| Message ID | 20200303094522.23180-2-geert+renesas@glider.be (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | spi: dt-bindings: spi-controller: Slave mode fixes | expand |
On Tue, Mar 3, 2020 at 3:45 AM Geert Uytterhoeven <geert+renesas@glider.be> 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@renesas.com> > Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be> > --- > 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? Seems like we should make it mutually exclusive with 'spi-slave'. Rob
Hi Rob, On Tue, Mar 3, 2020 at 10:05 PM Rob Herring <robh+dt@kernel.org> wrote: > On Tue, Mar 3, 2020 at 3:45 AM Geert Uytterhoeven > <geert+renesas@glider.be> 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@renesas.com> > > Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be> > > --- > > 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
On Wed, Mar 4, 2020 at 6:50 AM Geert Uytterhoeven <geert@linux-m68k.org> wrote: > > Hi Rob, > > On Tue, Mar 3, 2020 at 10:05 PM Rob Herring <robh+dt@kernel.org> wrote: > > On Tue, Mar 3, 2020 at 3:45 AM Geert Uytterhoeven > > <geert+renesas@glider.be> 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@renesas.com> > > > Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be> > > > --- > > > 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? oneOf: - required: - "#address-cells" - required: - spi-slave Rob
diff --git a/Documentation/devicetree/bindings/spi/spi-controller.yaml b/Documentation/devicetree/bindings/spi/spi-controller.yaml index 1e0ca6ccf64bbd0a..401d41a97562dc8d 100644 --- a/Documentation/devicetree/bindings/spi/spi-controller.yaml +++ b/Documentation/devicetree/bindings/spi/spi-controller.yaml @@ -20,7 +20,7 @@ properties: pattern: "^spi(@.*|-[0-9a-f])*$" "#address-cells": - const: 1 + enum: [0, 1] "#size-cells": const: 0 @@ -52,6 +52,18 @@ properties: description: The SPI controller acts as a slave, instead of a master. +if: + required: + - spi-slave +then: + properties: + "#address-cells": + const: 0 +else: + properties: + "#address-cells": + const: 1 + patternProperties: "^slave$": type: object
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@renesas.com> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be> --- 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 Just removing #address-cells (using /delete-property/ in the board DTS) to fix this warning causes: Warning (spi_bus_bridge): /soc/spi@e6e10000: incorrect #address-cells for SPI bus as spi_bus_bridge() uses node_addr_cells(), which defaults to 2 (due to dtc's ppc64 heritage? But node_size_cells() defaults to 1, not 2?). How should this be fixed? Should "#address-cells = <0>" be left out or not? Should node_{addr,size}_cells() in dtc default to zero? Thanks! --- .../devicetree/bindings/spi/spi-controller.yaml | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-)