vfs_vnops.c revision 1.188
1/* $NetBSD: vfs_vnops.c,v 1.188 2014/01/23 10:13:57 hannken 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 1.188 2014/01/23 10:13:57 hannken 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 103int (*vn_union_readdir_hook) (struct vnode **, struct file *, struct lwp *); 104 105#include <sys/verified_exec.h> 106 107static int vn_read(file_t *fp, off_t *offset, struct uio *uio, 108 kauth_cred_t cred, int flags); 109static int vn_write(file_t *fp, off_t *offset, struct uio *uio, 110 kauth_cred_t cred, int flags); 111static int vn_closefile(file_t *fp); 112static int vn_poll(file_t *fp, int events); 113static int vn_fcntl(file_t *fp, u_int com, void *data); 114static int vn_statfile(file_t *fp, struct stat *sb); 115static int vn_ioctl(file_t *fp, u_long com, void *data); 116 117const 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 */ 133int 134vn_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 vput(ndp->ni_dvp); 188 if (error) 189 goto out; 190 fmode &= ~O_TRUNC; 191 vp = ndp->ni_vp; 192 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 193 } else { 194 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 195 if (ndp->ni_dvp == ndp->ni_vp) 196 vrele(ndp->ni_dvp); 197 else 198 vput(ndp->ni_dvp); 199 ndp->ni_dvp = NULL; 200 vp = ndp->ni_vp; 201 if (fmode & O_EXCL) { 202 error = EEXIST; 203 goto bad; 204 } 205 fmode &= ~O_CREAT; 206 } 207 } else { 208 vp = ndp->ni_vp; 209 } 210 if (vp->v_type == VSOCK) { 211 error = EOPNOTSUPP; 212 goto bad; 213 } 214 if (ndp->ni_vp->v_type == VLNK) { 215 error = EFTYPE; 216 goto bad; 217 } 218 219 if ((fmode & O_CREAT) == 0) { 220 error = vn_openchk(vp, cred, fmode); 221 if (error != 0) 222 goto bad; 223 } 224 225 if (fmode & O_TRUNC) { 226 vattr_null(&va); 227 va.va_size = 0; 228 error = VOP_SETATTR(vp, &va, cred); 229 if (error != 0) 230 goto bad; 231 } 232 if ((error = VOP_OPEN(vp, fmode, cred)) != 0) 233 goto bad; 234 if (fmode & FWRITE) { 235 mutex_enter(vp->v_interlock); 236 vp->v_writecount++; 237 mutex_exit(vp->v_interlock); 238 } 239 240bad: 241 if (error) 242 vput(vp); 243out: 244 pathbuf_stringcopy_put(ndp->ni_pathbuf, pathstring); 245 return (error); 246} 247 248/* 249 * Check for write permissions on the specified vnode. 250 * Prototype text segments cannot be written. 251 */ 252int 253vn_writechk(struct vnode *vp) 254{ 255 256 /* 257 * If the vnode is in use as a process's text, 258 * we can't allow writing. 259 */ 260 if (vp->v_iflag & VI_TEXT) 261 return (ETXTBSY); 262 return (0); 263} 264 265int 266vn_openchk(struct vnode *vp, kauth_cred_t cred, int fflags) 267{ 268 int permbits = 0; 269 int error; 270 271 if ((fflags & O_DIRECTORY) != 0 && vp->v_type != VDIR) 272 return ENOTDIR; 273 274 if ((fflags & FREAD) != 0) { 275 permbits = VREAD; 276 } 277 if ((fflags & (FWRITE | O_TRUNC)) != 0) { 278 permbits |= VWRITE; 279 if (vp->v_type == VDIR) { 280 error = EISDIR; 281 goto bad; 282 } 283 error = vn_writechk(vp); 284 if (error != 0) 285 goto bad; 286 } 287 error = VOP_ACCESS(vp, permbits, cred); 288bad: 289 return error; 290} 291 292/* 293 * Mark a vnode as having executable mappings. 