1 /* 2 * Copyright (c) 2013 The DragonFly Project. All rights reserved. 3 * 4 * This code is derived from software contributed to The DragonFly Project 5 * by Antonio Huete Jimenez <tuxillo@quantumachine.net> 6 * by Matthew Dillon <dillon@dragonflybsd.org> 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in 16 * the documentation and/or other materials provided with the 17 * distribution. 18 * 3. Neither the name of The DragonFly Project nor the names of its 19 * contributors may be used to endorse or promote products derived 20 * from this software without specific, prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 23 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 24 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 25 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 26 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 27 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, 28 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 29 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 30 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 31 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 32 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 */ 36 37 #include <sys/vfsops.h> 38 #include <sys/mount.h> 39 #include <sys/types.h> 40 #include <sys/systm.h> 41 #include <sys/param.h> 42 #include <sys/module.h> 43 #include <sys/kernel.h> 44 #include <sys/malloc.h> 45 #include <sys/sysctl.h> 46 #include <sys/queue.h> 47 #include <sys/spinlock2.h> 48 #include <sys/sysref2.h> 49 #include <sys/ktr.h> 50 51 #include <string.h> 52 53 #include "dirfs.h" 54 55 MALLOC_DEFINE(M_DIRFS, "dirfs", "dirfs mount allocation"); 56 MALLOC_DEFINE(M_DIRFS_NODE, "dirfs nodes", "dirfs nodes memory allocation"); 57 MALLOC_DEFINE(M_DIRFS_MISC, "dirfs misc", "dirfs miscellaneous allocation"); 58 59 /* 60 * Kernel tracing facilities 61 */ 62 KTR_INFO_MASTER(dirfs); 63 64 KTR_INFO(KTR_DIRFS, dirfs, root, 31, 65 "DIRFS(root dnp=%p vnode=%p hostdir=%s fd=%d error=%d)", 66 dirfs_node_t dnp, struct vnode *vp, char *hostdir, int fd, int error); 67 68 /* System wide sysctl stuff */ 69 int debuglvl = 0; 70 int dirfs_fd_limit = 100; 71 int dirfs_fd_used = 0; 72 long passive_fd_list_miss = 0; 73 long passive_fd_list_hits = 0; 74 75 SYSCTL_NODE(_vfs, OID_AUTO, dirfs, CTLFLAG_RW, 0, 76 "dirfs filesystem for vkernels"); 77 SYSCTL_INT(_vfs_dirfs, OID_AUTO, debug, CTLFLAG_RW, 78 &debuglvl, 0, "dirfs debug level"); 79 SYSCTL_INT(_vfs_dirfs, OID_AUTO, fd_limit, CTLFLAG_RW, 80 &dirfs_fd_limit, 0, "Maximum number of passive nodes to cache"); 81 SYSCTL_INT(_vfs_dirfs, OID_AUTO, fd_used, CTLFLAG_RD, 82 &dirfs_fd_used, 0, "Current number of passive nodes cached"); 83 SYSCTL_LONG(_vfs_dirfs, OID_AUTO, passive_fd_list_miss, CTLFLAG_RD, 84 &passive_fd_list_miss, 0, "Passive fd list cache misses"); 85 SYSCTL_LONG(_vfs_dirfs, OID_AUTO, passive_fd_list_hits, CTLFLAG_RD, 86 &passive_fd_list_hits, 0, "Passive fd list cache misses"); 87 88 static int dirfs_statfs(struct mount *, struct