Skip to contentSkip to conversation

qcow2: Implement zstd cluster compression methodi

PATCH Superseded Denis Plotnikov 5 participants 19 messages Apr 21, 2020 – Apr 28, 2020 Block layer

v20 of 24 Superseded Latest is v24 →

lore
Share & tools
lore
#0 lore raw

v20 cover qcow2: Implement zstd cluster compression methodi

v20:
04: fix a number of flaws [Vladimir]
* don't use $RAND_FILE passing to qemu-io,
so check $TEST_DIR is redundant
* re-arrage $RAND_FILE writing
* fix a typo

v19:
04: fix a number of flaws [Eric]
* remove rudundant test case descriptions
* fix stdout redirect
* don't use (())
* use peek_file_be instead of od
* check $TEST_DIR for spaces and other before using
* use $RAND_FILE safer

v18:
* 04: add quotes to all file name variables [Vladimir]
* 04: add Vladimir's comment according to "qemu-io write -s" option issue.

v17:
* 03: remove incorrect comment in zstd decompress [Vladimir]
* 03: remove "paraniod" and rewrite the comment on decompress [Vladimir]
* 03: fix dead assignment [Vladimir]
* 04: add and remove quotes [Vladimir]
* 04: replace long offset form with the short one [Vladimir]

v16:
* 03: ssize_t for ret, size_t for zstd_ret [Vladimir]
* 04: small fixes according to the comments [Vladimir]

v15:
* 01: aiming qemu 5.1 [Eric]
* 03: change zstd_res definition place [Vladimir]
* 04: add two new test cases [Eric]
1. test adjacent cluster compression with zstd
2. test incompressible cluster processing
* 03, 04: many rewording and gramma fixing [Eric]

v14:
* fix bug on compression - looping until compress == 0 [Me]
* apply reworked Vladimir's suggestions:
1. not mixing ssize_t with size_t
2. safe check for ENOMEM in compression part - avoid overflow
3. tolerate sanity check allow zstd to make progress only on one of the buffers
v13:
* 03: add progress sanity check to decompression loop [Vladimir] 03: add successful decompression check [Me]

v12:
* 03: again, rework compression and decompression loops
to make them more correct [Vladimir]
03: move assert in compression to more appropriate place
[Vladimir]
v11:
* 03: the loops don't need "do{}while" form anymore and
the they were buggy (missed "do" in the beginning)
replace them with usual "while(){}" loops [Vladimir]
v10:
* 03: fix zstd (de)compressed loops for multi-frame
cases [Vladimir]
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            | 228 +++++++++++++++++++++++++++++--
 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           | 146 ++++++++++++++++++++
 tests/qemu-iotests/287.out       |  67 +++++++++
 tests/qemu-iotests/common.filter |   3 +-
 tests/qemu-iotests/group         |   1 +
 22 files changed, 708 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 v20 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 part:

---
 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 943df1926a..1522e2983f 100644
@@ -78,6 +78,8 @@
 #
 # @bitmaps: A list of qcow2 bitmap details (since 4.0)
 #
+# @compression-type: the image cluster compression method (since 5.1)
+#
 # Since: 1.7
 ##
 { 'struct': 'ImageInfoSpecificQCow2',
@@ -89,7 +91,8 @@
       '*corrupt': 'bool',
       'refcount-bits': 'int',
       '*encrypt': 'ImageInfoSpecificQCow2Encryption',
-      '*bitmaps': ['Qcow2BitmapInfo']
+      '*bitmaps': ['Qcow2BitmapInfo'],
+      'compression-type': 'Qcow2CompressionType'
   } }
 
 ##
@@ -4284,6 +4287,18 @@
   'data': [ 'v2', 'v3' ] }
 
 
+##
+# @Qcow2CompressionType:
+#
+# Compression type used in qcow2 image file
+#
+# @zlib: zlib compression, see <http://zlib.net/>
+#
+# Since: 5.1
+##
+{ 'enum': 'Qcow2CompressionType',
+  'data': [ 'zlib' ] }
+
 ##
 # @BlockdevCreateOptionsQcow2:
 #
@@ -4307,6 +4322,8 @@
 #                 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 5.1)
 #
 # Since: 2.12
 ##
@@ -4322,7 +4339,8 @@
             '*cluster-size':    'size',
             '*preallocation':   'PreallocMode',
             '*lazy-refcounts':  'bool',
-            '*refcount-bits':   'int' } }
+            '*refcount-bits':   'int',
+            '*compression-type':'Qcow2CompressionType' } }
 
 ##
 # @BlockdevCreateOptionsQed:
block/qcow2.h +19 -1
index f4de0a27d5..6a8b82e6cc 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(!QEMU_IS_ALIGNED(sizeof(QCowHeader), 8));
+
 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 */
@@ -366,6 +377,13 @@ 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 4c3587ea19..bc91b69f46 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 46f97c5a4e..5a4beda6a2 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                    288
+length                    336
 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                    288
+length                    336
 data                      <binary>
 
 Header extension:
@@ -164,7 +164,7 @@ No errors were found on the image.
 
 magic                     0x514649fb
 version                   3
-backing_file_offset       0x1d8
+backing_file_offset       0x210
 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                    288
+length                    336
 data                      <binary>
 
 Header extension:
tests/qemu-iotests/036.out +2 -2
index 23b699ce06..e409acf60e 100644
@@ -26,7 +26,7 @@ compatible_features       []
 autoclear_features        [63]
 Header extension:
 magic                     0x6803f857
-length                    288
+length                    336
 data                      <binary>
 
 
@@ -38,7 +38,7 @@ compatible_features       []
 autoclear_features        []
 Header extension:
 magic                     0x6803f857
-length                    288
+length                    336
 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 09caaea865..be5f8707a3 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 413cc4e0f4..45c029228c 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                    288
+length                    336
 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                    288
+length                    336
 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                    288
+length                    336
 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                    288
+length                    336
 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                    288
+length                    336
 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                    288
+length                    336
 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                    288
+length                    336
 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 v20 1/4
#2 lore raw

Re: [PATCH v20 1/4] qcow2: introduce compression type feature

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 5 lines

QAPI part:

Berto

Reviewed-by on v20 1/4
#3 lore raw

Re: [PATCH v20 1/4] qcow2: introduce compression type feature

In reply to Denis Plotnikov
Quoted 45 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 part:

