Re: [Qemu-devel] [PATCH] virtio: Fix setting up host notifiers for vhost

[Marc-André Lureau]

Hi

On Wed, Jun 29, 2016 at 2:17 PM, Cornelia Huck <address@hidden> wrote:
> When setting up host notifiers, virtio_bus_set_host_notifier()
> simply switches the handler. This will only work, however, if
> the ioeventfd has already been setup; this is true for dataplane,
> but not for vhost.
>
> Fix this by starting the ioeventfd if that has not happened
> before.
>
> While we're at it, also fixup the unsetting path of
> set_host_notifier_internal().
>
> Fixes: 6798e245a3 ("virtio-bus: common ioeventfd infrastructure")
> Reported-by: Jason Wang <address@hidden>
> Reported-by: Marc-André Lureau <address@hidden>
> Signed-off-by: Cornelia Huck <address@hidden>
> ---
>
> This fixes the vhost regression for me, while dataplane continues
> to work.
>

That doesn't work here,
QTEST_QEMU_BINARY=x86_64-softmmu/qemu-system-x86_64  tests/vhost-user-test

hangs in /x86_64/vhost-user/migrate

> Peter, does this help with your iSCSI regression?
>
> ---
>  hw/virtio/virtio-bus.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/hw/virtio/virtio-bus.c b/hw/virtio/virtio-bus.c
> index 1313760..0136242 100644
> --- a/hw/virtio/virtio-bus.c
> +++ b/hw/virtio/virtio-bus.c
> @@ -176,8 +176,8 @@ static int set_host_notifier_internal(DeviceState *proxy, 
> VirtioBusState *bus,
>              return r;
>          }
>      } else {
> -        virtio_queue_set_host_notifier_fd_handler(vq, false, false);
>          k->ioeventfd_assign(proxy, notifier, n, assign);
> +        virtio_queue_set_host_notifier_fd_handler(vq, false, false);
>          event_notifier_cleanup(notifier);
>      }
>      return r;
> @@ -258,6 +258,9 @@ int virtio_bus_set_host_notifier(VirtioBusState *bus, int 
> n, bool assign)
>          return -ENOSYS;
>      }
>      if (assign) {
> +        if (!k->ioeventfd_started(proxy)) {
> +            virtio_bus_start_ioeventfd(bus);
> +        }
>          /*
>           * Stop using the generic ioeventfd, we are doing eventfd handling
>           * ourselves below
> --
> 2.6.6
>



-- 
Marc-André Lureau