| 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 | * $FreeBSD: src/sys/sys/vnode.h,v 1.111.2.19 2002/12/29 18:19:53 dillon Exp $ |
| 35 | * $DragonFly: src/sys/sys/vnode.h,v 1.59 2006/06/02 04:59:53 dillon Exp $ |
| 36 | */ |
| 37 | |
| 38 | #ifndef _SYS_VNODE_H_ |
| 39 | #define _SYS_VNODE_H_ |
| 40 | |
| 41 | #if defined(_KERNEL) || defined(_KERNEL_STRUCTURES) |
| 42 | |
| 43 | #ifndef _SYS_QUEUE_H_ |
| 44 | #include <sys/queue.h> |
| 45 | #endif |
| 46 | #ifndef _SYS_LOCK_H_ |
| 47 | #include <sys/lock.h> |
| 48 | #endif |
| 49 | #ifndef _SYS_SELECT_H_ |
| 50 | #include <sys/select.h> |
| 51 | #endif |
| 52 | #ifndef _SYS_BIOTRACK_H_ |
| 53 | #include <sys/biotrack.h> |
| 54 | #endif |
| 55 | #ifndef _SYS_UIO_H_ |
| 56 | #include <sys/uio.h> |
| 57 | #endif |
| 58 | #ifndef _SYS_ACL_H_ |
| 59 | #include <sys/acl.h> |
| 60 | #endif |
| 61 | #ifndef _SYS_NAMECACHE_H_ |
| 62 | #include <sys/namecache.h> |
| 63 | #endif |
| 64 | #ifndef _SYS_THREAD_H_ |
| 65 | #include <sys/thread.h> |
| 66 | #endif |
| 67 | #ifndef _SYS_VFSOPS_H_ |
| 68 | #include <sys/vfsops.h> |
| 69 | #endif |
| 70 | #ifndef _SYS_VFSCACHE_H_ |
| 71 | #include <sys/vfscache.h> |
| 72 | #endif |
| 73 | #ifndef _SYS_TREE_H_ |
| 74 | #include <sys/tree.h> |
| 75 | #endif |
| 76 | #ifndef _MACHINE_LOCK_H_ |
| 77 | #include <machine/lock.h> |
| 78 | #endif |
| 79 | |
| 80 | /* |
| 81 | * The vnode is the focus of all file activity in UNIX. There is a |
| 82 | * unique vnode allocated for each active file, each current directory, |
| 83 | * each mounted-on file, text file, and the root. |
| 84 | */ |
| 85 | |
| 86 | /* |
| 87 | * Each underlying filesystem allocates its own private area and hangs |
| 88 | * it from v_data. If non-null, this area is freed in getnewvnode(). |
| 89 | */ |
| 90 | TAILQ_HEAD(buflists, buf); |
| 91 | |
| 92 | /* |
| 93 | * Range locks protect offset ranges in files and directories at a high |
| 94 | * level, allowing the actual I/O to be broken down into smaller pieces. |
| 95 | * Range locks will eventually be integrated into the clustered cache |
| 96 | * coherency infrastructure. |
| 97 | * |
| 98 | * We use a simple data structure for now, but eventually this should |
| 99 | * probably be a btree or red-black tree. |
| 100 | */ |
| 101 | struct vrangelock; |
| 102 | |
| 103 | TAILQ_HEAD(vrangelock_list, vrangelock); |
| 104 | |
| 105 | struct vrangehead { |
| 106 | struct vrangelock_list vh_list; |
| 107 | }; |
| 108 | |
| 109 | struct vrangelock { |
| 110 | TAILQ_ENTRY(vrangelock) vr_node; |
| 111 | int vr_flags; |
| 112 | off_t vr_offset; |
| 113 | off_t vr_length; |
| 114 | }; |
| 115 | |
| 116 | #define RNGL_WAITING 0x0001 /* waiting for lock, else has lock */ |
| 117 | #define RNGL_CHECK 0x0002 /* check for work on unlock */ |
| 118 | #define RNGL_SHARED 0x0004 /* shared lock, else exclusive */ |
| 119 | #define RNGL_ONLIST 0x0008 /* sanity check */ |
| 120 | |
| 121 | static __inline |
| 122 | void |
| 123 | vrange_init(struct vrangelock *vr, int flags, off_t offset, off_t length) |
| 124 | { |
| 125 | vr->vr_flags = flags; |
| 126 | vr->vr_offset = offset; |
| 127 | vr->vr_length = length; |
| 128 | } |
| 129 | |
| 130 | #ifdef _KERNEL |
| 131 | |
| 132 | void vrange_lock(struct vnode *vp, struct vrangelock *vr); |
| 133 | void vrange_unlock(struct vnode *vp, struct vrangelock *vr); |
| 134 | |
| 135 | static __inline |
| 136 | void |
