vnode.h revision b819cea2f73f98c5662230cc9affc8cc84f77fcf
1/*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21
22/*
23 * Copyright (c) 1988, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Copyright (c) 2013, Joyent, Inc. All rights reserved.
25 */
26
27/*	Copyright (c) 1983, 1984, 1985, 1986, 1987, 1988, 1989 AT&T	*/
28/*	  All Rights Reserved  	*/
29
30/*
31 * University Copyright- Copyright (c) 1982, 1986, 1988
32 * The Regents of the University of California
33 * All Rights Reserved
34 *
35 * University Acknowledgment- Portions of this document are derived from
36 * software developed by the University of California, Berkeley, and its
37 * contributors.
38 */
39
40#ifndef _SYS_VNODE_H
41#define	_SYS_VNODE_H
42
43#include <sys/types.h>
44#include <sys/t_lock.h>
45#include <sys/time_impl.h>
46#include <sys/cred.h>
47#include <sys/uio.h>
48#include <sys/resource.h>
49#include <vm/seg_enum.h>
50#include <sys/kstat.h>
51#include <sys/kmem.h>
52#include <sys/list.h>
53#include <sys/avl.h>
54#ifdef	_KERNEL
55#include <sys/rwstlock.h>
56#include <sys/buf.h>
57#endif	/* _KERNEL */
58
59#ifdef	__cplusplus
60extern "C" {
61#endif
62
63/*
64 * Statistics for all vnode operations.
65 * All operations record number of ops (since boot/mount/zero'ed).
66 * Certain I/O operations (read, write, readdir) also record number
67 * of bytes transferred.
68 * This appears in two places in the system: one is embedded in each
69 * vfs_t.  There is also an array of vopstats_t structures allocated
70 * on a per-fstype basis.
71 */
72
73#define	VOPSTATS_STR	"vopstats_"	/* Initial string for vopstat kstats */
74
75typedef struct vopstats {
76	kstat_named_t	nopen;		/* VOP_OPEN */
77	kstat_named_t	nclose;		/* VOP_CLOSE */
78	kstat_named_t	nread;		/* VOP_READ */
79	kstat_named_t	read_bytes;
80	kstat_named_t	nwrite;		/* VOP_WRITE */
81	kstat_named_t	write_bytes;
82	kstat_named_t	nioctl;		/* VOP_IOCTL */
83	kstat_named_t	nsetfl;		/* VOP_SETFL */
84	kstat_named_t	ngetattr;	/* VOP_GETATTR */
85	kstat_named_t	nsetattr;	/* VOP_SETATTR */
86	kstat_named_t	naccess;	/* VOP_ACCESS */
87	kstat_named_t	nlookup;	/* VOP_LOOKUP */
88	kstat_named_t	ncreate;	/* VOP_CREATE */
89	kstat_named_t	nremove;	/* VOP_REMOVE */
90	kstat_named_t	nlink;		/* VOP_LINK */
91	kstat_named_t	nrename;	/* VOP_RENAME */
92	kstat_named_t	nmkdir;		/* VOP_MKDIR */
93	kstat_named_t	nrmdir;		/* VOP_RMDIR */
94	kstat_named_t	nreaddir;	/* VOP_READDIR */
95	kstat_named_t	readdir_bytes;
96	kstat_named_t	nsymlink;	/* VOP_SYMLINK */
97	kstat_named_t	nreadlink;	/* VOP_READLINK */
98	kstat_named_t	nfsync;		/* VOP_FSYNC */
99	kstat_named_t	ninactive;	/* VOP_INACTIVE */
100	kstat_named_t	nfid;		/* VOP_FID */
101	kstat_named_t	nrwlock;	/* VOP_RWLOCK */
102	kstat_named_t	nrwunlock;	/* VOP_RWUNLOCK */
103	kstat_named_t	nseek;		/* VOP_SEEK */
104	kstat_named_t	ncmp;		/* VOP_CMP */
105	kstat_named_t	nfrlock;	/* VOP_FRLOCK */
106	kstat_named_t	nspace;		/* VOP_SPACE */
107	kstat_named_t	nrealvp;	/* VOP_REALVP */
108	kstat_named_t	ngetpage;	/* VOP_GETPAGE */
109	kstat_named_t	nputpage;	/* VOP_PUTPAGE */
110	kstat_named_t	nmap;		/* VOP_MAP */
111	kstat_named_t	naddmap;	/* VOP_ADDMAP */
112	kstat_named_t	ndelmap;	/* VOP_DELMAP */
113	kstat_named_t	npoll;		/* VOP_POLL */
114	kstat_named_t	ndump;		/* VOP_DUMP */
115	kstat_named_t	npathconf;	/* VOP_PATHCONF */
116	kstat_named_t	npageio;	/* VOP_PAGEIO */
117	kstat_named_t	ndumpctl;	/* VOP_DUMPCTL */
118	kstat_named_t	ndispose;	/* VOP_DISPOSE */
119	kstat_named_t	nsetsecattr;	/* VOP_SETSECATTR */
120	kstat_named_t	ngetsecattr;	/* VOP_GETSECATTR */
121	kstat_named_t	nshrlock;	/* VOP_SHRLOCK */
122	kstat_named_t	nvnevent;	/* VOP_VNEVENT */
123	kstat_named_t	nreqzcbuf;	/* VOP_REQZCBUF */
124	kstat_named_t	nretzcbuf;	/* VOP_RETZCBUF */
125} vopstats_t;
126
127/*
128 * The vnode is the focus of all file activity in UNIX.
129 * A vnode is allocated for each active file, each current
130 * directory, each mounted-on file, and the root.
131 *
132 * Each vnode is usually associated with a file-system-specific node (for
133 * UFS, this is the in-memory inode).  Generally, a vnode and an fs-node
134 * should be created and destroyed together as a pair.
135 *
136 * If a vnode is reused for a new file, it should be reinitialized by calling
137 * either vn_reinit() or vn_recycle().
138 *
139 * vn_reinit() resets the entire vnode as if it was returned by vn_alloc().
140 * The caller is responsible for setting up the entire vnode after calling
141 * vn_reinit().  This is important when using kmem caching where the vnode is
142 * allocated by a constructor, for instance.
143 *
144 * vn_recycle() is used when the file system keeps some state around in both
145 * the vnode and the associated FS-node.  In UFS, for example, the inode of
146 * a deleted file can be reused immediately.  The v_data, v_vfsp, v_op, etc.
147 * remains the same but certain fields related to the previous instance need
148 * to be reset.  In particular:
149 *	v_femhead
150 *	v_path
151 *	v_rdcnt, v_wrcnt
152 *	v_mmap_read, v_mmap_write
153 */
154
155/*
156 * vnode types.  VNON means no type.  These values are unrelated to
157 * values in on-disk inodes.
158 */
159typedef enum vtype {
160	VNON	= 0,
161	VREG	= 1,
162	VDIR	= 2,
163	VBLK	= 3,
164	VCHR	= 4,
165	VLNK	= 5,
166	VFIFO	= 6,
167	VDOOR	= 7,
168	VPROC	= 8,
169	VSOCK	= 9,
170	VPORT	= 10,
171	VBAD	= 11
172} vtype_t;
173
174/*
175 * VSD - Vnode Specific Data
176 * Used to associate additional private data with a vnode.
177 */
178struct vsd_node {
179	list_node_t vs_nodes;		/* list of all VSD nodes */
180	uint_t vs_nkeys;		/* entries in value array */
181	void **vs_value;		/* array of value/key */
182};
183
184/*
185 * Many of the fields in the vnode are read-only once they are initialized
186 * at vnode creation time.  Other fields are protected by locks.
187 *
188 * IMPORTANT: vnodes should be created ONLY by calls to vn_alloc().  They
189 * may not be embedded into the file-system specific node (inode).  The
190 * size of vnodes may change.
191 *
192 * The v_lock protects:
193 *   v_flag
194 *   v_stream
195 *   v_count
196 *   v_shrlocks
197 *   v_path
198 *   v_vsd
199 *   v_xattrdir
200 *
201 * A special lock (implemented by vn_vfswlock in vnode.c) protects:
202 *   v_vfsmountedhere
203 *
204 * The global flock_lock mutex (in flock.c) protects:
205 *   v_filocks
206 *
207 * IMPORTANT NOTE:
208 *
209 *   The following vnode fields are considered public and may safely be
210 *   accessed by file systems or other consumers:
211 *
212 *     v_lock
213 *     v_flag
214 *     v_count
215 *     v_data
216 *     v_vfsp
217 *     v_stream
218 *     v_type
219 *     v_rdev
220 *
221 * ALL OTHER FIELDS SHOULD BE ACCESSED ONLY BY THE OWNER OF THAT FIELD.
222 * In particular, file systems should not access other fields; they may
223 * change or even be removed.  The functionality which was once provided
224 * by these fields is available through vn_* functions.
