On Tue, Aug 15, 2017 at 3:07 PM, Fam Zheng <f...@redhat.com> wrote: > v2: Don't leak blk->vmsh when BB is deleted before the callback is called. > [Stefan] > From stub functions, don't return g_malloc0(1) which is risky, return > NULL. > [Eric]
Thanks Fam Zheng and Kevin Wolf, retesting on -rc3 clearly got me further. Unfortunately the overall test of a libvirt controlled migration with --copy-storage-all or --copy-storage-inc still does not work. I have no good pointers yet where to look at, but wanted to give a heads up. I'm tracking what I have in [1] so far and added upstream qemu as bug task as well until we know better. [1]: https://bugs.launchpad.net/qemu/+bug/1711602