source: mainline/uspace/srv/fs/mfs/mfs_ops.c@ 84239b1

/*
 * Copyright (c) 2011 Maurizio Lombardi
 * 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 fs
 * @{
 */

#include <stdlib.h>
#include <fibril_synch.h>
#include <align.h>
#include <adt/hash_table.h>
#include <adt/hash.h>
#include <str.h>
#include "mfs.h"


static bool check_magic_number(uint16_t magic, bool *native,
    mfs_version_t *version, bool *longfilenames);
static errno_t mfs_node_core_get(fs_node_t **rfn, struct mfs_instance *inst,
    fs_index_t index);
static errno_t mfs_node_put(fs_node_t *fsnode);
static errno_t mfs_node_open(fs_node_t *fsnode);
static fs_index_t mfs_index_get(fs_node_t *fsnode);
static unsigned mfs_lnkcnt_get(fs_node_t *fsnode);
static bool mfs_is_directory(fs_node_t *fsnode);
static bool mfs_is_file(fs_node_t *fsnode);
static errno_t mfs_has_children(bool *has_children, fs_node_t *fsnode);
static errno_t mfs_root_get(fs_node_t **rfn, service_id_t service_id);
static service_id_t mfs_service_get(fs_node_t *fsnode);
static aoff64_t mfs_size_get(fs_node_t *node);
static errno_t mfs_match(fs_node_t **rfn, fs_node_t *pfn, const char *component);
static errno_t mfs_create_node(fs_node_t **rfn, service_id_t service_id, int flags);
static errno_t mfs_link(fs_node_t *pfn, fs_node_t *cfn, const char *name);
static errno_t mfs_unlink(fs_node_t *, fs_node_t *, const char *name);
static errno_t mfs_destroy_node(fs_node_t *fn);
static errno_t mfs_node_get(fs_node_t **rfn, service_id_t service_id,
    fs_index_t index);
static errno_t mfs_instance_get(service_id_t service_id,
    struct mfs_instance **instance);
static errno_t mfs_check_sanity(struct mfs_sb_info *sbi);
static bool is_power_of_two(uint32_t n);
static errno_t mfs_size_block(service_id_t service_id, uint32_t *size);
static errno_t mfs_total_block_count(service_id_t service_id, uint64_t *count);
static errno_t mfs_free_block_count(service_id_t service_id, uint64_t *count);

static hash_table_t open_nodes;
static FIBRIL_MUTEX_INITIALIZE(open_nodes_lock);

libfs_ops_t mfs_libfs_ops = {
        .size_get = mfs_size_get,
        .root_get = mfs_root_get,
        .service_get = mfs_service_get,
        .is_directory = mfs_is_directory,
        .is_file = mfs_is_file,
        .node_get = mfs_node_get,
        .node_put = mfs_node_put,
        .node_open = mfs_node_open,
        .index_get = mfs_index_get,
        .match = mfs_match,
        .create = mfs_create_node,
        .link = mfs_link,
        .unlink = mfs_unlink,
        .destroy = mfs_destroy_node,
        .has_children = mfs_has_children,
        .lnkcnt_get = mfs_lnkcnt_get,
        .size_block = mfs_size_block,
        .total_block_count = mfs_total_block_count,
        .free_block_count = mfs_free_block_count
};

/* Hash table interface for open nodes hash table */
typedef struct {
        service_id_t service_id;
        fs_index_t index;
} node_key_t;

static size_t
open_nodes_key_hash(void *key)
{
        node_key_t *node_key = (node_key_t*)key;
        return hash_combine(node_key->service_id, node_key->index);
}

static size_t
open_nodes_hash(const ht_link_t *item)
{
        struct mfs_node *m = hash_table_get_inst(item, struct mfs_node, link);
        return hash_combine(m->instance->service_id, m->ino_i->index);
}

static bool
open_nodes_key_equal(void *key, const ht_link_t *item)
{
        node_key_t *node_key = (node_key_t*)key;
        struct mfs_node *mnode = hash_table_get_inst(item, struct mfs_node, link);

        return node_key->service_id == mnode->instance->service_id
                && node_key->index == mnode->ino_i->index;
}

static hash_table_ops_t open_nodes_ops = {
        .hash = open_nodes_hash,
        .key_hash = open_nodes_key_hash,
        .key_equal = open_nodes_key_equal,
        .equal = NULL,
        .remove_callback = NULL,
};

errno_t
mfs_global_init(void)
{
        if (!hash_table_create(&open_nodes, 0, 0, &open_nodes_ops)) {
                return ENOMEM;
        }
        return EOK;
}

