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 */
  33
  34#ifndef _SYS_VNODE_H_
  35#define	_SYS_VNODE_H_
  36
  37#include <sys/bufobj.h>
  38#include <sys/queue.h>
  39#include <sys/lock.h>
  40#include <sys/lockmgr.h>
  41#include <sys/mutex.h>
  42#include <sys/rangelock.h>
  43#include <sys/selinfo.h>
  44#include <sys/uio.h>
  45#include <sys/acl.h>
  46#include <sys/ktr.h>
  47#include <sys/_seqc.h>
  48
  49/*
  50 * The vnode is the focus of all file activity in UNIX.  There is a
  51 * unique vnode allocated for each active file, each current directory,
  52 * each mounted-on file, text file, and the root.
  53 */
  54
  55/*
  56 * Vnode types.  VNON means no type.
  57 */
  58__enum_uint8_decl(vtype) {
  59	VNON,
  60	VREG,
  61	VDIR,
  62	VBLK,
  63	VCHR,
  64	VLNK,
  65	VSOCK,
  66	VFIFO,
  67	VBAD,
  68	VMARKER,
  69	VLASTTYPE = VMARKER,
  70};
  71
  72__enum_uint8_decl(vstate) {
  73	VSTATE_UNINITIALIZED,
  74	VSTATE_CONSTRUCTED,
  75	VSTATE_DESTROYING,
  76	VSTATE_DEAD,
  77	VLASTSTATE = VSTATE_DEAD,
  78};
  79
  80enum vgetstate {
  81	VGET_NONE,
  82	VGET_HOLDCNT,
  83	VGET_USECOUNT,
  84};
  85
  86/*
  87 * Each underlying filesystem allocates its own private area and hangs
  88 * it from v_data.  If non-null, this area is freed in getnewvnode().
  89 */
  90
  91struct namecache;
  92struct cache_fpl;
  93
  94struct vpollinfo {
  95	struct	mtx vpi_lock;		/* lock to protect below */
  96	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
  97	short	vpi_events;		/* what they are looking for */
  98	short	vpi_revents;		/* what has happened */
  99};
 100
 101/*
 102 * Reading or writing any of these items requires holding the appropriate lock.
 103 *
 104 * Lock reference:
 105 *	c - namecache mutex
 106 *	i - interlock
 107 *	l - mp mnt_listmtx or freelist mutex
 108 *	I - updated with atomics, 0->1 and 1->0 transitions with interlock held
 109 *	m - mount point interlock
 110 *	p - pollinfo lock
 111 *	u - Only a reference to the vnode is needed to read.
 112 *	v - vnode lock
 113 *
 114 * Vnodes may be found on many lists.  The general way to deal with operating
 115 * on a vnode that is on a list is:
 116 *	1) Lock the list and find the vnode.
 117 *	2) Lock interlock so that the vnode does not go away.
 118 *	3) Unlock the list to avoid lock order reversals.
 119 *	4) vget with LK_INTERLOCK and check for ENOENT, or
 120 *	5) Check for DOOMED if the vnode lock is not required.
 121 *	6) Perform your operation, then vput().
 122 */
 123
 124#if defined(_KERNEL) || defined(_KVM_VNODE)
 125
 126struct vnode {
 127	/*
 128	 * Fields which define the identity of the vnode.  These fields are
 129	 * owned by the filesystem (XXX: and vgone() ?)
 130	 */
 131	__enum_uint8(vtype) v_type;		/* u vnode type */
 132	__enum_uint8(vstate) v_state;		/* u vnode state */
 133	short	v_irflag;			/* i frequently read flags */
 134	seqc_t	v_seqc;				/* i modification count */
 135	uint32_t v_nchash;			/* u namecache hash */
 136	u_int	v_hash;
 137	struct	vop_vector *v_op;		/* u vnode operations vector */
 138	void	*v_data;			/* u private data for fs */
 139
 140	/*
 141	 * Filesystem instance stuff
 142	 */
 143	struct	mount *v_mount;			/* u ptr to vfs we are in */
 144	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
 145
 146	/*
 147	 * Type specific fields, only one applies to any given vnode.
 148	 */
 149	union {
 150		struct mount	*v_mountedhere;	/* v ptr to mountpoint (VDIR) */
 151		struct unpcb	*v_unpcb;	/* v unix domain net (VSOCK) */
 152		struct cdev	*v_rdev; 	/* v device (VCHR, VBLK) */
 153		struct fifoinfo	*v_fifoinfo;	/* v fifo (VFIFO) */
 154	};
 155
 156	/*
 157	 * vfs_hash: (mount + inode) -> vnode hash.  The hash value
 158	 * itself is grouped with other int fields, to avoid padding.
 159	 */
 160	LIST_ENTRY(vnode)	v_hashlist;
 161
 162	/*
 163	 * VFS_namecache stuff
 164	 */
 165	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
 166	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
 167	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
 168
 169	/*
 170	 * Locking
 171	 */
 172	struct	lock v_lock;			/* u (if fs don't have one) */
 173	struct	mtx v_interlock;		/* lock for "i" things */
 174	struct	lock *v_vnlock;			/* u pointer to vnode lock */
 175
 176	/*
 177	 * The machinery of being a vnode
 178	 */
 179	TAILQ_ENTRY(vnode) v_vnodelist;		/* l vnode lists */
 180	TAILQ_ENTRY(vnode) v_lazylist;		/* l vnode lazy list */
 181	struct bufobj	v_bufobj;		/* * Buffer cache object */
 182
 183	/*
 184	 * Hooks for various subsystems and features.
