dir.c 16.9 KB
Newer Older
1
/*
2
 *  linux/fs/ext4/dir.c
3 4 5 6 7 8 9 10 11 12 13 14
 *
 * Copyright (C) 1992, 1993, 1994, 1995
 * Remy Card (card@masi.ibp.fr)
 * Laboratoire MASI - Institut Blaise Pascal
 * Universite Pierre et Marie Curie (Paris VI)
 *
 *  from
 *
 *  linux/fs/minix/dir.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
15
 *  ext4 directory handling functions
16 17 18 19 20 21 22 23 24
 *
 *  Big-endian to little-endian byte-swapping/bitmaps by
 *        David S. Miller (davem@caip.rutgers.edu), 1995
 *
 * Hash Tree Directory indexing (c) 2001  Daniel Phillips
 *
 */

#include <linux/fs.h>
25
#include <linux/jbd2.h>
26 27 28
#include <linux/buffer_head.h>
#include <linux/slab.h>
#include <linux/rbtree.h>
29
#include "ext4.h"
30

31
static unsigned char ext4_filetype_table[] = {
32 33 34
	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
};

35 36
static int ext4_dx_readdir(struct file *filp,
			   void *dirent, filldir_t filldir);
37 38 39

static unsigned char get_dtype(struct super_block *sb, int filetype)
{
40 41
	if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
	    (filetype >= EXT4_FT_MAX))
42 43
		return DT_UNKNOWN;

44
	return (ext4_filetype_table[filetype]);
45 46
}

47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66
/**
 * Check if the given dir-inode refers to an htree-indexed directory
 * (or a directory which chould potentially get coverted to use htree
 * indexing).
 *
 * Return 1 if it is a dx dir, 0 if not
 */
static int is_dx_dir(struct inode *inode)
{
	struct super_block *sb = inode->i_sb;

	if (EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
		     EXT4_FEATURE_COMPAT_DIR_INDEX) &&
	    ((ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) ||
	     ((inode->i_size >> sb->s_blocksize_bits) == 1)))
		return 1;

	return 0;
}

67 68 69 70 71
/*
 * Return 0 if the directory entry is OK, and 1 if there is a problem
 *
 * Note: this is the opposite of what ext2 and ext3 historically returned...
 */
72
int __ext4_check_dir_entry(const char *function, unsigned int line,
73
			   struct inode *dir, struct file *filp,
74 75 76
			   struct ext4_dir_entry_2 *de,
			   struct buffer_head *bh,
			   unsigned int offset)
77
{
78
	const char *error_msg = NULL;
79 80
	const int rlen = ext4_rec_len_from_disk(de->rec_len,
						dir->i_sb->s_blocksize);
81

82
	if (unlikely(rlen < EXT4_DIR_REC_LEN(1)))
83
		error_msg = "rec_len is smaller than minimal";
84
	else if (unlikely(rlen % 4 != 0))
85
		error_msg = "rec_len % 4 != 0";
86
	else if (unlikely(rlen < EXT4_DIR_REC_LEN(de->name_len)))
87
		error_msg = "rec_len is too small for name_len";
88 89
	else if (unlikely(((char *) de - bh->b_data) + rlen >
			  dir->i_sb->s_blocksize))
90
		error_msg = "directory entry across blocks";
91 92
	else if (unlikely(le32_to_cpu(de->inode) >
			le32_to_cpu(EXT4_SB(dir->i_sb)->s_es->s_inodes_count)))
93
		error_msg = "inode out of bounds";
94 95 96
	else
		return 0;