/** Read the superblock.
 */
static errno_t mfs_read_sb(service_id_t service_id, struct mfs_sb_info **rsbi)
{
        struct mfs_superblock *sb = NULL;
        struct mfs3_superblock *sb3 = NULL;
        struct mfs_sb_info *sbi;
        size_t bsize;
        bool native, longnames;
        mfs_version_t version;
        uint16_t magic;
        errno_t rc;

        /* Allocate space for generic MFS superblock */
        sbi = malloc(sizeof(*sbi));
        if (!sbi) {
                rc = ENOMEM;
                goto out_error;
        }

        sb = malloc(MFS_SUPERBLOCK_SIZE);
        if (!sb) {
                rc = ENOMEM;
                goto out_error;
        }

        rc = block_get_bsize(service_id, &bsize);
        if (rc != EOK) {
                rc = EIO;
                goto out_error;
        }

        /* We don't support other block size than 512 */
        if (bsize != 512) {
                rc = ENOTSUP;
                goto out_error;
        }

        /* Read the superblock */
        rc = block_read_direct(service_id, MFS_SUPERBLOCK << 1, 2, sb);
        if (rc != EOK)
                goto out_error;

        sb3 = (struct mfs3_superblock *) sb;

        if (check_magic_number(sb->s_magic, &native, &version, &longnames)) {
                /* This is a V1 or V2 Minix filesystem */
                magic = sb->s_magic;
        } else if (check_magic_number(sb3->s_magic, &native,
            &version, &longnames)) {
                /* This is a V3 Minix filesystem */
                magic = sb3->s_magic;
        } else {
                /* Not recognized */
                mfsdebug("magic number not recognized\n");
                rc = ENOTSUP;
                goto out_error;
        }

        mfsdebug("magic number recognized = %04x\n", magic);

        /* Fill superblock info structure */

        sbi->fs_version = version;
        sbi->long_names = longnames;
        sbi->native = native;
        sbi->magic = magic;
        sbi->isearch = 0;
        sbi->zsearch = 0;
        sbi->nfree_zones_valid = false;
        sbi->nfree_zones = 0;

        if (version == MFS_VERSION_V3) {
                sbi->ninodes = conv32(native, sb3->s_ninodes);
                sbi->ibmap_blocks = conv16(native, sb3->s_ibmap_blocks);
                sbi->zbmap_blocks = conv16(native, sb3->s_zbmap_blocks);
                sbi->firstdatazone = conv16(native, sb3->s_first_data_zone);
                sbi->log2_zone_size = conv16(native, sb3->s_log2_zone_size);
                sbi->max_file_size = conv32(native, sb3->s_max_file_size);
                sbi->nzones = conv32(native, sb3->s_nzones);
                sbi->block_size = conv16(native, sb3->s_block_size);
                sbi->ino_per_block = V3_INODES_PER_BLOCK(sbi->block_size);
                sbi->dirsize = MFS3_DIRSIZE;
                sbi->max_name_len = MFS3_MAX_NAME_LEN;
        } else {
                sbi->ninodes = conv16(native, sb->s_ninodes);
                sbi->ibmap_blocks = conv16(native, sb->s_ibmap_blocks);
                sbi->zbmap_blocks = conv16(native, sb->s_zbmap_blocks);
                sbi->firstdatazone = conv16(native, sb->s_first_data_zone);
                sbi->log2_zone_size = conv16(native, sb->s_log2_zone_size);
                sbi->max_file_size = conv32(native, sb->s_max_file_size);
                sbi->block_size = MFS_BLOCKSIZE;
                if (version == MFS_VERSION_V2) {
                        sbi->nzones = conv32(native, sb->s_nzones2);
                        sbi->ino_per_block = V2_INODES_PER_BLOCK;
                } else {
                        sbi->nzones = conv16(native, sb->s_nzones);
                        sbi->ino_per_block = V1_INODES_PER_BLOCK;
                }
                sbi->dirsize = longnames ? MFSL_DIRSIZE : MFS_DIRSIZE;
                sbi->max_name_len = longnames ? MFS_L_MAX_NAME_LEN :
                    MFS_MAX_NAME_LEN;
        }

        if (sbi->log2_zone_size != 0) {
                /* In MFS, file space is allocated per zones.
                 * Zones are a collection of consecutive blocks on disk.
                 *
                 * The current MFS implementation supports only filesystems
                 * where the size of a zone is equal to the
                 * size of a block.
                 */
                rc = ENOTSUP;
                goto out_error;
        }

        sbi->itable_off = 2 + sbi->ibmap_blocks + sbi->zbmap_blocks;
        if ((rc = mfs_check_sanity(sbi)) != EOK) {
                fprintf(stderr, "Filesystem corrupted, invalid superblock");
                goto out_error;
        }

        mfsdebug("read superblock successful\n");

        free(sb);
        *rsbi = sbi;
        return EOK;

out_error:
        if (sb)
                free(sb);
        if (sbi)
                free(sbi);
        return rc;
}


static errno_t mfs_fsprobe(service_id_t service_id, vfs_fs_probe_info_t *info)
{
        struct mfs_sb_info *sbi = NULL;
        errno_t rc;

