xref: /titanic_50/usr/src/uts/common/sys/vnode.h (revision 634e26ec75c89095090605284938356a3145f2b8)
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  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 /*	Copyright (c) 1983, 1984, 1985, 1986, 1987, 1988, 1989 AT&T	*/
27 /*	  All Rights Reserved  	*/
28 
29 /*
30  * University Copyright- Copyright (c) 1982, 1986, 1988
31  * The Regents of the University of California
32  * All Rights Reserved
33  *
34  * University Acknowledgment- Portions of this document are derived from
35  * software developed by the University of California, Berkeley, and its
36  * contributors.
37  */
38 
39 #ifndef _SYS_VNODE_H
40 #define	_SYS_VNODE_H
41 
42 #include <sys/types.h>
43 #include <sys/t_lock.h>
44 #include <sys/rwstlock.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 #ifdef	_KERNEL
54 #include <sys/buf.h>
55 #endif	/* _KERNEL */
56 
57 #ifdef	__cplusplus
58 extern "C" {
59 #endif
60 
61 /*
62  * Statistics for all vnode operations.
63  * All operations record number of ops (since boot/mount/zero'ed).
64  * Certain I/O operations (read, write, readdir) also record number
65  * of bytes transferred.
66  * This appears in two places in the system: one is embedded in each
67  * vfs_t.  There is also an array of vopstats_t structures allocated
68  * on a per-fstype basis.
69  */
70 
71 #define	VOPSTATS_STR	"vopstats_"	/* Initial string for vopstat kstats */
72 
73 typedef struct vopstats {
74 	kstat_named_t	nopen;		/* VOP_OPEN */
75 	kstat_named_t	nclose;		/* VOP_CLOSE */
76 	kstat_named_t	nread;		/* VOP_READ */
77 	kstat_named_t	read_bytes;
78 	kstat_named_t	nwrite;		/* VOP_WRITE */
79 	kstat_named_t	write_bytes;
80 	kstat_named_t	nioctl;		/* VOP_IOCTL */
81 	kstat_named_t	nsetfl;		/* VOP_SETFL */
82 	kstat_named_t	ngetattr;	/* VOP_GETATTR */
83 	kstat_named_t	nsetattr;	/* VOP_SETATTR */
84 	kstat_named_t	naccess;	/* VOP_ACCESS */
85 	kstat_named_t	nlookup;	/* VOP_LOOKUP */
86 	kstat_named_t	ncreate;	/* VOP_CREATE */
87 	kstat_named_t	nremove;	/* VOP_REMOVE */
88 	kstat_named_t	nlink;		/* VOP_LINK */
89 	kstat_named_t	nrename;	/* VOP_RENAME */
90 	kstat_named_t	nmkdir;		/* VOP_MKDIR */
91 	kstat_named_t	nrmdir;		/* VOP_RMDIR */
92 	kstat_named_t	nreaddir;	/* VOP_READDIR */
93 	kstat_named_t	readdir_bytes;
94 	kstat_named_t	nsymlink;	/* VOP_SYMLINK */
95 	kstat_named_t	nreadlink;	/* VOP_READLINK */
96 	kstat_named_t	nfsync;		/* VOP_FSYNC */
97 	kstat_named_t	ninactive;	/* VOP_INACTIVE */
98 	kstat_named_t	nfid;		/* VOP_FID */
99 	kstat_named_t	nrwlock;	/* VOP_RWLOCK */
100 	kstat_named_t	nrwunlock;	/* VOP_RWUNLOCK */
101 	kstat_named_t	nseek;		/* VOP_SEEK */
102 	kstat_named_t	ncmp;		/* VOP_CMP */
103 	kstat_named_t	nfrlock;	/* VOP_FRLOCK */
104 	kstat_named_t	nspace;		/* VOP_SPACE */
105 	kstat_named_t	nrealvp;	/* VOP_REALVP */
106 	kstat_named_t	ngetpage;	/* VOP_GETPAGE */
107 	kstat_named_t	nputpage;	/* VOP_PUTPAGE */
108 	kstat_named_t	nmap;		/* VOP_MAP */
109 	kstat_named_t	naddmap;	/* VOP_ADDMAP */
110 	kstat_named_t	ndelmap;	/* VOP_DELMAP */
111 	kstat_named_t	npoll;		/* VOP_POLL */
112 	kstat_named_t	ndump;		/* VOP_DUMP */
113 	kstat_named_t	npathconf;	/* VOP_PATHCONF */
114 	kstat_named_t	npageio;	/* VOP_PAGEIO */
115 	kstat_named_t	ndumpctl;	/* VOP_DUMPCTL */
116 	kstat_named_t	ndispose;	/* VOP_DISPOSE */
117 	kstat_named_t	nsetsecattr;	/* VOP_SETSECATTR */
118 	kstat_named_t	ngetsecattr;	/* VOP_GETSECATTR */
119 	kstat_named_t	nshrlock;	/* VOP_SHRLOCK */
120 	kstat_named_t	nvnevent;	/* VOP_VNEVENT */
121 } vopstats_t;
122 
123 /*
124  * The vnode is the focus of all file activity in UNIX.
125  * A vnode is allocated for each active file, each current
126  * directory, each mounted-on file, and the root.
127  *
128  * Each vnode is usually associated with a file-system-specific node (for
129  * UFS, this is the in-memory inode).  Generally, a vnode and an fs-node
130  * should be created and destroyed together as a pair.
131  *
132  * If a vnode is reused for a new file, it should be reinitialized by calling
133  * either vn_reinit() or vn_recycle().
134  *
135  * vn_reinit() resets the entire vnode as if it was returned by vn_alloc().
136  * The caller is responsible for setting up the entire vnode after calling
137  * vn_reinit().  This is important when using kmem caching where the vnode is
138  * allocated by a constructor, for instance.
139  *
140  * vn_recycle() is used when the file system keeps some state around in both
141  * the vnode and the associated FS-node.  In UFS, for example, the inode of
142  * a deleted file can be reused immediately.  The v_data, v_vfsp, v_op, etc.
143  * remains the same but certain fields related to the previous instance need
144  * to be reset.  In particular:
145  *	v_femhead
146  *	v_path
147  *	v_rdcnt, v_wrcnt
148  *	v_mmap_read, v_mmap_write
149  */
150 
151 /*
152  * vnode types.  VNON means no type.  These values are unrelated to
153  * values in on-disk inodes.
154  */
155 typedef enum vtype {
156 	VNON	= 0,
157 	VREG	= 1,
158 	VDIR	= 2,
159 	VBLK	= 3,
160 	VCHR	= 4,
161 	VLNK	= 5,
162 	VFIFO	= 6,
163 	VDOOR	= 7,
164 	VPROC	= 8,
165 	VSOCK	= 9,
166 	VPORT	= 10,
167 	VBAD	= 11
168 } vtype_t;
169 
170 /*
171  * VSD - Vnode Specific Data
172  * Used to associate additional private data with a vnode.
173  */
174 struct vsd_node {
175 	list_node_t vs_nodes;		/* list of all VSD nodes */
176 	uint_t vs_nkeys;		/* entries in value array */
177 	void **vs_value;		/* array of value/key */
178 };
179 
180 /*
181  * Many of the fields in the vnode are read-only once they are initialized
182  * at vnode creation time.  Other fields are protected by locks.
183  *
184  * IMPORTANT: vnodes should be created ONLY by calls to vn_alloc().  They
185  * may not be embedded into the file-system specific node (inode).  The
186  * size of vnodes may change.
187  *
188  * The v_lock protects:
189  *   v_flag
190  *   v_stream
191  *   v_count
192  *   v_shrlocks
193  *   v_path
194  *   v_vsd
195  *   v_xattrdir
196  *
197  * A special lock (implemented by vn_vfswlock in vnode.c) protects:
198  *   v_vfsmountedhere
199  *
200  * The global flock_lock mutex (in flock.c) protects:
201  *   v_filocks
202  *
203  * IMPORTANT NOTE:
204  *
205  *   The following vnode fields are considered public and may safely be
206  *   accessed by file systems or other consumers:
207  *
208  *     v_lock
209  *     v_flag
210  *     v_count
211  *     v_data
212  *     v_vfsp
213  *     v_stream
214  *     v_type
215  *     v_rdev
216  *
217  * ALL OTHER FIELDS SHOULD BE ACCESSED ONLY BY THE OWNER OF THAT FIELD.
218  * In particular, file systems should not access other fields; they may
219  * change or even be removed.  The functionality which was once provided
220  * by these fields is available through vn_* functions.
