On Mon, 31 Dec 2018 13:12:15 +0000, Dimitry Andric wrote: > On 31 Dec 2018, at 13:08, Matthias Apitz <guru_at_unixarea.de> wrote: >> >> On Mon, 31 Dec 2018 12:27:55 +0100, Matthias Apitz wrote: >>> >>> Hello, >>> >>> I'm updating the usual way a system from r314251 to tje >>> actual CURRENT r342378. >>> The 'make installworld' failes in single user mode with: >>> >>> ld-elf.so.1: /usr/obj/usr/src/amd64.amd64/tmp/legacy/usr/sbin/rm: >>> Undefined symbol lstat1_at_FBSD_1.5 >>> >>> What can I do? >> >> As I had no other idead, I run the 'make installworld' again, >> now with success. >> >> What does this mean? > > I think what may have happened was the following. > > The "legacy" rm which failed was likely a symlink to /bin/rm, like so: > > $ ls -l /usr/obj/usr/src/amd64.amd64/tmp/legacy/usr/bin/rm > lrwxr-xr-x 1 dim dim 7 2018-12-29 16:22:12 > /usr/obj/usr.old/src/amd64.amd64/tmp/legacy/usr/bin/rm_at_ -> > /bin/rm > > For some reason, /bin/rm may have been overwritten before your > /lib/libc.so.7, which caused the "Undefined symbol" error from > ld-elf.so.1. > > Alternatively, installing /lib/libc.so.7 may have failed without the > Makefile noticing it, after which the first /bin/rm would show that same > error. > > When you re-ran "make installworld", the installation of /lib/libc.so.7 > likely succeeded, and after that the symlinked commands in > /usr/obj/usr/src/amd64.amd64/tmp/legacy/usr/bin which used new symbols > started to work. > I still have two identical updates scheduled, based on extactly the same src and precompiled obj tree. I will investigate the next and file a PR. matthias -- Sent using Dekko from my Ubuntu device http://www.unixarea.de/ +49 176 38902045Received on Mon Dec 31 2018 - 13:20:12 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:19 UTC