2009-05-28 18:07:05 +04:00
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
2016-01-18 21:01:42 +03:00
|
|
|
#include "qemu/osdep.h"
|
2009-05-28 18:07:05 +04:00
|
|
|
#include <zlib.h>
|
|
|
|
|
include/qemu/osdep.h: Don't include qapi/error.h
Commit 57cb38b included qapi/error.h into qemu/osdep.h to get the
Error typedef. Since then, we've moved to include qemu/osdep.h
everywhere. Its file comment explains: "To avoid getting into
possible circular include dependencies, this file should not include
any other QEMU headers, with the exceptions of config-host.h,
compiler.h, os-posix.h and os-win32.h, all of which are doing a
similar job to this file and are under similar constraints."
qapi/error.h doesn't do a similar job, and it doesn't adhere to
similar constraints: it includes qapi-types.h. That's in excess of
100KiB of crap most .c files don't actually need.
Add the typedef to qemu/typedefs.h, and include that instead of
qapi/error.h. Include qapi/error.h in .c files that need it and don't
get it now. Include qapi-types.h in qom/object.h for uint16List.
Update scripts/clean-includes accordingly. Update it further to match
reality: replace config.h by config-target.h, add sysemu/os-posix.h,
sysemu/os-win32.h. Update the list of includes in the qemu/osdep.h
comment quoted above similarly.
This reduces the number of objects depending on qapi/error.h from "all
of them" to less than a third. Unfortunately, the number depending on
qapi-types.h shrinks only a little. More work is needed for that one.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
[Fix compilation without the spice devel packages. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2016-03-14 11:01:28 +03:00
|
|
|
#include "qapi/error.h"
|
2009-05-28 18:07:05 +04:00
|
|
|
#include "qemu-common.h"
|
2012-12-17 21:19:44 +04:00
|
|
|
#include "block/block_int.h"
|
2009-05-28 18:07:05 +04:00
|
|
|
#include "block/qcow2.h"
|
2016-03-15 19:22:36 +03:00
|
|
|
#include "qemu/bswap.h"
|
2012-03-01 21:36:21 +04:00
|
|
|
#include "trace.h"
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2013-05-14 18:14:33 +04:00
|
|
|
int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
|
|
|
|
bool exact_size)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-05-14 18:14:33 +04:00
|
|
|
int new_l1_size2, ret, i;
|
2009-05-28 18:07:05 +04:00
|
|
|
uint64_t *new_l1_table;
|
2013-09-30 19:57:21 +04:00
|
|
|
int64_t old_l1_table_offset, old_l1_size;
|
2013-05-14 18:14:33 +04:00
|
|
|
int64_t new_l1_table_offset, new_l1_size;
|
2009-05-28 18:07:05 +04:00
|
|
|
uint8_t data[12];
|
|
|
|
|
2010-10-18 19:53:53 +04:00
|
|
|
if (min_size <= s->l1_size)
|
2009-05-28 18:07:05 +04:00
|
|
|
return 0;
|
2010-10-18 19:53:53 +04:00
|
|
|
|
qcow2: Check min_size in qcow2_grow_l1_table()
First, new_l1_size is an int64_t, whereas min_size is a uint64_t.
Therefore, during the loop which adjusts new_l1_size until it equals or
exceeds min_size, new_l1_size might overflow and become negative. The
comparison in the loop condition however will take it as an unsigned
value (because min_size is unsigned) and therefore recognize it as
exceeding min_size. Therefore, the loop is left with a negative
new_l1_size, which is not correct. This could be fixed by making
new_l1_size uint64_t.
On the other hand, however, by doing this, the while loop may take
forever. If min_size is e.g. UINT64_MAX, it will take new_l1_size
probably multiple overflows to reach the exact same value (if it reaches
it at all). Then, right after the loop, new_l1_size will be recognized
as being too big anyway.
Both problems require a ridiculously high min_size value, which is very
unlikely to occur; but both problems are also simply avoided by checking
whether min_size is sane before calculating new_l1_size (which should
still be checked separately, though).
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2014-04-29 21:03:14 +04:00
|
|
|
/* Do a sanity check on min_size before trying to calculate new_l1_size
|
|
|
|
* (this prevents overflows during the while loop for the calculation of
|
|
|
|
* new_l1_size) */
|
|
|
|
if (min_size > INT_MAX / sizeof(uint64_t)) {
|
|
|
|
return -EFBIG;
|
|
|
|
}
|
|
|
|
|
2010-10-18 19:53:53 +04:00
|
|
|
if (exact_size) {
|
|
|
|
new_l1_size = min_size;
|
|
|
|
} else {
|
|
|
|
/* Bump size up to reduce the number of times we have to grow */
|
|
|
|
new_l1_size = s->l1_size;
|
|
|
|
if (new_l1_size == 0) {
|
|
|
|
new_l1_size = 1;
|
|
|
|
}
|
|
|
|
while (min_size > new_l1_size) {
|
|
|
|
new_l1_size = (new_l1_size * 3 + 1) / 2;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
2010-10-18 19:53:53 +04:00
|
|
|
|
2016-06-15 18:36:30 +03:00
|
|
|
QEMU_BUILD_BUG_ON(QCOW_MAX_L1_SIZE > INT_MAX);
|
|
|
|
if (new_l1_size > QCOW_MAX_L1_SIZE / sizeof(uint64_t)) {
|
2013-05-14 18:14:33 +04:00
|
|
|
return -EFBIG;
|
|
|
|
}
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
#ifdef DEBUG_ALLOC2
|
2013-05-14 18:14:33 +04:00
|
|
|
fprintf(stderr, "grow l1_table from %d to %" PRId64 "\n",
|
|
|
|
s->l1_size, new_l1_size);
|
2009-05-28 18:07:05 +04:00
|
|
|
#endif
|
|
|
|
|
|
|
|
new_l1_size2 = sizeof(uint64_t) * new_l1_size;
|
2015-06-16 15:19:22 +03:00
|
|
|
new_l1_table = qemu_try_blockalign(bs->file->bs,
|
2014-05-20 19:12:47 +04:00
|
|
|
align_offset(new_l1_size2, 512));
|
|
|
|
if (new_l1_table == NULL) {
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
memset(new_l1_table, 0, align_offset(new_l1_size2, 512));
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
memcpy(new_l1_table, s->l1_table, s->l1_size * sizeof(uint64_t));
|
|
|
|
|
|
|
|
/* write new table (align to cluster) */
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ALLOC_TABLE);
|
2009-05-28 18:07:07 +04:00
|
|
|
new_l1_table_offset = qcow2_alloc_clusters(bs, new_l1_size2);
|
2010-01-20 17:04:01 +03:00
|
|
|
if (new_l1_table_offset < 0) {
|
2014-05-20 19:12:47 +04:00
|
|
|
qemu_vfree(new_l1_table);
|
2010-01-20 17:04:01 +03:00
|
|
|
return new_l1_table_offset;
|
|
|
|
}
|
2011-01-10 19:17:28 +03:00
|
|
|
|
|
|
|
ret = qcow2_cache_flush(bs, s->refcount_block_cache);
|
|
|
|
if (ret < 0) {
|
2011-06-01 12:50:00 +04:00
|
|
|
goto fail;
|
2011-01-10 19:17:28 +03:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2013-08-30 16:34:26 +04:00
|
|
|
/* the L1 position has not yet been updated, so these clusters must
|
|
|
|
* indeed be completely free */
|
2013-10-10 13:09:23 +04:00
|
|
|
ret = qcow2_pre_write_overlap_check(bs, 0, new_l1_table_offset,
|
|
|
|
new_l1_size2);
|
2013-08-30 16:34:26 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_WRITE_TABLE);
|
2009-05-28 18:07:05 +04:00
|
|
|
for(i = 0; i < s->l1_size; i++)
|
|
|
|
new_l1_table[i] = cpu_to_be64(new_l1_table[i]);
|
2016-06-20 21:09:15 +03:00
|
|
|
ret = bdrv_pwrite_sync(bs->file, new_l1_table_offset,
|
2015-06-16 15:19:22 +03:00
|
|
|
new_l1_table, new_l1_size2);
|
2010-06-16 19:44:35 +04:00
|
|
|
if (ret < 0)
|
2009-05-28 18:07:05 +04:00
|
|
|
goto fail;
|
|
|
|
for(i = 0; i < s->l1_size; i++)
|
|
|
|
new_l1_table[i] = be64_to_cpu(new_l1_table[i]);
|
|
|
|
|
|
|
|
/* set new table */
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ACTIVATE_TABLE);
|
2016-06-16 19:06:17 +03:00
|
|
|
stl_be_p(data, new_l1_size);
|
2013-11-05 20:38:36 +04:00
|
|
|
stq_be_p(data + 4, new_l1_table_offset);
|
2016-06-20 21:09:15 +03:00
|
|
|
ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, l1_size),
|
2015-06-16 15:19:22 +03:00
|
|
|
data, sizeof(data));
|
2010-06-16 19:44:35 +04:00
|
|
|
if (ret < 0) {
|
2009-05-28 18:07:05 +04:00
|
|
|
goto fail;
|
2010-01-20 17:02:58 +03:00
|
|
|
}
|
2014-05-20 19:12:47 +04:00
|
|
|
qemu_vfree(s->l1_table);
|
2013-09-30 19:57:21 +04:00
|
|
|
old_l1_table_offset = s->l1_table_offset;
|
2009-05-28 18:07:05 +04:00
|
|
|
s->l1_table_offset = new_l1_table_offset;
|
|
|
|
s->l1_table = new_l1_table;
|
2013-09-30 19:57:21 +04:00
|
|
|
old_l1_size = s->l1_size;
|
2009-05-28 18:07:05 +04:00
|
|
|
s->l1_size = new_l1_size;
|
2013-09-30 19:57:21 +04:00
|
|
|
qcow2_free_clusters(bs, old_l1_table_offset, old_l1_size * sizeof(uint64_t),
|
|
|
|
QCOW2_DISCARD_OTHER);
|
2009-05-28 18:07:05 +04:00
|
|
|
return 0;
|
|
|
|
fail:
|
2014-05-20 19:12:47 +04:00
|
|
|
qemu_vfree(new_l1_table);
|
2013-06-19 15:44:18 +04:00
|
|
|
qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2,
|
|
|
|
QCOW2_DISCARD_OTHER);
|
2010-06-16 19:44:35 +04:00
|
|
|
return ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
2010-05-21 20:25:20 +04:00
|
|
|
static int l2_load(BlockDriverState *bs, uint64_t l2_offset,
|
|
|
|
uint64_t **l2_table)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-06-14 00:57:58 +03:00
|
|
|
return qcow2_cache_get(bs, s->l2_table_cache, l2_offset,
|
|
|
|
(void **)l2_table);
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2009-06-16 13:31:28 +04:00
|
|
|
/*
|
|
|
|
* Writes one sector of the L1 table to the disk (can't update single entries
|
|
|
|
* and we really don't want bdrv_pread to perform a read-modify-write)
|
|
|
|
*/
|
|
|
|
#define L1_ENTRIES_PER_SECTOR (512 / 8)
|
2013-08-30 16:34:28 +04:00
|
|
|
int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index)
|
2009-06-16 13:31:28 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2014-10-16 17:25:56 +04:00
|
|
|
uint64_t buf[L1_ENTRIES_PER_SECTOR] = { 0 };
|
2009-06-16 13:31:28 +04:00
|
|
|
int l1_start_index;
|
2010-03-23 19:28:22 +03:00
|
|
|
int i, ret;
|
2009-06-16 13:31:28 +04:00
|
|
|
|
|
|
|
l1_start_index = l1_index & ~(L1_ENTRIES_PER_SECTOR - 1);
|
2014-10-16 17:25:56 +04:00
|
|
|
for (i = 0; i < L1_ENTRIES_PER_SECTOR && l1_start_index + i < s->l1_size;
|
|
|
|
i++)
|
|
|
|
{
|
2009-06-16 13:31:28 +04:00
|
|
|
buf[i] = cpu_to_be64(s->l1_table[l1_start_index + i]);
|
|
|
|
}
|
|
|
|
|
2013-10-10 13:09:23 +04:00
|
|
|
ret = qcow2_pre_write_overlap_check(bs, QCOW2_OL_ACTIVE_L1,
|
2013-08-30 16:34:26 +04:00
|
|
|
s->l1_table_offset + 8 * l1_start_index, sizeof(buf));
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L1_UPDATE);
|
2016-06-20 21:09:15 +03:00
|
|
|
ret = bdrv_pwrite_sync(bs->file,
|
2015-06-16 15:19:22 +03:00
|
|
|
s->l1_table_offset + 8 * l1_start_index,
|
|
|
|
buf, sizeof(buf));
|
2010-03-23 19:28:22 +03:00
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
2009-06-16 13:31:28 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
/*
|
|
|
|
* l2_allocate
|
|
|
|
*
|
|
|
|
* Allocate a new l2 entry in the file. If l1_index points to an already
|
|
|
|
* used entry in the L2 table (i.e. we are doing a copy on write for the L2
|
|
|
|
* table) copy the contents of the old L2 table into the newly allocated one.
