ext4_journal: forcibly flush data to disk when stop journalling.
[lwext4.git] / lwext4 / ext4_dir.c
index acbd0816555aca1c64ee5a9b6229477397e6197d..b22e88bcb93f1ee4aab2a0aad675c8a19cef2d65 100644 (file)
-/*\r
- * Copyright (c) 2013 Grzegorz Kostka (kostka.grzegorz@gmail.com)\r
- *\r
- *\r
- * HelenOS:\r
- * Copyright (c) 2012 Martin Sucha\r
- * Copyright (c) 2012 Frantisek Princ\r
- * All rights reserved.\r
- *\r
- * Redistribution and use in source and binary forms, with or without\r
- * modification, are permitted provided that the following conditions\r
- * are met:\r
- *\r
- * - Redistributions of source code must retain the above copyright\r
- *   notice, this list of conditions and the following disclaimer.\r
- * - Redistributions in binary form must reproduce the above copyright\r
- *   notice, this list of conditions and the following disclaimer in the\r
- *   documentation and/or other materials provided with the distribution.\r
- * - The name of the author may not be used to endorse or promote products\r
- *   derived from this software without specific prior written permission.\r
- *\r
- * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\r
- * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\r
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\r
- * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\r
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT\r
- * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,\r
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY\r
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT\r
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF\r
- * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\r
- */\r
-\r
-/** @addtogroup lwext4\r
- * @{\r
- */\r
-/**\r
- * @file  ext4_dir.h\r
- * @brief Directory handle procedures.\r
- */\r
-\r
-#include <ext4_config.h>\r
-#include <ext4_dir.h>\r
-#include <ext4_dir_idx.h>\r
-#include <ext4_inode.h>\r
-#include <ext4_fs.h>\r
-\r
-#include <string.h>\r
-\r
-uint32_t ext4_dir_entry_ll_get_inode(struct ext4_directory_entry_ll *de)\r
-{\r
-    return to_le32(de->inode);\r
-}\r
-\r
-void ext4_dir_entry_ll_set_inode(struct ext4_directory_entry_ll *de,\r
-    uint32_t inode)\r
-{\r
-    de->inode = to_le32(inode);\r
-}\r
-\r
-\r
-uint16_t ext4_dir_entry_ll_get_entry_length(struct ext4_directory_entry_ll *de)\r
-{\r
-    return to_le16(de->entry_length);\r
-}\r
-\r
-void ext4_dir_entry_ll_set_entry_length(struct ext4_directory_entry_ll *de,\r
-    uint16_t len)\r
-{\r
-    de->entry_length = to_le16(len);\r
-}\r
-\r
-\r
-uint16_t ext4_dir_entry_ll_get_name_length(struct ext4_sblock *sb,\r
-    struct ext4_directory_entry_ll *de)\r
-{\r
-    uint16_t v = de->name_length;\r
-\r
-    if ((ext4_get32(sb, rev_level) == 0) &&\r
-            (ext4_get32(sb, minor_rev_level) < 5))\r
-        v |= ((uint16_t)de->in.name_length_high) << 8;\r
-\r
-    return v;\r
-}\r
-void ext4_dir_entry_ll_set_name_length(struct ext4_sblock *sb,\r
-    struct ext4_directory_entry_ll *de, uint16_t len)\r
-{\r
-    de->name_length = (len << 8) >> 8;\r
-\r
-    if ((ext4_get32(sb, rev_level) == 0) &&\r
-            (ext4_get32(sb, minor_rev_level) < 5))\r
-        de->in.name_length_high = len >> 8;\r
-}\r
-\r
-\r
-\r
-uint8_t ext4_dir_entry_ll_get_inode_type(struct ext4_sblock *sb,\r
-    struct ext4_directory_entry_ll *de)\r
-{\r
-    if ((ext4_get32(sb, rev_level) > 0) ||\r
-            (ext4_get32(sb, minor_rev_level) >= 5))\r
-        return de->in.inode_type;\r
-\r
-    return EXT4_DIRECTORY_FILETYPE_UNKNOWN;\r
-}\r
-\r
-\r
-void ext4_dir_entry_ll_set_inode_type(struct ext4_sblock *sb,\r
-    struct ext4_directory_entry_ll *de, uint8_t type)\r
-{\r
-    if ((ext4_get32(sb, rev_level) > 0) ||\r
-            (ext4_get32(sb, minor_rev_level) >= 5))\r
-        de->in.inode_type = type;\r
-}\r
-\r
-/****************************************************************************/\r
-\r
-/**@brief Do some checks before returning iterator.\r
- * @param it Iterator to be checked\r
- * @param block_size Size of data block\r
- * @return Error code\r
- */\r
-static int ext4_dir_iterator_set(struct ext4_directory_iterator *it,\r
-    uint32_t block_size)\r
-{\r
-    it->current = NULL;\r
-\r
-    uint32_t offset_in_block = it->current_offset % block_size;\r
-\r
-    /* Ensure proper alignment */\r
-    if ((offset_in_block % 4) != 0)\r
-        return EIO;\r
-\r
-    /* Ensure that the core of the entry does not overflow the block */\r
-    if (offset_in_block > block_size - 8)\r
-        return EIO;\r
-\r
-    struct ext4_directory_entry_ll *entry =\r
-            (void *)(it->current_block.data + offset_in_block);\r
-\r
-    /* Ensure that the whole entry does not overflow the block */\r
-    uint16_t length = ext4_dir_entry_ll_get_entry_length(entry);\r
-    if (offset_in_block + length > block_size)\r
-        return EIO;\r
-\r
-    /* Ensure the name length is not too large */\r
-    if (ext4_dir_entry_ll_get_name_length(\r
-            &it->inode_ref->fs->sb, entry) > length-8)\r
-        return EIO;\r
-\r
-    /* Everything OK - "publish" the entry */\r
-    it->current = entry;\r
-    return EOK;\r
-}\r
-\r
-/**@brief Seek to next valid directory entry.\r
- *        Here can be jumped to the next data block.\r
- * @param it  Initialized iterator\r
- * @param pos Position of the next entry\r
- * @return Error code\r
- */\r
