xref: /titanic_50/usr/src/uts/common/sys/vnode.h (revision 6343349bf9d42501532d1ee2e75718d86ee366fe)
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 2007 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 #pragma ident	"%Z%%M%	%I%	%E% SMI"
43 
44 #include <sys/types.h>
45 #include <sys/t_lock.h>
46 #include <sys/rwstlock.h>
47 #include <sys/time_impl.h>
48 #include <sys/cred.h>
49 #include <sys/uio.h>
50 #include <sys/resource.h>
51 #include <vm/seg_enum.h>
52 #include <sys/kstat.h>
53 #include <sys/kmem.h>
54 #include <sys/list.h>
55 #ifdef	_KERNEL
56 #include <sys/buf.h>
57 #endif	/* _KERNEL */
58 
59 #ifdef	__cplusplus
60 extern "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 
75 typedef 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 } vopstats_t;
124 
125 /*
126  * The vnode is the focus of all file activity in UNIX.
127  * A vnode is allocated for each active file, each current
128  * directory, each mounted-on file, and the root.
129  *
130  * Each vnode is usually associated with a file-system-specific node (for
131  * UFS, this is the in-memory inode).  Generally, a vnode and an fs-node
132  * should be created and destroyed together as a pair.
133  *
134  * If a vnode is reused for a new file, it should be reinitialized by calling
135  * either vn_reinit() or vn_recycle().
136  *
137  * vn_reinit() resets the entire vnode as if it was returned by vn_alloc().
138  * The caller is responsible for setting up the entire vnode after calling
139  * vn_reinit().  This is important when using kmem caching where the vnode is
140  * allocated by a constructor, for instance.
141  *
142  * vn_recycle() is used when the file system keeps some state around in both
143  * the vnode and the associated FS-node.  In UFS, for example, the inode of
144  * a deleted file can be reused immediately.  The v_data, v_vfsp, v_op, etc.
145  * remains the same but certain fields related to the previous instance need
146  * to be reset.  In particular:
147  *	v_femhead
148  *	v_path
149  *	v_rdcnt, v_wrcnt
150  *	v_mmap_read, v_mmap_write
151  */
152 
153 /*
154  * vnode types.  VNON means no type.  These values are unrelated to
155  * values in on-disk inodes.
156  */
157 typedef enum vtype {
158 	VNON	= 0,
159 	VREG	= 1,
160 	VDIR	= 2,
161 	VBLK	= 3,
162 	VCHR	= 4,
163 	VLNK	= 5,
164 	VFIFO	= 6,
165 	VDOOR	= 7,
166 	VPROC	= 8,
167 	VSOCK	= 9,
168 	VPORT	= 10,
169 	VBAD	= 11
170 } vtype_t;
171 
172 /*
173  * VSD - Vnode Specific Data
174  * Used to associate additional private data with a vnode.
175  */
176 struct vsd_node {
177 	list_node_t vs_nodes;		/* list of all VSD nodes */
178 	uint_t vs_nkeys;		/* entries in value array */
179 	void **vs_value;		/* array of value/key */
180 };
181 
182 /*
183  * Many of the fields in the vnode are read-only once they are initialized
184  * at vnode creation time.  Other fields are protected by locks.
185  *
186  * IMPORTANT: vnodes should be created ONLY by calls to vn_alloc().  They
187  * may not be embedded into the file-system specific node (inode).  The
188  * size of vnodes may change.
189  *
190  * The v_lock protects:
191  *   v_flag
192  *   v_stream
193  *   v_count
194  *   v_shrlocks
195  *   v_path
196  *   v_vsd
197  *   v_xattrdir
198  *
199  * A special lock (implemented by vn_vfswlock in vnode.c) protects:
200  *   v_vfsmountedhere
201  *
202  * The global flock_lock mutex (in flock.c) protects:
203  *   v_filocks
204  *
205  * IMPORTANT NOTE:
206  *
207  *   The following vnode fields are considered public and may safely be
208  *   accessed by file systems or other consumers:
209  *
210  *     v_lock
211  *     v_flag
212  *     v_count
213  *     v_data
214  *     v_vfsp
215  *     v_stream
216  *     v_type
217  *     v_rdev
218  *
219  * ALL OTHER FIELDS SHOULD BE ACCESSED ONLY BY THE OWNER OF THAT FIELD.
220  * In particular, file systems should not access other fields; they may
221  * change or even be removed.  The functionality which was once provided
222  * by these fields is available through vn_* functions.
223  */
224 
225 struct fem_head;	/* from fem.h */
226 
227 typedef struct vnode {
228 	kmutex_t	v_lock;		/* protects vnode fields */
229 	uint_t		v_flag;		/* vnode flags (see below) */
230 	uint_t		v_count;	/* reference count */
231 	void		*v_data;	/* private data for fs */
232 	struct vfs	*v_vfsp;	/* ptr to containing VFS */
233 	struct stdata	*v_stream;	/* associated stream */
234 	enum vtype	v_type;		/* vnode type */
235 	dev_t		v_rdev;		/* device (VCHR, VBLK) */
236 
237 	/* PRIVATE FIELDS BELOW - DO NOT USE */
238 
239 	struct vfs	*v_vfsmountedhere; /* ptr to vfs mounted here */
240 	struct vnodeops	*v_op;		/* vnode operations */
241 	struct page	*v_pages;	/* vnode pages list */
242 	pgcnt_t		v_npages;	/* # pages on this vnode */
243 	pgcnt_t		v_msnpages;	/* # pages charged to v_mset */
244 	struct page	*v_scanfront;	/* scanner front hand */
245 	struct page	*v_scanback;	/* scanner back hand */
246 	struct filock	*v_filocks;	/* ptr to filock list */
247 	struct shrlocklist *v_shrlocks;	/* ptr to shrlock list */
248 	krwlock_t	v_nbllock;	/* sync for NBMAND locks */
249 	kcondvar_t	v_cv;		/* synchronize locking */
250 	void		*v_locality;	/* hook for locality info */
251 	struct fem_head	*v_femhead;	/* fs monitoring */
252 	char		*v_path;	/* cached path */
253 	uint_t		v_rdcnt;	/* open for read count  (VREG only) */
254 	uint_t		v_wrcnt;	/* open for write count (VREG only) */
255 	u_longlong_t	v_mmap_read;	/* mmap read count */
256 	u_longlong_t	v_mmap_write;	/* mmap write count */
257 	void		*v_mpssdata;	/* info for large page mappings */
258 	hrtime_t	v_scantime;	/* last time this vnode was scanned */
259 	ushort_t	v_mset;		/* memory set ID */
260 	uint_t		v_msflags;	/* memory set flags */
261 	struct vnode	*v_msnext;	/* list of vnodes on an mset */
262 	struct vnode	*v_msprev;	/* list of vnodes on an mset */
263 	krwlock_t	v_mslock;	/* protects v_mset */
264 	void		*v_fopdata;	/* list of file ops event watches */
265 	struct vsd_node *v_vsd;		/* vnode specific data */
266 	struct vnode	*v_xattrdir;	/* unnamed extended attr dir (GFS) */
267 } vnode_t;
268 
269 #define	IS_DEVVP(vp)	\
270 	((vp)->v_type == VCHR || (vp)->v_type == VBLK || (vp)->v_type == VFIFO)
271 
272 /*
273  * vnode flags.
