vfs_vnops.c revision 1.187
1 /* $NetBSD: vfs_vnops.c,v 1.187 2014/01/17 10:55:02 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 1.187 2014/01/17 10:55:02 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 102 103 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 vput(ndp->ni_dvp); 188 if (error) 189 goto out; 190 fmode &= ~O_TRUNC; 191 vp = ndp->ni_vp; 192 } else { 193 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 194 if (ndp->ni_dvp == ndp->ni_vp) 195 vrele(ndp->ni_dvp); 196 else 197 vput(ndp->ni_dvp); 198 ndp->ni_dvp = NULL; 199 vp = ndp->ni_vp; 200 if (fmode & O_EXCL) { 201 error = EEXIST; 202 goto bad; 203 } 204 fmode &= ~O_CREAT; 205 } 206 } else { 207 vp = ndp->ni_vp; 208 } 209 if (vp->v_type == VSOCK) { 210 error = EOPNOTSUPP; 211 goto bad; 212 } 213 if (ndp->ni_vp->v_type == VLNK) { 214 error = EFTYPE; 215 goto bad; 216 } 217 218 if ((fmode & O_CREAT) == 0) { 219 error = vn_openchk(vp, cred, fmode); 220 if (error != 0) 221 goto bad; 222 } 223 224 if (fmode & O_TRUNC) { 225 vattr_null(&va); 226 va.va_size = 0; 227 error = VOP_SETATTR(vp, &va, cred); 228 if (error != 0) 229 goto bad; 230 } 231 if ((error = VOP_OPEN(vp, fmode, cred)) != 0) 232 goto bad; 233 if (fmode & FWRITE) { 234 mutex_enter(vp->v_interlock); 235 vp->v_writecount++; 236 mutex_exit(vp->v_interlock); 237 } 238 239 bad: 240 if (error) 241 vput(vp); 242 out: 243 pathbuf_stringcopy_put(ndp->ni_pathbuf, pathstring); 244 return (error); 245 } 246 247 /* 248 * Check for write permissions on the specified vnode. 249 * Prototype text segments cannot be written. 250 */ 251 int 252 vn_writechk(struct vnode *vp) 253 { 254 255 /* 256 * If the vnode is in use as a process's text, 257 * we can't allow writing. 258 */ 259 if (vp->v_iflag & VI_TEXT) 260 return (ETXTBSY); 261 return (0); 262 } 263 264 int 265 vn_openchk(struct vnode *vp, kauth_cred_t cred, int fflags) 266 { 267 int permbits = 0; 268 int error; 269 270 if ((fflags & O_DIRECTORY) != 0 && vp->v_type != VDIR) 271 return ENOTDIR; 272 273 if ((fflags & FREAD) != 0) { 274 permbits = VREAD; 275 } 276 if ((fflags & (FWRITE | O_TRUNC)) != 0) { 277 permbits |= VWRITE; 278 if (vp->v_type == VDIR) { 279 error = EISDIR; 280 goto bad; 281 } 282 error = vn_writechk(vp); 283 if (error != 0) 284 goto bad; 285 } 286 error = VOP_ACCESS(vp, permbits, cred); 287 bad: 288 return error; 289 } 290 291 /* 292 * Mark a vnode as having executable mappings. 293 */ 294 void 295 vn_markexec(struct vnode *vp) 296 { 297 298 if ((vp->v_iflag & VI_EXECMAP) != 0) { 299 /* Safe unlocked, as long as caller holds a reference. */ 300 return; 301 } 302 303 mutex_enter(vp->v_interlock); 304 if ((vp->v_iflag & VI_EXECMAP) == 0) { 305 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 306 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 307 vp->v_iflag |= VI_EXECMAP; 308 } 309 mutex_exit(vp->v_interlock); 310 } 311 312 /* 313 * Mark a vnode as being the text of a process. 314 * Fail if the vnode is currently writable. 