expfs.c 12.2 KB
Newer Older
C
Christoph Hellwig 已提交
1 2 3 4 5 6 7 8 9 10
/*
 * Copyright (C) Neil Brown 2002
 * Copyright (C) Christoph Hellwig 2007
 *
 * This file contains the code mapping from inodes to NFS file handles,
 * and for mapping back from file handles to dentries.
 *
 * For details on why we do all the strange and hairy things in here
 * take a look at Documentation/filesystems/Exporting.
 */
11
#include <linux/exportfs.h>
L
Linus Torvalds 已提交
12 13 14
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/module.h>
15
#include <linux/mount.h>
L
Linus Torvalds 已提交
16 17
#include <linux/namei.h>

18
#define dprintk(fmt, args...) do{}while(0)
L
Linus Torvalds 已提交
19 20


C
Christoph Hellwig 已提交
21
static int get_name(struct vfsmount *mnt, struct dentry *dentry, char *name,
22 23 24
		struct dentry *child);


C
Christoph Hellwig 已提交
25 26
static int exportfs_get_name(struct vfsmount *mnt, struct dentry *dir,
		char *name, struct dentry *child)
27
{
28
	const struct export_operations *nop = dir->d_sb->s_export_op;
29 30 31 32

	if (nop->get_name)
		return nop->get_name(dir, name, child);
	else
C
Christoph Hellwig 已提交
33
		return get_name(mnt, dir, name, child);
34
}
L
Linus Torvalds 已提交
35

36 37 38
/*
 * Check if the dentry or any of it's aliases is acceptable.
 */
39 40 41 42 43 44 45
static struct dentry *
find_acceptable_alias(struct dentry *result,
		int (*acceptable)(void *context, struct dentry *dentry),
		void *context)
{
	struct dentry *dentry, *toput = NULL;

46 47 48
	if (acceptable(context, result))
		return result;

49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68
	spin_lock(&dcache_lock);
	list_for_each_entry(dentry, &result->d_inode->i_dentry, d_alias) {
		dget_locked(dentry);
		spin_unlock(&dcache_lock);
		if (toput)
			dput(toput);
		if (dentry != result && acceptable(context, dentry)) {
			dput(result);
			return dentry;
		}
		spin_lock(&dcache_lock);
		toput = dentry;
	}
	spin_unlock(&dcache_lock);

	if (toput)
		dput(toput);
	return NULL;
}

69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
/*
 * Find root of a disconnected subtree and return a reference to it.
 */
static struct dentry *
find_disconnected_root(struct dentry *dentry)
{
	dget(dentry);
	spin_lock(&dentry->d_lock);
	while (!IS_ROOT(dentry) &&
	       (dentry->d_parent->d_flags & DCACHE_DISCONNECTED)) {
		struct dentry *parent = dentry->d_parent;
		dget(parent);
		spin_unlock(&dentry->d_lock);
		dput(dentry);
		dentry = parent;
		spin_lock(&dentry->d_lock);
	}
	spin_unlock(&dentry->d_lock);
	return dentry;
}

90 91 92

/*
 * Make sure target_dir is fully connected to the dentry tree.
L
Linus Torvalds 已提交
93
 *
94
 * It may already be, as the flag isn't always updated when connection happens.
L
Linus Torvalds 已提交
95
 */
96
static int
C
Christoph Hellwig 已提交
97
reconnect_path(struct vfsmount *mnt, struct dentry *target_dir)
L
Linus Torvalds 已提交
98 99
{
	char nbuf[NAME_MAX+1];
100 101
	int noprogress = 0;
	int err = -ESTALE;
L
Linus Torvalds 已提交
102 103

	/*
104
	 * It is possible that a confused file system might not let us complete
L
Linus Torvalds 已提交
105 106 107 108 109 110 111
	 * the path to the root.  For example, if get_parent returns a directory
	 * in which we cannot find a name for the child.  While this implies a
	 * very sick filesystem we don't want it to cause knfsd to spin.  Hence
	 * the noprogress counter.  If we go through the loop 10 times (2 is
	 * probably enough) without getting anywhere, we just give up
	 */
	while (target_dir->d_flags & DCACHE_DISCONNECTED && noprogress++ < 10) {
112
		struct dentry *pd = find_disconnected_root(target_dir);
L
Linus Torvalds 已提交
113 114 115 116 117 118 119

