Skip to contentSkip to conversation

Replacing VFAT as filesystem on removeable media

DISCUSSION Martin Steigerwald 7 participants 12 messages Mar 31, 2009 – Apr 6, 2009 VFS

lore
Share & tools
lore
#0 lore raw

Hi!

Now as TomTom appears to have surrendered to Microsoft and Microsoft seems to have accepted this deal probably in order to not find out that their patents are void, I think replacing VFAT as standard cross platform filesystem on removeable media would be an even better idea than before [1].

Granted I believe the patents will be made void one day, but even then I think it would make sense to replace VFAT for technical reasons.

Such a filesystem IMHO should have the following features:
- cross platform with implementations for Linux, Windows, Mac OS X at least, also the varios BSD variantes come to my mind
- open source
- probably some support for flash media without or with bad wear leveling
- some kind of journaling or other metadata consistency guarentee

People use ext2 as an alternative, but that lacks journaling.

I could also think of UDF with write support, but I am not sure whether Windows and Mac OS X has write support.

Or probably even a new filesystem as long as people appear to write one filesystem after another these days.

Or some kind of FAT *without* compatibility hacks, but I think this would still be an inferior solution as long as it doesn't provide for metadata consistency.

Maybe this could become some kind of Linux Foundation or FSF joint effort? Together with advertising and advocacy of free software users this could probably really replace VFAT in the long term.

What do you think? What other features would make sense to for such a filesystem.

I am willing to test such a filesystem and help with documentation as well as advocacy. But for coding I better start with something easier ;-).

[1]
http://www.h-online.com/open/TomTom-Microsoft-settle-patent-dispute--/news/112964

Ciao,

Signature · 2 lines
Martin 'Helios' Steigerwald - http://www.Lichtvoll.de
GPG: 03B0 0D6C 0040 0710 4AFA  B82F 991B EAAC A599 84C7
#1 lore raw

Another approach which occurs to me would be to come up with a "filesystem over USB" protocol so that the actual on-disk format is no longer relevant. With a standard or reasonably common FSoUSB implementation, a device maker could choose a disk filesystem that's appropriate to their purpose and hide the implementation details within the device. For instance, an MP3 player could use a flash-specific disk filesystem but would be able hide the details of this from the host PC.

This might not be so desirable for simple devices (like USB card readers, perhaps) but I imagine it would fairly straightforward on media players, GPSes, etc that run a "proper" OS. The most pain would arguably be in *somebody* having to write a filesystem driver for Windows but at least it would only need writing once ;-)

I've seen some talk of this sort of approach online but nothing too specific. A Linux USB website suggested that a certain camera protocol might be suitable as a basis for files-over-USB (there's even a FUSE module for this, I think). Another possibility that occurs would be to write a USB transport for 9P, which looked like it might be fairly straightforward...

Just a thought, anyhow.

Cheers,
Mark

Quoted 45 lines · On Tuesday 31 March 2009 09:15:28 Martin Steigerwald wrote:

Hi!

Now as TomTom appears to have surrendered to Microsoft and Microsoft seems to have accepted this deal probably in order to not find out that their patents are void, I think replacing VFAT as standard cross platform filesystem on removeable media would be an even better idea than before [1].

Granted I believe the patents will be made void one day, but even then I think it would make sense to replace VFAT for technical reasons.

Such a filesystem IMHO should have the following features:
- cross platform with implementations for Linux, Windows, Mac OS X at least, also the varios BSD variantes come to my mind
- open source
- probably some support for flash media without or with bad wear leveling
- some kind of journaling or other metadata consistency guarentee

People use ext2 as an alternative, but that lacks journaling.

I could also think of UDF with write support, but I am not sure whether Windows and Mac OS X has write support.

Or probably even a new filesystem as long as people appear to write one filesystem after another these days.

Or some kind of FAT *without* compatibility hacks, but I think this would still be an inferior solution as long as it doesn't provide for metadata consistency.

Maybe this could become some kind of Linux Foundation or FSF joint effort? Together with advertising and advocacy of free software users this could probably really replace VFAT in the long term.

