[dpdk-dev] [PATCH v2 05/32] net/i40e: set TX loopback from PF

Lu, Wenzhuo wenzhuo.lu at intel.com
Tue Dec 20 09:16:33 CET 2016


Hi Ferruh,


> -----Original Message-----
> From: Yigit, Ferruh
> Sent: Wednesday, December 7, 2016 9:05 PM
> To: Lu, Wenzhuo; dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2 05/32] net/i40e: set TX loopback from PF
> 
> On 12/7/2016 3:31 AM, Wenzhuo Lu wrote:
> > Support enabling/disabling TX loopback from PF.
> > User can call the API on PF to enable/disable TX loopback for all the
> > PF and VFs.
> >
> > Signed-off-by: Wenzhuo Lu <wenzhuo.lu at intel.com>
> > ---
> >  drivers/net/i40e/i40e_ethdev.c            | 219
> ++++++++++++++++++++++++++++++
> >  drivers/net/i40e/rte_pmd_i40e.h           |  16 +++
> >  drivers/net/i40e/rte_pmd_i40e_version.map |   1 +
> >  3 files changed, 236 insertions(+)
> >
> > diff --git a/drivers/net/i40e/i40e_ethdev.c
> > b/drivers/net/i40e/i40e_ethdev.c index ec863b9..8bd0d70 100644
> > --- a/drivers/net/i40e/i40e_ethdev.c
> > +++ b/drivers/net/i40e/i40e_ethdev.c
> > @@ -9938,3 +9938,222 @@ static void i40e_set_default_mac_addr(struct
> > rte_eth_dev *dev,
> >
> >  	return ret;
> >  }
> > +
> 
> <...>
> 
> > +
> > +static int
> > +i40e_vsi_set_tx_loopback(struct i40e_vsi *vsi, uint8_t on) {
> > +	struct i40e_vsi_context ctxt;
> > +	struct i40e_hw *hw;
> > +	int ret;
> > +
> > +	hw = I40E_VSI_TO_HW(vsi);
> > +
> > +	/* Use the FW API if FW >= v5.0 */
> > +	if (hw->aq.fw_maj_ver < 5) {
> > +		PMD_INIT_LOG(ERR, "FW < v5.0, cannot enable loopback");
> > +		return -ENOTSUP;
> > +	}
> > +
> > +	/* Check if it has been already on or off */
> > +	if (vsi->info.valid_sections &
> > +		rte_cpu_to_le_16(I40E_AQ_VSI_PROP_SWITCH_VALID)) {
> > +		if (on) {
> > +			if ((vsi->info.switch_id &
> > +			     I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB) ==
> > +			    I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB)
> > +				return 0; /* already on */
> > +		} else {
> > +			if ((vsi->info.switch_id &
> > +			     I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB) == 0)
> > +				return 0; /* already off */
> > +		}
> > +	}
> > +
> > +	/* remove all the MACs first */
> > +	ret = i40e_vsi_rm_mac_filter(vsi);
> > +	if (ret)
> > +		return ret;
> > +
> > +	vsi->info.valid_sections =
> cpu_to_le16(I40E_AQ_VSI_PROP_SWITCH_VALID);
> > +	if (on)
> > +		vsi->info.switch_id |= I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB;
> > +	else
> > +		vsi->info.switch_id &= ~I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB;
> > +
> 
> ---->
> 
> > +	memset(&ctxt, 0, sizeof(ctxt));
> > +	(void)rte_memcpy(&ctxt.info, &vsi->info, sizeof(vsi->info));
> > +	ctxt.seid = vsi->seid;
> > +
> > +	ret = i40e_aq_update_vsi_params(hw, &ctxt, NULL);
> > +	if (ret != I40E_SUCCESS) {
> > +		PMD_DRV_LOG(ERR, "Failed to update VSI params");
> > +		return ret;
> > +	}
> 
> <----
> 
> This part is now duplicated in a few functions, does it make sense to make it
> separate function, in the first patch it appeared 3/32 ?
There's already a function 'i40e_aq_update_vsi_params'. The duplicate code is for preparing the parameter for the functions. It looks bad to me if we add a function for that.

