Microsoft Readies exFAT Patents For Linux and Open Source (zdnet.com) 119
An anonymous reader writes: For years, Microsoft used its patents as a way to profit from open-source products. The poster-child for Microsoft's intellectual property aggression were the File Allocation Table (FAT) patents. But the Microsoft of then is not the Microsoft of now. First, Microsoft open-sourced its entire patent portfolio and now Microsoft is explicitly making its last remaining FAT intellectual property, the exFAT patents, available to Linux and open source via the Open Invention Network (OIN). Microsoft announced that it now loves Linux and "we say that a lot, and we mean it! Today we're pleased to announce that Microsoft is supporting the addition of Microsoft's exFAT (Extended File Allocation Table) technology to the Linux kernel." ExFAT is based on FAT, one of the first floppy disk file systems. Over time, FAT became Microsoft's files ystem of choice for MS-DOS and Windows. It would become the default file system for many applications. Microsoft extended FAT to flash memory storage devices such as USB drives and SD cards in 2006 with exFAT. While FAT isn't commonly used today, exFAT is used in hundreds of millions of storage device. Indeed, exFAT is the official file system for SD Card Association's standard large capacity SD cards.
Now, Microsoft states: "It's important to us that the Linux community can make use of exFAT included in the Linux kernel with confidence. To this end, we will be making Microsoft's technical specification for exFAT publicly available to facilitate the development of conformant, interoperable implementations. We also support the eventual inclusion of a Linux kernel with exFAT support in a future revision of the Open Invention Network's Linux System Definition, where, once accepted, the code will benefit from the defensive patent commitments of OIN's 3040+ members and licensees." Specifically, according to a Microsoft representative, "Microsoft is supporting the addition of the exFAT file system to the Linux kernel and the eventual inclusion of a Linux kernel with exFAT support in a future revision of the Open Invention Network's Linux System Definition."
Now, Microsoft states: "It's important to us that the Linux community can make use of exFAT included in the Linux kernel with confidence. To this end, we will be making Microsoft's technical specification for exFAT publicly available to facilitate the development of conformant, interoperable implementations. We also support the eventual inclusion of a Linux kernel with exFAT support in a future revision of the Open Invention Network's Linux System Definition, where, once accepted, the code will benefit from the defensive patent commitments of OIN's 3040+ members and licensees." Specifically, according to a Microsoft representative, "Microsoft is supporting the addition of the exFAT file system to the Linux kernel and the eventual inclusion of a Linux kernel with exFAT support in a future revision of the Open Invention Network's Linux System Definition."
It is simple (Score:4, Informative)
Running on Azure.
Re:It is simple (Score:4, Interesting)
Which happened to be very good for everyone involved.
Until the CEO change, Microsoft was Windows first, and Windows only. They had reluctant support on Mac OS, and barely existing one on Linux (via Skype and a few other things).
Now it is software first, and the platform does not matter. They still want to sell Windows licenses, and Azure subscriptions, however if that does not hinder Linux, or even supplements it that is very welcome.
(That does not mean that it would stay the same in the next decade or another).
Re: (Score:1)
Correction: Now it is *spyware* first. For all of Ballmer's faults, he never spied on anyone or put ads in the OS or pulled shady shit like the forced Windows 10 "upgrade".
It's only a matter of time before Windows 10 users are given the choice of paying a subscription or receiving increased amounts of ads baked into the OS. Users paying the subscription will still be subjected to some amount of ads and still won't be able to disable the spyware or control updates.
Prediction: Windows will be migrated to Linux (Score:4, Interesting)
Here is my Prediction:
Next year:
1) Windows and Windows Server (as brands) will still exist for years to come but will be fully migrated to Linux
2) Win32 will be open sourced and migrated to full compatibility with Linux, Virtually all legacy windows application will run natively in Linux
3) The full Office 365 suite will be available as a Linux native application
Re: (Score:2)
The full Office 365 suite will be available as a Linux native application
But Ubuntu will offer it only as an overbloated "snap".
Re: Prediction: Windows will be migrated to Linux (Score:4, Insightful)
No, MS will release their own special package system and expect all other distros to implement it.
Ironically, this will serve to make MS's application package system the default for all Linux applications.
