1 /* $NetBSD: vfs_vnops.c,v 1.180 2010/11/19 06:44:45 dholland Exp $ */ 2 3 /*- 4 * Copyright (c) 2009 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Andrew Doran. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 29 * POSSIBILITY OF SUCH DAMAGE. 30 */ 31 32 /* 33 * Copyright (c) 1982, 1986, 1989, 1993 34 * The Regents of the University of California. All rights reserved. 35 * (c) UNIX System Laboratories, Inc. 36 * All or some portions of this file are derived from material licensed 37 * to the University of California by American Telephone and Telegraph 38 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 39 * the permission of UNIX System Laboratories, Inc. 40 * 41 * Redistribution and use in source and binary forms, with or without 42 * modification, are permitted provided that the following conditions 43 * are met: 44 * 1. Redistributions of source code must retain the above copyright 45 * notice, this list of conditions and the following disclaimer. 46 * 2. Redistributions in binary form must reproduce the above copyright 47 * notice, this list of conditions and the following disclaimer in the 48 * documentation and/or other materials provided with the distribution. 49 * 3. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)vfs_vnops.c 8.14 (Berkeley) 6/15/95 66 */ 67 68 #include <sys/cdefs.h> 69 __KERNEL_RCSID(0, "$NetBSD: vfs_vnops.c,v 1.180 2010/11/19 06:44:45 dholland Exp $"); 70 71 #include "veriexec.h" 72 73 #include <sys/param.h> 74 #include <sys/systm.h> 75 #include <sys/kernel.h> 76 #include <sys/file.h> 77 #include <sys/stat.h> 78 #include <sys/buf.h> 79 #include <sys/proc.h> 80 #include <sys/mount.h> 81 #include <sys/namei.h> 82 #include <sys/vnode.h> 83 #include <sys/ioctl.h> 84 #include <sys/tty.h> 85 #include <sys/poll.h> 86 #include <sys/kauth.h> 87 #include <sys/syslog.h> 88 #include <sys/fstrans.h> 89 #include <sys/atomic.h> 90 #include <sys/filedesc.h> 91 #include <sys/wapbl.h> 92 93 #include <miscfs/specfs/specdev.h> 94 #include <miscfs/fifofs/fifo.h> 95 96 #include <uvm/uvm_extern.h> 97 #include <uvm/uvm_readahead.h> 98 99 #ifdef UNION 100 #include <fs/union/union.h> 101 #endif 102 103 int (*vn_union_readdir_hook) (struct vnode **, struct file *, struct lwp *); 104 105 #include <sys/verified_exec.h> 106 107 static int vn_read(file_t *fp, off_t *offset, struct uio *uio, 108 kauth_cred_t cred, int flags); 109 static int vn_write(file_t *fp, off_t *offset, struct uio *uio, 110 kauth_cred_t cred, int flags); 111 static int vn_closefile(file_t *fp); 112 static int vn_poll(file_t *fp, int events); 113 static int vn_fcntl(file_t *fp, u_int com, void *data); 114 static int vn_statfile(file_t *fp, struct stat *sb); 115 static int vn_ioctl(file_t *fp, u_long com, void *data); 116 117 const struct fileops vnops = { 118 .fo_read = vn_read, 119 .fo_write = vn_write, 120 .fo_ioctl = vn_ioctl, 121 .fo_fcntl = vn_fcntl, 122 .fo_poll = vn_poll, 123 .fo_stat = vn_statfile, 124 .fo_close = vn_closefile, 125 .fo_kqfilter = vn_kqfilter, 126 .fo_restart = fnullop_restart, 127 }; 128 129 /* 130 * Common code for vnode open operations. 