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) 2017, Joyent, Inc.
  25  * Copyright (c) 2011, 2017 by Delphix. All rights reserved.
  26  * Copyright 2017 RackTop Systems.
  27  */
  28 
  29 /*      Copyright (c) 1983, 1984, 1985, 1986, 1987, 1988, 1989 AT&T */
  30 /*        All Rights Reserved   */
  31 
  32 /*
  33  * University Copyright- Copyright (c) 1982, 1986, 1988
  34  * The Regents of the University of California
  35  * All Rights Reserved
  36  *
  37  * University Acknowledgment- Portions of this document are derived from
  38  * software developed by the University of California, Berkeley, and its
  39  * contributors.
  40  */
  41 
  42 #ifndef _SYS_VNODE_H
  43 #define _SYS_VNODE_H
  44 
  45 #include <sys/types.h>
  46 #include <sys/t_lock.h>
  47 #include <sys/rwstlock.h>
  48 #include <sys/time_impl.h>
  49 #include <sys/cred.h>
  50 #include <sys/uio.h>
  51 #include <sys/resource.h>
  52 #include <vm/seg_enum.h>
  53 #include <sys/kstat.h>
  54 #include <sys/kmem.h>
  55 #include <sys/list.h>
  56 #ifdef  _KERNEL
  57 #include <sys/buf.h>
  58 #include <sys/sdt.h>
  59 #endif  /* _KERNEL */
  60 
  61 #ifdef  __cplusplus
  62 extern "C" {
  63 #endif
  64 
  65 /*
  66  * Statistics for all vnode operations.
  67  * All operations record number of ops (since boot/mount/zero'ed).
  68  * Certain I/O operations (read, write, readdir) also record number
  69  * of bytes transferred.
  70  * This appears in two places in the system: one is embedded in each
  71  * vfs_t.  There is also an array of vopstats_t structures allocated
  72  * on a per-fstype basis.
  73  */
  74 
  75 #define VOPSTATS_STR    "vopstats_"     /* Initial string for vopstat kstats */
  76 
  77 typedef struct vopstats {
  78         kstat_named_t   nopen;          /* VOP_OPEN */
  79         kstat_named_t   nclose;         /* VOP_CLOSE */
  80         kstat_named_t   nread;          /* VOP_READ */
  81         kstat_named_t   read_bytes;
  82         kstat_named_t   nwrite;         /* VOP_WRITE */
  83         kstat_named_t   write_bytes;
  84         kstat_named_t   nioctl;         /* VOP_IOCTL */
  85         kstat_named_t   nsetfl;         /* VOP_SETFL */
  86         kstat_named_t   ngetattr;       /* VOP_GETATTR */
  87         kstat_named_t   nsetattr;       /* VOP_SETATTR */
  88         kstat_named_t   naccess;        /* VOP_ACCESS */
  89         kstat_named_t   nlookup;        /* VOP_LOOKUP */
  90         kstat_named_t   ncreate;        /* VOP_CREATE */
  91         kstat_named_t   nremove;        /* VOP_REMOVE */
  92         kstat_named_t   nlink;          /* VOP_LINK */
  93         kstat_named_t   nrename;        /* VOP_RENAME */
  94         kstat_named_t   nmkdir;         /* VOP_MKDIR */
  95         kstat_named_t   nrmdir;         /* VOP_RMDIR */
  96         kstat_named_t   nreaddir;       /* VOP_READDIR */
  97         kstat_named_t   readdir_bytes;
  98         kstat_named_t   nsymlink;       /* VOP_SYMLINK */
  99         kstat_named_t   nreadlink;      /* VOP_READLINK */
 100         kstat_named_t   nfsync;         /* VOP_FSYNC */
 101         kstat_named_t   ninactive;      /* VOP_INACTIVE */
 102         kstat_named_t   nfid;           /* VOP_FID */
 103         kstat_named_t   nrwlock;        /* VOP_RWLOCK */
 104         kstat_named_t   nrwunlock;      /* VOP_RWUNLOCK */
 105         kstat_named_t   nseek;          /* VOP_SEEK */
 106         kstat_named_t   ncmp;           /* VOP_CMP */
 107         kstat_named_t   nfrlock;        /* VOP_FRLOCK */
 108         kstat_named_t   nspace;         /* VOP_SPACE */
 109         kstat_named_t   nrealvp;        /* VOP_REALVP */
 110         kstat_named_t   ngetpage;       /* VOP_GETPAGE */
 111         kstat_named_t   nputpage;       /* VOP_PUTPAGE */
 112         kstat_named_t   nmap;           /* VOP_MAP */
 113         kstat_named_t   naddmap;        /* VOP_ADDMAP */
 114         kstat_named_t   ndelmap;        /* VOP_DELMAP */
 115         kstat_named_t   npoll;          /* VOP_POLL */
 116         kstat_named_t   ndump;          /* VOP_DUMP */
 117         kstat_named_t   npathconf;      /* VOP_PATHCONF */
 118         kstat_named_t   npageio;        /* VOP_PAGEIO */
 119         kstat_named_t   ndumpctl;       /* VOP_DUMPCTL */
 120         kstat_named_t   ndispose;       /* VOP_DISPOSE */
 121         kstat_named_t   nsetsecattr;    /* VOP_SETSECATTR */
 122         kstat_named_t   ngetsecattr;    /* VOP_GETSECATTR */
 123         kstat_named_t   nshrlock;       /* VOP_SHRLOCK */
 124         kstat_named_t   nvnevent;       /* VOP_VNEVENT */
 125         kstat_named_t   nreqzcbuf;      /* VOP_REQZCBUF */
 126         kstat_named_t   nretzcbuf;      /* VOP_RETZCBUF */
 127 } vopstats_t;
 128 
 129 /*
 130  * The vnode is the focus of all file activity in UNIX.
 131  * A vnode is allocated for each active file, each current
 132  * directory, each mounted-on file, and the root.
 133  *
 134  * Each vnode is usually associated with a file-system-specific node (for
 135  * UFS, this is the in-memory inode).  Generally, a vnode and an fs-node
 136  * should be created and destroyed together as a pair.
 137  *
 138  * If a vnode is reused for a new file, it should be reinitialized by calling
 139  * either vn_reinit() or vn_recycle().
 140  *
 141  * vn_reinit() resets the entire vnode as if it was returned by vn_alloc().
 142  * The caller is responsible for setting up the entire vnode after calling
 143  * vn_reinit().  This is important when using kmem caching where the vnode is
 144  * allocated by a constructor, for instance.
 145  *
 146  * vn_recycle() is used when the file system keeps some state around in both
 147  * the vnode and the associated FS-node.  In UFS, for example, the inode of
 148  * a deleted file can be reused immediately.  The v_data, v_vfsp, v_op, etc.
 149  * remains the same but certain fields related to the previous instance need
 150  * to be reset.  In particular:
 151  *      v_femhead
 152  *      v_path
 153  *      v_rdcnt, v_wrcnt
 154  *      v_mmap_read, v_mmap_write
 155  */
 156 
 157 /*
 158  * vnode types.  VNON means no type.  These values are unrelated to
 159  * values in on-disk inodes.
 160  */
 161 typedef enum vtype {
 162         VNON    = 0,
 163         VREG    = 1,
 164         VDIR    = 2,
 165         VBLK    = 3,
 166         VCHR    = 4,
 167         VLNK    = 5,
 168         VFIFO   = 6,
 169         VDOOR   = 7,
 170         VPROC   = 8,
 171         VSOCK   = 9,
 172         VPORT   = 10,
 173         VBAD    = 11
 174 } vtype_t;
 175 
 176 /*
 177  * VSD - Vnode Specific Data
 178  * Used to associate additional private data with a vnode.
 179  */
 180 struct vsd_node {
 181         list_node_t vs_nodes;           /* list of all VSD nodes */
 182         uint_t vs_nkeys;                /* entries in value array */
 183         void **vs_value;                /* array of value/key */
 184 };
 185 
 186 /*
 187  * Many of the fields in the vnode are read-only once they are initialized
 188  * at vnode creation time.  Other fields are protected by locks.
 189  *
 190  * IMPORTANT: vnodes should be created ONLY by calls to vn_alloc().  They
 191  * may not be embedded into the file-system specific node (inode).  The
 192  * size of vnodes may change.
 193  *
 194  * The v_lock protects:
 195  *   v_flag
 196  *   v_stream
 197  *   v_count
 198  *   v_shrlocks
 199  *   v_path
 200  *   v_vsd
 201  *   v_xattrdir
 202  *
 203  * A special lock (implemented by vn_vfswlock in vnode.c) protects:
 204  *   v_vfsmountedhere
 205  *
 206  * The global flock_lock mutex (in flock.c) protects:
 207  *   v_filocks
 208  *
 209  * IMPORTANT NOTE:
 210  *
 211  *   The following vnode fields are considered public and may safely be
 212  *   accessed by file systems or other consumers:
 213  *
 214  *     v_lock
 215  *     v_flag
 216  *     v_count
 217  *     v_data
 218  *     v_vfsp
 219  *     v_stream
 220  *     v_type
 221  *     v_rdev
 222  *
 223  * ALL OTHER FIELDS SHOULD BE ACCESSED ONLY BY THE OWNER OF THAT FIELD.
 224  * In particular, file systems should not access other fields; they may
 225  * change or even be removed.  The functionality which was once provided
 226  * by these fields is available through vn_* functions.
 227  *
 228  * VNODE PATH THEORY:
 229  * In each vnode, the v_path field holds a cached version of the canonical
 230  * filesystem path which that node represents.  Because vnodes lack contextual
 231  * information about their own name or position in the VFS hierarchy, this path
 232  * must be calculated when the vnode is instantiated by operations such as
 233  * fop_create, fop_lookup, or fop_mkdir.  During said operations, both the
 234  * parent vnode (and its cached v_path) and future name are known, so the
 235  * v_path of the resulting object can easily be set.
 236  *
 237  * The caching nature of v_path is complicated in the face of directory
 238  * renames.  Filesystem drivers are responsible for calling vn_renamepath when
 239  * a fop_rename operation succeeds.  While the v_path on the renamed vnode will
 240  * be updated, existing children of the directory (direct, or at deeper levels)
 241  * will now possess v_path caches which are stale.
 242  *
 243  * It is expensive (and for non-directories, impossible) to recalculate stale
 244  * v_path entries during operations such as vnodetopath.  The best time during
 245  * which to correct such wrongs is the same as when v_path is first
 246  * initialized: during fop_create/fop_lookup/fop_mkdir/etc, where adequate
 247  * context is available to generate the current path.
