[Tinyos-host-mote-wg] Re: [Tinyos-2.0wg] TOS2 tasks code size

[Philip Levis]

On Oct 9, 2005, at 2:55 PM, Cory Sharp wrote:

> On 10/9/05, Philip Levis <pal at cs.stanford.edu> wrote:
>
>> Ah, that is interesting. This sounds like something that the nesC
>> compiler could know and optimize for (as it used to).
>>
>> The one trick is that it's now possible to post a task through
>> directly accessing the interface and using an ID. Hrm.
>>
>
> Well, it's not the posting that's the problem, but how the event
> handler ("runTask") function body is referenced.
>
> In TOS1, only the individual post commands reference their respective
> task bodies.  So, no post, then no reference to the task body, and the
> task is elided.  The scheduler only keeps runtime reference to the
> task bodies.
>
> With TOS2, the scheduler dispatcher has a full table (of some form) of
> all runTask events.
>
> I see no easy path to elide entries from the task body table except to
> 1) somehow be able to tell nesC that runTask is an exclusive
> consequence of postTask.  Or, 2) get rid of the task body table, have
> postTask pass reference to its runTask, and have the scheduler keep a
> runtime array of runTask references.

The compiler could not automatically generate interface wirings and  
entries for tasks that are not posted...

Phil
_______________________________________________
Tinyos-2.0wg mailing list
Tinyos-2.0wg at Mail.Millennium.Berkeley.EDU
https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-2.0wg