131 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 132 */ 133 int 134 vn_open(struct nameidata *ndp, int fmode, int cmode) 135 { 136 struct vnode *vp; 137 struct lwp *l = curlwp; 138 kauth_cred_t cred = l->l_cred; 139 struct vattr va; 140 int error; 141 const char *pathstring; 142 143 if ((fmode & (O_CREAT | O_DIRECTORY)) == (O_CREAT | O_DIRECTORY)) 144 return EINVAL; 145 146 ndp->ni_cnd.cn_flags &= TRYEMULROOT | NOCHROOT; 147 148 if (fmode & O_CREAT) { 149 ndp->ni_cnd.cn_nameiop = CREATE; 150 ndp->ni_cnd.cn_flags |= LOCKPARENT | LOCKLEAF; 151 if ((fmode & O_EXCL) == 0 && 152 ((fmode & O_NOFOLLOW) == 0)) 153 ndp->ni_cnd.cn_flags |= FOLLOW; 154 } else { 155 ndp->ni_cnd.cn_nameiop = LOOKUP; 156 ndp->ni_cnd.cn_flags |= LOCKLEAF; 157 if ((fmode & O_NOFOLLOW) == 0) 158 ndp->ni_cnd.cn_flags |= FOLLOW; 159 } 160 161 pathstring = pathbuf_stringcopy_get(ndp->ni_pathbuf); 162 if (pathstring == NULL) { 163 return ENOMEM; 164 } 165 166 error = namei(ndp); 167 if (error) 168 goto out; 169 170 vp = ndp->ni_vp; 171 172 #if NVERIEXEC > 0 173 error = veriexec_openchk(l, ndp->ni_vp, pathstring, fmode); 174 if (error) 175 goto bad; 176 #endif /* NVERIEXEC > 0 */ 177 178 if (fmode & O_CREAT) { 179 if (ndp->ni_vp == NULL) { 180 vattr_null(&va); 181 va.va_type = VREG; 182 va.va_mode = cmode; 183 if (fmode & O_EXCL) 184 va.va_vaflags |= VA_EXCLUSIVE; 185 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 186 &ndp->ni_cnd, &va); 187 if (error) 188 goto out; 189 fmode &= ~O_TRUNC; 190 vp = ndp->ni_vp; 191 } else { 192 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 193 if (ndp->ni_dvp == ndp->ni_vp) 194 vrele(ndp->ni_dvp); 195 else 196 vput(ndp->ni_dvp); 197 ndp->ni_dvp = NULL; 198 vp = ndp->ni_vp; 199 if (fmode & O_EXCL) { 200 error = EEXIST; 201 goto bad; 202 } 203 fmode &= ~O_CREAT; 204 } 205 } else { 206 vp = ndp->ni_vp; 207 } 208 if (vp->v_type == VSOCK) { 209 error = EOPNOTSUPP; 210 goto bad; 211 } 212 if (ndp->ni_vp->v_type == VLNK) { 213 error = EFTYPE; 214 goto bad; 215 } 216 217 if ((fmode & O_CREAT) == 0) { 218 error = vn_openchk(vp, cred, fmode); 219 if (error != 0) 220 goto bad; 221 } 222 223 if (fmode & O_TRUNC) { 224 vattr_null(&va); 225 va.va_size = 0; 226 error = VOP_SETATTR(vp, &va, cred); 227 if (error != 0) 228 goto bad; 229 } 230 if ((error = VOP_OPEN(vp, fmode, cred)) != 0) 231 goto bad; 232 if (fmode & FWRITE) { 233 mutex_enter(&vp->v_interlock); 234 vp->v_writecount++; 235 mutex_exit(&vp->v_interlock); 236 } 237 238 bad: 239 if (error) 240 vput(vp); 241 out: 242 pathbuf_stringcopy_put(ndp->ni_pathbuf, pathstring); 243 return (error); 244 } 245 246 /* 247 * Check for write permissions on the specified vnode. 248 * Prototype text segments cannot be written. 249 */ 250 int 251 vn_writechk(struct vnode *vp) 252 { 253 254 /* 255 * If the vnode is in use as a process's text, 256 * we can't allow writing. 257 */ 258 if (vp->v_iflag & VI_TEXT) 259 return (ETXTBSY); 260 return (0); 261 } 262 263 int 264 vn_openchk(struct vnode *vp, kauth_cred_t cred, int fflags) 265 { 266 int permbits = 0; 267 int error; 268 269 if ((fflags & O_DIRECTORY) != 0 && vp->v_type != VDIR) 270 return ENOTDIR; 271 272 if ((fflags & FREAD) != 0) { 273 permbits = VREAD; 274 } 275 if ((fflags & (FWRITE | O_TRUNC)) != 0) { 276 permbits |= VWRITE; 277 if (vp->v_type == VDIR) { 278 error = EISDIR; 279 goto bad; 280 } 281 error = vn_writechk(vp); 282 if (error != 0) 283 goto bad; 284 } 285 error = VOP_ACCESS(vp, permbits, cred); 286 bad: 287 return error; 288 } 289 290 /* 291 * Mark a vnode as having executable mappings. 