Re: (Score:2)
Re: (Score:1)
It won't be ubuntu. That's too broken. It'll be Fedora and RedHat. What they're already using.
Re:Prediction: Windows will be migrated to Linux (Score:5, Funny)
Re: Prediction: Windows will be migrated to Linux (Score:2)
Not confirmed, but I'll bet on it.
Re: (Score:2)
1) Windows and Windows Server (as brands) will still exist for years to come but will be fully migrated to Linux
That would be an interesting alliance again . . . Microsoft and IBM, as IBM now owns RedHat!
Now make a reality TV show out of their "partnership"!
Re: (Score:1)
What does RedHat have to do with Linux? You seem to be confused. RedHat distributions "contain" Linux, but Linux distributions do not "contain" RedHat.
Sort of like "Hot Dogs" contain pork, but pork (as in a nice shoulder roast) does not contain "Hot Dogs".
Re: (Score:2)
Not to be pedantic, but I respectfully disagree. RedHat is a major Kernel and GNU utilities contributor. Linux distributions very much do contain RedHat. That's a good thing.
Re: (Score:1)
Re: (Score:2)
Why 2?
Why would open source windows have any need to migrate over?
Re: Prediction: Windows will be migrated to Linux (Score:2)
2 is the most important part. Many many many companies and governments that MS needs to stay in good graces with run legacy mission critical software. This software is an unsustainable pain for MS, because if Windows stops supporting the legacy software it looks bad on MS. However, the legacy software is becoming exponentially missed more expensive for MS to support. That was always the unkeepable promise MS Windows made, your software today will run on Windows forever.
By open sourcing Win32, MS gets to wal
Re: (Score:1)
Win32 is not the problem. The Win32 API has not changed in decades. I have software that was written before the turn of the century that still runs perfectly fine on current versions of Windows. The problem is all the other fly-by-night shit that Microsoft promulgates so they can have planned obsolescence, and the reams of fanbois that insist on using the latest crappola from Microsoft because they cherish instability and the planned obsolescence means they get to sell the same old shit over and over and
Re: Prediction: Windows will be migrated to Linux (Score:2)
" I have software that was written before the turn of the century that still runs perfectly fine on current versions of Windows."
That's literally exactly what I'm talking about. Keeping Windows secure while still fully supporting decades of legacy software (much of it written at a time when security wasn't a thing people did) is exponentially more are more expensive for MS.
So, MS needs to walk away but can't without upsetting many big important clients. Open sourcing that stuff and kicking the legacy suppo
Re: (Score:3)
And we still won't use it due to its security holes and guis that look and work like a little demented leprechaun wanker on the Lucky Charms commercials.
Re: (Score:2)
Re: (Score:2)
Next year:
Microsoft has been on the Linux bandwagon for the best part of this decade and only just got around to opening up patents to exFAT. If your prediction holds any weight at all it will be for something due to happen in 10+ years.
Re: (Score:2)
10+ years is easy. My prediction is next year.
MS already said Windows 10 is the last Windows version. That's a odd thing to say if you think about it. Sure, it's received major service pack updates, but based on MS's historic time line, it should be due for a replacement soon. Enter Linux Windows.
Re: (Score:2)
Mac OS X has been the latest Mac OS version for nearly twenty years. Microsoft has adopted the same development pattern. Windows 10 doesn't get service packs, it gets annual or seasonal updates. Changes that used to wait for new OS releases, like memory compression or replacing the windows terminal, are now done in those updates.
Re: (Score:2)
Seems more like they are going the other way, i.e. making Linux stuff work on the Windows kernel, e.g. WSL.
Re: (Score:2)
My friend, you define optimism.
Re: Prediction: Windows will be migrated to Linux (Score:2)
I don't think it's simple optimism.
The problem with Windows for MS is that MS made an unsustainable Windows promise that they couldn't keep. Every single version of Windows has to perfectly run every single third party legacy software written for Windows. The cost of that support grows exponentially with every Windows version. It's unsustainable. MS has to cut off its legacy promise, but doing so will look very bad on MS and will likely make them lose major clients.
So, MS is slowly open sourcing all of the
Re: (Score:2)
This all seems to be predicated on your assumption that Microsoft wants to cut off their PC/Windows business, which contrary to popular belief here on Slashdot, is still making them quite a bit of money. Their Personal Computing division, which includes Windows sales, grew 4% last quarter and earned over $11 billion, which is about a third of their earnings.
