Re: [PATCH net-next v2 2/9] phy: add hwtstamp_get callback to phy drivers

[Vadim Fedorenko <vadim.fedorenko@linux.dev>]

On 13/11/2025 13:36, Andrew Lunn wrote:
>> I was planning to remove SIOCSHWTSTAMP/SIOCGHWTSTAMP dev_eth_ioctl calls
>> later once everything has landed and we have tests confirming that ioctl
>> and netlink interfaces work exactly the same way.
> 
> I don't think you can remove SIOCSHWTSTAMP, it is ABI. All you can
> really do is change the implementation so that it uses the same path
> as the netlink code.

Probably wrong explanation from my side. The plan is not to remove ABI,
but to let it go through netlink path in the core. As you suggest.

> You can avoid this for _get by never adding it in the first
> place. Only support the netlink API for PHYs.
Andrew, could you please check if I understand things correctly. PHY
devices are not exposed to user-space directly, and thus cannot be
configured via ioctl without netdev's .ndo_eth_ioctl(). The core netdev 
part falls back to ioctl for SIOCSHWTSTAMP/SIOCGHWTSTAMP only in case
when there is no ndo_hwtstamp_get/ndo_hwtstamp_set implemented in
network device driver. Once all network drivers are converted, there
will be no way user-space SIOCSHWTSTAMP/SIOCGHWTSTAMP ABI can reach
phy_mii_ioctl().

If the above is correct, then yes, there is no reason to implement
SIOCGHWTSTAMP, and even more, SIOCSHWTSTAMP can be technically removed
as a dead code.