[RFC PATCHv4 1/6] arm: TI-Nspire platform code

Daniel Tang dt.tangr at gmail.com
Wed May 29 01:14:41 EDT 2013


On 28/05/2013, at 1:15 AM, Arnd Bergmann <arnd at arndb.de> wrote:

> On Monday 27 May 2013, Daniel Tang wrote:
>> Before any peripheral is accessed. I.e. before the clocksource and irqchip
>> drivers.
> 
> The irqchip comes first, and by that time, you can actually call
> of_iomap().
> 
>> The write to the port is supposed to ensure all mmio peripherals can be
>> accessed. Without it, access to certain peripherals will result in
>> undefined reads or ignored writes.
>> 
>> On second thoughts, would this actually be the job of the boot loader?
> 
> Doing it in the boot loader would certainly simplify things. I wonder
> about the dynamic aspects of power management though: It might be
> better to expose the individual bits of this register through a proper
> driver. The boot loader can start out enabling everything, but then
> you turn off everything that is not needed when that driver gets
> loaded.

That's the idea for the long term.

For now though, I'll probably just let the bootloader enable everything and work on a proper driver for power management later.

> 
> I'm still not sure what the register actually does: Does it
> control reset lines, clock signals, voltage regulators or something
> else? These things all have their own subsystems, and then there
> is also the power domain framework.

To be perfectly honest, I'm not too sure. The documentation for the TI-Nspire is all gathered from reverse engineering and all it says is that register "disables bus access to peripherals".

> 
> 	Arnd

Cheers,
Daniel Tang


More information about the linux-arm-kernel mailing list