Brent Brown wrote:
> Don't miss the other point though - a boolean flag (single bit, 0/1)
> could be used but if the flag itself was randomly corrupted (software
> bug, ESD, EMI, whatever) there would only be a 50/50 chance of
> detection.
>
> Instead here an 8 bit token is used. It is initialised only once,
> thereafter modified by each section of code in a prescribed fashion.
> Somewhat analogous to a game of
> Chinese whispers. Corruption of the token itself now has a 255/256
> chance of detection.

I'm no C whiz, but it looks to me like your method degenerates to the same
thing as using a single flag bit.

> During ISR:
> if watchcat = 0xAA
>   watchcat = !watchcat (invert the value)
>
> At end of mainline code:
> if watchcat = 0x55
>   watchcat = !watchcat
>   reset watchdog

I could be wrong about this, but I thought the C "!" operator used in this
context does a boolean inversion, not a integer one's complement.  If so,
you no longer know what the value is after the first watchcat = !watchcat,
only that it is now zero if it wasn't and non-zero if it was.  If you wanted
to check for random corruption you'd have to take the one's complement or
XOR with FFh.  (Of course that may be what the C "!" does.  Like I said, I'm
no C whiz and usually have to look these things up on the rare occasions I
write C).


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist