[Qemu-devel] Re: [PATCH-RFC 01/13] virtio: export virtqueue structure

["Michael S. Tsirkin" <mst@redhat.com>]

On Tue, Jan 12, 2010 at 04:32:44PM -0600, Anthony Liguori wrote:
> On 01/11/2010 11:16 AM, Michael S. Tsirkin wrote:
>> vhost needs physical addresses for
>> ring so expose that structure.
>>
>> Signed-off-by: Michael S. Tsirkin<mst@redhat.com>
>>    
>
> I think accessor functions might make more sense.
>
> Regards,
>
> Anthony Liguori

Well, take a look:

typedef struct VRing
{
    unsigned int num;
    target_phys_addr_t desc;
    target_phys_addr_t avail;
    target_phys_addr_t used;
} VRing;

struct VirtQueue
{
    VRing vring;
    target_phys_addr_t pa;
    uint16_t last_avail_idx;
    int inuse;
    uint16_t vector;
    void (*handle_output)(VirtIODevice *vdev, VirtQueue *vq);
    VirtIODevice *vdev;
    EventNotifier guest_notifier;
    EventNotifier host_notifier;
}

(Notifiers are added by patches I will shortly post).
We need at least:

    unsigned int num;
    target_phys_addr_t desc;
    target_phys_addr_t avail;
    target_phys_addr_t used;
    VRing vring;
    target_phys_addr_t pa;
    uint16_t vector;
    VirtIODevice *vdev;
    EventNotifier guest_notifier;
    EventNotifier host_notifier;

We do not need:
    uint16_t last_avail_idx;
    int inuse;
    void (*handle_output)(VirtIODevice *vdev, VirtQueue *vq);

(last_avail_idx will be needed if we ever want to
 move a running guest from kernel to userspace
 or back).
IOW, most of VirtQueue needs to be exposed.
So - do we really want accessors?

-- 
MST