Re: [dpdk-dev] [PATCH] ethdev: unify error code if port ID is invalid

[Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>]

On 10/13/20 7:12 PM, Ferruh Yigit wrote:
> On 10/13/2020 4:39 PM, Andrew Rybchenko wrote:
>> On 10/13/20 6:32 PM, Ferruh Yigit wrote:
>>> On 10/13/2020 3:53 PM, Andrew Rybchenko wrote:
>>>> Use ENODEV as the error code if specified port ID is invalid.
>>>>
>>>> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
>>>> ---
>>>>    lib/librte_ethdev/rte_ethdev.c | 44 ++++++++++++++++----------------
>>>>    lib/librte_ethdev/rte_ethdev.h | 46
>>>> +++++++++++++++++++++++-----------
>>>>    2 files changed, 54 insertions(+), 36 deletions(-)
>>>>
>>>> diff --git a/lib/librte_ethdev/rte_ethdev.c
>>>> b/lib/librte_ethdev/rte_ethdev.c
>>>> index 5b7979a3b8..1f862f918a 100644
>>>> --- a/lib/librte_ethdev/rte_ethdev.c
>>>> +++ b/lib/librte_ethdev/rte_ethdev.c
>>>> @@ -784,7 +784,7 @@ rte_eth_dev_get_name_by_port(uint16_t port_id,
>>>> char *name)
>>>>    {
>>>>        char *tmp;
>>>>    -    RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -EINVAL);
>>>> +    RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
>>>
>>> Thanks Andrew, +1 to this error unification.
>>>
>>> This will be API change without deprecation notice, cc'ed techboard
>>> for it.
>>
>> Yes, thanks.
>>
>>>
>>> If this should (almost) always return '-ENODEV', does it make sense to
>>> make another wrapper macro for it, to prevent later other error types
>>> used again.
>>
>> Unlikely, since most likely the line will be simply copied.
>> RTE_ETH_VALID_PORTID_OR_ERR_RET will remain in any case, so
>> it will be possible to misuse it anyway.
>>
> 
> Agree it won't prevent misuse completely but may help, anyway I don't
> have a strong opinion here, if you think that is not needed, that is OK.
> 
>>>
>>> And there are a few instances returning '-1', are they left
>>> intentionally?
>>
>> Yes. Inside ethdev it is either socket_id or fd in these cases.
>>
> 
> Can't those two also updated to return '-ENODEV' when 'port_id' is not
> valid?

I think no.
1. rte_eth_dev_socket_id() should not return -ENODEV since it
   can return -1 even if port ID is valid if fact (I see
   printouts from time to time if I'm not mistaken) and
   typically handled as unspecified NUMA node ID.
2. rte_eth_dev_rx_intr_ctl_q_get_fd() explicitly says that -1
   is returned on error. The function is still experimental
   and we can change it, but I'd say that -1 match typical
   behavior for functions returning file descriptor.

Let's limit the changeset to switch from EINVAL to ENODEV.