On Fri, 31 Jul 2009, Alan B. Pearce wrote:

> >If the consumer task is the only task that writes to the tail
> >index (your IndexOut) and the producer task is the only task
> >that writes to the head index (your IndexIn) *** AND *** you
> >don't modify other counts (like your BytesInBuf and BytesFree)
> >then a FIFO can be used efficently and safely to comunicate
> >between and interrupt handler and some other task without the
> >need to disable interrupts *** EVER *** while updating the FIFO
> >(provided of cource that the head and tail indecies use the
> >native word width of the CPU - this would be 8 bits for a 16F
> >and 16 bits for a 24F).
> 
> This, of course assumes that each routine is modifying only the head or tail 
> pointer as appropriate, and not updating a full/empty state variable as 
> well, just recalculating it every time. The routines I posted the link to 
> have a head pointer and count. 

Yes I agree but I find there is no need to maintain a full/empty state 
variable or count. Please see my other recent post showing the code I use 
to implement FIFOs.

Friendly Regards
Sergio Masci
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist