You are not logged in.
Pages: 1
Does this mean that using LUKS on OPAL NVMe (I don't mean BIOS HDD password) will offload all en/decryption computing to NVMe hardware, hence removing the penalty on CPU and enable LUKS to reach full PCI4.0 NVMe speeds?
Last edited by Lockheed (2023-11-14 09:39:56)
Offline
If you use cryptsetup with that merge included which is not yet part of any release and you follow the instructions to use the experimental OPAL support then encryption/decryption will be performed by the device not the kernel.
Offline
Which, if I understand correctly, means "YES".
Thanks!
Offline
If you use cryptsetup with that merge included which is not yet part of any release
Which, if I understand correctly, means "YES".
So YES if you are building from a git checkout of cryptsetup rather than the package supplied by Arch and not using linux-lts and using --hw-opal not --hw-opal-only as that does not use dmcrypt. This is also based on the assumption the CPU is a bottle neck.
Offline
Well, I am thinking from a perspective of few months into the future. When do you suppose we can expect the release of dmcrypt supporting that?
Offline
Pages: 1