1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21
22 /*
23 * Copyright (c) 1988, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Copyright (c) 2013, Joyent, Inc. All rights reserved.
25 * Copyright 2017 Nexenta Systems, Inc. All rights reserved.
26 */
27
28 /* Copyright (c) 1983, 1984, 1985, 1986, 1987, 1988, 1989 AT&T */
29 /* All Rights Reserved */
30
31 /*
32 * University Copyright- Copyright (c) 1982, 1986, 1988
33 * The Regents of the University of California
34 * All Rights Reserved
35 *
36 * University Acknowledgment- Portions of this document are derived from
37 * software developed by the University of California, Berkeley, and its
38 * contributors.
39 */
40
41 #ifndef _SYS_VNODE_H
42 #define _SYS_VNODE_H
43
44 #include <sys/types.h>
45 #include <sys/t_lock.h>
46 #include <sys/time_impl.h>
47 #include <sys/cred.h>
48 #include <sys/uio.h>
49 #include <sys/resource.h>
50 #include <vm/seg_enum.h>
51 #include <sys/kstat.h>
52 #include <sys/kmem.h>
53 #include <sys/list.h>
54 #include <sys/avl.h>
55 #ifdef _KERNEL
56 #include <sys/rwstlock.h>
57 #include <sys/buf.h>
58 #endif /* _KERNEL */
59
60 #ifdef __cplusplus
61 extern "C" {
62 #endif
63
64 #if defined(_KERNEL) || defined(_FAKE_KERNEL)
65
66 /*
67 * Statistics for all vnode operations.
68 * All operations record number of ops (since boot/mount/zero'ed).
69 * Certain I/O operations (read, write, readdir) also record number
70 * of bytes transferred.
71 * This appears in two places in the system: one is embedded in each
72 * vfs_t. There is also an array of vopstats_t structures allocated
73 * on a per-fstype basis.
74 */
75
76 #define VOPSTATS_STR "vopstats_" /* Initial string for vopstat kstats */
77
78 typedef struct vopstats {
79 kstat_named_t nopen; /* VOP_OPEN */
80 kstat_named_t nclose; /* VOP_CLOSE */
81 kstat_named_t nread; /* VOP_READ */
82 kstat_named_t read_bytes;
83 kstat_named_t nwrite; /* VOP_WRITE */
84 kstat_named_t write_bytes;
85 kstat_named_t nioctl; /* VOP_IOCTL */
86 kstat_named_t nsetfl; /* VOP_SETFL */
87 kstat_named_t ngetattr; /* VOP_GETATTR */
88 kstat_named_t nsetattr; /* VOP_SETATTR */
89 kstat_named_t naccess; /* VOP_ACCESS */
90 kstat_named_t nlookup; /* VOP_LOOKUP */
91 kstat_named_t ncreate; /* VOP_CREATE */
92 kstat_named_t nremove; /* VOP_REMOVE */
93 kstat_named_t nlink; /* VOP_LINK */
94 kstat_named_t nrename; /* VOP_RENAME */
95 kstat_named_t nmkdir; /* VOP_MKDIR */
96 kstat_named_t nrmdir; /* VOP_RMDIR */
97 kstat_named_t nreaddir; /* VOP_READDIR */
98 kstat_named_t readdir_bytes;
99 kstat_named_t nsymlink; /* VOP_SYMLINK */
100 kstat_named_t nreadlink; /* VOP_READLINK */
101 kstat_named_t nfsync; /* VOP_FSYNC */
102 kstat_named_t ninactive; /* VOP_INACTIVE */
103 kstat_named_t nfid; /* VOP_FID */
104 kstat_named_t nrwlock; /* VOP_RWLOCK */
105 kstat_named_t nrwunlock; /* VOP_RWUNLOCK */
106 kstat_named_t nseek; /* VOP_SEEK */
107 kstat_named_t ncmp; /* VOP_CMP */
108 kstat_named_t nfrlock; /* VOP_FRLOCK */
109 kstat_named_t nspace; /* VOP_SPACE */
110 kstat_named_t nrealvp; /* VOP_REALVP */
111 kstat_named_t ngetpage; /* VOP_GETPAGE */
112 kstat_named_t nputpage; /* VOP_PUTPAGE */
113 kstat_named_t nmap; /* VOP_MAP */
114 kstat_named_t naddmap; /* VOP_ADDMAP */
115 kstat_named_t ndelmap; /* VOP_DELMAP */
116 kstat_named_t npoll; /* VOP_POLL */
117 kstat_named_t ndump; /* VOP_DUMP */
118 kstat_named_t npathconf; /* VOP_PATHCONF */
119 kstat_named_t npageio; /* VOP_PAGEIO */
120 kstat_named_t ndumpctl; /* VOP_DUMPCTL */
121 kstat_named_t ndispose; /* VOP_DISPOSE */
122 kstat_named_t nsetsecattr; /* VOP_SETSECATTR */
123 kstat_named_t ngetsecattr; /* VOP_GETSECATTR */
124 kstat_named_t nshrlock; /* VOP_SHRLOCK */
125 kstat_named_t nvnevent; /* VOP_VNEVENT */
126 kstat_named_t nreqzcbuf; /* VOP_REQZCBUF */
127 kstat_named_t nretzcbuf; /* VOP_RETZCBUF */
128 } vopstats_t;
129 #endif // defined(_KERNEL) || defined(_FAKE_KERNEL)
130
131 /*
132 * The vnode is the focus of all file activity in UNIX.
133 * A vnode is allocated for each active file, each current
134 * directory, each mounted-on file, and the root.
135 *
136 * Each vnode is usually associated with a file-system-specific node (for
137 * UFS, this is the in-memory inode). Generally, a vnode and an fs-node
138 * should be created and destroyed together as a pair.
139 *
140 * If a vnode is reused for a new file, it should be reinitialized by calling
141 * either vn_reinit() or vn_recycle().
142 *
143 * vn_reinit() resets the entire vnode as if it was returned by vn_alloc().
144 * The caller is responsible for setting up the entire vnode after calling
145 * vn_reinit(). This is important when using kmem caching where the vnode is
146 * allocated by a constructor, for instance.
147 *
148 * vn_recycle() is used when the file system keeps some state around in both
149 * the vnode and the associated FS-node. In UFS, for example, the inode of
150 * a deleted file can be reused immediately. The v_data, v_vfsp, v_op, etc.
151 * remains the same but certain fields related to the previous instance need
152 * to be reset. In particular:
153 * v_femhead
154 * v_path
155 * v_rdcnt, v_wrcnt
156 * v_mmap_read, v_mmap_write
157 */
158
159 /*
160 * vnode types. VNON means no type. These values are unrelated to
161 * values in on-disk inodes.
162 */
163 typedef enum vtype {
164 VNON = 0,
165 VREG = 1,
166 VDIR = 2,
167 VBLK = 3,
168 VCHR = 4,
169 VLNK = 5,
170 VFIFO = 6,
171 VDOOR = 7,
172 VPROC = 8,
173 VSOCK = 9,
174 VPORT = 10,
175 VBAD = 11
176 } vtype_t;
177
178 #if defined(_KERNEL) || defined(_FAKE_KERNEL)
179
180 /*
181 * VSD - Vnode Specific Data
182 * Used to associate additional private data with a vnode.
183 */
184 struct vsd_node {
185 list_node_t vs_nodes; /* list of all VSD nodes */
186 uint_t vs_nkeys; /* entries in value array */
187 void **vs_value; /* array of value/key */
188 };
189
190 /*
191 * Many of the fields in the vnode are read-only once they are initialized
192 * at vnode creation time. Other fields are protected by locks.
193 *
194 * IMPORTANT: vnodes should be created ONLY by calls to vn_alloc(). They
195 * may not be embedded into the file-system specific node (inode). The
196 * size of vnodes may change.
197 *
198 * The v_lock protects:
199 * v_flag
200 * v_stream
201 * v_count
202 * v_shrlocks
203 * v_path
204 * v_vsd
205 * v_xattrdir
206 *
207 * A special lock (implemented by vn_vfswlock in vnode.c) protects:
208 * v_vfsmountedhere
209 *
210 * The global flock_lock mutex (in flock.c) protects:
211 * v_filocks
212 *
213 * IMPORTANT NOTE:
214 *
215 * The following vnode fields are considered public and may safely be
216 * accessed by file systems or other consumers:
217 *
218 * v_lock
219 * v_flag
220 * v_count
221 * v_data
222 * v_vfsp
223 * v_stream
224 * v_type
225 * v_rdev
226 *
227 * ALL OTHER FIELDS SHOULD BE ACCESSED ONLY BY THE OWNER OF THAT FIELD.
228 * In particular, file systems should not access other fields; they may
229 * change or even be removed. The functionality which was once provided
230 * by these fields is available through vn_* functions.
