vfs_vnops.c revision 1.156.2.5
1 /* $NetBSD: vfs_vnops.c,v 1.156.2.5 2010/10/09 03:32:33 yamt 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 1.156.2.5 2010/10/09 03:32:33 yamt 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 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 VERIEXEC_PATH_GET(ndp->ni_dirp, ndp->ni_segflg, ndp->ni_dirp, path); 162 163 error = namei(ndp); 164 if (error) 165 goto out; 166 167 vp = ndp->ni_vp; 168 169 #if NVERIEXEC > 0 170 error = veriexec_openchk(l, ndp->ni_vp, ndp->ni_dirp, fmode); 171 if (error) 172 goto bad; 173 #endif /* NVERIEXEC > 0 */ 174 175 if (fmode & O_CREAT) { 176 if (ndp->ni_vp == NULL) { 177 vattr_null(&va); 178 va.va_type = VREG; 179 va.va_mode = cmode; 180 if (fmode & O_EXCL) 181 va.va_vaflags |= VA_EXCLUSIVE; 182 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 183 &ndp->ni_cnd, &va); 184 if (error) 185 goto out; 186 fmode &= ~O_TRUNC; 187 vp = ndp->ni_vp; 188 } else { 189 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 190 if (ndp->ni_dvp == ndp->ni_vp) 191 vrele(ndp->ni_dvp); 192 else 193 vput(ndp->ni_dvp); 194 ndp->ni_dvp = NULL; 195 vp = ndp->ni_vp; 196 if (fmode & O_EXCL) { 197 error = EEXIST; 198 goto bad; 199 } 200 fmode &= ~O_CREAT; 201 } 202 } else { 203 vp = ndp->ni_vp; 204 } 205 if (vp->v_type == VSOCK) { 206 error = EOPNOTSUPP; 207 goto bad; 208 } 209 if (ndp->ni_vp->v_type == VLNK) { 210 error = EFTYPE; 211 goto bad; 212 } 213 214 if ((fmode & O_CREAT) == 0) { 215 error = vn_openchk(vp, cred, fmode); 216 if (error != 0) 217 goto bad; 218 } 219 220 if (fmode & O_TRUNC) { 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 & O_DIRECTORY) != 0 && vp->v_type != VDIR) 267 return ENOTDIR; 268 269 if ((fflags & FREAD) != 0) { 270 permbits = VREAD; 271 } 272 if ((fflags & (FWRITE | O_TRUNC)) != 0) { 273 permbits |= VWRITE; 274 if (vp->v_type == VDIR) { 275 error = EISDIR; 276 goto bad; 277 } 278 error = vn_writechk(vp); 279 if (error != 0) 280 goto bad; 281 } 282 error = VOP_ACCESS(vp, permbits, cred); 283 bad: 284 return error; 285 } 286 287 /* 288 * Mark a vnode as having executable mappings. 289 */ 290 void 291 vn_markexec(struct vnode *vp) 292 { 293 294 if ((vp->v_iflag & VI_EXECMAP) != 0) { 295 /* Safe unlocked, as long as caller holds a reference. */ 296 return; 297 } 298 299 mutex_enter(&vp->v_interlock); 300 if ((vp->v_iflag & VI_EXECMAP) == 0) { 301 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 302 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 303 vp->v_iflag |= VI_EXECMAP; 304 } 305 mutex_exit(&vp->v_interlock); 306 } 307 308 /* 309 * Mark a vnode as being the text of a process. 310 * Fail if the vnode is currently writable. 311 */ 312 int 313 vn_marktext(struct vnode *vp) 314 { 315 316 if ((vp->v_iflag & (VI_TEXT|VI_EXECMAP)) == (VI_TEXT|VI_EXECMAP)) { 317 /* Safe unlocked, as long as caller holds a reference. */ 318 return (0); 319 } 320 321 mutex_enter(&vp->v_interlock); 322 if (vp->v_writecount != 0) { 323 KASSERT((vp->v_iflag & VI_TEXT) == 0); 324 mutex_exit(&vp->v_interlock); 325 return (ETXTBSY); 326 } 327 if ((vp->v_iflag & VI_EXECMAP) == 0) { 328 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 329 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 330 } 331 vp->v_iflag |= (VI_TEXT | VI_EXECMAP); 332 mutex_exit(&vp->v_interlock); 333 return (0); 334 } 335 336 /* 337 * Vnode close call 338 * 339 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 340 */ 341 int 342 vn_close(struct vnode *vp, int flags, kauth_cred_t cred) 343 { 344 int error; 345 346 if (flags & FWRITE) { 347 mutex_enter(&vp->v_interlock); 348 vp->v_writecount--; 349 mutex_exit(&vp->v_interlock); 350 } 351 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 352 error = VOP_CLOSE(vp, flags, cred); 353 vput(vp); 354 return (error); 355 } 356 357 static int 358 enforce_rlimit_fsize(struct vnode *vp, struct uio *uio, int ioflag) 359 { 360 struct lwp *l = curlwp; 361 off_t testoff; 362 363 if (uio->uio_rw != UIO_WRITE || vp->v_type != VREG) 364 return 0; 365 366 KASSERT(VOP_ISLOCKED(vp) == LK_EXCLUSIVE); 367 if (ioflag & IO_APPEND) 368 testoff = vp->v_size; 369 else 370 testoff = uio->uio_offset; 371 372 if (testoff + uio->uio_resid > 373 l->l_proc->p_rlimit[RLIMIT_FSIZE].rlim_cur) { 374 mutex_enter(proc_lock); 375 psignal(l->l_proc, SIGXFSZ); 376 mutex_exit(proc_lock); 377 return EFBIG; 378 } 379 380 return 0; 381 } 382 383 /* 384 * Package up an I/O request on a vnode into a uio and do it. 385 */ 386 int 387 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 388 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 389 struct lwp *l) 390 { 391 struct uio auio; 392 struct iovec aiov; 393 int error; 394 395 if ((ioflg & IO_NODELOCKED) == 0) { 396 if (rw == UIO_READ) { 397 vn_lock(vp, LK_SHARED | LK_RETRY); 398 } else /* UIO_WRITE */ { 399 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 400 } 401 } 402 auio.uio_iov = &aiov; 403 auio.uio_iovcnt = 1; 404 aiov.iov_base = base; 405 aiov.iov_len = len; 406 auio.uio_resid = len; 407 auio.uio_offset = offset; 408 auio.uio_rw = rw; 409 if (segflg == UIO_SYSSPACE) { 410 UIO_SETUP_SYSSPACE(&auio); 411 } else { 412 auio.uio_vmspace = l->l_proc->p_vmspace; 413 } 414 415 if ((error = enforce_rlimit_fsize(vp, &auio, ioflg)) != 0) 416 goto out; 417 418 if (rw == UIO_READ) { 419 error = VOP_READ(vp, &auio, ioflg, cred); 420 } else { 421 error = VOP_WRITE(vp, &auio, ioflg, cred); 422 } 423 424 if (aresid) 425 *aresid = auio.uio_resid; 426 else 427 if (auio.uio_resid && error == 0) 428 error = EIO; 429 430 out: 431 if ((ioflg & IO_NODELOCKED) == 0) { 432 VOP_UNLOCK(vp); 433 } 434 return (error); 435 } 436 437 int 438 vn_readdir(file_t *fp, char *bf, int segflg, u_int count, int *done, 439 struct lwp *l, off_t **cookies, int *ncookies) 440 { 441 struct vnode *vp = (struct vnode *)fp->f_data; 442 struct iovec aiov; 443 struct uio auio; 444 int error, eofflag; 445 446 /* Limit the size on any kernel buffers used by VOP_READDIR */ 447 count = min(MAXBSIZE, count); 448 449 unionread: 450 if (vp->v_type != VDIR) 451 return (EINVAL); 452 aiov.iov_base = bf; 453 aiov.iov_len = count; 454 auio.uio_iov = &aiov; 455 auio.uio_iovcnt = 1; 456 auio.uio_rw = UIO_READ; 457 if (segflg == UIO_SYSSPACE) { 458 UIO_SETUP_SYSSPACE(&auio); 459 } else { 460 KASSERT(l == curlwp); 461 auio.uio_vmspace = l->l_proc->p_vmspace; 462 } 463 auio.uio_resid = count; 464 vn_lock(vp, LK_SHARED | LK_RETRY); 465 auio.uio_offset = fp->f_offset; 466 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 467 ncookies); 468 mutex_enter(&fp->f_lock); 469 fp->f_offset = auio.uio_offset; 470 mutex_exit(&fp->f_lock); 471 VOP_UNLOCK(vp); 472 if (error) 473 return (error); 474 475 if (count == auio.uio_resid && vn_union_readdir_hook) { 476 struct vnode *ovp = vp; 477 478 error = (*vn_union_readdir_hook)(&vp, fp, l); 479 if (error) 480 return (error); 481 if (vp != ovp) 482 goto unionread; 483 } 484 485 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 486 (vp->v_mount->mnt_flag & MNT_UNION)) { 487 struct vnode *tvp = vp; 488 vp = vp->v_mount->mnt_vnodecovered; 489 vref(vp); 490 mutex_enter(&fp->f_lock); 491 fp->f_data = vp; 492 fp->f_offset = 0; 493 mutex_exit(&fp->f_lock); 494 vrele(tvp); 495 goto unionread; 496 } 497 *done = count - auio.uio_resid; 498 return error; 499 } 500 501 /* 502 * File table vnode read routine. 