221  */
222 
223 struct fem_head;	/* from fem.h */
224 
225 typedef struct vnode {
226 	kmutex_t	v_lock;		/* protects vnode fields */
227 	uint_t		v_flag;		/* vnode flags (see below) */
228 	uint_t		v_count;	/* reference count */
229 	void		*v_data;	/* private data for fs */
230 	struct vfs	*v_vfsp;	/* ptr to containing VFS */
231 	struct stdata	*v_stream;	/* associated stream */
232 	enum vtype	v_type;		/* vnode type */
233 	dev_t		v_rdev;		/* device (VCHR, VBLK) */
234 
235 	/* PRIVATE FIELDS BELOW - DO NOT USE */
236 
237 	struct vfs	*v_vfsmountedhere; /* ptr to vfs mounted here */
238 	struct vnodeops	*v_op;		/* vnode operations */
239 	struct page	*v_pages;	/* vnode pages list */
240 	struct filock	*v_filocks;	/* ptr to filock list */
241 	struct shrlocklist *v_shrlocks;	/* ptr to shrlock list */
242 	krwlock_t	v_nbllock;	/* sync for NBMAND locks */
243 	kcondvar_t	v_cv;		/* synchronize locking */
244 	void		*v_locality;	/* hook for locality info */
245 	struct fem_head	*v_femhead;	/* fs monitoring */
246 	char		*v_path;	/* cached path */
247 	uint_t		v_rdcnt;	/* open for read count  (VREG only) */
248 	uint_t		v_wrcnt;	/* open for write count (VREG only) */
249 	u_longlong_t	v_mmap_read;	/* mmap read count */
250 	u_longlong_t	v_mmap_write;	/* mmap write count */
251 	void		*v_mpssdata;	/* info for large page mappings */
252 	void		*v_fopdata;	/* list of file ops event watches */
253 	kmutex_t	v_vsd_lock;	/* protects v_vsd field */
254 	struct vsd_node *v_vsd;		/* vnode specific data */
255 	struct vnode	*v_xattrdir;	/* unnamed extended attr dir (GFS) */
256 	uint_t		v_count_dnlc;	/* dnlc reference count */
257 } vnode_t;
258 
259 #define	IS_DEVVP(vp)	\
260 	((vp)->v_type == VCHR || (vp)->v_type == VBLK || (vp)->v_type == VFIFO)
261 
262 /*
263  * vnode flags.
264  */
265 #define	VROOT		0x01	/* root of its file system */
266 #define	VNOCACHE	0x02	/* don't keep cache pages on vnode */
267 #define	VNOMAP		0x04	/* file cannot be mapped/faulted */
268 #define	VDUP		0x08	/* file should be dup'ed rather then opened */
269 #define	VNOSWAP		0x10	/* file cannot be used as virtual swap device */
270 #define	VNOMOUNT	0x20	/* file cannot be covered by mount */
271 #define	VISSWAP		0x40	/* vnode is being used for swap */
272 #define	VSWAPLIKE	0x80	/* vnode acts like swap (but may not be) */
273 
274 #define	IS_SWAPVP(vp)	(((vp)->v_flag & (VISSWAP | VSWAPLIKE)) != 0)
275 
276 typedef struct vn_vfslocks_entry {
277 	rwstlock_t ve_lock;
278 	void *ve_vpvfs;
279 	struct vn_vfslocks_entry *ve_next;
280 	uint32_t ve_refcnt;
281 	char pad[64 - sizeof (rwstlock_t) - 2 * sizeof (void *) - \
282 	    sizeof (uint32_t)];
283 } vn_vfslocks_entry_t;
284 
285 /*
286  * The following two flags are used to lock the v_vfsmountedhere field
287  */
288 #define	VVFSLOCK	0x100
289 #define	VVFSWAIT	0x200
290 
291 /*
292  * Used to serialize VM operations on a vnode
293  */
294 #define	VVMLOCK		0x400
295 
296 /*
297  * Tell vn_open() not to fail a directory open for writing but
298  * to go ahead and call VOP_OPEN() to let the filesystem check.
299  */
300 #define	VDIROPEN	0x800
301 
302 /*
303  * Flag to let the VM system know that this file is most likely a binary
304  * or shared library since it has been mmap()ed EXEC at some time.
305  */
306 #define	VVMEXEC		0x1000
307 
308 #define	VPXFS		0x2000  /* clustering: global fs proxy vnode */
309 
310 #define	IS_PXFSVP(vp)	((vp)->v_flag & VPXFS)
311 
312 #define	V_XATTRDIR	0x4000	/* attribute unnamed directory */
313 
314 #define	IS_XATTRDIR(vp)	((vp)->v_flag & V_XATTRDIR)
315 
316 #define	V_LOCALITY	0x8000	/* whether locality aware */
317 
318 /*
319  * Flag that indicates the VM should maintain the v_pages list with all modified
320  * pages on one end and unmodified pages at the other. This makes finding dirty
321  * pages to write back to disk much faster at the expense of taking a minor
322  * fault on the first store instruction which touches a writable page.
323  */
324 #define	VMODSORT	(0x10000)
325 #define	IS_VMODSORT(vp) \
326 	(pvn_vmodsort_supported != 0 && ((vp)->v_flag  & VMODSORT) != 0)
327 
328 #define	VISSWAPFS	0x20000	/* vnode is being used for swapfs */
329 
330 /*
331  * The mdb memstat command assumes that IS_SWAPFSVP only uses the
332  * vnode's v_flag field.  If this changes, cache the additional
333  * fields in mdb; see vn_get in mdb/common/modules/genunix/memory.c
334  */
335 #define	IS_SWAPFSVP(vp)	(((vp)->v_flag & VISSWAPFS) != 0)
336 
337 #define	V_SYSATTR	0x40000	/* vnode is a GFS system attribute */
338 
339 /*
340  * Vnode attributes.  A bit-mask is supplied as part of the
341  * structure to indicate the attributes the caller wants to
342  * set (setattr) or extract (getattr).
343  */
344 
345 /*
346  * Note that va_nodeid and va_nblocks are 64bit data type.
347  * We support large files over NFSV3. With Solaris client and
348  * Server that generates 64bit ino's and sizes these fields
349  * will overflow if they are 32 bit sizes.
350  */
351 
352 typedef struct vattr {
353 	uint_t		va_mask;	/* bit-mask of attributes */
354 	vtype_t		va_type;	/* vnode type (for create) */
355 	mode_t		va_mode;	/* file access mode */
356 	uid_t		va_uid;		/* owner user id */
357 	gid_t		va_gid;		/* owner group id */
358 	dev_t		va_fsid;	/* file system id (dev for now) */
359 	u_longlong_t	va_nodeid;	/* node id */
360 	nlink_t		va_nlink;	/* number of references to file */
361 	u_offset_t	va_size;	/* file size in bytes */
362 	timestruc_t	va_atime;	/* time of last access */
363 	timestruc_t	va_mtime;	/* time of last modification */
364 	timestruc_t	va_ctime;	/* time of last status change */
365 	dev_t		va_rdev;	/* device the file represents */
366 	uint_t		va_blksize;	/* fundamental block size */
367 	u_longlong_t	va_nblocks;	/* # of blocks allocated */
368 	uint_t		va_seq;		/* sequence number */
369 } vattr_t;
370 
371 #define	AV_SCANSTAMP_SZ	32		/* length of anti-virus scanstamp */
372 
373 /*
374  * Structure of all optional attributes.
375  */
376 typedef struct xoptattr {
377 	timestruc_t	xoa_createtime;	/* Create time of file */
378 	uint8_t		xoa_archive;
379 	uint8_t		xoa_system;
380 	uint8_t		xoa_readonly;
381 	uint8_t		xoa_hidden;
382 	uint8_t		xoa_nounlink;
383 	uint8_t		xoa_immutable;
384 	uint8_t		xoa_appendonly;
385 	uint8_t		xoa_nodump;
386 	uint8_t		xoa_opaque;
387 	uint8_t		xoa_av_quarantined;
388 	uint8_t		xoa_av_modified;
389 	uint8_t		xoa_av_scanstamp[AV_SCANSTAMP_SZ];
390 	uint8_t		xoa_reparse;
391 } xoptattr_t;
392 
393 /*
394  * The xvattr structure is really a variable length structure that
395  * is made up of:
396  * - The classic vattr_t (xva_vattr)
397  * - a 32 bit quantity (xva_mapsize) that specifies the size of the
398  *   attribute bitmaps in 32 bit words.
399  * - A pointer to the returned attribute bitmap (needed because the
400  *   previous element, the requested attribute bitmap) is variable lenth.
401  * - The requested attribute bitmap, which is an array of 32 bit words.
402  *   Callers use the XVA_SET_REQ() macro to set the bits corresponding to
403  *   the attributes that are being requested.
404  * - The returned attribute bitmap, which is an array of 32 bit words.
405  *   File systems that support optional attributes use the XVA_SET_RTN()
406  *   macro to set the bits corresponding to the attributes that are being
407  *   returned.
408  * - The xoptattr_t structure which contains the attribute values
409  *
410  * xva_mapsize determines how many words in the attribute bitmaps.
411  * Immediately following the attribute bitmaps is the xoptattr_t.
412  * xva_getxoptattr() is used to get the pointer to the xoptattr_t
413  * section.
