spec_vnops.c revision 1.176
1 /* $NetBSD: spec_vnops.c,v 1.176 2019/09/22 22:59:39 christos Exp $ */ 2 3 /*- 4 * Copyright (c) 2008 The NetBSD Foundation, Inc. 5 * 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 * 16 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 17 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 18 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 19 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 20 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 21 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 22 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 23 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 24 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 25 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 26 * POSSIBILITY OF SUCH DAMAGE. 27 */ 28 29 /* 30 * Copyright (c) 1989, 1993 31 * The Regents of the University of California. All rights reserved. 32 * 33 * Redistribution and use in source and binary forms, with or without 34 * modification, are permitted provided that the following conditions 35 * are met: 36 * 1. Redistributions of source code must retain the above copyright 37 * notice, this list of conditions and the following disclaimer. 38 * 2. Redistributions in binary form must reproduce the above copyright 39 * notice, this list of conditions and the following disclaimer in the 40 * documentation and/or other materials provided with the distribution. 41 * 3. Neither the name of the University nor the names of its contributors 42 * may be used to endorse or promote products derived from this software 43 * without specific prior written permission. 44 * 45 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 46 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 47 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 48 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 49 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 50 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 51 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 52 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 53 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 54 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 55 * SUCH DAMAGE. 56 * 57 * @(#)spec_vnops.c 8.15 (Berkeley) 7/14/95 58 */ 59 60 #include <sys/cdefs.h> 61 __KERNEL_RCSID(0, "$NetBSD: spec_vnops.c,v 1.176 2019/09/22 22:59:39 christos Exp $"); 62 63 #include <sys/param.h> 64 #include <sys/proc.h> 65 #include <sys/systm.h> 66 #include <sys/kernel.h> 67 #include <sys/conf.h> 68 #include <sys/buf.h> 69 #include <sys/mount.h> 70 #include <sys/namei.h> 71 #include <sys/vnode_impl.h> 72 #include <sys/stat.h> 73 #include <sys/errno.h> 74 #include <sys/ioctl.h> 75 #include <sys/poll.h> 76 #include <sys/file.h> 77 #include <sys/disklabel.h> 78 #include <sys/disk.h> 79 #include <sys/lockf.h> 80 #include <sys/tty.h> 81 #include <sys/kauth.h> 82 #include <sys/fstrans.h> 83 #include <sys/module.h> 84 85 #include <miscfs/genfs/genfs.h> 86 #include <miscfs/specfs/specdev.h> 87 88 /* symbolic sleep message strings for devices */ 89 const char devopn[] = "devopn"; 90 const char devio[] = "devio"; 91 const char devwait[] = "devwait"; 92 const char devin[] = "devin"; 93 const char devout[] = "devout"; 94 const char devioc[] = "devioc"; 95 const char devcls[] = "devcls"; 96 97 #define SPECHSZ 64 98 #if ((SPECHSZ&(SPECHSZ-1)) == 0) 99 #define SPECHASH(rdev) (((rdev>>5)+(rdev))&(SPECHSZ-1)) 100 #else 101 #define SPECHASH(rdev) (((unsigned)((rdev>>5)+(rdev)))%SPECHSZ) 102 #endif 103 104 static vnode_t *specfs_hash[SPECHSZ]; 105 extern struct mount *dead_rootmount; 106 107 /* 108 * This vnode operations vector is used for special device nodes 109 * created from whole cloth by the kernel. For the ops vector for 110 * vnodes built from special devices found in a filesystem, see (e.g) 111 * ffs_specop_entries[] in ffs_vnops.c or the equivalent for other 112 * filesystems. 