503 */ 504 static int 505 vn_read(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 506 int flags) 507 { 508 struct vnode *vp = (struct vnode *)fp->f_data; 509 int count, error, ioflag, fflag; 510 511 ioflag = IO_ADV_ENCODE(fp->f_advice); 512 fflag = fp->f_flag; 513 if (fflag & FNONBLOCK) 514 ioflag |= IO_NDELAY; 515 if ((fflag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 516 ioflag |= IO_SYNC; 517 if (fflag & FALTIO) 518 ioflag |= IO_ALTSEMANTICS; 519 if (fflag & FDIRECT) 520 ioflag |= IO_DIRECT; 521 vn_lock(vp, LK_SHARED | LK_RETRY); 522 uio->uio_offset = *offset; 523 count = uio->uio_resid; 524 error = VOP_READ(vp, uio, ioflag, cred); 525 if (flags & FOF_UPDATE_OFFSET) 526 *offset += count - uio->uio_resid; 527 VOP_UNLOCK(vp); 528 return (error); 529 } 530 531 /* 532 * File table vnode write routine. 533 */ 534 static int 535 vn_write(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 536 int flags) 537 { 538 struct vnode *vp = (struct vnode *)fp->f_data; 539 int count, error, ioflag, fflag; 540 541 ioflag = IO_ADV_ENCODE(fp->f_advice) | IO_UNIT; 542 fflag = fp->f_flag; 543 if (vp->v_type == VREG && (fflag & O_APPEND)) 544 ioflag |= IO_APPEND; 545 if (fflag & FNONBLOCK) 546 ioflag |= IO_NDELAY; 547 if (fflag & FFSYNC || 548 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 549 ioflag |= IO_SYNC; 550 else if (fflag & FDSYNC) 551 ioflag |= IO_DSYNC; 552 if (fflag & FALTIO) 553 ioflag |= IO_ALTSEMANTICS; 554 if (fflag & FDIRECT) 555 ioflag |= IO_DIRECT; 556 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 557 uio->uio_offset = *offset; 558 count = uio->uio_resid; 559 560 if ((error = enforce_rlimit_fsize(vp, uio, ioflag)) != 0) 561 goto out; 562 563 error = VOP_WRITE(vp, uio, ioflag, cred); 564 565 if (flags & FOF_UPDATE_OFFSET) { 566 if (ioflag & IO_APPEND) { 567 /* 568 * SUSv3 describes behaviour for count = 0 as following: 569 * "Before any action ... is taken, and if nbyte is zero 570 * and the file is a regular file, the write() function 571 * ... in the absence of errors ... shall return zero 572 * and have no other results." 573 */ 574 if (count) 575 *offset = uio->uio_offset; 576 } else 577 *offset += count - uio->uio_resid; 578 } 579 580 out: 581 VOP_UNLOCK(vp); 582 return (error); 583 } 584 585 /* 586 * File table vnode stat routine. 587 */ 588 static int 589 vn_statfile(file_t *fp, struct stat *sb) 590 { 591 struct vnode *vp = fp->f_data; 592 int error; 593 594 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 595 error = vn_stat(vp, sb); 596 VOP_UNLOCK(vp); 597 return error; 598 } 599 600 int 601 vn_stat(struct vnode *vp, struct stat *sb) 602 { 603 struct vattr va; 604 int error; 605 mode_t mode; 606 607 memset(&va, 0, sizeof(va)); 608 error = VOP_GETATTR(vp, &va, kauth_cred_get()); 609 if (error) 610 return (error); 611 /* 612 * Copy from vattr table 613 */ 614 sb->st_dev = va.va_fsid; 615 sb->st_ino = va.va_fileid; 616 mode = va.va_mode; 617 switch (vp->v_type) { 618 case VREG: 619 mode |= S_IFREG; 620 break; 621 case VDIR: 622 mode |= S_IFDIR; 623 break; 624 case VBLK: 625 mode |= S_IFBLK; 626 break; 627 case VCHR: 628 mode |= S_IFCHR; 629 break; 630 case VLNK: 631 mode |= S_IFLNK; 632 break; 633 case VSOCK: 634 mode |= S_IFSOCK; 635 break; 636 case VFIFO: 637 mode |= S_IFIFO; 638 break; 639 default: 640 return (EBADF); 641 }; 642 sb->st_mode = mode; 643 sb->st_nlink = va.va_nlink; 644 sb->st_uid = va.va_uid; 645 sb->st_gid = va.va_gid; 646 sb->st_rdev = va.va_rdev; 647 sb->st_size = va.va_size; 648 sb->st_atimespec = va.va_atime; 649 sb->st_mtimespec = va.va_mtime; 650 sb->st_ctimespec = va.va_ctime; 651 sb->st_birthtimespec = va.va_birthtime; 652 sb->st_blksize = va.va_blocksize; 653 sb->st_flags = va.va_flags; 654 sb->st_gen = 0; 655 sb->st_blocks = va.va_bytes / S_BLKSIZE; 656 memset(sb->st_spare, 0, sizeof(sb->st_spare)); 657 return (0); 658 } 659 660 /* 661 * File table vnode fcntl routine. 662 */ 663 static int 664 vn_fcntl(file_t *fp, u_int com, void *data) 665 { 666 struct vnode *vp = fp->f_data; 667 int error; 668 669 error = VOP_FCNTL(vp, com, data, fp->f_flag, kauth_cred_get()); 670 return (error); 671 } 672 673 /* 674 * File table vnode ioctl routine. 675 */ 676 static int 677 vn_ioctl(file_t *fp, u_long com, void *data) 678 { 679 struct vnode *vp = fp->f_data, *ovp; 680 struct vattr vattr; 681 int error; 682 683 switch (vp->v_type) { 684 685 case VREG: 686 case VDIR: 687 if (com == FIONREAD) { 688 error = VOP_GETATTR(vp, &vattr, 689 kauth_cred_get()); 690 if (error) 691 return (error); 692 *(int *)data = vattr.va_size - fp->f_offset; 693 return (0); 694 } 695 if ((com == FIONWRITE) || (com == FIONSPACE)) { 696 /* 697 * Files don't have send queues, so there never 698 * are any bytes in them, nor is there any 699 * open space in them. 700 */ 701 *(int *)data = 0; 702 return (0); 703 } 704 if (com == FIOGETBMAP) { 705 daddr_t *block; 706 707 if (*(daddr_t *)data < 0) 708 return (EINVAL); 709 block = (daddr_t *)data; 710 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 711 } 712 if (com == OFIOGETBMAP) { 713 daddr_t ibn, obn; 714 715 if (*(int32_t *)data < 0) 716 return (EINVAL); 717 ibn = (daddr_t)*(int32_t *)data; 718 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 719 *(int32_t *)data = (int32_t)obn; 720 return error; 721 } 722 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 723 return (0); /* XXX */ 724 /* fall into ... */ 725 case VFIFO: 726 case VCHR: 727 case VBLK: 728 error = VOP_IOCTL(vp, com, data, fp->f_flag, 729 kauth_cred_get()); 730 if (error == 0 && com == TIOCSCTTY) { 731 vref(vp); 732 mutex_enter(proc_lock); 733 ovp = curproc->p_session->s_ttyvp; 734 curproc->p_session->s_ttyvp = vp; 735 mutex_exit(proc_lock); 736 if (ovp != NULL) 737 vrele(ovp); 738 } 739 return (error); 740 741 default: 742 return (EPASSTHROUGH); 743 } 744 } 745 746 /* 747 * File table vnode poll routine. 748 */ 749 static int 750 vn_poll(file_t *fp, int events) 751 { 752 753 return (VOP_POLL(fp->f_data, events)); 754 } 755 756 /* 757 * File table vnode kqfilter routine. 758 */ 759 int 760 vn_kqfilter(file_t *fp, struct knote *kn) 761 { 762 763 return (VOP_KQFILTER(fp->f_data, kn)); 764 } 765 766 /* 767 * Check that the vnode is still valid, and if so 768 * acquire requested lock. 769 */ 770 int 771 vn_lock(struct vnode *vp, int flags) 772 { 773 int error; 774 775 #if 0 776 KASSERT(vp->v_usecount > 0 || (vp->v_iflag & VI_ONWORKLST) != 0); 777 #endif 778 KASSERT((flags & ~(LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT|LK_RETRY)) == 0); 779 KASSERT(!mutex_owned(&vp->v_interlock)); 780 781 #ifdef DIAGNOSTIC 782 if (wapbl_vphaswapbl(vp)) 783 WAPBL_JUNLOCK_ASSERT(wapbl_vptomp(vp)); 784 #endif 785 786 do { 787 /* 788 * XXX PR 37706 forced unmount of file systems is unsafe. 789 * Race between vclean() and this the remaining problem. 