Re: [PATCH] spi: Support high CS when using descriptors

[Linus Walleij <linus.walleij@xxxxxxxxxx>]

On Thu, Jan 10, 2019 at 4:52 PM Janek Kotas <jank@xxxxxxxxxxx> wrote:

> >> However, the chip select signal is inverted.
> >> The moment the SPI HW is initialized it goes low.
> >> It is driven high during SPI transmissions.
> >
> > Ouch! Which device tree file is this?
>
> I’m using a custom one for out test system.
> Here’s the SPI part:
>
> spi0: spi@fd0b0000 {
>     compatible = "cdns,spi-r1p6";
>     reg = <0x0 0xfd0b0000 0x0 0x1000>;
>     clock-names = "ref_clk", "pclk";
>     clocks = <&sysclock>, <&sysclock>;
>     interrupts = <0 4 IRQ_TYPE_LEVEL_HIGH>;
>     interrupt-parent = <&gic>;
>     num-cs = <4>;
>     is-decoded-cs = <0>;
>     #address-cells = <1>;
>     #size-cells = <0>;
>
>     eeprom@0 {
>         compatible = "at,at25";
>         reg = <0>;
>         spi-max-frequency = <10000000>;
>         /* spi-cpha; */
>         /* spi-cpol; */
>         pagesize = <16>;
>         size = <1024>;
>         address-width = <16>;
>     };
> };

Ah! You're using native chip selects!

And now I enforce CS as high, and that will not just
affect the GPIO CS but also the native CS.
How sloppy of me.

I will send a separate patch for that that you can
test toghether with the first patch (which I think
is fine in itself).

> > My intention was that of_gpio_flags_quirks() in
> > drivers/gpio/gpiolib-of.c would activate and you would
> > get a print like this in dmesg:
> >
> > of_gpio_flags_quirks: enforce active low on chipselect handle
>
> I haven’t seen that in my logs.

Not strange as you're not using GPIOs.

Yours,
Linus Walleij