231 */
232
233 struct fem_head; /* from fem.h */
234
235 typedef struct vnode {
236 kmutex_t v_lock; /* protects vnode fields */
237 uint_t v_flag; /* vnode flags (see below) */
238 uint_t v_count; /* reference count */
239 void *v_data; /* private data for fs */
240 struct vfs *v_vfsp; /* ptr to containing VFS */
241 struct stdata *v_stream; /* associated stream */
242 enum vtype v_type; /* vnode type */
243 dev_t v_rdev; /* device (VCHR, VBLK) */
244
245 /* PRIVATE FIELDS BELOW - DO NOT USE */
246
247 struct vfs *v_vfsmountedhere; /* ptr to vfs mounted here */
248 struct vnodeops *v_op; /* vnode operations */
249 krwlock_t v_nbllock; /* sync for NBMAND locks */
250 char *v_path; /* cached path */
251 uint_t v_rdcnt; /* open for read count (VREG only) */
252 uint_t v_wrcnt; /* open for write count (VREG only) */
253 struct vnode *v_xattrdir; /* unnamed extended attr dir (GFS) */
254
255 /* Private to the fake vnode impl. */
256
257 int v_fd;
258 dev_t v_st_dev;
259 ino_t v_st_ino;
260 avl_node_t v_avl_node;
261 int v_vfsrlocks;
262 } vnode_t;
263
264 #define IS_DEVVP(vp) \
265 ((vp)->v_type == VCHR || (vp)->v_type == VBLK || (vp)->v_type == VFIFO)
266
267 #define VNODE_ALIGN 16
268
269 /*
270 * vnode flags.
271 */
272 #define VROOT 0x01 /* root of its file system */
273 #define VNOCACHE 0x02 /* don't keep cache pages on vnode */
274 #define VNOMAP 0x04 /* file cannot be mapped/faulted */
275 #define VDUP 0x08 /* file should be dup'ed rather then opened */
276 #define VNOSWAP 0x10 /* file cannot be used as virtual swap device */
277 #define VNOMOUNT 0x20 /* file cannot be covered by mount */
278 #define VISSWAP 0x40 /* vnode is being used for swap */
279 #define VSWAPLIKE 0x80 /* vnode acts like swap (but may not be) */
280
281 #define IS_SWAPVP(vp) (((vp)->v_flag & (VISSWAP | VSWAPLIKE)) != 0)
282
283 #else // defined(_KERNEL) || defined(_FAKE_KERNEL)
284 typedef struct vnode vnode_t;
285 #endif // defined(_KERNEL) || defined(_FAKE_KERNEL)
286
287 #if defined(_KERNEL)
288 typedef struct vn_vfslocks_entry {
289 rwstlock_t ve_lock;
290 void *ve_vpvfs;
291 struct vn_vfslocks_entry *ve_next;
292 uint32_t ve_refcnt;
293 char pad[64 - sizeof (rwstlock_t) - 2 * sizeof (void *) - \
294 sizeof (uint32_t)];
295 } vn_vfslocks_entry_t;
296 #endif /* _KERNEL */
297
298 /*
299 * The following two flags are used to lock the v_vfsmountedhere field
300 */
301 #define VVFSLOCK 0x100
302 #define VVFSWAIT 0x200
303
304 /*
305 * Used to serialize VM operations on a vnode
306 */
307 #define VVMLOCK 0x400
308
309 /*
310 * Tell vn_open() not to fail a directory open for writing but
311 * to go ahead and call VOP_OPEN() to let the filesystem check.
312 */
313 #define VDIROPEN 0x800
314
315 /*
316 * Flag to let the VM system know that this file is most likely a binary
317 * or shared library since it has been mmap()ed EXEC at some time.
318 */
319 #define VVMEXEC 0x1000
320
321 #define VPXFS 0x2000 /* clustering: global fs proxy vnode */
322
323 #define IS_PXFSVP(vp) ((vp)->v_flag & VPXFS)
324
325 #define V_XATTRDIR 0x4000 /* attribute unnamed directory */
326
327 #define IS_XATTRDIR(vp) ((vp)->v_flag & V_XATTRDIR)
328
329 #define V_LOCALITY 0x8000 /* whether locality aware */
330
331 /*
332 * Flag that indicates the VM should maintain the v_pages list with all modified
333 * pages on one end and unmodified pages at the other. This makes finding dirty
334 * pages to write back to disk much faster at the expense of taking a minor
335 * fault on the first store instruction which touches a writable page.
336 */
337 #define VMODSORT (0x10000)
338 #define IS_VMODSORT(vp) \
339 (pvn_vmodsort_supported != 0 && ((vp)->v_flag & VMODSORT) != 0)
340
341 #define VISSWAPFS 0x20000 /* vnode is being used for swapfs */
342
343 /*
344 * The mdb memstat command assumes that IS_SWAPFSVP only uses the
345 * vnode's v_flag field. If this changes, cache the additional
346 * fields in mdb; see vn_get in mdb/common/modules/genunix/memory.c
347 */
348 #define IS_SWAPFSVP(vp) (((vp)->v_flag & VISSWAPFS) != 0)
349
350 #define V_SYSATTR 0x40000 /* vnode is a GFS system attribute */
351
352 /*
353 * Vnode attributes. A bit-mask is supplied as part of the
354 * structure to indicate the attributes the caller wants to
355 * set (setattr) or extract (getattr).
356 */
357
358 /*
359 * Note that va_nodeid and va_nblocks are 64bit data type.
360 * We support large files over NFSV3. With Solaris client and
361 * Server that generates 64bit ino's and sizes these fields
362 * will overflow if they are 32 bit sizes.
363 */
364
365 typedef struct vattr {
366 uint_t va_mask; /* bit-mask of attributes */
367 vtype_t va_type; /* vnode type (for create) */
368 mode_t va_mode; /* file access mode */
369 uid_t va_uid; /* owner user id */
370 gid_t va_gid; /* owner group id */
371 dev_t va_fsid; /* file system id (dev for now) */
372 u_longlong_t va_nodeid; /* node id */
373 nlink_t va_nlink; /* number of references to file */
374 u_offset_t va_size; /* file size in bytes */
375 timestruc_t va_atime; /* time of last access */
376 timestruc_t va_mtime; /* time of last modification */
377 timestruc_t va_ctime; /* time of last status change */
378 dev_t va_rdev; /* device the file represents */
379 uint_t va_blksize; /* fundamental block size */
380 u_longlong_t va_nblocks; /* # of blocks allocated */
381 uint_t va_seq; /* sequence number */
382 } vattr_t;
383
384 #define AV_SCANSTAMP_SZ 32 /* length of anti-virus scanstamp */
385
386 /*
387 * Structure of all optional attributes.
388 */
389 typedef struct xoptattr {
390 timestruc_t xoa_createtime; /* Create time of file */
391 uint8_t xoa_archive;
392 uint8_t xoa_system;
393 uint8_t xoa_readonly;
394 uint8_t xoa_hidden;
395 uint8_t xoa_nounlink;
396 uint8_t xoa_immutable;
397 uint8_t xoa_appendonly;
398 uint8_t xoa_nodump;
399 uint8_t xoa_opaque;
400 uint8_t xoa_av_quarantined;
401 uint8_t xoa_av_modified;
402 uint8_t xoa_av_scanstamp[AV_SCANSTAMP_SZ];
403 uint8_t xoa_reparse;
404 uint64_t xoa_generation;
405 uint8_t xoa_offline;
406 uint8_t xoa_sparse;
407 } xoptattr_t;
408
409 /*
410 * The xvattr structure is really a variable length structure that
411 * is made up of:
412 * - The classic vattr_t (xva_vattr)
413 * - a 32 bit quantity (xva_mapsize) that specifies the size of the
414 * attribute bitmaps in 32 bit words.
415 * - A pointer to the returned attribute bitmap (needed because the
416 * previous element, the requested attribute bitmap) is variable lenth.
417 * - The requested attribute bitmap, which is an array of 32 bit words.
418 * Callers use the XVA_SET_REQ() macro to set the bits corresponding to
419 * the attributes that are being requested.
420 * - The returned attribute bitmap, which is an array of 32 bit words.
421 * File systems that support optional attributes use the XVA_SET_RTN()
422 * macro to set the bits corresponding to the attributes that are being
423 * returned.
424 * - The xoptattr_t structure which contains the attribute values
425 *
426 * xva_mapsize determines how many words in the attribute bitmaps.
427 * Immediately following the attribute bitmaps is the xoptattr_t.
428 * xva_getxoptattr() is used to get the pointer to the xoptattr_t
429 * section.
430 */
431
432 #define XVA_MAPSIZE 3 /* Size of attr bitmaps */
433 #define XVA_MAGIC 0x78766174 /* Magic # for verification */
434
435 /*
436 * The xvattr structure is an extensible structure which permits optional
437 * attributes to be requested/returned. File systems may or may not support
438 * optional attributes. They do so at their own discretion but if they do
439 * support optional attributes, they must register the VFSFT_XVATTR feature
440 * so that the optional attributes can be set/retrived.
