Re: [PATCH v1 1/1] power: Add simple gpio-restart driver

[Olof Johansson]

Hi,



On Tue, Aug 26, 2014 at 4:45 PM, David Riley <davidriley@xxxxxxxxxxxx> wrote:
> This driver registers a restart handler to set a GPIO line high/low
> to reset a board based on devicetree bindings.
>
> Signed-off-by: David Riley <davidriley@xxxxxxxxxxxx>
> ---
>  .../devicetree/bindings/gpio/gpio-restart.txt      |  48 +++++++
>  drivers/power/reset/Kconfig                        |   8 ++
>  drivers/power/reset/Makefile                       |   1 +
>  drivers/power/reset/gpio-restart.c                 | 142 +++++++++++++++++++++
>  4 files changed, 199 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/gpio/gpio-restart.txt
>  create mode 100644 drivers/power/reset/gpio-restart.c
>
> diff --git a/Documentation/devicetree/bindings/gpio/gpio-restart.txt b/Documentation/devicetree/bindings/gpio/gpio-restart.txt
> new file mode 100644
> index 0000000..7cd58788
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/gpio/gpio-restart.txt
> @@ -0,0 +1,48 @@
> +Driver a GPIO line that can be used to restart the system as a
> +restart handler.
> +
> +The driver supports both level triggered and edge triggered power off.
> +At driver load time, the driver will request the given gpio line and
> +install a restart handler. If the optional properties 'input' is
> +not found, the GPIO line will be driven in the inactive state.
> +Otherwise its configured as an input.
> +
> +When do_kernel_restart is called the various restart handlers will be tried
> +in order.

The above sentence documents the kernel behavior, not the hardware
description/binding.

> +The gpio is configured as an output, and drive active, so
> +triggering a level triggered power off condition. This will also cause an
> +inactive->active edge condition, so triggering positive edge triggered
> +power off.

> + After a delay of 100ms, the GPIO is set to inactive, thus
> +causing an active->inactive edge, triggering negative edge triggered power
> +off. After another 100ms delay the GPIO is driver active again. If the
> +power is still on and the CPU still running after a 3000ms delay, a
> +WARN_ON(1) is emitted.

It's possible that this behavior is inadequate for some hardware in
the future -- if so they can amend the binding (i.e. this comment is
an attempt at preemptive bikeshed avoidance :)

> +
> +Required properties:
> +- compatible : should be "gpio-restart".
> +- gpios : The GPIO to set high/low, see "gpios property" in
> +  Documentation/devicetree/bindings/gpio/gpio.txt. If the pin should be
> +  low to power down the board set it to "Active Low", otherwise set
> +  gpio to "Active High".
> +
> +Optional properties:
> +- input : Initially configure the GPIO line as an input. Only reconfigure
> +  it to an output when the machine_restart function is called. If this optional
> +  property is not specified, the GPIO is initialized as an output in its
> +  inactive state.

Isn't this the same as configuring the pin as tristate? I think that
should probably be controlled by pinmux setup instead?

> +- priority : A priority ranging from 0 to 255 (default 128) according to
> +  the following guidelines:
> +       0:      Restart handler of last resort, with limited restart
> +               capabilities
> +       128:    Default restart handler; use if no other restart handler is
> +               expected to be available, and/or if restart functionality is
> +               sufficient to restart the entire system
> +       255:    Highest priority restart handler, will preempt all other
> +               restart handlers

This is sort of leaking linux implementation, but it's also a useful
feature to have in the description. It seems sane enough to me to use.

> +
> +Examples:
> +
> +gpio-restart {
> +       compatible = "gpio-restart";
> +       gpios = <&gpio 4 0>;
> +       priority = /bits/ 8 <200>;

I think it makes sense to just have this as a regular cell instead of
doing an 8-bit value -- it's how we normally handle these elsewhere.


-Olof
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/