Re: [Qemu-block] [PATCH 4/5] qemu-iotests: Test commit block job where top has two parents

[Eric Blake]

On 09/25/2017 07:28 AM, Kevin Wolf wrote:
> Signed-off-by: Kevin Wolf <address@hidden>
> ---
>  tests/qemu-iotests/191     | 152 +++++++++
>  tests/qemu-iotests/191.out | 827 
> +++++++++++++++++++++++++++++++++++++++++++++
>  tests/qemu-iotests/group   |   1 +
>  3 files changed, 980 insertions(+)
>  create mode 100755 tests/qemu-iotests/191
>  create mode 100644 tests/qemu-iotests/191.out
> 

> +
> +_cleanup()
> +{
> +    rm -f "${TEST_IMG}.mid"
> +    rm -f "${TEST_IMG}.ovl2"
> +    rm -f "${TEST_IMG}.ovl3"
> +    _cleanup_test_img
> +    _cleanup_qemu
> +}
> +trap "_cleanup; exit \$status" 0 1 2 3 15

Semantic conflict with Jeff's work to allow preservation of intermediate
files via './check -s'.

> +++ b/tests/qemu-iotests/191.out

> +{
> +    "timestamp": {
> +        "seconds":  TIMESTAMP,
> +        "microseconds":  TIMESTAMP
> +    },
> +    "event": "BLOCK_JOB_COMPLETED",
> +    "data": {
> +        "device": "commit0",
> +        "len": 67108864,
> +        "offset": 67108864,
> +        "speed": 0,
> +        "type": "commit"
> +    }
> +}

This may be sensitive to ordering if I ever finish my patches that allow
for QAPI->JSON conversion without having to first go through QObject (as
it is, it's dependent on QDict's hash being stable, whereas my code
switches things to be stable according to QAPI ordering).  But that's
not a problem for your patch.


> +                        "data": {
> +                            "compat": "1.1",

You should make the test specifically exclude compat=0.10 images, or
else have further filtering in place if we can still support this on old
images.


-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

signature.asc
Description: OpenPGP digital signature