Hi Richard,
I thought it was OK because systemd requires usrmerge, but I've sent a new
version of the patch using "env sleep" instead of "/usr/bin/sleep" in order
to be safe. I hope this also fixes the test failure which was happening on
CI.
One reason this could be failing in CI is that this functionality currently
requires user-namespaces enabled in the kernel (e.g. core-image-sato does
have CONFIG_USER_NS, thus it works with this image). This is not really by
design, and I'm working on changing this in the upstream systemd:
https://github.com/systemd/systemd/pull/33043 . I can then backport this on
scarthgap if the change is merged upstream.

Étienne

On Wed, May 29, 2024 at 4:54 PM Richard Purdie <
[email protected]> wrote:

> On Wed, 2024-05-29 at 12:08 +0200, Etienne Cordonnier via
> lists.openembedded.org wrote:
> > From: Etienne Cordonnier <[email protected]>
> >
> > The builtin command "sleep" was getting executed instead of the busybox
> binary.
> > Hence "pidof sleep" was empty and the test was failing.
> >
> > Signed-off-by: Etienne Cordonnier <[email protected]>
> > ---
> >  meta/lib/oeqa/runtime/cases/systemd.py | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/meta/lib/oeqa/runtime/cases/systemd.py
> b/meta/lib/oeqa/runtime/cases/systemd.py
> > index 5481e1d840..a2229848e7 100644
> > --- a/meta/lib/oeqa/runtime/cases/systemd.py
> > +++ b/meta/lib/oeqa/runtime/cases/systemd.py
> > @@ -145,7 +145,7 @@ class SystemdServiceTests(SystemdTest):
> >          Verify that call-stacks generated by systemd-coredump contain
> symbolicated call-stacks,
> >          extracted from the minidebuginfo metadata (.gnu_debugdata elf
> section).
> >          """
> > -        t_thread = threading.Thread(target=self.target.run,
> args=("ulimit -c unlimited && sleep 1000",))
> > +        t_thread = threading.Thread(target=self.target.run,
> args=("ulimit -c unlimited && /usr/bin/sleep 1000",))
> >          t_thread.start()
> >          time.sleep(1)
>
> I get a bit nervous about hardcoding paths like "/usr/bin",
> particularly in the context of some distros liking to move things
> around. We may want to put a shutil.which("sleep") call in instead (and
> an import shutil if needed)?
>
> Cheers,
>
> Richard
>
>
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#200026): 
https://lists.openembedded.org/g/openembedded-core/message/200026
Mute This Topic: https://lists.openembedded.org/mt/106365956/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to