qemu-e2k/block/qcow.c

1126 lines
35 KiB
C
Raw Normal View History

/*
* Block driver for the QCOW 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/osdep.h"
2016-03-14 09:01:28 +01:00
#include "qapi/error.h"
#include "qemu-common.h"
2016-03-21 15:11:48 +01:00
#include "qemu/error-report.h"
#include "block/block_int.h"
#include "sysemu/block-backend.h"
#include "qemu/module.h"
#include "qemu/bswap.h"
#include <zlib.h>
#include "qapi/qmp/qerror.h"
#include "qapi/qmp/qstring.h"
#include "crypto/block.h"
#include "migration/blocker.h"
#include "block/crypto.h"
/**************************************************************/
/* QEMU COW block driver with compression and encryption support */
#define QCOW_MAGIC (('Q' << 24) | ('F' << 16) | ('I' << 8) | 0xfb)
#define QCOW_VERSION 1
#define QCOW_CRYPT_NONE 0
#define QCOW_CRYPT_AES 1
#define QCOW_OFLAG_COMPRESSED (1LL << 63)
typedef struct QCowHeader {
uint32_t magic;
uint32_t version;
uint64_t backing_file_offset;
uint32_t backing_file_size;
uint32_t mtime;
uint64_t size; /* in bytes */
uint8_t cluster_bits;
uint8_t l2_bits;
uint16_t padding;
uint32_t crypt_method;
uint64_t l1_table_offset;
} QEMU_PACKED QCowHeader;
#define L2_CACHE_SIZE 16
typedef struct BDRVQcowState {
int cluster_bits;
int cluster_size;
int cluster_sectors;
int l2_bits;
int l2_size;
unsigned int l1_size;
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;
QCryptoBlock *crypto; /* Disk encryption format driver */
uint32_t crypt_method_header;
CoMutex lock;
Error *migration_blocker;
} BDRVQcowState;
static int decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset);
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 QemuOptsList qcow_runtime_opts = {
.name = "qcow",
.head = QTAILQ_HEAD_INITIALIZER(qcow_runtime_opts.head),
.desc = {
BLOCK_CRYPTO_OPT_DEF_QCOW_KEY_SECRET("encrypt."),
{ /* end of list */ }
},
};
static int qcow_open(BlockDriverState *bs, QDict *options, int flags,
Error **errp)
{
BDRVQcowState *s = bs->opaque;
unsigned int len, i, shift;
int ret;
QCowHeader header;
Error *local_err = NULL;
QCryptoBlockOpenOptions *crypto_opts = NULL;
unsigned int cflags = 0;
QDict *encryptopts = NULL;
const char *encryptfmt;
qdict_extract_subqdict(options, &encryptopts, "encrypt.");
encryptfmt = qdict_get_try_str(encryptopts, "format");
bs->file = bdrv_open_child(NULL, options, "file", bs, &child_file,
false, errp);
if (!bs->file) {
ret = -EINVAL;
goto fail;
}
ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
if (ret < 0) {
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);
be32_to_cpus(&header.mtime);
be64_to_cpus(&header.size);
be32_to_cpus(&header.crypt_method);
be64_to_cpus(&header.l1_table_offset);
if (header.magic != QCOW_MAGIC) {
error_setg(errp, "Image not in qcow format");
ret = -EINVAL;
goto fail;
}
if (header.version != QCOW_VERSION) {
error_setg(errp, "Unsupported qcow version %" PRIu32, header.version);
ret = -ENOTSUP;
goto fail;
}
if (header.size <= 1) {
error_setg(errp, "Image size is too small (must be at least 2 bytes)");
ret = -EINVAL;
goto fail;
}
if (header.cluster_bits < 9 || header.cluster_bits > 16) {
error_setg(errp, "Cluster size must be between 512 and 64k");
ret = -EINVAL;
goto fail;
}
/* l2_bits specifies number of entries; storing a uint64_t in each entry,
* so bytes = num_entries << 3. */
if (header.l2_bits < 9 - 3 || header.l2_bits > 16 - 3) {
error_setg(errp, "L2 table size must be between 512 and 64k");
ret = -EINVAL;
goto fail;
}
s->crypt_method_header = header.crypt_method;
if (s->crypt_method_header) {
2016-03-21 15:11:48 +01:00
if (bdrv_uses_whitelist() &&
s->crypt_method_header == QCOW_CRYPT_AES) {
