1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 /* 22 * Copyright 2006 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #pragma ident "%Z%%M% %I% %E% SMI" 27 28 /* 29 * This is the /dev (hence, the sdev_ prefix) filesystem. 30 */ 31 32 #include <sys/types.h> 33 #include <sys/param.h> 34 #include <sys/sysmacros.h> 35 #include <sys/systm.h> 36 #include <sys/kmem.h> 37 #include <sys/time.h> 38 #include <sys/pathname.h> 39 #include <sys/vfs.h> 40 #include <sys/vnode.h> 41 #include <sys/file.h> 42 #include <sys/stat.h> 43 #include <sys/uio.h> 44 #include <sys/stat.h> 45 #include <sys/errno.h> 46 #include <sys/cmn_err.h> 47 #include <sys/cred.h> 48 #include <sys/statvfs.h> 49 #include <sys/policy.h> 50 #include <sys/mount.h> 51 #include <sys/debug.h> 52 #include <sys/modctl.h> 53 #include <sys/mkdev.h> 54 #include <fs/fs_subr.h> 55 #include <sys/fs/sdev_impl.h> 56 #include <sys/fs/sdev_node.h> 57 #include <sys/fs/snode.h> 58 #include <sys/fs/dv_node.h> 59 #include <sys/sunndi.h> 60 #include <sys/mntent.h> 61 62 /* 63 * /dev vfs operations. 64 */ 65 66 /* 67 * globals 68 */ 69 struct sdev_data *sdev_origins; /* mount info for origins under /dev */ 70 71 /* 72 * static 73 */ 74 static kmutex_t sdev_lock; /* protects global data */ 75 static major_t devmajor; /* the fictitious major we live on */ 76 static major_t devminor; /* the fictitious minor of this instance */ 77 static struct sdev_data *sdev_mntinfo = NULL; /* linked list of instances */ 78 static struct vnode *sdev_stale_attrvp; /* stale root attrvp after remount */ 79 static int sdev_mntinfo_cnt; /* mntinfo reference count */ 80 81 static int sdev_mount(struct vfs *, struct vnode *, struct mounta *, 82 struct cred *); 83 static int sdev_unmount(struct vfs *, int, struct cred *); 84 static int sdev_root(struct vfs *, struct vnode **); 85 static int sdev_statvfs(struct vfs *, struct statvfs64 *); 86 static void sdev_insert_mntinfo(struct sdev_data *); 87 static int devinit(int, char *); 88 89 static vfsdef_t sdev_vfssw = { 90 VFSDEF_VERSION, 91 "dev", /* type name string */ 92 devinit, /* init routine */ 93 VSW_CANREMOUNT, /* flags */ 94 NULL /* mount options table prototype */ 95 }; 96 97 98 /* 99 * Module linkage information 100 */ 101 static struct modlfs modlfs = { 102 &mod_fsops, "/dev filesystem %I%", &sdev_vfssw 103 }; 104 105 static struct modlinkage modlinkage = { 106 MODREV_1, (void *)&modlfs, NULL 107 }; 108 109 int 110 _init(void) 111 { 112 int e; 113 114 mutex_init(&sdev_lock, NULL, MUTEX_DEFAULT, NULL); 115 sdev_node_cache_init(); 116 sdev_devfsadm_lockinit(); 117 if ((e = mod_install(&modlinkage)) != 0) { 118 sdev_devfsadm_lockdestroy(); 119 sdev_node_cache_fini(); 120 mutex_destroy(&sdev_lock); 121 return (e); 122 } 123 return (0); 124 } 125 126 /* 127 * dev module remained loaded for the global /dev instance 128 */ 129 int 130 _fini(void) 131 { 132 return (EBUSY); 133 } 134 135 int 136 _info(struct modinfo *modinfop) 137 { 138 return (mod_info(&modlinkage, modinfop)); 139 } 140 141 /*ARGSUSED*/ 142 static int 143 devinit(int fstype, char *name) 144 { 145 static const fs_operation_def_t