gcc: tighten condition for inhibit_libc=true

[ghost]

Description of changes

The situation described in the comment preceding export inhibit_libc=true does not match the conditional which follows it. Specifically, the comment says that this line is meant for "clang builds gcc" situations, yet it is enabled even when !stdenv.cc.isClang.

This commit tightens the conditional to make it match the situation described in the comment.

Needed for:

• gccWithoutTargetLibc: if glibc, enable threads and pass headers #241208

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, 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.

[trofi]

Why the change is needed? Does it fix anything?

If it does then I suggest update the the description to say so. And maybe comment should hint why it's undesirable for other cases.

gcc -> clang is only one example where config does not change. Another one would be pkgsCross.gnu64:

nix-repl> pkgsCross.gnu64.stdenv.hostPlatform == stdenv.hostPlatform
false

pkgsCross.gnu64 requires building a new libc without reliance on the host libc.

[ghost]

Why the change is needed?

Because what the comment says and what the code does are not the same. One of them must change. If you prefer to change the comment please submit a PR that explains why we inhibit_libc in such a huge class of situations.

If it does then I suggest update the the description to say so.

I think it's the other way around -- right now we have this inhibit_libc=true with no description of why it's there (except if stdenv.cc.isClang).

Does it fix anything?

Yes, right now there's no way to not turn on inhibit_libc for the first-stage cross compiler, like #241208 does.

inhibit_libc doesn't just prevent gcc from using the buildPlatform headers -- it prevents it from using any headers at all. So if we already have the targetPlatform headers, this inhibit_libc is blocking us from being able to use them.

[trofi]

I'll remove myself from reviewers then.

[ghost]

Rebased.

[Artturin]

Rebuilds caused by

The conditional was added in ea8e124 #182666

No diff in pkgsLLVM.hello

pkgsCross.aarch64-multiplatform.hello no longer has inhibit_libc=true

rnix-lsp
pkgsCross.aarch64-multiplatform.hello
pkgsCross.aarch64-multiplatform.pkgsLLVM.stdenv.cc

work

[yu-re-ka]

This broke pkgsMusl.nix:

       error: attribute 'useLLVM' missing

       at /nix/store/m2vqij78lfy9rvrmws264jpr7q78q98w-source/pkgs/development/compilers/gcc/common/pre-configure.nix:125:45:

          124|                       targetPlatform.config == hostPlatform.config &&
          125|                       (stdenv.cc.isClang || stdenv.targetPlatform.useLLVM)) ''
             |                                             ^
          126|   export inhibit_libc=true

$ nix eval -f . pkgsMusl.pkgsStatic.stdenv.targetPlatform.useLLVM
error: attribute 'useLLVM' in selection path 'pkgsMusl.pkgsStatic.stdenv.targetPlatform.useLLVM' not found

$ 

[Artturin]

#259592

[yu-re-ka]

This also broke pkgsMusl.pkgsStatic.buildPackages.gccWithoutTargetLibc.cc, which is also required for pkgsMusl.nix: https://termbin.com/wf5i

[alyssais]

#260181