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 #define VLASTTYPE VMARKER 62 63 enum vstate { VSTATE_UNINITIALIZED, VSTATE_CONSTRUCTED, VSTATE_DESTROYING, 64 VSTATE_DEAD }; 65 #define VLASTSTATE VSTATE_DEAD 66 67 enum vgetstate { VGET_NONE, VGET_HOLDCNT, VGET_USECOUNT }; 68 /* 69 * Each underlying filesystem allocates its own private area and hangs 70 * it from v_data. If non-null, this area is freed in getnewvnode(). 71 */ 72 73 struct namecache; 74 struct cache_fpl; 75 76 struct vpollinfo { 77 struct mtx vpi_lock; /* lock to protect below */ 78 struct selinfo vpi_selinfo; /* identity of poller(s) */ 79 short vpi_events; /* what they are looking for */ 80 short vpi_revents; /* what has happened */ 81 }; 82 83 /* 84 * Reading or writing any of these items requires holding the appropriate lock. 85 * 86 * Lock reference: 87 * c - namecache mutex 88 * i - interlock 89 * l - mp mnt_listmtx or freelist mutex 90 * I - updated with atomics, 0->1 and 1->0 transitions with interlock held 91 * m - mount point interlock 92 * p - pollinfo lock 93 * u - Only a reference to the vnode is needed to read. 94 * v - vnode lock 95 * 96 * Vnodes may be found on many lists. The general way to deal with operating 97 * on a vnode that is on a list is: 98 * 1) Lock the list and find the vnode. 99 * 2) Lock interlock so that the vnode does not go away. 100 * 3) Unlock the list to avoid lock order reversals. 101 * 4) vget with LK_INTERLOCK and check for ENOENT, or 102 * 5) Check for DOOMED if the vnode lock is not required. 103 * 6) Perform your operation, then vput(). 104 */ 105 106 #if defined(_KERNEL) || defined(_KVM_VNODE) 107 108 struct vnode { 109 /* 110 * Fields which define the identity of the vnode. These fields are 111 * owned by the filesystem (XXX: and vgone() ?) 112 */ 113 enum vtype v_type:8; /* u vnode type */ 114 enum vstate v_state:8; /* u vnode state */ 115 short v_irflag; /* i frequently read flags */ 116 seqc_t v_seqc; /* i modification count */ 117 uint32_t v_nchash; /* u namecache hash */ 118 u_int v_hash; 119 struct vop_vector *v_op; /* u vnode operations vector */ 120 void *v_data; /* u private data for fs */ 121 122 /* 123 * Filesystem instance stuff 124 */ 125 struct mount *v_mount; /* u ptr to vfs we are in */ 126 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */ 127 128 /* 129 * Type specific fields, only one applies to any given vnode. 130 */ 131 union { 132 struct mount *v_mountedhere; /* v ptr to mountpoint (VDIR) */ 133 struct unpcb *v_unpcb; /* v unix domain net (VSOCK) */ 134 struct cdev *v_rdev; /* v device (VCHR, VBLK) */ 135 struct fifoinfo *v_fifoinfo; /* v fifo (VFIFO) */ 136 }; 137 138 /* 139 * vfs_hash: (mount + inode) -> vnode hash. The hash value 140 * itself is grouped with other int fields, to avoid padding. 141 */ 142 LIST_ENTRY(vnode) v_hashlist; 143 144 /* 145 * VFS_namecache stuff 146 */ 147 LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */ 148 TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */ 149 struct namecache *v_cache_dd; /* c Cache entry for .. vnode */ 150 151 /* 152 * Locking 153 */ 154 struct lock v_lock; /* u (if fs don't have one) */ 155 struct mtx v_interlock; /* lock for "i" things */ 156 struct lock *v_vnlock; /* u pointer to vnode lock */ 157 158 /* 159 * The machinery of being a vnode 160 */ 161 TAILQ_ENTRY(vnode) v_vnodelist; /* l vnode lists */ 162 TAILQ_ENTRY(vnode) v_lazylist; /* l vnode lazy list */ 163 struct bufobj v_bufobj; /* * Buffer cache object */ 164 165 /* 166 * Hooks for various subsystems and features. 167 */ 168 struct vpollinfo *v_pollinfo; /* i Poll events, p for *v_pi */ 169 struct label *v_label; /* MAC label for vnode */ 170 struct lockf *v_lockf; /* Byte-level advisory lock list */ 171 struct rangelock v_rl; /* Byte-range lock */ 172 173 u_int v_holdcnt; /* I prevents recycling. */ 174 u_int v_usecount; /* I ref count of users */ 175 u_short v_iflag; /* i vnode flags (see below) */ 176 u_short v_vflag; /* v vnode flags */ 177 u_short v_mflag; /* l mnt-specific vnode flags */ 178 short v_dbatchcpu; /* i LRU requeue deferral batch */ 179 int v_writecount; /* I ref count of writers or 180 (negative) text users */ 181 int v_seqc_users; /* i modifications pending */ 182 }; 183 184 #ifndef DEBUG_LOCKS 185 #ifdef _LP64 186 /* 187 * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields 188 * to the structure and there is nothing which can be done to prevent growth 189 * then so be it. But don't grow it without a good reason. 