common/mlx5: fix completion queue entry size configuration
authorMichael Baum <michaelba@nvidia.com>
Wed, 6 Jan 2021 08:19:23 +0000 (08:19 +0000)
committerFerruh Yigit <ferruh.yigit@intel.com>
Thu, 14 Jan 2021 09:12:36 +0000 (10:12 +0100)
commita2521c8f983727449bb776133e98bd4ed7534a47
tree1b1161dc76d5f7fc495fb00e607584584d166178
parent19e13263ed2fc47a32e9ec3ddcfdbf44168d72ac
common/mlx5: fix completion queue entry size configuration

According to the current data-path implementation in the PMD the CQE
size must follow the cache-line size.
So, the configuration of the CQE size should be depended in
RTE_CACHE_LINE_SIZE.

Wrongly, part of the CQE creations didn't follow it exactly what caused
an incompatibility between HW and SW in the data-path when working in
128B cache-line size systems.

Adjust the rule for any CQE creation.
Remove the cqe_size attribute from the DevX CQ creation command and set
it inside the command translation according to the cache-line size.

Fixes: 79a7e409a2f6 ("common/mlx5: prepare support of packet pacing")
Fixes: 5cd0a83f413e ("common/mlx5: support more fields in DevX CQ create")
Cc: stable@dpdk.org
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
drivers/common/mlx5/mlx5_devx_cmds.c
drivers/common/mlx5/mlx5_devx_cmds.h
drivers/net/mlx5/mlx5_devx.c
drivers/net/mlx5/mlx5_txpp.c