| 21 Apr 2025 |
 Tristan Ross | But the original PR just works basically | 00:40:12 |
| Tristan Ross | 🤷 | 00:40:27 |
 dramforever | oh i didn't realize you were testing a PR
| 01:22:05 |
| Tristan Ross | I'm picking up development of the PR, I hope to fix it in 25.05 | 04:54:32 |
|  @creepinson:matrix.org left the room. | 04:58:48 |
 Grimmauld (any/all) | wait can't i use pkgsCross to compile from aarch64 for x86_64-linux? | 06:57:46 |
 K900 | You can unless cross blows up | 06:57:57 |
| Grimmauld (any/all) | its not listing as something in the attrset | 06:58:57 |
| Grimmauld (any/all) | [grimmauld@build05:~/nixpkgs]$ nix build .#pkgsCross.x86_64-
.#pkgsCross.x86_64-darwin .#pkgsCross.x86_64-freebsd .#pkgsCross.x86_64-netbsd-llvm .#pkgsCross.x86_64-unknown-redox
.#pkgsCross.x86_64-embedded .#pkgsCross.x86_64-netbsd .#pkgsCross.x86_64-openbsd
[grimmauld@build05:~/nixpkgs]$ nix build .#pkgsCross.x86_64-linux.libapparmor
error: flake 'git+file:///home/grimmauld/nixpkgs' does not provide attribute 'packages.aarch64-linux.pkgsCross.x86_64-linux.libapparmor', 'legacyPackages.aarch64-linux.pkgsCross.x86_64-linux.libapparmor' or 'pkgsCross.x86_64-linux.libapparmor'
[grimmauld@build05:~/nixpkgs]$ nix build .#pkgsCross.x86_64-linux
error: flake 'git+file:///home/grimmauld/nixpkgs' does not provide attribute 'packages.aarch64-linux.pkgsCross.x86_64-linux', 'legacyPackages.aarch64-linux.pkgsCross.x86_64-linux' or 'pkgsCross.x86_64-linux'
[grimmauld@build05:~/nixpkgs]$
| 06:59:34 |
| K900 | It's pkgsCross.gnu64 | 06:59:51 |
| K900 | The subset names are stupid | 06:59:56 |
| Grimmauld (any/all) | huh | 07:00:21 |
| Grimmauld (any/all) | thanks | 07:00:23 |
| Grimmauld (any/all) | time to test whether i can build apparmor as cross now | 07:01:51 |
| Grimmauld (any/all) | I am working on https://github.com/NixOS/nixpkgs/pull/400430
Everything works, except apparmor-pam. That one doesn't find the pkg-config things. It was broken before too, i didn't break it - but i'd love to fix it! Any ideas? | 09:41:41 |
| Grimmauld (any/all) | Relevant log:
@nix { "action": "setPhase", "phase": "buildPhase" }
build flags: SHELL=/nix/store/si7wv0psd4df7bqq71ppcfw7sprlz9j8-bash-5.2p37/bin/bash USE_SYSTEM=1
/nix/store/si7wv0psd4df7bqq71ppcfw7sprlz9j8-bash-5.2p37/bin/bash: line 1: pkg-config: command not found
/nix/store/si7wv0psd4df7bqq71ppcfw7sprlz9j8-bash-5.2p37/bin/bash: line 1: /nix/store/w4gm8546mvar4v259b57yj6hm537pql9-glibc-x86_64-unknown-linux-gnu-2.40-66-bin/bin/ldconfig: cannot execute binary file: Exec format error
Makefile:64: ***
************************************************************************
Unable to find libapparmor installed on this system; either
install libapparmor devel packages, set the LIBAPPARMOR variable
manually, or build against in-tree libapparmor.
************************************************************************
. Stop.
| 09:51:14 |
| Grimmauld (any/all) | is this because it tries to run target pkg-config instead of build machine pkg-config, maybe? | 09:53:03 |
| Grimmauld (any/all) | * is this because it tries to run target pkg-config instead of build machine pkg-config, maybe? (building x86_64 from aarch64) | 09:55:58 |
| Grimmauld (any/all) | Ok, might be fixed | 10:03:55 |
|  tealquaternion joined the room. | 13:27:04 |
 emily | we set PKG_CONFIG etc. variables for cross | 13:44:17 |
| Grimmauld (any/all) | https://github.com/NixOS/nixpkgs/pull/400430/files#diff-def63f7c5751b7e899bdb59c3eea509acb9e4bac8a02a281ec588127c37b988cR16-R19
yeah, marcin-serwin saw the issue and i got it fixed | 13:48:38 |
| Grimmauld (any/all) | Now all of apparmor supports cross, LLVM, and Musl | 13:49:20 |
| emily | packaging looks clean, nice | 13:54:56 |
| Grimmauld (any/all) | Thank you, better than whatever this was: https://github.com/NixOS/nixpkgs/blob/nixos-unstable/pkgs/os-specific/linux/apparmor/default.nix
But i do expect some review to still come along and request some change because of some oversight. Building my system against this is not viable as it rebuilds systemd and dbus | 14:05:41 |
| emily | there comes a time in every Nixpkgs contributor's life where they must purchase a Big Computer | 14:06:20 |
|  9lore joined the room. | 15:18:56 |
 rosssmyth | I have a job that runs my builds against latest unstable each weekend, and this weekend the bump from GCC 14.2.1 20241116 to GCC 14.2.1 20250322 has changed something with relocations as I now get a bunch of dangerous relocation: unsupported relocation. Mainly because relocations are not supported on the target I used (arm-embedded, armv6-m), but I pass in hardeningDisable = [ "all" ];. I expected something to break due to the bump based upon that issue about GCC breakage, but not this. Anyone have any ideas what may have changed? | 16:42:03 |
| Tristan Ross | In reply to @rosssmyth:matrix.org
I have a job that runs my builds against latest unstable each weekend, and this weekend the bump from GCC 14.2.1 20241116 to GCC 14.2.1 20250322 has changed something with relocations as I now get a bunch of dangerous relocation: unsupported relocation. Mainly because relocations are not supported on the target I used (arm-embedded, armv6-m), but I pass in hardeningDisable = [ "all" ];. I expected something to break due to the bump based upon that issue about GCC breakage, but not this. Anyone have any ideas what may have changed? What derivation broke? | 17:17:44 |
| Tristan Ross | Do you have an easy reproducer? | 17:17:58 |
