vfs_vnops.c revision 1.135.4.1
1 /* $NetBSD: vfs_vnops.c,v 1.135.4.1 2007/07/11 20:10:25 mjf 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 1.135.4.1 2007/07/11 20:10:25 mjf 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 (vp->v_type == VREG && 206 uvn_attach(vp, fmode & FWRITE ? VM_PROT_WRITE : 0) == NULL) { 207 error = EIO; 208 goto bad; 209 } 210 if (fmode & FWRITE) 211 vp->v_writecount++; 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_flag & VTEXT) 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 if ((vp->v_flag & VEXECMAP) == 0) { 245 uvmexp.filepages -= vp->v_uobj.uo_npages; 246 uvmexp.execpages += vp->v_uobj.uo_npages; 247 } 248 vp->v_flag |= VEXECMAP; 249 } 250 251 /* 252 * Mark a vnode as being the text of a process. 253 * Fail if the vnode is currently writable. 254 */ 255 int 256 vn_marktext(struct vnode *vp) 257 { 258 259 if (vp->v_writecount != 0) { 260 KASSERT((vp->v_flag & VTEXT) == 0); 261 return (ETXTBSY); 262 } 263 vp->v_flag |= VTEXT; 264 vn_markexec(vp); 265 return (0); 266 } 267 268 /* 269 * Vnode close call 270 * 271 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 272 */ 273 int 274 vn_close(struct vnode *vp, int flags, kauth_cred_t cred, struct lwp *l) 275 { 276 int error; 277 278 if (flags & FWRITE) 279 vp->v_writecount--; 280 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 281 error = VOP_CLOSE(vp, flags, cred, l); 282 vput(vp); 283 return (error); 284 } 285 286 /* 287 * Package up an I/O request on a vnode into a uio and do it. 288 */ 289 int 290 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 291 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 292 struct lwp *l) 293 { 294 struct uio auio; 295 struct iovec aiov; 296 int error; 297 298 if ((ioflg & IO_NODELOCKED) == 0) { 299 if (rw == UIO_READ) { 300 vn_lock(vp, LK_SHARED | LK_RETRY); 301 } else /* UIO_WRITE */ { 302 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 303 } 304 } 305 auio.uio_iov = &aiov; 306 auio.uio_iovcnt = 1; 307 aiov.iov_base = base; 308 aiov.iov_len = len; 309 auio.uio_resid = len; 310 auio.uio_offset = offset; 311 auio.uio_rw = rw; 312 if (segflg == UIO_SYSSPACE) { 313 UIO_SETUP_SYSSPACE(&auio); 314 } else { 315 auio.uio_vmspace = l->l_proc->p_vmspace; 316 } 317 if (rw == UIO_READ) { 318 error = VOP_READ(vp, &auio, ioflg, cred); 319 } else { 320 error = VOP_WRITE(vp, &auio, ioflg, cred); 321 } 322 if (aresid) 323 *aresid = auio.uio_resid; 324 else 325 if (auio.uio_resid && error == 0) 326 error = EIO; 327 if ((ioflg & IO_NODELOCKED) == 0) { 328 VOP_UNLOCK(vp, 0); 329 } 330 return (error); 331 } 332 333 int 334 vn_readdir(struct file *fp, char *bf, int segflg, u_int count, int *done, 335 struct lwp *l, off_t **cookies, int *ncookies) 336 { 337 struct vnode *vp = (struct vnode *)fp->f_data; 338 struct iovec aiov; 339 struct uio auio; 340 int error, eofflag; 341 342 /* Limit the size on any kernel buffers used by VOP_READDIR */ 343 count = min(MAXBSIZE, count); 344 345 unionread: 346 if (vp->v_type != VDIR) 347 return (EINVAL); 348 aiov.iov_base = bf; 349 aiov.iov_len = count; 350 auio.uio_iov = &aiov; 351 auio.uio_iovcnt = 1; 352 auio.uio_rw = UIO_READ; 353 if (segflg == UIO_SYSSPACE) { 354 UIO_SETUP_SYSSPACE(&auio); 355 } else { 