        /* Initialize libblock */
        rc = block_init(service_id, 4096);
        if (rc != EOK)
                return rc;

        /* Read the superblock */
        rc = mfs_read_sb(service_id, &sbi);
        block_fini(service_id);

        return rc;
}

static errno_t
mfs_mounted(service_id_t service_id, const char *opts, fs_index_t *index,
    aoff64_t *size)
{
        enum cache_mode cmode;
        struct mfs_sb_info *sbi = NULL;
        struct mfs_instance *instance = NULL;
        errno_t rc;

        /* Check for option enabling write through. */
        if (str_cmp(opts, "wtcache") == 0)
                cmode = CACHE_MODE_WT;
        else
                cmode = CACHE_MODE_WB;

        /* Initialize libblock */
        rc = block_init(service_id, 4096);
        if (rc != EOK)
                return rc;

        /* Allocate space for filesystem instance */
        instance = malloc(sizeof(*instance));
        if (!instance) {
                rc = ENOMEM;
                goto out_error;
        }

        /* Read the superblock */
        rc = mfs_read_sb(service_id, &sbi);
        if (rc != EOK)
                goto out_error;

        rc = block_cache_init(service_id, sbi->block_size, 0, cmode);
        if (rc != EOK) {
                mfsdebug("block cache initialization failed\n");
                rc = EINVAL;
                goto out_error;
        }

        /* Initialize the instance structure and remember it */
        instance->service_id = service_id;
        instance->sbi = sbi;
        instance->open_nodes_cnt = 0;
        rc = fs_instance_create(service_id, instance);
        if (rc != EOK) {
                block_cache_fini(service_id);
                mfsdebug("fs instance creation failed\n");
                goto out_error;
        }

        mfsdebug("mount successful\n");

        fs_node_t *fn;
        mfs_node_get(&fn, service_id, MFS_ROOT_INO);

        struct mfs_node *mroot = fn->data;

        *index = mroot->ino_i->index;
        *size = mroot->ino_i->i_size;

        return mfs_node_put(fn);

out_error:
        block_fini(service_id);
        if (sbi)
                free(sbi);
        if(instance)
                free(instance);
        return rc;
}

static errno_t
mfs_unmounted(service_id_t service_id)
{
        struct mfs_instance *inst;

        mfsdebug("%s()\n", __FUNCTION__);

        errno_t r = mfs_instance_get(service_id, &inst);
        if (r != EOK)
                return r;

        if (inst->open_nodes_cnt != 0)
                return EBUSY;

        (void) block_cache_fini(service_id);
        block_fini(service_id);

        /* Remove and destroy the instance */
        (void) fs_instance_destroy(service_id);
        free(inst->sbi);
        free(inst);
        return EOK;
}

service_id_t
mfs_service_get(fs_node_t *fsnode)
{
        struct mfs_node *node = fsnode->data;
        return node->instance->service_id;
}

static errno_t
mfs_create_node(fs_node_t **rfn, service_id_t service_id, int flags)
{
        errno_t r;
        struct mfs_instance *inst;
        struct mfs_node *mnode;
        fs_node_t *fsnode;
        uint32_t inum;

        r = mfs_instance_get(service_id, &inst);
        if (r != EOK)
                return r;

