namei.c 64.3 KB
Newer Older
1
/*
2
 *  linux/fs/ext4/namei.c
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
 *
 * 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/namei.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
 *  Big-endian to little-endian byte-swapping/bitmaps by
 *        David S. Miller (davem@caip.rutgers.edu), 1995
 *  Directory entry file type support and forward compatibility hooks
 *	for B-tree directories by Theodore Ts'o (tytso@mit.edu), 1998
 *  Hash Tree Directory indexing (c)
 *	Daniel Phillips, 2001
 *  Hash Tree Directory indexing porting
 *	Christopher Li, 2002
 *  Hash Tree Directory indexing cleanup
 *	Theodore Ts'o, 2002
 */

#include <linux/fs.h>
#include <linux/pagemap.h>
29
#include <linux/jbd2.h>
30
#include <linux/time.h>
31
#include <linux/ext4_fs.h>
32
#include <linux/ext4_jbd2.h>
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
#include <linux/fcntl.h>
#include <linux/stat.h>
#include <linux/string.h>
#include <linux/quotaops.h>
#include <linux/buffer_head.h>
#include <linux/bio.h>

#include "namei.h"
#include "xattr.h"
#include "acl.h"

/*
 * define how far ahead to read directories while searching them.
 */
#define NAMEI_RA_CHUNKS  2
#define NAMEI_RA_BLOCKS  4
#define NAMEI_RA_SIZE        (NAMEI_RA_CHUNKS * NAMEI_RA_BLOCKS)
#define NAMEI_RA_INDEX(c,b)  (((c) * NAMEI_RA_BLOCKS) + (b))

52
static struct buffer_head *ext4_append(handle_t *handle,
53 54 55 56 57 58 59
					struct inode *inode,
					u32 *block, int *err)
{
	struct buffer_head *bh;

	*block = inode->i_size >> inode->i_sb->s_blocksize_bits;

60
	if ((bh = ext4_bread(handle, inode, *block, 1, err))) {
61
		inode->i_size += inode->i_sb->s_blocksize;
62 63
		EXT4_I(inode)->i_disksize = inode->i_size;
		ext4_journal_get_write_access(handle,bh);
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
	}
	return bh;
}

#ifndef assert
#define assert(test) J_ASSERT(test)
#endif

#ifndef swap
#define swap(x, y) do { typeof(x) z = x; x = y; y = z; } while (0)
#endif

#ifdef DX_DEBUG
#define dxtrace(command) command
#else
#define dxtrace(command)
#endif

struct fake_dirent
{
	__le32 inode;
	__le16 rec_len;
	u8 name_len;
	u8 file_type;
};

struct dx_countlimit
{
	__le16 limit;
	__le16 count;
};

struct dx_entry
{
	__le32 hash;
	__le32 block;
};

/*
 * dx_root_info is laid out so that if it should somehow get overlaid by a
 * dirent the two low bits of the hash version will be zero.  Therefore, the
 * hash version mod 4 should never be 0.  Sincerely, the paranoia department.
 */

struct dx_root
{
	struct fake_dirent dot;
	char dot_name[4];
	struct fake_dirent dotdot;
	char dotdot_name[4];
	struct dx_root_info
	{
		__le32 reserved_zero;
		u8 hash_version;
		u8 info_length; /* 8 */
		u8 indirect_levels;
		u8 unused_flags;
	}
	info;
	struct dx_entry	entries[0];
};

struct dx_node
{
	struct fake_dirent fake;
	struct dx_entry	entries[0];
};


struct dx_frame
{
	struct buffer_head *bh;
	struct dx_entry *entries;
	struct dx_entry *at;
};

struct dx_map_entry
{
	u32 hash;
	u32 offs;
};

146
#ifdef CONFIG_EXT4_INDEX
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
static inline unsigned dx_get_block (struct dx_entry *entry);
static void dx_set_block (struct dx_entry *entry, unsigned value);
static inline unsigned dx_get_hash (struct dx_entry *entry);
static void dx_set_hash (struct dx_entry *entry, unsigned value);
static unsigned dx_get_count (struct dx_entry *entries);
static unsigned dx_get_limit (struct dx_entry *entries);
static void dx_set_count (struct dx_entry *entries, unsigned value);
static void dx_set_limit (struct dx_entry *entries, unsigned value);
static unsigned dx_root_limit (struct inode *dir, unsigned infosize);
static unsigned dx_node_limit (struct inode *dir);
static struct dx_frame *dx_probe(struct dentry *dentry,
				 struct inode *dir,
				 struct dx_hash_info *hinfo,
				 struct dx_frame *frame,
				 int *err);
static void dx_release (struct dx_frame *frames);
163
static int dx_make_map (struct ext4_dir_entry_2 *de, int size,
164 165
			struct dx_hash_info *hinfo, struct dx_map_entry map[]);
static void dx_sort_map(struct dx_map_entry *map, unsigned count);
166
static struct ext4_dir_entry_2 *dx_move_dirents (char *from, char *to,
167
		struct dx_map_entry *offsets, int count);
168
static struct ext4_dir_entry_2* dx_pack_dirents (char *base, int size);
169
static void dx_insert_block (struct dx_frame *frame, u32 hash, u32 block);
170
static int ext4_htree_next_block(struct inode *dir, __u32 hash,
171 172 173
				 struct dx_frame *frame,
				 struct dx_frame *frames,
				 __u32 *start_hash);
174 175 176
static struct buffer_head * ext4_dx_find_entry(struct dentry *dentry,
		       struct ext4_dir_entry_2 **res_dir, int *err);
static int ext4_dx_add_entry(handle_t *handle, struct dentry *dentry,
177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225
			     struct inode *inode);

/*
 * Future: use high four bits of block for coalesce-on-delete flags
 * Mask them off for now.
 */

static inline unsigned dx_get_block (struct dx_entry *entry)
{
	return le32_to_cpu(entry->block) & 0x00ffffff;
}

static inline void dx_set_block (struct dx_entry *entry, unsigned value)
{
	entry->block = cpu_to_le32(value);
}

static inline unsigned dx_get_hash (struct dx_entry *entry)
{
	return le32_to_cpu(entry->hash);
}

static inline void dx_set_hash (struct dx_entry *entry, unsigned value)
{
	entry->hash = cpu_to_le32(value);
}

static inline unsigned dx_get_count (struct dx_entry *entries)
{
	return le16_to_cpu(((struct dx_countlimit *) entries)->count);
}

static inline unsigned dx_get_limit (struct dx_entry *entries)
{
	return le16_to_cpu(((struct dx_countlimit *) entries)->limit);
}

static inline void dx_set_count (struct dx_entry *entries, unsigned value)
{
	((struct dx_countlimit *) entries)->count = cpu_to_le16(value);
}

static inline void dx_set_limit (struct dx_entry *entries, unsigned value)
{
	((struct dx_countlimit *) entries)->limit = cpu_to_le16(value);
}

static inline unsigned dx_root_limit (struct inode *dir, unsigned infosize)
{
226 227
	unsigned entry_space = dir->i_sb->s_blocksize - EXT4_DIR_REC_LEN(1) -
		EXT4_DIR_REC_LEN(2) - infosize;
228 229 230 231 232
	return 0? 20: entry_space / sizeof(struct dx_entry);
}

static inline unsigned dx_node_limit (struct inode *dir)
{
233
	unsigned entry_space = dir->i_sb->s_blocksize - EXT4_DIR_REC_LEN(0);
234 235 236 237 238 239 240 241 242
	return 0? 22: entry_space / sizeof(struct dx_entry);
}

/*
 * Debug
 */
#ifdef DX_DEBUG
static void dx_show_index (char * label, struct dx_entry *entries)
{
A
Andrew Morton 已提交
243
	int i, n = dx_get_count (entries);
244
        printk("%s index ", label);
A
Andrew Morton 已提交
245 246 247 248 249
	for (i = 0; i < n; i++) {
		printk("%x->%u ", i? dx_get_hash(entries + i) :
				0, dx_get_block(entries + i));
	}
	printk("\n");
250 251 252 253 254 255 256 257 258
}

struct stats
{
	unsigned names;
	unsigned space;
	unsigned bcount;
};

259
static struct stats dx_show_leaf(struct dx_hash_info *hinfo, struct ext4_dir_entry_2 *de,
260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275
				 int size, int show_names)
{
	unsigned names = 0, space = 0;
	char *base = (char *) de;
	struct dx_hash_info h = *hinfo;

	printk("names: ");
	while ((char *) de < base + size)
	{
		if (de->inode)
		{
			if (show_names)
			{
				int len = de->name_len;
				char *name = de->name;
				while (len--) printk("%c", *name++);
276
				ext4fs_dirhash(de->name, de->name_len, &h);
277 278 279
				printk(":%x.%u ", h.hash,
				       ((char *) de - base));
			}
280
			space += EXT4_DIR_REC_LEN(de->name_len);
281 282
			names++;
		}
283
		de = (struct ext4_dir_entry_2 *) ((char *) de + le16_to_cpu(de->rec_len));
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303
	}
	printk("(%i)\n", names);
	return (struct stats) { names, space, 1 };
}

struct stats dx_show_entries(struct dx_hash_info *hinfo, struct inode *dir,
			     struct dx_entry *entries, int levels)
{
	unsigned blocksize = dir->i_sb->s_blocksize;
	unsigned count = dx_get_count (entries), names = 0, space = 0, i;
	unsigned bcount = 0;
	struct buffer_head *bh;
	int err;
	printk("%i indexed blocks...\n", count);
	for (i = 0; i < count; i++, entries++)
	{
		u32 block = dx_get_block(entries), hash = i? dx_get_hash(entries): 0;
		u32 range = i < count - 1? (dx_get_hash(entries + 1) - hash): ~hash;
		struct stats stats;
		printk("%s%3u:%03u hash %8x/%8x ",levels?"":"   ", i, block, hash, range);
304
		if (!(bh = ext4_bread (NULL,dir, block, 0,&err))) continue;
305 306
		stats = levels?
		   dx_show_entries(hinfo, dir, ((struct dx_node *) bh->b_data)->entries, levels - 1):
307
		   dx_show_leaf(hinfo, (struct ext4_dir_entry_2 *) bh->b_data, blocksize, 0);
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
		names += stats.names;
		space += stats.space;
		bcount += stats.bcount;
		brelse (bh);
	}
	if (bcount)
		printk("%snames %u, fullness %u (%u%%)\n", levels?"":"   ",
			names, space/bcount,(space/bcount)*100/blocksize);
	return (struct stats) { names, space, bcount};
}
#endif /* DX_DEBUG */

