Re: [dpdk-dev] [EXT] Re: [PATCH v2 05/11] net/octeontx_ep: Add dev info get and configure

-----Original Message-----
From: Ferruh Yigit <redacted> 
Sent: Tuesday, January 26, 2021 9:01 PM
To: Pradeep Kumar Nalla <redacted>; Radha Chintakuntla <redacted>; Veerasenareddy Burru <vburru@marvell.com>
Cc: Jerin Jacob Kollanukkaran <redacted>; dev@dpdk.org; Satananda Burla <sburla@marvell.com>
Subject: [EXT] Re: [dpdk-dev] [PATCH v2 05/11] net/octeontx_ep: Add dev info get and configure

External Email

----------------------------------------------------------------------
On 1/18/2021 9:35 AM, Nalla Pradeep wrote:

Add device information get and device configure operations.

Signed-off-by: Nalla Pradeep <redacted>

<...>

+static int
+otx_ep_dev_info_get(struct rte_eth_dev *eth_dev,
+		    struct rte_eth_dev_info *devinfo) {
+	struct otx_ep_device *otx_epvf;
+	struct rte_pci_device *pdev;
+	uint32_t dev_id;
+
+	otx_epvf = OTX_EP_DEV(eth_dev);
+	pdev = otx_epvf->pdev;
+	dev_id = pdev->id.device_id;
+
+	devinfo->speed_capa = ETH_LINK_SPEED_10G;

Is it always 10G independent from the device type and PHY?

Right now it is fixed for 10G. When communication infrastructure is incorporated with PF, will get information from PF.

Also can you please update the featutes file, octeontx_ep.ini, as the same patch features are >added.
For example the 'Speed capabilities' feature can be set as P (partially
supported) with this patch.

+	devinfo->max_rx_queues = otx_epvf->max_rx_queues;
+	devinfo->max_tx_queues = otx_epvf->max_tx_queues;
+
+	devinfo->min_rx_bufsize = OTX_EP_MIN_RX_BUF_SIZE;
+	if (dev_id == PCI_DEVID_OCTEONTX_EP_VF ||
+	    dev_id == PCI_DEVID_OCTEONTX2_EP_NET_VF ||
+	    dev_id == PCI_DEVID_CN98XX_EP_NET_VF) {
+		devinfo->max_rx_pktlen = OTX_EP_MAX_PKT_SZ;
+		devinfo->rx_offload_capa = DEV_RX_OFFLOAD_JUMBO_FRAME;
+		devinfo->rx_offload_capa |= DEV_RX_OFFLOAD_SCATTER;
+		devinfo->tx_offload_capa = DEV_TX_OFFLOAD_MULTI_SEGS;
+	}

The above list seems all the devices this driver probes, is there a reason to check for them?

<...>

+static int
+otx_ep_dev_configure(struct rte_eth_dev *eth_dev) {
+	struct otx_ep_device *otx_epvf = OTX_EP_DEV(eth_dev);
+	struct rte_eth_dev_data *data = eth_dev->data;
+	struct rte_eth_rxmode *rxmode;
+	struct rte_eth_txmode *txmode;
+	struct rte_eth_conf *conf;
+	uint32_t ethdev_queues;
+
+	conf = &data->dev_conf;
+	rxmode = &conf->rxmode;
+	txmode = &conf->txmode;
+	ethdev_queues = (uint32_t)(otx_epvf->sriov_info.rings_per_vf);
+	if (eth_dev->data->nb_rx_queues > ethdev_queues ||
+	    eth_dev->data->nb_tx_queues > ethdev_queues) {

In dev_info, 'max_rx_queues' & 'max_tx_queues' set by 'otx_epvf->max_rx_queues' 
& 'otx_epvf->max_tx_queues' respectively, but here they are checked against 'otx_epvf->sriov_info.rings_per_vf' are they same values? Or should dev_info chaged to set correct limits?