namei.c

/*
 *  linux/fs/ext4/namei.c
 *
 * 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>
#include <linux/jbd2.h>
#include <linux/time.h>
#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 "ext4.h"
#include "ext4_jbd2.h"

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

#include <trace/events/ext4.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)

static struct buffer_head *ext4_append(handle_t *handle,
					struct inode *inode,
					ext4_lblk_t *block)
{
	struct buffer_head *bh;
	int err = 0;

	if (unlikely(EXT4_SB(inode->i_sb)->s_max_dir_size_kb &&
		     ((inode->i_size >> 10) >=
		      EXT4_SB(inode->i_sb)->s_max_dir_size_kb)))
		return ERR_PTR(-ENOSPC);

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

	bh = ext4_bread(handle, inode, *block, 1, &err);
	if (!bh)
		return ERR_PTR(err);
	inode->i_size += inode->i_sb->s_blocksize;
	EXT4_I(inode)->i_disksize = inode->i_size;
	err = ext4_journal_get_write_access(handle, bh);
	if (err) {
		brelse(bh);
		ext4_std_error(inode->i_sb, err);
		return ERR_PTR(err);
	}
	return bh;
}

static int ext4_dx_csum_verify(struct inode *inode,
			       struct ext4_dir_entry *dirent);

typedef enum {
	EITHER, INDEX, DIRENT
} dirblock_type_t;

#define ext4_read_dirblock(inode, block, type) \
	__ext4_read_dirblock((inode), (block), (type), __LINE__)

static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
					      ext4_lblk_t block,
					      dirblock_type_t type,
					      unsigned int line)
{
	struct buffer_head *bh;
	struct ext4_dir_entry *dirent;
	int err = 0, is_dx_block = 0;

	bh = ext4_bread(NULL, inode, block, 0, &err);
	if (!bh) {
		if (err == 0) {
			ext4_error_inode(inode, __func__, line, block,
					       "Directory hole found");
			return ERR_PTR(-EIO);
		}
		__ext4_warning(inode->i_sb, __func__, line,
			       "error reading directory block "
			       "(ino %lu, block %lu)", inode->i_ino,
			       (unsigned long) block);
		return ERR_PTR(err);
	}
	dirent = (struct ext4_dir_entry *) bh->b_data;
	/* Determine whether or not we have an index block */
	if (is_dx(inode)) {
		if (block == 0)
			is_dx_block = 1;
		else if (ext4_rec_len_from_disk(dirent->rec_len,
						inode->i_sb->s_blocksize) ==
			 inode->i_sb->s_blocksize)
			is_dx_block = 1;
	}
	if (!is_dx_block && type == INDEX) {
		ext4_error_inode(inode, __func__, line, block,
		       "directory leaf block found instead of index block");
		return ERR_PTR(-EIO);
	}
	if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) ||
	    buffer_verified(bh))
		return bh;

	/*
	 * An empty leaf block can get mistaken for a index block; for
	 * this reason, we can only check the index checksum when the
	 * caller is sure it should be an index block.
	 */
	if (is_dx_block && type == INDEX) {
		if (ext4_dx_csum_verify(inode, dirent))
			set_buffer_verified(bh);
		else {
			ext4_error_inode(inode, __func__, line, block,
				"Directory index failed checksum");
			brelse(bh);
			return ERR_PTR(-EIO);
		}
	}
	if (!is_dx_block) {
		if (ext4_dirent_csum_verify(inode, dirent))
			set_buffer_verified(bh);
		else {
			ext4_error_inode(inode, __func__, line, block,
				"Directory block failed checksum");
			brelse(bh);
			return ERR_PTR(-EIO);
		}
	}
	return bh;
}

#ifndef assert
#define assert(test) J_ASSERT(test)
#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;
	u16 offs;
	u16 size;
};