|
|
|
|
* Otherwise the new table is initialized with zeros.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2010-03-23 19:41:24 +03:00
|
|
|
static int l2_allocate(BlockDriverState *bs, int l1_index, uint64_t **table)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2009-06-16 13:31:28 +04:00
|
|
|
uint64_t old_l2_offset;
|
2013-09-25 18:37:18 +04:00
|
|
|
uint64_t *l2_table = NULL;
|
2010-02-02 17:20:57 +03:00
|
|
|
int64_t l2_offset;
|
2010-03-23 19:41:24 +03:00
|
|
|
int ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
old_l2_offset = s->l1_table[l1_index];
|
|
|
|
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_l2_allocate(bs, l1_index);
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
/* allocate a new l2 entry */
|
|
|
|
|
2009-05-28 18:07:07 +04:00
|
|
|
l2_offset = qcow2_alloc_clusters(bs, s->l2_size * sizeof(uint64_t));
|
2010-01-20 17:04:01 +03:00
|
|
|
if (l2_offset < 0) {
|
2013-09-25 18:37:20 +04:00
|
|
|
ret = l2_offset;
|
|
|
|
goto fail;
|
2010-01-20 17:04:01 +03:00
|
|
|
}
|
2011-01-10 19:17:28 +03:00
|
|
|
|
|
|
|
ret = qcow2_cache_flush(bs, s->refcount_block_cache);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
/* allocate a new entry in the l2 cache */
|
|
|
|
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_l2_allocate_get_empty(bs, l1_index);
|
2011-01-10 19:17:28 +03:00
|
|
|
ret = qcow2_cache_get_empty(bs, s->l2_table_cache, l2_offset, (void**) table);
|
|
|
|
if (ret < 0) {
|
2013-09-25 18:37:20 +04:00
|
|
|
goto fail;
|
2011-01-10 19:17:28 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
l2_table = *table;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2012-02-23 18:40:55 +04:00
|
|
|
if ((old_l2_offset & L1E_OFFSET_MASK) == 0) {
|
2009-05-28 18:07:05 +04:00
|
|
|
/* if there was no old l2 table, clear the new table */
|
|
|
|
memset(l2_table, 0, s->l2_size * sizeof(uint64_t));
|
|
|
|
} else {
|
2011-01-10 19:17:28 +03:00
|
|
|
uint64_t* old_table;
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
/* if there was an old l2 table, read it from the disk */
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L2_ALLOC_COW_READ);
|
2012-02-23 18:40:55 +04:00
|
|
|
ret = qcow2_cache_get(bs, s->l2_table_cache,
|
|
|
|
old_l2_offset & L1E_OFFSET_MASK,
|
2011-01-10 19:17:28 +03:00
|
|
|
(void**) &old_table);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
memcpy(l2_table, old_table, s->cluster_size);
|
|
|
|
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &old_table);
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
2011-01-10 19:17:28 +03:00
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
/* write the l2 table to the file */
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L2_ALLOC_WRITE);
|
2011-01-10 19:17:28 +03:00
|
|
|
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_l2_allocate_write_l2(bs, l1_index);
|
2015-05-11 15:54:53 +03:00
|
|
|
qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, l2_table);
|
2011-01-10 19:17:28 +03:00
|
|
|
ret = qcow2_cache_flush(bs, s->l2_table_cache);
|
2010-03-23 19:41:24 +03:00
|
|
|
if (ret < 0) {
|
2010-05-12 18:23:26 +04:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* update the L1 entry */
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_l2_allocate_write_l1(bs, l1_index);
|
2010-05-12 18:23:26 +04:00
|
|
|
s->l1_table[l1_index] = l2_offset | QCOW_OFLAG_COPIED;
|
2013-08-30 16:34:28 +04:00
|
|
|
ret = qcow2_write_l1_entry(bs, l1_index);
|
2010-05-12 18:23:26 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
2010-03-23 19:41:24 +03:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2010-03-23 19:41:24 +03:00
|
|
|
*table = l2_table;
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_l2_allocate_done(bs, l1_index, 0);
|
2010-03-23 19:41:24 +03:00
|
|
|
return 0;
|
2010-05-12 18:23:26 +04:00
|
|
|
|
|
|
|
fail:
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_l2_allocate_done(bs, l1_index, ret);
|
2013-09-25 18:37:18 +04:00
|
|
|
if (l2_table != NULL) {
|
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void**) table);
|
|
|
|
}
|
2010-06-07 18:43:22 +04:00
|
|
|
s->l1_table[l1_index] = old_l2_offset;
|
2013-09-25 18:37:19 +04:00
|
|
|
if (l2_offset > 0) {
|
|
|
|
qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
|
|
|
|
QCOW2_DISCARD_ALWAYS);
|
|
|
|
}
|
2010-05-12 18:23:26 +04:00
|
|
|
return ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2012-03-15 19:37:40 +04:00
|
|
|
/*
|
|
|
|
* Checks how many clusters in a given L2 table are contiguous in the image
|
|
|
|
* file. As soon as one of the flags in the bitmask stop_flags changes compared
|
|
|
|
* to the first cluster, the search is stopped and the cluster is not counted
|
|
|
|
* as contiguous. (This allows it, for example, to stop at the first compressed
|
|
|
|
* cluster which may require a different handling)
|
|
|
|
*/
|
2015-09-14 17:39:47 +03:00
|
|
|
static int count_contiguous_clusters(int nb_clusters, int cluster_size,
|
2013-09-27 15:36:11 +04:00
|
|
|
uint64_t *l2_table, uint64_t stop_flags)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
|
|
|
int i;
|
2013-11-12 16:48:07 +04:00
|
|
|
uint64_t mask = stop_flags | L2E_OFFSET_MASK | QCOW_OFLAG_COMPRESSED;
|
2013-09-27 14:14:15 +04:00
|
|
|
uint64_t first_entry = be64_to_cpu(l2_table[0]);
|
|
|
|
uint64_t offset = first_entry & mask;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
if (!offset)
|
|
|
|
return 0;
|
|
|
|
|
2015-11-04 20:16:24 +03:00
|
|
|
assert(qcow2_get_cluster_type(first_entry) == QCOW2_CLUSTER_NORMAL);
|
2013-09-27 14:14:15 +04:00
|
|
|
|
2013-09-27 15:36:11 +04:00
|
|
|
for (i = 0; i < nb_clusters; i++) {
|
2012-03-15 19:37:40 +04:00
|
|
|
uint64_t l2_entry = be64_to_cpu(l2_table[i]) & mask;
|
|
|
|
if (offset + (uint64_t) i * cluster_size != l2_entry) {
|
2009-05-28 18:07:05 +04:00
|
|
|
break;
|
2012-03-15 19:37:40 +04:00
|
|
|
}
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2013-09-27 15:36:11 +04:00
|
|
|
return i;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2015-11-04 20:16:24 +03:00
|
|
|
static int count_contiguous_clusters_by_type(int nb_clusters,
|
|
|
|
uint64_t *l2_table,
|
|
|
|
int wanted_type)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2012-03-15 19:37:40 +04:00
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < nb_clusters; i++) {
|
|
|
|
int type = qcow2_get_cluster_type(be64_to_cpu(l2_table[i]));
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2015-11-04 20:16:24 +03:00
|
|
|
if (type != wanted_type) {
|
2012-03-15 19:37:40 +04:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
return i;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* The crypt function is compatible with the linux cryptoloop
|
|
|
|
algorithm for < 4 GB images. NOTE: out_buf == in_buf is
|
|
|
|
supported */
|
2015-09-07 18:12:56 +03:00
|
|
|
int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num,
|
2015-07-01 20:10:37 +03:00
|
|
|
uint8_t *out_buf, const uint8_t *in_buf,
|
|
|
|
int nb_sectors, bool enc,
|
|
|
|
Error **errp)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
|
|
|
union {
|
|
|
|
uint64_t ll[2];
|
|
|
|
uint8_t b[16];
|
|
|
|
} ivec;
|
|
|
|
int i;
|
2015-07-01 20:10:37 +03:00
|
|
|
int ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
for(i = 0; i < nb_sectors; i++) {
|
|
|
|
ivec.ll[0] = cpu_to_le64(sector_num);
|
|
|
|
ivec.ll[1] = 0;
|
2015-07-01 20:10:37 +03:00
|
|
|
if (qcrypto_cipher_setiv(s->cipher,
|
|
|
|
ivec.b, G_N_ELEMENTS(ivec.b),
|
|
|
|
errp) < 0) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
if (enc) {
|
|
|
|
ret = qcrypto_cipher_encrypt(s->cipher,
|
|
|
|
in_buf,
|
|
|
|
out_buf,
|
|
|
|
512,
|
|
|
|
errp);
|
|
|
|
} else {
|
|
|
|
ret = qcrypto_cipher_decrypt(s->cipher,
|
|
|
|
in_buf,
|
|
|
|
out_buf,
|
|
|
|
512,
|
|
|
|
errp);
|
|
|
|
}
|
|
|
|
if (ret < 0) {
|
|
|
|
return -1;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
sector_num++;
|
|
|
|
in_buf += 512;
|
|
|
|
out_buf += 512;
|
|
|
|
}
|
2015-07-01 20:10:37 +03:00
|
|
|
return 0;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2016-06-01 16:21:05 +03:00
|
|
|
static int coroutine_fn do_perform_cow(BlockDriverState *bs,
|
|
|
|
uint64_t src_cluster_offset,
|
|
|
|
uint64_t cluster_offset,
|
|
|
|
int offset_in_cluster,
|
|
|
|
int bytes)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2011-11-30 16:23:41 +04:00
|
|
|
QEMUIOVector qiov;
|
|
|
|
struct iovec iov;
|
2016-06-01 16:21:05 +03:00
|
|
|
int ret;
|
2011-09-19 13:26:48 +04:00
|
|
|
|
2016-06-01 16:21:05 +03:00
|
|
|
iov.iov_len = bytes;
|
2014-05-20 19:12:47 +04:00
|
|
|
iov.iov_base = qemu_try_blockalign(bs, iov.iov_len);
|
|
|
|
if (iov.iov_base == NULL) {
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
2011-11-30 16:23:41 +04:00
|
|
|
|
|
|
|
qemu_iovec_init_external(&qiov, &iov, 1);
|
2011-09-19 13:26:48 +04:00
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_COW_READ);
|
2011-11-30 16:23:41 +04:00
|
|
|
|
2014-03-11 02:44:07 +04:00
|
|
|
if (!bs->drv) {
|
2014-05-19 19:56:01 +04:00
|
|
|
ret = -ENOMEDIUM;
|
|
|
|
goto out;
|
2014-03-11 02:44:07 +04:00
|
|
|
}
|
|
|
|
|
2011-11-30 16:23:41 +04:00
|
|
|
/* Call .bdrv_co_readv() directly instead of using the public block-layer
|
|
|
|
* interface. This avoids double I/O throttling and request tracking,
|
|
|
|
* which can lead to deadlock when block layer copy-on-read is enabled.
|
|
|
|
*/
|
2016-06-01 16:21:05 +03:00
|
|
|
ret = bs->drv->bdrv_co_preadv(bs, src_cluster_offset + offset_in_cluster,
|
|
|
|
bytes, &qiov, 0);
|
2011-09-19 13:26:48 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
qcow2/qcow: protect against uninitialized encryption key
When a qcow[2] file is opened, if the header reports an
encryption method, this is used to set the 'crypt_method_header'
field on the BDRVQcow[2]State struct, and the 'encrypted' flag
in the BDRVState struct.
When doing I/O operations, the 'crypt_method' field on the
BDRVQcow[2]State struct is checked to determine if encryption
needs to be applied.
The crypt_method_header value is copied into crypt_method when
the bdrv_set_key() method is called.
The QEMU code which opens a block device is expected to always
do a check
if (bdrv_is_encrypted(bs)) {
bdrv_set_key(bs, ....key...);
}
If code forgets to do this, then 'crypt_method' is never set
and so when I/O is performed, QEMU writes plain text data
into a sector which is expected to contain cipher text, or
when reading, will return cipher text instead of plain
text.
Change the qcow[2] code to consult bs->encrypted when deciding
whether encryption is required, and assert(s->crypt_method)
to protect against cases where the caller forgets to set the
encryption key.
