[Intel-wired-lan] [PATCH bpf-next v3 4/5] xsk: introduce batched Tx descriptor interfaces
John Fastabend
john.fastabend at gmail.com
Tue Nov 17 19:07:22 UTC 2020
Magnus Karlsson wrote:
> From: Magnus Karlsson <magnus.karlsson at intel.com>
>
> Introduce batched descriptor interfaces in the xsk core code for the
> Tx path to be used in the driver to write a code path with higher
> performance. This interface will be used by the i40e driver in the
> next patch. Though other drivers would likely benefit from this new
> interface too.
>
> Note that batching is only implemented for the common case when
> there is only one socket bound to the same device and queue id. When
> this is not the case, we fall back to the old non-batched version of
> the function.
>
> Signed-off-by: Magnus Karlsson <magnus.karlsson at intel.com>
> ---
> include/net/xdp_sock_drv.h | 7 ++++
> net/xdp/xsk.c | 57 +++++++++++++++++++++++++++++
> net/xdp/xsk_queue.h | 89 +++++++++++++++++++++++++++++++++++++++-------
> 3 files changed, 140 insertions(+), 13 deletions(-)
>
Acked-by: John Fastabend <john.fastabend at gmail.com>
> +
> +u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, struct xdp_desc *descs,
> + u32 max_entries)
> +{
> + struct xdp_sock *xs;
> + u32 nb_pkts;
> +
> + rcu_read_lock();
> + if (!list_is_singular(&pool->xsk_tx_list)) {
> + /* Fallback to the non-batched version */
I'm going to ask even though I believe its correct.
If we fallback here and then an entry is added to the list while we are
in the fallback logic everything should still be OK, correct?
> + rcu_read_unlock();
> + return xsk_tx_peek_release_fallback(pool, descs, max_entries);
> + }
> +
> + xs = list_first_or_null_rcu(&pool->xsk_tx_list, struct xdp_sock, tx_list);
> + if (!xs) {
> + nb_pkts = 0;
> + goto out;
> + }
> +
> + nb_pkts = xskq_cons_peek_desc_batch(xs->tx, descs, pool, max_entries);
> + if (!nb_pkts) {
> + xs->tx->queue_empty_descs++;
> + goto out;
> + }
> +
> + /* This is the backpressure mechanism for the Tx path. Try to
> + * reserve space in the completion queue for all packets, but
> + * if there are fewer slots available, just process that many
> + * packets. This avoids having to implement any buffering in
> + * the Tx path.
> + */
> + nb_pkts = xskq_prod_reserve_addr_batch(pool->cq, descs, nb_pkts);
> + if (!nb_pkts)
> + goto out;
> +
> + xskq_cons_release_n(xs->tx, nb_pkts);
> + __xskq_cons_release(xs->tx);
> + xs->sk.sk_write_space(&xs->sk);
> +
> +out:
> + rcu_read_unlock();
> + return nb_pkts;
> +}
> +EXPORT_SYMBOL(xsk_tx_peek_release_desc_batch);
> +
More information about the Intel-wired-lan
mailing list