441 *
442 * The fields of the xvattr structure are:
443 *
444 * xva_vattr - The first element of an xvattr is a legacy vattr structure
445 * which includes the common attributes. If AT_XVATTR is set in the va_mask
446 * then the entire structure is treated as an xvattr. If AT_XVATTR is not
447 * set, then only the xva_vattr structure can be used.
448 *
449 * xva_magic - 0x78766174 (hex for "xvat"). Magic number for verification.
450 *
451 * xva_mapsize - Size of requested and returned attribute bitmaps.
452 *
453 * xva_rtnattrmapp - Pointer to xva_rtnattrmap[]. We need this since the
454 * size of the array before it, xva_reqattrmap[], could change which means
455 * the location of xva_rtnattrmap[] could change. This will allow unbundled
456 * file systems to find the location of xva_rtnattrmap[] when the sizes change.
457 *
458 * xva_reqattrmap[] - Array of requested attributes. Attributes are
459 * represented by a specific bit in a specific element of the attribute
460 * map array. Callers set the bits corresponding to the attributes
461 * that the caller wants to get/set.
462 *
463 * xva_rtnattrmap[] - Array of attributes that the file system was able to
464 * process. Not all file systems support all optional attributes. This map
465 * informs the caller which attributes the underlying file system was able
466 * to set/get. (Same structure as the requested attributes array in terms
467 * of each attribute corresponding to specific bits and array elements.)
468 *
469 * xva_xoptattrs - Structure containing values of optional attributes.
470 * These values are only valid if the corresponding bits in xva_reqattrmap
471 * are set and the underlying file system supports those attributes.
472 */
473 typedef struct xvattr {
474 vattr_t xva_vattr; /* Embedded vattr structure */
475 uint32_t xva_magic; /* Magic Number */
476 uint32_t xva_mapsize; /* Size of attr bitmap (32-bit words) */
477 uint32_t *xva_rtnattrmapp; /* Ptr to xva_rtnattrmap[] */
478 uint32_t xva_reqattrmap[XVA_MAPSIZE]; /* Requested attrs */
479 uint32_t xva_rtnattrmap[XVA_MAPSIZE]; /* Returned attrs */
480 xoptattr_t xva_xoptattrs; /* Optional attributes */
481 } xvattr_t;
482
483 #ifdef _SYSCALL32
484 /*
485 * For bigtypes time_t changed to 64 bit on the 64-bit kernel.
486 * Define an old version for user/kernel interface
487 */
488
489 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
490 #pragma pack(4)
491 #endif
492
493 typedef struct vattr32 {
494 uint32_t va_mask; /* bit-mask of attributes */
495 vtype_t va_type; /* vnode type (for create) */
496 mode32_t va_mode; /* file access mode */
497 uid32_t va_uid; /* owner user id */
498 gid32_t va_gid; /* owner group id */
499 dev32_t va_fsid; /* file system id (dev for now) */
500 u_longlong_t va_nodeid; /* node id */
501 nlink_t va_nlink; /* number of references to file */
502 u_offset_t va_size; /* file size in bytes */
503 timestruc32_t va_atime; /* time of last access */
504 timestruc32_t va_mtime; /* time of last modification */
505 timestruc32_t va_ctime; /* time of last status change */
506 dev32_t va_rdev; /* device the file represents */
507 uint32_t va_blksize; /* fundamental block size */
508 u_longlong_t va_nblocks; /* # of blocks allocated */
509 uint32_t va_seq; /* sequence number */
510 } vattr32_t;
511
512 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
513 #pragma pack()
514 #endif
515
516 #else /* not _SYSCALL32 */
517 #define vattr32 vattr
518 typedef vattr_t vattr32_t;
519 #endif /* _SYSCALL32 */
520
521 /*
522 * Attributes of interest to the caller of setattr or getattr.
523 */
524 #define AT_TYPE 0x00001
525 #define AT_MODE 0x00002
526 #define AT_UID 0x00004
527 #define AT_GID 0x00008
528 #define AT_FSID 0x00010
529 #define AT_NODEID 0x00020
530 #define AT_NLINK 0x00040
531 #define AT_SIZE 0x00080
532 #define AT_ATIME 0x00100
533 #define AT_MTIME 0x00200
534 #define AT_CTIME 0x00400
535 #define AT_RDEV 0x00800
536 #define AT_BLKSIZE 0x01000
537 #define AT_NBLOCKS 0x02000
538 /* 0x04000 */ /* unused */
539 #define AT_SEQ 0x08000
540 /*
541 * If AT_XVATTR is set then there are additional bits to process in
542 * the xvattr_t's attribute bitmap. If this is not set then the bitmap
543 * MUST be ignored. Note that this bit must be set/cleared explicitly.
544 * That is, setting AT_ALL will NOT set AT_XVATTR.
545 */
546 #define AT_XVATTR 0x10000
547
548 #define AT_ALL (AT_TYPE|AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|\
549 AT_NLINK|AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|\
550 AT_RDEV|AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
551
552 #define AT_STAT (AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|AT_NLINK|\
553 AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|AT_RDEV|AT_TYPE)
554
555 #define AT_TIMES (AT_ATIME|AT_MTIME|AT_CTIME)
556
557 #define AT_NOSET (AT_NLINK|AT_RDEV|AT_FSID|AT_NODEID|AT_TYPE|\
558 AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
559
560 /*
561 * Attribute bits used in the extensible attribute's (xva's) attribute
562 * bitmaps. Note that the bitmaps are made up of a variable length number
563 * of 32-bit words. The convention is to use XAT{n}_{attrname} where "n"
564 * is the element in the bitmap (starting at 1). This convention is for
565 * the convenience of the maintainer to keep track of which element each
566 * attribute belongs to.
567 *
568 * NOTE THAT CONSUMERS MUST *NOT* USE THE XATn_* DEFINES DIRECTLY. CONSUMERS
569 * MUST USE THE XAT_* DEFINES.
570 */
571 #define XAT0_INDEX 0LL /* Index into bitmap for XAT0 attrs */
572 #define XAT0_CREATETIME 0x00000001 /* Create time of file */
573 #define XAT0_ARCHIVE 0x00000002 /* Archive */
574 #define XAT0_SYSTEM 0x00000004 /* System */
575 #define XAT0_READONLY 0x00000008 /* Readonly */
576 #define XAT0_HIDDEN 0x00000010 /* Hidden */
577 #define XAT0_NOUNLINK 0x00000020 /* Nounlink */
578 #define XAT0_IMMUTABLE 0x00000040 /* immutable */
579 #define XAT0_APPENDONLY 0x00000080 /* appendonly */
580 #define XAT0_NODUMP 0x00000100 /* nodump */
581 #define XAT0_OPAQUE 0x00000200 /* opaque */
582 #define XAT0_AV_QUARANTINED 0x00000400 /* anti-virus quarantine */
583 #define XAT0_AV_MODIFIED 0x00000800 /* anti-virus modified */
584 #define XAT0_AV_SCANSTAMP 0x00001000 /* anti-virus scanstamp */
585 #define XAT0_REPARSE 0x00002000 /* FS reparse point */
586 #define XAT0_GEN 0x00004000 /* object generation number */
587 #define XAT0_OFFLINE 0x00008000 /* offline */
588 #define XAT0_SPARSE 0x00010000 /* sparse */
589
590 #define XAT0_ALL_ATTRS (XAT0_CREATETIME|XAT0_ARCHIVE|XAT0_SYSTEM| \
591 XAT0_READONLY|XAT0_HIDDEN|XAT0_NOUNLINK|XAT0_IMMUTABLE|XAT0_APPENDONLY| \
592 XAT0_NODUMP|XAT0_OPAQUE|XAT0_AV_QUARANTINED| XAT0_AV_MODIFIED| \
593 XAT0_AV_SCANSTAMP|XAT0_REPARSE|XATO_GEN|XAT0_OFFLINE|XAT0_SPARSE)
594
595 /* Support for XAT_* optional attributes */
596 #define XVA_MASK 0xffffffff /* Used to mask off 32 bits */
597 #define XVA_SHFT 32 /* Used to shift index */
598
599 /*
600 * Used to pry out the index and attribute bits from the XAT_* attributes
601 * defined below. Note that we're masking things down to 32 bits then
602 * casting to uint32_t.
603 */
604 #define XVA_INDEX(attr) ((uint32_t)(((attr) >> XVA_SHFT) & XVA_MASK))
605 #define XVA_ATTRBIT(attr) ((uint32_t)((attr) & XVA_MASK))
606
607 /*
608 * The following defines present a "flat namespace" so that consumers don't
609 * need to keep track of which element belongs to which bitmap entry.
