This is the mail archive of the
libc-help@sourceware.org
mailing list for the glibc project.
RE: Allin Cottrell: ld-2.10.1.so
- From: gabriele balducci <balducci at dschgrazlin2 dot units dot it>
- To: libc-help at sourceware dot org
- Cc: cottrell at wfu dot edu
- Date: Fri, 15 May 2009 16:18:50 +0200
- Subject: RE: Allin Cottrell: ld-2.10.1.so
Hi,
> I wonder, can anyone replicate, or explain away (or account for in
> any way) the segfaults in ld-2.10.1.so that I reported here
> (libc-help archive, May 13)?
>
> As I mentioned, the segfaults resulted from dereferencing a NULL
> value of the pointer variable "dl_random" in the function
> _dl_setup_stack_chk_guard (defined in
> sysdeps/unix/sysv/linux/dl-osinfo.h). And the problem went away
> when I manually undef'd __ASSUME_AT_RANDOM, the definition of
> which is the trigger for accessing dl_random (in the function
> security_init in elf/rtld.c, which includes dl-osinfo.h).
>
> FWIW, I note that in security_init, dl_random is set to NULL after
> it is used.
>
I do not know if this makes sense for you, but I noticed that you built
with gcc-4.4.0:
> waverley:~$ gcc -v
> Using built-in specs.
> Target: i686-pc-linux-gnu
> Configured with: ../configure --prefix=/opt/gcc4 --with-arch=i486
> --infodir=/usr/share/info --mandir=/usr/share/man --disable-nls
> --enable-languages=c,c++,fortran
> Thread model: posix
> gcc version 4.4.0 (GCC)
This caused troubles for me with 2.10.1: various programs started
segfaulting immediately after installation.
After cleaning and rebuilding with gcc-4.2.1 everything was (and is
still) fine. This on two different (gnu-linux) boxes.
I did not go through deeper than this: using gcc-4.2.1 instead of latest
gcc-4.4.0 just fixed everything.
Actually, the INSTALL file says:
* GCC 3.4 or newer, GCC 4.1 recommended
Hope this helps
ciao
gabriele
--
Gabriele Balducci - Dipartimento di Scienze Chimiche - Via L. Giorgieri 1
I-34127 TRIESTE tel: I-040-5583957 fax: I-040-5583903 e-mail: balducci@units.it
Please, if possible, don't send me MS Word or PowerPoint attachments
Why? See: http://www.gnu.org/philosophy/no-word-attachments.html