block: drop support for using qcow[2] encryption with system emulators Back in the 2.3.0 release we declared qcow[2] encryption as deprecated, warning people that it would be removed in a future release. commit a1f688f4152e65260b94f37543521ceff8bfebe4 Author: Markus Armbruster <armbru@redhat.com> Date: Fri Mar 13 21:09:40 2015 +0100 block: Deprecate QCOW/QCOW2 encryption The code still exists today, but by a (happy?) accident we entirely broke the ability to use qcow[2] encryption in the system emulators in the 2.4.0 release due to commit 8336aafae1451d54c81dd2b187b45f7c45d2428e Author: Daniel P. Berrange <berrange@redhat.com> Date: Tue May 12 17:09:18 2015 +0100 qcow2/qcow: protect against uninitialized encryption key This commit was designed to prevent future coding bugs which might cause QEMU to read/write data on an encrypted block device in plain text mode before a decryption key is set. It turns out this preventative measure was a little too good, because we already had a long standing bug where QEMU read encrypted data in plain text mode during system emulator startup, in order to guess disk geometry: Thread 10 (Thread 0x7fffd3fff700 (LWP 30373)): #0 0x00007fffe90b1a28 in raise () at /lib64/libc.so.6 #1 0x00007fffe90b362a in abort () at /lib64/libc.so.6 #2 0x00007fffe90aa227 in __assert_fail_base () at /lib64/libc.so.6 #3 0x00007fffe90aa2d2 in () at /lib64/libc.so.6 #4 0x000055555587ae19 in qcow2_co_readv (bs=0x5555562accb0, sector_num=0, remaining_sectors=1, qiov=0x7fffffffd260) at block/qcow2.c:1229 #5 0x000055555589b60d in bdrv_aligned_preadv (bs=bs@entry=0x5555562accb0, req=req@entry=0x7fffd3ffea50, offset=offset@entry=0, bytes=bytes@entry=512, align=align@entry=512, qiov=qiov@entry=0x7fffffffd260, flags=0) at block/io.c:908 #6 0x000055555589b8bc in bdrv_co_do_preadv (bs=0x5555562accb0, offset=0, bytes=512, qiov=0x7fffffffd260, flags=<optimized out>) at block/io.c:999 #7 0x000055555589c375 in bdrv_rw_co_entry (opaque=0x7fffffffd210) at block/io.c:544 #8 0x000055555586933b in coroutine_thread (opaque=0x555557876310) at coroutine-gthread.c:134 #9 0x00007ffff64e1835 in g_thread_proxy (data=0x5555562b5590) at gthread.c:778 #10 0x00007ffff6bb760a in start_thread () at /lib64/libpthread.so.0 #11 0x00007fffe917f59d in clone () at /lib64/libc.so.6 Thread 1 (Thread 0x7ffff7ecab40 (LWP 30343)): #0 0x00007fffe91797a9 in syscall () at /lib64/libc.so.6 #1 0x00007ffff64ff87f in g_cond_wait (cond=cond@entry=0x555555e085f0 <coroutine_cond>, mutex=mutex@entry=0x555555e08600 <coroutine_lock>) at gthread-posix.c:1397 #2 0x00005555558692c3 in qemu_coroutine_switch (co=<optimized out>) at coroutine-gthread.c:117 #3 0x00005555558692c3 in qemu_coroutine_switch (from_=0x5555562b5e30, to_=to_@entry=0x555557876310, action=action@entry=COROUTINE_ENTER) at coroutine-gthread.c:175 #4 0x0000555555868a90 in qemu_coroutine_enter (co=0x555557876310, opaque=0x0) at qemu-coroutine.c:116 #5 0x0000555555859b84 in thread_pool_completion_bh (opaque=0x7fffd40010e0) at thread-pool.c:187 #6 0x0000555555859514 in aio_bh_poll (ctx=ctx@entry=0x5555562953b0) at async.c:85 #7 0x0000555555864d10 in aio_dispatch (ctx=ctx@entry=0x5555562953b0) at aio-posix.c:135 #8 0x0000555555864f75 in aio_poll (ctx=ctx@entry=0x5555562953b0, blocking=blocking@entry=true) at aio-posix.c:291 #9 0x000055555589c40d in bdrv_prwv_co (bs=bs@entry=0x5555562accb0, offset=offset@entry=0, qiov=qiov@entry=0x7fffffffd260, is_write=is_write@entry=false, flags=flags@entry=(unknown: 0)) at block/io.c:591 #10 0x000055555589c503 in bdrv_rw_co (bs=bs@entry=0x5555562accb0, sector_num=sector_num@entry=0, buf=buf@entry=0x7fffffffd2e0 "\321,", nb_sectors=nb_sectors@entry=21845, is_write=is_write@entry=false, flags=flags@entry=(unknown: 0)) at block/io.c:614 #11 0x000055555589c562 in bdrv_read_unthrottled (nb_sectors=21845, buf=0x7fffffffd2e0 "\321,", sector_num=0, bs=0x5555562accb0) at block/io.c:622 #12 0x000055555589c562 in bdrv_read_unthrottled (bs=0x5555562accb0, sector_num=sector_num@entry=0, buf=buf@entry=0x7fffffffd2e0 "\321,", nb_sectors=nb_sectors@entry=21845) at block/io.c:634 nb_sectors@entry=1) at block/block-backend.c:504 #14 0x0000555555752e9f in guess_disk_lchs (blk=blk@entry=0x5555562a5290, pcylinders=pcylinders@entry=0x7fffffffd52c, pheads=pheads@entry=0x7fffffffd530, psectors=psectors@entry=0x7fffffffd534) at hw/block/hd-geometry.c:68 #15 0x0000555555752ff7 in hd_geometry_guess (blk=0x5555562a5290, pcyls=pcyls@entry=0x555557875d1c, pheads=pheads@entry=0x555557875d20, psecs=psecs@entry=0x555557875d24, ptrans=ptrans@entry=0x555557875d28) at hw/block/hd-geometry.c:133 #16 0x0000555555752b87 in blkconf_geometry (conf=conf@entry=0x555557875d00, ptrans=ptrans@entry=0x555557875d28, cyls_max=cyls_max@entry=65536, heads_max=heads_max@entry=16, secs_max=secs_max@entry=255, errp=errp@entry=0x7fffffffd5e0) at hw/block/block.c:71 #17 0x0000555555799bc4 in ide_dev_initfn (dev=0x555557875c80, kind=IDE_HD) at hw/ide/qdev.c:174 #18 0x0000555555768394 in device_realize (dev=0x555557875c80, errp=0x7fffffffd640) at hw/core/qdev.c:247 #19 0x0000555555769a81 in device_set_realized (obj=0x555557875c80, value=<optimized out>, errp=0x7fffffffd730) at hw/core/qdev.c:1058 #20 0x00005555558240ce in property_set_bool (obj=0x555557875c80, v=<optimized out>, opaque=0x555557875de0, name=<optimized out>, errp=0x7fffffffd730) at qom/object.c:1514 #21 0x0000555555826c87 in object_property_set_qobject (obj=obj@entry=0x555557875c80, value=value@entry=0x55555784bcb0, name=name@entry=0x55555591cb3d "realized", errp=errp@entry=0x7fffffffd730) at qom/qom-qobject.c:24 #22 0x0000555555825760 in object_property_set_bool (obj=obj@entry=0x555557875c80, value=value@entry=true, name=name@entry=0x55555591cb3d "realized", errp=errp@entry=0x7fffffffd730) at qom/object.c:905 #23 0x000055555576897b in qdev_init_nofail (dev=dev@entry=0x555557875c80) at hw/core/qdev.c:380 #24 0x0000555555799ead in ide_create_drive (bus=bus@entry=0x555557629630, unit=unit@entry=0, drive=0x5555562b77e0) at hw/ide/qdev.c:122 #25 0x000055555579a746 in pci_ide_create_devs (dev=dev@entry=0x555557628db0, hd_table=hd_table@entry=0x7fffffffd830) at hw/ide/pci.c:440 #26 0x000055555579b165 in pci_piix3_ide_init (bus=<optimized out>, hd_table=0x7fffffffd830, devfn=<optimized out>) at hw/ide/piix.c:218 #27 0x000055555568ca55 in pc_init1 (machine=0x5555562960a0, pci_enabled=1, kvmclock_enabled=<optimized out>) at /home/berrange/src/virt/qemu/hw/i386/pc_piix.c:256 #28 0x0000555555603ab2 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4249 So the safety net is correctly preventing QEMU reading cipher text as if it were plain text, during startup and aborting QEMU to avoid bad usage of this data. For added fun this bug only happens if the encrypted qcow2 file happens to have data written to the first cluster, otherwise the cluster won't be allocated and so qcow2 would not try the decryption routines at all, just return all 0's. That no one even noticed, let alone reported, this bug that has shipped in 2.4.0, 2.5.0 and 2.6.0 shows that the number of actual users of encrypted qcow2 is approximately zero. So rather than fix the crash, and backport it to stable releases, just go ahead with what we have warned users about and disable any use of qcow2 encryption in the system emulators. qemu-img/qemu-io/qemu-nbd are still able to access qcow2 encrypted images for the sake of data conversion. In the future, qcow2 will gain support for the alternative luks format, but when this happens it'll be using the '-object secret' infrastructure for getting keys, which avoids this problematic scenario entirely. Signed-off-by: Daniel P. Berrange <berrange@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2016-06-13 13:30:09 +02:00
error_setg(errp,
"Use of AES-CBC encrypted qcow images is no longer "
"supported in system emulators");
error_append_hint(errp,
"You can use 'qemu-img convert' to convert your "
"image to an alternative supported format, such "
"as unencrypted qcow, or raw with the LUKS "
"format instead.\n");
ret = -ENOSYS;
goto fail;
2016-03-21 15:11:48 +01:00
}
if (s->crypt_method_header == QCOW_CRYPT_AES) {
if (encryptfmt && !g_str_equal(encryptfmt, "aes")) {
error_setg(errp,
"Header reported 'aes' encryption format but "
"options specify '%s'", encryptfmt);
ret = -EINVAL;
goto fail;
}
qdict_del(encryptopts, "format");
crypto_opts = block_crypto_open_opts_init(
Q_CRYPTO_BLOCK_FORMAT_QCOW, encryptopts, errp);
if (!crypto_opts) {
ret = -EINVAL;
goto fail;
}
2016-03-21 15:11:48 +01:00
if (flags & BDRV_O_NO_IO) {
cflags |= QCRYPTO_BLOCK_OPEN_NO_IO;