/*
 * Probe for a directory leaf block to search.
 *
 * dx_probe can return ERR_BAD_DX_DIR, which means there was a format
 * error in the directory index, and the caller should fall back to
 * searching the directory normally.  The callers of dx_probe **MUST**
 * check for this error code, and make sure it never gets reflected
 * back to userspace.
 */
static struct dx_frame *
dx_probe(struct dentry *dentry, struct inode *dir,
	 struct dx_hash_info *hinfo, struct dx_frame *frame_in, int *err)
{
	unsigned count, indirect;
	struct dx_entry *at, *entries, *p, *q, *m;
	struct dx_root *root;
	struct buffer_head *bh;
	struct dx_frame *frame = frame_in;
	u32 hash;

	frame->bh = NULL;
	if (dentry)
		dir = dentry->d_parent->d_inode;
343
	if (!(bh = ext4_bread (NULL,dir, 0, 0, err)))
344 345 346 347 348
		goto fail;
	root = (struct dx_root *) bh->b_data;
	if (root->info.hash_version != DX_HASH_TEA &&
	    root->info.hash_version != DX_HASH_HALF_MD4 &&
	    root->info.hash_version != DX_HASH_LEGACY) {
349
		ext4_warning(dir->i_sb, __FUNCTION__,
350 351 352 353 354 355 356
			     "Unrecognised inode hash code %d",
			     root->info.hash_version);
		brelse(bh);
		*err = ERR_BAD_DX_DIR;
		goto fail;
	}
	hinfo->hash_version = root->info.hash_version;
357
	hinfo->seed = EXT4_SB(dir->i_sb)->s_hash_seed;
358
	if (dentry)
359
		ext4fs_dirhash(dentry->d_name.name, dentry->d_name.len, hinfo);
360 361 362
	hash = hinfo->hash;

	if (root->info.unused_flags & 1) {
363
		ext4_warning(dir->i_sb, __FUNCTION__,
364 365 366 367 368 369 370 371
			     "Unimplemented inode hash flags: %#06x",
			     root->info.unused_flags);
		brelse(bh);
		*err = ERR_BAD_DX_DIR;
		goto fail;
	}

	if ((indirect = root->info.indirect_levels) > 1) {
372
		ext4_warning(dir->i_sb, __FUNCTION__,
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422
			     "Unimplemented inode hash depth: %#06x",
			     root->info.indirect_levels);
		brelse(bh);
		*err = ERR_BAD_DX_DIR;
		goto fail;
	}

	entries = (struct dx_entry *) (((char *)&root->info) +
				       root->info.info_length);
	assert(dx_get_limit(entries) == dx_root_limit(dir,
						      root->info.info_length));
	dxtrace (printk("Look up %x", hash));
	while (1)
	{
		count = dx_get_count(entries);
		assert (count && count <= dx_get_limit(entries));
		p = entries + 1;
		q = entries + count - 1;
		while (p <= q)
		{
			m = p + (q - p)/2;
			dxtrace(printk("."));
			if (dx_get_hash(m) > hash)
				q = m - 1;
			else
				p = m + 1;
		}

		if (0) // linear search cross check
		{
			unsigned n = count - 1;
			at = entries;
			while (n--)
			{
				dxtrace(printk(","));
				if (dx_get_hash(++at) > hash)
				{
					at--;
					break;
				}
			}
			assert (at == p - 1);
		}

		at = p - 1;
		dxtrace(printk(" %x->%u\n", at == entries? 0: dx_get_hash(at), dx_get_block(at)));
		frame->bh = bh;
		frame->entries = entries;
		frame->at = at;
		if (!indirect--) return frame;
423
		if (!(bh = ext4_bread (NULL,dir, dx_get_block(at), 0, err)))
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
			goto fail2;
		at = entries = ((struct dx_node *) bh->b_data)->entries;
		assert (dx_get_limit(entries) == dx_node_limit (dir));
		frame++;
	}
fail2:
	while (frame >= frame_in) {
		brelse(frame->bh);
		frame--;
	}
fail:
	return NULL;
}

static void dx_release (struct dx_frame *frames)
{
	if (frames[0].bh == NULL)
		return;

	if (((struct dx_root *) frames[0].bh->b_data)->info.indirect_levels)
		brelse(frames[1].bh);
	brelse(frames[0].bh);
}

/*
 * This function increments the frame pointer to search the next leaf
 * block, and reads in the necessary intervening nodes if the search
 * should be necessary.  Whether or not the search is necessary is
 * controlled by the hash parameter.  If the hash value is even, then
 * the search is only continued if the next block starts with that
 * hash value.  This is used if we are searching for a specific file.
 *
 * If the hash value is HASH_NB_ALWAYS, then always go to the next block.
 *
 * This function returns 1 if the caller should continue to search,
 * or 0 if it should not.  If there is an error reading one of the
 * index blocks, it will a negative error code.
 *
 * If start_hash is non-null, it will be filled in with the starting
 * hash of the next page.
 */
465
static int ext4_htree_next_block(struct inode *dir, __u32 hash,
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
				 struct dx_frame *frame,
				 struct dx_frame *frames,
				 __u32 *start_hash)
{
	struct dx_frame *p;
	struct buffer_head *bh;
	int err, num_frames = 0;
	__u32 bhash;

	p = frame;
	/*
	 * Find the next leaf page by incrementing the frame pointer.
	 * If we run out of entries in the interior node, loop around and
	 * increment pointer in the parent node.  When we break out of
	 * this loop, num_frames indicates the number of interior
	 * nodes need to be read.
	 */
	while (1) {
		if (++(p->at) < p->entries + dx_get_count(p->entries))
			break;
		if (p == frames)
			return 0;
		num_frames++;
		p--;
	}

	/*
	 * If the hash is 1, then continue only if the next page has a
	 * continuation hash of any value.  This is used for readdir
	 * handling.  Otherwise, check to see if the hash matches the
	 * desired contiuation hash.  If it doesn't, return since
	 * there's no point to read in the successive index pages.
	 */
	bhash = dx_get_hash(p->at);
	if (start_hash)
		*start_hash = bhash;
	if ((hash & 1) == 0) {
		if ((bhash & ~1) != hash)
			return 0;
	}
	/*
	 * If the hash is HASH_NB_ALWAYS, we always go to the next
	 * block so no check is necessary
	 */
	while (num_frames--) {
511
		if (!(bh = ext4_bread(NULL, dir, dx_get_block(p->at),
512 513 514 515 516 517 518 519 520 521 522 523 524 525
				      0, &err)))
			return err; /* Failure */
		p++;
		brelse (p->bh);
		p->bh = bh;
		p->at = p->entries = ((struct dx_node *) bh->b_data)->entries;
	}
	return 1;
}


/*
 * p is at least 6 bytes before the end of page
 */
526
static inline struct ext4_dir_entry_2 *ext4_next_entry(struct ext4_dir_entry_2 *p)
527
{
528
	return (struct ext4_dir_entry_2 *)((char*)p + le16_to_cpu(p->rec_len));
529 530 531 532 533 534 535 536 537 538 539 540 541
}

/*
 * This function fills a red-black tree with information from a
 * directory block.  It returns the number directory entries loaded
 * into the tree.  If there is an error it is returned in err.
 */
static int htree_dirblock_to_tree(struct file *dir_file,
				  struct inode *dir, int block,
				  struct dx_hash_info *hinfo,
				  __u32 start_hash, __u32 start_minor_hash)
{
	struct buffer_head *bh;
542
	struct ext4_dir_entry_2 *de, *top;
543 544 545
	int err, count = 0;

	dxtrace(printk("In htree dirblock_to_tree: block %d\n", block));
546
	if (!(bh = ext4_bread (NULL, dir, block, 0, &err)))
547 548
		return err;

549 550
	de = (struct ext4_dir_entry_2 *) bh->b_data;
	top = (struct ext4_dir_entry_2 *) ((char *) de +
551
					   dir->i_sb->s_blocksize -
552 553
					   EXT4_DIR_REC_LEN(0));
	for (; de < top; de = ext4_next_entry(de)) {
554 555 556 557 558 559 560 561 562
		if (!ext4_check_dir_entry("htree_dirblock_to_tree", dir, de, bh,
					(block<<EXT4_BLOCK_SIZE_BITS(dir->i_sb))
						+((char *)de - bh->b_data))) {
			/* On error, skip the f_pos to the next block. */
			dir_file->f_pos = (dir_file->f_pos |
					(dir->i_sb->s_blocksize - 1)) + 1;
			brelse (bh);
			return count;
		}
563
		ext4fs_dirhash(de->name, de->name_len, hinfo);
564 565 566 567 568 569
		if ((hinfo->hash < start_hash) ||
		    ((hinfo->hash == start_hash) &&
		     (hinfo->minor_hash < start_minor_hash)))
			continue;
		if (de->inode == 0)
			continue;
570
		if ((err = ext4_htree_store_dirent(dir_file,
571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589
				   hinfo->hash, hinfo->minor_hash, de)) != 0) {
			brelse(bh);
			return err;
		}
		count++;
	}
	brelse(bh);
	return count;
}


/*
 * This function fills a red-black tree with information from a
 * directory.  We start scanning the directory in hash order, starting
 * at start_hash and start_minor_hash.
 *
 * This function returns the number of entries inserted into the tree,
 * or a negative error code.
 */
590
int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
591 592 593
			 __u32 start_minor_hash, __u32 *next_hash)
{
	struct dx_hash_info hinfo;
594
	struct ext4_dir_entry_2 *de;
595 596 597 598 599 600 601 602 603
	struct dx_frame frames[2], *frame;
	struct inode *dir;
	int block, err;
	int count = 0;
	int ret;
	__u32 hashval;

	dxtrace(printk("In htree_fill_tree, start hash: %x:%x\n", start_hash,
		       start_minor_hash));
604
	dir = dir_file->f_path.dentry->d_inode;
605 606 607
	if (!(EXT4_I(dir)->i_flags & EXT4_INDEX_FL)) {
		hinfo.hash_version = EXT4_SB(dir->i_sb)->s_def_hash_version;
		hinfo.seed = EXT4_SB(dir->i_sb)->s_hash_seed;
608 609 610 611 612 613 614
		count = htree_dirblock_to_tree(dir_file, dir, 0, &hinfo,
					       start_hash, start_minor_hash);
		*next_hash = ~0;
		return count;
	}
	hinfo.hash = start_hash;
	hinfo.minor_hash = 0;
615
	frame = dx_probe(NULL, dir_file->f_path.dentry->d_inode, &hinfo, frames, &err);
616 617 618 619 620
	if (!frame)
		return err;

	/* Add '.' and '..' from the htree header */
	if (!start_hash && !start_minor_hash) {
621 622
		de = (struct ext4_dir_entry_2 *) frames[0].bh->b_data;
		if ((err = ext4_htree_store_dirent(dir_file, 0, 0, de)) != 0)
623 624 625 626
			goto errout;
		count++;
	}
	if (start_hash < 2 || (start_hash ==2 && start_minor_hash==0)) {
627 628 629
		de = (struct ext4_dir_entry_2 *) frames[0].bh->b_data;
		de = ext4_next_entry(de);
		if ((err = ext4_htree_store_dirent(dir_file, 2, 0, de)) != 0)
630 631 632 633 634 635 636 637 638 639 640 641 642 643
			goto errout;
		count++;
	}

	while (1) {
		block = dx_get_block(frame->at);
		ret = htree_dirblock_to_tree(dir_file, dir, block, &hinfo,
					     start_hash, start_minor_hash);
		if (ret < 0) {
			err = ret;
			goto errout;
		}
		count += ret;
		hashval = ~0;
644
		ret = ext4_htree_next_block(dir, HASH_NB_ALWAYS,
645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673
					    frame, frames, &hashval);
		*next_hash = hashval;
		if (ret < 0) {
			err = ret;
			goto errout;
		}
		/*
		 * Stop if:  (a) there are no more entries, or
		 * (b) we have inserted at least one entry and the
		 * next hash value is not a continuation
		 */
		if ((ret == 0) ||
		    (count && ((hashval & 1) == 0)))
			break;
	}
	dx_release(frames);
	dxtrace(printk("Fill tree: returned %d entries, next hash: %x\n",
		       count, *next_hash));
	return count;
errout:
	dx_release(frames);
	return (err);
}


