spec_vnops.c revision 1.16
1 /* $NetBSD: spec_vnops.c,v 1.16 1994/06/29 06:35:03 cgd 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 8.6 (Berkeley) 4/9/94 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 #include <miscfs/specfs/specdev.h> 53 54 /* symbolic sleep message strings for devices */ 55 char devopn[] = "devopn"; 56 char devio[] = "devio"; 57 char devwait[] = "devwait"; 58 char devin[] = "devin"; 59 char devout[] = "devout"; 60 char devioc[] = "devioc"; 61 char devcls[] = "devcls"; 62 63 int (**spec_vnodeop_p)(); 64 struct vnodeopv_entry_desc spec_vnodeop_entries[] = { 65 { &vop_default_desc, vn_default_error }, 66 { &vop_lookup_desc, spec_lookup }, /* lookup */ 67 { &vop_create_desc, spec_create }, /* create */ 68 { &vop_mknod_desc, spec_mknod }, /* mknod */ 69 { &vop_open_desc, spec_open }, /* open */ 70 { &vop_close_desc, spec_close }, /* close */ 71 { &vop_access_desc, spec_access }, /* access */ 72 { &vop_getattr_desc, spec_getattr }, /* getattr */ 73 { &vop_setattr_desc, spec_setattr }, /* setattr */ 74 { &vop_read_desc, spec_read }, /* read */ 75 { &vop_write_desc, spec_write }, /* write */ 76 { &vop_ioctl_desc, spec_ioctl }, /* ioctl */ 77 { &vop_select_desc, spec_select }, /* select */ 78 { &vop_mmap_desc, spec_mmap }, /* mmap */ 79 { &vop_fsync_desc, spec_fsync }, /* fsync */ 80 { &vop_seek_desc, spec_seek }, /* seek */ 81 { &vop_remove_desc, spec_remove }, /* remove */ 82 { &vop_link_desc, spec_link }, /* link */ 83 { &vop_rename_desc, spec_rename }, /* rename */ 84 { &vop_mkdir_desc, spec_mkdir }, /* mkdir */ 85 { &vop_rmdir_desc, spec_rmdir }, /* rmdir */ 86 { &vop_symlink_desc, spec_symlink }, /* symlink */ 87 { &vop_readdir_desc, spec_readdir }, /* readdir */ 88 { &vop_readlink_desc, spec_readlink }, /* readlink */ 89 { &vop_abortop_desc, spec_abortop }, /* abortop */ 90 { &vop_inactive_desc, spec_inactive }, /* inactive */ 91 { &vop_reclaim_desc, spec_reclaim }, /* reclaim */ 92 { &vop_lock_desc, spec_lock }, /* lock */ 93 { &vop_unlock_desc, spec_unlock }, /* unlock */ 94 { &vop_bmap_desc, spec_bmap }, /* bmap */ 95 { &vop_strategy_desc, spec_strategy }, /* strategy */ 96 { &vop_print_desc, spec_print }, /* print */ 97 { &vop_islocked_desc, spec_islocked }, /* islocked */ 98 { &vop_pathconf_desc, spec_pathconf }, /* pathconf */ 99 { &vop_advlock_desc, spec_advlock }, /* advlock */ 100 { &vop_blkatoff_desc, spec_blkatoff }, /* blkatoff */ 101 { &vop_valloc_desc, spec_valloc }, /* valloc */ 102 { &vop_vfree_desc, spec_vfree }, /* vfree */ 103 { &vop_truncate_desc, spec_truncate }, /* truncate */ 104 { &vop_update_desc, spec_update }, /* update */ 105 { &vop_bwrite_desc, spec_bwrite }, /* bwrite */ 106 { (struct vnodeop_desc*)NULL, (int(*)())NULL } 107 }; 108 struct vnodeopv_desc spec_vnodeop_opv_desc = 109 { &spec_vnodeop_p, spec_vnodeop_entries }; 110 111 /* 112 * Trivial lookup routine that always fails. 113 */ 114 int 115 spec_lookup(ap) 116 struct vop_lookup_args /* { 117 struct vnode *a_dvp; 118 struct vnode **a_vpp; 119 struct componentname *a_cnp; 120 } */ *ap; 121 { 122 123 *ap->a_vpp = NULL; 124 return (ENOTDIR); 125 } 126 127 /* 128 * Open a special file. 129 */ 130 /* ARGSUSED */ 131 spec_open(ap) 132 struct vop_open_args /* { 133 struct vnode *a_vp; 134 int a_mode; 135 struct ucred *a_cred; 136 struct proc *a_p; 137 } */ *ap; 138 { 139 struct vnode *bvp, *vp = ap->a_vp; 140 dev_t bdev, dev = (dev_t)vp->v_rdev; 141 register int maj = major(dev); 142 int error; 143 144 /* 145 * Don't allow open if fs is mounted -nodev. 