1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1992, 1993, 1995
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This code is derived from software donated to Berkeley by
8 * Jan-Simon Pendry.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 */
34
35 /*
36 * /dev/fd Filesystem
37 */
38
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/filedesc.h>
42 #include <sys/kernel.h>
43 #include <sys/jail.h>
44 #include <sys/lock.h>
45 #include <sys/mutex.h>
46 #include <sys/malloc.h>
47 #include <sys/mount.h>
48 #include <sys/proc.h>
49 #include <sys/racct.h>
50 #include <sys/resourcevar.h>
51 #include <sys/vnode.h>
52
53 #include <fs/fdescfs/fdesc.h>
54
55 static MALLOC_DEFINE(M_FDESCMNT, "fdesc_mount", "FDESC mount structure");
56
57 static vfs_cmount_t fdesc_cmount;
58 static vfs_mount_t fdesc_mount;
59 static vfs_unmount_t fdesc_unmount;
60 static vfs_statfs_t fdesc_statfs;
61 static vfs_root_t fdesc_root;
62
63 /*
64 * Compatibility shim for old mount(2) system call.
65 */
66 int
fdesc_cmount(struct mntarg * ma,void * data,uint64_t flags)67 fdesc_cmount(struct mntarg *ma, void *data, uint64_t flags)
68 {
69
70 return kernel_mount(ma, flags);
71 }
72
73 /*
74 * Mount the per-process file descriptors (/dev/fd)
75 */
76 static int
fdesc_mount(struct mount * mp)77 fdesc_mount(struct mount *mp)
78 {
79 struct fdescmount *fmp;
80 struct vnode *rvp;
81 int error;
82
83 /*
84 * Update is a no-op
85 */
86 if (mp->mnt_flag & (MNT_UPDATE | MNT_ROOTFS))
87 return (EOPNOTSUPP);
88
89 fmp = malloc(sizeof(struct fdescmount), M_FDESCMNT, M_WAITOK);
90
91 /*
92 * We need to initialize a few bits of our local mount point struct to
93 * avoid confusion in allocvp.
94 */
95 mp->mnt_data = fmp;
96 fmp->flags = 0;
97 if (vfs_getopt(mp->mnt_optnew, "linrdlnk", NULL, NULL) == 0)
98 fmp->flags |= FMNT_LINRDLNKF;
99 if (vfs_getopt(mp->mnt_optnew, "rdlnk", NULL, NULL) == 0)
100 fmp->flags |= FMNT_RDLNKF;
101 if (vfs_getopt(mp->mnt_optnew, "nodup", NULL, NULL) == 0)
102 fmp->flags |= FMNT_NODUP;
103 error = fdesc_allocvp(Froot, -1, FD_ROOT, mp, &rvp);
104 if (error) {
105 free(fmp, M_FDESCMNT);
106 mp->mnt_data = NULL;
107 return (error);
108 }
109 VN_LOCK_ASHARE(rvp);
110 rvp->v_type = VDIR;
111 rvp->v_vflag |= VV_ROOT;
112 fmp->f_root = rvp;
113 VOP_UNLOCK(rvp);
114
115 MNT_ILOCK(mp);
116 /* XXX -- don't mark as local to work around fts() problems */
117 /*mp->mnt_flag |= MNT_LOCAL;*/
118 /*
119 * Enable shared locking so that there is no contention on the root
120 * vnode. Note only root vnode enables shared locking for itself,
121 * so this end up being a nop for the rest.
