vfs_vnops.c revision 1.145
1 /* $NetBSD: vfs_vnops.c,v 1.145 2007/11/29 18:07:11 ad Exp $ */ 2 3 /* 4 * Copyright (c) 1982, 1986, 1989, 1993 5 * The Regents of the University of California. All rights reserved. 6 * (c) UNIX System Laboratories, Inc. 7 * All or some portions of this file are derived from material licensed 8 * to the University of California by American Telephone and Telegraph 9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 10 * the permission of UNIX System Laboratories, Inc. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 3. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * @(#)vfs_vnops.c 8.14 (Berkeley) 6/15/95 37 */ 38 39 #include <sys/cdefs.h> 40 __KERNEL_RCSID(0, "$NetBSD: vfs_vnops.c,v 1.145 2007/11/29 18:07:11 ad Exp $"); 41 42 #include "fs_union.h" 43 #include "veriexec.h" 44 45 #include <sys/param.h> 46 #include <sys/systm.h> 47 #include <sys/kernel.h> 48 #include <sys/file.h> 49 #include <sys/stat.h> 50 #include <sys/buf.h> 51 #include <sys/proc.h> 52 #include <sys/malloc.h> 53 #include <sys/mount.h> 54 #include <sys/namei.h> 55 #include <sys/vnode.h> 56 #include <sys/ioctl.h> 57 #include <sys/tty.h> 58 #include <sys/poll.h> 59 #include <sys/kauth.h> 60 #include <sys/syslog.h> 61 #include <sys/fstrans.h> 62 #include <sys/atomic.h> 63 64 #include <miscfs/specfs/specdev.h> 65 66 #include <uvm/uvm_extern.h> 67 #include <uvm/uvm_readahead.h> 68 69 #ifdef UNION 70 #include <fs/union/union.h> 71 #endif 72 73 #if defined(LKM) || defined(UNION) 74 int (*vn_union_readdir_hook) (struct vnode **, struct file *, struct lwp *); 75 #endif 76 77 #include <sys/verified_exec.h> 78 79 static int vn_read(struct file *fp, off_t *offset, struct uio *uio, 80 kauth_cred_t cred, int flags); 81 static int vn_write(struct file *fp, off_t *offset, struct uio *uio, 82 kauth_cred_t cred, int flags); 83 static int vn_closefile(struct file *fp, struct lwp *l); 84 static int vn_poll(struct file *fp, int events, struct lwp *l); 85 static int vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l); 86 static int vn_statfile(struct file *fp, struct stat *sb, struct lwp *l); 87 static int vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l); 88 89 const struct fileops vnops = { 90 vn_read, vn_write, vn_ioctl, vn_fcntl, vn_poll, 91 vn_statfile, vn_closefile, vn_kqfilter 92 }; 93 94 /* 95 * Common code for vnode open operations. 96 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 97 */ 98 int 99 vn_open(struct nameidata *ndp, int fmode, int cmode) 100 { 101 struct vnode *vp; 102 struct lwp *l = ndp->ni_cnd.cn_lwp; 103 kauth_cred_t cred = l->l_cred; 104 struct vattr va; 105 int error; 106 char *path; 107 108 ndp->ni_cnd.cn_flags &= TRYEMULROOT; 109 110 if (fmode & O_CREAT) { 111 ndp->ni_cnd.cn_nameiop = CREATE; 112 ndp->ni_cnd.cn_flags |= LOCKPARENT | LOCKLEAF; 113 if ((fmode & O_EXCL) == 0 && 114 ((fmode & O_NOFOLLOW) == 0)) 115 ndp->ni_cnd.cn_flags |= FOLLOW; 116 } else { 117 ndp->ni_cnd.cn_nameiop = LOOKUP; 118 ndp->ni_cnd.cn_flags |= LOCKLEAF; 119 if ((fmode & O_NOFOLLOW) == 