Sender | Message | Time |
---|---|---|
3 Oct 2024 | ||
raitobezarius | So it's the FAT32 of Apple | 23:07:08 |
raitobezarius | The only interesting question is: is it journaled? | 23:07:30 |
emily | it might actually be a real APFS tree, not sure | 23:08:20 |
ElvishJerricco | honestly if I needed to implement the simplest possible file system for firmware to boot off of I'd probably just make it like ZFS's ZIL | 23:08:22 |
emily | all iBoot1 does is to read a boot policy and chainload into the appropriate iBoot2 basically | 23:09:16 |
ElvishJerricco | which would slot nicely into a CoW container format like APFS / ZFS | 23:09:21 |
emily | I know they use C with added memory safety sprinkled in for iBoot implementation but I don't know if they just have a whole APFS in there or what | 23:09:31 |
4 Oct 2024 | ||
mjm | ElvishJerricco: bcachefs generator doesn't seem to agree with my impermanence setup. all i've done is apply the patch of the PR and disable clevis. my bcachefs volume mounts at /persist, and i get prompted to unlock /sysroot/persist, but then it's blocked on 3 of the bind mounts, which just seem to sit there doing nothing until they timeout (i think it's /nix , /var/log and /var/lib/nixos that it's waiting on) | 00:48:50 |
ElvishJerricco | hm, interesting | 00:49:38 |
ElvishJerricco | I'm struggling to think of anything that would cause that... | 00:50:18 |
ElvishJerricco | mjm: Can you share your file system nix config? | 00:51:57 |
ElvishJerricco | or better yet, the initrd's fstab file | 00:52:53 |
ElvishJerricco | it'd be the config.boot.initrd.systemd.managerEnvironment.SYSTEMD_SYSROOT_FSTAB I think | 00:53:17 |
mjm | well first, here's my normal fstab:
| 00:53:46 |
ElvishJerricco | oh, yea, that ought to include the same stuff too I guess | 00:54:24 |
mjm |
| 00:54:34 |
mjm | that's the initrd one | 00:54:39 |
ElvishJerricco | mjm: ok, yea, there's some serious bugs here that don't make sense to me | 01:00:53 |
ElvishJerricco | thanks for the fstab | 01:00:59 |
ElvishJerricco | but suffice to say, you cannot use it right now :P | 01:01:12 |
mjm | np, let me know anything else i can provide that's useful | 01:01:13 |
ElvishJerricco | no this is plenty | 01:01:25 |
mjm | cool, i can test again whenever you have something | 01:02:16 |
ElvishJerricco | mjm: one issue I do understand: I had it work on auto mounts because I figured the condition I added to skip the unit if the device isn't a bcachefs device would be enough. But it's flipping out on your bind mounts because of that :P | 01:02:50 |
ElvishJerricco | so I'll make it only work on explicit bcachefs type | 01:03:03 |
ElvishJerricco | actually that might be the only issue | 01:03:12 |
ElvishJerricco | that might explain why it's creating non-sysroot mount units for those things too | 01:03:24 |
ElvishJerricco | I'll have to test to see and then think on why | 01:03:39 |
mjm | interesting | 01:03:55 |
ElvishJerricco | nope, that doesn't explain it. It's also trying to create the deps for your persist mount without the sysroot prefix, in addition to the sysroot one | 01:05:40 |