Microsoft FAT Patent Upheld

theodp writes "After initially rejecting Microsoft's File Allocation Table (FAT) patents, the USPTO has ruled them valid. From the article: 'Microsoft has won a debate where they were the only party allowed to speak, in that the patent re-examination process bars the public from rebutting arguments made by Microsoft, said unimpressed Public Patent Foundation President Dan Ravicher.'"

FAT's valuable

[typical]

A patent on FAT doesn't really have much of a use for them now; at least none that I can think of. Just let the filesystem become an open standard now, MS.

USB HID Mass Storage devices apparently usually use FAT.

Now, granted, I don't know whether they implement long filename support (which is what Microsoft's patent is on, IIRC), but FAT is still very relevant in the embedded device world, even if desktop boxes are now using NTFS instead of FAT.

Re:So now...

[tpgp]

What does that mean to companies that sell stuff like USB flash drives or CF cards? They'll obviously have to pay royalties, of course,

Yep, they will pricing has been set to 25c per unit [dpreview.com].

Utterly crippling in the low margin, high volume USB storage market (especially at the low end)

and that means a mass migration to a new filesystem to avoid such payments.

And exactly what filesystem could that be? That is supported out of the box by 95% of desktop PCs?

This - if anyone was still wondering why a monopoly is so dangerous in the hands of an immoral company like MS.

You can use your overwhelming advantage in one market (desktop PCs) to exert influence in another.

But what new FS will that be? FAT32? EXT2/3?

Fat32? Patents cover it.

EXT2/3? Get real. Who wants to install 3rd party drivers every time you plugin your USB device?

Re:So now...

[tpgp]

It would be stupid for Microsoft to enforce this patent because of the migration issue. If they were smart, they'd immediately turn around and put this into the public domain. If they don't, I can't see the marketplace relying on the hope that someday Microsoft won't try to enforce the patent. So if they were protecting their own interests that's fine, but they need to send a clear message that this move was only done to make sure that nobody would screw them

Wishful thinking aside - Microsoft have allready stated they're going to enforce the patent:

From Microsoft's FAT licensing page: [microsoft.com]

A license for manufacturers of certain consumer electronics devices--Pricing for this license is $0.25 per unit for each of the following types of devices that use removable solid state media to store data:

        * Portable digital still cameras
        * Portable digital video cameras
        * Portable digital still/video cameras
        * Portable digital audio players
        * Portable digital video players
        * Portable digital audio and video players
        * Multifunction printers
        * Electronic photo frames
        * Electronic musical instruments
        * Standard televisions

At 25c a unity, thats going to add up to a helluva lot of money.

Re:Less important than it sounds

[bosson]

Get your facts right. They are using FAT-patents to get license fees from storage manufacturers. And they started using it *after* storage manufacturers where using vfat as a standard for flash drives.

So the methods bears all the marks of asserting broad patents against standardization initiatives. The set of patents they hold could just as easy be used to kill off mozilla or any other competitor, but they should be playing it safe not to upset any legislators too early.

Re:So now...

[Anonymous Coward]

This - if anyone was still wondering why a monopoly is so dangerous in the hands of an immoral company like MS.

The problem here, if any problem exists to begin with, is not monopolies, but patents. $ony holds a patent on the CD, and gets a royalty payment for every single CD sold out there. Is that any better?

Utterly crippling in the low margin, high volume USB storage market (especially at the low end)

Not really - the extra cost will just get passed on to the consumer. Those who had >25c/unit margins before may get a slight advantage at the cost of decreased profits if they don't increase the price, but I'm going to bet every single manufacturer will unilaterally raise prices by 35c-50c (after all, if they're going to raise prices, might as well make some extra profit out of it) per unit and be done with it. That is assuming the current profit margins aren't large enough to just absorb the cost outright.
Unless you can find storage media that go for just a few dollars per unit, this ultimately won't even matter in the long run.

Re:The patents

[Anonymous Coward]

The problem is that FAT 8.3 directory entry holds the actual data (first cluster, size, time & date, attributes) and LFN entries are linked to that one. No 8.3 name entry = no file/directory.

Good Thing?

