On 3/30/22 10:59, Andrea Bolognani wrote:
On Mon, Mar 28, 2022 at 02:38:38PM -0600, Jim Fehlig wrote:
> A downstream packaging bug resulted in a scenario where no aarch64 emulator
> binary was installed on a kvm host. Querying capabilities on the host
> succeeds and the capabilities correctly report that no <guest>'s are
> supported, but the following error is logged
>
> libvirtd: Cannot check QEMU binary /usr/libexec/qemu-kvm: No such file or directory
>
> This error is confusing and not very helpful. Additionally, comments in the
> associated code note that /usr/libexec/qemu-kvm is disto-specific, which
> suggests the logic is better suited for a downstream patch. Removing the
> check for /usr/libexec/qemu-kvm leaves virQEMUCapsGetDefaultEmulator() as
> nothing more than a needless wrapper around virQEMUCapsFindBinaryForArch.
> Drop virQEMUCapsGetDefaultEmulator() and call virQEMUCapsFindBinaryForArch()
> directly in its place, which squelches the unhelpful error.
I agree that the message being logged is not very useful, but I don't
think the approach you take here is the correct one: we want upstream
libvirt to work out of the box when built on a variety of distros,
including RHEL and derivatives, and your patch breaks that.
Do the minimum RHEL and derivatives supported upstream still provide
/usr/libexec/qemu-kvm?
Regards,
Jim