On Fri, Oct 12, 2018 at 10:48:07AM +0200, Boris Brezillon wrote:

> First question that might come to mind is why should we support such
> stateful modes? If you think about it, the gain of transmitting the
> opcode on 8 IO lines is rather small compared to the pain it is to
> teach the SPI NOR framework how to deal with that. The problem is,
> some SPI NOR manufacturers (Macronix for instance) only implement 1-1-1
> and 8-8-8 (or 8D-8D-8D), and they want to be able to use their NORs in
> 8-8-8 mode when the controller supports it.

The SPI framework changes definitely look OK to me, if everyone agrees
that this is a good way to go from a MTD point of view I'm happy to
apply them.  I have no strong opinion on the MTD bits of the series.