reprepro feeds bad data to unzstd with libarchive 3.6.2 on s390x

[Simon Chopin]

I was only able to reproduce this using a single-threaded s390x VM with 2GB of RAM.

See https://autopkgtest.ubuntu.com/results/autopkgtest-lunar/lunar/s390x/r/reprepro/20230223_161449_e244d@/log.gz

autopkgtest [16:14:36]: test import-hello.sh: [-----------------------
Get:1 http://ftpmaster.internal/ubuntu lunar/main s390x hello s390x 2.10-3 [26.1 kB]
Fetched 26.1 kB in 1s (50.4 kB/s)
zstd: /*stdout*\: Broken pipe
Exporting indices...

[Simon Chopin (schopin)]

After a long investigation, it turns out the issue is not related to libarchive 3.6.2, but rather to a race condition between reprepro and the external decompression tool it invokes.

See https://salsa.debian.org/debian/reprepro/-/merge_requests/6 for a fix, currently building in https://launchpad.net/~schopin/+archive/ubuntu/reprepro

[Launchpad Janitor (janitor)]

This bug was fixed in the package reprepro - 5.3.1-1ubuntu1

---------------
reprepro (5.3.1-1ubuntu1) lunar; urgency=medium

  * d/p/0001-uncompress-wait-until-the-child-as-exited-to-close-t.patch:
    Fix a race condition when using external decompressors (LP: #2008508)

 -- Simon Chopin <email address hidden> Tue, 28 Feb 2023 12:35:59 +0100

[Christoph Fiehe (cfiehe)]

I am not sure, if this fix is related to this issue: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1056380

We are facing the same problem, that is described in the Debian bug report with every version of reprepro >= 5.3.1-1ubuntu1 on Ubuntu. reprepro occasionally hangs indefinitely waiting on unzstd.