Re: [dpdk-dev] [PATCH v1 04/11] drivers/raw/ifpga_rawdev: add IPN3KE support for IFPGA Rawdev

["Xu, Rosen" <rosen.xu@intel.com>]

Hi,

> -----Original Message-----
> From: Yigit, Ferruh
> Sent: Wednesday, March 06, 2019 20:31
> To: Xu, Rosen <rosen.xu@intel.com>; dev@dpdk.org
> Cc: Zhang, Tianfei <tianfei.zhang@intel.com>; Wei, Dan
> <dan.wei@intel.com>; Pei, Andy <andy.pei@intel.com>; Yang, Qiming
> <qiming.yang@intel.com>; Wang, Haiyue <haiyue.wang@intel.com>; Chen,
> Santos <santos.chen@intel.com>; Zhang, Zhang <zhang.zhang@intel.com>;
> Hemant Agrawal <hemant.agrawal@nxp.com>; Shreyansh Jain
> <shreyansh.jain@nxp.com>
> Subject: Re: [PATCH v1 04/11] drivers/raw/ifpga_rawdev: add IPN3KE
> support for IFPGA Rawdev
> 
> On 2/28/2019 7:13 AM, Rosen Xu wrote:
> > Add Intel FPGA Acceleration NIC IPN3KE support for IFPGA Rawdev.
> >
> > Signed-off-by: Rosen Xu <rosen.xu@intel.com>
> > Signed-off-by: Tianfei Zhang <tianfei.zhang@intel.com>
> > Signed-off-by: Andy Pei <andy.pei@intel.com>
> <...>
> 
> > +static int
> > +ifgpa_rawdev_get_attr(struct rte_rawdev *dev,
> > +		     const char *attr_name,
> > +		     uint64_t *attr_value)
> > +{
> > +	struct ifpga_rawdev_mac_info *mac_info;
> > +	struct ifpga_rawdevg_retimer_info *retimer_info;
> > +	struct opae_retimer_info or_info;
> > +	struct opae_adapter *adapter;
> > +	struct opae_manager *mgr;
> > +	struct ifpga_rawdevg_link_info *linfo;
> > +	struct opae_retimer_status rstatus;
> > +
> > +	IFPGA_RAWDEV_PMD_FUNC_TRACE();
> > +
> > +	if (!dev || !attr_name || !attr_value) {
> > +		IFPGA_BUS_ERR("Invalid arguments for getting attributes");
> > +		return -1;
> > +	}
> > +
> > +	adapter = ifpga_rawdev_get_priv(dev);
> > +	if (!adapter)
> > +		return -1;
> > +
> > +	mgr = opae_adapter_get_mgr(adapter);
> > +	if (!mgr)
> > +		return -1;
> > +
> > +	if (!strcmp(attr_name, "retimer_info")) {
> > +		retimer_info = (struct ifpga_rawdevg_retimer_info
> *)attr_value;
> 
> How you are using the 'get_attr' & 'set_attr' is a little odd.
> 
> I think the intention with these APIs to provide a map/dictionary like data
> structure, save/get 'attr_values' by 'attr_name'.
> 
> How you are using is, instead of getting 'attr_value', you are passing valid
> pointers via 'attr_value' and make this function to fill that struct. Why you
> don't have a specific function for this purpose instead of using 'get_attr' &
> 'set_attr' ops? I believe this will be source of confusion in long term.
> 

Good suggestions, I will modify it in v2 patch set.

