Re: [Qemu-devel] [PATCH] migration: Correctly handle subsections with no 'needed' function

[Juan Quintela]

Peter Maydell <address@hidden> wrote:
D> On 7 August 2018 at 16:05, Juan Quintela <address@hidden> wrote:
>> Peter Maydell <address@hidden> wrote:
>>> So it's better to use a needed-always subsection, which will give
>>> the desired behaviour:
>>>  * new QEMU -> new QEMU: state is always migrated
>>>  * old QEMU -> new QEMU: migration doesn't fail, and guest
>>>    will work assuming it didn't care about this corner of the
>>>    device's functionality (the device will end up with state
>>>    as it was at reset, or possibly special-cased via
>>>    pre_load/post_load hooks for the "section not present" case)
>>
>> we break by definiton new QEMU -M <old machine type> into old QEMU.
>
> Is that supposed to work? I always thought that we never
> supported migration back to an older QEMU version like that.

It is a "best effort" case.  As David told, downstream we try very
hard.  For upstream we do when I found that, but not anywhere else.

> In any case, for all the devices here the machine types are
> not versioned.

This is part of the problem.  For some architectures, we don't even care
about migraiton.  For others, we care about migration but not
cross-version.  And yet in others we care very much (downstream
specially).

And it is not clear what devices support each of them.

Later, Juan.