[PATCHv3] virtio-serial-bus: use bh for unthrottling

Instead of calling flush_queued_data when unthrottling, schedule
a bh. That way we can return immediately to the caller, and the
flush uses the same call path as a have_data for callbackee.

No migration change is required because bh are called from vm_stop.
---
 hw/virtio-serial-bus.c |   12 ++++++++++--
 hw/virtio-serial.h     |    5 +++++
 2 files changed, 15 insertions(+), 2 deletions(-)

On (Fri) 29 Apr 2011 [14:25:06], Alon Levy wrote:
> Instead of calling flush_queued_data when unthrottling, schedule
> a bh. That way we can return immediately to the caller, and the
> flush uses the same call path as a have_data for callbackee.
> 
> No migration change is required because bh are called from vm_stop.

This one looks OK; I'll put it through some testing.

However, I'd like confirmation that accessing (and modifying) vq
elements is fine in bh context.  Stefan / Michael?

> ---
>  hw/virtio-serial-bus.c |   12 ++++++++++--
>  hw/virtio-serial.h     |    5 +++++
>  2 files changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/hw/virtio-serial-bus.c b/hw/virtio-serial-bus.c
> index f10d48f..ca0581b 100644
> --- a/hw/virtio-serial-bus.c
> +++ b/hw/virtio-serial-bus.c
> @@ -285,6 +285,13 @@ size_t virtio_serial_guest_ready(VirtIOSerialPort *port)
>      return 0;
>  }
>  
> +static void flush_queued_data_bh(void *opaque)
> +{
> +    VirtIOSerialPort *port = opaque;
> +
> +    flush_queued_data(port);
> +}
> +
>  void virtio_serial_throttle_port(VirtIOSerialPort *port, bool throttle)
>  {
>      if (!port) {
> @@ -295,8 +302,7 @@ void virtio_serial_throttle_port(VirtIOSerialPort *port, bool throttle)
>      if (throttle) {
>          return;
>      }
> -
> -    flush_queued_data(port);
> +    qemu_bh_schedule(port->bh);
>  }
>  
>  /* Guest wants to notify us of some event */
> @@ -726,6 +732,7 @@ static int virtser_port_qdev_init(DeviceState *qdev, DeviceInfo *base)
>      bool plugging_port0;
>  
>      port->vser = bus->vser;
> +    port->bh = qemu_bh_new(flush_queued_data_bh, port);
>  
>      /*
>       * Is the first console port we're seeing? If so, put it up at
> @@ -792,6 +799,7 @@ static int virtser_port_qdev_exit(DeviceState *qdev)
>      VirtIOSerialPort *port = DO_UPCAST(VirtIOSerialPort, dev, qdev);
>      VirtIOSerial *vser = port->vser;
>  
> +    qemu_bh_delete(port->bh);
>      remove_port(port->vser, port->id);
>  
>      QTAILQ_REMOVE(&vser->ports, port, next);
> diff --git a/hw/virtio-serial.h b/hw/virtio-serial.h
> index 5eb948e..0fa03d1 100644
> --- a/hw/virtio-serial.h
> +++ b/hw/virtio-serial.h
> @@ -119,6 +119,11 @@ struct VirtIOSerialPort {
>      uint32_t iov_idx;
>      uint64_t iov_offset;
>  
> +    /*
> +     * When unthrottling we use a buttomhalf to call flush_queued_data.
> +     */
> +    QEMUBH *bh;
> +
>      /* Identify if this is a port that binds with hvc in the guest */
>      uint8_t is_console;
>  
> -- 
> 1.7.5
> 

		Amit

On (Fri) 29 Apr 2011 [14:25:06], Alon Levy wrote:
> Instead of calling flush_queued_data when unthrottling, schedule
> a bh. That way we can return immediately to the caller, and the
> flush uses the same call path as a have_data for callbackee.
> 
> No migration change is required because bh are called from vm_stop.

No sign-off?

The initial tests look good, no fireworks yet.

		Amit

On Fri, Apr 29, 2011 at 06:26:49PM +0530, Amit Shah wrote:
> On (Fri) 29 Apr 2011 [14:25:06], Alon Levy wrote:
> > Instead of calling flush_queued_data when unthrottling, schedule
> > a bh. That way we can return immediately to the caller, and the
> > flush uses the same call path as a have_data for callbackee.
> > 
> > No migration change is required because bh are called from vm_stop.
> 
> No sign-off?
> 
> The initial tests look good, no fireworks yet.
> 
> 		Amit

Can I reply with a sign off?

Signed-off-by: Alon Levy <alevy@redhat.com>

On (Sat) 30 Apr 2011 [11:23:42], Alon Levy wrote:
> On Fri, Apr 29, 2011 at 06:26:49PM +0530, Amit Shah wrote:
> > On (Fri) 29 Apr 2011 [14:25:06], Alon Levy wrote:
> > > Instead of calling flush_queued_data when unthrottling, schedule
> > > a bh. That way we can return immediately to the caller, and the
> > > flush uses the same call path as a have_data for callbackee.
> > > 
> > > No migration change is required because bh are called from vm_stop.
> > 
> > No sign-off?
> > 
> > The initial tests look good, no fireworks yet.
> > 
> > 		Amit
> 
> Can I reply with a sign off?
> 
> Signed-off-by: Alon Levy <alevy@redhat.com>

Yes, this is fine.

		Amit