}
s->crypto = qcrypto_block_open(crypto_opts, "encrypt.",
NULL, NULL, cflags, errp);
if (!s->crypto) {
ret = -EINVAL;
goto fail;
}
} else {
error_setg(errp, "invalid encryption method in qcow header");
ret = -EINVAL;
goto fail;
}
bs->encrypted = true;
} else {
if (encryptfmt) {
error_setg(errp, "No encryption in image header, but options "
"specified format '%s'", encryptfmt);
ret = -EINVAL;
goto fail;
}
}
s->cluster_bits = header.cluster_bits;
s->cluster_size = 1 << s->cluster_bits;
s->cluster_sectors = 1 << (s->cluster_bits - 9);
s->l2_bits = header.l2_bits;
s->l2_size = 1 << s->l2_bits;
bs->total_sectors = header.size / 512;
s->cluster_offset_mask = (1LL << (63 - s->cluster_bits)) - 1;
/* read the level 1 table */
shift = s->cluster_bits + s->l2_bits;
if (header.size > UINT64_MAX - (1LL << shift)) {
error_setg(errp, "Image too large");
ret = -EINVAL;
goto fail;
} else {
uint64_t l1_size = (header.size + (1LL << shift) - 1) >> shift;
if (l1_size > INT_MAX / sizeof(uint64_t)) {
error_setg(errp, "Image too large");
ret = -EINVAL;
goto fail;
}
s->l1_size = l1_size;
}
s->l1_table_offset = header.l1_table_offset;
s->l1_table = g_try_new(uint64_t, s->l1_size);
if (s->l1_table == NULL) {
error_setg(errp, "Could not allocate memory for L1 table");
ret = -ENOMEM;
goto fail;
}
ret = bdrv_pread(bs->file, s->l1_table_offset, s->l1_table,
s->l1_size * sizeof(uint64_t));
if (ret < 0) {
goto fail;
}
for(i = 0;i < s->l1_size; i++) {
be64_to_cpus(&s->l1_table[i]);
}
/* alloc L2 cache (max. 64k * 16 * 8 = 8 MB) */
s->l2_cache =
qemu_try_blockalign(bs->file->bs,
s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));
if (s->l2_cache == NULL) {
error_setg(errp, "Could not allocate L2 table cache");
ret = -ENOMEM;
goto fail;
}
s->cluster_cache = g_malloc(s->cluster_size);
s->cluster_data = g_malloc(s->cluster_size);
s->cluster_cache_offset = -1;
/* read the backing file name */
if (header.backing_file_offset != 0) {
len = header.backing_file_size;
if (len > 1023 || len >= sizeof(bs->backing_file)) {
error_setg(errp, "Backing file name too long");
ret = -EINVAL;
goto fail;
}
ret = bdrv_pread(bs->file, header.backing_file_offset,
bs->backing_file, len);
if (ret < 0) {
goto fail;
}
bs->backing_file[len] = '\0';
}
/* Disable migration when qcow images are used */
error_setg(&s->migration_blocker, "The qcow format used by node '%s' "
"does not support live migration",
bdrv_get_device_or_node_name(bs));
ret = migrate_add_blocker(s->migration_blocker, &local_err);
if (local_err) {
error_propagate(errp, local_err);
error_free(s->migration_blocker);
goto fail;
}
QDECREF(encryptopts);
qapi_free_QCryptoBlockOpenOptions(crypto_opts);
qemu_co_mutex_init(&s->lock);
return 0;
fail:
g_free(s->l1_table);
qemu_vfree(s->l2_cache);
g_free(s->cluster_cache);
g_free(s->cluster_data);
qcrypto_block_free(s->crypto);
QDECREF(encryptopts);
qapi_free_QCryptoBlockOpenOptions(crypto_opts);
return ret;
}
/* We have nothing to do for QCOW reopen, stubs just return
* success */
static int qcow_reopen_prepare(BDRVReopenState *state,
BlockReopenQueue *queue, Error **errp)
{
return 0;
}
/* 'allocate' is:
*
* 0 to not allocate.
*
* 1 to allocate a normal cluster (for sector indexes 'n_start' to
* 'n_end')
*
* 2 to allocate a compressed cluster of size
* 'compressed_size'. 'compressed_size' must be > 0 and <
* cluster_size
*
* return 0 if not allocated, 1 if *result is assigned, and negative
* errno on failure.
*/
static int get_cluster_offset(BlockDriverState *bs,
uint64_t offset, int allocate,
int compressed_size,
int n_start, int n_end, uint64_t *result)
{
BDRVQcowState *s = bs->opaque;
int min_index, i, j, l1_index, l2_index, ret;
int64_t l2_offset;
uint64_t *l2_table, cluster_offset, tmp;
uint32_t min_count;
int new_l2_table;
*result = 0;
l1_index = offset >> (s->l2_bits + s->cluster_bits);
l2_offset = s->l1_table[l1_index];
new_l2_table = 0;
if (!l2_offset) {
if (!allocate)
return 0;
/* allocate a new l2 entry */
l2_offset = bdrv_getlength(bs->file->bs);
if (l2_offset < 0) {
return l2_offset;
}
/* round to cluster size */
