* [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing
@ 2014-10-22 12:09 Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0() Max Reitz
` (17 more replies)
0 siblings, 18 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
As can be seen in the final patch of this series, there are certain
cases where the current repair implementation of qcow2 actually damages
the image further because it allocates new clusters for the refcount
structure which overlap with existing but according to the on-disk
refcounts (which are assumed to be wrong to begin with) unallocated
clusters.
This series fixes this by completely recreating the refcount structure
based on the in-memory information calculated during the check operation
if the possibility of damaging the image while repairing the refcount
structures in-place exists.
v8:
- Patches 7 to 10 let inc_refcounts() resize the reftable; if this isn't
done, rebuilding the refcount structures may lead to data clusters
conflicting with refcount structures [Kevin]
- Patch 7: Just use int64_t instead of int throughout all check
functions for the in-memory refcount table size so that the
following patches are smaller
- Patch 8: The fail path in the L1 and L2 check functions was not a
pure cleanup path, but also hard-coded what errors might occur (only
EIO), this fixes that
- Patch 9: Patch 10 requires inc_refcounts() to be able to return an
error, so this patch prepares for that
- Patch 10: Allows inc_refcounts() to resize the in-memory refcount
table
- Patch 12 (prev. 8):
- Drop a superfluous validation [Kevin]
- Fix leak of *refcount_table due to g_try_realloc() (which I forgot
to do here in v7) [Kevin]
- Rebase conflicts due to patch 9
- Patch 13 (prev. 9): Rebase conflicts due to patch 9
- Patch 14 (prev. 10): Keep *imrt_nb_clusters in case g_try_realloc()
fails [Kevin]
- Patch 15 (prev. 11): res->leaks does not account for fixed leaks, so
don't subtract them to get the number of unfixed leaks [Kevin]
git-backport-diff against v7:
Key:
[----] : patches are identical
[####] : number of functional differences between upstream/downstream patch
[down] : patch is downstream-only
The flags [FC] indicate (F)unctional and (C)ontextual differences, respectively
001/17:[----] [--] 'block: Add qemu_{,try_}blockalign0()'
002/17:[----] [--] 'qcow2: Calculate refcount block entry count'
003/17:[----] [--] 'qcow2: Fix leaks in dirty images'
004/17:[----] [--] 'qcow2: Split qcow2_check_refcounts()'
005/17:[----] [--] 'qcow2: Use sizeof(**refcount_table)'
006/17:[----] [--] 'qcow2: Pull check_refblocks() up'
007/17:[down] 'qcow2: Use int64_t for in-memory reftable size'
008/17:[down] 'qcow2: Split fail code in L1 and L2 checks'
009/17:[down] 'qcow2: Let inc_refcounts() return -errno'
010/17:[down] 'qcow2: Let inc_refcounts() resize the reftable'
011/17:[----] [--] 'qcow2: Reuse refcount table in calculate_refcounts()'
012/17:[0022] [FC] 'qcow2: Fix refcount blocks beyond image end'
013/17:[0007] [FC] 'qcow2: Do not perform potentially damaging repairs'
014/17:[0001] [FC] 'qcow2: Rebuild refcount structure during check'
015/17:[0002] [FC] 'qcow2: Clean up after refcount rebuild'
016/17:[----] [--] 'iotests: Fix test outputs'
017/17:[----] [--] 'iotests: Add test for potentially damaging repairs'
Max Reitz (17):
block: Add qemu_{,try_}blockalign0()
qcow2: Calculate refcount block entry count
qcow2: Fix leaks in dirty images
qcow2: Split qcow2_check_refcounts()
qcow2: Use sizeof(**refcount_table)
qcow2: Pull check_refblocks() up
qcow2: Use int64_t for in-memory reftable size
qcow2: Split fail code in L1 and L2 checks
qcow2: Let inc_refcounts() return -errno
qcow2: Let inc_refcounts() resize the reftable
qcow2: Reuse refcount table in calculate_refcounts()
qcow2: Fix refcount blocks beyond image end
qcow2: Do not perform potentially damaging repairs
qcow2: Rebuild refcount structure during check
qcow2: Clean up after refcount rebuild
iotests: Fix test outputs
iotests: Add test for potentially damaging repairs
block.c | 16 +
block/qcow2-refcount.c | 841 ++++++++++++++++++++++++++++++++-------------
block/qcow2.c | 5 +-
block/qcow2.h | 2 +
include/block/block.h | 2 +
tests/qemu-iotests/039.out | 10 +-
tests/qemu-iotests/060.out | 10 +-
tests/qemu-iotests/061.out | 18 +-
tests/qemu-iotests/108 | 141 ++++++++
tests/qemu-iotests/108.out | 110 ++++++
tests/qemu-iotests/group | 1 +
11 files changed, 907 insertions(+), 249 deletions(-)
create mode 100755 tests/qemu-iotests/108
create mode 100644 tests/qemu-iotests/108.out
--
1.9.3
^ permalink raw reply [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0()
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 14:02 ` Eric Blake
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 02/17] qcow2: Calculate refcount block entry count Max Reitz
` (16 subsequent siblings)
17 siblings, 1 reply; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
These functions call their non-0-counterparts and then fill the
allocated buffer with 0 (if the allocation has been successful).
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block.c | 16 ++++++++++++++++
include/block/block.h | 2 ++
2 files changed, 18 insertions(+)
diff --git a/block.c b/block.c
index 773e87e..bbb04e7 100644
--- a/block.c
+++ b/block.c
@@ -5200,6 +5200,11 @@ void *qemu_blockalign(BlockDriverState *bs, size_t size)
return qemu_memalign(bdrv_opt_mem_align(bs), size);
}
+void *qemu_blockalign0(BlockDriverState *bs, size_t size)
+{
+ return memset(qemu_blockalign(bs, size), 0, size);
+}
+
void *qemu_try_blockalign(BlockDriverState *bs, size_t size)
{
size_t align = bdrv_opt_mem_align(bs);
@@ -5213,6 +5218,17 @@ void *qemu_try_blockalign(BlockDriverState *bs, size_t size)
return qemu_try_memalign(align, size);
}
+void *qemu_try_blockalign0(BlockDriverState *bs, size_t size)
+{
+ void *mem = qemu_try_blockalign(bs, size);
+
+ if (mem) {
+ memset(mem, 0, size);
+ }
+
+ return mem;
+}
+
/*
* Check if all memory in this vector is sector aligned.
*/
diff --git a/include/block/block.h b/include/block/block.h
index c9ec0ab..341054d 100644
--- a/include/block/block.h
+++ b/include/block/block.h
@@ -412,7 +412,9 @@ void bdrv_img_create(const char *filename, const char *fmt,
size_t bdrv_opt_mem_align(BlockDriverState *bs);
void bdrv_set_guest_block_size(BlockDriverState *bs, int align);
void *qemu_blockalign(BlockDriverState *bs, size_t size);
+void *qemu_blockalign0(BlockDriverState *bs, size_t size);
void *qemu_try_blockalign(BlockDriverState *bs, size_t size);
+void *qemu_try_blockalign0(BlockDriverState *bs, size_t size);
bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov);
struct HBitmapIter;
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 02/17] qcow2: Calculate refcount block entry count
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0() Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 03/17] qcow2: Fix leaks in dirty images Max Reitz
` (15 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
The size of a refblock entry is (in theory) variable; calculate
therefore the number of entries per refblock and the according bit shift
(1 << x == entry count) when opening an image.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
---
block/qcow2.c | 3 +++
block/qcow2.h | 2 ++
2 files changed, 5 insertions(+)
diff --git a/block/qcow2.c b/block/qcow2.c
index 156a219..3c8b881 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -698,6 +698,9 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
s->l2_bits = s->cluster_bits - 3; /* L2 is always one cluster */
s->l2_size = 1 << s->l2_bits;
+ /* 2^(s->refcount_order - 3) is the refcount width in bytes */
+ s->refcount_block_bits = s->cluster_bits - (s->refcount_order - 3);
+ s->refcount_block_size = 1 << s->refcount_block_bits;
bs->total_sectors = header.size / 512;
s->csize_shift = (62 - (s->cluster_bits - 8));
s->csize_mask = (1 << (s->cluster_bits - 8)) - 1;
diff --git a/block/qcow2.h b/block/qcow2.h
index 7d61e61..47cd4b3 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -223,6 +223,8 @@ typedef struct BDRVQcowState {
int l2_size;
int l1_size;
int l1_vm_state_index;
+ int refcount_block_bits;
+ int refcount_block_size;
int csize_shift;
int csize_mask;
uint64_t cluster_offset_mask;
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 03/17] qcow2: Fix leaks in dirty images
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0() Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 02/17] qcow2: Calculate refcount block entry count Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 04/17] qcow2: Split qcow2_check_refcounts() Max Reitz
` (14 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
When opening dirty images, qcow2's repair function should not only
repair errors but leaks as well.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
---
block/qcow2.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/block/qcow2.c b/block/qcow2.c
index 3c8b881..7a2c66f 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -910,7 +910,7 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
(s->incompatible_features & QCOW2_INCOMPAT_DIRTY)) {
BdrvCheckResult result = {0};
- ret = qcow2_check(bs, &result, BDRV_FIX_ERRORS);
+ ret = qcow2_check(bs, &result, BDRV_FIX_ERRORS | BDRV_FIX_LEAKS);
if (ret < 0) {
error_setg_errno(errp, -ret, "Could not repair dirty image");
goto fail;
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 04/17] qcow2: Split qcow2_check_refcounts()
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (2 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 03/17] qcow2: Fix leaks in dirty images Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 05/17] qcow2: Use sizeof(**refcount_table) Max Reitz
` (13 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
Put the code for calculating the reference counts and comparing them
during qemu-img check into own functions.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
---
block/qcow2-refcount.c | 153 ++++++++++++++++++++++++++++++++-----------------
1 file changed, 102 insertions(+), 51 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index c31d85a..e5f7876 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1535,71 +1535,70 @@ done:
return new_offset;
}
+static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
+ BdrvCheckMode fix, uint16_t **refcount_table,
+ int64_t *nb_clusters);
+
/*
- * Checks an image for refcount consistency.
- *
- * Returns 0 if no errors are found, the number of errors in case the image is
- * detected as corrupted, and -errno when an internal error occurred.
+ * Calculates an in-memory refcount table.
