vnode.h revision 28954
1/* 2 * Copyright (c) 1989, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)vnode.h 8.7 (Berkeley) 2/4/94 34 * $Id: vnode.h,v 1.46 1997/08/26 07:32:46 phk Exp $ 35 */ 36 37#ifndef _SYS_VNODE_H_ 38#define _SYS_VNODE_H_ 39 40#include <sys/lock.h> 41#include <sys/queue.h> 42 43#ifdef SMP 44#include <machine/smp.h> 45#endif 46 47/* 48 * The vnode is the focus of all file activity in UNIX. There is a 49 * unique vnode allocated for each active file, each current directory, 50 * each mounted-on file, text file, and the root. 51 */ 52 53/* 54 * Vnode types. VNON means no type. 55 */ 56enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD }; 57 58/* 59 * Vnode tag types. 60 * These are for the benefit of external programs only (e.g., pstat) 61 * and should NEVER be inspected by the kernel. 62 */ 63enum vtagtype { 64 VT_NON, VT_UFS, VT_NFS, VT_MFS, VT_PC, VT_LFS, VT_LOFS, VT_FDESC, 65 VT_PORTAL, VT_NULL, VT_UMAP, VT_KERNFS, VT_PROCFS, VT_AFS, VT_ISOFS, 66 VT_UNION, VT_MSDOSFS, VT_DEVFS, VT_TFS 67}; 68 69/* 70 * Each underlying filesystem allocates its own private area and hangs 71 * it from v_data. If non-null, this area is freed in getnewvnode(). 72 */ 73LIST_HEAD(buflists, buf); 74 75typedef int vop_t __P((void *)); 76struct vm_object; 77struct namecache; 78 79/* 80 * Reading or writing any of these items requires holding the appropriate lock. 81 * v_freelist is locked by the global vnode_free_list simple lock. 82 * v_mntvnodes is locked by the global mntvnodes simple lock. 83 * v_flag, v_usecount, v_holdcount and v_writecount are 84 * locked by the v_interlock simple lock. 85 */ 86struct vnode { 87 u_long v_flag; /* vnode flags (see below) */ 88 int v_usecount; /* reference count of users */ 89 int v_writecount; /* reference count of writers */ 90 int v_holdcnt; /* page & buffer references */ 91 daddr_t v_lastr; /* last read (read-ahead) */ 92 u_long v_id; /* capability identifier */ 93 struct mount *v_mount; /* ptr to vfs we are in */ 94 vop_t **v_op; /* vnode operations vector */ 95 TAILQ_ENTRY(vnode) v_freelist; /* vnode freelist */ 96 LIST_ENTRY(vnode) v_mntvnodes; /* vnodes for mount point */ 97 struct buflists v_cleanblkhd; /* clean blocklist head */ 98 struct buflists v_dirtyblkhd; /* dirty blocklist head */ 99 long v_numoutput; /* num of writes in progress */ 100 enum vtype v_type; /* vnode type */ 101 union { 102 struct mount *vu_mountedhere;/* ptr to mounted vfs (VDIR) */ 103 struct socket *vu_socket; /* unix ipc (VSOCK) */ 104 struct specinfo *vu_specinfo; /* device (VCHR, VBLK) */ 105 struct fifoinfo *vu_fifoinfo; /* fifo (VFIFO) */ 106 } v_un; 107 struct nqlease *v_lease; /* Soft reference to lease */ 108 daddr_t v_lastw; /* last write (write cluster) */ 109 daddr_t v_cstart; /* start block of cluster */ 110 daddr_t v_lasta; /* last allocation */ 111 int v_clen; /* length of current cluster */ 112 struct vm_object *v_object; /* Place to store VM object */ 113 struct simplelock v_interlock; /* lock on usecount and flag */ 114 struct lock *v_vnlock; /* used for non-locking fs's */ 115 enum vtagtype v_tag; /* type of underlying data */ 116 void *v_data; /* private data for fs */ 117 LIST_HEAD(, namecache) v_cache_src; /* Cache entries from us */ 118 TAILQ_HEAD(, namecache) v_cache_dst; /* Cache entries to us */ 119 struct vnode *v_dd; /* .. vnode */ 120 u_long v_ddid; /* .. capability identifier */ 121}; 122#define v_mountedhere v_un.vu_mountedhere 123#define v_socket v_un.vu_socket 124#define v_specinfo v_un.vu_specinfo 125#define v_fifoinfo v_un.vu_fifoinfo 126 127/* 128 * Vnode flags. 