l2_offset = QEMU_ALIGN_UP(l2_offset, s->cluster_size);
/* update the L1 entry */
s->l1_table[l1_index] = l2_offset;
tmp = cpu_to_be64(l2_offset);
ret = bdrv_pwrite_sync(bs->file,
s->l1_table_offset + l1_index * sizeof(tmp),
&tmp, sizeof(tmp));
if (ret < 0) {
return ret;
}
new_l2_table = 1;
}
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;
}
}
l2_table = s->l2_cache + (i << s->l2_bits);
goto found;
}
}
/* not found: load 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;
}
}
l2_table = s->l2_cache + (min_index << s->l2_bits);
if (new_l2_table) {
memset(l2_table, 0, s->l2_size * sizeof(uint64_t));
ret = bdrv_pwrite_sync(bs->file, l2_offset, l2_table,
s->l2_size * sizeof(uint64_t));
if (ret < 0) {
return ret;
}
} else {
ret = bdrv_pread(bs->file, l2_offset, l2_table,
s->l2_size * sizeof(uint64_t));
if (ret < 0) {
return ret;
}
}
s->l2_cache_offsets[min_index] = l2_offset;
s->l2_cache_counts[min_index] = 1;
found:
l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
cluster_offset = be64_to_cpu(l2_table[l2_index]);
if (!cluster_offset ||
((cluster_offset & QCOW_OFLAG_COMPRESSED) && allocate == 1)) {
if (!allocate)
return 0;
/* allocate a new cluster */
if ((cluster_offset & QCOW_OFLAG_COMPRESSED) &&
(n_end - n_start) < s->cluster_sectors) {
/* if the cluster is already compressed, we must
decompress it in the case it is not completely
overwritten */
if (decompress_cluster(bs, cluster_offset) < 0) {
return -EIO;
}
cluster_offset = bdrv_getlength(bs->file->bs);
if ((int64_t) cluster_offset < 0) {
return cluster_offset;
}
cluster_offset = QEMU_ALIGN_UP(cluster_offset, s->cluster_size);
/* write the cluster content */
ret = bdrv_pwrite(bs->file, cluster_offset, s->cluster_cache,
s->cluster_size);
if (ret < 0) {
return ret;
}
} else {
cluster_offset = bdrv_getlength(bs->file->bs);
if ((int64_t) cluster_offset < 0) {
return cluster_offset;
}
if (allocate == 1) {
/* round to cluster size */
cluster_offset = QEMU_ALIGN_UP(cluster_offset, s->cluster_size);
if (cluster_offset + s->cluster_size > INT64_MAX) {
return -E2BIG;
}
ret = bdrv_truncate(bs->file, cluster_offset + s->cluster_size,
PREALLOC_MODE_OFF, NULL);
if (ret < 0) {
return ret;
}
/* if encrypted, we must initialize the cluster
content which won't be written */
if (bs->encrypted &&
(n_end - n_start) < s->cluster_sectors) {
uint64_t start_sect;
assert(s->crypto);
start_sect = (offset & ~(s->cluster_size - 1)) >> 9;
for(i = 0; i < s->cluster_sectors; i++) {
if (i < n_start || i >= n_end) {
memset(s->cluster_data, 0x00, 512);
if (qcrypto_block_encrypt(s->crypto, start_sect + i,
s->cluster_data,
BDRV_SECTOR_SIZE,
NULL) < 0) {
return -EIO;
}
ret = bdrv_pwrite(bs->file,
cluster_offset + i * 512,
s->cluster_data, 512);
if (ret < 0) {
return ret;
}
}
}
}
} else if (allocate == 2) {
cluster_offset |= QCOW_OFLAG_COMPRESSED |
(uint64_t)compressed_size << (63 - s->cluster_bits);
}
}
/* update L2 table */
tmp = cpu_to_be64(cluster_offset);
l2_table[l2_index] = tmp;
ret = bdrv_pwrite_sync(bs->file, l2_offset + l2_index * sizeof(tmp),
&tmp, sizeof(tmp));
if (ret < 0) {
return ret;
}
}
*result = cluster_offset;
return 1;
}
static int64_t coroutine_fn qcow_co_get_block_status(BlockDriverState *bs,
int64_t sector_num, int nb_sectors, int *pnum, BlockDriverState **file)
{
BDRVQcowState *s = bs->opaque;
int index_in_cluster, n, ret;
uint64_t cluster_offset;
qemu_co_mutex_lock(&s->lock);
ret = get_cluster_offset(bs, sector_num << 9, 0, 0, 0, 0, &cluster_offset);
qemu_co_mutex_unlock(&s->lock);
if (ret < 0) {
return ret;
}
index_in_cluster = sector_num & (s->cluster_sectors - 1);
n = s->cluster_sectors - index_in_cluster;
if (n > nb_sectors)
n = nb_sectors;
*pnum = n;
if (!cluster_offset) {
return 0;
}
if ((cluster_offset & QCOW_OFLAG_COMPRESSED) || s->crypto) {
return BDRV_BLOCK_DATA;
}
cluster_offset |= (index_in_cluster << BDRV_SECTOR_BITS);
*file = bs->file->bs;
return BDRV_BLOCK_DATA | BDRV_BLOCK_OFFSET_VALID | cluster_offset;
}
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;
}
static int decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
