vfs_vnops.c revision 1.139.2.1
1 /* $NetBSD: vfs_vnops.c,v 1.139.2.1 2007/08/15 13:49:25 skrll 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 1.139.2.1 2007/08/15 13:49:25 skrll 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 62 #include <miscfs/specfs/specdev.h> 63 64 #include <uvm/uvm_extern.h> 65 #include <uvm/uvm_readahead.h> 66 67 #ifdef UNION 68 #include <fs/union/union.h> 69 #endif 70 71 #if defined(LKM) || defined(UNION) 72 int (*vn_union_readdir_hook) (struct vnode **, struct file *, struct lwp *); 73 #endif 74 75 #include <sys/verified_exec.h> 76 77 static int vn_read(struct file *fp, off_t *offset, struct uio *uio, 78 kauth_cred_t cred, int flags); 79 static int vn_write(struct file *fp, off_t *offset, struct uio *uio, 80 kauth_cred_t cred, int flags); 81 static int vn_closefile(struct file *fp, struct lwp *l); 82 static int vn_poll(struct file *fp, int events, struct lwp *l); 83 static int vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l); 84 static int vn_statfile(struct file *fp, struct stat *sb, struct lwp *l); 85 static int vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l); 86 87 const struct fileops vnops = { 88 vn_read, vn_write, vn_ioctl, vn_fcntl, vn_poll, 89 vn_statfile, vn_closefile, vn_kqfilter 90 }; 91 92 /* 93 * Common code for vnode open operations. 94 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 95 */ 96 int 97 vn_open(struct nameidata *ndp, int fmode, int cmode) 98 { 99 struct vnode *vp; 100 struct lwp *l = ndp->ni_cnd.cn_lwp; 101 kauth_cred_t cred = l->l_cred; 102 struct vattr va; 103 int error; 104 char *path; 105 106 ndp->ni_cnd.cn_flags &= TRYEMULROOT; 107 108 if (fmode & O_CREAT) { 109 ndp->ni_cnd.cn_nameiop = CREATE; 110 ndp->ni_cnd.cn_flags |= LOCKPARENT | LOCKLEAF; 111 if ((fmode & O_EXCL) == 0 && 112 ((fmode & O_NOFOLLOW) == 0)) 113 ndp->ni_cnd.cn_flags |= FOLLOW; 114 } else { 115 ndp->ni_cnd.cn_nameiop = LOOKUP; 116 ndp->ni_cnd.cn_flags |= LOCKLEAF; 117 if ((fmode & O_NOFOLLOW) == 0) 118 ndp->ni_cnd.cn_flags |= FOLLOW; 119 } 120 121 VERIEXEC_PATH_GET(ndp->ni_dirp, ndp->ni_segflg, ndp->ni_dirp, path); 122 123 error = namei(ndp); 124 if (error) 125 goto out; 126 127 vp = ndp->ni_vp; 128 129 #if NVERIEXEC > 0 130 error = veriexec_openchk(l, ndp->ni_vp, ndp->ni_dirp, fmode); 131 if (error) 132 goto bad; 133 #endif /* NVERIEXEC > 0 */ 134 135 if (fmode & O_CREAT) { 136 if (ndp->ni_vp == NULL) { 137 VATTR_NULL(&va); 138 va.va_type = VREG; 139 va.va_mode = cmode; 140 if (fmode & O_EXCL) 141 va.va_vaflags |= VA_EXCLUSIVE; 142 VOP_LEASE(ndp->ni_dvp, l, cred, LEASE_WRITE); 143 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 144 &ndp->ni_cnd, &va); 145 if (error) 146 goto out; 147 fmode &= ~O_TRUNC; 148 vp = ndp->ni_vp; 149 } else { 150 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 151 if (ndp->ni_dvp == ndp->ni_vp) 152 vrele(ndp->ni_dvp); 153 else 154 vput(ndp->ni_dvp); 155 ndp->ni_dvp = NULL; 156 vp = ndp->ni_vp; 157 if (fmode & O_EXCL) { 158 error = EEXIST; 159 goto bad; 160 } 161 fmode &= ~O_CREAT; 162 } 163 } else { 164 vp = ndp->ni_vp; 165 } 166 if (vp->v_type == VSOCK) { 167 error = EOPNOTSUPP; 168 goto bad; 169 } 170 if (ndp->ni_vp->v_type == VLNK) { 171 error = EFTYPE; 172 goto bad; 173 } 174 175 