All of lore.kernel.org
 help / color / mirror / Atom feed
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
To: Denis Plotnikov <dplotnikov@virtuozzo.com>, qemu-devel@nongnu.org
Cc: kwolf@redhat.com, qemu-block@nongnu.org, den@vrtuozzo.com,
	armbru@redhat.com, mreitz@redhat.com
Subject: Re: [PATCH v1 3/8] qcow2: add zstd cluster compression
Date: Thu, 27 Feb 2020 17:18:33 +0300	[thread overview]
Message-ID: <85e9b76e-80f8-c49d-8db0-545cd0871647@virtuozzo.com> (raw)
In-Reply-To: <583362b9-76df-089c-8c1b-0b2fbda7533c@virtuozzo.com>

27.02.2020 17:11, Denis Plotnikov wrote:
> 
> 
> On 27.02.2020 12:55, Vladimir Sementsov-Ogievskiy wrote:
>> 27.02.2020 10:29, Denis Plotnikov wrote:
>>> 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
>>>
>>> Signed-off-by: Denis Plotnikov <dplotnikov@virtuozzo.com>
>>> ---
>>>   block/qcow2-threads.c  | 122 +++++++++++++++++++++++++++++++++++++++++
>>>   block/qcow2.c          |   7 +++
>>>   configure              |  29 ++++++++++
>>>   docs/interop/qcow2.txt |  18 ++++++
>>>   qapi/block-core.json   |   3 +-
>>>   5 files changed, 178 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/block/qcow2-threads.c b/block/qcow2-threads.c
>>> index 1c128e9840..e942c4d7e5 100644
>>> --- a/block/qcow2-threads.c
>>> +++ b/block/qcow2-threads.c
>>> @@ -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"
>>> @@ -164,6 +169,113 @@ static ssize_t qcow2_zlib_decompress(void *dest, size_t dest_size,
>>>       return ret;
>>>   }
>>>   +#ifdef CONFIG_ZSTD
>>> +
>>> +#define ZSTD_LEN_BUF 4
>>> +
>>> +/*
>>> + * 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
>>
>> This doesn't match qcow2_co_compress definition. You should return 0 on success.
> does it? I'd rather say it doesn't match to qcow2_co_compress description in the function comment, which we can change actually,
> because qcow2_co_compress is used like:

Oh, yes, you are right. Then we should change the comment.

