| 13 Sep 2022 |
@elvishjerricco:matrix.org | I wouldn't say generally, since it's not something we've ever done so I don't think nixos users have an expectation of it | 01:57:48 |
@elvishjerricco:matrix.org | It doesn't seem like a bad thing to do though | 01:58:01 |
@elvishjerricco:matrix.org | so long as it doesn't explode the size of your initrd | 01:58:22 |
@oxalica:matrix.org | In reply to @elvishjerricco:matrix.org so long as it doesn't explode the size of your initrd 🤔 It do doubles the initrd, to 32MiB now | 01:58:51 |
@elvishjerricco:matrix.org | oof | 01:58:57 |
@elvishjerricco:matrix.org | then that's a tradeoff you'll have to decide on for yourself | 01:59:13 |
| 15 Sep 2022 |
| @lassulus:nixos.dev left the room. | 13:19:06 |
| 18 Sep 2022 |
@kranzes:matrix.org | https://github.com/NixOS/nixpkgs/pull/189676 | 21:10:25 |
@kranzes:matrix.org | Can we get this reviewed and merged? | 21:10:43 |
| @kranzes:matrix.org changed their display name from Kranzes to Ilan Joselevich (Kranzes). | 21:12:55 |
| 20 Sep 2022 |
| @m_algery:leohoo.xyz joined the room. | 15:18:07 |
| @m_algery:leohoo.xyz left the room. | 15:18:10 |
| 23 Sep 2022 |
@elvishjerricco:matrix.org | Zhaofeng Li: While I'm largely OK with https://github.com/NixOS/nixpkgs/pull/189676/ I had another thought about the patch. Couldn't we just change that hard coded absolute path to something in /etc and configure said directory in NixOS with environment.etc rather than taking on the risk that we could be wrong about relative paths being safe? | 01:23:13 |
@elvishjerricco:matrix.org | Just seems like changing the path and populating it would be a less severe change | 01:24:31 |
Zhaofeng Li | If we add it in a fixed runtime path, something like /run/opengl-driver (I remember an RFC to change the name) seems better than /etc | 02:14:24 |
@elvishjerricco:matrix.org | sure, either way | 02:14:45 |
@elvishjerricco:matrix.org | just seemed better than patching actual code, y'know? | 02:15:04 |
Zhaofeng Li | Actually /run/opengl-driver doesn't appear to be very extensible at the moment 🥲 | 02:17:29 |
@elvishjerricco:matrix.org | yea i've had ideas for things to do similar to that but there isn't really a good mechanism in place. That'd be a nice improvement to make to nixos | 02:18:38 |
@oxalica:matrix.org | In reply to @zhaofeng:zhaofeng.li If we add it in a fixed runtime path, something like /run/opengl-driver (I remember an RFC to change the name) seems better than /etc is /run already populated in initrd stage? | 02:40:36 |
@elvishjerricco:matrix.org | Not exactly so we'd have to also set it up separately in initrd before luks devices start coming online. | 02:47:23 |
@oxalica:matrix.org | In reply to @elvishjerricco:matrix.org Not exactly so we'd have to also set it up separately in initrd before luks devices start coming online. mounting another tmpfs just for this seems not convincing | 03:44:41 |
@elvishjerricco:matrix.org | i mean it's the same tmpfs | 03:45:08 |
@elvishjerricco:matrix.org | /run is the same from before anything starts in stage 1 all the way to the very end of shutdown | 03:45:33 |
@elvishjerricco:matrix.org | well, systemd starts as pid 1 in stage 1 and mounts /run, /dev, /proc, and /sys before starting anything | 03:45:58 |
@elvishjerricco:matrix.org | but yea we would have to do the setup twice, once in stage 1 and once in stage 2 | 03:46:32 |
@janne.hess:helsinki-systems.de | But who populates /run in stage 1? | 08:15:46 |
@janne.hess:helsinki-systems.de | We're about to rebuild the activation script from stage 2 here | 08:16:25 |
@elvishjerricco:matrix.org | Janne Heß: It'd just have to be a service with Before=cryptsetup-pre.target | 08:16:35 |
@elvishjerricco:matrix.org | In reply to @janne.hess:helsinki-systems.de We're about to rebuild the activation script from stage 2 here (as a brief aside, arguably stage 1 and stage 2 should actually be almost the same infrastructure, and it should simply be possible to make a NixOS config that's as stripped down as stage 1 needs to be) | 08:17:56 |