-static int ext4_dir_iterator_seek(struct ext4_directory_iterator *it,\r
-    uint64_t pos)\r
-{\r
-    uint64_t size = ext4_inode_get_size(&it->inode_ref->fs->sb,\r
-            it->inode_ref->inode);\r
-\r
-    /* The iterator is not valid until we seek to the desired position */\r
-    it->current = NULL;\r
-\r
-    /* Are we at the end? */\r
-    if (pos >= size) {\r
-        if (it->current_block.lb_id) {\r
-\r
-            int rc = ext4_block_set(it->inode_ref->fs->bdev,\r
-                    &it->current_block);\r
-            it->current_block.lb_id = 0;\r
-\r
-            if (rc != EOK)\r
-                return rc;\r
-        }\r
-\r
-        it->current_offset = pos;\r
-        return EOK;\r
-    }\r
-\r
-    /* Compute next block address */\r
-    uint32_t block_size =\r
-            ext4_sb_get_block_size(&it->inode_ref->fs->sb);\r
-    uint64_t current_block_idx = it->current_offset / block_size;\r
-    uint64_t next_block_idx = pos / block_size;\r
-\r
-    /*\r
-     * If we don't have a block or are moving accross block boundary,\r
-     * we need to get another block\r
-     */\r
-    if ((it->current_block.lb_id == 0) ||\r
-            (current_block_idx != next_block_idx)) {\r
-        if (it->current_block.lb_id) {\r
-            int rc = ext4_block_set(it->inode_ref->fs->bdev, &it->current_block);\r
-            it->current_block.lb_id = 0;\r
-\r
-            if (rc != EOK)\r
-                return rc;\r
-        }\r
-\r
-        uint32_t next_block_phys_idx;\r
-        int rc = ext4_fs_get_inode_data_block_index(it->inode_ref,\r
-                next_block_idx, &next_block_phys_idx);\r
-        if (rc != EOK)\r
-            return rc;\r
-\r
-\r
-        rc = ext4_block_get(it->inode_ref->fs->bdev, &it->current_block,\r
-                next_block_phys_idx);\r
-        if (rc != EOK) {\r
-            it->current_block.lb_id = 0;\r
-            return rc;\r
-        }\r
-    }\r
-\r
-    it->current_offset = pos;\r
-\r
-    return ext4_dir_iterator_set(it, block_size);\r
-}\r
-\r
-\r
-int ext4_dir_iterator_init(struct ext4_directory_iterator *it,\r
-    struct ext4_inode_ref *inode_ref, uint64_t pos)\r
-{\r
-    it->inode_ref = inode_ref;\r
-    it->current = 0;\r
-    it->current_offset = 0;\r
-    it->current_block.lb_id = 0;\r
-\r
-    return ext4_dir_iterator_seek(it, pos);\r
-}\r
-\r
-int ext4_dir_iterator_next(struct ext4_directory_iterator *it)\r
-{\r
-    int r = EOK;\r
-    uint16_t skip;\r
-\r
-    while(r == EOK){\r
-        skip = ext4_dir_entry_ll_get_entry_length(it->current);\r
-        r = ext4_dir_iterator_seek(it, it->current_offset + skip);\r
-\r
-        if(!it->current)\r
-            break;\r
-        /*Skip NULL referenced entry*/\r
-        if(it->current->inode != 0)\r
-            break;\r
-    }\r
-\r
-    return r;\r
-}\r
-\r
-int ext4_dir_iterator_fini(struct ext4_directory_iterator *it)\r
-{\r
-    it->current = 0;\r
-\r
-    if (it->current_block.lb_id)\r
-        return ext4_block_set(it->inode_ref->fs->bdev, &it->current_block);\r
-\r
-    return EOK;\r
-}\r
-\r
-void ext4_dir_write_entry(struct ext4_sblock *sb,\r
-    struct ext4_directory_entry_ll *entry, uint16_t entry_len,\r
-    struct ext4_inode_ref *child,  const char *name, size_t name_len)\r
-{\r
-    /* Check maximum entry length */\r
-    uint32_t block_size = ext4_sb_get_block_size(sb);\r
-    ext4_assert(entry_len <= block_size);\r
-\r
-    /* Set basic attributes */\r
-    ext4_dir_entry_ll_set_inode(entry, child->index);\r
-    ext4_dir_entry_ll_set_entry_length(entry, entry_len);\r
-    ext4_dir_entry_ll_set_name_length(sb, entry, name_len);\r
-\r
-    /* Write name */\r
-    memcpy(entry->name, name, name_len);\r
-\r
-    /* Set type of entry */\r
-    if (ext4_inode_is_type(sb, child->inode, EXT4_INODE_MODE_DIRECTORY))\r
-        ext4_dir_entry_ll_set_inode_type(sb, entry,\r
-                EXT4_DIRECTORY_FILETYPE_DIR);\r
-    else\r
-        ext4_dir_entry_ll_set_inode_type(sb, entry,\r
-                EXT4_DIRECTORY_FILETYPE_REG_FILE);\r
-}\r
-\r
-int ext4_dir_add_entry(struct ext4_inode_ref *parent, const char *name,\r
-    uint32_t name_len, struct ext4_inode_ref *child)\r
-{\r
-    struct ext4_fs *fs = parent->fs;\r
-\r
-#if CONFIG_DIR_INDEX_ENABLE\r
-    /* Index adding (if allowed) */\r
-    if ((ext4_sb_check_feature_compatible(&fs->sb,\r
-            EXT4_FEATURE_COMPAT_DIR_INDEX)) &&\r
-            (ext4_inode_has_flag(parent->inode, EXT4_INODE_FLAG_INDEX))) {\r
-        int rc = ext4_dir_dx_add_entry(parent, child, name);\r
-\r
-        /* Check if index is not corrupted */\r
-        if (rc != EXT4_ERR_BAD_DX_DIR) {\r
-            if (rc != EOK)\r
-                return rc;\r
-\r
-            return EOK;\r
-        }\r
-\r
-        /* Needed to clear dir index flag if corrupted */\r
-        ext4_inode_clear_flag(parent->inode, EXT4_INODE_FLAG_INDEX);\r
-        parent->dirty = true;\r
-    }\r
-#endif\r
-\r
-    /* Linear algorithm */\r
-    uint32_t iblock = 0;\r
-    uint32_t fblock = 0;\r
-    uint32_t block_size = ext4_sb_get_block_size(&fs->sb);\r
-    uint32_t inode_size = ext4_inode_get_size(&fs->sb, parent->inode);\r
-    uint32_t total_blocks = inode_size / block_size;\r
-\r
-\r
-    /* Find block, where is space for new entry and try to add */\r
-    bool success = false;\r
-    for (iblock = 0; iblock < total_blocks; ++iblock) {\r