 185	 */
 186	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
 187	struct label *v_label;			/* MAC label for vnode */
 188	struct lockf *v_lockf;		/* Byte-level advisory lock list */
 189	struct rangelock v_rl;			/* Byte-range lock */
 190
 191	u_int	v_holdcnt;			/* I prevents recycling. */
 192	u_int	v_usecount;			/* I ref count of users */
 193	u_short	v_iflag;			/* i vnode flags (see below) */
 194	u_short	v_vflag;			/* v vnode flags */
 195	u_short	v_mflag;			/* l mnt-specific vnode flags */
 196	short	v_dbatchcpu;			/* i LRU requeue deferral batch */
 197	int	v_writecount;			/* I ref count of writers or
 198						   (negative) text users */
 199	int	v_seqc_users;			/* i modifications pending */
 200};
 201
 202#ifndef DEBUG_LOCKS
 203#ifdef _LP64
 204/*
 205 * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields
 206 * to the structure and there is nothing which can be done to prevent growth
 207 * then so be it. But don't grow it without a good reason.
 208 */
 209_Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes");
 210#endif
 211#endif
 212
 213#endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
 214
 215#define	bo2vnode(bo)	__containerof((bo), struct vnode, v_bufobj)
 216
 217/* XXX: These are temporary to avoid a source sweep at this time */
 218#define v_object	v_bufobj.bo_object
 219
 220/* We don't need to lock the knlist */
 221#define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
 222	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
 223
 224#define VN_KNOTE(vp, b, a)					\
 225	do {							\
 226		if (!VN_KNLIST_EMPTY(vp))			\
 227			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
 228			    (a) | KNF_NOKQLOCK);		\
 229	} while (0)
 230#define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
 231#define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
 232
 233/*
 234 * Vnode flags.
 235 *	VI flags are protected by interlock and live in v_iflag
 236 *	VIRF flags are protected by interlock and live in v_irflag
 237 *	VV flags are protected by the vnode lock and live in v_vflag
 238 *
 239 *	VIRF_DOOMED is doubly protected by the interlock and vnode lock.  Both
 240 *	are required for writing but the status may be checked with either.
 241 */
 242#define	VHOLD_NO_SMR	(1<<29)	/* Disable vhold_smr */
 243#define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
 244
 245#define	VIRF_DOOMED	0x0001	/* This vnode is being recycled */
 246#define	VIRF_PGREAD	0x0002	/* Direct reads from the page cache are permitted,
 247				   never cleared once set */
 248#define	VIRF_MOUNTPOINT	0x0004	/* This vnode is mounted on */
 249#define	VIRF_TEXT_REF	0x0008	/* Executable mappings ref the vnode */
 250#define	VIRF_CROSSMP	0x0010	/* Cross-mp vnode, no locking */
 251
 252#define	VI_UNUSED0	0x0001	/* unused */
 253#define	VI_MOUNT	0x0002	/* Mount in progress */
 254#define	VI_DOINGINACT	0x0004	/* VOP_INACTIVE is in progress */
 255#define	VI_OWEINACT	0x0008	/* Need to call inactive */
 256#define	VI_DEFINACT	0x0010	/* deferred inactive */
 257#define	VI_FOPENING	0x0020	/* In open, with opening process having the
 258				   first right to advlock file */
 259
 260#define	VV_ROOT		0x0001	/* root of its filesystem */
 261#define	VV_ISTTY	0x0002	/* vnode represents a tty */
 262#define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
 263#define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
 264#define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
 265#define	VV_VMSIZEVNLOCK	0x0020	/* object size check requires vnode lock */
 266#define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
 267#define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
 268#define	VV_PROCDEP	0x0100	/* vnode is process dependent */
 269#define	VV_UNLINKED	0x0200	/* unlinked but stil open directory */
 270#define	VV_DELETED	0x0400	/* should be removed */
 271#define	VV_MD		0x0800	/* vnode backs the md device */
 272#define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
 273#define	VV_READLINK	0x2000	/* fdescfs linux vnode */
 274#define	VV_UNREF	0x4000	/* vunref, do not drop lock in inactive() */
 275#define	VV_CROSSLOCK	0x8000	/* vnode lock is shared w/ root mounted here */
 276
 277#define	VMP_LAZYLIST	0x0001	/* Vnode is on mnt's lazy list */
 278
 279/*
 280 * Vnode attributes.  A field value of VNOVAL represents a field whose value
 281 * is unavailable (getattr) or which is not to be changed (setattr).
 282 */
 283struct vattr {
 284	__enum_uint8(vtype)	va_type;	/* vnode type (for create) */
 285	u_short		va_mode;	/* files access mode and type */
 286	u_short		va_padding0;
 287	uid_t		va_uid;		/* owner user id */
 288	gid_t		va_gid;		/* owner group id */
 289	nlink_t		va_nlink;	/* number of references to file */
 290	dev_t		va_fsid;	/* filesystem id */
 291	ino_t		va_fileid;	/* file id */
 292	u_quad_t	va_size;	/* file size in bytes */
 293	long		va_blocksize;	/* blocksize preferred for i/o */
 294	struct timespec	va_atime;	/* time of last access */
 295	struct timespec	va_mtime;	/* time of last modification */
 296	struct timespec	va_ctime;	/* time file changed */
 297	struct timespec	va_birthtime;	/* time file created */
 298	u_long		va_gen;		/* generation number of file */
 299	u_long		va_flags;	/* flags defined for file */
 300	dev_t		va_rdev;	/* device the special file represents */
 301	u_quad_t	va_bytes;	/* bytes of disk space held by file */
 302	u_quad_t	va_filerev;	/* file modification number */
 303	u_int		va_vaflags;	/* operations flags, see below */
 304	long		va_spare;	/* remain quad aligned */
 305};
 306
 307/*
 308 * Flags for va_vaflags.