[TwentyQuestions]

I'm actually glade MS won this. I think it will help clear the way for more devices to use more secure and open-source friendly file systems. But I doubt MS will try to crack the whip on people making technology to read FAT. It just doesnt make sense, plus the income would be so low. And as for drives coming preformatted with FAT. Alot of the flash drives and even some MP3 players I have received from Japan use FAT but dont come preformatted.

Re:Less important than it sounds

[a_n_d_e_r_s]

The reason anyone hasn't heard much about it is that NO ONE wants to be on the other end of a lawsuit from Microsoft. Its way too expensive.

There are free software developers whos been contacted by Microsoft. So yes Microsoft enforce their upatents. One example is Virtualdub and the patented ASF format.

Re:Linux?

[Anonymous Coward]

The claims in US patent No. 5579517 - the patnet that was subject to re-examination - are rather strange, and to my reasding are not infringed by a Linux system reading or writing a vfat file system. The analysys is not straightforward, but as a clue to those used to looking at patent claims, think about the effect of the opening words of the claim: "In a computer system having a processor running an operating system..." followed by the words "said short filename including at most a maximum number of characters that is permissible by the operating system", i.e., not some other operating system but by the executing operating system.

US Patent 5758352 is more of a worry, because it relates to the way in which long and short filenames are stored in a directory structure by an (i.e., any) operating system. I cannot find any reference to this potentially much more damaging patent having been re-examined.

Note that the claims are not infringed by any system that does not support both long and short filenames. It is not FAT per se that is being protected, it is the backwards-compatible DOS filenames and the particular manner in which they are stored. You have to read the claims to understand this.

So the question about Linux etc., requires an analysis of the claims with an understanding of how the Linux FS driver works.

HTH
Anonymous European Patent Attorney

Re:oups, vfat gone?

[Cus]

Why wait for xfs/reiser support? I needed a file system that could handle >4GB files and read/writable from Windows and Linux - I ended up installing Ext2FSD [sourceforge.net] and it does the job nicely.

Chain of events

[daBass]

1. Microsoft spearheads USB standard
2. "Mass Storage Class" added to USB that is so low level, the OS uses it as any disk, needing to support it's file systems
3. 95% of computers run windows and the ones that support USB only support FAT, forcing device manufacturers to use that as filesystem.
4. Patent filesystem and demand royalties after the fact
5. No need for "???"
6. Profit!

Yup, they planned this all along, the sneaky bastards.

Re:Food chain

[dabraun]

Actually, it's hard to see how the flash drives are even impacted by this issue - they don't have a filesystem (unless the manufacturer formats them which they really don't have to) - the filesystem is used by the software that reads and writes to them. So, it may impact digital cameras, or other OS's that write to FAT, or even printers that can read directly from memory cards - but I don't see how it would impact the card itself any more than it would impact a hard drive or other form of generic storage.

play them at their own game

[Stanneh]

why dont the manufacurers team up make their own file system and force microsoft to pay them loyalty's to support it in windows?

Re:Short file names?

[tpgp]

Every digital camera I've ever used (which is only about 5 or 6, so I may be wrong here) has only used short filenames. 'IMG_1234.JPG'. These aren't even covered by the patent.

From the page I linked to:

Additionally, the FAT file system licensing package includes rights to FAT file system innovations for which Microsoft has filed a claim for a patent that the U.S. Patent Office has not yet granted. This licensing program also provides licensees with rights to Microsoft FAT file system issued and pending patents outside the United States and to the Microsoft FAT file system specification and certain test specifications.

So why do they pay royalties?

To prevent the chance that Microsoft will sue you in the future or in an overseas market.

Re:Ship unformatted dammit.

[Vo0k]

-o noadate

generic FAT is a very bad idea for flash media. Every write operation no matter where on the media causes a write operation in one area. Your flash media can survive a million writes so theoretically you can write 100 million files as long as you randomize/distribute their locations evenly over the media. Sorry, with each file write, no matter where, FAT gets updated, a single location gets written. After a million writes it dies. Bummer.
Manufacturers overcome it by placing FAT on separate chip of extended lifetime or virtually mapping 'flat diskspace' to physical chip address space and migrating it from time to time, so that FAT "wanders". Won't be quite possible if you format the media at home.

Re:Food chain

[redhog]

There are Free Software ext2 drivers for all major OSes:
Windows: http://freshmeat.net/projects/ext2ifs/ [freshmeat.net]
MacOS X: http://freshmeat.net/projects/ext2fs/ [freshmeat.net]
OS/2: http://freshmeat.net/projects/ext2-os2/ [freshmeat.net]

The problem is, they don't come pre-installed...

