Re: [PATCH] vhost-user: add support for VHOST_USER_SET_STATUS

[Jason Wang <jasowang@redhat.com>]

On 2020/5/14 下午3:33, Maxime Coquelin wrote:
> It is usefull for the Vhost-user backend to know
> about about the Virtio device status updates,
> especially when the driver sets the DRIVER_OK bit.
>
> With that information, no more need to do hazardous
> assumptions on when the driver is done with the
> device configuration.
>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> ---
>
> This patch applies on top of Cindy's "vDPA support in qemu"
> series, which introduces the .vhost_set_state vhost-backend
> ops.
>
>   docs/interop/vhost-user.rst | 12 ++++++++++++
>   hw/net/vhost_net.c          | 10 +++++-----
>   hw/virtio/vhost-user.c      | 35 +++++++++++++++++++++++++++++++++++
>   3 files changed, 52 insertions(+), 5 deletions(-)
>
> diff --git a/docs/interop/vhost-user.rst b/docs/interop/vhost-user.rst
> index 3b1b6602c7..f108de7458 100644
> --- a/docs/interop/vhost-user.rst
> +++ b/docs/interop/vhost-user.rst
> @@ -815,6 +815,7 @@ Protocol features
>     #define VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD       12
>     #define VHOST_USER_PROTOCOL_F_RESET_DEVICE         13
>     #define VHOST_USER_PROTOCOL_F_INBAND_NOTIFICATIONS 14
> +  #define VHOST_USER_PROTOCOL_F_STATUS               15
>   
>   Master message types
>   --------------------
> @@ -1263,6 +1264,17 @@ Master message types
>   
>     The state.num field is currently reserved and must be set to 0.
>   
> +``VHOST_USER_SET_STATUS``
> +  :id: 36
> +  :equivalent ioctl: VHOST_VDPA_SET_STATUS
> +  :slave payload: N/A
> +  :master payload: ``u64``
> +
> +  When the ``VHOST_USER_PROTOCOL_F_STATUS`` protocol feature has been
> +  successfully negotiated, this message is submitted by the master to
> +  notify the backend with updated device status as defined in the Virtio
> +  specification.
> +
>   Slave message types
>   -------------------
>   
> diff --git a/hw/net/vhost_net.c b/hw/net/vhost_net.c
> index 463e333531..37f3156dbc 100644
> --- a/hw/net/vhost_net.c
> +++ b/hw/net/vhost_net.c
> @@ -517,10 +517,10 @@ int vhost_set_state(NetClientState *nc, int state)
>   {
>       struct vhost_net *net = get_vhost_net(nc);
>       struct vhost_dev *hdev = &net->dev;
> -    if (nc->info->type == NET_CLIENT_DRIVER_VHOST_VDPA) {
> -        if (hdev->vhost_ops->vhost_set_state) {
> -                return hdev->vhost_ops->vhost_set_state(hdev, state);
> -             }
> -        }
> +
> +    if (hdev->vhost_ops->vhost_set_state) {
> +        return hdev->vhost_ops->vhost_set_state(hdev, state);
> +    }
> +
>       return 0;
>   }
> diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
> index ec21e8fbe8..b7e52d97fc 100644
> --- a/hw/virtio/vhost-user.c
> +++ b/hw/virtio/vhost-user.c
> @@ -59,6 +59,7 @@ enum VhostUserProtocolFeature {
>       VHOST_USER_PROTOCOL_F_HOST_NOTIFIER = 11,
>       VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD = 12,
>       VHOST_USER_PROTOCOL_F_RESET_DEVICE = 13,
> +    VHOST_USER_PROTOCOL_F_STATUS = 15,
>       VHOST_USER_PROTOCOL_F_MAX
>   };
>   
> @@ -100,6 +101,7 @@ typedef enum VhostUserRequest {
>       VHOST_USER_SET_INFLIGHT_FD = 32,
>       VHOST_USER_GPU_SET_SOCKET = 33,
>       VHOST_USER_RESET_DEVICE = 34,
> +    VHOST_USER_SET_STATUS = 36,
>       VHOST_USER_MAX
>   } VhostUserRequest;
>   
> @@ -1886,6 +1888,38 @@ static int vhost_user_set_inflight_fd(struct vhost_dev *dev,
>       return 0;
>   }
>   
> +static int vhost_user_set_state(struct vhost_dev *dev, int state)
> +{
> +    bool reply_supported = virtio_has_feature(dev->protocol_features,
> +                                              VHOST_USER_PROTOCOL_F_REPLY_ACK);
> +
> +    VhostUserMsg msg = {
> +        .hdr.request = VHOST_USER_SET_STATUS,
> +        .hdr.flags = VHOST_USER_VERSION,
> +        .hdr.size = sizeof(msg.payload.u64),
> +        .payload.u64 = (uint64_t)state,
> +    };
> +
> +    if (!virtio_has_feature(dev->protocol_features,
> +                VHOST_USER_PROTOCOL_F_STATUS)) {
> +        return -1;
> +    }
> +
> +    if (reply_supported) {
> +        msg.hdr.flags |= VHOST_USER_NEED_REPLY_MASK;
> +    }
> +
> +    if (vhost_user_write(dev, &msg, NULL, 0) < 0) {
> +        return -1;
> +    }
> +
> +    if (reply_supported) {
> +        return process_message_reply(dev, &msg);
> +    }
> +
> +    return 0;
> +}


Interesting, I wonder how vm stop will be handled in this case.

In the case of vDPA kernel, we probable don't want to mirror the virtio 
device status to vdpa device status directly. Since qemu may stop 
vhost-vdpa device through e.g resting vdpa device, but in the mean time, 
guest should not detect such condition in virtio device status.

So in the new version of vDPA support, we probably need to do:

static int vhost_vdpa_set_state(struct vhost_dev *dev, bool started)
{
     if (started) {
         uint8_t status = 0;

         vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_DRIVER_OK);
         vhost_vdpa_call(dev, VHOST_VDPA_GET_STATUS, &status);

         return !(status & VIRTIO_CONFIG_S_DRIVER_OK);
     } else {
         vhost_vdpa_reset_device(dev);
         vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_ACKNOWLEDGE |
                                    VIRTIO_CONFIG_S_DRIVER);
         return 0;
     }
}

And vhost_set_state() will be called from vhost_dev_start()/stop().

Does this work for vhost-user as well?

Thanks


> +
>   bool vhost_user_init(VhostUserState *user, CharBackend *chr, Error **errp)
>   {
>       if (user->chr) {
> @@ -1947,4 +1981,5 @@ const VhostOps user_ops = {
>           .vhost_backend_mem_section_filter = vhost_user_mem_section_filter,
>           .vhost_get_inflight_fd = vhost_user_get_inflight_fd,
>           .vhost_set_inflight_fd = vhost_user_set_inflight_fd,
> +        .vhost_set_state = vhost_user_set_state,
>   };