On Mon, 7 Apr 2025 12:35:37 +0000 Marek Pazdan <mpaz...@arista.com> wrote:
> Signal Definition section (Other signals) of SFF-8636 Spec mentions that > additional signals like reset and module present may be implemented for > a specific hardware. There is currently no user space API for control of > those signals so user space management applications have no chance to > perform some diagnostic or configuration operations. This patch uses > get_phy_tunable/set_phy_tunable ioctl API to provide above control to > user space. Despite ethtool reset option seems to be more suitable for > transceiver module reset control, ethtool reset doesn't allow for reset > pin assertion and deassertion. Userspace API may require control over > when pin will be asserte and deasserted so we cannot trigger reset and > leave it to the driver when deassert should be perfromed. nit: performed ETHTOOL_PHY_G/STUNABLE IOCTLs are targeting the PHY of the NIC but IIUC in your case you are targeting the reset of the QSFP module. Maybe phylink API is more appropriate for this feature. You have to add net-next prefix in the subject like this [PATCH net-next 1/2] when you add new support to net subsystem. Regards, -- Köry Maincent, Bootlin Embedded Linux and kernel engineering https://bootlin.com