97
	if (filp)
98
		ext4_error_file(filp, function, line, bh->b_blocknr,
99 100
				"bad entry in directory: %s - offset=%u(%u), "
				"inode=%u, rec_len=%d, name_len=%d",
101
				error_msg, (unsigned) (offset % bh->b_size),
102 103 104
				offset, le32_to_cpu(de->inode),
				rlen, de->name_len);
	else
105
		ext4_error_inode(dir, function, line, bh->b_blocknr,
106 107
				"bad entry in directory: %s - offset=%u(%u), "
				"inode=%u, rec_len=%d, name_len=%d",
108
				error_msg, (unsigned) (offset % bh->b_size),
109 110 111
				offset, le32_to_cpu(de->inode),
				rlen, de->name_len);

112
	return 1;
113 114
}

115 116
static int ext4_readdir(struct file *filp,
			 void *dirent, filldir_t filldir)
117 118
{
	int error = 0;
119
	unsigned int offset;
120
	int i, stored;
121
	struct ext4_dir_entry_2 *de;
122
	int err;
123
	struct inode *inode = filp->f_path.dentry->d_inode;
124
	struct super_block *sb = inode->i_sb;
125
	int ret = 0;
126
	int dir_has_error = 0;
127

128
	if (is_dx_dir(inode)) {
129
		err = ext4_dx_readdir(filp, dirent, filldir);
130 131 132 133 134 135 136 137
		if (err != ERR_BAD_DX_DIR) {
			ret = err;
			goto out;
		}
		/*
		 * We don't set the inode dirty flag since it's not
		 * critical that it get flushed back to the disk.
		 */
138 139
		ext4_clear_inode_flag(filp->f_path.dentry->d_inode,
				      EXT4_INODE_INDEX);
140 141 142 143 144
	}
	stored = 0;
	offset = filp->f_pos & (sb->s_blocksize - 1);

	while (!error && !stored && filp->f_pos < inode->i_size) {
145
		struct ext4_map_blocks map;
146 147
		struct buffer_head *bh = NULL;

148 149 150
		map.m_lblk = filp->f_pos >> EXT4_BLOCK_SIZE_BITS(sb);
		map.m_len = 1;
		err = ext4_map_blocks(NULL, inode, &map, 0);
151
		if (err > 0) {
152
			pgoff_t index = map.m_pblk >>
153 154
					(PAGE_CACHE_SHIFT - inode->i_blkbits);
			if (!ra_has_index(&filp->f_ra, index))
155
				page_cache_sync_readahead(
156 157
					sb->s_bdev->bd_inode->i_mapping,
					&filp->f_ra, filp,
158
					index, 1);
159
			filp->f_ra.prev_pos = (loff_t)index << PAGE_CACHE_SHIFT;
160
			bh = ext4_bread(NULL, inode, map.m_lblk, 0, &err);
161 162 163 164 165 166 167
		}

		/*
		 * We ignore I/O errors on directories so users have a chance
		 * of recovering data when there's a bad sector
		 */
		if (!bh) {
168
			if (!dir_has_error) {
169 170 171
				EXT4_ERROR_FILE(filp, 0,
						"directory contains a "
						"hole at offset %llu",
172 173 174
					   (unsigned long long) filp->f_pos);
				dir_has_error = 1;
			}
175 176 177
			/* corrupt size?  Maybe no more blocks to read */
			if (filp->f_pos > inode->i_blocks << 9)
				break;
178 179 180 181 182 183 184 185 186 187 188
			filp->f_pos += sb->s_blocksize - offset;
			continue;
		}

revalidate:
		/* If the dir block has changed since the last call to
		 * readdir(2), then we might be pointing to an invalid
		 * dirent right now.  Scan from the start of the block
		 * to make sure. */
		if (filp->f_version != inode->i_version) {
			for (i = 0; i < sb->s_blocksize && i < offset; ) {
189
				de = (struct ext4_dir_entry_2 *)
190 191 192 193 194 195 196
					(bh->b_data + i);
				/* It's too expensive to do a full
				 * dirent test each time round this
				 * loop, but we do have to test at
				 * least that it is non-zero.  A
				 * failure will be detected in the
				 * dirent test below. */
197 198
				if (ext4_rec_len_from_disk(de->rec_len,
					sb->s_blocksize) < EXT4_DIR_REC_LEN(1))
199
					break;
200 201
				i += ext4_rec_len_from_disk(de->rec_len,
							    sb->s_blocksize);
202 203 204 205 206 207 208 209 210
			}
			offset = i;
			filp->f_pos = (filp->f_pos & ~(sb->s_blocksize - 1))
				| offset;
			filp->f_version = inode->i_version;
		}

