Send me all your floppies! They nourish me.

Recently I had another data recovery case that involved a comically large number of floppy disks, as in… more than five hundred (split evenly between 3.5” and 5.25” disks). We’re talking several large USPS boxes packed to the brim with floppies.

Of the numerous 3.5” floppies, only about 10% had one or more bad sectors, and none of them were completely unreadable.  The same was true for the 5.25” floppies, even though some of them were physically bent or warped, to the point where I had to cut them open and transplant the disk itself into a new container.  Some of the oldest files on these disks dated all the way back to 1986!

The recovery was performed using two older PCs, each of which have both 3.5” and 5.25” internal floppy drives, allowing the reading to be done somewhat in parallel.

There are actually plenty of cheapo floppy drives that connect over USB that can be purchased even now for as little as $15, but these drives are not, I repeat not suitable for recovering data from actual old floppy disks.  They must be read by a proper original floppy drive, preferably from the same era as the disks themselves.

Anyway, when floppy disks were in widespread use in the 1980s and 1990s, they weren’t really intended or marketed as a long-term storage solution, but they’re proving to be quite resilient as time goes by.  I’m not nearly as optimistic that today’s USB flash drives or SD cards will be readable in 30 years.

To be fair, these old disks have a much lower data density than modern storage media, so it makes sense that they would be more resilient to wear and tear. But still, it’s impressive that even what seems like mediocre-quality floppy disks still hold up to this day.

Despite these excellent outcomes, this still underscores how important it is to recover this data now, rather than waiting any longer and risking these disks developing any more bad sectors. So, let this be a call to action: if you have any old floppies lying around (or old tapes, Zip disks, Jaz disks, or anything else!), contact me for details on how to send them over, and I’ll recover the data from them for a fraction of the cost of other companies.

More data recovery war stories: recovering QIC-80 tape backups

I’ve got another recent data recovery job that is worth mentioning! (Quick reminder: I offer a service to read data from super-old media such as tapes and floppies)  This particular client had a number of QIC-80 tapes from the mid-1990s that contained backups of a desktop workstation.

To read QIC-80 tapes, I have a Conner 3200 SCSI drive which worked for me in the past, but when I tried to use this drive from Linux this time, I kept getting I/O errors regardless of what I tried to do with it.  I had a backup plan, though: In my collection I have an ancient Colorado 250MB drive which accepts QIC 80 cartridges, but here’s the problem: this drive connects to the floppy disk controller on the motherboard, which means that it must need very specific drivers or software to communicate with the drive.

In Linux, the way to communicate with such drives is using the ftape driver, which used to be included with the Linux kernel. However, ftape was removed from the kernel in 2006, citing too many bugs and too few users.

Since the tape’s label indicated that it came from a PC system, I decided to try keep it simple, and to recreate an actual DOS PC with this drive attached to the floppy controller. All I would need is to find the software that might communicate with this drive properly. After some serious dredging of old internet forums, I found a download site that contains the oldest versions of the Colorado backup software.

Setting up the boot disk

As I’ve done in the past, I used bochs to create a generously-sized disk image of 1 GB, and installed MS-DOS 6.22 onto it within the emulated environment of bochs. Then I copied over the installation files for the Colorado backup software. (I planned to do the actual installing of the Colorado software on the live PC instead of the emulated PC, since it might do some of the hardware detection during setup.)

Once the emulated disk image was complete, I wrote it onto a USB flash drive, plugged the drive into the old PC with the tape drive attached, and booted from the USB drive. (Thankfully the old PC has a very versatile motherboard that can boot from pretty much anything.)

The old PC booted successfully into MS-DOS, and I proceeded to set up the Colorado backup software. This did not present any issues, and the setup completed successfully without any non-default configuration.


When I ran the backup software, it went through a first-time setup process where the first sign of hope appeared: the software said that the tape drive was detected!

The next step was to rebuild the backup catalog from the tape, while praying that the catalog is compatible with this version of the backup software.  And what do you know — the catalog rebuilt successfully, and I could see the directory tree of the backup. The final step is to perform the actual restoring of the files, which I did directly onto the current boot disk.

There was not a single hiccup during the actual reading of the tape. I continue to be amazed by the resiliency of tape backup media, as well as the durability of the drive hardware, which still works flawlessly after 25 years.

