1 /* $NetBSD: spec_vnops.c,v 1.34 1996/10/13 02:21:41 christos Exp $ */ 2 3 /* 4 * Copyright (c) 1989, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * @(#)spec_vnops.c 8.8 (Berkeley) 11/21/94 36 */ 37 38 #include <sys/param.h> 39 #include <sys/proc.h> 40 #include <sys/systm.h> 41 #include <sys/kernel.h> 42 #include <sys/conf.h> 43 #include <sys/buf.h> 44 #include <sys/mount.h> 45 #include <sys/namei.h> 46 #include <sys/vnode.h> 47 #include <sys/stat.h> 48 #include <sys/errno.h> 49 #include <sys/ioctl.h> 50 #include <sys/file.h> 51 #include <sys/disklabel.h> 52 53 #include <miscfs/genfs/genfs.h> 54 #include <miscfs/specfs/specdev.h> 55 56 /* symbolic sleep message strings for devices */ 57 char devopn[] = "devopn"; 58 char devio[] = "devio"; 59 char devwait[] = "devwait"; 60 char devin[] = "devin"; 61 char devout[] = "devout"; 62 char devioc[] = "devioc"; 63 char devcls[] = "devcls"; 64 65 int (**spec_vnodeop_p) __P((void *)); 66 struct vnodeopv_entry_desc spec_vnodeop_entries[] = { 67 { &vop_default_desc, vn_default_error }, 68 { &vop_lookup_desc, spec_lookup }, /* lookup */ 69 { &vop_create_desc, spec_create }, /* create */ 70 { &vop_mknod_desc, spec_mknod }, /* mknod */ 71 { &vop_open_desc, spec_open }, /* open */ 72 { &vop_close_desc, spec_close }, /* close */ 73 { &vop_access_desc, spec_access }, /* access */ 74 { &vop_getattr_desc, spec_getattr }, /* getattr */ 75 { &vop_setattr_desc, spec_setattr }, /* setattr */ 76 { &vop_read_desc, spec_read }, /* read */ 77 { &vop_write_desc, spec_write }, /* write */ 78 { &vop_lease_desc, spec_lease_check }, /* lease */ 79 { &vop_ioctl_desc, spec_ioctl }, /* ioctl */ 80 { &vop_poll_desc, spec_poll }, /* poll */ 81 { &vop_mmap_desc, spec_mmap }, /* mmap */ 82 { &vop_fsync_desc, spec_fsync }, /* fsync */ 83 { &vop_seek_desc, spec_seek }, /* seek */ 84 { &vop_remove_desc, spec_remove }, /* remove */ 85 { &vop_link_desc, spec_link }, /* link */ 86 { &vop_rename_desc, spec_rename }, /* rename */ 87 { &vop_mkdir_desc, spec_mkdir }, /* mkdir */ 88 { &vop_rmdir_desc, spec_rmdir }, /* rmdir */ 89 { &vop_symlink_desc, spec_symlink }, /* symlink */ 90 { &vop_readdir_desc, spec_readdir }, /* readdir */ 91 { &vop_readlink_desc, spec_readlink }, /* readlink */ 92 { &vop_abortop_desc, spec_abortop }, /* abortop */ 93 { &vop_inactive_desc, spec_inactive }, /* inactive */ 94 { &vop_reclaim_desc, spec_reclaim }, /* reclaim */ 95 { &vop_lock_desc, spec_lock }, /* lock */ 96 { &vop_unlock_desc, spec_unlock }, /* unlock */ 97 { &vop_bmap_desc, spec_bmap }, /* bmap */ 98 { &vop_strategy_desc, spec_strategy }, /* strategy */ 99 { &vop_print_desc, spec_print }, /* print */ 100 { &vop_islocked_desc, spec_islocked }, /* islocked */ 101 { &vop_pathconf_desc, spec_pathconf }, /* pathconf */ 102 { &vop_advlock_desc, spec_advlock }, /* advlock */ 103 { &vop_blkatoff_desc, spec_blkatoff }, /* blkatoff */ 104 { &vop_valloc_desc, spec_valloc }, /* valloc */ 105 { &vop_vfree_desc, spec_vfree }, /* vfree */ 106 { &vop_truncate_desc, spec_truncate }, /* truncate */ 107 { &vop_update_desc, spec_update }, /* update */ 108 { &vop_bwrite_desc, spec_bwrite }, /* bwrite */ 109 { (struct vnodeop_desc*)NULL, (int(*) __P((void *)))NULL } 110 }; 111 struct vnodeopv_desc spec_vnodeop_opv_desc = 112 { &spec_vnodeop_p, spec_vnodeop_entries }; 113 114 /* 115 * Trivial lookup routine that always fails. 