17c478bd9Sstevel@tonic-gate /* 27c478bd9Sstevel@tonic-gate * CDDL HEADER START 37c478bd9Sstevel@tonic-gate * 47c478bd9Sstevel@tonic-gate * The contents of this file are subject to the terms of the 55a59a8b3Srsb * Common Development and Distribution License (the "License"). 65a59a8b3Srsb * You may not use this file except in compliance with the License. 77c478bd9Sstevel@tonic-gate * 87c478bd9Sstevel@tonic-gate * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 97c478bd9Sstevel@tonic-gate * or http://www.opensolaris.org/os/licensing. 107c478bd9Sstevel@tonic-gate * See the License for the specific language governing permissions 117c478bd9Sstevel@tonic-gate * and limitations under the License. 127c478bd9Sstevel@tonic-gate * 137c478bd9Sstevel@tonic-gate * When distributing Covered Code, include this CDDL HEADER in each 147c478bd9Sstevel@tonic-gate * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 157c478bd9Sstevel@tonic-gate * If applicable, add the following below this CDDL HEADER, with the 167c478bd9Sstevel@tonic-gate * fields enclosed by brackets "[]" replaced with your own identifying 177c478bd9Sstevel@tonic-gate * information: Portions Copyright [yyyy] [name of copyright owner] 187c478bd9Sstevel@tonic-gate * 197c478bd9Sstevel@tonic-gate * CDDL HEADER END 207c478bd9Sstevel@tonic-gate */ 217c478bd9Sstevel@tonic-gate /* 2279a28c7aSmarks * Copyright 2008 Sun Microsystems, Inc. All rights reserved. 237c478bd9Sstevel@tonic-gate * Use is subject to license terms. 247c478bd9Sstevel@tonic-gate */ 257c478bd9Sstevel@tonic-gate 267c478bd9Sstevel@tonic-gate #pragma ident "%Z%%M% %I% %E% SMI" 277c478bd9Sstevel@tonic-gate 287c478bd9Sstevel@tonic-gate #include <sys/param.h> 297c478bd9Sstevel@tonic-gate #include <sys/errno.h> 307c478bd9Sstevel@tonic-gate #include <sys/vfs.h> 31aa59c4cbSrsb #include <sys/vfs_opreg.h> 327c478bd9Sstevel@tonic-gate #include <sys/vnode.h> 337c478bd9Sstevel@tonic-gate #include <sys/uio.h> 347c478bd9Sstevel@tonic-gate #include <sys/pathname.h> 357c478bd9Sstevel@tonic-gate #include <sys/kmem.h> 367c478bd9Sstevel@tonic-gate #include <sys/cred.h> 377c478bd9Sstevel@tonic-gate #include <sys/statvfs.h> 387c478bd9Sstevel@tonic-gate #include <sys/fs/lofs_info.h> 397c478bd9Sstevel@tonic-gate #include <sys/fs/lofs_node.h> 407c478bd9Sstevel@tonic-gate #include <sys/mount.h> 417c478bd9Sstevel@tonic-gate #include <sys/mntent.h> 427c478bd9Sstevel@tonic-gate #include <sys/mkdev.h> 4345916cd2Sjpk #include <sys/priv.h> 447c478bd9Sstevel@tonic-gate #include <sys/sysmacros.h> 457c478bd9Sstevel@tonic-gate #include <sys/systm.h> 467c478bd9Sstevel@tonic-gate #include <sys/cmn_err.h> 477c478bd9Sstevel@tonic-gate #include <sys/policy.h> 4845916cd2Sjpk #include <sys/tsol/label.h> 497c478bd9Sstevel@tonic-gate #include "fs/fs_subr.h" 507c478bd9Sstevel@tonic-gate 517c478bd9Sstevel@tonic-gate /* 527c478bd9Sstevel@tonic-gate * This is the loadable module wrapper. 537c478bd9Sstevel@tonic-gate */ 547c478bd9Sstevel@tonic-gate #include <sys/modctl.h> 557c478bd9Sstevel@tonic-gate 567c478bd9Sstevel@tonic-gate static mntopts_t lofs_mntopts; 577c478bd9Sstevel@tonic-gate 587c478bd9Sstevel@tonic-gate static int lofsinit(int, char *); 597c478bd9Sstevel@tonic-gate 607c478bd9Sstevel@tonic-gate static vfsdef_t vfw = { 617c478bd9Sstevel@tonic-gate VFSDEF_VERSION, 627c478bd9Sstevel@tonic-gate "lofs", 637c478bd9Sstevel@tonic-gate lofsinit, 645a59a8b3Srsb VSW_HASPROTO|VSW_STATS, 657c478bd9Sstevel@tonic-gate &lofs_mntopts 667c478bd9Sstevel@tonic-gate }; 677c478bd9Sstevel@tonic-gate 687c478bd9Sstevel@tonic-gate /* 697c478bd9Sstevel@tonic-gate * LOFS mount options table 707c478bd9Sstevel@tonic-gate */ 717c478bd9Sstevel@tonic-gate static char *xattr_cancel[] = { MNTOPT_NOXATTR, NULL }; 727c478bd9Sstevel@tonic-gate static char *noxattr_cancel[] = { MNTOPT_XATTR, NULL }; 737c478bd9Sstevel@tonic-gate static char *sub_cancel[] = { MNTOPT_LOFS_NOSUB, NULL }; 747c478bd9Sstevel@tonic-gate static char *nosub_cancel[] = { MNTOPT_LOFS_SUB, NULL }; 757c478bd9Sstevel@tonic-gate 767c478bd9Sstevel@tonic-gate static mntopt_t mntopts[] = { 777c478bd9Sstevel@tonic-gate /* 787c478bd9Sstevel@tonic-gate * option name cancel option default arg flags 797c478bd9Sstevel@tonic-gate * private data 807c478bd9Sstevel@tonic-gate */ 817c478bd9Sstevel@tonic-gate { MNTOPT_XATTR, xattr_cancel, NULL, 0, 827c478bd9Sstevel@tonic-gate (void *)0 }, 837c478bd9Sstevel@tonic-gate { MNTOPT_NOXATTR, noxattr_cancel, NULL, 0, 847c478bd9Sstevel@tonic-gate (void *)0 }, 857c478bd9Sstevel@tonic-gate { MNTOPT_LOFS_SUB, sub_cancel, NULL, 0, 867c478bd9Sstevel@tonic-gate (void *)0 }, 877c478bd9Sstevel@tonic-gate { MNTOPT_LOFS_NOSUB, nosub_cancel, NULL, 0, 887c478bd9Sstevel@tonic-gate (void *)0 }, 897c478bd9Sstevel@tonic-gate }; 907c478bd9Sstevel@tonic-gate 917c478bd9Sstevel@tonic-gate static mntopts_t lofs_mntopts = { 927c478bd9Sstevel@tonic-gate sizeof (mntopts) / sizeof (mntopt_t), 937c478bd9Sstevel@tonic-gate mntopts 947c478bd9Sstevel@tonic-gate }; 957c478bd9Sstevel@tonic-gate 967c478bd9Sstevel@tonic-gate /* 977c478bd9Sstevel@tonic-gate * Module linkage information for the kernel. 987c478bd9Sstevel@tonic-gate */ 997c478bd9Sstevel@tonic-gate 1007c478bd9Sstevel@tonic-gate static struct modlfs modlfs = { 1017c478bd9Sstevel@tonic-gate &mod_fsops, "filesystem for lofs", &vfw 1027c478bd9Sstevel@tonic-gate }; 1037c478bd9Sstevel@tonic-gate 1047c478bd9Sstevel@tonic-gate static struct modlinkage modlinkage = { 1057c478bd9Sstevel@tonic-gate MODREV_1, (void *)&modlfs, NULL 1067c478bd9Sstevel@tonic-gate }; 1077c478bd9Sstevel@tonic-gate 1087c478bd9Sstevel@tonic-gate /* 1097c478bd9Sstevel@tonic-gate * This is the module initialization routine. 1107c478bd9Sstevel@tonic-gate */ 11145916cd2Sjpk 1127c478bd9Sstevel@tonic-gate int 11345916cd2Sjpk _init(void) 1147c478bd9Sstevel@tonic-gate { 1157c478bd9Sstevel@tonic-gate int status; 1167c478bd9Sstevel@tonic-gate 1177c478bd9Sstevel@tonic-gate lofs_subrinit(); 1187c478bd9Sstevel@tonic-gate status = mod_install(&modlinkage); 1197c478bd9Sstevel@tonic-gate if (status != 0) { 1207c478bd9Sstevel@tonic-gate /* 1217c478bd9Sstevel@tonic-gate * Cleanup previously initialized work. 1227c478bd9Sstevel@tonic-gate */ 1237c478bd9Sstevel@tonic-gate lofs_subrfini(); 1247c478bd9Sstevel@tonic-gate } 1257c478bd9Sstevel@tonic-gate 1267c478bd9Sstevel@tonic-gate return (status); 1277c478bd9Sstevel@tonic-gate } 1287c478bd9Sstevel@tonic-gate 1297c478bd9Sstevel@tonic-gate /* 1307c478bd9Sstevel@tonic-gate * Don't allow the lofs module to be unloaded for now. 1317c478bd9Sstevel@tonic-gate * There is a memory leak if it gets unloaded. 1327c478bd9Sstevel@tonic-gate */ 13345916cd2Sjpk 1347c478bd9Sstevel@tonic-gate int 13545916cd2Sjpk _fini(void) 1367c478bd9Sstevel@tonic-gate { 1377c478bd9Sstevel@tonic-gate return (EBUSY); 1387c478bd9Sstevel@tonic-gate } 1397c478bd9Sstevel@tonic-gate 1407c478bd9Sstevel@tonic-gate int 1417c478bd9Sstevel@tonic-gate _info(struct modinfo *modinfop) 1427c478bd9Sstevel@tonic-gate { 1437c478bd9Sstevel@tonic-gate return (mod_info(&modlinkage, modinfop)); 1447c478bd9Sstevel@tonic-gate } 1457c478bd9Sstevel@tonic-gate 1467c478bd9Sstevel@tonic-gate 1477c478bd9Sstevel@tonic-gate static int lofsfstype; 1487c478bd9Sstevel@tonic-gate vfsops_t *lo_vfsops; 1497c478bd9Sstevel@tonic-gate 1507c478bd9Sstevel@tonic-gate /* 1517c478bd9Sstevel@tonic-gate * lo mount vfsop 1527c478bd9Sstevel@tonic-gate * Set up mount info record and attach it to vfs struct. 