/*
 * This goes at the end of each htree block.
 */
struct dx_tail {
	u32 dt_reserved;
	__le32 dt_checksum;	/* crc32c(uuid+inum+dirblock) */
};

static inline ext4_lblk_t dx_get_block(struct dx_entry *entry);
static void dx_set_block(struct dx_entry *entry, ext4_lblk_t 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(const struct qstr *d_name,
				 struct inode *dir,
				 struct dx_hash_info *hinfo,
				 struct dx_frame *frame,
				 int *err);
static void dx_release(struct dx_frame *frames);
static int dx_make_map(struct ext4_dir_entry_2 *de, unsigned blocksize,
		       struct dx_hash_info *hinfo, struct dx_map_entry map[]);
static void dx_sort_map(struct dx_map_entry *map, unsigned count);
static struct ext4_dir_entry_2 *dx_move_dirents(char *from, char *to,
		struct dx_map_entry *offsets, int count, unsigned blocksize);
static struct ext4_dir_entry_2* dx_pack_dirents(char *base, unsigned blocksize);
static void dx_insert_block(struct dx_frame *frame,
					u32 hash, ext4_lblk_t block);
static int ext4_htree_next_block(struct inode *dir, __u32 hash,
				 struct dx_frame *frame,
				 struct dx_frame *frames,
				 __u32 *start_hash);
static struct buffer_head * ext4_dx_find_entry(struct inode *dir,
		const struct qstr *d_name,
		struct ext4_dir_entry_2 **res_dir,
		int *err);
static int ext4_dx_add_entry(handle_t *handle, struct dentry *dentry,
			     struct inode *inode);

/* checksumming functions */
void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
			    unsigned int blocksize)
{
	memset(t, 0, sizeof(struct ext4_dir_entry_tail));
	t->det_rec_len = ext4_rec_len_to_disk(
			sizeof(struct ext4_dir_entry_tail), blocksize);
	t->det_reserved_ft = EXT4_FT_DIR_CSUM;
}

/* Walk through a dirent block to find a checksum "dirent" at the tail */
static struct ext4_dir_entry_tail *get_dirent_tail(struct inode *inode,
						   struct ext4_dir_entry *de)
{
	struct ext4_dir_entry_tail *t;

#ifdef PARANOID
	struct ext4_dir_entry *d, *top;

	d = de;
	top = (struct ext4_dir_entry *)(((void *)de) +
		(EXT4_BLOCK_SIZE(inode->i_sb) -
		sizeof(struct ext4_dir_entry_tail)));
	while (d < top && d->rec_len)
		d = (struct ext4_dir_entry *)(((void *)d) +
		    le16_to_cpu(d->rec_len));

	if (d != top)
		return NULL;

	t = (struct ext4_dir_entry_tail *)d;
#else
	t = EXT4_DIRENT_TAIL(de, EXT4_BLOCK_SIZE(inode->i_sb));
#endif

	if (t->det_reserved_zero1 ||
	    le16_to_cpu(t->det_rec_len) != sizeof(struct ext4_dir_entry_tail) ||
	    t->det_reserved_zero2 ||
	    t->det_reserved_ft != EXT4_FT_DIR_CSUM)
		return NULL;

	return t;
}

static __le32 ext4_dirent_csum(struct inode *inode,
			       struct ext4_dir_entry *dirent, int size)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	struct ext4_inode_info *ei = EXT4_I(inode);
	__u32 csum;

	csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)dirent, size);
	return cpu_to_le32(csum);
}

static void warn_no_space_for_csum(struct inode *inode)
{
	ext4_warning(inode->i_sb, "no space in directory inode %lu leaf for "
		     "checksum.  Please run e2fsck -D.", inode->i_ino);
}

int ext4_dirent_csum_verify(struct inode *inode, struct ext4_dir_entry *dirent)
{
	struct ext4_dir_entry_tail *t;

	if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		return 1;

	t = get_dirent_tail(inode, dirent);
	if (!t) {
		warn_no_space_for_csum(inode);
		return 0;
	}

	if (t->det_checksum != ext4_dirent_csum(inode, dirent,
						(void *)t - (void *)dirent))
		return 0;

	return 1;
}

static void ext4_dirent_csum_set(struct inode *inode,
				 struct ext4_dir_entry *dirent)
{
	struct ext4_dir_entry_tail *t;

	if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		return;

	t = get_dirent_tail(inode, dirent);
	if (!t) {
		warn_no_space_for_csum(inode);
		return;
	}

	t->det_checksum = ext4_dirent_csum(inode, dirent,
					   (void *)t - (void *)dirent);
}

int ext4_handle_dirty_dirent_node(handle_t *handle,
				  struct inode *inode,
				  struct buffer_head *bh)
{
	ext4_dirent_csum_set(inode, (struct ext4_dir_entry *)bh->b_data);
	return ext4_handle_dirty_metadata(handle, inode, bh);
}

