/*

 * 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 "qemu-common.h"

#include "block_int.h"

#include <zlib.h>

#include "aes.h"

#include <assert.h>

/*

  Differences with QCOW:

  - Support for multiple incremental snapshots.

  - Memory management by reference counts.

  - Clusters which have a reference count of one have the bit

    QCOW_OFLAG_COPIED to optimize write performance.

  - Size of compressed clusters is stored in sectors to reduce bit usage

    in the cluster offsets.

  - Support for storing additional data (such as the VM state) in the

    snapshots.

  - If a backing store is used, the cluster size is not constrained

    (could be backported to QCOW).

  - L2 tables have always a size of one cluster.

*/

//#define DEBUG_ALLOC

//#define DEBUG_ALLOC2

#define QCOW_MAGIC (('Q' << 24) | ('F' << 16) | ('I' << 8) | 0xfb)

#define QCOW_VERSION 2

#define QCOW_CRYPT_NONE 0

#define QCOW_CRYPT_AES  1

#define QCOW_MAX_CRYPT_CLUSTERS 32

/* indicate that the refcount of the referenced cluster is exactly one. */

#define QCOW_OFLAG_COPIED     (1LL << 63)

/* indicate that the cluster is compressed (they never have the copied flag) */

#define QCOW_OFLAG_COMPRESSED (1LL << 62)

#define REFCOUNT_SHIFT 1 /* refcount size is 2 bytes */

typedef struct QCowHeader {

    uint32_t magic;

    uint32_t version;

    uint64_t backing_file_offset;

    uint32_t backing_file_size;

    uint32_t cluster_bits;

    uint64_t size; /* in bytes */

    uint32_t crypt_method;

    uint32_t l1_size; /* XXX: save number of clusters instead ? */

    uint64_t l1_table_offset;

    uint64_t refcount_table_offset;

    uint32_t refcount_table_clusters;

    uint32_t nb_snapshots;

    uint64_t snapshots_offset;

} QCowHeader;

typedef struct __attribute__((packed)) QCowSnapshotHeader {

    /* header is 8 byte aligned */

    uint64_t l1_table_offset;

    uint32_t l1_size;

    uint16_t id_str_size;

    uint16_t name_size;

    uint32_t date_sec;

    uint32_t date_nsec;

    uint64_t vm_clock_nsec;

    uint32_t vm_state_size;

    uint32_t extra_data_size; /* for extension */

    /* extra data follows */

    /* id_str follows */

    /* name follows  */

} QCowSnapshotHeader;

#define L2_CACHE_SIZE 16

typedef struct QCowSnapshot {

    uint64_t l1_table_offset;

    uint32_t l1_size;

    char *id_str;

    char *name;

    uint32_t vm_state_size;

    uint32_t date_sec;

    uint32_t date_nsec;

    uint64_t vm_clock_nsec;

} QCowSnapshot;

typedef struct BDRVQcowState {

    BlockDriverState *hd;

    int cluster_bits;

    int cluster_size;

    int cluster_sectors;

    int l2_bits;

    int l2_size;

    int l1_size;

    int l1_vm_state_index;

    int csize_shift;

    int csize_mask;

    uint64_t cluster_offset_mask;

    uint64_t l1_table_offset;

    uint64_t *l1_table;

    uint64_t *l2_cache;

    uint64_t l2_cache_offsets[L2_CACHE_SIZE];

    uint32_t l2_cache_counts[L2_CACHE_SIZE];

    uint8_t *cluster_cache;

    uint8_t *cluster_data;

    uint64_t cluster_cache_offset;

    uint64_t *refcount_table;

    uint64_t refcount_table_offset;

    uint32_t refcount_table_size;

    uint64_t refcount_block_cache_offset;

    uint16_t *refcount_block_cache;

    int64_t free_cluster_index;

    int64_t free_byte_offset;

    uint32_t crypt_method; /* current crypt method, 0 if no key yet */

    uint32_t crypt_method_header;

    AES_KEY aes_encrypt_key;

    AES_KEY aes_decrypt_key;

    uint64_t snapshots_offset;

    int snapshots_size;

    int nb_snapshots;

    QCowSnapshot *snapshots;