-        int rc = ext4_fs_get_inode_data_block_index(parent,\r
-                iblock, &fblock);\r
-        if (rc != EOK)\r
-            return rc;\r
-\r
-\r
-        struct ext4_block block;\r
-        rc = ext4_block_get(fs->bdev, &block, fblock);\r
-        if (rc != EOK)\r
-            return rc;\r
-\r
-        /* If adding is successful, function can finish */\r
-        rc = ext4_dir_try_insert_entry(&fs->sb, &block,\r
-                child, name, name_len);\r
-        if (rc == EOK)\r
-            success = true;\r
-\r
-        rc = ext4_block_set(fs->bdev, &block);\r
-        if (rc != EOK)\r
-            return rc;\r
-\r
-        if (success)\r
-            return EOK;\r
-    }\r
-\r
-    /* No free block found - needed to allocate next data block */\r
-\r
-    iblock = 0;\r
-    fblock = 0;\r
-    int rc = ext4_fs_append_inode_block(parent, &fblock, &iblock);\r
-    if (rc != EOK)\r
-        return rc;\r
-\r
-    /* Load new block */\r
-    struct ext4_block new_block;\r
-\r
-    rc = ext4_block_get(fs->bdev, &new_block, fblock);\r
-    if (rc != EOK)\r
-        return rc;\r
-\r
-    /* Fill block with zeroes */\r
-    memset(new_block.data, 0, block_size);\r
-    struct ext4_directory_entry_ll *block_entry = (void *)new_block.data;\r
-    ext4_dir_write_entry(&fs->sb, block_entry, block_size,\r
-        child, name, name_len);\r
-\r
-    /* Save new block */\r
-    new_block.dirty = true;\r
-    rc = ext4_block_set(fs->bdev, &new_block);\r
-\r
-    return rc;\r
-}\r
-\r
-int ext4_dir_find_entry(struct ext4_directory_search_result *result,\r
-    struct ext4_inode_ref *parent, const char *name, uint32_t name_len)\r
-{\r
-    struct ext4_sblock *sb = &parent->fs->sb;\r
-\r
-\r
-#if CONFIG_DIR_INDEX_ENABLE\r
-    /* Index search */\r
-    if ((ext4_sb_check_feature_compatible(sb,\r
-            EXT4_FEATURE_COMPAT_DIR_INDEX)) &&\r
-            (ext4_inode_has_flag(parent->inode, EXT4_INODE_FLAG_INDEX))) {\r
-        int rc = ext4_dir_dx_find_entry(result, parent, name_len,\r
-                name);\r
-\r
-        /* Check if index is not corrupted */\r
-        if (rc != EXT4_ERR_BAD_DX_DIR) {\r
-            if (rc != EOK)\r
-                return rc;\r
-\r
-            return EOK;\r
-        }\r
-\r
-        /* Needed to clear dir index flag if corrupted */\r
-        ext4_inode_clear_flag(parent->inode, EXT4_INODE_FLAG_INDEX);\r
-        parent->dirty = true;\r
-    }\r
-#endif\r
-\r
-    /* Linear algorithm */\r
-\r
-    uint32_t iblock;\r
-    uint32_t fblock;\r
-    uint32_t block_size = ext4_sb_get_block_size(sb);\r
-    uint32_t inode_size = ext4_inode_get_size(sb, parent->inode);\r
-    uint32_t total_blocks = inode_size / block_size;\r
-\r
-    /* Walk through all data blocks */\r
-    for (iblock = 0; iblock < total_blocks; ++iblock) {\r
-        /* Load block address */\r
-        int rc = ext4_fs_get_inode_data_block_index(parent, iblock,\r
-                &fblock);\r
-        if (rc != EOK)\r
-            return rc;\r
-\r
-        /* Load data block */\r
-        struct ext4_block block;\r
-        rc = ext4_block_get( parent->fs->bdev, &block, fblock);\r
-        if (rc != EOK)\r
-            return rc;\r
-\r
-        /* Try to find entry in block */\r
-        struct ext4_directory_entry_ll *res_entry;\r
-        rc = ext4_dir_find_in_block(&block, sb, name_len, name,\r
-                &res_entry);\r
-        if (rc == EOK) {\r
-            result->block = block;\r
-            result->dentry = res_entry;\r
-            return EOK;\r
-        }\r
-\r
-        /* Entry not found - put block and continue to the next block */\r
-\r
-        rc = ext4_block_set(parent->fs->bdev, &block);\r
-        if (rc != EOK)\r
-            return rc;\r
-    }\r
-\r
-    /* Entry was not found */\r
-\r
-    result->block.lb_id = 0;\r
-    result->dentry = NULL;\r
-\r
-    return ENOENT;\r
-}\r
-\r
-int ext4_dir_remove_entry(struct ext4_inode_ref *parent, const char *name,\r
-    uint32_t name_len)\r
-{\r
-    /* Check if removing from directory */\r
-    if (!ext4_inode_is_type(&parent->fs->sb, parent->inode,\r
-            EXT4_INODE_MODE_DIRECTORY))\r
-        return ENOTDIR;\r
-\r
-    /* Try to find entry */\r
-    struct ext4_directory_search_result result;\r
-    int rc = ext4_dir_find_entry(&result, parent, name, name_len);\r
-    if (rc != EOK)\r
-        return rc;\r
-\r
-    /* Invalidate entry */\r
-    ext4_dir_entry_ll_set_inode(result.dentry, 0);\r
-\r
-    /* Store entry position in block */\r
-    uint32_t pos = (uint8_t *) result.dentry - result.block.data;\r
-\r
-    /*\r
-     * If entry is not the first in block, it must be merged\r
-     * with previous entry\r
-     */\r
-    if (pos != 0) {\r
-        uint32_t offset = 0;\r
-\r
-        /* Start from the first entry in block */\r
-        struct ext4_directory_entry_ll *tmp_dentry = (void *)result.block.data;\r
-        uint16_t tmp_dentry_length =\r
-                ext4_dir_entry_ll_get_entry_length(tmp_dentry);\r
-\r
-        /* Find direct predecessor of removed entry */\r
-        while ((offset + tmp_dentry_length) < pos) {\r
-            offset +=\r
-                    ext4_dir_entry_ll_get_entry_length(tmp_dentry);\r
-            tmp_dentry = (void *)(result.block.data + offset);\r
-            tmp_dentry_length =\r
-                    ext4_dir_entry_ll_get_entry_length(tmp_dentry);\r
-        }\r
-\r
