https://infosec.exchange/@alexandreborges/115766847703175693
@wklj @bruienne fair point, I had forgotten about the always-on transparent encryption in Apple Silicon. I'm on the fence as to whether I'd quite count the T2 SEP as being equivalent to what they're doing here, but that's mostly architectural semantics.
still, neat to see the idea expanding to other platforms.
@gsuberland @bruienne I believe the I/O path on T2 Macs went through the T2 all the time, and nothing unencrypted was available to the Intel side. But that was long ago.
Definitely agreed it's good to have it on all platforms -- it's nice to have stuff encrypted all the time and pay no performance penalty for software encryption! Nobody should have to pay a big overhead to have their data secured!
@wklj @bruienne personally the main thing I want to see out of this is a path to proper attestation in hardware FDE, with a standard implementation.
one of the main issues I have with OPAL etc. is that you're kinda stuck with a "just trust me bro" situation with disk vendors; it only really resolves the superficial compliance side of the problem through being able to say the disks were certified, and I've seen enough "certified secure" stuff in my time to know how that plays out.
@wklj @gsuberland And even then it takes years (at this point) to get FIPS certifications for all the crypto modules processed 😵💫
@gsuberland @wklj Bill said it well, but here's some more reading material re: the AES in Secure Enclave:
https://support.apple.com/guide/security/volume-encryption-with-filevault-sec4c6dc1b6e/1/web/1
https://support.apple.com/guide/security/secure-enclave-sec59b0b31ff/web
TL;DR even when FileVault is "disabled" on T2 Macs and later the FDE functionally acts as Bitlocker in TPM-only mode. Enable FileVault and we're roughly in TPM+PIN mode (though we recommend passphrases, not PINs) 🔐