| 137 | vrange_lock_shared(struct vnode *vp, struct vrangelock *vr, |
| 138 | off_t offset, off_t length) |
| 139 | { |
| 140 | vrange_init(vr, RNGL_SHARED, offset, length); |
| 141 | vrange_lock(vp, vr); |
| 142 | } |
| 143 | |
| 144 | static __inline |
| 145 | void |
| 146 | vrange_lock_excl(struct vnode *vp, struct vrangelock *vr, |
| 147 | off_t offset, off_t length) |
| 148 | { |
| 149 | vrange_init(vr, 0, offset, length); |
| 150 | vrange_lock(vp, vr); |
| 151 | } |
| 152 | |
| 153 | #endif |
| 154 | |
| 155 | /* |
| 156 | * The vnode infrastructure is being reorgranized. Most reference-related |
| 157 | * fields are locked by the BGL, and most file I/O related operations and |
| 158 | * vnode teardown functions are locked by the vnode lock. |
| 159 | * |
| 160 | * File read operations require a shared lock, file write operations require |
| 161 | * an exclusive lock. Most directory operations (read or write) currently |
| 162 | * require an exclusive lock due to the side effects stored in the directory |
| 163 | * inode (which we intend to fix). |
| 164 | * |
| 165 | * File reads and writes are further protected by a range lock. The intention |
| 166 | * is to be able to break I/O operations down into more easily managed pieces |
| 167 | * so vm_page arrays can be passed through rather then UIOs. This work will |
| 168 | * occur in multiple stages. The range locks will also eventually be used to |
| 169 | * deal with clustered cache coherency issues and, more immediately, to |
| 170 | * protect operations associated with the kernel-managed journaling module. |
| 171 | * |
| 172 | * NOTE: The vnode operations vector, v_ops, is a double-indirect that |
| 173 | * typically points to &v_mount->mnt_vn_use_ops. We use a double |
| 174 | * pointer because mnt_vn_use_ops may change dynamically when e.g. |
| 175 | * journaling is turned on or off. |
| 176 | * |
| 177 | * NOTE: v_filesize is currently only applicable when a VM object is |
| 178 | * associated with the vnode. Otherwise it will be set to NOOFFSET. |
| 179 | */ |
| 180 | RB_HEAD(buf_rb_tree, buf); |
| 181 | RB_HEAD(buf_rb_hash, buf); |
| 182 | |
| 183 | struct vnode { |
| 184 | int v_flag; /* vnode flags (see below) */ |
| 185 | int v_usecount; /* reference count of users */ |
| 186 | int v_writecount; |
| 187 | int v_holdcnt; /* page & buffer references */ |
| 188 | int v_opencount; /* number of explicit opens */ |
| 189 | struct bio_track v_track_read; /* track I/O's in progress */ |
| 190 | struct bio_track v_track_write; /* track I/O's in progress */ |
| 191 | struct mount *v_mount; /* ptr to vfs we are in */ |
| 192 | struct vop_ops **v_ops; /* vnode operations vector */ |
| 193 | TAILQ_ENTRY(vnode) v_freelist; /* vnode freelist */ |
| 194 | TAILQ_ENTRY(vnode) v_nmntvnodes; /* vnodes for mount point */ |
| 195 | struct buf_rb_tree v_rbclean_tree; /* RB tree of clean bufs */ |
| 196 | struct buf_rb_tree v_rbdirty_tree; /* RB tree of dirty bufs */ |
| 197 | struct buf_rb_hash v_rbhash_tree; /* RB tree general lookup */ |
| 198 | LIST_ENTRY(vnode) v_synclist; /* vnodes with dirty buffers */ |
| 199 | enum vtype v_type; /* vnode type */ |
| 200 | union { |
| 201 | struct mount *vu_mountedhere;/* ptr to mounted vfs (VDIR) */ |
| 202 | struct socket *vu_socket; /* unix ipc (VSOCK) */ |
| 203 | struct { |
| 204 | udev_t vu_udev; /* device number for attach */ |
| 205 | struct specinfo *vu_specinfo; /* device (VCHR, VBLK) */ |
