> Sorry, I don't understand. Looking into the code again, the major problem > I see is that you want to model fan install status and fan fault > status as gpio pins. The same is true for psu information (installed, > ac, dc flags). > If you want to do this, fine, but then get the status from the gpio > driver and don't export anything to the gpio driver. The kernel supports > means to do that (look at gpiod_get and similar functions). It makes the > code more complex, but I assume you know what you are doing. Greetings Guenter and Andy, I have pursued this approach and I have found that while the PSU and FAN drivers can consume the presence and fail info from the GPIO driver, the host is unable to read the read only GPIOs. In OpenBMC we have a GPIO consumer that sits and waits for GPIOs changes then takes action on it. To resolve this issue would it be acceptable for the GPIO driver to poll the relevant GPIOs and share the necessary fan presence/failure and psu presence/failure information via a global shared variable? This would be the alternative to the drivers consuming GPIOs. Thanks you for your time and assistance, -Nick Hawkins