225 */
226
227struct fem_head;	/* from fem.h */
228
229typedef struct vnode {
230	kmutex_t	v_lock;		/* protects vnode fields */
231	uint_t		v_flag;		/* vnode flags (see below) */
232	uint_t		v_count;	/* reference count */
233	void		*v_data;	/* private data for fs */
234	struct vfs	*v_vfsp;	/* ptr to containing VFS */
235	struct stdata	*v_stream;	/* associated stream */
236	enum vtype	v_type;		/* vnode type */
237	dev_t		v_rdev;		/* device (VCHR, VBLK) */
238
239	/* PRIVATE FIELDS BELOW - DO NOT USE */
240
241	krwlock_t	v_nbllock;	/* sync for NBMAND locks */
242	char		*v_path;	/* cached path */
243	uint_t		v_rdcnt;	/* open for read count  (VREG only) */
244	uint_t		v_wrcnt;	/* open for write count (VREG only) */
245	struct vnode	*v_xattrdir;	/* unnamed extended attr dir (GFS) */
246
247	/* Private to the fake vnode impl. */
248
249	int		v_fd;
250	dev_t		v_st_dev;
251	ino_t		v_st_ino;
252	avl_node_t	v_avl_node;
253} vnode_t;
254
255#define	IS_DEVVP(vp)	\
256	((vp)->v_type == VCHR || (vp)->v_type == VBLK || (vp)->v_type == VFIFO)
257
258#define	VNODE_ALIGN	64
259/* Count of low-order 0 bits in a vnode *, based on size and alignment. */
260#if defined(_LP64)
261#define	VNODE_ALIGN_LOG2	8
262#else
263#define	VNODE_ALIGN_LOG2	7
264#endif
265
266/*
267 * vnode flags.
268 */
269#define	VROOT		0x01	/* root of its file system */
270#define	VNOCACHE	0x02	/* don't keep cache pages on vnode */
271#define	VNOMAP		0x04	/* file cannot be mapped/faulted */
272#define	VDUP		0x08	/* file should be dup'ed rather then opened */
273#define	VNOSWAP		0x10	/* file cannot be used as virtual swap device */
274#define	VNOMOUNT	0x20	/* file cannot be covered by mount */
275#define	VISSWAP		0x40	/* vnode is being used for swap */
276#define	VSWAPLIKE	0x80	/* vnode acts like swap (but may not be) */
277
278#define	IS_SWAPVP(vp)	(((vp)->v_flag & (VISSWAP | VSWAPLIKE)) != 0)
279
280#if defined(_KERNEL)
281typedef struct vn_vfslocks_entry {
282	rwstlock_t ve_lock;
283	void *ve_vpvfs;
284	struct vn_vfslocks_entry *ve_next;
285	uint32_t ve_refcnt;
286	char pad[64 - sizeof (rwstlock_t) - 2 * sizeof (void *) - \
287	    sizeof (uint32_t)];
288} vn_vfslocks_entry_t;
289#endif	/* _KERNEL */
290
291/*
292 * The following two flags are used to lock the v_vfsmountedhere field
293 */
294#define	VVFSLOCK	0x100
295#define	VVFSWAIT	0x200
296
297/*
298 * Used to serialize VM operations on a vnode
299 */
300#define	VVMLOCK		0x400
301
302/*
303 * Tell vn_open() not to fail a directory open for writing but
304 * to go ahead and call VOP_OPEN() to let the filesystem check.
305 */
306#define	VDIROPEN	0x800
307
308/*
309 * Flag to let the VM system know that this file is most likely a binary
310 * or shared library since it has been mmap()ed EXEC at some time.
311 */
312#define	VVMEXEC		0x1000
313
314#define	VPXFS		0x2000  /* clustering: global fs proxy vnode */
315
316#define	IS_PXFSVP(vp)	((vp)->v_flag & VPXFS)
317
318#define	V_XATTRDIR	0x4000	/* attribute unnamed directory */
319
320#define	IS_XATTRDIR(vp)	((vp)->v_flag & V_XATTRDIR)
321
322#define	V_LOCALITY	0x8000	/* whether locality aware */
323
324/*
325 * Flag that indicates the VM should maintain the v_pages list with all modified
326 * pages on one end and unmodified pages at the other. This makes finding dirty
327 * pages to write back to disk much faster at the expense of taking a minor
328 * fault on the first store instruction which touches a writable page.
329 */
330#define	VMODSORT	(0x10000)
331#define	IS_VMODSORT(vp) \
332	(pvn_vmodsort_supported != 0 && ((vp)->v_flag  & VMODSORT) != 0)
333
334#define	VISSWAPFS	0x20000	/* vnode is being used for swapfs */
335
336/*
337 * The mdb memstat command assumes that IS_SWAPFSVP only uses the
338 * vnode's v_flag field.  If this changes, cache the additional
339 * fields in mdb; see vn_get in mdb/common/modules/genunix/memory.c
340 */
341#define	IS_SWAPFSVP(vp)	(((vp)->v_flag & VISSWAPFS) != 0)
342
343#define	V_SYSATTR	0x40000	/* vnode is a GFS system attribute */
344
345/*
346 * Vnode attributes.  A bit-mask is supplied as part of the
347 * structure to indicate the attributes the caller wants to
348 * set (setattr) or extract (getattr).
349 */
350
351/*
352 * Note that va_nodeid and va_nblocks are 64bit data type.
353 * We support large files over NFSV3. With Solaris client and
354 * Server that generates 64bit ino's and sizes these fields
355 * will overflow if they are 32 bit sizes.
356 */
357
358typedef struct vattr {
359	uint_t		va_mask;	/* bit-mask of attributes */
360	vtype_t		va_type;	/* vnode type (for create) */
361	mode_t		va_mode;	/* file access mode */
362	uid_t		va_uid;		/* owner user id */
363	gid_t		va_gid;		/* owner group id */
364	dev_t		va_fsid;	/* file system id (dev for now) */
365	u_longlong_t	va_nodeid;	/* node id */
366	nlink_t		va_nlink;	/* number of references to file */
367	u_offset_t	va_size;	/* file size in bytes */
368	timestruc_t	va_atime;	/* time of last access */
369	timestruc_t	va_mtime;	/* time of last modification */
370	timestruc_t	va_ctime;	/* time of last status change */
371	dev_t		va_rdev;	/* device the file represents */
372	uint_t		va_blksize;	/* fundamental block size */
373	u_longlong_t	va_nblocks;	/* # of blocks allocated */
374	uint_t		va_seq;		/* sequence number */
375} vattr_t;
376
377#define	AV_SCANSTAMP_SZ	32		/* length of anti-virus scanstamp */
378
379/*
380 * Structure of all optional attributes.
381 */
382typedef struct xoptattr {
383	timestruc_t	xoa_createtime;	/* Create time of file */
384	uint8_t		xoa_archive;
385	uint8_t		xoa_system;
386	uint8_t		xoa_readonly;
387	uint8_t		xoa_hidden;
388	uint8_t		xoa_nounlink;
389	uint8_t		xoa_immutable;
390	uint8_t		xoa_appendonly;
391	uint8_t		xoa_nodump;
392	uint8_t		xoa_opaque;
393	uint8_t		xoa_av_quarantined;
394	uint8_t		xoa_av_modified;
395	uint8_t		xoa_av_scanstamp[AV_SCANSTAMP_SZ];
396	uint8_t		xoa_reparse;
397	uint64_t	xoa_generation;
398	uint8_t		xoa_offline;
399	uint8_t		xoa_sparse;
400} xoptattr_t;
401
402/*
403 * The xvattr structure is really a variable length structure that
404 * is made up of:
405 * - The classic vattr_t (xva_vattr)
406 * - a 32 bit quantity (xva_mapsize) that specifies the size of the
407 *   attribute bitmaps in 32 bit words.
408 * - A pointer to the returned attribute bitmap (needed because the
409 *   previous element, the requested attribute bitmap) is variable lenth.
410 * - The requested attribute bitmap, which is an array of 32 bit words.
411 *   Callers use the XVA_SET_REQ() macro to set the bits corresponding to
412 *   the attributes that are being requested.
413 * - The returned attribute bitmap, which is an array of 32 bit words.
414 *   File systems that support optional attributes use the XVA_SET_RTN()
415 *   macro to set the bits corresponding to the attributes that are being
416 *   returned.
417 * - The xoptattr_t structure which contains the attribute values
418 *
419 * xva_mapsize determines how many words in the attribute bitmaps.
420 * Immediately following the attribute bitmaps is the xoptattr_t.
421 * xva_getxoptattr() is used to get the pointer to the xoptattr_t
422 * section.
423 */
424
425#define	XVA_MAPSIZE	3		/* Size of attr bitmaps */
426#define	XVA_MAGIC	0x78766174	/* Magic # for verification */
427
428/*
429 * The xvattr structure is an extensible structure which permits optional
430 * attributes to be requested/returned.  File systems may or may not support
431 * optional attributes.  They do so at their own discretion but if they do
432 * support optional attributes, they must register the VFSFT_XVATTR feature
433 * so that the optional attributes can be set/retrived.
