2006-08-06 01:14:20 +04:00
|
|
|
/*
|
|
|
|
* Block driver for the QCOW version 2 format
|
2007-09-17 01:08:06 +04:00
|
|
|
*
|
2006-08-06 01:14:20 +04:00
|
|
|
* Copyright (c) 2004-2006 Fabrice Bellard
|
2007-09-17 01:08:06 +04:00
|
|
|
*
|
2006-08-06 01:14:20 +04:00
|
|
|
* 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.
|
|
|
|
*/
|
2007-11-11 05:51:17 +03:00
|
|
|
#include "qemu-common.h"
|
2006-08-06 01:14:20 +04:00
|
|
|
#include "block_int.h"
|
2009-05-10 02:03:42 +04:00
|
|
|
#include "module.h"
|
2006-08-06 01:14:20 +04:00
|
|
|
#include <zlib.h>
|
|
|
|
#include "aes.h"
|
2009-05-28 18:07:04 +04:00
|
|
|
#include "block/qcow2.h"
|
2010-06-11 23:37:37 +04:00
|
|
|
#include "qemu-error.h"
|
2011-02-09 13:11:07 +03:00
|
|
|
#include "qerror.h"
|
2006-08-06 01:14:20 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
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.
|
2007-09-17 01:08:06 +04:00
|
|
|
- Size of compressed clusters is stored in sectors to reduce bit usage
|
2006-08-06 01:14:20 +04:00
|
|
|
in the cluster offsets.
|
|
|
|
- Support for storing additional data (such as the VM state) in the
|
2007-09-17 12:09:54 +04:00
|
|
|
snapshots.
|
2006-08-06 01:14:20 +04:00
|
|
|
- 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.
|
|
|
|
*/
|
|
|
|
|
2009-03-28 20:55:06 +03:00
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
uint32_t magic;
|
|
|
|
uint32_t len;
|
|
|
|
} QCowExtension;
|
2010-12-17 18:02:39 +03:00
|
|
|
#define QCOW2_EXT_MAGIC_END 0
|
|
|
|
#define QCOW2_EXT_MAGIC_BACKING_FORMAT 0xE2792ACA
|
2009-03-28 20:55:06 +03:00
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_probe(const uint8_t *buf, int buf_size, const char *filename)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
|
|
|
const QCowHeader *cow_header = (const void *)buf;
|
2007-09-17 12:09:54 +04:00
|
|
|
|
2006-08-06 01:14:20 +04:00
|
|
|
if (buf_size >= sizeof(QCowHeader) &&
|
|
|
|
be32_to_cpu(cow_header->magic) == QCOW_MAGIC &&
|
2011-02-09 13:11:07 +03:00
|
|
|
be32_to_cpu(cow_header->version) >= QCOW_VERSION)
|
2006-08-06 01:14:20 +04:00
|
|
|
return 100;
|
|
|
|
else
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2009-03-28 20:55:06 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* read qcow2 extension and fill bs
|
|
|
|
* start reading from start_offset
|
|
|
|
* finish reading upon magic of value 0 or when end_offset reached
|
|
|
|
* unknown magic is skipped (future extension this version knows nothing about)
|
|
|
|
* return 0 upon success, non-0 otherwise
|
|
|
|
*/
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
|
|
|
|
uint64_t end_offset)
|
2009-03-28 20:55:06 +03:00
|
|
|
{
|
|
|
|
QCowExtension ext;
|
|
|
|
uint64_t offset;
|
|
|
|
|
|
|
|
#ifdef DEBUG_EXT
|
2010-12-17 18:02:39 +03:00
|
|
|
printf("qcow2_read_extensions: start=%ld end=%ld\n", start_offset, end_offset);
|
2009-03-28 20:55:06 +03:00
|
|
|
#endif
|
|
|
|
offset = start_offset;
|
|
|
|
while (offset < end_offset) {
|
|
|
|
|
|
|
|
#ifdef DEBUG_EXT
|
2011-08-04 21:22:10 +04:00
|
|
|
BDRVQcowState *s = bs->opaque;
|
2009-03-28 20:55:06 +03:00
|
|
|
/* Sanity check */
|
|
|
|
if (offset > s->cluster_size)
|
2010-12-17 18:02:39 +03:00
|
|
|
printf("qcow2_read_extension: suspicious offset %lu\n", offset);
|
2009-03-28 20:55:06 +03:00
|
|
|
|
2011-11-22 14:06:25 +04:00
|
|
|
printf("attempting to read extended header in offset %lu\n", offset);
|
2009-03-28 20:55:06 +03:00
|
|
|
#endif
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
if (bdrv_pread(bs->file, offset, &ext, sizeof(ext)) != sizeof(ext)) {
|
2010-12-17 18:02:39 +03:00
|
|
|
fprintf(stderr, "qcow2_read_extension: ERROR: "
|
2010-05-22 12:02:12 +04:00
|
|
|
"pread fail from offset %" PRIu64 "\n",
|
|
|
|
offset);
|
2009-03-28 20:55:06 +03:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
be32_to_cpus(&ext.magic);
|
|
|
|
be32_to_cpus(&ext.len);
|
|
|
|
offset += sizeof(ext);
|
|
|
|
#ifdef DEBUG_EXT
|
|
|
|
printf("ext.magic = 0x%x\n", ext.magic);
|
|
|
|
#endif
|
|
|
|
switch (ext.magic) {
|
2010-12-17 18:02:39 +03:00
|
|
|
case QCOW2_EXT_MAGIC_END:
|
2009-03-28 20:55:06 +03:00
|
|
|
return 0;
|
2009-03-28 20:55:14 +03:00
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
case QCOW2_EXT_MAGIC_BACKING_FORMAT:
|
2009-03-28 20:55:14 +03:00
|
|
|
if (ext.len >= sizeof(bs->backing_format)) {
|
|
|
|
fprintf(stderr, "ERROR: ext_backing_format: len=%u too large"
|
2009-03-29 05:31:56 +04:00
|
|
|
" (>=%zu)\n",
|
2009-03-28 20:55:14 +03:00
|
|
|
ext.len, sizeof(bs->backing_format));
|
|
|
|
return 2;
|
|
|
|
}
|
2010-04-14 16:17:38 +04:00
|
|
|
if (bdrv_pread(bs->file, offset , bs->backing_format,
|
2009-03-28 20:55:14 +03:00
|
|
|
ext.len) != ext.len)
|
|
|
|
return 3;
|
|
|
|
bs->backing_format[ext.len] = '\0';
|
|
|
|
#ifdef DEBUG_EXT
|
|
|
|
printf("Qcow2: Got format extension %s\n", bs->backing_format);
|
|
|
|
#endif
|
2009-11-26 16:03:42 +03:00
|
|
|
offset = ((offset + ext.len + 7) & ~7);
|
2009-03-28 20:55:14 +03:00
|
|
|
break;
|
|
|
|
|
2009-03-28 20:55:06 +03:00
|
|
|
default:
|
|
|
|
/* unknown magic -- just skip it */
|
2009-11-26 16:03:42 +03:00
|
|
|
offset = ((offset + ext.len + 7) & ~7);
|
2009-03-28 20:55:06 +03:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_open(BlockDriverState *bs, int flags)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
2010-12-17 18:02:40 +03:00
|
|
|
int len, i, ret = 0;
|
2006-08-06 01:14:20 +04:00
|
|
|
QCowHeader header;
|
2009-03-28 20:55:06 +03:00
|
|
|
uint64_t ext_end;
|
2011-01-10 19:17:28 +03:00
|
|
|
bool writethrough;
|
2006-08-06 01:14:20 +04:00
|
|
|
|
2010-12-17 18:02:40 +03:00
|
|
|
ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
|
|
|
|
if (ret < 0) {
