Nix + dotnet | 126 Members | |
| 24 Servers |
| Sender | Message | Time |
|---|---|---|
| 2 May 2023 | ||
| builds successfully on nixos and opensuse wsl | 07:12:23 | |
| but my nixos installation is still on 22.11 | 07:14:08 | |
| System rollback did not fix the issue unfortunately, I'm also getting an error building the package on my laptop as well. | 07:27:56 | |
| * System rollback did not fix the issue unfortunately, I'm also getting the same error building the package on my laptop as well. | 07:28:21 | |
| * System rollback did not fix the issue unfortunately, I'm also getting the same error building the package on my laptop as well. Doesn't have the exact same config, but does use the same base as usual. | 07:29:52 | |
| testing on my servers, I do get some interesting behavior | 07:48:20 | |
| I have two oracle vms rn, one I've updated to my latest flake, the other is running an older version with an older version of nixpkgs | 07:48:50 | |
| It builds properly on the one I did not update and is running an older version, but fails on the one I just updated.
| 07:49:23 | |
| so it seems it was caused by a system update | 07:49:39 | |
| can reproduce now: $ nix build github:huantianad/nixos-config#tjaplayer3-f error: builder for '/nix/store/rscfr5ivym33v9w1n7p20zbarn5cif77-tjaplayer3-f-1.8.2.1-dependencies-source.drv' failed with exit code 139; last 1 log lines: > /nix/store/a4v3az5zmqfc2x3gyjgds5azgdhf9b54-stdenv-linux/setup: line 1560: 7 Segmentation fault (core dumped) nuget init "/nix/store/gv7vh34zghgnmla0qjgfgzxmvjxb6k2d-tjaplayer3-f-1.8.2.1-nuget-deps" "$out/lib" For full logs, run 'nix log /nix/store/rscfr5ivym33v9w1n7p20zbarn5cif77-tjaplayer3-f-1.8.2.1-dependencies-source.drv'. nixos-unstable d6b863fd9b7bb962e6f9fdf292419a775e772891 worked, 1a411f23ba299db155a5b45d5e145b85a7aafc42 is broken. will try to bisect this | 17:14:57 | |
| * can reproduce now:
nixos-unstable d6b863fd9b7bb962e6f9fdf292419a775e772891 worked, 1a411f23ba299db155a5b45d5e145b85a7aafc42 is broken. will try to bisect this | 17:15:18 | |
| Redacted or Malformed Event | 18:02:23 | |
| 18:26:34 | |
| ugh mkNugetSource is using mono to run nuget | 18:44:46 | |
| possibly related: https://lore.kernel.org/all/32f156ba80010fd97dbaf0a0cdfc84366608624d.camel@intel.com/ | 18:46:07 | |
| * possibly related: https://lore.kernel.org/all/cb8dc31a-fef2-1d09-f133-e9f7b9f9e77a@sony.com/ | 18:47:36 | |
| I suppose kernel regression would make sense for this issue, wonder if anyone has reported this for mono as well | 19:54:45 | |
In reply to @raphi:tapesoftware.netI think Nuget itself is packaged with mono, as MS only seems to release a .exe for it? | 19:55:30 | |
the bug seems to only happen if the stars align. if i run nuget init outside of the nix sandbox it succeeds. i can also run a highly nontrivial mono application just fine. i can also run a tiny c program that calls mmap(... MAP_32BIT) inside of the nix sandbox. | 19:56:53 | |
yeah i was looking into it. the closest thing the .net core CLI has is dotnet nuget push which doesn't have the same semantics | 19:57:15 | |
| Wish there was a good way to git bisect through kernel versions to check if it is a regression lol | 20:04:10 | |
| * Wish there was a good way to git bisect through kernel versions to check if it is a kernel regression lol | 20:05:40 | |
In reply to @raphi:tapesoftware.netMan it is so weird that the nuget CLI requires mono on linux still... I suppose ms expects people to mostly use the .net core cli? | 20:19:09 | |
| 3 May 2023 | ||
| 14:57:42 | ||
| 16:58:40 | ||
| o/ did anyone figure out whether that kernel bug is causing the nuget segfault? i asked a couple of friends to try building xivlauncher and it works for them on kernel 6.1.23, but fails for me on 6.3.0. (fwiw, the reproducer from the mailing list doesn't reproduce for me, though I didn't try running it in the nix sandbox (no idea how to do that)) | 17:07:22 | |
| 100% confident that it's MAP_32BIT related no evidence that it's this bug in particular | 17:35:49 | |
| 23:56:08 | ||
| 4 May 2023 | ||
| It being a kernel regression makes sense though, with my system rollback not fixing it. I would’ve have to gone back 2 system updates for the rollback to fix it, if it is a kernel regression | 00:34:21 | |
| https://github.com/NixOS/nixpkgs/pull/229851 | 08:41:03 | |