610 *
611 * NOTE THAT THESE MUST NEVER BE OR-ed TOGETHER
612 */
613 #define XAT_CREATETIME ((XAT0_INDEX << XVA_SHFT) | XAT0_CREATETIME)
614 #define XAT_ARCHIVE ((XAT0_INDEX << XVA_SHFT) | XAT0_ARCHIVE)
615 #define XAT_SYSTEM ((XAT0_INDEX << XVA_SHFT) | XAT0_SYSTEM)
616 #define XAT_READONLY ((XAT0_INDEX << XVA_SHFT) | XAT0_READONLY)
617 #define XAT_HIDDEN ((XAT0_INDEX << XVA_SHFT) | XAT0_HIDDEN)
618 #define XAT_NOUNLINK ((XAT0_INDEX << XVA_SHFT) | XAT0_NOUNLINK)
619 #define XAT_IMMUTABLE ((XAT0_INDEX << XVA_SHFT) | XAT0_IMMUTABLE)
620 #define XAT_APPENDONLY ((XAT0_INDEX << XVA_SHFT) | XAT0_APPENDONLY)
621 #define XAT_NODUMP ((XAT0_INDEX << XVA_SHFT) | XAT0_NODUMP)
622 #define XAT_OPAQUE ((XAT0_INDEX << XVA_SHFT) | XAT0_OPAQUE)
623 #define XAT_AV_QUARANTINED ((XAT0_INDEX << XVA_SHFT) | XAT0_AV_QUARANTINED)
624 #define XAT_AV_MODIFIED ((XAT0_INDEX << XVA_SHFT) | XAT0_AV_MODIFIED)
625 #define XAT_AV_SCANSTAMP ((XAT0_INDEX << XVA_SHFT) | XAT0_AV_SCANSTAMP)
626 #define XAT_REPARSE ((XAT0_INDEX << XVA_SHFT) | XAT0_REPARSE)
627 #define XAT_GEN ((XAT0_INDEX << XVA_SHFT) | XAT0_GEN)
628 #define XAT_OFFLINE ((XAT0_INDEX << XVA_SHFT) | XAT0_OFFLINE)
629 #define XAT_SPARSE ((XAT0_INDEX << XVA_SHFT) | XAT0_SPARSE)
630
631 /*
632 * The returned attribute map array (xva_rtnattrmap[]) is located past the
633 * requested attribute map array (xva_reqattrmap[]). Its location changes
634 * when the array sizes change. We use a separate pointer in a known location
635 * (xva_rtnattrmapp) to hold the location of xva_rtnattrmap[]. This is
636 * set in xva_init()
637 */
638 #define XVA_RTNATTRMAP(xvap) ((xvap)->xva_rtnattrmapp)
639
640 /*
641 * XVA_SET_REQ() sets an attribute bit in the proper element in the bitmap
642 * of requested attributes (xva_reqattrmap[]).
643 */
644 #define XVA_SET_REQ(xvap, attr) \
645 ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR); \
646 ASSERT((xvap)->xva_magic == XVA_MAGIC); \
647 (xvap)->xva_reqattrmap[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
648 /*
649 * XVA_CLR_REQ() clears an attribute bit in the proper element in the bitmap
650 * of requested attributes (xva_reqattrmap[]).
651 */
652 #define XVA_CLR_REQ(xvap, attr) \
653 ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR); \
654 ASSERT((xvap)->xva_magic == XVA_MAGIC); \
655 (xvap)->xva_reqattrmap[XVA_INDEX(attr)] &= ~XVA_ATTRBIT(attr)
656
657 /*
658 * XVA_SET_RTN() sets an attribute bit in the proper element in the bitmap
659 * of returned attributes (xva_rtnattrmap[]).
660 */
661 #define XVA_SET_RTN(xvap, attr) \
662 ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR); \
663 ASSERT((xvap)->xva_magic == XVA_MAGIC); \
664 (XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
665
666 /*
667 * XVA_ISSET_REQ() checks the requested attribute bitmap (xva_reqattrmap[])
668 * to see of the corresponding attribute bit is set. If so, returns non-zero.
669 */
670 #define XVA_ISSET_REQ(xvap, attr) \
671 ((((xvap)->xva_vattr.va_mask | AT_XVATTR) && \
672 ((xvap)->xva_magic == XVA_MAGIC) && \
673 ((xvap)->xva_mapsize > XVA_INDEX(attr))) ? \
674 ((xvap)->xva_reqattrmap[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) : 0)
675
676 /*
677 * XVA_ISSET_RTN() checks the returned attribute bitmap (xva_rtnattrmap[])
678 * to see of the corresponding attribute bit is set. If so, returns non-zero.
679 */
680 #define XVA_ISSET_RTN(xvap, attr) \
681 ((((xvap)->xva_vattr.va_mask | AT_XVATTR) && \
682 ((xvap)->xva_magic == XVA_MAGIC) && \
683 ((xvap)->xva_mapsize > XVA_INDEX(attr))) ? \
684 ((XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) : 0)
685
686 /*
687 * Modes. Some values same as S_xxx entries from stat.h for convenience.
688 */
689 #define VSUID 04000 /* set user id on execution */
690 #define VSGID 02000 /* set group id on execution */
691 #define VSVTX 01000 /* save swapped text even after use */
692
693 /*
694 * Permissions.
695 */
696 #define VREAD 00400
697 #define VWRITE 00200
698 #define VEXEC 00100
699
700 #define MODEMASK 07777 /* mode bits plus permission bits */
701 #define PERMMASK 00777 /* permission bits */
702
703 /*
704 * VOP_ACCESS flags
705 */
706 #define V_ACE_MASK 0x1 /* mask represents NFSv4 ACE permissions */
707 #define V_APPEND 0x2 /* want to do append only check */
708
709 /*
710 * Check whether mandatory file locking is enabled.
711 */
712
713 #define MANDMODE(mode) (((mode) & (VSGID|(VEXEC>>3))) == VSGID)
714 #define MANDLOCK(vp, mode) ((vp)->v_type == VREG && MANDMODE(mode))
715
716 /*
717 * Flags for vnode operations.
718 */
719 enum rm { RMFILE, RMDIRECTORY }; /* rm or rmdir (remove) */
720 enum symfollow { NO_FOLLOW, FOLLOW }; /* follow symlinks (or not) */
721 enum vcexcl { NONEXCL, EXCL }; /* (non)excl create */
722 enum create { CRCREAT, CRMKNOD, CRMKDIR }; /* reason for create */
723
724 typedef enum rm rm_t;
725 typedef enum symfollow symfollow_t;
726 typedef enum vcexcl vcexcl_t;
727 typedef enum create create_t;
728
729 /* Vnode Events - Used by VOP_VNEVENT */
730 typedef enum vnevent {
731 VE_SUPPORT = 0, /* Query */
732 VE_RENAME_SRC = 1, /* Rename, with vnode as source */
733 VE_RENAME_DEST = 2, /* Rename, with vnode as target/destination */
734 VE_REMOVE = 3, /* Remove of vnode's name */
735 VE_RMDIR = 4, /* Remove of directory vnode's name */
736 VE_CREATE = 5, /* Create with vnode's name which exists */
737 VE_LINK = 6, /* Link with vnode's name as source */
738 VE_RENAME_DEST_DIR = 7, /* Rename with vnode as target dir */
739 VE_MOUNTEDOVER = 8, /* File or Filesystem got mounted over vnode */
740 VE_TRUNCATE = 9 /* Truncate */
741 } vnevent_t;
742
743 /*
744 * Values for checking vnode open and map counts
745 */
746 enum v_mode { V_READ, V_WRITE, V_RDORWR, V_RDANDWR };
747
748 typedef enum v_mode v_mode_t;
749
750 #define V_TRUE 1
751 #define V_FALSE 0
752
753 /*
754 * Structure used on VOP_GETSECATTR and VOP_SETSECATTR operations
755 */
756
757 typedef struct vsecattr {
758 uint_t vsa_mask; /* See below */
759 int vsa_aclcnt; /* ACL entry count */
760 void *vsa_aclentp; /* pointer to ACL entries */
761 int vsa_dfaclcnt; /* default ACL entry count */
762 void *vsa_dfaclentp; /* pointer to default ACL entries */
763 size_t vsa_aclentsz; /* ACE size in bytes of vsa_aclentp */
764 uint_t vsa_aclflags; /* ACE ACL flags */
765 } vsecattr_t;
766
767 /* vsa_mask values */
768 #define VSA_ACL 0x0001
769 #define VSA_ACLCNT 0x0002
770 #define VSA_DFACL 0x0004
771 #define VSA_DFACLCNT 0x0008
772 #define VSA_ACE 0x0010
773 #define VSA_ACECNT 0x0020
774 #define VSA_ACE_ALLTYPES 0x0040
775 #define VSA_ACE_ACLFLAGS 0x0080 /* get/set ACE ACL flags */
776
777 /*
778 * Structure used by various vnode operations to determine
779 * the context (pid, host, identity) of a caller.
780 *
781 * The cc_caller_id is used to identify one or more callers who invoke
782 * operations, possibly on behalf of others. For example, the NFS
783 * server could have it's own cc_caller_id which can be detected by
784 * vnode/vfs operations or (FEM) monitors on those operations. New
785 * caller IDs are generated by fs_new_caller_id().