Also put an assert in the set_key methods to protect against
the case where the caller sets an encryption key on a block
device that does not have encryption
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2015-05-12 19:09:18 +03:00
|
|
|
if (bs->encrypted) {
|
2015-07-01 20:10:37 +03:00
|
|
|
Error *err = NULL;
|
2016-06-01 16:21:05 +03:00
|
|
|
int64_t sector = (cluster_offset + offset_in_cluster)
|
|
|
|
>> BDRV_SECTOR_BITS;
|
2015-07-01 20:10:37 +03:00
|
|
|
assert(s->cipher);
|
2016-06-01 16:21:05 +03:00
|
|
|
assert((offset_in_cluster & ~BDRV_SECTOR_MASK) == 0);
|
|
|
|
assert((bytes & ~BDRV_SECTOR_MASK) == 0);
|
|
|
|
if (qcow2_encrypt_sectors(s, sector, iov.iov_base, iov.iov_base,
|
|
|
|
bytes >> BDRV_SECTOR_BITS, true, &err) < 0) {
|
2015-07-01 20:10:37 +03:00
|
|
|
ret = -EIO;
|
|
|
|
error_free(err);
|
|
|
|
goto out;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
2011-09-19 13:26:48 +04:00
|
|
|
|
2013-10-10 13:09:23 +04:00
|
|
|
ret = qcow2_pre_write_overlap_check(bs, 0,
|
2016-06-01 16:21:05 +03:00
|
|
|
cluster_offset + offset_in_cluster, bytes);
|
2013-08-30 16:34:26 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_COW_WRITE);
|
2016-06-20 22:31:46 +03:00
|
|
|
ret = bdrv_co_pwritev(bs->file, cluster_offset + offset_in_cluster,
|
2016-06-01 16:21:05 +03:00
|
|
|
bytes, &qiov, 0);
|
2011-09-19 13:26:48 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
out:
|
2011-11-30 16:23:41 +04:00
|
|
|
qemu_vfree(iov.iov_base);
|
2011-09-19 13:26:48 +04:00
|
|
|
return ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* get_cluster_offset
|
|
|
|
*
|
2016-05-31 17:13:07 +03:00
|
|
|
* For a given offset of the virtual disk, find the cluster type and offset in
|
|
|
|
* the qcow2 file. The offset is stored in *cluster_offset.
|
2009-05-28 18:07:05 +04:00
|
|
|
*
|
2016-05-31 17:13:07 +03:00
|
|
|
* On entry, *bytes is the maximum number of contiguous bytes starting at
|
|
|
|
* offset that we are interested in.
|
2009-05-28 18:07:05 +04:00
|
|
|
*
|
2016-05-31 17:13:07 +03:00
|
|
|
* On exit, *bytes is the number of bytes starting at offset that have the same
|
|
|
|
* cluster type and (if applicable) are stored contiguously in the image file.
|
|
|
|
* Compressed clusters are always returned one by one.
|
2009-05-28 18:07:05 +04:00
|
|
|
*
|
2012-03-14 22:15:03 +04:00
|
|
|
* Returns the cluster type (QCOW2_CLUSTER_*) on success, -errno in error
|
|
|
|
* cases.
|
2009-05-28 18:07:05 +04:00
|
|
|
*/
|
2010-05-21 19:59:36 +04:00
|
|
|
int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
|
2016-05-31 17:13:07 +03:00
|
|
|
unsigned int *bytes, uint64_t *cluster_offset)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-05-14 18:14:33 +04:00
|
|
|
unsigned int l2_index;
|
|
|
|
uint64_t l1_index, l2_offset, *l2_table;
|
2009-05-28 18:07:05 +04:00
|
|
|
int l1_bits, c;
|
2016-06-20 17:26:23 +03:00
|
|
|
unsigned int offset_in_cluster;
|
|
|
|
uint64_t bytes_available, bytes_needed, nb_clusters;
|
2010-05-21 20:25:20 +04:00
|
|
|
int ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-05-31 17:41:09 +03:00
|
|
|
offset_in_cluster = offset_into_cluster(s, offset);
|
2016-05-31 17:13:07 +03:00
|
|
|
bytes_needed = (uint64_t) *bytes + offset_in_cluster;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-05-31 17:41:09 +03:00
|
|
|
l1_bits = s->l2_bits + s->cluster_bits;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-05-31 17:41:09 +03:00
|
|
|
/* compute how many bytes there are between the start of the cluster
|
|
|
|
* containing offset and the end of the l1 entry */
|
|
|
|
bytes_available = (1ULL << l1_bits) - (offset & ((1ULL << l1_bits) - 1))
|
|
|
|
+ offset_in_cluster;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-05-31 17:41:09 +03:00
|
|
|
if (bytes_needed > bytes_available) {
|
|
|
|
bytes_needed = bytes_available;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2010-05-21 19:59:36 +04:00
|
|
|
*cluster_offset = 0;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2015-08-26 14:17:13 +03:00
|
|
|
/* seek to the l2 offset in the l1 table */
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
l1_index = offset >> l1_bits;
|
2012-03-14 22:15:03 +04:00
|
|
|
if (l1_index >= s->l1_size) {
|
|
|
|
ret = QCOW2_CLUSTER_UNALLOCATED;
|
2009-05-28 18:07:05 +04:00
|
|
|
goto out;
|
2012-03-14 22:15:03 +04:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2012-03-14 22:15:03 +04:00
|
|
|
l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
|
|
|
|
if (!l2_offset) {
|
|
|
|
ret = QCOW2_CLUSTER_UNALLOCATED;
|
2009-05-28 18:07:05 +04:00
|
|
|
goto out;
|
2012-03-14 22:15:03 +04:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2014-09-05 18:07:18 +04:00
|
|
|
if (offset_into_cluster(s, l2_offset)) {
|
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "L2 table offset %#" PRIx64
|
|
|
|
" unaligned (L1 index: %#" PRIx64 ")",
|
|
|
|
l2_offset, l1_index);
|
|
|
|
return -EIO;
|
|
|
|
}
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
/* load the l2 table in memory */
|
|
|
|
|
2010-05-21 20:25:20 +04:00
|
|
|
ret = l2_load(bs, l2_offset, &l2_table);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
2010-05-21 19:59:36 +04:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
/* find the cluster offset for the given disk offset */
|
|
|
|
|
|
|
|
l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
|
2010-05-21 19:59:36 +04:00
|
|
|
*cluster_offset = be64_to_cpu(l2_table[l2_index]);
|
2015-09-14 17:39:47 +03:00
|
|
|
|
2016-05-31 17:41:09 +03:00
|
|
|
nb_clusters = size_to_clusters(s, bytes_needed);
|
2016-06-20 17:26:23 +03:00
|
|
|
/* bytes_needed <= *bytes + offset_in_cluster, both of which are unsigned
|
|
|
|
* integers; the minimum cluster size is 512, so this assertion is always
|
|
|
|
* true */
|
|
|
|
assert(nb_clusters <= INT_MAX);
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2012-03-14 22:15:03 +04:00
|
|
|
ret = qcow2_get_cluster_type(*cluster_offset);
|
|
|
|
switch (ret) {
|
|
|
|
case QCOW2_CLUSTER_COMPRESSED:
|
|
|
|
/* Compressed clusters can only be processed one by one */
|
|
|
|
c = 1;
|
|
|
|
*cluster_offset &= L2E_COMPRESSED_OFFSET_SIZE_MASK;
|
|
|
|
break;
|
2012-03-16 18:02:38 +04:00
|
|
|
case QCOW2_CLUSTER_ZERO:
|
2013-03-06 21:02:01 +04:00
|
|
|
if (s->qcow_version < 3) {
|
2014-09-05 18:07:18 +04:00
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "Zero cluster entry found"
|
|
|
|
" in pre-v3 image (L2 offset: %#" PRIx64
|
|
|
|
", L2 index: %#x)", l2_offset, l2_index);
|
|
|
|
ret = -EIO;
|
|
|
|
goto fail;
|
2013-03-06 21:02:01 +04:00
|
|
|
}
|
2015-11-04 20:16:24 +03:00
|
|
|
c = count_contiguous_clusters_by_type(nb_clusters, &l2_table[l2_index],
|
|
|
|
QCOW2_CLUSTER_ZERO);
|
2012-03-16 18:02:38 +04:00
|
|
|
*cluster_offset = 0;
|
|
|
|
break;
|
2012-03-14 22:15:03 +04:00
|
|
|
case QCOW2_CLUSTER_UNALLOCATED:
|
2009-05-28 18:07:05 +04:00
|
|
|
/* how many empty clusters ? */
|
2015-11-04 20:16:24 +03:00
|
|
|
c = count_contiguous_clusters_by_type(nb_clusters, &l2_table[l2_index],
|
|
|
|
QCOW2_CLUSTER_UNALLOCATED);
|
2012-03-14 22:15:03 +04:00
|
|
|
*cluster_offset = 0;
|
|
|
|
break;
|
|
|
|
case QCOW2_CLUSTER_NORMAL:
|
2009-05-28 18:07:05 +04:00
|
|
|
/* how many allocated clusters ? */
|
|
|
|
c = count_contiguous_clusters(nb_clusters, s->cluster_size,
|
2013-09-27 15:36:11 +04:00
|
|
|
&l2_table[l2_index], QCOW_OFLAG_ZERO);
|
2012-03-14 22:15:03 +04:00
|
|
|
*cluster_offset &= L2E_OFFSET_MASK;
|
2014-09-05 18:07:18 +04:00
|
|
|
if (offset_into_cluster(s, *cluster_offset)) {
|
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "Data cluster offset %#"
|
|
|
|
PRIx64 " unaligned (L2 offset: %#" PRIx64
|
|
|
|
", L2 index: %#x)", *cluster_offset,
|
|
|
|
l2_offset, l2_index);
|
|
|
|
ret = -EIO;
|
|
|
|
goto fail;
|
|
|
|
}
|
2012-03-14 22:15:03 +04:00
|
|
|
break;
|
2012-06-15 15:43:18 +04:00
|
|
|
default:
|
|
|
|
abort();
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2011-01-10 19:17:28 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
|
|
|
|
|
2016-06-20 17:26:23 +03:00
|
|
|
bytes_available = (int64_t)c * s->cluster_size;
|
2012-03-14 22:15:03 +04:00
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
out:
|
2016-05-31 17:41:09 +03:00
|
|
|
if (bytes_available > bytes_needed) {
|
|
|
|
bytes_available = bytes_needed;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-06-20 17:26:23 +03:00
|
|
|
/* bytes_available <= bytes_needed <= *bytes + offset_in_cluster;
|
|
|
|
* subtracting offset_in_cluster will therefore definitely yield something
|
|
|
|
* not exceeding UINT_MAX */
|
|
|
|
assert(bytes_available - offset_in_cluster <= UINT_MAX);
|
2016-05-31 17:13:07 +03:00
|
|
|
*bytes = bytes_available - offset_in_cluster;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2012-03-14 22:15:03 +04:00
|
|
|
return ret;
|
2014-09-05 18:07:18 +04:00
|
|
|
|
|
|
|
fail:
|
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **)&l2_table);
|
|
|
|
return ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* get_cluster_table
|
|
|
|
*
|
|
|
|
* for a given disk offset, load (and allocate if needed)
|
|
|
|
* the l2 table.
|
|
|
|
*
|
|
|
|
* the l2 table offset in the qcow2 file and the cluster index
|
|
|
|
* in the l2 table are given to the caller.
|
|
|
|
*
|
2010-01-20 17:03:00 +03:00
|
|
|
* Returns 0 on success, -errno in failure case
|
2009-05-28 18:07:05 +04:00
|
|
|
*/
|
|
|
|
static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
|
|
|
|
uint64_t **new_l2_table,
|
|
|
|
int *new_l2_index)
|
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-05-14 18:14:33 +04:00
|
|
|
unsigned int l2_index;
|
|
|
|
uint64_t l1_index, l2_offset;
|
2010-03-23 19:41:24 +03:00
|
|
|
uint64_t *l2_table = NULL;
|
2009-09-15 14:30:43 +04:00
|
|
|
int ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2015-08-26 14:17:13 +03:00
|
|
|
/* seek to the l2 offset in the l1 table */
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
l1_index = offset >> (s->l2_bits + s->cluster_bits);
|
|
|
|
if (l1_index >= s->l1_size) {
|
2010-10-18 19:53:53 +04:00
|
|
|
ret = qcow2_grow_l1_table(bs, l1_index + 1, false);
|
2010-01-20 17:03:00 +03:00
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
2012-02-23 18:40:55 +04:00
|
|
|
|
2013-05-14 18:14:33 +04:00
|
|
|
assert(l1_index < s->l1_size);
|
2012-02-23 18:40:55 +04:00
|
|
|
l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
|
2014-09-05 18:07:18 +04:00
|
|
|
if (offset_into_cluster(s, l2_offset)) {
|
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "L2 table offset %#" PRIx64
|
|
|
|
" unaligned (L1 index: %#" PRIx64 ")",
|
|
|
|
l2_offset, l1_index);
|
|
|
|
return -EIO;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
/* seek the l2 table of the given l2 offset */
|
|
|
|
|
2012-02-23 18:40:55 +04:00
|
|
|
if (s->l1_table[l1_index] & QCOW_OFLAG_COPIED) {
|
2009-05-28 18:07:05 +04:00
|
|
|
/* load the l2 table in memory */
|
2010-05-21 20:25:20 +04:00
|
|
|
ret = l2_load(bs, l2_offset, &l2_table);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
2010-01-20 17:03:00 +03:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
} else {
|
2011-02-09 19:36:19 +03:00
|
|
|
/* First allocate a new L2 table (and do COW if needed) */
|
2010-03-23 19:41:24 +03:00
|
|
|
ret = l2_allocate(bs, l1_index, &l2_table);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
2010-01-20 17:03:00 +03:00
|
|
|
}
|
2011-02-09 19:36:19 +03:00
|
|
|
|
|
|
|
/* Then decrease the refcount of the old table */
|
|
|
|
if (l2_offset) {
|
2013-06-19 15:44:18 +04:00
|
|
|
qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
|
|
|
|
QCOW2_DISCARD_OTHER);
|
2011-02-09 19:36:19 +03:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/* find the cluster offset for the given disk offset */
|
|
|
|
|
|
|
|
l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
|
|
|
|
|
|
|
|
*new_l2_table = l2_table;
|
|
|
|
*new_l2_index = l2_index;
|
|
|
|
|
2010-01-20 17:03:00 +03:00
|
|
|
return 0;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* alloc_compressed_cluster_offset
|
|
|
|
*
|
|
|
|
* For a given offset of the disk image, return cluster offset in
|
|
|
|
* qcow2 file.
