a resulting diff between a "good" 040 kernel may show something, also
a diff at the same level against the same kernel from 1.16 may also
turn some thing up..
I don't think this is really necessary. Unless you don't want to compare kernels from MMU/stack frame point of view, there's a very few changes, even related to Aranym. You can use full sources with no problem. But if you really want to, you can always make simple shell/perl/python script which cut out everything non-related to 040/aranym/whatever (i.e. checking for that #ifdefs)