Re: [PATCH qemu-server v4 06/17] cpu: register standard option for CPU format

[Arthur Bied-Charreton <a.bied-charreton@proxmox.com>]

On Thu, May 07, 2026 at 04:08:27PM +0200, Fiona Ebner wrote:
> Am 07.05.26 um 3:59 PM schrieb Arthur Bied-Charreton:
> > On Thu, May 07, 2026 at 02:11:13PM +0200, Fiona Ebner wrote:
> >> Am 30.04.26 um 6:00 PM schrieb Arthur Bied-Charreton:
> >>> Expose $cpu_fmt as a standard option to allow using it in endpoint
> >>> definitions.
> >>>
> >>> Signed-off-by: Arthur Bied-Charreton <a.bied-charreton@proxmox.com>
> >>> ---
> >>>  src/PVE/QemuServer/CPUConfig.pm | 1 +
> >>>  1 file changed, 1 insertion(+)
> >>>
> >>> diff --git a/src/PVE/QemuServer/CPUConfig.pm b/src/PVE/QemuServer/CPUConfig.pm
> >>> index 65ced94f..78603acc 100644
> >>> --- a/src/PVE/QemuServer/CPUConfig.pm
> >>> +++ b/src/PVE/QemuServer/CPUConfig.pm
> >>> @@ -344,6 +344,7 @@ my $cpu_fmt = {
> >>>          optional => 1,
> >>>      },
> >>>  };
> >>> +PVE::JSONSchema::register_standard_option('pve-qemu-cpu', $cpu_fmt);
> >>
> >> I wanted to say: all the others for QEMU use pve-qm-*, so we should
> >> stick to that. But actually, there already is pve-qemu-machine as an
> >> exception, so meh. Might still be better to stick to the more common prefix.
> >>
> > I agree, the problem is just that QemuServer.pm:855 already registers
> > pve-qm-cpu:
> > 
> > ```
> > while (my ($k, $v) = each %$confdesc) {
> >     PVE::JSONSchema::register_standard_option("pve-qm-$k", $v);
> > }
> > ```
> 
> Aha!
> 
> > 
> > AFAICT, no other part of the code depends on that. We could just remove
> > that, or maybe append some prefix to the new pve-qm-cpu, like
> > pve-qm-cpu-conf?
> 
> What do you think about 'pve-qm-custom-cpu-model'? Hopefully reduces the
> confusion.
> 
Yes, sounds good :)
> > 
> >>>  
> >>>  my $sev_fmt = {
> >>>      type => {
> >>
>