> 
> block/qcow2.c:
> 
> static coroutine_fn int
> qcow2_co_pwritev_compressed_task(BlockDriverState *bs,
>                                   uint64_t offset, uint64_t bytes,
>                                   QEMUIOVector *qiov, size_t qiov_offset)
> {
>      ...
>      out_buf = g_malloc(s->cluster_size);
> 
>      out_len = qcow2_co_compress(bs, out_buf, s->cluster_size - 1,
>                                  buf, s->cluster_size);
>      if (out_len == -ENOMEM) {
>          /* could not compress: write normal cluster */
>          ret = qcow2_co_pwritev_part(bs, offset, bytes, qiov, qiov_offset, 0);
>          if (ret < 0) {
>              goto fail;
>          }
>          goto success;
>      } else if (out_len < 0) {
>          ret = -EINVAL;
>          goto fail;
>      }
> 
>      qemu_co_mutex_lock(&s->lock);
>      ret = qcow2_alloc_compressed_cluster_offset(bs, offset, out_len, <<<<<<<<<<<<
> &cluster_offset);
>      ...
> }
> 
>>
>>> + *          -ENOMEM destination buffer is not enough to store compressed data
>>> + *          -EIO    on any other error
>>> + */
>>> +
>>> +static ssize_t qcow2_zstd_compress(void *dest, size_t dest_size,
>>> +                                   const void *src, size_t src_size)
>>> +{
>>> +    size_t ret;
>>> +
>>> +    /*
>>> +     * steal ZSTD_LEN_BUF bytes in the very beginng of the buffer
>>
>> beginning
>>
>>> +     * to store compressed chunk size
>>> +     */
>>> +    char *d_buf = ((char *) dest) + ZSTD_LEN_BUF;
>>> +
>>> +    /*
>>> +     * sanity check that we can store the compressed data length,
>>> +     * and there is some space left for the compressor buffer
>>> +     */
>>> +    if (dest_size <= ZSTD_LEN_BUF) {
>>> +        return -ENOMEM;
>>> +    }
>>> +
>>> +    dest_size -= ZSTD_LEN_BUF;
>>> +
>>> +    ret = ZSTD_compress(d_buf, dest_size, src, src_size, 5);
>>> +
>>> +    if (ZSTD_isError(ret)) {
>>> +        if (ZSTD_getErrorCode(ret) == ZSTD_error_dstSize_tooSmall) {
>>> +            return -ENOMEM;
>>> +        } else {
>>> +            return -EIO;
>>> +        }
>>> +    }
>>> +
>>> +    /* paraniod sanity check that we can store the commpressed size */
>>> +    if (ret > UINT_MAX) {
>>> +        return -ENOMEM;
>>> +    }
>>
>> I'd use UINT32_MAX, possibly even more paranoid)
> ok
>>
>>
>>> +
>>> +    /* store the compressed chunk size in the very beginning of the buffer */
>>> +    stl_be_p(dest, ret);
>>> +
>>> +    return ret + ZSTD_LEN_BUF;
>>
>> return 0;
>>
>>> +}
>>> +
>>> +/*
>>> + * qcow2_zstd_decompress()
>>> + *
>>> + * Decompress some data (not more than @src_size bytes) to produce exactly
>>> + * @dest_size bytes using zstd compression method
>>> + *
>>> + * @dest - destination buffer, @dest_size bytes
>>> + * @src - source buffer, @src_size bytes
>>> + *
>>> + * Returns: 0 on success
>>> + *          -EIO on any error
>>> + */
>>> +
>>
>> extra empty line.
> will be removed
>>
>>
>>> +static ssize_t qcow2_zstd_decompress(void *dest, size_t dest_size,
>>> +                                     const void *src, size_t src_size)
>>> +{
>>> +    /*
>>> +     * zstd decompress wants to know the exact length of the data.
>>> +     * For that purpose, on compression, the length is stored in
>>> +     * the very beginning of the compressed buffer
>>> +     */
>>> +    size_t s_size;
>>> +    const char *s_buf = ((const char *) src) + ZSTD_LEN_BUF;
>>> +
>>> +    /*
>>> +     * sanity check that we can read 4 byte the content length and
>>> +     * and there is some content to decompress