Re:alternatives and extent

[m50d]

What about one of the ISO filesystems? There's an ISO for CDROM filesystems, and I imagine that thing isn't always read-only. If anyone has a flash disk and wants to format it as an ISO9660 filesystem and see if Windows can read/write it, that would be nice of them. I don't have either.

ISO9660 is completely non-writeable - the filesystem is designed in such a way that you simply can't write to it. However, its successor, UDF, is writeable, and is already being used by flash drives which are too big for FAT (>32GB).

Second, what product is hit by this? People are going on about shipping unformatted media, but think about it: most devices that use the media have to speak FAT as well. Your camera can't write a file to the flash card if it doesn't understand how to read and write to it, even if rudimentary. The unformatted argument only works for media that will only be used on a PC, which seems like it will be a small minority.

Absolutely. Anything that has to access its own disk is at risk - the main things I see are cameras, MP3 players and possibly PDAs. A camera could just use another filesystem and be accessible via PTP, and since that just specifies how to transfer files, I suppose in theory it could be used for MP3 players as well, it has support from all major OSes.

Embedded filesystems library affected

[Yseboodt]

I'm the author of the Embedded filesystems library. (http://sf.net/projects/efsl [sf.net])

I've read the patents, they all cover the long filenames ability in the FAT filesystem. So basically as long as I do not implement long filesystem support, the EFSL should be free from patent problems.

If anyone with a deeper understanding of legalese is willing to comment on this, I and the users of EFSL would be grateful.

Since EFSL is targetted at embedded devices, it is used commercially (I am using it in a commercial product as well, and I know of several other projects that are doing the same) and thus the companies using it should know wheter or not they can use EFSL without paying a fee to microsoft.

FAT is about the ugliest filesystem around, it's a shame they dare to ask licensing fees for it.

Not for the Mac...

[John Nowak]

There is no ext2/ext3 support for 10.4. It only exists for earlier versions.

The Patents

[Midnight Thunder]

I think to understand what it means to companies, then we need to look at what the patents are:

Patent: 5,579,517 [164.195.100.11]
Title: Common name space for long and short filenames
Filed: 24 April 1995

An operating system provides a common name space for both long filenames and short filenames. In this common namespace, a long filename and a short filename are provided for each file. Each file has a short filename directory entry and may have at least one long filename directory entry associated with it. The number of long filename directory entries that are associated with a file depends on the number of characters in the long filename of the file. The long filename directory entries are configured to minimize compatibility problems with existing installed program bases.

Patent: 5,758,352 [164.195.100.11]
Title: Common name space for long and short filenames
Filed: 5 September 1996

An operating system provides a common name space for both long filenames and short filenames. In this common namespace, a long filename and a short filename are provided for each file. Each file has a short filename directory entry and may have at least one long filename directory entry associated with it. The number of long filename directory entries that are associated with a file depends on the number of characters in the long filename of the file. The long filename directory entries are configured to minimize compatibility problems with existing installed program bases.

Patent: 6,286,013 [164.195.100.11]
Title: Method and system for providing a common name space for long and short file names in an operating system
Filed: 28 January 1997

An operating system provides a common name space for both long filenames and short filenames. In this common namespace, a long filename and a short filename are provided for each file. Each file has a short filename directory entry and may have at least one long filename directory entry associated with it. The number of long filename directory entries that are associated with a file depends on the number of characters in the long filename of the file. The long filename directory entries are configured to minimize compatibility problems with existing installed program bases.

So the patents in question all cover the same issue of a "common name space for long and short filenames". This would effect anyone using vfat and also potentially effect Rockridge and Joliet extensions for ISO 9660.

One thing to note, from looking at the licensing page, is that only "consumer electronics devices" and "removable solid state media manufacturers" are targeted. For the moment operating systems aren't listed.

One thing I have to ask myself whether makers of digital cameras would be legaly required to have to pay this license, despite them being listed in the "consumer electronics devices" section. The reason I ask this is because all the digital cameras I have seen to date still use 8.3 format file names (for example my Nikon is DSCN0000.jpg), therefore they are not using the technologies referenced by the patents.