356 KASSERT(l == curlwp); 357 auio.uio_vmspace = l->l_proc->p_vmspace; 358 } 359 auio.uio_resid = count; 360 vn_lock(vp, LK_SHARED | LK_RETRY); 361 auio.uio_offset = fp->f_offset; 362 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 363 ncookies); 364 fp->f_offset = auio.uio_offset; 365 VOP_UNLOCK(vp, 0); 366 if (error) 367 return (error); 368 369 #if defined(UNION) || defined(LKM) 370 if (count == auio.uio_resid && vn_union_readdir_hook) { 371 struct vnode *ovp = vp; 372 373 error = (*vn_union_readdir_hook)(&vp, fp, l); 374 if (error) 375 return (error); 376 if (vp != ovp) 377 goto unionread; 378 } 379 #endif /* UNION || LKM */ 380 381 if (count == auio.uio_resid && (vp->v_flag & VROOT) && 382 (vp->v_mount->mnt_flag & MNT_UNION)) { 383 struct vnode *tvp = vp; 384 vp = vp->v_mount->mnt_vnodecovered; 385 VREF(vp); 386 fp->f_data = vp; 387 fp->f_offset = 0; 388 vrele(tvp); 389 goto unionread; 390 } 391 *done = count - auio.uio_resid; 392 return error; 393 } 394 395 /* 396 * File table vnode read routine. 397 */ 398 static int 399 vn_read(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 400 int flags) 401 { 402 struct vnode *vp = (struct vnode *)fp->f_data; 403 int count, error, ioflag; 404 struct lwp *l = curlwp; 405 406 VOP_LEASE(vp, l, cred, LEASE_READ); 407 ioflag = IO_ADV_ENCODE(fp->f_advice); 408 if (fp->f_flag & FNONBLOCK) 409 ioflag |= IO_NDELAY; 410 if ((fp->f_flag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 411 ioflag |= IO_SYNC; 412 if (fp->f_flag & FALTIO) 413 ioflag |= IO_ALTSEMANTICS; 414 if (fp->f_flag & FDIRECT) 415 ioflag |= IO_DIRECT; 416 vn_lock(vp, LK_SHARED | LK_RETRY); 417 uio->uio_offset = *offset; 418 count = uio->uio_resid; 419 error = VOP_READ(vp, uio, ioflag, cred); 420 if (flags & FOF_UPDATE_OFFSET) 421 *offset += count - uio->uio_resid; 422 VOP_UNLOCK(vp, 0); 423 return (error); 424 } 425 426 /* 427 * File table vnode write routine. 428 */ 429 static int 430 vn_write(struct file *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 431 int flags) 432 { 433 struct vnode *vp = (struct vnode *)fp->f_data; 434 int count, error, ioflag = IO_UNIT; 435 struct lwp *l = curlwp; 436 437 if (vp->v_type == VREG && (fp->f_flag & O_APPEND)) 438 ioflag |= IO_APPEND; 439 if (fp->f_flag & FNONBLOCK) 440 ioflag |= IO_NDELAY; 441 if (fp->f_flag & FFSYNC || 442 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 443 ioflag |= IO_SYNC; 444 else if (fp->f_flag & FDSYNC) 445 ioflag |= IO_DSYNC; 446 if (fp->f_flag & FALTIO) 447 ioflag |= IO_ALTSEMANTICS; 448 if (fp->f_flag & FDIRECT) 449 ioflag |= IO_DIRECT; 450 VOP_LEASE(vp, l, cred, LEASE_WRITE); 451 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 452 uio->uio_offset = *offset; 453 count = uio->uio_resid; 454 error = VOP_WRITE(vp, uio, ioflag, cred); 455 if (flags & FOF_UPDATE_OFFSET) { 456 if (ioflag & IO_APPEND) 457 *offset = uio->uio_offset; 458 else 459 *offset += count - uio->uio_resid; 460 } 461 VOP_UNLOCK(vp, 0); 462 return (error); 463 } 464 465 /* 466 * File table vnode stat routine. 