 309 */
 310#define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
 311#define	VA_EXCLUSIVE	0x02		/* exclusive create request */
 312#define	VA_SYNC		0x04		/* O_SYNC truncation */
 313
 314/*
 315 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
 316 * help with write clustering)
 317 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
 318 */
 319#define	IO_UNIT		0x0001		/* do I/O as atomic unit */
 320#define	IO_APPEND	0x0002		/* append write to end */
 321#define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
 322#define	IO_NODELOCKED	0x0008		/* underlying node already locked */
 323#define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
 324#define	IO_VMIO		0x0020		/* data already in VMIO space */
 325#define	IO_INVAL	0x0040		/* invalidate after I/O */
 326#define	IO_SYNC		0x0080		/* do I/O synchronously */
 327#define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
 328#define	IO_NOREUSE	0x0200		/* VMIO data won't be reused */
 329#define	IO_EXT		0x0400		/* operate on external attributes */
 330#define	IO_NORMAL	0x0800		/* operate on regular data */
 331#define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
 332#define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
 333#define	IO_RANGELOCKED	0x4000		/* range locked */
 334#define	IO_DATASYNC	0x8000		/* do only data I/O synchronously */
 335
 336#define IO_SEQMAX	0x7F		/* seq heuristic max value */
 337#define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
 338
 339/*
 340 * Flags for accmode_t.
 341 */
 342#define	VEXEC			000000000100 /* execute/search permission */
 343#define	VWRITE			000000000200 /* write permission */
 344#define	VREAD			000000000400 /* read permission */
 345#define	VADMIN			000000010000 /* being the file owner */
 346#define	VAPPEND			000000040000 /* permission to write/append */
 347/*
 348 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
 349 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
 350 * and 0 otherwise.  This never happens with ordinary unix access rights
 351 * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
 352 * some other V* constant.
 353 */
 354#define	VEXPLICIT_DENY		000000100000
 355#define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
 356#define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
 357#define	VDELETE_CHILD	 	000001000000
 358#define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
 359#define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
 360#define	VDELETE		 	000010000000
 361#define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
 362#define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
 363#define	VWRITE_OWNER	 	000100000000 /* change file owner */
 364#define	VSYNCHRONIZE	 	000200000000 /* not used */
 365#define	VCREAT			000400000000 /* creating new file */
 366#define	VVERIFY			001000000000 /* verification required */
 367
 368/*
 369 * Permissions that were traditionally granted only to the file owner.
 370 */
 371#define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
 372    VWRITE_OWNER)
 373
 374/*
 375 * Permissions that were traditionally granted to everyone.
 376 */
 377#define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
 378
 379/*
 380 * Permissions that allow to change the state of the file in any way.
 381 */
 382#define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
 383    VDELETE)
 384
 385/*
 386 * Token indicating no attribute value yet assigned.
 387 */
 388#define	VNOVAL	(-1)
 389
 390/*
 391 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
 392 */
 393#define VLKTIMEOUT	(hz / 20 + 1)
 394
 395#ifdef _KERNEL
 396
 397#ifdef MALLOC_DECLARE
 398MALLOC_DECLARE(M_VNODE);
 399#endif
 400
 401extern u_int ncsizefactor;
 402extern const u_int io_hold_cnt;
 403
 404/*
 405 * Convert between vnode types and inode formats (since POSIX.1
 406 * defines mode word of stat structure in terms of inode formats).
 407 */
 408extern __enum_uint8(vtype)	iftovt_tab[];
 409extern int		vttoif_tab[];
 410#define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
 411#define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
 412#define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
 413
 414/*
 415 * Flags to various vnode functions.
 416 */
 417#define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
 418#define	FORCECLOSE	0x0002	/* vflush: force file closure */
 419#define	WRITECLOSE	0x0004	/* vflush: only close writable files */
 420#define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
 421#define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
 422#define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
 423#define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
 424#define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
 425#define	V_VMIO		0x0010	/* vinvalbuf: called during pageout */
 426#define	V_ALLOWCLEAN	0x0020	/* vinvalbuf: allow clean buffers after flush */
 427#define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
 428#define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
 429#define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
 430#define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
 431#define	V_PCATCH	0x0008	/* vn_start_write: make the sleep interruptible */
 432#define	V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH)
 433
 434#define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
 435#define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
 436
 437#define	VS_SKIP_UNMOUNT	0x0001	/* vfs_write_suspend: fail if the
 438				   filesystem is being unmounted */
 439
 440#define	VREF(vp)	vref(vp)
 441
 442#ifdef DIAGNOSTIC
 443#define	VATTR_NULL(vap)	vattr_null(vap)
 444#else
 445#define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
 446#endif /* DIAGNOSTIC */
 447
 448#define	NULLVP	((struct vnode *)NULL)
 449
 450/*
 451 * Global vnode data.
 452 */
 453extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
 454extern	struct mount *rootdevmp;	/* "/dev" mount */
 455extern	u_long desiredvnodes;		/* number of vnodes desired */
 456extern	struct uma_zone *namei_zone;
 457extern	struct vattr va_null;		/* predefined null vattr structure */
 458
 459extern	u_int vn_lock_pair_pause_max;
 460
 461#define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
 462#define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
 463#define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
 464#define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
 465#define	VI_MTX(vp)	(&(vp)->v_interlock)
 466
 467#define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
 468#define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
 469#define	VN_LOCK_DSHARE(vp)	lockdisableshare((vp)->v_vnlock)
 470
 471#endif /* _KERNEL */
 472
 473/*
 474 * Mods for extensibility.
 475 */
 476
 477/*
 478 * Flags for vdesc_flags:
 479 */
 480#define	VDESC_MAX_VPS		16
 481/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
 482#define	VDESC_VP0_WILLRELE	0x0001
 483#define	VDESC_VP1_WILLRELE	0x0002
 484#define	VDESC_VP2_WILLRELE	0x0004
 485#define	VDESC_VP3_WILLRELE	0x0008
 486
 487/*
 488 * A generic structure.
 489 * This can be used by bypass routines to identify generic arguments.
