stdenv Rust fixes

[alyssais]

Description of changes

See commit messages. Fixes a couple of regressions from #233628.

wasm32-rustc is still not ideal. As followup, I'd like to investigate whether we could just always include the wasm32-unknown-unknown target in rustc builds, since it doesn't require any libc dependency or anything.

Things done

• 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/)
• 23.11 Release Notes (or backporting 23.05 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.

[Ericson2314]

Yeah agreed this seems strictly better.

[andresilva]

FWIW this works fine with a package that I maintain that depends on rustc-wasm32. The wasi config threw me off at first until I read the commit message, maybe adding a small comment there is warranted.

[pbsds]

took a while to build, but pagefind functions as expected

[alyssais]

@Ericson2314 can I get your ack for dropping the assertion that at most one of "rust" and "rustc" is given? Caused eval to fail; reason now explained in the first commit message.

[ghost]

Suggested change

	};
	} // (args.rust or {});

Otherwise people have no way to override any of this stuff without editing this file.

Previously // args was the last thing in the //-chain, which provided an escape hatch.

[alyssais]

I don't think that works. If you pass { rust.platform = { arch = …; os = …; }; }, target-family is supposed to be filled in automatically. If we did a merge like this at the end, the whole platform attrset would be overridden with just the values you passed, so target-family wouldn't end up being there.

[alyssais]

I'm also not confident that this needs to be supported, at least not right now. It was never supported when these were in rust.lib, except by using an overlay, which you could also use in this situation. It would be a bit ugly, but making workarounds for things be ugly when they're fixable bugs in the underlying code is not the worst thing, because it encourages fixing the thing rather than working around it.

[ghost]

@ofborg eval

[alyssais]

@Ericson2314 can I get your ack for dropping the assertion that at most one of "rust" and "rustc" is given? Caused eval to fail; reason now explained in the first commit message.

Also had to make it possible to pass in rustcTargetSpec for similar reasons, also now explained in commit message.

[emilylange]

Pardon the question, but what is left to get this merged?
Given this neatly resolves the currently present breakage of pagefind, polkadot and lldap :)

[alyssais]

I would still like @Ericson2314's ack for the changes I mentioned.

[bendlas]

I'm currently running this patchset to fix lldap for my deployments.

[andresilva]

IMO we should merge this as-is. It's strictly better than the current situation, since there are broken packages depending on this fix, and if @Ericson2314 has any concerns they can be addressed in a follow-up PR.

[github-actions]

Successfully created backport PR for release-23.11:

• [Backport release-23.11] stdenv Rust fixes #269615

[misuzu]

This change broke rustc on armv7l-linux and riscv64-linux when evaluating with NixOS system:

62f7a6dcc1a2baf50e757a6a0bce1083692e0e56 is the first bad commit
commit 62f7a6dcc1a2baf50e757a6a0bce1083692e0e56
Author: Alyssa Ross <[email protected]>
Date:   Fri Nov 17 20:16:22 2023 +0100

    lib.systems.elaborate: fix passing `rust`

    Usually, attributes passed explicitly to elaborate take precedence
    over the elaborated ones, but since we also elaborate the nested
    "rust" attrset, we need to push that one level down, so the rest of
    "rust" is still filled in if you just pass
    { rust = { config = ... } }.

    I've had to drop the assertion that checked that at most one of "rust"
    and "rustc" was part of the un-elaborated system, because doing this
    broke passing an elaborated system in, which should be idempotent.

    For the same reason, I've also had to make it possible for
    rust.rustcTargetSpec to be passed in.  Otherwise, on the second call,
    since platform was filled in by the first, the custom target file
    would be constructed.  The only other way to avoid this would be to
    compare the platform attrs to all built in Rust targets to check it
    wasn't one of those, and that isn't feasible.

    Fixes: e3e57b8f1885 ("lib.systems: elaborate Rust metadata")

 lib/systems/default.nix | 184 ++++++++++++++++++++++++------------------------
 1 file changed, 92 insertions(+), 92 deletions(-)

Used commands:

git bisect start
git bisect good 19cbff5
git bisect bad e92039b55bcd
git bisect run nix-build nixos -A pkgs.rustc --argstr system riscv64-linux -I nixos-config=test.nix --dry-run

test.nix looks like this:

{
  boot.loader.grub.device = "/dev/sda";
  fileSystems."/".device = "/dev/sda1";
  services.sshd.enable = true;
}

See #271000 for more info