vfs_vnops.c revision 1.151
1 /* $NetBSD: vfs_vnops.c,v 1.151 2008/01/25 06:32:20 pooka 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 1.151 2008/01/25 06:32:20 pooka 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 #include <sys/filedesc.h> 64 65 #include <miscfs/specfs/specdev.h> 66 67 #include <uvm/uvm_extern.h> 68 #include <uvm/uvm_readahead.h> 69 70 #ifdef UNION 71 #include <fs/union/union.h> 72 #endif 73 74 #if defined(LKM) || defined(UNION) 75 int (*vn_union_readdir_hook) (struct vnode **, struct file *, struct lwp *); 76 #endif 77 78 #include <sys/verified_exec.h> 79 80 static int vn_read(struct file *fp, off_t *offset, struct uio *uio, 81 kauth_cred_t cred, int flags); 82 static int vn_write(struct file *fp, off_t *offset, struct uio *uio, 83 kauth_cred_t cred, int flags); 84 static int vn_closefile(struct file *fp, struct lwp *l); 85 static int vn_poll(struct file *fp, int events, struct lwp *l); 86 static int vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l); 87 static int vn_statfile(struct file *fp, struct stat *sb, struct lwp *l); 88 static int vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l); 89 90 const struct fileops vnops = { 91 vn_read, vn_write, vn_ioctl, vn_fcntl, vn_poll, 92 vn_statfile, vn_closefile, vn_kqfilter 93 }; 94 95 /* 96 * Common code for vnode open operations. 97 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 98 */ 99 int 100 vn_open(struct nameidata *ndp, int fmode, int cmode) 101 { 102 struct vnode *vp; 103 struct lwp *l = curlwp; 104 kauth_cred_t cred = l->l_cred; 105 struct vattr va; 106 int error; 107 char *path; 108 109 ndp->ni_cnd.cn_flags &= TRYEMULROOT; 110 111 if (fmode & O_CREAT) { 112 ndp->ni_cnd.cn_nameiop = CREATE; 113 ndp->ni_cnd.cn_flags |= LOCKPARENT | LOCKLEAF; 114 if ((fmode & O_EXCL) == 0 && 115 ((fmode & O_NOFOLLOW) == 0)) 116 ndp->ni_cnd.cn_flags |= FOLLOW; 117 } else { 118 ndp->ni_cnd.cn_nameiop = LOOKUP; 119 ndp->ni_cnd.cn_flags |= LOCKLEAF; 120 if ((fmode & O_NOFOLLOW) == 0) 121 ndp->ni_cnd.cn_flags |= FOLLOW; 122 } 123 124 VERIEXEC_PATH_GET(ndp->ni_dirp, ndp->ni_segflg, ndp->ni_dirp, path); 125 126 error = namei(ndp); 127 if (error) 128 goto out; 129 130 vp = ndp->ni_vp; 131 132 #if NVERIEXEC > 0 133 error = veriexec_openchk(l, ndp->ni_vp, ndp->ni_dirp, fmode); 134 if (error) 135 goto bad; 136 #endif /* NVERIEXEC > 0 */ 137 138 if (fmode & O_CREAT) { 139 if (ndp->ni_vp == NULL) { 140 VATTR_NULL(&va); 141 va.va_type = VREG; 142 va.va_mode = cmode; 143 if (fmode & O_EXCL) 144 va.va_vaflags |= VA_EXCLUSIVE; 145 VOP_LEASE(ndp->ni_dvp, cred, LEASE_WRITE); 146 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 147 &ndp->ni_cnd, &va); 148 if (error) 149 goto out; 150 fmode &= ~O_TRUNC; 151 vp = ndp->ni_vp; 152 } else { 153 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 154 if (ndp->ni_dvp == ndp->ni_vp) 155 vrele(ndp->ni_dvp); 156 else 157 vput(ndp->ni_dvp); 158 ndp->ni_dvp = NULL; 159 vp = ndp->ni_vp; 160 if (fmode & O_EXCL) { 161 error = EEXIST; 162 goto bad; 163 } 164 fmode &= ~O_CREAT; 165 } 166 } else { 167 vp = ndp->ni_vp; 168 } 169 if (vp->v_type == VSOCK) { 170 error = EOPNOTSUPP; 171 goto bad; 172 } 173 if (ndp->ni_vp->v_type == VLNK) { 174 error = EFTYPE; 175 goto bad; 176 } 177 178 