gcc14: fix internal compiler error on aarch64

[prusnak]

This PR combines fixes of these 3 upstream issues:

• https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118501
• https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118892
• https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119133

All 3 are related to segfault (internal compiler error) caused by the force_subreg function and are fixed by using the new force_lowpart_subreg function

Hopefully all 3 fixes will become part of the 14.3.0 release but until then we should patch them locally since the ICE can be triggered by compiling torch 2.6.0 on aarch64 and thus this blocks torch update in #377785.

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/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 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.

[prusnak]

I rebuilt #377785 on top of this branch and confirmed the GCC 14 patch fixed the torch 2.6.0 build issue on aarch64.

[prusnak]

what do you think? @trofi @emilazy @vcunat

[GaetanLepage]

NB: This is blocking the pytorch update.

[emilazy]

Perhaps we should bump the snapshot again, if these fixes are already upstream?

[prusnak]

Perhaps we should bump the snapshot again, if these fixes are already upstream?

They are upstream, but only in trunk (gcc 15) - they have not been backported to gcc 14 branch yet as can be seen in the upstream bugzilla.

[vcunat]

I don't object. I just don't have capacity to verify that backporting this makes sense (probably it's OK).

[flokli]

Successfully built this on aarch64, and considering it's in gcc trunk like this and fixes aarch64 builds, I think it's fine to land this 👍