[dpdk-dev] [PATCH v5 4/8]librte_ether:add a common filter API

[Liu, Jijiang]

> -----Original Message-----
> From: Thomas Monjalon [mailto:thomas.monjalon at 6wind.com]
> Sent: Friday, October 17, 2014 4:10 AM
> To: Liu, Jijiang
> Cc: dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v5 4/8]librte_ether:add a common filter API
> 
> I don't review the common API as it should be done in an unique place and
> there are many copies in different patchsets. Let's focus on tunnels.
> 
> 2014-10-11 13:55, Jijiang Liu:
> > +/**** TUNNEL FILTER DATA DEFINATION *** */
> 
> We cannot miss this comment :)
> 
> > +#define ETH_TUNNEL_FILTER_OMAC  0x01
> > +#define ETH_TUNNEL_FILTER_OIP   0x02
> > +#define ETH_TUNNEL_FILTER_TENID 0x04
> > +#define ETH_TUNNEL_FILTER_IMAC  0x08
> > +#define ETH_TUNNEL_FILTER_IVLAN 0x10
> > +#define ETH_TUNNEL_FILTER_IIP   0x20
> > +
> > +#define RTE_TUNNEL_FLAGS_TO_QUEUE 1
> 
> These values requires some comments.
OK, add comments for these MACROs
> > +/*
> > + * Tunneled filter type
> > + */
> > +enum rte_tunnel_filter_type {
> > +	RTE_TUNNEL_FILTER_TYPE_NONE = 0,
> > +	RTE_TUNNEL_FILTER_OIP = ETH_TUNNEL_FILTER_OIP,
> > +	RTE_TUNNEL_FILTER_IMAC_IVLAN =
> > +		ETH_TUNNEL_FILTER_IMAC | ETH_TUNNEL_FILTER_IVLAN,
> > +	RTE_TUNNEL_FILTER_IMAC_IVLAN_TENID =
> > +		ETH_TUNNEL_FILTER_IMAC | ETH_TUNNEL_FILTER_IVLAN |
> > +		ETH_TUNNEL_FILTER_TENID,
> > +	RTE_TUNNEL_FILTER_IMAC_TENID =
> > +		ETH_TUNNEL_FILTER_IMAC | ETH_TUNNEL_FILTER_TENID,
> > +	RTE_TUNNEL_FILTER_IMAC = ETH_TUNNEL_FILTER_IMAC,
> > +	RTE_TUNNEL_FILTER_OMAC_TENID_IMAC =
> > +		ETH_TUNNEL_FILTER_OMAC | ETH_TUNNEL_FILTER_TENID |
> > +		ETH_TUNNEL_FILTER_IMAC,
> > +	RTE_TUNNEL_FILTER_IIP = ETH_TUNNEL_FILTER_IIP,
> > +	RTE_TUNNEL_FILTER_TYPE_MAX,
> > +};
> 
> It's absolutely impossible to understand. Keep in mind the first goal of an
> API: be used (which imply to be understood by users).
> And I really don't understand why you define values for combination of
> previous flags. Please, keep it simple.

The goal of defining values for combination of filter type in order to easily distinguish/check if the mandatory parameters are valid for a specific filter type,
for example, if the filter type is RTE_TUNNEL_FILTER_IMAC_IVLAN,  we just need to check if the inner MAC address and inner VLAN ID are valid.
To limit sanity checks to valid parameters the rte_tunnel_filter_type enumeration can be replaced/initialized by bit mask.

Furthermore, please look at i40e_tunnel_filter_param_check () function in "[PATCH v5 5/8]i40e:implement API of VxLAN packet filter in librte_pmd_i40e" patch.
static int
+i40e_tunnel_filter_param_check(struct i40e_pf *pf,
+			struct rte_eth_tunnel_filter_conf *filter) {
+    ...

+	if ((filter->filter_type & ETH_TUNNEL_FILTER_OMAC) &&
+		(is_zero_ether_addr(filter->outer_mac))) {
+		PMD_DRV_LOG(ERR, "Cannot add NULL outer MAC address\n");
+		return -EINVAL;
+	}
+
+	if ((filter->filter_type & ETH_TUNNEL_FILTER_IMAC) &&
+		(is_zero_ether_addr(filter->inner_mac))) {
+		PMD_DRV_LOG(ERR, "Cannot add NULL inner MAC address\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}

Actually, If you really don't like rte_tunnel_filter_type definition style, and I can change it.

 
> --
> Thomas