vhost: block migration if backend does not log memory

[qemu.git] / block / qcow2-cluster.c

/*
 * Block driver for the QCOW version 2 format
 *
 * Copyright (c) 2004-2006 Fabrice Bellard
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

#include <zlib.h>

#include "qemu-common.h"
#include "block/block_int.h"
#include "block/qcow2.h"
#include "trace.h"

int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
                        bool exact_size)
{
    BDRVQcowState *s = bs->opaque;
    int new_l1_size2, ret, i;
    uint64_t *new_l1_table;
    int64_t old_l1_table_offset, old_l1_size;
    int64_t new_l1_table_offset, new_l1_size;
    uint8_t data[12];

    if (min_size <= s->l1_size)
        return 0;

    /* Do a sanity check on min_size before trying to calculate new_l1_size
     * (this prevents overflows during the while loop for the calculation of
     * new_l1_size) */
    if (min_size > INT_MAX / sizeof(uint64_t)) {
        return -EFBIG;
    }

    if (exact_size) {
        new_l1_size = min_size;
    } else {
        /* Bump size up to reduce the number of times we have to grow */
        new_l1_size = s->l1_size;
        if (new_l1_size == 0) {
            new_l1_size = 1;
        }
        while (min_size > new_l1_size) {
            new_l1_size = (new_l1_size * 3 + 1) / 2;
        }
    }

    if (new_l1_size > INT_MAX / sizeof(uint64_t)) {
        return -EFBIG;
    }

#ifdef DEBUG_ALLOC2
    fprintf(stderr, "grow l1_table from %d to %" PRId64 "\n",
            s->l1_size, new_l1_size);
#endif

    new_l1_size2 = sizeof(uint64_t) * new_l1_size;
    new_l1_table = g_malloc0(align_offset(new_l1_size2, 512));
    memcpy(new_l1_table, s->l1_table, s->l1_size * sizeof(uint64_t));

    /* write new table (align to cluster) */
    BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ALLOC_TABLE);
    new_l1_table_offset = qcow2_alloc_clusters(bs, new_l1_size2);
    if (new_l1_table_offset < 0) {
        g_free(new_l1_table);
        return new_l1_table_offset;
    }

    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
    if (ret < 0) {
        goto fail;
    }

    /* the L1 position has not yet been updated, so these clusters must
     * indeed be completely free */
    ret = qcow2_pre_write_overlap_check(bs, 0, new_l1_table_offset,
                                        new_l1_size2);
    if (ret < 0) {
        goto fail;
    }

    BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_WRITE_TABLE);
    for(i = 0; i < s->l1_size; i++)
        new_l1_table[i] = cpu_to_be64(new_l1_table[i]);
    ret = bdrv_pwrite_sync(bs->file, new_l1_table_offset, new_l1_table, new_l1_size2);
    if (ret < 0)
        goto fail;
    for(i = 0; i < s->l1_size; i++)
        new_l1_table[i] = be64_to_cpu(new_l1_table[i]);

    /* set new table */
    BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ACTIVATE_TABLE);
    cpu_to_be32w((uint32_t*)data, new_l1_size);
    stq_be_p(data + 4, new_l1_table_offset);
    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, l1_size), data,sizeof(data));
    if (ret < 0) {
        goto fail;
    }
    g_free(s->l1_table);
    old_l1_table_offset = s->l1_table_offset;
    s->l1_table_offset = new_l1_table_offset;
    s->l1_table = new_l1_table;
    old_l1_size = s->l1_size;
    s->l1_size = new_l1_size;
    qcow2_free_clusters(bs, old_l1_table_offset, old_l1_size * sizeof(uint64_t),
                        QCOW2_DISCARD_OTHER);
    return 0;
 fail:
    g_free(new_l1_table);
    qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2,
                        QCOW2_DISCARD_OTHER);
    return ret;
}

/*
 * l2_load
 *
 * Loads a L2 table into memory. If the table is in the cache, the cache
 * is used; otherwise the L2 table is loaded from the image file.
 *
 * Returns a pointer to the L2 table on success, or NULL if the read from
 * the image file failed.
 */

static int l2_load(BlockDriverState *bs, uint64_t l2_offset,
    uint64_t **l2_table)
{
    BDRVQcowState *s = bs->opaque;
    int ret;

    ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset, (void**) l2_table);

    return ret;
}

/*
 * Writes one sector of the L1 table to the disk (can't update single entries
 * and we really don't want bdrv_pread to perform a read-modify-write)
 */
#define L1_ENTRIES_PER_SECTOR (512 / 8)
int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t buf[L1_ENTRIES_PER_SECTOR];
    int l1_start_index;
    int i, ret;

    l1_start_index = l1_index & ~(L1_ENTRIES_PER_SECTOR - 1);
    for (i = 0; i < L1_ENTRIES_PER_SECTOR; i++) {
        buf[i] = cpu_to_be64(s->l1_table[l1_start_index + i]);
    }

    ret = qcow2_pre_write_overlap_check(bs, QCOW2_OL_ACTIVE_L1,
            s->l1_table_offset + 8 * l1_start_index, sizeof(buf));
    if (ret < 0) {
        return ret;
    }

    BLKDBG_EVENT(bs->file, BLKDBG_L1_UPDATE);
    ret = bdrv_pwrite_sync(bs->file, s->l1_table_offset + 8 * l1_start_index,
        buf, sizeof(buf));
    if (ret < 0) {
        return ret;
    }

    return 0;
}

/*
 * l2_allocate
 *
 * Allocate a new l2 entry in the file. If l1_index points to an already
 * used entry in the L2 table (i.e. we are doing a copy on write for the L2
 * table) copy the contents of the old L2 table into the newly allocated one.
 * Otherwise the new table is initialized with zeros.
 *
 */

static int l2_allocate(BlockDriverState *bs, int l1_index, uint64_t **table)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t old_l2_offset;
    uint64_t *l2_table = NULL;
    int64_t l2_offset;
    int ret;

    old_l2_offset = s->l1_table[l1_index];

    trace_qcow2_l2_allocate(bs, l1_index);

    /* allocate a new l2 entry */

    l2_offset = qcow2_alloc_clusters(bs, s->l2_size * sizeof(uint64_t));
    if (l2_offset < 0) {
        ret = l2_offset;
        goto fail;
    }

    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
    if (ret < 0) {
        goto fail;
    }

    /* allocate a new entry in the l2 cache */

    trace_qcow2_l2_allocate_get_empty(bs, l1_index);
    ret = qcow2_cache_get_empty(bs, s->l2_table_cache, l2_offset, (void**) table);
    if (ret < 0) {
        goto fail;
    }

    l2_table = *table;

    if ((old_l2_offset & L1E_OFFSET_MASK) == 0) {
        /* if there was no old l2 table, clear the new table */
        memset(l2_table, 0, s->l2_size * sizeof(uint64_t));
    } else {
        uint64_t* old_table;