Microsoft is still investing in Windows, notably making it a more jack-of-all-trades desktop by incorporating Linux compatibility, which I suspect is l
Re: (Score:3)
Windows Server (as brands) will still exist for years to come but will be fully migrated to Linux
I'm highly doubtful that Windows, at least in the server sense, would ever switch to a kernel it doesn't control. Google and it's history with Android is a great indicator of why that is. Hell IBM is a husk of what it used to be and they still develop actively their AS400 OS to this day. Now when it comes to desktop, maybe, maybe not.
Win32 will be open sourced and migrated
Well I'm not so sure about that. Some of the older API isn't Microsoft's. DDE, OLE, Some of the NT4 API, DCOM's de/serializing APIs, and so on. If we're sticking to moder
Re: (Score:2)
Timeline is negotiable, but it looks very much like MS is planning make Windows a Window Manager (+associated tools) on top of the Linux kernel. It would make a lot of sense technologically and from business-side.
Re: (Score:2)
I'm calling it.
2020 is the year to do it. MS will release GNU/LINUX Windows.
Re: (Score:2)
Well, it would be really cool if you turn out to be correct!
Re: Prediction: Windows will be migrated to Linux (Score:2)
Re: (Score:1)
Re: (Score:2)
Re: (Score:1)
Here is my Prediction:
Next year:
1) Windows and Windows Server (as brands) will still exist for years to come but will be fully migrated to Linux
2) Win32 will be open sourced and migrated to full compatibility with Linux, Virtually all legacy windows application will run natively in Linux
3) The full Office 365 suite will be available as a Linux native application
I doubt the Office 365 suite, they want to sell that online subscription !! They will migrate more of the office features to the O365 of the office suite.
Link missing? (Score:2)
Did something go wrong and there is no source link?
Re:Link missing? (Score:4, Informative)
Re: (Score:2)
On my portrait monitor, the link is underneath of Slashdot's tape reel icon. I will forgive the layout snafu, but I cannot understand why an article about Microsoft file system patents gets that icon.
Re: (Score:2)
Slashdot is run by a bunch of dumbasses, so they put links in the header title instead of the summary.
Re:Link missing? (Score:5, Informative)
Re:Link missing? (Score:5, Informative)
Source, just in case:
https://cloudblogs.microsoft.c... [microsoft.com]
Microsoft now loves Linux (Score:4, Insightful)
Microsoft announced that it now loves Linux
And Father Christmas will give me a pony. Because he said so.
How about NTFS (Score:5, Insightful)
Re: (Score:2)
Re: (Score:2)
I don't think Microsoft really knows how it works either. They just winged it and got lucky ;)
Re: (Score:2)
Re: (Score:3)
As for why Microsoft hasn't implemented it, it's probably because they haven't really had a need to until now. I suspect you might see something like ext support coming in the next year or two just because they're starting to push Linux integration.
Re: (Score:2)
/ as exFAT!! (Score:3)
Re:/ as exFAT!! (Score:4)
Why yes-- performance should increase if you are using flash media for your storage.
Just, don't expect any of your linux-specific permissions sets to have any meaning, and hope you never have a power failure, because exFAT only keeps a single copy of the FAT, instead of keeping a failsafe copy, like older versions of FAT.
(Better idea is to properly format your flash storage device with EXT4, using stride and stripe-width parameters, so that whole erase blocks are used on each committed write operation.)
Re: (Score:2)
I don't know about that - within the past couple of years I've lost EXT4 file systems on two separate occasions due to an unexpected loss of power. I've never had that issue with EXT3 or XFS and I've since banned EXT4 from all of the systems I administer at home and at work.
Re: (Score:2)
> within the past couple of years I've lost EXT4 file systems on two separate occasions due to an unexpected loss of power
--I suspect you might be doing something wrong, or there is some issue with the firmware on your drive... I've had unrecoverable btrfs (stopped using it entirely in favor of ZFS) and data loss on exfat, but no problems recovering ext4 with noatime.
--Of course, backups are important tho...
Re: (Score:2)
And UPS likely helps too.
