[v1] net/iavf: support FDIR TCP/UDP pattern without input set

  This patch adds an input set refinement function to support outer
and inner TCP/UDP patterns without input set for flow director filter.

Signed-off-by: Yuying Zhang <yuying.zhang@intel.com>
---
 drivers/net/iavf/iavf_fdir.c | 67 ++++++++++++++++++++++++++++++------
 1 file changed, 56 insertions(+), 11 deletions(-)
  

> -----Original Message-----
> From: Zhang, Yuying <yuying.zhang@intel.com>
> Sent: Tuesday, December 1, 2020 4:32 PM
> To: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Wu, Jingjing
> <jingjing.wu@intel.com>
> Cc: Zhang, Yuying <yuying.zhang@intel.com>
> Subject: [PATCH v1] net/iavf: support FDIR TCP/UDP pattern without input set
> 
> This patch adds an input set refinement function to support outer and inner
> TCP/UDP patterns without input set for flow director filter.
> 
> Signed-off-by: Yuying Zhang <yuying.zhang@intel.com>

Acked-by: Qi Zhang <qi.z.zhang@intel.com>

Applied to dpdk-next-net-intel.

Thanks
Qi
  

On 12/1/2020 8:32 AM, Yuying Zhang wrote:
> This patch adds an input set refinement function to support outer
> and inner TCP/UDP patterns without input set for flow director filter.
> 

Can you please describe more what is supported now, it can be even better to 
give same samples which patter is supported now, which was not supported before?

> Signed-off-by: Yuying Zhang <yuying.zhang@intel.com>
> ---
>   drivers/net/iavf/iavf_fdir.c | 67 ++++++++++++++++++++++++++++++------
>   1 file changed, 56 insertions(+), 11 deletions(-)
> 
> diff --git a/drivers/net/iavf/iavf_fdir.c b/drivers/net/iavf/iavf_fdir.c
> index 7054bde0b..2c57313e4 100644
> --- a/drivers/net/iavf/iavf_fdir.c
> +++ b/drivers/net/iavf/iavf_fdir.c
> @@ -448,9 +448,56 @@ iavf_fdir_parse_action(struct iavf_adapter *ad,
>   	return 0;
>   }
>   
> +static int
> +iavf_fdir_refine_input_set(const uint64_t input_set,
> +			   const uint64_t input_set_mask,
> +			   struct iavf_fdir_conf *filter)
> +{
> +	struct virtchnl_proto_hdr *hdr, *hdr_last;
> +	struct rte_flow_item_ipv4 ipv4_spec;
> +	struct rte_flow_item_ipv6 ipv6_spec;
> +	int last_layer;
> +	uint8_t proto_id;
> +
> +	if (input_set & ~input_set_mask)
> +		return -rte_errno;

What is the value of the 'rte_errno' at this stage, why returning it instead of 
setting it?

> +	else if (input_set)
> +		return 0;
> +
> +	last_layer = filter->add_fltr.rule_cfg.proto_hdrs.count - 1;
> +	hdr_last = &filter->add_fltr.rule_cfg.proto_hdrs.proto_hdr[last_layer];
> +	if (hdr_last->type == VIRTCHNL_PROTO_HDR_TCP)
> +		proto_id = 6;
> +	else if (hdr_last->type == VIRTCHNL_PROTO_HDR_UDP)
> +		proto_id = 17;
> +	else
> +		return -rte_errno;
> +
> +	hdr = &filter->add_fltr.rule_cfg.proto_hdrs.proto_hdr[last_layer - 1];

Do we know here that 'last_layer' > 0 ?

> +	switch (hdr->type) {
> +	case VIRTCHNL_PROTO_HDR_IPV4:
> +		VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr, IPV4, PROT);
> +		memset(&ipv4_spec, 0, sizeof(ipv4_spec));
> +		ipv4_spec.hdr.next_proto_id = proto_id;
> +		rte_memcpy(hdr->buffer, &ipv4_spec.hdr,
> +			   sizeof(ipv4_spec.hdr));
> +		return 0;
> +	case VIRTCHNL_PROTO_HDR_IPV6:
> +		VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr, IPV6, PROT);
> +		memset(&ipv6_spec, 0, sizeof(ipv6_spec));
> +		ipv6_spec.hdr.proto = proto_id;
> +		rte_memcpy(hdr->buffer, &ipv6_spec.hdr,
> +			   sizeof(ipv6_spec.hdr));
> +		return 0;
> +	default:
> +		return -rte_errno;
> +	}
> +}
> +

<...>