        /* if there was an old l2 table, read it from the disk */
        BLKDBG_EVENT(bs->file, BLKDBG_L2_ALLOC_COW_READ);
        ret = qcow2_cache_get(bs, s->l2_table_cache,
            old_l2_offset & L1E_OFFSET_MASK,
            (void**) &old_table);
        if (ret < 0) {
            goto fail;
        }

        memcpy(l2_table, old_table, s->cluster_size);

        ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &old_table);
        if (ret < 0) {
            goto fail;
        }
    }

    /* write the l2 table to the file */
    BLKDBG_EVENT(bs->file, BLKDBG_L2_ALLOC_WRITE);

    trace_qcow2_l2_allocate_write_l2(bs, l1_index);
    qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
    ret = qcow2_cache_flush(bs, s->l2_table_cache);
    if (ret < 0) {
        goto fail;
    }

    /* update the L1 entry */
    trace_qcow2_l2_allocate_write_l1(bs, l1_index);
    s->l1_table[l1_index] = l2_offset | QCOW_OFLAG_COPIED;
    ret = qcow2_write_l1_entry(bs, l1_index);
    if (ret < 0) {
        goto fail;
    }

    *table = l2_table;
    trace_qcow2_l2_allocate_done(bs, l1_index, 0);
    return 0;

fail:
    trace_qcow2_l2_allocate_done(bs, l1_index, ret);
    if (l2_table != NULL) {
        qcow2_cache_put(bs, s->l2_table_cache, (void**) table);
    }
    s->l1_table[l1_index] = old_l2_offset;
    if (l2_offset > 0) {
        qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
                            QCOW2_DISCARD_ALWAYS);
    }
    return ret;
}

/*
 * Checks how many clusters in a given L2 table are contiguous in the image
 * file. As soon as one of the flags in the bitmask stop_flags changes compared
 * to the first cluster, the search is stopped and the cluster is not counted
 * as contiguous. (This allows it, for example, to stop at the first compressed
 * cluster which may require a different handling)
 */
static int count_contiguous_clusters(uint64_t nb_clusters, int cluster_size,
        uint64_t *l2_table, uint64_t stop_flags)
{
    int i;
    uint64_t mask = stop_flags | L2E_OFFSET_MASK | QCOW_OFLAG_COMPRESSED;
    uint64_t first_entry = be64_to_cpu(l2_table[0]);
    uint64_t offset = first_entry & mask;

    if (!offset)
        return 0;

    assert(qcow2_get_cluster_type(first_entry) != QCOW2_CLUSTER_COMPRESSED);

    for (i = 0; i < nb_clusters; i++) {
        uint64_t l2_entry = be64_to_cpu(l2_table[i]) & mask;
        if (offset + (uint64_t) i * cluster_size != l2_entry) {
            break;
        }
    }

        return i;
}

static int count_contiguous_free_clusters(uint64_t nb_clusters, uint64_t *l2_table)
{
    int i;

    for (i = 0; i < nb_clusters; i++) {
        int type = qcow2_get_cluster_type(be64_to_cpu(l2_table[i]));

        if (type != QCOW2_CLUSTER_UNALLOCATED) {
            break;
        }
    }

    return i;
}

/* The crypt function is compatible with the linux cryptoloop
   algorithm for < 4 GB images. NOTE: out_buf == in_buf is
   supported */
void qcow2_encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
                           uint8_t *out_buf, const uint8_t *in_buf,
                           int nb_sectors, int enc,
                           const AES_KEY *key)
{
    union {
        uint64_t ll[2];
        uint8_t b[16];
    } ivec;
    int i;

    for(i = 0; i < nb_sectors; i++) {
        ivec.ll[0] = cpu_to_le64(sector_num);
        ivec.ll[1] = 0;
        AES_cbc_encrypt(in_buf, out_buf, 512, key,
                        ivec.b, enc);
        sector_num++;
        in_buf += 512;
        out_buf += 512;
    }
}

static int coroutine_fn copy_sectors(BlockDriverState *bs,
                                     uint64_t start_sect,
                                     uint64_t cluster_offset,
                                     int n_start, int n_end)
{
    BDRVQcowState *s = bs->opaque;
    QEMUIOVector qiov;
    struct iovec iov;
    int n, ret;

    n = n_end - n_start;
    if (n <= 0) {
        return 0;
    }

    iov.iov_len = n * BDRV_SECTOR_SIZE;
    iov.iov_base = qemu_blockalign(bs, iov.iov_len);

    qemu_iovec_init_external(&qiov, &iov, 1);

    BLKDBG_EVENT(bs->file, BLKDBG_COW_READ);

    if (!bs->drv) {
        ret = -ENOMEDIUM;
        goto out;
    }

    /* Call .bdrv_co_readv() directly instead of using the public block-layer
     * interface.  This avoids double I/O throttling and request tracking,
     * which can lead to deadlock when block layer copy-on-read is enabled.
     */
    ret = bs->drv->bdrv_co_readv(bs, start_sect + n_start, n, &qiov);
    if (ret < 0) {
        goto out;
    }

    if (s->crypt_method) {
        qcow2_encrypt_sectors(s, start_sect + n_start,
                        iov.iov_base, iov.iov_base, n, 1,
                        &s->aes_encrypt_key);
    }

    ret = qcow2_pre_write_overlap_check(bs, 0,
            cluster_offset + n_start * BDRV_SECTOR_SIZE, n * BDRV_SECTOR_SIZE);
    if (ret < 0) {
        goto out;
    }

    BLKDBG_EVENT(bs->file, BLKDBG_COW_WRITE);
    ret = bdrv_co_writev(bs->file, (cluster_offset >> 9) + n_start, n, &qiov);
    if (ret < 0) {
        goto out;
    }

    ret = 0;
out:
    qemu_vfree(iov.iov_base);
    return ret;
}


/*
 * get_cluster_offset
 *
 * For a given offset of the disk image, find the cluster offset in
 * qcow2 file. The offset is stored in *cluster_offset.
 *
 * on entry, *num is the number of contiguous sectors we'd like to
 * access following offset.
 *
 * on exit, *num is the number of contiguous sectors we can read.
 *
 * Returns the cluster type (QCOW2_CLUSTER_*) on success, -errno in error
 * cases.
 */
int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
    int *num, uint64_t *cluster_offset)
{
    BDRVQcowState *s = bs->opaque;
    unsigned int l2_index;
    uint64_t l1_index, l2_offset, *l2_table;
    int l1_bits, c;
    unsigned int index_in_cluster, nb_clusters;
    uint64_t nb_available, nb_needed;
    int ret;

    index_in_cluster = (offset >> 9) & (s->cluster_sectors - 1);
    nb_needed = *num + index_in_cluster;

    l1_bits = s->l2_bits + s->cluster_bits;