/*
 * Directory block splitting, compacting
 */

674
static int dx_make_map (struct ext4_dir_entry_2 *de, int size,
675 676 677 678 679 680 681 682 683
			struct dx_hash_info *hinfo, struct dx_map_entry *map_tail)
{
	int count = 0;
	char *base = (char *) de;
	struct dx_hash_info h = *hinfo;

	while ((char *) de < base + size)
	{
		if (de->name_len && de->inode) {
684
			ext4fs_dirhash(de->name, de->name_len, &h);
685 686 687 688 689 690 691
			map_tail--;
			map_tail->hash = h.hash;
			map_tail->offs = (u32) ((char *) de - base);
			count++;
			cond_resched();
		}
		/* XXX: do we need to check rec_len == 0 case? -Chris */
692
		de = (struct ext4_dir_entry_2 *) ((char *) de + le16_to_cpu(de->rec_len));
693 694 695 696 697 698
	}
	return count;
}

static void dx_sort_map (struct dx_map_entry *map, unsigned count)
{
A
Andrew Morton 已提交
699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
	struct dx_map_entry *p, *q, *top = map + count - 1;
	int more;
	/* Combsort until bubble sort doesn't suck */
	while (count > 2) {
		count = count*10/13;
		if (count - 9 < 2) /* 9, 10 -> 11 */
			count = 11;
		for (p = top, q = p - count; q >= map; p--, q--)
			if (p->hash < q->hash)
				swap(*p, *q);
	}
	/* Garden variety bubble sort */
	do {
		more = 0;
		q = top;
		while (q-- > map) {
			if (q[1].hash >= q[0].hash)
716
				continue;
A
Andrew Morton 已提交
717 718
			swap(*(q+1), *q);
			more = 1;
719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738
		}
	} while(more);
}

static void dx_insert_block(struct dx_frame *frame, u32 hash, u32 block)
{
	struct dx_entry *entries = frame->entries;
	struct dx_entry *old = frame->at, *new = old + 1;
	int count = dx_get_count(entries);

	assert(count < dx_get_limit(entries));
	assert(old < entries + count);
	memmove(new + 1, new, (char *)(entries + count) - (char *)(new));
	dx_set_hash(new, hash);
	dx_set_block(new, block);
	dx_set_count(entries, count + 1);
}
#endif


739
static void ext4_update_dx_flag(struct inode *inode)
740
{
741 742 743
	if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
				     EXT4_FEATURE_COMPAT_DIR_INDEX))
		EXT4_I(inode)->i_flags &= ~EXT4_INDEX_FL;
744 745 746
}

/*
747
 * NOTE! unlike strncmp, ext4_match returns 1 for success, 0 for failure.
748
 *
749
 * `len <= EXT4_NAME_LEN' is guaranteed by caller.
750 751
 * `de != NULL' is guaranteed by caller.
 */
752 753
static inline int ext4_match (int len, const char * const name,
			      struct ext4_dir_entry_2 * de)
754 755 756 757 758 759 760 761 762 763 764 765 766 767 768
{
	if (len != de->name_len)
		return 0;
	if (!de->inode)
		return 0;
	return !memcmp(name, de->name, len);
}

/*
 * Returns 0 if not found, -1 on failure, and 1 on success
 */
static inline int search_dirblock(struct buffer_head * bh,
				  struct inode *dir,
				  struct dentry *dentry,
				  unsigned long offset,
769
				  struct ext4_dir_entry_2 ** res_dir)
770
{
771
	struct ext4_dir_entry_2 * de;
772 773 774 775 776
	char * dlimit;
	int de_len;
	const char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;

777
	de = (struct ext4_dir_entry_2 *) bh->b_data;
778 779 780 781 782 783
	dlimit = bh->b_data + dir->i_sb->s_blocksize;
	while ((char *) de < dlimit) {
		/* this code is executed quadratically often */
		/* do minimal checking `by hand' */

		if ((char *) de + namelen <= dlimit &&
784
		    ext4_match (namelen, name, de)) {
785
			/* found a match - just to be sure, do a full check */
786
			if (!ext4_check_dir_entry("ext4_find_entry",
787 788 789 790 791 792 793 794 795 796
						  dir, de, bh, offset))
				return -1;
			*res_dir = de;
			return 1;
		}
		/* prevent looping on a bad block */
		de_len = le16_to_cpu(de->rec_len);
		if (de_len <= 0)
			return -1;
		offset += de_len;
797
		de = (struct ext4_dir_entry_2 *) ((char *) de + de_len);
798 799 800 801 802 803
	}
	return 0;
}


/*
804
 *	ext4_find_entry()
805 806 807 808 809 810 811 812 813
 *
 * finds an entry in the specified directory with the wanted name. It
 * returns the cache buffer in which the entry was found, and the entry
 * itself (as a parameter - res_dir). It does NOT read the inode of the
 * entry - you'll have to do that yourself if you want to.
 *
 * The returned buffer_head has ->b_count elevated.  The caller is expected
 * to brelse() it when appropriate.
 */
814 815
static struct buffer_head * ext4_find_entry (struct dentry *dentry,
					struct ext4_dir_entry_2 ** res_dir)
816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836
{
	struct super_block * sb;
	struct buffer_head * bh_use[NAMEI_RA_SIZE];
	struct buffer_head * bh, *ret = NULL;
	unsigned long start, block, b;
	int ra_max = 0;		/* Number of bh's in the readahead
				   buffer, bh_use[] */
	int ra_ptr = 0;		/* Current index into readahead
				   buffer */
	int num = 0;
	int nblocks, i, err;
	struct inode *dir = dentry->d_parent->d_inode;
	int namelen;
	const u8 *name;
	unsigned blocksize;

	*res_dir = NULL;
	sb = dir->i_sb;
	blocksize = sb->s_blocksize;
	namelen = dentry->d_name.len;
	name = dentry->d_name.name;
837
	if (namelen > EXT4_NAME_LEN)
838
		return NULL;
839
#ifdef CONFIG_EXT4_INDEX
840
	if (is_dx(dir)) {
841
		bh = ext4_dx_find_entry(dentry, res_dir, &err);
842 843 844 845 846 847 848
		/*
		 * On success, or if the error was file not found,
		 * return.  Otherwise, fall back to doing a search the
		 * old fashioned way.
		 */
		if (bh || (err != ERR_BAD_DX_DIR))
			return bh;
849
		dxtrace(printk("ext4_find_entry: dx failed, falling back\n"));
850 851
	}
#endif
852 853
	nblocks = dir->i_size >> EXT4_BLOCK_SIZE_BITS(sb);
	start = EXT4_I(dir)->i_dir_start_lookup;
854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876
	if (start >= nblocks)
		start = 0;
	block = start;
restart:
	do {
		/*
		 * We deal with the read-ahead logic here.
		 */
		if (ra_ptr >= ra_max) {
			/* Refill the readahead buffer */
			ra_ptr = 0;
			b = block;
			for (ra_max = 0; ra_max < NAMEI_RA_SIZE; ra_max++) {
				/*
				 * Terminate if we reach the end of the
				 * directory and must wrap, or if our
				 * search has finished at this block.
				 */
				if (b >= nblocks || (num && block == start)) {
					bh_use[ra_max] = NULL;
					break;
				}
				num++;
877
				bh = ext4_getblk(NULL, dir, b++, 0, &err);
878 879 880 881 882 883 884 885 886 887
				bh_use[ra_max] = bh;
				if (bh)
					ll_rw_block(READ_META, 1, &bh);
			}
		}
		if ((bh = bh_use[ra_ptr++]) == NULL)
			goto next;
		wait_on_buffer(bh);
		if (!buffer_uptodate(bh)) {
			/* read error, skip block & hope for the best */
888
			ext4_error(sb, __FUNCTION__, "reading directory #%lu "
889 890 891 892 893
				   "offset %lu", dir->i_ino, block);
			brelse(bh);
			goto next;
		}
		i = search_dirblock(bh, dir, dentry,
894
			    block << EXT4_BLOCK_SIZE_BITS(sb), res_dir);
895
		if (i == 1) {
896
			EXT4_I(dir)->i_dir_start_lookup = block;
897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913
			ret = bh;
			goto cleanup_and_exit;
		} else {
			brelse(bh);
			if (i < 0)
				goto cleanup_and_exit;
		}
	next:
		if (++block >= nblocks)
			block = 0;
	} while (block != start);

	/*
	 * If the directory has grown while we were searching, then
	 * search the last part of the directory before giving up.
	 */
	block = nblocks;
914
	nblocks = dir->i_size >> EXT4_BLOCK_SIZE_BITS(sb);
915 916 917 918 919 920 921 922 923 924 925 926
	if (block < nblocks) {
		start = 0;
		goto restart;
	}

cleanup_and_exit:
	/* Clean up the read-ahead blocks */
	for (; ra_ptr < ra_max; ra_ptr++)
		brelse (bh_use[ra_ptr]);
	return ret;
}

927 928 929
#ifdef CONFIG_EXT4_INDEX
static struct buffer_head * ext4_dx_find_entry(struct dentry *dentry,
		       struct ext4_dir_entry_2 **res_dir, int *err)
930 931 932 933 934
{
	struct super_block * sb;
	struct dx_hash_info	hinfo;
	u32 hash;
	struct dx_frame frames[2], *frame;
935
	struct ext4_dir_entry_2 *de, *top;
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
	struct buffer_head *bh;
	unsigned long block;
	int retval;
	int namelen = dentry->d_name.len;
	const u8 *name = dentry->d_name.name;
	struct inode *dir = dentry->d_parent->d_inode;

	sb = dir->i_sb;
	/* NFS may look up ".." - look at dx_root directory block */
	if (namelen > 2 || name[0] != '.'||(name[1] != '.' && name[1] != '\0')){
		if (!(frame = dx_probe(dentry, NULL, &hinfo, frames, err)))
			return NULL;
	} else {
		frame = frames;
		frame->bh = NULL;			/* for dx_release() */
		frame->at = (struct dx_entry *)frames;	/* hack for zero entry*/
		dx_set_block(frame->at, 0);		/* dx_root block is 0 */
	}
	hash = hinfo.hash;
	do {
		block = dx_get_block(frame->at);
957
		if (!(bh = ext4_bread (NULL,dir, block, 0, err)))
958
			goto errout;
959 960 961 962 963 964
		de = (struct ext4_dir_entry_2 *) bh->b_data;
		top = (struct ext4_dir_entry_2 *) ((char *) de + sb->s_blocksize -
				       EXT4_DIR_REC_LEN(0));
		for (; de < top; de = ext4_next_entry(de))
		if (ext4_match (namelen, name, de)) {
			if (!ext4_check_dir_entry("ext4_find_entry",
965
						  dir, de, bh,
966
				  (block<<EXT4_BLOCK_SIZE_BITS(sb))
967 968
					  +((char *)de - bh->b_data))) {
				brelse (bh);
969
				*err = ERR_BAD_DX_DIR;
970 971 972 973 974 975 976 977
				goto errout;
			}
			*res_dir = de;
			dx_release (frames);
			return bh;
		}
		brelse (bh);
		/* Check to see if we should continue to search */
978
		retval = ext4_htree_next_block(dir, hash, frame,
979 980
					       frames, NULL);
		if (retval < 0) {
981
			ext4_warning(sb, __FUNCTION__,
982 983 984 985 986 987 988 989 990 991 992 993 994 995 996
			     "error reading index page in directory #%lu",
			     dir->i_ino);
			*err = retval;
			goto errout;
		}
	} while (retval == 1);