274  */
275 #define	VROOT		0x01	/* root of its file system */
276 #define	VNOCACHE	0x02	/* don't keep cache pages on vnode */
277 #define	VNOMAP		0x04	/* file cannot be mapped/faulted */
278 #define	VDUP		0x08	/* file should be dup'ed rather then opened */
279 #define	VNOSWAP		0x10	/* file cannot be used as virtual swap device */
280 #define	VNOMOUNT	0x20	/* file cannot be covered by mount */
281 #define	VISSWAP		0x40	/* vnode is being used for swap */
282 #define	VSWAPLIKE	0x80	/* vnode acts like swap (but may not be) */
283 
284 #define	IS_SWAPVP(vp)	(((vp)->v_flag & (VISSWAP | VSWAPLIKE)) != 0)
285 
286 typedef struct vn_vfslocks_entry {
287 	rwstlock_t ve_lock;
288 	void *ve_vpvfs;
289 	struct vn_vfslocks_entry *ve_next;
290 	uint32_t ve_refcnt;
291 	char pad[64 - sizeof (rwstlock_t) - 2 * sizeof (void *) - \
292 	    sizeof (uint32_t)];
293 } vn_vfslocks_entry_t;
294 
295 /*
296  * The following two flags are used to lock the v_vfsmountedhere field
297  */
298 #define	VVFSLOCK	0x100
299 #define	VVFSWAIT	0x200
300 
301 /*
302  * Used to serialize VM operations on a vnode
303  */
304 #define	VVMLOCK		0x400
305 
306 /*
307  * Tell vn_open() not to fail a directory open for writing but
308  * to go ahead and call VOP_OPEN() to let the filesystem check.
309  */
310 #define	VDIROPEN	0x800
311 
312 /*
313  * Flag to let the VM system know that this file is most likely a binary
314  * or shared library since it has been mmap()ed EXEC at some time.
315  */
316 #define	VVMEXEC		0x1000
317 
318 #define	VPXFS		0x2000  /* clustering: global fs proxy vnode */
319 
320 #define	IS_PXFSVP(vp)	((vp)->v_flag & VPXFS)
321 
322 #define	V_XATTRDIR	0x4000	/* attribute unnamed directory */
323 
324 #define	V_LOCALITY	0x8000	/* whether locality aware */
325 
326 /*
327  * Flag that indicates the VM should maintain the v_pages list with all modified
328  * pages on one end and unmodified pages at the other. This makes finding dirty
329  * pages to write back to disk much faster at the expense of taking a minor
330  * fault on the first store instruction which touches a writable page.
331  */
332 #define	VMODSORT	(0x10000)
333 #define	IS_VMODSORT(vp) \
334 	(pvn_vmodsort_supported != 0 && ((vp)->v_flag  & VMODSORT) != 0)
335 
336 #define	VISSWAPFS	0x20000	/* vnode is being used for swapfs */
337 #define	IS_SWAPFSVP(vp)	(((vp)->v_flag & VISSWAPFS) != 0)
338 
339 #define	V_SYSATTR	0x40000	/* vnode is a GFS system attribute */
340 
341 /*
342  * Vnode attributes.  A bit-mask is supplied as part of the
343  * structure to indicate the attributes the caller wants to
344  * set (setattr) or extract (getattr).
345  */
346 
347 /*
348  * Note that va_nodeid and va_nblocks are 64bit data type.
349  * We support large files over NFSV3. With Solaris client and
350  * Server that generates 64bit ino's and sizes these fields
351  * will overflow if they are 32 bit sizes.
352  */
353 
354 typedef struct vattr {
355 	uint_t		va_mask;	/* bit-mask of attributes */
356 	vtype_t		va_type;	/* vnode type (for create) */
357 	mode_t		va_mode;	/* file access mode */
358 	uid_t		va_uid;		/* owner user id */
359 	gid_t		va_gid;		/* owner group id */
360 	dev_t		va_fsid;	/* file system id (dev for now) */
361 	u_longlong_t	va_nodeid;	/* node id */
362 	nlink_t		va_nlink;	/* number of references to file */
363 	u_offset_t	va_size;	/* file size in bytes */
364 	timestruc_t	va_atime;	/* time of last access */
365 	timestruc_t	va_mtime;	/* time of last modification */
366 	timestruc_t	va_ctime;	/* time of last status change */
367 	dev_t		va_rdev;	/* device the file represents */
368 	uint_t		va_blksize;	/* fundamental block size */
369 	u_longlong_t	va_nblocks;	/* # of blocks allocated */
370 	uint_t		va_seq;		/* sequence number */
371 } vattr_t;
372 
373 #define	AV_SCANSTAMP_SZ	32		/* length of anti-virus scanstamp */
374 
375 /*
376  * Structure of all optional attributes.
377  */
378 typedef struct xoptattr {
379 	timestruc_t	xoa_createtime;	/* Create time of file */
380 	uint8_t		xoa_archive;
381 	uint8_t		xoa_system;
382 	uint8_t		xoa_readonly;
383 	uint8_t		xoa_hidden;
384 	uint8_t		xoa_nounlink;
385 	uint8_t		xoa_immutable;
386 	uint8_t		xoa_appendonly;
387 	uint8_t		xoa_nodump;
388 	uint8_t		xoa_opaque;
389 	uint8_t		xoa_av_quarantined;
390 	uint8_t		xoa_av_modified;
391 	uint8_t		xoa_av_scanstamp[AV_SCANSTAMP_SZ];
392 } xoptattr_t;
393 
394 /*
395  * The xvattr structure is really a variable length structure that
396  * is made up of:
397  * - The classic vattr_t (xva_vattr)
398  * - a 32 bit quantity (xva_mapsize) that specifies the size of the
399  *   attribute bitmaps in 32 bit words.
400  * - A pointer to the returned attribute bitmap (needed because the
401  *   previous element, the requested attribute bitmap) is variable lenth.
402  * - The requested attribute bitmap, which is an array of 32 bit words.
403  *   Callers use the XVA_SET_REQ() macro to set the bits corresponding to
404  *   the attributes that are being requested.
405  * - The returned attribute bitmap, which is an array of 32 bit words.
406  *   File systems that support optional attributes use the XVA_SET_RTN()
407  *   macro to set the bits corresponding to the attributes that are being
408  *   returned.
409  * - The xoptattr_t structure which contains the attribute values
410  *
411  * xva_mapsize determines how many words in the attribute bitmaps.
412  * Immediately following the attribute bitmaps is the xoptattr_t.
413  * xva_getxoptattr() is used to get the pointer to the xoptattr_t
414  * section.
415  */
416 
417 #define	XVA_MAPSIZE	3		/* Size of attr bitmaps */
418 #define	XVA_MAGIC	0x78766174	/* Magic # for verification */
419 
420 /*
421  * The xvattr structure is an extensible structure which permits optional
422  * attributes to be requested/returned.  File systems may or may not support
423  * optional attributes.  They do so at their own discretion but if they do
424  * support optional attributes, they must register the VFSFT_XVATTR feature
425  * so that the optional attributes can be set/retrived.
426  *
427  * The fields of the xvattr structure are:
428  *
429  * xva_vattr - The first element of an xvattr is a legacy vattr structure
430  * which includes the common attributes.  If AT_XVATTR is set in the va_mask
431  * then the entire structure is treated as an xvattr.  If AT_XVATTR is not
432  * set, then only the xva_vattr structure can be used.
433  *
434  * xva_magic - 0x78766174 (hex for "xvat"). Magic number for verification.
435  *
436  * xva_mapsize - Size of requested and returned attribute bitmaps.
437  *
438  * xva_rtnattrmapp - Pointer to xva_rtnattrmap[].  We need this since the
439  * size of the array before it, xva_reqattrmap[], could change which means
440  * the location of xva_rtnattrmap[] could change.  This will allow unbundled
441  * file systems to find the location of xva_rtnattrmap[] when the sizes change.
442  *
443  * xva_reqattrmap[] - Array of requested attributes.  Attributes are
444  * represented by a specific bit in a specific element of the attribute
445  * map array.  Callers set the bits corresponding to the attributes
446  * that the caller wants to get/set.
447  *
448  * xva_rtnattrmap[] - Array of attributes that the file system was able to
449  * process.  Not all file systems support all optional attributes.  This map
450  * informs the caller which attributes the underlying file system was able
451  * to set/get.  (Same structure as the requested attributes array in terms
452  * of each attribute  corresponding to specific bits and array elements.)