---
 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 ++++++----
Quoted 8 lines
 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(-)
#4 lore raw

PATCH v20 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
Reviewed-by on v20 2/4
#5 lore raw

Re: [PATCH v20 2/4] qcow2: rework the cluster compression routine

Quoted 10 lines · 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(-)
#6 lore raw

PATCH v20 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  | 157 +++++++++++++++++++++++++++++++++++++++++
 block/qcow2.c          |   7 ++
 5 files changed, 168 insertions(+), 2 deletions(-)
docs/interop/qcow2.txt +1 -0
index 640e0eca40..18a77f737e 100644
@@ -209,6 +209,7 @@ 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 23b5e93752..4e3a1690ea 100755
@@ -1861,7 +1861,7 @@ 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 1522e2983f..6fbacddab2 100644
@@ -4293,11 +4293,12 @@
 # Compression type used in qcow2 image file
 #
 # @zlib: zlib compression, see <http://zlib.net/>
+# @zstd: zstd compression, see <http://github.com/facebook/zstd>
 #
 # Since: 5.1
 ##
 { 'enum': 'Qcow2CompressionType',
-  'data': [ 'zlib' ] }
+  'data': [ 'zlib', { 'name': 'zstd', 'if': 'defined(CONFIG_ZSTD)' } ] }
 
 ##
 # @BlockdevCreateOptionsQcow2:
block/qcow2-threads.c +157 -0

181 lines not on this page · raw message

block/qcow2.c +7 -0
index 9c0b20c912..21f281b7f8 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:
@@ -3478,6 +3481,10 @@ 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 v20 3/4] qcow2: add zstd cluster compression

Quoted 44 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:

---
 docs/interop/qcow2.txt |   1 +
 configure              |   2 +-
 qapi/block-core.json   |   3 +-
 block/qcow2-threads.c  | 157 +++++++++++++++++++++++++++++++++++++++++
 block/qcow2.c          |   7 ++
 5 files changed, 168 insertions(+), 2 deletions(-)

[...]

Quoted 30 lines
block/qcow2-threads.c +15 -0
index 7dbaf53489..0525718704 100644
@@ -28,6 +28,11 @@
 #define ZLIB_CONST
 #include <zlib.h>
 
+#ifdef CONFIG_ZSTD
+#include <zstd.h>
+#include <zstd_errors.h>
+#endif
+
 #include "qcow2.h"
 #include "block/thread-pool.h"
 #include "crypto.h"
@@ -166,6 +171,148 @@ static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size,
     return ret;
 }
 
+#ifdef CONFIG_ZSTD
+
+/*
+ * 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
+ *

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 193

Quoted 10 lines
block/qcow2-threads.c +10 -0
+ * 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)
+{
+    ssize_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };

Minor style note: I think it’d be nicer to use designated initializers here.

Quoted 16 lines
block/qcow2-threads.c +16 -0
+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * Use the zstd streamed interface for symmetry with decompression,
+     * where streaming is essential since we don't record the exact
+     * compressed size.
+     *
+     * In the loop, we try to compress all the data into one zstd frame.
+     * ZSTD_compressStream2 potentially can finish a frame earlier
+     * than the full input data is consumed. That's why we are looping
+     * until all the input data is consumed.
+     */
+    while (input.pos < input.size) {

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 219

Quoted 10 lines
block/qcow2-threads.c +10 -0
+        size_t zstd_ret;
+        /*
+         * 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". We assume that "start a new frame"
+         * means call ZSTD_compressStream2 in the very beginning or when
+         * ZSTD_compressStream2 has returned with 0.
+         */
+        do {
+            zstd_ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);

The spec makes it sound to me like ZSTD_e_end will always complete in a single call if there’s enough space in the output buffer. So the only team we have to loop would be when there isn’t enough space anyway:

It says this about ZSTD_e_end:

Quoted 2 lines

flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.

Those rules being:

Quoted 2 lines

Note that, if output->size is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).

So it seems like it will only return a value > 0 if the output buffer is definitely too small.

The spec also notes that the return value is greater than 0 if:

Quoted 2 lines
Quoted 1 line

0 if some data still present within internal buffer (the value is

minimal estimation of remaining size),

So it’s a minimum estimate. That’s another point that heavily implies to me that if the return value were less than what’s left in the buffer, the function wouldn’t return but still try to write it out, until it realizes that there isn’t enough space in the output buffer, and then return a value that exceeds the remaining output buffer size.

(Because if the function just played it safe, I would expect it to return a maximum estimate.)

OTOH, if it were actually possible for ZSTD_e_end to finish a frame earlier than the end of the input, I think it would make more sense to use ZSTD_e_continue until the input is done and then finish with ZSTD_e_end, like the spec seems to propose. That way, we’d always end up with a single frame to make decompression simpler (and I think it would also make more sense overall).

But anyway. From how I understand the spec, this code simply always ends up creating a single frame or erroring out, without looping ever. So it isn’t exactly wrong, it just seems overly complicated. (Again, assuming I understand the spec correctly. Which seems like a tough thing to assume, because the spec is not exactly obvious to read...)

(Running some quick tests by converting some images with zstd compression seems to confirm that whenever ZSTD_compressStream2() returns, either zstd_ret > output.size - output.pos, or zstd_ret == 0 and input.pos == input.size. So none of the loops ever loop.)

Max

Quoted 19 lines
block/qcow2-threads.c +19 -0
+
+            if (ZSTD_isError(zstd_ret)) {
+                ret = -EIO;
+                goto out;
+            }
+            /* Dest buffer isn't big enough to store compressed content */
+            if (zstd_ret > output.size - output.pos) {
+                ret = -ENOMEM;
+                goto out;
+            }
+        } while (zstd_ret);
+    }
+    /* make sure we can safely return compressed buffer size with ssize_t */
+    assert(output.pos <= SSIZE_MAX);
+    ret = output.pos;
+out:
+    ZSTD_freeCCtx(cctx);
+    return ret;
+}
#8 lore raw

Re: [PATCH v20 3/4] qcow2: add zstd cluster compression

Quoted 155 lines · Max Reitz
Quoted 44 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:

---
  docs/interop/qcow2.txt |   1 +
  configure              |   2 +-
  qapi/block-core.json   |   3 +-
  block/qcow2-threads.c  | 157 +++++++++++++++++++++++++++++++++++++++++
  block/qcow2.c          |   7 ++
  5 files changed, 168 insertions(+), 2 deletions(-)

[...]