    /* compute how many bytes there are between the offset and
     * the end of the l1 entry
     */

    nb_available = (1ULL << l1_bits) - (offset & ((1ULL << l1_bits) - 1));

    /* compute the number of available sectors */

    nb_available = (nb_available >> 9) + index_in_cluster;

    if (nb_needed > nb_available) {
        nb_needed = nb_available;
    }

    *cluster_offset = 0;

    /* seek the the l2 offset in the l1 table */

    l1_index = offset >> l1_bits;
    if (l1_index >= s->l1_size) {
        ret = QCOW2_CLUSTER_UNALLOCATED;
        goto out;
    }

    l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
    if (!l2_offset) {
        ret = QCOW2_CLUSTER_UNALLOCATED;
        goto out;
    }

    /* load the l2 table in memory */

    ret = l2_load(bs, l2_offset, &l2_table);
    if (ret < 0) {
        return ret;
    }

    /* find the cluster offset for the given disk offset */

    l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
    *cluster_offset = be64_to_cpu(l2_table[l2_index]);
    nb_clusters = size_to_clusters(s, nb_needed << 9);

    ret = qcow2_get_cluster_type(*cluster_offset);
    switch (ret) {
    case QCOW2_CLUSTER_COMPRESSED:
        /* Compressed clusters can only be processed one by one */
        c = 1;
        *cluster_offset &= L2E_COMPRESSED_OFFSET_SIZE_MASK;
        break;
    case QCOW2_CLUSTER_ZERO:
        if (s->qcow_version < 3) {
            qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
            return -EIO;
        }
        c = count_contiguous_clusters(nb_clusters, s->cluster_size,
                &l2_table[l2_index], QCOW_OFLAG_ZERO);
        *cluster_offset = 0;
        break;
    case QCOW2_CLUSTER_UNALLOCATED:
        /* how many empty clusters ? */
        c = count_contiguous_free_clusters(nb_clusters, &l2_table[l2_index]);
        *cluster_offset = 0;
        break;
    case QCOW2_CLUSTER_NORMAL:
        /* how many allocated clusters ? */
        c = count_contiguous_clusters(nb_clusters, s->cluster_size,
                &l2_table[l2_index], QCOW_OFLAG_ZERO);
        *cluster_offset &= L2E_OFFSET_MASK;
        break;
    default:
        abort();
    }

    qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);

    nb_available = (c * s->cluster_sectors);

out:
    if (nb_available > nb_needed)
        nb_available = nb_needed;

    *num = nb_available - index_in_cluster;

    return ret;
}

/*
 * get_cluster_table
 *
 * for a given disk offset, load (and allocate if needed)
 * the l2 table.
 *
 * the l2 table offset in the qcow2 file and the cluster index
 * in the l2 table are given to the caller.
 *
 * Returns 0 on success, -errno in failure case
 */
static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
                             uint64_t **new_l2_table,
                             int *new_l2_index)
{
    BDRVQcowState *s = bs->opaque;
    unsigned int l2_index;
    uint64_t l1_index, l2_offset;
    uint64_t *l2_table = NULL;
    int ret;

    /* seek the the l2 offset in the l1 table */

    l1_index = offset >> (s->l2_bits + s->cluster_bits);
    if (l1_index >= s->l1_size) {
        ret = qcow2_grow_l1_table(bs, l1_index + 1, false);
        if (ret < 0) {
            return ret;
        }
    }

    assert(l1_index < s->l1_size);
    l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;

    /* seek the l2 table of the given l2 offset */

    if (s->l1_table[l1_index] & QCOW_OFLAG_COPIED) {
        /* load the l2 table in memory */
        ret = l2_load(bs, l2_offset, &l2_table);
        if (ret < 0) {
            return ret;
        }
    } else {
        /* First allocate a new L2 table (and do COW if needed) */
        ret = l2_allocate(bs, l1_index, &l2_table);
        if (ret < 0) {
            return ret;
        }

        /* Then decrease the refcount of the old table */
        if (l2_offset) {
            qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
                                QCOW2_DISCARD_OTHER);
        }
    }

    /* find the cluster offset for the given disk offset */

    l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);

    *new_l2_table = l2_table;
    *new_l2_index = l2_index;

    return 0;
}

/*
 * alloc_compressed_cluster_offset
 *
 * For a given offset of the disk image, return cluster offset in
 * qcow2 file.
 *
 * If the offset is not found, allocate a new compressed cluster.
 *
 * Return the cluster offset if successful,
 * Return 0, otherwise.
 *
 */

uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
                                               uint64_t offset,
                                               int compressed_size)
{
    BDRVQcowState *s = bs->opaque;
    int l2_index, ret;
    uint64_t *l2_table;
    int64_t cluster_offset;
    int nb_csectors;

    ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
    if (ret < 0) {
        return 0;
    }

    /* Compression can't overwrite anything. Fail if the cluster was already
     * allocated. */
    cluster_offset = be64_to_cpu(l2_table[l2_index]);
    if (cluster_offset & L2E_OFFSET_MASK) {
        qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
        return 0;
    }

    cluster_offset = qcow2_alloc_bytes(bs, compressed_size);
    if (cluster_offset < 0) {
        qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
        return 0;
    }

    nb_csectors = ((cluster_offset + compressed_size - 1) >> 9) -
                  (cluster_offset >> 9);

    cluster_offset |= QCOW_OFLAG_COMPRESSED |
                      ((uint64_t)nb_csectors << s->csize_shift);

    /* update L2 table */

    /* compressed clusters never have the copied flag */

    BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE_COMPRESSED);
    qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
    l2_table[l2_index] = cpu_to_be64(cluster_offset);
    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    if (ret < 0) {
        return 0;
    }

    return cluster_offset;
}

static int perform_cow(BlockDriverState *bs, QCowL2Meta *m, Qcow2COWRegion *r)
{
    BDRVQcowState *s = bs->opaque;
    int ret;

    if (r->nb_sectors == 0) {
        return 0;
    }

    qemu_co_mutex_unlock(&s->lock);
    ret = copy_sectors(bs, m->offset / BDRV_SECTOR_SIZE, m->alloc_offset,
                       r->offset / BDRV_SECTOR_SIZE,
                       r->offset / BDRV_SECTOR_SIZE + r->nb_sectors);
    qemu_co_mutex_lock(&s->lock);

    if (ret < 0) {
        return ret;
    }

    /*
     * Before we update the L2 table to actually point to the new cluster, we
     * need to be sure that the refcounts have been increased and COW was
     * handled.
     */
    qcow2_cache_depends_on_flush(s->l2_table_cache);

    return 0;
}

int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
{
    BDRVQcowState *s = bs->opaque;
    int i, j = 0, l2_index, ret;
    uint64_t *old_cluster, *l2_table;
    uint64_t cluster_offset = m->alloc_offset;

    trace_qcow2_cluster_link_l2(qemu_coroutine_self(), m->nb_clusters);
    assert(m->nb_clusters > 0);

    old_cluster = g_malloc(m->nb_clusters * sizeof(uint64_t));

