On Sat, Nov 24, 2012 at 11:13:28AM +0900, Sangbeom Kim wrote:
There's some problems with this binding. The main one is the gpios property the format of which isn't specified at all.
All of above gpio property is i2s. That is,
- gpios = <&gpz 0 2 0 0>, -> SCLK
<&gpz 1 2 0 0>, -> CDCLK<&gpz 2 2 0 0>, -> LRCK<&gpz 3 2 0 0>, -> SDI<&gpz 4 2 0 0>, -> SDO[0]<&gpz 5 2 0 0>, -> SDO[1]<&gpz 6 2 0 0>; -> SDO[2]
Do you want like a below one? +sclk-gpios = <&gpz 0 2 0 0>, +cdclk-gpios = <&gpz 1 2 0 0>, ...
That would be nice but what I was really looking for was some indiciation in the binding document as to what all this stuff means - it should really say what gpz is (though I can figure that out) and it definitely needs to say what the four numbers are.
The requirement for an alias is also very odd, where does that come from?
I don't know that Which one is odd. Please let me know.
Having them at all is odd - it's not something other DT bindings have needed and there was nothing saying what it was for.
Some of the code also looks very peculiar, like the fact that it's generating a clock name i2s_opclk%d rather than hard coding the clock, the physical clock would normally be resolved based on the struct device.
This is to handle all of Samsung SOCs i2c clock mux. Please look at below clk_lookup table
In case of 6410, clk_lookup
- CLKDEV_INIT("samsung-i2s.0", "i2s_opclk0", &clk_i2s0),
- CLKDEV_INIT("samsung-i2s.0", "i2s_opclk1", &clk_audio_bus0.clk),
- CLKDEV_INIT("samsung-i2s.1", "i2s_opclk0", &clk_i2s1),
- CLKDEV_INIT("samsung-i2s.1", "i2s_opclk1", &clk_audio_bus1.clk),
+#ifdef CONFIG_CPU_S3C6410
- CLKDEV_INIT("samsung-i2s.2", "i2s_opclk0", &clk_i2s2),
- CLKDEV_INIT("samsung-i2s.2", "i2s_opclk1", &clk_audio_bus2.clk),
In case of exynos5, clk_lookup
- CLKDEV_INIT("samsung-i2s.0", "i2s_opclk0", &exynos5_clk_sclk_i2s.clk),
- CLKDEV_INIT("samsung-i2s.0", "i2s_opclk1", &exynos5_clk_i2s_bus.clk),
We try to handle clock source of i2s by only i2s_opclk0 and i2s_opclk1. Each SOCs have different clock source. Is this wrong approach?
That makes sense but why is the driver not just requesting by name rather than having code to generate the names, and why is this mixed in with the patch adding DT support?