[PATCH 4/4] arm/dts: am33xx: Add CPSW and MDIO module nodes for AM33XX

[Cousson, Benoit]

On 11/1/2012 8:45 AM, Richard Cochran wrote:
> On Wed, Oct 31, 2012 at 04:17:27PM +0100, Benoit Cousson wrote:
>>> +			compatible = "ti,cpsw";
>>> +			ti,hwmods = "cpgmac0";
>>> +			cpdma_channels = <8>;
>>> +			host_port_no = <0>;
>>> +			cpdma_reg_ofs = <0x800>;
>>> +			cpdma_sram_ofs = <0xa00>;
>>> +			ale_reg_ofs = <0xd00>;
>>> +			ale_entries = <1024>;
>>> +			host_port_reg_ofs = <0x108>;
>>> +			hw_stats_reg_ofs = <0x900>;
>>> +			bd_ram_ofs = <0x2000>;
>>> +			bd_ram_size = <0x2000>;
>>> +			no_bd_ram = <0>;
>>> +			rx_descs = <64>;
>>> +			mac_control = <0x20>;
>>
>> Do you have to store all these data in the DTS? Cannot it be in the driver?
>>
>> Do you expect to have several instance of the same IP with different
>> parameters here?
>
> As I understand it, there are only two different layouts for the CPSW,
> the one in the dm814x and the one in the am335x. So I think it would
> work to put only the version register offet in the DT, and the let the
> driver figure out the rest from there.

Yes, that's indeed better. We did that for other IPs already (GPIO, I2C...)

> But if TI is planning on reordering the registers with each new
> silicon revision, again and again, then it might make sense to keep
> the offsets in the DT.

Yeah, let's assume they will do a better job in the future.

All these offset registers information does belong to the driver, and 
even if the HW change a lot, I still rather hide that in the driver.
It will always be cleaner, most efficient, and will reduce the size if 
the blob.

> [ I really wonder why the hardware people think that reshuffling the
>    register layout constitutes an improvement. ]

I've been wondering that for ten years :-(
I'm always hoping it will be better some day.

Regards,
Benoit