Re: [Qemu-devel] [PATCH v5 1/2] vhost user: add support of live migration

[Marc-André Lureau]

Hi Thibaut

On Mon, Aug 3, 2015 at 11:22 AM, Thibaut Collet
<address@hidden> wrote:
> Some vhost user backends are able to support live migration.
> To provide this service the following features must be added:
> 1. Add the VIRTIO_NET_F_GUEST_ANNOUNCE capability to vhost-net when netdev
>    backend is vhost-user.
> 2. Provide a nop receive callback to vhost-user.
>    This callback is used only by qemu_announce_self after a migration to send
>    fake RARP to avoid network outage for peers talking to the migrated guest.
>    - For guest with GUEST_ANNOUNCE capabilities these packets must be 
> discarded,
>      GARP are sent by the guest thanks the GUEST_ANNOUNCE.
>    - For guest without GUEST_ANNOUNCE capabilities these packets are discarded
>      too, migration termination is notified when the guest sends packets.
>
> Signed-off-by: Thibaut Collet <address@hidden>
> ---
>  hw/net/vhost_net.c |    2 ++
>  net/vhost-user.c   |   25 +++++++++++++++++++++++--
>  2 files changed, 25 insertions(+), 2 deletions(-)
>
> diff --git a/hw/net/vhost_net.c b/hw/net/vhost_net.c
> index c864237..9850520 100644
> --- a/hw/net/vhost_net.c
> +++ b/hw/net/vhost_net.c
> @@ -85,6 +85,8 @@ static const int user_feature_bits[] = {
>      VIRTIO_NET_F_CTRL_MAC_ADDR,
>      VIRTIO_NET_F_CTRL_GUEST_OFFLOADS,
>
> +    VIRTIO_NET_F_GUEST_ANNOUNCE,
> +
>      VIRTIO_NET_F_MQ,
>
>      VHOST_INVALID_FEATURE_BIT
> diff --git a/net/vhost-user.c b/net/vhost-user.c
> index 93dcecd..2290271 100644
> --- a/net/vhost-user.c
> +++ b/net/vhost-user.c
> @@ -65,6 +65,28 @@ static void vhost_user_stop(VhostUserState *s)
>      s->vhost_net = 0;
>  }
>
> +static ssize_t vhost_user_receive(NetClientState *nc, const uint8_t *buf,
> +                                  size_t size)
> +{
> +    /* A live migration is done. Display an error if the packet is not a 
> RARP.
> +     * RARP are just discarded: guest is already notified about live 
> migration
> +     * by the virtio-net NIC or by the vhost-user backend.
> +     */
> +    if (size != 60) {
> +        static int display_trace = 1;
> +
> +        if (display_trace) {
> +            fprintf(stderr,
> +                    "Vhost user expects only RARP (size 60)."
> +                    "Receives unexpected packets with size %lu\n",
> +                    size);
> +            fflush(stderr);
> +            display_trace = 0;
> +        }
> +    }
> +    return size;
> +}

This warning appears during a dummy boot with vapp: qemu-system-x86_64
-m 512 -object 
memory-backend-file,id=mem,size=512M,mem-path=/hugetlbfs/,share=on
-numa node,memdev=mem -netdev
vhost-user,id=net0,chardev=chr0,vhostforce -device
virtio-net-pci,netdev=net0 -chardev
socket,id=chr0,path=/tmp/vapp.sock.

I think silentely returning 0 (without warning) is the right way to
keep the packets queued in the ring instead. But I must say I am quite
confused as this is triggered by a VIRTIO_PCI_QUEUE_NOTIFY ioport
write, and I fail to see how vhost-user handles it then without
eventfd kick...

>  static void vhost_user_cleanup(NetClientState *nc)
>  {
>      VhostUserState *s = DO_UPCAST(VhostUserState, nc, nc);
> @@ -90,6 +112,7 @@ static bool vhost_user_has_ufo(NetClientState *nc)
>  static NetClientInfo net_vhost_user_info = {
>          .type = NET_CLIENT_OPTIONS_KIND_VHOST_USER,
>          .size = sizeof(VhostUserState),
> +        .receive = vhost_user_receive,
>          .cleanup = vhost_user_cleanup,
>          .has_vnet_hdr = vhost_user_has_vnet_hdr,
>          .has_ufo = vhost_user_has_ufo,
> @@ -143,8 +166,6 @@ static int net_vhost_user_init(NetClientState *peer, 
> const char *device,
>
>      s = DO_UPCAST(VhostUserState, nc, nc);
>
> -    /* We don't provide a receive callback */
> -    s->nc.receive_disabled = 1;
>      s->chr = chr;
>
>      qemu_chr_add_handlers(s->chr, NULL, NULL, net_vhost_user_event, s);
> --
> 1.7.10.4
>



-- 
Marc-André Lureau