129 */ 130#define VROOT 0x00001 /* root of its file system */ 131#define VTEXT 0x00002 /* vnode is a pure text prototype */ 132#define VSYSTEM 0x00004 /* vnode being used by kernel */ 133#define VISTTY 0x00008 /* vnode represents a tty */ 134#define VXLOCK 0x00100 /* vnode is locked to change underlying type */ 135#define VXWANT 0x00200 /* process is waiting for vnode */ 136#define VBWAIT 0x00400 /* waiting for output to complete */ 137#define VALIASED 0x00800 /* vnode has an alias */ 138#define VDIROP 0x01000 /* LFS: vnode is involved in a directory op */ 139#define VVMIO 0x02000 /* VMIO flag */ 140#define VNINACT 0x04000 /* LFS: skip ufs_inactive() in lfs_vunref */ 141#define VAGE 0x08000 /* Insert vnode at head of free list */ 142#define VOLOCK 0x10000 /* vnode is locked waiting for an object */ 143#define VOWANT 0x20000 /* a process is waiting for VOLOCK */ 144#define VDOOMED 0x40000 /* This vnode is being recycled */ 145#define VFREE 0x80000 /* This vnode is on the freelist */ 146 147/* 148 * Vnode attributes. A field value of VNOVAL represents a field whose value 149 * is unavailable (getattr) or which is not to be changed (setattr). 150 */ 151struct vattr { 152 enum vtype va_type; /* vnode type (for create) */ 153 u_short va_mode; /* files access mode and type */ 154 short va_nlink; /* number of references to file */ 155 uid_t va_uid; /* owner user id */ 156 gid_t va_gid; /* owner group id */ 157 long va_fsid; /* file system id (dev for now) */ 158 long va_fileid; /* file id */ 159 u_quad_t va_size; /* file size in bytes */ 160 long va_blocksize; /* blocksize preferred for i/o */ 161 struct timespec va_atime; /* time of last access */ 162 struct timespec va_mtime; /* time of last modification */ 163 struct timespec va_ctime; /* time file changed */ 164 u_long va_gen; /* generation number of file */ 165 u_long va_flags; /* flags defined for file */ 166 dev_t va_rdev; /* device the special file represents */ 167 u_quad_t va_bytes; /* bytes of disk space held by file */ 168 u_quad_t va_filerev; /* file modification number */ 169 u_int va_vaflags; /* operations flags, see below */ 170 long va_spare; /* remain quad aligned */ 171}; 172 173/* 174 * Flags for va_vaflags. 175 */ 176#define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */ 177#define VA_EXCLUSIVE 0x02 /* exclusive create request */ 178 179/* 180 * Flags for ioflag. 181 */ 182#define IO_UNIT 0x01 /* do I/O as atomic unit */ 183#define IO_APPEND 0x02 /* append write to end */ 184#define IO_SYNC 0x04 /* do I/O synchronously */ 185#define IO_NODELOCKED 0x08 /* underlying node already locked */ 186#define IO_NDELAY 0x10 /* FNDELAY flag set in file table */ 187#define IO_VMIO 0x20 /* data already in VMIO space */ 188 189/* 190 * Modes. Some values same as Ixxx entries from inode.h for now. 191 */ 192#define VSUID 04000 /* set user id on execution */ 193#define VSGID 02000 /* set group id on execution */ 194#define VSVTX 01000 /* save swapped text even after use */ 195#define VREAD 00400 /* read, write, execute permissions */ 196#define VWRITE 00200 197#define VEXEC 00100 198 199/* 200 * Token indicating no attribute value yet assigned. 201 */ 202#define VNOVAL (-1) 203 204#ifdef KERNEL 205/* 206 * Convert between vnode types and inode formats (since POSIX.1 207 * defines mode word of stat structure in terms of inode formats). 