Re: Pinmux bindings proposal V2

* Shawn Guo [off-list ref] [120126 22:37]:

On Thu, Jan 26, 2012 at 09:51:23AM -0800, Tony Lindgren wrote:

quoted

Hi,

* Shawn Guo [off-list ref] [120126 00:53]:

quoted

On Mon, Jan 23, 2012 at 01:00:52PM -0800, Tony Lindgren wrote:
...

quoted

So to summarize: I suggest we'll just stick to basics to get the system
booting and devices working using device tree. In most cases the device
drivers should be able to configure the suspend and off states in a generic
way using pinctrl API. Everything else, like debugging, we can probably
do with userspace tools.

This would mean just using a minimal subset of your binding, probably
very close to what you originally suggested.

IMHO, as a generic device tree binding, it should be able to cope with
different use cases.  It's really free for you to use the minimal
subset of the binding as your need, but we should not make the binding
design just be that minimal subset to force that everyone else can
only use the minimal subset.

The main issue I have is that the example posted in this thread repeats
the same registers five times for one driver entry alone in the device
tree data. The repeated registers are TEGRA_PMX_PG_DTA and TEGRA_PMX_PG_DTD
in the example.

The alternative values are something that the pinmux/pinconf driver can
set based on state changes communicated from the driver using these pins.

I'm against the idea that requires client device drivers call pinctrl
APIs on pin basis.  They do not even know or care about the pins that
they are using.  That's all pinctrl driver's business.  All client
driver needs to do when it wants to change pinctrl setup should just
be calling pinmux/pinconf_get/put[enable/disable](dev, "state").

That should work fine too. As long as the states are generic, such as
PMX_STATE_ENABLED, PMX_STATE_IDLE, PMX_STATE_SUSPEND, PMX_STATE_OFF
and so on. You really don't want to change states with some string
names if you need to toggle something every time your enter and exit
idle. And it should be possible to set it on per pin or pingroup basis.
For wake-up events you typically just toggle one pin for the device,
such as serial rx line.

I guess the analog we should follow here is clk_get and clk_set_rate,
except we would have pinconf_set_state with predefined states.

Regards,

Tony