Re:Food chain

[baadger]

There are infact several choices of upto date maintained Ext2/3 drivers for Windows.

1. ext2fsd [sourceforge.net] which has support for Win64 (x64)
2. ext2ifs [swin.edu.au] by John Newbigin (the one linked by parent). It says on the website "This version probably does not work under XP SP2".
3. ext2ifs [fs-driver.org] by Stephan Schreiber. It's freeware but doesn't appear to be Open Source (so presumeably contains no GPL'd code). There are Windows XP screenshots on the site and it's x86 only.

Is it possible M$ wants a slice of the iPod pie?

[NimbleSquirrel]

I know the obvious impications for this patent include M$ trying to shut out OSes such as Linux, but I don't think that is as likely as everyone makes out. For a start it seems that M$ are going after device manufacturers that have physical products that use FAT filesystems. Bringing a lawsuit against software that just interacts with FAT would be a little harder to fight (especially with the likes of IBM and Novell to potentially fight against). I think preventing people from interacting with FAT filesystems would have M$ in trouble from an antitrust standpoint, so they are going after devices where there is an actual FAT filesystem rather than the potential a mere interaction with one (if that view makes sense - it was rather late when I wrote this)

There is one device that I can think off that needs the FAT filesystem preinstalled: the humble iPod.

All the recent iPods come with FAT32 as the filesystem (originally added for Windows users). They originally used HFS+, but that is no longer the case (and hasn't been for quite some time).

http://en.wikipedia.org/wiki/IPod [wikipedia.org]

I am aware that the FAT Licensing page [microsoft.com] puts a supposed $250,000 cap, but this is M$ and they can change their minds or have other nasty clauses in Licensing agreements that would be unfavourable to companies like Apple. Towards the bottom of the page they even say: "Sometimes, companies may want to negotiate broader or narrower rights than the standard Microsoft license for FAT file systems. In this case, prices may vary." M$ could easily use this to shut iPods out of the Windows market (if they are forced to return to using HFS+ filesystems).

These patents could be very handy iPod killers if M$ wants to use them as such.

Re:The Patents

[rwhamann]

Cameras will be affected if they can read long filenames. I've changed the names of pictures on my SD cards before taking them into the photo printer. When I put the card back in the camera and scroll through the pix, the new filename shows up.

Re:Food chain

[Marillion]

A device like a USB key fob is blank storage. Like a really big floppy. It doesn't violate because it doesn't have an algorithm that implements FAT. Cameras, on the other hand, have to save their images in a structured way. They do implement a FAT algorithm.

Re:Is it possible M$ wants a slice of the iPod pie

[gellenburg]

The one fortunate thing about this is not only does it apply to iPods, it also applies to every single other media player out there too.

Now, if M$ chooses not to enforce their patent against WMA devices, things could get interesting. Legally, they could. However, I think you'll see a huge public outcry and backlash if they chose to.

MS Fat patents have been enforceable since 1996

[kansas1051]

The MS FAT patents weren't originally "rejected" by the USPTO. They were examined, allowed, and issued. The patents were placed into ex parte reexamination after issuance (by the USPTO at the request of PubPat) due to various prior art that the USPTO didn't consider. During the reexamination, the USPTO issued an initial rejection (as is always the case) which Microsoft was able to overcome. The FAT patents were never invalidated or rendered unenforceable. The patents at issue were filed in 1995 and issued in 1996, so your argument that these patents were somehow hidden or unenforceable during that time is entirely baseless.

Re:I knew it

[Anonymous Coward]

Happen to actually catch the CES show? He's not as skinny as Gates, but he's hardly at the weight he used to be.

Re:USB Sticks and CF cards

[deander2]

Actually someone named George B. Selden (who had never built an automobile) held a patent on a "road engine". All American car manufacturers paid royalties to him until Henry Ford came along, who blatantly ignored it (and later got it overturned).

Patent trolls are nothing new to society.

Re:Food chain

[diamondsw]

The problem is, they don't come supported either. From the web page:

It is written for OS X 10.2-10.3.

No, they didn't just omit 10.4 accidentally, or not update the page. It doesn't work at all:

Apple completely changed the kernel interfaces in Tiger and as such, a lot of work needs to be done to get the Ext2 driver running on Tiger. I started some of this work last year after WWDC, but there is still a lot to do and I don't have the time to finish things up right now.