dev_vfsops_tbl[] = { 146 VFSNAME_MOUNT, sdev_mount, /* mount file system */ 147 VFSNAME_UNMOUNT, sdev_unmount, /* unmount file system */ 148 VFSNAME_ROOT, sdev_root, /* get root vnode */ 149 VFSNAME_STATVFS, sdev_statvfs, /* get file system statistics */ 150 NULL, NULL 151 }; 152 153 int error; 154 extern major_t getudev(void); 155 156 devtype = fstype; 157 158 error = vfs_setfsops(fstype, dev_vfsops_tbl, NULL); 159 if (error != 0) { 160 cmn_err(CE_WARN, "devinit: bad vfs ops tbl"); 161 return (error); 162 } 163 164 error = vn_make_ops("dev", sdev_vnodeops_tbl, &sdev_vnodeops); 165 if (error != 0) { 166 (void) vfs_freevfsops_by_type(fstype); 167 cmn_err(CE_WARN, "devinit: bad vnode ops tbl"); 168 return (error); 169 } 170 171 if ((devmajor = getudev()) == (major_t)-1) { 172 cmn_err(CE_WARN, "%s: can't get unique dev", sdev_vfssw.name); 173 return (1); 174 } 175 176 /* initialize negative cache */ 177 sdev_ncache_init(); 178 179 return (0); 180 } 181 182 /* 183 * Both mount point and backing store directory name are 184 * passed in from userland 185 */ 186 static int 187 sdev_mount(struct vfs *vfsp, struct vnode *mvp, struct mounta *uap, 188 struct cred *cr) 189 { 190 struct sdev_data *sdev_data; 191 struct vnode *avp; 192 struct sdev_node *dv; 193 struct sdev_mountargs *args = NULL; 194 int error = 0; 195 dev_t devdev; 196 197 /* 198 * security check 199 */ 200 if ((secpolicy_fs_mount(cr, mvp, vfsp) != 0) || 201 (secpolicy_sys_devices(cr) != 0)) 202 return (EPERM); 203 204 /* 205 * Sanity check the mount point 206 */ 207 if (mvp->v_type != VDIR) 208 return (ENOTDIR); 209 210 /* 211 * Sanity Check for overlay mount. 212 */ 213 mutex_enter(&mvp->v_lock); 214 if ((uap->flags & MS_OVERLAY) == 0 && 215 (uap->flags & MS_REMOUNT) == 0 && 216 (mvp->v_count > 1 || (mvp->v_flag & VROOT))) { 217 mutex_exit(&mvp->v_lock); 218 return (EBUSY); 219 } 220 mutex_exit(&mvp->v_lock); 221 222 args = kmem_zalloc(sizeof (*args), KM_SLEEP); 223 224 if ((uap->flags & MS_DATA) && 225 (uap->datalen != 0 && uap->dataptr != NULL)) { 226 /* copy in the arguments */ 227 if (error = sdev_copyin_mountargs(uap, args)) 228 goto cleanup; 229 } 230 231 /* 232 * Sanity check the backing store 233 */ 234 if (args->sdev_attrdir) { 235 /* user supplied an attribute store */ 236 if (error = lookupname((char *)(uintptr_t)args->sdev_attrdir, 237 UIO_USERSPACE, FOLLOW, NULLVPP, &avp)) { 238 cmn_err(CE_NOTE, "/dev fs: lookup on attribute " 239 "directory %s failed", 240 (char *)(uintptr_t)args->sdev_attrdir); 241 goto cleanup; 242 } 243 244 if (avp->v_type != VDIR) { 245 VN_RELE(avp); 246 error = ENOTDIR; 247 goto cleanup; 248 } 249 } else { 250 /* use mountp as the attribute store */ 251 avp = mvp; 252 VN_HOLD(avp); 253 } 254 255 mutex_enter(&sdev_lock); 256 257 /* 258 * handling installation 259 */ 260 if (uap->flags & MS_REMOUNT) { 261 sdev_data = (struct sdev_data *)vfsp->vfs_data; 262 ASSERT(sdev_data); 263 264 dv = sdev_data->sdev_root; 265 ASSERT(dv == dv->sdev_dotdot); 266 267 /* 268 * mark all existing sdev_nodes (except root node) stale 269 */ 270 sdev_stale(dv); 271 272 /* Reset previous