[dpdk-dev] [PATCH v7] vhost: support virtqueue interrupt/notification suppression

[Yao, Lei A]

> -----Original Message-----
> From: Chen, Junjie J
> Sent: Tuesday, January 9, 2018 7:04 PM
> To: yliu at fridaylinux.org; maxime.coquelin at redhat.com; Wang, Xiao W
> <xiao.w.wang at intel.com>; Bie, Tiwei <tiwei.bie at intel.com>; Yao, Lei A
> <lei.a.yao at intel.com>
> Cc: dev at dpdk.org; Chen, Junjie J <junjie.j.chen at intel.com>
> Subject: [PATCH v7] vhost: support virtqueue interrupt/notification
> suppression
> 
> The driver can suppress interrupt when VIRTIO_F_EVENT_IDX feature bit is
> negotiated. The driver set vring flags to 0, and MAY use used_event in
> available ring to advise device interrupt util reach an index specified
> by used_event. The device ignore the lower bit of vring flags, and send
> an interrupt when index reach used_event.
> 
> The device can suppress notification in a manner analogous to the ways
> driver suppress interrupt. The device manipulates flags or avail_event in
> the used ring in the same way the driver manipulates flags or used_event in
> available ring.
> 
> Signed-off-by: Junjie Chen <junjie.j.chen at intel.com>
Tested-by: Lei Yao <lei.a.yao at intel.com>
VM2VM Iperf test has been executed with virtio-net driver after apply this patch.
No performance drop. 
CPU: Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz
Host OS: Ubuntu 16.04
Guest OS: Ubuntu 16.04
Kernel: 4.4.0
> ---
> v7:
> Add vhost_need_event definition and update code for next virtio.
> 
> v6:
> Use volatile qualifier to access avail event idx.
> 
> v5:
> Remove updating avail event index in backend.
> 
> v2-v4:
> Use definition of VIRTIO_F_EVENT_IDX from kernel.
> 
>  lib/librte_vhost/vhost.c      |  2 +-
>  lib/librte_vhost/vhost.h      | 44
> ++++++++++++++++++++++++++++++++++++++-----
>  lib/librte_vhost/virtio_net.c |  6 +++---
>  3 files changed, 43 insertions(+), 9 deletions(-)
> 
> diff --git a/lib/librte_vhost/vhost.c b/lib/librte_vhost/vhost.c
> index 6f7ef7f..400caa0 100644
> --- a/lib/librte_vhost/vhost.c
> +++ b/lib/librte_vhost/vhost.c
> @@ -538,7 +538,7 @@ rte_vhost_vring_call(int vid, uint16_t vring_idx)
>  	if (!vq)
>  		return -1;
> 
> -	vhost_vring_call(vq);
> +	vhost_vring_call(dev, vq);
>  	return 0;
>  }
> 
> diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
> index 1d9366e..ec79991 100644
> --- a/lib/librte_vhost/vhost.h
> +++ b/lib/librte_vhost/vhost.h
> @@ -103,6 +103,8 @@ struct vhost_virtqueue {
> 
>  	uint16_t		last_avail_idx;
>  	uint16_t		last_used_idx;
> +	/* Last used index we notify to front end. */
> +	uint16_t		signalled_used;
>  #define VIRTIO_INVALID_EVENTFD		(-1)
>  #define VIRTIO_UNINITIALIZED_EVENTFD	(-2)
> 
> @@ -214,6 +216,7 @@ struct vhost_msg {
>  				(1ULL << VIRTIO_NET_F_GUEST_TSO6) | \
>  				(1ULL << VIRTIO_NET_F_GUEST_UFO) | \
>  				(1ULL << VIRTIO_RING_F_INDIRECT_DESC) |
> \
> +				(1ULL << VIRTIO_RING_F_EVENT_IDX) | \
>  				(1ULL << VIRTIO_NET_F_MTU) | \
>  				(1ULL << VIRTIO_F_IOMMU_PLATFORM))
> 
> @@ -399,16 +402,47 @@ vhost_iova_to_vva(struct virtio_net *dev, struct
> vhost_virtqueue *vq,
>  	return __vhost_iova_to_vva(dev, vq, iova, size, perm);
>  }
> 
> +#define vhost_used_event(vr) \
> +	(*(volatile uint16_t*)&(vr)->avail->ring[(vr)->size])
> +
> +/*
> + * The following is used with VIRTIO_RING_F_EVENT_IDX.
