Although there's no functional need for them I considered keeping it for
maintainability reasons:
1. Keep the call flow more aligned with pcap (file) that has separated
open_rx_pcap()/open_tx_pcap()
2. If in future there'll rise a need for different functionality between rx and
tx then it will be a good place to hook it in.
e.g. if we'll want to force PCAP_D_IN/OUT for rx_iface/tx_iface as you
suggested then it can go in those functions
(I'd like to go back to that direction issue after this fix is done)
-----Original Message-----
From: Ferruh Yigit <ferruh.yi...@intel.com>
Sent: Wednesday, June 20, 2018 8:41 PM
To: Ido Goshen <i...@cgstowernetworks.com>
Cc: dev@dpdk.org
Subject: Re: [PATCH v2 2/2] net/pcap: duplicate code consolidation
On 6/19/2018 3:37 PM, ido goshen wrote:
> Signed-off-by: ido goshen <i...@cgstowernetworks.com>
<...>
> /*
> + * Opens a NIC for reading packets from it */ static inline int
> +open_rx_iface(const char *key, const char *value, void *extra_args) {
> + return open_iface(key, value, extra_args); }
> +
> +/*
> * Opens a NIC for writing packets to it
> */
> static int
> open_tx_iface(const char *key, const char *value, void *extra_args)
> {
> - const char *iface = value;
> - struct pmd_devargs *tx = extra_args;
> - pcap_t *pcap;
> -
> - if (tx->num_of_queue >= RTE_PMD_PCAP_MAX_QUEUES)
> - return -1;
> - if (open_single_iface(iface, &pcap) < 0)
> - return -1;
> - tx->queue[tx->num_of_queue].pcap = pcap;
> - tx->queue[tx->num_of_queue].name = iface;
> - tx->queue[tx->num_of_queue].type = key;
> - tx->num_of_queue++;
> -
> - return 0;
> + return open_iface(key, value, extra_args);
> }
>
> static struct rte_vdev_driver pmd_pcap_drv;
>
Is there a reason to keep open_tx_iface() and open_rx_iface(), they both are
wrapper to open_iface().
Why not use open_iface() directly as callback function?
