Multi-Packet Rx Queue (MPRQ a.k.a Striding RQ) can further save PCIe bandwidth by posting a single large buffer for multiple packets. Instead of posting a buffer per a packet, one large buffer is posted in order to receive multiple packets on the buffer. A MPRQ buffer consists of multiple fixed-size strides and each stride receives one packet.
Rx packet is mem-copied to a user-provided mbuf if the size of Rx packet is comparatively small, or PMD attaches the Rx packet to the mbuf by external buffer attachment - rte_pktmbuf_attach_extbuf(). A mempool for external buffers will be allocated and managed by PMD. This patchset should be applied after: net/mlx5: add new Memory Region support v2: * add mprq_log_stride_num parameter to configure the number of strides. * enable configuring larger max_rx_pkt_len than the size of user-provided mbuf. The size of the stride isn't static anymore but dynamically determined. * enhance documentation. * fix issues from review. Yongseok Koh (3): net/mlx5: separate filling Rx flags net/mlx5: add a function to rdma-core glue net/mlx5: add Multi-Packet Rx support doc/guides/nics/mlx5.rst | 58 +++++ drivers/net/mlx5/Makefile | 5 + drivers/net/mlx5/mlx5.c | 83 +++++++ drivers/net/mlx5/mlx5.h | 11 + drivers/net/mlx5/mlx5_defs.h | 18 ++ drivers/net/mlx5/mlx5_ethdev.c | 3 + drivers/net/mlx5/mlx5_glue.c | 16 ++ drivers/net/mlx5/mlx5_glue.h | 8 + drivers/net/mlx5/mlx5_prm.h | 15 ++ drivers/net/mlx5/mlx5_rxq.c | 494 ++++++++++++++++++++++++++++++++++++--- drivers/net/mlx5/mlx5_rxtx.c | 307 +++++++++++++++++++++--- drivers/net/mlx5/mlx5_rxtx.h | 37 ++- drivers/net/mlx5/mlx5_rxtx_vec.c | 4 + drivers/net/mlx5/mlx5_rxtx_vec.h | 3 +- drivers/net/mlx5/mlx5_trigger.c | 6 +- 15 files changed, 992 insertions(+), 76 deletions(-) -- 2.11.0
