treewide: don’t set ulimit -n in builds

[emilazy]

This avoids having to work around the incredibly low Darwin default in a bunch of builds. The Nix daemon is meant to get a higher limit from launchd, but it seems to not work or not be passed down to builds for whatever reason. For now, this is a simple and expedient fix for a recurring problem.

This was a problem on the NixOS infra that was fixed so now these are redundant even without the stdenv patch.

See:

• increase max file descriptors on darwin by default #173657
• darwin launchd NumberOfFiles not applying nix#6557

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.

[emilazy]

It turns out that this was actually caused by a Hydra misconfiguration that affected Linux too (albeit to a lesser degree). We’re in the process of testing a fix and if it all goes well then we can probably do the tree‐wide portion of this without the stdenv change.

[paparodeo]

looks good sans the change to setup.sh

[paparodeo]

Suggested change

	# Increase the soft file descriptor limit if it’s too low. This is
	# mostly for the benefit of Darwin, where the default is a positively
	# anaemic 256 FDs by default. We use 10240 because it’s the default
	# value of `kern.maxfilesperproc` on macOS 10.12, but this can be
	# increased when the minimum macOS version is bumped.
	if [[ "$(ulimit -n)" < 10240 ]]; then
	ulimit -S -n 10240 || true
	fi

[philiptaron]

Simple as.

[emilazy]

I’ve dropped the stdenv commit. This is still likely to make things break more under sudo nix build … and such? But arguably that’s a footgun, especially if you don’t configure good limits yourself. Hydra having low limits would have kept it mostly working in the past, but all bets are off now.

[emilazy]

Fixed the Darwin glibc cross patch probably. Embarrassing mistake. Someone should verify the build.

[paparodeo]

Fixed the Darwin glibc cross patch probably. Embarrassing mistake. Someone should verify the build.

missed that in reviewing it too. diffs of diffs are the worst. and it also seems like the formatter switches rules on where to put the '' and indention based on if there is a + or not, which is cluttering the diffs too.

edit: tho hide whitespace takes care of formatter sins.

[philiptaron]

I also missed the glibc thing when reading through. 🫣

[emilazy]

The patch needs fixing.