| @@ -1,1435 +1,1448 @@ | | | @@ -1,1435 +1,1448 @@ |
1 | /* $NetBSD: vfs_mount.c,v 1.48 2017/02/22 09:50:13 hannken Exp $ */ | | 1 | /* $NetBSD: vfs_mount.c,v 1.49 2017/03/06 10:10:43 hannken Exp $ */ |
2 | | | 2 | |
3 | /*- | | 3 | /*- |
4 | * Copyright (c) 1997-2011 The NetBSD Foundation, Inc. | | 4 | * Copyright (c) 1997-2011 The NetBSD Foundation, Inc. |
5 | * All rights reserved. | | 5 | * All rights reserved. |
6 | * | | 6 | * |
7 | * This code is derived from software contributed to The NetBSD Foundation | | 7 | * This code is derived from software contributed to The NetBSD Foundation |
8 | * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, | | 8 | * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, |
9 | * NASA Ames Research Center, by Charles M. Hannum, and by Andrew Doran. | | 9 | * NASA Ames Research Center, by Charles M. Hannum, and by Andrew Doran. |
10 | * | | 10 | * |
11 | * Redistribution and use in source and binary forms, with or without | | 11 | * Redistribution and use in source and binary forms, with or without |
12 | * modification, are permitted provided that the following conditions | | 12 | * modification, are permitted provided that the following conditions |
13 | * are met: | | 13 | * are met: |
14 | * 1. Redistributions of source code must retain the above copyright | | 14 | * 1. Redistributions of source code must retain the above copyright |
15 | * notice, this list of conditions and the following disclaimer. | | 15 | * notice, this list of conditions and the following disclaimer. |
16 | * 2. Redistributions in binary form must reproduce the above copyright | | 16 | * 2. Redistributions in binary form must reproduce the above copyright |
17 | * notice, this list of conditions and the following disclaimer in the | | 17 | * notice, this list of conditions and the following disclaimer in the |
18 | * documentation and/or other materials provided with the distribution. | | 18 | * documentation and/or other materials provided with the distribution. |
19 | * | | 19 | * |
20 | * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS | | 20 | * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS |
21 | * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED | | 21 | * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED |
22 | * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR | | 22 | * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
23 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS | | 23 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS |
24 | * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR | | 24 | * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
25 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF | | 25 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
26 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS | | 26 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
27 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN | | 27 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
28 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) | | 28 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
29 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE | | 29 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
30 | * POSSIBILITY OF SUCH DAMAGE. | | 30 | * POSSIBILITY OF SUCH DAMAGE. |
31 | */ | | 31 | */ |
32 | | | 32 | |
33 | /* | | 33 | /* |
34 | * Copyright (c) 1989, 1993 | | 34 | * Copyright (c) 1989, 1993 |
35 | * The Regents of the University of California. All rights reserved. | | 35 | * The Regents of the University of California. All rights reserved. |
36 | * (c) UNIX System Laboratories, Inc. | | 36 | * (c) UNIX System Laboratories, Inc. |
37 | * All or some portions of this file are derived from material licensed | | 37 | * All or some portions of this file are derived from material licensed |
38 | * to the University of California by American Telephone and Telegraph | | 38 | * to the University of California by American Telephone and Telegraph |
39 | * Co. or Unix System Laboratories, Inc. and are reproduced herein with | | 39 | * Co. or Unix System Laboratories, Inc. and are reproduced herein with |
40 | * the permission of UNIX System Laboratories, Inc. | | 40 | * the permission of UNIX System Laboratories, Inc. |
41 | * | | 41 | * |
42 | * Redistribution and use in source and binary forms, with or without | | 42 | * Redistribution and use in source and binary forms, with or without |
43 | * modification, are permitted provided that the following conditions | | 43 | * modification, are permitted provided that the following conditions |
44 | * are met: | | 44 | * are met: |
45 | * 1. Redistributions of source code must retain the above copyright | | 45 | * 1. Redistributions of source code must retain the above copyright |
46 | * notice, this list of conditions and the following disclaimer. | | 46 | * notice, this list of conditions and the following disclaimer. |
47 | * 2. Redistributions in binary form must reproduce the above copyright | | 47 | * 2. Redistributions in binary form must reproduce the above copyright |
48 | * notice, this list of conditions and the following disclaimer in the | | 48 | * notice, this list of conditions and the following disclaimer in the |
49 | * documentation and/or other materials provided with the distribution. | | 49 | * documentation and/or other materials provided with the distribution. |
50 | * 3. Neither the name of the University nor the names of its contributors | | 50 | * 3. Neither the name of the University nor the names of its contributors |
51 | * may be used to endorse or promote products derived from this software | | 51 | * may be used to endorse or promote products derived from this software |
52 | * without specific prior written permission. | | 52 | * without specific prior written permission. |
53 | * | | 53 | * |
54 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND | | 54 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
55 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | | 55 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
56 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | | 56 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
57 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE | | 57 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
58 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | | 58 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
59 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS | | 59 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
60 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) | | 60 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
61 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT | | 61 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
62 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY | | 62 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
63 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF | | 63 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
64 | * SUCH DAMAGE. | | 64 | * SUCH DAMAGE. |
65 | * | | 65 | * |
66 | * @(#)vfs_subr.c 8.13 (Berkeley) 4/18/94 | | 66 | * @(#)vfs_subr.c 8.13 (Berkeley) 4/18/94 |
67 | */ | | 67 | */ |
68 | | | 68 | |
69 | #include <sys/cdefs.h> | | 69 | #include <sys/cdefs.h> |
70 | __KERNEL_RCSID(0, "$NetBSD: vfs_mount.c,v 1.48 2017/02/22 09:50:13 hannken Exp $"); | | 70 | __KERNEL_RCSID(0, "$NetBSD: vfs_mount.c,v 1.49 2017/03/06 10:10:43 hannken Exp $"); |
71 | | | 71 | |
72 | #include <sys/param.h> | | 72 | #include <sys/param.h> |
73 | #include <sys/kernel.h> | | 73 | #include <sys/kernel.h> |
74 | | | 74 | |
75 | #include <sys/atomic.h> | | 75 | #include <sys/atomic.h> |
76 | #include <sys/buf.h> | | 76 | #include <sys/buf.h> |
77 | #include <sys/conf.h> | | 77 | #include <sys/conf.h> |
78 | #include <sys/fcntl.h> | | 78 | #include <sys/fcntl.h> |
79 | #include <sys/filedesc.h> | | 79 | #include <sys/filedesc.h> |
80 | #include <sys/device.h> | | 80 | #include <sys/device.h> |
81 | #include <sys/kauth.h> | | 81 | #include <sys/kauth.h> |
82 | #include <sys/kmem.h> | | 82 | #include <sys/kmem.h> |
83 | #include <sys/module.h> | | 83 | #include <sys/module.h> |
84 | #include <sys/mount.h> | | 84 | #include <sys/mount.h> |
85 | #include <sys/fstrans.h> | | 85 | #include <sys/fstrans.h> |
86 | #include <sys/namei.h> | | 86 | #include <sys/namei.h> |
87 | #include <sys/extattr.h> | | 87 | #include <sys/extattr.h> |
88 | #include <sys/syscallargs.h> | | 88 | #include <sys/syscallargs.h> |
89 | #include <sys/sysctl.h> | | 89 | #include <sys/sysctl.h> |
90 | #include <sys/systm.h> | | 90 | #include <sys/systm.h> |
91 | #include <sys/vfs_syscalls.h> | | 91 | #include <sys/vfs_syscalls.h> |
92 | #include <sys/vnode_impl.h> | | 92 | #include <sys/vnode_impl.h> |
93 | | | 93 | |
94 | #include <miscfs/genfs/genfs.h> | | 94 | #include <miscfs/genfs/genfs.h> |
95 | #include <miscfs/specfs/specdev.h> | | 95 | #include <miscfs/specfs/specdev.h> |
96 | | | 96 | |
97 | static struct vnode *vfs_vnode_iterator_next1(struct vnode_iterator *, | | 97 | static struct vnode *vfs_vnode_iterator_next1(struct vnode_iterator *, |
98 | bool (*)(void *, struct vnode *), void *, bool); | | 98 | bool (*)(void *, struct vnode *), void *, bool); |
99 | | | 99 | |
100 | /* Root filesystem. */ | | 100 | /* Root filesystem. */ |
101 | vnode_t * rootvnode; | | 101 | vnode_t * rootvnode; |
102 | | | 102 | |
103 | /* Mounted filesystem list. */ | | 103 | /* Mounted filesystem list. */ |
104 | struct mntlist mountlist; | | 104 | struct mntlist mountlist; |
105 | kmutex_t mountlist_lock; | | 105 | kmutex_t mountlist_lock; |
106 | int vnode_offset_next_by_mount /* XXX: ugly hack for pstat.c */ | | 106 | int vnode_offset_next_by_mount /* XXX: ugly hack for pstat.c */ |
107 | = offsetof(vnode_impl_t, vi_mntvnodes.tqe_next); | | 107 | = offsetof(vnode_impl_t, vi_mntvnodes.tqe_next); |
108 | | | 108 | |
109 | kmutex_t mntvnode_lock; | | 109 | kmutex_t mntvnode_lock; |
110 | kmutex_t vfs_list_lock; | | 110 | kmutex_t vfs_list_lock; |
111 | | | 111 | |
112 | static specificdata_domain_t mount_specificdata_domain; | | 112 | static specificdata_domain_t mount_specificdata_domain; |
113 | static kmutex_t mntid_lock; | | 113 | static kmutex_t mntid_lock; |
114 | | | 114 | |
115 | static kmutex_t mountgen_lock; | | 115 | static kmutex_t mountgen_lock; |
116 | static uint64_t mountgen; | | 116 | static uint64_t mountgen; |
117 | | | 117 | |
118 | void | | 118 | void |
119 | vfs_mount_sysinit(void) | | 119 | vfs_mount_sysinit(void) |
120 | { | | 120 | { |
121 | | | 121 | |
122 | TAILQ_INIT(&mountlist); | | 122 | TAILQ_INIT(&mountlist); |
123 | mutex_init(&mountlist_lock, MUTEX_DEFAULT, IPL_NONE); | | 123 | mutex_init(&mountlist_lock, MUTEX_DEFAULT, IPL_NONE); |
124 | mutex_init(&mntvnode_lock, MUTEX_DEFAULT, IPL_NONE); | | 124 | mutex_init(&mntvnode_lock, MUTEX_DEFAULT, IPL_NONE); |
125 | mutex_init(&vfs_list_lock, MUTEX_DEFAULT, IPL_NONE); | | 125 | mutex_init(&vfs_list_lock, MUTEX_DEFAULT, IPL_NONE); |
126 | | | 126 | |
127 | mount_specificdata_domain = specificdata_domain_create(); | | 127 | mount_specificdata_domain = specificdata_domain_create(); |