0) 120 ndp->ni_cnd.cn_flags |= FOLLOW; 121 } 122 123 VERIEXEC_PATH_GET(ndp->ni_dirp, ndp->ni_segflg, ndp->ni_dirp, path); 124 125 error = namei(ndp); 126 if (error) 127 goto out; 128 129 vp = ndp->ni_vp; 130 131 #if NVERIEXEC > 0 132 error = veriexec_openchk(l, ndp->ni_vp, ndp->ni_dirp, fmode); 133 if (error) 134 goto bad; 135 #endif /* NVERIEXEC > 0 */ 136 137 if (fmode & O_CREAT) { 138 if (ndp->ni_vp == NULL) { 139 VATTR_NULL(&va); 140 va.va_type = VREG; 141 va.va_mode = cmode; 142 if (fmode & O_EXCL) 143 va.va_vaflags |= VA_EXCLUSIVE; 144 VOP_LEASE(ndp->ni_dvp, cred, LEASE_WRITE); 145 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 146 &ndp->ni_cnd, &va); 147 if (error) 148 goto out; 149 fmode &= ~O_TRUNC; 150 vp = ndp->ni_vp; 151 } else { 152 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 153 if (ndp->ni_dvp == ndp->ni_vp) 154 vrele(ndp->ni_dvp); 155 else 156 vput(ndp->ni_dvp); 157 ndp->ni_dvp = NULL; 158 vp = ndp->ni_vp; 159 if (fmode & O_EXCL) { 160 error = EEXIST; 161 goto bad; 162 } 163 fmode &= ~O_CREAT; 164 } 165 } else { 166 vp = ndp->ni_vp; 167 } 168 if (vp->v_type == VSOCK) { 169 error = EOPNOTSUPP; 170 goto bad; 171 } 172 if (ndp->ni_vp->v_type == VLNK) { 173 error = EFTYPE; 174 goto bad; 175 } 176 177 if ((fmode & O_CREAT) == 0) { 178 if (fmode & FREAD) { 179 if ((error = VOP_ACCESS(vp, VREAD, cred)) != 0) 180 goto bad; 181 } 182 183 if (fmode & (FWRITE | O_TRUNC)) { 184 if (vp->v_type == VDIR) { 185 error = EISDIR; 186 goto bad; 187 } 188 if ((error = vn_writechk(vp)) != 0 || 189 (error = VOP_ACCESS(vp, VWRITE, cred)) != 0) 190 goto bad; 191 } 192 } 193 194 if (fmode & O_TRUNC) { 195 VOP_UNLOCK(vp, 0); /* XXX */ 196 197 VOP_LEASE(vp, cred, LEASE_WRITE); 198 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */ 199 VATTR_NULL(&va); 200 va.va_size = 0; 201 error = VOP_SETATTR(vp, &va, cred); 202 if (error != 0) 203 goto bad; 204 } 205 if ((error = VOP_OPEN(vp, fmode, cred)) != 0) 206 goto bad; 207 if (fmode & FWRITE) { 208 simple_lock(&vp->v_interlock); 209 vp->v_writecount++; 210 simple_unlock(&vp->v_interlock); 211 } 212 213 bad: 214 if (error) 215 vput(vp); 216 out: 217 VERIEXEC_PATH_PUT(path); 218 return (error); 219 } 220 221 /* 222 * Check for write permissions on the specified vnode. 223 * Prototype text segments cannot be written. 224 */ 225 int 226 vn_writechk(struct vnode *vp) 227 { 228 229 /* 230 * If the vnode is in use as a process's text, 231 * we can't allow writing. 232 */ 233 if (vp->v_iflag & VI_TEXT) 234 return (ETXTBSY); 235 return (0); 236 } 237 238 /* 239 * Mark a vnode as having executable mappings. 240 */ 241 void 242 vn_markexec(struct vnode *vp) 243 { 244 245 LOCK_ASSERT(simple_lock_held(&vp->v_interlock)); 246 247 if ((vp->v_iflag & VI_EXECMAP) == 0) { 248 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 249 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 250 } 251 vp->v_iflag |= VI_EXECMAP; 252 } 253 254 /* 255 * Mark a vnode as being the text of a process. 256 * Fail if the vnode is currently writable. 