That was the only way I could figure out for doing it. You can have a separate and parallel entity which tracks what is read or written by the host and figures out which file a read block is a part of, even how far through, and it can regenerate its database after a write has occurred.An alternative approach, again assuming read only access and that you can get the block ID/address, might be to generate a look up table from the meta data and use that rather than working backwards through the meta data on each access. Of course, you'd have to regenerate that if the drive's content changes.
But that was based on an RP2 MSD drive where one could easily do that. I believe it would be much trickier doing that under Linux.
Statistics: Posted by hippy — Sun Nov 23, 2025 7:28 pm