    /* copy content of unmodified sectors */
    ret = perform_cow(bs, m, &m->cow_start);
    if (ret < 0) {
        goto err;
    }

    ret = perform_cow(bs, m, &m->cow_end);
    if (ret < 0) {
        goto err;
    }

    /* Update L2 table. */
    if (s->use_lazy_refcounts) {
        qcow2_mark_dirty(bs);
    }
    if (qcow2_need_accurate_refcounts(s)) {
        qcow2_cache_set_dependency(bs, s->l2_table_cache,
                                   s->refcount_block_cache);
    }

    ret = get_cluster_table(bs, m->offset, &l2_table, &l2_index);
    if (ret < 0) {
        goto err;
    }
    qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);

    assert(l2_index + m->nb_clusters <= s->l2_size);
    for (i = 0; i < m->nb_clusters; i++) {
        /* if two concurrent writes happen to the same unallocated cluster
         * each write allocates separate cluster and writes data concurrently.
         * The first one to complete updates l2 table with pointer to its
         * cluster the second one has to do RMW (which is done above by
         * copy_sectors()), update l2 table with its cluster pointer and free
         * old cluster. This is what this loop does */
        if(l2_table[l2_index + i] != 0)
            old_cluster[j++] = l2_table[l2_index + i];

        l2_table[l2_index + i] = cpu_to_be64((cluster_offset +
                    (i << s->cluster_bits)) | QCOW_OFLAG_COPIED);
     }


    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    if (ret < 0) {
        goto err;
    }

    /*
     * If this was a COW, we need to decrease the refcount of the old cluster.
     * Also flush bs->file to get the right order for L2 and refcount update.
     *
     * Don't discard clusters that reach a refcount of 0 (e.g. compressed
     * clusters), the next write will reuse them anyway.
     */
    if (j != 0) {
        for (i = 0; i < j; i++) {
            qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1,
                                    QCOW2_DISCARD_NEVER);
        }
    }

    ret = 0;
err:
    g_free(old_cluster);
    return ret;
 }

/*
 * Returns the number of contiguous clusters that can be used for an allocating
 * write, but require COW to be performed (this includes yet unallocated space,
 * which must copy from the backing file)
 */
static int count_cow_clusters(BDRVQcowState *s, int nb_clusters,
    uint64_t *l2_table, int l2_index)
{
    int i;

    for (i = 0; i < nb_clusters; i++) {
        uint64_t l2_entry = be64_to_cpu(l2_table[l2_index + i]);
        int cluster_type = qcow2_get_cluster_type(l2_entry);

        switch(cluster_type) {
        case QCOW2_CLUSTER_NORMAL:
            if (l2_entry & QCOW_OFLAG_COPIED) {
                goto out;
            }
            break;
        case QCOW2_CLUSTER_UNALLOCATED:
        case QCOW2_CLUSTER_COMPRESSED:
        case QCOW2_CLUSTER_ZERO:
            break;
        default:
            abort();
        }
    }

out:
    assert(i <= nb_clusters);
    return i;
}

/*
 * Check if there already is an AIO write request in flight which allocates
 * the same cluster. In this case we need to wait until the previous
 * request has completed and updated the L2 table accordingly.
 *
 * Returns:
 *   0       if there was no dependency. *cur_bytes indicates the number of
 *           bytes from guest_offset that can be read before the next
 *           dependency must be processed (or the request is complete)
 *
 *   -EAGAIN if we had to wait for another request, previously gathered
 *           information on cluster allocation may be invalid now. The caller
 *           must start over anyway, so consider *cur_bytes undefined.
 */
static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
    uint64_t *cur_bytes, QCowL2Meta **m)
{
    BDRVQcowState *s = bs->opaque;
    QCowL2Meta *old_alloc;
    uint64_t bytes = *cur_bytes;

    QLIST_FOREACH(old_alloc, &s->cluster_allocs, next_in_flight) {

        uint64_t start = guest_offset;
        uint64_t end = start + bytes;
        uint64_t old_start = l2meta_cow_start(old_alloc);
        uint64_t old_end = l2meta_cow_end(old_alloc);

        if (end <= old_start || start >= old_end) {
            /* No intersection */
        } else {
            if (start < old_start) {
                /* Stop at the start of a running allocation */
                bytes = old_start - start;
            } else {
                bytes = 0;
            }

            /* Stop if already an l2meta exists. After yielding, it wouldn't
             * be valid any more, so we'd have to clean up the old L2Metas
             * and deal with requests depending on them before starting to
             * gather new ones. Not worth the trouble. */
            if (bytes == 0 && *m) {
                *cur_bytes = 0;
                return 0;
            }

            if (bytes == 0) {
                /* Wait for the dependency to complete. We need to recheck
                 * the free/allocated clusters when we continue. */
                qemu_co_mutex_unlock(&s->lock);
                qemu_co_queue_wait(&old_alloc->dependent_requests);
                qemu_co_mutex_lock(&s->lock);
                return -EAGAIN;
            }
        }
    }

    /* Make sure that existing clusters and new allocations are only used up to
     * the next dependency if we shortened the request above */
    *cur_bytes = bytes;

    return 0;
}

/*
 * Checks how many already allocated clusters that don't require a copy on
 * write there are at the given guest_offset (up to *bytes). If
 * *host_offset is not zero, only physically contiguous clusters beginning at
 * this host offset are counted.
 *
 * Note that guest_offset may not be cluster aligned. In this case, the
 * returned *host_offset points to exact byte referenced by guest_offset and
 * therefore isn't cluster aligned as well.
 *
 * Returns:
 *   0:     if no allocated clusters are available at the given offset.
 *          *bytes is normally unchanged. It is set to 0 if the cluster
 *          is allocated and doesn't need COW, but doesn't have the right
 *          physical offset.
 *
 *   1:     if allocated clusters that don't require a COW are available at
 *          the requested offset. *bytes may have decreased and describes
 *          the length of the area that can be written to.
 *
 *  -errno: in error cases
 */
static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
    uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
{
    BDRVQcowState *s = bs->opaque;
    int l2_index;
    uint64_t cluster_offset;
    uint64_t *l2_table;
    unsigned int nb_clusters;
    unsigned int keep_clusters;
    int ret, pret;

    trace_qcow2_handle_copied(qemu_coroutine_self(), guest_offset, *host_offset,
                              *bytes);

    assert(*host_offset == 0 ||    offset_into_cluster(s, guest_offset)
                                == offset_into_cluster(s, *host_offset));