190 */ 191 _Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes"); 192 #endif 193 #endif 194 195 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */ 196 197 #define bo2vnode(bo) __containerof((bo), struct vnode, v_bufobj) 198 199 /* XXX: These are temporary to avoid a source sweep at this time */ 200 #define v_object v_bufobj.bo_object 201 202 /* We don't need to lock the knlist */ 203 #define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \ 204 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note)) 205 206 #define VN_KNOTE(vp, b, a) \ 207 do { \ 208 if (!VN_KNLIST_EMPTY(vp)) \ 209 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \ 210 (a) | KNF_NOKQLOCK); \ 211 } while (0) 212 #define VN_KNOTE_LOCKED(vp, b) VN_KNOTE(vp, b, KNF_LISTLOCKED) 213 #define VN_KNOTE_UNLOCKED(vp, b) VN_KNOTE(vp, b, 0) 214 215 /* 216 * Vnode flags. 217 * VI flags are protected by interlock and live in v_iflag 218 * VV flags are protected by the vnode lock and live in v_vflag 219 * 220 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both 221 * are required for writing but the status may be checked with either. 222 */ 223 #define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */ 224 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR) 225 226 #define VIRF_DOOMED 0x0001 /* This vnode is being recycled */ 227 #define VIRF_PGREAD 0x0002 /* Direct reads from the page cache are permitted, 228 never cleared once set */ 229 #define VIRF_MOUNTPOINT 0x0004 /* This vnode is mounted on */ 230 #define VIRF_TEXT_REF 0x0008 /* Executable mappings ref the vnode */ 231 #define VIRF_CROSSMP 0x0010 /* Cross-mp vnode, no locking */ 232 233 #define VI_UNUSED0 0x0001 /* unused */ 234 #define VI_MOUNT 0x0002 /* Mount in progress */ 235 #define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */ 236 #define VI_OWEINACT 0x0008 /* Need to call inactive */ 237 #define VI_DEFINACT 0x0010 /* deferred inactive */ 238 #define VI_FOPENING 0x0020 /* In open, with opening process having the 239 first right to advlock file */ 240 241 #define VV_ROOT 0x0001 /* root of its filesystem */ 242 #define VV_ISTTY 0x0002 /* vnode represents a tty */ 243 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */ 244 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */ 245 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */ 246 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */ 247 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */ 248 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */ 249 #define VV_PROCDEP 0x0100 /* vnode is process dependent */ 250 #define VV_UNLINKED 0x0200 /* unlinked but stil open directory */ 251 #define VV_DELETED 0x0400 /* should be removed */ 252 #define VV_MD 0x0800 /* vnode backs the md device */ 253 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */ 254 #define VV_READLINK 0x2000 /* fdescfs linux vnode */ 255 #define VV_UNREF 0x4000 /* vunref, do not drop lock in inactive() */ 256 #define VV_CROSSLOCK 0x8000 /* vnode lock is shared w/ root mounted here */ 257 258 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */ 259 260 /* 261 * Vnode attributes. A field value of VNOVAL represents a field whose value 262 * is unavailable (getattr) or which is not to be changed (setattr). 263 */ 264 struct vattr { 265 enum vtype va_type; /* vnode type (for create) */ 266 u_short va_mode; /* files access mode and type */ 267 u_short va_padding0; 268 uid_t va_uid; /* owner user id */ 269 gid_t va_gid; /* owner group id */ 270 nlink_t va_nlink; /* number of references to file */ 271 dev_t va_fsid; /* filesystem id */ 272 ino_t va_fileid; /* file id */ 273 u_quad_t va_size; /* file size in bytes */ 274 long va_blocksize; /* blocksize preferred for i/o */ 275 struct timespec va_atime; /* time of last access */ 276 struct timespec va_mtime; /* time of last modification */ 277 struct timespec va_ctime; /* time file changed */ 278 struct timespec va_birthtime; /* time file created */ 279 u_long va_gen; /* generation number of file */ 280 u_long va_flags; /* flags defined for file */ 281 dev_t va_rdev; /* device the special file represents */ 282 u_quad_t va_bytes; /* bytes of disk space held by file */ 283 u_quad_t va_filerev; /* file modification number */ 284 u_int va_vaflags; /* operations flags, see below */ 285 long va_spare; /* remain quad aligned */ 286 }; 287 288 /* 289 * Flags for va_vaflags. 