Eric Blake <ebl...@redhat.com> writes:
> Now that the QMP output visitor supports an explicit null
> output, we should utilize it to make it easier to diagnose
> the difference between a missing fdt vs. a present-but-empty
> one.
Please spell out that we change the value from {} to null.
>
> (Note that this reverts the behavior of commit ab8bf1d, taking
> us back to the behavior of commit 6c2f9a1 [which in turn
> stemmed from a crash fix in 1d10b44]; but that this time,
> the change is intentional and not an accidental side-effect.)
>
> Signed-off-by: Eric Blake <ebl...@redhat.com>
> Acked-by: David Gibson <da...@gibson.dropbear.id.au>
>
> ---
> v9: improved commit message
> v8: rebase to 'name' motion
> v7: new patch, based on discussion about spapr_drc.c
> ---
> hw/ppc/spapr_drc.c | 6 +-----
> 1 file changed, 1 insertion(+), 5 deletions(-)
>
> diff --git a/hw/ppc/spapr_drc.c b/hw/ppc/spapr_drc.c
> index ffc2cd9..831ce23 100644
> --- a/hw/ppc/spapr_drc.c
> +++ b/hw/ppc/spapr_drc.c
> @@ -259,11 +259,7 @@ static void prop_get_fdt(Object *obj, Visitor *v, const
> char *name,
> void *fdt;
>
> if (!drc->fdt) {
> - visit_start_struct(v, name, NULL, 0, &err);
> - if (!err) {
> - visit_end_struct(v, &err);
> - }
> - error_propagate(errp, err);
> + visit_type_null(v, NULL, errp);
> return;
> }
Easier to read, always welcome :)