if ((fmode & O_CREAT) == 0) { 179 error = vn_openchk(vp, cred, fmode); 180 if (error != 0) 181 goto bad; 182 } 183 184 if (fmode & O_TRUNC) { 185 VOP_UNLOCK(vp, 0); /* XXX */ 186 187 VOP_LEASE(vp, cred, LEASE_WRITE); 188 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */ 189 VATTR_NULL(&va); 190 va.va_size = 0; 191 error = VOP_SETATTR(vp, &va, cred); 192 if (error != 0) 193 goto bad; 194 } 195 if ((error = VOP_OPEN(vp, fmode, cred)) != 0) 196 goto bad; 197 if (fmode & FWRITE) { 198 mutex_enter(&vp->v_interlock); 199 vp->v_writecount++; 200 mutex_exit(&vp->v_interlock); 201 } 202 203 bad: 204 if (error) 205 vput(vp); 206 out: 207 VERIEXEC_PATH_PUT(path); 208 return (error); 209 } 210 211 /* 212 * Check for write permissions on the specified vnode. 213 * Prototype text segments cannot be written. 214 */ 215 int 216 vn_writechk(struct vnode *vp) 217 { 218 219 /* 220 * If the vnode is in use as a process's text, 221 * we can't allow writing. 222 */ 223 if (vp->v_iflag & VI_TEXT) 224 return (ETXTBSY); 225 return (0); 226 } 227 228 int 229 vn_openchk(struct vnode *vp, kauth_cred_t cred, int fflags) 230 { 231 int permbits = 0; 232 int error; 233 234 if ((fflags & FREAD) != 0) { 235 permbits = VREAD; 236 } 237 if ((fflags & (FWRITE | O_TRUNC)) != 0) { 238 permbits |= VWRITE; 239 if (vp->v_type == VDIR) { 240 error = EISDIR; 241 goto bad; 242 } 243 error = vn_writechk(vp); 244 if (error != 0) 245 goto bad; 246 } 247 error = VOP_ACCESS(vp, permbits, cred); 248 bad: 249 return error; 250 } 251 252 /* 253 * Mark a vnode as having executable mappings. 254 */ 255 void 256 vn_markexec(struct vnode *vp) 257 { 258 259 KASSERT(mutex_owned(&vp->v_interlock)); 260 261 if ((vp->v_iflag & VI_EXECMAP) == 0) { 262 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 263 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 264 } 265 vp->v_iflag |= VI_EXECMAP; 266 } 267 268 /* 269 * Mark a vnode as being the text of a process. 270 * Fail if the vnode is currently writable. 271 */ 272 int 273 vn_marktext(struct vnode *vp) 274 { 275 276 mutex_enter(&vp->v_interlock); 277 if (vp->v_writecount != 0) { 278 KASSERT((vp->v_iflag & VI_TEXT) == 0); 279 mutex_exit(&vp->v_interlock); 280 return (ETXTBSY); 281 } 282 vp->v_iflag |= VI_TEXT; 283 vn_markexec(vp); 284 mutex_exit(&vp->v_interlock); 285 return (0); 286 } 287 288 /* 289 * Vnode close call 290 * 291 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 292 */ 293 int 294 vn_close(struct vnode *vp, int flags, kauth_cred_t cred, struct lwp *l) 295 { 296 int error; 297 298 mutex_enter(&vp->v_interlock); 299 if (flags & FWRITE) 300 vp->v_writecount--; 301 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_INTERLOCK); 302 error = VOP_CLOSE(vp, flags, cred); 303 vput(vp); 304 return (error); 305 } 306 307 /* 308 * Package up an I/O request on a vnode into a uio and do it. 309 */ 310 int 311 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 312 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 313 struct lwp *l) 314 { 315 struct uio auio; 316 struct iovec aiov; 317 int error; 318 319 if ((ioflg & IO_NODELOCKED) == 0) { 320 if (rw == UIO_READ) { 321 vn_lock(vp, LK_SHARED | LK_RETRY); 322 } else /* UIO_WRITE */ { 323 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 