*/
-int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
- BdrvCheckMode fix)
+static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
+ BdrvCheckMode fix, uint16_t **refcount_table,
+ int64_t *nb_clusters)
{
BDRVQcowState *s = bs->opaque;
- int64_t size, i, highest_cluster, nb_clusters;
- int refcount1, refcount2;
+ int64_t i;
QCowSnapshot *sn;
- uint16_t *refcount_table;
int ret;
- size = bdrv_getlength(bs->file);
- if (size < 0) {
- res->check_errors++;
- return size;
- }
-
- nb_clusters = size_to_clusters(s, size);
- if (nb_clusters > INT_MAX) {
- res->check_errors++;
- return -EFBIG;
- }
-
- refcount_table = g_try_new0(uint16_t, nb_clusters);
- if (nb_clusters && refcount_table == NULL) {
+ *refcount_table = g_try_new0(uint16_t, *nb_clusters);
+ if (*nb_clusters && *refcount_table == NULL) {
res->check_errors++;
return -ENOMEM;
}
- res->bfi.total_clusters =
- size_to_clusters(s, bs->total_sectors * BDRV_SECTOR_SIZE);
-
/* header */
- inc_refcounts(bs, res, refcount_table, nb_clusters,
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
0, s->cluster_size);
/* current L1 table */
- ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
+ ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
s->l1_table_offset, s->l1_size, CHECK_FRAG_INFO);
if (ret < 0) {
- goto fail;
+ return ret;
}
/* snapshots */
- for(i = 0; i < s->nb_snapshots; i++) {
+ for (i = 0; i < s->nb_snapshots; i++) {
sn = s->snapshots + i;
- ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
+ ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
sn->l1_table_offset, sn->l1_size, 0);
if (ret < 0) {
- goto fail;
+ return ret;
}
}
- inc_refcounts(bs, res, refcount_table, nb_clusters,
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
s->snapshots_offset, s->snapshots_size);
/* refcount data */
- inc_refcounts(bs, res, refcount_table, nb_clusters,
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
s->refcount_table_offset,
s->refcount_table_size * sizeof(uint64_t));
+ return check_refblocks(bs, res, fix, refcount_table, nb_clusters);
+}
+
+/*
+ * Checks consistency of refblocks and accounts for each refblock in
+ * *refcount_table.
+ */
+static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
+ BdrvCheckMode fix, uint16_t **refcount_table,
+ int64_t *nb_clusters)
+{
+ BDRVQcowState *s = bs->opaque;
+ int64_t i;
+
for(i = 0; i < s->refcount_table_size; i++) {
uint64_t offset, cluster;
offset = s->refcount_table[i];
@@ -1613,7 +1612,7 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
continue;
}
- if (cluster >= nb_clusters) {
+ if (cluster >= *nb_clusters) {
fprintf(stderr, "ERROR refcount block %" PRId64
" is outside image\n", i);
res->corruptions++;
@@ -1621,14 +1620,14 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
if (offset != 0) {
- inc_refcounts(bs, res, refcount_table, nb_clusters,
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
offset, s->cluster_size);
- if (refcount_table[cluster] != 1) {
+ if ((*refcount_table)[cluster] != 1) {
fprintf(stderr, "%s refcount block %" PRId64
" refcount=%d\n",
fix & BDRV_FIX_ERRORS ? "Repairing" :
"ERROR",
- i, refcount_table[cluster]);
+ i, (*refcount_table)[cluster]);
if (fix & BDRV_FIX_ERRORS) {
int64_t new_offset;
@@ -1640,17 +1639,18 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
/* update refcounts */
- if ((new_offset >> s->cluster_bits) >= nb_clusters) {
+ if ((new_offset >> s->cluster_bits) >= *nb_clusters) {
/* increase refcount_table size if necessary */
- int old_nb_clusters = nb_clusters;
- nb_clusters = (new_offset >> s->cluster_bits) + 1;
- refcount_table = g_renew(uint16_t, refcount_table,
- nb_clusters);
- memset(&refcount_table[old_nb_clusters], 0, (nb_clusters
- - old_nb_clusters) * sizeof(uint16_t));
+ int old_nb_clusters = *nb_clusters;
+ *nb_clusters = (new_offset >> s->cluster_bits) + 1;
+ *refcount_table = g_renew(uint16_t, *refcount_table,
+ *nb_clusters);
+ memset(&(*refcount_table)[old_nb_clusters], 0,
+ (*nb_clusters - old_nb_clusters) *
+ sizeof(uint16_t));
}
- refcount_table[cluster]--;
- inc_refcounts(bs, res, refcount_table, nb_clusters,
+ (*refcount_table)[cluster]--;
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
new_offset, s->cluster_size);
res->corruptions_fixed++;
@@ -1661,8 +1661,22 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
}
- /* compare ref counts */
- for (i = 0, highest_cluster = 0; i < nb_clusters; i++) {
+ return 0;
+}
+
+/*
+ * Compares the actual reference count for each cluster in the image against the
+ * refcount as reported by the refcount structures on-disk.
+ */
+static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
+ BdrvCheckMode fix, int64_t *highest_cluster,
+ uint16_t *refcount_table, int64_t nb_clusters)
+{
+ BDRVQcowState *s = bs->opaque;
+ int64_t i;
+ int refcount1, refcount2, ret;
+
+ for (i = 0, *highest_cluster = 0; i < nb_clusters; i++) {
refcount1 = get_refcount(bs, i);
if (refcount1 < 0) {
fprintf(stderr, "Can't get refcount for cluster %" PRId64 ": %s\n",
@@ -1674,11 +1688,10 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
refcount2 = refcount_table[i];
if (refcount1 > 0 || refcount2 > 0) {
- highest_cluster = i;
+ *highest_cluster = i;
}
if (refcount1 != refcount2) {
-
/* Check if we're allowed to fix the mismatch */
int *num_fixed = NULL;
if (refcount1 > refcount2 && (fix & BDRV_FIX_LEAKS)) {
@@ -1711,6 +1724,44 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
}
}
+}
+
+/*
+ * Checks an image for refcount consistency.
+ *
+ * Returns 0 if no errors are found, the number of errors in case the image is
+ * detected as corrupted, and -errno when an internal error occurred.
+ */
+int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
+ BdrvCheckMode fix)
+{
+ BDRVQcowState *s = bs->opaque;
+ int64_t size, highest_cluster, nb_clusters;
+ uint16_t *refcount_table;
+ int ret;
+
+ size = bdrv_getlength(bs->file);
+ if (size < 0) {
+ res->check_errors++;
+ return size;
+ }
+
+ nb_clusters = size_to_clusters(s, size);
+ if (nb_clusters > INT_MAX) {
+ res->check_errors++;
+ return -EFBIG;
+ }
+
+ res->bfi.total_clusters =
+ size_to_clusters(s, bs->total_sectors * BDRV_SECTOR_SIZE);
+
+ ret = calculate_refcounts(bs, res, fix, &refcount_table, &nb_clusters);
+ if (ret < 0) {
+ goto fail;
+ }
+
+ compare_refcounts(bs, res, fix, &highest_cluster, refcount_table,
+ nb_clusters);
/* check OFLAG_COPIED */
ret = check_oflag_copied(bs, res, fix);
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 05/17] qcow2: Use sizeof(**refcount_table)
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (3 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 04/17] qcow2: Split qcow2_check_refcounts() Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 14:38 ` Eric Blake
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 06/17] qcow2: Pull check_refblocks() up Max Reitz
` (12 subsequent siblings)
17 siblings, 1 reply; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
When implementing variable refcounts, we want to be able to easily find
all the places in qemu which are tied to a certain refcount order.
Replace sizeof(uint16_t) in the check code by sizeof(**refcount_table)
so we can later find it more easily.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index e5f7876..e8b9df9 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1647,7 +1647,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
*nb_clusters);
memset(&(*refcount_table)[old_nb_clusters], 0,
(*nb_clusters - old_nb_clusters) *
- sizeof(uint16_t));
+ sizeof(**refcount_table));
}
(*refcount_table)[cluster]--;
inc_refcounts(bs, res, *refcount_table, *nb_clusters,
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 06/17] qcow2: Pull check_refblocks() up
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (4 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 05/17] qcow2: Use sizeof(**refcount_table) Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 07/17] qcow2: Use int64_t for in-memory reftable size Max Reitz
` (11 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
Pull check_refblocks() before calculate_refcounts() so we can drop its
static declaration.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
---
block/qcow2-refcount.c | 102 ++++++++++++++++++++++++-------------------------
1 file changed, 49 insertions(+), 53 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index e8b9df9..24f297f 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1535,59 +1535,6 @@ done:
return new_offset;
}
-static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
- BdrvCheckMode fix, uint16_t **refcount_table,
- int64_t *nb_clusters);
-
-/*
- * Calculates an in-memory refcount table.
- */
-static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
- BdrvCheckMode fix, uint16_t **refcount_table,
- int64_t *nb_clusters)
-{
- BDRVQcowState *s = bs->opaque;
- int64_t i;
- QCowSnapshot *sn;
- int ret;
-
- *refcount_table = g_try_new0(uint16_t, *nb_clusters);
- if (*nb_clusters && *refcount_table == NULL) {
- res->check_errors++;
- return -ENOMEM;
- }
-
- /* header */
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- 0, s->cluster_size);
-
- /* current L1 table */
- ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
- s->l1_table_offset, s->l1_size, CHECK_FRAG_INFO);
- if (ret < 0) {
- return ret;
- }
-
- /* snapshots */
- for (i = 0; i < s->nb_snapshots; i++) {
- sn = s->snapshots + i;
- ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
- sn->l1_table_offset, sn->l1_size, 0);
- if (ret < 0) {
- return ret;
- }
- }
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- s->snapshots_offset, s->snapshots_size);
-
- /* refcount data */
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- s->refcount_table_offset,
- s->refcount_table_size * sizeof(uint64_t));
-
- return check_refblocks(bs, res, fix, refcount_table, nb_clusters);
-}
-
/*
* Checks consistency of refblocks and accounts for each refblock in
* *refcount_table.
@@ -1665,6 +1612,55 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
}
/*
+ * Calculates an in-memory refcount table.