-        ext4_assert(tmp_dentry_length + offset == pos);\r
-\r
-        /* Add to removed entry length to predecessor's length */\r
-        uint16_t del_entry_length =\r
-                ext4_dir_entry_ll_get_entry_length(result.dentry);\r
-        ext4_dir_entry_ll_set_entry_length(tmp_dentry,\r
-            tmp_dentry_length + del_entry_length);\r
-    }\r
-\r
-    result.block.dirty = true;\r
-\r
-    return ext4_dir_destroy_result(parent, &result);\r
-}\r
-\r
-int ext4_dir_try_insert_entry(struct ext4_sblock *sb,\r
-    struct ext4_block *target_block, struct ext4_inode_ref *child,\r
-    const char *name, uint32_t name_len)\r
-{\r
-    /* Compute required length entry and align it to 4 bytes */\r
-    uint32_t block_size = ext4_sb_get_block_size(sb);\r
-    uint16_t required_len = sizeof(struct ext4_fake_directory_entry) + name_len;\r
-\r
-    if ((required_len % 4) != 0)\r
-        required_len += 4 - (required_len % 4);\r
-\r
-    /* Initialize pointers, stop means to upper bound */\r
-    struct ext4_directory_entry_ll *dentry  = (void *)target_block->data;\r
-    struct ext4_directory_entry_ll *stop =\r
-            (void *)(target_block->data + block_size);\r
-\r
-    /*\r
-     * Walk through the block and check for invalid entries\r
-     * or entries with free space for new entry\r
-     */\r
-    while (dentry < stop) {\r
-        uint32_t inode = ext4_dir_entry_ll_get_inode(dentry);\r
-        uint16_t rec_len = ext4_dir_entry_ll_get_entry_length(dentry);\r
-\r
-        /* If invalid and large enough entry, use it */\r
-        if ((inode == 0) && (rec_len >= required_len)) {\r
-            ext4_dir_write_entry(sb, dentry, rec_len, child,\r
-                    name, name_len);\r
-            target_block->dirty = true;\r
-\r
-            return EOK;\r
-        }\r
-\r
-        /* Valid entry, try to split it */\r
-        if (inode != 0) {\r
-            uint16_t used_name_len =\r
-                    ext4_dir_entry_ll_get_name_length(sb, dentry);\r
-\r
-            uint16_t used_space =\r
-                    sizeof(struct ext4_fake_directory_entry) + used_name_len;\r
-\r
-            if ((used_name_len % 4) != 0)\r
-                used_space += 4 - (used_name_len % 4);\r
-\r
-            uint16_t free_space = rec_len - used_space;\r
-\r
-            /* There is free space for new entry */\r
-            if (free_space >= required_len) {\r
-                /* Cut tail of current entry */\r
-                ext4_dir_entry_ll_set_entry_length(dentry, used_space);\r
-                struct ext4_directory_entry_ll *new_entry =\r
-                        (void *) ((uint8_t *)dentry + used_space);\r
-                ext4_dir_write_entry(sb, new_entry,\r
-                    free_space, child, name, name_len);\r
-\r
-                target_block->dirty = true;\r
-                return EOK;\r
-            }\r
-        }\r
-\r
-        /* Jump to the next entry */\r
-        dentry = (void *) ((uint8_t *)dentry + rec_len);\r
-    }\r
-\r
-    /* No free space found for new entry */\r
-    return ENOSPC;\r
-}\r
-\r
-\r
-int ext4_dir_find_in_block(struct ext4_block *block, struct ext4_sblock *sb,\r
-    size_t name_len, const char *name,\r
-    struct ext4_directory_entry_ll **res_entry)\r
-{\r
-    /* Start from the first entry in block */\r
-    struct ext4_directory_entry_ll *dentry =\r
-            (struct ext4_directory_entry_ll *) block->data;\r
-\r
-    /* Set upper bound for cycling */\r
-    uint8_t *addr_limit = block->data + ext4_sb_get_block_size(sb);\r
-\r
-    /* Walk through the block and check entries */\r
-    while ((uint8_t *) dentry < addr_limit) {\r
-        /* Termination condition */\r
-        if ((uint8_t *) dentry + name_len > addr_limit)\r
-            break;\r
-\r
-        /* Valid entry - check it */\r
-        if (dentry->inode != 0) {\r
-            /* For more effectivity compare firstly only lengths */\r
-            if (ext4_dir_entry_ll_get_name_length(sb, dentry) ==\r
-                    name_len) {\r
-                /* Compare names */\r
-                if (memcmp((uint8_t *) name, dentry->name, name_len) == 0) {\r
-                    *res_entry = dentry;\r
-                    return EOK;\r
-                }\r
-            }\r
-        }\r
-\r
-        uint16_t dentry_len =\r
-                ext4_dir_entry_ll_get_entry_length(dentry);\r
-\r
-        /* Corrupted entry */\r
-        if (dentry_len == 0)\r
-            return EINVAL;\r
-\r
-        /* Jump to next entry */\r
-        dentry = (struct ext4_directory_entry_ll *)\r
-                        ((uint8_t *) dentry + dentry_len);\r
-    }\r
-\r
-    /* Entry not found */\r
-    return ENOENT;\r
-}\r
-\r
-int ext4_dir_destroy_result(struct ext4_inode_ref *parent,\r
-    struct ext4_directory_search_result *result)\r
-{\r
-    if (result->block.lb_id)\r
-        return ext4_block_set(parent->fs->bdev, &result->block);\r
-\r
-    return EOK;\r
-}\r
-\r
-/**\r
- * @}\r
- */\r
+/*
+ * Copyright (c) 2013 Grzegorz Kostka (kostka.grzegorz@gmail.com)
+ *
+ *
+ * HelenOS:
+ * Copyright (c) 2012 Martin Sucha
+ * Copyright (c) 2012 Frantisek Princ
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * - Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * - Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in the
+ *   documentation and/or other materials provided with the distribution.