116 */ 117 int 118 spec_lookup(v) 119 void *v; 120 { 121 struct vop_lookup_args /* { 122 struct vnode *a_dvp; 123 struct vnode **a_vpp; 124 struct componentname *a_cnp; 125 } */ *ap = v; 126 127 *ap->a_vpp = NULL; 128 return (ENOTDIR); 129 } 130 131 /* 132 * Open a special file. 133 */ 134 /* ARGSUSED */ 135 int 136 spec_open(v) 137 void *v; 138 { 139 struct vop_open_args /* { 140 struct vnode *a_vp; 141 int a_mode; 142 struct ucred *a_cred; 143 struct proc *a_p; 144 } */ *ap = v; 145 struct vnode *bvp, *vp = ap->a_vp; 146 dev_t bdev, dev = (dev_t)vp->v_rdev; 147 register int maj = major(dev); 148 int error; 149 150 /* 151 * Don't allow open if fs is mounted -nodev. 152 */ 153 if (vp->v_mount && (vp->v_mount->mnt_flag & MNT_NODEV)) 154 return (ENXIO); 155 156 switch (vp->v_type) { 157 158 case VCHR: 159 if ((u_int)maj >= nchrdev) 160 return (ENXIO); 161 if (ap->a_cred != FSCRED && (ap->a_mode & FWRITE)) { 162 /* 163 * When running in very secure mode, do not allow 164 * opens for writing of any disk character devices. 165 */ 166 if (securelevel >= 2 && cdevsw[maj].d_type == D_DISK) 167 return (EPERM); 168 /* 169 * When running in secure mode, do not allow opens 170 * for writing of /dev/mem, /dev/kmem, or character 171 * devices whose corresponding block devices are 172 * currently mounted. 173 */ 174 if (securelevel >= 1) { 175 if ((bdev = chrtoblk(dev)) != NODEV && 176 vfinddev(bdev, VBLK, &bvp) && 177 bvp->v_usecount > 0 && 178 (error = vfs_mountedon(bvp))) 179 return (error); 180 if (iskmemdev(dev)) 181 return (EPERM); 182 } 183 } 184 if (cdevsw[maj].d_type == D_TTY) 185 vp->v_flag |= VISTTY; 186 VOP_UNLOCK(vp); 187 error = (*cdevsw[maj].d_open)(dev, ap->a_mode, S_IFCHR, ap->a_p); 188 VOP_LOCK(vp); 189 return (error); 190 191 case VBLK: 192 if ((u_int)maj >= nblkdev) 193 return (ENXIO); 194 /* 195 * When running in very secure mode, do not allow 196 * opens for writing of any disk block devices. 197 */ 198 if (securelevel >= 2 && ap->a_cred != FSCRED && 199 (ap->a_mode & FWRITE) && bdevsw[maj].d_type == D_DISK) 200 return (EPERM); 201 /* 202 * Do not allow opens of block devices that are 203 * currently mounted. 204 */ 205 if ((error = vfs_mountedon(vp)) != 0) 206 return (error); 207 return ((*bdevsw[maj].d_open)(dev, ap->a_mode, S_IFBLK, ap->a_p)); 208 case VNON: 209 case VLNK: 210 case VDIR: 211 case VREG: 212 case VBAD: 213 case VFIFO: 214 case VSOCK: 215 break; 216 } 217 return (0); 218 } 219 220 /* 221 * Vnode op for read 222 */ 223 /* ARGSUSED */ 224 int 225 spec_read(v) 226 void *v; 227 { 228 struct vop_read_args /* { 229 struct vnode *a_vp; 230 struct uio *a_uio; 231 int a_ioflag; 232 struct ucred *a_cred; 233 } */ *ap = v; 234 register struct vnode *vp = ap->a_vp; 235 register struct uio *uio = ap->a_uio; 236 struct proc *p = uio->uio_procp; 237 struct buf *bp; 238 daddr_t bn, nextbn; 239 long bsize, bscale, ssize; 240 struct partinfo dpart; 241 int n, on, majordev; 242 int (*ioctl) __P((dev_t, u_long, caddr_t, int, struct proc *)); 243 int error = 0; 244 245 #ifdef DIAGNOSTIC 246 if (uio->uio_rw != UIO_READ) 247 panic("spec_read mode"); 248 