453  *
454  * xva_xoptattrs - Structure containing values of optional attributes.
455  * These values are only valid if the corresponding bits in xva_reqattrmap
456  * are set and the underlying file system supports those attributes.
457  */
458 typedef struct xvattr {
459 	vattr_t		xva_vattr;	/* Embedded vattr structure */
460 	uint32_t	xva_magic;	/* Magic Number */
461 	uint32_t	xva_mapsize;	/* Size of attr bitmap (32-bit words) */
462 	uint32_t	*xva_rtnattrmapp;	/* Ptr to xva_rtnattrmap[] */
463 	uint32_t	xva_reqattrmap[XVA_MAPSIZE];	/* Requested attrs */
464 	uint32_t	xva_rtnattrmap[XVA_MAPSIZE];	/* Returned attrs */
465 	xoptattr_t	xva_xoptattrs;	/* Optional attributes */
466 } xvattr_t;
467 
468 #ifdef _SYSCALL32
469 /*
470  * For bigtypes time_t changed to 64 bit on the 64-bit kernel.
471  * Define an old version for user/kernel interface
472  */
473 
474 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
475 #pragma pack(4)
476 #endif
477 
478 typedef struct vattr32 {
479 	uint32_t	va_mask;	/* bit-mask of attributes */
480 	vtype_t		va_type;	/* vnode type (for create) */
481 	mode32_t	va_mode;	/* file access mode */
482 	uid32_t		va_uid;		/* owner user id */
483 	gid32_t		va_gid;		/* owner group id */
484 	dev32_t		va_fsid;	/* file system id (dev for now) */
485 	u_longlong_t	va_nodeid;	/* node id */
486 	nlink_t		va_nlink;	/* number of references to file */
487 	u_offset_t	va_size;	/* file size in bytes */
488 	timestruc32_t	va_atime;	/* time of last access */
489 	timestruc32_t	va_mtime;	/* time of last modification */
490 	timestruc32_t	va_ctime;	/* time of last status change */
491 	dev32_t		va_rdev;	/* device the file represents */
492 	uint32_t	va_blksize;	/* fundamental block size */
493 	u_longlong_t	va_nblocks;	/* # of blocks allocated */
494 	uint32_t	va_seq;		/* sequence number */
495 } vattr32_t;
496 
497 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
498 #pragma pack()
499 #endif
500 
501 #else  /* not _SYSCALL32 */
502 #define	vattr32		vattr
503 typedef vattr_t		vattr32_t;
504 #endif /* _SYSCALL32 */
505 
506 /*
507  * Attributes of interest to the caller of setattr or getattr.
508  */
509 #define	AT_TYPE		0x00001
510 #define	AT_MODE		0x00002
511 #define	AT_UID		0x00004
512 #define	AT_GID		0x00008
513 #define	AT_FSID		0x00010
514 #define	AT_NODEID	0x00020
515 #define	AT_NLINK	0x00040
516 #define	AT_SIZE		0x00080
517 #define	AT_ATIME	0x00100
518 #define	AT_MTIME	0x00200
519 #define	AT_CTIME	0x00400
520 #define	AT_RDEV		0x00800
521 #define	AT_BLKSIZE	0x01000
522 #define	AT_NBLOCKS	0x02000
523 /*			0x04000 */	/* unused */
524 #define	AT_SEQ		0x08000
525 /*
526  * If AT_XVATTR is set then there are additional bits to process in
527  * the xvattr_t's attribute bitmap.  If this is not set then the bitmap
528  * MUST be ignored.  Note that this bit must be set/cleared explicitly.
529  * That is, setting AT_ALL will NOT set AT_XVATTR.
530  */
531 #define	AT_XVATTR	0x10000
532 
533 #define	AT_ALL		(AT_TYPE|AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|\
534 			AT_NLINK|AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|\
535 			AT_RDEV|AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
536 
537 #define	AT_STAT		(AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|AT_NLINK|\
538 			AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|AT_RDEV|AT_TYPE)
539 
540 #define	AT_TIMES	(AT_ATIME|AT_MTIME|AT_CTIME)
541 
542 #define	AT_NOSET	(AT_NLINK|AT_RDEV|AT_FSID|AT_NODEID|AT_TYPE|\
543 			AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
544 
545 /*
546  * Attribute bits used in the extensible attribute's (xva's) attribute
547  * bitmaps.  Note that the bitmaps are made up of a variable length number
548  * of 32-bit words.  The convention is to use XAT{n}_{attrname} where "n"
549  * is the element in the bitmap (starting at 1).  This convention is for
550  * the convenience of the maintainer to keep track of which element each
551  * attribute belongs to.
552  *
553  * NOTE THAT CONSUMERS MUST *NOT* USE THE XATn_* DEFINES DIRECTLY.  CONSUMERS
554  * MUST USE THE XAT_* DEFINES.
555  */
556 #define	XAT0_INDEX	0LL		/* Index into bitmap for XAT0 attrs */
557 #define	XAT0_CREATETIME	0x00000001	/* Create time of file */
558 #define	XAT0_ARCHIVE	0x00000002	/* Archive */
559 #define	XAT0_SYSTEM	0x00000004	/* System */
560 #define	XAT0_READONLY	0x00000008	/* Readonly */
561 #define	XAT0_HIDDEN	0x00000010	/* Hidden */
562 #define	XAT0_NOUNLINK	0x00000020	/* Nounlink */
563 #define	XAT0_IMMUTABLE	0x00000040	/* immutable */
564 #define	XAT0_APPENDONLY	0x00000080	/* appendonly */
565 #define	XAT0_NODUMP	0x00000100	/* nodump */
566 #define	XAT0_OPAQUE	0x00000200	/* opaque */
567 #define	XAT0_AV_QUARANTINED	0x00000400	/* anti-virus quarantine */
568 #define	XAT0_AV_MODIFIED	0x00000800	/* anti-virus modified */
569 #define	XAT0_AV_SCANSTAMP	0x00001000	/* anti-virus scanstamp */
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)
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 
608 /*
609  * The returned attribute map array (xva_rtnattrmap[]) is located past the
610  * requested attribute map array (xva_reqattrmap[]).  Its location changes
611  * when the array sizes change.  We use a separate pointer in a known location
612  * (xva_rtnattrmapp) to hold the location of xva_rtnattrmap[].  This is
613  * set in xva_init()
614  */
615 #define	XVA_RTNATTRMAP(xvap)	((xvap)->xva_rtnattrmapp)
616 
617 /*
618  * XVA_SET_REQ() sets an attribute bit in the proper element in the bitmap
619  * of requested attributes (xva_reqattrmap[]).
620  */
621 #define	XVA_SET_REQ(xvap, attr)					\
622 	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
623 	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
624 	(xvap)->xva_reqattrmap[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
625 
626 /*
627  * XVA_SET_RTN() sets an attribute bit in the proper element in the bitmap
628  * of returned attributes (xva_rtnattrmap[]).
629  */
630 #define	XVA_SET_RTN(xvap, attr)					\
631 	ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);		\
632 	ASSERT((xvap)->xva_magic == XVA_MAGIC);			\
633 	(XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
634 
635 /*
636  * XVA_ISSET_REQ() checks the requested attribute bitmap (xva_reqattrmap[])
637  * to see of the corresponding attribute bit is set.  If so, returns non-zero.
638  */
639 #define	XVA_ISSET_REQ(xvap, attr)					\
640 	((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&			\
641 		((xvap)->xva_magic == XVA_MAGIC) &&			\
642 		((xvap)->xva_mapsize > XVA_INDEX(attr))) ?		\
643 	((xvap)->xva_reqattrmap[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) :	0)
644 
645 /*
646  * XVA_ISSET_RTN() checks the returned attribute bitmap (xva_rtnattrmap[])
647  * to see of the corresponding attribute bit is set.  If so, returns non-zero.