294 */ 295void 296vn_markexec(struct vnode *vp) 297{ 298 299 if ((vp->v_iflag & VI_EXECMAP) != 0) { 300 /* Safe unlocked, as long as caller holds a reference. */ 301 return; 302 } 303 304 mutex_enter(vp->v_interlock); 305 if ((vp->v_iflag & VI_EXECMAP) == 0) { 306 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 307 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 308 vp->v_iflag |= VI_EXECMAP; 309 } 310 mutex_exit(vp->v_interlock); 311} 312 313/* 314 * Mark a vnode as being the text of a process. 315 * Fail if the vnode is currently writable. 316 */ 317int 318vn_marktext(struct vnode *vp) 319{ 320 321 if ((vp->v_iflag & (VI_TEXT|VI_EXECMAP)) == (VI_TEXT|VI_EXECMAP)) { 322 /* Safe unlocked, as long as caller holds a reference. */ 323 return (0); 324 } 325 326 mutex_enter(vp->v_interlock); 327 if (vp->v_writecount != 0) { 328 KASSERT((vp->v_iflag & VI_TEXT) == 0); 329 mutex_exit(vp->v_interlock); 330 return (ETXTBSY); 331 } 332 if ((vp->v_iflag & VI_EXECMAP) == 0) { 333 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 334 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 335 } 336 vp->v_iflag |= (VI_TEXT | VI_EXECMAP); 337 mutex_exit(vp->v_interlock); 338 return (0); 339} 340 341/* 342 * Vnode close call 343 * 344 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 345 */ 346int 347vn_close(struct vnode *vp, int flags, kauth_cred_t cred) 348{ 349 int error; 350 351 if (flags & FWRITE) { 352 mutex_enter(vp->v_interlock); 353 KASSERT(vp->v_writecount > 0); 354 vp->v_writecount--; 355 mutex_exit(vp->v_interlock); 356 } 357 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 358 error = VOP_CLOSE(vp, flags, cred); 359 vput(vp); 360 return (error); 361} 362 363static int 364enforce_rlimit_fsize(struct vnode *vp, struct uio *uio, int ioflag) 365{ 366 struct lwp *l = curlwp; 367 off_t testoff; 368 369 if (uio->uio_rw != UIO_WRITE || vp->v_type != VREG) 370 return 0; 371 372 KASSERT(VOP_ISLOCKED(vp) == LK_EXCLUSIVE); 373 if (ioflag & IO_APPEND) 374 testoff = vp->v_size; 375 else 376 testoff = uio->uio_offset; 377 378 if (testoff + uio->uio_resid > 379 l->l_proc->p_rlimit[RLIMIT_FSIZE].rlim_cur) { 380 mutex_enter(proc_lock); 381 psignal(l->l_proc, SIGXFSZ); 382 mutex_exit(proc_lock); 383 return EFBIG; 384 } 385 386 return 0; 387} 388 389/* 390 * Package up an I/O request on a vnode into a uio and do it. 391 */ 392int 393vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 394 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 395 struct lwp *l) 396{ 397 struct uio auio; 398 struct iovec aiov; 399 int error; 400 401 if ((ioflg & IO_NODELOCKED) == 0) { 402 if (rw == UIO_READ) { 403 vn_lock(vp, LK_SHARED | LK_RETRY); 404 } else /* UIO_WRITE */ { 405 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 406 } 407 } 408 auio.uio_iov = &aiov; 409 auio.uio_iovcnt = 1; 410 aiov.iov_base = base; 411 aiov.iov_len = len; 412 auio.uio_resid = len; 413 auio.uio_offset = offset; 414 auio.uio_rw = rw; 415 if (segflg == UIO_SYSSPACE) { 416 UIO_SETUP_SYSSPACE(&auio); 417 } else { 418 auio.uio_vmspace = l->l_proc->p_vmspace; 419 } 420 421 if ((error = enforce_rlimit_fsize(vp, &auio, ioflg)) != 0) 422 goto out; 423 424 if (rw == UIO_READ) { 425 error = VOP_READ(vp, &auio, ioflg, cred); 426 } else { 427 error = VOP_WRITE(vp, &auio, ioflg, cred); 428 } 429 430 if (aresid) 431 *aresid = auio.uio_resid; 432 else 433 if (auio.uio_resid && error == 0) 434 error = EIO; 435 436 out: 437 if ((ioflg & IO_NODELOCKED) == 0) { 438 VOP_UNLOCK(vp); 439 } 440 return (error); 441} 442 443int 444vn_readdir(file_t *fp, char *bf, int segflg, u_int count, int *done, 445 struct lwp *l, off_t **cookies, int *ncookies) 446{ 447 struct vnode *vp = (struct vnode *)fp->f_data; 448 struct iovec aiov; 449 struct uio auio; 450 int error, eofflag; 451 452 /* Limit the size on any kernel buffers used by VOP_READDIR */ 453 count = min(MAXBSIZE, count); 454 455unionread: 456 if (vp->v_type != VDIR) 457 return (EINVAL); 458 aiov.iov_base = bf; 459 aiov.iov_len = count; 460 auio.uio_iov = &aiov; 461 auio.uio_iovcnt = 1; 462 auio.uio_rw = UIO_READ; 463 if (segflg == UIO_SYSSPACE) { 464 UIO_SETUP_SYSSPACE(&auio); 465 } else { 466 KASSERT(l == curlwp); 467 auio.uio_vmspace = l->l_proc->p_vmspace; 468 } 469 auio.uio_resid = count; 470 vn_lock(vp, LK_SHARED | LK_RETRY); 471 auio.uio_offset = fp->f_offset; 472 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 473 ncookies); 474 mutex_enter(&fp->f_lock); 475 fp->f_offset = auio.uio_offset; 476 mutex_exit(&fp->f_lock); 477 VOP_UNLOCK(vp); 478 if (error) 479 return (error); 480 481 if (count == auio.uio_resid && vn_union_readdir_hook) { 482 struct vnode *ovp = vp; 483 484 error = (*vn_union_readdir_hook)(&vp, fp, l); 485 if (error) 486 return (error); 487 if (vp != ovp) 488 goto unionread; 489 } 490 491 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 492 (vp->v_mount->mnt_flag & MNT_UNION)) { 493 struct vnode *tvp = vp; 494 vp = vp->v_mount->mnt_vnodecovered; 495 vref(vp); 496 mutex_enter(&fp->f_lock); 497 fp->f_data = vp; 498 fp->f_offset = 0; 499 mutex_exit(&fp->f_lock); 500 vrele(tvp); 501 goto unionread; 502 } 503 *done = count - auio.uio_resid; 504 return error; 505} 506 507/* 508 * File table vnode read routine. 509 */ 510static int 511vn_read(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 512 int flags) 513{ 514 struct vnode *vp = (struct vnode *)fp->f_data; 515 int error, ioflag, fflag; 516 size_t count; 517 518 ioflag = IO_ADV_ENCODE(fp->f_advice); 519 fflag = fp->f_flag; 520 if (fflag & FNONBLOCK) 521 ioflag |= IO_NDELAY; 522 if ((fflag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 523 ioflag |= IO_SYNC; 524 if (fflag & FALTIO) 525 ioflag |= IO_ALTSEMANTICS; 526 if (fflag & FDIRECT) 527 ioflag |= IO_DIRECT; 528 vn_lock(vp, LK_SHARED | LK_RETRY); 529 uio->uio_offset = *offset; 530 count = uio->uio_resid; 531 error = VOP_READ(vp, uio, ioflag, cred); 532 if (flags & FOF_UPDATE_OFFSET) 533 *offset += count - uio->uio_resid; 534 VOP_UNLOCK(vp); 535 return (error); 536} 537 538/* 539 * File table vnode write routine. 