1 /* $NetBSD: vfs_vnops.c,v 1.50 2001/09/21 08:02:55 chs 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. All advertising materials mentioning features or use of this software 21 * must display the following acknowledgement: 22 * This product includes software developed by the University of 23 * California, Berkeley and its contributors. 24 * 4. Neither the name of the University nor the names of its contributors 25 * may be used to endorse or promote products derived from this software 26 * without specific prior written permission. 27 * 28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 31 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 38 * SUCH DAMAGE. 39 * 40 * @(#)vfs_vnops.c 8.14 (Berkeley) 6/15/95 41 */ 42 43 #include "fs_union.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/mount.h> 53 #include <sys/namei.h> 54 #include <sys/vnode.h> 55 #include <sys/ioctl.h> 56 #include <sys/tty.h> 57 #include <sys/poll.h> 58 59 #include <uvm/uvm_extern.h> 60 61 #ifdef UNION 62 #include <miscfs/union/union.h> 63 #endif 64 65 static int vn_statfile __P((struct file *fp, struct stat *sb, struct proc *p)); 66 67 struct fileops vnops = { 68 vn_read, vn_write, vn_ioctl, vn_fcntl, vn_poll, 69 vn_statfile, vn_closefile 70 }; 71 72 /* 73 * Common code for vnode open operations. 74 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 75 */ 76 int 77 vn_open(ndp, fmode, cmode) 78 struct nameidata *ndp; 79 int fmode, cmode; 80 { 81 struct vnode *vp; 82 struct proc *p = ndp->ni_cnd.cn_proc; 83 struct ucred *cred = p->p_ucred; 84 struct vattr va; 85 int error; 86 87 if (fmode & O_CREAT) { 88 ndp->ni_cnd.cn_nameiop = CREATE; 89 ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF; 90 if ((fmode & O_EXCL) == 0 && 91 ((fmode & FNOSYMLINK) == 0)) 92 ndp->ni_cnd.cn_flags |= FOLLOW; 93 if ((error = namei(ndp)) != 0) 94 return (error); 95 if (ndp->ni_vp == NULL) { 96 VATTR_NULL(&va); 97 va.va_type = VREG; 98 va.va_mode = cmode; 99 if (fmode & O_EXCL) 100 va.va_vaflags |= VA_EXCLUSIVE; 101 VOP_LEASE(ndp->ni_dvp, p, cred, LEASE_WRITE); 102 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 103 &ndp->ni_cnd, &va); 104 if (error) 105 return (error); 106 fmode &= ~O_TRUNC; 107 vp = ndp->ni_vp; 108 } else { 109 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 110 if (ndp->ni_dvp == ndp->ni_vp) 111 vrele(ndp->ni_dvp); 112 else 113 vput(ndp->ni_dvp); 114 ndp->ni_dvp = NULL; 115 vp = ndp->ni_vp; 116 if (fmode & O_EXCL) { 117 error = EEXIST; 118 goto bad; 119 } 120 if (ndp->ni_vp->v_type == VLNK) { 121 error = EFTYPE; 122 goto bad; 123 } 124 fmode &= ~O_CREAT; 125 } 126 } else { 127 ndp->ni_cnd.cn_nameiop = LOOKUP; 128 ndp->ni_cnd.cn_flags = FOLLOW | LOCKLEAF; 129 if ((error = namei(ndp)) != 0) 130 return (error); 131 vp = ndp->ni_vp; 132 } 133 if (vp->v_type == VSOCK) { 134 error = EOPNOTSUPP; 135 goto bad; 136 } 137 if ((fmode & O_CREAT) == 0) { 138 if (fmode & FREAD) { 139 