1 /*-
2  * Copyright (c) 1989, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 4. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)vnode.h	8.7 (Berkeley) 2/4/94
30  * $FreeBSD: stable/9/sys/sys/vnode.h 288168 2015-09-24 09:35:35Z smh $
31  */
32 
33 #ifndef _SYS_VNODE_H_
34 #define	_SYS_VNODE_H_
35 
36 #include <sys/bufobj.h>
37 #include <sys/queue.h>
38 #include <sys/lock.h>
39 #include <sys/lockmgr.h>
40 #include <sys/mutex.h>
41 #include <sys/rangelock.h>
42 #include <sys/selinfo.h>
43 #include <sys/uio.h>
44 #include <sys/acl.h>
45 #include <sys/ktr.h>
46 
47 /*
48  * The vnode is the focus of all file activity in UNIX.  There is a
49  * unique vnode allocated for each active file, each current directory,
50  * each mounted-on file, text file, and the root.
51  */
52 
53 /*
54  * Vnode types.  VNON means no type.
55  */
56 enum vtype	{ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
57 		  VMARKER };
58 
59 /*
60  * Each underlying filesystem allocates its own private area and hangs
61  * it from v_data.  If non-null, this area is freed in getnewvnode().
62  */
63 
64 struct namecache;
65 
66 struct vpollinfo {
67 	struct	mtx vpi_lock;		/* lock to protect below */
68 	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
69 	short	vpi_events;		/* what they are looking for */
70 	short	vpi_revents;		/* what has happened */
71 };
72 
73 /*
74  * Reading or writing any of these items requires holding the appropriate lock.
75  *
76  * Lock reference:
77  *	c - namecache mutex
78  *	f - freelist mutex
79  *	i - interlock
80  *	m - mount point interlock
81  *	p - pollinfo lock
82  *	u - Only a reference to the vnode is needed to read.
83  *	v - vnode lock
84  *
85  * Vnodes may be found on many lists.  The general way to deal with operating
86  * on a vnode that is on a list is:
87  *	1) Lock the list and find the vnode.
88  *	2) Lock interlock so that the vnode does not go away.
89  *	3) Unlock the list to avoid lock order reversals.
90  *	4) vget with LK_INTERLOCK and check for ENOENT, or
91  *	5) Check for DOOMED if the vnode lock is not required.
92  *	6) Perform your operation, then vput().
93  */
94 
95 #if defined(_KERNEL) || defined(_KVM_VNODE)
96 
97 struct vnode {
98 	/*
99 	 * Fields which define the identity of the vnode.  These fields are
100 	 * owned by the filesystem (XXX: and vgone() ?)
101 	 */
102 	enum	vtype v_type;			/* u vnode type */
103 	const char *v_tag;			/* u type of underlying data */
104 	struct	vop_vector *v_op;		/* u vnode operations vector */
105 	void	*v_data;			/* u private data for fs */
106 
107 	/*
108 	 * Filesystem instance stuff
109 	 */
110 	struct	mount *v_mount;			/* u ptr to vfs we are in */
111 	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
112 
113 	/*
114 	 * Type specific fields, only one applies to any given vnode.
115 	 * See #defines below for renaming to v_* namespace.
116 	 */
117 	union {
118 		struct mount	*vu_mount;	/* v ptr to mountpoint (VDIR) */
119 		struct socket	*vu_socket;	/* v unix domain net (VSOCK) */
120 		struct cdev	*vu_cdev; 	/* v device (VCHR, VBLK) */
121 		struct fifoinfo	*vu_fifoinfo;	/* v fifo (VFIFO) */
122 	} v_un;
123 
124 	/*
125 	 * vfs_hash:  (mount + inode) -> vnode hash.