        /* Alloc a new inode */
        r = mfs_alloc_inode(inst, &inum);
        if (r != EOK)
                return r;

        struct mfs_ino_info *ino_i;

        ino_i = malloc(sizeof(*ino_i));
        if (!ino_i) {
                r = ENOMEM;
                goto out_err;
        }

        mnode = malloc(sizeof(*mnode));
        if (!mnode) {
                r = ENOMEM;
                goto out_err_1;
        }

        fsnode = malloc(sizeof(fs_node_t));
        if (!fsnode) {
                r = ENOMEM;
                goto out_err_2;
        }

        if (flags & L_DIRECTORY) {
                ino_i->i_mode = S_IFDIR;
                ino_i->i_nlinks = 1; /* This accounts for the '.' dentry */
        } else {
                ino_i->i_mode = S_IFREG;
                ino_i->i_nlinks = 0;
        }

        ino_i->i_uid = 0;
        ino_i->i_gid = 0;
        ino_i->i_size = 0;
        ino_i->i_atime = 0;
        ino_i->i_mtime = 0;
        ino_i->i_ctime = 0;

        memset(ino_i->i_dzone, 0, sizeof(uint32_t) * V2_NR_DIRECT_ZONES);
        memset(ino_i->i_izone, 0, sizeof(uint32_t) * V2_NR_INDIRECT_ZONES);

        mfsdebug("new node idx = %d\n", (int) inum);

        ino_i->index = inum;
        ino_i->dirty = true;
        mnode->ino_i = ino_i;
        mnode->instance = inst;
        mnode->refcnt = 1;

        fibril_mutex_lock(&open_nodes_lock);
        hash_table_insert(&open_nodes, &mnode->link);
        fibril_mutex_unlock(&open_nodes_lock);
        inst->open_nodes_cnt++;

        mnode->ino_i->dirty = true;

        fs_node_initialize(fsnode);
        fsnode->data = mnode;
        mnode->fsnode = fsnode;
        *rfn = fsnode;

        return EOK;

out_err_2:
        free(mnode);
out_err_1:
        free(ino_i);
out_err:
        mfs_free_inode(inst, inum);
        return r;
}

static errno_t
mfs_match(fs_node_t **rfn, fs_node_t *pfn, const char *component)
{
        struct mfs_node *mnode = pfn->data;
        struct mfs_ino_info *ino_i = mnode->ino_i;
        struct mfs_dentry_info d_info;
        errno_t r;

        if (!S_ISDIR(ino_i->i_mode))
                return ENOTDIR;

        struct mfs_sb_info *sbi = mnode->instance->sbi;
        const size_t comp_size = str_size(component);

        unsigned i;
        for (i = 0; i < mnode->ino_i->i_size / sbi->dirsize; ++i) {
                r = mfs_read_dentry(mnode, &d_info, i);
                if (r != EOK)
                        return r;

                if (!d_info.d_inum) {
                        /* This entry is not used */
                        continue;
                }

                const size_t dentry_name_size = str_size(d_info.d_name);

                if (comp_size == dentry_name_size &&
                    memcmp(component, d_info.d_name, dentry_name_size) == 0) {
                        /* Hit! */
                        mfs_node_core_get(rfn, mnode->instance,
                            d_info.d_inum);
                        goto found;
                }
        }
        *rfn = NULL;
found:
        return EOK;
}

static aoff64_t
mfs_size_get(fs_node_t *node)
{
        const struct mfs_node *mnode = node->data;
        return mnode->ino_i->i_size;
}

static errno_t
mfs_node_get(fs_node_t **rfn, service_id_t service_id,
    fs_index_t index)
{
        errno_t rc;
        struct mfs_instance *instance;

        rc = mfs_instance_get(service_id, &instance);
        if (rc != EOK)
                return rc;

        return mfs_node_core_get(rfn, instance, index);
}

static errno_t
mfs_node_put(fs_node_t *fsnode)
{
        errno_t rc = EOK;
        struct mfs_node *mnode = fsnode->data;

        fibril_mutex_lock(&open_nodes_lock);

        assert(mnode->refcnt > 0);
        mnode->refcnt--;
        if (mnode->refcnt == 0) {
                hash_table_remove_item(&open_nodes, &mnode->link);
                assert(mnode->instance->open_nodes_cnt > 0);
                mnode->instance->open_nodes_cnt--;
                rc = mfs_put_inode(mnode);
                free(mnode->ino_i);
                free(mnode);
                free(fsnode);
        }

        fibril_mutex_unlock(&open_nodes_lock);
        return rc;
}

static errno_t
mfs_node_open(fs_node_t *fsnode)
{
        /*
         * Opening a file is stateless, nothing
         * to be done here.
         */
        return EOK;
}

static fs_index_t
mfs_index_get(fs_node_t *fsnode)
{
        struct mfs_node *mnode = fsnode->data;
        return mnode->ino_i->index;
}

static unsigned
mfs_lnkcnt_get(fs_node_t *fsnode)
{
        struct mfs_node *mnode = fsnode->data;

        mfsdebug("%s() %d\n", __FUNCTION__, mnode->ino_i->i_nlinks);

        if (S_ISDIR(mnode->ino_i->i_mode)) {
                if (mnode->ino_i->i_nlinks > 1)
                        return 1;
                else
                        return 0;
        } else
                return mnode->ino_i->i_nlinks;
}

static errno_t
mfs_node_core_get(fs_node_t **rfn, struct mfs_instance *inst,
    fs_index_t index)
{
        fs_node_t *node = NULL;
        struct mfs_node *mnode = NULL;
        errno_t rc;

        fibril_mutex_lock(&open_nodes_lock);

