RE: [net-next 2/2] ena: Link queues to NAPIs

From: Arinzon, David
Date: Tue Oct 01 2024 - 05:06:31 EST


> Link queues to NAPIs using the netdev-genl API so this information is
> queryable.
>
> $ ./tools/net/ynl/cli.py --spec Documentation/netlink/specs/netdev.yaml \
> --dump queue-get --json='{"ifindex": 2}'
>
> [{'id': 0, 'ifindex': 2, 'napi-id': 8201, 'type': 'rx'},
> {'id': 1, 'ifindex': 2, 'napi-id': 8202, 'type': 'rx'},
> {'id': 2, 'ifindex': 2, 'napi-id': 8203, 'type': 'rx'},
> {'id': 3, 'ifindex': 2, 'napi-id': 8204, 'type': 'rx'},
> {'id': 4, 'ifindex': 2, 'napi-id': 8205, 'type': 'rx'},
> {'id': 5, 'ifindex': 2, 'napi-id': 8206, 'type': 'rx'},
> {'id': 6, 'ifindex': 2, 'napi-id': 8207, 'type': 'rx'},
> {'id': 7, 'ifindex': 2, 'napi-id': 8208, 'type': 'rx'},
> {'id': 0, 'ifindex': 2, 'napi-id': 8201, 'type': 'tx'},
> {'id': 1, 'ifindex': 2, 'napi-id': 8202, 'type': 'tx'},
> {'id': 2, 'ifindex': 2, 'napi-id': 8203, 'type': 'tx'},
> {'id': 3, 'ifindex': 2, 'napi-id': 8204, 'type': 'tx'},
> {'id': 4, 'ifindex': 2, 'napi-id': 8205, 'type': 'tx'},
> {'id': 5, 'ifindex': 2, 'napi-id': 8206, 'type': 'tx'},
> {'id': 6, 'ifindex': 2, 'napi-id': 8207, 'type': 'tx'},
> {'id': 7, 'ifindex': 2, 'napi-id': 8208, 'type': 'tx'}]
>
> Signed-off-by: Joe Damato <jdamato@xxxxxxxxxx>
> ---
> drivers/net/ethernet/amazon/ena/ena_netdev.c | 26
> +++++++++++++++++---
> 1 file changed, 22 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/ethernet/amazon/ena/ena_netdev.c
> b/drivers/net/ethernet/amazon/ena/ena_netdev.c
> index e88de5e426ef..1c59aedaa5d5 100644
> --- a/drivers/net/ethernet/amazon/ena/ena_netdev.c
> +++ b/drivers/net/ethernet/amazon/ena/ena_netdev.c
> @@ -1821,20 +1821,38 @@ static void ena_napi_disable_in_range(struct
> ena_adapter *adapter,
> int first_index,
> int count) {
> + struct napi_struct *napi;

Is this variable necessary? It has been defined in the enable function because
it is needed in netif_queue_set_napi() API as well as in napi_enable(),
and it makes sense in order to avoid long lines
In here, the variable is only used in a call to napi_disable(), can the code
be kept as it is? don't see a need to shorten the napi_disable() call line.

> int i;
>
> - for (i = first_index; i < first_index + count; i++)
> - napi_disable(&adapter->ena_napi[i].napi);
> + for (i = first_index; i < first_index + count; i++) {
> + napi = &adapter->ena_napi[i].napi;
> + if (!ENA_IS_XDP_INDEX(adapter, i)) {
> + netif_queue_set_napi(adapter->netdev, i,
> + NETDEV_QUEUE_TYPE_TX, NULL);
> + netif_queue_set_napi(adapter->netdev, i,
> + NETDEV_QUEUE_TYPE_RX, NULL);
> + }
> + napi_disable(napi);
> + }
> }
>
> static void ena_napi_enable_in_range(struct ena_adapter *adapter,
> int first_index,
> int count) {
> + struct napi_struct *napi;
> int i;
>
> - for (i = first_index; i < first_index + count; i++)
> - napi_enable(&adapter->ena_napi[i].napi);
> + for (i = first_index; i < first_index + count; i++) {
> + napi = &adapter->ena_napi[i].napi;
> + napi_enable(napi);
> + if (!ENA_IS_XDP_INDEX(adapter, i)) {

Can you share some info on why you decided to set the queue to napi association
only in non-xdp case?
In XDP, there's a napi poll function that's executed and it runs on the TX queue.
I am assuming that it's because XDP is not yet supported in the framework? If so,
there's a need to add an explicit comment above if (!ENA_IS_XDP_INDEX(adapter, i)) {
explaining this in order to avoid confusion with the rest of the code.

> + netif_queue_set_napi(adapter->netdev, i,
> + NETDEV_QUEUE_TYPE_RX, napi);
> + netif_queue_set_napi(adapter->netdev, i,
> + NETDEV_QUEUE_TYPE_TX, napi);
> + }
> + }
> }
>
> /* Configure the Rx forwarding */
> --
> 2.43.0

Thank you for uploading this patch.