Re: [PATCH] spi: bcm2835: Convert to use CS GPIO descriptors

[Stefan Wahren <stefan.wahren@xxxxxxxx>]

Am 20.04.19 um 17:46 schrieb Martin Sperl:
>
>> On 20.04.2019, at 13:08, Linus Walleij <linus.walleij@xxxxxxxxxx> wrote:
>>
>> This converts the BCM2835 SPI master driver to use GPIO
>> descriptors for chip select handling.
>>
>> The BCM2835 driver was relying on the core to drive the
>> CS high/low so very small changes were needed for this
>> part. If it managed to request the CS from the device tree
>> node, all is pretty straight forward.
>>
>> However for native GPIOs this driver has a quite unorthodox
>> loopback to request some GPIOs from the SoC GPIO chip by
>> looking it up from the device tree using gpiochip_find()
>> and then offseting hard into its numberspace. This has
>> been augmented a bit by using gpiochip_request_own_desc()
>> but this code really needs to be verified. If "native CS"
>> is actually an SoC GPIO, why is it even done this way?
>> Should this GPIO not just be defined in the device tree
>> like any other CS GPIO? I'm confused.
> It has been done to keep dt-backward compatibility.
>
> When the driver was written dt had configured native-cs.

This is done in a very opaque way. The bcm2835-rpi.dtsi defines a pinmux
group alt0 which init SPI pins (including native CS). So we better use
spi0_gpio7 (including native CS) or even better create new groups
without native CS and use them.

Even worse the alt0 group is applied for all the RPi Zero, 1 and 2
boards, but not for the RPi 3 ones.

> A native-cs implementation comes with lots of
> limitations and some of which - especially dma mode
> has some quirks with regards to native-cs (glitches,
> deasserts after each finished dma transfer,...)
>
> The only way to make this work properly was to use
> Gpio.

So we better extend the current binding with cs-gpios as optional
property in order to point the user the right direction. And gave the
same warnings like in bcm2835-spi-aux.

Stefan

>
> So to properly support old device-trees with native-cs 
> this gpio translation is needed, as we should not break
> things for old dt...
>
> Maybe this rational helps...
>
> Martin
>