 248  *
 249  * In order to quickly detect stale v_path entries (without full lookup
 250  * verification) to trigger a v_path update, the v_path_stamp field has been
 251  * added to vnode_t.  As part of successful fop_create/fop_lookup/fop_mkdir
 252  * operations, where the name and parent vnode are available, the following
 253  * rules are used to determine updates to the child:
 254  *
 255  * 1. If the parent lacks a v_path, clear any existing v_path and v_path_stamp
 256  *    on the child.  Until the parent v_path is refreshed to a valid state, the
 257  *    child v_path must be considered invalid too.
 258  *
 259  * 2. If the child lacks a v_path (implying v_path_stamp == 0), it inherits the
 260  *    v_path_stamp value from its parent and its v_path is updated.
 261  *
 262  * 3. If the child v_path_stamp is less than v_path_stamp in the parent, it is
 263  *    an indication that the child v_path is stale.  The v_path is updated and
 264  *    v_path_stamp in the child is set to the current hrtime().
 265  *
 266  *    It does _not_ inherit the parent v_path_stamp in order to propagate the
 267  *    the time of v_path invalidation through the directory structure.  This
 268  *    prevents concurrent invalidations (operating with a now-incorrect v_path)
 269  *    at deeper levels in the tree from persisting.
 270  *
 271  * 4. If the child v_path_stamp is greater or equal to the parent, no action
 272  *    needs to be taken.
 273  *
 274  * Note that fop_rename operations do not follow this ruleset.  They perform an
 275  * explicit update of v_path and v_path_stamp (setting it to the current time)
 276  *
 277  * With these constraints in place, v_path invalidations and updates should
 278  * proceed in a timely manner as vnodes are accessed.  While there still are
 279  * limited cases where vnodetopath operations will fail, the risk is minimized.
 280  */
 281 
 282 struct fem_head;        /* from fem.h */
 283 
 284 typedef struct vnode {
 285         kmutex_t        v_lock;         /* protects vnode fields */
 286         uint_t          v_flag;         /* vnode flags (see below) */
 287         uint_t          v_count;        /* reference count */
 288         void            *v_data;        /* private data for fs */
 289         struct vfs      *v_vfsp;        /* ptr to containing VFS */
 290         struct stdata   *v_stream;      /* associated stream */
 291         enum vtype      v_type;         /* vnode type */
 292         dev_t           v_rdev;         /* device (VCHR, VBLK) */
 293 
 294         /* PRIVATE FIELDS BELOW - DO NOT USE */
 295 
 296         struct vfs      *v_vfsmountedhere; /* ptr to vfs mounted here */
 297         struct vnodeops *v_op;          /* vnode operations */
 298         struct page     *v_pages;       /* vnode pages list */
 299         struct filock   *v_filocks;     /* ptr to filock list */
 300         struct shrlocklist *v_shrlocks; /* ptr to shrlock list */
 301         krwlock_t       v_nbllock;      /* sync for NBMAND locks */
 302         kcondvar_t      v_cv;           /* synchronize locking */
 303         void            *v_locality;    /* hook for locality info */
 304         struct fem_head *v_femhead;     /* fs monitoring */
 305         char            *v_path;        /* cached path */
 306         hrtime_t        v_path_stamp;   /* timestamp for cached path */
 307         uint_t          v_rdcnt;        /* open for read count  (VREG only) */
 308         uint_t          v_wrcnt;        /* open for write count (VREG only) */
 309         u_longlong_t    v_mmap_read;    /* mmap read count */
 310         u_longlong_t    v_mmap_write;   /* mmap write count */
 311         void            *v_mpssdata;    /* info for large page mappings */
 312         void            *v_fopdata;     /* list of file ops event watches */
 313         kmutex_t        v_vsd_lock;     /* protects v_vsd field */
 314         struct vsd_node *v_vsd;         /* vnode specific data */
 315         struct vnode    *v_xattrdir;    /* unnamed extended attr dir (GFS) */
 316         uint_t          v_count_dnlc;   /* dnlc reference count */
 317 } vnode_t;
 318 
 319 #define IS_DEVVP(vp)    \
 320         ((vp)->v_type == VCHR || (vp)->v_type == VBLK || (vp)->v_type == VFIFO)
 321 
 322 #define VNODE_ALIGN     64
 323 /* Count of low-order 0 bits in a vnode *, based on size and alignment. */
 324 #if defined(_LP64)
 325 #define VNODE_ALIGN_LOG2        8
 326 #else
 327 #define VNODE_ALIGN_LOG2        7
 328 #endif
 329 
 330 /*
 331  * vnode flags.
 332  */
 333 #define VROOT           0x01    /* root of its file system */
 334 #define VNOCACHE        0x02    /* don't keep cache pages on vnode */
 335 #define VNOMAP          0x04    /* file cannot be mapped/faulted */
 336 #define VDUP            0x08    /* file should be dup'ed rather then opened */
 337 #define VNOSWAP         0x10    /* file cannot be used as virtual swap device */
 338 #define VNOMOUNT        0x20    /* file cannot be covered by mount */
 339 #define VISSWAP         0x40    /* vnode is being used for swap */
 340 #define VSWAPLIKE       0x80    /* vnode acts like swap (but may not be) */
 341 
 342 #define IS_SWAPVP(vp)   (((vp)->v_flag & (VISSWAP | VSWAPLIKE)) != 0)
 343 
 344 #ifdef _KERNEL
 345 typedef struct vn_vfslocks_entry {
 346         rwstlock_t ve_lock;
 347         void *ve_vpvfs;
 348         struct vn_vfslocks_entry *ve_next;
 349         uint32_t ve_refcnt;
 350         char pad[64 - sizeof (rwstlock_t) - 2 * sizeof (void *) - \
 351             sizeof (uint32_t)];
 352 } vn_vfslocks_entry_t;
 353 #endif
 354 
 355 /*
 356  * The following two flags are used to lock the v_vfsmountedhere field
 357  */
 358 #define VVFSLOCK        0x100
 359 #define VVFSWAIT        0x200
 360 
 361 /*
 362  * Used to serialize VM operations on a vnode
 363  */
 364 #define VVMLOCK         0x400
 365 
 366 /*
 367  * Tell vn_open() not to fail a directory open for writing but
 368  * to go ahead and call VOP_OPEN() to let the filesystem check.
 369  */
 370 #define VDIROPEN        0x800
 371 
 372 /*
 373  * Flag to let the VM system know that this file is most likely a binary
 374  * or shared library since it has been mmap()ed EXEC at some time.
 375  */
 376 #define VVMEXEC         0x1000
 377 
 378 #define VPXFS           0x2000  /* clustering: global fs proxy vnode */
 379 
 380 #define IS_PXFSVP(vp)   ((vp)->v_flag & VPXFS)
 381 
 382 #define V_XATTRDIR      0x4000  /* attribute unnamed directory */
 383 
 384 #define IS_XATTRDIR(vp) ((vp)->v_flag & V_XATTRDIR)
 385 
 386 #define V_LOCALITY      0x8000  /* whether locality aware */
 387 
 388 /*
 389  * Flag that indicates the VM should maintain the v_pages list with all modified
 390  * pages on one end and unmodified pages at the other. This makes finding dirty
 391  * pages to write back to disk much faster at the expense of taking a minor
 392  * fault on the first store instruction which touches a writable page.
 393  */
 394 #define VMODSORT        (0x10000)
 395 #define IS_VMODSORT(vp) \
 396         (pvn_vmodsort_supported != 0 && ((vp)->v_flag  & VMODSORT) != 0)
 397 
 398 #define VISSWAPFS       0x20000 /* vnode is being used for swapfs */
 399 
 400 /*
 401  * The mdb memstat command assumes that IS_SWAPFSVP only uses the
 402  * vnode's v_flag field.  If this changes, cache the additional
 403  * fields in mdb; see vn_get in mdb/common/modules/genunix/memory.c
 404  */
 405 #define IS_SWAPFSVP(vp) (((vp)->v_flag & VISSWAPFS) != 0)
 406 
 407 #define V_SYSATTR       0x40000 /* vnode is a GFS system attribute */
 408 
 409 /*
 410  * Indication that VOP_LOOKUP operations on this vnode may yield results from a
 411  * different VFS instance.  The main use of this is to suppress v_path
 412  * calculation logic when filesystems such as procfs emit results which defy
 413  * expectations about normal VFS behavior.
 414  */
 415 #define VTRAVERSE       0x80000
 416 
 417 /*
 418  * Indicates the vnode is a zone's root ("/" in-zone). A zone's root does NOT
 419  * have to be a filesystem root, even if many zone brands guaranteed that.
 420  */
 421 #define VZONEROOT       0x100000
 422 
 423 
 424 /*
 425  * Vnode attributes.  A bit-mask is supplied as part of the
 426  * structure to indicate the attributes the caller wants to
 427  * set (setattr) or extract (getattr).
 428  */
 429 
 430 /*
 431  * Note that va_nodeid and va_nblocks are 64bit data type.
 432  * We support large files over NFSV3. With Solaris client and
 433  * Server that generates 64bit ino's and sizes these fields
 434  * will overflow if they are 32 bit sizes.
 435  */
 436 
 437 typedef struct vattr {
 438         uint_t          va_mask;        /* bit-mask of attributes */
 439         vtype_t         va_type;        /* vnode type (for create) */
 440         mode_t          va_mode;        /* file access mode */
 441         uid_t           va_uid;         /* owner user id */
 442         gid_t           va_gid;         /* owner group id */
 443         dev_t           va_fsid;        /* file system id (dev for now) */
 444         u_longlong_t    va_nodeid;      /* node id */
 445         nlink_t         va_nlink;       /* number of references to file */
 446         u_offset_t      va_size;        /* file size in bytes */
 447         timestruc_t     va_atime;       /* time of last access */
 448         timestruc_t     va_mtime;       /* time of last modification */
 449         timestruc_t     va_ctime;       /* time of last status change */
 450         dev_t           va_rdev;        /* device the file represents */
 451         uint_t          va_blksize;     /* fundamental block size */
 452         u_longlong_t    va_nblocks;     /* # of blocks allocated */
 453         uint_t          va_seq;         /* sequence number */
 454 } vattr_t;
 455 
 456 #define AV_SCANSTAMP_SZ 32              /* length of anti-virus scanstamp */
 457 
 458 /*
 459  * Structure of all optional attributes.
