#asahi-alt · 2024-05-13 15:35
floklihurr durr#asahi-alt · 2024-05-13 15:08
LarstiQaaand webkitgtk, also `/build/webkitgtk-2.44.1/build/WTF/Headers/wtf/Vector.h:1080:26: internal compiler error: Segmentation fault`, so not completely deterministic which of the builds fails first, but pretty reproducible just by `nixos-rebuild --show-trace switch |& nom` when I have zig, chromium and webkitgtk needing rebuilds#asahi-alt · 2024-05-13 14:57
LarstiQtrying again it's zig that hits `/nix/store/i77y0adr2a79fd1ipswzmrl43a0zbh5f-clang-16.0.6-dev/include/clang/AST/ExprCXX.h:2414:3: internal compiler error: in tree_node_structure_for_code, at tree.cc:523` and now firefox is no longer responding to input#asahi-alt · 2024-05-13 14:41
LarstiQhuh, build failed due to: error: ca hash mismatch importing path '/nix/store/3z420xrnjjv5bvxlyd7kv6gsj26jq6l5-chromium-124.0.6367.118.tar.zstd';#asahi-alt · 2024-05-13 14:04
LarstiQI'll try switching from replace to overlay, that causes quite a bit of load and maybe more chance of a haunting#asahi-alt · 2024-05-13 13:57
flokli🤷#asahi-alt · 2024-05-13 13:57
flokliok#asahi-alt · 2024-05-13 13:57
LarstiQyeah#asahi-alt · 2024-05-13 13:57
flokliLarstiQ: but only some are haunted the way these are?#asahi-alt · 2024-05-13 13:56
LarstiQflokli: that seems to happen in all my mesa-24.1.0 derivations#asahi-alt · 2024-05-13 13:51
flokliis this normal and supposed to happen? (from the build log I just attached)#asahi-alt · 2024-05-13 13:51
floklipatching script interpreter paths in /nix/store/sns2xgw85787sghv25b95fmcckxf0601-mesa-24.1.0/bin/mesa-overlay-control.py#asahi-alt · 2024-05-13 13:51
flokliTry 'mv --help' for more information.#asahi-alt · 2024-05-13 13:51
floklimv: missing file operand#asahi-alt · 2024-05-13 13:51
floklistripping (with command strip and flags -S -p) in /nix/store/4l0i08n1v3rwhmznwl821binph1k2nvn-mesa-24.1.0-debug/lib#asahi-alt · 2024-05-13 13:45
flokliLarstiQ: tpw_rules: I uploaded my cursed path to the issue#asahi-alt · 2024-05-13 13:45
LarstiQfwiw, I don't have any remote builders. The bad generations got gced already though so can't confirm with the entropy trick#asahi-alt · 2024-05-13 13:43
tpw_rulesany gentoo folks or asahi devs seen something like this? i suspect nixos people have compiled mesa the most #asahi-alt · 2024-05-13 13:42
LarstiQoho, I've had the same thing as flokli is having, I thought some change finally fixed it but maybe it's ghosts#asahi-alt · 2024-05-13 13:41
tpw_rulesbah, i've seen that before, posted a bit more in that issue#asahi-alt · 2024-05-13 13:41
floklicopied the comment over#asahi-alt · 2024-05-13 13:33
yukaflokli: ^#asahi-alt · 2024-05-13 13:32
yukahttps://github.com/tpwrules/nixos-apple-silicon/issues/199#asahi-alt · 2024-05-13 13:23
yukaokay!#asahi-alt · 2024-05-13 13:23
flokliI'm happy to drop a comment into the issue#asahi-alt · 2024-05-13 13:23
yukafor completeness: which nix version are you running? (locally and on the remote build machine)#asahi-alt · 2024-05-13 13:22
yukaI suspect that it happens when mesa is built under load (while simultaneously building a kernel)#asahi-alt · 2024-05-13 13:21
yukaflokli: should work. or add an overlay that changes the derivation hash of mesa somehow (for example : "final: prev: { mesa-asahi-edge = prev.mesa-asahi-edge.overrideAttrs (oldAttrs: { src = lib.cleanSource oldAttrs.src; }); }"#asahi-alt · 2024-05-13 13:20
flokliI however don't know if I built it on the M1 box or another aarch64 builder#asahi-alt · 2024-05-13 13:20
floklithis is cursed as hell#asahi-alt · 2024-05-13 13:19
flokliand I guess I can reboot into an older generation without that, delete the store path and then build again and it might fix itself?#asahi-alt · 2024-05-13 13:19
yukahttps://git.lix.systems/lix-project/lix/issues/248 <-- this was previously the issue#asahi-alt · 2024-05-13 13:19
yukawell, I guess i'll transfer the issue from lix to nixos-apple-silicon now#asahi-alt · 2024-05-13 13:18
flokliis there an issue tracking this somehow?#asahi-alt · 2024-05-13 13:18
yukawe don't really understand what's going on there, but you are at least the third person to run into this#asahi-alt · 2024-05-13 13:18
yukaokay#asahi-alt · 2024-05-13 13:17
flokliyes#asahi-alt · 2024-05-13 13:17
yuka(add a "echo $i" in the loop to find out which one it is)#asahi-alt · 2024-05-13 13:16
yukalet me guess, the one with 2.5M is the one referenced by your current system?#asahi-alt · 2024-05-13 13:16
yukayeah that looks suspicious#asahi-alt · 2024-05-13 13:16
flokli2513936#asahi-alt · 2024-05-13 13:16
flokli3176884#asahi-alt · 2024-05-13 13:16
flokli3039308#asahi-alt · 2024-05-13 13:16
flokli3175560#asahi-alt · 2024-05-13 13:16
floklifor i in /nix/store/*-mesa-24.1.0-drivers/lib/dri/apple_dri.so ; do cat $i | xz -9 | wc -c ; done #asahi-alt · 2024-05-13 13:15
yukaif this is the case it would be tremendously useful because it means this issue is not lix specific#asahi-alt · 2024-05-13 13:15
yukaif any of the paths has a significantly lower entropy than the others, your store is haunted#asahi-alt · 2024-05-13 13:15
yukaquick check: "for i in /nix/store/*-mesa-24.1.0-drivers/lib/dri/apple_dri.so ; do cat $i | xz -9 | wc -c ; done"#asahi-alt · 2024-05-13 13:14
flokliI use this replaceRuntimeDependencies thing#asahi-alt · 2024-05-13 13:14
yukagood to know!