113 */ 114 115 int (**spec_vnodeop_p)(void *); 116 const struct vnodeopv_entry_desc spec_vnodeop_entries[] = { 117 { &vop_default_desc, vn_default_error }, 118 { &vop_lookup_desc, spec_lookup }, /* lookup */ 119 { &vop_create_desc, spec_create }, /* create */ 120 { &vop_mknod_desc, spec_mknod }, /* mknod */ 121 { &vop_open_desc, spec_open }, /* open */ 122 { &vop_close_desc, spec_close }, /* close */ 123 { &vop_access_desc, spec_access }, /* access */ 124 { &vop_getattr_desc, spec_getattr }, /* getattr */ 125 { &vop_setattr_desc, spec_setattr }, /* setattr */ 126 { &vop_read_desc, spec_read }, /* read */ 127 { &vop_write_desc, spec_write }, /* write */ 128 { &vop_fallocate_desc, spec_fallocate }, /* fallocate */ 129 { &vop_fdiscard_desc, spec_fdiscard }, /* fdiscard */ 130 { &vop_fcntl_desc, spec_fcntl }, /* fcntl */ 131 { &vop_ioctl_desc, spec_ioctl }, /* ioctl */ 132 { &vop_poll_desc, spec_poll }, /* poll */ 133 { &vop_kqfilter_desc, spec_kqfilter }, /* kqfilter */ 134 { &vop_revoke_desc, spec_revoke }, /* revoke */ 135 { &vop_mmap_desc, spec_mmap }, /* mmap */ 136 { &vop_fsync_desc, spec_fsync }, /* fsync */ 137 { &vop_seek_desc, spec_seek }, /* seek */ 138 { &vop_remove_desc, spec_remove }, /* remove */ 139 { &vop_link_desc, spec_link }, /* link */ 140 { &vop_rename_desc, spec_rename }, /* rename */ 141 { &vop_mkdir_desc, spec_mkdir }, /* mkdir */ 142 { &vop_rmdir_desc, spec_rmdir }, /* rmdir */ 143 { &vop_symlink_desc, spec_symlink }, /* symlink */ 144 { &vop_readdir_desc, spec_readdir }, /* readdir */ 145 { &vop_readlink_desc, spec_readlink }, /* readlink */ 146 { &vop_abortop_desc, spec_abortop }, /* abortop */ 147 { &vop_inactive_desc, spec_inactive }, /* inactive */ 148 { &vop_reclaim_desc, spec_reclaim }, /* reclaim */ 149 { &vop_lock_desc, spec_lock }, /* lock */ 150 { &vop_unlock_desc, spec_unlock }, /* unlock */ 151 { &vop_bmap_desc, spec_bmap }, /* bmap */ 152 { &vop_strategy_desc, spec_strategy }, /* strategy */ 153 { &vop_print_desc, spec_print }, /* print */ 154 { &vop_islocked_desc, spec_islocked }, /* islocked */ 155 { &vop_pathconf_desc, spec_pathconf }, /* pathconf */ 156 { &vop_advlock_desc, spec_advlock }, /* advlock */ 157 { &vop_bwrite_desc, spec_bwrite }, /* bwrite */ 158 { &vop_getpages_desc, spec_getpages }, /* getpages */ 159 { &vop_putpages_desc, spec_putpages }, /* putpages */ 160 { NULL, NULL } 161 }; 162 const struct vnodeopv_desc spec_vnodeop_opv_desc = 163 { &spec_vnodeop_p, spec_vnodeop_entries }; 164 165 static kauth_listener_t rawio_listener; 166 167 /* Returns true if vnode is /dev/mem or /dev/kmem. */ 168 bool 169 iskmemvp(struct vnode *vp) 170 { 171 return ((vp->v_type == VCHR) && iskmemdev(vp->v_rdev)); 172 } 173 174 /* 175 * Returns true if dev is /dev/mem or /dev/kmem. 176 */ 177 int 178 iskmemdev(dev_t dev) 179 { 180 /* mem_no is emitted by config(8) to generated devsw.c */ 181 extern const int mem_no; 182 183 /* minor 14 is /dev/io on i386 with COMPAT_10 */ 184 return (major(dev) == mem_no && (minor(dev) < 2 || minor(dev) == 14)); 185 } 186 187 static int 188 rawio_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie, 189 void *arg0, void *arg1, void *arg2, void *arg3) 190 { 191 int result; 192 193 result = KAUTH_RESULT_DEFER; 194 195 if ((action != KAUTH_DEVICE_RAWIO_SPEC) && 196 (action != KAUTH_DEVICE_RAWIO_PASSTHRU)) 197 return result; 198 199 /* Access is mandated by permissions. */ 200 result = KAUTH_RESULT_ALLOW; 201 202 return result; 203 } 204 205 void 206 spec_init(void) 207 { 208 209 rawio_listener = kauth_listen_scope(KAUTH_SCOPE_DEVICE, 210 rawio_listener_cb, NULL); 211 } 212 213 /* 214 * Initialize a vnode that represents a device. 215 */ 216 void 217 spec_node_init(vnode_t *vp, dev_t rdev) 218 { 219 specnode_t *sn; 220 specdev_t *sd; 221 vnode_t *vp2; 222 vnode_t **vpp; 223 224 KASSERT(vp->v_type == VBLK || vp->v_type == VCHR); 225 KASSERT(vp->v_specnode == NULL); 226 227 /* 228 * Search the hash table for this device. If known, add a 229 * reference to the device structure. If not known, create 230 * a new entry to represent the device. In all cases add 231 * the vnode to the hash table. 