467 */ 468 static int 469 vn_statfile(struct file *fp, struct stat *sb, struct lwp *l) 470 { 471 struct vnode *vp = (struct vnode *)fp->f_data; 472 473 return vn_stat(vp, sb, l); 474 } 475 476 int 477 vn_stat(struct vnode *vp, struct stat *sb, struct lwp *l) 478 { 479 struct vattr va; 480 int error; 481 mode_t mode; 482 483 error = VOP_GETATTR(vp, &va, l->l_cred, l); 484 if (error) 485 return (error); 486 /* 487 * Copy from vattr table 488 */ 489 sb->st_dev = va.va_fsid; 490 sb->st_ino = va.va_fileid; 491 mode = va.va_mode; 492 switch (vp->v_type) { 493 case VREG: 494 mode |= S_IFREG; 495 break; 496 case VDIR: 497 mode |= S_IFDIR; 498 break; 499 case VBLK: 500 mode |= S_IFBLK; 501 break; 502 case VCHR: 503 mode |= S_IFCHR; 504 break; 505 case VLNK: 506 mode |= S_IFLNK; 507 break; 508 case VSOCK: 509 mode |= S_IFSOCK; 510 break; 511 case VFIFO: 512 mode |= S_IFIFO; 513 break; 514 default: 515 return (EBADF); 516 }; 517 sb->st_mode = mode; 518 sb->st_nlink = va.va_nlink; 519 sb->st_uid = va.va_uid; 520 sb->st_gid = va.va_gid; 521 sb->st_rdev = va.va_rdev; 522 sb->st_size = va.va_size; 523 sb->st_atimespec = va.va_atime; 524 sb->st_mtimespec = va.va_mtime; 525 sb->st_ctimespec = va.va_ctime; 526 sb->st_birthtimespec = va.va_birthtime; 527 sb->st_blksize = va.va_blocksize; 528 sb->st_flags = va.va_flags; 529 sb->st_gen = 0; 530 sb->st_blocks = va.va_bytes / S_BLKSIZE; 531 return (0); 532 } 533 534 /* 535 * File table vnode fcntl routine. 536 */ 537 static int 538 vn_fcntl(struct file *fp, u_int com, void *data, struct lwp *l) 539 { 540 struct vnode *vp = ((struct vnode *)fp->f_data); 541 int error; 542 543 error = VOP_FCNTL(vp, com, data, fp->f_flag, l->l_cred, l); 544 return (error); 545 } 546 547 /* 548 * File table vnode ioctl routine. 549 */ 550 static int 551 vn_ioctl(struct file *fp, u_long com, void *data, struct lwp *l) 552 { 553 struct vnode *vp = ((struct vnode *)fp->f_data), *ovp; 554 struct proc *p = l->l_proc; 555 struct vattr vattr; 556 int error; 557 558 switch (vp->v_type) { 559 560 case VREG: 561 case VDIR: 562 if (com == FIONREAD) { 563 error = VOP_GETATTR(vp, &vattr, l->l_cred, l); 564 if (error) 565 return (error); 566 *(int *)data = vattr.va_size - fp->f_offset; 567 return (0); 568 } 569 if ((com == FIONWRITE) || (com == FIONSPACE)) { 570 /* 571 * Files don't have send queues, so there never 572 * are any bytes in them, nor is there any 573 * open space in them. 574 */ 575 *(int *)data = 0; 576 return (0); 577 } 578 if (com == FIOGETBMAP) { 579 daddr_t *block; 580 581 if (*(daddr_t *)data < 0) 582 return (EINVAL); 583 block = (daddr_t *)data; 584 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 585 } 586 if (com == OFIOGETBMAP) { 587 daddr_t ibn, obn; 588 589 if (*(int32_t *)data < 0) 590 return (EINVAL); 591 ibn = (daddr_t)*(int32_t *)data; 592 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 593 *(int32_t *)data = (int32_t)obn; 594 return error; 595 } 596 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 597 return (0); /* XXX */ 598 /* fall into ... */ 599 case VFIFO: 600 case VCHR: 601 case VBLK: 602 error = VOP_IOCTL(vp, com, data, fp->f_flag, 603 l->l_cred, l); 604 if (error == 0 && com == TIOCSCTTY) { 605 VREF(vp); 606 mutex_enter(&proclist_lock); 607 ovp = p->p_session->s_ttyvp; 608 p->p_session->s_ttyvp = vp; 609 mutex_exit(&proclist_lock); 610 if (ovp != NULL) 611 vrele(ovp); 612 } 613 return (error); 614 615 default: 616 return (EPASSTHROUGH); 617 } 618 } 619 620 /* 621 * File table vnode poll routine. 