Concerns Raised Over The 'New' NTFS Linux Driver That Merged Last Year (phoronix.com) 90
UnknowingFool writes: In 2020, Paragon Software announced they wanted to upstream their previously proprietary NTFS driver into Linux. After a year of review, the NTFS3 driver was added to the Linux 5.15 kernel. While Paragon pledged to maintain their driver, there have been no major updates to the driver despite a growing list of patches that have submitted. Developer Kari Argillander has raised his concerns on the mailing list that the driver is orphaned, and that the Paragon maintainer has not responded to any messages about fixes. An offer to co-maintain the driver has also been met with "radio silence".
NTFS? (Score:2)
Re:NTFS? (Score:4, Funny)
NTFS?
Gate in a dress,
Would no one impress:
How 'bout XFS?
Burma Shave
Lack of breaks make the troll look fake.
Re: (Score:1)
Re: (Score:2)
RedHat ring any bells to you?
Re: (Score:2)
Re: (Score:2)
Because no OS, proprietary or open source, has ever had to deal with orphaned drivers before...
Re: (Score:2)
Re: (Score:3)
Re:Haha the folly of open source. (Score:4, Informative)
Re: (Score:2)
That's only sort of true. You can generally direct your system to only use open source drives and software, but proprietary software is clearly still used. Nvidia driver comes to mind first.
Re: (Score:2)
Linux and BSD requires the driver to be open source
They do not either. If you want to plug in your own driver then you can do that by explicit public interpretation of the license, with plenty of precedent. As SP says, nvidia, and there's notably also been fglrx. You only need to "open source" (actually GPL) your driver if you want it mainlined. If you want to track the kernel and maintain it outside of the tree that's your business. That's been done even for open source code with incompatible licenses, like the only Linux ZFS implementation was for ages.
Bi
Re: Haha the folly of open source. (Score:2)
Re: (Score:2)
I don't know if this is supposed to be funny or not, because the world is incredibly stupid right now, but ...
There is a cautionary tail to be told about the decision to make Linux a monolithic kernel.
Re: (Score:3)
I don't know if this is supposed to be funny or not, because the world is incredibly stupid right now, but ...
There is a cautionary tail to be told about the decision to make Linux a monolithic kernel.
Everybody knows Linux should be used by keyboard, not by mouse.
Re: (Score:2)
Shit happens.
Re: (Score:2)
Is there?
Linux can have filesystems running as separate processes, and in fact the old NTFS driver was one such thing running in user land.
Re: (Score:2)
It's an odd point to defend, as even Linus agrees that, both technically and aesthetically, microkernel architecture is superior. I have to wonder if that's the way he'd have gone, had he known what his hobby project would turn into or even just that it would completely eclipse Hurd.
When Linux was new, it might have made sense as it was generally assumed that monolithic designs had better performance, and that was an important consideration with the hardware of the day. These days, security is a lot more
Re: (Score:2)
User space drivers are a thing now, which is great, but that sort of like claiming victory after you've surrendered.
I'm not a kernel developer, neither do I play one on TV. So, I don't really have a horse in this race, so victory or defeat isn't really applicable, to me anyway. I'm also not going to particularly defend any design choices in Linux.
I like userspace drivers, especially for anything not performance critical. libusb is awesome, as is FUSE and developing without crashing the machine is just a ve
Re: (Score:2)
Re: (Score:2)
They have file systems in user space, and I think there's always been a user space element to USB, but in general I don't think you can really say Linux has user space drivers as a generic thing.
USB hasn't always had a userspace component. Now libusb exists so you can write userspace drivers.
It's amazing how many USB Wi-Fi adapters that "have Linux drivers" are currently unusable because (1) there's no such thing (right now) as a userspace USB Wi-Fi driver and (2) the kernel's own APIs have changed enough t
Re: (Score:2)
I was curious so I searched and found this
https://github.com/emmericp/ix... [github.com]
https://www.net.in.tum.de/file... [in.tum.de]
Google apparently really wanted me to know about it, it put both of those links at the top of the results when I searched for userspace network driver. But those links refer to other projects as well, so they're useful whether it's interesting or not.
Re: (Score:2)
Re: (Score:2)
Ah yeah that's a point. I was focussed on the USB and packet part, not the additional control interfaces.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
obvious troll is boring troll (Score:2)
All of the Top500 supercomputers run Linux.
A there are billions [businessofapps.com] of smartphones with the Linux kernel although not the GNU operating system.
