linuxmemes

I used to be a reseller of this highly specialized navigation software. The licensing was handled via hardware fingerprinting of the harddrive. I'm not 100% sure on how it worked, but it grabbed some raw data off of the boot device, and from that generated a fingerprint. This fingerprint was then sent to the guys who made the software, and they would then send us a license key and a hefty bill. The license key only worked with syatems running off of that particular harddrive. If a customer had a harddrive failure, we had to send them the actual harddrive for them to verify, so they would issue a free key to whichever replacement drive was used.

I did a lot of experimenting with that software. It was linux based and very tweakable, but the licensing part of it was a bit of a mystery. I managed to crack it through some surprisingly simple out-of-the-box thinking, but one limitation I could never figure out how to circumvent was its refusal to generate a fingerprint from virtual drives.

For starters, it only worked with drives registering as /dev/hdX or /dev/sdX. Anything outside of that and it wouldn't generate a fingerprint.

This was especially frustrating when a well-paying customer offered a nice bonus if we could install it on a macbook for him. After a few days of tweaking I managed to install and run it, only to discover that fingerprinting the drive couldn't be done due to the device node being/dev/nvmeSomething. And after avfew more days of hacking I managed to fake that too, and they outright refused to issue a license due to them not wanting to support our unofficial hacks.

Where was I going with this? Oh, right, vmware.. i never managed to get it to run in vmware. We had this other well paying customer who wanted the ability to alt-tab between the software and Windows. Unfortunately, any fingerprinting done from within vmware, regardless how I set up the storage, resulted in a fingerprint file with no data.

Fun fact: the software ui was written in raw xlib. I got to know the owner and lead dev fairly well, and he hinted that the codebase was a complete mess to the point where something as simple as an input dialog for a config option I recommended was A LOT of work.