146 */ 147 if (vp->v_mount && (vp->v_mount->mnt_flag & MNT_NODEV)) 148 return (ENXIO); 149 150 switch (vp->v_type) { 151 152 case VCHR: 153 if ((u_int)maj >= nchrdev) 154 return (ENXIO); 155 if (ap->a_cred != FSCRED && (ap->a_mode & FWRITE)) { 156 /* 157 * When running in very secure mode, do not allow 158 * opens for writing of any disk character devices. 159 */ 160 if (securelevel >= 2 && isdisk(dev, VCHR)) 161 return (EPERM); 162 /* 163 * When running in secure mode, do not allow opens 164 * for writing of /dev/mem, /dev/kmem, or character 165 * devices whose corresponding block devices are 166 * currently mounted. 167 */ 168 if (securelevel >= 1) { 169 if ((bdev = chrtoblk(dev)) != NODEV && 170 vfinddev(bdev, VBLK, &bvp) && 171 bvp->v_usecount > 0 && 172 (error = vfs_mountedon(bvp))) 173 return (error); 174 if (iskmemdev(dev)) 175 return (EPERM); 176 } 177 } 178 VOP_UNLOCK(vp); 179 error = (*cdevsw[maj].d_open)(dev, ap->a_mode, S_IFCHR, ap->a_p); 180 VOP_LOCK(vp); 181 return (error); 182 183 case VBLK: 184 if ((u_int)maj >= nblkdev) 185 return (ENXIO); 186 /* 187 * When running in very secure mode, do not allow 188 * opens for writing of any disk block devices. 189 */ 190 if (securelevel >= 2 && ap->a_cred != FSCRED && 191 (ap->a_mode & FWRITE) && isdisk(dev, VBLK)) 192 return (EPERM); 193 /* 194 * Do not allow opens of block devices that are 195 * currently mounted. 196 */ 197 if (error = vfs_mountedon(vp)) 198 return (error); 199 return ((*bdevsw[maj].d_open)(dev, ap->a_mode, S_IFBLK, ap->a_p)); 200 } 201 return (0); 202 } 203 204 /* 205 * Vnode op for read 206 */ 207 /* ARGSUSED */ 208 spec_read(ap) 209 struct vop_read_args /* { 210 struct vnode *a_vp; 211 struct uio *a_uio; 212 int a_ioflag; 213 struct ucred *a_cred; 214 } */ *ap; 215 { 216 register struct vnode *vp = ap->a_vp; 217 register struct uio *uio = ap->a_uio; 218 struct proc *p = uio->uio_procp; 219 struct buf *bp; 220 daddr_t bn, nextbn; 221 long bsize, bscale; 222 struct partinfo dpart; 223 int n, on, majordev, (*ioctl)(); 224 int error = 0; 225 dev_t dev; 226 227 #ifdef DIAGNOSTIC 228 if (uio->uio_rw != UIO_READ) 229 panic("spec_read mode"); 230 if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc) 231 panic("spec_read proc"); 232 #endif 233 if (uio->uio_resid == 0) 234 return (0); 235 236 switch (vp->v_type) { 237 238 case VCHR: 239 VOP_UNLOCK(vp); 240 error = (*cdevsw[major(vp->v_rdev)].d_read) 241 (vp->v_rdev, uio, ap->a_ioflag); 242 VOP_LOCK(vp); 243 return (error); 244 245 case VBLK: 246 if (uio->uio_offset < 0) 247 return (EINVAL); 248 bsize = BLKDEV_IOSIZE; 249 dev = vp->v_rdev; 250 if ((majordev = major(dev)) < nblkdev && 251 (ioctl = bdevsw[majordev].d_ioctl) != NULL && 252 (*ioctl)(dev, DIOCGPART, (caddr_t)&dpart, FREAD, p) == 0 && 253 dpart.part->p_fstype == FS_BSDFFS && 254 dpart.part->p_frag != 0 && dpart.part->p_fsize != 0) 255 bsize = dpart.part->p_frag * dpart.part->p_fsize; 256 bscale = bsize / DEV_BSIZE; 257 do { 258 bn = (uio->uio_offset / DEV_BSIZE) &~ (bscale - 1); 259 