>>> +     */
>>> +    if (src_size <= ZSTD_LEN_BUF) {
>>> +        return -EIO;
>>> +    }
>>> +
>>> +    s_size = ldl_be_p(src);
>>> +
>>> +    /* sanity check that the buffer is big enough to read the content from */
>>> +    if (src_size - ZSTD_LEN_BUF < s_size) {
>>> +        return -EIO;
>>> +    }
>>> +
>>> +    if (ZSTD_isError(
>>> +            ZSTD_decompress(dest, dest_size, s_buf, s_size))) {
>>
>> hmm, it fit into one line actually
> will be fixed if so
>>
>>
>>> +        return -EIO;
>>> +    }
>>> +
>>> +    return 0;
>>> +}
>>> +#endif
>>> +
>>>   static int qcow2_compress_pool_func(void *opaque)
>>>   {
>>>       Qcow2CompressData *data = opaque;
>>> @@ -215,6 +327,11 @@ qcow2_co_compress(BlockDriverState *bs, void *dest, size_t dest_size,
>>>           fn = qcow2_zlib_compress;
>>>           break;
>>>   +#ifdef CONFIG_ZSTD
>>> +    case QCOW2_COMPRESSION_TYPE_ZSTD:
>>> +        fn = qcow2_zstd_compress;
>>> +        break;
>>> +#endif
>>>       default:
>>>           return -ENOTSUP;
>>>       }
>>> @@ -247,6 +364,11 @@ qcow2_co_decompress(BlockDriverState *bs, void *dest, size_t dest_size,
>>>           fn = qcow2_zlib_decompress;
>>>           break;
>>>   +#ifdef CONFIG_ZSTD
>>> +    case QCOW2_COMPRESSION_TYPE_ZSTD:
>>> +        fn = qcow2_zstd_decompress;
>>> +        break;
>>> +#endif
>>>       default:
>>>           return -ENOTSUP;
>>>       }
>>> diff --git a/block/qcow2.c b/block/qcow2.c
>>> index 2ccb2cabd1..9c8ad9d580 100644
>>> --- a/block/qcow2.c
>>> +++ b/block/qcow2.c
>>> @@ -1257,6 +1257,9 @@ 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:
>>> @@ -3506,6 +3509,10 @@ 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_errno(errp, -EINVAL, "Unknown compression type");
>>>               goto out;
>>> diff --git a/configure b/configure
>>> index 48d6f89d57..4690a7ea9f 100755
>>> --- a/configure
>>> +++ b/configure
>>> @@ -444,6 +444,7 @@ opengl_dmabuf="no"
>>>   cpuid_h="no"
>>>   avx2_opt=""
>>>   zlib="yes"
>>> +zstd=""
>>>   capstone=""
>>>   lzo=""
>>>   snappy=""
>>> @@ -1371,6 +1372,10 @@ for opt do
>>>     ;;
>>>     --disable-lzfse) lzfse="no"
>>>     ;;
>>> +  --enable-zstd) zstd="yes"
>>> +  ;;
>>> +  --disable-zstd) zstd="no"
>>> +  ;;
>>>     --enable-guest-agent) guest_agent="yes"
>>>     ;;
>>>     --disable-guest-agent) guest_agent="no"
>>> @@ -1829,6 +1834,7 @@ disabled with --disable-FEATURE, default is enabled if available:
>>>                     (for reading bzip2-compressed dmg images)
>>>     lzfse           support of lzfse compression library
>>>                     (for reading lzfse-compressed dmg images)
>>> +  zstd            support of zstd compression library
>>>     seccomp         seccomp support
>>>     coroutine-pool  coroutine freelist (better performance)
>>>     glusterfs       GlusterFS backend
>>> @@ -2453,6 +2459,25 @@ EOF
>>>       fi
>>>   fi
>>>   +#########################################
>>> +# zstd check
>>> +
>>> +if test "$zstd" != "no" ; then
>>> +    cat > $TMPC << EOF
>>> +#include <zstd.h>
>>> +int main(void) { ZSTD_versionNumber(); return 0; }
>>> +EOF
>>> +    if compile_prog "" "-lzstd" ; then
>>> +        LIBS="$LIBS -lzstd"
>>> +        zstd="yes"
>>> +    else
>>> +        if test "$zstd" = "yes"; then
>>> +            feature_not_found "zstd" "Install libzstd-devel"
>>
>> to correspond to style used around: s/libzstd-devel/libzstd devel/
> My intention was to give a package name for a user encountered this message to copy-paste it to "yum install".