if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc) 249 panic("spec_read proc"); 250 #endif 251 if (uio->uio_resid == 0) 252 return (0); 253 254 switch (vp->v_type) { 255 256 case VCHR: 257 VOP_UNLOCK(vp); 258 error = (*cdevsw[major(vp->v_rdev)].d_read) 259 (vp->v_rdev, uio, ap->a_ioflag); 260 VOP_LOCK(vp); 261 return (error); 262 263 case VBLK: 264 if (uio->uio_resid == 0) 265 return (0); 266 if (uio->uio_offset < 0) 267 return (EINVAL); 268 bsize = BLKDEV_IOSIZE; 269 ssize = DEV_BSIZE; 270 if ((majordev = major(vp->v_rdev)) < nblkdev && 271 (ioctl = bdevsw[majordev].d_ioctl) != NULL && 272 (*ioctl)(vp->v_rdev, DIOCGPART, (caddr_t)&dpart, FREAD, p) == 0) { 273 if (dpart.part->p_fstype == FS_BSDFFS && 274 dpart.part->p_frag != 0 && dpart.part->p_fsize != 0) 275 bsize = dpart.part->p_frag * 276 dpart.part->p_fsize; 277 if (dpart.disklab->d_secsize != 0) 278 ssize = dpart.disklab->d_secsize; 279 } 280 bscale = bsize / ssize; 281 do { 282 bn = (uio->uio_offset / ssize) &~ (bscale - 1); 283 on = uio->uio_offset % bsize; 284 n = min((unsigned)(bsize - on), uio->uio_resid); 285 if (vp->v_lastr + bscale == bn) { 286 nextbn = bn + bscale; 287 error = breadn(vp, bn, (int)bsize, &nextbn, 288 (int *)&bsize, 1, NOCRED, &bp); 289 } else 290 error = bread(vp, bn, (int)bsize, NOCRED, &bp); 291 vp->v_lastr = bn; 292 n = min(n, bsize - bp->b_resid); 293 if (error) { 294 brelse(bp); 295 return (error); 296 } 297 error = uiomove((char *)bp->b_data + on, n, uio); 298 brelse(bp); 299 } while (error == 0 && uio->uio_resid > 0 && n != 0); 300 return (error); 301 302 default: 303 panic("spec_read type"); 304 } 305 /* NOTREACHED */ 306 } 307 308 /* 309 * Vnode op for write 310 */ 311 /* ARGSUSED */ 312 int 313 spec_write(v) 314 void *v; 315 { 316 struct vop_write_args /* { 317 struct vnode *a_vp; 318 struct uio *a_uio; 319 int a_ioflag; 320 struct ucred *a_cred; 321 } */ *ap = v; 322 register struct vnode *vp = ap->a_vp; 323 register struct uio *uio = ap->a_uio; 324 struct proc *p = uio->uio_procp; 325 struct buf *bp; 326 daddr_t bn; 327 long bsize, bscale, ssize; 328 struct partinfo dpart; 329 int n, on, majordev; 330 int (*ioctl) __P((dev_t, u_long, caddr_t, int, struct proc *)); 331 int error = 0; 332 333 #ifdef DIAGNOSTIC 334 if (uio->uio_rw != UIO_WRITE) 335 panic("spec_write mode"); 336 if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc) 337 panic("spec_write proc"); 338 #endif 339 340 switch (vp->v_type) { 341 342 case VCHR: 343 VOP_UNLOCK(vp); 344 error = (*cdevsw[major(vp->v_rdev)].d_write) 345 (vp->v_rdev, uio, ap->a_ioflag); 346 VOP_LOCK(vp); 347 return (error); 348 349 case VBLK: 350 if (uio->uio_resid == 0) 351 return (0); 352 if (uio->uio_offset < 0) 353 return (EINVAL); 354 bsize = BLKDEV_IOSIZE; 355 ssize = DEV_BSIZE; 356 if ((majordev = major(vp->v_rdev)) < nblkdev && 357 (ioctl = bdevsw[majordev].d_ioctl) != NULL && 358 (*ioctl)(vp->v_rdev, DIOCGPART, (caddr_t)&dpart, FREAD, p) == 0) { 359 if (dpart.part->p_fstype == FS_BSDFFS && 360 dpart.part->p_frag != 0 && dpart.part->p_fsize != 0) 361 bsize = dpart.part->p_frag * 362 dpart.part->p_fsize; 363 if (dpart.disklab->d_secsize != 0) 364 ssize = dpart.disklab->d_secsize; 365 } 366 bscale = bsize / ssize; 367 do { 368 bn = (uio->uio_offset / ssize) &~ (bscale - 1); 369 on = uio->uio_offset % bsize; 370 n = min((unsigned)(bsize - on), uio->uio_resid); 371 if (n == bsize) 372 bp = getblk(vp, bn, bsize, 0, 0); 373 else 374 error = bread(vp, bn, bsize, NOCRED, &bp); 375 n = min(n, bsize - bp->b_resid); 376 if (error) { 377 brelse(bp); 378 return (error); 379 } 380 error = uiomove((char *)bp->b_data + on, n, uio); 381 if (n + on == bsize) 382 bawrite(bp); 383 else 384 bdwrite(bp); 385 } while (error == 0 && uio->uio_resid > 0 && n != 0); 386 return (error); 387 388 default: 389 panic("spec_write type"); 390 } 391 /* NOTREACHED */ 392 } 393 394 /* 395 * Device ioctl operation. 396 */ 397 /* ARGSUSED */ 398 int 399 spec_ioctl(v) 400 void *v; 401 { 402 struct vop_ioctl_args /* { 403 struct vnode *a_vp; 404 u_long a_command; 405 caddr_t a_data; 406 int a_fflag; 407 struct ucred *a_cred; 408 struct proc *a_p; 409 } */ *ap = v; 410 dev_t dev = ap->a_vp->v_rdev; 411 int maj = major(dev); 412 413 switch (ap->a_vp->v_type) { 414 415 case VCHR: 416 return ((*cdevsw[maj].d_ioctl)(dev, ap->a_command, ap->a_data, 417 ap->a_fflag, ap->a_p)); 418 419 case VBLK: 420 if (ap->a_command == 0 && (long)ap->a_data == B_TAPE) 421 if (bdevsw[maj].d_type == D_TAPE) 422 return (0); 423 else 424 return (1); 425 return ((*bdevsw[maj].d_ioctl)(dev, ap->a_command, ap->a_data, 426 ap->a_fflag, ap->a_p)); 427 428 default: 429 panic("spec_ioctl"); 430 /* NOTREACHED */ 431 } 432 } 433 434 /* ARGSUSED */ 435 int 436 spec_poll(v) 437 void *v; 438 { 439 struct vop_poll_args /* { 440 struct vnode *a_vp; 441 int a_events; 442 struct proc *a_p; 443 } */ *ap = v; 444 register dev_t dev; 445 446 switch (ap->a_vp->v_type) { 447 448 case VCHR: 449 dev = ap->a_vp->v_rdev; 450 return (*cdevsw[major(dev)].d_poll)(dev, ap->a_events, ap->a_p); 451 452 default: 453 return (genfs_poll(v)); 454 } 455 } 456 /* 457 * Synch buffers associated with a block device 458 */ 459 /* ARGSUSED */ 460 int 461 spec_fsync(v) 462 void *v; 463 { 464 struct vop_fsync_args /* { 465 struct vnode *a_vp; 466 struct ucred *a_cred; 467 int a_waitfor; 468 struct proc *a_p; 469 } */ *ap = v; 470 register struct vnode *vp = ap->a_vp; 471 472 if (vp->v_type == VBLK) 473 vflushbuf(vp, ap->a_waitfor == MNT_WAIT); 474 return (0); 475 } 476 477 /* 478 * Just call the device strategy routine 479 */ 480 int 481 spec_strategy(v) 482 void *v; 483 { 484 struct vop_strategy_args /* { 485 struct buf *a_bp; 486 } */ *ap = v; 487 488 (*bdevsw[major(ap->a_bp->b_dev)].d_strategy)(ap->a_bp); 489 return (0); 490 } 491 492 /* 493 * This is a noop, simply returning what one has been given. 494 */ 495 int 496 spec_bmap(v) 497 void *v; 498 { 499 struct vop_bmap_args /* { 500 struct vnode *a_vp; 501 daddr_t a_bn; 502 struct vnode **a_vpp; 503 daddr_t *a_bnp; 504 } */ *ap = v; 505 506 if (ap->a_vpp != NULL) 507 *ap->a_vpp = ap->a_vp; 508 if (ap->a_bnp != NULL) 509 *ap->a_bnp = ap->a_bn; 510 return (0); 511 } 512 513 /* 514 * At the moment we do not do any locking. 