[PATCH 09/10] nvme/042: test bi-directional authentication

Hannes Reinecke hare at suse.de
Mon Mar 28 06:29:30 PDT 2022


On 3/28/22 14:02, Hannes Reinecke wrote:
> On 3/28/22 13:20, Sagi Grimberg wrote:
>>
>>>>>>>> This test case has a weird behavior. Why isn't _nvme_connect_subsys
>>>>>>>> failing?
>>>>>>>
>>>>>>> Oh, but it is.
>>>>>>> The line 'nvme controller found' is not part of the .out file, hence
>>>>>>> the test will fail.
>>>>>>>
>>>>>>> Call it an obscure feature of blktests :-)
>>>>>>>
>>>>>>> But yeah, I can make it more obvious that this is the failure case.
>>>>>>
>>>>>> Hannes, my question is why isn't nvme connect not failing (i.e.
>>>>>> returning a non-zero status code)?
>>>>>>
>>>>> I guess it does, but we're never actually checking the return value.
>>>>> Let me check.
>>>>
>>>> I tested it manually, and the connect does not fail, but rather 
>>>> succeeds
>>>> and the controller is removed - which is strange.
>>>
>>> Curious. "That shouldn't have happened (tm)".
>>
>> Hannes, before I take v10 to a test drive, did you fix this particular
>> issue?
> 
> Now that you mention it: Not sure. Will be checking.
> 
Found it. One shouldn't overwrite the error code with 'success' after 
sending failure2 ...

Cheers,

Hannes



More information about the Linux-nvme mailing list