on = uio->uio_offset % bsize; 260 n = min((unsigned)(bsize - on), uio->uio_resid); 261 if (vp->v_lastr + bscale == bn) { 262 nextbn = bn + bscale; 263 error = breadn(vp, bn, (int)bsize, &nextbn, 264 (int *)&bsize, 1, NOCRED, &bp); 265 } else 266 error = bread(vp, bn, (int)bsize, NOCRED, &bp); 267 vp->v_lastr = bn; 268 n = min(n, bsize - bp->b_resid); 269 if (error) { 270 brelse(bp); 271 return (error); 272 } 273 error = uiomove((char *)bp->b_data + on, n, uio); 274 if (n + on == bsize) 275 bp->b_flags |= B_AGE; 276 brelse(bp); 277 } while (error == 0 && uio->uio_resid > 0 && n != 0); 278 return (error); 279 280 default: 281 panic("spec_read type"); 282 } 283 /* NOTREACHED */ 284 } 285 286 /* 287 * Vnode op for write 288 */ 289 /* ARGSUSED */ 290 spec_write(ap) 291 struct vop_write_args /* { 292 struct vnode *a_vp; 293 struct uio *a_uio; 294 int a_ioflag; 295 struct ucred *a_cred; 296 } */ *ap; 297 { 298 register struct vnode *vp = ap->a_vp; 299 register struct uio *uio = ap->a_uio; 300 struct proc *p = uio->uio_procp; 301 struct buf *bp; 302 daddr_t bn; 303 int bsize, blkmask; 304 struct partinfo dpart; 305 register int n, on; 306 int error = 0; 307 308 #ifdef DIAGNOSTIC 309 if (uio->uio_rw != UIO_WRITE) 310 panic("spec_write mode"); 311 if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc) 312 panic("spec_write proc"); 313 #endif 314 315 switch (vp->v_type) { 316 317 case VCHR: 318 VOP_UNLOCK(vp); 319 error = (*cdevsw[major(vp->v_rdev)].d_write) 320 (vp->v_rdev, uio, ap->a_ioflag); 321 VOP_LOCK(vp); 322 return (error); 323 324 case VBLK: 325 if (uio->uio_resid == 0) 326 return (0); 327 if (uio->uio_offset < 0) 328 return (EINVAL); 329 bsize = BLKDEV_IOSIZE; 330 if ((*bdevsw[major(vp->v_rdev)].d_ioctl)(vp->v_rdev, DIOCGPART, 331 (caddr_t)&dpart, FREAD, p) == 0) { 332 if (dpart.part->p_fstype == FS_BSDFFS && 333 dpart.part->p_frag != 0 && dpart.part->p_fsize != 0) 334 bsize = dpart.part->p_frag * 335 dpart.part->p_fsize; 336 } 337 blkmask = (bsize / DEV_BSIZE) - 1; 338 do { 339 bn = (uio->uio_offset / DEV_BSIZE) &~ blkmask; 340 on = uio->uio_offset % bsize; 341 n = min((unsigned)(bsize - on), uio->uio_resid); 342 if (n == bsize) 343 bp = getblk(vp, bn, bsize, 0, 0); 344 else 345 error = bread(vp, bn, bsize, NOCRED, &bp); 346 n = min(n, bsize - bp->b_resid); 347 if (error) { 348 brelse(bp); 349 return (error); 350 } 351 error = uiomove((char *)bp->b_data + on, n, uio); 352 if (n + on == bsize) { 353 bp->b_flags |= B_AGE; 354 bawrite(bp); 355 } else 356 bdwrite(bp); 357 } while (error == 0 && uio->uio_resid > 0 && n != 0); 358 return (error); 359 360 default: 361 panic("spec_write type"); 362 } 363 /* NOTREACHED */ 364 } 365 366 /* 367 * Device ioctl operation. 368 */ 369 /* ARGSUSED */ 370 spec_ioctl(ap) 371 struct vop_ioctl_args /* { 372 struct vnode *a_vp; 373 int a_command; 374 caddr_t a_data; 375 int a_fflag; 376 struct ucred *a_cred; 377 struct proc *a_p; 378 } */ *ap; 379 { 380 dev_t dev = ap->a_vp->v_rdev; 381 382 switch (ap->a_vp->v_type) { 383 384 case VCHR: 385 return ((*cdevsw[major(dev)].d_ioctl)(dev, ap->a_command, ap->a_data, 386 ap->a_fflag, ap->a_p)); 387 388 case VBLK: 389 if (ap->a_command == 0 && (int)ap->a_data == B_TAPE) 390 if (bdevsw[major(dev)].d_flags & B_TAPE) 391 return (0); 392 else 393 