290 */ 291 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */ 292 #define VA_EXCLUSIVE 0x02 /* exclusive create request */ 293 #define VA_SYNC 0x04 /* O_SYNC truncation */ 294 295 /* 296 * Flags for ioflag. (high 16 bits used to ask for read-ahead and 297 * help with write clustering) 298 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h 299 */ 300 #define IO_UNIT 0x0001 /* do I/O as atomic unit */ 301 #define IO_APPEND 0x0002 /* append write to end */ 302 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */ 303 #define IO_NODELOCKED 0x0008 /* underlying node already locked */ 304 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */ 305 #define IO_VMIO 0x0020 /* data already in VMIO space */ 306 #define IO_INVAL 0x0040 /* invalidate after I/O */ 307 #define IO_SYNC 0x0080 /* do I/O synchronously */ 308 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */ 309 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */ 310 #define IO_EXT 0x0400 /* operate on external attributes */ 311 #define IO_NORMAL 0x0800 /* operate on regular data */ 312 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */ 313 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */ 314 #define IO_RANGELOCKED 0x4000 /* range locked */ 315 #define IO_DATASYNC 0x8000 /* do only data I/O synchronously */ 316 317 #define IO_SEQMAX 0x7F /* seq heuristic max value */ 318 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */ 319 320 /* 321 * Flags for accmode_t. 322 */ 323 #define VEXEC 000000000100 /* execute/search permission */ 324 #define VWRITE 000000000200 /* write permission */ 325 #define VREAD 000000000400 /* read permission */ 326 #define VADMIN 000000010000 /* being the file owner */ 327 #define VAPPEND 000000040000 /* permission to write/append */ 328 /* 329 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only 330 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL, 331 * and 0 otherwise. This never happens with ordinary unix access rights 332 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with 333 * some other V* constant. 334 */ 335 #define VEXPLICIT_DENY 000000100000 336 #define VREAD_NAMED_ATTRS 000000200000 /* not used */ 337 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */ 338 #define VDELETE_CHILD 000001000000 339 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */ 340 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */ 341 #define VDELETE 000010000000 342 #define VREAD_ACL 000020000000 /* read ACL and file mode */ 343 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */ 344 #define VWRITE_OWNER 000100000000 /* change file owner */ 345 #define VSYNCHRONIZE 000200000000 /* not used */ 346 #define VCREAT 000400000000 /* creating new file */ 347 #define VVERIFY 001000000000 /* verification required */ 348 349 /* 350 * Permissions that were traditionally granted only to the file owner. 351 */ 352 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \ 353 VWRITE_OWNER) 354 355 /* 356 * Permissions that were traditionally granted to everyone. 357 */ 358 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL) 359 360 /* 361 * Permissions that allow to change the state of the file in any way. 362 */ 363 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \ 364 VDELETE) 365 366 /* 367 * Token indicating no attribute value yet assigned. 368 */ 369 #define VNOVAL (-1) 370 371 /* 372 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon) 373 */ 374 #define VLKTIMEOUT (hz / 20 + 1) 375 376 #ifdef _KERNEL 377 378 #ifdef MALLOC_DECLARE 379 MALLOC_DECLARE(M_VNODE); 380 #endif 381 382 extern u_int ncsizefactor; 383 extern const u_int io_hold_cnt; 384 385 /* 386 * Convert between vnode types and inode formats (since POSIX.1 387 * defines mode word of stat structure in terms of inode formats). 388 */ 389 extern enum vtype iftovt_tab[]; 390 extern int vttoif_tab[]; 391 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12]) 392 #define VTTOIF(indx) (vttoif_tab[(int)(indx)]) 393 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode)) 394 395 /* 396 * Flags to various vnode functions. 