| 7 Oct 2025 |
 Grimmauld (any/all) | * tbh if asan+pie causes trouble depending on build kernel, i'd vote to just skip that check | 15:42:20 |
 Willi Butz | huh. I can't on my local workstation. the above happened on a hetzner ax162
so maybe ignore for now | 15:43:00 |
 Vladimír Čunát | Depends on your kernel perhaps. | 15:43:17 |
| Willi Butz | hetzner 6.12, local 6.16 | 15:44:03 |
| Vladimír Čunát | Doesn't seem suspicious. | 15:44:23 |
| Vladimír Čunát | Hydra's builders use 6.12, BTW. | 15:45:13 |
| Vladimír Čunát | (stable NixOS default generally) | 15:45:23 |
| Grimmauld (any/all) | i am on 6.12 here | 15:45:23 |
| Vladimír Čunát | I'm trying with 6.6 now. | 15:46:12 |
| Willi Butz | ah, so the builder runs some "hardened" variant, not sure if its just linuxPackages_6_12_hardened | 15:49:36 |
| Vladimír Čunát | That might be it. | 15:50:16 |
| Vladimír Čunát | This kind of layout changes feel potentially useful for hardening. | 15:50:42 |
| Vladimír Čunát |
Shadow memory range interleaves with an existing memory mapping.
| 15:51:08 |
| Vladimír Čunát | Still, if the test is unreliable like this, I'm perfectly fine with skipping it. | 15:51:57 |
| Vladimír Čunát | It's the kind of thing that might make sense to keep upstream, as e.g. their CI will be known not to suffer from the hardening. But not suitable as a default in a distro. | 15:53:00 |
| Willi Butz | just saw that it is part of the logs 🤦
https://gist.github.com/WilliButz/afd8ca68c45c5e6a96216f6b92658d23#file-build-log-L1012 | 15:53:59 |
 dramforever | this is like a test-only problem right | 15:54:06 |
| Vladimír Čunát | It certainly looks that way to me. | 15:54:32 |
| dramforever | does it mean it's building two copies of everything just to test | 15:56:46 |
| Vladimír Čunát | 🤔 right, it shouldn't have ASAN in the default build. | 15:57:41 |
| Vladimír Čunát | * 🤔 right, it shouldn't have ASAN in the default build. (though I don't know audit in particular really) | 15:58:11 |
| Grimmauld (any/all) | i believe it builds specific test binaries that are built with asan | 15:59:56 |
| Vladimír Čunát | I don't reproduce the issue on normal 6.6 kernel. | 16:01:08 |
| Grimmauld (any/all) | hydra doesn't run into this issue, but we should probably still skip the test for the sake of reproducibility | 16:02:11 |
| Grimmauld (any/all) | give me a few minutes until the dependencies are built on my machine, then i'll do the skip and open a PR | 16:04:08 |
| Grimmauld (any/all) | i'll target -next i guess, we don't have anything built. But this should be merged soon-ish, audit rebuilds systemd and pam and dbus and stuff iirc | 16:04:41 |
| Grimmauld (any/all) | or should i target not-next and call hydra not running into it good enough? | 16:05:10 |
| Vladimír Čunát | Yes, Hydra is about to start building stuff in minutes. | 16:05:16 |
| Vladimír Čunát | Either is fine, I'd say. | 16:05:32 |
| Vladimír Čunát | If staging-next, I'd prefer to merged that within one hour. | 16:06:18 |
