Le 09/09/2019 à 18:18, Dr. David Alan Gilbert a écrit :
> * Peter Maydell (peter.mayd...@linaro.org) wrote:
>> On Mon, 9 Sep 2019 at 17:05, Laurent Vivier <laur...@vivier.eu> wrote:
>>>
>>> Co-developed-by: Mark Cave-Ayland <mark.cave-ayl...@ilande.co.uk>
>>> Signed-off-by: Mark Cave-Ayland <mark.cave-ayl...@ilande.co.uk>
>>> Signed-off-by: Laurent Vivier <laur...@vivier.eu>
>>> Reviewed-by: Hervé Poussineau <hpous...@reactos.org>
>>
>>> +static void sysbus_swim_class_init(ObjectClass *oc, void *data)
>>> +{
>>> + DeviceClass *dc = DEVICE_CLASS(oc);
>>> +
>>> + dc->realize = sysbus_swim_realize;
>>> +}
>>
>> Missing reset and VMState for migration. These should be
>> baseline requirements for adding new device models to the
>> tree, because in an ideal world every device would support
>> both -- we should be gradually fixing the existing devices
>> which are missing these, and not letting new devices in,
>> so the situation gets gradually better, not worse.
>
> I'm happy to see things marked unmigratable instead for obscure
> devices; I don't see the need to worry about migration for stuff
> that's not going to be migrated.
>
I think in this case it's better to mark it unmigratable because the
structure will change in the future to really manage floppy and it
avoids to introduces a new version of the migration data stream.
Thanks,
Laurent