622 */ 623 static int 624 vn_poll(struct file *fp, int events, struct lwp *l) 625 { 626 627 return (VOP_POLL(((struct vnode *)fp->f_data), events, l)); 628 } 629 630 /* 631 * File table vnode kqfilter routine. 632 */ 633 int 634 vn_kqfilter(struct file *fp, struct knote *kn) 635 { 636 637 return (VOP_KQFILTER((struct vnode *)fp->f_data, kn)); 638 } 639 640 /* 641 * Check that the vnode is still valid, and if so 642 * acquire requested lock. 643 */ 644 int 645 vn_lock(struct vnode *vp, int flags) 646 { 647 int error; 648 649 #if 0 650 KASSERT(vp->v_usecount > 0 || (flags & LK_INTERLOCK) != 0 651 || (vp->v_flag & VONWORKLST) != 0); 652 #endif 653 KASSERT((flags & 654 ~(LK_INTERLOCK|LK_SHARED|LK_EXCLUSIVE|LK_DRAIN|LK_NOWAIT|LK_RETRY| 655 LK_SETRECURSE|LK_CANRECURSE)) 656 == 0); 657 658 do { 659 if ((flags & LK_INTERLOCK) == 0) 660 simple_lock(&vp->v_interlock); 661 if (vp->v_flag & VXLOCK) { 662 if (flags & LK_NOWAIT) { 663 simple_unlock(&vp->v_interlock); 664 return EBUSY; 665 } 666 vp->v_flag |= VXWANT; 667 ltsleep(vp, PINOD | PNORELOCK, 668 "vn_lock", 0, &vp->v_interlock); 669 error = ENOENT; 670 } else { 671 error = VOP_LOCK(vp, 672 (flags & ~LK_RETRY) | LK_INTERLOCK); 673 if (error == 0 || error == EDEADLK || error == EBUSY) 674 return (error); 675 } 676 flags &= ~LK_INTERLOCK; 677 } while (flags & LK_RETRY); 678 return (error); 679 } 680 681 /* 682 * File table vnode close routine. 683 */ 684 static int 685 vn_closefile(struct file *fp, struct lwp *l) 686 { 687 688 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 689 fp->f_cred, l)); 690 } 691 692 /* 693 * Enable LK_CANRECURSE on lock. Return prior status. 694 */ 695 u_int 696 vn_setrecurse(struct vnode *vp) 697 { 698 struct lock *lkp = &vp->v_lock; 699 u_int retval = lkp->lk_flags & LK_CANRECURSE; 700 701 lkp->lk_flags |= LK_CANRECURSE; 702 return retval; 703 } 704 705 /* 706 * Called when done with locksetrecurse. 707 */ 708 void 709 vn_restorerecurse(struct vnode *vp, u_int flags) 710 { 711 struct lock *lkp = &vp->v_lock; 712 713 lkp->lk_flags &= ~LK_CANRECURSE; 714 lkp->lk_flags |= flags; 715 } 716 717 int 718 vn_cow_establish(struct vnode *vp, 719 int (*func)(void *, struct buf *), void *cookie) 720 { 721 int s; 722 struct spec_cow_entry *e; 723 724 MALLOC(e, struct spec_cow_entry *, sizeof(struct spec_cow_entry), 725 M_DEVBUF, M_WAITOK); 726 e->ce_func = func; 727 e->ce_cookie = cookie; 728 729 SPEC_COW_LOCK(vp->v_specinfo, s); 730 vp->v_spec_cow_req++; 731 while (vp->v_spec_cow_count > 0) 732 ltsleep(&vp->v_spec_cow_req, PRIBIO, "cowlist", 0, 733 &vp->v_spec_cow_slock); 734 735 SLIST_INSERT_HEAD(&vp->v_spec_cow_head, e, ce_list); 736 737 vp->v_spec_cow_req--; 738 if (vp->v_spec_cow_req == 0) 739 wakeup(&vp->v_spec_cow_req); 740 SPEC_COW_UNLOCK(vp->v_specinfo, s); 741 742 return 0; 743 } 744 745 int 746 vn_cow_disestablish(struct vnode *vp, 747 int (*func)(void *, struct buf *), void *cookie) 748 { 749 int s; 750 struct spec_cow_entry *e; 751 752 SPEC_COW_LOCK(vp->v_specinfo, s); 753 vp->v_spec_cow_req++; 754 while (vp->v_spec_cow_count > 0) 755 ltsleep(&vp->v_spec_cow_req, PRIBIO, "cowlist", 0, 756 &vp->v_spec_cow_slock); 757 758 SLIST_FOREACH(e, &vp->v_spec_cow_head, ce_list) 759 if (e->ce_func == func && e->ce_cookie == cookie) { 760 SLIST_REMOVE(&vp->v_spec_cow_head, e, 761 spec_cow_entry, ce_list); 762 FREE(e, M_DEVBUF); 763 break; 764 } 765 766 vp->v_spec_cow_req--; 767 if (vp->v_spec_cow_req == 0) 768 wakeup(&vp->v_spec_cow_req); 769 SPEC_COW_UNLOCK(vp->v_specinfo, s); 770 771 return e ? 