540 */ 541static int 542vn_write(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 543 int flags) 544{ 545 struct vnode *vp = (struct vnode *)fp->f_data; 546 int error, ioflag, fflag; 547 size_t count; 548 549 ioflag = IO_ADV_ENCODE(fp->f_advice) | IO_UNIT; 550 fflag = fp->f_flag; 551 if (vp->v_type == VREG && (fflag & O_APPEND)) 552 ioflag |= IO_APPEND; 553 if (fflag & FNONBLOCK) 554 ioflag |= IO_NDELAY; 555 if (fflag & FFSYNC || 556 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 557 ioflag |= IO_SYNC; 558 else if (fflag & FDSYNC) 559 ioflag |= IO_DSYNC; 560 if (fflag & FALTIO) 561 ioflag |= IO_ALTSEMANTICS; 562 if (fflag & FDIRECT) 563 ioflag |= IO_DIRECT; 564 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 565 uio->uio_offset = *offset; 566 count = uio->uio_resid; 567 568 if ((error = enforce_rlimit_fsize(vp, uio, ioflag)) != 0) 569 goto out; 570 571 error = VOP_WRITE(vp, uio, ioflag, cred); 572 573 if (flags & FOF_UPDATE_OFFSET) { 574 if (ioflag & IO_APPEND) { 575 /* 576 * SUSv3 describes behaviour for count = 0 as following: 577 * "Before any action ... is taken, and if nbyte is zero 578 * and the file is a regular file, the write() function 579 * ... in the absence of errors ... shall return zero 580 * and have no other results." 581 */ 582 if (count) 583 *offset = uio->uio_offset; 584 } else 585 *offset += count - uio->uio_resid; 586 } 587 588 out: 589 VOP_UNLOCK(vp); 590 return (error); 591} 592 593/* 594 * File table vnode stat routine. 595 */ 596static int 597vn_statfile(file_t *fp, struct stat *sb) 598{ 599 struct vnode *vp = fp->f_data; 600 int error; 601 602 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 603 error = vn_stat(vp, sb); 604 VOP_UNLOCK(vp); 605 return error; 606} 607 608int 609vn_stat(struct vnode *vp, struct stat *sb) 610{ 611 struct vattr va; 612 int error; 613 mode_t mode; 614 615 memset(&va, 0, sizeof(va)); 616 error = VOP_GETATTR(vp, &va, kauth_cred_get()); 617 if (error) 618 return (error); 619 /* 620 * Copy from vattr table 621 */ 622 memset(sb, 0, sizeof(*sb)); 623 sb->st_dev = va.va_fsid; 624 sb->st_ino = va.va_fileid; 625 mode = va.va_mode; 626 switch (vp->v_type) { 627 case VREG: 628 mode |= S_IFREG; 629 break; 630 case VDIR: 631 mode |= S_IFDIR; 632 break; 633 case VBLK: 634 mode |= S_IFBLK; 635 break; 636 case VCHR: 637 mode |= S_IFCHR; 638 break; 639 case VLNK: 640 mode |= S_IFLNK; 641 break; 642 case VSOCK: 643 mode |= S_IFSOCK; 644 break; 645 case VFIFO: 646 mode |= S_IFIFO; 647 break; 648 default: 649 return (EBADF); 650 }; 651 sb->st_mode = mode; 652 sb->st_nlink = va.va_nlink; 653 sb->st_uid = va.va_uid; 654 sb->st_gid = va.va_gid; 655 sb->st_rdev = va.va_rdev; 656 sb->st_size = va.va_size; 657 sb->st_atimespec = va.va_atime; 658 sb->st_mtimespec = va.va_mtime; 659 sb->st_ctimespec = va.va_ctime; 660 sb->st_birthtimespec = va.va_birthtime; 661 sb->st_blksize = va.va_blocksize; 662 sb->st_flags = va.va_flags; 663 sb->st_gen = 0; 664 sb->st_blocks = va.va_bytes / S_BLKSIZE; 665 return (0); 666} 667 668/* 669 * File table vnode fcntl routine. 670 */ 671static int 672vn_fcntl(file_t *fp, u_int com, void *data) 673{ 674 struct vnode *vp = fp->f_data; 675 int error; 676 677 error = VOP_FCNTL(vp, com, data, fp->f_flag, kauth_cred_get()); 678 return (error); 679} 680 681/* 682 * File table vnode ioctl routine. 