315 */ 316 int 317 vn_marktext(struct vnode *vp) 318 { 319 320 if ((vp->v_iflag & (VI_TEXT|VI_EXECMAP)) == (VI_TEXT|VI_EXECMAP)) { 321 /* Safe unlocked, as long as caller holds a reference. */ 322 return (0); 323 } 324 325 mutex_enter(vp->v_interlock); 326 if (vp->v_writecount != 0) { 327 KASSERT((vp->v_iflag & VI_TEXT) == 0); 328 mutex_exit(vp->v_interlock); 329 return (ETXTBSY); 330 } 331 if ((vp->v_iflag & VI_EXECMAP) == 0) { 332 atomic_add_int(&uvmexp.filepages, -vp->v_uobj.uo_npages); 333 atomic_add_int(&uvmexp.execpages, vp->v_uobj.uo_npages); 334 } 335 vp->v_iflag |= (VI_TEXT | VI_EXECMAP); 336 mutex_exit(vp->v_interlock); 337 return (0); 338 } 339 340 /* 341 * Vnode close call 342 * 343 * Note: takes an unlocked vnode, while VOP_CLOSE takes a locked node. 344 */ 345 int 346 vn_close(struct vnode *vp, int flags, kauth_cred_t cred) 347 { 348 int error; 349 350 if (flags & FWRITE) { 351 mutex_enter(vp->v_interlock); 352 KASSERT(vp->v_writecount > 0); 353 vp->v_writecount--; 354 mutex_exit(vp->v_interlock); 355 } 356 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 357 error = VOP_CLOSE(vp, flags, cred); 358 vput(vp); 359 return (error); 360 } 361 362 static int 363 enforce_rlimit_fsize(struct vnode *vp, struct uio *uio, int ioflag) 364 { 365 struct lwp *l = curlwp; 366 off_t testoff; 367 368 if (uio->uio_rw != UIO_WRITE || vp->v_type != VREG) 369 return 0; 370 371 KASSERT(VOP_ISLOCKED(vp) == LK_EXCLUSIVE); 372 if (ioflag & IO_APPEND) 373 testoff = vp->v_size; 374 else 375 testoff = uio->uio_offset; 376 377 if (testoff + uio->uio_resid > 378 l->l_proc->p_rlimit[RLIMIT_FSIZE].rlim_cur) { 379 mutex_enter(proc_lock); 380 psignal(l->l_proc, SIGXFSZ); 381 mutex_exit(proc_lock); 382 return EFBIG; 383 } 384 385 return 0; 386 } 387 388 /* 389 * Package up an I/O request on a vnode into a uio and do it. 390 */ 391 int 392 vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base, int len, off_t offset, 393 enum uio_seg segflg, int ioflg, kauth_cred_t cred, size_t *aresid, 394 struct lwp *l) 395 { 396 struct uio auio; 397 struct iovec aiov; 398 int error; 399 400 if ((ioflg & IO_NODELOCKED) == 0) { 401 if (rw == UIO_READ) { 402 vn_lock(vp, LK_SHARED | LK_RETRY); 403 } else /* UIO_WRITE */ { 404 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 405 } 406 } 407 auio.uio_iov = &aiov; 408 auio.uio_iovcnt = 1; 409 aiov.iov_base = base; 410 aiov.iov_len = len; 411 auio.uio_resid = len; 412 auio.uio_offset = offset; 413 auio.uio_rw = rw; 414 if (segflg == UIO_SYSSPACE) { 415 UIO_SETUP_SYSSPACE(&auio); 416 } else { 417 auio.uio_vmspace = l->l_proc->p_vmspace; 418 } 419 420 if ((error = enforce_rlimit_fsize(vp, &auio, ioflg)) != 0) 421 goto out; 422 423 if (rw == UIO_READ) { 424 error = VOP_READ(vp, &auio, ioflg, cred); 425 } else { 426 error = VOP_WRITE(vp, &auio, ioflg, cred); 427 } 428 429 if (aresid) 430 *aresid = auio.uio_resid; 431 else 432 if (auio.uio_resid && error == 0) 433 error = EIO; 434 435 out: 436 if ((ioflg & IO_NODELOCKED) == 0) { 437 VOP_UNLOCK(vp); 438 } 439 return (error); 440 } 441 442 int 443 vn_readdir(file_t *fp, char *bf, int