1. 29 Dec, 2014 1 commit
  2. 16 Nov, 2014 1 commit
  3. 25 Sep, 2014 1 commit
    • Mark Fasheh's avatar
      Large hash buckets (in the 10's of thousands range) are costly to walk · dad97112
      Mark Fasheh authored
      in find_all_dups. Instead, we catch those buckets during hash insert
      time and add filerec pointers (for each unique filerec) to the
      dupe_blocks_list structure. Later, when find_all_dups() encounters
      these it walks the filerecs directly instead of the blocks list.
      
      This patch halves the time (from 16s to 8s) to calculate extents
      from a sample hash file (created from my build partition).
      
      The value chosen at the moment (30000) is arbitrary. We might want to
      make it user selectable. There is also the question as to whether it
      makes sense to _always_ do this but the downside is we then lose the
      block_ever_seen() shortcut in find_all_dups.
      Signed-off-by: 's avatarMark Fasheh <mfasheh@suse.de>
      dad97112
  4. 04 Sep, 2014 2 commits
  5. 24 Apr, 2014 1 commit
  6. 11 Apr, 2014 1 commit
  7. 25 May, 2013 1 commit
  8. 24 Apr, 2013 1 commit
  9. 18 Apr, 2013 1 commit
  10. 16 Apr, 2013 1 commit