statfs *, struct ucred *); 89 90 static int 91 dirfs_mount(struct mount *mp, char *path, caddr_t data, struct ucred *cred) 92 { 93 dirfs_mount_t dmp; 94 struct stat st; 95 size_t done, nlen; 96 int error; 97 98 debug_called(); 99 100 if (mp->mnt_flag & MNT_UPDATE) { 101 dmp = VFS_TO_DIRFS(mp); 102 if (dmp->dm_rdonly == 0 && (mp->mnt_flag & MNT_RDONLY)) { 103 /* XXX We should make sure all writes are synced */ 104 dmp->dm_rdonly = 1; 105 debug(2, "dirfs read-write -> read-only\n"); 106 } 107 108 if (dmp->dm_rdonly && (mp->mnt_kern_flag & MNTK_WANTRDWR)) { 109 debug(2, "dirfs read-only -> read-write\n"); 110 dmp->dm_rdonly = 0; 111 } 112 return 0; 113 } 114 115 dmp = kmalloc(sizeof(*dmp), M_DIRFS, M_WAITOK | M_ZERO); 116 mp->mnt_data = (qaddr_t)dmp; 117 dmp->dm_mount = mp; 118 119 error = copyinstr(data, &dmp->dm_path, MAXPATHLEN, &done); 120 if (error) { 121 /* Attempt to copy from kernel address */ 122 error = copystr(data, &dmp->dm_path, MAXPATHLEN, &done); 123 if (error) { 124 kfree(dmp, M_DIRFS); 125 return error; 126 } 127 } 128 129 /* Strip / character at the end to avoid problems */ 130 nlen = strnlen(dmp->dm_path, MAXPATHLEN); 131 if (dmp->dm_path[nlen-1] == '/') 132 dmp->dm_path[nlen-1] = 0; 133 134 /* Make sure host directory exists and it is indeed a directory. */ 135 if ((stat(dmp->dm_path, &st)) == 0) { 136 if (!S_ISDIR(st.st_mode)) { 137 kfree(dmp, M_DIRFS); 138 return EINVAL; 139 } 140 } else { 141 return errno; 142 } 143 144 lockinit(&dmp->dm_lock, "dfsmnt", 0, LK_CANRECURSE); 145 146 vfs_add_vnodeops(mp, &dirfs_vnode_vops, &mp->mnt_vn_norm_ops); 147 vfs_getnewfsid(mp); 148 149 /* Who is running the vkernel */ 150 dmp->dm_uid = getuid(); 151 dmp->dm_gid = getgid(); 152 153 TAILQ_INIT(&dmp->dm_fdlist); 154 RB_INIT(&dmp->dm_inotree); 155 156 kmalloc_raise_limit(M_DIRFS_NODE, 0); 157 158 dirfs_statfs(mp, &mp->mnt_stat, cred); 159 160 dbg(5, "%s mounted. dmp=%p mp=%p\n", dmp->dm_path, dmp, mp); 161 162 return 0; 163 } 164 165 static int 166 dirfs_unmount(struct mount *mp, int mntflags) 167 { 168 dirfs_mount_t dmp; 169 dirfs_node_t dnp; 170 int cnt; 171 int error; 172 173 debug_called(); 174 cnt = 0; 175 dmp = VFS_TO_DIRFS(mp); 176 177 error = vflush(mp, 0, 0); 178 if (error) 179 return error; 180 181 /* 182 * Clean up dm_fdlist. There should be no vnodes left so the 183 * only ref should be from the fdlist. 184 */ 185 while ((dnp = TAILQ_FIRST(&dmp->dm_fdlist)) != NULL) { 186 dirfs_node_setpassive(dmp, dnp, 0); 187 } 188 189 /* 190 * Cleanup root node. In the case the filesystem is mounted 191 * but no operation is done on it, there will be no call to 192 * VFS_ROOT() so better check dnp is not NULL before attempting 193 * to release it. 194 */ 195 dnp = dmp->dm_root; 196 if (dnp != NULL) { 197 dirfs_close_helper(dnp); 198 debug_node2(dnp); 199 dirfs_node_drop(dmp, dnp); /* last ref should free structure */ 200 } 201 kfree(dmp, M_DIRFS); 202 mp->mnt_data = (qaddr_t) 0; 203 204 dbg(5, "dirfs umounted successfully\n"); 205 206 return 0; 207 } 208 209 static