[PATCH v8 4/9] davinci: McASP configuration for Omapl138-Hawkboard

[Victor Rodriguez]

On Thu, Nov 18, 2010 at 5:57 PM, Kevin Hilman
<khilman at deeprootsystems.com> wrote:
> "Nori, Sekhar" <nsekhar at ti.com> writes:
>
>> Hi Michael,
>>
>> On Wed, Nov 17, 2010 at 02:12:53, Michael Williamson wrote:
>>>
>>> Help me out.  Why do we need generic pin lists?
>>>
>>
>> They might help in cases where all boards will use the same set of
>> pins. For example, every one who uses I2C will most likely both the
>> clock and data pins from the IP. For more complex peripherals with
>> different pins options they serve a documentation purpose at best.
>>
>>> It seems to me that the "generic pin list" for da850.c isn't practical for most
>>> (if not all) of the peripherals.  They should be done using __initdata in
>>> each board file.
>>
>> Yes, agreed.
>>
>>>
>>> Just a cursory glance at what's in da850.c highlights several items being set
>>> up for the EVM and not generically.  For example:
>>>
>>> - da850_uart1_pins and da850_uart2_pins: I believe both have RTS/CTS pins which
>>>   for a generic definition should be included as for UART0, but would then
>>>   be unused as the EVM doesn't use these pins in this function.
>>
>> Yes, the generic pin list should have RTS and CTS pins defined for UART1
>> and UART2. This needs fixing.
>>
>>>
>>> - da850_mcasp_pins: if generic, must include all 16 AXR pins.  I think you'd
>>>   be hard pressed to find a board configuration that would use all 16 AXR pins
>>>   for the McASP.  I'm fairly sure the EVM uses the pins called out, and uses
>>>   other pins for other functions.  So it's likely this structure wouldn't get used.
>>
>> Yes, the generic pin list should either be completed or removed
>> altogether and the existing pin list da850_mcasp_pins should be
>> copied into the board file and called da850_evm_mcasp_pins.
>>
>>>
>>> - da850_mmcsd0_pins : includes 2 GPIO pins (specific to the EVM, though possible for
>>>   other boards) for the card detect and write protect signals.  These pins are
>>>   completely arbitrary for that particular board design. I also believe that
>>>   the complete mmcsd0 port has 4 more data lines as part of it's peripheral, although
>>>   the driver doesn't support using them.
>>
>> This is incorrect again. The generic pin list should be completed
>> (or removed) and the existing list should be copied into the EVM board
>> file as da850_evm_mmcsd0_pins.
>>
>>>
>>> - da850_emif25_pins interface doesn't include the generic pins for some of
>>>   the SDRAM functions.
>>
>> Yes, this should be completed (or removed). This list is unused anyway.
>>
>>>
>>> - da850_cpgmac_pins defines both RMII and MII pins.  I don't think any board
>>>   would want to configure both sets at the same time.  Seems like this should
>>>   never get used...
>>
>> Agreed.
>>
>>>
>>> It's also incomplete.  What about the uPP pin list?  Or the VPIF?  Etc.
>>
>> These should be added as the drivers for these devices are
>> supported.
>>
>>>
>>> I think a board file author should be familiar enough with the SoC to understand
>>> what peripheral pins he should be configuring for his/her particular hardware setup
>>> and explicitly specify them in the board file.
>>
>> Agree.
>>
>>>
>>> If you remove the common pin-mux lists and move them to a board file, then once you
>>> configure your specific platform, is there any more memory used than with
>>> the common scheme?  Of course, there would be replication of pin-mux code in the board
>>
>> There is no memory wastage. All the pin lists are init data.
>>
>> I too prefer all generic pin lists which are most likely not
>> going to be used at all to be removed. Unused stuff like this
>> will only make code difficult to read.
>
> FWIW, I agree.
>
> Now, who wants to tackle it?
>
> Kevin
>

I could but I will need a little help and from all of you. :)

Regards

Victor Rodriguez


> _______________________________________________
> Davinci-linux-open-source mailing list
> Davinci-linux-open-source at linux.davincidsp.com
> http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
>