On 5/19/26 4:09 PM, David Lechner wrote: > It seems a bit odd to use regulator_set_enable_if_allowed() > since the point of the binding is to provide a clock that > is gated by a regulator. Why would someone use this binding > with a regulator that can't be enabled/disabled?
Hi David, Jonas, Two concrete data points for the choice in this driver, on top of the general pattern Jonas mentioned: - pcie_dw_rockchip, the consumer that will drive this clock on rock-5-itx, uses regulator_set_enable_if_allowed() itself for its own vpcie3v3 supply, so the gated-fixed-clock driver runs the exact same code path through the regulator core. - vcc3v3_mkey on rock-5-itx is currently marked regulator-always-on in the u-boot dtsi, so the strict variant would surface -EACCES on every probe. Thanks for the review. Daniele