 490 */
 491struct vop_generic_args {
 492	struct vnodeop_desc *a_desc;
 493	/* other random data follows, presumably */
 494};
 495
 496typedef int vop_bypass_t(struct vop_generic_args *);
 497
 498/*
 499 * VDESC_NO_OFFSET is used to identify the end of the offset list
 500 * and in places where no such field exists.
 501 */
 502#define VDESC_NO_OFFSET -1
 503
 504/*
 505 * This structure describes the vnode operation taking place.
 506 */
 507struct vnodeop_desc {
 508	char	*vdesc_name;		/* a readable name for debugging */
 509	int	 vdesc_flags;		/* VDESC_* flags */
 510	int	vdesc_vop_offset;
 511	vop_bypass_t	*vdesc_call;	/* Function to call */
 512
 513	/*
 514	 * These ops are used by bypass routines to map and locate arguments.
 515	 * Creds and procs are not needed in bypass routines, but sometimes
 516	 * they are useful to (for example) transport layers.
 517	 * Nameidata is useful because it has a cred in it.
 518	 */
 519	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
 520	int	vdesc_vpp_offset;	/* return vpp location */
 521	int	vdesc_cred_offset;	/* cred location, if any */
 522	int	vdesc_thread_offset;	/* thread location, if any */
 523	int	vdesc_componentname_offset; /* if any */
 524};
 525
 526#ifdef _KERNEL
 527/*
 528 * A list of all the operation descs.
 529 */
 530extern struct vnodeop_desc *vnodeop_descs[];
 531
 532#define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
 533#define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
 534    ((s_type)(((char*)(struct_p)) + (s_offset)))
 535
 536#ifdef DEBUG_VFS_LOCKS
 537/*
 538 * Support code to aid in debugging VFS locking problems.  Not totally
 539 * reliable since if the thread sleeps between changing the lock
 540 * state and checking it with the assert, some other thread could
 541 * change the state.  They are good enough for debugging a single
 542 * filesystem using a single-threaded test.  Note that the unreliability is
 543 * limited to false negatives; efforts were made to ensure that false
 544 * positives cannot occur.
 545 */
 546void	assert_vi_locked(struct vnode *vp, const char *str);
 547void	assert_vi_unlocked(struct vnode *vp, const char *str);
 548void	assert_vop_elocked(struct vnode *vp, const char *str);
 549void	assert_vop_locked(struct vnode *vp, const char *str);
 550void	assert_vop_unlocked(struct vnode *vp, const char *str);
 551
 552#define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
 553#define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
 554#define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
 555#define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
 556#define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
 557
 558#define ASSERT_VOP_IN_SEQC(vp)	do {				\
 559	struct vnode *_vp = (vp);				\
 560								\
 561	VNPASS(seqc_in_modify(_vp->v_seqc), _vp);		\
 562} while (0)
 563
 564#define ASSERT_VOP_NOT_IN_SEQC(vp)	do {			\
 565	struct vnode *_vp = (vp);				\
 566								\
 567	VNPASS(!seqc_in_modify(_vp->v_seqc), _vp);		\
 568} while (0)
 569
 570#else /* !DEBUG_VFS_LOCKS */
 571
 572#define	ASSERT_VI_LOCKED(vp, str)	((void)0)
 573#define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
 574#define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
 575#define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
 576#define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
 577
 578#define ASSERT_VOP_IN_SEQC(vp)		((void)0)
 579#define ASSERT_VOP_NOT_IN_SEQC(vp)	((void)0)
 580
 581#endif /* DEBUG_VFS_LOCKS */
 582
 583/*
 584 * This call works for vnodes in the kernel.
 585 */
 586#define VCALL(c) ((c)->a_desc->vdesc_call(c))
 587
 588#define DOINGASYNC(vp)	   					\
 589	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
 590	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
 591
 592/*
 593 * VMIO support inline
 594 */
 595
 596extern int vmiodirenable;
 597
 598static __inline int
 599vn_canvmio(struct vnode *vp)
 600{
 601      if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
 602		return(TRUE);
 603	return(FALSE);
 604}
 605
 606/*
 607 * Finally, include the default set of vnode operations.
 608 */
 609typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
 610#include "vnode_if.h"
 611
 612/* vn_open_flags */
 613#define	VN_OPEN_NOAUDIT		0x00000001
 614#define	VN_OPEN_NOCAPCHECK	0x00000002
 615#define	VN_OPEN_NAMECACHE	0x00000004
 616#define	VN_OPEN_INVFS		0x00000008
 617#define	VN_OPEN_WANTIOCTLCAPS	0x00000010
 618
 619/* copy_file_range kernel flags */
 620#define	COPY_FILE_RANGE_KFLAGS		0xff000000
 621#define	COPY_FILE_RANGE_TIMEO1SEC	0x01000000	/* Return after 1sec. */
 622
 623/*
 624 * Public vnode manipulation functions.