 460  */
 461 typedef struct xoptattr {
 462         timestruc_t     xoa_createtime; /* Create time of file */
 463         uint8_t         xoa_archive;
 464         uint8_t         xoa_system;
 465         uint8_t         xoa_readonly;
 466         uint8_t         xoa_hidden;
 467         uint8_t         xoa_nounlink;
 468         uint8_t         xoa_immutable;
 469         uint8_t         xoa_appendonly;
 470         uint8_t         xoa_nodump;
 471         uint8_t         xoa_opaque;
 472         uint8_t         xoa_av_quarantined;
 473         uint8_t         xoa_av_modified;
 474         uint8_t         xoa_av_scanstamp[AV_SCANSTAMP_SZ];
 475         uint8_t         xoa_reparse;
 476         uint64_t        xoa_generation;
 477         uint8_t         xoa_offline;
 478         uint8_t         xoa_sparse;
 479         uint8_t         xoa_projinherit;
 480         uint64_t        xoa_projid;
 481 } xoptattr_t;
 482 
 483 /*
 484  * The xvattr structure is really a variable length structure that
 485  * is made up of:
 486  * - The classic vattr_t (xva_vattr)
 487  * - a 32 bit quantity (xva_mapsize) that specifies the size of the
 488  *   attribute bitmaps in 32 bit words.
 489  * - A pointer to the returned attribute bitmap (needed because the
 490  *   previous element, the requested attribute bitmap) is variable lenth.
 491  * - The requested attribute bitmap, which is an array of 32 bit words.
 492  *   Callers use the XVA_SET_REQ() macro to set the bits corresponding to
 493  *   the attributes that are being requested.
 494  * - The returned attribute bitmap, which is an array of 32 bit words.
 495  *   File systems that support optional attributes use the XVA_SET_RTN()
 496  *   macro to set the bits corresponding to the attributes that are being
 497  *   returned.
 498  * - The xoptattr_t structure which contains the attribute values
 499  *
 500  * xva_mapsize determines how many words in the attribute bitmaps.
 501  * Immediately following the attribute bitmaps is the xoptattr_t.
 502  * xva_getxoptattr() is used to get the pointer to the xoptattr_t
 503  * section.
 504  */
 505 
 506 #define XVA_MAPSIZE     3               /* Size of attr bitmaps */
 507 #define XVA_MAGIC       0x78766174      /* Magic # for verification */
 508 
 509 /*
 510  * The xvattr structure is an extensible structure which permits optional
 511  * attributes to be requested/returned.  File systems may or may not support
 512  * optional attributes.  They do so at their own discretion but if they do
 513  * support optional attributes, they must register the VFSFT_XVATTR feature
 514  * so that the optional attributes can be set/retrived.
 515  *
 516  * The fields of the xvattr structure are:
 517  *
 518  * xva_vattr - The first element of an xvattr is a legacy vattr structure
 519  * which includes the common attributes.  If AT_XVATTR is set in the va_mask
 520  * then the entire structure is treated as an xvattr.  If AT_XVATTR is not
 521  * set, then only the xva_vattr structure can be used.
 522  *
 523  * xva_magic - 0x78766174 (hex for "xvat"). Magic number for verification.
 524  *
 525  * xva_mapsize - Size of requested and returned attribute bitmaps.
 526  *
 527  * xva_rtnattrmapp - Pointer to xva_rtnattrmap[].  We need this since the
 528  * size of the array before it, xva_reqattrmap[], could change which means
 529  * the location of xva_rtnattrmap[] could change.  This will allow unbundled
 530  * file systems to find the location of xva_rtnattrmap[] when the sizes change.
 531  *
 532  * xva_reqattrmap[] - Array of requested attributes.  Attributes are
 533  * represented by a specific bit in a specific element of the attribute
 534  * map array.  Callers set the bits corresponding to the attributes
 535  * that the caller wants to get/set.
 536  *
 537  * xva_rtnattrmap[] - Array of attributes that the file system was able to
 538  * process.  Not all file systems support all optional attributes.  This map
 539  * informs the caller which attributes the underlying file system was able
 540  * to set/get.  (Same structure as the requested attributes array in terms
 541  * of each attribute  corresponding to specific bits and array elements.)
 542  *
 543  * xva_xoptattrs - Structure containing values of optional attributes.
 544  * These values are only valid if the corresponding bits in xva_reqattrmap
 545  * are set and the underlying file system supports those attributes.
 546  */
 547 typedef struct xvattr {
 548         vattr_t         xva_vattr;      /* Embedded vattr structure */
 549         uint32_t        xva_magic;      /* Magic Number */
 550         uint32_t        xva_mapsize;    /* Size of attr bitmap (32-bit words) */
 551         uint32_t        *xva_rtnattrmapp;       /* Ptr to xva_rtnattrmap[] */
 552         uint32_t        xva_reqattrmap[XVA_MAPSIZE];    /* Requested attrs */
 553         uint32_t        xva_rtnattrmap[XVA_MAPSIZE];    /* Returned attrs */
 554         xoptattr_t      xva_xoptattrs;  /* Optional attributes */
 555 } xvattr_t;
 556 
 557 #ifdef _SYSCALL32
 558 /*
 559  * For bigtypes time_t changed to 64 bit on the 64-bit kernel.
 560  * Define an old version for user/kernel interface
 561  */
 562 
 563 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
 564 #pragma pack(4)
 565 #endif
 566 
 567 typedef struct vattr32 {
 568         uint32_t        va_mask;        /* bit-mask of attributes */
 569         vtype_t         va_type;        /* vnode type (for create) */
 570         mode32_t        va_mode;        /* file access mode */
 571         uid32_t         va_uid;         /* owner user id */
 572         gid32_t         va_gid;         /* owner group id */
 573         dev32_t         va_fsid;        /* file system id (dev for now) */
 574         u_longlong_t    va_nodeid;      /* node id */
 575         nlink_t         va_nlink;       /* number of references to file */
 576         u_offset_t      va_size;        /* file size in bytes */
 577         timestruc32_t   va_atime;       /* time of last access */
 578         timestruc32_t   va_mtime;       /* time of last modification */
 579         timestruc32_t   va_ctime;       /* time of last status change */
 580         dev32_t         va_rdev;        /* device the file represents */
 581         uint32_t        va_blksize;     /* fundamental block size */
 582         u_longlong_t    va_nblocks;     /* # of blocks allocated */
 583         uint32_t        va_seq;         /* sequence number */
 584 } vattr32_t;
 585 
 586 #if _LONG_LONG_ALIGNMENT == 8 && _LONG_LONG_ALIGNMENT_32 == 4
 587 #pragma pack()
 588 #endif
 589 
 590 #else  /* not _SYSCALL32 */
 591 #define vattr32         vattr
 592 typedef vattr_t         vattr32_t;
 593 #endif /* _SYSCALL32 */
 594 
 595 /*
 596  * Attributes of interest to the caller of setattr or getattr.
 597  */
 598 #define AT_TYPE         0x00001
 599 #define AT_MODE         0x00002
 600 #define AT_UID          0x00004
 601 #define AT_GID          0x00008
 602 #define AT_FSID         0x00010
 603 #define AT_NODEID       0x00020
 604 #define AT_NLINK        0x00040
 605 #define AT_SIZE         0x00080
 606 #define AT_ATIME        0x00100
 607 #define AT_MTIME        0x00200
 608 #define AT_CTIME        0x00400
 609 #define AT_RDEV         0x00800
 610 #define AT_BLKSIZE      0x01000
 611 #define AT_NBLOCKS      0x02000
 612 /*                      0x04000 */      /* unused */
 613 #define AT_SEQ          0x08000
 614 /*
 615  * If AT_XVATTR is set then there are additional bits to process in
 616  * the xvattr_t's attribute bitmap.  If this is not set then the bitmap
 617  * MUST be ignored.  Note that this bit must be set/cleared explicitly.
 618  * That is, setting AT_ALL will NOT set AT_XVATTR.
 619  */
 620 #define AT_XVATTR       0x10000
 621 
 622 #define AT_ALL          (AT_TYPE|AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|\
 623                         AT_NLINK|AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|\
 624                         AT_RDEV|AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
 625 
 626 #define AT_STAT         (AT_MODE|AT_UID|AT_GID|AT_FSID|AT_NODEID|AT_NLINK|\
 627                         AT_SIZE|AT_ATIME|AT_MTIME|AT_CTIME|AT_RDEV|AT_TYPE)
 628 
 629 #define AT_TIMES        (AT_ATIME|AT_MTIME|AT_CTIME)
 630 
 631 #define AT_NOSET        (AT_NLINK|AT_RDEV|AT_FSID|AT_NODEID|AT_TYPE|\
 632                         AT_BLKSIZE|AT_NBLOCKS|AT_SEQ)
 633 
 634 /*
 635  * Attribute bits used in the extensible attribute's (xva's) attribute
 636  * bitmaps.  Note that the bitmaps are made up of a variable length number
 637  * of 32-bit words.  The convention is to use XAT{n}_{attrname} where "n"
 638  * is the element in the bitmap (starting at 1).  This convention is for
 639  * the convenience of the maintainer to keep track of which element each
 640  * attribute belongs to.
 641  *
 642  * NOTE THAT CONSUMERS MUST *NOT* USE THE XATn_* DEFINES DIRECTLY.  CONSUMERS
 643  * MUST USE THE XAT_* DEFINES.