128 | mutex_init(&mntid_lock, MUTEX_DEFAULT, IPL_NONE); | | 128 | mutex_init(&mntid_lock, MUTEX_DEFAULT, IPL_NONE); |
129 | mutex_init(&mountgen_lock, MUTEX_DEFAULT, IPL_NONE); | | 129 | mutex_init(&mountgen_lock, MUTEX_DEFAULT, IPL_NONE); |
130 | mountgen = 0; | | 130 | mountgen = 0; |
131 | } | | 131 | } |
132 | | | 132 | |
133 | struct mount * | | 133 | struct mount * |
134 | vfs_mountalloc(struct vfsops *vfsops, vnode_t *vp) | | 134 | vfs_mountalloc(struct vfsops *vfsops, vnode_t *vp) |
135 | { | | 135 | { |
136 | struct mount *mp; | | 136 | struct mount *mp; |
137 | int error __diagused; | | 137 | int error __diagused; |
138 | | | 138 | |
139 | mp = kmem_zalloc(sizeof(*mp), KM_SLEEP); | | 139 | mp = kmem_zalloc(sizeof(*mp), KM_SLEEP); |
140 | if (mp == NULL) | | 140 | if (mp == NULL) |
141 | return NULL; | | 141 | return NULL; |
142 | | | 142 | |
143 | mp->mnt_op = vfsops; | | 143 | mp->mnt_op = vfsops; |
144 | mp->mnt_refcnt = 1; | | 144 | mp->mnt_refcnt = 1; |
145 | TAILQ_INIT(&mp->mnt_vnodelist); | | 145 | TAILQ_INIT(&mp->mnt_vnodelist); |
146 | mutex_init(&mp->mnt_unmounting, MUTEX_DEFAULT, IPL_NONE); | | 146 | mutex_init(&mp->mnt_unmounting, MUTEX_DEFAULT, IPL_NONE); |
147 | mutex_init(&mp->mnt_renamelock, MUTEX_DEFAULT, IPL_NONE); | | 147 | mutex_init(&mp->mnt_renamelock, MUTEX_DEFAULT, IPL_NONE); |
148 | mutex_init(&mp->mnt_updating, MUTEX_DEFAULT, IPL_NONE); | | 148 | mutex_init(&mp->mnt_updating, MUTEX_DEFAULT, IPL_NONE); |
149 | error = vfs_busy(mp, NULL); | | 149 | error = vfs_busy(mp, NULL); |
150 | KASSERT(error == 0); | | 150 | KASSERT(error == 0); |
151 | mp->mnt_vnodecovered = vp; | | 151 | mp->mnt_vnodecovered = vp; |
152 | mount_initspecific(mp); | | 152 | mount_initspecific(mp); |
153 | | | 153 | |
154 | mutex_enter(&mountgen_lock); | | 154 | mutex_enter(&mountgen_lock); |
155 | mp->mnt_gen = mountgen++; | | 155 | mp->mnt_gen = mountgen++; |
156 | mutex_exit(&mountgen_lock); | | 156 | mutex_exit(&mountgen_lock); |
157 | | | 157 | |
158 | return mp; | | 158 | return mp; |
159 | } | | 159 | } |
160 | | | 160 | |
161 | /* | | 161 | /* |
162 | * vfs_rootmountalloc: lookup a filesystem type, and if found allocate and | | 162 | * vfs_rootmountalloc: lookup a filesystem type, and if found allocate and |
163 | * initialize a mount structure for it. | | 163 | * initialize a mount structure for it. |
164 | * | | 164 | * |
165 | * Devname is usually updated by mount(8) after booting. | | 165 | * Devname is usually updated by mount(8) after booting. |
166 | */ | | 166 | */ |
167 | int | | 167 | int |
168 | vfs_rootmountalloc(const char *fstypename, const char *devname, | | 168 | vfs_rootmountalloc(const char *fstypename, const char *devname, |
169 | struct mount **mpp) | | 169 | struct mount **mpp) |
170 | { | | 170 | { |
171 | struct vfsops *vfsp = NULL; | | 171 | struct vfsops *vfsp = NULL; |
172 | struct mount *mp; | | 172 | struct mount *mp; |
173 | | | 173 | |
174 | mutex_enter(&vfs_list_lock); | | 174 | mutex_enter(&vfs_list_lock); |
175 | LIST_FOREACH(vfsp, &vfs_list, vfs_list) | | 175 | LIST_FOREACH(vfsp, &vfs_list, vfs_list) |
176 | if (!strncmp(vfsp->vfs_name, fstypename, | | 176 | if (!strncmp(vfsp->vfs_name, fstypename, |
177 | sizeof(mp->mnt_stat.f_fstypename))) | | 177 | sizeof(mp->mnt_stat.f_fstypename))) |
178 | break; | | 178 | break; |
179 | if (vfsp == NULL) { | | 179 | if (vfsp == NULL) { |
180 | mutex_exit(&vfs_list_lock); | | 180 | mutex_exit(&vfs_list_lock); |
181 | return (ENODEV); | | 181 | return (ENODEV); |
182 | } | | 182 | } |
183 | vfsp->vfs_refcount++; | | 183 | vfsp->vfs_refcount++; |
184 | mutex_exit(&vfs_list_lock); | | 184 | mutex_exit(&vfs_list_lock); |
185 | | | 185 | |
186 | if ((mp = vfs_mountalloc(vfsp, NULL)) == NULL) | | 186 | if ((mp = vfs_mountalloc(vfsp, NULL)) == NULL) |
187 | return ENOMEM; | | 187 | return ENOMEM; |
188 | mp->mnt_flag = MNT_RDONLY; | | 188 | mp->mnt_flag = MNT_RDONLY; |
189 | (void)strlcpy(mp->mnt_stat.f_fstypename, vfsp->vfs_name, | | 189 | (void)strlcpy(mp->mnt_stat.f_fstypename, vfsp->vfs_name, |
190 | sizeof(mp->mnt_stat.f_fstypename)); | | 190 | sizeof(mp->mnt_stat.f_fstypename)); |
191 | mp->mnt_stat.f_mntonname[0] = '/'; | | 191 | mp->mnt_stat.f_mntonname[0] = '/'; |
192 | mp->mnt_stat.f_mntonname[1] = '\0'; | | 192 | mp->mnt_stat.f_mntonname[1] = '\0'; |
193 | mp->mnt_stat.f_mntfromname[sizeof(mp->mnt_stat.f_mntfromname) - 1] = | | 193 | mp->mnt_stat.f_mntfromname[sizeof(mp->mnt_stat.f_mntfromname) - 1] = |
194 | '\0'; | | 194 | '\0'; |
195 | (void)copystr(devname, mp->mnt_stat.f_mntfromname, | | 195 | (void)copystr(devname, mp->mnt_stat.f_mntfromname, |
196 | sizeof(mp->mnt_stat.f_mntfromname) - 1, 0); | | 196 | sizeof(mp->mnt_stat.f_mntfromname) - 1, 0); |
197 | *mpp = mp; | | 197 | *mpp = mp; |
198 | return 0; | | 198 | return 0; |
199 | } | | 199 | } |
200 | | | 200 | |
201 | /* | | 201 | /* |
202 | * vfs_getnewfsid: get a new unique fsid. | | 202 | * vfs_getnewfsid: get a new unique fsid. |
203 | */ | | 203 | */ |
204 | void | | 204 | void |
205 | vfs_getnewfsid(struct mount *mp) | | 205 | vfs_getnewfsid(struct mount *mp) |
206 | { | | 206 | { |
207 | static u_short xxxfs_mntid; | | 207 | static u_short xxxfs_mntid; |
208 | fsid_t tfsid; | | 208 | fsid_t tfsid; |
209 | int mtype; | | 209 | int mtype; |
210 | | | 210 | |
211 | mutex_enter(&mntid_lock); | | 211 | mutex_enter(&mntid_lock); |
212 | mtype = makefstype(mp->mnt_op->vfs_name); | | 212 | mtype = makefstype(mp->mnt_op->vfs_name); |
213 | mp->mnt_stat.f_fsidx.__fsid_val[0] = makedev(mtype, 0); | | 213 | mp->mnt_stat.f_fsidx.__fsid_val[0] = makedev(mtype, 0); |
214 | mp->mnt_stat.f_fsidx.__fsid_val[1] = mtype; | | 214 | mp->mnt_stat.f_fsidx.__fsid_val[1] = mtype; |
215 | mp->mnt_stat.f_fsid = mp->mnt_stat.f_fsidx.__fsid_val[0]; | | 215 | mp->mnt_stat.f_fsid = mp->mnt_stat.f_fsidx.__fsid_val[0]; |
216 | if (xxxfs_mntid == 0) | | 216 | if (xxxfs_mntid == 0) |
217 | ++xxxfs_mntid; | | 217 | ++xxxfs_mntid; |
218 | tfsid.__fsid_val[0] = makedev(mtype & 0xff, xxxfs_mntid); | | 218 | tfsid.__fsid_val[0] = makedev(mtype & 0xff, xxxfs_mntid); |
219 | tfsid.__fsid_val[1] = mtype; | | 219 | tfsid.__fsid_val[1] = mtype; |
220 | if (!TAILQ_EMPTY(&mountlist)) { | | 220 | if (!TAILQ_EMPTY(&mountlist)) { |
221 | while (vfs_getvfs(&tfsid)) { | | 221 | while (vfs_getvfs(&tfsid)) { |
222 | tfsid.__fsid_val[0]++; | | 222 | tfsid.__fsid_val[0]++; |
223 | xxxfs_mntid++; | | 223 | xxxfs_mntid++; |
224 | } | | 224 | } |
225 | } | | 225 | } |
226 | mp->mnt_stat.f_fsidx.__fsid_val[0] = tfsid.__fsid_val[0]; | | 226 | mp->mnt_stat.f_fsidx.__fsid_val[0] = tfsid.__fsid_val[0]; |
227 | mp->mnt_stat.f_fsid = mp->mnt_stat.f_fsidx.__fsid_val[0]; | | 227 | mp->mnt_stat.f_fsid = mp->mnt_stat.f_fsidx.__fsid_val[0]; |
228 | mutex_exit(&mntid_lock); | | 228 | mutex_exit(&mntid_lock); |
229 | } | | 229 | } |
230 | | | 230 | |
231 | /* | | 231 | /* |
232 | * Lookup a mount point by filesystem identifier. | | 232 | * Lookup a mount point by filesystem identifier. |
233 | * | | 233 | * |
234 | * XXX Needs to add a reference to the mount point. | | 234 | * XXX Needs to add a reference to the mount point. |
235 | */ | | 235 | */ |
236 | struct mount * | | 236 | struct mount * |
237 | vfs_getvfs(fsid_t *fsid) | | 237 | vfs_getvfs(fsid_t *fsid) |
238 | { | | 238 | { |
239 | struct mount *mp; | | 239 | struct mount *mp; |
240 | | | 240 | |
241 | mutex_enter(&mountlist_lock); | | 241 | mutex_enter(&mountlist_lock); |
242 | TAILQ_FOREACH(mp, &mountlist, mnt_list) { | | 242 | TAILQ_FOREACH(mp, &mountlist, mnt_list) { |
243 | if (mp->mnt_stat.f_fsidx.__fsid_val[0] == fsid->__fsid_val[0] && | | 243 | if (mp->mnt_stat.f_fsidx.__fsid_val[0] == fsid->__fsid_val[0] && |
244 | mp->mnt_stat.f_fsidx.__fsid_val[1] == fsid->__fsid_val[1]) { | | 244 | mp->mnt_stat.f_fsidx.__fsid_val[1] == fsid->__fsid_val[1]) { |
245 | mutex_exit(&mountlist_lock); | | 245 | mutex_exit(&mountlist_lock); |
246 | return (mp); | | 246 | return (mp); |
247 | } | | 247 | } |
248 | } | | 248 | } |
249 | mutex_exit(&mountlist_lock); | | 249 | mutex_exit(&mountlist_lock); |
250 | return NULL; | | 250 | return NULL; |
251 | } | | 251 | } |
252 | | | 252 | |
253 | /* | | 253 | /* |
254 | * Drop a reference to a mount structure, freeing if the last reference. | | 254 | * Drop a reference to a mount structure, freeing if the last reference. |
255 | */ | | 255 | */ |
256 | void | | 256 | void |
257 | vfs_destroy(struct mount *mp) | | 257 | vfs_destroy(struct mount *mp) |
258 | { | | 258 | { |
259 | | | 259 | |
260 | if (__predict_true((int)atomic_dec_uint_nv(&mp->mnt_refcnt) > 0)) { | | 260 | if (__predict_true((int)atomic_dec_uint_nv(&mp->mnt_refcnt) > 0)) { |
261 | return; | | 261 | return; |
262 | } | | 262 | } |
263 | | | 263 | |
264 | /* | | 264 | /* |
265 | * Nothing else has visibility of the mount: we can now | | 265 | * Nothing else has visibility of the mount: we can now |
266 | * free the data structures. | | 266 | * free the data structures. |
267 | */ | | 267 | */ |
268 | KASSERT(mp->mnt_refcnt == 0); | | 268 | KASSERT(mp->mnt_refcnt == 0); |
269 | specificdata_fini(mount_specificdata_domain, &mp->mnt_specdataref); | | 269 | specificdata_fini(mount_specificdata_domain, &mp->mnt_specdataref); |
270 | mutex_destroy(&mp->mnt_unmounting); | | 270 | mutex_destroy(&mp->mnt_unmounting); |
271 | mutex_destroy(&mp->mnt_updating); | | 271 | mutex_destroy(&mp->mnt_updating); |
272 | mutex_destroy(&mp->mnt_renamelock); | | 272 | mutex_destroy(&mp->mnt_renamelock); |
273 | if (mp->mnt_op != NULL) { | | 273 | if (mp->mnt_op != NULL) { |
274 | vfs_delref(mp->mnt_op); | | 274 | vfs_delref(mp->mnt_op); |
275 | } | | 275 | } |
276 | kmem_free(mp, sizeof(*mp)); | | 276 | kmem_free(mp, sizeof(*mp)); |
277 | } | | 277 | } |
278 | | | 278 | |
279 | /* | | 279 | /* |
280 | * Mark a mount point as busy, and gain a new reference to it. Used to | | 280 | * Mark a mount point as busy, and gain a new reference to it. Used to |
281 | * prevent the file system from being unmounted during critical sections. | | 281 | * prevent the file system from being unmounted during critical sections. |
282 | * | | 282 | * |
283 | * vfs_busy can be called multiple times and by multiple threads | | 283 | * vfs_busy can be called multiple times and by multiple threads |
284 | * and must be accompanied by the same number of vfs_unbusy calls. | | 284 | * and must be accompanied by the same number of vfs_unbusy calls. |
285 | * | | 285 | * |
286 | * => The caller must hold a pre-existing reference to the mount. | | 286 | * => The caller must hold a pre-existing reference to the mount. |
287 | * => Will fail if the file system is being unmounted, or is unmounted. | | 287 | * => Will fail if the file system is being unmounted, or is unmounted. |
288 | */ | | 288 | */ |
289 | int | | 289 | int |
290 | vfs_busy(struct mount *mp, struct mount **nextp) | | 290 | vfs_busy(struct mount *mp, struct mount **nextp) |
291 | { | | 291 | { |
292 | | | 292 | |
293 | KASSERT(mp->mnt_refcnt > 0); | | 293 | KASSERT(mp->mnt_refcnt > 0); |
294 | | | 294 | |
295 | mutex_enter(&mp->mnt_unmounting); | | 295 | mutex_enter(&mp->mnt_unmounting); |
296 | if (__predict_false((mp->mnt_iflag & IMNT_GONE) != 0)) { | | 296 | if (__predict_false((mp->mnt_iflag & IMNT_GONE) != 0)) { |
297 | mutex_exit(&mp->mnt_unmounting); | | 297 | mutex_exit(&mp->mnt_unmounting); |
