Re: [Qemu-devel] [PATCH] set VIRTIO_BALLOON_F_MUST_TELL_HOST unconditionally

[Anthony Liguori <aliguori@us.ibm.com>]

On 04/15/2011 09:44 AM, Dave Hansen wrote:
> There's a patch pending on LKML at the moment:
>
> 	http://lkml.org/lkml/2011/4/7/101
>
> The virtio balloon driver has a VIRTIO_BALLOON_F_MUST_TELL_HOST
> feature bit.  As of now, qemu-kvm defines the bit, but doesn't set it.
> feature bit.  Whenever the bit is set, the guest kernel must always
> tell the host before we free pages back to the allocator.  Without
> this feature, we might free a page (and have another user touch it)
> while the hypervisor is unprepared for it.
>
> But, if the bit is _not_ set, there is no obligation to reverse the
> order; we're under no obligation to do _anything_.
>
> This patch makes the "tell host first" logic the only case.  This
> should make everybody happy, and reduce the amount of untested or
> untestable code in the kernel.

It doesn't make me happy.

Why would we do this in QEMU?  This prevents the guest from doing 
ballooning reclaim during OOM.

Regards,

Anthony Liguori

> Signed-off-by: Dave Hansen<dave@linux.vnet.ibm.com>
> Cc: Amit Shah<amit.shah@redhat.com>
> Cc: Anthony Liguori<aliguori@linux.vnet.ibm.com>
> ---
>   hw/virtio-balloon.c |    4 ++++
>   1 files changed, 4 insertions(+), 0 deletions(-)
>
> diff --git a/hw/virtio-balloon.c b/hw/virtio-balloon.c
> index 1e6be18..71b0864 100644
> --- a/hw/virtio-balloon.c
> +++ b/hw/virtio-balloon.c
> @@ -198,6 +198,10 @@ static void virtio_balloon_set_config(VirtIODevice *vdev,
>   static uint32_t virtio_balloon_get_features(VirtIODevice *vdev, uint32_t f)
>   {
>       f |= (1<<  VIRTIO_BALLOON_F_STATS_VQ);
> +    /* Kernels>= 2.6.40 do this unconditionally, so set
> +     * the bit to make old kernels do the same thing
> +     */
> +    f |= (1<<  VIRTIO_BALLOON_F_MUST_TELL_HOST);
>       return f;
>   }
>