[Tinyos-help] [TinyOs 2.x] ACK Bug?

[Philip Levis]

On Jan 2, 2009, at 11:19 AM, Michael Schippling wrote:

> Hmm...interesting. Why would #2 happen? Message overrun?

The current (2.1.0) code flushes only when a node receives a packet  
whose length is longer than the FIFO (128 bytes). Older stacks would  
flush on additional error conditions.

>
>
> If it was only #1 it would seem that you would have at worst
> a 1::2^16 chance of a falsely correct CRC and then an even
> smaller chance of it being the type field that was corrupted.

You'd think so, but... I'm not 100% sure. There could be other things  
going on in the radio stack. *shrug* If I had a free day or two I  
might try to get the bottom of it. But since assuming an L2 ack  
implies an L3 delivery is a bad idea to start with, I made TOSSIM keep  
this in account.

>
>
> That said, the chance of missing a valid ACK is much greater
> than getting a false one.

Yes. False positives are rare, but they happen.

Phil