[RFC PATCH 5/7] ARM: davinci: i2c: add OF support

[Heiko Schocher]

Hello Grant,

Grant Likely wrote:
> On Tue, Jan 24, 2012 at 10:51:50AM +0100, Sylwester Nawrocki wrote:
>> Hello Heiko,
>>
>> On 01/24/2012 08:18 AM, Heiko Schocher wrote:
>>>> On 01/23/2012 09:56 AM, Heiko Schocher wrote:
>>>>> add of support for the davinci i2c driver.
>>>>>
>>>>> Signed-off-by: Heiko Schocher<hs at denx.de>
>>>>> Cc: davinci-linux-open-source at linux.davincidsp.com
>>>>> Cc: linux-arm-kernel at lists.infradead.org
>>>>> Cc: devicetree-discuss at lists.ozlabs.org
>>>>> Cc: linux-i2c at vger.kernel.org
>>>>> Cc: Ben Dooks<ben-linux at fluff.org>
>>>>> Cc: Wolfram Sang<w.sang at pengutronix.de>
>>>>> Cc: Grant Likely<grant.likely at secretlab.ca>
>>>>> Cc: Sekhar Nori<nsekhar at ti.com>
>>>>> Cc: Wolfgang Denk<wd at denx.de>
>>>>> ---
>>>>>   .../devicetree/bindings/arm/davinci/i2c.txt        |   39 ++++++++++++++++++
>>>>>   drivers/i2c/busses/i2c-davinci.c                   |   43 ++++++++++++++++++++
>>>>>   2 files changed, 82 insertions(+), 0 deletions(-)
>>>>>   create mode 100644 Documentation/devicetree/bindings/arm/davinci/i2c.txt
[...]
>> Moreover you seem to overwrite platform device name and id,
>>
>> 		if (!of_property_read_u32(pdev->dev.of_node, "id",
>> +			&prop)) {
>> +			pdev->id = prop;
>> +			pdev->dev.init_name = kzalloc(20, GFP_KERNEL);
>> +			sprintf((char *)pdev->dev.init_name,
>> +				"i2c_davinci.%d", pdev->id);
>>
>> I'm not sure if it is a good practice.
> 
> No, it is not good practice.  At this point the device is already
> registered.  Changing the value of pdev->id will cause problems in the
> core code because the /sysfs files will have already been created.
> 
>> If you want to pre-define platform
>> device name (likely for the clock API to work), it might be more appropriate
>> to use OF_DEV_AUXDATA in the machine code, until there are clock bindings
>> available.
> 
> Yes, use OF_DEV_AUXDATA, but *only* if you need it for hooking up
> clocks, regulators, or similar.  Don't use it just because you want
> the device to have a different name for cosmetic reasons.  The need
> for AUXDATA should go away once the DT clock binding code is merged.

I now use OF_DEV_AUXDATA in board code, "id" nor "cell-index" used:

struct of_dev_auxdata enbw_cmc_auxdata_lookup[] __initdata = {
        OF_DEV_AUXDATA("ti,davinci-i2c", 0x01c22000, "i2c_davinci.1", NULL),
        OF_DEV_AUXDATA("ti,davinci-i2c", 0x01e28000, "i2c_davinci.2", NULL),
        {}
};

This is needed for hooking up clocks, see:
arch/arm/mach-davinci/da850.c
static struct clk_lookup da850_clks
[...]
        CLK("i2c_davinci.1",    NULL,           &i2c0_clk),
[...]
        CLK("i2c_davinci.2",    NULL,           &i2c1_clk),

bye,
Heiko
-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany