Re: [Qemu-devel] [PATCH v2 2/3] hw/sd: model a power-up delay, as a workaround for an EDK2 bug

[Andrew Baumann]

> From: Peter Maydell [mailto:address@hidden
> Sent: Wednesday, 23 December 2015 11:21
> On 21 December 2015 at 22:25, Andrew Baumann
> <address@hidden> wrote:
> 
> >> If you change the VMSTATE layout, you need to bump the version. As
> >> this is very common code, it may have stricter version bump
> >> requirements. Last I knew however, there was a way to add new fields
> >> at the end of VMSD without breaking backwards compatibility. Peter or
> >> Juan may know more.
> >
> > I'll admit that I didn't think about these issues when adding the
> > fields, but after a (quick) look at vmstate_save_state() and
> > vmstate_load_state(), they seem to be using named fields in a json
> > format, so I don't think the order of fields should matter if we are
> > adding new ones.
> 
> The on-the-wire format is not json, and order is important; field
> names are not sent.
> 
> If we care about migration compatibility I think the recommendation
  ^^^^^^^^^^
> is to use subsections, rather than marking fields as only existing
> in particular versions. (see docs/migration.txt for a discussion
> of subsections).

Do we care about migration compatibility for this code?

(As far as I can tell, this only matters for migration of SD interfaces across 
qemu versions, where the previous state was saved in the window between reset 
and driver initialisation.)

Andrew