> Is this a known issue and the reason why acme is using bit banging > from user space to interface the PCF8591? (timing problems?)
They clearly state the following: <snippet> This is the most quick-and-dirty method to connect an I2C device to the FOX Board. It is a crude programming solution, imperfect and inelegant but easy and useful to do experiments directly with the chips without going into too much knowledge details about device drivers. <end snippet> This means: it works, but it's not the best solution. If you use John's examples, those are using the I2C driver that has been delivered with the Phrozen SDK. However, this is also "bit banging". The only difference is that it in this case happens from kernel space and not from user space. The Etrax chip doesn't have HW I2C on board, so you have no other option than using "bit banging". It could be that they were not using the I2C driver, since the original one could only read and write one byte at a time, AFAIK. The "new" version is able to read and write whole messages and that's an incredible improvement! Best rgds, --Geert
