Please close this bug as it was caused due to a bad RAM chip and is invalid. As it seems the kernel modules were always loaded at the same (bad) physical addresses and the same errors were repeated each time. Sorry for the confusion.