OSDev.org

I recently had to deal with a data recovery issue where I was able to use my os driver code for ntfs (yeah for hobby helping with job). As a part of this I realized that I had several ReFS volumes that could have problems in the future. So I've been working on researching for writing a driver out of curiosity. Unfortunately I haven't had any luck finding complete (and recent) information on it. Anyone else looked into this?

Why couldn't you just use a Linux live disk for recovering the NTFS filesystem? Also, considering that ReFS is a proprietary Microsoft technology for which there are better alternatives, I doubt that anyone will be writing a free or open-source driver for it any time soon; a free driver would probably be developed single-handedly, which is very difficult when trying to reverse-engineer propietary technologies, and an open-source driver would need to attract sufficient interest which isn't likely when ReFS isn't a mainstream desktop filesystem (yet) and in the server market there are superior filesystems that developers would rather focus their attention on.

Sorry for the late reply, but:

onlyonemac wrote:considering that ReFS is a proprietary Microsoft technology

More to the point, part of it is patented technology (though at least one of the patents is questionable at best in light of prior art), while other parts are trade secret technology, which means that even if you had enough information to roll your own driver for it, it would be illegal to do so from one angle (at least in the nations where such things are covered), and illegal just to know how from another (because it would require either industrial espionage and/or reverse engineering to get that information).

While this was also true with NTFS (though the patents relevant to NTFS 4.0 are now expired, I think - IIRC, they were developed at DEC by Dave Cutler's team, and were bought by MS as part of a technology and talent exchange on Cutler's request, as most of the 'new technology' operating system in 1993 was basically just a port the VMS kernel from BLISS to C - hence the development of ReFS in the first place), and plenty of FOSS drivers have been written for it without any legal actions against them AFAIK, the risk exists, and while the chance of you ending up on their radar is still minimal, it is significantly higher today than it was 15-20 years ago.

However, I don't think that the OP was talking about using ReFS for the data recovery; that was just what led the OP to the question of how the existing ReFS systems could be accessed from this hobby OS.

(OK, so I was being facetious about the NT 3.1 kernel being a flat-out port of the VMS kernel - I am pretty sure that the kernel, drivers, and file system for VAX VMS were all written in VAX MACRO-32 assembly language, with only non-critical utilities being in BLISS-32, and they were too closely tied to the DEC hardware to be easily ported until the OpenVMS re-write in the mid-1990s. Also, while Cutler did borrow heavily from his design for VMS, he made significant refinements and additions based on newer ideas of his and others, and there were some significant influences from UNIX, MACH, Amoeba, and other systems as well - while it was hardly a micro-kernel, it was something of a hybrid kernel, more out of the necessity to support diverse hardware than for any philosophical purpose - not to mention facilities to make is play nice with at least some MS-DOS and Win16 software. Still, the original design for NT was very close to the VMS of the late 1980s internally.)