	*err = -ENOENT;
errout:
	dxtrace(printk("%s not found\n", name));
	dx_release (frames);
	return NULL;
}
#endif

997
static struct dentry *ext4_lookup(struct inode * dir, struct dentry *dentry, struct nameidata *nd)
998 999
{
	struct inode * inode;
1000
	struct ext4_dir_entry_2 * de;
1001 1002
	struct buffer_head * bh;

1003
	if (dentry->d_name.len > EXT4_NAME_LEN)
1004 1005
		return ERR_PTR(-ENAMETOOLONG);

1006
	bh = ext4_find_entry(dentry, &de);
1007 1008 1009 1010
	inode = NULL;
	if (bh) {
		unsigned long ino = le32_to_cpu(de->inode);
		brelse (bh);
1011 1012
		if (!ext4_valid_inum(dir->i_sb, ino)) {
			ext4_error(dir->i_sb, "ext4_lookup",
1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
				   "bad inode number: %lu", ino);
			inode = NULL;
		} else
			inode = iget(dir->i_sb, ino);

		if (!inode)
			return ERR_PTR(-EACCES);
	}
	return d_splice_alias(inode, dentry);
}


1025
struct dentry *ext4_get_parent(struct dentry *child)
1026 1027 1028 1029 1030
{
	unsigned long ino;
	struct dentry *parent;
	struct inode *inode;
	struct dentry dotdot;
1031
	struct ext4_dir_entry_2 * de;
1032 1033 1034 1035 1036 1037
	struct buffer_head *bh;

	dotdot.d_name.name = "..";
	dotdot.d_name.len = 2;
	dotdot.d_parent = child; /* confusing, isn't it! */

1038
	bh = ext4_find_entry(&dotdot, &de);
1039 1040 1041 1042 1043 1044
	inode = NULL;
	if (!bh)
		return ERR_PTR(-ENOENT);
	ino = le32_to_cpu(de->inode);
	brelse(bh);

1045 1046
	if (!ext4_valid_inum(child->d_inode->i_sb, ino)) {
		ext4_error(child->d_inode->i_sb, "ext4_get_parent",
1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
			   "bad inode number: %lu", ino);
		inode = NULL;
	} else
		inode = iget(child->d_inode->i_sb, ino);

	if (!inode)
		return ERR_PTR(-EACCES);

	parent = d_alloc_anon(inode);
	if (!parent) {
		iput(inode);
		parent = ERR_PTR(-ENOMEM);
	}
	return parent;
}

#define S_SHIFT 12
1064 1065 1066 1067 1068 1069 1070 1071
static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
1072 1073
};

1074 1075
static inline void ext4_set_de_type(struct super_block *sb,
				struct ext4_dir_entry_2 *de,
1076
				umode_t mode) {
1077 1078
	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1079 1080
}

1081 1082
#ifdef CONFIG_EXT4_INDEX
static struct ext4_dir_entry_2 *
1083 1084 1085 1086 1087
dx_move_dirents(char *from, char *to, struct dx_map_entry *map, int count)
{
	unsigned rec_len = 0;

	while (count--) {
1088 1089
		struct ext4_dir_entry_2 *de = (struct ext4_dir_entry_2 *) (from + map->offs);
		rec_len = EXT4_DIR_REC_LEN(de->name_len);
1090
		memcpy (to, de, rec_len);
1091
		((struct ext4_dir_entry_2 *) to)->rec_len =
1092 1093 1094 1095 1096
				cpu_to_le16(rec_len);
		de->inode = 0;
		map++;
		to += rec_len;
	}
1097
	return (struct ext4_dir_entry_2 *) (to - rec_len);
1098 1099
}

1100
static struct ext4_dir_entry_2* dx_pack_dirents(char *base, int size)
1101
{
1102
	struct ext4_dir_entry_2 *next, *to, *prev, *de = (struct ext4_dir_entry_2 *) base;
1103 1104 1105 1106
	unsigned rec_len = 0;

	prev = to = de;
	while ((char*)de < base + size) {
1107
		next = (struct ext4_dir_entry_2 *) ((char *) de +
1108 1109
						    le16_to_cpu(de->rec_len));
		if (de->inode && de->name_len) {
1110
			rec_len = EXT4_DIR_REC_LEN(de->name_len);
1111 1112 1113 1114
			if (de > to)
				memmove(to, de, rec_len);
			to->rec_len = cpu_to_le16(rec_len);
			prev = to;
1115
			to = (struct ext4_dir_entry_2 *) (((char *) to) + rec_len);
1116 1117 1118 1119 1120 1121
		}
		de = next;
	}
	return prev;
}

1122
static struct ext4_dir_entry_2 *do_split(handle_t *handle, struct inode *dir,
1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133
			struct buffer_head **bh,struct dx_frame *frame,
			struct dx_hash_info *hinfo, int *error)
{
	unsigned blocksize = dir->i_sb->s_blocksize;
	unsigned count, continued;
	struct buffer_head *bh2;
	u32 newblock;
	u32 hash2;
	struct dx_map_entry *map;
	char *data1 = (*bh)->b_data, *data2;
	unsigned split;
1134
	struct ext4_dir_entry_2 *de = NULL, *de2;
1135
	int	err = 0;
1136

1137
	bh2 = ext4_append (handle, dir, &newblock, &err);
1138 1139 1140 1141 1142 1143 1144
	if (!(bh2)) {
		brelse(*bh);
		*bh = NULL;
		goto errout;
	}

	BUFFER_TRACE(*bh, "get_write_access");
1145
	err = ext4_journal_get_write_access(handle, *bh);
1146 1147 1148
	if (err)
		goto journal_error;

1149
	BUFFER_TRACE(frame->bh, "get_write_access");
1150
	err = ext4_journal_get_write_access(handle, frame->bh);
1151 1152 1153 1154 1155 1156 1157
	if (err)
		goto journal_error;

	data2 = bh2->b_data;

	/* create map in the end of data2 block */
	map = (struct dx_map_entry *) (data2 + blocksize);
1158
	count = dx_make_map ((struct ext4_dir_entry_2 *) data1,
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
			     blocksize, hinfo, map);
	map -= count;
	split = count/2; // need to adjust to actual middle
	dx_sort_map (map, count);
	hash2 = map[split].hash;
	continued = hash2 == map[split - 1].hash;
	dxtrace(printk("Split block %i at %x, %i/%i\n",
		dx_get_block(frame->at), hash2, split, count-split));

	/* Fancy dance to stay within two buffers */
	de2 = dx_move_dirents(data1, data2, map + split, count - split);
	de = dx_pack_dirents(data1,blocksize);
	de->rec_len = cpu_to_le16(data1 + blocksize - (char *) de);
	de2->rec_len = cpu_to_le16(data2 + blocksize - (char *) de2);
1173 1174
	dxtrace(dx_show_leaf (hinfo, (struct ext4_dir_entry_2 *) data1, blocksize, 1));
	dxtrace(dx_show_leaf (hinfo, (struct ext4_dir_entry_2 *) data2, blocksize, 1));
1175 1176 1177 1178 1179 1180 1181 1182

	/* Which block gets the new entry? */
	if (hinfo->hash >= hash2)
	{
		swap(*bh, bh2);
		de = de2;
	}
	dx_insert_block (frame, hash2 + continued, newblock);
1183
	err = ext4_journal_dirty_metadata (handle, bh2);
1184 1185
	if (err)
		goto journal_error;
1186
	err = ext4_journal_dirty_metadata (handle, frame->bh);
1187 1188 1189 1190 1191
	if (err)
		goto journal_error;
	brelse (bh2);
	dxtrace(dx_show_index ("frame", frame->entries));
	return de;
1192 1193 1194 1195 1196 1197 1198 1199 1200

journal_error:
	brelse(*bh);
	brelse(bh2);
	*bh = NULL;
	ext4_std_error(dir->i_sb, err);
errout:
	*error = err;
	return NULL;
1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
}
#endif


/*
 * Add a new entry into a directory (leaf) block.  If de is non-NULL,
 * it points to a directory entry which is guaranteed to be large
 * enough for new directory entry.  If de is NULL, then
 * add_dirent_to_buf will attempt search the directory block for
 * space.  It will return -ENOSPC if no space is available, and -EIO
 * and -EEXIST if directory entry already exists.
 *
 * NOTE!  bh is NOT released in the case where ENOSPC is returned.  In
 * all other cases bh is released.
 */
static int add_dirent_to_buf(handle_t *handle, struct dentry *dentry,
1217
			     struct inode *inode, struct ext4_dir_entry_2 *de,
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
			     struct buffer_head * bh)
{
	struct inode	*dir = dentry->d_parent->d_inode;
	const char	*name = dentry->d_name.name;
	int		namelen = dentry->d_name.len;
	unsigned long	offset = 0;
	unsigned short	reclen;
	int		nlen, rlen, err;
	char		*top;

1228
	reclen = EXT4_DIR_REC_LEN(namelen);
1229
	if (!de) {
1230
		de = (struct ext4_dir_entry_2 *)bh->b_data;
1231 1232
		top = bh->b_data + dir->i_sb->s_blocksize - reclen;
		while ((char *) de <= top) {
1233
			if (!ext4_check_dir_entry("ext4_add_entry", dir, de,
1234 1235 1236 1237
						  bh, offset)) {
				brelse (bh);
				return -EIO;
			}
1238
			if (ext4_match (namelen, name, de)) {
1239 1240 1241
				brelse (bh);
				return -EEXIST;
			}
1242
			nlen = EXT4_DIR_REC_LEN(de->name_len);
1243 1244 1245
			rlen = le16_to_cpu(de->rec_len);
			if ((de->inode? rlen - nlen: rlen) >= reclen)
				break;
1246
			de = (struct ext4_dir_entry_2 *)((char *)de + rlen);
1247 1248 1249 1250 1251 1252
			offset += rlen;
		}
		if ((char *) de > top)
			return -ENOSPC;
	}
	BUFFER_TRACE(bh, "get_write_access");
1253
	err = ext4_journal_get_write_access(handle, bh);
1254
	if (err) {
1255
		ext4_std_error(dir->i_sb, err);
1256 1257 1258 1259 1260
		brelse(bh);
		return err;
	}

	/* By now the buffer is marked for journaling */
1261
	nlen = EXT4_DIR_REC_LEN(de->name_len);
1262 1263
	rlen = le16_to_cpu(de->rec_len);
	if (de->inode) {
1264
		struct ext4_dir_entry_2 *de1 = (struct ext4_dir_entry_2 *)((char *)de + nlen);
1265 1266 1267 1268
		de1->rec_len = cpu_to_le16(rlen - nlen);
		de->rec_len = cpu_to_le16(nlen);
		de = de1;
	}
1269
	de->file_type = EXT4_FT_UNKNOWN;
1270 1271
	if (inode) {
		de->inode = cpu_to_le32(inode->i_ino);
1272
		ext4_set_de_type(dir->i_sb, de, inode->i_mode);
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282
	} else
		de->inode = 0;
	de->name_len = namelen;
	memcpy (de->name, name, namelen);
	/*
	 * XXX shouldn't update any times until successful
	 * completion of syscall, but too many callers depend
	 * on this.
	 *
	 * XXX similarly, too many callers depend on
1283
	 * ext4_new_inode() setting the times, but error
1284 1285 1286 1287 1288
	 * recovery deletes the inode, so the worst that can
	 * happen is that the times are slightly out of date
	 * and/or different from the directory change time.
	 */
	dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC;
1289
	ext4_update_dx_flag(dir);
1290
	dir->i_version++;
1291 1292 1293
	ext4_mark_inode_dirty(handle, dir);
	BUFFER_TRACE(bh, "call ext4_journal_dirty_metadata");
	err = ext4_journal_dirty_metadata(handle, bh);
1294
	if (err)
1295
		ext4_std_error(dir->i_sb, err);
1296 1297 1298 1299
	brelse(bh);
	return 0;
}