648  */
649 #define	XVA_ISSET_RTN(xvap, attr)					\
650 	((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&			\
651 		((xvap)->xva_magic == XVA_MAGIC) &&			\
652 		((xvap)->xva_mapsize > XVA_INDEX(attr))) ?		\
653 	((XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) : 0)
654 
655 /*
656  *  Modes.  Some values same as S_xxx entries from stat.h for convenience.
657  */
658 #define	VSUID		04000		/* set user id on execution */
659 #define	VSGID		02000		/* set group id on execution */
660 #define	VSVTX		01000		/* save swapped text even after use */
661 
662 /*
663  * Permissions.
664  */
665 #define	VREAD		00400
666 #define	VWRITE		00200
667 #define	VEXEC		00100
668 
669 #define	MODEMASK	07777		/* mode bits plus permission bits */
670 #define	PERMMASK	00777		/* permission bits */
671 
672 /*
673  * VOP_ACCESS flags
674  */
675 #define	V_ACE_MASK	0x1	/* mask represents  NFSv4 ACE permissions */
676 #define	V_APPEND	0x2	/* want to do append only check */
677 
678 /*
679  * Check whether mandatory file locking is enabled.
680  */
681 
682 #define	MANDMODE(mode)		(((mode) & (VSGID|(VEXEC>>3))) == VSGID)
683 #define	MANDLOCK(vp, mode)	((vp)->v_type == VREG && MANDMODE(mode))
684 
685 /*
686  * Flags for vnode operations.
687  */
688 enum rm		{ RMFILE, RMDIRECTORY };	/* rm or rmdir (remove) */
689 enum symfollow	{ NO_FOLLOW, FOLLOW };		/* follow symlinks (or not) */
690 enum vcexcl	{ NONEXCL, EXCL };		/* (non)excl create */
691 enum create	{ CRCREAT, CRMKNOD, CRMKDIR };	/* reason for create */
692 
693 typedef enum rm		rm_t;
694 typedef enum symfollow	symfollow_t;
695 typedef enum vcexcl	vcexcl_t;
696 typedef enum create	create_t;
697 
698 /* Vnode Events - Used by VOP_VNEVENT */
699 typedef enum vnevent	{
700 	VE_SUPPORT	= 0,	/* Query */
701 	VE_RENAME_SRC	= 1,	/* Rename, with vnode as source */
702 	VE_RENAME_DEST	= 2,	/* Rename, with vnode as target/destination */
703 	VE_REMOVE	= 3,	/* Remove of vnode's name */
704 	VE_RMDIR	= 4,	/* Remove of directory vnode's name */
705 	VE_CREATE	= 5,	/* Create with vnode's name which exists */
706 	VE_LINK		= 6, 	/* Link with vnode's name as source */
707 	VE_RENAME_DEST_DIR	= 7, 	/* Rename with vnode as target dir */
708 	VE_MOUNTEDOVER	= 8 	/* File or Filesystem got mounted over vnode */
709 } vnevent_t;
710 
711 /*
712  * Values for checking vnode open and map counts
713  */
714 enum v_mode { V_READ, V_WRITE, V_RDORWR, V_RDANDWR };
715 
716 typedef enum v_mode v_mode_t;
717 
718 #define	V_TRUE	1
719 #define	V_FALSE	0
720 
721 /*
722  * Structure used on VOP_GETSECATTR and VOP_SETSECATTR operations
723  */
724 
725 typedef struct vsecattr {
726 	uint_t		vsa_mask;	/* See below */
727 	int		vsa_aclcnt;	/* ACL entry count */
728 	void		*vsa_aclentp;	/* pointer to ACL entries */
729 	int		vsa_dfaclcnt;	/* default ACL entry count */
730 	void		*vsa_dfaclentp;	/* pointer to default ACL entries */
731 	size_t		vsa_aclentsz;	/* ACE size in bytes of vsa_aclentp */
732 	uint_t		vsa_aclflags;	/* ACE ACL flags */
733 } vsecattr_t;
734 
735 /* vsa_mask values */
736 #define	VSA_ACL			0x0001
737 #define	VSA_ACLCNT		0x0002
738 #define	VSA_DFACL		0x0004
739 #define	VSA_DFACLCNT		0x0008
740 #define	VSA_ACE			0x0010
741 #define	VSA_ACECNT		0x0020
742 #define	VSA_ACE_ALLTYPES	0x0040
743 #define	VSA_ACE_ACLFLAGS	0x0080	/* get/set ACE ACL flags */
744 
745 /*
746  * Structure used by various vnode operations to determine
747  * the context (pid, host, identity) of a caller.
748  *
749  * The cc_caller_id is used to identify one or more callers who invoke
750  * operations, possibly on behalf of others.  For example, the NFS
751  * server could have it's own cc_caller_id which can be detected by
752  * vnode/vfs operations or (FEM) monitors on those operations.  New
753  * caller IDs are generated by fs_new_caller_id().
754  */
755 typedef struct caller_context {
756 	pid_t		cc_pid;		/* Process ID of the caller */
757 	int		cc_sysid;	/* System ID, used for remote calls */
758 	u_longlong_t	cc_caller_id;	/* Identifier for (set of) caller(s) */
759 	ulong_t		cc_flags;
760 } caller_context_t;
761 
762 /*
763  * Flags for caller context.  The caller sets CC_DONTBLOCK if it does not
764  * want to block inside of a FEM monitor.  The monitor will set CC_WOULDBLOCK
765  * and return EAGAIN if the operation would have blocked.
766  */
767 #define	CC_WOULDBLOCK	0x01
768 #define	CC_DONTBLOCK	0x02
769 
770 /*
771  * Structure tags for function prototypes, defined elsewhere.
772  */
773 struct pathname;
774 struct fid;
775 struct flock64;
776 struct flk_callback;
777 struct shrlock;
778 struct page;
779 struct seg;
780 struct as;
781 struct pollhead;
782 
783 #ifdef	_KERNEL
784 
785 /*
786  * VNODE_OPS defines all the vnode operations.  It is used to define
787  * the vnodeops structure (below) and the fs_func_p union (vfs_opreg.h).
