On 19/12/22 14:31, Jonah Palmer wrote:
On 12/13/22 06:17, Philippe Mathieu-Daudé wrote:
The monitor decoders are the only functions using the CONFIG_xxx
definitions declared in the target specific CONFIG_DEVICES header.
Signed-off-by: Philippe Mathieu-Daudé<phi...@linaro.org>
---
hw/virtio/meson.build | 2 +-
hw/virtio/virtio-qmp.c | 659 +++++++++++++++++++++++++++++++++++++++++
hw/virtio/virtio-qmp.h | 20 ++
hw/virtio/virtio.c | 635 +--------------------------------------
4 files changed, 682 insertions(+), 634 deletions(-)
create mode 100644 hw/virtio/virtio-qmp.c
create mode 100644 hw/virtio/virtio-qmp.h
I haven't tried this myself, but is there a reason why we're not also including
the 'qmp_decode_vring_desc_flags' function here?
Because we use VirtQueue as an opaque forward-declared structure.
The structure is declared in hw/virtio/virtio.c, which is where
the internal fields are accessed.
If I move qmp_x_query_virtio_queue_status() and
qmp_x_query_virtio_queue_element(), which calls
qmp_decode_vring_desc_flags(), I get:
../hw/virtio/virtio-qmp.c:874:35: error: subscript of pointer to
incomplete type 'VirtQueue' (aka 'struct VirtQueue')
status->queue_index = vdev->vq[queue].queue_index;
~~~~~~~~^
include/hw/virtio/virtio.h:39:8: note: forward declaration of 'struct
VirtQueue'
struct VirtQueue;
^
Now you are right the QOM-generic functions (qmp_x_query_virtio,
qmp_x_query_virtio_status and qmp_x_query_virtio_vhost_queue_status)
can be moved.
Thanks for your review,
Phil.
