| 15 Nov 2022 |
@andreas.schraegle:helsinki-systems.de | https://github.com/NixOS/nixpkgs/blob/master/nixos/modules/tasks/filesystems/btrfs.nix#L56 | 22:35:36 |
@andreas.schraegle:helsinki-systems.de | yup. every module just does that afaict. | 22:35:45 |
@elvishjerricco:matrix.org | yea so probably need to go through and add a bunch of explicit definitions of fs packages for initrd in each of those... | 22:36:30 |
@elvishjerricco:matrix.org | did not want to have to do that | 22:36:34 |
K900 | Looks like they already do that for extraUtils | 22:37:29 |
K900 | Just awkwardly | 22:37:32 |
@elvishjerricco:matrix.org | yea | 22:37:34 |
@elvishjerricco:matrix.org | i'll probably go ahead and make a boot.initrd.systemd.fsPackages option or something, in case we decide to make it smarter later, e.g. by only copying fsck.${fsType} and mount.${fsType} or something | 22:38:31 |
@elvishjerricco:matrix.org | or more likely I'll just use extraBin to specify the exact commands already specified for scripted initrd in those modules | 22:39:55 |
@elvishjerricco:matrix.org | or... could just... make dumb fix... and add util-linux in a more special place instead of in system.fsPackages | 22:43:37 |
@elvishjerricco:matrix.org | Could literally just do this:
diff --git a/nixos/modules/system/boot/systemd.nix b/nixos/modules/system/boot/systemd.nix
index d28e6ed0e27..2999b84428f 100644
--- a/nixos/modules/system/boot/systemd.nix
+++ b/nixos/modules/system/boot/systemd.nix
@@ -558,7 +558,7 @@ in
# Environment of PID 1
systemd.managerEnvironment = {
# Doesn't contain systemd itself - everything works so it seems to use the compiled-in value for its tools
- PATH = lib.makeBinPath config.system.fsPackages;
+ PATH = lib.makeBinPath (config.system.fsPackages ++ [cfg.package.util-linux]);
LOCALE_ARCHIVE = "/run/current-system/sw/lib/locale/locale-archive";
TZDIR = "/etc/zoneinfo";
# If SYSTEMD_UNIT_PATH ends with an empty component (":"), the usual unit load path will be appended to the contents of the variable
diff --git a/nixos/modules/tasks/filesystems.nix b/nixos/modules/tasks/filesystems.nix
index 97f02a8c963..399ea9eabe0 100644
--- a/nixos/modules/tasks/filesystems.nix
+++ b/nixos/modules/tasks/filesystems.nix
@@ -300,11 +300,7 @@ in
boot.supportedFilesystems = map (fs: fs.fsType) fileSystems;
# Add the mount helpers to the system path so that `mount' can find them.
- system.fsPackages = [
- pkgs.dosfstools
- # This is needed for the main fsck utility wrapping the fs-specific ones.
- pkgs.util-linux
- ];
+ system.fsPackages = [ pkgs.dosfstools ];
environment.systemPackages = with pkgs; [ fuse3 fuse ] ++ config.system.fsPackages;
| 22:46:41 |
@elvishjerricco:matrix.org | https://github.com/NixOS/nixpkgs/pull/201396 | 23:01:31 |
@elvishjerricco:matrix.org | @ofborg test fsck systemd-initrd-simple would be how I make it do the tests, right? | 23:02:16 |
@elvishjerricco:matrix.org | I do not understand the ofborg-eval failure on the PR... | 23:05:52 |
@elvishjerricco:matrix.org | I based my branch on nixos-unstable-small | 23:06:33 |
@andreas.schraegle:helsinki-systems.de | it's unrelated. a conflict between two PRs (https://github.com/NixOS/nixpkgs/pull/200930 and https://github.com/NixOS/nixpkgs/pull/179406) that got merged independently | 23:11:59 |
@andreas.schraegle:helsinki-systems.de | interesting that nixos-unstable-small is affected by this. I'd expect it not to advance, because the tarball job fails. oh well. | 23:13:02 |
| 16 Nov 2022 |
| @omlet:matrix.org joined the room. | 20:34:21 |
@uep:matrix.org | just got the same for unstable, nice to see already in hand | 21:14:32 |
| @omlet:matrix.org left the room. | 22:05:04 |
@uep:matrix.org | (I just commented out the systemd stage 1 for now, in part because I don't actually need to reboot but I do want the firefox update) | 22:26:19 |
| 18 Nov 2022 |
| * colemickens doesn't fully understand why only some of his initrds failed to build pre-201396 | 02:37:44 |
colemickens | I assume I'm not the only one, but specialisations are a nice way to test this feature. | 03:00:45 |
| 19 Nov 2022 |
@elvishjerricco:matrix.org | Oh, heh, I thought the hibernate test was failing because of some crazy unit dependency issue. Turns out the pre-hibernate boot was the one failing, and that's just because it's lacking the ext3 kernel module | 23:16:48 |
@elvishjerricco:matrix.org | I dunno why that works with scripted initrd. Does that just always include certain kernel modules or something? | 23:17:21 |
| 21 Nov 2022 |
@elvishjerricco:matrix.org | Ah I see what happened. Here, the ext4 kernel module was made conditional on if it was inInitrd, but only for systemd stage 1 | 02:50:33 |
@elvishjerricco:matrix.org | tbh I'm not sure if that's the right thing to do. I mean I know it's better, but that is technically a compatibility break | 02:53:34 |
@elvishjerricco:matrix.org | Oh and in the same diff, fsPackages = [e2fsprogs] was made conditional on systemd -> inInitrd. I think this is definitely incorrect though since fsPackages is relevant to stage 2 as well | 02:54:37 |
@elvishjerricco:matrix.org | https://github.com/NixOS/nixpkgs/pull/202132 | 03:37:38 |
| 22 Nov 2022 |
| @mlyx:matrix.org joined the room. | 21:05:33 |