Re: [dpdk-dev] [PATCH] net/nfp: support new HW offloads API

[Alejandro Lucero <alejandro.lucero@netronome.com>]

On Tue, Mar 27, 2018 at 7:25 PM, Ferruh Yigit <ferruh.yigit@intel.com>
wrote:

> On 3/15/2018 2:30 PM, Alejandro Lucero wrote:
> > In next 18.05 the old hw offload API will be removed. This patch adds
> > support for just the new hw offload API.
> >
> > Signed-off-by: Alejandro Lucero <alejandro.lucero@netronome.com>
>
> <...>
>
> > -     if (rxmode->enable_scatter) {
> > +     if (!(rxmode->offloads & DEV_RX_OFFLOAD_CRC_STRIP))
> > +             PMD_INIT_LOG(INFO, "HW does strip CRC. No configurable!");
>
> Also this behavior is changing, not providing CRC_STRIP will be by default
> stripping [2], this check should be removed.
>
> [2]
> https://dpdk.org/ml/archives/dev/2018-March/093489.
> htmlhttps://dpdk.org/ml/archives/dev/2018-March/093489.html
>
> <...>
>
> > +     dev_conf = &dev->data->dev_conf;
> > +     rxmode = &dev_conf->rxmode;
> > +
> > +     if (rx_conf->offloads != rxmode->offloads) {
> > +             RTE_LOG(ERR, PMD, "queue %u rx offloads not as port
> offloads\n",
> > +                               queue_idx);
> > +             RTE_LOG(ERR, PMD, "\tport: %" PRIx64 "\n",
> rxmode->offloads);
> > +             RTE_LOG(ERR, PMD, "\tqueue: %" PRIx64 "\n",
> rx_conf->offloads);
> > +             return -EINVAL;
> > +     }
>
> The offload API is changing [1], this check is no more valid. If PMD is not
> supporting queue specific offload app will not send port offloads here.
>
> [1]
> https://dpdk.org/ml/archives/dev/2018-March/092978.html
>
> <...>
>
> > +     dev_conf = &dev->data->dev_conf;
> > +     txmode = &dev_conf->txmode;
> > +
> > +     if (tx_conf->offloads != txmode->offloads) {
> > +             RTE_LOG(ERR, PMD, "queue %u tx offloads not as port
> offloads",
> > +                               queue_idx);
> > +             return -EINVAL;
> > +     }
> > +
>
> Same as above.
>

Thank you for the review.

I will remove those checks and send a new path version.