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
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);
|
2007-09-17 12:09:54 +04:00
|
|
|
|
2011-02-09 13:11:07 +03:00
|
|
|
if (header.magic != QCOW_MAGIC) {
|
2010-12-17 18:02:40 +03:00
|
|
|
ret = -EINVAL;
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2011-02-09 13:11:07 +03:00
|
|
|
if (header.version != QCOW_VERSION) {
|
|
|
|
char version[64];
|
|
|
|
snprintf(version, sizeof(version), "QCOW version %d", header.version);
|
|
|
|
qerror_report(QERR_UNKNOWN_BLOCK_FORMAT_FEATURE,
|
|
|
|
bs->device_name, "qcow2", version);
|
|
|
|
ret = -ENOTSUP;
|
|
|
|
goto fail;
|
|
|
|
}
|
2009-10-26 18:11:16 +03:00
|
|
|
if (header.cluster_bits < MIN_CLUSTER_BITS ||
|
2010-12-17 18:02:40 +03:00
|
|
|
header.cluster_bits > MAX_CLUSTER_BITS) {
|
|
|
|
ret = -EINVAL;
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
|
|
|
if (header.crypt_method > QCOW_CRYPT_AES) {
|
|
|
|
ret = -EINVAL;
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
s->crypt_method_header = header.crypt_method;
|
2010-12-17 18:02:40 +03:00
|
|
|
if (s->crypt_method_header) {
|
2006-08-06 01:14:20 +04:00
|
|
|
bs->encrypted = 1;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
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;
|
2007-09-17 01:08:06 +04:00
|
|
|
s->refcount_table_size =
|
2006-08-06 01:14:20 +04:00
|
|
|
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;
|
2010-04-28 14:36:11 +04:00
|
|
|
s->l1_vm_state_index = size_to_l1(s, header.size);
|
2006-08-06 01:14:20 +04:00
|
|
|
/* the L1 table must contain at least enough entries to put
|
|
|
|
header.size bytes */
|
2010-12-17 18:02:40 +03:00
|
|
|
if (s->l1_size < s->l1_vm_state_index) {
|
|
|
|
ret = -EINVAL;
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
s->l1_table_offset = header.l1_table_offset;
|
2009-10-26 18:11:16 +03:00
|
|
|
if (s->l1_size > 0) {
|
2011-08-21 07:09:37 +04:00
|
|
|
s->l1_table = g_malloc0(
|
2009-10-26 18:11:16 +03:00
|
|
|
align_offset(s->l1_size * sizeof(uint64_t), 512));
|
2010-12-17 18:02:40 +03:00
|
|
|
ret = bdrv_pread(bs->file, s->l1_table_offset, s->l1_table,
|
|
|
|
s->l1_size * sizeof(uint64_t));
|
|
|
|
if (ret < 0) {
|
2009-10-26 18:11:16 +03:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2009-10-26 18:11:16 +03:00
|
|
|
for(i = 0;i < s->l1_size; i++) {
|
|
|
|
be64_to_cpus(&s->l1_table[i]);
|
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
}
|
2011-01-10 19:17:28 +03:00
|
|
|
|
|
|
|
/* alloc L2 table/refcount block cache */
|
2011-05-17 20:04:06 +04:00
|
|
|
writethrough = ((flags & BDRV_O_CACHE_WB) == 0);
|
2011-01-10 19:17:28 +03:00
|
|
|
s->l2_table_cache = qcow2_cache_create(bs, L2_CACHE_SIZE, writethrough);
|
|
|
|
s->refcount_block_cache = qcow2_cache_create(bs, REFCOUNT_CACHE_SIZE,
|
|
|
|
writethrough);
|
|
|
|
|
2011-08-21 07:09:37 +04:00
|
|
|
s->cluster_cache = g_malloc(s->cluster_size);
|
2006-08-06 01:14:20 +04:00
|
|
|
/* one more sector for decompressed data alignment */
|
2011-09-10 12:59:16 +04:00
|
|
|
s->cluster_data = qemu_blockalign(bs, QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size
|
2008-08-14 22:10:28 +04:00
|
|
|
+ 512);
|
2006-08-06 01:14:20 +04:00
|
|
|
s->cluster_cache_offset = -1;
|
2011-11-15 01:09:46 +04:00
|
|
|
s->flags = flags;
|
2007-09-17 12:09:54 +04:00
|
|
|
|
2010-12-17 18:02:40 +03:00
|
|
|
ret = qcow2_refcount_init(bs);
|
|
|
|
if (ret != 0) {
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
|
2009-09-12 11:36:22 +04:00
|
|
|
QLIST_INIT(&s->cluster_allocs);
|
2009-08-31 18:48:49 +04:00
|
|
|
|
2009-03-28 20:55:06 +03:00
|
|
|
/* read qcow2 extensions */
|
2010-12-17 18:02:40 +03:00
|
|
|
if (header.backing_file_offset) {
|
2009-03-28 20:55:06 +03:00
|
|
|
ext_end = header.backing_file_offset;
|
2010-12-17 18:02:40 +03:00
|
|
|
} else {
|
2009-03-28 20:55:06 +03:00
|
|
|
ext_end = s->cluster_size;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
|
|
|
if (qcow2_read_extensions(bs, sizeof(header), ext_end)) {
|
|
|
|
ret = -EINVAL;
|
2009-03-28 20:55:06 +03:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2009-03-28 20:55:06 +03:00
|
|
|
|
2006-08-06 01:14:20 +04:00
|
|
|
/* read the backing file name */
|
|
|
|
if (header.backing_file_offset != 0) {
|
|
|
|
len = header.backing_file_size;
|
2010-12-17 18:02:40 +03:00
|
|
|
if (len > 1023) {
|
2006-08-06 01:14:20 +04:00
|
|
|
len = 1023;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
|
|
|
ret = bdrv_pread(bs->file, header.backing_file_offset,
|
|
|
|
bs->backing_file, len);
|
|
|
|
if (ret < 0) {
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
bs->backing_file[len] = '\0';
|
|
|
|
}
|
2011-11-16 14:43:28 +04:00
|
|
|
|
|
|
|
ret = qcow2_read_snapshots(bs);
|
|
|
|
if (ret < 0) {
|
2006-08-06 01:14:20 +04:00
|
|
|
goto fail;
|
2010-12-17 18:02:40 +03:00
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
|
2011-06-30 19:42:09 +04:00
|
|
|
/* Initialise locks */
|
|
|
|
qemu_co_mutex_init(&s->lock);
|
|
|
|
|
2006-08-06 01:14:20 +04:00
|
|
|
#ifdef DEBUG_ALLOC
|
2011-08-04 21:22:10 +04:00
|
|
|
{
|
|
|
|
BdrvCheckResult result = {0};
|
|
|
|
qcow2_check_refcounts(bs, &result);
|
|
|
|
}
|
2006-08-06 01:14:20 +04:00
|
|
|
#endif
|
2010-12-17 18:02:40 +03:00
|
|
|
return ret;
|
2006-08-06 01:14:20 +04:00
|
|
|
|
|
|
|
fail:
|
2009-05-28 18:07:07 +04:00
|
|
|
qcow2_free_snapshots(bs);
|
|
|
|
qcow2_refcount_close(bs);
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(s->l1_table);
|
2011-01-10 19:17:28 +03:00
|
|
|
if (s->l2_table_cache) {
|
|
|
|
qcow2_cache_destroy(bs, s->l2_table_cache);
|
|
|
|
}
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(s->cluster_cache);
|
2011-09-10 12:59:16 +04:00
|
|
|
qemu_vfree(s->cluster_data);
|
