mfs_vfsops.c revision 1.78
1 /* $NetBSD: mfs_vfsops.c,v 1.78 2007/03/04 06:03:46 christos 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. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 * @(#)mfs_vfsops.c 8.11 (Berkeley) 6/19/95 32 */ 33 34 #include <sys/cdefs.h> 35 __KERNEL_RCSID(0, "$NetBSD: mfs_vfsops.c,v 1.78 2007/03/04 06:03:46 christos Exp $"); 36 37 #if defined(_KERNEL_OPT) 38 #include "opt_compat_netbsd.h" 39 #endif 40 41 #include <sys/param.h> 42 #include <sys/systm.h> 43 #include <sys/sysctl.h> 44 #include <sys/time.h> 45 #include <sys/kernel.h> 46 #include <sys/proc.h> 47 #include <sys/buf.h> 48 #include <sys/bufq.h> 49 #include <sys/mount.h> 50 #include <sys/signalvar.h> 51 #include <sys/vnode.h> 52 #include <sys/malloc.h> 53 54 #include <miscfs/syncfs/syncfs.h> 55 56 #include <ufs/ufs/quota.h> 57 #include <ufs/ufs/inode.h> 58 #include <ufs/ufs/ufsmount.h> 59 #include <ufs/ufs/ufs_extern.h> 60 61 #include <ufs/ffs/fs.h> 62 #include <ufs/ffs/ffs_extern.h> 63 64 #include <ufs/mfs/mfsnode.h> 65 #include <ufs/mfs/mfs_extern.h> 66 67 void * mfs_rootbase; /* address of mini-root in kernel virtual memory */ 68 u_long mfs_rootsize; /* size of mini-root in bytes */ 69 70 static int mfs_minor; /* used for building internal dev_t */ 71 72 extern int (**mfs_vnodeop_p)(void *); 73 74 MALLOC_DEFINE(M_MFSNODE, "MFS node", "MFS vnode private part"); 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_statvfs, 95 ffs_sync, 96 ffs_vget, 97 ffs_fhtovp, 98 ffs_vptofh, 99 mfs_init, 100 mfs_reinit, 101 mfs_done, 102 NULL, 103 (int (*)(struct mount *, struct vnode *, struct timespec *)) eopnotsupp, 104 vfs_stdextattrctl, 105 vfs_stdsuspendctl, 106 mfs_vnodeopv_descs, 107 0, 108 { NULL, NULL }, 109 }; 110 VFS_ATTACH(mfs_vfsops); 111 112 SYSCTL_SETUP(sysctl_vfs_mfs_setup, "sysctl vfs.mfs subtree setup") 113 { 114 115 sysctl_createv(clog, 0, NULL, NULL, 116 CTLFLAG_PERMANENT, 117 CTLTYPE_NODE, "vfs", NULL, 118 NULL, 0, NULL, 0, 119 CTL_VFS, CTL_EOL); 120 sysctl_createv(clog, 0, NULL, NULL, 121 CTLFLAG_PERMANENT|CTLFLAG_ALIAS, 122 CTLTYPE_NODE, "mfs", 123 SYSCTL_DESCR("Memory based file system"), 124 NULL, 1, NULL, 0, 125 CTL_VFS, 3, CTL_EOL); 126 /* 127 * XXX the "1" and the "3" above could be dynamic, thereby 128 * eliminating one more instance of the "number to vfs" 129 * mapping problem, but they are in order as taken from 130 * sys/mount.h 131 */ 132 } 133 134 /* 135 * Memory based filesystem initialization. 136 */ 137 void 138 mfs_init(void) 139 { 140 #ifdef _LKM 141 malloc_type_attach(M_MFSNODE); 142 #endif 143 /* 144 * ffs_init() ensures to initialize necessary resources 145 * only once. 146 */ 147 ffs_init(); 148 } 149 150 void 151 mfs_reinit(void) 152 { 153 ffs_reinit(); 154 } 155 156 void 157 mfs_done(void) 158 { 159 /* 160 * ffs_done() ensures to free necessary resources 161 * only once, when it's no more needed. 