mountargs */ 273 if (sdev_data->sdev_mountargs) { 274 kmem_free(sdev_data->sdev_mountargs, 275 sizeof (struct sdev_mountargs)); 276 } 277 sdev_data->sdev_mountargs = args; 278 args = NULL; /* so it won't be freed below */ 279 280 sdev_stale_attrvp = dv->sdev_attrvp; 281 dv->sdev_attrvp = avp; 282 vfsp->vfs_mtime = ddi_get_time(); 283 284 mutex_exit(&sdev_lock); 285 goto cleanup; /* we're done */ 286 } 287 288 /* 289 * Create and initialize the vfs-private data. 290 */ 291 devdev = makedevice(devmajor, devminor); 292 while (vfs_devismounted(devdev)) { 293 devminor = (devminor + 1) & MAXMIN32; 294 295 /* 296 * All the minor numbers are used up. 297 */ 298 if (devminor == 0) { 299 mutex_exit(&sdev_lock); 300 VN_RELE(avp); 301 error = ENODEV; 302 goto cleanup; 303 } 304 305 devdev = makedevice(devmajor, devminor); 306 } 307 308 dv = sdev_mkroot(vfsp, devdev, mvp, avp, cr); 309 sdev_data = kmem_zalloc(sizeof (struct sdev_data), KM_SLEEP); 310 vfsp->vfs_dev = devdev; 311 vfsp->vfs_data = (caddr_t)sdev_data; 312 vfsp->vfs_fstype = devtype; 313 vfsp->vfs_bsize = DEV_BSIZE; 314 vfsp->vfs_mtime = ddi_get_time(); 315 vfs_make_fsid(&vfsp->vfs_fsid, vfsp->vfs_dev, devtype); 316 317 ASSERT(dv == dv->sdev_dotdot); 318 319 sdev_data->sdev_vfsp = vfsp; 320 sdev_data->sdev_root = dv; 321 sdev_data->sdev_mountargs = args; 322 323 /* get acl flavor from attribute dir */ 324 if (VOP_PATHCONF(avp, _PC_ACL_ENABLED, &sdev_data->sdev_acl_flavor, 325 kcred) != 0 || sdev_data->sdev_acl_flavor == 0) 326 sdev_data->sdev_acl_flavor = _ACL_ACLENT_ENABLED; 327 328 args = NULL; /* so it won't be freed below */ 329 sdev_insert_mntinfo(sdev_data); 330 mutex_exit(&sdev_lock); 331 332 if (!SDEV_IS_GLOBAL(dv)) { 333 ASSERT(sdev_origins); 334 dv->sdev_flags &= ~SDEV_GLOBAL; 335 dv->sdev_origin = sdev_origins->sdev_root; 336 } else { 337 sdev_ncache_setup(); 338 } 339 340 sdev_update_timestamps(dv->sdev_attrvp, 341 cr, AT_CTIME|AT_MTIME|AT_ATIME); 342 343 cleanup: 344 if (args) 345 kmem_free(args, sizeof (*args)); 346 return (error); 347 } 348 349 /* 350 * unmounting the non-global /dev instances, e.g. when deleting a Kevlar zone. 351 */ 352 static int 353 sdev_unmount(struct vfs *vfsp, int flag, struct cred *cr) 354 { 355 struct sdev_node *dv; 356 int error; 357 struct sdev_data *sdev_data, *prev, *next; 358 359 /* 360 * enforce the security policies 361 */ 362 if ((secpolicy_fs_unmount(cr, vfsp) != 0) || 363 (secpolicy_sys_devices(cr) != 0)) 364 return (EPERM); 365 366 if (flag & MS_FORCE) 367 return (ENOTSUP); 368 369 mutex_enter(&sdev_lock); 370 dv = VFSTOSDEVFS(vfsp)->sdev_root; 371 ASSERT(dv == dv->sdev_dotdot); 372 if (SDEVTOV(dv)->v_count > 1) { 373 mutex_exit(&sdev_lock); 374 return (EBUSY); 375 } 376 377 /* 378 * global instance remains mounted 379 */ 380 if (SDEV_IS_GLOBAL(dv)) { 381 mutex_exit(&sdev_lock); 382 return (EBUSY); 383 } 384 mutex_exit(&sdev_lock); 385 386 /* verify the v_count */ 387 if ((error = sdev_cleandir(dv, NULL, 0)) != 0) { 388 return (error); 389 } 390 ASSERT(SDEVTOV(dv)->v_count == 1); 391 392 /* release hold on root node and destroy it */ 393 SDEV_RELE(dv); 