Re: [Qemu-devel] [PATCH] qemu-img convert: Fix copy length for the iteration

[Max Reitz]

On 2015-02-11 at 06:08, Kevin Wolf wrote:
> qemu-img converts already checks the block status of the next sectors to
> copy and skips them if it knows that they don't need to be copied (e.g.
> because they are zero and the destination is zero-initialised).
>
> It doesn't, however, limit the number of sectors read to the number of
> allocated sectors when copying data yet. This means that when using -B,
> some data is needlessly copied from the backing file into the overlay.
> For images without backing files, it means that more data is read than
> is necessary, even though the zero buffer detection will still keep
> those sectors sparse.
>
> This patch changes qemu-img to respect the number of allocated sectors
> and limit its requests accordingly when processing data, which fixes -B
> to expose the expected behaviour and is a performance optimisation for
> other cases.
>
> Reported-by: Max Reitz <address@hidden>
> Signed-off-by: Kevin Wolf <address@hidden>
> ---
>   qemu-img.c                 |  1 +
>   tests/qemu-iotests/122     | 56 ++++++++++++++++++++++++++++++++++++++++++++++
>   tests/qemu-iotests/122.out | 11 +++++++++
>   tests/qemu-iotests/group   |  1 +
>   4 files changed, 69 insertions(+)
>   create mode 100755 tests/qemu-iotests/122
>   create mode 100644 tests/qemu-iotests/122.out
>
> diff --git a/qemu-img.c b/qemu-img.c
> index e148af8..b94afe2 100644
> --- a/qemu-img.c
> +++ b/qemu-img.c
> @@ -1809,6 +1809,7 @@ restart:
>               }
>   
>               n = MIN(nb_sectors, bufsectors);
> +            n = MIN(n, sector_num_next_status - sector_num);

This needs to be done only if sector_num_next_status > sector_num, 
otherwise we'll run into an infinite loop.

Apparently I accidentally broke -n for qemu-img convert, but once I get 
it working again, you can see that this command:

$ qemu-img convert -n -O raw test.qcow2 nbd://localhost:10809

does not terminate (because out_baseimg is NULL and has_zero_init is 
false due to nbd not supporting that, which means that 
sector_num_next_status is never set to anything but 0).

Other than that, this looks good to me.

Max

>               /* round down request length to an aligned sector, but
>                * do not bother doing this on short requests. They happen
> diff --git a/tests/qemu-iotests/122 b/tests/qemu-iotests/122
> new file mode 100755
> index 0000000..0106e8a
> --- /dev/null
> +++ b/tests/qemu-iotests/122
> @@ -0,0 +1,56 @@
> +#!/bin/bash
> +#
> +# Test image sparseness after qemu-img convert
> +#
> +# Copyright (C) 2015 Red Hat, Inc.
> +#
> +# 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
> address@hidden
> +
> +seq="$(basename $0)"
> +echo "QA output created by $seq"
> +
> +here="$PWD"
> +tmp=/tmp/$$
> +status=1       # failure is the default!
> +
> +_cleanup()
> +{
> +       _cleanup_test_img
> +}
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +# get standard environment, filters and checks
> +. ./common.rc
> +. ./common.filter
> +
> +_supported_fmt qcow2
> +_supported_proto generic
> +_supported_os Linux
> +
> +
> +TEST_IMG=$TEST_IMG.base _make_test_img 64M
> +_make_test_img -b $TEST_IMG.base
> +$QEMU_IO -c "write -P 0x11 0 64M" $TEST_IMG.base 2>&1 | _filter_qemu_io | 
> _filter_testdir
> +$QEMU_IO -c "write -P 0x22 0 3M" $TEST_IMG 2>&1 | _filter_qemu_io | 
> _filter_testdir
> +$QEMU_IMG convert -O $IMGFMT -B $TEST_IMG.base $TEST_IMG $TEST_IMG.orig
> +$QEMU_IMG map $TEST_IMG.orig | _filter_qemu_img_map
> +
> +# success, all done
> +echo '*** done'
> +rm -f $seq.full
> +status=0
> diff --git a/tests/qemu-iotests/122.out b/tests/qemu-iotests/122.out
> new file mode 100644
> index 0000000..6e4a760
> --- /dev/null
> +++ b/tests/qemu-iotests/122.out
> @@ -0,0 +1,11 @@
> +QA output created by 122
> +Formatting 'TEST_DIR/t.IMGFMT.base', fmt=IMGFMT size=67108864
> +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864 
> backing_file='TEST_DIR/t.IMGFMT.base'
> +wrote 67108864/67108864 bytes at offset 0
> +64 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +wrote 3145728/3145728 bytes at offset 0
> +3 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +Offset          Length          File
> +0               0x300000        TEST_DIR/t.IMGFMT.orig
> +0x300000        0x3d00000       TEST_DIR/t.IMGFMT.base
> +*** done
> diff --git a/tests/qemu-iotests/group b/tests/qemu-iotests/group
> index 4b2b93b..28c78e5 100644
> --- a/tests/qemu-iotests/group
> +++ b/tests/qemu-iotests/group
> @@ -117,3 +117,4 @@
>   113 rw auto quick
>   114 rw auto quick
>   116 rw auto quick
> +122 rw auto quick