    /*
     * Calculate the number of clusters to look for. We stop at L2 table
     * boundaries to keep things simple.
     */
    nb_clusters =
        size_to_clusters(s, offset_into_cluster(s, guest_offset) + *bytes);

    l2_index = offset_to_l2_index(s, guest_offset);
    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);

    /* Find L2 entry for the first involved cluster */
    ret = get_cluster_table(bs, guest_offset, &l2_table, &l2_index);
    if (ret < 0) {
        return ret;
    }

    cluster_offset = be64_to_cpu(l2_table[l2_index]);

    /* Check how many clusters are already allocated and don't need COW */
    if (qcow2_get_cluster_type(cluster_offset) == QCOW2_CLUSTER_NORMAL
        && (cluster_offset & QCOW_OFLAG_COPIED))
    {
        /* If a specific host_offset is required, check it */
        bool offset_matches =
            (cluster_offset & L2E_OFFSET_MASK) == *host_offset;

        if (*host_offset != 0 && !offset_matches) {
            *bytes = 0;
            ret = 0;
            goto out;
        }

        /* We keep all QCOW_OFLAG_COPIED clusters */
        keep_clusters =
            count_contiguous_clusters(nb_clusters, s->cluster_size,
                                      &l2_table[l2_index],
                                      QCOW_OFLAG_COPIED | QCOW_OFLAG_ZERO);
        assert(keep_clusters <= nb_clusters);

        *bytes = MIN(*bytes,
                 keep_clusters * s->cluster_size
                 - offset_into_cluster(s, guest_offset));

        ret = 1;
    } else {
        ret = 0;
    }

    /* Cleanup */
out:
    pret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    if (pret < 0) {
        return pret;
    }

    /* Only return a host offset if we actually made progress. Otherwise we
     * would make requirements for handle_alloc() that it can't fulfill */
    if (ret) {
        *host_offset = (cluster_offset & L2E_OFFSET_MASK)
                     + offset_into_cluster(s, guest_offset);
    }

    return ret;
}

/*
 * Allocates new clusters for the given guest_offset.
 *
 * At most *nb_clusters are allocated, and on return *nb_clusters is updated to
 * contain the number of clusters that have been allocated and are contiguous
 * in the image file.
 *
 * If *host_offset is non-zero, it specifies the offset in the image file at
 * which the new clusters must start. *nb_clusters can be 0 on return in this
 * case if the cluster at host_offset is already in use. If *host_offset is
 * zero, the clusters can be allocated anywhere in the image file.
 *
 * *host_offset is updated to contain the offset into the image file at which
 * the first allocated cluster starts.
 *
 * Return 0 on success and -errno in error cases. -EAGAIN means that the
 * function has been waiting for another request and the allocation must be
 * restarted, but the whole request should not be failed.
 */
static int do_alloc_cluster_offset(BlockDriverState *bs, uint64_t guest_offset,
    uint64_t *host_offset, unsigned int *nb_clusters)
{
    BDRVQcowState *s = bs->opaque;

    trace_qcow2_do_alloc_clusters_offset(qemu_coroutine_self(), guest_offset,
                                         *host_offset, *nb_clusters);

    /* Allocate new clusters */
    trace_qcow2_cluster_alloc_phys(qemu_coroutine_self());
    if (*host_offset == 0) {
        int64_t cluster_offset =
            qcow2_alloc_clusters(bs, *nb_clusters * s->cluster_size);
        if (cluster_offset < 0) {
            return cluster_offset;
        }
        *host_offset = cluster_offset;
        return 0;
    } else {
        int ret = qcow2_alloc_clusters_at(bs, *host_offset, *nb_clusters);
        if (ret < 0) {
            return ret;
        }
        *nb_clusters = ret;
        return 0;
    }
}

/*
 * Allocates new clusters for an area that either is yet unallocated or needs a
 * copy on write. If *host_offset is non-zero, clusters are only allocated if
 * the new allocation can match the specified host offset.
 *
 * Note that guest_offset may not be cluster aligned. In this case, the
 * returned *host_offset points to exact byte referenced by guest_offset and
 * therefore isn't cluster aligned as well.
 *
 * Returns:
 *   0:     if no clusters could be allocated. *bytes is set to 0,
 *          *host_offset is left unchanged.
 *
 *   1:     if new clusters were allocated. *bytes may be decreased if the
 *          new allocation doesn't cover all of the requested area.
 *          *host_offset is updated to contain the host offset of the first
 *          newly allocated cluster.
 *
 *  -errno: in error cases
 */
static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
    uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
{
    BDRVQcowState *s = bs->opaque;
    int l2_index;
    uint64_t *l2_table;
    uint64_t entry;
    unsigned int nb_clusters;
    int ret;

    uint64_t alloc_cluster_offset;

    trace_qcow2_handle_alloc(qemu_coroutine_self(), guest_offset, *host_offset,
                             *bytes);
    assert(*bytes > 0);

    /*
     * Calculate the number of clusters to look for. We stop at L2 table
     * boundaries to keep things simple.
     */
    nb_clusters =
        size_to_clusters(s, offset_into_cluster(s, guest_offset) + *bytes);

    l2_index = offset_to_l2_index(s, guest_offset);
    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);

    /* Find L2 entry for the first involved cluster */
    ret = get_cluster_table(bs, guest_offset, &l2_table, &l2_index);
    if (ret < 0) {
        return ret;
    }

    entry = be64_to_cpu(l2_table[l2_index]);

    /* For the moment, overwrite compressed clusters one by one */
    if (entry & QCOW_OFLAG_COMPRESSED) {
        nb_clusters = 1;
    } else {
        nb_clusters = count_cow_clusters(s, nb_clusters, l2_table, l2_index);
    }

    /* This function is only called when there were no non-COW clusters, so if
     * we can't find any unallocated or COW clusters either, something is
     * wrong with our code. */
    assert(nb_clusters > 0);

    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    if (ret < 0) {
        return ret;
    }

    /* Allocate, if necessary at a given offset in the image file */
    alloc_cluster_offset = start_of_cluster(s, *host_offset);
    ret = do_alloc_cluster_offset(bs, guest_offset, &alloc_cluster_offset,
                                  &nb_clusters);
    if (ret < 0) {
        goto fail;
    }

    /* Can't extend contiguous allocation */
    if (nb_clusters == 0) {
        *bytes = 0;
        return 0;
    }

