Re: [PATCH v2 for-next 2/2] RDMA/hns: Support userspace configuring congestion control algorithm with QP granularity

[Junxian Huang <huangjunxian6@hisilicon.com>]

On 2024/2/26 16:09, Leon Romanovsky wrote:
> On Thu, Feb 22, 2024 at 03:06:20PM +0800, Junxian Huang wrote:
>>
>>
>> On 2024/2/21 23:52, Jason Gunthorpe wrote:
>>> On Thu, Feb 08, 2024 at 11:50:38AM +0800, Junxian Huang wrote:
>>>> Support userspace configuring congestion control algorithm with
>>>> QP granularity. If the algorithm is not specified in userspace,
>>>> use the default one.
>>>>
>>>> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
>>>> ---
>>>>  drivers/infiniband/hw/hns/hns_roce_device.h | 23 +++++--
>>>>  drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 14 +---
>>>>  drivers/infiniband/hw/hns/hns_roce_hw_v2.h  |  3 +-
>>>>  drivers/infiniband/hw/hns/hns_roce_main.c   |  3 +
>>>>  drivers/infiniband/hw/hns/hns_roce_qp.c     | 71 +++++++++++++++++++++
>>>>  include/uapi/rdma/hns-abi.h                 | 17 +++++
>>>>  6 files changed, 112 insertions(+), 19 deletions(-)
> 
> <...>
> 
>>>> +
>>>> +enum hns_roce_create_qp_comp_mask {
>>>> +	HNS_ROCE_CREATE_QP_MASK_CONGEST_TYPE = 1 << 1,
>>>
>>> Why 1<<1 not 1<<0?
>>
>> This is to keep consistent with our internal ABI, there are some
>> features not upstream yet.
>>
> 
> <...>
> 
>>>> @@ -114,6 +128,9 @@ struct hns_roce_ib_alloc_ucontext_resp {
>>>>  	__u32	reserved;
>>>>  	__u32	config;
>>>>  	__u32	max_inline_data;
>>>> +	__u8	reserved0;
>>>> +	__u8	congest_type;
>>>
>>> Why this layout?
>>>> Jason
>>
>> Same as the 1<<1 issue, to keep consistent with our internal ABI.
> 
> We are talking about upstream kernel UAPI, there is no internal ABI here.
> 
> Please fix it.
> 
> Thanks
> 

Sure. Will fix it in next version.

Thanks,
Junxian

>>
>> Thanks,
>> Junxian