|
|
|
|
*
|
|
|
|
* If the offset is not found, allocate a new compressed cluster.
|
|
|
|
*
|
|
|
|
* Return the cluster offset if successful,
|
|
|
|
* Return 0, otherwise.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2009-05-28 18:07:07 +04:00
|
|
|
uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
|
|
|
|
uint64_t offset,
|
|
|
|
int compressed_size)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2009-05-28 18:07:05 +04:00
|
|
|
int l2_index, ret;
|
2012-03-12 20:46:51 +04:00
|
|
|
uint64_t *l2_table;
|
2010-02-02 17:20:57 +03:00
|
|
|
int64_t cluster_offset;
|
2009-05-28 18:07:05 +04:00
|
|
|
int nb_csectors;
|
|
|
|
|
2012-03-12 20:46:51 +04:00
|
|
|
ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
|
2010-01-20 17:03:00 +03:00
|
|
|
if (ret < 0) {
|
2009-05-28 18:07:05 +04:00
|
|
|
return 0;
|
2010-01-20 17:03:00 +03:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2012-03-15 20:20:11 +04:00
|
|
|
/* Compression can't overwrite anything. Fail if the cluster was already
|
|
|
|
* allocated. */
|
2009-05-28 18:07:05 +04:00
|
|
|
cluster_offset = be64_to_cpu(l2_table[l2_index]);
|
2012-03-15 20:20:11 +04:00
|
|
|
if (cluster_offset & L2E_OFFSET_MASK) {
|
2011-10-18 19:12:44 +04:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
|
|
|
|
return 0;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2009-05-28 18:07:07 +04:00
|
|
|
cluster_offset = qcow2_alloc_bytes(bs, compressed_size);
|
2010-01-20 17:04:01 +03:00
|
|
|
if (cluster_offset < 0) {
|
2011-01-10 19:17:28 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
|
2010-01-20 17:04:01 +03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
nb_csectors = ((cluster_offset + compressed_size - 1) >> 9) -
|
|
|
|
(cluster_offset >> 9);
|
|
|
|
|
|
|
|
cluster_offset |= QCOW_OFLAG_COMPRESSED |
|
|
|
|
((uint64_t)nb_csectors << s->csize_shift);
|
|
|
|
|
|
|
|
/* update L2 table */
|
|
|
|
|
|
|
|
/* compressed clusters never have the copied flag */
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE_COMPRESSED);
|
2015-05-11 15:54:53 +03:00
|
|
|
qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, l2_table);
|
2009-05-28 18:07:05 +04:00
|
|
|
l2_table[l2_index] = cpu_to_be64(cluster_offset);
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2009-06-16 13:31:29 +04:00
|
|
|
|
2011-01-10 19:17:28 +03:00
|
|
|
return cluster_offset;
|
2009-06-16 13:31:29 +04:00
|
|
|
}
|
|
|
|
|
2012-12-07 21:08:43 +04:00
|
|
|
static int perform_cow(BlockDriverState *bs, QCowL2Meta *m, Qcow2COWRegion *r)
|
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2012-12-07 21:08:43 +04:00
|
|
|
int ret;
|
|
|
|
|
2016-06-01 15:08:56 +03:00
|
|
|
if (r->nb_bytes == 0) {
|
2012-12-07 21:08:43 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
2016-06-01 15:08:56 +03:00
|
|
|
ret = do_perform_cow(bs, m->offset, m->alloc_offset, r->offset, r->nb_bytes);
|
2012-12-07 21:08:43 +04:00
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Before we update the L2 table to actually point to the new cluster, we
|
|
|
|
* need to be sure that the refcounts have been increased and COW was
|
|
|
|
* handled.
|
|
|
|
*/
|
|
|
|
qcow2_cache_depends_on_flush(s->l2_table_cache);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-01-20 17:03:01 +03:00
|
|
|
int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2009-05-28 18:07:05 +04:00
|
|
|
int i, j = 0, l2_index, ret;
|
2012-12-07 21:08:43 +04:00
|
|
|
uint64_t *old_cluster, *l2_table;
|
2012-03-02 17:10:54 +04:00
|
|
|
uint64_t cluster_offset = m->alloc_offset;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2012-03-01 21:36:21 +04:00
|
|
|
trace_qcow2_cluster_link_l2(qemu_coroutine_self(), m->nb_clusters);
|
2012-12-07 21:08:46 +04:00
|
|
|
assert(m->nb_clusters > 0);
|
2009-05-28 18:07:05 +04:00
|
|
|
|
block: Use g_new() & friends where that makes obvious sense
g_new(T, n) is neater than g_malloc(sizeof(T) * n). It's also safer,
for two reasons. One, it catches multiplication overflowing size_t.
Two, it returns T * rather than void *, which lets the compiler catch
more type errors.
Patch created with Coccinelle, with two manual changes on top:
* Add const to bdrv_iterate_format() to keep the types straight
* Convert the allocation in bdrv_drop_intermediate(), which Coccinelle
inexplicably misses
Coccinelle semantic patch:
@@
type T;
@@
-g_malloc(sizeof(T))
+g_new(T, 1)
@@
type T;
@@
-g_try_malloc(sizeof(T))
+g_try_new(T, 1)
@@
type T;
@@
-g_malloc0(sizeof(T))
+g_new0(T, 1)
@@
type T;
@@
-g_try_malloc0(sizeof(T))
+g_try_new0(T, 1)
@@
type T;
expression n;
@@
-g_malloc(sizeof(T) * (n))
+g_new(T, n)
@@
type T;
expression n;
@@
-g_try_malloc(sizeof(T) * (n))
+g_try_new(T, n)
@@
type T;
expression n;
@@
-g_malloc0(sizeof(T) * (n))
+g_new0(T, n)
@@
type T;
expression n;
@@
-g_try_malloc0(sizeof(T) * (n))
+g_try_new0(T, n)
@@
type T;
expression p, n;
@@
-g_realloc(p, sizeof(T) * (n))
+g_renew(T, p, n)
@@
type T;
expression p, n;
@@
-g_try_realloc(p, sizeof(T) * (n))
+g_try_renew(T, p, n)
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2014-08-19 12:31:08 +04:00
|
|
|
old_cluster = g_try_new(uint64_t, m->nb_clusters);
|
2014-05-20 19:12:47 +04:00
|
|
|
if (old_cluster == NULL) {
|
|
|
|
ret = -ENOMEM;
|
|
|
|
goto err;
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
/* copy content of unmodified sectors */
|
2012-12-07 21:08:43 +04:00
|
|
|
ret = perform_cow(bs, m, &m->cow_start);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto err;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
2012-12-07 21:08:43 +04:00
|
|
|
ret = perform_cow(bs, m, &m->cow_end);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto err;
|
2011-01-10 19:17:28 +03:00
|
|
|
}
|
|
|
|
|
2012-12-07 21:08:43 +04:00
|
|
|
/* Update L2 table. */
|
2013-03-15 13:35:08 +04:00
|
|
|
if (s->use_lazy_refcounts) {
|
2012-12-07 21:08:47 +04:00
|
|
|
qcow2_mark_dirty(bs);
|
|
|
|
}
|
2012-07-27 12:05:22 +04:00
|
|
|
if (qcow2_need_accurate_refcounts(s)) {
|
|
|
|
qcow2_cache_set_dependency(bs, s->l2_table_cache,
|
|
|
|
s->refcount_block_cache);
|
|
|
|
}
|
2012-12-07 21:08:47 +04:00
|
|
|
|
2012-03-12 20:46:51 +04:00
|
|
|
ret = get_cluster_table(bs, m->offset, &l2_table, &l2_index);
|
2010-01-20 17:03:00 +03:00
|
|
|
if (ret < 0) {
|
2009-05-28 18:07:05 +04:00
|
|
|
goto err;
|
2010-01-20 17:03:00 +03:00
|
|
|
}
|
2015-05-11 15:54:53 +03:00
|
|
|
qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, l2_table);
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2013-09-25 19:48:55 +04:00
|
|
|
assert(l2_index + m->nb_clusters <= s->l2_size);
|
2009-05-28 18:07:05 +04:00
|
|
|
for (i = 0; i < m->nb_clusters; i++) {
|
|
|
|
/* if two concurrent writes happen to the same unallocated cluster
|
2016-06-01 16:21:05 +03:00
|
|
|
* each write allocates separate cluster and writes data concurrently.
|
|
|
|
* The first one to complete updates l2 table with pointer to its
|
|
|
|
* cluster the second one has to do RMW (which is done above by
|
|
|
|
* perform_cow()), update l2 table with its cluster pointer and free
|
|
|
|
* old cluster. This is what this loop does */
|
|
|
|
if (l2_table[l2_index + i] != 0) {
|
2009-05-28 18:07:05 +04:00
|
|
|
old_cluster[j++] = l2_table[l2_index + i];
|
2016-06-01 16:21:05 +03:00
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
l2_table[l2_index + i] = cpu_to_be64((cluster_offset +
|
|
|
|
(i << s->cluster_bits)) | QCOW_OFLAG_COPIED);
|
|
|
|
}
|
|
|
|
|
2010-09-17 19:02:09 +04:00
|
|
|
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2010-09-01 14:40:52 +04:00
|
|
|
/*
|
|
|
|
* If this was a COW, we need to decrease the refcount of the old cluster.
|
2013-06-19 15:44:18 +04:00
|
|
|
*
|
|
|
|
* Don't discard clusters that reach a refcount of 0 (e.g. compressed
|
|
|
|
* clusters), the next write will reuse them anyway.
|
2010-09-01 14:40:52 +04:00
|
|
|
*/
|
|
|
|
if (j != 0) {
|
|
|
|
for (i = 0; i < j; i++) {
|
2013-06-19 15:44:18 +04:00
|
|
|
qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1,
|
|
|
|
QCOW2_DISCARD_NEVER);
|
2010-09-01 14:40:52 +04:00
|
|
|
}
|
|
|
|
}
|
2009-05-28 18:07:05 +04:00
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
err:
|
2011-08-21 07:09:37 +04:00
|
|
|
g_free(old_cluster);
|
2009-05-28 18:07:05 +04:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2012-03-02 22:27:53 +04:00
|
|
|
/*
|
|
|
|
* Returns the number of contiguous clusters that can be used for an allocating
|
|
|
|
* write, but require COW to be performed (this includes yet unallocated space,
|
|
|
|
* which must copy from the backing file)
|
|
|
|
*/
|
2015-09-07 18:12:56 +03:00
|
|
|
static int count_cow_clusters(BDRVQcow2State *s, int nb_clusters,
|
2012-03-02 22:27:53 +04:00
|
|
|
uint64_t *l2_table, int l2_index)
|
|
|
|
{
|
2012-03-27 15:17:22 +04:00
|
|
|
int i;
|
2012-03-02 22:27:53 +04:00
|
|
|
|
2012-03-27 15:17:22 +04:00
|
|
|
for (i = 0; i < nb_clusters; i++) {
|
|
|
|
uint64_t l2_entry = be64_to_cpu(l2_table[l2_index + i]);
|
|
|
|
int cluster_type = qcow2_get_cluster_type(l2_entry);
|
|
|
|
|
|
|
|
switch(cluster_type) {
|
|
|
|
case QCOW2_CLUSTER_NORMAL:
|
|
|
|
if (l2_entry & QCOW_OFLAG_COPIED) {
|
|
|
|
goto out;
|
|
|
|
}
|
2012-03-02 22:27:53 +04:00
|
|
|
break;
|
2012-03-27 15:17:22 +04:00
|
|
|
case QCOW2_CLUSTER_UNALLOCATED:
|
|
|
|
case QCOW2_CLUSTER_COMPRESSED:
|
2012-03-16 18:02:38 +04:00
|
|
|
case QCOW2_CLUSTER_ZERO:
|
2012-03-02 22:27:53 +04:00
|
|
|
break;
|
2012-03-27 15:17:22 +04:00
|
|
|
default:
|
|
|
|
abort();
|
|
|
|
}
|
2012-03-02 22:27:53 +04:00
|
|
|
}
|
|
|
|
|
2012-03-27 15:17:22 +04:00
|
|
|
out:
|
2012-03-02 22:27:53 +04:00
|
|
|
assert(i <= nb_clusters);
|
|
|
|
return i;
|
|
|
|
}
|
|
|
|
|
2012-03-02 17:10:54 +04:00
|
|
|
/*
|
2012-12-07 21:08:49 +04:00
|
|
|
* Check if there already is an AIO write request in flight which allocates
|
|
|
|
* the same cluster. In this case we need to wait until the previous
|
|
|
|
* request has completed and updated the L2 table accordingly.