Let this be a cautionary tale on depending on Linus' monstrosity.
I agree with you there. I think every user should write their own operating system. Not only for the mental exercise but because nobody knows what is best for you better than yourself. Besides the security through obscurity would be amazingly effective with billions of mostly incompatible and undocumented operating systems.
Re: (Score:2)
Re: (Score:3)
GPL wasn't prepare to protect people from voluntarily putting all their personal information into a big marketing vacuum cleaner.
Re: (Score:2)
Only the NTFS code in the story was contributed by a corporation, and it's the weekend hobbyists who are identifying the problems with it...
So it's a problem, but is it a show stopper? (Score:4, Insightful)
If someone has volunteered to co-maintain the driver, perhaps it can be forked and someone can be found to maintain the fork (maybe the same guy.) It must be licensed in such a way that this is possible if it was mainlined, right?
Re:So it's a problem, but is it a show stopper? (Score:5, Informative)
If the driver was merged into mainline Linux, it's under the GPL v2. It doesn't really need to be forked; they just need to change maintainers to somebody who will actually do the job. The kernel already has the infrastructure to do this, but it's good to follow the process and give the existing maintainer a chance to respond.
Re: (Score:2)
Also, developers on free projects are always in short supply. Effectively pushing out the old maintainer should only be done if it is certain that he will no longer contribute.
Re: (Score:3)
Yeah, Kari is being vague about last contact and being dramatic with suggesting it should be dropped.
Not helpful.
1. when was last contact?
2. did you call him at the office? Maybe he got the koof - who knows. Maybe he got drafted. Talk to the secretary.
Don't catastrophize. This isn't twitter, people.
Re: (Score:3)
Re: (Score:3, Interesting)
If the Paragon driver were being maintained outside the kernel, it would need to be forked. But since it was merged, it's part of the kernel and can't be forked without forking the whole kernel. If the maintainer disappears, fails to respond to requests, or otherwise is seen as not doing an adequate job, the people higher in the hierarchy can name a replacement. That's seen as a drastic step, so the higher ups will usually try other things first, but it doesn't require a fork.
Re: (Score:2)
My understanding is that the maintainer(s) is the only one to make changes to this branch for consistency and governance.
But it's not on a separate branch, it's on the master branch.
If Paragon never updates the driver again, it most likely be forked with someone else as the maintainer.
Fork what? It's part of the Linux kernel master branch [github.com], there's nothing to fork.
I'm not quite sure what all the fuss is about here, the mantra of the open source community toward proprietary developers has always been "just open source your code and the community will maintain it".
Re: (Score:2)
Re: (Score:1)
At the end of the day, NTFS is mainly only used on desktop and workstation drives where linux isnt the OS, leaving only the rare cases to be covered, and I suspect most of the rare cases are already situations where read-only in satisfactory
In my case I have an NTFS drive on the network, and linux/icrap devices like Roku and smart phones only need to read from it.
Yes, the network drive could use a different filesystem, but h
Re: (Score:2)
If it's on a network drive, just use SMB to access it.
Re: (Score:2, Interesting)
Well, even here there are issues. SMBv1 is actively disabled since Windows 10 1709 becuse it has security issues you could drive a bus through, yet it's still the default protocol used by SAMBA in Linux.
SMBv2 has been a thing since Windows Vista and Windows Server 2008 and SMBv3 has been a thing since Windows 8 and Windows Server 2012. Linux really needs to stop dragging its heels here and make SMBv2 the default with an eye to making SMBv3 the default in the near term (2-3 years max).
Re: (Score:2)
They did.
SMBv2 has been the default since SAMBA 4.11 released in 2019, however, it takes a while for the linux distributions to pickup the newer SAMBA releases and those to get out there.
The problem is that a lot of people's build scripts and configs still use SMBv1, or they use older methods such as WINBIND instead of SSSD as they still work.
Things are changing now because security auditors are chasing people up, but it's slow going.
Re: (Score:2)
I would assume anyone that wants to share data back and forth between a Linux machine and a Windows machine might want NTFS read/write ability. You could format the drive differently but Windows likes to default to NTFS.
Re: (Score:2)
Network filesystems generally don't care what the underlying filesystem on the server is.
The main usecases for a NTFS driver on linux are.
1. Multiboot systems, if i'm dual-booting windows and Linux i want a way to transfer files between them, in both directions.