434 *
435 * The fields of the xvattr structure are:
436 *
437 * xva_vattr - The first element of an xvattr is a legacy vattr structure
438 * which includes the common attributes.  If AT_XVATTR is set in the va_mask
439 * then the entire structure is treated as an xvattr.  If AT_XVATTR is not
440 * set, then only the xva_vattr structure can be used.
441 *
442 * xva_magic - 0x78766174 (hex for "xvat"). Magic number for verification.
443 *
444 * xva_mapsize - Size of requested and returned attribute bitmaps.
445 *
446 * xva_rtnattrmapp - Pointer to xva_rtnattrmap[].  We need this since the
447 * size of the array before it, xva_reqattrmap[], could change which means
448 * the location of xva_rtnattrmap[] could change.  This will allow unbundled
449 * file systems to find the location of xva_rtnattrmap[] when the sizes change.
450 *
451 * xva_reqattrmap[] - Array of requested attributes.  Attributes are
452 * represented by a specific bit in a specific element of the attribute
453 * map array.  Callers set the bits corresponding to the attributes
454 * that the caller wants to get/set.
455 *
456 * xva_rtnattrmap[] - Array of attributes that the file system was able to
457 * process.  Not all file systems support all optional attributes.  This map
458 * informs the caller which attributes the underlying file system was able
459 * to set/get.  (Same structure as the requested attributes array in terms
460 * of each attribute  corresponding to specific bits and array elements.)
461 *
462 * xva_xoptattrs - Structure containing values of optional attributes.
463 * These values are only valid if the corresponding bits in xva_reqattrmap
464 * are set and the underlying file system supports those attributes.
465 */
466typedef struct xvattr {
467	vattr_t		xva_vattr;	/* Embedded vattr structure */
468	uint32_t	xva_magic;	/* Magic Number */
469	uint32_t	xva_mapsize;	/* Size of attr bitmap (32-bit words) */
470	uint32_t	*xva_rtnattrmapp;	/* Ptr to xva_rtnattrmap[] */
471	uint32_t	xva_reqattrmap[XVA_MAPSIZE];	/* Requested attrs */
472	uint32_t	xva_rtnattrmap[XVA_MAPSIZE];	/* Returned attrs */
473	xoptattr_t	xva_xoptattrs;	/* Optional attributes */
474} xvattr_t;
475
476#ifdef _SYSCALL32
477/*
478 * For bigtypes time_t changed to 64 bit on the 64-bit kernel.
479 * Define an old version for user/kernel interface
480 */
481
482#if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
483#pragma pack(4)
484#endif
485
486typedef struct vattr32 {
487	uint32_t	va_mask;	/* bit-mask of attributes */
488	vtype_t		va_type;	/* vnode type (for create) */
489	mode32_t	va_mode;	/* file access mode */
490	uid32_t		va_uid;		/* owner user id */
491	gid32_t		va_gid;		/* owner group id */
492	dev32_t		va_fsid;	/* file system id (dev for now) */
493	u_longlong_t	va_nodeid;	/* node id */
494	nlink_t		va_nlink;	/* number of references to file */
495	u_offset_t	va_size;	/* file size in bytes */
496	timestruc32_t	va_atime;	/* time of last access */
497	timestruc32_t	va_mtime;	/* time of last modification */
498	timestruc32_t	va_ctime;	/* time of last status change */
499	dev32_t		va_rdev;	/* device the file represents */
500	uint32_t	va_blksize;	/* fundamental block size */
501	u_longlong_t	va_nblocks;	/* # of blocks allocated */
502	uint32_t	va_seq;		/* sequence number */
503} vattr32_t;
504
505#if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
506#pragma pack()
507#endif
508
509#else  /* not _SYSCALL32 */
510#define	vattr32		vattr
511typedef vattr_t		vattr32_t;
512#endif /* _SYSCALL32 */
513
514/*
515 * Attributes of interest to the caller of setattr or getattr.
516 */
517#define	AT_TYPE		0x00001
518#define	AT_MODE		0x00002
519#define	AT_UID		0x00004
520#define	AT_GID		0x00008
521#define	AT_FSID		0x00010
522#define	AT_NODEID	0x00020
523#define	AT_NLINK	0x00040
524#define	AT_SIZE		0x00080
525#define	AT_ATIME	0x00100
526#define	AT_MTIME	0x00200
527#define	AT_CTIME	0x00400
528#define	AT_RDEV		0x00800
529#define	AT_BLKSIZE	0x01000
530#define	AT_NBLOCKS	0x02000
531/*			0x04000 */	/* unused */
532#define	AT_SEQ		0x08000
533/*
534 * If AT_XVATTR is set then there are additional bits to process in
535 * the xvattr_t's attribute bitmap.  If this is not set then the bitmap
536 * MUST be ignored.  Note that this bit must be set/cleared explicitly.
537 * That is, setting AT_ALL will NOT set AT_XVATTR.
538 */
539#define	AT_XVATTR	0x10000
540
541#define	AT_ALL		(AT_TYPE|AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|\
542			AT_NLINK|AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|\
543			AT_RDEV|AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
544
545#define	AT_STAT		(AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|AT_NLINK|\
546			AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|AT_RDEV|AT_TYPE)
547
548#define	AT_TIMES	(AT_ATIME|AT_MTIME|AT_CTIME)
549
550#define	AT_NOSET	(AT_NLINK|AT_RDEV|AT_FSID|AT_NODEID|AT_TYPE|\
551			AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
552
553/*
554 * Attribute bits used in the extensible attribute's (xva's) attribute
555 * bitmaps.  Note that the bitmaps are made up of a variable length number
556 * of 32-bit words.  The convention is to use XAT{n}_{attrname} where "n"
557 * is the element in the bitmap (starting at 1).  This convention is for
558 * the convenience of the maintainer to keep track of which element each
559 * attribute belongs to.
560 *
561 * NOTE THAT CONSUMERS MUST *NOT* USE THE XATn_* DEFINES DIRECTLY.  CONSUMERS
562 * MUST USE THE XAT_* DEFINES.
563 */
564#define	XAT0_INDEX	0LL		/* Index into bitmap for XAT0 attrs */
565#define	XAT0_CREATETIME	0x00000001	/* Create time of file */
566#define	XAT0_ARCHIVE	0x00000002	/* Archive */
567#define	XAT0_SYSTEM	0x00000004	/* System */
568#define	XAT0_READONLY	0x00000008	/* Readonly */
569#define	XAT0_HIDDEN	0x00000010	/* Hidden */
570#define	XAT0_NOUNLINK	0x00000020	/* Nounlink */
571#define	XAT0_IMMUTABLE	0x00000040	/* immutable */
572#define	XAT0_APPENDONLY	0x00000080	/* appendonly */
573#define	XAT0_NODUMP	0x00000100	/* nodump */
574#define	XAT0_OPAQUE	0x00000200	/* opaque */
575#define	XAT0_AV_QUARANTINED	0x00000400	/* anti-virus quarantine */
576#define	XAT0_AV_MODIFIED	0x00000800	/* anti-virus modified */
577#define	XAT0_AV_SCANSTAMP	0x00001000	/* anti-virus scanstamp */
578#define	XAT0_REPARSE	0x00002000	/* FS reparse point */
579#define	XAT0_GEN	0x00004000	/* object generation number */
580#define	XAT0_OFFLINE	0x00008000	/* offline */
581#define	XAT0_SPARSE	0x00010000	/* sparse */
582
583#define	XAT0_ALL_ATTRS	(XAT0_CREATETIME|XAT0_ARCHIVE|XAT0_SYSTEM| \
584    XAT0_READONLY|XAT0_HIDDEN|XAT0_NOUNLINK|XAT0_IMMUTABLE|XAT0_APPENDONLY| \
585    XAT0_NODUMP|XAT0_OPAQUE|XAT0_AV_QUARANTINED|  XAT0_AV_MODIFIED| \
586    XAT0_AV_SCANSTAMP|XAT0_REPARSE|XATO_GEN|XAT0_OFFLINE|XAT0_SPARSE)
587
588/* Support for XAT_* optional attributes */
589#define	XVA_MASK		0xffffffff	/* Used to mask off 32 bits */
590#define	XVA_SHFT		32		/* Used to shift index */
591
592/*
593 * Used to pry out the index and attribute bits from the XAT_* attributes
594 * defined below.  Note that we're masking things down to 32 bits then
595 * casting to uint32_t.
596 */
597#define	XVA_INDEX(attr)		((uint32_t)(((attr) >> XVA_SHFT) & XVA_MASK))
598#define	XVA_ATTRBIT(attr)	((uint32_t)((attr) & XVA_MASK))
599
600/*
601 * The following defines present a "flat namespace" so that consumers don't
602 * need to keep track of which element belongs to which bitmap entry.