+ */
+static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
+ BdrvCheckMode fix, uint16_t **refcount_table,
+ int64_t *nb_clusters)
+{
+ BDRVQcowState *s = bs->opaque;
+ int64_t i;
+ QCowSnapshot *sn;
+ int ret;
+
+ *refcount_table = g_try_new0(uint16_t, *nb_clusters);
+ if (*nb_clusters && *refcount_table == NULL) {
+ res->check_errors++;
+ return -ENOMEM;
+ }
+
+ /* header */
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ 0, s->cluster_size);
+
+ /* current L1 table */
+ ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
+ s->l1_table_offset, s->l1_size, CHECK_FRAG_INFO);
+ if (ret < 0) {
+ return ret;
+ }
+
+ /* snapshots */
+ for (i = 0; i < s->nb_snapshots; i++) {
+ sn = s->snapshots + i;
+ ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
+ sn->l1_table_offset, sn->l1_size, 0);
+ if (ret < 0) {
+ return ret;
+ }
+ }
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ s->snapshots_offset, s->snapshots_size);
+
+ /* refcount data */
+ inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ s->refcount_table_offset,
+ s->refcount_table_size * sizeof(uint64_t));
+
+ return check_refblocks(bs, res, fix, refcount_table, nb_clusters);
+}
+
+/*
* Compares the actual reference count for each cluster in the image against the
* refcount as reported by the refcount structures on-disk.
*/
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 07/17] qcow2: Use int64_t for in-memory reftable size
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (5 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 06/17] qcow2: Pull check_refblocks() up Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 08/17] qcow2: Split fail code in L1 and L2 checks Max Reitz
` (10 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
Use int64_t for the entry count of the in-memory refcount table
throughout the check functions.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 24f297f..a3f4d47 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1084,7 +1084,7 @@ fail:
static void inc_refcounts(BlockDriverState *bs,
BdrvCheckResult *res,
uint16_t *refcount_table,
- int refcount_table_size,
+ int64_t refcount_table_size,
int64_t offset, int64_t size)
{
BDRVQcowState *s = bs->opaque;
@@ -1127,7 +1127,7 @@ enum {
* error occurred.
*/
static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
- uint16_t *refcount_table, int refcount_table_size, int64_t l2_offset,
+ uint16_t *refcount_table, int64_t refcount_table_size, int64_t l2_offset,
int flags)
{
BDRVQcowState *s = bs->opaque;
@@ -1237,7 +1237,7 @@ fail:
static int check_refcounts_l1(BlockDriverState *bs,
BdrvCheckResult *res,
uint16_t *refcount_table,
- int refcount_table_size,
+ int64_t refcount_table_size,
int64_t l1_table_offset, int l1_size,
int flags)
{
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 08/17] qcow2: Split fail code in L1 and L2 checks
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (6 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 07/17] qcow2: Use int64_t for in-memory reftable size Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 09/17] qcow2: Let inc_refcounts() return -errno Max Reitz
` (9 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
Instead of printing out an error message, incrementing check_errors and
returning a fixed -errno, just do cleanups and return -ret, with ret set
by the code which threw the exception (jumped to the fail label).
Also, increment check_errors on error in check_refcounts_l2().
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 23 ++++++++++++++---------
1 file changed, 14 insertions(+), 9 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index a3f4d47..59fb400 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1133,14 +1133,18 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
BDRVQcowState *s = bs->opaque;
uint64_t *l2_table, l2_entry;
uint64_t next_contiguous_offset = 0;
- int i, l2_size, nb_csectors;
+ int i, l2_size, nb_csectors, ret;
/* Read L2 table from disk */
l2_size = s->l2_size * sizeof(uint64_t);
l2_table = g_malloc(l2_size);
- if (bdrv_pread(bs->file, l2_offset, l2_table, l2_size) != l2_size)
+ ret = bdrv_pread(bs->file, l2_offset, l2_table, l2_size);
+ if (ret < 0) {
+ fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n");
+ res->check_errors++;
goto fail;
+ }
/* Do the actual checks */
for(i = 0; i < s->l2_size; i++) {
@@ -1221,9 +1225,8 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
return 0;
fail:
- fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n");
g_free(l2_table);
- return -EIO;
+ return ret;
}
/*
@@ -1258,11 +1261,15 @@ static int check_refcounts_l1(BlockDriverState *bs,
l1_table = g_try_malloc(l1_size2);
if (l1_table == NULL) {
ret = -ENOMEM;
+ res->check_errors++;
goto fail;
}
- if (bdrv_pread(bs->file, l1_table_offset,
- l1_table, l1_size2) != l1_size2)
+ ret = bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2);
+ if (ret < 0) {
+ fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
+ res->check_errors++;
goto fail;
+ }
for(i = 0;i < l1_size; i++)
be64_to_cpus(&l1_table[i]);
}
@@ -1295,10 +1302,8 @@ static int check_refcounts_l1(BlockDriverState *bs,
return 0;
fail:
- fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
- res->check_errors++;
g_free(l1_table);
- return -EIO;
+ return ret;
}
/*
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 09/17] qcow2: Let inc_refcounts() return -errno
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (7 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 08/17] qcow2: Split fail code in L1 and L2 checks Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 10/17] qcow2: Let inc_refcounts() resize the reftable Max Reitz
` (8 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
As of a future patch, inc_refcounts() will have to throw errors which
are generally signaled by returning -errno. Therefore, let it return an
integer which is either 0 for success or -errno and handle the -errno
case in all callers.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 91 +++++++++++++++++++++++++++++++++-----------------
1 file changed, 60 insertions(+), 31 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 59fb400..6001c85 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1081,17 +1081,18 @@ fail:
*
* Modifies the number of errors in res.
*/
-static void inc_refcounts(BlockDriverState *bs,
- BdrvCheckResult *res,
- uint16_t *refcount_table,
- int64_t refcount_table_size,
- int64_t offset, int64_t size)
+static int inc_refcounts(BlockDriverState *bs,
+ BdrvCheckResult *res,
+ uint16_t *refcount_table,
+ int64_t refcount_table_size,
+ int64_t offset, int64_t size)
{
BDRVQcowState *s = bs->opaque;
uint64_t start, last, cluster_offset, k;
- if (size <= 0)
- return;
+ if (size <= 0) {
+ return 0;
+ }
start = start_of_cluster(s, offset);
last = start_of_cluster(s, offset + size - 1);
@@ -1111,6 +1112,8 @@ static void inc_refcounts(BlockDriverState *bs,
}
}
}
+
+ return 0;
}
/* Flags for check_refcounts_l1() and check_refcounts_l2() */
@@ -1165,8 +1168,11 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
nb_csectors = ((l2_entry >> s->csize_shift) &
s->csize_mask) + 1;
l2_entry &= s->cluster_offset_mask;
- inc_refcounts(bs, res, refcount_table, refcount_table_size,
- l2_entry & ~511, nb_csectors * 512);
+ ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
+ l2_entry & ~511, nb_csectors * 512);
+ if (ret < 0) {
+ goto fail;
+ }
if (flags & CHECK_FRAG_INFO) {
res->bfi.allocated_clusters++;
@@ -1201,8 +1207,11 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
}
/* Mark cluster as used */
- inc_refcounts(bs, res, refcount_table,refcount_table_size,
- offset, s->cluster_size);
+ ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
+ offset, s->cluster_size);
+ if (ret < 0) {
+ goto fail;
+ }
/* Correct offsets are cluster aligned */
if (offset_into_cluster(s, offset)) {
@@ -1245,19 +1254,20 @@ static int check_refcounts_l1(BlockDriverState *bs,
int flags)
{
BDRVQcowState *s = bs->opaque;
- uint64_t *l1_table, l2_offset, l1_size2;
+ uint64_t *l1_table = NULL, l2_offset, l1_size2;
int i, ret;
l1_size2 = l1_size * sizeof(uint64_t);
/* Mark L1 table as used */
- inc_refcounts(bs, res, refcount_table, refcount_table_size,
- l1_table_offset, l1_size2);
+ ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
+ l1_table_offset, l1_size2);
+ if (ret < 0) {
+ goto fail;
+ }
/* Read L1 table entries from disk */
- if (l1_size2 == 0) {
- l1_table = NULL;
- } else {
+ if (l1_size2 > 0) {
l1_table = g_try_malloc(l1_size2);
if (l1_table == NULL) {
ret = -ENOMEM;
@@ -1280,8 +1290,11 @@ static int check_refcounts_l1(BlockDriverState *bs,
if (l2_offset) {
/* Mark L2 table as used */
l2_offset &= L1E_OFFSET_MASK;
- inc_refcounts(bs, res, refcount_table, refcount_table_size,
- l2_offset, s->cluster_size);
+ ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
+ l2_offset, s->cluster_size);
+ if (ret < 0) {
+ goto fail;
+ }
/* L2 tables are cluster aligned */
if (offset_into_cluster(s, l2_offset)) {
@@ -1550,6 +1563,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
{
BDRVQcowState *s = bs->opaque;
int64_t i;
+ int ret;
for(i = 0; i < s->refcount_table_size; i++) {
uint64_t offset, cluster;
@@ -1572,8 +1586,11 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
}
if (offset != 0) {
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- offset, s->cluster_size);
+ ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ offset, s->cluster_size);
+ if (ret < 0) {
+ return ret;
+ }
if ((*refcount_table)[cluster] != 1) {
fprintf(stderr, "%s refcount block %" PRId64
" refcount=%d\n",
@@ -1602,8 +1619,11 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
sizeof(**refcount_table));
}
(*refcount_table)[cluster]--;
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- new_offset, s->cluster_size);
+ ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ new_offset, s->cluster_size);
+ if (ret < 0) {
+ return ret;
+ }
res->corruptions_fixed++;
} else {
@@ -1635,8 +1655,11 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
/* header */
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- 0, s->cluster_size);
+ ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ 0, s->cluster_size);
+ if (ret < 0) {
+ return ret;
+ }
/* current L1 table */
ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
@@ -1649,18 +1672,24 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
for (i = 0; i < s->nb_snapshots; i++) {
sn = s->snapshots + i;
ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
- sn->l1_table_offset, sn->l1_size, 0);
+ sn->l1_table_offset, sn->l1_size, 0);
if (ret < 0) {
return ret;
}
}
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- s->snapshots_offset, s->snapshots_size);
+ ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ s->snapshots_offset, s->snapshots_size);
+ if (ret < 0) {
+ return ret;
+ }
/* refcount data */
- inc_refcounts(bs, res, *refcount_table, *nb_clusters,
- s->refcount_table_offset,
- s->refcount_table_size * sizeof(uint64_t));
+ ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ s->refcount_table_offset,
+ s->refcount_table_size * sizeof(uint64_t));
+ if (ret < 0) {
+ return ret;
+ }
return check_refblocks(bs, res, fix, refcount_table, nb_clusters);
}
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 10/17] qcow2: Let inc_refcounts() resize the reftable
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (8 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 09/17] qcow2: Let inc_refcounts() return -errno Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 11/17] qcow2: Reuse refcount table in calculate_refcounts() Max Reitz
` (7 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
Now that the refcount table can be passed around by reference, do that
for inc_refcounts() (and subsequently check_refcounts_l1() and
check_refcounts_l2()) and use it for resizing it when a cluster after
the image end is encountered.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 57 +++++++++++++++++++++++++++++++-------------------
1 file changed, 35 insertions(+), 22 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 6001c85..d653124 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1083,8 +1083,8 @@ fail:
*/
static int inc_refcounts(BlockDriverState *bs,
BdrvCheckResult *res,
- uint16_t *refcount_table,
- int64_t refcount_table_size,
+ uint16_t **refcount_table,
+ int64_t *refcount_table_size,
int64_t offset, int64_t size)
{
BDRVQcowState *s = bs->opaque;
@@ -1099,17 +1099,30 @@ static int inc_refcounts(BlockDriverState *bs,
for(cluster_offset = start; cluster_offset <= last;
cluster_offset += s->cluster_size) {
k = cluster_offset >> s->cluster_bits;
- if (k >= refcount_table_size) {
- fprintf(stderr, "Warning: cluster offset=0x%" PRIx64 " is after "
- "the end of the image file, can't properly check refcounts.\n",
- cluster_offset);
- res->check_errors++;
- } else {
- if (++refcount_table[k] == 0) {
- fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64
- "\n", cluster_offset);
- res->corruptions++;
+ if (k >= *refcount_table_size) {
+ int64_t old_refcount_table_size = *refcount_table_size;
+ uint16_t *new_refcount_table;
+
+ *refcount_table_size = k + 1;
+ new_refcount_table = g_try_realloc(*refcount_table,
+ *refcount_table_size *
+ sizeof(**refcount_table));
+ if (!new_refcount_table) {
+ *refcount_table_size = old_refcount_table_size;
+ res->check_errors++;
+ return -ENOMEM;
}
+ *refcount_table = new_refcount_table;
+
+ memset(*refcount_table + old_refcount_table_size, 0,
+ (*refcount_table_size - old_refcount_table_size) *
+ sizeof(**refcount_table));
+ }
+
+ if (++(*refcount_table)[k] == 0) {
+ fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64
+ "\n", cluster_offset);
+ res->corruptions++;
}
}
@@ -1130,7 +1143,7 @@ enum {
* error occurred.
