Re: "make installworld" fail r363660 -> r363689 (amd64)

On Thu, Jul 30, 2020 at 6:24 AM David Wolfskill <david_at_catwhisker.org> wrote:
>
> build{worl,kernel} and installkernel were uneventful, but:
>
> ...
> install -l h -o root -g wheel -m 555  /usr/bin/mail /usr/bin/Mail
> install -l h -o root -g wheel -m 555  /usr/bin/mail /usr/bin/mailx
> ===> usr.bin/msgs (install)
> install  -s -o root -g wheel -m 555   msgs /usr/bin/msgs
> install  -o root -g wheel -m 444 msgs.1.gz  /usr/share/man/man1/
> ===> usr.bin/bmake (install)
> install  -s -o root -g wheel -m 555   make /usr/bin/make
> install  -o root -g wheel -m 444 make.1.gz  /usr/share/man/man1/
> ===> usr.bin/bmake/tests (install)
> ld-elf.so.1: /common/S4/obj/usr/src/amd64.amd64/tmp/legacy/usr/sbin/make: Undefined symbol "regcomp_at_FBSD_1.6"
> *** Error code 1
> *** Error code 1
> *** Error code 1
> *** Error code 1
> *** Error code 1
>
> Stop.
> make[1]: stopped in /usr/src
> *** Error code 1
>
> Running:
> FreeBSD freebeast.catwhisker.org 13.0-CURRENT FreeBSD 13.0-CURRENT #986 r363660M/363660: Wed Jul 29 03:52:03 PDT 2020     root_at_freebeast.catwhisker.org:/common/S4/obj/usr/src/amd64.amd64/sys/GENERIC  amd64 1300102 1300102
>
> src/UPDATING has a recent entry:
> 20200729:
>         r363679 has redefined some undefined behavior in regcomp(3); notably,
>         extraneous escapes of most ordinary characters will no longer be
>         accepted.  An exp-run has identified all of the problems with this in
>         ports, but other non-ports software may need extra escapes removed to
>         continue to function.
>
> but while there does seem to be some relevance, I'm not seeing what
> sort of evasive maneuvers I need to make.
>
> I would appreciate a suggestion.
>

Hi,

Can you describe the environment in which you're running installworld,
please? i.e. is it just a raw installworld directly in your shell, or
something more complicated?

I observed this in testing an exceptional scenario; running
installworld in a buildenv. installworld injects .WAIT between lib and
libexec + other subdirs, which is supposed to prevent stuff like this
(new binary got installed linked against new libc before new libc).
Running in a buildenv set SYSROOT and stripped out the .WAITs, leaving
me with an annoyance where I had to installworld twice.

Thanks,

Kyle Evans