This uses a greedy algorithm and can, in theory, overpredict the starting address of the glabels per file. However, this should at the very least be a good starting point for separating out the data sections into their corresponding c files.