232 */ 233 sn = kmem_alloc(sizeof(*sn), KM_SLEEP); 234 sd = kmem_alloc(sizeof(*sd), KM_SLEEP); 235 mutex_enter(&device_lock); 236 vpp = &specfs_hash[SPECHASH(rdev)]; 237 for (vp2 = *vpp; vp2 != NULL; vp2 = vp2->v_specnext) { 238 KASSERT(vp2->v_specnode != NULL); 239 if (rdev == vp2->v_rdev && vp->v_type == vp2->v_type) { 240 break; 241 } 242 } 243 if (vp2 == NULL) { 244 /* No existing record, create a new one. */ 245 sd->sd_rdev = rdev; 246 sd->sd_mountpoint = NULL; 247 sd->sd_lockf = NULL; 248 sd->sd_refcnt = 1; 249 sd->sd_opencnt = 0; 250 sd->sd_bdevvp = NULL; 251 sn->sn_dev = sd; 252 sd = NULL; 253 } else { 254 /* Use the existing record. */ 255 sn->sn_dev = vp2->v_specnode->sn_dev; 256 sn->sn_dev->sd_refcnt++; 257 } 258 /* Insert vnode into the hash chain. */ 259 sn->sn_opencnt = 0; 260 sn->sn_rdev = rdev; 261 sn->sn_gone = false; 262 vp->v_specnode = sn; 263 vp->v_specnext = *vpp; 264 *vpp = vp; 265 mutex_exit(&device_lock); 266 267 /* Free the record we allocated if unused. */ 268 if (sd != NULL) { 269 kmem_free(sd, sizeof(*sd)); 270 } 271 } 272 273 /* 274 * Lookup a vnode by device number and return it referenced. 275 */ 276 int 277 spec_node_lookup_by_dev(enum vtype type, dev_t dev, vnode_t **vpp) 278 { 279 int error; 280 vnode_t *vp; 281 282 mutex_enter(&device_lock); 283 for (vp = specfs_hash[SPECHASH(dev)]; vp; vp = vp->v_specnext) { 284 if (type == vp->v_type && dev == vp->v_rdev) { 285 mutex_enter(vp->v_interlock); 286 /* If clean or being cleaned, then ignore it. */ 287 if (vdead_check(vp, VDEAD_NOWAIT) == 0) 288 break; 289 mutex_exit(vp->v_interlock); 290 } 291 } 292 KASSERT(vp == NULL || mutex_owned(vp->v_interlock)); 293 if (vp == NULL) { 294 mutex_exit(&device_lock); 295 return ENOENT; 296 } 297 /* 298 * If it is an opened block device return the opened vnode. 299 */ 300 if (type == VBLK && vp->v_specnode->sn_dev->sd_bdevvp != NULL) { 301 mutex_exit(vp->v_interlock); 302 vp = vp->v_specnode->sn_dev->sd_bdevvp; 303 mutex_enter(vp->v_interlock); 304 } 305 mutex_exit(&device_lock); 306 error = vcache_vget(vp); 307 if (error != 0) 308 return error; 309 *vpp = vp; 310 311 return 0; 312 } 313 314 /* 315 * Lookup a vnode by file system mounted on and return it referenced. 316 */ 317 int 318 spec_node_lookup_by_mount(struct mount *mp, vnode_t **vpp) 319 { 320 int i, error; 321 vnode_t *vp, *vq; 322 323 mutex_enter(&device_lock); 324 for (i = 0, vq = NULL; i < SPECHSZ && vq == NULL; i++) { 325 for (vp = specfs_hash[i]; vp; vp = vp->v_specnext) { 326 if (vp->v_type != VBLK) 327 continue; 328 vq = vp->v_specnode->sn_dev->sd_bdevvp; 329 if (vq != NULL && 330 vq->v_specnode->sn_dev->sd_mountpoint == mp) 331 break; 332 vq = NULL; 333 } 334 } 335 if (vq == NULL) { 336 mutex_exit(&device_lock); 337 return ENOENT; 338 } 339 mutex_enter(vq->v_interlock); 340 mutex_exit(&device_lock); 341 error = vcache_vget(vq); 342 if (error != 0) 343 return error; 344 *vpp = vq; 345 346 return 0; 347 348 } 349 350 /* 351 * Get the file system mounted on this block device. 352 */ 353 struct mount * 354 spec_node_getmountedfs(vnode_t *devvp) 355 { 356 struct mount *mp; 357 358 KASSERT(devvp->v_type == VBLK); 359 mp = devvp->v_specnode->sn_dev->sd_mountpoint; 360 361 return mp; 362 } 363 364 /* 365 * Set the file system mounted on this block device. 366 */ 367 void 368 spec_node_setmountedfs(vnode_t *devvp, struct mount *mp) 369 { 370 struct dkwedge_info dkw; 371 372 KASSERT(devvp->v_type == VBLK); 373 KASSERT(devvp->v_specnode->sn_dev->sd_mountpoint == NULL || mp == NULL); 374 devvp->v_specnode->sn_dev->sd_mountpoint = mp; 375 if (mp == NULL) 376 return; 377 378 if (bdev_ioctl(devvp->v_rdev, DIOCGWEDGEINFO, &dkw, FREAD, curlwp) != 0) 379 return; 380 381 strlcpy(mp->mnt_stat.f_mntfromlabel, dkw.dkw_wname, 382 sizeof(mp->mnt_stat.f_mntfromlabel)); 383 } 384 385 /* 386 * A vnode representing a special device is going away. Close 387 * the device if the vnode holds it open. 388 */ 389 void 390 spec_node_revoke(vnode_t *vp) 391 { 392 specnode_t *sn; 393 specdev_t *sd; 394 395 sn = vp->v_specnode; 396 sd = sn->sn_dev; 397 398 KASSERT(vp->v_type == VBLK || vp->v_type == VCHR); 399 KASSERT(vp->v_specnode != NULL); 400 KASSERT(sn->sn_gone == false); 401 402 mutex_enter(&device_lock); 403 KASSERT(sn->sn_opencnt <= sd->sd_opencnt); 404 if (sn->sn_opencnt != 0) { 405 sd->sd_opencnt -= (sn->sn_opencnt - 1); 406 sn->sn_opencnt = 1; 407 sn->sn_gone = true; 408 mutex_exit(&device_lock); 409 410 VOP_CLOSE(vp, FNONBLOCK, NOCRED); 411 412 mutex_enter(&device_lock); 413 KASSERT(sn->sn_opencnt == 0); 414 } 415 mutex_exit(&device_lock); 416 } 417 418 /* 419 * A vnode representing a special device is being recycled. 420 * Destroy the specfs component. 