static struct dx_countlimit *get_dx_countlimit(struct inode *inode,
					       struct ext4_dir_entry *dirent,
					       int *offset)
{
	struct ext4_dir_entry *dp;
	struct dx_root_info *root;
	int count_offset;

	if (le16_to_cpu(dirent->rec_len) == EXT4_BLOCK_SIZE(inode->i_sb))
		count_offset = 8;
	else if (le16_to_cpu(dirent->rec_len) == 12) {
		dp = (struct ext4_dir_entry *)(((void *)dirent) + 12);
		if (le16_to_cpu(dp->rec_len) !=
		    EXT4_BLOCK_SIZE(inode->i_sb) - 12)
			return NULL;
		root = (struct dx_root_info *)(((void *)dp + 12));
		if (root->reserved_zero ||
		    root->info_length != sizeof(struct dx_root_info))
			return NULL;
		count_offset = 32;
	} else
		return NULL;

	if (offset)
		*offset = count_offset;
	return (struct dx_countlimit *)(((void *)dirent) + count_offset);
}

static __le32 ext4_dx_csum(struct inode *inode, struct ext4_dir_entry *dirent,
			   int count_offset, int count, struct dx_tail *t)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	struct ext4_inode_info *ei = EXT4_I(inode);
	__u32 csum;
	__le32 save_csum;
	int size;

	size = count_offset + (count * sizeof(struct dx_entry));
	save_csum = t->dt_checksum;
	t->dt_checksum = 0;
	csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)dirent, size);
	csum = ext4_chksum(sbi, csum, (__u8 *)t, sizeof(struct dx_tail));
	t->dt_checksum = save_csum;

	return cpu_to_le32(csum);
}

static int ext4_dx_csum_verify(struct inode *inode,
			       struct ext4_dir_entry *dirent)
{
	struct dx_countlimit *c;
	struct dx_tail *t;
	int count_offset, limit, count;

	if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		return 1;

	c = get_dx_countlimit(inode, dirent, &count_offset);
	if (!c) {
		EXT4_ERROR_INODE(inode, "dir seems corrupt?  Run e2fsck -D.");
		return 1;
	}
	limit = le16_to_cpu(c->limit);
	count = le16_to_cpu(c->count);
	if (count_offset + (limit * sizeof(struct dx_entry)) >
	    EXT4_BLOCK_SIZE(inode->i_sb) - sizeof(struct dx_tail)) {
		warn_no_space_for_csum(inode);
		return 1;
	}
	t = (struct dx_tail *)(((struct dx_entry *)c) + limit);

	if (t->dt_checksum != ext4_dx_csum(inode, dirent, count_offset,
					    count, t))
		return 0;
	return 1;
}

static void ext4_dx_csum_set(struct inode *inode, struct ext4_dir_entry *dirent)
{
	struct dx_countlimit *c;
	struct dx_tail *t;
	int count_offset, limit, count;

	if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		return;

	c = get_dx_countlimit(inode, dirent, &count_offset);
	if (!c) {
		EXT4_ERROR_INODE(inode, "dir seems corrupt?  Run e2fsck -D.");
		return;
	}
	limit = le16_to_cpu(c->limit);
	count = le16_to_cpu(c->count);
	if (count_offset + (limit * sizeof(struct dx_entry)) >
	    EXT4_BLOCK_SIZE(inode->i_sb) - sizeof(struct dx_tail)) {
		warn_no_space_for_csum(inode);
		return;
	}
	t = (struct dx_tail *)(((struct dx_entry *)c) + limit);

	t->dt_checksum = ext4_dx_csum(inode, dirent, count_offset, count, t);
}

static inline int ext4_handle_dirty_dx_node(handle_t *handle,
					    struct inode *inode,
					    struct buffer_head *bh)
{
	ext4_dx_csum_set(inode, (struct ext4_dir_entry *)bh->b_data);
	return ext4_handle_dirty_metadata(handle, inode, bh);
}

/*
 * p is at least 6 bytes before the end of page
 */
static inline struct ext4_dir_entry_2 *
ext4_next_entry(struct ext4_dir_entry_2 *p, unsigned long blocksize)
{
	return (struct ext4_dir_entry_2 *)((char *)p +
		ext4_rec_len_from_disk(p->rec_len, blocksize));
}