if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0) 140 goto bad; 141 } 142 if (fmode & (FWRITE | O_TRUNC)) { 143 if (vp->v_type == VDIR) { 144 error = EISDIR; 145 goto bad; 146 } 147 if ((error = vn_writechk(vp)) != 0 || 148 (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0) 149 goto bad; 150 } 151 } 152 if (fmode & O_TRUNC) { 153 VOP_UNLOCK(vp, 0); /* XXX */ 154 VOP_LEASE(vp, p, cred, LEASE_WRITE); 155 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */ 156 VATTR_NULL(&va); 157 va.va_size = 0; 158 if ((error = VOP_SETATTR(vp, &va, cred, p)) != 0) 159 goto bad; 160 } 161 if ((error = VOP_OPEN(vp, fmode, cred, p)) != 0) 162 goto bad; 163 if (vp->v_type == VREG && 164 uvn_attach(vp, fmode & FWRITE ? VM_PROT_WRITE : 0) == NULL) { 165 error = EIO; 166 goto bad; 167 } 168 if (fmode & FWRITE) 169 vp->v_writecount++; 170 171 return (0); 172 bad: 173 vput(vp); 174 return (error); 175 } 176 177 /* 178 * Check for write permissions on the specified vnode. 179 * Prototype text segments cannot be written. 180 */ 181 int 182 vn_writechk(vp) 183 struct vnode *vp; 184 { 185 186 /* 187 * If the vnode is in use as a process's text, 188 * we can't allow writing. 189 */ 190 if (vp->v_flag & VTEXT) 191 return (ETXTBSY); 192 return (0); 193 } 194 195 /* 196 * Mark a vnode as being the text image of a running process. 197 */ 198 void 199 vn_marktext(vp) 200 struct vnode *vp; 201 { 202 if ((vp->v_flag & VTEXT) == 0) { 203 uvmexp.vnodepages -= vp->v_uobj.uo_npages; 204 uvmexp.vtextpages += vp->v_uobj.uo_npages; 205 } 206 vp->v_flag |= VTEXT; 207 } 208 209 /* 210 * Vnode close call 211 * 212 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 213 */ 214 int 215 vn_close(vp, flags, cred, p) 216 struct vnode *vp; 217 int flags; 218 struct ucred *cred; 219 struct proc *p; 220 { 221 int error; 222 223 if (flags & FWRITE) 224 vp->v_writecount--; 225 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 226 error = VOP_CLOSE(vp, flags, cred, p); 227 vput(vp); 228 return (error); 229 } 230 231 /* 232 * Package up an I/O request on a vnode into a uio and do it. 233 */ 234 int 235 vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, cred, aresid, p) 236 enum uio_rw rw; 237 struct vnode *vp; 238 caddr_t base; 239 int len; 240 off_t offset; 241 enum uio_seg segflg; 242 int ioflg; 243 struct ucred *cred; 244 size_t *aresid; 245 struct proc *p; 246 { 247 struct uio auio; 248 struct iovec aiov; 249 int error; 250 251 if ((ioflg & IO_NODELOCKED) == 0) { 252 if (rw == UIO_READ) { 253 vn_lock(vp, LK_SHARED | LK_RETRY); 254 } else { 255 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 256 } 257 } 258 auio.uio_iov = &aiov; 259 auio.uio_iovcnt = 1; 260 aiov.iov_base = base; 261 aiov.iov_len = len; 262 auio.uio_resid = len; 263 auio.uio_offset = offset; 264 auio.uio_segflg = segflg; 265 auio.uio_rw = rw; 266 auio.uio_procp = p; 267 if (rw == UIO_READ) { 268 error = VOP_READ(vp, &auio, ioflg, cred); 269 } else { 270 error = VOP_WRITE(vp, &auio, ioflg, cred); 271 } 272 if (aresid) 273 *aresid = auio.uio_resid; 274 else 275 if (auio.uio_resid && error == 0) 276 error = EIO; 277 if ((ioflg & IO_NODELOCKED) == 0) 278 VOP_UNLOCK(vp, 0); 279 return (error); 280 } 281 282 int 283 vn_readdir(fp, buf, segflg, count, done, p, cookies, ncookies) 284 struct file *fp; 285 char *buf; 286 int segflg, *done, *ncookies; 287 u_int count; 288 struct proc *p; 289 off_t **cookies; 290 { 291 struct vnode *vp = (struct vnode *)fp->f_data; 292 struct iovec aiov; 293 struct uio auio; 294 int error, eofflag; 295 296 unionread: 297 if (vp->v_type != VDIR) 298 return (EINVAL); 299 aiov.iov_base = buf; 300 aiov.iov_len = count; 301 auio.uio_iov = &aiov; 302 auio.uio_iovcnt = 1; 303 auio.uio_rw = UIO_READ; 304 auio.uio_segflg = segflg; 305 auio.uio_procp = p; 306 auio.uio_resid = count; 307 vn_lock(vp, LK_SHARED | LK_RETRY); 308 auio.uio_offset = fp->f_offset; 309 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 310 ncookies); 311 fp->f_offset = auio.uio_offset; 312 VOP_UNLOCK(vp, 0); 313 if (error) 314 return (error); 315 316 #ifdef UNION 317 { 318 extern struct vnode *union_dircache __P((struct vnode *)); 319 320 if (count == auio.uio_resid && (vp->v_op == union_vnodeop_p)) { 321 struct vnode *lvp; 322 323 lvp = union_dircache(vp); 324 if (lvp != NULLVP) { 325 struct vattr va; 326 327 /* 328 * If the directory is opaque, 329 * then don't show lower entries 330 */ 331 error = VOP_GETATTR(vp, &va, fp->f_cred, p); 332 if (va.va_flags & OPAQUE) { 333 vput(lvp); 334 lvp = NULL; 335 } 336 } 337 338 if (lvp != NULLVP) { 339 error = VOP_OPEN(lvp, FREAD, fp->f_cred, p); 340 if (error) { 341 vput(lvp); 342 return (error); 343 } 344 VOP_UNLOCK(lvp, 0); 345 fp->f_data = (caddr_t) lvp; 346 fp->f_offset = 0; 347 error = vn_close(vp, FREAD, fp->f_cred, p); 348 if (error) 349 return (error); 350 vp = lvp; 351 goto unionread; 352 } 353 } 354 } 355 #endif /* UNION */ 356 357 if (count == auio.uio_resid && (vp->v_flag & VROOT) && 358 (vp->v_mount->mnt_flag & MNT_UNION)) { 359 struct vnode *tvp = vp; 360 vp = vp->v_mount->mnt_vnodecovered; 361 VREF(vp); 362 fp->f_data = (caddr_t) vp; 363 fp->f_offset = 0; 364 vrele(tvp); 365 goto unionread; 366 } 367 *done = count - auio.uio_resid; 368 return error; 369 } 370 371 /* 372 * File table vnode read routine. 373 */ 374 int 375 vn_read(fp, offset, uio, cred, flags) 376 struct file *fp; 377 off_t *offset; 378 struct uio *uio; 379 struct ucred *cred; 380 int flags; 381 { 382 struct vnode *vp = (struct vnode *)fp->f_data; 383 int count, error, ioflag = 0; 384 385 VOP_LEASE(vp, uio->uio_procp, cred, LEASE_READ); 386 if (fp->f_flag & FNONBLOCK) 387 ioflag |= IO_NDELAY; 388 if ((fp->f_flag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 389 ioflag |= IO_SYNC; 390 if (fp->f_flag & FALTIO) 391 ioflag |= IO_ALTSEMANTICS; 392 vn_lock(vp, LK_SHARED | LK_RETRY); 393 uio->uio_offset = *offset; 394 count = uio->uio_resid; 395 error = VOP_READ(vp, uio, ioflag, cred); 396 if (flags & FOF_UPDATE_OFFSET) 397 *offset += count - uio->uio_resid; 398 VOP_UNLOCK(vp, 0); 399 return (error); 400 } 401 402 /* 403 * File table vnode write routine. 