"Daniel P. Berrange" <berra...@redhat.com> writes: > On Fri, Dec 15, 2017 at 05:38:00PM +0100, Max Reitz wrote: > >> Image creation in qemu-system-* vs. qemu-img: >> In order to get proper introspection for qemu-img create, we need a >> QAPI schema. If we have a QAPI schema, we might as well add >> blockdev-create to QMP. >> As long as we do not have a really-none (null, void, ...) machine type >> for qemu-system-*, launching such a process just for creating an image >> will bring quite a bit of overhead (e.g. with -M none -accel qtest). >> However, as for libvirt, this is not exactly a regression since >> libvirt currently cannot create images at all (apart from implicitly >> through drive-mirror etc.). Further work on voidifying qemu-system-* >> will improve performance. > > In terms of the I/O operations involved, image creation is a already a > pretty slow process, particularly if pre-allocation is used which is > common. So even QEMU's current slow (circa 300ms) startup time is a > complete non-issue for image creation IMHO - it'll be dwarfed by the > time to actually create the image. > >> On the other side, we can also add QAPI introspection to qemu-img. >> (qemu-img already links to QAPI, so this should not be too hard.) >> qemu-img will also need command-line introspection, though. > > I figure the qapi-ificiation is the hard & time consuming bit of > work. Once that's done exporting it via both qemu-img & qemu-system* > is quite straighforward.
qemu-system-*: trivial. qemu-img, via command line: straightforward qemu-img, via QMP: more difficult, since QMP is entangled with HMP, character devices, ... If libvirt really wants to use QMP for the job, *and* doesn't want to use the qemu-system-* that's running a guest, the easy solution is running another qemu-system-* without a guest. >> Plan B: >> libvirt can use qemu-img now with the currently supported options, >> and as soon as libvirt needs anything better, we will have something >> better done. >> (Also, there is "qemu-img create -f $format -o help"! Because >> parsing help texts has worked so well in the past.) > > Regards, > Daniel