		while (!error && filp->f_pos < inode->i_size
		       && offset < sb->s_blocksize) {
211
			de = (struct ext4_dir_entry_2 *) (bh->b_data + offset);
212
			if (ext4_check_dir_entry(inode, filp, de,
213
						 bh, offset)) {
214 215 216
				/*
				 * On error, skip the f_pos to the next block
				 */
217 218
				filp->f_pos = (filp->f_pos |
						(sb->s_blocksize - 1)) + 1;
219
				brelse(bh);
220 221 222
				ret = stored;
				goto out;
			}
223 224
			offset += ext4_rec_len_from_disk(de->rec_len,
					sb->s_blocksize);
225 226 227 228 229 230 231 232
			if (le32_to_cpu(de->inode)) {
				/* We might block in the next section
				 * if the data destination is
				 * currently swapped out.  So, use a
				 * version stamp to detect whether or
				 * not the directory has been modified
				 * during the copy operation.
				 */
233
				u64 version = filp->f_version;
234 235 236 237 238 239 240 241 242 243

				error = filldir(dirent, de->name,
						de->name_len,
						filp->f_pos,
						le32_to_cpu(de->inode),
						get_dtype(sb, de->file_type));
				if (error)
					break;
				if (version != filp->f_version)
					goto revalidate;
244
				stored++;
245
			}
246 247
			filp->f_pos += ext4_rec_len_from_disk(de->rec_len,
						sb->s_blocksize);
248 249
		}
		offset = 0;
250
		brelse(bh);
251 252 253 254 255
	}
out:
	return ret;
}

256 257 258 259 260 261 262 263 264
static inline int is_32bit_api(void)
{
#ifdef CONFIG_COMPAT
	return is_compat_task();
#else
	return (BITS_PER_LONG == 32);
#endif
}

265 266
/*
 * These functions convert from the major/minor hash to an f_pos
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
 * value for dx directories
 *
 * Upper layer (for example NFS) should specify FMODE_32BITHASH or
 * FMODE_64BITHASH explicitly. On the other hand, we allow ext4 to be mounted
 * directly on both 32-bit and 64-bit nodes, under such case, neither
 * FMODE_32BITHASH nor FMODE_64BITHASH is specified.
 */
static inline loff_t hash2pos(struct file *filp, __u32 major, __u32 minor)
{
	if ((filp->f_mode & FMODE_32BITHASH) ||
	    (!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
		return major >> 1;
	else
		return ((__u64)(major >> 1) << 32) | (__u64)minor;
}

static inline __u32 pos2maj_hash(struct file *filp, loff_t pos)
{
	if ((filp->f_mode & FMODE_32BITHASH) ||
	    (!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
		return (pos << 1) & 0xffffffff;
	else
		return ((pos >> 32) << 1) & 0xffffffff;
}

static inline __u32 pos2min_hash(struct file *filp, loff_t pos)
{
	if ((filp->f_mode & FMODE_32BITHASH) ||
	    (!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
		return 0;
	else
		return pos & 0xffffffff;
}

/*
 * Return 32- or 64-bit end-of-file for dx directories
 */
static inline loff_t ext4_get_htree_eof(struct file *filp)
{
	if ((filp->f_mode & FMODE_32BITHASH) ||
	    (!(filp->f_mode & FMODE_64BITHASH) && is_32bit_api()))
		return EXT4_HTREE_EOF_32BIT;
	else
		return EXT4_HTREE_EOF_64BIT;
}


