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 * @(#)fdesc_vfsops.c 8.4 (Berkeley) 1/21/94
35 */
36
37 /*
38 * /dev/fd Filesystem
39 */
40
41 #include <sys/param.h>
42 #include <sys/systm.h>
43 #include <sys/filedesc.h>
44 #include <sys/kernel.h>
45 #include <sys/jail.h>
46 #include <sys/lock.h>
47 #include <sys/mutex.h>
48 #include <sys/malloc.h>
49 #include <sys/mount.h>
50 #include <sys/proc.h>
51 #include <sys/racct.h>
52 #include <sys/resourcevar.h>
53 #include <sys/vnode.h>
54
55 #include <fs/fdescfs/fdesc.h>
56
57 static MALLOC_DEFINE(M_FDESCMNT, "fdesc_mount", "FDESC mount structure");
58
59 static vfs_cmount_t fdesc_cmount;
60 static vfs_mount_t fdesc_mount;
61 static vfs_unmount_t fdesc_unmount;
62 static vfs_statfs_t fdesc_statfs;
63 static vfs_root_t fdesc_root;
64
65 /*
66 * Compatibility shim for old mount(2) system call.
67 */
68 int
fdesc_cmount(struct mntarg * ma,void * data,uint64_t flags)69 fdesc_cmount(struct mntarg *ma, void *data, uint64_t flags)
70 {
71
72 return kernel_mount(ma, flags);
73 }
74
75 /*
76 * Mount the per-process file descriptors (/dev/fd)
77 */
78 static int
fdesc_mount(struct mount * mp)79 fdesc_mount(struct mount *mp)
80 {
81 struct fdescmount *fmp;
82 struct vnode *rvp;
83 int error;
84
85 /*
86 * Update is a no-op
87 */
88 if (mp->mnt_flag & (MNT_UPDATE | MNT_ROOTFS))
89 return (EOPNOTSUPP);
90
91 fmp = malloc(sizeof(struct fdescmount), M_FDESCMNT, M_WAITOK);
92
93 /*
94 * We need to initialize a few bits of our local mount point struct to
95 * avoid confusion in allocvp.
96 */
97 mp->mnt_data = fmp;
98 fmp->flags = 0;
99 if (vfs_getopt(mp->mnt_optnew, "linrdlnk", NULL, NULL) == 0)
100 fmp->flags |= FMNT_LINRDLNKF;
101 if (vfs_getopt(mp->mnt_optnew, "rdlnk", NULL, NULL) == 0)
102 fmp->flags |= FMNT_RDLNKF;
103 if (vfs_getopt(mp->mnt_optnew, "nodup", NULL, NULL) == 0)
104 fmp->flags |= FMNT_NODUP;
105 error = fdesc_allocvp(Froot, -1, FD_ROOT, mp, &rvp);
106 if (error) {
107 free(fmp, M_FDESCMNT);
108 mp->mnt_data = NULL;
109 return (error);
110 }
111 VN_LOCK_ASHARE(rvp);
112 rvp->v_type = VDIR;
113 rvp->v_vflag |= VV_ROOT;
114 fmp->f_root = rvp;
115 VOP_UNLOCK(rvp);
116
117 MNT_ILOCK(mp);
118 /* XXX -- don't mark as local to work around fts() problems */
119 /*mp->mnt_flag |= MNT_LOCAL;*/
120 /*
121 * Enable shared locking so that there is no contention on the root
122 * vnode. Note only root vnode enables shared locking for itself,
123 * so this end up being a nop for the rest.