 644  */
 645 #define XAT0_INDEX      0LL             /* Index into bitmap for XAT0 attrs */
 646 #define XAT0_CREATETIME 0x00000001      /* Create time of file */
 647 #define XAT0_ARCHIVE    0x00000002      /* Archive */
 648 #define XAT0_SYSTEM     0x00000004      /* System */
 649 #define XAT0_READONLY   0x00000008      /* Readonly */
 650 #define XAT0_HIDDEN     0x00000010      /* Hidden */
 651 #define XAT0_NOUNLINK   0x00000020      /* Nounlink */
 652 #define XAT0_IMMUTABLE  0x00000040      /* immutable */
 653 #define XAT0_APPENDONLY 0x00000080      /* appendonly */
 654 #define XAT0_NODUMP     0x00000100      /* nodump */
 655 #define XAT0_OPAQUE     0x00000200      /* opaque */
 656 #define XAT0_AV_QUARANTINED     0x00000400      /* anti-virus quarantine */
 657 #define XAT0_AV_MODIFIED        0x00000800      /* anti-virus modified */
 658 #define XAT0_AV_SCANSTAMP       0x00001000      /* anti-virus scanstamp */
 659 #define XAT0_REPARSE    0x00002000      /* FS reparse point */
 660 #define XAT0_GEN        0x00004000      /* object generation number */
 661 #define XAT0_OFFLINE    0x00008000      /* offline */
 662 #define XAT0_SPARSE     0x00010000      /* sparse */
 663 #define XAT0_PROJINHERIT        0x00020000      /* Create with parent projid */
 664 #define XAT0_PROJID     0x00040000      /* Project ID */
 665 
 666 #define XAT0_ALL_ATTRS  (XAT0_CREATETIME|XAT0_ARCHIVE|XAT0_SYSTEM| \
 667     XAT0_READONLY|XAT0_HIDDEN|XAT0_NOUNLINK|XAT0_IMMUTABLE|XAT0_APPENDONLY| \
 668     XAT0_NODUMP|XAT0_OPAQUE|XAT0_AV_QUARANTINED|  XAT0_AV_MODIFIED| \
 669     XAT0_AV_SCANSTAMP|XAT0_REPARSE|XATO_GEN|XAT0_OFFLINE|XAT0_SPARSE| \
 670     XAT0_PROJINHERIT | XAT0_PROJID)
 671 
 672 /* Support for XAT_* optional attributes */
 673 #define XVA_MASK                0xffffffff      /* Used to mask off 32 bits */
 674 #define XVA_SHFT                32              /* Used to shift index */
 675 
 676 /*
 677  * Used to pry out the index and attribute bits from the XAT_* attributes
 678  * defined below.  Note that we're masking things down to 32 bits then
 679  * casting to uint32_t.
 680  */
 681 #define XVA_INDEX(attr)         ((uint32_t)(((attr) >> XVA_SHFT) & XVA_MASK))
 682 #define XVA_ATTRBIT(attr)       ((uint32_t)((attr) & XVA_MASK))
 683 
 684 /*
 685  * The following defines present a "flat namespace" so that consumers don't
 686  * need to keep track of which element belongs to which bitmap entry.
 687  *
 688  * NOTE THAT THESE MUST NEVER BE OR-ed TOGETHER
 689  */
 690 #define XAT_CREATETIME          ((XAT0_INDEX << XVA_SHFT) | XAT0_CREATETIME)
 691 #define XAT_ARCHIVE             ((XAT0_INDEX << XVA_SHFT) | XAT0_ARCHIVE)
 692 #define XAT_SYSTEM              ((XAT0_INDEX << XVA_SHFT) | XAT0_SYSTEM)
 693 #define XAT_READONLY            ((XAT0_INDEX << XVA_SHFT) | XAT0_READONLY)
 694 #define XAT_HIDDEN              ((XAT0_INDEX << XVA_SHFT) | XAT0_HIDDEN)
 695 #define XAT_NOUNLINK            ((XAT0_INDEX << XVA_SHFT) | XAT0_NOUNLINK)
 696 #define XAT_IMMUTABLE           ((XAT0_INDEX << XVA_SHFT) | XAT0_IMMUTABLE)
 697 #define XAT_APPENDONLY          ((XAT0_INDEX << XVA_SHFT) | XAT0_APPENDONLY)
 698 #define XAT_NODUMP              ((XAT0_INDEX << XVA_SHFT) | XAT0_NODUMP)
 699 #define XAT_OPAQUE              ((XAT0_INDEX << XVA_SHFT) | XAT0_OPAQUE)
 700 #define XAT_AV_QUARANTINED      ((XAT0_INDEX << XVA_SHFT) | XAT0_AV_QUARANTINED)
 701 #define XAT_AV_MODIFIED         ((XAT0_INDEX << XVA_SHFT) | XAT0_AV_MODIFIED)
 702 #define XAT_AV_SCANSTAMP        ((XAT0_INDEX << XVA_SHFT) | XAT0_AV_SCANSTAMP)
 703 #define XAT_REPARSE             ((XAT0_INDEX << XVA_SHFT) | XAT0_REPARSE)
 704 #define XAT_GEN                 ((XAT0_INDEX << XVA_SHFT) | XAT0_GEN)
 705 #define XAT_OFFLINE             ((XAT0_INDEX << XVA_SHFT) | XAT0_OFFLINE)
 706 #define XAT_SPARSE              ((XAT0_INDEX << XVA_SHFT) | XAT0_SPARSE)
 707 #define XAT_PROJINHERIT         ((XAT0_INDEX << XVA_SHFT) | XAT0_PROJINHERIT)
 708 #define XAT_PROJID              ((XAT0_INDEX << XVA_SHFT) | XAT0_PROJID)
 709 
 710 /*
 711  * The returned attribute map array (xva_rtnattrmap[]) is located past the
 712  * requested attribute map array (xva_reqattrmap[]).  Its location changes
 713  * when the array sizes change.  We use a separate pointer in a known location
 714  * (xva_rtnattrmapp) to hold the location of xva_rtnattrmap[].  This is
 715  * set in xva_init()
 716  */
 717 #define XVA_RTNATTRMAP(xvap)    ((xvap)->xva_rtnattrmapp)
 718 
 719 /*
 720  * XVA_SET_REQ() sets an attribute bit in the proper element in the bitmap
 721  * of requested attributes (xva_reqattrmap[]).
 722  */
 723 #define XVA_SET_REQ(xvap, attr)                                 \
 724         ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);               \
 725         ASSERT((xvap)->xva_magic == XVA_MAGIC);                      \
 726         (xvap)->xva_reqattrmap[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
 727 /*
 728  * XVA_CLR_REQ() clears an attribute bit in the proper element in the bitmap
 729  * of requested attributes (xva_reqattrmap[]).
 730  */
 731 #define XVA_CLR_REQ(xvap, attr)                                 \
 732         ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);               \
 733         ASSERT((xvap)->xva_magic == XVA_MAGIC);                      \
 734         (xvap)->xva_reqattrmap[XVA_INDEX(attr)] &= ~XVA_ATTRBIT(attr)
 735 
 736 /*
 737  * XVA_SET_RTN() sets an attribute bit in the proper element in the bitmap
 738  * of returned attributes (xva_rtnattrmap[]).
 739  */
 740 #define XVA_SET_RTN(xvap, attr)                                 \
 741         ASSERT((xvap)->xva_vattr.va_mask | AT_XVATTR);               \
 742         ASSERT((xvap)->xva_magic == XVA_MAGIC);                      \
 743         (XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] |= XVA_ATTRBIT(attr)
 744 
 745 /*
 746  * XVA_ISSET_REQ() checks the requested attribute bitmap (xva_reqattrmap[])
 747  * to see of the corresponding attribute bit is set.  If so, returns non-zero.
 748  */
 749 #define XVA_ISSET_REQ(xvap, attr)                                       \
 750         ((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&                 \
 751                 ((xvap)->xva_magic == XVA_MAGIC) &&                  \
 752                 ((xvap)->xva_mapsize > XVA_INDEX(attr))) ?                \
 753         ((xvap)->xva_reqattrmap[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) :  0)
 754 
 755 /*
 756  * XVA_ISSET_RTN() checks the returned attribute bitmap (xva_rtnattrmap[])
 757  * to see of the corresponding attribute bit is set.  If so, returns non-zero.
 758  */
 759 #define XVA_ISSET_RTN(xvap, attr)                                       \
 760         ((((xvap)->xva_vattr.va_mask | AT_XVATTR) &&                 \
 761                 ((xvap)->xva_magic == XVA_MAGIC) &&                  \
 762                 ((xvap)->xva_mapsize > XVA_INDEX(attr))) ?                \
 763         ((XVA_RTNATTRMAP(xvap))[XVA_INDEX(attr)] & XVA_ATTRBIT(attr)) : 0)
 764 
 765 /*
 766  *  Modes.  Some values same as S_xxx entries from stat.h for convenience.
 767  */
 768 #define VSUID           04000           /* set user id on execution */
 769 #define VSGID           02000           /* set group id on execution */
 770 #define VSVTX           01000           /* save swapped text even after use */
 771 
 772 /*
 773  * Permissions.
 774  */
 775 #define VREAD           00400
 776 #define VWRITE          00200
 777 #define VEXEC           00100
 778 
 779 #define MODEMASK        07777           /* mode bits plus permission bits */
 780 #define PERMMASK        00777           /* permission bits */
 781 
 782 /*
 783  * VOP_ACCESS flags
 784  */
 785 #define V_ACE_MASK      0x1     /* mask represents  NFSv4 ACE permissions */
 786 #define V_APPEND        0x2     /* want to do append only check */
 787 
 788 /*
 789  * Check whether mandatory file locking is enabled.
 790  */
 791 
 792 #define MANDMODE(mode)          (((mode) & (VSGID|(VEXEC>>3))) == VSGID)
 793 #define MANDLOCK(vp, mode)      ((vp)->v_type == VREG && MANDMODE(mode))
 794 
 795 /*
 796  * Flags for vnode operations.
 797  */
 798 enum rm         { RMFILE, RMDIRECTORY };        /* rm or rmdir (remove) */
 799 enum symfollow  { NO_FOLLOW, FOLLOW };          /* follow symlinks (or not) */
 800 enum vcexcl     { NONEXCL, EXCL };              /* (non)excl create */
 801 enum create     { CRCREAT, CRMKNOD, CRMKDIR };  /* reason for create */
 802 
 803 typedef enum rm         rm_t;
 804 typedef enum symfollow  symfollow_t;
 805 typedef enum vcexcl     vcexcl_t;
 806 typedef enum create     create_t;
 807 
 808 /*
 809  * Vnode Events - Used by VOP_VNEVENT
 810  * The VE_PRE_RENAME_* events fire before the rename operation and are
 811  * primarily used for specialized applications, such as NFSv4 delegation, which
 812  * need to know about rename before it occurs.
