Joerg Reisenweber wrote: > I don't think this direct CPU GPIO thing would be a better alternative > to an emergency driver directly accessing I2C to force some few bytes > into the LED controller.
A device can block the I2C bus indefinitely under some circumstances, e.g., if you're unlucky enough to crash while in the middle of an I2C transfer, SCL is low, and the device is sending a 0 or an ACK. Interestingly, not even a start condition is a reliable way to clear a stuck bus. (As I found out when experimenting with I2C from IDBG.) We used to have I2C bus issues fairly often in the past, perhaps also due to excessively large pull-ups causing badly deformed signals on some earlier boards, but it seems that this subsystem is reasonably under control now. - Werner _______________________________________________ hardware mailing list [email protected] http://lists.openmoko.org/mailman/listinfo/hardware