> 
> > +
> > +	/* add all the MACs back */
> > +	ret = i40e_vsi_restore_mac_filter(vsi);
> > +	if (ret)
> > +		return ret;
> > +
> > +	return ret;
> > +}
> > +
> > +int
> > +rte_pmd_i40e_set_tx_loopback(uint8_t port, uint8_t on) {
> > +	struct rte_eth_dev *dev;
> > +	struct i40e_pf *pf;
> > +	struct i40e_pf_vf *vf;
> > +	struct i40e_vsi *vsi;
> > +	uint16_t vf_id;
> > +	int ret;
> > +
> > +	RTE_ETH_VALID_PORTID_OR_ERR_RET(port, -ENODEV);
> > +
> > +	dev = &rte_eth_devices[port];
> > +
> > +	pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
> > +
> > +	/* setup PF TX loopback */
> > +	vsi = pf->main_vsi;
> > +	ret = i40e_vsi_set_tx_loopback(vsi, on);
> > +	if (ret)
> > +		return ret;
> > +
> > +	/* setup TX loopback for all the VFs */
> > +	if (!pf->vfs) {
> > +		PMD_DRV_LOG(ERR, "Invalid argument.");
> > +		return -EINVAL;
> > +	}
> > +
> > +	for (vf_id = 0; vf_id < pf->vf_num; vf_id++) {
> > +		vf = &pf->vfs[vf_id];
> > +		vsi = vf->vsi;
> > +
> > +		ret = i40e_vsi_set_tx_loopback(vsi, on);
> > +		if (ret)
> > +			return ret;
> > +	}
> > +
> > +	return ret;
> > +}
> > diff --git a/drivers/net/i40e/rte_pmd_i40e.h
> > b/drivers/net/i40e/rte_pmd_i40e.h index c8736c8..3c65be4 100644
> > --- a/drivers/net/i40e/rte_pmd_i40e.h
> > +++ b/drivers/net/i40e/rte_pmd_i40e.h
> > @@ -114,4 +114,20 @@ int rte_pmd_i40e_set_vf_vlan_anti_spoof(uint8_t
> port,
> >  					uint16_t vf_id,
> >  					uint8_t on);
> >
> > +/**
> > + * Enable/Disable TX loopback on all the PF and VFs.
> > + *
> > + * @param port
> > + *    The port identifier of the Ethernet device.
> > + * @param on
> > + *    1 - Enable TX loopback.
> > + *    0 - Disable TX loopback.
> > + * @return
> > + *   - (0) if successful.
> > + *   - (-ENODEV) if *port* invalid.
> > + *   - (-EINVAL) if bad parameter.
> > + */
> > +int rte_pmd_i40e_set_tx_loopback(uint8_t port,
> > +				 uint8_t on);
> > +
> >  #endif /* _PMD_I40E_H_ */
> > diff --git a/drivers/net/i40e/rte_pmd_i40e_version.map
> > b/drivers/net/i40e/rte_pmd_i40e_version.map
> > index fff6cf9..3da04d3 100644
> > --- a/drivers/net/i40e/rte_pmd_i40e_version.map
> > +++ b/drivers/net/i40e/rte_pmd_i40e_version.map
> > @@ -9,4 +9,5 @@ DPDK_17.02 {
> >  	rte_pmd_i40e_ping_vfs;
> >  	rte_pmd_i40e_set_vf_mac_anti_spoof;
> >  	rte_pmd_i40e_set_vf_vlan_anti_spoof;
> > +	rte_pmd_i40e_set_tx_loopback;
> 
> We can add these alphabetically.
> This may not worth to create a new version of the patch itself, but can be fixed
> if a new version already required..
Yes, will do it.

> 
> >  } DPDK_2.0;
> >



More information about the dev mailing list