122 */
123 mp->mnt_kern_flag |= MNTK_LOOKUP_SHARED | MNTK_EXTENDED_SHARED;
124 MNT_IUNLOCK(mp);
125
126 vfs_getnewfsid(mp);
127
128 vfs_mountedfrom(mp, "fdescfs");
129 return (0);
130 }
131
132 static int
fdesc_unmount(struct mount * mp,int mntflags)133 fdesc_unmount(struct mount *mp, int mntflags)
134 {
135 struct fdescmount *fmp;
136 int error, flags;
137
138 flags = 0;
139 fmp = mp->mnt_data;
140 if (mntflags & MNT_FORCE) {
141 /* The hash mutex protects the private mount flags. */
142 mtx_lock(&fdesc_hashmtx);
143 fmp->flags |= FMNT_UNMOUNTF;
144 mtx_unlock(&fdesc_hashmtx);
145 flags |= FORCECLOSE;
146 }
147
148 /*
149 * Clear out buffer cache. I don't think we
150 * ever get anything cached at this level at the
151 * moment, but who knows...
152 *
153 * There is 1 extra root vnode reference corresponding
154 * to f_root.
155 */
156 if ((error = vflush(mp, 1, flags, curthread)) != 0)
157 return (error);
158
159 /*
160 * Finally, throw away the fdescmount structure.
161 */
162 mp->mnt_data = NULL;
163 free(fmp, M_FDESCMNT);
164 return (0);
165 }
166
167 static int
fdesc_root(struct mount * mp,int flags,struct vnode ** vpp)168 fdesc_root(struct mount *mp, int flags, struct vnode **vpp)
169 {
170 struct vnode *vp;
171
172 /*
173 * Return locked reference to root.
174 */
175 vp = VFSTOFDESC(mp)->f_root;
176 vget(vp, flags | LK_RETRY);
177 *vpp = vp;
178 return (0);
179 }
180
181 static int
fdesc_statfs(struct mount * mp,struct statfs * sbp)182 fdesc_statfs(struct mount *mp, struct statfs *sbp)
183 {
184 struct thread *td;
185 struct filedesc *fdp;
186 int lim;
187 int i;
188 int last;
189 int freefd;
190 uint64_t limit;
191
192 td = curthread;
193
194 /*
195 * Compute number of free file descriptors.
196 * [ Strange results will ensue if the open file
197 * limit is ever reduced below the current number
198 * of open files... ]
199 */
200 lim = lim_cur(td, RLIMIT_NOFILE);
201 fdp = td->td_proc->p_fd;
202 FILEDESC_SLOCK(fdp);
203 limit = racct_get_limit(td->td_proc, RACCT_NOFILE);
204 if (lim > limit)
205 lim = limit;
206 last = min(fdp->fd_nfiles, lim);
207 freefd = 0;
208 for (i = fdp->fd_freefile; i < last; i++)
209 if (fdp->fd_ofiles[i].fde_file == NULL)
210 freefd++;
211
212 /*
213 * Adjust for the fact that the fdesc array may not
214 * have been fully allocated yet.
215 */
216 if (fdp->fd_nfiles < lim)
217 freefd += (lim - fdp->fd_nfiles);
218 FILEDESC_SUNLOCK(fdp);
219
220 sbp->f_flags = mp->mnt_flag & MNT_IGNORE;
221 sbp->f_bsize = DEV_BSIZE;
222 sbp->f_iosize = DEV_BSIZE;
223 sbp->f_blocks = 2; /* 1K to keep df happy */
224 sbp->f_bfree = 2;
225 sbp->f_bavail = 2;
226 sbp->f_files = lim + 1; /* Allow for "." */
227 sbp->f_ffree = freefd; /* See comments above */
228 return (0);
229 }
230
231 static struct vfsops fdesc_vfsops = {
232 .vfs_cmount = fdesc_cmount,
233 .vfs_init = fdesc_init,
234 .vfs_mount = fdesc_mount,
235 .vfs_root = fdesc_root,
236 .vfs_statfs = fdesc_statfs,
237 .vfs_uninit = fdesc_uninit,
238 .vfs_unmount = fdesc_unmount,
239 };
240
241 VFS_SET(fdesc_vfsops, fdescfs, VFCF_SYNTHETIC | VFCF_JAIL);
242