Skip to contentSkip to conversation

qcow2: Implement zstd cluster compression method

PATCH Superseded Denis Plotnikov 3 participants 11 messages Mar 23, 2020 – Mar 30, 2020 Block layer

v9 of 24 Superseded Latest is v24 →

lore
Share & tools
lore
#0 lore raw

v9 cover qcow2: Implement zstd cluster compression method

v9:
* 01: fix error checking and reporting in qcow2_amend compression type part [Vladimir]
* 03: replace asserts with -EIO in qcow2_zstd_decompression [Vladimir, Alberto]
* 03: reword/amend/add comments, fix typos [Vladimir]

v8:
* 03: switch zstd API from simple to stream [Eric] No need to state a special cluster layout for zstd compressed clusters.
v7:
* use qapi_enum_parse instead of the open-coding [Eric]
* fix wording, typos and spelling [Eric]

v6:
* "block/qcow2-threads: fix qcow2_decompress" is removed from the series since it has been accepted by Max already
* add compile time checking for Qcow2Header to be a multiple of 8 [Max, Alberto]
* report error on qcow2 amending when the compression type is actually chnged [Max]
* remove the extra space and the extra new line [Max]
* re-arrange acks and signed-off-s [Vladimir]

v5:
* replace -ENOTSUP with abort in qcow2_co_decompress [Vladimir]
* set cluster size for all test cases in the beginning of the 287 test

v4:
* the series is rebased on top of 01 "block/qcow2-threads: fix qcow2_decompress"
* 01 is just a no-change resend to avoid extra dependencies. Still, it may be merged in separate

v3:
* remove redundant max compression type value check [Vladimir, Eric] (the switch below checks everything)
* prevent compression type changing on "qemu-img amend" [Vladimir]
* remove zstd config setting, since it has been added already by "migration" patches [Vladimir]
* change the compression type error message [Vladimir]
* fix alignment and 80-chars exceeding [Vladimir]

v2:
* rework compression type setting [Vladimir]
* squash iotest changes to the compression type introduction patch [Vladimir, Eric]
* fix zstd availability checking in zstd iotest [Vladimir]
* remove unnecessry casting [Eric]
* remove rudundant checks [Eric]
* fix compressed cluster layout in qcow2 spec [Vladimir]
* fix wording [Eric, Vladimir]
* fix compression type filtering in iotests [Eric]

v1:
the initial series

Denis Plotnikov (4):
qcow2: introduce compression type feature
qcow2: rework the cluster compression routine
qcow2: add zstd cluster compression
iotests: 287: add qcow2 compression type test

 docs/interop/qcow2.txt           |   1 +
 configure                        |   2 +-
 qapi/block-core.json             |  23 +++-
 block/qcow2.h                    |  20 ++-
 include/block/block_int.h        |   1 +
 block/qcow2-threads.c            | 205 +++++++++++++++++++++++++++++--
 block/qcow2.c                    | 120 ++++++++++++++++++
 tests/qemu-iotests/031.out       |  14 +--
 tests/qemu-iotests/036.out       |   4 +-
 tests/qemu-iotests/049.out       | 102 +++++++--------
 tests/qemu-iotests/060.out       |   1 +
 tests/qemu-iotests/061.out       |  34 ++---
 tests/qemu-iotests/065           |  28 +++--
 tests/qemu-iotests/080           |   2 +-
 tests/qemu-iotests/144.out       |   4 +-
 tests/qemu-iotests/182.out       |   2 +-
 tests/qemu-iotests/242.out       |   5 +
 tests/qemu-iotests/255.out       |   8 +-
 tests/qemu-iotests/287           | 128 +++++++++++++++++++
 tests/qemu-iotests/287.out       |  43 +++++++
 tests/qemu-iotests/common.filter |   3 +-
 tests/qemu-iotests/group         |   1 +
 22 files changed, 643 insertions(+), 108 deletions(-)
 create mode 100755 tests/qemu-iotests/287
 create mode 100644 tests/qemu-iotests/287.out
Signature · 1 line
2.17.0
#1 lore raw

PATCH v9 1/4 qcow2: introduce compression type feature

The patch adds some preparation parts for incompatible compression type feature to qcow2 allowing the use different compression methods for image clusters (de)compressing.

It is implied that the compression type is set on the image creation and can be changed only later by image conversion, thus compression type defines the only compression algorithm used for the image, and thus, for all image clusters.

The goal of the feature is to add support of other compression methods to qcow2. For example, ZSTD which is more effective on compression than ZLIB.

The default compression is ZLIB. Images created with ZLIB compression type are backward compatible with older qemu versions.

Adding of the compression type breaks a number of tests because now the compression type is reported on image creation and there are some changes in the qcow2 header in size and offsets.

The tests are fixed in the following ways:

    * filter out compression_type for many tests
    * fix header size, feature table size and backing file offset
      affected tests: 031, 036, 061, 080
      header_size +=8: 1 byte compression type
                       7 bytes padding
      feature_table += 48: incompatible feature compression type
      backing_file_offset += 56 (8 + 48 -> header_change + feature_table_change)
    * add "compression type" for test output matching when it isn't filtered
      affected tests: 049, 060, 061, 065, 144, 182, 242, 255
---
 qapi/block-core.json             |  22 +++++-
 block/qcow2.h                    |  20 +++++-
 include/block/block_int.h        |   1 +
 block/qcow2.c                    | 113 +++++++++++++++++++++++++++++++
 tests/qemu-iotests/031.out       |  14 ++--
 tests/qemu-iotests/036.out       |   4 +-
 tests/qemu-iotests/049.out       | 102 ++++++++++++++--------------
 tests/qemu-iotests/060.out       |   1 +
 tests/qemu-iotests/061.out       |  34 ++++++----
 tests/qemu-iotests/065           |  28 +++++---
 tests/qemu-iotests/080           |   2 +-
 tests/qemu-iotests/144.out       |   4 +-
 tests/qemu-iotests/182.out       |   2 +-
 tests/qemu-iotests/242.out       |   5 ++
 tests/qemu-iotests/255.out       |   8 +--
 tests/qemu-iotests/common.filter |   3 +-
 16 files changed, 267 insertions(+), 96 deletions(-)
qapi/block-core.json +20 -2
index 85e27bb61f..a306484973 100644
@@ -78,6 +78,8 @@
 #
 # @bitmaps: A list of qcow2 bitmap details (since 4.0)
 #
+# @compression-type: the image cluster compression method (since 5.0)
+#
 # Since: 1.7
 ##
 { 'struct': 'ImageInfoSpecificQCow2',
@@ -89,7 +91,8 @@
       '*corrupt': 'bool',
       'refcount-bits': 'int',
       '*encrypt': 'ImageInfoSpecificQCow2Encryption',
-      '*bitmaps': ['Qcow2BitmapInfo']
+      '*bitmaps': ['Qcow2BitmapInfo'],
+      'compression-type': 'Qcow2CompressionType'
   } }
 
 ##
@@ -4392,6 +4395,18 @@
   'data': [ 'v2', 'v3' ] }
 
 
+##
+# @Qcow2CompressionType:
+#
+# Compression type used in qcow2 image file
+#
+# @zlib: zlib compression, see <http://zlib.net/>
+#
+# Since: 5.0
+##
+{ 'enum': 'Qcow2CompressionType',
+  'data': [ 'zlib' ] }
+
 ##
 # @BlockdevCreateOptionsQcow2:
 #
@@ -4415,6 +4430,8 @@
 #                 allowed values: off, falloc, full, metadata)
 # @lazy-refcounts: True if refcounts may be updated lazily (default: off)
 # @refcount-bits: Width of reference counts in bits (default: 16)
+# @compression-type: The image cluster compression method
+#                    (default: zlib, since 5.0)
 #
 # Since: 2.12
 ##
@@ -4430,7 +4447,8 @@
             '*cluster-size':    'size',
             '*preallocation':   'PreallocMode',
             '*lazy-refcounts':  'bool',
-            '*refcount-bits':   'int' } }
+            '*refcount-bits':   'int',
+            '*compression-type':'Qcow2CompressionType' } }
 