    /*
     * Save info needed for meta data update.
     *
     * requested_sectors: Number of sectors from the start of the first
     * newly allocated cluster to the end of the (possibly shortened
     * before) write request.
     *
     * avail_sectors: Number of sectors from the start of the first
     * newly allocated to the end of the last newly allocated cluster.
     *
     * nb_sectors: The number of sectors from the start of the first
     * newly allocated cluster to the end of the area that the write
     * request actually writes to (excluding COW at the end)
     */
    int requested_sectors =
        (*bytes + offset_into_cluster(s, guest_offset))
        >> BDRV_SECTOR_BITS;
    int avail_sectors = nb_clusters
                        << (s->cluster_bits - BDRV_SECTOR_BITS);
    int alloc_n_start = offset_into_cluster(s, guest_offset)
                        >> BDRV_SECTOR_BITS;
    int nb_sectors = MIN(requested_sectors, avail_sectors);
    QCowL2Meta *old_m = *m;

    *m = g_malloc0(sizeof(**m));

    **m = (QCowL2Meta) {
        .next           = old_m,

        .alloc_offset   = alloc_cluster_offset,
        .offset         = start_of_cluster(s, guest_offset),
        .nb_clusters    = nb_clusters,
        .nb_available   = nb_sectors,

        .cow_start = {
            .offset     = 0,
            .nb_sectors = alloc_n_start,
        },
        .cow_end = {
            .offset     = nb_sectors * BDRV_SECTOR_SIZE,
            .nb_sectors = avail_sectors - nb_sectors,
        },
    };
    qemu_co_queue_init(&(*m)->dependent_requests);
    QLIST_INSERT_HEAD(&s->cluster_allocs, *m, next_in_flight);

    *host_offset = alloc_cluster_offset + offset_into_cluster(s, guest_offset);
    *bytes = MIN(*bytes, (nb_sectors * BDRV_SECTOR_SIZE)
                         - offset_into_cluster(s, guest_offset));
    assert(*bytes != 0);

    return 1;

fail:
    if (*m && (*m)->nb_clusters > 0) {
        QLIST_REMOVE(*m, next_in_flight);
    }
    return ret;
}

/*
 * alloc_cluster_offset
 *
 * For a given offset on the virtual disk, find the cluster offset in qcow2
 * file. If the offset is not found, allocate a new cluster.
 *
 * If the cluster was already allocated, m->nb_clusters is set to 0 and
 * other fields in m are meaningless.
 *
 * If the cluster is newly allocated, m->nb_clusters is set to the number of
 * contiguous clusters that have been allocated. In this case, the other
 * fields of m are valid and contain information about the first allocated
 * cluster.
 *
 * If the request conflicts with another write request in flight, the coroutine
 * is queued and will be reentered when the dependency has completed.
 *
 * Return 0 on success and -errno in error cases
 */
int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
    int *num, uint64_t *host_offset, QCowL2Meta **m)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t start, remaining;
    uint64_t cluster_offset;
    uint64_t cur_bytes;
    int ret;

    trace_qcow2_alloc_clusters_offset(qemu_coroutine_self(), offset, *num);

    assert((offset & ~BDRV_SECTOR_MASK) == 0);

again:
    start = offset;
    remaining = *num << BDRV_SECTOR_BITS;
    cluster_offset = 0;
    *host_offset = 0;
    cur_bytes = 0;
    *m = NULL;

    while (true) {

        if (!*host_offset) {
            *host_offset = start_of_cluster(s, cluster_offset);
        }

        assert(remaining >= cur_bytes);

        start           += cur_bytes;
        remaining       -= cur_bytes;
        cluster_offset  += cur_bytes;

        if (remaining == 0) {
            break;
        }

        cur_bytes = remaining;

        /*
         * Now start gathering as many contiguous clusters as possible:
         *
         * 1. Check for overlaps with in-flight allocations
         *
         *      a) Overlap not in the first cluster -> shorten this request and
         *         let the caller handle the rest in its next loop iteration.
         *
         *      b) Real overlaps of two requests. Yield and restart the search
         *         for contiguous clusters (the situation could have changed
         *         while we were sleeping)
         *
         *      c) TODO: Request starts in the same cluster as the in-flight
         *         allocation ends. Shorten the COW of the in-fight allocation,
         *         set cluster_offset to write to the same cluster and set up
         *         the right synchronisation between the in-flight request and
         *         the new one.
         */
        ret = handle_dependencies(bs, start, &cur_bytes, m);
        if (ret == -EAGAIN) {
            /* Currently handle_dependencies() doesn't yield if we already had
             * an allocation. If it did, we would have to clean up the L2Meta
             * structs before starting over. */
            assert(*m == NULL);
            goto again;
        } else if (ret < 0) {
            return ret;
        } else if (cur_bytes == 0) {
            break;
        } else {
            /* handle_dependencies() may have decreased cur_bytes (shortened
             * the allocations below) so that the next dependency is processed
             * correctly during the next loop iteration. */
        }

        /*
         * 2. Count contiguous COPIED clusters.
         */
        ret = handle_copied(bs, start, &cluster_offset, &cur_bytes, m);
        if (ret < 0) {
            return ret;
        } else if (ret) {
            continue;
        } else if (cur_bytes == 0) {
            break;
        }

        /*
         * 3. If the request still hasn't completed, allocate new clusters,
         *    considering any cluster_offset of steps 1c or 2.
         */
        ret = handle_alloc(bs, start, &cluster_offset, &cur_bytes, m);
        if (ret < 0) {
            return ret;
        } else if (ret) {
            continue;
        } else {
            assert(cur_bytes == 0);
            break;
        }
    }

    *num -= remaining >> BDRV_SECTOR_BITS;
    assert(*num > 0);
    assert(*host_offset != 0);

    return 0;
}

static int decompress_buffer(uint8_t *out_buf, int out_buf_size,
                             const uint8_t *buf, int buf_size)
{
    z_stream strm1, *strm = &strm1;
    int ret, out_len;

    memset(strm, 0, sizeof(*strm));

    strm->next_in = (uint8_t *)buf;
    strm->avail_in = buf_size;
    strm->next_out = out_buf;
    strm->avail_out = out_buf_size;

    ret = inflateInit2(strm, -12);
    if (ret != Z_OK)
        return -1;
    ret = inflate(strm, Z_FINISH);
    out_len = strm->next_out - out_buf;
    if ((ret != Z_STREAM_END && ret != Z_BUF_ERROR) ||
        out_len != out_buf_size) {
        inflateEnd(strm);
        return -1;
    }
    inflateEnd(strm);
    return 0;
}

int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
{
    BDRVQcowState *s = bs->opaque;
    int ret, csize, nb_csectors, sector_offset;
    uint64_t coffset;

    coffset = cluster_offset & s->cluster_offset_mask;
    if (s->cluster_cache_offset != coffset) {
        nb_csectors = ((cluster_offset >> s->csize_shift) & s->csize_mask) + 1;
        sector_offset = coffset & 511;
        csize = nb_csectors * 512 - sector_offset;
        BLKDBG_EVENT(bs->file, BLKDBG_READ_COMPRESSED);
        ret = bdrv_read(bs->file, coffset >> 9, s->cluster_data, nb_csectors);
        if (ret < 0) {
            return ret;
        }
        if (decompress_buffer(s->cluster_cache, s->cluster_size,
                              s->cluster_data + sector_offset, csize) < 0) {
            return -EIO;
        }
        s->cluster_cache_offset = coffset;
    }
    return 0;
}