786 */
787 typedef struct caller_context {
788 pid_t cc_pid; /* Process ID of the caller */
789 int cc_sysid; /* System ID, used for remote calls */
790 u_longlong_t cc_caller_id; /* Identifier for (set of) caller(s) */
791 ulong_t cc_flags;
792 } caller_context_t;
793
794 /*
795 * Flags for caller context. The caller sets CC_DONTBLOCK if it does not
796 * want to block inside of a FEM monitor. The monitor will set CC_WOULDBLOCK
797 * and return EAGAIN if the operation would have blocked.
798 */
799 #define CC_WOULDBLOCK 0x01
800 #define CC_DONTBLOCK 0x02
801
802 /*
803 * Structure tags for function prototypes, defined elsewhere.
804 */
805 struct pathname;
806 struct fid;
807 struct flock64;
808 struct flk_callback;
809 struct shrlock;
810 struct page;
811 struct seg;
812 struct as;
813 struct pollhead;
814 struct taskq;
815
816 #if defined(_KERNEL) || defined(_FAKE_KERNEL)
817
818 /*
819 * VNODE_OPS defines all the vnode operations. It is used to define
820 * the vnodeops structure (below) and the fs_func_p union (vfs_opreg.h).
821 */
822 #define VNODE_OPS \
823 int (*vop_open)(vnode_t **, int, cred_t *, \
824 caller_context_t *); \
825 int (*vop_close)(vnode_t *, int, int, offset_t, cred_t *, \
826 caller_context_t *); \
827 int (*vop_read)(vnode_t *, uio_t *, int, cred_t *, \
828 caller_context_t *); \
829 int (*vop_write)(vnode_t *, uio_t *, int, cred_t *, \
830 caller_context_t *); \
831 int (*vop_ioctl)(vnode_t *, int, intptr_t, int, cred_t *, \
832 int *, caller_context_t *); \
833 int (*vop_setfl)(vnode_t *, int, int, cred_t *, \
834 caller_context_t *); \
835 int (*vop_getattr)(vnode_t *, vattr_t *, int, cred_t *, \
836 caller_context_t *); \
837 int (*vop_setattr)(vnode_t *, vattr_t *, int, cred_t *, \
838 caller_context_t *); \
839 int (*vop_access)(vnode_t *, int, int, cred_t *, \
840 caller_context_t *); \
841 int (*vop_lookup)(vnode_t *, char *, vnode_t **, \
842 struct pathname *, \
843 int, vnode_t *, cred_t *, \
844 caller_context_t *, int *, \
845 struct pathname *); \
846 int (*vop_create)(vnode_t *, char *, vattr_t *, vcexcl_t, \
847 int, vnode_t **, cred_t *, int, \
848 caller_context_t *, vsecattr_t *); \
849 int (*vop_remove)(vnode_t *, char *, cred_t *, \
850 caller_context_t *, int); \
851 int (*vop_link)(vnode_t *, vnode_t *, char *, cred_t *, \
852 caller_context_t *, int); \
853 int (*vop_rename)(vnode_t *, char *, vnode_t *, char *, \
854 cred_t *, caller_context_t *, int); \
855 int (*vop_mkdir)(vnode_t *, char *, vattr_t *, vnode_t **, \
856 cred_t *, caller_context_t *, int, \
857 vsecattr_t *); \
858 int (*vop_rmdir)(vnode_t *, char *, vnode_t *, cred_t *, \
859 caller_context_t *, int); \
860 int (*vop_readdir)(vnode_t *, uio_t *, cred_t *, int *, \
861 caller_context_t *, int); \
862 int (*vop_symlink)(vnode_t *, char *, vattr_t *, char *, \
863 cred_t *, caller_context_t *, int); \
864 int (*vop_readlink)(vnode_t *, uio_t *, cred_t *, \
865 caller_context_t *); \
866 int (*vop_fsync)(vnode_t *, int, cred_t *, \
867 caller_context_t *); \
868 void (*vop_inactive)(vnode_t *, cred_t *, \
869 caller_context_t *); \
870 int (*vop_fid)(vnode_t *, struct fid *, \
871 caller_context_t *); \
872 int (*vop_rwlock)(vnode_t *, int, caller_context_t *); \
873 void (*vop_rwunlock)(vnode_t *, int, caller_context_t *); \
874 int (*vop_seek)(vnode_t *, offset_t, offset_t *, \
875 caller_context_t *); \
876 int (*vop_cmp)(vnode_t *, vnode_t *, caller_context_t *); \
877 int (*vop_frlock)(vnode_t *, int, struct flock64 *, \
878 int, offset_t, \
879 struct flk_callback *, cred_t *, \
880 caller_context_t *); \
881 int (*vop_space)(vnode_t *, int, struct flock64 *, \
882 int, offset_t, \
883 cred_t *, caller_context_t *); \
884 int (*vop_realvp)(vnode_t *, vnode_t **, \
885 caller_context_t *); \
886 int (*vop_getpage)(vnode_t *, offset_t, size_t, uint_t *, \
887 struct page **, size_t, struct seg *, \
888 caddr_t, enum seg_rw, cred_t *, \
889 caller_context_t *); \
890 int (*vop_putpage)(vnode_t *, offset_t, size_t, \
891 int, cred_t *, caller_context_t *); \
892 int (*vop_map)(vnode_t *, offset_t, struct as *, \
893 caddr_t *, size_t, \
894 uchar_t, uchar_t, uint_t, cred_t *, \
895 caller_context_t *); \
896 int (*vop_addmap)(vnode_t *, offset_t, struct as *, \
897 caddr_t, size_t, \
898 uchar_t, uchar_t, uint_t, cred_t *, \
899 caller_context_t *); \
900 int (*vop_delmap)(vnode_t *, offset_t, struct as *, \
901 caddr_t, size_t, \
902 uint_t, uint_t, uint_t, cred_t *, \
903 caller_context_t *); \
904 int (*vop_poll)(vnode_t *, short, int, short *, \
905 struct pollhead **, \
906 caller_context_t *); \
907 int (*vop_dump)(vnode_t *, caddr_t, offset_t, offset_t, \
908 caller_context_t *); \
909 int (*vop_pathconf)(vnode_t *, int, ulong_t *, cred_t *, \
910 caller_context_t *); \
911 int (*vop_pageio)(vnode_t *, struct page *, \
912 u_offset_t, size_t, int, cred_t *, \
913 caller_context_t *); \
914 int (*vop_dumpctl)(vnode_t *, int, offset_t *, \
915 caller_context_t *); \
916 void (*vop_dispose)(vnode_t *, struct page *, \
917 int, int, cred_t *, \
918 caller_context_t *); \
919 int (*vop_setsecattr)(vnode_t *, vsecattr_t *, \
920 int, cred_t *, caller_context_t *); \
921 int (*vop_getsecattr)(vnode_t *, vsecattr_t *, \
922 int, cred_t *, caller_context_t *); \
923 int (*vop_shrlock)(vnode_t *, int, struct shrlock *, \
924 int, cred_t *, caller_context_t *); \
925 int (*vop_vnevent)(vnode_t *, vnevent_t, vnode_t *, \
926 char *, caller_context_t *); \
927 int (*vop_reqzcbuf)(vnode_t *, enum uio_rw, xuio_t *, \
928 cred_t *, caller_context_t *); \
929 int (*vop_retzcbuf)(vnode_t *, xuio_t *, cred_t *, \
930 caller_context_t *)
931 /* NB: No ";" */
932
933 /*
934 * Operations on vnodes. Note: File systems must never operate directly
935 * on a 'vnodeops' structure -- it WILL change in future releases! They
936 * must use vn_make_ops() to create the structure.