|
2013-03-26 20:49:58 +04:00
|
|
|
*
|
|
|
|
* Returns:
|
|
|
|
* 0 if there was no dependency. *cur_bytes indicates the number of
|
|
|
|
* bytes from guest_offset that can be read before the next
|
|
|
|
* dependency must be processed (or the request is complete)
|
|
|
|
*
|
|
|
|
* -EAGAIN if we had to wait for another request, previously gathered
|
|
|
|
* information on cluster allocation may be invalid now. The caller
|
|
|
|
* must start over anyway, so consider *cur_bytes undefined.
|
2012-03-02 17:10:54 +04:00
|
|
|
*/
|
2012-12-07 21:08:49 +04:00
|
|
|
static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
|
2013-03-27 14:43:49 +04:00
|
|
|
uint64_t *cur_bytes, QCowL2Meta **m)
|
2012-03-02 17:10:54 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2012-03-02 17:10:54 +04:00
|
|
|
QCowL2Meta *old_alloc;
|
2013-03-26 20:49:58 +04:00
|
|
|
uint64_t bytes = *cur_bytes;
|
2012-03-02 17:10:54 +04:00
|
|
|
|
|
|
|
QLIST_FOREACH(old_alloc, &s->cluster_allocs, next_in_flight) {
|
|
|
|
|
2013-03-26 20:49:58 +04:00
|
|
|
uint64_t start = guest_offset;
|
|
|
|
uint64_t end = start + bytes;
|
|
|
|
uint64_t old_start = l2meta_cow_start(old_alloc);
|
|
|
|
uint64_t old_end = l2meta_cow_end(old_alloc);
|
2012-03-02 17:10:54 +04:00
|
|
|
|
2013-03-26 20:49:57 +04:00
|
|
|
if (end <= old_start || start >= old_end) {
|
2012-03-02 17:10:54 +04:00
|
|
|
/* No intersection */
|
|
|
|
} else {
|
|
|
|
if (start < old_start) {
|
|
|
|
/* Stop at the start of a running allocation */
|
2013-03-26 20:49:58 +04:00
|
|
|
bytes = old_start - start;
|
2012-03-02 17:10:54 +04:00
|
|
|
} else {
|
2013-03-26 20:49:58 +04:00
|
|
|
bytes = 0;
|
2012-03-02 17:10:54 +04:00
|
|
|
}
|
|
|
|
|
2013-03-27 14:43:49 +04:00
|
|
|
/* Stop if already an l2meta exists. After yielding, it wouldn't
|
|
|
|
* be valid any more, so we'd have to clean up the old L2Metas
|
|
|
|
* and deal with requests depending on them before starting to
|
|
|
|
* gather new ones. Not worth the trouble. */
|
|
|
|
if (bytes == 0 && *m) {
|
|
|
|
*cur_bytes = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:49:58 +04:00
|
|
|
if (bytes == 0) {
|
2012-03-02 17:10:54 +04:00
|
|
|
/* Wait for the dependency to complete. We need to recheck
|
|
|
|
* the free/allocated clusters when we continue. */
|
|
|
|
qemu_co_mutex_unlock(&s->lock);
|
|
|
|
qemu_co_queue_wait(&old_alloc->dependent_requests);
|
|
|
|
qemu_co_mutex_lock(&s->lock);
|
|
|
|
return -EAGAIN;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:49:58 +04:00
|
|
|
/* Make sure that existing clusters and new allocations are only used up to
|
|
|
|
* the next dependency if we shortened the request above */
|
|
|
|
*cur_bytes = bytes;
|
2012-03-02 17:10:54 +04:00
|
|
|
|
2012-12-07 21:08:49 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:05 +04:00
|
|
|
/*
|
|
|
|
* Checks how many already allocated clusters that don't require a copy on
|
|
|
|
* write there are at the given guest_offset (up to *bytes). If
|
|
|
|
* *host_offset is not zero, only physically contiguous clusters beginning at
|
|
|
|
* this host offset are counted.
|
|
|
|
*
|
2013-03-26 20:50:09 +04:00
|
|
|
* Note that guest_offset may not be cluster aligned. In this case, the
|
|
|
|
* returned *host_offset points to exact byte referenced by guest_offset and
|
|
|
|
* therefore isn't cluster aligned as well.
|
2013-03-26 20:50:05 +04:00
|
|
|
*
|
|
|
|
* Returns:
|
|
|
|
* 0: if no allocated clusters are available at the given offset.
|
|
|
|
* *bytes is normally unchanged. It is set to 0 if the cluster
|
|
|
|
* is allocated and doesn't need COW, but doesn't have the right
|
|
|
|
* physical offset.
|
|
|
|
*
|
|
|
|
* 1: if allocated clusters that don't require a COW are available at
|
|
|
|
* the requested offset. *bytes may have decreased and describes
|
|
|
|
* the length of the area that can be written to.
|
|
|
|
*
|
|
|
|
* -errno: in error cases
|
|
|
|
*/
|
|
|
|
static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
|
2013-03-26 20:50:07 +04:00
|
|
|
uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
|
2013-03-26 20:50:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-03-26 20:50:05 +04:00
|
|
|
int l2_index;
|
|
|
|
uint64_t cluster_offset;
|
|
|
|
uint64_t *l2_table;
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t nb_clusters;
|
2013-03-26 20:50:07 +04:00
|
|
|
unsigned int keep_clusters;
|
2015-05-11 15:54:58 +03:00
|
|
|
int ret;
|
2013-03-26 20:50:05 +04:00
|
|
|
|
|
|
|
trace_qcow2_handle_copied(qemu_coroutine_self(), guest_offset, *host_offset,
|
|
|
|
*bytes);
|
|
|
|
|
2013-03-26 20:50:09 +04:00
|
|
|
assert(*host_offset == 0 || offset_into_cluster(s, guest_offset)
|
|
|
|
== offset_into_cluster(s, *host_offset));
|
|
|
|
|
2013-03-26 20:50:06 +04:00
|
|
|
/*
|
|
|
|
* Calculate the number of clusters to look for. We stop at L2 table
|
|
|
|
* boundaries to keep things simple.
|
|
|
|
*/
|
|
|
|
nb_clusters =
|
|
|
|
size_to_clusters(s, offset_into_cluster(s, guest_offset) + *bytes);
|
|
|
|
|
|
|
|
l2_index = offset_to_l2_index(s, guest_offset);
|
|
|
|
nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
|
2015-09-14 17:39:47 +03:00
|
|
|
assert(nb_clusters <= INT_MAX);
|
2013-03-26 20:50:06 +04:00
|
|
|
|
2013-03-26 20:50:05 +04:00
|
|
|
/* Find L2 entry for the first involved cluster */
|
|
|
|
ret = get_cluster_table(bs, guest_offset, &l2_table, &l2_index);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
cluster_offset = be64_to_cpu(l2_table[l2_index]);
|
|
|
|
|
|
|
|
/* Check how many clusters are already allocated and don't need COW */
|
|
|
|
if (qcow2_get_cluster_type(cluster_offset) == QCOW2_CLUSTER_NORMAL
|
|
|
|
&& (cluster_offset & QCOW_OFLAG_COPIED))
|
|
|
|
{
|
2013-03-26 20:50:08 +04:00
|
|
|
/* If a specific host_offset is required, check it */
|
|
|
|
bool offset_matches =
|
|
|
|
(cluster_offset & L2E_OFFSET_MASK) == *host_offset;
|
|
|
|
|
2014-09-05 18:07:18 +04:00
|
|
|
if (offset_into_cluster(s, cluster_offset & L2E_OFFSET_MASK)) {
|
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "Data cluster offset "
|
|
|
|
"%#llx unaligned (guest offset: %#" PRIx64
|
|
|
|
")", cluster_offset & L2E_OFFSET_MASK,
|
|
|
|
guest_offset);
|
|
|
|
ret = -EIO;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:08 +04:00
|
|
|
if (*host_offset != 0 && !offset_matches) {
|
|
|
|
*bytes = 0;
|
|
|
|
ret = 0;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:05 +04:00
|
|
|
/* We keep all QCOW_OFLAG_COPIED clusters */
|
2013-03-26 20:50:07 +04:00
|
|
|
keep_clusters =
|
2013-03-26 20:50:06 +04:00
|
|
|
count_contiguous_clusters(nb_clusters, s->cluster_size,
|
2013-09-27 15:36:11 +04:00
|
|
|
&l2_table[l2_index],
|
2013-03-26 20:50:05 +04:00
|
|
|
QCOW_OFLAG_COPIED | QCOW_OFLAG_ZERO);
|
2013-03-26 20:50:07 +04:00
|
|
|
assert(keep_clusters <= nb_clusters);
|
|
|
|
|
|
|
|
*bytes = MIN(*bytes,
|
|
|
|
keep_clusters * s->cluster_size
|
|
|
|
- offset_into_cluster(s, guest_offset));
|
2013-03-26 20:50:05 +04:00
|
|
|
|
|
|
|
ret = 1;
|
|
|
|
} else {
|
|
|
|
ret = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Cleanup */
|
2013-03-26 20:50:08 +04:00
|
|
|
out:
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2013-03-26 20:50:05 +04:00
|
|
|
|
2013-03-26 20:50:08 +04:00
|
|
|
/* Only return a host offset if we actually made progress. Otherwise we
|
|
|
|
* would make requirements for handle_alloc() that it can't fulfill */
|
2014-09-05 18:07:18 +04:00
|
|
|
if (ret > 0) {
|
2013-03-26 20:50:09 +04:00
|
|
|
*host_offset = (cluster_offset & L2E_OFFSET_MASK)
|
|
|
|
+ offset_into_cluster(s, guest_offset);
|
2013-03-26 20:50:08 +04:00
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:05 +04:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2012-12-07 21:08:49 +04:00
|
|
|
/*
|
|
|
|
* Allocates new clusters for the given guest_offset.
|
|
|
|
*
|
|
|
|
* At most *nb_clusters are allocated, and on return *nb_clusters is updated to
|
|
|
|
* contain the number of clusters that have been allocated and are contiguous
|
|
|
|
* in the image file.
|
|
|
|
*
|
|
|
|
* If *host_offset is non-zero, it specifies the offset in the image file at
|
|
|
|
* which the new clusters must start. *nb_clusters can be 0 on return in this
|
|
|
|
* case if the cluster at host_offset is already in use. If *host_offset is
|
|
|
|
* zero, the clusters can be allocated anywhere in the image file.
|
|
|
|
*
|
|
|
|
* *host_offset is updated to contain the offset into the image file at which
|
|
|
|
* the first allocated cluster starts.
|
|
|
|
*
|
|
|
|
* Return 0 on success and -errno in error cases. -EAGAIN means that the
|
|
|
|
* function has been waiting for another request and the allocation must be
|
|
|
|
* restarted, but the whole request should not be failed.
|
|
|
|
*/
|
|
|
|
static int do_alloc_cluster_offset(BlockDriverState *bs, uint64_t guest_offset,
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t *host_offset, uint64_t *nb_clusters)
|
2012-12-07 21:08:49 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2012-12-07 21:08:49 +04:00
|
|
|
|
|
|
|
trace_qcow2_do_alloc_clusters_offset(qemu_coroutine_self(), guest_offset,
|
|
|
|
*host_offset, *nb_clusters);
|
|
|
|
|
2012-03-02 17:10:54 +04:00
|
|
|
/* Allocate new clusters */
|
|
|
|
trace_qcow2_cluster_alloc_phys(qemu_coroutine_self());
|
|
|
|
if (*host_offset == 0) {
|
2012-05-24 14:56:32 +04:00
|
|
|
int64_t cluster_offset =
|
|
|
|
qcow2_alloc_clusters(bs, *nb_clusters * s->cluster_size);
|
|
|
|
if (cluster_offset < 0) {
|
|
|
|
return cluster_offset;
|
|
|
|
}
|
|
|
|
*host_offset = cluster_offset;
|
|
|
|
return 0;
|
2012-03-02 17:10:54 +04:00
|
|
|
} else {
|
2015-09-14 17:39:47 +03:00
|
|
|
int64_t ret = qcow2_alloc_clusters_at(bs, *host_offset, *nb_clusters);
|
2012-05-24 14:56:32 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
*nb_clusters = ret;
|
|
|
|
return 0;
|
2012-03-02 17:10:54 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:00 +04:00
|
|
|
/*
|
|
|
|
* Allocates new clusters for an area that either is yet unallocated or needs a
|
|
|
|
* copy on write. If *host_offset is non-zero, clusters are only allocated if
|
|
|
|
* the new allocation can match the specified host offset.
|
|
|
|
*
|
2013-03-26 20:50:09 +04:00
|
|
|
* Note that guest_offset may not be cluster aligned. In this case, the
|
|
|
|
* returned *host_offset points to exact byte referenced by guest_offset and
|
|
|
|
* therefore isn't cluster aligned as well.
|
2013-03-26 20:50:00 +04:00
|
|
|
*
|
|
|
|
* Returns:
|
|
|
|
* 0: if no clusters could be allocated. *bytes is set to 0,
|
|
|
|
* *host_offset is left unchanged.
|
|
|
|
*
|
|
|
|
* 1: if new clusters were allocated. *bytes may be decreased if the
|
|
|
|
* new allocation doesn't cover all of the requested area.