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

static inline ext4_lblk_t dx_get_block(struct dx_entry *entry)
{
	return le32_to_cpu(entry->block) & 0x00ffffff;
}

static inline void dx_set_block(struct dx_entry *entry, ext4_lblk_t 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)
{
	unsigned entry_space = dir->i_sb->s_blocksize - EXT4_DIR_REC_LEN(1) -
		EXT4_DIR_REC_LEN(2) - infosize;

	if (EXT4_HAS_RO_COMPAT_FEATURE(dir->i_sb,
				       EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		entry_space -= sizeof(struct dx_tail);
	return entry_space / sizeof(struct dx_entry);
}

static inline unsigned dx_node_limit(struct inode *dir)
{
	unsigned entry_space = dir->i_sb->s_blocksize - EXT4_DIR_REC_LEN(0);

	if (EXT4_HAS_RO_COMPAT_FEATURE(dir->i_sb,
				       EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		entry_space -= sizeof(struct dx_tail);
	return entry_space / sizeof(struct dx_entry);
}

/*
 * Debug
 */
#ifdef DX_DEBUG
static void dx_show_index(char * label, struct dx_entry *entries)
{
	int i, n = dx_get_count (entries);
	printk(KERN_DEBUG "%s index ", label);
	for (i = 0; i < n; i++) {
		printk("%x->%lu ", i ? dx_get_hash(entries + i) :
				0, (unsigned long)dx_get_block(entries + i));
	}
	printk("\n");
}

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

static struct stats dx_show_leaf(struct dx_hash_info *hinfo, struct ext4_dir_entry_2 *de,
				 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++);
				ext4fs_dirhash(de->name, de->name_len, &h);
				printk(":%x.%u ", h.hash,
				       (unsigned) ((char *) de - base));
			}
			space += EXT4_DIR_REC_LEN(de->name_len);
			names++;
		}
		de = ext4_next_entry(de, size);
	}
	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++)
	{
		ext4_lblk_t block = dx_get_block(entries);
		ext4_lblk_t 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);
		if (!(bh = ext4_bread (NULL,dir, block, 0,&err))) continue;
		stats = levels?
		   dx_show_entries(hinfo, dir, ((struct dx_node *) bh->b_data)->entries, levels - 1):
		   dx_show_leaf(hinfo, (struct ext4_dir_entry_2 *) bh->b_data, blocksize, 0);
		names += stats.names;
		space += stats.space;
		bcount += stats.bcount;
		brelse(bh);
	}
	if (bcount)
		printk(KERN_DEBUG "%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(const struct qstr *d_name, 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;
	bh = ext4_read_dirblock(dir, 0, INDEX);
	if (IS_ERR(bh)) {
		*err = PTR_ERR(bh);
		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) {
		ext4_warning(dir->i_sb, "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;
	if (hinfo->hash_version <= DX_HASH_TEA)
		hinfo->hash_version += EXT4_SB(dir->i_sb)->s_hash_unsigned;
	hinfo->seed = EXT4_SB(dir->i_sb)->s_hash_seed;
	if (d_name)
		ext4fs_dirhash(d_name->name, d_name->len, hinfo);
	hash = hinfo->hash;

	if (root->info.unused_flags & 1) {
		ext4_warning(dir->i_sb, "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) {
		ext4_warning(dir->i_sb, "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);

	if (dx_get_limit(entries) != dx_root_limit(dir,
						   root->info.info_length)) {
		ext4_warning(dir->i_sb, "dx entry: limit != root limit");
		brelse(bh);
		*err = ERR_BAD_DX_DIR;
		goto fail;
	}

	dxtrace(printk("Look up %x", hash));
	while (1)
	{
		count = dx_get_count(entries);
		if (!count || count > dx_get_limit(entries)) {
			ext4_warning(dir->i_sb,
				     "dx entry: no count or count > limit");
			brelse(bh);
			*err = ERR_BAD_DX_DIR;
			goto fail2;
		}

