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 /* 203 * Userland version of struct vnode, for sysctl. 204 */ 205 struct xvnode { 206 size_t xv_size; /* sizeof(struct xvnode) */ 207 void *xv_vnode; /* address of real vnode */ 208 u_long xv_flag; /* vnode vflags */ 209 int xv_usecount; /* reference count of users */ 210 int xv_writecount; /* reference count of writers */ 211 int xv_holdcnt; /* page & buffer references */ 212 u_long xv_id; /* capability identifier */ 213 void *xv_mount; /* address of parent mount */ 214 long xv_numoutput; /* num of writes in progress */ 215 enum vtype xv_type; /* vnode type */ 216 union { 217 void *xvu_socket; /* unpcb, if VSOCK */ 218 void *xvu_fifo; /* fifo, if VFIFO */ 219 dev_t xvu_rdev; /* maj/min, if VBLK/VCHR */ 220 struct { 221 dev_t xvu_dev; /* device, if VDIR/VREG/VLNK */ 222 ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */ 223 } xv_uns; 224 } xv_un; 225 }; 226 #define xv_socket xv_un.xvu_socket 227 #define xv_fifo xv_un.xvu_fifo 228 #define xv_rdev xv_un.xvu_rdev 229 #define xv_dev xv_un.xv_uns.xvu_dev 230 #define xv_ino xv_un.xv_uns.xvu_ino 231 232 /* We don't need to lock the knlist */ 233 #define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \ 234 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note)) 235 236 #define VN_KNOTE(vp, b, a) \ 237 do { \ 238 if (!VN_KNLIST_EMPTY(vp)) \ 239 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \ 240 (a) | KNF_NOKQLOCK); \ 241 } while (0) 242 #define VN_KNOTE_LOCKED(vp, b) VN_KNOTE(vp, b, KNF_LISTLOCKED) 243 #define VN_KNOTE_UNLOCKED(vp, b) VN_KNOTE(vp, b, 0) 244 245 /* 246 * Vnode flags. 247 * VI flags are protected by interlock and live in v_iflag 248 * VV flags are protected by the vnode lock and live in v_vflag 249 * 250 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both 251 * are required for writing but the status may be checked with either. 252 */ 253 #define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */ 254 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR) 255 256 #define VIRF_DOOMED 0x0001 /* This vnode is being recycled */ 257 #define VIRF_PGREAD 0x0002 /* Direct reads from the page cache are permitted, 258 never cleared once set */ 259 #define VIRF_MOUNTPOINT 0x0004 /* This vnode is mounted on */ 260 #define VIRF_TEXT_REF 0x0008 /* Executable mappings ref the vnode */ 261 262 #define VI_UNUSED0 0x0001 /* unused */ 263 #define VI_MOUNT 0x0002 /* Mount in progress */ 264 #define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */ 265 #define VI_OWEINACT 0x0008 /* Need to call inactive */ 266 #define VI_DEFINACT 0x0010 /* deferred inactive */ 267 #define VI_FOPENING 0x0020 /* In open, with opening process having the 268 first right to advlock file */ 269 270 #define VV_ROOT 0x0001 /* root of its filesystem */ 271 #define VV_ISTTY 0x0002 /* vnode represents a tty */ 272 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */ 273 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */ 274 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */ 275 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */ 276 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */ 277 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */ 278 #define VV_PROCDEP 0x0100 /* vnode is process dependent */ 279 #define VV_UNLINKED 0x0200 /* unlinked but stil open directory */ 280 #define VV_DELETED 0x0400 /* should be removed */ 281 #define VV_MD 0x0800 /* vnode backs the md device */ 282 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */ 283 #define VV_READLINK 0x2000 /* fdescfs linux vnode */ 284 #define VV_UNREF 0x4000 /* vunref, do not drop lock in inactive() */ 285 #define VV_CROSSLOCK 0x8000 /* vnode lock is shared w/ root mounted here */ 286 287 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */ 288 289 /* 290 * Vnode attributes. A field value of VNOVAL represents a field whose value 291 * is unavailable (getattr) or which is not to be changed (setattr). 292 */ 293 struct vattr { 294 enum vtype va_type; /* vnode type (for create) */ 295 u_short va_mode; /* files access mode and type */ 296 u_short va_padding0; 297 uid_t va_uid; /* owner user id */ 298 gid_t va_gid; /* owner group id */ 299 nlink_t va_nlink; /* number of references to file */ 300 dev_t va_fsid; /* filesystem id */ 301 ino_t va_fileid; /* file id */ 302 u_quad_t va_size; /* file size in bytes */ 303 long va_blocksize; /* blocksize preferred for i/o */ 304 struct timespec va_atime; /* time of last access */ 305 struct timespec va_mtime; /* time of last modification */ 306 struct timespec va_ctime; /* time file changed */ 307 struct timespec va_birthtime; /* time file created */ 308 u_long va_gen; /* generation number of file */ 309 u_long va_flags; /* flags defined for file */ 310 dev_t va_rdev; /* device the special file represents */ 311 u_quad_t va_bytes; /* bytes of disk space held by file */ 312 u_quad_t va_filerev; /* file modification number */ 313 u_int va_vaflags; /* operations flags, see below */ 314 long va_spare; /* remain quad aligned */ 315 }; 316 317 /* 318 * Flags for va_vaflags. 