414  */
415 
416 #define	XVA_MAPSIZE	3		/* Size of attr bitmaps */
417 #define	XVA_MAGIC	0x78766174	/* Magic # for verification */
418 
419 /*
420  * The xvattr structure is an extensible structure which permits optional
421  * attributes to be requested/returned.  File systems may or may not support
422  * optional attributes.  They do so at their own discretion but if they do
423  * support optional attributes, they must register the VFSFT_XVATTR feature
424  * so that the optional attributes can be set/retrived.
425  *
426  * The fields of the xvattr structure are:
427  *
428  * xva_vattr - The first element of an xvattr is a legacy vattr structure
429  * which includes the common attributes.  If AT_XVATTR is set in the va_mask
430  * then the entire structure is treated as an xvattr.  If AT_XVATTR is not
431  * set, then only the xva_vattr structure can be used.
432  *
433  * xva_magic - 0x78766174 (hex for "xvat"). Magic number for verification.
434  *
435  * xva_mapsize - Size of requested and returned attribute bitmaps.
436  *
437  * xva_rtnattrmapp - Pointer to xva_rtnattrmap[].  We need this since the
438  * size of the array before it, xva_reqattrmap[], could change which means
439  * the location of xva_rtnattrmap[] could change.  This will allow unbundled
440  * file systems to find the location of xva_rtnattrmap[] when the sizes change.
441  *
442  * xva_reqattrmap[] - Array of requested attributes.  Attributes are
443  * represented by a specific bit in a specific element of the attribute
444  * map array.  Callers set the bits corresponding to the attributes
445  * that the caller wants to get/set.
446  *
447  * xva_rtnattrmap[] - Array of attributes that the file system was able to
448  * process.  Not all file systems support all optional attributes.  This map
449  * informs the caller which attributes the underlying file system was able
450  * to set/get.  (Same structure as the requested attributes array in terms
451  * of each attribute  corresponding to specific bits and array elements.)
452  *
453  * xva_xoptattrs - Structure containing values of optional attributes.
454  * These values are only valid if the corresponding bits in xva_reqattrmap
455  * are set and the underlying file system supports those attributes.
456  */
457 typedef struct xvattr {
458 	vattr_t		xva_vattr;	/* Embedded vattr structure */
459 	uint32_t	xva_magic;	/* Magic Number */
460 	uint32_t	xva_mapsize;	/* Size of attr bitmap (32-bit words) */
461 	uint32_t	*xva_rtnattrmapp;	/* Ptr to xva_rtnattrmap[] */
462 	uint32_t	xva_reqattrmap[XVA_MAPSIZE];	/* Requested attrs */
463 	uint32_t	xva_rtnattrmap[XVA_MAPSIZE];	/* Returned attrs */
464 	xoptattr_t	xva_xoptattrs;	/* Optional attributes */
465 } xvattr_t;
466 
467 #ifdef _SYSCALL32
468 /*
469  * For bigtypes time_t changed to 64 bit on the 64-bit kernel.
470  * Define an old version for user/kernel interface
471  */
472 
473 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
474 #pragma pack(4)
475 #endif
476 
477 typedef struct vattr32 {
478 	uint32_t	va_mask;	/* bit-mask of attributes */
479 	vtype_t		va_type;	/* vnode type (for create) */
480 	mode32_t	va_mode;	/* file access mode */
481 	uid32_t		va_uid;		/* owner user id */
482 	gid32_t		va_gid;		/* owner group id */
483 	dev32_t		va_fsid;	/* file system id (dev for now) */
484 	u_longlong_t	va_nodeid;	/* node id */
485 	nlink_t		va_nlink;	/* number of references to file */
486 	u_offset_t	va_size;	/* file size in bytes */
487 	timestruc32_t	va_atime;	/* time of last access */
488 	timestruc32_t	va_mtime;	/* time of last modification */
489 	timestruc32_t	va_ctime;	/* time of last status change */
490 	dev32_t		va_rdev;	/* device the file represents */
491 	uint32_t	va_blksize;	/* fundamental block size */
492 	u_longlong_t	va_nblocks;	/* # of blocks allocated */
493 	uint32_t	va_seq;		/* sequence number */
494 } vattr32_t;
495 
496 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
497 #pragma pack()
498 #endif
499 
500 #else  /* not _SYSCALL32 */
501 #define	vattr32		vattr
502 typedef vattr_t		vattr32_t;
503 #endif /* _SYSCALL32 */
504 
505 /*
506  * Attributes of interest to the caller of setattr or getattr.
507  */
508 #define	AT_TYPE		0x00001
509 #define	AT_MODE		0x00002
510 #define	AT_UID		0x00004
511 #define	AT_GID		0x00008
512 #define	AT_FSID		0x00010
513 #define	AT_NODEID	0x00020
514 #define	AT_NLINK	0x00040
515 #define	AT_SIZE		0x00080
516 #define	AT_ATIME	0x00100
517 #define	AT_MTIME	0x00200
518 #define	AT_CTIME	0x00400
519 #define	AT_RDEV		0x00800
520 #define	AT_BLKSIZE	0x01000
521 #define	AT_NBLOCKS	0x02000
522 /*			0x04000 */	/* unused */
523 #define	AT_SEQ		0x08000
524 /*
525  * If AT_XVATTR is set then there are additional bits to process in
526  * the xvattr_t's attribute bitmap.  If this is not set then the bitmap
527  * MUST be ignored.  Note that this bit must be set/cleared explicitly.
528  * That is, setting AT_ALL will NOT set AT_XVATTR.
529  */
530 #define	AT_XVATTR	0x10000
531 
532 #define	AT_ALL		(AT_TYPE|AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|\
533 			AT_NLINK|AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|\
534 			AT_RDEV|AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
535 
536 #define	AT_STAT		(AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|AT_NLINK|\
537 			AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|AT_RDEV|AT_TYPE)
538 
539 #define	AT_TIMES	(AT_ATIME|AT_MTIME|AT_CTIME)
540 
541 #define	AT_NOSET	(AT_NLINK|AT_RDEV|AT_FSID|AT_NODEID|AT_TYPE|\
542 			AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
543 
544 /*
545  * Attribute bits used in the extensible attribute's (xva's) attribute
546  * bitmaps.  Note that the bitmaps are made up of a variable length number
547  * of 32-bit words.  The convention is to use XAT{n}_{attrname} where "n"
548  * is the element in the bitmap (starting at 1).  This convention is for
549  * the convenience of the maintainer to keep track of which element each
550  * attribute belongs to.
551  *
552  * NOTE THAT CONSUMERS MUST *NOT* USE THE XATn_* DEFINES DIRECTLY.  CONSUMERS
553  * MUST USE THE XAT_* DEFINES.
554  */
555 #define	XAT0_INDEX	0LL		/* Index into bitmap for XAT0 attrs */
556 #define	XAT0_CREATETIME	0x00000001	/* Create time of file */
557 #define	XAT0_ARCHIVE	0x00000002	/* Archive */
558 #define	XAT0_SYSTEM	0x00000004	/* System */
559 #define	XAT0_READONLY	0x00000008	/* Readonly */
560 #define	XAT0_HIDDEN	0x00000010	/* Hidden */
561 #define	XAT0_NOUNLINK	0x00000020	/* Nounlink */
562 #define	XAT0_IMMUTABLE	0x00000040	/* immutable */
563 #define	XAT0_APPENDONLY	0x00000080	/* appendonly */
564 #define	XAT0_NODUMP	0x00000100	/* nodump */
565 #define	XAT0_OPAQUE	0x00000200	/* opaque */
566 #define	XAT0_AV_QUARANTINED	0x00000400	/* anti-virus quarantine */
567 #define	XAT0_AV_MODIFIED	0x00000800	/* anti-virus modified */
568 #define	XAT0_AV_SCANSTAMP	0x00001000	/* anti-virus scanstamp */
569 #define	XAT0_REPARSE	0x00002000	/* FS reparse point */
570 
571 #define	XAT0_ALL_ATTRS	(XAT0_CREATETIME|XAT0_ARCHIVE|XAT0_SYSTEM| \
572     XAT0_READONLY|XAT0_HIDDEN|XAT0_NOUNLINK|XAT0_IMMUTABLE|XAT0_APPENDONLY| \
573     XAT0_NODUMP|XAT0_OPAQUE|XAT0_AV_QUARANTINED| \
574     XAT0_AV_MODIFIED|XAT0_AV_SCANSTAMP|XAT0_REPARSE)
575 
576 /* Support for XAT_* optional attributes */
577 #define	XVA_MASK		0xffffffff	/* Used to mask off 32 bits */
578 #define	XVA_SHFT		32		/* Used to shift index */
579 
580 /*
581  * Used to pry out the index and attribute bits from the XAT_* attributes
582  * defined below.  Note that we're masking things down to 32 bits then
583  * casting to uint32_t.
584  */
585 #define	XVA_INDEX(attr)		((uint32_t)(((attr) >> XVA_SHFT) & XVA_MASK))
586 #define	XVA_ATTRBIT(attr)	((uint32_t)((attr) & XVA_MASK))
587 
588 /*
589  * The following defines present a "flat namespace" so that consumers don't
590  * need to keep track of which element belongs to which bitmap entry.
