Re: [Qemu-devel] [PATCH 12/11] virtio-pci: avoid extra pointer dereferences on fast path

[Frederic Konrad <fred.konrad@greensocs.com>]

Hi Paolo,

This looks ok, but I don't find the branch where it applies?

Seems it's already fixed on master.

Thanks,
Fred

On 14/10/2013 18:23, Paolo Bonzini wrote:
> We do not need to access vdev on the MSI-X fast path of virtio_pci_notify.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
> 	It is possible to get rid of BusChild altogether, but even then
> 	this would be one less pointer dereference, and it's a simpler
> 	patch.  So let's do this instead for 1.7.
>
>   hw/virtio/virtio-pci.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 1a179ea..a191c24 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -113,12 +113,13 @@ static inline VirtIOPCIProxy *to_virtio_pci_proxy_fast(DeviceState *d)
>   static void virtio_pci_notify(DeviceState *d, uint16_t vector)
>   {
>       VirtIOPCIProxy *proxy = to_virtio_pci_proxy_fast(d);
> -    VirtIODevice *vdev = virtio_bus_get_device(&proxy->bus);
>   
>       if (msix_enabled(&proxy->pci_dev))
>           msix_notify(&proxy->pci_dev, vector);
> -    else
> +    else {
> +        VirtIODevice *vdev = virtio_bus_get_device(&proxy->bus);
>           qemu_set_irq(proxy->pci_dev.irq[0], vdev->isr & 1);
> +    }
>   }
>   
>   static void virtio_pci_save_config(DeviceState *d, QEMUFile *f)