+ * - The name of the author may not be used to endorse or promote products
+ *   derived from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
+ * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+ * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/** @addtogroup lwext4
+ * @{
+ */
+/**
+ * @file  ext4_dir.h
+ * @brief Directory handle procedures.
+ */
+
+#include "ext4_config.h"
+#include "ext4_dir.h"
+#include "ext4_dir_idx.h"
+#include "ext4_crc32c.h"
+#include "ext4_inode.h"
+#include "ext4_fs.h"
+
+#include <string.h>
+
+/****************************************************************************/
+
+/* Walk through a dirent block to find a checksum "dirent" at the tail */
+static struct ext4_dir_entry_tail *
+ext4_dir_get_tail(struct ext4_inode_ref *inode_ref,
+               struct ext4_dir_en *de)
+{
+       struct ext4_dir_entry_tail *t;
+       struct ext4_sblock *sb = &inode_ref->fs->sb;
+
+       t = EXT4_DIRENT_TAIL(de, ext4_sb_get_block_size(sb));
+
+       if (t->reserved_zero1 || t->reserved_zero2)
+               return NULL;
+       if (to_le16(t->rec_len) != sizeof(struct ext4_dir_entry_tail))
+               return NULL;
+       if (t->reserved_ft != EXT4_DIRENTRY_DIR_CSUM)
+               return NULL;
+
+       return t;
+}
+
+#if CONFIG_META_CSUM_ENABLE
+static uint32_t ext4_dir_csum(struct ext4_inode_ref *inode_ref,
+                             struct ext4_dir_en *dirent, int size)
+{
+       uint32_t csum;
+       struct ext4_sblock *sb = &inode_ref->fs->sb;
+       uint32_t ino_index = to_le32(inode_ref->index);
+       uint32_t ino_gen = to_le32(ext4_inode_get_generation(inode_ref->inode));
+
+       /* First calculate crc32 checksum against fs uuid */
+       csum = ext4_crc32c(EXT4_CRC32_INIT, sb->uuid, sizeof(sb->uuid));
+       /* Then calculate crc32 checksum against inode number
+        * and inode generation */
+       csum = ext4_crc32c(csum, &ino_index, sizeof(ino_index));
+       csum = ext4_crc32c(csum, &ino_gen, sizeof(ino_gen));
+       /* Finally calculate crc32 checksum against directory entries */
+       csum = ext4_crc32c(csum, dirent, size);
+       return csum;
+}
+#else
+#define ext4_dir_csum(...) 0
+#endif
+
+bool ext4_dir_csum_verify(struct ext4_inode_ref *inode_ref,
+                             struct ext4_dir_en *dirent)
+{
+#ifdef CONFIG_META_CSUM_ENABLE
+       struct ext4_dir_entry_tail *t;
+       struct ext4_sblock *sb = &inode_ref->fs->sb;
+
+       /* Compute the checksum only if the filesystem supports it */
+       if (ext4_sb_feature_ro_com(sb, EXT4_FRO_COM_METADATA_CSUM)) {
+               t = ext4_dir_get_tail(inode_ref, dirent);
+               if (!t) {
+                       /* There is no space to hold the checksum */
+                       return false;
+               }
+
+               ptrdiff_t __unused diff = (char *)t - (char *)dirent;
+               uint32_t csum = ext4_dir_csum(inode_ref, dirent, diff);
+               if (t->checksum != to_le32(csum))
+                       return false;
+
+       }
+#endif
+       return true;
+}
+
+void ext4_dir_init_entry_tail(struct ext4_dir_entry_tail *t)
+{
+       memset(t, 0, sizeof(struct ext4_dir_entry_tail));
+       t->rec_len = to_le16(sizeof(struct ext4_dir_entry_tail));
+       t->reserved_ft = EXT4_DIRENTRY_DIR_CSUM;
+}
+
+void ext4_dir_set_csum(struct ext4_inode_ref *inode_ref,
+                          struct ext4_dir_en *dirent)
+{
+       struct ext4_dir_entry_tail *t;
+       struct ext4_sblock *sb = &inode_ref->fs->sb;
+
+       /* Compute the checksum only if the filesystem supports it */
+       if (ext4_sb_feature_ro_com(sb, EXT4_FRO_COM_METADATA_CSUM)) {
+               t = ext4_dir_get_tail(inode_ref, dirent);
+               if (!t) {
+                       /* There is no space to hold the checksum */
+                       return;
+               }
+
+               ptrdiff_t __unused diff = (char *)t - (char *)dirent;
+               uint32_t csum = ext4_dir_csum(inode_ref, dirent, diff);
+               t->checksum = to_le32(csum);
+       }
+}
+
+/**@brief Do some checks before returning iterator.
+ * @param it Iterator to be checked
+ * @param block_size Size of data block
+ * @return Error code
+ */
+static int ext4_dir_iterator_set(struct ext4_dir_iter *it,
+                                uint32_t block_size)
+{
+       uint32_t off_in_block = it->curr_off % block_size;
+       struct ext4_sblock *sb = &it->inode_ref->fs->sb;
+
+       it->curr = NULL;
+
+       /* Ensure proper alignment */
+       if ((off_in_block % 4) != 0)
+               return EIO;
+
+       /* Ensure that the core of the entry does not overflow the block */
+       if (off_in_block > block_size - 8)
+               return EIO;
+
+       struct ext4_dir_en *en;
+       en = (void *)(it->curr_blk.data + off_in_block);
+
+       /* Ensure that the whole entry does not overflow the block */
+       uint16_t length = ext4_dir_en_get_entry_len(en);
+       if (off_in_block + length > block_size)
+               return EIO;
+
+       /* Ensure the name length is not too large */
+       if (ext4_dir_en_get_name_len(sb, en) > length - 8)
+               return EIO;
+
+       /* Everything OK - "publish" the entry */
+       it->curr = en;
+       return EOK;
+}
+
+/**@brief Seek to next valid directory entry.