603 *
604 * NOTE THAT THESE MUST NEVER BE OR-ed TOGETHER
605 */
606#define	XAT_CREATETIME		((XAT0_INDEX << XVA_SHFT) | XAT0_CREATETIME)
607#define	XAT_ARCHIVE		((XAT0_INDEX << XVA_SHFT) | XAT0_ARCHIVE)
608#define	XAT_SYSTEM		((XAT0_INDEX << XVA_SHFT) | XAT0_SYSTEM)
609#define	XAT_READONLY		((XAT0_INDEX << XVA_SHFT) | XAT0_READONLY)
610#define	XAT_HIDDEN		((XAT0_INDEX << XVA_SHFT) | XAT0_HIDDEN)
611#define	XAT_NOUNLINK		((XAT0_INDEX << XVA_SHFT) | XAT0_NOUNLINK)
612#define	XAT_IMMUTABLE		((XAT0_INDEX << XVA_SHFT) | XAT0_IMMUTABLE)
613#define	XAT_APPENDONLY		((XAT0_INDEX << XVA_SHFT) | XAT0_APPENDONLY)
614#define	XAT_NODUMP		((XAT0_INDEX << XVA_SHFT) | XAT0_NODUMP)
615#define	XAT_OPAQUE		((XAT0_INDEX << XVA_SHFT) | XAT0_OPAQUE)
616#define	XAT_AV_QUARANTINED	((XAT0_INDEX << XVA_SHFT) | XAT0_AV_QUARANTINED)
617#define	XAT_AV_MODIFIED		((XAT0_INDEX << XVA_SHFT) | XAT0_AV_MODIFIED)
618#define	XAT_AV_SCANSTAMP	((XAT0_INDEX << XVA_SHFT) | XAT0_AV_SCANSTAMP)
619#define	XAT_REPARSE		((XAT0_INDEX << XVA_SHFT) | XAT0_REPARSE)
620#define	XAT_GEN			((XAT0_INDEX << XVA_SHFT) | XAT0_GEN)
621#define	XAT_OFFLINE		((XAT0_INDEX << XVA_SHFT) | XAT0_OFFLINE)
622#define	XAT_SPARSE		((XAT0_INDEX << XVA_SHFT) | XAT0_SPARSE)
623
624/*
625 * The returned attribute map array (xva_rtnattrmap[]) is located past the
626 * requested attribute map array (xva_reqattrmap[]).  Its location changes
627 * when the array sizes change.  We use a separate pointer in a known location
628 * (xva_rtnattrmapp) to hold the location of xva_rtnattrmap[].  This is
629 * set in xva_init()
630 */
631#define	XVA_RTNATTRMAP(xvap)	((xvap)->xva_rtnattrmapp)
632
633/*
634 * XVA_SET_REQ() sets an attribute bit in the proper element in the bitmap
635 * of requested attributes (xva_reqattrmap[]).
636 */
637#define	XVA_SET_REQ(xvap, attr)					\
638	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
639	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
640	(xvap)->xva_reqattrmap[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
641/*
642 * XVA_CLR_REQ() clears an attribute bit in the proper element in the bitmap
643 * of requested attributes (xva_reqattrmap[]).
644 */
645#define	XVA_CLR_REQ(xvap, attr)					\
646	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
647	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
648	(xvap)->xva_reqattrmap[XVA_INDEX(attr)] &= ~XVA_ATTRBIT(attr)
649
650/*
651 * XVA_SET_RTN() sets an attribute bit in the proper element in the bitmap
652 * of returned attributes (xva_rtnattrmap[]).
653 */
654#define	XVA_SET_RTN(xvap, attr)					\
655	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
656	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
657	(XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
658
659/*
660 * XVA_ISSET_REQ() checks the requested attribute bitmap (xva_reqattrmap[])
661 * to see of the corresponding attribute bit is set.  If so, returns non-zero.
662 */
663#define	XVA_ISSET_REQ(xvap, attr)					\
664	((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&			\
665		((xvap)->xva_magic == XVA_MAGIC) &&			\
666		((xvap)->xva_mapsize > XVA_INDEX(attr))) ?		\
667	((xvap)->xva_reqattrmap[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) :	0)
668
669/*
670 * XVA_ISSET_RTN() checks the returned attribute bitmap (xva_rtnattrmap[])
671 * to see of the corresponding attribute bit is set.  If so, returns non-zero.
672 */
673#define	XVA_ISSET_RTN(xvap, attr)					\
674	((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&			\
675		((xvap)->xva_magic == XVA_MAGIC) &&			\
676		((xvap)->xva_mapsize > XVA_INDEX(attr))) ?		\
677	((XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) : 0)
678
679/*
680 *  Modes.  Some values same as S_xxx entries from stat.h for convenience.
681 */
682#define	VSUID		04000		/* set user id on execution */
683#define	VSGID		02000		/* set group id on execution */
684#define	VSVTX		01000		/* save swapped text even after use */
685
686/*
687 * Permissions.
688 */
689#define	VREAD		00400
690#define	VWRITE		00200
691#define	VEXEC		00100
692
693#define	MODEMASK	07777		/* mode bits plus permission bits */
694#define	PERMMASK	00777		/* permission bits */
695
696/*
697 * VOP_ACCESS flags
698 */
699#define	V_ACE_MASK	0x1	/* mask represents  NFSv4 ACE permissions */
700#define	V_APPEND	0x2	/* want to do append only check */
701
702/*
703 * Check whether mandatory file locking is enabled.
704 */
705
706#define	MANDMODE(mode)		(((mode) & (VSGID|(VEXEC>>3))) == VSGID)
707#define	MANDLOCK(vp, mode)	((vp)->v_type == VREG && MANDMODE(mode))
708
709/*
710 * Flags for vnode operations.
711 */
712enum rm		{ RMFILE, RMDIRECTORY };	/* rm or rmdir (remove) */
713enum symfollow	{ NO_FOLLOW, FOLLOW };		/* follow symlinks (or not) */
714enum vcexcl	{ NONEXCL, EXCL };		/* (non)excl create */
715enum create	{ CRCREAT, CRMKNOD, CRMKDIR };	/* reason for create */
716
717typedef enum rm		rm_t;
718typedef enum symfollow	symfollow_t;
719typedef enum vcexcl	vcexcl_t;
720typedef enum create	create_t;
721
722/* Vnode Events - Used by VOP_VNEVENT */
723typedef enum vnevent	{
724	VE_SUPPORT	= 0,	/* Query */
725	VE_RENAME_SRC	= 1,	/* Rename, with vnode as source */
726	VE_RENAME_DEST	= 2,	/* Rename, with vnode as target/destination */
727	VE_REMOVE	= 3,	/* Remove of vnode's name */
728	VE_RMDIR	= 4,	/* Remove of directory vnode's name */
729	VE_CREATE	= 5,	/* Create with vnode's name which exists */
730	VE_LINK		= 6, 	/* Link with vnode's name as source */
731	VE_RENAME_DEST_DIR	= 7, 	/* Rename with vnode as target dir */
732	VE_MOUNTEDOVER	= 8, 	/* File or Filesystem got mounted over vnode */
733	VE_TRUNCATE = 9		/* Truncate */
734} vnevent_t;
735
736/*
737 * Values for checking vnode open and map counts
738 */
739enum v_mode { V_READ, V_WRITE, V_RDORWR, V_RDANDWR };
740
741typedef enum v_mode v_mode_t;
742
743#define	V_TRUE	1
744#define	V_FALSE	0
745
746/*
747 * Structure used on VOP_GETSECATTR and VOP_SETSECATTR operations
748 */
749
750typedef struct vsecattr {
751	uint_t		vsa_mask;	/* See below */
752	int		vsa_aclcnt;	/* ACL entry count */
753	void		*vsa_aclentp;	/* pointer to ACL entries */
754	int		vsa_dfaclcnt;	/* default ACL entry count */
755	void		*vsa_dfaclentp;	/* pointer to default ACL entries */
756	size_t		vsa_aclentsz;	/* ACE size in bytes of vsa_aclentp */
757	uint_t		vsa_aclflags;	/* ACE ACL flags */
758} vsecattr_t;
759
760/* vsa_mask values */
761#define	VSA_ACL			0x0001
762#define	VSA_ACLCNT		0x0002
763#define	VSA_DFACL		0x0004
764#define	VSA_DFACLCNT		0x0008
765#define	VSA_ACE			0x0010
766#define	VSA_ACECNT		0x0020
767#define	VSA_ACE_ALLTYPES	0x0040
768#define	VSA_ACE_ACLFLAGS	0x0080	/* get/set ACE ACL flags */
769
770/*
771 * Structure used by various vnode operations to determine
772 * the context (pid, host, identity) of a caller.
773 *
774 * The cc_caller_id is used to identify one or more callers who invoke
775 * operations, possibly on behalf of others.  For example, the NFS
776 * server could have it's own cc_caller_id which can be detected by
777 * vnode/vfs operations or (FEM) monitors on those operations.  New
778 * caller IDs are generated by fs_new_caller_id().