324 } 325 } 326 auio.uio_iov = &aiov; 327 auio.uio_iovcnt = 1; 328 aiov.iov_base = base; 329 aiov.iov_len = len; 330 auio.uio_resid = len; 331 auio.uio_offset = offset; 332 auio.uio_rw = rw; 333 if (segflg == UIO_SYSSPACE) { 334 UIO_SETUP_SYSSPACE(&auio); 335 } else { 336 auio.uio_vmspace = l->l_proc->p_vmspace; 337 } 338 if (rw == UIO_READ) { 339 error = VOP_READ(vp, &auio, ioflg, cred); 340 } else { 341 error = VOP_WRITE(vp, &auio, ioflg, cred); 342 } 343 if (aresid) 344 *aresid = auio.uio_resid; 345 else 346 if (auio.uio_resid && error == 0) 347 error = EIO; 348 if ((ioflg & IO_NODELOCKED) == 0) { 349 VOP_UNLOCK(vp, 0); 350 } 351 return (error); 352 } 353 354 int 355 vn_readdir(struct file *fp, char *bf, int segflg, u_int count, int *done, 356 struct lwp *l, off_t **cookies, int *ncookies) 357 { 358 struct vnode *vp = (struct vnode *)fp->f_data; 359 struct iovec aiov; 360 struct uio auio; 361 int error, eofflag; 362 363 /* Limit the size on any kernel buffers used by VOP_READDIR */ 364 count = min(MAXBSIZE, count); 365 366 unionread: 367 if (vp->v_type != VDIR) 368 return (EINVAL); 369 aiov.iov_base = bf; 370 aiov.iov_len = count; 371 auio.uio_iov = &aiov; 372 auio.uio_iovcnt = 1; 373 auio.uio_rw = UIO_READ; 374 if (segflg == UIO_SYSSPACE) { 375 UIO_SETUP_SYSSPACE(&auio); 376 } else { 377 KASSERT(l == curlwp); 378 auio.uio_vmspace = l->l_proc->p_vmspace; 379 } 380 auio.uio_resid = count; 381 vn_lock(vp, LK_SHARED | LK_RETRY); 382 auio.uio_offset = fp->f_offset; 383 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 384 ncookies); 385 FILE_LOCK(fp); 386 fp->f_offset = auio.uio_offset; 387 FILE_UNLOCK(fp); 388 VOP_UNLOCK(vp, 0); 389 if (error) 390 return (error); 391 392 #if defined(UNION) || defined(LKM) 393 if (count == auio.uio_resid && vn_union_readdir_hook) { 394 struct vnode *ovp = vp; 395 396 error = (*vn_union_readdir_hook)(&vp, fp, l); 397 if (error) 398 return (error); 399 if (vp != ovp) 400 goto unionread; 401 } 402 #endif /* UNION || LKM */ 403 404 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 405 (vp->v_mount->mnt_flag & MNT_UNION)) { 406 struct vnode *tvp = vp; 407 vp = vp->v_mount->mnt_vnodecovered; 408 VREF(vp); 409 FILE_LOCK(fp); 410 fp->f_data = vp; 411 fp->f_offset = 0; 412 FILE_UNLOCK(fp); 413 vrele(tvp); 414 goto unionread; 415 } 416 *done = count - auio.uio_resid; 417 return error; 418 } 419 420 /* 421 * File table vnode read routine. 422 */ 423 static int 424 vn_read(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 425 int flags) 426 { 427 struct vnode *vp = (struct vnode *)fp->f_data; 428 int count, error, ioflag; 429 430 VOP_LEASE(vp, cred, LEASE_READ); 431 FILE_LOCK(fp); 432 ioflag = IO_ADV_ENCODE(fp->f_advice); 433 if (fp->f_flag & FNONBLOCK) 434 ioflag |= IO_NDELAY; 435 if ((fp->f_flag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 436 ioflag |= IO_SYNC; 437 if (fp->f_flag & FALTIO) 438 ioflag |= IO_ALTSEMANTICS; 439 if (fp->f_flag & FDIRECT) 440 ioflag |= IO_DIRECT; 441 FILE_UNLOCK(fp); 442 vn_lock(vp, LK_SHARED | LK_RETRY); 443 uio->uio_offset = *offset; 444 count = uio->uio_resid; 445 error = VOP_READ(vp, uio, ioflag, cred); 446 if (flags & FOF_UPDATE_OFFSET) 447 *offset += count - uio->uio_resid; 448 VOP_UNLOCK(vp, 0); 449 return (error); 450 } 451 452 /* 453 * File table vnode write routine. 