This tape was a full backup of a PC workstation in 1996, so the final “bonus” step is to boot into the backup within an emulated environment, and see this workstation running in all its glory:

Chinese intellectual property theft hits home

Rampant piracy is pretty much a “feature” of publishing software on the internet, and any author of a semi-popular app will be well-acquainted with it. Not a day goes by when I don’t see pirated copies of my app shared via Google Drive, Dropbox, and all kinds of shadier file sharing sites.

But now I’ve discovered an even more disturbing dimension to this seediness. The problem with Android apps (and any Java-based software, really) is that it’s very easy to reverse-engineer, even when the code is obfuscated. Given enough time and manpower, it’s possible to recreate nearly the original source code from the compiled app. It’s analogous to reconstructing a shredded piece of paper — it simply takes some time to find all the strips and glue them together.

So I recently received a communication from a Chinese user who alerted me that this is precisely what’s happening in China. My app is being deconstructed and repackaged under a different name. But worst of all, they have inserted their own payment mechanism into the app, which requires the user to submit a payment before the app can be used!  According to my Chinese whistle blower, this counterfeit version of DiskDigger is being used by tech support departments that send the app to the user, make them install it, and then wait for payment before assisting the user further.


I generally turn a blind eye to most of the piracy of my apps, since it’s humanly impossible to continually track down violators, but this new kind of perversion makes me feel truly powerless. Without having any legal representation in China, there is literally nothing I can do to combat these bad actors. Perhaps it’s time to research how to establish a legal presence in China. If you have any experience with this, feel free to contact me. And in the meantime, when looking for DiskDigger, insist on the original!

A deep dive into optical disk file systems

It’s not often that I come across a data recovery story in my own personal life, but recently I came across just such a story, and a rather unusual one at that.

You see, my mother-in-law has several video recordings of my wife from her middle school and high school years, which I naturally couldn’t wait to watch, much to her embarrassment. These recordings are saved on a number of DVD-R discs. I’m guessing that my mother-in-law recorded the videos on a camcorder (onto compact tapes), and then hooked up the camcorder to a DVD recorder and burned DVDs from the contents of the tapes. (In the early-ish days of DVDs, there were standalone DVD recording devices into which you could plug in a video input, and it will continuously write the video to the DVD.)

But, to my disappointment, when I inserted these discs into the DVD drive in my computer, they appeared completely blank. One after another, the same thing: the disc contains no files, and the system reports it having a capacity of 0 MB (with no errors or warnings), even though it was visually apparent by the burn marks that the discs had data on them. I tried reading them on a different computer, with the same result.

Since the problem seemed to be affecting all of the discs, we can conjecture that it was the DVD recorder’s fault, where it might have somehow recorded the data incorrectly, or failed to close the recording session, etc. But can there be a way to access the data that was written to the disc?

The standard way to get the total size of a disk (using the Windows API) is to call the DeviceIoControl function and get a DISK_GEOMETRY_EX structure that contains the dimensions of the disk. Calling this function on these discs was returning a size of just 2048 bytes, or just one sector, since optical discs usually have a sector size of 2048 bytes.

But just because the OS is telling us that the disk is a certain size doesn’t mean we can’t attempt to explicitly read beyond that limit.  We can use the ReadFile function to brute-force the system into reading the disk at any location we specify. It may simply be that the driver is reporting an incorrect total size for the disk, while other areas of the disk might still be accessible.

So, I attempted to read the disc beyond the first sector. Reading the second, third, fourth, etc. sectors was returning errors, as might be expected, but I continued reading, and at around the 16000th sector, it started returning data! It’s almost as if the disc’s contents didn’t start until sector 16384. Onward from that point, the data could be read successfully all the way to the end of the disc.

Now, in order to actually recover the files present on the disc, we could potentially use DiskDigger to scan and carve any recoverable files from the raw data.  But I wanted to go a step further: up until this point, DiskDigger did not support any optical disc file systems, and since I haven’t dealt with a lot of CD/DVD recovery cases, I admittedly wasn’t totally familiar with the file systems used in optical discs, which presents a perfect opportunity to learn.

