On Fri, 2019-05-31 at 14:53 +0200, Paolo Abeni wrote: > We can avoid another indirect call per packet wrapping the rx > handler call with the proper helper. > > To ensure that even the last listed direct call experience > measurable gain, despite the additional conditionals we must > traverse before reaching it, I tested reversing the order of the > listed options, with performance differences below noise level. > > Together with the previous indirect call patch, this gives > ~6% performance improvement in raw UDP tput. >
Nice ! I like it. > Signed-off-by: Paolo Abeni <[email protected]> > --- > drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c > b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c > index 0fe5f13d07cc..c3752dbe00c8 100644 > --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c > @@ -1333,7 +1333,9 @@ int mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int > budget) > > mlx5_cqwq_pop(cqwq); > > - rq->handle_rx_cqe(rq, cqe); > + INDIRECT_CALL_4(rq->handle_rx_cqe, > mlx5e_handle_rx_cqe_mpwrq, > + mlx5e_handle_rx_cqe, > mlx5e_handle_rx_cqe_rep, > + mlx5e_ipsec_handle_rx_cqe, rq, cqe); you missed mlx5i_handle_rx_cqe, anyway don't add INDIRECT_CALL_5 :D just replace mlx5e_handle_rx_cqe_rep with mlx5i_handle_rx_cqe, mlx5e_handle_rx_cqe_rep is actually a slow path of switchdev mode. Maybe define the list somewhere in en.h where it is visible for every one: #define MLX5_RX_INDIRECT_CALL_LIST \ mlx5e_handle_rx_cqe_mpwrq, mlx5e_handle_rx_cqe, mlx5i_handle_rx_cqe, mlx5e_ipsec_handle_rx_cqe and here: INDIRECT_CALL_4(rq->handle_rx_cqe, MLX5_RX_INDIRECT_CALL_LIST, rq, cqe); > } while ((++work_done < budget) && (cqe = > mlx5_cqwq_get_cqe(cqwq))); > > out:
