It's a great step, and ext4 is probably a good place to start.
It seems to me that encryption in a file system is easy to do. We have plenty of libs to handle encryption, and any filesystem that has built in configurable compression already has a pipeline to handle encryption libraries.
But there are plenty of difficulties:
1. We're talking about an important attack vector, so it has to be done correctly, which requires the right experience and knowledge;
2. How to provide correct credentials to a low-level system, that works securely in both server and desktops environments (think both boot time and runtime, cross platform)
3. There are many ways to configure encryption, and many ways to configure a fs, so any approach has to be able to handle varied approaches.
I think that there are many people who have wanted to get into integrating encryption into their fs development, but beside it being relatively easy to do, it's easy to get wrong, that it's a really big task.
These first steps are great to see, but we should probably think of this as one of those "write one to throw away" things that we can use to start the whole process.
The questions I have are:
- do file systems work in similar enough manners that encryption approaches can be shared?
Announcement
Collapse
No announcement yet.
Encryption Support For EXT4
Collapse
X
-
Originally posted by not.sure View PostPerhaps there are scenarios where you just want to encrypt a couple of directories? And not deal with the hassles of FDE? And keep a system bootable and working with basic functionality even if the user forgets the password? Like.. dunno.. a phone?
Leave a comment:
-
Originally posted by not.sure View PostPerhaps there are scenarios where you just want to encrypt a couple of directories?
Leave a comment:
-
Encryption should be available and easy to get set up in order for honest folks to keep their sensitive information protected from prying eyes of hackers and even NSA. Why should the government regulate encryption? There's enough bogus reasons out there and easy to ram a truck through!
Leave a comment:
-
Originally posted by darkbasic View PostI'm tired having to use hundreds of layers on top of layers: native encryption is welcome
Code:# mkfs.ext4 -j -encrypt /dev/sda3
Leave a comment:
-
We need default encryption in the wild ahead of any key escrow law
The front page of the Washington Post today had cops and FBI whining about not being able to get into locked iPhones and proposing key escrow. The Post speculated that the Snowden case would make key escrow legislation impossible to pass. They also suggested key disclosure laws (rubber hose decryption), though those have run into constitutional issued in US courts. It looks like the FBI's strategy is to have a key escrow proposal waiting to go and then wait for any serious crime or terrorism incident they can plausably blame on encryption, then press for the bill. We need to beat them to the punch.
The best insurance against key esroiw in the futue is to have installable Android images with default high security encryption circulating over the Internet as soon as possible and reaching servers out of the reach of governments that might want to regulate encryption. I speak as someone who has sucessfully used disk encryption on a desktop to defeat a police search warrant aimed at protesters. I do not at this time use any smartphone as I do not trust the carriers not to do things like deploy keyloggers against encryption. This could be prevented by having an image that cuts power to the baseband radio until after the encrypted disks are unlocked and blocking all carrier-pushed updates to anything.
Needless to say, I would refuse to comply with a key disclosure law, following the example of British anti-vivisection protesters who defied a key disclosure (UK law allows these) and got away with it. One of the most important reasons for encryption is to lock out police seeking to muzzle political dissent and "plumbers" sent by politicians to punish whisteblowers.
Leave a comment:
-
Originally posted by bitman View PostDoesnt sound like partition would not be fully encrypted, but more like parts of metadata + data. Given that makes me wonder why would anyone want to use it when there is LUKS with everything encrypted. Google can keep it..
Leave a comment:
-
Originally posted by Redi44 View PostWould be more useful if it was developed for BTRFS...
As for BTRFS what I would really love to see would be encryption for specific subvolumes or even directories.
Leave a comment:
-
Originally posted by Modu View PostBtrfs does support encryption but native one.
https://btrfs.wiki.kernel.org/index...._encryption.3F
You shouldn't say IIRC about the info that can be easily reached IMO
(no, disk/file-level encryption doesn't count - he was asking about filesystem-level encryption)
Leave a comment:
-
Originally posted by not.sure View PostPerhaps there are scenarios where you just want to encrypt a couple of directories? And not deal with the hassles of FDE? And keep a system bootable and working with basic functionality even if the user forgets the password? Like.. dunno.. a phone?
Leave a comment:
Leave a comment: