On Tue, 2024-01-23 at 21:09 +0100, Gerhard Engleder wrote:
> The fill ring of the XDP socket may contain not enough buffers to
> completey fill the RX queue during socket creation. In this case the
> flag XDP_RING_NEED_WAKEUP is not set as this flag is only set if the RX
> queue is not completely filled during polling.
>
> Set XDP_RING_NEED_WAKEUP flag also if RX queue is not completely filled
> during XDP socket creation.
>
> Fixes: 3fc2333933fd ("tsnep: Add XDP socket zero-copy RX support")
> Signed-off-by: Gerhard Engleder <gerhard@xxxxxxxxxxxxxxxxxxxxx>
> ---
> drivers/net/ethernet/engleder/tsnep_main.c | 13 +++++++++++++
> 1 file changed, 13 insertions(+)
>
> diff --git a/drivers/net/ethernet/engleder/tsnep_main.c b/drivers/net/ethernet/engleder/tsnep_main.c
> index d380a407e175..ae0b8b37b9bf 100644
> --- a/drivers/net/ethernet/engleder/tsnep_main.c
> +++ b/drivers/net/ethernet/engleder/tsnep_main.c
> @@ -1764,6 +1764,19 @@ static void tsnep_rx_reopen_xsk(struct tsnep_rx *rx)
> allocated--;
> }
> }
> +
> + /* set need wakeup flag immediately if ring is not filled completely,
> + * first polling would be too late as need wakeup signalisation would
> + * be delayed for an indefinite time
> + */
> + if (xsk_uses_need_wakeup(rx->xsk_pool)) {
> + int desc_available = tsnep_rx_desc_available(rx);
> +
> + if (desc_available)
> + xsk_set_rx_need_wakeup(rx->xsk_pool);
> + else
> + xsk_clear_rx_need_wakeup(rx->xsk_pool);
> + }
> }
>
> static bool tsnep_pending(struct tsnep_queue *queue)
The patch LGTM, but there is a very similar chunk of code in
tsnep_rx_poll_zc(). You should consider a net-next follow-up
consolidating the code in a common helper.
Cheers,
Paolo