404 */ 405 int 406 vn_write(fp, offset, uio, cred, flags) 407 struct file *fp; 408 off_t *offset; 409 struct uio *uio; 410 struct ucred *cred; 411 int flags; 412 { 413 struct vnode *vp = (struct vnode *)fp->f_data; 414 int count, error, ioflag = IO_UNIT; 415 416 if (vp->v_type == VREG && (fp->f_flag & O_APPEND)) 417 ioflag |= IO_APPEND; 418 if (fp->f_flag & FNONBLOCK) 419 ioflag |= IO_NDELAY; 420 if (fp->f_flag & FFSYNC || 421 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 422 ioflag |= IO_SYNC; 423 else if (fp->f_flag & FDSYNC) 424 ioflag |= IO_DSYNC; 425 if (fp->f_flag & FALTIO) 426 ioflag |= IO_ALTSEMANTICS; 427 VOP_LEASE(vp, uio->uio_procp, cred, LEASE_WRITE); 428 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 429 uio->uio_offset = *offset; 430 count = uio->uio_resid; 431 error = VOP_WRITE(vp, uio, ioflag, cred); 432 if (flags & FOF_UPDATE_OFFSET) { 433 if (ioflag & IO_APPEND) 434 *offset = uio->uio_offset; 435 else 436 *offset += count - uio->uio_resid; 437 } 438 VOP_UNLOCK(vp, 0); 439 return (error); 440 } 441 442 /* 443 * File table vnode stat routine. 444 */ 445 static int 446 vn_statfile(fp, sb, p) 447 struct file *fp; 448 struct stat *sb; 449 struct proc *p; 450 { 451 struct vnode *vp = (struct vnode *)fp->f_data; 452 453 return vn_stat(vp, sb, p); 454 } 455 456 int 457 vn_stat(fdata, sb, p) 458 void *fdata; 459 struct stat *sb; 460 struct proc *p; 461 { 462 struct vnode *vp = fdata; 463 struct vattr va; 464 int error; 465 mode_t mode; 466 467 error = VOP_GETATTR(vp, &va, p->p_ucred, p); 468 if (error) 469 return (error); 470 /* 471 * Copy from vattr table 472 */ 473 sb->st_dev = va.va_fsid; 474 sb->st_ino = va.va_fileid; 475 mode = va.va_mode; 476 switch (vp->v_type) { 477 case VREG: 478 mode |= S_IFREG; 479 break; 480 case VDIR: 481 mode |= S_IFDIR; 482 break; 483 case VBLK: 484 mode |= S_IFBLK; 485 break; 486 case VCHR: 487 mode |= S_IFCHR; 488 break; 489 case VLNK: 490 mode |= S_IFLNK; 491 break; 492 case VSOCK: 493 mode |= S_IFSOCK; 494 break; 495 case VFIFO: 496 mode |= S_IFIFO; 497 break; 498 default: 499 return (EBADF); 500 }; 501 sb->st_mode = mode; 502 sb->st_nlink = va.va_nlink; 503 sb->st_uid = va.va_uid; 504 sb->st_gid = va.va_gid; 505 sb->st_rdev = va.va_rdev; 506 sb->st_size = va.va_size; 507 sb->st_atimespec = va.va_atime; 508 sb->st_mtimespec = va.va_mtime; 509 sb->st_ctimespec = va.va_ctime; 510 sb->st_blksize = va.va_blocksize; 511 sb->st_flags = va.va_flags; 512 sb->st_gen = 0; 513 sb->st_blocks = va.va_bytes / S_BLKSIZE; 514 return (0); 515 } 516 517 /* 518 * File table vnode fcntl routine. 