...

I've started getting back to bringing up the driver on Tiger. Progress is going well, everything is compiling (but not necessarily running) except for the vnops file. I still have to implement locking and then testing before a release can happen.

Re:What about UDF?

[moyix]

Update:

• This page indicates that Windows does *not* have UDF write support [aumha.org] without the use of a third party program, like DirectCD or Nero's offering, InCD.
• Linux does have write support for UDF filesystems [chalmers.se], but it's not turned on in the default kernel config.
• It's difficult to find anything definitive about OSX, but the consensus from the interweb is that OS X can't currently handle writing to UDF without third party software.

So this probably won't work as a universal filesystem unless some pressure is put on MS and Apple to get native support for writing to UDF, unfortunately :\

Re:More accurate history of FAT

[swilver]

It is silly to use on harddisks :) With FAT16 for example, a 2 GB partition was forced to use 32kB blocks, resulting in huge overhead for small files, not to mention having a ~60000 file limit for each partition.

FAT32 isn't much better. For a modern 120 GB harddisk, the FAT table would consume 125 MB (using 4 kB blocks), which is a bit too big to keep in RAM all the time. Large file performance would start to suffer, especially seek performance. Although the concept of FAT is nice, for larger disks using bitmaps + extents results in far better performance with far less memory overhead.

Re:More accurate history of FAT

[po8]

Actually, FAT was a pretty straightforward clone of the CP/M filesystem, with a bit of "optimization" for the 8088. I'm not sure who at DR did the CP/M filesystem.

(I had a CP/M box back in the day, and my first internship job was to build FAT (PC-DOS 1.1) filesystem tools for a UNIX workstation. Microsoft not only didn't charge us royalties (AFAIK), they provided us with internal documentation on how FAT worked. Those were the days...)

Re:It's Non-obvious If...Allow me to correct here

[Nom du Keyboard]

The original fat file system was based on the cp/m filesystem. Remember that the original MS-DOS was based on an 8086 operating system that was in turn based on cp/m. So the non-long file name file system is covered (or should be) by prior art. However MS DID invent the extensions for long file names.

Allow me to correct here.

It's not the 8.3 length filenames that are patented. It's the way directories are structured and free space are handled in the FAT (File Allocation Table) file system.

As for MS-DOS being based on CP/M (Control Program for Microprocessors), PC-DOS (first incarnation) and MS-DOS (non-IBM version) are based on QDOS (Quick & Dirty Operating System) which was written as a CP/M work-alike that Bill Gates was able to purchase rights to on very short notice. While it is likely that QDOS used the CP/M file format on floppy discs so that it could read CP/M discs, I've never heard that confirmed.

But if MS-DOS, by way of QDOS, was based on CP/M floppy discs, than that should be more than sufficient prior art to invalidate at least the original FAT system, which was originally FAT12 (FAT 12-bit). And FAT16/32 are very obvious extensions of FAT12.

But there's another reason MS doesn't deserve a FAT file patent. IMHO and IANAL, they waited far too long to file for it. You shouldn't be able to patent something so far after the fact that everyone has adopted using it because it wasn't patented. That's just Malum In Se (wrong in itself).

Re:So now...

[Anonymous Coward]

You wouldn't want to use standard journalling on a flash drive. IIRC for each write cycle at least 3 write actions are required: log in the journal that a write will be done (has to be synced to the disk), do the write, log in the journal that the write action ended successful. With flash, where you can only erase block-wise, this is not a good idea - for one its very slow, and on the other hand, the flash supports only so many write cycles. For journalling, special handling is needed as implemented e.g. in jffs2.

And FAT is better? It does at least three writes, one to the directory structure, one to the FAT, and then the file itself. Both the directory entry and the FAT require block erasures. If anything, a log based filesystem would be optimal, because the journalling would be inherant. Only valid updates to the log would be considered as part of the filesystem, so interrupted writes would simply disappear from the filesystem without modifying what was on it before. Fragmentation is not an issue because of the flat access time, and you just pack the log wherever it fits and update your pointer of logs every so many writes, it is easy to set aside a list of blocks to use for future log entry writes once and write that to the superblock, and then check all entries for valid log entries (checksummed or similar). I don't know of any filesystems that actually do this right now, and certainly none that is widely used.