|
|
|
|
* *host_offset is updated to contain the host offset of the first
|
|
|
|
* newly allocated cluster.
|
|
|
|
*
|
|
|
|
* -errno: in error cases
|
|
|
|
*/
|
|
|
|
static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
|
2013-03-26 20:50:03 +04:00
|
|
|
uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
|
2013-03-26 20:50:00 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-03-26 20:50:00 +04:00
|
|
|
int l2_index;
|
|
|
|
uint64_t *l2_table;
|
|
|
|
uint64_t entry;
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t nb_clusters;
|
2013-03-26 20:50:00 +04:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
uint64_t alloc_cluster_offset;
|
|
|
|
|
|
|
|
trace_qcow2_handle_alloc(qemu_coroutine_self(), guest_offset, *host_offset,
|
|
|
|
*bytes);
|
|
|
|
assert(*bytes > 0);
|
|
|
|
|
2013-03-26 20:50:01 +04:00
|
|
|
/*
|
|
|
|
* Calculate the number of clusters to look for. We stop at L2 table
|
|
|
|
* boundaries to keep things simple.
|
|
|
|
*/
|
2013-03-26 20:50:03 +04:00
|
|
|
nb_clusters =
|
|
|
|
size_to_clusters(s, offset_into_cluster(s, guest_offset) + *bytes);
|
|
|
|
|
2013-03-26 20:50:01 +04:00
|
|
|
l2_index = offset_to_l2_index(s, guest_offset);
|
2013-03-26 20:50:03 +04:00
|
|
|
nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
|
2015-09-14 17:39:47 +03:00
|
|
|
assert(nb_clusters <= INT_MAX);
|
2013-03-26 20:50:01 +04:00
|
|
|
|
2013-03-26 20:50:00 +04:00
|
|
|
/* Find L2 entry for the first involved cluster */
|
|
|
|
ret = get_cluster_table(bs, guest_offset, &l2_table, &l2_index);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:02 +04:00
|
|
|
entry = be64_to_cpu(l2_table[l2_index]);
|
2013-03-26 20:50:00 +04:00
|
|
|
|
|
|
|
/* For the moment, overwrite compressed clusters one by one */
|
|
|
|
if (entry & QCOW_OFLAG_COMPRESSED) {
|
|
|
|
nb_clusters = 1;
|
|
|
|
} else {
|
2013-03-26 20:50:02 +04:00
|
|
|
nb_clusters = count_cow_clusters(s, nb_clusters, l2_table, l2_index);
|
2013-03-26 20:50:00 +04:00
|
|
|
}
|
|
|
|
|
2013-03-27 14:43:49 +04:00
|
|
|
/* This function is only called when there were no non-COW clusters, so if
|
|
|
|
* we can't find any unallocated or COW clusters either, something is
|
|
|
|
* wrong with our code. */
|
|
|
|
assert(nb_clusters > 0);
|
|
|
|
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2013-03-26 20:50:00 +04:00
|
|
|
|
|
|
|
/* Allocate, if necessary at a given offset in the image file */
|
2013-03-26 20:50:09 +04:00
|
|
|
alloc_cluster_offset = start_of_cluster(s, *host_offset);
|
2013-03-26 20:50:04 +04:00
|
|
|
ret = do_alloc_cluster_offset(bs, guest_offset, &alloc_cluster_offset,
|
2013-03-26 20:50:00 +04:00
|
|
|
&nb_clusters);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:04 +04:00
|
|
|
/* Can't extend contiguous allocation */
|
|
|
|
if (nb_clusters == 0) {
|
2013-03-26 20:50:00 +04:00
|
|
|
*bytes = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-08-08 00:47:53 +04:00
|
|
|
/* !*host_offset would overwrite the image header and is reserved for "no
|
|
|
|
* host offset preferred". If 0 was a valid host offset, it'd trigger the
|
|
|
|
* following overlap check; do that now to avoid having an invalid value in
|
|
|
|
* *host_offset. */
|
|
|
|
if (!alloc_cluster_offset) {
|
|
|
|
ret = qcow2_pre_write_overlap_check(bs, 0, alloc_cluster_offset,
|
|
|
|
nb_clusters * s->cluster_size);
|
|
|
|
assert(ret < 0);
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2013-03-26 20:50:04 +04:00
|
|
|
/*
|
|
|
|
* Save info needed for meta data update.
|
|
|
|
*
|
2016-06-01 15:08:56 +03:00
|
|
|
* requested_bytes: Number of bytes from the start of the first
|
2013-03-26 20:50:04 +04:00
|
|
|
* newly allocated cluster to the end of the (possibly shortened
|
|
|
|
* before) write request.
|
|
|
|
*
|
2016-06-01 15:08:56 +03:00
|
|
|
* avail_bytes: Number of bytes from the start of the first
|
2013-03-26 20:50:04 +04:00
|
|
|
* newly allocated to the end of the last newly allocated cluster.
|
|
|
|
*
|
2016-06-01 15:08:56 +03:00
|
|
|
* nb_bytes: The number of bytes from the start of the first
|
2013-03-26 20:50:04 +04:00
|
|
|
* newly allocated cluster to the end of the area that the write
|
|
|
|
* request actually writes to (excluding COW at the end)
|
|
|
|
*/
|
2016-06-01 15:08:56 +03:00
|
|
|
uint64_t requested_bytes = *bytes + offset_into_cluster(s, guest_offset);
|
|
|
|
int avail_bytes = MIN(INT_MAX, nb_clusters << s->cluster_bits);
|
|
|
|
int nb_bytes = MIN(requested_bytes, avail_bytes);
|
2013-03-26 20:50:11 +04:00
|
|
|
QCowL2Meta *old_m = *m;
|
2013-03-26 20:50:04 +04:00
|
|
|
|
|
|
|
*m = g_malloc0(sizeof(**m));
|
|
|
|
|
|
|
|
**m = (QCowL2Meta) {
|
2013-03-26 20:50:11 +04:00
|
|
|
.next = old_m,
|
|
|
|
|
2013-03-26 20:50:09 +04:00
|
|
|
.alloc_offset = alloc_cluster_offset,
|
2013-03-26 20:50:04 +04:00
|
|
|
.offset = start_of_cluster(s, guest_offset),
|
|
|
|
.nb_clusters = nb_clusters,
|
|
|
|
|
|
|
|
.cow_start = {
|
|
|
|
.offset = 0,
|
2016-06-01 15:08:56 +03:00
|
|
|
.nb_bytes = offset_into_cluster(s, guest_offset),
|
2013-03-26 20:50:04 +04:00
|
|
|
},
|
|
|
|
.cow_end = {
|
2016-06-01 15:08:56 +03:00
|
|
|
.offset = nb_bytes,
|
|
|
|
.nb_bytes = avail_bytes - nb_bytes,
|
2013-03-26 20:50:04 +04:00
|
|
|
},
|
|
|
|
};
|
|
|
|
qemu_co_queue_init(&(*m)->dependent_requests);
|
|
|
|
QLIST_INSERT_HEAD(&s->cluster_allocs, *m, next_in_flight);
|
|
|
|
|
2013-03-26 20:50:09 +04:00
|
|
|
*host_offset = alloc_cluster_offset + offset_into_cluster(s, guest_offset);
|
2016-06-01 15:08:56 +03:00
|
|
|
*bytes = MIN(*bytes, nb_bytes - offset_into_cluster(s, guest_offset));
|
2013-03-26 20:50:04 +04:00
|
|
|
assert(*bytes != 0);
|
|
|
|
|
2013-03-26 20:50:00 +04:00
|
|
|
return 1;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
if (*m && (*m)->nb_clusters > 0) {
|
|
|
|
QLIST_REMOVE(*m, next_in_flight);
|
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2009-05-28 18:07:05 +04:00
|
|
|
/*
|
|
|
|
* alloc_cluster_offset
|
|
|
|
*
|
2012-03-02 17:10:54 +04:00
|
|
|
* For a given offset on the virtual disk, find the cluster offset in qcow2
|
|
|
|
* file. If the offset is not found, allocate a new cluster.
|
2009-05-28 18:07:05 +04:00
|
|
|
*
|
2012-03-02 17:10:54 +04:00
|
|
|
* If the cluster was already allocated, m->nb_clusters is set to 0 and
|
2011-09-07 18:19:17 +04:00
|
|
|
* other fields in m are meaningless.
|
2010-01-20 17:03:01 +03:00
|
|
|
*
|
|
|
|
* If the cluster is newly allocated, m->nb_clusters is set to the number of
|
2011-06-30 19:42:09 +04:00
|
|
|
* contiguous clusters that have been allocated. In this case, the other
|
|
|
|
* fields of m are valid and contain information about the first allocated
|
|
|
|
* cluster.
|
2009-05-28 18:07:05 +04:00
|
|
|
*
|
2011-06-30 19:42:09 +04:00
|
|
|
* If the request conflicts with another write request in flight, the coroutine
|
|
|
|
* is queued and will be reentered when the dependency has completed.
|
2010-01-20 17:03:01 +03:00
|
|
|
*
|
|
|
|
* Return 0 on success and -errno in error cases
|
2009-05-28 18:07:05 +04:00
|
|
|
*/
|
2010-02-02 17:20:57 +03:00
|
|
|
int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
|
2016-06-01 17:55:05 +03:00
|
|
|
unsigned int *bytes, uint64_t *host_offset,
|
|
|
|
QCowL2Meta **m)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-03-26 20:50:10 +04:00
|
|
|
uint64_t start, remaining;
|
2012-03-02 17:10:54 +04:00
|
|
|
uint64_t cluster_offset;
|
2013-03-26 20:49:58 +04:00
|
|
|
uint64_t cur_bytes;
|
2013-03-26 20:50:10 +04:00
|
|
|
int ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2016-06-01 17:55:05 +03:00
|
|
|
trace_qcow2_alloc_clusters_offset(qemu_coroutine_self(), offset, *bytes);
|
2013-03-26 20:50:10 +04:00
|
|
|
|
2012-04-24 18:10:56 +04:00
|
|
|
again:
|
2014-01-26 07:12:37 +04:00
|
|
|
start = offset;
|
2016-06-01 17:55:05 +03:00
|
|
|
remaining = *bytes;
|
2013-03-26 20:50:05 +04:00
|
|
|
cluster_offset = 0;
|
|
|
|
*host_offset = 0;
|
2013-03-27 14:43:49 +04:00
|
|
|
cur_bytes = 0;
|
|
|
|
*m = NULL;
|
2013-03-26 20:50:05 +04:00
|
|
|
|
2013-03-26 20:50:12 +04:00
|
|
|
while (true) {
|
2013-03-27 14:43:49 +04:00
|
|
|
|
|
|
|
if (!*host_offset) {
|
|
|
|
*host_offset = start_of_cluster(s, cluster_offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(remaining >= cur_bytes);
|
|
|
|
|
|
|
|
start += cur_bytes;
|
|
|
|
remaining -= cur_bytes;
|
|
|
|
cluster_offset += cur_bytes;
|
|
|
|
|
|
|
|
if (remaining == 0) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
cur_bytes = remaining;
|
|
|
|
|
2013-03-26 20:50:12 +04:00
|
|
|
/*
|
|
|
|
* Now start gathering as many contiguous clusters as possible:
|
|
|
|
*
|
|
|
|
* 1. Check for overlaps with in-flight allocations
|
|
|
|
*
|
|
|
|
* a) Overlap not in the first cluster -> shorten this request and
|
|
|
|
* let the caller handle the rest in its next loop iteration.
|
|
|
|
*
|
|
|
|
* b) Real overlaps of two requests. Yield and restart the search
|
|
|
|
* for contiguous clusters (the situation could have changed
|
|
|
|
* while we were sleeping)
|
|
|
|
*
|
|
|
|
* c) TODO: Request starts in the same cluster as the in-flight
|
|
|
|
* allocation ends. Shorten the COW of the in-fight allocation,
|
|
|
|
* set cluster_offset to write to the same cluster and set up
|
|
|
|
* the right synchronisation between the in-flight request and
|
|
|
|
* the new one.
|
|
|
|
*/
|
2013-03-27 14:43:49 +04:00
|
|
|
ret = handle_dependencies(bs, start, &cur_bytes, m);
|
2013-03-26 20:50:12 +04:00
|
|
|
if (ret == -EAGAIN) {
|
2013-03-27 14:43:49 +04:00
|
|
|
/* Currently handle_dependencies() doesn't yield if we already had
|
|
|
|
* an allocation. If it did, we would have to clean up the L2Meta
|
|
|
|
* structs before starting over. */
|
|
|
|
assert(*m == NULL);
|
2013-03-26 20:50:12 +04:00
|
|
|
goto again;
|
|
|
|
} else if (ret < 0) {
|
|
|
|
return ret;
|
2013-03-27 14:43:49 +04:00
|
|
|
} else if (cur_bytes == 0) {
|
|
|
|
break;
|
2013-03-26 20:50:12 +04:00
|
|
|
} else {
|
|
|
|
/* handle_dependencies() may have decreased cur_bytes (shortened
|
|
|
|
* the allocations below) so that the next dependency is processed
|
|
|
|
* correctly during the next loop iteration. */
|
2013-03-26 20:50:05 +04:00
|
|
|
}
|
2013-03-26 20:50:10 +04:00
|
|
|
|
2013-03-26 20:50:12 +04:00
|
|
|
/*
|
|
|
|
* 2. Count contiguous COPIED clusters.
