1 /* $NetBSD: vfs_vnops.c,v 1.175 2010/07/13 15:38:15 pooka 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 1.175 2010/07/13 15:38:15 pooka 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 char *path; 142 143 ndp->ni_cnd.cn_flags &= TRYEMULROOT | NOCHROOT; 144 145 if (fmode & O_CREAT) { 146 ndp->ni_cnd.cn_nameiop = CREATE; 147 ndp->ni_cnd.cn_flags |= LOCKPARENT | LOCKLEAF; 148 if ((fmode & O_EXCL) == 0 && 149 ((fmode & O_NOFOLLOW) == 0)) 150 ndp->ni_cnd.cn_flags |= FOLLOW; 151 } else { 152 ndp->ni_cnd.cn_nameiop = LOOKUP; 153 ndp->ni_cnd.cn_flags |= LOCKLEAF; 154 if ((fmode & O_NOFOLLOW) == 0) 155 ndp->ni_cnd.cn_flags |= FOLLOW; 156 } 157 158 VERIEXEC_PATH_GET(ndp->ni_dirp, ndp->ni_segflg, ndp->ni_dirp, path); 159 160 error = namei(ndp); 161 if (error) 162 goto out; 163 164 vp = ndp->ni_vp; 165 166 #if NVERIEXEC > 0 167 error = veriexec_openchk(l, ndp->ni_vp, ndp->ni_dirp, fmode); 168 if (error) 169 goto bad; 170 #endif /* NVERIEXEC > 0 */ 171 172 if (fmode & O_CREAT) { 173 if (ndp->ni_vp == NULL) { 174 vattr_null(&va); 175 va.va_type = VREG; 176 va.va_mode = cmode; 177 if (fmode & O_EXCL) 178 va.va_vaflags |= VA_EXCLUSIVE; 179 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 180 &ndp->ni_cnd, &va); 181 if (error) 182 goto out; 183 fmode &= ~O_TRUNC; 184 vp = ndp->ni_vp; 185 } else { 186 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 187 if (ndp->ni_dvp == ndp->ni_vp) 188 vrele(ndp->ni_dvp); 189 else 190 vput(ndp->ni_dvp); 191 ndp->ni_dvp = NULL; 192 vp = ndp->ni_vp; 193 if (fmode & O_EXCL) { 194 error = EEXIST; 195 goto bad; 196 } 197 fmode &= ~O_CREAT; 198 } 199 } else { 200 vp = ndp->ni_vp; 201 } 202 if (vp->v_type == VSOCK) { 203 error = EOPNOTSUPP; 204 goto bad; 205 } 206 if (ndp->ni_vp->v_type == VLNK) { 207 error = EFTYPE; 208 goto bad; 209 } 210 211 if ((fmode & O_CREAT) == 0) { 212 error = vn_openchk(vp, cred, fmode); 213 if (error != 0) 214 goto bad; 215 } 216 217 if (fmode & O_TRUNC) { 218 VOP_UNLOCK(vp); /* XXX */ 219 220 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* XXX */ 221 vattr_null(&va); 222 va.va_size = 0; 223 error = VOP_SETATTR(vp, &va, cred); 224 if (error != 0) 225 goto bad; 226 } 227 if ((error = VOP_OPEN(vp, fmode, cred)) != 0) 228 goto bad; 229 if (fmode & FWRITE) { 230 mutex_enter(&vp->v_interlock); 231 vp->v_writecount++; 232 mutex_exit(&vp->v_interlock); 233 } 234 235 bad: 236 if (error) 237 vput(vp); 238 out: 239 VERIEXEC_PATH_PUT(path); 240 return (error); 241 } 242 243 /* 244 * Check for write permissions on the specified vnode. 245 * Prototype text segments cannot be written. 246 */ 247 int 248 vn_writechk(struct vnode *vp) 249 { 250 251 /* 252 * If the vnode is in use as a process's text, 253 * we can't allow writing. 254 */ 255 if (vp->v_iflag & VI_TEXT) 256 return (ETXTBSY); 257 return (0); 258 } 259 260 int 261 vn_openchk(struct vnode *vp, kauth_cred_t cred, int fflags) 262 { 263 int permbits = 0; 264 int error; 265 266 if ((fflags & FREAD) != 0) { 267 permbits = VREAD; 268 } 269 if ((fflags & (FWRITE | O_TRUNC)) != 0) { 270 permbits |= VWRITE; 271 if (vp->v_type == VDIR) { 272 error = EISDIR; 273 goto bad; 274 } 275 error = vn_writechk(vp); 276 if (error != 0) 277 goto bad; 278 } 279 error = VOP_ACCESS(vp, permbits, cred); 280 bad: 281 return error; 282 } 283 284 /* 285 * Mark a vnode as having executable mappings. 