> This IBM is a good candidate to be a guinea pig. I understand the > consequences. (It's a lease from work, and I do the ordering...hehehe). > > Let me know how and I'll try it out and send you the results. Well, as you wish. I've recently modified sensors-detect and i2cdetect so as to prevent any kind of corruption of the AT24RF08. There was a different fix previously, but it was known to possibly not work under special circumstances. The new one is safer but could have additional side effects. So, if you want to help and have no fear, you could try both tools on your IBM system, with a reboot between each test. If the EEPROM is present and got corrupted, you'll now it because the system won't boot anymore. Tests are also welcome on non-IBM systems so as to verify that I did not introduce nasty side effects. Please get the latest CVS version of lm_sensors: http://secure.netroedge.com/~lm78/download.html#cvs Compile and install, then edit /usr/local/sbin/sensors-detect. Near line 4497, replace "exit;" with "return 1;" (at the end of function safe_system_vendor). Thid will bypass the IBM test. Then run sensors-detect. It should detect two busses and propose you to load i2c-i801 and i2c-i810. Accept (hopefully you have these modules availablable on your system). Then let the script scan them. Please provide a full output of the run ("script" may help). After that, reboot and check that everything still works. Then you can do the same for i2cdetect. There is no IBM test to bypass there. Just run "i2cdetect N" where N is the number of the i801 bus. You can do the same for the i810 bus if you wish. Again, full output of the run(s) is requested. Reboot again and ensure that nothing bad happened. If the new corruption prevention mechanism seems to work with no additional drawbacks, a similar change will be done in i2c-core (Linux 2.6 only) some times later and the lm_sensors on IBM nightmare will hopefully be gone forever. Thanks for your help. -- Jean Delvare http://khali.linux-fr.org/