 ##
 # @BlockdevCreateOptionsQed:
block/qcow2.h +19 -1
index 0942126232..cb6bf2ab83 100644
@@ -146,8 +146,16 @@ typedef struct QCowHeader {
 
     uint32_t refcount_order;
     uint32_t header_length;
+
+    /* Additional fields */
+    uint8_t compression_type;
+
+    /* header must be a multiple of 8 */
+    uint8_t padding[7];
 } QEMU_PACKED QCowHeader;
 
+QEMU_BUILD_BUG_ON(sizeof(QCowHeader) % 8 != 0);
+
 typedef struct QEMU_PACKED QCowSnapshotHeader {
     /* header is 8 byte aligned */
     uint64_t l1_table_offset;
@@ -216,13 +224,16 @@ enum {
     QCOW2_INCOMPAT_DIRTY_BITNR      = 0,
     QCOW2_INCOMPAT_CORRUPT_BITNR    = 1,
     QCOW2_INCOMPAT_DATA_FILE_BITNR  = 2,
+    QCOW2_INCOMPAT_COMPRESSION_BITNR = 3,
     QCOW2_INCOMPAT_DIRTY            = 1 << QCOW2_INCOMPAT_DIRTY_BITNR,
     QCOW2_INCOMPAT_CORRUPT          = 1 << QCOW2_INCOMPAT_CORRUPT_BITNR,
     QCOW2_INCOMPAT_DATA_FILE        = 1 << QCOW2_INCOMPAT_DATA_FILE_BITNR,
+    QCOW2_INCOMPAT_COMPRESSION      = 1 << QCOW2_INCOMPAT_COMPRESSION_BITNR,
 
     QCOW2_INCOMPAT_MASK             = QCOW2_INCOMPAT_DIRTY
                                     | QCOW2_INCOMPAT_CORRUPT
-                                    | QCOW2_INCOMPAT_DATA_FILE,
+                                    | QCOW2_INCOMPAT_DATA_FILE
+                                    | QCOW2_INCOMPAT_COMPRESSION,
 };
 
 /* Compatible feature bits */
@@ -369,6 +380,13 @@ typedef struct BDRVQcow2State {
 
     bool metadata_preallocation_checked;
     bool metadata_preallocation;
+    /*
+     * Compression type used for the image. Default: 0 - ZLIB
+     * The image compression type is set on image creation.
+     * For now, the only way to change the compression type
+     * is to convert the image with the desired compression type set.
+     */
+    Qcow2CompressionType compression_type;
 } BDRVQcow2State;
 
 typedef struct Qcow2COWRegion {
include/block/block_int.h +1 -0
index 6f9fd5e20e..2c6bb9dc99 100644
@@ -57,6 +57,7 @@
 #define BLOCK_OPT_REFCOUNT_BITS     "refcount_bits"
 #define BLOCK_OPT_DATA_FILE         "data_file"
 #define BLOCK_OPT_DATA_FILE_RAW     "data_file_raw"
+#define BLOCK_OPT_COMPRESSION_TYPE  "compression_type"
 
 #define BLOCK_PROBE_BUF_SIZE        512
 
block/qcow2.c +113 -0

191 lines not on this page · raw message

tests/qemu-iotests/031.out +7 -7
index d535e407bc..ed51afe9ce 100644
@@ -113,11 +113,11 @@ incompatible_features     []
 compatible_features       []
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 Header extension:
@@ -146,11 +146,11 @@ incompatible_features     []
 compatible_features       []
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 Header extension:
@@ -164,7 +164,7 @@ No errors were found on the image.
 
 magic                     0x514649fb
 version                   3
-backing_file_offset       0x178
+backing_file_offset       0x1b0
 backing_file_size         0x17
 cluster_bits              16
 size                      67108864
@@ -179,7 +179,7 @@ incompatible_features     []
 compatible_features       []
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0xe2792aca
@@ -188,7 +188,7 @@ data                      'host_device'
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 Header extension:
tests/qemu-iotests/036.out +2 -2
index 0b52b934e1..fb509f6357 100644
@@ -26,7 +26,7 @@ compatible_features       []
 autoclear_features        [63]
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 
@@ -38,7 +38,7 @@ compatible_features       []
 autoclear_features        []
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 *** done
tests/qemu-iotests/049.out +51 -51

225 lines not on this page · raw message

tests/qemu-iotests/060.out +1 -0
index d27692a33c..3e47f537e8 100644
@@ -17,6 +17,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     corrupt: true
tests/qemu-iotests/061.out +20 -14
index 8b3091a412..c913f02ad6 100644
@@ -22,11 +22,11 @@ incompatible_features     []
 compatible_features       [0]
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 magic                     0x514649fb
@@ -80,11 +80,11 @@ incompatible_features     []
 compatible_features       [0]
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 magic                     0x514649fb
@@ -136,11 +136,11 @@ incompatible_features     [0]
 compatible_features       [0]
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 ERROR cluster 5 refcount=0 reference=1
@@ -191,11 +191,11 @@ incompatible_features     []
 compatible_features       [42]
 autoclear_features        [42]
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 magic                     0x514649fb
@@ -260,11 +260,11 @@ incompatible_features     []
 compatible_features       [0]
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 read 65536/65536 bytes at offset 44040192
@@ -294,11 +294,11 @@ incompatible_features     [0]
 compatible_features       [0]
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 ERROR cluster 5 refcount=0 reference=1
@@ -323,11 +323,11 @@ incompatible_features     []
 compatible_features       []
 autoclear_features        []
 refcount_order            4
-header_length             104
+header_length             112
 
 Header extension:
 magic                     0x6803f857
-length                    192
+length                    240
 data                      <binary>
 
 read 131072/131072 bytes at offset 0
@@ -491,6 +491,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     data file: TEST_DIR/t.IMGFMT.data
@@ -511,6 +512,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     data file: foo
@@ -524,6 +526,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     data file raw: false
@@ -538,6 +541,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     data file: TEST_DIR/t.IMGFMT.data
@@ -550,6 +554,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     data file: TEST_DIR/t.IMGFMT.data
@@ -563,6 +568,7 @@ virtual size: 64 MiB (67108864 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     data file: TEST_DIR/t.IMGFMT.data
tests/qemu-iotests/065 +18 -10
index 6426474271..18dc488c7a 100755
@@ -88,24 +88,30 @@ class TestQMP(TestImageInfoSpecific):
 class TestQCow2(TestQemuImgInfo):
     '''Testing a qcow2 version 2 image'''
     img_options = 'compat=0.10'
-    json_compare = { 'compat': '0.10', 'refcount-bits': 16 }
-    human_compare = [ 'compat: 0.10', 'refcount bits: 16' ]
+    json_compare = { 'compat': '0.10', 'refcount-bits': 16,
+                     'compression-type': 'zlib' }
+    human_compare = [ 'compat: 0.10', 'compression type: zlib',
+                      'refcount bits: 16' ]
 
