Re: [poky] Error from package.bbclass?

Paul Barker <[email protected]> Wed, 07 Jan 2026 18:36:55 +0000
Newsgroups org.yoctoproject.lists.poky,org.yoctoproject.lists.yocto
Message-ID <[email protected]>
On Wed, 2026-01-07 at 09:50 -0800, shawn.adams via
lists.yoctoproject.org wrote:
> I'm getting an error when building core-image-minimal for qemuarm64 using the scarthgap branch on a CentOS 9 VM.
> 
> From what I can tell, the error occurs when perform_packagecopy() attempts to copy the linux-libc-headers source files into tmp/work/cortexa57-poky-linux/linux-libc-headers/6.6/package. Line 355 in poky/meta/classes-global/package.bbclass runs two tar commands from Python's subprocess.check_output() which throw an exception. The text of the exception is
> 
> > 
> > Exception: subprocess.CalledProcessError: Command 'tar
> > --exclude=./sysroot-only -cf - -C
> > /builds/build/qemuarm64/tmp/work/cortexa57-poky-linux/linux-libc-headers/6.6/image
> > -p -S . | tar -xf - -C
> > /builds/build/qemuarm64/tmp/work/cortexa57-poky-linux/linux-libc-headers/6.6/package'
> > returned non-zero exit status 2.
> > 
> 
> The second tar command is what fails (the one extracting into the package directory). I have tar 1.34 installed on my VM and have verified that Bitbake uses the same version when it runs.
> 
> When I run the tar command manually from a terminal it works fine, so the problem seems to be unique to the Bitbake run-time environment. I have also changed perform_packagecopy() a couple different ways to see if I can get it to work:
> * I changed it to use Python's tarfile module to do what the tar command is doing. This avoids running the tar command in a shell via subprocess.check_output() and seems to work.
> * I added a shell function to package.bbclass which runs the same tar command and call it via bb.build.exec_func() instead of running the tar command in subprocess.check_output(). This produces the same error.
> I've attached the full Bitbake log file in case that helps.
> 
> Does anyone have ideas about what could be causing this problem? Could it be a problem with software versions installed on my VM?

([email protected] as that's a more relevant list)

This looks like an issue that we're seeing on our autobuilder with the
CentOS 9 worker [1].

It also seems similar to RHEL issue "tar --xattrs -xf failure" [2], but
that may not be an exact match as I'm unsure if we're using xattrs here
without refreshing my memory on the details.

[1]: https://bugzilla.yoctoproject.org/show_bug.cgi?id=16117
[2]: https://issues.redhat.com/browse/RHEL-136277

Thanks,

-- 
Paul Barker
signature.asc (application/pgp-signature, 252 B)
-----BEGIN PGP SIGNATURE-----

iIcEABYKAC8WIQSzjPXf5Y1BDWhU2iCrY1Tsnbr0bgUCaV6nxxEccGF1bEBwYmFy
a2VyLmRldgAKCRCrY1Tsnbr0btG2AQDGzWMxwm1OHM6Q+FW+RvjuhCtG7pI91aZI
EmpR2+I/zAEA1tA0lJ/0Do8HU1wTcJvu248+09xZ2grwqgaoFePqdws=
=dvHX
-----END PGP SIGNATURE-----