| 206 | SLIST_ENTRY(vnode) vu_specnext; |
| 207 | } vu_spec; |
| 208 | struct fifoinfo *vu_fifoinfo; /* fifo (VFIFO) */ |
| 209 | } v_un; |
| 210 | off_t v_filesize; /* file EOF or NOOFFSET */ |
| 211 | off_t v_lazyw; /* lazy write iterator */ |
| 212 | off_t v_lastw; /* last write (write cluster) */ |
| 213 | off_t v_cstart; /* start block of cluster */ |
| 214 | off_t v_lasta; /* last allocation */ |
| 215 | int v_clen; /* length of current cluster */ |
| 216 | struct vm_object *v_object; /* Place to store VM object */ |
| 217 | struct lock v_lock; /* file/dir ops lock */ |
| 218 | enum vtagtype v_tag; /* type of underlying data */ |
| 219 | void *v_data; /* private data for fs */ |
| 220 | struct namecache_list v_namecache; /* associated nc entries */ |
| 221 | struct { |
| 222 | struct lwkt_token vpi_token; /* lock to protect below */ |
| 223 | struct selinfo vpi_selinfo; /* identity of poller(s) */ |
| 224 | short vpi_events; /* what they are looking for */ |
| 225 | short vpi_revents; /* what has happened */ |
| 226 | } v_pollinfo; |
| 227 | struct vmresident *v_resident; /* optional vmresident */ |
| 228 | struct vrangehead v_range; /* range lock */ |
| 229 | #ifdef DEBUG_LOCKS |
| 230 | const char *filename; /* Source file doing locking */ |
| 231 | int line; /* Line number doing locking */ |
| 232 | #endif |
| 233 | void *v_xaddr; |
| 234 | }; |
| 235 | #define v_mountedhere v_un.vu_mountedhere |
| 236 | #define v_socket v_un.vu_socket |
| 237 | #define v_udev v_un.vu_spec.vu_udev |
| 238 | #define v_rdev v_un.vu_spec.vu_specinfo |
| 239 | #define v_specnext v_un.vu_spec.vu_specnext |
| 240 | #define v_fifoinfo v_un.vu_fifoinfo |
| 241 | #define v_spinlock v_lock.lk_spinlock |
| 242 | |
| 243 | #define VN_POLLEVENT(vp, events) \ |
| 244 | do { \ |
| 245 | if ((vp)->v_pollinfo.vpi_events & (events)) \ |
| 246 | vn_pollevent((vp), (events)); \ |
| 247 | } while (0) |
| 248 | |
| 249 | /* |
| 250 | * Vnode flags. |
| 251 | */ |
| 252 | #define VROOT 0x00001 /* root of its file system */ |
| 253 | #define VTEXT 0x00002 /* vnode is a pure text prototype */ |
| 254 | #define VSYSTEM 0x00004 /* vnode being used by kernel */ |
| 255 | #define VISTTY 0x00008 /* vnode represents a tty */ |
| 256 | #define VCTTYISOPEN 0x00010 /* controlling terminal tty is open */ |
| 257 | #define VCKPT 0x00020 /* checkpoint-restored vnode */ |
| 258 | #define VFSMID 0x00040 /* request FSMID update */ |
| 259 | #define VMAYHAVELOCKS 0x00080 /* there may be posix or flock locks on vp */ |
| 260 | /* open for business 0x00100 */ |
| 261 | /* open for business 0x00200 */ |
| 262 | /* open for business 0x00400 */ |
| 263 | /* open for business 0x00800 */ |
| 264 | /* open for business 0x01000 */ |
| 265 | #define VOBJBUF 0x02000 /* Allocate buffers in VM object */ |
| 266 | #define VINACTIVE 0x04000 /* The vnode is inactive */ |
| 267 | #define VAGE 0x08000 /* Insert vnode at head of free list */ |
| 268 | #define VOLOCK 0x10000 /* vnode is locked waiting for an object */ |
| 269 | #define VOWANT 0x20000 /* a process is waiting for VOLOCK */ |
| 270 | #define VRECLAIMED 0x40000 /* This vnode has been destroyed */ |
| 271 | #define VFREE 0x80000 /* This vnode is on the freelist */ |
| 272 | /* open for business 0x100000 */ |
| 273 | #define VONWORKLST 0x200000 /* On syncer work-list */ |
| 274 | #define VMOUNT 0x400000 /* Mount in progress */ |
| 275 | #define VOBJDIRTY 0x800000 /* object might be dirty */ |
| 276 | |