/*
 * This discards as many clusters of nb_clusters as possible at once (i.e.
 * all clusters in the same L2 table) and returns the number of discarded
 * clusters.
 */
static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
    unsigned int nb_clusters, enum qcow2_discard_type type)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l2_table;
    int l2_index;
    int ret;
    int i;

    ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
    if (ret < 0) {
        return ret;
    }

    /* Limit nb_clusters to one L2 table */
    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);

    for (i = 0; i < nb_clusters; i++) {
        uint64_t old_l2_entry;

        old_l2_entry = be64_to_cpu(l2_table[l2_index + i]);

        /*
         * Make sure that a discarded area reads back as zeroes for v3 images
         * (we cannot do it for v2 without actually writing a zero-filled
         * buffer). We can skip the operation if the cluster is already marked
         * as zero, or if it's unallocated and we don't have a backing file.
         *
         * TODO We might want to use bdrv_get_block_status(bs) here, but we're
         * holding s->lock, so that doesn't work today.
         */
        switch (qcow2_get_cluster_type(old_l2_entry)) {
            case QCOW2_CLUSTER_UNALLOCATED:
                if (!bs->backing_hd) {
                    continue;
                }
                break;

            case QCOW2_CLUSTER_ZERO:
                continue;

            case QCOW2_CLUSTER_NORMAL:
            case QCOW2_CLUSTER_COMPRESSED:
                break;

            default:
                abort();
        }

        /* First remove L2 entries */
        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
        if (s->qcow_version >= 3) {
            l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
        } else {
            l2_table[l2_index + i] = cpu_to_be64(0);
        }

        /* Then decrease the refcount */
        qcow2_free_any_clusters(bs, old_l2_entry, 1, type);
    }

    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    if (ret < 0) {
        return ret;
    }

    return nb_clusters;
}

int qcow2_discard_clusters(BlockDriverState *bs, uint64_t offset,
    int nb_sectors, enum qcow2_discard_type type)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t end_offset;
    unsigned int nb_clusters;
    int ret;

    end_offset = offset + (nb_sectors << BDRV_SECTOR_BITS);

    /* Round start up and end down */
    offset = align_offset(offset, s->cluster_size);
    end_offset = start_of_cluster(s, end_offset);

    if (offset > end_offset) {
        return 0;
    }

    nb_clusters = size_to_clusters(s, end_offset - offset);

    s->cache_discards = true;

    /* Each L2 table is handled by its own loop iteration */
    while (nb_clusters > 0) {
        ret = discard_single_l2(bs, offset, nb_clusters, type);
        if (ret < 0) {
            goto fail;
        }

        nb_clusters -= ret;
        offset += (ret * s->cluster_size);
    }

    ret = 0;
fail:
    s->cache_discards = false;
    qcow2_process_discards(bs, ret);

    return ret;
}

/*
 * This zeroes as many clusters of nb_clusters as possible at once (i.e.
 * all clusters in the same L2 table) and returns the number of zeroed
 * clusters.
 */
static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
    unsigned int nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l2_table;
    int l2_index;
    int ret;
    int i;

    ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
    if (ret < 0) {
        return ret;
    }

    /* Limit nb_clusters to one L2 table */
    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);

    for (i = 0; i < nb_clusters; i++) {
        uint64_t old_offset;

        old_offset = be64_to_cpu(l2_table[l2_index + i]);

        /* Update L2 entries */
        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
        if (old_offset & QCOW_OFLAG_COMPRESSED) {
            l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
            qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
        } else {
            l2_table[l2_index + i] |= cpu_to_be64(QCOW_OFLAG_ZERO);
        }
    }

    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    if (ret < 0) {
        return ret;
    }

    return nb_clusters;
}

int qcow2_zero_clusters(BlockDriverState *bs, uint64_t offset, int nb_sectors)
{
    BDRVQcowState *s = bs->opaque;
    unsigned int nb_clusters;
    int ret;

    /* The zero flag is only supported by version 3 and newer */
    if (s->qcow_version < 3) {
        return -ENOTSUP;
    }

    /* Each L2 table is handled by its own loop iteration */
    nb_clusters = size_to_clusters(s, nb_sectors << BDRV_SECTOR_BITS);

    s->cache_discards = true;

    while (nb_clusters > 0) {
        ret = zero_single_l2(bs, offset, nb_clusters);
        if (ret < 0) {
            goto fail;
        }

        nb_clusters -= ret;
        offset += (ret * s->cluster_size);
    }

    ret = 0;
fail:
    s->cache_discards = false;
    qcow2_process_discards(bs, ret);

    return ret;
}

/*
 * Expands all zero clusters in a specific L1 table (or deallocates them, for
 * non-backed non-pre-allocated zero clusters).
 *
 * expanded_clusters is a bitmap where every bit corresponds to one cluster in
 * the image file; a bit gets set if the corresponding cluster has been used for
 * zero expansion (i.e., has been filled with zeroes and is referenced from an
 * L2 table). nb_clusters contains the total cluster count of the image file,
 * i.e., the number of bits in expanded_clusters.
 */
static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                                      int l1_size, uint8_t **expanded_clusters,
                                      uint64_t *nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    bool is_active_l1 = (l1_table == s->l1_table);
    uint64_t *l2_table = NULL;
    int ret;
    int i, j;

    if (!is_active_l1) {
        /* inactive L2 tables require a buffer to be stored in when loading
         * them from disk */
        l2_table = qemu_blockalign(bs, s->cluster_size);
    }

    for (i = 0; i < l1_size; i++) {
        uint64_t l2_offset = l1_table[i] & L1E_OFFSET_MASK;
        bool l2_dirty = false;

        if (!l2_offset) {
            /* unallocated */
            continue;
        }

        if (is_active_l1) {
            /* get active L2 tables from cache */
            ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset,
                    (void **)&l2_table);
        } else {
            /* load inactive L2 tables from disk */
            ret = bdrv_read(bs->file, l2_offset / BDRV_SECTOR_SIZE,
                    (void *)l2_table, s->cluster_sectors);
        }
        if (ret < 0) {
            goto fail;
        }

        for (j = 0; j < s->l2_size; j++) {
            uint64_t l2_entry = be64_to_cpu(l2_table[j]);
            int64_t offset = l2_entry & L2E_OFFSET_MASK, cluster_index;
            int cluster_type = qcow2_get_cluster_type(l2_entry);
            bool preallocated = offset != 0;

