On Sun, Feb 25, 2018 at 8:08 PM, Andrew Lunn <and...@lunn.ch> wrote: > On Sun, Feb 25, 2018 at 01:51:27PM +0100, Linus Walleij wrote: >> This kills off the platform data support from the bitbanged >> GPIO-based MDIO driver and moves it over to using GPIO >> descriptors exclusively. > > Hi Linus > > I like where this ends up. I wounder about the path it takes to get > there. There seems to be quite a lot of code which gets moved around > and then in the end deleted. Maybe changing the order of the patches > would help. Converting to devm_gpiod_get_index() first would remove > all the active_low flags. Then remove all the unused reset callback, > irq, etc?
It's mainly a matter of which order we want bisects to end up... I was thinking to order the patches from least-likely to cause regressions to most-likely to cause regressions so that it is later possible to revert the more likely regression cause without getting into conflict with the rest. Yours, Linus Walleij