Re: [PATCH 3/3] hw/acpi/cpu_hotplug: Convert 'Object *device' -> 'DeviceState *parent'

[Philippe Mathieu-Daudé]

On 28/2/23 23:09, Michael S. Tsirkin wrote:
> On Tue, Feb 28, 2023 at 11:05:24PM +0100, Philippe Mathieu-Daudé wrote:
> > On 28/2/23 22:41, Michael S. Tsirkin wrote:
> > > On Fri, Feb 03, 2023 at 05:30:21PM +0100, Philippe Mathieu-Daudé wrote:
> > > > ACPI CPU hotplug parent can't be any QOM object, it must be a QDev.
> > > > Convert AcpiCpuHotplug::device field as QDev to enforce this.
> > > > Rename 'owner' and 'device' variables as 'parent'.
> > > >
> > > > Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> > >
> > >
> > > So instead of plain gpe->device we now have OBJECT all over
> > > the place. Why is this an improvement?
> >
> >  From QOM PoV, in this prototype change:
> >
> >    void legacy_acpi_cpu_hotplug_init(MemoryRegion *container,
> > - Object *owner,
> > + DeviceState *parent,
> >
> > we promote the parent/owner argument from generic Object to
> > Device, which is more restrictive.
> >
> > Thus now you can not pass any QOM object, it has to be a QDev.
>
> None of the functions really seem to care what kind of object
> they get. Anything that results in less casting is a win
> in my book. More casting - a loss.

OK we are discussing QOM design then. I'll go back to this
in a future RFC. Sorry for having wasted your time meanwhile.