0 : EINVAL; 772 } 773 774 /* 775 * Simplified in-kernel wrapper calls for extended attribute access. 776 * Both calls pass in a NULL credential, authorizing a "kernel" access. 777 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 778 */ 779 int 780 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 781 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 782 { 783 struct uio auio; 784 struct iovec aiov; 785 int error; 786 787 aiov.iov_len = *buflen; 788 aiov.iov_base = bf; 789 790 auio.uio_iov = &aiov; 791 auio.uio_iovcnt = 1; 792 auio.uio_rw = UIO_READ; 793 auio.uio_offset = 0; 794 auio.uio_resid = *buflen; 795 UIO_SETUP_SYSSPACE(&auio); 796 797 if ((ioflg & IO_NODELOCKED) == 0) 798 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 799 800 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL, 801 l); 802 803 if ((ioflg & IO_NODELOCKED) == 0) 804 VOP_UNLOCK(vp, 0); 805 806 if (error == 0) 807 *buflen = *buflen - auio.uio_resid; 808 809 return (error); 810 } 811 812 /* 813 * XXX Failure mode if partially written? 814 */ 815 int 816 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 817 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 818 { 819 struct uio auio; 820 struct iovec aiov; 821 int error; 822 823 aiov.iov_len = buflen; 824 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 825 826 auio.uio_iov = &aiov; 827 auio.uio_iovcnt = 1; 828 auio.uio_rw = UIO_WRITE; 829 auio.uio_offset = 0; 830 auio.uio_resid = buflen; 831 UIO_SETUP_SYSSPACE(&auio); 832 833 if ((ioflg & IO_NODELOCKED) == 0) { 834 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 835 } 836 837 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, l); 838 839 if ((ioflg & IO_NODELOCKED) == 0) { 840 VOP_UNLOCK(vp, 0); 841 } 842 843 return (error); 844 } 845 846 int 847 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 848 const char *attrname, struct lwp *l) 849 { 850 int error; 851 852 if ((ioflg & IO_NODELOCKED) == 0) { 853 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 854 } 855 856 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL, l); 857 if (error == EOPNOTSUPP) 858 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, 859 NULL, l); 860 861 if ((ioflg & IO_NODELOCKED) == 0) { 862 VOP_UNLOCK(vp, 0); 863 } 864 865 return (error); 866 } 867 868 void 869 vn_ra_allocctx(struct vnode *vp) 870 { 871 struct uvm_ractx *ra = NULL; 872 873 if (vp->v_type != VREG) { 874 return; 875 } 876 if (vp->v_ractx != NULL) { 877 return; 878 } 879 simple_lock(&vp->v_interlock); 880 if (vp->v_ractx == NULL) { 881 simple_unlock(&vp->v_interlock); 882 ra = uvm_ra_allocctx(); 883 simple_lock(&vp->v_interlock); 884 if (ra != NULL && vp->v_ractx == NULL) { 885 vp->v_ractx = ra; 886 ra = NULL; 887 } 888 } 889 simple_unlock(&vp->v_interlock); 890 if (ra != NULL) { 891 uvm_ra_freectx(ra); 892 } 893 } 894
Indexes created Sat Sep 20 22:09:52 GMT 2025