| 277 | /* |
| 278 | * vmntvnodescan() flags |
| 279 | */ |
| 280 | #define VMSC_GETVP 1 |
| 281 | #define VMSC_GETVX 2 |
| 282 | #define VMSC_REFVP 3 |
| 283 | #define VMSC_NOWAIT 0x10 |
| 284 | |
| 285 | /* |
| 286 | * Flags for ioflag. (high 16 bits used to ask for read-ahead and |
| 287 | * help with write clustering) |
| 288 | */ |
| 289 | #define IO_UNIT 0x0001 /* do I/O as atomic unit */ |
| 290 | #define IO_APPEND 0x0002 /* append write to end */ |
| 291 | #define IO_SYNC 0x0004 /* do I/O synchronously */ |
| 292 | #define IO_NODELOCKED 0x0008 /* underlying node already locked */ |
| 293 | #define IO_NDELAY 0x0010 /* FNDELAY flag set in file table */ |
| 294 | #define IO_VMIO 0x0020 /* data already in VMIO space */ |
| 295 | #define IO_INVAL 0x0040 /* invalidate after I/O */ |
| 296 | #define IO_ASYNC 0x0080 /* bawrite rather then bdwrite */ |
| 297 | #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */ |
| 298 | #define IO_NOWDRAIN 0x0200 /* do not block on wdrain */ |
| 299 | #define IO_CORE 0x0400 /* I/O is part of core dump */ |
| 300 | |
| 301 | #define IO_SEQMAX 0x7F /* seq heuristic max value */ |
| 302 | #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */ |
| 303 | |
| 304 | /* |
| 305 | * Modes. Note that these V-modes must match file S_I*USR, SUID, SGID, |
| 306 | * and SVTX flag bits. |
| 307 | * |
| 308 | * VCREATE, VDELETE, and VEXCL may only be used in naccess() calls. |
| 309 | */ |
| 310 | #define VDELETE 040000 /* delete if the file/dir exists */ |
| 311 | #define VCREATE 020000 /* create if the file/dir does not exist */ |
| 312 | #define VEXCL 010000 /* error if the file/dir already exists */ |
| 313 | |
| 314 | #define VSUID 04000 /* set user id on execution */ |
| 315 | #define VSGID 02000 /* set group id on execution */ |
| 316 | #define VSVTX 01000 /* save swapped text even after use */ |
| 317 | #define VREAD 00400 /* read, write, execute permissions */ |
| 318 | #define VWRITE 00200 |
| 319 | #define VEXEC 00100 |
| 320 | |
| 321 | /* |
| 322 | * Token indicating no attribute value yet assigned. |
| 323 | */ |
| 324 | #define VNOVAL (-1) |
| 325 | |
| 326 | /* |
| 327 | * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon) |
| 328 | */ |
| 329 | #define VLKTIMEOUT (hz / 20 + 1) |
| 330 | |
| 331 | #ifdef _KERNEL |
| 332 | |
| 333 | /* |
| 334 | * Convert between vnode types and inode formats (since POSIX.1 |
| 335 | * defines mode word of stat structure in terms of inode formats). |
| 336 | */ |
| 337 | extern enum vtype iftovt_tab[]; |
| 338 | extern int vttoif_tab[]; |
| 339 | #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12]) |
| 340 | #define VTTOIF(indx) (vttoif_tab[(int)(indx)]) |
| 341 | #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode)) |
| 342 | |
| 343 | /* |
| 344 | * Flags to various vnode functions. |
| 345 | */ |
| 346 | #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */ |
| 347 | #define FORCECLOSE 0x0002 /* vflush: force file closure */ |
| 348 | #define WRITECLOSE 0x0004 /* vflush: only close writable files */ |
| 349 | #define DOCLOSE 0x0008 /* vclean: close active files */ |
| 350 | #define V_SAVE 0x0001 /* vinvalbuf: sync file first */ |
| 351 | #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */ |
| 352 | |
| 353 | #ifdef DIAGNOSTIC |
| 354 | #define VATTR_NULL(vap) vattr_null(vap) |
| 355 | #else |
| 356 | #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */ |
| 357 | #endif /* DIAGNOSTIC */ |
| 358 | |
| 359 | #define NULLVP ((struct vnode *)NULL) |
| 360 | |
| 361 | #define VNODEOP_SET(f) \ |
| 362 | C_SYSINIT(f##init, SI_SUB_VFS, SI_ORDER_SECOND, vfs_add_vnodeops_sysinit, &f); \ |
| 363 | C_SYSUNINIT(f##uninit, SI_SUB_VFS, SI_ORDER_SECOND,vfs_rm_vnodeops_sysinit, &f); |
| 364 | |
| 365 | /* |
| 366 | * Global vnode data. |
| 367 | */ |
| 368 | struct objcache; |
| 369 | |
| 370 | extern struct vnode *rootvnode; /* root (i.e. "/") vnode */ |
| 371 | extern struct namecache *rootncp; /* root (i.e. "/") namecache */ |
| 372 | extern int desiredvnodes; /* number of vnodes desired */ |
| 373 | extern time_t syncdelay; /* max time to delay syncing data */ |
| 374 | extern time_t filedelay; /* time to delay syncing files */ |
| 375 | extern time_t dirdelay; /* time to delay syncing directories */ |
| 376 | extern time_t metadelay; /* time to delay syncing metadata */ |
| 377 | extern struct objcache *namei_oc; |
| 378 | extern int prtactive; /* nonzero to call vprint() */ |
| 379 | extern struct vattr va_null; /* predefined null vattr structure */ |
| 380 | extern int vfs_ioopt; |
| 381 | extern int numvnodes; |
| 382 | extern int freevnodes; |
| 383 | extern int vfs_fastdev; /* fast specfs device access */ |
| 384 | |
| 385 | #endif /* _KERNEL */ |
| 386 | |
| 387 | /* |
| 388 | * Mods for extensibility. |
| 389 | */ |
| 390 | |
| 391 | /* |
| 392 | * Flags for vdesc_flags: |
| 393 | */ |
| 394 | #define VDESC_MAX_VPS 8 |
| 395 | /* Low order 8 flag bits are reserved for willrele flags for vp arguments. */ |
| 396 | #define VDESC_VP0_WILLRELE 0x00000001 |
| 397 | #define VDESC_VP1_WILLRELE 0x00000002 |
| 398 | #define VDESC_VP2_WILLRELE 0x00000004 |
| 399 | #define VDESC_VP3_WILLRELE 0x00000008 |
| 400 | #define VDESC_VP4_WILLRELE 0x00000010 |
| 401 | #define VDESC_VP5_WILLRELE 0x00000020 |
| 402 | #define VDESC_VP6_WILLRELE 0x00000040 |
| 403 | #define VDESC_VP7_WILLRELE 0x00000080 |
| 404 | #define VDESC_NOMAP_VPP 0x00000100 |
| 405 | #define VDESC_VPP_WILLRELE 0x00000200 |
| 406 | #define VDESC_VP0_WILLUNLOCK 0x00010000 |
| 407 | #define VDESC_VP1_WILLUNLOCK 0x00020000 |
| 408 | #define VDESC_VP2_WILLUNLOCK 0x00040000 |
| 409 | #define VDESC_VP3_WILLUNLOCK 0x00080000 |
| 410 | #define VDESC_VP4_WILLUNLOCK 0x00100000 |
| 411 | #define VDESC_VP5_WILLUNLOCK 0x00200000 |
| 412 | #define VDESC_VP6_WILLUNLOCK 0x00400000 |
| 413 | #define VDESC_VP7_WILLUNLOCK 0x00800000 |
| 414 | |
| 415 | /* |
| 416 | * VDESC_NO_OFFSET is used to identify the end of the offset list |
| 417 | * and in places where no such field exists. |
| 418 | */ |
| 419 | #define VDESC_NO_OFFSET -1 |
| 420 | |
| 421 | /* |
| 422 | * This structure describes the vnode operation taking place. |
| 423 | */ |
| 424 | struct vnodeop_desc { |
| 425 | int vdesc_offset; /* offset in vector--first for speed */ |
| 426 | char *vdesc_name; /* a readable name for debugging */ |
| 427 | int vdesc_unused01; |
| 428 | int *vdesc_unused02; |
| 429 | int vdesc_unused03; |
| 430 | int vdesc_unused04; |
| 431 | int vdesc_unused05; |
| 432 | int vdesc_unused06; |
| 433 | }; |
| 434 | |
| 435 | #ifdef _KERNEL |
| 436 | /* |
| 437 | * A list of all the operation descs. |
| 438 | */ |
| 439 | extern struct vnodeop_desc *vnodeop_descs[]; |
| 440 | |
| 441 | /* |
| 442 | * Interlock for scanning list of vnodes attached to a mountpoint |
| 443 | */ |
| 444 | extern struct lwkt_token mntvnode_token; |
| 445 | |