319 */ 320 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */ 321 #define VA_EXCLUSIVE 0x02 /* exclusive create request */ 322 #define VA_SYNC 0x04 /* O_SYNC truncation */ 323 324 /* 325 * Flags for ioflag. (high 16 bits used to ask for read-ahead and 326 * help with write clustering) 327 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h 328 */ 329 #define IO_UNIT 0x0001 /* do I/O as atomic unit */ 330 #define IO_APPEND 0x0002 /* append write to end */ 331 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */ 332 #define IO_NODELOCKED 0x0008 /* underlying node already locked */ 333 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */ 334 #define IO_VMIO 0x0020 /* data already in VMIO space */ 335 #define IO_INVAL 0x0040 /* invalidate after I/O */ 336 #define IO_SYNC 0x0080 /* do I/O synchronously */ 337 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */ 338 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */ 339 #define IO_EXT 0x0400 /* operate on external attributes */ 340 #define IO_NORMAL 0x0800 /* operate on regular data */ 341 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */ 342 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */ 343 #define IO_RANGELOCKED 0x4000 /* range locked */ 344 #define IO_DATASYNC 0x8000 /* do only data I/O synchronously */ 345 346 #define IO_SEQMAX 0x7F /* seq heuristic max value */ 347 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */ 348 349 /* 350 * Flags for accmode_t. 351 */ 352 #define VEXEC 000000000100 /* execute/search permission */ 353 #define VWRITE 000000000200 /* write permission */ 354 #define VREAD 000000000400 /* read permission */ 355 #define VADMIN 000000010000 /* being the file owner */ 356 #define VAPPEND 000000040000 /* permission to write/append */ 357 /* 358 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only 359 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL, 360 * and 0 otherwise. This never happens with ordinary unix access rights 361 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with 362 * some other V* constant. 363 */ 364 #define VEXPLICIT_DENY 000000100000 365 #define VREAD_NAMED_ATTRS 000000200000 /* not used */ 366 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */ 367 #define VDELETE_CHILD 000001000000 368 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */ 369 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */ 370 #define VDELETE 000010000000 371 #define VREAD_ACL 000020000000 /* read ACL and file mode */ 372 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */ 373 #define VWRITE_OWNER 000100000000 /* change file owner */ 374 #define VSYNCHRONIZE 000200000000 /* not used */ 375 #define VCREAT 000400000000 /* creating new file */ 376 #define VVERIFY 001000000000 /* verification required */ 377 378 /* 379 * Permissions that were traditionally granted only to the file owner. 380 */ 381 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \ 382 VWRITE_OWNER) 383 384 /* 385 * Permissions that were traditionally granted to everyone. 386 */ 387 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL) 388 389 /* 390 * Permissions that allow to change the state of the file in any way. 391 */ 392 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \ 393 VDELETE) 394 395 /* 396 * Token indicating no attribute value yet assigned. 397 */ 398 #define VNOVAL (-1) 399 400 /* 401 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon) 402 */ 403 #define VLKTIMEOUT (hz / 20 + 1) 404 405 #ifdef _KERNEL 406 407 #ifdef MALLOC_DECLARE 408 MALLOC_DECLARE(M_VNODE); 409 #endif 410 411 extern u_int ncsizefactor; 412 extern const u_int io_hold_cnt; 413 414 /* 415 * Convert between vnode types and inode formats (since POSIX.1 416 * defines mode word of stat structure in terms of inode formats). 417 */ 418 extern enum vtype iftovt_tab[]; 419 extern int vttoif_tab[]; 420 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12]) 421 #define VTTOIF(indx) (vttoif_tab[(int)(indx)]) 422 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode)) 423 424 /* 425 * Flags to various vnode functions. 426 */ 427 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */ 428 #define FORCECLOSE 0x0002 /* vflush: force file closure */ 429 #define WRITECLOSE 0x0004 /* vflush: only close writable files */ 430 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */ 431 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */ 432 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */ 433 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */ 434 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */ 435 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */ 436 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */ 437 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */ 438 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */ 439 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */ 440 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */ 441 #define V_PCATCH 0x0008 /* vn_start_write: make the sleep interruptible */ 442 #define V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH) 443 444 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */ 445 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */ 446 447 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the 448 filesystem is being unmounted */ 449 450 #define VREF(vp) vref(vp) 451 452 #ifdef DIAGNOSTIC 453 #define VATTR_NULL(vap) vattr_null(vap) 454 #else 455 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */ 456 #endif /* DIAGNOSTIC */ 457 458 #define NULLVP ((struct vnode *)NULL) 459 460 /* 461 * Global vnode data. 462 */ 463 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */ 464 extern struct mount *rootdevmp; /* "/dev" mount */ 465 extern u_long desiredvnodes; /* number of vnodes desired */ 466 extern struct uma_zone *namei_zone; 467 extern struct vattr va_null; /* predefined null vattr structure */ 468 469 extern u_int vn_lock_pair_pause_max; 470 471 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock) 472 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags)) 473 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock) 474 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock) 475 #define VI_MTX(vp) (&(vp)->v_interlock) 476 477 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock) 478 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock) 479 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock) 480 481 #endif /* _KERNEL */ 482 483 /* 484 * Mods for extensibility. 485 */ 486 487 /* 488 * Flags for vdesc_flags: 489 */ 490 #define VDESC_MAX_VPS 16 491 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */ 492 #define VDESC_VP0_WILLRELE 0x0001 493 #define VDESC_VP1_WILLRELE 0x0002 494 #define VDESC_VP2_WILLRELE 0x0004 495 #define VDESC_VP3_WILLRELE 0x0008 496 497 /* 498 * A generic structure. 499 * This can be used by bypass routines to identify generic arguments. 500 */ 501 struct vop_generic_args { 502 struct vnodeop_desc *a_desc; 503 /* other random data follows, presumably */ 504 }; 505 506 typedef int vop_bypass_t(struct vop_generic_args *); 507 508 /* 509 * VDESC_NO_OFFSET is used to identify the end of the offset list 510 * and in places where no such field exists. 511 */ 512 #define VDESC_NO_OFFSET -1 513 514 /* 515 * This structure describes the vnode operation taking place. 516 */ 517 struct vnodeop_desc { 518 char *vdesc_name; /* a readable name for debugging */ 519 int vdesc_flags; /* VDESC_* flags */ 520 int vdesc_vop_offset; 521 vop_bypass_t *vdesc_call; /* Function to call */ 522 523 /* 524 * These ops are used by bypass routines to map and locate arguments. 525 * Creds and procs are not needed in bypass routines, but sometimes 526 * they are useful to (for example) transport layers. 527 * Nameidata is useful because it has a cred in it. 528 */ 529 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */ 530 int vdesc_vpp_offset; /* return vpp location */ 531 int vdesc_cred_offset; /* cred location, if any */ 532 int vdesc_thread_offset; /* thread location, if any */ 533 int vdesc_componentname_offset; /* if any */ 534 }; 535 536 #ifdef _KERNEL 537 /* 538 * A list of all the operation descs. 539 */ 540 extern struct vnodeop_desc *vnodeop_descs[]; 541 542 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field) 543 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \ 544 ((s_type)(((char*)(struct_p)) + (s_offset))) 545 546 #ifdef DEBUG_VFS_LOCKS 547 /* 548 * Support code to aid in debugging VFS locking problems. Not totally 549 * reliable since if the thread sleeps between changing the lock 550 * state and checking it with the assert, some other thread could 551 * change the state. They are good enough for debugging a single 552 * filesystem using a single-threaded test. Note that the unreliability is 553 * limited to false negatives; efforts were made to ensure that false 554 * positives cannot occur. 555 */ 556 void assert_vi_locked(struct vnode *vp, const char *str); 557 void assert_vi_unlocked(struct vnode *vp, const char *str); 558 void assert_vop_elocked(struct vnode *vp, const char *str); 559 void assert_vop_locked(struct vnode *vp, const char *str); 560 void assert_vop_unlocked(struct vnode *vp, const char *str); 561 562 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str)) 563 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str)) 564 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str)) 565 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str)) 566 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str)) 567 568 #define ASSERT_VOP_IN_SEQC(vp) do { \ 569 struct vnode *_vp = (vp); \ 570 \ 571 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \ 572 } while (0) 573 574 #define ASSERT_VOP_NOT_IN_SEQC(vp) do { \ 575 struct vnode *_vp = (vp); \ 576 \ 577 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \ 578 } while (0) 579 580 #else /* !DEBUG_VFS_LOCKS */ 581 582 #define ASSERT_VI_LOCKED(vp, str) ((void)0) 583 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0) 584 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0) 585 #define ASSERT_VOP_LOCKED(vp, str) ((void)0) 586 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0) 587 588 #define ASSERT_VOP_IN_SEQC(vp) ((void)0) 589 #define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0) 590 591 #endif /* DEBUG_VFS_LOCKS */ 592 593 /* 594 * This call works for vnodes in the kernel. 