+ *        Here can be jumped to the next data block.
+ * @param it  Initialized iterator
+ * @param pos Position of the next entry
+ * @return Error code
+ */
+static int ext4_dir_iterator_seek(struct ext4_dir_iter *it, uint64_t pos)
+{
+       struct ext4_sblock *sb = &it->inode_ref->fs->sb;
+       struct ext4_inode *inode = it->inode_ref->inode;
+       struct ext4_blockdev *bdev = it->inode_ref->fs->bdev;
+       uint64_t size = ext4_inode_get_size(sb, inode);
+       int r;
+
+       /* The iterator is not valid until we seek to the desired position */
+       it->curr = NULL;
+
+       /* Are we at the end? */
+       if (pos >= size) {
+               if (it->curr_blk.lb_id) {
+
+                       r = ext4_block_set(bdev, &it->curr_blk);
+                       it->curr_blk.lb_id = 0;
+                       if (r != EOK)
+                               return r;
+               }
+
+               it->curr_off = pos;
+               return EOK;
+       }
+
+       /* Compute next block address */
+       uint32_t block_size = ext4_sb_get_block_size(sb);
+       uint64_t current_blk_idx = it->curr_off / block_size;
+       uint32_t next_blk_idx = pos / block_size;
+
+       /*
+        * If we don't have a block or are moving across block boundary,
+        * we need to get another block
+        */
+       if ((it->curr_blk.lb_id == 0) ||
+           (current_blk_idx != next_blk_idx)) {
+               if (it->curr_blk.lb_id) {
+                       r = ext4_block_set(bdev, &it->curr_blk);
+                       it->curr_blk.lb_id = 0;
+
+                       if (r != EOK)
+                               return r;
+               }
+
+               ext4_fsblk_t next_blk;
+               r = ext4_fs_get_inode_dblk_idx(it->inode_ref, next_blk_idx,
+                                              &next_blk, false);
+               if (r != EOK)
+                       return r;
+
+               r = ext4_block_get(bdev, &it->curr_blk, next_blk);
+               if (r != EOK) {
+                       it->curr_blk.lb_id = 0;
+                       return r;
+               }
+       }
+
+       it->curr_off = pos;
+       return ext4_dir_iterator_set(it, block_size);
+}
+
+int ext4_dir_iterator_init(struct ext4_dir_iter *it,
+                          struct ext4_inode_ref *inode_ref, uint64_t pos)
+{
+       it->inode_ref = inode_ref;
+       it->curr = 0;
+       it->curr_off = 0;
+       it->curr_blk.lb_id = 0;
+
+       return ext4_dir_iterator_seek(it, pos);
+}
+
+int ext4_dir_iterator_next(struct ext4_dir_iter *it)
+{
+       int r = EOK;
+       uint16_t skip;
+
+       while (r == EOK) {
+               skip = ext4_dir_en_get_entry_len(it->curr);
+               r = ext4_dir_iterator_seek(it, it->curr_off + skip);
+
+               if (!it->curr)
+                       break;
+               /*Skip NULL referenced entry*/
+               if (ext4_dir_en_get_inode(it->curr) != 0)
+                       break;
+       }
+
+       return r;
+}
+
+int ext4_dir_iterator_fini(struct ext4_dir_iter *it)
+{
+       it->curr = 0;
+
+       if (it->curr_blk.lb_id)
+               return ext4_block_set(it->inode_ref->fs->bdev, &it->curr_blk);
+
+       return EOK;
+}
+
+void ext4_dir_write_entry(struct ext4_sblock *sb, struct ext4_dir_en *en,
+                         uint16_t entry_len, struct ext4_inode_ref *child,
+                         const char *name, size_t name_len)
+{
+       /* Check maximum entry length */
+       ext4_assert(entry_len <= ext4_sb_get_block_size(sb));
+
+       /* Set type of entry */
+       switch (ext4_inode_type(sb, child->inode)) {
+       case EXT4_INODE_MODE_DIRECTORY:
+               ext4_dir_en_set_inode_type(sb, en, EXT4_DE_DIR);
+               break;
+       case EXT4_INODE_MODE_FILE:
+               ext4_dir_en_set_inode_type(sb, en, EXT4_DE_REG_FILE);
+               break;
+       case EXT4_INODE_MODE_SOFTLINK:
+               ext4_dir_en_set_inode_type(sb, en, EXT4_DE_SYMLINK);
+               break;
+       default:
+               /* FIXME: right now we only support 3 inode type. */
+               ext4_assert(0);
+       }
+
+       /* Set basic attributes */
+       ext4_dir_en_set_inode(en, child->index);
+       ext4_dir_en_set_entry_len(en, entry_len);
+       ext4_dir_en_set_name_len(sb, en, name_len);
+
+       /* Write name */
+       memcpy(en->name, name, name_len);
+}
+
+int ext4_dir_add_entry(struct ext4_inode_ref *parent, const char *name,
+                      uint32_t name_len, struct ext4_inode_ref *child)
+{
+       int r;
+       struct ext4_fs *fs = parent->fs;
+       struct ext4_sblock *sb = &parent->fs->sb;
+
+#if CONFIG_DIR_INDEX_ENABLE
+       /* Index adding (if allowed) */
+       if ((ext4_sb_feature_com(sb, EXT4_FCOM_DIR_INDEX)) &&
+           (ext4_inode_has_flag(parent->inode, EXT4_INODE_FLAG_INDEX))) {
+               r = ext4_dir_dx_add_entry(parent, child, name);
+
+               /* Check if index is not corrupted */
+               if (r != EXT4_ERR_BAD_DX_DIR) {
+                       if (r != EOK)
+                               return r;
+
+                       return EOK;
+               }
+
+               /* Needed to clear dir index flag if corrupted */
+               ext4_inode_clear_flag(parent->inode, EXT4_INODE_FLAG_INDEX);
+               parent->dirty = true;
+       }
+#endif
+
+       /* Linear algorithm */
+       uint32_t iblock = 0;
+       ext4_fsblk_t fblock = 0;
+       uint32_t block_size = ext4_sb_get_block_size(sb);
+       uint32_t inode_size = ext4_inode_get_size(sb, parent->inode);
+       uint32_t total_blocks = inode_size / block_size;
+
+       /* Find block, where is space for new entry and try to add */
+       bool success = false;
+       for (iblock = 0; iblock < total_blocks; ++iblock) {
+               r = ext4_fs_get_inode_dblk_idx(parent, iblock, &fblock, false);
+               if (r != EOK)
+                       return r;
+
+               struct ext4_block block;
+               r = ext4_block_get(fs->bdev, &block, fblock);
+               if (r != EOK)
+                       return r;
+
+               if (!ext4_dir_csum_verify(parent, (void *)block.data)) {
+                       ext4_dbg(DEBUG_DIR,
+                                DBG_WARN "Leaf block checksum failed."
