tags:

views:

1088

answers:

7

Is there some open-sourced, well-documented and used DRM framework/library?

I want to write some framework for buying and selling digital stuff, where I want to implement, somehow, for the seller to have the possibility to lock the files with some sort of DRM, where only authorised computers would be able to open it (something like iTunes FairPlay).

It can, and probably has to, involve contacting my server with some login credentials.

On the other hand, I want the client to be open-sourced, and probably the server too .. is that even possible? Security through obscurity does not work, but DRM is not exactly "security"...

All I was able to find is this discussion on slashdot with the exact same problem, but it ended with "DRM IS BAD", and Sun's DReaM project, but I have no clue how to get to the actual code/usage of the framework on their site.

If you think Open Sourced DRM is not possible, tell me so.

+5  A: 

For an open-source DRM solution there will likely be an open-source crack.

Many of DRM solutions actually work based on the "security-by-obscurity" principle. Meaning some of them not broken yet because their models are kept secret. Even that does not prevent major DRM system from getting broken.

User
Agreed. Proprietary DRM is broken very, very quickly. I wouldn't be optimistic about the strength of an open source solution. Of course, DRM is significantly flawed to begin with. It doesn't matter if your DRM is good enough to stop 99.999% of users from making a copy or using a program. One person figures out how to break it and your DRM is worthless. The file is shared, the program has a crack available, etc. You really need a hardware solution to 'succeed' and even then it's not a sure thing.
Steven Richards
well ... I thought, there are really strong and virtually un-breakable crypting algorithms, both symetrical and asymetrical ... so, there can be some DRM system, that works (maybe built in a way I haven't thought about) and is unbreakable by nature. But, probably not
Karel Bílek
Well, the thing is... your player will need the original content to be able to play it. So at some stage of the game the file gets decrypted. Then writing the crack would actually mean updating the algorithm to decrypt the content offine rather than in live streaming mode.
User
+7  A: 

Open-source DRM is virtually impossible.

The point of DRM is to prevent a user from decrypting some data, while at the same time allowing them to do so under some circumstances. The theoretical framework of encryption makes this absurd - how can the user only have the key to decrypt some data if they're using it for approved purposes?

The solution adopted by existing DRM systems is to go to ridiculous lengths to hide the key - a perfect example of security by obscurity - but if the source code is available it is trivial to simply modify the code to hand the key to the user. At this point it doesn't matter how good your encryption is, the user has everything they need to break it.

One solution to all this is using the Trusted Platform Module on some machines to verify the binary image of all software that might be able to access the key, and ensure the key itself is inaccessible (encrypted by a key derived by a secret burned-into-hardware value). However I'm not sure how common TPMs are, and moreover this is very much defeating the point of open-sourcing the software in the first place, as you would be unable to modify it in any way without losing access to the DRM keys.

Finally, on a more practical note, DRM seems to be on its way out anyway - eg, iTunes has gotten rid of DRM entirely, and the movie industry seems to be moving in that direction as well...

bdonlan
"if the source code is available it is trivial to simply modify the code to hand the key to the user" -- assuming you can load code on the device, which is why DRM tends to work better on mobile phones than on PCs, and why GPL3 includes that anti-tivo clause.
Steve Jessop
Yes, the TPM would serve to prevent loading code in the PC case. However if one can't load any code at all, then making DRM is fairly trivial, since you don't have to hide the key very well.
bdonlan
"if one can't load any code at all, then making DRM is fairly trivial". In practice, the approach on e.g. Symbian is that the user can load code, but such code is prevented by the kernel from accessing the DRM APIs. So it's not that you "can't load code at all", just that you can't load DRM-related code at all. Whether this is "trivial" probably depends whether you're the one who has to implement the Symbian kernel.
Steve Jessop
Oh, and the key is hidden by process boundaries. So a little bit like a TPM in some ways, but different approach in others.
Steve Jessop
+3  A: 

This claims to be an open source implementation of OMA DRM2. I assume it contains the software components needed to build the server and client, leaving the hardware as an exercise for the reader:

http://sourceforge.net/projects/openipmp

License is MPL, which is a non-GPL-compatible FOSS license.

I have no experience of this implementation, but a little of OMA DRM, and it seemed at the time to be a workable DRM scheme, as much as any DRM scheme is workable. The OMA DRM standard is well-documented, and is (or at least has been) widely-used by the mobile phone industry.

The fundamental problem with open-source DRM is that although all of the algorithms and source code can be published without harming the scheme, client devices have to be "trusted" by the rights issuer to respect the rights, i.e. not do anything forbidden. This is incompatible with FOSS, which says that the user of a device should have full control over what it does.

Security through obscurity does not work, but DRM is not exactly "security"

Security through obscurity of algorithms is usually weak. Security through secrecy of information is the only way to do crypto, signing, etc. DRM does not require obscurity of algorithms (which is why OMA DRM is a published standard, and how come the source for an implementation can be published and freely usable), but it does require that the player device have access to information (some kind of key) which the user of the device does not, and which is not part of the algorithm/source.

Normally, security protects the owner/user of a device from a threat model of external attackers. In the DRM threat model, the owner/user of the device is the attacker, and the rights owner is being defended. If the device's user has full control over it, then clearly in principle this is game over.

In practice it may not be quite that immediate, but in the open source case, allowing people to write their own DRM clients which prevent them from copying your rights-protected data would be asking them to be astonishingly honest.

Users can sometimes be persuaded to be law-abiding, in which case DRM takes on the role of reminding them that if they're jumping through hoops to work around the restrictions, then they may be breaking the law.

Steve Jessop
openipmp was last updated 2006 - wondering if it actually runs on Android and if somebody has actually already implemented it successfully on it.
Mathias Lin
+1  A: 

Isn't 'Open-source DRM' kind of a oxymoron?

Nick Whaley
A: 

If you consider that DRM can also be used to protect private content and to make sure that only certain authenticated users can read it then the view is completely different. You can store a video on the server and encrypt the content using a user private key. Only this user can then read the video: even someone hacking the server would not be able to read it without the owner's unique key. In this case, open source DRM does make sense.

lec
A: 

Its worth to mention it depends on what content you want to protect. If you want to protect content that you yourself own, and the damages if content leaks is not devastating, then you could for example look in the previously mentioned Open IPMP. Suns DReaM project is still availible online, it is however worth to notice that the project is closed due to inactivity. A problem arises when you want to protect content that you dont own. It is very possible that the content owners will not allow their content to be protected by an open source drm solution. Content owners tend to go with proprietary solutions like Microsoft WMDRM/PlayReady, Marlin or Helix. So if your framework will be used to sell for example Hollywood content, I would not recommend putting time in an open source drm solution.

Clean
+1  A: 

DRM is broken because it relies on "security-by-obscurity" principle.

There are other solutions like watermarking.
When you sell something to a client, you watermark it with the name of the client. If you find this watermarking on sharing network, you will know where is the leak. If the client knowns there is watermarking (I think it is more honest to warn him), he will probably not share.

webreac