595 */ 596 #define VCALL(c) ((c)->a_desc->vdesc_call(c)) 597 598 #define DOINGASYNC(vp) \ 599 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \ 600 ((curthread->td_pflags & TDP_SYNCIO) == 0)) 601 602 /* 603 * VMIO support inline 604 */ 605 606 extern int vmiodirenable; 607 608 static __inline int 609 vn_canvmio(struct vnode *vp) 610 { 611 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR))) 612 return(TRUE); 613 return(FALSE); 614 } 615 616 /* 617 * Finally, include the default set of vnode operations. 618 */ 619 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int); 620 #include "vnode_if.h" 621 622 /* vn_open_flags */ 623 #define VN_OPEN_NOAUDIT 0x00000001 624 #define VN_OPEN_NOCAPCHECK 0x00000002 625 #define VN_OPEN_NAMECACHE 0x00000004 626 #define VN_OPEN_INVFS 0x00000008 627 #define VN_OPEN_WANTIOCTLCAPS 0x00000010 628 629 /* copy_file_range kernel flags */ 630 #define COPY_FILE_RANGE_KFLAGS 0xff000000 631 #define COPY_FILE_RANGE_TIMEO1SEC 0x01000000 /* Return after 1sec. */ 632 633 /* 634 * Public vnode manipulation functions. 635 */ 636 struct componentname; 637 struct file; 638 struct mount; 639 struct nameidata; 640 struct ostat; 641 struct freebsd11_stat; 642 struct thread; 643 struct proc; 644 struct stat; 645 struct nstat; 646 struct ucred; 647 struct uio; 648 struct vattr; 649 struct vfsops; 650 struct vnode; 651 652 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **); 653 654 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn, 655 daddr_t endn); 656 /* cache_* may belong in namei.h. */ 657 void cache_changesize(u_long newhashsize); 658 659 #define VFS_CACHE_DROPOLD 0x1 660 661 void cache_enter_time_flags(struct vnode *dvp, struct vnode *vp, 662 struct componentname *cnp, struct timespec *tsp, 663 struct timespec *dtsp, int flags); 664 #define cache_enter(dvp, vp, cnp) \ 665 cache_enter_time(dvp, vp, cnp, NULL, NULL) 666 void cache_enter_time(struct vnode *dvp, struct vnode *vp, 667 struct componentname *cnp, struct timespec *tsp, 668 struct timespec *dtsp); 669 int cache_lookup(struct vnode *dvp, struct vnode **vpp, 670 struct componentname *cnp, struct timespec *tsp, int *ticksp); 671 void cache_vnode_init(struct vnode *vp); 672 void cache_purge(struct vnode *vp); 673 void cache_purge_vgone(struct vnode *vp); 674 void cache_purge_negative(struct vnode *vp); 675 void cache_purgevfs(struct mount *mp); 676 char *cache_symlink_alloc(size_t size, int flags); 677 void cache_symlink_free(char *string, size_t size); 678 int cache_symlink_resolve(struct cache_fpl *fpl, const char *string, 679 size_t len); 680 void cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp, 681 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp); 682 void cache_vop_rmdir(struct vnode *dvp, struct vnode *vp); 683 #ifdef INVARIANTS 684 void cache_validate(struct vnode *dvp, struct vnode *vp, 685 struct componentname *cnp); 686 void cache_assert_no_entries(struct vnode *vp); 687 #else 688 static inline void 689 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp) 690 { 691 } 692 693 static inline void 694 cache_assert_no_entries(struct vnode *vp) 695 { 696 } 697 #endif 698 void cache_fast_lookup_enabled_recalc(void); 699 int change_dir(struct vnode *vp, struct thread *td); 700 void cvtstat(struct stat *st, struct ostat *ost); 701 int freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb); 702 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost); 703 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops, 704 struct vnode **vpp); 705 void getnewvnode_reserve(void); 706 void getnewvnode_drop_reserve(void); 707 int insmntque(struct vnode *vp, struct mount *mp); 708 int insmntque1(struct vnode *vp, struct mount *mp); 709 u_quad_t init_va_filerev(void); 710 int speedup_syncer(void); 711 int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen); 712 int vn_getcwd(char *buf, char **retbuf, size_t *buflen); 713 int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf); 714 int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf); 715 int vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp, 716 const char *hdrl_name, size_t hrdl_name_length, char **retbuf, 717 char **freebuf, size_t *buflen); 718 struct vnode * 719 vn_dir_dd_ino(struct vnode *vp); 720 int vn_commname(struct vnode *vn, char *buf, u_int buflen); 721 int vn_path_to_global_path(struct thread *td, struct vnode *vp, 722 char *path, u_int pathlen); 723 int vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp, 724 struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name, 