??? 03/11/10 14:00 Read: times |
#173991 - Unlikely to be "random" Responding to: ???'s previous message |
Baptiste Pomechicot said:
I'm sure to not feed the watchdog in an infinite loop. I activate it on the first instruction of my main() and I never disable it in my soft. You mean you never intentionally disable it in your software?! Clearly something is wrong - so how can you be sure that the "something" doesn't include disabling (possibly inadvertently) the watchdog? That's precisely why Kai said that a watchdog that can be enabled/disabled by software is such a Bad Idea! The other possibility, as I've said before, is that you scheme for "feeding" the watchdog is flawed and allows the software to keep "feeding" it while it is in this "fozen" state... It's realy random... No, I very much doubt that it is random at all; there is almost certainly a direct cause - the tricky bit is to find that cause... Quite possibly, something is "marginal" so, for one board it is just "close enough" to work - but the other board is just far enough out for you to see it failing occasionally... Or you have some other fault on the board (possibly an intermittent fault), and your software does not correctly handle the results of that fault - but ends up in this "frozen" state. Again, just because you haven't seen the problem in the other board(s) doesn't mean that it isn't there waiting to bite you when just the "wrong" set of conditions conspires to happen together... |