Re: [PATCH for-next 01/13] RDMA/hns: Encapsulate some lines for setting sq size in user mode

[oulijun <oulijun@huawei.com>]

在 2019/7/30 20:17, Jason Gunthorpe 写道:
> On Tue, Jul 30, 2019 at 02:16:01PM +0300, Gal Pressman wrote:
>> On 30/07/2019 11:56, Lijun Ou wrote:
>>> It needs to check the sq size with integrity when configures
>>> the relatived parameters of sq. Here moves the relatived code
>>> into a special function.
>>>
>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>  drivers/infiniband/hw/hns/hns_roce_qp.c | 29 ++++++++++++++++++++++-------
>>>  1 file changed, 22 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
>>> index 9c272c2..35ef7e2 100644
>>> +++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
>>> @@ -324,16 +324,12 @@ static int hns_roce_set_rq_size(struct hns_roce_dev *hr_dev,
>>>  	return 0;
>>>  }
>>>  
>>> -static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev,
>>> -				     struct ib_qp_cap *cap,
>>> -				     struct hns_roce_qp *hr_qp,
>>> -				     struct hns_roce_ib_create_qp *ucmd)
>>> +static int check_sq_size_with_integrity(struct hns_roce_dev *hr_dev,
>>> +					struct ib_qp_cap *cap,
>>> +					struct hns_roce_ib_create_qp *ucmd)
>>>  {
>>>  	u32 roundup_sq_stride = roundup_pow_of_two(hr_dev->caps.max_sq_desc_sz);
>>>  	u8 max_sq_stride = ilog2(roundup_sq_stride);
>>> -	u32 ex_sge_num;
>>> -	u32 page_size;
>>> -	u32 max_cnt;
>>>  
>>>  	/* Sanity check SQ size before proceeding */
>>>  	if ((u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes ||
>>> @@ -349,6 +345,25 @@ static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev,
>>>  		return -EINVAL;
>>>  	}
>>>  
>>> +	return 0;
>>> +}
>>> +
>>> +static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev,
>>> +				     struct ib_qp_cap *cap,
>>> +				     struct hns_roce_qp *hr_qp,
>>> +				     struct hns_roce_ib_create_qp *ucmd)
>>> +{
>>> +	u32 ex_sge_num;
>>> +	u32 page_size;
>>> +	u32 max_cnt;
>>> +	int ret;
>>> +
>>> +	ret = check_sq_size_with_integrity(hr_dev, cap, ucmd);
>>> +	if (ret) {
>>> +		dev_err(hr_dev->dev, "Sanity check sq size fail\n");
>> Consider using ibdev_err, same applies for other patches.
> It would be good if driver authors would convert their drivers to use
> the new interfaces
>
> Jason
>
> .
Yes, I am testing the effects of both according to the recommendations.
I think we should consider replacement spparately,  Do you agree?