*/
static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
- uint16_t *refcount_table, int64_t refcount_table_size, int64_t l2_offset,
+ uint16_t **refcount_table, int64_t *refcount_table_size, int64_t l2_offset,
int flags)
{
BDRVQcowState *s = bs->opaque;
@@ -1248,8 +1261,8 @@ fail:
*/
static int check_refcounts_l1(BlockDriverState *bs,
BdrvCheckResult *res,
- uint16_t *refcount_table,
- int64_t refcount_table_size,
+ uint16_t **refcount_table,
+ int64_t *refcount_table_size,
int64_t l1_table_offset, int l1_size,
int flags)
{
@@ -1586,7 +1599,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
}
if (offset != 0) {
- ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
offset, s->cluster_size);
if (ret < 0) {
return ret;
@@ -1619,7 +1632,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
sizeof(**refcount_table));
}
(*refcount_table)[cluster]--;
- ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
new_offset, s->cluster_size);
if (ret < 0) {
return ret;
@@ -1655,14 +1668,14 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
/* header */
- ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
0, s->cluster_size);
if (ret < 0) {
return ret;
}
/* current L1 table */
- ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
+ ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
s->l1_table_offset, s->l1_size, CHECK_FRAG_INFO);
if (ret < 0) {
return ret;
@@ -1671,20 +1684,20 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
/* snapshots */
for (i = 0; i < s->nb_snapshots; i++) {
sn = s->snapshots + i;
- ret = check_refcounts_l1(bs, res, *refcount_table, *nb_clusters,
+ ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
sn->l1_table_offset, sn->l1_size, 0);
if (ret < 0) {
return ret;
}
}
- ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
s->snapshots_offset, s->snapshots_size);
if (ret < 0) {
return ret;
}
/* refcount data */
- ret = inc_refcounts(bs, res, *refcount_table, *nb_clusters,
+ ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
s->refcount_table_offset,
s->refcount_table_size * sizeof(uint64_t));
if (ret < 0) {
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 11/17] qcow2: Reuse refcount table in calculate_refcounts()
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (9 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 10/17] qcow2: Let inc_refcounts() resize the reftable Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 12/17] qcow2: Fix refcount blocks beyond image end Max Reitz
` (6 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
We will later call calculate_refcounts multiple times, so reuse the
refcount table if possible.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
---
block/qcow2-refcount.c | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index d653124..c92e1fc 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1661,10 +1661,12 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
QCowSnapshot *sn;
int ret;
- *refcount_table = g_try_new0(uint16_t, *nb_clusters);
- if (*nb_clusters && *refcount_table == NULL) {
- res->check_errors++;
- return -ENOMEM;
+ if (!*refcount_table) {
+ *refcount_table = g_try_new0(uint16_t, *nb_clusters);
+ if (*nb_clusters && *refcount_table == NULL) {
+ res->check_errors++;
+ return -ENOMEM;
+ }
}
/* header */
@@ -1780,7 +1782,7 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
{
BDRVQcowState *s = bs->opaque;
int64_t size, highest_cluster, nb_clusters;
- uint16_t *refcount_table;
+ uint16_t *refcount_table = NULL;
int ret;
size = bdrv_getlength(bs->file);
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 12/17] qcow2: Fix refcount blocks beyond image end
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (10 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 11/17] qcow2: Reuse refcount table in calculate_refcounts() Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 13/17] qcow2: Do not perform potentially damaging repairs Max Reitz
` (5 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
If the qcow2 check function detects a refcount block located beyond the
image end, grow the image appropriately. This cannot break anything and
is the logical fix for such a case.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 67 +++++++++++++++++++++++++++++++++++++++++++++++---
1 file changed, 63 insertions(+), 4 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index c92e1fc..b87eafc 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1575,7 +1575,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
int64_t *nb_clusters)
{
BDRVQcowState *s = bs->opaque;
- int64_t i;
+ int64_t i, size;
int ret;
for(i = 0; i < s->refcount_table_size; i++) {
@@ -1592,9 +1592,68 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
}
if (cluster >= *nb_clusters) {
- fprintf(stderr, "ERROR refcount block %" PRId64
- " is outside image\n", i);
- res->corruptions++;
+ fprintf(stderr, "%s refcount block %" PRId64 " is outside image\n",
+ fix & BDRV_FIX_ERRORS ? "Repairing" : "ERROR", i);
+
+ if (fix & BDRV_FIX_ERRORS) {
+ int64_t old_nb_clusters = *nb_clusters;
+ uint16_t *new_refcount_table;
+
+ if (offset > INT64_MAX - s->cluster_size) {
+ ret = -EINVAL;
+ goto resize_fail;
+ }
+
+ ret = bdrv_truncate(bs->file, offset + s->cluster_size);
+ if (ret < 0) {
+ goto resize_fail;
+ }
+ size = bdrv_getlength(bs->file);
+ if (size < 0) {
+ ret = size;
+ goto resize_fail;
+ }
+
+ *nb_clusters = size_to_clusters(s, size);
+ assert(*nb_clusters >= old_nb_clusters);
+
+ new_refcount_table = g_try_realloc(*refcount_table,
+ *nb_clusters *
+ sizeof(**refcount_table));
+ if (!new_refcount_table) {
+ *nb_clusters = old_nb_clusters;
+ res->check_errors++;
+ return -ENOMEM;
+ }
+ *refcount_table = new_refcount_table;
+
+ memset(*refcount_table + old_nb_clusters, 0,
+ (*nb_clusters - old_nb_clusters) *
+ sizeof(**refcount_table));
+
+ if (cluster >= *nb_clusters) {
+ ret = -EINVAL;
+ goto resize_fail;
+ }
+
+ res->corruptions_fixed++;
+ ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
+ offset, s->cluster_size);
+ if (ret < 0) {
+ return ret;
+ }
+ /* No need to check whether the refcount is now greater than 1:
+ * This area was just allocated and zeroed, so it can only be
+ * exactly 1 after inc_refcounts() */
+ continue;
+
+resize_fail:
+ res->corruptions++;
+ fprintf(stderr, "ERROR could not resize image: %s\n",
+ strerror(-ret));
+ } else {
+ res->corruptions++;
+ }
continue;
}
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 13/17] qcow2: Do not perform potentially damaging repairs
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (11 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 12/17] qcow2: Fix refcount blocks beyond image end Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 14/17] qcow2: Rebuild refcount structure during check Max Reitz
` (4 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
If a referenced cluster has a refcount of 0, increasing its refcount may
result in clusters being allocated for the refcount structures. This may
overwrite the referenced cluster, therefore we cannot simply increase
the refcount then.
In such cases, we can either try to replicate all the refcount
operations solely for the check operation, basing the allocations on the
in-memory refcount table; or we can simply rebuild the whole refcount
structure based on the in-memory refcount table. Since the latter will
be much easier, do that.
To prepare for this, introduce a "rebuild" boolean which should be set
to true whenever a fix is rather dangerous or too complicated using the
current refcount structures. Another example for this is refcount blocks
being referenced more than once.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 188 +++++++------------------------------------------
1 file changed, 27 insertions(+), 161 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index b87eafc..e964666 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1452,127 +1452,12 @@ fail:
}
/*
- * Writes one sector of the refcount table to the disk
- */
-#define RT_ENTRIES_PER_SECTOR (512 / sizeof(uint64_t))
-static int write_reftable_entry(BlockDriverState *bs, int rt_index)
-{
- BDRVQcowState *s = bs->opaque;
- uint64_t buf[RT_ENTRIES_PER_SECTOR];
- int rt_start_index;
- int i, ret;
-
- rt_start_index = rt_index & ~(RT_ENTRIES_PER_SECTOR - 1);
- for (i = 0; i < RT_ENTRIES_PER_SECTOR; i++) {
- buf[i] = cpu_to_be64(s->refcount_table[rt_start_index + i]);
- }
-
- ret = qcow2_pre_write_overlap_check(bs, QCOW2_OL_REFCOUNT_TABLE,
- s->refcount_table_offset + rt_start_index * sizeof(uint64_t),
- sizeof(buf));
- if (ret < 0) {
- return ret;
- }
-
- BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_UPDATE);
- ret = bdrv_pwrite_sync(bs->file, s->refcount_table_offset +
- rt_start_index * sizeof(uint64_t), buf, sizeof(buf));
- if (ret < 0) {
- return ret;
- }
-
- return 0;
-}
-
-/*
- * Allocates a new cluster for the given refcount block (represented by its
- * offset in the image file) and copies the current content there. This function
- * does _not_ decrement the reference count for the currently occupied cluster.
