On Tue, Jun 7, 2022 at 7:16 PM Maciej Fijalkowski
<maciej.fijalkowski@xxxxxxxxx> wrote:
>
> Xdpxceiver run on a AF_XDP ZC enabled driver revealed a problem with XSK
> Tx batching API. There is a test that checks how invalid Tx descriptors
> are handled by AF_XDP. Each valid descriptor is followed by invalid one
> on Tx side whereas the Rx side expects only to receive a set of valid
> descriptors.
>
> In current xsk_tx_peek_release_desc_batch() function, the amount of
> available descriptors is hidden inside xskq_cons_peek_desc_batch(). This
> can be problematic in cases where invalid descriptors are present due to
> the fact that xskq_cons_peek_desc_batch() returns only a count of valid
> descriptors. This means that it is impossible to properly update XSK
> ring state when calling xskq_cons_release_n().
>
> To address this issue, pull out the contents of
> xskq_cons_peek_desc_batch() so that callers (currently only
> xsk_tx_peek_release_desc_batch()) will always be able to update the
> state of ring properly, as total count of entries is now available and
> use this value as an argument in xskq_cons_release_n(). By
> doing so, xskq_cons_peek_desc_batch() can be dropped altogether.
Thank you for catching this Maciej!
Acked-by: Magnus Karlsson <magnus.karlsson@xxxxxxxxx>
> Fixes: 9349eb3a9d2a ("xsk: Introduce batched Tx descriptor interfaces")
> Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@xxxxxxxxx>
> ---
> net/xdp/xsk.c | 5 +++--
> net/xdp/xsk_queue.h | 8 --------
> 2 files changed, 3 insertions(+), 10 deletions(-)
>
> diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
> index e0a4526ab66b..19ac872a6624 100644
> --- a/net/xdp/xsk.c
> +++ b/net/xdp/xsk.c
> @@ -373,7 +373,8 @@ u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, u32 max_entries)
> goto out;
> }
>
> - nb_pkts = xskq_cons_peek_desc_batch(xs->tx, pool, max_entries);
> + max_entries = xskq_cons_nb_entries(xs->tx, max_entries);
> + nb_pkts = xskq_cons_read_desc_batch(xs->tx, pool, max_entries);
> if (!nb_pkts) {
> xs->tx->queue_empty_descs++;
> goto out;
> @@ -389,7 +390,7 @@ u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, u32 max_entries)
> if (!nb_pkts)
> goto out;
>
> - xskq_cons_release_n(xs->tx, nb_pkts);
> + xskq_cons_release_n(xs->tx, max_entries);
> __xskq_cons_release(xs->tx);
> xs->sk.sk_write_space(&xs->sk);
>
> diff --git a/net/xdp/xsk_queue.h b/net/xdp/xsk_queue.h
> index a794410989cc..fb20bf7207cf 100644
> --- a/net/xdp/xsk_queue.h
> +++ b/net/xdp/xsk_queue.h
> @@ -282,14 +282,6 @@ static inline bool xskq_cons_peek_desc(struct xsk_queue *q,
> return xskq_cons_read_desc(q, desc, pool);
> }
>
> -static inline u32 xskq_cons_peek_desc_batch(struct xsk_queue *q, struct xsk_buff_pool *pool,
> - u32 max)
> -{
> - u32 entries = xskq_cons_nb_entries(q, max);
> -
> - return xskq_cons_read_desc_batch(q, pool, entries);
> -}
> -
> /* To improve performance in the xskq_cons_release functions, only update local state here.
> * Reflect this to global state when we get new entries from the ring in
> * xskq_cons_get_entries() and whenever Rx or Tx processing are completed in the NAPI loop.
> --
> 2.27.0
>