208 */ 209extern enum vtype iftovt_tab[]; 210extern int vttoif_tab[]; 211#define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12]) 212#define VTTOIF(indx) (vttoif_tab[(int)(indx)]) 213#define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode)) 214 215/* 216 * Flags to various vnode functions. 217 */ 218#define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */ 219#define FORCECLOSE 0x0002 /* vflush: force file closure */ 220#define WRITECLOSE 0x0004 /* vflush: only close writable files */ 221#define DOCLOSE 0x0008 /* vclean: close active files */ 222#define V_SAVE 0x0001 /* vinvalbuf: sync file first */ 223#define V_SAVEMETA 0x0002 /* vinvalbuf: leave indirect blocks */ 224#define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */ 225 226#define VREF(vp) vref(vp) 227 228#ifdef DIAGNOSTIC 229#define VATTR_NULL(vap) vattr_null(vap) 230#else 231#define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */ 232#endif /* DIAGNOSTIC */ 233 234#define NULLVP ((struct vnode *)NULL) 235 236#ifdef VFS_LKM 237#define VNODEOP_SET(f) DATA_SET(MODVNOPS,f) 238#else 239#define VNODEOP_SET(f) DATA_SET(vfs_opv_descs_,f) 240#endif 241 242/* 243 * Global vnode data. 244 */ 245extern struct vnode *rootvnode; /* root (i.e. "/") vnode */ 246extern int desiredvnodes; /* number of vnodes desired */ 247extern int prtactive; /* nonzero to call vprint() */ 248extern struct vattr va_null; /* predefined null vattr structure */ 249 250/* 251 * Macro/function to check for client cache inconsistency w.r.t. leasing. 252 */ 253#define LEASE_READ 0x1 /* Check lease for readers */ 254#define LEASE_WRITE 0x2 /* Check lease for modifiers */ 255 256 257extern void (*lease_updatetime) __P((int deltat)); 258 259#ifdef NFS 260#ifdef NQNFS 261#define LEASE_UPDATETIME(dt) lease_updatetime(dt) 262#else 263#define LEASE_UPDATETIME(dt) 264#endif /* NQNFS */ 265#else 266#define LEASE_UPDATETIME(dt) \ 267 do { if(lease_updatetime) lease_updatetime(dt); } while(0) 268#endif /* NFS */ 269 270#define VSHOULDFREE(vp) \ 271 (!((vp)->v_flag & (VFREE|VDOOMED)) && \ 272 !(vp)->v_holdcnt && !(vp)->v_usecount) 273 274#define VSHOULDBUSY(vp) \ 275 (((vp)->v_flag & VFREE) && \ 276 ((vp)->v_holdcnt || (vp)->v_usecount)) 277 278 279#endif /* KERNEL */ 280 281 282/* 283 * Mods for extensibility. 284 */ 285 286/* 287 * Flags for vdesc_flags: 288 */ 289#define VDESC_MAX_VPS 16 290/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */ 291#define VDESC_VP0_WILLRELE 0x0001 292#define VDESC_VP1_WILLRELE 0x0002 293#define VDESC_VP2_WILLRELE 0x0004 294#define VDESC_VP3_WILLRELE 0x0008 295#define VDESC_NOMAP_VPP 0x0100 296#define VDESC_VPP_WILLRELE 0x0200 297 298/* 299 * VDESC_NO_OFFSET is used to identify the end of the offset list 300 * and in places where no such field exists. 301 */ 302#define VDESC_NO_OFFSET -1 303 304/* 305 * This structure describes the vnode operation taking place. 306 */ 307struct vnodeop_desc { 308 int vdesc_offset; /* offset in vector--first for speed */ 309 char *vdesc_name; /* a readable name for debugging */ 310 int vdesc_flags; /* VDESC_* flags */ 311 312 /* 313 * These ops are used by bypass routines to map and locate arguments. 314 * Creds and procs are not needed in bypass routines, but sometimes 315 * they are useful to (for example) transport layers. 