/*
 * ext4_dir_llseek() based on generic_file_llseek() to handle both
 * non-htree and htree directories, where the "offset" is in terms
 * of the filename hash value instead of the byte offset.
318
 *
319 320
 * NOTE: offsets obtained *before* ext4_set_inode_flag(dir, EXT4_INODE_INDEX)
 *       will be invalid once the directory was converted into a dx directory
321
 */
322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
loff_t ext4_dir_llseek(struct file *file, loff_t offset, int origin)
{
	struct inode *inode = file->f_mapping->host;
	loff_t ret = -EINVAL;
	int dx_dir = is_dx_dir(inode);

	mutex_lock(&inode->i_mutex);

	/* NOTE: relative offsets with dx directories might not work
	 *       as expected, as it is difficult to figure out the
	 *       correct offset between dx hashes */

	switch (origin) {
	case SEEK_END:
		if (unlikely(offset > 0))
			goto out_err; /* not supported for directories */

		/* so only negative offsets are left, does that have a
		 * meaning for directories at all? */
		if (dx_dir)
			offset += ext4_get_htree_eof(file);
		else
			offset += inode->i_size;
		break;
	case SEEK_CUR:
		/*
		 * Here we special-case the lseek(fd, 0, SEEK_CUR)
		 * position-querying operation.  Avoid rewriting the "same"
		 * f_pos value back to the file because a concurrent read(),
		 * write() or lseek() might have altered it
		 */
		if (offset == 0) {
			offset = file->f_pos;
			goto out_ok;
		}

		offset += file->f_pos;
		break;
	}

	if (unlikely(offset < 0))
		goto out_err;

	if (!dx_dir) {
		if (offset > inode->i_sb->s_maxbytes)
			goto out_err;
	} else if (offset > ext4_get_htree_eof(file))
		goto out_err;

	/* Special lock needed here? */
	if (offset != file->f_pos) {
		file->f_pos = offset;
		file->f_version = 0;
	}

out_ok:
	ret = offset;
out_err:
	mutex_unlock(&inode->i_mutex);

	return ret;
}
384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411

/*
 * This structure holds the nodes of the red-black tree used to store
 * the directory entry in hash order.
 */
struct fname {
	__u32		hash;
	__u32		minor_hash;
	struct rb_node	rb_hash;
	struct fname	*next;
	__u32		inode;
	__u8		name_len;
	__u8		file_type;
	char		name[0];
};

/*
 * This functoin implements a non-recursive way of freeing all of the
 * nodes in the red-black tree.
 */
static void free_rb_tree_fname(struct rb_root *root)
{
	struct rb_node	*n = root->rb_node;
	struct rb_node	*parent;
	struct fname	*fname;

	while (n) {
		/* Do the node's children first */
412
		if (n->rb_left) {
413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
			n = n->rb_left;
			continue;
		}
		if (n->rb_right) {
			n = n->rb_right;
			continue;
		}
		/*
		 * The node has no children; free it, and then zero
		 * out parent's link to it.  Finally go to the
		 * beginning of the loop and try to free the parent
		 * node.
		 */
		parent = rb_parent(n);
		fname = rb_entry(n, struct fname, rb_hash);
		while (fname) {
429
			struct fname *old = fname;
430
			fname = fname->next;
431
			kfree(old);
432 433
		}
		if (!parent)
434
			*root = RB_ROOT;
435 436 437 438 439 440 441 442 443
		else if (parent->rb_left == n)
			parent->rb_left = NULL;
		else if (parent->rb_right == n)
			parent->rb_right = NULL;
		n = parent;
	}
}


444 445
static struct dir_private_info *ext4_htree_create_dir_info(struct file *filp,
							   loff_t pos)
446 447 448
{
	struct dir_private_info *p;

