-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Package: g++-4.3
Version: 4.3.1-4
Severity: important

Good morning,

it seems as there's an older bug in (maybe) g++.
The package znc won't build on the following architectures:

hurd-i386
kfreebsd-i386
kfreebsd-amd64

It FTBFS with the following error message:

> i486-gnu-g++ -Wl,--export-dynamic -o znc String.o Csocket.o main.o
znc.o User.o IRCSock.o Client.o DCCBounce.o DCCSock.o Chan.o Nick.o
Server.o Modules.o MD5.o Buffer.o Utils.o FileUtils.o HTTPSock.o
Template.o -ldl -lssl -lcrypto
> znc.o: In function `CLockFile::TryExLock(CString const&, bool)':
>
znc.cpp.text._ZN9CLockFile9TryExLockERK7CStringb[CLockFile::TryExLock(CString
const&, bool)]+0x6a): undefined reference to `__open_2'

How you can see at [1] it just happens with -D_FORTIFY_SOURCE=2 -
without fortify_source it's compiling, linking and working well.
So on FORTIFY_SOURCE=2 is optimizing something in a quite bad way?

This is at znc the evil line which causes the FTBFS:
m_fd = open(sFile.c_str(), bRw ? O_RDWR : O_RDONLY);

You can find it in Utils.h arround line 112.


The author of the software znc Uli Schlachter has written the
"testcase.cpp" where you could also test it (with some notes).

As the build logs on [1] say, this bug also exists in stable and lenny,
but I haven't got a hurd/kfreebsd stable/testing buildd.


[1]: http://buildd.debian-ports.org/status/package.php?p=znc

- --
/*
Mit freundlichem Gruß / With kind regards,
Patrick Matthäi

E-Mail: patrick.matthaei@web.de
*/

// Comment:
// Always if we think we are right,
// we were maybe wrong.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkhvOp8ACgkQ2XA5inpabMfnggCfSKSjPEHs6C 5utZTC4SHRrzIe
OXoAoJPiIMa4lWzQOuGBCF/fAxgLOolR
=yrdE
-----END PGP SIGNATURE-----