788  */
789 #define	VNODE_OPS							\
790 	int	(*vop_open)(vnode_t **, int, cred_t *,			\
791 				caller_context_t *);			\
792 	int	(*vop_close)(vnode_t *, int, int, offset_t, cred_t *,	\
793 				caller_context_t *);			\
794 	int	(*vop_read)(vnode_t *, uio_t *, int, cred_t *,		\
795 				caller_context_t *);			\
796 	int	(*vop_write)(vnode_t *, uio_t *, int, cred_t *,		\
797 				caller_context_t *);			\
798 	int	(*vop_ioctl)(vnode_t *, int, intptr_t, int, cred_t *,	\
799 				int *, caller_context_t *);		\
800 	int	(*vop_setfl)(vnode_t *, int, int, cred_t *,		\
801 				caller_context_t *);			\
802 	int	(*vop_getattr)(vnode_t *, vattr_t *, int, cred_t *,	\
803 				caller_context_t *);			\
804 	int	(*vop_setattr)(vnode_t *, vattr_t *, int, cred_t *,	\
805 				caller_context_t *);			\
806 	int	(*vop_access)(vnode_t *, int, int, cred_t *,		\
807 				caller_context_t *);			\
808 	int	(*vop_lookup)(vnode_t *, char *, vnode_t **,		\
809 				struct pathname *,			\
810 				int, vnode_t *, cred_t *,		\
811 				caller_context_t *, int *,		\
812 				struct pathname *);			\
813 	int	(*vop_create)(vnode_t *, char *, vattr_t *, vcexcl_t,	\
814 				int, vnode_t **, cred_t *, int,		\
815 				caller_context_t *, vsecattr_t *);	\
816 	int	(*vop_remove)(vnode_t *, char *, cred_t *,		\
817 				caller_context_t *, int);		\
818 	int	(*vop_link)(vnode_t *, vnode_t *, char *, cred_t *,	\
819 				caller_context_t *, int);		\
820 	int	(*vop_rename)(vnode_t *, char *, vnode_t *, char *,	\
821 				cred_t *, caller_context_t *, int);	\
822 	int	(*vop_mkdir)(vnode_t *, char *, vattr_t *, vnode_t **,	\
823 				cred_t *, caller_context_t *, int,	\
824 				vsecattr_t *);				\
825 	int	(*vop_rmdir)(vnode_t *, char *, vnode_t *, cred_t *,	\
826 				caller_context_t *, int);		\
827 	int	(*vop_readdir)(vnode_t *, uio_t *, cred_t *, int *,	\
828 				caller_context_t *, int);		\
829 	int	(*vop_symlink)(vnode_t *, char *, vattr_t *, char *,	\
830 				cred_t *, caller_context_t *, int);	\
831 	int	(*vop_readlink)(vnode_t *, uio_t *, cred_t *,		\
832 				caller_context_t *);			\
833 	int	(*vop_fsync)(vnode_t *, int, cred_t *,			\
834 				caller_context_t *);			\
835 	void	(*vop_inactive)(vnode_t *, cred_t *,			\
836 				caller_context_t *);			\
837 	int	(*vop_fid)(vnode_t *, struct fid *,			\
838 				caller_context_t *);			\
839 	int	(*vop_rwlock)(vnode_t *, int, caller_context_t *);	\
840 	void	(*vop_rwunlock)(vnode_t *, int, caller_context_t *);	\
841 	int	(*vop_seek)(vnode_t *, offset_t, offset_t *,		\
842 				caller_context_t *);			\
843 	int	(*vop_cmp)(vnode_t *, vnode_t *, caller_context_t *);	\
844 	int	(*vop_frlock)(vnode_t *, int, struct flock64 *,		\
845 				int, offset_t,				\
846 				struct flk_callback *, cred_t *,	\
847 				caller_context_t *);			\
848 	int	(*vop_space)(vnode_t *, int, struct flock64 *,		\
849 				int, offset_t,				\
850 				cred_t *, caller_context_t *);		\
851 	int	(*vop_realvp)(vnode_t *, vnode_t **,			\
852 				caller_context_t *);			\
853 	int	(*vop_getpage)(vnode_t *, offset_t, size_t, uint_t *,	\
854 				struct page **, size_t, struct seg *,	\
855 				caddr_t, enum seg_rw, cred_t *,		\
856 				caller_context_t *);			\
857 	int	(*vop_putpage)(vnode_t *, offset_t, size_t,		\
858 				int, cred_t *, caller_context_t *);	\
859 	int	(*vop_map)(vnode_t *, offset_t, struct as *,		\
860 				caddr_t *, size_t,			\
861 				uchar_t, uchar_t, uint_t, cred_t *,	\
862 				caller_context_t *);			\
863 	int	(*vop_addmap)(vnode_t *, offset_t, struct as *,		\
864 				caddr_t, size_t,			\
865 				uchar_t, uchar_t, uint_t, cred_t *,	\
866 				caller_context_t *);			\
867 	int	(*vop_delmap)(vnode_t *, offset_t, struct as *,		\
868 				caddr_t, size_t,			\
869 				uint_t, uint_t, uint_t, cred_t *,	\
870 				caller_context_t *);			\
871 	int	(*vop_poll)(vnode_t *, short, int, short *,		\
872 				struct pollhead **,			\
873 				caller_context_t *);			\
874 	int	(*vop_dump)(vnode_t *, caddr_t, int, int,		\
875 				caller_context_t *);			\
876 	int	(*vop_pathconf)(vnode_t *, int, ulong_t *, cred_t *,	\
877 				caller_context_t *);			\
878 	int	(*vop_pageio)(vnode_t *, struct page *,			\
879 				u_offset_t, size_t, int, cred_t *,	\
880 				caller_context_t *);			\
881 	int	(*vop_dumpctl)(vnode_t *, int, int *,			\
882 				caller_context_t *);			\
883 	void	(*vop_dispose)(vnode_t *, struct page *,		\
884 				int, int, cred_t *,			\
885 				caller_context_t *);			\
886 	int	(*vop_setsecattr)(vnode_t *, vsecattr_t *,		\
887 				int, cred_t *, caller_context_t *);	\
888 	int	(*vop_getsecattr)(vnode_t *, vsecattr_t *,		\
889 				int, cred_t *, caller_context_t *);	\
890 	int	(*vop_shrlock)(vnode_t *, int, struct shrlock *,	\
891 				int, cred_t *, caller_context_t *);	\
892 	int	(*vop_vnevent)(vnode_t *, vnevent_t, vnode_t *,		\
893 				char *, caller_context_t *)
894 	/* NB: No ";" */
895 
896 /*
897  * Operations on vnodes.  Note: File systems must never operate directly
898  * on a 'vnodeops' structure -- it WILL change in future releases!  They
899  * must use vn_make_ops() to create the structure.
900  */
901 typedef struct vnodeops {
902 	const char *vnop_name;
903 	VNODE_OPS;	/* Signatures of all vnode operations (vops) */
904 } vnodeops_t;
905 
906 typedef int (*fs_generic_func_p) ();	/* Generic vop/vfsop/femop/fsemop ptr */
907 
908 extern int	fop_open(vnode_t **, int, cred_t *, caller_context_t *);
909 extern int	fop_close(vnode_t *, int, int, offset_t, cred_t *,
910 				caller_context_t *);
911 extern int	fop_read(vnode_t *, uio_t *, int, cred_t *, caller_context_t *);
912 extern int	fop_write(vnode_t *, uio_t *, int, cred_t *,
913 				caller_context_t *);
914 extern int	fop_ioctl(vnode_t *, int, intptr_t, int, cred_t *, int *,
915 				caller_context_t *);
916 extern int	fop_setfl(vnode_t *, int, int, cred_t *, caller_context_t *);
917 extern int	fop_getattr(vnode_t *, vattr_t *, int, cred_t *,
918 				caller_context_t *);
919 extern int	fop_setattr(vnode_t *, vattr_t *, int, cred_t *,
920 				caller_context_t *);
921 extern int	fop_access(vnode_t *, int, int, cred_t *, caller_context_t *);