segflg, u_int count, int *done, 444 struct lwp *l, off_t **cookies, int *ncookies) 445 { 446 struct vnode *vp = (struct vnode *)fp->f_data; 447 struct iovec aiov; 448 struct uio auio; 449 int error, eofflag; 450 451 /* Limit the size on any kernel buffers used by VOP_READDIR */ 452 count = min(MAXBSIZE, count); 453 454 unionread: 455 if (vp->v_type != VDIR) 456 return (EINVAL); 457 aiov.iov_base = bf; 458 aiov.iov_len = count; 459 auio.uio_iov = &aiov; 460 auio.uio_iovcnt = 1; 461 auio.uio_rw = UIO_READ; 462 if (segflg == UIO_SYSSPACE) { 463 UIO_SETUP_SYSSPACE(&auio); 464 } else { 465 KASSERT(l == curlwp); 466 auio.uio_vmspace = l->l_proc->p_vmspace; 467 } 468 auio.uio_resid = count; 469 vn_lock(vp, LK_SHARED | LK_RETRY); 470 auio.uio_offset = fp->f_offset; 471 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, cookies, 472 ncookies); 473 mutex_enter(&fp->f_lock); 474 fp->f_offset = auio.uio_offset; 475 mutex_exit(&fp->f_lock); 476 VOP_UNLOCK(vp); 477 if (error) 478 return (error); 479 480 if (count == auio.uio_resid && vn_union_readdir_hook) { 481 struct vnode *ovp = vp; 482 483 error = (*vn_union_readdir_hook)(&vp, fp, l); 484 if (error) 485 return (error); 486 if (vp != ovp) 487 goto unionread; 488 } 489 490 if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && 491 (vp->v_mount->mnt_flag & MNT_UNION)) { 492 struct vnode *tvp = vp; 493 vp = vp->v_mount->mnt_vnodecovered; 494 vref(vp); 495 mutex_enter(&fp->f_lock); 496 fp->f_data = vp; 497 fp->f_offset = 0; 498 mutex_exit(&fp->f_lock); 499 vrele(tvp); 500 goto unionread; 501 } 502 *done = count - auio.uio_resid; 503 return error; 504 } 505 506 /* 507 * File table vnode read routine. 508 */ 509 static int 510 vn_read(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 511 int flags) 512 { 513 struct vnode *vp = (struct vnode *)fp->f_data; 514 int error, ioflag, fflag; 515 size_t count; 516 517 ioflag = IO_ADV_ENCODE(fp->f_advice); 518 fflag = fp->f_flag; 519 if (fflag & FNONBLOCK) 520 ioflag |= IO_NDELAY; 521 if ((fflag & (FFSYNC | FRSYNC)) == (FFSYNC | FRSYNC)) 522 ioflag |= IO_SYNC; 523 if (fflag & FALTIO) 524 ioflag |= IO_ALTSEMANTICS; 525 if (fflag & FDIRECT) 526 ioflag |= IO_DIRECT; 527 vn_lock(vp, LK_SHARED | LK_RETRY); 528 uio->uio_offset = *offset; 529 count = uio->uio_resid; 530 error = VOP_READ(vp, uio, ioflag, cred); 531 if (flags & FOF_UPDATE_OFFSET) 532 *offset += count - uio->uio_resid; 533 VOP_UNLOCK(vp); 534 return (error); 535 } 536 537 /* 538 * File table vnode write routine. 539 */ 540 static int 541 vn_write(file_t *fp, off_t *offset, struct uio *uio, kauth_cred_t cred, 542 int flags) 543 { 544 struct vnode *vp = (struct vnode *)fp->f_data; 545 int error, ioflag, fflag; 546 size_t count; 547 548 ioflag = IO_ADV_ENCODE(fp->f_advice) | IO_UNIT; 549 fflag = fp->f_flag; 550 if (vp->v_type == VREG && (fflag & O_APPEND)) 551 ioflag |= IO_APPEND; 552 if (fflag & FNONBLOCK) 553 ioflag |= IO_NDELAY; 554 if (fflag & FFSYNC || 555 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 556 ioflag |= IO_SYNC; 557 else if (fflag & FDSYNC) 558 ioflag |= IO_DSYNC; 559 if (fflag & FALTIO) 560 ioflag |= IO_ALTSEMANTICS; 561 if (fflag & FDIRECT) 562 ioflag |= IO_DIRECT; 563 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 564 uio->uio_offset = *offset; 565 count = uio->uio_resid; 566 567 if ((error = enforce_rlimit_fsize(vp, uio, ioflag)) != 0) 568 goto out; 569 570 error = VOP_WRITE(vp, uio, ioflag, cred); 571 572 if (flags & FOF_UPDATE_OFFSET) { 573 if (ioflag & IO_APPEND) { 574 /* 575 * SUSv3 describes behaviour for count = 0 as following: 576 * "Before any action ... is taken, and if nbyte is zero 577 * and the file is a regular file, the write() function 578 * ... in the absence of errors ... shall return zero 579 * and have no other results." 580 */ 581 if (count) 582 *offset = uio->uio_offset; 583 } else 584 *offset += count - uio->uio_resid; 585 } 586 587 out: 588 VOP_UNLOCK(vp); 589 return (error); 590 } 591 592 /* 593 * File table vnode stat routine. 594 */ 595 static int 596 vn_statfile(file_t *fp, struct stat *sb) 597 { 598 struct vnode *vp = fp->f_data; 599 int error; 600 601 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 602 error = vn_stat(vp, sb); 603 VOP_UNLOCK(vp); 604 return error; 605 } 606 607 int 608 vn_stat(struct vnode *vp, struct stat *sb) 609 { 610 struct vattr va; 611 int error; 612 mode_t mode; 613 614 memset(&va, 0, sizeof(va)); 615 error = VOP_GETATTR(vp, &va, kauth_cred_get()); 616 if (error) 617 return (error); 618 /* 619 * Copy from vattr table 620 */ 621 memset(sb, 0, sizeof(*sb)); 622 sb->st_dev = va.va_fsid; 623 sb->st_ino = va.va_fileid; 624 mode = va.va_mode; 625 switch (vp->v_type) { 626 case VREG: 627 mode |= S_IFREG; 628 break; 629 case VDIR: 630 mode |= S_IFDIR; 631 break; 632 case VBLK: 633 mode |= S_IFBLK; 634 break; 635 case VCHR: 636 mode |= S_IFCHR; 637 break; 638 case VLNK: 639 mode |= S_IFLNK; 640 break; 641 case VSOCK: 642 mode |= S_IFSOCK; 643 break; 644 case VFIFO: 645 mode |= S_IFIFO; 646 break; 647 default: 648 return (EBADF); 649 }; 650 sb->st_mode = mode; 651 sb->st_nlink = va.va_nlink; 652 sb->st_uid = va.va_uid; 653 sb->st_gid = va.va_gid; 654 sb->st_rdev = va.va_rdev; 655 sb->st_size = va.va_size; 656 sb->st_atimespec = va.va_atime; 657 sb->st_mtimespec = va.va_mtime; 658 sb->st_ctimespec = va.va_ctime; 659 sb->st_birthtimespec = va.va_birthtime; 660 sb->st_blksize = va.va_blocksize; 661 sb->st_flags = va.va_flags; 662 sb->st_gen = 0; 663 sb->st_blocks = va.va_bytes / S_BLKSIZE; 664 return (0); 665 } 666 667 /* 668 * File table vnode fcntl routine. 669 */ 670 static int 671 vn_fcntl(file_t *fp, u_int com, void *data) 672 { 673 struct vnode *vp = fp->f_data; 674 int error; 675 676 error = VOP_FCNTL(vp, com, data, fp->f_flag, kauth_cred_get()); 677 return (error); 678 } 679 680 /* 681 * File table vnode ioctl routine. 