I've added a 'c' mode to i2cdump in CVS. If 'i2cdump 0 0x50 c' returns different data than 'i2cdump 0 0x50' then the problem is in the nforce2 driver. Otherwise it's in eeprom.c. Mark Studebaker wrote: > > don't rule out a bus driver problem. They're both using nforce2. > Note that i2cdump uses read_byte_data > while eeprom.c writes the address then uses read_byte. > There's also a couple of failure paths in eeprom that don't have > printks. > I'll add them to CVS now. > > We have had previous reports (months ago) that the ddcmon driver > was reading off-by-one, we never figured that out. > This looks different though (off-by-slices). > > mds > > McClintock William J Contr MCOM wrote: > > > > > Then, unload the old eeprom module, load the new one, and do "cat > > > /proc/sys/dev/sensors/eeprom-i2c-0-50/*" again. Then take a look at > > > dmesg (or whichever log the debug stuff will have gone into), and tell > > > me what you see there, from the moment the module was loaded. Hopefully, > > > this will help me locate where the bug could hide. > > > > Hope this is it... >