779 */
780typedef struct caller_context {
781	pid_t		cc_pid;		/* Process ID of the caller */
782	int		cc_sysid;	/* System ID, used for remote calls */
783	u_longlong_t	cc_caller_id;	/* Identifier for (set of) caller(s) */
784	ulong_t		cc_flags;
785} caller_context_t;
786
787/*
788 * Flags for caller context.  The caller sets CC_DONTBLOCK if it does not
789 * want to block inside of a FEM monitor.  The monitor will set CC_WOULDBLOCK
790 * and return EAGAIN if the operation would have blocked.
791 */
792#define	CC_WOULDBLOCK	0x01
793#define	CC_DONTBLOCK	0x02
794
795/*
796 * Structure tags for function prototypes, defined elsewhere.
797 */
798struct pathname;
799struct fid;
800struct flock64;
801struct flk_callback;
802struct shrlock;
803struct page;
804struct seg;
805struct as;
806struct pollhead;
807struct taskq;
808
809#if defined(_KERNEL) || defined(_FAKE_KERNEL)
810
811/*
812 * VNODE_OPS defines all the vnode operations.  It is used to define
813 * the vnodeops structure (below) and the fs_func_p union (vfs_opreg.h).
814 */
815#define	VNODE_OPS							\
816	int	(*vop_open)(vnode_t **, int, cred_t *,			\
817				caller_context_t *);			\
818	int	(*vop_close)(vnode_t *, int, int, offset_t, cred_t *,	\
819				caller_context_t *);			\
820	int	(*vop_read)(vnode_t *, uio_t *, int, cred_t *,		\
821				caller_context_t *);			\
822	int	(*vop_write)(vnode_t *, uio_t *, int, cred_t *,		\
823				caller_context_t *);			\
824	int	(*vop_ioctl)(vnode_t *, int, intptr_t, int, cred_t *,	\
825				int *, caller_context_t *);		\
826	int	(*vop_setfl)(vnode_t *, int, int, cred_t *,		\
827				caller_context_t *);			\
828	int	(*vop_getattr)(vnode_t *, vattr_t *, int, cred_t *,	\
829				caller_context_t *);			\
830	int	(*vop_setattr)(vnode_t *, vattr_t *, int, cred_t *,	\
831				caller_context_t *);			\
832	int	(*vop_access)(vnode_t *, int, int, cred_t *,		\
833				caller_context_t *);			\
834	int	(*vop_lookup)(vnode_t *, char *, vnode_t **,		\
835				struct pathname *,			\
836				int, vnode_t *, cred_t *,		\
837				caller_context_t *, int *,		\
838				struct pathname *);			\
839	int	(*vop_create)(vnode_t *, char *, vattr_t *, vcexcl_t,	\
840				int, vnode_t **, cred_t *, int,		\
841				caller_context_t *, vsecattr_t *);	\
842	int	(*vop_remove)(vnode_t *, char *, cred_t *,		\
843				caller_context_t *, int);		\
844	int	(*vop_link)(vnode_t *, vnode_t *, char *, cred_t *,	\
845				caller_context_t *, int);		\
846	int	(*vop_rename)(vnode_t *, char *, vnode_t *, char *,	\
847				cred_t *, caller_context_t *, int);	\
848	int	(*vop_mkdir)(vnode_t *, char *, vattr_t *, vnode_t **,	\
849				cred_t *, caller_context_t *, int,	\
850				vsecattr_t *);				\
851	int	(*vop_rmdir)(vnode_t *, char *, vnode_t *, cred_t *,	\
852				caller_context_t *, int);		\
853	int	(*vop_readdir)(vnode_t *, uio_t *, cred_t *, int *,	\
854				caller_context_t *, int);		\
855	int	(*vop_symlink)(vnode_t *, char *, vattr_t *, char *,	\
856				cred_t *, caller_context_t *, int);	\
857	int	(*vop_readlink)(vnode_t *, uio_t *, cred_t *,		\
858				caller_context_t *);			\
859	int	(*vop_fsync)(vnode_t *, int, cred_t *,			\
860				caller_context_t *);			\
861	void	(*vop_inactive)(vnode_t *, cred_t *,			\
862				caller_context_t *);			\
863	int	(*vop_fid)(vnode_t *, struct fid *,			\
864				caller_context_t *);			\
865	int	(*vop_rwlock)(vnode_t *, int, caller_context_t *);	\
866	void	(*vop_rwunlock)(vnode_t *, int, caller_context_t *);	\
867	int	(*vop_seek)(vnode_t *, offset_t, offset_t *,		\
868				caller_context_t *);			\
869	int	(*vop_cmp)(vnode_t *, vnode_t *, caller_context_t *);	\
870	int	(*vop_frlock)(vnode_t *, int, struct flock64 *,		\
871				int, offset_t,				\
872				struct flk_callback *, cred_t *,	\
873				caller_context_t *);			\
874	int	(*vop_space)(vnode_t *, int, struct flock64 *,		\
875				int, offset_t,				\
876				cred_t *, caller_context_t *);		\
877	int	(*vop_realvp)(vnode_t *, vnode_t **,			\
878				caller_context_t *);			\
879	int	(*vop_getpage)(vnode_t *, offset_t, size_t, uint_t *,	\
880				struct page **, size_t, struct seg *,	\
881				caddr_t, enum seg_rw, cred_t *,		\
882				caller_context_t *);			\
883	int	(*vop_putpage)(vnode_t *, offset_t, size_t,		\
884				int, cred_t *, caller_context_t *);	\
885	int	(*vop_map)(vnode_t *, offset_t, struct as *,		\
886				caddr_t *, size_t,			\
887				uchar_t, uchar_t, uint_t, cred_t *,	\
888				caller_context_t *);			\
889	int	(*vop_addmap)(vnode_t *, offset_t, struct as *,		\
890				caddr_t, size_t,			\
891				uchar_t, uchar_t, uint_t, cred_t *,	\
892				caller_context_t *);			\
893	int	(*vop_delmap)(vnode_t *, offset_t, struct as *,		\
894				caddr_t, size_t,			\
895				uint_t, uint_t, uint_t, cred_t *,	\
896				caller_context_t *);			\
897	int	(*vop_poll)(vnode_t *, short, int, short *,		\
898				struct pollhead **,			\
899				caller_context_t *);			\
900	int	(*vop_dump)(vnode_t *, caddr_t, offset_t, offset_t,	\
901				caller_context_t *);			\
902	int	(*vop_pathconf)(vnode_t *, int, ulong_t *, cred_t *,	\
903				caller_context_t *);			\
904	int	(*vop_pageio)(vnode_t *, struct page *,			\
905				u_offset_t, size_t, int, cred_t *,	\
906				caller_context_t *);			\
907	int	(*vop_dumpctl)(vnode_t *, int, offset_t *,		\
908				caller_context_t *);			\
909	void	(*vop_dispose)(vnode_t *, struct page *,		\
910				int, int, cred_t *,			\
911				caller_context_t *);			\
912	int	(*vop_setsecattr)(vnode_t *, vsecattr_t *,		\
913				int, cred_t *, caller_context_t *);	\
914	int	(*vop_getsecattr)(vnode_t *, vsecattr_t *,		\
915				int, cred_t *, caller_context_t *);	\
916	int	(*vop_shrlock)(vnode_t *, int, struct shrlock *,	\
917				int, cred_t *, caller_context_t *);	\
918	int	(*vop_vnevent)(vnode_t *, vnevent_t, vnode_t *,		\
919				char *, caller_context_t *);		\
920	int	(*vop_reqzcbuf)(vnode_t *, enum uio_rw, xuio_t *,	\
921				cred_t *, caller_context_t *);		\
922	int	(*vop_retzcbuf)(vnode_t *, xuio_t *, cred_t *,		\
923				caller_context_t *)
924	/* NB: No ";" */
925
926/*
927 * Operations on vnodes.  Note: File systems must never operate directly
928 * on a 'vnodeops' structure -- it WILL change in future releases!  They
929 * must use vn_make_ops() to create the structure.
