RE: [PATCH 2/2] ARM: dts: AM33XX: Add lis331dlh device tree data to am335x-evm

+Daniel

On Fri, Sep 14, 2012 at 15:58:37, Arnd Bergmann wrote:

On Friday 14 September 2012, AnilKumar, Chimata wrote:

quoted

quoted

On Fri, Sep 14, 2012 at 13:56:06, Arnd Bergmann wrote:
On Thursday 13 September 2012, AnilKumar Ch wrote:

quoted

quoted

Why do you put the "reg" property here

Here I specified reg property because lis331dlh I2C slave address is 0x18.

quoted

quoted

 		dcan1: d_can at 481d0000 {
 			status = "okay";
 			pinctrl-names = "default";

@@ -61,6 +70,39 @@
 		regulator-max-microvolt = <5000000>;
 		regulator-boot-on;
 	};
+
+	lis3_reg: fixedregulator at 1 {
+		compatible = "regulator-fixed";
+		regulator-name = "lis3_reg";
+		regulator-boot-on;
+	};
+};
+&lis331dlh {
+	compatible = "st,lis3lv02d-i2c";

and all the rest here? At least I would expect the "compatible" property
to be in the same place above.

This data is appended to above one, to make it readable I moved remaining
properties to here.

I don't follow how this is making things more readable.

Basic lis3 parameter is in i2c2 node to tell the hierarchy of
nodes and remaining parameters at the bottom. With this way
we can understand easily what is start & end braces so that
node hierarchy is cleaner. (This is my view)

Maybe a more logical way to do this would be use the existing i2c2 label
and write all the additions as

i2c2: {
	status = "okay";
	clock-frequency = <400000>;

	lis331dlh at 18 {
		compatible = "st,lis3lv02d";
		reg = <0x18>;

		vdd-supply = <&lis3_reg>;
		vdd-io-supply = <&lis3_reg>;
		...
	};

If this is better/preferable way then I will change.

quoted

quoted

Also, I think you should remove the "-i2c" postfix from the name, that
is already implied by the parent bus.

I will remove, but in case of spi the compatible name is lis3lv02d_spi.
By mistake I have uses "-i2c" instead of "_i2c".

Then, I will change to 'st,lis3lv02d' in lis3lv02d_i2c driver
and same will added to .dts file.

Small question here, in my v2 version I have specified both
the compatible names lis3lv02d and lis331dlh is it fine or
only one is sufficient?

+static struct of_device_id lis3lv02d_i2c_dt_ids[] = {
+	{ .compatible = "st,lis3lv02d" },
+	{ .compatible = "st,lis331dlh" },
+	{}
+};

The normal convention is to use '-', not '_', so that part was ok.

I think naming the other one lis3lv02d_spi was a mistake, it should
be named 'st,lis3lv02d' independent of the bus IMHO.

Yes, initially I expected the same. I think driver name is used
as a compatible string.

quoted

Document is already present, 
http://git.kernel.org/?p=linux/kernel/git/next/linux-next.git;a=commit;h=2f2ff3cc8d930493f9a598b9192706c09403e12e

Some minor changes in docs, in my next version I will update document
as well. I will send V3 if there are no comments on v2.

Can you just quickly review v2 series? So that I will send
V3.

Thanks
AnilKumar

On Friday 14 September 2012, AnilKumar, Chimata wrote:

Small question here, in my v2 version I have specified both
the compatible names lis3lv02d and lis331dlh is it fine or
only one is sufficient?

+static struct of_device_id lis3lv02d_i2c_dt_ids[] = {
+       { .compatible = "st,lis3lv02d" },
+       { .compatible = "st,lis331dlh" },
+       {}
+};

That's ok. In most cases, people will need the data field to point to a
data structure with the differences between two chips, but not if they are
identical from the software side.
It's also fine if you just list one entry here and put both values as
"compatible" in the device tree, to signify that the device is backwards
compatible with the older variant.

	Arnd