725 size_t leaf_length); 726 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid, 727 gid_t file_gid, accmode_t accmode, struct ucred *cred); 728 int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid, 729 struct ucred *cred); 730 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid, 731 struct acl *aclp, accmode_t accmode, struct ucred *cred); 732 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid, 733 gid_t file_gid, struct acl *acl, accmode_t accmode, 734 struct ucred *cred); 735 void vattr_null(struct vattr *vap); 736 void vlazy(struct vnode *); 737 void vdrop(struct vnode *); 738 void vdropl(struct vnode *); 739 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td); 740 int vget(struct vnode *vp, int flags); 741 enum vgetstate vget_prep_smr(struct vnode *vp); 742 enum vgetstate vget_prep(struct vnode *vp); 743 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs); 744 void vget_finish_ref(struct vnode *vp, enum vgetstate vs); 745 void vget_abort(struct vnode *vp, enum vgetstate vs); 746 void vgone(struct vnode *vp); 747 void vhold(struct vnode *); 748 void vholdnz(struct vnode *); 749 bool vhold_smr(struct vnode *); 750 int vinactive(struct vnode *vp); 751 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo); 752 int vtruncbuf(struct vnode *vp, off_t length, int blksize); 753 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn, 754 int blksize); 755 void vunref(struct vnode *); 756 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3); 757 int vrecycle(struct vnode *vp); 758 int vrecyclel(struct vnode *vp); 759 int vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off, 760 struct ucred *cred); 761 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off, 762 struct ucred *cred); 763 int vn_close(struct vnode *vp, 764 int flags, struct ucred *file_cred, struct thread *td); 765 int vn_copy_file_range(struct vnode *invp, off_t *inoffp, 766 struct vnode *outvp, off_t *outoffp, size_t *lenp, 767 unsigned int flags, struct ucred *incred, struct ucred *outcred, 768 struct thread *fsize_td); 769 int vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags, 770 int ioflg, struct ucred *active_cred, struct ucred *file_cred); 771 void vn_finished_write(struct mount *mp); 772 void vn_finished_secondary_write(struct mount *mp); 773 int vn_fsync_buf(struct vnode *vp, int waitfor); 774 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp, 775 struct vnode *outvp, off_t *outoffp, size_t *lenp, 776 unsigned int flags, struct ucred *incred, struct ucred *outcred, 777 struct thread *fsize_td); 778 int vn_need_pageq_flush(struct vnode *vp); 779 bool vn_isdisk_error(struct vnode *vp, int *errp); 780 bool vn_isdisk(struct vnode *vp); 781 int _vn_lock(struct vnode *vp, int flags, const char *file, int line); 782 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__) 783 void vn_lock_pair(struct vnode *vp1, bool vp1_locked, struct vnode *vp2, 784 bool vp2_locked); 785 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp); 786 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode, 787 u_int vn_open_flags, struct ucred *cred, struct file *fp); 788 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred, 789 struct thread *td, struct file *fp); 790 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end); 791 void vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start, 792 vm_pindex_t end); 793 int vn_pollrecord(struct vnode *vp, struct thread *p, int events); 794 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, 795 int len, off_t offset, enum uio_seg segflg, int ioflg, 796 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid, 797 struct thread *td); 798 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base, 799 size_t len, off_t offset, enum uio_seg segflg, int ioflg, 800 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid, 801 struct thread *td); 802 int vn_read_from_obj(struct vnode *vp, struct uio *uio); 803 int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio, 804 struct thread *td); 805 int vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio, 806 off_t maxfsz, ssize_t *resid_adj, struct thread *td); 807 void vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj); 808 int vn_rlimit_trunc(u_quad_t size, struct thread *td); 809 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags); 810 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp, 811 int flags); 812 int vn_truncate_locked(struct vnode *vp, off_t length, bool sync, 813 struct ucred *cred); 814 int vn_writechk(struct vnode *vp); 815 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 816 const char *attrname, int *buflen, char *buf, struct thread *td); 817 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 818 const char *attrname, int buflen, char *buf, struct thread *td); 819 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 820 const char *attrname, struct thread *td); 821 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags, 822 struct vnode **rvp); 