        /* Check if the node is not already open */
        node_key_t key = {
                .service_id = inst->service_id,
                .index = index
        };

        ht_link_t *already_open = hash_table_find(&open_nodes, &key);

        if (already_open) {
                mnode = hash_table_get_inst(already_open, struct mfs_node, link);

                *rfn = mnode->fsnode;
                mnode->refcnt++;

                fibril_mutex_unlock(&open_nodes_lock);
                return EOK;
        }

        node = malloc(sizeof(fs_node_t));
        if (!node) {
                rc = ENOMEM;
                goto out_err;
        }

        fs_node_initialize(node);

        mnode = malloc(sizeof(*mnode));
        if (!mnode) {
                rc = ENOMEM;
                goto out_err;
        }

        struct mfs_ino_info *ino_i;

        rc = mfs_get_inode(inst, &ino_i, index);
        if (rc != EOK)
                goto out_err;

        ino_i->index = index;
        mnode->ino_i = ino_i;
        mnode->refcnt = 1;

        mnode->instance = inst;
        node->data = mnode;
        mnode->fsnode = node;
        *rfn = node;

        hash_table_insert(&open_nodes, &mnode->link);
        inst->open_nodes_cnt++;

        fibril_mutex_unlock(&open_nodes_lock);

        return EOK;

out_err:
        if (node)
                free(node);
        if (mnode)
                free(mnode);
        fibril_mutex_unlock(&open_nodes_lock);
        return rc;
}

static bool
mfs_is_directory(fs_node_t *fsnode)
{
        const struct mfs_node *node = fsnode->data;
        return S_ISDIR(node->ino_i->i_mode);
}

static bool
mfs_is_file(fs_node_t *fsnode)
{
        struct mfs_node *node = fsnode->data;
        return S_ISREG(node->ino_i->i_mode);
}

static errno_t
mfs_root_get(fs_node_t **rfn, service_id_t service_id)
{
        errno_t rc = mfs_node_get(rfn, service_id, MFS_ROOT_INO);
        return rc;
}

static errno_t
mfs_link(fs_node_t *pfn, fs_node_t *cfn, const char *name)
{
        struct mfs_node *parent = pfn->data;
        struct mfs_node *child = cfn->data;
        struct mfs_sb_info *sbi = parent->instance->sbi;
        bool destroy_dentry = false;

        if (str_size(name) > sbi->max_name_len)
                return ENAMETOOLONG;

        errno_t r = mfs_insert_dentry(parent, name, child->ino_i->index);
        if (r != EOK)
                return r;

        if (S_ISDIR(child->ino_i->i_mode)) {
                if (child->ino_i->i_nlinks != 1) {
                        /* It's not possible to hardlink directories in MFS */
                        destroy_dentry = true;
                        r = EMLINK;
                        goto exit;
                }
                r = mfs_insert_dentry(child, ".", child->ino_i->index);
                if (r != EOK) {
                        destroy_dentry = true;
                        goto exit;
                }

                r = mfs_insert_dentry(child, "..", parent->ino_i->index);
                if (r != EOK) {
                        mfs_remove_dentry(child, ".");
                        destroy_dentry = true;
                        goto exit;
                }

                parent->ino_i->i_nlinks++;
                parent->ino_i->dirty = true;
        }

exit:
        if (destroy_dentry) {
                errno_t r2 = mfs_remove_dentry(parent, name);
                if (r2 != EOK)
                        r = r2;
        } else {
                child->ino_i->i_nlinks++;
                child->ino_i->dirty = true;
        }
        return r;
}

static errno_t
mfs_unlink(fs_node_t *pfn, fs_node_t *cfn, const char *name)
{
        struct mfs_node *parent = pfn->data;
        struct mfs_node *child = cfn->data;
        bool has_children;
        errno_t r;

        if (!parent)
                return EBUSY;

        r = mfs_has_children(&has_children, cfn);
        if (r != EOK)
                return r;

        if (has_children)
                return ENOTEMPTY;

        r = mfs_remove_dentry(parent, name);
        if (r != EOK)
                return r;

