Re: [PATCH net-next 1/2] net: dsa: Add support for a switch reset gpio

[Andrew Lunn <andrew@lunn.ch>]

On Thu, Nov 19, 2015 at 09:39:11AM +0100, Neil Armstrong wrote:
> Hi Andrew,
> 
> On 11/19/2015 12:29 AM, Andrew Lunn wrote:
> > +		gpio = of_get_named_gpio_flags(child, "reset-gpios", 0,
> > +					       &flags);
> > +		if (gpio_is_valid(gpio)) {
> > +			ret = devm_gpio_request_one(dev, gpio, flags,
> > +						    "switch_reset");
> > +			if (ret)
> > +				goto out_free_chip;
> > +
> > +			cd->reset = gpio;
> > +			cd->reset_flags = flags;
> > +			off = (flags && OF_GPIO_ACTIVE_LOW ? 1 : 0);
> > +			gpio_direction_output(cd->reset, off);
> > +		}
> > +
> >  		for_each_available_child_of_node(child, port) {
> >  			port_reg = of_get_property(port, "reg", NULL);
> >  			if (!port_reg)
> > 
> 
> You could also use :
> gpio = of_get_named_gpio(child, "reset-gpios", 0)
> devm_gpio_request(dev, gpio, "switch_reset")
> 
> and :
> cd->reset = gpio_to_desc(gpio);
> 
> and cd->switch reset to struct gpio_desc *reset
> 
> to use the gpiod calls afterward. The flags are no more needed.

Hi Neil

I really wanted to use gpiod, since as you said, it takes care of
active low/active high. I tried all sorts of combinations. The problem
with this one is that desc->flags is set in gpiod_parse_flags(), which
is only called from gpiod_get_index(), which is limited to gpios in
the root of the devices subtree. I was 'lucky' in that my reset is
active low, so i noticed the problem when my switch failed to probe,
being held in reset, because the ACTIVE_LOW flag in DT is being ignored.

It seems like gpiod is backwards compatible to gpio, but gpio is not
forward compatible to gpiod.

	Andrew