} BDRVQcowState;

static int decompress_cluster(BDRVQcowState *s, uint64_t cluster_offset);

static int qcow_read(BlockDriverState *bs, int64_t sector_num,

                     uint8_t *buf, int nb_sectors);

static int qcow_read_snapshots(BlockDriverState *bs);

static void qcow_free_snapshots(BlockDriverState *bs);

static int refcount_init(BlockDriverState *bs);

static void refcount_close(BlockDriverState *bs);

static int get_refcount(BlockDriverState *bs, int64_t cluster_index);

static int update_cluster_refcount(BlockDriverState *bs,

                                   int64_t cluster_index,

                                   int addend);

static void update_refcount(BlockDriverState *bs,

                            int64_t offset, int64_t length,

                            int addend);

static int64_t alloc_clusters(BlockDriverState *bs, int64_t size);

static int64_t alloc_bytes(BlockDriverState *bs, int size);

static void free_clusters(BlockDriverState *bs,

                          int64_t offset, int64_t size);

#ifdef DEBUG_ALLOC

static void check_refcounts(BlockDriverState *bs);

#endif

static int qcow_probe(const uint8_t *buf, int buf_size, const char *filename)

{

    const QCowHeader *cow_header = (const void *)buf;

    if (buf_size >= sizeof(QCowHeader) &&

        be32_to_cpu(cow_header->magic) == QCOW_MAGIC &&

        be32_to_cpu(cow_header->version) == QCOW_VERSION)

        return 100;

    else

        return 0;

}

static int qcow_open(BlockDriverState *bs, const char *filename, int flags)

{

    BDRVQcowState *s = bs->opaque;

    int len, i, shift, ret;

    QCowHeader header;

    /* Performance is terrible right now with cache=writethrough due mainly

     * to reference count updates.  If the user does not explicitly specify

     * a caching type, force to writeback caching.

     */

    if ((flags & BDRV_O_CACHE_DEF)) {

        flags |= BDRV_O_CACHE_WB;

        flags &= ~BDRV_O_CACHE_DEF;

    }

    ret = bdrv_file_open(&s->hd, filename, flags);

    if (ret < 0)

        return ret;

    if (bdrv_pread(s->hd, 0, &header, sizeof(header)) != sizeof(header))

        goto fail;

    be32_to_cpus(&header.magic);

    be32_to_cpus(&header.version);

    be64_to_cpus(&header.backing_file_offset);

    be32_to_cpus(&header.backing_file_size);

    be64_to_cpus(&header.size);

    be32_to_cpus(&header.cluster_bits);

    be32_to_cpus(&header.crypt_method);

    be64_to_cpus(&header.l1_table_offset);

    be32_to_cpus(&header.l1_size);

    be64_to_cpus(&header.refcount_table_offset);

    be32_to_cpus(&header.refcount_table_clusters);

    be64_to_cpus(&header.snapshots_offset);

    be32_to_cpus(&header.nb_snapshots);

    if (header.magic != QCOW_MAGIC || header.version != QCOW_VERSION)

        goto fail;

    if (header.size <= 1 ||

        header.cluster_bits < 9 ||

        header.cluster_bits > 16)

        goto fail;

    if (header.crypt_method > QCOW_CRYPT_AES)

        goto fail;

    s->crypt_method_header = header.crypt_method;

    if (s->crypt_method_header)

        bs->encrypted = 1;

    s->cluster_bits = header.cluster_bits;

    s->cluster_size = 1 << s->cluster_bits;

    s->cluster_sectors = 1 << (s->cluster_bits - 9);

    s->l2_bits = s->cluster_bits - 3; /* L2 is always one cluster */

    s->l2_size = 1 << s->l2_bits;

    bs->total_sectors = header.size / 512;

    s->csize_shift = (62 - (s->cluster_bits - 8));

    s->csize_mask = (1 << (s->cluster_bits - 8)) - 1;

    s->cluster_offset_mask = (1LL << s->csize_shift) - 1;

    s->refcount_table_offset = header.refcount_table_offset;

    s->refcount_table_size =

        header.refcount_table_clusters << (s->cluster_bits - 3);

    s->snapshots_offset = header.snapshots_offset;

    s->nb_snapshots = header.nb_snapshots;

    /* read the level 1 table */

    s->l1_size = header.l1_size;