162 */ 163 ffs_done(); 164 #ifdef _LKM 165 malloc_type_detach(M_MFSNODE); 166 #endif 167 } 168 169 /* 170 * Called by main() when mfs is going to be mounted as root. 171 */ 172 173 int 174 mfs_mountroot(void) 175 { 176 struct fs *fs; 177 struct mount *mp; 178 struct lwp *l = curlwp; /* XXX */ 179 struct ufsmount *ump; 180 struct mfsnode *mfsp; 181 int error = 0; 182 183 if ((error = vfs_rootmountalloc(MOUNT_MFS, "mfs_root", &mp))) { 184 vrele(rootvp); 185 return (error); 186 } 187 188 mfsp = malloc(sizeof *mfsp, M_MFSNODE, M_WAITOK); 189 rootvp->v_data = mfsp; 190 rootvp->v_op = mfs_vnodeop_p; 191 rootvp->v_tag = VT_MFS; 192 mfsp->mfs_baseoff = mfs_rootbase; 193 mfsp->mfs_size = mfs_rootsize; 194 mfsp->mfs_vnode = rootvp; 195 mfsp->mfs_proc = NULL; /* indicate kernel space */ 196 mfsp->mfs_shutdown = 0; 197 bufq_alloc(&mfsp->mfs_buflist, "fcfs", 0); 198 if ((error = ffs_mountfs(rootvp, mp, l)) != 0) { 199 mp->mnt_op->vfs_refcount--; 200 vfs_unbusy(mp); 201 bufq_free(mfsp->mfs_buflist); 202 free(mp, M_MOUNT); 203 free(mfsp, M_MFSNODE); 204 return (error); 205 } 206 simple_lock(&mountlist_slock); 207 CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list); 208 simple_unlock(&mountlist_slock); 209 mp->mnt_vnodecovered = NULLVP; 210 ump = VFSTOUFS(mp); 211 fs = ump->um_fs; 212 (void) copystr(mp->mnt_stat.f_mntonname, fs->fs_fsmnt, MNAMELEN - 1, 0); 213 (void)ffs_statvfs(mp, &mp->mnt_stat, l); 214 vfs_unbusy(mp); 215 return (0); 216 } 217 218 /* 219 * This is called early in boot to set the base address and size 220 * of the mini-root. 221 */ 222 int 223 mfs_initminiroot(void *base) 224 { 225 struct fs *fs = (struct fs *)((char *)base + SBLOCK_UFS1); 226 227 /* check for valid super block */ 228 if (fs->fs_magic != FS_UFS1_MAGIC || fs->fs_bsize > MAXBSIZE || 229 fs->fs_bsize < sizeof(struct fs)) 230 return (0); 231 mountroot = mfs_mountroot; 232 mfs_rootbase = base; 233 mfs_rootsize = fs->fs_fsize * fs->fs_size; 234 rootdev = makedev(255, mfs_minor); 235 mfs_minor++; 236 return (mfs_rootsize); 237 } 238 239 /* 240 * VFS Operations. 241 * 242 * mount system call 243 */ 244 /* ARGSUSED */ 245 int 246 mfs_mount(struct mount *mp, const char *path, void *data, 247 struct nameidata *ndp, struct lwp *l) 248 { 249 struct vnode *devvp; 250 struct mfs_args args; 251 struct ufsmount *ump; 252 struct fs *fs; 253 struct mfsnode *mfsp; 254 struct proc *p; 255 int flags, error; 256 257 p = l->l_proc; 258 if (mp->mnt_flag & MNT_GETARGS) { 259 struct vnode *vp; 260 261 ump = VFSTOUFS(mp); 262 if (ump == NULL) 263 return EIO; 264 265 vp = ump->um_devvp; 266 if (vp == NULL) 267 return EIO; 268 269 mfsp = VTOMFS(vp); 270 if (mfsp == NULL) 271 return EIO; 272 273 args.fspec = NULL; 274 args.base = mfsp->mfs_baseoff; 275 args.size = mfsp->mfs_size; 276 return copyout(&args, data, sizeof(args)); 277 } 278 /* 279 * XXX turn off async to avoid hangs when writing lots of data. 280 * the problem is that MFS needs to allocate pages to clean pages, 281 * so if we wait until the last minute to clean pages then there 282 * may not be any pages available to do the cleaning. 283 * ... and since the default partially-synchronous mode turns out 284 * to not be sufficient under heavy load, make it full synchronous. 285 */ 286 mp->mnt_flag &= ~MNT_ASYNC; 287 mp->mnt_flag |= MNT_SYNCHRONOUS; 288 289 error = copyin(data, (void *)&args, sizeof (struct mfs_args)); 290 if (error) 291 return (error); 292 293 /* 294 * If updating, check whether changing from read-only to 295 * read/write; if there is no device name, that's all we do. 296 */ 297 if (mp->mnt_flag & MNT_UPDATE) { 298 ump = VFSTOUFS(mp); 299 fs = ump->um_fs; 300 if (fs->fs_ronly == 0 && (mp->mnt_flag & MNT_RDONLY)) { 301 flags = WRITECLOSE; 302 if (mp->mnt_flag & MNT_FORCE) 303 flags |= FORCECLOSE; 304 error = ffs_flushfiles(mp, flags, l); 305 if (error) 306 return (error); 307 } 308 