RDMA/hns: Fix the endian problem for hns
authoroulijun <oulijun@huawei.com>
Mon, 5 Feb 2018 13:14:00 +0000 (21:14 +0800)
committerDoug Ledford <dledford@redhat.com>
Mon, 5 Feb 2018 15:48:48 +0000 (10:48 -0500)
commit8b9b8d143b467ec9c65f87b7c2596dc2aabe6737
treee7e8bfd3bbaaf03ee1435cf97c74447b141400c6
parente9d1e389273aac7fc50c53f0d4492dc190185882
RDMA/hns: Fix the endian problem for hns

The hip06 and hip08 run on a little endian ARM, it needs to
revise the annotations to indicate that the HW uses little
endian data in the various DMA buffers, and flow the necessary
swaps throughout.

The imm_data use big endian mode. The cpu_to_le32/le32_to_cpu
swaps are no-op for this, which makes the only substantive
change the handling of imm_data which is now mandatory swapped.

This also keep match with the userspace hns driver and resolve
the warning by sparse.

Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
drivers/infiniband/hw/hns/hns_roce_common.h
drivers/infiniband/hw/hns/hns_roce_device.h
drivers/infiniband/hw/hns/hns_roce_hw_v1.c
drivers/infiniband/hw/hns/hns_roce_hw_v1.h
drivers/infiniband/hw/hns/hns_roce_hw_v2.c
drivers/infiniband/hw/hns/hns_roce_hw_v2.h
drivers/infiniband/hw/hns/hns_roce_main.c
drivers/infiniband/hw/hns/hns_roce_qp.c