126 	 */
127 	LIST_ENTRY(vnode)	v_hashlist;
128 	u_int			v_hash;
129 
130 	/*
131 	 * VFS_namecache stuff
132 	 */
133 	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
134 	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
135 	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
136 
137 	/*
138 	 * clustering stuff
139 	 */
140 	daddr_t	v_cstart;			/* v start block of cluster */
141 	daddr_t	v_lasta;			/* v last allocation  */
142 	daddr_t	v_lastw;			/* v last write  */
143 	int	v_clen;				/* v length of cur. cluster */
144 
145 	/*
146 	 * Locking
147 	 */
148 	struct	lock v_lock;			/* u (if fs don't have one) */
149 	struct	mtx v_interlock;		/* lock for "i" things */
150 	struct	lock *v_vnlock;			/* u pointer to vnode lock */
151 	int	v_holdcnt;			/* i prevents recycling. */
152 	int	v_usecount;			/* i ref count of users */
153 	u_long	v_iflag;			/* i vnode flags (see below) */
154 	u_long	v_vflag;			/* v vnode flags */
155 	int	v_writecount;			/* v ref count of writers */
156 
157 	/*
158 	 * The machinery of being a vnode
159 	 */
160 	TAILQ_ENTRY(vnode) v_actfreelist;	/* f vnode active/free lists */
161 	struct bufobj	v_bufobj;		/* * Buffer cache object */
162 
163 	/*
164 	 * Hooks for various subsystems and features.
165 	 */
166 	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
167 	struct label *v_label;			/* MAC label for vnode */
168 	struct lockf *v_lockf;		/* Byte-level advisory lock list */
169 	struct rangelock v_rl;			/* Byte-range lock */
170 };
171 
172 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
173 
174 #define	v_mountedhere	v_un.vu_mount
175 #define	v_socket	v_un.vu_socket
176 #define	v_rdev		v_un.vu_cdev
177 #define	v_fifoinfo	v_un.vu_fifoinfo
178 
179 /* XXX: These are temporary to avoid a source sweep at this time */
180 #define v_object	v_bufobj.bo_object
181 
182 /*
183  * Userland version of struct vnode, for sysctl.
184  */
185 struct xvnode {
186 	size_t	xv_size;			/* sizeof(struct xvnode) */
187 	void	*xv_vnode;			/* address of real vnode */
188 	u_long	xv_flag;			/* vnode vflags */
189 	int	xv_usecount;			/* reference count of users */
190 	int	xv_writecount;			/* reference count of writers */
191 	int	xv_holdcnt;			/* page & buffer references */
192 	u_long	xv_id;				/* capability identifier */
193 	void	*xv_mount;			/* address of parent mount */
194 	long	xv_numoutput;			/* num of writes in progress */
195 	enum	vtype xv_type;			/* vnode type */
196 	union {
197 		void	*xvu_socket;		/* socket, if VSOCK */
198 		void	*xvu_fifo;		/* fifo, if VFIFO */
199 		dev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
200 		struct {
201 			dev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
202 			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
203 		} xv_uns;
204 	} xv_un;
205 };
206 #define xv_socket	xv_un.xvu_socket
207 #define xv_fifo		xv_un.xvu_fifo
208 #define xv_rdev		xv_un.xvu_rdev
209 #define xv_dev		xv_un.xv_uns.xvu_dev
210 #define xv_ino		xv_un.xv_uns.xvu_ino
211 
212 /* We don't need to lock the knlist */
213 #define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
214 	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
215 
216 #define VN_KNOTE(vp, b, a)					\
217 	do {							\
218 		if (!VN_KNLIST_EMPTY(vp))			\
219 			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
220 			    (a) | KNF_NOKQLOCK);		\
221 	} while (0)
222 #define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
223 #define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
224 
225 /*
226  * Vnode flags.
227  *	VI flags are protected by interlock and live in v_iflag
228  *	VV flags are protected by the vnode lock and live in v_vflag
229  *
230  *	VI_DOOMED is doubly protected by the interlock and vnode lock.  Both
231  *	are required for writing but the status may be checked with either.
232  */
233 #define	VI_MOUNT	0x0020	/* Mount in progress */
234 #define	VI_AGE		0x0040	/* Insert vnode at head of free list */
235 #define	VI_DOOMED	0x0080	/* This vnode is being recycled */
236 #define	VI_FREE		0x0100	/* This vnode is on the freelist */
237 #define	VI_ACTIVE	0x0200	/* This vnode is on the active list */
238 #define	VI_DOINGINACT	0x0800	/* VOP_INACTIVE is in progress */
239 #define	VI_OWEINACT	0x1000	/* Need to call inactive */
240 
241 #define	VV_ROOT		0x0001	/* root of its filesystem */
242 #define	VV_ISTTY	0x0002	/* vnode represents a tty */
243 #define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
244 #define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
245 #define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
246 #define	VV_TEXT		0x0020	/* vnode is a pure text prototype */
247 #define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
248 #define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
249 #define	VV_PROCDEP	0x0100	/* vnode is process dependent */
250 #define	VV_NOKNOTE	0x0200	/* don't activate knotes on this vnode */
251 #define	VV_DELETED	0x0400	/* should be removed */
252 #define	VV_MD		0x0800	/* vnode backs the md device */
253 #define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
254 
255 /*
256  * Vnode attributes.  A field value of VNOVAL represents a field whose value
257  * is unavailable (getattr) or which is not to be changed (setattr).
