Re: [PATCH v2 4/4] ARM: dts: mt8135: Add pinctrl node for mt8135.

[Sascha Hauer <s.hauer@pengutronix.de>]

Hi Linus,

On Wed, Sep 24, 2014 at 01:23:09PM +0200, Linus Walleij wrote:
> On Tue, Sep 23, 2014 at 5:39 AM, Hongzhou.Yang
> <srv_hongzhou.yang@mediatek.com> wrote:
> 
> > From: Hongzhou Yang <hongzhou.yang@mediatek.com>
> >
> > Add pinctrl node to mt8135.dtsi.
> >
> > Signed-off-by: Hongzhou Yang <hongzhou.yang@mediatek.com>
> (...)
> > +#define MT8135_PIN_0_MSDC0_DAT7__FUNC_GPIO0 (MT_PIN_NO(0) | 0)
> > +#define MT8135_PIN_0_MSDC0_DAT7__FUNC_MSDC0_DAT7 (MT_PIN_NO(0) | 1)
> > +#define MT8135_PIN_0_MSDC0_DAT7__FUNC_EINT49 (MT_PIN_NO(0) | 2)
> > +#define MT8135_PIN_0_MSDC0_DAT7__FUNC_I2SOUT_DAT (MT_PIN_NO(0) | 3)
> 
> I haven't got to reviewing the driver, but this looks just wrong.
> 
> Have the magic numbers in the driver.
> 
> Use strings to describe functions, not integers.

Interrupts, clocks, gpios, dma channels, nearly everything in the device tree is
arbitrarily numbered. Instead of "irq-i2c0" we have <0 36 IRQ_TYPE_LEVEL_HIGH>
in the device tree. These numbers can be resolved efficiently in the
driver by shifting them to get a bitmask or by adding them as offset to
a register base.
Why do you want to make pinctrl different? Thanks to the recently
introduced defines in the device trees these numbers are not magic at
all anymore.

> 
> We need to move toward standardized device tree bindings
> for this stuff, and that means using strings, not magic
> numbers.

Agreed for standardized device tree bindings, but not for using strings.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |