On 19/09/2019 17:03:15+0100, Mark Brown wrote: > On Thu, Sep 19, 2019 at 05:38:47PM +0200, Gregory CLEMENT wrote: > > > With this patch, when using a gpio CS, the hardware CS0 is always > > used. Thanks to this, there is no more limitation for the number of > > gpio CS we can use. > > This is going to break any system where we use both a GPIO chip select > and chip select 0. Ideally we'd try to figure out an unused chip select > to use here... The point is that this use case is already broken and this patch fixes the crash and is easily backportable. Fixing the CS + gpio CS should probably be done in a separate patch. -- Alexandre Belloni, Bootlin Embedded Linux and Kernel engineering https://bootlin.com