937 */
938 typedef struct vnodeops {
939 const char *vnop_name;
940 VNODE_OPS; /* Signatures of all vnode operations (vops) */
941 } vnodeops_t;
942
943 typedef int (*fs_generic_func_p) (); /* Generic vop/vfsop/femop/fsemop ptr */
944
945 extern int fop_open(vnode_t **, int, cred_t *, caller_context_t *);
946 extern int fop_close(vnode_t *, int, int, offset_t, cred_t *,
947 caller_context_t *);
948 extern int fop_read(vnode_t *, uio_t *, int, cred_t *, caller_context_t *);
949 extern int fop_write(vnode_t *, uio_t *, int, cred_t *,
950 caller_context_t *);
951 extern int fop_ioctl(vnode_t *, int, intptr_t, int, cred_t *, int *,
952 caller_context_t *);
953 extern int fop_setfl(vnode_t *, int, int, cred_t *, caller_context_t *);
954 extern int fop_getattr(vnode_t *, vattr_t *, int, cred_t *,
955 caller_context_t *);
956 extern int fop_setattr(vnode_t *, vattr_t *, int, cred_t *,
957 caller_context_t *);
958 extern int fop_access(vnode_t *, int, int, cred_t *, caller_context_t *);
959 extern int fop_lookup(vnode_t *, char *, vnode_t **, struct pathname *,
960 int, vnode_t *, cred_t *, caller_context_t *,
961 int *, struct pathname *);
962 extern int fop_create(vnode_t *, char *, vattr_t *, vcexcl_t, int,
963 vnode_t **, cred_t *, int, caller_context_t *,
964 vsecattr_t *);
965 extern int fop_remove(vnode_t *vp, char *, cred_t *, caller_context_t *,
966 int);
967 extern int fop_link(vnode_t *, vnode_t *, char *, cred_t *,
968 caller_context_t *, int);
969 extern int fop_rename(vnode_t *, char *, vnode_t *, char *, cred_t *,
970 caller_context_t *, int);
971 extern int fop_mkdir(vnode_t *, char *, vattr_t *, vnode_t **, cred_t *,
972 caller_context_t *, int, vsecattr_t *);
973 extern int fop_rmdir(vnode_t *, char *, vnode_t *, cred_t *,
974 caller_context_t *, int);
975 extern int fop_readdir(vnode_t *, uio_t *, cred_t *, int *,
976 caller_context_t *, int);
977 extern int fop_symlink(vnode_t *, char *, vattr_t *, char *, cred_t *,
978 caller_context_t *, int);
979 extern int fop_readlink(vnode_t *, uio_t *, cred_t *, caller_context_t *);
980 extern int fop_fsync(vnode_t *, int, cred_t *, caller_context_t *);
981 extern void fop_inactive(vnode_t *, cred_t *, caller_context_t *);
982 extern int fop_fid(vnode_t *, struct fid *, caller_context_t *);
983 extern int fop_rwlock(vnode_t *, int, caller_context_t *);
984 extern void fop_rwunlock(vnode_t *, int, caller_context_t *);
985 extern int fop_seek(vnode_t *, offset_t, offset_t *, caller_context_t *);
986 extern int fop_cmp(vnode_t *, vnode_t *, caller_context_t *);
987 extern int fop_frlock(vnode_t *, int, struct flock64 *, int, offset_t,
988 struct flk_callback *, cred_t *,
989 caller_context_t *);
990 extern int fop_space(vnode_t *, int, struct flock64 *, int, offset_t,
991 cred_t *, caller_context_t *);
992 extern int fop_realvp(vnode_t *, vnode_t **, caller_context_t *);
993 extern int fop_getpage(vnode_t *, offset_t, size_t, uint_t *,
994 struct page **, size_t, struct seg *,
995 caddr_t, enum seg_rw, cred_t *,
996 caller_context_t *);
997 extern int fop_putpage(vnode_t *, offset_t, size_t, int, cred_t *,
998 caller_context_t *);
999 extern int fop_map(vnode_t *, offset_t, struct as *, caddr_t *, size_t,
1000 uchar_t, uchar_t, uint_t, cred_t *cr,
1001 caller_context_t *);
1002 extern int fop_addmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
1003 uchar_t, uchar_t, uint_t, cred_t *,
1004 caller_context_t *);
1005 extern int fop_delmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
1006 uint_t, uint_t, uint_t, cred_t *,
1007 caller_context_t *);
1008 extern int fop_poll(vnode_t *, short, int, short *, struct pollhead **,
1009 caller_context_t *);
1010 extern int fop_dump(vnode_t *, caddr_t, offset_t, offset_t,
1011 caller_context_t *);
1012 extern int fop_pathconf(vnode_t *, int, ulong_t *, cred_t *,
1013 caller_context_t *);
1014 extern int fop_pageio(vnode_t *, struct page *, u_offset_t, size_t, int,
1015 cred_t *, caller_context_t *);
1016 extern int fop_dumpctl(vnode_t *, int, offset_t *, caller_context_t *);
1017 extern void fop_dispose(vnode_t *, struct page *, int, int, cred_t *,
1018 caller_context_t *);
1019 extern int fop_setsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
1020 caller_context_t *);
1021 extern int fop_getsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
1022 caller_context_t *);
1023 extern int fop_shrlock(vnode_t *, int, struct shrlock *, int, cred_t *,
1024 caller_context_t *);
1025 extern int fop_vnevent(vnode_t *, vnevent_t, vnode_t *, char *,
1026 caller_context_t *);
1027 extern int fop_reqzcbuf(vnode_t *, enum uio_rw, xuio_t *, cred_t *,
1028 caller_context_t *);
1029 extern int fop_retzcbuf(vnode_t *, xuio_t *, cred_t *, caller_context_t *);
1030
1031 #endif /* _KERNEL */
1032
1033 #define VOP_OPEN(vpp, mode, cr, ct) \
1034 fop_open(vpp, mode, cr, ct)
1035 #define VOP_CLOSE(vp, f, c, o, cr, ct) \
1036 fop_close(vp, f, c, o, cr, ct)
1037 #define VOP_READ(vp, uiop, iof, cr, ct) \
1038 fop_read(vp, uiop, iof, cr, ct)
1039 #define VOP_WRITE(vp, uiop, iof, cr, ct) \
1040 fop_write(vp, uiop, iof, cr, ct)
1041 #define VOP_IOCTL(vp, cmd, a, f, cr, rvp, ct) \
1042 fop_ioctl(vp, cmd, a, f, cr, rvp, ct)
1043 #define VOP_SETFL(vp, f, a, cr, ct) \
1044 fop_setfl(vp, f, a, cr, ct)
1045 #define VOP_GETATTR(vp, vap, f, cr, ct) \
1046 fop_getattr(vp, vap, f, cr, ct)
1047 #define VOP_SETATTR(vp, vap, f, cr, ct) \
1048 fop_setattr(vp, vap, f, cr, ct)
1049 #define VOP_ACCESS(vp, mode, f, cr, ct) \
1050 fop_access(vp, mode, f, cr, ct)
1051 #define VOP_LOOKUP(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp) \
1052 fop_lookup(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp)
1053 #define VOP_CREATE(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap) \
1054 fop_create(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap)
1055 #define VOP_REMOVE(dvp, p, cr, ct, f) \
1056 fop_remove(dvp, p, cr, ct, f)
1057 #define VOP_LINK(tdvp, fvp, p, cr, ct, f) \
1058 fop_link(tdvp, fvp, p, cr, ct, f)
1059 #define VOP_RENAME(fvp, fnm, tdvp, tnm, cr, ct, f) \
1060 fop_rename(fvp, fnm, tdvp, tnm, cr, ct, f)
1061 #define VOP_MKDIR(dp, p, vap, vpp, cr, ct, f, vsap) \
1062 fop_mkdir(dp, p, vap, vpp, cr, ct, f, vsap)
1063 #define VOP_RMDIR(dp, p, cdir, cr, ct, f) \
1064 fop_rmdir(dp, p, cdir, cr, ct, f)
1065 #define VOP_READDIR(vp, uiop, cr, eofp, ct, f) \
1066 fop_readdir(vp, uiop, cr, eofp, ct, f)
1067 #define VOP_SYMLINK(dvp, lnm, vap, tnm, cr, ct, f) \
1068 fop_symlink(dvp, lnm, vap, tnm, cr, ct, f)
1069 #define VOP_READLINK(vp, uiop, cr, ct) \
1070 fop_readlink(vp, uiop, cr, ct)
1071 #define VOP_FSYNC(vp, syncflag, cr, ct) \
1072 fop_fsync(vp, syncflag, cr, ct)
1073 #define VOP_INACTIVE(vp, cr, ct) \
1074 fop_inactive(vp, cr, ct)
1075 #define VOP_FID(vp, fidp, ct) \
1076 fop_fid(vp, fidp, ct)
1077 #define VOP_RWLOCK(vp, w, ct) \
1078 fop_rwlock(vp, w, ct)
1079 #define VOP_RWUNLOCK(vp, w, ct) \
1080 fop_rwunlock(vp, w, ct)