if (fs->fs_ronly && (mp->mnt_iflag & IMNT_WANTRDWR)) 309 fs->fs_ronly = 0; 310 if (args.fspec == NULL) 311 return EINVAL; 312 return (0); 313 } 314 error = getnewvnode(VT_MFS, (struct mount *)0, mfs_vnodeop_p, &devvp); 315 if (error) 316 return (error); 317 devvp->v_type = VBLK; 318 if (checkalias(devvp, makedev(255, mfs_minor), (struct mount *)0)) 319 panic("mfs_mount: dup dev"); 320 mfs_minor++; 321 mfsp = (struct mfsnode *)malloc(sizeof *mfsp, M_MFSNODE, M_WAITOK); 322 devvp->v_data = mfsp; 323 mfsp->mfs_baseoff = args.base; 324 mfsp->mfs_size = args.size; 325 mfsp->mfs_vnode = devvp; 326 mfsp->mfs_proc = p; 327 mfsp->mfs_shutdown = 0; 328 bufq_alloc(&mfsp->mfs_buflist, "fcfs", 0); 329 if ((error = ffs_mountfs(devvp, mp, l)) != 0) { 330 mfsp->mfs_shutdown = 1; 331 vrele(devvp); 332 return (error); 333 } 334 ump = VFSTOUFS(mp); 335 fs = ump->um_fs; 336 error = set_statvfs_info(path, UIO_USERSPACE, args.fspec, 337 UIO_USERSPACE, mp, l); 338 if (error) 339 return error; 340 (void)strncpy(fs->fs_fsmnt, mp->mnt_stat.f_mntonname, 341 sizeof(fs->fs_fsmnt)); 342 fs->fs_fsmnt[sizeof(fs->fs_fsmnt) - 1] = '\0'; 343 /* XXX: cleanup on error */ 344 return 0; 345 } 346 347 int mfs_pri = PWAIT | PCATCH; /* XXX prob. temp */ 348 349 /* 350 * Used to grab the process and keep it in the kernel to service 351 * memory filesystem I/O requests. 352 * 353 * Loop servicing I/O requests. 354 * Copy the requested data into or out of the memory filesystem 355 * address space. 356 */ 357 /* ARGSUSED */ 358 int 359 mfs_start(struct mount *mp, int flags, struct lwp *l) 360 { 361 struct vnode *vp = VFSTOUFS(mp)->um_devvp; 362 struct mfsnode *mfsp = VTOMFS(vp); 363 struct proc *p; 364 struct buf *bp; 365 void *base; 366 int sleepreturn = 0; 367 ksiginfoq_t kq; 368 369 base = mfsp->mfs_baseoff; 370 while (mfsp->mfs_shutdown != 1) { 371 while ((bp = BUFQ_GET(mfsp->mfs_buflist)) != NULL) { 372 mfs_doio(bp, base); 373 wakeup((void *)bp); 374 } 375 /* 376 * If a non-ignored signal is received, try to unmount. 377 * If that fails, or the filesystem is already in the 378 * process of being unmounted, clear the signal (it has been 379 * "processed"), otherwise we will loop here, as tsleep 380 * will always return EINTR/ERESTART. 381 */ 382 if (sleepreturn != 0) { 383 /* 384 * XXX Freeze syncer. Must do this before locking 385 * the mount point. See dounmount() for details. 386 */ 387 mutex_enter(&syncer_mutex); 388 if (vfs_busy(mp, LK_NOWAIT, 0) != 0) 389 mutex_exit(&syncer_mutex); 390 else if (dounmount(mp, 0, l) != 0) { 391 p = l->l_proc; 392 ksiginfo_queue_init(&kq); 393 mutex_enter(&p->p_smutex); 394 sigclearall(p, NULL, &kq); 395 mutex_exit(&p->p_smutex); 396 ksiginfo_queue_drain(&kq); 397 } 398 sleepreturn = 0; 399 continue; 400 } 401 402 sleepreturn = tsleep(vp, mfs_pri, "mfsidl", 0); 403 } 404 KASSERT(BUFQ_PEEK(mfsp->mfs_buflist) == NULL); 405 bufq_free(mfsp->mfs_buflist); 406 return (sleepreturn); 407 } 408 409 /* 410 * Get file system statistics. 411 */ 412 int 413 mfs_statvfs(struct mount *mp, struct statvfs *sbp, struct lwp *l) 414 { 415 int error; 416 417 error = ffs_statvfs(mp, sbp, l); 418 if (error) 419 return error; 420 (void)strncpy(sbp->f_fstypename, mp->mnt_op->vfs_name, 421 sizeof(sbp->f_fstypename)); 422 sbp->f_fstypename[sizeof(sbp->f_fstypename) - 1] = '\0'; 423 return 0; 424 } 425
Indexes created Fri Sep 26 20:09:58 GMT 2025