        struct mfs_ino_info *chino = child->ino_i;

        assert(chino->i_nlinks >= 1);
        chino->i_nlinks--;
        mfsdebug("Links: %d\n", chino->i_nlinks);

        if (chino->i_nlinks <= 1 && S_ISDIR(chino->i_mode)) {
                /* The child directory will be destroyed, decrease the
                 * parent hard links counter.
                 */
                parent->ino_i->i_nlinks--;
                parent->ino_i->dirty = true;
        }

        chino->dirty = true;

        return r;
}

static errno_t
mfs_has_children(bool *has_children, fs_node_t *fsnode)
{
        struct mfs_node *mnode = fsnode->data;
        struct mfs_sb_info *sbi = mnode->instance->sbi;
        errno_t r;

        *has_children = false;

        if (!S_ISDIR(mnode->ino_i->i_mode))
                goto out;

        struct mfs_dentry_info d_info;

        /* The first two dentries are always . and .. */
        unsigned i;
        for (i = 2; i < mnode->ino_i->i_size / sbi->dirsize; ++i) {
                r = mfs_read_dentry(mnode, &d_info, i);
                if (r != EOK)
                        return r;

                if (d_info.d_inum) {
                        /* A valid entry has been found */
                        *has_children = true;
                        break;
                }
        }
out:

        return EOK;
}

static errno_t
mfs_read(service_id_t service_id, fs_index_t index, aoff64_t pos,
    size_t *rbytes)
{
        errno_t rc;
        errno_t tmp;
        fs_node_t *fn = NULL;

        rc = mfs_node_get(&fn, service_id, index);
        if (rc != EOK)
                return rc;
        if (!fn)
                return ENOENT;

        struct mfs_node *mnode;
        struct mfs_ino_info *ino_i;
        size_t len, bytes = 0;
        ipc_callid_t callid;

        mnode = fn->data;
        ino_i = mnode->ino_i;

        if (!async_data_read_receive(&callid, &len)) {
                rc = EINVAL;
                goto out_error;
        }

        if (S_ISDIR(ino_i->i_mode)) {
                aoff64_t spos = pos;
                struct mfs_dentry_info d_info;
                struct mfs_sb_info *sbi = mnode->instance->sbi;

                if (pos < 2) {
                        /* Skip the first two dentries ('.' and '..') */
                        pos = 2;
                }

                for (; pos < mnode->ino_i->i_size / sbi->dirsize; ++pos) {
                        rc = mfs_read_dentry(mnode, &d_info, pos);
                        if (rc != EOK)
                                goto out_error;

                        if (d_info.d_inum) {
                                /* Dentry found! */
                                goto found;
                        }
                }

                rc = mfs_node_put(fn);
                async_answer_0(callid, rc != EOK ? rc : ENOENT);
                return rc;
found:
                async_data_read_finalize(callid, d_info.d_name,
                    str_size(d_info.d_name) + 1);
                bytes = ((pos - spos) + 1);
        } else {
                struct mfs_sb_info *sbi = mnode->instance->sbi;

                if (pos >= (size_t) ino_i->i_size) {
                        /* Trying to read beyond the end of file */
                        bytes = 0;
                        (void) async_data_read_finalize(callid, NULL, 0);
                        goto out_success;
                }

                bytes = min(len, sbi->block_size - pos % sbi->block_size);
                bytes = min(bytes, ino_i->i_size - pos);

                uint32_t zone;
                block_t *b;

                rc = mfs_read_map(&zone, mnode, pos);
                if (rc != EOK)
                        goto out_error;

                if (zone == 0) {
                        /* sparse file */
                        uint8_t *buf = malloc(sbi->block_size);
                        if (!buf) {
                                rc = ENOMEM;
                                goto out_error;
                        }
                        memset(buf, 0, sizeof(sbi->block_size));
                        async_data_read_finalize(callid,
                            buf + pos % sbi->block_size, bytes);
                        free(buf);
                        goto out_success;
                }

                rc = block_get(&b, service_id, zone, BLOCK_FLAGS_NONE);
                if (rc != EOK)
                        goto out_error;

                async_data_read_finalize(callid, b->data +
                    pos % sbi->block_size, bytes);

                rc = block_put(b);
                if (rc != EOK) {
                        mfs_node_put(fn);
                        return rc;
                }
        }
out_success:
        rc = mfs_node_put(fn);
        *rbytes = bytes;
        return rc;
out_error:
        tmp = mfs_node_put(fn);
        async_answer_0(callid, tmp != EOK ? tmp : rc);
        return tmp != EOK ? tmp : rc;
}