 813  */
 814 typedef enum vnevent    {
 815         VE_SUPPORT      = 0,    /* Query */
 816         VE_RENAME_SRC   = 1,    /* Rename, with vnode as source */
 817         VE_RENAME_DEST  = 2,    /* Rename, with vnode as target/destination */
 818         VE_REMOVE       = 3,    /* Remove of vnode's name */
 819         VE_RMDIR        = 4,    /* Remove of directory vnode's name */
 820         VE_CREATE       = 5,    /* Create with vnode's name which exists */
 821         VE_LINK         = 6,    /* Link with vnode's name as source */
 822         VE_RENAME_DEST_DIR      = 7,    /* Rename with vnode as target dir */
 823         VE_MOUNTEDOVER  = 8,    /* File or Filesystem got mounted over vnode */
 824         VE_TRUNCATE = 9,        /* Truncate */
 825         VE_PRE_RENAME_SRC = 10, /* Pre-rename, with vnode as source */
 826         VE_PRE_RENAME_DEST = 11, /* Pre-rename, with vnode as target/dest. */
 827         VE_PRE_RENAME_DEST_DIR = 12 /* Pre-rename with vnode as target dir */
 828 } vnevent_t;
 829 
 830 /*
 831  * Values for checking vnode open and map counts
 832  */
 833 enum v_mode { V_READ, V_WRITE, V_RDORWR, V_RDANDWR };
 834 
 835 typedef enum v_mode v_mode_t;
 836 
 837 #define V_TRUE  1
 838 #define V_FALSE 0
 839 
 840 /*
 841  * Structure used on VOP_GETSECATTR and VOP_SETSECATTR operations
 842  */
 843 
 844 typedef struct vsecattr {
 845         uint_t          vsa_mask;       /* See below */
 846         int             vsa_aclcnt;     /* ACL entry count */
 847         void            *vsa_aclentp;   /* pointer to ACL entries */
 848         int             vsa_dfaclcnt;   /* default ACL entry count */
 849         void            *vsa_dfaclentp; /* pointer to default ACL entries */
 850         size_t          vsa_aclentsz;   /* ACE size in bytes of vsa_aclentp */
 851         uint_t          vsa_aclflags;   /* ACE ACL flags */
 852 } vsecattr_t;
 853 
 854 /* vsa_mask values */
 855 #define VSA_ACL                 0x0001
 856 #define VSA_ACLCNT              0x0002
 857 #define VSA_DFACL               0x0004
 858 #define VSA_DFACLCNT            0x0008
 859 #define VSA_ACE                 0x0010
 860 #define VSA_ACECNT              0x0020
 861 #define VSA_ACE_ALLTYPES        0x0040
 862 #define VSA_ACE_ACLFLAGS        0x0080  /* get/set ACE ACL flags */
 863 
 864 /*
 865  * Structure used by various vnode operations to determine
 866  * the context (pid, host, identity) of a caller.
 867  *
 868  * The cc_caller_id is used to identify one or more callers who invoke
 869  * operations, possibly on behalf of others.  For example, the NFS
 870  * server could have it's own cc_caller_id which can be detected by
 871  * vnode/vfs operations or (FEM) monitors on those operations.  New
 872  * caller IDs are generated by fs_new_caller_id().
 873  */
 874 typedef struct caller_context {
 875         pid_t           cc_pid;         /* Process ID of the caller */
 876         int             cc_sysid;       /* System ID, used for remote calls */
 877         u_longlong_t    cc_caller_id;   /* Identifier for (set of) caller(s) */
 878         ulong_t         cc_flags;
 879 } caller_context_t;
 880 
 881 /*
 882  * Flags for caller context.  The caller sets CC_DONTBLOCK if it does not
 883  * want to block inside of a FEM monitor.  The monitor will set CC_WOULDBLOCK
 884  * and return EAGAIN if the operation would have blocked.
 885  */
 886 #define CC_WOULDBLOCK   0x01
 887 #define CC_DONTBLOCK    0x02
 888 
 889 /*
 890  * Structure tags for function prototypes, defined elsewhere.
 891  */
 892 struct pathname;
 893 struct fid;
 894 struct flock64;
 895 struct flk_callback;
 896 struct shrlock;
 897 struct page;
 898 struct seg;
 899 struct as;
 900 struct pollhead;
 901 struct taskq;
 902 
 903 #ifdef  _KERNEL
 904 
 905 /*
 906  * VNODE_OPS defines all the vnode operations.  It is used to define
 907  * the vnodeops structure (below) and the fs_func_p union (vfs_opreg.h).
 908  */
 909 #define VNODE_OPS                                                       \
 910         int     (*vop_open)(vnode_t **, int, cred_t *,                  \
 911                                 caller_context_t *);                    \
 912         int     (*vop_close)(vnode_t *, int, int, offset_t, cred_t *,   \
 913                                 caller_context_t *);                    \
 914         int     (*vop_read)(vnode_t *, uio_t *, int, cred_t *,          \
 915                                 caller_context_t *);                    \
 916         int     (*vop_write)(vnode_t *, uio_t *, int, cred_t *,         \
 917                                 caller_context_t *);                    \
 918         int     (*vop_ioctl)(vnode_t *, int, intptr_t, int, cred_t *,   \
 919                                 int *, caller_context_t *);             \
 920         int     (*vop_setfl)(vnode_t *, int, int, cred_t *,             \
 921                                 caller_context_t *);                    \
 922         int     (*vop_getattr)(vnode_t *, vattr_t *, int, cred_t *,     \
 923                                 caller_context_t *);                    \
 924         int     (*vop_setattr)(vnode_t *, vattr_t *, int, cred_t *,     \
 925                                 caller_context_t *);                    \
 926         int     (*vop_access)(vnode_t *, int, int, cred_t *,            \
 927                                 caller_context_t *);                    \
 928         int     (*vop_lookup)(vnode_t *, char *, vnode_t **,            \
 929                                 struct pathname *,                      \
 930                                 int, vnode_t *, cred_t *,               \
 931                                 caller_context_t *, int *,              \
 932                                 struct pathname *);                     \
 933         int     (*vop_create)(vnode_t *, char *, vattr_t *, vcexcl_t,   \
 934                                 int, vnode_t **, cred_t *, int,         \
 935                                 caller_context_t *, vsecattr_t *);      \
 936         int     (*vop_remove)(vnode_t *, char *, cred_t *,              \
 937                                 caller_context_t *, int);               \
 938         int     (*vop_link)(vnode_t *, vnode_t *, char *, cred_t *,     \
 939                                 caller_context_t *, int);               \
 940         int     (*vop_rename)(vnode_t *, char *, vnode_t *, char *,     \
 941                                 cred_t *, caller_context_t *, int);     \
 942         int     (*vop_mkdir)(vnode_t *, char *, vattr_t *, vnode_t **,  \
 943                                 cred_t *, caller_context_t *, int,      \
 944                                 vsecattr_t *);                          \
 945         int     (*vop_rmdir)(vnode_t *, char *, vnode_t *, cred_t *,    \
 946                                 caller_context_t *, int);               \
 947         int     (*vop_readdir)(vnode_t *, uio_t *, cred_t *, int *,     \
 948                                 caller_context_t *, int);               \
 949         int     (*vop_symlink)(vnode_t *, char *, vattr_t *, char *,    \
 950                                 cred_t *, caller_context_t *, int);     \
 951         int     (*vop_readlink)(vnode_t *, uio_t *, cred_t *,           \
 952                                 caller_context_t *);                    \
 953         int     (*vop_fsync)(vnode_t *, int, cred_t *,                  \
 954                                 caller_context_t *);                    \
 955         void    (*vop_inactive)(vnode_t *, cred_t *,                    \
 956                                 caller_context_t *);                    \
 957         int     (*vop_fid)(vnode_t *, struct fid *,                     \
 958                                 caller_context_t *);                    \
 959         int     (*vop_rwlock)(vnode_t *, int, caller_context_t *);      \
 960         void    (*vop_rwunlock)(vnode_t *, int, caller_context_t *);    \
 961         int     (*vop_seek)(vnode_t *, offset_t, offset_t *,            \
 962                                 caller_context_t *);                    \
 963         int     (*vop_cmp)(vnode_t *, vnode_t *, caller_context_t *);   \
 964         int     (*vop_frlock)(vnode_t *, int, struct flock64 *,         \
 965                                 int, offset_t,                          \
 966                                 struct flk_callback *, cred_t *,        \
 967                                 caller_context_t *);                    \
 968         int     (*vop_space)(vnode_t *, int, struct flock64 *,          \
 969                                 int, offset_t,                          \
 970                                 cred_t *, caller_context_t *);          \
 971         int     (*vop_realvp)(vnode_t *, vnode_t **,                    \
 972                                 caller_context_t *);                    \
 973         int     (*vop_getpage)(vnode_t *, offset_t, size_t, uint_t *,   \
 974                                 struct page **, size_t, struct seg *,   \
 975                                 caddr_t, enum seg_rw, cred_t *,         \
 976                                 caller_context_t *);                    \
 977         int     (*vop_putpage)(vnode_t *, offset_t, size_t,             \
 978                                 int, cred_t *, caller_context_t *);     \
 979         int     (*vop_map)(vnode_t *, offset_t, struct as *,            \
 980                                 caddr_t *, size_t,                      \
 981                                 uchar_t, uchar_t, uint_t, cred_t *,     \
 982                                 caller_context_t *);                    \
 983         int     (*vop_addmap)(vnode_t *, offset_t, struct as *,         \
 984                                 caddr_t, size_t,                        \
 985                                 uchar_t, uchar_t, uint_t, cred_t *,     \
 986                                 caller_context_t *);                    \
 987         int     (*vop_delmap)(vnode_t *, offset_t, struct as *,         \
 988                                 caddr_t, size_t,                        \
 989                                 uint_t, uint_t, uint_t, cred_t *,       \
 990                                 caller_context_t *);                    \
 991         int     (*vop_poll)(vnode_t *, short, int, short *,             \
 992                                 struct pollhead **,                     \
 993                                 caller_context_t *);                    \
 994         int     (*vop_dump)(vnode_t *, caddr_t, offset_t, offset_t,     \
 995                                 caller_context_t *);                    \
 996         int     (*vop_pathconf)(vnode_t *, int, ulong_t *, cred_t *,    \
 997                                 caller_context_t *);                    \
 998         int     (*vop_pageio)(vnode_t *, struct page *,                 \
 999                                 u_offset_t, size_t, int, cred_t *,      \
1000                                 caller_context_t *);                    \
1001         int     (*vop_dumpctl)(vnode_t *, int, offset_t *,              \
1002                                 caller_context_t *);                    \
1003         void    (*vop_dispose)(vnode_t *, struct page *,                \
1004                                 int, int, cred_t *,                     \
1005                                 caller_context_t *);                    \
1006         int     (*vop_setsecattr)(vnode_t *, vsecattr_t *,              \
1007                                 int, cred_t *, caller_context_t *);     \
1008         int     (*vop_getsecattr)(vnode_t *, vsecattr_t *,              \
1009                                 int, cred_t *, caller_context_t *);     \
1010         int     (*vop_shrlock)(vnode_t *, int, struct shrlock *,        \
1011                                 int, cred_t *, caller_context_t *);     \
1012         int     (*vop_vnevent)(vnode_t *, vnevent_t, vnode_t *,         \
1013                                 char *, caller_context_t *);            \
1014         int     (*vop_reqzcbuf)(vnode_t *, enum uio_rw, xuio_t *,       \
1015                                 cred_t *, caller_context_t *);          \
1016         int     (*vop_retzcbuf)(vnode_t *, xuio_t *, cred_t *,          \
1017                                 caller_context_t *)
1018         /* NB: No ";" */
1019 
1020 /*
1021  * Operations on vnodes.  Note: File systems must never operate directly
1022  * on a 'vnodeops' structure -- it WILL change in future releases!  They
1023  * must use vn_make_ops() to create the structure.
