Re: [RFC PATCH v8 04/10] dpll: netlink: Add DPLL framework base functions

From: Jiri Pirko
Date: Fri Jun 23 2023 - 03:48:11 EST


Fri, Jun 23, 2023 at 02:56:24AM CEST, arkadiusz.kubalewski@xxxxxxxxx wrote:
>>From: Jiri Pirko <jiri@xxxxxxxxxxx>
>>Sent: Wednesday, June 21, 2023 3:08 PM
>>
>>Wed, Jun 21, 2023 at 01:53:24PM CEST, jiri@xxxxxxxxxxx wrote:
>>>Wed, Jun 21, 2023 at 01:18:59PM CEST, poros@xxxxxxxxxx wrote:
>>>>Arkadiusz Kubalewski píše v Pá 09. 06. 2023 v 14:18 +0200:
>>>>> From: Vadim Fedorenko <vadim.fedorenko@xxxxxxxxx>
>>>
>>>[...]
>>>
>>>Could you perhaps cut out the text you don't comment? Saves some time
>>>finding your reply.
>>>
>>>
>>>>> +static int
>>>>> +dpll_set_from_nlattr(struct dpll_device *dpll, struct genl_info
>>>>> *info)
>>>>> +{
>>>>> +       const struct dpll_device_ops *ops = dpll_device_ops(dpll);
>>>>> +       struct nlattr *tb[DPLL_A_MAX + 1];
>>>>> +       int ret = 0;
>>>>> +
>>>>> +       nla_parse(tb, DPLL_A_MAX, genlmsg_data(info->genlhdr),
>>>>> +                 genlmsg_len(info->genlhdr), NULL, info->extack);
>>>>> +       if (tb[DPLL_A_MODE]) {
>>>>Hi,
>>>>
>>>>Here should be something like:
>>>> if (!ops->mode_set)
>>>> return -EOPNOTSUPP;
>>>
>>>Why? All drivers implement that.
>>>I believe that it's actullaly better that way. For a called setting up
>>>the same mode it is the dpll in, there should be 0 return by the driver.
>>>Note that driver holds this value. I'd like to keep this code as it is.
>>
>>Actually, you are correct Petr, my mistake. Actually, no driver
>>implements this. Arkadiusz, could you please remove this op and
>>possibly any other unused op? It will be added when needed.
>>
>>Thanks!
>>
>
>Sorry, didn't have time for such change, added only check as suggested by
>Petr.
>If you think this is a big issue, we could change it for next version.

It's odd to carry on ops which are unused. I would prefer that to be
removed now and only introduced when they are actually needed.


>
>Thank you!
>Arkadiusz
>
>>
>>>
>>>[...]