> + * Assuming a given event_idx value from the other size, if we have
> + * just incremented index from old to new_idx, should we trigger an
> + * event?
> + */
> +static __rte_always_inline int
> +vhost_need_event(uint16_t event_idx, uint16_t new_idx, uint16_t old)
> +{
> +	return (uint16_t)(new_idx - event_idx - 1) < (uint16_t)(new_idx -
> old);
> +}
> +
>  static __rte_always_inline void
> -vhost_vring_call(struct vhost_virtqueue *vq)
> +vhost_vring_call(struct virtio_net *dev, struct vhost_virtqueue *vq)
>  {
>  	/* Flush used->idx update before we read avail->flags. */
>  	rte_mb();
> 
> -	/* Kick the guest if necessary. */
> -	if (!(vq->avail->flags & VRING_AVAIL_F_NO_INTERRUPT)
> -			&& (vq->callfd >= 0))
> -		eventfd_write(vq->callfd, (eventfd_t)1);
> +	/* Don't kick guest if we don't reach index specified by guest. */
> +	if (dev->features & (1ULL << VIRTIO_RING_F_EVENT_IDX)) {
> +		uint16_t old = vq->signalled_used;
> +		uint16_t new = vq->last_used_idx;
> +
> +		LOG_DEBUG(VHOST_DATA, "%s: used_event_idx=%d,
> old=%d, new=%d\n",
> +			__func__,
> +			vhost_used_event(vq),
> +			old, new);
> +		if (vhost_need_event(vhost_used_event(vq), new, old)
> +			&& (vq->callfd >= 0)) {
> +			vq->signalled_used = vq->last_used_idx;
> +			eventfd_write(vq->callfd, (eventfd_t) 1);
> +		}
> +	} else {
> +		/* Kick the guest if necessary. */
> +		if (!(vq->avail->flags & VRING_AVAIL_F_NO_INTERRUPT)
> +				&& (vq->callfd >= 0))
> +			eventfd_write(vq->callfd, (eventfd_t)1);
> +	}
>  }
> 
>  #endif /* _VHOST_NET_CDEV_H_ */
> diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
> index eb496e1..d0f9221 100644
> --- a/lib/librte_vhost/virtio_net.c
> +++ b/lib/librte_vhost/virtio_net.c
> @@ -413,7 +413,7 @@ virtio_dev_rx(struct virtio_net *dev, uint16_t
> queue_id,
>  		offsetof(struct vring_used, idx),
>  		sizeof(vq->used->idx));
> 
> -	vhost_vring_call(vq);
> +	vhost_vring_call(dev, vq);
>  out:
>  	if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM))
>  		vhost_user_iotlb_rd_unlock(vq);
> @@ -700,7 +700,7 @@ virtio_dev_merge_rx(struct virtio_net *dev, uint16_t
> queue_id,
> 
>  	if (likely(vq->shadow_used_idx)) {
>  		flush_shadow_used_ring(dev, vq);
> -		vhost_vring_call(vq);
> +		vhost_vring_call(dev, vq);
>  	}
> 
>  out:
> @@ -1104,7 +1104,7 @@ update_used_idx(struct virtio_net *dev, struct
> vhost_virtqueue *vq,
>  	vq->used->idx += count;
>  	vhost_log_used_vring(dev, vq, offsetof(struct vring_used, idx),
>  			sizeof(vq->used->idx));
> -	vhost_vring_call(vq);
> +	vhost_vring_call(dev, vq);
>  }
> 
>  static __rte_always_inline struct zcopy_mbuf *
> --
> 2.0.1