Quoted 30 lines
block/qcow2-threads.c +15 -0
index 7dbaf53489..0525718704 100644
@@ -28,6 +28,11 @@
  #define ZLIB_CONST
  #include <zlib.h>
 
+#ifdef CONFIG_ZSTD
+#include <zstd.h>
+#include <zstd_errors.h>
+#endif
+
  #include "qcow2.h"
  #include "block/thread-pool.h"
  #include "crypto.h"
@@ -166,6 +171,148 @@ static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size,
      return ret;
  }
 
+#ifdef CONFIG_ZSTD
+
+/*
+ * 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
+ *

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 193

Quoted 10 lines
block/qcow2-threads.c +10 -0
+ * 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)
+{
+    ssize_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };

Minor style note: I think it’d be nicer to use designated initializers here.

Quoted 16 lines
block/qcow2-threads.c +16 -0
+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * Use the zstd streamed interface for symmetry with decompression,
+     * where streaming is essential since we don't record the exact
+     * compressed size.
+     *
+     * In the loop, we try to compress all the data into one zstd frame.
+     * ZSTD_compressStream2 potentially can finish a frame earlier
+     * than the full input data is consumed. That's why we are looping
+     * until all the input data is consumed.
+     */
+    while (input.pos < input.size) {

block/qcow2-threads.c static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 219

Quoted 10 lines
block/qcow2-threads.c +10 -0
+        size_t zstd_ret;
+        /*
+         * 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". We assume that "start a new frame"
+         * means call ZSTD_compressStream2 in the very beginning or when
+         * ZSTD_compressStream2 has returned with 0.
+         */
+        do {
+            zstd_ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);

The spec makes it sound to me like ZSTD_e_end will always complete in a single call if there’s enough space in the output buffer. So the only team we have to loop would be when there isn’t enough space anyway:

It says this about ZSTD_e_end:

Quoted 2 lines

flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.

Those rules being:

Quoted 2 lines

Note that, if output->size is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).

So it seems like it will only return a value > 0 if the output buffer is definitely too small.

The spec also notes that the return value is greater than 0 if:

Quoted 2 lines
Quoted 1 line

0 if some data still present within internal buffer (the value is

minimal estimation of remaining size),

So it’s a minimum estimate. That’s another point that heavily implies to me that if the return value were less than what’s left in the buffer, the function wouldn’t return but still try to write it out, until it realizes that there isn’t enough space in the output buffer, and then return a value that exceeds the remaining output buffer size.

(Because if the function just played it safe, I would expect it to return a maximum estimate.)

OTOH, if it were actually possible for ZSTD_e_end to finish a frame earlier than the end of the input, I think it would make more sense to use ZSTD_e_continue until the input is done and then finish with ZSTD_e_end, like the spec seems to propose. That way, we’d always end up with a single frame to make decompression simpler (and I think it would also make more sense overall).

But anyway. From how I understand the spec, this code simply always ends up creating a single frame or erroring out, without looping ever. So it isn’t exactly wrong, it just seems overly complicated. (Again, assuming I understand the spec correctly. Which seems like a tough thing to assume, because the spec is not exactly obvious to read...)

Quoted 6 lines

(Running some quick tests by converting some images with zstd compression seems to confirm that whenever ZSTD_compressStream2() returns, either zstd_ret > output.size - output.pos, or zstd_ret == 0 and input.pos == input.size. So none of the loops ever loop.)

Max

So, what should we do?

1. Rely on the test that there's no need for the loop:
   * make one ZSTD_compressStream2() call
   * make sure it returned with zstd_ret == 0 and input.pos == input.size.
     if so, return with the size
   * if not, check that zstd_ret > output.size - output.pos if so, return with -ENOMEM
   * if none above return with -EIO

   This should cover the majority of the compressing cases

2. Leave the loop as is, because of the documentation:
   "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."

   This fulfills the documentation requirements.

3. Any other option?

Denis

Quoted 20 lines
Quoted 20 lines
block/qcow2-threads.c +19 -0
+
+            if (ZSTD_isError(zstd_ret)) {
+                ret = -EIO;
+                goto out;
+            }
+            /* Dest buffer isn't big enough to store compressed content */
+            if (zstd_ret > output.size - output.pos) {
+                ret = -ENOMEM;
+                goto out;
+            }
+        } while (zstd_ret);
+    }
+    /* make sure we can safely return compressed buffer size with ssize_t */
+    assert(output.pos <= SSIZE_MAX);
+    ret = output.pos;
+out:
+    ZSTD_freeCCtx(cctx);
+    return ret;
+}
#9 lore raw

Re: [PATCH v20 3/4] qcow2: add zstd cluster compression

Quoted 181 lines · Denis Plotnikov
Quoted 160 lines · Max Reitz
Quoted 44 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:

---
  docs/interop/qcow2.txt |   1 +
  configure              |   2 +-
  qapi/block-core.json   |   3 +-
  block/qcow2-threads.c  | 157 +++++++++++++++++++++++++++++++++++++++++ block/qcow2.c          |   7 ++

  5 files changed, 168 insertions(+), 2 deletions(-)

[...]

Quoted 34 lines
block/qcow2-threads.c +0 -0
index 7dbaf53489..0525718704 100644
@@ -28,6 +28,11 @@
   #define ZLIB_CONST
   #include <zlib.h>

  +#ifdef CONFIG_ZSTD

block/qcow2-threads.c +4 -0
+#include <zstd.h>
+#include <zstd_errors.h>
+#endif
+

  #include "qcow2.h"
  #include "block/thread-pool.h"
  #include "crypto.h"

block/qcow2-threads.c +0 -0
@@ -166,6 +171,148 @@ static ssize_t qcow2_zlib_decompress(void
 *dest, size_t dest_size,
       return ret;
   }

  +#ifdef CONFIG_ZSTD

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 static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 193

Quoted 6 lines
block/qcow2-threads.c +6 -0
+static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
+{
+    ssize_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };

Minor style note: I think it’d be nicer to use designated initializers here.