2010-12-17 18:02:40 +03:00
|
|
|
return ret;
|
2006-08-06 01:14:20 +04:00
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_set_key(BlockDriverState *bs, const char *key)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
uint8_t keybuf[16];
|
|
|
|
int len, i;
|
2007-09-17 12:09:54 +04:00
|
|
|
|
2006-08-06 01:14:20 +04:00
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2011-11-14 16:44:21 +04:00
|
|
|
static int coroutine_fn qcow2_co_is_allocated(BlockDriverState *bs,
|
|
|
|
int64_t sector_num, int nb_sectors, int *pnum)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
2011-11-14 16:44:21 +04:00
|
|
|
BDRVQcowState *s = bs->opaque;
|
2006-08-06 01:14:20 +04:00
|
|
|
uint64_t cluster_offset;
|
2010-05-21 19:59:36 +04:00
|
|
|
int ret;
|
2006-08-06 01:14:20 +04:00
|
|
|
|
2008-08-14 22:10:28 +04:00
|
|
|
*pnum = nb_sectors;
|
2011-11-14 16:44:21 +04:00
|
|
|
/* FIXME We can get errors here, but the bdrv_co_is_allocated interface
|
|
|
|
* can't pass them on today */
|
|
|
|
qemu_co_mutex_lock(&s->lock);
|
2010-05-21 19:59:36 +04:00
|
|
|
ret = qcow2_get_cluster_offset(bs, sector_num << 9, pnum, &cluster_offset);
|
2011-11-14 16:44:21 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2010-05-21 19:59:36 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
*pnum = 0;
|
|
|
|
}
|
2008-08-14 22:10:28 +04:00
|
|
|
|
2006-08-06 01:14:20 +04:00
|
|
|
return (cluster_offset != 0);
|
|
|
|
}
|
|
|
|
|
2006-08-06 17:34:04 +04:00
|
|
|
/* handle reading after the end of the backing file */
|
2010-09-13 20:08:52 +04:00
|
|
|
int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov,
|
|
|
|
int64_t sector_num, int nb_sectors)
|
2006-08-06 17:34:04 +04:00
|
|
|
{
|
|
|
|
int n1;
|
|
|
|
if ((sector_num + nb_sectors) <= bs->total_sectors)
|
|
|
|
return nb_sectors;
|
|
|
|
if (sector_num >= bs->total_sectors)
|
|
|
|
n1 = 0;
|
|
|
|
else
|
|
|
|
n1 = bs->total_sectors - sector_num;
|
2010-09-13 20:08:52 +04:00
|
|
|
|
2011-02-03 18:12:49 +03:00
|
|
|
qemu_iovec_memset_skip(qiov, 0, 512 * (nb_sectors - n1), 512 * n1);
|
2010-09-13 20:08:52 +04:00
|
|
|
|
2006-08-06 17:34:04 +04:00
|
|
|
return n1;
|
|
|
|
}
|
|
|
|
|
2011-11-10 12:23:22 +04:00
|
|
|
static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
|
2011-08-23 17:21:19 +04:00
|
|
|
int remaining_sectors, QEMUIOVector *qiov)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
2006-08-06 17:34:04 +04:00
|
|
|
int index_in_cluster, n1;
|
2011-06-30 19:42:09 +04:00
|
|
|
int ret;
|
2011-08-23 17:21:14 +04:00
|
|
|
int cur_nr_sectors; /* number of sectors in current iteration */
|
2011-08-23 17:21:16 +04:00
|
|
|
uint64_t cluster_offset = 0;
|
2011-08-23 17:21:19 +04:00
|
|
|
uint64_t bytes_done = 0;
|
|
|
|
QEMUIOVector hd_qiov;
|
|
|
|
uint8_t *cluster_data = NULL;
|
2006-08-06 01:14:20 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_init(&hd_qiov, qiov->niov);
|
|
|
|
|
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
|
|
|
|
while (remaining_sectors != 0) {
|
2010-09-13 20:08:52 +04:00
|
|
|
|
2011-08-23 17:21:18 +04:00
|
|
|
/* prepare next request */
|
2011-08-23 17:21:19 +04:00
|
|
|
cur_nr_sectors = remaining_sectors;
|
2011-08-23 17:21:18 +04:00
|
|
|
if (s->crypt_method) {
|
|
|
|
cur_nr_sectors = MIN(cur_nr_sectors,
|
|
|
|
QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors);
|
2006-08-06 01:14:20 +04:00
|
|
|
}
|
2011-08-23 17:21:18 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
ret = qcow2_get_cluster_offset(bs, sector_num << 9,
|
2011-08-23 17:21:18 +04:00
|
|
|
&cur_nr_sectors, &cluster_offset);
|
2011-02-09 12:26:06 +03:00
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2011-02-09 12:26:06 +03:00
|
|
|
}
|
2010-09-13 20:08:52 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
index_in_cluster = sector_num & (s->cluster_sectors - 1);
|
2009-04-07 22:43:20 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_reset(&hd_qiov);
|
|
|
|
qemu_iovec_copy(&hd_qiov, qiov, bytes_done,
|
2011-08-23 17:21:18 +04:00
|
|
|
cur_nr_sectors * 512);
|
|
|
|
|
|
|
|
if (!cluster_offset) {
|
|
|
|
|
|
|
|
if (bs->backing_hd) {
|
|
|
|
/* read from the base image */
|
2011-08-23 17:21:19 +04:00
|
|
|
n1 = qcow2_backing_read1(bs->backing_hd, &hd_qiov,
|
|
|
|
sector_num, cur_nr_sectors);
|
2011-08-23 17:21:18 +04:00
|
|
|
if (n1 > 0) {
|
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO);
|
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2011-08-23 17:21:19 +04:00
|
|
|
ret = bdrv_co_readv(bs->backing_hd, sector_num,
|
|
|
|
n1, &hd_qiov);
|
2011-08-23 17:21:18 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
/* Note: in this case, no need to wait */
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_memset(&hd_qiov, 0, 512 * cur_nr_sectors);
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
|
|
|
} else if (cluster_offset & QCOW_OFLAG_COMPRESSED) {
|
|
|
|
/* add AIO support for compressed blocks ? */
|
|
|
|
ret = qcow2_decompress_cluster(bs, cluster_offset);
|
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2010-09-13 20:08:52 +04:00
|
|
|
}
|
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_from_buffer(&hd_qiov,
|
2011-08-23 17:21:18 +04:00
|
|
|
s->cluster_cache + index_in_cluster * 512,
|
2011-08-23 17:21:14 +04:00
|
|
|
512 * cur_nr_sectors);
|
2011-08-23 17:21:18 +04:00
|
|
|
} else {
|
|
|
|
if ((cluster_offset & 511) != 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
ret = -EIO;
|
|
|
|
goto fail;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
2010-09-13 20:08:52 +04:00
|
|
|
|
2011-08-23 17:21:18 +04:00
|
|
|
if (s->crypt_method) {
|
|
|
|
/*
|
|
|
|
* For encrypted images, read everything into a temporary
|
|
|
|
* contiguous buffer on which the AES functions can work.