		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;
		bh = ext4_read_dirblock(dir, dx_get_block(at), INDEX);
		if (IS_ERR(bh)) {
			*err = PTR_ERR(bh);
			goto fail2;
		}
		entries = ((struct dx_node *) bh->b_data)->entries;

		if (dx_get_limit(entries) != dx_node_limit (dir)) {
			ext4_warning(dir->i_sb,
				     "dx entry: limit != node limit");
			brelse(bh);
			*err = ERR_BAD_DX_DIR;
			goto fail2;
		}
		frame++;
		frame->bh = NULL;
	}
fail2:
	while (frame >= frame_in) {
		brelse(frame->bh);
		frame--;
	}
fail:
	if (*err == ERR_BAD_DX_DIR)
		ext4_warning(dir->i_sb,
			     "Corrupt dir inode %lu, running e2fsck is "
			     "recommended.", dir->i_ino);
	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.
 */
static int ext4_htree_next_block(struct inode *dir, __u32 hash,
				 struct dx_frame *frame,
				 struct dx_frame *frames,
				 __u32 *start_hash)
{
	struct dx_frame *p;
	struct buffer_head *bh;
	int 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--) {
		bh = ext4_read_dirblock(dir, dx_get_block(p->at), INDEX);
		if (IS_ERR(bh))
			return PTR_ERR(bh);
		p++;
		brelse(p->bh);
		p->bh = bh;
		p->at = p->entries = ((struct dx_node *) bh->b_data)->entries;
	}
	return 1;
}


/*
 * 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, ext4_lblk_t block,
				  struct dx_hash_info *hinfo,
				  __u32 start_hash, __u32 start_minor_hash)
{
	struct buffer_head *bh;
	struct ext4_dir_entry_2 *de, *top;
	int err = 0, count = 0;

	dxtrace(printk(KERN_INFO "In htree dirblock_to_tree: block %lu\n",
							(unsigned long)block));
	bh = ext4_read_dirblock(dir, block, DIRENT);
	if (IS_ERR(bh))
		return PTR_ERR(bh);

	de = (struct ext4_dir_entry_2 *) bh->b_data;
	top = (struct ext4_dir_entry_2 *) ((char *) de +
					   dir->i_sb->s_blocksize -
					   EXT4_DIR_REC_LEN(0));
	for (; de < top; de = ext4_next_entry(de, dir->i_sb->s_blocksize)) {
		if (ext4_check_dir_entry(dir, NULL, de, bh,
				bh->b_data, bh->b_size,
				(block<<EXT4_BLOCK_SIZE_BITS(dir->i_sb))
					 + ((char *)de - bh->b_data))) {
			/* silently ignore the rest of the block */
			break;
		}
		ext4fs_dirhash(de->name, de->name_len, hinfo);
		if ((hinfo->hash < start_hash) ||
		    ((hinfo->hash == start_hash) &&
		     (hinfo->minor_hash < start_minor_hash)))
			continue;
		if (de->inode == 0)
			continue;
		if ((err = ext4_htree_store_dirent(dir_file,
				   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.
 */
int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
			 __u32 start_minor_hash, __u32 *next_hash)
{
	struct dx_hash_info hinfo;
	struct ext4_dir_entry_2 *de;
	struct dx_frame frames[2], *frame;
	struct inode *dir;
	ext4_lblk_t block;
	int count = 0;
	int ret, err;
	__u32 hashval;

	dxtrace(printk(KERN_DEBUG "In htree_fill_tree, start hash: %x:%x\n",
		       start_hash, start_minor_hash));
	dir = file_inode(dir_file);
	if (!(ext4_test_inode_flag(dir, EXT4_INODE_INDEX))) {
		hinfo.hash_version = EXT4_SB(dir->i_sb)->s_def_hash_version;
		if (hinfo.hash_version <= DX_HASH_TEA)
			hinfo.hash_version +=
				EXT4_SB(dir->i_sb)->s_hash_unsigned;
		hinfo.seed = EXT4_SB(dir->i_sb)->s_hash_seed;
		if (ext4_has_inline_data(dir)) {
			int has_inline_data = 1;
			count = htree_inlinedir_to_tree(dir_file, dir, 0,
							&hinfo, start_hash,
							start_minor_hash,
							&has_inline_data);
			if (has_inline_data) {
				*next_hash = ~0;
				return count;
			}
		}
		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;
	frame = dx_probe(NULL, dir, &hinfo, frames, &err);
	if (!frame)
		return err;