421 */ 422 void 423 spec_node_destroy(vnode_t *vp) 424 { 425 specnode_t *sn; 426 specdev_t *sd; 427 vnode_t **vpp, *vp2; 428 int refcnt; 429 430 sn = vp->v_specnode; 431 sd = sn->sn_dev; 432 433 KASSERT(vp->v_type == VBLK || vp->v_type == VCHR); 434 KASSERT(vp->v_specnode != NULL); 435 KASSERT(sn->sn_opencnt == 0); 436 437 mutex_enter(&device_lock); 438 /* Remove from the hash and destroy the node. */ 439 vpp = &specfs_hash[SPECHASH(vp->v_rdev)]; 440 for (vp2 = *vpp;; vp2 = vp2->v_specnext) { 441 if (vp2 == NULL) { 442 panic("spec_node_destroy: corrupt hash"); 443 } 444 if (vp2 == vp) { 445 KASSERT(vp == *vpp); 446 *vpp = vp->v_specnext; 447 break; 448 } 449 if (vp2->v_specnext == vp) { 450 vp2->v_specnext = vp->v_specnext; 451 break; 452 } 453 } 454 sn = vp->v_specnode; 455 vp->v_specnode = NULL; 456 refcnt = sd->sd_refcnt--; 457 KASSERT(refcnt > 0); 458 mutex_exit(&device_lock); 459 460 /* If the device is no longer in use, destroy our record. */ 461 if (refcnt == 1) { 462 KASSERT(sd->sd_opencnt == 0); 463 KASSERT(sd->sd_bdevvp == NULL); 464 kmem_free(sd, sizeof(*sd)); 465 } 466 kmem_free(sn, sizeof(*sn)); 467 } 468 469 /* 470 * Trivial lookup routine that always fails. 471 */ 472 int 473 spec_lookup(void *v) 474 { 475 struct vop_lookup_v2_args /* { 476 struct vnode *a_dvp; 477 struct vnode **a_vpp; 478 struct componentname *a_cnp; 479 } */ *ap = v; 480 481 *ap->a_vpp = NULL; 482 return (ENOTDIR); 483 } 484 485 typedef int (*spec_ioctl_t)(dev_t, u_long, void *, int, struct lwp *); 486 487 /* 488 * Open a special file. 489 */ 490 /* ARGSUSED */ 491 int 492 spec_open(void *v) 493 { 494 struct vop_open_args /* { 495 struct vnode *a_vp; 496 int a_mode; 497 kauth_cred_t a_cred; 498 } */ *ap = v; 499 struct lwp *l; 500 struct vnode *vp; 501 dev_t dev; 502 int error; 503 enum kauth_device_req req; 504 specnode_t *sn; 505 specdev_t *sd; 506 spec_ioctl_t ioctl; 507 u_int gen; 508 const char *name; 509 struct partinfo pi; 510 511 l = curlwp; 512 vp = ap->a_vp; 513 dev = vp->v_rdev; 514 sn = vp->v_specnode; 515 sd = sn->sn_dev; 516 name = NULL; 517 gen = 0; 518 519 /* 520 * Don't allow open if fs is mounted -nodev. 521 */ 522 if (vp->v_mount && (vp->v_mount->mnt_flag & MNT_NODEV)) 523 return (ENXIO); 524 525 switch (ap->a_mode & (FREAD | FWRITE)) { 526 case FREAD | FWRITE: 527 req = KAUTH_REQ_DEVICE_RAWIO_SPEC_RW; 528 break; 529 case FWRITE: 530 req = KAUTH_REQ_DEVICE_RAWIO_SPEC_WRITE; 531 break; 532 default: 533 req = KAUTH_REQ_DEVICE_RAWIO_SPEC_READ; 534 break; 535 } 536 537 switch (vp->v_type) { 538 case VCHR: 539 error = kauth_authorize_device_spec(ap->a_cred, req, vp); 540 if (error != 0) 541 return (error); 542 543 /* 544 * Character devices can accept opens from multiple 545 * vnodes. 546 */ 547 mutex_enter(&device_lock); 548 if (sn->sn_gone) { 549 mutex_exit(&device_lock); 550 return (EBADF); 551 } 552 sd->sd_opencnt++; 553 sn->sn_opencnt++; 554 mutex_exit(&device_lock); 555 if (cdev_type(dev) == D_TTY) 556 vp->v_vflag |= VV_ISTTY; 557 VOP_UNLOCK(vp); 558 do { 559 const struct cdevsw *cdev; 560 561 gen = module_gen; 562 error = cdev_open(dev, ap->a_mode, S_IFCHR, l); 563 if (error != ENXIO) 564 break; 565 566 /* Check if we already have a valid driver */ 567 mutex_enter(&device_lock); 568 cdev = cdevsw_lookup(dev); 569 mutex_exit(&device_lock); 570 if (cdev != NULL) 571 break; 572 573 /* Get device name from devsw_conv array */ 574 if ((name = cdevsw_getname(major(dev))) == NULL) 575 break; 576 577 /* Try to autoload device module */ 578 (void) module_autoload(name, MODULE_CLASS_DRIVER); 579 } while (gen != module_gen); 580 581 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 582 break; 583 584 case VBLK: 585 error = kauth_authorize_device_spec(ap->a_cred, req, vp); 586 if (error != 0) 587 return (error); 588 589 /* 590 * For block devices, permit only one open. The buffer 591 * cache cannot remain self-consistent with multiple 592 * vnodes holding a block device open. 593 * 594 * Treat zero opencnt with non-NULL mountpoint as open. 595 * This may happen after forced detach of a mounted device. 