| 446 | /* |
| 447 | * This macro is very helpful in defining those offsets in the vdesc struct. |
| 448 | * |
| 449 | * This is stolen from X11R4. I ignored all the fancy stuff for |
| 450 | * Crays, so if you decide to port this to such a serious machine, |
| 451 | * you might want to consult Intrinsic.h's XtOffset{,Of,To}. |
| 452 | */ |
| 453 | #define VOPARG_OFFSET(p_type,field) \ |
| 454 | ((int) (((char *) (&(((p_type)NULL)->field))) - ((char *) NULL))) |
| 455 | #define VOPARG_OFFSETOF(s_type,field) \ |
| 456 | VOPARG_OFFSET(s_type*,field) |
| 457 | #define VOPARG_OFFSETTO(S_TYPE,S_OFFSET,STRUCT_P) \ |
| 458 | ((S_TYPE)(((char*)(STRUCT_P))+(S_OFFSET))) |
| 459 | |
| 460 | typedef int (*vnodeopv_entry_t)(struct vop_generic_args *); |
| 461 | |
| 462 | /* |
| 463 | * This structure is used to configure the new vnodeops vector. The entry |
| 464 | * descriptor describes a patricular VOP function while the operations |
| 465 | * vector descriptor recursively describes arrays of entry descriptors. |
| 466 | */ |
| 467 | struct vnodeopv_entry_desc { |
| 468 | struct vnodeop_desc *opve_op; |
| 469 | vnodeopv_entry_t opve_func; |
| 470 | }; |
| 471 | |
| 472 | struct vnodeopv_desc { |
| 473 | struct vop_ops **opv_desc_vector; /* vect to allocate/fill*/ |
| 474 | struct vnodeopv_entry_desc *opv_desc_ops; /* null terminated list */ |
| 475 | int opv_flags; |
| 476 | }; |
| 477 | |
| 478 | struct vnodeopv_node { |
| 479 | TAILQ_ENTRY(vnodeopv_node) entry; |
| 480 | struct vop_ops *ops; /* allocated vector */ |
| 481 | struct vnodeopv_entry_desc *descs; /* null terminated list */ |
| 482 | }; |
| 483 | |
| 484 | #ifdef DEBUG_VFS_LOCKS |
| 485 | /* |
| 486 | * Macros to aid in tracing VFS locking problems. Not totally |
| 487 | * reliable since if the process sleeps between changing the lock |
| 488 | * state and checking it with the assert, some other process could |
| 489 | * change the state. They are good enough for debugging a single |
| 490 | * filesystem using a single-threaded test. I find that 'cvs co src' |
| 491 | * is a pretty good test. |
| 492 | */ |
| 493 | |
| 494 | #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked(vp, str) |
| 495 | #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked(vp, str); |
| 496 | |
| 497 | void assert_vop_locked(struct vnode *vp, const char *str); |
| 498 | void assert_vop_unlocked(struct vnode *vp, const char *str); |
| 499 | |
| 500 | #else |
| 501 | |
| 502 | #define ASSERT_VOP_LOCKED(vp, str) |
| 503 | #define ASSERT_VOP_UNLOCKED(vp, str) |
| 504 | |
| 505 | #endif /* DEBUG_VFS_LOCKS */ |
| 506 | |
| 507 | /* |
| 508 | * VOCALL calls an op given an ops vector. We break it out because BSD's |
| 509 | * vclean changes the ops vector and then wants to call ops with the old |
| 510 | * vector. |
| 511 | */ |
| 512 | |
| 513 | typedef int (*vocall_func_t)(struct vop_generic_args *); |
| 514 | |
| 515 | /* |
| 516 | * This call executes the vops vector for the offset stored in the ap's |
| 517 | * descriptor of the passed vops rather then the one related to the |
| 518 | * ap's vop_ops structure. It is used to chain VOPS calls on behalf of |
| 519 | * filesystems from a VFS's context ONLY (that is, from a VFS's own vops |
| 520 | * vector function). |
| 521 | */ |
| 522 | #define VOCALL(vops, ap) \ |
| 523 | (*(vocall_func_t *)((char *)(vops)+((ap)->a_desc->vdesc_offset)))(ap) |
| 524 | |
| 525 | #define VDESC(OP) (& __CONCAT(OP,_desc)) |