397 */ 398 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */ 399 #define FORCECLOSE 0x0002 /* vflush: force file closure */ 400 #define WRITECLOSE 0x0004 /* vflush: only close writable files */ 401 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */ 402 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */ 403 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */ 404 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */ 405 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */ 406 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */ 407 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */ 408 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */ 409 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */ 410 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */ 411 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */ 412 #define V_PCATCH 0x0008 /* vn_start_write: make the sleep interruptible */ 413 #define V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH) 414 415 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */ 416 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */ 417 418 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the 419 filesystem is being unmounted */ 420 421 #define VREF(vp) vref(vp) 422 423 #ifdef DIAGNOSTIC 424 #define VATTR_NULL(vap) vattr_null(vap) 425 #else 426 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */ 427 #endif /* DIAGNOSTIC */ 428 429 #define NULLVP ((struct vnode *)NULL) 430 431 /* 432 * Global vnode data. 433 */ 434 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */ 435 extern struct mount *rootdevmp; /* "/dev" mount */ 436 extern u_long desiredvnodes; /* number of vnodes desired */ 437 extern struct uma_zone *namei_zone; 438 extern struct vattr va_null; /* predefined null vattr structure */ 439 440 extern u_int vn_lock_pair_pause_max; 441 442 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock) 443 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags)) 444 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock) 445 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock) 446 #define VI_MTX(vp) (&(vp)->v_interlock) 447 448 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock) 449 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock) 450 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock) 451 452 #endif /* _KERNEL */ 453 454 /* 455 * Mods for extensibility. 456 */ 457 458 /* 459 * Flags for vdesc_flags: 460 */ 461 #define VDESC_MAX_VPS 16 462 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */ 463 #define VDESC_VP0_WILLRELE 0x0001 464 #define VDESC_VP1_WILLRELE 0x0002 465 #define VDESC_VP2_WILLRELE 0x0004 466 #define VDESC_VP3_WILLRELE 0x0008 467 468 /* 469 * A generic structure. 470 * This can be used by bypass routines to identify generic arguments. 471 */ 472 struct vop_generic_args { 473 struct vnodeop_desc *a_desc; 474 /* other random data follows, presumably */ 475 }; 476 477 typedef int vop_bypass_t(struct vop_generic_args *); 478 479 /* 480 * VDESC_NO_OFFSET is used to identify the end of the offset list 481 * and in places where no such field exists. 482 */ 483 #define VDESC_NO_OFFSET -1 484 485 /* 486 * This structure describes the vnode operation taking place. 487 */ 488 struct vnodeop_desc { 489 char *vdesc_name; /* a readable name for debugging */ 490 int vdesc_flags; /* VDESC_* flags */ 491 int vdesc_vop_offset; 492 vop_bypass_t *vdesc_call; /* Function to call */ 493 494 /* 495 * These ops are used by bypass routines to map and locate arguments. 496 * Creds and procs are not needed in bypass routines, but sometimes 497 * they are useful to (for example) transport layers. 498 * Nameidata is useful because it has a cred in it. 499 */ 500 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */ 501 int vdesc_vpp_offset; /* return vpp location */ 502 int vdesc_cred_offset; /* cred location, if any */ 503 int vdesc_thread_offset; /* thread location, if any */ 504 int vdesc_componentname_offset; /* if any */ 505 }; 506 507 #ifdef _KERNEL 508 /* 509 * A list of all the operation descs. 