298 | if (nextp != NULL) { | | 298 | if (nextp != NULL) { |
299 | KASSERT(mutex_owned(&mountlist_lock)); | | 299 | KASSERT(mutex_owned(&mountlist_lock)); |
300 | *nextp = TAILQ_NEXT(mp, mnt_list); | | 300 | *nextp = TAILQ_NEXT(mp, mnt_list); |
301 | } | | 301 | } |
302 | return ENOENT; | | 302 | return ENOENT; |
303 | } | | 303 | } |
304 | ++mp->mnt_busynest; | | 304 | ++mp->mnt_busynest; |
305 | KASSERT(mp->mnt_busynest != 0); | | 305 | KASSERT(mp->mnt_busynest != 0); |
306 | mutex_exit(&mp->mnt_unmounting); | | 306 | mutex_exit(&mp->mnt_unmounting); |
307 | if (nextp != NULL) { | | 307 | if (nextp != NULL) { |
308 | mutex_exit(&mountlist_lock); | | 308 | mutex_exit(&mountlist_lock); |
309 | } | | 309 | } |
310 | atomic_inc_uint(&mp->mnt_refcnt); | | 310 | atomic_inc_uint(&mp->mnt_refcnt); |
311 | return 0; | | 311 | return 0; |
312 | } | | 312 | } |
313 | | | 313 | |
314 | /* | | 314 | /* |
315 | * Unbusy a busy filesystem. | | 315 | * Unbusy a busy filesystem. |
316 | * | | 316 | * |
317 | * Every successful vfs_busy() call must be undone by a vfs_unbusy() call. | | 317 | * Every successful vfs_busy() call must be undone by a vfs_unbusy() call. |
318 | * | | 318 | * |
319 | * => If keepref is true, preserve reference added by vfs_busy(). | | 319 | * => If keepref is true, preserve reference added by vfs_busy(). |
320 | * => If nextp != NULL, acquire mountlist_lock. | | 320 | * => If nextp != NULL, acquire mountlist_lock. |
321 | */ | | 321 | */ |
322 | void | | 322 | void |
323 | vfs_unbusy(struct mount *mp, bool keepref, struct mount **nextp) | | 323 | vfs_unbusy(struct mount *mp, bool keepref, struct mount **nextp) |
324 | { | | 324 | { |
325 | | | 325 | |
326 | KASSERT(mp->mnt_refcnt > 0); | | 326 | KASSERT(mp->mnt_refcnt > 0); |
327 | | | 327 | |
328 | if (nextp != NULL) { | | 328 | if (nextp != NULL) { |
329 | mutex_enter(&mountlist_lock); | | 329 | mutex_enter(&mountlist_lock); |
330 | } | | 330 | } |
331 | mutex_enter(&mp->mnt_unmounting); | | 331 | mutex_enter(&mp->mnt_unmounting); |
332 | KASSERT(mp->mnt_busynest != 0); | | 332 | KASSERT(mp->mnt_busynest != 0); |
333 | mp->mnt_busynest--; | | 333 | mp->mnt_busynest--; |
334 | mutex_exit(&mp->mnt_unmounting); | | 334 | mutex_exit(&mp->mnt_unmounting); |
335 | if (!keepref) { | | 335 | if (!keepref) { |
336 | vfs_destroy(mp); | | 336 | vfs_destroy(mp); |
337 | } | | 337 | } |
338 | if (nextp != NULL) { | | 338 | if (nextp != NULL) { |
339 | KASSERT(mutex_owned(&mountlist_lock)); | | 339 | KASSERT(mutex_owned(&mountlist_lock)); |
340 | *nextp = TAILQ_NEXT(mp, mnt_list); | | 340 | *nextp = TAILQ_NEXT(mp, mnt_list); |
341 | } | | 341 | } |
342 | } | | 342 | } |
343 | | | 343 | |
344 | struct vnode_iterator { | | 344 | struct vnode_iterator { |
345 | vnode_impl_t vi_vnode; | | 345 | vnode_impl_t vi_vnode; |
346 | }; | | 346 | }; |
347 | | | 347 | |
348 | void | | 348 | void |
349 | vfs_vnode_iterator_init(struct mount *mp, struct vnode_iterator **vnip) | | 349 | vfs_vnode_iterator_init(struct mount *mp, struct vnode_iterator **vnip) |
350 | { | | 350 | { |
351 | vnode_t *vp; | | 351 | vnode_t *vp; |
352 | vnode_impl_t *vip; | | 352 | vnode_impl_t *vip; |
353 | | | 353 | |
354 | vp = vnalloc_marker(mp); | | 354 | vp = vnalloc_marker(mp); |
355 | vip = VNODE_TO_VIMPL(vp); | | 355 | vip = VNODE_TO_VIMPL(vp); |
356 | | | 356 | |
357 | mutex_enter(&mntvnode_lock); | | 357 | mutex_enter(&mntvnode_lock); |
358 | TAILQ_INSERT_HEAD(&mp->mnt_vnodelist, vip, vi_mntvnodes); | | 358 | TAILQ_INSERT_HEAD(&mp->mnt_vnodelist, vip, vi_mntvnodes); |
359 | vp->v_usecount = 1; | | 359 | vp->v_usecount = 1; |
360 | mutex_exit(&mntvnode_lock); | | 360 | mutex_exit(&mntvnode_lock); |
361 | | | 361 | |
362 | *vnip = (struct vnode_iterator *)vip; | | 362 | *vnip = (struct vnode_iterator *)vip; |
363 | } | | 363 | } |
364 | | | 364 | |
365 | void | | 365 | void |
366 | vfs_vnode_iterator_destroy(struct vnode_iterator *vni) | | 366 | vfs_vnode_iterator_destroy(struct vnode_iterator *vni) |
367 | { | | 367 | { |
368 | vnode_impl_t *mvip = &vni->vi_vnode; | | 368 | vnode_impl_t *mvip = &vni->vi_vnode; |
369 | vnode_t *mvp = VIMPL_TO_VNODE(mvip); | | 369 | vnode_t *mvp = VIMPL_TO_VNODE(mvip); |
370 | | | 370 | |
371 | mutex_enter(&mntvnode_lock); | | 371 | mutex_enter(&mntvnode_lock); |
372 | KASSERT(vnis_marker(mvp)); | | 372 | KASSERT(vnis_marker(mvp)); |
373 | if (mvp->v_usecount != 0) { | | 373 | if (mvp->v_usecount != 0) { |
374 | TAILQ_REMOVE(&mvp->v_mount->mnt_vnodelist, mvip, vi_mntvnodes); | | 374 | TAILQ_REMOVE(&mvp->v_mount->mnt_vnodelist, mvip, vi_mntvnodes); |
375 | mvp->v_usecount = 0; | | 375 | mvp->v_usecount = 0; |
376 | } | | 376 | } |
377 | mutex_exit(&mntvnode_lock); | | 377 | mutex_exit(&mntvnode_lock); |
378 | vnfree_marker(mvp); | | 378 | vnfree_marker(mvp); |
379 | } | | 379 | } |
380 | | | 380 | |
381 | static struct vnode * | | 381 | static struct vnode * |
382 | vfs_vnode_iterator_next1(struct vnode_iterator *vni, | | 382 | vfs_vnode_iterator_next1(struct vnode_iterator *vni, |
383 | bool (*f)(void *, struct vnode *), void *cl, bool do_wait) | | 383 | bool (*f)(void *, struct vnode *), void *cl, bool do_wait) |
384 | { | | 384 | { |
385 | vnode_impl_t *mvip = &vni->vi_vnode; | | 385 | vnode_impl_t *mvip = &vni->vi_vnode; |
386 | struct mount *mp = VIMPL_TO_VNODE(mvip)->v_mount; | | 386 | struct mount *mp = VIMPL_TO_VNODE(mvip)->v_mount; |
387 | vnode_t *vp; | | 387 | vnode_t *vp; |
388 | vnode_impl_t *vip; | | 388 | vnode_impl_t *vip; |
389 | int error; | | 389 | int error; |
390 | | | 390 | |
391 | KASSERT(vnis_marker(VIMPL_TO_VNODE(mvip))); | | 391 | KASSERT(vnis_marker(VIMPL_TO_VNODE(mvip))); |
392 | | | 392 | |
393 | do { | | 393 | do { |
394 | mutex_enter(&mntvnode_lock); | | 394 | mutex_enter(&mntvnode_lock); |
395 | vip = TAILQ_NEXT(mvip, vi_mntvnodes); | | 395 | vip = TAILQ_NEXT(mvip, vi_mntvnodes); |
396 | TAILQ_REMOVE(&mp->mnt_vnodelist, mvip, vi_mntvnodes); | | 396 | TAILQ_REMOVE(&mp->mnt_vnodelist, mvip, vi_mntvnodes); |
397 | VIMPL_TO_VNODE(mvip)->v_usecount = 0; | | 397 | VIMPL_TO_VNODE(mvip)->v_usecount = 0; |
398 | again: | | 398 | again: |
399 | vp = VIMPL_TO_VNODE(vip); | | 399 | vp = VIMPL_TO_VNODE(vip); |
400 | if (vp == NULL) { | | 400 | if (vp == NULL) { |
401 | mutex_exit(&mntvnode_lock); | | 401 | mutex_exit(&mntvnode_lock); |
402 | return NULL; | | 402 | return NULL; |
403 | } | | 403 | } |
404 | mutex_enter(vp->v_interlock); | | 404 | mutex_enter(vp->v_interlock); |
405 | if (vnis_marker(vp) || | | 405 | if (vnis_marker(vp) || |
406 | vdead_check(vp, (do_wait ? 0 : VDEAD_NOWAIT)) || | | 406 | vdead_check(vp, (do_wait ? 0 : VDEAD_NOWAIT)) || |
407 | (f && !(*f)(cl, vp))) { | | 407 | (f && !(*f)(cl, vp))) { |
408 | mutex_exit(vp->v_interlock); | | 408 | mutex_exit(vp->v_interlock); |
409 | vip = TAILQ_NEXT(vip, vi_mntvnodes); | | 409 | vip = TAILQ_NEXT(vip, vi_mntvnodes); |
410 | goto again; | | 410 | goto again; |
411 | } | | 411 | } |
412 | | | 412 | |
413 | TAILQ_INSERT_AFTER(&mp->mnt_vnodelist, vip, mvip, vi_mntvnodes); | | 413 | TAILQ_INSERT_AFTER(&mp->mnt_vnodelist, vip, mvip, vi_mntvnodes); |
414 | VIMPL_TO_VNODE(mvip)->v_usecount = 1; | | 414 | VIMPL_TO_VNODE(mvip)->v_usecount = 1; |
415 | mutex_exit(&mntvnode_lock); | | 415 | mutex_exit(&mntvnode_lock); |
416 | error = vcache_vget(vp); | | 416 | error = vcache_vget(vp); |
417 | KASSERT(error == 0 || error == ENOENT); | | 417 | KASSERT(error == 0 || error == ENOENT); |
418 | } while (error != 0); | | 418 | } while (error != 0); |
419 | | | 419 | |
420 | return vp; | | 420 | return vp; |
421 | } | | 421 | } |
422 | | | 422 | |
423 | struct vnode * | | 423 | struct vnode * |
424 | vfs_vnode_iterator_next(struct vnode_iterator *vni, | | 424 | vfs_vnode_iterator_next(struct vnode_iterator *vni, |
425 | bool (*f)(void *, struct vnode *), void *cl) | | 425 | bool (*f)(void *, struct vnode *), void *cl) |
426 | { | | 426 | { |
427 | | | 427 | |
428 | return vfs_vnode_iterator_next1(vni, f, cl, false); | | 428 | return vfs_vnode_iterator_next1(vni, f, cl, false); |
429 | } | | 429 | } |
430 | | | 430 | |
431 | /* | | 431 | /* |
432 | * Move a vnode from one mount queue to another. | | 432 | * Move a vnode from one mount queue to another. |
433 | */ | | 433 | */ |
434 | void | | 434 | void |
435 | vfs_insmntque(vnode_t *vp, struct mount *mp) | | 435 | vfs_insmntque(vnode_t *vp, struct mount *mp) |
436 | { | | 436 | { |
437 | vnode_impl_t *vip = VNODE_TO_VIMPL(vp); | | 437 | vnode_impl_t *vip = VNODE_TO_VIMPL(vp); |
438 | struct mount *omp; | | 438 | struct mount *omp; |
439 | | | 439 | |
440 | KASSERT(mp == NULL || (mp->mnt_iflag & IMNT_UNMOUNT) == 0 || | | 440 | KASSERT(mp == NULL || (mp->mnt_iflag & IMNT_UNMOUNT) == 0 || |
441 | vp->v_tag == VT_VFS); | | 441 | vp->v_tag == VT_VFS); |
442 | | | 442 | |
443 | mutex_enter(&mntvnode_lock); | | 443 | mutex_enter(&mntvnode_lock); |
444 | /* | | 444 | /* |
445 | * Delete from old mount point vnode list, if on one. | | 445 | * Delete from old mount point vnode list, if on one. |
446 | */ | | 446 | */ |
447 | if ((omp = vp->v_mount) != NULL) | | 447 | if ((omp = vp->v_mount) != NULL) |
448 | TAILQ_REMOVE(&vp->v_mount->mnt_vnodelist, vip, vi_mntvnodes); | | 448 | TAILQ_REMOVE(&vp->v_mount->mnt_vnodelist, vip, vi_mntvnodes); |
449 | /* | | 449 | /* |
450 | * Insert into list of vnodes for the new mount point, if | | 450 | * Insert into list of vnodes for the new mount point, if |
451 | * available. The caller must take a reference on the mount | | 451 | * available. The caller must take a reference on the mount |
452 | * structure and donate to the vnode. | | 452 | * structure and donate to the vnode. |
453 | */ | | 453 | */ |
454 | if ((vp->v_mount = mp) != NULL) | | 454 | if ((vp->v_mount = mp) != NULL) |
455 | TAILQ_INSERT_TAIL(&mp->mnt_vnodelist, vip, vi_mntvnodes); | | 455 | TAILQ_INSERT_TAIL(&mp->mnt_vnodelist, vip, vi_mntvnodes); |
456 | mutex_exit(&mntvnode_lock); | | 456 | mutex_exit(&mntvnode_lock); |
457 | | | 457 | |
458 | if (omp != NULL) { | | 458 | if (omp != NULL) { |
459 | /* Release reference to old mount. */ | | 459 | /* Release reference to old mount. */ |
460 | vfs_destroy(omp); | | 460 | vfs_destroy(omp); |
461 | } | | 461 | } |
462 | } | | 462 | } |
463 | | | 463 | |
464 | /* | | 464 | /* |
465 | * Remove any vnodes in the vnode table belonging to mount point mp. | | 465 | * Remove any vnodes in the vnode table belonging to mount point mp. |
466 | * | | 466 | * |
467 | * If FORCECLOSE is not specified, there should not be any active ones, | | 467 | * If FORCECLOSE is not specified, there should not be any active ones, |
468 | * return error if any are found (nb: this is a user error, not a | | 468 | * return error if any are found (nb: this is a user error, not a |
469 | * system error). If FORCECLOSE is specified, detach any active vnodes | | 469 | * system error). If FORCECLOSE is specified, detach any active vnodes |
470 | * that are found. | | 470 | * that are found. |
471 | * | | 471 | * |
472 | * If WRITECLOSE is set, only flush out regular file vnodes open for | | 472 | * If WRITECLOSE is set, only flush out regular file vnodes open for |
473 | * writing. | | 473 | * writing. |
474 | * | | 474 | * |