922 extern int	fop_lookup(vnode_t *, char *, vnode_t **, struct pathname *,
923 				int, vnode_t *, cred_t *, caller_context_t *,
924 				int *, struct pathname *);
925 extern int	fop_create(vnode_t *, char *, vattr_t *, vcexcl_t, int,
926 				vnode_t **, cred_t *, int, caller_context_t *,
927 				vsecattr_t *);
928 extern int	fop_remove(vnode_t *vp, char *, cred_t *, caller_context_t *,
929 				int);
930 extern int	fop_link(vnode_t *, vnode_t *, char *, cred_t *,
931 				caller_context_t *, int);
932 extern int	fop_rename(vnode_t *, char *, vnode_t *, char *, cred_t *,
933 				caller_context_t *, int);
934 extern int	fop_mkdir(vnode_t *, char *, vattr_t *, vnode_t **, cred_t *,
935 				caller_context_t *, int, vsecattr_t *);
936 extern int	fop_rmdir(vnode_t *, char *, vnode_t *, cred_t *,
937 				caller_context_t *, int);
938 extern int	fop_readdir(vnode_t *, uio_t *, cred_t *, int *,
939 				caller_context_t *, int);
940 extern int	fop_symlink(vnode_t *, char *, vattr_t *, char *, cred_t *,
941 				caller_context_t *, int);
942 extern int	fop_readlink(vnode_t *, uio_t *, cred_t *, caller_context_t *);
943 extern int	fop_fsync(vnode_t *, int, cred_t *, caller_context_t *);
944 extern void	fop_inactive(vnode_t *, cred_t *, caller_context_t *);
945 extern int	fop_fid(vnode_t *, struct fid *, caller_context_t *);
946 extern int	fop_rwlock(vnode_t *, int, caller_context_t *);
947 extern void	fop_rwunlock(vnode_t *, int, caller_context_t *);
948 extern int	fop_seek(vnode_t *, offset_t, offset_t *, caller_context_t *);
949 extern int	fop_cmp(vnode_t *, vnode_t *, caller_context_t *);
950 extern int	fop_frlock(vnode_t *, int, struct flock64 *, int, offset_t,
951 				struct flk_callback *, cred_t *,
952 				caller_context_t *);
953 extern int	fop_space(vnode_t *, int, struct flock64 *, int, offset_t,
954 				cred_t *, caller_context_t *);
955 extern int	fop_realvp(vnode_t *, vnode_t **, caller_context_t *);
956 extern int	fop_getpage(vnode_t *, offset_t, size_t, uint_t *,
957 				struct page **, size_t, struct seg *,
958 				caddr_t, enum seg_rw, cred_t *,
959 				caller_context_t *);
960 extern int	fop_putpage(vnode_t *, offset_t, size_t, int, cred_t *,
961 				caller_context_t *);
962 extern int	fop_map(vnode_t *, offset_t, struct as *, caddr_t *, size_t,
963 				uchar_t, uchar_t, uint_t, cred_t *cr,
964 				caller_context_t *);
965 extern int	fop_addmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
966 				uchar_t, uchar_t, uint_t, cred_t *,
967 				caller_context_t *);
968 extern int	fop_delmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
969 				uint_t, uint_t, uint_t, cred_t *,
970 				caller_context_t *);
971 extern int	fop_poll(vnode_t *, short, int, short *, struct pollhead **,
972 				caller_context_t *);
973 extern int	fop_dump(vnode_t *, caddr_t, int, int, caller_context_t *);
974 extern int	fop_pathconf(vnode_t *, int, ulong_t *, cred_t *,
975 				caller_context_t *);
976 extern int	fop_pageio(vnode_t *, struct page *, u_offset_t, size_t, int,
977 				cred_t *, caller_context_t *);
978 extern int	fop_dumpctl(vnode_t *, int, int *, caller_context_t *);
979 extern void	fop_dispose(vnode_t *, struct page *, int, int, cred_t *,
980 				caller_context_t *);
981 extern int	fop_setsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
982 				caller_context_t *);
983 extern int	fop_getsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
984 				caller_context_t *);
985 extern int	fop_shrlock(vnode_t *, int, struct shrlock *, int, cred_t *,
986 				caller_context_t *);
987 extern int	fop_vnevent(vnode_t *, vnevent_t, vnode_t *, char *,
988 				caller_context_t *);
989 
990 #endif	/* _KERNEL */
991 
992 #define	VOP_OPEN(vpp, mode, cr, ct) \
993 	fop_open(vpp, mode, cr, ct)
994 #define	VOP_CLOSE(vp, f, c, o, cr, ct) \
995 	fop_close(vp, f, c, o, cr, ct)
996 #define	VOP_READ(vp, uiop, iof, cr, ct) \
997 	fop_read(vp, uiop, iof, cr, ct)
998 #define	VOP_WRITE(vp, uiop, iof, cr, ct) \
999 	fop_write(vp, uiop, iof, cr, ct)
1000 #define	VOP_IOCTL(vp, cmd, a, f, cr, rvp, ct) \
1001 	fop_ioctl(vp, cmd, a, f, cr, rvp, ct)
1002 #define	VOP_SETFL(vp, f, a, cr, ct) \
1003 	fop_setfl(vp, f, a, cr, ct)
1004 #define	VOP_GETATTR(vp, vap, f, cr, ct) \
1005 	fop_getattr(vp, vap, f, cr, ct)
1006 #define	VOP_SETATTR(vp, vap, f, cr, ct) \
1007 	fop_setattr(vp, vap, f, cr, ct)
1008 #define	VOP_ACCESS(vp, mode, f, cr, ct) \
1009 	fop_access(vp, mode, f, cr, ct)
1010 #define	VOP_LOOKUP(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp) \
1011 	fop_lookup(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp)
1012 #define	VOP_CREATE(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap) \
1013 	fop_create(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap)
1014 #define	VOP_REMOVE(dvp, p, cr, ct, f) \
1015 	fop_remove(dvp, p, cr, ct, f)
1016 #define	VOP_LINK(tdvp, fvp, p, cr, ct, f) \
1017 	fop_link(tdvp, fvp, p, cr, ct, f)
1018 #define	VOP_RENAME(fvp, fnm, tdvp, tnm, cr, ct, f) \
1019 	fop_rename(fvp, fnm, tdvp, tnm, cr, ct, f)
1020 #define	VOP_MKDIR(dp, p, vap, vpp, cr, ct, f, vsap) \
1021 	fop_mkdir(dp, p, vap, vpp, cr, ct, f, vsap)
1022 #define	VOP_RMDIR(dp, p, cdir, cr, ct, f) \
1023 	fop_rmdir(dp, p, cdir, cr, ct, f)
1024 #define	VOP_READDIR(vp, uiop, cr, eofp, ct, f) \
1025 	fop_readdir(vp, uiop, cr, eofp, ct, f)
1026 #define	VOP_SYMLINK(dvp, lnm, vap, tnm, cr, ct, f) \
1027 	fop_symlink(dvp, lnm, vap, tnm, cr, ct, f)
1028 #define	VOP_READLINK(vp, uiop, cr, ct) \
1029 	fop_readlink(vp, uiop, cr, ct)
1030 #define	VOP_FSYNC(vp, syncflag, cr, ct) \
1031 	fop_fsync(vp, syncflag, cr, ct)
1032 #define	VOP_INACTIVE(vp, cr, ct) \
1033 	fop_inactive(vp, cr, ct)
1034 #define	VOP_FID(vp, fidp, ct) \
1035 	fop_fid(vp, fidp, ct)
1036 #define	VOP_RWLOCK(vp, w, ct) \
1037 	fop_rwlock(vp, w, ct)
1038 #define	VOP_RWUNLOCK(vp, w, ct) \
1039 	fop_rwunlock(vp, w, ct)
1040 #define	VOP_SEEK(vp, ooff, noffp, ct) \
1041 	fop_seek(vp, ooff, noffp, ct)
1042 #define	VOP_CMP(vp1, vp2, ct) \
1043 	fop_cmp(vp1, vp2, ct)