|
|
|
|
*/
|
|
|
|
ret = handle_copied(bs, start, &cluster_offset, &cur_bytes, m);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
} else if (ret) {
|
2013-03-27 14:43:49 +04:00
|
|
|
continue;
|
2013-03-26 20:50:12 +04:00
|
|
|
} else if (cur_bytes == 0) {
|
|
|
|
break;
|
|
|
|
}
|
2012-12-07 21:08:45 +04:00
|
|
|
|
2013-03-26 20:50:12 +04:00
|
|
|
/*
|
|
|
|
* 3. If the request still hasn't completed, allocate new clusters,
|
|
|
|
* considering any cluster_offset of steps 1c or 2.
|
|
|
|
*/
|
|
|
|
ret = handle_alloc(bs, start, &cluster_offset, &cur_bytes, m);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
} else if (ret) {
|
2013-03-27 14:43:49 +04:00
|
|
|
continue;
|
2013-03-26 20:50:12 +04:00
|
|
|
} else {
|
|
|
|
assert(cur_bytes == 0);
|
|
|
|
break;
|
|
|
|
}
|
2013-03-26 20:50:01 +04:00
|
|
|
}
|
2013-03-26 20:50:00 +04:00
|
|
|
|
2016-06-01 17:55:05 +03:00
|
|
|
*bytes -= remaining;
|
|
|
|
assert(*bytes > 0);
|
2013-03-26 20:50:10 +04:00
|
|
|
assert(*host_offset != 0);
|
2009-05-28 18:07:05 +04:00
|
|
|
|
2010-01-20 17:03:01 +03:00
|
|
|
return 0;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2010-04-14 16:17:38 +04:00
|
|
|
int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
|
2009-05-28 18:07:05 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2009-05-28 18:07:05 +04:00
|
|
|
int ret, csize, nb_csectors, sector_offset;
|
|
|
|
uint64_t coffset;
|
|
|
|
|
|
|
|
coffset = cluster_offset & s->cluster_offset_mask;
|
|
|
|
if (s->cluster_cache_offset != coffset) {
|
|
|
|
nb_csectors = ((cluster_offset >> s->csize_shift) & s->csize_mask) + 1;
|
|
|
|
sector_offset = coffset & 511;
|
|
|
|
csize = nb_csectors * 512 - sector_offset;
|
2010-04-14 16:17:38 +04:00
|
|
|
BLKDBG_EVENT(bs->file, BLKDBG_READ_COMPRESSED);
|
2016-05-30 17:48:35 +03:00
|
|
|
ret = bdrv_read(bs->file, coffset >> 9, s->cluster_data,
|
2015-06-16 15:19:22 +03:00
|
|
|
nb_csectors);
|
2009-05-28 18:07:05 +04:00
|
|
|
if (ret < 0) {
|
2011-02-09 12:26:06 +03:00
|
|
|
return ret;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
if (decompress_buffer(s->cluster_cache, s->cluster_size,
|
|
|
|
s->cluster_data + sector_offset, csize) < 0) {
|
2011-02-09 12:26:06 +03:00
|
|
|
return -EIO;
|
2009-05-28 18:07:05 +04:00
|
|
|
}
|
|
|
|
s->cluster_cache_offset = coffset;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
2011-01-26 18:56:48 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This discards as many clusters of nb_clusters as possible at once (i.e.
|
|
|
|
* all clusters in the same L2 table) and returns the number of discarded
|
|
|
|
* clusters.
|
|
|
|
*/
|
|
|
|
static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t nb_clusters, enum qcow2_discard_type type,
|
|
|
|
bool full_discard)
|
2011-01-26 18:56:48 +03:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2012-03-12 20:46:51 +04:00
|
|
|
uint64_t *l2_table;
|
2011-01-26 18:56:48 +03:00
|
|
|
int l2_index;
|
|
|
|
int ret;
|
|
|
|
int i;
|
|
|
|
|
2012-03-12 20:46:51 +04:00
|
|
|
ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
|
2011-01-26 18:56:48 +03:00
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Limit nb_clusters to one L2 table */
|
|
|
|
nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
|
2015-09-14 17:39:47 +03:00
|
|
|
assert(nb_clusters <= INT_MAX);
|
2011-01-26 18:56:48 +03:00
|
|
|
|
|
|
|
for (i = 0; i < nb_clusters; i++) {
|
2014-04-29 18:12:30 +04:00
|
|
|
uint64_t old_l2_entry;
|
2011-01-26 18:56:48 +03:00
|
|
|
|
2014-04-29 18:12:30 +04:00
|
|
|
old_l2_entry = be64_to_cpu(l2_table[l2_index + i]);
|
2014-02-08 17:38:33 +04:00
|
|
|
|
|
|
|
/*
|
2014-10-24 17:57:30 +04:00
|
|
|
* If full_discard is false, make sure that a discarded area reads back
|
|
|
|
* as zeroes for v3 images (we cannot do it for v2 without actually
|
|
|
|
* writing a zero-filled buffer). We can skip the operation if the
|
|
|
|
* cluster is already marked as zero, or if it's unallocated and we
|
|
|
|
* don't have a backing file.
|
2014-02-08 17:38:33 +04:00
|
|
|
*
|
|
|
|
* TODO We might want to use bdrv_get_block_status(bs) here, but we're
|
|
|
|
* holding s->lock, so that doesn't work today.
|
2014-10-24 17:57:30 +04:00
|
|
|
*
|
|
|
|
* If full_discard is true, the sector should not read back as zeroes,
|
|
|
|
* but rather fall through to the backing file.
|
2014-02-08 17:38:33 +04:00
|
|
|
*/
|
2014-04-29 18:12:30 +04:00
|
|
|
switch (qcow2_get_cluster_type(old_l2_entry)) {
|
|
|
|
case QCOW2_CLUSTER_UNALLOCATED:
|
2015-06-17 15:55:21 +03:00
|
|
|
if (full_discard || !bs->backing) {
|
2014-04-29 18:12:30 +04:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case QCOW2_CLUSTER_ZERO:
|
2014-10-24 17:57:30 +04:00
|
|
|
if (!full_discard) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
break;
|
2014-04-29 18:12:30 +04:00
|
|
|
|
|
|
|
case QCOW2_CLUSTER_NORMAL:
|
|
|
|
case QCOW2_CLUSTER_COMPRESSED:
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
abort();
|
2011-01-26 18:56:48 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
/* First remove L2 entries */
|
2015-05-11 15:54:53 +03:00
|
|
|
qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, l2_table);
|
2014-10-24 17:57:30 +04:00
|
|
|
if (!full_discard && s->qcow_version >= 3) {
|
2014-02-08 17:38:33 +04:00
|
|
|
l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
|
|
|
|
} else {
|
|
|
|
l2_table[l2_index + i] = cpu_to_be64(0);
|
|
|
|
}
|
2011-01-26 18:56:48 +03:00
|
|
|
|
|
|
|
/* Then decrease the refcount */
|
2014-04-29 18:12:30 +04:00
|
|
|
qcow2_free_any_clusters(bs, old_l2_entry, 1, type);
|
2011-01-26 18:56:48 +03:00
|
|
|
}
|
|
|
|
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2011-01-26 18:56:48 +03:00
|
|
|
|
|
|
|
return nb_clusters;
|
|
|
|
}
|
|
|
|
|
|
|
|
int qcow2_discard_clusters(BlockDriverState *bs, uint64_t offset,
|
2014-10-24 17:57:30 +04:00
|
|
|
int nb_sectors, enum qcow2_discard_type type, bool full_discard)
|
2011-01-26 18:56:48 +03:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2011-01-26 18:56:48 +03:00
|
|
|
uint64_t end_offset;
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t nb_clusters;
|
2011-01-26 18:56:48 +03:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
end_offset = offset + (nb_sectors << BDRV_SECTOR_BITS);
|
|
|
|
|
|
|
|
/* Round start up and end down */
|
|
|
|
offset = align_offset(offset, s->cluster_size);
|
2013-12-05 10:32:34 +04:00
|
|
|
end_offset = start_of_cluster(s, end_offset);
|
2011-01-26 18:56:48 +03:00
|
|
|
|
|
|
|
if (offset > end_offset) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
nb_clusters = size_to_clusters(s, end_offset - offset);
|
|
|
|
|
2013-06-19 15:44:20 +04:00
|
|
|
s->cache_discards = true;
|
|
|
|
|
2011-01-26 18:56:48 +03:00
|
|
|
/* Each L2 table is handled by its own loop iteration */
|
|
|
|
while (nb_clusters > 0) {
|
2014-10-24 17:57:30 +04:00
|
|
|
ret = discard_single_l2(bs, offset, nb_clusters, type, full_discard);
|
2011-01-26 18:56:48 +03:00
|
|
|
if (ret < 0) {
|
2013-06-19 15:44:20 +04:00
|
|
|
goto fail;
|
2011-01-26 18:56:48 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
nb_clusters -= ret;
|
|
|
|
offset += (ret * s->cluster_size);
|
|
|
|
}
|
|
|
|
|
2013-06-19 15:44:20 +04:00
|
|
|
ret = 0;
|
|
|
|
fail:
|
|
|
|
s->cache_discards = false;
|
|
|
|
qcow2_process_discards(bs, ret);
|
|
|
|
|
|
|
|
return ret;
|
2011-01-26 18:56:48 +03:00
|
|
|
}
|
2012-03-20 18:12:58 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This zeroes as many clusters of nb_clusters as possible at once (i.e.
|
|
|
|
* all clusters in the same L2 table) and returns the number of zeroed
|
|
|
|
* clusters.
|
|
|
|
*/
|
|
|
|
static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t nb_clusters)
|
2012-03-20 18:12:58 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2012-03-20 18:12:58 +04:00
|
|
|
uint64_t *l2_table;
|
|
|
|
int l2_index;
|
|
|
|
int ret;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
|
|
|
|
if (ret < 0) {
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Limit nb_clusters to one L2 table */
|
|
|
|
nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
|
2015-09-14 17:39:47 +03:00
|
|
|
assert(nb_clusters <= INT_MAX);
|
2012-03-20 18:12:58 +04:00
|
|
|
|
|
|
|
for (i = 0; i < nb_clusters; i++) {
|
|
|
|
uint64_t old_offset;
|
|
|
|
|
|
|
|
old_offset = be64_to_cpu(l2_table[l2_index + i]);
|
|
|
|
|
|
|
|
/* Update L2 entries */
|
2015-05-11 15:54:53 +03:00
|
|
|
qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, l2_table);
|
2012-03-20 18:12:58 +04:00
|
|
|
if (old_offset & QCOW_OFLAG_COMPRESSED) {
|
|
|
|
l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
|
2013-06-19 15:44:18 +04:00
|
|
|
qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
|
2012-03-20 18:12:58 +04:00
|
|
|
} else {
|
|
|
|
l2_table[l2_index + i] |= cpu_to_be64(QCOW_OFLAG_ZERO);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2012-03-20 18:12:58 +04:00
|
|
|
|
|
|
|
return nb_clusters;
|
|
|
|
}
|
|
|
|
|
|
|
|
int qcow2_zero_clusters(BlockDriverState *bs, uint64_t offset, int nb_sectors)
|
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2015-09-14 17:39:47 +03:00
|
|
|
uint64_t nb_clusters;
|
2012-03-20 18:12:58 +04:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
/* The zero flag is only supported by version 3 and newer */
|
|
|
|
if (s->qcow_version < 3) {
|
|
|
|
return -ENOTSUP;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Each L2 table is handled by its own loop iteration */
|
|
|
|
nb_clusters = size_to_clusters(s, nb_sectors << BDRV_SECTOR_BITS);
|
|
|
|
|
2013-06-19 15:44:20 +04:00
|
|
|
s->cache_discards = true;
|
|
|
|
|
2012-03-20 18:12:58 +04:00
|
|
|
while (nb_clusters > 0) {
|
|
|
|
ret = zero_single_l2(bs, offset, nb_clusters);
|
|
|
|
if (ret < 0) {
|
2013-06-19 15:44:20 +04:00
|
|
|
goto fail;
|
2012-03-20 18:12:58 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
nb_clusters -= ret;
|
|
|
|
offset += (ret * s->cluster_size);
|
|
|
|
}
|
|
|
|
|
2013-06-19 15:44:20 +04:00
|
|
|
ret = 0;
|
|
|
|
fail:
|
|
|
|
s->cache_discards = false;
|
|
|
|
qcow2_process_discards(bs, ret);
|
|
|
|
|
|
|
|
return ret;
|
2012-03-20 18:12:58 +04:00
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Expands all zero clusters in a specific L1 table (or deallocates them, for
|
|
|
|
* non-backed non-pre-allocated zero clusters).
|
|
|
|
*
|
2014-10-27 13:12:53 +03:00
|
|
|
* l1_entries and *visited_l1_entries are used to keep track of progress for
|
|
|
|
* status_cb(). l1_entries contains the total number of L1 entries and
|
|
|
|
* *visited_l1_entries counts all visited L1 entries.
