Am 20.04.2018 um 21:20 hat Peter Krempa geschrieben:
On Fri, Apr 20, 2018 at 13:55:35 +0200, Kevin Wolf wrote:
Am 19.04.2018 um 17:25 hat Peter Krempa geschrieben:
Similarly to the 'raw' case add tests for bochs, cloop, dmg, ploop, vdi vhd, and vpc. Covering all supproted non-backing formats.
Note that the JSON name for 'ploop' maps to 'parallels' and 'vhd' maps to 'vhdx'.
Your -drive lines below mention format=ploop/vhd, though. That wouldn't actually work.
So, is it something that we should actually forbid? I did not actually try all of the weird formats, so I just assumed that if we'd happily generate the -drive command line and the 'blockdev' version exists it's equivalent.
I thought it's just a mistake in the commit message and that libvirt actually does the translation described here. So it would use -drive format=parallels instead of ploop.
(Also 'vhd' as an alias for 'vhdx' is super confusing, because VHD is really the name of the format implemented by QEMU's 'vpc' driver - which is already a source of confusion on its own.)
Hmmm, maybe that is a bug in my implementation. Since libvirt has a VPC format and also a VHD format I thohght that VHD is just another name for 'vhxd'. If they are different, we maybe should forbid it altogether. I confess that I did a simple prefix match rather than any complex analysis.
Maybe double-check, but your assumption might be right. There would be little reason to have both VPC and VHD, when both are names for the same thing (except as aliases, possibly). Kevin