596 */ 597 mutex_enter(&device_lock); 598 if (sn->sn_gone) { 599 mutex_exit(&device_lock); 600 return (EBADF); 601 } 602 if (sd->sd_opencnt != 0 || sd->sd_mountpoint != NULL) { 603 mutex_exit(&device_lock); 604 return EBUSY; 605 } 606 sn->sn_opencnt = 1; 607 sd->sd_opencnt = 1; 608 sd->sd_bdevvp = vp; 609 mutex_exit(&device_lock); 610 do { 611 const struct bdevsw *bdev; 612 613 gen = module_gen; 614 error = bdev_open(dev, ap->a_mode, S_IFBLK, l); 615 if (error != ENXIO) 616 break; 617 618 /* Check if we already have a valid driver */ 619 mutex_enter(&device_lock); 620 bdev = bdevsw_lookup(dev); 621 mutex_exit(&device_lock); 622 if (bdev != NULL) 623 break; 624 625 /* Get device name from devsw_conv array */ 626 if ((name = bdevsw_getname(major(dev))) == NULL) 627 break; 628 629 VOP_UNLOCK(vp); 630 631 /* Try to autoload device module */ 632 (void) module_autoload(name, MODULE_CLASS_DRIVER); 633 634 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 635 } while (gen != module_gen); 636 637 break; 638 639 case VNON: 640 case VLNK: 641 case VDIR: 642 case VREG: 643 case VBAD: 644 case VFIFO: 645 case VSOCK: 646 default: 647 return 0; 648 } 649 650 mutex_enter(&device_lock); 651 if (sn->sn_gone) { 652 if (error == 0) 653 error = EBADF; 654 } else if (error != 0) { 655 sd->sd_opencnt--; 656 sn->sn_opencnt--; 657 if (vp->v_type == VBLK) 658 sd->sd_bdevvp = NULL; 659 660 } 661 mutex_exit(&device_lock); 662 663 if (cdev_type(dev) != D_DISK || error != 0) 664 return error; 665 666 667 ioctl = vp->v_type == VCHR ? cdev_ioctl : bdev_ioctl; 668 error = (*ioctl)(vp->v_rdev, DIOCGPARTINFO, &pi, FREAD, curlwp); 669 if (error == 0) 670 uvm_vnp_setsize(vp, (voff_t)pi.pi_secsize * pi.pi_size); 671 672 return 0; 673 } 674 675 /* 676 * Vnode op for read 677 */ 678 /* ARGSUSED */ 679 int 680 spec_read(void *v) 681 { 682 struct vop_read_args /* { 683 struct vnode *a_vp; 684 struct uio *a_uio; 685 int a_ioflag; 686 kauth_cred_t a_cred; 687 } */ *ap = v; 688 struct vnode *vp = ap->a_vp; 689 struct uio *uio = ap->a_uio; 690 struct lwp *l = curlwp; 691 struct buf *bp; 692 daddr_t bn; 693 int bsize, bscale; 694 struct partinfo pi; 695 int n, on; 696 int error = 0; 697 698 KASSERT(uio->uio_rw == UIO_READ); 699 KASSERTMSG(VMSPACE_IS_KERNEL_P(uio->uio_vmspace) || 700 uio->uio_vmspace == curproc->p_vmspace, 701 "vmspace belongs to neither kernel nor curproc"); 702 703 if (uio->uio_resid == 0) 704 return (0); 705 706 switch (vp->v_type) { 707 708 case VCHR: 709 VOP_UNLOCK(vp); 710 error = cdev_read(vp->v_rdev, uio, ap->a_ioflag); 711 vn_lock(vp, LK_SHARED | LK_RETRY); 712 return (error); 713 714 case VBLK: 715 KASSERT(vp == vp->v_specnode->sn_dev->sd_bdevvp); 716 if (uio->uio_offset < 0) 717 return (EINVAL); 718 719 if (bdev_ioctl(vp->v_rdev, DIOCGPARTINFO, &pi, FREAD, l) == 0) 720 bsize = pi.pi_bsize; 721 else 722 bsize = BLKDEV_IOSIZE; 723 724 bscale = bsize >> DEV_BSHIFT; 725 do { 726 bn = (uio->uio_offset >> DEV_BSHIFT) &~ (bscale - 1); 727 on = uio->uio_offset % bsize; 728 n = uimin((unsigned)(bsize - on), uio->uio_resid); 729 error = bread(vp, bn, bsize, 0, &bp); 730 if (error) { 731 return (error); 732 } 733 n = uimin(n, bsize - bp->b_resid); 734 error = uiomove((char *)bp->b_data + on, n, uio); 735 brelse(bp, 0); 736 } while (error == 0 && uio->uio_resid > 0 && n != 0); 737 return (error); 738 739 default: 740 panic("spec_read type"); 741 } 742 /* NOTREACHED */ 743 } 744 745 /* 746 * Vnode op for write 747 */ 748 /* ARGSUSED */ 749 int 750 spec_write(void *v) 751 { 752 struct vop_write_args /* { 753 struct vnode *a_vp; 754 struct uio *a_uio; 755 int a_ioflag; 756 kauth_cred_t a_cred; 757 } */ *ap = v; 758 struct vnode *vp = ap->a_vp; 759 struct uio *uio = ap->a_uio; 760 struct lwp *l = curlwp; 761 struct buf *bp; 762 daddr_t bn; 763 int bsize, bscale; 764 struct partinfo pi; 765 int n, on; 766 int error = 0; 767 768 KASSERT(uio->uio_rw == UIO_WRITE); 769 KASSERTMSG(VMSPACE_IS_KERNEL_P(uio->uio_vmspace) || 770 uio->uio_vmspace == curproc->p_vmspace, 771 "vmspace belongs to neither kernel nor curproc"); 772 773 switch (vp->v_type) { 774 775 case VCHR: 776 VOP_UNLOCK(vp); 777 error = cdev_write(vp->v_rdev, uio, ap->a_ioflag); 778 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 779 return (error); 780 781 case VBLK: 782 KASSERT(vp == vp->v_specnode->sn_dev->sd_bdevvp); 783 if (uio->uio_resid == 0) 784 return (0); 785 if (uio->uio_offset < 0) 786 return (EINVAL); 787 788 if (bdev_ioctl(vp->v_rdev, DIOCGPARTINFO, &pi, FREAD, l) == 0) 789 bsize = pi.pi_bsize; 790 else 791 bsize = BLKDEV_IOSIZE; 792 793 bscale = bsize >> DEV_BSHIFT; 794 do { 795 bn = (uio->uio_offset >> DEV_BSHIFT) &~ (bscale - 1); 796 on = uio->uio_offset % bsize; 797 n = uimin((unsigned)(bsize - on), uio->uio_resid); 798 if (n == bsize) 799 bp = getblk(vp, bn, bsize, 0, 0); 800 else 801 error = bread(vp, bn, bsize, B_MODIFY, &bp); 802 if (error) { 803 return (error); 804 } 805 n = uimin(n, bsize - bp->b_resid); 806 error = uiomove((char *)bp->b_data + on, n, uio); 807 if (error) 808 brelse(bp, 0); 809 else { 810 if (n + on == bsize) 811 bawrite(bp); 812 else 813 bdwrite(bp); 814 error = bp->b_error; 815 } 816 } while (error == 0 && uio->uio_resid > 0 && n != 0); 817 return (error); 818 819 default: 820 panic("spec_write type"); 821 } 822 /* NOTREACHED */ 823 } 824 825 /* 826 * fdiscard, which on disk devices becomes TRIM. 827 */ 828 int 829 spec_fdiscard(void *v) 830 { 831 struct vop_fdiscard_args /* { 832 struct vnode *a_vp; 833 off_t a_pos; 834 off_t a_len; 835 } */ *ap = v; 836 struct vnode *vp; 837 dev_t dev; 838 839 vp = ap->a_vp; 840 dev = NODEV; 841 842 mutex_enter(vp->v_interlock); 843 if (vdead_check(vp, VDEAD_NOWAIT) == 0 && vp->v_specnode != NULL) { 844 dev = vp->v_rdev; 845 } 846 mutex_exit(vp->v_interlock); 847 848 if (dev == NODEV) { 849 return ENXIO; 850 } 851 852 switch (vp->v_type) { 853 case VCHR: 854 // this is not stored for character devices 855 //KASSERT(vp == vp->v_specnode->sn_dev->sd_cdevvp); 856 return cdev_discard(dev, ap->a_pos, ap->a_len); 857 case VBLK: 858 KASSERT(vp == vp->v_specnode->sn_dev->sd_bdevvp); 859 return bdev_discard(dev, ap->a_pos, ap->a_len); 860 default: 861 panic("spec_fdiscard: not a device\n"); 862 } 863 } 864 865 /* 866 * Device ioctl operation. 867 */ 868 /* ARGSUSED */ 869 int 870 spec_ioctl(void *v) 871 { 872 struct vop_ioctl_args /* { 873 struct vnode *a_vp; 874 u_long a_command; 875 void *a_data; 876 int a_fflag; 877 kauth_cred_t a_cred; 878 } */ *ap = v; 879 struct vnode *vp; 880 dev_t dev; 881 882 /* 883 * Extract all the info we need from the vnode, taking care to 884 * avoid a race with VOP_REVOKE(). 885 */ 886 887 vp = ap->a_vp; 888 dev = NODEV; 889 mutex_enter(vp->v_interlock); 890 if (vdead_check(vp, VDEAD_NOWAIT) == 0 && vp->v_specnode) { 891 dev = vp->v_rdev; 892 } 893 mutex_exit(vp->v_interlock); 894 if (dev == NODEV) { 895 return ENXIO; 896 } 897 898 switch (vp->v_type) { 899 900 case VCHR: 901 return cdev_ioctl(dev, ap->a_command, ap->a_data, 902 ap->a_fflag, curlwp); 903 904 case VBLK: 905 KASSERT(vp == vp->v_specnode->sn_dev->sd_bdevvp); 906 return bdev_ioctl(dev, ap->a_command, ap->a_data, 907 ap->a_fflag, curlwp); 908 909 default: 910 panic("spec_ioctl"); 911 /* NOTREACHED */ 912 } 913 } 914 915 /* ARGSUSED */ 916 int 917 spec_poll(void *v) 918 { 919 struct vop_poll_args /* { 920 struct vnode *a_vp; 921 int a_events; 922 } */ *ap = v; 923 struct vnode *vp; 924 dev_t dev; 925 926 /* 927 * Extract all the info we need from the vnode, taking care to 928 * avoid a race with VOP_REVOKE(). 929 */ 930 931 vp = ap->a_vp; 932 dev = NODEV; 933 mutex_enter(vp->v_interlock); 934 if (vdead_check(vp, VDEAD_NOWAIT) == 0 && vp->v_specnode) { 935 dev = vp->v_rdev; 936 } 937 mutex_exit(vp->v_interlock); 938 if (dev == NODEV) { 939 return POLLERR; 940 } 941 942 switch (vp->v_type) { 943 944 case VCHR: 945 return cdev_poll(dev, ap->a_events, curlwp); 946 947 default: 948 return (genfs_poll(v)); 949 } 950 } 951 952 /* ARGSUSED */ 953 int 954 spec_kqfilter(void *v) 955 { 956 struct vop_kqfilter_args /* { 957 struct vnode *a_vp; 958 struct proc *a_kn; 959 } */ *ap = v; 960 dev_t dev; 961 962 switch (ap->a_vp->v_type) { 963 964 case VCHR: 965 dev = ap->a_vp->v_rdev; 966 return cdev_kqfilter(dev, ap->a_kn); 967 default: 968 /* 969 * Block devices don't support kqfilter, and refuse it 970 * for any other files (like those vflush()ed) too. 971 */ 972 return (EOPNOTSUPP); 973 } 974 } 975 976 /* 977 * Allow mapping of only D_DISK. This is called only for VBLK. 978 */ 979 int 980 spec_mmap(void *v) 981 { 982 struct vop_mmap_args /* { 983 struct vnode *a_vp; 984 vm_prot_t a_prot; 985 kauth_cred_t a_cred; 986 } */ *ap = v; 987 struct vnode *vp = ap->a_vp; 988 989 KASSERT(vp->v_type == VBLK); 990 if (bdev_type(vp->v_rdev) != D_DISK) 991 return EINVAL; 992 993 return 0; 994 } 995 996 /* 997 * Synch buffers associated with a block device 998 */ 999 /* ARGSUSED */ 1000 int 1001 spec_fsync(void *v) 1002 { 1003 struct vop_fsync_args /* { 1004 struct vnode *a_vp; 1005 kauth_cred_t a_cred; 1006 int a_flags; 1007 off_t offlo; 1008 off_t offhi; 1009 } */ *ap = v; 1010 struct vnode *vp = ap->a_vp; 1011 struct mount *mp; 1012 int error; 1013 1014 if (vp->v_type == VBLK) { 1015 if ((mp = spec_node_getmountedfs(vp)) != NULL) { 1016 error = VFS_FSYNC(mp, vp, ap->a_flags); 1017 if (error != EOPNOTSUPP) 1018 return error; 1019 } 1020 return vflushbuf(vp, ap->a_flags); 1021 } 1022 return (0); 1023 } 1024 1025 /* 1026 * Just call the device strategy routine 1027 */ 1028 int 1029 spec_strategy(void *v) 1030 { 1031 struct vop_strategy_args /* { 1032 struct vnode *a_vp; 1033 struct buf *a_bp; 1034 } */ *ap = v; 1035 struct vnode *vp = ap->a_vp; 1036 struct buf *bp = ap->a_bp; 1037 dev_t dev; 1038 int error; 1039 1040 dev = NODEV; 1041 1042 /* 1043 * Extract all the info we need from the vnode, taking care to 1044 * avoid a race with VOP_REVOKE(). 1045 */ 1046 1047 mutex_enter(vp->v_interlock); 1048 if (vdead_check(vp, VDEAD_NOWAIT) == 0 && vp->v_specnode != NULL) { 1049 KASSERT(vp == vp->v_specnode->sn_dev->sd_bdevvp); 1050 dev = vp->v_rdev; 1051 } 1052 mutex_exit(vp->v_interlock); 1053 1054 if (dev == NODEV) { 1055 error = ENXIO; 1056 goto out; 1057 } 1058 bp->b_dev = dev; 1059 1060 if (!(bp->b_flags & B_READ)) { 1061 #ifdef DIAGNOSTIC 1062 if (bp->b_vp && bp->b_vp->v_type == VBLK) { 1063 struct mount *mp = spec_node_getmountedfs(bp->b_vp); 1064 1065 if (mp && (mp->mnt_flag & MNT_RDONLY)) { 1066 printf("%s blk %"PRId64" written while ro!\n", 1067 mp->mnt_stat.f_mntonname, bp->b_blkno); 1068 } 1069 } 1070 #endif /* DIAGNOSTIC */ 1071 error = fscow_run(bp, false); 1072 if (error) 1073 goto out; 1074 } 1075 bdev_strategy(bp); 1076 1077 return 0; 1078 1079 out: 1080 bp->b_error = error; 1081 bp->b_resid = bp->b_bcount; 1082 biodone(bp); 1083 1084 return error; 1085 } 1086 1087 int 1088 spec_inactive(void *v) 1089 { 1090 struct vop_inactive_v2_args /* { 1091 struct vnode *a_vp; 1092 struct bool *a_recycle; 1093 } */ *ap = v; 1094 1095 KASSERT(ap->a_vp->v_mount == dead_rootmount); 1096 *ap->a_recycle = true; 1097 1098 return 0; 1099 } 1100 1101 int 1102 spec_reclaim(void *v) 1103 { 1104 struct vop_reclaim_v2_args /* { 1105 struct vnode *a_vp; 1106 } */ *ap = v; 1107 struct vnode *vp = ap->a_vp; 1108 1109 VOP_UNLOCK(vp); 1110 1111 KASSERT(vp->v_mount == dead_rootmount); 1112 return 0; 1113 } 1114 1115 /* 1116 * This is a noop, simply returning what one has been given. 1117 */ 1118 int 1119 spec_bmap(void *v) 1120 { 1121 struct vop_bmap_args /* { 1122 struct vnode *a_vp; 1123 daddr_t a_bn; 1124 struct vnode **a_vpp; 1125 daddr_t *a_bnp; 1126 int *a_runp; 1127 } */ *ap = v; 1128 1129 if (ap->a_vpp != NULL) 1130 *ap->a_vpp = ap->a_vp; 1131 if (ap->a_bnp != NULL) 1132 *ap->a_bnp = ap->a_bn; 1133 if (ap->a_runp != NULL) 1134 *ap->a_runp = (MAXBSIZE >> DEV_BSHIFT) - 1; 1135 return (0); 1136 } 1137 1138 /* 1139 * Device close routine 1140 */ 1141 /* ARGSUSED */ 1142 int 1143 spec_close(void *v) 1144 { 1145 struct vop_close_args /* { 1146 struct vnode *a_vp; 1147 int a_fflag; 1148 kauth_cred_t a_cred; 1149 } */ *ap = v; 1150 struct vnode *vp = ap->a_vp; 1151 struct session *sess; 1152 dev_t dev = vp->v_rdev; 1153 int flags = ap->a_fflag; 1154 int mode, error, count; 1155 specnode_t *sn; 1156 specdev_t *sd; 1157 1158 mutex_enter(vp->v_interlock); 1159 sn = vp->v_specnode; 1160 sd = sn->sn_dev; 1161 /* 1162 * If we're going away soon, make this non-blocking. 1163 * Also ensures that we won't wedge in vn_lock below. 1164 */ 1165 if (vdead_check(vp, VDEAD_NOWAIT) != 0) 1166 flags |= FNONBLOCK; 1167 mutex_exit(vp->v_interlock); 1168 1169 switch (vp->v_type) { 1170 1171 case VCHR: 1172 /* 1173 * Hack: a tty device that is a controlling terminal 1174 * has a reference from the session structure. We 1175 * cannot easily tell that a character device is a 1176 * controlling terminal, unless it is the closing 1177 * process' controlling terminal. In that case, if the 1178 * open count is 1 release the reference from the 1179 * session. Also, remove the link from the tty back to 1180 * the session and pgrp. 1181 * 1182 * XXX V. fishy. 