[libvirt] [PATCH 1/1] viruuid: On POWER, qemu exports uuid as 'vm, uuid' devicetree node.

[Christian Ehrhardt]

On Mon, Apr 15, 2019 at 4:16 PM Dimitri John Ledkov <xnox at ubuntu.com> wrote:
>
> Signed-off-by: Dimitri John Ledkov <xnox at ubuntu.com>
> ---
>  src/util/viruuid.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/src/util/viruuid.c b/src/util/viruuid.c
> index 0c12ddcc3e..8553a68965 100644
> --- a/src/util/viruuid.c
> +++ b/src/util/viruuid.c
> @@ -1,3 +1,4 @@
> +
>  /*
>   * viruuid.c: helper APIs for dealing with UUIDs
>   *
> @@ -201,6 +202,7 @@ getDMISystemUUID(char *uuid, int len)
>      const char *paths[] = {
>          "/sys/devices/virtual/dmi/id/product_uuid",
>          "/sys/class/dmi/id/product_uuid",
> +        "/sys/firmware/devicetree/base/vm,uuid",

This will only be available in KVM guests, so while correct and useful
there will only help for nested cases. I was concerned about that, but
Dimitri and I discussed that on IRC and he was aware of that and the
intention is indeed just for nested.

That said while I think the overall share of systems making use of
that will be low it seems ok to me.

Thanks for also pushing a similar change to systemd [1], that way down
the road old-libvirt+new-systemd setups will still benefit as
machine-id is a fallback for the system UUID.

P.S. do you have (or need) a test build of libvirt with the change
setting the host-id accordingly to show the function here?

Reviewed-by: Christian Ehrhardt <christian.ehrhardt at canonical.com>

[1]: https://github.com/systemd/systemd/pull/12321

>          NULL
>      };
>
> --
> 2.20.1
>
> --
> libvir-list mailing list
> libvir-list at redhat.com
> https://www.redhat.com/mailman/listinfo/libvir-list



-- 
Christian Ehrhardt
Software Engineer, Ubuntu Server
Canonical Ltd