475 | * SKIPSYSTEM causes any vnodes marked VV_SYSTEM to be skipped. | | 475 | * SKIPSYSTEM causes any vnodes marked VV_SYSTEM to be skipped. |
476 | */ | | 476 | */ |
477 | #ifdef DEBUG | | 477 | #ifdef DEBUG |
478 | int busyprt = 0; /* print out busy vnodes */ | | 478 | int busyprt = 0; /* print out busy vnodes */ |
479 | struct ctldebug debug1 = { "busyprt", &busyprt }; | | 479 | struct ctldebug debug1 = { "busyprt", &busyprt }; |
480 | #endif | | 480 | #endif |
481 | | | 481 | |
482 | static vnode_t * | | 482 | static vnode_t * |
483 | vflushnext(struct vnode_iterator *marker, int *when) | | 483 | vflushnext(struct vnode_iterator *marker, int *when) |
484 | { | | 484 | { |
485 | if (hardclock_ticks > *when) { | | 485 | if (hardclock_ticks > *when) { |
486 | yield(); | | 486 | yield(); |
487 | *when = hardclock_ticks + hz / 10; | | 487 | *when = hardclock_ticks + hz / 10; |
488 | } | | 488 | } |
489 | return vfs_vnode_iterator_next1(marker, NULL, NULL, true); | | 489 | return vfs_vnode_iterator_next1(marker, NULL, NULL, true); |
490 | } | | 490 | } |
491 | | | 491 | |
492 | /* | | 492 | /* |
493 | * Flush one vnode. Referenced on entry, unreferenced on return. | | 493 | * Flush one vnode. Referenced on entry, unreferenced on return. |
494 | */ | | 494 | */ |
495 | static int | | 495 | static int |
496 | vflush_one(vnode_t *vp, vnode_t *skipvp, int flags) | | 496 | vflush_one(vnode_t *vp, vnode_t *skipvp, int flags) |
497 | { | | 497 | { |
498 | int error; | | 498 | int error; |
499 | struct vattr vattr; | | 499 | struct vattr vattr; |
500 | | | 500 | |
501 | if (vp == skipvp || | | 501 | if (vp == skipvp || |
502 | ((flags & SKIPSYSTEM) && (vp->v_vflag & VV_SYSTEM))) { | | 502 | ((flags & SKIPSYSTEM) && (vp->v_vflag & VV_SYSTEM))) { |
503 | vrele(vp); | | 503 | vrele(vp); |
504 | return 0; | | 504 | return 0; |
505 | } | | 505 | } |
506 | /* | | 506 | /* |
507 | * If WRITECLOSE is set, only flush out regular file | | 507 | * If WRITECLOSE is set, only flush out regular file |
508 | * vnodes open for writing or open and unlinked. | | 508 | * vnodes open for writing or open and unlinked. |
509 | */ | | 509 | */ |
510 | if ((flags & WRITECLOSE)) { | | 510 | if ((flags & WRITECLOSE)) { |
511 | if (vp->v_type != VREG) { | | 511 | if (vp->v_type != VREG) { |
512 | vrele(vp); | | 512 | vrele(vp); |
513 | return 0; | | 513 | return 0; |
514 | } | | 514 | } |
515 | error = vn_lock(vp, LK_EXCLUSIVE); | | 515 | error = vn_lock(vp, LK_EXCLUSIVE); |
516 | if (error) { | | 516 | if (error) { |
517 | KASSERT(error == ENOENT); | | 517 | KASSERT(error == ENOENT); |
518 | vrele(vp); | | 518 | vrele(vp); |
519 | return 0; | | 519 | return 0; |
520 | } | | 520 | } |
521 | error = VOP_FSYNC(vp, curlwp->l_cred, FSYNC_WAIT, 0, 0); | | 521 | error = VOP_FSYNC(vp, curlwp->l_cred, FSYNC_WAIT, 0, 0); |
522 | if (error == 0) | | 522 | if (error == 0) |
523 | error = VOP_GETATTR(vp, &vattr, curlwp->l_cred); | | 523 | error = VOP_GETATTR(vp, &vattr, curlwp->l_cred); |
524 | VOP_UNLOCK(vp); | | 524 | VOP_UNLOCK(vp); |
525 | if (error) { | | 525 | if (error) { |
526 | vrele(vp); | | 526 | vrele(vp); |
527 | return error; | | 527 | return error; |
528 | } | | 528 | } |
529 | if (vp->v_writecount == 0 && vattr.va_nlink > 0) { | | 529 | if (vp->v_writecount == 0 && vattr.va_nlink > 0) { |
530 | vrele(vp); | | 530 | vrele(vp); |
531 | return 0; | | 531 | return 0; |
532 | } | | 532 | } |
533 | } | | 533 | } |
534 | /* | | 534 | /* |
535 | * First try to recycle the vnode. | | 535 | * First try to recycle the vnode. |
536 | */ | | 536 | */ |
537 | if (vrecycle(vp)) | | 537 | if (vrecycle(vp)) |
538 | return 0; | | 538 | return 0; |
539 | /* | | 539 | /* |
540 | * If FORCECLOSE is set, forcibly close the vnode. | | 540 | * If FORCECLOSE is set, forcibly close the vnode. |
541 | */ | | 541 | */ |
542 | if (flags & FORCECLOSE) { | | 542 | if (flags & FORCECLOSE) { |
543 | vgone(vp); | | 543 | vgone(vp); |
544 | return 0; | | 544 | return 0; |
545 | } | | 545 | } |
546 | vrele(vp); | | 546 | vrele(vp); |
547 | return EBUSY; | | 547 | return EBUSY; |
548 | } | | 548 | } |
549 | | | 549 | |
550 | int | | 550 | int |
551 | vflush(struct mount *mp, vnode_t *skipvp, int flags) | | 551 | vflush(struct mount *mp, vnode_t *skipvp, int flags) |
552 | { | | 552 | { |
553 | vnode_t *vp; | | 553 | vnode_t *vp; |
554 | struct vnode_iterator *marker; | | 554 | struct vnode_iterator *marker; |
555 | int busy, error, when, retries = 2; | | 555 | int busy, error, when, retries = 2; |
556 | | | 556 | |
557 | do { | | 557 | do { |
558 | busy = error = when = 0; | | 558 | busy = error = when = 0; |
559 | | | 559 | |
560 | /* | | 560 | /* |
561 | * First, flush out any vnode references from the | | 561 | * First, flush out any vnode references from the |
562 | * deferred vrele list. | | 562 | * deferred vrele list. |
563 | */ | | 563 | */ |
564 | vfs_drainvnodes(); | | 564 | vfs_drainvnodes(); |
565 | | | 565 | |
566 | vfs_vnode_iterator_init(mp, &marker); | | 566 | vfs_vnode_iterator_init(mp, &marker); |
567 | | | 567 | |
568 | while ((vp = vflushnext(marker, &when)) != NULL) { | | 568 | while ((vp = vflushnext(marker, &when)) != NULL) { |
569 | error = vflush_one(vp, skipvp, flags); | | 569 | error = vflush_one(vp, skipvp, flags); |
570 | if (error == EBUSY) { | | 570 | if (error == EBUSY) { |
571 | error = 0; | | 571 | error = 0; |
572 | busy++; | | 572 | busy++; |
573 | #ifdef DEBUG | | 573 | #ifdef DEBUG |
574 | if (busyprt && retries == 0) | | 574 | if (busyprt && retries == 0) |
575 | vprint("vflush: busy vnode", vp); | | 575 | vprint("vflush: busy vnode", vp); |
576 | #endif | | 576 | #endif |
577 | } else if (error != 0) { | | 577 | } else if (error != 0) { |
578 | break; | | 578 | break; |
579 | } | | 579 | } |
580 | } | | 580 | } |
581 | | | 581 | |
582 | vfs_vnode_iterator_destroy(marker); | | 582 | vfs_vnode_iterator_destroy(marker); |
583 | } while (error == 0 && busy > 0 && retries-- > 0); | | 583 | } while (error == 0 && busy > 0 && retries-- > 0); |
584 | | | 584 | |
585 | if (error) | | 585 | if (error) |
586 | return error; | | 586 | return error; |
587 | if (busy) | | 587 | if (busy) |
588 | return EBUSY; | | 588 | return EBUSY; |
589 | return 0; | | 589 | return 0; |
590 | } | | 590 | } |
591 | | | 591 | |
592 | /* | | 592 | /* |
593 | * Mount a file system. | | 593 | * Mount a file system. |
594 | */ | | 594 | */ |
595 | | | 595 | |
596 | /* | | 596 | /* |
597 | * Scan all active processes to see if any of them have a current or root | | 597 | * Scan all active processes to see if any of them have a current or root |
598 | * directory onto which the new filesystem has just been mounted. If so, | | 598 | * directory onto which the new filesystem has just been mounted. If so, |
599 | * replace them with the new mount point. | | 599 | * replace them with the new mount point. |
600 | */ | | 600 | */ |
601 | static void | | 601 | static void |
602 | mount_checkdirs(vnode_t *olddp) | | 602 | mount_checkdirs(vnode_t *olddp) |
603 | { | | 603 | { |
604 | vnode_t *newdp, *rele1, *rele2; | | 604 | vnode_t *newdp, *rele1, *rele2; |
605 | struct cwdinfo *cwdi; | | 605 | struct cwdinfo *cwdi; |
606 | struct proc *p; | | 606 | struct proc *p; |
607 | bool retry; | | 607 | bool retry; |
608 | | | 608 | |
609 | if (olddp->v_usecount == 1) { | | 609 | if (olddp->v_usecount == 1) { |
610 | return; | | 610 | return; |
611 | } | | 611 | } |
612 | if (VFS_ROOT(olddp->v_mountedhere, &newdp)) | | 612 | if (VFS_ROOT(olddp->v_mountedhere, &newdp)) |
613 | panic("mount: lost mount"); | | 613 | panic("mount: lost mount"); |
614 | | | 614 | |
615 | do { | | 615 | do { |
616 | retry = false; | | 616 | retry = false; |
617 | mutex_enter(proc_lock); | | 617 | mutex_enter(proc_lock); |
618 | PROCLIST_FOREACH(p, &allproc) { | | 618 | PROCLIST_FOREACH(p, &allproc) { |
619 | if ((cwdi = p->p_cwdi) == NULL) | | 619 | if ((cwdi = p->p_cwdi) == NULL) |
620 | continue; | | 620 | continue; |
621 | /* | | 621 | /* |
622 | * Cannot change to the old directory any more, | | 622 | * Cannot change to the old directory any more, |
623 | * so even if we see a stale value it is not a | | 623 | * so even if we see a stale value it is not a |
624 | * problem. | | 624 | * problem. |
625 | */ | | 625 | */ |
626 | if (cwdi->cwdi_cdir != olddp && | | 626 | if (cwdi->cwdi_cdir != olddp && |
627 | cwdi->cwdi_rdir != olddp) | | 627 | cwdi->cwdi_rdir != olddp) |
628 | continue; | | 628 | continue; |
629 | retry = true; | | 629 | retry = true; |
630 | rele1 = NULL; | | 630 | rele1 = NULL; |
631 | rele2 = NULL; | | 631 | rele2 = NULL; |
632 | atomic_inc_uint(&cwdi->cwdi_refcnt); | | 632 | atomic_inc_uint(&cwdi->cwdi_refcnt); |
633 | mutex_exit(proc_lock); | | 633 | mutex_exit(proc_lock); |
634 | rw_enter(&cwdi->cwdi_lock, RW_WRITER); | | 634 | rw_enter(&cwdi->cwdi_lock, RW_WRITER); |
635 | if (cwdi->cwdi_cdir == olddp) { | | 635 | if (cwdi->cwdi_cdir == olddp) { |
636 | rele1 = cwdi->cwdi_cdir; | | 636 | rele1 = cwdi->cwdi_cdir; |
637 | vref(newdp); | | 637 | vref(newdp); |
638 | cwdi->cwdi_cdir = newdp; | | 638 | cwdi->cwdi_cdir = newdp; |
639 | } | | 639 | } |
640 | if (cwdi->cwdi_rdir == olddp) { | | 640 | if (cwdi->cwdi_rdir == olddp) { |
641 | rele2 = cwdi->cwdi_rdir; | | 641 | rele2 = cwdi->cwdi_rdir; |
642 | vref(newdp); | | 642 | vref(newdp); |
643 | cwdi->cwdi_rdir = newdp; | | 643 | cwdi->cwdi_rdir = newdp; |
644 | } | | 644 | } |
645 | rw_exit(&cwdi->cwdi_lock); | | 645 | rw_exit(&cwdi->cwdi_lock); |
646 | cwdfree(cwdi); | | 646 | cwdfree(cwdi); |
647 | if (rele1 != NULL) | | 647 | if (rele1 != NULL) |
648 | vrele(rele1); | | 648 | vrele(rele1); |
649 | if (rele2 != NULL) | | 649 | if (rele2 != NULL) |
650 | vrele(rele2); | | 650 | vrele(rele2); |
651 | mutex_enter(proc_lock); | | 651 | mutex_enter(proc_lock); |
652 | break; | | 652 | break; |
653 | } | | 653 | } |
654 | mutex_exit(proc_lock); | | 654 | mutex_exit(proc_lock); |
655 | } while (retry); | | 655 | } while (retry); |
656 | | | 656 | |
657 | if (rootvnode == olddp) { | | 657 | if (rootvnode == olddp) { |
658 | vrele(rootvnode); | | 658 | vrele(rootvnode); |
659 | vref(newdp); | | 659 | vref(newdp); |
660 | rootvnode = newdp; | | 660 | rootvnode = newdp; |
661 | } | | 661 | } |
662 | vput(newdp); | | 662 | vput(newdp); |
663 | } | | 663 | } |
664 | | | 664 | |
665 | /* | | 665 | /* |
666 | * Start extended attributes | | 666 | * Start extended attributes |
667 | */ | | 667 | */ |
668 | static int | | 668 | static int |
669 | start_extattr(struct mount *mp) | | 669 | start_extattr(struct mount *mp) |
670 | { | | 670 | { |
671 | int error; | | 671 | int error; |
672 | | | 672 | |
673 | error = VFS_EXTATTRCTL(mp, EXTATTR_CMD_START, NULL, 0, NULL); | | 673 | error = VFS_EXTATTRCTL(mp, EXTATTR_CMD_START, NULL, 0, NULL); |
674 | if (error) | | 674 | if (error) |
675 | printf("%s: failed to start extattr: error = %d\n", | | 675 | printf("%s: failed to start extattr: error = %d\n", |
676 | mp->mnt_stat.f_mntonname, error); | | 676 | mp->mnt_stat.f_mntonname, error); |
677 | | | 677 | |
678 | return error; | | 678 | return error; |
679 | } | | 679 | } |
680 | | | 680 | |
681 | int | | 681 | int |
682 | mount_domount(struct lwp *l, vnode_t **vpp, struct vfsops *vfsops, | | 682 | mount_domount(struct lwp *l, vnode_t **vpp, struct vfsops *vfsops, |
