I have been questioning people in #gentoo with the following problem for quite some time: error while loading shared libraries: libstdc++.so.5: cannot open shared object file: no such file or directory All of them did emerge something before it happened (heavy load), some were failed emerges, some successful. All of them used reiserfs on their / partition except one who used xfs. All of them either used gentoo-sources 2.4.x or 2.6-testx. All, who managed to tell, had preempt enabled. The problem is fixed by this command: ls -d /usr/lib/gcc-lib/*/* > /etc/ld.so.conf; ldconfig; env-update Pure coincidence? I think not. I had that problem myself with gentoo-sources, never had it since I solely use vanilla-sources (2.4.20 2.4.21). Alltogether I asked ~40-50 people within the last two months. The question is does gentoo's compiling trigger some bug with reiser+preempt under heavy load? What other information might be relevant in finding the reason for that common problem? roger55 Reproducible: Always Steps to Reproduce: 1. 2. 3.
there were some bugs with gcc updates and env.d not being updated properly ... if you run `env-update ; . /etc/profile` does the bug return ?
No reply... NEEDINFO.
<vagwin> roger55: ok. here u go... <roger55> vagwin: ... waiting ... <vagwin> usr/bin/python: error while loading shared libraries: libstdc++.so.5: cannot open shared object file: no such file or directory. <roger55> vagwin: ok <roger55> vagwin: reiserfs? <vagwin> yup <roger55> vagwin: what sources? <vagwin> kernel? <roger55> yes <vagwin> 2.4.22-gentootest <vagwin> -r1 <roger55> preempt enables? <vagwin> yup <roger55> vagwin: hang on please... <vagwin> roger55: k. thanks. <roger55> vagwin: if you run `env-update ; source /etc/profile` does it work again? <vagwin> nope <roger55> vagwin: really? <vagwin> cant run env-update. get same error. <vagwin> its a pyython thing... so it's not fixed by Spanky's advice.
seems like a broken gcc spider@Darkmere> locate libstdc++.so.5 /usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/libstdc++.so.5.0.5 /usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/libstdc++.so.5 spider@Darkmere> qpkg -v -f /usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/libstdc++.so.5 sys-devel/gcc-3.3.2-r3 *
some more info: <roger55> I want to see the contents of ld.so.conf <-- buggy (~bug@dyn-202-89-162-69.arach.net.au) has left #flood ("Leaving") <vagwin> ok. i will be typing it tho. <vagwin> usr/local/lib <vagwin> usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2 <vagwin> usr/lib/opengl/nvidia/lib <vagwin> usr/lib/mozill <vagwin> *mozilla <vagwin> usr/X11R6/lib <vagwin> opt/blackdown-jdk-1.4.1/jre/lib/i386/ <vagwin> ditto native_threads <vagwin> ditto classic <vagwin> usr/qt/3/lib <vagwin> usr/kde/3.1/lib <vagwin> thats it. .... <roger55> tell me this first: cat /etc/env.d/05gcc <vagwin> cannot k <vagwin> PATH="/usr/i686-pc-linux-gnu/gcc-bin/3.3 <vagwin> ROOTPATH="/usr/i686-pc-linux-gnu/gcc-bin/3.3 <vagwin> (which looks wrong) <vagwin> MANPATH="/usr/share/gcc-data/i686-pc-gnu/3.3/man <vagwin> INFOPATH="/usr/share/gcc-data/i686-pc-linux-gnu/3.3/info <vagwin> CC="gcc" <vagwin> CXX="g++" <vagwin> LDPATH="/usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2" <roger55> vagwin: locate libstdc++.so.5 <vagwin> env-update still gives python missing libstdc++ errors. <vagwin> locate libstdc++.so.5 <vagwin> ooops wrong keyb... <roger55> vagwin: :) <vagwin> 3 entries: /opt/openoffice1.1.1/program/libstdc++.so.5 <vagwin> usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/libstdc++.so.5 <vagwin> ditto .so.5.0.5
<vagwin> PATH="/usr/i686-pc-linux-gnu/gcc-bin/3.3 <vagwin> ROOTPATH="/usr/i686-pc-linux-gnu/gcc-bin/3.3 That doesn't match reality. Have him change those to the real matching path : ROOTPATH="/usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/" (or make a symlink between 3.3 and 3.3.2 in both places. )
in other words this looks like the gcc upgrade bug, not some kind of bug related to reiserfs