454 */ 455 static int 456 vn_write(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 457 int flags) 458 { 459 struct vnode *vp = (struct vnode *)fp->f_data; 460 int count, error, ioflag; 461 462 FILE_LOCK(fp); 463 ioflag = IO_ADV_ENCODE(fp->f_advice) | IO_UNIT; 464 if (vp->v_type == VREG && (fp->f_flag & O_APPEND)) 465 ioflag |= IO_APPEND; 466 if (fp->f_flag & FNONBLOCK) 467 ioflag |= IO_NDELAY; 468 if (fp->f_flag & FFSYNC || 469 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 470 ioflag |= IO_SYNC; 471 else if (fp->f_flag & FDSYNC) 472 ioflag |= IO_DSYNC; 473 if (fp->f_flag & FALTIO) 474 ioflag |= IO_ALTSEMANTICS; 475 if (fp->f_flag & FDIRECT) 476 ioflag |= IO_DIRECT; 477 FILE_UNLOCK(fp); 478 VOP_LEASE(vp, cred, LEASE_WRITE); 479 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 480 uio->uio_offset = *offset; 481 count = uio->uio_resid; 482 error = VOP_WRITE(vp, uio, ioflag, cred); 483 if (flags & FOF_UPDATE_OFFSET) { 484 if (ioflag & IO_APPEND) 485 *offset = uio->uio_offset; 486 else 487 *offset += count - uio->uio_resid; 488 } 489 VOP_UNLOCK(vp, 0); 490 return (error); 491 } 492 493 /* 494 * File table vnode stat routine. 495 */ 496 static int 497 vn_statfile(struct file *fp, struct stat *sb, struct lwp *l) 498 { 499 struct vnode *vp = (struct vnode *)fp->f_data; 500 501 return vn_stat(vp, sb, l); 502 } 503 504 int 505 vn_stat(struct vnode *vp, struct stat *sb, struct lwp *l) 506 { 507 struct vattr va; 508 int error; 509 mode_t mode; 510 511 error = VOP_GETATTR(vp, &va, l->l_cred); 512 if (error) 513 return (error); 514 /* 515 * Copy from vattr table 516 */ 517 sb->st_dev = va.va_fsid; 518 sb->st_ino = va.va_fileid; 519 mode = va.va_mode; 520 switch (vp->v_type) { 521 case VREG: 522 mode |= S_IFREG; 523 break; 524 case VDIR: 525 mode |= S_IFDIR; 526 break; 527 case VBLK: 528 mode |= S_IFBLK; 529 break; 530 case VCHR: 531 mode |= S_IFCHR; 532 break; 533 case VLNK: 534 mode |= S_IFLNK; 535 break; 536 case VSOCK: 537 mode |= S_IFSOCK; 538 break; 539 case VFIFO: 540 mode |= S_IFIFO; 541 break; 542 default: 543 return (EBADF); 544 }; 545 sb->st_mode = mode; 546 sb->st_nlink = va.va_nlink; 547 sb->st_uid = va.va_uid; 548 sb->st_gid = va.va_gid; 549 sb->st_rdev = va.va_rdev; 550 sb->st_size = va.va_size; 551 sb->st_atimespec = va.va_atime; 552 sb->st_mtimespec = va.va_mtime; 553 sb->st_ctimespec = va.va_ctime; 554 sb->st_birthtimespec = va.va_birthtime; 555 sb->st_blksize = va.va_blocksize; 556 sb->st_flags = va.va_flags; 557 sb->st_gen = 0; 558 sb->st_blocks = va.va_bytes / S_BLKSIZE; 559 return (0); 560 } 561 562 /* 563 * File table vnode fcntl routine. 564 */ 565 static int 566 vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l) 567 { 568 struct vnode *vp = ((struct vnode *)fp->f_data); 569 int error; 570 571 error = VOP_FCNTL(vp, com, data, fp->f_flag, l->l_cred); 572 return (error); 573 } 574 575 /* 576 * File table vnode ioctl routine. 