On 04/20/2017 09:05 AM, Fam Zheng wrote: > On Tue, 02/28 14:35, Peter Lieven wrote: >> img_convert has been around before there was an ImgConvertState or >> a block backend, but it has never been modified to directly use >> these structs. Change this by parsing parameters directly into >> the ImgConvertState and directly use BlockBackend where possible. >> Futhermore variable initalization has been reworked and sorted. >> >> Signed-off-by: Peter Lieven <p...@kamp.de> > > I see an iotest failure with this patch, in Kevin's block-next tree: > > 019 1s ... - output mismatch (see 019.out.bad) > --- /stor/work/qemu/tests/qemu-iotests/019.out 2017-04-17 > 16:19:56.523968474 +0800 > +++ 019.out.bad 2017-04-20 22:03:29.868216955 +0800 > @@ -1086,8 +1086,8 @@ > > Checking if backing clusters are allocated when they shouldn't > > -0/128 sectors allocated at offset 1 MiB > -0/128 sectors allocated at offset 4.001 GiB > +128/128 sectors allocated at offset 1 MiB > +128/128 sectors allocated at offset 4.001 GiB
Hmm - I wonder if my patch to make 'zero with unmap' on an image with no backing file prefer pure unallocated clusters over a reads-as-zero would make a difference. https://lists.gnu.org/archive/html/qemu-devel/2017-04/msg01728.html Testing now... -- Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
signature.asc
Description: OpenPGP digital signature