 625 */
 626struct componentname;
 627struct file;
 628struct mount;
 629struct nameidata;
 630struct ostat;
 631struct freebsd11_stat;
 632struct thread;
 633struct proc;
 634struct stat;
 635struct nstat;
 636struct ucred;
 637struct uio;
 638struct vattr;
 639struct vfsops;
 640struct vnode;
 641
 642typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
 643
 644int	bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
 645	    daddr_t endn);
 646/* cache_* may belong in namei.h. */
 647void	cache_changesize(u_long newhashsize);
 648
 649#define	VFS_CACHE_DROPOLD	0x1
 650
 651void	cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
 652	    struct componentname *cnp, struct timespec *tsp,
 653	    struct timespec *dtsp, int flags);
 654#define	cache_enter(dvp, vp, cnp)					\
 655	cache_enter_time(dvp, vp, cnp, NULL, NULL)
 656void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
 657	    struct componentname *cnp, struct timespec *tsp,
 658	    struct timespec *dtsp);
 659int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
 660	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
 661void	cache_vnode_init(struct vnode *vp);
 662void	cache_purge(struct vnode *vp);
 663void	cache_purge_vgone(struct vnode *vp);
 664void	cache_purge_negative(struct vnode *vp);
 665void	cache_purgevfs(struct mount *mp);
 666char	*cache_symlink_alloc(size_t size, int flags);
 667void	cache_symlink_free(char *string, size_t size);
 668int	cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
 669	    size_t len);
 670void	cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
 671    struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
 672void	cache_vop_rmdir(struct vnode *dvp, struct vnode *vp);
 673void	cache_vop_vector_register(struct vop_vector *);
 674#ifdef INVARIANTS
 675void	cache_validate(struct vnode *dvp, struct vnode *vp,
 676	    struct componentname *cnp);
 677void	cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops);
 678void	cache_assert_no_entries(struct vnode *vp);
 679#else
 680static inline void
 681cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
 682{
 683}
 684
 685static inline void
 686cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops)
 687{
 688}
 689
 690static inline void
 691cache_assert_no_entries(struct vnode *vp)
 692{
 693}
 694#endif
 695void	cache_fast_lookup_enabled_recalc(void);
 696int	change_dir(struct vnode *vp, struct thread *td);
 697void	cvtstat(struct stat *st, struct ostat *ost);
 698int	freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
 699int	freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
 700int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
 701	    struct vnode **vpp);
 702void	getnewvnode_reserve(void);
 703void	getnewvnode_drop_reserve(void);
 704int	insmntque(struct vnode *vp, struct mount *mp);
 705int	insmntque1(struct vnode *vp, struct mount *mp);
 706u_quad_t init_va_filerev(void);
 707int	speedup_syncer(void);
 708int	vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
 709int	vn_getcwd(char *buf, char **retbuf, size_t *buflen);
 710int	vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
 711int	vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
 712int	vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
 713	    const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
 714	    char **freebuf, size_t *buflen);
 715struct vnode *
 716	vn_dir_dd_ino(struct vnode *vp);
 717int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
 718int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
 719	    char *path, u_int pathlen);
 720int	vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp,
 721	    struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name,
 722	    size_t leaf_length);
 723int	vaccess(__enum_uint8(vtype) type, mode_t file_mode, uid_t file_uid,
 724	    gid_t file_gid, accmode_t accmode, struct ucred *cred);
 725int	vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
 726	    struct ucred *cred);
 727int	vaccess_acl_nfs4(__enum_uint8(vtype) type, uid_t file_uid, gid_t file_gid,
 728	    struct acl *aclp, accmode_t accmode, struct ucred *cred);
 729int	vaccess_acl_posix1e(__enum_uint8(vtype) type, uid_t file_uid,
 730	    gid_t file_gid, struct acl *acl, accmode_t accmode,
 731	    struct ucred *cred);
 732void	vattr_null(struct vattr *vap);
 733void	vlazy(struct vnode *);
 734void	vdrop(struct vnode *);
 735void	vdropl(struct vnode *);
 736int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
 737int	vget(struct vnode *vp, int flags);
 738enum vgetstate	vget_prep_smr(struct vnode *vp);
 739enum vgetstate	vget_prep(struct vnode *vp);
 740int	vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
 741void	vget_finish_ref(struct vnode *vp, enum vgetstate vs);
 742void	vget_abort(struct vnode *vp, enum vgetstate vs);
 743void	vgone(struct vnode *vp);
 744void	vhold(struct vnode *);
 745void	vholdnz(struct vnode *);
 746bool	vhold_smr(struct vnode *);
 747int	vinactive(struct vnode *vp);
 748int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
 749int	vtruncbuf(struct vnode *vp, off_t length, int blksize);
 750void	v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
 751	    int blksize);
 752void	vunref(struct vnode *);
 753void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
 754int	vrecycle(struct vnode *vp);
 755int	vrecyclel(struct vnode *vp);
 756int	vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off,
 757	    struct ucred *cred);
 758int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
 759	    struct ucred *cred);
 760int	vn_close(struct vnode *vp,
 761	    int flags, struct ucred *file_cred, struct thread *td);
 762int	vn_copy_file_range(struct vnode *invp, off_t *inoffp,
 763	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
 764	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
 765	    struct thread *fsize_td);
 766int	vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags,
 767	    int ioflg, struct ucred *active_cred, struct ucred *file_cred);
 768void	vn_finished_write(struct mount *mp);
 769void	vn_finished_secondary_write(struct mount *mp);