823 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc, 824 void *alloc_arg, int lkflags, struct vnode **rvp); 825 int vn_utimes_perm(struct vnode *vp, struct vattr *vap, 826 struct ucred *cred, struct thread *td); 827 828 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio); 829 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize, 830 struct uio *uio); 831 832 void vn_seqc_write_begin_locked(struct vnode *vp); 833 void vn_seqc_write_begin(struct vnode *vp); 834 void vn_seqc_write_end_locked(struct vnode *vp); 835 void vn_seqc_write_end(struct vnode *vp); 836 #define vn_seqc_read_any(vp) seqc_read_any(&(vp)->v_seqc) 837 #define vn_seqc_read_notmodify(vp) seqc_read_notmodify(&(vp)->v_seqc) 838 #define vn_seqc_consistent(vp, seq) seqc_consistent(&(vp)->v_seqc, seq) 839 840 #define vn_rangelock_unlock(vp, cookie) \ 841 rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp)) 842 #define vn_rangelock_unlock_range(vp, cookie, start, end) \ 843 rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), \ 844 VI_MTX(vp)) 845 #define vn_rangelock_rlock(vp, start, end) \ 846 rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 847 #define vn_rangelock_tryrlock(vp, start, end) \ 848 rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 849 #define vn_rangelock_wlock(vp, start, end) \ 850 rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 851 #define vn_rangelock_trywlock(vp, start, end) \ 852 rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp)) 853 854 #define vn_irflag_read(vp) atomic_load_short(&(vp)->v_irflag) 855 void vn_irflag_set_locked(struct vnode *vp, short toset); 856 void vn_irflag_set(struct vnode *vp, short toset); 857 void vn_irflag_set_cond_locked(struct vnode *vp, short toset); 858 void vn_irflag_set_cond(struct vnode *vp, short toset); 859 void vn_irflag_unset_locked(struct vnode *vp, short tounset); 860 void vn_irflag_unset(struct vnode *vp, short tounset); 861 862 int vfs_cache_lookup(struct vop_lookup_args *ap); 863 int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp); 864 void vfs_timestamp(struct timespec *); 865 void vfs_write_resume(struct mount *mp, int flags); 866 int vfs_write_suspend(struct mount *mp, int flags); 867 int vfs_write_suspend_umnt(struct mount *mp); 868 struct vnode *vnlru_alloc_marker(void); 869 void vnlru_free_marker(struct vnode *); 870 void vnlru_free_vfsops(int, struct vfsops *, struct vnode *); 871 int vop_stdbmap(struct vop_bmap_args *); 872 int vop_stdfdatasync_buf(struct vop_fdatasync_args *); 873 int vop_stdfsync(struct vop_fsync_args *); 874 int vop_stdgetwritemount(struct vop_getwritemount_args *); 875 int vop_stdgetpages(struct vop_getpages_args *); 876 int vop_stdinactive(struct vop_inactive_args *); 877 int vop_stdioctl(struct vop_ioctl_args *); 878 int vop_stdneed_inactive(struct vop_need_inactive_args *); 879 int vop_stdkqfilter(struct vop_kqfilter_args *); 880 int vop_stdlock(struct vop_lock1_args *); 881 int vop_stdunlock(struct vop_unlock_args *); 882 int vop_stdislocked(struct vop_islocked_args *); 883 int vop_lock(struct vop_lock1_args *); 884 int vop_unlock(struct vop_unlock_args *); 885 int vop_islocked(struct vop_islocked_args *); 886 int vop_stdputpages(struct vop_putpages_args *); 887 int vop_nopoll(struct vop_poll_args *); 888 int vop_stdaccess(struct vop_access_args *ap); 889 int vop_stdaccessx(struct vop_accessx_args *ap); 890 int vop_stdadvise(struct vop_advise_args *ap); 891 int vop_stdadvlock(struct vop_advlock_args *ap); 892 int vop_stdadvlockasync(struct vop_advlockasync_args *ap); 893 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap); 894 int vop_stdallocate(struct vop_allocate_args *ap); 895 int vop_stddeallocate(struct vop_deallocate_args *ap); 896 int vop_stdset_text(struct vop_set_text_args *ap); 897 int vop_stdpathconf(struct vop_pathconf_args *); 898 int vop_stdpoll(struct vop_poll_args *); 899 int vop_stdvptocnp(struct vop_vptocnp_args *ap); 900 int vop_stdvptofh(struct vop_vptofh_args *ap); 901 int vop_stdunp_bind(struct vop_unp_bind_args *ap); 902 int vop_stdunp_connect(struct vop_unp_connect_args *ap); 903 int vop_stdunp_detach(struct vop_unp_detach_args *ap); 904 int vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap); 905 int vop_eopnotsupp(struct vop_generic_args *ap); 906 int vop_ebadf(struct vop_generic_args *ap); 907 int vop_einval(struct vop_generic_args *ap); 908 int vop_enoent(struct vop_generic_args *ap); 909 int vop_enotty(struct vop_generic_args *ap); 910 int vop_eagain(struct vop_generic_args *ap); 911 int vop_null(struct vop_generic_args *ap); 912 int vop_panic(struct vop_generic_args *ap); 913 int dead_poll(struct vop_poll_args *ap); 914 int dead_read(struct vop_read_args *ap); 915 int dead_write(struct vop_write_args *ap); 916 917 /* These are called from within the actual VOPS. */ 918 void vop_close_post(void *a, int rc); 919 void vop_create_pre(void *a); 920 void vop_create_post(void *a, int rc); 921 void vop_whiteout_pre(void *a); 922 void vop_whiteout_post(void *a, int