[PATCH] net/bonding: fix destroy dedicated queues flow

[PATCH] net/bonding: fix destroy dedicated queues flow

[Chaoyong He]

From: Long Wu <long.wu@corigine.com>

Bonding port in mode 4 enables dedicated queues, we
will create a flow for this feature. So we need to
destroy this flow when we remove the member port.

Furthermore if we don't destroy the flow that created
for dedicated queues when we remove the member port,
maybe we couldn't add the member port to a new bonding
port that needed dedicated queues.

We add the destroy action in removing member port
function.

Fixes: 49dad9028e2a ("net/bonding: support flow API")
Cc: matan@mellanox.com
Cc: stable@dpdk.org

Signed-off-by: Long Wu <long.wu@corigine.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@corigine.com>
Reviewed-by: Chaoyong He <chaoyong.he@corigine.com>
---
 drivers/net/bonding/rte_eth_bond_api.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/drivers/net/bonding/rte_eth_bond_api.c b/drivers/net/bonding/rte_eth_bond_api.c
index c0178369b4..85d0528b7c 100644
--- a/drivers/net/bonding/rte_eth_bond_api.c
+++ b/drivers/net/bonding/rte_eth_bond_api.c
@@ -712,6 +712,16 @@ __eth_bond_slave_remove_lock_free(uint16_t bonded_port_id,
 		}
 	}
 
+	/* Remove the dedicated queues flow */
+	if (internals->mode == BONDING_MODE_8023AD &&
+		internals->mode4.dedicated_queues.enabled == 1 &&
+		internals->mode4.dedicated_queues.flow[slave_port_id] != NULL) {
+		rte_flow_destroy(slave_port_id,
+				internals->mode4.dedicated_queues.flow[slave_port_id],
+				&flow_error);
+		internals->mode4.dedicated_queues.flow[slave_port_id] = NULL;
+	}
+
 	slave_eth_dev = &rte_eth_devices[slave_port_id];
 	slave_remove(internals, slave_eth_dev);
 	slave_eth_dev->data->dev_flags &= (~RTE_ETH_DEV_BONDED_SLAVE);
-- 
2.39.1

Re: [PATCH] net/bonding: fix destroy dedicated queues flow

[humin (Q)]

Acked-by: Min Hu (Connor) <humin29@huawei.com>

在 2023/6/8 10:59, Chaoyong He 写道:
> From: Long Wu <long.wu@corigine.com>
>
> Bonding port in mode 4 enables dedicated queues, we
> will create a flow for this feature. So we need to
> destroy this flow when we remove the member port.
>
> Furthermore if we don't destroy the flow that created
> for dedicated queues when we remove the member port,
> maybe we couldn't add the member port to a new bonding
> port that needed dedicated queues.
>
> We add the destroy action in removing member port
> function.
>
> Fixes: 49dad9028e2a ("net/bonding: support flow API")
> Cc: matan@mellanox.com
> Cc: stable@dpdk.org
>
> Signed-off-by: Long Wu <long.wu@corigine.com>
> Reviewed-by: Niklas Söderlund <niklas.soderlund@corigine.com>
> Reviewed-by: Chaoyong He <chaoyong.he@corigine.com>
> ---
>   drivers/net/bonding/rte_eth_bond_api.c | 10 ++++++++++
>   1 file changed, 10 insertions(+)
>
> diff --git a/drivers/net/bonding/rte_eth_bond_api.c b/drivers/net/bonding/rte_eth_bond_api.c
> index c0178369b4..85d0528b7c 100644
> --- a/drivers/net/bonding/rte_eth_bond_api.c
> +++ b/drivers/net/bonding/rte_eth_bond_api.c
> @@ -712,6 +712,16 @@ __eth_bond_slave_remove_lock_free(uint16_t bonded_port_id,
>   		}
>   	}
>   
> +	/* Remove the dedicated queues flow */
> +	if (internals->mode == BONDING_MODE_8023AD &&
> +		internals->mode4.dedicated_queues.enabled == 1 &&
> +		internals->mode4.dedicated_queues.flow[slave_port_id] != NULL) {
> +		rte_flow_destroy(slave_port_id,
> +				internals->mode4.dedicated_queues.flow[slave_port_id],
> +				&flow_error);
> +		internals->mode4.dedicated_queues.flow[slave_port_id] = NULL;
> +	}
> +
>   	slave_eth_dev = &rte_eth_devices[slave_port_id];
>   	slave_remove(internals, slave_eth_dev);
>   	slave_eth_dev->data->dev_flags &= (~RTE_ETH_DEV_BONDED_SLAVE);

Re: [PATCH] net/bonding: fix destroy dedicated queues flow

[Ferruh Yigit]

On 6/20/2023 4:02 AM, humin (Q) wrote:

> 在 2023/6/8 10:59, Chaoyong He 写道:
>> From: Long Wu <long.wu@corigine.com>
>>
>> Bonding port in mode 4 enables dedicated queues, we
>> will create a flow for this feature. So we need to
>> destroy this flow when we remove the member port.
>>
>> Furthermore if we don't destroy the flow that created
>> for dedicated queues when we remove the member port,
>> maybe we couldn't add the member port to a new bonding
>> port that needed dedicated queues.
>>
>> We add the destroy action in removing member port
>> function.
>>
>> Fixes: 49dad9028e2a ("net/bonding: support flow API")
>> Cc: matan@mellanox.com
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Long Wu <long.wu@corigine.com>
>> Reviewed-by: Niklas Söderlund <niklas.soderlund@corigine.com>
>> Reviewed-by: Chaoyong He <chaoyong.he@corigine.com>
>> 
>
> Acked-by: Min Hu (Connor) <humin29@huawei.com>
>

Applied to dpdk-next-net/main, thanks.