1537c478bd9Sstevel@tonic-gate */ 1547c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 1557c478bd9Sstevel@tonic-gate static int 1567c478bd9Sstevel@tonic-gate lo_mount(struct vfs *vfsp, 1577c478bd9Sstevel@tonic-gate struct vnode *vp, 1587c478bd9Sstevel@tonic-gate struct mounta *uap, 1597c478bd9Sstevel@tonic-gate struct cred *cr) 1607c478bd9Sstevel@tonic-gate { 1617c478bd9Sstevel@tonic-gate int error; 1627c478bd9Sstevel@tonic-gate struct vnode *srootvp = NULL; /* the server's root */ 1637c478bd9Sstevel@tonic-gate struct vnode *realrootvp; 1647c478bd9Sstevel@tonic-gate struct loinfo *li; 1657c478bd9Sstevel@tonic-gate int nodev; 1667c478bd9Sstevel@tonic-gate 1677c478bd9Sstevel@tonic-gate nodev = vfs_optionisset(vfsp, MNTOPT_NODEVICES, NULL); 1687c478bd9Sstevel@tonic-gate 1697c478bd9Sstevel@tonic-gate if ((error = secpolicy_fs_mount(cr, vp, vfsp)) != 0) 1707c478bd9Sstevel@tonic-gate return (EPERM); 1717c478bd9Sstevel@tonic-gate 1727c478bd9Sstevel@tonic-gate /* 1737c478bd9Sstevel@tonic-gate * Loopback devices which get "nodevices" added can be done without 1747c478bd9Sstevel@tonic-gate * "nodevices" set because we cannot import devices into a zone 1757c478bd9Sstevel@tonic-gate * with loopback. Note that we have all zone privileges when 1767c478bd9Sstevel@tonic-gate * this happens; if not, we'd have gotten "nosuid". 1777c478bd9Sstevel@tonic-gate */ 1787c478bd9Sstevel@tonic-gate if (!nodev && vfs_optionisset(vfsp, MNTOPT_NODEVICES, NULL)) 1797c478bd9Sstevel@tonic-gate vfs_setmntopt(vfsp, MNTOPT_DEVICES, NULL, VFS_NODISPLAY); 1807c478bd9Sstevel@tonic-gate 1817c478bd9Sstevel@tonic-gate mutex_enter(&vp->v_lock); 1827c478bd9Sstevel@tonic-gate if (!(uap->flags & MS_OVERLAY) && 18345916cd2Sjpk (vp->v_count != 1 || (vp->v_flag & VROOT))) { 1847c478bd9Sstevel@tonic-gate mutex_exit(&vp->v_lock); 1857c478bd9Sstevel@tonic-gate return (EBUSY); 1867c478bd9Sstevel@tonic-gate } 1877c478bd9Sstevel@tonic-gate mutex_exit(&vp->v_lock); 1887c478bd9Sstevel@tonic-gate 1897c478bd9Sstevel@tonic-gate /* 1907c478bd9Sstevel@tonic-gate * Find real root, and make vfs point to real vfs 1917c478bd9Sstevel@tonic-gate */ 19279a28c7aSmarks 1937c478bd9Sstevel@tonic-gate if (error = lookupname(uap->spec, (uap->flags & MS_SYSSPACE) ? 19479a28c7aSmarks UIO_SYSSPACE : UIO_USERSPACE, FOLLOW, NULLVPP, &realrootvp)) 1957c478bd9Sstevel@tonic-gate return (error); 1967c478bd9Sstevel@tonic-gate 19745916cd2Sjpk /* 19845916cd2Sjpk * Enforce MAC policy if needed. 19945916cd2Sjpk * 20045916cd2Sjpk * Loopback mounts must not allow writing up. The dominance test 20145916cd2Sjpk * is intended to prevent a global zone caller from accidentally 20245916cd2Sjpk * creating write-up conditions between two labeled zones. 20345916cd2Sjpk * Local zones can't violate MAC on their own without help from 20445916cd2Sjpk * the global zone because they can't name a pathname that 20545916cd2Sjpk * they don't already have. 20645916cd2Sjpk * 20745916cd2Sjpk * The special case check for the NET_MAC_AWARE process flag is 20845916cd2Sjpk * to support the case of the automounter in the global zone. We 20945916cd2Sjpk * permit automounting of local zone directories such as home 21045916cd2Sjpk * directories, into the global zone as required by setlabel, 21145916cd2Sjpk * zonecopy, and saving of desktop sessions. Such mounts are 21245916cd2Sjpk * trusted not to expose the contents of one zone's directories 21345916cd2Sjpk * to another by leaking them through the global zone. 21445916cd2Sjpk */ 21545916cd2Sjpk if (is_system_labeled() && crgetzoneid(cr) == GLOBAL_ZONEID) { 21666aa579dSrica char specname[MAXPATHLEN]; 21766aa579dSrica zone_t *from_zptr; 21866aa579dSrica zone_t *to_zptr; 21966aa579dSrica 22066aa579dSrica if (vnodetopath(NULL, realrootvp, specname, 221*7f66f2f8Srica sizeof (specname), CRED()) != 0) { 222*7f66f2f8Srica VN_RELE(realrootvp); 22366aa579dSrica return (EACCES); 224*7f66f2f8Srica } 22545916cd2Sjpk 22666aa579dSrica from_zptr = zone_find_by_path(specname); 22745916cd2Sjpk to_zptr = zone_find_by_path(refstr_value(vfsp->vfs_mntpt)); 22845916cd2Sjpk 22945916cd2Sjpk /* 23045916cd2Sjpk * Special case for zone devfs: the zone for /dev will 23145916cd2Sjpk * incorrectly appear as the global zone since it's not 23245916cd2Sjpk * under the zone rootpath. So for zone devfs check allow 23345916cd2Sjpk * read-write mounts. 23448451833Scarlsonj * 23548451833Scarlsonj * Second special case for scratch zones used for Live Upgrade: 23648451833Scarlsonj * this is used to mount the zone's root from /root to /a in 23748451833Scarlsonj * the scratch zone. As with the other special case, this 23848451833Scarlsonj * appears to be outside of the zone because it's not under 23948451833Scarlsonj * the zone rootpath, which is $ZONEPATH/lu in the scratch 24048451833Scarlsonj * zone case. 24145916cd2Sjpk */ 24245916cd2Sjpk 24352782930Sszhou if (from_zptr != to_zptr && 24448451833Scarlsonj !(to_zptr->zone_flags & ZF_IS_SCRATCH)) { 24545916cd2Sjpk /* 24645916cd2Sjpk * We know at this point that the labels aren't equal 24745916cd2Sjpk * because the zone pointers aren't equal, and zones 24845916cd2Sjpk * can't share a label. 24945916cd2Sjpk * 25045916cd2Sjpk * If the source is the global zone then making 25145916cd2Sjpk * it available to a local zone must be done in 25245916cd2Sjpk * read-only mode as the label will become admin_low. 25345916cd2Sjpk * 25445916cd2Sjpk * If it is a mount between local zones then if 25545916cd2Sjpk * the current process is in the global zone and has 25645916cd2Sjpk * the NET_MAC_AWARE flag, then regular read-write 25745916cd2Sjpk * access is allowed. If it's in some other zone, but 25845916cd2Sjpk * the label on the mount point dominates the original 25945916cd2Sjpk * source, then allow the mount as read-only 26045916cd2Sjpk * ("read-down"). 26145916cd2Sjpk */ 26245916cd2Sjpk if (from_zptr->zone_id == GLOBAL_ZONEID) { 26345916cd2Sjpk /* make the mount read-only */ 26445916cd2Sjpk vfs_setmntopt(vfsp, MNTOPT_RO, NULL, 0); 26545916cd2Sjpk } else { /* cross-zone mount */ 26645916cd2Sjpk if (to_zptr->zone_id == GLOBAL_ZONEID && 26745916cd2Sjpk /* LINTED: no consequent */ 26845916cd2Sjpk getpflags(NET_MAC_AWARE, cr) != 0) { 26945916cd2Sjpk /* Allow the mount as read-write */ 27045916cd2Sjpk } else if (bldominates( 27145916cd2Sjpk label2bslabel(to_zptr->zone_slabel), 27245916cd2Sjpk label2bslabel(from_zptr->zone_slabel))) { 27345916cd2Sjpk /* make the mount read-only */ 27445916cd2Sjpk vfs_setmntopt(vfsp, MNTOPT_RO, NULL, 0); 27545916cd2Sjpk } else { 276*7f66f2f8Srica VN_RELE(realrootvp); 27745916cd2Sjpk zone_rele(to_zptr); 27845916cd2Sjpk zone_rele(from_zptr); 27945916cd2Sjpk return (EACCES); 28045916cd2Sjpk } 28145916cd2Sjpk } 28245916cd2Sjpk } 28345916cd2Sjpk zone_rele(to_zptr); 28445916cd2Sjpk zone_rele(from_zptr); 28545916cd2Sjpk } 28645916cd2Sjpk 2877c478bd9Sstevel@tonic-gate /* 2887c478bd9Sstevel@tonic-gate * realrootvp may be an AUTOFS node, in which case we 2897c478bd9Sstevel@tonic-gate * perform a VOP_ACCESS() to trigger the mount of the 2907c478bd9Sstevel@tonic-gate * intended filesystem, so we loopback mount the intended 2917c478bd9Sstevel@tonic-gate * filesystem instead of the AUTOFS filesystem. 2927c478bd9Sstevel@tonic-gate */ 293da6c28aaSamw (void) VOP_ACCESS(realrootvp, 0, 0, cr, NULL); 2947c478bd9Sstevel@tonic-gate 2957c478bd9Sstevel@tonic-gate /* 2967c478bd9Sstevel@tonic-gate * We're interested in the top most filesystem. 