316 * Nameidata is useful because it has a cred in it. 317 */ 318 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */ 319 int vdesc_vpp_offset; /* return vpp location */ 320 int vdesc_cred_offset; /* cred location, if any */ 321 int vdesc_proc_offset; /* proc location, if any */ 322 int vdesc_componentname_offset; /* if any */ 323 /* 324 * Finally, we've got a list of private data (about each operation) 325 * for each transport layer. (Support to manage this list is not 326 * yet part of BSD.) 327 */ 328 caddr_t *vdesc_transports; 329}; 330 331#ifdef KERNEL 332/* 333 * A list of all the operation descs. 334 */ 335extern struct vnodeop_desc *vnodeop_descs[]; 336 337/* 338 * Interlock for scanning list of vnodes attached to a mountpoint 339 */ 340extern struct simplelock mntvnode_slock; 341 342/* 343 * This macro is very helpful in defining those offsets in the vdesc struct. 344 * 345 * This is stolen from X11R4. I ignored all the fancy stuff for 346 * Crays, so if you decide to port this to such a serious machine, 347 * you might want to consult Intrinsic.h's XtOffset{,Of,To}. 348 */ 349#define VOPARG_OFFSET(p_type,field) \ 350 ((int) (((char *) (&(((p_type)NULL)->field))) - ((char *) NULL))) 351#define VOPARG_OFFSETOF(s_type,field) \ 352 VOPARG_OFFSET(s_type*,field) 353#define VOPARG_OFFSETTO(S_TYPE,S_OFFSET,STRUCT_P) \ 354 ((S_TYPE)(((char*)(STRUCT_P))+(S_OFFSET))) 355 356 357/* 358 * This structure is used to configure the new vnodeops vector. 359 */ 360struct vnodeopv_entry_desc { 361 struct vnodeop_desc *opve_op; /* which operation this is */ 362 vop_t *opve_impl; /* code implementing this operation */ 363}; 364struct vnodeopv_desc { 365 /* ptr to the ptr to the vector where op should go */ 366 vop_t ***opv_desc_vector_p; 367 struct vnodeopv_entry_desc *opv_desc_ops; /* null terminated list */ 368}; 369 370/* 371 * A default routine which just returns an error. 372 */ 373int vn_default_error __P((void)); 374 375/* 376 * A generic structure. 377 * This can be used by bypass routines to identify generic arguments. 378 */ 379struct vop_generic_args { 380 struct vnodeop_desc *a_desc; 381 /* other random data follows, presumably */ 382}; 383 384#ifdef DEBUG_VFS_LOCKS 385/* 386 * Macros to aid in tracing VFS locking problems. Not totally 387 * reliable since if the process sleeps between changing the lock 388 * state and checking it with the assert, some other process could 389 * change the state. They are good enough for debugging a single 390 * filesystem using a single-threaded test. I find that 'cvs co src' 391 * is a pretty good test. 392 */ 393 394/* 395 * [dfr] Kludge until I get around to fixing all the vfs locking. 396 */ 397#define IS_LOCKING_VFS(vp) ((vp)->v_tag == VT_UFS \ 398 || (vp)->v_tag == VT_MFS \ 399 || (vp)->v_tag == VT_NFS \ 400 || (vp)->v_tag == VT_LFS \ 401 || (vp)->v_tag == VT_ISOFS \ 402 || (vp)->v_tag == VT_MSDOSFS \ 403 || (vp)->v_tag == VT_DEVFS) 404 405#define ASSERT_VOP_LOCKED(vp, str) \ 406 if ((vp) && IS_LOCKING_VFS(vp) && !VOP_ISLOCKED(vp)) { \ 407 panic("%s: %x is not locked but should be", str, vp); \ 408 } 409 410#define ASSERT_VOP_UNLOCKED(vp, str) \ 411 if ((vp) && IS_LOCKING_VFS(vp) && VOP_ISLOCKED(vp)) { \ 412 panic("%s: %x is locked but shouldn't be", str, vp); \ 413 } 414 415#else 416 417#define ASSERT_VOP_LOCKED(vp, str) 418#define ASSERT_VOP_UNLOCKED(vp, str) 419 420#endif 421 422/* 423 * VOCALL calls an op given an ops vector. We break it out because BSD's 424 * vclean changes the ops vector and then wants to call ops with the old 425 * vector. 