What do you think? What other features would make sense to for such a filesystem.

I am willing to test such a filesystem and help with documentation as well as advocacy. But for coding I better start with something easier ;-).

[1]
http://www.h-online.com/open/TomTom-Microsoft-settle-patent-dispute--/news/
112964

Ciao,

#2 lore raw
Quoted 2 lines · Mark Williamson

Another approach which occurs to me would be to come up with a "filesystem over USB" protocol so that the actual on-disk format is no longer relevant.

Look for MTP. It's what you want, made by Microsoft, and enables all sorts of DRM niceties.

Xav

#3 lore raw

Hi Xav,

Quoted 7 lines · On Tuesday 31 March 2009 14:47:07 Xavier Bestel wrote:
Quoted 3 lines · Mark Williamson

Another approach which occurs to me would be to come up with a "filesystem over USB" protocol so that the actual on-disk format is no longer relevant.

Look for MTP. It's what you want, made by Microsoft, and enables all sorts of DRM niceties.

Ah, thanks for that. It looks fairly close to what I had in mind, although I think there are some differences (it transfers whole files at a time and incorporates device control stuff - neither of which is really necessary in the scheme I imagined).

According to Wikipedia it is now an official USB device class, which sounds attractive. I've not heard of Linux providing an MTP responder ("server"), which I find mildly surprising as it sounds like it would be useful, particularly given the range of advanced filesystems Linux devices might like to leverage under the hood. It seems odd if an embedded Linux company hasn't already coded this up for a device ... ?

I can see how a file-based protocol would be "useful" in enforcing DRM-like protections in a device implementation but I don't think that such protocols fundamentally enables any nastiness that couldn't be enforced in other ways by a motivated pro-DRM device manufacturer. Sounds like MTP is designed to facilitate DRM if the communications endpoints want it; not something I'm keen on but also not something that rules it out as a useful standard.

A file-based transfer protocol seems like it would be useful to have for "sufficiently intelligent" devices, whether it's MTP or something else.

Cheers,
Mark

#4 lore raw
Quoted 5 lines · Mark Williamson

I've not heard of Linux providing an MTP responder ("server"), which I find mildly surprising as it sounds like it would be useful, particularly given the range of advanced filesystems Linux devices might like to leverage under the hood. It seems odd if an embedded Linux company hasn't already coded this up for a device ... ?

I'm speaking out of my ass here, but seeing how Microsoft managed to sneak some patents into something as trivial as FAT, I'm pretty sure MTP is a hell of a minefield.

Xav

#5 lore raw
Quoted 3 lines · On Tuesday 31 March 2009 16:21:37 Xavier Bestel wrote:

I'm speaking out of my ass here, but seeing how Microsoft managed to sneak some patents into something as trivial as FAT, I'm pretty sure MTP is a hell of a minefield.

That did occur to me too but I was somewhat hopeful that perhaps this would be mitigated by the (eventual?) adoption as a usb.org standard.

The spec and an adopters' legal agreement around it is here:
http://www.usb.org/developers/devclass_docs/MTP_1.0.zip

I scanned the agreement and it didn't look like it was obviously evil but I'm not really qualified to make that judgement ;-) It had some promising words in it like "zero royalty" but I would be much happier if someone with legal knowhow (and preferably experience of the usb.org standards procedures) decoded it for me / us!

Assuming the spec is legally "safe" to implement, I would have thought it would be generally beneficial for device manufacturers to support both "initiator" and "responder" endpoints under Linux.

Cheers,
Mark

#6 lore raw
In reply to Martin Steigerwald
Quoted 1 line

Now as TomTom appears to have surrendered to Microsoft and Microsoft

seems

Quoted 7 lines

to have accepted this deal probably in order to not find out that their patents are void, I think replacing VFAT as standard cross platform filesystem on removeable media would be an even better idea than before [1].

[1]

http://www.h-online.com/open/TomTom-Microsoft-settle-patent-dispute--/news/112964

How do you mean "replace"?

