New NTFS Linux Driver Spun A Ninth Time, Still Under Review

[jacob]

Windows XP SP1 source code leaked a few weeks ago. Would be pretty easy to look through the source code to get this driver working perfectly.

It would be easy but illegal.

[eydee]

Windows XP SP1 source code leaked a few weeks ago. Would be pretty easy to look through the source code to get this driver working perfectly.

Disregarding all the legal issues, significant changes happened to NTFS between XP and Vista. In fact, XP SP1 can't even read an NTFS volume made by Vista, XP need newer updates for that. Not that useful, isn't it?

[stiiixy]

Agreed, XP isn't exactly a place I'd want to look for idea gathering, and taking a leaf from it might do more harm than good in the long run. Using more modern idea's applied against reverse engineering a new driver compared to what was done twenty years ago should prove more fruitful.

Would most certainly be interesting to see how the old beast worked however.

[ultimA]

Does anybody have any suspicions as to why Paragon is looking to mainline their NTFS driver? Wouldn't this sabotage their commerical driver income? For something that would presumably be working against them for future revenue, they sure are keeping at this project!

Pure speculation: Perhaps Microsoft is secretly working towards rebasing Windows to run off of Linux? Maybe Microsoft is funding Paragon's efforts behind the scenes? Mainlining a full R/W commercial NTFS driver in Linux would help "Microsoft Linux" become closer to reality. This is the only explanation I can come up with as to why Paragon is suddenly so keen on doing this work...

First we got systemd, which makes services under Linux work a lot more like how services operate in Windows. Next we got exFAT file system support. Now we're getting NTFS file system support. Valve is working on WINE for playing old Windows games under Linux, despite very low Linux marketshare by their own statistics. Yesterday, we got Microsoft's open sourced Calculator ported to Linux...

It's like a giant jigsaw puzzle slowly being put together, piece by piece...

ntfs-3g already works pretty well for read-write, even if slower, so I guess not many users remained who needed the extra features of the Paragon driver. The Windows-Linux filesystem-interoperability gap also narrowed recently with Microsoft freeing exFat for Linux, so it might also have been a factor. I would guess (as other have also guessed here) that Paragon's ntfs driver has lost most of its commercial value and is not worth privately maintaining anymore. Now they had two choices: either let it die silently and go to waste, or open-source it. The latter scores plus-points with the community and also makes a lot of marketing for the company, so they obviously chose it. A bit like, "if this thing's commercial utility is about to burn out anyway, then at least let it burn out in a flash", or so they thought. Anyway, I think this is good, a win-win for all parties (for both Paragon and the community). Thumbs up for them for doing this.

[jacob]

Does anybody have any suspicions as to why Paragon is looking to mainline their NTFS driver? Wouldn't this sabotage their commerical driver income? For something that would presumably be working against them for future revenue, they sure are keeping at this project!

Pure speculation: Perhaps Microsoft is secretly working towards rebasing Windows to run off of Linux? Maybe Microsoft is funding Paragon's efforts behind the scenes? Mainlining a full R/W commercial NTFS driver in Linux would help "Microsoft Linux" become closer to reality. This is the only explanation I can come up with as to why Paragon is suddenly so keen on doing this work...

First we got systemd, which makes services under Linux work a lot more like how services operate in Windows. Next we got exFAT file system support. Now we're getting NTFS file system support. Valve is working on WINE for playing old Windows games under Linux, despite very low Linux marketshare by their own statistics. Yesterday, we got Microsoft's open sourced Calculator ported to Linux...

It's like a giant jigsaw puzzle slowly being put together, piece by piece...

I don't see any of that happening (it wouldn't make much sense as others pointed out, if MS wanted official NTFS support in Linux they could simply port over their own driver), but assuming for the sake of the argument that it was all true, I don't see any reason to be suspicious about it. All it would mean is that Windows would actually get a FOSS kernel instead of its current non-free one, and wouldn't that be good news? Linux itself can't be made proprietary, we have the GPL to thank for that.