            if (cluster_type == QCOW2_CLUSTER_NORMAL) {
                cluster_index = offset >> s->cluster_bits;
                assert((cluster_index >= 0) && (cluster_index < *nb_clusters));
                if ((*expanded_clusters)[cluster_index / 8] &
                    (1 << (cluster_index % 8))) {
                    /* Probably a shared L2 table; this cluster was a zero
                     * cluster which has been expanded, its refcount
                     * therefore most likely requires an update. */
                    ret = qcow2_update_cluster_refcount(bs, cluster_index, 1,
                                                        QCOW2_DISCARD_NEVER);
                    if (ret < 0) {
                        goto fail;
                    }
                    /* Since we just increased the refcount, the COPIED flag may
                     * no longer be set. */
                    l2_table[j] = cpu_to_be64(l2_entry & ~QCOW_OFLAG_COPIED);
                    l2_dirty = true;
                }
                continue;
            }
            else if (qcow2_get_cluster_type(l2_entry) != QCOW2_CLUSTER_ZERO) {
                continue;
            }

            if (!preallocated) {
                if (!bs->backing_hd) {
                    /* not backed; therefore we can simply deallocate the
                     * cluster */
                    l2_table[j] = 0;
                    l2_dirty = true;
                    continue;
                }

                offset = qcow2_alloc_clusters(bs, s->cluster_size);
                if (offset < 0) {
                    ret = offset;
                    goto fail;
                }
            }

            ret = qcow2_pre_write_overlap_check(bs, 0, offset, s->cluster_size);
            if (ret < 0) {
                if (!preallocated) {
                    qcow2_free_clusters(bs, offset, s->cluster_size,
                                        QCOW2_DISCARD_ALWAYS);
                }
                goto fail;
            }

            ret = bdrv_write_zeroes(bs->file, offset / BDRV_SECTOR_SIZE,
                                    s->cluster_sectors, 0);
            if (ret < 0) {
                if (!preallocated) {
                    qcow2_free_clusters(bs, offset, s->cluster_size,
                                        QCOW2_DISCARD_ALWAYS);
                }
                goto fail;
            }

            l2_table[j] = cpu_to_be64(offset | QCOW_OFLAG_COPIED);
            l2_dirty = true;

            cluster_index = offset >> s->cluster_bits;

            if (cluster_index >= *nb_clusters) {
                uint64_t old_bitmap_size = (*nb_clusters + 7) / 8;
                uint64_t new_bitmap_size;
                /* The offset may lie beyond the old end of the underlying image
                 * file for growable files only */
                assert(bs->file->growable);
                *nb_clusters = size_to_clusters(s, bs->file->total_sectors *
                                                BDRV_SECTOR_SIZE);
                new_bitmap_size = (*nb_clusters + 7) / 8;
                *expanded_clusters = g_realloc(*expanded_clusters,
                                               new_bitmap_size);
                /* clear the newly allocated space */
                memset(&(*expanded_clusters)[old_bitmap_size], 0,
                       new_bitmap_size - old_bitmap_size);
            }

            assert((cluster_index >= 0) && (cluster_index < *nb_clusters));
            (*expanded_clusters)[cluster_index / 8] |= 1 << (cluster_index % 8);
        }

        if (is_active_l1) {
            if (l2_dirty) {
                qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
                qcow2_cache_depends_on_flush(s->l2_table_cache);
            }
            ret = qcow2_cache_put(bs, s->l2_table_cache, (void **)&l2_table);
            if (ret < 0) {
                l2_table = NULL;
                goto fail;
            }
        } else {
            if (l2_dirty) {
                ret = qcow2_pre_write_overlap_check(bs,
                        QCOW2_OL_INACTIVE_L2 | QCOW2_OL_ACTIVE_L2, l2_offset,
                        s->cluster_size);
                if (ret < 0) {
                    goto fail;
                }

                ret = bdrv_write(bs->file, l2_offset / BDRV_SECTOR_SIZE,
                        (void *)l2_table, s->cluster_sectors);
                if (ret < 0) {
                    goto fail;
                }
            }
        }
    }

    ret = 0;

fail:
    if (l2_table) {
        if (!is_active_l1) {
            qemu_vfree(l2_table);
        } else {
            if (ret < 0) {
                qcow2_cache_put(bs, s->l2_table_cache, (void **)&l2_table);
            } else {
                ret = qcow2_cache_put(bs, s->l2_table_cache,
                        (void **)&l2_table);
            }
        }
    }
    return ret;
}

/*
 * For backed images, expands all zero clusters on the image. For non-backed
 * images, deallocates all non-pre-allocated zero clusters (and claims the
 * allocation for pre-allocated ones). This is important for downgrading to a
 * qcow2 version which doesn't yet support metadata zero clusters.
 */
int qcow2_expand_zero_clusters(BlockDriverState *bs)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l1_table = NULL;
    uint64_t nb_clusters;
    uint8_t *expanded_clusters;
    int ret;
    int i, j;

    nb_clusters = size_to_clusters(s, bs->file->total_sectors *
                                   BDRV_SECTOR_SIZE);
    expanded_clusters = g_malloc0((nb_clusters + 7) / 8);

    ret = expand_zero_clusters_in_l1(bs, s->l1_table, s->l1_size,
                                     &expanded_clusters, &nb_clusters);
    if (ret < 0) {
        goto fail;
    }

    /* Inactive L1 tables may point to active L2 tables - therefore it is
     * necessary to flush the L2 table cache before trying to access the L2
     * tables pointed to by inactive L1 entries (else we might try to expand
     * zero clusters that have already been expanded); furthermore, it is also
     * necessary to empty the L2 table cache, since it may contain tables which
     * are now going to be modified directly on disk, bypassing the cache.
     * qcow2_cache_empty() does both for us. */
    ret = qcow2_cache_empty(bs, s->l2_table_cache);
    if (ret < 0) {
        goto fail;
    }

    for (i = 0; i < s->nb_snapshots; i++) {
        int l1_sectors = (s->snapshots[i].l1_size * sizeof(uint64_t) +
                BDRV_SECTOR_SIZE - 1) / BDRV_SECTOR_SIZE;

        l1_table = g_realloc(l1_table, l1_sectors * BDRV_SECTOR_SIZE);

        ret = bdrv_read(bs->file, s->snapshots[i].l1_table_offset /
                BDRV_SECTOR_SIZE, (void *)l1_table, l1_sectors);
        if (ret < 0) {
            goto fail;
        }

        for (j = 0; j < s->snapshots[i].l1_size; j++) {
            be64_to_cpus(&l1_table[j]);
        }

        ret = expand_zero_clusters_in_l1(bs, l1_table, s->snapshots[i].l1_size,
                                         &expanded_clusters, &nb_clusters);
        if (ret < 0) {
            goto fail;
        }
    }

    ret = 0;

fail:
    g_free(expanded_clusters);
    g_free(l1_table);
    return ret;
}