1183 */ 1184 mutex_enter(proc_lock); 1185 sess = curlwp->l_proc->p_session; 1186 if (sn->sn_opencnt == 1 && vp == sess->s_ttyvp) { 1187 mutex_spin_enter(&tty_lock); 1188 sess->s_ttyvp = NULL; 1189 if (sess->s_ttyp->t_session != NULL) { 1190 sess->s_ttyp->t_pgrp = NULL; 1191 sess->s_ttyp->t_session = NULL; 1192 mutex_spin_exit(&tty_lock); 1193 /* Releases proc_lock. */ 1194 proc_sessrele(sess); 1195 } else { 1196 mutex_spin_exit(&tty_lock); 1197 if (sess->s_ttyp->t_pgrp != NULL) 1198 panic("spec_close: spurious pgrp ref"); 1199 mutex_exit(proc_lock); 1200 } 1201 vrele(vp); 1202 } else 1203 mutex_exit(proc_lock); 1204 1205 /* 1206 * If the vnode is locked, then we are in the midst 1207 * of forcably closing the device, otherwise we only 1208 * close on last reference. 1209 */ 1210 mode = S_IFCHR; 1211 break; 1212 1213 case VBLK: 1214 KASSERT(vp == vp->v_specnode->sn_dev->sd_bdevvp); 1215 /* 1216 * On last close of a block device (that isn't mounted) 1217 * we must invalidate any in core blocks, so that 1218 * we can, for instance, change floppy disks. 1219 */ 1220 error = vinvalbuf(vp, V_SAVE, ap->a_cred, curlwp, 0, 0); 1221 if (error) 1222 return (error); 1223 /* 1224 * We do not want to really close the device if it 1225 * is still in use unless we are trying to close it 1226 * forcibly. Since every use (buffer, vnode, swap, cmap) 1227 * holds a reference to the vnode, and because we mark 1228 * any other vnodes that alias this device, when the 1229 * sum of the reference counts on all the aliased 1230 * vnodes descends to one, we are on last close. 1231 */ 1232 mode = S_IFBLK; 1233 break; 1234 1235 default: 1236 panic("spec_close: not special"); 1237 } 1238 1239 mutex_enter(&device_lock); 1240 sn->sn_opencnt--; 1241 count = --sd->sd_opencnt; 1242 if (vp->v_type == VBLK) 1243 sd->sd_bdevvp = NULL; 1244 mutex_exit(&device_lock); 1245 1246 if (count != 0 && (vp->v_type != VCHR || !(cdev_flags(dev) & D_MCLOSE))) 1247 return 0; 1248 1249 /* 1250 * If we're able to block, release the vnode lock & reacquire. We 1251 * might end up sleeping for someone else who wants our queues. They 1252 * won't get them if we hold the vnode locked. 1253 */ 1254 if (!(flags & FNONBLOCK)) 1255 VOP_UNLOCK(vp); 1256 1257 if (vp->v_type == VBLK) 1258 error = bdev_close(dev, flags, mode, curlwp); 1259 else 1260 error = cdev_close(dev, flags, mode, curlwp); 1261 1262 if (!(flags & FNONBLOCK)) 1263 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 1264 1265 return (error); 1266 } 1267 1268 /* 1269 * Print out the contents of a special device vnode. 1270 */ 1271 int 1272 spec_print(void *v) 1273 { 1274 struct vop_print_args /* { 1275 struct vnode *a_vp; 1276 } */ *ap = v; 1277 1278 printf("dev %llu, %llu\n", (unsigned long long)major(ap->a_vp->v_rdev), 1279 (unsigned long long)minor(ap->a_vp->v_rdev)); 1280 return 0; 1281 } 1282 1283 /* 1284 * Return POSIX pathconf information applicable to special devices. 1285 */ 1286 int 1287 spec_pathconf(void *v) 1288 { 1289 struct vop_pathconf_args /* { 1290 struct vnode *a_vp; 1291 int a_name; 1292 register_t *a_retval; 1293 } */ *ap = v; 1294 1295 switch (ap->a_name) { 1296 case _PC_LINK_MAX: 1297 *ap->a_retval = LINK_MAX; 1298 return (0); 1299 case _PC_MAX_CANON: 1300 *ap->a_retval = MAX_CANON; 1301 return (0); 1302 case _PC_MAX_INPUT: 1303 *ap->a_retval = MAX_INPUT; 1304 return (0); 1305 case _PC_PIPE_BUF: 1306 *ap->a_retval = PIPE_BUF; 1307 return (0); 1308 case _PC_CHOWN_RESTRICTED: 1309 *ap->a_retval = 1; 1310 return (0); 1311 case _PC_VDISABLE: 1312 *ap->a_retval = _POSIX_VDISABLE; 1313 return (0); 1314 case _PC_SYNC_IO: 1315 *ap->a_retval = 1; 1316 return (0); 1317 default: 1318 return (EINVAL); 1319 } 1320 /* NOTREACHED */ 1321 } 1322 1323 /* 1324 * Advisory record locking support. 1325 */ 1326 int 1327 spec_advlock(void *v) 1328 { 1329 struct vop_advlock_args /* { 1330 struct vnode *a_vp; 1331 void *a_id; 1332 int a_op; 1333 struct flock *a_fl; 1334 int a_flags; 1335 } */ *ap = v; 1336 struct vnode *vp = ap->a_vp; 1337 1338 return lf_advlock(ap, &vp->v_speclockf, (off_t)0); 1339 } 1340
Indexes created Fri Sep 26 20:09:58 GMT 2025