|
|
|
|
*/
|
2011-08-23 17:21:19 +04:00
|
|
|
if (!cluster_data) {
|
|
|
|
cluster_data =
|
2011-09-10 12:59:16 +04:00
|
|
|
qemu_blockalign(bs, QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size);
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
assert(cur_nr_sectors <=
|
|
|
|
QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors);
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_reset(&hd_qiov);
|
|
|
|
qemu_iovec_add(&hd_qiov, cluster_data,
|
2011-08-23 17:21:18 +04:00
|
|
|
512 * cur_nr_sectors);
|
|
|
|
}
|
|
|
|
|
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
|
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
|
|
|
ret = bdrv_co_readv(bs->file,
|
|
|
|
(cluster_offset >> 9) + index_in_cluster,
|
2011-08-23 17:21:19 +04:00
|
|
|
cur_nr_sectors, &hd_qiov);
|
2011-08-23 17:21:18 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
|
|
|
if (s->crypt_method) {
|
2011-08-23 17:21:19 +04:00
|
|
|
qcow2_encrypt_sectors(s, sector_num, cluster_data,
|
|
|
|
cluster_data, cur_nr_sectors, 0, &s->aes_decrypt_key);
|
|
|
|
qemu_iovec_reset(&hd_qiov);
|
|
|
|
qemu_iovec_copy(&hd_qiov, qiov, bytes_done,
|
2011-08-23 17:21:18 +04:00
|
|
|
cur_nr_sectors * 512);
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_from_buffer(&hd_qiov, cluster_data,
|
2011-08-23 17:21:18 +04:00
|
|
|
512 * cur_nr_sectors);
|
|
|
|
}
|
2011-08-23 17:21:14 +04:00
|
|
|
}
|
2009-04-07 22:43:24 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
remaining_sectors -= cur_nr_sectors;
|
|
|
|
sector_num += cur_nr_sectors;
|
|
|
|
bytes_done += cur_nr_sectors * 512;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
2011-08-23 17:21:19 +04:00
|
|
|
ret = 0;
|
2011-08-23 17:21:14 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
fail:
|
2011-06-30 19:42:09 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2011-06-07 17:04:32 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_destroy(&hd_qiov);
|
2011-09-10 12:59:16 +04:00
|
|
|
qemu_vfree(cluster_data);
|
2011-06-30 19:42:09 +04:00
|
|
|
|
|
|
|
return ret;
|
2006-08-06 01:14:20 +04:00
|
|
|
}
|
|
|
|
|
2011-06-30 19:42:09 +04:00
|
|
|
static void run_dependent_requests(BDRVQcowState *s, QCowL2Meta *m)
|
2009-08-31 18:48:49 +04:00
|
|
|
{
|
|
|
|
/* Take the request off the list of running requests */
|
|
|
|
if (m->nb_clusters != 0) {
|
2009-09-12 11:36:22 +04:00
|
|
|
QLIST_REMOVE(m, next_in_flight);
|
2009-08-31 18:48:49 +04:00
|
|
|
}
|
|
|
|
|
2010-04-15 17:11:35 +04:00
|
|
|
/* Restart all dependent requests */
|
2011-06-30 19:42:09 +04:00
|
|
|
if (!qemu_co_queue_empty(&m->dependent_requests)) {
|
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2011-11-17 17:40:26 +04:00
|
|
|
qemu_co_queue_restart_all(&m->dependent_requests);
|
2011-06-30 19:42:09 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
2009-08-31 18:48:49 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-11-10 12:23:22 +04:00
|
|
|
static coroutine_fn int qcow2_co_writev(BlockDriverState *bs,
|
2011-08-23 17:21:19 +04:00
|
|
|
int64_t sector_num,
|
|
|
|
int remaining_sectors,
|
|
|
|
QEMUIOVector *qiov)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int index_in_cluster;
|
2008-08-14 22:10:28 +04:00
|
|
|
int n_end;
|
2011-06-30 19:42:09 +04:00
|
|
|
int ret;
|
2011-08-23 17:21:14 +04:00
|
|
|
int cur_nr_sectors; /* number of sectors in current iteration */
|
2011-08-23 17:21:16 +04:00
|
|
|
uint64_t cluster_offset;
|
2011-08-23 17:21:19 +04:00
|
|
|
QEMUIOVector hd_qiov;
|
|
|
|
uint64_t bytes_done = 0;
|
|
|
|
uint8_t *cluster_data = NULL;
|
2011-09-01 17:00:42 +04:00
|
|
|
QCowL2Meta l2meta = {
|
|
|
|
.nb_clusters = 0,
|
|
|
|
};
|
2011-08-23 17:21:15 +04:00
|
|
|
|
|
|
|
qemu_co_queue_init(&l2meta.dependent_requests);
|
2006-08-06 01:14:20 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_init(&hd_qiov, qiov->niov);
|
|
|
|
|
|
|
|
s->cluster_cache_offset = -1; /* disable compressed cache */
|
2007-09-17 12:09:54 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
|
|
|
|
while (remaining_sectors != 0) {
|
|
|
|
|
|
|
|
index_in_cluster = sector_num & (s->cluster_sectors - 1);
|
|
|
|
n_end = index_in_cluster + remaining_sectors;
|
2011-08-23 17:21:18 +04:00
|
|
|
if (s->crypt_method &&
|
|
|
|
n_end > QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors) {
|
|
|
|
n_end = QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors;
|
|
|
|
}
|
2008-08-14 22:10:28 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
ret = qcow2_alloc_cluster_offset(bs, sector_num << 9,
|
2011-08-23 17:21:18 +04:00
|
|
|
index_in_cluster, n_end, &cur_nr_sectors, &l2meta);
|
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
2010-01-20 17:03:01 +03:00
|
|
|
|
2011-08-23 17:21:18 +04:00
|
|
|
cluster_offset = l2meta.cluster_offset;
|
|
|
|
assert((cluster_offset & 511) == 0);
|
2010-01-20 17:03:01 +03:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_reset(&hd_qiov);
|
|
|
|
qemu_iovec_copy(&hd_qiov, qiov, bytes_done,
|
2011-08-23 17:21:18 +04:00
|
|
|
cur_nr_sectors * 512);
|
2010-09-13 20:24:10 +04:00
|
|
|
|
2011-08-23 17:21:18 +04:00
|
|
|
if (s->crypt_method) {
|
2011-08-23 17:21:19 +04:00
|
|
|
if (!cluster_data) {
|
2011-09-10 12:59:16 +04:00
|
|
|
cluster_data = qemu_blockalign(bs, QCOW_MAX_CRYPT_CLUSTERS *
|
2011-08-23 17:21:18 +04:00
|
|
|
s->cluster_size);
|
|
|
|
}
|
2010-09-13 20:24:10 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
assert(hd_qiov.size <=
|
2011-08-23 17:21:18 +04:00
|
|
|
QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size);
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_to_buffer(&hd_qiov, cluster_data);
|
2010-09-13 20:24:10 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qcow2_encrypt_sectors(s, sector_num, cluster_data,
|
|
|
|
cluster_data, cur_nr_sectors, 1, &s->aes_encrypt_key);
|
2010-09-13 20:24:10 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_reset(&hd_qiov);
|
|
|
|
qemu_iovec_add(&hd_qiov, cluster_data,
|
2011-08-23 17:21:18 +04:00
|
|
|
cur_nr_sectors * 512);
|
|
|
|
}
|
2010-09-13 20:24:10 +04:00
|
|
|
|
2011-08-23 17:21:18 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO);
|
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
|
|
|
ret = bdrv_co_writev(bs->file,
|
|
|
|
(cluster_offset >> 9) + index_in_cluster,
|
2011-08-23 17:21:19 +04:00
|
|
|
cur_nr_sectors, &hd_qiov);
|
2011-08-23 17:21:18 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
2009-04-07 22:43:24 +04:00
|
|
|
|
2011-08-23 17:21:18 +04:00
|
|
|
ret = qcow2_alloc_cluster_link_l2(bs, &l2meta);
|
|
|
|
if (ret < 0) {
|
2011-08-23 17:21:19 +04:00
|
|
|
goto fail;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
2011-08-23 17:21:14 +04:00
|
|
|
|
2011-09-01 17:02:13 +04:00
|
|
|
run_dependent_requests(s, &l2meta);
|
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
remaining_sectors -= cur_nr_sectors;
|
|
|
|
sector_num += cur_nr_sectors;
|
|
|
|
bytes_done += cur_nr_sectors * 512;
|
2011-08-23 17:21:18 +04:00
|
|
|
}
|
2011-08-23 17:21:19 +04:00
|
|
|
ret = 0;
|
2011-08-23 17:21:14 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
fail:
|
2011-09-01 17:02:13 +04:00
|
|
|
run_dependent_requests(s, &l2meta);
|
|
|
|
|
2011-06-30 19:42:09 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2007-09-17 12:09:54 +04:00
|
|
|
|
2011-08-23 17:21:19 +04:00
|
|
|
qemu_iovec_destroy(&hd_qiov);
|
2011-09-10 12:59:16 +04:00
|
|
|
qemu_vfree(cluster_data);
|
