bintools-wrapper: eliminate build references when cross-compiling

[lopsided98]

Description of changes

The wrapper had references to build platform bash (stdenv.shell) and expand-response-params. I'm not sure if this is the best way to fix it, but it seems to work. I was able to build both the native x86_64 stdenv and cross-compiled armv6l stdenv.

A similar fix is likely needed for cc-wrapper.

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/)
• 22.05 Release Notes (or backporting 21.11 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
  • (Release notes changes) Ran nixos/doc/manual/md-to-db.sh to update generated release notes
• Fits CONTRIBUTING.md.

[Mindavi]

The shell change seems ok, but not sure about the buildPackages change. Can't that be changed in the derivation file itself?

[lopsided98]

Can't that be changed in the derivation file itself?

That would also require several more complex changes in the bootstrapping code for each platform.

[lopsided98]

cc @Ericson2314, since you might have some familiarity with this

[AlexandreTunstall]

I think I've encountered the issue this PR is trying to fix while trying to cross-compile GHC.

A similar fix is needed for cc-wrapper.

$ nix build nixpkgs#pkgsCross.riscv64.gcc -o result-gcc
$ head -n1 result-gcc/bin/gcc
#! /nix/store/51sszqz1d9kpx480scb1vllc00kxlx79-bash-5.2-p15/bin/bash
$ nix run nixpkgs#file -- -L /nix/store/51sszqz1d9kpx480scb1vllc00kxlx79-bash-5.2-p15/bin/bash
/nix/store/51sszqz1d9kpx480scb1vllc00kxlx79-bash-5.2-p15/bin/bash: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /nix/store/3n58xw4373jp0ljirf06d8077j15pc4j-glibc-2.37-8/lib/ld-linux-x86-64.so.2, BuildID[sha1]=4fa92b70b17d9f0c3f7715fdd952e56cc1efc118, for GNU/Linux 3.10.0, not stripped

As shown above, pkgsCross.riscv64.gcc produces a cc-wrapper that does not run on riscv64.

[leifhelm]

@Mindavi buildPackages is only used to compile expand-response-params (in both bintools-wrapper and cc-wrapper). I implemented it in the derivation. It’s just an if that uses pkgs instead of buildPackages when cross compiling.
With a lot of if statements the big rebuild can be avoided. i.e. Only change stuff when cross-compiling.
Is that neater?

[leifhelm]

After adding

    shell = runtimeShell;
    buildPackages = pkgs;

to wrapCCWith as well, nix path-info -r .#pkgsCross.riscv64.gcc lists only riscv64 packages.

/nix/store/372zfza5hk7zhwiyjky382zv15y42cgw-riscv64-unknown-linux-gnu-stage-static-gcc-12.3.0-libgcc
/nix/store/q5v876nk37wxryqwkbp7a4pi3swgp1kq-glibc-riscv64-unknown-linux-gnu-2.37-8
/nix/store/zrjx5jdyj382cqy0xr98bwdc6s2k0hh5-zlib-riscv64-unknown-linux-gnu-1.2.13
/nix/store/19gsi9r8r2qznv2s909f77r0ammv9hwg-binutils-riscv64-unknown-linux-gnu-2.40-lib
/nix/store/243qy2cc408hbhzgl2d03wcrr5ydzxsv-pcre2-riscv64-unknown-linux-gnu-10.42
/nix/store/5d10w6w90k0kkz7cs94zn6sprkwlf81i-linux-headers-6.4
/nix/store/628wlhmn912wp9ssl8827mnidyhls15k-glibc-riscv64-unknown-linux-gnu-2.37-8-bin
/nix/store/6vpbh764zwa41sv69ih0dl2ln1zjz6qb-glibc-riscv64-unknown-linux-gnu-2.37-8-dev
/nix/store/yv5y6x78vkwj8iarfdadqwj280nyx3ci-riscv64-unknown-linux-gnu-stage-final-gcc-12.3.0-libgcc
/nix/store/cw2x43p6qr0a1m6jy3l6vd9v07zl3d4x-riscv64-unknown-linux-gnu-stage-final-gcc-12.3.0-lib
/nix/store/9zmm4qw8jls9b0xmz31x1dnfrh3xb0jy-gmp-with-cxx-riscv64-unknown-linux-gnu-6.2.1
/nix/store/ib8rbz77wp7q29qb66cp5xmw1fsxd123-isl-riscv64-unknown-linux-gnu-0.20
/nix/store/v4cb7pk25qc3wfrsdqvi4m290yyspz7h-mpfr-riscv64-unknown-linux-gnu-4.2.0
/nix/store/ijc0s193yr6dspzd1b5gp2ix862xxffi-libmpc-riscv64-unknown-linux-gnu-1.3.1
/nix/store/879knymhp35xd1l7x7nw08rawmygspcb-gcc-riscv64-unknown-linux-gnu-12.3.0-libgcc
/nix/store/r2wp6szfjqr9bfz2j7lgp7kny4avj48k-gcc-riscv64-unknown-linux-gnu-12.3.0-lib
/nix/store/7k9mg81z8jd28q5r6w3k0zmsgjz59kq0-gcc-riscv64-unknown-linux-gnu-12.3.0
/nix/store/qcs9wc8ncrdmn1kpa6f9z69a74zr2303-gcc-riscv64-unknown-linux-gnu-12.3.0-man
/nix/store/9qdvszlfyd9d6pg3w3k4xskmvz54yxgi-gcc-wrapper-12.3.0-man
/nix/store/kabjm6bgbg20bjp75p16zqy8laggl141-gnugrep-riscv64-unknown-linux-gnu-3.11
/nix/store/q93y53hhh2rrn5kw8m1aq0dnv066cvhc-binutils-riscv64-unknown-linux-gnu-2.40
/nix/store/n6vnp7381idsaki1hym1mmq6d67mg8bb-attr-riscv64-unknown-linux-gnu-2.5.1
/nix/store/hq8rh39fgxya8aypdhlpm8g62xasks82-acl-riscv64-unknown-linux-gnu-2.3.1
/nix/store/ssgsrr7yb9salalx0hw4f5rfw5dypnvx-coreutils-riscv64-unknown-linux-gnu-9.3
/nix/store/v84ifs4qqa32xw7y0vfdmw6fccfr2pr9-bash-5.2-p15-riscv64-unknown-linux-gnu
/nix/store/x6wr3wskk10vv5m1pmf9q0szxhgvsmdf-expand-response-params-riscv64-unknown-linux-gnu
/nix/store/rc5hkf5axsfgmqwwy8is532v8srwnx90-binutils-wrapper-2.40
/nix/store/cbq45bn0csibjx97fzdkrjsarl254jlj-gcc-wrapper-12.3.0

[sternenseemann]

Best solution is to fix the incorrect logic in the wrappers themselves and adjust the stdenv bootstrapping accordingly: #304861

[sternenseemann]

Should be resolved on master!