Let me guess... (Score:2)
Re: (Score:3)
exFAT's only been around since 2006.
Re: (Score:2)
The patents were bogus anyway, and no one would use this system anyway except that it's become the defacto standard for flash based filesystems that need to be compatible across systems - thus choose the worst system because it's popular rather than the best system. A file system is mostly a set of structures and protocols, and really nothing should even be possible to be patented in this. For example, a hashing scheme for quick filename lookup is a mathematical operation, and mathematical operations cann
Good. (Score:2)
FUSE is great for (relatively) low speed media like network shares and SD cards. It really starts showing its limitations if you start pushing large files around on external SSDs that are meant to be shared between systems. FAT32 hasn't really been viable for this in years.
Hilarious Thought (Score:1)
What if Microsoft transitioned to the Linux Kernel since Apple went BSD any the *nix Wars began anew?
Re: (Score:3, Interesting)
Unix derived systems like Linux use a lightweight process creation model (the fork() system call is blazing fast) but the (post-NT) windows kernel is derived from DEC's VMS, and as such has such heavyweight process creation that threaded code is actually a good idea.
So Microsoft will simply brand and develop their own Linux distribution (they are already well down that path) and sell it in parallel with Windows. It's extremely unlikely that they'll try to replace the NT kernel, it's not worth the level of
Re: (Score:2)
Unix derived systems like Linux use a lightweight process creation model (the fork() system call is blazing fast) but the (post-NT) windows kernel is derived from DEC's VMS, and as such has such heavyweight process creation that threaded code is actually a good idea.
The only difference between a process and a thread in the NT kernel is the protection boundaries. Threads are just faster in general because you can share objects between threads without serializing them. This is why threads are easy to work with in any modern programming language, and processes ... aren't.
So Microsoft will simply brand and develop their own Linux distribution (they are already well down that path) and sell it in parallel with Windows. It's extremely unlikely that they'll try to replace the NT kernel, it's not worth the level of effort required when they can just sell linux at the same time.
Replacing the NT kernel with the Linux kernel would be trivial for MS to do. It's more or less what WINE did, but much easier to do from the inside, as it were. Windows userland software talks to the
Re: (Score:2)
That isn't true... at all. Hell, there isn't even just one type of thread. I mean, why would you just make shit up like that?
Replacing the NT kernel with the Linux kernel would be trivial for MS to do.
WTF? No it wouldn't be trivial. Every single thing about the two are different, from the driver interface, to even the protection ring levels, to processes and threads, to every other bloody thing.
FFS. Get o
Re: (Score:2)
That isn't true... at all. Hell, there isn't even just one type of thread. I mean, why would you just make shit up like that?
I've worked with several people who have worked on the NT kernel. Process boundaries aren't part of the thread dispatch and anti-starvation logic in NT, it's just a bucket of threads.
Launching a new process obviously requires the overhead of setting up protection boundaries, which could be expensive depending on what you're ding.
WTF? No it wouldn't be trivial. Every single thing about the two are different, from the driver interface, to even the protection ring levels, to processes and threads, to every other bloody thing.
You do understand we're talking about implementing win32 over the Linux kernel APIs, right? What WINE already did?
You don't want compatibility for stuff that uses the kernel APIs,
Re: (Score:2)
> Replacing the NT kernel with the Linux kernel would be trivial for MS to do
That's literally the stupidest thing I've heard all week. You obviously know NOTHING about backwards compatibility requirements on the Windows platform, much less Licensing.
Re: (Score:2)
Not "Windows platform", but "Windows userland".
MS would have a big win if the vast bulk of legacy Windows userland software would work with the Linux kernel. That just requires getting win32 to work with the Linux kernel (which WINE has mostly done) and getting the most commonly used "cheats" to work with the Linux kernel (which WINE has mostly done).
MS could make WINE bulletproof, as they of course have the relevant source code to get to the bottom of the crufty corner cases.
Re: (Score:2)
None of that is really accurate. It's almost jaw-droppingly wrong. [wikipedia.org]
I have worked with kernel interactive code on linux, DEC RSX-11 and RT-11, VAX-VMS and post-NT Windows. That experience is what informed
Finally! (Score:3, Insightful)
Thank you, Microsoft. Better late than never.
exFAT is really terrible (Score:5, Informative)
It is even more terrible than the older FAT32 it has supplanted, because it only has one copy of the table, instead of two.