2011-06-07 17:04:32 +04:00
|
|
|
|
2011-06-30 19:42:09 +04:00
|
|
|
return ret;
|
2006-08-06 01:14:20 +04:00
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static void qcow2_close(BlockDriverState *bs)
|
2006-08-06 01:14:20 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(s->l1_table);
|
2011-01-10 19:17:28 +03:00
|
|
|
|
|
|
|
qcow2_cache_flush(bs, s->l2_table_cache);
|
|
|
|
qcow2_cache_flush(bs, s->refcount_block_cache);
|
|
|
|
|
|
|
|
qcow2_cache_destroy(bs, s->l2_table_cache);
|
|
|
|
qcow2_cache_destroy(bs, s->refcount_block_cache);
|
|
|
|
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(s->cluster_cache);
|
2011-09-10 12:59:16 +04:00
|
|
|
qemu_vfree(s->cluster_data);
|
2009-05-28 18:07:07 +04:00
|
|
|
qcow2_refcount_close(bs);
|
2006-08-06 01:14:20 +04:00
|
|
|
}
|
|
|
|
|
2011-11-15 01:09:46 +04:00
|
|
|
static void qcow2_invalidate_cache(BlockDriverState *bs)
|
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int flags = s->flags;
|
|
|
|
AES_KEY aes_encrypt_key;
|
|
|
|
AES_KEY aes_decrypt_key;
|
|
|
|
uint32_t crypt_method = 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Backing files are read-only which makes all of their metadata immutable,
|
|
|
|
* that means we don't have to worry about reopening them here.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (s->crypt_method) {
|
|
|
|
crypt_method = s->crypt_method;
|
|
|
|
memcpy(&aes_encrypt_key, &s->aes_encrypt_key, sizeof(aes_encrypt_key));
|
|
|
|
memcpy(&aes_decrypt_key, &s->aes_decrypt_key, sizeof(aes_decrypt_key));
|
|
|
|
}
|
|
|
|
|
|
|
|
qcow2_close(bs);
|
|
|
|
|
|
|
|
memset(s, 0, sizeof(BDRVQcowState));
|
|
|
|
qcow2_open(bs, flags);
|
|
|
|
|
|
|
|
if (crypt_method) {
|
|
|
|
s->crypt_method = crypt_method;
|
|
|
|
memcpy(&s->aes_encrypt_key, &aes_encrypt_key, sizeof(aes_encrypt_key));
|
|
|
|
memcpy(&s->aes_decrypt_key, &aes_decrypt_key, sizeof(aes_decrypt_key));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-01-12 14:55:17 +03:00
|
|
|
/*
|
|
|
|
* Updates the variable length parts of the qcow2 header, i.e. the backing file
|
|
|
|
* name and all extensions. qcow2 was not designed to allow such changes, so if
|
|
|
|
* we run out of space (we can only use the first cluster) this function may
|
|
|
|
* fail.
|
|
|
|
*
|
|
|
|
* Returns 0 on success, -errno in error cases.
|
|
|
|
*/
|
|
|
|
static int qcow2_update_ext_header(BlockDriverState *bs,
|
|
|
|
const char *backing_file, const char *backing_fmt)
|
|
|
|
{
|
|
|
|
size_t backing_file_len = 0;
|
|
|
|
size_t backing_fmt_len = 0;
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
QCowExtension ext_backing_fmt = {0, 0};
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
/* Backing file format doesn't make sense without a backing file */
|
|
|
|
if (backing_fmt && !backing_file) {
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Prepare the backing file format extension if needed */
|
|
|
|
if (backing_fmt) {
|
|
|
|
ext_backing_fmt.len = cpu_to_be32(strlen(backing_fmt));
|
2010-12-17 18:02:39 +03:00
|
|
|
ext_backing_fmt.magic = cpu_to_be32(QCOW2_EXT_MAGIC_BACKING_FORMAT);
|
2010-01-12 14:55:17 +03:00
|
|
|
backing_fmt_len = ((sizeof(ext_backing_fmt)
|
|
|
|
+ strlen(backing_fmt) + 7) & ~7);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Check if we can fit the new header into the first cluster */
|
|
|
|
if (backing_file) {
|
|
|
|
backing_file_len = strlen(backing_file);
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t header_size = sizeof(QCowHeader) + backing_file_len
|
|
|
|
+ backing_fmt_len;
|
|
|
|
|
|
|
|
if (header_size > s->cluster_size) {
|
|
|
|
return -ENOSPC;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Rewrite backing file name and qcow2 extensions */
|
|
|
|
size_t ext_size = header_size - sizeof(QCowHeader);
|
|
|
|
uint8_t buf[ext_size];
|
|
|
|
size_t offset = 0;
|
|
|
|
size_t backing_file_offset = 0;
|
|
|
|
|
|
|
|
if (backing_file) {
|
|
|
|
if (backing_fmt) {
|
|
|
|
int padding = backing_fmt_len -
|
|
|
|
(sizeof(ext_backing_fmt) + strlen(backing_fmt));
|
|
|
|
|
|
|
|
memcpy(buf + offset, &ext_backing_fmt, sizeof(ext_backing_fmt));
|
|
|
|
offset += sizeof(ext_backing_fmt);
|
|
|
|
|
|
|
|
memcpy(buf + offset, backing_fmt, strlen(backing_fmt));
|
|
|
|
offset += strlen(backing_fmt);
|
|
|
|
|
|
|
|
memset(buf + offset, 0, padding);
|
|
|
|
offset += padding;
|
|
|
|
}
|
|
|
|
|
|
|
|
memcpy(buf + offset, backing_file, backing_file_len);
|
|
|
|
backing_file_offset = sizeof(QCowHeader) + offset;
|
|
|
|
}
|
|
|
|
|
2010-06-16 19:44:35 +04:00
|
|
|
ret = bdrv_pwrite_sync(bs->file, sizeof(QCowHeader), buf, ext_size);
|
2010-01-12 14:55:17 +03:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Update header fields */
|
|
|
|
uint64_t be_backing_file_offset = cpu_to_be64(backing_file_offset);
|
|
|
|
uint32_t be_backing_file_size = cpu_to_be32(backing_file_len);
|
|
|
|
|
2010-06-16 19:44:35 +04:00
|
|
|
ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, backing_file_offset),
|
2010-01-12 14:55:17 +03:00
|
|
|
&be_backing_file_offset, sizeof(uint64_t));
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2010-06-16 19:44:35 +04:00
|
|
|
ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, backing_file_size),
|
2010-01-12 14:55:17 +03:00
|
|
|
&be_backing_file_size, sizeof(uint32_t));
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
fail:
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int qcow2_change_backing_file(BlockDriverState *bs,
|
|
|
|
const char *backing_file, const char *backing_fmt)
|
|
|
|
{
|
|
|
|
return qcow2_update_ext_header(bs, backing_file, backing_fmt);
|
|
|
|
}
|
|
|
|
|
2009-08-17 17:50:10 +04:00
|
|
|
static int preallocate(BlockDriverState *bs)
|
|
|
|
{
|
|
|
|
uint64_t nb_sectors;
|
|
|
|
uint64_t offset;
|
|
|
|
int num;
|
2010-01-20 17:03:01 +03:00
|
|
|
int ret;
|
2009-08-17 17:50:10 +04:00
|
|
|
QCowL2Meta meta;
|
|
|
|
|
|
|
|
nb_sectors = bdrv_getlength(bs) >> 9;
|
|
|
|
offset = 0;
|
2011-06-30 19:42:09 +04:00
|
|
|
qemu_co_queue_init(&meta.dependent_requests);
|
2010-01-20 17:03:01 +03:00
|
|
|
meta.cluster_offset = 0;
|
2009-08-17 17:50:10 +04:00
|
|
|
|
|
|
|
while (nb_sectors) {
|
|
|
|
num = MIN(nb_sectors, INT_MAX >> 9);
|
2010-01-20 17:03:01 +03:00
|
|
|
ret = qcow2_alloc_cluster_offset(bs, offset, 0, num, &num, &meta);
|
|
|
|
if (ret < 0) {
|
2010-06-22 18:59:46 +04:00
|
|
|
return ret;
|
2009-08-17 17:50:10 +04:00
|
|
|
}
|
|
|
|
|
2010-06-22 18:59:46 +04:00
|
|
|
ret = qcow2_alloc_cluster_link_l2(bs, &meta);
|
|
|
|
if (ret < 0) {
|
2010-01-20 17:03:01 +03:00
|
|
|
qcow2_free_any_clusters(bs, meta.cluster_offset, meta.nb_clusters);
|
2010-06-22 18:59:46 +04:00
|
|
|
return ret;
|
2009-08-17 17:50:10 +04:00
|
|
|
}
|
|
|
|
|
2009-08-31 18:48:49 +04:00
|
|
|
/* There are no dependent requests, but we need to remove our request
|
|
|
|
* from the list of in-flight requests */
|
2011-06-30 19:42:09 +04:00
|
|
|
run_dependent_requests(bs->opaque, &meta);
|
2009-08-31 18:48:49 +04:00
|
|
|
|
2009-08-17 17:50:10 +04:00
|
|
|
/* TODO Preallocate data if requested */
|
|
|
|
|
|
|
|
nb_sectors -= num;
|
|
|
|
offset += num << 9;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* It is expected that the image file is large enough to actually contain
|
|
|
|
* all of the allocated clusters (otherwise we get failing reads after
|
|
|
|
* EOF). Extend the image to the last allocated sector.