	/* Add '.' and '..' from the htree header */
	if (!start_hash && !start_minor_hash) {
		de = (struct ext4_dir_entry_2 *) frames[0].bh->b_data;
		if ((err = ext4_htree_store_dirent(dir_file, 0, 0, de)) != 0)
			goto errout;
		count++;
	}
	if (start_hash < 2 || (start_hash ==2 && start_minor_hash==0)) {
		de = (struct ext4_dir_entry_2 *) frames[0].bh->b_data;
		de = ext4_next_entry(de, dir->i_sb->s_blocksize);
		if ((err = ext4_htree_store_dirent(dir_file, 2, 0, de)) != 0)
			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;
		ret = ext4_htree_next_block(dir, HASH_NB_ALWAYS,
					    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(KERN_DEBUG "Fill tree: returned %d entries, "
		       "next hash: %x\n", count, *next_hash));
	return count;
errout:
	dx_release(frames);
	return (err);
}

static inline int search_dirblock(struct buffer_head *bh,
				  struct inode *dir,
				  const struct qstr *d_name,
				  unsigned int offset,
				  struct ext4_dir_entry_2 **res_dir)
{
	return search_dir(bh, bh->b_data, dir->i_sb->s_blocksize, dir,
			  d_name, offset, res_dir);
}

/*
 * Directory block splitting, compacting
 */

/*
 * Create map of hash values, offsets, and sizes, stored at end of block.
 * Returns number of entries mapped.
 */
static int dx_make_map(struct ext4_dir_entry_2 *de, unsigned blocksize,
		       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 + blocksize) {
		if (de->name_len && de->inode) {
			ext4fs_dirhash(de->name, de->name_len, &h);
			map_tail--;
			map_tail->hash = h.hash;
			map_tail->offs = ((char *) de - base)>>2;
			map_tail->size = le16_to_cpu(de->rec_len);
			count++;
			cond_resched();
		}
		/* XXX: do we need to check rec_len == 0 case? -Chris */
		de = ext4_next_entry(de, blocksize);
	}
	return count;
}

/* Sort map by hash value */
static void dx_sort_map (struct dx_map_entry *map, unsigned count)
{
	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)
				continue;
			swap(*(q+1), *q);
			more = 1;
		}
	} while(more);
}

static void dx_insert_block(struct dx_frame *frame, u32 hash, ext4_lblk_t 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);
}

/*
 * NOTE! unlike strncmp, ext4_match returns 1 for success, 0 for failure.
 *
 * `len <= EXT4_NAME_LEN' is guaranteed by caller.
 * `de != NULL' is guaranteed by caller.
 */
static inline int ext4_match (int len, const char * const name,
			      struct ext4_dir_entry_2 * de)
{
	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
 */
int search_dir(struct buffer_head *bh,
	       char *search_buf,
	       int buf_size,
	       struct inode *dir,
	       const struct qstr *d_name,
	       unsigned int offset,
	       struct ext4_dir_entry_2 **res_dir)
{
	struct ext4_dir_entry_2 * de;
	char * dlimit;
	int de_len;
	const char *name = d_name->name;
	int namelen = d_name->len;

	de = (struct ext4_dir_entry_2 *)search_buf;
	dlimit = search_buf + buf_size;
	while ((char *) de < dlimit) {
		/* this code is executed quadratically often */
		/* do minimal checking `by hand' */

		if ((char *) de + namelen <= dlimit &&
		    ext4_match (namelen, name, de)) {
			/* found a match - just to be sure, do a full check */
			if (ext4_check_dir_entry(dir, NULL, de, bh, bh->b_data,
						 bh->b_size, offset))
				return -1;
			*res_dir = de;
			return 1;
		}
		/* prevent looping on a bad block */
		de_len = ext4_rec_len_from_disk(de->rec_len,
						dir->i_sb->s_blocksize);
		if (de_len <= 0)
			return -1;
		offset += de_len;
		de = (struct ext4_dir_entry_2 *) ((char *) de + de_len);
	}
	return 0;
}

static int is_dx_internal_node(struct inode *dir, ext4_lblk_t block,
			       struct ext4_dir_entry *de)
{
	struct super_block *sb = dir->i_sb;

	if (!is_dx(dir))
		return 0;
	if (block == 0)
		return 1;
	if (de->inode == 0 &&
	    ext4_rec_len_from_disk(de->rec_len, sb->s_blocksize) ==
			sb->s_blocksize)
		return 1;
	return 0;
}