Quoted 22 lines

+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * Use the zstd streamed interface for symmetry with decompression, +     * where streaming is essential since we don't record the exact +     * compressed size.
+     *
+     * In the loop, we try to compress all the data into one zstd frame.
+     * ZSTD_compressStream2 potentially can finish a frame earlier +     * than the full input data is consumed. That's why we are looping +     * until all the input data is consumed.
+     */
+    while (input.pos < input.size) {
+        size_t zstd_ret;
+        /*
+         * 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". We assume that "start a new

Quoted 8 lines

frame"
+         * means call ZSTD_compressStream2 in the very beginning or when
+         * ZSTD_compressStream2 has returned with 0.
+         */
+        do {
+            zstd_ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);

The spec makes it sound to me like ZSTD_e_end will always complete in a single call if there’s enough space in the output buffer.  So the only team we have to loop would be when there isn’t enough space anyway:

It says this about ZSTD_e_end:

Quoted 2 lines

flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.

Those rules being:

Quoted 2 lines

Note that, if output->size is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).

So it seems like it will only return a value > 0 if the output buffer is definitely too small.

The spec also notes that the return value is greater than 0 if:

Quoted 2 lines
Quoted 1 line

0 if some data still present within internal buffer (the value is

minimal estimation of remaining size),

So it’s a minimum estimate.  That’s another point that heavily implies to me that if the return value were less than what’s left in the buffer, the function wouldn’t return but still try to write it out, until it realizes that there isn’t enough space in the output buffer, and then return a value that exceeds the remaining output buffer size.

(Because if the function just played it safe, I would expect it to return a maximum estimate.)

OTOH, if it were actually possible for ZSTD_e_end to finish a frame earlier than the end of the input, I think it would make more sense to use ZSTD_e_continue until the input is done and then finish with ZSTD_e_end, like the spec seems to propose.  That way, we’d always end up with a single frame to make decompression simpler (and I think it would also make more sense overall).

But anyway.  From how I understand the spec, this code simply always ends up creating a single frame or erroring out, without looping ever. So it isn’t exactly wrong, it just seems overly complicated.  (Again, assuming I understand the spec correctly.  Which seems like a tough thing to assume, because the spec is not exactly obvious to read...)

Quoted 6 lines

(Running some quick tests by converting some images with zstd compression seems to confirm that whenever ZSTD_compressStream2() returns, either zstd_ret > output.size - output.pos, or zstd_ret == 0 and input.pos == input.size.  So none of the loops ever loop.)

Max

So, what should we do?

1. Rely on the test that there's no need for the loop:
   * make one ZSTD_compressStream2() call
   * make sure it returned with zstd_ret == 0 and input.pos == input.size.
     if so, return with the size
   * if not, check that zstd_ret > output.size - output.pos if so, return with -ENOMEM
   * if none above return with -EIO

Quoted 1 line

   This should cover the majority of the compressing cases

According to how I interpret the spec, “none of the above” should never happen except for ZSTD_isError(zstd_ret), so this should cover all compressing cases, actually.

Quoted 4 lines

2. Leave the loop as is, because of the documentation:
   "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."

As far as I can see, the return value is always 0 or greater than the remaining buffer space, so this will always be satisfied even without a loop. (We will always have one of three cases: (1) Success and all input has been consumed, (2) ZSTD_isError(zstd_ret), so we return -EIO,
(3) zstd_ret > output.size - output.pos, so we return -ENOMEM.

I interpret the “You *must* continue until it returns 0” as “If there is no sufficient space in the output buffer, this function will return a value greater than 0 indicating how much space is at least still required. The caller is free to supply a greater output buffer for the next call (by supplying a different ZSTD_outBuffer structure), and then we’ll try again.”
(I.e., retrying with the same ZSTD_outBuffer will make the function return immediately because it knows that it’s insufficient.)

Max

#10 lore raw

Re: [PATCH v20 3/4] qcow2: add zstd cluster compression

Quoted 204 lines · Max Reitz
Quoted 179 lines · Denis Plotnikov
Quoted 160 lines · Max Reitz
Quoted 44 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:

---
   docs/interop/qcow2.txt |   1 +
   configure              |   2 +-
   qapi/block-core.json   |   3 +-
   block/qcow2-threads.c  | 157 +++++++++++++++++++++++++++++++++++++++++
   block/qcow2.c          |   7 ++
   5 files changed, 168 insertions(+), 2 deletions(-)

[...]

Quoted 34 lines
block/qcow2-threads.c +15 -0
index 7dbaf53489..0525718704 100644
@@ -28,6 +28,11 @@
   #define ZLIB_CONST
   #include <zlib.h>
   +#ifdef CONFIG_ZSTD
+#include <zstd.h>
+#include <zstd_errors.h>
+#endif
+
   #include "qcow2.h"
   #include "block/thread-pool.h"
   #include "crypto.h"
@@ -166,6 +171,148 @@ static ssize_t qcow2_zlib_decompress(void
 *dest, size_t dest_size,
       return ret;
   }
   +#ifdef CONFIG_ZSTD
+
+/*
+ * 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 static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 193

Quoted 6 lines
block/qcow2-threads.c +6 -0
+static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
+{
+    ssize_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };

Minor style note: I think it’d be nicer to use designated initializers here.

Quoted 22 lines

+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * Use the zstd streamed interface for symmetry with decompression, +     * where streaming is essential since we don't record the exact +     * compressed size.
+     *
+     * In the loop, we try to compress all the data into one zstd frame.
+     * ZSTD_compressStream2 potentially can finish a frame earlier +     * than the full input data is consumed. That's why we are looping +     * until all the input data is consumed.
+     */
+    while (input.pos < input.size) {
+        size_t zstd_ret;
+        /*
+         * 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". We assume that "start a new

Quoted 8 lines

frame"
+         * means call ZSTD_compressStream2 in the very beginning or when
+         * ZSTD_compressStream2 has returned with 0.
+         */
+        do {
+            zstd_ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);

The spec makes it sound to me like ZSTD_e_end will always complete in a single call if there’s enough space in the output buffer.  So the only team we have to loop would be when there isn’t enough space anyway:

It says this about ZSTD_e_end:

Quoted 2 lines

flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.

Those rules being:

Quoted 2 lines

Note that, if output->size is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).

So it seems like it will only return a value > 0 if the output buffer is definitely too small.