449
	p = kzalloc(sizeof(struct dir_private_info), GFP_KERNEL);
450 451
	if (!p)
		return NULL;
452 453
	p->curr_hash = pos2maj_hash(filp, pos);
	p->curr_minor_hash = pos2min_hash(filp, pos);
454 455 456
	return p;
}

457
void ext4_htree_free_dir_info(struct dir_private_info *p)
458 459 460 461 462 463 464 465
{
	free_rb_tree_fname(&p->root);
	kfree(p);
}

/*
 * Given a directory entry, enter it into the fname rb tree.
 */
466
int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
467
			     __u32 minor_hash,
468
			     struct ext4_dir_entry_2 *dirent)
469 470
{
	struct rb_node **p, *parent = NULL;
471
	struct fname *fname, *new_fn;
472 473 474
	struct dir_private_info *info;
	int len;

475
	info = dir_file->private_data;
476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
	p = &info->root.rb_node;

	/* Create and allocate the fname structure */
	len = sizeof(struct fname) + dirent->name_len + 1;
	new_fn = kzalloc(len, GFP_KERNEL);
	if (!new_fn)
		return -ENOMEM;
	new_fn->hash = hash;
	new_fn->minor_hash = minor_hash;
	new_fn->inode = le32_to_cpu(dirent->inode);
	new_fn->name_len = dirent->name_len;
	new_fn->file_type = dirent->file_type;
	memcpy(new_fn->name, dirent->name, dirent->name_len);
	new_fn->name[dirent->name_len] = 0;

	while (*p) {
		parent = *p;
		fname = rb_entry(parent, struct fname, rb_hash);

		/*
		 * If the hash and minor hash match up, then we put
		 * them on a linked list.  This rarely happens...
		 */
		if ((new_fn->hash == fname->hash) &&
		    (new_fn->minor_hash == fname->minor_hash)) {
			new_fn->next = fname->next;
			fname->next = new_fn;
			return 0;
		}

		if (new_fn->hash < fname->hash)
			p = &(*p)->rb_left;
		else if (new_fn->hash > fname->hash)
			p = &(*p)->rb_right;
		else if (new_fn->minor_hash < fname->minor_hash)
			p = &(*p)->rb_left;
		else /* if (new_fn->minor_hash > fname->minor_hash) */
			p = &(*p)->rb_right;
	}

	rb_link_node(&new_fn->rb_hash, parent, p);
	rb_insert_color(&new_fn->rb_hash, &info->root);
	return 0;
}



/*
524
 * This is a helper function for ext4_dx_readdir.  It calls filldir
525 526 527
 * for all entres on the fname linked list.  (Normally there is only
 * one entry on the linked list, unless there are 62 bit hash collisions.)
 */
528
static int call_filldir(struct file *filp, void *dirent,
529 530 531 532
			filldir_t filldir, struct fname *fname)
{
	struct dir_private_info *info = filp->private_data;
	loff_t	curr_pos;
533
	struct inode *inode = filp->f_path.dentry->d_inode;
534
	struct super_block *sb;
535 536 537 538 539
	int error;

	sb = inode->i_sb;

	if (!fname) {
540 541 542
		ext4_msg(sb, KERN_ERR, "%s:%d: inode #%lu: comm %s: "
			 "called with null fname?!?", __func__, __LINE__,
			 inode->i_ino, current->comm);
543 544
		return 0;
	}
545
	curr_pos = hash2pos(filp, fname->hash, fname->minor_hash);
546 547 548 549 550 551 552
	while (fname) {
		error = filldir(dirent, fname->name,
				fname->name_len, curr_pos,
				fname->inode,
				get_dtype(sb, fname->file_type));
		if (error) {
			filp->f_pos = curr_pos;
553
			info->extra_fname = fname;
554 555 556 557 558 559 560
			return error;
		}
		fname = fname->next;
	}
	return 0;
}

