[PATCH v2] riscv: dts: canaan: Fix SPI3 bus width

[Niklas Cassel]

On Tue, Apr 05, 2022 at 02:26:53PM +0200, Geert Uytterhoeven wrote:
> Hi Niklas,
> 
> On Tue, Mar 8, 2022 at 2:30 PM Niklas Cassel <Niklas.Cassel at wdc.com> wrote:
> > From: Niklas Cassel <niklas.cassel at wdc.com>
> > According to the K210 Standalone SDK Programming guide:
> > https://canaan-creative.com/wp-content/uploads/2020/03/kendryte_standalone_programming_guide_20190311144158_en.pdf
> >
> > Section 15.4.3.3:
> > SPI0 and SPI1 supports: standard, dual, quad and octal transfers.
> > SPI3 supports: standard, dual and quad transfers (octal is not supported).
> >
> > In order to support quad transfers (Quad SPI), SPI3 must have four IO wires
> > connected to the SPI flash.
> >
> > Update the device tree to specify the correct bus width.
> >
> > Tested on maix bit, maix dock and maixduino, which all have the same
> > SPI flash (gd25lq128d) connected to SPI3. maix go is untested, but it
> > would not make sense for this k210 board to be designed differently.
> >
> > Signed-off-by: Niklas Cassel <niklas.cassel at wdc.com>
> > ---
> > Changes since v1:
> > -Add the new properties directly after spi-max-frequency for all DT board
> >  files.
> 
> Thanks for your patch, which is now commit 6846d656106add3a ("riscv:
> dts: canaan: Fix SPI3 bus width") in v5.18-rc1.
> 
> > --- a/arch/riscv/boot/dts/canaan/sipeed_maix_bit.dts
> > +++ b/arch/riscv/boot/dts/canaan/sipeed_maix_bit.dts
> > @@ -203,6 +203,8 @@ flash at 0 {
> >                 compatible = "jedec,spi-nor";
> >                 reg = <0>;
> >                 spi-max-frequency = <50000000>;
> > +               spi-tx-bus-width = <4>;
> > +               spi-rx-bus-width = <4>;
> >                 m25p,fast-read;
> >                 broken-flash-reset;
> >         };
> 
> On MAiX BiT, I get:
> 
>     +spi spi1.0: setup: ignoring unsupported mode bits a00
>      spi-nor spi1.0: gd25lq128d (16384 Kbytes)

Hello Geert,

The device tree is supposed to describe the hardware.

The Synopsys SPI controller and the Gigadevice SPI flash both support quad
transfers.

It would be incorrect to adapt the device tree based on current limitations
of the drivers/spi/spi-dw-core.c driver.

Likewise, we shouldn't need to update the device tree if the dwc driver
ever adds support for quad transfers.

However, I do agree that it is a bit weird that the kernel outputs a
warning for this case.

I understand that the warning is supposed to be there to warn that a
controller does not support a mode required by the driver, but if it
is the driver and not the controller that lacks support, is a warning
really warranted? I'm not so sure.

Adding Mark Brown to hopefully hear his opinion.


Kind regards,
Niklas