683 | const char *path, int flags, void *data, size_t *data_len) | | 683 | const char *path, int flags, void *data, size_t *data_len) |
684 | { | | 684 | { |
685 | vnode_t *vp = *vpp; | | 685 | vnode_t *vp = *vpp; |
686 | struct mount *mp; | | 686 | struct mount *mp; |
687 | struct pathbuf *pb; | | 687 | struct pathbuf *pb; |
688 | struct nameidata nd; | | 688 | struct nameidata nd; |
689 | int error; | | 689 | int error; |
690 | | | 690 | |
691 | error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MOUNT, | | 691 | error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MOUNT, |
692 | KAUTH_REQ_SYSTEM_MOUNT_NEW, vp, KAUTH_ARG(flags), data); | | 692 | KAUTH_REQ_SYSTEM_MOUNT_NEW, vp, KAUTH_ARG(flags), data); |
693 | if (error) { | | 693 | if (error) { |
694 | vfs_delref(vfsops); | | 694 | vfs_delref(vfsops); |
695 | return error; | | 695 | return error; |
696 | } | | 696 | } |
697 | | | 697 | |
698 | /* Cannot make a non-dir a mount-point (from here anyway). */ | | 698 | /* Cannot make a non-dir a mount-point (from here anyway). */ |
699 | if (vp->v_type != VDIR) { | | 699 | if (vp->v_type != VDIR) { |
700 | vfs_delref(vfsops); | | 700 | vfs_delref(vfsops); |
701 | return ENOTDIR; | | 701 | return ENOTDIR; |
702 | } | | 702 | } |
703 | | | 703 | |
704 | if (flags & MNT_EXPORTED) { | | 704 | if (flags & MNT_EXPORTED) { |
705 | vfs_delref(vfsops); | | 705 | vfs_delref(vfsops); |
706 | return EINVAL; | | 706 | return EINVAL; |
707 | } | | 707 | } |
708 | | | 708 | |
709 | if ((mp = vfs_mountalloc(vfsops, vp)) == NULL) { | | 709 | if ((mp = vfs_mountalloc(vfsops, vp)) == NULL) { |
710 | vfs_delref(vfsops); | | 710 | vfs_delref(vfsops); |
711 | return ENOMEM; | | 711 | return ENOMEM; |
712 | } | | 712 | } |
713 | | | 713 | |
714 | if ((error = fstrans_mount(mp)) != 0) { | | 714 | if ((error = fstrans_mount(mp)) != 0) { |
715 | vfs_unbusy(mp, false, NULL); | | 715 | vfs_unbusy(mp, false, NULL); |
716 | vfs_destroy(mp); | | 716 | vfs_destroy(mp); |
717 | return error; | | 717 | return error; |
718 | } | | 718 | } |
719 | | | 719 | |
720 | mp->mnt_stat.f_owner = kauth_cred_geteuid(l->l_cred); | | 720 | mp->mnt_stat.f_owner = kauth_cred_geteuid(l->l_cred); |
721 | | | 721 | |
722 | /* | | 722 | /* |
723 | * The underlying file system may refuse the mount for | | 723 | * The underlying file system may refuse the mount for |
724 | * various reasons. Allow the user to force it to happen. | | 724 | * various reasons. Allow the user to force it to happen. |
725 | * | | 725 | * |
726 | * Set the mount level flags. | | 726 | * Set the mount level flags. |
727 | */ | | 727 | */ |
728 | mp->mnt_flag = flags & (MNT_BASIC_FLAGS | MNT_FORCE | MNT_IGNORE); | | 728 | mp->mnt_flag = flags & (MNT_BASIC_FLAGS | MNT_FORCE | MNT_IGNORE); |
729 | | | 729 | |
730 | mutex_enter(&mp->mnt_updating); | | 730 | mutex_enter(&mp->mnt_updating); |
731 | error = VFS_MOUNT(mp, path, data, data_len); | | 731 | error = VFS_MOUNT(mp, path, data, data_len); |
732 | mp->mnt_flag &= ~MNT_OP_FLAGS; | | 732 | mp->mnt_flag &= ~MNT_OP_FLAGS; |
733 | | | 733 | |
734 | if (error != 0) | | 734 | if (error != 0) |
735 | goto err_unmounted; | | 735 | goto err_unmounted; |
736 | | | 736 | |
737 | /* | | 737 | /* |
738 | * Validate and prepare the mount point. | | 738 | * Validate and prepare the mount point. |
739 | */ | | 739 | */ |
740 | error = pathbuf_copyin(path, &pb); | | 740 | error = pathbuf_copyin(path, &pb); |
741 | if (error != 0) { | | 741 | if (error != 0) { |
742 | goto err_mounted; | | 742 | goto err_mounted; |
743 | } | | 743 | } |
744 | NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | TRYEMULROOT, pb); | | 744 | NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | TRYEMULROOT, pb); |
745 | error = namei(&nd); | | 745 | error = namei(&nd); |
746 | pathbuf_destroy(pb); | | 746 | pathbuf_destroy(pb); |
747 | if (error != 0) { | | 747 | if (error != 0) { |
748 | goto err_mounted; | | 748 | goto err_mounted; |
749 | } | | 749 | } |
750 | if (nd.ni_vp != vp) { | | 750 | if (nd.ni_vp != vp) { |
751 | vput(nd.ni_vp); | | 751 | vput(nd.ni_vp); |
752 | error = EINVAL; | | 752 | error = EINVAL; |
753 | goto err_mounted; | | 753 | goto err_mounted; |
754 | } | | 754 | } |
755 | if (vp->v_mountedhere != NULL) { | | 755 | if (vp->v_mountedhere != NULL) { |
756 | vput(nd.ni_vp); | | 756 | vput(nd.ni_vp); |
757 | error = EBUSY; | | 757 | error = EBUSY; |
758 | goto err_mounted; | | 758 | goto err_mounted; |
759 | } | | 759 | } |
760 | error = vinvalbuf(vp, V_SAVE, l->l_cred, l, 0, 0); | | 760 | error = vinvalbuf(vp, V_SAVE, l->l_cred, l, 0, 0); |
761 | if (error != 0) { | | 761 | if (error != 0) { |
762 | vput(nd.ni_vp); | | 762 | vput(nd.ni_vp); |
763 | goto err_mounted; | | 763 | goto err_mounted; |
764 | } | | 764 | } |
765 | | | 765 | |
766 | /* | | 766 | /* |
767 | * Put the new filesystem on the mount list after root. | | 767 | * Put the new filesystem on the mount list after root. |
768 | */ | | 768 | */ |
769 | cache_purge(vp); | | 769 | cache_purge(vp); |
770 | mp->mnt_iflag &= ~IMNT_WANTRDWR; | | 770 | mp->mnt_iflag &= ~IMNT_WANTRDWR; |
771 | | | 771 | |
772 | mutex_enter(&mountlist_lock); | | 772 | mutex_enter(&mountlist_lock); |
773 | TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list); | | 773 | TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list); |
774 | mutex_exit(&mountlist_lock); | | 774 | mutex_exit(&mountlist_lock); |
775 | if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) | | 775 | if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) |
776 | vfs_syncer_add_to_worklist(mp); | | 776 | vfs_syncer_add_to_worklist(mp); |
777 | vp->v_mountedhere = mp; | | 777 | vp->v_mountedhere = mp; |
778 | vput(nd.ni_vp); | | 778 | vput(nd.ni_vp); |
779 | | | 779 | |
780 | mount_checkdirs(vp); | | 780 | mount_checkdirs(vp); |
781 | mutex_exit(&mp->mnt_updating); | | 781 | mutex_exit(&mp->mnt_updating); |
782 | | | 782 | |
783 | /* Hold an additional reference to the mount across VFS_START(). */ | | 783 | /* Hold an additional reference to the mount across VFS_START(). */ |
784 | vfs_unbusy(mp, true, NULL); | | 784 | vfs_unbusy(mp, true, NULL); |
785 | (void) VFS_STATVFS(mp, &mp->mnt_stat); | | 785 | (void) VFS_STATVFS(mp, &mp->mnt_stat); |
786 | error = VFS_START(mp, 0); | | 786 | error = VFS_START(mp, 0); |
787 | if (error) { | | 787 | if (error) { |
788 | vrele(vp); | | 788 | vrele(vp); |
789 | } else if (flags & MNT_EXTATTR) { | | 789 | } else if (flags & MNT_EXTATTR) { |
790 | (void)start_extattr(mp); | | 790 | (void)start_extattr(mp); |
791 | } | | 791 | } |
792 | /* Drop reference held for VFS_START(). */ | | 792 | /* Drop reference held for VFS_START(). */ |
793 | vfs_destroy(mp); | | 793 | vfs_destroy(mp); |
794 | *vpp = NULL; | | 794 | *vpp = NULL; |
795 | return error; | | 795 | return error; |
796 | | | 796 | |
797 | err_mounted: | | 797 | err_mounted: |
798 | if (VFS_UNMOUNT(mp, MNT_FORCE) != 0) | | 798 | if (VFS_UNMOUNT(mp, MNT_FORCE) != 0) |
799 | panic("Unmounting fresh file system failed"); | | 799 | panic("Unmounting fresh file system failed"); |
800 | | | 800 | |
801 | err_unmounted: | | 801 | err_unmounted: |
802 | vp->v_mountedhere = NULL; | | 802 | vp->v_mountedhere = NULL; |
803 | mutex_exit(&mp->mnt_updating); | | 803 | mutex_exit(&mp->mnt_updating); |
804 | fstrans_unmount(mp); | | 804 | fstrans_unmount(mp); |
805 | vfs_unbusy(mp, false, NULL); | | 805 | vfs_unbusy(mp, false, NULL); |
806 | vfs_destroy(mp); | | 806 | vfs_destroy(mp); |
807 | | | 807 | |
808 | return error; | | 808 | return error; |
809 | } | | 809 | } |
810 | | | 810 | |
811 | /* | | 811 | /* |
812 | * Do the actual file system unmount. File system is assumed to have | | 812 | * Do the actual file system unmount. File system is assumed to have |
813 | * been locked by the caller. | | 813 | * been locked by the caller. |
814 | * | | 814 | * |
815 | * => Caller hold reference to the mount, explicitly for dounmount(). | | 815 | * => Caller hold reference to the mount, explicitly for dounmount(). |
816 | */ | | 816 | */ |
817 | int | | 817 | int |
818 | dounmount(struct mount *mp, int flags, struct lwp *l) | | 818 | dounmount(struct mount *mp, int flags, struct lwp *l) |
819 | { | | 819 | { |
| | | 820 | struct mount *cmp; |
820 | vnode_t *coveredvp; | | 821 | vnode_t *coveredvp; |
821 | int error, async, used_syncer, used_extattr; | | 822 | int error, async, used_syncer, used_extattr; |
822 | | | 823 | |
823 | #if NVERIEXEC > 0 | | 824 | #if NVERIEXEC > 0 |
824 | error = veriexec_unmountchk(mp); | | 825 | error = veriexec_unmountchk(mp); |
825 | if (error) | | 826 | if (error) |
826 | return (error); | | 827 | return (error); |
827 | #endif /* NVERIEXEC > 0 */ | | 828 | #endif /* NVERIEXEC > 0 */ |
828 | | | 829 | |
829 | /* | | 830 | /* |
| | | 831 | * No unmount below layered mounts. |
| | | 832 | */ |
| | | 833 | mutex_enter(&mountlist_lock); |
| | | 834 | TAILQ_FOREACH(cmp, &mountlist, mnt_list) { |
| | | 835 | if (cmp->mnt_lower == mp) { |
| | | 836 | mutex_exit(&mountlist_lock); |
| | | 837 | return EBUSY; |
| | | 838 | } |
| | | 839 | } |
| | | 840 | mutex_exit(&mountlist_lock); |
| | | 841 | |
| | | 842 | /* |
830 | * XXX Freeze syncer. Must do this before locking the | | 843 | * XXX Freeze syncer. Must do this before locking the |
831 | * mount point. See dounmount() for details. | | 844 | * mount point. See dounmount() for details. |
832 | */ | | 845 | */ |
833 | mutex_enter(&syncer_mutex); | | 846 | mutex_enter(&syncer_mutex); |
834 | | | 847 | |
835 | /* | | 848 | /* |
836 | * Abort unmount attempt when the filesystem is in use | | 849 | * Abort unmount attempt when the filesystem is in use |
837 | */ | | 850 | */ |
838 | mutex_enter(&mp->mnt_unmounting); | | 851 | mutex_enter(&mp->mnt_unmounting); |
839 | if (mp->mnt_busynest != 0) { | | 852 | if (mp->mnt_busynest != 0) { |
840 | mutex_exit(&mp->mnt_unmounting); | | 853 | mutex_exit(&mp->mnt_unmounting); |
841 | mutex_exit(&syncer_mutex); | | 854 | mutex_exit(&syncer_mutex); |
842 | return EBUSY; | | 855 | return EBUSY; |
843 | } | | 856 | } |
844 | | | 857 | |
845 | /* | | 858 | /* |
846 | * Abort unmount attempt when the filesystem is not mounted | | 859 | * Abort unmount attempt when the filesystem is not mounted |
847 | */ | | 860 | */ |
848 | if ((mp->mnt_iflag & IMNT_GONE) != 0) { | | 861 | if ((mp->mnt_iflag & IMNT_GONE) != 0) { |
849 | mutex_exit(&mp->mnt_unmounting); | | 862 | mutex_exit(&mp->mnt_unmounting); |
850 | mutex_exit(&syncer_mutex); | | 863 | mutex_exit(&syncer_mutex); |
851 | return ENOENT; | | 864 | return ENOENT; |
852 | } | | 865 | } |
853 | | | 866 | |
854 | used_syncer = (mp->mnt_iflag & IMNT_ONWORKLIST) != 0; | | 867 | used_syncer = (mp->mnt_iflag & IMNT_ONWORKLIST) != 0; |
855 | used_extattr = mp->mnt_flag & MNT_EXTATTR; | | 868 | used_extattr = mp->mnt_flag & MNT_EXTATTR; |
856 | | | 869 | |
857 | /* | | 870 | /* |
858 | * XXX Syncer must be frozen when we get here. This should really | | 871 | * XXX Syncer must be frozen when we get here. This should really |
859 | * be done on a per-mountpoint basis, but the syncer doesn't work | | 872 | * be done on a per-mountpoint basis, but the syncer doesn't work |
860 | * like that. | | 873 | * like that. |
861 | * | | 874 | * |