257 */ 258 int 259 vn_marktext(struct vnode *vp) 260 { 261 262 simple_lock(&vp->v_interlock); 263 if (vp->v_writecount != 0) { 264 KASSERT((vp->v_iflag & VI_TEXT) == 0); 265 simple_unlock(&vp->v_interlock); 266 return (ETXTBSY); 267 } 268 vp->v_iflag |= VI_TEXT; 269 vn_markexec(vp); 270 simple_unlock(&vp->v_interlock); 271 return (0); 272 } 273 274 /* 275 * Vnode close call 276 * 277 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 278 */ 279 int 280 vn_close(struct vnode *vp, int flags, kauth_cred_t cred, struct lwp *l) 281 { 282 int error; 283 284 simple_lock(&vp->v_interlock); 285 if (flags & FWRITE) 286 vp->v_writecount--; 287 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_INTERLOCK); 288 error = VOP_CLOSE(vp, flags, cred); 289 vput(vp); 290 return (error); 291 } 292 293 /* 294 * Package up an I/O request on a vnode into a uio and do it. 295 */ 296 int 297 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 298 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 299 struct lwp *l) 300 { 301 struct uio auio; 302 struct iovec aiov; 303 int error; 304 305 if ((ioflg & IO_NODELOCKED) == 0) { 306 if (rw == UIO_READ) { 307 vn_lock(vp, LK_SHARED | LK_RETRY); 308 } else /* UIO_WRITE */ { 309 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 310 } 311 } 312 auio.uio_iov = &aiov; 313 auio.uio_iovcnt = 1; 314 aiov.iov_base = base; 315 aiov.iov_len = len; 316 auio.uio_resid = len; 317 auio.uio_offset = offset; 318 auio.uio_rw = rw; 319 if (segflg == UIO_SYSSPACE) { 320 UIO_SETUP_SYSSPACE(&auio); 321 } else { 322 auio.uio_vmspace = l->l_proc->p_vmspace; 323 } 324 if (rw == UIO_READ) { 325 error = VOP_READ(vp, &auio, ioflg, cred); 326 } else { 327 error = VOP_WRITE(vp, &auio, ioflg, cred); 328 } 329 if (aresid) 330 *aresid = auio.uio_resid; 331 else 332 if (auio.uio_resid && error == 0) 333 error = EIO; 334 if ((ioflg & IO_NODELOCKED) == 0) { 335 VOP_UNLOCK(vp, 0); 336 } 337 return (error); 338 } 339 340 int 341 vn_readdir(struct file *fp, char *bf, int segflg, u_int count, int *done, 342 struct lwp *l, off_t **cookies, int *ncookies) 343 { 344 struct vnode *vp = (struct vnode *)fp->f_data; 345 struct iovec aiov; 346 struct uio auio; 347 int error, eofflag; 348 349 /* Limit the size on any kernel buffers used by VOP_READDIR */ 350 count = min(MAXBSIZE, count); 351 352 unionread: 353 if (vp->v_type != VDIR) 354 return (EINVAL); 355 aiov.iov_base = bf; 356 aiov.iov_len = count; 357 auio.uio_iov = &aiov; 358 auio.uio_iovcnt = 1; 359 auio.uio_rw = UIO_READ; 360 if (segflg == UIO_SYSSPACE) { 361 UIO_SETUP_SYSSPACE(&auio); 362 } else { 363 KASSERT(l == curlwp); 364 auio.uio_vmspace = l->l_proc->p_vmspace; 365 } 366 auio.uio_resid = count; 367 vn_lock(vp, LK_SHARED | LK_RETRY); 368 auio.uio_offset = fp->f_offset; 369 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 370 ncookies); 371 mutex_enter(&fp->f_lock); 372 fp->f_offset = auio.uio_offset; 373 mutex_exit(&fp->f_lock); 374 VOP_UNLOCK(vp, 0); 375 if (error) 376 return (error); 377 378 #if defined(UNION) || defined(LKM) 379 if (count == auio.uio_resid && vn_union_readdir_hook) { 380 struct vnode *ovp = vp; 381 382 error = (*vn_union_readdir_hook)(&vp, fp, l); 383 if (error) 384 return (error); 385 if (vp != ovp) 386 goto unionread; 387 } 388 #endif /* UNION || LKM */ 389 390 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 391 (vp->v_mount->mnt_flag & MNT_UNION)) { 392 struct vnode *tvp = vp; 393 vp = vp->v_mount->mnt_vnodecovered; 394 VREF(vp); 395 mutex_enter(&fp->f_lock); 396 fp->f_data = vp; 397 fp->f_offset = 0; 398 mutex_exit(&fp->f_lock); 399 vrele(tvp); 400 goto unionread; 401 } 402 *done = count - auio.uio_resid; 403 return error; 404 } 405 406 /* 407 * File table vnode read routine. 408 */ 409 static int 410 vn_read(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 411 int flags) 412 { 413 struct vnode *vp = (struct vnode *)fp->f_data; 414 int count, error, ioflag; 415 416 VOP_LEASE(vp, cred, LEASE_READ); 417 mutex_enter(&fp->f_lock); 418 ioflag = IO_ADV_ENCODE(fp->f_advice); 419 if (fp->f_flag & FNONBLOCK) 420 ioflag |= IO_NDELAY; 421 if ((fp->f_flag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 422 ioflag |= IO_SYNC; 423 if (fp->f_flag & FALTIO) 424 ioflag |= IO_ALTSEMANTICS; 425 if (fp->f_flag & FDIRECT) 426 ioflag |= IO_DIRECT; 427 mutex_exit(&fp->f_lock); 428 vn_lock(vp, LK_SHARED | LK_RETRY); 429 uio->uio_offset = *offset; 430 count = uio->uio_resid; 431 error = VOP_READ(vp, uio, ioflag, cred); 432 if (flags & FOF_UPDATE_OFFSET) 433 *offset += count - uio->uio_resid; 434 VOP_UNLOCK(vp, 0); 435 return (error); 436 } 437 438 /* 439 * File table vnode write routine. 440 */ 441 static int 442 vn_write(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 443 int flags) 444 { 445 struct vnode *vp = (struct vnode *)fp->f_data; 446 int count, error, ioflag = IO_UNIT; 447 448 mutex_enter(&fp->f_lock); 449 if (vp->v_type == VREG && (fp->f_flag & O_APPEND)) 450 ioflag |= IO_APPEND; 451 if (fp->f_flag & FNONBLOCK) 452 ioflag |= IO_NDELAY; 453 if (fp->f_flag & FFSYNC || 454 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 455 ioflag |= IO_SYNC; 456 else if (fp->f_flag & FDSYNC) 457 ioflag |= IO_DSYNC; 458 if (fp->f_flag & FALTIO) 459 ioflag |= IO_ALTSEMANTICS; 460 if (fp->f_flag & FDIRECT) 461 ioflag |= IO_DIRECT; 462 mutex_exit(&fp->f_lock); 463 VOP_LEASE(vp, cred, LEASE_WRITE); 464 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 465 uio->uio_offset = *offset; 466 count = uio->uio_resid; 467 error = VOP_WRITE(vp, uio, ioflag, cred); 468 if (flags & FOF_UPDATE_OFFSET) { 469 if (ioflag & IO_APPEND) 470 *offset = uio->uio_offset; 471 else 472 *offset += count - uio->uio_resid; 473 } 474 VOP_UNLOCK(vp, 0); 475 return (error); 476 } 477 478 /* 479 * File table vnode stat routine. 