| 526 | |
| 527 | /* |
| 528 | * Public vnode manipulation functions. |
| 529 | */ |
| 530 | struct file; |
| 531 | struct mount; |
| 532 | struct nlookupdata; |
| 533 | struct proc; |
| 534 | struct thread; |
| 535 | struct stat; |
| 536 | struct ucred; |
| 537 | struct uio; |
| 538 | struct vattr; |
| 539 | struct vnode; |
| 540 | |
| 541 | void addaliasu (struct vnode *vp, udev_t nvp_udev); |
| 542 | int v_associate_rdev(struct vnode *vp, dev_t dev); |
| 543 | void v_release_rdev(struct vnode *vp); |
| 544 | int bdevvp (dev_t dev, struct vnode **vpp); |
| 545 | struct vnode *allocvnode(int lktimeout, int lkflags); |
| 546 | int getnewvnode (enum vtagtype tag, struct mount *mp, |
| 547 | struct vnode **vpp, int timo, int lkflags); |
| 548 | int getspecialvnode (enum vtagtype tag, struct mount *mp, |
| 549 | struct vop_ops **ops, struct vnode **vpp, int timo, |
| 550 | int lkflags); |
| 551 | int spec_vnoperate (struct vop_generic_args *); |
| 552 | int speedup_syncer (void); |
| 553 | void vattr_null (struct vattr *vap); |
| 554 | int vcount (struct vnode *vp); |
| 555 | int vfinddev (dev_t dev, enum vtype type, struct vnode **vpp); |
| 556 | void vfs_add_vnodeops_sysinit (const void *); |
| 557 | void vfs_rm_vnodeops_sysinit (const void *); |
| 558 | void vfs_add_vnodeops(struct mount *, struct vop_ops **, |
| 559 | struct vnodeopv_entry_desc *, int); |
| 560 | void vfs_rm_vnodeops(struct vop_ops **); |
| 561 | int vflush (struct mount *mp, int rootrefs, int flags); |
| 562 | int vmntvnodescan(struct mount *mp, int flags, |
| 563 | int (*fastfunc)(struct mount *mp, struct vnode *vp, void *data), |
| 564 | int (*slowfunc)(struct mount *mp, struct vnode *vp, void *data), |
| 565 | void *data); |
| 566 | void insmntque(struct vnode *vp, struct mount *mp); |
| 567 | |
| 568 | void vclean (struct vnode *vp, int flags); |
| 569 | void vgone (struct vnode *vp); |
| 570 | void vupdatefsmid (struct vnode *vp); |
| 571 | int vinvalbuf (struct vnode *vp, int save, int slpflag, int slptimeo); |
| 572 | int vtruncbuf (struct vnode *vp, off_t length, int blksize); |
| 573 | int vfsync(struct vnode *vp, int waitfor, int passes, |
| 574 | int (*checkdef)(struct buf *), |
| 575 | int (*waitoutput)(struct vnode *, struct thread *)); |
| 576 | int vinitvmio(struct vnode *vp, off_t filesize); |
| 577 | void vprint (char *label, struct vnode *vp); |
| 578 | int vrecycle (struct vnode *vp); |
| 579 | void vn_strategy(struct vnode *vp, struct bio *bio); |
| 580 | int vn_close (struct vnode *vp, int flags); |
| 581 | int vn_isdisk (struct vnode *vp, int *errp); |
| 582 | int vn_lock (struct vnode *vp, int flags); |
| 583 | #ifdef DEBUG_LOCKS |
| 584 | int debug_vn_lock (struct vnode *vp, int flags, |
| 585 | const char *filename, int line); |
| 586 | #define vn_lock(vp,flags) debug_vn_lock(vp, flags, __FILE__, __LINE__) |
| 587 | #endif |
| 588 | |
| 589 | int vn_get_namelen(struct vnode *, int *); |
| 590 | void vn_setspecops (struct file *fp); |
| 591 | int vn_fullpath (struct proc *p, struct vnode *vn, char **retbuf, char **freebuf); |
| 592 | int vn_open (struct nlookupdata *ndp, struct file *fp, int fmode, int cmode); |
| 593 | void vn_pollevent (struct vnode *vp, int events); |
| 594 | void vn_pollgone (struct vnode *vp); |
| 595 | int vn_pollrecord (struct vnode *vp, int events); |
| 596 | int vn_rdwr (enum uio_rw rw, struct vnode *vp, caddr_t base, |
