[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <8bb40f98-2f1f-c331-23d4-ed94a6a1ce76@redhat.com>
Date: Tue, 5 Apr 2022 15:04:17 +0200
From: Jesper Dangaard Brouer <jbrouer@...hat.com>
To: Maciej Fijalkowski <maciej.fijalkowski@...el.com>,
bpf@...r.kernel.org, ast@...nel.org, daniel@...earbox.net,
magnus.karlsson@...el.com, bjorn@...nel.org
Cc: brouer@...hat.com, netdev@...r.kernel.org, maximmi@...dia.com,
alexandr.lobakin@...el.com,
Toke Hoiland Jorgensen <toke@...hat.com>
Subject: Re: [PATCH bpf-next 04/10] i40e: xsk: terminate NAPI when XSK Rx
queue gets full
On 05/04/2022 13.06, Maciej Fijalkowski wrote:
> Correlate -ENOBUFS that was returned from xdp_do_redirect() with a XSK
> Rx queue being full. In such case, terminate the softirq processing and
> let the user space to consume descriptors from XSK Rx queue so that
> there is room that driver can use later on.
>
> Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
> ---
> .../ethernet/intel/i40e/i40e_txrx_common.h | 1 +
> drivers/net/ethernet/intel/i40e/i40e_xsk.c | 21 ++++++++++++-------
> 2 files changed, 15 insertions(+), 7 deletions(-)
>
[...]
I noticed you are only doing this for the Zero-Copy variants.
Wouldn't this also be a benefit for normal AF_XDP ?
> diff --git a/drivers/net/ethernet/intel/i40e/i40e_xsk.c b/drivers/net/ethernet/intel/i40e/i40e_xsk.c
> index c1d25b0b0ca2..9f9e4ce9a24d 100644
> --- a/drivers/net/ethernet/intel/i40e/i40e_xsk.c
> +++ b/drivers/net/ethernet/intel/i40e/i40e_xsk.c
> @@ -161,9 +161,10 @@ static int i40e_run_xdp_zc(struct i40e_ring *rx_ring, struct xdp_buff *xdp)
>
> if (likely(act == XDP_REDIRECT)) {
> err = xdp_do_redirect(rx_ring->netdev, xdp, xdp_prog);
> - if (err)
> - goto out_failure;
> - return I40E_XDP_REDIR;
> + if (!err)
> + return I40E_XDP_REDIR;
> + result = (err == -ENOBUFS) ? I40E_XDP_EXIT : I40E_XDP_CONSUMED;
> + goto out_failure;
> }
>
> switch (act) {
> @@ -175,6 +176,9 @@ static int i40e_run_xdp_zc(struct i40e_ring *rx_ring, struct xdp_buff *xdp)
> if (result == I40E_XDP_CONSUMED)
> goto out_failure;
> break;
> + case XDP_DROP:
> + result = I40E_XDP_CONSUMED;
> + break;
> default:
> bpf_warn_invalid_xdp_action(rx_ring->netdev, xdp_prog, act);
> fallthrough;
> @@ -182,9 +186,6 @@ static int i40e_run_xdp_zc(struct i40e_ring *rx_ring, struct xdp_buff *xdp)
> out_failure:
> trace_xdp_exception(rx_ring->netdev, xdp_prog, act);
> fallthrough; /* handle aborts by dropping packet */
> - case XDP_DROP:
> - result = I40E_XDP_CONSUMED;
> - break;
> }
> return result;
> }
> @@ -370,6 +371,12 @@ int i40e_clean_rx_irq_zc(struct i40e_ring *rx_ring, int budget)
> xsk_buff_dma_sync_for_cpu(bi, rx_ring->xsk_pool);
>
> xdp_res = i40e_run_xdp_zc(rx_ring, bi);
> + if (xdp_res == I40E_XDP_EXIT) {
> + failure = true;
> + xsk_buff_free(bi);
> + next_to_clean = (next_to_clean + 1) & count_mask;
> + break;
> + }
> i40e_handle_xdp_result_zc(rx_ring, bi, rx_desc, &rx_packets,
> &rx_bytes, size, xdp_res);
> total_rx_packets += rx_packets;
> @@ -382,7 +389,7 @@ int i40e_clean_rx_irq_zc(struct i40e_ring *rx_ring, int budget)
> cleaned_count = (next_to_clean - rx_ring->next_to_use - 1) & count_mask;
>
> if (cleaned_count >= I40E_RX_BUFFER_WRITE)
> - failure = !i40e_alloc_rx_buffers_zc(rx_ring, cleaned_count);
> + failure |= !i40e_alloc_rx_buffers_zc(rx_ring, cleaned_count);
>
> i40e_finalize_xdp_rx(rx_ring, xdp_xmit);
> i40e_update_rx_stats(rx_ring, total_rx_bytes, total_rx_packets);
>
Powered by blists - more mailing lists