diff mbox series

[v1] ena_ethdev: don't override the user provided queue length value

Message ID 20200305222333.18784-1-vladz@scylladb.com (mailing list archive)
State Changes Requested, archived
Delegated to: Ferruh Yigit
Headers
Series [v1] ena_ethdev: don't override the user provided queue length value |

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/iol-intel-Performance success Performance Testing PASS
ci/Intel-compilation success Compilation OK
ci/iol-testing success Testing PASS
ci/travis-robot success Travis build: passed
ci/iol-mellanox-Performance success Performance Testing PASS

Commit Message

Vladislav Zolotarov March 5, 2020, 10:23 p.m. UTC 
  There is a funny logic that seems to be outdated which tries to detect
a situation when a user requests a default size of the queue and
configures a device specific default value.

This seems to be not in line with the DPDK current design that expects
the user to go via the rte_eth_dev_info_get() in order to get device
specific values.

And if the user doesn't care it should pass 0 as a queue length and this
case is also being taken care of on the rte_eth level.

Signed-off-by: Vlad Zolotarov <vladz@scylladb.com>
---
 drivers/net/ena/ena_ethdev.c | 6 ------
 1 file changed, 6 deletions(-)

Comments

Vladislav Zolotarov March 5, 2020, 10:28 p.m. UTC | #1 
Igor, Evgeny,
Please, review.

On 3/5/20 5:23 PM, Vlad Zolotarov wrote:
> There is a funny logic that seems to be outdated which tries to detect
> a situation when a user requests a default size of the queue and
> configures a device specific default value.
>
> This seems to be not in line with the DPDK current design that expects
> the user to go via the rte_eth_dev_info_get() in order to get device
> specific values.
>
> And if the user doesn't care it should pass 0 as a queue length and this
> case is also being taken care of on the rte_eth level.
>
> Signed-off-by: Vlad Zolotarov <vladz@scylladb.com>
> ---
>   drivers/net/ena/ena_ethdev.c | 6 ------
>   1 file changed, 6 deletions(-)
>
> diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
> index 665afee4f..bf632dee0 100644
> --- a/drivers/net/ena/ena_ethdev.c
> +++ b/drivers/net/ena/ena_ethdev.c
> @@ -1221,9 +1221,6 @@ static int ena_tx_queue_setup(struct rte_eth_dev *dev,
>   		return -EINVAL;
>   	}
>   
> -	if (nb_desc == RTE_ETH_DEV_FALLBACK_TX_RINGSIZE)
> -		nb_desc = adapter->tx_ring_size;
> -
>   	txq->port_id = dev->data->port_id;
>   	txq->next_to_clean = 0;
>   	txq->next_to_use = 0;
> @@ -1292,9 +1289,6 @@ static int ena_rx_queue_setup(struct rte_eth_dev *dev,
>   		return ENA_COM_FAULT;
>   	}
>   
> -	if (nb_desc == RTE_ETH_DEV_FALLBACK_RX_RINGSIZE)
> -		nb_desc = adapter->rx_ring_size;
> -
>   	if (!rte_is_power_of_2(nb_desc)) {
>   		PMD_DRV_LOG(ERR,
>   			"Unsupported size of RX queue: %d is not a power of 2.\n",
Chauskin, Igor March 9, 2020, 10:07 a.m. UTC | #2 
Acked inline

From: Vladislav Zolotarov <vladz@scylladb.com>
Sent: Friday, March 6, 2020 12:28 AM
To: dev@dpdk.org; Schmeilin, Evgeny <evgenys@amazon.com>; Chauskin, Igor <igorch@amazon.com>
Subject: RE: [EXTERNAL][dpdk-dev] [PATCH v1] ena_ethdev: don't override the user provided queue length value

Igor, Evgeny,
Please, review.
On 3/5/20 5:23 PM, Vlad Zolotarov wrote:

There is a funny logic that seems to be outdated which tries to detect

a situation when a user requests a default size of the queue and

configures a device specific default value.



This seems to be not in line with the DPDK current design that expects

the user to go via the rte_eth_dev_info_get() in order to get device

specific values.



And if the user doesn't care it should pass 0 as a queue length and this

case is also being taken care of on the rte_eth level.



Signed-off-by: Vlad Zolotarov <vladz@scylladb.com><mailto:vladz@scylladb.com>

Acked-by: Igor Chauskin igorch@amazon.com<mailto:igorch@amazon.com>

---

 drivers/net/ena/ena_ethdev.c | 6 ------

 1 file changed, 6 deletions(-)



diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c

index 665afee4f..bf632dee0 100644

--- a/drivers/net/ena/ena_ethdev.c

+++ b/drivers/net/ena/ena_ethdev.c

@@ -1221,9 +1221,6 @@ static int ena_tx_queue_setup(struct rte_eth_dev *dev,

         return -EINVAL;

  }



- if (nb_desc == RTE_ETH_DEV_FALLBACK_TX_RINGSIZE)

-        nb_desc = adapter->tx_ring_size;

-

  txq->port_id = dev->data->port_id;

  txq->next_to_clean = 0;

  txq->next_to_use = 0;

@@ -1292,9 +1289,6 @@ static int ena_rx_queue_setup(struct rte_eth_dev *dev,

         return ENA_COM_FAULT;

  }



- if (nb_desc == RTE_ETH_DEV_FALLBACK_RX_RINGSIZE)

-        nb_desc = adapter->rx_ring_size;

-

  if (!rte_is_power_of_2(nb_desc)) {

         PMD_DRV_LOG(ERR,

                 "Unsupported size of RX queue: %d is not a power of 2.\n",
Ferruh Yigit March 11, 2020, 1:30 p.m. UTC | #3 
On 3/9/2020 10:07 AM, Chauskin, Igor wrote:
> Acked inline
> 
> From: Vladislav Zolotarov <vladz@scylladb.com>
> Sent: Friday, March 6, 2020 12:28 AM
> To: dev@dpdk.org; Schmeilin, Evgeny <evgenys@amazon.com>; Chauskin, Igor <igorch@amazon.com>
> Subject: RE: [EXTERNAL][dpdk-dev] [PATCH v1] ena_ethdev: don't override the user provided queue length value
> 
> Igor, Evgeny,
> Please, review.
> On 3/5/20 5:23 PM, Vlad Zolotarov wrote:
> 
> There is a funny logic that seems to be outdated which tries to detect
> 
> a situation when a user requests a default size of the queue and
> 
> configures a device specific default value.
> 
> 
> 
> This seems to be not in line with the DPDK current design that expects
> 
> the user to go via the rte_eth_dev_info_get() in order to get device
> 
> specific values.
> 
> 
> 
> And if the user doesn't care it should pass 0 as a queue length and this
> 
> case is also being taken care of on the rte_eth level.
> 
> 
> 
> Signed-off-by: Vlad Zolotarov <vladz@scylladb.com><mailto:vladz@scylladb.com>
> 
> Acked-by: Igor Chauskin igorch@amazon.com<mailto:igorch@amazon.com>
> 
> ---
> 
>  drivers/net/ena/ena_ethdev.c | 6 ------
> 
>  1 file changed, 6 deletions(-)
> 
> 
> 
> diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
> 
> index 665afee4f..bf632dee0 100644
> 
> --- a/drivers/net/ena/ena_ethdev.c
> 
> +++ b/drivers/net/ena/ena_ethdev.c
> 
> @@ -1221,9 +1221,6 @@ static int ena_tx_queue_setup(struct rte_eth_dev *dev,
> 
>          return -EINVAL;
> 
>   }
> 
> 
> 
> - if (nb_desc == RTE_ETH_DEV_FALLBACK_TX_RINGSIZE)
> 
> -        nb_desc = adapter->tx_ring_size;
> 
> -

Logic in ethdev is;

/* Use default specified by driver, if nb_tx_desc is zero */
if (nb_tx_desc == 0) {
        nb_tx_desc = dev_info.default_txportconf.ring_size;
        /* If driver default is zero, fall back on EAL default */
        if (nb_tx_desc == 0)
                nb_tx_desc = RTE_ETH_DEV_FALLBACK_TX_RINGSIZE;
}

User may provide, '0' nb_tx_desc to request driver default value. If there is no
driver default value 'RTE_ETH_DEV_FALLBACK_TX_RINGSIZE' is used.

As driver doesn't support the default value, the ring size still can be set to
'RTE_ETH_DEV_FALLBACK_TX_RINGSIZE'.
Wouldn't be good to add driver default value when removing this check?

> 
>   txq->port_id = dev->data->port_id;
> 
>   txq->next_to_clean = 0;
> 
>   txq->next_to_use = 0;
> 
> @@ -1292,9 +1289,6 @@ static int ena_rx_queue_setup(struct rte_eth_dev *dev,
> 
>          return ENA_COM_FAULT;
> 
>   }
> 
> 
> 
> - if (nb_desc == RTE_ETH_DEV_FALLBACK_RX_RINGSIZE)
> 
> -        nb_desc = adapter->rx_ring_size;
> 
> -
> 
>   if (!rte_is_power_of_2(nb_desc)) {
> 
>          PMD_DRV_LOG(ERR,
> 
>                  "Unsupported size of RX queue: %d is not a power of 2.\n",
>
diff mbox series

Patch

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 665afee4f..bf632dee0 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1221,9 +1221,6 @@  static int ena_tx_queue_setup(struct rte_eth_dev *dev,
 		return -EINVAL;
 	}
 
-	if (nb_desc == RTE_ETH_DEV_FALLBACK_TX_RINGSIZE)
-		nb_desc = adapter->tx_ring_size;
-
 	txq->port_id = dev->data->port_id;
 	txq->next_to_clean = 0;
 	txq->next_to_use = 0;
@@ -1292,9 +1289,6 @@  static int ena_rx_queue_setup(struct rte_eth_dev *dev,
 		return ENA_COM_FAULT;
 	}
 
-	if (nb_desc == RTE_ETH_DEV_FALLBACK_RX_RINGSIZE)
-		nb_desc = adapter->rx_ring_size;
-
 	if (!rte_is_power_of_2(nb_desc)) {
 		PMD_DRV_LOG(ERR,
 			"Unsupported size of RX queue: %d is not a power of 2.\n",