862 | * The caller of dounmount() must acquire syncer_mutex because | | 875 | * The caller of dounmount() must acquire syncer_mutex because |
863 | * the syncer itself acquires locks in syncer_mutex -> vfs_busy | | 876 | * the syncer itself acquires locks in syncer_mutex -> vfs_busy |
864 | * order, and we must preserve that order to avoid deadlock. | | 877 | * order, and we must preserve that order to avoid deadlock. |
865 | * | | 878 | * |
866 | * So, if the file system did not use the syncer, now is | | 879 | * So, if the file system did not use the syncer, now is |
867 | * the time to release the syncer_mutex. | | 880 | * the time to release the syncer_mutex. |
868 | */ | | 881 | */ |
869 | if (used_syncer == 0) { | | 882 | if (used_syncer == 0) { |
870 | mutex_exit(&syncer_mutex); | | 883 | mutex_exit(&syncer_mutex); |
871 | } | | 884 | } |
872 | mp->mnt_iflag |= IMNT_UNMOUNT; | | 885 | mp->mnt_iflag |= IMNT_UNMOUNT; |
873 | mutex_enter(&mp->mnt_updating); | | 886 | mutex_enter(&mp->mnt_updating); |
874 | async = mp->mnt_flag & MNT_ASYNC; | | 887 | async = mp->mnt_flag & MNT_ASYNC; |
875 | mp->mnt_flag &= ~MNT_ASYNC; | | 888 | mp->mnt_flag &= ~MNT_ASYNC; |
876 | cache_purgevfs(mp); /* remove cache entries for this file sys */ | | 889 | cache_purgevfs(mp); /* remove cache entries for this file sys */ |
877 | if (used_syncer) | | 890 | if (used_syncer) |
878 | vfs_syncer_remove_from_worklist(mp); | | 891 | vfs_syncer_remove_from_worklist(mp); |
879 | error = 0; | | 892 | error = 0; |
880 | if (((mp->mnt_flag & MNT_RDONLY) == 0) && ((flags & MNT_FORCE) == 0)) { | | 893 | if (((mp->mnt_flag & MNT_RDONLY) == 0) && ((flags & MNT_FORCE) == 0)) { |
881 | error = VFS_SYNC(mp, MNT_WAIT, l->l_cred); | | 894 | error = VFS_SYNC(mp, MNT_WAIT, l->l_cred); |
882 | } | | 895 | } |
883 | if (error == 0 || (flags & MNT_FORCE)) { | | 896 | if (error == 0 || (flags & MNT_FORCE)) { |
884 | error = VFS_UNMOUNT(mp, flags); | | 897 | error = VFS_UNMOUNT(mp, flags); |
885 | } | | 898 | } |
886 | if (error) { | | 899 | if (error) { |
887 | mp->mnt_iflag &= ~IMNT_UNMOUNT; | | 900 | mp->mnt_iflag &= ~IMNT_UNMOUNT; |
888 | mutex_exit(&mp->mnt_unmounting); | | 901 | mutex_exit(&mp->mnt_unmounting); |
889 | if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) | | 902 | if ((mp->mnt_flag & (MNT_RDONLY | MNT_ASYNC)) == 0) |
890 | vfs_syncer_add_to_worklist(mp); | | 903 | vfs_syncer_add_to_worklist(mp); |
891 | mp->mnt_flag |= async; | | 904 | mp->mnt_flag |= async; |
892 | mutex_exit(&mp->mnt_updating); | | 905 | mutex_exit(&mp->mnt_updating); |
893 | if (used_syncer) | | 906 | if (used_syncer) |
894 | mutex_exit(&syncer_mutex); | | 907 | mutex_exit(&syncer_mutex); |
895 | if (used_extattr) { | | 908 | if (used_extattr) { |
896 | if (start_extattr(mp) != 0) | | 909 | if (start_extattr(mp) != 0) |
897 | mp->mnt_flag &= ~MNT_EXTATTR; | | 910 | mp->mnt_flag &= ~MNT_EXTATTR; |
898 | else | | 911 | else |
899 | mp->mnt_flag |= MNT_EXTATTR; | | 912 | mp->mnt_flag |= MNT_EXTATTR; |
900 | } | | 913 | } |
901 | return (error); | | 914 | return (error); |
902 | } | | 915 | } |
903 | mutex_exit(&mp->mnt_updating); | | 916 | mutex_exit(&mp->mnt_updating); |
904 | | | 917 | |
905 | /* | | 918 | /* |
906 | * release mnt_umounting lock here, because other code calls | | 919 | * release mnt_umounting lock here, because other code calls |
907 | * vfs_busy() while holding the mountlist_lock. | | 920 | * vfs_busy() while holding the mountlist_lock. |
908 | * | | 921 | * |
909 | * mark filesystem as gone to prevent further umounts | | 922 | * mark filesystem as gone to prevent further umounts |
910 | * after mnt_umounting lock is gone, this also prevents | | 923 | * after mnt_umounting lock is gone, this also prevents |
911 | * vfs_busy() from succeeding. | | 924 | * vfs_busy() from succeeding. |
912 | */ | | 925 | */ |
913 | mp->mnt_iflag |= IMNT_GONE; | | 926 | mp->mnt_iflag |= IMNT_GONE; |
914 | mutex_exit(&mp->mnt_unmounting); | | 927 | mutex_exit(&mp->mnt_unmounting); |
915 | | | 928 | |
916 | if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) { | | 929 | if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) { |
917 | vn_lock(coveredvp, LK_EXCLUSIVE | LK_RETRY); | | 930 | vn_lock(coveredvp, LK_EXCLUSIVE | LK_RETRY); |
918 | coveredvp->v_mountedhere = NULL; | | 931 | coveredvp->v_mountedhere = NULL; |
919 | VOP_UNLOCK(coveredvp); | | 932 | VOP_UNLOCK(coveredvp); |
920 | } | | 933 | } |
921 | mutex_enter(&mountlist_lock); | | 934 | mutex_enter(&mountlist_lock); |
922 | TAILQ_REMOVE(&mountlist, mp, mnt_list); | | 935 | TAILQ_REMOVE(&mountlist, mp, mnt_list); |
923 | mutex_exit(&mountlist_lock); | | 936 | mutex_exit(&mountlist_lock); |
924 | if (TAILQ_FIRST(&mp->mnt_vnodelist) != NULL) | | 937 | if (TAILQ_FIRST(&mp->mnt_vnodelist) != NULL) |
925 | panic("unmount: dangling vnode"); | | 938 | panic("unmount: dangling vnode"); |
926 | if (used_syncer) | | 939 | if (used_syncer) |
927 | mutex_exit(&syncer_mutex); | | 940 | mutex_exit(&syncer_mutex); |
928 | vfs_hooks_unmount(mp); | | 941 | vfs_hooks_unmount(mp); |
929 | | | 942 | |
930 | fstrans_unmount(mp); | | 943 | fstrans_unmount(mp); |
931 | vfs_destroy(mp); /* reference from mount() */ | | 944 | vfs_destroy(mp); /* reference from mount() */ |
932 | if (coveredvp != NULLVP) { | | 945 | if (coveredvp != NULLVP) { |
933 | vrele(coveredvp); | | 946 | vrele(coveredvp); |
934 | } | | 947 | } |
935 | return (0); | | 948 | return (0); |
936 | } | | 949 | } |
937 | | | 950 | |
938 | /* | | 951 | /* |
939 | * Unmount all file systems. | | 952 | * Unmount all file systems. |
940 | * We traverse the list in reverse order under the assumption that doing so | | 953 | * We traverse the list in reverse order under the assumption that doing so |
941 | * will avoid needing to worry about dependencies. | | 954 | * will avoid needing to worry about dependencies. |
942 | */ | | 955 | */ |
943 | bool | | 956 | bool |
944 | vfs_unmountall(struct lwp *l) | | 957 | vfs_unmountall(struct lwp *l) |
945 | { | | 958 | { |
946 | | | 959 | |
947 | printf("unmounting file systems...\n"); | | 960 | printf("unmounting file systems...\n"); |
948 | return vfs_unmountall1(l, true, true); | | 961 | return vfs_unmountall1(l, true, true); |
949 | } | | 962 | } |
950 | | | 963 | |
951 | static void | | 964 | static void |
952 | vfs_unmount_print(struct mount *mp, const char *pfx) | | 965 | vfs_unmount_print(struct mount *mp, const char *pfx) |
953 | { | | 966 | { |
954 | | | 967 | |
955 | aprint_verbose("%sunmounted %s on %s type %s\n", pfx, | | 968 | aprint_verbose("%sunmounted %s on %s type %s\n", pfx, |
956 | mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname, | | 969 | mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname, |
957 | mp->mnt_stat.f_fstypename); | | 970 | mp->mnt_stat.f_fstypename); |
958 | } | | 971 | } |
959 | | | 972 | |
960 | bool | | 973 | bool |
961 | vfs_unmount_forceone(struct lwp *l) | | 974 | vfs_unmount_forceone(struct lwp *l) |
962 | { | | 975 | { |
963 | struct mount *mp, *nmp; | | 976 | struct mount *mp, *nmp; |
964 | int error; | | 977 | int error; |
965 | | | 978 | |
966 | nmp = NULL; | | 979 | nmp = NULL; |
967 | | | 980 | |
968 | TAILQ_FOREACH_REVERSE(mp, &mountlist, mntlist, mnt_list) { | | 981 | TAILQ_FOREACH_REVERSE(mp, &mountlist, mntlist, mnt_list) { |
969 | if (nmp == NULL || mp->mnt_gen > nmp->mnt_gen) { | | 982 | if (nmp == NULL || mp->mnt_gen > nmp->mnt_gen) { |
970 | nmp = mp; | | 983 | nmp = mp; |
971 | } | | 984 | } |
972 | } | | 985 | } |
973 | if (nmp == NULL) { | | 986 | if (nmp == NULL) { |
974 | return false; | | 987 | return false; |
975 | } | | 988 | } |
976 | | | 989 | |
977 | #ifdef DEBUG | | 990 | #ifdef DEBUG |
978 | printf("forcefully unmounting %s (%s)...\n", | | 991 | printf("forcefully unmounting %s (%s)...\n", |
979 | nmp->mnt_stat.f_mntonname, nmp->mnt_stat.f_mntfromname); | | 992 | nmp->mnt_stat.f_mntonname, nmp->mnt_stat.f_mntfromname); |
980 | #endif | | 993 | #endif |
981 | atomic_inc_uint(&nmp->mnt_refcnt); | | 994 | atomic_inc_uint(&nmp->mnt_refcnt); |
982 | if ((error = dounmount(nmp, MNT_FORCE, l)) == 0) { | | 995 | if ((error = dounmount(nmp, MNT_FORCE, l)) == 0) { |
983 | vfs_unmount_print(nmp, "forcefully "); | | 996 | vfs_unmount_print(nmp, "forcefully "); |
984 | return true; | | 997 | return true; |
985 | } else { | | 998 | } else { |
986 | vfs_destroy(nmp); | | 999 | vfs_destroy(nmp); |
987 | } | | 1000 | } |
988 | | | 1001 | |
989 | #ifdef DEBUG | | 1002 | #ifdef DEBUG |
990 | printf("forceful unmount of %s failed with error %d\n", | | 1003 | printf("forceful unmount of %s failed with error %d\n", |
991 | nmp->mnt_stat.f_mntonname, error); | | 1004 | nmp->mnt_stat.f_mntonname, error); |
992 | #endif | | 1005 | #endif |
993 | | | 1006 | |
994 | return false; | | 1007 | return false; |
995 | } | | 1008 | } |
996 | | | 1009 | |
997 | bool | | 1010 | bool |
998 | vfs_unmountall1(struct lwp *l, bool force, bool verbose) | | 1011 | vfs_unmountall1(struct lwp *l, bool force, bool verbose) |
999 | { | | 1012 | { |
1000 | struct mount *mp, *nmp; | | 1013 | struct mount *mp, *nmp; |
1001 | bool any_error = false, progress = false; | | 1014 | bool any_error = false, progress = false; |
1002 | int error; | | 1015 | int error; |
1003 | | | 1016 | |
1004 | TAILQ_FOREACH_REVERSE_SAFE(mp, &mountlist, mntlist, mnt_list, nmp) { | | 1017 | TAILQ_FOREACH_REVERSE_SAFE(mp, &mountlist, mntlist, mnt_list, nmp) { |
1005 | #ifdef DEBUG | | 1018 | #ifdef DEBUG |
1006 | printf("unmounting %p %s (%s)...\n", | | 1019 | printf("unmounting %p %s (%s)...\n", |
1007 | (void *)mp, mp->mnt_stat.f_mntonname, | | 1020 | (void *)mp, mp->mnt_stat.f_mntonname, |
1008 | mp->mnt_stat.f_mntfromname); | | 1021 | mp->mnt_stat.f_mntfromname); |
1009 | #endif | | 1022 | #endif |
1010 | atomic_inc_uint(&mp->mnt_refcnt); | | 1023 | atomic_inc_uint(&mp->mnt_refcnt); |
1011 | if ((error = dounmount(mp, force ? MNT_FORCE : 0, l)) == 0) { | | 1024 | if ((error = dounmount(mp, force ? MNT_FORCE : 0, l)) == 0) { |
1012 | vfs_unmount_print(mp, ""); | | 1025 | vfs_unmount_print(mp, ""); |
1013 | progress = true; | | 1026 | progress = true; |
1014 | } else { | | 1027 | } else { |
1015 | vfs_destroy(mp); | | 1028 | vfs_destroy(mp); |
1016 | if (verbose) { | | 1029 | if (verbose) { |
1017 | printf("unmount of %s failed with error %d\n", | | 1030 | printf("unmount of %s failed with error %d\n", |
1018 | mp->mnt_stat.f_mntonname, error); | | 1031 | mp->mnt_stat.f_mntonname, error); |
1019 | } | | 1032 | } |
1020 | any_error = true; | | 1033 | any_error = true; |
1021 | } | | 1034 | } |
1022 | } | | 1035 | } |
1023 | if (verbose) { | | 1036 | if (verbose) { |
1024 | printf("unmounting done\n"); | | 1037 | printf("unmounting done\n"); |
1025 | } | | 1038 | } |
1026 | if (any_error && verbose) { | | 1039 | if (any_error && verbose) { |
1027 | printf("WARNING: some file systems would not unmount\n"); | | 1040 | printf("WARNING: some file systems would not unmount\n"); |
1028 | } | | 1041 | } |
1029 | return progress; | | 1042 | return progress; |
1030 | } | | 1043 | } |
1031 | | | 1044 | |
1032 | void | | 1045 | void |
1033 | vfs_sync_all(struct lwp *l) | | 1046 | vfs_sync_all(struct lwp *l) |
1034 | { | | 1047 | { |
1035 | printf("syncing disks... "); | | 1048 | printf("syncing disks... "); |
1036 | | | 1049 | |
