[staging] coreutils: 8.32 -> 9.0

[dasJ]

Motivation for this change

This might speed up cp and mv in nix builds (on certain file systems) since the tools use copy_file_range now.

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandbox = true set in nix.conf? (See Nix manual)
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests) → I built the small channel on x86_64-linux
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• 21.11 Release Notes (or backporting 21.05 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.

[r-burns]

Release notes: https://lists.gnu.org/archive/html/coreutils-announce/2021-09/msg00000.html

[dasJ]

Marking this as ready since a lot of relevant builds are succeeding on the systems I have easy access to (x86_64-linux, aarch64-linux).

I'm still building the entire nixos/release-small.nix for these systems but apart from unrelated eval errors, everything seems to be going well.
I will report back when the builds are done.

[r-burns]

Cherry-picked 97c7af3f4dab4fb715b65786e3be5ac1b507618e (this PR) back to 7c86d10 (master).

Encountered during x86_64-darwin stdenv bootstrap: make[2]: *** No rule to make target '-lintl', needed by 'src/coreutils'. Stop.

[dasJ]

@r-burns Fixed. Currently building stdenv on x86_64-darwin

[r-burns]

Okay, one very long rebuild later, darwin stdenv is building.

Also tested on x86_64-linux: stdenv, pkgsMusl.coreutils, pkgsStatic.coreutils, pkgsCross.aarch64-multiplatform.coreutils. Problematic leaf packages will probably need to be addressed individually in staging-next.

[dasJ]

Still seems to fail on aarch64-linux while building coreutils in the checkPhase

FAIL: tests/tail-2/assert
=========================

x
==> a <==
x

==> foo <==
FAIL tests/tail-2/assert.sh (exit status: 1)

Test seems flaky on aarch64

[dasJ]

This is probably good to go now. I was able to build:

• The small channel on x86_64-linux
• stdenv for x86_64-darwin
• Parts of the small channel on aarch64 due to my limited computing resources on that platform. No builds failed, just most of them are still running. Succeeding are stdenv, gettext, and vim

[r-burns]

I noticed some Go packages were segfaulting in #141684, and the segfaults reproduce locally. I bisected the jx failure https://hydra.nixos.org/build/156019589 to this bump. Any ideas?

[vcunat]

So... we revert this until someone manages to resolve these issues somehow (or they disappear with so further upstream update)? #141684 (comment)

[dasJ]

I'll take a look at it and report back in the next couple of hours

[vcunat]

For reference, reverted on *-darwin only in PR #142836

[vcunat]

And there's some Linux regression as well :-( #119110 (comment)