The W5100 chip has a sort of alarming errata posted on their site regarding a lockup during closely timed UDP transmit/receives (link at the bottom).
Does antbody know if this is an ongoing issue with the W5100? (Or does the manufacturer fix issues like this when they discover them)?
If it hasn't been fixed, it seems like this would mean that this chip has a high likelyhood of failure in high-bandwitdh (or high crosstalk) UDP scenarios. Why would anybody tolerate this class of malfunction in a new, out of the box IC?
Also, does the UDP library account for/mitigate this issue?
Do you still have this problem? I'm currently working on code for W5100 and came accross this thread. I've introduced a workaround for the problem, but I'm still nowhere near a complete solution.
Reply if you're still interested in discussing this.
I actually never ended up using UDP with the Ethernet Shield, so I'm afraid I can't really comment either way on the problem.
I've introduced a workaround for the problem, but I'm still nowhere near a complete solution.
I assume that you've read over the recommendation in the 2.3 errata document? It reads to me like your only option with this chip is to camp out in your SEND primitive and wait for things to normalize before returning control to the caller.
Is your solution hosted on an Arduino, or are you just working with a W5100?