 class TestQCow3NotLazy(TestQemuImgInfo):
     '''Testing a qcow2 version 3 image with lazy refcounts disabled'''
     img_options = 'compat=1.1,lazy_refcounts=off'
     json_compare = { 'compat': '1.1', 'lazy-refcounts': False,
-                     'refcount-bits': 16, 'corrupt': False }
-    human_compare = [ 'compat: 1.1', 'lazy refcounts: false',
-                      'refcount bits: 16', 'corrupt: false' ]
+                     'refcount-bits': 16, 'corrupt': False,
+                     'compression-type': 'zlib' }
+    human_compare = [ 'compat: 1.1', 'compression type: zlib',
+                      'lazy refcounts: false', 'refcount bits: 16',
+                      'corrupt: false' ]
 
 class TestQCow3Lazy(TestQemuImgInfo):
     '''Testing a qcow2 version 3 image with lazy refcounts enabled'''
     img_options = 'compat=1.1,lazy_refcounts=on'
     json_compare = { 'compat': '1.1', 'lazy-refcounts': True,
-                     'refcount-bits': 16, 'corrupt': False }
-    human_compare = [ 'compat: 1.1', 'lazy refcounts: true',
-                      'refcount bits: 16', 'corrupt: false' ]
+                     'refcount-bits': 16, 'corrupt': False,
+                     'compression-type': 'zlib' }
+    human_compare = [ 'compat: 1.1', 'compression type: zlib',
+                      'lazy refcounts: true', 'refcount bits: 16',
+                      'corrupt: false' ]
 
 class TestQCow3NotLazyQMP(TestQMP):
     '''Testing a qcow2 version 3 image with lazy refcounts disabled, opening
@@ -113,7 +119,8 @@ class TestQCow3NotLazyQMP(TestQMP):
     img_options = 'compat=1.1,lazy_refcounts=off'
     qemu_options = 'lazy-refcounts=on'
     compare = { 'compat': '1.1', 'lazy-refcounts': False,
-                'refcount-bits': 16, 'corrupt': False }
+                'refcount-bits': 16, 'corrupt': False,
+                'compression-type': 'zlib' }
 
 
 class TestQCow3LazyQMP(TestQMP):
@@ -122,7 +129,8 @@ class TestQCow3LazyQMP(TestQMP):
     img_options = 'compat=1.1,lazy_refcounts=on'
     qemu_options = 'lazy-refcounts=off'
     compare = { 'compat': '1.1', 'lazy-refcounts': True,
-                'refcount-bits': 16, 'corrupt': False }
+                'refcount-bits': 16, 'corrupt': False,
+                'compression-type': 'zlib' }
 
 TestImageInfoSpecific = None
 TestQemuImgInfo = None
tests/qemu-iotests/080 +1 -1
index a3d13c414e..7588c63b6c 100755
@@ -45,7 +45,7 @@ _supported_os Linux
 # - This is generally a test for compat=1.1 images
 _unsupported_imgopts 'refcount_bits=1[^0-9]' data_file 'compat=0.10'
 
-header_size=104
+header_size=112
 
 offset_backing_file_offset=8
 offset_backing_file_size=16
tests/qemu-iotests/144.out +2 -2
index c7aa2e4820..885a8874a5 100644
@@ -9,7 +9,7 @@ Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=536870912
 { 'execute': 'qmp_capabilities' }
 {"return": {}}
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'virtio0', 'snapshot-file':'TEST_DIR/tmp.IMGFMT', 'format': 'IMGFMT' } }
-Formatting 'TEST_DIR/tmp.qcow2', fmt=qcow2 size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/tmp.qcow2', fmt=qcow2 size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 {"return": {}}
 
 === Performing block-commit on active layer ===
@@ -31,6 +31,6 @@ Formatting 'TEST_DIR/tmp.qcow2', fmt=qcow2 size=536870912 backing_file=TEST_DIR/
 === Performing Live Snapshot 2 ===
 
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'virtio0', 'snapshot-file':'TEST_DIR/tmp2.IMGFMT', 'format': 'IMGFMT' } }
-Formatting 'TEST_DIR/tmp2.qcow2', fmt=qcow2 size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/tmp2.qcow2', fmt=qcow2 size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 {"return": {}}
 *** done
tests/qemu-iotests/182.out +1 -1
index a8eea166c3..ae43654d32 100644
@@ -13,7 +13,7 @@ Is another process using the image [TEST_DIR/t.qcow2]?
 {'execute': 'blockdev-add', 'arguments': { 'node-name': 'node0', 'driver': 'file', 'filename': 'TEST_DIR/t.IMGFMT', 'locking': 'on' } }
 {"return": {}}
 {'execute': 'blockdev-snapshot-sync', 'arguments': { 'node-name': 'node0', 'snapshot-file': 'TEST_DIR/t.IMGFMT.overlay', 'snapshot-node-name': 'node1' } }
-Formatting 'TEST_DIR/t.qcow2.overlay', fmt=qcow2 size=197120 backing_file=TEST_DIR/t.qcow2 backing_fmt=file cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2.overlay', fmt=qcow2 size=197120 backing_file=TEST_DIR/t.qcow2 backing_fmt=file cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 {"return": {}}
 {'execute': 'blockdev-add', 'arguments': { 'node-name': 'node1', 'driver': 'file', 'filename': 'TEST_DIR/t.IMGFMT', 'locking': 'on' } }
 {"return": {}}
tests/qemu-iotests/242.out +5 -0
index 7ac8404d11..091b9126ce 100644
@@ -12,6 +12,7 @@ virtual size: 1 MiB (1048576 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
@@ -32,6 +33,7 @@ virtual size: 1 MiB (1048576 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     bitmaps:
         [0]:
@@ -64,6 +66,7 @@ virtual size: 1 MiB (1048576 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     bitmaps:
         [0]:
@@ -104,6 +107,7 @@ virtual size: 1 MiB (1048576 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     bitmaps:
         [0]:
@@ -153,6 +157,7 @@ virtual size: 1 MiB (1048576 bytes)
 cluster_size: 65536
 Format specific information:
     compat: 1.1
+    compression type: zlib
     lazy refcounts: false
     bitmaps:
         [0]:
tests/qemu-iotests/255.out +4 -4
index 348909fdef..a3c99fd62e 100644
@@ -3,9 +3,9 @@ Finishing a commit job with background reads
 
 === Create backing chain and start VM ===
 
-Formatting 'TEST_DIR/PID-t.qcow2.mid', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-t.qcow2.mid', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 
-Formatting 'TEST_DIR/PID-t.qcow2', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-t.qcow2', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 
 === Start background read requests ===
 
@@ -23,9 +23,9 @@ Closing the VM while a job is being cancelled
 
 === Create images and start VM ===
 
-Formatting 'TEST_DIR/PID-src.qcow2', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-src.qcow2', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 
-Formatting 'TEST_DIR/PID-dst.qcow2', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-dst.qcow2', fmt=qcow2 size=134217728 cluster_size=65536 lazy_refcounts=off refcount_bits=16 compression_type=zlib
 
 wrote 1048576/1048576 bytes at offset 0
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
tests/qemu-iotests/common.filter +2 -1
index 3f8ee3e5f7..279e0bbb0d 100644
@@ -152,7 +152,8 @@ _filter_img_create()
         -e "s# refcount_bits=[0-9]\\+##g" \
         -e "s# key-secret=[a-zA-Z0-9]\\+##g" \
         -e "s# iter-time=[0-9]\\+##g" \
-        -e "s# force_size=\\(on\\|off\\)##g"
+        -e "s# force_size=\\(on\\|off\\)##g" \
+        -e "s# compression_type=[a-zA-Z0-9]\\+##g"
 }
 
 _filter_img_info()
Signature · 1 line
2.17.0
Reviewed-by on v9 1/4
#2 lore raw

Re: [PATCH v9 1/4] qcow2: introduce compression type feature

Quoted 50 lines · Denis Plotnikov

The patch adds some preparation parts for incompatible compression type feature to qcow2 allowing the use different compression methods for image clusters (de)compressing.

