I just tried out 20200818-systemd to see the new organization in action. A few comments:
- The biggest issue I ran into was that the "chown -R tester ..." commands initially just output "retaining ownership as root" on all files. I eventually traced down the cause to tester's uid being assigned as 0 due to my environment. In my case, the cause of that was my habit of running systemd-nspawn --as-pid2 -D "$LFS" in place of chroot "$LFS" (and sometimes, I also add --private-network, though not in this run through). But I could also imagine that being the case if the LFS build were run on a Linux console where somebody logged in directly as root. Then, I changed the UID manually to 1000 which mostly resolved the issues, except in a few cases where the test suite expected to be able to open $(tty) for writing. (And there were a few extra test failures because of running in a seccomp environment, but that was completely expected.) - I wonder whether in section 7.14.1, it would make sense to use $LFS/tools/bin/$LFS_TGT-strip instead of the host system's strip. - I wonder if there's any good reason to continue building telnet from inetutils. At this point, about the only valid use I could think of for that executable would be directly connecting to some other service type using a text-based protocol for testing; and for that use case, there are better dedicated tools for that purpose, e.g. socat. -- Daniel Schepler -- http://lists.linuxfromscratch.org/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