480 */ 481 static int 482 vn_statfile(struct file *fp, struct stat *sb, struct lwp *l) 483 { 484 struct vnode *vp = (struct vnode *)fp->f_data; 485 486 return vn_stat(vp, sb, l); 487 } 488 489 int 490 vn_stat(struct vnode *vp, struct stat *sb, struct lwp *l) 491 { 492 struct vattr va; 493 int error; 494 mode_t mode; 495 496 error = VOP_GETATTR(vp, &va, l->l_cred); 497 if (error) 498 return (error); 499 /* 500 * Copy from vattr table 501 */ 502 sb->st_dev = va.va_fsid; 503 sb->st_ino = va.va_fileid; 504 mode = va.va_mode; 505 switch (vp->v_type) { 506 case VREG: 507 mode |= S_IFREG; 508 break; 509 case VDIR: 510 mode |= S_IFDIR; 511 break; 512 case VBLK: 513 mode |= S_IFBLK; 514 break; 515 case VCHR: 516 mode |= S_IFCHR; 517 break; 518 case VLNK: 519 mode |= S_IFLNK; 520 break; 521 case VSOCK: 522 mode |= S_IFSOCK; 523 break; 524 case VFIFO: 525 mode |= S_IFIFO; 526 break; 527 default: 528 return (EBADF); 529 }; 530 sb->st_mode = mode; 531 sb->st_nlink = va.va_nlink; 532 sb->st_uid = va.va_uid; 533 sb->st_gid = va.va_gid; 534 sb->st_rdev = va.va_rdev; 535 sb->st_size = va.va_size; 536 sb->st_atimespec = va.va_atime; 537 sb->st_mtimespec = va.va_mtime; 538 sb->st_ctimespec = va.va_ctime; 539 sb->st_birthtimespec = va.va_birthtime; 540 sb->st_blksize = va.va_blocksize; 541 sb->st_flags = va.va_flags; 542 sb->st_gen = 0; 543 sb->st_blocks = va.va_bytes / S_BLKSIZE; 544 return (0); 545 } 546 547 /* 548 * File table vnode fcntl routine. 549 */ 550 static int 551 vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l) 552 { 553 struct vnode *vp = ((struct vnode *)fp->f_data); 554 int error; 555 556 error = VOP_FCNTL(vp, com, data, fp->f_flag, l->l_cred); 557 return (error); 558 } 559 560 /* 561 * File table vnode ioctl routine. 562 */ 563 static int 564 vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l) 565 { 566 struct vnode *vp = ((struct vnode *)fp->f_data), *ovp; 567 struct proc *p = l->l_proc; 568 struct vattr vattr; 569 int error; 570 571 switch (vp->v_type) { 572 573 case VREG: 574 case VDIR: 575 if (com == FIONREAD) { 576 error = VOP_GETATTR(vp, &vattr, l->l_cred); 577 if (error) 578 return (error); 579 *(int *)data = vattr.va_size - fp->f_offset; 580 return (0); 581 } 582 if ((com == FIONWRITE) || (com == FIONSPACE)) { 583 /* 584 * Files don't have send queues, so there never 585 * are any bytes in them, nor is there any 586 * open space in them. 587 */ 588 *(int *)data = 0; 589 return (0); 590 } 591 if (com == FIOGETBMAP) { 592 daddr_t *block; 593 594 if (*(daddr_t *)data < 0) 595 return (EINVAL); 596 block = (daddr_t *)data; 597 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 598 } 599 if (com == OFIOGETBMAP) { 600 daddr_t ibn, obn; 601 602 if (*(int32_t *)data < 0) 603 return (EINVAL); 604 ibn = (daddr_t)*(int32_t *)data; 605 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 606 *(int32_t *)data = (int32_t)obn; 607 return error; 608 } 609 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 610 return (0); /* XXX */ 611 /* fall into ... */ 612 case VFIFO: 613 case VCHR: 614 case VBLK: 615 error = VOP_IOCTL(vp, com, data, fp->f_flag, l->l_cred); 616 if (error == 0 && com == TIOCSCTTY) { 617 VREF(vp); 618 mutex_enter(&proclist_lock); 619 ovp = p->p_session->s_ttyvp; 620 p->p_session->s_ttyvp = vp; 621 mutex_exit(&proclist_lock); 622 if (ovp != NULL) 623 vrele(ovp); 624 } 625 return (error); 626 627 default: 628 return (EPASSTHROUGH); 629 } 630 } 631 632 /* 633 * File table vnode poll routine. 