Re: [libvirt] [REPOSTv2 PATCH v3 0/6] Add support for VM Generation ID (vmgenid)
[...]
>> of flags. If you fix the capability check like I'm
suggesting in 4/6 you
>> have my ACK for the whole patch set.
>
> Ironically, as part of review of v2 patch 10, I was asked to remove the
> check from qemuBuildVMGenIDCommandLine:
>
> https://www.redhat.com/archives/libvir-list/2018-April/msg02253.html
>
> Perhaps it's best to just move it from qemu_process to qemu_command -
> that just means passing @qemuCaps into qemuBuildVMGenIDCommandLine and
> using it as well as removing @priv from qemuProcessGenID.
>
> It really doesn't matter in qemu_process whether the capability exists,
> it was an optimization to put it there rather than an error path in
> qemu_command... and with that change GenIDCommandLine now could return
> -1 or 0; whereas, without the caps check it was only ever returning 0.
>
> I can repost the entire series if so desired.
I don't think it's needed. Just fix and push.
OK - thanks. The removal of the caps check will be done in patch 3 since
that's where qemuProcessGenID was introduced, I've also beefed up the
[sparse] comments to the function to be:
* If this domain is requesting to use genid, then update the GUID
* value if the VIR_QEMU_PROCESS_START_GEN_VMID flag is set. This
* flag is set on specific paths during domain start processing when
* there is the possibility that the VM is potentially re-executing
* something that has already been executed before.
FWIW: The last part of the wording is taken from Daniel's v2 response.
John