292 */ 293 void 294 vn_markexec(struct vnode *vp) 295 { 296 297 if ((vp->v_iflag & VI_EXECMAP) != 0) { 298 /* Safe unlocked, as long as caller holds a reference. */ 299 return; 300 } 301 302 mutex_enter(&vp->v_interlock); 303 if ((vp->v_iflag & VI_EXECMAP) == 0) { 304 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 305 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 306 vp->v_iflag |= VI_EXECMAP; 307 } 308 mutex_exit(&vp->v_interlock); 309 } 310 311 /* 312 * Mark a vnode as being the text of a process. 313 * Fail if the vnode is currently writable. 314 */ 315 int 316 vn_marktext(struct vnode *vp) 317 { 318 319 if ((vp->v_iflag & (VI_TEXT|VI_EXECMAP)) == (VI_TEXT|VI_EXECMAP)) { 320 /* Safe unlocked, as long as caller holds a reference. */ 321 return (0); 322 } 323 324 mutex_enter(&vp->v_interlock); 325 if (vp->v_writecount != 0) { 326 KASSERT((vp->v_iflag & VI_TEXT) == 0); 327 mutex_exit(&vp->v_interlock); 328 return (ETXTBSY); 329 } 330 if ((vp->v_iflag & VI_EXECMAP) == 0) { 331 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 332 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 333 } 334 vp->v_iflag |= (VI_TEXT | VI_EXECMAP); 335 mutex_exit(&vp->v_interlock); 336 return (0); 337 } 338 339 /* 340 * Vnode close call 341 * 342 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 343 */ 344 int 345 vn_close(struct vnode *vp, int flags, kauth_cred_t cred) 346 { 347 int error; 348 349 if (flags & FWRITE) { 350 mutex_enter(&vp->v_interlock); 351 vp->v_writecount--; 352 mutex_exit(&vp->v_interlock); 353 } 354 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 355 error = VOP_CLOSE(vp, flags, cred); 356 vput(vp); 357 return (error); 358 } 359 360 static int 361 enforce_rlimit_fsize(struct vnode *vp, struct uio *uio, int ioflag) 362 { 363 struct lwp *l = curlwp; 364 off_t testoff; 365 366 if (uio->uio_rw != UIO_WRITE || vp->v_type != VREG) 367 return 0; 368 369 KASSERT(VOP_ISLOCKED(vp) == LK_EXCLUSIVE); 370 if (ioflag & IO_APPEND) 371 testoff = vp->v_size; 372 else 373 testoff = uio->uio_offset; 374 375 if (testoff + uio->uio_resid > 376 l->l_proc->p_rlimit[RLIMIT_FSIZE].rlim_cur) { 377 mutex_enter(proc_lock); 378 psignal(l->l_proc, SIGXFSZ); 379 mutex_exit(proc_lock); 380 return EFBIG; 381 } 382 383 return 0; 384 } 385 386 /* 387 * Package up an I/O request on a vnode into a uio and do it. 388 */ 389 int 390 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 391 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 392 struct lwp *l) 393 { 394 struct uio auio; 395 struct iovec aiov; 396 int error; 397 398 if ((ioflg & IO_NODELOCKED) == 0) { 399 if (rw == UIO_READ) { 400 vn_lock(vp, LK_SHARED | LK_RETRY); 401 } else /* UIO_WRITE */ { 402 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 403 } 404 } 405 auio.uio_iov = &aiov; 406 auio.uio_iovcnt = 1; 407 aiov.iov_base = base; 408 aiov.iov_len = len; 409 auio.uio_resid = len; 410 auio.uio_offset = offset; 411 auio.uio_rw = rw; 412 if (segflg == UIO_SYSSPACE) { 413 UIO_SETUP_SYSSPACE(&auio); 414 } else { 415 auio.uio_vmspace = l->l_proc->p_vmspace; 416 } 417 418 if ((error = enforce_rlimit_fsize(vp, &auio, ioflg)) != 0) 419 goto out; 420 421 if (rw == UIO_READ) { 422 error = VOP_READ(vp, &auio, ioflg, cred); 423 } else { 424 error = VOP_WRITE(vp, &auio, ioflg, cred); 425 } 426 427 if (aresid) 428 *aresid = auio.uio_resid; 429 else 430 if (auio.uio_resid && error == 0) 431 error = EIO; 432 433 out: 434 if ((ioflg & IO_NODELOCKED) == 0) { 435 VOP_UNLOCK(vp); 436 } 437 return (error); 438 } 439 440 int 441 vn_readdir(file_t *fp, char *bf, int segflg, u_int count, int *done, 442 struct lwp *l, off_t **cookies, int *ncookies) 443 { 444 struct vnode *vp = (struct vnode *)fp->f_data; 445 struct iovec aiov; 446 struct uio auio; 447 int error, eofflag; 448 449 /* Limit the size on any kernel buffers used by VOP_READDIR */ 450 count = min(MAXBSIZE, count); 451 452 unionread: 453 if (vp->v_type != VDIR) 454 return (EINVAL); 455 aiov.iov_base = bf; 456 aiov.iov_len = count; 457 auio.uio_iov = &aiov; 458 auio.uio_iovcnt = 1; 459 auio.uio_rw = UIO_READ; 460 if (segflg == UIO_SYSSPACE) { 461 UIO_SETUP_SYSSPACE(&auio); 462 } else { 463 KASSERT(l == curlwp); 464 auio.uio_vmspace = l->l_proc->p_vmspace; 465 } 466 auio.uio_resid = count; 467 vn_lock(vp, LK_SHARED | LK_RETRY); 468 auio.uio_offset = fp->f_offset; 469 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 470 ncookies); 471 mutex_enter(&fp->f_lock); 472 fp->f_offset = auio.uio_offset; 473 mutex_exit(&fp->f_lock); 474 VOP_UNLOCK(vp); 475 if (error) 476 return (error); 477 478 if (count == auio.uio_resid && vn_union_readdir_hook) { 479 struct vnode *ovp = vp; 480 481 error = (*vn_union_readdir_hook)(&vp, fp, l); 482 if (error) 483 return (error); 484 if (vp != ovp) 485 goto unionread; 486 } 487 488 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 489 (vp->v_mount->mnt_flag & MNT_UNION)) { 490 struct vnode *tvp = vp; 491 vp = vp->v_mount->mnt_vnodecovered; 492 vref(vp); 493 mutex_enter(&fp->f_lock); 494 fp->f_data = vp; 495 fp->f_offset = 0; 496 mutex_exit(&fp->f_lock); 497 vrele(tvp); 498 goto unionread; 499 } 500 *done = count - auio.uio_resid; 501 return error; 502 } 503 504 /* 505 * File table vnode read routine. 506 */ 507 static int 508 vn_read(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 509 int flags) 510 { 511 struct vnode *vp = (struct vnode *)fp->f_data; 512 int count, error, ioflag, fflag; 513 514 ioflag = IO_ADV_ENCODE(fp->f_advice); 515 fflag = fp->f_flag; 516 if (fflag & FNONBLOCK) 517 ioflag |= IO_NDELAY; 518 if ((fflag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 519 ioflag |= IO_SYNC; 520 if (fflag & FALTIO) 521 ioflag |= IO_ALTSEMANTICS; 522 if (fflag & FDIRECT) 523 ioflag |= IO_DIRECT; 524 vn_lock(vp, LK_SHARED | LK_RETRY); 525 uio->uio_offset = *offset; 526 count = uio->uio_resid; 527 error = VOP_READ(vp, uio, ioflag, cred); 528 if (flags & FOF_UPDATE_OFFSET) 529 *offset += count - uio->uio_resid; 530 VOP_UNLOCK(vp); 531 return (error); 532 } 533 534 /* 535 * File table vnode write routine. 