make-initrd: use closureInfo again

[xaverdh]

nix 1.* is quite old now (the original issue #36268 was closed many years ago)

Things done

This is basically a revert of 165b32d

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[philiptaron]

The diff looks good to me. How should it be tested? I ran a few nixos tests, since it looks like this is version of making the initrd is the one used there. They all came out fine.

@mweinelt told me that things that make all nixos tests rebuild ought to go to staging, even if they have small rebuild counts here.

[xaverdh]

The diff looks good to me. How should it be tested? I ran a few nixos tests, since it looks like this is version of making the initrd is the one used there. They all came out fine.

@mweinelt told me that things that make all nixos tests rebuild ought to go to staging, even if they have small rebuild counts here.

retargeted it at staging and removed pathsFromGraph

I tested it by building my own initrd both from a local checkout of nixpkgs and then cherry-picking the changes here and building from that. When comparing the outputs with diff they were bit-for-bit identical actually.

[xaverdh]

Also when testing, make sure the initrd you are building actually uses this code path and not make-initrd-ng, as e.g. systems with boot.initrd.systemd.enable = true.

[philiptaron]

I'm sad to report that you've triggered the following bug in Nix:

• NixOS/nix#11503

For details, read:

• #369694

To show this, run the nixosTests.installer.simple test. With this PR, it fails with an inscrutable curl error (per the bug).

How to fix? I think it's something like:

• #369151 (comment)

But I don't see the line that's triggering the bug immediately.

[Mic92]

@philiptaron how do you see that it triggers: NixOS/nix#11503 ?

[ElvishJerricco]

@philiptaron I believe the issue isn't NixOS/nix#11503 but rather just an ordinary case of a missing dependency in the installer test. closureInfo depends on jq, and while the installation-device.nix module adds jq to system.extraDependencies, it's not adding the right output. It needs to add lib.getDev jq instead. This appears to fix it:

diff --git a/nixos/modules/profiles/installation-device.nix b/nixos/modules/profiles/installation-device.nix
index a4e5a4aac790..f01867c7e29c 100644
--- a/nixos/modules/profiles/installation-device.nix
+++ b/nixos/modules/profiles/installation-device.nix
@@ -102,7 +102,7 @@ with lib;
         stdenv
         stdenvNoCC # for runCommand
         busybox
-        jq # for closureInfo
+        (lib.getDev jq) # for closureInfo
         # For boot.initrd.systemd
         makeInitrdNGTool
       ];

Alternatively, we could append jq.all to that extraDependencies list just to make sure we always have the whole thing.

[philiptaron]

@philiptaron how do you see that it triggers: NixOS/nix#11503 ?

Jörg, @Mic92, My evidence is circumstantial: most triggers of this bug cause curl to be erroneously redownloaded, and that's what I saw here when running the nixosTests.installer.simple test. Per Will, @ElvishJerricco, there are multiple ways to hit that, and my sniff test failed in this instance.

Still need to make the nixosTests.installer.simple test pass

[ElvishJerricco]

See above comment about jq for the installer tests.

nixosTests.installer.simple passes now.

[nixos-discourse]

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/issue-building-linux-kernel-modules-after-flake-update/62322/8

[trofi]

Proposed a minor cleanup to drop no longer used toValidStoreName as:

• make-initrd: drop unused toValidStoreName local #482823