I couldn't rest with Disk Block 815 looking so recoverable, so I grabbed a few more captures until I had another three with a few more mismatches and set to work...
To start with, I was looking at a total of 28 mismatches between two or more of five captures, and by a process of elimination including looking for similar code in other parts of the same file, weighing the probabilities according to how many captures contained each of the candidate bytes and comparing the current and target checksums for each MOD(4) column, I eventually made just
17 1-bit changes to my original capture to match the target checksum and thus fix it.
Quote:
Originally Posted by dirkies
woohoo, this starts to read like an addictive detective story  |
Let's hope it continues that way. To put today's achievement into perspective, Disk Block 797 required
7 1-bit changes, and Disk Block 845 just
1!
However, the disk blocks I have fixed so far plus those fixed earlier by importing valid files to replace the damaged ones are near the edges of the area covered by the stain and least affected, but it's about to get a whole lot more difficult because the remaining five (Disk Blocks 642, 661, 691, 711 and 778) are in the central area and are worst affected. For example, I don't, as yet, even have a target checksum for Disk Block 711.
Quote:
Originally Posted by dirkies
|
As long as I can make measureable progress, I will keep trying, but I wish I had as much confidence as you!
I shall be taking a short break from this task from tomorrow, because a space has finally been cleared where I can set up a workstation to build a PC and leave it set up until I can dump an image of an old
Kyocera KC 30B hard disk which has travelled all the way from Salzburg for the occasion.