258  */
259 struct vattr {
260 	enum vtype	va_type;	/* vnode type (for create) */
261 	u_short		va_mode;	/* files access mode and type */
262 	short		va_nlink;	/* number of references to file */
263 	uid_t		va_uid;		/* owner user id */
264 	gid_t		va_gid;		/* owner group id */
265 	dev_t		va_fsid;	/* filesystem id */
266 	long		va_fileid;	/* file id */
267 	u_quad_t	va_size;	/* file size in bytes */
268 	long		va_blocksize;	/* blocksize preferred for i/o */
269 	struct timespec	va_atime;	/* time of last access */
270 	struct timespec	va_mtime;	/* time of last modification */
271 	struct timespec	va_ctime;	/* time file changed */
272 	struct timespec	va_birthtime;	/* time file created */
273 	u_long		va_gen;		/* generation number of file */
274 	u_long		va_flags;	/* flags defined for file */
275 	dev_t		va_rdev;	/* device the special file represents */
276 	u_quad_t	va_bytes;	/* bytes of disk space held by file */
277 	u_quad_t	va_filerev;	/* file modification number */
278 	u_int		va_vaflags;	/* operations flags, see below */
279 	long		va_spare;	/* remain quad aligned */
280 };
281 
282 /*
283  * Flags for va_vaflags.
284  */
285 #define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
286 #define	VA_EXCLUSIVE	0x02		/* exclusive create request */
287 
288 /*
289  * Flags for ioflag. (high 16 bits used to ask for read-ahead and
290  * help with write clustering)
291  * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
292  */
293 #define	IO_UNIT		0x0001		/* do I/O as atomic unit */
294 #define	IO_APPEND	0x0002		/* append write to end */
295 #define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
296 #define	IO_NODELOCKED	0x0008		/* underlying node already locked */
297 #define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
298 #define	IO_VMIO		0x0020		/* data already in VMIO space */
299 #define	IO_INVAL	0x0040		/* invalidate after I/O */
300 #define	IO_SYNC		0x0080		/* do I/O synchronously */
301 #define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
302 #define	IO_EXT		0x0400		/* operate on external attributes */
303 #define	IO_NORMAL	0x0800		/* operate on regular data */
304 #define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
305 #define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
306 
307 #define IO_SEQMAX	0x7F		/* seq heuristic max value */
308 #define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
309 
310 /*
311  * Flags for accmode_t.
312  */
313 #define	VEXEC			000000000100 /* execute/search permission */
314 #define	VWRITE			000000000200 /* write permission */
315 #define	VREAD			000000000400 /* read permission */
316 #define	VADMIN			000000010000 /* being the file owner */
317 #define	VAPPEND			000000040000 /* permission to write/append */
318 /*
319  * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
320  * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
321  * and 0 otherwise.  This never happens with ordinary unix access rights
322  * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
323  * some other V* constant.
324  */
325 #define	VEXPLICIT_DENY		000000100000
326 #define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
327 #define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
328 #define	VDELETE_CHILD	 	000001000000
329 #define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
330 #define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
331 #define	VDELETE		 	000010000000
332 #define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
333 #define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
334 #define	VWRITE_OWNER	 	000100000000 /* change file owner */
335 #define	VSYNCHRONIZE	 	000200000000 /* not used */
336 
337 /*
338  * Permissions that were traditionally granted only to the file owner.
339  */
340 #define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
341     VWRITE_OWNER)
342 
343 /*
344  * Permissions that were traditionally granted to everyone.
345  */
346 #define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
347 
348 /*
349  * Permissions that allow to change the state of the file in any way.
350  */
351 #define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
352     VDELETE)
353 
354 /*
355  * Token indicating no attribute value yet assigned.
