On Thu, Apr 19, 2018 at 11:48:39PM +0800, Xueming Li wrote:
> Add supported RSS hash function check in device configuration to
> have better error verbosity for application developers.
> 
> Signed-off-by: Xueming Li <xuemi...@mellanox.com>

In general, I do not like API/wrappers generating log messages on their own,
as those can't be caught and interpreted by applications. I believe error
codes should be explicit enough. However in this specific instance,
rte_eth_dev_configure() already does it a lot, so I guess it's fine:

Acked-by: Adrien Mazarguil <adrien.mazarg...@6wind.com>

One nit below.

> ---
>  lib/librte_ether/rte_ethdev.c | 23 +++++++++++++++++++++++
>  1 file changed, 23 insertions(+)
> 
> diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
> index 3c049ef43..4ac0fadae 100644
> --- a/lib/librte_ether/rte_ethdev.c
> +++ b/lib/librte_ether/rte_ethdev.c
> @@ -1179,6 +1179,18 @@ rte_eth_dev_configure(uint16_t port_id, uint16_t 
> nb_rx_q, uint16_t nb_tx_q,
>                                                       ETHER_MAX_LEN;
>       }
>  
> +     /* Check that device supports requested rss hash functions. */
> +     if ((dev_info.flow_type_rss_offloads |
> +          dev_conf->rx_adv_conf.rss_conf.rss_hf) !=
> +         dev_info.flow_type_rss_offloads) {
> +             RTE_PMD_DEBUG_TRACE("ethdev port_id=%d invalid rss_hf: "
> +                                 "0x%"PRIx64", valid value: 0x%"PRIx64"\n",
> +                                 port_id,
> +                                 dev_conf->rx_adv_conf.rss_conf.rss_hf,
> +                                 dev_info.flow_type_rss_offloads);
> +             return -EINVAL;
> +     }
> +
>       /*
>        * Setup new number of RX/TX queues and reconfigure device.
>        */
> @@ -2835,9 +2847,20 @@ rte_eth_dev_rss_hash_update(uint16_t port_id,
>                           struct rte_eth_rss_conf *rss_conf)
>  {
>       struct rte_eth_dev *dev;
> +     struct rte_eth_dev_info dev_info = {0};

A "{0}" initializer for structures is nonstandard and may trigger compiler
warnings. I encourage you to initialize one field C99-style instead:

 struct rte_eth_dev_info dev_info = { .device = NULL };

> 
>       RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
>       dev = &rte_eth_devices[port_id];
> +     rte_eth_dev_info_get(port_id, &dev_info);
> +     if ((dev_info.flow_type_rss_offloads | rss_conf->rss_hf) !=
> +         dev_info.flow_type_rss_offloads) {
> +             RTE_PMD_DEBUG_TRACE("ethdev port_id=%d invalid rss_hf: "
> +                                 "0x%"PRIx64", valid value: 0x%"PRIx64"\n",
> +                                 port_id,
> +                                 rss_conf->rss_hf,
> +                                 dev_info.flow_type_rss_offloads);
> +             return -EINVAL;
> +     }
>       RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rss_hash_update, -ENOTSUP);
>       return eth_err(port_id, (*dev->dev_ops->rss_hash_update)(dev,
>                                                                rss_conf));
> -- 
> 2.13.3
> 

-- 
Adrien Mazarguil
6WIND

Reply via email to