Hi Rob,
On Tue, Mar 3, 2020 at 10:05 PM Rob Herring <robh+dt@xxxxxxxxxx> wrote:
> On Tue, Mar 3, 2020 at 3:45 AM Geert Uytterhoeven
> <geert+renesas@xxxxxxxxx> wrote:
> > Currently, the DT bindings for an SPI controller specify that
> > "#address-cells" must be fixed to one. However, that applies to an SPI
> > controller in master mode only. When running in SPI slave mode,
> > "#address-cells" should be zero.
> >
> > Fix this making the value of "#address-cells" dependent on the presence
> > of "spi-slave".
> >
> > Fixes: 0a1b929356830257 ("spi: Add YAML schemas for the generic SPI options")
> > Reported-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@xxxxxxxxxxx>
> > Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>
> > ---
> > v2:
> > - Use "enum: [0, 1]" instead of min/max limit,
> > - use "- spi-slave" instead of "[ spi-slave ]".
> >
> > As of dtc commit 403cc79f06a135ae ("checks: Update SPI bus check for
> > 'spi-slave'") and Linux commit c2e7075ca8303631 ("scripts/dtc: Update to
> > upstream version v1.4.7-57-gf267e674d145"), dtc knows about SPI slave.
> >
> > However, when using "#address-cells = <0>" with W=1:
> >
> > Warning (avoid_unnecessary_addr_size): /soc/spi@e6e10000: unnecessary #address-cells/#size-cells without "ranges" or child "reg" property
>
> What was the point in having #address-cells in the first place for
> slaves?
I don't know, commit a8830cb19cfea04e ("spi: Document DT bindings for
SPI controllers in slave mode") doesn't require any #address-cells for
slave mode.
Perhaps because node_addr_cells() in dtc defaults to 2?
Or because of_bus_n_addr_cells() walks up the parent chain and thus
defaults to the first found parent value?
> Seems like we should make it mutually exclusive with 'spi-slave'.
Sounds like a good idea. How to express that in yaml?
Thanks!
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