930 */
931typedef struct vnodeops {
932	const char *vnop_name;
933	VNODE_OPS;	/* Signatures of all vnode operations (vops) */
934} vnodeops_t;
935
936typedef int (*fs_generic_func_p) ();	/* Generic vop/vfsop/femop/fsemop ptr */
937
938extern int	fop_open(vnode_t **, int, cred_t *, caller_context_t *);
939extern int	fop_close(vnode_t *, int, int, offset_t, cred_t *,
940				caller_context_t *);
941extern int	fop_read(vnode_t *, uio_t *, int, cred_t *, caller_context_t *);
942extern int	fop_write(vnode_t *, uio_t *, int, cred_t *,
943				caller_context_t *);
944extern int	fop_ioctl(vnode_t *, int, intptr_t, int, cred_t *, int *,
945				caller_context_t *);
946extern int	fop_setfl(vnode_t *, int, int, cred_t *, caller_context_t *);
947extern int	fop_getattr(vnode_t *, vattr_t *, int, cred_t *,
948				caller_context_t *);
949extern int	fop_setattr(vnode_t *, vattr_t *, int, cred_t *,
950				caller_context_t *);
951extern int	fop_access(vnode_t *, int, int, cred_t *, caller_context_t *);
952extern int	fop_lookup(vnode_t *, char *, vnode_t **, struct pathname *,
953				int, vnode_t *, cred_t *, caller_context_t *,
954				int *, struct pathname *);
955extern int	fop_create(vnode_t *, char *, vattr_t *, vcexcl_t, int,
956				vnode_t **, cred_t *, int, caller_context_t *,
957				vsecattr_t *);
958extern int	fop_remove(vnode_t *vp, char *, cred_t *, caller_context_t *,
959				int);
960extern int	fop_link(vnode_t *, vnode_t *, char *, cred_t *,
961				caller_context_t *, int);
962extern int	fop_rename(vnode_t *, char *, vnode_t *, char *, cred_t *,
963				caller_context_t *, int);
964extern int	fop_mkdir(vnode_t *, char *, vattr_t *, vnode_t **, cred_t *,
965				caller_context_t *, int, vsecattr_t *);
966extern int	fop_rmdir(vnode_t *, char *, vnode_t *, cred_t *,
967				caller_context_t *, int);
968extern int	fop_readdir(vnode_t *, uio_t *, cred_t *, int *,
969				caller_context_t *, int);
970extern int	fop_symlink(vnode_t *, char *, vattr_t *, char *, cred_t *,
971				caller_context_t *, int);
972extern int	fop_readlink(vnode_t *, uio_t *, cred_t *, caller_context_t *);
973extern int	fop_fsync(vnode_t *, int, cred_t *, caller_context_t *);
974extern void	fop_inactive(vnode_t *, cred_t *, caller_context_t *);
975extern int	fop_fid(vnode_t *, struct fid *, caller_context_t *);
976extern int	fop_rwlock(vnode_t *, int, caller_context_t *);
977extern void	fop_rwunlock(vnode_t *, int, caller_context_t *);
978extern int	fop_seek(vnode_t *, offset_t, offset_t *, caller_context_t *);
979extern int	fop_cmp(vnode_t *, vnode_t *, caller_context_t *);
980extern int	fop_frlock(vnode_t *, int, struct flock64 *, int, offset_t,
981				struct flk_callback *, cred_t *,
982				caller_context_t *);
983extern int	fop_space(vnode_t *, int, struct flock64 *, int, offset_t,
984				cred_t *, caller_context_t *);
985extern int	fop_realvp(vnode_t *, vnode_t **, caller_context_t *);
986extern int	fop_getpage(vnode_t *, offset_t, size_t, uint_t *,
987				struct page **, size_t, struct seg *,
988				caddr_t, enum seg_rw, cred_t *,
989				caller_context_t *);
990extern int	fop_putpage(vnode_t *, offset_t, size_t, int, cred_t *,
991				caller_context_t *);
992extern int	fop_map(vnode_t *, offset_t, struct as *, caddr_t *, size_t,
993				uchar_t, uchar_t, uint_t, cred_t *cr,
994				caller_context_t *);
995extern int	fop_addmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
996				uchar_t, uchar_t, uint_t, cred_t *,
997				caller_context_t *);
998extern int	fop_delmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
999				uint_t, uint_t, uint_t, cred_t *,
1000				caller_context_t *);
1001extern int	fop_poll(vnode_t *, short, int, short *, struct pollhead **,
1002				caller_context_t *);
1003extern int	fop_dump(vnode_t *, caddr_t, offset_t, offset_t,
1004    caller_context_t *);
1005extern int	fop_pathconf(vnode_t *, int, ulong_t *, cred_t *,
1006				caller_context_t *);
1007extern int	fop_pageio(vnode_t *, struct page *, u_offset_t, size_t, int,
1008				cred_t *, caller_context_t *);
1009extern int	fop_dumpctl(vnode_t *, int, offset_t *, caller_context_t *);
1010extern void	fop_dispose(vnode_t *, struct page *, int, int, cred_t *,
1011				caller_context_t *);
1012extern int	fop_setsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
1013				caller_context_t *);
1014extern int	fop_getsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
1015				caller_context_t *);
1016extern int	fop_shrlock(vnode_t *, int, struct shrlock *, int, cred_t *,
1017				caller_context_t *);
1018extern int	fop_vnevent(vnode_t *, vnevent_t, vnode_t *, char *,
1019				caller_context_t *);
1020extern int	fop_reqzcbuf(vnode_t *, enum uio_rw, xuio_t *, cred_t *,
1021				caller_context_t *);
1022extern int	fop_retzcbuf(vnode_t *, xuio_t *, cred_t *, caller_context_t *);
1023
1024#endif	/* _KERNEL */
1025
1026#define	VOP_OPEN(vpp, mode, cr, ct) \
1027	fop_open(vpp, mode, cr, ct)
1028#define	VOP_CLOSE(vp, f, c, o, cr, ct) \
1029	fop_close(vp, f, c, o, cr, ct)
1030#define	VOP_READ(vp, uiop, iof, cr, ct) \
1031	fop_read(vp, uiop, iof, cr, ct)
1032#define	VOP_WRITE(vp, uiop, iof, cr, ct) \
1033	fop_write(vp, uiop, iof, cr, ct)
1034#define	VOP_IOCTL(vp, cmd, a, f, cr, rvp, ct) \
1035	fop_ioctl(vp, cmd, a, f, cr, rvp, ct)
1036#define	VOP_SETFL(vp, f, a, cr, ct) \
1037	fop_setfl(vp, f, a, cr, ct)
1038#define	VOP_GETATTR(vp, vap, f, cr, ct) \
1039	fop_getattr(vp, vap, f, cr, ct)
1040#define	VOP_SETATTR(vp, vap, f, cr, ct) \
1041	fop_setattr(vp, vap, f, cr, ct)
1042#define	VOP_ACCESS(vp, mode, f, cr, ct) \
1043	fop_access(vp, mode, f, cr, ct)
1044#define	VOP_LOOKUP(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp) \
1045	fop_lookup(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp)
1046#define	VOP_CREATE(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap) \
1047	fop_create(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap)
1048#define	VOP_REMOVE(dvp, p, cr, ct, f) \
1049	fop_remove(dvp, p, cr, ct, f)
