On Thu, Dec 14, 2017 at 9:59 AM, Philippe Mathieu-Daudé <f4...@amsat.org> wrote: > On Thu, Dec 14, 2017 at 2:50 PM, Alistair Francis > <alistair.fran...@xilinx.com> wrote: >> On Wed, Dec 13, 2017 at 11:58 AM, Philippe Mathieu-Daudé >> <f4...@amsat.org> wrote: >>> Now only SD 'producers' are able to use the "sd-internal.h" API, >>> while SD 'consumers' are restricted to the "hw/sd/sd.h" 'public' API. >>> >>> Signed-off-by: Philippe Mathieu-Daudé <f4...@amsat.org> >> >> I don't see what this gets us. Why bother moving this code into an >> internal header? > > This helped me to sort out the QOM design, however it is indeed not necessary.
If it makes development easier that is a good reason. I just don't see in this series how it helps. > > I later use the registerfields API to add a bunch of other registers > and bitfields, so I prefer to still move the register #defines in > another internal header, else this API header get pretty unreadable. Ah, that makes some more sense. Alistair > I'll respin this series without this split. >
