gpio-mxc gpio get always returns 0 for outputs for IMX6

[Benoît Thébaudeau]

On Wed, Jul 16, 2014 at 8:01 AM, Shawn Guo <shawn.guo at freescale.com> wrote:
> On Tue, Jul 15, 2014 at 10:28:52AM -0700, Tim Harvey wrote:
>> On Mon, Jul 14, 2014 at 7:05 AM, Shawn Guo <shawn.guo at freescale.com> wrote:
>> > On Mon, Jul 14, 2014 at 02:56:06PM +0200, Benoît Thébaudeau wrote:
>> >> > So the question comes down to, for an output GPIO, whether the value in
>> >> > GPIO_DR register is always identical to what we see on the pad.  If yes,
>> >> > your patch makes sense to me.  Otherwise, we have to set SION bit to
>> >> > read the value of an output GPIO from GPIO_PSR.
>> >>
>> >> Both are not always the same. They should be the same only for sane
>> >> hardware. GPIO_DR indicates the level that the output pad tries to
>> >> drive, while GPIO_PSR indicates the level sensed on the pad. Thus,
>> >> they may differ if there is an external level conflict on the pin.
>> >> This difference may be useful in order to test if the board is
>> >> behaving as expected or if there is anything going wrong, which is
>> >> typically useful for board prototyping.
>> >>
>> >> The software "knows" what it has set on the output pad, so having a
>> >> get accessor for GPIO_PSR is more useful than for GPIO_DR, and it
>> >> better fits the actual meaning of this function. GPIO_DR would
>> >> correspond to the non-existing gpio_get_set_value() function.
>> >
>> > Thanks for the input, Benoît.  That's helpful.
>> >
>> > Shawn
>>
>> Great - thanks for the clarification everyone. I will not submit such
>> a patch and instead work this from the pinmux perspective setting SION
>> on GPIO outputs.
>>
>> That said, can anyone shed light on the subject of what is 'proper'
>> for GPIO pinmux in the bootloader vs the kernel device-tree? I see
>> that among the various imx6qdl*.dtsi files, some boards configure
>> MX6QDL_PAD_GPIO* as 0x80000000 (no config, keeping power-on or
>> bootloader settings) instead of manually configuring them. Perhaps
>> there is some rule of thumb that should be followed?
>
> Although there are quite some cases where 0x80000000 is being used, we
> should really configure it properly to avoid the dependency on
> bootloader or reset value.
>
> Instead of specifying SION setup in the config cell case by case, I
> think we can do that directly in pin function ID for all cases where
> GPIO mode is selected.  I just sent a patch to do that.  Please take
> a look to see if it works for you.

I don't find this patch on the ML.

Note that enabling SION for all pads in GPIO mode will have some
impact on the power consumption. That's why SION is designed to be
used on a per-pad basis, only when an output GPIO needs to be sensed.

Benoît