286 */ 287 void 288 vn_markexec(struct vnode *vp) 289 { 290 291 if ((vp->v_iflag & VI_EXECMAP) != 0) { 292 /* Safe unlocked, as long as caller holds a reference. */ 293 return; 294 } 295 296 mutex_enter(&vp->v_interlock); 297 if ((vp->v_iflag & VI_EXECMAP) == 0) { 298 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 299 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 300 vp->v_iflag |= VI_EXECMAP; 301 } 302 mutex_exit(&vp->v_interlock); 303 } 304 305 /* 306 * Mark a vnode as being the text of a process. 307 * Fail if the vnode is currently writable. 308 */ 309 int 310 vn_marktext(struct vnode *vp) 311 { 312 313 if ((vp->v_iflag & (VI_TEXT|VI_EXECMAP)) == (VI_TEXT|VI_EXECMAP)) { 314 /* Safe unlocked, as long as caller holds a reference. */ 315 return (0); 316 } 317 318 mutex_enter(&vp->v_interlock); 319 if (vp->v_writecount != 0) { 320 KASSERT((vp->v_iflag & VI_TEXT) == 0); 321 mutex_exit(&vp->v_interlock); 322 return (ETXTBSY); 323 } 324 if ((vp->v_iflag & VI_EXECMAP) == 0) { 325 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 326 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 327 } 328 vp->v_iflag |= (VI_TEXT | VI_EXECMAP); 329 mutex_exit(&vp->v_interlock); 330 return (0); 331 } 332 333 /* 334 * Vnode close call 335 * 336 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 337 */ 338 int 339 vn_close(struct vnode *vp, int flags, kauth_cred_t cred) 340 { 341 int error; 342 343 if (flags & FWRITE) { 344 mutex_enter(&vp->v_interlock); 345 vp->v_writecount--; 346 mutex_exit(&vp->v_interlock); 347 } 348 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 349 error = VOP_CLOSE(vp, flags, cred); 350 vput(vp); 351 return (error); 352 } 353 354 static int 355 enforce_rlimit_fsize(struct vnode *vp, struct uio *uio, int ioflag) 356 { 357 struct lwp *l = curlwp; 358 off_t testoff; 359 360 if (uio->uio_rw != UIO_WRITE || vp->v_type != VREG) 361 return 0; 362 363 KASSERT(VOP_ISLOCKED(vp) == LK_EXCLUSIVE); 364 if (ioflag & IO_APPEND) 365 testoff = vp->v_size; 366 else 367 testoff = uio->uio_offset; 368 369 if (testoff + uio->uio_resid > 370 l->l_proc->p_rlimit[RLIMIT_FSIZE].rlim_cur) { 371 mutex_enter(proc_lock); 372 psignal(l->l_proc, SIGXFSZ); 373 mutex_exit(proc_lock); 374 return EFBIG; 375 } 376 377 return 0; 378 } 379 380 /* 381 * Package up an I/O request on a vnode into a uio and do it. 382 */ 383 int 384 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 385 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 386 struct lwp *l) 387 { 388 struct uio auio; 389 struct iovec aiov; 390 int error; 391 392 if ((ioflg & IO_NODELOCKED) == 0) { 393 if (rw == UIO_READ) { 394 vn_lock(vp, LK_SHARED | LK_RETRY); 395 } else /* UIO_WRITE */ { 396 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 397 } 398 } 399 auio.uio_iov = &aiov; 400 auio.uio_iovcnt = 1; 401 aiov.iov_base = base; 402 aiov.iov_len = len; 403 auio.uio_resid = len; 404 auio.uio_offset = offset; 405 auio.uio_rw = rw; 406 if (segflg == UIO_SYSSPACE) { 407 UIO_SETUP_SYSSPACE(&auio); 408 } else { 409 auio.uio_vmspace = l->l_proc->p_vmspace; 410 } 411 412 if ((error = enforce_rlimit_fsize(vp, &auio, ioflg)) != 0) 413 goto out; 414 415 if (rw == UIO_READ) { 416 error = VOP_READ(vp, &auio, ioflg, cred); 417 } else { 418 error = VOP_WRITE(vp, &auio, ioflg, cred); 419 } 420 421 if (aresid) 422 *aresid = auio.uio_resid; 423 else 424 if (auio.uio_resid && error == 0) 425 error = EIO; 426 427 out: 428 if ((ioflg & IO_NODELOCKED) == 0) { 429 VOP_UNLOCK(vp); 430 } 431 return (error); 432 } 433 434 int 435 vn_readdir(file_t *fp, char *bf, int segflg, u_int count, int *done, 436 struct lwp *l, off_t **cookies, int *ncookies) 437 { 438 struct vnode *vp = (struct vnode *)fp->f_data; 439 struct iovec aiov; 440 struct uio auio; 441 int error, eofflag; 442 443 /* Limit the size on any kernel buffers used by VOP_READDIR */ 444 count = min(MAXBSIZE, count); 445 446 unionread: 447 if (vp->v_type != VDIR) 448 return (EINVAL); 449 aiov.iov_base = bf; 450 aiov.iov_len = count; 451 auio.uio_iov = &aiov; 452 auio.uio_iovcnt = 1; 453 auio.uio_rw = UIO_READ; 454 if (segflg == UIO_SYSSPACE) { 455 UIO_SETUP_SYSSPACE(&auio); 456 } else { 457 KASSERT(l == curlwp); 458 auio.uio_vmspace = l->l_proc->p_vmspace; 459 } 460 auio.uio_resid = count; 461 vn_lock(vp, LK_SHARED | LK_RETRY); 462 auio.uio_offset = fp->f_offset; 463 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 464 ncookies); 465 mutex_enter(&fp->f_lock); 466 fp->f_offset = auio.uio_offset; 467 mutex_exit(&fp->f_lock); 468 VOP_UNLOCK(vp); 469 if (error) 470 return (error); 471 472 if (count == auio.uio_resid && vn_union_readdir_hook) { 473 struct vnode *ovp = vp; 474 475 error = (*vn_union_readdir_hook)(&vp, fp, l); 476 if (error) 477 return (error); 478 if (vp != ovp) 479 goto unionread; 480 } 481 482 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 483 (vp->v_mount->mnt_flag & MNT_UNION)) { 484 struct vnode *tvp = vp; 485 vp = vp->v_mount->mnt_vnodecovered; 486 vref(vp); 487 mutex_enter(&fp->f_lock); 488 fp->f_data = vp; 489 fp->f_offset = 0; 490 mutex_exit(&fp->f_lock); 491 vrele(tvp); 492 goto unionread; 493 } 494 *done = count - auio.uio_resid; 495 return error; 496 } 497 498 /* 499 * File table vnode read routine. 500 */ 501 static int 502 vn_read(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 503 int flags) 504 { 505 struct vnode *vp = (struct vnode *)fp->f_data; 506 int count, error, ioflag, fflag; 507 508 ioflag = IO_ADV_ENCODE(fp->f_advice); 509 fflag = fp->f_flag; 510 if (fflag & FNONBLOCK) 511 ioflag |= IO_NDELAY; 512 if ((fflag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 513 ioflag |= IO_SYNC; 514 if (fflag & FALTIO) 515 ioflag |= IO_ALTSEMANTICS; 516 if (fflag & FDIRECT) 517 ioflag |= IO_DIRECT; 518 vn_lock(vp, LK_SHARED | LK_RETRY); 519 uio->uio_offset = *offset; 520 count = uio->uio_resid; 521 error = VOP_READ(vp, uio, ioflag, cred); 522 if (flags & FOF_UPDATE_OFFSET) 523 *offset += count - uio->uio_resid; 524 VOP_UNLOCK(vp); 525 return (error); 526 } 527 528 /* 529 * File table vnode write routine. 