On Wed, Sep 05, 2018 at 03:09:42PM +0200, Linus Walleij wrote: > On Wed, Sep 5, 2018 at 11:39 AM Mark Brown <broonie@xxxxxxxxxx> wrote: > > AFAICT from the rest of the series the root cause here is that you're > > trying to work around the GPIO controller setting the wrong flags rather > > than an actual fix here - there's no need for any change that I can see. > I think that may be up to the interpretation of > SPI_[MASTER|CONTROLLER]_NO_[RX|TX] flags. > From the code and the bitbanging inlines it is clear that > the actual semantics of these flags are: > SPI_MASTER_NO_RX == does not have a MISO line > SPI_MASTER_NO_TX == does not have a MOSI line I would be very surprised if whoever wrote the DT parsing code in the bitbanging driver even considered the three wire case. That's clearly not a sensible reading of NO_[RT]X from a natural langauge point of view. > Maybe I should make a patch renaming the flags > as SPI_*_NO_MISO, SPI_*_NO_MOSI > as that is how they are used in the code. That's how they're used in that specific bit of code, it's certainly not how the validation code which I wrote uses them, and I'm not convinced that it's a particularly useful thing to tell the rest of the world about - users will care if they can do the operation, it's not *super* important to them which wires it uses to do that.
Attachment:
signature.asc
Description: PGP signature