1037 | /* remove user processes from run queue */ | | 1050 | /* remove user processes from run queue */ |
1038 | suspendsched(); | | 1051 | suspendsched(); |
1039 | (void)spl0(); | | 1052 | (void)spl0(); |
1040 | | | 1053 | |
1041 | /* avoid coming back this way again if we panic. */ | | 1054 | /* avoid coming back this way again if we panic. */ |
1042 | doing_shutdown = 1; | | 1055 | doing_shutdown = 1; |
1043 | | | 1056 | |
1044 | do_sys_sync(l); | | 1057 | do_sys_sync(l); |
1045 | | | 1058 | |
1046 | /* Wait for sync to finish. */ | | 1059 | /* Wait for sync to finish. */ |
1047 | if (buf_syncwait() != 0) { | | 1060 | if (buf_syncwait() != 0) { |
1048 | #if defined(DDB) && defined(DEBUG_HALT_BUSY) | | 1061 | #if defined(DDB) && defined(DEBUG_HALT_BUSY) |
1049 | Debugger(); | | 1062 | Debugger(); |
1050 | #endif | | 1063 | #endif |
1051 | printf("giving up\n"); | | 1064 | printf("giving up\n"); |
1052 | return; | | 1065 | return; |
1053 | } else | | 1066 | } else |
1054 | printf("done\n"); | | 1067 | printf("done\n"); |
1055 | } | | 1068 | } |
1056 | | | 1069 | |
1057 | /* | | 1070 | /* |
1058 | * Sync and unmount file systems before shutting down. | | 1071 | * Sync and unmount file systems before shutting down. |
1059 | */ | | 1072 | */ |
1060 | void | | 1073 | void |
1061 | vfs_shutdown(void) | | 1074 | vfs_shutdown(void) |
1062 | { | | 1075 | { |
1063 | lwp_t *l = curlwp; | | 1076 | lwp_t *l = curlwp; |
1064 | | | 1077 | |
1065 | vfs_sync_all(l); | | 1078 | vfs_sync_all(l); |
1066 | | | 1079 | |
1067 | /* | | 1080 | /* |
1068 | * If we have paniced - do not make the situation potentially | | 1081 | * If we have paniced - do not make the situation potentially |
1069 | * worse by unmounting the file systems. | | 1082 | * worse by unmounting the file systems. |
1070 | */ | | 1083 | */ |
1071 | if (panicstr != NULL) { | | 1084 | if (panicstr != NULL) { |
1072 | return; | | 1085 | return; |
1073 | } | | 1086 | } |
1074 | | | 1087 | |
1075 | /* Unmount file systems. */ | | 1088 | /* Unmount file systems. */ |
1076 | vfs_unmountall(l); | | 1089 | vfs_unmountall(l); |
1077 | } | | 1090 | } |
1078 | | | 1091 | |
1079 | /* | | 1092 | /* |
1080 | * Print a list of supported file system types (used by vfs_mountroot) | | 1093 | * Print a list of supported file system types (used by vfs_mountroot) |
1081 | */ | | 1094 | */ |
1082 | static void | | 1095 | static void |
1083 | vfs_print_fstypes(void) | | 1096 | vfs_print_fstypes(void) |
1084 | { | | 1097 | { |
1085 | struct vfsops *v; | | 1098 | struct vfsops *v; |
1086 | int cnt = 0; | | 1099 | int cnt = 0; |
1087 | | | 1100 | |
1088 | mutex_enter(&vfs_list_lock); | | 1101 | mutex_enter(&vfs_list_lock); |
1089 | LIST_FOREACH(v, &vfs_list, vfs_list) | | 1102 | LIST_FOREACH(v, &vfs_list, vfs_list) |
1090 | ++cnt; | | 1103 | ++cnt; |
1091 | mutex_exit(&vfs_list_lock); | | 1104 | mutex_exit(&vfs_list_lock); |
1092 | | | 1105 | |
1093 | if (cnt == 0) { | | 1106 | if (cnt == 0) { |
1094 | printf("WARNING: No file system modules have been loaded.\n"); | | 1107 | printf("WARNING: No file system modules have been loaded.\n"); |
1095 | return; | | 1108 | return; |
1096 | } | | 1109 | } |
1097 | | | 1110 | |
1098 | printf("Supported file systems:"); | | 1111 | printf("Supported file systems:"); |
1099 | mutex_enter(&vfs_list_lock); | | 1112 | mutex_enter(&vfs_list_lock); |
1100 | LIST_FOREACH(v, &vfs_list, vfs_list) { | | 1113 | LIST_FOREACH(v, &vfs_list, vfs_list) { |
1101 | printf(" %s", v->vfs_name); | | 1114 | printf(" %s", v->vfs_name); |
1102 | } | | 1115 | } |
1103 | mutex_exit(&vfs_list_lock); | | 1116 | mutex_exit(&vfs_list_lock); |
1104 | printf("\n"); | | 1117 | printf("\n"); |
1105 | } | | 1118 | } |
1106 | | | 1119 | |
1107 | /* | | 1120 | /* |
1108 | * Mount the root file system. If the operator didn't specify a | | 1121 | * Mount the root file system. If the operator didn't specify a |
1109 | * file system to use, try all possible file systems until one | | 1122 | * file system to use, try all possible file systems until one |
1110 | * succeeds. | | 1123 | * succeeds. |
1111 | */ | | 1124 | */ |
1112 | int | | 1125 | int |
1113 | vfs_mountroot(void) | | 1126 | vfs_mountroot(void) |
1114 | { | | 1127 | { |
1115 | struct vfsops *v; | | 1128 | struct vfsops *v; |
1116 | int error = ENODEV; | | 1129 | int error = ENODEV; |
1117 | | | 1130 | |
1118 | if (root_device == NULL) | | 1131 | if (root_device == NULL) |
1119 | panic("vfs_mountroot: root device unknown"); | | 1132 | panic("vfs_mountroot: root device unknown"); |
1120 | | | 1133 | |
1121 | switch (device_class(root_device)) { | | 1134 | switch (device_class(root_device)) { |
1122 | case DV_IFNET: | | 1135 | case DV_IFNET: |
1123 | if (rootdev != NODEV) | | 1136 | if (rootdev != NODEV) |
1124 | panic("vfs_mountroot: rootdev set for DV_IFNET " | | 1137 | panic("vfs_mountroot: rootdev set for DV_IFNET " |
1125 | "(0x%llx -> %llu,%llu)", | | 1138 | "(0x%llx -> %llu,%llu)", |
1126 | (unsigned long long)rootdev, | | 1139 | (unsigned long long)rootdev, |
1127 | (unsigned long long)major(rootdev), | | 1140 | (unsigned long long)major(rootdev), |
1128 | (unsigned long long)minor(rootdev)); | | 1141 | (unsigned long long)minor(rootdev)); |
1129 | break; | | 1142 | break; |
1130 | | | 1143 | |
1131 | case DV_DISK: | | 1144 | case DV_DISK: |
1132 | if (rootdev == NODEV) | | 1145 | if (rootdev == NODEV) |
1133 | panic("vfs_mountroot: rootdev not set for DV_DISK"); | | 1146 | panic("vfs_mountroot: rootdev not set for DV_DISK"); |
1134 | if (bdevvp(rootdev, &rootvp)) | | 1147 | if (bdevvp(rootdev, &rootvp)) |
1135 | panic("vfs_mountroot: can't get vnode for rootdev"); | | 1148 | panic("vfs_mountroot: can't get vnode for rootdev"); |
1136 | error = VOP_OPEN(rootvp, FREAD, FSCRED); | | 1149 | error = VOP_OPEN(rootvp, FREAD, FSCRED); |
1137 | if (error) { | | 1150 | if (error) { |
1138 | printf("vfs_mountroot: can't open root device\n"); | | 1151 | printf("vfs_mountroot: can't open root device\n"); |
1139 | return (error); | | 1152 | return (error); |
1140 | } | | 1153 | } |
1141 | break; | | 1154 | break; |
1142 | | | 1155 | |
1143 | case DV_VIRTUAL: | | 1156 | case DV_VIRTUAL: |
1144 | break; | | 1157 | break; |
1145 | | | 1158 | |
1146 | default: | | 1159 | default: |
1147 | printf("%s: inappropriate for root file system\n", | | 1160 | printf("%s: inappropriate for root file system\n", |
1148 | device_xname(root_device)); | | 1161 | device_xname(root_device)); |
1149 | return (ENODEV); | | 1162 | return (ENODEV); |
1150 | } | | 1163 | } |
1151 | | | 1164 | |
1152 | /* | | 1165 | /* |
1153 | * If user specified a root fs type, use it. Make sure the | | 1166 | * If user specified a root fs type, use it. Make sure the |
1154 | * specified type exists and has a mount_root() | | 1167 | * specified type exists and has a mount_root() |
1155 | */ | | 1168 | */ |
1156 | if (strcmp(rootfstype, ROOT_FSTYPE_ANY) != 0) { | | 1169 | if (strcmp(rootfstype, ROOT_FSTYPE_ANY) != 0) { |
1157 | v = vfs_getopsbyname(rootfstype); | | 1170 | v = vfs_getopsbyname(rootfstype); |
1158 | error = EFTYPE; | | 1171 | error = EFTYPE; |
1159 | if (v != NULL) { | | 1172 | if (v != NULL) { |
1160 | if (v->vfs_mountroot != NULL) { | | 1173 | if (v->vfs_mountroot != NULL) { |
1161 | error = (v->vfs_mountroot)(); | | 1174 | error = (v->vfs_mountroot)(); |
1162 | } | | 1175 | } |
1163 | v->vfs_refcount--; | | 1176 | v->vfs_refcount--; |
1164 | } | | 1177 | } |
1165 | goto done; | | 1178 | goto done; |
1166 | } | | 1179 | } |
1167 | | | 1180 | |
1168 | /* | | 1181 | /* |
1169 | * Try each file system currently configured into the kernel. | | 1182 | * Try each file system currently configured into the kernel. |
1170 | */ | | 1183 | */ |
1171 | mutex_enter(&vfs_list_lock); | | 1184 | mutex_enter(&vfs_list_lock); |
1172 | LIST_FOREACH(v, &vfs_list, vfs_list) { | | 1185 | LIST_FOREACH(v, &vfs_list, vfs_list) { |
1173 | if (v->vfs_mountroot == NULL) | | 1186 | if (v->vfs_mountroot == NULL) |
1174 | continue; | | 1187 | continue; |
1175 | #ifdef DEBUG | | 1188 | #ifdef DEBUG |
1176 | aprint_normal("mountroot: trying %s...\n", v->vfs_name); | | 1189 | aprint_normal("mountroot: trying %s...\n", v->vfs_name); |
1177 | #endif | | 1190 | #endif |
1178 | v->vfs_refcount++; | | 1191 | v->vfs_refcount++; |
1179 | mutex_exit(&vfs_list_lock); | | 1192 | mutex_exit(&vfs_list_lock); |
1180 | error = (*v->vfs_mountroot)(); | | 1193 | error = (*v->vfs_mountroot)(); |
1181 | mutex_enter(&vfs_list_lock); | | 1194 | mutex_enter(&vfs_list_lock); |
1182 | v->vfs_refcount--; | | 1195 | v->vfs_refcount--; |
1183 | if (!error) { | | 1196 | if (!error) { |
1184 | aprint_normal("root file system type: %s\n", | | 1197 | aprint_normal("root file system type: %s\n", |
1185 | v->vfs_name); | | 1198 | v->vfs_name); |
1186 | break; | | 1199 | break; |
1187 | } | | 1200 | } |
1188 | } | | 1201 | } |
1189 | mutex_exit(&vfs_list_lock); | | 1202 | mutex_exit(&vfs_list_lock); |
1190 | | | 1203 | |
1191 | if (v == NULL) { | | 1204 | if (v == NULL) { |
1192 | vfs_print_fstypes(); | | 1205 | vfs_print_fstypes(); |
1193 | printf("no file system for %s", device_xname(root_device)); | | 1206 | printf("no file system for %s", device_xname(root_device)); |
1194 | if (device_class(root_device) == DV_DISK) | | 1207 | if (device_class(root_device) == DV_DISK) |
1195 | printf(" (dev 0x%llx)", (unsigned long long)rootdev); | | 1208 | printf(" (dev 0x%llx)", (unsigned long long)rootdev); |
1196 | printf("\n"); | | 1209 | printf("\n"); |
1197 | error = EFTYPE; | | 1210 | error = EFTYPE; |
1198 | } | | 1211 | } |
1199 | | | 1212 | |
1200 | done: | | 1213 | done: |
1201 | if (error && device_class(root_device) == DV_DISK) { | | 1214 | if (error && device_class(root_device) == DV_DISK) { |
1202 | VOP_CLOSE(rootvp, FREAD, FSCRED); | | 1215 | VOP_CLOSE(rootvp, FREAD, FSCRED); |
1203 | vrele(rootvp); | | 1216 | vrele(rootvp); |
1204 | } | | 1217 | } |
1205 | if (error == 0) { | | 1218 | if (error == 0) { |
1206 | struct mount *mp; | | 1219 | struct mount *mp; |
1207 | extern struct cwdinfo cwdi0; | | 1220 | extern struct cwdinfo cwdi0; |
1208 | | | 1221 | |
1209 | mp = TAILQ_FIRST(&mountlist); | | 1222 | mp = TAILQ_FIRST(&mountlist); |
1210 | mp->mnt_flag |= MNT_ROOTFS; | | 1223 | mp->mnt_flag |= MNT_ROOTFS; |
1211 | mp->mnt_op->vfs_refcount++; | | 1224 | mp->mnt_op->vfs_refcount++; |
1212 | error = fstrans_mount(mp); | | 1225 | error = fstrans_mount(mp); |
1213 | KASSERT(error == 0); | | 1226 | KASSERT(error == 0); |
1214 | | | 1227 | |
1215 | /* | | 1228 | /* |
1216 | * Get the vnode for '/'. Set cwdi0.cwdi_cdir to | | 1229 | * Get the vnode for '/'. Set cwdi0.cwdi_cdir to |
1217 | * reference it. | | 1230 | * reference it. |
1218 | */ | | 1231 | */ |
1219 | error = VFS_ROOT(mp, &rootvnode); | | 1232 | error = VFS_ROOT(mp, &rootvnode); |
1220 | if (error) | | 1233 | if (error) |
1221 | panic("cannot find root vnode, error=%d", error); | | 1234 | panic("cannot find root vnode, error=%d", error); |
1222 | cwdi0.cwdi_cdir = rootvnode; | | 1235 | cwdi0.cwdi_cdir = rootvnode; |
1223 | vref(cwdi0.cwdi_cdir); | | 1236 | vref(cwdi0.cwdi_cdir); |
1224 | VOP_UNLOCK(rootvnode); | | 1237 | VOP_UNLOCK(rootvnode); |
1225 | cwdi0.cwdi_rdir = NULL; | | 1238 | cwdi0.cwdi_rdir = NULL; |
1226 | | | 1239 | |
