Bug#1103303: src:glibc FTBFS with test failure in elf/tst-audit10 on system without AVX

[Jochen Sprickerhof]

Source: glibc
Version: 2.41-6
Severity: normal
Tags: ftbfs
X-Debbugs-Cc: debian-amd64 at lists.debian.org, reproducible-bugs at lists.alioth.debian.org
User: debian-amd64 at lists.debian.org
Usertags: amd64
User: reproducible-builds at lists.alioth.debian.org
Usertags: environment

from #debian-devel:

2025-04-16 10:26:45     jochensp        aurel32: hi, could you help me make glibc build on reproduce.d.n? 2.41-7 fails for me with FAIL: elf/tst-audit10 (whereas it is UNSUPPORTED: elf/tst-audit10 on the buildd). I believe the relevant log is: https://paste.debian.net/1369722/
2025-04-16 10:30:55     jochensp        (note that this is only on amd64, other archs are fine)
2025-04-16 10:35:23     aurel32 jochensp: from a quick look, this test requires a cpu with AVX512, otherwise it is marked as unsupported
2025-04-16 10:35:35     aurel32 the question is if you are able to reproduce the issue with -6
2025-04-16 10:36:26     jochensp        aurel32: lscpu | grep avx returns nothing on the reproduce.d.n host
2025-04-16 10:36:42     jochensp        will run a test with -6
2025-04-16 10:39:27     aurel32 hmm that's strange, basic avx should be present on most modern cpus. What about /proc/cpuinfo?
2025-04-16 10:40:12     jochensp        nothing as well
2025-04-16 10:40:25     jochensp        it is a Intel(R) Xeon(R) CPU           X5690  @ 3.47GHz
2025-04-16 10:41:23     aurel32 ok, error code 132 is SIGILL, so it seems to wrongly detect AVX512 and then fails to execute AVX512 code
2025-04-16 10:41:47     aurel32 is it a VM?
2025-04-16 10:42:13     aurel32 and indeed that CPU doesn't have AVX
2025-04-16 10:42:32     jochensp        systemd-detect-virt returns none
2025-04-16 10:47:07     aurel32 jochensp, h01ger: which kernel does it run?
2025-04-16 10:47:20     jochensp        6.12.21-amd64
2025-04-16 10:50:00     jochensp        running ./build-tree/amd64-libc/elf/tst-audit10 inside the failed build chroot exits with 77 though
2025-04-16 11:51:53     jochensp        aurel32: -6 fails the same way. Do you want a bug report for that?
2025-04-16 11:52:58     aurel32 jochensp: yes, that said I am unlikely to be able to do anything with it if I can't reproduce it
2025-04-16 11:53:41     jochensp        do you want to have access to that system? (I guess Holger would be fine with that)
2025-04-16 11:54:49     jochensp        aurel32: can I quote the conversation here for the bug report?
2025-04-16 11:55:17     aurel32 yes

Cheers Jochen