It is implied that the compression type is set on the image creation and can be changed only later by image conversion, thus compression type defines the only compression algorithm used for the image, and thus, for all image clusters.

The goal of the feature is to add support of other compression methods to qcow2. For example, ZSTD which is more effective on compression than ZLIB.

The default compression is ZLIB. Images created with ZLIB compression type are backward compatible with older qemu versions.

Adding of the compression type breaks a number of tests because now the compression type is reported on image creation and there are some changes in the qcow2 header in size and offsets.

The tests are fixed in the following ways:

     * filter out compression_type for many tests
     * fix header size, feature table size and backing file offset
       affected tests: 031, 036, 061, 080
       header_size +=8: 1 byte compression type
                        7 bytes padding
       feature_table += 48: incompatible feature compression type
       backing_file_offset += 56 (8 + 48 -> header_change + feature_table_change)
     * add "compression type" for test output matching when it isn't filtered
       affected tests: 049, 060, 061, 065, 144, 182, 242, 255
---
  qapi/block-core.json             |  22 +++++-
  block/qcow2.h                    |  20 +++++-
  include/block/block_int.h        |   1 +
  block/qcow2.c                    | 113 +++++++++++++++++++++++++++++++
  tests/qemu-iotests/031.out       |  14 ++--
  tests/qemu-iotests/036.out       |   4 +-
  tests/qemu-iotests/049.out       | 102 ++++++++++++++--------------
  tests/qemu-iotests/060.out       |   1 +
  tests/qemu-iotests/061.out       |  34 ++++++----
  tests/qemu-iotests/065           |  28 +++++---
  tests/qemu-iotests/080           |   2 +-
  tests/qemu-iotests/144.out       |   4 +-
  tests/qemu-iotests/182.out       |   2 +-
  tests/qemu-iotests/242.out       |   5 ++
  tests/qemu-iotests/255.out       |   8 +--
  tests/qemu-iotests/common.filter |   3 +-
  16 files changed, 267 insertions(+), 96 deletions(-)

[..]

Quoted 5 lines
block/qcow2.h +0 -0
@@ -146,8 +146,16 @@ typedef struct QCowHeader {
 
      uint32_t refcount_order;

block/qcow2.h typedef struct QCowHeader { · line 157

Quoted 10 lines
block/qcow2.h +7 -0
      uint32_t header_length;
+
+    /* Additional fields */
+    uint8_t compression_type;
+
+    /* header must be a multiple of 8 */
+    uint8_t padding[7];
  } QEMU_PACKED QCowHeader;
 
+QEMU_BUILD_BUG_ON(sizeof(QCowHeader) % 8 != 0);

Better write

QEMU_BUILD_BUG_ON(QEMU_IS_ALIGNED(sizeof(QCowHeader), 8);

with or without this:

Quoted 36 lines
block/qcow2.h +12 -1
+
  typedef struct QEMU_PACKED QCowSnapshotHeader {
      /* header is 8 byte aligned */
      uint64_t l1_table_offset;
@@ -216,13 +224,16 @@ enum {
      QCOW2_INCOMPAT_DIRTY_BITNR      = 0,
      QCOW2_INCOMPAT_CORRUPT_BITNR    = 1,
      QCOW2_INCOMPAT_DATA_FILE_BITNR  = 2,
+    QCOW2_INCOMPAT_COMPRESSION_BITNR = 3,
      QCOW2_INCOMPAT_DIRTY            = 1 << QCOW2_INCOMPAT_DIRTY_BITNR,
      QCOW2_INCOMPAT_CORRUPT          = 1 << QCOW2_INCOMPAT_CORRUPT_BITNR,
      QCOW2_INCOMPAT_DATA_FILE        = 1 << QCOW2_INCOMPAT_DATA_FILE_BITNR,
+    QCOW2_INCOMPAT_COMPRESSION      = 1 << QCOW2_INCOMPAT_COMPRESSION_BITNR,
 
      QCOW2_INCOMPAT_MASK             = QCOW2_INCOMPAT_DIRTY
                                      | QCOW2_INCOMPAT_CORRUPT
-                                    | QCOW2_INCOMPAT_DATA_FILE,
+                                    | QCOW2_INCOMPAT_DATA_FILE
+                                    | QCOW2_INCOMPAT_COMPRESSION,
  };
 
  /* Compatible feature bits */
@@ -369,6 +380,13 @@ typedef struct BDRVQcow2State {
 
      bool metadata_preallocation_checked;
      bool metadata_preallocation;
+    /*
+     * Compression type used for the image. Default: 0 - ZLIB
+     * The image compression type is set on image creation.
+     * For now, the only way to change the compression type
+     * is to convert the image with the desired compression type set.
+     */
+    Qcow2CompressionType compression_type;
  } BDRVQcow2State;
 
  typedef struct Qcow2COWRegion {

[..]

Signature · 2 lines
Best regards,
Vladimir
Acked-by on v9 1/4
#3 lore raw

Re: [PATCH v9 1/4] qcow2: introduce compression type feature

In reply to Denis Plotnikov
Quoted 30 lines · Denis Plotnikov

The patch adds some preparation parts for incompatible compression type feature to qcow2 allowing the use different compression methods for image clusters (de)compressing.

It is implied that the compression type is set on the image creation and can be changed only later by image conversion, thus compression type defines the only compression algorithm used for the image, and thus, for all image clusters.

The goal of the feature is to add support of other compression methods to qcow2. For example, ZSTD which is more effective on compression than ZLIB.

The default compression is ZLIB. Images created with ZLIB compression type are backward compatible with older qemu versions.

Adding of the compression type breaks a number of tests because now the compression type is reported on image creation and there are some changes in the qcow2 header in size and offsets.

The tests are fixed in the following ways:

    * filter out compression_type for many tests
    * fix header size, feature table size and backing file offset
      affected tests: 031, 036, 061, 080
      header_size +=8: 1 byte compression type
                       7 bytes padding
      feature_table += 48: incompatible feature compression type
      backing_file_offset += 56 (8 + 48 -> header_change + feature_table_change)
    * add "compression type" for test output matching when it isn't filtered
      affected tests: 049, 060, 061, 065, 144, 182, 242, 255
Quoted 1 line

Hmm, haven't I've seen this before? ... Yes, I provided my QAPI part:
Acked-by for v5, and the QAPI part hasn't changed since. You can save me a bit of labor by carrying my Acked-by forward.

Once again, QAPI part:

#4 lore raw

Re: [PATCH v9 1/4] qcow2: introduce compression type feature

Quoted 32 lines · Markus Armbruster
Quoted 30 lines · Denis Plotnikov

The patch adds some preparation parts for incompatible compression type feature to qcow2 allowing the use different compression methods for image clusters (de)compressing.

It is implied that the compression type is set on the image creation and can be changed only later by image conversion, thus compression type defines the only compression algorithm used for the image, and thus, for all image clusters.

The goal of the feature is to add support of other compression methods to qcow2. For example, ZSTD which is more effective on compression than ZLIB.

The default compression is ZLIB. Images created with ZLIB compression type are backward compatible with older qemu versions.

Adding of the compression type breaks a number of tests because now the compression type is reported on image creation and there are some changes in the qcow2 header in size and offsets.