2977c478bd9Sstevel@tonic-gate * This is specially important when uap->spec is a trigger 2987c478bd9Sstevel@tonic-gate * AUTOFS node, since we're really interested in mounting the 2997c478bd9Sstevel@tonic-gate * filesystem AUTOFS mounted as result of the VOP_ACCESS() 3007c478bd9Sstevel@tonic-gate * call not the AUTOFS node itself. 3017c478bd9Sstevel@tonic-gate */ 3027c478bd9Sstevel@tonic-gate if (vn_mountedvfs(realrootvp) != NULL) { 3037c478bd9Sstevel@tonic-gate if (error = traverse(&realrootvp)) { 3047c478bd9Sstevel@tonic-gate VN_RELE(realrootvp); 3057c478bd9Sstevel@tonic-gate return (error); 3067c478bd9Sstevel@tonic-gate } 3077c478bd9Sstevel@tonic-gate } 3087c478bd9Sstevel@tonic-gate 3097c478bd9Sstevel@tonic-gate /* 3107c478bd9Sstevel@tonic-gate * Allocate a vfs info struct and attach it 3117c478bd9Sstevel@tonic-gate */ 3127c478bd9Sstevel@tonic-gate li = kmem_zalloc(sizeof (struct loinfo), KM_SLEEP); 3137c478bd9Sstevel@tonic-gate li->li_realvfs = realrootvp->v_vfsp; 3147c478bd9Sstevel@tonic-gate li->li_mountvfs = vfsp; 3157c478bd9Sstevel@tonic-gate 3167c478bd9Sstevel@tonic-gate /* 3177c478bd9Sstevel@tonic-gate * Set mount flags to be inherited by loopback vfs's 3187c478bd9Sstevel@tonic-gate */ 3197c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_RO, NULL)) { 3207c478bd9Sstevel@tonic-gate li->li_mflag |= VFS_RDONLY; 3217c478bd9Sstevel@tonic-gate } 3227c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_NOSUID, NULL)) { 3237c478bd9Sstevel@tonic-gate li->li_mflag |= (VFS_NOSETUID|VFS_NODEVICES); 3247c478bd9Sstevel@tonic-gate } 3257c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_NODEVICES, NULL)) { 3267c478bd9Sstevel@tonic-gate li->li_mflag |= VFS_NODEVICES; 3277c478bd9Sstevel@tonic-gate } 3287c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_NOSETUID, NULL)) { 3297c478bd9Sstevel@tonic-gate li->li_mflag |= VFS_NOSETUID; 3307c478bd9Sstevel@tonic-gate } 3317c478bd9Sstevel@tonic-gate /* 3327c478bd9Sstevel@tonic-gate * Permissive flags are added to the "deny" bitmap. 3337c478bd9Sstevel@tonic-gate */ 3347c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_NOXATTR, NULL)) { 3357c478bd9Sstevel@tonic-gate li->li_dflag |= VFS_XATTR; 3367c478bd9Sstevel@tonic-gate } 3377c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_NONBMAND, NULL)) { 3387c478bd9Sstevel@tonic-gate li->li_dflag |= VFS_NBMAND; 3397c478bd9Sstevel@tonic-gate } 3407c478bd9Sstevel@tonic-gate 3417c478bd9Sstevel@tonic-gate /* 3427c478bd9Sstevel@tonic-gate * Propagate inheritable mount flags from the real vfs. 3437c478bd9Sstevel@tonic-gate */ 3447c478bd9Sstevel@tonic-gate if ((li->li_realvfs->vfs_flag & VFS_RDONLY) && 3457c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_RO, NULL)) 3467c478bd9Sstevel@tonic-gate vfs_setmntopt(vfsp, MNTOPT_RO, NULL, 3477c478bd9Sstevel@tonic-gate VFS_NODISPLAY); 3487c478bd9Sstevel@tonic-gate if ((li->li_realvfs->vfs_flag & VFS_NOSETUID) && 3497c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_NOSETUID, NULL)) 3507c478bd9Sstevel@tonic-gate vfs_setmntopt(vfsp, MNTOPT_NOSETUID, NULL, 3517c478bd9Sstevel@tonic-gate VFS_NODISPLAY); 3527c478bd9Sstevel@tonic-gate if ((li->li_realvfs->vfs_flag & VFS_NODEVICES) && 3537c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_NODEVICES, NULL)) 3547c478bd9Sstevel@tonic-gate vfs_setmntopt(vfsp, MNTOPT_NODEVICES, NULL, 3557c478bd9Sstevel@tonic-gate VFS_NODISPLAY); 3567c478bd9Sstevel@tonic-gate /* 3577c478bd9Sstevel@tonic-gate * Permissive flags such as VFS_XATTR, as opposed to restrictive flags 3587c478bd9Sstevel@tonic-gate * such as VFS_RDONLY, are handled differently. An explicit 3597c478bd9Sstevel@tonic-gate * MNTOPT_NOXATTR should override the underlying filesystem's VFS_XATTR. 