[v4,0/8] net/mlx5: consolidate Tx datapath

  This patchset introduces the new implementation of tx_burst
routine of mlx5 PMD. The existing implementation is based on the
several branches of tx_burst routines optimizied for most
common sets of Tx hardware offload, supported by Mellanox NICs.

It was not very easy to update, support and develop such kind
of code - multiple branches impose multiple points to process.
Also, many of frequently requested offload combinations are not
supported yet in the most efficient way. That leads to selecting
of not completely matching tx_burst routine and harms the performance.

The new Tx datapath:
  - introduces the unified template for tx_burst routine to generate
    on compile time the most efficient instances, tuned for specified
    sets of hardware offloads
  - the best matching instance is chosen in run-time at sending queue
    configuration
  - almost all possible sets of Tx are supported
  - the best possible method to send each packet is chosen on runtime
  - introduces new devargs for more accurate tuning of data inlining 

Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

---
v4:
  - inline first mbufs for multi-segment packets
  - revert deprecated devargs support to not break device probing
  - HW VLAN insertion from WQE bug
  - rebase

v3:
  - http://patches.dpdk.org/cover/56568/
  - add reporting max number of segments in packet
  - minor fixes with inline data length setup
  - rebase on the current repo top

v2: 
  - http://patches.dpdk.org/cover/56451/
  - minor performance issues resolved
  - typos in comments and docs

v1: 
  - http://patches.dpdk.org/patch/56091/

Viacheslav Ovsiienko (8):

*** BLURB HERE ***

Viacheslav Ovsiienko (8):
  net/mlx5: remove Tx datapath implementation
  net/mlx5: add Tx datapath related devargs
  net/mlx5: update Tx datapath definitions
  net/mlx5: extend NIC attributes query via DevX
  net/mlx5: add Tx datapath configuration and setup
  net/mlx5: introduce Tx burst routine template
  net/mlx5: implement Tx burst template
  net/mlx5: report supported max number of mbuf segments

 doc/guides/nics/mlx5.rst               |  173 +-
 doc/guides/rel_notes/release_19_08.rst |    2 +
 drivers/net/mlx5/mlx5.c                |  187 +-
 drivers/net/mlx5/mlx5.h                |   14 +-
 drivers/net/mlx5/mlx5_defs.h           |   36 +-
 drivers/net/mlx5/mlx5_devx_cmds.c      |  102 +-
 drivers/net/mlx5/mlx5_ethdev.c         |   95 +-
 drivers/net/mlx5/mlx5_prm.h            |  357 ++-
 drivers/net/mlx5/mlx5_rxtx.c           | 5338 ++++++++++++++++++++++----------
 drivers/net/mlx5/mlx5_rxtx.h           |  334 +-
 drivers/net/mlx5/mlx5_rxtx_vec.c       |  175 --
 drivers/net/mlx5/mlx5_rxtx_vec_neon.h  |  289 --
 drivers/net/mlx5/mlx5_rxtx_vec_sse.h   |  284 --
 drivers/net/mlx5/mlx5_txq.c            |  289 +-
 14 files changed, 4638 insertions(+), 3037 deletions(-)
  

Hi,

> -----Original Message-----
> From: dev <dev-bounces@dpdk.org> On Behalf Of Viacheslav Ovsiienko
> Sent: Sunday, July 21, 2019 5:25 PM
> To: dev@dpdk.org
> Cc: Yongseok Koh <yskoh@mellanox.com>
> Subject: [dpdk-dev] [PATCH v4 0/8] net/mlx5: consolidate Tx datapath
> 
> This patchset introduces the new implementation of tx_burst routine of mlx5
> PMD. The existing implementation is based on the several branches of
> tx_burst routines optimizied for most common sets of Tx hardware offload,
> supported by Mellanox NICs.
> 
> It was not very easy to update, support and develop such kind of code -
> multiple branches impose multiple points to process.
> Also, many of frequently requested offload combinations are not supported
> yet in the most efficient way. That leads to selecting of not completely
> matching tx_burst routine and harms the performance.
> 
> The new Tx datapath:
>   - introduces the unified template for tx_burst routine to generate
>     on compile time the most efficient instances, tuned for specified
>     sets of hardware offloads
>   - the best matching instance is chosen in run-time at sending queue
>     configuration
>   - almost all possible sets of Tx are supported
>   - the best possible method to send each packet is chosen on runtime
>   - introduces new devargs for more accurate tuning of data inlining
> 
> Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
> 

Series applied to next-net-mlx

Kindest regards
Raslan Darawsheh