+                                "Inode: %" PRIu32", "
+                                "Block: %" PRIu32"\n",
+                                parent->index,
+                                iblock);
+               }
+
+               /* If adding is successful, function can finish */
+               r = ext4_dir_try_insert_entry(sb, parent, &block, child,
+                                               name, name_len);
+               if (r == EOK)
+                       success = true;
+
+               r = ext4_block_set(fs->bdev, &block);
+               if (r != EOK)
+                       return r;
+
+               if (success)
+                       return EOK;
+       }
+
+       /* No free block found - needed to allocate next data block */
+
+       iblock = 0;
+       fblock = 0;
+       r = ext4_fs_append_inode_dblk(parent, &fblock, &iblock);
+       if (r != EOK)
+               return r;
+
+       /* Load new block */
+       struct ext4_block b;
+
+       r = ext4_block_get_noread(fs->bdev, &b, fblock);
+       if (r != EOK)
+               return r;
+
+       /* Fill block with zeroes */
+       memset(b.data, 0, block_size);
+       struct ext4_dir_en *blk_en = (void *)b.data;
+
+       /* Save new block */
+       if (ext4_sb_feature_ro_com(sb, EXT4_FRO_COM_METADATA_CSUM)) {
+               uint16_t el = block_size - sizeof(struct ext4_dir_entry_tail);
+               ext4_dir_write_entry(sb, blk_en, el, child, name, name_len);
+               ext4_dir_init_entry_tail(EXT4_DIRENT_TAIL(b.data, block_size));
+       } else {
+               ext4_dir_write_entry(sb, blk_en, block_size, child, name,
+                               name_len);
+       }
+
+       ext4_dir_set_csum(parent, (void *)b.data);
+       ext4_bcache_set_dirty(b.buf);
+       r = ext4_block_set(fs->bdev, &b);
+
+       return r;
+}
+
+int ext4_dir_find_entry(struct ext4_dir_search_result *result,
+                       struct ext4_inode_ref *parent, const char *name,
+                       uint32_t name_len)
+{
+       int r;
+       struct ext4_sblock *sb = &parent->fs->sb;
+
+       /* Entry clear */
+       result->block.lb_id = 0;
+       result->dentry = NULL;
+
+#if CONFIG_DIR_INDEX_ENABLE
+       /* Index search */
+       if ((ext4_sb_feature_com(sb, EXT4_FCOM_DIR_INDEX)) &&
+           (ext4_inode_has_flag(parent->inode, EXT4_INODE_FLAG_INDEX))) {
+               r = ext4_dir_dx_find_entry(result, parent, name_len, name);
+               /* Check if index is not corrupted */
+               if (r != EXT4_ERR_BAD_DX_DIR) {
+                       if (r != EOK)
+                               return r;
+
+                       return EOK;
+               }
+
+               /* Needed to clear dir index flag if corrupted */
+               ext4_inode_clear_flag(parent->inode, EXT4_INODE_FLAG_INDEX);
+               parent->dirty = true;
+       }
+#endif
+
+       /* Linear algorithm */
+
+       uint32_t iblock;
+       ext4_fsblk_t fblock;
+       uint32_t block_size = ext4_sb_get_block_size(sb);
+       uint32_t inode_size = ext4_inode_get_size(sb, parent->inode);
+       uint32_t total_blocks = inode_size / block_size;
+
+       /* Walk through all data blocks */
+       for (iblock = 0; iblock < total_blocks; ++iblock) {
+               /* Load block address */
+               r = ext4_fs_get_inode_dblk_idx(parent, iblock, &fblock, false);
+               if (r != EOK)
+                       return r;
+
+               /* Load data block */
+               struct ext4_block b;
+               r = ext4_block_get(parent->fs->bdev, &b, fblock);
+               if (r != EOK)
+                       return r;
+
+               if (!ext4_dir_csum_verify(parent, (void *)b.data)) {
+                       ext4_dbg(DEBUG_DIR,
+                                DBG_WARN "Leaf block checksum failed."