530 */ 531 static int 532 vn_write(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 533 int flags) 534 { 535 struct vnode *vp = (struct vnode *)fp->f_data; 536 int count, error, ioflag, fflag; 537 538 ioflag = IO_ADV_ENCODE(fp->f_advice) | IO_UNIT; 539 fflag = fp->f_flag; 540 if (vp->v_type == VREG && (fflag & O_APPEND)) 541 ioflag |= IO_APPEND; 542 if (fflag & FNONBLOCK) 543 ioflag |= IO_NDELAY; 544 if (fflag & FFSYNC || 545 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 546 ioflag |= IO_SYNC; 547 else if (fflag & FDSYNC) 548 ioflag |= IO_DSYNC; 549 if (fflag & FALTIO) 550 ioflag |= IO_ALTSEMANTICS; 551 if (fflag & FDIRECT) 552 ioflag |= IO_DIRECT; 553 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 554 uio->uio_offset = *offset; 555 count = uio->uio_resid; 556 557 if ((error = enforce_rlimit_fsize(vp, uio, ioflag)) != 0) 558 goto out; 559 560 error = VOP_WRITE(vp, uio, ioflag, cred); 561 562 if (flags & FOF_UPDATE_OFFSET) { 563 if (ioflag & IO_APPEND) { 564 /* 565 * SUSv3 describes behaviour for count = 0 as following: 566 * "Before any action ... is taken, and if nbyte is zero 567 * and the file is a regular file, the write() function 568 * ... in the absence of errors ... shall return zero 569 * and have no other results." 570 */ 571 if (count) 572 *offset = uio->uio_offset; 573 } else 574 *offset += count - uio->uio_resid; 575 } 576 577 out: 578 VOP_UNLOCK(vp); 579 return (error); 580 } 581 582 /* 583 * File table vnode stat routine. 584 */ 585 static int 586 vn_statfile(file_t *fp, struct stat *sb) 587 { 588 struct vnode *vp = fp->f_data; 589 int error; 590 591 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 592 error = vn_stat(vp, sb); 593 VOP_UNLOCK(vp); 594 return error; 595 } 596 597 int 598 vn_stat(struct vnode *vp, struct stat *sb) 599 { 600 struct vattr va; 601 int error; 602 mode_t mode; 603 604 memset(&va, 0, sizeof(va)); 605 error = VOP_GETATTR(vp, &va, kauth_cred_get()); 606 if (error) 607 return (error); 608 /* 609 * Copy from vattr table 610 */ 611 sb->st_dev = va.va_fsid; 612 sb->st_ino = va.va_fileid; 613 mode = va.va_mode; 614 switch (vp->v_type) { 615 case VREG: 616 mode |= S_IFREG; 617 break; 618 case VDIR: 619 mode |= S_IFDIR; 620 break; 621 case VBLK: 622 mode |= S_IFBLK; 623 break; 624 case VCHR: 625 mode |= S_IFCHR; 626 break; 627 case VLNK: 628 mode |= S_IFLNK; 629 break; 630 case VSOCK: 631 mode |= S_IFSOCK; 632 break; 633 case VFIFO: 634 mode |= S_IFIFO; 635 break; 636 default: 637 return (EBADF); 638 }; 639 sb->st_mode = mode; 640 sb->st_nlink = va.va_nlink; 641 sb->st_uid = va.va_uid; 642 sb->st_gid = va.va_gid; 643 sb->st_rdev = va.va_rdev; 644 sb->st_size = va.va_size; 645 sb->st_atimespec = va.va_atime; 646 sb->st_mtimespec = va.va_mtime; 647 sb->st_ctimespec = va.va_ctime; 648 sb->st_birthtimespec = va.va_birthtime; 649 sb->st_blksize = va.va_blocksize; 650 sb->st_flags = va.va_flags; 651 sb->st_gen = 0; 652 sb->st_blocks = va.va_bytes / S_BLKSIZE; 653 memset(sb->st_spare, 0, sizeof(sb->st_spare)); 654 return (0); 655 } 656 657 /* 658 * File table vnode fcntl routine. 659 */ 660 static int 661 vn_fcntl(file_t *fp, u_int com, void *data) 662 { 663 struct vnode *vp = fp->f_data; 664 int error; 665 666 error = VOP_FCNTL(vp, com, data, fp->f_flag, kauth_cred_get()); 667 return (error); 668 } 669 670 /* 671 * File table vnode ioctl routine. 