darwin: remove -arch from cc/ld wrapper

[paparodeo]

-arch in the cc/ld wrapper causes problems when specifying different --target options, which would succeed except for the conflicting -arch specifier.

Reverts fc0456b and b26e0ba

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/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 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.

[paparodeo]

cc @emilazy

[emilazy]

@thefloweringash do you remember why you added these originally?

[thefloweringash]

@thefloweringash do you remember why you added these originally?

It was long enough ago that some details have been lost. There's a little background in #114817. At the time I was focused on bringing up aarch64-darwin by cross compiling stdenv from x86_64-darwin. For the CC wrapper I don't see why this couldn't have been covered by -target, but perhaps that was somehow inconvenient? For the bintools wrapper, it seems important that invocations of ld use the correct architecture. When invoked through clang the Darwin toolchain driver will pass -arch, but I would want to be sure that a direct invocation in a cross compiling environment used the correct arch value.

There's precedent in the nixpkgs clang wrapper for making some flags conditional on the absence of the --target argument, see #291901. If the flag is still required for cross completion, perhaps it could be similarly conditional?

[emilazy]

Thanks for the reply! We’ve tested that basic Darwin–Darwin cross still works on this branch, and looking at the ld64 source code it seems it infers the default architecture from the format of its input files rather than the platform it’s running on or anything like that. So I think that hopefully we should be good to land this, as any invocation that doesn’t result in a default being computed would presumably fail with the Xcode toolchain anyway.