vhost: fix packed ring dequeue offloading

When vhost doing dequeue offloading, will parse ethernet and l3/l4
header of the packet. Then vhost will set corresponded value in mbuf
attributes. Thus mean offloading action should after packet data copy.

Fixes: 75ed51697820 ("vhost: add packed ring batch dequeue")
Cc: stable@dpdk.org

Signed-off-by: Marvin Liu <yong.liu@intel.com>

On 2/5/21 8:47 AM, Marvin Liu wrote:
> When vhost doing dequeue offloading, will parse ethernet and l3/l4
> header of the packet. Then vhost will set corresponded value in mbuf

corresponding*

> attributes. Thus mean offloading action should after packet data copy.

It means*

> 
> Fixes: 75ed51697820 ("vhost: add packed ring batch dequeue")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Marvin Liu <yong.liu@intel.com>
> 

Please check your git patch formatting config, diff stats are missing.

> diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
> index 730b92e478..0a7d008a91 100644
> --- a/lib/librte_vhost/virtio_net.c
> +++ b/lib/librte_vhost/virtio_net.c
> @@ -2267,7 +2267,6 @@ vhost_reserve_avail_batch_packed(struct virtio_net *dev,
>  {
>  	bool wrap = vq->avail_wrap_counter;
>  	struct vring_packed_desc *descs = vq->desc_packed;
> -	struct virtio_net_hdr *hdr;
>  	uint64_t lens[PACKED_BATCH_SIZE];
>  	uint64_t buf_lens[PACKED_BATCH_SIZE];
>  	uint32_t buf_offset = sizeof(struct virtio_net_hdr_mrg_rxbuf);
> @@ -2324,13 +2323,6 @@ vhost_reserve_avail_batch_packed(struct virtio_net *dev,
>  		ids[i] = descs[avail_idx + i].id;
>  	}
>  
> -	if (virtio_net_with_host_offload(dev)) {
> -		vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
> -			hdr = (struct virtio_net_hdr *)(desc_addrs[i]);
> -			vhost_dequeue_offload(hdr, pkts[i]);
> -		}
> -	}
> -
>  	return 0;
>  
>  free_buf:
> @@ -2348,6 +2340,7 @@ virtio_dev_tx_batch_packed(struct virtio_net *dev,
>  {
>  	uint16_t avail_idx = vq->last_avail_idx;
>  	uint32_t buf_offset = sizeof(struct virtio_net_hdr_mrg_rxbuf);
> +	struct virtio_net_hdr *hdr;
>  	uintptr_t desc_addrs[PACKED_BATCH_SIZE];
>  	uint16_t ids[PACKED_BATCH_SIZE];
>  	uint16_t i;
> @@ -2364,6 +2357,13 @@ virtio_dev_tx_batch_packed(struct virtio_net *dev,
>  			   (void *)(uintptr_t)(desc_addrs[i] + buf_offset),
>  			   pkts[i]->pkt_len);
>  
> +	if (virtio_net_with_host_offload(dev)) {
> +		vhost_for_each_try_unroll(i, 0, PACKED_BATCH_SIZE) {
> +			hdr = (struct virtio_net_hdr *)(desc_addrs[i]);
> +			vhost_dequeue_offload(hdr, pkts[i]);
> +		}
> +	}
> +
>  	if (virtio_net_is_inorder(dev))
>  		vhost_shadow_dequeue_batch_packed_inorder(vq,
>  			ids[PACKED_BATCH_SIZE - 1]);
> 

Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

I'll fixup the commit message while applying.

Thanks,
Maxime

09/02/2021 15:52, Maxime Coquelin:
> 
> On 2/5/21 8:47 AM, Marvin Liu wrote:
> > When vhost doing dequeue offloading, will parse ethernet and l3/l4
> > header of the packet. Then vhost will set corresponded value in mbuf
> 
> corresponding*
> 
> > attributes. Thus mean offloading action should after packet data copy.
> 
> It means*
> 
> > 
> > Fixes: 75ed51697820 ("vhost: add packed ring batch dequeue")
> > Cc: stable@dpdk.org
> > 
> > Signed-off-by: Marvin Liu <yong.liu@intel.com>
> 
> Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> 
> I'll fixup the commit message while applying.

Fixed english wording and applied, thanks.