• Mark Fasheh's avatar
    debug: calculate stats on find_all_dupes() coverage · 6cd42053
    Mark Fasheh authored
    In order to further hack on find_all_dupes() and remove_overlapping_extents(),
    we need to be able to see how much 'coverage' of duplicated space a given set
    of changes gives us. The most natural way to do this is per-file. So for each
    file we compare the sum of our duplicated blocks (hash-tree blocks) with the
    sum of our dedupe extents.
    The data is printed in csv, there's also a block and extent dump on verbose
    for immediate comparison. This is all compile-time as it costs cpu and for
    now, is only interesting to anyone hacking on duperemove.
    Signed-off-by: 's avatarMark Fasheh <mfasheh@suse.de>