Hi Andy, On 2/10/23 16:35, Andy Shevchenko wrote: > On Fri, Feb 10, 2023 at 01:05:52PM +0200, Laurent Pinchart wrote: >> On Fri, Feb 10, 2023 at 12:53:43PM +0200, Andy Shevchenko wrote: >>> On Fri, Feb 10, 2023 at 12:47:55PM +0200, Sakari Ailus wrote: >>>> On Fri, Feb 10, 2023 at 12:29:19PM +0200, Laurent Pinchart wrote: >>>>> On Fri, Feb 10, 2023 at 12:21:15PM +0200, Sakari Ailus wrote: >>>>>> On Thu, Feb 09, 2023 at 06:11:12PM +0200, Laurent Pinchart wrote: > > ... > >>>>>> I took a look at this some time ago, too, and current regmap API is a poor >>>>>> fit for CCI devices. CCI works on top of e.g. both I²C and I3C so something >>>>>> on top of regmap is a better approach indeed. >>>>> >>>>> I'm confused, is regmap a poor fit, or a better approach ? >>>> >>>> I'm proposing having something on top of regmap, but not changing regmap >>>> itself. >>> >>> I don't understand why we can't change regmap? regmap has a facility called >>> regmap bus which we can provide specifically for these types of devices. What's >>> wrong to see it done? >> >> How would that work ? > > If I'm not mistaken, you may introduce something like regmal CCI and then > > regmap_init_cci(); > > > regmap_read()/regmap_write() > regmap_update_bits() > regmap_bulk_*() > > at your service without changing a bit in the drivers (they will use plain > regmap APIs instead of custom ones). regmap_bus is for low-level busses like i2c, i3c, spi, etc. We could "abuse" this to overwrite the standard regmap read/write helpers with bus specific ones, but then we loose the actual bus abstraction and we would need separate regmap-cci implementations for i2c/i3c/spi. > Again, we can extend regmap to have something like > > int (*reg_width)(regmap *, offset) > > callback added that will tell the regmap bus underneath what size to use. > > In the driver one will define the respective method to return these widths. That won't work internal helpers to marshall raw-buffers with both reg-addr + reg value(s) to pass to the low-level (i2c/i3c/spi) drivers use an internal regmap.format struct which gets filled with reg-width specific info once from __regmap_init() what you are suggesting really requires major surgery to the whole regmap core. CCI really is an extra protocol layer on top of lower-level / more primitive busses and it is best to have a helper library with a few helpers using regmap underneath to abstract the raw bus accesses away. Note this helper library can be quite thin and small though :) Regards, Hans
