Quoting Peter Maydell (2017-03-30 10:13:17) > On 23 March 2017 at 19:59, Michael Roth <mdr...@linux.vnet.ibm.com> wrote: > > Quoting Peter Maydell (2017-03-23 10:13:13) > >> Hi; I find that qga-test fails on Linux/sparc host: > > > I think the code is assuming that it'll be the virtual block device > > itself that's mounted, rather than a partition within it. This is > > certainly a false assumption, so we shouldn't be making this a > > fatal error. > > > > It does make me wonder if we *should* be looking at the parent to > > get the slave/device information though, which in turn means we > > might want logic to traverse up the whole tree for nested cases. > > But the GuestDiskAddress field of GuestFilesystemInfo is documented > > as possibly being empty for certain disk types, so I think for now > > we can simply relax the error. Will send a patch for 2.9. > > Did you find time to write a patch for this yet? > It would be nice to be able to add the sparc machine to the > build test set for 2.9.
Sorry, the patch was trivial, but I'd gotten distracted before I had a chance to test. I think I've recreated the scenario with loopback devices though, but you might want to give it a run just to double-check (I've cc'd you on the patch). Either way, I will send a pull by EOD. > > thanks > -- PMM >