672 */ 673 static int 674 vn_ioctl(file_t *fp, u_long com, void *data) 675 { 676 struct vnode *vp = fp->f_data, *ovp; 677 struct vattr vattr; 678 int error; 679 680 switch (vp->v_type) { 681 682 case VREG: 683 case VDIR: 684 if (com == FIONREAD) { 685 error = VOP_GETATTR(vp, &vattr, 686 kauth_cred_get()); 687 if (error) 688 return (error); 689 *(int *)data = vattr.va_size - fp->f_offset; 690 return (0); 691 } 692 if ((com == FIONWRITE) || (com == FIONSPACE)) { 693 /* 694 * Files don't have send queues, so there never 695 * are any bytes in them, nor is there any 696 * open space in them. 697 */ 698 *(int *)data = 0; 699 return (0); 700 } 701 if (com == FIOGETBMAP) { 702 daddr_t *block; 703 704 if (*(daddr_t *)data < 0) 705 return (EINVAL); 706 block = (daddr_t *)data; 707 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 708 } 709 if (com == OFIOGETBMAP) { 710 daddr_t ibn, obn; 711 712 if (*(int32_t *)data < 0) 713 return (EINVAL); 714 ibn = (daddr_t)*(int32_t *)data; 715 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 716 *(int32_t *)data = (int32_t)obn; 717 return error; 718 } 719 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 720 return (0); /* XXX */ 721 /* fall into ... */ 722 case VFIFO: 723 case VCHR: 724 case VBLK: 725 error = VOP_IOCTL(vp, com, data, fp->f_flag, 726 kauth_cred_get()); 727 if (error == 0 && com == TIOCSCTTY) { 728 vref(vp); 729 mutex_enter(proc_lock); 730 ovp = curproc->p_session->s_ttyvp; 731 curproc->p_session->s_ttyvp = vp; 732 mutex_exit(proc_lock); 733 if (ovp != NULL) 734 vrele(ovp); 735 } 736 return (error); 737 738 default: 739 return (EPASSTHROUGH); 740 } 741 } 742 743 /* 744 * File table vnode poll routine. 745 */ 746 static int 747 vn_poll(file_t *fp, int events) 748 { 749 750 return (VOP_POLL(fp->f_data, events)); 751 } 752 753 /* 754 * File table vnode kqfilter routine. 755 */ 756 int 757 vn_kqfilter(file_t *fp, struct knote *kn) 758 { 759 760 return (VOP_KQFILTER(fp->f_data, kn)); 761 } 762 763 /* 764 * Check that the vnode is still valid, and if so 765 * acquire requested lock. 766 */ 767 int 768 vn_lock(struct vnode *vp, int flags) 769 { 770 int error; 771 772 #if 0 773 KASSERT(vp->v_usecount > 0 || (flags & LK_INTERLOCK) != 0 774 || (vp->v_iflag & VI_ONWORKLST) != 0); 775 #endif 776 KASSERT((flags & 777 ~(LK_INTERLOCK|LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT|LK_RETRY)) 778 == 0); 779 780 #ifdef DIAGNOSTIC 781 if (wapbl_vphaswapbl(vp)) 782 WAPBL_JUNLOCK_ASSERT(wapbl_vptomp(vp)); 783 #endif 784 785 do { 786 /* 787 * XXX PR 37706 forced unmount of file systems is unsafe. 788 * Race between vclean() and this the remaining problem. 789 */ 790 if (vp->v_iflag & VI_XLOCK) { 791 if ((flags & LK_INTERLOCK) == 0) { 792 mutex_enter(&vp->v_interlock); 793 } 794 flags &= ~LK_INTERLOCK; 795 if (flags & LK_NOWAIT) { 796 mutex_exit(&vp->v_interlock); 797 return EBUSY; 798 } 799 vwait(vp, VI_XLOCK); 800 mutex_exit(&vp->v_interlock); 801 error = ENOENT; 802 } else { 803 if ((flags & LK_INTERLOCK) != 0) { 804 mutex_exit(&vp->v_interlock); 805 } 806 flags &= ~LK_INTERLOCK; 807 error = VOP_LOCK(vp, (flags & ~LK_RETRY)); 808 if (error == 0 || error == EDEADLK || error == EBUSY) 809 return (error); 810 } 811 } while (flags & LK_RETRY); 812 return (error); 813 } 814 815 /* 816 * File table vnode close routine. 