124 */
125 mp->mnt_kern_flag |= MNTK_LOOKUP_SHARED | MNTK_EXTENDED_SHARED;
126 MNT_IUNLOCK(mp);
127
128 vfs_getnewfsid(mp);
129
130 vfs_mountedfrom(mp, "fdescfs");
131 return (0);
132 }
133
134 static int
fdesc_unmount(struct mount * mp,int mntflags)135 fdesc_unmount(struct mount *mp, int mntflags)
136 {
137 struct fdescmount *fmp;
138 int error, flags;
139
140 flags = 0;
141 fmp = mp->mnt_data;
142 if (mntflags & MNT_FORCE) {
143 /* The hash mutex protects the private mount flags. */
144 mtx_lock(&fdesc_hashmtx);
145 fmp->flags |= FMNT_UNMOUNTF;
146 mtx_unlock(&fdesc_hashmtx);
147 flags |= FORCECLOSE;
148 }
149
150 /*
151 * Clear out buffer cache. I don't think we
152 * ever get anything cached at this level at the
153 * moment, but who knows...
154 *
155 * There is 1 extra root vnode reference corresponding
156 * to f_root.
157 */
158 if ((error = vflush(mp, 1, flags, curthread)) != 0)
159 return (error);
160
161 /*
162 * Finally, throw away the fdescmount structure.
163 */
164 mp->mnt_data = NULL;
165 free(fmp, M_FDESCMNT);
166 return (0);
167 }
168
169 static int
fdesc_root(struct mount * mp,int flags,struct vnode ** vpp)170 fdesc_root(struct mount *mp, int flags, struct vnode **vpp)
171 {
172 struct vnode *vp;
173
174 /*
175 * Return locked reference to root.
176 */
177 vp = VFSTOFDESC(mp)->f_root;
178 vget(vp, flags | LK_RETRY);
179 *vpp = vp;
180 return (0);
181 }
182
183 static int
fdesc_statfs(struct mount * mp,struct statfs * sbp)184 fdesc_statfs(struct mount *mp, struct statfs *sbp)
185 {
186 struct thread *td;
187 struct filedesc *fdp;
188 int lim;
189 int i;
190 int last;
191 int freefd;
192 uint64_t limit;
193
194 td = curthread;
195
196 /*
197 * Compute number of free file descriptors.
198 * [ Strange results will ensue if the open file
199 * limit is ever reduced below the current number
200 * of open files... ]
201 */
202 lim = lim_cur(td, RLIMIT_NOFILE);
203 fdp = td->td_proc->p_fd;
204 FILEDESC_SLOCK(fdp);
205 limit = racct_get_limit(td->td_proc, RACCT_NOFILE);
206 if (lim > limit)
207 lim = limit;
208 last = min(fdp->fd_nfiles, lim);
209 freefd = 0;
210 for (i = fdp->fd_freefile; i < last; i++)
211 if (fdp->fd_ofiles[i].fde_file == NULL)
212 freefd++;
213
214 /*
215 * Adjust for the fact that the fdesc array may not
216 * have been fully allocated yet.
217 */
218 if (fdp->fd_nfiles < lim)
219 freefd += (lim - fdp->fd_nfiles);
220 FILEDESC_SUNLOCK(fdp);
221
222 sbp->f_flags = mp->mnt_flag & MNT_IGNORE;
223 sbp->f_bsize = DEV_BSIZE;
224 sbp->f_iosize = DEV_BSIZE;
225 sbp->f_blocks = 2; /* 1K to keep df happy */
226 sbp->f_bfree = 2;
227 sbp->f_bavail = 2;
228 sbp->f_files = lim + 1; /* Allow for "." */
229 sbp->f_ffree = freefd; /* See comments above */
230 return (0);
231 }
232
233 static struct vfsops fdesc_vfsops = {
234 .vfs_cmount = fdesc_cmount,
235 .vfs_init = fdesc_init,
236 .vfs_mount = fdesc_mount,
237 .vfs_root = fdesc_root,
238 .vfs_statfs = fdesc_statfs,
239 .vfs_uninit = fdesc_uninit,
240 .vfs_unmount = fdesc_unmount,
241 };
242
243 VFS_SET(fdesc_vfsops, fdescfs, VFCF_SYNTHETIC | VFCF_JAIL);
244