| 597 | int len, off_t offset, enum uio_seg segflg, int ioflg, |
| 598 | struct ucred *cred, int *aresid); |
| 599 | int vn_rdwr_inchunks (enum uio_rw rw, struct vnode *vp, caddr_t base, |
| 600 | int len, off_t offset, enum uio_seg segflg, int ioflg, |
| 601 | struct ucred *cred, int *aresid); |
| 602 | int vn_stat (struct vnode *vp, struct stat *sb, struct ucred *cred); |
| 603 | dev_t vn_todev (struct vnode *vp); |
| 604 | void vfs_timestamp (struct timespec *); |
| 605 | int vn_writechk (struct vnode *vp); |
| 606 | int vop_stdopen (struct vop_open_args *ap); |
| 607 | int vop_stdclose (struct vop_close_args *ap); |
| 608 | int vop_stdislocked (struct vop_islocked_args *ap); |
| 609 | int vop_stdlock (struct vop_lock_args *ap); |
| 610 | int vop_stdrlock (struct vop_lock_args *ap); |
| 611 | int vop_stdunlock (struct vop_unlock_args *ap); |
| 612 | int vop_nopoll (struct vop_poll_args *ap); |
| 613 | int vop_stdpathconf (struct vop_pathconf_args *ap); |
| 614 | int vop_stdpoll (struct vop_poll_args *ap); |
| 615 | int vop_stdrevoke (struct vop_revoke_args *ap); |
| 616 | int vop_eopnotsupp (struct vop_generic_args *ap); |
| 617 | int vop_ebadf (struct vop_generic_args *ap); |
| 618 | int vop_einval (struct vop_generic_args *ap); |
| 619 | int vop_enotty (struct vop_generic_args *ap); |
| 620 | int vop_defaultop (struct vop_generic_args *ap); |
| 621 | int vop_null (struct vop_generic_args *ap); |
| 622 | int vop_panic (struct vop_generic_args *ap); |
| 623 | int vop_write_dirent(int *, struct uio *, ino_t, uint8_t, uint16_t, |
| 624 | const char *); |
| 625 | |
| 626 | int vop_compat_nresolve(struct vop_nresolve_args *ap); |
| 627 | int vop_compat_nlookupdotdot(struct vop_nlookupdotdot_args *ap); |
| 628 | int vop_compat_ncreate(struct vop_ncreate_args *ap); |
| 629 | int vop_compat_nmkdir(struct vop_nmkdir_args *ap); |
| 630 | int vop_compat_nmknod(struct vop_nmknod_args *ap); |
| 631 | int vop_compat_nlink(struct vop_nlink_args *ap); |
| 632 | int vop_compat_nsymlink(struct vop_nsymlink_args *ap); |
| 633 | int vop_compat_nwhiteout(struct vop_nwhiteout_args *ap); |
| 634 | int vop_compat_nremove(struct vop_nremove_args *ap); |
| 635 | int vop_compat_nrmdir(struct vop_nrmdir_args *ap); |
| 636 | int vop_compat_nrename(struct vop_nrename_args *ap); |
| 637 | |
| 638 | int vx_lock (struct vnode *vp); |
| 639 | void vx_unlock (struct vnode *vp); |
| 640 | int vx_get (struct vnode *vp); |
| 641 | int vx_get_nonblock (struct vnode *vp); |
| 642 | void vx_put (struct vnode *vp); |
| 643 | int vget (struct vnode *vp, int lockflag); |
| 644 | void vput (struct vnode *vp); |
| 645 | void vhold (struct vnode *); |
| 646 | void vdrop (struct vnode *); |
| 647 | void vref (struct vnode *vp); |
| 648 | void vrele (struct vnode *vp); |
| 649 | void vsetflags (struct vnode *vp, int flags); |
| 650 | void vclrflags (struct vnode *vp, int flags); |
| 651 | |
| 652 | void vfs_subr_init(void); |
| 653 | void vfs_mount_init(void); |
| 654 | void vfs_lock_init(void); |
| 655 | void vfs_sync_init(void); |
| 656 | |
| 657 | void vn_syncer_add_to_worklist(struct vnode *, int); |
| 658 | void vnlru_proc_wait(void); |
| 659 | |
| 660 | extern struct vop_ops *default_vnode_vops; |
| 661 | extern struct vop_ops *spec_vnode_vops; |
| 662 | extern struct vop_ops *dead_vnode_vops; |
| 663 | |
| 664 | #endif /* _KERNEL */ |
| 665 | |
| 666 | #endif /* _KERNEL || _KERNEL_STRUCTURES */ |
| 667 | #endif /* !_SYS_VNODE_H_ */ |