682 */ 683 static int 684 vn_ioctl(file_t *fp, u_long com, void *data) 685 { 686 struct vnode *vp = fp->f_data, *ovp; 687 struct vattr vattr; 688 int error; 689 690 switch (vp->v_type) { 691 692 case VREG: 693 case VDIR: 694 if (com == FIONREAD) { 695 vn_lock(vp, LK_SHARED | LK_RETRY); 696 error = VOP_GETATTR(vp, &vattr, kauth_cred_get()); 697 VOP_UNLOCK(vp); 698 if (error) 699 return (error); 700 *(int *)data = vattr.va_size - fp->f_offset; 701 return (0); 702 } 703 if ((com == FIONWRITE) || (com == FIONSPACE)) { 704 /* 705 * Files don't have send queues, so there never 706 * are any bytes in them, nor is there any 707 * open space in them. 708 */ 709 *(int *)data = 0; 710 return (0); 711 } 712 if (com == FIOGETBMAP) { 713 daddr_t *block; 714 715 if (*(daddr_t *)data < 0) 716 return (EINVAL); 717 block = (daddr_t *)data; 718 return (VOP_BMAP(vp, *block, NULL, block, NULL)); 719 } 720 if (com == OFIOGETBMAP) { 721 daddr_t ibn, obn; 722 723 if (*(int32_t *)data < 0) 724 return (EINVAL); 725 ibn = (daddr_t)*(int32_t *)data; 726 error = VOP_BMAP(vp, ibn, NULL, &obn, NULL); 727 *(int32_t *)data = (int32_t)obn; 728 return error; 729 } 730 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 731 return (0); /* XXX */ 732 /* fall into ... */ 733 case VFIFO: 734 case VCHR: 735 case VBLK: 736 error = VOP_IOCTL(vp, com, data, fp->f_flag, 737 kauth_cred_get()); 738 if (error == 0 && com == TIOCSCTTY) { 739 vref(vp); 740 mutex_enter(proc_lock); 741 ovp = curproc->p_session->s_ttyvp; 742 curproc->p_session->s_ttyvp = vp; 743 mutex_exit(proc_lock); 744 if (ovp != NULL) 745 vrele(ovp); 746 } 747 return (error); 748 749 default: 750 return (EPASSTHROUGH); 751 } 752 } 753 754 /* 755 * File table vnode poll routine. 756 */ 757 static int 758 vn_poll(file_t *fp, int events) 759 { 760 761 return (VOP_POLL(fp->f_data, events)); 762 } 763 764 /* 765 * File table vnode kqfilter routine. 766 */ 767 int 768 vn_kqfilter(file_t *fp, struct knote *kn) 769 { 770 771 return (VOP_KQFILTER(fp->f_data, kn)); 772 } 773 774 /* 775 * Check that the vnode is still valid, and if so 776 * acquire requested lock. 777 */ 778 int 779 vn_lock(struct vnode *vp, int flags) 780 { 781 int error; 782 783 #if 0 784 KASSERT(vp->v_usecount > 0 || (vp->v_iflag & VI_ONWORKLST) != 0); 785 #endif 786 KASSERT((flags & ~(LK_SHARED|LK_EXCLUSIVE|LK_NOWAIT|LK_RETRY)) == 0); 787 KASSERT(!mutex_owned(vp->v_interlock)); 788 789 #ifdef DIAGNOSTIC 790 if (wapbl_vphaswapbl(vp)) 791 WAPBL_JUNLOCK_ASSERT(wapbl_vptomp(vp)); 792 #endif 793 794 do { 795 /* 796 * XXX PR 37706 forced unmount of file systems is unsafe. 797 * Race between vclean() and this the remaining problem. 798 */ 799 mutex_enter(vp->v_interlock); 800 if (vp->v_iflag & VI_XLOCK) { 801 if (flags & LK_NOWAIT) { 802 mutex_exit(vp->v_interlock); 803 return EBUSY; 804 } 805 vwait(vp, VI_XLOCK); 806 mutex_exit(vp->v_interlock); 807 error = ENOENT; 808 } else { 809 mutex_exit(vp->v_interlock); 810 error = VOP_LOCK(vp, (flags & ~LK_RETRY)); 811 if (error == 0 || error == EDEADLK || error == EBUSY) 812 return (error); 813 } 814 } while (flags & LK_RETRY); 815 return (error); 816 } 817 818 /* 819 * File table vnode close routine. 820 */ 821 static int 822 vn_closefile(file_t *fp) 823 { 824 825 return vn_close(fp->f_data, fp->f_flag, fp->f_cred); 826 } 827 828 /* 829 * Simplified in-kernel wrapper calls for extended attribute access. 