The spec also notes that the return value is greater than 0 if:

Quoted 2 lines
Quoted 1 line

0 if some data still present within internal buffer (the value is

minimal estimation of remaining size),

So it’s a minimum estimate.  That’s another point that heavily implies to me that if the return value were less than what’s left in the buffer, the function wouldn’t return but still try to write it out, until it realizes that there isn’t enough space in the output buffer, and then return a value that exceeds the remaining output buffer size.

(Because if the function just played it safe, I would expect it to return a maximum estimate.)

OTOH, if it were actually possible for ZSTD_e_end to finish a frame earlier than the end of the input, I think it would make more sense to use ZSTD_e_continue until the input is done and then finish with ZSTD_e_end, like the spec seems to propose.  That way, we’d always end up with a single frame to make decompression simpler (and I think it would also make more sense overall).

But anyway.  From how I understand the spec, this code simply always ends up creating a single frame or erroring out, without looping ever. So it isn’t exactly wrong, it just seems overly complicated.  (Again, assuming I understand the spec correctly.  Which seems like a tough thing to assume, because the spec is not exactly obvious to read...)

Quoted 6 lines

(Running some quick tests by converting some images with zstd compression seems to confirm that whenever ZSTD_compressStream2() returns, either zstd_ret > output.size - output.pos, or zstd_ret == 0 and input.pos == input.size.  So none of the loops ever loop.)

Max

So, what should we do?

1. Rely on the test that there's no need for the loop:
   * make one ZSTD_compressStream2() call
   * make sure it returned with zstd_ret == 0 and input.pos == input.size.
     if so, return with the size
   * if not, check that zstd_ret > output.size - output.pos if so, return with -ENOMEM
   * if none above return with -EIO

Quoted 1 line

   This should cover the majority of the compressing cases

According to how I interpret the spec, “none of the above” should never happen except for ZSTD_isError(zstd_ret), so this should cover all compressing cases, actually.

Quoted 4 lines

2. Leave the loop as is, because of the documentation:
   "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."

As far as I can see, the return value is always 0 or greater than the remaining buffer space, so this will always be satisfied even without a loop. (We will always have one of three cases: (1) Success and all input has been consumed, (2) ZSTD_isError(zstd_ret), so we return -EIO,
(3) zstd_ret > output.size - output.pos, so we return -ENOMEM.

I interpret the “You *must* continue until it returns 0” as “If there is no sufficient space in the output buffer, this function will return a value greater than 0 indicating how much space is at least still required. The caller is free to supply a greater output buffer for the next call (by supplying a different ZSTD_outBuffer structure), and then we’ll try again.”
(I.e., retrying with the same ZSTD_outBuffer will make the function return immediately because it knows that it’s insufficient.)

Quoted 1 line

Max

ok, removing the loop sounds reasonable.
My only concern is that *must* in the doc.
Could ZSTD-lib change the logic in the future relying on the fact that they make users use ZSTD_compressStream() in a loop. Honestly, I can't imagine the case when they would want to do that, but still.
Without the loop we're protected even in this case. The worst thing could happen because of that is qcow2_zstd_compress() would return with -EIO more frequently.

So, if I understand correctly, you are ok with removing the loop.

Denis

Quoted 1 line
#11 lore raw

Re: [PATCH v20 3/4] qcow2: add zstd cluster compression

Quoted 214 lines · Denis Plotnikov
Quoted 211 lines · Max Reitz
Quoted 186 lines · Denis Plotnikov
Quoted 167 lines · Max Reitz
Quoted 46 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:

---
   docs/interop/qcow2.txt |   1 +
   configure              |   2 +-
   qapi/block-core.json   |   3 +-
   block/qcow2-threads.c  | 157
+++++++++++++++++++++++++++++++++++++++++
   block/qcow2.c          |   7 ++

   5 files changed, 168 insertions(+), 2 deletions(-)

[...]

Quoted 34 lines
block/qcow2-threads.c +0 -0
index 7dbaf53489..0525718704 100644
@@ -28,6 +28,11 @@
    #define ZLIB_CONST
    #include <zlib.h>

   +#ifdef CONFIG_ZSTD

block/qcow2-threads.c +4 -0
+#include <zstd.h>
+#include <zstd_errors.h>
+#endif
+

   #include "qcow2.h"
   #include "block/thread-pool.h"
   #include "crypto.h"

block/qcow2-threads.c +0 -0
@@ -166,6 +171,148 @@ static ssize_t qcow2_zlib_decompress(void
 *dest, size_t dest_size,
        return ret;
    }

   +#ifdef CONFIG_ZSTD

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 static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size, · line 193

Quoted 6 lines
block/qcow2-threads.c +6 -0
+static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
+                                   const void *src, size_t src_size)
+{
+    ssize_t ret;
+    ZSTD_outBuffer output = { dest, dest_size, 0 };
+    ZSTD_inBuffer input = { src, src_size, 0 };

Minor style note: I think it’d be nicer to use designated initializers here.

Quoted 25 lines

+    ZSTD_CCtx *cctx = ZSTD_createCCtx();
+
+    if (!cctx) {
+        return -EIO;
+    }
+    /*
+     * Use the zstd streamed interface for symmetry with decompression,
+     * where streaming is essential since we don't record the exact +     * compressed size.
+     *
+     * In the loop, we try to compress all the data into one zstd frame.
+     * ZSTD_compressStream2 potentially can finish a frame earlier +     * than the full input data is consumed. That's why we are looping
+     * until all the input data is consumed.
+     */
+    while (input.pos < input.size) {
+        size_t zstd_ret;
+        /*
+         * 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". We assume that "start a new

Quoted 8 lines

frame"
+         * means call ZSTD_compressStream2 in the very beginning or when
+         * ZSTD_compressStream2 has returned with 0.
+         */
+        do {
+            zstd_ret = ZSTD_compressStream2(cctx, &output, &input, ZSTD_e_end);

The spec makes it sound to me like ZSTD_e_end will always complete in a single call if there’s enough space in the output buffer.  So the only team we have to loop would be when there isn’t enough space anyway:

It says this about ZSTD_e_end:

Quoted 2 lines

flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.

Those rules being:

Quoted 2 lines

Note that, if output->size is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).

So it seems like it will only return a value > 0 if the output buffer is
definitely too small.

The spec also notes that the return value is greater than 0 if:

Quoted 2 lines
Quoted 1 line

0 if some data still present within internal buffer (the value is

minimal estimation of remaining size),

So it’s a minimum estimate.  That’s another point that heavily implies to me that if the return value were less than what’s left in the buffer,
the function wouldn’t return but still try to write it out, until it realizes that there isn’t enough space in the output buffer, and then return a value that exceeds the remaining output buffer size.