return (1); 394 return ((*bdevsw[major(dev)].d_ioctl)(dev, ap->a_command, ap->a_data, 395 ap->a_fflag, ap->a_p)); 396 397 default: 398 panic("spec_ioctl"); 399 /* NOTREACHED */ 400 } 401 } 402 403 /* ARGSUSED */ 404 spec_select(ap) 405 struct vop_select_args /* { 406 struct vnode *a_vp; 407 int a_which; 408 int a_fflags; 409 struct ucred *a_cred; 410 struct proc *a_p; 411 } */ *ap; 412 { 413 register dev_t dev; 414 415 switch (ap->a_vp->v_type) { 416 417 default: 418 return (1); /* XXX */ 419 420 case VCHR: 421 dev = ap->a_vp->v_rdev; 422 return (*cdevsw[major(dev)].d_select)(dev, ap->a_which, ap->a_p); 423 } 424 } 425 /* 426 * Synch buffers associated with a block device 427 */ 428 /* ARGSUSED */ 429 int 430 spec_fsync(ap) 431 struct vop_fsync_args /* { 432 struct vnode *a_vp; 433 struct ucred *a_cred; 434 int a_waitfor; 435 struct proc *a_p; 436 } */ *ap; 437 { 438 register struct vnode *vp = ap->a_vp; 439 register struct buf *bp; 440 struct buf *nbp; 441 int s; 442 443 if (vp->v_type == VCHR) 444 return (0); 445 /* 446 * Flush all dirty buffers associated with a block device. 447 */ 448 loop: 449 s = splbio(); 450 for (bp = vp->v_dirtyblkhd.lh_first; bp; bp = nbp) { 451 nbp = bp->b_vnbufs.le_next; 452 if ((bp->b_flags & B_BUSY)) 453 continue; 454 if ((bp->b_flags & B_DELWRI) == 0) 455 panic("spec_fsync: not dirty"); 456 bremfree(bp); 457 bp->b_flags |= B_BUSY; 458 splx(s); 459 bawrite(bp); 460 goto loop; 461 } 462 if (ap->a_waitfor == MNT_WAIT) { 463 while (vp->v_numoutput) { 464 vp->v_flag |= VBWAIT; 465 sleep((caddr_t)&vp->v_numoutput, PRIBIO + 1); 466 } 467 #ifdef DIAGNOSTIC 468 if (vp->v_dirtyblkhd.lh_first) { 469 vprint("spec_fsync: dirty", vp); 470 goto loop; 471 } 472 #endif 473 } 474 splx(s); 475 return (0); 476 } 477 478 /* 479 * Just call the device strategy routine 480 */ 481 spec_strategy(ap) 482 struct vop_strategy_args /* { 483 struct buf *a_bp; 484 } */ *ap; 485 { 486 487 (*bdevsw[major(ap->a_bp->b_dev)].d_strategy)(ap->a_bp); 488 return (0); 489 } 490 491 /* 492 * This is a noop, simply returning what one has been given. 493 */ 494 spec_bmap(ap) 495 struct vop_bmap_args /* { 496 struct vnode *a_vp; 497 daddr_t a_bn; 498 struct vnode **a_vpp; 499 daddr_t *a_bnp; 500 } */ *ap; 501 { 502 503 if (ap->a_vpp != NULL) 504 *ap->a_vpp = ap->a_vp; 505 if (ap->a_bnp != NULL) 506 *ap->a_bnp = ap->a_bn; 507 return (0); 508 } 509 510 /* 511 * At the moment we do not do any locking. 512 */ 513 /* ARGSUSED */ 514 spec_lock(ap) 515 struct vop_lock_args /* { 516 struct vnode *a_vp; 517 } */ *ap; 518 { 519 520 return (0); 521 } 522 523 /* ARGSUSED */ 524 spec_unlock(ap) 525 struct vop_unlock_args /* { 526 struct vnode *a_vp; 527 } */ *ap; 528 { 529 530 return (0); 531 } 532 533 /* 534 * Device close routine 535 */ 536 /* ARGSUSED */ 537 spec_close(ap) 538 struct vop_close_args /* { 539 struct vnode *a_vp; 540 int a_fflag; 541 struct ucred *a_cred; 542 struct proc *a_p; 543 } */ *ap; 544 { 545 register struct vnode *vp = ap->a_vp; 546 dev_t dev = vp->v_rdev; 547 int (*devclose) __P((dev_t, int, int, struct proc *)); 548 int mode, error; 549 550 switch (vp->v_type) { 551 552 case VCHR: 553 /* 554 * Hack: a tty device that is a controlling terminal 555 * has a reference from the session structure. 