Exotic Nix Targets | 320 Members | |
| 99 Servers |
| Sender | Message | Time |
|---|---|---|
| 10 Jul 2023 | ||
I disabled sandbox fallback in the overlay with https://github.com/trofi/nix-guix-gentoo/commit/88050fcf0f3a11a7a6763494eda580b707be1267 and added a README entry to diagnose it: https://github.com/trofi/nix-guix-gentoo/commit/8109131192dc340e314d9e41159b176bfb5b82c8 | 08:33:40 | |
What environment did you use to run nix? Real machine, a VM or some container? | 08:34:20 | |
In reply to @trofi:matrix.orgreal machine, no VMs, no containers | 08:56:22 | |
| also my /etc/nix/nix.conf already has `sandbox-fallback = false` | 08:58:48 | |
| * also my /etc/nix/nix.conf already has `sandbox-fallback = false` without issue | 08:59:01 | |
| * also my /etc/nix/nix.conf already has `sandbox-fallback = false` without issue. i dont know when i added that but it's been there | 08:59:53 | |
| 09:05:21 | ||
| 09:39:14 | ||
| Must be something else then (or a poisoned local cache when was built without a sandbox, before nix.conf tweaks) | 09:46:47 | |
| how coud i clean local cache to ttest that theory | 10:41:41 | |
| Good question. I would "uninstall" everything by cleaning all gc roots in /nix/var/nix/gcroots and run nix-xollect-garbage -d. | 10:56:09 | |
| * Good question. I would "uninstall" everything by cleaning all gc roots in /nix/var/nix/gcroots and run nix-collect-garbage -d. | 10:56:33 | |
| Alternatively you can `nix-build --check` all paths in the problematic closure. But it will require checking for false positives manually. | 11:00:46 | |
| i will do the first option | 11:25:01 | |
| 11 Jul 2023 | ||
In reply to @alex:tunstall.xyzI've gotten Linux 5.15 to compile but it doesn't boot, doesn't log to journalctl, and doesn't output to HDMI, so I've folded and ordered the necessary hardware to debug over serial. | 10:41:37 | |
| * I've gotten Linux 5.15 (VF2 dev branch) to compile but it doesn't boot, doesn't log to journalctl, and doesn't output to HDMI, so I've folded and ordered the necessary hardware to debug over serial. | 10:42:00 | |
In reply to @alex:tunstall.xyzI have a serial adapter, could you publish your patches so I can try them on the star64? | 17:43:14 | |
| 12 Jul 2023 | ||
In reply to @fgaz:matrix.org Sure: https://github.com/AlexandreTunstall/nixos-vf2/tree/linux-devel This isn't the exact code I'm building (my VF2 configuration files are a bit of a mess), but it should work when built natively or at the very least allow you to build the same kernel. | 04:34:43 | |
| My main machine (x86 computer) is dead so don't have enough resources to CXX but if I could get an image, I can check for the issue... | 06:58:27 | |
In reply to @thefossguy:matrix.orgI might be able to generate a NAR of the closure. | 07:15:53 | |
| This should work. https://github.com/AlexandreTunstall/nixos-vf2/releases/tag/broken-v1 If needed, I can spend extra time to build a disk image instead. | 07:39:40 | |
| By the way, if you don't already have NixOS on the VF2, you can get a bootable image from here: https://github.com/misuzu/nixos-vf2/releases (set boot mode to flash). | 07:45:19 | |
| Ah, this is tricky for me... I'll see what I can do | 10:20:08 | |
| I don't have a big enough SD Card that can handle more than one generation :( | 10:20:31 | |
In reply to @thefossguy:matrix.org 4GB should be enough, the closure is ~850MiB and misuzu's images are <2GiB. You can also download/inflate it onto tmpfs if you barely have enough space to import it into the Nix store. | 10:24:39 | |
| Ah okay | 10:46:37 | |
| Concerning the VF2 CPU frequency issue on the "upstream" branch: a Starfive employee has responded to say that it'll be fixed once the JH7110 PLL patchset is accepted upstream. If anyone doesn't want to wait for the vendor's fix, the issue is that the PLL0 clock runs at 1GHz instead of 1.5GHz. I'd write a patch for it myself, but I can't figure out what change needs to be made, so good luck to anyone willing to try. If you do find a fix, please share it in the GitHub issue. https://github.com/starfive-tech/VisionFive2/issues/55#issuecomment-1631996288 | 16:27:44 | |
| They're here: https://lore.kernel.org/all/20230613125852.211636-1-xingyu.wu@starfivetech.com/ | 17:01:16 | |
In reply to @alex:tunstall.xyz I see the extracted
| 17:05:39 | |
| Oh my bad | 17:06:45 | |