634 */ 635 static int 636 vn_poll(struct file *fp, int events, struct lwp *l) 637 { 638 639 return (VOP_POLL(((struct vnode *)fp->f_data), events)); 640 } 641 642 /* 643 * File table vnode kqfilter routine. 644 */ 645 int 646 vn_kqfilter(struct file *fp, struct knote *kn) 647 { 648 649 return (VOP_KQFILTER((struct vnode *)fp->f_data, kn)); 650 } 651 652 /* 653 * Check that the vnode is still valid, and if so 654 * acquire requested lock. 655 */ 656 int 657 vn_lock(struct vnode *vp, int flags) 658 { 659 int error; 660 661 #if 0 662 KASSERT(vp->v_usecount > 0 || (flags & LK_INTERLOCK) != 0 663 || (vp->v_iflag & VI_ONWORKLST) != 0); 664 #endif 665 KASSERT((flags & 666 ~(LK_INTERLOCK|LK_SHARED|LK_EXCLUSIVE|LK_DRAIN|LK_NOWAIT|LK_RETRY| 667 LK_SETRECURSE|LK_CANRECURSE)) 668 == 0); 669 670 do { 671 if ((flags & LK_INTERLOCK) == 0) 672 simple_lock(&vp->v_interlock); 673 if (vp->v_iflag & VI_XLOCK) { 674 if (flags & LK_NOWAIT) { 675 simple_unlock(&vp->v_interlock); 676 return EBUSY; 677 } 678 vp->v_iflag |= VI_XWANT; 679 ltsleep(vp, PINOD | PNORELOCK, 680 "vn_lock", 0, &vp->v_interlock); 681 error = ENOENT; 682 } else { 683 error = VOP_LOCK(vp, 684 (flags & ~LK_RETRY) | LK_INTERLOCK); 685 if (error == 0 || error == EDEADLK || error == EBUSY) 686 return (error); 687 } 688 flags &= ~LK_INTERLOCK; 689 } while (flags & LK_RETRY); 690 return (error); 691 } 692 693 /* 694 * File table vnode close routine. 695 */ 696 static int 697 vn_closefile(struct file *fp, struct lwp *l) 698 { 699 700 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 701 fp->f_cred, l)); 702 } 703 704 /* 705 * Enable LK_CANRECURSE on lock. Return prior status. 706 */ 707 u_int 708 vn_setrecurse(struct vnode *vp) 709 { 710 struct lock *lkp = &vp->v_lock; 711 u_int retval; 712 713 simple_lock(&lkp->lk_interlock); 714 retval = lkp->lk_flags & LK_CANRECURSE; 715 lkp->lk_flags |= LK_CANRECURSE; 716 simple_unlock(&lkp->lk_interlock); 717 718 return retval; 719 } 720 721 /* 722 * Called when done with locksetrecurse. 723 */ 724 void 725 vn_restorerecurse(struct vnode *vp, u_int flags) 726 { 727 struct lock *lkp = &vp->v_lock; 728 729 simple_lock(&lkp->lk_interlock); 730 lkp->lk_flags &= ~LK_CANRECURSE; 731 lkp->lk_flags |= flags; 732 simple_unlock(&lkp->lk_interlock); 733 } 734 735 /* 736 * Obsolete: this function will be removed from 6.0 737 * Please use fscow_establish() instead. 738 */ 739 int 740 vn_cow_establish(struct vnode *vp, 741 int (*func)(void *, struct buf *), void *cookie) 742 { 743 static int firstrun = 1; 744 745 if (firstrun) { 746 printf("%s: this function is obsolete.\n", __FUNCTION__); 747 firstrun = 0; 748 } 749 if (vp->v_type == VBLK) 750 return fscow_establish(vp->v_specmountpoint, func, cookie); 751 else 752 return fscow_establish(vp->v_mount, func, cookie); 753 } 754 755 /* 756 * Obsolete: this function will be removed from 6.0 757 * Please use fscow_disestablish() instead. 