1044 #define	VOP_FRLOCK(vp, cmd, a, f, o, cb, cr, ct) \
1045 	fop_frlock(vp, cmd, a, f, o, cb, cr, ct)
1046 #define	VOP_SPACE(vp, cmd, a, f, o, cr, ct) \
1047 	fop_space(vp, cmd, a, f, o, cr, ct)
1048 #define	VOP_REALVP(vp1, vp2, ct) \
1049 	fop_realvp(vp1, vp2, ct)
1050 #define	VOP_GETPAGE(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct) \
1051 	fop_getpage(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct)
1052 #define	VOP_PUTPAGE(vp, of, sz, fl, cr, ct) \
1053 	fop_putpage(vp, of, sz, fl, cr, ct)
1054 #define	VOP_MAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1055 	fop_map(vp, of, as, a, sz, p, mp, fl, cr, ct)
1056 #define	VOP_ADDMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1057 	fop_addmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1058 #define	VOP_DELMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1059 	fop_delmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1060 #define	VOP_POLL(vp, events, anyyet, reventsp, phpp, ct) \
1061 	fop_poll(vp, events, anyyet, reventsp, phpp, ct)
1062 #define	VOP_DUMP(vp, addr, bn, count, ct) \
1063 	fop_dump(vp, addr, bn, count, ct)
1064 #define	VOP_PATHCONF(vp, cmd, valp, cr, ct) \
1065 	fop_pathconf(vp, cmd, valp, cr, ct)
1066 #define	VOP_PAGEIO(vp, pp, io_off, io_len, flags, cr, ct) \
1067 	fop_pageio(vp, pp, io_off, io_len, flags, cr, ct)
1068 #define	VOP_DUMPCTL(vp, action, blkp, ct) \
1069 	fop_dumpctl(vp, action, blkp, ct)
1070 #define	VOP_DISPOSE(vp, pp, flag, dn, cr, ct) \
1071 	fop_dispose(vp, pp, flag, dn, cr, ct)
1072 #define	VOP_GETSECATTR(vp, vsap, f, cr, ct) \
1073 	fop_getsecattr(vp, vsap, f, cr, ct)
1074 #define	VOP_SETSECATTR(vp, vsap, f, cr, ct) \
1075 	fop_setsecattr(vp, vsap, f, cr, ct)
1076 #define	VOP_SHRLOCK(vp, cmd, shr, f, cr, ct) \
1077 	fop_shrlock(vp, cmd, shr, f, cr, ct)
1078 #define	VOP_VNEVENT(vp, vnevent, dvp, fnm, ct) \
1079 	fop_vnevent(vp, vnevent, dvp, fnm, ct)
1080 
1081 #define	VOPNAME_OPEN		"open"
1082 #define	VOPNAME_CLOSE		"close"
1083 #define	VOPNAME_READ		"read"
1084 #define	VOPNAME_WRITE		"write"
1085 #define	VOPNAME_IOCTL		"ioctl"
1086 #define	VOPNAME_SETFL		"setfl"
1087 #define	VOPNAME_GETATTR		"getattr"
1088 #define	VOPNAME_SETATTR		"setattr"
1089 #define	VOPNAME_ACCESS		"access"
1090 #define	VOPNAME_LOOKUP		"lookup"
1091 #define	VOPNAME_CREATE		"create"
1092 #define	VOPNAME_REMOVE		"remove"
1093 #define	VOPNAME_LINK		"link"
1094 #define	VOPNAME_RENAME		"rename"
1095 #define	VOPNAME_MKDIR		"mkdir"
1096 #define	VOPNAME_RMDIR		"rmdir"
1097 #define	VOPNAME_READDIR		"readdir"
1098 #define	VOPNAME_SYMLINK		"symlink"
1099 #define	VOPNAME_READLINK	"readlink"
1100 #define	VOPNAME_FSYNC		"fsync"
1101 #define	VOPNAME_INACTIVE	"inactive"
1102 #define	VOPNAME_FID		"fid"
1103 #define	VOPNAME_RWLOCK		"rwlock"
1104 #define	VOPNAME_RWUNLOCK	"rwunlock"
1105 #define	VOPNAME_SEEK		"seek"
1106 #define	VOPNAME_CMP		"cmp"
1107 #define	VOPNAME_FRLOCK		"frlock"
1108 #define	VOPNAME_SPACE		"space"
1109 #define	VOPNAME_REALVP		"realvp"
1110 #define	VOPNAME_GETPAGE		"getpage"
1111 #define	VOPNAME_PUTPAGE		"putpage"
1112 #define	VOPNAME_MAP		"map"
1113 #define	VOPNAME_ADDMAP		"addmap"
1114 #define	VOPNAME_DELMAP		"delmap"
1115 #define	VOPNAME_POLL		"poll"
1116 #define	VOPNAME_DUMP		"dump"
1117 #define	VOPNAME_PATHCONF	"pathconf"
1118 #define	VOPNAME_PAGEIO		"pageio"
1119 #define	VOPNAME_DUMPCTL		"dumpctl"
1120 #define	VOPNAME_DISPOSE		"dispose"
1121 #define	VOPNAME_GETSECATTR	"getsecattr"
1122 #define	VOPNAME_SETSECATTR	"setsecattr"
1123 #define	VOPNAME_SHRLOCK		"shrlock"
1124 #define	VOPNAME_VNEVENT		"vnevent"
1125 
1126 /*
1127  * Flags for VOP_LOOKUP
1128  *
1129  * Defined in file.h, but also possible, FIGNORECASE
1130  *
1131  */
1132 #define	LOOKUP_DIR		0x01	/* want parent dir vp */
1133 #define	LOOKUP_XATTR		0x02	/* lookup up extended attr dir */
1134 #define	CREATE_XATTR_DIR	0x04	/* Create extended attr dir */
1135 #define	LOOKUP_HAVE_SYSATTR_DIR	0x08	/* Already created virtual GFS dir */
1136 
1137 /*
1138  * Flags for VOP_READDIR
1139  */
1140 #define	V_RDDIR_ENTFLAGS	0x01	/* request dirent flags */
1141 
1142 /*
1143  * Flags for VOP_RWLOCK/VOP_RWUNLOCK
1144  * VOP_RWLOCK will return the flag that was actually set, or -1 if none.
1145  */
1146 #define	V_WRITELOCK_TRUE	(1)	/* Request write-lock on the vnode */
1147 #define	V_WRITELOCK_FALSE	(0)	/* Request read-lock on the vnode */
1148 
1149 /*
1150  * Flags for VOP_DUMPCTL
1151  */
1152 #define	DUMP_ALLOC	0
1153 #define	DUMP_FREE	1
1154 #define	DUMP_SCAN	2
1155 
1156 /*
1157  * Public vnode manipulation functions.
1158  */
1159 #ifdef	_KERNEL
1160 
1161 vnode_t *vn_alloc(int);
1162 void	vn_reinit(vnode_t *);
1163 void	vn_recycle(vnode_t *);
1164 void	vn_free(vnode_t *);
1165 
1166 int	vn_is_readonly(vnode_t *);
1167 int   	vn_is_opened(vnode_t *, v_mode_t);
1168 int   	vn_is_mapped(vnode_t *, v_mode_t);
1169 int   	vn_has_other_opens(vnode_t *, v_mode_t);
1170 void	vn_open_upgrade(vnode_t *, int);
1171 void	vn_open_downgrade(vnode_t *, int);
1172 
1173 int	vn_can_change_zones(vnode_t *vp);
1174 
1175 int	vn_has_flocks(vnode_t *);
1176 int	vn_has_mandatory_locks(vnode_t *, int);
1177 int	vn_has_cached_data(vnode_t *);
1178 
1179 void	vn_setops(vnode_t *, vnodeops_t *);
1180 vnodeops_t *vn_getops(vnode_t *);
1181 int	vn_matchops(vnode_t *, vnodeops_t *);
1182 int	vn_matchopval(vnode_t *, char *, fs_generic_func_p);
1183 int	vn_ismntpt(vnode_t *);
1184 
1185 struct vfs *vn_mountedvfs(vnode_t *);
1186 
1187 void	vn_create_cache(void);
1188 void	vn_destroy_cache(void);
1189 
1190 void	vn_freevnodeops(vnodeops_t *);
1191 
1192 int	vn_open(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1193 		struct vnode **vpp, enum create crwhy, mode_t umask);
1194 int	vn_openat(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1195 		struct vnode **vpp, enum create crwhy,
1196 		mode_t umask, struct vnode *startvp, int fd);
1197 int	vn_create(char *pnamep, enum uio_seg seg, struct vattr *vap,
1198 		enum vcexcl excl, int mode, struct vnode **vpp,
1199 		enum create why, int flag, mode_t umask);
