021 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \ 1022 error_)); \ 1023 } while (0) 1024 #define VOP_UNSET_TEXT_CHECKED(vp) \ 1025 do { \ 1026 int error_; \ 1027 \ 1028 error_ = VOP_UNSET_TEXT((vp)); \ 1029 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \ 1030 error_)); \ 1031 } while (0) 1032 #else 1033 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt)) 1034 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp)) 1035 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp)) 1036 #endif 1037 1038 #define VN_IS_DOOMED(vp) __predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0) 1039 1040 void vput(struct vnode *vp); 1041 void vrele(struct vnode *vp); 1042 void vref(struct vnode *vp); 1043 void vrefact(struct vnode *vp); 1044 void v_addpollinfo(struct vnode *vp); 1045 static __inline int 1046 vrefcnt(struct vnode *vp) 1047 { 1048 1049 return (vp->v_usecount); 1050 } 1051 1052 #define vholdl(vp) do { \ 1053 ASSERT_VI_LOCKED(vp, __func__); \ 1054 vhold(vp); \ 1055 } while (0) 1056 1057 #define vrefl(vp) do { \ 1058 ASSERT_VI_LOCKED(vp, __func__); \ 1059 vref(vp); \ 1060 } while (0) 1061 1062 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td); 1063 void vnode_destroy_vobject(struct vnode *vp); 1064 1065 extern struct vop_vector fifo_specops; 1066 extern struct vop_vector dead_vnodeops; 1067 extern struct vop_vector default_vnodeops; 1068 1069 #define VOP_PANIC ((void*)(uintptr_t)vop_panic) 1070 #define VOP_NULL ((void*)(uintptr_t)vop_null) 1071 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf) 1072 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty) 1073 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval) 1074 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent) 1075 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp) 1076 #define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain) 1077 1078 /* fifo_vnops.c */ 1079 int fifo_printinfo(struct vnode *); 1080 1081 /* vfs_hash.c */ 1082 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg); 1083 1084 void vfs_hash_changesize(u_long newhashsize); 1085 int vfs_hash_get(const struct mount *mp, u_int hash, int flags, 1086 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1087 u_int vfs_hash_index(struct vnode *vp); 1088 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, 1089 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1090 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td, 1091 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1092 void vfs_hash_rehash(struct vnode *vp, u_int hash); 1093 void vfs_hash_remove(struct vnode *vp); 1094 1095 int vfs_kqfilter(struct vop_kqfilter_args *); 1096 struct dirent; 1097 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off); 1098 int vfs_emptydir(struct vnode *vp); 1099 1100 int vfs_unixify_accmode(accmode_t *accmode); 1101 1102 void vfs_unp_reclaim(struct vnode *vp); 1103 1104 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode); 1105 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid, 1106 gid_t gid); 1107 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, 1108 struct thread *td); 1109 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred, 1110 struct thread *td); 1111 1112 void vn_fsid(struct vnode *vp, struct vattr *va); 1113 1114 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp); 1115 int vn_lktype_write(struct mount *mp, struct vnode *vp); 1116 1117 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \ 1118 struct vnode *_vp = (vp); \ 1119 int _flags = (flags); \ 1120 int _error; \ 1121 \ 1122 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \ 1123 panic("%s: unsupported flags %x\n", __func__, flags); \ 1124 _error = VOP_UNLOCK(_vp); \ 1125 if (_flags & LK_INTERLOCK) \ 1126 VI_UNLOCK(_vp); \ 1127 _error; \ 1128 }) 1129 1130 #include <sys/kernel.h> 1131 1132 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \ 1133 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \ 1134 vfs_vector_op_register, &vnodeops) 1135 1136 #define VFS_SMR_DECLARE \ 1137 extern smr_t vfs_smr 1138 1139 #define VFS_SMR() vfs_smr 1140 #define vfs_smr_enter() smr_enter(VFS_SMR()) 1141 #define vfs_smr_exit() smr_exit(VFS_SMR()) 1142 #define vfs_smr_synchronize() smr_synchronize(VFS_SMR()) 1143 #define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR()) 1144 #define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR()) 1145 #define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR()) 1146 #define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR()) 1147 1148 #define vn_load_v_data_smr(vp) ({ \ 1149 struct vnode *_vp = (vp); \ 1150 \ 1151 VFS_SMR_ASSERT_ENTERED(); \ 1152 atomic_load_consume_ptr(&(_vp)->v_data);\ 1153 }) 1154 1155 #endif /* _KERNEL */ 1156 1157 #endif /* !_SYS_VNODE_H_ */ 1158