|
|
|
|
*/
|
2010-01-20 17:03:01 +03:00
|
|
|
if (meta.cluster_offset != 0) {
|
2009-08-31 14:26:57 +04:00
|
|
|
uint8_t buf[512];
|
|
|
|
memset(buf, 0, 512);
|
2010-06-22 18:59:46 +04:00
|
|
|
ret = bdrv_write(bs->file, (meta.cluster_offset >> 9) + num - 1, buf, 1);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
2009-08-17 17:50:10 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_create2(const char *filename, int64_t total_size,
|
|
|
|
const char *backing_file, const char *backing_format,
|
|
|
|
int flags, size_t cluster_size, int prealloc,
|
|
|
|
QEMUOptionParameter *options)
|
2010-06-11 23:37:37 +04:00
|
|
|
{
|
2011-11-22 14:06:25 +04:00
|
|
|
/* Calculate cluster_bits */
|
2010-06-11 23:37:37 +04:00
|
|
|
int cluster_bits;
|
|
|
|
cluster_bits = ffs(cluster_size) - 1;
|
|
|
|
if (cluster_bits < MIN_CLUSTER_BITS || cluster_bits > MAX_CLUSTER_BITS ||
|
|
|
|
(1 << cluster_bits) != cluster_size)
|
|
|
|
{
|
|
|
|
error_report(
|
2011-06-22 16:03:54 +04:00
|
|
|
"Cluster size must be a power of two between %d and %dk",
|
2010-06-11 23:37:37 +04:00
|
|
|
1 << MIN_CLUSTER_BITS, 1 << (MAX_CLUSTER_BITS - 10));
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Open the image file and write a minimal qcow2 header.
|
|
|
|
*
|
|
|
|
* We keep things simple and start with a zero-sized image. We also
|
|
|
|
* do without refcount blocks or a L1 table for now. We'll fix the
|
|
|
|
* inconsistency later.
|
|
|
|
*
|
|
|
|
* We do need a refcount table because growing the refcount table means
|
|
|
|
* allocating two new refcount blocks - the seconds of which would be at
|
|
|
|
* 2 GB for 64k clusters, and we don't want to have a 2 GB initial file
|
|
|
|
* size for any qcow2 image.
|
|
|
|
*/
|
|
|
|
BlockDriverState* bs;
|
|
|
|
QCowHeader header;
|
|
|
|
uint8_t* refcount_table;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = bdrv_create_file(filename, options);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = bdrv_file_open(&bs, filename, BDRV_O_RDWR);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Write the header */
|
|
|
|
memset(&header, 0, sizeof(header));
|
|
|
|
header.magic = cpu_to_be32(QCOW_MAGIC);
|
|
|
|
header.version = cpu_to_be32(QCOW_VERSION);
|
|
|
|
header.cluster_bits = cpu_to_be32(cluster_bits);
|
|
|
|
header.size = cpu_to_be64(0);
|
|
|
|
header.l1_table_offset = cpu_to_be64(0);
|
|
|
|
header.l1_size = cpu_to_be32(0);
|
|
|
|
header.refcount_table_offset = cpu_to_be64(cluster_size);
|
|
|
|
header.refcount_table_clusters = cpu_to_be32(1);
|
|
|
|
|
|
|
|
if (flags & BLOCK_FLAG_ENCRYPT) {
|
|
|
|
header.crypt_method = cpu_to_be32(QCOW_CRYPT_AES);
|
|
|
|
} else {
|
|
|
|
header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = bdrv_pwrite(bs, 0, &header, sizeof(header));
|
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Write an empty refcount table */
|
2011-08-21 07:09:37 +04:00
|
|
|
refcount_table = g_malloc0(cluster_size);
|
2010-06-11 23:37:37 +04:00
|
|
|
ret = bdrv_pwrite(bs, cluster_size, refcount_table, cluster_size);
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(refcount_table);
|
2010-06-11 23:37:37 +04:00
|
|
|
|
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
bdrv_close(bs);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* And now open the image and make it consistent first (i.e. increase the
|
|
|
|
* refcount of the cluster that is occupied by the header and the refcount
|
|
|
|
* table)
|
|
|
|
*/
|
|
|
|
BlockDriver* drv = bdrv_find_format("qcow2");
|
|
|
|
assert(drv != NULL);
|
2011-01-27 18:46:01 +03:00
|
|
|
ret = bdrv_open(bs, filename,
|
|
|
|
BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_NO_FLUSH, drv);
|
2010-06-11 23:37:37 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = qcow2_alloc_clusters(bs, 2 * cluster_size);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
} else if (ret != 0) {
|
|
|
|
error_report("Huh, first cluster in empty image is already in use?");
|
|
|
|
abort();
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Okay, now that we have a valid image, let's give it the right size */
|
|
|
|
ret = bdrv_truncate(bs, total_size * BDRV_SECTOR_SIZE);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Want a backing file? There you go.*/
|
|
|
|
if (backing_file) {
|
|
|
|
ret = bdrv_change_backing_file(bs, backing_file, backing_format);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* And if we're supposed to preallocate metadata, do that now */
|
|
|
|
if (prealloc) {
|
|
|
|
ret = preallocate(bs);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
out:
|
|
|
|
bdrv_delete(bs);
|
|
|
|
return ret;
|
|
|
|
}
|
2010-05-07 14:43:45 +04:00
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_create(const char *filename, QEMUOptionParameter *options)
|
2010-05-07 14:43:45 +04:00
|
|
|
{
|
|
|
|
const char *backing_file = NULL;
|
|
|
|
const char *backing_fmt = NULL;
|
|
|
|
uint64_t sectors = 0;
|
|
|
|
int flags = 0;
|
2011-05-31 17:01:46 +04:00
|
|
|
size_t cluster_size = DEFAULT_CLUSTER_SIZE;
|
2010-05-07 14:43:45 +04:00
|
|
|
int prealloc = 0;
|
|
|
|
|
|
|
|
/* Read out options */
|
|
|
|
while (options && options->name) {
|
|
|
|
if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
|
|
|
|
sectors = options->value.n / 512;
|
|
|
|
} else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
|
|
|
|
backing_file = options->value.s;
|
|
|
|
} else if (!strcmp(options->name, BLOCK_OPT_BACKING_FMT)) {
|
|
|
|
backing_fmt = options->value.s;
|
|
|
|
} else if (!strcmp(options->name, BLOCK_OPT_ENCRYPT)) {
|
|
|
|
flags |= options->value.n ? BLOCK_FLAG_ENCRYPT : 0;
|
|
|
|
} else if (!strcmp(options->name, BLOCK_OPT_CLUSTER_SIZE)) {
|
|
|
|
if (options->value.n) {
|
|
|
|
cluster_size = options->value.n;
|
|
|
|
}
|
|
|
|
} else if (!strcmp(options->name, BLOCK_OPT_PREALLOC)) {
|
|
|
|
if (!options->value.s || !strcmp(options->value.s, "off")) {
|
|
|
|
prealloc = 0;
|
|
|
|
} else if (!strcmp(options->value.s, "metadata")) {
|
|
|
|
prealloc = 1;
|
|
|
|
} else {
|
|
|
|
fprintf(stderr, "Invalid preallocation mode: '%s'\n",
|
|
|
|
options->value.s);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
options++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (backing_file && prealloc) {
|
|
|
|
fprintf(stderr, "Backing file and preallocation cannot be used at "