static errno_t
mfs_write(service_id_t service_id, fs_index_t index, aoff64_t pos,
    size_t *wbytes, aoff64_t *nsize)
{
        fs_node_t *fn;
        errno_t r;
        int flags = BLOCK_FLAGS_NONE;

        r = mfs_node_get(&fn, service_id, index);
        if (r != EOK)
                return r;
        if (!fn)
                return ENOENT;

        ipc_callid_t callid;
        size_t len;

        if (!async_data_write_receive(&callid, &len)) {
                r = EINVAL;
                goto out_err;
        }

        struct mfs_node *mnode = fn->data;
        struct mfs_sb_info *sbi = mnode->instance->sbi;
        struct mfs_ino_info *ino_i = mnode->ino_i;
        const size_t bs = sbi->block_size;
        size_t bytes = min(len, bs - (pos % bs));
        uint32_t block;

        if (bytes == bs)
                flags = BLOCK_FLAGS_NOREAD;

        r = mfs_read_map(&block, mnode, pos);
        if (r != EOK)
                goto out_err;

        if (block == 0) {
                uint32_t dummy;

                r = mfs_alloc_zone(mnode->instance, &block);
                if (r != EOK)
                        goto out_err;

                r = mfs_write_map(mnode, pos, block, &dummy);
                if (r != EOK) {
                        mfs_free_zone(mnode->instance, block);
                        goto out_err;
                }

                flags = BLOCK_FLAGS_NOREAD;
        }

        block_t *b;
        r = block_get(&b, service_id, block, flags);
        if (r != EOK)
                goto out_err;

        if (flags == BLOCK_FLAGS_NOREAD)
                memset(b->data, 0, sbi->block_size);

        async_data_write_finalize(callid, b->data + (pos % bs), bytes);
        b->dirty = true;

        r = block_put(b);
        if (r != EOK) {
                mfs_node_put(fn);
                return r;
        }

        if (pos + bytes > ino_i->i_size) {
                ino_i->i_size = pos + bytes;
                ino_i->dirty = true;
        }
        r = mfs_node_put(fn);
        *nsize = ino_i->i_size;
        *wbytes = bytes;
        return r;

out_err:
        mfs_node_put(fn);
        async_answer_0(callid, r);
        return r;
}

static errno_t
mfs_destroy(service_id_t service_id, fs_index_t index)
{
        fs_node_t *fn = NULL;
        errno_t r;

        r = mfs_node_get(&fn, service_id, index);
        if (r != EOK)
                return r;
        if (!fn)
                return ENOENT;

        /* Destroy the inode */
        return mfs_destroy_node(fn);
}

static errno_t
mfs_destroy_node(fs_node_t *fn)
{
        struct mfs_node *mnode = fn->data;
        bool has_children;
        errno_t r;

        mfsdebug("mfs_destroy_node %d\n", mnode->ino_i->index);

        r = mfs_has_children(&has_children, fn);
        if (r != EOK)
                goto out;

        assert(!has_children);

        /* Free the entire inode content */
        r = mfs_inode_shrink(mnode, mnode->ino_i->i_size);
        if (r != EOK)
                goto out;

        /* Mark the inode as free in the bitmap */
        r = mfs_free_inode(mnode->instance, mnode->ino_i->index);

out:
        mfs_node_put(fn);
        return r;
}

static errno_t
mfs_truncate(service_id_t service_id, fs_index_t index, aoff64_t size)
{
        fs_node_t *fn;
        errno_t r;

        r = mfs_node_get(&fn, service_id, index);
        if (r != EOK)
                return r;
        if (!fn)
                return r;

        struct mfs_node *mnode = fn->data;
        struct mfs_ino_info *ino_i = mnode->ino_i;

        if (ino_i->i_size == size)
                r = EOK;
        else
                r = mfs_inode_shrink(mnode, ino_i->i_size - size);

        mfs_node_put(fn);
        return r;
}

static errno_t
mfs_instance_get(service_id_t service_id, struct mfs_instance **instance)
{
        void *data;
        errno_t rc;

        rc = fs_instance_get(service_id, &data);
        if (rc == EOK)
                *instance = (struct mfs_instance *) data;
        else {
                mfsdebug("instance not found\n");
        }

        return rc;
}

static bool
check_magic_number(uint16_t magic, bool *native,
                mfs_version_t *version, bool *longfilenames)
{
        bool rc = true;
        *longfilenames = false;

