The TPM acts as a second factor (assume PIN is required); if you pulled the disk out and placed it another PC you cannot brute force the plaintext password, even using the $5 wrench method.
I understand the XKCD reference but I don't understand why it does not apply to this case. Does that mean that you can't move your disk to a new machine of yours?
As for the second question, I suppose it depends on the implementation; I suspect there is a way to migrate and transfer the keys for the device for the person who already has access to all of the factors.
If I use a Yubikey and LUKS for example, I can re-use the Yubikey if I move the disk to another machine. I'm not sure how this would work for integrated TPMs and Bitlocker (I don't use Windows), but I believe they have recovery key-files you could use.
And if there is a procedure to let me move the disk to a new machine, somebody can wrench me until I perform that procedure for them.
If you want to you can just not save the recovery password. In that case I guess they'll just beat you to death with that $5 wrench.
I think the implication is that they (the attacker) don't have the machine, or you, or whatever "thing" you have/use to unlock it.
If someone stole your disk, or took a copy of it, they can't offline attack it.
An external hardware device seems preferable for this scenario; if someone steals/snatches your laptop, they won't have your Yubikey or whatever.
In any scenario where the wrenching is an option, I suspect all bets are off, but if wrenching is a real risk, and whatever you're protecting is worth taking a wrenching for (as opposed to just giving it up at the threat of a wrenching, there should probably be some tougher protections which make it clear that the wrenching will gain an attacker nothing, and hopefully save you from it to begin with.
I'm not sure what or how that would work.
I sincerely hope that the guy with the $5 wrench knows this and don't continue hitting until he has the key...
But how does that compare the exactly, performance wise, against Bitlocker and LUKS? (assuming similar strength encryption algorithms are selected)
Veracrypt has a neat benchmark tool so you know the speed beforehand. I suppose most CPUs have native support for the popular algorithms, so the bottleneck really is the disk, not CPU itself or the software.
The principal difference is the native speed of raw IO - NVmEs are an order of magnitude faster than SSDs. TC/VC don't use hardware acceleration, so all the encryption work falls on the CPU. On a machine with a reasonably modern CPU, TC/VC run nearly at drive's native speed.
I assume by that you mean it doesn't use the AES instructions? That's odd to me.
On the other hand, trying to backdoor XOR would give you an astronomically higher amount of white noise. Finding important data mixed in with all the other things a computer XORs would be much harder.
If the attacker has access to your computer, you've already lost. If the attacker built your computer, it was never even a fight.
Given two USB drives, one encrypted with BitLocker and another with TC or VC, chances of master key recovery by Microsoft are definitely not the same.
I don't think those two sentences hold water when put together. In practice, if your risk is master key leakage and theft of the encrypted data by microsoft, you shouldn't be using windows. If you suspect that, MS can have a kernel mode driver masquerading as anything else, and it can just siphon your master key whenever you enter it.
I like VC for the portability of the encrypted .tc files. Keep all my backups as tc files, and recovered from more than one failure using them.
Once had an issue where dropbox corrupted the duplicates, so dont use dropbox anymore.
just gonna drop that like that aint a $10k+ implication, gonna need a src or ref thanks
unless you mean by 3letters, which dont exactly give their backdoors to randoms.
randoms aint coming across 3letter's backdoors, not active/modern ones anyway
it's probably more than a $10k implication..
Is file corruption still an issue?
Never had problems. FWIW, I'm using AES/256 bit keys/HMAC SHA-512 on all of my volumes.
I use static password plus a keyfile. Since the keyfile can be anything you can use something innocent looking which you just have laying around on the machine, like a picture.
This works well on a file you decrypt often. What can happen with infrequently used files is that many years pass, you upgrade computers and/or operating systems, remember to copy your encrypted volumes. When you try to decrypt your volume, you remember the passcode, but forget which "innocuous" file you used as a keyfile, even after you boot up your old computer you had handed down to your cousin.
You can use any file, even something you can quickly download from the Internet. Like the Windows 3.11 file manager icon or your favorite Linux torrent. Only the first 1024 bytes are used so be careful with file types which have a big header.
There are better ways, often the the key can be stored on a hardware token(yubi key etc... ). which presumably offers additional security guarantees around copying the key. but a keydrive is a very easy way to start.
Off tangent(I doubt you will be using openbsd), but the openbsd encryption layer offers some nice features around using keydrives, http://www.openbsd.org/faq/faq14.html#softraidFDE
Performance is great, nearly "metal" SSD speeds. I use it with a hidden partition of an external HDD, and works great too, once I got Windows to stop asking me to formally that partition.
The only feature I wish it could have is better Yubikey (and alike) integration. I can use those static passwords as an encryption key, but I'd need something better.