This series improves mlx4 Tx performance and fix and clean some Tx code. 1. 10% MPPS improvement for 1 queue, 1 core, 64B packets, txonly mode. 2. 20% MPPS improvement for 1 queue, 1 core, 32B*4(segs) packets, txonly mode.
V2: Add missed function descriptions. Accurate descriptions. Change Tx descriptor alignment to be like Rx. Move mlx4_fill_tx_data_seg to mlx4_rxtx.c and use rte_be32_t for byte count. Change remain_size type to uin32_t. Poisoning with memset. V3: Accurate descriptions. Fix poisoning from v2. Matan Azrad (8): net/mlx4: fix Tx packet drop application report net/mlx4: remove unnecessary Tx wraparound checks net/mlx4: remove restamping from Tx error path net/mlx4: optimize Tx multi-segment case net/mlx4: merge Tx queue rings management net/mlx4: mitigate Tx send entry size calculations net/mlx4: align Tx descriptors number net/mlx4: remove Tx completion elements counter drivers/net/mlx4/mlx4_prm.h | 20 +- drivers/net/mlx4/mlx4_rxtx.c | 492 +++++++++++++++++++++---------------------- drivers/net/mlx4/mlx4_rxtx.h | 5 +- drivers/net/mlx4/mlx4_txq.c | 37 ++-- 4 files changed, 279 insertions(+), 275 deletions(-) -- 1.8.3.1