790 */ 791 mutex_enter(&vp->v_interlock); 792 if (vp->v_iflag & VI_XLOCK) { 793 if (flags & LK_NOWAIT) { 794 mutex_exit(&vp->v_interlock); 795 return EBUSY; 796 } 797 vwait(vp, VI_XLOCK); 798 mutex_exit(&vp->v_interlock); 799 error = ENOENT; 800 } else { 801 mutex_exit(&vp->v_interlock); 802 error = VOP_LOCK(vp, (flags & ~LK_RETRY)); 803 if (error == 0 || error == EDEADLK || error == EBUSY) 804 return (error); 805 } 806 } while (flags & LK_RETRY); 807 return (error); 808 } 809 810 /* 811 * File table vnode close routine. 812 */ 813 static int 814 vn_closefile(file_t *fp) 815 { 816 817 return vn_close(fp->f_data, fp->f_flag, fp->f_cred); 818 } 819 820 /* 821 * Simplified in-kernel wrapper calls for extended attribute access. 822 * Both calls pass in a NULL credential, authorizing a "kernel" access. 823 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 824 */ 825 int 826 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 827 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 828 { 829 struct uio auio; 830 struct iovec aiov; 831 int error; 832 833 aiov.iov_len = *buflen; 834 aiov.iov_base = bf; 835 836 auio.uio_iov = &aiov; 837 auio.uio_iovcnt = 1; 838 auio.uio_rw = UIO_READ; 839 auio.uio_offset = 0; 840 auio.uio_resid = *buflen; 841 UIO_SETUP_SYSSPACE(&auio); 842 843 if ((ioflg & IO_NODELOCKED) == 0) 844 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 845 846 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 847 848 if ((ioflg & IO_NODELOCKED) == 0) 849 VOP_UNLOCK(vp); 850 851 if (error == 0) 852 *buflen = *buflen - auio.uio_resid; 853 854 return (error); 855 } 856 857 /* 858 * XXX Failure mode if partially written? 859 */ 860 int 861 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 862 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 863 { 864 struct uio auio; 865 struct iovec aiov; 866 int error; 867 868 aiov.iov_len = buflen; 869 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 870 871 auio.uio_iov = &aiov; 872 auio.uio_iovcnt = 1; 873 auio.uio_rw = UIO_WRITE; 874 auio.uio_offset = 0; 875 auio.uio_resid = buflen; 876 UIO_SETUP_SYSSPACE(&auio); 877 878 if ((ioflg & IO_NODELOCKED) == 0) { 879 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 880 } 881 882 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 883 884 if ((ioflg & IO_NODELOCKED) == 0) { 885 VOP_UNLOCK(vp); 886 } 887 888 return (error); 889 } 890 891 int 892 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 893 const char *attrname, struct lwp *l) 894 { 895 int error; 896 897 if ((ioflg & IO_NODELOCKED) == 0) { 898 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 899 } 900 901 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 902 if (error == EOPNOTSUPP) 903 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 904 905 if ((ioflg & IO_NODELOCKED) == 0) { 906 VOP_UNLOCK(vp); 907 } 908 909 return (error); 910 } 911 912 void 913 vn_ra_allocctx(struct vnode *vp) 914 { 915 struct uvm_ractx *ra = NULL; 916 917 KASSERT(mutex_owned(&vp->v_interlock)); 918 919 if (vp->v_type != VREG) { 920 return; 921 } 922 if (vp->v_ractx != NULL) { 923 return; 924 } 925 if (vp->v_ractx == NULL) { 926 mutex_exit(&vp->v_interlock); 927 ra = uvm_ra_allocctx(); 928 mutex_enter(&vp->v_interlock); 929 if (ra != NULL && vp->v_ractx == NULL) { 930 vp->v_ractx = ra; 931 ra = NULL; 932 } 933 } 934 if (ra != NULL) { 935 uvm_ra_freectx(ra); 936 } 937 } 938 939 int 940 vn_fifo_bypass(void *v) 941 { 942 struct vop_generic_args *ap = v; 943 944 return VOCALL(fifo_vnodeop_p, ap->a_desc->vdesc_offset, v); 945 } 946
Indexes created Sat Sep 20 22:09:52 GMT 2025