The most basic and original file system used in these discs is ISO 9660, otherwise known as ECMA-119. This is a very simple file system without any special affordances like journaling, access control, etc., which is perfectly adequate for read-only media where the data is written once, and will not need to be modified again.

Later, Microsoft developed the Joliet extensions to the ISO 9660 file system, which basically added support for Unicode file names, while remaining backwards-compatible by introducing a supplementary volume descriptor. This way, systems that support only the original ISO 9660 would continue to use the original volume descriptor, and systems that support Joliet will know to look for the new volume descriptor.  So basically, Joliet-formatted disks have two directory trees (one Unicode, the other non-Unicode), with the same file entries in each tree pointing to the same content on the disk.

And finally, by the time DVDs came around, the UDF file system (Universal Disk Format), also known as ECMA-167, was standardized.  It’s not backwards-compatible with ISO 9660, but discs that are formatted with UDF usually also have a stub ISO 9660 volume that tells the reader to look for a UDF volume on the same disk.  UDF is quite a bit more sophisticated, since it’s intended to be suitable for re-writable media, as well as multiple sessions on the same disk, but it’s still not nearly as complex as NTFS or ext4.

By the way, UDF can also be used on regular disks, not just optical disks. Here’s a little-known trick: it’s actually possible to format any disk as UDF by executing this command in an elevated command prompt: format <drive>: /fs:UDF

So, after poring over the ECMA specifications (real page-turners, I assure you), I implemented support for these file systems in DiskDigger, as well as in my FileSystemAnalyzer tool.

When you use DiskDigger to scan a CD or DVD (or an .ISO image), it will simply dump the contents of the disk and make all the files available for you to save.  The ISO 9660 and UDF file systems don’t really have a concept of “deleted” files, so DiskDigger will present all the files for recovery, even if they are still accessible by normal means. The benefit of this is that DiskDigger can now also scan the disk beyond the size reported by the OS (which was the issue I detailed above), and find these file systems in the space of the disk that is not accessible by normal means. You can do this by launching DiskDigger, going to the Advanced tab, and selecting the “Detect disk size manually” checkbox.

And in FileSystemAnalyzer, you can now examine these file systems in great detail. When you open an optical disk (or disk image), if it contains an ISO 9660 file system, it lets you examine and navigate it. If the disk contains a Joliet file system, it lets you examine it as either Joliet or ISO 9660, by letting you select which volume descriptor to use.  And if the disk contains a UDF file system, it lets you examine it or the stub ISO 9660 volume that usually comes along with it.

This rounds out support for optical disk file systems in DiskDigger and FileSystemAnalyzer! It’s admittedly a bit late, and also a bit overkill, since it’s true that data recovery cases involving optical disks are few and far between, but it’s good to know that even these sorts of cases can now be handled easily and smoothly.

FAT12 is alive and well!

One would have thought that the FAT12 file system was safely a relic of the 1980s and 1990s, when it was used as the default file system for floppy disks and very early hard disks. FAT12 would be entirely impractical today, since it can only cover a maximum of 32 MB of disk space. However, I was surprised to find it very much alive today, in the most unlikely of places.When I go running, I use my trusty Garmin Forerunner 10 watch, which uses GPS to record my position and pace during the run. When I connect the watch to the USB port on my PC, it appears as a mass storage device, and allows me to retrieve the workout files (stored in the FIT format). It hadn’t occurred to me until now to check out the finer details of this mass storage device, but there were a few things that surprised me:

  • The entire size of the watch’s flash memory is actually just one megabyte! I’m guessing this is because they want to discourage users from using the watch for general storage (i.e. dumping of photos, documents, and so on), thereby unnecessarily wearing out the flash memory. It also encourages the user to offload the workout files relatively often, in case the memory ever gets corrupted or the watch is lost. It’s also possible that this watch uses a more expensive type of flash memory (one that is more resilient to wear and tear), which would make it prohibitively expensive to provide multi-gigabyte sizes that we breezily expect in today’s USB flash drives.
  • You guessed it: it uses FAT12 to organize the files in the flash memory. Because why not! With a total disk space of 1 MB, this is really the simplest and most compatible solution they could have chosen.

Therefore, hats off to Garmin for not overcomplicating things, and making use of a tried and tested solution that is sure to remain compatible and future-proof.