[EN] Compiler Compilation Complication = SNAFU

Posted on August 27, 2008
Filed Under English

After upgrading my server from FreeBSD 7.0-STABLE to 7.1-PRERELEASE, I ran into some serious problems today. The server came back online, but it just didn’t feel right. When I tried to recompile Sendmail (I use a heavily modified version) and BIND (I use the ports version), I ran into some unclear errors. Then I decided to ‘csup’ again and rebuild the entire operating system. The ‘make buildworld’ failed on the first bit of C code. Ah .. the C compiler was fscked. That’s a pretty hefty setback if you compile your own stuff, including your own compiler 😉

After consuming some coffee and tobacco, I stared at the ceiling for a few minutes (no C compiler there). And then it hit me (‘and it felt like a kiss’): I had my brand-spanking-new laptop with me, with a very recent FreeBSD 7 on it!

So I copied ‘cc’, put it on the server, and symlinked it from its original location. Now it complained about ‘cc1’. Ok, copy and symlink. And the same for ‘cc1obj’, ‘cc1plus’, ‘ld’, and ‘as’. And then .. it compiled! I’m still in the ‘make buildworld’ phase, so fingers crossed. But I managed to compile Sendmail and BIND in the meantime, so I guess it’s ok.

Anyway, there’s a lesson in there: always install FreeBSD on a laptop and keep it within reach!

Alas: something appears to be seriously wrong at FreeBSD HQ. The programs that were failing before (see above) returned 8 bytes bigger than before, but the same problem happened again, even though a ‘csup’ did fetch quite a lot of new sources. A compiler error message read: “C compiler cannot create executables”. If I’m not mistaken, that’s what it’s for … Time to hit some bulletin boards and Bugzilla pages. Meanwhile, I’m trying a third time with a ‘make clean’ in /usr/src.

The third attempt was successful. I saw an updated src/Makefile creep in during a csup. It may have something to do with it. I also cleared out the CPUTYPE setting from /etc/make.conf, though I hardly think that was the problem, since it’s been in there for over a hundred builds on dozens of machines, ever since the Pentium4 target was added. Anyhoo, back in business.

Oh, and this is what bit me.


Comments are closed.

Bad Behavior has blocked 11 access attempts in the last 7 days.

Better Tag Cloud