1200 int	vn_createat(char *pnamep, enum uio_seg seg, struct vattr *vap,
1201 		enum vcexcl excl, int mode, struct vnode **vpp,
1202 		enum create why, int flag, mode_t umask, struct vnode *startvp);
1203 int	vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base, ssize_t len,
1204 		offset_t offset, enum uio_seg seg, int ioflag, rlim64_t ulimit,
1205 		cred_t *cr, ssize_t *residp);
1206 void	vn_rele(struct vnode *vp);
1207 void	vn_rele_stream(struct vnode *vp);
1208 int	vn_link(char *from, char *to, enum uio_seg seg);
1209 int	vn_rename(char *from, char *to, enum uio_seg seg);
1210 int	vn_renameat(vnode_t *fdvp, char *fname, vnode_t *tdvp, char *tname,
1211 		enum uio_seg seg);
1212 int	vn_remove(char *fnamep, enum uio_seg seg, enum rm dirflag);
1213 int	vn_removeat(vnode_t *startvp, char *fnamep, enum uio_seg seg,
1214 		enum rm dirflag);
1215 int	vn_compare(vnode_t *vp1, vnode_t *vp2);
1216 int	vn_vfswlock(struct vnode *vp);
1217 int	vn_vfswlock_wait(struct vnode *vp);
1218 int	vn_vfsrlock(struct vnode *vp);
1219 int	vn_vfsrlock_wait(struct vnode *vp);
1220 void	vn_vfsunlock(struct vnode *vp);
1221 int	vn_vfswlock_held(struct vnode *vp);
1222 vnode_t *specvp(struct vnode *vp, dev_t dev, vtype_t type, struct cred *cr);
1223 vnode_t *makespecvp(dev_t dev, vtype_t type);
1224 vn_vfslocks_entry_t *vn_vfslocks_getlock(void *);
1225 void	vn_vfslocks_rele(vn_vfslocks_entry_t *);
1226 
1227 void vn_copypath(struct vnode *src, struct vnode *dst);
1228 void vn_setpath_str(struct vnode *vp, const char *str, size_t len);
1229 void vn_setpath(vnode_t *rootvp, struct vnode *startvp, struct vnode *vp,
1230     const char *path, size_t plen);
1231 
1232 /* Vnode event notification */
1233 void	vnevent_rename_src(vnode_t *, vnode_t *, char *, caller_context_t *);
1234 void	vnevent_rename_dest(vnode_t *, vnode_t *, char *, caller_context_t *);
1235 void	vnevent_remove(vnode_t *, vnode_t *, char *, caller_context_t *);
1236 void	vnevent_rmdir(vnode_t *, vnode_t *, char *, caller_context_t *);
1237 void	vnevent_create(vnode_t *, caller_context_t *);
1238 void	vnevent_link(vnode_t *, caller_context_t *);
1239 void	vnevent_rename_dest_dir(vnode_t *, caller_context_t *ct);
1240 void	vnevent_mountedover(vnode_t *, caller_context_t *);
1241 int	vnevent_support(vnode_t *, caller_context_t *);
1242 
1243 /* Vnode specific data */
1244 void vsd_create(uint_t *, void (*)(void *));
1245 void vsd_destroy(uint_t *);
1246 void *vsd_get(vnode_t *, uint_t);
1247 int vsd_set(vnode_t *, uint_t, void *);
1248 void vsd_free(vnode_t *);
1249 
1250 /*
1251  * Extensible vnode attribute (xva) routines:
1252  * xva_init() initializes an xvattr_t (zero struct, init mapsize, set AT_XATTR)
1253  * xva_getxoptattr() returns a ponter to the xoptattr_t section of xvattr_t
1254  */
1255 void		xva_init(xvattr_t *);
1256 xoptattr_t	*xva_getxoptattr(xvattr_t *);	/* Get ptr to xoptattr_t */
1257 
1258 void xattr_init(void);		/* Initialize vnodeops for xattrs */
1259 
1260 /* GFS tunnel for xattrs */
1261 int xattr_dir_lookup(vnode_t *, vnode_t **, int, cred_t *);
1262 
1263 /* Context identification */
1264 u_longlong_t	fs_new_caller_id();
1265 
1266 int	vn_vmpss_usepageio(vnode_t *);
1267 
1268 /*
1269  * Needed for use of IS_VMODSORT() in kernel.
1270  */
1271 extern uint_t pvn_vmodsort_supported;
1272 
1273 #define	VN_HOLD(vp)	{ \
1274 	mutex_enter(&(vp)->v_lock); \
1275 	(vp)->v_count++; \
1276 	mutex_exit(&(vp)->v_lock); \
1277 }
1278 
1279 #define	VN_RELE(vp)	{ \
1280 	vn_rele(vp); \
1281 }
1282 
1283 #define	VN_SET_VFS_TYPE_DEV(vp, vfsp, type, dev)	{ \
1284 	(vp)->v_vfsp = (vfsp); \
1285 	(vp)->v_type = (type); \
1286 	(vp)->v_rdev = (dev); \
1287 }
1288 
1289 /*
1290  * Compare two vnodes for equality.  In general this macro should be used
1291  * in preference to calling VOP_CMP directly.
1292  */
1293 #define	VN_CMP(VP1, VP2)	((VP1) == (VP2) ? 1 : 	\
1294 	((VP1) && (VP2) && (vn_getops(VP1) == vn_getops(VP2)) ? \
1295 	VOP_CMP(VP1, VP2, NULL) : 0))
1296 
1297 extern struct vnode kvp;
1298 extern struct vnode zvp;
1299 
1300 #define	VN_ISKAS(vp)		((vp) == &kvp || (vp) == &zvp)
1301 
1302 #endif	/* _KERNEL */
1303 
1304 /*
1305  * Flags to VOP_SETATTR/VOP_GETATTR.
1306  */
1307 #define	ATTR_UTIME	0x01	/* non-default utime(2) request */
1308 #define	ATTR_EXEC	0x02	/* invocation from exec(2) */
1309 #define	ATTR_COMM	0x04	/* yield common vp attributes */
1310 #define	ATTR_HINT	0x08	/* information returned will be `hint' */
1311 #define	ATTR_REAL	0x10	/* yield attributes of the real vp */
1312 #define	ATTR_NOACLCHECK	0x20	/* Don't check ACL when checking permissions */
1313 #define	ATTR_TRIGGER	0x40	/* Mount first if vnode is a trigger mount */
1314 /*
1315  * Generally useful macros.
1316  */
1317 #define	VBSIZE(vp)	((vp)->v_vfsp->vfs_bsize)
1318 
1319 #define	VTOZONE(vp)	((vp)->v_vfsp->vfs_zone)
1320 
1321 #define	NULLVP		((struct vnode *)0)
1322 #define	NULLVPP		((struct vnode **)0)
1323 
1324 #ifdef	_KERNEL
1325 
1326 /*
1327  * Structure used while handling asynchronous VOP_PUTPAGE operations.
1328  */
1329 struct async_reqs {
1330 	struct async_reqs *a_next;	/* pointer to next arg struct */
1331 	struct vnode *a_vp;		/* vnode pointer */
1332 	u_offset_t a_off;			/* offset in file */
1333 	uint_t a_len;			/* size of i/o request */
1334 	int a_flags;			/* flags to indicate operation type */
1335 	struct cred *a_cred;		/* cred pointer	*/
1336 	ushort_t a_prealloced;		/* set if struct is pre-allocated */
1337 };
1338 
1339 /*
1340  * VN_DISPOSE() -- given a page pointer, safely invoke VOP_DISPOSE().
1341  */
1342 #define	VN_DISPOSE(pp, flag, dn, cr)	{ \
1343 	extern struct vnode kvp; \
1344 	if ((pp)->p_vnode != NULL && !VN_ISKAS((pp)->p_vnode)) \
1345 		VOP_DISPOSE((pp)->p_vnode, (pp), (flag), (dn), (cr), NULL); \
1346 	else if ((flag) == B_FREE) \
1347 		page_free((pp), (dn)); \
1348 	else \
1349 		page_destroy((pp), (dn)); \
1350 	}
1351 
1352 #endif	/* _KERNEL */
1353 
1354 #ifdef	__cplusplus
1355 }
1356 #endif
1357 
1358 #endif	/* _SYS_VNODE_H */
1359