1300
#ifdef CONFIG_EXT4_INDEX
1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
/*
 * This converts a one block unindexed directory to a 3 block indexed
 * directory, and adds the dentry to the indexed directory.
 */
static int make_indexed_dir(handle_t *handle, struct dentry *dentry,
			    struct inode *inode, struct buffer_head *bh)
{
	struct inode	*dir = dentry->d_parent->d_inode;
	const char	*name = dentry->d_name.name;
	int		namelen = dentry->d_name.len;
	struct buffer_head *bh2;
	struct dx_root	*root;
	struct dx_frame	frames[2], *frame;
	struct dx_entry *entries;
1315
	struct ext4_dir_entry_2	*de, *de2;
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
	char		*data1, *top;
	unsigned	len;
	int		retval;
	unsigned	blocksize;
	struct dx_hash_info hinfo;
	u32		block;
	struct fake_dirent *fde;

	blocksize =  dir->i_sb->s_blocksize;
	dxtrace(printk("Creating index\n"));
1326
	retval = ext4_journal_get_write_access(handle, bh);
1327
	if (retval) {
1328
		ext4_std_error(dir->i_sb, retval);
1329 1330 1331 1332 1333
		brelse(bh);
		return retval;
	}
	root = (struct dx_root *) bh->b_data;

1334
	bh2 = ext4_append (handle, dir, &block, &retval);
1335 1336 1337 1338
	if (!(bh2)) {
		brelse(bh);
		return retval;
	}
1339
	EXT4_I(dir)->i_flags |= EXT4_INDEX_FL;
1340 1341 1342 1343
	data1 = bh2->b_data;

	/* The 0th block becomes the root, move the dirents out */
	fde = &root->dotdot;
1344
	de = (struct ext4_dir_entry_2 *)((char *)fde + le16_to_cpu(fde->rec_len));
1345 1346
	len = ((char *) root) + blocksize - (char *) de;
	memcpy (data1, de, len);
1347
	de = (struct ext4_dir_entry_2 *) data1;
1348 1349 1350 1351 1352
	top = data1 + len;
	while ((char *)(de2=(void*)de+le16_to_cpu(de->rec_len)) < top)
		de = de2;
	de->rec_len = cpu_to_le16(data1 + blocksize - (char *) de);
	/* Initialize the root; the dot dirents already exist */
1353 1354
	de = (struct ext4_dir_entry_2 *) (&root->dotdot);
	de->rec_len = cpu_to_le16(blocksize - EXT4_DIR_REC_LEN(2));
1355 1356
	memset (&root->info, 0, sizeof(root->info));
	root->info.info_length = sizeof(root->info);
1357
	root->info.hash_version = EXT4_SB(dir->i_sb)->s_def_hash_version;
1358 1359 1360 1361 1362 1363 1364
	entries = root->entries;
	dx_set_block (entries, 1);
	dx_set_count (entries, 1);
	dx_set_limit (entries, dx_root_limit(dir, sizeof(root->info)));

	/* Initialize as for dx_probe */
	hinfo.hash_version = root->info.hash_version;
1365 1366
	hinfo.seed = EXT4_SB(dir->i_sb)->s_hash_seed;
	ext4fs_dirhash(name, namelen, &hinfo);
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381
	frame = frames;
	frame->entries = entries;
	frame->at = entries;
	frame->bh = bh;
	bh = bh2;
	de = do_split(handle,dir, &bh, frame, &hinfo, &retval);
	dx_release (frames);
	if (!(de))
		return retval;

	return add_dirent_to_buf(handle, dentry, inode, de, bh);
}
#endif

/*
1382
 *	ext4_add_entry()
1383 1384
 *
 * adds a file entry to the specified directory, using the same
1385
 * semantics as ext4_find_entry(). It returns NULL if it failed.
1386 1387 1388 1389 1390
 *
 * NOTE!! The inode part of 'de' is left at 0 - which means you
 * may not sleep between calling this and putting something into
 * the entry, as someone else might have used it while you slept.
 */
1391
static int ext4_add_entry (handle_t *handle, struct dentry *dentry,
1392 1393 1394 1395 1396
	struct inode *inode)
{
	struct inode *dir = dentry->d_parent->d_inode;
	unsigned long offset;
	struct buffer_head * bh;
1397
	struct ext4_dir_entry_2 *de;
1398 1399
	struct super_block * sb;
	int	retval;
1400
#ifdef CONFIG_EXT4_INDEX
1401 1402 1403 1404 1405 1406 1407 1408 1409
	int	dx_fallback=0;
#endif
	unsigned blocksize;
	u32 block, blocks;

	sb = dir->i_sb;
	blocksize = sb->s_blocksize;
	if (!dentry->d_name.len)
		return -EINVAL;
1410
#ifdef CONFIG_EXT4_INDEX
1411
	if (is_dx(dir)) {
1412
		retval = ext4_dx_add_entry(handle, dentry, inode);
1413 1414
		if (!retval || (retval != ERR_BAD_DX_DIR))
			return retval;
1415
		EXT4_I(dir)->i_flags &= ~EXT4_INDEX_FL;
1416
		dx_fallback++;
1417
		ext4_mark_inode_dirty(handle, dir);
1418 1419 1420 1421
	}
#endif
	blocks = dir->i_size >> sb->s_blocksize_bits;
	for (block = 0, offset = 0; block < blocks; block++) {
1422
		bh = ext4_bread(handle, dir, block, 0, &retval);
1423 1424 1425 1426 1427 1428
		if(!bh)
			return retval;
		retval = add_dirent_to_buf(handle, dentry, inode, NULL, bh);
		if (retval != -ENOSPC)
			return retval;

1429
#ifdef CONFIG_EXT4_INDEX
1430
		if (blocks == 1 && !dx_fallback &&
1431
		    EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_DIR_INDEX))
1432 1433 1434 1435
			return make_indexed_dir(handle, dentry, inode, bh);
#endif
		brelse(bh);
	}
1436
	bh = ext4_append(handle, dir, &block, &retval);
1437 1438
	if (!bh)
		return retval;
1439
	de = (struct ext4_dir_entry_2 *) bh->b_data;
1440 1441 1442 1443 1444
	de->inode = 0;
	de->rec_len = cpu_to_le16(blocksize);
	return add_dirent_to_buf(handle, dentry, inode, de, bh);
}

1445
#ifdef CONFIG_EXT4_INDEX
1446 1447 1448
/*
 * Returns 0 for success, or a negative error value
 */
1449
static int ext4_dx_add_entry(handle_t *handle, struct dentry *dentry,
1450 1451 1452 1453 1454 1455 1456 1457
			     struct inode *inode)
{
	struct dx_frame frames[2], *frame;
	struct dx_entry *entries, *at;
	struct dx_hash_info hinfo;
	struct buffer_head * bh;
	struct inode *dir = dentry->d_parent->d_inode;
	struct super_block * sb = dir->i_sb;
1458
	struct ext4_dir_entry_2 *de;
1459 1460 1461 1462 1463 1464 1465 1466
	int err;

	frame = dx_probe(dentry, NULL, &hinfo, frames, &err);
	if (!frame)
		return err;
	entries = frame->entries;
	at = frame->at;

1467
	if (!(bh = ext4_bread(handle,dir, dx_get_block(frame->at), 0, &err)))
1468 1469 1470
		goto cleanup;

	BUFFER_TRACE(bh, "get_write_access");
1471
	err = ext4_journal_get_write_access(handle, bh);
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494
	if (err)
		goto journal_error;

	err = add_dirent_to_buf(handle, dentry, inode, NULL, bh);
	if (err != -ENOSPC) {
		bh = NULL;
		goto cleanup;
	}

	/* Block full, should compress but for now just split */
	dxtrace(printk("using %u of %u node entries\n",
		       dx_get_count(entries), dx_get_limit(entries)));
	/* Need to split index? */
	if (dx_get_count(entries) == dx_get_limit(entries)) {
		u32 newblock;
		unsigned icount = dx_get_count(entries);
		int levels = frame - frames;
		struct dx_entry *entries2;
		struct dx_node *node2;
		struct buffer_head *bh2;

		if (levels && (dx_get_count(frames->entries) ==
			       dx_get_limit(frames->entries))) {
1495
			ext4_warning(sb, __FUNCTION__,
1496 1497 1498 1499
				     "Directory index full!");
			err = -ENOSPC;
			goto cleanup;
		}
1500
		bh2 = ext4_append (handle, dir, &newblock, &err);
1501 1502 1503 1504 1505 1506 1507
		if (!(bh2))
			goto cleanup;
		node2 = (struct dx_node *)(bh2->b_data);
		entries2 = node2->entries;
		node2->fake.rec_len = cpu_to_le16(sb->s_blocksize);
		node2->fake.inode = 0;
		BUFFER_TRACE(frame->bh, "get_write_access");
1508
		err = ext4_journal_get_write_access(handle, frame->bh);
1509 1510 1511 1512 1513 1514 1515 1516
		if (err)
			goto journal_error;
		if (levels) {
			unsigned icount1 = icount/2, icount2 = icount - icount1;
			unsigned hash2 = dx_get_hash(entries + icount1);
			dxtrace(printk("Split index %i/%i\n", icount1, icount2));

			BUFFER_TRACE(frame->bh, "get_write_access"); /* index root */
1517
			err = ext4_journal_get_write_access(handle,
1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537
							     frames[0].bh);
			if (err)
				goto journal_error;

			memcpy ((char *) entries2, (char *) (entries + icount1),
				icount2 * sizeof(struct dx_entry));
			dx_set_count (entries, icount1);
			dx_set_count (entries2, icount2);
			dx_set_limit (entries2, dx_node_limit(dir));

			/* Which index block gets the new entry? */
			if (at - entries >= icount1) {
				frame->at = at = at - entries - icount1 + entries2;
				frame->entries = entries = entries2;
				swap(frame->bh, bh2);
			}
			dx_insert_block (frames + 0, hash2, newblock);
			dxtrace(dx_show_index ("node", frames[1].entries));
			dxtrace(dx_show_index ("node",
			       ((struct dx_node *) bh2->b_data)->entries));
1538
			err = ext4_journal_dirty_metadata(handle, bh2);
1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557
			if (err)
				goto journal_error;
			brelse (bh2);
		} else {
			dxtrace(printk("Creating second level index...\n"));
			memcpy((char *) entries2, (char *) entries,
			       icount * sizeof(struct dx_entry));
			dx_set_limit(entries2, dx_node_limit(dir));

			/* Set up root */
			dx_set_count(entries, 1);
			dx_set_block(entries + 0, newblock);
			((struct dx_root *) frames[0].bh->b_data)->info.indirect_levels = 1;