2. Removable media/external drives. I want my external drives to work on both my windows computers and my Linux computers. Fat32 used to be the obvious solution but as everything gets bigger the 4GB filesize limit it becomes increasingly problematic,
Re: (Score:2)
NTFS is not relevant on a network drive, unless you're exporting block level storage (Eg iSCSI). Otherwise, file sharing protocols like NFS or SMB abstract away the actual underlying filesystem.
The main use case for NTFS is probably portable drives, followed by repair of broken windows systems by booting a linux livecd.
Love Hate... (Score:5, Funny)
This is why I love, hate love, hate, love open source.
On one hand, It's ultimate freedom.
On the other hand it's ultimate freedom.
Sooner or later I find myself maintaining some elderly source code due to abandonment. That sucks!
Sooner or later I find myself maintaining some elderly source code due to abandonment. That totally rocks!
There are always moments. Then sometimes there's not.
Re: (Score:3)
Re: (Score:2)
There have been lots of times that commercial software wasn't well maintained, or was abandoned temporarily or permanently. It's not that FOSS isn't ever abandoned, it's that this isn't a differentiating factor. What is of course is that if you can get your hands on the sources, and you've got the will and the way, you at least can pick it up again.
Re: (Score:2)
Remove the orphaned driver (Score:2, Troll)
Re: Remove the orphaned driver (Score:2)
Re: (Score:3)
No. If a response can't be obtained, then a new maintainer should be chosen. If nobody is available, then mark it deprecated. Eventually it should be dropped, but unless there are dangerous bugs, that should be "eventually", not "now". Perhaps someone will show up to maintain it.
It is, however, reason not to trust promises by Paragon.
Maintainer is in Russia (Score:5, Interesting)
The NTFS3g maintainer lives in Russia. In fact most of the world's NTFS experts are Russian for that matter, since that's where it was developed.
One might argue that this shouldn't really be a big deal from an OSS standpoint, but given that he works for a non-Russian company, it's reasonable to assume that current events may be having some negative effects on his job and his ability to openly communicate.
In any case it's a hell of a detail to gloss over, even given the understanding that things weren't particularly awesome before.
Re: (Score:1)
The NTFS3g maintainer lives in Russia. In fact most of the world's NTFS experts are Russian for that matter, since that's where it was developed.
It's lobsters all the way down. [antipope.org]
He's about to purchase a ticket when a messenger window blinks open. "Manfred Macx?"
"Ack?"
"Am sorry about yesterday. Analysis dictat incomprehension mutualized."
"Are you the same KGB AI that phoned me yesterday?"
"Da. However, believe you misconceptionized me. External Intelligence Services of Russian Federation am now called FSB. Komitet Gosudarstvennoy Bezopasnosti name canceled in 1991."
"You're the –" Manfred spawns a quick search bot, gapes when he sees the answer – "Moscow Windows NT User Group? Okhni NT?"
"Da. Am needing help in defecting."
Manfred scratches his head. "Oh. That's different, then. I thought you were trying to 419 me. This will take some thinking. Why do you want to defect, and who to? Have you thought about where you're going? Is it ideological or strictly economic?"
"Neither – is biological. Am wanting to go away from humans, away from light cone of impending singularity. Take us to the ocean."
"Us?" Something is tickling Manfred's mind: This is where he went wrong yesterday, not researching the background of people he was dealing with. It was bad enough then, without the somatic awareness of Pamela's whiplash love burning at his nerve endings. Now he's not at all sure he knows what he's doing. "Are you a collective or something? A gestalt?"
"Am – were – Panulirus interruptus, with lexical engine and good mix of parallel hidden level neural simulation for logical inference of networked data sources. Is escape channel from processor cluster inside Bezier-Soros Pty. Am was awakened from noise of billion chewing stomachs: product of uploading research technology. Rapidity swallowed expert system, hacked Okhni NT webserver. Swim away! Swim away! Must escape. Will help, you?"
Manfred leans against a black-painted cast-iron bollard next to a cycle rack; he feels dizzy. He stares into the nearest antique shop window at a display of traditional hand-woven Afghan rugs: It's all MiGs and Kalashnikovs and wobbly helicopter gunships against a backdrop of camels.
"Let me get this straight. You're uploads – nervous system state vectors – from spiny lobsters? The Moravec operation; take a neuron, map its synapses, replace with microelectrodes that deliver identical outputs from a simulation of the nerve. Repeat for entire brain, until you've got a working map of it in your simulator. That right?"