if ((fmode & O_CREAT) == 0) { 176 if (fmode & FREAD) { 177 if ((error = VOP_ACCESS(vp, VREAD, cred, l)) != 0) 178 goto bad; 179 } 180 181 if (fmode & (FWRITE | O_TRUNC)) { 182 if (vp->v_type == VDIR) { 183 error = EISDIR; 184 goto bad; 185 } 186 if ((error = vn_writechk(vp)) != 0 || 187 (error = VOP_ACCESS(vp, VWRITE, cred, l)) != 0) 188 goto bad; 189 } 190 } 191 192 if (fmode & O_TRUNC) { 193 VOP_UNLOCK(vp, 0); /* XXX */ 194 195 VOP_LEASE(vp, l, cred, LEASE_WRITE); 196 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */ 197 VATTR_NULL(&va); 198 va.va_size = 0; 199 error = VOP_SETATTR(vp, &va, cred, l); 200 if (error != 0) 201 goto bad; 202 } 203 if ((error = VOP_OPEN(vp, fmode, cred, l)) != 0) 204 goto bad; 205 if (fmode & FWRITE) 206 vp->v_writecount++; 207 208 bad: 209 if (error) 210 vput(vp); 211 out: 212 VERIEXEC_PATH_PUT(path); 213 return (error); 214 } 215 216 /* 217 * Check for write permissions on the specified vnode. 218 * Prototype text segments cannot be written. 219 */ 220 int 221 vn_writechk(struct vnode *vp) 222 { 223 224 /* 225 * If the vnode is in use as a process's text, 226 * we can't allow writing. 227 */ 228 if (vp->v_flag & VTEXT) 229 return (ETXTBSY); 230 return (0); 231 } 232 233 /* 234 * Mark a vnode as having executable mappings. 235 */ 236 void 237 vn_markexec(struct vnode *vp) 238 { 239 if ((vp->v_flag & VEXECMAP) == 0) { 240 uvmexp.filepages -= vp->v_uobj.uo_npages; 241 uvmexp.execpages += vp->v_uobj.uo_npages; 242 } 243 vp->v_flag |= VEXECMAP; 244 } 245 246 /* 247 * Mark a vnode as being the text of a process. 248 * Fail if the vnode is currently writable. 249 */ 250 int 251 vn_marktext(struct vnode *vp) 252 { 253 254 if (vp->v_writecount != 0) { 255 KASSERT((vp->v_flag & VTEXT) == 0); 256 return (ETXTBSY); 257 } 258 vp->v_flag |= VTEXT; 259 vn_markexec(vp); 260 return (0); 261 } 262 263 /* 264 * Vnode close call 265 * 266 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 267 */ 268 int 269 vn_close(struct vnode *vp, int flags, kauth_cred_t cred, struct lwp *l) 270 { 271 int error; 272 273 if (flags & FWRITE) 274 vp->v_writecount--; 275 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 276 error = VOP_CLOSE(vp, flags, cred, l); 277 vput(vp); 278 return (error); 279 } 280 281 /* 282 * Package up an I/O request on a vnode into a uio and do it. 283 */ 284 int 285 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 286 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 287 struct lwp *l) 288 { 289 struct uio auio; 290 struct iovec aiov; 291 int error; 292 293 if ((ioflg & IO_NODELOCKED) == 0) { 294 if (rw == UIO_READ) { 295 vn_lock(vp, LK_SHARED | LK_RETRY); 296 } else /* UIO_WRITE */ { 297 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 298 } 299 } 300 auio.uio_iov = &aiov; 301 auio.uio_iovcnt = 1; 302 aiov.iov_base = base; 303 aiov.iov_len = len; 304 auio.uio_resid = len; 305 auio.uio_offset = offset; 306 auio.uio_rw = rw; 307 if (segflg == UIO_SYSSPACE) { 308 UIO_SETUP_SYSSPACE(&auio); 309 } else { 310 auio.uio_vmspace = l->l_proc->p_vmspace; 311 } 312 if (rw == UIO_READ) { 313 error = VOP_READ(vp, &auio, ioflg, cred); 314 } else { 315 error = VOP_WRITE(vp, &auio, ioflg, cred); 316 } 317 if (aresid) 318 *aresid = auio.uio_resid; 