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

    s->l1_vm_state_index = (header.size + (1LL << shift) - 1) >> shift;

    /* the L1 table must contain at least enough entries to put

       header.size bytes */

    if (s->l1_size < s->l1_vm_state_index)

        goto fail;

    s->l1_table_offset = header.l1_table_offset;

    s->l1_table = qemu_malloc(s->l1_size * sizeof(uint64_t));

    if (!s->l1_table)

        goto fail;

    if (bdrv_pread(s->hd, s->l1_table_offset, s->l1_table, s->l1_size * sizeof(uint64_t)) !=

        s->l1_size * sizeof(uint64_t))

        goto fail;

    for(i = 0;i < s->l1_size; i++) {

        be64_to_cpus(&s->l1_table[i]);

    }

    /* alloc L2 cache */

    s->l2_cache = qemu_malloc(s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));

    if (!s->l2_cache)

        goto fail;

    s->cluster_cache = qemu_malloc(s->cluster_size);

    if (!s->cluster_cache)

        goto fail;

    /* one more sector for decompressed data alignment */

    s->cluster_data = qemu_malloc(QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size

                                  + 512);

    if (!s->cluster_data)

        goto fail;

    s->cluster_cache_offset = -1;

    if (refcount_init(bs) < 0)

        goto fail;

    /* read the backing file name */

    if (header.backing_file_offset != 0) {

        len = header.backing_file_size;

        if (len > 1023)

            len = 1023;

        if (bdrv_pread(s->hd, header.backing_file_offset, bs->backing_file, len) != len)

            goto fail;

        bs->backing_file[len] = '\0';

    }

    if (qcow_read_snapshots(bs) < 0)

        goto fail;

#ifdef DEBUG_ALLOC

    check_refcounts(bs);

#endif

    return 0;

 fail:

    qcow_free_snapshots(bs);

    refcount_close(bs);

    qemu_free(s->l1_table);

    qemu_free(s->l2_cache);

    qemu_free(s->cluster_cache);

    qemu_free(s->cluster_data);

    bdrv_delete(s->hd);

    return -1;

}

static int qcow_set_key(BlockDriverState *bs, const char *key)

{

    BDRVQcowState *s = bs->opaque;

    uint8_t keybuf[16];

    int len, i;

    memset(keybuf, 0, 16);

    len = strlen(key);

    if (len > 16)

        len = 16;

    /* XXX: we could compress the chars to 7 bits to increase

       entropy */

    for(i = 0;i < len;i++) {

        keybuf[i] = key[i];

    }

    s->crypt_method = s->crypt_method_header;

    if (AES_set_encrypt_key(keybuf, 128, &s->aes_encrypt_key) != 0)

        return -1;

    if (AES_set_decrypt_key(keybuf, 128, &s->aes_decrypt_key) != 0)

        return -1;

#if 0

    /* test */

    {

        uint8_t in[16];

        uint8_t out[16];

        uint8_t tmp[16];

        for(i=0;i<16;i++)

            in[i] = i;

        AES_encrypt(in, tmp, &s->aes_encrypt_key);

        AES_decrypt(tmp, out, &s->aes_decrypt_key);

        for(i = 0; i < 16; i++)

            printf(" %02x", tmp[i]);

        printf("\n");

        for(i = 0; i < 16; i++)

            printf(" %02x", out[i]);

        printf("\n");

    }

#endif

    return 0;

}

/* The crypt function is compatible with the linux cryptoloop

   algorithm for < 4 GB images. NOTE: out_buf == in_buf is

   supported */

static void 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 copy_sectors(BlockDriverState *bs, uint64_t start_sect,

                        uint64_t cluster_offset, int n_start, int n_end)

{

    BDRVQcowState *s = bs->opaque;

    int n, ret;

    n = n_end - n_start;

    if (n <= 0)

        return 0;

    ret = qcow_read(bs, start_sect + n_start, s->cluster_data, n);

    if (ret < 0)

        return ret;

    if (s->crypt_method) {

        encrypt_sectors(s, start_sect + n_start,

                        s->cluster_data,

                        s->cluster_data, n, 1,

                        &s->aes_encrypt_key);

    }

    ret = bdrv_write(s->hd, (cluster_offset >> 9) + n_start,

                     s->cluster_data, n);

    if (ret < 0)

        return ret;

    return 0;

