-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Duncan wrote: > Alternatively, you may be able to work with someone who knows a quite a > lot more than I do about gcc and/or glibc, and be able to find the > problem. For a start, after the broken compile, you should have a > config.log file in the build dir, that will tell you what lines of the > configure script actually went wrong and how, and you (or someone else who > knows about it) can work from there at discovering the problem. >
Haven't solved the problem completely, but before somebody else invests too much effort brainstorming this I think I've narrowed it down. Besides the aforementioned logs the build process also left behind the the xgcc binary. I ran it with the options listed in the error output and it did in fact compile a test.c fine. A 64-bit a.out worked great, but a 32-bit a.out gave an error about some missing symbol. I noticed that the path to the libc was in my 32-bit chroot though. Sure enough ldd showed that it had been linked against the /usr/lib/libc in my chroot and not the emul-linux-86 libc. Well, my 32-bit glibc had been upgraded just the other day, and I'm guessing that this is the source of the problem. Not sure why stuff is getting linked against my chroot. I checked ld.so.conf and the chroot is listed way at the bottom. So, quick solution is to chmod the chroot so that portage can't see it and rebuild. I'll probably leave the permissions that way while I do my general system rebuild so that nothing else gets linked this way. Still rebuilding, but I see that it is using the bootstrap compiler so I think I'm in the clear... -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFE+X4ZG4/rWKZmVWkRAqSlAJ9J5RS4w3oavBUYGV5/W6uLerGK9gCfT41f 8ZkA1yTHFVT9PZHmLpXTdnU= =Kh2q -----END PGP SIGNATURE-----
smime.p7s
Description: S/MIME Cryptographic Signature
