Recompiling with these few changes helped and the networking drivers _seems_ to be working on macOS.
At least I got past the error I was getting initially.

I'm just now dealing with a slightly different set of problems related to libvirt sockets on macOS.
The Homebrew formula ships a launchd plist which doesn't seem to do socket activation, unlike systemd.
I worked around that by setting the socket dir path manually in the config:

unix_sock_dir = "/usr/local/var/run/libvirt"

which did make libvirtd create "libvirt-sock" in that directory as expected, but per docs there should also be a secondary read-only socket "libvirt-sock-ro" https://libvirt.org/daemons.html#monolithic-sockets and it's not there for some reason. Is there any additional configuration needed to activate that secondary socket?

Feel free to tell me if this is getting too off-topic btw. - The big picture is that I'm trying to get libvirt working on darwin and automate RPi virtualization via Vagrant (vagrant-libvirt) and taking it step by step.

Radek Simko


On Thu, Apr 1, 2021 at 2:54 PM Andrea Bolognani <abologna@redhat.com> wrote:
On Wed, 2021-03-31 at 21:39 +0100, Radek Simko wrote:
> I was attempting to virtualize Raspberry Pi on qemu via libvirt,
> where my host machine is macOS (both qemu and libvirt installed via
> Homebrew) and I while trying to set up the network I stumbled upon
> this:
>
> $ virsh -c qemu:///system?socket=/usr/local/var/run/libvirt/libvirt-sock net-list --all

Aside: if you have libvirt >= 6.9.0, which at this point is very
likely, you shouldn't need to include the 'socket' parameter in the
connection URI; in fact, for local connections, it should have worked
without the parameter even before that.

--
Andrea Bolognani / Red Hat / Virtualization