591  *
592  * NOTE THAT THESE MUST NEVER BE OR-ed TOGETHER
593  */
594 #define	XAT_CREATETIME		((XAT0_INDEX << XVA_SHFT) | XAT0_CREATETIME)
595 #define	XAT_ARCHIVE		((XAT0_INDEX << XVA_SHFT) | XAT0_ARCHIVE)
596 #define	XAT_SYSTEM		((XAT0_INDEX << XVA_SHFT) | XAT0_SYSTEM)
597 #define	XAT_READONLY		((XAT0_INDEX << XVA_SHFT) | XAT0_READONLY)
598 #define	XAT_HIDDEN		((XAT0_INDEX << XVA_SHFT) | XAT0_HIDDEN)
599 #define	XAT_NOUNLINK		((XAT0_INDEX << XVA_SHFT) | XAT0_NOUNLINK)
600 #define	XAT_IMMUTABLE		((XAT0_INDEX << XVA_SHFT) | XAT0_IMMUTABLE)
601 #define	XAT_APPENDONLY		((XAT0_INDEX << XVA_SHFT) | XAT0_APPENDONLY)
602 #define	XAT_NODUMP		((XAT0_INDEX << XVA_SHFT) | XAT0_NODUMP)
603 #define	XAT_OPAQUE		((XAT0_INDEX << XVA_SHFT) | XAT0_OPAQUE)
604 #define	XAT_AV_QUARANTINED	((XAT0_INDEX << XVA_SHFT) | XAT0_AV_QUARANTINED)
605 #define	XAT_AV_MODIFIED		((XAT0_INDEX << XVA_SHFT) | XAT0_AV_MODIFIED)
606 #define	XAT_AV_SCANSTAMP	((XAT0_INDEX << XVA_SHFT) | XAT0_AV_SCANSTAMP)
607 #define	XAT_REPARSE		((XAT0_INDEX << XVA_SHFT) | XAT0_REPARSE)
608 
609 /*
610  * The returned attribute map array (xva_rtnattrmap[]) is located past the
611  * requested attribute map array (xva_reqattrmap[]).  Its location changes
612  * when the array sizes change.  We use a separate pointer in a known location
613  * (xva_rtnattrmapp) to hold the location of xva_rtnattrmap[].  This is
614  * set in xva_init()
615  */
616 #define	XVA_RTNATTRMAP(xvap)	((xvap)->xva_rtnattrmapp)
617 
618 /*
619  * XVA_SET_REQ() sets an attribute bit in the proper element in the bitmap
620  * of requested attributes (xva_reqattrmap[]).
621  */
622 #define	XVA_SET_REQ(xvap, attr)					\
623 	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
624 	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
625 	(xvap)->xva_reqattrmap[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
626 /*
627  * XVA_CLR_REQ() clears an attribute bit in the proper element in the bitmap
628  * of requested attributes (xva_reqattrmap[]).
629  */
630 #define	XVA_CLR_REQ(xvap, attr)					\
631 	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
632 	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
633 	(xvap)->xva_reqattrmap[XVA_INDEX(attr)] &= ~XVA_ATTRBIT(attr)
634 
635 /*
636  * XVA_SET_RTN() sets an attribute bit in the proper element in the bitmap
637  * of returned attributes (xva_rtnattrmap[]).
638  */
639 #define	XVA_SET_RTN(xvap, attr)					\
640 	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
641 	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
642 	(XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
643 
644 /*
645  * XVA_ISSET_REQ() checks the requested attribute bitmap (xva_reqattrmap[])
646  * to see of the corresponding attribute bit is set.  If so, returns non-zero.
647  */
648 #define	XVA_ISSET_REQ(xvap, attr)					\
649 	((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&			\
650 		((xvap)->xva_magic == XVA_MAGIC) &&			\
651 		((xvap)->xva_mapsize > XVA_INDEX(attr))) ?		\
652 	((xvap)->xva_reqattrmap[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) :	0)
653 
654 /*
655  * XVA_ISSET_RTN() checks the returned attribute bitmap (xva_rtnattrmap[])
656  * to see of the corresponding attribute bit is set.  If so, returns non-zero.
657  */
658 #define	XVA_ISSET_RTN(xvap, attr)					\
659 	((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&			\
660 		((xvap)->xva_magic == XVA_MAGIC) &&			\
661 		((xvap)->xva_mapsize > XVA_INDEX(attr))) ?		\
662 	((XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) : 0)
663 
664 /*
665  *  Modes.  Some values same as S_xxx entries from stat.h for convenience.
666  */
667 #define	VSUID		04000		/* set user id on execution */
668 #define	VSGID		02000		/* set group id on execution */
669 #define	VSVTX		01000		/* save swapped text even after use */
670 
671 /*
672  * Permissions.
673  */
674 #define	VREAD		00400
675 #define	VWRITE		00200
676 #define	VEXEC		00100
677 
678 #define	MODEMASK	07777		/* mode bits plus permission bits */
679 #define	PERMMASK	00777		/* permission bits */
680 
681 /*
682  * VOP_ACCESS flags
683  */
684 #define	V_ACE_MASK	0x1	/* mask represents  NFSv4 ACE permissions */
685 #define	V_APPEND	0x2	/* want to do append only check */
686 
687 /*
688  * Check whether mandatory file locking is enabled.
689  */
690 
691 #define	MANDMODE(mode)		(((mode) & (VSGID|(VEXEC>>3))) == VSGID)
692 #define	MANDLOCK(vp, mode)	((vp)->v_type == VREG && MANDMODE(mode))
693 
694 /*
695  * Flags for vnode operations.
696  */
697 enum rm		{ RMFILE, RMDIRECTORY };	/* rm or rmdir (remove) */
698 enum symfollow	{ NO_FOLLOW, FOLLOW };		/* follow symlinks (or not) */
699 enum vcexcl	{ NONEXCL, EXCL };		/* (non)excl create */
700 enum create	{ CRCREAT, CRMKNOD, CRMKDIR };	/* reason for create */
701 
702 typedef enum rm		rm_t;
703 typedef enum symfollow	symfollow_t;
704 typedef enum vcexcl	vcexcl_t;
705 typedef enum create	create_t;
706 
707 /* Vnode Events - Used by VOP_VNEVENT */
708 typedef enum vnevent	{
709 	VE_SUPPORT	= 0,	/* Query */
710 	VE_RENAME_SRC	= 1,	/* Rename, with vnode as source */
711 	VE_RENAME_DEST	= 2,	/* Rename, with vnode as target/destination */
712 	VE_REMOVE	= 3,	/* Remove of vnode's name */
713 	VE_RMDIR	= 4,	/* Remove of directory vnode's name */
714 	VE_CREATE	= 5,	/* Create with vnode's name which exists */
715 	VE_LINK		= 6, 	/* Link with vnode's name as source */
716 	VE_RENAME_DEST_DIR	= 7, 	/* Rename with vnode as target dir */
717 	VE_MOUNTEDOVER	= 8 	/* File or Filesystem got mounted over vnode */
718 } vnevent_t;
719 
720 /*
721  * Values for checking vnode open and map counts
722  */
723 enum v_mode { V_READ, V_WRITE, V_RDORWR, V_RDANDWR };
724 
725 typedef enum v_mode v_mode_t;
726 
727 #define	V_TRUE	1
728 #define	V_FALSE	0
729 
730 /*
731  * Structure used on VOP_GETSECATTR and VOP_SETSECATTR operations
732  */
733 
734 typedef struct vsecattr {
735 	uint_t		vsa_mask;	/* See below */
736 	int		vsa_aclcnt;	/* ACL entry count */
737 	void		*vsa_aclentp;	/* pointer to ACL entries */
738 	int		vsa_dfaclcnt;	/* default ACL entry count */
739 	void		*vsa_dfaclentp;	/* pointer to default ACL entries */
740 	size_t		vsa_aclentsz;	/* ACE size in bytes of vsa_aclentp */
741 	uint_t		vsa_aclflags;	/* ACE ACL flags */
742 } vsecattr_t;
743 
744 /* vsa_mask values */
745 #define	VSA_ACL			0x0001
746 #define	VSA_ACLCNT		0x0002
747 #define	VSA_DFACL		0x0004
748 #define	VSA_DFACLCNT		0x0008
749 #define	VSA_ACE			0x0010
750 #define	VSA_ACECNT		0x0020
751 #define	VSA_ACE_ALLTYPES	0x0040
752 #define	VSA_ACE_ACLFLAGS	0x0080	/* get/set ACE ACL flags */
753 
754 /*
755  * Structure used by various vnode operations to determine
756  * the context (pid, host, identity) of a caller.
757  *
758  * The cc_caller_id is used to identify one or more callers who invoke
759  * operations, possibly on behalf of others.  For example, the NFS
760  * server could have it's own cc_caller_id which can be detected by
761  * vnode/vfs operations or (FEM) monitors on those operations.  New
762  * caller IDs are generated by fs_new_caller_id().
