If radio is busy, I think CC2420Config.sync will fail, or may return SUCCESS and signal syncDone at some time later. I do not understand what you said "if sync were to never signal done, then it is unlikely that the radio is receiving/transmitting packets." Did you mean if syncDone is not signalled, then radio is likely to be free? I am confused since free radio should be easier for switching channels.
For the tests you mentioned before, I found some channels can be switched to while some cannot. But I had not run thorough tests on all channels.
The bug that I mentioned, and has been fixed by Tinyos team, is in CC2420ControlP.nc line 273, SyncResource.release() is added.
Answering your last question, I am not using SPI for other things like large storage op; what I am doing is purely radio communications and computations.
Rgds,
Tie
On 10/25/06, Philip Levis <[EMAIL PROTECTED]> wrote:
Tie,
Did you run the tests I mentioned on the list? Or this part of that?
CC2420Config.sync is a split-phase operation; it might signal
completion a little while later, especially if there is contention
for the SPI bus. However, if sync were to never signal done, then it
is unlikely that the radio is receiving/transmitting packets. Which
not releasing resource bug are you referring to?
Are you using the SPI bus for anything else? E.g., large storage
operations?
Phil
On Oct 25, 2006, at 2:11 AM, Tie Luo wrote:
> if (call CC2420Config.sync()==SUCCESS) var=1;
> ....
> and in RadioConfig.syncDone(error_t err) I reset var=2 regardless
> of the value of err.
>
> However, sometimes the var is always 1, even if I check var
> periodically.
>
> This shows syncDone may not be called. Is that a bug? I recall
> previously a similar bug (not releasing resource) has been fixed.
> Is this a similar problem?
>
--
Regards,
Tie
_______________________________________________ Tinyos-help mailing list [email protected] https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