577 */ 578 static int 579 vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l) 580 { 581 struct vnode *vp = ((struct vnode *)fp->f_data), *ovp; 582 struct proc *p = l->l_proc; 583 struct vattr vattr; 584 int error; 585 586 switch (vp->v_type) { 587 588 case VREG: 589 case VDIR: 590 if (com == FIONREAD) { 591 error = VOP_GETATTR(vp, &vattr, l->l_cred); 592 if (error) 593 return (error); 594 *(int *)data = vattr.va_size - fp->f_offset; 595 return (0); 596 } 597 if ((com == FIONWRITE) || (com == FIONSPACE)) { 598 /* 599 * Files don't have send queues, so there never 600 * are any bytes in them, nor is there any 601 * open space in them. 602 */ 603 *(int *)data = 0; 604 return (0); 605 } 606 if (com == FIOGETBMAP) { 607 daddr_t *block; 608 609 if (*(daddr_t *)data < 0) 610 return (EINVAL); 611 block = (daddr_t *)data; 612 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 613 } 614 if (com == OFIOGETBMAP) { 615 daddr_t ibn, obn; 616 617 if (*(int32_t *)data < 0) 618 return (EINVAL); 619 ibn = (daddr_t)*(int32_t *)data; 620 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 621 *(int32_t *)data = (int32_t)obn; 622 return error; 623 } 624 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 625 return (0); /* XXX */ 626 /* fall into ... */ 627 case VFIFO: 628 case VCHR: 629 case VBLK: 630 error = VOP_IOCTL(vp, com, data, fp->f_flag, l->l_cred); 631 if (error == 0 && com == TIOCSCTTY) { 632 VREF(vp); 633 mutex_enter(&proclist_lock); 634 ovp = p->p_session->s_ttyvp; 635 p->p_session->s_ttyvp = vp; 636 mutex_exit(&proclist_lock); 637 if (ovp != NULL) 638 vrele(ovp); 639 } 640 return (error); 641 642 default: 643 return (EPASSTHROUGH); 644 } 645 } 646 647 /* 648 * File table vnode poll routine. 649 */ 650 static int 651 vn_poll(struct file *fp, int events, struct lwp *l) 652 { 653 654 return (VOP_POLL(((struct vnode *)fp->f_data), events)); 655 } 656 657 /* 658 * File table vnode kqfilter routine. 659 */ 660 int 661 vn_kqfilter(struct file *fp, struct knote *kn) 662 { 663 664 return (VOP_KQFILTER((struct vnode *)fp->f_data, kn)); 665 } 666 667 /* 668 * Check that the vnode is still valid, and if so 669 * acquire requested lock. 670 */ 671 int 672 vn_lock(struct vnode *vp, int flags) 673 { 674 int error; 675 676 #if 0 677 KASSERT(vp->v_usecount > 0 || (flags & LK_INTERLOCK) != 0 678 || (vp->v_iflag & VI_ONWORKLST) != 0); 679 #endif 680 KASSERT((flags & 681 ~(LK_INTERLOCK|LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT|LK_RETRY| 682 LK_CANRECURSE)) 683 == 0); 684 685 do { 686 if ((flags & LK_INTERLOCK) == 0) 687 mutex_enter(&vp->v_interlock); 688 if (vp->v_iflag & VI_XLOCK) { 689 if (flags & LK_NOWAIT) { 690 mutex_exit(&vp->v_interlock); 691 return EBUSY; 692 } 693 vwait(vp, VI_XLOCK); 694 mutex_exit(&vp->v_interlock); 695 error = ENOENT; 696 } else { 697 error = VOP_LOCK(vp, 698 (flags & ~LK_RETRY) | LK_INTERLOCK); 699 if (error == 0 || error == EDEADLK || error == EBUSY) 700 return (error); 701 } 702 flags &= ~LK_INTERLOCK; 703 } while (flags & LK_RETRY); 704 return (error); 705 } 706 707 /* 708 * File table vnode close routine. 709 */ 710 static int 711 vn_closefile(struct file *fp, struct lwp *l) 712 { 713 714 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 715 fp->f_cred, l)); 716 } 717 718 /* 719 * Enable LK_CANRECURSE on lock. Return prior status. 