[rdma-core,0/2] CQ ignore overrun

Message

Yishai Hadas Aug. 14, 2018, 3:09 p.m. UTC

This series enables creating a CQ which doesn't pass to error state if overrun,
CQE always will be written to the next entry by the device.

This allows low-latency applications to achieve better performance by
shortening the data path around polling CQ.

The kernel part was accepted long time ago, this is the supplementary user
space part.

PR was sent:
https://github.com/linux-rdma/rdma-core/pull/369

Artemy Kovalyov (2):
  verbs: Add IBV_CREATE_CQ_ATTR_IGNORE_OVERRUN support
  mlx5: Add IBV_UVERBS_CQ_FLAGS_IGNORE_OVERRUN support

 libibverbs/cmd_cq.c               |  3 ++
 libibverbs/man/ibv_create_cq_ex.3 |  1 +
 libibverbs/verbs.h                |  1 +
 providers/mlx5/mlx5-abi.h         |  2 ++
 providers/mlx5/verbs.c            | 60 +++++++++++++++++++++++++++------------
 5 files changed, 49 insertions(+), 18 deletions(-)