>Четверг, 8 февраля 2018, 11:58 +03:00 от Linus Walleij <linus.walleij@xxxxxxxxxx>:
>
>Hi Alexander,
>
>I converted your driver for GPIO watchdog to use GPIO descriptors
>in the last merge window.
Hi!
Unfortunately I can not test the changes on the hardware in the near future.
In any case, Your changes look reasonable. If something does not work as expected,
we can always rollback this change, therefore, if it possible, it is better to make this
change a separate patch.
>It seems there is no upstream device tree for this driver, so I guess
>you have some system using it, and it would be nice to hear if you
>can confirm that it still works for you.
>
>I was thinking a bit about this code:
>
> /* Put GPIO back to tristate */
> if (priv->hw_algo == HW_ALGO_TOGGLE)
> gpiod_direction_input(priv->gpiod);
>
>I think the right thing is to remove these lines and instead use
>this in the device tree:
>
>gpios = <&gpio0 5 (GPIO_ACTIVE_HIGH|GPIO_OPEN_DRAIN)>;
>
>Also we should update the DTS example to use open drain.
>
>This will fall back to software emulation setting the line as input
>to drive it high just like the driver does if the hardware doesn't have
>explicit open drain support.
>
>The semantic effect would be that the open drain semantic will
>be used also when pinging the watchdog, i.e the line will be set
>as input to pull the line high. AFAICT this should work, but
>obviously it needs to be tested on hardware.
---
��.n��������+%������w��{.n�����{���rh���ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
