This project has moved. For the latest updates, please go here.

Observations while encrypting a system disk.

Topics: Feature Requests
Dec 9, 2014 at 11:17 AM
Could there be some sort of indication as to if there has been enough entropy gathered please ? This would also be a nice feature when creating volumes. If you look at this excellent password generator there is a nice, easy to understand display of the entropy gathered.

When encrypting an entire system disk, the user is offered the option to overwrite while encrypting. The minimum overwrite is 3 passes. It would be nice to have the option of 1 single pass. As far as I know there has never been a successful recovery of data after a single pass. Considering VC will go on to overwrite the single pass with encryption almost immediately it is effectively a double overwrite as it is. I am almost certain this is a secure method to use for most people, it will save time and hard drive wear.

I can't remember just now if I saw the 35 pass Gutmann method there. If so can that be removed please ? It is a ridiculous wipe method even Peter Gutmann says so LOL. It makes VC look paranoid and a little odd having such an overwrite pattern. personally I would just have a single pass and a 3 pass option. It's always nice to reduce code :)

Mounir please remember to move these requests to this thread and accept or reject them, thanks ;)
Dec 9, 2014 at 3:02 PM
Hello L0ck,

Would having the high number of passes options available be any value to trying to encrypted a system drive using SSD?

I know the recommend approach is to encrypt a brand new SSD immediately, however in the real world we need to account for SSDs that are being used as a system drive that are being encrypted by VC. :)

Kind Regards,
Dec 9, 2014 at 6:33 PM
With SSD you don't actually need any passes if you are 100% encrypting the drive.

Overwriting is for magnetic disks.
Dec 9, 2014 at 7:39 PM
You misunderstood my point. See the link below.

Dec 10, 2014 at 12:56 PM
Edited Dec 10, 2014 at 12:56 PM
If you are 100% encrypting your drive it should not make any difference. Also many overwrites would wear out SSD's unnecessarily.

If you 100% encrypt your drive VC will write to every sector, or should do. With the single pass VC will write to encrypt the drive the data could be assumed to be gone. There is a feature on modern SSD drives called trim, this is a forensic nightmare but good for us :)

There were rumours on forensic sites / forums that there was spare capacity on a lot of SSD drives which is like a dumping ground for the firmware. If you are into security I suggest you avoid SSD drives as not enough is known about them, particularly when you take many different manufacturers / firmware into account.

I personally don't use SSD drives as they are not reliable enough yet and the capacities are nowhere near what I need.

I thought I would find a link for you, I did this quickly due to time restraints and I am sure there are better pages but it should give you an idea of trim and how different SSD drives are.
Dec 10, 2014 at 11:13 PM
Thank you for the link.

Here is an article about encrypting on SSD from Q3 2012.
Dec 10, 2014 at 11:42 PM
Good link thanks.

I notice that article talks about the hidden 25% space I mentioned above. A nightmare for security minded people.

You will be pleased to know about a previous feature request I made for a product called VeraWipe. This will help SSD users as it will not only provide Magnetic disk users with plausible deniability but also offer the ATA Secure Erase command. So far I think that is the only way to totally wipe a SSD.

The part where that article talks about encryption they seem to be talking about volumes and not WDE. I think you would be much better with WDE when using SSD.

If you are really interested in security and think your threat model requires it, I suggest you buy some magnetic drives before they stop producing them :)

If you really need speed you can try Raptors, as I believe they are very fast.

I personally would not use SSD.
Dec 12, 2014 at 1:09 AM
I have just converted another drive from TrueCrypt to VeraCrypt ... YAY !

I did notice this time that the 35 pass Gutmann method was offered. As mentioned before this is not really a sensible method, even Peter Gutmann agrees :)

Mounir, could this method please be removed and possibly swapped for a single overwrite option ?

Thank you.
Dec 12, 2014 at 11:32 PM
OK, I'll add an option for a single overwrite while leaving the other options (3 passes, 35-passes).

Here is a link that talks about this:
Dec 13, 2014 at 1:20 PM
Thank you, 1 pass is much more sensible :)

I am surprised you have decided to retain the 35 pass method, especially considering the link you provided.

Will you please give it some more consideration ? My concern is that professionals may point to the 35 method and make fun of VeraCrypt. The Gutmann method is almost an industry joke and is often commented on as paranoia (of the worst kind) and it also demonstrates a lack of understanding about overwriting.

I don't want people to be able to point to anything within VeraCrypt in an effort to discredit it. If you really must offer more overwrites how about.....

1 Pass

3 Pass

7 Pass

ideally I personally would prefer ..

1 Pass

3 Pass

... as it would hopefully reduce code in VeraCrypt and also demonstrate we understand about overwriting.

Obviously it is your decision in the end and I will go along with it.

Dec 13, 2014 at 11:26 PM
I know that 35 passes are too much and not needed at all but my approach is to let the user choose what he wants. If there are paranoid people out there who trust only this number of passes and can wait for days, then it is there call and I don't want to impose any thing.