356  */
357 #define	VNOVAL	(-1)
358 
359 /*
360  * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
361  */
362 #define VLKTIMEOUT	(hz / 20 + 1)
363 
364 #ifdef _KERNEL
365 
366 #ifdef MALLOC_DECLARE
367 MALLOC_DECLARE(M_VNODE);
368 #endif
369 
370 /*
371  * Convert between vnode types and inode formats (since POSIX.1
372  * defines mode word of stat structure in terms of inode formats).
373  */
374 extern enum vtype	iftovt_tab[];
375 extern int		vttoif_tab[];
376 #define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
377 #define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
378 #define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
379 
380 /*
381  * Flags to various vnode functions.
382  */
383 #define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
384 #define	FORCECLOSE	0x0002	/* vflush: force file closure */
385 #define	WRITECLOSE	0x0004	/* vflush: only close writable files */
386 #define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
387 #define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
388 #define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
389 #define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
390 #define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
391 #define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
392 #define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
393 #define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
394 #define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
395 
396 #define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
397 #define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
398 
399 #define	VREF(vp)	vref(vp)
400 
401 #ifdef DIAGNOSTIC
402 #define	VATTR_NULL(vap)	vattr_null(vap)
403 #else
404 #define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
405 #endif /* DIAGNOSTIC */
406 
407 #define	NULLVP	((struct vnode *)NULL)
408 
409 /*
410  * Global vnode data.
411  */
412 extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
413 extern	int async_io_version;		/* 0 or POSIX version of AIO i'face */
414 extern	int desiredvnodes;		/* number of vnodes desired */
415 extern	struct uma_zone *namei_zone;
416 extern	struct vattr va_null;		/* predefined null vattr structure */
417 
418 #define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
419 #define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
420 #define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
421 #define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
422 #define	VI_MTX(vp)	(&(vp)->v_interlock)
423 
424 #define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
425 #define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
426 
427 #endif /* _KERNEL */
428 
429 /*
430  * Mods for extensibility.
431  */
432 
433 /*
434  * Flags for vdesc_flags:
435  */
436 #define	VDESC_MAX_VPS		16
437 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
438 #define	VDESC_VP0_WILLRELE	0x0001
439 #define	VDESC_VP1_WILLRELE	0x0002
440 #define	VDESC_VP2_WILLRELE	0x0004
441 #define	VDESC_VP3_WILLRELE	0x0008
442 #define	VDESC_NOMAP_VPP		0x0100
443 #define	VDESC_VPP_WILLRELE	0x0200
444 
445 /*
446  * A generic structure.
447  * This can be used by bypass routines to identify generic arguments.
448  */
449 struct vop_generic_args {
450 	struct vnodeop_desc *a_desc;
451 	/* other random data follows, presumably */
452 };
453 
454 typedef int vop_bypass_t(struct vop_generic_args *);
455 
456 /*
457  * VDESC_NO_OFFSET is used to identify the end of the offset list
458  * and in places where no such field exists.
459  */
460 #define VDESC_NO_OFFSET -1
461 
462 /*
463  * This structure describes the vnode operation taking place.
464  */
465 struct vnodeop_desc {
466 	char	*vdesc_name;		/* a readable name for debugging */
467 	int	 vdesc_flags;		/* VDESC_* flags */
468 	vop_bypass_t	*vdesc_call;	/* Function to call */
469 
470 	/*
471 	 * These ops are used by bypass routines to map and locate arguments.
472 	 * Creds and procs are not needed in bypass routines, but sometimes
473 	 * they are useful to (for example) transport layers.
474 	 * Nameidata is useful because it has a cred in it.
475 	 */
476 	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
477 	int	vdesc_vpp_offset;	/* return vpp location */
478 	int	vdesc_cred_offset;	/* cred location, if any */
479 	int	vdesc_thread_offset;	/* thread location, if any */
480 	int	vdesc_componentname_offset; /* if any */
481 };
482 
483 #ifdef _KERNEL
484 /*
485  * A list of all the operation descs.
486  */
487 extern struct vnodeop_desc *vnodeop_descs[];
488 
489 #define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
490 #define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
491     ((s_type)(((char*)(struct_p)) + (s_offset)))
492 
493 
494 #ifdef DEBUG_VFS_LOCKS
495 /*
496  * Support code to aid in debugging VFS locking problems.  Not totally
497  * reliable since if the thread sleeps between changing the lock
498  * state and checking it with the assert, some other thread could
499  * change the state.  They are good enough for debugging a single
500  * filesystem using a single-threaded test.