426 */ 427#define VOCALL(OPSV,OFF,AP) (( *((OPSV)[(OFF)])) (AP)) 428 429/* 430 * This call works for vnodes in the kernel. 431 */ 432#define VCALL(VP,OFF,AP) VOCALL((VP)->v_op,(OFF),(AP)) 433#define VDESC(OP) (& __CONCAT(OP,_desc)) 434#define VOFFSET(OP) (VDESC(OP)->vdesc_offset) 435 436/* 437 * Finally, include the default set of vnode operations. 438 */ 439#include "vnode_if.h" 440 441/* 442 * Public vnode manipulation functions. 443 */ 444struct componentname; 445struct file; 446struct mount; 447struct nameidata; 448struct ostat; 449struct proc; 450struct stat; 451struct ucred; 452struct uio; 453struct vattr; 454struct vnode; 455struct vop_bwrite_args; 456 457extern int (*lease_check_hook) __P((struct vop_lease_args *)); 458 459int bdevvp __P((dev_t dev, struct vnode **vpp)); 460/* cache_* may belong in namei.h. */ 461void cache_enter __P((struct vnode *dvp, struct vnode *vp, 462 struct componentname *cnp)); 463int cache_lookup __P((struct vnode *dvp, struct vnode **vpp, 464 struct componentname *cnp)); 465void cache_purge __P((struct vnode *vp)); 466void cache_purgevfs __P((struct mount *mp)); 467void cvtstat __P((struct stat *st, struct ostat *ost)); 468int getnewvnode __P((enum vtagtype tag, 469 struct mount *mp, vop_t **vops, struct vnode **vpp)); 470void insmntque __P((struct vnode *vp, struct mount *mp)); 471int lease_check __P((struct vop_lease_args *ap)); 472void vattr_null __P((struct vattr *vap)); 473void vbusy __P((struct vnode *)); 474int vcount __P((struct vnode *vp)); 475void vdrop __P((struct vnode *)); 476int vfinddev __P((dev_t dev, enum vtype type, struct vnode **vpp)); 477void vfree __P((struct vnode *)); 478void vfs_opv_init __P((struct vnodeopv_desc **them)); 479int vflush __P((struct mount *mp, struct vnode *skipvp, int flags)); 480int vget __P((struct vnode *vp, int lockflag, struct proc *p)); 481void vgone __P((struct vnode *vp)); 482void vhold __P((struct vnode *)); 483int vinvalbuf __P((struct vnode *vp, int save, struct ucred *cred, 484 struct proc *p, int slpflag, int slptimeo)); 485void vprint __P((char *label, struct vnode *vp)); 486int vrecycle __P((struct vnode *vp, struct simplelock *inter_lkp, 487 struct proc *p)); 488int vn_bwrite __P((struct vop_bwrite_args *ap)); 489int vn_close __P((struct vnode *vp, 490 int flags, struct ucred *cred, struct proc *p)); 491int vn_lock __P((struct vnode *vp, int flags, struct proc *p)); 492int vn_open __P((struct nameidata *ndp, int fmode, int cmode)); 493int vn_rdwr __P((enum uio_rw rw, struct vnode *vp, caddr_t base, 494 int len, off_t offset, enum uio_seg segflg, int ioflg, 495 struct ucred *cred, int *aresid, struct proc *p)); 496int vn_stat __P((struct vnode *vp, struct stat *sb, struct proc *p)); 497int vfs_cache_lookup __P((struct vop_lookup_args *ap)); 498int vfs_object_create __P((struct vnode *vp, struct proc *p, 499 struct ucred *cred, int waslocked)); 500int vn_writechk __P((struct vnode *vp)); 501int vop_noislocked __P((struct vop_islocked_args *)); 502int vop_nolock __P((struct vop_lock_args *)); 503int vop_nounlock __P((struct vop_unlock_args *)); 504int vop_sharedlock __P((struct vop_lock_args *)); 505int vop_revoke __P((struct vop_revoke_args *)); 506struct vnode * 507 checkalias __P((struct vnode *vp, dev_t nvp_rdev, struct mount *mp)); 508void vput __P((struct vnode *vp)); 509void vref __P((struct vnode *vp)); 510void vrele __P((struct vnode *vp)); 511#endif /* KERNEL */ 512 513#endif /* !_SYS_VNODE_H_ */ 514