763  */
764 typedef struct caller_context {
765 	pid_t		cc_pid;		/* Process ID of the caller */
766 	int		cc_sysid;	/* System ID, used for remote calls */
767 	u_longlong_t	cc_caller_id;	/* Identifier for (set of) caller(s) */
768 	ulong_t		cc_flags;
769 } caller_context_t;
770 
771 /*
772  * Flags for caller context.  The caller sets CC_DONTBLOCK if it does not
773  * want to block inside of a FEM monitor.  The monitor will set CC_WOULDBLOCK
774  * and return EAGAIN if the operation would have blocked.
775  */
776 #define	CC_WOULDBLOCK	0x01
777 #define	CC_DONTBLOCK	0x02
778 
779 /*
780  * Structure tags for function prototypes, defined elsewhere.
781  */
782 struct pathname;
783 struct fid;
784 struct flock64;
785 struct flk_callback;
786 struct shrlock;
787 struct page;
788 struct seg;
789 struct as;
790 struct pollhead;
791 struct taskq;
792 
793 #ifdef	_KERNEL
794 
795 /*
796  * VNODE_OPS defines all the vnode operations.  It is used to define
797  * the vnodeops structure (below) and the fs_func_p union (vfs_opreg.h).
798  */
799 #define	VNODE_OPS							\
800 	int	(*vop_open)(vnode_t **, int, cred_t *,			\
801 				caller_context_t *);			\
802 	int	(*vop_close)(vnode_t *, int, int, offset_t, cred_t *,	\
803 				caller_context_t *);			\
804 	int	(*vop_read)(vnode_t *, uio_t *, int, cred_t *,		\
805 				caller_context_t *);			\
806 	int	(*vop_write)(vnode_t *, uio_t *, int, cred_t *,		\
807 				caller_context_t *);			\
808 	int	(*vop_ioctl)(vnode_t *, int, intptr_t, int, cred_t *,	\
809 				int *, caller_context_t *);		\
810 	int	(*vop_setfl)(vnode_t *, int, int, cred_t *,		\
811 				caller_context_t *);			\
812 	int	(*vop_getattr)(vnode_t *, vattr_t *, int, cred_t *,	\
813 				caller_context_t *);			\
814 	int	(*vop_setattr)(vnode_t *, vattr_t *, int, cred_t *,	\
815 				caller_context_t *);			\
816 	int	(*vop_access)(vnode_t *, int, int, cred_t *,		\
817 				caller_context_t *);			\
818 	int	(*vop_lookup)(vnode_t *, char *, vnode_t **,		\
819 				struct pathname *,			\
820 				int, vnode_t *, cred_t *,		\
821 				caller_context_t *, int *,		\
822 				struct pathname *);			\
823 	int	(*vop_create)(vnode_t *, char *, vattr_t *, vcexcl_t,	\
824 				int, vnode_t **, cred_t *, int,		\
825 				caller_context_t *, vsecattr_t *);	\
826 	int	(*vop_remove)(vnode_t *, char *, cred_t *,		\
827 				caller_context_t *, int);		\
828 	int	(*vop_link)(vnode_t *, vnode_t *, char *, cred_t *,	\
829 				caller_context_t *, int);		\
830 	int	(*vop_rename)(vnode_t *, char *, vnode_t *, char *,	\
831 				cred_t *, caller_context_t *, int);	\
832 	int	(*vop_mkdir)(vnode_t *, char *, vattr_t *, vnode_t **,	\
833 				cred_t *, caller_context_t *, int,	\
834 				vsecattr_t *);				\
835 	int	(*vop_rmdir)(vnode_t *, char *, vnode_t *, cred_t *,	\
836 				caller_context_t *, int);		\
837 	int	(*vop_readdir)(vnode_t *, uio_t *, cred_t *, int *,	\
838 				caller_context_t *, int);		\
839 	int	(*vop_symlink)(vnode_t *, char *, vattr_t *, char *,	\
840 				cred_t *, caller_context_t *, int);	\
841 	int	(*vop_readlink)(vnode_t *, uio_t *, cred_t *,		\
842 				caller_context_t *);			\
843 	int	(*vop_fsync)(vnode_t *, int, cred_t *,			\
844 				caller_context_t *);			\
845 	void	(*vop_inactive)(vnode_t *, cred_t *,			\
846 				caller_context_t *);			\
847 	int	(*vop_fid)(vnode_t *, struct fid *,			\
848 				caller_context_t *);			\
849 	int	(*vop_rwlock)(vnode_t *, int, caller_context_t *);	\
850 	void	(*vop_rwunlock)(vnode_t *, int, caller_context_t *);	\
851 	int	(*vop_seek)(vnode_t *, offset_t, offset_t *,		\
852 				caller_context_t *);			\
853 	int	(*vop_cmp)(vnode_t *, vnode_t *, caller_context_t *);	\
854 	int	(*vop_frlock)(vnode_t *, int, struct flock64 *,		\
855 				int, offset_t,				\
856 				struct flk_callback *, cred_t *,	\
857 				caller_context_t *);			\
858 	int	(*vop_space)(vnode_t *, int, struct flock64 *,		\
859 				int, offset_t,				\
860 				cred_t *, caller_context_t *);		\
861 	int	(*vop_realvp)(vnode_t *, vnode_t **,			\
862 				caller_context_t *);			\
863 	int	(*vop_getpage)(vnode_t *, offset_t, size_t, uint_t *,	\
864 				struct page **, size_t, struct seg *,	\
865 				caddr_t, enum seg_rw, cred_t *,		\
866 				caller_context_t *);			\
867 	int	(*vop_putpage)(vnode_t *, offset_t, size_t,		\
868 				int, cred_t *, caller_context_t *);	\
869 	int	(*vop_map)(vnode_t *, offset_t, struct as *,		\
870 				caddr_t *, size_t,			\
871 				uchar_t, uchar_t, uint_t, cred_t *,	\
872 				caller_context_t *);			\
873 	int	(*vop_addmap)(vnode_t *, offset_t, struct as *,		\
874 				caddr_t, size_t,			\
875 				uchar_t, uchar_t, uint_t, cred_t *,	\
876 				caller_context_t *);			\
877 	int	(*vop_delmap)(vnode_t *, offset_t, struct as *,		\
878 				caddr_t, size_t,			\
879 				uint_t, uint_t, uint_t, cred_t *,	\
880 				caller_context_t *);			\
881 	int	(*vop_poll)(vnode_t *, short, int, short *,		\
882 				struct pollhead **,			\
883 				caller_context_t *);			\
884 	int	(*vop_dump)(vnode_t *, caddr_t, offset_t, offset_t,	\
885 				caller_context_t *);			\
886 	int	(*vop_pathconf)(vnode_t *, int, ulong_t *, cred_t *,	\
887 				caller_context_t *);			\
888 	int	(*vop_pageio)(vnode_t *, struct page *,			\
889 				u_offset_t, size_t, int, cred_t *,	\
890 				caller_context_t *);			\
891 	int	(*vop_dumpctl)(vnode_t *, int, offset_t *,		\
892 				caller_context_t *);			\
893 	void	(*vop_dispose)(vnode_t *, struct page *,		\
894 				int, int, cred_t *,			\
895 				caller_context_t *);			\
896 	int	(*vop_setsecattr)(vnode_t *, vsecattr_t *,		\
897 				int, cred_t *, caller_context_t *);	\
898 	int	(*vop_getsecattr)(vnode_t *, vsecattr_t *,		\
899 				int, cred_t *, caller_context_t *);	\
900 	int	(*vop_shrlock)(vnode_t *, int, struct shrlock *,	\
901 				int, cred_t *, caller_context_t *);	\
902 	int	(*vop_vnevent)(vnode_t *, vnevent_t, vnode_t *,		\
903 				char *, caller_context_t *)
904 	/* NB: No ";" */
905 
906 /*
907  * Operations on vnodes.  Note: File systems must never operate directly
908  * on a 'vnodeops' structure -- it WILL change in future releases!  They
909  * must use vn_make_ops() to create the structure.