1050#define	VOP_LINK(tdvp, fvp, p, cr, ct, f) \
1051	fop_link(tdvp, fvp, p, cr, ct, f)
1052#define	VOP_RENAME(fvp, fnm, tdvp, tnm, cr, ct, f) \
1053	fop_rename(fvp, fnm, tdvp, tnm, cr, ct, f)
1054#define	VOP_MKDIR(dp, p, vap, vpp, cr, ct, f, vsap) \
1055	fop_mkdir(dp, p, vap, vpp, cr, ct, f, vsap)
1056#define	VOP_RMDIR(dp, p, cdir, cr, ct, f) \
1057	fop_rmdir(dp, p, cdir, cr, ct, f)
1058#define	VOP_READDIR(vp, uiop, cr, eofp, ct, f) \
1059	fop_readdir(vp, uiop, cr, eofp, ct, f)
1060#define	VOP_SYMLINK(dvp, lnm, vap, tnm, cr, ct, f) \
1061	fop_symlink(dvp, lnm, vap, tnm, cr, ct, f)
1062#define	VOP_READLINK(vp, uiop, cr, ct) \
1063	fop_readlink(vp, uiop, cr, ct)
1064#define	VOP_FSYNC(vp, syncflag, cr, ct) \
1065	fop_fsync(vp, syncflag, cr, ct)
1066#define	VOP_INACTIVE(vp, cr, ct) \
1067	fop_inactive(vp, cr, ct)
1068#define	VOP_FID(vp, fidp, ct) \
1069	fop_fid(vp, fidp, ct)
1070#define	VOP_RWLOCK(vp, w, ct) \
1071	fop_rwlock(vp, w, ct)
1072#define	VOP_RWUNLOCK(vp, w, ct) \
1073	fop_rwunlock(vp, w, ct)
1074#define	VOP_SEEK(vp, ooff, noffp, ct) \
1075	fop_seek(vp, ooff, noffp, ct)
1076#define	VOP_CMP(vp1, vp2, ct) \
1077	fop_cmp(vp1, vp2, ct)
1078#define	VOP_FRLOCK(vp, cmd, a, f, o, cb, cr, ct) \
1079	fop_frlock(vp, cmd, a, f, o, cb, cr, ct)
1080#define	VOP_SPACE(vp, cmd, a, f, o, cr, ct) \
1081	fop_space(vp, cmd, a, f, o, cr, ct)
1082#define	VOP_REALVP(vp1, vp2, ct) \
1083	fop_realvp(vp1, vp2, ct)
1084#define	VOP_GETPAGE(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct) \
1085	fop_getpage(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct)
1086#define	VOP_PUTPAGE(vp, of, sz, fl, cr, ct) \
1087	fop_putpage(vp, of, sz, fl, cr, ct)
1088#define	VOP_MAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1089	fop_map(vp, of, as, a, sz, p, mp, fl, cr, ct)
1090#define	VOP_ADDMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1091	fop_addmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1092#define	VOP_DELMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1093	fop_delmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1094#define	VOP_POLL(vp, events, anyyet, reventsp, phpp, ct) \
1095	fop_poll(vp, events, anyyet, reventsp, phpp, ct)
1096#define	VOP_DUMP(vp, addr, bn, count, ct) \
1097	fop_dump(vp, addr, bn, count, ct)
1098#define	VOP_PATHCONF(vp, cmd, valp, cr, ct) \
1099	fop_pathconf(vp, cmd, valp, cr, ct)
1100#define	VOP_PAGEIO(vp, pp, io_off, io_len, flags, cr, ct) \
1101	fop_pageio(vp, pp, io_off, io_len, flags, cr, ct)
1102#define	VOP_DUMPCTL(vp, action, blkp, ct) \
1103	fop_dumpctl(vp, action, blkp, ct)
1104#define	VOP_DISPOSE(vp, pp, flag, dn, cr, ct) \
1105	fop_dispose(vp, pp, flag, dn, cr, ct)
1106#define	VOP_GETSECATTR(vp, vsap, f, cr, ct) \
1107	fop_getsecattr(vp, vsap, f, cr, ct)
1108#define	VOP_SETSECATTR(vp, vsap, f, cr, ct) \
1109	fop_setsecattr(vp, vsap, f, cr, ct)
1110#define	VOP_SHRLOCK(vp, cmd, shr, f, cr, ct) \
1111	fop_shrlock(vp, cmd, shr, f, cr, ct)
1112#define	VOP_VNEVENT(vp, vnevent, dvp, fnm, ct) \
1113	fop_vnevent(vp, vnevent, dvp, fnm, ct)
1114#define	VOP_REQZCBUF(vp, rwflag, xuiop, cr, ct) \
1115	fop_reqzcbuf(vp, rwflag, xuiop, cr, ct)
1116#define	VOP_RETZCBUF(vp, xuiop, cr, ct) \
1117	fop_retzcbuf(vp, xuiop, cr, ct)
1118
1119#define	VOPNAME_OPEN		"open"
1120#define	VOPNAME_CLOSE		"close"
1121#define	VOPNAME_READ		"read"
1122#define	VOPNAME_WRITE		"write"
1123#define	VOPNAME_IOCTL		"ioctl"
1124#define	VOPNAME_SETFL		"setfl"
1125#define	VOPNAME_GETATTR		"getattr"
1126#define	VOPNAME_SETATTR		"setattr"
1127#define	VOPNAME_ACCESS		"access"
1128#define	VOPNAME_LOOKUP		"lookup"
1129#define	VOPNAME_CREATE		"create"
1130#define	VOPNAME_REMOVE		"remove"
1131#define	VOPNAME_LINK		"link"
1132#define	VOPNAME_RENAME		"rename"
1133#define	VOPNAME_MKDIR		"mkdir"
1134#define	VOPNAME_RMDIR		"rmdir"
1135#define	VOPNAME_READDIR		"readdir"
1136#define	VOPNAME_SYMLINK		"symlink"
1137#define	VOPNAME_READLINK	"readlink"
1138#define	VOPNAME_FSYNC		"fsync"
1139#define	VOPNAME_INACTIVE	"inactive"
1140#define	VOPNAME_FID		"fid"
1141#define	VOPNAME_RWLOCK		"rwlock"
1142#define	VOPNAME_RWUNLOCK	"rwunlock"
1143#define	VOPNAME_SEEK		"seek"
1144#define	VOPNAME_CMP		"cmp"
1145#define	VOPNAME_FRLOCK		"frlock"
1146#define	VOPNAME_SPACE		"space"
1147#define	VOPNAME_REALVP		"realvp"
1148#define	VOPNAME_GETPAGE		"getpage"
1149#define	VOPNAME_PUTPAGE		"putpage"
1150#define	VOPNAME_MAP		"map"
1151#define	VOPNAME_ADDMAP		"addmap"
1152#define	VOPNAME_DELMAP		"delmap"
1153#define	VOPNAME_POLL		"poll"
1154#define	VOPNAME_DUMP		"dump"
1155#define	VOPNAME_PATHCONF	"pathconf"
1156#define	VOPNAME_PAGEIO		"pageio"
1157#define	VOPNAME_DUMPCTL		"dumpctl"
1158#define	VOPNAME_DISPOSE		"dispose"
1159#define	VOPNAME_GETSECATTR	"getsecattr"
1160#define	VOPNAME_SETSECATTR	"setsecattr"
1161#define	VOPNAME_SHRLOCK		"shrlock"
1162#define	VOPNAME_VNEVENT		"vnevent"
1163#define	VOPNAME_REQZCBUF	"reqzcbuf"
1164#define	VOPNAME_RETZCBUF	"retzcbuf"
1165
1166/*
1167 * Flags for VOP_LOOKUP
1168 *
1169 * Defined in file.h, but also possible, FIGNORECASE and FSEARCH
1170 *
1171 */
1172#define	LOOKUP_DIR		0x01	/* want parent dir vp */
1173#define	LOOKUP_XATTR		0x02	/* lookup up extended attr dir */
1174#define	CREATE_XATTR_DIR	0x04	/* Create extended attr dir */
1175#define	LOOKUP_HAVE_SYSATTR_DIR	0x08	/* Already created virtual GFS dir */
1176
1177/*
1178 * Flags for VOP_READDIR
1179 */
1180#define	V_RDDIR_ENTFLAGS	0x01	/* request dirent flags */
1181#define	V_RDDIR_ACCFILTER	0x02	/* filter out inaccessible dirents */
1182
1183/*
1184 * Flags for VOP_RWLOCK/VOP_RWUNLOCK
1185 * VOP_RWLOCK will return the flag that was actually set, or -1 if none.
1186 */
1187#define	V_WRITELOCK_TRUE	(1)	/* Request write-lock on the vnode */
1188#define	V_WRITELOCK_FALSE	(0)	/* Request read-lock on the vnode */
1189
1190/*
1191 * Flags for VOP_DUMPCTL
1192 */
1193#define	DUMP_ALLOC	0
1194#define	DUMP_FREE	1
1195#define	DUMP_SCAN	2
1196
1197/*
1198 * Public vnode manipulation functions.