		if (!IS_ROOT(pd)) {
			/* must have found a connected parent - great */
			spin_lock(&pd->d_lock);
			pd->d_flags &= ~DCACHE_DISCONNECTED;
			spin_unlock(&pd->d_lock);
			noprogress = 0;
C
Christoph Hellwig 已提交
120
		} else if (pd == mnt->mnt_sb->s_root) {
L
Linus Torvalds 已提交
121 122 123 124 125 126
			printk(KERN_ERR "export: Eeek filesystem root is not connected, impossible\n");
			spin_lock(&pd->d_lock);
			pd->d_flags &= ~DCACHE_DISCONNECTED;
			spin_unlock(&pd->d_lock);
			noprogress = 0;
		} else {
127 128 129 130 131 132 133 134 135 136 137 138 139 140
			/*
			 * We have hit the top of a disconnected path, try to
			 * find parent and connect.
			 *
			 * Racing with some other process renaming a directory
			 * isn't much of a problem here.  If someone renames
			 * the directory, it will end up properly connected,
			 * which is what we want
			 *
			 * Getting the parent can't be supported generically,
			 * the locking is too icky.
			 *
			 * Instead we just return EACCES.  If server reboots
			 * or inodes get flushed, you lose
L
Linus Torvalds 已提交
141
			 */
142
			struct dentry *ppd = ERR_PTR(-EACCES);
L
Linus Torvalds 已提交
143 144
			struct dentry *npd;

145
			mutex_lock(&pd->d_inode->i_mutex);
C
Christoph Hellwig 已提交
146 147
			if (mnt->mnt_sb->s_export_op->get_parent)
				ppd = mnt->mnt_sb->s_export_op->get_parent(pd);
148
			mutex_unlock(&pd->d_inode->i_mutex);
L
Linus Torvalds 已提交
149 150 151

			if (IS_ERR(ppd)) {
				err = PTR_ERR(ppd);
152 153
				dprintk("%s: get_parent of %ld failed, err %d\n",
					__FUNCTION__, pd->d_inode->i_ino, err);
L
Linus Torvalds 已提交
154 155 156
				dput(pd);
				break;
			}
157 158 159

			dprintk("%s: find name of %lu in %lu\n", __FUNCTION__,
				pd->d_inode->i_ino, ppd->d_inode->i_ino);
C
Christoph Hellwig 已提交
160
			err = exportfs_get_name(mnt, ppd, nbuf, pd);
L
Linus Torvalds 已提交
161 162 163 164 165 166 167 168 169 170
			if (err) {
				dput(ppd);
				dput(pd);
				if (err == -ENOENT)
					/* some race between get_parent and
					 * get_name?  just try again
					 */
					continue;
				break;
			}
171
			dprintk("%s: found name: %s\n", __FUNCTION__, nbuf);
172
			mutex_lock(&ppd->d_inode->i_mutex);
L
Linus Torvalds 已提交
173
			npd = lookup_one_len(nbuf, ppd, strlen(nbuf));
174
			mutex_unlock(&ppd->d_inode->i_mutex);
L
Linus Torvalds 已提交
175 176
			if (IS_ERR(npd)) {
				err = PTR_ERR(npd);
177 178
				dprintk("%s: lookup failed: %d\n",
					__FUNCTION__, err);
L
Linus Torvalds 已提交
179 180 181 182 183 184 185 186 187 188 189 190
				dput(ppd);
				dput(pd);
				break;
			}
			/* we didn't really want npd, we really wanted
			 * a side-effect of the lookup.
			 * hopefully, npd == pd, though it isn't really
			 * a problem if it isn't
			 */
			if (npd == pd)
				noprogress = 0;
			else
191
				printk("%s: npd != pd\n", __FUNCTION__);
L
Linus Torvalds 已提交
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
			dput(npd);
			dput(ppd);
			if (IS_ROOT(pd)) {
				/* something went wrong, we have to give up */
				dput(pd);
				break;
			}
		}
		dput(pd);
	}

	if (target_dir->d_flags & DCACHE_DISCONNECTED) {
		/* something went wrong - oh-well */
		if (!err)
			err = -ESTALE;
207
		return err;
L
Linus Torvalds 已提交
208
	}
209 210 211 212

	return 0;
}

L
Linus Torvalds 已提交
213 214 215 216 217 218 219 220 221 222 223 224 225
struct getdents_callback {
	char *name;		/* name that was found. It already points to a
				   buffer NAME_MAX+1 is size */
	unsigned long ino;	/* the inum we are looking for */
	int found;		/* inode matched? */
	int sequence;		/* sequence counter */
};