        if (magic == MFS_MAGIC_V1 || magic == MFS_MAGIC_V1R) {
                *native = magic == MFS_MAGIC_V1;
                *version = MFS_VERSION_V1;
        } else if (magic == MFS_MAGIC_V1L || magic == MFS_MAGIC_V1LR) {
                *native = magic == MFS_MAGIC_V1L;
                *version = MFS_VERSION_V1;
                *longfilenames = true;
        } else if (magic == MFS_MAGIC_V2 || magic == MFS_MAGIC_V2R) {
                *native = magic == MFS_MAGIC_V2;
                *version = MFS_VERSION_V2;
        } else if (magic == MFS_MAGIC_V2L || magic == MFS_MAGIC_V2LR) {
                *native = magic == MFS_MAGIC_V2L;
                *version = MFS_VERSION_V2;
                *longfilenames = true;
        } else if (magic == MFS_MAGIC_V3 || magic == MFS_MAGIC_V3R) {
                *native = magic == MFS_MAGIC_V3;
                *version = MFS_VERSION_V3;
        } else
                rc = false;

        return rc;
}

/** Filesystem sanity check
 *
 * @param Pointer to the MFS superblock.
 *
 * @return EOK on success, ENOTSUP otherwise.
 */
static errno_t
mfs_check_sanity(struct mfs_sb_info *sbi)
{
        if (!is_power_of_two(sbi->block_size) ||
            sbi->block_size < MFS_MIN_BLOCKSIZE ||
            sbi->block_size > MFS_MAX_BLOCKSIZE)
                return ENOTSUP;
        else if (sbi->ibmap_blocks == 0 || sbi->zbmap_blocks == 0)
                return ENOTSUP;
        else if (sbi->ninodes == 0 || sbi->nzones == 0)
                return ENOTSUP;
        else if (sbi->firstdatazone == 0)
                return ENOTSUP;

        return EOK;
}

static errno_t
mfs_close(service_id_t service_id, fs_index_t index)
{
        return 0;
}

static errno_t
mfs_sync(service_id_t service_id, fs_index_t index)
{
        fs_node_t *fn = NULL;
        errno_t rc = mfs_node_get(&fn, service_id, index);
        if (rc != EOK)
                return rc;
        if (!fn)
                return ENOENT;

        struct mfs_node *mnode = fn->data;
        mnode->ino_i->dirty = true;

        return mfs_node_put(fn);
}

/** Check if a given number is a power of two.
 *
 * @param n     The number to check.
 *
 * @return      true if it is a power of two, false otherwise.
 */
static bool
is_power_of_two(uint32_t n)
{
        if (n == 0)
                return false;

        return (n & (n - 1)) == 0;
}

static errno_t
mfs_size_block(service_id_t service_id, uint32_t *size)
{
        struct mfs_instance *inst;
        errno_t rc;

        rc = mfs_instance_get(service_id, &inst);
        if (rc != EOK)
                return rc;

        if (NULL == inst)
                return ENOENT;

        *size = inst->sbi->block_size;

        return EOK;
}

static errno_t
mfs_total_block_count(service_id_t service_id, uint64_t *count)
{
        struct mfs_instance *inst;
        errno_t rc;

        rc = mfs_instance_get(service_id, &inst);
        if (rc != EOK)
                return rc;

        if (NULL == inst)
                return ENOENT;

        *count = (uint64_t) MFS_BMAP_SIZE_BITS(inst->sbi, BMAP_ZONE);

        return EOK;
}

static errno_t
mfs_free_block_count(service_id_t service_id, uint64_t *count)
{
        uint32_t block_free;

        struct mfs_instance *inst;
        errno_t rc = mfs_instance_get(service_id, &inst);
        if (rc != EOK)
                return rc;

        struct mfs_sb_info *sbi = inst->sbi;

        if (!sbi->nfree_zones_valid) {
                /* The cached number of free zones is not valid,
                 * we need to scan the bitmap to retrieve the
                 * current value.
                 */

                rc = mfs_count_free_zones(inst, &block_free);
                if (rc != EOK)
                        return rc;

                sbi->nfree_zones = block_free;
                sbi->nfree_zones_valid = true;
        }

        *count = sbi->nfree_zones;

        return rc;
}

vfs_out_ops_t mfs_ops = {
        .fsprobe = mfs_fsprobe,
        .mounted = mfs_mounted,
        .unmounted = mfs_unmounted,
        .read = mfs_read,
        .write = mfs_write,
        .truncate = mfs_truncate,
        .close = mfs_close,
        .destroy = mfs_destroy,
        .sync = mfs_sync,
};

/**
 * @}
 */