[Old Grouch]

Ntfs-3g is already fully compatible. Your issue probably comes from you not disabling the so-called fast startup feature. It is a semi-hibernation feature that's being used instead of normal shutdown to fake a faster OS startup. It's enabled by default, and leaves all file systems in a "dirty" state. When dual-booting anything with widows, it's vital to have this turned off.

It is a consequence of Microsoft writing code that makes an explicit assumption that it is the only user of the storage device, which is false in the case of dual-boot devices, and when portable media formatted as NTFS is moved between systems.

As eydee says, this can be resolved by turning off 'fast startup' in Windows. The alternative is for the Linux driver to be modified to re-implement the 'fast start' feature, which I don't think anyone is in a hurry to do, as the workaround of disabling the feature in Windows works. It has the consequence of making it harder for non-technical people to use Linux, as from their point of view, it appears to induce problems on their storage devices that work perfectly well under Windows.

Cynical people might see this as a reimplementation of the Microsoft strategy against DR-DOS
( https://en.wikipedia.org/wiki/AARD_code )

[kpedersen]

- Seeing stuff like this isn't super uncommon, sometimes revenue falls and they may not have enough users to justify the (rather high) maintenance overhead of an out of tree driver at this point.

Usually commercial companies opt to just take the software (and source) to the grave with them. There must still be something a little more to it.
Perhaps they have existing contracts that they would need to maintain and instead they have managed to amend the support contracts to accept upstream maintained code? Who knows.

[skeevy420]

Linux itself can't be made proprietary, we have the GPL to thank for that.

No reason someone can't be evil and use the Linux kernel with proprietary kernel modules.

Also, see Android. That's literally proprietary Linux.

They took a standard Linux OS, hid that from the user, and gave us a Java front-end tied to Google servers and services with an app store that rips developers off just as bad as Steam. They didn't even need to be evil and hide behind proprietary kernel modules. They just hid the entire Linux kernel and userspace with VM-Droid.

Linux can totally be proprietary and the GPL doesn't prevent it at all. All the GPL does is ensure that companies might release some fustercluck of a kernel source like Motorola and Samsung do with their Android phones. GPL and Open Source doesn't prevent using locked bootloaders, the (IMHO) criminal act of tying unlocking the bootloader to agreeing to give up one's warranty, and using that locked bootloader to fire up a proprietary software stack designed to keep you from using Linux and the standard Linux userspace tools.

[Volta]

No reason someone can't be evil and use the Linux kernel with proprietary kernel modules.

Also, see Android. That's literally proprietary Linux.

They took a standard Linux OS, hid that from the user, and gave us a Java front-end tied to Google servers and services with an app store that rips developers off just as bad as Steam. They didn't even need to be evil and hide behind proprietary kernel modules. They just hid the entire Linux kernel and userspace with VM-Droid.

Linux can totally be proprietary and the GPL doesn't prevent it at all. All the GPL does is ensure that companies might release some fustercluck of a kernel source like Motorola and Samsung do with their Android phones. GPL and Open Source doesn't prevent using locked bootloaders, the (IMHO) criminal act of tying unlocking the bootloader to agreeing to give up one's warranty, and using that locked bootloader to fire up a proprietary software stack designed to keep you from using Linux and the standard Linux userspace tools.

Closing Linux in proprietary environment doesn't make Linux proprietary. Going to the bank doesn't make you a banker.

[skeevy420]

Usually commercial companies opt to just take the software (and source) to the grave with them. There must still be something a little more to it.
Perhaps they have existing contracts that they would need to maintain and instead they have managed to amend the support contracts to accept upstream maintained code? Who knows.

I'm thinking it's a maintenance burden over the long term for Paragon. If they can get the driver upstreamed then they can focus on all their proprietary software that works with those drivers. Imagine doing what ZoL does, only closed source with no help at all from the community. Sounds like a motherf*cker.

Plus they still have their closed source APFS drivers. I imagine that those have a lot of paying users since Apple users don't have a problem paying for stuff.