|
|
|
|
"the same time\n");
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
return qcow2_create2(filename, sectors, backing_file, backing_fmt, flags,
|
|
|
|
cluster_size, prealloc, options);
|
2010-05-07 14:43:45 +04:00
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_make_empty(BlockDriverState *bs)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
#if 0
|
|
|
|
/* XXX: not correct */
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
uint32_t l1_length = s->l1_size * sizeof(uint64_t);
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
memset(s->l1_table, 0, l1_length);
|
2010-04-14 16:17:38 +04:00
|
|
|
if (bdrv_pwrite(bs->file, s->l1_table_offset, s->l1_table, l1_length) < 0)
|
2010-04-24 00:19:47 +04:00
|
|
|
return -1;
|
2010-04-14 16:17:38 +04:00
|
|
|
ret = bdrv_truncate(bs->file, s->l1_table_offset + l1_length);
|
2010-04-24 00:19:47 +04:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
l2_cache_reset(bs);
|
|
|
|
#endif
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2011-10-20 15:16:25 +04:00
|
|
|
static coroutine_fn int qcow2_co_discard(BlockDriverState *bs,
|
|
|
|
int64_t sector_num, int nb_sectors)
|
2011-01-26 18:56:48 +03:00
|
|
|
{
|
2011-10-20 15:16:25 +04:00
|
|
|
int ret;
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
|
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
ret = qcow2_discard_clusters(bs, sector_num << BDRV_SECTOR_BITS,
|
2011-01-26 18:56:48 +03:00
|
|
|
nb_sectors);
|
2011-10-20 15:16:25 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
|
|
|
return ret;
|
2011-01-26 18:56:48 +03:00
|
|
|
}
|
|
|
|
|
2010-04-28 14:36:11 +04:00
|
|
|
static int qcow2_truncate(BlockDriverState *bs, int64_t offset)
|
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int ret, new_l1_size;
|
|
|
|
|
|
|
|
if (offset & 511) {
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* cannot proceed if image has snapshots */
|
|
|
|
if (s->nb_snapshots) {
|
|
|
|
return -ENOTSUP;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* shrinking is currently not supported */
|
|
|
|
if (offset < bs->total_sectors * 512) {
|
|
|
|
return -ENOTSUP;
|
|
|
|
}
|
|
|
|
|
|
|
|
new_l1_size = size_to_l1(s, offset);
|
2010-10-18 19:53:53 +04:00
|
|
|
ret = qcow2_grow_l1_table(bs, new_l1_size, true);
|
2010-04-28 14:36:11 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* write updated header.size */
|
|
|
|
offset = cpu_to_be64(offset);
|
2010-06-16 19:44:35 +04:00
|
|
|
ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, size),
|
|
|
|
&offset, sizeof(uint64_t));
|
2010-04-28 14:36:11 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
s->l1_vm_state_index = new_l1_size;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-04-24 00:19:47 +04:00
|
|
|
/* XXX: put compressed sectors first, then all the cluster aligned
|
|
|
|
tables to avoid losing bytes in alignment */
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_write_compressed(BlockDriverState *bs, int64_t sector_num,
|
|
|
|
const uint8_t *buf, int nb_sectors)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
z_stream strm;
|
|
|
|
int ret, out_len;
|
|
|
|
uint8_t *out_buf;
|
|
|
|
uint64_t cluster_offset;
|
|
|
|
|
|
|
|
if (nb_sectors == 0) {
|
|
|
|
/* align end of file to a sector boundary to ease reading with
|
|
|
|
sector based I/Os */
|
2010-04-14 16:17:38 +04:00
|
|
|
cluster_offset = bdrv_getlength(bs->file);
|
2010-04-24 00:19:47 +04:00
|
|
|
cluster_offset = (cluster_offset + 511) & ~511;
|
2010-04-14 16:17:38 +04:00
|
|
|
bdrv_truncate(bs->file, cluster_offset);
|
2010-04-24 00:19:47 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (nb_sectors != s->cluster_sectors)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2011-08-21 07:09:37 +04:00
|
|
|
out_buf = g_malloc(s->cluster_size + (s->cluster_size / 1000) + 128);
|
2010-04-24 00:19:47 +04:00
|
|
|
|
|
|
|
/* best compression, small window, no zlib header */
|
|
|
|
memset(&strm, 0, sizeof(strm));
|
|
|
|
ret = deflateInit2(&strm, Z_DEFAULT_COMPRESSION,
|
|
|
|
Z_DEFLATED, -12,
|
|
|
|
9, Z_DEFAULT_STRATEGY);
|
|
|
|
if (ret != 0) {
|
2011-10-18 19:12:44 +04:00
|
|
|
ret = -EINVAL;
|
|
|
|
goto fail;
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
strm.avail_in = s->cluster_size;
|
|
|
|
strm.next_in = (uint8_t *)buf;
|
|
|
|
strm.avail_out = s->cluster_size;
|
|
|
|
strm.next_out = out_buf;
|
|
|
|
|
|
|
|
ret = deflate(&strm, Z_FINISH);
|
|
|
|
if (ret != Z_STREAM_END && ret != Z_OK) {
|
|
|
|
deflateEnd(&strm);
|
2011-10-18 19:12:44 +04:00
|
|
|
ret = -EINVAL;
|
|
|
|
goto fail;
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
out_len = strm.next_out - out_buf;
|
|
|
|
|
|
|
|
deflateEnd(&strm);
|
|
|
|
|
|
|
|
if (ret != Z_STREAM_END || out_len >= s->cluster_size) {
|
|
|
|
/* could not compress: write normal cluster */
|
2011-10-18 19:12:44 +04:00
|
|
|
ret = bdrv_write(bs, sector_num, buf, s->cluster_sectors);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
2010-04-24 00:19:47 +04:00
|
|
|
} else {
|
|
|
|
cluster_offset = qcow2_alloc_compressed_cluster_offset(bs,
|
|
|
|
sector_num << 9, out_len);
|
2011-10-18 19:12:44 +04:00
|
|
|
if (!cluster_offset) {
|
|
|
|
ret = -EIO;
|
|
|
|
goto fail;
|
|
|
|
}
|
2010-04-24 00:19:47 +04:00
|
|
|
cluster_offset &= s->cluster_offset_mask;
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_WRITE_COMPRESSED);
|
2011-10-18 19:12:44 +04:00
|
|
|
ret = bdrv_pwrite(bs->file, cluster_offset, out_buf, out_len);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-10-18 19:12:44 +04:00
|
|
|
ret = 0;
|
|
|
|
fail:
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(out_buf);
|
2011-10-18 19:12:44 +04:00
|
|
|
return ret;
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
|
2011-11-10 12:23:22 +04:00
|
|
|
static coroutine_fn int qcow2_co_flush_to_os(BlockDriverState *bs)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
2011-01-10 19:17:28 +03:00
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int ret;
|
|
|
|
|
2011-10-20 15:16:24 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
2011-01-10 19:17:28 +03:00
|
|
|
ret = qcow2_cache_flush(bs, s->l2_table_cache);
|
|
|
|
if (ret < 0) {
|
2011-10-27 13:22:28 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2011-10-20 15:16:24 +04:00
|
|
|
return ret;
|
2011-01-10 19:17:28 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
