Alan Cox wrote: > > > Some mechanism in the i2c core telling the attached i2c device (or > > module) something abut the i2c host/driver to setup the lowlevel stuff > > would help - but its this possible? > > If you know what i2c id you found and what to load for it - yes that could > be done via request_module > This is what I currently do in my 0.7b inofficial patch. I just "use" the i2c-bus in the zoran code to scan every address (uahh ugly..). Then I use 'request_module' before the 'i2c_register_bus' which finally attaches them. I have not looked at the new i2c implementation, but I hope a bus-scan is possible there via a regular interface without this hack. But then there is still the problem how the saaXXXX module knows if it is "Pauline's driver" requesting it or "Serguei's driver" - and set up the timing correctly. The post-setup method Serguei is proposing sounds good. Bye, Wolfgang