519 */ 520 int 521 vn_fcntl(fp, com, data, p) 522 struct file *fp; 523 u_int com; 524 caddr_t data; 525 struct proc *p; 526 { 527 struct vnode *vp = ((struct vnode *)fp->f_data); 528 int error; 529 530 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 531 error = VOP_FCNTL(vp, com, data, fp->f_flag, p->p_ucred, p); 532 VOP_UNLOCK(vp, 0); 533 return (error); 534 } 535 536 /* 537 * File table vnode ioctl routine. 538 */ 539 int 540 vn_ioctl(fp, com, data, p) 541 struct file *fp; 542 u_long com; 543 caddr_t data; 544 struct proc *p; 545 { 546 struct vnode *vp = ((struct vnode *)fp->f_data); 547 struct vattr vattr; 548 int error; 549 550 switch (vp->v_type) { 551 552 case VREG: 553 case VDIR: 554 if (com == FIONREAD) { 555 error = VOP_GETATTR(vp, &vattr, p->p_ucred, p); 556 if (error) 557 return (error); 558 *(int *)data = vattr.va_size - fp->f_offset; 559 return (0); 560 } 561 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 562 return (0); /* XXX */ 563 /* fall into ... */ 564 565 default: 566 return (ENOTTY); 567 568 case VFIFO: 569 case VCHR: 570 case VBLK: 571 error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p); 572 if (error == 0 && com == TIOCSCTTY) { 573 if (p->p_session->s_ttyvp) 574 vrele(p->p_session->s_ttyvp); 575 p->p_session->s_ttyvp = vp; 576 VREF(vp); 577 } 578 return (error); 579 } 580 } 581 582 /* 583 * File table vnode poll routine. 584 */ 585 int 586 vn_poll(fp, events, p) 587 struct file *fp; 588 int events; 589 struct proc *p; 590 { 591 592 return (VOP_POLL(((struct vnode *)fp->f_data), events, p)); 593 } 594 595 /* 596 * Check that the vnode is still valid, and if so 597 * acquire requested lock. 598 */ 599 int 600 vn_lock(vp, flags) 601 struct vnode *vp; 602 int flags; 603 { 604 int error; 605 606 do { 607 if ((flags & LK_INTERLOCK) == 0) 608 simple_lock(&vp->v_interlock); 609 if (vp->v_flag & VXLOCK) { 610 if (flags & LK_NOWAIT) { 611 simple_unlock(&vp->v_interlock); 612 return EBUSY; 613 } 614 vp->v_flag |= VXWANT; 615 ltsleep(vp, PINOD | PNORELOCK, 616 "vn_lock", 0, &vp->v_interlock); 617 error = ENOENT; 618 } else { 619 error = VOP_LOCK(vp, flags | LK_INTERLOCK); 620 if (error == 0 || error == EDEADLK || error == EBUSY) 621 return (error); 622 } 623 flags &= ~LK_INTERLOCK; 624 } while (flags & LK_RETRY); 625 return (error); 626 } 627 628 /* 629 * File table vnode close routine. 630 */ 631 int 632 vn_closefile(fp, p) 633 struct file *fp; 634 struct proc *p; 635 { 636 637 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 638 fp->f_cred, p)); 639 } 640 641 /* 642 * Enable LK_CANRECURSE on lock. Return prior status. 643 */ 644 u_int 645 vn_setrecurse(vp) 646 struct vnode *vp; 647 { 648 struct lock *lkp = &vp->v_lock; 649 u_int retval = lkp->lk_flags & LK_CANRECURSE; 650 651 lkp->lk_flags |= LK_CANRECURSE; 652 return retval; 653 } 654 655 /* 656 * Called when done with locksetrecurse. 657 */ 658 void 659 vn_restorerecurse(vp, flags) 660 struct vnode *vp; 661 u_int flags; 662 { 663 struct lock *lkp = &vp->v_lock; 664 665 lkp->lk_flags &= ~LK_CANRECURSE; 666 lkp->lk_flags |= flags; 667 } 668