[ofa-general] Re: impossibility to bind a device/port with the rdma-cm when the port is down

[Yossi Etigin]

Sean Hefty wrote:
>> @@ -2167,6 +2170,12 @@ static int cma_sidr_rep_handler(struct i
>> 			event.status = ib_event->param.sidr_rep_rcvd.status;
>> 			break;
>> 		}
>> +		ret = cma_set_qkey(id_priv);
>> +		if (ret) {
>> +			event.event = RDMA_CM_EVENT_ADDR_ERROR;
>> +			event.status = -EINVAL;
>> +			break;
>> +		}
>> 		if (id_priv->qkey != rep->qkey) {
>> 			event.event = RDMA_CM_EVENT_UNREACHABLE;
>> 			event.status = -EINVAL;
>> @@ -2446,10 +2455,14 @@ static int cma_send_sidr_rep(struct rdma
>> 			     const void *private_data, int private_data_len)
>> {
>> 	struct ib_cm_sidr_rep_param rep;
>> +	int ret;
>>
>> 	memset(&rep, 0, sizeof rep);
>> 	rep.status = status;
>> 	if (status == IB_SIDR_SUCCESS) {
>> +		ret = cma_set_qkey(id_priv);
>> +		if (ret)
>> +			return ret;
>> 		rep.qp_num = id_priv->qp_num;
>> 		rep.qkey = id_priv->qkey;
>> 	}
> 
> Looking at this, I keep wanting to set the qkey when sending or receiving the
> sidr req, not rep.  This is earlier than the qkey is needed, but catching the
> error sooner in this case seems better to me than deferring.  Thoughts?
> 
> - Sean
> 

It might be better to catch errors earlier, but there is the risk that the
flow might change somehow, and losing the (now obvious) logical connection 
between retrieving the qkey and actually using it.

-- 
--Yossi