| 18 Apr 2024 |
 adamcstephens | ─❯ incus config set mighty-anteater limits.cpu=2
Error: Key "limits.cpu" cannot be updated when VM is running
| 13:51:43 |
| adamcstephens | aarch64.... | 13:51:50 |
| adamcstephens | maybe i'll leave the x86 condition in the udev rule | 13:51:57 |
| adamcstephens | was trying to dig into aarch64 nested virtualization because i'd love to be able to run VM tests on this architecture, but i don't even think qemu supports this for aarch64 | 14:16:23 |
 hexa | huh | 14:16:42 |
| hexa | I have no aarch64 hosts, but works on x86_64-linux 🙂 | 14:16:53 |
| adamcstephens | yeah x86 is good | 14:17:03 |
| adamcstephens | there's this patchset... https://patchew.org/QEMU/20230227163718.62003-1-miguel.luis@oracle.com/ | 14:17:16 |
| adamcstephens | for an older qemu | 14:17:29 |
| adamcstephens | [root@nixos:/etc/udev/rules.d]# dmesg | grep -i el1
[ 0.006279] CPU features: detected: 32-bit EL1 Support
[ 0.006368] CPU: All CPU(s) started at EL1
| 14:17:44 |
| adamcstephens | virtualization requires EL2... | 14:18:00 |
| adamcstephens | (that's inside a VM) | 14:18:12 |
| adamcstephens | meanwhile on the host (an rpi4)
─❯ sudo dmesg | rg el2
[ 0.031475] CPU: All CPU(s) started at EL2
| 14:18:47 |
| adamcstephens | based on the qemu commit log it seems like people are working on this stuff, so maybe one day | 14:25:53 |
| adamcstephens | i was just trying to see if it's something we have disabled | 14:26:56 |
| hexa | cool | 14:27:08 |
| adamcstephens | https://github.com/NixOS/nixpkgs/pull/304919 | 14:28:14 |
| hexa | the x86 check is redundant, no? 🙂 | 14:32:50 |
| hexa | and please no pkgs.stdenv indirection | 14:33:06 |
| adamcstephens | just figured no need to put the file on all platforms. i can remove the condition, but how would you condition this instead? | 14:34:40 |
| hexa | either you deploy the rule only when stdenv.hostPlaform.isx86_64 | 14:35:05 |
| hexa | and then the arch check in the rule is redundant | 14:35:19 |
| hexa | or you deploy it always and keep the udev arch checkl | 14:35:28 |
| hexa | * or you deploy it always and keep the udev arch check | 14:35:30 |
| hexa | imo. | 14:35:31 |
| adamcstephens | that's fine. i guess i'm asking what the alternative to pkgs.stdenv.hostPlatform.isx86 is? | 14:35:58 |
| adamcstephens | e.g. what would i reference instead of that boolean? | 14:38:53 |
| hexa | hm? | 14:40:16 |
| hexa | isx86_64? | 14:40:22 |
| adamcstephens | you said "and please no pkgs.stdenv indirection" so i'm just trying to understand how you'd write the condition instead :) | 14:40:38 |
