On 1/18/22 15:08, Andrea Bolognani wrote:
On Tue, Jan 18, 2022 at 02:13:23PM +0100, Michal Prívozník wrote:
On 1/18/22 14:05, Michal Prívozník wrote:
On 1/18/22 13:57, Andrea Bolognani wrote:
Can you please try a version of this series with your original dnsmasq mocking approach in CI and see whether ASAN is happy with it? If so, we can just go ahead with that one.
Will do. Although, since virCommandRun() wouldn't actually execute anything I don't expect ASAN to raise any issues. Meanwhile, I'm testing the approach I've outlined:
https://gitlab.com/MichalPrivoznik/libvirt/-/pipelines/450382098
Let's see how it runs.
Aaand I have the results:
failed to create the fake capabilities: internal error: Child process (LC_ALL=C LD_PRELOAD=/builds/MichalPrivoznik/libvirt/build/tests/libvirdnsmasqmock.so PATH=/builds/MichalPrivoznik/libvirt/build/tests:/usr/libexec/ccache-wrappers:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin HOME=/bad-test-used-env-home /usr/local/sbin/dnsmasqmock.py --version) unexpected exit status 127: /usr/bin/env: symbol lookup error: /builds/MichalPrivoznik/libvirt/build/tests/../src/libvirt.so.0: undefined symbol: __asan_option_detect_stack_use_after_return
Why does dnsmasqmock.py try to link with libvirt.so.0 or why there's a missing symbol is beyond me. So let me just stick with what I suggested initially.
Yeah, that sounds good. Can you please make sure the result passes CI and post it to the list? IIRC you've shuffled patches around in the meantime, so we should do one last quick sanity check before pushing.
Alright, although the only change I did really was in the patch 10/13 which I replaced with the corresponding patch from my earlier version. But anyway, let's respin another version (hopefully the last). The pipeline's green: https://gitlab.com/MichalPrivoznik/libvirt/-/pipelines/450418184 Michal