			/* Add new access path frame */
			frame = frames + 1;
			frame->at = at = at - entries + entries2;
			frame->entries = entries = entries2;
			frame->bh = bh2;
1558
			err = ext4_journal_get_write_access(handle,
1559 1560 1561 1562
							     frame->bh);
			if (err)
				goto journal_error;
		}
1563
		ext4_journal_dirty_metadata(handle, frames[0].bh);
1564 1565 1566 1567 1568 1569 1570 1571 1572
	}
	de = do_split(handle, dir, &bh, frame, &hinfo, &err);
	if (!de)
		goto cleanup;
	err = add_dirent_to_buf(handle, dentry, inode, de, bh);
	bh = NULL;
	goto cleanup;

journal_error:
1573
	ext4_std_error(dir->i_sb, err);
1574 1575 1576 1577 1578 1579 1580 1581 1582
cleanup:
	if (bh)
		brelse(bh);
	dx_release(frames);
	return err;
}
#endif

/*
1583
 * ext4_delete_entry deletes a directory entry by merging it with the
1584 1585
 * previous entry
 */
1586
static int ext4_delete_entry (handle_t *handle,
1587
			      struct inode * dir,
1588
			      struct ext4_dir_entry_2 * de_del,
1589 1590
			      struct buffer_head * bh)
{
1591
	struct ext4_dir_entry_2 * de, * pde;
1592 1593 1594 1595
	int i;

	i = 0;
	pde = NULL;
1596
	de = (struct ext4_dir_entry_2 *) bh->b_data;
1597
	while (i < bh->b_size) {
1598
		if (!ext4_check_dir_entry("ext4_delete_entry", dir, de, bh, i))
1599 1600 1601
			return -EIO;
		if (de == de_del)  {
			BUFFER_TRACE(bh, "get_write_access");
1602
			ext4_journal_get_write_access(handle, bh);
1603 1604 1605 1606 1607 1608 1609
			if (pde)
				pde->rec_len =
					cpu_to_le16(le16_to_cpu(pde->rec_len) +
						    le16_to_cpu(de->rec_len));
			else
				de->inode = 0;
			dir->i_version++;
1610 1611
			BUFFER_TRACE(bh, "call ext4_journal_dirty_metadata");
			ext4_journal_dirty_metadata(handle, bh);
1612 1613 1614 1615
			return 0;
		}
		i += le16_to_cpu(de->rec_len);
		pde = de;
1616
		de = (struct ext4_dir_entry_2 *)
1617 1618 1619 1620 1621
			((char *) de + le16_to_cpu(de->rec_len));
	}
	return -ENOENT;
}

1622
static int ext4_add_nondir(handle_t *handle,
1623 1624
		struct dentry *dentry, struct inode *inode)
{
1625
	int err = ext4_add_entry(handle, dentry, inode);
1626
	if (!err) {
1627
		ext4_mark_inode_dirty(handle, inode);
1628 1629 1630
		d_instantiate(dentry, inode);
		return 0;
	}
1631
	drop_nlink(inode);
1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
	iput(inode);
	return err;
}

/*
 * By the time this is called, we already have created
 * the directory cache entry for the new file, but it
 * is so far negative - it has no inode.
 *
 * If the create succeeds, we fill in the inode information
 * with d_instantiate().
 */
1644
static int ext4_create (struct inode * dir, struct dentry * dentry, int mode,
1645 1646 1647 1648 1649 1650 1651
		struct nameidata *nd)
{
	handle_t *handle;
	struct inode * inode;
	int err, retries = 0;

retry:
1652 1653 1654
	handle = ext4_journal_start(dir, EXT4_DATA_TRANS_BLOCKS(dir->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 3 +
					2*EXT4_QUOTA_INIT_BLOCKS(dir->i_sb));
1655 1656 1657 1658 1659 1660
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

1661
	inode = ext4_new_inode (handle, dir, mode);
1662 1663
	err = PTR_ERR(inode);
	if (!IS_ERR(inode)) {
1664 1665 1666 1667
		inode->i_op = &ext4_file_inode_operations;
		inode->i_fop = &ext4_file_operations;
		ext4_set_aops(inode);
		err = ext4_add_nondir(handle, dentry, inode);
1668
	}
1669 1670
	ext4_journal_stop(handle);
	if (err == -ENOSPC && ext4_should_retry_alloc(dir->i_sb, &retries))
1671 1672 1673 1674
		goto retry;
	return err;
}

1675
static int ext4_mknod (struct inode * dir, struct dentry *dentry,
1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
			int mode, dev_t rdev)
{
	handle_t *handle;
	struct inode *inode;
	int err, retries = 0;

	if (!new_valid_dev(rdev))
		return -EINVAL;

retry:
1686 1687 1688
	handle = ext4_journal_start(dir, EXT4_DATA_TRANS_BLOCKS(dir->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 3 +
					2*EXT4_QUOTA_INIT_BLOCKS(dir->i_sb));
1689 1690 1691 1692 1693 1694
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

1695
	inode = ext4_new_inode (handle, dir, mode);
1696 1697 1698
	err = PTR_ERR(inode);
	if (!IS_ERR(inode)) {
		init_special_inode(inode, inode->i_mode, rdev);
1699 1700
#ifdef CONFIG_EXT4DEV_FS_XATTR
		inode->i_op = &ext4_special_inode_operations;
1701
#endif
1702
		err = ext4_add_nondir(handle, dentry, inode);
1703
	}
1704 1705
	ext4_journal_stop(handle);
	if (err == -ENOSPC && ext4_should_retry_alloc(dir->i_sb, &retries))
1706 1707 1708 1709
		goto retry;
	return err;
}

1710
static int ext4_mkdir(struct inode * dir, struct dentry * dentry, int mode)
1711 1712 1713 1714
{
	handle_t *handle;
	struct inode * inode;
	struct buffer_head * dir_block;
1715
	struct ext4_dir_entry_2 * de;
1716 1717
	int err, retries = 0;

1718
	if (dir->i_nlink >= EXT4_LINK_MAX)
1719 1720 1721
		return -EMLINK;

retry:
1722 1723 1724
	handle = ext4_journal_start(dir, EXT4_DATA_TRANS_BLOCKS(dir->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 3 +
					2*EXT4_QUOTA_INIT_BLOCKS(dir->i_sb));
1725 1726 1727 1728 1729 1730
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

1731
	inode = ext4_new_inode (handle, dir, S_IFDIR | mode);
1732 1733 1734 1735
	err = PTR_ERR(inode);
	if (IS_ERR(inode))
		goto out_stop;

1736 1737 1738 1739
	inode->i_op = &ext4_dir_inode_operations;
	inode->i_fop = &ext4_dir_operations;
	inode->i_size = EXT4_I(inode)->i_disksize = inode->i_sb->s_blocksize;
	dir_block = ext4_bread (handle, inode, 0, 1, &err);
1740 1741
	if (!dir_block) {
		drop_nlink(inode); /* is this nlink == 0? */
1742
		ext4_mark_inode_dirty(handle, inode);
1743 1744 1745 1746
		iput (inode);
		goto out_stop;
	}
	BUFFER_TRACE(dir_block, "get_write_access");
1747 1748
	ext4_journal_get_write_access(handle, dir_block);
	de = (struct ext4_dir_entry_2 *) dir_block->b_data;
1749 1750
	de->inode = cpu_to_le32(inode->i_ino);
	de->name_len = 1;
1751
	de->rec_len = cpu_to_le16(EXT4_DIR_REC_LEN(de->name_len));
1752
	strcpy (de->name, ".");
1753 1754
	ext4_set_de_type(dir->i_sb, de, S_IFDIR);
	de = (struct ext4_dir_entry_2 *)
1755 1756
			((char *) de + le16_to_cpu(de->rec_len));
	de->inode = cpu_to_le32(dir->i_ino);
1757
	de->rec_len = cpu_to_le16(inode->i_sb->s_blocksize-EXT4_DIR_REC_LEN(1));
1758 1759
	de->name_len = 2;
	strcpy (de->name, "..");
1760
	ext4_set_de_type(dir->i_sb, de, S_IFDIR);
1761
	inode->i_nlink = 2;
1762 1763
	BUFFER_TRACE(dir_block, "call ext4_journal_dirty_metadata");
	ext4_journal_dirty_metadata(handle, dir_block);
1764
	brelse (dir_block);
1765 1766
	ext4_mark_inode_dirty(handle, inode);
	err = ext4_add_entry (handle, dentry, inode);
1767 1768
	if (err) {
		inode->i_nlink = 0;
1769
		ext4_mark_inode_dirty(handle, inode);
1770 1771 1772 1773
		iput (inode);
		goto out_stop;
	}
	inc_nlink(dir);
1774 1775
	ext4_update_dx_flag(dir);
	ext4_mark_inode_dirty(handle, dir);
1776 1777
	d_instantiate(dentry, inode);
out_stop:
1778 1779
	ext4_journal_stop(handle);
	if (err == -ENOSPC && ext4_should_retry_alloc(dir->i_sb, &retries))
1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790
		goto retry;
	return err;
}

/*
 * routine to check that the specified directory is empty (for rmdir)
 */
static int empty_dir (struct inode * inode)
{
	unsigned long offset;
	struct buffer_head * bh;
1791
	struct ext4_dir_entry_2 * de, * de1;
1792 1793 1794 1795
	struct super_block * sb;
	int err = 0;

	sb = inode->i_sb;
1796 1797
	if (inode->i_size < EXT4_DIR_REC_LEN(1) + EXT4_DIR_REC_LEN(2) ||
	    !(bh = ext4_bread (NULL, inode, 0, 0, &err))) {
1798
		if (err)
1799
			ext4_error(inode->i_sb, __FUNCTION__,
1800 1801 1802
				   "error %d reading directory #%lu offset 0",
				   err, inode->i_ino);
		else
1803
			ext4_warning(inode->i_sb, __FUNCTION__,
1804 1805 1806 1807
				     "bad directory (dir #%lu) - no data block",
				     inode->i_ino);
		return 1;
	}
1808 1809
	de = (struct ext4_dir_entry_2 *) bh->b_data;
	de1 = (struct ext4_dir_entry_2 *)
1810 1811 1812 1813 1814
			((char *) de + le16_to_cpu(de->rec_len));
	if (le32_to_cpu(de->inode) != inode->i_ino ||
			!le32_to_cpu(de1->inode) ||
			strcmp (".", de->name) ||
			strcmp ("..", de1->name)) {
1815
		ext4_warning (inode->i_sb, "empty_dir",
1816 1817 1818 1819 1820 1821
			      "bad directory (dir #%lu) - no `.' or `..'",
			      inode->i_ino);
		brelse (bh);
		return 1;
	}
	offset = le16_to_cpu(de->rec_len) + le16_to_cpu(de1->rec_len);
1822
	de = (struct ext4_dir_entry_2 *)
1823 1824 1825 1826 1827 1828
			((char *) de1 + le16_to_cpu(de1->rec_len));
	while (offset < inode->i_size ) {
		if (!bh ||
			(void *) de >= (void *) (bh->b_data+sb->s_blocksize)) {
			err = 0;
			brelse (bh);
1829 1830
			bh = ext4_bread (NULL, inode,
				offset >> EXT4_BLOCK_SIZE_BITS(sb), 0, &err);
1831 1832
			if (!bh) {
				if (err)
1833
					ext4_error(sb, __FUNCTION__,
1834 1835 1836 1837 1838 1839
						   "error %d reading directory"
						   " #%lu offset %lu",
						   err, inode->i_ino, offset);
				offset += sb->s_blocksize;
				continue;
			}
1840
			de = (struct ext4_dir_entry_2 *) bh->b_data;
1841
		}
1842 1843
		if (!ext4_check_dir_entry("empty_dir", inode, de, bh, offset)) {
			de = (struct ext4_dir_entry_2 *)(bh->b_data +
1844 1845 1846 1847 1848 1849 1850 1851 1852
							 sb->s_blocksize);
			offset = (offset | (sb->s_blocksize - 1)) + 1;
			continue;
		}
		if (le32_to_cpu(de->inode)) {
			brelse (bh);
			return 0;
		}
		offset += le16_to_cpu(de->rec_len);
1853
		de = (struct ext4_dir_entry_2 *)
1854 1855 1856 1857 1858 1859
				((char *) de + le16_to_cpu(de->rec_len));
	}
	brelse (bh);
	return 1;
}

