Bug 876070 – sys-libs/uid_wrapper-1.2.9 fails tests (MUSL)

• Status: CONFIRMED
• Product: Gentoo Linux

[Description Agostino Sarubbo 2022-10-08 10:26:36 UTC]

https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/

Issue: sys-libs/uid_wrapper-1.2.9 fails tests (MUSL).
Discovered on: amd64 (internal ref: tinderbox_musl)

NOTE:
This machine uses MUSL libc

[Comment 1 Agostino Sarubbo 2022-10-08 10:26:39 UTC]

Created attachment 822811 [details]
build.log

build log and emerge --info

[Comment 2 Agostino Sarubbo 2022-10-08 10:26:40 UTC]

Created attachment 822814 [details]
1-LastTest.log

1-LastTest.log

[Comment 3 Agostino Sarubbo 2022-10-08 10:26:40 UTC]

Error(s) that match a know pattern:


	 15 - test_syscall (Failed)

[Comment 4 matoro 2023-11-23 21:52:59 UTC]

Somehow timezone-related, but the usual good old TZ=UTC has no effect.

[Comment 5 matoro 2023-11-23 22:21:07 UTC]

This seems to be somewhat unpredictable, on amd64 it is MUSL-specific, but it also appears under other platforms with glibc.

For glibc, unaffected platforms:  arm64, ppc64, loong
For glibc, affected platforms:  ia64, sparc, alpha, mips, hppa, riscv

This note in the man page for gettimeofday(2) might be relevant:

VERSIONS
   C library/kernel differences
       On some architectures, an implementation of gettimeofday() is provided in the vdso(7).
 
       The kernel accepts NULL for both tv and tz.  The timezone argument is ignored by glibc and musl, and not passed to/from the kernel.  Android’s bionic passes the timezone argument to/from the kernel, but Android does not update the kernel timezone based on the device timezone in Settings, so the kernel’s timezone is typically UTC.


I tried looking for a pattern matching this in vdso(7), but did not find one.  Also, given that the issue is present with MUSL under amd64, it's likely not a kernel issue.