/*
 *	ext4_find_entry()
 *
 * 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.
 */
static struct buffer_head * ext4_find_entry (struct inode *dir,
					const struct qstr *d_name,
					struct ext4_dir_entry_2 **res_dir,
					int *inlined)
{
	struct super_block *sb;
	struct buffer_head *bh_use[NAMEI_RA_SIZE];
	struct buffer_head *bh, *ret = NULL;
	ext4_lblk_t start, block, b;
	const u8 *name = d_name->name;
	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;
	ext4_lblk_t  nblocks;
	int i, err;
	int namelen;

	*res_dir = NULL;
	sb = dir->i_sb;
	namelen = d_name->len;
	if (namelen > EXT4_NAME_LEN)
		return NULL;

	if (ext4_has_inline_data(dir)) {
		int has_inline_data = 1;
		ret = ext4_find_inline_entry(dir, d_name, res_dir,
					     &has_inline_data);
		if (has_inline_data) {
			if (inlined)
				*inlined = 1;
			return ret;
		}
	}

	if ((namelen <= 2) && (name[0] == '.') &&
	    (name[1] == '.' || name[1] == '\0')) {
		/*
		 * "." or ".." will only be in the first block
		 * NFS may look up ".."; "." should be handled by the VFS
		 */
		block = start = 0;
		nblocks = 1;
		goto restart;
	}
	if (is_dx(dir)) {
		bh = ext4_dx_find_entry(dir, d_name, res_dir, &err);
		/*
		 * 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;
		dxtrace(printk(KERN_DEBUG "ext4_find_entry: dx failed, "
			       "falling back\n"));
	}
	nblocks = dir->i_size >> EXT4_BLOCK_SIZE_BITS(sb);
	start = EXT4_I(dir)->i_dir_start_lookup;
	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++;
				bh = ext4_getblk(NULL, dir, b++, 0, &err);
				bh_use[ra_max] = bh;
				if (bh)
					ll_rw_block(READ | REQ_META | REQ_PRIO,
						    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 */
			EXT4_ERROR_INODE(dir, "reading directory lblock %lu",
					 (unsigned long) block);
			brelse(bh);
			goto next;
		}
		if (!buffer_verified(bh) &&
		    !is_dx_internal_node(dir, block,
					 (struct ext4_dir_entry *)bh->b_data) &&
		    !ext4_dirent_csum_verify(dir,
				(struct ext4_dir_entry *)bh->b_data)) {
			EXT4_ERROR_INODE(dir, "checksumming directory "
					 "block %lu", (unsigned long)block);
			brelse(bh);
			goto next;
		}
		set_buffer_verified(bh);
		i = search_dirblock(bh, dir, d_name,
			    block << EXT4_BLOCK_SIZE_BITS(sb), res_dir);
		if (i == 1) {
			EXT4_I(dir)->i_dir_start_lookup = block;
			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;
	nblocks = dir->i_size >> EXT4_BLOCK_SIZE_BITS(sb);
	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;
}

static struct buffer_head * ext4_dx_find_entry(struct inode *dir, const struct qstr *d_name,
		       struct ext4_dir_entry_2 **res_dir, int *err)
{
	struct super_block * sb = dir->i_sb;
	struct dx_hash_info	hinfo;
	struct dx_frame frames[2], *frame;
	struct buffer_head *bh;
	ext4_lblk_t block;
	int retval;

	if (!(frame = dx_probe(d_name, dir, &hinfo, frames, err)))
		return NULL;
	do {
		block = dx_get_block(frame->at);
		bh = ext4_read_dirblock(dir, block, DIRENT);
		if (IS_ERR(bh)) {
			*err = PTR_ERR(bh);
			goto errout;
		}
		retval = search_dirblock(bh, dir, d_name,
					 block << EXT4_BLOCK_SIZE_BITS(sb),
					 res_dir);
		if (retval == 1) { 	/* Success! */
			dx_release(frames);
			return bh;
		}
		brelse(bh);
		if (retval == -1) {
			*err = ERR_BAD_DX_DIR;
			goto errout;
		}

		/* Check to see if we should continue to search */
		retval = ext4_htree_next_block(dir, hinfo.hash, frame,
					       frames, NULL);
		if (retval < 0) {
			ext4_warning(sb,
			     "error reading index page in directory #%lu",
			     dir->i_ino);
			*err = retval;
			goto errout;
		}
	} while (retval == 1);

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

static struct dentry *ext4_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
{
	struct inode *inode;
	struct ext4_dir_entry_2 *de;
	struct buffer_head *bh;