1860
/* ext4_orphan_add() links an unlinked or truncated inode into a list of
1861 1862 1863 1864 1865
 * such inodes, starting at the superblock, in case we crash before the
 * file is closed/deleted, or in case the inode truncate spans multiple
 * transactions and the last transaction is not recovered after a crash.
 *
 * At filesystem recovery time, we walk this list deleting unlinked
1866
 * inodes and truncating linked inodes in ext4_orphan_cleanup().
1867
 */
1868
int ext4_orphan_add(handle_t *handle, struct inode *inode)
1869 1870
{
	struct super_block *sb = inode->i_sb;
1871
	struct ext4_iloc iloc;
1872 1873 1874
	int err = 0, rc;

	lock_super(sb);
1875
	if (!list_empty(&EXT4_I(inode)->i_orphan))
1876 1877 1878 1879 1880 1881
		goto out_unlock;

	/* Orphan handling is only valid for files with data blocks
	 * being truncated, or files being unlinked. */

	/* @@@ FIXME: Observation from aviro:
1882 1883
	 * I think I can trigger J_ASSERT in ext4_orphan_add().  We block
	 * here (on lock_super()), so race with ext4_link() which might bump
1884 1885 1886 1887 1888 1889
	 * ->i_nlink. For, say it, character device. Not a regular file,
	 * not a directory, not a symlink and ->i_nlink > 0.
	 */
	J_ASSERT ((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
		S_ISLNK(inode->i_mode)) || inode->i_nlink == 0);

1890 1891
	BUFFER_TRACE(EXT4_SB(sb)->s_sbh, "get_write_access");
	err = ext4_journal_get_write_access(handle, EXT4_SB(sb)->s_sbh);
1892 1893 1894
	if (err)
		goto out_unlock;

1895
	err = ext4_reserve_inode_write(handle, inode, &iloc);
1896 1897 1898 1899
	if (err)
		goto out_unlock;

	/* Insert this inode at the head of the on-disk orphan list... */
1900 1901 1902 1903
	NEXT_ORPHAN(inode) = le32_to_cpu(EXT4_SB(sb)->s_es->s_last_orphan);
	EXT4_SB(sb)->s_es->s_last_orphan = cpu_to_le32(inode->i_ino);
	err = ext4_journal_dirty_metadata(handle, EXT4_SB(sb)->s_sbh);
	rc = ext4_mark_iloc_dirty(handle, inode, &iloc);
1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915
	if (!err)
		err = rc;

	/* Only add to the head of the in-memory list if all the
	 * previous operations succeeded.  If the orphan_add is going to
	 * fail (possibly taking the journal offline), we can't risk
	 * leaving the inode on the orphan list: stray orphan-list
	 * entries can cause panics at unmount time.
	 *
	 * This is safe: on error we're going to ignore the orphan list
	 * anyway on the next recovery. */
	if (!err)
1916
		list_add(&EXT4_I(inode)->i_orphan, &EXT4_SB(sb)->s_orphan);
1917 1918 1919 1920 1921 1922

	jbd_debug(4, "superblock will point to %lu\n", inode->i_ino);
	jbd_debug(4, "orphan inode %lu will point to %d\n",
			inode->i_ino, NEXT_ORPHAN(inode));
out_unlock:
	unlock_super(sb);
1923
	ext4_std_error(inode->i_sb, err);
1924 1925 1926 1927
	return err;
}

/*
1928
 * ext4_orphan_del() removes an unlinked or truncated inode from the list
1929 1930
 * of such inodes stored on disk, because it is finally being cleaned up.
 */
1931
int ext4_orphan_del(handle_t *handle, struct inode *inode)
1932 1933
{
	struct list_head *prev;
1934 1935
	struct ext4_inode_info *ei = EXT4_I(inode);
	struct ext4_sb_info *sbi;
1936
	unsigned long ino_next;
1937
	struct ext4_iloc iloc;
1938 1939 1940 1941 1942 1943 1944 1945 1946 1947
	int err = 0;

	lock_super(inode->i_sb);
	if (list_empty(&ei->i_orphan)) {
		unlock_super(inode->i_sb);
		return 0;
	}

	ino_next = NEXT_ORPHAN(inode);
	prev = ei->i_orphan.prev;
1948
	sbi = EXT4_SB(inode->i_sb);
1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960

	jbd_debug(4, "remove inode %lu from orphan list\n", inode->i_ino);

	list_del_init(&ei->i_orphan);

	/* If we're on an error path, we may not have a valid
	 * transaction handle with which to update the orphan list on
	 * disk, but we still need to remove the inode from the linked
	 * list in memory. */
	if (!handle)
		goto out;

1961
	err = ext4_reserve_inode_write(handle, inode, &iloc);
1962 1963 1964 1965 1966 1967
	if (err)
		goto out_err;

	if (prev == &sbi->s_orphan) {
		jbd_debug(4, "superblock will point to %lu\n", ino_next);
		BUFFER_TRACE(sbi->s_sbh, "get_write_access");
1968
		err = ext4_journal_get_write_access(handle, sbi->s_sbh);
1969 1970 1971
		if (err)
			goto out_brelse;
		sbi->s_es->s_last_orphan = cpu_to_le32(ino_next);
1972
		err = ext4_journal_dirty_metadata(handle, sbi->s_sbh);
1973
	} else {
1974
		struct ext4_iloc iloc2;
1975
		struct inode *i_prev =
1976
			&list_entry(prev, struct ext4_inode_info, i_orphan)->vfs_inode;
1977 1978 1979

		jbd_debug(4, "orphan inode %lu will point to %lu\n",
			  i_prev->i_ino, ino_next);
1980
		err = ext4_reserve_inode_write(handle, i_prev, &iloc2);
1981 1982 1983
		if (err)
			goto out_brelse;
		NEXT_ORPHAN(i_prev) = ino_next;
1984
		err = ext4_mark_iloc_dirty(handle, i_prev, &iloc2);
1985 1986 1987 1988
	}
	if (err)
		goto out_brelse;
	NEXT_ORPHAN(inode) = 0;
1989
	err = ext4_mark_iloc_dirty(handle, inode, &iloc);
1990 1991

out_err:
1992
	ext4_std_error(inode->i_sb, err);
1993 1994 1995 1996 1997 1998 1999 2000 2001
out:
	unlock_super(inode->i_sb);
	return err;

out_brelse:
	brelse(iloc.bh);
	goto out_err;
}

2002
static int ext4_rmdir (struct inode * dir, struct dentry *dentry)
2003 2004 2005 2006
{
	int retval;
	struct inode * inode;
	struct buffer_head * bh;
2007
	struct ext4_dir_entry_2 * de;
2008 2009 2010 2011 2012
	handle_t *handle;

	/* Initialize quotas before so that eventual writes go in
	 * separate transaction */
	DQUOT_INIT(dentry->d_inode);
2013
	handle = ext4_journal_start(dir, EXT4_DELETE_TRANS_BLOCKS(dir->i_sb));
2014 2015 2016 2017
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	retval = -ENOENT;
2018
	bh = ext4_find_entry (dentry, &de);
2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034
	if (!bh)
		goto end_rmdir;

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

	inode = dentry->d_inode;

	retval = -EIO;
	if (le32_to_cpu(de->inode) != inode->i_ino)
		goto end_rmdir;

	retval = -ENOTEMPTY;
	if (!empty_dir (inode))
		goto end_rmdir;

2035
	retval = ext4_delete_entry(handle, dir, de, bh);
2036 2037 2038
	if (retval)
		goto end_rmdir;
	if (inode->i_nlink != 2)
2039
		ext4_warning (inode->i_sb, "ext4_rmdir",
2040 2041 2042 2043 2044 2045 2046 2047
			      "empty directory has nlink!=2 (%d)",
			      inode->i_nlink);
	inode->i_version++;
	clear_nlink(inode);
	/* There's no need to set i_disksize: the fact that i_nlink is
	 * zero will ensure that the right thing happens during any
	 * recovery. */
	inode->i_size = 0;
2048
	ext4_orphan_add(handle, inode);
2049
	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME_SEC;
2050
	ext4_mark_inode_dirty(handle, inode);
2051
	drop_nlink(dir);
2052 2053
	ext4_update_dx_flag(dir);
	ext4_mark_inode_dirty(handle, dir);
2054 2055

end_rmdir:
2056
	ext4_journal_stop(handle);
2057 2058 2059 2060
	brelse (bh);
	return retval;
}

2061
static int ext4_unlink(struct inode * dir, struct dentry *dentry)
2062 2063 2064 2065
{
	int retval;
	struct inode * inode;
	struct buffer_head * bh;
2066
	struct ext4_dir_entry_2 * de;
2067 2068 2069 2070 2071
	handle_t *handle;

	/* Initialize quotas before so that eventual writes go
	 * in separate transaction */
	DQUOT_INIT(dentry->d_inode);
2072
	handle = ext4_journal_start(dir, EXT4_DELETE_TRANS_BLOCKS(dir->i_sb));
2073 2074 2075 2076 2077 2078 2079
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

	retval = -ENOENT;
2080
	bh = ext4_find_entry (dentry, &de);
2081 2082 2083 2084 2085 2086 2087 2088 2089 2090
	if (!bh)
		goto end_unlink;

	inode = dentry->d_inode;

	retval = -EIO;
	if (le32_to_cpu(de->inode) != inode->i_ino)
		goto end_unlink;

	if (!inode->i_nlink) {
2091
		ext4_warning (inode->i_sb, "ext4_unlink",
2092 2093 2094 2095
			      "Deleting nonexistent file (%lu), %d",
			      inode->i_ino, inode->i_nlink);
		inode->i_nlink = 1;
	}
2096
	retval = ext4_delete_entry(handle, dir, de, bh);
2097 2098 2099
	if (retval)
		goto end_unlink;
	dir->i_ctime = dir->i_mtime = CURRENT_TIME_SEC;
2100 2101
	ext4_update_dx_flag(dir);
	ext4_mark_inode_dirty(handle, dir);
2102 2103
	drop_nlink(inode);
	if (!inode->i_nlink)
2104
		ext4_orphan_add(handle, inode);
2105
	inode->i_ctime = dir->i_ctime;
2106
	ext4_mark_inode_dirty(handle, inode);
2107 2108 2109
	retval = 0;

end_unlink:
2110
	ext4_journal_stop(handle);
2111 2112 2113 2114
	brelse (bh);
	return retval;
}

2115
static int ext4_symlink (struct inode * dir,
2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126
		struct dentry *dentry, const char * symname)
{
	handle_t *handle;
	struct inode * inode;
	int l, err, retries = 0;

	l = strlen(symname)+1;
	if (l > dir->i_sb->s_blocksize)
		return -ENAMETOOLONG;

retry:
2127 2128 2129
	handle = ext4_journal_start(dir, EXT4_DATA_TRANS_BLOCKS(dir->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 5 +
					2*EXT4_QUOTA_INIT_BLOCKS(dir->i_sb));
2130 2131 2132 2133 2134 2135
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