- *
- * This function prints an informative message to stderr on error (and returns
- * -errno); on success, the offset of the newly allocated cluster is returned.
- */
-static int64_t realloc_refcount_block(BlockDriverState *bs, int reftable_index,
- uint64_t offset)
-{
- BDRVQcowState *s = bs->opaque;
- int64_t new_offset = 0;
- void *refcount_block = NULL;
- int ret;
-
- /* allocate new refcount block */
- new_offset = qcow2_alloc_clusters(bs, s->cluster_size);
- if (new_offset < 0) {
- fprintf(stderr, "Could not allocate new cluster: %s\n",
- strerror(-new_offset));
- ret = new_offset;
- goto done;
- }
-
- /* fetch current refcount block content */
- ret = qcow2_cache_get(bs, s->refcount_block_cache, offset, &refcount_block);
- if (ret < 0) {
- fprintf(stderr, "Could not fetch refcount block: %s\n", strerror(-ret));
- goto fail_free_cluster;
- }
-
- /* new block has not yet been entered into refcount table, therefore it is
- * no refcount block yet (regarding this check) */
- ret = qcow2_pre_write_overlap_check(bs, 0, new_offset, s->cluster_size);
- if (ret < 0) {
- fprintf(stderr, "Could not write refcount block; metadata overlap "
- "check failed: %s\n", strerror(-ret));
- /* the image will be marked corrupt, so don't even attempt on freeing
- * the cluster */
- goto done;
- }
-
- /* write to new block */
- ret = bdrv_write(bs->file, new_offset / BDRV_SECTOR_SIZE, refcount_block,
- s->cluster_sectors);
- if (ret < 0) {
- fprintf(stderr, "Could not write refcount block: %s\n", strerror(-ret));
- goto fail_free_cluster;
- }
-
- /* update refcount table */
- assert(!offset_into_cluster(s, new_offset));
- s->refcount_table[reftable_index] = new_offset;
- ret = write_reftable_entry(bs, reftable_index);
- if (ret < 0) {
- fprintf(stderr, "Could not update refcount table: %s\n",
- strerror(-ret));
- goto fail_free_cluster;
- }
-
- goto done;
-
-fail_free_cluster:
- qcow2_free_clusters(bs, new_offset, s->cluster_size, QCOW2_DISCARD_OTHER);
-
-done:
- if (refcount_block) {
- /* This should never fail, as it would only do so if the given refcount
- * block cannot be found in the cache. As this is impossible as long as
- * there are no bugs, assert the success. */
- int tmp = qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block);
- assert(tmp == 0);
- }
-
- if (ret < 0) {
- return ret;
- }
-
- return new_offset;
-}
-
-/*
* Checks consistency of refblocks and accounts for each refblock in
* *refcount_table.
*/
static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
- BdrvCheckMode fix, uint16_t **refcount_table,
- int64_t *nb_clusters)
+ BdrvCheckMode fix, bool *rebuild,
+ uint16_t **refcount_table, int64_t *nb_clusters)
{
BDRVQcowState *s = bs->opaque;
int64_t i, size;
@@ -1588,6 +1473,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
fprintf(stderr, "ERROR refcount block %" PRId64 " is not "
"cluster aligned; refcount table entry corrupted\n", i);
res->corruptions++;
+ *rebuild = true;
continue;
}
@@ -1649,6 +1535,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
resize_fail:
res->corruptions++;
+ *rebuild = true;
fprintf(stderr, "ERROR could not resize image: %s\n",
strerror(-ret));
} else {
@@ -1664,43 +1551,10 @@ resize_fail:
return ret;
}
if ((*refcount_table)[cluster] != 1) {
- fprintf(stderr, "%s refcount block %" PRId64
- " refcount=%d\n",
- fix & BDRV_FIX_ERRORS ? "Repairing" :
- "ERROR",
- i, (*refcount_table)[cluster]);
-
- if (fix & BDRV_FIX_ERRORS) {
- int64_t new_offset;
-
- new_offset = realloc_refcount_block(bs, i, offset);
- if (new_offset < 0) {
- res->corruptions++;
- continue;
- }
-
- /* update refcounts */
- if ((new_offset >> s->cluster_bits) >= *nb_clusters) {
- /* increase refcount_table size if necessary */
- int old_nb_clusters = *nb_clusters;
- *nb_clusters = (new_offset >> s->cluster_bits) + 1;
- *refcount_table = g_renew(uint16_t, *refcount_table,
- *nb_clusters);
- memset(&(*refcount_table)[old_nb_clusters], 0,
- (*nb_clusters - old_nb_clusters) *
- sizeof(**refcount_table));
- }
- (*refcount_table)[cluster]--;
- ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
- new_offset, s->cluster_size);
- if (ret < 0) {
- return ret;
- }
-
- res->corruptions_fixed++;
- } else {
- res->corruptions++;
- }
+ fprintf(stderr, "ERROR refcount block %" PRId64
+ " refcount=%d\n", i, (*refcount_table)[cluster]);
+ res->corruptions++;
+ *rebuild = true;
}
}
}
@@ -1712,8 +1566,8 @@ resize_fail:
* Calculates an in-memory refcount table.
*/
static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
- BdrvCheckMode fix, uint16_t **refcount_table,
- int64_t *nb_clusters)
+ BdrvCheckMode fix, bool *rebuild,
+ uint16_t **refcount_table, int64_t *nb_clusters)
{
BDRVQcowState *s = bs->opaque;
int64_t i;
@@ -1765,7 +1619,7 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
return ret;
}
- return check_refblocks(bs, res, fix, refcount_table, nb_clusters);
+ return check_refblocks(bs, res, fix, rebuild, refcount_table, nb_clusters);
}
/*
@@ -1773,7 +1627,8 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
* refcount as reported by the refcount structures on-disk.
*/
static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
- BdrvCheckMode fix, int64_t *highest_cluster,
+ BdrvCheckMode fix, bool *rebuild,
+ int64_t *highest_cluster,
uint16_t *refcount_table, int64_t nb_clusters)
{
BDRVQcowState *s = bs->opaque;
@@ -1798,7 +1653,9 @@ static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
if (refcount1 != refcount2) {
/* Check if we're allowed to fix the mismatch */
int *num_fixed = NULL;
- if (refcount1 > refcount2 && (fix & BDRV_FIX_LEAKS)) {
+ if (refcount1 == 0) {
+ *rebuild = true;
+ } else if (refcount1 > refcount2 && (fix & BDRV_FIX_LEAKS)) {
num_fixed = &res->leaks_fixed;
} else if (refcount1 < refcount2 && (fix & BDRV_FIX_ERRORS)) {
num_fixed = &res->corruptions_fixed;
@@ -1842,6 +1699,7 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
BDRVQcowState *s = bs->opaque;
int64_t size, highest_cluster, nb_clusters;
uint16_t *refcount_table = NULL;
+ bool rebuild = false;
int ret;
size = bdrv_getlength(bs->file);
@@ -1859,14 +1717,22 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
res->bfi.total_clusters =
size_to_clusters(s, bs->total_sectors * BDRV_SECTOR_SIZE);
- ret = calculate_refcounts(bs, res, fix, &refcount_table, &nb_clusters);
+ ret = calculate_refcounts(bs, res, fix, &rebuild, &refcount_table,
+ &nb_clusters);
if (ret < 0) {
goto fail;
}
- compare_refcounts(bs, res, fix, &highest_cluster, refcount_table,
+ compare_refcounts(bs, res, fix, &rebuild, &highest_cluster, refcount_table,
nb_clusters);
+ if (rebuild) {
+ fprintf(stderr, "ERROR need to rebuild refcount structures\n");
+ res->check_errors++;
+ /* Just carry on, the rest does not rely on the on-disk refcount
+ * structures */
+ }
+
/* check OFLAG_COPIED */
ret = check_oflag_copied(bs, res, fix);
if (ret < 0) {
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 14/17] qcow2: Rebuild refcount structure during check
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (12 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 13/17] qcow2: Do not perform potentially damaging repairs Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 15/17] qcow2: Clean up after refcount rebuild Max Reitz
` (3 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
The previous commit introduced the "rebuild" variable to qcow2's
implementation of the image consistency check. Now make use of this by
adding a function which creates a completely new refcount structure
based solely on the in-memory information gathered before.
The old refcount structure will be leaked, however. This leak will be
dealt with in a follow-up commit.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 311 ++++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 305 insertions(+), 6 deletions(-)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index e964666..9bfc75e 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -1688,6 +1688,285 @@ static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
}
/*
+ * Allocates clusters using an in-memory refcount table (IMRT) in contrast to
+ * the on-disk refcount structures.
+ *
+ * On input, *first_free_cluster tells where to start looking, and need not
+ * actually be a free cluster; the returned offset will not be before that
+ * cluster. On output, *first_free_cluster points to the first gap found, even
+ * if that gap was too small to be used as the returned offset.
+ *
+ * Note that *first_free_cluster is a cluster index whereas the return value is
+ * an offset.