"Da. Is-am assimilate expert system – use for self-awareness and contact with net at large – then hack into Moscow Windows NT User Group website. Am wanting to defect. Must repeat? Okay?"
Re: (Score:2)
Re: (Score:2)
While it's true current events might be having an effect now, it looks like there's been fixes queued that haven't been acted on since October 2021.
Re:Maintainer is in Russia (Score:4, Informative)
Russian? NTFS was developed by Tom Miller, Gary Kimura, Brian Andrew, and David Goebel at Microsoft in the US.
Re: Maintainer is in Russia (Score:3, Informative)
You are misinformed. NTFS wasn't developed by Russians. The original developers were American:
Tom Miller, Gary Kimura, Brian Andrew, and David Goebel.
Go read the Wikipedia entry if you really care.
Re: (Score:2)
You've never read Shakespeare until you've read it in the original Russian [wikipedia.org].
Re: (Score:1)
Perhaps Shakespeare is Russian inwhention?
Thank you Ensign Chekov
Go watch "The Trouble With Tribbles" episode to understand the reference
Re: (Score:2)
You are misinformed. NTFS wasn't developed by Russians. The original developers were American:
Tom Miller, Gary Kimura, Brian Andrew, and David Goebel.
Go read the Wikipedia entry if you really care.
The fine developers you reference are listed on the NTFS Wikipedia entry, having developed a filesystem driver that runs in Windows NT.
It's a different driver from the NTFS3g drivers which run under Linux, and give access to the filesystems written by NTFS under Windows.
Re: (Score:3)
NTFS was developed in america yes, but it was proprietary.
It was russians who reverse engineered it and developed their own implementations separate from the proprietary microsoft one.
Given the sanctions russia now faces, developing their own implementations makes even more sense otherwise they could potentially be cut off from accessing their own data being held to ransom inside proprietary filesystems.
Re: (Score:3)
No idea where you are getting that NTFS was reverse engineered by Russians. Tuxera, the company behind NTFS3g is Finnish, and from a quick Google search, the original developer is Szabolcs Szakacsits, who is Hungarian based upon the university he attended. There is an 'Advanced version' that is maintained by Jean-Pierre Andre.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
What makes you think Szabolcs Szakacsits lives in Russia? Why would a Hungarian want to live in Russia?
Re: So the real story is: (Score:3)
Request for MS & Apple to use Ext4 (Score:4, Insightful)
Re: (Score:2)
Words don't describe how much I hate proprietary file systems.
Ext4 is nice and all, but Apple and MS have needs that ext4 does not cover (see APFS, ReFS and NTFS 3.1).
Maybe when BTRFS Comes of age.
Or if you want it like RIGHT NOW tm , you can lobby for XFS or JFS
Re: (Score:2)
There are other open source filesystems... ZFS, XFS, JFS, BRTFS etc.
Something as fundamental and low level as a filesystem should always be open source in any case. The filesystem is used to hold *your* data, you don't want someone else to own that filesystem and have leverage over you or just abandon it entirely and leave you with drives full of data you cant access.
Re: (Score:2)
Why would you no longer be able to access data on a drive just because the company that built the filesystem stopped maintaining it?
Re: (Score:2)
Planned obsolescence, or the device you use to read the drive has failed and you can't get a replacement etc.
There is already software out there which requires continuous online activation or it shuts down, or requires online activation during initial install. As soon as the vendor shuts down those servers, the software is dead. If you've no other way to access the data, then your data has effectively been taken by ransomware.
CONCERN raised. Singular. (Score:4, Informative)
This is basically an email that one person posted on a kernel.org mailing list earlier today. As of yet, there have been no replies. This story seems premature, to say the least.
Re: CONCERN raised. Singular. (Score:3)
Re: (Score:2)
LKML being all like "Yes, Konstantin is not perfect, duh, no need to be dick about it. We've had far worse FS maintainers. There's plenty of time before it would need any sort of ReiserFS intervention.".
It's perhaps time to remind people why companies mainline their code: To cut costs.
FWIW, there's been a lot of drama wrt the initial merge over fstests too. It's a legitimate concern, but anything that talks ntfs in useable manner is sought after at this point.
I dare people to run ntfs as their / fs
Never buying Paragon stuff ever again (Score:5, Informative)
Fork It (Score:1)
Project turning to abandonware? Lots of push requests and no response?
Have linux.org organize an election from volunteers for a leader, leader forks the ntfs driver. Everyone incorporate the push requests. Switch the repositories over to the new work. Problem solved!!