On 2019-06-18 12:58, Linus Walleij wrote: > This switches the i801 GPIO mux to use GPIO descriptors for > handling the GPIO lines. The previous hack which was reaching > inside the GPIO chips etc cannot live on. We pass descriptors > along with the GPIO mux device at creation instead. > > The GPIO mux was only used by way of platform data with a > platform device from one place in the kernel: the i801 i2c bus > driver. Let's just associate the GPIO descriptor table with > the actual device like everyone else and dynamically create > a descriptor table passed along with the GPIO i2c mux. > > This enables simplification of the GPIO i2c mux driver to > use only the descriptor API and the OF probe path gets > simplified in the process. > > The i801 driver was registering the GPIO i2c mux with > PLATFORM_DEVID_AUTO which would make it hard to predict the > device name and assign the descriptor table properly, but > this seems to be a mistake to begin with: all of the > GPIO mux devices are hardcoded to look up GPIO lines from > the "gpio_ich" GPIO chip. If there are more than one mux, > there is certainly more than one gpio chip as well, and > then we have more serious problems. Switch to > PLATFORM_DEVID_NONE instead. There can be only one. > > Cc: Mika Westerberg <mika.westerberg@xxxxxxxxxxxxxxx> > Cc: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> > Cc: Peter Rosin <peda@xxxxxxxxxx> > Cc: Jean Delvare <jdelvare@xxxxxxxx> > Signed-off-by: Serge Semin <fancer.lancer@xxxxxxxxx> > Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxx> Patch added to the i2c-mux/for-next branch with the newline removed as suggested by Andy. Thanks to all involved for the effort to get this done! [BTW, Andy, I do not seem to get any mails from you delivered, and only noticed your review by looking at patchwork. Strange] Cheers, Peter