(Because if the function just played it safe, I would expect it to return a maximum estimate.)

OTOH, if it were actually possible for ZSTD_e_end to finish a frame earlier than the end of the input, I think it would make more sense to use ZSTD_e_continue until the input is done and then finish with ZSTD_e_end, like the spec seems to propose.  That way, we’d always end up with a single frame to make decompression simpler (and I think it would also make more sense overall).

But anyway.  From how I understand the spec, this code simply always ends up creating a single frame or erroring out, without looping ever. So it isn’t exactly wrong, it just seems overly complicated.  (Again, assuming I understand the spec correctly.  Which seems like a tough thing to assume, because the spec is not exactly obvious to read...)

Quoted 6 lines

(Running some quick tests by converting some images with zstd compression seems to confirm that whenever ZSTD_compressStream2() returns, either zstd_ret > output.size - output.pos, or zstd_ret == 0 and input.pos == input.size.  So none of the loops ever loop.)

Max

So, what should we do?

1. Rely on the test that there's no need for the loop:
    * make one ZSTD_compressStream2() call
    * make sure it returned with zstd_ret == 0 and input.pos == input.size.
      if so, return with the size
    * if not, check that zstd_ret > output.size - output.pos if so, return with -ENOMEM
    * if none above return with -EIO

Quoted 1 line

    This should cover the majority of the compressing cases

According to how I interpret the spec, “none of the above” should never happen except for ZSTD_isError(zstd_ret), so this should cover all compressing cases, actually.

Quoted 4 lines

2. Leave the loop as is, because of the documentation:
    "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."

As far as I can see, the return value is always 0 or greater than the remaining buffer space, so this will always be satisfied even without a loop.  (We will always have one of three cases: (1) Success and all input has been consumed, (2) ZSTD_isError(zstd_ret), so we return -EIO,
(3) zstd_ret > output.size - output.pos, so we return -ENOMEM.

I interpret the “You *must* continue until it returns 0” as “If there is no sufficient space in the output buffer, this function will return a value greater than 0 indicating how much space is at least still required.  The caller is free to supply a greater output buffer for the next call (by supplying a different ZSTD_outBuffer structure), and then we’ll try again.”
(I.e., retrying with the same ZSTD_outBuffer will make the function return immediately because it knows that it’s insufficient.)

Quoted 4 lines
Quoted 1 line

Max

ok, removing the loop sounds reasonable.
My only concern is that *must* in the doc.

Well, if we just return an error whenever we get a return value != 0, then we shouldn’t have to care what we must and mustn’t do, because we’ll just abort the compression process then.

Quoted 2 lines

Could ZSTD-lib change the logic in the future relying on the fact that they make users use ZSTD_compressStream() in a loop.

It isn’t like I just wondered whether the loop is necessary and saw that with the current implementation, it didn’t seem necessary for any of the test images I have.

My reasoning is based on the specification, which says for ZSTD_e_flush that it will only return a value > 0 if output->size is too small; and that ZSTD_e_end follows the same rules.

So I think if they were to change behavior, they’d violate the spec.

Quoted 5 lines

Honestly, I can't imagine the case when they would want to do that, but still.
Without the loop we're protected even in this case. The worst thing could happen because of that is qcow2_zstd_compress() would return with -EIO more frequently.

I think so, too.

Quoted 1 line

So, if I understand correctly, you are ok with removing the loop.

Yes.

Max

#12 lore raw

PATCH v20 4/4 iotests: 287: add qcow2 compression type test

In reply to Denis Plotnikov

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

---
 tests/qemu-iotests/287     | 146 +++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/287.out |  67 +++++++++++++++++
 tests/qemu-iotests/group   |   1 +
 3 files changed, 214 insertions(+)
 create mode 100755 tests/qemu-iotests/287
 create mode 100644 tests/qemu-iotests/287.out
tests/qemu-iotests/287 new file +146 -0
new file mode 100755
index 0000000000..156acc40ad
@@ -0,0 +1,146 @@
+#!/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 $seq"
+
+status=1	# failure is the default!
+
+# standard environment
+. ./common.rc
+. ./common.filter
+
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux
+
+COMPR_IMG="$TEST_IMG.compressed"
+RAND_FILE="$TEST_DIR/rand_data"
+
+_cleanup()
+{
+	_cleanup_test_img
+	rm -f "$COMPR_IMG"
+	rm -f "$RAND_FILE"
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# for all the cases
+CLUSTER_SIZE=65536
+
+# Check if we can run this test.
+if IMGOPTS='compression_type=zstd' _make_test_img 64M |
+    grep "Invalid parameter 'zstd'"; then
+    _notrun "ZSTD is disabled"
+fi
+
+echo
+echo "=== Testing compression type incompatible bit setting for zlib ==="
+echo
+IMGOPTS='compression_type=zlib' _make_test_img 64M
+$PYTHON qcow2.py "$TEST_IMG" dump-header | grep incompatible_features
+
+echo
+echo "=== Testing compression type incompatible bit setting for zstd ==="
+echo
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+$PYTHON qcow2.py "$TEST_IMG" dump-header | grep incompatible_features
+
+echo
+echo "=== Testing zlib with incompatible bit set ==="
+echo
+IMGOPTS='compression_type=zlib' _make_test_img 64M
+$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
+
+if $QEMU_IMG info "$TEST_IMG" >/dev/null 2>&1 ; then
+    echo "Error: The image opened successfully. The image must not be opened."
+fi
+
+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
+
+if $QEMU_IMG info "$TEST_IMG" >/dev/null 2>&1 ; then
+    echo "Error: The image opened successfully. The image must not be opened."
+fi
+
+echo
+echo "=== Testing compression type values ==="
+echo
+# zlib=0
+IMGOPTS='compression_type=zlib' _make_test_img 64M
+peek_file_be "$TEST_IMG" 104 1
+echo
+
+# zstd=1
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+peek_file_be "$TEST_IMG" 104 1
+echo
+
+echo
+echo "=== Testing simple reading and writing with zstd ==="
+echo
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+$QEMU_IO -c "write -c -P 0xAC 64K 64K " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -P 0xAC 64K 64K " "$TEST_IMG" | _filter_qemu_io
+# read on the cluster boundaries
+$QEMU_IO -c "read -v 131070 8 " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -v 65534 8" "$TEST_IMG" | _filter_qemu_io
+
+echo
+echo "=== Testing adjacent clusters reading and writing with zstd ==="
+echo
+IMGOPTS='compression_type=zstd' _make_test_img 64M
+$QEMU_IO -c "write -c -P 0xAB 0 64K " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "write -c -P 0xAC 64K 64K " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "write -c -P 0xAD 128K 64K " "$TEST_IMG" | _filter_qemu_io
+
+$QEMU_IO -c "read -P 0xAB 0 64k " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -P 0xAC 64K 64k " "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c "read -P 0xAD 128K 64k " "$TEST_IMG" | _filter_qemu_io
+
+echo
+echo "=== Testing incompressible cluster processing with zstd ==="
+echo
+# create a 2M image and fill it with 1M likely incompressible data
+# and 1M compressible data
+dd if=/dev/urandom of="$RAND_FILE" bs=1M count=1 seek=1
+QEMU_IO_OPTIONS="$QEMU_IO_OPTIONS_NO_FMT" \
+$QEMU_IO -f raw -c "write -P 0xFA 0 1M" "$RAND_FILE" | _filter_qemu_io
+$QEMU_IMG convert -f raw -O $IMGFMT -c "$RAND_FILE" "$TEST_IMG" | _filter_qemu_io
+
+$QEMU_IMG convert -O $IMGFMT -c -o compression_type=zstd \
+                  "$TEST_IMG" "$COMPR_IMG"
+$QEMU_IMG compare "$TEST_IMG" "$COMPR_IMG"
+
+# success, all done
+echo "*** done"
+rm -f $seq.full
+status=0
tests/qemu-iotests/287.out new file +67 -0

