Re: [PATCH v12 0/6] Driver for at91 usart in spi mode

[Lee Jones]

On Tue, 11 Sep 2018, Geert Uytterhoeven wrote:
> On Tue, Sep 11, 2018 at 8:40 PM Lee Jones <lee.jones@xxxxxxxxxx> wrote:
> > On Tue, 11 Sep 2018, Geert Uytterhoeven wrote:
> > > On Tue, Sep 11, 2018 at 5:36 PM Alexandre Belloni
> > > <alexandre.belloni@xxxxxxxxxxx> wrote:
> > > > On 11/09/2018 16:59:09+0200, Geert Uytterhoeven wrote:
> > > > > On Tue, Sep 11, 2018 at 11:40 AM Alexandre Belloni
> > > > > <alexandre.belloni@xxxxxxxxxxx> wrote:
> > > > > > On 11/09/2018 10:33:56+0100, Lee Jones wrote:
> > > > > > > On Tue, 04 Sep 2018, Radu Pirea wrote:
> > > > > > > > Radu Pirea (6):
> > > > > > > >   MAINTAINERS: add at91 usart mfd driver
> > > > > > > >   dt-bindings: add binding for atmel-usart in SPI mode
> > > > > > > >   mfd: at91-usart: added mfd driver for usart
> > > > > > > >   MAINTAINERS: add at91 usart spi driver
> > > > > > > >   spi: at91-usart: add driver for at91-usart as spi
> > > > > > > >   tty/serial: atmel: change the driver to work under at91-usart mfd
> > > > > > > >
> > > > > > > >  .../bindings/{serial => mfd}/atmel-usart.txt  |  25 +-
> > > > > > > >  MAINTAINERS                                   |  16 +
> > > > > > > >  drivers/mfd/Kconfig                           |   9 +
> > > > > > > >  drivers/mfd/Makefile                          |   1 +
> > > > > > > >  drivers/mfd/at91-usart.c                      |  71 +++
> > > > > > > >  drivers/spi/Kconfig                           |   8 +
> > > > > > > >  drivers/spi/Makefile                          |   1 +
> > > > > > > >  drivers/spi/spi-at91-usart.c                  | 432 ++++++++++++++++++
> > > > > > > >  drivers/tty/serial/Kconfig                    |   1 +
> > > > > > > >  drivers/tty/serial/atmel_serial.c             |  42 +-
> > > > > > > >  include/dt-bindings/mfd/at91-usart.h          |  17 +
> > > > > > > >  11 files changed, 606 insertions(+), 17 deletions(-)
> > > > > > > >  rename Documentation/devicetree/bindings/{serial => mfd}/atmel-usart.txt (76%)
> > > > > > > >  create mode 100644 drivers/mfd/at91-usart.c
> > > > > > > >  create mode 100644 drivers/spi/spi-at91-usart.c
> > > > > > > >  create mode 100644 include/dt-bindings/mfd/at91-usart.h
> > > > > > >
> > > > > > > Seeing as this patch-set has caused some issues this morning, I took
> > > > > > > the liberty to peruse back into its history to figure out where things
> > > > > > > started to go wrong.  I also re-reviewed the MFD driver - and I'm glad
> > > > > > > I did!
> > > > > > >
> > > > > > > My Acked-by has been attached to the MFD portion since v5, which is
> > > > > > > why the code hasn't caught my eye before today.  I reviewed the
> > > > > > > relocation of the *binding document* (serial => mfd with no changes)
> > > > > > > in v4 and nothing else.  It appears as though you mistakenly added it
> > > > > > > to the *MFD driver* instead.  This explains my confusion in v10 when I
> > > > > > > told you I'd already reviewed the binding document.
> > > > > > >
> > > > > > > As I said, I have re-reviewed the MFD driver and I'm afraid to say
> > > > > > > that I do not like what I see.  Besides the missing header file and
> > > > > > > the whitespace tabbing errors, I do not agree with the implementation.
> > > > > > > Using MFD as a shim to hack around driver selection is not a valid
> > > > > > > use-case.
> > > > > > >
> > > > > > > What's stopping you from just using the compatible string directly to
> > > > > > > select which driver you need to probe?
> > > > > > >
> > > > > >
> > > > > > Then you'd have multiple compatible strings for the same IP which is a
> > > > > > big no-no.
> > > > >
> > > > > It's still the same hardware device, isn't?
> > > > > What if the SPI or UART slave is not on-board, but on an expansion board?
> > > > > Then the SoC-specific .dtsi has no idea what mode should be used.
> > > > >
> > > > > Hence shouldn't the software derive the hardware mode from the full
> > > > > hardware description in DT? If that's impossible (I didn't look into detail
> > > > > whether an SPI bus can easily be distinguished from a UART bus), perhaps
> > > > > a mode property should be added?
> > > >
> > > > Yes, this is exactly what is done:
> > > >
> > > > https://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd.git/tree/drivers/mfd/at91-usart.c?h=ib-mfd-spi-tty-4.20-1#n33
> > >
> > > OK.
> > >
> > > I guess the main "hackish" part is that the mfd_cell uses of_compatible,
> > > which thus requires having additional compatible values?
> > >
> > > I think those can just be removed. AFAICS, the SPI and serial drivers already
> > > match against the "at91_usart_spi" resp. "atmel_usart_serial" platform device
> > > names?
> >
> > The hackish part of this driver is that it's using MFD for something
> > which is clearly not an MFD.  It's a USART device.  Nothing more,
> > nothing less.
> >
> > Does anyone have the datasheet to hand?
> 
> I haven't read it, but I believe it's not unlike Renesas SCIF, which is
> served by both drivers/tty/serial/sh-sci.c and drivers/spi/spi-sh-sci.c.
> But the latter is not used from DT, so we haven't experienced (and solved)
> the similar issue yet.
> 
> Would it work if the UART driver and SPI driver would match against the
> same compatible value, but the UART driver would do in its probe()
> function:
> 
>     device_property_read_u32(&pdev->dev, "atmel,usart-mode", &opmode);
>     if (opmode != AT91_USART_MODE_SERIAL)
>         return ENODEV;
> 
> while the SPI driver would do:
> 
>     device_property_read_u32(&pdev->dev, "atmel,usart-mode", &opmode);
>     if (opmode != AT91_USART_MODE_SPI)
>         return ENODEV;
> 
> ? No MFD driver involved.

I haven't looked at the code in a while, but if memory serves I
believe platform code gives up once it has found its first match, so
by doing this, one of the drivers will never be matched/probed.

It's midnight here, so cracking out the datasheet isn't going to
happen just now, but it's my current belief that if the IP serves 2
very different modes of operation, even if the registers are in a
shared space, they could have their own compatible strings in DT.

That is what the MFD driver provides after all.  Why would it be okay
to allocate different compatible strings from the MFD, but not in the
Device Tree?

It would be the easiest solution.

Has Rob commented on this yet?

-- 
Lee Jones [æçæ]
Linaro Services Technical Lead
Linaro.org â Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog