Re: [dpdk-dev] [PATCH v4 1/2] net/i40e: support VF request more queues

["Yan, Zhirun" <zhirun.yan@intel.com>]

> -----Original Message-----
> From: Zhang, Qi Z
> Sent: Friday, December 14, 2018 8:00 PM
> To: Yan, Zhirun <zhirun.yan@intel.com>; dev@dpdk.org
> Cc: Wang, Haiyue <haiyue.wang@intel.com>
> Subject: RE: [PATCH v4 1/2] net/i40e: support VF request more queues
> 
> 
> 
> > -----Original Message-----
> > From: Yan, Zhirun
> > Sent: Friday, December 14, 2018 10:37 PM
> > To: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> > Cc: Yan, Zhirun <zhirun.yan@intel.com>; Wang, Haiyue
> > <haiyue.wang@intel.com>
> > Subject: [PATCH v4 1/2] net/i40e: support VF request more queues
> >
> > Before this patch, VF gets a default number of queues from the PF.
> > This patch enables VF to request a different number. When VF
> > configures more queues, it will send VIRTCHNL_OP_REQUEST_QUEUES to PF
> > to request more queues, if success, PF will reset the VF.
> >
> > User can run "port stop all", "port config port_id rxq/txq queue_num"
> > and "port start all" to reconfigure queue number.
> >
> > Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
> > Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> > ---
> >  drivers/net/i40e/i40e_ethdev_vf.c | 62
> > ++++++++++++++++++++++++++++++-
> >  1 file changed, 60 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/i40e/i40e_ethdev_vf.c
> > b/drivers/net/i40e/i40e_ethdev_vf.c
> > index 05dc6596b..a568fb528 100644
> > --- a/drivers/net/i40e/i40e_ethdev_vf.c
> > +++ b/drivers/net/i40e/i40e_ethdev_vf.c
> > @@ -359,6 +359,25 @@ i40evf_execute_vf_cmd(struct rte_eth_dev *dev,
> > struct vf_cmd_info *args)
> >  		} while (i++ < MAX_TRY_TIMES);
> >  		_clear_cmd(vf);
> >  		break;
> > +	case VIRTCHNL_OP_REQUEST_QUEUES:
> > +		 /*ignore async reply, only wait for system message,*/
> > +		 /*vf_reset = true if get
> VIRTCHNL_EVENT_RESET_IMPENDING,*/
> > +		 /*if not, means request queues failed */
> > +		err = -1;
> > +		do {
> > +			ret = i40evf_read_pfmsg(dev, &info);
> > +			vf->cmd_retval = info.result;
> > +			if (ret == I40EVF_MSG_SYS && vf->vf_reset) {
> > +				err = 0;
> > +				break;
> > +			} else if (ret == I40EVF_MSG_ERR) {
> 
> Base on patch 2/2, in the case some error happen for example that no more
> free queue available, I40EVF_MSG_CMD will be returned.
> I think it should be "else if (ret == I40EVF_MSG_ERR || ret ==
> I40EVF_MSG_CMD)" here So we don't need to wait until end of loop in that
> case.

Yes,  I will modify it in the next version. Thanks.

> > +				break;
> > +			}
> > +			rte_delay_ms(ASQ_DELAY_MS);
> > +			/* If don't read msg or read sys event, continue */
> > +		} while (i++ < MAX_TRY_TIMES);
> > +		_clear_cmd(vf);
> > +		break;
> >
> >  	default:
> >  		/* for other adminq in running time, waiting the cmd done flag
> */
> > @@
> > -1012,6 +1031,28 @@ i40evf_add_vlan(struct rte_eth_dev *dev, uint16_t
> vlanid)
> >  	return err;
> >  }
> >
> > +static int
> > +i40evf_request_queues(struct rte_eth_dev *dev, uint16_t num) {
> > +	struct i40e_vf *vf =
> > I40EVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
> > +	struct virtchnl_vf_res_request vfres;
> > +	struct vf_cmd_info args;
> > +	int err;
> > +
> > +	vfres.num_queue_pairs = num;
> > +
> > +	args.ops = VIRTCHNL_OP_REQUEST_QUEUES;
> > +	args.in_args = (u8 *)&vfres;
> > +	args.in_args_size = sizeof(vfres);
> > +	args.out_buffer = vf->aq_resp;
> > +	args.out_size = I40E_AQ_BUF_SZ;
> > +	err = i40evf_execute_vf_cmd(dev, &args);
> > +	if (err)
> > +		PMD_DRV_LOG(ERR, "fail to execute command
> > OP_REQUEST_QUEUES");
> > +
> > +	return err;
> > +}
> > +
> >  static int
> >  i40evf_del_vlan(struct rte_eth_dev *dev, uint16_t vlanid)  { @@
> > -1516,8
> > +1557,11 @@ RTE_PMD_REGISTER_KMOD_DEP(net_i40e_vf, "* igb_uio |
> > vfio-pci");  static int  i40evf_dev_configure(struct rte_eth_dev *dev)
> > {
> > +	struct i40e_vf *vf =
> > I40EVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
> >  	struct i40e_adapter *ad =
> >  		I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> > +	uint16_t num_queue_pairs = RTE_MAX(dev->data->nb_rx_queues,
> > +				dev->data->nb_tx_queues);
> >
> >  	/* Initialize to TRUE. If any of Rx queues doesn't meet the bulk
> >  	 * allocation or vector Rx preconditions we will reset it.
> > @@ -1527,6 +1571,20 @@ i40evf_dev_configure(struct rte_eth_dev *dev)
> >  	ad->tx_simple_allowed = true;
> >  	ad->tx_vec_allowed = true;
> >
> > +	if (num_queue_pairs > vf->vsi_res->num_queue_pairs) {
> > +		int ret = 0;
> > +
> > +		PMD_DRV_LOG(INFO, "change queue pairs from %u to %u",
> > +			    vf->vsi_res->num_queue_pairs, num_queue_pairs);
> > +		ret = i40evf_request_queues(dev, num_queue_pairs);
> > +		if (ret != 0)
> > +			return ret;
> > +
> > +		ret = i40evf_dev_reset(dev);
> > +		if (ret != 0)
> > +			return ret;
> > +	}
> > +
> >  	return i40evf_init_vlan(dev);
> >  }
> >
> > @@ -2145,8 +2203,8 @@ i40evf_dev_info_get(struct rte_eth_dev *dev,
> > struct rte_eth_dev_info *dev_info)  {
> >  	struct i40e_vf *vf =
> > I40EVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
> >
> > -	dev_info->max_rx_queues = vf->vsi_res->num_queue_pairs;
> > -	dev_info->max_tx_queues = vf->vsi_res->num_queue_pairs;
> > +	dev_info->max_rx_queues = I40E_MAX_QP_NUM_PER_VF;
> > +	dev_info->max_tx_queues = I40E_MAX_QP_NUM_PER_VF;
> >  	dev_info->min_rx_bufsize = I40E_BUF_SIZE_MIN;
> >  	dev_info->max_rx_pktlen = I40E_FRAME_SIZE_MAX;
> >  	dev_info->hash_key_size = (I40E_VFQF_HKEY_MAX_INDEX + 1) *
> > sizeof(uint32_t);
> > --
> > 2.17.1