Re: [PATCH v2 3/4] hw/display/vga-mmio: QOM'ify vga_mmio_init() as TYPE_VGA_MMIO

[Philippe Mathieu-Daudé]

On 12/14/21 13:52, Daniel P. Berrangé wrote:
> On Tue, Dec 14, 2021 at 01:05:29PM +0100, Juan Quintela wrote:
>> Thomas Huth <thuth@redhat.com> wrote:
>>> On 06/12/2021 23.45, Philippe Mathieu-Daudé wrote:
>>>> Introduce TYPE_VGA_MMIO, a sysbus device.
>>>> While there is no change in the vga_mmio_init()
>>>> interface, this is a migration compatibility break
>>>> of the MIPS Acer Pica 61 Jazz machine (pica61).
>>>
>>> It's unfortunate, but as far as I know, it would be pretty difficult
>>> or even impossible to get this done without versioned machine types?
>>> So IMHO it's ok to break this in this case here.
>>
>> Hi
>>
>> My understanding is that outside of x86*, and now ppc, arm and s390,
>> no one else really cares about migration compatibility.  I am not even
>> sure if they really care about migration at all O:-)
>>
>> So, if the code is better for other reasons, I will not wonder about
>> migration compatibility.
> 
> Essentially if it has versioned machine types, then migration ABI
> compat is mandatory. If it doesn't have versioned machine types
> then migration API compat explicitly doesn't exist.
> 
> There are no versioned machine types for MIPS, so migration compat
> is a non-issue.

Thank you, this is a clear explanation worth being added on
docs/devel/migration.rst FAQ ("General advice for device developers"
section).