[dpdk-dev] [PATCH 4/4] ethdev: add helpers to move to the new offloads API

[Thomas Monjalon]

04/09/2017 15:25, Ananyev, Konstantin:
> Hi Shahaf,
> 
> > +/**
> > + * A conversion function from rxmode offloads API to rte_eth_rxq_conf
> > + * offloads API.
> > + */
> > +static void
> > +rte_eth_convert_rxmode_offloads(struct rte_eth_rxmode *rxmode,
> > +				struct rte_eth_rxq_conf *rxq_conf)
> > +{
> > +	if (rxmode->header_split == 1)
> > +		rxq_conf->offloads |= DEV_RX_OFFLOAD_HEADER_SPLIT;
> > +	if (rxmode->hw_ip_checksum == 1)
> > +		rxq_conf->offloads |= DEV_RX_OFFLOAD_CHECKSUM;
> > +	if (rxmode->hw_vlan_filter == 1)
> > +		rxq_conf->offloads |= DEV_RX_OFFLOAD_VLAN_FILTER;
> 
> Thinking on it a bit more:
> VLAN_FILTER is definitely one per device, as it would affect VFs also.
> At least that's what we have for Intel devices (ixgbe, i40e) right now.
> For Intel devices VLAN_STRIP is also per device and
> will also be  applied to all corresponding VFs.
> In fact, right now it is possible to query/change these 3 vlan offload flags on the fly
> (after dev_start) on  port basis by rte_eth_dev_(get|set)_vlan_offload API.
> So, I think at least these 3 flags need to be remained on a port basis.

I don't understand how it helps to be able to configure the same thing
in 2 places.
I think you are just describing a limitation of these HW: some offloads
must be the same for all queues.
It does not prevent from configuring them in the per-queue setup.

> In fact, why can't we have both per port and per queue RX offload:
> - dev_configure() will accept RX_OFFLOAD_* flags and apply them on a port basis.
> - rx_queue_setup() will also accept RX_OFFLOAD_* flags and apply them on a queue basis.
> - if particular RX_OFFLOAD flag for that device couldn't be setup on a queue basis  -
>    rx_queue_setup() will return an error.

The queue setup can work while the value is the same for every queues.

> - rte_eth_rxq_info can be extended to provide information which RX_OFFLOADs
>   can be configured on a per queue basis.

Yes the PMD should advertise its limitations like being forced to
apply the same configuration to all its queues.

> BTW - in that case we probably wouldn't need ignore flag inside rx_conf anymore.