1081 #define VOP_SEEK(vp, ooff, noffp, ct) \
1082 fop_seek(vp, ooff, noffp, ct)
1083 #define VOP_CMP(vp1, vp2, ct) \
1084 fop_cmp(vp1, vp2, ct)
1085 #define VOP_FRLOCK(vp, cmd, a, f, o, cb, cr, ct) \
1086 fop_frlock(vp, cmd, a, f, o, cb, cr, ct)
1087 #define VOP_SPACE(vp, cmd, a, f, o, cr, ct) \
1088 fop_space(vp, cmd, a, f, o, cr, ct)
1089 #define VOP_REALVP(vp1, vp2, ct) \
1090 fop_realvp(vp1, vp2, ct)
1091 #define VOP_GETPAGE(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct) \
1092 fop_getpage(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct)
1093 #define VOP_PUTPAGE(vp, of, sz, fl, cr, ct) \
1094 fop_putpage(vp, of, sz, fl, cr, ct)
1095 #define VOP_MAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1096 fop_map(vp, of, as, a, sz, p, mp, fl, cr, ct)
1097 #define VOP_ADDMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1098 fop_addmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1099 #define VOP_DELMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1100 fop_delmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1101 #define VOP_POLL(vp, events, anyyet, reventsp, phpp, ct) \
1102 fop_poll(vp, events, anyyet, reventsp, phpp, ct)
1103 #define VOP_DUMP(vp, addr, bn, count, ct) \
1104 fop_dump(vp, addr, bn, count, ct)
1105 #define VOP_PATHCONF(vp, cmd, valp, cr, ct) \
1106 fop_pathconf(vp, cmd, valp, cr, ct)
1107 #define VOP_PAGEIO(vp, pp, io_off, io_len, flags, cr, ct) \
1108 fop_pageio(vp, pp, io_off, io_len, flags, cr, ct)
1109 #define VOP_DUMPCTL(vp, action, blkp, ct) \
1110 fop_dumpctl(vp, action, blkp, ct)
1111 #define VOP_DISPOSE(vp, pp, flag, dn, cr, ct) \
1112 fop_dispose(vp, pp, flag, dn, cr, ct)
1113 #define VOP_GETSECATTR(vp, vsap, f, cr, ct) \
1114 fop_getsecattr(vp, vsap, f, cr, ct)
1115 #define VOP_SETSECATTR(vp, vsap, f, cr, ct) \
1116 fop_setsecattr(vp, vsap, f, cr, ct)
1117 #define VOP_SHRLOCK(vp, cmd, shr, f, cr, ct) \
1118 fop_shrlock(vp, cmd, shr, f, cr, ct)
1119 #define VOP_VNEVENT(vp, vnevent, dvp, fnm, ct) \
1120 fop_vnevent(vp, vnevent, dvp, fnm, ct)
1121 #define VOP_REQZCBUF(vp, rwflag, xuiop, cr, ct) \
1122 fop_reqzcbuf(vp, rwflag, xuiop, cr, ct)
1123 #define VOP_RETZCBUF(vp, xuiop, cr, ct) \
1124 fop_retzcbuf(vp, xuiop, cr, ct)
1125
1126 #define VOPNAME_OPEN "open"
1127 #define VOPNAME_CLOSE "close"
1128 #define VOPNAME_READ "read"
1129 #define VOPNAME_WRITE "write"
1130 #define VOPNAME_IOCTL "ioctl"
1131 #define VOPNAME_SETFL "setfl"
1132 #define VOPNAME_GETATTR "getattr"
1133 #define VOPNAME_SETATTR "setattr"
1134 #define VOPNAME_ACCESS "access"
1135 #define VOPNAME_LOOKUP "lookup"
1136 #define VOPNAME_CREATE "create"
1137 #define VOPNAME_REMOVE "remove"
1138 #define VOPNAME_LINK "link"
1139 #define VOPNAME_RENAME "rename"
1140 #define VOPNAME_MKDIR "mkdir"
1141 #define VOPNAME_RMDIR "rmdir"
1142 #define VOPNAME_READDIR "readdir"
1143 #define VOPNAME_SYMLINK "symlink"
1144 #define VOPNAME_READLINK "readlink"
1145 #define VOPNAME_FSYNC "fsync"
1146 #define VOPNAME_INACTIVE "inactive"
1147 #define VOPNAME_FID "fid"
1148 #define VOPNAME_RWLOCK "rwlock"
1149 #define VOPNAME_RWUNLOCK "rwunlock"
1150 #define VOPNAME_SEEK "seek"
1151 #define VOPNAME_CMP "cmp"
1152 #define VOPNAME_FRLOCK "frlock"
1153 #define VOPNAME_SPACE "space"
1154 #define VOPNAME_REALVP "realvp"
1155 #define VOPNAME_GETPAGE "getpage"
1156 #define VOPNAME_PUTPAGE "putpage"
1157 #define VOPNAME_MAP "map"
1158 #define VOPNAME_ADDMAP "addmap"
1159 #define VOPNAME_DELMAP "delmap"
1160 #define VOPNAME_POLL "poll"
1161 #define VOPNAME_DUMP "dump"
1162 #define VOPNAME_PATHCONF "pathconf"
1163 #define VOPNAME_PAGEIO "pageio"
1164 #define VOPNAME_DUMPCTL "dumpctl"
1165 #define VOPNAME_DISPOSE "dispose"
1166 #define VOPNAME_GETSECATTR "getsecattr"
1167 #define VOPNAME_SETSECATTR "setsecattr"
1168 #define VOPNAME_SHRLOCK "shrlock"
1169 #define VOPNAME_VNEVENT "vnevent"
1170 #define VOPNAME_REQZCBUF "reqzcbuf"
1171 #define VOPNAME_RETZCBUF "retzcbuf"
1172
1173 /*
1174 * Flags for VOP_LOOKUP
1175 *
1176 * Defined in file.h, but also possible, FIGNORECASE and FSEARCH
1177 *
1178 */
1179 #define LOOKUP_DIR 0x01 /* want parent dir vp */
1180 #define LOOKUP_XATTR 0x02 /* lookup up extended attr dir */
1181 #define CREATE_XATTR_DIR 0x04 /* Create extended attr dir */
1182 #define LOOKUP_HAVE_SYSATTR_DIR 0x08 /* Already created virtual GFS dir */
1183
1184 /*
1185 * Flags for VOP_READDIR
1186 */
1187 #define V_RDDIR_ENTFLAGS 0x01 /* request dirent flags */
1188 #define V_RDDIR_ACCFILTER 0x02 /* filter out inaccessible dirents */
1189
1190 /*
1191 * Flags for VOP_RWLOCK/VOP_RWUNLOCK
1192 * VOP_RWLOCK will return the flag that was actually set, or -1 if none.
1193 */
1194 #define V_WRITELOCK_TRUE (1) /* Request write-lock on the vnode */
1195 #define V_WRITELOCK_FALSE (0) /* Request read-lock on the vnode */
1196
1197 /*
1198 * Flags for VOP_DUMPCTL
1199 */
1200 #define DUMP_ALLOC 0
1201 #define DUMP_FREE 1
1202 #define DUMP_SCAN 2
1203
1204 /*
1205 * Public vnode manipulation functions.
1206 */
1207 #if defined(_KERNEL) || defined(_FAKE_KERNEL)
1208
1209 vnode_t *vn_alloc(int);
1210 void vn_reinit(vnode_t *);
1211 void vn_recycle(vnode_t *);
1212 void vn_free(vnode_t *);
1213
1214 int vn_is_readonly(vnode_t *);
1215 int vn_is_opened(vnode_t *, v_mode_t);
1216 int vn_is_mapped(vnode_t *, v_mode_t);
1217 int vn_has_other_opens(vnode_t *, v_mode_t);
1218 void vn_open_upgrade(vnode_t *, int);
1219 void vn_open_downgrade(vnode_t *, int);
1220
1221 int vn_can_change_zones(vnode_t *vp);
1222
1223 int vn_has_flocks(vnode_t *);
1224 int vn_has_mandatory_locks(vnode_t *, int);
1225 int vn_has_cached_data(vnode_t *);
1226
1227 void vn_setops(vnode_t *, vnodeops_t *);
1228 vnodeops_t *vn_getops(vnode_t *);
1229 int vn_matchops(vnode_t *, vnodeops_t *);
1230 int vn_matchopval(vnode_t *, char *, fs_generic_func_p);
1231 int vn_ismntpt(vnode_t *);
1232
1233 struct vfs *vn_mountedvfs(vnode_t *);
1234
1235 int vn_in_dnlc(vnode_t *);
1236
1237 void vn_create_cache(void);
1238 void vn_destroy_cache(void);
1239
1240 void vn_freevnodeops(vnodeops_t *);
1241
1242 int vn_open(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1243 struct vnode **vpp, enum create crwhy, mode_t umask);
1244 int vn_openat(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1245 struct vnode **vpp, enum create crwhy,
1246 mode_t umask, struct vnode *startvp, int fd);
1247 int vn_create(char *pnamep, enum uio_seg seg, struct vattr *vap,
1248 enum vcexcl excl, int mode, struct vnode **vpp,
1249 enum create why, int flag, mode_t umask);
1250 int vn_createat(char *pnamep, enum uio_seg seg, struct vattr *vap,
1251 enum vcexcl excl, int mode, struct vnode **vpp,
1252 enum create why, int flag, mode_t umask, struct vnode *startvp);
1253 int vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base, ssize_t len,
1254 offset_t offset, enum uio_seg seg, int ioflag, rlim64_t ulimit,
1255 cred_t *cr, ssize_t *residp);
1256
1257 void vn_hold(struct vnode *vp);
1258 void vn_rele(struct vnode *vp);
