On 05/03/2012 06:24 AM, Wolfram Sang wrote:

On Tue, May 01, 2012 at 11:23:31AM -0600, Stephen Warren wrote:

quoted

From: Stephen Warren <redacted>

This is useful for SoCs whose I2C module's signals can be routed
to different sets of pins at run-time, using the pinctrl API.

+-----+  +-----+ | dev |  | dev | +------------------------+
+-----+  +-----+ | SoC                    |      |        | |
/----|------+--------+ |   +---+   +------+     | child bus A, on
first set of pins |   |I2C|---|Pinmux|     | |   +---+   +------+
| child bus B, on second set of pins |
\----|------+--------+--------+ |                        |      |
|        | +------------------------+  +-----+  +-----+  +-----+ 
| dev |  | dev |  | dev | +-----+  +-----+  +-----+

Signed-off-by: Stephen Warren <redacted> Acked-by:
Linus Walleij [off-list ref] --- v3: Renamed
pinctrl-i2cmux.c to i2c-mux-pinctrl.c to match recent changes to
other I2C mux files.

Thanks for doing the rename already. While I think I could follow
your argument regarding the compatible-binding, I'd still like to
have an ack from one of the device tree maintainers. The bindings
are non-trivial and I am missing the insight to judge them.

Rob, Grant, any chance you could take a quick look at the bindings for
this? I'd love to see the driver in 3.5 so I could make use of it for
Tegra in 3.6 without cross-subsystem dependencies. Thanks.