1 /* $NetBSD: mfs_vfsops.c,v 1.42 2002/10/24 16:41:00 chs Exp $ */ 2 3 /* 4 * Copyright (c) 1989, 1990, 1993, 1994 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * @(#)mfs_vfsops.c 8.11 (Berkeley) 6/19/95 36 */ 37 38 #include <sys/cdefs.h> 39 __KERNEL_RCSID(0, "$NetBSD: mfs_vfsops.c,v 1.42 2002/10/24 16:41:00 chs Exp $"); 40 41 #if defined(_KERNEL_OPT) 42 #include "opt_compat_netbsd.h" 43 #endif 44 45 #include <sys/param.h> 46 #include <sys/systm.h> 47 #include <sys/time.h> 48 #include <sys/kernel.h> 49 #include <sys/proc.h> 50 #include <sys/buf.h> 51 #include <sys/mount.h> 52 #include <sys/signalvar.h> 53 #include <sys/vnode.h> 54 #include <sys/malloc.h> 55 56 #include <miscfs/syncfs/syncfs.h> 57 58 #include <ufs/ufs/quota.h> 59 #include <ufs/ufs/inode.h> 60 #include <ufs/ufs/ufsmount.h> 61 #include <ufs/ufs/ufs_extern.h> 62 63 #include <ufs/ffs/fs.h> 64 #include <ufs/ffs/ffs_extern.h> 65 66 #include <ufs/mfs/mfsnode.h> 67 #include <ufs/mfs/mfs_extern.h> 68 69 caddr_t mfs_rootbase; /* address of mini-root in kernel virtual memory */ 70 u_long mfs_rootsize; /* size of mini-root in bytes */ 71 72 static int mfs_minor; /* used for building internal dev_t */ 73 74 extern int (**mfs_vnodeop_p) __P((void *)); 75 76 /* 77 * mfs vfs operations. 78 */ 79 80 extern const struct vnodeopv_desc mfs_vnodeop_opv_desc; 81 82 const struct vnodeopv_desc * const mfs_vnodeopv_descs[] = { 83 &mfs_vnodeop_opv_desc, 84 NULL, 85 }; 86 87 struct vfsops mfs_vfsops = { 88 MOUNT_MFS, 89 mfs_mount, 90 mfs_start, 91 ffs_unmount, 92 ufs_root, 93 ufs_quotactl, 94 mfs_statfs, 95 ffs_sync, 96 ffs_vget, 97 ffs_fhtovp, 98 ffs_vptofh, 99 mfs_init, 100 mfs_reinit, 101 mfs_done, 102 ffs_sysctl, 103 NULL, 104 ufs_check_export, 105 mfs_vnodeopv_descs, 106 }; 107 108 /* 109 * Memory based filesystem initialization. 110 */ 111 void 112 mfs_init() 113 { 114 /* 115 * ffs_init() ensures to initialize necessary resources 116 * only once. 117 */ 118 ffs_init(); 119 } 120 121 void 122 mfs_reinit() 123 { 124 ffs_reinit(); 125 } 126 127 void 128 mfs_done() 129 { 130 /* 131 * ffs_done() ensures to free necessary resources 132 * only once, when it's no more needed. 133 */ 134 ffs_done(); 135 } 136 137 /* 138 * Called by main() when mfs is going to be mounted as root. 139 */ 140 141 int 142 mfs_mountroot() 143 { 144 struct fs *fs; 145 struct mount *mp; 146 struct proc *p = curproc; /* XXX */ 147 struct ufsmount *ump; 148 struct mfsnode *mfsp; 149 int error = 0; 150 151 /* 152 * Get vnodes for rootdev. 153 */ 154 if (bdevvp(rootdev, &rootvp)) { 155 printf("mfs_mountroot: can't setup bdevvp's"); 156 return (error); 157 } 158 159 if ((error = vfs_rootmountalloc(MOUNT_MFS, "mfs_root", &mp))) { 160 vrele(rootvp); 161 return (error); 162 } 163 164 mfsp = malloc(sizeof *mfsp, M_MFSNODE, M_WAITOK); 165 rootvp->v_data = mfsp; 166 rootvp->v_op = mfs_vnodeop_p; 167 rootvp->v_tag = VT_MFS; 168 mfsp->mfs_baseoff = mfs_rootbase; 169 mfsp->mfs_size = mfs_rootsize; 170 mfsp->mfs_vnode = rootvp; 171 mfsp->mfs_proc = NULL; /* indicate kernel space */ 172 mfsp->mfs_shutdown = 0; 173 bufq_alloc(&mfsp->mfs_buflist, BUFQ_FCFS); 174 if ((error = ffs_mountfs(rootvp, mp, p)) != 0) { 175 mp->mnt_op->vfs_refcount--; 