556 * We cannot easily tell that a character device is 557 * a controlling terminal, unless it is the closing 558 * process' controlling terminal. In that case, 559 * if the reference count is 2 (this last descriptor 560 * plus the session), release the reference from the session. 561 */ 562 if (vcount(vp) == 2 && ap->a_p && 563 vp == ap->a_p->p_session->s_ttyvp) { 564 vrele(vp); 565 ap->a_p->p_session->s_ttyvp = NULL; 566 } 567 /* 568 * If the vnode is locked, then we are in the midst 569 * of forcably closing the device, otherwise we only 570 * close on last reference. 571 */ 572 if (vcount(vp) > 1 && (vp->v_flag & VXLOCK) == 0) 573 return (0); 574 devclose = cdevsw[major(dev)].d_close; 575 mode = S_IFCHR; 576 break; 577 578 case VBLK: 579 /* 580 * On last close of a block device (that isn't mounted) 581 * we must invalidate any in core blocks, so that 582 * we can, for instance, change floppy disks. 583 */ 584 if (error = vinvalbuf(vp, V_SAVE, ap->a_cred, ap->a_p, 0, 0)) 585 return (error); 586 /* 587 * We do not want to really close the device if it 588 * is still in use unless we are trying to close it 589 * forcibly. Since every use (buffer, vnode, swap, cmap) 590 * holds a reference to the vnode, and because we mark 591 * any other vnodes that alias this device, when the 592 * sum of the reference counts on all the aliased 593 * vnodes descends to one, we are on last close. 594 */ 595 if (vcount(vp) > 1 && (vp->v_flag & VXLOCK) == 0) 596 return (0); 597 devclose = bdevsw[major(dev)].d_close; 598 mode = S_IFBLK; 599 break; 600 601 default: 602 panic("spec_close: not special"); 603 } 604 605 return ((*devclose)(dev, ap->a_fflag, mode, ap->a_p)); 606 } 607 608 /* 609 * Print out the contents of a special device vnode. 610 */ 611 spec_print(ap) 612 struct vop_print_args /* { 613 struct vnode *a_vp; 614 } */ *ap; 615 { 616 617 printf("tag VT_NON, dev %d, %d\n", major(ap->a_vp->v_rdev), 618 minor(ap->a_vp->v_rdev)); 619 } 620 621 /* 622 * Return POSIX pathconf information applicable to special devices. 623 */ 624 spec_pathconf(ap) 625 struct vop_pathconf_args /* { 626 struct vnode *a_vp; 627 int a_name; 628 int *a_retval; 629 } */ *ap; 630 { 631 632 switch (ap->a_name) { 633 case _PC_LINK_MAX: 634 *ap->a_retval = LINK_MAX; 635 return (0); 636 case _PC_MAX_CANON: 637 *ap->a_retval = MAX_CANON; 638 return (0); 639 case _PC_MAX_INPUT: 640 *ap->a_retval = MAX_INPUT; 641 return (0); 642 case _PC_PIPE_BUF: 643 *ap->a_retval = PIPE_BUF; 644 return (0); 645 case _PC_CHOWN_RESTRICTED: 646 *ap->a_retval = 1; 647 return (0); 648 case _PC_VDISABLE: 649 *ap->a_retval = _POSIX_VDISABLE; 650 return (0); 651 default: 652 return (EINVAL); 653 } 654 /* NOTREACHED */ 655 } 656 657 /* 658 * Special device advisory byte-level locks. 659 */ 660 /* ARGSUSED */ 661 spec_advlock(ap) 662 struct vop_advlock_args /* { 663 struct vnode *a_vp; 664 caddr_t a_id; 665 int a_op; 666 struct flock *a_fl; 667 int a_flags; 668 } */ *ap; 669 { 670 671 return (EOPNOTSUPP); 672 } 673 674 /* 675 * Special device failed operation 676 */ 677 spec_ebadf() 678 { 679 680 return (EBADF); 681 } 682 683 /* 684 * Special device bad operation 685 */ 686 spec_badop() 687 { 688 689 panic("spec_badop called"); 690 /* NOTREACHED */ 691 } 692
Indexes created Tue Sep 23 20:09:58 GMT 2025