Additionally, it was adopted by the SDCard association because of two basic features:
1) I allows OBSCENELY sized clusters, which just so happens to correspond with the native erase block sizes of larger SDcard media, because of how flash storage works under the hood, allowing exFAT to not have write amplification.
2) Its simplicity means it can blast that data into the card very quickly, if unreliably. (again, a single FAT.)
This is because Microsoft and Pals just cant bother themselves to make a proper journaling file system that works natively with obscenely huge physical sectors, and instead just threw some old shit at the wall, and it stuck.
Linux CAN use SDCards "correctly" (ahem) by abusing the RAID functionality baked into EXT3/4. Just nobody seems that keen on doing this except the really geeky, because the media manufacturers seem completely unwilling to disclose what the physical allocation sizes are of their consumer grade media devices. (Forcing people to have to resort to esoteric interrogation of the media to suss out what that geometry actually is.) This means there is no quick and easy way to automatically determine what stripe and stride-width values to use to properly use the card, and so no linux tools helpfully jump in to go "Hey, dumbass-- that's an SDCard, let me format it PROPERLY for you."
Accepting exFAT is accepting the lowest common denominator here.
Use larger powers of two, get all smaller as side- (Score:3)
> media manufacturers seem completely unwilling to disclose what the physical allocation sizes are of their consumer grade media devices. (Forcing people to have to resort to esoteric interrogation of the media to suss out what that geometry actually is.) This means there is no quick and easy way to automatically determine what stripe and stride-width values to use to properly use the card
It occurs to me that if you align on 8MB, you will also be aligned on 4MB, 2MB, 1MB, 512K, 128K, etc. Just use a powe
Re: (Score:2)
The alignment is the easy part. The thing is, you need to know the actual block size. If you choose wrong, you will wear the flash out much faster and get much worse performance as well as wasting a lot of capacity in the form of unneeded slack space.
Re:exFAT is really terrible (Score:5, Informative)
Why: http://lwn.net/Articles/428584/ [lwn.net]
How: https://thelastmaimou.wordpress.com/2013/05/04/magic-soup-ext4-with-ssd-stripes-and-strides/ [wordpress.com]
Yet more reading: https://wiki.gentoo.org/wiki/SDCard [gentoo.org]
Re:exFAT is really terrible (Score:4, Informative)
As for the "Esoteric interrogation" part, there's a program for linux called flashbench.
https://github.com/bradfa/flas... [github.com]
It reads/writes a block of data with some arbitrary parameters the user provides, to evaluate the time needed for the operation to complete. Since SDcards (and flash memory in general) contain a small micro controller and a small amount of RAM to buffer reads/writes, the controller can accept and queue up such operations, but then issues a wait/busy response while it handles those requests. This introduces latency between the command being issued, and getting the data back (on read) and "OK/ready" response from the controller.
By analyzing those latency times, you can kinda-sorta suss out what the buffer size and erase block size are.
There's a complication in recent SDCards though, because the manufacturers have "different" alignments at the first 4mb or so of the device, because that's where exFAT's FAT gets stored, and FAT does a lot of operations on the table. (Which would wear the media out prematurely otherwise). So, you need to run several runs of flashbench to get a statistical mean to get a better grasp of what the alignment should be for EXT3/4,
It would all be solved cleanly and neatly if the SDCard assc. decided that they needed the likes of PNY and SanDisk to disclose what the buffer size and the erase block sizes are. But they haven't, because .. reasons..
Obvious reasoning (Score:1)
Re: (Score:3)
the vast majority of Linux kernel work is done by large companies like google, facebook, etc
Re: (Score:2)
... And Microsoft.
Re: (Score:2)
Last I heard, there was only one year when that was true.
Does it mean android will finally support exFAT ? (Score:2)
Re: (Score:1)
Windows will transition to Linux kernel, Wayland (Score:2)
This overall is a good thing. it shows Microsofts commitment to Linux.
Microsoft is planning to transition Windows over to Wayland and a Linux kernel, and systemd. It will happen. This will include an system call emulation facility for running Windows apps on the Linux kernel (WINE is being considered) , and a compatibility facility for running Windows drivers on the Linux kernel. This is good for everyone since Microsoft can reduce development costs and as well the community gets more collaborative effort.
