ADC overflows on fast IOPs may cause timing errors.
Lee McCuller reports:
This is most likely a red herring, but I just want to mention our suspicion that ADC overflows on our 512kHz Cymac system cause timeouts or buffer underruns. This seems like an odd thing to accuse but maybe incrementing the counters at 512kHz thrashes atomic writes on shared memory. Figured I'd point it out if it might help devilish debugging at the sites.
Edited by Erik von Reis