319 else 320 if (auio.uio_resid && error == 0) 321 error = EIO; 322 if ((ioflg & IO_NODELOCKED) == 0) { 323 VOP_UNLOCK(vp, 0); 324 } 325 return (error); 326 } 327 328 int 329 vn_readdir(struct file *fp, char *bf, int segflg, u_int count, int *done, 330 struct lwp *l, off_t **cookies, int *ncookies) 331 { 332 struct vnode *vp = (struct vnode *)fp->f_data; 333 struct iovec aiov; 334 struct uio auio; 335 int error, eofflag; 336 337 /* Limit the size on any kernel buffers used by VOP_READDIR */ 338 count = min(MAXBSIZE, count); 339 340 unionread: 341 if (vp->v_type != VDIR) 342 return (EINVAL); 343 aiov.iov_base = bf; 344 aiov.iov_len = count; 345 auio.uio_iov = &aiov; 346 auio.uio_iovcnt = 1; 347 auio.uio_rw = UIO_READ; 348 if (segflg == UIO_SYSSPACE) { 349 UIO_SETUP_SYSSPACE(&auio); 350 } else { 351 KASSERT(l == curlwp); 352 auio.uio_vmspace = l->l_proc->p_vmspace; 353 } 354 auio.uio_resid = count; 355 vn_lock(vp, LK_SHARED | LK_RETRY); 356 auio.uio_offset = fp->f_offset; 357 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 358 ncookies); 359 fp->f_offset = auio.uio_offset; 360 VOP_UNLOCK(vp, 0); 361 if (error) 362 return (error); 363 364 #if defined(UNION) || defined(LKM) 365 if (count == auio.uio_resid && vn_union_readdir_hook) { 366 struct vnode *ovp = vp; 367 368 error = (*vn_union_readdir_hook)(&vp, fp, l); 369 if (error) 370 return (error); 371 if (vp != ovp) 372 goto unionread; 373 } 374 #endif /* UNION || LKM */ 375 376 if (count == auio.uio_resid && (vp->v_flag & VROOT) && 377 (vp->v_mount->mnt_flag & MNT_UNION)) { 378 struct vnode *tvp = vp; 379 vp = vp->v_mount->mnt_vnodecovered; 380 VREF(vp); 381 fp->f_data = vp; 382 fp->f_offset = 0; 383 vrele(tvp); 384 goto unionread; 385 } 386 *done = count - auio.uio_resid; 387 return error; 388 } 389 390 /* 391 * File table vnode read routine. 392 */ 393 static int 394 vn_read(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 395 int flags) 396 { 397 struct vnode *vp = (struct vnode *)fp->f_data; 398 int count, error, ioflag; 399 struct lwp *l = curlwp; 400 401 VOP_LEASE(vp, l, cred, LEASE_READ); 402 ioflag = IO_ADV_ENCODE(fp->f_advice); 403 if (fp->f_flag & FNONBLOCK) 404 ioflag |= IO_NDELAY; 405 if ((fp->f_flag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 406 ioflag |= IO_SYNC; 407 if (fp->f_flag & FALTIO) 408 ioflag |= IO_ALTSEMANTICS; 409 if (fp->f_flag & FDIRECT) 410 ioflag |= IO_DIRECT; 411 vn_lock(vp, LK_SHARED | LK_RETRY); 412 uio->uio_offset = *offset; 413 count = uio->uio_resid; 414 error = VOP_READ(vp, uio, ioflag, cred); 415 if (flags & FOF_UPDATE_OFFSET) 416 *offset += count - uio->uio_resid; 417 VOP_UNLOCK(vp, 0); 418 return (error); 419 } 420 421 /* 422 * File table vnode write routine. 423 */ 424 static int 425 vn_write(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 426 int flags) 427 { 428 struct vnode *vp = (struct vnode *)fp->f_data; 429 int count, error, ioflag = IO_UNIT; 430 struct lwp *l = curlwp; 431 432 if (vp->v_type == VREG && (fp->f_flag & O_APPEND)) 433 ioflag |= IO_APPEND; 434 if (fp->f_flag & FNONBLOCK) 435 ioflag |= IO_NDELAY; 436 if (fp->f_flag & FFSYNC || 437 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 438 ioflag |= IO_SYNC; 439 else if (fp->f_flag & FDSYNC) 440 ioflag |= IO_DSYNC; 441 if (fp->f_flag & FALTIO) 442 ioflag |= IO_ALTSEMANTICS; 443 if (fp->f_flag & FDIRECT) 444 ioflag |= IO_DIRECT; 445 VOP_LEASE(vp, l, cred, LEASE_WRITE); 446 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 447 uio->uio_offset = *offset; 448 count = uio->uio_resid; 449 error = VOP_WRITE(vp, uio, ioflag, cred); 450 if (flags & FOF_UPDATE_OFFSET) { 451 if (ioflag & IO_APPEND) 452 *offset = uio->uio_offset; 453 else 454 *offset += count - uio->uio_resid; 455 } 456 VOP_UNLOCK(vp, 0); 457 return (error); 458 } 459 460 /* 461 * File table vnode stat routine. 462 */ 463 static int 464 vn_statfile(struct file *fp, struct stat *sb, struct lwp *l) 465 { 466 struct vnode *vp = (struct vnode *)fp->f_data; 467 468 return vn_stat(vp, sb, l); 469 } 470 471 int 472 vn_stat(struct vnode *vp, struct stat *sb, struct lwp *l) 473 { 474 struct vattr va; 475 int error; 476 mode_t mode; 477 478 error = VOP_GETATTR(vp, &va, l->l_cred, l); 479 if (error) 480 return (error); 481 /* 482 * Copy from vattr table 483 */ 484 sb->st_dev = va.va_fsid; 485 sb->st_ino = va.va_fileid; 486 mode = va.va_mode; 487 switch (vp->v_type) { 488 case VREG: 489 mode |= S_IFREG; 490 break; 491 case VDIR: 492 mode |= S_IFDIR; 493 break; 494 case VBLK: 495 mode |= S_IFBLK; 496 break; 497 case VCHR: 498 mode |= S_IFCHR; 499 break; 500 case VLNK: 501 mode |= S_IFLNK; 502 break; 503 case VSOCK: 504 mode |= S_IFSOCK; 505 break; 506 case VFIFO: 507 mode |= S_IFIFO; 508 break; 509 default: 510 return (EBADF); 511 }; 512 sb->st_mode = mode; 513 sb->st_nlink = va.va_nlink; 514 sb->st_uid = va.va_uid; 515 sb->st_gid = va.va_gid; 516 sb->st_rdev = va.va_rdev; 517 sb->st_size = va.va_size; 518 sb->st_atimespec = va.va_atime; 519 sb->st_mtimespec = va.va_mtime; 520 sb->st_ctimespec = va.va_ctime; 521 sb->st_birthtimespec = va.va_birthtime; 522 sb->st_blksize = va.va_blocksize; 523 sb->st_flags = va.va_flags; 524 sb->st_gen = 0; 525 sb->st_blocks = va.va_bytes / S_BLKSIZE; 526 return (0); 527 } 528 529 /* 530 * File table vnode fcntl routine. 531 */ 532 static int 533 vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l) 534 { 535 struct vnode *vp = ((struct vnode *)fp->f_data); 536 int error; 537 538 error = VOP_FCNTL(vp, com, data, fp->f_flag, l->l_cred, l); 539 return (error); 540 } 541 542 /* 543 * File table vnode ioctl routine. 544 */ 545 static int 546 vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l) 547 { 548 struct vnode *vp = ((struct vnode *)fp->f_data), *ovp; 549 struct proc *p = l->l_proc; 550 struct vattr vattr; 551 int error; 552 553 switch (vp->v_type) { 554 555 case VREG: 556 case VDIR: 557 if (com == FIONREAD) { 558 error = VOP_GETATTR(vp, &vattr, l->l_cred, l); 559 if (error) 560 return (error); 561 *(int *)data = vattr.va_size - fp->f_offset; 562 return (0); 563 } 564 if ((com == FIONWRITE) || (com == FIONSPACE)) { 565 /* 566 * Files don't have send queues, so there never 567 * are any bytes in them, nor is there any 568 * open space in them. 569 */ 570 *(int *)data = 0; 571 return (0); 572 } 573 if (com == FIOGETBMAP) { 574 daddr_t *block; 575 576 if (*(daddr_t *)data < 0) 577 return (EINVAL); 578 block = (daddr_t *)data; 579 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 580 } 581 