> > +		if (opae_manager_get_retimer_info(mgr, &or_info))
> > +			return -1;
> > +
> > +		retimer_info->retimer_num = or_info.num_retimer;
> > +		retimer_info->port_num = or_info.num_port;
> > +		switch (or_info.support_speed) {
> > +		case MXD_10GB:
> > +			retimer_info->mac_type =
> > +			IFPGA_RAWDEV_RETIMER_MAC_TYPE_10GE_XFI;
> > +			break;
> > +		case MXD_25GB:
> > +			retimer_info->mac_type =
> > +			IFPGA_RAWDEV_RETIMER_MAC_TYPE_25GE_25GAUI;
> > +			break;
> > +		case MXD_40GB:
> > +			retimer_info->mac_type =
> > +			IFPGA_RAWDEVG_RETIMER_MAC_TYPE_40GE_XLAUI;
> > +			break;
> > +		case MXD_100GB:
> > +			retimer_info->mac_type =
> > +			IFPGA_RAWDEV_RETIMER_MAC_TYPE_100GE_CAUI;
> > +			break;
> > +		default:
> > +			retimer_info->mac_type =
> > +			IFPGA_RAWDEV_RETIMER_MAC_TYPE_UNKNOWN;
> > +			break;
> > +		}
> > +		return 0;
> > +	} else if (!strcmp(attr_name, "default_mac")) {
> > +		/* not implement by MAX */
> > +		mac_info = (struct ifpga_rawdev_mac_info *)attr_value;
> > +		mac_info->addr.addr_bytes[0] = 0;
> > +		mac_info->addr.addr_bytes[1] = 0;
> > +		mac_info->addr.addr_bytes[2] = 0;
> > +		mac_info->addr.addr_bytes[3] = 0;
> > +		mac_info->addr.addr_bytes[4] = 0;
> > +		mac_info->addr.addr_bytes[5] = 0xA + mac_info->port_id;
> > +
> > +		return 0;
> > +	} else if (!strcmp(attr_name, "retimer_linkstatus")) {
> > +		linfo = (struct ifpga_rawdevg_link_info *)attr_value;
> > +		linfo->link_up = 0;
> > +		linfo->link_speed = IFPGA_RAWDEV_LINK_SPEED_UNKNOWN;
> > +
> > +		if (opae_manager_get_retimer_status(mgr, linfo->port,
> &rstatus))
> > +			return -1;
> > +
> > +		linfo->link_up = rstatus.line_link;
> > +		switch (rstatus.speed) {
> > +		case MXD_10GB:
> > +			linfo->link_speed =
> > +				IFPGA_RAWDEV_LINK_SPEED_10GB;
> > +			break;
> > +		case MXD_25GB:
> > +			linfo->link_speed =
> > +				IFPGA_RAWDEV_LINK_SPEED_25GB;
> > +			break;
> > +		case MXD_40GB:
> > +			linfo->link_speed =
> > +				IFPGA_RAWDEV_LINK_SPEED_40GB;
> > +			break;
> > +		default:
> > +			linfo->link_speed =
> > +				IFPGA_RAWDEV_LINK_SPEED_UNKNOWN;
> > +			break;
> > +		}
> > +
> > +		return 0;
> > +	} else
> > +		return -1;
> > +
> > +	/* Attribute not found */
> > +	return -1;
> > +}
> > +
> > +static int ifgpa_rawdev_set_attr(struct rte_rawdev *dev,
> > +				     const char *attr_name,
> > +				     const uint64_t attr_value)
> > +{
> > +	struct opae_adapter *adapter;
> > +	struct opae_manager *mgr;
> > +	/*struct ifpga_rawdevg_link_info *linfo;*/
> > +	/*struct opae_retimer_status rstatus;*/
> 
> Please remove commented out code.

Applied.

> > +
> > +	IFPGA_RAWDEV_PMD_FUNC_TRACE();
> > +
> > +	if (!dev || !attr_name) {
> > +		IFPGA_BUS_ERR("Invalid arguments for setting attributes");
> > +		return -1;
> > +	}
> > +
> > +	adapter = ifpga_rawdev_get_priv(dev);
> > +	if (!adapter)
> > +		return -1;
> > +
> > +	mgr = opae_adapter_get_mgr(adapter);
> > +	if (!mgr)
> > +		return -1;
> 
> nothing done with 'adapter' & 'mgr' ? Why getting them?

Currently we are not using set_attr, I will remove it.

> > +
> > +	if (!strcmp(attr_name, "retimer_linkstatus"))
> > +		printf("ifgpa_rawdev_set_attr_func %lx\n", attr_value);
> 
> %lx usage is wrong for 32bits, it is giving a build warning [1], also please don't
> use 'printf' for logging.
Fixed in v2 patch.
> And why there is a specific check for a value and log, in a generic function like
> set_attr?

Currently we are not using set_attr, I will remove it.

> [1]
> .../drivers/raw/ifpga_rawdev/ifpga_rawdev.c: In function
> ‘ifgpa_rawdev_set_attr’:
> .../drivers/raw/ifpga_rawdev/ifpga_rawdev.c:465:40: error: format ‘%lx’
> expects argument of type ‘long unsigned int’, but argument 2 has type
> ‘uint64_t’ {aka ‘const long long unsigned int’} [-We$ ror=format=]
>    printf("ifgpa_rawdev_set_attr_func %lx\n", attr_value);
>                                       ~~^     ~~~~~~~~~~
>                                       %llx

It's just for debug.
Currently we are not using set_attr, I will remove it.

> > +
> > +	return -1;
> 
> Will function always fail?
> And are you aware the you did not set anything in this function? Just ignored
> the 'attr_value'?
> 
> <...>

Currently we are not using set_attr, I will remove it.

> > @@ -419,7 +559,7 @@
> >
> >  	rawdev->dev_ops = &ifpga_rawdev_ops;
> >  	rawdev->device = &pci_dev->device;
> > -	rawdev->driver_name = pci_dev->device.driver->name;
> > +	rawdev->driver_name = pci_dev->driver->driver.name;
> 
> They are both same right?
Yes, they are same.

> <...>
> 
> > +
> > +#include <rte_ether.h>
> > +
> > +#ifndef ETH_ALEN
> > +#define ETH_ALEN	6
> > +#endif
> 
> You can use "ETHER_ADDR_LEN" instead, it is defined in above included
> 'rte_ether.h' header.

Thanks, I will replace " ETH_ALEN " with "ETHER_ADDR_LEN".