[Tinyos-devel] GeneralIO interface

[Philip Levis]

On Sep 11, 2006, at 1:48 PM, Martin Leopold wrote:

> Hi All.
> I've started implementing an 8051 platform for TOS 2.0 (in a separate
> repository) and first up in my list of things to do is the pins
> definitions, which leads me to a few questions:
>
>  * The way I understood TEP 117 each platform must export a HIL level
> interface to the pins conforming to the GeneralIO interface and should
> do so in the config GeneralIOC. Maybe I'm blind, but the only place  
> that
> I can find such a config was in chips directory for the pxa27x chip
> (iMote?).
>
> As far as I can tell neither atm128 nor msp430 have a similar
> definition, but only export HPL level interfaces (even though atm128
> reuses the same interface the file i still named Hpl...).
>
> Am I missing something here?
>
>  * The atm128 and msp430 implementations both seem to create a "port"
> abstraction somewhere in the gut (grouping 8 pins into a port) only to
> use this to export a long list of named pins. To me it seems that the
> ports abstraction is mostly unused, and is not exported to upper
> components, which seems a shame - since it sounds rather intuitive. Is
> there some practical reason for not doing that?
>
>  * In TEP 117 I can't seem to read how to export the entire set of  
> pins
> - the way I read it it only says that I must use the GeneralIO
> interface. In ./tos/chips/pxa27x/gpio/GeneralIOC.nc this seems to be
> done by parameterised interface, it that the way to go about it,  
> rather
> than say exporting a list of port-names (like atm128 does for HPL)?

I think the issue you're encountering is that 117 is a very initial  
draft. Phil B. wrote it as part of implementing the pxa27x, and the  
WG has yet to fully review it and bring other platforms in line with  
it. Given how critical and low-level it is, I think it makes sense to  
push it forward in the schedule: I'll put it in front of 111 and put  
it on the WG agenda this week.

There's a bit of a tradeoff here. On one hand, the WG would like  
people to be able to read TEPs and give feedback as early as  
possible. On the other, it needs to be clear what TEPs are solid and  
unlikely to change and what TEPs are early drafts. As we move more  
TEPs to Final state this will become clearer, but right now we have  
many in process and none that are yet finalized.

Phil