758 */ 759 int 760 vn_cow_disestablish(struct vnode *vp, 761 int (*func)(void *, struct buf *), void *cookie) 762 { 763 if (vp->v_type == VBLK) 764 return fscow_disestablish(vp->v_specmountpoint, func, cookie); 765 else 766 return fscow_disestablish(vp->v_mount, func, cookie); 767 } 768 769 /* 770 * Simplified in-kernel wrapper calls for extended attribute access. 771 * Both calls pass in a NULL credential, authorizing a "kernel" access. 772 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 773 */ 774 int 775 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 776 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 777 { 778 struct uio auio; 779 struct iovec aiov; 780 int error; 781 782 aiov.iov_len = *buflen; 783 aiov.iov_base = bf; 784 785 auio.uio_iov = &aiov; 786 auio.uio_iovcnt = 1; 787 auio.uio_rw = UIO_READ; 788 auio.uio_offset = 0; 789 auio.uio_resid = *buflen; 790 UIO_SETUP_SYSSPACE(&auio); 791 792 if ((ioflg & IO_NODELOCKED) == 0) 793 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 794 795 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 796 797 if ((ioflg & IO_NODELOCKED) == 0) 798 VOP_UNLOCK(vp, 0); 799 800 if (error == 0) 801 *buflen = *buflen - auio.uio_resid; 802 803 return (error); 804 } 805 806 /* 807 * XXX Failure mode if partially written? 808 */ 809 int 810 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 811 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 812 { 813 struct uio auio; 814 struct iovec aiov; 815 int error; 816 817 aiov.iov_len = buflen; 818 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 819 820 auio.uio_iov = &aiov; 821 auio.uio_iovcnt = 1; 822 auio.uio_rw = UIO_WRITE; 823 auio.uio_offset = 0; 824 auio.uio_resid = buflen; 825 UIO_SETUP_SYSSPACE(&auio); 826 827 if ((ioflg & IO_NODELOCKED) == 0) { 828 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 829 } 830 831 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 832 833 if ((ioflg & IO_NODELOCKED) == 0) { 834 VOP_UNLOCK(vp, 0); 835 } 836 837 return (error); 838 } 839 840 int 841 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 842 const char *attrname, struct lwp *l) 843 { 844 int error; 845 846 if ((ioflg & IO_NODELOCKED) == 0) { 847 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 848 } 849 850 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 851 if (error == EOPNOTSUPP) 852 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 853 854 if ((ioflg & IO_NODELOCKED) == 0) { 855 VOP_UNLOCK(vp, 0); 856 } 857 858 return (error); 859 } 860 861 void 862 vn_ra_allocctx(struct vnode *vp) 863 { 864 struct uvm_ractx *ra = NULL; 865 866 if (vp->v_type != VREG) { 867 return; 868 } 869 if (vp->v_ractx != NULL) { 870 return; 871 } 872 simple_lock(&vp->v_interlock); 873 if (vp->v_ractx == NULL) { 874 simple_unlock(&vp->v_interlock); 875 ra = uvm_ra_allocctx(); 876 simple_lock(&vp->v_interlock); 877 if (ra != NULL && vp->v_ractx == NULL) { 878 vp->v_ractx = ra; 879 ra = NULL; 880 } 881 } 882 simple_unlock(&vp->v_interlock); 883 if (ra != NULL) { 884 uvm_ra_freectx(ra); 885 } 886 } 887
Indexes created Sat Sep 20 22:09:52 GMT 2025