Currently if we have 2 extents which have the same hash but different
size it will trigger an abort in dedupe_extent_list. The reason is that
the query used to get dupe extents in dbfile_load_extent_hashes ignores
the len of the extent and instead only matches on digest. This would
result in creating 2 distinct struct dupe_extents with only a single
entry, this in turn results in triggering an assertion in
dedupe_extent_list.
Fix this by changing to the SQL statement responsible for loading dupes
to consider the 'len' column as well.
Currently if we have 2 extents which have the same hash but different size it will trigger an abort in dedupe_extent_list. The reason is that the query used to get dupe extents in dbfile_load_extent_hashes ignores the len of the extent and instead only matches on digest. This would result in creating 2 distinct struct dupe_extents with only a single entry, this in turn results in triggering an assertion in dedupe_extent_list.
Fix this by changing to the SQL statement responsible for loading dupes to consider the 'len' column as well.