[v5,1/2] net/i40e: support VF request more queues
diff mbox series

Message ID 20181217111054.45414-2-zhirun.yan@intel.com
State Superseded, archived
Delegated to: Qi Zhang
Headers show
Series
  • Support request more queues
Related show

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/Intel-compilation success Compilation OK
ci/mellanox-Performance-Testing success Performance Testing PASS
ci/intel-Performance-Testing success Performance Testing PASS

Commit Message

Yan, Zhirun Dec. 17, 2018, 11:10 a.m. UTC
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(-)

Comments

Ferruh Yigit Dec. 17, 2018, 2:28 p.m. UTC | #1
On 12/17/2018 11:10 AM, Zhirun Yan wrote:
> 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.

Is this replaces "ETH_I40E_QUEUE_NUM_PER_VF_ARG (queue-num-per-vf)" devarg?
Should we remove this devarg?

> 
> Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>

<...>
Ferruh Yigit Dec. 17, 2018, 2:31 p.m. UTC | #2
On 12/17/2018 11:10 AM, Zhirun Yan wrote:
> 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>

<...>

> @@ -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 */

Don't need comment terminators for each line, can have block comment, would you
mind if I fix while merging?
Zhang, Qi Z Dec. 18, 2018, 1:44 a.m. UTC | #3
> -----Original Message-----
> From: Yigit, Ferruh
> Sent: Monday, December 17, 2018 10:28 PM
> To: Yan, Zhirun <zhirun.yan@intel.com>; dev@dpdk.org; Zhang, Qi Z
> <qi.z.zhang@intel.com>
> Cc: Wang, Haiyue <haiyue.wang@intel.com>
> Subject: Re: [dpdk-dev] [PATCH v5 1/2] net/i40e: support VF request more
> queues
> 
> On 12/17/2018 11:10 AM, Zhirun Yan wrote:
> > 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.
> 
> Is this replaces "ETH_I40E_QUEUE_NUM_PER_VF_ARG (queue-num-per-vf)"
> devarg?
> Should we remove this devarg?

queue-num-per-vf is the reserved queue number by PF for VF, it fit the situation when fixed queue number is required, so I think we still can keep it.
but I think the document could be modified properly , since queue-num-per-vf is never be the upper boundary of the queue number per vf, 
pf still can allocate max to 16 queues to exceed, we can add this explanation.


> 
> >
> > Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
> > Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> 
> <...>

Patch
diff mbox series

diff --git a/drivers/net/i40e/i40e_ethdev_vf.c b/drivers/net/i40e/i40e_ethdev_vf.c
index 05dc6596b..498e86649 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 || ret == I40EVF_MSG_CMD) {
+				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);