501  */
502 void	assert_vi_locked(struct vnode *vp, const char *str);
503 void	assert_vi_unlocked(struct vnode *vp, const char *str);
504 void	assert_vop_elocked(struct vnode *vp, const char *str);
505 #if 0
506 void	assert_vop_elocked_other(struct vnode *vp, const char *str);
507 #endif
508 void	assert_vop_locked(struct vnode *vp, const char *str);
509 #if 0
510 voi0	assert_vop_slocked(struct vnode *vp, const char *str);
511 #endif
512 void	assert_vop_unlocked(struct vnode *vp, const char *str);
513 
514 #define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
515 #define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
516 #define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
517 #if 0
518 #define	ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
519 #endif
520 #define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
521 #if 0
522 #define	ASSERT_VOP_SLOCKED(vp, str)	assert_vop_slocked((vp), (str))
523 #endif
524 #define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
525 
526 #else /* !DEBUG_VFS_LOCKS */
527 
528 #define	ASSERT_VI_LOCKED(vp, str)	((void)0)
529 #define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
530 #define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
531 #if 0
532 #define	ASSERT_VOP_ELOCKED_OTHER(vp, str)
533 #endif
534 #define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
535 #if 0
536 #define	ASSERT_VOP_SLOCKED(vp, str)
537 #endif
538 #define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
539 #endif /* DEBUG_VFS_LOCKS */
540 
541 
542 /*
543  * This call works for vnodes in the kernel.
544  */
545 #define VCALL(c) ((c)->a_desc->vdesc_call(c))
546 
547 #define DOINGASYNC(vp)	   					\
548 	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
549 	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
550 
551 /*
552  * VMIO support inline
553  */
554 
555 extern int vmiodirenable;
556 
557 static __inline int
vn_canvmio(struct vnode * vp)558 vn_canvmio(struct vnode *vp)
559 {
560       if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
561 		return(TRUE);
562 	return(FALSE);
563 }
564 
565 /*
566  * Finally, include the default set of vnode operations.
567  */
568 #include "vnode_if.h"
569 
570 /* vn_open_flags */
571 #define	VN_OPEN_NOAUDIT		0x00000001
572 
573 /*
574  * Public vnode manipulation functions.
575  */
576 struct componentname;
577 struct file;
578 struct mount;
579 struct nameidata;
580 struct ostat;
581 struct thread;
582 struct proc;
583 struct stat;
584 struct nstat;
585 struct ucred;
586 struct uio;
587 struct vattr;
588 struct vnode;
589 
590 /* cache_* may belong in namei.h. */
591 #define	cache_enter(dvp, vp, cnp)					\
592 	cache_enter_time(dvp, vp, cnp, NULL, NULL)
593 void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
594 	    struct componentname *cnp, struct timespec *tsp,
595 	    struct timespec *dtsp);
596 #define	cache_lookup(dvp, vpp, cnp)					\
597 	cache_lookup_times(dvp, vpp, cnp, NULL, NULL)
598 int	cache_lookup_times(struct vnode *dvp, struct vnode **vpp,
599 	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
600 void	cache_purge(struct vnode *vp);
601 void	cache_purge_negative(struct vnode *vp);
602 void	cache_purgevfs(struct mount *mp);
603 int	change_dir(struct vnode *vp, struct thread *td);
604 int	change_root(struct vnode *vp, struct thread *td);
605 void	cvtstat(struct stat *st, struct ostat *ost);
606 void	cvtnstat(struct stat *sb, struct nstat *nsb);
607 int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
608 	    struct vnode **vpp);
609 void	getnewvnode_reserve(u_int count);
610 void	getnewvnode_drop_reserve(void);
611 int	insmntque1(struct vnode *vp, struct mount *mp,
612 	    void (*dtr)(struct vnode *, void *), void *dtr_arg);
613 int	insmntque(struct vnode *vp, struct mount *mp);
614 u_quad_t init_va_filerev(void);
615 int	speedup_syncer(void);
616 int	vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf,
617 	    u_int *buflen);
618 #define textvp_fullpath(p, rb, rfb) \
619 	vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
620 int	vn_fullpath(struct thread *td, struct vnode *vn,
621 	    char **retbuf, char **freebuf);
622 int	vn_fullpath_global(struct thread *td, struct vnode *vn,
623 	    char **retbuf, char **freebuf);
624 struct vnode *
625 	vn_dir_dd_ino(struct vnode *vp);
626 int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
627 int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
628 	    char *path, u_int pathlen);
629 int	vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
630 	    gid_t file_gid, accmode_t accmode, struct ucred *cred,
631 	    int *privused);
632 int	vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