394 dv->sdev_nlink -= 2; 395 sdev_nodedestroy(dv, 0); 396 397 sdev_data = (struct sdev_data *)vfsp->vfs_data; 398 vfsp->vfs_data = (caddr_t)0; 399 400 /* 401 * XXX separate it into sdev_delete_mntinfo() if useful 402 */ 403 mutex_enter(&sdev_lock); 404 prev = sdev_data->sdev_prev; 405 next = sdev_data->sdev_next; 406 if (prev) 407 prev->sdev_next = next; 408 else 409 sdev_mntinfo = next; 410 if (next) 411 next->sdev_prev = prev; 412 mutex_exit(&sdev_lock); 413 414 if (sdev_data->sdev_mountargs) { 415 kmem_free(sdev_data->sdev_mountargs, 416 sizeof (struct sdev_mountargs)); 417 } 418 kmem_free(sdev_data, sizeof (struct sdev_data)); 419 return (0); 420 } 421 422 /* 423 * return root vnode for given vfs 424 */ 425 static int 426 sdev_root(struct vfs *vfsp, struct vnode **vpp) 427 { 428 *vpp = SDEVTOV(VFSTOSDEVFS(vfsp)->sdev_root); 429 VN_HOLD(*vpp); 430 return (0); 431 } 432 433 /* 434 * return 'generic superblock' information to userland. 435 * 436 * not much that we can usefully admit to here 437 */ 438 static int 439 sdev_statvfs(struct vfs *vfsp, struct statvfs64 *sbp) 440 { 441 dev32_t d32; 442 443 bzero(sbp, sizeof (*sbp)); 444 sbp->f_frsize = sbp->f_bsize = vfsp->vfs_bsize; 445 sbp->f_files = kmem_cache_stat(sdev_node_cache, "alloc"); 446 447 /* no illusions that free/avail files is relevant to dev */ 448 sbp->f_ffree = 0; 449 sbp->f_favail = 0; 450 451 /* no illusions that blocks are relevant to devfs */ 452 sbp->f_bfree = 0; 453 sbp->f_bavail = 0; 454 sbp->f_blocks = 0; 455 456 (void) cmpldev(&d32, vfsp->vfs_dev); 457 sbp->f_fsid = d32; 458 (void) strcpy(sbp->f_basetype, vfssw[devtype].vsw_name); 459 sbp->f_flag = vf_to_stf(vfsp->vfs_flag); 460 sbp->f_namemax = MAXNAMELEN - 1; 461 (void) strcpy(sbp->f_fstr, "dev"); 462 463 return (0); 464 } 465 466 int 467 sdev_module_register(char *mod_name, struct devname_ops *dev_ops) 468 { 469 struct devname_nsmap *map = NULL; 470 471 if (strcmp(mod_name, DEVNAME_NSCONFIG) == 0) { 472 devname_ns_ops = dev_ops; 473 return (0); 474 } 475 476 map = sdev_get_nsmap_by_module(mod_name); 477 if (map == NULL) 478 return (EFAULT); 479 480 rw_enter(&map->dir_lock, RW_WRITER); 481 map->dir_ops = dev_ops; 482 rw_exit(&map->dir_lock); 483 return (0); 484 } 485 486 static void 487 sdev_insert_mntinfo(struct sdev_data *data) 488 { 489 ASSERT(mutex_owned(&sdev_lock)); 490 data->sdev_next = sdev_mntinfo; 491 data->sdev_prev = NULL; 492 if (sdev_mntinfo) { 493 sdev_mntinfo->sdev_prev = data; 494 } else { 495 sdev_origins = data; 496 } 497 sdev_mntinfo = data; 498 } 499 500 struct sdev_data * 501 sdev_find_mntinfo(char *mntpt) 502 { 503 struct sdev_data *mntinfo; 504 505 mutex_enter(&sdev_lock); 506 mntinfo = sdev_mntinfo; 507 while (mntinfo) { 508 if (strcmp(mntpt, mntinfo->sdev_root->sdev_name) == 0) { 509 SDEVTOV(mntinfo->sdev_root)->v_count++; 510 break; 511 } 512 mntinfo = mntinfo->sdev_next; 513 } 514 mutex_exit(&sdev_lock); 515 return (mntinfo); 516 } 517 518 void 519 sdev_mntinfo_rele(struct sdev_data *mntinfo) 520 { 521 mutex_enter(&sdev_lock); 522 SDEVTOV(mntinfo->sdev_root)->v_count--; 523 mutex_exit(&sdev_lock); 524 } 525