diff options
| author | Sam James <sam@gentoo.org> | 2025-12-18 10:18:07 +0000 |
|---|---|---|
| committer | Sam James <sam@gentoo.org> | 2025-12-18 10:21:20 +0000 |
| commit | 414f21a6b310ec7d4e0885dd7ac5e379cd49dfe2 (patch) | |
| tree | 3b782c032f8fc569d2b3f55f894fee146c23f0ec /dev-python/google-api-python-client/google-api-python-client-2.185.0.ebuild | |
| parent | cbb0901e1530860c7d37193dc8ccdf21e23405b2 (diff) | |
| download | gentoo-414f21a6b310ec7d4e0885dd7ac5e379cd49dfe2.tar.gz gentoo-414f21a6b310ec7d4e0885dd7ac5e379cd49dfe2.tar.bz2 gentoo-414f21a6b310ec7d4e0885dd7ac5e379cd49dfe2.zip | |
sys-libs/glibc: ignore CFLAGS+LDFLAGS in do_run_test for binpkgs
With binpkgs, CFLAGS and LDFLAGS will come from the builder environment
and may not be compatible with an old glibc on the system before a new
one is merged.
In the case reported to me by Kerin, an old glibc was installed w/o
DT_RELR support, and LDFLAGS on the binhost builder had -Wl,-z,pack-relative-relocs,
so the built test binary (which would run against the system's old glibc)
couldn't be executed, meaning the "syscall test" failed.
Set CFLAGS to just -O2 and blank LDFLAGS to avoid this.
It did also wrongly diagnose the failure because the function both
compiles + runs and the error message can't distinguish between those. If
we fail to build it, things are likely fine, there's only a problem if we
fail to run it. I haven't addressed that here.
Bug: https://bugs.gentoo.org/324685
Signed-off-by: Sam James <sam@gentoo.org>
Diffstat (limited to 'dev-python/google-api-python-client/google-api-python-client-2.185.0.ebuild')
0 files changed, 0 insertions, 0 deletions