ret = qcow2_cache_flush(bs, s->refcount_block_cache);
|
|
|
|
if (ret < 0) {
|
2011-10-27 13:22:28 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2011-10-20 15:16:24 +04:00
|
|
|
return ret;
|
2011-01-10 19:17:28 +03:00
|
|
|
}
|
2011-10-20 15:16:24 +04:00
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2011-01-10 19:17:28 +03:00
|
|
|
|
2011-11-10 21:10:11 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2011-11-10 12:23:22 +04:00
|
|
|
static coroutine_fn int qcow2_co_flush_to_disk(BlockDriverState *bs)
|
2011-11-10 21:10:11 +04:00
|
|
|
{
|
2011-10-20 15:16:24 +04:00
|
|
|
return bdrv_co_flush(bs->file);
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int64_t qcow2_vm_state_offset(BDRVQcowState *s)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
return (int64_t)s->l1_vm_state_index << (s->cluster_bits + s->l2_bits);
|
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
bdi->cluster_size = s->cluster_size;
|
2010-12-17 18:02:39 +03:00
|
|
|
bdi->vm_state_offset = qcow2_vm_state_offset(s);
|
2010-04-24 00:19:47 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_check(BlockDriverState *bs, BdrvCheckResult *result)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
2010-06-29 14:37:54 +04:00
|
|
|
return qcow2_check_refcounts(bs, result);
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
#if 0
|
|
|
|
static void dump_refcounts(BlockDriverState *bs)
|
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int64_t nb_clusters, k, k1, size;
|
|
|
|
int refcount;
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
size = bdrv_getlength(bs->file);
|
2010-04-24 00:19:47 +04:00
|
|
|
nb_clusters = size_to_clusters(s, size);
|
|
|
|
for(k = 0; k < nb_clusters;) {
|
|
|
|
k1 = k;
|
|
|
|
refcount = get_refcount(bs, k);
|
|
|
|
k++;
|
|
|
|
while (k < nb_clusters && get_refcount(bs, k) == refcount)
|
|
|
|
k++;
|
2010-05-22 12:02:12 +04:00
|
|
|
printf("%" PRId64 ": refcount=%d nb=%" PRId64 "\n", k, refcount,
|
|
|
|
k - k1);
|
2010-04-24 00:19:47 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
|
|
|
|
int64_t pos, int size)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int growable = bs->growable;
|
|
|
|
int ret;
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_VMSTATE_SAVE);
|
2010-04-24 00:19:47 +04:00
|
|
|
bs->growable = 1;
|
2010-12-17 18:02:39 +03:00
|
|
|
ret = bdrv_pwrite(bs, qcow2_vm_state_offset(s) + pos, buf, size);
|
2010-04-24 00:19:47 +04:00
|
|
|
bs->growable = growable;
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static int qcow2_load_vmstate(BlockDriverState *bs, uint8_t *buf,
|
|
|
|
int64_t pos, int size)
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
BDRVQcowState *s = bs->opaque;
|
|
|
|
int growable = bs->growable;
|
|
|
|
int ret;
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_VMSTATE_LOAD);
|
2010-04-24 00:19:47 +04:00
|
|
|
bs->growable = 1;
|
2010-12-17 18:02:39 +03:00
|
|
|
ret = bdrv_pread(bs, qcow2_vm_state_offset(s) + pos, buf, size);
|
2010-04-24 00:19:47 +04:00
|
|
|
bs->growable = growable;
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
static QEMUOptionParameter qcow2_create_options[] = {
|
2010-04-24 00:19:47 +04:00
|
|
|
{
|
|
|
|
.name = BLOCK_OPT_SIZE,
|
|
|
|
.type = OPT_SIZE,
|
|
|
|
.help = "Virtual disk size"
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.name = BLOCK_OPT_BACKING_FILE,
|
|
|
|
.type = OPT_STRING,
|
|
|
|
.help = "File name of a base image"
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.name = BLOCK_OPT_BACKING_FMT,
|
|
|
|
.type = OPT_STRING,
|
|
|
|
.help = "Image format of the base image"
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.name = BLOCK_OPT_ENCRYPT,
|
|
|
|
.type = OPT_FLAG,
|
|
|
|
.help = "Encrypt the image"
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.name = BLOCK_OPT_CLUSTER_SIZE,
|
|
|
|
.type = OPT_SIZE,
|
2011-05-31 17:01:46 +04:00
|
|
|
.help = "qcow2 cluster size",
|
|
|
|
.value = { .n = DEFAULT_CLUSTER_SIZE },
|
2010-04-24 00:19:47 +04:00
|
|
|
},
|
|
|
|
{
|
|
|
|
.name = BLOCK_OPT_PREALLOC,
|
|
|
|
.type = OPT_STRING,
|
|
|
|
.help = "Preallocation mode (allowed values: off, metadata)"
|
|
|
|
},
|
|
|
|
{ NULL }
|
|
|
|
};
|
|
|
|
|
|
|
|
static BlockDriver bdrv_qcow2 = {
|
2010-12-17 18:02:39 +03:00
|
|
|
.format_name = "qcow2",
|
|
|
|
.instance_size = sizeof(BDRVQcowState),
|
|
|
|
.bdrv_probe = qcow2_probe,
|
|
|
|
.bdrv_open = qcow2_open,
|
|
|
|
.bdrv_close = qcow2_close,
|
|
|
|
.bdrv_create = qcow2_create,
|
2011-11-14 16:44:21 +04:00
|
|
|
.bdrv_co_is_allocated = qcow2_co_is_allocated,
|
2010-12-17 18:02:39 +03:00
|
|
|
.bdrv_set_key = qcow2_set_key,
|
|
|
|
.bdrv_make_empty = qcow2_make_empty,
|
|
|
|
|
2011-11-10 20:25:44 +04:00
|
|
|
.bdrv_co_readv = qcow2_co_readv,
|
|
|
|
.bdrv_co_writev = qcow2_co_writev,
|
2011-11-10 21:10:11 +04:00
|
|
|
.bdrv_co_flush_to_os = qcow2_co_flush_to_os,
|
|
|
|
.bdrv_co_flush_to_disk = qcow2_co_flush_to_disk,
|
2010-04-28 14:36:11 +04:00
|
|
|
|
2011-10-20 15:16:25 +04:00
|
|
|
.bdrv_co_discard = qcow2_co_discard,
|
2010-04-28 14:36:11 +04:00
|
|
|
.bdrv_truncate = qcow2_truncate,
|
2010-12-17 18:02:39 +03:00
|
|
|
.bdrv_write_compressed = qcow2_write_compressed,
|
2010-04-24 00:19:47 +04:00
|
|
|
|
|
|
|
.bdrv_snapshot_create = qcow2_snapshot_create,
|
|
|
|
.bdrv_snapshot_goto = qcow2_snapshot_goto,
|
|
|
|
.bdrv_snapshot_delete = qcow2_snapshot_delete,
|
|
|
|
.bdrv_snapshot_list = qcow2_snapshot_list,
|
2010-09-22 06:58:41 +04:00
|
|
|
.bdrv_snapshot_load_tmp = qcow2_snapshot_load_tmp,
|
2010-12-17 18:02:39 +03:00
|
|
|
.bdrv_get_info = qcow2_get_info,
|
2010-04-24 00:19:47 +04:00
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
.bdrv_save_vmstate = qcow2_save_vmstate,
|
|
|
|
.bdrv_load_vmstate = qcow2_load_vmstate,
|
2010-04-24 00:19:47 +04:00
|
|
|
|
|
|
|
.bdrv_change_backing_file = qcow2_change_backing_file,
|
|
|
|
|
2011-11-15 01:09:46 +04:00
|
|
|
.bdrv_invalidate_cache = qcow2_invalidate_cache,
|
|
|
|
|
2010-12-17 18:02:39 +03:00
|
|
|
.create_options = qcow2_create_options,
|
|
|
|
.bdrv_check = qcow2_check,
|
2010-04-24 00:19:47 +04:00
|
|
|
};
|
|
|
|
|
2009-05-10 02:03:42 +04:00
|
|
|
static void bdrv_qcow2_init(void)
|
|
|
|
{
|
|
|
|
bdrv_register(&bdrv_qcow2);
|
|
|
|
}
|
|
|
|
|
|
|
|
block_init(bdrv_qcow2_init);
|