Andy Green wrote: > what might be worth effort > is bringing it under or extending the mainline SPI somehow so it can > operate inside interrupt context.
Yup, that would be nice. I guess a real-life example where it makes a big different would help with lobbying such a change into mainline. An incremental way to implement this would be to make a driver-specific function for asynchronous requests, and maybe have some fiendishly inefficient generic fallback. That way one driver could be changed at a time, without having to redesign the SPI core. - Werner