3607c478bd9Sstevel@tonic-gate */ 3617c478bd9Sstevel@tonic-gate if ((li->li_realvfs->vfs_flag & VFS_XATTR) && 3627c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_NOXATTR, NULL) && 3637c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_XATTR, NULL)) 3647c478bd9Sstevel@tonic-gate vfs_setmntopt(vfsp, MNTOPT_XATTR, NULL, 3657c478bd9Sstevel@tonic-gate VFS_NODISPLAY); 3667c478bd9Sstevel@tonic-gate if ((li->li_realvfs->vfs_flag & VFS_NBMAND) && 3677c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_NBMAND, NULL) && 3687c478bd9Sstevel@tonic-gate !vfs_optionisset(vfsp, MNTOPT_NONBMAND, NULL)) 3697c478bd9Sstevel@tonic-gate vfs_setmntopt(vfsp, MNTOPT_NBMAND, NULL, 3707c478bd9Sstevel@tonic-gate VFS_NODISPLAY); 3717c478bd9Sstevel@tonic-gate 3727c478bd9Sstevel@tonic-gate li->li_refct = 0; 3737c478bd9Sstevel@tonic-gate vfsp->vfs_data = (caddr_t)li; 3747c478bd9Sstevel@tonic-gate vfsp->vfs_bcount = 0; 3757c478bd9Sstevel@tonic-gate vfsp->vfs_fstype = lofsfstype; 3767c478bd9Sstevel@tonic-gate vfsp->vfs_bsize = li->li_realvfs->vfs_bsize; 3777c478bd9Sstevel@tonic-gate 37852782930Sszhou vfsp->vfs_dev = li->li_realvfs->vfs_dev; 37952782930Sszhou vfsp->vfs_fsid.val[0] = li->li_realvfs->vfs_fsid.val[0]; 38052782930Sszhou vfsp->vfs_fsid.val[1] = li->li_realvfs->vfs_fsid.val[1]; 3817c478bd9Sstevel@tonic-gate 3827c478bd9Sstevel@tonic-gate if (vfs_optionisset(vfsp, MNTOPT_LOFS_NOSUB, NULL)) { 3837c478bd9Sstevel@tonic-gate li->li_flag |= LO_NOSUB; 3847c478bd9Sstevel@tonic-gate } 3857c478bd9Sstevel@tonic-gate 38679a28c7aSmarks /* 38779a28c7aSmarks * Propagate any VFS features 38879a28c7aSmarks */ 38979a28c7aSmarks 39079a28c7aSmarks vfs_propagate_features(li->li_realvfs, vfsp); 39179a28c7aSmarks 3927c478bd9Sstevel@tonic-gate /* 3937c478bd9Sstevel@tonic-gate * Setup the hashtable. If the root of this mount isn't a directory, 3947c478bd9Sstevel@tonic-gate * there's no point in allocating a large hashtable. A table with one 3957c478bd9Sstevel@tonic-gate * bucket is sufficient. 3967c478bd9Sstevel@tonic-gate */ 3977c478bd9Sstevel@tonic-gate if (realrootvp->v_type != VDIR) 3987c478bd9Sstevel@tonic-gate lsetup(li, 1); 3997c478bd9Sstevel@tonic-gate else 4007c478bd9Sstevel@tonic-gate lsetup(li, 0); 4017c478bd9Sstevel@tonic-gate 4027c478bd9Sstevel@tonic-gate /* 4037c478bd9Sstevel@tonic-gate * Make the root vnode 4047c478bd9Sstevel@tonic-gate */ 405b431137cSowenr srootvp = makelonode(realrootvp, li, 0); 4067c478bd9Sstevel@tonic-gate srootvp->v_flag |= VROOT; 4077c478bd9Sstevel@tonic-gate li->li_rootvp = srootvp; 4087c478bd9Sstevel@tonic-gate 4097c478bd9Sstevel@tonic-gate #ifdef LODEBUG 4107c478bd9Sstevel@tonic-gate lo_dprint(4, "lo_mount: vfs %p realvfs %p root %p realroot %p li %p\n", 4117c478bd9Sstevel@tonic-gate vfsp, li->li_realvfs, srootvp, realrootvp, li); 4127c478bd9Sstevel@tonic-gate #endif 4137c478bd9Sstevel@tonic-gate return (0); 4147c478bd9Sstevel@tonic-gate } 4157c478bd9Sstevel@tonic-gate 4167c478bd9Sstevel@tonic-gate /* 4177c478bd9Sstevel@tonic-gate * Undo loopback mount 4187c478bd9Sstevel@tonic-gate */ 4197c478bd9Sstevel@tonic-gate static int 4207c478bd9Sstevel@tonic-gate lo_unmount(struct vfs *vfsp, int flag, struct cred *cr) 4217c478bd9Sstevel@tonic-gate { 4227c478bd9Sstevel@tonic-gate struct loinfo *li; 4237c478bd9Sstevel@tonic-gate 4247c478bd9Sstevel@tonic-gate if (secpolicy_fs_unmount(cr, vfsp) != 0) 4257c478bd9Sstevel@tonic-gate return (EPERM); 4267c478bd9Sstevel@tonic-gate 4277c478bd9Sstevel@tonic-gate /* 4287c478bd9Sstevel@tonic-gate * Forced unmount is not supported by this file system 4297c478bd9Sstevel@tonic-gate * and thus, ENOTSUP, is being returned. 