On Sat, Dec 01, 2012 at 11:59:07PM +1300, Simon Geard wrote: > On Fri, 2012-11-30 at 11:17 -0500, Chris Staub wrote: > > First, check how Bash is linked: "readelf -l /bin/bash | grep > > interpret". Of course, this should say that it's looking for the dynamic > > linker in /lib. Then verify you actually have all the right libraries > > for Ncurses: "ls -la {/usr,}/lib/*ncurses*" > > Yes, it's using /lib64/ld-linux-x86-64.so.2, and yes, the chapter 6 > version of ncurses *is* correctly installed. But that copy of ncurses is > the wide-char version, and bash has managed, somehow, to link to the > non-wide version. My assumption, as I said, is that instead of finding > the linker scripts in /usr/lib which would point it to the wide version, > it's found the non-wide version in /tools instead. >
I don't remember seeing anything like this before. In the distant past, I've had my own fun and games with ncurses when I fubar'd some of the moves and symlinks. So, is your _wide_ version of libncurses ok ? I'm thinking perhaps something went wrong when you moved it to /lib, so that the /usr/lib/libncursesw.so symlink is pointing to a non-existent file. Looking at my own *logs* from bash, it uses -lcurses so also check that /usr/lib/libcurses.so is a symlink to libncurses.so. If you ran the verification tests when you adjusted the toolchain, and after installing gcc, then my money is on an error *after* that. Of course, if you didn't run those checks (the SEARCH_DIR etc) then all bets are off. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page