if (com == OFIOGETBMAP) { 582 daddr_t ibn, obn; 583 584 if (*(int32_t *)data < 0) 585 return (EINVAL); 586 ibn = (daddr_t)*(int32_t *)data; 587 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 588 *(int32_t *)data = (int32_t)obn; 589 return error; 590 } 591 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 592 return (0); /* XXX */ 593 /* fall into ... */ 594 case VFIFO: 595 case VCHR: 596 case VBLK: 597 error = VOP_IOCTL(vp, com, data, fp->f_flag, 598 l->l_cred, l); 599 if (error == 0 && com == TIOCSCTTY) { 600 VREF(vp); 601 mutex_enter(&proclist_lock); 602 ovp = p->p_session->s_ttyvp; 603 p->p_session->s_ttyvp = vp; 604 mutex_exit(&proclist_lock); 605 if (ovp != NULL) 606 vrele(ovp); 607 } 608 return (error); 609 610 default: 611 return (EPASSTHROUGH); 612 } 613 } 614 615 /* 616 * File table vnode poll routine. 617 */ 618 static int 619 vn_poll(struct file *fp, int events, struct lwp *l) 620 { 621 622 return (VOP_POLL(((struct vnode *)fp->f_data), events, l)); 623 } 624 625 /* 626 * File table vnode kqfilter routine. 627 */ 628 int 629 vn_kqfilter(struct file *fp, struct knote *kn) 630 { 631 632 return (VOP_KQFILTER((struct vnode *)fp->f_data, kn)); 633 } 634 635 /* 636 * Check that the vnode is still valid, and if so 637 * acquire requested lock. 638 */ 639 int 640 vn_lock(struct vnode *vp, int flags) 641 { 642 int error; 643 644 #if 0 645 KASSERT(vp->v_usecount > 0 || (flags & LK_INTERLOCK) != 0 646 || (vp->v_flag & VONWORKLST) != 0); 647 #endif 648 KASSERT((flags & 649 ~(LK_INTERLOCK|LK_SHARED|LK_EXCLUSIVE|LK_DRAIN|LK_NOWAIT|LK_RETRY| 650 LK_SETRECURSE|LK_CANRECURSE)) 651 == 0); 652 653 do { 654 if ((flags & LK_INTERLOCK) == 0) 655 simple_lock(&vp->v_interlock); 656 if (vp->v_flag & VXLOCK) { 657 if (flags & LK_NOWAIT) { 658 simple_unlock(&vp->v_interlock); 659 return EBUSY; 660 } 661 vp->v_flag |= VXWANT; 662 ltsleep(vp, PINOD | PNORELOCK, 663 "vn_lock", 0, &vp->v_interlock); 664 error = ENOENT; 665 } else { 666 error = VOP_LOCK(vp, 667 (flags & ~LK_RETRY) | LK_INTERLOCK); 668 if (error == 0 || error == EDEADLK || error == EBUSY) 669 return (error); 670 } 671 flags &= ~LK_INTERLOCK; 672 } while (flags & LK_RETRY); 673 return (error); 674 } 675 676 /* 677 * File table vnode close routine. 678 */ 679 static int 680 vn_closefile(struct file *fp, struct lwp *l) 681 { 682 683 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 684 fp->f_cred, l)); 685 } 686 687 /* 688 * Enable LK_CANRECURSE on lock. Return prior status. 689 */ 690 u_int 691 vn_setrecurse(struct vnode *vp) 692 { 693 struct lock *lkp = &vp->v_lock; 694 u_int retval = lkp->lk_flags & LK_CANRECURSE; 695 696 lkp->lk_flags |= LK_CANRECURSE; 697 return retval; 698 } 699 700 /* 701 * Called when done with locksetrecurse. 702 */ 703 void 704 vn_restorerecurse(struct vnode *vp, u_int flags) 705 { 706 struct lock *lkp = &vp->v_lock; 707 708 lkp->lk_flags &= ~LK_CANRECURSE; 709 lkp->lk_flags |= flags; 710 } 711 712 int 713 vn_cow_establish(struct vnode *vp, 714 int (*func)(void *, struct buf *), void *cookie) 715 { 716 int s; 717 struct spec_cow_entry *e; 718 719 MALLOC(e, struct spec_cow_entry *, sizeof(struct spec_cow_entry), 720 M_DEVBUF, M_WAITOK); 721 e->ce_func = func; 722 e->ce_cookie = cookie; 723 724 