633 	    struct acl *aclp, accmode_t accmode, struct ucred *cred,
634 	    int *privused);
635 int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
636 	    gid_t file_gid, struct acl *acl, accmode_t accmode,
637 	    struct ucred *cred, int *privused);
638 void	vattr_null(struct vattr *vap);
639 int	vcount(struct vnode *vp);
640 void	vdrop(struct vnode *);
641 void	vdropl(struct vnode *);
642 int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
643 int	vget(struct vnode *vp, int lockflag, struct thread *td);
644 void	vgone(struct vnode *vp);
645 void	vhold(struct vnode *);
646 void	vholdl(struct vnode *);
647 void	vinactive(struct vnode *, struct thread *);
648 int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
649 int	vtruncbuf(struct vnode *vp, struct ucred *cred, struct thread *td,
650 	    off_t length, int blksize);
651 void	vunref(struct vnode *);
652 void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
653 #define vprint(label, vp) vn_printf((vp), "%s\n", (label))
654 int	vrecycle(struct vnode *vp, struct thread *td);
655 int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
656 	    struct ucred *cred);
657 int	vn_close(struct vnode *vp,
658 	    int flags, struct ucred *file_cred, struct thread *td);
659 void	vn_finished_write(struct mount *mp);
660 void	vn_finished_secondary_write(struct mount *mp);
661 int	vn_isdisk(struct vnode *vp, int *errp);
662 int	_vn_lock(struct vnode *vp, int flags, char *file, int line);
663 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
664 int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
665 int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
666 	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
667 void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
668 int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
669 int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
670 	    int len, off_t offset, enum uio_seg segflg, int ioflg,
671 	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
672 	    struct thread *td);
673 int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
674 	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
675 	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
676 	    struct thread *td);
677 int	vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio,
678 	    const struct thread *td);
679 int	vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
680 	    struct ucred *file_cred, struct thread *td);
681 int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
682 int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
683 	    int flags);
684 int	vn_writechk(struct vnode *vp);
685 int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
686 	    const char *attrname, int *buflen, char *buf, struct thread *td);
687 int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
688 	    const char *attrname, int buflen, char *buf, struct thread *td);
689 int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
690 	    const char *attrname, struct thread *td);
691 int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
692 	    struct vnode **rvp);
693 
694 int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
695 int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
696 	    struct uio *uio);
697 
698 #define	vn_rangelock_unlock(vp, cookie)					\
699 	rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
700 #define	vn_rangelock_unlock_range(vp, cookie, start, end)		\
701 	rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), 	\
702 	    VI_MTX(vp))
703 #define	vn_rangelock_rlock(vp, start, end)				\
704 	rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
705 #define	vn_rangelock_wlock(vp, start, end)				\
706 	rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
707 
708 int	vfs_cache_lookup(struct vop_lookup_args *ap);
709 void	vfs_timestamp(struct timespec *);
710 void	vfs_write_resume(struct mount *mp);
711 void	vfs_write_resume_flags(struct mount *mp, int flags);
712 int	vfs_write_suspend(struct mount *mp);
713 int	vop_stdbmap(struct vop_bmap_args *);
714 int	vop_stdfsync(struct vop_fsync_args *);
715 int	vop_stdgetwritemount(struct vop_getwritemount_args *);
716 int	vop_stdgetpages(struct vop_getpages_args *);
717 int	vop_stdinactive(struct vop_inactive_args *);
718 int	vop_stdislocked(struct vop_islocked_args *);
719 int	vop_stdkqfilter(struct vop_kqfilter_args *);
720 int	vop_stdlock(struct vop_lock1_args *);
721 int	vop_stdputpages(struct vop_putpages_args *);
722 int	vop_stdunlock(struct vop_unlock_args *);
723 int	vop_nopoll(struct vop_poll_args *);
724 int	vop_stdaccess(struct vop_access_args *ap);