4307c478bd9Sstevel@tonic-gate */ 4317c478bd9Sstevel@tonic-gate if (flag & MS_FORCE) 4327c478bd9Sstevel@tonic-gate return (ENOTSUP); 4337c478bd9Sstevel@tonic-gate 4347c478bd9Sstevel@tonic-gate li = vtoli(vfsp); 4357c478bd9Sstevel@tonic-gate #ifdef LODEBUG 4367c478bd9Sstevel@tonic-gate lo_dprint(4, "lo_unmount(%p) li %p\n", vfsp, li); 4377c478bd9Sstevel@tonic-gate #endif 4387c478bd9Sstevel@tonic-gate if (li->li_refct != 1 || li->li_rootvp->v_count != 1) { 4397c478bd9Sstevel@tonic-gate #ifdef LODEBUG 4407c478bd9Sstevel@tonic-gate lo_dprint(4, "refct %d v_ct %d\n", li->li_refct, 4417c478bd9Sstevel@tonic-gate li->li_rootvp->v_count); 4427c478bd9Sstevel@tonic-gate #endif 4437c478bd9Sstevel@tonic-gate return (EBUSY); 4447c478bd9Sstevel@tonic-gate } 4457c478bd9Sstevel@tonic-gate VN_RELE(li->li_rootvp); 4467c478bd9Sstevel@tonic-gate return (0); 4477c478bd9Sstevel@tonic-gate } 4487c478bd9Sstevel@tonic-gate 4497c478bd9Sstevel@tonic-gate /* 4507c478bd9Sstevel@tonic-gate * Find root of lofs mount. 4517c478bd9Sstevel@tonic-gate */ 4527c478bd9Sstevel@tonic-gate static int 4537c478bd9Sstevel@tonic-gate lo_root(struct vfs *vfsp, struct vnode **vpp) 4547c478bd9Sstevel@tonic-gate { 4557c478bd9Sstevel@tonic-gate *vpp = vtoli(vfsp)->li_rootvp; 4567c478bd9Sstevel@tonic-gate #ifdef LODEBUG 4577c478bd9Sstevel@tonic-gate lo_dprint(4, "lo_root(0x%p) = %p\n", vfsp, *vpp); 4587c478bd9Sstevel@tonic-gate #endif 4597c478bd9Sstevel@tonic-gate /* 4607c478bd9Sstevel@tonic-gate * If the root of the filesystem is a special file, return the specvp 4617c478bd9Sstevel@tonic-gate * version of the vnode. We don't save the specvp vnode in our 4627c478bd9Sstevel@tonic-gate * hashtable since that's exclusively for lnodes. 4637c478bd9Sstevel@tonic-gate */ 4647c478bd9Sstevel@tonic-gate if (IS_DEVVP(*vpp)) { 4657c478bd9Sstevel@tonic-gate struct vnode *svp; 4667c478bd9Sstevel@tonic-gate 4677c478bd9Sstevel@tonic-gate svp = specvp(*vpp, (*vpp)->v_rdev, (*vpp)->v_type, kcred); 4687c478bd9Sstevel@tonic-gate if (svp == NULL) 4697c478bd9Sstevel@tonic-gate return (ENOSYS); 4707c478bd9Sstevel@tonic-gate *vpp = svp; 4717c478bd9Sstevel@tonic-gate } else { 4727c478bd9Sstevel@tonic-gate VN_HOLD(*vpp); 4737c478bd9Sstevel@tonic-gate } 4747c478bd9Sstevel@tonic-gate 4757c478bd9Sstevel@tonic-gate return (0); 4767c478bd9Sstevel@tonic-gate } 4777c478bd9Sstevel@tonic-gate 4787c478bd9Sstevel@tonic-gate /* 4797c478bd9Sstevel@tonic-gate * Get file system statistics. 4807c478bd9Sstevel@tonic-gate */ 4817c478bd9Sstevel@tonic-gate static int 4827c478bd9Sstevel@tonic-gate lo_statvfs(register struct vfs *vfsp, struct statvfs64 *sbp) 4837c478bd9Sstevel@tonic-gate { 4847c478bd9Sstevel@tonic-gate vnode_t *realrootvp; 4857c478bd9Sstevel@tonic-gate 4867c478bd9Sstevel@tonic-gate #ifdef LODEBUG 4877c478bd9Sstevel@tonic-gate lo_dprint(4, "lostatvfs %p\n", vfsp); 4887c478bd9Sstevel@tonic-gate #endif 4897c478bd9Sstevel@tonic-gate /* 4907c478bd9Sstevel@tonic-gate * Using realrootvp->v_vfsp (instead of the realvfsp that was 4917c478bd9Sstevel@tonic-gate * cached) is necessary to make lofs work woth forced UFS unmounts. 4927c478bd9Sstevel@tonic-gate * In the case of a forced unmount, UFS stores a set of dummy vfsops 4937c478bd9Sstevel@tonic-gate * in all the (i)vnodes in the filesystem. The dummy ops simply 4947c478bd9Sstevel@tonic-gate * returns back EIO. 4957c478bd9Sstevel@tonic-gate */ 4967c478bd9Sstevel@tonic-gate (void) lo_realvfs(vfsp, &realrootvp); 4977c478bd9Sstevel@tonic-gate if (realrootvp != NULL) 4987c478bd9Sstevel@tonic-gate return (VFS_STATVFS(realrootvp->v_vfsp, sbp)); 4997c478bd9Sstevel@tonic-gate else 5007c478bd9Sstevel@tonic-gate return (EIO); 5017c478bd9Sstevel@tonic-gate } 5027c478bd9Sstevel@tonic-gate 5037c478bd9Sstevel@tonic-gate /* 5047c478bd9Sstevel@tonic-gate * LOFS doesn't have any data or metadata to flush, pending I/O on the 5057c478bd9Sstevel@tonic-gate * underlying filesystem will be flushed when such filesystem is synched. 