!PSmBFWNKoXmlQBzUQf:helsinki-systems.de

Stage 1 systemd

81 Members
systemd in NixOs's stage 1, replacing the current bash tooling https://github.com/NixOS/nixpkgs/projects/5127 Servers

Load older messages


SenderMessageTime
11 Jun 2023
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgbut sealing a key is just as good06:48:01
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgor, no, you want to do the actual auth crypto on the tpm and never on the host06:48:17
@elvishjerricco:matrix.org@elvishjerricco:matrix.orghm06:48:23
@k900:0upti.meK900Yep06:48:26
@k900:0upti.meK900You should be able to06:48:32
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgyea I dunno the details well enough to be sure06:48:41
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgbut it sounds possible06:48:48
@k900:0upti.meK900 One problem with that is that I'm not sure you can authenticate as the same machine more than once on Tailscale 06:52:21
@k900:0upti.meK900At least with Headscale you have to manually remove the old node06:52:38
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgoh, wait, which level of auth are we talking about? The SSO thing where you actually login for the first time and add the node? Or the auth that occurs just be starting tailscaled back up again?06:53:37
@elvishjerricco:matrix.org@elvishjerricco:matrix.org * oh, wait, which level of auth are we talking about? The SSO thing where you actually login for the first time and add the node? Or the auth that occurs just by starting tailscaled back up again?06:53:50
@k900:0upti.meK900The SSO auth06:54:30
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgI see06:54:37
@k900:0upti.meK900The second auth just uses the token that's persisted on disk06:54:40
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgright I thought you were talking about persisting that token in the TPM instead somehow06:54:58
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgbecause I'm guessing that token is generated locally and then certified by the SSO?06:55:42
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgIf it's generated locally you can do that part on the TPM and leave it there06:55:52
@k900:0upti.meK900I don't know if it is06:55:59
@k900:0upti.meK900But I don't think so?06:56:01
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgbummer06:56:10
@elvishjerricco:matrix.org@elvishjerricco:matrix.orgsecrets should always be generated locally06:56:26
13 Jun 2023
@k900:0upti.meK900 changed their display name from K900 to K900 (Old).20:51:50
@k900:0upti.meK900 invited @k900:conduit.0upti.meK900 (deprecated).21:09:07
@k900:conduit.0upti.meK900 (deprecated) joined the room.21:11:29
14 Jun 2023
@raphi:tapesoftware.net@raphi:tapesoftware.net changed their display name from raphi (element unread channel fix when) to raphi.07:03:17
@arianvp:matrix.orgArianoidc is realy meany for third-party auth though12:32:46
@arianvp:matrix.orgArianTailscale recently also added Webauthn support for first-party log in12:32:56
@arianvp:matrix.orgArianand converting a TPM attestation to a Webauthn sig is actually defined in the CTAP/Webauthn spec12:33:08
@arianvp:matrix.orgArianBut if you wanna go the OIDC route. if you're on Google Cloud they now have an attestation service that allows you to exchange TPM quotes for OIDC tokens signed by accounts.google.com,12:34:02
@arianvp:matrix.orgArianthey'll check the EKCert and see if the hardware is signed by google and then give you an oidc token back12:34:31

Show newer messages


Back to Room ListRoom Version: 6