The metadata overlap checks introduced in a40f1c2add help detect
corruption in the qcow2 image by verifying that data writes don't
overlap with existing metadata sections.
The 'refcount-block' check in particular iterates over the refcount
table in order to get the addresses of all refcount blocks and check
that none of them overlap with the region where we want to write.
The problem with the refcount table is that since it always occupies
complete clusters its size is usually very big. With the default
values of cluster_size=64KB and refcount_bits=16 this table holds 8192
entries, each one of them enough to map 2GB worth of host clusters.
So unless we're using images with several TB of allocated data this
table is going to be mostly empty, and iterating over it is a waste of
CPU. If the storage backend is fast enough this can have an effect on
I/O performance.
This patch keeps the index of the last used (i.e. non-zero) entry in
the refcount table and updates it every time the table changes. The
refcount-block overlap check then uses that index instead of reading
the whole table.
In my tests with a 4GB qcow2 file stored in RAM this doubles the
amount of write IOPS.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Message-id: 20170201123828.4815-1-berto@igalia.com
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
It was from the time when none of the global functions had a qcow2_
prefix.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Handling this is similar to what is done to the L2 entry in the case of
compressed clusters.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Unused function declarations were found using a simple gcc plugin and
manually verified by grepping the sources.
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This changes qcow2 to implement the byte-based .bdrv_co_pwritev
interface rather than the sector-based old one.
As preallocation uses the same allocation function as normal writes, and
the interface of that function needs to be changed, it is converted in
the same patch.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reading from qcow2 images is now byte granularity.
Most of the affected code in qcow2 actually gets simpler with this
change. The only exception is encryption, which is fixed on 512 bytes
blocks; in order to keep this working, bs->request_alignment is set for
encrypted images.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
The problem with excessive flushing was found by a couple of performance
tests:
- parallel directory tree creation (from 2 processes)
- 32 cached writes + fsync at the end in a loop
For the first one results improved from 2.6 loops/sec to 3.5 loops/sec.
Each loop creates 10^3 directories with 10 files in each.
For the second one results improved from ~600 fsync/sec to ~1100
fsync/sec. Though, it was run on SSD so it probably won't show such
performance gain on rotational media.
qcow2_cache_flush() calls bdrv_flush() unconditionally after writing
cache entries of a particular cache. This can lead to as many as
2 additional fdatasyncs inside bdrv_flush.
We can simply skip all fdatasync calls inside qcow2_co_flush_to_os
as bdrv_flush for sure will do the job. These flushes are necessary to
keep the right order of writes to the different caches. Though this is
not necessary in the current code base as this ordering is ensured through
the flush in qcow2_cache_flush_dependency().
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Pavel Borzenkov <pborzenkov@virtuozzo.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add a function qcow2_change_refcount_order() which allows changing the
refcount order of a qcow2 image.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add an opaque value which is to be passed to the bdrv_amend_options()
status callback.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The coroutine files are currently referenced by the block-obj-y
variable. The coroutine functionality though is already used by
more than just the block code. eg migration code uses coroutine
yield. In the future the I/O channel code will also use the
coroutine yield functionality. Since the coroutine code is nicely
self-contained it can be easily built as part of the libqemuutil.a
library, making it widely available.
The headers are also moved into include/qemu, instead of the
include/block directory, since they are now part of the util
codebase, and the impl was never in the block/ directory
either.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
This struct doesn't exist any more since commit 3fc48d09 in August 2011,
it's about time to remove its forward declaration.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Sadly, some images may have more clusters than what can be represented
using a plain int. We should be prepared for that case (in
qcow2_check_refcounts() we actually were trying to catch that case, but
since size_to_clusters() truncated the returned value, that check never
did anything useful).
Cc: qemu-stable <qemu-stable@nongnu.org>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
BDRVQcowState is already used by qcow1, and gdb is always confused which
one to use. Rename the qcow2 one so they can be distinguished.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
This adds a new 'cache-clean-interval' option that cleans all qcow2
cache entries that haven't been used in a certain interval, given in
seconds.
This allows setting a large L2 cache size so it can handle scenarios
with lots of I/O and at the same time use little memory during periods
of inactivity.
