On 7/25/19 11:57 AM, Max Reitz wrote:
> Several vmdk subformats do not work with iotest 126, so disable them.
>
> (twoGbMaxExtentSparse actually should work, but fixing that is a bit
> difficult. The problem is that the vmdk descriptor file will contain a
> referenc to "image:base.vmdk", which the block layer cannot open because
reference
> it does not know the protocol "image". This is not trivial to solve,
> because I suppose real protocols like "http://"; should be supported.
> Making vmdk treat all paths with a potential protocol prefix that the
> block layer does not recognize as plain files seems a bit weird,
> though. Ignoring this problem does not seem too bad.)
>
> Signed-off-by: Max Reitz <mre...@redhat.com>
> ---
> tests/qemu-iotests/126 | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/tests/qemu-iotests/126 b/tests/qemu-iotests/126
> index 9b0dcf9255..8e55d7c843 100755
> --- a/tests/qemu-iotests/126
> +++ b/tests/qemu-iotests/126
> @@ -33,6 +33,12 @@ status=1 # failure is the default!
>
> # Needs backing file support
> _supported_fmt qcow qcow2 qed vmdk
> +# (1) Flat vmdk images do not support backing files
> +# (2) Split vmdk images simply fail this test right now. Fixing that
> +# is left for another day.
Which one? :)
> +_unsupported_imgopts "subformat=monolithicFlat" \
> + "subformat=twoGbMaxExtentFlat" \
> + "subformat=twoGbMaxExtentSparse"
> # This is the default protocol (and we want to test the difference between
> # colons which separate a protocol prefix from the rest and colons which are
> # just part of the filename, so we cannot test protocols which require a
> prefix)
>
What exactly fails? Does the VMDK driver see `image:` and think it's a
special filename it needs to handle and fails to do so?