67 lines not on this page · raw message

tests/qemu-iotests/group +1 -0
index 435dccd5af..3bbe02c88d 100644
@@ -294,6 +294,7 @@
 283 auto quick
 284 rw
 286 rw quick
+287 auto quick
 288 quick
 289 rw quick
 290 rw auto quick
Signature · 1 line
2.17.0
Reviewed-by Tested-by on v20 4/4
#13 lore raw

Re: [PATCH v20 4/4] iotests: 287: add qcow2 compression type test

Quoted 4 lines · Denis Plotnikov

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

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

Re: [PATCH v20 4/4] iotests: 287: add qcow2 compression type test

In reply to Denis Plotnikov
Quoted 44 lines · Denis Plotnikov

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

---
 tests/qemu-iotests/287     | 146 +++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/287.out |  67 +++++++++++++++++
 tests/qemu-iotests/group   |   1 +
 3 files changed, 214 insertions(+)
 create mode 100755 tests/qemu-iotests/287
 create mode 100644 tests/qemu-iotests/287.out
tests/qemu-iotests/287 new file +26 -0
new file mode 100755
index 0000000000..156acc40ad
@@ -0,0 +1,146 @@
+#!/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"
+

tests/qemu-iotests/287 line 36

Quoted 10 lines
tests/qemu-iotests/287 new file +10 -0
+status=1	# failure is the default!
+
+# standard environment
+. ./common.rc
+. ./common.filter
+
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux

This test doesn’t work with compat=0.10 (because we can’t store a non-default compression type there) or data_file (because those don’t support compression), so those options should be marked as unsupported.

(It does seem to work with any refcount_bits, though.)

tests/qemu-iotests/287 line 44

Quoted 8 lines
tests/qemu-iotests/287 new file +8 -0
+
+COMPR_IMG="$TEST_IMG.compressed"
+RAND_FILE="$TEST_DIR/rand_data"
+
+_cleanup()
+{
+	_cleanup_test_img
+	rm -f "$COMPR_IMG"

Using _rm_test_img() would be nicer. There shouldn’t be a functional difference here because there’d only be one with external data files (I think), which won’t work with this test, but still.

Quoted 7 lines
tests/qemu-iotests/287 new file +7 -0
+	rm -f "$RAND_FILE"
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# for all the cases
+CLUSTER_SIZE=65536
+

tests/qemu-iotests/287 line 61

Quoted 10 lines
tests/qemu-iotests/287 new file +10 -0
+# Check if we can run this test.
+if IMGOPTS='compression_type=zstd' _make_test_img 64M |
+    grep "Invalid parameter 'zstd'"; then
+    _notrun "ZSTD is disabled"
+fi
+
+echo
+echo "=== Testing compression type incompatible bit setting for zlib ==="
+echo
+IMGOPTS='compression_type=zlib' _make_test_img 64M

Please use -o so user options are still considered.

(i.e., _make_test_img -o compression_type=zlib)

[...]

tests/qemu-iotests/287 line 140

Quoted 12 lines
tests/qemu-iotests/287 new file +12 -0
+echo
+echo "=== Testing incompressible cluster processing with zstd ==="
+echo
+# create a 2M image and fill it with 1M likely incompressible data
+# and 1M compressible data
+dd if=/dev/urandom of="$RAND_FILE" bs=1M count=1 seek=1
+QEMU_IO_OPTIONS="$QEMU_IO_OPTIONS_NO_FMT" \
+$QEMU_IO -f raw -c "write -P 0xFA 0 1M" "$RAND_FILE" | _filter_qemu_io
+$QEMU_IMG convert -f raw -O $IMGFMT -c "$RAND_FILE" "$TEST_IMG" | _filter_qemu_io
+
+$QEMU_IMG convert -O $IMGFMT -c -o compression_type=zstd \
+                  "$TEST_IMG" "$COMPR_IMG"

Again, it would be nice to not discard the user-supplied options here, and maybe it would also be nicer to explicitly pass the compression type for the first convert, too. So we’d use
-o "$(_optstr_add "$IMGOPTS" "compression_type=zlib")"
for the first convert, and
-o "$(_optstr_add "$IMGOPTS" "compression_type=zstd")"
for the second one.

