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