int 210 dirfs_root(struct mount *mp, struct vnode **vpp) 211 { 212 dirfs_mount_t dmp; 213 dirfs_node_t dnp; 214 int fd; 215 int error; 216 217 debug_called(); 218 219 dmp = VFS_TO_DIRFS(mp); 220 KKASSERT(dmp != NULL); 221 222 if (dmp->dm_root == NULL) { 223 /* 224 * dm_root holds the root dirfs node. Allocate a new one since 225 * there is none. Also attempt to lstat(2) it, in order to set 226 * data for VOP_ACCESS() 227 */ 228 dnp = dirfs_node_alloc(mp); 229 error = dirfs_node_stat(DIRFS_NOFD, dmp->dm_path, dnp); 230 if (error != 0) { 231 dirfs_node_free(dmp, dnp); 232 return error; 233 } 234 dirfs_node_ref(dnp); /* leave inact for life of mount */ 235 236 /* Root inode's parent is NULL, used for verification */ 237 dnp->dn_parent = NULL; 238 dmp->dm_root = dnp; 239 dirfs_node_setflags(dnp, DIRFS_ROOT); 240 241 /* 242 * Maintain an open descriptor on the root dnp. The 243 * normal open/close/cache does not apply for the root 244 * so the descriptor is ALWAYS available. 245 */ 246 fd = open(dmp->dm_path, O_DIRECTORY); 247 if (fd == -1) { 248 dbg(5, "failed to open ROOT node\n"); 249 dirfs_free_vp(dmp, dnp); 250 dirfs_node_free(dmp, dnp); 251 return errno; 252 } 253 dnp->dn_fd = fd; 254 dnp->dn_type = VDIR; 255 } else { 256 dnp = dmp->dm_root; 257 } 258 259 /* 260 * Acquire the root vnode (dn_type already set above). This 261 * call will handle any races and return a locked vnode. 262 */ 263 dirfs_alloc_vp(mp, vpp, LK_CANRECURSE, dnp); 264 KTR_LOG(dirfs_root, dnp, *vpp, dmp->dm_path, dnp->dn_fd, error); 265 266 return 0; 267 } 268 269 static int 270 dirfs_fhtovp(struct mount *mp, struct vnode *rootvp, struct fid *fhp, struct vnode **vpp) 271 { 272 debug_called(); 273 274 return EOPNOTSUPP; 275 } 276 277 static int 278 dirfs_statfs(struct mount *mp, struct statfs *sbp, struct ucred *cred) 279 { 280 dirfs_mount_t dmp = VFS_TO_DIRFS(mp); 281 struct statfs st; 282 283 debug_called(); 284 285 if((statfs(dmp->dm_path, &st)) == -1) 286 return errno; 287 288 ksnprintf(st.f_mntfromname, MNAMELEN - 1, "dirfs@%s", dmp->dm_path); 289 bcopy(&st, sbp, sizeof(st)); 290 strlcpy(sbp->f_fstypename, mp->mnt_vfc->vfc_name, MFSNAMELEN); 291 dbg(5, "iosize = %zd\n", sbp->f_iosize); 292 293 return 0; 294 } 295 296 static int 297 dirfs_vptofh(struct vnode *vp, struct fid *fhp) 298 { 299 dirfs_node_t dnp; 300 301 dnp = VP_TO_NODE(vp); 302 debug_node2(dnp); 303 debug_called(); 304 305 return EOPNOTSUPP; 306 } 307 308 static int 309 dirfs_checkexp(struct mount *mp, struct sockaddr *nam, int *exflagsp, 310 struct ucred **credanonp) 311 { 312 debug_called(); 313 314 return EOPNOTSUPP; 315 } 316 317 static struct vfsops dirfs_vfsops = { 318 .vfs_mount = dirfs_mount, 319 .vfs_unmount = dirfs_unmount, 320 .vfs_root = dirfs_root, 321 .vfs_vget = vfs_stdvget, 322 .vfs_statfs = dirfs_statfs, 323 .vfs_fhtovp = dirfs_fhtovp, 324 .vfs_vptofh = dirfs_vptofh, 325 .vfs_sync = vfs_stdsync, 326 .vfs_checkexp = dirfs_checkexp 327 }; 328 329 VFS_SET(dirfs_vfsops, dirfs, 0); 330 MODULE_VERSION(dirfs, 1); 331