This feature currently relies on MADV_DONTNEED to free that memory, so
it is not useful in systems that don't follow that behavior.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: a70d12da60433df9360ada648b3f34b8f6f354ce.1438690126.git.berto@igalia.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
Switch the qcow/qcow2 block driver over to use the generic cipher
API, this allows it to use the pluggable AES implementations,
instead of being hardcoded to use QEMU's built-in impl.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1435770638-25715-10-git-send-email-berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To prepare for a generic internal cipher API, move the
built-in AES implementation into the crypto/ directory
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1435770638-25715-3-git-send-email-berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If a relatively large cluster size is chosen, the default of 1 MB L2
cache is not really appropriate. In this case, unless overridden by the
user, the default cache size should not be determined by its size in
bytes but by the number of L2 tables (clusters) it is supposed to
contain.
Note that without this patch, MIN_L2_CACHE_SIZE will effectively take
over the same role. However, providing space for just two L2 tables is
not enough to be the default.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The L2 cache must cover at least two L2 tables, because during COW two
L2 tables are accessed simultaneously.
Reported-by: Alexander Graf <agraf@suse.de>
Cc: qemu-stable <qemu-stable@nongnu.org>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Tested-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This function never receives an invalid table pointer, so we can make
it void and remove all the error checking code.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The qcow2 L2/refcount cache contains one separate table for each cache
entry. Doing one allocation per table adds unnecessary overhead and it
also requires us to store the address of each table separately.
Since the size of the cache is constant during its lifetime, it's
better to have an array that contains all the tables using one single
allocation.
In my tests measuring freshly created caches with sizes 128MB (L2) and
32MB (refcount) this uses around 10MB of RAM less.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In recent qemu versions, it is possible to override the backing file
name and format that is stored in the image file with values given at
runtime. In such cases, the temporary override could end up in the
image header if the qcow2 header was updated, while obviously correct
behaviour would be to leave the on-disk backing file path/format
unchanged.
Fix this and add a test case for it.
Reported-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1428411796-2852-1-git-send-email-kwolf@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Since refcounts do not always have to be a uint16_t, all refcount blocks
and arrays in memory should not have a specific type (thus they become
pointers to void) and for accessing them, two helper functions are used
(a getter and a setter). Those functions are called indirectly through
function pointers in the BDRVQcowState so they may later be exchanged
for different refcount orders.
With the check and repair functions using this function, the refcount
array they are creating will be in big endian byte order; additionally,
using realloc_refcount_array() makes the size of this refcount array
always cluster-aligned. Both combined allow rebuild_refcount_structure()
to drop the bounce buffer which was used to convert parts of the
refcount array to big endian byte order and store them on disk. Instead,
those parts can now be written directly.
[ kwolf: Fixed a build failure on 32 bit and another with old glib ]
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Refcounts may have a width of up to 64 bits, so qemu should use the same
width to represent refcount values internally.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
update_refcount() and qcow2_update_cluster_refcount() currently take a
signed addend. At least one caller passes a value directly derived from
an absolute refcount that should be reached ("l2_refcount - 1" in
expand_zero_clusters_in_l1()). Therefore, the addend should be unsigned
as well; this will be especially important for 64 bit refcounts.
Because update_refcount() then no longer knows whether the refcount
should be increased or decreased, it now requires an additional flag
which specified exactly that. The same applies to
qcow2_update_cluster_refcount().
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Refcounts can theoretically be of type uint64_t; in order to be able to
represent the full range, qcow2_get_refcount() cannot use a single
variable to represent both all refcount values and also keep some values
reserved for errors.
One solution would be to add an Error pointer parameter to
qcow2_get_refcount(); however, no caller could (currently) pass that
error message, so it would have to be emitted immediately and be
passed to the next caller by returning -EIO or something similar.
Therefore, an Error parameter does not offer any advantages here.
The solution applied by this patch is simpler to use. Because no caller
would be able to pass the error message, they would have to print it and
free it, whereas with this patch the caller only needs to pass the
returned integer (which is often a no-op from the code perspective,
because that integer will be stored in a variable "ret" which will be
returned by the fail path of many callers).
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add two new fields regarding refcount information (the bit width of
every entry and the maximum refcount value) to the BDRVQcowState.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The only user went away five years ago with commit a9420734 ('qcow2:
Simplify image creation'). It's about time to remove it.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reading the refcount of a cluster is an operation which can be useful in
all of the qcow2 code, so make that function globally available.