515 */ 516 /* ARGSUSED */ 517 int 518 spec_lock(v) 519 void *v; 520 { 521 522 return (0); 523 } 524 525 /* ARGSUSED */ 526 int 527 spec_unlock(v) 528 void *v; 529 { 530 531 return (0); 532 } 533 534 /* 535 * Device close routine 536 */ 537 /* ARGSUSED */ 538 int 539 spec_close(v) 540 void *v; 541 { 542 struct vop_close_args /* { 543 struct vnode *a_vp; 544 int a_fflag; 545 struct ucred *a_cred; 546 struct proc *a_p; 547 } */ *ap = v; 548 register struct vnode *vp = ap->a_vp; 549 dev_t dev = vp->v_rdev; 550 int (*devclose) __P((dev_t, int, int, struct proc *)); 551 int mode, error; 552 553 switch (vp->v_type) { 554 555 case VCHR: 556 /* 557 * Hack: a tty device that is a controlling terminal 558 * has a reference from the session structure. 559 * We cannot easily tell that a character device is 560 * a controlling terminal, unless it is the closing 561 * process' controlling terminal. In that case, 562 * if the reference count is 2 (this last descriptor 563 * plus the session), release the reference from the session. 564 */ 565 if (vcount(vp) == 2 && ap->a_p && 566 vp == ap->a_p->p_session->s_ttyvp) { 567 vrele(vp); 568 ap->a_p->p_session->s_ttyvp = NULL; 569 } 570 /* 571 * If the vnode is locked, then we are in the midst 572 * of forcably closing the device, otherwise we only 573 * close on last reference. 574 */ 575 if (vcount(vp) > 1 && (vp->v_flag & VXLOCK) == 0) 576 return (0); 577 devclose = cdevsw[major(dev)].d_close; 578 mode = S_IFCHR; 579 break; 580 581 case VBLK: 582 /* 583 * On last close of a block device (that isn't mounted) 584 * we must invalidate any in core blocks, so that 585 * we can, for instance, change floppy disks. 586 */ 587 error = vinvalbuf(vp, V_SAVE, ap->a_cred, ap->a_p, 0, 0); 588 if (error) 589 return (error); 590 /* 591 * We do not want to really close the device if it 592 * is still in use unless we are trying to close it 593 * forcibly. Since every use (buffer, vnode, swap, cmap) 594 * holds a reference to the vnode, and because we mark 595 * any other vnodes that alias this device, when the 596 * sum of the reference counts on all the aliased 597 * vnodes descends to one, we are on last close. 598 */ 599 if (vcount(vp) > 1 && (vp->v_flag & VXLOCK) == 0) 600 return (0); 601 devclose = bdevsw[major(dev)].d_close; 602 mode = S_IFBLK; 603 break; 604 605 default: 606 panic("spec_close: not special"); 607 } 608 609 return ((*devclose)(dev, ap->a_fflag, mode, ap->a_p)); 610 } 611 612 /* 613 * Print out the contents of a special device vnode. 614 */ 615 int 616 spec_print(v) 617 void *v; 618 { 619 struct vop_print_args /* { 620 struct vnode *a_vp; 621 } */ *ap = v; 622 623 printf("tag VT_NON, dev %d, %d\n", major(ap->a_vp->v_rdev), 624 minor(ap->a_vp->v_rdev)); 625 return 0; 626 } 627 628 /* 629 * Return POSIX pathconf information applicable to special devices. 630 */ 631 int 632 spec_pathconf(v) 633 void *v; 634 { 635 struct vop_pathconf_args /* { 636 struct vnode *a_vp; 637 int a_name; 638 register_t *a_retval; 639 } */ *ap = v; 640 641 switch (ap->a_name) { 642 case _PC_LINK_MAX: 643 *ap->a_retval = LINK_MAX; 644 return (0); 645 case _PC_MAX_CANON: 646 *ap->a_retval = MAX_CANON; 647 return (0); 648 case _PC_MAX_INPUT: 649 *ap->a_retval = MAX_INPUT; 650 return (0); 651 case _PC_PIPE_BUF: 652 *ap->a_retval = PIPE_BUF; 653 return (0); 654 case _PC_CHOWN_RESTRICTED: 655 *ap->a_retval = 1; 656 return (0); 657 case _PC_VDISABLE: 658 *ap->a_retval = _POSIX_VDISABLE; 659 return (0); 660 default: 661 return (EINVAL); 662 } 663 /* NOTREACHED */ 664 } 665