Max

#15 lore raw

Re: [PATCH v20 4/4] iotests: 287: add qcow2 compression type test

Quoted 59 lines · Max Reitz
Quoted 44 lines · Denis Plotnikov

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

---
  tests/qemu-iotests/287     | 146 +++++++++++++++++++++++++++++++++++++
  tests/qemu-iotests/287.out |  67 +++++++++++++++++
  tests/qemu-iotests/group   |   1 +
  3 files changed, 214 insertions(+)

create mode 100755 tests/qemu-iotests/287
create mode 100644 tests/qemu-iotests/287.out

tests/qemu-iotests/287 new file +26 -0
new file mode 100755
index 0000000000..156acc40ad
@@ -0,0 +1,146 @@
+#!/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"
+

tests/qemu-iotests/287 line 36

Quoted 10 lines
tests/qemu-iotests/287 new file +10 -0
+status=1	# failure is the default!
+
+# standard environment
+. ./common.rc
+. ./common.filter
+
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux

This test doesn’t work with compat=0.10 (because we can’t store a non-default compression type there) or data_file (because those don’t support compression), so those options should be marked as unsupported.

Quoted 1 line

(It does seem to work with any refcount_bits, though.)

Could I ask how to achieve that?
I can't find any _supported_* related.

Denis

Quoted 58 lines

tests/qemu-iotests/287 line 44

Quoted 9 lines
tests/qemu-iotests/287 new file +8 -0
+
+COMPR_IMG="$TEST_IMG.compressed"
+RAND_FILE="$TEST_DIR/rand_data"
+
+_cleanup()
+{
+	_cleanup_test_img
+	rm -f "$COMPR_IMG"

Using _rm_test_img() would be nicer. There shouldn’t be a functional difference here because there’d only be one with external data files (I think), which won’t work with this test, but still.

Quoted 7 lines
tests/qemu-iotests/287 new file +7 -0
+	rm -f "$RAND_FILE"
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# for all the cases
+CLUSTER_SIZE=65536
+

tests/qemu-iotests/287 line 61

Quoted 10 lines
tests/qemu-iotests/287 new file +10 -0
+# Check if we can run this test.
+if IMGOPTS='compression_type=zstd' _make_test_img 64M |
+    grep "Invalid parameter 'zstd'"; then
+    _notrun "ZSTD is disabled"
+fi
+
+echo
+echo "=== Testing compression type incompatible bit setting for zlib ==="
+echo
+IMGOPTS='compression_type=zlib' _make_test_img 64M

Please use -o so user options are still considered.

(i.e., _make_test_img -o compression_type=zlib)

[...]

tests/qemu-iotests/287 line 140

Quoted 12 lines
tests/qemu-iotests/287 new file +12 -0
+echo
+echo "=== Testing incompressible cluster processing with zstd ==="
+echo
+# create a 2M image and fill it with 1M likely incompressible data
+# and 1M compressible data
+dd if=/dev/urandom of="$RAND_FILE" bs=1M count=1 seek=1
+QEMU_IO_OPTIONS="$QEMU_IO_OPTIONS_NO_FMT" \
+$QEMU_IO -f raw -c "write -P 0xFA 0 1M" "$RAND_FILE" | _filter_qemu_io
+$QEMU_IMG convert -f raw -O $IMGFMT -c "$RAND_FILE" "$TEST_IMG" | _filter_qemu_io
+
+$QEMU_IMG convert -O $IMGFMT -c -o compression_type=zstd \
+                  "$TEST_IMG" "$COMPR_IMG"

Again, it would be nice to not discard the user-supplied options here, and maybe it would also be nicer to explicitly pass the compression type for the first convert, too. So we’d use
-o "$(_optstr_add "$IMGOPTS" "compression_type=zlib")"
for the first convert, and
-o "$(_optstr_add "$IMGOPTS" "compression_type=zstd")"
for the second one.

Max

#16 lore raw

Re: [PATCH v20 4/4] iotests: 287: add qcow2 compression type test

Quoted 62 lines · Denis Plotnikov
Quoted 59 lines · Max Reitz
Quoted 44 lines · Denis Plotnikov

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

---
  tests/qemu-iotests/287     | 146 +++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/287.out |  67 +++++++++++++++++ tests/qemu-iotests/group   |   1 +

  3 files changed, 214 insertions(+)

  create mode 100755 tests/qemu-iotests/287
  create mode 100644 tests/qemu-iotests/287.out

tests/qemu-iotests/287 new file +26 -0
new file mode 100755
index 0000000000..156acc40ad
@@ -0,0 +1,146 @@
+#!/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"
+

tests/qemu-iotests/287 line 36

Quoted 10 lines
tests/qemu-iotests/287 new file +10 -0
+status=1    # failure is the default!
+
+# standard environment
+. ./common.rc
+. ./common.filter
+
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux

This test doesn’t work with compat=0.10 (because we can’t store a non-default compression type there) or data_file (because those don’t support compression), so those options should be marked as unsupported.

Quoted 4 lines
Quoted 1 line

(It does seem to work with any refcount_bits, though.)

Could I ask how to achieve that?
I can't find any _supported_* related.

It’s _unsupported_imgopts.

Max

#17 lore raw

Re: [PATCH v20 4/4] iotests: 287: add qcow2 compression type test

Quoted 8 lines · Max Reitz
Quoted 8 lines
Quoted 8 lines

tests/qemu-iotests/287 line 36

Quoted 4 lines
tests/qemu-iotests/287 new file +4 -0
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux

This test doesn’t work with compat=0.10 (because we can’t store a non-default compression type there) or data_file (because those don’t support compression), so those options should be marked as unsupported.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

(It does seem to work with any refcount_bits, though.)

Could I ask how to achieve that?
I can't find any _supported_* related.

It’s _unsupported_imgopts.

Test 036 is an example of this.

Signature · 3 lines
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org
#18 lore raw

Re: [PATCH v20 4/4] iotests: 287: add qcow2 compression type test

Quoted 12 lines · Eric Blake
Quoted 10 lines · Max Reitz
Quoted 10 lines
Quoted 9 lines

tests/qemu-iotests/287 line 36

Quoted 4 lines
tests/qemu-iotests/287 new file +4 -0
+# This tests qocw2-specific low-level functionality
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux

This test doesn’t work with compat=0.10 (because we can’t store a non-default compression type there) or data_file (because those don’t support compression), so those options should be marked as unsupported.

(It does seem to work with any refcount_bits, though.)

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Could I ask how to achieve that?
I can't find any _supported_* related.

It’s _unsupported_imgopts.

Test 036 is an example of this.

Max, Eric

Thanks!

Denis

Quoted 2 lines