{
BDRVQcowState *s = bs->opaque;
int ret, csize;
uint64_t coffset;
coffset = cluster_offset & s->cluster_offset_mask;
if (s->cluster_cache_offset != coffset) {
csize = cluster_offset >> (63 - s->cluster_bits);
csize &= (s->cluster_size - 1);
ret = bdrv_pread(bs->file, coffset, s->cluster_data, csize);
if (ret != csize)
return -1;
if (decompress_buffer(s->cluster_cache, s->cluster_size,
s->cluster_data, csize) < 0) {
return -1;
}
s->cluster_cache_offset = coffset;
}
return 0;
}
static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num,
int nb_sectors, QEMUIOVector *qiov)
{
BDRVQcowState *s = bs->opaque;
int index_in_cluster;
int ret = 0, n;
uint64_t cluster_offset;
struct iovec hd_iov;
QEMUIOVector hd_qiov;
uint8_t *buf;
void *orig_buf;
if (qiov->niov > 1) {
buf = orig_buf = qemu_try_blockalign(bs, qiov->size);
if (buf == NULL) {
return -ENOMEM;
}
} else {
orig_buf = NULL;
buf = (uint8_t *)qiov->iov->iov_base;
}
qemu_co_mutex_lock(&s->lock);
while (nb_sectors != 0) {
/* prepare next request */
ret = get_cluster_offset(bs, sector_num << 9,
0, 0, 0, 0, &cluster_offset);
if (ret < 0) {
break;
}
index_in_cluster = sector_num & (s->cluster_sectors - 1);
n = s->cluster_sectors - index_in_cluster;
if (n > nb_sectors) {
n = nb_sectors;
}
if (!cluster_offset) {
if (bs->backing) {
/* read from the base image */
hd_iov.iov_base = (void *)buf;
hd_iov.iov_len = n * 512;
qemu_iovec_init_external(&hd_qiov, &hd_iov, 1);
qemu_co_mutex_unlock(&s->lock);
ret = bdrv_co_readv(bs->backing, sector_num, n, &hd_qiov);
qemu_co_mutex_lock(&s->lock);
if (ret < 0) {
break;
}
} else {
/* Note: in this case, no need to wait */
memset(buf, 0, 512 * n);
}
} else if (cluster_offset & QCOW_OFLAG_COMPRESSED) {
/* add AIO support for compressed blocks ? */
if (decompress_cluster(bs, cluster_offset) < 0) {
ret = -EIO;
break;
}
memcpy(buf,
s->cluster_cache + index_in_cluster * 512, 512 * n);
} else {
if ((cluster_offset & 511) != 0) {
ret = -EIO;
break;
}
hd_iov.iov_base = (void *)buf;
hd_iov.iov_len = n * 512;
qemu_iovec_init_external(&hd_qiov, &hd_iov, 1);
qemu_co_mutex_unlock(&s->lock);
ret = bdrv_co_readv(bs->file,
(cluster_offset >> 9) + index_in_cluster,
n, &hd_qiov);
qemu_co_mutex_lock(&s->lock);
if (ret < 0) {
break;
}
if (bs->encrypted) {
assert(s->crypto);
if (qcrypto_block_decrypt(s->crypto, sector_num, buf,
n * BDRV_SECTOR_SIZE, NULL) < 0) {
ret = -EIO;
break;
}
}
}
ret = 0;
nb_sectors -= n;
sector_num += n;
buf += n * 512;
}
qemu_co_mutex_unlock(&s->lock);
if (qiov->niov > 1) {
qemu_iovec_from_buf(qiov, 0, orig_buf, qiov->size);
qemu_vfree(orig_buf);
}
return ret;
}
static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num,
int nb_sectors, QEMUIOVector *qiov)
{
BDRVQcowState *s = bs->opaque;
int index_in_cluster;
uint64_t cluster_offset;
int ret = 0, n;
struct iovec hd_iov;
QEMUIOVector hd_qiov;
uint8_t *buf;
void *orig_buf;
s->cluster_cache_offset = -1; /* disable compressed cache */
/* We must always copy the iov when encrypting, so we
* don't modify the original data buffer during encryption */
if (bs->encrypted || qiov->niov > 1) {
buf = orig_buf = qemu_try_blockalign(bs, qiov->size);
if (buf == NULL) {
return -ENOMEM;
}
qemu_iovec_to_buf(qiov, 0, buf, qiov->size);
} else {
orig_buf = NULL;
buf = (uint8_t *)qiov->iov->iov_base;
}
qemu_co_mutex_lock(&s->lock);
while (nb_sectors != 0) {
index_in_cluster = sector_num & (s->cluster_sectors - 1);
n = s->cluster_sectors - index_in_cluster;
if (n > nb_sectors) {
n = nb_sectors;
}
ret = get_cluster_offset(bs, sector_num << 9, 1, 0,
index_in_cluster,
index_in_cluster + n, &cluster_offset);
if (ret < 0) {
break;
}
if (!cluster_offset || (cluster_offset & 511) != 0) {
ret = -EIO;
break;
}
if (bs->encrypted) {
assert(s->crypto);
if (qcrypto_block_encrypt(s->crypto, sector_num, buf,
n * BDRV_SECTOR_SIZE, NULL) < 0) {
ret = -EIO;
break;
}
}
hd_iov.iov_base = (void *)buf;
hd_iov.iov_len = n * 512;
qemu_iovec_init_external(&hd_qiov, &hd_iov, 1);
qemu_co_mutex_unlock(&s->lock);
ret = bdrv_co_writev(bs->file,
(cluster_offset >> 9) + index_in_cluster,