510 */ 511 extern struct vnodeop_desc *vnodeop_descs[]; 512 513 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field) 514 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \ 515 ((s_type)(((char*)(struct_p)) + (s_offset))) 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 #define ASSERT_VOP_IN_SEQC(vp) do { \ 540 struct vnode *_vp = (vp); \ 541 \ 542 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \ 543 } while (0) 544 545 #define ASSERT_VOP_NOT_IN_SEQC(vp) do { \ 546 struct vnode *_vp = (vp); \ 547 \ 548 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \ 549 } while (0) 550 551 #else /* !DEBUG_VFS_LOCKS */ 552 553 #define ASSERT_VI_LOCKED(vp, str) ((void)0) 554 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0) 555 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0) 556 #define ASSERT_VOP_LOCKED(vp, str) ((void)0) 557 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0) 558 559 #define ASSERT_VOP_IN_SEQC(vp) ((void)0) 560 #define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0) 561 562 #endif /* DEBUG_VFS_LOCKS */ 563 564 /* 565 * This call works for vnodes in the kernel. 566 */ 567 #define VCALL(c) ((c)->a_desc->vdesc_call(c)) 568 569 #define DOINGASYNC(vp) \ 570 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \ 571 ((curthread->td_pflags & TDP_SYNCIO) == 0)) 572 573 /* 574 * VMIO support inline 575 */ 576 577 extern int vmiodirenable; 578 579 static __inline int 580 vn_canvmio(struct vnode *vp) 581 { 582 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR))) 583 return(TRUE); 584 return(FALSE); 585 } 586 587 /* 588 * Finally, include the default set of vnode operations. 589 */ 590 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int); 591 #include "vnode_if.h" 592 593 /* vn_open_flags */ 594 #define VN_OPEN_NOAUDIT 0x00000001 595 #define VN_OPEN_NOCAPCHECK 0x00000002 596 #define VN_OPEN_NAMECACHE 0x00000004 597 #define VN_OPEN_INVFS 0x00000008 598 #define VN_OPEN_WANTIOCTLCAPS 0x00000010 599 600 /* copy_file_range kernel flags */ 601 #define COPY_FILE_RANGE_KFLAGS 0xff000000 602 #define COPY_FILE_RANGE_TIMEO1SEC 0x01000000 /* Return after 1sec. */ 603 604 /* 605 * Public vnode manipulation functions. 606 */ 607 struct componentname; 608 struct file; 609 struct mount; 610 struct nameidata; 611 struct ostat; 612 struct freebsd11_stat; 613 struct thread; 614 struct proc; 615 struct stat; 616 struct nstat; 617 struct ucred; 618 struct uio; 619 struct vattr; 620 struct vfsops; 621 struct vnode; 622 623 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **); 624 625 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn, 626 daddr_t endn); 627 /* cache_* may belong in namei.h. */ 628 void cache_changesize(u_long newhashsize); 629 630 #define VFS_CACHE_DROPOLD 0x1 631 632 void cache_enter_time_flags(struct vnode *dvp, struct vnode *vp, 633 struct componentname *cnp, struct timespec *tsp, 634 struct timespec *dtsp, int flags); 635 #define cache_enter(dvp, vp, cnp) \ 636 cache_enter_time(dvp, vp, cnp, NULL, NULL) 637 void cache_enter_time(struct vnode *dvp, struct vnode *vp, 638 struct componentname *cnp, struct timespec *tsp, 639 struct timespec *dtsp); 640 int cache_lookup(struct vnode *dvp, struct vnode **vpp, 641 struct componentname *cnp, struct timespec *tsp, int *ticksp); 642 void cache_vnode_init(struct vnode *vp); 643 void cache_purge(struct vnode *vp); 644 void cache_purge_vgone(struct vnode *vp); 645 void cache_purge_negative(struct vnode *vp); 646 void cache_purgevfs(struct mount *mp); 647 char *cache_symlink_alloc(size_t size, int flags); 648 void cache_symlink_free(char *string, size_t size); 649 int cache_symlink_resolve(struct cache_fpl *fpl, const char *string, 650 size_t len); 651 void cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp, 652 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp); 653 void cache_vop_rmdir(struct vnode *dvp, struct vnode *vp); 654 void cache_vop_vector_register(struct vop_vector *); 655 #ifdef INVARIANTS 656 void cache_validate(struct vnode *dvp, struct vnode *vp, 657 struct componentname *cnp); 658 void cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops); 659 void cache_assert_no_entries(struct vnode *vp); 660 #else 661 static inline void 662 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp) 663 { 664 } 665 666 static inline void 667 cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops) 668 { 669 } 670 671 static inline void 672 cache_assert_no_entries(struct vnode *vp) 673 { 674 } 675 #endif 676 void cache_fast_lookup_enabled_recalc(void); 677 int change_dir(struct vnode *vp, struct thread *td); 678 void cvtstat(struct stat *st, struct ostat *ost); 679 int freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb); 680 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost); 681 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops, 682 struct vnode **vpp); 683 void getnewvnode_reserve(void); 684 void getnewvnode_drop_reserve(void); 685 int insmntque(struct vnode *vp, struct mount *mp); 686 int insmntque1(struct vnode *vp, struct mount *mp); 687 u_quad_t init_va_filerev(void); 688 int speedup_syncer(void); 689 int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen); 690 int vn_getcwd(char *buf, char **retbuf, size_t *buflen); 691 int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf); 692 int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf); 693 int vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp, 694 const char *hdrl_name, size_t hrdl_name_length, char **retbuf, 695 char **freebuf, size_t *buflen); 696 struct vnode * 697 vn_dir_dd_ino(struct vnode *vp); 698 int vn_commname(struct vnode *vn, char *buf, u_int buflen); 699 int vn_path_to_global_path(struct thread *td, struct vnode *vp, 700 char *path, u_int pathlen); 701 int vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp, 702 struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name, 703 size_t leaf_length); 704 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid, 705 gid_t file_gid, accmode_t accmode, struct ucred *cred); 706 int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid, 707 struct ucred *cred); 708 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid, 709 struct acl *aclp, accmode_t accmode, struct ucred *cred); 710 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid, 711 gid_t file_gid, struct acl *acl, accmode_t accmode, 712 struct ucred *cred); 713 void vattr_null(struct vattr *vap); 714 void vlazy(struct vnode *); 715 void vdrop(struct vnode *); 716 void vdropl(struct vnode *); 717 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td); 718 int vget(struct vnode *vp, int flags); 719 enum vgetstate vget_prep_smr(struct vnode *vp); 720 enum vgetstate vget_prep(struct vnode *vp); 721 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs); 722 void vget_finish_ref(struct vnode *vp, enum vgetstate vs); 723 void vget_abort(struct vnode *vp, enum vgetstate vs); 724 void vgone(struct vnode *vp); 725 void vhold(struct vnode *); 726 void vholdnz(struct vnode *); 727 bool vhold_smr(struct vnode *); 728 int vinactive(struct vnode *vp); 729 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo); 730 int vtruncbuf(struct vnode *vp, off_t length, int blksize); 731 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn, 732 int blksize); 733 void vunref(struct vnode *); 734 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3); 735 int vrecycle(struct vnode *vp); 736 int vrecyclel(struct vnode *vp); 737 int vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off, 738 struct ucred *cred); 739 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off, 740 struct ucred *cred); 741 int vn_close(struct vnode *vp, 742 int flags, struct ucred *file_cred, struct thread *td); 743 int vn_copy_file_range(struct vnode *invp, off_t *inoffp, 744 struct vnode *outvp, off_t *outoffp, size_t *lenp, 745 unsigned int flags, struct ucred *incred, struct ucred *outcred, 746 struct thread *fsize_td); 747 int vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags, 748 int ioflg, struct ucred *active_cred, struct ucred *file_cred); 749 void vn_finished_write(struct mount *mp); 750 void vn_finished_secondary_write(struct mount *mp); 751 int vn_fsync_buf(struct vnode *vp, int waitfor); 752 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp, 753 struct vnode *outvp, off_t *outoffp, size_t *lenp, 754 unsigned int flags, struct ucred *incred, struct ucred *outcred, 755 struct thread *fsize_td); 756 int vn_need_pageq_flush(struct vnode *vp); 757 bool vn_isdisk_error(struct vnode *vp, int *errp); 758 bool vn_isdisk(struct vnode *vp); 759 int _vn_lock(struct vnode *vp, int flags, const char *file, int line); 