+ */
+static int64_t alloc_clusters_imrt(BlockDriverState *bs,
+ int cluster_count,
+ uint16_t **refcount_table,
+ int64_t *imrt_nb_clusters,
+ int64_t *first_free_cluster)
+{
+ BDRVQcowState *s = bs->opaque;
+ int64_t cluster = *first_free_cluster, i;
+ bool first_gap = true;
+ int contiguous_free_clusters;
+
+ /* Starting at *first_free_cluster, find a range of at least cluster_count
+ * continuously free clusters */
+ for (contiguous_free_clusters = 0;
+ cluster < *imrt_nb_clusters &&
+ contiguous_free_clusters < cluster_count;
+ cluster++)
+ {
+ if (!(*refcount_table)[cluster]) {
+ contiguous_free_clusters++;
+ if (first_gap) {
+ /* If this is the first free cluster found, update
+ * *first_free_cluster accordingly */
+ *first_free_cluster = cluster;
+ first_gap = false;
+ }
+ } else if (contiguous_free_clusters) {
+ contiguous_free_clusters = 0;
+ }
+ }
+
+ /* If contiguous_free_clusters is greater than zero, it contains the number
+ * of continuously free clusters until the current cluster; the first free
+ * cluster in the current "gap" is therefore
+ * cluster - contiguous_free_clusters */
+
+ /* If no such range could be found, grow the in-memory refcount table
+ * accordingly to append free clusters at the end of the image */
+ if (contiguous_free_clusters < cluster_count) {
+ int64_t old_imrt_nb_clusters = *imrt_nb_clusters;
+ uint16_t *new_refcount_table;
+
+ /* contiguous_free_clusters clusters are already empty at the image end;
+ * we need cluster_count clusters; therefore, we have to allocate
+ * cluster_count - contiguous_free_clusters new clusters at the end of
+ * the image (which is the current value of cluster; note that cluster
+ * may exceed old_imrt_nb_clusters if *first_free_cluster pointed beyond
+ * the image end) */
+ *imrt_nb_clusters = cluster + cluster_count - contiguous_free_clusters;
+ new_refcount_table = g_try_realloc(*refcount_table,
+ *imrt_nb_clusters *
+ sizeof(**refcount_table));
+ if (!new_refcount_table) {
+ *imrt_nb_clusters = old_imrt_nb_clusters;
+ return -ENOMEM;
+ }
+ *refcount_table = new_refcount_table;
+
+ memset(*refcount_table + old_imrt_nb_clusters, 0,
+ (*imrt_nb_clusters - old_imrt_nb_clusters) *
+ sizeof(**refcount_table));
+ }
+
+ /* Go back to the first free cluster */
+ cluster -= contiguous_free_clusters;
+ for (i = 0; i < cluster_count; i++) {
+ (*refcount_table)[cluster + i] = 1;
+ }
+
+ return cluster << s->cluster_bits;
+}
+
+/*
+ * Creates a new refcount structure based solely on the in-memory information
+ * given through *refcount_table. All necessary allocations will be reflected
+ * in that array.
+ *
+ * On success, the old refcount structure is leaked (it will be covered by the
+ * new refcount structure).
+ */
+static int rebuild_refcount_structure(BlockDriverState *bs,
+ BdrvCheckResult *res,
+ uint16_t **refcount_table,
+ int64_t *nb_clusters)
+{
+ BDRVQcowState *s = bs->opaque;
+ int64_t first_free_cluster = 0, reftable_offset = -1, cluster = 0;
+ int64_t refblock_offset, refblock_start, refblock_index;
+ uint32_t reftable_size = 0;
+ uint64_t *on_disk_reftable = NULL;
+ uint16_t *on_disk_refblock;
+ int i, ret = 0;
+ struct {
+ uint64_t reftable_offset;
+ uint32_t reftable_clusters;
+ } QEMU_PACKED reftable_offset_and_clusters;
+
+ qcow2_cache_empty(bs, s->refcount_block_cache);
+
+write_refblocks:
+ for (; cluster < *nb_clusters; cluster++) {
+ if (!(*refcount_table)[cluster]) {
+ continue;
+ }
+
+ refblock_index = cluster >> s->refcount_block_bits;
+ refblock_start = refblock_index << s->refcount_block_bits;
+
+ /* Don't allocate a cluster in a refblock already written to disk */
+ if (first_free_cluster < refblock_start) {
+ first_free_cluster = refblock_start;
+ }
+ refblock_offset = alloc_clusters_imrt(bs, 1, refcount_table,
+ nb_clusters, &first_free_cluster);
+ if (refblock_offset < 0) {
+ fprintf(stderr, "ERROR allocating refblock: %s\n",
+ strerror(-refblock_offset));
+ res->check_errors++;
+ ret = refblock_offset;
+ goto fail;
+ }
+
+ if (reftable_size <= refblock_index) {
+ uint32_t old_reftable_size = reftable_size;
+ uint64_t *new_on_disk_reftable;
+
+ reftable_size = ROUND_UP((refblock_index + 1) * sizeof(uint64_t),
+ s->cluster_size) / sizeof(uint64_t);
+ new_on_disk_reftable = g_try_realloc(on_disk_reftable,
+ reftable_size *
+ sizeof(uint64_t));
+ if (!new_on_disk_reftable) {
+ res->check_errors++;
+ ret = -ENOMEM;
+ goto fail;
+ }
+ on_disk_reftable = new_on_disk_reftable;
+
+ memset(on_disk_reftable + old_reftable_size, 0,
+ (reftable_size - old_reftable_size) * sizeof(uint64_t));
+
+ /* The offset we have for the reftable is now no longer valid;
+ * this will leak that range, but we can easily fix that by running
+ * a leak-fixing check after this rebuild operation */
+ reftable_offset = -1;
+ }
+ on_disk_reftable[refblock_index] = refblock_offset;
+
+ /* If this is apparently the last refblock (for now), try to squeeze the
+ * reftable in */
+ if (refblock_index == (*nb_clusters - 1) >> s->refcount_block_bits &&
+ reftable_offset < 0)
+ {
+ uint64_t reftable_clusters = size_to_clusters(s, reftable_size *
+ sizeof(uint64_t));
+ reftable_offset = alloc_clusters_imrt(bs, reftable_clusters,
+ refcount_table, nb_clusters,
+ &first_free_cluster);
+ if (reftable_offset < 0) {
+ fprintf(stderr, "ERROR allocating reftable: %s\n",
+ strerror(-reftable_offset));
+ res->check_errors++;
+ ret = reftable_offset;
+ goto fail;
+ }
+ }
+
+ ret = qcow2_pre_write_overlap_check(bs, 0, refblock_offset,
+ s->cluster_size);
+ if (ret < 0) {
+ fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret));
+ goto fail;
+ }
+
+ on_disk_refblock = qemu_blockalign0(bs->file, s->cluster_size);
+ for (i = 0; i < s->refcount_block_size &&
+ refblock_start + i < *nb_clusters; i++)
+ {
+ on_disk_refblock[i] =
+ cpu_to_be16((*refcount_table)[refblock_start + i]);
+ }
+
+ ret = bdrv_write(bs->file, refblock_offset / BDRV_SECTOR_SIZE,
+ (void *)on_disk_refblock, s->cluster_sectors);
+ qemu_vfree(on_disk_refblock);
+ if (ret < 0) {
+ fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret));
+ goto fail;
+ }
+
+ /* Go to the end of this refblock */
+ cluster = refblock_start + s->refcount_block_size - 1;
+ }
+
+ if (reftable_offset < 0) {
+ uint64_t post_refblock_start, reftable_clusters;
+
+ post_refblock_start = ROUND_UP(*nb_clusters, s->refcount_block_size);
+ reftable_clusters = size_to_clusters(s,
+ reftable_size * sizeof(uint64_t));
+ /* Not pretty but simple */
+ if (first_free_cluster < post_refblock_start) {
+ first_free_cluster = post_refblock_start;
+ }
+ reftable_offset = alloc_clusters_imrt(bs, reftable_clusters,
+ refcount_table, nb_clusters,
+ &first_free_cluster);
+ if (reftable_offset < 0) {
+ fprintf(stderr, "ERROR allocating reftable: %s\n",
+ strerror(-reftable_offset));
+ res->check_errors++;
+ ret = reftable_offset;
+ goto fail;
+ }
+
+ goto write_refblocks;
+ }
+
+ assert(on_disk_reftable);
+
+ for (refblock_index = 0; refblock_index < reftable_size; refblock_index++) {
+ cpu_to_be64s(&on_disk_reftable[refblock_index]);
+ }
+
+ ret = qcow2_pre_write_overlap_check(bs, 0, reftable_offset,
+ reftable_size * sizeof(uint64_t));
+ if (ret < 0) {
+ fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret));
+ goto fail;
+ }
+
+ assert(reftable_size < INT_MAX / sizeof(uint64_t));
+ ret = bdrv_pwrite(bs->file, reftable_offset, on_disk_reftable,
+ reftable_size * sizeof(uint64_t));
+ if (ret < 0) {
+ fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret));
+ goto fail;
+ }
+
+ /* Enter new reftable into the image header */
+ cpu_to_be64w(&reftable_offset_and_clusters.reftable_offset,
+ reftable_offset);
+ cpu_to_be32w(&reftable_offset_and_clusters.reftable_clusters,
+ size_to_clusters(s, reftable_size * sizeof(uint64_t)));
+ ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader,
+ refcount_table_offset),
+ &reftable_offset_and_clusters,
+ sizeof(reftable_offset_and_clusters));
+ if (ret < 0) {
+ fprintf(stderr, "ERROR setting reftable: %s\n", strerror(-ret));
+ goto fail;
+ }
+
+ for (refblock_index = 0; refblock_index < reftable_size; refblock_index++) {
+ be64_to_cpus(&on_disk_reftable[refblock_index]);
+ }
+ s->refcount_table = on_disk_reftable;
+ s->refcount_table_offset = reftable_offset;
+ s->refcount_table_size = reftable_size;
+
+ return 0;
+
+fail:
+ g_free(on_disk_reftable);
+ return ret;
+}
+
+/*
* Checks an image for refcount consistency.