The tests are fixed in the following ways:

     * filter out compression_type for many tests
     * fix header size, feature table size and backing file offset
       affected tests: 031, 036, 061, 080
       header_size +=8: 1 byte compression type
                        7 bytes padding
       feature_table += 48: incompatible feature compression type
       backing_file_offset += 56 (8 + 48 -> header_change + feature_table_change)
     * add "compression type" for test output matching when it isn't filtered
       affected tests: 049, 060, 061, 065, 144, 182, 242, 255
Quoted 4 lines
Quoted 1 line

Hmm, haven't I've seen this before? ... Yes, I provided my QAPI part:
Acked-by for v5, and the QAPI part hasn't changed since. You can save me a bit of labor by carrying my Acked-by forward.

For sure, next time. Thanks!

Quoted 4 lines

Once again, QAPI part:

#5 lore raw

PATCH v9 2/4 qcow2: rework the cluster compression routine

In reply to Denis Plotnikov

The patch enables processing the image compression type defined for the image and chooses an appropriate method for image clusters (de)compression.

---
 block/qcow2-threads.c | 71 ++++++++++++++++++++++++++++++++++++-------
 1 file changed, 60 insertions(+), 11 deletions(-)
block/qcow2-threads.c +60 -11
index a68126f291..7dbaf53489 100644
@@ -74,7 +74,9 @@ typedef struct Qcow2CompressData {
 } Qcow2CompressData;
 
 /*
- * qcow2_compress()
+ * qcow2_zlib_compress()
+ *
+ * Compress @src_size bytes of data using zlib compression method
  *
  * @dest - destination buffer, @dest_size bytes
  * @src - source buffer, @src_size bytes
@@ -83,8 +85,8 @@ typedef struct Qcow2CompressData {
  *          -ENOMEM destination buffer is not enough to store compressed data
  *          -EIO    on any other error
  */
-static ssize_t qcow2_compress(void *dest, size_t dest_size,
-                              const void *src, size_t src_size)
+static ssize_t qcow2_zlib_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
 {
     ssize_t ret;
     z_stream strm;
@@ -119,10 +121,10 @@ static ssize_t qcow2_compress(void *dest, size_t dest_size,
 }
 
 /*
- * qcow2_decompress()
+ * qcow2_zlib_decompress()
  *
  * Decompress some data (not more than @src_size bytes) to produce exactly
- * @dest_size bytes.
+ * @dest_size bytes using zlib compression method
  *
  * @dest - destination buffer, @dest_size bytes
  * @src - source buffer, @src_size bytes
@@ -130,8 +132,8 @@ static ssize_t qcow2_compress(void *dest, size_t dest_size,
  * Returns: 0 on success
  *          -EIO on fail
  */
-static ssize_t qcow2_decompress(void *dest, size_t dest_size,
-                                const void *src, size_t src_size)
+static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size,
+                                     const void *src, size_t src_size)
 {
     int ret;
     z_stream strm;
@@ -191,20 +193,67 @@ qcow2_co_do_compress(BlockDriverState *bs, void *dest, size_t dest_size,
     return arg.ret;
 }
 
+/*
+ * qcow2_co_compress()
+ *
+ * Compress @src_size bytes of data using the compression
+ * method defined by the image compression type
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: compressed size on success
+ *          a negative error code on failure
+ */
 ssize_t coroutine_fn
 qcow2_co_compress(BlockDriverState *bs, void *dest, size_t dest_size,
                   const void *src, size_t src_size)
 {
-    return qcow2_co_do_compress(bs, dest, dest_size, src, src_size,
-                                qcow2_compress);
+    BDRVQcow2State *s = bs->opaque;
+    Qcow2CompressFunc fn;
+
+    switch (s->compression_type) {
+    case QCOW2_COMPRESSION_TYPE_ZLIB:
+        fn = qcow2_zlib_compress;
+        break;
+
+    default:
+        abort();
+    }
+
+    return qcow2_co_do_compress(bs, dest, dest_size, src, src_size, fn);
 }
 
+/*
+ * qcow2_co_decompress()
+ *
+ * Decompress some data (not more than @src_size bytes) to produce exactly
+ * @dest_size bytes using the compression method defined by the image
+ * compression type
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: 0 on success
+ *          a negative error code on failure
+ */
 ssize_t coroutine_fn
 qcow2_co_decompress(BlockDriverState *bs, void *dest, size_t dest_size,
                     const void *src, size_t src_size)
 {
-    return qcow2_co_do_compress(bs, dest, dest_size, src, src_size,
-                                qcow2_decompress);
+    BDRVQcow2State *s = bs->opaque;
+    Qcow2CompressFunc fn;
+
+    switch (s->compression_type) {
+    case QCOW2_COMPRESSION_TYPE_ZLIB:
+        fn = qcow2_zlib_decompress;
+        break;
+
+    default:
+        abort();
+    }
+
+    return qcow2_co_do_compress(bs, dest, dest_size, src, src_size, fn);
 }
 
 
Signature · 1 line
2.17.0
#6 lore raw

PATCH v9 3/4 qcow2: add zstd cluster compression

In reply to Denis Plotnikov

zstd significantly reduces cluster compression time. It provides better compression performance maintaining the same level of the compression ratio in comparison with zlib, which, at the moment, is the only compression method available.

The performance test results:
Test compresses and decompresses qemu qcow2 image with just installed rhel-7.6 guest.
Image cluster size: 64K. Image on disk size: 2.2G

The test was conducted with brd disk to reduce the influence of disk subsystem to the test results.
The results is given in seconds.

compress cmd:
time ./qemu-img convert -O qcow2 -c -o compression_type=[zlib|zstd]
src.img [zlib|zstd]_compressed.img
decompress cmd
time ./qemu-img convert -O qcow2
[zlib|zstd]_compressed.img uncompressed.img

compression decompression
zlib zstd zlib zstd
------------------------------------------------------------

real     65.5       16.3 (-75 %)    1.9          1.6 (-16 %)
user     65.0       15.8            5.3          2.5
sys       3.3        0.2            2.0          2.0

Both ZLIB and ZSTD gave the same compression ratio: 1.57 compressed image size in both cases: 1.4G

QAPI part:

---
 docs/interop/qcow2.txt |   1 +
 configure              |   2 +-
 qapi/block-core.json   |   3 +-
 block/qcow2-threads.c  | 134 +++++++++++++++++++++++++++++++++++++++++
 block/qcow2.c          |   7 +++
 5 files changed, 145 insertions(+), 2 deletions(-)
docs/interop/qcow2.txt +1 -0
index 5597e24474..795dbb21dd 100644
@@ -208,6 +208,7 @@ version 2.
 