 770int	vn_fsync_buf(struct vnode *vp, int waitfor);
 771int	vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
 772	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
 773	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
 774	    struct thread *fsize_td);
 775int	vn_need_pageq_flush(struct vnode *vp);
 776bool	vn_isdisk_error(struct vnode *vp, int *errp);
 777bool	vn_isdisk(struct vnode *vp);
 778int	_vn_lock(struct vnode *vp, int flags, const char *file, int line);
 779#define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
 780void	vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1,
 781	    struct vnode *vp2, bool vp2_locked, int lkflags2);
 782int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
 783int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
 784	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
 785int	vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
 786	    struct thread *td, struct file *fp);
 787void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
 788void	vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
 789	    vm_pindex_t end);
 790int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
 791int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
 792	    int len, off_t offset, enum uio_seg segflg, int ioflg,
 793	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
 794	    struct thread *td);
 795int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
 796	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
 797	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
 798	    struct thread *td);
 799int	vn_read_from_obj(struct vnode *vp, struct uio *uio);
 800int	vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
 801	    struct thread *td);
 802int	vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
 803	    off_t maxfsz, ssize_t *resid_adj, struct thread *td);
 804void	vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
 805int	vn_rlimit_trunc(u_quad_t size, struct thread *td);
 806int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
 807int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
 808	    int flags);
 809int	vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
 810	    struct ucred *cred);
 811int	vn_writechk(struct vnode *vp);
 812int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
 813	    const char *attrname, int *buflen, char *buf, struct thread *td);
 814int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
 815	    const char *attrname, int buflen, char *buf, struct thread *td);
 816int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
 817	    const char *attrname, struct thread *td);
 818int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
 819	    struct vnode **rvp);
 820int	vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
 821	    void *alloc_arg, int lkflags, struct vnode **rvp);
 822int	vn_utimes_perm(struct vnode *vp, struct vattr *vap,
 823	    struct ucred *cred, struct thread *td);
 824int	vn_cmp(struct file *, struct file *, struct thread *td);
 825
 826int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
 827int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
 828	    struct uio *uio);
 829
 830void	vn_seqc_write_begin_locked(struct vnode *vp);
 831void	vn_seqc_write_begin(struct vnode *vp);
 832void	vn_seqc_write_end_locked(struct vnode *vp);
 833void	vn_seqc_write_end(struct vnode *vp);
 834#define	vn_seqc_read_any(vp)		seqc_read_any(&(vp)->v_seqc)
 835#define	vn_seqc_read_notmodify(vp)	seqc_read_notmodify(&(vp)->v_seqc)
 836#define	vn_seqc_consistent(vp, seq)	seqc_consistent(&(vp)->v_seqc, seq)
 837
 838#define	vn_rangelock_unlock(vp, cookie)					\
 839	rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
 840#define	vn_rangelock_unlock_range(vp, cookie, start, end)		\
 841	rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), 	\
 842	    VI_MTX(vp))
 843#define	vn_rangelock_rlock(vp, start, end)				\
 844	rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
 845#define	vn_rangelock_tryrlock(vp, start, end)				\
 846	rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
 847#define	vn_rangelock_wlock(vp, start, end)				\
 848	rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
 849#define	vn_rangelock_trywlock(vp, start, end)				\
 850	rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
 851
 852#define	vn_irflag_read(vp)	atomic_load_short(&(vp)->v_irflag)
 853void	vn_irflag_set_locked(struct vnode *vp, short toset);
 854void	vn_irflag_set(struct vnode *vp, short toset);
 855void	vn_irflag_set_cond_locked(struct vnode *vp, short toset);
 856void	vn_irflag_set_cond(struct vnode *vp, short toset);
 857void	vn_irflag_unset_locked(struct vnode *vp, short tounset);
 858void	vn_irflag_unset(struct vnode *vp, short tounset);
 859
 860int	vfs_cache_lookup(struct vop_lookup_args *ap);
 861int	vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
 862void	vfs_timestamp(struct timespec *);
 863void	vfs_write_resume(struct mount *mp, int flags);
 864int	vfs_write_suspend(struct mount *mp, int flags);
 865int	vfs_write_suspend_umnt(struct mount *mp);
 866struct vnode *vnlru_alloc_marker(void);
 867void	vnlru_free_marker(struct vnode *);
 868void	vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
 869int	vop_stdbmap(struct vop_bmap_args *);
 870int	vop_stdfdatasync_buf(struct vop_fdatasync_args *);
 871int	vop_stdfsync(struct vop_fsync_args *);
 872int	vop_stdgetwritemount(struct vop_getwritemount_args *);
 873int	vop_stdgetpages(struct vop_getpages_args *);
 874int	vop_stdinactive(struct vop_inactive_args *);
 875int	vop_stdioctl(struct vop_ioctl_args *);
 876int	vop_stdneed_inactive(struct vop_need_inactive_args *);
 877int	vop_stdkqfilter(struct vop_kqfilter_args *);
 878int	vop_stdlock(struct vop_lock1_args *);
 879int	vop_stdunlock(struct vop_unlock_args *);
 880int	vop_stdislocked(struct vop_islocked_args *);
 881int	vop_lock(struct vop_lock1_args *);
 882int	vop_unlock(struct vop_unlock_args *);
 883int	vop_islocked(struct vop_islocked_args *);
 884int	vop_stdputpages(struct vop_putpages_args *);
 885int	vop_nopoll(struct vop_poll_args *);
 886int	vop_stdaccess(struct vop_access_args *ap);