817 */ 818 static int 819 vn_closefile(file_t *fp) 820 { 821 822 return vn_close(fp->f_data, fp->f_flag, fp->f_cred); 823 } 824 825 /* 826 * Simplified in-kernel wrapper calls for extended attribute access. 827 * Both calls pass in a NULL credential, authorizing a "kernel" access. 828 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 829 */ 830 int 831 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 832 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 833 { 834 struct uio auio; 835 struct iovec aiov; 836 int error; 837 838 aiov.iov_len = *buflen; 839 aiov.iov_base = bf; 840 841 auio.uio_iov = &aiov; 842 auio.uio_iovcnt = 1; 843 auio.uio_rw = UIO_READ; 844 auio.uio_offset = 0; 845 auio.uio_resid = *buflen; 846 UIO_SETUP_SYSSPACE(&auio); 847 848 if ((ioflg & IO_NODELOCKED) == 0) 849 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 850 851 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 852 853 if ((ioflg & IO_NODELOCKED) == 0) 854 VOP_UNLOCK(vp); 855 856 if (error == 0) 857 *buflen = *buflen - auio.uio_resid; 858 859 return (error); 860 } 861 862 /* 863 * XXX Failure mode if partially written? 864 */ 865 int 866 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 867 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 868 { 869 struct uio auio; 870 struct iovec aiov; 871 int error; 872 873 aiov.iov_len = buflen; 874 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 875 876 auio.uio_iov = &aiov; 877 auio.uio_iovcnt = 1; 878 auio.uio_rw = UIO_WRITE; 879 auio.uio_offset = 0; 880 auio.uio_resid = buflen; 881 UIO_SETUP_SYSSPACE(&auio); 882 883 if ((ioflg & IO_NODELOCKED) == 0) { 884 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 885 } 886 887 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 888 889 if ((ioflg & IO_NODELOCKED) == 0) { 890 VOP_UNLOCK(vp); 891 } 892 893 return (error); 894 } 895 896 int 897 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 898 const char *attrname, struct lwp *l) 899 { 900 int error; 901 902 if ((ioflg & IO_NODELOCKED) == 0) { 903 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 904 } 905 906 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 907 if (error == EOPNOTSUPP) 908 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 909 910 if ((ioflg & IO_NODELOCKED) == 0) { 911 VOP_UNLOCK(vp); 912 } 913 914 return (error); 915 } 916 917 void 918 vn_ra_allocctx(struct vnode *vp) 919 { 920 struct uvm_ractx *ra = NULL; 921 922 KASSERT(mutex_owned(&vp->v_interlock)); 923 924 if (vp->v_type != VREG) { 925 return; 926 } 927 if (vp->v_ractx != NULL) { 928 return; 929 } 930 if (vp->v_ractx == NULL) { 931 mutex_exit(&vp->v_interlock); 932 ra = uvm_ra_allocctx(); 933 mutex_enter(&vp->v_interlock); 934 if (ra != NULL && vp->v_ractx == NULL) { 935 vp->v_ractx = ra; 936 ra = NULL; 937 } 938 } 939 if (ra != NULL) { 940 uvm_ra_freectx(ra); 941 } 942 } 943 944 int 945 vn_fifo_bypass(void *v) 946 { 947 struct vop_generic_args *ap = v; 948 949 return VOCALL(fifo_vnodeop_p, ap->a_desc->vdesc_offset, v); 950 } 951