683 */ 684static int 685vn_ioctl(file_t *fp, u_long com, void *data) 686{ 687 struct vnode *vp = fp->f_data, *ovp; 688 struct vattr vattr; 689 int error; 690 691 switch (vp->v_type) { 692 693 case VREG: 694 case VDIR: 695 if (com == FIONREAD) { 696 vn_lock(vp, LK_SHARED | LK_RETRY); 697 error = VOP_GETATTR(vp, &vattr, kauth_cred_get()); 698 VOP_UNLOCK(vp); 699 if (error) 700 return (error); 701 *(int *)data = vattr.va_size - fp->f_offset; 702 return (0); 703 } 704 if ((com == FIONWRITE) || (com == FIONSPACE)) { 705 /* 706 * Files don't have send queues, so there never 707 * are any bytes in them, nor is there any 708 * open space in them. 709 */ 710 *(int *)data = 0; 711 return (0); 712 } 713 if (com == FIOGETBMAP) { 714 daddr_t *block; 715 716 if (*(daddr_t *)data < 0) 717 return (EINVAL); 718 block = (daddr_t *)data; 719 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 720 } 721 if (com == OFIOGETBMAP) { 722 daddr_t ibn, obn; 723 724 if (*(int32_t *)data < 0) 725 return (EINVAL); 726 ibn = (daddr_t)*(int32_t *)data; 727 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 728 *(int32_t *)data = (int32_t)obn; 729 return error; 730 } 731 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 732 return (0); /* XXX */ 733 /* fall into ... */ 734 case VFIFO: 735 case VCHR: 736 case VBLK: 737 error = VOP_IOCTL(vp, com, data, fp->f_flag, 738 kauth_cred_get()); 739 if (error == 0 && com == TIOCSCTTY) { 740 vref(vp); 741 mutex_enter(proc_lock); 742 ovp = curproc->p_session->s_ttyvp; 743 curproc->p_session->s_ttyvp = vp; 744 mutex_exit(proc_lock); 745 if (ovp != NULL) 746 vrele(ovp); 747 } 748 return (error); 749 750 default: 751 return (EPASSTHROUGH); 752 } 753} 754 755/* 756 * File table vnode poll routine. 757 */ 758static int 759vn_poll(file_t *fp, int events) 760{ 761 762 return (VOP_POLL(fp->f_data, events)); 763} 764 765/* 766 * File table vnode kqfilter routine. 767 */ 768int 769vn_kqfilter(file_t *fp, struct knote *kn) 770{ 771 772 return (VOP_KQFILTER(fp->f_data, kn)); 773} 774 775/* 776 * Check that the vnode is still valid, and if so 777 * acquire requested lock. 778 */ 779int 780vn_lock(struct vnode *vp, int flags) 781{ 782 int error; 783 784#if 0 785 KASSERT(vp->v_usecount > 0 || (vp->v_iflag & VI_ONWORKLST) != 0); 786#endif 787 KASSERT((flags & ~(LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT|LK_RETRY)) == 0); 788 KASSERT(!mutex_owned(vp->v_interlock)); 789 790#ifdef DIAGNOSTIC 791 if (wapbl_vphaswapbl(vp)) 792 WAPBL_JUNLOCK_ASSERT(wapbl_vptomp(vp)); 793#endif 794 795 do { 796 /* 797 * XXX PR 37706 forced unmount of file systems is unsafe. 798 * Race between vclean() and this the remaining problem. 799 */ 800 mutex_enter(vp->v_interlock); 801 if (vp->v_iflag & VI_XLOCK) { 802 if (flags & LK_NOWAIT) { 803 mutex_exit(vp->v_interlock); 804 return EBUSY; 805 } 806 vwait(vp, VI_XLOCK); 807 mutex_exit(vp->v_interlock); 808 error = ENOENT; 809 } else { 810 mutex_exit(vp->v_interlock); 811 error = VOP_LOCK(vp, (flags & ~LK_RETRY)); 812 if (error == 0 || error == EDEADLK || error == EBUSY) 813 return (error); 814 } 815 } while (flags & LK_RETRY); 816 return (error); 817} 818 819/* 820 * File table vnode close routine. 