Re: (Score:2)
Re: (Score:2)
That's one huge speculation.
It's also not good for Microsoft. Yes, lower development costs - but also a lower cost for users who might seek to transition away from Windows entirely. Plenty of businesses would really love to save on those license fees, but right now the fees are cheaper than the cost of porting applications and training staff. The more linux-like Windows is, the better the prospect looks.
Any Benefit for Other FLOSS OSes (Score:2)
This is great for Linux but does it offer any protection or benefit to the FLOSS non-Linux operating systems out there? Random examples include FreeBSD, NetBSD, Haiku, ReactOS, RTEMS, HelenOS, etc.
bingo (Score:2)
This is yet another way for them to control the open source world, and choose (pro-microsoft hegemony) winners.
Comment removed (Score:4, Interesting)
Re: (Score:2)
Microsoft bet on software, IBM bet on hardware - and now Microsoft is looking enviously at Apple, who established the 'ecosystem' model in the consumer space, where every product promotes every other. It's a nice apple pie, and Microsoft wants a slice. You can't use Windows 10 stock without it making frequent, gentle but insistent nudges in the direction of the Microsoft store and OneDrive. You can't even install Windows 10 without it asking you to create a Microsoft account - though you can skip that if yo
M$ doesn't do Open Source (Score:2)
I will believe their commitment to Open Source when they open source all of their products.
That will never happen.
Would Microsoft open source include Xbox? (Score:2)
I will believe their commitment to Open Source when they open source all of their products.
Could this be part of why Microsoft recently rebranded Microsoft Studios to Xbox Game Studios? If Microsoft became a free software company, a rebrand would at least make it easier to spin out the Xbox business without having to come up with a business model for developing original, high-production-value video games for release as free software from day one.
Hell Hath Frozen Over (Score:2)
Well done MS (Score:2)
Microsoft loves Linux? (Score:1)
The fuck it isn't, this is the same company that is currently extorting revenue from Android phone makers. It's understandable why your reader would want to remain anonymous. Here's a Microsoft front trying to patent SMS messages [unifiedpatents.com].
Dominion Harb [intellectualventures.com]
SD Card Association (Score:2)
Now lets see the SD Card association open source their IP so they not bilking small businesses $3k/yr to use an SD card in their product.
Funny /. fortune (Score:1)
As of this moment, my fortune at the bottom of /. reads We are Microsoft. Unix is irrelevant. Openness is futile. Prepare to be assimilated.
Has anyone actually seen an exFat SDCard? (Score:2)
Genuine question. I remember Microsoft releasing exFat in 2006. After looking at the description and reading that Microsoft planned to charge people who used it, I was left scratching my head. Why would someone use it in preference to fat which was free, or one of the numerous open source file systems out there which were not only free, but fast and robust as well?
It's now 13 years down the track and I've never seen an exFat formatted anything - which meant everything has panned out pretty much as expect
Re: (Score:1)
It's a trap! (Score:2)
It's a trap!
Re: (Score:3)
I read it as exFAT Patients and wondered if MS was getting into medical care
Re: (Score:1)
Re: (Score:2)
Re: (Score:3)
Microsoft would rather you use their subscription services on Linux than somebody else's subscription services on Windows. They don't care about being the only OS in the world.
A Windows license is a hundred bucks every 5+ years, and probably free for future upgrades. An Azure instance running Linux could easily be a hundred bucks a month, every month. Yes, Linux on Azure isn't replacing desktop, but you probably use some services or sites that are running Linux on Azure, or you might have a paid subscriptio
Re: (Score:2)
A Windows license is a hundred bucks every 5+ years, and probably free for future upgrades.
For the desktop OS, sure if you buy OEM and not enterprise. Now Windows Server, that's a very different story. Go play here: https://wintelguy.com/windows2... [wintelguy.com]
So yea, they don't care because they know they will get the desktop . And they have even a subscription for that (for Enterprise customers), Microsoft 365, for that, or you can buy E3/E5 one time, if you don't get SA that is. For Windows Server, if you run some as VMs in your environment you better just go shell out for Datacenter for each host. T
Re: (Score:1)