5067c478bd9Sstevel@tonic-gate */ 5077c478bd9Sstevel@tonic-gate /* ARGSUSED */ 5087c478bd9Sstevel@tonic-gate static int 5097c478bd9Sstevel@tonic-gate lo_sync(struct vfs *vfsp, 5107c478bd9Sstevel@tonic-gate short flag, 5117c478bd9Sstevel@tonic-gate struct cred *cr) 5127c478bd9Sstevel@tonic-gate { 5137c478bd9Sstevel@tonic-gate #ifdef LODEBUG 5147c478bd9Sstevel@tonic-gate lo_dprint(4, "lo_sync: %p\n", vfsp); 5157c478bd9Sstevel@tonic-gate #endif 5167c478bd9Sstevel@tonic-gate return (0); 5177c478bd9Sstevel@tonic-gate } 5187c478bd9Sstevel@tonic-gate 5197c478bd9Sstevel@tonic-gate /* 5207c478bd9Sstevel@tonic-gate * Obtain the vnode from the underlying filesystem. 5217c478bd9Sstevel@tonic-gate */ 5227c478bd9Sstevel@tonic-gate static int 5237c478bd9Sstevel@tonic-gate lo_vget(struct vfs *vfsp, struct vnode **vpp, struct fid *fidp) 5247c478bd9Sstevel@tonic-gate { 5257c478bd9Sstevel@tonic-gate vnode_t *realrootvp; 5267c478bd9Sstevel@tonic-gate 5277c478bd9Sstevel@tonic-gate #ifdef LODEBUG 5287c478bd9Sstevel@tonic-gate lo_dprint(4, "lo_vget: %p\n", vfsp); 5297c478bd9Sstevel@tonic-gate #endif 5307c478bd9Sstevel@tonic-gate (void) lo_realvfs(vfsp, &realrootvp); 5317c478bd9Sstevel@tonic-gate if (realrootvp != NULL) 5327c478bd9Sstevel@tonic-gate return (VFS_VGET(realrootvp->v_vfsp, vpp, fidp)); 5337c478bd9Sstevel@tonic-gate else 5347c478bd9Sstevel@tonic-gate return (EIO); 5357c478bd9Sstevel@tonic-gate } 5367c478bd9Sstevel@tonic-gate 5377c478bd9Sstevel@tonic-gate /* 5387c478bd9Sstevel@tonic-gate * Free mount-specific data. 5397c478bd9Sstevel@tonic-gate */ 5407c478bd9Sstevel@tonic-gate static void 5417c478bd9Sstevel@tonic-gate lo_freevfs(struct vfs *vfsp) 5427c478bd9Sstevel@tonic-gate { 5437c478bd9Sstevel@tonic-gate struct loinfo *li = vtoli(vfsp); 5447c478bd9Sstevel@tonic-gate 5457c478bd9Sstevel@tonic-gate ldestroy(li); 5467c478bd9Sstevel@tonic-gate kmem_free(li, sizeof (struct loinfo)); 5477c478bd9Sstevel@tonic-gate } 5487c478bd9Sstevel@tonic-gate 5497c478bd9Sstevel@tonic-gate static int 5507c478bd9Sstevel@tonic-gate lofsinit(int fstyp, char *name) 5517c478bd9Sstevel@tonic-gate { 5527c478bd9Sstevel@tonic-gate static const fs_operation_def_t lo_vfsops_template[] = { 553aa59c4cbSrsb VFSNAME_MOUNT, { .vfs_mount = lo_mount }, 554aa59c4cbSrsb VFSNAME_UNMOUNT, { .vfs_unmount = lo_unmount }, 555aa59c4cbSrsb VFSNAME_ROOT, { .vfs_root = lo_root }, 556aa59c4cbSrsb VFSNAME_STATVFS, { .vfs_statvfs = lo_statvfs }, 557aa59c4cbSrsb VFSNAME_SYNC, { .vfs_sync = lo_sync }, 558aa59c4cbSrsb VFSNAME_VGET, { .vfs_vget = lo_vget }, 559aa59c4cbSrsb VFSNAME_FREEVFS, { .vfs_freevfs = lo_freevfs }, 560aa59c4cbSrsb NULL, NULL 5617c478bd9Sstevel@tonic-gate }; 5627c478bd9Sstevel@tonic-gate int error; 5637c478bd9Sstevel@tonic-gate 5647c478bd9Sstevel@tonic-gate error = vfs_setfsops(fstyp, lo_vfsops_template, &lo_vfsops); 5657c478bd9Sstevel@tonic-gate if (error != 0) { 5667c478bd9Sstevel@tonic-gate cmn_err(CE_WARN, "lofsinit: bad vfs ops template"); 5677c478bd9Sstevel@tonic-gate return (error); 5687c478bd9Sstevel@tonic-gate } 5697c478bd9Sstevel@tonic-gate 5707c478bd9Sstevel@tonic-gate error = vn_make_ops(name, lo_vnodeops_template, &lo_vnodeops); 5717c478bd9Sstevel@tonic-gate if (error != 0) { 5727c478bd9Sstevel@tonic-gate (void) vfs_freevfsops_by_type(fstyp); 5737c478bd9Sstevel@tonic-gate cmn_err(CE_WARN, "lofsinit: bad vnode ops template"); 5747c478bd9Sstevel@tonic-gate return (error); 5757c478bd9Sstevel@tonic-gate } 5767c478bd9Sstevel@tonic-gate 5777c478bd9Sstevel@tonic-gate lofsfstype = fstyp; 5787c478bd9Sstevel@tonic-gate 5797c478bd9Sstevel@tonic-gate return (0); 5807c478bd9Sstevel@tonic-gate } 581