 887int	vop_stdaccessx(struct vop_accessx_args *ap);
 888int	vop_stdadvise(struct vop_advise_args *ap);
 889int	vop_stdadvlock(struct vop_advlock_args *ap);
 890int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
 891int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
 892int	vop_stdallocate(struct vop_allocate_args *ap);
 893int	vop_stddeallocate(struct vop_deallocate_args *ap);
 894int	vop_stdset_text(struct vop_set_text_args *ap);
 895int	vop_stdpathconf(struct vop_pathconf_args *);
 896int	vop_stdpoll(struct vop_poll_args *);
 897int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
 898int	vop_stdvptofh(struct vop_vptofh_args *ap);
 899int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
 900int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
 901int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
 902int	vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap);
 903int	vop_eopnotsupp(struct vop_generic_args *ap);
 904int	vop_ebadf(struct vop_generic_args *ap);
 905int	vop_einval(struct vop_generic_args *ap);
 906int	vop_enoent(struct vop_generic_args *ap);
 907int	vop_enotty(struct vop_generic_args *ap);
 908int	vop_eagain(struct vop_generic_args *ap);
 909int	vop_null(struct vop_generic_args *ap);
 910int	vop_panic(struct vop_generic_args *ap);
 911int	dead_poll(struct vop_poll_args *ap);
 912int	dead_read(struct vop_read_args *ap);
 913int	dead_write(struct vop_write_args *ap);
 914
 915/* These are called from within the actual VOPS. */
 916void	vop_close_post(void *a, int rc);
 917void	vop_create_pre(void *a);
 918void	vop_create_post(void *a, int rc);
 919void	vop_whiteout_pre(void *a);
 920void	vop_whiteout_post(void *a, int rc);
 921void	vop_deleteextattr_pre(void *a);
 922void	vop_deleteextattr_post(void *a, int rc);
 923void	vop_link_pre(void *a);
 924void	vop_link_post(void *a, int rc);
 925void	vop_lookup_post(void *a, int rc);
 926void	vop_lookup_pre(void *a);
 927void	vop_mkdir_pre(void *a);
 928void	vop_mkdir_post(void *a, int rc);
 929void	vop_mknod_pre(void *a);
 930void	vop_mknod_post(void *a, int rc);
 931void	vop_open_post(void *a, int rc);
 932void	vop_read_post(void *a, int rc);
 933void	vop_read_pgcache_post(void *ap, int rc);
 934void	vop_readdir_post(void *a, int rc);
 935void	vop_reclaim_post(void *a, int rc);
 936void	vop_remove_pre(void *a);
 937void	vop_remove_post(void *a, int rc);
 938void	vop_rename_post(void *a, int rc);
 939void	vop_rename_pre(void *a);
 940void	vop_rmdir_pre(void *a);
 941void	vop_rmdir_post(void *a, int rc);
 942void	vop_setattr_pre(void *a);
 943void	vop_setattr_post(void *a, int rc);
 944void	vop_setacl_pre(void *a);
 945void	vop_setacl_post(void *a, int rc);
 946void	vop_setextattr_pre(void *a);
 947void	vop_setextattr_post(void *a, int rc);
 948void	vop_symlink_pre(void *a);
 949void	vop_symlink_post(void *a, int rc);
 950int	vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
 951
 952#ifdef DEBUG_VFS_LOCKS
 953void	vop_fdatasync_debugpre(void *a);
 954void	vop_fdatasync_debugpost(void *a, int rc);
 955void	vop_fplookup_vexec_debugpre(void *a);
 956void	vop_fplookup_vexec_debugpost(void *a, int rc);
 957void	vop_fplookup_symlink_debugpre(void *a);
 958void	vop_fplookup_symlink_debugpost(void *a, int rc);
 959void	vop_fsync_debugpre(void *a);
 960void	vop_fsync_debugpost(void *a, int rc);
 961void	vop_strategy_debugpre(void *a);
 962void	vop_lock_debugpre(void *a);
 963void	vop_lock_debugpost(void *a, int rc);
 964void	vop_unlock_debugpre(void *a);
 965void	vop_need_inactive_debugpre(void *a);
 966void	vop_need_inactive_debugpost(void *a, int rc);
 967void	vop_mkdir_debugpost(void *a, int rc);
 968#else
 969#define	vop_fdatasync_debugpre(x)		do { } while (0)
 970#define	vop_fdatasync_debugpost(x, y)		do { } while (0)
 971#define	vop_fplookup_vexec_debugpre(x)		do { } while (0)
 972#define	vop_fplookup_vexec_debugpost(x, y)	do { } while (0)
 973#define	vop_fplookup_symlink_debugpre(x)	do { } while (0)
 974#define	vop_fplookup_symlink_debugpost(x, y)	do { } while (0)
 975#define	vop_fsync_debugpre(x)			do { } while (0)
 976#define	vop_fsync_debugpost(x, y)		do { } while (0)
 977#define	vop_strategy_debugpre(x)		do { } while (0)
 978#define	vop_lock_debugpre(x)			do { } while (0)
 979#define	vop_lock_debugpost(x, y)		do { } while (0)
 980#define	vop_unlock_debugpre(x)			do { } while (0)
 981#define	vop_need_inactive_debugpre(x)		do { } while (0)
 982#define	vop_need_inactive_debugpost(x, y)	do { } while (0)
 983#define	vop_mkdir_debugpost(x, y)		do { } while (0)
 984#endif
 985
 986void	vop_rename_fail(struct vop_rename_args *ap);
 987
 988#define	vop_stat_helper_pre(ap)	({						\
 989	struct vop_stat_args *_ap = (ap);					\
 990	int _error;								\
 991	AUDIT_ARG_VNODE1(ap->a_vp);						\
 992	_error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\
 993	if (__predict_true(_error == 0)) {					\
 994		ap->a_sb->st_padding0 = 0;					\
 995		ap->a_sb->st_padding1 = 0;					\
 996		bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare));	\
 997	}									\
 998	_error;									\
 999})
1000
1001#define	vop_stat_helper_post(ap, error)	({					\
1002	struct vop_stat_args *_ap = (ap);					\
1003	int _error = (error);							\
1004	if (priv_check_cred_vfs_generation(_ap->a_active_cred))			\
1005		_ap->a_sb->st_gen = 0;						\
1006	_error;									\
1007})
1008
1009#define	VOP_WRITE_PRE(ap)						\
1010	struct vattr va;						\
1011	int error;							\
1012	off_t osize, ooffset, noffset;					\
1013									\
1014	osize = ooffset = noffset = 0;					\
1015	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
1016		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
1017		if (error)						\
1018			return (error);					\
1019		ooffset = (ap)->a_uio->uio_offset;			\
1020		osize = (off_t)va.va_size;				\
1021	}
1022
1023#define VOP_WRITE_POST(ap, ret)						\
1024	noffset = (ap)->a_uio->uio_offset;				\
1025	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
1026		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
1027		    | (noffset > osize ? NOTE_EXTEND : 0));		\
1028	}
1029
1030#define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1031