n, &hd_qiov);
qemu_co_mutex_lock(&s->lock);
if (ret < 0) {
break;
}
ret = 0;
nb_sectors -= n;
sector_num += n;
buf += n * 512;
}
qemu_co_mutex_unlock(&s->lock);
qemu_vfree(orig_buf);
return ret;
}
static void qcow_close(BlockDriverState *bs)
{
BDRVQcowState *s = bs->opaque;
qcrypto_block_free(s->crypto);
s->crypto = NULL;
g_free(s->l1_table);
qemu_vfree(s->l2_cache);
g_free(s->cluster_cache);
g_free(s->cluster_data);
migrate_del_blocker(s->migration_blocker);
error_free(s->migration_blocker);
}
static int qcow_create(const char *filename, QemuOpts *opts, Error **errp)
{
int header_size, backing_filename_len, l1_size, shift, i;
QCowHeader header;
uint8_t *tmp;
int64_t total_size = 0;
char *backing_file = NULL;
Error *local_err = NULL;
int ret;
BlockBackend *qcow_blk;
char *encryptfmt = NULL;
QDict *options;
QDict *encryptopts = NULL;
QCryptoBlockCreateOptions *crypto_opts = NULL;
QCryptoBlock *crypto = NULL;
/* Read out options */
total_size = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
BDRV_SECTOR_SIZE);
if (total_size == 0) {
error_setg(errp, "Image size is too small, cannot be zero length");
ret = -EINVAL;
goto cleanup;
}
backing_file = qemu_opt_get_del(opts, BLOCK_OPT_BACKING_FILE);
encryptfmt = qemu_opt_get_del(opts, BLOCK_OPT_ENCRYPT_FORMAT);
if (encryptfmt) {
if (qemu_opt_get(opts, BLOCK_OPT_ENCRYPT)) {
error_setg(errp, "Options " BLOCK_OPT_ENCRYPT " and "
BLOCK_OPT_ENCRYPT_FORMAT " are mutually exclusive");
ret = -EINVAL;
goto cleanup;
}
} else if (qemu_opt_get_bool_del(opts, BLOCK_OPT_ENCRYPT, false)) {
encryptfmt = g_strdup("aes");
}
ret = bdrv_create_file(filename, opts, &local_err);
if (ret < 0) {
error_propagate(errp, local_err);
goto cleanup;
}
qcow_blk = blk_new_open(filename, NULL, NULL,
BDRV_O_RDWR | BDRV_O_RESIZE | BDRV_O_PROTOCOL,
&local_err);
if (qcow_blk == NULL) {
error_propagate(errp, local_err);
ret = -EIO;
goto cleanup;
}
blk_set_allow_write_beyond_eof(qcow_blk, true);
ret = blk_truncate(qcow_blk, 0, PREALLOC_MODE_OFF, errp);
if (ret < 0) {
goto exit;
}
memset(&header, 0, sizeof(header));
header.magic = cpu_to_be32(QCOW_MAGIC);
header.version = cpu_to_be32(QCOW_VERSION);
header.size = cpu_to_be64(total_size);
header_size = sizeof(header);
backing_filename_len = 0;
if (backing_file) {
if (strcmp(backing_file, "fat:")) {
header.backing_file_offset = cpu_to_be64(header_size);
backing_filename_len = strlen(backing_file);
header.backing_file_size = cpu_to_be32(backing_filename_len);
header_size += backing_filename_len;
} else {
/* special backing file for vvfat */
g_free(backing_file);
backing_file = NULL;
}
header.cluster_bits = 9; /* 512 byte cluster to avoid copying
unmodified sectors */
header.l2_bits = 12; /* 32 KB L2 tables */
} else {
header.cluster_bits = 12; /* 4 KB clusters */
header.l2_bits = 9; /* 4 KB L2 tables */
}
header_size = (header_size + 7) & ~7;
shift = header.cluster_bits + header.l2_bits;
l1_size = (total_size + (1LL << shift) - 1) >> shift;
header.l1_table_offset = cpu_to_be64(header_size);
options = qemu_opts_to_qdict(opts, NULL);
qdict_extract_subqdict(options, &encryptopts, "encrypt.");
QDECREF(options);
if (encryptfmt) {
if (!g_str_equal(encryptfmt, "aes")) {
error_setg(errp, "Unknown encryption format '%s', expected 'aes'",
encryptfmt);
ret = -EINVAL;
goto exit;
}
header.crypt_method = cpu_to_be32(QCOW_CRYPT_AES);
crypto_opts = block_crypto_create_opts_init(
Q_CRYPTO_BLOCK_FORMAT_QCOW, encryptopts, errp);
if (!crypto_opts) {
ret = -EINVAL;
goto exit;
}
crypto = qcrypto_block_create(crypto_opts, "encrypt.",
NULL, NULL, NULL, errp);
if (!crypto) {
ret = -EINVAL;
goto exit;
}
} else {
header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
}
/* write all the data */
ret = blk_pwrite(qcow_blk, 0, &header, sizeof(header), 0);
if (ret != sizeof(header)) {
goto exit;
}
if (backing_file) {
ret = blk_pwrite(qcow_blk, sizeof(header),
backing_file, backing_filename_len, 0);
if (ret != backing_filename_len) {
goto exit;
}
}
tmp = g_malloc0(BDRV_SECTOR_SIZE);
for (i = 0; i < DIV_ROUND_UP(sizeof(uint64_t) * l1_size, BDRV_SECTOR_SIZE);
i++) {