                     Available compression type values:
                         0: zlib <https://www.zlib.net/>
+                        1: zstd <http://github.com/facebook/zstd>
 
 
 === Header padding ===
configure +1 -1
index caa65f5883..b2a0aa241a 100755
@@ -1835,7 +1835,7 @@ disabled with --disable-FEATURE, default is enabled if available:
   lzfse           support of lzfse compression library
                   (for reading lzfse-compressed dmg images)
   zstd            support for zstd compression library
-                  (for migration compression)
+                  (for migration compression and qcow2 cluster compression)
   seccomp         seccomp support
   coroutine-pool  coroutine freelist (better performance)
   glusterfs       GlusterFS backend
qapi/block-core.json +2 -1
index a306484973..8953451818 100644
@@ -4401,11 +4401,12 @@
 # Compression type used in qcow2 image file
 #
 # @zlib: zlib compression, see <http://zlib.net/>
+# @zstd: zstd compression, see <http://github.com/facebook/zstd>
 #
 # Since: 5.0
 ##
 { 'enum': 'Qcow2CompressionType',
-  'data': [ 'zlib' ] }
+  'data': [ 'zlib', { 'name': 'zstd', 'if': 'defined(CONFIG_ZSTD)' } ] }
 
 ##
 # @BlockdevCreateOptionsQcow2:
block/qcow2-threads.c +134 -0

158 lines not on this page · raw message

block/qcow2.c +7 -0
index eb30cfe811..de8e3f976d 100644
@@ -1246,6 +1246,9 @@ static int validate_compression_type(BDRVQcow2State *s, Error **errp)
 {
     switch (s->compression_type) {
     case QCOW2_COMPRESSION_TYPE_ZLIB:
+#ifdef CONFIG_ZSTD
+    case QCOW2_COMPRESSION_TYPE_ZSTD:
+#endif
         break;
 
     default:
@@ -3454,6 +3457,10 @@ qcow2_co_create(BlockdevCreateOptions *create_options, Error **errp)
         }
 
         switch (qcow2_opts->compression_type) {
+#ifdef CONFIG_ZSTD
+        case QCOW2_COMPRESSION_TYPE_ZSTD:
+            break;
+#endif
         default:
             error_setg(errp, "Unknown compression type");
             goto out;
Signature · 1 line
2.17.0
#7 lore raw

Re: [PATCH v9 3/4] qcow2: add zstd cluster compression

Should we note somehow in qcow2 spec that we use streamed version of zstd with specific end byte?

Quoted 36 lines · Denis Plotnikov

zstd significantly reduces cluster compression time. It provides better compression performance maintaining the same level of the compression ratio in comparison with zlib, which, at the moment, is the only compression method available.

The performance test results:
Test compresses and decompresses qemu qcow2 image with just installed rhel-7.6 guest.
Image cluster size: 64K. Image on disk size: 2.2G

The test was conducted with brd disk to reduce the influence of disk subsystem to the test results.
The results is given in seconds.

compress cmd:
time ./qemu-img convert -O qcow2 -c -o compression_type=[zlib|zstd]
src.img [zlib|zstd]_compressed.img
decompress cmd
time ./qemu-img convert -O qcow2
[zlib|zstd]_compressed.img uncompressed.img

compression decompression
zlib zstd zlib zstd
------------------------------------------------------------

real     65.5       16.3 (-75 %)    1.9          1.6 (-16 %)
user     65.0       15.8            5.3          2.5
sys       3.3        0.2            2.0          2.0

Both ZLIB and ZSTD gave the same compression ratio: 1.57 compressed image size in both cases: 1.4G

QAPI part:

---

[..]

Quoted 93 lines
block/qcow2-threads.c +93 -0
+
+/*
+ * qcow2_zstd_compress()
+ *
+ * Compress @src_size bytes of data using zstd compression method
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: compressed size on success
+ *          -ENOMEM destination buffer is not enough to store compressed data
+ *          -EIO    on any other error
+ */
+static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
+{
+    size_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };
+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * ZSTD spec: "You must continue calling ZSTD_compressStream2()
+     * with ZSTD_e_end until it returns 0, at which point you are
+     * free to start a new frame"
+     */
+    {
+        /*
+         * zstd simple interface requires the exact compressed size.
+         * zstd stream interface reads the comressed size from
+         * the compressed stream frame.
+         * Instruct zstd to compress the whole buffer and write
+         * the frame which includes the compressed size.
+         * This allows as to use zstd streaming semantics and
+         * don't store the compressed size for the zstd decompression.
+         */
+        ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);
+        if (ZSTD_isError(ret)) {
+            ret = -EIO;
+            goto out;
+        }
+        /* Dest buffer isn't big enough to store compressed content */
+        if (output.pos + ret > output.size) {
+            ret = -ENOMEM;
+            goto out;
+        }
+    } while (ret);
+
+    /* if no error, the input data must be fully consumed */
+    assert(input.pos == input.size);
+    /* make sure we can safely return compressed buffer size with ssize_t *//z
+    assert(output.pos <= SSIZE_MAX);
+    ret = output.pos;
+
+out:
+    ZSTD_freeCCtx(cctx);
+    return ret;
+}
+
+/*
+ * qcow2_zstd_decompress()
+ *
+ * Decompress some data (not more than @src_size bytes) to produce exactly
+ * @dest_size bytes using zstd compression method
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: 0 on success
+ *          -EIO on any error
+ */
+static ssize_t qcow2_zstd_decompress(void *dest, size_t dest_size,
+                                     const void *src, size_t src_size)
+{
+    size_t ret = 0;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };
+    ZSTD_DCtx *dctx = ZSTD_createDCtx();
+
+    if (!dctx) {
+        return -EIO;
+    }
+
+    {
+        ret = ZSTD_decompressStream(dctx, &output, &input);
+        if (ZSTD_isError(ret)) {
+            ret = -EIO;
+            goto out;
+        }
+        /*

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 277

Quoted 10 lines
block/qcow2-threads.c +10 -0
+         * Dest buffer size is the image cluster size.
+         * It should be big enough to store uncompressed content.
+         * There shouldn't be any cases when the decompressed content
+         * size is greater then the cluster size.
+         */
+        if (output.pos + ret > output.size) {
+            ret = -EIO;
+            goto out;
+        }
+    } while (ret);

Hmm. Unfortunately, zstd spec is not enough verbose to understand how to use these functions :).

But I found this in comment in https://github.com/facebook/zstd/blob/dev/examples/streaming_decompression.c :

/* The return code is zero if the frame is complete, but there may
* be multiple frames concatenated together.

so, I think it would be safer to move to "while (input.pos < input.size)" loop like in example.

and drop next assertion. And possibly do same for compression to be consistent (and safer?).

Quoted 19 lines
block/qcow2-threads.c +15 -0
+
+    /*
+     * If decompression went fine we must have the compressed
+     * cluster fully consumed and flushed
+     */
+    if (output.pos != output.size) {
+        ret = -EIO;
+    }
+out:
+    ZSTD_freeDCtx(dctx);
+    return ret;
+
+}
+#endif
+
  static int qcow2_compress_pool_func(void *opaque)
  {
      Qcow2CompressData *data = opaque;
@@ -217,6 +341,11 @@ qcow2_co_compress(BlockDriverState *bs, void *dest, size_t dest_size,

[..]

Signature · 2 lines
Best regards,
Vladimir
#8 lore raw

Re: [PATCH v9 3/4] qcow2: add zstd cluster compression

Quoted 1 line · Vladimir Sementsov-Ogievskiy

Should we note somehow in qcow2 spec that we use streamed version of zstd with specific end byte?

We didn't do it for zlib. zstd does it the same way as zlib, saves the compression output to some buffer.

Quoted 152 lines
Quoted 36 lines · Denis Plotnikov

zstd significantly reduces cluster compression time. It provides better compression performance maintaining the same level of the compression ratio in comparison with zlib, which, at the moment, is the only compression method available.

The performance test results:
Test compresses and decompresses qemu qcow2 image with just installed rhel-7.6 guest.
Image cluster size: 64K. Image on disk size: 2.2G

The test was conducted with brd disk to reduce the influence of disk subsystem to the test results.
The results is given in seconds.

compress cmd:
   time ./qemu-img convert -O qcow2 -c -o compression_type=[zlib|zstd] src.img [zlib|zstd]_compressed.img decompress cmd
   time ./qemu-img convert -O qcow2 [zlib|zstd]_compressed.img uncompressed.img

            compression               decompression zlib       zstd           zlib         zstd
------------------------------------------------------------
real     65.5       16.3 (-75 %)    1.9          1.6 (-16 %) user     65.0       15.8            5.3          2.5 sys       3.3        0.2            2.0          2.0

Both ZLIB and ZSTD gave the same compression ratio: 1.57 compressed image size in both cases: 1.4G

QAPI part:

---

[..]