561 562
static int ext4_dx_readdir(struct file *filp,
			 void *dirent, filldir_t filldir)
563 564
{
	struct dir_private_info *info = filp->private_data;
565
	struct inode *inode = filp->f_path.dentry->d_inode;
566 567 568 569
	struct fname *fname;
	int	ret;

	if (!info) {
570
		info = ext4_htree_create_dir_info(filp, filp->f_pos);
571 572 573 574 575
		if (!info)
			return -ENOMEM;
		filp->private_data = info;
	}

576
	if (filp->f_pos == ext4_get_htree_eof(filp))
577 578 579 580 581 582 583
		return 0;	/* EOF */

	/* Some one has messed with f_pos; reset the world */
	if (info->last_pos != filp->f_pos) {
		free_rb_tree_fname(&info->root);
		info->curr_node = NULL;
		info->extra_fname = NULL;
584 585
		info->curr_hash = pos2maj_hash(filp, filp->f_pos);
		info->curr_minor_hash = pos2min_hash(filp, filp->f_pos);
586 587 588 589 590 591
	}

	/*
	 * If there are any leftover names on the hash collision
	 * chain, return them first.
	 */
592 593 594 595
	if (info->extra_fname) {
		if (call_filldir(filp, dirent, filldir, info->extra_fname))
			goto finished;
		info->extra_fname = NULL;
596
		goto next_node;
597
	} else if (!info->curr_node)
598 599 600 601 602 603 604 605 606 607 608 609 610
		info->curr_node = rb_first(&info->root);

	while (1) {
		/*
		 * Fill the rbtree if we have no more entries,
		 * or the inode has changed since we last read in the
		 * cached entries.
		 */
		if ((!info->curr_node) ||
		    (filp->f_version != inode->i_version)) {
			info->curr_node = NULL;
			free_rb_tree_fname(&info->root);
			filp->f_version = inode->i_version;
611
			ret = ext4_htree_fill_tree(filp, info->curr_hash,
612 613 614 615 616
						   info->curr_minor_hash,
						   &info->next_hash);
			if (ret < 0)
				return ret;
			if (ret == 0) {
617
				filp->f_pos = ext4_get_htree_eof(filp);
618 619 620 621 622 623 624 625 626 627
				break;
			}
			info->curr_node = rb_first(&info->root);
		}

		fname = rb_entry(info->curr_node, struct fname, rb_hash);
		info->curr_hash = fname->hash;
		info->curr_minor_hash = fname->minor_hash;
		if (call_filldir(filp, dirent, filldir, fname))
			break;
628
	next_node:
629
		info->curr_node = rb_next(info->curr_node);
630 631 632 633 634 635
		if (info->curr_node) {
			fname = rb_entry(info->curr_node, struct fname,
					 rb_hash);
			info->curr_hash = fname->hash;
			info->curr_minor_hash = fname->minor_hash;
		} else {
636
			if (info->next_hash == ~0) {
637
				filp->f_pos = ext4_get_htree_eof(filp);
638 639 640 641 642 643 644 645 646 647 648
				break;
			}
			info->curr_hash = info->next_hash;
			info->curr_minor_hash = 0;
		}
	}
finished:
	info->last_pos = filp->f_pos;
	return 0;
}

649
static int ext4_release_dir(struct inode *inode, struct file *filp)
650
{
651
	if (filp->private_data)
652
		ext4_htree_free_dir_info(filp->private_data);
653 654 655

	return 0;
}
656 657 658 659 660 661 662 663 664 665 666 667

const struct file_operations ext4_dir_operations = {
	.llseek		= ext4_dir_llseek,
	.read		= generic_read_dir,
	.readdir	= ext4_readdir,
	.unlocked_ioctl = ext4_ioctl,
#ifdef CONFIG_COMPAT
	.compat_ioctl	= ext4_compat_ioctl,
#endif
	.fsync		= ext4_sync_file,
	.release	= ext4_release_dir,
};