Re: [PATCH 2/4] migration: Introduce VMStateOffset

On Thu, Jul 30, 2026 at 10:35:41AM -0400, Michael S. Tsirkin wrote:
> > diff --git a/migration/savevm.c b/migration/savevm.c
> > index 34dd06f9f73..a272bcfd0b4 100644
> > --- a/migration/savevm.c
> > +++ b/migration/savevm.c
> > @@ -873,14 +873,14 @@ static void vmstate_check(const VMStateDescription *vmsd)
> >                       * Size must be provided because dest QEMU needs that
> >                       * info to know what to allocate
> >                       */
> > -                    assert(field->size || field->size_offset);
> > +                    assert(field->size || field->size_offset.size != 0);
> 
> Why is field->size checked by coersing it to bool, but
> field->size_offset.size - by comparison to 0?
> field->size != 0 for consistency?

Indeed it might be better to add a comment, but for another reason: IIUC
old code was wrong.. size_offset can be 0 if the size field is the 1st
element.. so this may have fixed a bug.

-- 
Peter Xu