Quoted 96 lines
block/qcow2-threads.c +11 -0
+
+/*
+ * qcow2_zstd_compress()
+ *
+ * Compress @src_size bytes of data using zstd compression method
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: compressed size on success
+ *          -ENOMEM destination buffer is not enough to store 

compressed data
+ *          -EIO    on any other error
+ */

block/qcow2-threads.c +41 -0
+static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
+{
+    size_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };
+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * ZSTD spec: "You must continue calling ZSTD_compressStream2()
+     * with ZSTD_e_end until it returns 0, at which point you are
+     * free to start a new frame"
+     */
+    {
+        /*
+         * zstd simple interface requires the exact compressed size.
+         * zstd stream interface reads the comressed size from
+         * the compressed stream frame.
+         * Instruct zstd to compress the whole buffer and write
+         * the frame which includes the compressed size.
+         * This allows as to use zstd streaming semantics and
+         * don't store the compressed size for the zstd decompression.
+         */
+        ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);
+        if (ZSTD_isError(ret)) {
+            ret = -EIO;
+            goto out;
+        }
+        /* Dest buffer isn't big enough to store compressed content */
+        if (output.pos + ret > output.size) {
+            ret = -ENOMEM;
+            goto out;
+        }
+    } while (ret);
+
+    /* if no error, the input data must be fully consumed */
+    assert(input.pos == input.size);
+    /* make sure we can safely return compressed buffer size with 

ssize_t *//z
+    assert(output.pos <= SSIZE_MAX);
+    ret = output.pos;

block/qcow2-threads.c +10 -0
+
+out:
+    ZSTD_freeCCtx(cctx);
+    return ret;
+}
+
+/*
+ * qcow2_zstd_decompress()
+ *
+ * Decompress some data (not more than @src_size bytes) to produce 

exactly
+ * @dest_size bytes using zstd compression method
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: 0 on success
+ *          -EIO on any error
+ */

block/qcow2-threads.c +19 -0
+static ssize_t qcow2_zstd_decompress(void *dest, size_t dest_size,
+                                     const void *src, size_t src_size)
+{
+    size_t ret = 0;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };
+    ZSTD_DCtx *dctx = ZSTD_createDCtx();
+
+    if (!dctx) {
+        return -EIO;
+    }
+
+    {
+        ret = ZSTD_decompressStream(dctx, &output, &input);
+        if (ZSTD_isError(ret)) {
+            ret = -EIO;
+            goto out;
+        }
+        /*

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 277

Quoted 10 lines
block/qcow2-threads.c +10 -0
+         * Dest buffer size is the image cluster size.
+         * It should be big enough to store uncompressed content.
+         * There shouldn't be any cases when the decompressed content
+         * size is greater then the cluster size.
+         */
+        if (output.pos + ret > output.size) {
+            ret = -EIO;
+            goto out;
+        }
+    } while (ret);

Hmm. Unfortunately, zstd spec is not enough verbose to understand how to use these functions :).

Quoted 4 lines

But I found this in comment in https://github.com/facebook/zstd/blob/dev/examples/streaming_decompression.c :

/* The return code is zero if the frame is complete, but there may
             * be multiple frames concatenated together.

In our case, there should be the only frame as we stored it with ZSTD_compressStream2(..., ZSTD_e_end)
If explicitly state in the spec, that "zstd compressed cluster must contain the only zstd compressed frame"
On the other hand, we actually don't need such a restriction and we should be able to decompress whatever is in the buffer (cluster) if it's zstd stream.

Quoted 2 lines

so, I think it would be safer to move to "while (input.pos < input.size)" loop like in example.

yes, I think It worth doing

Quoted 29 lines

and drop next assertion. And possibly do same for compression to be consistent (and safer?).

Quoted 20 lines
block/qcow2-threads.c +15 -0
+
+    /*
+     * If decompression went fine we must have the compressed
+     * cluster fully consumed and flushed
+     */
+    if (output.pos != output.size) {
+        ret = -EIO;
+    }
+out:
+    ZSTD_freeDCtx(dctx);
+    return ret;
+
+}
+#endif
+

  static int qcow2_compress_pool_func(void *opaque) {
      Qcow2CompressData *data = opaque;

(quoted diff) +0 -0
@@ -217,6 +341,11 @@ qcow2_co_compress(BlockDriverState *bs, void 
 *dest, size_t dest_size,

[..]

#9 lore raw

Re: [PATCH v9 3/4] qcow2: add zstd cluster compression

Quoted 5 lines · Denis Plotnikov
Quoted 1 line · Vladimir Sementsov-Ogievskiy

Should we note somehow in qcow2 spec that we use streamed version of zstd with specific end byte?

We didn't do it for zlib. zstd does it the same way as zlib, saves the compression output to some buffer.

Looked through zstd format spec, I don't thing that there are may be any principal difference between data compressed by simple api or by streamed, so now I don't think that we need any additional comment.

Quoted 191 lines
Quoted 150 lines
Quoted 36 lines · Denis Plotnikov

zstd significantly reduces cluster compression time. It provides better compression performance maintaining the same level of the compression ratio in comparison with zlib, which, at the moment, is the only compression method available.

The performance test results:
Test compresses and decompresses qemu qcow2 image with just installed rhel-7.6 guest.
Image cluster size: 64K. Image on disk size: 2.2G

The test was conducted with brd disk to reduce the influence of disk subsystem to the test results.
The results is given in seconds.

compress cmd:
   time ./qemu-img convert -O qcow2 -c -o compression_type=[zlib|zstd] src.img [zlib|zstd]_compressed.img decompress cmd
   time ./qemu-img convert -O qcow2 [zlib|zstd]_compressed.img uncompressed.img

            compression               decompression zlib       zstd           zlib         zstd
------------------------------------------------------------
real     65.5       16.3 (-75 %)    1.9          1.6 (-16 %) user     65.0       15.8            5.3          2.5 sys       3.3        0.2            2.0          2.0

Both ZLIB and ZSTD gave the same compression ratio: 1.57 compressed image size in both cases: 1.4G

QAPI part:

---

[..]

Quoted 93 lines
block/qcow2-threads.c +93 -0
+
+/*
+ * qcow2_zstd_compress()
+ *
+ * Compress @src_size bytes of data using zstd compression method
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: compressed size on success
+ *          -ENOMEM destination buffer is not enough to store compressed data
+ *          -EIO    on any other error
+ */
+static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
+{
+    size_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };
+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * ZSTD spec: "You must continue calling ZSTD_compressStream2()
+     * with ZSTD_e_end until it returns 0, at which point you are
+     * free to start a new frame"
+     */
+    {
+        /*
+         * zstd simple interface requires the exact compressed size.
+         * zstd stream interface reads the comressed size from
+         * the compressed stream frame.
+         * Instruct zstd to compress the whole buffer and write
+         * the frame which includes the compressed size.
+         * This allows as to use zstd streaming semantics and
+         * don't store the compressed size for the zstd decompression.
+         */
+        ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);
+        if (ZSTD_isError(ret)) {
+            ret = -EIO;
+            goto out;
+        }
+        /* Dest buffer isn't big enough to store compressed content */
+        if (output.pos + ret > output.size) {
+            ret = -ENOMEM;
+            goto out;
+        }
+    } while (ret);
+
+    /* if no error, the input data must be fully consumed */
+    assert(input.pos == input.size);
+    /* make sure we can safely return compressed buffer size with ssize_t *//z
+    assert(output.pos <= SSIZE_MAX);
+    ret = output.pos;
+
+out:
+    ZSTD_freeCCtx(cctx);
+    return ret;
+}
+
+/*
+ * qcow2_zstd_decompress()
+ *
+ * Decompress some data (not more than @src_size bytes) to produce exactly
+ * @dest_size bytes using zstd compression method
+ *
+ * @dest - destination buffer, @dest_size bytes
+ * @src - source buffer, @src_size bytes
+ *
+ * Returns: 0 on success
+ *          -EIO on any error
+ */
+static ssize_t qcow2_zstd_decompress(void *dest, size_t dest_size,
+                                     const void *src, size_t src_size)
+{
+    size_t ret = 0;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };
+    ZSTD_DCtx *dctx = ZSTD_createDCtx();
+
+    if (!dctx) {
+        return -EIO;
+    }
+
+    {
+        ret = ZSTD_decompressStream(dctx, &output, &input);
+        if (ZSTD_isError(ret)) {
+            ret = -EIO;
+            goto out;
+        }
+        /*