/*
 * A rather strange filldir function to capture
 * the name matching the specified inode number.
 */
static int filldir_one(void * __buf, const char * name, int len,
226
			loff_t pos, u64 ino, unsigned int d_type)
L
Linus Torvalds 已提交
227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249
{
	struct getdents_callback *buf = __buf;
	int result = 0;

	buf->sequence++;
	if (buf->ino == ino) {
		memcpy(buf->name, name, len);
		buf->name[len] = '\0';
		buf->found = 1;
		result = -1;
	}
	return result;
}

/**
 * get_name - default export_operations->get_name function
 * @dentry: the directory in which to find a name
 * @name:   a pointer to a %NAME_MAX+1 char buffer to store the name
 * @child:  the dentry for the child directory.
 *
 * calls readdir on the parent until it finds an entry with
 * the same inode number as the child, and returns that.
 */
C
Christoph Hellwig 已提交
250 251
static int get_name(struct vfsmount *mnt, struct dentry *dentry,
		char *name, struct dentry *child)
L
Linus Torvalds 已提交
252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
{
	struct inode *dir = dentry->d_inode;
	int error;
	struct file *file;
	struct getdents_callback buffer;

	error = -ENOTDIR;
	if (!dir || !S_ISDIR(dir->i_mode))
		goto out;
	error = -EINVAL;
	if (!dir->i_fop)
		goto out;
	/*
	 * Open the directory ...
	 */
C
Christoph Hellwig 已提交
267
	file = dentry_open(dget(dentry), mntget(mnt), O_RDONLY);
L
Linus Torvalds 已提交
268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
	error = PTR_ERR(file);
	if (IS_ERR(file))
		goto out;

	error = -EINVAL;
	if (!file->f_op->readdir)
		goto out_close;

	buffer.name = name;
	buffer.ino = child->d_inode->i_ino;
	buffer.found = 0;
	buffer.sequence = 0;
	while (1) {
		int old_seq = buffer.sequence;

		error = vfs_readdir(file, filldir_one, &buffer);

		if (error < 0)
			break;

		error = 0;
		if (buffer.found)
			break;
		error = -ENOENT;
		if (old_seq == buffer.sequence)
			break;
	}

out_close:
	fput(file);
out:
	return error;
}

/**
 * export_encode_fh - default export_operations->encode_fh function
 * @dentry:  the dentry to encode
 * @fh:      where to store the file handle fragment
 * @max_len: maximum length to store there
 * @connectable: whether to store parent information
 *
 * This default encode_fh function assumes that the 32 inode number
 * is suitable for locating an inode, and that the generation number
 * can be used to check that it is still valid.  It places them in the
 * filehandle fragment where export_decode_fh expects to find them.
 */
C
Christoph Hellwig 已提交
314 315
static int export_encode_fh(struct dentry *dentry, struct fid *fid,
		int *max_len, int connectable)
L
Linus Torvalds 已提交
316 317 318
{
	struct inode * inode = dentry->d_inode;
	int len = *max_len;
C
Christoph Hellwig 已提交
319
	int type = FILEID_INO32_GEN;
L
Linus Torvalds 已提交
320 321 322 323 324
	
	if (len < 2 || (connectable && len < 4))
		return 255;

	len = 2;
C
Christoph Hellwig 已提交
325 326
	fid->i32.ino = inode->i_ino;
	fid->i32.gen = inode->i_generation;
L
Linus Torvalds 已提交
327 328 329 330 331
	if (connectable && !S_ISDIR(inode->i_mode)) {
		struct inode *parent;

		spin_lock(&dentry->d_lock);
		parent = dentry->d_parent->d_inode;
C
Christoph Hellwig 已提交
332 333
		fid->i32.parent_ino = parent->i_ino;
		fid->i32.parent_gen = parent->i_generation;
L
Linus Torvalds 已提交
334 335
		spin_unlock(&dentry->d_lock);
		len = 4;
C
Christoph Hellwig 已提交
336
		type = FILEID_INO32_GEN_PARENT;
L
Linus Torvalds 已提交
337 338 339 340 341
	}
	*max_len = len;
	return type;
}

C
Christoph Hellwig 已提交
342
int exportfs_encode_fh(struct dentry *dentry, struct fid *fid, int *max_len,
343 344
		int connectable)
{
345
	const struct export_operations *nop = dentry->d_sb->s_export_op;
346
	int error;
347

348
	if (nop->encode_fh)
C
Christoph Hellwig 已提交
349
		error = nop->encode_fh(dentry, fid->raw, max_len, connectable);
350
	else
C
Christoph Hellwig 已提交
351
		error = export_encode_fh(dentry, fid, max_len, connectable);
352 353

	return error;
354 355 356
}
EXPORT_SYMBOL_GPL(exportfs_encode_fh);