rc); 923 void vop_deleteextattr_pre(void *a); 924 void vop_deleteextattr_post(void *a, int rc); 925 void vop_link_pre(void *a); 926 void vop_link_post(void *a, int rc); 927 void vop_lookup_post(void *a, int rc); 928 void vop_lookup_pre(void *a); 929 void vop_mkdir_pre(void *a); 930 void vop_mkdir_post(void *a, int rc); 931 void vop_mknod_pre(void *a); 932 void vop_mknod_post(void *a, int rc); 933 void vop_open_post(void *a, int rc); 934 void vop_read_post(void *a, int rc); 935 void vop_read_pgcache_post(void *ap, int rc); 936 void vop_readdir_post(void *a, int rc); 937 void vop_reclaim_post(void *a, int rc); 938 void vop_remove_pre(void *a); 939 void vop_remove_post(void *a, int rc); 940 void vop_rename_post(void *a, int rc); 941 void vop_rename_pre(void *a); 942 void vop_rmdir_pre(void *a); 943 void vop_rmdir_post(void *a, int rc); 944 void vop_setattr_pre(void *a); 945 void vop_setattr_post(void *a, int rc); 946 void vop_setacl_pre(void *a); 947 void vop_setacl_post(void *a, int rc); 948 void vop_setextattr_pre(void *a); 949 void vop_setextattr_post(void *a, int rc); 950 void vop_symlink_pre(void *a); 951 void vop_symlink_post(void *a, int rc); 952 int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a); 953 954 #ifdef DEBUG_VFS_LOCKS 955 void vop_fdatasync_debugpre(void *a); 956 void vop_fdatasync_debugpost(void *a, int rc); 957 void vop_fplookup_vexec_debugpre(void *a); 958 void vop_fplookup_vexec_debugpost(void *a, int rc); 959 void vop_fplookup_symlink_debugpre(void *a); 960 void vop_fplookup_symlink_debugpost(void *a, int rc); 961 void vop_fsync_debugpre(void *a); 962 void vop_fsync_debugpost(void *a, int rc); 963 void vop_strategy_debugpre(void *a); 964 void vop_lock_debugpre(void *a); 965 void vop_lock_debugpost(void *a, int rc); 966 void vop_unlock_debugpre(void *a); 967 void vop_need_inactive_debugpre(void *a); 968 void vop_need_inactive_debugpost(void *a, int rc); 969 void vop_mkdir_debugpost(void *a, int rc); 970 #else 971 #define vop_fdatasync_debugpre(x) do { } while (0) 972 #define vop_fdatasync_debugpost(x, y) do { } while (0) 973 #define vop_fplookup_vexec_debugpre(x) do { } while (0) 974 #define vop_fplookup_vexec_debugpost(x, y) do { } while (0) 975 #define vop_fplookup_symlink_debugpre(x) do { } while (0) 976 #define vop_fplookup_symlink_debugpost(x, y) do { } while (0) 977 #define vop_fsync_debugpre(x) do { } while (0) 978 #define vop_fsync_debugpost(x, y) do { } while (0) 979 #define vop_strategy_debugpre(x) do { } while (0) 980 #define vop_lock_debugpre(x) do { } while (0) 981 #define vop_lock_debugpost(x, y) do { } while (0) 982 #define vop_unlock_debugpre(x) do { } while (0) 983 #define vop_need_inactive_debugpre(x) do { } while (0) 984 #define vop_need_inactive_debugpost(x, y) do { } while (0) 985 #define vop_mkdir_debugpost(x, y) do { } while (0) 986 #endif 987 988 void vop_rename_fail(struct vop_rename_args *ap); 989 990 #define vop_stat_helper_pre(ap) ({ \ 991 struct vop_stat_args *_ap = (ap); \ 992 int _error; \ 993 AUDIT_ARG_VNODE1(ap->a_vp); \ 994 _error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\ 995 if (__predict_true(_error == 0)) { \ 996 ap->a_sb->st_padding0 = 0; \ 997 ap->a_sb->st_padding1 = 0; \ 998 bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare)); \ 999 } \ 1000 _error; \ 1001 }) 1002 1003 #define vop_stat_helper_post(ap, error) ({ \ 1004 struct vop_stat_args *_ap = (ap); \ 1005 int _error = (error); \ 1006 if (priv_check_cred_vfs_generation(_ap->a_active_cred)) \ 1007 _ap->a_sb->st_gen = 0; \ 1008 _error; \ 1009 }) 1010 1011 #define VOP_WRITE_PRE(ap) \ 1012 struct vattr va; \ 1013 int error; \ 1014 off_t osize, ooffset, noffset; \ 1015 \ 1016 osize = ooffset = noffset = 0; \ 1017 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \ 1018 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \ 1019 if (error) \ 1020 return (error); \ 1021 ooffset = (ap)->a_uio->uio_offset; \ 1022 osize = (off_t)va.va_size; \ 1023 } 1024 1025 #define VOP_WRITE_POST(ap, ret) \ 1026 noffset = (ap)->a_uio->uio_offset; \ 1027 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \ 1028 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \ 1029 | (noffset > osize ? NOTE_EXTEND : 0)); \ 1030 } 1031 1032 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__) 1033 1034 #ifdef INVARIANTS 1035 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \ 1036 do { \ 1037 int error_; \ 1038 \ 1039 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \ 1040 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \ 1041 error_)); \ 1042 } while (0) 1043 #define VOP_SET_TEXT_CHECKED(vp) \ 1044 do { \ 1045 int error_; \ 1046 \ 1047 error_ = VOP_SET_TEXT((vp)); \ 1048 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \ 1049 error_)); \ 1050 } while (0) 1051 #define VOP_UNSET_TEXT_CHECKED(vp) \ 1052 do { \ 1053 int error_; \ 1054 \ 1055 error_ = VOP_UNSET_TEXT((vp)); \ 1056 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \ 1057 error_)); \ 1058 } while (0) 1059 #else 1060 