*
* Returns 0 if no errors are found, the number of errors in case the image is
@@ -1697,6 +1976,7 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
BdrvCheckMode fix)
{
BDRVQcowState *s = bs->opaque;
+ BdrvCheckResult pre_compare_res;
int64_t size, highest_cluster, nb_clusters;
uint16_t *refcount_table = NULL;
bool rebuild = false;
@@ -1723,14 +2003,33 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
goto fail;
}
- compare_refcounts(bs, res, fix, &rebuild, &highest_cluster, refcount_table,
+ /* In case we don't need to rebuild the refcount structure (but want to fix
+ * something), this function is immediately called again, in which case the
+ * result should be ignored */
+ pre_compare_res = *res;
+ compare_refcounts(bs, res, 0, &rebuild, &highest_cluster, refcount_table,
nb_clusters);
- if (rebuild) {
- fprintf(stderr, "ERROR need to rebuild refcount structures\n");
- res->check_errors++;
- /* Just carry on, the rest does not rely on the on-disk refcount
- * structures */
+ if (rebuild && (fix & BDRV_FIX_ERRORS)) {
+ fprintf(stderr, "Rebuilding refcount structure\n");
+ ret = rebuild_refcount_structure(bs, res, &refcount_table,
+ &nb_clusters);
+ if (ret < 0) {
+ goto fail;
+ }
+ } else if (fix) {
+ if (rebuild) {
+ fprintf(stderr, "ERROR need to rebuild refcount structures\n");
+ res->check_errors++;
+ ret = -EIO;
+ goto fail;
+ }
+
+ if (res->leaks || res->corruptions) {
+ *res = pre_compare_res;
+ compare_refcounts(bs, res, fix, &rebuild, &highest_cluster,
+ refcount_table, nb_clusters);
+ }
}
/* check OFLAG_COPIED */
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 15/17] qcow2: Clean up after refcount rebuild
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (13 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 14/17] qcow2: Rebuild refcount structure during check Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 16/17] iotests: Fix test outputs Max Reitz
` (2 subsequent siblings)
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
Because the old refcount structure will be leaked after having rebuilt
it, we need to recalculate the refcounts and run a leak-fixing operation
afterwards (if leaks should be fixed at all).
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2-refcount.c | 45 +++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 45 insertions(+)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 9bfc75e..6f1b118 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -2011,12 +2011,57 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
nb_clusters);
if (rebuild && (fix & BDRV_FIX_ERRORS)) {
+ BdrvCheckResult old_res = *res;
+ int fresh_leaks = 0;
+
fprintf(stderr, "Rebuilding refcount structure\n");
ret = rebuild_refcount_structure(bs, res, &refcount_table,
&nb_clusters);
if (ret < 0) {
goto fail;
}
+
+ res->corruptions = 0;
+ res->leaks = 0;
+
+ /* Because the old reftable has been exchanged for a new one the
+ * references have to be recalculated */
+ rebuild = false;
+ memset(refcount_table, 0, nb_clusters * sizeof(uint16_t));
+ ret = calculate_refcounts(bs, res, 0, &rebuild, &refcount_table,
+ &nb_clusters);
+ if (ret < 0) {
+ goto fail;
+ }
+
+ if (fix & BDRV_FIX_LEAKS) {
+ /* The old refcount structures are now leaked, fix it; the result
+ * can be ignored, aside from leaks which were introduced by
+ * rebuild_refcount_structure() that could not be fixed */
+ BdrvCheckResult saved_res = *res;
+ *res = (BdrvCheckResult){ 0 };
+
+ compare_refcounts(bs, res, BDRV_FIX_LEAKS, &rebuild,
+ &highest_cluster, refcount_table, nb_clusters);
+ if (rebuild) {
+ fprintf(stderr, "ERROR rebuilt refcount structure is still "
+ "broken\n");
+ }
+
+ /* Any leaks accounted for here were introduced by
+ * rebuild_refcount_structure() because that function has created a
+ * new refcount structure from scratch */
+ fresh_leaks = res->leaks;
+ *res = saved_res;
+ }
+
+ if (res->corruptions < old_res.corruptions) {
+ res->corruptions_fixed += old_res.corruptions - res->corruptions;
+ }
+ if (res->leaks < old_res.leaks) {
+ res->leaks_fixed += old_res.leaks - res->leaks;
+ }
+ res->leaks += fresh_leaks;
} else if (fix) {
if (rebuild) {
fprintf(stderr, "ERROR need to rebuild refcount structures\n");
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 16/17] iotests: Fix test outputs
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (14 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 15/17] qcow2: Clean up after refcount rebuild Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 17/17] iotests: Add test for potentially damaging repairs Max Reitz
2014-10-22 13:18 ` [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Kevin Wolf
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
039, 060 and 061 all create images with referenced clusters having a
refcount of 0. Because previous commits changed handling of such errors,
these tests now have a different output. Fix it.
Furthermore, 060 created a refblock with a refcount greater than one
which now results in having to rebuild the refcount structure as well.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
---
tests/qemu-iotests/039.out | 10 ++++++++--
tests/qemu-iotests/060.out | 10 ++++++++--
tests/qemu-iotests/061.out | 18 ++++++++++++------
3 files changed, 28 insertions(+), 10 deletions(-)
diff --git a/tests/qemu-iotests/039.out b/tests/qemu-iotests/039.out
index 67e7744..0adf153 100644
--- a/tests/qemu-iotests/039.out
+++ b/tests/qemu-iotests/039.out
@@ -25,7 +25,10 @@ read 512/512 bytes at offset 0
incompatible_features 0x1
== Repairing the image file must succeed ==
-Repairing cluster 5 refcount=0 reference=1
+ERROR cluster 5 refcount=0 reference=1
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
The following inconsistencies were found and repaired:
0 leaked clusters
@@ -45,7 +48,10 @@ wrote 512/512 bytes at offset 0
512 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
./039: Aborted ( ulimit -c 0; exec "$@" )
incompatible_features 0x1
-Repairing cluster 5 refcount=0 reference=1
+ERROR cluster 5 refcount=0 reference=1
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
wrote 512/512 bytes at offset 0
512 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
incompatible_features 0x0
diff --git a/tests/qemu-iotests/060.out b/tests/qemu-iotests/060.out
index cd679f9..9419da1 100644
--- a/tests/qemu-iotests/060.out
+++ b/tests/qemu-iotests/060.out
@@ -36,11 +36,15 @@ incompatible_features 0x0
qcow2: Marking image as corrupt: Preventing invalid write on metadata (overlaps with refcount block); further corruption events will be suppressed
write failed: Input/output error
incompatible_features 0x2
-Repairing refcount block 0 refcount=2
+ERROR refcount block 0 refcount=2
+ERROR cluster 2 refcount=1 reference=2
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=2 reference=1
The following inconsistencies were found and repaired:
0 leaked clusters
- 1 corruptions
+ 2 corruptions
Double checking the fixed image now...
No errors were found on the image.
@@ -68,6 +72,8 @@ incompatible_features 0x0
qcow2: Marking image as corrupt: Preventing invalid write on metadata (overlaps with inactive L2 table); further corruption events will be suppressed
write failed: Input/output error
incompatible_features 0x2
+ERROR cluster 4 refcount=1 reference=2
+Leaked cluster 9 refcount=1 reference=0
Repairing cluster 4 refcount=1 reference=2
Repairing cluster 9 refcount=1 reference=0
Repairing OFLAG_COPIED data cluster: l2_entry=8000000000040000 refcount=2
diff --git a/tests/qemu-iotests/061.out b/tests/qemu-iotests/061.out
index e372470..4ae6472 100644
--- a/tests/qemu-iotests/061.out
+++ b/tests/qemu-iotests/061.out
@@ -76,8 +76,11 @@ autoclear_features 0x0
refcount_order 4
header_length 104
-Repairing cluster 5 refcount=0 reference=1
-Repairing cluster 6 refcount=0 reference=1
+ERROR cluster 5 refcount=0 reference=1
+ERROR cluster 6 refcount=0 reference=1
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
magic 0x514649fb
version 2
backing_file_offset 0x0
@@ -87,7 +90,7 @@ size 67108864
crypt_method 0
l1_size 1
l1_table_offset 0x30000
-refcount_table_offset 0x10000
+refcount_table_offset 0x80000
refcount_table_clusters 1
nb_snapshots 0
snapshot_offset 0x0
@@ -230,8 +233,11 @@ autoclear_features 0x0
refcount_order 4
header_length 104
-Repairing cluster 5 refcount=0 reference=1
-Repairing cluster 6 refcount=0 reference=1
+ERROR cluster 5 refcount=0 reference=1
+ERROR cluster 6 refcount=0 reference=1
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
magic 0x514649fb
version 3
backing_file_offset 0x0
@@ -241,7 +247,7 @@ size 67108864
crypt_method 0
l1_size 1
l1_table_offset 0x30000
-refcount_table_offset 0x10000
+refcount_table_offset 0x80000
refcount_table_clusters 1
nb_snapshots 0
snapshot_offset 0x0
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [Qemu-devel] [PATCH v8 17/17] iotests: Add test for potentially damaging repairs
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (15 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 16/17] iotests: Fix test outputs Max Reitz
@ 2014-10-22 12:09 ` Max Reitz
2014-10-22 13:18 ` [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Kevin Wolf
17 siblings, 0 replies; 21+ messages in thread
From: Max Reitz @ 2014-10-22 12:09 UTC (permalink / raw)
To: qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi, Max Reitz
There are certain cases where repairing a qcow2 image might actually
damage it further (or rather, where repairing it has in fact damaged it
further with the old qcow2 check implementation). This should not
happen, so add a test for these cases.
Furthermore, the repair function now repairs refblocks beyond the image
end by resizing the image accordingly. Add several tests for this as
well.
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
tests/qemu-iotests/108 | 141 +++++++++++++++++++++++++++++++++++++++++++++
tests/qemu-iotests/108.out | 110 +++++++++++++++++++++++++++++++++++
tests/qemu-iotests/group | 1 +
3 files changed, 252 insertions(+)
create mode 100755 tests/qemu-iotests/108
create mode 100644 tests/qemu-iotests/108.out
diff --git a/tests/qemu-iotests/108 b/tests/qemu-iotests/108
new file mode 100755
index 0000000..12fc92a
--- /dev/null
+++ b/tests/qemu-iotests/108
@@ -0,0 +1,141 @@
+#!/bin/bash
+#
+# Test case for repairing qcow2 images which cannot be repaired using
+# the on-disk refcount structures
+#
+# Copyright (C) 2014 Red Hat, Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+# creator
+owner=mreitz@redhat.com
+
+seq="$(basename $0)"
+echo "QA output created by $seq"
+
+here="$PWD"
+tmp=/tmp/$$
+status=1 # failure is the default!