|
2013-09-03 12:09:52 +04:00
|
|
|
*/
|
|
|
|
static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
|
2014-10-27 13:12:55 +03:00
|
|
|
int l1_size, int64_t *visited_l1_entries,
|
2014-10-27 13:12:53 +03:00
|
|
|
int64_t l1_entries,
|
2015-07-27 18:51:32 +03:00
|
|
|
BlockDriverAmendStatusCB *status_cb,
|
|
|
|
void *cb_opaque)
|
2013-09-03 12:09:52 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-09-03 12:09:52 +04:00
|
|
|
bool is_active_l1 = (l1_table == s->l1_table);
|
|
|
|
uint64_t *l2_table = NULL;
|
|
|
|
int ret;
|
|
|
|
int i, j;
|
|
|
|
|
|
|
|
if (!is_active_l1) {
|
|
|
|
/* inactive L2 tables require a buffer to be stored in when loading
|
|
|
|
* them from disk */
|
2015-06-16 15:19:22 +03:00
|
|
|
l2_table = qemu_try_blockalign(bs->file->bs, s->cluster_size);
|
2014-05-20 19:12:47 +04:00
|
|
|
if (l2_table == NULL) {
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < l1_size; i++) {
|
|
|
|
uint64_t l2_offset = l1_table[i] & L1E_OFFSET_MASK;
|
|
|
|
bool l2_dirty = false;
|
2015-02-10 23:28:48 +03:00
|
|
|
uint64_t l2_refcount;
|
2013-09-03 12:09:52 +04:00
|
|
|
|
|
|
|
if (!l2_offset) {
|
|
|
|
/* unallocated */
|
2014-10-27 13:12:53 +03:00
|
|
|
(*visited_l1_entries)++;
|
|
|
|
if (status_cb) {
|
2015-07-27 18:51:32 +03:00
|
|
|
status_cb(bs, *visited_l1_entries, l1_entries, cb_opaque);
|
2014-10-27 13:12:53 +03:00
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2015-01-19 23:49:03 +03:00
|
|
|
if (offset_into_cluster(s, l2_offset)) {
|
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "L2 table offset %#"
|
|
|
|
PRIx64 " unaligned (L1 index: %#x)",
|
|
|
|
l2_offset, i);
|
|
|
|
ret = -EIO;
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2013-09-03 12:09:52 +04:00
|
|
|
if (is_active_l1) {
|
|
|
|
/* get active L2 tables from cache */
|
|
|
|
ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset,
|
|
|
|
(void **)&l2_table);
|
|
|
|
} else {
|
|
|
|
/* load inactive L2 tables from disk */
|
2016-05-30 17:48:35 +03:00
|
|
|
ret = bdrv_read(bs->file, l2_offset / BDRV_SECTOR_SIZE,
|
2015-06-16 15:19:22 +03:00
|
|
|
(void *)l2_table, s->cluster_sectors);
|
2013-09-03 12:09:52 +04:00
|
|
|
}
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2015-02-10 23:28:46 +03:00
|
|
|
ret = qcow2_get_refcount(bs, l2_offset >> s->cluster_bits,
|
|
|
|
&l2_refcount);
|
|
|
|
if (ret < 0) {
|
2014-10-27 13:12:55 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2013-09-03 12:09:52 +04:00
|
|
|
for (j = 0; j < s->l2_size; j++) {
|
|
|
|
uint64_t l2_entry = be64_to_cpu(l2_table[j]);
|
2014-10-27 13:12:55 +03:00
|
|
|
int64_t offset = l2_entry & L2E_OFFSET_MASK;
|
2013-09-03 12:09:52 +04:00
|
|
|
int cluster_type = qcow2_get_cluster_type(l2_entry);
|
2013-09-27 12:21:48 +04:00
|
|
|
bool preallocated = offset != 0;
|
2013-09-03 12:09:52 +04:00
|
|
|
|
2014-10-27 13:12:55 +03:00
|
|
|
if (cluster_type != QCOW2_CLUSTER_ZERO) {
|
2013-09-03 12:09:52 +04:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2013-09-27 12:21:48 +04:00
|
|
|
if (!preallocated) {
|
2015-06-17 15:55:21 +03:00
|
|
|
if (!bs->backing) {
|
2013-09-03 12:09:52 +04:00
|
|
|
/* not backed; therefore we can simply deallocate the
|
|
|
|
* cluster */
|
|
|
|
l2_table[j] = 0;
|
|
|
|
l2_dirty = true;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
offset = qcow2_alloc_clusters(bs, s->cluster_size);
|
|
|
|
if (offset < 0) {
|
|
|
|
ret = offset;
|
|
|
|
goto fail;
|
|
|
|
}
|
2014-10-27 13:12:55 +03:00
|
|
|
|
|
|
|
if (l2_refcount > 1) {
|
|
|
|
/* For shared L2 tables, set the refcount accordingly (it is
|
|
|
|
* already 1 and needs to be l2_refcount) */
|
|
|
|
ret = qcow2_update_cluster_refcount(bs,
|
2015-02-10 23:28:47 +03:00
|
|
|
offset >> s->cluster_bits,
|
|
|
|
refcount_diff(1, l2_refcount), false,
|
2014-10-27 13:12:55 +03:00
|
|
|
QCOW2_DISCARD_OTHER);
|
|
|
|
if (ret < 0) {
|
|
|
|
qcow2_free_clusters(bs, offset, s->cluster_size,
|
|
|
|
QCOW2_DISCARD_OTHER);
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
}
|
|
|
|
|
2015-01-19 23:49:03 +03:00
|
|
|
if (offset_into_cluster(s, offset)) {
|
|
|
|
qcow2_signal_corruption(bs, true, -1, -1, "Data cluster offset "
|
|
|
|
"%#" PRIx64 " unaligned (L2 offset: %#"
|
|
|
|
PRIx64 ", L2 index: %#x)", offset,
|
|
|
|
l2_offset, j);
|
|
|
|
if (!preallocated) {
|
|
|
|
qcow2_free_clusters(bs, offset, s->cluster_size,
|
|
|
|
QCOW2_DISCARD_ALWAYS);
|
|
|
|
}
|
|
|
|
ret = -EIO;
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2013-10-10 13:09:23 +04:00
|
|
|
ret = qcow2_pre_write_overlap_check(bs, 0, offset, s->cluster_size);
|
2013-09-03 12:09:52 +04:00
|
|
|
if (ret < 0) {
|
2013-09-27 12:21:48 +04:00
|
|
|
if (!preallocated) {
|
|
|
|
qcow2_free_clusters(bs, offset, s->cluster_size,
|
|
|
|
QCOW2_DISCARD_ALWAYS);
|
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2016-06-16 16:13:15 +03:00
|
|
|
ret = bdrv_pwrite_zeroes(bs->file, offset, s->cluster_size, 0);
|
2013-09-03 12:09:52 +04:00
|
|
|
if (ret < 0) {
|
2013-09-27 12:21:48 +04:00
|
|
|
if (!preallocated) {
|
|
|
|
qcow2_free_clusters(bs, offset, s->cluster_size,
|
|
|
|
QCOW2_DISCARD_ALWAYS);
|
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2014-10-27 13:12:55 +03:00
|
|
|
if (l2_refcount == 1) {
|
|
|
|
l2_table[j] = cpu_to_be64(offset | QCOW_OFLAG_COPIED);
|
|
|
|
} else {
|
|
|
|
l2_table[j] = cpu_to_be64(offset);
|
2013-09-25 14:07:22 +04:00
|
|
|
}
|
2014-10-27 13:12:55 +03:00
|
|
|
l2_dirty = true;
|
2013-09-03 12:09:52 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
if (is_active_l1) {
|
|
|
|
if (l2_dirty) {
|
2015-05-11 15:54:53 +03:00
|
|
|
qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, l2_table);
|
2013-09-03 12:09:52 +04:00
|
|
|
qcow2_cache_depends_on_flush(s->l2_table_cache);
|
|
|
|
}
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2013-09-03 12:09:52 +04:00
|
|
|
} else {
|
|
|
|
if (l2_dirty) {
|
2013-10-10 13:09:23 +04:00
|
|
|
ret = qcow2_pre_write_overlap_check(bs,
|
|
|
|
QCOW2_OL_INACTIVE_L2 | QCOW2_OL_ACTIVE_L2, l2_offset,
|
2013-09-03 12:09:52 +04:00
|
|
|
s->cluster_size);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2016-05-31 15:42:08 +03:00
|
|
|
ret = bdrv_write(bs->file, l2_offset / BDRV_SECTOR_SIZE,
|
2015-06-16 15:19:22 +03:00
|
|
|
(void *)l2_table, s->cluster_sectors);
|
2013-09-03 12:09:52 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2014-10-27 13:12:53 +03:00
|
|
|
|
|
|
|
(*visited_l1_entries)++;
|
|
|
|
if (status_cb) {
|
2015-07-27 18:51:32 +03:00
|
|
|
status_cb(bs, *visited_l1_entries, l1_entries, cb_opaque);
|
2014-10-27 13:12:53 +03:00
|
|
|
}
|
2013-09-03 12:09:52 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
if (l2_table) {
|
|
|
|
if (!is_active_l1) {
|
|
|
|
qemu_vfree(l2_table);
|
|
|
|
} else {
|
2015-05-11 15:54:58 +03:00
|
|
|
qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);
|
2013-09-03 12:09:52 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* For backed images, expands all zero clusters on the image. For non-backed
|
|
|
|
* images, deallocates all non-pre-allocated zero clusters (and claims the
|
|
|
|
* allocation for pre-allocated ones). This is important for downgrading to a
|
|
|
|
* qcow2 version which doesn't yet support metadata zero clusters.
|
|
|
|
*/
|
2014-10-27 13:12:53 +03:00
|
|
|
int qcow2_expand_zero_clusters(BlockDriverState *bs,
|
2015-07-27 18:51:32 +03:00
|
|
|
BlockDriverAmendStatusCB *status_cb,
|
|
|
|
void *cb_opaque)
|
2013-09-03 12:09:52 +04:00
|
|
|
{
|
2015-09-07 18:12:56 +03:00
|
|
|
BDRVQcow2State *s = bs->opaque;
|
2013-09-03 12:09:52 +04:00
|
|
|
uint64_t *l1_table = NULL;
|
2014-10-27 13:12:53 +03:00
|
|
|
int64_t l1_entries = 0, visited_l1_entries = 0;
|
2013-09-03 12:09:52 +04:00
|
|
|
int ret;
|
|
|
|
int i, j;
|
|
|
|
|
2014-10-27 13:12:53 +03:00
|
|
|
if (status_cb) {
|
|
|
|
l1_entries = s->l1_size;
|
|
|
|
for (i = 0; i < s->nb_snapshots; i++) {
|
|
|
|
l1_entries += s->snapshots[i].l1_size;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-09-03 12:09:52 +04:00
|
|
|
ret = expand_zero_clusters_in_l1(bs, s->l1_table, s->l1_size,
|
2014-10-27 13:12:53 +03:00
|
|
|
&visited_l1_entries, l1_entries,
|
2015-07-27 18:51:32 +03:00
|
|
|
status_cb, cb_opaque);
|
2013-09-03 12:09:52 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Inactive L1 tables may point to active L2 tables - therefore it is
|
|
|
|
* necessary to flush the L2 table cache before trying to access the L2
|
|
|
|
* tables pointed to by inactive L1 entries (else we might try to expand
|
|
|
|
* zero clusters that have already been expanded); furthermore, it is also
|
|
|
|
* necessary to empty the L2 table cache, since it may contain tables which
|
|
|
|
* are now going to be modified directly on disk, bypassing the cache.
|
|
|
|
* qcow2_cache_empty() does both for us. */
|
|
|
|
ret = qcow2_cache_empty(bs, s->l2_table_cache);
|
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < s->nb_snapshots; i++) {
|
2016-05-31 19:35:52 +03:00
|
|
|
int l1_sectors = DIV_ROUND_UP(s->snapshots[i].l1_size *
|
|
|
|
sizeof(uint64_t), BDRV_SECTOR_SIZE);
|
2013-09-03 12:09:52 +04:00
|
|
|
|
|
|
|
l1_table = g_realloc(l1_table, l1_sectors * BDRV_SECTOR_SIZE);
|
|
|
|
|
2016-05-30 17:48:35 +03:00
|
|
|
ret = bdrv_read(bs->file,
|
2015-06-16 15:19:22 +03:00
|
|
|
s->snapshots[i].l1_table_offset / BDRV_SECTOR_SIZE,
|
|
|
|
(void *)l1_table, l1_sectors);
|
2013-09-03 12:09:52 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (j = 0; j < s->snapshots[i].l1_size; j++) {
|
|
|
|
be64_to_cpus(&l1_table[j]);
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = expand_zero_clusters_in_l1(bs, l1_table, s->snapshots[i].l1_size,
|
2014-10-27 13:12:53 +03:00
|
|
|
&visited_l1_entries, l1_entries,
|
2015-07-27 18:51:32 +03:00
|
|
|
status_cb, cb_opaque);
|
2013-09-03 12:09:52 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
g_free(l1_table);
|
|
|
|
return ret;
|
|
|
|
}
|