910  */
911 typedef struct vnodeops {
912 	const char *vnop_name;
913 	VNODE_OPS;	/* Signatures of all vnode operations (vops) */
914 } vnodeops_t;
915 
916 typedef int (*fs_generic_func_p) ();	/* Generic vop/vfsop/femop/fsemop ptr */
917 
918 extern int	fop_open(vnode_t **, int, cred_t *, caller_context_t *);
919 extern int	fop_close(vnode_t *, int, int, offset_t, cred_t *,
920 				caller_context_t *);
921 extern int	fop_read(vnode_t *, uio_t *, int, cred_t *, caller_context_t *);
922 extern int	fop_write(vnode_t *, uio_t *, int, cred_t *,
923 				caller_context_t *);
924 extern int	fop_ioctl(vnode_t *, int, intptr_t, int, cred_t *, int *,
925 				caller_context_t *);
926 extern int	fop_setfl(vnode_t *, int, int, cred_t *, caller_context_t *);
927 extern int	fop_getattr(vnode_t *, vattr_t *, int, cred_t *,
928 				caller_context_t *);
929 extern int	fop_setattr(vnode_t *, vattr_t *, int, cred_t *,
930 				caller_context_t *);
931 extern int	fop_access(vnode_t *, int, int, cred_t *, caller_context_t *);
932 extern int	fop_lookup(vnode_t *, char *, vnode_t **, struct pathname *,
933 				int, vnode_t *, cred_t *, caller_context_t *,
934 				int *, struct pathname *);
935 extern int	fop_create(vnode_t *, char *, vattr_t *, vcexcl_t, int,
936 				vnode_t **, cred_t *, int, caller_context_t *,
937 				vsecattr_t *);
938 extern int	fop_remove(vnode_t *vp, char *, cred_t *, caller_context_t *,
939 				int);
940 extern int	fop_link(vnode_t *, vnode_t *, char *, cred_t *,
941 				caller_context_t *, int);
942 extern int	fop_rename(vnode_t *, char *, vnode_t *, char *, cred_t *,
943 				caller_context_t *, int);
944 extern int	fop_mkdir(vnode_t *, char *, vattr_t *, vnode_t **, cred_t *,
945 				caller_context_t *, int, vsecattr_t *);
946 extern int	fop_rmdir(vnode_t *, char *, vnode_t *, cred_t *,
947 				caller_context_t *, int);
948 extern int	fop_readdir(vnode_t *, uio_t *, cred_t *, int *,
949 				caller_context_t *, int);
950 extern int	fop_symlink(vnode_t *, char *, vattr_t *, char *, cred_t *,
951 				caller_context_t *, int);
952 extern int	fop_readlink(vnode_t *, uio_t *, cred_t *, caller_context_t *);
953 extern int	fop_fsync(vnode_t *, int, cred_t *, caller_context_t *);
954 extern void	fop_inactive(vnode_t *, cred_t *, caller_context_t *);
955 extern int	fop_fid(vnode_t *, struct fid *, caller_context_t *);
956 extern int	fop_rwlock(vnode_t *, int, caller_context_t *);
957 extern void	fop_rwunlock(vnode_t *, int, caller_context_t *);
958 extern int	fop_seek(vnode_t *, offset_t, offset_t *, caller_context_t *);
959 extern int	fop_cmp(vnode_t *, vnode_t *, caller_context_t *);
960 extern int	fop_frlock(vnode_t *, int, struct flock64 *, int, offset_t,
961 				struct flk_callback *, cred_t *,
962 				caller_context_t *);
963 extern int	fop_space(vnode_t *, int, struct flock64 *, int, offset_t,
964 				cred_t *, caller_context_t *);
965 extern int	fop_realvp(vnode_t *, vnode_t **, caller_context_t *);
966 extern int	fop_getpage(vnode_t *, offset_t, size_t, uint_t *,
967 				struct page **, size_t, struct seg *,
968 				caddr_t, enum seg_rw, cred_t *,
969 				caller_context_t *);
970 extern int	fop_putpage(vnode_t *, offset_t, size_t, int, cred_t *,
971 				caller_context_t *);
972 extern int	fop_map(vnode_t *, offset_t, struct as *, caddr_t *, size_t,
973 				uchar_t, uchar_t, uint_t, cred_t *cr,
974 				caller_context_t *);
975 extern int	fop_addmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
976 				uchar_t, uchar_t, uint_t, cred_t *,
977 				caller_context_t *);
978 extern int	fop_delmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
979 				uint_t, uint_t, uint_t, cred_t *,
980 				caller_context_t *);
981 extern int	fop_poll(vnode_t *, short, int, short *, struct pollhead **,
982 				caller_context_t *);
983 extern int	fop_dump(vnode_t *, caddr_t, offset_t, offset_t,
984     caller_context_t *);
985 extern int	fop_pathconf(vnode_t *, int, ulong_t *, cred_t *,
986 				caller_context_t *);
987 extern int	fop_pageio(vnode_t *, struct page *, u_offset_t, size_t, int,
988 				cred_t *, caller_context_t *);
989 extern int	fop_dumpctl(vnode_t *, int, offset_t *, caller_context_t *);
990 extern void	fop_dispose(vnode_t *, struct page *, int, int, cred_t *,
991 				caller_context_t *);
992 extern int	fop_setsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
993 				caller_context_t *);
994 extern int	fop_getsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
995 				caller_context_t *);
996 extern int	fop_shrlock(vnode_t *, int, struct shrlock *, int, cred_t *,
997 				caller_context_t *);
998 extern int	fop_vnevent(vnode_t *, vnevent_t, vnode_t *, char *,
999 				caller_context_t *);
1000 
1001 #endif	/* _KERNEL */
1002 
1003 #define	VOP_OPEN(vpp, mode, cr, ct) \
1004 	fop_open(vpp, mode, cr, ct)
1005 #define	VOP_CLOSE(vp, f, c, o, cr, ct) \
1006 	fop_close(vp, f, c, o, cr, ct)
1007 #define	VOP_READ(vp, uiop, iof, cr, ct) \
1008 	fop_read(vp, uiop, iof, cr, ct)
1009 #define	VOP_WRITE(vp, uiop, iof, cr, ct) \
1010 	fop_write(vp, uiop, iof, cr, ct)
1011 #define	VOP_IOCTL(vp, cmd, a, f, cr, rvp, ct) \
1012 	fop_ioctl(vp, cmd, a, f, cr, rvp, ct)
1013 #define	VOP_SETFL(vp, f, a, cr, ct) \
1014 	fop_setfl(vp, f, a, cr, ct)
1015 #define	VOP_GETATTR(vp, vap, f, cr, ct) \
1016 	fop_getattr(vp, vap, f, cr, ct)
1017 #define	VOP_SETATTR(vp, vap, f, cr, ct) \
1018 	fop_setattr(vp, vap, f, cr, ct)
1019 #define	VOP_ACCESS(vp, mode, f, cr, ct) \
1020 	fop_access(vp, mode, f, cr, ct)
1021 #define	VOP_LOOKUP(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp) \
1022 	fop_lookup(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp)
1023 #define	VOP_CREATE(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap) \
1024 	fop_create(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap)
1025 #define	VOP_REMOVE(dvp, p, cr, ct, f) \