1227 | /* | | 1240 | /* |
1228 | * Now that root is mounted, we can fixup initproc's CWD | | 1241 | * Now that root is mounted, we can fixup initproc's CWD |
1229 | * info. All other processes are kthreads, which merely | | 1242 | * info. All other processes are kthreads, which merely |
1230 | * share proc0's CWD info. | | 1243 | * share proc0's CWD info. |
1231 | */ | | 1244 | */ |
1232 | initproc->p_cwdi->cwdi_cdir = rootvnode; | | 1245 | initproc->p_cwdi->cwdi_cdir = rootvnode; |
1233 | vref(initproc->p_cwdi->cwdi_cdir); | | 1246 | vref(initproc->p_cwdi->cwdi_cdir); |
1234 | initproc->p_cwdi->cwdi_rdir = NULL; | | 1247 | initproc->p_cwdi->cwdi_rdir = NULL; |
1235 | /* | | 1248 | /* |
1236 | * Enable loading of modules from the filesystem | | 1249 | * Enable loading of modules from the filesystem |
1237 | */ | | 1250 | */ |
1238 | module_load_vfs_init(); | | 1251 | module_load_vfs_init(); |
1239 | | | 1252 | |
1240 | } | | 1253 | } |
1241 | return (error); | | 1254 | return (error); |
1242 | } | | 1255 | } |
1243 | | | 1256 | |
1244 | /* | | 1257 | /* |
1245 | * mount_specific_key_create -- | | 1258 | * mount_specific_key_create -- |
1246 | * Create a key for subsystem mount-specific data. | | 1259 | * Create a key for subsystem mount-specific data. |
1247 | */ | | 1260 | */ |
1248 | int | | 1261 | int |
1249 | mount_specific_key_create(specificdata_key_t *keyp, specificdata_dtor_t dtor) | | 1262 | mount_specific_key_create(specificdata_key_t *keyp, specificdata_dtor_t dtor) |
1250 | { | | 1263 | { |
1251 | | | 1264 | |
1252 | return specificdata_key_create(mount_specificdata_domain, keyp, dtor); | | 1265 | return specificdata_key_create(mount_specificdata_domain, keyp, dtor); |
1253 | } | | 1266 | } |
1254 | | | 1267 | |
1255 | /* | | 1268 | /* |
1256 | * mount_specific_key_delete -- | | 1269 | * mount_specific_key_delete -- |
1257 | * Delete a key for subsystem mount-specific data. | | 1270 | * Delete a key for subsystem mount-specific data. |
1258 | */ | | 1271 | */ |
1259 | void | | 1272 | void |
1260 | mount_specific_key_delete(specificdata_key_t key) | | 1273 | mount_specific_key_delete(specificdata_key_t key) |
1261 | { | | 1274 | { |
1262 | | | 1275 | |
1263 | specificdata_key_delete(mount_specificdata_domain, key); | | 1276 | specificdata_key_delete(mount_specificdata_domain, key); |
1264 | } | | 1277 | } |
1265 | | | 1278 | |
1266 | /* | | 1279 | /* |
1267 | * mount_initspecific -- | | 1280 | * mount_initspecific -- |
1268 | * Initialize a mount's specificdata container. | | 1281 | * Initialize a mount's specificdata container. |
1269 | */ | | 1282 | */ |
1270 | void | | 1283 | void |
1271 | mount_initspecific(struct mount *mp) | | 1284 | mount_initspecific(struct mount *mp) |
1272 | { | | 1285 | { |
1273 | int error __diagused; | | 1286 | int error __diagused; |
1274 | | | 1287 | |
1275 | error = specificdata_init(mount_specificdata_domain, | | 1288 | error = specificdata_init(mount_specificdata_domain, |
1276 | &mp->mnt_specdataref); | | 1289 | &mp->mnt_specdataref); |
1277 | KASSERT(error == 0); | | 1290 | KASSERT(error == 0); |
1278 | } | | 1291 | } |
1279 | | | 1292 | |
1280 | /* | | 1293 | /* |
1281 | * mount_finispecific -- | | 1294 | * mount_finispecific -- |
1282 | * Finalize a mount's specificdata container. | | 1295 | * Finalize a mount's specificdata container. |
1283 | */ | | 1296 | */ |
1284 | void | | 1297 | void |
1285 | mount_finispecific(struct mount *mp) | | 1298 | mount_finispecific(struct mount *mp) |
1286 | { | | 1299 | { |
1287 | | | 1300 | |
1288 | specificdata_fini(mount_specificdata_domain, &mp->mnt_specdataref); | | 1301 | specificdata_fini(mount_specificdata_domain, &mp->mnt_specdataref); |
1289 | } | | 1302 | } |
1290 | | | 1303 | |
1291 | /* | | 1304 | /* |
1292 | * mount_getspecific -- | | 1305 | * mount_getspecific -- |
1293 | * Return mount-specific data corresponding to the specified key. | | 1306 | * Return mount-specific data corresponding to the specified key. |
1294 | */ | | 1307 | */ |
1295 | void * | | 1308 | void * |
1296 | mount_getspecific(struct mount *mp, specificdata_key_t key) | | 1309 | mount_getspecific(struct mount *mp, specificdata_key_t key) |
1297 | { | | 1310 | { |
1298 | | | 1311 | |
1299 | return specificdata_getspecific(mount_specificdata_domain, | | 1312 | return specificdata_getspecific(mount_specificdata_domain, |
1300 | &mp->mnt_specdataref, key); | | 1313 | &mp->mnt_specdataref, key); |
1301 | } | | 1314 | } |
1302 | | | 1315 | |
1303 | /* | | 1316 | /* |
1304 | * mount_setspecific -- | | 1317 | * mount_setspecific -- |
1305 | * Set mount-specific data corresponding to the specified key. | | 1318 | * Set mount-specific data corresponding to the specified key. |
1306 | */ | | 1319 | */ |
1307 | void | | 1320 | void |
1308 | mount_setspecific(struct mount *mp, specificdata_key_t key, void *data) | | 1321 | mount_setspecific(struct mount *mp, specificdata_key_t key, void *data) |
1309 | { | | 1322 | { |
1310 | | | 1323 | |
1311 | specificdata_setspecific(mount_specificdata_domain, | | 1324 | specificdata_setspecific(mount_specificdata_domain, |
1312 | &mp->mnt_specdataref, key, data); | | 1325 | &mp->mnt_specdataref, key, data); |
1313 | } | | 1326 | } |
1314 | | | 1327 | |
1315 | /* | | 1328 | /* |
1316 | * Check to see if a filesystem is mounted on a block device. | | 1329 | * Check to see if a filesystem is mounted on a block device. |
1317 | */ | | 1330 | */ |
1318 | int | | 1331 | int |
1319 | vfs_mountedon(vnode_t *vp) | | 1332 | vfs_mountedon(vnode_t *vp) |
1320 | { | | 1333 | { |
1321 | vnode_t *vq; | | 1334 | vnode_t *vq; |
1322 | int error = 0; | | 1335 | int error = 0; |
1323 | | | 1336 | |
1324 | if (vp->v_type != VBLK) | | 1337 | if (vp->v_type != VBLK) |
1325 | return ENOTBLK; | | 1338 | return ENOTBLK; |
1326 | if (spec_node_getmountedfs(vp) != NULL) | | 1339 | if (spec_node_getmountedfs(vp) != NULL) |
1327 | return EBUSY; | | 1340 | return EBUSY; |
1328 | if (spec_node_lookup_by_dev(vp->v_type, vp->v_rdev, &vq) == 0) { | | 1341 | if (spec_node_lookup_by_dev(vp->v_type, vp->v_rdev, &vq) == 0) { |
1329 | if (spec_node_getmountedfs(vq) != NULL) | | 1342 | if (spec_node_getmountedfs(vq) != NULL) |
1330 | error = EBUSY; | | 1343 | error = EBUSY; |
1331 | vrele(vq); | | 1344 | vrele(vq); |
1332 | } | | 1345 | } |
1333 | | | 1346 | |
1334 | return error; | | 1347 | return error; |
1335 | } | | 1348 | } |
1336 | | | 1349 | |
1337 | /* | | 1350 | /* |
1338 | * Check if a device pointed to by vp is mounted. | | 1351 | * Check if a device pointed to by vp is mounted. |
1339 | * | | 1352 | * |
1340 | * Returns: | | 1353 | * Returns: |
1341 | * EINVAL if it's not a disk | | 1354 | * EINVAL if it's not a disk |
1342 | * EBUSY if it's a disk and mounted | | 1355 | * EBUSY if it's a disk and mounted |
1343 | * 0 if it's a disk and not mounted | | 1356 | * 0 if it's a disk and not mounted |
1344 | */ | | 1357 | */ |
1345 | int | | 1358 | int |
1346 | rawdev_mounted(vnode_t *vp, vnode_t **bvpp) | | 1359 | rawdev_mounted(vnode_t *vp, vnode_t **bvpp) |
1347 | { | | 1360 | { |
1348 | vnode_t *bvp; | | 1361 | vnode_t *bvp; |
1349 | dev_t dev; | | 1362 | dev_t dev; |
1350 | int d_type; | | 1363 | int d_type; |
1351 | | | 1364 | |
1352 | bvp = NULL; | | 1365 | bvp = NULL; |
1353 | d_type = D_OTHER; | | 1366 | d_type = D_OTHER; |
1354 | | | 1367 | |
1355 | if (iskmemvp(vp)) | | 1368 | if (iskmemvp(vp)) |
1356 | return EINVAL; | | 1369 | return EINVAL; |
1357 | | | 1370 | |
1358 | switch (vp->v_type) { | | 1371 | switch (vp->v_type) { |
1359 | case VCHR: { | | 1372 | case VCHR: { |
1360 | const struct cdevsw *cdev; | | 1373 | const struct cdevsw *cdev; |
1361 | | | 1374 | |
1362 | dev = vp->v_rdev; | | 1375 | dev = vp->v_rdev; |
1363 | cdev = cdevsw_lookup(dev); | | 1376 | cdev = cdevsw_lookup(dev); |
1364 | if (cdev != NULL) { | | 1377 | if (cdev != NULL) { |
1365 | dev_t blkdev; | | 1378 | dev_t blkdev; |
1366 | | | 1379 | |
1367 | blkdev = devsw_chr2blk(dev); | | 1380 | blkdev = devsw_chr2blk(dev); |
1368 | if (blkdev != NODEV) { | | 1381 | if (blkdev != NODEV) { |
1369 | if (vfinddev(blkdev, VBLK, &bvp) != 0) { | | 1382 | if (vfinddev(blkdev, VBLK, &bvp) != 0) { |
1370 | d_type = (cdev->d_flag & D_TYPEMASK); | | 1383 | d_type = (cdev->d_flag & D_TYPEMASK); |
1371 | /* XXX: what if bvp disappears? */ | | 1384 | /* XXX: what if bvp disappears? */ |
1372 | vrele(bvp); | | 1385 | vrele(bvp); |
1373 | } | | 1386 | } |
1374 | } | | 1387 | } |
1375 | } | | 1388 | } |
1376 | | | 1389 | |
1377 | break; | | 1390 | break; |
1378 | } | | 1391 | } |
1379 | | | 1392 | |
1380 | case VBLK: { | | 1393 | case VBLK: { |
1381 | const struct bdevsw *bdev; | | 1394 | const struct bdevsw *bdev; |
1382 | | | 1395 | |
1383 | dev = vp->v_rdev; | | 1396 | dev = vp->v_rdev; |
1384 | bdev = bdevsw_lookup(dev); | | 1397 | bdev = bdevsw_lookup(dev); |
1385 | if (bdev != NULL) | | 1398 | if (bdev != NULL) |
1386 | d_type = (bdev->d_flag & D_TYPEMASK); | | 1399 | d_type = (bdev->d_flag & D_TYPEMASK); |
1387 | | | 1400 | |
1388 | bvp = vp; | | 1401 | bvp = vp; |
1389 | | | 1402 | |
1390 | break; | | 1403 | break; |
1391 | } | | 1404 | } |
1392 | | | 1405 | |
1393 | default: | | 1406 | default: |
1394 | break; | | 1407 | break; |
1395 | } | | 1408 | } |
1396 | | | 1409 | |
1397 | if (d_type != D_DISK) | | 1410 | if (d_type != D_DISK) |
1398 | return EINVAL; | | 1411 | return EINVAL; |
1399 | | | 1412 | |
1400 | if (bvpp != NULL) | | 1413 | if (bvpp != NULL) |
1401 | *bvpp = bvp; | | 1414 | *bvpp = bvp; |
1402 | | | 1415 | |
1403 | /* | | 1416 | /* |
1404 | * XXX: This is bogus. We should be failing the request | | 1417 | * XXX: This is bogus. We should be failing the request |
1405 | * XXX: not only if this specific slice is mounted, but | | 1418 | * XXX: not only if this specific slice is mounted, but |
1406 | * XXX: if it's on a disk with any other mounted slice. | | 1419 | * XXX: if it's on a disk with any other mounted slice. |
1407 | */ | | 1420 | */ |
1408 | if (vfs_mountedon(bvp)) | | 1421 | if (vfs_mountedon(bvp)) |
1409 | return EBUSY; | | 1422 | return EBUSY; |
1410 | | | 1423 | |
1411 | return 0; | | 1424 | return 0; |
1412 | } | | 1425 | } |
1413 | | | 1426 | |
1414 | /* | | 1427 | /* |
1415 | * Make a 'unique' number from a mount type name. | | 1428 | * Make a 'unique' number from a mount type name. |
1416 | */ | | 1429 | */ |
1417 | long | | 1430 | long |
1418 | makefstype(const char *type) | | 1431 | makefstype(const char *type) |
1419 | { | | 1432 | { |
1420 | long rv; | | 1433 | long rv; |
1421 | | | 1434 | |
1422 | for (rv = 0; *type; type++) { | | 1435 | for (rv = 0; *type; type++) { |
1423 | rv <<= 2; | | 1436 | rv <<= 2; |
1424 | rv ^= *type; | | 1437 | rv ^= *type; |
1425 | } | | 1438 | } |
1426 | return rv; | | 1439 | return rv; |
1427 | } | | 1440 | } |
1428 | | | 1441 | |
1429 | void | | 1442 | void |
1430 | mountlist_append(struct mount *mp) | | 1443 | mountlist_append(struct mount *mp) |
1431 | { | | 1444 | { |
1432 | mutex_enter(&mountlist_lock); | | 1445 | mutex_enter(&mountlist_lock); |
1433 | TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list); | | 1446 | TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list); |
1434 | mutex_exit(&mountlist_lock); | | 1447 | mutex_exit(&mountlist_lock); |
1435 | } | | 1448 | } |