1024  */
1025 typedef struct vnodeops {
1026         const char *vnop_name;
1027         VNODE_OPS;      /* Signatures of all vnode operations (vops) */
1028 } vnodeops_t;
1029 
1030 typedef int (*fs_generic_func_p) ();    /* Generic vop/vfsop/femop/fsemop ptr */
1031 
1032 extern int      fop_open(vnode_t **, int, cred_t *, caller_context_t *);
1033 extern int      fop_close(vnode_t *, int, int, offset_t, cred_t *,
1034                                 caller_context_t *);
1035 extern int      fop_read(vnode_t *, uio_t *, int, cred_t *, caller_context_t *);
1036 extern int      fop_write(vnode_t *, uio_t *, int, cred_t *,
1037                                 caller_context_t *);
1038 extern int      fop_ioctl(vnode_t *, int, intptr_t, int, cred_t *, int *,
1039                                 caller_context_t *);
1040 extern int      fop_setfl(vnode_t *, int, int, cred_t *, caller_context_t *);
1041 extern int      fop_getattr(vnode_t *, vattr_t *, int, cred_t *,
1042                                 caller_context_t *);
1043 extern int      fop_setattr(vnode_t *, vattr_t *, int, cred_t *,
1044                                 caller_context_t *);
1045 extern int      fop_access(vnode_t *, int, int, cred_t *, caller_context_t *);
1046 extern int      fop_lookup(vnode_t *, char *, vnode_t **, struct pathname *,
1047                                 int, vnode_t *, cred_t *, caller_context_t *,
1048                                 int *, struct pathname *);
1049 extern int      fop_create(vnode_t *, char *, vattr_t *, vcexcl_t, int,
1050                                 vnode_t **, cred_t *, int, caller_context_t *,
1051                                 vsecattr_t *);
1052 extern int      fop_remove(vnode_t *vp, char *, cred_t *, caller_context_t *,
1053                                 int);
1054 extern int      fop_link(vnode_t *, vnode_t *, char *, cred_t *,
1055                                 caller_context_t *, int);
1056 extern int      fop_rename(vnode_t *, char *, vnode_t *, char *, cred_t *,
1057                                 caller_context_t *, int);
1058 extern int      fop_mkdir(vnode_t *, char *, vattr_t *, vnode_t **, cred_t *,
1059                                 caller_context_t *, int, vsecattr_t *);
1060 extern int      fop_rmdir(vnode_t *, char *, vnode_t *, cred_t *,
1061                                 caller_context_t *, int);
1062 extern int      fop_readdir(vnode_t *, uio_t *, cred_t *, int *,
1063                                 caller_context_t *, int);
1064 extern int      fop_symlink(vnode_t *, char *, vattr_t *, char *, cred_t *,
1065                                 caller_context_t *, int);
1066 extern int      fop_readlink(vnode_t *, uio_t *, cred_t *, caller_context_t *);
1067 extern int      fop_fsync(vnode_t *, int, cred_t *, caller_context_t *);
1068 extern void     fop_inactive(vnode_t *, cred_t *, caller_context_t *);
1069 extern int      fop_fid(vnode_t *, struct fid *, caller_context_t *);
1070 extern int      fop_rwlock(vnode_t *, int, caller_context_t *);
1071 extern void     fop_rwunlock(vnode_t *, int, caller_context_t *);
1072 extern int      fop_seek(vnode_t *, offset_t, offset_t *, caller_context_t *);
1073 extern int      fop_cmp(vnode_t *, vnode_t *, caller_context_t *);
1074 extern int      fop_frlock(vnode_t *, int, struct flock64 *, int, offset_t,
1075                                 struct flk_callback *, cred_t *,
1076                                 caller_context_t *);
1077 extern int      fop_space(vnode_t *, int, struct flock64 *, int, offset_t,
1078                                 cred_t *, caller_context_t *);
1079 extern int      fop_realvp(vnode_t *, vnode_t **, caller_context_t *);
1080 extern int      fop_getpage(vnode_t *, offset_t, size_t, uint_t *,
1081                                 struct page **, size_t, struct seg *,
1082                                 caddr_t, enum seg_rw, cred_t *,
1083                                 caller_context_t *);
1084 extern int      fop_putpage(vnode_t *, offset_t, size_t, int, cred_t *,
1085                                 caller_context_t *);
1086 extern int      fop_map(vnode_t *, offset_t, struct as *, caddr_t *, size_t,
1087                                 uchar_t, uchar_t, uint_t, cred_t *cr,
1088                                 caller_context_t *);
1089 extern int      fop_addmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
1090                                 uchar_t, uchar_t, uint_t, cred_t *,
1091                                 caller_context_t *);
1092 extern int      fop_delmap(vnode_t *, offset_t, struct as *, caddr_t, size_t,
1093                                 uint_t, uint_t, uint_t, cred_t *,
1094                                 caller_context_t *);
1095 extern int      fop_poll(vnode_t *, short, int, short *, struct pollhead **,
1096                                 caller_context_t *);
1097 extern int      fop_dump(vnode_t *, caddr_t, offset_t, offset_t,
1098     caller_context_t *);
1099 extern int      fop_pathconf(vnode_t *, int, ulong_t *, cred_t *,
1100                                 caller_context_t *);
1101 extern int      fop_pageio(vnode_t *, struct page *, u_offset_t, size_t, int,
1102                                 cred_t *, caller_context_t *);
1103 extern int      fop_dumpctl(vnode_t *, int, offset_t *, caller_context_t *);
1104 extern void     fop_dispose(vnode_t *, struct page *, int, int, cred_t *,
1105                                 caller_context_t *);
1106 extern int      fop_setsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
1107                                 caller_context_t *);
1108 extern int      fop_getsecattr(vnode_t *, vsecattr_t *, int, cred_t *,
1109                                 caller_context_t *);
1110 extern int      fop_shrlock(vnode_t *, int, struct shrlock *, int, cred_t *,
1111                                 caller_context_t *);
1112 extern int      fop_vnevent(vnode_t *, vnevent_t, vnode_t *, char *,
1113                                 caller_context_t *);
1114 extern int      fop_reqzcbuf(vnode_t *, enum uio_rw, xuio_t *, cred_t *,
1115                                 caller_context_t *);
1116 extern int      fop_retzcbuf(vnode_t *, xuio_t *, cred_t *, caller_context_t *);
1117 
1118 #endif  /* _KERNEL */
1119 
1120 #define VOP_OPEN(vpp, mode, cr, ct) \
1121         fop_open(vpp, mode, cr, ct)
1122 #define VOP_CLOSE(vp, f, c, o, cr, ct) \
1123         fop_close(vp, f, c, o, cr, ct)
1124 #define VOP_READ(vp, uiop, iof, cr, ct) \
1125         fop_read(vp, uiop, iof, cr, ct)
1126 #define VOP_WRITE(vp, uiop, iof, cr, ct) \
1127         fop_write(vp, uiop, iof, cr, ct)
1128 #define VOP_IOCTL(vp, cmd, a, f, cr, rvp, ct) \
1129         fop_ioctl(vp, cmd, a, f, cr, rvp, ct)
1130 #define VOP_SETFL(vp, f, a, cr, ct) \
1131         fop_setfl(vp, f, a, cr, ct)
1132 #define VOP_GETATTR(vp, vap, f, cr, ct) \
1133         fop_getattr(vp, vap, f, cr, ct)
1134 #define VOP_SETATTR(vp, vap, f, cr, ct) \
1135         fop_setattr(vp, vap, f, cr, ct)
1136 #define VOP_ACCESS(vp, mode, f, cr, ct) \
1137         fop_access(vp, mode, f, cr, ct)
1138 #define VOP_LOOKUP(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp) \
1139         fop_lookup(vp, cp, vpp, pnp, f, rdir, cr, ct, defp, rpnp)
1140 #define VOP_CREATE(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap) \
1141         fop_create(dvp, p, vap, ex, mode, vpp, cr, flag, ct, vsap)
1142 #define VOP_REMOVE(dvp, p, cr, ct, f) \
1143         fop_remove(dvp, p, cr, ct, f)
1144 #define VOP_LINK(tdvp, fvp, p, cr, ct, f) \
1145         fop_link(tdvp, fvp, p, cr, ct, f)
1146 #define VOP_RENAME(fvp, fnm, tdvp, tnm, cr, ct, f) \
1147         fop_rename(fvp, fnm, tdvp, tnm, cr, ct, f)
1148 #define VOP_MKDIR(dp, p, vap, vpp, cr, ct, f, vsap) \
1149         fop_mkdir(dp, p, vap, vpp, cr, ct, f, vsap)
1150 #define VOP_RMDIR(dp, p, cdir, cr, ct, f) \
1151         fop_rmdir(dp, p, cdir, cr, ct, f)
1152 #define VOP_READDIR(vp, uiop, cr, eofp, ct, f) \
1153         fop_readdir(vp, uiop, cr, eofp, ct, f)
1154 #define VOP_SYMLINK(dvp, lnm, vap, tnm, cr, ct, f) \
1155         fop_symlink(dvp, lnm, vap, tnm, cr, ct, f)
1156 #define VOP_READLINK(vp, uiop, cr, ct) \
1157         fop_readlink(vp, uiop, cr, ct)
1158 #define VOP_FSYNC(vp, syncflag, cr, ct) \
1159         fop_fsync(vp, syncflag, cr, ct)
1160 #define VOP_INACTIVE(vp, cr, ct) \
1161         fop_inactive(vp, cr, ct)
1162 #define VOP_FID(vp, fidp, ct) \
1163         fop_fid(vp, fidp, ct)
1164 #define VOP_RWLOCK(vp, w, ct) \
1165         fop_rwlock(vp, w, ct)