536 */ 537 static int 538 vn_write(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 539 int flags) 540 { 541 struct vnode *vp = (struct vnode *)fp->f_data; 542 int count, error, ioflag, fflag; 543 544 ioflag = IO_ADV_ENCODE(fp->f_advice) | IO_UNIT; 545 fflag = fp->f_flag; 546 if (vp->v_type == VREG && (fflag & O_APPEND)) 547 ioflag |= IO_APPEND; 548 if (fflag & FNONBLOCK) 549 ioflag |= IO_NDELAY; 550 if (fflag & FFSYNC || 551 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 552 ioflag |= IO_SYNC; 553 else if (fflag & FDSYNC) 554 ioflag |= IO_DSYNC; 555 if (fflag & FALTIO) 556 ioflag |= IO_ALTSEMANTICS; 557 if (fflag & FDIRECT) 558 ioflag |= IO_DIRECT; 559 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 560 uio->uio_offset = *offset; 561 count = uio->uio_resid; 562 563 if ((error = enforce_rlimit_fsize(vp, uio, ioflag)) != 0) 564 goto out; 565 566 error = VOP_WRITE(vp, uio, ioflag, cred); 567 568 if (flags & FOF_UPDATE_OFFSET) { 569 if (ioflag & IO_APPEND) { 570 /* 571 * SUSv3 describes behaviour for count = 0 as following: 572 * "Before any action ... is taken, and if nbyte is zero 573 * and the file is a regular file, the write() function 574 * ... in the absence of errors ... shall return zero 575 * and have no other results." 576 */ 577 if (count) 578 *offset = uio->uio_offset; 579 } else 580 *offset += count - uio->uio_resid; 581 } 582 583 out: 584 VOP_UNLOCK(vp); 585 return (error); 586 } 587 588 /* 589 * File table vnode stat routine. 590 */ 591 static int 592 vn_statfile(file_t *fp, struct stat *sb) 593 { 594 struct vnode *vp = fp->f_data; 595 int error; 596 597 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 598 error = vn_stat(vp, sb); 599 VOP_UNLOCK(vp); 600 return error; 601 } 602 603 int 604 vn_stat(struct vnode *vp, struct stat *sb) 605 { 606 struct vattr va; 607 int error; 608 mode_t mode; 609 610 memset(&va, 0, sizeof(va)); 611 error = VOP_GETATTR(vp, &va, kauth_cred_get()); 612 if (error) 613 return (error); 614 /* 615 * Copy from vattr table 616 */ 617 memset(sb, 0, sizeof(*sb)); 618 sb->st_dev = va.va_fsid; 619 sb->st_ino = va.va_fileid; 620 mode = va.va_mode; 621 switch (vp->v_type) { 622 case VREG: 623 mode |= S_IFREG; 624 break; 625 case VDIR: 626 mode |= S_IFDIR; 627 break; 628 case VBLK: 629 mode |= S_IFBLK; 630 break; 631 case VCHR: 632 mode |= S_IFCHR; 633 break; 634 case VLNK: 635 mode |= S_IFLNK; 636 break; 637 case VSOCK: 638 mode |= S_IFSOCK; 639 break; 640 case VFIFO: 641 mode |= S_IFIFO; 642 break; 643 default: 644 return (EBADF); 645 }; 646 sb->st_mode = mode; 647 sb->st_nlink = va.va_nlink; 648 sb->st_uid = va.va_uid; 649 sb->st_gid = va.va_gid; 650 sb->st_rdev = va.va_rdev; 651 sb->st_size = va.va_size; 652 sb->st_atimespec = va.va_atime; 653 sb->st_mtimespec = va.va_mtime; 654 sb->st_ctimespec = va.va_ctime; 655 sb->st_birthtimespec = va.va_birthtime; 656 sb->st_blksize = va.va_blocksize; 657 sb->st_flags = va.va_flags; 658 sb->st_gen = 0; 659 sb->st_blocks = va.va_bytes / S_BLKSIZE; 660 return (0); 661 } 662 663 /* 664 * File table vnode fcntl routine. 