1199 */
1200#if defined(_KERNEL) || defined(_FAKE_KERNEL)
1201
1202vnode_t *vn_alloc(int);
1203void	vn_reinit(vnode_t *);
1204void	vn_recycle(vnode_t *);
1205void	vn_free(vnode_t *);
1206
1207int	vn_is_readonly(vnode_t *);
1208int   	vn_is_opened(vnode_t *, v_mode_t);
1209int   	vn_is_mapped(vnode_t *, v_mode_t);
1210int   	vn_has_other_opens(vnode_t *, v_mode_t);
1211void	vn_open_upgrade(vnode_t *, int);
1212void	vn_open_downgrade(vnode_t *, int);
1213
1214int	vn_can_change_zones(vnode_t *vp);
1215
1216int	vn_has_flocks(vnode_t *);
1217int	vn_has_mandatory_locks(vnode_t *, int);
1218int	vn_has_cached_data(vnode_t *);
1219
1220void	vn_setops(vnode_t *, vnodeops_t *);
1221vnodeops_t *vn_getops(vnode_t *);
1222int	vn_matchops(vnode_t *, vnodeops_t *);
1223int	vn_matchopval(vnode_t *, char *, fs_generic_func_p);
1224int	vn_ismntpt(vnode_t *);
1225
1226struct vfs *vn_mountedvfs(vnode_t *);
1227
1228int	vn_in_dnlc(vnode_t *);
1229
1230void	vn_create_cache(void);
1231void	vn_destroy_cache(void);
1232
1233void	vn_freevnodeops(vnodeops_t *);
1234
1235int	vn_open(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1236		struct vnode **vpp, enum create crwhy, mode_t umask);
1237int	vn_openat(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1238		struct vnode **vpp, enum create crwhy,
1239		mode_t umask, struct vnode *startvp, int fd);
1240int	vn_create(char *pnamep, enum uio_seg seg, struct vattr *vap,
1241		enum vcexcl excl, int mode, struct vnode **vpp,
1242		enum create why, int flag, mode_t umask);
1243int	vn_createat(char *pnamep, enum uio_seg seg, struct vattr *vap,
1244		enum vcexcl excl, int mode, struct vnode **vpp,
1245		enum create why, int flag, mode_t umask, struct vnode *startvp);
1246int	vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base, ssize_t len,
1247		offset_t offset, enum uio_seg seg, int ioflag, rlim64_t ulimit,
1248		cred_t *cr, ssize_t *residp);
1249
1250void	vn_hold(struct vnode *vp);
1251void	vn_rele(struct vnode *vp);
1252void	vn_rele_async(struct vnode *vp, struct taskq *taskq);
1253void	vn_rele_dnlc(struct vnode *vp);
1254void	vn_rele_stream(struct vnode *vp);
1255int	vn_link(char *from, char *to, enum uio_seg seg);
1256int	vn_linkat(vnode_t *fstartvp, char *from, enum symfollow follow,
1257		vnode_t *tstartvp, char *to, enum uio_seg seg);
1258int	vn_rename(char *from, char *to, enum uio_seg seg);
1259int	vn_renameat(vnode_t *fdvp, char *fname, vnode_t *tdvp, char *tname,
1260		enum uio_seg seg);
1261int	vn_remove(char *fnamep, enum uio_seg seg, enum rm dirflag);
1262int	vn_removeat(vnode_t *startvp, char *fnamep, enum uio_seg seg,
1263		enum rm dirflag);
1264int	vn_compare(vnode_t *vp1, vnode_t *vp2);
1265int	vn_vfswlock(struct vnode *vp);
1266int	vn_vfswlock_wait(struct vnode *vp);
1267int	vn_vfsrlock(struct vnode *vp);
1268int	vn_vfsrlock_wait(struct vnode *vp);
1269void	vn_vfsunlock(struct vnode *vp);
1270int	vn_vfswlock_held(struct vnode *vp);
1271vnode_t *specvp(struct vnode *vp, dev_t dev, vtype_t type, struct cred *cr);
1272vnode_t *makespecvp(dev_t dev, vtype_t type);
1273
1274#if defined(_KERNEL)
1275vn_vfslocks_entry_t *vn_vfslocks_getlock(void *);
1276void	vn_vfslocks_rele(vn_vfslocks_entry_t *);
1277#endif
1278
1279boolean_t vn_is_reparse(vnode_t *, cred_t *, caller_context_t *);
1280
1281void vn_copypath(struct vnode *src, struct vnode *dst);
1282void vn_setpath_str(struct vnode *vp, const char *str, size_t len);
1283void vn_setpath(vnode_t *rootvp, struct vnode *startvp, struct vnode *vp,
1284    const char *path, size_t plen);
1285void vn_renamepath(vnode_t *dvp, vnode_t *vp, const char *nm, size_t len);
1286
1287/* Vnode event notification */
1288void	vnevent_rename_src(vnode_t *, vnode_t *, char *, caller_context_t *);
1289void	vnevent_rename_dest(vnode_t *, vnode_t *, char *, caller_context_t *);
1290void	vnevent_remove(vnode_t *, vnode_t *, char *, caller_context_t *);
1291void	vnevent_rmdir(vnode_t *, vnode_t *, char *, caller_context_t *);
1292void	vnevent_create(vnode_t *, caller_context_t *);
1293void	vnevent_link(vnode_t *, caller_context_t *);
1294void	vnevent_rename_dest_dir(vnode_t *, caller_context_t *ct);
1295void	vnevent_mountedover(vnode_t *, caller_context_t *);
1296void	vnevent_truncate(vnode_t *, caller_context_t *);
1297int	vnevent_support(vnode_t *, caller_context_t *);
1298
1299/* Vnode specific data */
1300void vsd_create(uint_t *, void (*)(void *));
1301void vsd_destroy(uint_t *);
1302void *vsd_get(vnode_t *, uint_t);
1303int vsd_set(vnode_t *, uint_t, void *);
1304void vsd_free(vnode_t *);
1305
1306/*
1307 * Extensible vnode attribute (xva) routines:
1308 * xva_init() initializes an xvattr_t (zero struct, init mapsize, set AT_XATTR)
1309 * xva_getxoptattr() returns a ponter to the xoptattr_t section of xvattr_t
1310 */
1311void		xva_init(xvattr_t *);
1312xoptattr_t	*xva_getxoptattr(xvattr_t *);	/* Get ptr to xoptattr_t */
1313
1314void xattr_init(void);		/* Initialize vnodeops for xattrs */
1315
1316/* GFS tunnel for xattrs */
1317int xattr_dir_lookup(vnode_t *, vnode_t **, int, cred_t *);
1318
1319/* Reparse Point */
1320void reparse_point_init(void);
1321
1322/* Context identification */
1323u_longlong_t	fs_new_caller_id();
1324
1325int	vn_vmpss_usepageio(vnode_t *);
1326
1327/*
1328 * Needed for use of IS_VMODSORT() in kernel.
1329 */
1330extern uint_t pvn_vmodsort_supported;
1331
1332#define	VN_HOLD(vp)	{ \
1333	vn_hold(vp); \
1334}
1335
1336#define	VN_RELE(vp)	{ \
1337	vn_rele(vp); \
1338}
1339
1340#define	VN_RELE_ASYNC(vp, taskq)	{ \
1341	vn_rele_async(vp, taskq); \
1342}
1343
1344#define	VN_SET_VFS_TYPE_DEV(vp, vfsp, type, dev)	{ \
1345	(vp)->v_vfsp = (vfsp); \
1346	(vp)->v_type = (type); \
1347	(vp)->v_rdev = (dev); \
1348}
1349
1350/*
1351 * Compare two vnodes for equality.  In general this macro should be used
1352 * in preference to calling VOP_CMP directly.
1353 */
1354#if defined(_FAKE_KERNEL)
1355#define	VN_CMP(VP1, VP2) \
1356	(((VP1) == (VP2)) ? 1 : VOP_CMP(VP1, VP2, NULL))
1357#else
1358#define	VN_CMP(VP1, VP2)	((VP1) == (VP2) ? 1 : 	\
1359	((VP1) && (VP2) && (vn_getops(VP1) == vn_getops(VP2)) ? \
1360	VOP_CMP(VP1, VP2, NULL) : 0))
1361#endif
1362
1363/*
1364 * Some well-known global vnodes used by the VM system to name pages.
1365 */
1366extern struct vnode kvps[];
1367
1368typedef enum {
1369	KV_KVP,		/* vnode for all segkmem pages */
1370	KV_ZVP,		/* vnode for all ZFS pages */
1371#if defined(__sparc)
1372	KV_MPVP,	/* vnode for all page_t meta-pages */
1373	KV_PROMVP,	/* vnode for all PROM pages */
1374#endif	/* __sparc */
1375	KV_MAX		/* total number of vnodes in kvps[] */
1376} kvps_index_t;
1377
1378#define	VN_ISKAS(vp)	((vp) >= &kvps[0] && (vp) < &kvps[KV_MAX])
1379
1380#endif	/* _KERNEL */
1381
1382/*
1383 * Flags to VOP_SETATTR/VOP_GETATTR.
1384 */
1385#define	ATTR_UTIME	0x01	/* non-default utime(2) request */
1386#define	ATTR_EXEC	0x02	/* invocation from exec(2) */
1387#define	ATTR_COMM	0x04	/* yield common vp attributes */
1388#define	ATTR_HINT	0x08	/* information returned will be `hint' */
1389#define	ATTR_REAL	0x10	/* yield attributes of the real vp */
1390#define	ATTR_NOACLCHECK	0x20	/* Don't check ACL when checking permissions */
1391#define	ATTR_TRIGGER	0x40	/* Mount first if vnode is a trigger mount */
1392/*
1393 * Generally useful macros.
1394 */
1395#define	VBSIZE(vp)	((vp)->v_vfsp->vfs_bsize)
1396
1397#define	VTOZONE(vp)	((vp)->v_vfsp->vfs_zone)
1398
1399#define	NULLVP		((struct vnode *)0)
1400#define	NULLVPP		((struct vnode **)0)
1401
1402#ifdef	_KERNEL
1403
1404/*
1405 * Structure used while handling asynchronous VOP_PUTPAGE operations.
1406 */
1407struct async_reqs {
1408	struct async_reqs *a_next;	/* pointer to next arg struct */
1409	struct vnode *a_vp;		/* vnode pointer */
1410	u_offset_t a_off;			/* offset in file */
1411	uint_t a_len;			/* size of i/o request */
1412	int a_flags;			/* flags to indicate operation type */
1413	struct cred *a_cred;		/* cred pointer	*/
1414	ushort_t a_prealloced;		/* set if struct is pre-allocated */
1415};
1416
1417/*
1418 * VN_DISPOSE() -- given a page pointer, safely invoke VOP_DISPOSE().
1419 * Note that there is no guarantee that the page passed in will be
1420 * freed.  If that is required, then a check after calling VN_DISPOSE would
1421 * be necessary to ensure the page was freed.
1422 */
1423#define	VN_DISPOSE(pp, flag, dn, cr)	{ \
1424	if ((pp)->p_vnode != NULL && !VN_ISKAS((pp)->p_vnode)) \
1425		VOP_DISPOSE((pp)->p_vnode, (pp), (flag), (dn), (cr), NULL); \
1426	else if ((flag) == B_FREE) \
1427		page_free((pp), (dn)); \
1428	else \
1429		page_destroy((pp), (dn)); \
1430	}
1431
1432#endif	/* _KERNEL */
1433
1434#ifdef	__cplusplus
1435}
1436#endif
1437
1438#endif	/* _SYS_VNODE_H */
1439