What about apt users? :)

> If this doesn't make things easier, because the package name can be changed in the future (although I doubt) I'll fix it.
> If the style isn't that important I'd rather leave it as is.

I don't care, keep it as is.

>>
>>> +        fi
>>> +        zstd="no"
>>> +    fi
>>> +fi
>>> +
>>>   ##########################################
>>>   # libseccomp check
>>>   @@ -6668,6 +6693,7 @@ echo "lzo support       $lzo"
>>>   echo "snappy support    $snappy"
>>>   echo "bzip2 support     $bzip2"
>>>   echo "lzfse support     $lzfse"
>>> +echo "zstd support      $zstd"
>>>   echo "NUMA host support $numa"
>>>   echo "libxml2           $libxml2"
>>>   echo "tcmalloc support  $tcmalloc"
>>> @@ -7559,6 +7585,9 @@ if test "$plugins" = "yes" ; then
>>>           "\$ld_exported_symbols_list should have been set to 'yes'."
>>>       fi
>>>   fi
>>> +if test "$zstd" = "yes" ; then
>>> +  echo "CONFIG_ZSTD=y" >> $config_host_mak
>>> +fi
>>>     if test "$tcg_interpreter" = "yes"; then
>>>     QEMU_INCLUDES="-iquote \$(SRC_PATH)/tcg/tci $QEMU_INCLUDES"
>>> diff --git a/docs/interop/qcow2.txt b/docs/interop/qcow2.txt
>>> index 5597e24474..aeca2ddebd 100644
>>> --- a/docs/interop/qcow2.txt
>>> +++ b/docs/interop/qcow2.txt
>>> @@ -208,6 +208,7 @@ version 2.
>>>                         Available compression type values:
>>>                           0: zlib <https://www.zlib.net/>
>>> +                        1: zstd <http://github.com/facebook/zstd>
>>>       === Header padding ===
>>> @@ -575,11 +576,28 @@ Compressed Clusters Descriptor (x = 62 - (cluster_bits - 8)):
>>>                       Another compressed cluster may map to the tail of the final
>>>                       sector used by this compressed cluster.
>>>   +                    The layout of the compressed data depends on the compression
>>> +                    type used for the image (see compressed cluster layout).
>>> +
>>>   If a cluster is unallocated, read requests shall read the data from the backing
>>>   file (except if bit 0 in the Standard Cluster Descriptor is set). If there is
>>>   no backing file or the backing file is smaller than the image, they shall read
>>>   zeros for all parts that are not covered by the backing file.
>>>   +=== Compressed Cluster Layout ===
>>> +
>>> +The compressed cluster data has a layout depending on the compression
>>> +type used for the image, as follows:
>>> +
>>> +Compressed data layout for the available compression types:
>>> +(x = data_space_length - 1)
>>
>> Note, that term 'data_space_lenght' is not defined in the spec..
>>
>> Hmm. And it's not trivial to define it correctly, all we have is offset and
>> "number of additional sectors", and may be not all bytes of final sector are
>> occupied... What about something like this:
>>
>> In the following layout description byte 0 corresponds to byte at host cluster offset,
>> as defined by "Compressed Clusters Descriptor" paragraph above. The whole layout occupy
>> space starting from this offset, using additional 512-byte sectors defined by
>> "Compressed Clusters Descriptor" paragraph, not necessarily occupying all of the bytes
>> in the final sector. Let x be number of last byte of the layout.
> Sounds too complicated for me, may be
> x - data chunk length available to store a compressed cluster. (for more details see "Compressed Clusters Descriptor")

Ok

>>
>>
>>> +
>>> +    0:  (default)  zlib <http://zlib.net/>:
>>> +            Byte  0 -  x:     the compressed data content
>>> +                              all the space provided used for compressed data
>>> +    1:  zstd <http://github.com/facebook/zstd>:
>>> +            Byte  0 -  3:     the length of compressed data in bytes
>>> +                  4 -  x:     the compressed data content
>>
>> Maybe, note that x+! == 4 + length, where length is value of first field of the layout.
>>
>>>     == Snapshots ==
>>>   diff --git a/qapi/block-core.json b/qapi/block-core.json
>>> index 873fbef3b5..4b6e576c44 100644
>>> --- a/qapi/block-core.json
>>> +++ b/qapi/block-core.json
>>> @@ -4401,11 +4401,12 @@
>>>   # Compression type used in qcow2 image file
>>>   #
>>>   # @zlib:  zlib compression, see <http://zlib.net/>
>>> +# @zstd:  zstd compression, see <http://github.com/facebook/zstd>
>>>   #
>>>   # Since: 5.0
>>>   ##
>>>   { 'enum': 'Qcow2CompressionType',
>>> -  'data': [ 'zlib' ] }
>>> +  'data': [ 'zlib', { 'name': 'zstd', 'if': 'defined(CONFIG_ZSTD)' } ] }
>>>     ##
>>>   # @BlockdevCreateOptionsQcow2:
>>>
>>
>>
> 


