[PATCH V3 3/5] ARM: tegra:add aliases and DMA requestor for serial nodes of Tegra114

[Stephen Warren]

On 03/15/2013 12:42 PM, Laxman Dewangan wrote:
> On Friday 15 March 2013 11:56 PM, Stephen Warren wrote:
>> On 03/13/2013 02:02 PM, Stephen Warren wrote:
>>> On 03/13/2013 01:49 PM, Laxman Dewangan wrote:
>>>> Add APB DMA requestor and serial aliases for serial controller.
>>>> There will be two serial driver i.e. 8250 based simple serial driver
>>>> and APB DMA based serial driver for higher baudrate and performace.
>>>>
>>>> The simple serial driver get enabled with compatible
>>>> "nvidia,tegra114-uart",
>>>> "nvidia,tegra20-uart" and APB DMA based driver will get enabled with
>>>> compatible "nvidia,tegra114-hsuart", "nvidia,tegra30-hsuart".
>>>> +    /*
>>>> +     * There are two serial driver i.e. 8250 based simple serial
>>>> +     * driver and APB DMA based serial driver for higher baudrate
>>>> +     * and performace. To enable the 8250 based driver, the compatible
>>>> +     * is "nvidia,tegra114-uart", "nvidia,tegra20-uart" and to enable
>>>> +     * the APB DMA based serial driver, the comptible is
>>>> +     * "nvidia,tegra114-hsuart", "nvidia,tegra30-hsuart".
>>>> +     */
>>> Again, that text says you want either of:
>>>
>>> compatible = "nvidia,tegra114-uart",   "nvidia,tegra20-uart";
>>> compatible = "nvidia,tegra114-hsuart", "nvidia,tegra30-hsuart";
>>>
>>> (note Tegra20-vs-Tegra30 in the second compatible value)
>>>
>>> Why isn't it instead:
>>>
>>> compatible = "nvidia,tegra114-uart",   "nvidia,tegra20-uart";
>>> compatible = "nvidia,tegra114-hsuart", "nvidia,tegra20-hsuart";
>>>
>>> (note both second compatible values say Tegra20)
>>>
>>> I assume this is a typo.
>>>
>>> I suppose I can fix this up when I apply it to avoid a resend, assuming
>>> it's wrong.
>> Since I haven't seen a reply to this, when I apply this, I'm going to
>> change the comment I quoted above to match the values I wrote above
>> under "why isn't it instead:".
> 
> Stephen,
> Sorry, I missed your comment to reply.
> I mean was that compatible should be
> compatible = "nvidia,tegra114-uart",   "nvidia,tegra20-uart";
> compatible = "nvidia,tegra114-hsuart", "nvidia,tegra30-hsuart";
> 
> not
> 
> compatible = "nvidia,tegra114-uart",   "nvidia,tegra20-uart";
> compatible = "nvidia,tegra114-hsuart", "nvidia,tegra20-hsuart";
> 
> 
> The reason is that, tegra30 has the clock divider in the CAR register
> set and it is 15.1 which gives more precise baudrate. tegra20 does not
> have the same.
> Tegra114 also have the clock divider in the CAR register.
> 
> All SoCs UART support 16.0 clock divider inside the uart controller as
> DLL/DLM.
> 
> Simple uart driver use the uart clock divider and it is fine here.
> 
> High speed uart driver uses the car register driver for better
> flexibility and better resolution.

OK, so I see that Tegra30 has an enhancement over Tegra20. However,
given your description, that enhancement is optional; a driver could
simply continue to use /just/ the in-UART divider, and ignore the CAR
divider, and still work just fine, albeit with (entirely
backwards-compatible) less accuracy than it might achieve if it used the
new feature.

As such, I think it's correct to mark the device as actually being
compatible with all 3: 114 (precise HW model), 30 (base model w/ extra
divider), 20 (base model that's compatible, albeit ignoring extra features).

That might be a bit excessive though, so I guess I'll just go with the
values in your patch. It'd be a good idea if you could post a follow-on
patch that updates the DT binding to explain this, and then removes the
comments from *.dtsi since this really should be explained in the
binding document not the .dtsi files, I think.

At most, I'd expect to see the following in the .dtsi files:

These nodes can either be compatible with nvidia,tegra114-uart, or
nvidia,tegra114-hsuart. See the bindings for details of the difference.