ret = blk_pwrite(qcow_blk, header_size + BDRV_SECTOR_SIZE * i,
tmp, BDRV_SECTOR_SIZE, 0);
if (ret != BDRV_SECTOR_SIZE) {
g_free(tmp);
goto exit;
}
}
g_free(tmp);
ret = 0;
exit:
blk_unref(qcow_blk);
cleanup:
QDECREF(encryptopts);
g_free(encryptfmt);
qcrypto_block_free(crypto);
qapi_free_QCryptoBlockCreateOptions(crypto_opts);
g_free(backing_file);
return ret;
}
static int qcow_make_empty(BlockDriverState *bs)
{
BDRVQcowState *s = bs->opaque;
uint32_t l1_length = s->l1_size * sizeof(uint64_t);
int ret;
memset(s->l1_table, 0, l1_length);
if (bdrv_pwrite_sync(bs->file, s->l1_table_offset, s->l1_table,
l1_length) < 0)
return -1;
ret = bdrv_truncate(bs->file, s->l1_table_offset + l1_length,
PREALLOC_MODE_OFF, NULL);
if (ret < 0)
return ret;
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));
return 0;
}
/* XXX: put compressed sectors first, then all the cluster aligned
tables to avoid losing bytes in alignment */
static coroutine_fn int
qcow_co_pwritev_compressed(BlockDriverState *bs, uint64_t offset,
uint64_t bytes, QEMUIOVector *qiov)
{
BDRVQcowState *s = bs->opaque;
QEMUIOVector hd_qiov;
struct iovec iov;
z_stream strm;
int ret, out_len;
uint8_t *buf, *out_buf;
uint64_t cluster_offset;
buf = qemu_blockalign(bs, s->cluster_size);
if (bytes != s->cluster_size) {
if (bytes > s->cluster_size ||
offset + bytes != bs->total_sectors << BDRV_SECTOR_BITS)
{
qemu_vfree(buf);
return -EINVAL;
}
/* Zero-pad last write if image size is not cluster aligned */
memset(buf + bytes, 0, s->cluster_size - bytes);
}
qemu_iovec_to_buf(qiov, 0, buf, qiov->size);
out_buf = g_malloc(s->cluster_size);
/* 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) {
ret = -EINVAL;
goto fail;
}
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);
ret = -EINVAL;
goto fail;
}
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 */
ret = qcow_co_writev(bs, offset >> BDRV_SECTOR_BITS,
bytes >> BDRV_SECTOR_BITS, qiov);
if (ret < 0) {
goto fail;
}
goto success;
}
qemu_co_mutex_lock(&s->lock);
ret = get_cluster_offset(bs, offset, 2, out_len, 0, 0, &cluster_offset);
qemu_co_mutex_unlock(&s->lock);
if (ret < 0) {
goto fail;
}
if (cluster_offset == 0) {
ret = -EIO;
goto fail;
}
cluster_offset &= s->cluster_offset_mask;
iov = (struct iovec) {
.iov_base = out_buf,
.iov_len = out_len,
};
qemu_iovec_init_external(&hd_qiov, &iov, 1);
ret = bdrv_co_pwritev(bs->file, cluster_offset, out_len, &hd_qiov, 0);
if (ret < 0) {
goto fail;
}
success:
ret = 0;
fail:
qemu_vfree(buf);
g_free(out_buf);
return ret;
}
static int qcow_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
BDRVQcowState *s = bs->opaque;
bdi->cluster_size = s->cluster_size;
return 0;
}
static QemuOptsList qcow_create_opts = {
.name = "qcow-create-opts",
.head = QTAILQ_HEAD_INITIALIZER(qcow_create_opts.head),
.desc = {
{
.name = BLOCK_OPT_SIZE,
.type = QEMU_OPT_SIZE,
.help = "Virtual disk size"
},
{
.name = BLOCK_OPT_BACKING_FILE,
.type = QEMU_OPT_STRING,
.help = "File name of a base image"
},
{
.name = BLOCK_OPT_ENCRYPT,
.type = QEMU_OPT_BOOL,
.help = "Encrypt the image with format 'aes'. (Deprecated "
"in favor of " BLOCK_OPT_ENCRYPT_FORMAT "=aes)",
},
{
.name = BLOCK_OPT_ENCRYPT_FORMAT,
.type = QEMU_OPT_STRING,
.help = "Encrypt the image, format choices: 'aes'",
},
BLOCK_CRYPTO_OPT_DEF_QCOW_KEY_SECRET("encrypt."),
{ /* end of list */ }
}
};
static BlockDriver bdrv_qcow = {
.format_name = "qcow",
.instance_size = sizeof(BDRVQcowState),
.bdrv_probe = qcow_probe,
.bdrv_open = qcow_open,
.bdrv_close = qcow_close,
.bdrv_child_perm = bdrv_format_default_perms,
.bdrv_reopen_prepare = qcow_reopen_prepare,
.bdrv_create = qcow_create,
.bdrv_has_zero_init = bdrv_has_zero_init_1,
.supports_backing = true,
.bdrv_co_readv = qcow_co_readv,
.bdrv_co_writev = qcow_co_writev,
.bdrv_co_get_block_status = qcow_co_get_block_status,
.bdrv_make_empty = qcow_make_empty,
.bdrv_co_pwritev_compressed = qcow_co_pwritev_compressed,
.bdrv_get_info = qcow_get_info,
.create_opts = &qcow_create_opts,
};
static void bdrv_qcow_init(void)
{
bdrv_register(&bdrv_qcow);
}
block_init(bdrv_qcow_init);