-- 
Best regards,
Vladimir


  reply	other threads:[~2020-02-27 14:19 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-02-27  7:29 [PATCH v1 0/8] qcow2: Implement zstd cluster compression method Denis Plotnikov
2020-02-27  7:29 ` [PATCH v1 1/8] qcow2: introduce compression type feature Denis Plotnikov
2020-02-27  8:21   ` Vladimir Sementsov-Ogievskiy
2020-02-27 13:24     ` Eric Blake
2020-02-27 13:48   ` Eric Blake
2020-02-27 13:59     ` Vladimir Sementsov-Ogievskiy
2020-02-27 14:13       ` Eric Blake
2020-02-27 14:30         ` Vladimir Sementsov-Ogievskiy
2020-02-27 14:39           ` Eric Blake
2020-02-28  8:34     ` Denis Plotnikov
2020-02-27  7:29 ` [PATCH v1 2/8] qcow2: rework the cluster compression routine Denis Plotnikov
2020-02-27  8:54   ` Vladimir Sementsov-Ogievskiy
2020-02-27  7:29 ` [PATCH v1 3/8] qcow2: add zstd cluster compression Denis Plotnikov
2020-02-27  9:55   ` Vladimir Sementsov-Ogievskiy
2020-02-27 14:11     ` Denis Plotnikov
2020-02-27 14:18       ` Vladimir Sementsov-Ogievskiy [this message]
2020-02-28 11:59         ` Denis Plotnikov
2020-02-27 14:01   ` Eric Blake
2020-02-28 11:40     ` Denis Plotnikov
2020-02-27  7:29 ` [PATCH v1 4/8] iotests: filter out compression_type Denis Plotnikov
2020-02-27  9:57   ` Vladimir Sementsov-Ogievskiy
2020-02-27 14:03   ` Eric Blake
2020-02-28 11:53     ` Denis Plotnikov
2020-02-27  7:29 ` [PATCH v1 5/8] iotests: fix header size, feature table size and backing file offset Denis Plotnikov
2020-02-27  9:59   ` Vladimir Sementsov-Ogievskiy
2020-02-27  7:29 ` [PATCH v1 6/8] iotests: add "compression type" for test output matching Denis Plotnikov
2020-02-27 10:04   ` Vladimir Sementsov-Ogievskiy
2020-02-27 10:09     ` Vladimir Sementsov-Ogievskiy
2020-02-27 14:06       ` Eric Blake
2020-02-28  8:13       ` Denis Plotnikov
2020-02-27  7:29 ` [PATCH v1 7/8] iotests: 080: update header size value because of adding compression type Denis Plotnikov
2020-02-27  7:29 ` [PATCH v1 8/8] iotests: 287: add qcow2 compression type test Denis Plotnikov
2020-02-27 10:29   ` Vladimir Sementsov-Ogievskiy
2020-02-28  8:23     ` Denis Plotnikov
2020-02-28  8:32       ` Vladimir Sementsov-Ogievskiy

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=85e9b76e-80f8-c49d-8db0-545cd0871647@virtuozzo.com \
    --to=vsementsov@virtuozzo.com \
    --cc=armbru@redhat.com \
    --cc=den@vrtuozzo.com \
    --cc=dplotnikov@virtuozzo.com \
    --cc=kwolf@redhat.com \
    --cc=mreitz@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.