| 10 Mar 2025 |
 @elvishjerricco:matrix.org | new development on the issue: The 256.7 version of systemd-boot that originally introduced the 4GB boundary bug successfully boots a 24.11 generation but not a nixos-unstable generation | 23:39:00 |
| @elvishjerricco:matrix.org | with the same failure as the 257.2 systemd-boot | 23:39:26 |
| @elvishjerricco:matrix.org | so given that the breakage definitely happens between 256.6 and 256.7, just like the 4GB issue, it's certainly related, but probably a little different since it doesn't happen on a 24.11 generation | 23:40:34 |
 eyJhb | In reply to @elvishjerricco:matrix.org
I don't suppose we keep an easily parsed list of commits the channel was bumped to, do we? I'm planning on starting such a list again. I'm so tired of it not being easily available | 23:56:54 |
| 11 Mar 2025 |
 raitobezarius | In reply to @elvishjerricco:matrix.org
so given that the breakage definitely happens between 256.6 and 256.7, just like the 4GB issue, it's certainly related, but probably a little different since it doesn't happen on a 24.11 generation exciting | 01:16:51 |
| * raitobezarius put one cent on d7ae71b9c187133db028a52cd05145f6c1ef8a50 and one cent on d64193a2a652b15db9cb9ed10c6b77a17ca46cd2 | 01:20:06 |
| @elvishjerricco:matrix.org | K900: I guess I should ask you if anything changed since 24.11 in how the kernels look to systemd-boot that could screw it up? | 02:55:41 |
| @elvishjerricco:matrix.org | because the same systemd-boot is only broken for kernels from unstable, not from 24.11 | 02:56:17 |
 K900 | In reply to @elvishjerricco:matrix.org
K900: I guess I should ask you if anything changed since 24.11 in how the kernels look to systemd-boot that could screw it up? Don't think so? | 05:41:08 |
| @elvishjerricco:matrix.org | yea I didn't think so... I wondered about file size too but the kernel and initrd are both only barely larger in unstable | 05:41:36 |
| @elvishjerricco:matrix.org | I mean certainly there's something different about these that causes the issue, as they are indeed the parameters that triggers the broken boot loader behavior | 05:42:13 |
| @elvishjerricco:matrix.org | just can't think of anything | 05:42:21 |
| @elvishjerricco:matrix.org | I don't know enough about how PEs work but it might be something as stupid as the virtual addresses used by the kernel sections? (Is that the right terminology for a kernel PE?) | 05:43:45 |
| raitobezarius | In reply to @elvishjerricco:matrix.org
I don't know enough about how PEs work but it might be something as stupid as the virtual addresses used by the kernel sections? (Is that the right terminology for a kernel PE?) bingrep might help | 11:39:13 |
 Arian | Didn't systemd-boot have some weird changes in stub | 15:15:20 |
| Arian | Not that this should affect a normal nixos install though.. | 15:16:45 |
| @elvishjerricco:matrix.org | jesus the guy has a 350M initrd | 17:59:00 |
| @elvishjerricco:matrix.org | https://discourse.nixos.org/t/systemd-boot-failing-to-boot-a-nixos-warbler-system-due-to-an-efi-assertion-failure/61450/13?u=elvishjerricco | 17:59:01 |
| @elvishjerricco:matrix.org | ok that's a bug of its own | 17:59:10 |
| @elvishjerricco:matrix.org | we shouldn't make initrds that big unless user knows what and why | 17:59:26 |
| @elvishjerricco:matrix.org | but also systemd-boot shouldn't die because of it | 17:59:55 |
 aloisw | Well 350M not fitting into memory after decompression sounds plausible. | 18:00:44 |
| @elvishjerricco:matrix.org | plausible for the 4G boundary, yes | 18:01:06 |
| @elvishjerricco:matrix.org | I wouldn't say likely | 18:01:11 |
| @elvishjerricco:matrix.org | it's like right on the edge | 18:01:14 |
| @elvishjerricco:matrix.org | when do you get a more than 10x compression ratio? | 18:01:50 |
| aloisw | It has to be contiguous memory below 4G. | 18:01:59 |
| @elvishjerricco:matrix.org | oh so it might be putting the uncompressed initrd near-to-but-not-right-at the boundary? | 18:02:30 |
| @elvishjerricco:matrix.org | close enough that it pretty much always fits, unless you have a massive initrd? | 18:02:47 |
| aloisw | In reply to @elvishjerricco:matrix.org
when do you get a more than 10x compression ratio? I have no idea what got put in their initrd, it's so out of the normal that it's plausible that it either compresses very well or not at all. (Really it should be figured out what got into there and the bug fixed.) | 18:03:36 |