1259 void vn_rele_async(struct vnode *vp, struct taskq *taskq);
1260 void vn_rele_dnlc(struct vnode *vp);
1261 void vn_rele_stream(struct vnode *vp);
1262 int vn_link(char *from, char *to, enum uio_seg seg);
1263 int vn_linkat(vnode_t *fstartvp, char *from, enum symfollow follow,
1264 vnode_t *tstartvp, char *to, enum uio_seg seg);
1265 int vn_rename(char *from, char *to, enum uio_seg seg);
1266 int vn_renameat(vnode_t *fdvp, char *fname, vnode_t *tdvp, char *tname,
1267 enum uio_seg seg);
1268 int vn_remove(char *fnamep, enum uio_seg seg, enum rm dirflag);
1269 int vn_removeat(vnode_t *startvp, char *fnamep, enum uio_seg seg,
1270 enum rm dirflag);
1271 int vn_compare(vnode_t *vp1, vnode_t *vp2);
1272 int vn_vfswlock(struct vnode *vp);
1273 int vn_vfswlock_wait(struct vnode *vp);
1274 int vn_vfsrlock(struct vnode *vp);
1275 int vn_vfsrlock_wait(struct vnode *vp);
1276 void vn_vfsunlock(struct vnode *vp);
1277 int vn_vfswlock_held(struct vnode *vp);
1278 vnode_t *specvp(struct vnode *vp, dev_t dev, vtype_t type, struct cred *cr);
1279 vnode_t *makespecvp(dev_t dev, vtype_t type);
1280
1281 #if defined(_KERNEL)
1282 vn_vfslocks_entry_t *vn_vfslocks_getlock(void *);
1283 void vn_vfslocks_rele(vn_vfslocks_entry_t *);
1284 #endif
1285
1286 boolean_t vn_is_reparse(vnode_t *, cred_t *, caller_context_t *);
1287
1288 void vn_copypath(struct vnode *src, struct vnode *dst);
1289 void vn_setpath_str(struct vnode *vp, const char *str, size_t len);
1290 void vn_setpath(vnode_t *rootvp, struct vnode *startvp, struct vnode *vp,
1291 const char *path, size_t plen);
1292 void vn_renamepath(vnode_t *dvp, vnode_t *vp, const char *nm, size_t len);
1293
1294 /* Vnode event notification */
1295 void vnevent_rename_src(vnode_t *, vnode_t *, char *, caller_context_t *);
1296 void vnevent_rename_dest(vnode_t *, vnode_t *, char *, caller_context_t *);
1297 void vnevent_remove(vnode_t *, vnode_t *, char *, caller_context_t *);
1298 void vnevent_rmdir(vnode_t *, vnode_t *, char *, caller_context_t *);
1299 void vnevent_create(vnode_t *, caller_context_t *);
1300 void vnevent_link(vnode_t *, caller_context_t *);
1301 void vnevent_rename_dest_dir(vnode_t *, caller_context_t *ct);
1302 void vnevent_mountedover(vnode_t *, caller_context_t *);
1303 void vnevent_truncate(vnode_t *, caller_context_t *);
1304 int vnevent_support(vnode_t *, caller_context_t *);
1305
1306 /* Vnode specific data */
1307 void vsd_create(uint_t *, void (*)(void *));
1308 void vsd_destroy(uint_t *);
1309 void *vsd_get(vnode_t *, uint_t);
1310 int vsd_set(vnode_t *, uint_t, void *);
1311 void vsd_free(vnode_t *);
1312
1313 /*
1314 * Extensible vnode attribute (xva) routines:
1315 * xva_init() initializes an xvattr_t (zero struct, init mapsize, set AT_XATTR)
1316 * xva_getxoptattr() returns a ponter to the xoptattr_t section of xvattr_t
1317 */
1318 void xva_init(xvattr_t *);
1319 xoptattr_t *xva_getxoptattr(xvattr_t *); /* Get ptr to xoptattr_t */
1320
1321 void xattr_init(void); /* Initialize vnodeops for xattrs */
1322
1323 /* GFS tunnel for xattrs */
1324 int xattr_dir_lookup(vnode_t *, vnode_t **, int, cred_t *);
1325
1326 /* Reparse Point */
1327 void reparse_point_init(void);
1328
1329 /* Context identification */
1330 u_longlong_t fs_new_caller_id();
1331
1332 int vn_vmpss_usepageio(vnode_t *);
1333
1334 /* Empty v_path placeholder */
1335 extern char *vn_vpath_empty;
1336
1337 /*
1338 * Needed for use of IS_VMODSORT() in kernel.
1339 */
1340 extern uint_t pvn_vmodsort_supported;
1341
1342 #define VN_HOLD(vp) { \
1343 vn_hold(vp); \
1344 }
1345
1346 #define VN_RELE(vp) { \
1347 vn_rele(vp); \
1348 }
1349
1350 #define VN_RELE_ASYNC(vp, taskq) { \
1351 vn_rele_async(vp, taskq); \
1352 }
1353
1354 #define VN_RELE_LOCKED(vp) { \
1355 (vp)->v_count--; \
1356 }
1357
1358 #define VN_SET_VFS_TYPE_DEV(vp, vfsp, type, dev) { \
1359 (vp)->v_vfsp = (vfsp); \
1360 (vp)->v_type = (type); \
1361 (vp)->v_rdev = (dev); \
1362 }
1363
1364 /*
1365 * Compare two vnodes for equality. In general this macro should be used
1366 * in preference to calling VOP_CMP directly.
1367 */
1368 #if defined(_FAKE_KERNEL)
1369 #define VN_CMP(VP1, VP2) \
1370 (((VP1) == (VP2)) ? 1 : VOP_CMP(VP1, VP2, NULL))
1371 #else
1372 #define VN_CMP(VP1, VP2) ((VP1) == (VP2) ? 1 : \
1373 ((VP1) && (VP2) && (vn_getops(VP1) == vn_getops(VP2)) ? \
1374 VOP_CMP(VP1, VP2, NULL) : 0))
1375 #endif
1376
1377 /*
1378 * Some well-known global vnodes used by the VM system to name pages.
1379 */
1380 extern struct vnode kvps[];
1381
1382 typedef enum {
1383 KV_KVP, /* vnode for all segkmem pages */
1384 KV_ZVP, /* vnode for all ZFS pages */
1385 #if defined(__sparc)
1386 KV_MPVP, /* vnode for all page_t meta-pages */
1387 KV_PROMVP, /* vnode for all PROM pages */
1388 #endif /* __sparc */
1389 KV_MAX /* total number of vnodes in kvps[] */
1390 } kvps_index_t;
1391
1392 #define VN_ISKAS(vp) ((vp) >= &kvps[0] && (vp) < &kvps[KV_MAX])
1393
1394 #endif /* _KERNEL */
1395
1396 /*
1397 * Flags to VOP_SETATTR/VOP_GETATTR.
1398 */
1399 #define ATTR_UTIME 0x01 /* non-default utime(2) request */
1400 #define ATTR_EXEC 0x02 /* invocation from exec(2) */
1401 #define ATTR_COMM 0x04 /* yield common vp attributes */
1402 #define ATTR_HINT 0x08 /* information returned will be `hint' */
1403 #define ATTR_REAL 0x10 /* yield attributes of the real vp */
1404 #define ATTR_NOACLCHECK 0x20 /* Don't check ACL when checking permissions */
1405 #define ATTR_TRIGGER 0x40 /* Mount first if vnode is a trigger mount */
1406 /*
1407 * Generally useful macros.
1408 */
1409 #define VBSIZE(vp) ((vp)->v_vfsp->vfs_bsize)
1410
1411 #define VTOZONE(vp) ((vp)->v_vfsp->vfs_zone)
1412
1413 #define NULLVP ((struct vnode *)0)
1414 #define NULLVPP ((struct vnode **)0)
1415
1416 #ifdef _KERNEL
1417
1418 /*
1419 * Structure used while handling asynchronous VOP_PUTPAGE operations.
1420 */
1421 struct async_reqs {
1422 struct async_reqs *a_next; /* pointer to next arg struct */
1423 struct vnode *a_vp; /* vnode pointer */
1424 u_offset_t a_off; /* offset in file */
1425 uint_t a_len; /* size of i/o request */
1426 int a_flags; /* flags to indicate operation type */
1427 struct cred *a_cred; /* cred pointer */
1428 ushort_t a_prealloced; /* set if struct is pre-allocated */
1429 };
1430
1431 /*
1432 * VN_DISPOSE() -- given a page pointer, safely invoke VOP_DISPOSE().
1433 * Note that there is no guarantee that the page passed in will be
1434 * freed. If that is required, then a check after calling VN_DISPOSE would
1435 * be necessary to ensure the page was freed.
1436 */
1437 #define VN_DISPOSE(pp, flag, dn, cr) { \
1438 if ((pp)->p_vnode != NULL && !VN_ISKAS((pp)->p_vnode)) \
1439 VOP_DISPOSE((pp)->p_vnode, (pp), (flag), (dn), (cr), NULL); \
1440 else if ((flag) == B_FREE) \
1441 page_free((pp), (dn)); \
1442 else \
1443 page_destroy((pp), (dn)); \
1444 }
1445
1446 #endif /* _KERNEL */
1447
1448 #ifdef __cplusplus
1449 }
1450 #endif
1451
1452 #endif /* _SYS_VNODE_H */