665 */ 666 static int 667 vn_fcntl(file_t *fp, u_int com, void *data) 668 { 669 struct vnode *vp = fp->f_data; 670 int error; 671 672 error = VOP_FCNTL(vp, com, data, fp->f_flag, kauth_cred_get()); 673 return (error); 674 } 675 676 /* 677 * File table vnode ioctl routine. 678 */ 679 static int 680 vn_ioctl(file_t *fp, u_long com, void *data) 681 { 682 struct vnode *vp = fp->f_data, *ovp; 683 struct vattr vattr; 684 int error; 685 686 switch (vp->v_type) { 687 688 case VREG: 689 case VDIR: 690 if (com == FIONREAD) { 691 error = VOP_GETATTR(vp, &vattr, 692 kauth_cred_get()); 693 if (error) 694 return (error); 695 *(int *)data = vattr.va_size - fp->f_offset; 696 return (0); 697 } 698 if ((com == FIONWRITE) || (com == FIONSPACE)) { 699 /* 700 * Files don't have send queues, so there never 701 * are any bytes in them, nor is there any 702 * open space in them. 703 */ 704 *(int *)data = 0; 705 return (0); 706 } 707 if (com == FIOGETBMAP) { 708 daddr_t *block; 709 710 if (*(daddr_t *)data < 0) 711 return (EINVAL); 712 block = (daddr_t *)data; 713 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 714 } 715 if (com == OFIOGETBMAP) { 716 daddr_t ibn, obn; 717 718 if (*(int32_t *)data < 0) 719 return (EINVAL); 720 ibn = (daddr_t)*(int32_t *)data; 721 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 722 *(int32_t *)data = (int32_t)obn; 723 return error; 724 } 725 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 726 return (0); /* XXX */ 727 /* fall into ... */ 728 case VFIFO: 729 case VCHR: 730 case VBLK: 731 error = VOP_IOCTL(vp, com, data, fp->f_flag, 732 kauth_cred_get()); 733 if (error == 0 && com == TIOCSCTTY) { 734 vref(vp); 735 mutex_enter(proc_lock); 736 ovp = curproc->p_session->s_ttyvp; 737 curproc->p_session->s_ttyvp = vp; 738 mutex_exit(proc_lock); 739 if (ovp != NULL) 740 vrele(ovp); 741 } 742 return (error); 743 744 default: 745 return (EPASSTHROUGH); 746 } 747 } 748 749 /* 750 * File table vnode poll routine. 751 */ 752 static int 753 vn_poll(file_t *fp, int events) 754 { 755 756 return (VOP_POLL(fp->f_data, events)); 757 } 758 759 /* 760 * File table vnode kqfilter routine. 761 */ 762 int 763 vn_kqfilter(file_t *fp, struct knote *kn) 764 { 765 766 return (VOP_KQFILTER(fp->f_data, kn)); 767 } 768 769 /* 770 * Check that the vnode is still valid, and if so 771 * acquire requested lock. 772 */ 773 int 774 vn_lock(struct vnode *vp, int flags) 775 { 776 int error; 777 778 #if 0 779 KASSERT(vp->v_usecount > 0 || (vp->v_iflag & VI_ONWORKLST) != 0); 780 #endif 781 KASSERT((flags & ~(LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT|LK_RETRY)) == 0); 782 KASSERT(!mutex_owned(&vp->v_interlock)); 783 784 #ifdef DIAGNOSTIC 785 if (wapbl_vphaswapbl(vp)) 786 WAPBL_JUNLOCK_ASSERT(wapbl_vptomp(vp)); 787 #endif 788 789 do { 790 /* 791 * XXX PR 37706 forced unmount of file systems is unsafe. 792 * Race between vclean() and this the remaining problem. 793 */ 794 mutex_enter(&vp->v_interlock); 795 if (vp->v_iflag & VI_XLOCK) { 796 if (flags & LK_NOWAIT) { 797 mutex_exit(&vp->v_interlock); 798 return EBUSY; 799 } 800 vwait(vp, VI_XLOCK); 801 mutex_exit(&vp->v_interlock); 802 error = ENOENT; 803 } else { 804 mutex_exit(&vp->v_interlock); 805 error = VOP_LOCK(vp, (flags & ~LK_RETRY)); 806 if (error == 0 || error == EDEADLK || error == EBUSY) 807 return (error); 808 } 809 } while (flags & LK_RETRY); 810 return (error); 811 } 812 813 /* 814 * File table vnode close routine. 