821 */ 822static int 823vn_closefile(file_t *fp) 824{ 825 826 return vn_close(fp->f_data, fp->f_flag, fp->f_cred); 827} 828 829/* 830 * Simplified in-kernel wrapper calls for extended attribute access. 831 * Both calls pass in a NULL credential, authorizing a "kernel" access. 832 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 833 */ 834int 835vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 836 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 837{ 838 struct uio auio; 839 struct iovec aiov; 840 int error; 841 842 aiov.iov_len = *buflen; 843 aiov.iov_base = bf; 844 845 auio.uio_iov = &aiov; 846 auio.uio_iovcnt = 1; 847 auio.uio_rw = UIO_READ; 848 auio.uio_offset = 0; 849 auio.uio_resid = *buflen; 850 UIO_SETUP_SYSSPACE(&auio); 851 852 if ((ioflg & IO_NODELOCKED) == 0) 853 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 854 855 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 856 857 if ((ioflg & IO_NODELOCKED) == 0) 858 VOP_UNLOCK(vp); 859 860 if (error == 0) 861 *buflen = *buflen - auio.uio_resid; 862 863 return (error); 864} 865 866/* 867 * XXX Failure mode if partially written? 868 */ 869int 870vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 871 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 872{ 873 struct uio auio; 874 struct iovec aiov; 875 int error; 876 877 aiov.iov_len = buflen; 878 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 879 880 auio.uio_iov = &aiov; 881 auio.uio_iovcnt = 1; 882 auio.uio_rw = UIO_WRITE; 883 auio.uio_offset = 0; 884 auio.uio_resid = buflen; 885 UIO_SETUP_SYSSPACE(&auio); 886 887 if ((ioflg & IO_NODELOCKED) == 0) { 888 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 889 } 890 891 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 892 893 if ((ioflg & IO_NODELOCKED) == 0) { 894 VOP_UNLOCK(vp); 895 } 896 897 return (error); 898} 899 900int 901vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 902 const char *attrname, struct lwp *l) 903{ 904 int error; 905 906 if ((ioflg & IO_NODELOCKED) == 0) { 907 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 908 } 909 910 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 911 if (error == EOPNOTSUPP) 912 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 913 914 if ((ioflg & IO_NODELOCKED) == 0) { 915 VOP_UNLOCK(vp); 916 } 917 918 return (error); 919} 920 921void 922vn_ra_allocctx(struct vnode *vp) 923{ 924 struct uvm_ractx *ra = NULL; 925 926 KASSERT(mutex_owned(vp->v_interlock)); 927 928 if (vp->v_type != VREG) { 929 return; 930 } 931 if (vp->v_ractx != NULL) { 932 return; 933 } 934 if (vp->v_ractx == NULL) { 935 mutex_exit(vp->v_interlock); 936 ra = uvm_ra_allocctx(); 937 mutex_enter(vp->v_interlock); 938 if (ra != NULL && vp->v_ractx == NULL) { 939 vp->v_ractx = ra; 940 ra = NULL; 941 } 942 } 943 if (ra != NULL) { 944 uvm_ra_freectx(ra); 945 } 946} 947 948int 949vn_fifo_bypass(void *v) 950{ 951 struct vop_generic_args *ap = v; 952 953 return VOCALL(fifo_vnodeop_p, ap->a_desc->vdesc_offset, v); 954} 955