1026 	fop_remove(dvp, p, cr, ct, f)
1027 #define	VOP_LINK(tdvp, fvp, p, cr, ct, f) \
1028 	fop_link(tdvp, fvp, p, cr, ct, f)
1029 #define	VOP_RENAME(fvp, fnm, tdvp, tnm, cr, ct, f) \
1030 	fop_rename(fvp, fnm, tdvp, tnm, cr, ct, f)
1031 #define	VOP_MKDIR(dp, p, vap, vpp, cr, ct, f, vsap) \
1032 	fop_mkdir(dp, p, vap, vpp, cr, ct, f, vsap)
1033 #define	VOP_RMDIR(dp, p, cdir, cr, ct, f) \
1034 	fop_rmdir(dp, p, cdir, cr, ct, f)
1035 #define	VOP_READDIR(vp, uiop, cr, eofp, ct, f) \
1036 	fop_readdir(vp, uiop, cr, eofp, ct, f)
1037 #define	VOP_SYMLINK(dvp, lnm, vap, tnm, cr, ct, f) \
1038 	fop_symlink(dvp, lnm, vap, tnm, cr, ct, f)
1039 #define	VOP_READLINK(vp, uiop, cr, ct) \
1040 	fop_readlink(vp, uiop, cr, ct)
1041 #define	VOP_FSYNC(vp, syncflag, cr, ct) \
1042 	fop_fsync(vp, syncflag, cr, ct)
1043 #define	VOP_INACTIVE(vp, cr, ct) \
1044 	fop_inactive(vp, cr, ct)
1045 #define	VOP_FID(vp, fidp, ct) \
1046 	fop_fid(vp, fidp, ct)
1047 #define	VOP_RWLOCK(vp, w, ct) \
1048 	fop_rwlock(vp, w, ct)
1049 #define	VOP_RWUNLOCK(vp, w, ct) \
1050 	fop_rwunlock(vp, w, ct)
1051 #define	VOP_SEEK(vp, ooff, noffp, ct) \
1052 	fop_seek(vp, ooff, noffp, ct)
1053 #define	VOP_CMP(vp1, vp2, ct) \
1054 	fop_cmp(vp1, vp2, ct)
1055 #define	VOP_FRLOCK(vp, cmd, a, f, o, cb, cr, ct) \
1056 	fop_frlock(vp, cmd, a, f, o, cb, cr, ct)
1057 #define	VOP_SPACE(vp, cmd, a, f, o, cr, ct) \
1058 	fop_space(vp, cmd, a, f, o, cr, ct)
1059 #define	VOP_REALVP(vp1, vp2, ct) \
1060 	fop_realvp(vp1, vp2, ct)
1061 #define	VOP_GETPAGE(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct) \
1062 	fop_getpage(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct)
1063 #define	VOP_PUTPAGE(vp, of, sz, fl, cr, ct) \
1064 	fop_putpage(vp, of, sz, fl, cr, ct)
1065 #define	VOP_MAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1066 	fop_map(vp, of, as, a, sz, p, mp, fl, cr, ct)
1067 #define	VOP_ADDMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1068 	fop_addmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1069 #define	VOP_DELMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1070 	fop_delmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1071 #define	VOP_POLL(vp, events, anyyet, reventsp, phpp, ct) \
1072 	fop_poll(vp, events, anyyet, reventsp, phpp, ct)
1073 #define	VOP_DUMP(vp, addr, bn, count, ct) \
1074 	fop_dump(vp, addr, bn, count, ct)
1075 #define	VOP_PATHCONF(vp, cmd, valp, cr, ct) \
1076 	fop_pathconf(vp, cmd, valp, cr, ct)
1077 #define	VOP_PAGEIO(vp, pp, io_off, io_len, flags, cr, ct) \
1078 	fop_pageio(vp, pp, io_off, io_len, flags, cr, ct)
1079 #define	VOP_DUMPCTL(vp, action, blkp, ct) \
1080 	fop_dumpctl(vp, action, blkp, ct)
1081 #define	VOP_DISPOSE(vp, pp, flag, dn, cr, ct) \
1082 	fop_dispose(vp, pp, flag, dn, cr, ct)
1083 #define	VOP_GETSECATTR(vp, vsap, f, cr, ct) \
1084 	fop_getsecattr(vp, vsap, f, cr, ct)
1085 #define	VOP_SETSECATTR(vp, vsap, f, cr, ct) \
1086 	fop_setsecattr(vp, vsap, f, cr, ct)
1087 #define	VOP_SHRLOCK(vp, cmd, shr, f, cr, ct) \
1088 	fop_shrlock(vp, cmd, shr, f, cr, ct)
1089 #define	VOP_VNEVENT(vp, vnevent, dvp, fnm, ct) \
1090 	fop_vnevent(vp, vnevent, dvp, fnm, ct)
1091 
1092 #define	VOPNAME_OPEN		"open"
1093 #define	VOPNAME_CLOSE		"close"
1094 #define	VOPNAME_READ		"read"
1095 #define	VOPNAME_WRITE		"write"
1096 #define	VOPNAME_IOCTL		"ioctl"
1097 #define	VOPNAME_SETFL		"setfl"
1098 #define	VOPNAME_GETATTR		"getattr"
1099 #define	VOPNAME_SETATTR		"setattr"
1100 #define	VOPNAME_ACCESS		"access"
1101 #define	VOPNAME_LOOKUP		"lookup"
1102 #define	VOPNAME_CREATE		"create"
1103 #define	VOPNAME_REMOVE		"remove"
1104 #define	VOPNAME_LINK		"link"
1105 #define	VOPNAME_RENAME		"rename"
1106 #define	VOPNAME_MKDIR		"mkdir"
1107 #define	VOPNAME_RMDIR		"rmdir"
1108 #define	VOPNAME_READDIR		"readdir"
1109 #define	VOPNAME_SYMLINK		"symlink"
1110 #define	VOPNAME_READLINK	"readlink"
1111 #define	VOPNAME_FSYNC		"fsync"
1112 #define	VOPNAME_INACTIVE	"inactive"
1113 #define	VOPNAME_FID		"fid"
1114 #define	VOPNAME_RWLOCK		"rwlock"
1115 #define	VOPNAME_RWUNLOCK	"rwunlock"
1116 #define	VOPNAME_SEEK		"seek"
1117 #define	VOPNAME_CMP		"cmp"
1118 #define	VOPNAME_FRLOCK		"frlock"
1119 #define	VOPNAME_SPACE		"space"
1120 #define	VOPNAME_REALVP		"realvp"
1121 #define	VOPNAME_GETPAGE		"getpage"
1122 #define	VOPNAME_PUTPAGE		"putpage"
1123 #define	VOPNAME_MAP		"map"
1124 #define	VOPNAME_ADDMAP		"addmap"
1125 #define	VOPNAME_DELMAP		"delmap"
1126 #define	VOPNAME_POLL		"poll"
1127 #define	VOPNAME_DUMP		"dump"
1128 #define	VOPNAME_PATHCONF	"pathconf"
1129 #define	VOPNAME_PAGEIO		"pageio"
1130 #define	VOPNAME_DUMPCTL		"dumpctl"
1131 #define	VOPNAME_DISPOSE		"dispose"
1132 #define	VOPNAME_GETSECATTR	"getsecattr"
1133 #define	VOPNAME_SETSECATTR	"setsecattr"
1134 #define	VOPNAME_SHRLOCK		"shrlock"
1135 #define	VOPNAME_VNEVENT		"vnevent"
1136 
1137 /*
1138  * Flags for VOP_LOOKUP
1139  *
1140  * Defined in file.h, but also possible, FIGNORECASE
1141  *
1142  */
1143 #define	LOOKUP_DIR		0x01	/* want parent dir vp */
1144 #define	LOOKUP_XATTR		0x02	/* lookup up extended attr dir */
1145 #define	CREATE_XATTR_DIR	0x04	/* Create extended attr dir */
1146 #define	LOOKUP_HAVE_SYSATTR_DIR	0x08	/* Already created virtual GFS dir */
1147 
1148 /*
1149  * Flags for VOP_READDIR
1150  */
1151 #define	V_RDDIR_ENTFLAGS	0x01	/* request dirent flags */
1152 #define	V_RDDIR_ACCFILTER	0x02	/* filter out inaccessible dirents */
1153 
1154 /*
1155  * Flags for VOP_RWLOCK/VOP_RWUNLOCK
1156  * VOP_RWLOCK will return the flag that was actually set, or -1 if none.
1157  */
1158 #define	V_WRITELOCK_TRUE	(1)	/* Request write-lock on the vnode */
1159 #define	V_WRITELOCK_FALSE	(0)	/* Request read-lock on the vnode */
1160 
1161 /*
1162  * Flags for VOP_DUMPCTL
1163  */
1164 #define	DUMP_ALLOC	0
1165 #define	DUMP_FREE	1
1166 #define	DUMP_SCAN	2
1167 
1168 /*
1169  * Public vnode manipulation functions.