2136
	inode = ext4_new_inode (handle, dir, S_IFLNK|S_IRWXUGO);
2137 2138 2139 2140
	err = PTR_ERR(inode);
	if (IS_ERR(inode))
		goto out_stop;

2141 2142 2143
	if (l > sizeof (EXT4_I(inode)->i_data)) {
		inode->i_op = &ext4_symlink_inode_operations;
		ext4_set_aops(inode);
2144
		/*
2145
		 * page_symlink() calls into ext4_prepare/commit_write.
2146 2147 2148 2149 2150 2151
		 * We have a transaction open.  All is sweetness.  It also sets
		 * i_size in generic_commit_write().
		 */
		err = __page_symlink(inode, symname, l,
				mapping_gfp_mask(inode->i_mapping) & ~__GFP_FS);
		if (err) {
2152
			drop_nlink(inode);
2153
			ext4_mark_inode_dirty(handle, inode);
2154 2155 2156 2157
			iput (inode);
			goto out_stop;
		}
	} else {
2158 2159
		inode->i_op = &ext4_fast_symlink_inode_operations;
		memcpy((char*)&EXT4_I(inode)->i_data,symname,l);
2160 2161
		inode->i_size = l-1;
	}
2162 2163
	EXT4_I(inode)->i_disksize = inode->i_size;
	err = ext4_add_nondir(handle, dentry, inode);
2164
out_stop:
2165 2166
	ext4_journal_stop(handle);
	if (err == -ENOSPC && ext4_should_retry_alloc(dir->i_sb, &retries))
2167 2168 2169 2170
		goto retry;
	return err;
}

2171
static int ext4_link (struct dentry * old_dentry,
2172 2173 2174 2175 2176 2177
		struct inode * dir, struct dentry *dentry)
{
	handle_t *handle;
	struct inode *inode = old_dentry->d_inode;
	int err, retries = 0;

2178
	if (inode->i_nlink >= EXT4_LINK_MAX)
2179
		return -EMLINK;
2180 2181 2182 2183 2184 2185
	/*
	 * Return -ENOENT if we've raced with unlink and i_nlink is 0.  Doing
	 * otherwise has the potential to corrupt the orphan inode list.
	 */
	if (inode->i_nlink == 0)
		return -ENOENT;
2186 2187

retry:
2188 2189
	handle = ext4_journal_start(dir, EXT4_DATA_TRANS_BLOCKS(dir->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS);
2190 2191 2192 2193 2194 2195 2196
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(dir))
		handle->h_sync = 1;

	inode->i_ctime = CURRENT_TIME_SEC;
2197
	inc_nlink(inode);
2198 2199
	atomic_inc(&inode->i_count);

2200 2201 2202
	err = ext4_add_nondir(handle, dentry, inode);
	ext4_journal_stop(handle);
	if (err == -ENOSPC && ext4_should_retry_alloc(dir->i_sb, &retries))
2203 2204 2205 2206 2207
		goto retry;
	return err;
}

#define PARENT_INO(buffer) \
2208 2209
	((struct ext4_dir_entry_2 *) ((char *) buffer + \
	le16_to_cpu(((struct ext4_dir_entry_2 *) buffer)->rec_len)))->inode
2210 2211 2212 2213 2214

/*
 * Anybody can rename anything with this: the permission checks are left to the
 * higher-level routines.
 */
2215
static int ext4_rename (struct inode * old_dir, struct dentry *old_dentry,
2216 2217 2218 2219 2220
			   struct inode * new_dir,struct dentry *new_dentry)
{
	handle_t *handle;
	struct inode * old_inode, * new_inode;
	struct buffer_head * old_bh, * new_bh, * dir_bh;
2221
	struct ext4_dir_entry_2 * old_de, * new_de;
2222 2223 2224 2225 2226 2227 2228 2229
	int retval;

	old_bh = new_bh = dir_bh = NULL;

	/* Initialize quotas before so that eventual writes go
	 * in separate transaction */
	if (new_dentry->d_inode)
		DQUOT_INIT(new_dentry->d_inode);
2230 2231 2232
	handle = ext4_journal_start(old_dir, 2 *
					EXT4_DATA_TRANS_BLOCKS(old_dir->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 2);
2233 2234 2235 2236 2237 2238
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	if (IS_DIRSYNC(old_dir) || IS_DIRSYNC(new_dir))
		handle->h_sync = 1;

2239
	old_bh = ext4_find_entry (old_dentry, &old_de);
2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
	/*
	 *  Check for inode number is _not_ due to possible IO errors.
	 *  We might rmdir the source, keep it as pwd of some process
	 *  and merrily kill the link to whatever was created under the
	 *  same name. Goodbye sticky bit ;-<
	 */
	old_inode = old_dentry->d_inode;
	retval = -ENOENT;
	if (!old_bh || le32_to_cpu(old_de->inode) != old_inode->i_ino)
		goto end_rename;

	new_inode = new_dentry->d_inode;
2252
	new_bh = ext4_find_entry (new_dentry, &new_de);
2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265
	if (new_bh) {
		if (!new_inode) {
			brelse (new_bh);
			new_bh = NULL;
		}
	}
	if (S_ISDIR(old_inode->i_mode)) {
		if (new_inode) {
			retval = -ENOTEMPTY;
			if (!empty_dir (new_inode))
				goto end_rename;
		}
		retval = -EIO;
2266
		dir_bh = ext4_bread (handle, old_inode, 0, 0, &retval);
2267 2268 2269 2270 2271 2272
		if (!dir_bh)
			goto end_rename;
		if (le32_to_cpu(PARENT_INO(dir_bh->b_data)) != old_dir->i_ino)
			goto end_rename;
		retval = -EMLINK;
		if (!new_inode && new_dir!=old_dir &&
2273
				new_dir->i_nlink >= EXT4_LINK_MAX)
2274 2275 2276
			goto end_rename;
	}
	if (!new_bh) {
2277
		retval = ext4_add_entry (handle, new_dentry, old_inode);
2278 2279 2280 2281
		if (retval)
			goto end_rename;
	} else {
		BUFFER_TRACE(new_bh, "get write access");
2282
		ext4_journal_get_write_access(handle, new_bh);
2283
		new_de->inode = cpu_to_le32(old_inode->i_ino);
2284 2285
		if (EXT4_HAS_INCOMPAT_FEATURE(new_dir->i_sb,
					      EXT4_FEATURE_INCOMPAT_FILETYPE))
2286 2287
			new_de->file_type = old_de->file_type;
		new_dir->i_version++;
2288 2289
		BUFFER_TRACE(new_bh, "call ext4_journal_dirty_metadata");
		ext4_journal_dirty_metadata(handle, new_bh);
2290 2291 2292 2293 2294 2295 2296 2297 2298
		brelse(new_bh);
		new_bh = NULL;
	}

	/*
	 * Like most other Unix systems, set the ctime for inodes on a
	 * rename.
	 */
	old_inode->i_ctime = CURRENT_TIME_SEC;
2299
	ext4_mark_inode_dirty(handle, old_inode);
2300 2301 2302 2303 2304 2305 2306

	/*
	 * ok, that's it
	 */
	if (le32_to_cpu(old_de->inode) != old_inode->i_ino ||
	    old_de->name_len != old_dentry->d_name.len ||
	    strncmp(old_de->name, old_dentry->d_name.name, old_de->name_len) ||
2307
	    (retval = ext4_delete_entry(handle, old_dir,
2308 2309 2310 2311 2312 2313
					old_de, old_bh)) == -ENOENT) {
		/* old_de could have moved from under us during htree split, so
		 * make sure that we are deleting the right entry.  We might
		 * also be pointing to a stale entry in the unused part of
		 * old_bh so just checking inum and the name isn't enough. */
		struct buffer_head *old_bh2;
2314
		struct ext4_dir_entry_2 *old_de2;
2315

2316
		old_bh2 = ext4_find_entry(old_dentry, &old_de2);
2317
		if (old_bh2) {
2318
			retval = ext4_delete_entry(handle, old_dir,
2319 2320 2321 2322 2323
						   old_de2, old_bh2);
			brelse(old_bh2);
		}
	}
	if (retval) {
2324
		ext4_warning(old_dir->i_sb, "ext4_rename",
2325 2326 2327 2328 2329 2330 2331 2332 2333
				"Deleting old file (%lu), %d, error=%d",
				old_dir->i_ino, old_dir->i_nlink, retval);
	}

	if (new_inode) {
		drop_nlink(new_inode);
		new_inode->i_ctime = CURRENT_TIME_SEC;
	}
	old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME_SEC;
2334
	ext4_update_dx_flag(old_dir);
2335 2336
	if (dir_bh) {
		BUFFER_TRACE(dir_bh, "get_write_access");
2337
		ext4_journal_get_write_access(handle, dir_bh);
2338
		PARENT_INO(dir_bh->b_data) = cpu_to_le32(new_dir->i_ino);
2339 2340
		BUFFER_TRACE(dir_bh, "call ext4_journal_dirty_metadata");
		ext4_journal_dirty_metadata(handle, dir_bh);
2341 2342 2343 2344 2345
		drop_nlink(old_dir);
		if (new_inode) {
			drop_nlink(new_inode);
		} else {
			inc_nlink(new_dir);
2346 2347
			ext4_update_dx_flag(new_dir);
			ext4_mark_inode_dirty(handle, new_dir);
2348 2349
		}
	}
2350
	ext4_mark_inode_dirty(handle, old_dir);
2351
	if (new_inode) {
2352
		ext4_mark_inode_dirty(handle, new_inode);
2353
		if (!new_inode->i_nlink)
2354
			ext4_orphan_add(handle, new_inode);
2355 2356 2357 2358 2359 2360 2361
	}
	retval = 0;

end_rename:
	brelse (dir_bh);
	brelse (old_bh);
	brelse (new_bh);
2362
	ext4_journal_stop(handle);
2363 2364 2365 2366 2367 2368
	return retval;
}

/*
 * directories can handle most operations...
 */
2369
const struct inode_operations ext4_dir_inode_operations = {
2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380
	.create		= ext4_create,
	.lookup		= ext4_lookup,
	.link		= ext4_link,
	.unlink		= ext4_unlink,
	.symlink	= ext4_symlink,
	.mkdir		= ext4_mkdir,
	.rmdir		= ext4_rmdir,
	.mknod		= ext4_mknod,
	.rename		= ext4_rename,
	.setattr	= ext4_setattr,
#ifdef CONFIG_EXT4DEV_FS_XATTR
2381 2382
	.setxattr	= generic_setxattr,
	.getxattr	= generic_getxattr,
2383
	.listxattr	= ext4_listxattr,
2384 2385
	.removexattr	= generic_removexattr,
#endif
2386
	.permission	= ext4_permission,
2387 2388
};

2389
const struct inode_operations ext4_special_inode_operations = {
2390 2391
	.setattr	= ext4_setattr,
#ifdef CONFIG_EXT4DEV_FS_XATTR
2392 2393
	.setxattr	= generic_setxattr,
	.getxattr	= generic_getxattr,
2394
	.listxattr	= ext4_listxattr,
2395 2396
	.removexattr	= generic_removexattr,
#endif
2397
	.permission	= ext4_permission,
2398
};