+                                "Inode: %" PRIu32", "
+                                "Block: %" PRIu32"\n",
+                                parent->index,
+                                iblock);
+               }
+
+               /* Try to find entry in block */
+               struct ext4_dir_en *res_entry;
+               r = ext4_dir_find_in_block(&b, sb, name_len, name, &res_entry);
+               if (r == EOK) {
+                       result->block = b;
+                       result->dentry = res_entry;
+                       return EOK;
+               }
+
+               /* Entry not found - put block and continue to the next block */
+
+               r = ext4_block_set(parent->fs->bdev, &b);
+               if (r != EOK)
+                       return r;
+       }
+
+       return ENOENT;
+}
+
+int ext4_dir_remove_entry(struct ext4_inode_ref *parent, const char *name,
+                         uint32_t name_len)
+{
+       struct ext4_sblock *sb = &parent->fs->sb;
+       /* Check if removing from directory */
+       if (!ext4_inode_is_type(sb, parent->inode, EXT4_INODE_MODE_DIRECTORY))
+               return ENOTDIR;
+
+       /* Try to find entry */
+       struct ext4_dir_search_result result;
+       int rc = ext4_dir_find_entry(&result, parent, name, name_len);
+       if (rc != EOK)
+               return rc;
+
+       /* Invalidate entry */
+       ext4_dir_en_set_inode(result.dentry, 0);
+
+       /* Store entry position in block */
+       uint32_t pos = (uint8_t *)result.dentry - result.block.data;
+
+       /*
+        * If entry is not the first in block, it must be merged
+        * with previous entry
+        */
+       if (pos != 0) {
+               uint32_t offset = 0;
+
+               /* Start from the first entry in block */
+               struct ext4_dir_en *tmp_de =(void *)result.block.data;
+               uint16_t de_len = ext4_dir_en_get_entry_len(tmp_de);
+
+               /* Find direct predecessor of removed entry */
+               while ((offset + de_len) < pos) {
+                       offset += ext4_dir_en_get_entry_len(tmp_de);
+                       tmp_de = (void *)(result.block.data + offset);
+                       de_len = ext4_dir_en_get_entry_len(tmp_de);
+               }
+
+               ext4_assert(de_len + offset == pos);
+
+               /* Add to removed entry length to predecessor's length */
+               uint16_t del_len;
+               del_len = ext4_dir_en_get_entry_len(result.dentry);
+               ext4_dir_en_set_entry_len(tmp_de, de_len + del_len);
+       }
+
+       ext4_dir_set_csum(parent,
+                       (struct ext4_dir_en *)result.block.data);
+       ext4_bcache_set_dirty(result.block.buf);
+
+       return ext4_dir_destroy_result(parent, &result);
+}
+
+int ext4_dir_try_insert_entry(struct ext4_sblock *sb,
+                             struct ext4_inode_ref *inode_ref,
+                             struct ext4_block *dst_blk,
+                             struct ext4_inode_ref *child, const char *name,
+                             uint32_t name_len)
+{
+       /* Compute required length entry and align it to 4 bytes */
+       uint32_t block_size = ext4_sb_get_block_size(sb);
+       uint16_t required_len = sizeof(struct ext4_fake_dir_entry) + name_len;
+
+       if ((required_len % 4) != 0)
+               required_len += 4 - (required_len % 4);
+
+       /* Initialize pointers, stop means to upper bound */
+       struct ext4_dir_en *start = (void *)dst_blk->data;
+       struct ext4_dir_en *stop = (void *)(dst_blk->data + block_size);
+
+       /*
+        * Walk through the block and check for invalid entries
+        * or entries with free space for new entry
+        */
+       while (start < stop) {
+               uint32_t inode = ext4_dir_en_get_inode(start);
+               uint16_t rec_len = ext4_dir_en_get_entry_len(start);
+               uint8_t itype = ext4_dir_en_get_inode_type(sb, start);
+
+               /* If invalid and large enough entry, use it */
+               if ((inode == 0) && (itype != EXT4_DIRENTRY_DIR_CSUM) &&
+                   (rec_len >= required_len)) {
+                       ext4_dir_write_entry(sb, start, rec_len, child, name,
+                                            name_len);
+                       ext4_dir_set_csum(inode_ref, (void *)dst_blk->data);
+                       ext4_bcache_set_dirty(dst_blk->buf);
+
+                       return EOK;
+               }
+
+               /* Valid entry, try to split it */
+               if (inode != 0) {
+                       uint16_t used_len;
+                       used_len = ext4_dir_en_get_name_len(sb, start);
+
+                       uint16_t sz;
+                       sz = sizeof(struct ext4_fake_dir_entry) + used_len;
+
+                       if ((used_len % 4) != 0)
+                               sz += 4 - (used_len % 4);
+
+                       uint16_t free_space = rec_len - sz;
+
+                       /* There is free space for new entry */
+                       if (free_space >= required_len) {
+                               /* Cut tail of current entry */
+                               struct ext4_dir_en * new_entry;
+                               new_entry = (void *)((uint8_t *)start + sz);
+                               ext4_dir_en_set_entry_len(start, sz);
+                               ext4_dir_write_entry(sb, new_entry, free_space,
+                                                    child, name, name_len);
+
+                               ext4_dir_set_csum(inode_ref,
+                                                 (void *)dst_blk->data);
+                               ext4_bcache_set_dirty(dst_blk->buf);
+                               return EOK;
+                       }
+               }
+
+               /* Jump to the next entry */
+               start = (void *)((uint8_t *)start + rec_len);
+       }
+
+       /* No free space found for new entry */
+       return ENOSPC;
+}
+
+int ext4_dir_find_in_block(struct ext4_block *block, struct ext4_sblock *sb,
+                          size_t name_len, const char *name,
+                          struct ext4_dir_en **res_entry)
+{
+       /* Start from the first entry in block */
+       struct ext4_dir_en *de = (struct ext4_dir_en *)block->data;
+
+       /* Set upper bound for cycling */
+       uint8_t *addr_limit = block->data + ext4_sb_get_block_size(sb);
+
+       /* Walk through the block and check entries */
+       while ((uint8_t *)de < addr_limit) {
+               /* Termination condition */
+               if ((uint8_t *)de + name_len > addr_limit)
+                       break;
+
+               /* Valid entry - check it */
+               if (ext4_dir_en_get_inode(de) != 0) {
+                       /* For more efficient compare only lengths firstly*/
+                       uint16_t el = ext4_dir_en_get_name_len(sb, de);
+                       if (el == name_len) {
+                               /* Compare names */
+                               if (memcmp(name, de->name, name_len) == 0) {
+                                       *res_entry = de;
+                                       return EOK;
+                               }
+                       }
+               }
+
+               uint16_t de_len = ext4_dir_en_get_entry_len(de);
+
+               /* Corrupted entry */
+               if (de_len == 0)
+                       return EINVAL;
+
+               /* Jump to next entry */
+               de = (struct ext4_dir_en *)((uint8_t *)de + de_len);
+       }
+
+       /* Entry not found */
+       return ENOENT;
+}
+
+int ext4_dir_destroy_result(struct ext4_inode_ref *parent,
+                           struct ext4_dir_search_result *result)
+{
+       if (result->block.lb_id)
+               return ext4_block_set(parent->fs->bdev, &result->block);
+
+       return EOK;
+}
+
+/**
+ * @}
+ */