725 int	vop_stdaccessx(struct vop_accessx_args *ap);
726 int	vop_stdadvise(struct vop_advise_args *ap);
727 int	vop_stdadvlock(struct vop_advlock_args *ap);
728 int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
729 int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
730 int	vop_stdallocate(struct vop_allocate_args *ap);
731 int	vop_stdpathconf(struct vop_pathconf_args *);
732 int	vop_stdpoll(struct vop_poll_args *);
733 int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
734 int	vop_stdvptofh(struct vop_vptofh_args *ap);
735 int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
736 int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
737 int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
738 int	vop_eopnotsupp(struct vop_generic_args *ap);
739 int	vop_ebadf(struct vop_generic_args *ap);
740 int	vop_einval(struct vop_generic_args *ap);
741 int	vop_enoent(struct vop_generic_args *ap);
742 int	vop_enotty(struct vop_generic_args *ap);
743 int	vop_null(struct vop_generic_args *ap);
744 int	vop_panic(struct vop_generic_args *ap);
745 
746 /* These are called from within the actual VOPS. */
747 void	vop_create_post(void *a, int rc);
748 void	vop_deleteextattr_post(void *a, int rc);
749 void	vop_link_post(void *a, int rc);
750 void	vop_lock_pre(void *a);
751 void	vop_lock_post(void *a, int rc);
752 void	vop_lookup_post(void *a, int rc);
753 void	vop_lookup_pre(void *a);
754 void	vop_mkdir_post(void *a, int rc);
755 void	vop_mknod_post(void *a, int rc);
756 void	vop_remove_post(void *a, int rc);
757 void	vop_rename_post(void *a, int rc);
758 void	vop_rename_pre(void *a);
759 void	vop_rmdir_post(void *a, int rc);
760 void	vop_setattr_post(void *a, int rc);
761 void	vop_setextattr_post(void *a, int rc);
762 void	vop_strategy_pre(void *a);
763 void	vop_symlink_post(void *a, int rc);
764 void	vop_unlock_post(void *a, int rc);
765 void	vop_unlock_pre(void *a);
766 
767 void	vop_rename_fail(struct vop_rename_args *ap);
768 
769 #define	VOP_WRITE_PRE(ap)						\
770 	struct vattr va;						\
771 	int error;							\
772 	off_t osize, ooffset, noffset;					\
773 									\
774 	osize = ooffset = noffset = 0;					\
775 	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
776 		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
777 		if (error)						\
778 			return (error);					\
779 		ooffset = (ap)->a_uio->uio_offset;			\
780 		osize = (off_t)va.va_size;				\
781 	}
782 
783 #define VOP_WRITE_POST(ap, ret)						\
784 	noffset = (ap)->a_uio->uio_offset;				\
785 	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
786 		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
787 		    | (noffset > osize ? NOTE_EXTEND : 0));		\
788 	}
789 
790 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
791 
792 
793 void	vput(struct vnode *vp);
794 void	vrele(struct vnode *vp);
795 void	vref(struct vnode *vp);
796 int	vrefcnt(struct vnode *vp);
797 void 	v_addpollinfo(struct vnode *vp);
798 
799 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
800 void vnode_destroy_vobject(struct vnode *vp);
801 
802 extern struct vop_vector fifo_specops;
803 extern struct vop_vector dead_vnodeops;
804 extern struct vop_vector default_vnodeops;
805 
806 #define VOP_PANIC	((void*)(uintptr_t)vop_panic)
807 #define VOP_NULL	((void*)(uintptr_t)vop_null)
808 #define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
809 #define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
810 #define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
811 #define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
812 #define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
813 
814 /* vfs_hash.c */
815 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
816 
817 int vfs_hash_get(const struct mount *mp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
818 u_int vfs_hash_index(struct vnode *vp);
819 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
820 void vfs_hash_rehash(struct vnode *vp, u_int hash);
821 void vfs_hash_remove(struct vnode *vp);
822 
823 int vfs_kqfilter(struct vop_kqfilter_args *);
824 void vfs_mark_atime(struct vnode *vp, struct ucred *cred);
825 struct dirent;
826 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
827 
828 int vfs_unixify_accmode(accmode_t *accmode);
829 
830 void vfs_unp_reclaim(struct vnode *vp);
831 
832 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
833 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
834     gid_t gid);
835 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
836     struct thread *td);
837 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
838     struct thread *td);
839 
840 #endif /* _KERNEL */
841 
842 #endif /* !_SYS_VNODE_H_ */
843