1166 #define VOP_RWUNLOCK(vp, w, ct) \
1167         fop_rwunlock(vp, w, ct)
1168 #define VOP_SEEK(vp, ooff, noffp, ct) \
1169         fop_seek(vp, ooff, noffp, ct)
1170 #define VOP_CMP(vp1, vp2, ct) \
1171         fop_cmp(vp1, vp2, ct)
1172 #define VOP_FRLOCK(vp, cmd, a, f, o, cb, cr, ct) \
1173         fop_frlock(vp, cmd, a, f, o, cb, cr, ct)
1174 #define VOP_SPACE(vp, cmd, a, f, o, cr, ct) \
1175         fop_space(vp, cmd, a, f, o, cr, ct)
1176 #define VOP_REALVP(vp1, vp2, ct) \
1177         fop_realvp(vp1, vp2, ct)
1178 #define VOP_GETPAGE(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct) \
1179         fop_getpage(vp, of, sz, pr, pl, ps, sg, a, rw, cr, ct)
1180 #define VOP_PUTPAGE(vp, of, sz, fl, cr, ct) \
1181         fop_putpage(vp, of, sz, fl, cr, ct)
1182 #define VOP_MAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1183         fop_map(vp, of, as, a, sz, p, mp, fl, cr, ct)
1184 #define VOP_ADDMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1185         fop_addmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1186 #define VOP_DELMAP(vp, of, as, a, sz, p, mp, fl, cr, ct) \
1187         fop_delmap(vp, of, as, a, sz, p, mp, fl, cr, ct)
1188 #define VOP_POLL(vp, events, anyyet, reventsp, phpp, ct) \
1189         fop_poll(vp, events, anyyet, reventsp, phpp, ct)
1190 #define VOP_DUMP(vp, addr, bn, count, ct) \
1191         fop_dump(vp, addr, bn, count, ct)
1192 #define VOP_PATHCONF(vp, cmd, valp, cr, ct) \
1193         fop_pathconf(vp, cmd, valp, cr, ct)
1194 #define VOP_PAGEIO(vp, pp, io_off, io_len, flags, cr, ct) \
1195         fop_pageio(vp, pp, io_off, io_len, flags, cr, ct)
1196 #define VOP_DUMPCTL(vp, action, blkp, ct) \
1197         fop_dumpctl(vp, action, blkp, ct)
1198 #define VOP_DISPOSE(vp, pp, flag, dn, cr, ct) \
1199         fop_dispose(vp, pp, flag, dn, cr, ct)
1200 #define VOP_GETSECATTR(vp, vsap, f, cr, ct) \
1201         fop_getsecattr(vp, vsap, f, cr, ct)
1202 #define VOP_SETSECATTR(vp, vsap, f, cr, ct) \
1203         fop_setsecattr(vp, vsap, f, cr, ct)
1204 #define VOP_SHRLOCK(vp, cmd, shr, f, cr, ct) \
1205         fop_shrlock(vp, cmd, shr, f, cr, ct)
1206 #define VOP_VNEVENT(vp, vnevent, dvp, fnm, ct) \
1207         fop_vnevent(vp, vnevent, dvp, fnm, ct)
1208 #define VOP_REQZCBUF(vp, rwflag, xuiop, cr, ct) \
1209         fop_reqzcbuf(vp, rwflag, xuiop, cr, ct)
1210 #define VOP_RETZCBUF(vp, xuiop, cr, ct) \
1211         fop_retzcbuf(vp, xuiop, cr, ct)
1212 
1213 #define VOPNAME_OPEN            "open"
1214 #define VOPNAME_CLOSE           "close"
1215 #define VOPNAME_READ            "read"
1216 #define VOPNAME_WRITE           "write"
1217 #define VOPNAME_IOCTL           "ioctl"
1218 #define VOPNAME_SETFL           "setfl"
1219 #define VOPNAME_GETATTR         "getattr"
1220 #define VOPNAME_SETATTR         "setattr"
1221 #define VOPNAME_ACCESS          "access"
1222 #define VOPNAME_LOOKUP          "lookup"
1223 #define VOPNAME_CREATE          "create"
1224 #define VOPNAME_REMOVE          "remove"
1225 #define VOPNAME_LINK            "link"
1226 #define VOPNAME_RENAME          "rename"
1227 #define VOPNAME_MKDIR           "mkdir"
1228 #define VOPNAME_RMDIR           "rmdir"
1229 #define VOPNAME_READDIR         "readdir"
1230 #define VOPNAME_SYMLINK         "symlink"
1231 #define VOPNAME_READLINK        "readlink"
1232 #define VOPNAME_FSYNC           "fsync"
1233 #define VOPNAME_INACTIVE        "inactive"
1234 #define VOPNAME_FID             "fid"
1235 #define VOPNAME_RWLOCK          "rwlock"
1236 #define VOPNAME_RWUNLOCK        "rwunlock"
1237 #define VOPNAME_SEEK            "seek"
1238 #define VOPNAME_CMP             "cmp"
1239 #define VOPNAME_FRLOCK          "frlock"
1240 #define VOPNAME_SPACE           "space"
1241 #define VOPNAME_REALVP          "realvp"
1242 #define VOPNAME_GETPAGE         "getpage"
1243 #define VOPNAME_PUTPAGE         "putpage"
1244 #define VOPNAME_MAP             "map"
1245 #define VOPNAME_ADDMAP          "addmap"
1246 #define VOPNAME_DELMAP          "delmap"
1247 #define VOPNAME_POLL            "poll"
1248 #define VOPNAME_DUMP            "dump"
1249 #define VOPNAME_PATHCONF        "pathconf"
1250 #define VOPNAME_PAGEIO          "pageio"
1251 #define VOPNAME_DUMPCTL         "dumpctl"
1252 #define VOPNAME_DISPOSE         "dispose"
1253 #define VOPNAME_GETSECATTR      "getsecattr"
1254 #define VOPNAME_SETSECATTR      "setsecattr"
1255 #define VOPNAME_SHRLOCK         "shrlock"
1256 #define VOPNAME_VNEVENT         "vnevent"
1257 #define VOPNAME_REQZCBUF        "reqzcbuf"
1258 #define VOPNAME_RETZCBUF        "retzcbuf"
1259 
1260 /*
1261  * Flags for VOP_LOOKUP
1262  *
1263  * Defined in file.h, but also possible, FIGNORECASE and FSEARCH
1264  *
1265  */
1266 #define LOOKUP_DIR              0x01    /* want parent dir vp */
1267 #define LOOKUP_XATTR            0x02    /* lookup up extended attr dir */
1268 #define CREATE_XATTR_DIR        0x04    /* Create extended attr dir */
1269 #define LOOKUP_HAVE_SYSATTR_DIR 0x08    /* Already created virtual GFS dir */
1270 
1271 /*
1272  * Flags for VOP_READDIR
1273  */
1274 #define V_RDDIR_ENTFLAGS        0x01    /* request dirent flags */
1275 #define V_RDDIR_ACCFILTER       0x02    /* filter out inaccessible dirents */
1276 
1277 /*
1278  * Flags for VOP_RWLOCK/VOP_RWUNLOCK
1279  * VOP_RWLOCK will return the flag that was actually set, or -1 if none.
1280  */
1281 #define V_WRITELOCK_TRUE        (1)     /* Request write-lock on the vnode */
1282 #define V_WRITELOCK_FALSE       (0)     /* Request read-lock on the vnode */
1283 
1284 /*
1285  * Flags for VOP_DUMPCTL
1286  */
1287 #define DUMP_ALLOC      0
1288 #define DUMP_FREE       1
1289 #define DUMP_SCAN       2
1290 
1291 /*
1292  * Public vnode manipulation functions.
1293  */
1294 #ifdef  _KERNEL
1295 
1296 vnode_t *vn_alloc(int);
1297 void    vn_reinit(vnode_t *);
1298 void    vn_recycle(vnode_t *);
1299 void    vn_free(vnode_t *);
1300 
1301 int     vn_is_readonly(vnode_t *);
1302 int     vn_is_opened(vnode_t *, v_mode_t);
1303 int     vn_is_mapped(vnode_t *, v_mode_t);
1304 int     vn_has_other_opens(vnode_t *, v_mode_t);
1305 void    vn_open_upgrade(vnode_t *, int);
1306 void    vn_open_downgrade(vnode_t *, int);
1307 
1308 int     vn_can_change_zones(vnode_t *vp);
1309 
1310 int     vn_has_flocks(vnode_t *);
1311 int     vn_has_mandatory_locks(vnode_t *, int);
1312 int     vn_has_cached_data(vnode_t *);
1313 
1314 void    vn_setops(vnode_t *, vnodeops_t *);
1315 vnodeops_t *vn_getops(vnode_t *);
1316 int     vn_matchops(vnode_t *, vnodeops_t *);
1317 int     vn_matchopval(vnode_t *, char *, fs_generic_func_p);
1318 int     vn_ismntpt(vnode_t *);
1319 
1320 struct vfs *vn_mountedvfs(vnode_t *);
1321 
1322 int     vn_in_dnlc(vnode_t *);
1323 
1324 void    vn_create_cache(void);
1325 void    vn_destroy_cache(void);
1326 
1327 void    vn_freevnodeops(vnodeops_t *);
1328 
1329 int     vn_open(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1330                 struct vnode **vpp, enum create crwhy, mode_t umask);
1331 int     vn_openat(char *pnamep, enum uio_seg seg, int filemode, int createmode,
1332                 struct vnode **vpp, enum create crwhy,
1333                 mode_t umask, struct vnode *startvp, int fd);
1334 int     vn_create(char *pnamep, enum uio_seg seg, struct vattr *vap,
1335                 enum vcexcl excl, int mode, struct vnode **vpp,
1336                 enum create why, int flag, mode_t umask);
1337 int     vn_createat(char *pnamep, enum uio_seg seg, struct vattr *vap,
1338                 enum vcexcl excl, int mode, struct vnode **vpp,
1339                 enum create why, int flag, mode_t umask, struct vnode *startvp);
1340 int     vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base, ssize_t len,
1341                 offset_t offset, enum uio_seg seg, int ioflag, rlim64_t ulimit,
1342                 cred_t *cr, ssize_t *residp);
1343 void    vn_rele(struct vnode *vp);
1344 void    vn_rele_async(struct vnode *vp, struct taskq *taskq);
1345 void    vn_rele_dnlc(struct vnode *vp);
1346 void    vn_rele_stream(struct vnode *vp);
1347 int     vn_link(char *from, char *to, enum uio_seg seg);
1348 int     vn_linkat(vnode_t *fstartvp, char *from, enum symfollow follow,
1349                 vnode_t *tstartvp, char *to, enum uio_seg seg);
1350 int     vn_rename(char *from, char *to, enum uio_seg seg);
1351 int     vn_renameat(vnode_t *fdvp, char *fname, vnode_t *tdvp, char *tname,
1352                 enum uio_seg seg);
1353 int     vn_remove(char *fnamep, enum uio_seg seg, enum rm dirflag);
1354 int     vn_removeat(vnode_t *startvp, char *fnamep, enum uio_seg seg,
