This patch add shrinking of the image file for qcow2. As a result, this allows
us to reduce the virtual image size and free up space on the disk without
copying the image. Image can be fragmented and shrink is done by punching holes
in the image file.
# ./qemu-img create -f qcow2 image.qcow2 4G
Formatting 'image.qcow2', fmt=qcow2 size=4294967296 encryption=off
cluster_size=65536 lazy_refcounts=off refcount_bits=16
# ./qemu-io -c "write -P 0x22 0 1G" image.qcow2
wrote 1073741824/1073741824 bytes at offset 0
1 GiB, 1 ops; 0:00:04.59 (222.886 MiB/sec and 0.2177 ops/sec)
# ./qemu-img resize image.qcow2 512M
warning: qemu-img: Shrinking an image will delete all data beyond the shrunken
image's end. Before performing such an operation, make sure there is no
important data there.
error: qemu-img: Use the --shrink option to perform a shrink operation.
# ./qemu-img resize --shrink image.qcow2 128M
Image resized.
# ./qemu-img info image.qcow2
image: image.qcow2
file format: qcow2
virtual size: 128M (134217728 bytes)
disk size: 128M
cluster_size: 65536
Format specific information:
compat: 1.1
lazy refcounts: false
refcount bits: 16
corrupt: false
# du -h image.qcow2
129M image.qcow2
Changes from v1:
- add --shrink flag for qemu-img resize
- add qcow2_cache_discard
- simplify qcow2_shrink_l1_table() to reduce the likelihood of image corruption
- add new qemu-iotests for shrinking images
Changes from v2:
- replace qprintf() on error_report() (1)
- rewrite warning messages (1)
- enforce --shrink flag for all formats except raw (1)
- split qcow2_cache_discard() (2)
- minor fixes according to comments (3)
- rewrite the last part of qcow2_shrink_reftable() to avoid
qcow2_free_clusters() calls inside (3)
- improve test for shrinking image (4)
Changes from v3:
- rebase on "Implement a warning_report function" Alistair's patch-set (1)
- spelling fixes (1)
- the man page fix according to the discussion (1)
- add call qcow2_signal_corruption() in case of image corruption (3)
Changes from v4:
- rebase on https://github.com/XanClic/qemu/commits/block Max's block branch
Changes from v5:
- the condition refcount == 0 should be enough to evict the l2/refcount cluster
from the cache (2)
- overwrite the l1/refcount table in memory with zeros, even if overwriting the
l1/refcount table on disk has failed (3)
- replace g_try_malloc() on g_malloc() for allocation reftable_tmp (3)
Pavel Butsykin (4):
qemu-img: add --shrink flag for resize
qcow2: add qcow2_cache_discard
qcow2: add shrink image support
qemu-iotests: add shrinking image test
block/qcow2-cache.c | 26 +++++++
block/qcow2-cluster.c | 50 +++++++++++++
block/qcow2-refcount.c | 140 ++++++++++++++++++++++++++++++++++++-
block/qcow2.c | 43 +++++++++---
block/qcow2.h | 17 +++++
qapi/block-core.json | 3 +-
qemu-img-cmds.hx | 4 +-
qemu-img.c | 23 ++++++
qemu-img.texi | 6 +-
tests/qemu-iotests/102 | 4 +-
tests/qemu-iotests/163 | 170 +++++++++++++++++++++++++++++++++++++++++++++
tests/qemu-iotests/163.out | 5 ++
tests/qemu-iotests/group | 1 +
13 files changed, 475 insertions(+), 17 deletions(-)
create mode 100644 tests/qemu-iotests/163
create mode 100644 tests/qemu-iotests/163.out
--
2.13.0