1032#ifdef INVARIANTS
1033#define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)				\
1034do {									\
1035	int error_;							\
1036									\
1037	error_ = VOP_ADD_WRITECOUNT((vp), (cnt));			\
1038	VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d",	\
1039	    error_));							\
1040} while (0)
1041#define	VOP_SET_TEXT_CHECKED(vp)					\
1042do {									\
1043	int error_;							\
1044									\
1045	error_ = VOP_SET_TEXT((vp));					\
1046	VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d",	\
1047	    error_));							\
1048} while (0)
1049#define	VOP_UNSET_TEXT_CHECKED(vp)					\
1050do {									\
1051	int error_;							\
1052									\
1053	error_ = VOP_UNSET_TEXT((vp));					\
1054	VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d",	\
1055	    error_));							\
1056} while (0)
1057#else
1058#define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)	VOP_ADD_WRITECOUNT((vp), (cnt))
1059#define	VOP_SET_TEXT_CHECKED(vp)		VOP_SET_TEXT((vp))
1060#define	VOP_UNSET_TEXT_CHECKED(vp)		VOP_UNSET_TEXT((vp))
1061#endif
1062
1063#define	VN_IS_DOOMED(vp)	__predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1064
1065void	vput(struct vnode *vp);
1066void	vrele(struct vnode *vp);
1067void	vref(struct vnode *vp);
1068void	vrefact(struct vnode *vp);
1069void 	v_addpollinfo(struct vnode *vp);
1070static __inline int
1071vrefcnt(struct vnode *vp)
1072{
1073
1074	return (vp->v_usecount);
1075}
1076
1077#define	vholdl(vp)	do {						\
1078	ASSERT_VI_LOCKED(vp, __func__);					\
1079	vhold(vp);							\
1080} while (0)
1081
1082#define	vrefl(vp)	do {						\
1083	ASSERT_VI_LOCKED(vp, __func__);					\
1084	vref(vp);							\
1085} while (0)
1086
1087int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1088void vnode_destroy_vobject(struct vnode *vp);
1089
1090extern struct vop_vector fifo_specops;
1091extern struct vop_vector dead_vnodeops;
1092extern struct vop_vector default_vnodeops;
1093
1094#define VOP_PANIC	((void*)(uintptr_t)vop_panic)
1095#define VOP_NULL	((void*)(uintptr_t)vop_null)
1096#define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
1097#define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
1098#define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
1099#define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
1100#define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
1101#define VOP_EAGAIN	((void*)(uintptr_t)vop_eagain)
1102
1103/* fifo_vnops.c */
1104int	fifo_printinfo(struct vnode *);
1105
1106/* vfs_hash.c */
1107typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1108
1109void vfs_hash_changesize(u_long newhashsize);
1110int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1111    struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1112u_int vfs_hash_index(struct vnode *vp);
1113int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1114    struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1115void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1116    struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1117void vfs_hash_rehash(struct vnode *vp, u_int hash);
1118void vfs_hash_remove(struct vnode *vp);
1119
1120int vfs_kqfilter(struct vop_kqfilter_args *);
1121struct dirent;
1122int vn_dir_next_dirent(struct vnode *vp, struct thread *td,
1123    char *dirbuf, size_t dirbuflen,
1124    struct dirent **dpp, size_t *len, off_t *off, int *eofflag);
1125int vn_dir_check_empty(struct vnode *vp);
1126int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1127
1128int vfs_unixify_accmode(accmode_t *accmode);
1129
1130void vfs_unp_reclaim(struct vnode *vp);
1131
1132int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1133int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1134    gid_t gid);
1135int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1136    struct thread *td);
1137int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1138    struct thread *td);
1139int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred);
1140int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred);
1141
1142void vn_fsid(struct vnode *vp, struct vattr *va);
1143
1144int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1145int vn_lktype_write(struct mount *mp, struct vnode *vp);
1146
1147#ifdef INVARIANTS
1148void vn_set_state_validate(struct vnode *vp, __enum_uint8(vstate) state);
1149#endif
1150
1151static inline void
1152vn_set_state(struct vnode *vp, __enum_uint8(vstate) state)
1153{
1154#ifdef INVARIANTS
1155	vn_set_state_validate(vp, state);
1156#endif
1157	vp->v_state = state;
1158}
1159
1160static inline __enum_uint8(vstate)
1161vn_get_state(struct vnode *vp)
1162{
1163	return (vp->v_state);
1164}
1165
1166#define VOP_UNLOCK_FLAGS(vp, flags)	({				\
1167	struct vnode *_vp = (vp);					\
1168	int _flags = (flags);						\
1169	int _error;							\
1170									\
1171        if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0)		\
1172                panic("%s: unsupported flags %x\n", __func__, flags);	\
1173        _error = VOP_UNLOCK(_vp);					\
1174        if (_flags & LK_INTERLOCK)					\
1175                VI_UNLOCK(_vp);						\
1176        _error;								\
1177})
1178
1179#include <sys/kernel.h>
1180
1181#define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1182	SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1183	    vfs_vector_op_register, &vnodeops)
1184
1185#define VFS_SMR_DECLARE				\
1186	extern smr_t vfs_smr
1187
1188#define VFS_SMR()	vfs_smr
1189#define vfs_smr_enter()	smr_enter(VFS_SMR())
1190#define vfs_smr_exit()	smr_exit(VFS_SMR())
1191#define vfs_smr_synchronize()	smr_synchronize(VFS_SMR())
1192#define vfs_smr_entered_load(ptr)	smr_entered_load((ptr), VFS_SMR())
1193#define VFS_SMR_ASSERT_ENTERED()	SMR_ASSERT_ENTERED(VFS_SMR())
1194#define VFS_SMR_ASSERT_NOT_ENTERED()	SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1195#define VFS_SMR_ZONE_SET(zone)	uma_zone_set_smr((zone), VFS_SMR())
1196
1197#define vn_load_v_data_smr(vp)	({		\
1198	struct vnode *_vp = (vp);		\
1199						\
1200	VFS_SMR_ASSERT_ENTERED();		\
1201	atomic_load_consume_ptr(&(_vp)->v_data);\
1202})
1203
1204#endif /* _KERNEL */
1205
1206#endif /* !_SYS_VNODE_H_ */