176 vfs_unbusy(mp); 177 bufq_free(&mfsp->mfs_buflist); 178 free(mp, M_MOUNT); 179 free(mfsp, M_MFSNODE); 180 vrele(rootvp); 181 return (error); 182 } 183 simple_lock(&mountlist_slock); 184 CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list); 185 simple_unlock(&mountlist_slock); 186 mp->mnt_vnodecovered = NULLVP; 187 ump = VFSTOUFS(mp); 188 fs = ump->um_fs; 189 (void) copystr(mp->mnt_stat.f_mntonname, fs->fs_fsmnt, MNAMELEN - 1, 0); 190 (void)ffs_statfs(mp, &mp->mnt_stat, p); 191 vfs_unbusy(mp); 192 inittodr((time_t)0); 193 return (0); 194 } 195 196 /* 197 * This is called early in boot to set the base address and size 198 * of the mini-root. 199 */ 200 int 201 mfs_initminiroot(base) 202 caddr_t base; 203 { 204 struct fs *fs = (struct fs *)(base + SBOFF); 205 206 /* check for valid super block */ 207 if (fs->fs_magic != FS_MAGIC || fs->fs_bsize > MAXBSIZE || 208 fs->fs_bsize < sizeof(struct fs)) 209 return (0); 210 mountroot = mfs_mountroot; 211 mfs_rootbase = base; 212 mfs_rootsize = fs->fs_fsize * fs->fs_size; 213 rootdev = makedev(255, mfs_minor); 214 mfs_minor++; 215 return (mfs_rootsize); 216 } 217 218 /* 219 * VFS Operations. 220 * 221 * mount system call 222 */ 223 /* ARGSUSED */ 224 int 225 mfs_mount(mp, path, data, ndp, p) 226 struct mount *mp; 227 const char *path; 228 void *data; 229 struct nameidata *ndp; 230 struct proc *p; 231 { 232 struct vnode *devvp; 233 struct mfs_args args; 234 struct ufsmount *ump; 235 struct fs *fs; 236 struct mfsnode *mfsp; 237 size_t size; 238 int flags, error; 239 240 if (mp->mnt_flag & MNT_GETARGS) { 241 struct vnode *vp; 242 struct mfsnode *mfsp; 243 244 ump = VFSTOUFS(mp); 245 if (ump == NULL) 246 return EIO; 247 248 vp = ump->um_devvp; 249 if (vp == NULL) 250 return EIO; 251 252 mfsp = VTOMFS(vp); 253 if (mfsp == NULL) 254 return EIO; 255 256 args.fspec = NULL; 257 vfs_showexport(mp, &args.export, &ump->um_export); 258 args.base = mfsp->mfs_baseoff; 259 args.size = mfsp->mfs_size; 260 return copyout(&args, data, sizeof(args)); 261 } 262 /* 263 * XXX turn off async to avoid hangs when writing lots of data. 264 * the problem is that MFS needs to allocate pages to clean pages, 265 * so if we wait until the last minute to clean pages then there 266 * may not be any pages available to do the cleaning. 267 * ... and since the default partially-synchronous mode turns out 268 * to not be sufficient under heavy load, make it full synchronous. 269 */ 270 mp->mnt_flag &= ~MNT_ASYNC; 271 mp->mnt_flag |= MNT_SYNCHRONOUS; 272 273 error = copyin(data, (caddr_t)&args, sizeof (struct mfs_args)); 274 if (error) 275 return (error); 276 277 /* 278 * If updating, check whether changing from read-only to 279 * read/write; if there is no device name, that's all we do. 280 */ 281 if (mp->mnt_flag & MNT_UPDATE) { 282 ump = VFSTOUFS(mp); 283 fs = ump->um_fs; 284 if (fs->fs_ronly == 0 && (mp->mnt_flag & MNT_RDONLY)) { 285 flags = WRITECLOSE; 286 if (mp->mnt_flag & MNT_FORCE) 287 flags |= FORCECLOSE; 288 error = ffs_flushfiles(mp, flags, p); 289 if (error) 290 return (error); 291 } 292 if (fs->fs_ronly && (mp->mnt_flag & MNT_WANTRDWR)) 293 fs->fs_ronly = 0; 294 if (args.fspec == 0) 295 return (vfs_export(mp, &ump->um_export, &args.export)); 296 return (0); 297 } 298 error = getnewvnode(VT_MFS, (struct mount *)0, mfs_vnodeop_p, &devvp); 299 if (error) 300 return (error); 301 devvp->v_type = VBLK; 302 if (checkalias(devvp, makedev(255, mfs_minor), (struct mount *)0)) 303 panic("mfs_mount: dup dev"); 304 mfs_minor++; 305 mfsp = (struct mfsnode *)malloc(sizeof *mfsp, M_MFSNODE, M_WAITOK); 306 devvp->v_data = mfsp; 307 mfsp->mfs_baseoff = args.base; 308 mfsp->mfs_size = args.size; 309 mfsp->mfs_vnode = devvp; 310 mfsp->mfs_proc = p; 311 mfsp->mfs_shutdown = 0; 312 bufq_alloc(&mfsp->mfs_buflist, BUFQ_FCFS); 313 if ((error = ffs_mountfs(devvp, mp, p)) != 0) { 314 mfsp->mfs_shutdown = 1; 315 vrele(devvp); 316 return (error); 317 } 318 ump = VFSTOUFS(mp); 319 fs = ump->um_fs; 320 (void) copyinstr(path, fs->fs_fsmnt, sizeof(fs->fs_fsmnt) - 1, &size); 321 memset(fs->fs_fsmnt + size, 0, sizeof(fs->fs_fsmnt) - size); 322 memcpy(mp->mnt_stat.f_mntonname, fs->fs_fsmnt, MNAMELEN); 323 (void) copyinstr(args.fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, 324 &size); 325 memset(mp->mnt_stat.f_mntfromname + size, 0, MNAMELEN - size); 326 return (0); 327 } 328 329 int mfs_pri = PWAIT | PCATCH; /* XXX prob. temp */ 330 331 /* 332 * Used to grab the process and keep it in the kernel to service 333 * memory filesystem I/O requests. 334 * 335 * Loop servicing I/O requests. 336 * Copy the requested data into or out of the memory filesystem 337 * address space. 338 */ 339 /* ARGSUSED */ 340 int 341 mfs_start(mp, flags, p) 342 struct mount *mp; 343 int flags; 344 struct proc *p; 345 { 346 struct vnode *vp = VFSTOUFS(mp)->um_devvp; 347 struct mfsnode *mfsp = VTOMFS(vp); 348 struct buf *bp; 349 caddr_t base; 350 int sleepreturn = 0; 351 352 base = mfsp->mfs_baseoff; 353 while (mfsp->mfs_shutdown != 1) { 354 while ((bp = BUFQ_GET(&mfsp->mfs_buflist)) != NULL) { 355 mfs_doio(bp, base); 356 wakeup((caddr_t)bp); 357 } 358 /* 359 * If a non-ignored signal is received, try to unmount. 360 * If that fails, or the filesystem is already in the 361 * process of being unmounted, clear the signal (it has been 362 * "processed"), otherwise we will loop here, as tsleep 363 * will always return EINTR/ERESTART. 364 */ 365 if (sleepreturn != 0) { 366 /* 367 * XXX Freeze syncer. Must do this before locking 368 * the mount point. See dounmount() for details. 369 */ 370 lockmgr(&syncer_lock, LK_EXCLUSIVE, NULL); 371 if (vfs_busy(mp, LK_NOWAIT, 0) != 0) 372 lockmgr(&syncer_lock, LK_RELEASE, NULL); 373 else if (dounmount(mp, 0, p) != 0) 374 CLRSIG(p, CURSIG(p)); 375 sleepreturn = 0; 376 continue; 377 } 378 379 sleepreturn = tsleep(vp, mfs_pri, "mfsidl", 0); 380 } 381 KASSERT(BUFQ_PEEK(&mfsp->mfs_buflist) == NULL); 382 bufq_free(&mfsp->mfs_buflist); 383 return (sleepreturn); 384 } 385 386 /* 387 * Get file system statistics. 388 */ 389 int 390 mfs_statfs(mp, sbp, p) 391 struct mount *mp; 392 struct statfs *sbp; 393 struct proc *p; 394 { 395 int error; 396 397 error = ffs_statfs(mp, sbp, p); 398 #ifdef COMPAT_09 399 sbp->f_type = 3; 400 #else 401 sbp->f_type = 0; 402 #endif 403 strncpy(&sbp->f_fstypename[0], mp->mnt_op->vfs_name, MFSNAMELEN); 404 return (error); 405 } 406