760 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__) 761 void vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1, 762 struct vnode *vp2, bool vp2_locked, int lkflags2); 763 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp); 764 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode, 765 u_int vn_open_flags, struct ucred *cred, struct file *fp); 766 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred, 767 struct thread *td, struct file *fp); 768 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end); 769 void vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start, 770 vm_pindex_t end); 771 int vn_pollrecord(struct vnode *vp, struct thread *p, int events); 772 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, 773 int len, off_t offset, enum uio_seg segflg, int ioflg, 774 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid, 775 struct thread *td); 776 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base, 777 size_t len, off_t offset, enum uio_seg segflg, int ioflg, 778 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid, 779 struct thread *td); 780 int vn_read_from_obj(struct vnode *vp, struct uio *uio); 781 int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio, 782 struct thread *td); 783 int vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio, 784 off_t maxfsz, ssize_t *resid_adj, struct thread *td); 785 void vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj); 786 int vn_rlimit_trunc(u_quad_t size, 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 struct vop_stat_args *_ap = (ap); \ 970 int _error; \ 971 AUDIT_ARG_VNODE1(ap->a_vp); \ 972 _error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\ 973 if (__predict_true(_error == 0)) { \ 974 ap->a_sb->st_padding0 = 0; \ 975 ap->a_sb->st_padding1 = 0; \ 976 bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare)); \ 977 } \ 978 _error; \ 979 }) 980 981 #define vop_stat_helper_post(ap, error) ({ \ 982 struct vop_stat_args *_ap = (ap); \ 983 int _error = (error); \ 984 if (priv_check_cred_vfs_generation(_ap->a_active_cred)) \ 985 _ap->a_sb->st_gen = 0; \ 986 _error; \ 987 }) 988 989 #define VOP_WRITE_PRE(ap) \ 990 struct vattr va; \ 991 int error; \ 992 off_t osize, ooffset, noffset; \ 993 \ 994 osize = ooffset = noffset = 0; \ 995 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \ 996 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \ 997 if (error) \ 998 return (error); \ 999 ooffset = (ap)->a_uio->uio_offset; \ 1000 osize = (off_t)va.va_size; \ 1001 } 1002 1003 #define VOP_WRITE_POST(ap, ret) \ 1004 noffset = (ap)->a_uio->uio_offset; \ 1005 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \ 1006 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \ 1007 | (noffset > osize ? NOTE_EXTEND : 0)); \ 1008 } 1009 1010 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__) 1011 1012 #ifdef INVARIANTS 1013 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \ 1014 do { \ 1015 int error_; \ 1016 \ 1017 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \ 1018 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \ 1019 error_)); \ 1020 } while (0) 1021 #define VOP_SET_TEXT_CHECKED(vp) \ 1022 do { \ 1023 int error_; \ 1024 \ 1025 error_ = VOP_SET_TEXT((vp)); \ 1026 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \ 1027 error_)); \ 1028 } while (0) 1029 #define VOP_UNSET_TEXT_CHECKED(vp) \ 1030 do { \ 1031 int error_; \ 1032 \ 1033 error_ = VOP_UNSET_TEXT((vp)); \ 1034 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \ 1035 error_)); \ 1036 } while (0) 1037 #else 1038 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt)) 1039 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp)) 1040 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp)) 1041 #endif 1042 1043 #define VN_IS_DOOMED(vp) __predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0) 1044 1045 void vput(struct vnode *vp); 1046 void vrele(struct vnode *vp); 1047 void vref(struct vnode *vp); 1048 void vrefact(struct vnode *vp); 1049 void v_addpollinfo(struct vnode *vp); 1050 static __inline int 1051 vrefcnt(struct vnode *vp) 1052 { 1053 1054 return (vp->v_usecount); 1055 } 1056 1057 #define vholdl(vp) do { \ 1058 ASSERT_VI_LOCKED(vp, __func__); \ 1059 vhold(vp); \ 1060 } while (0) 1061 1062 #define