C
Christoph Hellwig 已提交
357 358 359
struct dentry *exportfs_decode_fh(struct vfsmount *mnt, struct fid *fid,
		int fh_len, int fileid_type,
		int (*acceptable)(void *, struct dentry *), void *context)
360
{
361
	const struct export_operations *nop = mnt->mnt_sb->s_export_op;
C
Christoph Hellwig 已提交
362 363
	struct dentry *result, *alias;
	int err;
364

C
Christoph Hellwig 已提交
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
	/*
	 * Try to get any dentry for the given file handle from the filesystem.
	 */
	result = nop->fh_to_dentry(mnt->mnt_sb, fid, fh_len, fileid_type);
	if (!result)
		result = ERR_PTR(-ESTALE);
	if (IS_ERR(result))
		return result;

	if (S_ISDIR(result->d_inode->i_mode)) {
		/*
		 * This request is for a directory.
		 *
		 * On the positive side there is only one dentry for each
		 * directory inode.  On the negative side this implies that we
		 * to ensure our dentry is connected all the way up to the
		 * filesystem root.
		 */
		if (result->d_flags & DCACHE_DISCONNECTED) {
C
Christoph Hellwig 已提交
384
			err = reconnect_path(mnt, result);
C
Christoph Hellwig 已提交
385 386 387 388 389 390 391 392 393 394
			if (err)
				goto err_result;
		}

		if (!acceptable(context, result)) {
			err = -EACCES;
			goto err_result;
		}

		return result;
395
	} else {
C
Christoph Hellwig 已提交
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
		/*
		 * It's not a directory.  Life is a little more complicated.
		 */
		struct dentry *target_dir, *nresult;
		char nbuf[NAME_MAX+1];

		/*
		 * See if either the dentry we just got from the filesystem
		 * or any alias for it is acceptable.  This is always true
		 * if this filesystem is exported without the subtreecheck
		 * option.  If the filesystem is exported with the subtree
		 * check option there's a fair chance we need to look at
		 * the parent directory in the file handle and make sure
		 * it's connected to the filesystem root.
		 */
		alias = find_acceptable_alias(result, acceptable, context);
		if (alias)
			return alias;

		/*
		 * Try to extract a dentry for the parent directory from the
		 * file handle.  If this fails we'll have to give up.
		 */
		err = -ESTALE;
		if (!nop->fh_to_parent)
			goto err_result;

		target_dir = nop->fh_to_parent(mnt->mnt_sb, fid,
				fh_len, fileid_type);
		if (!target_dir)
			goto err_result;
		err = PTR_ERR(target_dir);
		if (IS_ERR(target_dir))
			goto err_result;

		/*
		 * And as usual we need to make sure the parent directory is
		 * connected to the filesystem root.  The VFS really doesn't
		 * like disconnected directories..
		 */
C
Christoph Hellwig 已提交
436
		err = reconnect_path(mnt, target_dir);
C
Christoph Hellwig 已提交
437 438 439 440 441 442 443 444 445 446
		if (err) {
			dput(target_dir);
			goto err_result;
		}

		/*
		 * Now that we've got both a well-connected parent and a
		 * dentry for the inode we're after, make sure that our
		 * inode is actually connected to the parent.
		 */
C
Christoph Hellwig 已提交
447
		err = exportfs_get_name(mnt, target_dir, nbuf, result);
C
Christoph Hellwig 已提交
448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
		if (!err) {
			mutex_lock(&target_dir->d_inode->i_mutex);
			nresult = lookup_one_len(nbuf, target_dir,
						 strlen(nbuf));
			mutex_unlock(&target_dir->d_inode->i_mutex);
			if (!IS_ERR(nresult)) {
				if (nresult->d_inode) {
					dput(result);
					result = nresult;
				} else
					dput(nresult);
			}
		}

		/*
		 * At this point we are done with the parent, but it's pinned
		 * by the child dentry anyway.
		 */
		dput(target_dir);

		/*
		 * And finally make sure the dentry is actually acceptable
		 * to NFSD.
		 */
		alias = find_acceptable_alias(result, acceptable, context);
		if (!alias) {
			err = -EACCES;
			goto err_result;
		}

		return alias;
479 480
	}

C
Christoph Hellwig 已提交
481 482 483
 err_result:
	dput(result);
	return ERR_PTR(err);
484 485 486
}
EXPORT_SYMBOL_GPL(exportfs_decode_fh);

L
Linus Torvalds 已提交
487
MODULE_LICENSE("GPL");