While touching this function, amend the comment describing the "addend"
parameter: It is (no longer, if it ever was) necessary to have it set to
-1 or 1; any value is fine.
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>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Message-id: 1414404776-4919-6-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The only really time-consuming operation potentially performed by
qcow2_amend_options() is zero cluster expansion when downgrading qcow2
images from compat=1.1 to compat=0.10, so report status of that
operation and that operation only through the status CB.
For this, approximate the progress as the number of L1 entries visited
during the operation.
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>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Message-id: 1414404776-4919-5-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Normally, discarded sectors should read back as zero. However, there are
cases in which a sector (or rather cluster) should be discarded as if
they were never written in the first place, that is, reading them should
fall through to the backing file again.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1414159063-25977-2-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
With BDRVQcowState.refcount_block_bits, we don't need REFCOUNT_SHIFT
anymore.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
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>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Being able to set the overlap-check option to a string and then refine
it via the overlap-check.* options is a nice idea for the command line
but does not work so well for non-flattened dicts. In that case, one can
only specify either but not both, so add a field to overlap-check.*
which does the same as directly specifying overlap-check but can be used
in conjunction with the other fields in non-flattened dicts.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1408557576-14574-4-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Add a helper function for easily marking an image corrupt (on fatal
corruptions) while outputting an informative message to stderr and via
QAPI.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Message-id: 1409926039-29044-3-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Add options for specifying the size of the metadata caches. This can
either be done directly for each cache (if only one is given, the other
will be derived according to a default ratio) or combined for both.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Specifying the metadata cache sizes in clusters results in less clusters
(and much less bytes) covered for small cluster sizes and vice versa.
Using a constant byte size reduces this difference, and makes it
possible to manually specify the cache size in an easily comprehensible
unit.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Even with a limit of 64k snapshots, each snapshot could have a filename
and an ID with up to 64k, which would still lead to pretty large
allocations, which could potentially lead to qemu aborting. Limit the
total size of the snapshot table to an average of 1k per entry when
the limit of 64k snapshots is fully used. This should be plenty for any
reasonable user.
This also fixes potential integer overflows of s->snapshot_size.
Suggested-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This avoids an unbounded allocation.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
In order to avoid integer overflows.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
If the size becomes larger than what qcow2_open() would accept, fail the
growing operation.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This avoid unbounded memory allocation and fixes a potential buffer
overflow on 32 bit hosts.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
n_start can be actually calculated from offset. The number of
sectors to be allocated(n_end - n_start) can be passed in in
num. By removing n_start and n_end, we can save two parameters.
The side effect is there is a bug in qcow2.c:preallocate() that
passes incorrect n_start to qcow2_alloc_cluster_offset() is
fixed. The bug can be triggerred by a larger cluster size than
the default value(65536), for example:
./qemu-img create -f qcow2 \
-o 'cluster_size=131072,preallocation=metadata' file.img 4G
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Accoring to qcow spec, the offset fields in l1e, l2e and ref table entry
start at bit 9. The offset is cluster offset, and the smallest possible
cluster size is 512 bytes.
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since later this function will be used so improve it. The only caller of it
now is qemu-img, and it is not impacted by introduce function
bdrv_snapshot_load_tmp_by_id_or_name() that call bdrv_snapshot_load_tmp()
twice to keep old search logic. bdrv_snapshot_load_tmp_by_id_or_name() return
int to let caller know the errno, and errno will be used later.
Also fix a typo in comments of bdrv_snapshot_delete().
Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Introduces the macros QCOW2_OL_CONSTANT and QCOW2_OL_ALL in addition to
the already existing QCOW2_OL_CACHED, signifying all metadata overlap
checks that can be performed in constant time (regardless of image size
etc.) and truly all available overlap checks, respectively.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add runtime options to tune the overlap checks to be performed before
write accesses.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Replace the QCOW2_OL_DEFAULT macro by a variable overlap_check in
BDRVQcowState.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In qcow2_check_metadata_overlap and qcow2_pre_write_overlap_check,
change the parameter signifying the checks to perform from its current
positive form to a negative one, i.e., it will no longer explicitly
specify every check to perform but rather a mask of checks not to
perform.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
QCowHeader and QCowExtension are structs that reside in the on-disk
image format, and are read and written directly via bdrv_pread()/write(),
and as such should be packed to avoid any unintentional struct padding.
Signed-off-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