720 */ 721 u_int 722 vn_setrecurse(struct vnode *vp) 723 { 724 struct lock *lkp = vp->v_vnlock; 725 u_int retval; 726 727 mutex_enter(&lkp->lk_interlock); 728 retval = lkp->lk_flags & LK_CANRECURSE; 729 lkp->lk_flags |= LK_CANRECURSE; 730 mutex_exit(&lkp->lk_interlock); 731 732 return retval; 733 } 734 735 /* 736 * Called when done with locksetrecurse. 737 */ 738 void 739 vn_restorerecurse(struct vnode *vp, u_int flags) 740 { 741 struct lock *lkp = vp->v_vnlock; 742 743 mutex_enter(&lkp->lk_interlock); 744 lkp->lk_flags &= ~LK_CANRECURSE; 745 lkp->lk_flags |= flags; 746 mutex_exit(&lkp->lk_interlock); 747 } 748 749 /* 750 * Simplified in-kernel wrapper calls for extended attribute access. 751 * Both calls pass in a NULL credential, authorizing a "kernel" access. 752 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 753 */ 754 int 755 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 756 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 757 { 758 struct uio auio; 759 struct iovec aiov; 760 int error; 761 762 aiov.iov_len = *buflen; 763 aiov.iov_base = bf; 764 765 auio.uio_iov = &aiov; 766 auio.uio_iovcnt = 1; 767 auio.uio_rw = UIO_READ; 768 auio.uio_offset = 0; 769 auio.uio_resid = *buflen; 770 UIO_SETUP_SYSSPACE(&auio); 771 772 if ((ioflg & IO_NODELOCKED) == 0) 773 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 774 775 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 776 777 if ((ioflg & IO_NODELOCKED) == 0) 778 VOP_UNLOCK(vp, 0); 779 780 if (error == 0) 781 *buflen = *buflen - auio.uio_resid; 782 783 return (error); 784 } 785 786 /* 787 * XXX Failure mode if partially written? 788 */ 789 int 790 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 791 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 792 { 793 struct uio auio; 794 struct iovec aiov; 795 int error; 796 797 aiov.iov_len = buflen; 798 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 799 800 auio.uio_iov = &aiov; 801 auio.uio_iovcnt = 1; 802 auio.uio_rw = UIO_WRITE; 803 auio.uio_offset = 0; 804 auio.uio_resid = buflen; 805 UIO_SETUP_SYSSPACE(&auio); 806 807 if ((ioflg & IO_NODELOCKED) == 0) { 808 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 809 } 810 811 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 812 813 if ((ioflg & IO_NODELOCKED) == 0) { 814 VOP_UNLOCK(vp, 0); 815 } 816 817 return (error); 818 } 819 820 int 821 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 822 const char *attrname, struct lwp *l) 823 { 824 int error; 825 826 if ((ioflg & IO_NODELOCKED) == 0) { 827 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 828 } 829 830 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 831 if (error == EOPNOTSUPP) 832 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 833 834 if ((ioflg & IO_NODELOCKED) == 0) { 835 VOP_UNLOCK(vp, 0); 836 } 837 838 return (error); 839 } 840 841 void 842 vn_ra_allocctx(struct vnode *vp) 843 { 844 struct uvm_ractx *ra = NULL; 845 846 KASSERT(mutex_owned(&vp->v_interlock)); 847 848 if (vp->v_type != VREG) { 849 return; 850 } 851 if (vp->v_ractx != NULL) { 852 return; 853 } 854 if (vp->v_ractx == NULL) { 855 mutex_exit(&vp->v_interlock); 856 ra = uvm_ra_allocctx(); 857 mutex_enter(&vp->v_interlock); 858 if (ra != NULL && vp->v_ractx == NULL) { 859 vp->v_ractx = ra; 860 ra = NULL; 861 } 862 } 863 if (ra != NULL) { 864 uvm_ra_freectx(ra); 865 } 866 } 867
Indexes created Wed Sep 24 05:09:52 GMT 2025