vrefl(vp) do { \ 1063 ASSERT_VI_LOCKED(vp, __func__); \ 1064 vref(vp); \ 1065 } while (0) 1066 1067 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td); 1068 void vnode_destroy_vobject(struct vnode *vp); 1069 1070 extern struct vop_vector fifo_specops; 1071 extern struct vop_vector dead_vnodeops; 1072 extern struct vop_vector default_vnodeops; 1073 1074 #define VOP_PANIC ((void*)(uintptr_t)vop_panic) 1075 #define VOP_NULL ((void*)(uintptr_t)vop_null) 1076 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf) 1077 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty) 1078 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval) 1079 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent) 1080 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp) 1081 #define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain) 1082 1083 /* fifo_vnops.c */ 1084 int fifo_printinfo(struct vnode *); 1085 1086 /* vfs_hash.c */ 1087 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg); 1088 1089 void vfs_hash_changesize(u_long newhashsize); 1090 int vfs_hash_get(const struct mount *mp, u_int hash, int flags, 1091 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1092 u_int vfs_hash_index(struct vnode *vp); 1093 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, 1094 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1095 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td, 1096 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1097 void vfs_hash_rehash(struct vnode *vp, u_int hash); 1098 void vfs_hash_remove(struct vnode *vp); 1099 1100 int vfs_kqfilter(struct vop_kqfilter_args *); 1101 struct dirent; 1102 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off); 1103 int vfs_emptydir(struct vnode *vp); 1104 1105 int vfs_unixify_accmode(accmode_t *accmode); 1106 1107 void vfs_unp_reclaim(struct vnode *vp); 1108 1109 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode); 1110 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid, 1111 gid_t gid); 1112 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, 1113 struct thread *td); 1114 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred, 1115 struct thread *td); 1116 int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred); 1117 int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred); 1118 1119 void vn_fsid(struct vnode *vp, struct vattr *va); 1120 1121 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp); 1122 int vn_lktype_write(struct mount *mp, struct vnode *vp); 1123 1124 #ifdef INVARIANTS 1125 void vn_set_state_validate(struct vnode *vp, enum vstate state); 1126 #endif 1127 1128 static inline void 1129 vn_set_state(struct vnode *vp, enum vstate state) 1130 { 1131 #ifdef INVARIANTS 1132 vn_set_state_validate(vp, state); 1133 #endif 1134 vp->v_state = state; 1135 } 1136 1137 static inline enum vstate 1138 vn_get_state(struct vnode *vp) 1139 { 1140 return (vp->v_state); 1141 } 1142 1143 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \ 1144 struct vnode *_vp = (vp); \ 1145 int _flags = (flags); \ 1146 int _error; \ 1147 \ 1148 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \ 1149 panic("%s: unsupported flags %x\n", __func__, flags); \ 1150 _error = VOP_UNLOCK(_vp); \ 1151 if (_flags & LK_INTERLOCK) \ 1152 VI_UNLOCK(_vp); \ 1153 _error; \ 1154 }) 1155 1156 #include <sys/kernel.h> 1157 1158 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \ 1159 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \ 1160 vfs_vector_op_register, &vnodeops) 1161 1162 #define VFS_SMR_DECLARE \ 1163 extern smr_t vfs_smr 1164 1165 #define VFS_SMR() vfs_smr 1166 #define vfs_smr_enter() smr_enter(VFS_SMR()) 1167 #define vfs_smr_exit() smr_exit(VFS_SMR()) 1168 #define vfs_smr_synchronize() smr_synchronize(VFS_SMR()) 1169 #define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR()) 1170 #define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR()) 1171 #define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR()) 1172 #define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR()) 1173 1174 #define vn_load_v_data_smr(vp) ({ \ 1175 struct vnode *_vp = (vp); \ 1176 \ 1177 VFS_SMR_ASSERT_ENTERED(); \ 1178 atomic_load_consume_ptr(&(_vp)->v_data);\ 1179 }) 1180 1181 #endif /* _KERNEL */ 1182 1183 #endif /* !_SYS_VNODE_H_ */ 1184