VFAT's standardness today comes from the fact that virtually every personal computer in the world today and for the foreseeable future is able to read and write VFAT. Do you envision some other filesystem format achieving that status?

--
Bryan Henderson IBM Almaden Research Center
San Jose CA Storage Systems

#7 lore raw

Am Donnerstag 02 April 2009 schrieb Bryan Henderson:

Quoted 14 lines
Quoted 1 line

Now as TomTom appears to have surrendered to Microsoft and Microsoft

seems

Quoted 6 lines

to have accepted this deal probably in order to not find out that their patents are void, I think replacing VFAT as standard cross platform filesystem on removeable media would be an even better idea than before [1].

[1]

http://www.h-online.com/open/TomTom-Microsoft-settle-patent-dispute--/n
ews/112964

Quoted 6 lines

How do you mean "replace"?

VFAT's standardness today comes from the fact that virtually every personal computer in the world today and for the foreseeable future is able to read and write VFAT. Do you envision some other filesystem format achieving that status?

That was my (possibly crazy) idea yes. ;)

Signature · 2 lines
Martin 'Helios' Steigerwald - http://www.Lichtvoll.de
GPG: 03B0 0D6C 0040 0710 4AFA  B82F 991B EAAC A599 84C7
#8 lore raw
Quoted 10 lines · Martin Steigerwald
Quoted 7 lines

How do you mean "replace"?

VFAT's standardness today comes from the fact that virtually every personal computer in the world today and for the foreseeable future is able to read and write VFAT. Do you envision some other filesystem format achieving that status?

That was my (possibly crazy) idea yes. ;)

Ironically enough, one of the better filesystems for being supported by many OSes is probably ext2. However, that doesn't mean it is even in the same rough ballpark as (V)FAT.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
#9 lore raw
Quoted 9 lines · El Sat, 04 Apr 2009 23:34:50 -0700 "H. Peter Anvin" <[email protected]> escribió:
Quoted 8 lines · Martin Steigerwald
Quoted 7 lines

How do you mean "replace"?

VFAT's standardness today comes from the fact that virtually every personal computer in the world today and for the foreseeable future is able to read and write VFAT. Do you envision some other filesystem format achieving that status?

Quoted 6 lines
Quoted 2 lines

That was my (possibly crazy) idea yes. ;)

Ironically enough, one of the better filesystems for being supported by many OSes is probably ext2. However, that doesn't mean it is even in the same rough ballpark as (V)FAT.

What about UDF ?

Quoted 3 lines

-hpa

#10 lore raw
Quoted 6 lines · Alejandro Riveira Fernández
Quoted 4 lines
Quoted 1 line

Ironically enough, one of the better filesystems for being supported by many OSes is probably ext2. However, that doesn't mean it is even in the same rough ballpark as (V)FAT.

What about UDF ?

Good point. UDF is probably widely supported, but again, not in the same ballpark. I don't personally know how complex UDF is to implement on a small memory device.

-hpa

Signature · 7 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#11 lore raw
Quoted 6 lines · H. Peter Anvin
Quoted 5 lines · Alejandro Riveira Fernández
Quoted 4 lines
Quoted 1 line

Ironically enough, one of the better filesystems for being supported by many OSes is probably ext2. However, that doesn't mean it is even in the same rough ballpark as (V)FAT.

Quoted 8 lines
Quoted 1 line

What about UDF ?

Good point. UDF is probably widely supported, but again, not in the same ballpark. I don't personally know how complex UDF is to implement on a small memory device.

-hpa

DVD players are small memory devices and they don't actually implement UDF.

(DVD movies have UDF filesystems, but they're constructed such that the files that the DVD player needs are at specific fixed offsets on the disc, and the DVD player never actually interprets the UDF filesystem.)

Also, there's several different revisions of UDF, which causes some level of operating system compatibility problems.

Fortunately, the UDF flavor likely to be used on magnetic/flash media is also the flavor that's the most compatible, so that makes things easier for everyone.

Signature · 6 lines
Nicholas Miell <[email protected]>

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html