[Tinyos-host-mote-wg] Re: [Tinyos-2.0wg] Resource Allocation RoundRobinArbiter

Kevin Klues klues at tkn.tu-berlin.de
Tue Apr 26 11:22:56 PDT 2005


I knew I wasn't exactly following the TEP here, and I planned on 
bringing up in the meeting on Wednesday why I chose to implement it 
differently.  We can talk about whether how I did it is better or worse 
and then either the implementation or the TEP can be changed based on 
what is said.

Kevin

David Gay wrote:

>On 4/21/05, Kevin Klues <klues at tkn.tu-berlin.de> wrote:
>  
>
>>The Resource and the ResourceUser interfaces according to TEP108 have
>>been committed to the interfaces directory.  The resource interface has
>>changed slightly to include an event for when the bus is released.  I
>>don't know how that wasn't included in the first place, but it cannot
>>work properly without it.  This needs to be added into the TEP.
>>    
>>
>
>See comments below.
>
>  
>
>>A generic RoundRobinArbiter module providing the Resource and the
>>ResourceUser interfaces has been implemented and committed to the sytems
>>directory.  A test application for verifying that it works has been
>>committed to the apps/TestArbiter directory.  It has only been tested on
>>the eyesIFX nodes and needs verification on the telos and micas.
>>    
>>
>
>There's a problem with this implementation: it's not actually
>implementing TEP 108. Specifically:
>
>1) If the resource is available when you call request, you're supposed
>to get a SUCCESS return and not get a granted event. The
>implementation does signal granted in this case.
>
>2) If the resource is not available when you call request, you're
>supposed to eventually get a granted event (yes, this requires a queue
>of pending requests). The implementation will never signal this.
>
>3) The "released" event is not necessary if 2) is done according to the TEP...
>
>4) request and release are async in the implementation, and not in the
>TEP (I'm fairly sure we should change the TEP here).
>
>So the basic question is: do we want to change the implementation, or
>should we be changing the TEP? (i.e., does the TEP specify the wrong
>thing? If so, we should have a discussion with the reasons...)
>
>David Gay
>
>_______________________________________________
>Tinyos-2.0wg mailing list
>Tinyos-2.0wg at Mail.Millennium.Berkeley.EDU
>http://Mail.Millennium.Berkeley.EDU/mailman/listinfo/tinyos-2.0wg
>
>
>  
>


_______________________________________________
Tinyos-2.0wg mailing list
Tinyos-2.0wg at Mail.Millennium.Berkeley.EDU
http://Mail.Millennium.Berkeley.EDU/mailman/listinfo/tinyos-2.0wg


More information about the Tinyos-host-mote-wg mailing list