+
+_cleanup()
+{
+ _cleanup_test_img
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common.rc
+. ./common.filter
+
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux
+
+echo
+echo '=== Repairing an image without any refcount table ==='
+echo
+
+_make_test_img 64M
+# just write some data
+$QEMU_IO -c 'write -P 42 0 64k' "$TEST_IMG" | _filter_qemu_io
+
+# refcount_table_offset
+poke_file "$TEST_IMG" $((0x30)) "\x00\x00\x00\x00\x00\x00\x00\x00"
+# refcount_table_clusters
+poke_file "$TEST_IMG" $((0x38)) "\x00\x00\x00\x00"
+
+_check_test_img -r all
+
+$QEMU_IO -c 'read -P 42 0 64k' "$TEST_IMG" | _filter_qemu_io
+
+echo
+echo '=== Repairing unreferenced data cluster in new refblock area ==='
+echo
+
+IMGOPTS='cluster_size=512' _make_test_img 64M
+# Allocate the first 128 kB in the image (first refblock)
+$QEMU_IO -c 'write 0 0x1b200' "$TEST_IMG" | _filter_qemu_io
+# should be 131072 == 0x20000
+stat -c '%s' "$TEST_IMG"
+
+# Enter a cluster at 128 kB (0x20000)
+# XXX: This should be the first free entry in the last L2 table, but we cannot
+# be certain
+poke_file "$TEST_IMG" $((0x1ccc8)) "\x80\x00\x00\x00\x00\x02\x00\x00"
+
+# Fill the cluster
+truncate -s $((0x20200)) "$TEST_IMG"
+$QEMU_IO -c "open -o driver=raw $TEST_IMG" -c 'write -P 42 128k 512' \
+ | _filter_qemu_io
+
+# The data should now appear at this guest offset
+$QEMU_IO -c 'read -P 42 0x1b200 512' "$TEST_IMG" | _filter_qemu_io
+
+# This cluster is unallocated; fix it
+_check_test_img -r all
+
+# This repair operation must have allocated a new refblock; and that refblock
+# should not overlap with the unallocated data cluster. If it does, the data
+# will be damaged, so check it.
+$QEMU_IO -c 'read -P 42 0x1b200 512' "$TEST_IMG" | _filter_qemu_io
+
+echo
+echo '=== Repairing refblock beyond the image end ==='
+echo
+
+echo
+echo '--- Otherwise clean ---'
+echo
+
+_make_test_img 64M
+# Normally, qemu doesn't create empty refblocks, so we just have to do it by
+# hand
+# XXX: This should be the entry for the second refblock
+poke_file "$TEST_IMG" $((0x10008)) "\x00\x00\x00\x00\x00\x10\x00\x00"
+# Mark that refblock as used
+# XXX: This should be the 17th entry (cluster 16) of the first
+# refblock
+poke_file "$TEST_IMG" $((0x20020)) "\x00\x01"
+_check_test_img -r all
+
+echo
+echo '--- Refblock is unallocated ---'
+echo
+
+_make_test_img 64M
+poke_file "$TEST_IMG" $((0x10008)) "\x00\x00\x00\x00\x00\x10\x00\x00"
+_check_test_img -r all
+
+echo
+echo '--- Signed overflow after the refblock ---'
+echo
+
+_make_test_img 64M
+poke_file "$TEST_IMG" $((0x10008)) "\x7f\xff\xff\xff\xff\xff\x00\x00"
+_check_test_img -r all
+
+echo
+echo '--- Unsigned overflow after the refblock ---'
+echo
+
+_make_test_img 64M
+poke_file "$TEST_IMG" $((0x10008)) "\xff\xff\xff\xff\xff\xff\x00\x00"
+_check_test_img -r all
+
+# success, all done
+echo '*** done'
+rm -f $seq.full
+status=0
diff --git a/tests/qemu-iotests/108.out b/tests/qemu-iotests/108.out
new file mode 100644
index 0000000..824d5cf
--- /dev/null
+++ b/tests/qemu-iotests/108.out
@@ -0,0 +1,110 @@
+QA output created by 108
+
+=== Repairing an image without any refcount table ===
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+wrote 65536/65536 bytes at offset 0
+64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+ERROR cluster 0 refcount=0 reference=1
+ERROR cluster 3 refcount=0 reference=1
+ERROR cluster 4 refcount=0 reference=1
+ERROR cluster 5 refcount=0 reference=1
+Rebuilding refcount structure
+The following inconsistencies were found and repaired:
+
+ 0 leaked clusters
+ 4 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+read 65536/65536 bytes at offset 0
+64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+
+=== Repairing unreferenced data cluster in new refblock area ===
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+wrote 111104/111104 bytes at offset 0
+108.500 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+131072
+wrote 512/512 bytes at offset 131072
+512 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 512/512 bytes at offset 111104
+512 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+ERROR cluster 256 refcount=0 reference=1
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
+The following inconsistencies were found and repaired:
+
+ 0 leaked clusters
+ 1 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+read 512/512 bytes at offset 111104
+512 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+
+=== Repairing refblock beyond the image end ===
+
+
+--- Otherwise clean ---
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+Repairing refcount block 1 is outside image
+The following inconsistencies were found and repaired:
+
+ 0 leaked clusters
+ 1 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+
+--- Refblock is unallocated ---
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+Repairing refcount block 1 is outside image
+ERROR cluster 16 refcount=0 reference=1
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
+Repairing cluster 16 refcount=1 reference=0
+The following inconsistencies were found and repaired:
+
+ 0 leaked clusters
+ 2 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+
+--- Signed overflow after the refblock ---
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+Repairing refcount block 1 is outside image
+ERROR could not resize image: Invalid argument
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
+The following inconsistencies were found and repaired:
+
+ 0 leaked clusters
+ 1 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+
+--- Unsigned overflow after the refblock ---
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+Repairing refcount block 1 is outside image
+ERROR could not resize image: Invalid argument
+Rebuilding refcount structure
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
+The following inconsistencies were found and repaired:
+
+ 0 leaked clusters
+ 1 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+*** done
diff --git a/tests/qemu-iotests/group b/tests/qemu-iotests/group
index b230996..be2054f 100644
--- a/tests/qemu-iotests/group
+++ b/tests/qemu-iotests/group
@@ -106,3 +106,4 @@
103 rw auto quick
104 rw auto
105 rw auto quick
+108 rw auto quick
--
1.9.3
^ permalink raw reply related [flat|nested] 21+ messages in thread
* Re: [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
` (16 preceding siblings ...)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 17/17] iotests: Add test for potentially damaging repairs Max Reitz
@ 2014-10-22 13:18 ` Kevin Wolf
17 siblings, 0 replies; 21+ messages in thread
From: Kevin Wolf @ 2014-10-22 13:18 UTC (permalink / raw)
To: Max Reitz; +Cc: qemu-devel, Stefan Hajnoczi, Benoît Canet
Am 22.10.2014 um 14:09 hat Max Reitz geschrieben:
> As can be seen in the final patch of this series, there are certain
> cases where the current repair implementation of qcow2 actually damages
> the image further because it allocates new clusters for the refcount
> structure which overlap with existing but according to the on-disk
> refcounts (which are assumed to be wrong to begin with) unallocated
> clusters.
>
> This series fixes this by completely recreating the refcount structure
> based on the in-memory information calculated during the check operation
> if the possibility of damaging the image while repairing the refcount
> structures in-place exists.
Thanks, applied to the block branch.
Kevin
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0()
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0() Max Reitz
@ 2014-10-22 14:02 ` Eric Blake
0 siblings, 0 replies; 21+ messages in thread
From: Eric Blake @ 2014-10-22 14:02 UTC (permalink / raw)
To: Max Reitz, qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi
[-- Attachment #1: Type: text/plain, Size: 514 bytes --]
On 10/22/2014 06:09 AM, Max Reitz wrote:
> These functions call their non-0-counterparts and then fill the
> allocated buffer with 0 (if the allocation has been successful).
>
> Signed-off-by: Max Reitz <mreitz@redhat.com>
> ---
> block.c | 16 ++++++++++++++++
> include/block/block.h | 2 ++
> 2 files changed, 18 insertions(+)
Reviewed-by: Eric Blake <eblake@redhat.com>
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 539 bytes --]
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [Qemu-devel] [PATCH v8 05/17] qcow2: Use sizeof(**refcount_table)
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 05/17] qcow2: Use sizeof(**refcount_table) Max Reitz
@ 2014-10-22 14:38 ` Eric Blake
0 siblings, 0 replies; 21+ messages in thread
From: Eric Blake @ 2014-10-22 14:38 UTC (permalink / raw)
To: Max Reitz, qemu-devel; +Cc: Kevin Wolf, Benoît Canet, Stefan Hajnoczi
[-- Attachment #1: Type: text/plain, Size: 602 bytes --]
On 10/22/2014 06:09 AM, Max Reitz wrote:
> When implementing variable refcounts, we want to be able to easily find
> all the places in qemu which are tied to a certain refcount order.
> Replace sizeof(uint16_t) in the check code by sizeof(**refcount_table)
> so we can later find it more easily.
>
> Signed-off-by: Max Reitz <mreitz@redhat.com>
> ---
> block/qcow2-refcount.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
Reviewed-by: Eric Blake <eblake@redhat.com>
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 539 bytes --]
^ permalink raw reply [flat|nested] 21+ messages in thread
end of thread, other threads:[~2014-10-22 14:38 UTC | newest]
Thread overview: 21+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-10-22 12:09 [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 01/17] block: Add qemu_{, try_}blockalign0() Max Reitz
2014-10-22 14:02 ` Eric Blake
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 02/17] qcow2: Calculate refcount block entry count Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 03/17] qcow2: Fix leaks in dirty images Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 04/17] qcow2: Split qcow2_check_refcounts() Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 05/17] qcow2: Use sizeof(**refcount_table) Max Reitz
2014-10-22 14:38 ` Eric Blake
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 06/17] qcow2: Pull check_refblocks() up Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 07/17] qcow2: Use int64_t for in-memory reftable size Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 08/17] qcow2: Split fail code in L1 and L2 checks Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 09/17] qcow2: Let inc_refcounts() return -errno Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 10/17] qcow2: Let inc_refcounts() resize the reftable Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 11/17] qcow2: Reuse refcount table in calculate_refcounts() Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 12/17] qcow2: Fix refcount blocks beyond image end Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 13/17] qcow2: Do not perform potentially damaging repairs Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 14/17] qcow2: Rebuild refcount structure during check Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 15/17] qcow2: Clean up after refcount rebuild Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 16/17] iotests: Fix test outputs Max Reitz
2014-10-22 12:09 ` [Qemu-devel] [PATCH v8 17/17] iotests: Add test for potentially damaging repairs Max Reitz
2014-10-22 13:18 ` [Qemu-devel] [PATCH v8 00/17] qcow2: Fix image repairing Kevin Wolf
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).