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 277

Quoted 10 lines
block/qcow2-threads.c +10 -0
+         * Dest buffer size is the image cluster size.
+         * It should be big enough to store uncompressed content.
+         * There shouldn't be any cases when the decompressed content
+         * size is greater then the cluster size.
+         */
+        if (output.pos + ret > output.size) {
+            ret = -EIO;
+            goto out;
+        }
+    } while (ret);

Hmm. Unfortunately, zstd spec is not enough verbose to understand how to use these functions :).

Quoted 4 lines

But I found this in comment in https://github.com/facebook/zstd/blob/dev/examples/streaming_decompression.c :

/* The return code is zero if the frame is complete, but there may
             * be multiple frames concatenated together.

In our case, there should be the only frame as we stored it with ZSTD_compressStream2(..., ZSTD_e_end)
If explicitly state in the spec, that "zstd compressed cluster must contain the only zstd compressed frame"
On the other hand, we actually don't need such a restriction and we should be able to decompress whatever is in the buffer (cluster) if it's zstd stream.

Quoted 2 lines

so, I think it would be safer to move to "while (input.pos < input.size)" loop like in example.

yes, I think It worth doing

Quoted 28 lines

and drop next assertion. And possibly do same for compression to be consistent (and safer?).

Quoted 19 lines
block/qcow2-threads.c +15 -0
+
+    /*
+     * If decompression went fine we must have the compressed
+     * cluster fully consumed and flushed
+     */
+    if (output.pos != output.size) {
+        ret = -EIO;
+    }
+out:
+    ZSTD_freeDCtx(dctx);
+    return ret;
+
+}
+#endif
+

  static int qcow2_compress_pool_func(void *opaque) {
      Qcow2CompressData *data = opaque;

(quoted diff) +0 -0
@@ -217,6 +341,11 @@ qcow2_co_compress(BlockDriverState *bs, void *dest, size_t dest_size,

[..]

Signature · 2 lines
Best regards,
Vladimir
#10 lore raw

PATCH v9 4/4 iotests: 287: add qcow2 compression type test

In reply to Denis Plotnikov

The test checks fulfilling qcow2 requiriements for the compression type feature and zstd compression type operability.

---
 tests/qemu-iotests/287     | 128 +++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/287.out |  43 +++++++++++++
 tests/qemu-iotests/group   |   1 +
 3 files changed, 172 insertions(+)
 create mode 100755 tests/qemu-iotests/287
 create mode 100644 tests/qemu-iotests/287.out
tests/qemu-iotests/287 new file +128 -0
new file mode 100755
index 0000000000..49d15b3d43
@@ -0,0 +1,128 @@
+#!/usr/bin/env bash
+#
+# Test case for an image using zstd compression
+#
+# Copyright (c) 2020 Virtuozzo International GmbH
+#
+# 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
[email protected]
+
+seq="$(basename $0)"
+echo "QA output created by $seq"
+
+status=1	# failure is the default!
+
+_cleanup()
+{
+	_cleanup_test_img
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# standard environment
+. ./common.rc
+. ./common.filter
+
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux
+
+# for all the cases
+CLUSTER_SIZE=65536
+
+# Check if we can run this test.
+
+IMGOPTS='compression_type=zstd' _make_test_img 64M | grep "Invalid parameter 'zstd'" 2>&1 1>/dev/null
+
+ZSTD_SUPPORTED=$?
+
+if (($ZSTD_SUPPORTED==0)); then
+    _notrun "ZSTD is disabled"
+fi
+
+# Test: when compression is zlib the incompatible bit is unset
+echo
+echo "=== Testing compression type incompatible bit setting for zlib ==="
+echo
+
+IMGOPTS='compression_type=zlib' _make_test_img 64M
+$PYTHON qcow2.py "$TEST_IMG" dump-header | grep incompatible_features
+
+# Test: when compression differs from zlib the incompatible bit is set
+echo
+echo "=== Testing compression type incompatible bit setting for zstd ==="
+echo
+
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+$PYTHON qcow2.py "$TEST_IMG" dump-header | grep incompatible_features
+
+# Test: an image can't be openned if compression type is zlib and
+#       incompatible feature compression type is set
+echo
+echo "=== Testing zlib with incompatible bit set  ==="
+echo
+
+IMGOPTS='compression_type=zlib' _make_test_img 64M
+$PYTHON qcow2.py "$TEST_IMG" set-feature-bit incompatible 3
+# to make sure the bit was actually set
+$PYTHON qcow2.py "$TEST_IMG" dump-header | grep incompatible_features
+$QEMU_IMG info "$TEST_IMG" 2>1 1>/dev/null
+if (($?==0)); then
+    echo "Error: The image openned successfully. The image must not be openned"
+fi
+
+# Test: an image can't be openned if compression type is NOT zlib and
+#       incompatible feature compression type is UNSET
+echo
+echo "=== Testing zstd with incompatible bit unset  ==="
+echo
+
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+$PYTHON qcow2.py "$TEST_IMG" set-header incompatible_features 0
+# to make sure the bit was actually unset
+$PYTHON qcow2.py "$TEST_IMG" dump-header | grep incompatible_features
+$QEMU_IMG info "$TEST_IMG" 2>1 1>/dev/null
+if (($?==0)); then
+    echo "Error: The image openned successfully. The image must not be openned"
+fi
+# Test: check compression type values
+echo
+echo "=== Testing compression type values  ==="
+echo
+# zlib=0
+IMGOPTS='compression_type=zlib' _make_test_img 64M
+od -j104 -N1 -An -vtu1 "$TEST_IMG"
+
+# zstd=1
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+od -j104 -N1 -An -vtu1 "$TEST_IMG"
+
+# Test: using zstd compression, write to and read from an image
+echo
+echo "=== Testing reading and writing with zstd ==="
+echo
+
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+$QEMU_IO -c "write -c -P 0xAC 65536 64k " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -P 0xAC 65536 65536 " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -v 131070 8 " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -v 65534 8" "$TEST_IMG" | _filter_qemu_io
+
+# success, all done
+echo "*** done"
+rm -f $seq.full
+status=0
tests/qemu-iotests/287.out new file +43 -0

43 lines not on this page · raw message

tests/qemu-iotests/group +1 -0
index 0317667695..5edbadef40 100644
@@ -293,3 +293,4 @@
 283 auto quick
 284 rw
 286 rw quick
+287 auto quick
Signature · 1 line
2.17.0