1355                 enum rm dirflag);
1356 int     vn_compare(vnode_t *vp1, vnode_t *vp2);
1357 int     vn_vfswlock(struct vnode *vp);
1358 int     vn_vfswlock_wait(struct vnode *vp);
1359 int     vn_vfsrlock(struct vnode *vp);
1360 int     vn_vfsrlock_wait(struct vnode *vp);
1361 void    vn_vfsunlock(struct vnode *vp);
1362 int     vn_vfswlock_held(struct vnode *vp);
1363 vnode_t *specvp(struct vnode *vp, dev_t dev, vtype_t type, struct cred *cr);
1364 vnode_t *makespecvp(dev_t dev, vtype_t type);
1365 vn_vfslocks_entry_t *vn_vfslocks_getlock(void *);
1366 void    vn_vfslocks_rele(vn_vfslocks_entry_t *);
1367 boolean_t vn_is_reparse(vnode_t *, cred_t *, caller_context_t *);
1368 
1369 void vn_copypath(struct vnode *src, struct vnode *dst);
1370 void vn_setpath_str(struct vnode *vp, const char *str, size_t len);
1371 void vn_setpath(vnode_t *rootvp, struct vnode *startvp, struct vnode *vp,
1372     const char *path, size_t plen);
1373 void vn_renamepath(vnode_t *dvp, vnode_t *vp, const char *nm, size_t len);
1374 
1375 /* Private vnode manipulation functions */
1376 void vn_clearpath(vnode_t *, hrtime_t);
1377 void vn_updatepath(vnode_t *, vnode_t *, const char *);
1378 
1379 
1380 /* Vnode event notification */
1381 void    vnevent_rename_src(vnode_t *, vnode_t *, char *, caller_context_t *);
1382 void    vnevent_rename_dest(vnode_t *, vnode_t *, char *, caller_context_t *);
1383 void    vnevent_remove(vnode_t *, vnode_t *, char *, caller_context_t *);
1384 void    vnevent_rmdir(vnode_t *, vnode_t *, char *, caller_context_t *);
1385 void    vnevent_create(vnode_t *, caller_context_t *);
1386 void    vnevent_link(vnode_t *, caller_context_t *);
1387 void    vnevent_rename_dest_dir(vnode_t *, caller_context_t *ct);
1388 void    vnevent_mountedover(vnode_t *, caller_context_t *);
1389 void    vnevent_truncate(vnode_t *, caller_context_t *);
1390 int     vnevent_support(vnode_t *, caller_context_t *);
1391 void    vnevent_pre_rename_src(vnode_t *, vnode_t *, char *,
1392             caller_context_t *);
1393 void    vnevent_pre_rename_dest(vnode_t *, vnode_t *, char *,
1394             caller_context_t *);
1395 void    vnevent_pre_rename_dest_dir(vnode_t *, vnode_t *, char *,
1396             caller_context_t *);
1397 
1398 /* Vnode specific data */
1399 void vsd_create(uint_t *, void (*)(void *));
1400 void vsd_destroy(uint_t *);
1401 void *vsd_get(vnode_t *, uint_t);
1402 int vsd_set(vnode_t *, uint_t, void *);
1403 void vsd_free(vnode_t *);
1404 
1405 /*
1406  * Extensible vnode attribute (xva) routines:
1407  * xva_init() initializes an xvattr_t (zero struct, init mapsize, set AT_XATTR)
1408  * xva_getxoptattr() returns a ponter to the xoptattr_t section of xvattr_t
1409  */
1410 void            xva_init(xvattr_t *);
1411 xoptattr_t      *xva_getxoptattr(xvattr_t *);   /* Get ptr to xoptattr_t */
1412 
1413 void xattr_init(void);          /* Initialize vnodeops for xattrs */
1414 
1415 /* GFS tunnel for xattrs */
1416 int xattr_dir_lookup(vnode_t *, vnode_t **, int, cred_t *);
1417 
1418 /* Reparse Point */
1419 void reparse_point_init(void);
1420 
1421 /* Context identification */
1422 u_longlong_t    fs_new_caller_id();
1423 
1424 int     vn_vmpss_usepageio(vnode_t *);
1425 
1426 /* Empty v_path placeholder */
1427 extern char *vn_vpath_empty;
1428 
1429 /*
1430  * Needed for use of IS_VMODSORT() in kernel.
1431  */
1432 extern uint_t pvn_vmodsort_supported;
1433 
1434 /*
1435  * All changes to v_count should be done through VN_HOLD() or VN_RELE(), or
1436  * one of their variants. This makes it possible to ensure proper locking,
1437  * and to guarantee that all modifications are accompanied by a firing of
1438  * the vn-hold or vn-rele SDT DTrace probe.
1439  *
1440  * Example DTrace command for tracing vnode references using these probes:
1441  *
1442  * dtrace -q -n 'sdt:::vn-hold,sdt:::vn-rele
1443  * {
1444  *      this->vp = (vnode_t *)arg0;
1445  *      printf("%s %s(%p[%s]) %d\n", execname, probename, this->vp,
1446  *          this->vp->v_path == NULL ? "NULL" : stringof(this->vp->v_path),
1447  *          this->vp->v_count)
1448  * }'
1449  */
1450 #define VN_HOLD_LOCKED(vp) {                    \
1451         ASSERT(mutex_owned(&(vp)->v_lock));      \
1452         (vp)->v_count++;                     \
1453         DTRACE_PROBE1(vn__hold, vnode_t *, vp); \
1454 }
1455 
1456 #define VN_HOLD(vp)     {               \
1457         mutex_enter(&(vp)->v_lock);      \
1458         VN_HOLD_LOCKED(vp);             \
1459         mutex_exit(&(vp)->v_lock);       \
1460 }
1461 
1462 #define VN_RELE(vp)     { \
1463         vn_rele(vp); \
1464 }
1465 
1466 #define VN_RELE_ASYNC(vp, taskq)        { \
1467         vn_rele_async(vp, taskq); \
1468 }
1469 
1470 #define VN_RELE_LOCKED(vp) {                    \
1471         ASSERT(mutex_owned(&(vp)->v_lock));      \
1472         ASSERT((vp)->v_count >= 1);               \
1473         (vp)->v_count--;                     \
1474         DTRACE_PROBE1(vn__rele, vnode_t *, vp); \
1475 }
1476 
1477 #define VN_SET_VFS_TYPE_DEV(vp, vfsp, type, dev)        { \
1478         (vp)->v_vfsp = (vfsp); \
1479         (vp)->v_type = (type); \
1480         (vp)->v_rdev = (dev); \
1481 }
1482 
1483 /*
1484  * Compare two vnodes for equality.  In general this macro should be used
1485  * in preference to calling VOP_CMP directly.
1486  */
1487 #define VN_CMP(VP1, VP2)        ((VP1) == (VP2) ? 1 :   \
1488         ((VP1) && (VP2) && (vn_getops(VP1) == vn_getops(VP2)) ? \
1489         VOP_CMP(VP1, VP2, NULL) : 0))
1490 
1491 /*
1492  * Some well-known global vnodes used by the VM system to name pages.
1493  */
1494 extern struct vnode kvps[];
1495 
1496 typedef enum {
1497         KV_KVP,         /* vnode for all segkmem pages */
1498         KV_ZVP,         /* vnode for all ZFS pages */
1499 #if defined(__sparc)
1500         KV_MPVP,        /* vnode for all page_t meta-pages */
1501         KV_PROMVP,      /* vnode for all PROM pages */
1502 #endif  /* __sparc */
1503         KV_MAX          /* total number of vnodes in kvps[] */
1504 } kvps_index_t;
1505 
1506 #define VN_ISKAS(vp)    ((vp) >= &kvps[0] && (vp) < &kvps[KV_MAX])
1507 
1508 #endif  /* _KERNEL */
1509 
1510 /*
1511  * Flags to VOP_SETATTR/VOP_GETATTR.
1512  */
1513 #define ATTR_UTIME      0x01    /* non-default utime(2) request */
1514 #define ATTR_EXEC       0x02    /* invocation from exec(2) */
1515 #define ATTR_COMM       0x04    /* yield common vp attributes */
1516 #define ATTR_HINT       0x08    /* information returned will be `hint' */
1517 #define ATTR_REAL       0x10    /* yield attributes of the real vp */
1518 #define ATTR_NOACLCHECK 0x20    /* Don't check ACL when checking permissions */
1519 #define ATTR_TRIGGER    0x40    /* Mount first if vnode is a trigger mount */
1520 /*
1521  * Generally useful macros.
1522  */
1523 #define VBSIZE(vp)      ((vp)->v_vfsp->vfs_bsize)
1524 
1525 #define VTOZONE(vp)     ((vp)->v_vfsp->vfs_zone)
1526 
1527 #define NULLVP          ((struct vnode *)0)
1528 #define NULLVPP         ((struct vnode **)0)
1529 
1530 #ifdef  _KERNEL
1531 
1532 /*
1533  * Structure used while handling asynchronous VOP_PUTPAGE operations.
1534  */
1535 struct async_reqs {
1536         struct async_reqs *a_next;      /* pointer to next arg struct */
1537         struct vnode *a_vp;             /* vnode pointer */
1538         u_offset_t a_off;                       /* offset in file */
1539         uint_t a_len;                   /* size of i/o request */
1540         int a_flags;                    /* flags to indicate operation type */
1541         struct cred *a_cred;            /* cred pointer */
1542         ushort_t a_prealloced;          /* set if struct is pre-allocated */
1543 };
1544 
1545 /*
1546  * VN_DISPOSE() -- given a page pointer, safely invoke VOP_DISPOSE().
1547  * Note that there is no guarantee that the page passed in will be
1548  * freed.  If that is required, then a check after calling VN_DISPOSE would
1549  * be necessary to ensure the page was freed.
1550  */
1551 #define VN_DISPOSE(pp, flag, dn, cr)    { \
1552         if ((pp)->p_vnode != NULL && !VN_ISKAS((pp)->p_vnode)) \
1553                 VOP_DISPOSE((pp)->p_vnode, (pp), (flag), (dn), (cr), NULL); \
1554         else if ((flag) == B_FREE) \
1555                 page_free((pp), (dn)); \
1556         else \
1557                 page_destroy((pp), (dn)); \
1558         }
1559 
1560 #endif  /* _KERNEL */
1561 
1562 #ifdef  __cplusplus
1563 }
1564 #endif
1565 
1566 #endif  /* _SYS_VNODE_H */