815 */ 816 static int 817 vn_closefile(file_t *fp) 818 { 819 820 return vn_close(fp->f_data, fp->f_flag, fp->f_cred); 821 } 822 823 /* 824 * Simplified in-kernel wrapper calls for extended attribute access. 825 * Both calls pass in a NULL credential, authorizing a "kernel" access. 826 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 827 */ 828 int 829 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 830 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 831 { 832 struct uio auio; 833 struct iovec aiov; 834 int error; 835 836 aiov.iov_len = *buflen; 837 aiov.iov_base = bf; 838 839 auio.uio_iov = &aiov; 840 auio.uio_iovcnt = 1; 841 auio.uio_rw = UIO_READ; 842 auio.uio_offset = 0; 843 auio.uio_resid = *buflen; 844 UIO_SETUP_SYSSPACE(&auio); 845 846 if ((ioflg & IO_NODELOCKED) == 0) 847 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 848 849 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 850 851 if ((ioflg & IO_NODELOCKED) == 0) 852 VOP_UNLOCK(vp); 853 854 if (error == 0) 855 *buflen = *buflen - auio.uio_resid; 856 857 return (error); 858 } 859 860 /* 861 * XXX Failure mode if partially written? 862 */ 863 int 864 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 865 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 866 { 867 struct uio auio; 868 struct iovec aiov; 869 int error; 870 871 aiov.iov_len = buflen; 872 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 873 874 auio.uio_iov = &aiov; 875 auio.uio_iovcnt = 1; 876 auio.uio_rw = UIO_WRITE; 877 auio.uio_offset = 0; 878 auio.uio_resid = buflen; 879 UIO_SETUP_SYSSPACE(&auio); 880 881 if ((ioflg & IO_NODELOCKED) == 0) { 882 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 883 } 884 885 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 886 887 if ((ioflg & IO_NODELOCKED) == 0) { 888 VOP_UNLOCK(vp); 889 } 890 891 return (error); 892 } 893 894 int 895 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 896 const char *attrname, struct lwp *l) 897 { 898 int error; 899 900 if ((ioflg & IO_NODELOCKED) == 0) { 901 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 902 } 903 904 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 905 if (error == EOPNOTSUPP) 906 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 907 908 if ((ioflg & IO_NODELOCKED) == 0) { 909 VOP_UNLOCK(vp); 910 } 911 912 return (error); 913 } 914 915 void 916 vn_ra_allocctx(struct vnode *vp) 917 { 918 struct uvm_ractx *ra = NULL; 919 920 KASSERT(mutex_owned(&vp->v_interlock)); 921 922 if (vp->v_type != VREG) { 923 return; 924 } 925 if (vp->v_ractx != NULL) { 926 return; 927 } 928 if (vp->v_ractx == NULL) { 929 mutex_exit(&vp->v_interlock); 930 ra = uvm_ra_allocctx(); 931 mutex_enter(&vp->v_interlock); 932 if (ra != NULL && vp->v_ractx == NULL) { 933 vp->v_ractx = ra; 934 ra = NULL; 935 } 936 } 937 if (ra != NULL) { 938 uvm_ra_freectx(ra); 939 } 940 } 941 942 int 943 vn_fifo_bypass(void *v) 944 { 945 struct vop_generic_args *ap = v; 946 947 return VOCALL(fifo_vnodeop_p, ap->a_desc->vdesc_offset, v); 948 } 949