1170  */
1171 #ifdef	_KERNEL
1172 
1173 vnode_t *vn_alloc(int);
1174 void	vn_reinit(vnode_t *);
1175 void	vn_recycle(vnode_t *);
1176 void	vn_free(vnode_t *);
1177 
1178 int	vn_is_readonly(vnode_t *);
1179 int   	vn_is_opened(vnode_t *, v_mode_t);
1180 int   	vn_is_mapped(vnode_t *, v_mode_t);
1181 int   	vn_has_other_opens(vnode_t *, v_mode_t);
1182 void	vn_open_upgrade(vnode_t *, int);
1183 void	vn_open_downgrade(vnode_t *, int);
1184 
1185 int	vn_can_change_zones(vnode_t *vp);
1186 
1187 int	vn_has_flocks(vnode_t *);
1188 int	vn_has_mandatory_locks(vnode_t *, int);
1189 int	vn_has_cached_data(vnode_t *);
1190 
1191 void	vn_setops(vnode_t *, vnodeops_t *);
1192 vnodeops_t *vn_getops(vnode_t *);
1193 int	vn_matchops(vnode_t *, vnodeops_t *);
1194 int	vn_matchopval(vnode_t *, char *, fs_generic_func_p);
1195 int	vn_ismntpt(vnode_t *);
1196 
1197 struct vfs *vn_mountedvfs(vnode_t *);
1198 
1199 int	vn_in_dnlc(vnode_t *);
1200 
1201 void	vn_create_cache(void);
1202 void	vn_destroy_cache(void);
1203 
1204 void	vn_freevnodeops(vnodeops_t *);
1205 
1206 int	vn_open(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1207 		struct vnode **vpp, enum create crwhy, mode_t umask);
1208 int	vn_openat(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1209 		struct vnode **vpp, enum create crwhy,
1210 		mode_t umask, struct vnode *startvp, int fd);
1211 int	vn_create(char *pnamep, enum uio_seg seg, struct vattr *vap,
1212 		enum vcexcl excl, int mode, struct vnode **vpp,
1213 		enum create why, int flag, mode_t umask);
1214 int	vn_createat(char *pnamep, enum uio_seg seg, struct vattr *vap,
1215 		enum vcexcl excl, int mode, struct vnode **vpp,
1216 		enum create why, int flag, mode_t umask, struct vnode *startvp);
1217 int	vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base, ssize_t len,
1218 		offset_t offset, enum uio_seg seg, int ioflag, rlim64_t ulimit,
1219 		cred_t *cr, ssize_t *residp);
1220 void	vn_rele(struct vnode *vp);
1221 void	vn_rele_async(struct vnode *vp, struct taskq *taskq);
1222 void	vn_rele_dnlc(struct vnode *vp);
1223 void	vn_rele_stream(struct vnode *vp);
1224 int	vn_link(char *from, char *to, enum uio_seg seg);
1225 int	vn_rename(char *from, char *to, enum uio_seg seg);
1226 int	vn_renameat(vnode_t *fdvp, char *fname, vnode_t *tdvp, char *tname,
1227 		enum uio_seg seg);
1228 int	vn_remove(char *fnamep, enum uio_seg seg, enum rm dirflag);
1229 int	vn_removeat(vnode_t *startvp, char *fnamep, enum uio_seg seg,
1230 		enum rm dirflag);
1231 int	vn_compare(vnode_t *vp1, vnode_t *vp2);
1232 int	vn_vfswlock(struct vnode *vp);
1233 int	vn_vfswlock_wait(struct vnode *vp);
1234 int	vn_vfsrlock(struct vnode *vp);
1235 int	vn_vfsrlock_wait(struct vnode *vp);
1236 void	vn_vfsunlock(struct vnode *vp);
1237 int	vn_vfswlock_held(struct vnode *vp);
1238 vnode_t *specvp(struct vnode *vp, dev_t dev, vtype_t type, struct cred *cr);
1239 vnode_t *makespecvp(dev_t dev, vtype_t type);
1240 vn_vfslocks_entry_t *vn_vfslocks_getlock(void *);
1241 void	vn_vfslocks_rele(vn_vfslocks_entry_t *);
1242 boolean_t vn_is_reparse(vnode_t *, cred_t *, caller_context_t *);
1243 
1244 void vn_copypath(struct vnode *src, struct vnode *dst);
1245 void vn_setpath_str(struct vnode *vp, const char *str, size_t len);
1246 void vn_setpath(vnode_t *rootvp, struct vnode *startvp, struct vnode *vp,
1247     const char *path, size_t plen);
1248 void vn_renamepath(vnode_t *dvp, vnode_t *vp, const char *nm, size_t len);
1249 
1250 /* Vnode event notification */
1251 void	vnevent_rename_src(vnode_t *, vnode_t *, char *, caller_context_t *);
1252 void	vnevent_rename_dest(vnode_t *, vnode_t *, char *, caller_context_t *);
1253 void	vnevent_remove(vnode_t *, vnode_t *, char *, caller_context_t *);
1254 void	vnevent_rmdir(vnode_t *, vnode_t *, char *, caller_context_t *);
1255 void	vnevent_create(vnode_t *, caller_context_t *);
1256 void	vnevent_link(vnode_t *, caller_context_t *);
1257 void	vnevent_rename_dest_dir(vnode_t *, caller_context_t *ct);
1258 void	vnevent_mountedover(vnode_t *, caller_context_t *);
1259 int	vnevent_support(vnode_t *, caller_context_t *);
1260 
1261 /* Vnode specific data */
1262 void vsd_create(uint_t *, void (*)(void *));
1263 void vsd_destroy(uint_t *);
1264 void *vsd_get(vnode_t *, uint_t);
1265 int vsd_set(vnode_t *, uint_t, void *);
1266 void vsd_free(vnode_t *);
1267 
1268 /*
1269  * Extensible vnode attribute (xva) routines:
1270  * xva_init() initializes an xvattr_t (zero struct, init mapsize, set AT_XATTR)
1271  * xva_getxoptattr() returns a ponter to the xoptattr_t section of xvattr_t
1272  */
1273 void		xva_init(xvattr_t *);
1274 xoptattr_t	*xva_getxoptattr(xvattr_t *);	/* Get ptr to xoptattr_t */
1275 
1276 void xattr_init(void);		/* Initialize vnodeops for xattrs */
1277 
1278 /* GFS tunnel for xattrs */
1279 int xattr_dir_lookup(vnode_t *, vnode_t **, int, cred_t *);
1280 
1281 /* Reparse Point */
1282 void reparse_point_init(void);
1283 
1284 /* Context identification */
1285 u_longlong_t	fs_new_caller_id();
1286 
1287 int	vn_vmpss_usepageio(vnode_t *);
1288 
1289 /*
1290  * Needed for use of IS_VMODSORT() in kernel.
1291  */
1292 extern uint_t pvn_vmodsort_supported;
1293 
1294 #define	VN_HOLD(vp)	{ \
1295 	mutex_enter(&(vp)->v_lock); \
1296 	(vp)->v_count++; \
1297 	mutex_exit(&(vp)->v_lock); \
1298 }
1299 
1300 #define	VN_RELE(vp)	{ \
1301 	vn_rele(vp); \
1302 }
1303 
1304 #define	VN_RELE_ASYNC(vp, taskq)	{ \
1305 	vn_rele_async(vp, taskq); \
1306 }
1307 
1308 #define	VN_SET_VFS_TYPE_DEV(vp, vfsp, type, dev)	{ \
1309 	(vp)->v_vfsp = (vfsp); \
1310 	(vp)->v_type = (type); \
1311 	(vp)->v_rdev = (dev); \
1312 }
1313 
1314 /*
1315  * Compare two vnodes for equality.  In general this macro should be used
1316  * in preference to calling VOP_CMP directly.
1317  */
1318 #define	VN_CMP(VP1, VP2)	((VP1) == (VP2) ? 1 : 	\
1319 	((VP1) && (VP2) && (vn_getops(VP1) == vn_getops(VP2)) ? \
1320 	VOP_CMP(VP1, VP2, NULL) : 0))
1321 
1322 /*
1323  * Some well-known global vnodes used by the VM system to name pages.
1324  */
1325 extern struct vnode kvps[];
1326 
1327 typedef enum {
1328 	KV_KVP,		/* vnode for all segkmem pages */
1329 	KV_ZVP,		/* vnode for all ZFS pages */
1330 #if defined(__sparc)
1331 	KV_MPVP,	/* vnode for all page_t meta-pages */
1332 	KV_PROMVP,	/* vnode for all PROM pages */
1333 #endif	/* __sparc */
1334 	KV_MAX		/* total number of vnodes in kvps[] */
1335 } kvps_index_t;
1336 
1337 #define	VN_ISKAS(vp)	((vp) >= &kvps[0] && (vp) < &kvps[KV_MAX])
1338 
1339 #endif	/* _KERNEL */
1340 
1341 /*
1342  * Flags to VOP_SETATTR/VOP_GETATTR.
1343  */
1344 #define	ATTR_UTIME	0x01	/* non-default utime(2) request */
1345 #define	ATTR_EXEC	0x02	/* invocation from exec(2) */
1346 #define	ATTR_COMM	0x04	/* yield common vp attributes */
1347 #define	ATTR_HINT	0x08	/* information returned will be `hint' */
1348 #define	ATTR_REAL	0x10	/* yield attributes of the real vp */
1349 #define	ATTR_NOACLCHECK	0x20	/* Don't check ACL when checking permissions */
1350 #define	ATTR_TRIGGER	0x40	/* Mount first if vnode is a trigger mount */
1351 /*
1352  * Generally useful macros.
1353  */
1354 #define	VBSIZE(vp)	((vp)->v_vfsp->vfs_bsize)
1355 
1356 #define	VTOZONE(vp)	((vp)->v_vfsp->vfs_zone)
1357 
1358 #define	NULLVP		((struct vnode *)0)
1359 #define	NULLVPP		((struct vnode **)0)
1360 
1361 #ifdef	_KERNEL
1362 
1363 /*
1364  * Structure used while handling asynchronous VOP_PUTPAGE operations.
1365  */
1366 struct async_reqs {
1367 	struct async_reqs *a_next;	/* pointer to next arg struct */
1368 	struct vnode *a_vp;		/* vnode pointer */
1369 	u_offset_t a_off;			/* offset in file */
1370 	uint_t a_len;			/* size of i/o request */
1371 	int a_flags;			/* flags to indicate operation type */
1372 	struct cred *a_cred;		/* cred pointer	*/
1373 	ushort_t a_prealloced;		/* set if struct is pre-allocated */
1374 };
1375 
1376 /*
1377  * VN_DISPOSE() -- given a page pointer, safely invoke VOP_DISPOSE().
1378  * Note that there is no guarantee that the page passed in will be
1379  * freed.  If that is required, then a check after calling VN_DISPOSE would
1380  * be necessary to ensure the page was freed.
1381  */
1382 #define	VN_DISPOSE(pp, flag, dn, cr)	{ \
1383 	if ((pp)->p_vnode != NULL && !VN_ISKAS((pp)->p_vnode)) \
1384 		VOP_DISPOSE((pp)->p_vnode, (pp), (flag), (dn), (cr), NULL); \
1385 	else if ((flag) == B_FREE) \
1386 		page_free((pp), (dn)); \
1387 	else \
1388 		page_destroy((pp), (dn)); \
1389 	}
1390 
1391 #endif	/* _KERNEL */
1392 
1393 #ifdef	__cplusplus
1394 }
1395 #endif
1396 
1397 #endif	/* _SYS_VNODE_H */
1398