}

static void l2_cache_reset(BlockDriverState *bs)

{

    BDRVQcowState *s = bs->opaque;

    memset(s->l2_cache, 0, s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));

    memset(s->l2_cache_offsets, 0, L2_CACHE_SIZE * sizeof(uint64_t));

    memset(s->l2_cache_counts, 0, L2_CACHE_SIZE * sizeof(uint32_t));

}

static inline int l2_cache_new_entry(BlockDriverState *bs)

{

    BDRVQcowState *s = bs->opaque;

    uint32_t min_count;

    int min_index, i;

    /* find a new entry in the least used one */

    min_index = 0;

    min_count = 0xffffffff;

    for(i = 0; i < L2_CACHE_SIZE; i++) {

        if (s->l2_cache_counts[i] < min_count) {

            min_count = s->l2_cache_counts[i];

            min_index = i;

        }

    }

    return min_index;

}

static int64_t align_offset(int64_t offset, int n)

{

    offset = (offset + n - 1) & ~(n - 1);

    return offset;

}

static int grow_l1_table(BlockDriverState *bs, int min_size)

{

    BDRVQcowState *s = bs->opaque;

    int new_l1_size, new_l1_size2, ret, i;

    uint64_t *new_l1_table;

    uint64_t new_l1_table_offset;

    uint8_t data[12];

    new_l1_size = s->l1_size;

    if (min_size <= new_l1_size)

        return 0;

    while (min_size > new_l1_size) {

        new_l1_size = (new_l1_size * 3 + 1) / 2;

    }

#ifdef DEBUG_ALLOC2

    printf("grow l1_table from %d to %d\n", s->l1_size, new_l1_size);

#endif

    new_l1_size2 = sizeof(uint64_t) * new_l1_size;

    new_l1_table = qemu_mallocz(new_l1_size2);

    if (!new_l1_table)

        return -ENOMEM;

    memcpy(new_l1_table, s->l1_table, s->l1_size * sizeof(uint64_t));

    /* write new table (align to cluster) */

    new_l1_table_offset = alloc_clusters(bs, new_l1_size2);

    for(i = 0; i < s->l1_size; i++)

        new_l1_table[i] = cpu_to_be64(new_l1_table[i]);

    ret = bdrv_pwrite(s->hd, new_l1_table_offset, new_l1_table, new_l1_size2);

    if (ret != new_l1_size2)

        goto fail;

    for(i = 0; i < s->l1_size; i++)

        new_l1_table[i] = be64_to_cpu(new_l1_table[i]);

    /* set new table */

    cpu_to_be32w((uint32_t*)data, new_l1_size);

    cpu_to_be64w((uint64_t*)(data + 4), new_l1_table_offset);

    if (bdrv_pwrite(s->hd, offsetof(QCowHeader, l1_size), data,

                sizeof(data)) != sizeof(data))

        goto fail;

    qemu_free(s->l1_table);

    free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t));

    s->l1_table_offset = new_l1_table_offset;

    s->l1_table = new_l1_table;

    s->l1_size = new_l1_size;

    return 0;

 fail:

    qemu_free(s->l1_table);

    return -EIO;

}

/*

 * seek_l2_table

 *

 * seek l2_offset in the l2_cache table

 * if not found, return NULL,

 * if found,

 *   increments the l2 cache hit count of the entry,

 *   if counter overflow, divide by two all counters

 *   return the pointer to the l2 cache entry

 *

 */

static uint64_t *seek_l2_table(BDRVQcowState *s, uint64_t l2_offset)

{

    int i, j;

    for(i = 0; i < L2_CACHE_SIZE; i++) {

        if (l2_offset == s->l2_cache_offsets[i]) {

            /* increment the hit count */

            if (++s->l2_cache_counts[i] == 0xffffffff) {

                for(j = 0; j < L2_CACHE_SIZE; j++) {

                    s->l2_cache_counts[j] >>= 1;

                }

            }

            return s->l2_cache + (i << s->l2_bits);

        }

    }

    return NULL;

}

/*

 * 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 uint64_t *l2_load(BlockDriverState *bs, uint64_t l2_offset)

{

    BDRVQcowState *s = bs->opaque;