Re: [PATCH] ARM: AM33XX: board-generic: Add of_dev_auxdata to fix dev_id for CAN module

[Rob Herring <robherring2@gmail.com>]

On 08/07/2012 10:53 AM, Hiremath, Vaibhav wrote:
> On Tue, Aug 07, 2012 at 20:49:48, Rob Herring wrote:
>> On 08/07/2012 08:37 AM, Vaibhav Hiremath wrote:
>>> If the module requires only one clocksource to function, then
>>> driver can very well call clk_get() api with "con_id = NULL",
>>>
>>>        clk = clk_get(dev, NULL);
>>>
>>> And platform code should respect this and return valid clk handle.
>>> That means, now the clk_get() api would rely on dev_id, and platform
>>> code should either have clk node with matching dev_id (with con_id=NULL)
>>> or return dummy clk node.
>>>
>>> With DT support, we can fix the dev_id for particular module
>>> using "struct of_dev_auxdata" to satisfy above clk framework requirement.
>>>
>>> In case of AM33XX, we required this for the DCAN module, so this
>>> patch adds "of_dev_auxdata" for both DCAN_0/1 instances.
>>>
>>> Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
>>> Cc: Tony Lindgren <tony@atomide.com>
>>> Cc: Paul Walmsley <paul@pwsan.com>
>>> Cc: Benoit Cousson <b-cousson@ti.com>
>>> Cc: Grant Likely <grant.likely@secretlab.ca>
>>> ---
>>> This patch is boot tested on BeagleBone platform, and checked for
>>> clk_get() return value in d_can module driver.
>>>
>>>  arch/arm/mach-omap2/board-generic.c |   18 +++++++++++++++++-
>>>  1 files changed, 17 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/arch/arm/mach-omap2/board-generic.c b/arch/arm/mach-omap2/board-generic.c
>>> index 6f93a20..c9b7903 100644
>>> --- a/arch/arm/mach-omap2/board-generic.c
>>> +++ b/arch/arm/mach-omap2/board-generic.c
>>> @@ -37,11 +37,27 @@ static struct of_device_id omap_dt_match_table[] __initdata = {
>>>  	{ }
>>>  };
>>>
>>> +/*
>>> + * Lookup table for attaching a specific name and platform_data pointer to
>>> + * devices as they get created by of_platform_populate(). Ideally this table
>>> + * would not exist, but the current clock implementation depends on some devices
>>> + * having a specific name OR to satisfy NULL con_id requirement from driver.
>>> + */
>>> +static const struct of_dev_auxdata am33xx_auxdata_lookup[] __initconst = {
>>> +	OF_DEV_AUXDATA("bosch,d_can", 0x481cc000, "d_can.0", NULL),
>>> +	OF_DEV_AUXDATA("bosch,d_can", 0x481d0000, "d_can.1", NULL),
>>> +	{ },
>>> +};
>>
>> Adding an additional clkdev lookup accomplishes the same thing and is a
>> cleaner solution.
>>
> 
> That is also required and I have submitted patch for the same -
> 
> http://www.spinics.net/lists/arm-kernel/msg187998.html

That only adds the non-DT name. What I'm saying is you can have 2 lookup
names for the same clock.

> 
> With DT support, I am getting dev_id as -
> 
>  - Without "reg" property: d_can.16 and d_can.17,
>    (I believe it changes dynamically here)
> 
>  - With "reg" property: 481cc000.d_can and 481d0000.d_can
> 
> Which is not so intuitive, I would like to see them as per Spec/TRM, so 
> doesn't this patch make sense?

The spec doesn't have a base address for the module? This name is going
to get used in sysfs anyway, and the old name should be going away.

> Similar thing has already been done for other platforms too.

Only ones I haven't reviewed.

Rob