[PATCH v2 1/1] nvme-rdma: Fix memory leak during queue allocation

[Max Gurtovoy]

On 11/13/2017 9:31 PM, Sagi Grimberg wrote:
> Hi Max,
> 
>>   out_destroy_qp:
>> -    rdma_destroy_qp(queue->cm_id);
>> +    ib_destroy_qp(queue->qp);
> 
> Why was this changed? Any specific reason?>
> 

In order to destroy QP using 1 API instead of 2.
We can leave it "rdma_destroy_qp(queue->cm_id);", here it's safe.

>>   out_destroy_ib_cq:
>>       ib_free_cq(queue->ib_cq);
>>   out_put_dev:
>> @@ -546,6 +555,7 @@ static int nvme_rdma_alloc_queue(struct 
>> nvme_rdma_ctrl *ctrl,
>>   out_destroy_cm_id:
>>       rdma_destroy_id(queue->cm_id);
>> +    nvme_rdma_destroy_queue_ib(queue);
>>       return ret;
>>   }
>> @@ -563,8 +573,8 @@ static void nvme_rdma_free_queue(struct 
>> nvme_rdma_queue *queue)
>>       if (!test_and_clear_bit(NVME_RDMA_Q_ALLOCATED, &queue->flags))
>>           return;
>> -    nvme_rdma_destroy_queue_ib(queue);
>>       rdma_destroy_id(queue->cm_id);
>> +    nvme_rdma_destroy_queue_ib(queue);
> 
> Why was this changed? What race are you preventing here?

No race here, just wanted to align the order of destruction and make 
sure we don't get any rdma_cm events during queue_ib destruction as we 
did above.

Do you prefer leaving these 2 lines "as is" and add comments in the code ?