830 * Both calls pass in a NULL credential, authorizing a "kernel" access. 831 * Set IO_NODELOCKED in ioflg if the vnode is already locked. 832 */ 833 int 834 vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace, 835 const char *attrname, size_t *buflen, void *bf, struct lwp *l) 836 { 837 struct uio auio; 838 struct iovec aiov; 839 int error; 840 841 aiov.iov_len = *buflen; 842 aiov.iov_base = bf; 843 844 auio.uio_iov = &aiov; 845 auio.uio_iovcnt = 1; 846 auio.uio_rw = UIO_READ; 847 auio.uio_offset = 0; 848 auio.uio_resid = *buflen; 849 UIO_SETUP_SYSSPACE(&auio); 850 851 if ((ioflg & IO_NODELOCKED) == 0) 852 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 853 854 error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL); 855 856 if ((ioflg & IO_NODELOCKED) == 0) 857 VOP_UNLOCK(vp); 858 859 if (error == 0) 860 *buflen = *buflen - auio.uio_resid; 861 862 return (error); 863 } 864 865 /* 866 * XXX Failure mode if partially written? 867 */ 868 int 869 vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace, 870 const char *attrname, size_t buflen, const void *bf, struct lwp *l) 871 { 872 struct uio auio; 873 struct iovec aiov; 874 int error; 875 876 aiov.iov_len = buflen; 877 aiov.iov_base = __UNCONST(bf); /* XXXUNCONST kills const */ 878 879 auio.uio_iov = &aiov; 880 auio.uio_iovcnt = 1; 881 auio.uio_rw = UIO_WRITE; 882 auio.uio_offset = 0; 883 auio.uio_resid = buflen; 884 UIO_SETUP_SYSSPACE(&auio); 885 886 if ((ioflg & IO_NODELOCKED) == 0) { 887 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 888 } 889 890 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL); 891 892 if ((ioflg & IO_NODELOCKED) == 0) { 893 VOP_UNLOCK(vp); 894 } 895 896 return (error); 897 } 898 899 int 900 vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace, 901 const char *attrname, struct lwp *l) 902 { 903 int error; 904 905 if ((ioflg & IO_NODELOCKED) == 0) { 906 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 907 } 908 909 error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL); 910 if (error == EOPNOTSUPP) 911 error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL, NULL); 912 913 if ((ioflg & IO_NODELOCKED) == 0) { 914 VOP_UNLOCK(vp); 915 } 916 917 return (error); 918 } 919 920 void 921 vn_ra_allocctx(struct vnode *vp) 922 { 923 struct uvm_ractx *ra = NULL; 924 925 KASSERT(mutex_owned(vp->v_interlock)); 926 927 if (vp->v_type != VREG) { 928 return; 929 } 930 if (vp->v_ractx != NULL) { 931 return; 932 } 933 if (vp->v_ractx == NULL) { 934 mutex_exit(vp->v_interlock); 935 ra = uvm_ra_allocctx(); 936 mutex_enter(vp->v_interlock); 937 if (ra != NULL && vp->v_ractx == NULL) { 938 vp->v_ractx = ra; 939 ra = NULL; 940 } 941 } 942 if (ra != NULL) { 943 uvm_ra_freectx(ra); 944 } 945 } 946 947 int 948 vn_fifo_bypass(void *v) 949 { 950 struct vop_generic_args *ap = v; 951 952 return VOCALL(fifo_vnodeop_p, ap->a_desc->vdesc_offset, v); 953 } 954
Indexes created Sat Sep 20 22:09:52 GMT 2025