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt)) 1061 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp)) 1062 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp)) 1063 #endif 1064 1065 #define VN_IS_DOOMED(vp) __predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0) 1066 1067 void vput(struct vnode *vp); 1068 void vrele(struct vnode *vp); 1069 void vref(struct vnode *vp); 1070 void vrefact(struct vnode *vp); 1071 void v_addpollinfo(struct vnode *vp); 1072 static __inline int 1073 vrefcnt(struct vnode *vp) 1074 { 1075 1076 return (vp->v_usecount); 1077 } 1078 1079 #define vholdl(vp) do { \ 1080 ASSERT_VI_LOCKED(vp, __func__); \ 1081 vhold(vp); \ 1082 } while (0) 1083 1084 #define vrefl(vp) do { \ 1085 ASSERT_VI_LOCKED(vp, __func__); \ 1086 vref(vp); \ 1087 } while (0) 1088 1089 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td); 1090 void vnode_destroy_vobject(struct vnode *vp); 1091 1092 extern struct vop_vector fifo_specops; 1093 extern struct vop_vector dead_vnodeops; 1094 extern struct vop_vector default_vnodeops; 1095 1096 #define VOP_PANIC ((void*)(uintptr_t)vop_panic) 1097 #define VOP_NULL ((void*)(uintptr_t)vop_null) 1098 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf) 1099 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty) 1100 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval) 1101 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent) 1102 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp) 1103 #define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain) 1104 1105 /* fifo_vnops.c */ 1106 int fifo_printinfo(struct vnode *); 1107 1108 /* vfs_hash.c */ 1109 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg); 1110 1111 void vfs_hash_changesize(u_long newhashsize); 1112 int vfs_hash_get(const struct mount *mp, u_int hash, int flags, 1113 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1114 u_int vfs_hash_index(struct vnode *vp); 1115 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, 1116 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1117 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td, 1118 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg); 1119 void vfs_hash_rehash(struct vnode *vp, u_int hash); 1120 void vfs_hash_remove(struct vnode *vp); 1121 1122 int vfs_kqfilter(struct vop_kqfilter_args *); 1123 struct dirent; 1124 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off); 1125 int vfs_emptydir(struct vnode *vp); 1126 1127 int vfs_unixify_accmode(accmode_t *accmode); 1128 1129 void vfs_unp_reclaim(struct vnode *vp); 1130 1131 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode); 1132 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid, 1133 gid_t gid); 1134 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, 1135 struct thread *td); 1136 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred, 1137 struct thread *td); 1138 int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred); 1139 int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred); 1140 1141 void vn_fsid(struct vnode *vp, struct vattr *va); 1142 1143 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp); 1144 int vn_lktype_write(struct mount *mp, struct vnode *vp); 1145 1146 #ifdef INVARIANTS 1147 void vn_set_state_validate(struct vnode *vp, enum vstate state); 1148 #endif 1149 1150 static inline void 1151 vn_set_state(struct vnode *vp, enum vstate state) 1152 { 1153 #ifdef INVARIANTS 1154 vn_set_state_validate(vp, state); 1155 #endif 1156 vp->v_state = state; 1157 } 1158 1159 static inline enum vstate 1160 vn_get_state(struct vnode *vp) 1161 { 1162 return (vp->v_state); 1163 } 1164 1165 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \ 1166 struct vnode *_vp = (vp); \ 1167 int _flags = (flags); \ 1168 int _error; \ 1169 \ 1170 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \ 1171 panic("%s: unsupported flags %x\n", __func__, flags); \ 1172 _error = VOP_UNLOCK(_vp); \ 1173 if (_flags & LK_INTERLOCK) \ 1174 VI_UNLOCK(_vp); \ 1175 _error; \ 1176 }) 1177 1178 #include <sys/kernel.h> 1179 1180 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \ 1181 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \ 1182 vfs_vector_op_register, &vnodeops) 1183 1184 #define VFS_SMR_DECLARE \ 1185 extern smr_t vfs_smr 1186 1187 #define VFS_SMR() vfs_smr 1188 #define vfs_smr_enter() smr_enter(VFS_SMR()) 1189 #define vfs_smr_exit() smr_exit(VFS_SMR()) 1190 #define vfs_smr_synchronize() smr_synchronize(VFS_SMR()) 1191 #define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR()) 1192 #define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR()) 1193 #define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR()) 1194 #define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR()) 1195 1196 #define vn_load_v_data_smr(vp) ({ \ 1197 struct vnode *_vp = (vp); \ 1198 \ 1199 VFS_SMR_ASSERT_ENTERED(); \ 1200 atomic_load_consume_ptr(&(_vp)->v_data);\ 1201 }) 1202 1203 #endif /* _KERNEL */ 1204 1205 #endif /* !_SYS_VNODE_H_ */ 1206