SPEC_COW_LOCK(vp->v_specinfo, s); 725 vp->v_spec_cow_req++; 726 while (vp->v_spec_cow_count > 0) 727 ltsleep(&vp->v_spec_cow_req, PRIBIO, "cowlist", 0, 728 &vp->v_spec_cow_slock); 729 730 SLIST_INSERT_HEAD(&vp->v_spec_cow_head, e, ce_list); 731 732 vp->v_spec_cow_req--; 733 if (vp->v_spec_cow_req == 0) 734 wakeup(&vp->v_spec_cow_req); 735 SPEC_COW_UNLOCK(vp->v_specinfo, s); 736 737 return 0; 738 } 739 740 int 741 vn_cow_disestablish(struct vnode *vp, 742 int (*func)(void *, struct buf *), void *cookie) 743 { 744 int s; 745 struct spec_cow_entry *e; 746 747 SPEC_COW_LOCK(vp->v_specinfo, s); 748 vp->v_spec_cow_req++; 749 while (vp->v_spec_cow_count > 0) 750 ltsleep(&vp->v_spec_cow_req, PRIBIO, "cowlist", 0, 751 &vp->v_spec_cow_slock); 752 753 SLIST_FOREACH(e, &vp->v_spec_cow_head, ce_list) 754 if (e->ce_func == func && e->ce_cookie == cookie) { 755 SLIST_REMOVE(&vp->v_spec_cow_head, e, 756 spec_cow_entry, ce_list); 757 FREE(e, M_DEVBUF); 758 break; 759 } 760 761 vp->v_spec_cow_req--; 762 if (vp->v_spec_cow_req == 0) 763 wakeup(&vp->v_spec_cow_req); 764 SPEC_COW_UNLOCK(vp->v_specinfo, s); 765 766 return e ? 0 : EINVAL; 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 l); 797 798 if ((ioflg & IO_NODELOCKED) == 0) 799 VOP_UNLOCK(vp, 0); 800 801 if (error == 0) 802 *buflen = *buflen - auio.uio_resid; 803 804 return (error); 805 } 806 807 /* 808 * XXX Failure mode if partially written? 809 */ 810 int 811 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 812 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 813 { 814 struct uio auio; 815 struct iovec aiov; 816 int error; 817 818 aiov.iov_len = buflen; 819 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 820 821 auio.uio_iov = &aiov; 822 auio.uio_iovcnt = 1; 823 auio.uio_rw = UIO_WRITE; 824 auio.uio_offset = 0; 825 auio.uio_resid = buflen; 826 UIO_SETUP_SYSSPACE(&auio); 827 828 if ((ioflg & IO_NODELOCKED) == 0) { 829 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 830 } 831 832 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, l); 833 834 if ((ioflg & IO_NODELOCKED) == 0) { 835 VOP_UNLOCK(vp, 0); 836 } 837 838 return (error); 839 } 840 841 int 842 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 843 const char *attrname, struct lwp *l) 844 { 845 int error; 846 847 if ((ioflg & IO_NODELOCKED) == 0) { 848 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 849 } 850 851 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL, l); 852 if (error == EOPNOTSUPP) 853 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, 854 NULL, l); 855 856 if ((ioflg & IO_NODELOCKED) == 0) { 857 VOP_UNLOCK(vp, 0); 858 } 859 860 return (error); 861 } 862 863 void 864 vn_ra_allocctx(struct vnode *vp) 865 { 866 struct uvm_ractx *ra = NULL; 867 868 if (vp->v_type != VREG) { 869 return; 870 } 871 if (vp->v_ractx != NULL) { 872 return; 873 } 874 simple_lock(&vp->v_interlock); 875 if (vp->v_ractx == NULL) { 876 simple_unlock(&vp->v_interlock); 877 ra = uvm_ra_allocctx(); 878 simple_lock(&vp->v_interlock); 879 if (ra != NULL && vp->v_ractx == NULL) { 880 vp->v_ractx = ra; 881 ra = NULL; 882 } 883 } 884 simple_unlock(&vp->v_interlock); 885 if (ra != NULL) { 886 uvm_ra_freectx(ra); 887 } 888 } 889
Indexes created Sun Sep 21 20:09:37 GMT 2025